Abstract The Human Leukocyte Antigen (HLA) class Ib molecules consist of three molecules – HLA-E, HLA-F and HLA-G [reviewed in Kochan et al., 2013]. HLA-G and HLA-F were first reported to be expressed by trophoblast cells in combination with HLA-E at the feto-maternal interface, where they protect the fetus against maternal immune attacks [reviewed in Hviid, 2006]. Later, it was observed that tumor cells also express HLA-G and this lead to the hypothesis that expression of HLA-G was a mechanism by which tumors could escape the immune system and immunosurveillance [Paul et al., 1998]. HLA-E and HLA-F are also often found over-expressed in a variety of malignancies [reviewed in Kochan et al., 2013]. In this thesis, we characterized the HLA class Ib mRNA and protein expression of seven malignant melanoma cell lines by Droplet Digital PCR and flow cytometry, respectively. The mRNA expression of different isoforms of HLA-G was further characterized by RT-PCR fragment length analysis. Messenger RNA expression of HLA class Ib was observed in all cell lines while protein expression was observed in some of these. The levels of mRNA and protein expression differed between the cell lines. HLA-G is the most extensively studied HLA class Ib molecule. Less is known about the function of HLA-E, while the function of HLA-F remains largely unknown. The function of HLA-E seems to be mainly through inhibition of NK and CD8+ T cells, while the function of HLA-G seems to be diverse [reviewed in Kochan et al., 2013]. One mechanism whereby HLA-G is thought to induce immune tolerance is through induction of regulatory T cells (Tregs). In this thesis, we performed functional pilot studies in which we co-cultured peripheral blood mononuclear cells (PBMCs) with malignant melanoma cell lines. In some cases, it seemed that the percentages of Tregs and of HLA-G-positive T cells were affected by the co-culture. In addition, we performed a pilot study investigating if the peripheral blood levels of Tregs and HLA-G-positive T cells were different between skin cancer patients and healthy individuals. The levels of Tregs did not differ between the two groups but the level of HLA-G-positive CD4+ T cells seemed to be increased while the level of HLA-G-positive CD8+ T cells seemed to be decreased in the peripheral blood of skin cancer patients compared to healthy individuals. However, additional studies have to be performed to confirm or disconfirm the findings of these pilot studies.

1

Dansk resumé Gruppen af Humant Leukocyt-Antigen (HLA) klasse Ib proteiner består af tre molekyler – HLA-E, HLA-F og HLA-G [review af Kochan et al., 2013]. Expression af HLA-G og HLA-F blev først observeret i trofoblast-celler i den føto-maternelle barriere, hvor de sammen med HLA-E beskytter fosteret imod angreb fra morens immunsystem [review af Hviid, 2006]. Senere blev HLA-G-ekspression også observeret i tumorceller, hvilket førte til den hypotese, at HLA-Gekspression er en mekanisme, hvorved tumorer undslipper immunsystemets overvågning, såkaldt ’tumor escape’ [Paul et al., 1998]. Overekspression af HLA-E og HLA-F observeres også ofte i en række maligne sygdomme [review af Kochan et al., 2013]. I denne specialerapport har vi, i syv malignt melanom cellelinier, karakteriseret ekspressionen af HLA klasse Ib mRNA og protein ved henholdsvis flowcytometri og Droplet Digital PCR. Vi karakteriserede yderligere mRNA-ekspressionen af forskellige HLA-Gisoformer vhja. RT-PCR-fragment-længde-analyse. Messenger RNA ekspression af HLA klasse Ib blev observeret i alle cellelinier, mens protein ekspression blev observeret i nogle af disse. Niveauerne af mRNA og protein expression var forskellige imellem cellelinierne. HLA-G er det mest undersøgte HLA klasse Ib-molekyle. Man kender lidt til funktionen af HLA-E, mens funktionen af HLA-F stort set er ukendt. Tilsyneladende udøver HLA-E hovedsageligt sin funktion ved at hæmme NK-celler og CD8+ T-celler, mens der er stigende evidens for, at HLA-G har mange forskellige funktioner [review af Kochan et al., 2013]. Det menes, at inducering af regulatoriske T-celler (Tregs) er en vigtig mekanisme, hvorved HLA-G inducerer immuntolerans. I denne specialerapport har vi udført funktionelle co-kultur-pilotforsøg med mononukleære celler fra perifert blod (PBMCs) og malignt melanom cellelinier. I nogle tilfælde viste disse foreløbige resultater, at procentdelen af Tregs og HLA-G-positive T-celler var påvirket af denne co-kultur. I et pilotforsøg undersøgte vi desuden, om niveauerne af Tregs og HLA-Gpositive T-celler i perifert blod var forskellige imellem hudcancer-patienter og raske kontrolindivider. Niveauerne af Tregs i perifert blod var ikke signifikant forskellige imellem de to grupper i de udførte pilotforsøg, men niveauet af HLA-G-positive CD4+ T-celler var øget mens niveauet af HLA-G-positive CD8+ T-celler var nedsat i hudcancer-patienter sammenlignet med de raske kontrol-individer. Dog kræves der flere forsøg for at reproducere eller afkræfte de foreløbige resultater af disse pilotforsøg.

2

Preface and acknowledgement The present Master Thesis was completed as part of our master studies at Roskilde University. The thesis was performed in collaboration with the Department of Clinical Biochemistry at Copenhagen University Hospital (Roskilde). All the experimental work in this thesis was performed in the laboratories at the Department of Clinical Biochemistry from September 2013 to June 2014. We first met our external supervisor Thomas V. F. Hviid, MD, PhD, DMSc, Associate Professor in April 2013. At this point, his research was mostly focusing on the HLA class Ib molecule HLA-G and preeclampsia. During the last decade, there has been a growing interest in the influence of the HLA class Ib molecules in cancer. As a possibility to investigate the interaction of the HLA class Ib molecules in the cancer environment occurred, Thomas V. F. Hviid saw his opportunity to expand the research field. This new project named CancerImmun-project (CIP) was approved by the local ethics committee of Region Sjælland in October 2013. The aim of the project is to investigate the role and influence of the HLA class Ib molecules in the cancer environment. The project includes skin, breast and ovarian cancer, but our thesis only included skin cancer, while PhD student Wenna Gleyce Araujo Do Nascimento works with breast and ovarian cancer. It has been exciting but also challenging to be a part of the establishment of this new project. Throughout the past 10 months, we have met some wonderful people, who we want to thank for their help and support. First of all we would like to thank our daily supervisor, Thomas V. F. Hviid, MD, PhD, DMSc, Associate Professor for giving us the opportunity to work in the laboratories at the Department of Clinical Biochemistry. We are grateful for the extraordinary supervision and support that we have got from you throughout the whole project. We are also very thankful for the big trust and open possibilities that you have given us. In addition, we would also like to thank our internal supervisor Cathy Mitchelmore, B.A., PhD, Associate Professor for her support and faith in our project. Besides our supervisors, we would like to thank the helpful staff in the Department of Clinical Biochemistry. A special thanks to Anja Steenfatt Torbensen, Technician who has helped us several times with purchasing materials for our experiments through a newly implemented

3

purchasing system that have caused us some serious headaches. Unfortunately, we have, for some orders, waited several months, which in many ways have delayed our experimental part of the thesis. Roskilde Hospital was due to complications with the new purchasing system listed as “bad payers” at many companies and this caught us since the financing of materials to our project has been through the Department of Clinical Biochemistry at Copenhagen University Hospital (Roskilde). A great thank to Nina Dahl Kjersgaard, Technician who introduced us to the basic skills of fragment length analysis and Droplet Digital PCR. Another person who has been very helpful, no matter our questions asked is Snezana Djurisic PhD student, MSc. We would like to thank Gitte Winther Thomsen, Technician and Hanne Normann Rasmussen, Technician for helping us with the practical part of flow cytometry. We have really appreciated their help and enjoyed their good working atmosphere and kindness. Furthermore, we would like to thank Gregor Borut Ernst Jemec, MD, DMSc, Professor, Head of Dept., Department of Dermatology and Jørgen Lock-Andersen, MD, DMSc, Chief Physician, Department of Plastic surgery for their collaboration and help with including patients in our study. Last but not least we want to thank Wenna Gleyce Araujo Do Nascimento, a PhD student at the Department of Clinical Biochemistry, who arrived from Brazil last year and started out the CancerImmun-Project with us. Working with the same project, though with different types of cancers, we have been able to share our knowledge and experience in the establishment and implementation of this project.

4

List of abbreviation APC

Antigen presenting cell

BCC

Basal cell carcinoma

CCL

Chemokine (C-C motif) ligand

CCR

Chemokine (C-C motif) receptor

KIR2DL4

Killer cell immunoglobulin like receptor 2DL4

mAb

Monoclonal antibody

MICA

MHC class I related chain A

MHC

Major histocompatibility complex

mRNA

Messenger ribonucleic acid

NK

Natural killer

NKG2

Natural killer group 2

cDNA

Complementary DNA

CIP

CancerImmmun-Project

CXCL

Chemokine (C-X-C motif) ligand

DC

Dendritic cell

nTreg

Natural Treg

ddPCR

Droplet digital PCR

pAb

Polyclonal antibody

DMEM

Dulbecco’s Modified Eagle Medium

PBMCs cells

Peripheral blood mononuclear

EDTA

Ethylenediaminetetraacetic acid

PCR

Polymerase chain reaction

EMEM

Eagle’s Minimun Essential Medium

PMT

Photo-multiplier tube

ROC

Receiver operating characteristic

F-12K

Kaighn’s Modification of Ham´s F-12 (F-12K) Medium

RPMI

FBS

Fetal bovine serum

Roswell Park Memorial Institute

HLA

Human Leukocyte Antigen

RT-PCR

Real time PCR

sHLA

Soluble HLA

IFN-

Interferon-

IL

Interleukin

TGF-

Transforming growth factor 

ILT

Immunoglobulin-like transcript receptor

Tm

Melting temperature

Tregs

T regulatory cells

iTreg

Inducible Treg

UTR

Untranslated regio

5

Table of contents

Abstract .......................................................................................................................................................................... 1 Dansk resumé .............................................................................................................................................................. 2 Preface and acknowledgement ............................................................................................................................. 3 List of abbreviation ................................................................................................................................................... 5 Introduction ................................................................................................................................................................. 9 Aim ................................................................................................................................................................................. 11 Human Leukocyte Antigens (HLA) .................................................................................................................... 12 Classical HLA class I molecules – HLA class Ia ......................................................................................... 12 Non-classical HLA class I molecules – HLA class Ib ............................................................................... 13 Human Leukocyte Antigen E (HLA-E) .................................................................................................... 14 Human Leukocyte Antigen F (HLA-F) .................................................................................................... 14 Human Leukocyte Antigen G (HLA-G) .................................................................................................... 15 Expression of HLA class Ib in cancer ................................................................................................................ 18 HLA-E expression in cancer ............................................................................................................................ 18 HLA-F expression in cancer............................................................................................................................. 18 HLA-G expression in cancer ............................................................................................................................ 19 Skin cancer .................................................................................................................................................................. 20 Basal cell carcinoma ........................................................................................................................................... 20 Studies investigating HLA class Ib expression in BCC ...................................................................... 21 Cutaneous malignant melanoma ................................................................................................................... 22 Studies investigating HLA class Ib expression in malignant melanoma ................................... 23 Immunoediting.......................................................................................................................................................... 25 Regulatory T cells (Tregs) .................................................................................................................................... 30 Clinical relevance of Tregs in cancer ........................................................................................................... 31 Dendritic cells (DCs) ............................................................................................................................................... 33 The interaction of HLA class Ib with immune cells ..................................................................................... 35 HLA class Ib and NK cells ................................................................................................................................. 36 HLA class Ib and dendritic cells ..................................................................................................................... 37 HLA class Ib and B cells ..................................................................................................................................... 38 HLA class Ib and T cells ..................................................................................................................................... 38 HLA class Ib and Tregs ...................................................................................................................................... 39

6

Establishment of a cell bank ................................................................................................................................ 41 Experiment 1 – Flowcytometric characterization of malignant melanoma cell lines ................... 44 Aim ............................................................................................................................................................................ 44 Hypothesis ............................................................................................................................................................. 44 Preparation ............................................................................................................................................................ 44 Materials and methods ...................................................................................................................................... 46 Results ..................................................................................................................................................................... 48 Discussion .............................................................................................................................................................. 54 Optimization of the experiment..................................................................................................................... 55 Experiment 2 – Droplet Digital PCR (ddPCR) characterization of malignant melanoma cell lines ............................................................................................................................................................................... 56 Introduction .......................................................................................................................................................... 56 Aim ............................................................................................................................................................................ 57 Hypothesis ............................................................................................................................................................. 57 Preparation ............................................................................................................................................................ 57 Materials and methods ...................................................................................................................................... 59 Results ..................................................................................................................................................................... 60 Discussion .............................................................................................................................................................. 63 Optimization of the experiment..................................................................................................................... 64 Experiment 3 – Fragment analysis of HLA-G isoforms in malignant melanoma cell lines .......... 66 Aim ............................................................................................................................................................................ 66 Hypothesis ............................................................................................................................................................. 66 Materials and methods ...................................................................................................................................... 66 Results ..................................................................................................................................................................... 68 Discussion .............................................................................................................................................................. 69 Optimization of the experiment..................................................................................................................... 70 Experiment 4 – Flow cytometric analysis of PBMCs co-cultured with malignant melanoma cells lines ..................................................................................................................................................................... 72 Aim ............................................................................................................................................................................ 72 Hypothesis ............................................................................................................................................................. 72 Preparation ............................................................................................................................................................ 72 Materials and Methods ...................................................................................................................................... 74 Results ..................................................................................................................................................................... 78 Discussion .............................................................................................................................................................. 80 Optimization of the experiment..................................................................................................................... 82 Experiment 5 – Flow cytometric analysis of PBMCs from skin cancer patients.............................. 84 7

Introduction .......................................................................................................................................................... 84 Aim ............................................................................................................................................................................ 84 Hypothesis ............................................................................................................................................................. 84 Preparation ............................................................................................................................................................ 84 Materials and methods ...................................................................................................................................... 85 Results ..................................................................................................................................................................... 87 Discussion .............................................................................................................................................................. 90 Optimization of the experiment..................................................................................................................... 91 Discussion ................................................................................................................................................................... 93 References................................................................................................................................................................ 100 Appendix 1 - Mycoplasma test of cell lines ................................................................................................. 115 Appendix 2 – Extracellular and intracellular characterization of cell lines ................................... 118 Appendix 3 - Droplet Digital PCR .................................................................................................................... 121 Appendix 4 - RNA purification with RNeasy® Mini Kit ......................................................................... 123 Appendix 5 - Protocol for First-Strand cDNA synthesis......................................................................... 125 Appendix 6 – Droplet Digital PCR results and calculations .................................................................. 126 Appendix 7 - Titration of antibodies.............................................................................................................. 130 Appendix 8 - Co-culture of cell lines and PBMCs ...................................................................................... 141 Appendix 9 - List of lot numbers ..................................................................................................................... 146

8

Introduction Early studies showed that expression of the non-classical HLA class I molecule HLA-G was restricted to trophoblasts at the feto-maternal interface [reviewed in; Hviid 2006, Carosella et al., 2008] In the condition of pregnancy HLA-G mediates immune tolerance by suppressing proliferation of alloreactive CD4+ T cell [Bainbridge et al., 2000] and inhibiting cytolysis mediated by NK- and T-cells [Rouas-Freiss et al., 1997; reviewed in Hviid, 2006]. Later, its expression was detected in immune privileged organs such as the cornea, thymus, pancreatic islets, endothelial cell precursors and erythroblasts as well [reviewed in Carosella et al., 2008]. Expression of HLA-G is also found in some pathological conditions including multiple sclerosis, inflammatory diseases, viral infections and in numerous types of tumor tissue [reviewed in Carosella et al., 2008]. Growing interest in the role of HLA-G in cancer rose, given its role in the protection of fetal trophoblasts from maternal immune attack [reviewed in Hviid, 2006]. The first study to report detection of HLA-G expression by tumor cells was by Paul et al. (1998). The authors suggested that expression of HLA-G, analogues to its role in the escape of trophoblasts from maternal allorecognition, is a strategy for the cancer cells to escape host immunosurveillance [Paul et al., 1998]. Since this, several studies have confirmed the expression of HLA-G in various cancers. The cancer types studied include cutaneous melanomas, renal cell carcinomas, colorectal cancer, ovarian carcinomas, gastric cancer and cutaneous T-cell lymphomas [reviewed in Donadi et al., 2011]. Two other class Ib molecules exists, HLA-E and HLA-F, that are also often found over-expressed in a variety of malignancies [reviewed in Kochan et al., 2013]. Tumor cells are often found to have down-regulated expression of the classical HLA class I molecules [reviewed in Ferrone & Marincola, 1995]. The expression of the classical HLA class I molecules is necessary for the induction of a T cell based anti-tumor specific immune response [reviewed in Rebmann et al., 2007]. Therefore, loss or down-regulation of the classical HLA class I molecules by tumors enables them to escape from the immunological surveillance [reviewed in Rebmann et al., 2007]. However, the down-regulation is in vitro associated with an increase in the susceptibility to NK cell-mediated lysis [Maio et al., 1991]. 9

So how come the tumor still escape the immune system? One of the mechanisms of escaping immune surveillance seems to be tumor expression of HLA-G. HLA-G can affect a variety of immune cells to induce a tolerogenic environment for the tumor [Kochan et al., 2013]. The mechanisms for this are not fully elucidated but it is shown, that HLA-G, in addition to its ability to inhibit proliferation and function of immune cells, can induce regulatory T cells and tolerogenic dendritic cells [Kochan et al., 2013]. These cells further inhibit immune cells and maintain a tolerogenic environment [Kochan et al., 2013]. The function of HLA-E and HLA-F in tumors is not investigated as extensively as the function of HLA-G. Therefore, a more detailed description of HLA-G and its functions will be presented in this thesis.

10

Aim In our thesis we will work with malignant melanoma cell lines and blood samples from skin cancer patients. The malignant melanoma cell lines - we will characterize the mRNA and protein expression of HLA class Ia and Ib molecules and the expression of the different HLA-G isoforms. The protein expression will be characterized by flow cytometri, the mRNA expression will be characterized using Droplet Digital PCR and the different HLA-G isoforms will be detected by fragment length analysis. Furthermore, we will investigate the effect on immune cells when co-culturing these with cancer cells expressing different levels of HLA class Ia and Ib and different isoforms of HLA-G. Patient blood samples – we will investigate if peripheral blood levels of different T cell subsets, including Tregs, differ between skin cancer patients and healthy individuals.

11

Human Leukocyte Antigens (HLA) The human Major Histocompatibility Complex (MHC), named the Human Leukocyte Antigen (HLA) system, is located on the short arm of chromosome 6 and contains many genes (see Figure 1) [Wood, 2011, p. 71]. Some of the genes are closely related, and their products have similar functions. On this basis, the genes, and the proteins encoded by these, have been divided into three classes: class I, class II and class III [Wood, 2011, p. 74]. The class I and II molecules are closely related. Most of them are surface proteins and are involved in T cell recognition of antigens [Wood, 2011, p. 74]. Class III molecules are not related to class I or II or to each other [Wood, 2011, p. 74]. They have some immune-related functions encoding several components of the complement system, which help activate and maintain the inflammatory process of an immune response [Wood, 2011, p. 74, 181-184]. The class I molecules are further divided into the classical and non-classical HLA molecules, also called class Ia and Ib molecules [Wood, 2011, p. 76]. A description of both of these groups will follow below.

