Influence of recombinant passenger properties and process conditions on surface expression using the

AIDA-I autotransporter

Martin Gustavsson

Royal Institute of Technology Stockholm 2013

ii

©Martin Gustavsson 2013 Division of Industrial Biotechnology School of Biotechnology Royal Institute of Technology AlbaNova University Centre Stockholm Sweden ISBN 978-91-7501-770-9 ISSN 1654-2312 TRITA-BIO Report 2013:9 Tryckt av Eprint AB 2013

iii

“If we knew what it was we were doing, it would not be called research, would it?”

-Albert Einstein

iv

©Martin Gustavsson (2013): Influence of recombinant passenger properties and process conditions on surface expression using the AIDA-I autotransporter. School of Biotechnology, Royal Institute of Technology (KTH), AlbaNova University Centre, Stockholm, Sweden

Abstract

Surface expression has attracted much recent interest, and it has been suggested for a variety of applications. Two such applications are whole-cell biocatalysis and the creation of live vaccines. For successful implementation of these applications there is a need for flexible surface expression systems that can yield a high level of expression with a variety of recombinant fusion proteins. The aim of this work was thus to create a surface expression system that would fulfil these requirements. A novel surface expression system based on the AIDA-I autotransporter was created with the key qualities being are good, protein-independent detection of the expression through the presence of two epitope tags flanking the recombinant protein, and full modularity of the different components of the expression cassette. To evaluate the flexibility of this construct, 8 different model proteins with potential use as live-vaccines or biocatalysts were expressed and their surface expression levels were analysed. Positive signals were detected for all of the studied proteins using antibody labelling followed by flow cytometric analysis, showing the functionality of the expression system. The ratio of the signal from the two epitope tags indicated that several of the studied proteins were present mainly in proteolytically degraded forms, which was confirmed by Western blot analysis of the outer membrane protein fraction. This proteolysis was suggested to be due to protein-dependent stalling of translocation intermediates in the periplasm, with indications that larger size and higher cysteine content had a negative impact on expression levels. Process design with reduced cultivation pH and temperature was used to increase total surface expression yield of one of the model proteins by 400 %, with a simultaneous reduction of proteolysis by a third. While not sufficient to completely remove proteolysis, this shows that process design can be used to greatly increase surface expression. Thus, it is recommended that future work combine this with engineering of the bacterial strain or the expression system in order to overcome the observed proteolysis and maximise the yield of surface expressed protein. Keywords: AIDA-I, autotransport, biocatalysis, E. coli, live vaccines, surface expression

v

©Martin Gustavsson (2013): Influence of recombinant passenger properties and process conditions on surface expression using the AIDA-I autotransporter. School of Biotechnology, Royal Institute of Technology (KTH), AlbaNova University Centre, Stockholm, Sweden

Sammanfattning

Ytexpression är ett forskningsområde som har visats stort intresse på senare tid. Två användingsområden som föreslagits för ytexpression är framställning av celler som fungerar som biokatalysatorer eller vaccin. För att kunna implementera ytexpression för detta behövs flexibla ytexpressionssystem som kan ge höga expressionsnivåer för proteiner med varierande egenskaper. Målet med den här avhandlingen var att skapa ett ytexpressionssystem som uppfyller detta. För att uppnå detta skapades ett nytt ytexpressionssystem baserat på autotransportören AIDA-I. Huvudegenskaperna för detta system är att det finns goda möjligheter att detektera ytuttrycket baserat på två epitoptaggar på vardera sidan om det rekombinanta proteinet, samt att expressionskassetten är modulär vilket möjliggör utbyte av enskilda komponenter. Åtta olika proteiner med potential att anvädas som vaccin eller biokatalysatorer klonades in i den nya vektorn för att utvärdera dess förmåga att uttrycka proteiner med olika egenskaper. Alla åtta proteinerna kunde detekteras på cellytan med hjälp av flödescytometri-mätning av celler märktamed antikroppar mot de två epitoptaggarna, vilket visar att expressionssystemet är funktionellt. Förhållandet mellan signalerna från de två taggarna indikerade dock att flera av proteinerna framförallt förekom i delvis proteolytiskt nedbrutna former, vilket bekräftades med Western blot-analys av yttermembranproteinfraktionen. Proteolysen föreslogs bero på att dessa proteiner fastnar i periplasman under transporten till cellytan, och det fanns indikationer på att större storlek och högre cysteininnehåll negativt påverkar ytexpressionsnivåerna. Processdesign med sänkt pH och temperatur under odlingen fanns vara en framgångsrik strategi för att öka ytexpressionen för ett av modellproteinen. Med denna strategi ökades uttrycksnivån för modellproteinet med 400 % med samtidig minskning av andelen proteolytiskt nedbrutna former med en tredjedel. Även om detta inte helt avlägsnade proteolysen så visar det att processdesign kan användas för att starkt öka ytexpressionsnivåerna. Därför rekommenderas att framtida studier kombinerar detta med modifiering av bakteriestammen eller ytexpressionssystemet för att avlägsna proteolysen och maximera ytuttrycksnivåerna. Nyckelord: AIDA-I, autotransport, biokatalys, Escherichia coli, live vaccines, ytexpression

vi

List of Publications This thesis is based on the following publications, which are referred to by their roman numerals:

I. Nhan N, Gonzalez de Valdivia E, Gustavsson M, Hai T, Larsson G (2011): Surface display of Salmonella epitopes in Escherichia coli and Staphylococcus carnosus. Microbial Cell Factories 10:22 II. Gustavsson M, Bäcklund E, Larsson G (2011): Optimisation of surface expression using the AIDA autotransporter. Microbial Cell Factories 10:72 III. Jarmander J, Gustavsson M, Thi-Huyen D, Samuelson P, Larsson G (2012). A dual tag system for facilitated detection of surface expressed proteins in Escherichia coli. Microbial Cell Factories 11:118 IV. Gustavsson M, Muralheedaran MN, Larsson G (2013). Surface expression of ω-transaminase in Escherichia coli. Pending revision. V. Jarmander J, Janoschek L, Lundh S, Larsson G and Gustavsson M (2013). Process design for improved yield of surface expressed protein in Escherichia coli. Submitted manuscript VI. Gustavsson M, Do T-H, Tran NT, Lundh S, Larsson G, Samuelson P (2013): Improved cell surface display of Salmonella enterica serovar Enteritidis antigens in Escherichia coli. Manuscript

vii

Contributions to papers I to VI

Paper I: Contributed to the experimental work, supervision of the work and the manuscript.

Paper II: Planned and performed the majority of the experiments.

Analysed the results and wrote the manuscript. Paper III: Planned and performed the experiments together with J. Jarmander (JJ). Contributed to the manuscript. Paper IV: Planned and performed the majority of the experiments.

Wrote the manuscript. Paper V: Responsible for the original concept, planning and

performance of the majority of the experiments together with JJ. Analysed and modelled the data. Wrote the manuscript.

Paper VI: Planned and performed the expression study. Contributed

to the manuscript.

viii

Table of contents

1 INTRODUCTION 1

2 THE BACTERIAL CELL ENVELOPE 2 2.1 STRUCTURE OF THE BACTERIAL CELL ENVELOPE 2 2.2 OUTER MEMBRANE PROTEIN (OMP) BIOGENESIS 4

2.2.1 The Sec system 4 2.2.2 The β-barrel assembly machinery 5 2.2.3 Periplasmic chaperones 5

3 SURFACE EXPRESSION 6 3.1 GRAM-POSITVE SURFACE EXPRESSION 6 3.2 GRAM-NEGATIVE SURFACE EXPRESSION 7 3.3 APPLICATIONS FOR SURFACE EXPRESSION 9

3.3.1 Library screening 9 3.3.2 Live vaccine development 10 3.3.3 Bioremediation 10 3.3.4 Biocatalysis 11 3.3.5 Biosensors 12 3.3.6 Biofuel production 12

4 AUTOTRANSPORT 13 4.1 STRUCTURE OF AUTOTRANSPORTERS 13

4.1.1 The AT translocation unit 14 4.1.2 The passenger domain 15 4.1.3 Autotransporter signal peptides 17

4.2 THE AT TRANSPORT MECHANISM 18 4.3 AUXILIARY PROTEINS INVOLVED IN AT BIOGENESIS 20

4.3.1 Influence of the β-barrel assembly machinery 20 4.3.2 Influence of periplasmic chaperones 21

4.4 RECOMBINANT SURFACE EXPRESSION USING ATS 21 4.4.1 Adhesin involved in diffuse adherence (AIDA-I) 22

5 PRESENT INVESTIGATION 24 5.1 AIM AND STRATEGY 24 5.2 MODEL PROTEINS 24

5.2.1 Protein Z 25 5.2.2 Salmonella epitopes 26 5.2.3 ω-transaminase 26 5.2.4 Tyrosinases 27

ix

5.3 METHODS 28 5.3.1 Bacterial cultivation and protein expression 28 5.3.2 Analysis of surface expression 29

5.4 EVALUATION OF AN EXISTING AT SYSTEM 30 5.4.1 Expression vectors 30 5.4.2 Selection of a suitable expression host (II) 30 5.4.3 Surface expression of Salmonella epitopes (I) 32

5.5 DESIGN OF AN IMPROVED SURFACE DISPLAY VECTOR (III) 35 5.6 SURFACE EXPRESSION OF Ω-TRANSAMINASE (IV) 40 5.7 ENGINEERING H:GM FOR IMPROVED EXPRESSION (VI) 42 5.8 SURFACE EXPRESSION OF TYROSINASES 45 5.9 INFLUENCE OF CULTIVATION CONDITIONS (V) 47 5.10 INFLUENCE OF THE RECOMBINANT PASSENGER PROTEIN (III, IV, V) 50

6 CONCLUDING REMARKS 53

7 ACKNOWLEDGEMENTS 56

8 REFERENCES 57

x

xi

List of abbreviations aa Amino acid(s) Ac-ω-TA Arthrobacter citreus ω-transaminase AIDA-I Adhesin involved in diffuse adherence AT Autotransport(er) ATP Adenosine triphosphate BAM β-barrel assembly machinery E. coli Escherichia coli ESP Extended signal peptide FACS Fluorescence-activated cell sorting IgG Immunoglobulin G IM Inner membrane IPTG Isopropyl β-D-1-thiogalactopyranoside kDa kiloDalton LPS Lipipolysaccharide OM Outer membrane OMP Integral outer membrane protein Pfam The Pfam protein families database SE Salmonella enterica serotype Enteritidis SP Signal peptide TU Translocation unit

1

1 Introduction

The use of enzymes for biocatalytic production of enantiomerically pure compounds is advantageous compared to traditional chemical catalysis, mainly due to the high catalytic efficiency and enantioselectivity of enzymes. These properties enable greener, less energy consuming processes that yield purer products of high quality [1]. However, the cost of the enzymatic catalyst is often a concern for overall process economy [2]. Thus, there is a demand for improvements leading to more economical production processes of enzymes, which should preferably be reusable. Reusability typically demands enzyme immobilisation on a solid support, which adds extra downstream steps and increases the cost. Ideally, whole cells should be directly usable as biocatalyst, since this removes the majority of the downstream steps and enables simple separation and subsequent reuse of the catalyst through centrifugation or filtration. However, the cell envelope presents a potent mass transfer barrier that slows or completely prevents the entry of reaction substrates into the cell, where the enzyme can catalyse the desired reaction [2,3]. Furthermore, the cell interior contains a multitude of enzymes that may catalyse undesired side reactions. An a possible way to avoid these drawbacks is the expression of enzymes on the surface of the host cell.

