Journal of Biotechnology 161 (2012) 92–103

Contents lists available at SciVerse ScienceDirect

Journal of Biotechnology journal homepage: www.elsevier.com/locate/jbiotec

Autodisplay of enzymes—Molecular basis and perspectives Joachim Jose a,∗ , Ruth Maria Maas b , Mark George Teese a a b

Institut für Pharmazeutische und Medizinische Chemie, Westfälische Wilhelms-Universität Münster, D-48149 Münster, Germany Autodisplay Biotech GmbH, Merowingerplatz 1a, D-40225 Düsseldorf, Germany

a r t i c l e

i n f o

Article history: Received 8 October 2011 Received in revised form 14 February 2012 Accepted 4 April 2012 Available online 30 April 2012 Keywords: Autodisplay Biocatalysis Synthesis Enzymes Whole cells

a b s t r a c t To display an enzyme on the surface of a living cell is an important step forward towards a broader use of biocatalysts. Enzymes immobilized on surfaces appeared to be more stable compared to free molecules. It is possible by standard techniques to let the bacterial cell (e.g. Escherichia coli) decorate its surface with the enzyme and produce it on high amounts with a minimum of costs and equipment. Moreover, these cells can be recovered and reused in several subsequent process cycles. Among other systems, autodisplay has some extra features that could overcome limitations in the industrial applications of enzymes. One major advantage of autodisplay is the motility of the anchoring domain. Enzyme subunits exposed at the cell surface having affinity to each other will spontaneously form dimers or multimers. Using autodisplay enzymes with prosthetic groups can be displayed, expanding the application of surface display to the industrial important P450 enzymes. Finally, up to 105 –106 enzyme molecules can be displayed on a single cell. In the present review, we summarize recent achievements in the autodisplay of enzymes with particular attention to industrial needs and process development. Applications that will provide sustainable solutions towards a bio-based industry are discussed. © 2012 Elsevier B.V. All rights reserved.

1. Introduction Enzymes as biocatalysts show some outstanding advantages for the synthesis of chemicals (Choi, 2009; Luetz et al., 2008), pharmaceutical and agrochemical intermediates (Fischer and Pietruszka, 2010; Tufvesson et al., 2011) as well as active pharmaceutical and agrochemical compounds (Ran et al., 2009). Unlike conventional organic chemistry, enzymes can be used under mild conditions concerning temperature, pressure and pH and usually they convert a substrate with high regio- and enantioselectivity without protecting and de-protecting steps as necessary in conventional organic chemistry. In many cases, the use of enzymes in chemical synthesis requires less substrate, less energy and reduces waste. Moreover enzyme discovery and improvement could lead to completely new processes, such as the use of cellulose for fuel production, the bioremediation of contaminated water and soil, or the production of polymers from non–petroleum sources, which under current conditions are not feasible. To date enzymes are already used in a distinct number of industrial processes (Busch et al., 2006). They are either applied as preparations of purified proteins (Goldberg et al., 2007a), or as microorganisms that produce the desired enzyme within the cell

∗ Corresponding author at: Institute for Pharmaceutical and Medicinal Chemistry, Westfälische Wilhelms-Universität, Münster, Hittorfstraße 58-62, D-48149 Münster, Germany. Tel.: +49 251 83 32210, fax: +49 251 83 32211. E-mail address: [email protected] (J. Jose). 0168-1656/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jbiotec.2012.04.001

(Goldberg et al., 2007b). Despite the advantages, there are some drawbacks that prevent a broader application of enzymes. The purification of enzymes is often a complex and costly production process. In most cases purified enzymes cannot be used in repeated reactions, they turn to waste after a single processing step. The use of microorganisms as whole cell biocatalysts avoids the costs associated with enzyme purification and ensures that the enzyme is working in an optimal environment, where all co-factors and regeneration networks are provided. Moreover, the enzyme as a biocatalyst is largely protected from destabilizing and degrading effects. However the intracellular location of the enzymes means that this method will only be successful if both the substrate and product can cross the membrane barrier. In addition, there is a tremendous consortium of other enzymes present within the cell. To obtain the product in a pure and unaltered form, whole cell biocatalysis is also limited to substrates and products that cannot be converted by these native enzymes. For many applications, the display of the enzyme at the cell surface of the microorganism is an advancement of the whole cell biocatalyst approach. Neither substrate nor product needs to be membrane permeable, and both could be excluded from any unwanted attack by other enzymes. Among the systems for the display of recombinant proteins on microorganisms, which include yeast (Kuroda and Ueda, 2011), gram positive (Kronqvist et al., 2010) and gram negative bacteria (van Bloois et al., 2011), the autodisplay system is a particularly elegant and efficient tool with some advantageous features for biotechnological and – if scaled up – industrial applications.

J. Jose et al. / Journal of Biotechnology 161 (2012) 92–103

93

domains remain anchored on the cell surface (Linke et al., 2006). For type Vc autotransporters, the passenger and the translocator domains are provided by separate genes (St Geme and Yeo, 2009). Both domains are transported across the inner membrane by the Sec machinery, and interact in the periplasm via a so-called POTRA (“polypeptide transport associated domain”) domain of the translocator, which initiates transport of the passenger across the outer membrane. This makes type Vc resembling similar transport systems existing in chloroplasts and mitochondria, which are supposed to be able to transport very complex and extended folded protein structures (Tommassen, 2007). Although these particular features of type Vb and Vc autotransporters make them interesting candidates for the surface display of enzymes, they have not been used for this purpose yet. Therefore this review focuses on the application of classical autotransporters in biotechnology.

3. Display of enzymes by classical autotransporters

Fig. 1. Model of the classical autotransporter secretion mechanism. (A) Autotransporters are synthesized as precursor protein containing all domains needed to transport the passenger to the cell surface. (B) By the aid of a classical signal peptide the precursor is transported across the inner membrane, which is cleaved off. Subsequently, the C terminal part folds into the outer membrane as a porin-like structure, a so-called ␤-barrel. The passenger is translocated to the cell surface by the aid of the ␤-barrel maintains an unfolded conformation during transport. According to this model surface translocation requires the formation of an interim hairpin structure, which was recently experimentally verified (Ieva and Bernstein, 2009). Surface translocation is supported by folding of the passenger at the cell surface.

2. Autodisplay Autodisplay is defined as the recombinant surface display of proteins or peptides by means of an autotransporter protein in any gram negative bacterium (Jose and Meyer, 2007). The autotransporter proteins are a large family of secreted proteins in gram negative bacteria and are divided into three subgroups, the classical autotransporters (secretion type Va), the trimeric autotransporter adhesins (Vb), and the two partner secretion systems (Vc) (Henderson et al., 2004). All classical autotransporters are thought to share a common general structure (Jose et al., 1995). They are produced as precursor proteins with a standard signal peptide at the very N terminus, which enables the transport of the precursor protein across the cytoplasmic membrane, most frequently by the Sec machinery (Fig. 1). The signal peptide, which is cleaved off as the protein crosses the inner membrane, is followed by the actual passenger, which will be transported to the cell surface. Outer membrane translocation is facilitated by the C–terminal part of the precursor, which forms a porin–like structure, a so-called ␤-barrel within the outer membrane that is frequently named translocator domain. Because the ␤-strand that closes the barrel is directed towards the periplasm, an additional “linker” domain is required in between the passenger and the translocator in order to enable full surface access of the passenger. There are over thirty examples where the coding sequence of the natural passenger in the autotransporter precursor protein has been replaced by the coding sequence of a recombinant protein, resulting in the transport of the recombinant protein to the cell surface. Maurer et al. (1997) convincingly showed this using the AIDA-I autotransporter, and it was the first report that used the term “autodisplay” for such purpose. The trimeric autotransporters (type Vb) show a similar organization as the classical autotransporters, with the difference that the ␤-domain is rather truncated and cannot function as a monomer. The translocator within the outer membrane is formed by the ␤domains of three precursors and as a consequence, three passenger

Before we come to the transport and the display of recombinant enzymes by the aid of an autotransporter, it appears worth to have a look on their natural passengers, which are enzymes as well. The prototype of an autotransporter protein and the first family member to be discovered middle of the eighties – although not named an autotransporter at that time – was IgA1 protease from Neisseria gonorrhoeae (Halter et al., 1984). Together with its structural description, the very elegant model for outer membrane translocation was proposed, without the requirement of energy or accessory factors (Pohlner et al., 1987), which is still valid as a concept today. Almost a decade later, the first publication to mention the term “autotransporter” listed ten first examples of this protein family (or eleven when IgA1 proteases form N. gonorrhoeae and N. meningitidis are considered to be different examples), among which five enzymes can be found (Jose et al., 1995). Nowadays the autotransporter family of proteins comprises more than 1000 members, among which a considerable number bear proteases or other hydrolases, in particular lipases as natural passengers (Benz and Schmidt, 2011; Wells et al., 2010; Wilhelm et al., 2011). At this point the question arises, for what reason natural autotransporters are not used more frequently for catalytic purposes (Wilhelm et al., 2011). Autotransporter proteins have been discovered first in pathogenic gram negative bacteria and are supposed to represent the largest protein family in this group of microorganisms (Kajava and Steven, 2006). This reflects first that pathogenic bacteria are a far more prominent subject in research as harmless commensals are. Secondly, pathogenicity or the degree of pathogenicity i.e. virulence of a gram negative bacterium often appears to be associated with its proteins displayed at the cell surface, in many cases as a part of an autotransporter. The application of pathogenic bacteria for catalytic purposes bears a safety problem and appears not recommendable for industrial purposes. Therefore it was necessary to express the autotransporter genes from a pathogenic bacterium in laboratory strains of E. coli. However this led in many cases to incompatibility problems and affected also the early work with IgA1 protease from N. gonorrhoeae expressed in E. coli. Although the potential of this enzyme for proteolytic purposes, in particular site specific cleavage of fusion proteins, and for the transport of recombinant proteins was anticipated from the beginning (Pohlner et al., 1992) and the pioneering work based thereon delivered the proof of principle for many following studies (Jose et al., 1996; Klauser et al., 1990, 1992), all these early experiments were performed with E. coli cells grown on agar plates, which is acceptable for basic research but cumbersome for most biotechnical applications. Incompatibility problems due to phylogenetic distant relationship could also account for the incomplete transport of a lipase and an esterase from Bacillus subtilis and Serratia marcescens by the aid of a lipase

