Sensing and Dynamics in DNA Nanotechnology - Actuators, Biosensors, and Chiroptical Switches

PhD Thesis by Martin Kristiansen Department of Chemistry Aarhus University November 2011

A BSTRACT

he following report has been divided into three parts, each consisting of a separate project. The projects are all in the context of DNA nanotechnology but aside from this, they do not have much in common. Preceding the description of these three projects is a chapter on DNA nanotechnology in general. The aim of this chapter is to give a general impression of the scope of the field and of how the field has developed through the years. In the first project, the facilitation of a novel linker for covalent attachment to electrode surfaces is described. The linker is based on the in situ generation of a diazonium salt from a more stable triazene compound. Following the conversion of the triazene by the addition of dimethyl sulfate, the liberated diazonium salt is electrochemically reduced following known grafting techniques. The linker was used in the immobilization of DNA onto electrode surfaces and subsequent problems in quantifying the immobilized DNA led to change in the linker design. By incorporating a ferrocene moiety, the immobilized DNA was easily quantified and the ferrocene unexpectedly also served as a reporter for hybridization of the immobilized DNA to a complementary strand. The work behind the second project described was performed during my visit in the groups of Prof. James Canary and Prof. Ned Seeman at New York University in the fall of 2010. In the project fluorophores were conjugated to a DNA sequence that in its double helical form is known to undergo a transition in secondary structure from B-DNA to Z-DNA. The transition was monitored using circular dichroism (CD), and it was shown that two fluorophores on the same piece of double stranded DNA produced a signal in the CD spectra, indicating that they combined to form a chiral exciton coupling. The intent was to show that chirality of the exciton would be inverted as the DNA changed from B- to Z-DNA however, while I believe that this would be possible, time was too short to complete the

T

iii

project. The third project was a continuation of a project in Kurt Gothelf’s group pertaining to the recently published DNA tile actuator. The DNA tile actuator is a set of two so-called piston strands that are both hybridized to each their side of a circular roller strand between them. Sequence homology allows for both piston strands to hybridize equally well to the two regions that lie between piston specific regions. As a consequence, the piston strands can “roll” on the roller strand while continuously changing hybridization. Adding so-called lock strands that hybridize to the unhybridized regions of the piston strands will have the effect of fixing the actuator in a specific geometry. The project described here consists of the expansion of this concept with the intent of showing that multiple actuators can be joined together to eventually produce larger, more complex dynamic systems. Thus far, results have indicated that two actuators have been joined together, and that the locking of one of the actuators, fixes the second actuator as well.

iv

R ESUM E´

ølgende ph.d.-afhandling er inddelt i tre dele, som hver især beskriver et projekt inden for feltet DNA-nanoteknologi. De tre projekter indg˚ar alle som en del af mit arbejde under mit ph.d.-forløb, og selvom de alle tre hører under DNA -nanoteknology, s˚a har de ikke meget andet til fælles. Forud for projektbeskrivelserne er et kapitel, der har til form˚al at give en generel introduktion to DNA -nanoteknologi. Dette kapitel er skrevet med henblik p˚a at give en indsigt i, hvor bredt feltet spænder, samt hvordan feltets historiske udvikling har været igennem a˚ rene. Det første projekt omtalt i denne afhandling omhandler udviklingen af en triazenelinker til at binde DNA kovalent til elektrodeoverflader med henblik p˚a fremstillingen af biosensorer. Teknikken beroer p˚a in situ-omdannelsen af triazene til det mere reaktive diazonium salt, som følgende kan reduceres elektrokemisk og danne en kovalent binding fra et kulstofatom i linker-molekylet direkte til elektrodeoverfladen. Den udviklede triazenelinker blev brugt til at immobilisere DNA p˚a en elektrodeoverflade, men problemer med kvantifiseringen af immobiliseret DNA for˚arsagede problemer. For at afhjælpe dette problem, blev linkerdesignet ændret s˚aledes, at selve linkeren indeholdt ferrocen. Det var h˚abet, at denne ændring ville kunne give muligheden for at m˚ale mængden af immobiliseret DNA direkte under antagelse af, at den var den samme som mængden af immobiliseret ferrocen, idet denne kan m˚ales elektrokemisk. Ved et heldigt tilfælde viste det sig, at det indkorporerede ferrocen ikke alene tjente dette form˚al vel, det redox-potentiale afhang ogs˚a direkte af, om det immobiliserede DNA var enkelt- eller dobbeltstrenget. Med andre ord, havde ferrocenlinkeren en sensoreffekt, som kunne p˚avise tilstedeværelsen af komplementær DNA i en opløsning. Arbejdet bag det andet projekt beskrevet i denne afhandling blev udført under et seks m˚aneders besøg p˚a New York University i efter˚aret 2010, hvor jeg besøgte

F

v

grupperne ledt af Prof. James Canary og Prof. Ned Seeman. Projektet gik i al sin enkelthed ud p˚a at konjugere fluoroforer til DNA-sekvenser, som i dobbeltstrenget form findes med to forskellige sekundære strukturer: den normale højredrejede B-form og den venstredrejede dobbelte helix kendt som Z-formen. Hvilken af de to former, den dobbeltstrengede DNA forefindes i, afhænger af koncentrationen af specifikke kationer i opløsningen, og en overgang fra den ene form til den anden kan kontrolleres direkte ved at styrre kationernes koncentration. Tanken var s˚aledes, at ved p˚ahæftningen af to fluoroforer p˚a samme DNA-helix, ville de to koble og danne et fælles eksiteret stadie, kaldet en eksiton, og kiraliteten af denne eksiton ville afhænge af kiraliteten af DNA-helixen. Man ville s˚aledes kunne probe den lokale sekundere struktur, hvor fluoroforerne var p˚ahæftede, ved hjælp af circulær dikroisme-m˚alinger (CD). Idet fluoroforerne ville afgive et signal ved en helt anden bølgelængde end DNA, ville den sekundære struktur for alt andet DNA i opløsningen ikke overdøve signalet fra fluoroforerne. Man ville s˚aledes kunne teste den sekundære struktur for et meget lille udsnit af DNA tilstede i en opløsning. Det blev vist, at fluoroforerne danner eksitoner, n˚ar der er to tilstede p˚a samme helix, og at disse eksitoner giver et stærkt signal i CD-m˚alinger. Dette betyder, at de ellers akirale fluoroforer har dannet en kiral eksiton, og kiraliteten m˚a nødvendigvis komme fra DNA. P˚a grund af den begrænsede tid til at afslutte projektet, lykkedes det aldrig at invertere eksitonens kiralitet sammen med overgangen fra B- til Z-DNA. Indledende eksperimenter virkede positive, men et endeligt bevis er endnu ikke fundet. Det tredje og sidste projekt beskrevet i denne afhandling var en opfølgning p˚a et projekt i Kurt Gothelfs gruppe, som fornyligt havde ledt til en publikation af en s˚akaldt DNA aktuator. Denne aktuator var i al sin enkelthed konstrueret af to DNA -strenge kaldet piston-strenge, som ved hjælp af homologi i basesekvenserne kunne “rulle” p˚a en cirkulært hybridiseret streng imellem de to. Fuldstændigt ligesom to tandstænger p˚a et tandhjul. Ved at tilsætte ekstra DNA-strenge, som var komplementære til de dele af piston-strengene, som ikke var hybridiseret, kunne aktuatoren l˚ases i en bestemt konformation efter ønske. Projektet beskrevet her var s˚aledes baseret p˚a at videreføre dette koncept og vise, at flere aktuatorer kunne sættes sammen, og at man herigennem potentielt kan fremstille komplekse, dynamiske DNA nanosystemer p˚a samme vis, som tandhjul kan sættes sammen til at danne større mekaniske systemer. I projektet blev to aktuatorer sat sammen til en dimer. Det blev endvidere vist ved FRET-eksperimenter, at man ved at l˚ase den ene ende af dimeren i en bestemt position ogs˚a dikterede, hvilken position den anden ende af dimeren var i.

vi

ACKNOWLEDGEMENT

would like to express my most profound gratitude to everyone who in some way or form has helped in the making of this report and the underlying research work. First of all, I would like to thank Prof. Kurt Vesterager Gothelf for his guidance, without which I would have been stranded several times, and for his belief in my skills as a scientist, which resulted in the funding of two thirds of my PhD program. For the last third of my program funding, I would like to extend my gratitude to AGSoS, The Faculty of Science, Aarhus University. Dr Menglin Chen deserves my fullest gratitude for all of her preliminary work, her help and her collaboration. For extensive guidance and assistance in the arts of electrochemistry, I would like to thank Ass. Prof. Steen Uttrup Petersen and Ass. Prof. Elena E. Ferapontova. For their tremendous help in my work with 32 P-labeled oligonucleotides, Dr Morten M. Nielsen and Dr Ebbe S. Andersen deserve a special thanks. Likewise does Mrs Eva M. Olsen for the synthesis of a ferrocene-labeled oligonucleotide. For accepting me as a visitor into their groups and providing guidance for the six months I was visiting New York University, I would like to extend my gratitude to Prof. Ned Seeman and Prof. James Canary. In addition I would like to thank the students in the groups for both helping me getting set in the lab and generally providing a good social environment during my visit. In both regards, I would especially like to emphasize Mr Miao Ye and Dr Francesca Gruppi. Following my departure from New York, my work was continued by Dr Zhaohua Dai, and this deserves a special recognition. With regards to the work on the DNA actuator dimer project in particular, I

I

vii

would like to thank Ms Mille B. L. Kryger and Dr Zhao Zhang for sharing with me their experiences helping me reflect on problems that have come up during the project. Their help has been a great contribution. Additionally, Dr Victoria Birkedal has assisted me greatly with FRET measurements and calculations. I am very thankful for her help. A thank is due to the people who helped me proofread this report, Mrs Eva Olsen, Mrs Mille Kryger and Dr Zhao Zhang. Their work is much appreciated, and any errors that might remain are solely my responsibility, not theirs. I would like to thank my girlfriend, Violeta C. Ashby, for her patience and support and generally for being the wonderful person she is. Last, I would like to thank all the members of Group 106 for a pleasant and stimulating working environment where a problem often can be solved by merely asking someone else for their opinion.

viii

P REFACE

he author of this thesis has been enrolled in the 4-year PhD program at the Aarhus Graduate School of Science (AGSoS) at Aarhus University since the summer 2007 and has been part of the Centre for DNA Nanotechnology (CDNA) during the same time span. The work presented here concludes the research work done during these studies. The research presented in this thesis has been conducted under the supervision of Prof. Kurt Gothelf with the exception of a six months visit in the fall of 2010 to the New York University which was under the supervision of Prof. James Canary and Prof. Ned Seeman. The majority of the work presented here has been performed by the author, however, in cases where fellow researchers have contributed to the projects, their work has also been included. In these cases, the individual contributions of participating parties is explicitly stated. The work presented in Chapter 2 has resulted in a scientific publication, which is included in Appendix D. In addition to this, the author synthesized the terpyridine in Fig. 1 as a contribution to the collaboration with the group of Chen Wang at the National Center for Nanoscience and Technology, Beijing 100190, China. This collaboration produced publishable results with a manuscript currently in preparation. This work will not be discussed within this report.

T

..................................... Martin Kristiansen November 2011 ix

N N N

Figure 1. Synthesized 4’-(4-ethynylphenyl)-2,2’:6’,2”-terpyridine for collaboration with Chen Wang at the National Center for Nanoscience and Technology, Beijing.

x

C ONTENTS

Abstract

iii

Resum´e

v

Acknowledgement

vii

Preface

ix

Contents

xi

Abbreviations

xv

1 Introduction 1.1 DNA . . . . . . . . . . . . . . . . 1.2 DNA Nanotechnology . . . . . . 1.2.1 DNA Tiles . . . . . . . . 1.2.2 DNA Origami . . . . . . . 1.2.3 DNA-Templated Reactions

. . . . .

1 1 3 6 8 11

. . . .

15 16 17 18 21

. . . . .

2 Immobilization of dna via triazenes 2.1 Introduction . . . . . . . . . . . . . 2.1.1 Common Grafting Methods 2.1.2 Diazonium Salts . . . . . . 2.1.3 Triazenes . . . . . . . . . . xi

. . . . .

. . . .

. . . . .

. . . .

. . . . .

. . . .

. . . . .

. . . .

. . . . .

. . . .

. . . . .

. . . .

. . . . .

. . . .

. . . . .

. . . .

. . . . .

. . . .

. . . . .

. . . .

. . . . .

. . . .

. . . . .

. . . .

. . . . .

. . . .

. . . . .

. . . .

. . . . .

. . . .

2.2 2.3 2.4

Results . . . . . . . . . . . . . Diazocoupling from Triazenes 2.3.1 Results . . . . . . . . Conclusion and Perspectives .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

3 Transfer of Helix Chirality to Fluorescein 3.1 Introduction . . . . . . . . . . . . . . . . . . . 3.1.1 B-DNA . . . . . . . . . . . . . . . . . 3.1.2 Z-DNA . . . . . . . . . . . . . . . . . 3.1.3 Examples of Z-dna in Nanotechnology 3.2 The Project . . . . . . . . . . . . . . . . . . . 3.2.1 Conjugation of Fluorophores . . . . . . 3.2.2 Click Reaction . . . . . . . . . . . . . 3.2.3 DNA Sequences . . . . . . . . . . . . 3.2.4 Circular Dichroism . . . . . . . . . . . 3.3 Conclusions and Perspectives . . . . . . . . . .

. . . .

. . . . . . . . . .

4 DNA Actuators 4.1 Introduction . . . . . . . . . . . . . . . . . . . . 4.1.1 Dynamic DNA Nanostructures . . . . . . 4.1.2 The DNA tile actuator . . . . . . . . . . 4.1.3 Three- and Four-way DNA Actuator Tiles 4.2 Project Aim . . . . . . . . . . . . . . . . . . . . 4.3 Actuator Design . . . . . . . . . . . . . . . . . . 4.3.1 The Actuator Dimer . . . . . . . . . . . 4.3.2 Generating Base Sequences . . . . . . . 4.4 Results . . . . . . . . . . . . . . . . . . . . . . . 4.4.1 The Unlocked Actuator Dimer . . . . . . 4.4.2 Adding Locks . . . . . . . . . . . . . . . 4.4.3 FRET . . . . . . . . . . . . . . . . . . . 4.5 Conclusion and Perspective . . . . . . . . . . . .

. . . .

. . . . . . . . . .

. . . . . . . . . . . . .

. . . .

. . . . . . . . . .

. . . . . . . . . . . . .

. . . .

. . . . . . . . . .

. . . . . . . . . . . . .

. . . .

. . . . . . . . . .

. . . . . . . . . . . . .

. . . .

. . . . . . . . . .

. . . . . . . . . . . . .

. . . .

. . . . . . . . . .

. . . . . . . . . . . . .

. . . .

. . . . . . . . . .

. . . . . . . . . . . . .

. . . .

. . . . . . . . . .

. . . . . . . . . . . . .

. . . .

24 31 33 40

. . . . . . . . . .

41 41 42 43 45 48 50 51 52 53 59

. . . . . . . . . . . . .

61 62 62 64 69 70 72 72 74 77 78 80 83 91

5 Epilogue

95

References

97 xii

A Experimental Section A.0.1 Immobilization of DNA via Triazenes . . . . . . . . . . . A.0.2 Transfer of Helix Chirality to Fluorescein . . . . . . . . . A.0.3 DNA Actuators . . . . . . . . . . . . . . . . . . . . . . .

107 107 111 114

B Python Code

117

C DNA Sequences

131

D Publications

135

xiii

A BBREVIATIONS

A AFM bp C CD CuAAC CV DAC DAE DAO DIPEA DMF DMTMM DNA DPV ds DX ECCD EDC FDCD FRET G

Adenine Atomic Force Microscopy base pair Cytosine Circular Dichroism Copper(I) mediated azido-alkyne [2+3]-cycloaddition Cyclic Voltammetry/Voltammogram Diammonium Citrate Double Crossover, Antiparallel, Even Spacing Double Crossover, Antiparallel, Odd Spacing Di-iso-propylethylamine Dimethyl Formamide 4-(4,6-Dimethoxy-1,3,5-triazin-2-yl)-4methylmorpholinium Deoxyribonucleic Acid Differential Pulse Voltammetry Double Stranded Double Cross-over Exciton Coupled Circular Dichroism 1-Ethyl-3-(3-dimethylaminopropyl) Carbodiimide Fluorescence Detected Circular Dichroism F¨orster Resonance Energy Transfer Guanine xv

GC HHAQ HPA HOPG MALDI-TOF-MS NHS PAGE PCR R.t. RP-HPLC SAM SWNT T THF TSTU UV/vis XOR

Glassy Carbon 1,2,3,5,6,7-hexahydroxyanthraquinone 3-Hydroxypicolinic Acid Highly Ordered Pyrolytic Graphite Matrix-Assisted Laser Desorption-Ionization Time of Flight Mass Spectrometry N-Hydroxysuccinimide Polyacrylamide Gel Electrophoresis Polymerase Chain Reaction Room Temperature Reverse Phase High Performence Liquid Chromatography Self-Assembled Monolayer Single-Walled Carbon Nanotubes Thymine Tetrahydrofuran 1,1,3,3-Tetramethyluronium Tetrafluoroborate Ultra Violet/Visual Range Exclusive Or

xvi

C HAPTER 1

I NTRODUCTION

n 1953, James D. Watson and Francis Crick proposed a model for the structure of double stranded DNA explaining the X-ray data collected by Rosalind Franklin and Maurice Wilkins. 1 This model was the DNA double helix that you now see on the first pages of every molecular biology textbook, and on a great extent of their book covers as well. Aside from revolutionizing the field of molecular biology (a term coined only a few years prior), 2 little did Watson and Crick know that their discovery also laid the foundation for what would eventually become DNA nanotechnology. Still a somewhat small field in comparison, but nonetheless an interesting one. In the following chapters, three separate examples of DNA nanotechnology based projects will be presented. First, a brief history of the field.

I

1.1 DNA n order to understand the mechanics of DNA nanotechnology, one must first understand the self-assembling properties and the sequence specificity of double stranded (ds-) DNA. As illustrated in Fig. 1.1, DNA consists of a poly-2’-deoxyribose phosphate backbone, where the deoxyribose units are linked from the 5’-oxygen via a phosphodiester bridge to the 3’-position on the next sugar moiety. On each sugar, one of the four nucleobases - adenosine (A), guanine (G), cytosine (C) and thymine (T) - is situated in the anomeric β-position. The famous double helical structure arises when two complementary DNAstrands pair up to form ds-DNA. The process where the two single strands pair up is referred to as hybridization of the strands.

I

1

2

C HAPTER 1. I NTRODUCTION

Upon hybridization of complementary DNA strands, the nucleobases associate in so-called Watson-Crick base pairs, where A and C match up with T and G, respectively. The base pairing is enforced by hydrogen bonding between the bases as illustrated in Fig. 1.1 on the right, making the pairing highly selective. It is this selectivity that makes DNA a unique tool that not only stores all of our genetic information, but additionally can be used to design and create nano-scaled structures with the DNA strands as building blocks. It allows us to specifically choose the exact position of each single DNA strand and which other strands they bind to in these structures containing hundreds of different strands. To add to the stability of the DNA helix, the π-systems of the bases interact in the form of π-stacking, resulting in a largely hydrophobic core with the sugar phosphate backbone wrapping around it in opposite directions, thus forming the helix. This separation of the hydrophobic core surrounded by the highly hydrophilic backbone adds to the high stability of the helix in aqueous media in the same way as a micell. 3 a)

b) Thymine

OH

O

Adenine

H

HN

H

N N

O

OO P O

O

N O H H N

Guanine

N

HN H

Cytosine

5' -

O

3'

O P O O-

H N

O

N

O

Adenine

H H N N N H

OO P O

N

H NH

Thymine O

O N

O O

N

O

N N

O P O O

O

Guanine HN H

O

N O

N H N

O

OO P O

O

N

O

N O P O O-

O

Cytosine

N

5'

O

N

N

3' -

O O P O O-

H

O

N N

OO P O O

N H

O

OH

Figure 1.1. a) A 3-dimensional rendering of a DNA helix illustrating the intertwining of the sugar phosphate backbones with the nucleobases pairing up along the core of the helix. b) A molecular schematic representation of the four nucleotides and how they base pair according to the Watson-Crick model. Nucleotides are color coded so that the colors in the 3D model correspond to the colors in the molecular illustration.

3

1.2 DNA Nanotechnology he term nanotechnology is often very loosely defined as the science of manipulating materials on an atomic or molecular scale. 4 As a result, the term can be argued to encompass the entire fields of chemistry, molecular biology and others. The concept of DNA nanotechnology can generally be seen as a subsection of nanotechnology, viz the part that includes DNA. Often, however, it is assumed that artificially designed DNA systems are implied, rather than naturally occurring ones. As such, the term can be applied to a very wide range of research work, and it is beyond the scope of this report to supply an exhaustive account of the work being done in the field. The conception of DNA nanotechnology is often attributed to Nadrian Seeman when he in the early 1980’s was attempting to crystallize proteins for X-ray diffraction studies. As can be seen in Seeman’s own, somewhat facetious illustration of the protein crystallization process (Fig. 1.2), this was not always a trivial task. In the hope of adding an additional arrow to the flow chart denoting successful crystallization, Seeman looked to DNA.

T

Figure 1.2. Nadrian Seeman’s original illustration of the protein crystallization protocol. The illustration can be found on Seeman’s website. 5

Reportedly, Seeman was inspired by Depth, a woodcut by M. C. Escher (See Fig. 1.3 left), and found it to have great similarity to an array of six-armed DNA

4

C HAPTER 1. I NTRODUCTION

junctions forming a crystal. In the woodcut, Escher has aligned what appears as an infinite quantity of fish in a (simple cubic) crystal-like lattice, head to tail and with orthogonal fins defining the two remaining cubic axes. Seeman’s theory was that a six-armed DNA junction could be made to auto-associate in the same fashion, creating a crystalline lattice defined by the design of the junction. If then, subsequently, the DNA could be made to arrange proteins within the lattice, as per Seeman’s original illustrations in Fig. 1.3 middle and right, the X-ray diffraction data would soon be within reach. a)

b)

Figure 1.3. a) M. C. Escher’s 1955 woodcut, Depth. b) Seeman’s original illustration of DNA crystal matrices for arranging proteins (illustration from Nature 2003). 6 * All M.C. Escher works © 2011 The M.C. Escher Company - the Netherlands. All rights reserved. Used by permission. www.mcescher.com

It was not until 2009, however, that Seeman and coworkers reported the first three-dimensional DNA lattice where the macroscopic crystal structure was dictated by the DNA design. The design was based on the tensegrity triangle illustrated in Fig. 1.4a, which let to crystals of dimensions visible to the naked eye. As can be see in Fig. 1.4b, the symmetry of the crystal clearly reflects the symmetry of the DNA design. 7 In the, almost, 30 years that had passed since Seeman’s original idea was conceived, many advances in the field of DNA nanotechnology had been made to reach this goal. In 1991, Seeman and coworkers published the fabrication of the five-arm and

5

a)

b)

Figure 1.4. The first example of a bottom-up designed macroscale object based on DNA. Figure a) illustrates schematically the design of the DNA sequences. Figure b) is an optical image of the crystals from the DNA strands in a). 7

six-arm DNA branched junctions that Seeman had derived from watching the Escher woodcut. At this point, however, they did not yet form periodic lattices as in Escher’s master piece. 8 Soon to follow, they published the first example of a man-made nanoscale DNA object: the DNA cube illustrated in Fig 1.5, 9 and a few years later, the truncated octahedron. 10

Figure 1.5. The first DNA nanoscaled object by Seeman and coworkers. (Illustration from Nature 2003) 6

While this established the discipline of structural DNA nanotechnology, two fundamental contributions still remained to be seen: DNA tiles and DNA origami. These two design strategies have since accounted for the majority of all larger, artificial DNA structures. The DNA tile was the first of the two to be used in DNA structure design. As the name implies, the concept is largely based on the same principles as Seeman’s DNA junction lattice, and this lattice would indeed be an example of tile assembly. However, to achieve a lattice structure, the tile structures

6

C HAPTER 1. I NTRODUCTION

need to be much more rigid.

1.2.1 DNA Tiles n the first attempts at constructing a sufficiently rigid DNA tile, two-dimensional propagation was the target. To achieve this, parallel DNA helices were interlinked by having the participating strands cross over between the helical domains. This way, motifs were formed that were more rigid than a single DNA double helix on its own. These types of double cross-over motifs were first described by Fu and Seeman in 1993. 11 It was Erik Winfree who, in collaboration with Nadrian Seeman and others, demonstrated that a DNA complex could function as a Wang tile - i.e. a rectangular shaped tile with its edges coded in such a way that only tiles with corresponding edges may pair up edge to edge on a regular grid. Winfree used two different designs, both based on Fu’s double crossover (DX) tiles from six years earlier. Each design consisted of two different tiles, named A and B, and these were designed to associate with each other as illustrated in Fig. 1.6a. The coding of the edges was executed by leaving unpaired, single stranded DNA, also known as sticky ends, protruding from the edges of the tile. Sticky ends of A tiles would be complementary to sticky ends on B tiles in places where the tile edges were meant to align. In Fig. 1.6a, edges with complementary sticky ends on A and B tiles, respectively, are drawn in the same colors. The two tile designs used by Winfree can be seen in Fig. 1.6b. The naming of the tile types are in reference to the geometries and number of helical turns in the DNA strands: double crossover, antiparallel, odd spacing (DAO) and double crossover, antiparallel, even spacing (DAE). 12 After Winfree’s proof of concept paper, other DNA tile designs emerged, 14,15 and the use of the tiles as logical gates was demonstrated only a few years later. 16 In 2004, Winfree demonstrated in collaboration with Paul Rothemund and Nick Papadakis how these logic gate DNA tiles could be used to provide structural control in DNA self-assembly systems. Utilizing the DX tiles from Winfree’s first paper on DNA tiles, they designed the tiles to self-assemble following a XOR premise leading to a Sierpinski triangle pattern. The logic gate premise, XOR (eXclusive OR), is explained in Fig. 1.7A where it can be seen how two adjacent tiles in row t = n decides the value of the tile in row t = n + 1 with which they both share an edge. The tiles are given face values of 0 and 1 respectively. If two adjacent tiles have the same value (both 1 or both 0), then the tile in the next row will be given a value of 0. Conversely, if they are

I

1.2.1. DNA T ILES

7

Figure 1.6. Winfree’s first DX based tile designs. a) depicts a schematic illustration of Winfree’s DNA version of a set of two Wang tiles named A and B. Edges of the same color on A and B tiles have complementary sticky ends protruding, causing the sides to associate. b) illustrates the two tile designs used by Winfree: the DAO design on the left and the DAE design on the right. The illustration is reproduced with modifications from Winfree’s original publication. 12

Figure 1.7. Logically programmed DNA tiles forming a Sierpinski triangle. A) illustrates the logic premise for the assembly, where the face values of adjacent tiles in row t = n dictate the face value of the tile to which the black arrows are pointing in row t = n + 1. B) provides a schematic illustration of how the tiles are designed to accommodate the logic premise. C) shows the four tiles following the same schematic key as used in B, and illustrates the assembly process. D) depicts a larger assembled system of tiles, showing the Sierpinski triangle. E) is an image averaging multiple AFM scans of the actual DNA tiles after self-assembly, arranged into a Sierpinski triangle. Red X’s denote errors in the algorithm, the lowest of which acts as a starting point for the triangle, while the two, upper left errors terminate the propagation. All illustrations are reproduced from the publication of Rothemund et al. 13

8

C HAPTER 1. I NTRODUCTION

different (0/1 or 1/0), the value for the tile in the next row will be 1. To make a DNA version of such an XOR tile, the inputs and outputs were controlled by the complementarity between unpaired DNA strands at edges of the tiles. The face value used for distinguishing between 0 and 1 tiles was produced by including a protruding DNA hairpin structure only on the 1 tiles. This protrusion would make the tile visibly taller in atomic force microscopy (AFM). Fig. 1.7B–D illustrate how the tiles are designed to accommodate the premise and how they assemble, and Fig. 1.7E depicts an AFM image of the self-assembled tiles forming a Sierpinski triangle, as predicted. 13

1.2.2 DNA Origami he second great landmark in structural DNA nanotechnology occurred in 2006 when Paul Rothemund publicized the technique of scaffolded DNA origami. This allowed for the folding of DNA strands into any desired two-dimensional shape. An inherent problem with DNA tiles and other assemblies of short DNA strands is the need for very accurate stoichiometry of the participating DNA strands. Due to experimental error and the requirement for often multiple purification steps, the overall yield of the formation of these nanoscale DNA objects is often low. The low yields, in turn, imposes limitations on the complexity of the objects. Rothemund was inspired by a publication in Nature from 2004 where Willian Shih et al. demonstrated the self-assembly of a DNA octahedron made from 1.7 kilobase single stranded DNA. 17 Rothemund argued that the success of this work suggested that the stoichiometry problems thus far associated with DNA architecture could be circumvented by the use of long strands. In his 2006 publication, Rothemund introduces the name scaffolded DNA origami to describe his and Shih’s work. The name of the technique derives from the Japanese art of folding paper into decorative shapes and figures, and this is very descriptive of Rothemund’s DNA structures. In place of paper, Rothemund folded a circular genomic DNA strand from the M13mp18 virus with a length of 7,249 nucleotides (referred to as the scaffold strand). To fold it, he used 200-250 short synthetic DNA oligonucleotides, also referred to as staple strands. In order to prove the versatility of the technique, he produced 11 different shapes of scaffolded DNA origami, six of which can be seen in Fig. 1.8. Rothemund’s publication resonated throughout the entire field of DNA nanotechnology, and since its introduction in 2006, the concept of DNA origami has spread widely. The technique’s high yields, low sensitivity to uneven stoichiometry, and few defects despite high complexity makes it a relatively simple means