Another application for proteins expressed on the cell surface is the creation of live vaccines [4]. Non-pathogenic bacteria expressing vaccine epitopes on the surface may serve as a cheap, edible vaccine. As an added benefit, surface proteins and oligosaccharides of bacteria are antigenic and may therefore act as adjuvants, leading to an increased immune response to the displayed antigens [5]. There is currently much research directed towards expression of proteins on the surface of different cells. However, few reports have focused on surface expression from a protein production point of view, a perspective that is crucial for the realisation of the use of surface expression in applications such as whole-cell biocatalysis or live vaccine development. In this work, the aim was to create a surface display system in the bacterium Escherichia coli using the autotransporter adhesin involved in diffuse

2

adherence (AIDA-I) [6] and to evaluate the suitability of this system for protein production, based on stability of expression, protein yield, and flexibility regarding the properties of the surface-expressed fusion protein. This thesis will begin by briefly describing the structure of the bacterial cell envelope and subsequently describe different ways that have been exploited for anchoring recombinant proteins on the surface of this structure. Then, a mechanistic description of autotransporter-based protein secretion will be given, with a focus on how autotransporters can be used for surface expression. Finally, the findings of the present work will be described.

2 The bacterial cell envelope

2.1 Structure of the bacterial cell envelope Bacteria are divided into two groups, Gram-positive and Gram-negative, based on the structure of their cell envelope. Gram-positive bacteria have an envelope consisting of a cytoplasmic phospholipid membrane, surrounded on the outside by a thick cell wall made up of cross-linked peptidoglycan [7]. This thick peptidoglycan layer provides mechanical stability to the cell, and is responsible for keeping its shape [7,8]. The peptidolycan consists of mixed polymers of N-acetyl glucosamine (NAG) and N-acetyl muramic acid (NAM), where NAM forms cross-links between different peptidoglycan polymers [8]. The cytoplasmic membrane is made from a mixture of phospholipids and forms a diffusion barrier into the cell. In addition, up to 70% of the mass of the membrane is proteins [8].

Gram-negative bacteria differ from Gram-positive, in that they have two cell membranes, aptly named the cytoplasmic or inner membrane (IM), and the outer membrane (OM). The volume located between the IM and OM is called the periplasm and contains a relatively thin layer of peptidoglycan [7]. This peptidoglycan layer is connected to the OM through covalent bonds made by an outer membrane lipoprotein, which is the most abundant

3

protein in E. coli with 700 000 copies per cell [8]. The IM is a typical lipid bilayer made up of phospholipids, while the OM is made up of both phospholipids and lipopolysaccharides (LPS) [9]. Due to the polar sugar moieties of LPS, the OM forms a diffusion barrier not only for hydrophilic molecules, but also hydrophobic ones [7,8]. However, small molecules (< 600 Da) can cross the OM through channels formed by pore-forming proteins (porins). These are a class of membrane-bound, β-barrel forming proteins found in Gram-negative bacteria. These porins form trimeric complexes in the outer membrane, and are typically present in the order of 105 copies per cell [7]. Figure 1 shows a comparison between the Gram-positive and Gram-negative cell envelope.

Figure 1: Comparison of the Gram negative (A) and Gram-positive (B) cell envelope. LPS: lipopolysaccharide, C: core oligosaccharide, CM: cytoplasmic membrane, LA: lipid A, OM: outer membrane, LPP: lipoprotein, PP: periplasm, TMP: transmembrane protein, IM: inner membreane

A) B)

OM

IM

O-antigen

LALPS

LPP

Peptidoglycan

C

TMP

PorinPP

CM

TMP

Peptidoglycan

4

2.2 Outer membrane protein (OMP) biogenesis Gram-negative outer membrane proteins (OMPs) are synthesised in the cytoplasm, with an N-terminal signal peptide that targets them for secretion to the periplasm through the Sec system [10,11]. Following secretion, this signal peptide is proteolytically cleaved. Finally, the OMPs are inserted into the OM. This insertion is aided by several periplasmic and membrane-bound proteins, which will be discussed in the following sections.

2.2.1 The Sec system The Sec system is an inner membrane protein complex responsible for transporting the majority of the secreted proteins from the cytoplasm into the periplasm, as well as inserting proteins in the IM [12]. The main component of Sec is the heterotrimeric complex SecYEG. SecYEG is an integral inner membrane protein that forms a translocation channel, through which unfolded proteins can be transported to the periplasm [12]. Proteins that are destined for secretion through Sec are tagged by N-terminal signal peptides. Depending on the properties of this signal peptide, secreted proteins may interact with cytosolic helper proteins such as the signal recognition particle (SRP), DnaK or SecB. Sec-mediated transport is dependent on the ATP hydrolase SecA, which is associated to SecYEG. Upon ATP hydrolysis, SecA undergoes a conformational change, reaching into the channel formed by SecYEG. Since SecA also binds to the polypeptide to be secreted, this leads to the polypeptide being pulled into the SecYEG channel at approximately 20-25 amino acids (aa) at a time [13]. An additional complex, SecDF is also needed for translocation. SecDF is an inner membrane protein with loops protruding into the periplasm. It has been suggested that these loops undergo a conformational change, driven by the simultaneous transport of protons, which helps translocation by pulling the polypeptide out of the SecYEG channel into the periplasm [12]. Figure 2 shows an overview of Sec-mediated translocation.

5

2.2.2 The β-barrel assembly machinery The β-barrel assembly machinery (BAM) is a complex of 5 separate proteins (BamA-BamE) located at the OM [14]. BamA is an integral membrane protein that forms a β-barrel in the OM and is the central part of the complex [14], while BamB-BamE are lipoproteins that are associated to the BamA β-barrel [15]. The exact mechanism of the BAM complex is currently unknown, but it has been implicated in the assembly of most outer membrane proteins [16].

2.2.3 Periplasmic chaperones Upon release into the periplasm, OMPs interact with several periplasmic chaperones. These include SurA, Skp and DegP [10]. SurA is a peptidyl-prolyl isomerase that also displays general chaperone activity [17]. DegP fills dual roles, as it acts as both a general chaperone and a protease involved in degradation of misfolded proteins in the periplasm [17]. Finally, Skp is a general chaperone that interacts with unfolded peptides as they exit the Sec system and prevents their periplasmic aggregation [11]. SurA, Skp and DegP are all thought to be involved in delivering OMPs to

SecYEG SecDF

SecAATP

ADP+Pi

SecB

Cytoplasm

Periplasm

N

C

Figure 2: Overview of Sec-mediated transport.

6

the BAM-complex [11] for subsequent OM insertion. This is divided in two pathways: the SurA pathway, and the Skp and DegP pathway. This is evidenced by the fact that Skp and DegP can make up for the lack of SurA, but knock-out of SurA in combination with either Skp or DegP prevents growth [18].

3 Surface expression Surface expression, or surface display, is the expression of a recombinant protein of interest on the surface of a host. This is achieved by genetically fusing the protein of interest with a surface protein of the host cell, and was initially reported in 1985 by Smith and co-workers who fused peptides to coat proteins of filamentous phages [19]. Shortly after the report by Smith et al., surface expression was adapted for use in bacteria [20,21], and later for yeast [22]. The use of cells has several advantages compared to phage display: cells replicate independently as opposed to phages; cells are comparably large, enabling easy separation through filtration or centrifugation as well as analysis and sorting using flow cytometry [23]; nevertheless, phage display have been used extensively for library screening purposes. However, the applications that are the aim of the present study benefit greatly from the improved separation associated with the larger size of cells compared to phages, and cell surface display will thus be the focus of this thesis.

3.1 Gram-positive surface expression Gram-positive bacteria have been suggested advantageous for surface display because of high structural stability due to their thick peptidoglycan cell wall [4,24]. Many Gram-positive surface proteins are covalently linked into this cell wall [24], leading to a firm attachment to the cell surface. Others instead link to the cytoplasmic cell membrane [4]. The most studied cell-wall binding protein is the Staphylococcal protein A from S. aureus. It consists of an N-terminal signal peptide for inner membrane transport, 4-5 functional, IgG-binding domains and a cell-wall anchoring domain (X) [4]. The anchoring domain X has been exploited for surface display by fusion to recombinant proteins of interest. Using this fusion, it

7

has been possible to surface-express proteins in Staphylococcus xylosis [25], Staphylococcus carnosus [26] and Lactococcus lactis. Other cell-wall anchoring proteins that have been used for surface display include the Streptococcus pyogenes protein M6 [27] and S. aureus fibronectin binding protein B [28]. Another class of Gram-positive proteins that have been studied for surface expression is spore coat proteins. For example, fusion of an ω-transaminase to Bacillus subtilis coat protein cotG resulted in spores with a 30-times increased transaminase activity [29]. Another example is display of a tetanus-toxin fragment by fusion to B. Subtilis protein CotB [30]. Finally, cell membrane anchored proteins have been used for protein display in Gram-positive bacteria, for instance using the lipoprotein DppE [4]. However, this strategy required the removal of the peptidoglycan cell-wall to achieve surface-exposure of the recombinant fusion protein [4].

3.2 Gram-negative surface expression A complication of surface expression in Gram-negative bacteria compared to Gram-positive is the structure of the cell envelope. Surface expressed proteins need to traverse two membranes, as compared to only one in Gram-positive bacteria, in order to reach the cell surface. To complicate things further, there is generally no ATP present in the periplasmic space, and neither is there a proton gradient across the OM to drive this translocation. Thus, the motive force for the OM translocation has to be supplied in another way if the secretion proceeds through a periplasmic intermediate [31]. Nevertheless, Gram-negative surface display has advantages as well, not the least in the well studied and widely used host organism E. coli. Both physiological and genetic properties of E. coli are relatively well known, and there are a wide variety of knockout mutants available. Furthermore, E. coli grows well on cheap, defined growth media, making it an excellent choice for large-scale processing. It is also easily manipulated using standard transformation methods.

8

Several different strategies have been developed for Gram-negative surface expression. The first reports utilised a sandwich fusion strategy, where a small peptide was inserted in an extracellular loop between two of the β-strands of the OMPs LamB and OmpA [21,32]. This strategy has later been used with different OMPs, such as E. coli OmpC [33], Pseudomonas aeruginosa OprF [34], and Vibrio cholerae OmpS [35]. OmpA is an interesting choice as an anchoring protein, since it has been shown to be expressed at high levels (105 copies per cell [36]). However, fusion to outer membrane β-barrel protein loops has a disadvantage in that only relatively small proteins can be expressed without disrupting the structure of the β-barrel. Improved results have been seen after further development of the system by fusing a truncated OmpA with the lipoprotein Lpp, and inserting the recombinant protein at the truncated OmpA C-terminus [37]. Using this chimera, termed Lpp-OmpA’, expression of large proteins (EGFP, 60 kDa [38] and cyclodextrin, 74 kDa [39]) have been reported, though the expression of cyclodextrin was only demonstrated using membrane fractionation and never using any whole-cell analysis [39]. A different strategy has been to fuse the protein of interest to either fimbrial (FimA and FimH) [40,41] or flagellar subunits (FlicC) [42]. The use of surface-appendage-based display is advantageous for vaccine development due to the high immunogenicity of fimbriae and flagella [24]. Additionally, fimbrial display promises the possibility of very high copy-numbers per cell (approximately 500 fimbriae exist on a single cell, and each is made up of 1000 subunits). However, only small peptides up to 30-odd amino acids have been successfully displayed using such systems [43]. Lipoproteins have also been used for Gram-negative surface display. Out of the different lipoproteins that have been used, the Pseudomonas syringae ice nucleation protein (INP) [44,45] is the most successful example. It has been used for surface expression of for instance enzymes [45,46] and vaccine epitopes [47], with sizes of up to 119 kDa [48]. A convenient feature of INP is that it contains of two small domains at the N- and C-terminus, and a central core domain consisting of repeats of 48 aa. This core domain can be varied in length by simply changing the number of repeats that are included, thus providing a variable-length spacer for presentation of recombinant proteins.

9

The final class of proteins that have been used for Gram-negative surface expression is the autotransporter (AT) family. In their mature form, ATs consist of a membrane-anchored β-barrel and a surface-expressed passenger domain [49]. This passenger domain can be substituted with a recombinant protein, which will then be displayed on the cell surface [50]. Since ATs are the main focus of this thesis, their transport mechanism and recombinant uses will be discussed in more detail in chapter 4.

3.3 Applications for surface expression Cell surface expression has been suggested for a variety of applications, including peptide library screening, whole-cell biocatalysis, live vaccine development, biosensors, bioremediation, and biofuel production. These different applications will be described in some detail in following sections.