Figure 1: Gene map of the Human Leukocyte Antigen (HLA) region [modified from Mehra & Kaur, 2003]. The HLA class I, II and III genes are located in the HLA region on the short arm of chromosome 6 [Wood, 2011, p. 71, 74]. The HLA class I and class II genes are closely related and most of them are involved in antigen presentation [Wood, 2011, p. 74]. The HLA class III genes are related neither to the other classes nor to each other [Wood, 2011, p. 74]. They have some immune related functions encoding several components of the complement system, which help activate and maintain the inflammatory process of an immune response [Wood, 2011, p. 74, 181-184].

Classical HLA class I molecules – HLA class Ia The HLA class Ia molecules are expressed on the cell surface of most nucleated cells [Wood, 2011, p. 74]. Their role is to bind peptides derived from antigens and present these peptides for recognition by CD8+ T lymphocytes [Wood, 2011, p. 84]. They consist of two protein

12

chains, the -chain that has three external domains called 1, 2 and 3, plus another chain called 2-microglobulin, which is non-covalently bound to the α-chain (see Figure 2) [Wood, 2011, p. 74-75]. The gene encoding 2-microglobulin is not located on chromosome 6, as the HLA genes, but on chromosome 15 [Wood, 2011, p. 74].

Figure 2: The structure of HLA class Ia molecules [modified from Hughes & Yeager, 1998]. HLA class Ia molecules have a transmembrane α-chain with three extracellular domains α1-3 [Wood, 2011, p. 74-75]. Most of the molecules have β2-microglobulin covalently bound to the α-chain [Wood, 2011, p. 74-75].

There are three class Ia genes called HLA-A, HLA-B and HLA-C [Wood, 2011, p. 75]. Each gene encodes an -chain that pairs with 2-microglobulin [Wood, 2011, p. 75]. They are unique in the extent of their polymorphism, exhibiting hundreds to thousands of alleles [reviewed in Donadi et al., 2011], which makes it difficult for pathogens to evolve antigens that do not bind any of the HLA molecules [Wood, 2011, p. 88].

Non-classical HLA class I molecules – HLA class Ib HLA class Ib consists of three molecules - HLA-E, HLA-F and HLA-G [reviewed in Kochan et al., 2013]. They are called non-classical since they differ from the classical HLA class I molecules by their genetic diversity, structure, expression and function [reviewed in; Carosella et al., 2008 & Kochan et al., 2013]. A more detailed description of each is presented below.

13

Human Leukocyte Antigen E (HLA-E) The HLA-E gene consists of eight exons and only four alleles of HLA-E have been described [reviewed in Kochan et al., 2013]. However, the existence of two of these could not be confirmed in recent studies [reviewed in Kochan et al., 2013]. The HLA-E protein is membrane-bound,

consisting

of

three

MHC

immunoglobulin-like

α

domains,

a

transmembrane domain and a cytoplasmic tail [reviewed in Kochan et al., 2013]. Like the classical HLA class I molecules, HLA-E forms a complex with 2-microglobulin [reviewed in Kochan et al., 2013]. HLA-E is transcribed in most cells, but protein expression is mainly observed on endothelial cells, T and B lymphocytes and macrophages [reviewed in Kochan et al., 2013]. Although the limited polymorphism of HLA-E clearly distinguishes it from the classical HLA molecules, HLA-E is able to bind and present peptides [reviewed in Adams & Luoma, 2013]. In fact, peptide binding is required for its stabilization of cell surface expression [reviewed in Adams & Luoma, 2013]. The peptides presented by HLA-E are derived from the leader sequences of classical HLA proteins or HLA-G [reviewed in: Hviid, 2006 & Adams & Luoma, 2013]. HLA-E is also observed in a secreted form (sHLA-E) and although the mechanism for HLA-E secretion is not known, it is a result of HLA-E upregulation [reviewed in Kochan et al., 2013]. HLA-E is, as well as HLA-G, expressed in placental tissue and plays a role in enabling the mother’s immune system to accept the fetus [reviewed in Hunt, 2006]. The primary function of HLA-E seems to be in the innate immune system, where it is recognized by inhibitory or activating immune receptors expressed by cytotoxic T lymphocytes, NK cells [reviewed in Rodgers & Cook, 2005]. The antigen presentation to cytotoxic T lymphocytes and NK cells play an important role in for example antiviral defense [reviewed in Kochan et al., 2013; reviewed in Adams & Luoma, 2013]. Some bacterial and viral infections down-regulate production of HLA class Ia proteins and this limits the availability of peptides for HLA-E to bind [reviewed in Adams & Luoma, 2013]. This leads to unstable HLA-E expression and thereby the NK cells will not obtain the inhibitory signal from HLA-E and they will be able to detect and attack the cell [reviewed in Adams & Luoma, 2013].

Human Leukocyte Antigen F (HLA-F) HLA-F was identified in 1990 by Geraghty and colleagues. The gene consists of eight exons and the protein is organized in a similar way as HLA-E, forming a complex with 214

microglobulin [reviewed in Kochan et al., 2013]. Three splicing variants of HLA-F have been described, leading to synthesis of three HLA-F isoforms that differ from each other in the length of the cytoplasmic tail [reviewed in Kochan et al., 2013]. Its cytoplasmic domain is shorter than that of the HLA class Ia molecules because exon 7 is excluded from the mRNA [Geraghty et al., 1990]. No studies have reported, whether HLA-F is capable of mediating antigen presentation [reviewed in Kochan et al., 2013]. Since HLA-F is expressed on extravillous trophoblast cells, it is possible that HLA-F, like HLAG and HLA-E, has a role in the modulation of immune responses during pregnancy [Ishitani et al., 2003]. HLA-F has been detected both as an intracellular protein and as a cell surface protein. It was first detected intracellularly in different cells and tissues, e.g. in peripheral blood B cells, B cell lines, spleen, bladder and skin [Lepin et al., 2000; Wainwright et al., 2000]. Later, it was observed that HLA-F also could be expressed on the cell surface of e.g. all activated B cells, NK cells and monocytes [Lee & Geraghty 2003; Lee et al., 2010].

Human Leukocyte Antigen G (HLA-G) HLA-G consists of eight exons like HLA-E and HLA-F (see Figure 3) [reviewed in Hviid, 2006]. However, exon 7 is always spliced out in the RNA and a stop codon in exon 6 results in an untranslated exon 8 [reviewed in Donadi et al., 2011]. The final part of exon 6 and exon 8 is called the 3´untranslated region (3´UTR) of the gene [reviewed in Donadi et al., 2011]. Exon 1 encodes a signal peptide, each of exons 2-3 encode an α-domain and exon 5 and 6 encode the transmembrane and the cytoplasmic domain [reviewed in Donadi et al., 2011]. In contrast to the highly polymorphic HLA class Ia molecules, the polymorphic diversity of HLA-G is limited to 46 alleles [reviewed in; Dahl & Hviid, 2012 & Kochan et al., 2013]. It can generate seven alternatively spliced mRNAs, four membrane-bound (HLA-G1-4) and three secreted isoforms (HLA-G5-7) (see Figure 3) [reviewed in Rebmann et al., 2007]. Only HLA-G1 and HLA-G5 contain a peptide-binding region and are the only isoforms able to present peptides [reviewed in; Amiot et al., 2011 & Kochan et al., 2013]. HLA-G1 represents the sole full-length version of the molecule and can be released as a soluble protein by proteolytic cleavage [reviewed in Kochan et al., 2013]. Similar to HLA-E, HLA-F and the HLA class Ia molecules, HLA-G1 and -G5 forms a complex with 2-microglobulin [reviewed in Kochan et al., 2013].

15

Figure 3: Structure of the HLA-G gene, mRNA and protein [modified from Yan, 2011]. The HLA-G gene is composed of eight exons, of which exon 1 encodes a signal peptide, each of exons 2-3 encode an α-domain and exon 5 and 6 encode the transmembrane and the cytoplasmic domain [reviewed in Hviid, 2006; Donadi et al., 2011]. The HLA-G RNA is alternatively spliced to yield seven transcripts encoding four membrane-bound (HLA-G1-4) and three soluble isoforms (HLA-G5-7) of HLA-G [reviewed in Hviid, 2006; Donadi et al., 2011]. Exon 7 is always spliced out from the RNA and a stop codon in exon 6 results in an untranslated exon 8 [reviewed in Donadi et al., 2011]. The final part of exon 6 and exon 8 is called the 3´untranslated region of the gene (3´UTR) [reviewed in Hviid, 2006; Donadi et al., 2011]. HLA-G1 and 5 contain all three α-domains and are non-covalently bound to β2-microglobulin [reviewed in Hviid, 2006; Donadi et al., 2011]. HLA-G2 and HLA-G6 contain the α1 and α3 domains, HLA-G3 and HLA-G7 only contain the α1 domain and HLA-G4 contains the α1 and α2 domains [reviewed in Hviid, 2006; Donadi et al., 2011].

The expression of HLA-G is modulated by many factors, including promoter efficiency, which can be altered by polymorphisms in the promoter region, and the rate of mRNA translation and degradation, which can be altered by polymorphisms in the 3’UTR [reviewed in: Hviid, 2006 & Donadi et al., 2011]. A major polymorphism in this region is the presence or absence (insertion or deletion) of a 14 bp fragment [reviewed in Hviid, 2006]. This polymorphism is called the 14 bp ins/del (rs66554220) polymorphism and it has been associated with stability of the mRNA and thereby the level of HLA-G expression [reviewed in Hviid, 2006]. However, the exact mechanism for this is not known [reviewed in: Donadi et al., 2011 & Dahl & Hviid, 2012]. The 14 bp insertion polymorphism has been associated with decreased levels of HLA-G in trophoblasts [reviewed in Hviid, 2006]. The mRNA containing the 14 bp insertion can be

16

further processed by removal of 92 bases from the mature mRNA [reviewed in: Hviid, 2006 & Donadi et al., 2011]. This results in a shorter transcript that has been shown to be more stable than the full length +14 bp mRNA [Rousseau et al., 2003; Svendsen et al., 2013]. HLA-G was first reported to be restricted to the placenta, where it contributes to fetomaternal tolerance by suppressing local immune responses, protecting the fetus against the immune system of the mother [reviewed in; Hviid, 2006, Carosella et al., 2008]. Later, it has been discovered to be expressed constitutively in immune privileged organs, such as the cornea, thymus, pancreatic islets, endothelial cell precursors and erythroblasts as well [reviewed in Carosella et al., 2008]. In healthy individuals, transcription of HLA-G takes place in most cells, but the protein expression is restricted to specific tissues [reviewed in Kochan et al., 2013]. It can be induced under pathological conditions such as multiple sclerosis, inflammatory diseases, viral infections and cancer [reviewed in Carosella et al., 2008].

17

Expression of HLA class Ib in cancer For a wide variety of malignant diseases, down-regulation or loss of HLA class Ia expression is observed [Geertsen et al., 1998, Demanet et al., 2004; del Campo et al., 2014; reviewed in Aptsiauri et al., 2013]. Various mechanisms contribute to this down-regulation or loss, one mechanism, described by del Campo et al., (2014) is by loss of β2-microglobulin, which is important for effective transport of HLA class Ia molecules to the cell surface [Vitiello et al., 1990]. The down-regulation or loss of HLA class Ia molecules renders the tumors less susceptible to attack from the adaptive immune system [reviewed in: Rebmann et al., 2007 & Aptsiauri et al., 2013]. However, NK cells recognize and attack cells without HLA class I molecules [Maio et al., 1991]. Expression of HLA class Ib molecules by tumor cells is thought to be a mechanism by which they prevent this [Paul et al., 1998]. In accordance, expression of HLA class Ib molecules is observed in various types of cancer as described in the following chapters. Functional aspects of this expression will be described in the chapter “The interaction of HLA class Ib with immune cells”.

HLA-E expression in cancer

HLA-E expression has been reported in several types of cancer tissue and cancer cell lines [Sasaki et al., 2014; Benevolo et al., 2011; Allard et al., 2011; Spaans et al., 2012; Marín et al., 2003; Lo Monaco et al., 2011], but the relevance of this expression is debated. Benevolo et al. (2011) found that high expression of HLA-E in colorectal carcinoma was associated with a favorable prognosis. This was supported by Spaans et al. (2012), who found that high expression in cervical cancer was associated with favorable long term and recurrence-free surival. High HLA-E expression was also found in a significant proportion of colorectal tumor biopsies compared to normal mucosae [Levy et al., 2008]. However, the study by Levy et al. (2008) suggested that HLA-E renders the tumor less susceptible to immune attack and that it could be a marker of shorter disease-free survival. In addition to high expression of membrane-bound HLA-E, increased sHLA-E was also found in sera of cancer patients compared to healthy subjects [Allard et al., 2011].

HLA-F expression in cancer

HLA-F mRNA has been detected in various cancer cell lines [Noguchi et al., 2004], and HLA-F protein expression has been observed in clinical cancer tissues of lung cancer, gastric cancer and esophageal cancer [Ishigami et al., 2013; Zhang et al., 2013a; Zhang et al., 2013b; Lin et al., 18

2011]. To our knowledge, only four studies have investigated HLA-F expression in cancer tissue. Three of the studies found that cancer patients with HLA-F-positive tumors had significantly worse prognosis compared to the patients with HLA-F-negative tumors [Lin et al., 2011; Ishigami et al., 2013; Zhang et al., 2013a]. In contrast, a fourth study did not observe any correlation between HLA-F expression and prognosis of the cancer patients [Zhang et al., 2013b].

HLA-G expression in cancer

HLA-G expression has been observed on tumor cells and tumor-infiltrating cells in many types of primary tumors, metastases and in malignant effusions [reviewed in Rouas-Freiss et al., 2014]. HLA-G is often expressed in tumor tissue but not in healthy surrounding tissue, indicating an association with malignant transformation [Adithi et al., 2006; reviewed in Rouas-Freiss et al., 2014]. It is often found expressed in solid tumors of high histological grades and advanced clinical stages [reviewed in Rouas-Freiss et al., 2014]. Among others, the expression has been found in tumor tissue from malignant melanoma [Paul et al., 1998], gastric cancer [Du et al., 2011], renal cell carcinoma [Ibrahim et al., 2001; Li et al., 2009], breast carcinoma [Lefebvre et al., 2002; Palmisano et al., 2002], glioblastoma [Wastowski et al., 2013], non-small cell lung cancer [Yie et al., 2007], cutaneous B-cell lymphomas [Urosevic et al., 2002], and B-chronic lymphocytic leukemia [Nückel et al., 2005]. Besides expression in primary tumors, metastases, and malignant effusions, increased levels of sHLA-G are found in blood of cancer patients compared to healthy subjects [Ugurel et al., 2001; Sebti et al., 2003; Rebmann et al., 2003; Gros et al., 2006; Cao et al., 2011; Jeong et al., 2014]. Furthermore, a study by Zhu et al. (2011) found that the soluble level of HLA-G could be an indicator distinguishing colorectal cancer from benign colorectal diseases. As described, HLA class Ib expression has been observed in a variety of malignant diseases but our focus in this thesis is mainly on skin cancer. Therefore, a description of skin cancer and two types that we focus on, basal cell carcinoma (BCC) and malignant melanoma, follows.

19

Skin cancer Skin cancer is one of the most frequent types of cancer in Denmark [Clemmensen et al., 2006, p. 12-13]. The different cancers of skin are often divided in malignant melanoma and nonmalignant melanomas, where the non-malignant melanomas rank highest on the list, as the most common cancer type for men, and the second most common cancer for women [Clemmensen et al., 2006, p. 68]. Malignant melanoma is in top ten of the most common cancers in Denmark [Clemmensen et al., 2006, p. 66]. In this thesis, we investigate two different kinds of skin cancers: basal cell carcinoma (BCC), which is a “non-malignant melanoma” skin cancer, and malignant melanoma. BCC originates in the lowest layer of the epidermis and is the least dangerous skin cancer [reviewed in de Gruijl, 1999]. It is relatively easy to treat but often relapse [reviewed in de Gruijl, 1999 & in Madan et al., 2010]. In Denmark, more than 10,000 new incidences occur each year [Kræftens Bekæmpelse (a), 09.06.14]. Malignant melanoma originates from melanocytes, the pigmentproducing cells, and is the least common, but most aggressive skin cancer [reviewed in de Gruijl, 1999]. If untreated, there is a high risk of metastasis with fatal outcome [reviewed in Diao & Lee, 2013]. In Denmark, more than 1,900 new incidences occur each year and about 160 cases have fatal outcome [Kræftens Bekæmpelse (b), 09.06.14]. The incidence of both cancers increases with age [Kræftens Bekæmpelse (a) & (b), 09.06.14]. We have chosen to investigate these two common types of cancers, due to their remarkable difference in aggressiveness. Below is a more detailed description of the two cancers.

Basal cell carcinoma BCC accounts for 75% of all skin cancers in the white population [reviewed in Dessinioti et al., 2010] and the incidence is increasing worldwide due to an aging population and sun exposure habits [reviewed in Kasper et al., 2012]. The etiology is unclear, but there is a strong association between BCC and exposure to UV radiation [reviewed in Kasper et al., 2012], which is believed to be the predominant causative risk factor [reviewed in Dessinioti et al., 2010]. The origin of BCC is controversial, but the phenotypic appearance of BCC is very diverse, which indicates that the cancer may originate from a stem or progenitor cell [reviewed in Kasper et al., 2012]. Although it can be locally aggressive, causing extensive tissue damage [reviewed in Tilli et al., 2005], it is a rather benign tumor, since metastases are 20

largely absent [reviewed in Kasper et al., 2012]. BCCs can be divided into three main types: nodular, superficial and morphoeic [Skin Cancer Clinic, 12.06.14]. Nodular BCC is the most common subtype and accounts for about 60% of the cases [LWW Oncology, 12.06.14]. The second most common subtype is the superficial BCC that accounts for about 15% of the cases [LWW Oncology, 12.06.14]. The less common morphoeic type is an aggressive type of BCC and is also called sclerosing BCC [Skin Cancer Clinic, 12.06.14]. This subtype is an aggressive tumor because it is locally invasive, have illdefined boarders and a high recurrence rate [Skin Cancer Clinic, 12.06.14 & Zagrodnik et al., 2003].

Studies investigating HLA class Ib expression in BCC To our knowledge, no studies have investigated HLA-E or HLA-F expression in BCC. Only two studies have investigated HLA-G expression in BCC: Urosevic et al. (2005) and Aractingi et al. (2003). Both have investigated the HLA-G expression by immunohistochemical staining. Urosevic et al. (2005) investigated the HLA-G expression in 38 primary BCCs that underwent radiotherapy and 14 secondary BCCs recurring on the primary site after radiotherapy. They found that collectively HLA-G was expressed in 90% of the tumors and that 60% displayed strong HLA-G expression. The HLA-G expression was decreased in secondary tumors, since 96% of the primary tumors expressed HLA-G while 86% of the secondary tumors expressed HLA-G. In 17% of the cases, all of which were primary tumors, they also observed HLA-G expression in tumor-infiltrating mononuclear cells. Nodular BCCs had the strongest HLA-G expression, but expression was most frequently observed in sclerosing BCCs. They suggest that

HLA-G

expression

might

associate

with

aggressive

phenotype.