94

J. Jose et al. / Journal of Biotechnology 161 (2012) 92–103

from Pseudomonas aeruginosa when expressed in E. coli (Becker et al., 2005). In order to understand what could account for these problems, we need to critically reflect the term autotransporter. It was assigned to this family of proteins based on the observation that surface translocation of the passenger was achieved by the mere transfer of the autotransporter gene from one gram negative bacterium to the other (Jose et al., 1995; Loveless and Saier, 1997; Pohlner et al., 1987). This led to the impression that this type of secretion does not require further accessory factors, and hence the assignment autotransporter. Today it is obvious that autotransporters are a common theme in gram negative bacteria (Benz and Schmidt, 2011). They have been found by sequence comparisons in most if not all pathogenic gram negative bacteria (Nishimura et al., 2010). Recently more than 215 autotransporter genes could be located within the 28 complete genome sequences of E. coli, and their basic structure could be analyzed and phylogenetically compared (Wells et al., 2010). This leads to the assumption that most probably all gram negative bacteria contain autotransporter proteins, and in case additional accessory factors are required for transport, they are available in these bacteria. As a consequence, if an autotransporter protein is expressed recombinantly in another gram negative bacterium than that where it is originated from, it can take use of this machinery being present as a biological principle (Walther et al., 2009). On one hand this would mimic a kind of self-facilitated process which initially made these proteins appointed to be “auto”-transporters, and which now requires some sort of reassessment. On the other hand, this would account also for the observed incompatibility problems in case an autotransporter gene is expressed in a foreign host background. There is a well documented example for this kind of interspecies incompatibility given by Robert et al. (2006), who could show that the restriction in recombinant expression of the neisserial ␤-barrel protein PorA could be overcome by exchanging the sequence of the last C terminal membrane ␤-strand with that of the corresponding E. coli sequence. They called this a C terminal signature sequence by virtue an outer membrane protein assembly machinery recognizes outer membrane proteins including autotransporters (Robert et al., 2006). Outer membrane translocation within the autotransporter secretion pathway is thought to require the assistance of chaperone proteins such as BamA, a component of a hetero-oligomeric complex (Bam complex) with several lipoproteins (BamB-E) (Knowles et al., 2009; Tommassen, 2007), SurA, Skp, DegP and DnaK (Benz and Schmidt, 2011) (Fig. 2). Moreover, intramolecular domains have been identified with chaperone–like function for the support of correct folding of the passenger (Kajava and Steven, 2006; Peterson et al., 2010; Renn and Clark, 2008), but we are still far away from understanding by what molecular mechanism the passenger of an autotransporter protein traverse the outer membrane. Recent studies on the autotransporter Esp (Ieva and Bernstein, 2009; Ieva et al., 2008) were able to experimentally verify the hairpin formation of the linker domain during transport, and also to demonstrate chaperone interactions. It cannot be excluded at the moment that two of the models (reflected by Fig. 1 and Fig. 2) will eventually be absorbed in a common scheme (Benz and Schmidt, 2011). Although these observations clearly indicate that autotransporters do not deserve the prefix “auto” and that their suffix “transporter” is under re-evaluation, we would prefer to keep this name to facilitate convenient discussion. To overcome the pathogenicity and incompatibility problems discussed above, a systematic approach to identify autotransporters with enzyme passengers in harmless commensals or soil bacteria would be helpful. These could serve as translocators for the construction of whole cell biocatalysts with surface displayed enzymes, or as the starting point for a molecular evolution approach

Fig. 2. Bam complex model of autotransporter secretion. In this model the precursor including the passenger folds at least partially already in the periplasm. Transport of the passenger domain across the outer membrane and ␤-barrel integration is facilitated by the Bam complex, thus enabling the surface translocation of folded passenger proteins. This model is according to the model initially developed for Omp85 (Tommassen, 2007). Beside the Bam complex consisting of BamA-E, several chaperones are involved, which appeared also to play a role in the classical secretion model (Fig. 1) (Benz and Schmidt, 2011).

in order to adapt the enzymatic activity of the autotransporter to a synthetic reaction that is in need (Wilhelm et al., 2007). There have been no comprehensive studies on native autotransporter functions, from strains which are not associated with pathogenicity. At this point it would be interesting to know whether there are other enzymatic activities manifested within autotransporter passenger domains beyond the proteolytic and hydrolytic activities identified so far (Henderson et al., 2004). From the thousands of putative autotransporter sequences in public databases to which no known function has been ascribed, we think it is highly likely that many of these will be ascribed enzymatic functions completely unrelated to pathogenicity (Wells et al., 2010). To solve incompatibility problems, one could consider identifying autotransporter–specific chaperones from the original organism, and co–expressing these in the non-pathogenic lab strain in order to optimize autotransporter mediated surface display of enzymes. Although there is experimental evidence for positive effects of such co-expression (Binder et al., 2010; Schlapschy and Skerra, 2011), this would be a case to case solution which might be too cumbersome for biocatalytic purposes directed towards industrial applications. 4. Autodisplay of recombinant enzymes The breakthrough in the autodisplay of recombinant enzymes appeared when a homologous autotransporter was used for expression in a homologous host, namely an E. coli autotransporter in an E. coli host strain. Although other E. coli autotransporters were known at that time, such as Tsh, (Jose et al., 1995) and EspP, (Brunder et al., 1997), the adhesin involved in diffuse adherence (AIDA-I) was chosen, most probably because it was the most thoroughly investigated candidate at that time (Benz and Schmidt, 1989, 1992a,b; Suhr et al., 1996). The term “autodisplay” was coined initially for the use of AIDA-I in the surface display of recombinant proteins in E. coli (Maurer et al., 1997). For autodisplay the ␤-barrel and the linker region of AIDA-I was employed in combination with the signal peptides of various origins (CTB, PelB, OmpA and also the original AIDA-I signal peptide) (Fig. 3). The DNA encoding the recombinant passenger was inserted in frame between the coding regions for the domains indispensable for transport and this resulted in proper surface translocation of the passenger. For successful surface display the artificial construct had to be expressed in an E. coli strain lacking the outer membrane protease (OmpT) such as UT5600 (Elish et al., 1988) or BL21, because it had been shown that this protease cleaved a region within the autodisplay linker, efficiently releasing recombinant passenger proteins from the cell

J. Jose et al. / Journal of Biotechnology 161 (2012) 92–103

95

Fig. 3. Typical structure of a fusion protein precursor for autodisplay by the example of CYP3A4. It consists of a signal peptide at the very N terminus followed by the passenger domain. The signal peptides in use were most frequently derived from CtxB, PelB or OmpA, as well as the natural signal peptide of AIDA-I (Jose and Meyer, 2007). Due to the cloning procedure several amino acids are added at the C terminus and at the N terminus of the passenger. The passenger domain is followed by the so-called linker, which optionally can contain protease cleavage site and tags for antibody detection or specific protein labeling (Jose and Handel, 2003).

surface into the supernatant (Jose et al., 2002; Maurer et al., 1997). Using this system, quite a number of enzymes have been displayed in a functional form on E. coli (Table 1), including ␤-lactamase from E. coli, adrenodoxin (Adx) from Bos taurus, sorbitol dehydrogenase (SDH) from Rhodobacter sphaeroides, different esterases (e.g. ApeE and EstA), nitrilases from Alcaligenes faecalis and Klebsiella pneumoniae, human hyaluronidases, isoprenyltransferase from Aspergillus formigatus and finally a P450 enzyme from Bacillus megaterium (CYP106A2) and the human P450 enzyme CYP3A4. From the list of enzymes which can be displayed (Table 1) it becomes obvious that the origin of the protein, whether it is bacterial or eukaryotic, does not really matter. This is remarkable, because eukaryotic and prokaryotic proteins are composed of different domain structures and are assumed to exhibit different folding behaviour (Netzer and Hartl, 1997). As outlined above,

outer membrane translocation by autodisplay is thought to involve the formation of a hairpin structure (Fig. 1), which would lead to a scenario, in which the C-terminus of the passenger reaches the surface first followed by its N-terminus. This is exactly the opposite of the order in which proteins are released from the ribosome. It would also require that during entire transport, the polypeptide chain needs to be kept in an unfolded conformation, a model which is experimentally supported by several examples (Ieva and Bernstein, 2009; Jose et al., 1996; Jose and Zangen, 2005). In the case of the displayed enzymes discussed here, this is obviously working for proteins that are not naturally secreted, independent whether they are of eukaryotic or prokaryotic origin. To answer the underlying questions, a more systematic approach would be helpful. In particular, it would be useful to understand the role of the linker domain, which contains the so-called autochaperones. It would be

Table 1 Autodisplay of enzymes using different autransporter proteins. Auto-transporter

Autotransporter origin

Enzyme

Passenger origin

Application

Reference

AIDA-I

E. coli

Escherichia coli Bos taurus Burkholderia gladioli Rhodobacter sphaeroides Salmonella enterica Alcaligenes faecalis Homo sapiens Aspergillus fumigatus Bacillus megaterium Homo sapiens Flavobacterium ATCC 27551

Translocation studies Whole cell biocatalysis Whole cell biocatalysis Whole-cell biocatalysis Whole cell biocatalysis Whole-cell biocatalysis Inhibitor screening Whole-cell biocatalysis Whole-cell biocatalysis Drug metabolism studies Bioremediation