3.3.1 Library screening A major application for surface display is creation and screening of protein or peptide libraries. Initially, phage display was used extensively for library screening, but as cell display systems have advanced their use has become more common [51]. The main advantage of cell display over phages is that cells can be propagated independently from a host. Furthermore, the larger size of cells compared to phages enables using whole-cell labelling methods coupled with fluorescence activated cell sorting (FACS) for rapid selection of cells displaying protein with the desired properties for further studies [23]. The beauty of phage and cell surface display for library screening is the direct connection between the phenotype of the displayed protein to the genotype stored inside the displaying cell or phage. Several systems for prokaryotic library display has been reported, mainly for E. coli [52-54] and Staphylococci [55,56]. In addition to bacterial display, yeast surface display has been used extensively for library screening [22,57,58].

10

3.3.2 Live vaccine development Another application of surface expression that has gathered much interest is the creation of live vaccines by expressing epitopes on the surface of non-pathogenic cells [4]. Live vaccines have several advantages over attenuated pathogens or purified protein subunits. Compared to attenuated vaccines, the most obvious advantage is increased safety. There is always a risk that attenuated strains may revert to their pathogenic form, which is not present if non-pathogenic strains expressing epitopes from the pathogenic strain is used instead. Surface proteins and lipopolysaccharides of live vaccines may also act as adjuvants, giving an increased immune response when compared to purified subunit vaccines [5,23]. Furthermore, downstream processing of the vaccine is greatly simplified compared to purification of a subunit vaccine. Also, the intended administration route for a live vaccine is either nasally or orally, removing the need for injection by trained personnel [59]. Finally, if the vaccine strain can colonize the recipient it is possible to increase the time of exposure, resulting in a stronger immunisation [5].

3.3.3 Bioremediation Surface expression has also been studied for use in bioremediation. Two different approaches have been suggested. The first is the display of enzymes that can degrade pollutants of interest. This is suitable if the target pollutant consist of organic compounds [60]. An example of this is the use of organophosphorous hydrolase for degradation of pesticides. Surface expression of the organophosphorous hydrolase was found to give seven times higher degradation rate for two tested pesticides compared to intracellular expression [61]. The second strategy for bioremediation using surface expression is the expression of peptides that are capable to bind pollutants to the cell surface. This approach is mainly intended for removal of non-degradable pollutants, such as heavy metals [62-64]. A similar approach has been investigated with intracellular binders, where the idea is that the host cell can take up the heavy metals. In this case the main use of the chelating peptides is to protect the cell from the toxic effects of the heavy metals

11

[65]. Usage of surface expression is advantageous compared to this approach due to the easier removal of the pollutants from the cell, enabling desorption of the pollutants and subsequent reuse of the expressing cells.

3.3.4 Biocatalysis The use of surface expression has also been suggested for biocatalysis. The idea is to create whole-cell biocatalysts displaying the enzyme or enzymes of interest. Whole-cell biocatalysts are often preferable over purified enzymes. One of the bigger advantages is the cost of preparing the biocatalyst, which can often be crucial for the total cost of the process [2]. The use of enzyme catalysts typically requires cell disruption followed by at least partial purification of the enzyme of interest. After purification it is also often preferable to immobilise the enzyme on a carrier to facilitate its separation from the reaction medium and enable its reuse [2,3]. This further adds to the cost of preparing the catalyst. The use of whole-cell biocatalysts reduces the required purification to simple centrifugation and washing of the cells. However, traditional whole-cell biocatalysts have a major disadvantage in the mass-transfer barrier associated with substrate diffusion over the cell membrane(s) [2,3]. By displaying the enzyme on the surface of the cell, one gets the typical advantages associated with whole-cell biocatalysis without any mass-transfer barrier, since the enzyme has free access to the reaction medium [4,24]. There are a number of reports of surface display of enzymes that have been used for catalysing a variety of syntheses in laboratory scale. This has been achieved using yeast, Gram-negative and Gram-positive bacteria, and even Bacillus spores. Some examples are the expression of sorbitol dehydrogenase using the AIDA-I autotransporter [66], expression of a lipase on the surface of Saccharomyces cerevisiae [67] and display of another lipase on the surface of S. carnosus [28].

12

3.3.5 Biosensors Another application for surface-expressed enzymes is biosensors. The idea is to surface-express an enzyme that can catalyse a reaction where the molecule that is to be detected is converted with a resulting colour, pH or redox change. Thus, if the cells expressing the enzyme are exposed to a sample containing the compound of interest there will be a measurable change. There are several examples where this has been achieved, including the use of the same organophosphorous hydrolase mentioned above [68].

3.3.6 Biofuel production Production of ethanol and other fuels through fermentation of lignocellulose waste is a much-studied method for biofuel production. In contrast to starch-based fermentation (such as corn starch fermentation), production of ethanol and other value-added compounds from lignocellulose-derived sugar does not compete with food production. However, yeast, a commonly used ethanol producer, is incapable of directly utilising cellulose as a carbon source [69]. Instead, several steps of pre-treatment of the cellulose is needed in order to release fermentable monosaccharides [70]. Cellulolytic fungi, such as Trichoderma reesei produce enzymes that are capable of degrading cellulose polymers. This requires a combination of endogluconases, cellobiohydrolases, and β-glucosidases. These are secreted and associated to binding proteins on the cell surface, forming cellulolytic complexes known as cellulosomes [71]. An elegant way of removing the need for enzymatic cellulose hydrolysis prior to ethanol fermentation is the transfer of these enzymes into ethanol-producing yeast or bacteria. Indeed, it has been shown that it is possible to produce ethanol through direct fermentation of cellulose, without separate enzymatic pre-treatment, using yeast cells displaying these three cellulolytic enzymes on the surface, at least in laboratory scale [72].

13

4 Autotransport Autotransporters (ATs) are a class of secreted proteins used by pathogenic Gram-negative bacteria to export various virulence factors. The first AT to be discovered was the IgA1 protease of Neisseria gonorrhoea, and its discovery was reported in 1987 [73]. Since then, a large number of ATs with diverse functions have been reported; at the time of writing, 8209 autotransporters are listed in Pfam [74]. This makes ATs the largest family of secreted proteins in Gram-negative bacteria [75]. Among the 7 different protein secretion families in Gram-negative bacteria, [76,77], autotransporters are classified as members of the type V family [78]. In addition to the classical autotransporters (type Va) that are the focus of this thesis, the type V secretion family also contains the two partner secretion system (type Vb) and the trimeric autotransporter adhesins (type Vc) [49].

4.1 Structure of autotransporters Autotransporters are synthesised as single molecules with three main components. These are an N-terminal signal peptide, a C-terminal domain, denoted the translocation unit (TU), and a passenger domain that is responsible for the functionality of each specific autotransporter [49]. This organisation was originally proposed by Pohlner et al. [73] and has since been confirmed both by sequence analysis and protein crystallisation [77]. Figure 3 shows an overview of the domain layout of a typical autotransporter.

14

4.1.1 The AT translocation unit As mentioned above, the translocation unit consists of the C-terminal part of the autotransporter [79]. This domain forms a β-barrel that integrates into the outer membrane of the host cell [73,80,81]. This β-barrel fulfils two main functions: firstly, it anchors the autotransporter to the outer membrane, and secondly it forms a pore that is believed to be responsible for transporting the passenger to the cell surface [73]. The mechanism for this transport will be further discussed below. While the sequence of the translocation unit is not conserved among different autotransporters, the β-barrel structure is. Typically, it consists of 12 antiparallel, amphiphilic β-strands, forming a hydrophobic exterior that is exposed to the lipid bilayer of the outer membrane, and a hydrophilic interior, forming a pore through the OM. This is consistent in all AT translocation units that have been crystallised so far [81-85]. The pore has been predicted and later confirmed by crystal structures to be filled by an α-helix, referred to as the ”linker” [49,86,87]. Here, it will be considered as being a part of the translocation unit domain as defined by Maurer et al. [79] and references to the translocation unit will thus be considered to include the α-helical linker. Figure 4 shows crystal structures of 3 AT translocation units.

Passenger

Translocationunit

Signalpeptide

C-terminal

N-terminal

A) B)

Figure 3: The structure of ATs. A) Overview of the domain layout of a typical AT. B) Structure of the AT Hpb, recreated from individual structures of the C-terminal translocation unit and the N-terminal passenger domain.

15

4.1.2 The passenger domain Compared to the translocation unit, the passenger domains of different ATs are much more diverse, with sizes ranging from at least 20 to 400 kDa and low sequence homology [88,89]. The reason for this diversity is that the passenger domains are responsible for the functionality of each individual autotransporter. There are examples of passengers with many different functions, including adherence, toxins, proteases, and serum resistance [89]. Table 1 shows a few ATs and their passenger functions.

Table 1: Selected ATs and their passenger functions.

Passenger Size

[kDa] Organism Function Reference AIDA-I 132 E. coli Adhesin [90]

BrkA 103 Bordetella pertussis Serum resistance [91]

Hbp 148 E. coli Haemoglobin protease [92]

IcsA 116 Shigella flexneri Intercellular spread [93] IgA1

protease 169 N. gonhorroea Protease [73]

Pertactin 60 B. pertussis Adhesin [94]

Pet 100 E. coli Protease [95]

         

NalPEspPHbp

Figure 4: Crystal structures of the translocation units of the ATs Hbp, EspP and NalP.

16

To date, the structure of at least 6 autotransporter passenger domains have been determined: the adhesin Hap from Haemophilus influenzae [96], the haemoglobin-binding serine protease Hbp from E. coli [92], the IgA protease (IgAP) of H. influenzae [97], pertactin from Bordetella pertussis [94], the vacuolating cytotoxin (VacA) of Heliobacter pylori [98], and the Pseudomanas aeruginosa esterase A (EstA) [83]. Out of these 6 structures, all but EstA share a common structural fold: a right-handed β-helix with three β-strands per turn (Figure 5). Despite the low sequence similarity between different passengers, sequence analysis of over 500 autotransporters by Junker et al. estimated that more than 97% of the analysed passengers contained this β-helix structure [99] , and Kajava et al. made similar predictions [100]. Such β-helix structures are relatively scarce in nature. Thus it seems unlikely that this uncommon structure would be so conserved among AT passengers unless it was of significance to the secretion mechanism. Indeed, it has been proposed that this structure plays an important role in AT biogenesis [99], as will be discussed in section 4.2.

Hbp Hap Pertactin

Figure 5 Crystal structures of the β-helical passenger domains of the ATs Hbp (left), Hap (center), and pertactin (right).

17

The C-terminal part of the β-helical passenger, termed the ”junction” region [101], has a significantly more stable fold than the rest of the β-helix [99,102]. It has been shown that the junction is necessary for correct transport and folding of several ATs [101,103]; hence it has also been referred to as an autochaperone domain [103]. The proposed mechanism is that the higher stability of the junction region means that it will form a stable fold on the cell surface, and subsequently acts as a scaffold for the folding of the rest of the passenger β-helix.

4.1.3 Autotransporter signal peptides The N-terminal part of ATs consists of a signal peptide that targets the protein for export to the periplasm via the Sec system. Generally, AT signal peptides are 20-30 aa in length, as is typical for signal peptides of proteins passing through the Sec system [77]. They have low sequence homology but follow a general pattern with a positively charged domain (N) followed by a hydrophobic (H) domain and, finally, a C domain that is recognized and cleaved by the signal peptidase [77]. However, a portion of the known autotransporters exhibit unusually long signal peptides, stretching from around 50 to 60 aa in length [49]. These so called extended signal peptides (ESPs) are characterized by the presence of a highly conserved extension in the N-terminal. This extension consists of an N-terminal charged region followed by a second hydrophobic region [49]. Bioinformatic analysis has shown that this ESP region is a feature unique to the type V protein secretion family [104]. The role of ESPs is not known. Initially it was suggested that it targets the AT to be assisted by signal recognition particle (SRP) during Sec-mediated translocation [105,106]. However, later evidence indicates that the ESPs in fact do not influence the targeting of the protein, but rather has an effect on the inner membrane translocation kinetics [107-109].