The second study by Aractingi et al. (2003) investigated HLA-G expression in tissue from organ transplant recipients. They detected HLA-G expression in various types of epithelial skin cancers, but observed HLA-G expression in only 1 of 7 of the BCCs. Furthermore, no HLA-G expression was observed in the benign tumors studied. In many cases the benign and malignant tumors were obtained from the same patients. Therefore, it seems that HLA-G expression is restricted to malignant tumors. The results from the two studies are very diverse as the fraction of HLA-G-expressing BCCs found by Aractingi et al. (2003) is very low compared to the results of Urosevic et al. (2005). The diversity might be due to the difference in the number of cases investigated. Second, the study by Aractingi et al. (2003) only includes

21

transplant recipients, who have gone through immunosuppressive treatments that could possibly have an influence on the expression of HLA-G. It can be hypothesized, that the expression of HLA-G is a contributing factor for the higher risk of malignancies in these patients. But this may not be the case for BCCs in general since only 1 of 7 showed HLA-G expression. More studies are needed to better characterize the expression of HLA-G in BCCs.

Cutaneous malignant melanoma Malignant melanoma develops from the melanin-producing melanocytes located in the stratum basale of the epidermis [reviewed in: Ingraffea, 2013 & Olbryt, 2013]. It is the 19 th most common cancer worldwide [Cancer Research UK (a), 12.06.14] and one of the most aggressive tumors [reviewed in Olbryt, 2013]. In Europe, Danish women have the highest agestandardised incidence rate [Cancer Research UK (a), 12.06.14]. Overall, the incidence is increasing, some due to increased surveillance, but also as mentioned for the BCCs, changes in sun-related behaviour [Cancer Research UK (b), 12.06.14]. An important factor of malignant melanoma etiology is UV-radiation [reviewed in Olbryt, 2013]. A study of malignant melanoma patients in the UK estimated that about 86% of the cases were linked to exposure of UV-radiation from sun and sunbeds [Cancer Research UK (b), 12.06.14]. In approximately 10% of the cases malignant melanoma is due to the influence from a familiar background [reviewed in Olbryt, 2013]. Several histological types of melanoma exist, the most common types being superficial spreading melanoma, nodular melanoma, acral lentiginous melanoma and lentigo maligna melanoma [Cancer Research UK (c), 12.06.14]. Superficial spreading melanoma is by far the most common of them all accounting for about 70% of all melanoma cases [Cancer Research UK (c), 12.06.14; Skin Cancer Foundation, 12.06.14]. It grows along the top layer of the skin but can become invasive [Cancer Research UK (c), 12.06.14]. Lentigo maligna melanoma and acral lentigious melanoma also grow in the top layer of the skin before invasive growth [Skin Cancer Foundation, 12.06.14]. Both grow very slowly and each account for 5-10% of melanoma cases [Cancer Research UK (c), 12.06.14 & Sundhed.dk, 12.06.14]. Nodular melanoma, which accounts for 10-15% of melanoma cases, is the most aggressive type of melanoma [Cancer Research UK (c), 12.06.14 & Skin Cancer Foundation, 12.06.14]. This cancer is fast-growing and is often invasive at the time of diagnosis [Skin Cancer Foundation, 12.06.14].

22

Studies investigating HLA class Ib expression in malignant melanoma Paul et al. (1998) were the first to discover an “aberrant” HLA-G mRNA and protein expression in malignant melanoma, which lead to a serial of studies investigating the role of HLA class Ib in the immunobiology of tumors. Many studies have investigated the expression of HLA-G and some the expression of HLA-E in malignant melanoma. To our knowledge, none have investigated expression of HLA-F. Malignant melanoma is the most tested tumor type in the analysis of HLA-G expression when it comes to both cell lines and surgically removed lesions [reviewed in Chang & Ferrone, 2003]. The expression of HLA-G on transcriptional and/or protein level has been analyzed in hundreds of surgically removed malignant melanoma lesions [reviewed in Chang & Ferrone, 2003]. The majority of studies found protein expression of HLA-G in many of the malignant melanoma lesions [Paul et al., 1999; Wagner et al., 2000; Ibrahim et al., 2004; Fang et al., 2008; Bezuhly et al., 2008]. However, other studies have not been able to detect protein expression of HLA-G in malignant melanoma lesions [Real et al., 1999; Frumento et al., 2000]. The two studies that did not detect HLA-G used HLA-G mAb 87G also used by Paul et al. (1999), who did detect HLA-G protein with this antibody. Thus, the conflicting results were not due to use of different HLA-G antibodies. HLA-G is present in primary and metastatic tumor sites, but not in healthy skin or in tumor regression sites [Paul et al., 1999; Fang et al., 2008; Ibrahim et al., 2004]. When studying HLA-G expression in malignant melanoma cell lines, it is only expressed in a minority of these [reviewed in Rebmann et al., 2007]. Interestingly, HLA-G mRNA and protein expression in malignant cell lines is generally low compared to tumor tissue [Du et al., 2011; reviewed in Rebmann et al., 2007]. It is important to note, that long-term culturing of cell lines derived from HLA-G-positive primary tumors can result in loss of HLA-G expression [Rouas-Freiss et al., 2005]. Rouas-Freiss discovered that expression of HLA-G1 started to wane around passage 66, and was completely lost at passage 70. It seems that tumor cells require certain factors in their microenvironment for the induction and maintenance of HLA-G, which is not present in culture medium. Both mRNA and protein expression of HLA-E is also observed in some malignant melanoma tissues and cell lines [Marín et al., 2003; Palmisano et al., 2005; Derré et al., 2006; Tremante et al., 2014]. However, the studies are conflicting regarding the level of HLA-E expression and its correlation with tumor progression. Results of Tremante et al. (2014) indicates that expression of HLA-E is up-regulated during tumor progression. They found that HLA-E 23

expression was largely absent in non-malignant tissue, increased in malignant melanoma tissue and further increased in metastases. In contrast to this, another study showed that HLA-E was expressed in healthy skin, that the expression was down-regulated in malignant melanoma tissue and further down-regulated in metastases [Derré et al., 2006]. It has been reported that melanoma cell lines produce soluble HLA-E and that it is generated by proteolytic cleavage of membrane-bound HLA-E [Derré et al., 2006]. The level of soluble HLA-G and HLA-E has also been investigated in melanoma patients. A study by Ugurel et al. (2001) found that the level of sHLA-G was significantly elevated in melanoma patients compared to healthy controls. The elevated level of sHLA-G did not seem to have impact on the prognosis of the patients [Ugurel et al., 2001]. Another study found that sHLA-E was significantly increased in sera of patients suffering from melanoma compared to healthy individuals, but they did not find a significant association between serum sHLA-E levels and melanoma stage [Allard et al., 2011].

24

Immunoediting The interaction between tumors and the immune system is very complex and not fully understood. In the following, an introduction to the interplay between tumors and the immune system is provided to help the understanding of how HLA class Ib molecules might contribute to tumor evasion from the immune system. In the early 1900´s Paul Ehrlich proposed the idea that the immune system is able to repress the development of cancer [reviewed in Dunn et al., 2002]. Not before the mid 1900’s this proposal was supported by the discovery of tumor specific antigens, which are important for the “cancer immunosurveillance” hypothesis formulated by Barnet and Thomas [reviewed in Dunn et al., 2002]. This hypothesis states that the tumor specific antigens on the cancer cells elicit an immune response that is able to eliminate the cancer cells and both Burnet and Thomas suggested that lymphocytes could recognize and eliminate continuously arising cancer cells [reviewed in Dunn et al., 2002]. Since this, several studies have confirmed that tumors are able to elicit immune responses against the tumor cells [reviewed in Tanchot et al., 2012]. However, a suppressive microenvironment develops in many tumors and the immune responses against the tumors are not sustained [reviewed in Tanchot et al., 2012]. According to the “cancer immunosurveillance” hypothesis, the immune system protects the host by eliminating developing tumors but it is now recognized that the interaction between the immune system and cancer cells is more complex than first assumed and that the interaction not only occurs during early tumor development [reviewed in Ikeda et al., 2002]. Although the exact molecular mechanisms and interactions between cells are not clear, some growing tumors are eliminated by immune cells, while others are immunetolerant [reviewed in Dunn et al., 2002]. To describe this, Ikeda and colleagues (2002) refined the “cancer immunosurveillance” hypothesis and introduced the model “cancer immunoediting”. Cancer immunoediting is a process of three steps: elimination, equilibrium and escape (see Figure 4). Immunosurveillance, where tumor cells are eliminated by cytotoxic T and NK cells, occurs in the elimination step, whereas tumor cells with reduced immunogenicity are selected in the equilibrium step, which results in an increased chance of tumor escape which allows tumor survival and maintenance [reviewed in Ikeda et al., 2002].

25

Figure 4: The three steps of cancer immunoediting [from Dunn et al., 2002]. The cancer immunoediting is divided in three steps: elimination, equilibrium and escape [reviewed in Dunn et al., 2002]. Elimination: normal cells (gray) are due to oncogenic stimuli transformed into tumor cells (the blue ones are developing tumors and the red ones are tumor cell variants) [reviewed in Dunn et al., 2002]. The tumor cells express tumor specific markers and generate proinflammatory signals, which are recognized by immune cells, initiating recruitment of macrophages, lymphocytes, NK and dendritic cells, initiating the first phase of elimination [reviewed in Dunn et al., 2002]. The immune cells fight the growing cancer, by inducing production of chemokines and cytokines (orange dots), and through cytotoxic activity of cytotoxic CD8+ T cells and NK cells (white flashes) [reviewed in Dunn et al., 2002]. If this process is not successful, the tumor cell may enter the equilibrium phase [reviewed in Dunn et al., 2002]. Equilibrium: in this phase many of the original tumor cells are destroyed, but new variants arise carrying more mutations that provide them with increased resistance to immune attacks, leading some tumor cells to tumor escape [reviewed in Dunn et al., 2002]. Escape: the tumor evades the immune system and expands in an uncontrolled manner [reviewed in Dunn et al., 2002]. Additional tumor variants formed as a result of the equilibrium process are in the figure shown in orange [reviewed in Dunn et al., 2002].

The elimination step: Based on studies from several groups, Ikeda and colleagues (2002) presented a model of the elimination step in which there is four phases (see Figure 5).

26

Figure 5: The four phases of the elimination step of cancer immunoediting [from Dunn et al., 2002]. A) The transformed cells (blue) are recognized by cells of the innate immune response that are stimulated to produce IFN- [reviewed in Dunn et al., 2002]. B) The production of IFN- initiates a cascade of innate immune reactions involving induction of chemokines (including CXCL10 (IP-10), CXCL9 (MIG) and CXCLII (I-TAC)), which block neovascularization in the tumor and induce the recruitment of immune effector cells [reviewed in Dunn et al., 2002]. Furthermore, IFN- elicits antiproliferative actions on the developing tumor cells, and activates the cytotoxic activity of macrophages (Mac) and NK cells (NK) [reviewed in Dunn et al., 2002]. All of the above result in the death of some of the tumor cells (white and grey cells surrounded by a dashed black line) [reviewed in Dunn et al., 2002]. Dead tumor cells or debris (illustrated by blue squares) are ingested by dendritic cells (DC) and transported to the draining lymph node [reviewed in Dunn et al., 2002]. C) NK cells and macrophages inhibit the tumor growth by their cytotoxic activities. Meanwhile, tumor antigen specific CD4+ and CD8+ T cells are developed in the draining lymph node [reviewed in Dunn et al., 2002]. These cells home to the tumor where they destroy the cells expressing distinctive tumor antigens [reviewed in Dunn et al., 2002].

In the first phase of elimination, the tumor grows and as it reach a certain size, it begins to grow invasively [reviewed in Dunn et al., 2002]. Therefore, it needs to induce angiogenesis to enhance its blood supply [reviewed in Dunn et al., 2002]. The invasive growth of the tumor

27

induces inflammation as it causes damage to the surrounding tissue that induce recruitment of immune cells such as macrophages, NK and dendritic cells [reviewed in Dunn et al., 2002]. The inflammatory lymphocytes are then stimulated to produce IFN-γ by the recognition of structures on the tumor cells [reviewed in Dunn et al., 2002]. In the second phase of elimination, the produced IFN-γ may induce death of some of the tumor cells by inhibiting proliferation and induce apoptosis [reviewed in Dunn et al., 2002]. Furthermore, it induces the tumor cells and surrounding tissue to produce the chemokines CXCL9 (MIG), CXCL10 (IP10) and CXCL11 (I-TAC) [reviewed in Dunn et al., 2002]. These – or at least some of these – chemokines have a capacity to block angiogenesis and this induces dead of more tumor cells [reviewed in Dunn et al., 2002]. The cell debris is ingested by dendritic cells that end up in draining lymph nodes [reviewed in Dunn et al., 2002]. During the escalation of the inflammatory response, chemokines are produced that recruit more NK cells and macrophages [reviewed in Dunn et al., 2002]. In the third phase of elimination, these cells activate each other by producing IFN-γ and IL-12 and kill more of the tumor cells by mechanisms involving tumor necrosis factor-related apoptosis-inducing ligand, perforin and reactive oxygen and nitrogen species [reviewed in Dunn et al., 2002]. In the draining lymph nodes, the dendritic cells induce tumor antigen specific CD4+ T helper cells that express IFN-γ [reviewed in Dunn et al., 2002]. These cells then facilitate the development of tumor specific CD8+ T cells [reviewed in Dunn et al., 2002]. In the fourth phase of elimination, these tumor specific CD4+ and CD8+ T cells migrate to the tumor site where the cytolytic CD8+ T cells attack the tumor cells, whose immunogenicity has been enhanced because of exposure to IFN-γ [reviewed in Dunn et al., 2002]. The equilibrium step: In this step, the remaining tumor cells and the immune cells enters a dynamic equilibrium where the lymphocytes and INF-γ exerts a selective pressure on the tumor cells [reviewed in Dunn et al., 2002]. This is enough to contain the rapidly mutating tumor cells [reviewed in Dunn et al., 2002]. Most of the tumor cells will be destroyed by the immune system but some of the tumor cell mutations will decrease their immunogenicity and result in increased resistance to immune attacks [reviewed in Dunn et al., 2002]. The escape step: In the last step, tumor cells that have not been eliminated start to grow in an uncontrolled manner and result in a clinically observable tumor [reviewed in Dunn et al., 2002].

28

Two of the immune cells involved in immunoediting will be described in the following chapters: “Regulatory T cells (Tregs)” and “Dendritic cells (DCs)”. Our focus is on these two, since it is the two cell types, which we will study further in the experimental part of this thesis.

29

Regulatory T cells (Tregs) In healthy individuals Tregs are important for the maintenance of self tolerance, preventing autoimmune reactions, as they suppress auto reactive T cells [reviewed in Nishikawa & Sakaguchi, 2010]. It has been shown that impairment of Tregs can lead to autoimmune diseases [reviewed in Nishikawa & Sakaguchi, 2010]. The exact mechanisms by which Tregs exert their immunosuppressive functions are not fully understood [reviewed in Nishikawa & Sakaguchi, 2010]. However, such mechanisms could be through cell-cell contact with DCs or effector T cells or by secretion of immunosuppressive cytokines such as IL-10 and transforming growth factor β (TGF-β) [reviewed in Nishikawa & Sakaguchi, 2010]. These cytokines are able to induce cell cycle arrest, apoptosis and influence B and T cell maturation [Bailey et al., 2006; reviewed in Kubiczkova et al., 2012]. Tregs are CD4+CD25+ lymphocytes expressing the transcription factor FoxP3, which is critical for their development, maintenance and function [Fontenot et al., 2003; reviewed in Oleinika et al., 2013]. Typically 5-10% of CD4+ T cells are Tregs but this fraction can be significantly elevated in tumor tissue and peripheral blood where Tregs can constitute 20-30% of the CD4+ cells [Gobert et al., 2009, Curiel et al., 2004; reviewed in Oleinika et al., 2013]. There are two major types of Tregs: natural Tregs (nTregs) and inducible Tregs (iTregs). The nTregs are developed during T cell development in the thymus. Here, some of the T cells that are specific for self-antigens during negative selection will undergo apoptosis (clonal deletion), while others will be induced to differentiate into nTregs [Wood, 2011, p. 244]. iTregs develops from naïve CD4+ T cells when they interact with tolerogenic antigenpresenting cells (APCs) [reviewed in Oleinika et al., 2013], such as the subtype DC-10, which will be described in the “Dendritic cells (DCs)”. Immunosuppression by Tregs is a major mechanism whereby tumors escape immune responses [reviewed in Tanchot et al., 2012]. The roles of the iTregs and the nTregs subset in tumor specific immune tolerance are not clear but it is likely that they both contribute to the immune tolerance [reviewed in Oleinika et al., 2013]. It seems that there are different mechanisms in which Tregs can accumulate in tumor tissues, here we will divide them into three mechanisms: 1) increased trafficking, 2) expansion of already existing Tregs, and 3) conversion of FoxP3- T cells into Tregs.

30

Increased trafficking: Some studies suggest that Tregs have a higher capacity to infiltrate tumor tissue than effector T cells and it has been observed that it is preferentially Tregs that are recruited to the tumor [reviewed in Oleinika et al., 2013]. The recruitment of Tregs occurs by chemotaxis [reviewed in Ménétrier-Caux et al., 2012]. Tregs express the chemokinereceptors CCR4 and CCR8 and they are attracted by the CCR4 ligand that has been shown to be produced by tumor cells and tumor-infiltrating macrophages [Iellem et al., 2001; Mizukami et al., 2008; Gobert et al., 2009; Maruyama et al., 2010; reviewed in Oleinika et al., 2013]. It is also shown that during hypoxia the chemokine CCL28 can be expressed by tumor cells and it is able to attract Tregs expressing the receptor CCR10 [Facciabene et al., 2011]. Expansion of already existing Tregs: Another mechanism, in which Tregs can accumulate in tumor tissues, is trough expansion of Tregs that are already in the tumor or in the tumordraining lymph nodes [reviewed in Oleinika et al., 2013]. It is proposed that IL-2, which is essential to development and homeostasis of Tregs, secreted by effector T cells in the tumor can promote proliferation of Tregs [reviewed in Oleinika et al., 2013]. A nuclear protein expressed by proliferating cells has been detected in some tumor-infiltrating Tregs, demonstrating proliferative potential [reviewed in Tanchot et al., 2012]. Conversion of FoxP3- T cells into Tregs: The last mechanism by which Tregs are able to accumulate in tumors is by conversion of FoxP3- T cells into Tregs [reviewed in Oleinika et al., 2013]. It is shown that TGF-β secreted by tumor cells can contribute to induction of Tregs [Liu et al., 2007]. Also indoleamine 2,3-dioxygenase has been shown to induce Tregs, as well as activate Tregs in tumor-draining lymph nodes [Curti et al., 2007; Sharma et al., 2007]. It is not clear which of the mechanisms contribute the most to Tregs accumulation in tumors and it may also vary between different cancers and between timepoints in the disease [reviewed in Oleinika et al., 2013].