T

1.2.2. DNA O RIGAMI

a)

b)

9

c)

d)

e)

f)

Figure 1.8. Six examples of scaffolded DNA origami as they were presented in Rothemund’s Nature publication. 18 The top row shows the path of the scaffold strand, and the two bottom rows are AFM images of the scaffolded DNA origamis. Images without scale bars are 165 nm x 165 nm. Scale bars for lower AFM images are: b, 1 µm; c–f, 100 µm.

to an end and thus ideal for many purposes where the DNA structure is considered the starting point, rather than the goal of the research. The first example of a functional RNA detection system based on a DNA origami platform was published by Ke et al. in 2008. In their work, they demonstrate the use of DNA origami as a scaffold on which single hybridized oligonucleotides can be imaged by AFM. Ke et al. exploited the fact that each of the staple strands in the origami has a unique position. By extending certain chosen staple strands with a segment that would hybridize with the target RNA, that particular RNA sequence would always bind to that same place on the origami. The rigidity of the formed RNA /DNA double strand made it visible in AFM, causing the positions in question to light up on the image if the target RNA was present. The large number of different staple strands meant that a single DNA origami could be designed to test for the presence of a variety of different RNA strands simultaneously. Different RNA strands simply caused different patterns to form on the image of the origami. 19 Winfree and Rothemund demonstrate in their 2008 publication how the edge of a DNA origami can be used as a seeding point to their programmable DX tiles rather than depending on assembly errors for this (vide supra). Ultimately, Voigt et al. recently showed how a variety of bond cleavage and bond formation reactions can be performed in prespecified locations on DNA origami while absorbed onto a mica surface. Specifically, the copper(I) mediated azido-alkyne [2+3]-cycloaddition (CuAAC) and the NHS-ester mediated amide

10

C HAPTER 1. I NTRODUCTION

formation were performed as examples of bond formation. As for bond cleavage reactions, disulfide reduction and olefin cleavage by triplet oxygen were the reactions of choice. 20 The latter of the two cleavage reactions was also used in the first example of a practical application of chemical reactions on DNA origami surfaces. Helmig et al. monitored the singlet oxygen diffusion originating from a single photosensitizer molecule placed in the center of a DNA origami. By irradiating the photosensitizer, singlet oxygen was produced in situ, diffusing from its center of origin into solution. Biotin was attached across the origami using linkers that would be cleaved upon contact with the singlet oxygen. By simply counting the number of cleaved linkers on recorded AFM images, a statistical basis was achieved for the diffusion of singlet oxygen. 21 As for the DNA origami itself, the technique made a leap from two to three dimensional in 2009 when Andersen et al. constructed the first nanoscaled DNA origami box. The box was designed as six interconnected squares that would ultimately assemble as the sides of the box. This approach would also give the advantage of making it possible to open and close the lid by either adding or removing the staple strands that fix the lid in position. 22 Ke et al. used the same basic principle of folding two-dimensional DNA origami into three-dimensional structure - in their case, a tetrahedron. 23 Douglas et al. demonstrated a different approach to the addition of a third dimension to DNA origami. Instead of weaving the one-dimensional DNA into a twodimensional structure and then, subsequently, folding the two-dimensional sheet into a three-dimensional object, they skipped the intermediate two-dimensional sheet and went directly from one to three dimensions. Where the two-dimensional DNA origami would align the tube-like DNA helices in flat sheets, one next to the other, Douglas’ approach was to order them into a honeycomb lattice type structure. This approach gives dense structures, rather than hollow ones, although cavities can be incorporated by omission of center helices. 24 Some examples of Douglas’ type three-dimensional DNA origami are depicted in Fig. 1.9. In order to achieve a honey comb lattice of DNA helices, Douglas took advantage of the fact that the angle between adjacent helices can be dictated by the number of base pairs between points where they are interconnected, the so-called crossover points. Standard B-DNA has approximately 10.5 base pairs per full helical twist, i.e. there are approximately 34.3° twist in the helix backbone from one base pair to the next. This means that the angle between DNA helices connected by crossovers can be calculated as the number of base pairs between crossovers multiplied by 34.3°. Rothermund’s two-dimensional DNA origami had 16 base pairs between crossovers giving an approximate angle of 180° between helices causing

1.2.3. DNA-T EMPLATED R EACTIONS

11

Figure 1.9. Douglas’ type three-dimensional origami. Illustration from an online interview with William Shih. 25

the structure to be flat. Douglas simply chose crossover spacings divisible by 7 base pairs giving angles consistent with a honey comb lattice. The most recent approach to three-dimensional origami, by Han et al., also builds on the principle of controlling angles between helices. Rather than basing their designs on honey comb lattices, they sculpt the shape of their origami by choosing the appropriate cross-over angle on a helix to helix basis. This makes the technique specially practical in the design of structures containing spherical elements and the likes. 26

1.2.3 DNA-Templated Reactions part of DNA nanotechnology, which also deserves to be mentioned, is DNAtemplated ligation. It certainly has its roots in a time where the term DNA nanotechnology had never been heard of before, but even if it was not originally a part of the field, it is a concept that has been frequently used within the field. In biology, DNA strands serve as templates or blueprints for the reactions of the body. In DNA-directed synthesis, this concept is transferred to the laboratory facilitating a control on the molecular level not seen elsewhere. D NA-templated reactions build on the concept of effective concentration and on DNA’s ability to form duplexes at very low concentrations. The rate of any chemical reaction with more than one participating reactant depends on the concentrations of the species involved. In a simple single-step reaction between two molecules, the rate of reaction is directly proportional to the concentrations of each of the two molecules. If the concentrations are lowered enough, the rate of reaction will eventually be so low that the reaction can be regarded as not occurring at all in given time frame. This is the concentration of interest in DNA-templated

A

12

C HAPTER 1. I NTRODUCTION

reactions. This is because DNA has the interesting property that even at such low concentrations, complementary DNA strands are able to hybridize and form a DNA duplex. If the two DNA strands are carrying reactive groups, these reactive groups will be confined within a relatively small volume as long as the two DNA strands are bound to each other. Since the volume available to the reactants to diffuse away from one another is reduced, it is said that the effective concentration of the reactants is increased. Consequently, the rate of reaction between reactants increases and a product is formed despite the low initial concentrations. The concept is often referred to as a pseudo-intramolecular reaction, since the DNA complex technically does not constitute a single molecule but the principles are the same as if they were. A key point to DNA-templated reactions is that the reaction only occurs when the DNA strands are complementary. This is what provides the control of when a reaction is to occur and when it is not. In 1966 Naylor and coworkers reported the first non-enzymatic DNA-templated reaction. The reaction was a ligation between thymidine hexanucleotides, employing a polyadenosine strand as a template for the reaction. While the product yield was low (approximately 5%), the results were the first of their kind to demonstrate the possibility of using DNA templating to control classic organic reactions through proximity effects caused by the binding to the template. 27 It was not until two decades later, however, that the nucleotide sequence dependency was demonstrated in an experiment by Orgel and coworkers. With a DNA oligonucleotide of five nucleotides in length (dGGCGG), they demonstrated how the complementary oligonucleotide could be synthesized by applying the original strand as a template. 28 Up until this point, the main focus in DNA-templated reactions had been on the ligation of smaller oligonucleotides into larger ones through phosphate ester linkage. Lynn and Goodwin changed this paradigm by exchanging the phosphate ester formation with an imine condensation, thus expanding the scope of chemical reactions templated by DNA (See Scheme 1.1). 29 However, it was David Liu and coworkers who revolutionized the field by introducing the concept of small molecule synthesis governed by DNA control strands conjugated to the reactants. In their first reports, they introduce their methodology by performing SN 2-substitutions and conjugate additions to α, βunsaturated carbonyl systems and to vinyl sulphones via species, which were conjugated to guiding DNA strands as illustrated in Fig. 1.10. 30,31 Liu deployed this new system in the construction of a small molecule screening system. By conjugating a thiol moiety to 1025 different oligonucleotides, where only one of these was ligated to biotin, and then combining them with 1025 different oligonucleotides with maleimide attached, they demonstrated the screen-

1.2.3. DNA-T EMPLATED R EACTIONS

13 O

NH2

NH2

O

N

N

Scheme 1.1. The use of an external template type system by Lynn and Goodwin O N O

O

SH N

O

NH

SH

O NH O

End Template

Hairpin Template

Figure 1.10. The first two types of templates reported by Liu and coworkers. In their original publications, Liu and coworkers used two different DNA template motifs, the End template and the Hairpin template. The two motifs differ in the placement of the reactants along the double helix. In the End template, the reactants are placed at the end of the helix, while the Hairpin template uses a hairpin structure to place the reacting moieties on the side of the helix. In their work, they found that both structures, despite their geometric differences, worked well.

ing potential in their new system. After a reaction time of 10 minutes at 25 °C, the products were selected in vitro for binding to immobilized streptavidine. After PCR amplification of the strands attached to the selected products, analyses of these revealed a 1000-fold enrichment of the biotin-labeled product in comparison with the unselected library. In other words, giving a ratio between biotin-labeled (desired product) and non-biotin-labeled strand of 1:1. 30,31 In the first article, 30 Liu and Gartner report that the number of unhybridized nucleotides between the reacting groups seems to have little to no relevance in whether the groups will react or not, as long as the rate determining step is the annealing of the nucleotides. Exploiting this interesting property, Liu and coworkers later performed 3 consecutive peptide linkages utilizing the DNA template scheme. 32 In recent years, the use of DNA-directed chemistry for the development of small molecule libraries to be used for screening purposes has resulted in the de-

14

C HAPTER 1. I NTRODUCTION

sign of new template motifs. While Liu and coworkers reported that distance between reactants on hybridized motifs is of less importance when the rate determining step is hybridization (vide infra), this is not the case for slower reactions. In an attempt to overcome this problem in combinatorial libraries facilitating more than one reaction per template, the so-called yoctoreactor was developed by Hansen et al. 33 The reactor is constructed in a way placing the reacting moieties half-way into the strand, dividing the strand in two parts. Upon assembly of the motif, one half of the first strand is complementary to one half of a second strand. The second half of this strand fits with a third strand, whose second half is complementary to the second half of the first strand (See Scheme 1.2). The reactions then take place in the center of the structure where the DNA-conjugated reactants meet up sequentially. The yoctoreactor is named after the approximate volume in which the reactants presumably are confined.

Scheme 1.2. Schematic illustration of the yoctoreactor in use. Illustration replicated with some modification from Hansen et al.’s original publication 33

C HAPTER 2

I MMOBILIZATION

OF DNA VIA TRIAZENES

he work presented in this chapter includes contributions from several students over a long period of time. It concerns the use of triazenes - a masked version of a diazonium salt - for immobilization on electrode surfaces, which is often referred to as grafting. The use of diazonium salts in immobilization on surfaces has been known for a few decades and has proven both efficient and versatile. The diazonium group, however, is a labile and reactive specie so combining diazonium chemistry and DNA is not a trivial task. Thus, the project began with Cindy Knudsen, a Master’s student at the time, synthesizing different triazenes, and Ass. Prof. Steen Uttrup Pedersen testing their applicability in grafting on electrode surfaces as illustrated in Scheme 2.1. One of the triazenes was subsequently selected for testing of the immobilization of DNA on electrodes.

T

R

N

N

N

Me2SO4 - Me3N

N

R

N

e- N2 R

Scheme 2.1. The grafting of an aryltriazene onto an electrode surface. In the first step, the diazonium salt is generated by methylation of the terminal nitrogen. In the second step, the diazonium salt is reduced electrochemically, which results in the bond formation between the aromat and the electrode surface.

The synthesis and stability tests of this triazene-DNA conjugate were per15

16

C HAPTER 2. I MMOBILIZATION

OF DNA VIA TRIAZENES

formed by Menglin Chen as part of her work towards her PhD. I became involved with the project when grafting of the triazene-DNA began. Our initial attempts at grafting the triazene-DNA were supervised by Ass. Prof. Steen Uttrup Pedersen. The electrode setup we used is best described as an attempt to miniaturize the setup we had used when grafting triazenes and diazonium salts that did not contain DNA. In those cases, the reacting compounds were abundant in comparison to the DNA samples, so large volumes and high concentrations were not a problem. Our lacking experience in working with electrochemistry on DNA species cost us a lot of extra work, and when Ass. Prof. Elena Ferapontava joined our group, her expertise in the area was welcome. The electrode setup was changed and the amount of triazene-DNA required per experiment was reduced dramatically. All results seemed to indicate that grafting was taking place, however, conclusive evidence was hard to achieve. Different techniques for verifying the grafting were utilized, but all had their weaknesses, and often reproducibility was an issue. A common weakness with all the techniques was the need for the grafted DNA to hybridize with a complementary strand before the coverage could be measured. When a problem arose in the final measuring step, it was hard to locate where the problem was. Addressing this problem, a slight alteration was made to the link between the triazene and the DNA: a ferrocene moiety was inserted. The point of this was to ensure that the presumably grafted DNA was indeed attached to the electrode, even in case its ability to hybridize had been compromised. PhD student Majken Hansen was given the task of synthesizing the new linker containing the ferrocene, and Elaheh Farjami, a visiting PhD student with experience in electro chemistry, performed the subsequent grafting and hybridization experiments. This addition of ferrocene to the link between the triazene and the DNA, worked beyond any hopes and expectations. Not only was the DNA successfully grafted to the surface, it also hybridized with complementary oligonucleotides. Moreover, the electrochemical redox potential of the ferrocene group changed detectably upon hybridization, making the system a fully functioning sensor device. This, finally, resulted in a publication in the Journal of Organic Chemistry in 2010 (see Appendix D).

2.1 Introduction iosensors currently attract a high amount of interest in the research community because they permit complex and time consuming laboratory analysis procedures to be replaced by simple testing kits administered by the general pub-

B

2.1.1. C OMMON G RAFTING M ETHODS

17

lic. These testing kits are often based on solid-state sensors that can be deployed directly without any prior preparations or purifications of the sample to be analyzed. An important quality for these sensors is high selectivity towards the analyte in question. D NA can provide such a selectivity. Lastly, the recognition element, i. e. the DNA, must be integrated with the transducing element in the sensor. 34 A biosensor is defined as a device which transduces biological events, e.g. conformational change of a peptide, the annealing of a set of complementary DNA strands etc., into a measurable electronic or optical signal. In many such systems, distances between active parties are of vital importance. For instance, in sensory systems based on F¨orster resonance energy transfer (FRET), the efficiency decreases with a factor of the distance between acceptor and donor entities to the sixth power. In electrochemically based biosensors, the read-out is registered as a change in the electrochemical potential measured via an electrode. In this case, it is the distance between the electro-active species, such as ferrocene or methylene blue, and the electrode that influences the sensitivity of the system. The signal is reduced exponentially with an increase of the distance, so immobilization of the electroactive species to the surface, and thus elimination of diffusion, ensures a high signal. In addition, many sensory systems have a design based on conformational changes, which in turn switch the read-out signal on and off by altering the distance of electron or energy transfer. An example of a such system can be seen in Fig. 2.1 where the distance between the electrode and the ferrocene tag is maintained low by a stem-loop DNA strand. In the presence of a complementary oligonucleotide, the rigidity of the hybridized structure will force the ferrocene tag away from the electrode surface, thus reducing the chance for electron transfer from the ferrocene and thereby suppressing the readout signal. 35 An alternative setup can be seen in Fig. 2.2 where the electron transfer is dependent on conductance through the hybridized oligonucleotide rather than the distance between surface and methylene blue. Using this setup, Boon et al. successfully managed to differentiate between hybridization with a complementary oligonucleotide (15 bases) and an oligonucleotide that contained a single mismatching base. 37 Amplification of the signal is attained by reoxidizing the methylene blue moiety continuously with ferric cyanide in solution.

2.1.1 Common Grafting Methods hile there exists a variety of mechanical methods for applying a substrate to an electrode surface (spin coating, deposition of Langmuir-Blodgett films, sublimation etc.), these will not be discussed here as it is beyond the scope of this

W

18

C HAPTER 2. I MMOBILIZATION

a)

OF DNA VIA TRIAZENES

b)

Figure 2.1. a) A stem-loop based DNA sensory system. The immobilized stem-loop DNA strand holds the ferrocene tag (the red dot) close to the electrode surface. When complementary DNA is added, the stem-loop is opened and the ferrocene is removed from the surface. b) Background-subtracted voltammograms (anodic scan) for the sensor in the presence of complementary DNA at 0 M, 30 pM, 500 pM, 30 nM, 800 nM, and 5 µM (from top to bottom). The hybridization time was fixed at 30 min. 36

report. One class of surface modifications is the self-assembling monolayers (SAM). When modifying electrodes using SAM, several possibilities exist. Among others are silane chemistry on glass and oxidized silicon surfaces, 39 photoactivated reaction of alkenes and aldehydes with hydride-terminated silicon surfaces 40 and of alkanes with hydride-terminated diamond surfaces. 41 The dominating methods, though, are based on alkylthiols and disulfides on gold surfaces. 34 The use of disulfides was first reported by Nuzzo et al. approximately a quarter of a century ago, 42 and the SAM obtained with thiols was characterized by Porter et al. a few years later. 43 The gold/sulfur bond is somewhat weak (approximately 44 kcal/mol 44 ) when compared to typical intramolecular bonding energies. A carbon-carbon bond, for instance, has about double the bonding energy (90 kcal/mol). 45 In fact, for the formation of a strong bond between a metal surface and an organic molecule without the use of heteroatoms, only two methods are available. One is the electrochemical reduction of vinylic compounds which leads to the formation of a thin polymeric medium covalently bound to the metal surface. The second method is the use of diazonium salts. 46

2.1.2 Diazonium Salts

I

n 1992, Jean-Michel Saveant and coworkers devised a novel method for covalent modification of carbon surfaces by grafting functionalized aryl radicals

2.1.2. D IAZONIUM S ALTS

19

Figure 2.2. On the left, electrochemical reduction of the methylene blue is not possible due to a base mismatch in the hybridization blocking the electron transfer. On the right, no mismatch is present, and the methylene blue is reduced by electron transfer from the electrode. The generated leucomethylene blue is continuously reoxidized by ferric cyanide in the solution. 38

produced from electrochemical reduction of diazonium salts (See Scheme 2.2). More specifically, the group grafted a layer of p-nitrophenyl groups onto a glassy carbon (GC) substrate by electrochemical reduction of p-nitrophenyldiazonium tetrafluoro-borate to the radical, which in turn spontaneously attached to the surface. 47 Five years later, the group reported an expansion of the scope of this technique with the grafting of a number of other aryldiazonium salts with different substituents administering similar grafting condition. 48

N

N + e-

- N2 GGGGGGGGGA

GGGGGGA R

R

R

Scheme 2.2. The grafting of an aryl group to a surface via reduction of the corresponding diazonium salt.

This use of diazonium salts for immobilization of organic molecules proved versatile with respect to the electrode material used, and this expanded the possibilities of covalent bonding to electrodes. Since its discovery, this method for covalent attachment of organic molecules to electrode surfaces has been widely employed. As mentioned above, glassy carbon was the initial electrode material, but since then grafting has been performed on highly ordered pyrolytic graphite (HOPG), 48,49 pyrolyzed Teflon, 50 carbon fibers 51 and nanotubes, 52 on diamond, 53 silicon 54,55 and GaAs, 55 and on metals (iron, 56 cobalt, 57 nickel, 57 copper, 57 zinc, 57 platinum, 57 gold 57 and palladium 55 ). This multitude of possi-

20

C HAPTER 2. I MMOBILIZATION

OF DNA VIA TRIAZENES

ble electrode materials combined with the relatively high ease of both synthesis and application make aryl diazonium salts well suited for many purposes. It was Peter Griess, a German chemist, who in the mid-nineteenth century, discovered that the treatment of amines with nitrous acid resulted in the amine exchanging three hydrogen atoms for one nitrogen. Also, he found that the product showed remarkable physical and chemical properties - especially their tendency to explode violently when heated. 58 Griess called his newfound class of compounds diazo compounds, owing to the two nitrogen atoms in the compounds (azote being the French name for nitrogen); today we know them as diazonium salts. The method of manufacture devised by Griess included the passing of a current of nitrous acid gas into an alcoholic solution of an amine. Not much has been changed since then. The traditional method for the preparation of diazonium compounds consists of treating an aniline with strong acid and sodium nitrite. First, the nitrosonium ions form, as depicted in Scheme 2.3 A, and this ion reacts with the aniline to form the diazonium compound (see Scheme 2.3 B). Generally, the preferred counter ion for diazonium salts is tetrafluoroborate when the diazonium salts are not used immediately (and especially if not kept in solution), since this eliminates the risk of explosion. 46 A) B)

HO

N

O

H+ H H N

NH2 N O -H+

N O

H O

N

H+

O

-H2O N O

N

N OH2

N

N

Scheme 2.3. A) Formation of nitrosonium ion from nitrite in strong acid. B) Diazotation of aniline

These harsh conditions obviously can present a problem in cases where the organic compound to be diazotated is acid labile. As an alternative, a Lewis acid can be used in place of the Brønsted acid in combination with tert-butyl nitrite in organic solvents. While these are less harsh conditions, the mildest conditions are obtained with the tetrafluoroborate of the nitrosonium ion. Since the harsh conditions are only used for the production of the nitrosonium ion, no Brønsted or Lewis acid is needed when the premade nitrosonium tetrafluoroborate is added. 59 Nitrosonium tetrafluoroborate is still a very reactive compound, but relatively easy to work with. Concurrently with our research, Harper et al. published an article where a GC electrode was directly modified with phenylmaleimide diazonium conjugated

2.1.3. T RIAZENES

21

to ferrocene labeled ss-DNA. The fabrication of the DNA-conjugated diazonium salt was conducted by ligation of N-4-(diazophenyl)maleimide tetrafluoroborate to sulfide modified DNA (See Scheme 2.4). Subsequently, the DNA was grafted to the GC electrode through electrochemical reduction of the diazonium group. 60 Fe Fe

O

N

O

O

N

Fe

O SH

NOBF4

S O

NH2

N+ N

N

O

+e-N2

S O

N

O

N+ N

Scheme 2.4. The approach of Harper et al. for immobilizing DNA on a GC electrode surface through diazonium chemistry. In the first step the diazonium salt is produced and subsequently, the thio-modified DNA strand is attached through a sulfide maleimide coupling. Finally, the DNA-diazonium is immobilized onto the electrode surface by the standard diazonium grafting procedure. The DNA has a ferrocene group attached for quantifying the amount of grafted DNA.

Prior to this, diazonium salts had already been used in the immobilization of DNA to electrode surfaces. Lee et al. reported the immobilization of DNA on single-walled carbon nanotubes (SWNT) through covalent bonding. They achieved this by initial spontaneous grafting of 4-nitrophenyl diazonium to the SWNT followed by reduction of the nitro group. The newly formed layer of amino functionalities on the SWNT was then in turn used to form amide linkages to terminally modified oligonucleotides, using standard succinimide chemistry (See Scheme 2.5). 61

2.1.3 Triazenes hile Harper et al. demonstrated successful immobilization of DNA via the diazonium salt, this approach still has some disadvantages. The major problem lies with the stability of the diazonium salts. While they maintain a halflife of approximately 5 days in acidic or aprotic solvents, they are not stable in aqueous solution above pH ≈ 2-3. 46 The reaction of diazonium salts with nucleobases may present another problem, since such reactions have been reported in the literature,. 62 This adds another aspect of instability. The approach of Lee

W

22

C HAPTER 2. I MMOBILIZATION

OF DNA VIA TRIAZENES

S 5'-DNA-3'

O

S 5'-DNA-3'

O N

N

O

O

O

N

O

NH2 O

O N

O

HN

O

O

O2N N2

e-

SO3

SDS

Scheme 2.5. Covalent immobilization of DNA onto SWNT by initial functionalization with nitrophenyl groups, which are reduced to aminophenyl and linked via succinimide chemistry.

et al. 61 avoids these problems by modification of the surface prior to immobilization of the DNA. This method, however, is likely to leave unreacted aminomodifications on the surface, which may not be desired, depending on application. The use of triazenes (not to be confused with triazines) bypasses a large portion of these issues. Triazenes are a class of compounds first investigated by Wallach back in the late nineteenth century. 63,64 They follow the general formula R1 N=N—NR2 R3 , where the nitrogen are numbered from left to right as N1, N2 and N3, respectively. 65 That is, N1 is bound to R1 and doubly bound to N2. In turn, N2 is bonded to N3, which is linked by single bonds to R2 and R3 , respectively. Typically, triazenes are synthesized from reaction of an amine (N3) with a diazonium compound (N1 and N2) as depicted in Scheme 2.6 A. Other methods, however, have been demonstrated, for instance, by Trost and Pearson back in the early 1980’s. In their work, they demonstrate the use of a triazene as an intermediate in the conversion of an alkyl halide into an amide or a primary amine. The halide is initially converted to the corresponding Grignard reagent, then reacted with an azide and quenched with an acid halide (as seen in Scheme 2.6 B) or a proton. Subsequent treatment of the formed triazene with hydroxide released the amine or amide. 66,67 The use of triazenes in grafting rely on their in situ conversion back to the diazonium salt during grafting. This principle was demonstrated by Tour and coworkers in grafting of molecules onto silicon 68,69 and later by immobilization of small molecules onto SWNT. 70 In both cases, the conversion was performed by lowering the pH to approximately 2. During formation of the triazene as described in Scheme 2.6 A, the two equilibria are drawn to the right by the high pH and an excess of the amine. By lowering the pH, the second equilibrium is pushed towards the protonated triazene, which in turn is cleaved into the amine

2.1.3. T RIAZENES

A)

R1

N

23

N

+

R2

H N

R2 H N N N R3 R1

R3

RMgBr

B)

PhS

N3

PhS

N

N

N

R

-H+ R1

N

R'COX PhS

N

N N

R2 N

R3

R N COR'

Scheme 2.6. A) Formation of a triazene from a diazonium compound and an amine. B) Alternative synthetic strategy as presented by Trost and Pearson. The method was used as a means to synthesize an amide containing R and R’. This would be released when treating the triazene with hydroxide. 66,67

and diazonium salt when the amine is no longer in excess. In preliminary studies of triazenes in grafting, a ferrocene labeled aryl triazene 1 had been synthesized and grafted to the surface of a GC electrode.i Instead of lowering pH, the triazene was converted back into the diazonium salt by addition of dimethyl sulfate. Methylation of N3 basically works in the same way as for the protonation, except that it is irreversible. Following analysis of the generated layer by cyclic voltammetry (CV) suggested that in situ formation of the diazonium salt from a triazene seemed a viable path. O O N

N

Fe N

1

Following up on these results, Chen synthesized a range of oligonucleotides with terminal triazene modifications in order to assess stability and effectiveness. 72 Ultimately, she arrived at 3, where the aryl triazene moiety was conjugated to amino-modified DNA. The conjugation of the succinimidyl carbonate 2 to DNA can be seen in Scheme 2.7. An alternative to Chen’s reaction conditions was primarily utilized because it provided a simpler purification procedure while maintaining a similar yield. It was observed that reaction yields dropped after prolonged storage of 2, due to degradation of the triazene. During a time span of one i

Synthesis performed by M.Sc. student at the time Cindy Knudsen, 71 Aarhus University, grafting by Ass. Prof. Steen Uttrup Pedersen, Aarhus University

24

C HAPTER 2. I MMOBILIZATION

OF DNA VIA TRIAZENES

year, the yield was reduced from 30% to approximately 7% for reactions with the same batch of 2, and Chen reported yields as high as 54% from the fresh batch. These times pertain to 2 being stored in DMF solution (0.2 mg/µL) at -18 °C. An increased lifetime might be achieved by storing neat, rather than as a frozen DMF solution. O O N

N

O O

N

O

N

H2N

DNA

6

Et3 N GGGGGGGGGGGGGGGGGA H2 O/DMF (1:1)

H N

O N

N

2

O

N

DNA 6

3

Scheme 2.7. Conjugation of aryltriazene moiety to DNA.

2.2 Results nitial grafting attempts were performed using a custom-made micro cell setup comprising of a hard polymer tube fitted with work and auxiliary electrodes in each end and a reference electrode through a hole in the middle (See Figure 2.3)ii . This micro cell not only allowed for the use of small volumes but also rendered the volume adjustable by altering the positions of the work and auxiliary electrodes in the tube. In this fashion, only a smaller quantity of 3 was necessary, and concentration requirements were not compromised at the expense of volume. Grafting in this setup was only attempted on glassy carbon.