18

4.2 The AT transport mechanism In the classical AT transport mechanism, originally described by Pohlner et al. [73], the AT polypeptide contain all components required for its own secretion to the cell surface, hence the term autotransporter. However, results in recent years have shown that this is likely a simplification of reality, and the biogenesis of ATs is more complex than initially suggested. This section will describe the classical AT mechanism, with the following sections dealing with the more recent additions. Like all E. coli proteins, ATs are synthesised in the cytoplasm. Next, the N-terminal signal peptide targets the protein for secretion to the periplasm through the Sec system. The signal peptide is then cleaved from the protein by a signal peptidase and the protein is released in the periplasm. Following this secretion, the C-terminal β-barrel folds and inserts into the outer membrane, forming a pore, as described in section 4.1.1. Next, the helical linker and passenger are believed to form a hairpin structure through the β-barrel pore. The passenger is then believed to pass through the pore in a C-to-N-terminal direction and finally fold on the cell surface with the N-terminal of the passenger facing away from the cell [77]. In many cases, such as the ATs AIDA-I [90] and EspP [110], there are protease cleavage sites present on the C-terminal side of the passenger protein. Passengers containing such cleavage sites are ultimately cleaved off from the anchoring β-barrel and thereby released to the medium. This model of translocation is supported by experimental evidence showing that the C-terminal part of the passenger is indeed the first part that becomes exposed on the cell surface [111]. Furthermore, the crystallized β-barrel structures mentioned in section 4.1.1 show that the TU pore is large enough to accommodate two peptide chains simultaneously, which is required for the hairpin model [81]. An overview of this mechanism is shown in Figure 6.

Though elegant, the AT mechanism implies some limitations. The proposed mechanism requires the passenger to remain in an unfolded conformation in the periplasm, as it would otherwise be unable to fit through the narrow pore formed by the TU. This can be seen from the previously mentioned crystal structures of TUs of different ATs [81-85].

19

Furthermore, the finding that cysteine-containing passengers generally are difficult to express using ATs, due to the promotion of disulphide-formation in the oxidative environment of the periplasm, further supports the requirement of the passenger maintaining an unfolded conformation [112-114]. In fact, it has been shown that it is possible to selectively stall translocation at specific positions by introduction of cysteines in the passenger domain [111].

Figure 6: The classical AT transport mechanism. A) Overview of the transport. The protein is synthesised in the cytoplasm and translocated through the Sec system. Next, the TU inserts into the OM and transports the passenger to the surface, where it may be realeased through autoproteolysis. B) Detailed view of the surface translocation. Passenger secretion is initiated by the formation of a hairpin structure pulling the passenger thrugh the pore to the surface, where it subsequently folds.

NH3+

NH3+

NH3+

NH3+

NH3+

OM

Periplasm

1. Protein synthesis

2. Translocation throughthe Sec system

3. Insertion into OM

4. Translocation andfolding on cell surfasce

OM

IM

5. Autoproteolysis and release of the passengerto the medium

A)

B)

20

Results for the AT pertactin show that the β-helical passenger structure folds slowly [115]. It has been suggested that the slow folding that this structure imposes is the reason that so many passengers adopts a β-helical fold. The rationale for this hypothesis is that the slow folding of the β-helix means that the passenger is kept in an unfolded conformation in the periplasm. Upon translocation to the surface the stable junction region folds and forms a scaffold that accelerates the folding of the rest of the passenger on the surface, as mentioned in section 4.1.2. This folding on the surface would also prevent the passenger from sliding back through the pore, leading to the suggestion that vectorial folding of the passenger on the cell surface leads to a directed diffusion through the pore, and thereby provides the driving force for passenger translocation [111,116].

Finally, it has been suggested that the N-terminal extension of some AT signal peptides also assist in keeping the passenger in a translocation-competent conformation in the periplasm. It has been shown that the extended signal peptides are processed by the signal peptidase at a slower rate compared to regular signal peptides, resulting in a slower release to the periplasm. This is believed to prevent premature folding of the passenger before translocation through the TU pore has been initiated [77].

4.3 Auxiliary proteins involved in AT biogenesis In contrast to the initial model proposed by Pohlner et al. [73], it is now known that the AT mechanism is more complex and dependent on additional proteins. There is evidence indicating the involvement of both periplasmic chaperones and the BAM complex.

4.3.1 Influence of the β-barrel assembly machinery It has recently been shown that the β-barrel assembly machinery (BAM) is involved in the insertion of the AT β-barrel. This is not surprising, considering that most known OMPs are dependent on the BAM complex, as discussed in section 2.2.2. Of the subunits making up the BAM complex, BamA and BamD has been shown to be required for secretion

21

of the ATs Pet and Ag43 [16], and BamA is required for secretion of the AT Hbp [117]. Furthermore, cross-linking experiments have shown that the AT EspP interacts with BamA during OM insertion[118,119].

4.3.2 Influence of periplasmic chaperones As mentioned in section 2.2.3, the main periplasmic chaperones involved in assisting OMPs are SurA, DegP and Skp. They are believed to keep OMPs from misfolding and aggregating in the periplasm, target them to the BAM complex, and protect them from periplasmic proteases [77]. ATs are no exception, as was initially indicated by the fact that disulphide-containing proteins are more readily expressed in the absence of the periplasmic disulphide-bond-forming chaperone DsbA [113], which suggested that ATs indeed can interact with chaperones during periplasmic transit. More recent studies have confirmed that DsbA is not an exception. Rather, AT passenger domains seem to also interact with the general chaperones present in the periplasm. This is exemplified by results showing that the passenger of Hbp accumulates in the periplasm of SurA mutant strains [117]. It has also been shown that the AT EspP is expressed at lower levels in SurA and Skp mutants, and that the unfolded EspP passenger interacts with SurA and DegP in binding assays [120]. SurA, DegP and Skp seem to not be required for correct folding of the TU, at least not for the ATs AIDA-I and EspP [120,121]. However, these results are in contrast with results for the TU of the AT NalP, which has been shown to interact with Skp in vitro [122].

4.4 Recombinant surface expression using ATs Soon after the first autotransporter was discovered, the potential of using autotransporters for recombinant surface expression was realised. The first successful use of an autotransporter for recombinant expression was reported by Klauser et al., who used the N. gonorrhoeae IgA1 protease to secrete cholera toxin B [123]. Initial work on recombinant surface expression using autotransporters was mainly focused on the previously mentioned IgA1 protease and the E. coli AT adhesin involved in diffuse adherence (AIDA-I) [80,113,123,124]. More recently, a variety of different

22

ATs have been explored for recombinant uses, including haemoglobin protease (Hbp) [125], antigen 43 (Ag43) [126], and IcsA [127]. Nevertheless, AIDA remains the most studied AT for recombinant expression, and some examples of its uses include display of enzymes [128], enzyme inhibitors [129], protein libraries [130] and antigens [131]. Table 2 shows a small selection of proteins that have been expressed using ATs.

Table 2: Selection of recombinant passenger proteins that have been expressed using ATs.

Recombinant Function Size AT Reference

passenger (kDa)

Aprotinin Enzyme inhibitor 7 AIDA-I [129]

Cholera toxin B Toxin 13 IgA1 protease/

AIDA-I [124]

ESAT6 Antigen 6 Hbp [125] Sorbitol

dehydrogenase Enzyme 28 AIDA-I [66]

β-lactamase Enzyme 29 IcsA [127]

 

4.4.1 Adhesin involved in diffuse adherence (AIDA-I) The adhesin involved in diffuse adherence (AIDA-I) is an AT found in enteropathogenic E. coli. It was initially isolated from weaning pigs suffering from diarrhoea [6]. Its function is to confer the ability for the host cell to adhere to intestinal epithelial cells [6]. It was one of the first AT that was used for recombinant surface expression [124]. Compared to the previously used N. gonorrhoeae IgA1 protease [123] it was considered advantageous due to being an E. coli native protein transporter [124]. AIDA-I consists of a functional passenger domain that is flanked by an N-terminal signal sequence and C-terminal TU. The passenger domain is 84.7 kDa and responsible for the adherence activity [6,90], while the TU domain, denoted AIDAc, is 47.5 kDa [132]. AIDAc is in turn divided into

23

three parts: two β-domains (β1 and β2) connected by an α-helix. β2 forms the typical autotransporter β-barrel pore, while β1 forms the junction region [89]. This domain structure is shown in Figure 7. Following translocation to the cell surface, AIDA-I is proteolytically cleaved to release the mature passenger protein [90]. This cleavage is autocatalysed by two aa residues (Glu897 and Asp878) in the junction region [133]. The cleavage site has been identified and marks the border between the mature passenger and AIDAc [134]. Following proteolytic cleavage, the passenger remains associated to the outer membrane of the host cell, rather than being released to the medium [132]. However, it is possible to dissociate the cleaved passenger from the cell membrane by brief heat-treatment [90]. The identification of the autoproteolytic cleavage site has enabled the use of AIDA-I for recombinant surface expression by disruption of the cleavage through point mutation of this site. The function of the AIDA-I passenger is dependent on O-glycosylation by the cytoplasmic autotransporter heptosyl transferase [135,136]. The aah gene encoding the heptosyl transferase is located directly upstream from the aidA gene encoding AIDA-I, and the two proteins are transcribed as bis-cistronic mRNA [137]. In addition, the gene for AIDA-I has its own promoter aidA [137]. These promoters are expressed constitutively, but there expression levels are influenced by environmental factors such as nutrition, oxygen and temperature. The mechanism behind this regulation is presently unknown [137].

Figure 7: A: Domains structure of AIDA-I. AIDA-I consists of an N-terminal signal peptide, a passenger domain, a junction region (β1) and a TU (β2). B: Structure of AIDAc, showing the two β domains and the α-helical linker spanning the β2 pore. The structure was modelled using the Phyre2 online tool [138].

SP Passenger ȕ1 ȕ2

AIDAcA

B

ǃ1

ǃ2

24

5 Present investigation

5.1 Aim and strategy The aim of the present work was to design a system for surface expression in E. coli and evaluate its suitability in whole-cell biocatalysis and live vaccine applications. Such a system needs to be flexible in regards to accepting a variety of passenger proteins, and give high, homogenous yields of protein expression on the cell surface. The study was divided in three main parts. In the first part, expression of three proteins was evaluated using a pre-existing AIDA-I surface display system (papers I and II) in order to gain understanding of the expression and develop analytical methods. Based on the results from this evaluation, an improved surface display system was created (paper III), with the main improvements being better possibilities for detection of the expressed proteins, more flexibility for changes in the expression construct and improved control of the expression by the use of an inducible promoter. Finally, the improved surface display construct was used to express 8 different model proteins in order to evaluate the influence of the recombinant passenger protein, as well as the cultivation conditions (papers IV-VI).

5.2 Model proteins Six different model proteins were used in this study. They include two Salmonella surface proteins with potential for use as live vaccines, one transaminase, which is of interest for biocatalytic chiral amine synthesis, two tyrosinases that can be used for melanin or L-DOPA synthesis, and the synthetic protein Z. In addition, two engineered variants of the Salmonella proteins, and one engineered tyrosinase were created and studied. Together, these represent proteins with different properties and potential applications, which makes them interesting for this study. Table 3 shows a summary of the model proteins.

25

Table 3: Model proteins used as recombinant passengers in this study.

Passenger

Size [kDa]

Number of cysteines

Wild-type passengers SefA 14.5 0

H:gm 53.2 0

AcωTA 53.2 3

B. megaterium tyrosinase 34.2 0

R. etli tyrosinase 67.2 2

Engineered passengers

Z 6.7 0

H:gmd 11.0 0

H:gmd-SefA 25.4 0

R. etli core tyrosinase 35.9 2

5.2.1 Protein Z Protein Z is the synthetic B-domain of staphylococcal protein A (SpA), and forms a compact three-helix bundle. It was chosen as a model protein due to its small size, stable fold and due to it being a naturally secreted protein [139]. Z is highly soluble, minimizing the risk of inclusion body formation and lacks disulphide bonds. Together, these factors suggest that Z should be a comparatively easy protein to express and export to the periplasm. Therefore, using Z as a model enables isolating effects related to the AT-based translocation of the passenger to the cell surface, since other limitations due to expression and Sec-mediated secretion are minimized. The functionality of Z is to bind to the Fc domain of IgG antibodies, which is an added advantage since it enables easy detection without specific antibodies. Furthermore, this can be used to assess the functionality of surface-expressed Z, since only cells displaying correctly folded Z will bind to the antibody Fc region.