Clinical relevance of Tregs in cancer

Studies have observed a correlation between the level of Tregs and prognosis of cancer patients. A study by Badoual et al. (2006) suggested that an increased level of tumorinfiltrating Tregs is associated with good prognosis in head and neck squamous cell carcinoma, but most studies found that a high amount of tumor-infiltrating Tregs is associated with poor prognosis in human malignant tumors [Sasada et al., 2003; Curiel et al. 2004; Fu et

31

al. 2007; Bates et al., 2006; Du et al., 2011]. An example is the study by Du et al., (2011), who found that gastric cancer patients with high level of Tregs had a significantly poorer survival five years after operation compared to those with low Treg levels (see Figure 6).

Figure 6: [from Du et al., 2011]. The figure illustrates the relationship between high and low levels of Tregs in relation to survival rate. Patients with high levels of Tregs had significantly poorer survival five years after operation [Du et al., 2011].

In accordance to the observation that high level of Tregs correlate with worse prognosis, it has been demonstrated that suppression or depletion of Tregs in tumors results in increased immune response against the tumor and inhibits tumor growth [Onizuka et al., 1999; Dannull et al., 2005; Yu et al., 2005].

32

Dendritic cells (DCs) Dendritic cells (DCs) play a major role in cancer immunosurveillance initiating the antitumor immunerespons as described in the chapter “Immunoeediting”. DCs are the most important antigen presenting cell (APC), playing a role in the link between innate and adaptive immune responses [reviewed in: Carosella et al., 2011 & Gregori, 2010]. Immature DCs circulate in the peripheral blood where they are characterized by low expression of HLA class II molecules and co-stimulatory molecules [reviewed in Gregori, 2010]. Upon encounter with foreign antigens, they are activated, increasing their expression of HLA class II molecules and co-stimulatory molecules, which enable them to effectively activate T cells [reviewed in Carosella et al., 2011]. Furthermore, they up-regulate the CCR7, which enable them to migrate to the spleen or lymph nodes [reviewed in Gregori, 2010]. On arrival they present the foreign antigen through either HLA class I or II [Wood, 2011, p. 141, 196]. Antigens presented by HLA class I activates CD8+ T cells, while antigens presented by HLA class II activates CD4+ T cells [Wood, 2011, p. 141, 196]. Furthermore, DCs are involved in central and peripheral tolerance. The mechanism of central tolerance plays an essential role in the developing of T cells, eliminating T cells that are directed against antigens of the body’s own tissue [Wood, 2011, p. 243-245]. Peripheral tolerance is tolerance induced after the T cells have left the thymus [Wood, 2011, p. 246-248]. This can be by induction of anergy or induction of iTreg [Wood, 2011, p. 246-248]. Dendritic cells (DCs) are not a single cell type but a heterogenous group of cells that arise from different bone-marrow derived hematopoietic lineages [reviewed in Gregori, 2010]. DCs are divided into two main groups: myeloid DCs and plasmacytoid DCs [Posch et al., 2013, p. 32]. Plasmacytoid DCs: The plasmacytoid DCs are also called non-conventional DCs. They develope in the bone-marrow, and are distinguished from the myeloid DCs by not expressing high levels of CD11c and CD14 [Posch et al., 2013, p. 32, 41]. These cells are involved in antiviral immune mechanisms [reviewed in Gregori, 2010]. Myeloid DCs: Myeloid DCs, which are also called conventional DCs, differentiate in lymphoid or peripheral tissue from progenitor cells that have migrated from the bone-marrow [Posch et al., 2013, p. 32]. They are further divided in migratory DCs and lymphoid resident DCs, the

33

latter of which reside in the lymphoid organs and lack the ability to migrate [Posch et al., 2013, p. 32]. Many different immature migratory DCs, characterized by expression of CD11c exist circulating in the blood [reviewed in Gregori, 2010]. The major CD11c+ subtype is also CD16+. DC-10: A more newly discovered subtype of circulating DCs is called DC-10 [Gregory et al., 2010].

These

cells

constitute

about

0.3%

of

human

PBMCs

and

are

CD14highCD16highCD11c+CD11b+HLA-DR+CD83+CD163+ CD1a-CD1c- [Gregory et al., 2010]. Furthermore, they express HLA-G and spontaneously produce IL-10 [Gregory et al., 2010]. Oppose to the other myeloid DCs, they have low capacity for inducing proliferation of CD4+ T cells and inducing their INF-γ production [Gregory et al., 2010]. Furthermore, they secrete low levels of IL-12, which is important in the activation of T helper cells [Gregory et al., 2010; Heufler et al., 1996]. DC-10 cells have been shown to strongly induce differentiation of regulatory T cells in vitro, by a mechanism that is unknown but dependent on IL-10, HLA-G and ILT-4 [Gregory et al., 2010]. The DCs express the two inhibitory receptors ILT-2 and ILT-4 for which HLA-G is a ligand [reviewed in Carosella et al., 2011]. The interaction between the DCs and HLA-G will be described further in the chapter “HLA class Ib and dendritic cells”.

34

The interaction of HLA class Ib with immune cells In the following we will describe the effects of the HLA class Ib molecules on immune cells in the situation of cancer. With focus on cancer, HLA-G is clearly the most studied class Ib molecule, hence the majority of this chapter will be about HLA-G. An overview of the function of HLA-G on various immune cells is presented in “Figure 7” below. It is important to note that, even though a lot of interactions between the HLA class Ib molecules and immune cells in cancer have been described, the mechanisms behind are still uncertain.

Figure 7: Overview of the function of HLA-G on immune cells [modified from; Yan, 2011 & Pistoia et al., 2007]. HLA-G can be expressed on the cell surface of the tumor cell and/or released in soluble form [reviewed in Rebmann et al., 2007]. HLA-G can interact with inhibitory receptors on the immune cells: KIR2DL4, ILT-2 and ILT-4 [reviewed in Amiot et al., 2011]. As described in this chapter, HLA-G is able to inhibit immune cells by various mechanisms.

35

HLA class Ib and NK cells HLA-G is known to interact with two different receptors on NK cells; killer cell immunoglobulin like receptor 2DL4 (KIR2DL4) and immunoglobulin-like transcript receptor 2 (ILT-2) [reviewed in Amiot et al., 2011 & Rajagopalan & Long, 2012]. Both membranebound and soluble HLA-G can bind KIR2DL4, which is expressed on the cell surface or endosomes of NK cells [reviewed in Rajagopalan & Long, 2012]. HLA-G is able to up-regulate these inhibitory receptors on NK cells, which may be an advantage for HLA-G-expressing tumors [LeMaoult et al., 2005]. The interaction of HLA-G with inhibitory receptor ILT-2 on NK cells inhibits their production of IFN-γ [Favier et al., 2010]. IFN-γ is an important part of the elimination step of immunoediting, inhibiting proliferation and inducing apoptosis of tumor cells, as mentioned in the chapter “Immunoediting” [reviewed in Dunn et al., 2002]. In vitro, studies have shown that tumor cell lines expressing HLA-G or transfected with HLA-G are protected against cytolysis by NK cells [Rouas-Freiss et al., 1997; Cabestré et al., 1999; Lin et al., 2010; Favier et al., 2010]. This function has been detected for both membrane-bound and soluble HLA-G: in 2009, Lesport et al. (2009) showed that transfection with the soluble HLA-G5 protected a melanoma cell line against the cytotoxicity of natural killer leukemia cells. It has been shown, that inhibition of NK cytolysis is caused by impaired actin reorganization and impaired polarization of perforin granules towards the area of contact with target cells [Lesport et al., 2009; Favier et al., 2010]. Favier et al. (2010) demonstrated that this inhibitory effect of HLA-G was mediated trough ILT-2. One study has shown that sHLA-G is also able to inhibit chemotaxis of NK cells towards the chemokines CXCL10 and CXCL11 among others [Morandi et al., 2011]. This is interesting since IFN-γ induces tumor cells to secrete these chemokines during the elimination phase of immunoediting, as described in the chapter “Immunoediting” [reviewed in Dunn et al., 2002]. Another way of HLA-G to inhibit cytolysis is by counteracting the activating interaction of the ligand MICA, expressed by some tumor cells, with the receptor NKG2D on NK cells [Menier et al., 2002]. It seems that the interaction of HLA-G with the inhibitory receptor ILT-2 overrules the activation mediated by MICA [Menier et al., 2002]. HLA-G is also able to inhibit NK cytolysis indirectly through HLA-E. A peptide derived from the leader sequence of HLA-G binds and stabilizes HLA-E, enabling HLA-E to inhibit NK cytolysis through interaction with the inhibitory receptor CD94/NKG2A [reviewed in Curigliano et al., 2013]. It is shown, that

36

HLA-G not only has the ability to inhibit NK cells but is also able to induce apoptosis of these cells [Contini et al., 2003; Lindaman et al., 2006]. Caumartin et al. (2007) showed that activated NK cells could acquire HLA-G1 from tumor cells by trogocytosis, and that this caused the NK cells to stop proliferating, to no longer be cytotoxic, and to behave as supressor cells [Caumartin et al., 2007]. These could as well inhibit the cytotoxic function of other NK cells. In this way, the tumor can escape the cytotoxic NK cells [Caumartin et al., 2007]. However, the acquired expression of HLA-G1 is temporary, as these cells do not transcribe HLA-G [Caumartin et al., 2007]. HLA-E can bind the inhibitory NKG2A and the activating receptor NKG2C expressed by NK cells [reviewed in Kochan et al., 2013]. HLA-E binds stronger to NKG2A than to NKG2C [ValésGómez et al., 1999]. In accordance, it is demonstrated that high expression of HLA-E protects melanoma cells against NK cytolysis [Derré et al., 2006; Tremante et al., 2014]. The knowledge about the function of HLA-F in cancer is limited and to our knowledge, a function on NK cells has not been reported.

HLA class Ib and dendritic cells

Interactions between HLA-E or HLA-F with dendritic cell is, to our knowledge, not reported. HLA-G can interact with, as well as up-regulate, two receptors on dendritic cells, the inhibitory receptors ILT-2 and ILT-4 [LeMaoult et al., 2005; reviewed in Rouas-Freiss et al., 2014]. It has been shown, that HLA-G is able to induce tolerogenic DCs by altering the expression of HLA class II molecules [Ristich et al., 2005]. Upon interaction between HLA-G and DCs, several genes involved in the HLA class II presentation pathway were down-regulated, decreasing their ability to present antigens [Ristich et al., 2005]. Induction of tolerogenic DCs involved inhibition of maturation and differentiation of myeloid DC [Ristich et al., 2005]. The tolerogenic DCs had increased ability to induce anergic and immunosuppressive T cells [Ristich et al., 2005]. Inhibition of DC maturation has been detected by decreased expression of activation markers including HLA-DR and secretion of IL-12 [Ristich et al., 2005; Gros et al., 2008]. This decreases their ability to activate NK cells [Gros et al., 2008]. HLA-G can be expressed on, and secreted or shed by DC-10s [Gregori et al., 2010; reviewed in Carosella et al., 2011]. These DCs can be regarded as suppressor cells that contribute to

37

generation and maintenance of a tolerogenic microenvironment [Gregori et al., 2010]. DC-10s can alter the function of T cells by inducing generation of Tregs [Gregori et al., 2010].

HLA class Ib and B cells Considering the interaction of HLA class Ib molecules with B cells, we have only found few studies regarding HLA-G. HLA-G can inhibit proliferation, Ig secretion and chemotaxis of B cells [Naji et al., 2014; reviewed in Rouas-Freiss et al., 2014]. HLA-G inhibits B cell proliferation through interaction with the inhibitory receptor ILT-2 [Naji et al., 2012]. The inhibition of the proliferation seems to be caused by cell cycle arrest, and not by induction of apoptosis or necrosis [Naji et al., 2012].

HLA class Ib and T cells

HLA-G is able to up-regulate the inhibitory receptors ILT-2 and ILT-4 on T cells [LeMaoult et al., 2005]. As described for NK cells, chemotaxis of T cells is also influenced by sHLA-G [Morandi et al., 2010]. Soluble HLA-G down-regulated chemokine receptors on CD4+ and CD8+ T cells, that are attracted towards the chemokines CXCL10 and CXCL11 among others [Morandi et al., 2010]. HLA-G5 inhibits proliferation of CD4+ and CD8+ T cells by inhibition of cell cycle progression [Bahri et al., 2006]. Furthermore, HLA-G5 reduces cytokine production by activated T cells [Bahri et al., 2006]. HLA-G1 is also shown to inhibit proliferation of CD4+ T cells [LeMaoult et al., 2004]. APCs transfected with HLA-G1 can, through trogocytosis, inhibit proliferation of CD4+ T cells, induce CD4+ T cell anergy and induce the differentiation of CD4+ T cells into Tregs [LeMaoult et al., 2004]. This will lead to a missing immune response and an impaired generation of cytotoxic T cells. HLA-G can also directly, by interference with ILT-2 and ILT-4, inhibit the cytotoxicity of CD8+ cells [Le Gal et al., 1999; reviewed in Rouas-Freiss et al., 2014]. In addition, sHLA-G is able to induce apoptosis of activated CD8+ T cells (cytotoxic T cells) through a FAS/FAS-L dependent pathway [Contini et al., 2003; Fournel et al., 2000]. However, another study has found that HLA-G5 does not induce apoptosis of CD4+ or CD8+ T cells [Bahri et al., 2006]. HLA-E can interact with the T cell receptor of CD8+ T cells, which might induce CD8+ Tregs [reviewed in Kochan et al., 2013]. Regarding HLA-F no functional mechanisms have been described.

38

HLA class Ib and Tregs Studies of tumor immunology indicate that increased HLA-G expression often correlates with and increased level of Tregs [Chen et al., 2010; Du et al., 2011; Tuncel et al., 2013; Rizzo et al., 2014]. The precise mechanism by which HLA-G induces Tregs is unclear [reviewed in Carosella et al., 2011]. As mentioned, DC-10s are able to induce Tregs and this is dependent on HLA-G/ILT-2 interaction [Gregori et al., 2010]. Trogocytosis is one mechanism described by LeMaoult et al., (2007). CD4+ and CD8+ T cells can become temporary Tregs by acquiring HLAG1 by trogocytosis from HLA-G1 expressing APCs [LeMaoult et al., 2007]. In a functional study by Du et al. (2011) they found that the level of Tregs was essentially unchanged after indirect co-culture of PBMCs with HLA-G+ cells, contrary to the results of direct co-culture. This suggests that the effect of HLA-G on Tregs might be through direct cell-cell contact. For mesenchymal stem cells, the soluble HLA-G5 has been shown to contribute to expansion of CD4+CD25highFOXP3+ Tregs, indicating the existence of another mechanism of Tregs induction than by trogocytosis or through cell-cell direct contact [Selmani et al., 2008]. As mentioned, studies have observed a correlation between increased HLA-G expression and increased level of Tregs in cancer patients [Chen et al., 2010; Du et al., 2011; Tuncel et al., 2013; Rizzo et al., 2014]. Du et al. (2011) have investigated this correlation in tumor tissue and peripheral blood from gastric cancer patients. They detected HLA-G expression in about half of the tissue samples investigated, and found a significantly correlation between HLA-G expression and the presence of tumor infiltrating Tregs. In peripheral blood of the gastric cancer patients, they found elevated levels of sHLA-G compared to healthy controls correlating with elevated levels of Tregs [Du et al., 2011]. This correlation found in the gastric cancer tissue and plasma was supported by a functional assay [Du et al., 2011]. In the assay, gastric cancer cells, overexpressing HLA-G, were directly co-cultured with PBMCs. They found that the overexpression of HLA-G significantly enhanced the frequency of Tregs [Du et al., 2011]. Chen et al, (2010) has investigated the correlation between sHLA-G expression and Treg levels in the peripheral of ductal breast cancer. In accordance with the finding by Du et al. (2011), they found that sHLA-G was dramatically increased in the breast cancer patients, correlating with a markedly increase in the level of Tregs.

39

As mentioned in the previous chapter “HLA class Ib and T cells”, HLA-E might induce CD8+ Tregs through interaction with the T cell receptor [reviewed in Kochan et al., 2013]. To our knowledge, there are no studies regarding HLA-F and Tregs. The mechanism by which HLA class Ib expression by tumor cells allow them to escape the immune system is not fully understood. However, the described interactions between HLA class Ib molecules and various immune cells represents possible ways of tumor cells to evade the antitumor immune response. In the experimental part of the thesis, we will characterize HLA class Ib expression by malignant melanoma cell lines. Furthermore, we will perform a functional assay co-culturing PBMCs and malignant melanoma cell lines with different expression patterns of HLA class Ib molecules. We will observe if the co-culture affects the levels of Treg and DC subsets of PBMCs. In addition, we will analyze blood samples from skin cancer patients and healthy individuals, investigating if the levels of different T cell subsets, including Tregs, differ between the two.

40

Establishment of a cell bank In our study, we performed experiments with nine different cell lines. We use two human choriocarcinoma cell lines, JEG-3 (Sigma-Aldrich, St. Louis, Missouri) and BeWo (SigmaAldrich, St. Louis, Missouri), as HLA class Ib positive controls (see “Table 1” p. 43) and seven human malignant melanoma cell lines, FM-6, FM-45, FM-55M2, FM56, SK-MEL-3, SK-MEL-28 (kindly provided by Prof. Mads Hald Andersen (Herlev Hospital, DK)) and IGR-1 (DSMZ, Braunschweig, Germany). Photos of the cell lines are seen in “Figure 8” below. FM-45, FM-56 and SK-MEL-28 originate from primary malignant melanomas whereas IGR-1, FM-6, FM55M2 and SK-MEL-3 originate from malignant melanoma lymph node metastases.

Figure 8: Photos of our cell lines. 40x Magnification. The size bars represent 50 µm.

JEG-3, BeWo and IGR-1 were tested free for mycoplasma at the companies, where we bought them. We performed a mycoplasma test on all the cell lines from Herlev Hospital (for detailed protocol see “Appendix 1”) using “MycoSensor QPCR Assay Kit” (Agilent Technologies, Santa Clara, California). As illustrated in “Figure 9” below, all the cell lines were negative for mycoplasma.

41

Figure 9: Results from the Mycoplasma test. All tested cell lines were free of mycoplasma. The graph is a plot of PCR cycles on the x-axis versus fluorescence on the y-axis. The samples named the cell line with an A are samples that contain a positive control template. The samples named only the name of the cell lines contain purified DNA and if the cells were infected with mycoplasma the mycoplasma DNA would be amplified.

We started with one frozen vial of each cell line and established a cell bank in which we established frozen cell stocks at three different passages, the last being our working cell stock. All cell lines were maintained in suitable media (see “Table 1”) with 10% fetal bovine serum (FBS) and 5% penicillin/streptomycin solution. However, when cells were thawed for use in experiments, we did not use antibiotics. The cell lines have different passages, but for all our experiments the same cell line has the same passage number (see “Table 1”). The passage number of the cell lines from Herlev Hospital is not known so we provided all our cell lines with new passage number as we thawed them for the first time. Therefore, the number does not contain information about the complete times of passage, but only the times of passage in our lab. As mentioned JEG-3 and BeWo are known to express HLA-E, -F and -G (see “Table 1”). The expression of HLA class Ib molecules by the other cell lines included in our study is not well characterized. It is known that IGR-1 expresses HLA-G, that SK-MEL-28 expresses HLA-E and that FM-56 does not express HLA-G (see “Table 1” below).