I

Figure 2.3. Micro cell setup

In order to evaluate the success of the grafting experiment, a CV of K3 Fe(CN)6 in solution was recorded both prior to and following the grafting procedure. As ii

Micro cell designed and supplied by Ass. Prof. Steen Uttrup Pedersen, Aarhus University

2.1.3. T RIAZENES

25

Figure 2.4. Cyclic voltammogram of K3 Fe(CN)6 in solution before and after grafting of 3 to the electrode surface. The set of redox waves from reduction and subsequent oxidation of ferric cyanide are evident prior to grafting. After grafting, the redox waves are no longer visible due to blocking of the electrode surface.

can be seen from the recorded CVs, both the oxidation- and reduction waves have disappeared post grafting (See Figure 2.4). This indicates isolation of the electrode from the solution which in turn suggests that the electrode has been covered during the grafting procedure. By means of an AFM scratching experiment, the thickness of the grafted layer was estimated by Chen (See Figure 2.5). She found the thickness of the layer to be consistent with that expected of a 15 nucleotide oligomer conjugated to the surface through the 4-(2-(hexylaminocarbonyloxy)-ethyl)-phenyl linkage resulting from grafting via the triazene 3. 72 Hybridization of the grafted layer of oligonucleotides with a ferrocene labeled complementary strand was attempted without success. This was attributed to unspecific binding between the grafted strands and the electrode surface, causing sterical issues and lowering the possible energy gain in hybridization with external strands. Studies by Pedano et al. corroborate our observations in this respect. 73 Grafting was then attempted on gold electrodes using a different setup that allowed for smaller volumes of the DNA solution to be used compared to the micro cell setup. In this setup, the working electrode was fixed vertically and the solution was placed as a droplet on top of it (See experimental section for details). Following grafting, the gold electrode was treated with 6-mercaptohexan-1-ol in order to minimize interaction between the grafted DNA and the electrode surface. The mercaptohexanol forms a highly ordered SAM on the gold blocking the surface and thereby reducing interaction between DNA and the electrode surface. This time, hybridization with the ferrocene labeled, complementary oligonucleotide appeared successful. Analyzing with differential pulse voltammetry (DPV), the

26

C HAPTER 2. I MMOBILIZATION

OF DNA VIA TRIAZENES

Figure 2.5. a) 2.5 × 1 µm AFM image after 1 × 1 µm scratch measurements of GC electrode grafted with 3, b) Mean cross section (averaged over 10 line scans) of the top view presented in (a). 72

a)

b)

c)

Fe

-e- +eFc Fe

e-

Figure 2.6. a) Illustration of grafted oligonucleotide hybridized with ferrocene labeled (Fc) complementary strand. On left, DPV of grafted oligonucleotide hybridized with ferrocene labeled complementary DNA. b) is the oxidation of ferrocene to ferrocenium, c) is the subsequent reduction back to ferrocene.

oxidative and reductive potential waves for ferrocene were observed (See Figure 2.6). In an attempt to get more quantitative data on the degree of coverage, the hybridization strategy was changed to include 32 P-labeled complementary strands rather than ferrocene-labeled. A complementary DNA strand was added a terminal 32 P-phosphate, and the strand was brought into contact with with the immobilized DNA . In the case of successful hybridization, the radiation from the 32 P can give an indication to the amount of immobilized DNAon the surface. In order to take unspecific binding in to account, a blank electrode was exposed to identical hybridization conditions as a reference. Preliminary data were obtained by hybridization with the 32 P-labeled complementary strands on two grafted electrodes and one blank electrode to serve as a

2.1.3. T RIAZENES

27

reference for non-specific binding to the electrode surface. The data verified that hybridization on GC electrodes was not easily attained since the blank electrode (f) seemed to bind more labeled DNA than one of the grafted electrodes (e) and less than the other (d) (See Figure 2.7 columns (d), (e) and (f)). On the contrary, hybridization on gold electrodes gave somewhat promising results in the sense that electrode (a) in Figure 2.7 bound more than fivefold the amount of DNA compared to the blank electrode (c). While electrode (b) held a much smaller amount of the labeled DNA than did electrode (a), it was still roughly twice as much as the blank electrode (c). In order to expand the data set and to make the estimation of the absolute amounts of hybridized DNA possible, a second experiment was devised. Three electrodes grafted with DNA was added complementary, 32 P-labeled DNA in order to estimate the coverage. As references for non-specific DNA /DNA binding, two electrodes, identical to the first three, was added non-complementary, 32 P-labeled DNA Finally, ˙ three blank electrodes were exposed to the 32 P-labeled DNA along with the same hybridizing conditions to serve as a reference for unspecific DNA binding to the electrode surface. All the electrodes were treated with 6-mercaptohexan-1-ol. The radioactivity was also measured of a known amount of labeled DNA from the same batch as used with the electrodes. The ratio between the recorded number of counts and the amount of DNA could then serve as a conversion factor for estimating the surface coverage. As can be seen in Figure 2.8, the grafted electrodes did not display any significant preference to binding to complementary oligonucleotides rather than the non-complementary sequence. Moreover, the blank electrodes (6), (7) and (8) all exhibited a higher affinity to the labeled DNA than did the grafted electrodes. It remains unknown why the results were inconclusive, however, observations made in other experiments performed in the group give strong evidence that hybridization may be the major issue. In an unrelated matter, Sarah W. Helmig, a student in our group, performed multiple hybridizations with 32 P-labeled oligonucleotides to immobilized DNA on gold surfaces. The immobilizations were performed using classic thiol on gold-immobilization techniques, and while successful hybridizations were at times achieved on the surfaces, they were inconsistent and never more than partial. Spots with a high surface coverage of the hybridized DNA was observed, whereas large surrounding areas for unknown reasons showed little to no coverage. Harper et al., who also reported on the use of diazonium chemistry for grafting of DNA onto electrodes (vide supra), also reported that DNA covered surfaces

28

C HAPTER 2. I MMOBILIZATION

OF DNA VIA TRIAZENES

Figure 2.7. Preliminary hybridization experiment with 32 P-labeled complementary DNA. (a) and (b) Grafted gold electrodes + complementary DNA, (c) Blank electrode + labeled DNA , (d) and (e) Grafted GC electrodes + complementary DNA , (f) Blank GC electrode + labeled DNA.

Figure 2.8. Second hybridization experiment with 32 P-labeled complementary DNA. (1), (3) and (5) are grafted gold electrodes + complementary DNA, (2) and (4) are grafted gold electrodes with non-complementary DNA, (6), (7) and (8) are blank gold electrodes with labeled DNA. The scale is normalized to pmol DNA.

2.1.3. T RIAZENES

29

achieved with this approach did not prove effective for hybridization with complementary DNA, nor for aptamers binding to proteins. Their conclusion was that the diazonium groups were reacting with the nucleobases on other DNA strands, rendering them ineffective for the above purposes. 60 As a parallel experiment in the attempt of overcoming the above problems, a slightly different approach was taken. Majken Hansen, a PhD student in our group at the time, synthesized 4, a variation of the DNA conjugated triazene linker containing a ferrocene in the linker portion of the molecule. By including the ferrocene as an electrochemical probe in the grafted DNA containing molecule, hybridization would no longer be necessary in order to test for successful grafting. H N N

N

N

O

Fe

O N H

6

DNA

4

The new DNA/triazene conjugate, 4, was designed with the same triazene moiety as the previous triazene, 3, making all previous grafting protocols reusable. Elaheh Farjami succeeded in grafting 4 onto both GC and Au electrodes using the established protocols. Additionally, she obtained CV’s of the grafted GC electrodes showing a redox peak at a mean potential of +231 ± 5 mV, corresponding well with that of the incorporated ferrocene reporter group. The recorded CV’s reflected a linear correlation of the peak currents to potential scan rate indicating independence of diffusion rates of the electroactive specie 74 - in other words, the ferrocene was not diffusing, it was statically immobilized on the electrode surface. Integration of the ferrocene redox peaks yielded an approximate surface density of ferrocene of 29 ± 4 pmol cm-2 . Comparing this value to those of DNA selfassembled monolayers using gold-thiol chemistry provided by Petrovykh et al., the values are of the same order of magnitude. The surface density achieved by the triazene is roughly comparable to that of a gold surface treated with thiol modified oligonucleotides for 10 minutes. Treating a gold surface with thiols for two hours will approximately double the coverage. 75 When she attempted to graft 4 to Au electrodes, the result was equally good. The recorded mean redox potential was +218 ± 4 mV, and the DNA monolayer coverage obtained from peak area integration was 31 ± 9 pmol cm-2 . The electron transfer rate constant, ks , was determined by the cathodic and anodic peak

30

C HAPTER 2. I MMOBILIZATION

OF DNA VIA TRIAZENES

Figure 2.9. (a) Schematic illustration of the hybridization and denaturing of the immobilized DNA. (b) Cyclic voltammogram of the ferrocene labeled DNA-modified GC electrode in PBS, pH 7, before (ss1) and after (ds1) hybridization, and after denaturation at 50 °C in water (ss2). Scan rate is 100 mV s-1 and CVs recorded after 5th scan are shown c) Normalized and background subtracted anodic scan currents derived from the corresponding CVs. 77

separation to be almost 1 s-1 . It was found to be higher than for the GC electrode, which was 0.28 ± 0.05 s-1 , and this is in accordance with the literature. 76 While this served as proof that 4 grafted successfully onto both GC and Au electrode surfaces, the grafted DNA’s capability to hybridize was still untested. When hybridization was performed, it was not only successful, the system proved to intrinsically supply a sensory function to the complementary DNA. As can be seen in Fig. 2.9, the ferrocene peak potentials shift when the attached DNA is hybridized with a complementary strand (ds1) as opposed to when it is single stranded (ss1 and ss2). In the case of the modified GC electrode, the size of the shift was -15 mV, and the transfer rate, ks was found to have increased slightly to 0.34 ± 0.04 s-1 . The registered amount of ferrocene on the surface remained unaltered post hybridization. The observed shift in the case of the gold electrodes, was consistent with the

2.1.3. T RIAZENES

31

observations from the GC electrodes at -14 mV. In this case the electron transfer rates, however, remained largely unchanged upon hybridization. As opposed to the data collected from the GC electrodes, the peak widths on the gold electrodes broadened slightly upon hybridization. This was attributed to a minor contribution from unhybridized, immobilized DNA. As the electrodes were heated to 50 °C in order to denature the immobilized DNA , the major advantage of the ability to immobilize on GC electrodes became apparent. While the ferrocene peaks on both electrodes reverted to the potential consistent with unhybridized surface DNA, the heating of the gold electrode caused partial breakage of the gold-carbon bond attaching the DNA to the surface. In the course of repeated cycles of hybridization and thermal denaturation of the DNA on the gold electrodes, the signal from the ferrocene redox label constantly decreased. The GC electrodes, however, showed no such degradation owing to the higher strength of the carbon-carbon bond (346 kJ mol-1 ). 76

2.3 Diazocoupling from Triazenes he use of triazenes in DNA-directed synthesis seemed a logical next step from the previous work done with this functional group. In synthesis work, the facilitation of triazenes fall into three categories: use as a protection group, as a tether and as a reactive group in itself. Primarily, triazenes have been used to a wide extent as masking groups in place of different functional groups (See Table 2.1 for conditions where triazenes are stable and labile, respectively). Gross et al. demonstrated this application in their work back in 1993. 78 Their focus was mainly on protection of aromatic amines but in addition they illustrated the possibility of exchanging the triazene with halogens, alkanes, cyanide or simply hydrogen using Sandmeyer chemistry. 79,80 While Gross only considered N1 in the triazene as the protected amino group, Lazny et al. demonstrated the protection of non-aromatic amines using triazenes merely by considering N3 the protected amino functionality. 81,82 Br¨ase and coworkers took the same approach to triazenes when they adapted these for use as tethers to solid supports. 83 In addition to producing amines upon cleavage, they demonstrated the use of triazene tethered solid-phase synthesis of guanidines, amides, ureas and thioureas. 84–86

T

32

C HAPTER 2. I MMOBILIZATION

OF DNA VIA TRIAZENES

Table 2.1. Reactivity of Triazenes 87

Triazenes Stability

Triazene Lability

Aqueous NaOH Methoxide Pd cross coupling BuLi Grignard Reagents Catechol borane

MeI (110 °C) Acids (e.g. Acetic Acid or TFA) Lewis Acids (e.e. AlCl3 , BCl3 , TfOH) Br2 Me2 SiI Strong Alkylating Agents

The use of triazenes as a reactive group in synthetic chemistry has been demonstrated by Wirschun et al. in the manufacture of heterocycles from triazenes. The heterocycles form through a [3+2] cycloaddition of an intermediate 1,3-diaza2-azoniallene ion and an alkene 88 (See Scheme 2.8), 1,3-butadiene, 89 alkynes, carbodiimides or cyanamide 90 yielding 1,2,3-triazolium and tetrazolium salts.

Ar

N

N

H N

Cl t

R

BuOCl -tBuOH

Ar

N

N

Cl N

Ar Lewis Acid N N N 60°C, CH Cl 2 2 R R

Ar N N N R

Scheme 2.8. The formation of heterocycles from triazenes as it was presented by Wirschun et al. 88

Many of the above features of triazenes may transfer well to use in DNAdirected synthesis. Until now, however, we have only investigated the use of triazenes as masked diazonium salts in reaction. In this manner, it is simply a question of reversing the formation of the triazene. As in the case of electrochemical immobilization (vide infra), methylation of N3 with dimethyl sulfate was taken into use to achieve this. The chemistry of diazonium salts has already been touched upon briefly at the mention of Sandmeyer’s transformations of these into halides and pseudohalides. In addition, diazonium salts also find a place in metal-mediated cross-linking. The Meerwein arylation is an early example of this, where an aryl diazonium salt is N

N

EWG

EWG

Cu(I)

Scheme 2.9. The Meerwein reaction.

2.3.1. R ESULTS

33

coupled to an electron-poor alkene, much to the same effect as in a Heck coupling (See Scheme 2.9). 91 While the mechanism is still not fully understood, a radical pathway is assumed. 92 Palladium-mediated coupling reactions are also possible with diazonium salts liberating nitrogen gas during the oxidative addition of palladium. 93 Kikukawa and Matsuda were the first to report this type of reaction back in 1977. In an attempt to find an alternative to the Meerwein arylation, they managed to arylate several olefins including some with electron-rich substituents in a Heck-like couplings using palladium(0) in MeCN/water. 94 Since then, palladium coupling reactions with diazonium salts have been reported for both Suzuki-Miyaura type, carbonylative, Stille type and carbon-heteroatom couplings amongst others (See Scheme 2.10). 93 R'

R' R

R'B(OH)2 R [Pd]

R' [Pd]

N

O

CO, [Pd] NuH Nu

N R'

R'4Sn

R

R [Pd]

Heterocoupling X-R'

R R

Scheme 2.10. An overview of some of the palladium-mediated cross-couplings with diazonium salts demonstrated in the literature 93

This rich versatility of the triazene functional group makes it useful in many aspects. Since the formation of a diazonium salt from a triazene had already been investigated in our previous work, this was chosen as a starting point for DNA directed reactions as well. In particular, diazo dyes were the target since the formation of these would be fairly simple and yield a highly visible product.

2.3.1 Results During preliminary studies performed by Menglin Chen, 72 the reaction between an aryltriazene 5 and 2-naphthol producing the diazo-compound 6 (see Scheme 2.11) was investigated. This was done in order to estimate whether hydrolysis of the added dimethyl sulfate would outmatch the desired reaction with the triazene in buffer solution. Due to solubility issues, the experiment was performed

34

C HAPTER 2. I MMOBILIZATION

OF DNA VIA TRIAZENES

in 1:1 mixture of acetonitrile and phosphate buffer solution (0.5M, pH 7). Chen monitored the formation of 6 by measuring the molecule’s absorbance at 480nm over time. The studies led to the conclusion that for concentrations of 5 as low as 10 µM with 100-fold excess of both napthol and dimethyl sulfate, the formation of 6 was observed (See Figure 2.10). Subsequently, the triazene conjugated to an oligonucleotide was tested under the same conditions and was found to react analogously. OH N

N

N

5

Me2 SO4 GGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGA Phosphate buffer/MeCN (1:1) pH 7.0

OH

N

N

OH

OH

N N

GGGGGGGGGGGGGGGGGGGA OH

6

Scheme 2.11. Triazene reacting with 2-naphthol.

For the testing of triazenes’ possible utilization in DNA-directed chemistry, it was thus an obvious choice to consider the diazo coupling to 2-naphthol as a viable reaction. The DNA-directed chemical reaction between the naphthol and the aryltriazene was tested on two different DNA systems, namely the End-templated system and the Externally templated system as depicted in Scheme 2.12. The former was performed as the conceptual proof that triazenes can be brought to react in extensive yields under ambient conditions through the DNA-controlled pseudo-intramolecular reaction. Conversely, the Externally templated system was chosen in the hope of demonstrating a possible application. Since the formation of the diazo-compound is easily monitored via UV/vis photospectrometry, the Externally templated diazo coupling between 2-naphthol and an aryltriazene seemed a good candidate for a colorimetric detection system. While both the DNA -conjugated naphthol and aryltriazene are always present in such a system, they will only react if the template strand is added. If so, the oligomers conjugated to the aryltriazene and the naphthol, respectively, will hybridize with the template strand. In the presence of dimethyl sulfate, this will enable the pseudo-

2.3.1. R ESULTS

35

Figure 2.10. Absorbance followed at 480 nm for reaction of 5 (10 µM), 2-naphthol (1 mM) and Me2 SO4 (1 mM)

intramolecular reaction between the conjugated species to take place, yielding a colorimetric readout. In both systems, the same triazene 3 as in the immobilization project was used. This was based on the 5’-C6-amino modified oligonucleotide A, so the Endtemplated system logically entailed the application of a 3’-C6-amino-modified, complementary oligonucleotide A’. For the Externally templated strand system, a set of oligonucleotide strands (Table 2.2) were devised with the intent of enabling separation of reactants and product via PAGE analysis of the reaction. A template strand, B, with a length of 28 nucleotides was chosen with a sequence complementary to the first 14 bases of A (counting from the 5’ end). An oligonucleotide, C, of 19 nucleotides in length was constructed to complement the remaining 14 bases in the template sequence, leaving an overhang of 5 nucleotides. Additionally, C was 3’-modified, placing the triazene and naphthol moieties in close proximity upon hybridization.

36

C HAPTER 2. I MMOBILIZATION

OF DNA VIA TRIAZENES

a) OH

O

O

O

HN

H N

O

HN N N

H N

N

Me2SO4

N

O

H N

O

N

N

O O

H N

N

O O

b) N N

OH

N HN

O O NH

Scheme 2.12. Types of DNA-directed reactions. a) The End-templated system, and b) the Externally templated system.

Table 2.2. Oligo Nucleotide Sequences

Oligo

Sequence (5’-3’)

Modifications

A A’ B C

GGC CTC GCA CTG GCG CGC CAG TGC GAG GCC GCC AGT GCG AGG CCA CCT GTA GAA CCG C CAC AGG CGG TTC TAC AGG T

5’-(CH2 )6 NH2 3’-(CH2 )6 NH2 3’-(CH2 )6 NH2

Since conjugation of a 2-naphthol moiety to DNA is essential to the setup, the following synthetic strategy was devised. To facilitate linking, 6-hydroxy-1naphthoic acid was used in place of 2-naphthol, and the oligomers were bought with a terminal C-6 amino modification. These were then conjugated via the succinimide 7 and subsequent amide bond formation with the amine terminated oligomer yielding the desired product 8. An alternative approach, utilizing the coupling reagent 1,1,3,3-Tetramethyluronium Tetrafluoroborate (TSTU) as reported by Knorr et al., 95 to produce the succinimide 7 was also attempted but abandoned due to inferior yields. Both in the conjugation of A’ and C to 7, yielding 8a and 8b , respectively, the products were purified on RP-HPLC and identified via MALDI-TOF-MS analysis.

2.3.1. R ESULTS

O

37

OH O OH

6-hydroxy-1naphthoic acid

O

N O

O

H2N

NHS, EDC GGGGGGGGGGGGGGGGGGGA THF/CH2 Cl2 (1:1)

OH

DNA

6

Et3 N GGGGGGGGGGGGGGGGGGGGGGGGGGA H2 O/DMF/MeCN (1:1:1)

7

HO

O N H

DNA

6

8a,b

Scheme 2.13. Conjugation of naphthol moiety to DNA. Ligation of 7 to oligonucleotide A’ yields 8a , and ligation to oligonucletide C results in 8b .

Yields were determined to 16% and 40%, respectively. For the testing of the End-templated system, the following experiment was devised. Equimolar amounts of the DNA-conjugated reactants were dissolved and combined in buffer giving a reactant concentration of 10 µM. After 30 minutes, a tenfold excess of Me2 SO4 was added, and the mixture was left to react for an additional 30 minutes. Analysis of the reaction through PAGE on a denaturizing gel (see Figure 2.11) registered the formation of a product band with a lower retention factor on the gel than the two reactants, which compose the other band. Additionally, in lanes (b) and (d) on the gel, the reactants, 3 and 8, respectively, were treated with Me2 SO4 for 30 minutes in order to elucidate on the possibility of the reactants reacting internally to produce the aforementioned product. Evidently, the product only forms in the presence of both reactants and no visible reaction occurs when the reactants are separated. The template system was examined in an analogous fashion where the template strand, B, was added at the time of the initial mixing of the reactants. An identical experiment, except without the addition of the template strand, was executed in parallel. This was done to demonstrate the need for a template strand in the system for the reaction to occur. As seen in Figure 2.12, an easily distinguishable product band is materializing in lane (d), which contains the reaction including the template strand. Aside from the product, also bands from the tem-

38

C HAPTER 2. I MMOBILIZATION

Lane

3

8a

Me2 SO4

(a) (b) (c) (d) (e)

+ + + -

+ + +

+ + + -

OF DNA VIA TRIAZENES

Figure 2.11. PAGE analysis of the End-templated system experiment

Lane

3

8b

(a) (b) (c) (d) (e)

+ + + -

+ + + -

a

Oligo Ba Me2 SO4 + -

+ + +

Template strand

Figure 2.12. PAGE analysis of the Externally templated system experiment

2.3.1. R ESULTS

39

plate strand and unreacted 8b are present. This suggests that 8b was in excess and that the reaction proceeded to completion since no 3 remains. In lane (c), which contains the reaction devoid of template strand, no product is observed and both reactants are still present. While this seemed a good start, reproducibility proved an issue. Some times the results would mirror those of depicted above other times, no product band formed. Attempts were made at isolating the product, but to no avail. When at one point the denaturing conditions of the polyacrylamide gels were improved by addition of formamide to the gel, the product band stopped appearing altogether. This could be an indication that the formerly used urea gels simply did not succeed in denaturing the samples completely and, in fact, the desired product was never formed. It could also be coincidental that the setup ceased working entirely at the same time as the gel composition was changed. It was decided to simplify the experiments in order to eliminate possible causes for the above inconsistencies. This was done by removing the element of DNA control and instead let the DNA -conjugated triazene react with an excess of different nucleophiles. This way the reactivity of the triazene was still being tested in the presence of DNA and under hybridization-friendly conditions. For the product to absorb visible light, the highly conjugated nucleophiles in Fig. 2.13 were chosen. a)

b)

c)

d) O HO

OH

OH OH

OH

OH OH

HO

OH OH O

Figure 2.13. Nucleophiles chosen for reaction with DNA: a) Phenol, b) 2-naphthol, c) 1,1’-bi-2-naphthol, and d) 1,2,3,5,6,7-hexahydroxyanthraquinone (HHAQ).

Nucleophiles of different sizes were chosen in the hope of achieving differently colored products. All of them had hydroxyl substituents to both increase solubility and reactivity. The DNA-trizene, 3, was added the nucleophile in 4500 equivalents along with dimethylsulfide, and after reacting over night, the reaction mixture was purified by HPLC. All peaks in the HPLC spectrum were isolated and analyzed by MALDI-TOF-MS, but no masses consistent with either the product or starting DNA was ever located. Reaction seemed to have taken place at least in some of the cases. In the reaction with phenol, the reaction mixture had taken on a weak yellow color, and with 2-naphthol, a very strong red color was visible even

40

C HAPTER 2. I MMOBILIZATION

OF DNA VIA TRIAZENES

in subnanomole amounts. Some samples were sent to DNA Technology A/S for analysis by QC-MS, however, they were not able to find any explainable masses either. Only in two cases did they find anything in the samples, and in these cases the found masses were much too low to have any significance.

2.4 Conclusion and Perspectives novel technique for electrochemical immobilization of DNA onto electrode surfaces by the application of a special triazene linker was devised. In initial attempts at immobilization, the immobilized DNA was troublesome to quantify on the electrode surface through indirect quantification by hybridization with labeled complementary strands. This led to the alteration of the linker to contain a ferrocene moiety that made direct detection possible. The redox potential of the ferrocene proved to be sensitive to whether the immobilised was single or double stranded. This sensitivity made the ferrocene linker directly applicable in the production of sensors for detection of antisense DNA. The triazene moiety was additionally investigated in a context of both DNA directed and non-directed synthesis. The results were indicative of a possible use in this respect but they were not conclusive. These studies were interrupted by my visit to New York University and upon my return, the development of the DNA actuator dimer was prioritized. Triazenes, however, could as a functional group find a very divers use in DNA chemistry, as it is itself very divers in reactivity and it is compatible with the buffered aqueous conditions needed for DNA. A possible use for its ability to readily produce carbon-carbon bonds was the DNA modification of sheets of graphene. Both improving solubility and adding a possible control element this could be an interesting application that seems likely to succeed. In our group, however, it was never attempted due to problems producing the needed graphene molecules.

A

C HAPTER 3

T RANSFER OF H ELIX C HIRALITY TO F LUORESCEIN

his chapter concerns the work I did while I was visiting the laboratories of Professors Seeman and Canary at New York University. The project aim was twofold. The primary objective was to attach two fluorophores to a double stranded DNA oligonucleotide and show that they upon excitation would create a coupled exciton, i.e. one excited state including both fluorophores, and that this exciton would adapt chirality from the DNA double helix. The secondary objective was applying this effect in finding a new way of detecting whether a DNA double helix was on the normal B-form or the Z-form, which is of left-handed chirality rather than right-handed.

T

3.1 Introduction ack in 1953, Watson and Crick 1 suggested that the X-ray diffraction patterns recorded for crystalline DNA fibers in the B-form, could be explained entirely by assuming that two different DNA strands associated with each other to form a right handed double helical structure. In their model the two strands coil around a center axis held together solely by the selective association of the bases with one another - adenine with thymine and guanine with cytosine - ever since known as the Watson-Crick base pairs. Since then, this double helix has had an almost iconic status in our collective minds. While the B-DNA, which Watson and Crick described, is considered to be the predominant structure of ds-DNA in eukaryotic cells, 96 many other conformations

B

41

42

C HAPTER 3. T RANSFER

OF

H ELIX C HIRALITY

TO

F LUORESCEIN

have been observed over the years. The A-form was first observed along with the B-form in the sodium salt fibers of calf thymus DNA where Franklin and Gosling 97 found that they obtained different X-ray diffraction patterns depending on, among other factors, the relative humidity of the sample. They thus named the structures associated with the two different diffraction patterns A and B for 75% and 92% relative humidity, respectively. The C-form of DNA was discovered in 1961 by Marvin et al. 98 as the lithium salt of calf thymus DNA, and the list goes on. In fact, as of 2003 according to Ghosh and Bansal, the many different types of DNA have been so abundant that only the letters F, Q, U, V and Y remain unused in the naming of DNA structures in the literature. 99

3.1.1 B-DNA atson and Crick’s original model for the B-form of DNA suggests that it consist of two intertwined, right-handed helices, held together by the aforementioned Watson-Crick base pairs, twisting uniformly along the common helix ˚ per full helical turn. This strucaxis 36° per base pair - i.e. 10 base pairs and 34 A ture gives rise to a major and a minor groove along the helix, and a helix diameter ˚ 1 The number of base pairs per helical turn have since been adjusted to of 20 A. 10.4-10.6 for double helical DNA in solution. 100 In 1978 single crystal X-ray diffraction allowed for atomic resolution studies of deoxynucleotides. Regular B-DNA however, was not what the first single crystal diffraction pattern revealed when Viswamitra et al. studied the selfcomplementary oligonucleotide d(pApTpApT). 101 Rather they described a set of mini-helices, a structure that since has been named a T-tract. 102 A year later when Wang et al. studied a sample of d(CpGpCpGpCpGp), B-DNA was still not observed. In fact, the X-ray data revealed a structure which seemed to be quite distinct from the familiar right-handed helical B-DNA. It was a left-handed, antiparallel double helix, where the phosphate/deoxyribose backbone seemed to be zigzagging along the helix, and so the name Z-DNA was proposed in reference to the zigzag pattern. 103 The Z-DNA structure arises with that particular sequence of alternating cytosine and guanine in the presence of certain cations in high enough concentrations and will be discussed in detail later. It was not until 1980 that Dickerson and co-workers produced the first preliminary report on single crystal X-ray study of a full helical turn of B-DNA, d(CpGpCpGpApApTpTpCpGpCpGp). 104 Dickerson chose that sequence in the belief that he would obtain the Z-form as observed by Wang (vide supra) at the ends where the alternating cytosine/guanine was placed, while the AATT center would be “right handed or melted out”[sic]. Instead, they reported what over-

W

3.1.2. Z-DNA

43

all appeared very similar to Watson and Cricks original structure in spite of their sample containing a high number of alternating cytosine and guanine. This, once again, fortified the paradigm of the ubiquitous B-DNA as the biologically relevant structure, making Z-DNA a curiosity that appears in vitro at high salinity. 105 However, the atomic resolution now available opened the possibilities for further scrutiny. Already in the preliminary report mentioned above, two major deviations from the Watson/Crick model were emphasized, namely a propeller twist in the base pairs and a 19° bend of the helix. While the first had already been proposed from theoretical calculations and circular dichroism studies, the latter was attributed to mild deformation during crystallization. In later elaborations on their findings, Dickerson and co-workers addressed the deviations from the ideal B-form, and concludes that “the ’ideal’ B helix in DNA is as much an oversimplification as is the ’ideal’ α helix in proteins”. 106,107 An interesting point is that while some of the deviations from the ideal BDNA stem from rather low energy penalties on deformation, making the DNA docile, other deviations correlate to the base sequence. These nucleoside dependent deviations from the ideal B-form have since their discovery been embraced and utilized in explaining phenomena such as intrinsic bending in mini-circles of kinetoplast DNA, to the recognition of promoter regions by polymerases. Rather than assigning an overall type to a structure, it has become common to name single nucleosides in a structure as A, B or C form. 105 It should always be kept in mind that DNA does not constrict itself to one specific form, that it is docile and readily adapts to a plethora of intermediate and even outlandish geometries depending on conditions and base sequences. However, simplified models are often still useful in cases where the minor deviations are not important. As such, the term B-DNA is still widely used to denominate “normal” double stranded DNA where the ideal model still suffices.