(Lattemann et al., 2000) (Jose et al., 2001, 2002) (Schultheiss et al., 2002) (Jose and von Schwichow, 2004a,b) (Schultheiss et al., 2008) (Detzel et al., 2011) (Kaessler et al., 2011) (Kranen et al., 2011) (Schumacher et al., 2012) (Schumacher and Jose, 2012) (Li et al., 2008)

EstA

P. aeruginosa*

ß-Lactamase (bla) bovine adrenodoxin (Adx) esterase A (EstA) sorbitol dehydrogenase esterase (ApeE) nitrilase hyaluronidase (hPH-20) prenyltransferase (FgaPT2) cytochrome P450 106A2 cytochrome P450 3A4 organophosphate hydrolase (opd) lipase (LipA) cutinase lipase

Bacillus subtilis Fusarium solani pisi Serratia marcescens Pseudomonas aeruginosa

Translocation studies

(Becker et al., 2005)

Enzyme refolding

(Wilhelm et al., 2007)

Escherichia coli Pseudomonas aeruginosa Burkholderia cepacia Pseudomonas fluorescens

Translocation studies Whole cell biocatalysis

(Yang et al., 2004) (Yang et al., 2010)

Escherichia coli

Translocation studies

(Suzuki et al., 1995)

P. putida*

IcsA (VirG) *

S. flexneri

foldase, lipase-specific (lipH) ß-Lactamase (bla) lipase (PAL), foldase lipase (BCL), foldase lipase (PFL) alkaline phosphatase (phoA)

There is 100% amino acid identity between the EstA gene of the closely related species, Pseudomonas aeruginosa and Pseudomonas putida.

96

J. Jose et al. / Journal of Biotechnology 161 (2012) 92–103

Fig. 4. Passenger driven dimerization of SDH on the surface of E. coli by autodisplay. Due to the motility of the ␤ barrel within the plane of the outer membrane, passenger domains that have affinity to each other will form stable dimers or multimers. As in the example of SDH, where the crystal structure showed that the two subunits bind in reverse, the linker has be long and flexible enough to allow such conformations.

reasonable to investigate the efficiency of functional enzyme display with one or two model enzymes i.e. one of prokaryotic and one of eukaryotic origin. An example of an enzyme of eukaryotic origin that failed to be expressed on the surface of E. coli by autodisplay is human steroid 5␣-reductase type II (Panter et al., 2005). 5␣-reductase type II is a natural integral inner membrane protein with five transmembrane spanning ␣ helices. When its coding sequence was inserted at the passenger site of the autotransporter precursor gene and this construct was transferred into an E. coli host, no expression product was detectable. This indicates that the surface display of a natural inner membrane protein by the autodisplay strategy is not possible. It possibly interferes with the proteins inherent signalling or membrane targeting. Therefore autodisplay appears to be restricted to the surface display of soluble proteins or – at least – the soluble domains of membrane proteins, so long as they are not involved in membrane embedding, targeting or signalling. 5. Autodisplay for P450 enzyme whole cell biocatalysis The first two enzymes to be autodisplayed, ␤-lactamase (Lattemann et al., 2000) and esterase EstA (Schultheiss et al., 2002), provided the proof of principle, but the first biocatalytic application of autodisplay appeared with the surface display of Adx (Jose et al., 2002; Jose et al., 2001). Adx, a bovine iron-sulfur protein, delivers electrons to the mitochondrial type of P450 enzymes in order to maintain their activity (Ewen et al., 2011). These electrons are provided by adrenodoxin reductase (AdR). In addition to containing an inorganic prosthetic group, Adx is a functional dimer (Pikuleva et al., 2000). Adx was expressed as a monomeric passenger protein by autodisplay and subsequently Adx dimers were detectable at the cell surface. One of the most convenient features of autodisplay is the membrane anchoring domain, the ␤-barrel, which is not covalently linked to the cell envelope as it is in other display systems, but instead is motile within the plane of the outer membrane. If the displayed passenger domains have affinity to each other, this will lead to a passenger driven or self-driven dimerization or multimerization as it has been shown in addition for SDH (Jose and von Schwichow, 2004a) (Fig. 4), nitrilase (Detzel et al., 2011), prenyltransferase (Kranen et al., 2011), or the lacZ domain of protein A (Jose et al., 2009). By electron spin resonance experiments it was shown that when it reached the surface, the Adx dimer was devoid of the iron-sulfur group and hence inactive (Jose et al., 2001). By a simple titration step under mild but anaerobic conditions, the iron sulfur group could be incorporated in apo–Adx displayed at the E. coli surface and after the addition of AdR and P450 enzyme (either CYP11A1 or CYP11B1) under aerobic conditions a whole cell

biocatalyst for the efficient conversion of different steroids was obtained (Fig. 5a) (Jose et al., 2002). Three important conclusions could be drawn from this investigation. First, cells displaying a recombinant protein on E. coli by the use of a homologous autotransporter were robust and cell viability was not disturbed by incorporation of an anorganic prosthetic group under unaerobic conditions followed by long time use under aerobic conditions (Jose et al., 2001). Second, the outer membrane of E. coli provides sufficient membrane environment to the investigated membrane associated P450 enzymes in order to be enzymatically active. The activity assay for Adx involved the external addition of a P450 enzyme, whose activity is thought to depend on membrane surrounding. The whole cell biocatalyst obtained by the surface display of Adx and the addition of AdR and P450 enzyme exhibited activities in the same range as they were obtained with traditional reconstituted membrane approaches (Jose and Meyer, 2007). Therefore autodisplay overcame all the obstacles to CYP activity, and is a promising tool for accessing the synthetic potential of P450 monooxygenases, an interesting but difficult to handle class of enzymes. Finally, by setting up a calibration curve with purified Adx, the number of functional Adx dimers at the cell surface was determined as 1.8 × 105 molecules per single cell. To put this in perspective, the diameter of a ␤-barrel was calculated to be 1.1 nm (a crystal structure of the AIDA-I ␤-barrel is not yet available). After estimating the cell surface area of an average E. coli cell, the mean distance between two ␤-barrels was determined to be around 8.6 nm in each direction (Jose and Meyer, 2007). Although this is a rough estimation, it makes clear, that such a large number of molecules displayed per single cell is not unreasonable. In addition a similar number was later experimentally verified for other passengers such as SDH (Jose and von Schwichow, 2004b). The major consequence from the results obtained with Adx was to investigate the autodisplay of a P450 enzyme. Cytochrome P450 monooxygenases (P450s or CYPs) play essential roles in the biosynthesis of prostaglandins, steroids or secondary metabolites of plants and microorganisms, as well as in the detoxification of a wide range of foreign compounds as drugs or chemical pollutants (Bernhardt, 2006). Most P450s are membrane-bound or need a membrane environment to gain functional conformation. All require at least one redox partner protein in addition to be active. As mentioned above, the redox partner proteins for the mammalian mitochondrial P450 enzymes (Class I) are Adx and AdR. The purification of Adx and AdR is laborious, and may be a reason why the use of the synthetic potential of P450s in bio-transformations or organic synthesis is not very common (Urlacher and Girhard, 2011). Nevertheless, proof of principle in the autodisplay of a P450 was achieved using naturally soluble CYP106A2 from Bacillus megaterium (Schumacher et al., 2012). CYP106A2 accepts bovine Adx and AdR as redox partners, and is known to catalyze the 15␤-hydroxylation of various steroids or steroid like compounds such as the diterpene abietic acid, as well as the N-demethylation of the antidepressant imipramine (Bleif et al., 2011). For autodisplay the CYP106A2 encoding sequence was inserted as described above and after expression of the corresponding gene construct in E. coli BL21(DE3), sufficient amounts of protein were detectable at the cell surface (Schumacher et al., 2012). Enzyme assays included the cells displaying CYP106A2, the redox partners Adx and AdR, and NADPH. Unexpectedly, cells displaying CYP106A2 showed enzymatic activity towards the substrate deoxycorticosterone without the need to externally incorporate the heme prosthetic group. As described above, the translocation mechanism of autodisplay is thought to occur while the protein is in an unfolded form. The lack of activity of autodisplayed Adx without the external addition of the prosthetic group was consistent with this translocation model (Jose et al., 2002, 2001). Experiments aiming to add the prosthetic group from the exterior did no yield a higher

J. Jose et al. / Journal of Biotechnology 161 (2012) 92–103

97

Fig. 5. Whole biocatalyst for steroid synthesis obtained by autodisplay of Adx and the addition of AdR and CYP enzyme for steroid biosynthesis (A, Jose et al., 2002). The next step is to display Adx, AdR and CYP enzyme as CYP106A2 on the surface of a single cell in order to have a complete and stable P450 biocatalyst without the need of adding components from exterior (B).