42

Table 1 Cell line JEG-3

Tumor type Choriocarcinoma

Suitable media EMEM + L-glutamine

Passage no. at

Known HLA class 1b

experiments

protein expression

9

HLA-E, -F and –G [Roby et al., 1994; Pangault et al., 1999; Tripathi & Agrawal, 2006; Jabeen et al., 2013]

BeWo

Choriocarcinoma

F-12K + L-glutamine

7

HLA-E, -F and –G [Tripathi & Agrawal, 2006; Shaikly et al., 2010; Ishitani et al., 2003]

IGR-1

MM lymph node

DMEM + GlutaMAX

8

metastasis FM-6

MM lymph node

HLA-G [Tripathi & Agrawal, 2006]

RPMI 1640 + GlutaMAX

8

metastasis FM-45

Primary MM

RPMI 1640 + GlutaMAX

7

FM-55M2

MM lymph node

RPMI 1640 + GlutaMAX

7

metastasis FM-56

Primary MM

RPMI 1640 + GlutaMAX

7

SK-MEL-3

MM lymph node

RPMI 1640 + GlutaMAX

6

RPMI 1640 + GlutaMAX

9

HLA-G-negative [ESTDAB, 10.06.14]

metastasis SK-MEL-28

Primary MM

HLA-E [Fuertes et al., 2008]

43

Experiment 1 – Flowcytometric characterization of malignant melanoma cell lines Aim The aim was to characterize the protein expression of HLA class Ia, HLA-E, HLA-F and HLA-G in malignant melanoma cell lines. We investigated both intracellular and extracellular HLA-F and membrane-bound and soluble HLA-G.

Hypothesis

The choriocarcinoma cell lines JEG-3 and BeWo that have been shown to express HLA-G, -E and F (see “Table 1” p. 43) were our HLA class Ib positive controls. Therefore, we expected to detect expression of all the HLA class Ib molecules in these cell lines. Some characterization of HLA class Ib have been made in few of the malignant melanoma cell lines included in this study. HLA-G expression has been observed in IGR-1 and HLA-E expression has been observed in SK-MEL-28. Furthermore, according to The European searchable Tumour Line Database (ESTDAB) FM-56 does not express HLA-G [ESTDAB, 10.06.14]. Thus, we expected to observe this pattern. HLA-G, -E and -F protein expression are observed in several types of cancer. Based on this, we expected to find expression of HLA class Ib molecules in at least some of the malignant melanoma cell lines FM-6, FM-45, FM-55M2, FM56, SK-MEL-3, SK-MEL28 and IGR-1.

Preparation

Experimental design HLA-F is, as described in the chapter “Human Leukocyt Antigen F (HLA-F)”, expressed both extra- and intracellularly. Therefore, we decided to create both an extra- and intracellular experimental design to characterize the malignant melanoma cells. As described in the chapter “Human Leukocyt Antigen G (HLA-G)”, both membrane-bound and soluble isoforms of HLA-G exist. The experimental design for the intracellular detection enables us to characterize the expression of the soluble forms of HLA-G with mouse mAb 5A6G7 (Exbio, Vestec, Czech Republic). This mAb reacts with the sequence encoded by intron 4, thus only HLA-G5 and HLA-G6 are detected using HLA-G mAb 5A6G7 [Le Rond et al., 2004]. With the use of the HLAG mAb MEM-G/9 membrane-bound HLA-G was detected. This mAb reacts with HLA-G1, HLAG3 and with the β2-microglobulin associated form of the soluble isoform HLA-G5 [Exbio, 19.06.14]. However, we did not expect binding to HLA-G5, since the cells were not 44

permeabilized when staining with HLA-G mAb MEM-G/9. In the extracellular experimental design, we included two HLA-ABC mAbs. We included W6/32 because it is a widely used clone. However, HLA-ABC mAb W6/32 recognize all β2microglobulin-associated HLA class I antigens including HLA-E and HLA-G1 and HLA-G5 [Ibrahim et al., 2001]. Therefore, we included HLA-ABC mAb B9.12.1, which recognize an epitope different than recognized by W6/32 [Rebaï & Malissen, 1983]. HLA-ABC mAb B9.12.1 is not reported to detect HLA class Ib molecules. The antibody against HLA-F is polyclonal, since no monoclonal antibody against native HLA-F was available. In addition, we conjugated this pAb with PerCP-CY5.5 ourselves. We used isotype controls for all the markers in both flow cytometry panels. The mix of antibodies are shown in “Table 2” and “Table 3” below. For the proteins that we expected to be the least expressed the antibodies were conjugated with the brightest (most intense) fluorochromes. Table 2: Antibody mixes for intracellular panel Laser Filter

Marker

Isotype control

Blue

Red

Violet

Alexa Fluor® 488

PE

PerCP-Cy5.5

APC

Pacific Blue

(530/30)

(585/40)

(695/40)

(660/20)

(440/40)

HLA-G

HLA-E

HLA-G

HLA-ABC

(5A6G7)

(3D12)

(PA5-252773, polyclonal)

(MEM-G/9)

(W6/32)

IgG1

IgG1

IgG

IgG1

IgG2a

(MOPC-21)

(MOPC-21)

(polyclonal)

(P3.6.2.8.1)

(MOPC-173)

Red

Violet

HLA-F

Table 3: Antibody mixes for extracellular panel Laser Filter

Marker

Isotype control

Blue FITC

PE

PerCP-Cy5.5

APC

Pacific Blue

(530/30)

(585/40)

(695/40)

(660/20)

(440/40)

HLA-ABC

HLA-E

HLA-F

HLA-G

HLA-ABC

(MEM-G/9)

(W6/32)

(B9.12.1)

(3D12)

(PA5-252773, polyclonal)

IgG2a

IgG1

IgG

IgG1

IgG2a

(MOPC-173)

(MOPC-21)

(polyclonal)

(P3.6.2.8.1)

(MOPC-173)

45

Setting photo-multiplier tube (PMT) voltage and color compensation Optimal voltage settings are important to resolve dim signals from background noise. We used isotype control and marker stained cells to set the PMT voltage. For the more bright fluorochromes, the optimal procedure is to adjust the cells stained with isotype control to appear in the first decade, while still being able to detect a positive signal in the marker stained tube. However, this is not the optimal for fluorochromes with longer emission, because the PMT detectors are less sensitive to these wavelengths. Furthermore, the unstained cells in this part of the spectrum emit some autofluorescence. The color compensation was made using BDTM CompBeads set Anti-Mouse Ig, κ (BD Bioscience, San Jose, California) according to the manufactures instructions. Since our HLA-F antibody is anti-rabbit, we could not use this for the color compensation. Although not optimal, we used another PerCP-Cy5.5 conjugated antibody.

Materials and methods Cell lines The study included the human malignant melanoma cell lines FM-6, FM-45, FM-55M2, FM56, SK-MEL-3, SK-MEL-28 kindly provided by Prof. Mads Hald Andersen (Herlev Hospital, DK) and IGR-1 (DSMZ, Braunschweig, Germany). Furthermore, two HLA class Ib positive controls were included in the study. JEG-3 (Sigma-Aldrich, St. Louis, Missouri) and BeWo (SigmaAldrich, St. Louis, Missouri), which are choriocarcinoma cell lines. Cell lines provided by Prof. Mads Hald Andersen were maintained in Roswell Park Memorial Institute (RPMI) 1640 + Glutamax with 10% FBS; IGR-1 were maintained in Dulbecco’s Modified Eagle Medium (DMEM) with 10% FBS; BeWo were maintained in Kaighn’s Modification of Ham’s F-12 (F12K) Medium with 10% FBS; and JEG-3 were maintained in Eagle’s Minimum Essential Medium (EMEM) with 10% FBS. Monoclonal and polyclonal antibodies for the extracellular staining FITC conjugated: HLA-ABC mouse mAb B9.12.1 (Beckman Coulter, Brea, California) and IgG2a mouse mAb MOPC-173 (Biolegend, San Diego, California). PE conjugated: HLA-E mouse mAb 3D12 (Biolegend, San Diego, California) and IgG1 mouse mAb MOPC-21 (BD pharmingen, San Jose, California). A LYNX Rapid PerCP-Cy5.5 antibody conjugation kit was purchased from

46

AbD Serotec (Kidlington, United Kingdom) and used to conjugate: HLA-F rabbit pAb PA5252773 (Thermofisher Scientific, Waltham, Massachusetts) and IgG rabbit F(ab’)2 pAb (Abcam, Cambridge, United Kingdom). APC conjugated: HLA-G mouse mAb MEM-G/9 (Exbio, Vestec, Czech Republic) and IgG1 mouse mAb P3.6.2.8.1 (eBioscience, San Diego, California). Pacific Blue conjugated: HLA-ABC mouse mAb W6/32 (Biolegend, San Diego, California) and IgG2a mouse mAb MOPC-173 (Biolegend, San Diego, California). Monoclonal and polyclonal antibodies for the intracellular staining Alexa fluor 488 conjugated: HLA-G mouse mAb 5A6G7 (Exbio, Vestec, Czech Republic) and IgG1 mouse mAb MOPC-21 (Biolegend, San Diego, California). A LYNX Rapid PerCP-Cy5.5 antibody conjugation kit was purchased from AbD Serotec (Kidlington, United Kingdom) and used to conjugate: HLA-F rabbit pAb PA5-252773 (Thermofisher Scientific, Waltham, Massachusetts) and IgG rabbit F(ab’)2 pAb (Abcam, Cambridge, United Kingdom). Flow Cytometry All cells were grown to exponential phase (80-90% confluency) and detached with 0.25% trypsin/EDTA. Cells were incubated with 10% mouse serum to inhibit non-specific binding of mouse antibodies. Cells from each cell line were stained with marker antibodies. Isotype matched mouse or rabbit IgGs were used as negative controls. When investigating intracellular expression, the cells were fixed and permeabilized using Intrastain (Dako, Glostrup, Denmark) prior to incubation with antibodies. Cells were resuspended in FACSFlow (BD Bioscience, San Jose, California) and immediately analyzed on BD FACSCantoTM II (BD Bioscience, San Jose, California). For detailed protocols of our extracellular and intracellular staining procedure, lot, stock and working concentrations of antibodies see “Appendix 2”. Analysis of flow cytometry data Flow cytometry analysis was carried out using BD FACSDiva Software v6.1.3 (BD Bioscience, San Jose, California). We used the isotype control tube to set a gate on a histogram plot of each fluorochrome-labeled isotype control. Setting this gate, we allowed maximum 1% of the events to be positive. The gates were then copied to the same plot in the corresponding marker tubes. From here, we noted the percentage of positive event minus the percentage of positive events from the isotype control tube. Cell lines in which ≤ 0.5% exhibited protein 47

expression were not considered positive.

Results

Extracellular characterization of cell lines For most of the cell lines tested a high percentage exhibited protein expression of the classical class Ia molecules (see “Figure 10” below). BeWo stands out, with only about 20% positive events detected with the HLA-ABC mAb W6/32 (Biolegend, San Diego, California) and no positive events with the HLA-ABC mouse mAb B9.12.1 (Beckman Coulter, Brea, California). The lowest percentage of HLA-class Ia expressing malignant melanoma cells (38%), detected with HLA-ABC mAb B9.12.1. (Beckman Coulter, Brea, California), is observed for SK-MEL-3.

Figure 10: Surface expression of HLA class Ia by malignant cell lines. A) The percentage of cells from each cell line stained with HLA-ABC mAb B9.12.1. B) The percentage of cell from each cell line stained with HLA-ABC mAb W6/32. For the cell lines FM-6, FM-45, FM-56 and SK-MEL-28 a high percentage of positive cells were observed using either of the mAbs. For SK-MEL-3 and IGR-1 the percentage of positive cells is remarkably lower for HLA-ABC mAb B9.12.1 compared to HLA-ABC mAb W6/32. No BeWo cells were positive for HLA-ABC mAb B9.12.1, but some were positive for HLA-ABC mAb W6/32. More than half of JEG-3 cells were positive using either of the mAbs.

The cell lines in which HLA-E protein expression was observed is the malignant melanoma cell lines; FM-55M2, FM-6, FM-45 and the choriocarcinoma cell line JEG-3 (see “Figure 11” below). The percentages of positive cells were 10.3%, 9.2%, 3.0% and 2.8%, respectively. Extracellular expression of HLA-F was found in four of the malignant melanoma cell lines; FM55M2, FM-6, FM-56, SK-MEL-28 (6.6%, 1.1%, 3.4% and 9.5%, respectively) and in the choriocarcinoma cell line JEG-3 (0.9%) (see “Figure 11” below).

48

Figure 11: Surface expression of HLA class Ib by malignant cell lines. A) The percentage of cells from each cell line stained with HLA-E mAb B) The percentage of cells from each cell line stained with HLA-F pAb. C) The percentage of cells from each cell line stained with HLA-G mAb. For JEG-3, FM-55M2, FM-6 and FM-45 some positive cells were observed, while for the other cell lines no or very few cells were positive for HLA-E. For JEG-3, FM-55M2, FM-6, FM-56 and SK-MEL-28 some cells were positive for HLA-F, while the other cell lines were negative for HLA-F. Most of JEG-3 cells were positive for HLA-G while only some of the cells of the malignant melanoma cell lines were positive. BeWo did not express HLA-G.

Consistent with earlier findings [Roby et al., 1994; Pangault et al., 1999], JEG-3 exhibited surface expression of HLA-G (93.4%). In contrast, BeWo did not exhibit surface expression of HLA-G. For all of the seven malignant melanoma cell lines some percentage of the cells exhibit HLA-G surface expression (see “Figure 11” above). The cell line where the highest percentage of positive cells was observed is FM-55M2, FM-56 and SK-MEL-28 (9.6%, 7.2%, and 5.4% respectively). For the other malignant melanoma cell lines, the percentage of positive cells was between 1.1% and 3.2%. To give an overview of the HLA class Ia and Ib expression pattern of each cell line, the results are combined in “Figure 12” below.

49

J E G -3

40

20

20

20

60

40

20

)

)

6 (W

9

M

.1

-G

2

/3

2

.1

/9

a n

H

H

L

L

A

A

-A

-A

B

B

C

C

(B

E (M

6 (W

H

H

L

L

A

A

-A

-A

B

B

C

C

(B

E

9

M

.1

-G

2

/3

2

.1

)

)

) /9

a n lo (M -G A

20

) /3

2

.1 2

(W

6

.1 9

C

(B

B

C

-A

B

A L H

L H

H

L

A

A

-A

-G

(p -F

)

) -G M E (M

ly o

-E A L H

/9

l) c

(3

lo

D

n

1

a

2

2 /3 6 (W C

A L H

A L H

)

)

) .1 2 -A

B -A

-G A L H

B

C

(M

(B

E

9

M

.1

-G

n lo c ly o (p

-F A L

)

a

2 1 D (3 -E A L H H

/9

l)

)

) 2 /3

A L H

A L H

L

40

0

6 (W C B -A

B -A

-G A L H

) .1 .1 9 (B

C

(M

ly

E

c

M

lo

-G

n

2

a

/9

l)

) 2 1 D o (p -F

H

60

0

)

0

80

A

20

80

L

40

% p o s it iv e e v e n t s

60

A

c -F

A L

IG R -1 100

H

% p o s it iv e e v e n t s

80

L

l)

) 2 A L H

H

H

H

L

L

A

A

-A

-A

B

B

C

C

(B

-E

9

(W

(3

6

.1

D

2

/3

1

2

.1

)

)

) -G M E (M

-G

H

H

L

L

A

A

-F

H

/9

l) a lo c ly

L

(p

A

o

-E

(W

(3

6

D

n

1

/3

2

2

)

)

) .1 C -A A L H

H

L

L

A

A

-A

-G

B

B

C

(M

(B

E

9

M

.1

-G

n lo c ly o (p

-F A L

2

a

/9

l)

) 2 1 D (3 -E A L H H

H

S K -M E L -3 100

(3

-G

20

0

S K -M E L -2 8 100

-E

H

40

0

)

0

A

L

A L H

40

60

ly

20

60

80

o

40

80

(p

60

L

A

-F

H

A L H

100

% p o s it iv e e v e n t s

% p o s it iv e e v e n t s

80

H

lo

D (3 -E A L

B -A

B -A A L H

F M -5 6

F M -4 5 100

H

l)

) 2 1

2 /3 6 (W C

(B C

(M -G A L H

)

) .1 2 9

M E

c ly o (p -F

A L H

.1

-G

n lo

D (3 -E A L H

)

a

2 1

/3 6 (W

H

/9

l)

)

) 2

.1 2 C -A A L H

H

L

L

A

A

-A

-G

B

B

C

(M

(B

E

9

M

.1

-G

n lo c ly o (p

-F A L

)

)

a

/9

l)

) 2 1 D (3 -E A L H H

F M -6

)

0

0

100

% p o s it iv e e v e n t s

40

c

0

60

ly

20

60

80

o

40

80

(p

60

% p o s it iv e e v e n t s

% p o s it iv e e v e n t s

% p o s it iv e e v e n t s

100

100

80

% p o s it iv e e v e n t s

F M -5 5 M 2

B eW o

100

Figure 12: HLA clas Ia and Ib surface expression pattern. Cells from each of the cell lines studied were stained with antibodies against HLA-E, HLA-F, HLA-G and two against HLA-ABC. HLA-ABC (B9.12.1) recognize β2-m-associated HLA class Ia molecules and HLA-ABC (W6/32) recognize all β2-m-associated HLA class I antigens including HLA-E and HLA-G1 and HLA-G5 [Rebaï & Malissen, 1983; Ibrahim et al., 2001]. The y-axis shows the percentage of positive events. Most JEG-3 cells was found to express HLA class Ia proteins and HLA-G while few JEG-3 cells expressed HLA-E or HLA-F. Relatively few BeWo cells were positive for HLA class Ia expression using HLA-ABC mAb W6/32 but when using the HLA-ABC mAb B9.12.1 positivity was absent. Furthermore, no HLA class Ib expression was detected. For FM-55M2 a high percentage of cells was found to express HLA class Ia proteins and a relatively high percentage expressed HLA-E, -F or -G compared with the other cell lines. In contrast the percentage of HLA-E, -F or -G positive FM-6 cells was low. A high percentage of the FM-6 cells expressed HLA-ABC. The same was observed FM-45, although no surface expression of HLA-F was detected. A high percentage of both FM-56 and SK-MEL-28 express HLA class Ia molecules. For both of these cell lines, some of the cells were positive for HLA-F and HLA-G, but not for HLA-E. A high percentage of SK-MEL-3 were positive for HLA class Ia proteins detected by HLA-ABC mAb W6/32. However, detected by HLA-ABC mAb B9.12.1 a relatively low percentage of the cells were positive for HLA class Ia proteins. Compared to some of the other cell lines, a low percentage of SK-MEL-3 expressed HLA-G. The same pattern is observed for IGR-1, except that the percentage of cells positive for HLA class Ia detected by HLA-ABC mAb B9.12.1 is about the same as the percentage detected by HLA-ABC mAb W6/32.