26

5.2.2 Salmonella epitopes As described in section 3.3.2, development of live vaccines is an area where surface expression has a big potential. It is of particular interest in poor regions of the world where expensive vaccines can not be afforded, since expensive purification steps can be avoided. Salmonella enterica serovar Enteritidis (SE) is, together with S. enterica serovar Typhimurium, a major cause of Salmonella infections [140]. SE is unique among Salmonella serovars by being the only human pathogen that routinely infects eggs [141]. This infection can occur without the egg-laying hen showing any symptoms [142]. Therefore, the prevention of SE infection of eggs is of high interest. One strategy to achieve this is to immunize chicken against SE. However, large-scale immunisation of chicken would require an efficient, low-cost vaccine. To this end, the expression of the SE surface proteins SefA [143] and H:gm [144] were expressed on the surface of E. coli using the AIDA-I surface expression system, as a step towards constructing a live vaccine against SE. The proteins SefA and H:gm are surface proteins from S. enterica serovar Enteritidis (SE). SefA is a fimbrial subunit protein and H:gm is a flagellar subunit. Being surface proteins that are specific for SE they are both interesting as potential epitopes for immunization against SE. SefA and H:gm are valuable as model proteins since they, like Z, are naturally expressed on the cell surface. Compared to Z they are larger and more complex proteins than the three-helix bundle of Z.

5.2.3 ω-transaminase Transaminases (EC 2.6.1.18) are enzymes catalysing the transfer of an amino group from one substrate (the amino donor) to a keto group on the second substrate (the amino acceptor), resulting in the formation of a new ketone and a new amine [145]. Figure 8 shows an example amine synthesis reaction using a transaminase. In nature, transaminases are often involved in amino acid metabolism, with substrates consisting of amino acids and keto acids. A subclass of transaminases, the ω-transaminases, accepts substrates lacking the acid moiety present in amino acids. The ω-transaminases are of industrial interest due to their ability to catalyse the

27

synthesis of chiral amines with high enantiomeric purity, a task that is difficult using traditional chemistry [146]. Thus, whole-cell transamination is an interesting application for surface expression.

In this study, an engineered Arthrobacter citreus ω-transaminase (Ac-ω-TA) variant [147] was used. Ac-ω-TA is large (53 kDa) and contains three cysteine residues. In addition, transaminases are cofactor-dependent enzymes - requiring the cofactor pyridoxal-5’-phosphate (PLP) - and also need to form homodimers in order to be active. Together, these factors make Ac-ω-TA a good model for proteins that can be expected to be difficult to express.

5.2.4 Tyrosinases Tyrosinases are copper-dependent enzymes that are classified as monophenol oxidases (EC 1.14.18.1), but also catalyse the reaction of diphenol oxidases (EC 1.10.3.1) [148], as shown in Figure 9. In nature, tyrosinases fill several functions, including production of melanin, detoxification of plant phenols in symbiotic bacteria and production of amino-acid-based antibiotics [149]. Tyrosinases are of interest for several biotechnological applications, such as production of L-DOPA [150], removal of phenolic compounds from waste water [151], and as biosensors for phenolic compounds [152]. Surface expression of tyrosinases could provide an economically competitive alternative to purified enzymes for all of these applications, and it is thus of interest.

ONH3

+ ONH2

transaminase

Amino-acceptor

Amino-donor Product amine Product ketone

Figure 8: Transaminase-catalysed chral amine synthesis, exemplified by production of (S)-methylbenzylamine from acetophenone and isopropylamine.

28

Two tyrosinases were expressed in this study: one from Bacillus megaterium [153] and one from Rhizobium etli [154]. These tyrosinases where interesting models since they are enzymes, meaning it is simple to evaluate whether they are expressed in a functional form using activity assays. They had also both been expressed in E. coli previously [153,154]. Furthermore, the R. etli tyrosinase is a large protein and also contains 5 cysteines, leading to the prediction that it might be the most difficult of all the proteins studied. However, its structure suggested that it might be possible to engineer a more suitable recombinant passenger based on this tyrosinase, which will be discussed further.

5.3 Methods

5.3.1 Bacterial cultivation and protein expression Unless otherwise stated, E. coli expressing the various AIDA fusion proteins have routinely been cultivated at 37 °C in a defined mineral salt medium with glucose as carbon source. With the exception of papers I and II where a constitutive promoter was used, protein expression has been induced by addition of isopropyl β-D-1-thiogalactopyranoside (IPTG) to 0.2 mM in the early exponential phase, and the cultures have typically been harvested after four generations of induction.

OH OH

OH

O

Omonophenol ortho-diphenol ortho-quinone

O2 H2O O2 H2O

Figure 9: Reactions catalysed by tyrosinases.

29

5.3.2 Analysis of surface expression In this study, two main methods have been employed for evaluation of expression of the different model proteins. These are analysis of antibody-labelled whole cells using flow cytometry, and fractionation of cellular protein followed by SDS-PAGE and Western blot analysis. Cell protein fractionation into soluble, IM and OM protein fractions, followed by electrophoretic separation, has been used mainly as a qualitative analysis of the expression. This method enables the evaluation of the localisation of the target protein within the host cell, which can be used as an indication for problems during protein secretion. This would be seen by the presence of the fusion protein in the IM or soluble fractions. Western blotting of the protein fractions also reveals the presence of proteolytic degradation products. Quantification and verification of the surface exposure has been done by flow cytometric analysis of cell samples labelled with fluorescent antibodies against detection tags present in the surface expression constructs. Flow cytometry is a single-cell analysis, giving as output a histogram showing the fluorescence distribution in the studied cell population. The mean fluorescence intensity of this population is proportional to the amount of protein present on the surface of the average cell. In other words, the mean fluorescence intensity corresponds to the yield of product per cell, given in mean fluorescence intensities. This has been used to compare the expression of the different proteins, as well as the influence of different expression conditions on the same protein. An overview of the flow cytometric analysis can be seen in Figure 10.

Figure 10: Flow cytometric analysis of cells labelled with antibodies against a surface-expressed protein. Red: negative cells without the target protein, blue: cells expressing the target protein.

30

5.4 Evaluation of an existing AT system

5.4.1 Expression vectors As a starting point for the work, the pre-exiting AIDA surface display vectors pMK90 [155] and its derivative pDT1 were used. These plasmids encode a construct based on wild-type AIDA-I with a truncated passenger domain. The expression cassette consists of the N-terminal signal peptide of AIDA-I, a multiple cloning site for recombinant passenger insertion, a linker consisting of the first 54 aa of the native AIDA-I passenger and the AIDAc translocation unit, under control of the native promoter aidA [137]. The two plasmids differ only by the inclusion of a His6-tag in the multiple cloning site of pDT1. This tag can be used together with anti-His6 antibodies to detect the expression of the passenger protein. pMK90 and pDT1 were used for an initial evaluation of the expression system (papers I and II).

5.4.2 Selection of a suitable expression host (II)

A common problem during protein expression is proteolytic degradation, and surface expression is no exception. For instance, it has previously been shown that the outer membrane protease OmpT can degrade proteins that are surface-expressed using ATs [124]. As such, OmpT negative strains are typically used, with the strain UT5600 being prevalent, along with BL21(DE3). However, both these strains have disadvantages for production. UT5600 is auxotrophic for several amino acids, leading to inability to grow in simple, defined salt media. BL21(DE3) grows readily on defined media, but has another disadvantage: usage of IPTG-induced expression plasmids leads to the simultaneous synthesis of T7 RNA polymerase. This diverts cellular resources from synthesis of the protein of interest. Thus we looked for an alternative E. coli strain without any of these drawbacks, but with an available OmpT-deletion mutant. E. coli K12 strain 0:17 [156] was a promising candidate, and after a preliminary study of growth in defined media it was chosen as the surface expression host for this work.

31

The plasmid pMK90-Z, carrying a fusion of the AIDA-I-based surface display system and Z, was cloned into the wild-type and OmpT-deletion mutant of 0:17 (0:17ΔOmpT) and the expression levels of Z on the surface was evaluated using flow cytometry of cells labelled with human IgGs binding to Z, and Western blotting with anti-AIDAc antibodies (Figure 11). As seen in the figure, both of the bacterial strains gave strong fluorescence signals, showing that the combination of the 0:17 strain and the AIDA system has a great potential for surface expression. A stronger fluorescence signal was found when the construct was expressed in the OmpT-mutant, indicating that OmpT degraded Z in the wild-type strain, which was expected. This was confirmed by the Western blot, where a weaker band corresponding to full-length AIDAc-Z fusion protein was present in the wild-type strain. A second band with a slightly higher molecular weight than the empty vector could also be seen in the wild-type sample indicating that the fusion protein was partially degraded. Z contains a known OmpT-cleavage site close to its C-terminus, and the size of the fragment on the Western blot conforms to the expected if Z was cleaved at this site. Together these results show that 0:17 is a suitable strain for surface expression using the AIDA-I-autotransporter and highlight the importance of using OmpT-negative hosts when expressing proteins on the cell surface. Thus 0:17ΔOmpT was subsequently used as an expression host in the present work.

32

5.4.3 Surface expression of Salmonella epitopes (I) To challenge the expression system with more complex protein, the proteins SefA and H:gm where cloned into the surface expression vector pDT1. pDT1 was chosen due to the presence of a His6-tag that enables antibody-based detection of the surface-expressed protein. The two Salmonella proteins were cloned with the His6-tag on their N-terminal side in order to ensure good exposure of the tag to the external environment. Expression of the two fusion proteins in the correct cellular compartment was evaluated by protein fractionation, and the resulting protein fractions were analysed using Western blotting, as described above. This showed the presence of both AIDAc-SefA and AIDAc-H:gm in the outer membrane protein fraction. While H:gm was found exclusively in this

Figure 11: Comparison of surface expression of protein Z in the wild type and in the OmpT negative E. coli strain 0:17. (A): Western blot of the outer membrane protein fraction, developed using antiserum against AIDAc. Lane 1: Marker, Lane 2: 0:17, Lane 3: 0:17ΔOmpT, Lane 4: 0:17 with pMK90, Lane 5: 0:17ΔOmpT with pMK90, Lane 6: 0:17 pMK90-Z, Lane 7: 0:17ΔOmpT with pMK90-Z. (B): Cells analysed by flow cytometry. Red: 0:17 with empty expression vector, Blue: 0:17 with pMK90-Z, Green: 0:17ΔOmpT with pMK90-Z.

33

protein fraction, trace amounts of SefA could be visualised in the soluble and inner membrane fractions (Figure 12). However, H:gm appeared only in degraded forms. Several bands corresponding to degradation products were also found in the OM fraction of cells expression AIDAc-SefA. Since the degradation was detectable using the AIDAc antiserum, the likely scenario is that the protein was degraded from the N-terminal end. This would mean that the proteolysis had to happen after translocation to the periplasm, due to the requirement of the N-terminal signal peptide for secretion through the Sec system.

Figure 12: Western blot of protein fractions from cells expressing H:gm-AIDAc and SefA-AIDAc, respectively. OM: outer membrane protein fraction, IM: inner membrane protein fraction.

H:gm - AIDAc

SefA - AIDAc

191 97 64 51 39 28 19

kDa H:gm

SefA

M

SefA

H:gm

M

SefA

H:gm M

OM IM Soluble

191

AIDAc

34

Although the presence of the fusions proteins was confirmed in the OM fraction, there was still the possibility that they were facing the periplasmic side instead of being correctly translocated to the cell surfaces. To investigate this, cells were labelled using fluorescent anti-His6 antibodies and subsequently analysed using flow cytometry. The resulting fluorescence histograms showed a clear increase in fluorescence for cells expressing AIDAc-SefA compared to the negative control (Figure 13). In contrast, no fluorescence was detected from cells expressing AIDAc-H:gm. Since H:gm was only present in degraded forms, we hypothesised that the lack of fluorescence was due to cleavage of the N-terminal His6-tag. This would explain the lack of fluorescence, since there would be no tag for the antibody to bind.