enzymatic activity (Schumacher et al., 2012). In principle this could be explained in two ways. Either, the poryphrin was incorporated during transport and – as a consequence – CYP106A2 was translocated to the cell surface along with the heme prosthetic group in a folded form, which contradicts the current theories regarding translocation. Or alternatively, CYP106A2 was translocated as an apoprotein without the heme and the prosthetic group was incorporated after transport. In this case the porphyrin must have been present in the supernatant, either as a component of the growth medium or released by the cells themselves. Very recently, the TolC channel protein was identified to be responsible for the active transport of porphyrins into the supernatant of E. coli (Tatsumi and Wachi, 2008). Based on these findings autodisplay of CYP106A2 was analyzed in the TolC negative mutant JW55301-1 of E. coli (Baba et al., 2006). The amount of protein displayed at the surface of the TolC mutant remained unaltered in comparison to the TolC positive host background, however, enzymatic activity was substantially reduced. This was an indirect but strong indication that CYP106A2 was translocated to the cell surface without poryphyrin, most probably in an unfolded form, consistent with the mechanism recently supported by data obtained with natural autotransporter passengers (Ieva and Bernstein, 2009). Finally by adding heme as a salt

solution to cells of E. coli displaying the CYP106A2 apoprotein the enzymatic activity in the TolC mutant could be completely restored. For biotechnology purposes, the origin of the heme group is of less importance than the clear activity towards substrates which mirror those previously described for the purified CYP106A2 enzyme. In addition to the substrate deoxycorticosterone, the whole cell biocatalyst obtained by autodisplay of CYP106A2 also hydroxylated abietic acid and enabled N-demethylation of imipramin (Schumacher et al., 2012). The successful display of CYP106A2 as a P450 model enzyme was not an isolated case, and it is likely that surface display can be applied to other P450s of interest for chemical synthesis and toxicology. The strategy described above was subsequently applied to the human P450 enzyme CYP3A4, a membrane associated liver enzyme of the human first-pass metabolism (Kato, 2008). The enzyme was displayed at the cell surface and activity was tested in a 2-component system, requiring the external addition of NADPH–P450–reductase (Schumacher and Jose, 2012). Whole cells displaying the enzyme converted the substrate testosterone to yield 6␤-OH-deoxycorticosterone (Schumacher and Jose, 2012). We would like to clearly state that the rate of CYP expression and the quite low enzymatic activity of the CYP3A4 whole cell

98

J. Jose et al. / Journal of Biotechnology 161 (2012) 92–103

biocatalyst, which is at least partly owed to the very low turnover number of CYP enzymes in general and that of CYP3A4 in particular, needs improvement. Nevertheless these experiments open the door for a broader application of P450 whole cell biocatalysis for the conversion of substrates and products that are not able to pass membrane barriers. For purposes of chemical synthesis the native form of the enzyme does not need to be retained, and it is likely that the application of sophisticated enzyme improvement techniques can select enzyme variants with increased rates of turnover. The next step will be the co-expression of the other proteins required for P450 activity on a single cell of E. coli, (e.g. Adx and AdR for class I, and NADPH-P450-reductase and cytochrome b5 for class II) in order to obtain a self-assembling functioning electron delivery system on the cell surface (Fig. 5B). Cells of these types could subsequently be combined with cells displaying redox co-factor (NADP/NADPH) regenerating enzymes to enable a continued enzymatic conversion of the substrates. Finally after this configuration has been successfully assembled and shown to operate for synthesis purposes in the lab scale, conditions need to be identified and optimized for the scale up of this process and finally would lead to products in the gram scale.

6. Enzyme autodisplay and product preparation The first chemical compound prepared and purified in the subgram range using the autodisplay technology was achieved with a whole cell biocatalyst displaying nitrilase from Alcaligenes faecalis (Detzel et al., 2011). Nitrilases (EC 3.5.5.1) are enzymes that convert nitriles to the corresponding carboxylic acid and ammonia in a single step, a reaction of substantial industrial interest. The carboxylic acids produced are used as intermediates in a great variety of chemical production processes. In most cases they are enantiomerically pure and can be produced under mild conditions. However, nitrilases are known to be rather labile and immobilization and aggregation has been attempted in order to adapt these enzymes to biocatalytic purposes (Martinkova and Mylerova, 2003). Using the whole cell biocatalyst displaying nitrilase from Alcaligenes faecalis, 0.4 g of R-mandelic acid with an ee value >99% could be produced with 120 h of a 1 l batch culture. Nevertheless, it was about 22-fold less than a similar culture of A. faecalis cells could have produced in the same time (Kaul et al., 2007). This could have been due to a lower number of enzyme molecules displayed on the surface (75,000) in comparison to those which were expressed in A. faecalis (not determined). It could also reflect that the conditions within the cell are more convenient for this enzyme reaction than on the cell surface and e.g. would allow higher degrees of multimerization than on the cell surface. Nitrilase has been found to increase its enzymatic activity with increasing states multimerization and nonamers up to dodecamers have been reported (Yamamoto et al., 1992). Although the motile ␤-barrel allows a multimerization of passenger domains as seen with various examples, this multimerization will be hindered by the extension of the barrel itself and will be limited by the operating distance that is provided by the linker region. Therefore it is at least doubtful whether dodecamers indeed can be formed on the cell surface after autodisplay, and up to now such have not been experimentally detected yet. Surface display of nitrilase from A. faecalis nevertheless proved the evidence obtained before that autodisplay allows multimerization, and gave an excellent indication of the suitability of autodisplay for industrial processes. The whole cell biocatalyst displaying nitrilase could be stored for 180 days at −70 ◦ C without any significant loss in enzymatic activity. Cells were reused in subsequent cycles of R-mandelic acid production in batch culture and it turned out that within the last

cycle, 55% of the intial activity remained after 120 h of enzymatic conversion (Detzel et al., 2011). In these experiments, the cell sediment harvested from one cycle was put into the next production cycle of R-mandelic acid without normalization on the cell number. Therefore, this loss in activity also includes the loss in cell material during harvesting and transfer. Similar results were obtained with a whole cell biocatalyst displaying prenyltransferase from Aspergillus niger for the efficient prenylation of indole derivatives, which was shown to convert the substrates indole-3-propionic acid and L␤-homotryptophan (Kranen et al., 2011). It could be stored one month at 8 ◦ C without loss in activity and could also be reused in cyclic synthesis protocols. For nitrilase from A. faecalis the whole cell biocatalyst obtained by autodisplay showed a KM value for mandelonitrile (3.6 mM), which was in accordance with the published KM value for the free enzyme (5.75 mM) (Yamamoto et al., 1992) as well as an identical pH optimum of 7.5. It also converted phenylacetonitrile five times faster than mandelonitrile (9.3 mM in 16 h) indicating the same substrate specificity as the free enzyme (Yamamoto et al., 1992). This would imply that autodisplay of the nitrilase of A. faecalis does not affect its substrate specificity. Such unaltered substrate specificity was also found when nitrilase from Klebsiella aerogenes was displayed on the cell surface of E. coli (Detzel and Jose, unpublished). However, this is not a general rule, because it was observed before in the autodisplay of sorbitol dehydrogenase from R. sphaeroides that the preferences for different substrates, polyols and sugars, was altered in comparison to the free enzyme (Jose and von Schwichow, 2004a). An altered substrate preference was also found for esterase ApeE when displayed at the cell surface of E. coli by autodisplay (Schultheiss et al., 2008). In principle for autodisplay, the enzyme remains connected to the cell surface via its C-terminus by fusion to the linker domain. This fusion can limit the flexibility of the enzyme resulting in an altered activity or substrate specificity. There are also a small number of extra amino acids at the N-terminus which remain after cleavage of the signal peptide. And finally, one cannot discount the possibility that the environment the enzyme molecule is facing at the cell surface is different from that of the cell interior, resulting in an altered activity or substrate specificity. In summary it can be concluded that the activity of an autodisplayed enzyme does not need to be altered a priori in comparison to the free enzyme, but it appears to depend on the enzyme displayed and must be investigated on a case by case basis. At this point it appears worth discussing how the activity of whole cell biocatalysts displaying an enzyme can be measured and compared to pure enzymes, or other whole cell biocatalyst, in particular which dimension should be used. The activity in most purified enzyme preparations is given as U mg−1 protein, which means ␮mol substrate converted per min per mg of enzyme. Also in case kcat would be used instead, an estimation of the amount of protein would be required. Different as in the case of a purified enzyme, where all proteins are supposed to be enzyme molecules, the enzyme displayed at the cell surface makes up only a tiny proportion of the whole cell protein, to which it remains intrinsically tied to. Therefore, to determine the activity of a surface displayed enzyme and set it into relation of the whole cell protein appears to be not reasonable and will give a wrong impression about the enzyme’s real activity. An enzyme displayed at the cell surface is excluded from the vast majority of cell protein which is expressed intracellular and as a consequence, to set the activity at the cell surface into relation to the entire cell protein would rather cover the real circumstances than being useful information. Another possibility is to calculate the amount of substrate that is converted by a single cell of E. coli displaying the enzyme of interest, as it has been done for sorbitol dehydrogenase (Jose and von Schwichow, 2004a) and prenyltransferase (Kranen et al., 2011). This allows to compare the activities of different whole cell

J. Jose et al. / Journal of Biotechnology 161 (2012) 92–103

biocatalysts, which is in the range of pU per single cell of E. coli, but does not allow to compare these activities with those of the corresponding free enzyme. For better comparison we suggest to indicate the enzymatic activity of a whole cell biocatalyst in U (␮mol min−1 ) per volume of a bacterial suspension with a defined optical density. This can then be compared with a solution of the purified enzyme preparation. For the autodisplay of enzymes, whole cell activities of 0.1 until up to 50 mU ml−1 of a cell culture with an OD578 of 1 were observed for various whole cell biocatalyst displaying enzymes including SDH (Jose and von Schwichow, 2004a), esterase (Schultheiss et al., 2008), lipase (Detzel, Kranen, Jose unpublished) and CYP106A2 (Schumacher et al., 2012). Preliminary experiments to increase the OD values by a continuous culture system and subsequent substrate fermentation have been performed with the whole cell biocatalyst displaying nitrilase from A. faecalis. The optical density could indeed be increased but the overall enzymatic activity was not higher than that obtained in the batch culture (Detzel and Jose, unpublished). This means, that transferring the substrate conversion with whole cell biocatalysts displaying enzymes at the cell surface from the lab scale to production scale or even industrial scale is an investigation area of its own and needs further efforts. It can also indicate that the observed product inhibition for nitrilase from A. faecalis accounts for this observation (Yamamoto et al., 1992) and systematic process development will be required to overcome this limitation.