50

Intracellular characterization of cell lines In all of the cell lines protein expression of the HLA class Ia molecules was still detected but for most of the cell lines the percentages of positive cells were considerable lower compared to the percentages found in the extracellular setup (see “Figure 10” p. 48 and “Figure 13” below). A percentage of the cells showed HLA-E surface expression for three of the malignant melanoma cell lines: FM-6 (1.8%), FM-55M2 (1.3%) and SK-MEL-3 (0.6%), as well as for the choriocarcinoma cell line JEG-3 (1.6%) (see “Figure 13” below).

Figure 13: Surface or intracellular expression of HLA class I proteins by malignant cell lines. A) The percentage of cells from each cell line stained with HLA-ABC mAb W6/32 B) The percentage of cells from each cell line stained with HLA-E mAb. C) The percentage of cells from each cell line stained with HLA-F pAb. For all cell lines, a percentage of cells expressed HLA-ABC, although the percentage of HLA-ABC-positive BeWo cells was much lower compared to the other cell lines. For all cell lines, expression of HLA-E was observed on few cells or was absent. For all cell lines, except IGR-1, intracellular HLA-F expression was observed, but the percentage of positive cells is several times higher for FM-6, SK-MEL-28 and SK-MEL-3 compared to the other cell lines.

Intracellular expression of HLA-F was observed in all cell lines except for IGR-1 (see “Figure 13” above). In three of the cell lines the percentage of positive cells was about 50% or higher: SK-MEL-28 (93.1%), FM-6 (60.2%) and SK-MEL-3 (58.6%). The percentage of positive cells in the other malignant melanoma cell lines: FM-45, FM-55M2 and FM-56 were 9.1%, 7.1% and 13.4% respectively. For both of the choriocarcinoma cell lines JEG-3 and BeWo, the percentage of positive cells was 9.8%. Surface expression of HLA-G, detected with the use of mAb MEM-G/9 (Exbio, Vestec, Czech Republic), was found on a considerably higher percentage of three of the cell lines FM-6, SKMEL-28 and SK-MEL-3 compared to that found in the extracellular setup (see “Figure 11” p. 49 and “Figure 14” below). Notably, this was coincident with the cell lines having a high percentage of cells positive for intracellular HLA-F. For the other cell lines, the percentage of cells positive for surface HLA-G, was near the same, or a little lower than in the extracellular setup. 51

Figure 14: Expression of membrane-bound HLA-G or soluble HLA-G by malignant cell lines. A) The percentage of cells from each cell line stained with HLA-G mAb MEM-G/9. B) The percentage of cells from each cell line stained with HLA-G mAb 5A6G7. Expression of membrane-bound HLA-G was observed in all the cell lines. JEG-3 had the highest percentage of cells expressing membrane-bound HLA-G. Of the malignant melanoma cell lines FM-5, SK-MEL-28 and SK-MEL-3 had the highest percentage of cells expressing membrane-bound HLA-G. FM-56 was the cell line with the highest percentage of cells positive for soluble HLA-G isoforms. No expression of soluble HLA-G was observed in FM6, SK-MEL-3 or IGR-1.

Soluble HLA-G, detected with mAb 5A6G7 (Exbio, Vestec, Czech Republic), was expressed by some percentages of four of the malignant melanoma cell lines: FM-56 (5.7%), FM-45 (2.5%), FM-55M2 (1.4%) and SK-MEL-28 (1.3%) as well the two choriocarcinoma cell lines JEG-3 (1.8%) and BeWo (0.6%) (see “Figure 14” above). To give an overview of the HLA class Ia and Ib expression pattern of each cell line, the results are combined in “Figure 15” below.

52

J E G -3

B eW o

40

20

20

40

20

60

40

20

80

60

40

20

0

G

/3

2

7

)

)

) /9

(W

L

H

A

L

-A

A

B

-G

C

E

(5

M

A

6

6

-G

n (M -G A L

H G

/3

2

7

)

)

) /9

(W

L

H

A

L

-A

A

B

-G

C

E

(5

M

A

6

6

-G

n (M -G A

H

20

)

)

/3

2

7 G

6

6

(W

A

C

(5

B -A

A H

L

A

L H

A L H

-G

(M -G

(p -F A L

) -G M E

c ly o

-E A L H H

/9

a n lo

D (3

6 (W C B -A

A L H

l)

) 1

/3

G 6 A (5 -G

A L H

2

2

7

)

)

) -G M E (M

-G H

L

A

-F A L H

/9

a n lo c ly o

(p

A L H

-A A L H

l)

) 2 1 D (3 -E

C B

-G A L H

L

40

) /3 6 (W

A (5

M E (M -G

A L

lo

60

0

2

7 G 6

-G

n lo c ly o H

)

)

a

/9

l)

) 2 1 D (3 (p -F

H

80

0

A

c -F

A L H

IG R -1 100

% p o s it iv e e v e n t s

% p o s it iv e e v e n t s

80

L

a

2 1 D (3 -E A L H

-A A L H

H

S K -M E L -3 100

H

l)

)

) /3 6 (W C

B

-G A L H

L

2

7 G 6 A (5

M E (M -G

A

-F A L H

)

) -G

n lo c ly o (p

L H

-A A L H

/9

a

2 1 D (3 -E A

C B

-G A L H

l)

)

) 2 /3 6 (W

A (5

M E (M -G

A L H

H

) 7 6

-G

n lo c ly o (p

-F A L

G

a

/9

l)

) 2 1 D (3 -E A L H

lo

20

0

S K -M E L -2 8 100

-E

H

40

0 )

0

60

ly

20

60

80

o

40

80

(p

60

A

c -F

A L H

F M -5 6 100

% p o s it iv e e v e n t s

% p o s it iv e e v e n t s

80

L

a

2 1 D (3 -E A L H

-A A L H

F M -4 5 100

H

l)

)

) /3 6 (W C

B

-G A L H

L H

H

2

7 G 6 A (5

E (M -G A

-F A L

)

) -G M

lo c ly o (p

L H

-A A L H

/9

a n

1 D (3 -E A

C B

-G A L H

l)

) 2

2 /3 6 (W

A (5

M E (M -G

A L H

H

)

) 7 6

-G

n lo c ly o (p

-F A L

G

a

/9

l)

) 2 1 D (3 -E A L H

0

F M -6 100

% p o s it iv e e v e n t s

40

0 )

0

60

ly

20

60

80

o

40

80

(p

60

100

% p o s it iv e e v e n t s

80

% p o s it iv e e v e n t s

F M -5 5 M 2

100

% p o s it iv e e v e n t s

% p o s it iv e e v e n t s

100

Figure 15: HLA class Ia and Ib surface and cytoplasmic expression pattern. Cells were stained extracellularly with antibodies against HLA-E, HLA-ABC and HLA-G (MEM-G/9) and intracellularly with antibodies against HLA-F and HLA-G (5A6G7). HLA-G (MEM-G/9) recognize membrane-bound isoforms of HLA-G [Exbio, 19.06.14] and HLA-G (5A6G7) recognize soluble isoforms of HLA-G (HLA-G5 and -G6) [Le Rond et al., 2004]. The y-axis shows the percentage of positive events. Most JEG-3 cells were found to express membrane-bound HLA-G, while very few expressed soluble isoforms of HLA-G. Some of the JEG-3 cells expressed HLA class Ia and intracellular HLA-F. Very few cells expressed HLA-E. Surface HLA class Ia expression and expression of intracellular HLA-F was detected in a low percentage of BeWo cells. Very few BeWo cells were positive for membrane-bound HLA-G and for soluble isoforms of HLA-G, while no cells were positive for HLA-E. For FM-55M2 a high percentage of cells expressed HLA class Ib, some of the cells expressed HLA-F and some expressed membrane-bound HLA-G. Very few of the cells expressed HLA-E or soluble isoforms of HLA-G. A high percentage of FM-6 was positive for HLA class Ia. More than half of the FM-6 cells were positive for HLA-F, some were positive for membrane-bound HLA-G and very few were positive for HLA-E. More than half of the FM-45 cells expressed HLA class Ia, some expressed HLA-F and few expressed membrane-bound or soluble isoforms of HLA-G. No HLA-E expression was observed in this cell line. For

53

FM-56, the result is similar, though the percentage of cells expressing HLA-F is higher. For SK-MEL-28, almost all cells express HLA-F, more than half of the cells express HLA class Ib and some of the cells expressed membrane- bound HLA-G. Very few of the cells expressed soluble forms of HLA-G. Expression of HLA class Ia, HLA-F and membranebound HLA-G was observed in about half of the SK-MEL-3 cells, while very few cells expressed HLA-E and no cells expressed soluble isoforms of HLA-G. For IGR-1 about half of the cells expressed HLA class Ia proteins and very few expressed membrane-bound HLA-G. No expression of HLA-E, HLA-F or soluble isoforms of HLA-G was detected.

Discussion

One of the control cell lines JEG-3 expressed all of the class Ib proteins but very low percentages were detected of HLA-E and HLA-F. Furthermore, JEG-3 was positive for HLAABC when staining with both HLA-ABC mAbs. Of the HLA class Ia molecules, HLA-C is the only one expressed by JEG-3 [Burt et al., 1991; reviewed in Jabeen et al., 2013]. Therefore, the positivity observed with HLA-ABC mAb B9.12.1 must be due to HLA-C while the positivity observed with HLA-ABC mAb W6/32 is due to both HLA-C and HLA-G. We detected HLA-G expression by IGR-1 and HLA-E expression by SK-MEL-28 as previously reported by others [Tripathi & Agrawal, 2006; Fuertes et al., 2008]. The percentage of positive events regarding HLA-E, HLA-F and HLA-G in the cell line BeWo is surprisingly low or not present in the extracellular and intracellular tubes. It is unexpected since this cell line was chosen as a positive control for the HLA class Ib molecules. Since we detect positive events in some of the other tested cell lines we exclude the possibility that missing events is due to low effectiveness of the mAbs. A possible explanation is the loss of expression due to long-term culturing of the cell line. It is known as mentioned in the chapter “Studies investigating HLA class Ib expression in malignant melanoma” that expression of HLA-G can wane and be lost after 70 passages [Rouas-Freiss et al., 2005]. However, BeWo was in passage 13 (passage number from the company + our passage number), making this explanation unlikely. Another unexpected observation was that FM-56 did express HLA-G. According to The European Searchable Tumour Line Database (ESTDAB) [ESTDAB, 10.06.14], FM-56 does not express HLA-G. However, they do not refer to a reference and we have not been able to find a description of the study. Therefore, we could not explain the contradictory result. Three of the malignant melanoma cell lines tested are established from primary tumors (FM45, FM-56, SK-MEL-28), while the four others are established from lymph node metastasis (IGR-1, FM-6, FM-55M2, SK-MEL-3). When we compared the HLA class I expression between these two groups, no obvious pattern was observed. However, it seems that a higher percentage of the cells established from primary tumors expressed soluble HLA-G. The

54

percentage of cells that were positive for HLA-G5/6 was generally low. However, we did only perform these experiments once. To our knowledge, we are the first to report HLA-F expression by malignant melanoma cell lines. HLA-F was expressed intracellularly by both choriocarcinoma cell lines JEG-3 and BeWo and by all the malignant melanoma cell lines except from IGR-1. Furthermore, we detected HLA-F surface expression by some of these cell lines as well. For three of the malignant melanoma cell lines (FM-6, SK-MEL-28 and SK-MEL-3) the percentages of cells expressing intracellular HLA-F were remarkably high. The same three cell lines showed increased expression of membrane-bound HLA-G in the intracellular experimental design compared to the extracellular experimental design. This cannot be explained by the experimental design. In the intracellular experimental design, fixation and permeabilization of the cells are performed after staining of extracellular markers. This treatment would rather decrease than increase the staining efficiency. These three cell lines (FM-6, SK-MEL-28 and SK-MEL-3) are the cell lines where a remarkably high percentage of cells expressed intracellular HLA-F. Therefore, we suspected that spectral overlap between PerCP-Cy5.5 (HLA-F pAb fluorochrome) and APC (HLA-G mAb MEM/G-9 fluorochrome) might be an explanation, so this could be a color compensation problem. When adjusting the compensation between these two fluorochromes we observed that the compensation settings were not optimal supporting that spectral overlap was the explanation for the increased percentage of HLA-G-positive cells.

Optimization of the experiment In our experiment, we incubated our cells with mouse serum to inhibit non-specific binding. Since the pAb detecting HLA-F was rabbit antibodies, the optimal would be to block with rabbit serum as well. This was not available in our lab, and partly due to economic reasons we decided not to implement this. Another improvement would be to do the experiment in one or more replicates, validating our results. This was not done due to the expense of the antibodies. For this experiment, we did not titrate the antibodies, because we lacked some clear positive controls. Therefore, we used the recommended amount, which are not always necessary or economically favorable. For each tested cell lines, we stained 500,000 cells. When analyzing these on the flow cytometer we were often only able to count 50,000 cells. Somehow, we lost many cells in the process of staining, but we have not been able to identify the reason for this yet.

55

Experiment 2 – Droplet Digital PCR (ddPCR) characterization of malignant melanoma cell lines Introduction The principles of Droplet Digital PCR Droplet Digital PCR (ddPCR) is a method for performing digital PCR that is based on a wateroil emulsion droplet system. A sample containing the cDNA, primers, a probe conjugated with either FAM or HEX and Bio-Rad ddPCR supermix, is mixed with oil in the QX100 droplet generator, which generates droplets and partitions target and non-target DNA randomly into these. Thousands of independent PCRs are run at the same time since amplifications of the cDNA occurs in each individual droplet. After PCR amplification to end point (40 or 50 cycles), the samples are run through the QX100 droplet reader, which works very similar to a flow cytometer. The droplet reader sips up the droplets, and streams them single file past a twocolor optical detection system that detects if the droplets provide a positive signal indicating the presence of the target cDNA. The positive and negative droplets are counted and each droplet in a sample is plotted on a graph of event fluorescence intensity vs event number and on a graph of frequency vs fluorescence intensity. On both graphs, the positive droplets are those above the threshold indicated by the pink line (see Figure 16).

Figure 16: Example of ddPCR result from QuantaSoft software v.1.3.2.0 (Bio-Rad, Hercules, California). The result is from a ddPCR on 2 µl of a 1:5 dilution BeWo cDNA with GUS primer/probe set (PCR program is in “Appendix 3”). The right plot shows amplitude vs event number and the left plot shows Frequency vs amplitude. Each dot on the right plot represent a droplet. Events above or on the right side of the pink line are positive.

The QuantaSoft software (Bio-Rad, Hercules, California) calculates the concentration of target cDNA as copies/µl using Poisson distribution analysis. When the droplets are generated, 56

template molecules are distributed randomly into droplets. Some droplets may contain more than one template, while others contain none. Because the partitioning is random, the data are well fit by a Poisson distribution that can predict the likelihood of 0, 1, 2, 3 or more template molecules present in each droplet. The number of target copies per droplet and thus the number of copies of target molecules in a 20 µl sample can be described using the equation ln(1-p), in which p is the fraction of positive droplets.

Aim

The aim was to characterize the mRNA expression of HLA-E, HLA-F and HLA-G in malignant melanoma cell lines. We investigated both the amount of total HLA-G mRNA and the combined amount of HLA-G5 and HLA-G6, two of the soluble isoforms of HLA-G.

Hypothesis We expected to observe mRNA expression of HLA-E, -F and -G in our positive control choriocarcinoma cell lines JEG-3 and BeWo. We expected that the malignant melanoma cell lines expressed HLA-E mRNA, since HLA-E is transcribed by most cells, as mentioned in the chapter “Human Leukocyte Antigen E”. Furthermore, mRNA expression of HLA-F and HLA-G are observed in various cancer cell lines, and in line with this, we expected that at least some of them expressed HLA-F and HLA-G mRNA. In “Experiment 1”, we detected protein expression of soluble isoforms of HLA-G by JEG-3, BeWo, FM-55M2, FM-45, FM-56 and SKMEL-28. Thus, we expected to detect mRNA of soluble isoforms of HLA-G in these cell lines.

Preparation Designing primers and probes For this experiment we designed the primers and probes for amplification of HLA-E, -F and –G (produced by TAG Copenhagen A/S, Frederiksberg, Denmark). The β-glucuronidase (GUS) primers and probe were designed by Europe Against Cancer (EAC) laboratories [Beillard et al., 2003] and produced by TAG Copenhagen A/S, Frederiksberg, Denmark. The sequences and locations of the primers and probes are showed in “Table 4” below.

57

Table 4: Primers and probes used in the ddPCR experiments Amplicon

Forward primer

Revers primer

Probe

HLA-E

5’TACTCCTCTCGGAGGCCCTG

5’TGTGTCTTTGGGGGCT-CCAGG 3’

5’-Fam-

G 3’ (exon 1)

(exon3/4)

AGTGGATGCATGGCTGCGAGCTGGGG C-BHQ-1-3’ (exon 3)

HLA-F

5’AGTATTGGGAGTGGACCACA

5’ CGCTGTAGCGTCTCCTTCCCATT

5’-Fam-

GGGT 3’ (exon 2)

3’ (exon 3)

CCGAGTGGCCCTGAGGAACCTGCTCCG C-BHQ-1-3’ (exon 2)

HLA-G all

HLA-G5/6

5’AGCAGTCTTCCCTGCCCACC

5’ CAAT-

5’-Fam-

3’ (exon 5)

CTGAGCTCTTCTTTCTCCACAGC 3’

CCTTGCAGCTGTAGTCACTGGAGCTGC

(exon 5/6)

GGTCG-BHQ-1-3’ (exon 5)

5’TTCACCTCCTTTCCCAG-

5’CGACCGCAGCTCCAGTGACT 3’

5’-Fam-

AGCAGTC 3’ (intron 4/exon 5)

(exon 5)

GCCCACCATCCCCATCATGGGTATCGT TGCTGG-BHQ-1-3’ (exon 5)

GUS

5’

5’ CCGAGTGAAGATCCCCTTTTTA ‘3

5’-Hex-

GAAAATATGTGGTTGGAGAGC

(exon 12)

CCAGCACTCTCGTCGGTGACTGTTCA-

TCATT ’3 (exon11)

BHQ-1-3’ (exon 11/12)

When designing the primers and probes, we made sure that the primers had approximatley the same Tm and the probe had a Tm of about 10°C higher. The HLA-G all primers are located so that all isoforms of HLA-G are detected. The HLA-G5/6 primer set is located so that the soluble isoforms HLA-G5 and HLA-G6 are detected (it was not possible to include detection of the last soluble isoform HLA-G7) and the HLA-F primers are located so that all three isoforms of HLA-F are detected. Furthermore, the primers and probes are located where there is no known sequence variation in the genes according to the Ensembl Genome Browser (10.11.14). To be sure that the primers do not amplify genomic DNA, one of the primers in each set spans an intron. This was unfortunaltely not possible for HLA-F. A Blastn was performed on all the primer and probe sequences to make sure that they were specific. The expected length of PCR amplified sequences by the primers (for HLA-E, -G all and -G5/6 only) using GoTag® qPCR Master Mix (Promega, Fitchburg, Wisconsin), were confirmed by electroforesis on a 2% agarose gel. Sequencing of the PCR products with BigDye® Terminator v1.1,v3.1 (Applied Biosystems, Foster City, California) on an ABI 3500 Genetic Analyzer (Applied Biosystems, Foster City, California) also confirmed that the primers were amplifying the correct sequences (data not shown).