To assess whether the His6-tag was indeed missing from H:gm, a second Western blot analysis of the OM protein fraction was performed with anti-His6 antibodies. The results confirmed our hypothesis, as full length AIDAc-SefA was clearly visible while no band could be detected from cells expressing AIDAc-H:gm (Figure 14). Thus, it could be concluded that both SefA and H:gm were transported to and incorporated in the E. coli OM. Additionally, SefA was correctly oriented towards the cell exterior, as confirmed by flow cytometric analysis, while the orientation of H:gm remained unknown due to the cleavage of the detection tag.

Figure 13: Fluorescence histograms for cells expressing AIDAc fused to SefA and H:gm, respectively.

35

5.5 Design of an improved surface display vector (III)

Based on the previous results (I,II) it was evident that some improvements needed to be made to the existing surface display system. From (II) it could be concluded that the control of expression from the wild-type AIDA-I promoter was unsatisfactory, and the results in (I) made clear the disadvantage of relying on a single tag for detection of passenger translocation. Consequently, the existing display vector was modified to add the desired functionality. Three main design criteria had to be fulfilled by the improved display vector:

i. Good possibilities for detection had to be provided, both for full-length and partially degraded passenger proteins.

ii. Protein production had to be under the control of an inducible promoter in order to obtain stable expression levels. Also, an inducible promoter would enable tuning of expression in order to avoid overloading the transport capacity of the cell.

iii. The expression cassette had to be modular, enabling simple exchange of any of the major parts (i.e. the signal peptide, passenger, linker and translocation unit).

His$tagged$SefA-AIDAC$64$

His-tagged$Lipase$A$(ref)$19$

kDa$

51$ His$tagged$AIDAC$

His$tagged$H:gm-AIDAC$missing!$

Figure 14: His6-tag Western blot of OM fraction of cells expressing SefA and H:gm.

36

The new surface display construct was designed using the plasmid pMK90 [155] as a starting point. First, the entire cloning region for passenger insertion was stripped from the sequence and replaced with only two restriction sites. Then, two small epitope detection tags (His6 and Myc [157]) where inserted flanking the passenger protein on the N- and C-terminal sides, respectively. Additionally, recognition sequences for two highly specific proteases from human rhinovirus (3C protease) and tobacco etch virus (TEV protease) where also inserted in flanking positions around the passenger. These protease sites were included to add the possibility to release the recombinant passenger from the cell surface by whole-cell treatment with TEV. This would in turn enable absolute quantification of the surface-expressed protein by combining His6-tag purification followed by conventional protein quantification assays. Finally, a total of eight unique recognition sites for common restriction enzymes where inserted between the major components of the construct. The construct was cloned into a pACYC184 vector backbone under the control of the LacUV5 promoter [158]. pACYC184 was chosen since it has a low-copy-number pA15 ori, giving approximately 20 copies per cell [159]. It was reasoned that a too high expression level would be deleterious to successful surface expression, since it has been shown that overexpression of high levels of membrane proteins is toxic to the cell, due to overloading of the capacity of the Sec translocon [160]. The new surface expression vector, denoted pAIDA1, is shown in Figure 15. The Salmonella proteins SefA and H:gm that where used in the initial study (section 5.4.3, paper I) were used as model proteins for the evaluation of the new display vector. They were chosen as they represent one protein (SefA) that was readily expressed and detected in the old vector and one that was not (H:gm). This enabled the comparison of expression levels between the two vectors based on SefA, as well as confirmation that the new vector enables detection of proteolytically degraded proteins on the cell surface (using H:gm).

37

After cultivation, cells harbouring pAIDA1-SefA and pAIDA1-H:gm were labelled using fluorescent antibodies against the two detection tags and analysed using flow cytometry. Figure 16A shows a comparison of the fluorescence histograms obtained from cells expressing SefA using the old and new vector, respectively. The histograms show a clear fluorescence signal corresponding to the His6-tag for both of the constructs. The signal from the new vector was approximately 50% higher than for the old vector. Furthermore, the signal from the construct pAIDA1-SefA showed a single, uniform cell population, while pDT1-SefA where divided into one expressing and one non-expressing population. The reason for this is presently unknown, but is speculated to have arisen due to heterogeneous induction of the aidA promoter in cells containing pDT1-SefA. Finally, cells harbouring pAIDA1-SefA had a strong signal from the Myc tag as well, showing that this tag could successfully be probed using the fluorescent antibodies. Notably, the Myc fluorescence histogram shows a population that is similar to the one obtained from the His6-tag.

pLacUV5 His6 3C site Passenger Linker AIDAcTEV site Myc

p15A oripAIDA15560 bp

CmR

LacI

Surface expression cassette

SPHindIII SalINdeI KpnI SacI XbaI NotI

His6

Myc

AIDAc

N

His6

AIDAc

N

pAIDA1

pDT1

Passenger

Passenger

Figure 15: Overview of the novel vector pAIDA1, and comparison of the detection tags present in pAIDA1 and pDT1.

38

Having confirmed that the detection tags worked for SefA, the next step was analysing the expression of H:gm from the new vector (Figure 16B). Compared to SefA, the His6-tag fluorescence obtained for both H:gm constructs was very low, as expected from our previous results. Notably, the signal was slightly higher from pAIDA1-H:gm, consistent with the comparison of the expression of SefA from the two vectors. Even though no signal was obtained from the His6-tag, probing cells expressing pAIDA1-H:gm produced a clear signal with a mean of approximately 50% of that obtained from pAIDA1-SefA. This confirmed our previous hypothesis that the degraded forms of H:gm were indeed expressed facing outwards on the cell surface, although the His6-tag was missing.

Figure 16: A: Surface expression of SefA using pDT1 (green) and pAIDA1 (blue), detected using His6-tag antibodies (left) and Myc-tag antibodies (right). Red: Negative control (E. coli without surface expression vector). B: Surface expression of H:gm using pDT1 (green) and pAIDA1 (blue), detected using His6-tag antibodies (left) and Myc-tag antibodies (right). Red: Negative control (E. coli without surface expression vector).

A

B

His-tag Myc-tag

Myc-tagHis-tag

Fluorescence Fluorescence

FluorescenceFluorescence

Count

Count

Count

Count

39

Protein from cells expressing the two new constructs was fractionated into soluble, inner membrane and outer membrane fractions. The resulting fractions were analysed using Western blot with antibodies against the His6- and Myc-tag, respectively (Figure 17). The resulting His6-tag blots showed similar results to the ones obtained previously with the old vector, confirming a similar surface expression. These results also correspond well with the ones obtained using flow cytometry: clear bands are present at the expected molecular weight for pAIDA1-SefA, showing that it was expressed as full-length protein, while no bands could be detected for either of the two H:gm constructs. The Myc-tag blots also confirmed that the SefA fusion was present as full-length protein, while only a very weak band could be found for H:gm. As expected H:gm was instead present in degraded forms. It should also be noted that all of the expressed SefA and H:gm was found in the correct protein fraction, i.e. the OM, and none was present in the IM or soluble fractions. These results indicate that the combination of a pA15 ori and the LacUV5 promoter leads to expression levels of the AIDAc fusion proteins do not overload the Sec translocon.

22511580

65

50

35

30

25

15

1 7652 3 4 1098kDa225

1158065

50

35

30

25

15

kDa 1 7652 3 4 1098

SefA H:gm

His-tagSefA H:gm

Myc-tag

Figure 17: Western blots of protein fractions from cells expressing AIDAc-SefA and AIDAc-H:gm fusions from the new vector pAIDA1. Left: Detection using anti-His6. Lanes 1 and 6: Size marker. Lanes 2 and 7: Outer membrane fraction of pAIDA1 without passenger (positive control). Lane 3: pAIDA1-SefA soluble fraction. Lane 4: pAIDA1-SefA inner membrane fraction. Lane 5: pAIDA1-SefA outer membrane fraction. Lane 8: pAIDA1-H:gm soluble fraction. Lane 9: pAIDA1-H:gm inner membrane fraction. Lane 10: pAIDA1-H:gm outer membrane fraction. Right: Detection using an anti-Myc antibody. Lanes 1 and 6: Size marker. Lane 2: pAIDA1-SefA soluble fraction. Lane 3: pAIDA1-SefA inner membrane fraction. Lane 4: pAIDA1-SefA outer membrane fraction. Lane 5 and 10: Outer membrane fraction of pAIDA1 without passenger (positive control). Lane 7: pAIDA1-H:gm soluble fraction. Lane 8: pAIDA1-H:gm inner membrane fraction. Lane 9: pAIDA1-H:gm outer membrane fraction.

40

5.6 Surface expression of ω-transaminase (IV) Having shown that the system could express the Salmonella epitopes it was of interest to investigate if it could also express a functional enzyme. Thus, the gene for an engineered variant of Arthrobacter citreus ω-transaminase (Ac-ω-TA) was cloned into the surface expression vector pAIDA1. The resulting AIDAc-Ac-ω-TA was expressed in the surface expression strain 0:17ΔOmpT. Surface expression of the fusion protein was assayed using flow cytometry and Western blotting of outer membrane protein isolations. Labelling of cells expressing Ac-ω-TA resulted of a detectable fluorescence signal for the c-Myc tag, while the signal from the His6-tag was relatively weak and almost comparable with the negative control (Figure 18). Western blotting with His6-antibodies confirmed that full-length transaminase was present exclusively in the outer membrane fraction (Figure 19). Probing identical Western blots with c-Myc antibodies again showed full-length transaminase in the outer membrane fraction, but also revealed large amounts of degraded forms of the transaminase (Figure 19). Nevertheless, since the transaminase was present in the OM, and to some degree present on the cell surface, it was deemed fruitful to assay the cells for activity.

Figure 18: Surface expression of Ac-ω-TA, measured using flow cytometry.

100 101 1031020

20

80

60

40

100

Cell c

oun

t

Myc-tag fluorescence

Fluorescence

100 101 1031020

20

80

60

40

100His-tag fluorescence

Cell c

oun

t

Fluorescence

41

An activity assay based on the conversion of a substituted (S)-aminotetraline to the corresponding tetralone was used to test whether the cells expressed active transaminase. This is a simple and rapid assay for detection of transaminase activity, where spontaneous oxidation of the formed tetralone into a coloured product indicates the presence of an active transaminase. Using this method, whole cells and cellular protein were analysed for the presence of active transaminase. After 4 days at 37 °C a brown colour had developed in the wells containing the OM protein fraction (Figure 20, confirming that the transaminase detected using Western blotting was active. Neither the IM or soluble protein fractions, nor the whole cells showed activity. Thus it could be concluded that the transaminase was expressed in the correct compartment but where not active while still attached to the cells. This is likely due to a too low degree of dimerization on the cell surface, especially since the probability of two full-length transaminases encountering each other is severely reduced due to the abundance of degraded forms.

Figure 19: Western blot of proteim fractions from cells expressing Ac-ω-TA.

225115

8065

50

35

30

25

15

His6-tag antibody

Mar

ker

pAID

A1 O

MpA

IDA1

-Ac-

ω-TA

OM

pAID

A1-A

c-ω

-TA IM

pAID

A1-A

c-ω

-TA S

olub

le

kDa225115

8065

50

35

30

25

15

Mar

ker

pAID

A1 O

MpA

IDA1

-Ac-

ω-TA

OM

pAID

A1-A

c-ω

-TA IM

pAID

A1-A

c-ω

-TA S

olub

le

Myc antibody

kDa

42

5.7 Engineering H:gm for improved expression (VI)

As discussed in section 5.5, use of the novel surface display vector pAIDA1 enabled verification that both SefA and H:gm were present on the cell surface. However, H:gm was mainly present in proteolytically degraded forms. Based on the expression results for H:gm and the A. citreus transaminase, it was hypothesized that it might be the large size of H:gm that lead to its low expression level. In addition to its size, the structure of H:gm, with its N- and C-terminals in close proximity (Figure 21), suggests that H:gm might be prone of folding into a hair-pin-like structure in the periplasm, which would prevent its C-terminal to enter the TU pore and trap H:gm in the periplasm, where it could subsequently be degraded. This would explain the size of the observed degradation products.