3.1.2 Z-DNA s mentioned above, the single crystal X-ray data for Z-DNA was first obtained by Wang et al. in 1979. The structure arises when a DNA sequence with alternating cytosines and guanines is exposed to certain conditions. More specifically, these conditions include high concentrations of sodium or magnesium chloride, 108 hexaamminecobalt(III) chloride, calcium chloride, barium chloride, and ethanol. 109 In cases where the cytosines are methylated in the 5-position, the transition from B- to Z-DNA is observed at lower concentrations. As an example, the transition for non-methylated DNA is reported at a magnesium chloride concentration of 700 mM while the magnesium concentration in the methylated

A

44

C HAPTER 3. T RANSFER

OF

H ELIX C HIRALITY

TO

F LUORESCEIN

version needs only be as high as 0.6 mM. The 5-methylated cytosines also makes the DNA susceptible to transitioning to the Z-form by addition of spermine, spermidine and range of alkyldiamines. 109 Negative supercoiling has been known to stabilize Z-DNA in closed circular DNA. 110 The Z-form of DNA differs significantly from the B-form in that it constitutes a left-handed rather than a right-handed double helix. While normal WatsonCrick base pairing is maintained, Wang observed that the deoxycytidines have O1’ pointing down where the deoxyguanosines have O1’ pointing up. Since the Z-DNA base sequences consisted of alternating guanine and cytosine, this makes the repeat unit along the Z-axis two base pairs, rather than one as seen in B-DNA. The phosphates in the backbone are aligned such that if the repeat unit is defined as CpGp, there is almost no backbone twist within the repeat unit itself (excepting the terminal phosphate), only between repeat units. The phosphates between repeat ˚ as opposed to units also reach further than the ones within the repeat unit - 15 A ˚ ˚ 12.5 A- and as such are much more comparable to those of B-DNA (17.5 A). Additionally, the base pairs in this repeat unit are not stacked in the way they are in B-DNA. Instead a lateral translation between the base pairs in a single repeat ˚ causes the two pyrimidines to partially stack, while the purines unit of almost 7 A are situated almost right above the O1’ of the preceding ribose residue. Between repeat units, however, the stacking is a lot more similar to that of B-DNA, albeit twisted in a left-handed fashion, rather than right-handed. 103 The base sequence of alternating cytosine and guanine mentioned above is the arch typical Z-DNA sequence. Strictly, however, the structure only needs a base sequence consisting of alternating purines and pyrimidines to potentially adopt the Z-form. Repetitive sequences of the form d(TG/CA)n in genomic DNA have been shown to adopt a Z-DNA structure stabilized by negative supercoiling. 110 However, linear polymers of d(TG/CA)n only change to the Z-form under extreme salt conditions, even for Z-DNA, and generally require a Z-favoring modification on the pyrimidine-C5. 111 Jovin et al. provides an extensive review on Z-DNA, and among other things lists different factors that have an effect on the transition from B- to Z-DNA. A simplified overview of these factors is listed in Table 3.1. In some cases, such as [d(G-io5 C)]n , the factors can be so Z-favoring that the polymer in solution only exists on the Z-form under all ionic conditions. However, it may be reverted to the B-form by addition of preferential ligands such as ethidium bromide. 111 Another very thorough review is by Rich et al. 112

3.1.3. E XAMPLES

OF

Z- DNA

IN

NANOTECHNOLOGY

45

Table 3.1. A list of factors that affect a DNA strand’s ability to appear in the Z form. 111 The symbol > denotes the order of decreasing efficiency for stabilizing the Z-complex. In most cases, the effects are additive.

Intrinsic factors DNA sequence Base substitution

Backbone:

[d(G-C)]n > [d(A-C)·d(G-T)]n ≫ [d(A-T)]n > non-[d(Pur-Pyr)]n C5-Pyr: iodo > bromo > methyl ≫ H ∼ = aza C8-Pur: bromo > H N7-Pur: methyl ≫ non-alkylated N2-Pur: G > I; n2 A > A O4-Pyr: S > O 5’-p-Pyr: S ≫ O 5’-p-Pur: O ≫ S Extrinsic factors

Salt:

Hydration: Temperaturea Ligands Supercoiling a

higher ionic strength > lower higher cationic valence > lower chaotropic salts > non-structure breakers site-specific > territorial binding mixed solvents > water lower > higher Z favoring > B favoring negative > relaxed

The temperature dependence provided by Jovin et al. is incorrect and has been replaced by that reported by Rich et al. 112 and later confirmed by Tashiro and Sugiyama. 113,114 The temperature dependence of Z-RNA is the reverse of Z-DNA. 114

3.1.3 Examples of Z-dna in Nanotechnology ne of the first DNA nanomechanical devices that could switch dynamically between defined geometric states was reported bye Mao et al. in 1999. 14 This device was based on the transition between regular B-DNA and the more exotic left-handed helix, Z-DNA. The device, which is depicted in Fig. 3.1a, consists of two static, rigid domains made from DX motifs. The DNA in these regions remain on the B-form throughout, and the two domains are interlinked by a single double helix. This interlinking piece of DNA is able to transition between the B- and the Z-form depending on external stimuli. This gives the device two

O

46

C HAPTER 3. T RANSFER

OF

H ELIX C HIRALITY

TO

F LUORESCEIN

distinct states where the static domains are aligned either syn or anti relative to each other. a)

b)

Figure 3.1. The first dynamic DNA nanodevices with two distinct states. Both devices are based on the switching between the B- and the Z-form. In Mao’s twisting system (a), the transition happens in the connecting helix (yellow) between the two static regions. The connecting segment is on its B-form in the above state and on its Z-form in the bottom state. The switch between states is detected via FRET between the attached fluorophores (green and pink spheres). 14 In Yang’s system (b), the B-/Z-transition induced torque is countered by a shift in the base pairing in the Holliday junction effectively causing the device to contract. Yang controlled the transition by adding the B-DNA favoring ligand, ethidium. 115

To detect the transition between the two distinct states, the device was designed to accommodate two fluorophores that would be in close proximity when the device was in the B-state. In this state FRET efficiency would thus be high as opposed to the Z-state, where the fluorophores would be further apart. The same group had the previous year published a different device which also built on the B-/Z-transition, except in this case the rotational torque from the transition was transformed into an overall contraction of the device as described in Fig. 3.1b. 115 These first ventures into DNA nanotechnology were preceded in 1993, when Wang et al. reported a DNA design which during assembly formed one of two possible structures depending on whether the a specific segment in the DNA was on the B- or the Z-form during a ligation step. 116 The structure is described in Fig. 3.2 and was referred to as a Tight Knot by the authors. As illustrated, the the DNA strand is designed to hybridize with itself in a way that caused the strand to twist around itself. The number of twists could be increased by inducing a segment in the strand to switch to the Z-form. The final ligation of the strand locks the knot so it cannot unfold and, as illustrated in Fig. 3.2, the strand can either be locked into a Trefoil Knot or a Figure-8 Knot depending on whether the segment is on

3.1.3. E XAMPLES

OF

Z- DNA

IN

NANOTECHNOLOGY

47

the B- or the Z-form. Wang et al. provided the evidence for the two distinct knots by gel electrophoresis.

Figure 3.2. The tight knot by Wang et al. can either assume a trefoil or a figure-8 profile depending whether ligation is performed under Z-promoting conditions or not. Wang adds hexammincobalt(III) to induce Z-formation. 116

An interesting application of Z-DNA in nanotechnology was reported by Tashiro and Sugiyama in 2003, where they exploit two unique features of Z-DNA. These features are the temperature dependence on the formation of Z-DNA and the fact that the π-stacking in the B-DNA is broken when the strand switches to the Z-form. Combining these two features, Tashiro and Sugiyama produced a nanoscaled thermometer with a fluorescence based output. 113 The DNA-thermometer was an duplex with a length of 10 base pairs with alternating cytosines and guanines. To produce the fluorescence read-out, one GC pair was replaced by a 2-aminopurine/thymine pair. The fluorescent properties of 2-aminopurine were known to be quenched in B-DNA and Tashiro and Sugiyama expected that the broken π stack in Z-DNA illustrated in Fig. 3.3 would be unable to quench its fluorescence. Monitoring the fluorescence of a bulk solution, Tashiro and Sugiyama found a direct and fully reversible correlation between temperature and fluorescence. Two years later, they expanded on the system by employing Z-RNA in place of the Z-DNA. The reverse effect of temperature on the transition from A- to Z-RNA resolted in the exact opposite correlation between temperature and fluorescence. The fluorescence as a function of temperature is depicted in Fig. 3.4a for the DNA system and 3.4b for the RNA system. Tashiro and Sugiyama reports this to be the first reversible switching devices, to their knowledge, that show an exact inverted response to the same stimulus. 114

48

C HAPTER 3. T RANSFER

OF

H ELIX C HIRALITY

TO

F LUORESCEIN

Figure 3.3. Molecular models by Tashiro and Sugiyama of B-DNA (left) and Z-DNA (right) demonstrating how the π stacking of the latter is broken into 4 bp segments. The two strands are in yellow and blue. 113

a)

b)

Figure 3.4. Fluorescence as a function of temperature on the DNA B-/Z-thermometer (a) and the RNA B-/Z-thermometer (b). 114

3.2 The Project ormally, the secondary structure of a DNA double helix in solution is determined using circular dichroism (CD). CD is basically measuring the difference in absorbance of left-handed and right-handed circularly polarized light, respectively. A difference in the two absorbances occurs when a chiral chromophore is present. The technique does not provide structural information in the sense that for instance X-ray diffraction data does, but it can provide a way to discriminate between two known chiral chromophores, be they fluorophores, DNA or something completely different. A single peak in a CD spectrum thus represents the polarized bias in light absorbance of a chromophore and is referred to as a Cotton

N

3.1.3. E XAMPLES

OF

Z- DNA

IN

NANOTECHNOLOGY

49

effect. Anyone who has ever measured the concentration of DNA in a sample knows that DNA has a characteristic light absorption peak around 260 nm. As a consequence of the inherent chiral nature of the DNA double helix, this absorption shows up as a sigmoid signal in a CD spectrum. The switch in chirality between B- and Z-DNA results in an inversion of the sigmoid CD signal, and so the technique is very straight forward when one wishes to determine whether an entire sample is on the B- or the Z-form. If, however, it is only a small part of the sample that is changing conformation, the signal from the unchanging remainder of the sample may easily drown out any signal changes. In DNA nanostructures and devices where the transition between B- and Z-DNA is used to add a dynamical element, this would typically be the case. The major part of the ensemble consists of B-DNA, while only a small part is able to switch to the Z-form. Assuming that an achiral chromophore that was conjugated to a DNA double helix could adopt the helix chirality the way the DNA bases do, the CD spectrum for the chromophore would contain the same information as that of the bases. The two major differences would be that the information is very specifically contained to the region the chromophore is positioned, and ideally that the wavelength could be tailored to meet any needs by choosing an appropriate chromophore. In addition, excitons - i.e. two or more chromophores combined to provide a single excited state together - tend to provide a strongly amplified signal in CD compared to single-molecule excited states. Also, it could be speculated that the exciton’s chirality could be even more dependent on changes in helical structure, since the two chromophores that make up the exciton follow the DNA backbone residues independently of each other. Exciton chirality arises from the spatial interaction between the chromophoric electric dipole transition of two or more chromophores when these are chirally arranged and close enough in both energy and space to interact. In the simple case with no external perturbations, the exciton produces two Cotton effects with opposite signs in the CD spectrum (an exciton couplet), where the couplet crossover point in the degenerate case (i.e. identical chromophores) coincides with the point of maximum absorbance. The terminology defines the sign of the couplet to be the same as that of the Cotton effect with the longest wavelength, and the sign of the couplet reflects the spatial angle between dipole transition moments. The strength of the exciton coupling is inversely proportional to the energy difference and spacial distance between chromophores and is also highly dependent on the angle between transition dipoles. For collinear or coplanar dipoles, the dipole coupling term is zero, while a maximum dipole interaction is achieved at a dihedral angle of 70° for dipoles lying in quasiparallel

50

C HAPTER 3. T RANSFER

OF

H ELIX C HIRALITY

TO

F LUORESCEIN

places. 117 As mentioned earlier, the aim of the current project was primarily to show that attaching two chromophores to a DNA double helix does indeed open the possibility for exciton coupled circular dichroism (ECCD) - that is to say, that the chromophores couple to form an exciton, and that the exciton inherits chirality from the DNA. The secondary aim was to make use of this phenomenon in detecting a secondary structural change of a DNA double helix. More specifically, it was the hope that switching between the B- and the Z-form would reflect in CD measurements of the attached chromophores. When choosing the chromophores for the project, the primary criterion was, of course, that its absorption wavelength be different from that of DNA. Secondly, there was a wish to eventually expand the scope of the project from simple CD to fluorescence detected CD (FDCD), a technique where the polarization of emitted light is measured, combining the structural information from CD with the high sensitivity from fluorescence detection. 117 Because of this, fluorophores were selected for the project. Fluorescein and Cy5 were the fluorophores of choice, and when preliminary results for fluorescein looked the most promising, further experiments were focused on fluorescein conjugates alone.

3.2.1 Conjugation of Fluorophores ttachment of the fluorophores to the DNA was accomplished by “clicking” the azido modified fluorophore onto a alkyne-modified DNA oligonucleotide via the Huisgen-Meldal-Sharpless copper(I) mediated azido-alkyne [2+3]-cycloaddition (CuAAC). The DNA-oligonucleotide in question had previously been synthesized using routine phosphoramidite procedures. 118 The alkynes needed for the CuAAC coupling had been situated at the desired positions in the sequence by substituting the regular nucleotide for the corresponding 2’-O-propargylribonucleotide.i This conjugation strategy allowed for simple conjugation of different fluorophores without the need to resynthesize the same oligonucleotide multiple times. It also had the advantage of the DNA synthesis conditions being independent of the choice of dye, and even base labile dyes are made possible candidates in any future expansions on our present work. 119 Placement of the conjugate in the 2’-position was chosen in the hope of minimizing interference with base pairing, and in the

A

i

All custom amidite DNA strands used in this chapter were synthesized by Dr Roujie Sha of the Seeman laboratory

3.2.2. C LICK R EACTION

51

hope that helix chirality would have a stronger influence when the conjugate was placed directly on the chiral saccharide. A significant difference in the DNA secondary structure between the B- and the Z-form is the fact that while the sugar residue on all cytidines remain as the anti rotamer relative to the base, the sugar residues on the guanosines changes to a syn conformation. Since all the sugars in B-DNA are anti relative to the bases, conjugation of the fluorophore to the sugar moiety on guanosine is a judicious choice when distinction between B- and Z-DNA is the object of the experiment.

3.2.2 Click Reaction s mentioned above, the azido-modified fluorophore was conjugated to the nucleoside utilizing the copper(I) mediated azide-alkyne [2+3]-cycloaddition (CuAAC) as reported by Meldal 120 and Sharpless. 121 The reaction is also known as one of the “click” reactions, a term coined by Sharpless in 2001, although it originally referred to the non-copper mediated Huisgen reaction. 122 As a catalyst and in order to stabilize the cupper(I) against disproportioning 123 and to prevent DNA degradation caused by reactive oxygen species generated in the presence of copper(I), 124 the copper(I)-binding ligand tris-(1-[3-hydroxypropyl]triazolyl-4-methyl)amine (THTA) was added to the reaction. 125 Previous work has demonstrated this particular ligand’s efficiency in CuAAC couplings to DNA and ease of use in these couplings due to high water solubility. 20,126 Conjugations of azidofluorescein to the propargyl modified DNA strands was initially attempted with different sources of commercially available copper(I) salts. This was, however, quickly abandoned in favor of in situ reduction of copper(II) sulfate with sodium ascorbate. This procedure was easily executed and showed full conversion on denaturing PAGE analysis. On an interesting side note, it should be mentioned that while the CuAAC reaction is generally considered pH independent, 127 full conversion for the fluorescein conjugation was only observed under slightly acidic conditions (pH 4.3). Reaction was still observed at neutral pH, however, unreacted starting DNA was also detected in the PAGE analysis. The mechanistics behind this was not been investigated, however, two different explanations were considered. Initially, a change in secondary structure of the single strand was considered to be the cause. Protonation of cytosine is known to be a prerequisite for some types of Hoogsteen base pairing, 128 so acidification could yield a secondary structure more accessible to reaction at the propargyl modification site. When, however, conjugation of Cy5 was subsequently attempted, the reaction conditions that proved to work best included buffered conditions at pH 7. With this in mind, the need for acidification

A

52

C HAPTER 3. T RANSFER

OF

H ELIX C HIRALITY

TO

F LUORESCEIN

seemed to be pertaining to the azido-fluorescein, itself. It has long been know that fluorescein’s ability to fluoresce is pH dependent, and seems likely that the change in electronic properties that affects fluorescence may affect reactivity as well. The pKa of the neutral fluorescein is 4.4, so the reduced reactivity at neutral pH compared to acidic implies that the neutral fluorescein specie is more reactive than the deprotonated. 129 Alternatively, it is simply a question of electrostatic repulsion of the deprotonated specie from the negatively charged DNA backbone. Conjugation of the Cy5-azide did not proceed as cleanly as in the case of fluorescein. PAGE analysis always produced two or three bands aside from the product band. It was found that premixing of copper(II) sulfate, THTA ligand, Cy5-azide, and sodium ascorbate before adding the alkyne-modified DNA gave the best results. While side products were evident in PAGE, the final isolated yield was comparable to that of the fluorescein conjugation.

3.2.3 DNA Sequences

s demonstrated in the literature, the oligonucleotide sequence most prone to undergo switching from the B- to the Z-form is a sequence of alternating dG and dC. 110 Hence, the project strands were designed from a d(CG)20 starting point and bases number 2 and 19 were switched from guanine to thymine. This was done to promote hybridization with the complementary strand over stem loop formation or hybridization between identical strands. Three different relative positions of the fluorophore modification were investigated: one design had both fluorophores positioned on the same strand in the double helix, six base pairs apart, and the remaining two designs had the fluorophores on opposite strands, three and five base pairs apart, respectively. The sequences are listed in Table 3.2, and Fig 3.5 illustrates the relative placement of the fluoresceins on the helices.

A

3.2.4. C IRCULAR D ICHROISM

Table 3.2.

Strands for Transfer of Helix Chirality

Name

Sequence

V V’ M8 M’8 M’10 D8,14

CTC GCG CGC GCG CGC GCG TC GAC GCG CGC GCG CGC GCG AG CTC GCG CG* C GCG CGC GCG TC GAC GCG CG* C GCG CGC GCG AG GAC GCG CGC G* CG CGC GCG AG CTC GCG CG* C GCG CG* C GCG TC

*

a)

DNA

53

signifies position of modification.

b)

c)

Figure 3.5. Placement of fluorophores on the double stranded DNA. In a) the two fluorophores are conjugated to the same strand (D8,14), in b) and c) the fluorophores are placed on opposite strands but with different distances. One fluorophore was conjugated to M8 and this was hybridized with either M’10 (in b) or M’8 (in c), each carrying a fluorophore as well.

The objective for testing these different relative placements of the fluorophores on the double helices was to find the optimum conditions for CD spectroscopy on the samples. Since the angle between the fluorphores would have a large impact on the CD signal, the three different fluorophore placements were expected to be quite distinct in the CD spectra – what the differences would be, we did not know. Additionally, we expected the B-/Z-transition to have a different amount of impact on the spectra.

3.2.4 Circular Dichroism nitially, CD spectra were recorded for the unmodified DNA to find fitting conditions for the transition between B- and Z-DNA. Hexaammincobalt(III) chloride had been chosen for inducing the switch to Z-DNA because it was reported

I

54

C HAPTER 3. T RANSFER

1


OF

H ELIX C HIRALITY

TO

F LUORESCEIN

! / M c

B D Z D

1




" / n




22


2


2


2






2






1

1


Figure 3.6. CD spectra of unmodified double stranded DNA of 20 base pairs in length. CD spectra are shown for the DNA both on the B- (blue) and the Z-form (red).

to work in very small concentrations compared to sodium or magnesium chloride. However, after numerous failed attempts at inducing the transition from B- to Z-DNA cobalt(III) concentrations ranging between 0.25–20 mM. This puts the highest tested cobalt(III) concentration three orders of magnitude higher than the needed concentration according to Behe and Felsenfeld. 109 While Behe’s figures are for poly(dG-dC) · poly(dG-dC), my experiments only involved a double stranded oligonucleotide 20 base pairs in length. Furthermore, Behe used tris buffer (5 mM, pH 8 with 50 mM NaCl) and I followed a protocol from Mao et al. using cacodylate buffer (10 mM, pH 7.5 with 10 mM magnesium chloride and 100 mM sodium chloride). 14 This, of course, means that the concentrations are not directly comparable. In Mao’s article, the DNA meant to undergo the B/Z-transition was made with 5-methyl cytosines, making it more prone to adopt the Z-form. In this case, Behe claims that cobalt(III) concentrations as low as 5 µM should suffice, yet Mao reports using a concentration of 250 µM for stabilizing the Z-form – fifty times higher than Behe’s. Testing the possibility that the cacodylate buffer causes the need for a higher cobalt(III) concentration, it was attempted to induce the Z-DNA formation in tris buffer as used by Behe, added hexaammincobalt(III) chloride until a final concentration of 20µM. Also this was to no avail. The reason that cobalt(III) could not effect the transition from B- to Z-DNA was never found. However, after abandoning cobalt(III) for 20 w/w-% sodium chloride, the transition to Z-DNA was observed immediately in CD (Fig. 3.6). Following conjugation of the fluoresceins to the oligonucleotides in the man-

3.2.4. C IRCULAR D ICHROISM

55

a) 6

b) 

!"/ M!1 c
!1

4



2 # / n


2 4 6 8 

!"  



22

2

32

3

42

4

52



 



 







 



    

Figure 3.7. First CD spectra of fluorescein conjugated double stranded DNA. In a) only one fluorescein is present on each double strand made from M8’ and V strands as named in Table 3.2. In b) two fluoresceins are conjugated to the same strand (D8,14), and this is hybridized to an unmodified strand (V’).

ner described in section 3.2.2, the modified oligonucleotides were paired up with each other or with non-modified complementary strands as described in Fig. 3.5 and hybridized following standard protocols. Subsequently, the resulting double helices were subjected to CD spectroscopic measurements. Initially, under conditions that favored the B-form, then the salinity was increased to favor Z-DNA formation and the CD measurements were repeated. The first CD spectra recorded for the fluorescein conjugated DNA on the Bform were interesting. Two parallel experiments with two different modified strands were being performed; one was the mono-modified M8’ strand, and the other was the dimodified D8,14 strand (See Table 3.2), both hybridized with an unmodified complementary strand. It seemed that the fluorescein did not give a visible signal in the spectra if only one fluorescein was present on the helix as in Fig. 3.7a. If, however, the double helix had two fluoresceins attached, a clear, strong sigmoid signal emerged in the range of 450–550 nm as in Fig. 3.7b. This overall means that the two fluoresceins on the DNA together couple to form an exciton. Some typical characteristics for coupled excitons in CD are the high strength and the sigmoid shape of the signal, where both a positive and a negative Cotton effect are present at the same time. The sole fluorescein in Fig 3.7a may easily have adopted a chiral environment from the DNA strand and thus have a chiral bias in absorption. If that is the case, however, the signal in the CD spectrum in Fig. 3.7a is simply not strong enough to be visible. When, however, two fluoresceins couple to form an exciton, the CD signal becomes very strong. As mentioned, the CD signal around 500 nm in Fig 3.7b comes from two fluoresceins on the DNA double helix. For comparison, the signal in the 225–

56

C HAPTER 3. T RANSFER

OF

H ELIX C HIRALITY

TO

F LUORESCEIN

300 nm range comes from the 40 nucleobases attached to the same double strand. In other words, the strength of the signal from the two fluoresceins is comparable to that of the 40 nucleobases. This is a vital quality in the ultimate goal of detecting local secondary structure changes in large systems. A similar experiment was executed with the Cy5-modified strands, but neither with one nor with two fluorophores did the CD spectra show any signal from the Cy5. Later characterization of the dimodified strand, however, indicated that only one Cy5 had been conjugated to the strand, so the CD spectra for a DNA double strand with two Cy5 was never recorded. Due to the promising results using fluorescein and the limited time for further experimentation, the use of Cy5 was not pursued further. Following the recording of the CD spectra of the fluorescein conjugated double strands on their B-form, the salinity was increased as mentioned above. This caused the double strands to change into the Z-form and CD-spectra were recorded once again. In Fig. 3.8, the superimposed spectra of the B- and the Z-form are presented, and the difference in the signals from the fluoresceins is quite distinct. ! / M!
c!


1

5 " / n

22

2

2



2



52

5

5

1

15

B f Bf Zf

Figure 3.8. CD spectra of fluorescein conjugated D8,14 hybridized with the unmodified V’. The blue curve is for the complex on its B-form, and the red is on its Z-form at high salinity. Note the difference in signal from the fluorescein around 500 nm between the two forms.

Exploring the changes in fluorescein interaction, two additional placements of the fluoresceins (Fig. 3.5b,c) were tested along with the original positioning (Fig. 3.5a). As was expected, the three had different effects on the CD spectra recorded (Fig. 3.9). While the fluoresceins produced a signal in all three cases, both under B- and Z-favoring conditions, the changes in signal between the B- and Z-form are dif-

3.2.4. C IRCULAR D ICHROISM

a)

b) ! / M!1 c
!1

6



" / n


 

42

4

Bf
Zf


2



   

"# $%"#

& ' " 



!





2

33

c) !  



4



57





 





"  *%

 ,- + ,+ ,

&'(

&)(

'(

)(

.,057% 8,057%

,)

Figure 3.9. CD spectra for the three different placements of fluoresceins on the DNA double helix. Only the regions of the spectra pertaining to the signal from fluorescein are depicted. The order of the spectra corresponds to the order of the fluorescein placement illustrated in Fig. 3.5: a) both fluoresceins are on the same strand six bases apart, b) fluoresceins are placed on opposite strands three bases apart, and c) fluoresceins are placed on opposite strands fives bases apart.

ferent. The design where the two fluoresceins are on opposite strands three bases apart seems the least affected by the switch between the B- and the Z-forms. Both the others, however, make the distinction between the two very clear. When both fluoresceins are on the same strand six bases apart, the B-form had a strong sigmoid signal which changed to two weak Cotton effects that were both positive in the spectrum. Monitoring at a wavelength of 525 nm would thus provide a strong change in signal when switching between the two DNA secondary structures, going from a strong negative signal in the B-form to almost no signal on the Z-form. Interestingly, for the fluoresceins on opposite strands five bases apart, a strong, positive Cotton effect is seen on the Z-form and only a weak sigmoid signal is seen in the B-form. As such, these both provide an excellent means for detecting changes in secondary structure. These results were obtained in the days leading up to my departure from New York. This limited the possibilities for pursuing further investigations, however, Dr Zhaohua Dai was able to assist in the acquisition of fluorescence detected circular dichroism (FDCD) data. For this, the design with both fluoresceins on the same oligonucleotide was chosen because it seemed to have the strongest signal from the fluorescein in CD and because a significant difference in CD spectra had been observed when changing between the B- and the Z-form. The FDCD spectra recorded, however, did not provide any significant distinction between the B- and the Z-form as seen in Fig. 3.10. Some quenching of the fluorescein was observed, however, a clear signal was

58

C HAPTER 3. T RANSFER

OF

H ELIX C HIRALITY

TO

F LUORESCEIN

Figure 3.10. The superimposed FDCD spectra of the fluorescein dimodified D8,14 hybridized with the unmodified V’. The spectra are recorded with the DNA double helix on the B-form (red) and on the Z-form (black).

detected in FDCD after some attempts. As it appears in Fig. 3.10, the signal, however, did not alter depending on whether the DNA was on the B-form or the Zform. This disappointment was later explained when the original CD experiment was repeated with a more concentrated sample prepare right before my departure. By measuring on the more concentrated samples, Dr Dai was able to obtain a better signal-to-noise ratio and the data, now much less sensitive to artifacts, revealed a different spectrum in the Z-form than was observed earlier. The spectra collected for both the B- and the Z-form as depicted in Fig. 3.11a held close to identical Cotton effects, only with slightly different amplitudes. a)

b) ! / M!1 c
!1

6



4



2

 " / n


 3 3  

!  