58

Materials and methods Cell lines The study included the human malignant melanoma cell lines FM-6, FM-45, FM-55M2, FM-56, SK-MEL-3, SK-MEL-28 kindly provided by Prof. Mads Hald Andersen (Herlev Hospital, DK) and IGR-1 (DSMZ, Braunschweig, Germany). Furthermore, two HLA class Ib positive controls were included in the study. JEG-3 (Sigma-Aldrich, St. Louis, Missouri) and BeWo (SigmaAldrich, St. Louis, Missouri), which are choriocarcinoma cell lines. Cell lines provided by Prof. Mads Hald Andersen were maintained in Roswell Park Memorial Institute (RPMI) 1640 + Glutamax with 10% FBS; IGR-1 were maintained in Dulbecco’s Modified Eagle Medium (DMEM) with 10% FBS; BeWo were maintained in Kaighn’s Modification of Ham’s F-12 (F12K) Medium with 10% FBS; and JEG-3 were maintained in Eagle’s Minimum Essential Medium (EMEM) with 10% FBS. RNA purification and cDNA synthesis From each cell line, total RNA was purified from one million cells using the RNeasy® Mini Kit (Qiagen). For detailed protocol see “Appendix 4”. Three µl (1209-2593.5 ng) of RNA (260/280 beween 1.97-2.08), measured by Thermoscientific µDrop™plate, were used for first strand cDNA synthesis with SuperScript® VILO™ cDNA synthesis kit (Invitrogen). For detailed protocol see “Appendix 5”. Droplet Digital PCR Each reaction contained 10 µl 2x ddPCRTM Supermix for Probes (Bio-Rad, Hercules, California), 20 pmol (0.5 µM) primers, 4 pmol (0.2 µM) probe and template cDNA in a total volume of 20 µl. For amplification with GUS primers 0.4 µl cDNA was used. For amplification with HLA-G all primers 0.1 µl (JEG-3), 0.2 µl (BeWo and IGR-1) or 5 µl (FM-6, FM-45, FM55M2, FM-56, SK-MEL-3 and SK-MEL-28) cDNA was used. For amplification with HLA-G5/6 primers 0.4 µl (JEG-3), 2 µl (BeWo) or 5 µl (IGR-1, FM-6, FM-45, FM-55M2, FM-56, SK-MEL-3 and SK-MEL-28) cDNA was used. For amplification with HLA-E primers 0.4 µl cDNA was used and for amplification with HLA-F primers 2 µl (IGR-1, FM-6, FM-45, FM-55M2, FM-56, SKMEL-3 and SK-MEL-28) or 5 µl (JEG-3 and BeWo) was used. The reaction mixtures were loaded on to a QX-100 Droplet Generator (Bio-Rad, Hercules, California) and afterwards the generated droplets were transferred to a 96 well plate for PCR in a thermal cycler. The ddPCR 59

for amplification with GUS, HLA-G all and HLA-G5/6 primers were run with following cycling conditions: 10 minutes at 95 °C, 40 cycles of: 15 second denaturing at 95 °C, 1 minute annealing at 62 °C and 1 minute extension at 72 °C, and finally a 10 minute step at 98 °C. The ddPCR for amplification with GUS, HLA-E and HLA-F primers were run with following cycling conditions: 10 minutes at 95 °C, 50 cycles of: 30 second denaturing at 95 °C, 1 minute annealing at 62 °C (GUS), 64 °C (HLA-F) or 66 °C (HLA-E) and 2 minute extension at 72 °C, and finally a 10 minute step at 98 °C. After cycling the plate was loaded on a QX100 Droplet reader (Bio-Rad, Hercules, California). Calculating copies of mRNA per 1x106 cells The procedure is illustrated in “Figure 17” below. The amount of cells, volumes of total RNA and cDNA are indicated in the figure. Considering the dilutions factors during the steps it is possible to calculate the total number of mRNA copies in 1x106 cells from the data given by the droplet reader. For example: the detection of all HLA-G isoforms in JEG-3 revealed a concentration of 817 copies/µl sample. Our total sample volume was 20 µl, which are 16,340 copies/20 µl sample. The 20 µl sample contained 0.1 µl cDNA, which are 163,400 copies/µl cDNA. For the cDNA synthesis, we used 3 µl of RNA, which are 1,089,333 copies/µl RNA. From 1x106 cells we purified 30 µl of RNA. Therefore, the total copies per 1x106 cells are 30x1,089,333=3.27x107. The calculations for all the cell lines can be found in “Appendix 6”.

Figure 17: Overview of the steps in this experiment [Modified from Biocompare, 13.06.14; Bio-Rad (a) & (b), 13.06.14; Fine4fit; 13.06.14 and our own material]. For each cell line, RNA was purified from 1x106 cells and eluted in 30 µl water. Three µl RNA was used for cDNA synthesis in a volume of 20 µl. Depending of the cell line and the target, 0.1-5 µl of cDNA was mixed with primers, probe and 2x ddPCRTM Supermix for Probes (Bio-Rad, Hercules, California). Before the PCR, droplets were made of the mix and oil in a QX-100 Droplet Generator (Bio-Rad, Hercules, California). The droplets were analyzed on QX100 Droplet reader (Bio-Rad, Hercules, California).

Results The results from this experiment are presented as mRNA copies per 1 million cells in “Figure 18” and “Figure 19” below. The exact values are found in “Table 5” in the end of the result paragraph.

60

6

8 .0  1 0

6

3 .0  1 0

6

6 .0  1 0

6

2 .0  1 0

6

4 .0  1 0

6

1 .0  1 0

6

2 .0  1 0

6

/6

ll

5 -G A

-G

L

A

H

L H

/6

ll

-G

5

A

A

-G

L

A

H

L

4 .0  1 0

6

4 .0  1 0

6

3 .0  1 0

6

3 .0  1 0

6

2 .0  1 0

6

2 .0  1 0

6

1 .0  1 0

6

1 .0  1 0

6

/6 L H

H

L

A

A

-G

-G

5

A

-F A L H

H

L

A

5 -G H

L H

L

A

A

L H

-E

/6

-F A

-E A L H

L

ll

0

/6 -G

A

-F H

-F A L

H 6

H

H

L

A

A

-G

L H

H

L H 5 .0  1 0

C o p ie s / 1 m c e lls

6

5

A

-F A

-E A

-E A

5 -G A

H

H

IG R -1

5 .0  1 0

0

ll

0

L

L

A L H

C o p ie s / 1 m c e lls /6

ll L

L

L H

6

H

A

-E

/6 5 -G A

1000000

S K -M E L -3

C o p ie s / 1 m c e lls

5 .0  1 0

A

-E A

5 -G H

L

L

A

A

-G

L H

H 7

2000000

0

/6

ll A

-F A

-E A L H

C o p ie s / 1 m c e lls

1 .0  1 0

3000000

0

S K -M E L -2 8 7

H

L H 6

0

1 .5  1 0

L

A

H

2 .0  1 0

-G

6

6

A

1 .0  1 0

4 .0  1 0

ll

6

6

A

2 .0  1 0

6 .0  1 0

-G

6

-F

3 .0  1 0

F M -5 6 4000000

A

6

6

L

4 .0  1 0

F M -4 5 8 .0  1 0

H

6

C o p ie s / 1 m c e lls

C o p ie s / 1 m c e lls

F M -6 5 .0  1 0

A -G

A L

-G

H

H

L

L

A

A

-G

L H

-E

/6 5

A

-F A

-E A L H

0

ll

0

ll

0

4 .0  1 0

C o p ie s / 1 m c e lls

7

7

-F

2 .0  1 0

1 .0  1 0

A

7

F M -5 5 M 2

6

L

4 .0  1 0

B eW o 5 .0  1 0

H

7

C o p ie s / 1 m c e lls

C o p ie s / 1 m c e lls

J E G -3 6 .0  1 0

Figure 18: HLA class Ib mRNA expression pattern of the cell lines included in the study. Analyzed results from ddPCR. The bars represent the mean of independent duplicates and +/- SD are indicated. The x-axis shows the number of copies per one million cells. Note that the values of the x-axis differ between the plots. JEG-3 is the cell line which express the highest level of HLA-G mRNA, both membrane-bound and soluble isoforms, and the lowest level of HLA-E and HLA-F mRNA. All the malignant melanoma cell lines and BeWo exhibit relative high expression of HLA-E compared to their expression of the other HLA class Ib. All the cell lines except JEG-3 had higher mRNA level of HLA-F compared to mRNA of the membrane-bound or soluble isoforms of HLA-G.

Of all the class Ib mRNA investigated, HLA-E is the most expressed among all cell lines, with SK-MEL-28 showing the highest expression (see “Figure 18” and “Figure 19”). This cell line is also the one showing the highest expression of HLA-F mRNA. For our positive controls JEG-3 and BeWo the HLA-F mRNA is so low, that it is not possible to see in either “Figure 18” or “Figure 19”. For the exact values see “Table 5”. JEG-3 HLA-G and HLA-G5/6 mRNA are far more expressed in comparison to the other cell lines. Therefore, we have chosen not to plot

61

JEG-3 HLA-G5 and HLA-G5/6 in the same graph as the other cell lines. Among the malignant melanoma cell lines IGR-1 showed the highest mRNA expression of HLA-G all, followed by FM55M2 and FM-56 respectively. Messenger RNA expression of HLA-G5/6 is not detected in FM56, which did show relatively high expression of HLA-G all mRNA. Among the malignant melanoma cell lines, the highest expression of HLA-G5/6 mRNA was found in IGR-1 and FM55M2. For some of the cell lines the level of HLA-F, HLA-G all and HLA-G5/6 is difficult to read on “Figure 18” and “Figure 19”. Therefore, we refer for all of these to “Table 5” for the exact values.

Figure 19: Comparison of HLA class Ib mRNA levels between cell lines included in the study. Analyzed results from ddPCR. The bars represent the mean of independent duplicates and +/- SD are indicated. The x-axis shows the number of copies per one million cells. Note that the values of the x-axis differ between the plots. The mRNA level of HLA-E did not differ as much between cell lines as did the mRNA level of the other HLA class Ib molecules. JEG-3 had a remarkably higher level of HLA-G mRNA, both membrane-bound and soluble isoforms, compared to the other cell lines. Conversely, when it comes to HLA-E and HLA-F, JEG-3 exhibited the lowest mRNA level. IGR-1, FM-55M2 and FM-56 had a relatively high mRNA level of membrane-bound isoforms of HLA-G compared to the other cell lines. This is also the case when it comes to the soluble isoforms, with the exception that FM-56 did not express mRNA of soluble isoforms of HLAG. Of all the cell lines included in the study, SK-MEL-28 had by far the highest expression of HLA.F mRNA. The choriocarcinoma cell lines JEG-3 and BeWo had the lowest levels HLA-F mRNA.

62

Table 5

HLA-G all

HLA-G5/6

HLA-E

HLA-F

JEG-3

3,91x107

4,55x106

2,77x106

5,24x103

BeWo

7,34x104

1,73x104

3,74x106

1,29x104

FM-55M2

1,23x105

1,87x104

7,67x106

1,72x106

FM-6

1,13x103

4,84x102

4,50x106

8,45x105

FM-45

1,24x103

5,35x102

6,52x106

6,79x104

FM-56

3,73x104

0,0x100

2,92x106

1,16x106

SK-MEL-28

3,61x103

3,96x101

9,04x106

5,91x106

SK-MEL-3

2,52x103

8,76x102

4,55x106

1,78x105

IGR-1

2,12x105

6,96x104

4,02x106

4,77x105

Discussion As expected, we detected HLA-E mRNA expression in all cell lines. The level of HLA-E was very similar between all the cell lines compared to HLA-F, HLA-G and HLA-G5/6. Transcripts of all HLA class Ib genes were detected in both of the control cell lines (JEG-3 and BeWo). However, the level of HLA-F was very low, compared to most of the malignant melanoma cell lines. IGR1 was negative for HLA-F protein expression (“Experiment 1”), but we observed that it does transcribe HLA-F. All cell lines exhibited mRNA expression of membrane-bound HLA-G and in most of them we also detected mRNA expression of soluble HLA-G, but not in FM-56. This is remarkable since FM-56 was the cell line with the highest percentage of cells positive for protein expression of soluble HLA-G. Our primer set for HLA-G5/6 does not detect HLA-G7 and therefore we thought that the protein expression observed in “Experiment 1” could be protein expression of HLA-G7. However, HLA-G mAb 5A6G7 is raised against a polypeptide corresponding to a sequence of intron 4 [Le Rond et al., 2004]. HLA-G7 should not be detected since it does not contain intron 4 (see “Figure 3” p. 16). The mRNA expression pattern is very different in JEG-3 compared to the other cell lines. The most abundant mRNA in JEG-3 is membrane-bound and soluble isoforms of HLA-G. In contrast, for all the other cell lines we observed that the most abundant mRNA was HLA-E.

63

Similar, the HLA-F transcript level in JEG-3 was very low, while the level of HLA-F in all the malignant melanoma cell lines was higher than their level of HLA-G. Although we found no protein expression of HLA-E in BeWo (“Experiment 1”) we here observed that BeWo does transcribe HLA-E. As described in the chapter “Human Leukocyte Antigen E”, stable surface expression of HLA-E requires peptides derived from HLA-G or HLA class Ia molecules. As observed in “Experiment 1”, a very low percentage of BeWo cells had protein expression of HLA class Ia and HLA-G. Therefore, this could explain why HLA-E is not expressed on the surface although HLA-E transcripts are present. In “Experiment 1”, we observed a possible correlation between primary malignant melanoma cell lines (FM-45, FM-56, SK-MEL-28) and protein expression of soluble HLA-G. However, this possible correlation was not observed for mRNA expression of HLA-G5 and -G6.

Optimization of the experiment For this experiment, we initially performed optimization experiments, where we determined the optimal PCR programs for each primer/probe set as well as the optimal amount of cDNA used in the PCR. The optimal annealing temperature differs between the primer sets and since the PCR products of the HLA-E and HLA-F primers are longer than those of the other primers, the denaturing and extension steps needs to be longer. The optimal would be to do a multiplex experiment in which our housekeeping cDNA and our target cDNA are both PCR amplified in the same sample. A multiplex like that would enable us to compare the expression levels of the different mRNAs by standardizing the values to the housekeeping gene. We tried to do this but it was not possible since the different primer sets and probes did not work under the same PCR conditions, or with the same concentration of cDNA. Another approach concerning the experimental procedure could be to use the same amounts instead of the same volumes. To do this we should use the same amount (ng) of RNA from each cell line, instead of the same volume when synthesizing cDNA. For the PCR, we should have measured the cDNA concentration and added an amount of cDNA, which resulted in the same cDNA concentrations for each PCR.

64

Using the Poisson distribution analysis, the optimal percentage of positive droplets is in the range of 60-70%. For some of the targets investigated, the RNA expression in the cells was so low, that this optimal percentage was simply not possible to reach. I addition, for some of the samples, we had problems with droplets with dim signals. We tried to optimize this, but without any luck. These droplets with dim signals made it difficult in some samples to set threshold values for positive/negative droplets.

65

Experiment 3 – Fragment analysis of HLA-G isoforms in malignant melanoma cell lines Aim The aim was to investigate the variation of HLA-G isoform expression in malignant melanoma cell lines. In addition, we will detect if the cell lines have the 14 bp insertion or deletion polymorphism in the 3’UTR of HLA-G.

Hypothesis We expected that there might be differences in the pattern of HLA-G mRNA isoforms among the malignant melanoma cell lines. The malignant melanoma metastasis might be more aggressive than the primary malignant melanoma tumors. This might be reflected by different expression patterns of HLA-G isoforms between the cell lines established from the metastases (FM-6, FM55M2, SK-MEL-3 and IGR-1) and the cell lines established from primary tumors (FM-45, FM-56 and SK-MEL-28). Furthermore, a 14 bp insertion polymorphism has been associated with reduced HLA-G mRNA expression [reviewed in Dahl & Hviid, 2012]. Based on this, we expected not to observe this insertion polymorphism in the cell lines for which we detected relatively high HLA-G mRNA expression by Droplet Digital PCR (FM-55M2, FM56 and IGR-1).

Materials and methods Cell lines The study included the human malignant melanoma cell lines FM-6, FM-45, FM-55M2, FM-56, SK-MEL-3, SK-MEL-28 kindly provided by Prof. Mads Hald Andersen (Herlev Hospital, DK) and IGR-1 (DSMZ, Braunschweig, Germany). Furthermore, two HLA class Ib positive controls were included in the study. JEG-3 (Sigma-Aldrich, St. Louis, Missouri) and BeWo (SigmaAldrich, St. Louis, Missouri), which are choriocarcinoma cell lines. Cell lines provided by Prof. Mads Hald Andersen were maintained in Roswell Park Memorial Institute (RPMI) 1640 + Glutamax with 10% FBS; IGR-1 were maintained in Dulbecco’s Modified Eagle Medium (DMEM) with 10% FBS; BeWo were maintained in Kaighn’s Modification of Ham’s F-12 (F12K) Medium with 10% FBS; and JEG-3 were maintained in Eagle’s Minimum Essential Medium (EMEM) with 10% FBS. All cell lines were 80-90% confluent when harvested for RNA purification.

66

RNA purification and cDNA synthesis From each cell line, total RNA was purified from 2x1 million cells using the RNeasy® Mini Kit (Qiagen). For detailed protocol see “Appendix 4”. Three µl (1209-2593.5 ng) of RNA (260/280 beween 1.97-2.08), measured by Thermoscientific µDrop™plate, were used for first strand cDNA synthesis with SuperScript® VILO™ cDNA synthesis kit (Invitrogen). For a detailed protocol see “Appendix 5”. PCR Separate PCRs were run for each of the one million cells purified per cell line. Each reaction contained 13.5 µl 2x Go Taq qPCR Master Mix (Promega, Fitchburg, Wisconsin), 10 pmol (2 µM) primers and 2.5 µl template cDNA in a total volume of 25 µl. For details on the primer set please see “Table 6”. The PCRs were run with following conditions: 2 minutes at 95 °C, 40 cycles of: 30 second denaturing at 94 °C, 1 minute annealing at 62 °C and 3 minute extension at 72 °C, and finally a 10 minute step at 72 °C. Table 6: Primers used in the experiment Amplicon HLA-G1-6

Forward primer

Revers primer

5G2RNA

RHG4CY5

5’-CACAGACTGACAGAATGAACCTGCA-3’

5’-Cy5-GGAAGGAATGCAGTTCAGCATGA - 3’

Fragment analysis For each cell line, we had duplicates of PCR product, originating from two different stocks of one million of cells. Furthermore, for each cell line we measured each of the PCR products in duplicates. For JEG-3 and BeWo 2 µl of PCR product and from the other cell lines 4 µl of PCR product was mixed with 0.5 µl GeneScanTM 1200 LIZ Size Standard (Applied Biosystems, Foster City, California) in a total volume of 12.5 µl. The samples were denatured, then cooled down and analyzed on ABI3500 (Applied Biosystems, Foster City, California). Analysis of fragment length data We analyzed the fragment length results using GeneMarker V2.4.0 (Softgenetics). When analyzing the results we identified if peaks were present at known positions for the different isoforms. This was necessary since the peak level for some of the isoforms, for example HLA-

67

G5 (+14 bp), was often not higher than the background noise, but due to consistency within the measured samples we have accepted these as positive peaks. For the different cell lines, an average peak height for each of the isoforms was calculated. The intensities are presented as “NC” meaning signal in ≤ half of the samples, “(+)” meaning signal in > half of the samples - but not all, “+” meaning signal lower than 1000, “++” meaning signal between 1000-4999, ”+++” meaning signal between 5000-9999, “++++” meaning signal ≥ 10,000.