7. Autodisplay of challenging enzymes Human hyaluronidases are interesting pharmaceutical targets particularly to address cancer diseases (Kovar et al., 2006; Lokeshwar et al., 2006). Specific inhibitors of this group of enzymes could allow new therapeutic options to treat such diseases. Moreover, hyaluronidases could be used for the production of selected fragments of the biopolymer hyaluronic acid (HA) for research and cosmetic purposes. HA consists of ␤-1,3 linked d-glucuronic acid and N-acetyl-d-glucosamine disaccharide units. Disaccharide units are ␤-1,4 linked and joined up to 25,000 times, reaching a molecular mass up to 4 × 109 Da. HA is the main component of the extracellular matrix and belongs to the glycosaminoglycans, but in contrast to heparin or chondroitin it is not sulfated. Its concentration depends on the balance between synthesis via hyaluronate synthases and degradation via human hyaluronidases, mainly hyaluronidase 1 (hHyal-1), hyaluronidase 2 (hHyal-2) and PH–20 (hPH-20) (Stern, 2005; Stern et al., 2006). Until today access to these enzymes and hence to specific inhibitors is limited because human hyaluronidases form inclusion bodies (IBs) when expressed in E. coli and need to be purified and refolded, as observed in expression studies using hHyal-1 (Hofinger et al., 2007b) and bee venom hyaluronidase (Soldatova et al., 1998). Expression in eukaryotic cells is slow, expensive and inefficient in comparison to E. coli. In the best case, several days are needed to obtain only low amounts of enzyme for activity determinations (Bookbinder et al., 2006; Hofinger et al., 2007b; Soldatova et al., 1998). Human hyaluronidase PH20 was displayed on the surface of E. coli by autodisplay and considerable enzymatic activity could be measured with whole cells (Kaessler et al., 2011). Autodisplay of hPH20 yielded a simple, reproducible and reliable source for this interesting enzyme and the first three inhibitors could be identified. However, when the enzyme was expressed on the surface of E. coli hosts strains with standard lipopolysaccharide (LPS) like BL21or UT5600, only marginal or no enzymatic activity was detectable. Only when autodisplay of hPH20 was performed in E. coli host strain F470 (Schop et al., 2000; Vinogradov et al., 1999), which

99

possesses restricted LPS at the cell surface, sufficient enzymatic activity was obtained. This was not surprising and could have been due to a competitive inhibition of hPH20, because the substrate of hPH20, HA and LPS share sufficient structural similarity. A similar competitive inhibition of LPS was observed with the surface display of sorbitol dehydrogenase (Jose and von Schwichow, 2004a). We cannot at this point exclude the possibility that LPS is a general problem when sugar modifying enzymes are displayed at the cell surface of E. coli. For hPH20 the enzymatic activity was significantly reduced in comparison to the free enzyme produced in eukaryotic cells. A similar reduction was found with surface display of nitrilase from A. faecalis in comparison to intracellular prokaryotic expression. But unlike the nitrilase, which is of bacterial origin, the human hPH20 is predicted to be glycosylated, and the reduced activity could have been due to the lack of eukaryotic-like glycosylation machinery in the E. coli host cell. It was demonstrated by the example of human hyaluronidase Hyal-1 that incubation with N-glycosidase (PNGase F) resulted in a reduction of enzyme activity to 60% (Hofinger et al., 2007a). Three N-glycosylation sites within hHyal-1 were identified and proposed to support correct protein folding (Chao et al., 2007). Within the hPH-20 amino acid sequence four possible glycosylation sites were found at positions 31, 96, 260 and 369 and it is supposed that like hHyal–1, the lack of glycosylation of hPH20 in E. coli is a reason for the lower enzyme activity. Despite some progress in understanding bacterial protein glycosylation (Benz and Schmidt, 2002; Nothaft and Szymanski, 2010), a major drawback of all mechanisms of expression in E. coli is the lack of mammalian-like protein glycosylation, and autodisplay is no exception. Nevertheless the experience with human hyaluronidase hPH20 shows that autodisplay is a viable alternative expression system for challenging enzymes, especially where intracellular expression results in inclusion bodies. Whole cells displaying hyaluronidases can be used to test for novel inhibitors, but also for the production of hyaluronic acids of defined chain length (Fig. 6), necessary for research of physiological function or for use in cosmetic or therapeutic preparations (Bogdan Allemann and Baumann, 2008; Burdick and Prestwich, 2011), currently restricted to crude hyaluronic acid preparations with a broad variation in chain length.

8. Autodisplay of recombinant enzymes by other autotransporters Beside AIDA-I, surface display of enzymes has been tested with only a few other autotransporters, none of which are of E. coli origin (Table 1). EstA, an esterase from Pseudomas spp., was the most often used autotransporter for such purpose. Autodisplay has been performed in several cases with the EstA of Pseudomonas aeruginosa origin (Becker et al., 2008, 2007, 2005; Wilhelm et al., 2007, 1999), and performed once with the EstA from Pseudomanas putida (Yang et al., 2004), but both EstA proteins share the identical amino acid sequence and can be considered as being the same autotransporter. EstA has been used for the surface display of lipases (Becker et al., 2005), a foldase (Wilhelm et al., 2007), and ␤-lactamase (Yang et al., 2004). The aim of most of these studies was to demonstrate the use of this autotransporter for surface display of recombinant proteins and also to investigate the translocation and the folding of the passenger protein. It was also used as a platform for clever screening approaches, either to identify cells displaying esterases with catalytic activity (Becker et al., 2007, 2004, 2005) or to identify variants with improved enantioselectivity (Becker et al., 2008). Therefore EstA can be considered alongside AIDA-I as the most commonly used autotransporter protein for biotechnological applications (Wilhelm et al., 2011). Also Yang et al. (2004) reported

100

J. Jose et al. / Journal of Biotechnology 161 (2012) 92–103

Fig. 6. HPLC monitoring of the production of hyaluronic acid (HA) with defined chain length by hyaluronidase. Crude HA extracts are incubated with hyaluronidase and samples were taken at 24 h and 48 h and analyzed by HPLC using a NH2 -column with a linear gradient of 16 mM to 800 mM NaH2 PO4 as solute. The smallest fragment that could be identified was a tetrasaccharide at a retention time of 10 min. The elution was stopped at 33 min with the appearance of a 22mer. Whether the fragment appearing at around 3 min was a monomer or a dimer was not possible to determine. The crude extract contained a 11mer (retention time around 18 min), which was not degradable.

on whole cell biocatalysis with different lipases displayed at the cell surface of E. coli by the aid of EstA. A very early example concerning autodisplay was a report of an alkaline phosphatase displayed on the E. coli cell surface by the aid of VirG autotransporter from Shigella flexneri, but its enzymatic activity remained obscure (Suzuki et al., 1995). 9. Perspectives The examples of autodisplay based whole cell biocatalysts presented here are promising, but have only been tested in the lab scale. Preliminary experiments with nitrilase displaying cells in a 2 l bioreactor were successsful, however, more systematic studies with whole cell biocatalysts displaying enzymes will be needed to identify conditions for up-scaling and for process development in order to bring the applications from the lab scale to scales of industrial interest. For whole cell biocatalysts displaying P450 enzymes, it was shown by several examples (CYP11A1, CYP11B1, CYP3A4) that the surface of E. coli provides sufficient membrane environment to these enzymes in order to be active and in the case of CYP3A4 and CYP106A2 it could be demonstrated that the heme group is provided by the E. coli host, most probably via TolC, and incorporated at the cell surface into the P450 apoprotein (Schumacher et al., 2012; Schumacher and Jose, 2012). This makes the autodisplay of P450 enzymes a convenient system to access the synthetic potential of these enzymes and it provides a new expression platform for these and other difficult to handle biocatalysts. The next steps will be the co-expression of those proteins that are needed to deliver the electrons to the P450 enzyme, such as Adx and AdR for the 3-component system, and CYP P450 reductase for the 2component system (Bernhardt, 2006; Urlacher and Girhard, 2011). Functional surface display of Adx has already been reported (Jose et al., 2002, 2001). It is suggested to be co-expressed with the

mitochondrial type of P450 enzyme and AdR on the surface of a single cell (Fig. 5b). AdR is an FAD containing enzyme and CYP P450 reductase to be co-expressed with type II P450s contains FMN in addition to FAD. It has already been shown that flavin containing enzymes can be expressed in an active form at the cell surface by autodisplay (Kranen and Jose, unpublished). As with the heme group of P450 enzymes, the flavin component was delivered by the host cell, most probably released via Tolc and incorporated into apoprotein displayed at the surface. In case it will be possible to create cells displaying different functional P450s including co-factors and partner proteins, modular synthetic systems are technically feasible combining different enzymatic activities. For example, cells displaying CYP3A4, cells displaying CYP2D6 and cells displaying CYP2C9, the major enzymes of human first pass metabolism in the liver could be used in concert in order to simulate human metabolism of drugs or drug like compounds. Moreover they could be used to prepare metabolic intermediates of new drugs to be provided to analytical purposes, i.e. as reference compounds. It would be most suitable to combine these modules with surface displayed P450 enzymes with another whole cell biocatalyst displaying another enzyme which is able to regenerate the redox equivalents NADP or NADPH in order to assure continues reactions. Another interesting approach, which is not restricted to P450 enzymes, is to use whole cell biocatalyst with autodisplayed enzymes as modules in a sequential row for synthesis (Fig. 7). One could imagine a start with an easy accessible compound, perhaps a plant ingredient or waste material, whether aromatic, heterocyclic, cyclic or non-cyclic, and lead it along a series of whole cell biocatalysts in order to produce a new valuable or bioactive compound (i.e. a drug). In a similar fashion, a series of whole cell biocatalysts could be used in bioremediation, to degrade persistent pollutants or toxins (Li et al., 2008; Scott et al., 2009). Many pollutants (e.g. isomers of hexachlorocyclohexane) are extremely resistant to degradation,

Fig. 7. Modular system of whole cell biocatalysts with autodisplayed enzymes for the synthesis of drugs or building blocks.