42

4

2



Bf
Zf


" 

   









 



Figure 3.11. Superimposed CD spectra for both the B-form (blue) and the Z-form (red) repeated with samples of higher concentration. The spectra are of (a) the fluorescein modified complexes D8,14·V’, and (b) M8·M’8

The relatively large difference observed earlier was attributed to a systematic error, and when the CD spectra are close to identical, the similarities between the

3.2.4. C IRCULAR D ICHROISM

59

FDCD spectra for the B- and Z-form is to be expected. While both CD and FDCD spectra for the D8,14·V’ complex verify the helix chirality transfer to the conjugated fluorophores, this chirality could just as easily arise from chiral glycosidic bond by which the fluorophores are attached. However, the recording of the CD spectrum for the M8·M’8 complex was also repeated with a higher sample concentration, and as seen in Fig. 3.11b, the original observation was affirmed in this case. For this placement of the fluorophores, there is a significant difference between the CD spectra for the B- and the Z-form, respectively. Since the covalent bonds in the structures do not change between the B- and the Z-form, this verifies the hypothesis that the formed fluorophore exciton adapts chirality from the DNA helix, and that it can be altered along with the chirality of the helix.

3.3 Conclusions and Perspectives luorophores were “clicked” to oligonucleotides with an oxopropargyl modification in the 2’ position on the nucleotide where the fluorophore was desired. It was demonstrated how two fluoresceins on a DNA double helix coupled to form a chiral exciton that was visible in CD spectroscopy while a single fluorescein did not produce a detectable signal. Initial results indicated that the signal from the chiral exciton was dependent on the secondary structure of the double strand in two of the three different fluorophore placements tested. One of these was chosen for further testing to see if the signal change could be detected with FDCD. A signal from the fluorescein was detected in the FDCD spectra, however, it did not change with the switch from B- to Z-DNA. The regular CD spectrum for the transition was repeated, and the change in signal seen in the early spectra was no longer visible. The initially detected signal change was attributed to artifacts. As stated above, only the results from the fluorophore placement chosen for further testing was refuted. The other design where a large signal difference between the B- and the Z-form was observed, the data was confirmed. This also verifies the hypothesis that chirality is transferred from the helix to the fluorophore exciton. It still remains to be tested whether this particular setup will produce distinct FDCD spectra for the B- and Z-form of the DNA. The conjugation of the fluorophores on the sugar residues of a guanosines was chosen deliberately to achieve the largest distinction between the fluorophore geometries in the B- and the Z-form, since these change from an anti to a syn rotameric position while those of cytidine remain anti. Alternatively, it could be argued that the largest difference would be achieved by conjugating only one of the flurophores to the sugar of guanosine while the other be conjugated onto that of

F

60

C HAPTER 3. T RANSFER

OF

H ELIX C HIRALITY

TO

F LUORESCEIN

a cytidine. In this fashion, only the orientation of the guanosine bound fluorophore would be inverted and the difference in their relative geometry might be increased. In elucidating this point, computational modeling might contribute as well as the experimental testing of different designs in the wet laboratory. When an appropriate placement of the fluorophores has been found, it opens the possibility for exploring the capabilities of this system. Obviously, it gives a direct possibility of exploring the chirality transfer of the DNA helix onto substituents, and in this context different linkers and substituent positions on the nucleotide should be investigated. A possibly interesting expansion on the transfer of helix chirality to substituents might be found in combination with DNA directed reactions. This way an element of chirality might be deliberately induced in a product of two achiral reactants conjugated to DNA. Recently, Kuzyk et al. reported on the production of a DNA six-helix bundle with nine gold nanoparticles attached in a helical pattern around and along the bundle. 130 Both a version where the gold nanoparticle helix was right-handed and one where it was left-handed was constructed, and it was demonstrated how the gold nanoparticles produced two distinct patterns in CD spectroscopy depending on the chirality of the helix shape produced by the nanoparticles. Kuzyk refers to an article by P. J. Pendry arguing that the highly coveted and thus far illusive negative index of refraction in theory can be achieved through chirality of a chromophore. 131 While this remains to be proven in the laboratory, a B-/Z-switching design as the one presented in this chapter might be of high interest in optical applications.

C HAPTER 4

DNA ACTUATORS

he following chapter describes a project that was started in our group many students ago, and recently had a major break-through in terms of results. The original project constitutes the design and construction of a DNA tile actuator with 11 discrete states. The actuator consists of two so-called piston strands linked together with a so-called roller strand. The base sequences in the strands are designed such that each piston strand hybridizes specifically onto opposite sides of the circular roller strand. In between the two specifically hybridized regions on the roller strand are regions that can hybridize with either of the two pistons. As a result, one can imagine the two pistons moving back and forth with the roller strand working as a DNA version of a sprocket in between the two - hence the names pistons and roller. While this macroscopic analogy is unlikely to transfer completely to the nanoscale behavior of DNA, the DNA tile actuator has been shown by Zhao Zhang, Mille B. Kryger and Eva M. Olsen in our group to lock in 11 discrete states, i.e. relative positions of the pistons. Additionally, they have shown that this could be used to control FRET and the reaction of small molecules conjugated to the piston strands. Expanding on the actuator paradigm, they also managed to design and construct both a 3-way and a 4-way version of the DNA tile actuator and show that these lock in discrete states in the same fashion as the above mentioned linear actuator. 132 To expand this concept further, I set out to couple multiple actuators together, opening the possibilities of constructing complex dynamic systems from block elements.

T

61

62

C HAPTER 4. DNA ACTUATORS

4.1 Introduction 4.1.1 Dynamic DNA Nanostructures n the previous chapter, the concept of dynamic DNA nanostructures was briefly touched upon, and as mentioned some of the first DNA nanodevices capable of dynamically altering their structure were published by the Seeman group in the 1990’s. The concept of the devices, which are depicted in the previous chapter in Fig. 3.1, was based on the transition between B- and Z-DNA giving two distinct, geometric conformations. 14,115 Other examples of environmentally controlled dynamic DNA devices include the pH controlled transition between regular B-DNA and the i-motif formed between cytosines at low pH. 133,134 This transition, which is illustrated in Fig 4.1, has been automated by Liedl and Simmel by applying an oscillating type reaction; a reaction that sequentially raises and lowers the pH of the solution causing the cycle in Fig. 4.1 to propagate over time. 135,136

I

Figure 4.1. A pH-dependent, two-state DNA device. The device forms a regular, straight double helix on the right for pH values above 6.5, while protonation of the cytosines under more acidic conditions causes the X* strand to form the i-motif on the left. 133

Other dynamic DNA devices have been designed invoking the dynamic element through in situ displacement of DNA strands in the structure. The first such design was constructed by Yurke et al. in 2000 and took on the shape of a set of DNA tweezers. 137 The DNA tweezers, as illustrated in Fig. 4.2a, have a closed and an open state. The two termini of the tweezer arms (black regions) have a fluorophore and a quencher attached, respectively, for detection purposes, and in the closed state, they are held together by strand F. Strand F, however, still has a small unhybridized section, referred to as a toehold (red), which allows for the

4.1.1. DYNAMIC DNA NANOSTRUCTURES

63

removal of the strand by adding the fully complementary strand F. This removes F from the tweezer complex and reverts its state to open. This toehold facilitated removal of a strand from a complex is known as strand displacement. a)

b)

Figure 4.2. Examples of strand displacement as a fuel in dynamic DNA nanodevices. In Yurke’s device (a), adding strand F brings the device to its closed state, while adding F removes F by forming the “waste” complex F·F. Removing F reverts the tweezer back to its open state. 137 Yan’s one dimensional array (b) changes between its anti and syn forms by strand displacement. 138

Strand displacement has since accounted for conformational changes in many systems. Simmel and Yurke expanded the tweezer setup to a three state system. 139 Yan et al. reported the use of the technique in switching between syn and a trans conformers of a rigid nanodevice. Not unlike the first B-/Z-controlled device from the Seeman group, however, instead of a two rigid domains, this device consisted of a whole one-dimensional array of rigid DNA tiles (Fig. 4.2b). 138 The device responsible for the switching was since incorporated into a two-dimensional array by Ding and Seeman. 140 Another building block in dynamic DNA nanostructures is the use of DNA hairpin loops. A hairpin loop is a segment of DNA where the one end of the segment is complementary to the other causing the strand to hybridize with itself and form the characteristic hairpin shape. These have their place in dynamic DNA devices as contracting elements that can be countered by adding a DNA strand that is fully complementary to the entire hairpin segment and, preferably, a toehold segment. An early example of this use of hairpins for dynamic contraction was reported by Feng et al. in 2003. A two-dimensional lattice was devised to contract via hairpin loops, and subsequently expand when a complementary strand was added as illustrated in Fig. 4.3a. 141 Aldaye and Sleiman applied the same principle in the contraction and expansion of their DNA polyhedra, however, as illustrated in Fig. 4.3b, they demonstrated how three discrete states, i.e. degrees of contraction,

64

C HAPTER 4. DNA ACTUATORS

could be achieved from this technique. 142 a)

b)

Figure 4.3. DNAstructural contraction by hairpin formation. The two-dimensional lattice (a) by Feng et al. has a contracted and an expanded state. The contracted state is produced by hairpin formation while the expanded state is achieved by opening the hairpin with a complementary strand. 141 The polyhedron (b) by Aldaye and Sleiman was designed to contract and expand in the same fashion, however, two levels of contraction were achieved with different complementary strands. 142

4.1.2 The DNA tile actuator n the spring of 2011, Zhang et al. published the first example of a fully functional DNA tile actuator with 11 discrete states. 132 The actuator was designed around the principles of the Holliday junction, a structure that plays a large role in the scheme of genetic recombination. 143 The Holliday junction consists of four DNA strands with a sequence homology that allows for a continuous exchange in hybridization between the four strands effectively moving the junction along the DNA. Breaking homology renders the junction immobile, in which case it is merely referred to as a four-way junction, a construct that has played a great role in DNA nanotechnology ever since the beginning. 144 Until the publication of the DNA tile actuator, however, Ned Seeman’s double cross-over tile (DX tile) from 2000 was the only artificial dynamic DNA nanostructure design that was based on

I

4.1.2. T HE DNA

TILE ACTUATOR

65

the Holliday junction. 145 The DNA tile actuator design is based on the same double cross-over motif as the DX-tile. In Fig. 4.4a is an illustration of the classic DX tile with two cross-overs, one in each end, and right below in Fig. 4.4b, is an illustration of the unlocked actuator. In the actuator nomenclature, the two strands labeled A and B are referred to as piston strands and the one labeled R is the roller strand. All red section of the piston strands have the same base sequence, introducing an element of homology that allows for the roller strand to continuously change hybridization from one piston strand to the other. This allows the pistons to freely “roll” back and forth on the roller strand as indicated in Fig. 4.4b where only the extreme positions are showed. Rolling beyond these extreme positions is prohibited by the intentionally lacking homology.

Figure 4.4. Schematic illustration and 3D model of a DNA actuator. a) DX tile inspired by the Holliday junction, where the actuator originates. b) The assembled, unlocked actuator consisting of piston-A (A), piston-B (B) and roller (R). The sequence homology in the red regions allows for the pistons to roll on the roller. c) A lock strand (L), which is complementary to specific parts of both pistons, is added to fix the complex into a specific state. d) A full-atom model of the actuator. The max distance between centers of two helices is set to 3 nm. 132

Each of the homologous regions contain 11 nucleobases, each providing one discrete step in the rolling from one extreme to the other. The actuator can be fixed in each of these intermediate states by adding two identical, U-shaped lock

66

C HAPTER 4. DNA ACTUATORS

strands. As illustrated in Fig. 4.4c, the U-locked complex resembles the DX-tile in Fig. 4.4a. They are, however, rotated 180° compared to the strands in the ends of the DX-tile, and this prevents changes in hybridization. While the schematic illustrations in Fig. 4.4a–c show the piston strands to be parallel, initial work by Seeman and coworkers on the DX-system 11 and later work by Rothemund 18 and Andersen et al. 146 in the context of DNA origami suggest reality to be slightly different. Electrostatic repulsion between DNA strands causes the actuator to deform slightly compared to the schematic representation. In Fig. 4.4d is a model by Zhang et al. of the locked actuator where electrostatic forces have been taken into account. The two piston strands bend away from each other on the middle. In this case, Zhang estimates the maximum helix center to center distance to be 3 nm. Aside from PAGE analysis of the complex showing complete assembly of the complex both with and without lock strands, Zhang et al. produced a set of experiments that would produce evidence that the actuator effectively locked in the states dictated by the lock strands. These consisted of two experiments using F¨orster Resonance Energy Transfer (FRET) and two with a chemical reaction between an amine and an carboxylic acid to form an amide bond. The experiment designs and the results are provided in Fig. 4.5. In the FRET experiments, each piston strand was conjugated to a fluorophore. In one case, the fluorophores were situated such that they would be closest in one of the extreme states named State 0 (Fig. 4.5a). In the other case, the two would be closest in state 5 (4.5d). The measured FRET data for the first scenario, where the fluorophores were closest in State 0, are depicted in Fig. 4.5b, and when comparing to the theoretically calculated values in Fig 4.5c, similarity in shape is clear. Similarly, for the case where the fluorophores are closest in state 5, the observed FRET values (Fig. 4.5e) are very similar in shape to the calculated values (Fig. 4.5f). In the experiments where chemical coupling between an amine and a carboxylic acid was used to indicate effective difference between states, the reacting groups were positioned where the fluorophores were in the FRET experiments. Following locking of the actuator in the desired state, the reaction vessel was added the coupling reagent 4-(4,6-dimethoxy-1,3,5-triazin-2-yl)-4-methylmorpholinium (DMTMM), which activates the carboxylic acid making it more prone to react with any present nucleophile. In states where the carboxylic acid and the amine were close enough to each other to react, the reaction would covalently bind the two piston strands together yielding a 128 nt single molecule, which would easily separate from the 64 nt piston strands and the 65 nt roller strand in denaturing gel electrophoresis. The gels in Fig. 4.5h and i are the re-

4.1.2. T HE DNA

TILE ACTUATOR

67

Figure 4.5. Data collected by Zhang et al. for 11 different states of the actuator. All data on the left are collected from actuators with the detection groups placed as illustrated in a). The illustration is in State 0, where the groups are closest and arrows indicate the motion of the pistons when moving to a higher state. b) depicts collected FRET data with a CyB/Alexa647 FRET pair. c) Theoretical FRET curve. d)illustrates the placement of detection groups for all data on the right. In this design, the groups are closest in State 5. e) depicts collected FRET data, where the fluorophores are placed as illustrated in d). f) shows the corresponding theoretical FRET values. g) is the amide coupling reaction between the amine and carboxylic acid conjugated to each of the piston strands. h) is the PAGE analysis of the amide coupling between the piston strands in the 11 different states, where the placement of the reacting groups were placed as illustrated in a). It is seen that reaction only occurs in states 0 and 1. i) contains the PAGE analysis, where the reacting groups were placed as illustrated in d). It is seen that reaction occurs in the states close to state 5. 132

68

C HAPTER 4. DNA ACTUATORS

sults from the two experiments where the reactive groups are closest in states 0 and 5 respectively. Again the results indicate a successful locking of the actuator in the different states. On an interesting side note, however, the gel in Fig. 4.5i, which shows the experiment where the reacting groups are closest in state 5, seems to indicate that the groups are close enough for reaction to occur in significant amounts in states 3–7. Meanwhile, in the case where the groups are closest in state 0, reaction only occurs in states 0 and 1 - a much narrower distribution. While the above FRET measurements were performed on separate samples locked in each their state, Zhang et al. also inspected the dynamic system where locks are removed by standard strand displacement before new locks are added. This was initially studied through gel electrophoresis by sequentially locking and unlocking the actuator and extracting an aliquot of the sample for analysis following each locking or unlocking operation. For these experiments, the actuator was only locked in states 0, 5, and 10.

Figure 4.6. Dynamic locking and unlocking of the actuator in states 0, 5, and 10. The sequential transition between locked states (S0, S5 and S10) and the unlocked state (ABR) is illustrated in a) and the corresponding PAGE analysis of the complexes is depicted in b). FRET experiments with two different placement of the fluorophores are given in c) and d), and corresponding time resolved FRET are depicted in e) and f). 132

The locking/unlocking sequence of the experiment is illustrated in Fig. 4.6a and the corresponding PAGE analysis in Fig. 4.6b shows how the retention of the locked complex is increased when locked, and how it reverts to its original retention upon unlocking. The lowest band on the gel is the accumulated lock/unlock

4.1.3. T HREE -

AND

F OUR - WAY DNA ACTUATOR T ILES

69

strand complex that forms whenever the complex is unlocked by strand displacement. Fig. 4.6c and d show the corresponding FRET experiments for the statically locked complexes. In Fig. 4.6e and f, however, the locking of the complex and the unlocking by strand displacement are monitored over time. These time resolved measurements confirm that not only can the lock strands be displaced dynamically, the actuator is also dynamically fixed in certain geometries upon locking and subsequently released when unlocking.

4.1.3 Three- and Four-way DNA Actuator Tiles n addition to Mille Kryger’s contributions in the work described above on the DNA tile actuator, she reported in her Master thesis the design and construction of two alternative actuators. 147 While the original actuator design was designed for movement in one dimension, Kryger’s triangular and quadrangular designs expand into two dimensions. The designs are based on the same principle as the original actuator, where piston strands are hybridized with a common roller strand and sequence homologies allow for the pistons to “role”. Instead of two piston strands per roller, however, Kryger’s designs, as illustrated in Fig. 4.7, had three and four pistons, respectively.

I

a)

b)

Figure 4.7. Kryger’s two alternative actuator designs: (a) the triangular and (b) the quadrangular actuators. The piston strands (blue) are identical and hybridize with the complementary regions on the roller strand (yellow). The number of piston complementary regions on the roller dictate the shape of the actuator. Lock strands (green) lock the complex analogously to the original, linear actuator. (Images provided by Mille Kryger)

For the sake of simplicity, Kryger’s design took the symmetric approach of

70

C HAPTER 4. DNA ACTUATORS

the original, linear actuator one step further and incorporated only a single piston strand sequence. This way, the roller strand sequence would dictate the number of pistons in the fully assembled actuator complex by the number of piston complementary regions. A roller strand sequence with three piston complementary regions was thus designed to form the triangular actuator, and a roller strand with four complementary regions was produced for the quadrangular actuator. Initially, Locking strands were designed in accordance with the U-lock system used in the linear actuator design. The locks were made to hybridize with 11 bases on each of two adjacent piston strands in the actuator complex in a given state. Additionally, 10 nt toe-hold segments were added to facilitate the removal of the locks by strand displacement, effectively unlocking the complex. The dynamic locking and unlocking of the quadrangular actuator can be seen in Fig. 4.8. Problems with aggregation of the locked complex, however, were attributed to unwanted interaction between the lock strands. This first led to a shortening of the toe hold region to 5 nt, and subsequently to a change in the lock design. The new lock strands consisted to two strands per locked state, rather than one. While the old locks would hybridize to two piston strands in a U-like motif, the new design provided a strand for each of the piston strands forming simple, straight double strands. The new straight locks can be considered identical to the U-locks cut in two at the cross-over point. Kryger’s theory was that the U-locks were interlinking different actuator complexes, rather than forming the desired U-motif. As reflected by the gel in Fig. 4.9, this change in design eliminated the problem with aggregation. Results for the triangular actuator were similar to those of the quadrangular one, and the straight lock design proved better in both cases.

4.2 Project Aim he aim of the project was to expand the paradigm of the dynamic DNA single tile actuator into larger dynamic systems by linking together two or more actuators. The linkage between the actuators should be designed such that the state of one actuator should dictate the states of all other actuators in the system. This would both give the possibility of creating larger, dynamic DNA nanostructures, and it would make possible the construction of nanomechanical devices where an input signal in the form of a lock strand could propagate from one actuator to another. Combined with the already available three- and four-way DNA tile actuators, this could eventually open the possibilities for the splitting and transduction of a mechanical signal along a complex assembly of actuators.

T

4.1.3. T HREE -

AND

F OUR - WAY DNA ACTUATOR T ILES

71

Figure 4.8. PAGE analysis of the quadrangular actuator dynamically being locked and unlocked with U-locks. Note the partial aggregation of the locked complex. The lanes from left to right begin with a DNA ladder, the piston strand (P), the roller strand (R), and the unlocked actuator complex (PR). This is followed by the complex locked in State 0 (L0), unlocked (LC0), locked in State 5 (L5), unlocked (LC5), locked in State 10 (L10), and finally unlocked (LC10). The illustrations to the right of each gel explain the contents of each band. The unlocking is performed with strand displacement and the formed complex between the lock and unlock strands is seen accumulating at the bottom of the gels.

Figure 4.9. PAGE analysis of the quadrangular actuator dynamically being locked and unlocked with straight locks. The lanes are ordered in the same fasion as in Fig 4.8. The use of straight locks eliminates the aggregation problem.

72

C HAPTER 4. DNA ACTUATORS

4.3 Actuator Design 4.3.1 The Actuator Dimer or the initial proof of concept, it was decided to make a dimer of the original linear actuator. While the original actuator, however, was designed to be symmetric with respect to the moving regions so that the same lock strands were required in both ends, this was not the case for the actuators in the dimer. Additionally, it was the hope that it would be enough to simply lock one end of the actuator dimer and have the locking propagate from one actuator to the other within the dimer. For this, two different actuators had to be designed: one with only one locking region and one with two, and none of the locking regions could be the same. For connecting the two actuators, a third roller strand would be added in between the two actuators, hybridizing to their locking regions. This would both serve the purpose of linking the two actuators together, and propagating the locking state from one actuator to the other. See Fig 4.10 for a schematic illustration of the actuator dimer.

F

Figure 4.10. Schematic illustration of DNA tile actuator dimer locked in a) state 0, b) state 5 and c) state 11. Black signifies regions of the strands with unique base sequences that only hybridize in one way - i.e. static regions, or in the case of black piston termini, locking regions. Colors signify dynamic regions where the roller strand can base pair with each piston region of same color. Vertical lines only illustrate connections, not nucleotides.

In order to avoid tension in the structure, it is important for all roller strands

4.3.1. T HE ACTUATOR D IMER

73

to have a length corresponding to an even number of helical turns. This is because both ends of the strand have to point in the same direction when crossing over from one piston strand to the other. Furthermore, the number of bases should be even since equal amounts of hybridization with the two piston strands is desired. Ultimately, the statically hybridized regions have to be long enough to allow for stable hybridization across the nick in the roller strand, since it is a linear, rather than a circular strand. The original actuator design utilized a roller strand of 64 nucleotides in length, allowing for 21 base pair static hybridization with each piston strand and 2 x 11 base pair dynamic hybridization regions. The dynamic hybridization thus provides the 12 different states of the actuator, and the static region provides good anchoring of the piston strands to the roller strand. Lastly, the length of the static region, provides approximately a full helical turn on each side of the nick in the roller strand, which should suffice in keeping the structure assembled in all the 12 states. It was decided to utilize the same length of the roller strand as in the original setup, since it had already proven successful for the single actuator assembly. It should be mentioned, that in the original paper, the actuator is referred to as having 11 discrete states. This is simply because one of the extreme position, State 11, has not been counted. A roller strand length of 64 nucleotides, however, demands for the number of base pairs per helical turn to be 10.67 if the entire length of the strand is to give an integral number of helical turns. While this has proven an acceptable approximation for the single DNA tile actuator, this number is slightly too high for the dimer. Thus, having three consecutive rollers in the dimer with 64 nucleotides in each might lead to strain and/or an overall twist in the structure. To counter this strain and twist, the length of the connecting roller strand - Roller X in Fig. 4.10 was set to 62 nucleotides. This corresponds to 10.3 base pairs per helical turn, i.e. too few, thus countering the twist from the two actuators. As for the nick in the connecting roller strand, Roller X, this is a possible weak point. Ideally, the nick should be situated far from both dynamic regions and other nicks, e.g. the piston-piston gap, since both may cause failure of hybridization across the nick. In the case of the connecting roller strand, 5 base pairs is the largest distance achievable to fulfill both requirements simultaneously for all locked states 0 through 11. To avoid having the connecting roller strand fold up on itself (according to computational predictions by mfold), 148 the nick had to be placed even closer to piston-piston gap, yielding only a 3-base-pair overlap between (Piston 3 · Roller X) and (Piston 4 · Roller X) (see Fig 4.10). After the roller strands have been designed, the piston strands are almost completely derivative, as per Fig 4.10. The only undefined part that remains is the

74

C HAPTER 4. DNA ACTUATORS

locking regions that are not hybridized to the connecting roller strand. For the locks to hybridize as illustrated in figure 4.10, the crossover from one piston to the other has to be an integral number of helical turns from where the roller strand has its crossover. One full helical turn on each piston strand had proven enough in the original actuator design, and so this was chosen for the dimer as well, approximated to 10 base pairs per helical turn. Opposed to the original actuator, however, a locking region was only added to one end of the dimer, Piston 2 and Piston 4. It was the hope that this would be enough, and that the locking state would propagate to the other end of the dimer through the connecting roller strand.

4.3.2 Generating Base Sequences n our group, generating strand base sequences for specific projects has mainly been accomplished through ad hoc adaptions of MATLAB scripts designed to minimize non-desired complementarities. The scripts, written by Niels V. Voigt, have so far been based on an iterative process with the following steps:

I

• Generating a random base sequences of the desired length. • Identifying any non-desired complementarities. • Assigning a score to the non-desired complementarities based on the estimated strength of the hybridization. • Save the sequence if the resulting score is within the top 20. If not, discard it. • Repeat. While this method has proven effective in prior projects, there has been some doubt as to whether it is the most efficient way. Typically, the script would have to iterate for a while, before it might seem reasonable to assume that the base sequence would not improve much further on additional iterations. Moreover, it was hard to get a good measure on when a base sequence was “good enough”. The number of iteration was mostly a matter of the time it took for the operator of the script to run out of patience after the assigned sequence score appeared to stagnate. Typically, one would then be left with a base sequence that was among the better, and hopefully one that was good enough to serve the purpose. It still left behind the question, whether this was the best possible sequence for the job. For this project it was thus decided to try a different approach. Instead of generating random sequences and trying them against one another, the program

4.3.2. G ENERATING BASE S EQUENCES

75

should attempt to optimize the sequence while it was being generated. The scripting language Python 2.7 was chosen on account of its ease of use as a generalpurpose language, its availability to the public free of charge, and its general popularity. The code for the script used to find base sequences will not be discussed in detail, as this is beyond the aim of this report. However, a brief overview of the used algorithm is provided below. The full source code can be viewed in Appendix B. Since Python is an object oriented language, the script was written as such. This should eventually mean simpler integration with other Python scripts, and that user interfaces can be easily written and later substituted as needed. A quick look through the source code provided in Appendix B reveals that there currently is no user interface. If the script is merely run by itself, nothing will happen. Currently, if one wishes to utilize the script, it has to be loaded into a Python environment from which the user can manually create the objects that generate the strand sequences. When generating base sequences, the script takes the simple approach that it randomly adds one base to the strand at a time. Each time it adds a base, it reads the last x number of bases in the sequence, including the recently added base, and checks whether this sequence snippet or its complementary sequence has already appeared in the entire sequence. If it has, the last added base is removed, else, the base is accepted, the script moves on to the next base in the sequence. The size of x, i.e. the minimum number of bases that are not allowed to repeat in the sequence, is defined by the user at run time. The smaller the number, the less self-complementarity is accepted. However, it should be kept in mind, that the longer the base sequence, the higher x has to be. If x is set to one, each base pair can only be used once, and so the sequence can only be two bases long. If x is set to four, that means that we allow up sequence bits up to three bases long to be identical or complementary to other places in the entire sequence, but any four adjacent bases in the entire sequence have to constitute a unique sequence not found elsewhere in the sequence, nor its complement. The number of unique sequences of length four is 44 /2 = 27 = 128. Since all these unique sequences have to overlap if made into one sequence (except the last three bases at the very end), this gives a theoretically maximum sequence length of 131 bases. This is, of course, assuming that all pieces fit together into one long sequence. Practically, this is, of course, not always the case, depending on the order the pieces are put together. If the requested sequence length is close to the theoretical maximum, the chances that the script runs out of possible solutions increase drastically. In such a case, it is often enough to simply reiterate the process.