Results

Figure 20: Results from a representative fragment analysis. The electropherogram is from fragment analysis on JEG3 cDNA. On the electropherogram it is marked which HLA-G isoforms were detected.

The results from one of the cell lines are shown in “Figure 20”, where we have marked the different isoforms detected. “Table 7” shows the analyzed results. Our results show that JEG-3 expressed both membrane-bound and soluble isoforms of HLA-G. The HLA-G isoforms expressed at the highest level by JEG-3 are HLA-G1, HLA-G3, and HLA-G2/4. The soluble isoforms of HLA-G5 and HLA-G6 were also detected. The +14 bp isoform was present for all the detected isoforms, the -92bp isoform was present for HLA-G1, HLA-G3, and HLA-G2/4. The -14 bp form was only detected for HLA-G3. BeWo as well expressed both membrane-

68

bound and soluble isoforms of HLA-G. The HLA-G isoforms expressed at the highest level by BeWo, was HLA-G1 and HLA-G3. HLA-G2/4 and the soluble isoform HLA-G5 were also detected. Whether or not BeWo expressed HLA-G6 was based on our results unclear. The +14 bp variant was present for all detected isoforms, the -92bp variant was present for HLA-G1. IGR-1 expressed both membrane-bound and soluble isoforms of HLA-G. The HLA-G isoforms expressed at the highest level by IGR-1, were HLA-G1, HLA-G3 and HLA-G2/4. HLA-G5 and HLA-G6 were also expressed by this cell line. The -14 bp variant was present for all detected isoforms. FM-6 may express HLA-G1 and HLA-G3, but the results were not consistent. FM-45 expresses the +14 bp HLA-G1. It may express HLA-G5, but the results were not consistent. FM55M2 expressed the membrane-bound isoforms HLA-G1 and HLA-G2/4 and it may express HLA-G3, HLA-G5 and HLA-G6, but the results were not consistent. For HLA-G1, both the -14 bp and the +14 bp variants were detected. For HLA-G2/4 the +14 bp variant was detected. FM-56 expressed the +14 bp HLA-G1 isoform. It may express all other isoforms but these results were not consistent. SK-MEL-3 and SK-MEL-28 may express HLA-G1 and HLA-G2/4, but the results were for both not consistent.

Table 7: Analyzed results of the HLA-G mRNA isoform fragment length analysis. “NC” = signal in ≤ half of the samples, “(+)” = signal in > half of the samples, but not all, “+” = signal lower than 1000, “++” = signal between 1000-4999, ”+++” = signal between 5000-9999, “++++” = signal ≥ 10,000. The intensities indicated in plusses are the mean of eight different fragment analyses. From each cell line, we have purified RNA from 2 x 1 million cells in the same passage and synthesized cDNA from this. We made two separate PCRs with each cDNA. For JEG-3 and BeWo we used 2 µl and from the other cell lines we used 4 µl of these PCR products for the fragment analysis. For each PCR product, we made a duplicate of the fragment analysis. The mean intensities of JEG-3 and BeWo are not directly comparable to the mean intensities of the other cell lines due to the different amounts of PCR product used in the fragment analysis. The PCR included 40 cycles to obtain any signal and because the exponential phase of HLA-G amplification ends at about 30 PCR cycles the results are only semi-quantitative. Clear mRNA expression of HLA-G membranebound and soluble isoforms were found for the cell lines JEG-3, BeWo and IGR-1. FM-45, FM-55M2 and FM-56 showed expression of membrane-bound HLA-G. The results for FM-6, SK-MEL-3 and SK-MEL-28 were not consistent.

Discussion Consistent with our results, JEG-3 has been shown to express the HLA-G1, -G2, -G3 and -G4 mRNA isoforms [Hiby et al., 1999], as well as soluble isoforms of HLA-G [Sangrouber et al.,

69

2007]. Furthermore, JEG-3 has been shown to be homozygous for an allele of HLA-G containing the 14 bp insertion polymorphism [Hiby et al., 1999]. In contrast to this, our fragment analysis revealed a fragment of the same size as the -14 bp HLA-G3. However, the signal was very weak and was most likely noise. It has been shown, that the 14 bp insertion polymorphism is associated with reduced expression of HLA-G mRNA [reviewed in Dahl & Hviid, 2012]. Of the malignant melanoma cell lines included in the study, IGR-1 and FM55M2 are the two cell lines in which the 14 bp deletion polymorphism is observed in all replicates. Interestingly, it is the two malignant melanoma cell lines that exhibited the highest level of HLA-G mRNA in “Experiment 2”. Moreover, from our results it seems that IGR-1, the cell line with the highest HLA-G mRNA expression is homozygote for the deletion of the 14 bp sequence since the only fragment that was observed in all replicates was from the -14 bp form of HLA-G. The detected HLA-G mRNA for the malignant melanoma cell lines (“Experiment 2”), except for IGR-1, were generally were low compared to the detected HLA-G mRNA for JEG-3. Because the level of HLA-G mRNA is low, it is difficult to detect the specific isoforms in the fragment analysis. For some of the fragments corresponding to specific HLA-G mRNA isoforms, the signal was so low that it was difficult to distinguish specific signals from the background noise.

Optimization of the experiment We tested the HLA-G primer set with JEG-3 in a real-time PCR setup and found that the exponential phase of HLA-G amplification was about 28 cycles (see “Figure 21”). However, we had to perform the PCR for 40 cycles to obtain any signal when analyzing the PCR products on the ABI3500 (Applied Biosystems, Foster City, California) instrument and the results are therefore only semi-quantitative.

70

Figure 21: Real-time PCR with HLA-G primers. The exponential phase of HLA-G amplification was about 28 cycles.

On the ABI3500 (Applied Biosystems, Foster City, California) there was several settings that we optimized during the first tests i.e. run and injection voltage and injection time. During the optimizing tests (data not included), we realized that the intensities of the fragments in our investigated cell lines were considerably lower than in our control cell lines and therefore decided to use different volumes of PCR product for the fragment analysis. For JEG-3 and BeWo we used 2 µl and from the other cell lines we used 4 µl. Due to the different amounts of PCR product used in the fragment analysis, the mean intensities of JEG-3 and BeWo are not directly comparable to the mean intensities of the other cell lines.

71

Experiment 4 – Flow cytometric analysis of PBMCs cocultured with malignant melanoma cells lines Aim The aim was to characterize possible differences in the amount and/or distribution of PBMC subtypes prior to and after co-culture with malignant melanoma cells.

Hypothesis In previous experiments (see “Experiment 1” and “Experiment 2”), we have characterized the protein and mRNA expression of HLA class Ia and Ib for the cell lines included in this experiment. We expected that co-culture of peripheral blood mononuclear cells (PBMCs) with malignant melanoma cell lines expressing HLA-G would increase the fraction of regulatory T cells (Tregs), a tolerogenic cell population. This hypothesis is based on findings of previous studies showing a correlation between high HLA-G expression and induction of Tregs [Du et al., 2011; Chen et al., 2010; Tuncel et al., 2013; Rizzo et al., 2014]. In vitro, DC-10, a tolerogenic subtype of dendritic cells (DCs), has been shown to induce differentiation of a specific group of Tregs [Gregori et al., 2010]. Thus, it can be hypothesized that the effect of HLA-G on Treg levels is trough induction of DC-10. If this is the case, we should observe a correlation between HLA-G and DC-10 as well. As mentioned in the chapter “The interaction of HLA class Ib with immune cells”, tumor cells can transfer HLA-G to immune cells by trogocytosis. Therefore, the level of HLA-G-positive immune cells could be increased by co-culture with cell lines expressing HLA-G. Furthermore, HLA-G could induce expression of its receptor, ILT-2, on immune cells. Many studies have focused on HLA-G expression, but HLA-E and HLA-F expression might also have an effect on the level of tolerogenic cells.

Preparation Titration of antibodies All antibodies were titrated before this pilot study. The results are shown in “Appendix 7”. Creating the experimental setup We designed panels for detection of Tregs (“Table 8”), immature DCs (“Table 9”) and DC-10s (“Table 10”). To awoid intracellular staining, we decided to use CD127 instead of FoxP3 for 72

the detection of Treg cells. In general, FoxP3-positive Tregs are defined based on expression of CD4, CD25 and FoxP3. However, a study by Liu et al. (2006) showed that CD127 expression inversely correlates with FoxP3, and that a combination of CD4+CD25+CD127low resulted in a highly purified population of Tregs, which support the use of CD127 as a biomarker for Treg. Table 8: Tregs flow cytometry panel

Laser

Filter

Blue

Maker

PE

(530/30)

(585/40)

(HIL-7RM21) IgG1

Isotype

PerCPCy5.5

FITC

CD127

(MOPC21)

Red PE-Cy7 (780/60)

(695/40)

ILT-2

CD45

CD25

(GHI/75)

(HI30)

(M-A251)

IgG2b

IgG1

(27-35)

(MOPC-21)

IgG1 (MOPC21)

APC (660/20) HLA-G (MEMG/9)

Violet APCeFluor 780 (780/60)

Pacific Blue

Krome Orange

(440/40)

(550/40)

CD4

CD3

CD8

(OKT4)

(UCHT1)

(B9.11)

IgG2b

IgG1 (P3.6.2.8.1)

(eBMG2b)

IgG1

IgG1

(MOPC21)

(679.1Mc7)

Table 9: Dendritic cells flow cytometry panel I

Laser

Filter

Maker

Blue

Red PerCPCy5.5

FITC

PE

(530/30)

(585/40)

CD83 (HB15e)

CD209

CCR7

CD11c

(DCN46)

(150503)

(B-ly6)

(695/40)

PE-Cy7 (780/60)

APC (660/20) HLA-G (MEMG/9)

Violet

APCeFluor 780 (780/60) HLA-DR (LN3)

Pacific Blue

Krome Orange

(440/40)

(550/40)

Lineage cocktail*

CD45 (J.33)

IgG1 IgG1 Isotype

(MOPC21)

IgG2b (27-35)

IgG2a (G155178)

IgG1 (MOPC21)

IgG1

IgG2b

(MOPC21)

IgG1

(P3.6.2.8.1)

(eBMG2b)

IgG2a

(679.1Mc7)

(MOPC173) *CD19 (HIB19), CD56 (HCD56) and CD3 (OKT3).

73

Table 10: Dendritic cells flow cytometry panel II

Laser

Filter

Maker

Blue

Red PerCPCy5.5

FITC

PE

(530/30)

(585/40)

CD83 (HB15e)

CD16

CD14

CD11c

(HI16a)

(MP9)

(B-ly6)

(695/40)

PE-Cy7 (780/60)

APC (660/20) HLA-G (MEMG/9)

Violet

APCeFluor 780 (780/60) HLA-DR (LN3)

Pacific Blue

Krome Orange

(440/40)

(550/40)

Lineage cocktail*

CD45 (J.33)

IgG1 IgG1 Isotype

(MOPC21)

IgG2b (27-35)

IgG2a (G155178)

IgG1 (MOPC21)

IgG1

IgG2b

(MOPC21)

IgG1

(P3.6.2.8.1)

(eBMG2b)

IgG2a

(679.1Mc7)

(MOPC173) *CD19 (HIB19), CD56 (HCD56) and CD3 (OKT3).

Materials and Methods Cell lines and PBMCs The study included the human malignant melanoma cell lines: FM-45, FM-55M2, FM-56, SKMEL-28 kindly provided by Prof. Mads Hald Andersen (Herlev Hospital, DK) and IGR-1 (DSMZ, Braunschweig, Germany). Furthermore, one HLA-G-positive control was included in the study, JEG-3 (Sigma-Aldrich, St. Louis, Missouri) which is a choriocarcinoma cell line. Cell lines provided by Prof. Mads Hald Andersen were maintained in Roswell Park Memorial Institute (RPMI) 1640 + Glutamax with 10% FBS; IGR-1 were maintained in Dulbecco’s Modified Eagle Medium (DMEM) with 10% FBS; and JEG-3 were maintained in Eagle’s Minimum Essential Medium (EMEM) with 10% FBS. All cell lines were 80-90% confluent when harvested for flow cytometry. Human peripheral blood mononuclear cells (PBMCs) were isolated from blood obtained from six healthy donors using Lymphoprep (Axis-Shiled PoC AS, Norton, Massachusetts). See data concerning the donors in “Table 11”. The PBMCs were frozen in liquid nitrogen after isolation, with approximately 11 million cells per cryotube.

74

Table 11 Donor

Gender

Age (years)

1

Female

56

2

Female

39

3

Female

32

4

Male

56

5

Male

57

6

Male

47

Co-culture experiments Day 1: From each of the six cell lines 50,000 cells, in a volume of 0.6 ml suitable media, were transferred to 54 wells in three 24 well plates and incubated for 24 hours at 37 °C/5% CO2. Day 2: After 24 hours one frozen vial of PBMCs from each of six donors were thawed and added directly to nine wells (0.6 ml suitable media) containing the cancer cells and plates were incubated at 37 °C/5% CO2 for 24 hours. Day 3: The media including PBMCs from the nine wells with PBMCs from the same donor were pooled. To detach the cells that adhere to the bottom of the plate (cancer cells and dendritic cells) the cells were trypsinized with 0.25% trypsin/EDTA. After trypsinization, the cells were pooled with the PBMC that did not attach. The PBMCs from the same donor were resuspended in 500 µl FACSFlow (BD Bioscience, San Jose, California) and 100 µl were aliquot to each of the tubes for the dendritic cell setups (four tubes) and 50 µl were aliquoted to each of the tubes for the Treg setup (two tubes). Since the cell lines were cultured in three different kinds of media, we made control plates with PBMCs growing in the three different media without cancer cells. For a detailed protocol on the co-culture experiments see “Appendix 8”.

Figure 22: Experimental setup of the co-culture experiments. A) The cell lines were cultured in three different media. Therefore, we included controls with PBMCs cultured in each of these three media without malignant cell lines. B) Cell lines were cultured for 24 hours before adding PBMCs (ratio 1:20). The cells were co-cultured for

75

additional 24 hours and harvested for flow cytometric analysis. C) Photo of co-culture with the malignant melanoma cell line FM-56 and PBMCs. 40x magnification.

Flow cytometry Cells were incubated with 10% mouse serum to inhibit non-specific binding of mouse antibodies. Cells from one tube of each setup were stained with marker mAbs. Cells from the other tubes were stained with isotype matched mouse mAbs for use as negative controls. Cells were resuspended in FACSFlow (BD Bioscience, San Jose, California) and immediately analyzed on BD FACSCantoTM II (BD Bioscience, San Jose, California). For a detailed protocol see “Appendix 8”. Analysis of flow cytometry data Analysis of flow cytometry data were performed with the use of BD FACSDiva Software v6.1.3 (BD Bioscience, San Jose, California). Tregs were gated as CD45+CD3+CD4+CD25highCD127low cells. Immature dendritic cells were gated as CD45+CD11c+lin-HLA-DR+CD209+CCR7-CD83cells. Tolerogenic DC-10s were gated as CD45+CD14highCD16highCD11c+lin-HLA-DR+CD83+HLAG+ cells. Gating of Tregs: In the marker tubes, lymphocytes were gated based on CD45 and SSC-A. In this population, CD4+ lymphocytes were gated based on a CD3 vs CD4 plot. To gate the Tregs from this population, CD25 vs CD127 were plotted. Since Tregs are CD27low or negative, the gate was placed according to this (see “Figure 23” below). In the isotype control tubes, we used the lymphocyte/monocyte population (gated by SSC-A and FSC-A) to gate on the plot of the isotype controls for CD25 and CD127. The gate on the CD25 isotype axis was adjusted until maximum 1% positive events were included. This gate was copied to the CD25 vs CD127 plot in the corresponding marker tubes and the percentage of positive events (CD25highCD127low) in the CD45+CD3+CD4+ population from this gate minus the percentage of positive events in the same gate of the isotope control tube was recorded. Gating of CD4+HLA-G+ T cells: The percentage of CD4+HLA-G+ T cells were determined by plotting the CD4+ T cells in a HLA-G vs CD4 plot. The gate was set in the isotype tube, where maximum 1% positive events were accepted. The percentage of positive events in the isotype control tube was subtracted from the percentage of positive events in the corresponding marker tubes. Gating of CD4+HLA-G+ILT-2+ T cells: The percentage of CD4+HLA-G+ILT-2+ T cells were

76

determined by plotting the CD4+ T cells in a HLA-G vs ILT-2 plot. The gate was set in the isotype tube, where maximum 1% positive events were accepted. The percentage of positive events in the isotype control tube was subtracted from the percentage of positive events in the corresponding marker tubes. Gating of CD8+HLA-G+ T cells and CD8+HLA-G+ILT-2+ T cells: From the lymphocyte population, we gated CD8+ T cells based on a CD3 vs CD8 plot. CD8+HLA-G+ and CD8+HLA-G+ILT-2+ T cells were gated using the same strategy as for CD4+HLA-G+ and CD4+HLA-G+ILT-2+ T cells. Gating of Immature dendritic cells: In the marker tubes, CD45+ cells were gated based on CD45 and SSC-A. This population was shown in a CD11c vs lin- plot, where no cells were negative for the lineage cocktail. Therefore, we could not gate further. Gating of DC-10s: In the marker tubes, CD45+ cells were gated based on CD45 and SSC-A. This population was shown in a CD14 vs CD16 plot, where no cells were double positive. Therefore, we could not gate further.

Figure 23: Gating strategy of HLA-G+CD4+ T cells. In the isotype control tubes, the lymphocyte/monocyte population was gated by SSC-A and FSC-A. In the marker tubes, lymphocytes were gated based on CD45 and SSC-A. In this population, CD4+ lymphocytes were gated based on a CD3 vs CD4 plot. To gate the CD4 +HLA-G+ T cells from this population, HLA-G vs CD4 were plotted. In the isotype control tubes, we used the lymphocyte/monocyte population (gated by SSC-A and FSC-A) to gate on the plot of the isotype controls for HLA-G and CD4. The gate on the HLA-G isotype axis was adjusted until maximum 1% positive events were included. This gate was copied to the HLA-G vs CD4 plot in the corresponding marker tubes. The percentage of positive events in the isotype control tube was subtracted from the percentage of positive events in the corresponding marker tubes.

Statistical analysis The GraphPad Prism software v6.02 was used for all statistical analysis. For all co-cultures, the results were analyzed in relation to the respective control and not in relation to the other co-cultures. To evaluate if co-culture of PBMCs with malignant cell lines increased the level of

77

CD4+CD25+CD127low Tregs a paired t-test assuming Gaussian distribution was performed. Other data from the experiment were analyzed using Wilcoxon matched-pairs signed rank test. Differences were considered significant when the p-value was