J. Jose et al. / Journal of Biotechnology 161 (2012) 92–103

requiring several steps before detoxification is achieved, and in some cases, multiple enzymes (Lal et al., 2010). Most of the whole cell biocatalysts obtained by the autodisplay of enzymes were using E. coli as a host organism. The advantages of E. coli as a host include that genetic engineering protocols and tools are at hand, detailed knowledge on physiology and protein function, and available expertise in industrial fermentation. But the expression of recombinant proteins in E. coli as a host organism also has disadvantages. E. coli contains LPS and the contamination of products, of peptides or proteins used for therapeutic or pharmaceutical purposes is inacceptable because it causes an impetuous immune response (Freudenberg et al., 2008). Although an increased rate of cell lysis in the case of autodisplay in E. coli using homologous autotransporters has not yet been observed (Kranen et al., 2011; Schumacher and Jose, 2012), it is unrealistic to expect an E. coli supernatant to remain completely free of LPS. In addition E. coli does not grow to such high densities as it has been reported for other gram negative bacteria and it appears to be not as robust as natural soil bacteria in remediation experiments in soil or other mechanically crude processes. Therefore it could be beneficial to establish autodisplay in other gram negative bacteria more suitable for such purposes, e.g. Rhodobacter capsulatus (Katzke et al., 2010) or Ralstonia eutropha (Valls et al., 2000). However, E. coli autotransporter constructs used for autodisplay in these organisms may encounter incompatibility problems in the host organism and it remains to be seen whether this is as efficient as in E. coli. An alternative would be to isolate and test naturally occurring autotransporters in these organisms, which could be used as a transport vehicle for recombinant enzymes. Autodisplay has the advantage, that neither substrate nor product need to cross a membrane barrier. This makes it most suitable for screening purposes in order to obtain taylor-made enzymes for pre-given reactions (Becker et al., 2008, 2007; Gratz and Jose, 2008, 2011). Molecular biology tools are used to create variations in the sequence coding the enzyme, and single cells can be selected by the enzyme variant displayed at the cell surface which will replicate and finally load themselves with the new enzyme identified. Autodisplay is also compatible with some high throughput screening approaches, for example agar plate clearing assays or fluorescence activated cell sorting (FACS). So long as the cells can be labelled based on their activity, FACS in particular will allow single cell high throughput screening (Becker et al., 2008; Jose et al., 2005). Such approach towards enzymes with new synthetic features could start with the surface display of a recombinant enzyme that is subsequently varied by random in order to get libraries of bacteria which bear a single variant in high numbers at the cell surface. It would also be possible to start with a natural enzyme passenger of an autotransporter as the template to introduce random variation in order to identify improve enzyme variant (Becker et al., 2008, 2005). For the latter case it would be interesting to start a more systematic approach in order to identify autotransporter proteins with enzymes as passengers in soil bacteria or harmless gram negative commensals. At least 53 putative autotransporters from gram negative commensals have already been identified (Wells et al., 2010). Autodisplay could also allow screening for novel catalytic activity, where improved variants of the enzyme confer a growth advantage on the organism. This can be achieved by nutrient limitation, where the product of the catalysis is the sole source of the particular nutrient (e.g. carbon, nitrogen or phosphate). Alternatively, the substrate of the reaction can be toxic to the cells (e.g. an antibiotic). The autodisplay of antibiotic-degrading enzymes such as Bla has been shown to convey resistance to ampicillin (Lattemann et al., 2000). Cells displaying enzymes with high turnover efficiency would have a selective advantage over cells expressing enzymes with poor efficiency. This “growth selection”

101

approach has been successfully used in combination with plate assays and the yeast model (Luthi et al., 2003; Murai et al., 1997). With autodisplay, the location of the enzyme on the cell envelope might also confer an advantage to the organism in liquid cultures. This allows the system to be self-selective, as the cells containing the enzyme variant with the most improved catalysis will grow to dominate the culture, reducing the number of variants which need to be investigated. Finally, cells with autodisplayed redox enzymes could be used to create novel microbial fuel cells (Fishilevich et al., 2009). Low value compounds such as waste products could be used as substrate and the electrons produced can be bled off by carbon electrodes in order to produce current. Similar systems could also be applied for analytical purposes such as a biosensor in which the current produced is a function of the concentration of substrate analyte. In conclusion, what can be learned from the past experience is that autodisplay can achieve much more than initially realized. Therefore it seems worthwhile to take the next step towards a large scale application of whole cell biocatalysts with autodisplayed enzymes and to adapt it to industrial needs. Acknowledgements The authors would like to thank all colleagues, former and current co-workers and collaboration partners for their contributions in the autodisplay of enzymes. We would like to apologize if not all was considered within the narrow focus of this review. References Baba, T., Ara, T., Hasegawa, M., Takai, Y., Okumura, Y., Baba, M., Datsenko, K.A., Tomita, M., Wanner, B.L., Mori, H., 2006. Construction of Escherichia coli K-12 in-frame, single-gene knockout mutants: the Keio collection. Molecular Systems Biology 2, 2006–2008. Becker, S., Hobenreich, H., Vogel, A., Knorr, J., Wilhelm, S., Rosenau, F., Jaeger, K.E., Reetz, M.T., Kolmar, H., 2008. Single-cell high-throughput screening to identify enantioselective hydrolytic enzymes. Angewandte Chemie International Edition 47, 5085–5088. Becker, S., Michalczyk, A., Wilhelm, S., Jaeger, K.E., Kolmar, H., 2007. Ultrahighthroughput screening to identify E. coli cells expressing functionally active enzymes on their surface. Chembiochem 8, 943–949. Becker, S., Schmoldt, H.U., Adams, T.M., Wilhelm, S., Kolmar, H., 2004. Ultra-highthroughput screening based on cell-surface display and fluorescence-activated cell sorting for the identification of novel biocatalysts. Current Opinion in Biotechnology 15, 323–329. Becker, S., Theile, S., Heppeler, N., Michalczyk, A., Wentzel, A., Wilhelm, S., Jaeger, K.E., Kolmar, H., 2005. A generic system for the Escherichia coli cell-surface display of lipolytic enzymes. FEBS Letters 579, 1177–1182. Benz, I., Schmidt, M.A., 1989. Cloning and expression of an adhesin (AIDA-I) involved in diffuse adherence of enteropathogenic Escherichia coli. Infection and Immunity 57, 1506–1511. Benz, I., Schmidt, M.A., 1992a. AIDA-I, the adhesin involved in diffuse adherence of the diarrhoeagenic Escherichia coli strain 2787 (O126:H27), is synthesized via a precursor molecule. Molecular Microbiology 6, 1539–1546. Benz, I., Schmidt, M.A., 1992b. Isolation and serologic characterization of AIDA-I, the adhesin mediating the diffuse adherence phenotype of the diarrhea-associated Escherichia coli strain 2787 (O126:H27). Infection and Immunity 60, 13–18. Benz, I., Schmidt, M.A., 2002. Never say never again: protein glycosylation in pathogenic bacteria. Molecular Microbiology 45, 267–276. Benz, I., Schmidt, M.A., 2011. Structures and functions of autotransporter proteins in microbial pathogens. International Journal of Medical Microbiology 301, 461–468. Bernhardt, R., 2006. Cytochromes P450 as versatile biocatalysts. Journal of Biotechnology 124, 128–145. Binder, U., Matschiner, G., Theobald, I., Skerra, A., 2010. High-throughput sorting of an anticalin library via EspP-mediated functional display on the Escherichia coli cell surface. Journal of Molecular Biology 400, 783–802. Bleif, S., Hannemann, F., Lisurek, M., von Kries, J.P., Zapp, J., Dietzen, M., Antes, I., Bernhardt, R., 2011. Identification of CYP106A2 as a regioselective allylic bacterial diterpene hydroxylase. Chembiochem. Bogdan Allemann, I., Baumann, L., 2008. Hyaluronic acid gel (Juvederm) preparations in the treatment of facial wrinkles and folds. Clinical Interventions in Aging 3, 629–634. Bookbinder, L.H., Hofer, A., Haller, M.F., Zepeda, M.L., Keller, G.A., Lim, J.E., Edgington, T.S., Shepard, H.M., Patton, J.S., Frost, G.I., 2006. A recombinant human enzyme for enhanced interstitial transport of therapeutics. Journal of Controlled Release 114, 230–241.