76

C HAPTER 4. DNA ACTUATORS

The above description of how the uniqueness of subsequences is assured has a major lack in the way it is described. Assuming that the last three bases added to a sequence is TAG, and assuming that both TAGT, TAGA, TAGG and TAGC are used already, then the program cannot add any more bases to the sequence. To avoid this scenario, a simple step-back mechanism was initially implemented. This allowed the program to delete a few bases at the end of the sequence whenever this happened and try again. While this in itself was not a very elegant solution, it also did not prove very efficient and was quickly abandoned. In its place, another system was devised. The new system simply ascertains that whenever both TAGT, TAGA, TAGG and TAGC are present, the sequence bit TAG is no longer allowed. The entire structure of the program is encapsulated in the Project class, and whenever a new project containing multiple strands is needed, one single Project object is created. New Strand objects can then be added using the Project.newStrand() method. By using this structure, the program itself can keep track of sequences that are already generated and ensure that no unwanted complementarities or similarities between strands accidentally occur. At its current state, the program already provides a means of producing sequences with reduced self-complementarity. There is, however, still the need for improvements. As it is, the program only tests for exact self-complementarities. This means that non-Watson Crick base pairing is not accounted for and that large “almost self-complementary” sections may not be detected. An example of “almost self-complementarity” could be a case where x = 4 and there is a single base pair mismatch for every 3 base pairs. This merely means that any sequence produced by the program should be manually inspected before use. Folding software such as mfold and the newer UNAfold can be used to test for problems as the ones mentioned above. 148 For a project such as designing new sequences for an actuator type structure, the program has another major lack. In our current actuator design there are sequences that need to be identical, and while the program does provide a way of dictating parts of the sequence, the code is not optimized for this. At its current state, the program merely produces a random sequence and subsequently changes the parts that were predefined by the user. This is, obviously, not an efficient approach, but the real problem lies in the transition between random and predefined parts of the sequence. The program currently has no way of testing whether unwanted self-complementarities form in the area where predefined and randomly generated sequences meet each other. This problem does also not seem to have any trivial solution, and manual inspection of the produced sequences is necessary.

4.3.2. G ENERATING BASE S EQUENCES

77

The original goal of the final program was to provide a means to generate base sequences that would suit a large variety of applications. Thus the above mentioned problems need to be solved. Additionally, a graphical user interface still needs to be made. All the sequences used in this project were made using the program mentioned above. The maximum allowed self-complementarity was set to 3 bases length (i.e. x = 4), and between strands it was 4 bases (i.e. x = 5). Pistons 1 and 2 were designed first as a 53- and a 63-mer respectively. Subsequently, Pistons 3 and 4 were designed by concatenating the sequences already dictated by the symmetry of the actuator (colored sections in Fig 1) with newly designed sequences from the Python script. The roller strands and lock strands were designed to be complementary to the piston strands as illustrated, making the sequences derivative of the piston sequences. All strands were tested for self-complementarity using m-fold (online version), 148 and the entire design process was reiterated until any secondary structures predicted by m-fold seemed less energetically unfavorable compared to the desired actuator structures. The actuator dimer schematic locked in state 0 with sequences included is depicted in Fig. 4.11. All the sequences can be found in Appendix C.3. P
 2

P
 1

CTTGTTCGTATGACACTGCGATGCAATTACTCGACGGCCTCAGCACCAGGTTGTTATGCACGTTGGTAATGTACTGACCTCGATCCGACTTCTCATCACCGCCACTCTGTCTTGCT GAACAAGCATACTGTGACGCTACGTTAATGAGCTGCCGGAGTCGTGGTCCAACAATACGTGCAACCATTACATGACTGGAGCTAGGCTGAAGAGTAGTGGCGGTG R 1

R X

R 2

L 0

CGTGACTCCTGCTAAACATCTCTGAGGCCGTCAGACTGGGAGCTATCCAGCGGTACATTACCATTTGTAGCGTCTGATAAGGGAAGTGGCGGTGAGGAGCGAAGG TATGCTTGTTCGCACTGAGGACGATTTGTAGAGACTCCGGCAGTCTGACCCTCGATAGGTCGCCATGTAATGGTAAACATCGCAGACTATTCCCTTCACCGCCACTCCTCGCTTCC P
 3

P
 4

Figure 4.11. Schematic of DNA tile actuator dimer locked in state 0 with DNA sequences included. Black signifies regions of the strands with unique base sequences that only base pair as depicted - i.e. static regions, or in the case of black piston termini, locking regions. Colors signify dynamic regions where the roller strand can base pair with each piston region of same color.

4.4 Results nitial hybridization experiments were aimed to show the successful assembly of the individual actuators along with the complete, unlocked actuator dimer. Subsequently, assembly of the actuator dimer with lock strands was attempted. The purity of the oligonucleotides was confirmed via RP-HPLC prior to the hybridization experiment, and the strands were used as provided by DNA Technolo-

I

78

C HAPTER 4. DNA ACTUATORS

gies A/S.

4.4.1 The Unlocked Actuator Dimer he concentrations of the oligonucleotides were estimated from UV-absorbance measurements and the strands were combined into 15 different intermediate levels of assembly. These ranged from the isolated single strands over the individual actuators to the complete actuator dimer containing all the strands together. The different combinations, as depicted in Table 4.1, underwent the same hybridization conditions and were subsequently analyzed by non-denaturing PAGE. The combinations are devised such that the combinations a–g contained only the strands pertaining to Actuator 1 (Pistons 1 and 2, and Roller 1), while combinations h–n contained only those pertaining to Actuator 2 (Pistons 2 and 4, and Roller 2). The final combination, o, contains both Actuators 1 and 2 along with the connecting roller strand, Roller X, in order to form the entire actuator dimer.

T

Table 4.1. Oligonucleotide combinations for first actuator dimer assembly. Rows represent different combinations of pistons (P) and rollers (R)

P1 a b c d e f g h i j k l m n o

P2

P3

P4

R1

R2

Rx

X X X X X

X X X X

X X

X X

X X X X

X

X X

X

X

X X X

X

X X X X

X

4.4.1. T HE U NLOCKED ACTUATOR D IMER

79

The PAGE analysis of combinations a–g (Actuator 1) along with combination o (actuator dimer) is depicted in Fig. 4.12a. Despite the fact that the lanes a, b and c should only contain a single oligonucleotide each (pistons 1 and 2, and Roller 1, respectively), it is quite evident that lane b has more than one band. In order to assure that the extra bands were not caused by impurities in the sample, all strands were purified by PAGE, and the experiment was repeated under the exact same conditions. No changes were observed (gels not shown). It seems that Piston 3 under hybridizing conditions partially folds up on itself (a possible folding structure can be seen in Fig. C.1 in Appendix C), but the fact that folding is only partial, suggests that the hybridization is not very strong and, hence, may not cause a problem. When looking at lane f, where Piston 3 and Roller 1 are combined, and lane g, where pistons 1 and 3, and Roller 1 are combined to form Actuator 1, an extra, weaker band is showing in both lanes. Since this extra band occurs in all lanes containing Piston 3, it is tempting to conclude that the fault lies with this particular strand. However, if one looks to Fig. 4.12b, which depicts the PAGE analysis of combinations h–o, i.e. the strands for Actuator 2 and the full actuator dimer, lane l contains an extra band similar to that found in lane f. None of the isolated oligonucleotides for Actuator 2, however, form any extra bands. A possible explanation can be found in the actuator design. Both lanes, f and l, contain a piston and a roller strand, where the piston hybridizes to the nicked side of the roller. When the second piston is not present for making the entire actuator, it seems that the roller strand does not hybridize fully with the piston on the nicked side. This does not seem entirely reasonable, since full hybridization of both ends of the roller strands onto the piston would require that the unhybridized part of the roller strand be completely stretched. Some entropic penalty is likely to be associated with such a stretch. Whether this unwarranted opening of the roller strands hybridization takes place in the full actuator assembly is hard to conclude entirely based on Fig. 4.12. There are distinguishable extra bands in both lanes g and n. Some of the extra bands, especially in lane n, seem to be consistent with bands found in other lanes, insinuating that they arise from stoichiometry inconsistencies. Some higher order bands, however, could easily be consistent with actuators opening on the nicked side of the roller strand. The higher order bands, however, are not as intense as in the cases with only one piston strand present. As for the fully assembled actuator dimer in lane o, the assembly seems fairly clean. There is some smear above and a few weak bands below the strongest band, which is likely to be the fully assembled actuator dimer.

80

C HAPTER 4. DNA ACTUATORS

a)

b)

Figure 4.12. PAGE analysis of the strand assemblies found in Table 4.1. Gel a) contains the strands used in assembly of Actuator 1, and gel b) the strands for Actuator 2. Both gels have the fully assembled actuator dimer as lane o on the far right, and a 25 bp ladder on the far left.

4.4.2 Adding Locks hile the assembly of the unlocked actuator dimer showed small impurities, it was decided to attempt the full actuator dimer assembly with lock strands in the hope that the extra rigidity provided by the lock strands might decrease unwanted bands. Like for the piston and roller strands, the concentrations of all lock strands were measured as precisely as possible using light absorption at 260 nm and extinction coefficients found using the nearest neighbor technique. Subsequently, samples of the actuator dimer locked with each of the 12 lock strands along with an unlocked sample were hybridized and analyzed via PAGE. The result can be seen in Fig. 4.13. As can be seen in the gels in Fig. 4.13, the experiment was inconclusive. In order to obtain good separation of the relatively large actuator dimers, the electrophoresis time had been extended, causing the bands to move further into the gel. The hope was to show that the locked and unlocked actuator dimers would elude with different retention on the gel. While it is hard to detect any unambiguous differences in retention between the supposedly locked and the unlocked actuator dimers, the prolonged electrophoresis time may be assumed to also have caused any unhybridized lock strands to have eluded from the gel. Overall, it seems that a fully assembled actuator dimer is still present. All bands in the gel move very similarly to the 250 bp mark in the ladder. Whether

W

4.4.2. A DDING L OCKS

a)

81

b)

Figure 4.13. PAGE analysis of the actuator dimer with and without lock strands. Gel a) contains an unlocked actuator dimer (X) for comparison purposes, locked actuator dimers with locks 0–5 (L0–L5), and a 25-500 bp ladder on the far right. Gel b) has a similar setup, except with lock strands L6–L11.

the actuator dimers are locked at this stage, cannot be concluded from the gels in Fig. 4.13, but conversely that they should remain unlocked in the presence of lock strands can also not be concluded. It is entirely possible that the addition of a 20 base long oligonucleotide to a structure containing a total 422 bases divided between 7 oligonucleotides would not cause any significant change in retention on the gel. The total charge increases with the lock strands, but the terminus of the actuator dimer also changes from a relatively floppy single stranded to a more rigid double strand region, which may equally well cause a change in retention. A change in the lock design was required for better separation. Following the theory that the locks were simply too small, hairpin sequences of 10 bp length with a “hinge” of 4 thymidines were introduced to simply add more mass. Additionally, Mille Kryger’s positive results with two straight lock strands in place of a single U-lock strand, inspired the adoption of the straight locks in the actuator dimer project as well. Mille Kryger had found that the straight locks generally provided better hybridization, and an additional feature of these was that a larger portion of the piston strands could be hybridized, since this was no longer limited by the geometric needs for cross-over points. The actuator dimer locked in State 0 with the new straight locks is illustrated in Fig. 4.14 below, where the red region of the strands denote a 5 nucleotide toehold section. This was incorporated in the new design to accommodate for the eventual need for strand displacement when unlocking the actuator dimer.

82

C HAPTER 4. DNA ACTUATORS

Figure 4.14. Actuator dimer locked in State 0 with straight locks. The red regions on the lock strands signify 5 nt toeholds for strand displacement.

The first attempts at locking the actuator dimer with these new straight locks caused cautious optimism. As seen in Fig. 4.15, the locked and unlocked actuator dimers did not separate convincingly, but with the eyes of an optimist, an observable separation was apparent. In an attempt to increase separation, the electrophoresis time was increased from 2 to 4 hours, but to no avail. The extra two hours only seemed to increase smear of the bands, and not separation (gel images are not shown here).

Figure 4.15. The first attempt at locking the actuator dimer with straight locks. The contain a) Locked state 0, b) unlocked, c) locked state 5, d) unlocked, e) locked state 10, and f) 25-500 bp ladder.

Generally when attempting to separate DNA strands of similar sizes, increasing the polyacrylamide percentage of the gel often produces the desired separation. In this particular case, however, we reflected that the actuator dimer complex would have trouble moving through the gel if it were made denser than 7.5 %. Instead, it was decided to increase the difference between the locked and unlocked state even further by attachment of streptavidin to the locked state. One of the straight locks (Lock 0a) was synthesized with a biotin modification in the 5’ end, and the actuator dimer was locked with this strand (along with the unmodified Lock 0b). The resulting PAGE analysis can be seen in Fig. 4.16 where the unlocked actuator dimer is compared to two versions of the locked actuator: one is only locked with the biotin modified lock, the other has the biotin modified lock and has been added streptavidin. Since streptavidin is a rather large protein (approx. 60 kDa) that binds strongly to biotin, the polyacrylamide percentage of the gel

4.4.3. FRET

83

was reduced to 6 % to ensure that the streptavidin bound complex would be able to migrate in the gel. As anticipated, the streptavidin bound complexes migrated slower on the gel and separated nicely from the unlocked complex. Interestingly, the 6% gel proved to increase separation even in absence of streptavidin. This could either be due to the biotin conjugated to the lock strand, or it could be a result of lowering the polyacrylamide percentage to 6%. Both PAGE analyses below in Fig. 4.17 were performed in 6% polyacrylamide. One is of the actuator mono- and dimer locked with unmodified lock strands, while the other is performed utilizing biotin-modified lock strands. From these two gels, it can be concluded that the straight locks hybridize well with both the actuator dimer and monomer. Additionally, it can be seen that the unmodified actuators separate better on the 6% gel, and that with biotin-modifications, the separation is even better. It can thus be concluded that both the lowering of the gel percentage and the biotin modification contribute to good separation between the locked and unlocked states. Since the original problem with U-locks had been that PAGE analysis could not distinguish between the locked and the unlocked actuator dimer, the experiment was repeated with biotin modified U-locks. The actuator dimer was locked in the states 0, 5 and 10 with biotin modified U-locks and treated with streptavidin prior to PAGE analysis. The result is shown in Fig. 4.18, and it is very clear that the band containing the unlocked complex disappears completely. This leaves no doubt that the U-locks hybridize successfully with the actuator dimer.

4.4.3 FRET Having evidence that lock strand hybridization was successful, the next logical step was to move on to the F¨orster Resonance Energy Transfer (FRET) experiments to test for propagation of the locking from one actuator unit to the other. Piston strands 1 and 2 were acquired in a version with an amino-linker modified nucleobase on each, situated in such a way that the two modifications would be closest in locked state 0 and furthest apart in state 11. The fluorophore Cy3B was conjugated to an amino-modified Piston 1 by a normal NHS coupling reaction, and Cy5 was conjugated to Piston 3 in the same manner. At this point, there was doubt about the locking abilities of the straight locks. Since these only hybridize to one piston strand each in the desired state, the complex is comparable to a Holliday junction where one strand has been nicked at the point of the junction. It is thus only the energy penalty of creating a cross-over that makes the desired state favorable energetically. The U-locks, however, also penalize deviations from the desired state by strain or eventually breeches in hy-

84

C HAPTER 4. DNA ACTUATORS

Figure 4.16. PAGE analysis of the actuator mono- and dimer locked with biotin-modified lock strands and added streptavidin. Lanes from left to right contain: a) unlocked actuator monomer, b) locked monomer, c) locked monomer with streptavidin, d) lock strand, e) lock strand with streptavidin, f) unlocked actuator dimer, g) locked dimer, h) locked dimer with streptavidin, and finally i) a 100-1000 bp ladder on the far right.

a)

b)

Figure 4.17. PAGE analysis (6% polyacrylamide) of the actuator mono- and dimer locked with biotin-modified straight lock strands (a) and unmodified straight lock strands (b). Lanes from left to right on both gels contain: a) state 0 locked monomer, b) unlocked monomer, c) state 5 locked monomer, d) unlocked monomer, e) state 10 locked monomer, f) state 0 locked dimer, g) unlocked dimer, h) state 5 locked dimer, i) unlocked dimer, j) and state 10 locked dimer.

Figure 4.18. PAGE analysis (6% polyacrylamide) of the actuator dimer locked with biotin-modified U-lock strands and then added streptavidin. Lanes from left to right contain: state 0 locked dimer (a), unlocked dimer (b), state 5 locked dimer (c), unlocked dimer (d), and state 10 locked dimer (e). The last lane (f) contains a 25-500 bp ladder.

4.4.3. FRET

85

Figure 4.19. FRET measurements of actuator dimer locked in 12 discrete states. Black data points represent FRET relative to Cy5 emission when Cy5 is excited directly. Blue data points represent FRET relative to the sum of Cy3B and Cy5 emission upon Cy3B excitation.

bridization in the cross-over region of the lock. It was thus decided to return to the U-lock design for the FRET experiments. The actuator dimer was assembled like before but with the fluorophore labeled piston strands. It was locked with U-locks in each of the 12 possible states, and the FRET was measured for each of the states. The relative FRET measurements are depicted in Fig 4.19. In Fig. 4.19, the measured FRET is given as the fraction of detected fluorescence emission from Cy5, the acceptor fluorophore. This fraction can either be defined relative to the maximum possible acceptor emission or relative to total emission detected, i.e. the sum of donor and acceptor emission during excitation of the donor. In the first case, the maximum possible acceptor emission is simply measured by exciting the acceptor directly and measuring the fluorescence. This definition is preferred if the donor fluorophore - in this case Cy3B - is in excess, since the acceptor (Cy5) then is the limiting factor. Conversely, if acceptor is in excess, this definition will be less practical because it will be a lot more sensitive to small changes of acceptor concentration between samples effectively lowering signal-to-noise ratio. In this case, defining FRET relative to the sum of donor and acceptor emission is preferred, because donor concentration then is the limiting factor. As can be seen in Fig. 4.19, both definitions yield a similar result indicating

86

C HAPTER 4. DNA ACTUATORS

that donor and acceptor are present in equal amounts. The FRET results in Fig. 4.19 support the hypothesis that locking of the actuator dimer in one end causes locking of the entire complex in the given state. As expected, the overall FRET is reduced from state 0 to state 11 as the distance between donor and acceptor is increased. Due to the helical nature of the DNA, the distance between the fluorophores is not only increased linearly, but also gains a sinusoidal component to their relative movement. This characteristic is clearly reflected in the measurements as well. The geometry of the complex forces an integral number of 360° turns between the two cross-overs of the roller strand. In state 0, the two fluorophores are placed a distance of 16 and 17 nucleotides from each of the cross-over points. This places them at an approximate angle of 180° relative to what their position would be at the cross-over points. Since this placement of the fluorophores was also used in the original, linear actuator, the FRET values are almost directly comparable. The published FRET data from the original actuator in Fig. 4.5b follows a curve very similar in shape to the data collected from the actuator dimer. The theoretically calculated FRET signal in Fig. 4.5c confirms that this is the curve shape expected for this setup, as well. However, the changes in FRET signal are very small for both the actuator dimer and Zhang’s actuator monomer compared to the theoretical data. The difference in FRET signal between the state with the highest and the one with the lowest signal is only about 0.065 in the case of the actuator dimer and only 0.14 for Zhang’s monomer. For comparison, Zhang’s actuator monomer where the fluorophores were placed as illustrated in Fig. 4.5d had FRET signals spanning approximately 0.40. This was a point of concern. A possible explanation to this very small change in FRET signal in the case of the actuator dimer could be that, while the signal change that we do see correlates well with the actuator dimer locking in the desired states, the majority of the bulk does not have the fluorophore-labeled part of the dimer locked. This could be because the initial locking does not occur efficiently, or it could be that the locking does not propagate from the one actuator to the other in the dimer in the bulk of the solution. A minor population of fully functioning locked dimers then provide the small observed signal change. A weakness in the dimer design lies in the linking of the two actuators by a roller that is not circular. As mentioned in the discussion on the design, the nick in the roller strand means that the system relies on four terminal nucleobases on the Roller X 5’ terminus to hybridize with the four nucleobases on the 5’ terminus of Piston 3. If these four base pairs break, the complex will still be held together so a change may not be visible in PAGE analysis, but the locking signal would no longer be able to propagate across the Roller X connection. The observed changes in FRET signal would come from the

4.4.3. FRET

87

population of properly hybridized actuator dimers and the “broken” dimers would produce a constant average FRET signal, since the distribution of states would not change. In an attempt to reduce the enthropic energy gained by breaking hybridization between the four base pairs, the FRET experiment was repeated at a lower temperature of 4°C. The FRET profile depicted in Fig. 4.20 is not very different from the original FRET that was measured at 25°C. However, the difference between lowest and the highest value has increased from 0.06 to 0.09.

Figure 4.20. FRET signal of the actuator in 12 discrete states measured at 4°C. The same samples were reused from the measurements in Fig. 4.19. Black data points represent FRET calculated relative to Cy5 emission, blue data points relative to the sum of Cy3B and Cy5 emission.

This seems to indicate that temperature does have an effect on the FRET efficiency. This, of course, does not necessarily mean that the above hypothesis holds true. If the above explanation was indeed correct, the ideal solution would be to substitute Roller X with a circular version. At this point, however, time was running short, and ligation to make circular strands had so far proven unreliable at best in our group. Instead, two new Roller X substitutes were bought, where the nick had been moved to two new positions. In both cases, the four base pair overlap expected to have been the problem was increased to 10 base pairs. This was the maximum number of bases possible the way the actuator dimer was designed and it was expected to be sufficient. The possible disadvantage to the approach was that for one of the new Roller X strands, the nick would be right where the roller strand crosses over, when the dimer is locked in state 0. For the other new

88

C HAPTER 4. DNA ACTUATORS

Roller X, it would happen in state 11. The hypothesis was that if this design caused problems, such as the actuator dimer having a tendency to seek towards states 0 and 11 for each of the new rollers, respectively, the problems were most likely to only be significant in states close to 0 and 11.

Figure 4.21. FRET signal of the actuator in 12 discrete states with two different Roller X strands.

The FRET data in Fig. 4.21 adhere well to the theoretical curve in Fig. 4.22. They do, however, look different from FRET profiles in Fig. 4.19 and Fig. 4.20 in that the linear contribution seems to be more pronounced than the sinusoid contribution. It is hard to say why this is, however, it would seem that the nick placement in Roller X affects the geometry of the actuators. If, for instance, the distance between piston strands was increased, the rotation of the fluorophores around the pistons would contribute less to the change in FRET than the linear increase in distance along the actuator. With respect to the risk of the actuator dimer obtaining a bias favoring the states where the nick in Roller X is situated in the cross-over, it does not seem to be a problem. If there had been a ubiquitous population of actuator dimers in this state, their contributions to the background FRET would have been very different for the case where State 0 was favored compare to where State 11 was favored. As seen in Fig. 4.21, the two FRET profiles have very similar values. It could be speculated that there is a small bias apparent, since the FRET signal for the State 0 favoring Roller X is a little higher than in the case where State 11 is expected to be favored. However, the difference is very small and more data would be needed to ascertain any statistical significance.

4.4.3. FRET

89

From the FRET data in Fig. 4.21, however, it is apparent that the change in signal has not been improved compared to measurements with the original Roller X. In other words, the hypothesis that the problem lay with the nick placement in Roller X does not seem valid. In an attempt to identify different factors’ effect on the FRET data, a simplified model of the system was constructed. The theoretical FRET values were estimated by simply calculating the distance between the fluorophores in an idealized system where the fluorophores rotate 34.4°/state around each their piston strand while distance between them increases linearly along the axis defined by the actuator dimer. Fig. 4.22 illustrates the theoretically expected FRET signal. The phase of the sinusoidal component seems to fit the experimental data best if the fluorophores extend from the helix with an approximate angle of 150° relative to the backbone of the complementary strand, i.e. the roller strand. The linear character of the graph predominates in cases where the rotational radius is small compared to the distance increase along the actuator axis. Additionally, any constant fluorescence contribution, i.e. one that does not change as a function of state, will dampen all changes in FRET signal, linear as well as sinusoidal. The data points in Fig. 4.22 have been calculated with a 1.5 nm rotational diameter for the fluorophores. This has by no means been optimized to fit the data, however, it seems a reasonable diameter for relatively large molecules like Cy3B and Cy5 with 12 atom long linker to the nucleobase attachment point. The FRET values were calculated from the theoretical relation between donor/acceptor distance and FRET signal: E = 1 orster distance of 1+(r/R0 )6 , where r is the donor/acceptor distance and R0 is the F¨ a particular donor/acceptor pair, where half the energy is transferred. The values used in Fig. 4.22 are FRET data calculated for R0 = 4.9 nm. This value is taken from Zhang’s theoretical calculations. 132 Obviously, the model in Fig. 4.22 does not fit the experimental data perfectly, but for a crude model that does not take into account any quantum mechanical effects or deviations from the idealized shape of the actuator, such as electrostatic repulsion between the hybridized pistons, the similarity is good. Good enough to conclude that locking the actuator dimer in one end causes the other end to be locked as well. As in Zhang’s theoretical model in Fig. 4.5c, this model also suggests that a much larger difference between the minimum and maximum FRET values is to be expected. Reading the actuator article by Zhang et al., the F¨orster distance reported for use with the Cy3B donor is a little inconsistent. In the article it is stated to be 4.9 nm, while the supporting material places it at 6.8 nm. The F¨orster distance, R0 , is constant defined as the distance between the fluo-

90

C HAPTER 4. DNA ACTUATORS

Figure 4.22. Theoretical FRET solely deduced from the geometry of the system.

rophores where the energy transfer is exactly half of the possible maximum. This constant is dependent on four factors: the quantum yield of the fluorophore donor (ϕD ), the overlap integral between the donor emission and acceptor absorbance spectra (JDA ), the refractive index of the solution (n), and the donor-acceptor orientation factor (κ). These four factors define the F¨orster distance through the following relation. 149 1

R0 = 8.79 · 10−28 · (JDA n−4 κ2 ϕD ) 6

Assuming that the fluorophores have no rotational constraints, which Zhang et al. showed was the case for the similar actuator monomer, κ2 has a value of 2/3. The refractive index, n, was measured by Zhang et al. to be 1.35 for the same buffer used for the actuator dimer. 132 The quantum yield for Cy3B is reported to be ϕD = 0.67 and have an approximately identical emission spectrum as that of Cy3. 150 Following this approximation, the overlap integral between Cy3B and Cy5 is approximated to be identical to that of Cy3 and Cy5. This gives JDA = 7.2 · 10−13 M-1 cm3 . 151 These values give a F¨orster distance of 6.6 nm. This makes it more plausible that the F¨orster distance of 6.8 nm was the correct one, rather than the 4.9 nm. In Fig. 4.23 the theoretical FRET efficiency is calculated with a F¨orster distance of 6.8 nm and the fluorophore radius reduced slightly to 1 nm. These two alterations alone reduce the expected differences between minimum and maximum FRET values to approximately 0.10. The reason for this is simply that the quantum yields of Cy3B is very high.

4.4.3. FRET

91

1 F


0 0,

0,

00,

0,

S  0









10

1

Figure 4.23. Theoretical FRET solely deduced from the geometry of the system.

For comparison, Alexa 555 has a quantum yield of 0.10, which with the other factors above unaltered gives a F¨orster distance of 4.8 nm. 152 In other words, the Cy3B is simply shining too bright for the relatively small distances in the actuator. This also explains why the FRET data for the actuator monomer collected with Alexa-fluorophores had a lot larger difference between the minimum and maximum FRET efficiencies.

4.5 Conclusion and Perspective wo asymmetrical linear actuators similar to that originally published by Zhang et al. was successfully designed and assembled, and the two were linked together by adding a third roller strand. The actuator dimer was designed with a locking region only in one end, and locking of the complex was explored both utilizing locking strands that take on a U-shape in the locked complex (U-locks), and locking strands that were not designed to make cross-overs (straight locks). Both types of lock strands were shown to effectively hybridize with the actuator dimer. In order to verify the locks capability for locking the actuator dimer in the twelve possible discrete states, the actuator monomer in the end opposite the locking region was equipped with a set of fluorophores for distinguishing the twelve state by FRET measurements. The fluorophore labeled actuator dimer was locked in the twelve states, and FRET was measured for each state. The FRET efficiencies plotted as a function of the state followed the seagull shaped curve known from the published actuator monomer. This reflected the geometry of the system well, however, the differences in FRET efficiency between the states that placed

T

92

C HAPTER 4. DNA ACTUATORS

the fluorophores closest and furthest apart, respectively, was very small. In some of the FRET data published with the actuator monomer, the changes in FRET efficiencies was comparatively small. This indicates that the problem does not lie particularly with the dimer design. Since the comparably low published FRET efficiencies only were seen in the data collected when the fluorophore placement was identical to one used with the actuator dimer, the problem might pertain to this particular setup. In this case, the fluorophores were situated exactly opposite each other on the piston strands in State 0, and the distance between them increased along with the state number. In the published FRET data for the actuator monomer, the changes in FRET efficiency increased significantly when the fluorophores were situated such that they were exactly opposite each other in State 5. It should be mentioned that the fluorophores in this case were Alexa555 and Alexa647, while the other setup had Cy3B and Alexa647. The FRET data for the actuator dimer were obtained with the fluorophores Cy3B and Cy5. This might also imply a problem with FRET from a the Cy3B donors. The collected data do not reflect quenching or bleaching problems with either the Cy3B nor the Cy5, but the very strong quantum efficiency of the Cy3B may be the cause of the problem. The high quantum yield simply makes the Cy3B too bright for the distances of the actuator. Obviously, the possibility should be considered that the actuator dimer does not have proper signal propagation from the locking region on the one monomer to the fluorophore carrying region on the other monomer. This would give a strong, unchanging FRET background signal, which would diminish the size of the signal from the actuator dimers that do work as intended. If this is the case, however, data presented here seem to indicate that the problem does not lie with the interlinking roller strand. Additionally, the fact that the published monomer FRET data exhibit the same problem indicate that the problem lies with the actuator monomers and not with the dimerisation. Finally, since the problem seems to be dependent on the placement and choice of the fluorophores, as mentioned above, it seems improbable that these correlate with the efficiency of the lock strands in fixing the complex geometry. In uncovering the above questions, there are some simple experiments that might help shed some light. While the nicks in the roller strands are unlikely to be any major cause of problems, if ligation can be accomplished simply and efficiently, it would eliminate any doubt. Ligation of piston strands post actuator dimer assembly would also provide an additional element of stability and might especially be feasible when larger assemblies of multiple actuators is considered. However, with respect to investigating the low FRET efficiencies, an experiment that would seem more eminent would be repeating the FRET measurements with

4.4.3. FRET

93

Alexa555 and Alexa647 as the conjugated fluorophores. It would also be interesting to conjugate a fluorophore FRET pair on the locking end of the actuator dimer. Comparing the FRET data collected from each end of the actuator dimer would reflect any problems conveying the locking signal from one end to the other. The next chapter in the actuator story should be the further investigation of interlinking already existing actuator monomers. Expanding from the dimer to a simple oligo- or polymer of actuators is a simple first step that we have considered. By conjugating either a FRET pair or a fluorophore/quencher pair on each of the actuator monomers, the input lock strand would thus produce a strongly amplified output signal compared to conventional systems where only one chromophore pair is affected per input strand. If pursued, the actuator system could come to play a significant role in future DNA controlled nanomechanics.