Figure 20: Colourimetric transaminase activity assay. -: Negative control, Sol: soluble protein fraction, IM: inner membrane protein fraction, OM: outer membrane protein fraction, AcωTA: purified transaminase (positive control).

- - -AcωTAOMIMSolWholecells

43

Based on these hypotheses, a smaller variant of H:gm, denoted H:gmd, was engineered. H:gmd consisted amino acid residues 254 to 351 of H:gm, giving a smaller size of approximately 11 kDa, compared to 53 kDa for the full H:gm. This was slightly smaller than the readily expressed SefA (14.5 kDa), which suggested that H:gmd had a good potential for successful expression, at least from a size perspective. Furthermore, H:gmd lacks the part of the H:gm structure that was predicted to be problematic for expression, which further suggested that it could be a good candidate. Finally, a previous study had shown that chicken exposed to SE develop antibodies against the H:gmd fragment, showing that it is antigenic and thereby suggesting its potential as a vaccine epitope [161]. In addition to expressing H:gmd, a fusion of H:gmd and SefA was made, with the rationale that the combined antigens might give rise to an enhanced immune response, which would be beneficial for creating a live vaccine. Furthermore, up to this point SefA was the protein that had given the highest expression. Thus it was reasoned that a fusion to SefA might assist in driving the secretion of H:gmd. Cells expressing the different Salmonella proteins were grown and subsequently analysed. Flow cytometric analysis of H:gmd and H:gmd-SefA revealed that both proteins were expressed on the cell surface as full-length protein (Figure 22), in contrast to H:gm that was previously shown in paper III to give only a weak signal for the His6-tag (Figure 16). Thus, engineering the shorter H:gmd was a success from a surface expression

Figure 21: Crystal structure of full-length H:gm, with the H:gmd fragment highlighted in cyan.

44

perspective. Surprisingly, both H:gmd and the H:gmd-SefA fusion showed significantly higher expression based on the His6-tag fluorescence compared to SefA, even though H:gmd-SefA was nearly twice the size of SefA alone.

This indicates that H:gmd possesses some properties that enhance the expression of the H:gmd-SefA fusion compared to SefA alone. At this point, the properties causing the increased expression are unknown. However, looking at Figure 21, the H:gmd fragment appear to have a rather loose fold, which can be assumed to not be very stable in the absence of the rest of H:gm. It is thus proposed that the enhanced expression of SefA when fused to H:gmd comes from a destabilisation due to the low stability of H:gmd, which prevents premature folding in the periplasm. This would be analogous to the proposed role of the slow-folding β-helical structure that most AT passengers adopt, as discussed in section 4.2. This hypothesis is supported by the recent finding by Renn et al. that reducing the folding stability of the passenger domain leads to an approximately linear increase in surface expression using the AT Pet [116]. Nevertheless, this hypothesis needs to be tested, for instance by studying the folding stability of SefA and the H:gmd-SefA fusion, respectively. Disregarding the exact mechanism, the increased expression of the H:gmd-SefA fusion is encouraging, since it implies that fusion to H:gmd, or other polypeptides with similar properties, might be a potential strategy for enhancing the surface expression also for other proteins than SefA.

Figure 22: Surface expression of the engineered passengers H:gmd (blue) and H:gmd-SefA (purple), compared with the previously used SefA (green). Red: negative control.

Fluorescence

Myc tagHis6-tag

Fluorescence

Co

un

t

Co

un

t

0 100

101

102

103 0 10

010

110

210

3

0

20

40

60

80

0

20

40

60

80

100

45

5.8 Surface expression of tyrosinases In addition to transaminases, tyrosinases were considered interesting models for whole-cell biocatalysis. As mentioned in section 5.2.4, they have potential fro use in synthesis of for L-DOPA and melanin, as well as degradation of phenolic pollutants. Furthermore, their ability to form melanin enables easy detection of their activity based on the dark colour developed. Two bacterial tyrosinases were chosen, one from B. megaterium and one from R. etli, since they have been expressed in E. coli before in active forms. The R. etli tyrosinase was the largest of the model proteins, and additionally contained 5 cysteines, which suggested that it might be difficult to express. However, this tyrosinase belongs to a family of tyrosinases with a specific domain structure. This structure consists of a core domain with tyrosinase activity and a C-terminal extension. This is typical in mushroom tyrosinases, where the extension is proteolytically cleaved, yielding a mature protein with increased activity [162]. It was recently demonstrated that engineering of a tyrosinase from the bacterium Verrucomicrobium spinosum by removing the extension resulted in increased activity, suggesting that the extension fills a similar role as in the mushroom tyrosinases [162]. It was thus reasoned that a similar engineering of the R. etli tyrosinase might yield an enzyme that is not only easier to express due to the smaller size and three fewer cysteines than its wild-type, but might also be more active. Homology modelling of the sequence of this core tyrosinase resulted in an almost identical structure as the crystallized B. Megaterium tyrosinase, suggesting that this was a feasible strategy (Figure 23).

46

The genes for the B. megaterium and R. etli tyrosinases, as well as the engineered R. etli core tyrosinase, were cloned into pAIDA1 and their expression was evaluated using flow cytometry (Figure 24). All three constructs showed similar fluorescence based on the Myc tag, while the His6-tag differed greatly. Both the B. megaterium and the R. etli core tyrosinase gave positive signals for His6-tag, while the full-length R. etli tyrosinase gave fluorescence comparable to the negative control. Thus it appeared that the large, cysteine-rich R. etli tyrosinase was not expressed as full length protein on the cell surface, while the other two tyrosinases were.

The catalytic activity of cells expressing the three tyrosinases was assayed by streaking the cells on minimal medium agar plates, additionally substituted with copper, tyrosine and IPTG (Figure 25) as previously described [154]. This resulted in a brown colour development of the colonies over time, indicating the conversion of tyrosine into melanin, which confirmed that the tyrosinases were expressed in active forms.

B. megaterium R. etli

Figure 23: Crystal structure of the B. megaterium tyrosinase (left) and homology model of the R. etli tyrosinases (right). The parts removed to create R. etli core tyrosinase are coloured grey.

47

R. etli core R. etli Negative controlB. megaterium

Figure 25: Tyrosinase activity assay. Cells expressing active tyrosinase develop a dark brown colour due to formation of melanin from tyrosine.

Figure 24: Surface expression of the studied tyrosinases (blue) compared to a negative control (red).

R. etli

R. etlicore

B. megaterium

His6-tag fluorescence Myc-tag fluorescence

0 100

101

102

103

0 100

101

102

103

0 100

101

102

103

0 100

101

102

103

0 100

101

102

103

0 100

101

102

103

0 100

101

102

103

0

20

40

60

80

100

0

20

40

60

80

100

0

20

40

60

80

100

0

20

40

60

80

100

0

20

40

60

80

100

0

20

40

60

80

100

48

5.9 Influence of cultivation conditions (V) Having found degraded forms of several of the studied passengers, an attempt was made to increase protein yield by lowering this proteolysis. It has previously been reported that periplasmic proteases (for instance DegP, which has been shown to degrade AT passengers that prematurely fold in the periplasm), are inhibited by pH below 6 [163]. It was thus hypothesised that decreasing the pH below this value during protein production would result in an increased yield of full-length protein on the cell surface. In addition to pH, the effects of temperature [163] and IPTG concentration on surface expression yield were studied. In order to obtain the maximum information from the experiments, a factorial design of experiments (DOE) approach was used to study the influence of the different factors. The use of DOE enabled studying interaction effects between the three parameters, in addition to their individual effects. The resulting surface expression levels were measured using flow cytometry, and models for surface expression yield, ratios of full-length to degraded protein, and cell growth rate were created using the software MODDE (Umetrix, Sweden). The model of the data shows a clear tendency of increased yield as pH and temperature decreases (Figure 26). In fact, the highest yield was obtained at the lower extreme of the studied conditions. Not surprisingly, the growth rate decreased markedly under these conditions, which was expected. Since no maximum for surface expression was found in the model, a second set of experiments were performed, where pH and temperature was decreased to 5.0 and 18 °C, respectively. From these data a second model was made. This model showed the same trend as the first one, with increasing yield at the lowest values for pH and temperature. However, these conditions led to very slow growth, and it was thus not considered realistic to further decrease the pH and temperature, especially since E. coli does not grow below pH 5 [164].

49

Finally, the expression at the optimum conditions (pH 5.0, 18 °C, 1250 µM IPTG) was compared to our previous standard conditions (pH 7.0, 37 °C, 200 µM IPTG). The resulting fluorescence histograms (Figure 27) showed a fourfold increase in yield of full-length protein, based on the His6-tag signal. At the same time, the fluorescence from the Myc-tag increased threefold. This indicates that the total expression increased, and that the level of degraded forms of SefA decreased by approximately one third relative to full-length protein. Nevertheless, there were still degradation products present in the OM fraction under all conditions, as seen by Western blots (data not shown). In other words, while process optimisation could reduce the relative abundance of degraded protein by a third, it is insufficient to completely remove the degradation. It is suggested that it is combined with engineering of either the expression host strain or the recombinant passenger for maximum effect.

Figure 26: Models of cell growth rate (A), surface expression of full-length SefA based on His6-tag fluorescence (B) and the ratio of full-length to total fusion protein on the cell surface (C). All models represent values predicted with an IPTG concentration of 1250 µM.

Growth rate (h-1) His-tag fluorescence His-tag/Myc-tag fluorescenceA B C

Tem

peratu

re (

°C

)

Tem

peratu

re (

°C

)

Tem

peratu

re (

°C

)

pH pH pH

50

5.10 Influence of the recombinant passenger protein (III, IV, V)

The presence of the His6- and Myc-tags in the new surface expression vector pAIDA1 enables detection of surface expression, independent of the chosen recombinant passenger. In contrast to using specific antibodies for each passenger, the usage of standard antibodies against the detection tags lead to fluorescence signals that are directly comparable, as the binding affinity of different antibodies is removed. Thus, the expression of the different model proteins in this study could be compared and used to estimate their relative expression levels. In order to evaluate the influence of the properties of the different passengers on surface expression, mean fluorescence intensities for the expressing populations were recorded for the 8 model proteins, excluding protein Z. To compensate for daily variance in the measurements, the mean fluorescence intensities were all normalised against a common positive control, which was chosen as cells expressing the AIDAc-SefA fusion protein. A diagram showing the resulting relative fluorescences for all of the model proteins is shown in Figure 28.

Figure 27: Comparison of surface expression at pre-optimised conditions (37 °C, pH 7.0, 200 µM IPTG, Blue) and the best condition found in this study (18°C, pH 5.0, 1250 µM IPTG, Green). Red: E. coli without plasmid (negative control).

Fluorescence

His-tag fluorescence

Cell c

oun

t

80

60

40

20

0

60

40

20

0

Cell c

oun

t

Fluorescence

Myc-tag fluorescence

0 100

101

102

103

0 100

101

102

103

51

From the figure it is evident that the two H:gmd fusions were expressed to the highest levels of all the studied proteins. Furthermore, it can be seen that Ac-ω-TA, the B. megaterium tyrosinase, H:gm and the R. etli tyrosinase have significantly lower normalised fluorescences for the His6-tag than the Myc-tag, indicating that they are not present on the cell surface as full length protein to any greater degree, which can also be seen on the Western blots of Ac-ω-TA and H:gm. The engineered core tyrosinase, on the other hand, is also only weakly expressed, but has a reversed proportion of fluorescence from the two tags compared to the B. megaterium tyrosinase, which has a similar size and fold. This indicates that the R. etli core tyrosinase either has a much lower total expression level in the cell but a very high efficiency of translocation to the cell surface, or that it is expressed to higher levels than expected because the Myc-tag is shielded by the tyrosinase, preventing the antibody from binding. Further investigation is needed to confirm which of these cases is true, but if it is the first case, then the engineered core tyrosinase would be very

B. m

egater

iumty

rosinas

e

AcƷ

TA

H:g

md

H:g

md-S

efA

H:g

m

R. etli

tyro

sinas

eR. et

li core

tyr

osinas

e

Sef

A

0

50

100

150

200

250

300

350

400

Rela

tive f

luo

rescen

ce (

%)

Figure 28: Surface expression of the different model proteins, determined by mean fluorescence intensities normalised against SefA. White bars: relative His6-tag fluorescence; grey bars: relative Myc-tag fluorescence.