102

J. Jose et al. / Journal of Biotechnology 161 (2012) 92–103

Brunder, W., Schmidt, H., Karch, H., 1997. EspP, a novel extracellular serine protease of enterohaemorrhagic Escherichia coli O157:H7 cleaves human coagulation factor V. Molecular Microbiology 24, 767–778. Burdick, J.A., Prestwich, G.D., 2011. Hyaluronic acid hydrogels for biomedical applications. Advanced Materials 23, H41–H56. Busch, R., Hirth, T., Liese, A., Nordhoff, S., Puls, J., Pulz, O., Sell, D., Syldatk, C., Ulber, R., 2006. The utilization of renewable resources in German industrial production. Biotechnology Journal 1, 770–776. Chao, K.L., Muthukumar, L., Herzberg, O., 2007. Structure of human hyaluronidase-1, a hyaluronan hydrolyzing enzyme involved in tumor growth and angiogenesis. Biochemistry 46, 6911–6920. Choi, W.J., 2009. Biotechnological production of enantiopure epoxides by enzymatic kinetic resolution. Applied Microbiology and Biotechnology 84, 239–247. Detzel, C., Maas, R., Jose, J., 2011. Autodisplay of nitrilase from Alcaligenes faecalis in E. coli yields a whole cell biocatalyst for the synthesis of enantiomerically pure R-mandelic acid. ChemCatChem 3, 719–725. Elish, M.E., Pierce, J.R., Earhart, C.F., 1988. Biochemical analysis of spontaneous fepA mutants of Escherichia coli. Journal of General Microbiology 134, 1355–1364. Ewen, K.M., Kleser, M., Bernhardt, R., 2011. Adrenodoxin: the archetype of vertebrate-type [2Fe-2S] cluster ferredoxins. Biochimica et Biophysica Acta 1814, 111–125. Fischer, T., Pietruszka, J., 2010. Key building blocks via enzyme-mediated synthesis. Topics in Current Chemistry 297, 1–43. Fishilevich, S., Amir, L., Fridman, Y., Aharoni, A., Alfonta, L., 2009. Surface display of redox enzymes in microbial fuel cells. Journal of the American Chemical Society 131, 12052–12053. Freudenberg, M.A., Tchaptchet, S., Keck, S., Fejer, G., Huber, M., Schutze, N., Beutler, B., Galanos, C., 2008. Lipopolysaccharide sensing an important factor in the innate immune response to Gram-negative bacterial infections: benefits and hazards of LPS hypersensitivity. Immunobiology 213, 193–203. Goldberg, K., Schroer, K., Lutz, S., Liese, A., 2007a. Biocatalytic ketone reduction—a powerful tool for the production of chiral alcohols—part I: processes with isolated enzymes. Applied Microbiology and Biotechnology 76, 237–248. Goldberg, K., Schroer, K., Lutz, S., Liese, A., 2007b. Biocatalytic ketone reduction—a powerful tool for the production of chiral alcohols-part II: whole-cell reductions. Applied Microbiology and Biotechnology 76, 249–255. Gratz, A., Jose, J., 2008. Protein domain library generation by overlap extension (PDLGO): a tool for enzyme engineering. Analytical Biochemistry 378, 171–176. Gratz, A., Jose, J., 2011. Focusing mutations within random libraries to distinct areas: protein domain library generation by overlap extension. Methods in Molecular Biology 729, 153–166. Halter, R., Pohlner, J., Meyer, T.F., 1984. IgA protease of Neisseria gonorrhoeae: isolation and characterization of the gene and its extracellular product. Embo Journal 3, 1595–1601. Henderson, I.R., Navarro-Garcia, F., Desvaux, M., Fernandez, R.C., Ala’Aldeen, D., 2004. Type V protein secretion pathway: the autotransporter story. Microbiology and Molecular Biology Reviews 68, 692–744. Hofinger, E.S., Bernhardt, G., Buschauer, A., 2007a. Kinetics of Hyal-1 and PH-20 hyaluronidases: comparison of minimal substrates and analysis of the transglycosylation reaction. Glycobiology 17, 963–971. Hofinger, E.S., Spickenreither, M., Oschmann, J., Bernhardt, G., Rudolph, R., Buschauer, A., 2007b. Recombinant human hyaluronidase Hyal-1: insect cells versus Escherichia coli as expression system and identification of low molecular weight inhibitors. Glycobiology 17, 444–453. Ieva, R., Bernstein, H.D., 2009. Interaction of an autotransporter passenger domain with BamA during its translocation across the bacterial outer membrane. Proceedings of the National Academy of Sciences of the United States of America 106, 19120–19125. Ieva, R., Skillman, K.M., Bernstein, H.D., 2008. Incorporation of a polypeptide segment into the beta-domain pore during the assembly of a bacterial autotransporter. Molecular Microbiology 67, 188–201. Jose, J., Bernhardt, R., Hannemann, F., 2002. Cellular surface display of dimeric Adx and whole cell P450-mediated steroid synthesis on E. coli. Journal of Biotechnology 95, 257–268. Jose, J., Betscheider, D., Zangen, D., 2005. Bacterial surface display library screening by target enzyme labeling: Identification of new human cathepsin G inhibitors. Analytical Biochemistry 346, 258–267. Jose, J., Chung, J.W., Jeon, B.J., Maas, R.M., Nam, C.H., Pyun, J.C., 2009. Escherichia coli with autodisplayed Z-domain of protein A for signal amplification of SPR biosensor. Biosensors and Bioelectronics 24, 1324–1329. Jose, J., Handel, S., 2003. Monitoring the surface display of recombinant proteins by cysteine labeling and flow cytometry. ChemBioChem 4, 396–405. Jose, J., Hannemann, F., Bernhardt, R., 2001. Functional display of active bovine adrenodoxin on the surface of E. coli by chemical incorporation of the [2Fe-2S] cluster. ChemBioChem 2, 695–701. Jose, J., Jahnig, F., Meyer, T.F., 1995. Common structural features of IgA1 proteaselike outer membrane protein autotransporters. Molecular Microbiology 18, 378–380. Jose, J., Kramer, J., Klauser, T., Pohlner, J., Meyer, T.F., 1996. Absence of periplasmic DsbA oxidoreductase facilitates export of cysteine-containing passenger proteins to the Escherichia coli cell surface via the Iga beta autotransporter pathway. Gene 178, 107–110. Jose, J., Meyer, T.F., 2007. The autodisplay story, from discovery to biotechnical and biomedical applications. Microbiology and Molecular Biology Reviews 71, 600–619.

Jose, J., von Schwichow, S., 2004a. Autodisplay of active sorbitol dehydrogenase (SDH) yields a whole cell biocatalyst for the synthesis of rare sugars. ChemBioChem 5, 100–108. Jose, J., von Schwichow, S., 2004b. Cystope tagging for labelling and detection of recombinant protein expression. Analytical Biochemistry 331, 267–274. Jose, J., Zangen, D., 2005. Autodisplay of the protease inhibitor aprotinin in Escherichia coli. Biochemical and Biophysical Research Communications 333, 1218–1226. Kaessler, A., Olgen, S., Jose, J., 2011. Autodisplay of catalytically active human hyaluronidase hPH-20 and testing of enzyme inhibitors. European Journal of Pharmaceutical Sciences 42, 138–147. Kajava, A.V., Steven, A.C., 2006. The turn of the screw: variations of the abundant beta-solenoid motif in passenger domains of Type V secretory proteins. Journal of Structural Biology 155, 306–315. Kato, M., 2008. Intestinal first-pass metabolism of CYP3A4 substrates. Drug Metabolism and Pharmacokinetics 23, 87–94. Katzke, N., Arvani, S., Bergmann, R., Circolone, F., Markert, A., Svensson, V., Jaeger, K.E., Heck, A., Drepper, T., 2010. A novel T7 RNA polymerase dependent expression system for high-level protein production in the phototrophic bacterium Rhodobacter capsulatus. Protein Expression and Purification 69, 137–146. Kaul, P., Stolz, A., Banerjee, U.C., 2007. Cross-linked amornitrilase aggregates for enantioselective nitrile phous Advanced Synthesis and Catalysis 349, 2167– hydrolysis. 2176. Klauser, T., Pohlner, J., Meyer, T.F., 1990. Extracellular transport of cholera toxin B subunit using Neisseria IgA protease beta-domain: conformation-dependent outer membrane translocation. Embo Journal 9, 1991–1999. Klauser, T., Pohlner, J., Meyer, T.F., 1992. Selective extracellular release of cholera toxin B subunit by Escherichia coli: dissection of Neisseria Iga beta-mediated outer membrane transport. Embo Journal 11, 2327–2335. Knowles, T.J., Scott-Tucker, A., Overduin, M., Henderson, I.R., 2009. Membrane protein architects: the role of the BAM complex in outer membrane protein assembly. Nature Reviews Microbiology 7, 206–214. Kovar, J.L., Johnson, M.A., Volcheck, W.M., Chen, J., Simpson, M.A., 2006. Hyaluronidase expression induces prostate tumor metastasis in an orthotopic mouse model. American Journal of Pathology 169, 1415–1426. Kranen, E., Steffan, N., Maas, R., Li, S., Jose, J., 2011. Development of a whole cell biocatalyst for the efficient prenylation of indole derivatives by Autodisplay of the aromatic prenyltransferase FgaPT2. ChemCatChem 3, 1200–1207. Kronqvist, N., Malm, M., Rockberg, J., Hjelm, B., Uhlen, M., Stahl, S., Lofblom, J., 2010. Staphylococcal surface display in combinatorial protein engineering and epitope mapping of antibodies. Recent Patents on Biotechnology 4, 171–182. Kuroda, K., Ueda, M., 2011. Cell surface engineering of yeast for applications in white biotechnology. Biotechnology Letters 33, 1–9. Lal, R., Pandey, G., Sharma, P., Kumari, K., Malhotra, S., Pandey, R., Raina, V., Kohler, H.P.E., Holliger, C., Jackson, C., Oakeshott, J.G., 2010. Biochemistry of microbial degradation of hexachlorocyclohexane and prospects for bioremediation. Microbiology and Molecular Biology Reviews 74, 58–80. Lattemann, C.T., Maurer, J., Gerland, E., Meyer, T.F., 2000. Autodisplay: functional display of active beta-lactamase on the surface of Escherichia coli by the AIDA-I autotransporter. Journal of Bacteriology 182, 3726–3733. Li, C., Zhu, Y., Benz, I., Schmidt, M.A., Chen, W., Mulchandani, A., Qiao, C., 2008. Presentation of functional organophosphorus hydrolase fusions on the surface of Escherichia coli by the AIDA-I autotransporter pathway. Biotechnology and Bioengineering 99, 485–490. Linke, D., Riess, T., Autenrieth, I.B., Lupas, A., Kempf, V.A., 2006. Trimeric autotransporter adhesins: variable structure, common function. Trends in Microbiology 14, 264–270. Lokeshwar, V.B., Estrella, V., Lopez, L., Kramer, M., Gomez, P., Soloway, M.S., Lokeshwar, B.L., 2006. HYAL1-v1, an alternatively spliced variant of HYAL1 hyaluronidase: a negative regulator of bladder cancer. Cancer Research 66, 11219–11227. Loveless, B.J., Saier Jr., M.H., 1997. A novel family of channel-forming, autotransporting, bacterial virulence factors. Molecular Membrane Biology 14, 113–123. Luetz, S., Giver, L., Lalonde, J., 2008. Engineered enzymes for chemical production. Biotechnology and Bioengineering 101, 647–653. Luthi, U., Schaerer-Brodbeck, C., Tanner, S., Middendorp, O., Edler, K., Barberis, A., 2003. Human beta-secretase activity in yeast detected by a novel cellular growth selection system. Biochimica et Biophysica Acta 1620, 167–178. Martinkova, L., Mylerova, V., 2003. Synthetic applications of nitrile-converting enyzmes. Current Organic Chemistry 7, 1279–1295. Maurer, J., Jose, J., Meyer, T.F., 1997. Autodisplay: one-component system for efficient surface display and release of soluble recombinant proteins from Escherichia coli. Journal of Bacteriology 179, 794–804. Murai, T., Ueda, M., Yamamura, M., Atomi, H., Shibasaki, Y., Kamasawa, N., Osumi, M., Amachi, T., Tanaka, A., 1997. Construction of a starch-utilizing yeast by cell surface engineering. Applied and Environmental Microbiology 63, 1362–1366. Netzer, W.J., Hartl, F.U., 1997. Recombination of protein domains facilitated by cotranslational folding in eukaryotes. Nature 388, 343–349. Nishimura, K., Tajima, N., Yoon, Y.H., Park, S.Y., Tame, J.R., 2010. Autotransporter passenger proteins: virulence factors with common structural themes. J Mol Med (Berl) 88, 451–458.