C HAPTER 5

E PILOGUE

hree different projects have been described in this report, all in the field of DNA nanotechnology. The three projects have provided three very different contributions to the field, and all have the potential to grow. The project concerning immobilization of DNA via triazenes resulted in a ferrocene containing triazene linker. This linker not only made it possible to immobilize a DNA strand to an electrode surface, the immobilization also directly resulted in a biosensor that enables the detection of the presence of complementary DNA strands. A key feature of the triazene linker is the versatility of the triazene group, which enables both gold and glassy carbon electrodes and possibly more for simple immobilization. Other possible applications of the triazene in DNA chemistry was briefly touched upon but to no extend fully explored. The transfer of helix chirality project proved how conjugation of two fluorescein molecules to the same DNA double helix resulted in the chiral exciton coupling between the fluoresceins. Additionally, the fluorescein exciton in one of the three tested cases had distinctly different CD spectra for the DNA on the B- and the Z-form. This proved that the exciton chirality originates in the DNA secondary structure. The fluorescein exciton was also detectable by fluorescence detected CD, a far more sensitive technique. The project leads the way both for studies of secondary DNA structures on very specific sites while still in solution, and for the use of helix chirality in DNA directed chemistry. Additionally, it may prove a new starting point for the pursuit of negative refractive index materials. In the final project, the concept from the original linear actuator with 11 discrete states by Zhang et al. was used in designing two asymmetric actuators which were interlinked by a roller strand to form a dimer. This approach proved to produce a dimer where locking one end of the dimer with lock strands successfully

T

95

96

R EFERENCES

fixed the entire actuator in a specific geometry. This has provided a proof of concept for the possibility of producing larger mechanical systems based on assembly of multiple actuators. The triangular and quadrangular actuators produced by Kryger creates a good foundation for producing these larger mechanical systems, where a single set of input lock strands can produce a signal that propagates throughout the entire complex. On an interesting side note, the FRET measurements for the single discrete states provide data on the secondary structure for the actuators. Since the actuator profile is very like those used in many other DNA nanostructures, this can prove an interesting element by itself.

R EFERENCES

[1] Watson, J. D.; Crick, F. H. Nature 1953, 171, 737–738. [2] Harper, D. http://www.etymonline.com/. [3] Berg, J. M.; Tymoczko, J. L.; Stryer, L. Biochemistry, 5th ed.; W. H. Freeman and Company, New York, USA, 2003. [4] http://www.merriam-webster.com/dictionary/nanotechnology. [5] http://seemanlab4.chem.nyu.edu/. [6] Seeman, N. C. Nature 2003, 421, 427–431. [7] Zheng, J.; Birktoft, J. J.; Chen, Y.; Wang, T.; Sha, R.; Constantinou, P. E.; Ginell, S. L.; Mao, C.; Seeman, N. C. Nature 2009, 461, 74–77. [8] Wang, Y.; Mueller, J. E.; Kemper, B.; Seeman, N. C. Biochemistry 1991, 30, 5667–5674. [9] Chen, J.; Seeman, N. C. Nature 1991, 350, 631–633. [10] Zhang, Y.; Seeman, N. C. J. Am. Chem. Soc. 1994, 116, 1661–1669. [11] Fu, T. J.; Seeman, N. C. Biochemistry 1993, 32, 3211–3220. [12] Winfree, E.; Liu, F.; Wenzler, L. A.; Seeman, N. C. Nature 1998, 394, 539–544. [13] Rothemund, P. W. K.; Papadakis, N.; Winfree, E. PLoS Biol. 2004, 2, 2041– 2053. 97

98

R EFERENCES

[14] Mao, C.; Sun, W.; Seeman, N. C. J. Am. Chem. Soc. 1999, 121, 5437–5443. [15] LaBean, T. H.; Yan, H.; Kopatsch, J.; Liu, F.; Winfree, E.; Reif, J. H.; Seeman, N. C. J. Am. Chem. Soc. 2000, 122, 1848–1860. [16] Mao, C.; LaBean, T. H.; Reif, J. H.; Seeman, N. C. Nature 2000, 407, 493–496. [17] Shih, W. M.; Quispe, J. D.; Joyce, G. F. Nature 2004, 427, 618–621. [18] Rothemund, P. W. K. Nature 2006, 440, 297–302. [19] Ke, Y.; Lindsay, S.; Chang, Y.; Liu, Y.; Yan, H. Science 2008, 319, 180– 183. [20] Voigt, N. V.; Tørring, T.; Rotaru, A.; Jacobsen, M. F.; Ravnsbæk, J. B.; Subramani, R.; Mamdouh, W.; Kjems, J.; Mokhir, A.; Besenbacher, F.; Gothelf, K. V. Nat. Nano 2010, 5, 200–203. [21] Helmig, S.; Rotaru, A.; Arian, D.; Kovbasyuk, L.; Arnbjerg, J.; Ogilby, P. R.; Kjems, J.; Mokhir, A.; Besenbacher, F.; Gothelf, K. V. ACS Nano 2010, 4, 7475–7480. [22] Andersen, E. S.; Dong, M.; Nielsen, M. M.; Jahn, K.; Subramani, R.; Mamdouh, W.; Golas, M. M.; Sander, B.; Stark, H.; Oliveira, C. L. P.; Pedersen, J. S.; Birkedal, V.; Besenbacher, F.; Gothelf, K. V.; Kjems, J. Nature 2009, 459, 73–76. [23] Ke, Y.; Sharma, J.; Liu, M.; Jahn, K.; Liu, Y.; Yan, H. Nano Lett. 2009, 9, 2445–2447. [24] Douglas, S. M.; Dietz, H.; Liedl, T.; H¨ogberg, B.; Graf, F.; Shih, W. M. Nature 2009, 459, 414–418. [25] http://en.dogeno.us/2009/10/william-shih-talks-about-how-to-buildnanoscale-molecular-parts-using-dna-double-helix-with-the-simple-atgcpairing-rule/. [26] Han, D.; Pal, S.; Nangreave, J.; Deng, Z.; Liu, Y.; Yan, H. Science 2011, 332, 342–346. [27] Naylor, R.; Gilham, P. T. Biochemistry 1966, 5, 2722–2728. [28] Inoue, T.; Joyce, G. F.; Grzeskowiak, K.; Orgel, L. E.; Brown, J. M.; Reese, C. B. J. Mol. Biol. 1984, 178, 669–676.

R EFERENCES

99

[29] Goodwin, J. T.; Lynn, D. G. J. Am. Chem. Soc. 1992, 114, 9197–9198. [30] Gartner, Z. J.; Liu, D. R. J. Am. Chem. Soc. 2001, 123, 6961–6963. [31] Gartner, Z. J.; Kanan, M. W.; Liu, D. R. Angew. Chem., Int. Ed. 2002, 41, 1796–1800. [32] Gartner, Z. J.; Kanan, M. W.; Liu, D. R. J. Am. Chem. Soc. 2002, 124, 10304–10306. [33] Hansen, M. H.; Blakskjær, P.; Petersen, L. K.; Hansen, T. H.; Højfeldt, J. W.; Gothelf, K. V.; Hansen, N. J. V. J. Am. Chem. Soc. 2009, 131, 1322–1327. [34] Gooding, J. Electroanal. 2008, 20, 573–582. [35] Fan, C.; Plaxco, K. W.; Heeger, A. J. Trends Biotechnol. 2005, 23, 186–192. [36] Fan, C.; Plaxco, K. W.; Heeger, A. J. Proc. Natl. Acad. Sci., USA 2003, 100, 9134–9137. [37] Boon, E. M.; Ceres, D. M.; Drummond, T. G.; Hill, M. G.; Barton, J. K. Nat. Biotech. 2000, 18, 1096–1100. [38] Drummond, T. G.; Hill, M. G.; Barton, J. K. Nat. Biotech. 2003, 21, 1192– 1199. [39] Halliwell, C. M.; Cass, A. E. G. Anal. Chem. 2001, 73, 2476–2483. [40] Effenberger, F.; G¨otz, G.; Bidlingmaier, B.; Wezstein, M. Angew. Chem., Int. Ed. 1998, 37, 2462–2464. [41] Yang, W.; Auciello, O.; Butler, J. E.; Cai, W.; Carlisle, J. A.; Gerbi, J. E.; Gruen, D. M.; Knickerbocker, T.; Lasseter, T. L.; Russell, J. N.; Smith, L. M.; Hamers, R. J. Nat. Mater. 2002, 1, 253–257. [42] Nuzzo, R. G.; Allara, D. L. J. Am. Chem. Soc. 1983, 105, 4481–4483. [43] Porter, M. D.; Bright, T. B.; Allara, D. L.; Chidsey, C. E. D. J. Am. Chem. Soc. 1987, 109, 3559–3568. [44] Dubois, L. H.; Nuzzo, R. G. Annu. Rev. Phys. Chem. 1992, 43, 437–467. [45] McMurry, J. Fundamentals of Organic Chemistry, 4th ed.; Brooks/Cole Publishing Company, 1998.

100

R EFERENCES

[46] Pinson, J.; Podvorica, F. Chem. Soc. Rev. 2005, 34, 429–439. [47] Delamar, M.; Hitmi, R.; Pinson, J.; Sav´eant, J. M. J. Am. Chem. Soc. 1992, 114, 5883–5884. [48] Allongue, P.; Delamar, M.; Desbat, B.; Fagebaume, O.; Hitmi, R.; Pinson, J.; Sav´eant, J.-M. J. Am. Chem. Soc. 1997, 119, 201–207. [49] Kariuki, J. K.; McDermott, M. T. Langmuir 2001, 17, 5947–5951. [50] Combellas, C.; Kanoufi, F.; Mazouzi, D.; Thi´ebault, A.; Bertrand, P.; M´edard, N. Polymer 2003, 44, 19–24. [51] Delamar, M.; D´esarmot, G.; Fagebaume, O.; Hitmi, R.; Pinson, J.; Sav´eant, J. M. Carbon 1997, 35, 801–807. [52] Dyke, C. A.; Tour, J. M. Nano Lett. 2003, 3, 1215–1218. [53] Wang, J.; Firestone, M. A.; Auciello, O.; Carlisle, J. A. Langmuir 2004, 20, 11450–11456. [54] de Villeneuve, C. H.; Pinson, J.; Bernard, M. C.; Allongue, P. J. Phys. Chem. B 1997, 101, 2415–2420. [55] Stewart, M. P.; Maya, F.; Kosynkin, D. V.; Dirk, S. M.; Stapleton, J. J.; McGuiness, C. L.; Allara, D. L.; Tour, J. M. J. Am. Chem. Soc. 2004, 126, 370–378. [56] Adenier, A.; Bernard, M.-C.; Chehimi, M. M.; Cabet-Deliry, E.; Desbat, B.; Fagebaume, O.; Pinson, J.; Podvorica, F. J. Am. Chem. Soc. 2001, 123, 4541–4549. [57] Bernard, M.-C.; Chauss´e, A.; Cabet-Deliry, E.; Chehimi, M. M.; Pinson, J.; Podvorica, F.; Vautrin-Ul, C. Chem. Mater. 2003, 15, 3450–3462. [58] Griess, P. J. Chem. Soc. 1865, 18, 268–272. [59] Kosynkin, D. V.; Tour, J. M. Org. Lett. 2001, 3, 993–995. [60] Harper, J. C.; Polsky, R.; Wheeler, D. R.; Brozik, S. M. Langmuir 2008, 24, 2206–2211. [61] Lee, C.-S.; Baker, S. E.; Marcus, M. S.; Yang, W.; Eriksson, M. A.; Hamers, R. J. Nano Lett. 2004, 4, 1713–1716.

R EFERENCES

101

[62] Dolan, P. L.; Wu, Y.; Ista, L. K.; Metzenberg, R. L.; Nelson, M. A.; Lopez, G. P. Nucl. Acids Res. 2001, 29, e107. [63] Wallach, O. Liebigs Ann. Chem. 1886, 235, 233. [64] Wallach, O. Liebigs Ann. Chem. 1888, 243, 219. [65] Kimball, D. B.; Haley, M. M. Angew. Chem., Int. Ed. 2002, 41, 3338–3351. [66] Trost, B. M.; Pearson, W. H. J. Am. Chem. Soc. 1981, 103, 2483–2485. [67] Trost, B. M.; Pearson, W. H. J. Am. Chem. Soc. 1983, 105, 1054–1056. [68] Chen, B.; Flatt, A. K.; Jian, H.; Hudson, J. L.; Tour, J. M. Chem. Mater. 2005, 17, 4832–4836. [69] Lu, M.; Chen, B.; He, T.; Li, Y.; Tour, J. M. Chem. Mater. 2007, 19, 4447– 4453. [70] Hudson, J. L.; Jian, H.; Leonard, A. D.; Stephenson, J. J.; Tour, J. M. Chem. Mater. 2006, 18, 2766–2770. [71] Knudsen, C. S. Syntese af nukleobase analoger til test for deres anticancereffekt og immobilisering ad DNA p˚a overflader. M.Sc. thesis, University of Aarhus, Denmark, 2007. [72] Chen, M. Conjugation of Reporter Groups and Aryltriazenes to Peptides and Oligonucleotides. Ph.D. thesis, University of Aarhus, Denmark, 2008. [73] Pedano, M. L.; Rivas, G. A. Sensors 2005, 5, 424–447. [74] Bard, A. J.; Faulkner, L. R. Electrochemical Methods – Fundamentals and Applications, 2nd ed.; Wiley: New York, 1980. [75] Petrovykh, D. Y.; Kimura-Suda, H.; Whitman, L. J.; Tarlov, M. J. J. Am. Chem. Soc. 2003, 125, 5219–5226. [76] Rayner-Canham, G. Descriptive Inorganic Chemistry, 2nd ed.; W. H. Freeman and Company, New York, USA, 2000. [77] Hansen, M. N.; Farjami, E.; Kristiansen, M.; Clima, L.; Pedersen, S. U.; Daasbjerg, K.; Ferapontova, E. E.; Gothelf, K. V. J. Org. Chem. 2010, 75, 2474–2481. [78] Gross, M. L.; Blank, D. H.; Welch, W. M. J. Org. Chem. 1993, 58, 2104– 2109.

102

R EFERENCES

[79] Sandmeyer, T. Ber. Dtsch. Chem. Ges. 1884, 17, 1633–1635. [80] Sandmeyer, T. Ber. Dtsch. Chem. Ges. 1884, 17, 2650–2653. [81] Lazny, R.; Poplawski, J.; K¨obberling, J.; Enders, D.; Br¨ase, S. Synlett 1999, 1999, 1304–1306. [82] Lazny, R.; Sienkiewicz, M.; Br¨ase, S. Tetrahedron 2001, 57, 5825–5832. [83] Br¨ase, S.; K¨obberling, J.; Enders, D.; Lazny, R.; Wang, M.; Brandtner, S. Tetrahedron Lett. 1999, 40, 2105–2108. [84] Br¨ase, S.; Dahmen, S.; Pfefferkorn, M. J. Comb. Chem. 2000, 2, 710–715. [85] Dahmen, S.; Br¨ase, S. Angew. Chem., Int. Ed. 2000, 39, 3681–3683. [86] Dahmen, S.; Br¨ase, S. Org. Lett. 2000, 2, 3563–3565. [87] Burgess, K. Solid-Phase Organic Synthesis; WileyBlackwell, 1999. [88] Wirschun, W.; Jochims, J. C. Synthesis 1997, 1997, 233–241. [89] Wirschun, W.; Maier, G.-M.; Jochims, J. C. Tetrahedron 1997, 53, 5755– 5766. [90] Wirschun, W.; Winkler, M.; Lutz, K.; Jochims, J. C. J. Chem. Soc., Perkin Trans. 1 1998, 1755–1762. [91] Meerwein, H.; B¨uchner, E.; van Emsterk, K. J. Prakt. Chem. 1939, 152, 237. [92] Kochi, J. K. J. Am. Chem. Soc. 1955, 77, 5090–5092. [93] Roglans, A.; Pla-Quintana, A.; Moreno-Manas, M. Chem. Rev. 2006, 106, 4622–4643. [94] Kikukawa, K.; Matsuda, T. Chem. Lett. 1977, 159–162. [95] Knorr, R.; Trzeciak, A.; Bannwarh, W.; Gillesen, D. Tetrahedron Lett. 1989, 30, 1927–1930. [96] Richmond, T. J.; Davey, C. A. Nature 2003, 423, 145–150. [97] Franklin, R. E.; Gosling, R. G. Acta Crystallogr. D 1953, 6, 673–677. [98] Marvin, D.; Spencer, M.; Wilkins, M.; Hamilton, L. J. Mol. Biol. 1961, 3, 547–565.

R EFERENCES

103

[99] Ghosh, A.; Bansal, M. Acta Crystallogr. D 2003, 59, 620–626. [100] Wang, J. C. Proc. Natl. Acad. Sci., USA 1979, 76, 200–203. [101] Viswamitra, M. A.; Kennard, O.; Jones, P. G.; Sheldrick, G. M.; Salisbury, S.; Falvello, L.; Shakked, Z. Nature 1978, 273, 687–688. [102] Suter, B.; Schnappauf, G.; Thoma, F. Nucl. Acids Res. 2000, 28, 4083– 4089. [103] Wang, A. H.-J.; Quigley, G. J.; Kolpak, F. J.; Crawford, J. L.; van Boom, J. H.; van der Marel, G.; Rich, A. Nature 1979, 282, 680–686. [104] Wing, R.; Drew, H.; Takano, T.; Broka, C.; Tanaka, S.; Itakura, K.; Dickerson, R. E. Nature 1980, 287, 755–758. [105] Bansal, M. Curr. Sci. 1999, 76, 1178–1181. [106] Drew, H. R.; Wing, R. M.; Takano, T.; Broka, C.; Tanaka, S.; Itakura, K.; Dickerson, R. E. Proc. Natl. Acad. Sci., USA 1981, 78, 2179–2183. [107] Dickerson, R. E.; Drew, H. R. J. Mol. Biol. 1981, 149, 761–786. [108] Pohl, F. M.; Jovin, T. M. J. Mol. Biol. 1972, 67, 375–396. [109] Behe, M.; Felsenfeld, G. Proc. Natl. Acad. Sci., USA 1981, 78, 1619–1623. [110] Ho, P. S. Proc. Natl. Acad. Sci., USA 1994, 91, 9549–9553. [111] Jovin, T. M.; McIntosh, L. P.; Arndt Jovin, D. J.; Zarling, D. A.; Nicoud, M. R.; van De Sande, J. H.; Jorgenson, K. F.; Eckstein, F. J. Biomol. Struct. Dyn. 1983, 1, 21–58. [112] Rich, A.; Nordheimer, A.; Wang, A. H.-J. Ann. Rev. Biochem. 1984, 53, 791–846. [113] Tashiro, R.; Sugiyama, H. Angew. Chem., Int. Ed. 2003, 42, 6018–6020. [114] Tashiro, R.; Sugiyama, H. J. Am. Chem. Soc. 2005, 127, 2094–2097. [115] Yang, X.; Vologodskii, A. V.; Liu, B.; Kemper, B.; Seeman, N. C. Biopolymers 1998, 45, 69–83. [116] Wang, H.; Du, S. M.; Seeman, N. C. J. Biomol. Stuct. Dyn. 1993, 10, 853– 863.

104

R EFERENCES

[117] Tanaka, K.; Pescitelli, G.; Nakanishi, K.; Berova, N. Monatsh. Chem. 2005, 136, 367–395. [118] Caruthers, M. H. Science 1985, 230, 281–285. [119] Berndl, S.; Herzig, N.; Kele, P.; Lachmann, D.; Li, X.; Wolfbeis, O. S.; Wagenknecht, H.-A. Bioconjugate Chem. 2009, 20, 558–564. [120] Tornøe, C. W.; Christensen, C.; Meldal, M. J. Org. Chem. 2002, 67, 3057– 3064. [121] Lewis, W. G.; Green, L. G.; Grynszpan, F.; Radi´c, Z.; Carlier, P. R.; Taylor, P.; Finn, M. G.; Sharpless, K. B. Angew. Chem., Int. Ed. 2002, 41, 1053–1057. [122] Kolb, H. C.; Finn, M. G.; Sharpless, K. B. Angew. Chem., Int. Ed. 2001, 40, 2004–2021. [123] Ciavatta, L.; Ferri, D.; Palombari, R. J. Inorg. Nucl. Chem. 1980, 42, 593– 598. [124] Gunther, M. R.; Hanna, P. M.; Mason, R. P.; Cohen, M. S. Arch. Biochem. Biophys. 1995, 316, 515–522. [125] Chan, T. R.; Hilgraf, R.; Sharpless, K. B.; Fokin, V. V. Org. Lett. 2004, 6, 2853–2855. [126] Jacobsen, M. F.; Ravnsbæk, J. B.; Gothelf, K. V. Org. Biomol. Chem. 2010, 8, 50–52. [127] Himo, F.; Lovell, T.; Hilgraf, R.; Rostovtsev, V. V.; Noodleman, L.; Sharpless, K. B.; Fokin, V. V. J. Am. Chem. Soc. 2004, 127, 210–216. [128] Vasquez, K. M.; Glazer, P. M. Q. Rev. Biophys. 2002, 35, 89–107. [129] Martin, M. M.; Lindqvist, L. J. Lumin. 1975, 10, 381–390. [130] Kuzyk, A.; Schreiber, R.; Fan, Z.; Pardatscher, G.; Roller, E.-M.; H¨ogele, A.; Simmel, F. C.; Govorov, A. O.; Liedl, T. DNA-based SelfAssembly of Chiral Plasmonic Nanostructures with Tailored Optical Response. 2011; http://arxiv.org/abs/1108.3752v1, arXiv:1108.3752v1 [condmat.mes-hall]. [131] Pendry, J. B. Science 2004, 306, 1353–1355.

R EFERENCES

105

[132] Zhang, Z.; Olsen, E. M.; Kryger, M.; Voigt, N. V.; Tørring, T.; G¨ultekin, E.; Nielsen, M.; MohammadZadegan, R.; Andersen, E. S.; Nielsen, M. M.; Kjems, J.; Birkedal, V.; Gothelf, K. V. Angew. Chem., Int. Ed. 2011, 50, 3983–3987. [133] Liu, D.; Balasubramanian, S. Angew. Chem., Int. Ed. 2003, 42, 5734–5736. [134] Liu, D.; Bruckbauer, A.; Abell, C.; Balasubramanian, S.; Kang, D.-J.; Klenerman, D.; Zhou, D. J. Am. Chem. Soc. 2006, 128, 2067–2071. [135] Liedl, T.; Simmel, F. C. Nano Lett. 2005, 5, 1894–1898. [136] Liedl, T.; Olapinski, M.; Simmel, F. C. Angew. Chem., Int. Ed. 2006, 45, 5007–5010. [137] Yurke, B.; Turberfield, A. J.; Mills, A. P.; Simmel, F. C.; Neumann, J. L. Nature 2000, 406, 605–608. [138] Yan, H.; Zhang, X.; Shen, Z.; Seeman, N. C. Nature 2002, 415, 62–65. [139] Simmel, F.; Yurke, B. Appl. Phys. Lett. 2002, 80, 883–885. [140] Ding, B.; Seeman, N. C. Science 2006, 314, 1583–1585. [141] Feng, L.; Park, S. H.; Reif, J. H.; Yan, H. Angew. Chem., Int. Ed. 2003, 42, 4342–4346. [142] Aldaye, F. A.; Sleiman, H. F. J. Am. Chem. Soc. 2007, 129, 13376–13377. [143] Holliday, R. Genet. Res. 1964, 5, 282–304. [144] Kallenbach, N. R.; Ma, R.-I.; Seeman, N. C. Nature 1983, 305, 829–831. [145] Sha, R.; Liu, F.; Seeman, N. C. Biochemistry 2000, 39, 11514–11522. [146] Andersen, E. S.; Dong, M.; Nielsen, M. M.; Jahn, K.; Lind-Thomsen, A.; Mamdouh, W.; Gothelf, K. V.; Besenbacher, F.; Kjems, J. ACS Nano 2008, 2, 1213–1218. [147] Kryger, M. B. L. Design and Functionalization of DNA Nanostructures. M.Sc. thesis, Aarhus University, 2011. [148] http://mfold.rna.albany.edu/. [149] Jares-Erijman, E. A.; Jovin, T. M. J. Mol. Biol. 1996, 257, 597–617.

106

R EFERENCES

[150] Cooper, M.; Ebner, A.; Briggs, M.; Burrows, M.; Gardner, N.; Richardson, R.; West, R. J. Fluoresc. 2004, 14, 145–150. [151] Iqbal, A.; Arslan, S.; Okumus, B.; Wilson, T. J.; Giraud, G.; Norman, D. G.; Ha, T.; Lilley, D. M. J. Proc. Natl. Acad. Sci., USA 2008, 105, 11176– 11181. [152] http://www.invitrogen.com/site/us/en/home/References/Molecular-ProbesThe-Handbook/tables/Fluorescence-quantum-yields-and-lifetimes-forAlexa-Fluor-dyes.html. [153] Gottlieb, H.; Kotlyar, V.; Nudelman, A. J. Org. Chem. 1997, 62, 7512– 7515. [154] Ferapontova, E. E.; Olsen, E. M.; Gothelf, K. V. J. Am. Chem. Soc. 2008, 130, 4256–4258. [155] Chan, T. R.; Hilgraf, R.; Sharpless, K. B.; Fokin, V. V. Org. Lett. 2004, 6, 2853–2855. [156] Quemener, D.; Davis, T. P.; Barner-Kowollik, C.; Stenzel, M. H. Chem. Commun. 2006, 5051–5053. [157] Rotman, A.; Heldman, J. Biochemistry 1981, 20, 5995–5999.

A PPENDIX A

E XPERIMENTAL S ECTION

A.0.1 Immobilization of DNA via Triazenes General methods and materials All DNA sequences were synthesized by DNA Technology A/S, Risskov, Denmark. All other purchased chemicals were acquired from Sigma-Aldrich Co. and used without further purification. All utilized organic solvents were HPLC grade and all water was purified on a Milli-Q™ Biocell system by Millipore. All purifications on RP-HPLC were performed on a Hewlett Packard Agilent instrument with autosampler and fraction collector and fitted with an Phenomenex Clarity 3µm Oligo-RP 50 x 4.60 mm column. Eluent was a mixture of acetonitrile and TEAA (0.1 M, pH 7.0) ramping from 0-20% acetonitrile in the first 3 minutes, then to 30% in the next 13 minutes. Yields of products conjugated to DNA were determined by measurement of light absorption (λ = 260 nm) in aqueous solution under the assumption that the DNA conjugates have little to no effect on the extinction coefficient at the given wave length. A NanoDrop ND-1000 photospectrometer was used with the included software. MALDI-TOF-MS mass spectra were acquired on a Bruker Autoflex instrument set in reflective negative mode. The applied matrix was HPA/DAC on an AnchorPlate target. NMR spectra were recorded on a Varian Mercury (9.4 T) spectrometer (1 H NMR at 400 MHz) at ambient temperatures. The chemical shifts are reported in ppm relative to the solvent residual peak (CDCl3 unless otherwise stated). 153 Cyclic voltammetry (CV) and differential pulse voltammetry (DPV) experiments were done with a three-electrode potentiostat AUTOLAB PGSTAT 30 (Eco Chemie B.V., Utrecht, Netherlands) equipped with GPES 4.9.006 software. In 107

108

A PPENDIX A. E XPERIMENTAL S ECTION

all cases, voltage was measured against a reference electrode of Ag/AgCl. Two different experimental setups were applied during grafting. The micro cell setup consisted of a polymer-based tube (length: 25 mm, inner diameter: 7 mm) fitted with auxiliary and working electrode in each end and reference electrode through a hole (Ø 0.5 mm) in the middle of the tube (See Figure 2.3). The working electrode was a glassy carbon disk electrode (Ø 5.0 mm) fitted in a polymer casing (external diameter: 7 mm). The auxiliary electrode was a gold disk electrode of identical dimensions as the working electrode. Placement of the electrodes in tube was adjustable, providing the setup a variable inner volume of the cell. The droplet setup was based around a vertically affixed disk electrode with the electrode face pointing upwards. The auxiliary electrode (Pt-wire, Ø 0.1 mm) was bent in place around the perimeter of the working electrode and the reference electrode was suspended above it. Ultimately, the solution containing the compound to be grafted was placed as a droplet in full contact with all three electrodes. Working electrodes used in this setup include custom-made gold disk electrodes (Ø 1 mm) and gold disk electrodes (Ø 2 mm) bought from CH Instruments, USA. The custommade electrodes were constructed by casting of the electrode material in epoxy inside a glass tube. The newly made electrodes were severed with a diamond blade and ground mechanically with wet silicon carbide paper (Struers A/S), grit sizes 180, 500, 1000, 2400 and 4000, 15 minutes each. This was followed by sequential, mechanical polishing with suspensions of polycrystalline diamonds with particle sizes 9 m, 3 m, 1 m and 0.25 m, respectively (Struers A/S), 15 minutes with each particle size. The bought electrodes were mechanically polished to a mirror luster stepwise in 1 µm diamond- and in 0.1 µm alumna slurry (both from Struers, Denmark) on microcloth (Buehler, Germany). All electrodes were subsequently ultrasonicated in ethanol/water (1:1) for 5 minutes, and ultimately each electrode was exposed to ten cyclovoltammetric scans in sulfuric acid (0.5 M) at a scan rate of 300 mV/s, in the interval -300 mV to +1700 mV relative to Ag/AgCl. The electrodes were rinsed thoroughly with deionized water and stored in ethanol until usage. The buffer utilized in all experiments was prepared as a 20 mM NaH2 PO4 , 150 mM NaCl, 10 mM MgCl2 solution with pH adjusted to 7.0 by addition of concentrated hydrochloric acid.