J. Jose et al. / Journal of Biotechnology 161 (2012) 92–103 Nothaft, H., Szymanski, C.M., 2010. Protein glycosylation in bacteria: sweeter than ever. Nature Reviews Microbiology 8, 765–778. Panter, B.U., Jose, J., Hartmann, R.W., 2005. 5alpha-reductase in human embryonic kidney cell line HEK293: evidence for type II enzyme expression and activity. Molecular and Cellular Biochemistry 270, 201–208. Peterson, J.H., Tian, P., Ieva, R., Dautin, N., Bernstein, H.D., 2010. Secretion of a bacterial virulence factor is driven by the folding of a C-terminal segment. Proceedings of the National Academy of Sciences of United States of America 107, 17739–17744. Pikuleva, I.A., Tesh, K., Waterman, M.R., Kim, Y., 2000. The tertiary structure of fulllength bovine adrenodoxin suggests functional dimers. Archives of Biochemistry and Biophysics 373, 44–55. Pohlner, J., Halter, R., Beyreuther, K., Meyer, T.F., 1987. Gene structure and extracellular secretion of Neisseria gonorrhoeae IgA protease. Nature 325, 458– 462. Pohlner, J., Klauser, T., Kuttler, E., Halter, R., 1992. Sequence-specific cleavage of protein fusions using a recombinant Neisseria type 2 IgA protease. Biotechnology (NY) 10, 799–804. Ran, N., Rui, E., Liu, J., Tao, J., 2009. Chemoenzymatic synthesis of small molecule human therapeutics. Current Pharmaceutical Design 15, 134–152. Renn, J.P., Clark, P.L., 2008. A conserved stable core structure in the passenger domain beta-helix of autotransporter virulence proteins. Biopolymers 89, 420– 427. Robert, V., Volokhina, E.B., Senf, F., Bos, M.P., Van Gelder, P., Tommassen, J., 2006. Assembly factor Omp85 recognizes its outer membrane protein substrates by a species-specific C-terminal motif. PLoS Biology 4, e377. Schlapschy, M., Skerra, A., 2011. Periplasmic chaperones used to enhance functional secretion of proteins in E. coli. Methods in Molecular Biology 705, 211– 224. Schop, H., Wiese, M., Cordes, H.P., Seydel, J.K., 2000. Partial resistance of E. coli mutants against 2,4-diamino-5-benzylpyrimidines by interactions with bacterial membrane lipopolysaccharides. Derivation of quantitative structurebinding relationships. European Journal of Medicinal Chemistry 35, 619–634. Schultheiss, E., Paar, C., Schwab, H., Jose, J., 2002. Functional esterase surface display by the autotransporter pathway in Escherichia coli. Journal of Molecular Catalysis B Enzymatic 18, 89–97. Schultheiss, E., Weiss, S., Winterer, E., Maas, R., Heinzle, E., Jose, J., 2008. Esterase autodisplay: enzyme engineering and whole-cell activity determination in microplates with pH sensors. Applied and Environmental Microbiology 74, 4782–4791. Schumacher, S.D., Hannemann, F., Teese, M.G., Bernhardt, R., Jose, J., 2012. Autodisplay of functional CYP106A2 in Escherichia coli. Journal of Biotechnology, http://dx.doi.org/10.1016/j.jbiotec.2012.02.018. Schumacher, S.D., Jose, J., 2012. Expression of active human P450 3A4 on the cell surface of Escherichia coli by Autodisplay. Journal of Biotechnology, http://dx.doi.org/10.1016/j.jbiotec.2012.01.031. Scott, C., Pandey, G., Hartley, C.J., Jackson, C.J., Cheesman, M.J., Taylor, M.C., Pandey, R., Khurana, J.L., Teese, M.G., Coppin, C.W., Weir, K.M., Jain, R.K., Lal, R., Russell, R.J., Oakeshott, J.G., 2009. The enzymatic basis for pesticide bioremediation. Indian Journal of Microbiology 48, 65–79. Soldatova, L.N., Crameri, R., Gmachl, M., Kemeny, D.M., Schmidt, M., Weber, M., Mueller, U.R., 1998. Superior biologic activity of the recombinant bee venom allergen hyaluronidase expressed in baculovirus-infected insect cells as compared with Escherichia coli. Journal of Allergy and Clinical Immunology 101, 691–698.

103

St Geme, J.W., 3rd, Yeo, H.J., 2009. A prototype two-partner secretion pathway: the Haemophilus influenzae HMW1 and HMW2 adhesin systems. Trends in Microbiology 17, 355–360. Stern, R., 2005. Hyaluronan metabolism: a major paradox in cancer biology. Pathology Biology (Paris) 53, 372–382. Stern, R., Asari, A.A., Sugahara, K.N., 2006. Hyaluronan fragments: an informationrich system. European Journal of Cellular Biology 85, 699–715. Suhr, M., Benz, I., Schmidt, M.A., 1996. Processing of the AIDA-I precursor: removal of AIDAc and evidence for the outer membrane anchoring as a beta-barrel structure. Molecular Microbiology 22, 31–42. Suzuki, T., Lett, M.C., Sasakawa, C., 1995. Extracellular transport of VirG protein in Shigella. Journal of Biological Chemistry 270, 30874–30880. Tatsumi, R., Wachi, M., 2008. TolC-dependent exclusion of porphyrins in Escherichia coli. Journal of Bacteriology 190, 6228–6233. Tommassen, J., 2007. Biochemistry. Getting into and through the outer membrane. Science 317, 903–904. Tufvesson, P., Lima-Ramos, J., Jensen, J.S., Al-Haque, N., Neto, W., Woodley, J.M., 2011. Process considerations for the asymmetric synthesis of chiral amines using transaminases. Biotechnology and Bioengineering 108, 1479–1493. Urlacher, V.B., Girhard, M., 2011. Cytochrome P450 monooxygenases: an update on perspectives for synthetic application. Trends in Biotechnol. Valls, M., Atrian, S., de Lorenzo, V., Fernandez, L.A., 2000. Engineering a mouse metallothionein on the cell surface of Ralstonia eutropha CH34 for immobilization of heavy metals in soil. Nature Biotechnology 18, 661–665. van Bloois, E., Winter, R.T., Kolmar, H., Fraaije, M.W., 2011. Decorating microbes: surface display of proteins on Escherichia coli. Trends in Biotechnology 29, 79–86. Vinogradov, E.V., Van Der Drift, K., Thomas-Oates, J.E., Meshkov, S., Brade, H., Holst, O., 1999. The structures of the carbohydrate backbones of the lipopolysaccharides from Escherichia coli rough mutants F470 (R1 core type) and F576 (R2 core type). European Journal of Biochemistry 261, 629–639. Walther, D.M., Rapaport, D., Tommassen, J., 2009. Biogenesis of beta-barrel membrane proteins in bacteria and eukaryotes: evolutionary conservation and divergence. Cellular and Molecular Life Sciences 66, 2789–2804. Wells, T.J., Totsika, M., Schembri, M.A., 2010. Autotransporters of Escherichia coli: a sequence-based characterization. Microbiology 156, 2459–2469. Wilhelm, S., Rosenau, F., Becker, S., Buest, S., Hausmann, S., Kolmar, H., Jaeger, K.E., 2007. Functional cell-surface display of a lipase-specific chaperone. Chembiochem 8, 55–60. Wilhelm, S., Rosenau, F., Kolmar, H., Jaeger, K.E., 2011. Autotransporters with GDSL passenger domains: molecular physiology and biotechnological applications. Chembiochem 12, 1476–1485. Wilhelm, S., Tommassen, J., Jaeger, K.E., 1999. A novel lipolytic enzyme located in the outer membrane of Pseudomonas aeruginosa. Journal of Bacteriology 181, 6977–6986. Yamamoto, K., Fujimatsu, I., Komatsu, K.-I., 1992. Purification and characterization of the nitrilase from Alcaligenes faecalis ATCC 8750 responsible for enantioselective hydrolysis of mandelonitrile. Journal of Fermentation and Bioengineering 73, 425–430. Yang, T., Pan, J., Seo, Y., Rhee, J., 2004. Use of Pseudomonas putida EstA as an anchoring motif for display of a periplasmic enzyme on the surface of Escherichia coli. Applied and Environmental Microbiology 70, 6968–6976. Yang, T.H., Kwon, M.A., Song, J.K., Pan, J.G., Rhee, J.S., 2010. Functional display of Pseudomonas and Burkholderia lipases using a translocator domain of EstA autotransporter on the cell surface of Escherichia coli. Journal of Biotechnology 146, 126–129.