Grafting of 3 on glassy carbon, Micro cell setup A solution of K3 Fe(CN)6 (0.1 M, 150 µL) in phosphate buffer (0.1 M, K3 PO4 /Na2 PO4 , pH 7.0) was injected into the cell through the hole for the reference electrode. The reference electrode was inserted, and the total volume of

A.0.1. I MMOBILIZATION

OF

DNA

VIA

T RIAZENES

109

the cell adjusted to fit the volume of the injected solution. A CV was measured at scan rate 10 mV/s from -200 mV to 500 mV testing for the presence of the redox wave set for ferric cyanide. To a solution of 3 (30 µM, 148.5 µL) in phosphate buffer (as above) was added dimethyl sulfate solution (12.5 mg/mL, 1.5 µL) and the entire volume was immediately transferred to the empty, clean micro cell via syringe. After replacing the reference electrode and adjusting the cell volume to match that of the solution, a potential of 500 mV was applied over 30 minutes. Subsequently, the electrode was washed in buffer solution and ultrasonicated for 10 minutes, and the measurement on the K3 Fe(CN)6 was repeated under the same conditions as before. For AFM techniques applied in the imaging of the grafted electrode, please see Chen’s work. 72 Standard grafting conditions: Droplet setup In an Eppendorf tube, 10 µL triazene modified DNA (20 µM in triply distilled water) was diluted with 10 µL PBS and added 1 µL dimethyl sulfate solution (12.5 mg/mL in PBS). The solution was then transferred via pipette to the surface of the working electrode, and reference and counter electrodes were placed in contact with the droplet. To ensure containment of the droplet to the electrode surface area, parafilm was applied around the electrode to increase hydrophobicity of the surrounding area. Penultimately, a negative potential (-500 mV) was applied for 20 minutes while monitoring the current for short-outs. After ended grafting, the electrodes were washed in PBS, followed by submersion in aqueous 6-mercaptohexan-1-ol solution (1 mg/mL) for 30 minutes. General procedure for hybridization with labeled DNA An unmodified oligomer complementary to oligomer A was radioactively labeled in the 5’ end using γ-32 P-ATP and T4 polynucleotide kinasei . Ferrocene labeling was performed by amide linkage of ferrocene carboxylic acid to a 5’-C6-aminomodified strand complementary to oligomer A via the NHS ester of the priorii . The radioactively labeled oligonucleotide (2 µL, approximately 2.5 µM assuming 20% loss during workup) was added unlabeled oligonucleotide (2 µL, 209 µM) and PBS (200 µL). The resulting solution was used as stock solution for the hybridization experiments. A similar stock solution of the ferrocene-labeled complementary strand was produced by dilution with PBS until a total concentration of 1.2 µM. From the stock solution, 20 µL was transferred to the electrode and i ii

Labeling and imaging performed by Morten M. Nielsen and Ebbe S. Andersen Labeling performed by Eva Olsen analogously to her published work 154

110

A PPENDIX A. E XPERIMENTAL S ECTION

left to hybridize for 40 minutes at room temperature. Subsequently, the droplet was removed via pipette, and the electrode washed with PBS buffer (5 x 20 µL, until no radiation was monitored in the washing in the radioactively labeled case). Radioactivity was measured by mounting a phosphor imager screen on top of a fixed array of electrodes adding a reference of stock solution (1 µL) absorbed on filter paper. Array with mounted screen was left in a dark box from 2 hours to overnight. The screen was subsequently scanned in a phosphor imager (Molecular Dynamics), and the recorded counts were integrated per area consistent with one electrode using Imagequant (Molecular Dynamics)i . A non-complementary oligonucleotide for assessing the degree of non-specific binding to grafted electrodes was used the following sequence (5’ to 3’): CAG CGT CCA CCA TCT TAC AC. Signal from the ferrocene labeled strands was detected via DPV in PBS buffer relative to Ag/AgCl.

Conjugation of 2 to DNA (3) An Eppendorf tube was charged with 2 (1.2 µmol) in DMF (12 µL) and oligonucleotide A (10 µL, 250 µM, aq) and triethylamine (1 µL) was added. The mixture was stirred by vortex and left at r.t. overnight. Subsequently, the solution was added water (100 µL), washed with CH2 Cl2 (3 × 50 µL), freeze-dried and redissolved in TEAA buffer (100 µL, 0.1 M). Finally, the product was separated by preparative RP-HPLC (retention time 13.7 minutes) in a yield of 30%.

Conjugation of 7 to DNA (8a,b) An Eppendorf tube was charged with the oligonucleotide (20 µL, 5000 pmol), 7 dissolved in DMF (20 µL, 47 µmol), DIPEA (1 µL) and acetonitrile (20 µL), the mixture was stirred by vortex, centrifuged and left to react for 3 hours. Water (20 µL) was added, the formed precipitate was removed by centrifugation. The supernatant was isolated and added sodium acetate buffer (3 M, pH 5.7, 12 µL) and ethanol (330 µL). The solution was cooled to 4 °C overnight, and the precipitated DNA spun down at 4 °C over 2.5 hours. The supernatant liquid was removed, and the precipitated DNA was redissolved in TEAA buffer (0.1 M, 100 µL) and purified via RP-HPLC (RT 9.69 min in 16% yield for 8a , and RT 9.26 min in 40% yield for 8b ). The final products were identified by MALDI-TOF-MS. Mass calc. for 8a [M+ ]: 4928 m/z, found: 4926 m/z. Mass calc. for 8b [M+ ]: 6178 m/z, found: 6174 m/z.

A.0.2. T RANSFER

OF

H ELIX C HIRALITY

TO

F LUORESCEIN

111

DNA directed coupling of 8 and 3 Initially, 8b (20 pmol) and 3 were freeze-dried and redissolved in PBS buffer (1 µL each). The two solutions were combined, and in the coupling where a template strand was neede, the mixture was divided in two, and one part was added the template strand, oligonucleotide B, (0.2 µL, 47 µM). The other part was left as a reference without the template strand. After 30 minutes, Me2 SO4 solution (0.2 µL, 5.3 mM in PBS buffer) was added to both solutions and they were left to stand for an additional 30 minutes. The reaction mixtures were analyzed by PAGE on a 15% polyacrylamide gel with urea for denaturation at 150 V for 1 hour 20 minutes. Coupling of the template-free 8a and 3 was achieved using the same protocol as above. However, two additional reference reactions were made where the reactants 8a and 3, respectively, were treated with Me2 SO4 solution (vide supra) for 30 minutes.

A.0.2 Transfer of Helix Chirality to Fluorescein All unmodified DNA sequences were synthesized by Integrated DNA Technologies, Inc., USA. All modified DNA sequences were synthesized by Dr Rujie Sha of the Seeman group at New York University on an Applied Biosystems 380B automatic DNA synthesizer, removed from the support, and deprotected using routine phosphoramidite procedures. 118 All solvents and reagents were purchased from commercial sources and used as received without further purification. Water was distilled before use. A DMSO solution of Cy5-azide (10 mM, 1-(6-((3-azidopropyl)amino)-6-oxohexyl)-3,3-dimethyl-2-(5(1,3,3-trimethylindolin-2-ylidene)penta-1,3-dien-1-yl)-3H-indol-1-ium chloride) was purchased from Lumiprobe, LLC. The copper(I) binding ligand tris-(1[3-hydroxypropyl]triazolyl-4-methyl)amine (THTA) was synthesized by Miao Ye following literature procedure. 155,156 Azido-fluorescein was synthesized by Francesca Gruppi following literature procedure. 157 Mass spectroscopy was performed on a MALDI-TOF UltrafleXtreme mass spectrometer with a matrix consisting of 2’,4’,6’-trihydroxyacetophenone (THAP) and dibasic ammonium citrate (DAC) in a w/w ratio of 3:5. Conjugation of fluorescein onto DNA via the Huisgen-Meldahl-Sharpless coupling An Eppendorf tube charged with the oligonucleotide (2 nmol, dry or concentrated aqueous) was added sodium ascorbate (0.01 M, aq, 200 µL), THTA ligand (0.02 M, aq, 70 µL), water (doubly distilled), PBS (0.1 M, 0.8 M, pH 4.3, 20 µL),

112

A PPENDIX A. E XPERIMENTAL S ECTION

azido-fluorescein (0.01 M, DMSO/tert-butanol = 3:1, 20 µL), and copper(II) sulphate (0.01 M, aq, 20 µL) in that order. The amount of water added was chosen to give a final DNA concentration of 2 µM. The reaction vessel was wrapped in aluminum foil and left shaking over night. Subsequently, excess, unreacted azidofluorescein was extracted with n-butanol, the remaining mixture was concentrated on YM-3 Microcon® centrifugal filter (Millipore), added denaturing loading dye, and purified by denaturing PAGE (20%) at 600 V for 1 hour and 20 minutes. Staining was done with ethidium bromide, and the product band was excised and extracted with elusion buffer (500 mM ammonium acetate, 10 mM magnesium acetate, 2 mM EDTA) at 4°C over night. Excess ethidium bromide was extracted and the total buffer volume was reduced by sequential washing with n-butanol, and the product was precipitated with ethanol at -78°C. PAGE always showed full conversion, and approximately 17-24% of the final, pure yellow product was isolated after PAGE purification (estimated from absorbance at 260 nm). Product mass, calc.: 6926, found: 6927. Attachment of Cy5 onto DNA via the Huisgen-Meldahl-Sharpless coupling An Eppendorf tube was charged with premixed cupper/ligand solution (22.5 µL, 2.2 mM copper(II) sulphate, 15.6 mM THTA) and Cy5-azide (4 µL, 750 µM, aq), TEAA buffer (10 µL, 1 M, pH 7.5), sodium ascorbate (15 µL, 10 mM) and water (doubly distilled, 950 µL) were added and subsequently mixed by shaking. The entire mixture was transferred to an Eppendorf tube containing the dry oligonucleotide (2 nmol), and the reaction was left wrapped in aluminum foil, shaking over night. Excess Cy5-azide was extracted from the solution by sequential washing with n-butanol. The remaining solution was concentrated on YM-3 Microcon® centrifugal filter (Millipore), added denaturing loading dye, and purified by denaturing PAGE (20%) at 600 V for 1 hour and 20 minutes. Staining was done with ethidium bromide, and the product band was excised and extracted with elusion buffer (500 mM ammonium acetate, 10 mM magnesium acetate, 2 mM EDTA) at 4°C over night. Excess ethidium bromide was extracted and the total buffer volume was reduced by sequential washing with n-butanol, and the product was precipitated with ethanol. PAGE showed impurity bands which implied degradation of the product, leaving possibility of optimization. Quantities of impurities are hard to assess because the fluorescent properties of Cy5 seem to interact with those of ethidium bromide making direct comparison between the bands hard. Due to calibration problems, the exact product mass was never confirmed by MALDI-TOF-MS (calc.: 6790, found: 6780), but assuming the correct product was isolated, the reaction yield was 1.3 nmol product from 6 nmol starting DNA (22%, estimated from absorbance at 260 nm). The product is blue.

A.0.2. T RANSFER

OF

H ELIX C HIRALITY

TO

F LUORESCEIN

113

Hybridization of DNA Duplexes Complexes were formed by mixing a stoichiometric amounts of each strand (2 nM), as estimated by OD260, in TAE/Mg buffer (40 mM Tris-HCl (pH 8.0), 20 mM acetic acid, 2 mM EDTA and 12.5 mM magnesium acetate, 500 µL). This mixture was then heated to 95°C and cooled to room temperature over night in a 1 L hot water bath insulated in a styrofoam box.

Switching between B- and Z-DNA To obtain the B-form of the DNA duplex, regular TAE/Mg buffer (see above). The same buffer was used for the Z-form, except with added NaCl (20 w/w%). When switching between the two forms, the hybridized DNA duplex solution was concentrated on a YM-3 Microcon® centrifugal filter (Millipore) to remove the previous buffer. When switching from B- to Z-form, the NaCl containing buffer was simply added in a concentration to obtain 20 w/w% NaCl and a total volume of 500 µL. When switching from Z- to B-form, the DNA was washed twice with the NaCl-free buffer solution before the final dilution with buffer to a total volume of 500 µL.

Circular dichroism All CD spectra were recorded on an Aviv 212SF CD spectrometer at 298 K with a spectral bandwidth of 1 nm, a step length of 1 nm, an averaging time of 0.5 seconds per step, and a settling time of 0.333 seconds between steps. The cuvette was of quartz with a light path length of 3 mm. DNA solutions were measured at a concentration of 2 µM (as estimated by OD260) and a total volume of 500 µL. For each sample, 16 scans for recorded and averaged. All recorded spectra had background subtracted (i.e. spectrum of buffer in absence of DNA) and were smoothed using the built-in function in the acquisition software.

Fluorescence detected CD All FDCD measurements were performed by Dr Zhaohua Dai. Fluorescence measurements were performed on a Hitachi F-2500 spectrophotometer. Excitation wavelength (spectral bandwidth: 5 nm) was set at 494 nm and emission spectra (spectral bandwidth: 5 nm) between 500 nm and 900 nm were recorded. In all measurements, 1 cm quartz cells were used. All measurements were performed at 298 K. The spectra were collected from samples in the same buffers as used for regular CD, however, with DNA concentrations of 0.2 µM.

114

A PPENDIX A. E XPERIMENTAL S ECTION

Denaturing polyacrylamide gel electrophoresis Gels contained 20% acrylamide (19:1 acrylamide/bisacrylamide), 8.3 M urea, TBE buffer (89 mM Tris base, 89 mM boric acid, 2 mM EDTA, pH 7.5). During purifaction, each lane was loaded with 1 nmol DNA (as estimated per OD260) in denaturing loading dye (10 mM NaOH, 1 mM disodium EDTA, in 90% formamide, 10% water containing trace amounts of xylene cyanole FF and bromophenol blue tracking dyes), and electrophoresis was performed on a Hoefer SE 600 electrophoresis unit at 600 V, 55°C in TBE running buffer. Subsequent staining was done with ethidium bromide.

A.0.3 DNA Actuators All DNA sequences were synthesized by DNA Technology A/S, Risskov, Denmark. All other purchased chemicals were acquired from Sigma-Aldrich Co. and used without further purification. All utilized organic solvents were HPLC grade and all water was purified on a Milli-Q™ Biocell system by Millipore. All purifications on RP-HPLC were performed on a Hewlett Packard Agilent instrument with autosampler and fraction collector and fitted with an Phenomenex Clarity 3µm Oligo-RP 50 x 4.60 mm column. Eluent was a mixture of acetonitrile and TEAA (0.1 M, pH 7.0) ramping from 0-20% acetonitrile in the first 3 minutes, then to 30% in the next 13 minutes. Yields of products conjugated to DNA were determined by measurement of light absorption (λ = 260 nm) in aqueous solution under the assumption that the DNA conjugates have little to no effect on the extinction coefficient at the given wave length. A NanoDrop ND-1000 photospectrometer was used with the included software. The monoreactive Cy3B-NHS and Cy5-NHS dyes were purchased from GE Healthcare Conjugation of fluorophores to DNA The monoreactive NHS ester of the dye to be conjugated (0.1 mg) was dissolved in DMF (5 µL) and added to an aqueous solution of the amino-modified DNA (2 nmol, 5 µL) along with acetonitrile (5 µL) and triethylamine (0.2 µL). The reaction mixture was covered in aluminium foil and agitated for 2 hours. Subsequently, the reaction mixture was added sodium acetate (1.5 µL, 3M, pH 5.2) and ethanol (45 µL, 96%, cold) and was cooled on dry ice for 45 min. The precipitated DNA was isolated by centrifugation (1 hour, 14000 rpm) and washed with ethanol (45 uL, 70%, cold). The isolated DNA was dissolved in TEAA buffer (0.1 M, pH 7) and purified by RP-HPLC. Conjugation with Cy3B proceeded with very poor

A.0.3. DNA ACTUATORS

115

yield (69 pmol, 3.5%). Conjugation with Cy5 showed close to full conversion. Actuator mono- and dimer assembly - general procedure A PCR tube was charged with piston and roller strands in stoichiometric amounts. If a locked complex was desired, the appropriate lock strands were added in 5 equivalents. TAE/Mg buffer and water were added until a final buffer concentration of 1x (40 mM Tris-HCl (pH 8.0), 20 mM acetic acid, 2 mM EDTA and 12.5 mM magnesium acetate) and a DNA concentration of 2 µM for each of the piston and roller strands. Subsequent hybridization was performed on Thermocycler by first heating quickly to 80°C and subsequently cooling slowly. Streptavidin binding Streptavidin was bound to the locked actuator complex in some experiments to increase retention in PAGE analysis compared to the unlocked complex. In these experiments, biotin modified lock strands were used. In the case of straight locks, only the strand designed to hybridize with Piston 2 was biotin modified. The locked complexes were assembled using the general procedure and then subsequently added 5 equivalents (20 µM in TAE/Mg buffer) streptavidin relative to the biotin modified lock strand. The samples were kept at 4°C for 1 hour before subsequent PAGE analysis. Polyacrylamide gel electrophoresis Gels contained either 7.5% or 6% polyacrylamide (from 19:1 acrylamide/bisacrylamide), TBE buffer (89 mM Tris base, 89 mM boric acid, 2 mM EDTA, pH 7.5), and magnesium chloride (12.5 mM). Prior to loading of samples onto the gel, the samples were added an equal volume of loading dye (60% glycerol in TAE/Mg and trace amounts of bromophenol blue tracking dye). Electrophoresis was performed at 70 V, 4°C in TBE running buffer for 2 hours. Subsequent staining was done with ethidium bromide, and images were recorded under UV-irradiation. In the initial assembly of the unlocked actuator mono- and dimer, each lane on the gels was loaded samples containing 10 pmol of each strand present. Subsequent gels were loaded samples containing 4 pmol. FRET measurements Samples of the actuator dimer intended for FRET analysis were assembled following the general procedure with Cy3B-modified Piston 1, Cy5-modified Piston

116

A PPENDIX A. E XPERIMENTAL S ECTION

3, and unmodified pistons 2 and 4 and rollers 1, 2, and X. For locking, unmodified U-lock strands were used in excess (5 equivalents). A sample locked in each of the states 0–11 was produced, each containing 2 pmol complex, and following hybridization, TAE/Mg buffer was added to a total volume of 70 µL. FRET measurements were performed on a scanning spectrofluorometer (Fluoro-Max-3, HORIBA Jobin Yvon Inc.) in a quartz cuvette, which was washed twice with TAE/Mg between different samples. Excitation was performed at 530 nm (Cy3B) and at 600 nm (Cy5), and spectra of the locked actuator dimers were recorded with 0.5 s integration time and 1 nm wavelength intervals. Spectral intensities were corrected for the excitation lamp intensity fluctuations and for instrumental change of detection efficiency as a function of wavelength. Temperature was set to either 25°C or 4°C during measurements. Subsequent PAGE analysis of FRET samples was performed as described above. Each lane contained approximately 0.3 pmol complex on a 6% gel, and imaging was done on a Typhoon scanner (Amersham Biosciences) without staining. Gel imaging consisted of three scans. Two scans were with fluorophore excitation at 532 nm (Cy3B), where emission intensity was recorded in the ranges 565–595 nm (filter peak transmission at 580 nm) and 655– 685 nm (filter peak transmission at 670 nm). One scan was performed with fluorophore excitation at 633 nm (Cy5), and emission was recorded through the same 670 nm filter as above. Scanned gels were analyzed using ImageQuant TL 7.0 (GE Healthcare). Lane creation and band detection was done manually.

A PPENDIX B

P YTHON C ODE

Script for Finding DNA Base Sequences The Python script below is a work in progress and is provided as it looked when it was used, except escaped line breaks have been added to fit the text to the page. Unfortunately, this means that the code is not always easily read, and not all unused parts have been removed. Some unused parts have been added with the aim of eventual expansion of the program’s capabilities, others are old versions that were simply not deleted in the event that they would become useful again. The Python version is 2.7.

117

2 3

#!c:/Python27/python.exe from __future__ import division import random, bisect

118

1

4 5 6 7 8

class Project(): """Defines ScaffoldGroup and a given set of Strand object generated against that ScaffoldGroup. Each Strand is sequentially added to the ScaffoldGroup as they are generated, ensuring orthogonality between strands.

9 10 11

strands must be a list of dictionaries. Each dictionary defines the parameters for one strand. Example: strand = [{"length":8, "wordSize":3, ...}, ]"""

12 13 14 15 16

18 19 20 21 22 23 24 25 26

def printStrands(self): for strand in self.strands: print "".join(strand.seq())

A PPENDIX B. P YTHON C ODE

17

def __init__(self, scaffolds=[], strands=[]): self.scaffold = ScaffoldGroup(scaffolds) self.strands = [] self.addScaf = self.scaffold.addScaf #Makes referencing easier if scaffolds: self.addScaf(scaffolds) for item in strands: self.newStrand(**item) #Add strands one by one. #**item parses all the individual keyword=value pairs # in the list item.

28

def newStrand(self, length, selfWordSize, scafWordSize, predefSeq=None, scafGroup=None, GCweight=0.5):

29 30

if scafGroup is None: scafGroup = self.scaffold

31 32 33 34 35 36 37 38

self.strands.append(Strand(length=length, wordSize=selfWordSize, predefSeq=predefSeq, scafGroup=scafGroup, GCweight=GCweight)) self.addScaf([scafWordSize, "".join(self.strands[-1].seq()), "".join(self.strands[-1].complement())]) #Add the new strand to the scaffold

A PPENDIX B. P YTHON C ODE

27

39 40

self.scaffold.reduceScaf()

41 42 43 44 45 46 47 48 49 50 51

119

52

class ScaffoldGroup(): """ Takes scaffolds = [[wordSize:int, seq:str, seq2:str,...],...]""" def __init__(self, scaffolds=None): self.wordSizeList = [] self.scaffolds = {} if scaffolds is None: scaffolds = [[1, "x"]] for item in scaffolds: self.addScaf(item) self.reduceScaf()

54 55 56 57 58 59 60 61

120

53

def addScaf(self, scaf, reduce=True): if scaf[0] > 0: for seq in scaf[1:]: if all([seq[:wS] not in self.scaffolds[wS].words for wS in self.wordSizeList]): try: self.scaffolds[scaf[0]].addSequence(seq) except KeyError: self.wordSizeList.append(scaf[0]) self.scaffolds[scaf[0]] = StaticScaffold(sequences=scaf[1:],

62 63 64 65 66

if reduce: self.reduceScaf() else: raise RuntimeError("Parsed Scaffold size is not allowed. scaf[0] = " + str(scaf[0]))

67 68

70 71 72 73

def addScafGroup(self, scafGroup): if scafGroup.__class__.__name__ == ’ScaffoldGroup’: for wS in scafGroup.wordSizeList: self.addScaf([wS] + scafGroup.scaffolds[wS].sequences) else: raise RuntimeError(’Object parsed is not of class DNA.ScaffoldGroup’)

74 75 76 77 78

def reduceScaf(self): #print "reducing" wordSizeList = sorted(self.wordSizeList) wordSizeList.reverse()

A PPENDIX B. P YTHON C ODE

69

80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97

for size in wordSizeList: for word in self.scaffolds[size].words: if all( \ [any( \ [word[-wS:-1] + base in self.scaffolds[wS].words \ for wS in self.wordSizeList]) for base in "ACTG"]): try: self.addScaf([size-1, word[:size-1]], reduce=False) except RuntimeError as err: print str(err) print self.scaffolds[1].words raise if size-1 in self.wordSizeList: for word in self.scaffolds[size-1].words: for base in "ACTG": word + base in self.scaffolds[size].words and \ self.scaffolds[size].words.remove(word + base) if sorted(wordSizeList) != sorted(self.wordSizeList): self.reduceScaf()

A PPENDIX B. P YTHON C ODE

79

98 99 100 101 102 103

121

104

class StaticScaffold: """Takes a set of static strands and creates a library of wordSize codons sequences is a list of strings containing the sequences wordSize is an integer defining the codon length words is the library of codons.

106 107 108 109 110 111 112 113 114 115 116

Functions: verifyDNA() """ def __init__(self, sequences, wordSize): self.sequences = [self.verifyDNA(seq) for seq in sequences] self.words = [] self.wordSize = wordSize for seq in self.sequences: codons = [seq[item:item+self.wordSize] for \ item in xrange(len(seq)-self.wordSize+1)] for codon in codons: self.words.append(codon) # Produce wordSize overlapping codons self.words = [codon for codon in self.uniqueWords(self.words)] #Remove duplicates

122

105

117 118 119 120

122 123 124 125 126 127 128

def verifyDNA(self, strand): """Verifies that strand is a string containing only the bases A, T, G, and C In addition it returns the input strand with only capital letters and stripped of spaces."""

129 130

strand = ’’.join(strand.upper().split(’ ’)) # capitalize and remove spaces

A PPENDIX B. P YTHON C ODE

121

def addSequence(self, sequence): sequence = self.verifyDNA(sequence) self.sequences.append(sequence) self.words += [sequence[item:item+self.wordSize] for \ item in xrange(len(sequence)-self.wordSize+1)] self.words = [codon for codon in self.uniqueWords(self.words)] #Remove duplicates

132 133 134 135 136 137 138

if (reduce(lambda x, y: x and y, \ [base in ’ACTG’ for base in strand]) or strand == ’X’): # reduce takes a function and a list. # The function is given the previous result and the next element in the list. return strand else: raise RuntimeError( \ ’Input sequences contains non-nucleotide characters (A, T, C, G):\n’ \ + strand)

139 140 141 142 143 144 145 146

def uniqueWords(self, wordList): seen = set() for codon in wordList: if codon in seen: continue seen.add(codon) yield codon

A PPENDIX B. P YTHON C ODE

131

147 148 149 150 151 152 153 154 155

123

156

class Strand: """Defines a DNA strand. length is an integer containing the number of bases predefSeq is a dictionary containing predefined strand sequences on the form {index:"seq", ..}. index is an integer containing the number of bases before the predefined section seq is a string containing the base sequence.

158 159 160 161 162

Functions: generateSeq() insertSeq() complement() seq() verifyDNA()"""

124

157

163 164 165 166 167 168 169 170 171 172

174 175 176 177 178 179 180 181 182

def generateSeq(self, length, ownWordSize, scafGroup=None, GCweight=0.5): """Generates and returns a sequence with length nucleotides. If a scaffold (of class StaticScaffold) is parsed, the returned strand consists of codons not found in the scaffold.

A PPENDIX B. P YTHON C ODE

173

def __init__(self, length, wordSize, predefSeq=None, scafGroup=None, GCweight=0.5): self.length = length self.GCweight = GCweight self.wordSize = wordSize self.predefSeq = predefSeq self.sequence = [[base, None, None, False, False] for base in \ self.generateSmartSeq(length, wordSize, scafGroup,\ GCweight=self.GCweight)] #Create random strand with *length* bases if predefSeq: self.insertSeq(predefSeq) #sequence = [[Base, compStrand, compIndex, static, slave], [Base, ....]...] self.GCcontent = sum([base in "GC" for base in self.seq()])/self.length

184 185 186 187 188 189 190 191 192 193 194 195 196 197 198

returns a string""" if scafGroup: if self.predefSeq: scafGroup.addScaf([wordSize] + \ [self.predefSeq[index] for index in self.predefSeq] + \ [self.complement(self.predefSeq[index])\ for index in self.predefSeq]) # ˆ--- This needs to be tested. scafWSList = scafGroup.wordSizeList strandWords = [] strand = self.randomBase(ownWordSize-1, GCweight) try: while (len(strand)