52

interesting to study further to elucidate the mechanism behind its high secretion efficiency. The two H:gmd constructs also show significantly higher relative fluorescences from the His6-tag, with the same implications as the core tyrosinase. Finally, it is notable that the three proteins containing cysteines (Ac-ω-TA and the two R. etli tyrosinase variants) all show very low expression levels. This is in line with previous reports that have shown cysteine-containing proteins difficult to express. That the cysteines negatively affects the expression of Ac-ω-TA was confirmed by growing cells expressing the transaminase in the presence of β-mercaptoethanol, which lead to somewhat increased His6-tag fluorescence, although at the cost of significantly impaired cell growth (data not shown). Based on the early result from expression of Z, SefA, H:gm and Ac-ω-TA it was hypothesised that increased passenger size might have a negative impact on surface expression levels. To check whether this hypothesis was valid, mean fluorescences from the different model proteins were plotted versus the size of the protein (Figure 29). It was decided to include the detection tags and TEV and 3C protease sites in the calculation of the size of the respective passengers. The rationale behind this decision was that everything on the C-terminal side of the linker region was non-native to AIDA-I and could thus potentially have a negative impact on the expression levels. The plot of the Myc-tag fluorescences show a trend of decreased fluorescence as protein size increases. Notable exceptions are the B. megaterium tyrosinase and the core tyrosinase domains located at approximately 40 kDa, which are expressed at significantly lower levels than expected based on the other model proteins, and SefA (second leftmost data point) which also lies lower than the projected line. Whether this is due to low expression of SefA or exceptionally high expression of the two H:gmd fusion proteins is unknown. The trend is not as evident for the expression of His6-tag on the cell surface, but the tendency is the same, with the smallest proteins having the highest expression and a very sharp drop to the levels of the B. megaterium and core tyrosinases.

53

6 Concluding remarks

The aim of this study was to create a flexible system for surface expression of proteins in E. coli, with a focus on expression of enzymes and vaccine epitopes. A total of 5 unique proteins and 3 of their engineered derivatives were expressed using the novel vector. The presence of two generic, protein-independent detection tags in the expression system allowed quantitative analysis of the relative expression levels of the different proteins. To the authors’ knowledge, this makes the present work the largest comparative study of expression of different proteins using ATs to date. A key criterion for use of surface expression technology for the suggested applications is that the selected system is flexible in regard to the properties of the passenger protein. Unfortunately, the data from this study shows that the expression levels obtained using the AIDA-I autotransporter vary greatly for different recombinant passengers. There is thus more work needed before the desired flexibility can be obtained.

0 20 40 60 800

50

100

150

200

250

300

350

400

Rela

tive f

luo

rescen

ce (

%)

0 20 40 60 800

20

40

60

80

100

120

140

160

180

Passenger size (kDa)

Rela

tive f

luo

rescen

ce (

%)

Myc-tagHis6-tag

Passenger size (kDa)

Figure 29: Surface expression as a function of recombinant passenger protein size.

54

All of the studied proteins could be detected either by Western blot analysis of OM protein fractions or directly on the cell surface using whole-cell labelling with fluorescent antibodies. As mentioned, the expression levels of individual proteins varied, and several of the proteins were present to a large extent in proteolytically degraded forms. In fact, even the passengers that expressed well were found to contain degradation products. This is not unique for this study, but has been seen in several other reports where ATs are used [127,131,165,166]. This indicates that proteolysis is a common occurrence when using ATs, and needs to be more thoroughly investigated. This is of particular importance since there is limited room for protein over expression in the OM, and for applications such as whole cell biocatalysis it is critical to maximise the yield of protein per cell in order to achieve a whole-cell catalyst with sufficient activity. A tempting solution to solve this proteolysis would be to perform knock-out of different periplasmic proteases. However, it could be argued that the proteolysis is not the main problem, but rather a symptom of another underlying problem. The general belief is that ATs transport their passenger through the pore formed by the TU, as discussed in section 4.2. This pore is too small to accommodate folded passengers, which has been demonstrated by the difficulty of expressing cysteine-containing proteins. Thus, the conclusion is that if a passenger forms a stable fold in the periplasm it will get trapped. This has in turn been shown to lead to proteolytic degradation of the passenger by, for instance DegP [111,113,167]. Therefore, trying to increase expression levels by deleting periplasmic proteases might just result in an increase of misfolded passenger protein stuck in the periplasm. In the author’s opinion, a better approach would be to either engineer the host cell to increase the expression of the periplasmic chaperones discussed in section 2.2.3, or to further engineer the surface expression construct. Interesting parts of the surface expression cassette could be the signal peptide, and the linker region. ESPs have been suggested to assist in surface translocation through decreased rates for transport through the Sec system and signal peptide cleavage. This has been proposed as a mechanism to prevent aggregation or misfolding of the AT in the periplasm (section 4.2).

55

Modification of the linking region could be done by including native AT passengers on the C-terminal side of the recombinant passenger. Most native passengers consist of β-helical structures, which strongly suggest that this fold has a central role in AT biogenesis. In fact, it has recently been shown that the inclusion of the native passengers of the ATs IcsA and Hbp promote surface expression of recombinant proteins using these two ATs [125,127]. This is an encouraging finding that might lead to increased protein yields even for more difficult recombinant passengers. Finally, a surprising, yet encouraging, finding of the present study is that the H:gmd fragment appears to promote the expression of SefA if fused on the N-terminal side of SefA (paper V). An increase of the His6-tag-based fluorescence signal, corresponding to full-length protein on the cell surface, of approximately 50 % was detected in this fusion when compared to SefA alone. This shows that there is great potential in this kind of modifications that is yet to be realised.

56

7 Acknowledgements I would like to thank everyone who has made the work presented in this thesis possible. First and foremost, thank you Gen for accepting me as a PhD student and for your guidance and support. Thank you Johan for our collaboration on papers III and V, and for many long and fruitful discussion. Emma, thanks for all your help, all the conversations we have had and your contribution to the second paper. Patrik, thank you for help with solving cloning problems, for writing the last paper and for your support during the writing of this thesis. Susanna, thank you for your work with the tyrosinases and your contribution to papers V and VI. Ernesto, thank you for your help with cloning and your contribution to paper I.

Thanks to John, Filippa and Ken for all help with flow cytometry during these years. Thanks to the collaborators from Vietnam, Huyen, Tan and Professor Hai, for their contributions to the study of the Salmonella proteins. Andres and Johan, thank you for our many nice lunch time discussions. Thanks to all partners in the BIO-AMINES project for the collaboration during the first years of the work.

I am thankful to the diploma workers Cecilia, Nhan, Alberto, Mattia, Rafaela and Madhu who have all made contributions to the work presented here. Thanks to all colleagues at the Division of Industrial Biotechnology, past and present, for providing a nice work environment during the years.

Finally I would like to thank my family for their constant support Especially Ella for your support and putting up with my disappearance to the lab for long hours, and for being who you are. I love you.

57

8 References

[1] W.-D. Fessner, N.J. Turner, M.-X. Wang, Adv. Synth. Catal. 353 (2011) 2189.

[2] P. Tufvesson, W. Fu, J.S. Jensen, J.M. Woodley, Food Bioprod Process 88 (2010) 3.

[3] C.C.C.R. de Carvalho, Biotechnol Adv 29 (2011) 75. [4] P. Samuelson, E. Gunneriusson, P.-Å. Nygren, S. Ståhl, J Biotechnol

96 (2002) 129. [5] M.T. Dertzbaugh, Plasmid 39 (1998) 100. [6] I. Benz, M. Schmidt, Infect Immun 57 (1989) 1506. [7] J. Lengeler, G. Drews, H. Schlegel, Biology of the Prokaryotes, Wiley-

Blackwell, 1999. [8] F.C. Neidhardt, J.L. Ingraham, M. Schaechter, Physiology of the

Bacterial Cell: a Molecular Approach, Sinauer Associates Inc, 1990. [9] M.T. Madigan, J.M. Martinko, J. Parker, Brock Biology of

Microorganisms, Tenth, Pearson Educacion, 2003. [10] T.J. Knowles, A. Scott-Tucker, M. Overduin, I.R. Henderson, Nat Rev

Microbiol 7 (2009) 206. [11] N. Ruiz, D. Kahne, T.J. Silhavy, Nat Rev Microbiol 4 (2006) 57. [12] J.A. Lycklama A Nijeholt, A.J.M. Driessen, Philos. Trans. R. Soc.

Lond., B, Biol. Sci. 367 (2012) 1016. [13] J.P. van der Wolk, J.G. de Wit, A.J. Driessen, Embo J 16 (1997) 7297. [14] V. Robert, E.B. Volokhina, F. Senf, M.P. Bos, P. Van Gelder, J.

Tommassen, PLoS Biol. 4 (2006) e377. [15] K.H. Kim, S. Aulakh, M. Paetzel, Protein Sci 21 (2012) 751. [16] A.E. Rossiter, D.L. Leyton, K. Tveen-Jensen, D.F. Browning, Y.

Sevastsyanovich, T.J. Knowles, K.B. Nichols, A.F. Cunningham, M. Overduin, M.A. Schembri, I.R. Henderson, J. Bacteriol. 193 (2011) 4250.

[17] J.G. Sklar, T. Wu, D. Kahne, T.J. Silhavy, Genes Dev. 21 (2007) 2473. [18] A.E. Rizzitello, J.R. Harper, T.J. Silhavy, J. Bacteriol. 183 (2001) 6794. [19] G.P. Smith, Science 228 (1985) 1315. [20] A. Charbit, J.C. Boulain, A. Ryter, M. Hofnung, Embo J 5 (1986) 3029. [21] R. Freudl, S. MacIntyre, M. Degen, U. Henning, J. Mol. Biol. 188

(1986) 491. [22] E.T. Boder, K.D. Wittrup, Nat Biotechnol 15 (1997) 553. [23] G. Georgiou, C. Stathopoulos, P.S. Daugherty, A.R. Nayak, B.L.

Iverson, R. Curtiss, Nat Biotechnol 15 (1997) 29. [24] S.Y. Lee, J.H. Choi, Z. Xu, Trends Biotechnol 21 (2003) 45. [25] M. Hansson, S. Ståhl, T.N. Nguyen, T. Bächi, A. Robert, H. Binz, A.

Sjölander, M. Uhlén, J. Bacteriol. 174 (1992) 4239. [26] P. Samuelson, M. Hansson, N. Ahlborg, C. Andréoni, F. Götz, T.

Bächi, T.N. Nguyen, H. Binz, M. Uhlén, S. Ståhl, J. Bacteriol. 177

58

(1995) 1470. [27] D. Medaglini, G. Pozzi, T.P. King, V.A. Fischetti, P Natl Acad Sci

USA 92 (1995) 6868. [28] A. Strauss, F. Götz, Mol Microbiol 21 (1996) 491. [29] B.-Y. Hwang, B.-G. Kim, J.-H. Kim, Biosci Biotechnol Biochem 75

(2011) 1862. [30] R. Isticato, G. Cangiano, H.T. Tran, A. Ciabattini, D. Medaglini, M.R.

Oggioni, M. De Felice, G. Pozzi, E. Ricca, J. Bacteriol. 183 (2001) 6294.

[31] D.G. Thanassi, C. Stathopoulos, A. Karkal, H. Li, Mol Membr Biol 22 (2005) 63.

[32] A. Charbit, E. Sobczak, M.L. Michel, A. Molla, P. Tiollais, M. Hofnung, J. Immunol. 139 (1987) 1658.

[33] Z. Xu, S.Y. Lee, Appl Environ Microb 65 (1999) 5142. [34] R.S. Wong, R.A. Wirtz, R.E. Hancock, Gene 158 (1995) 55. [35] H. Lång, M. Mäki, A. Rantakari, T.K. Korhonen, Eur J Biochem 267

(2000) 163. [36] R. Koebnik, K.P. Locher, P. Van Gelder, Mol Microbiol 37 (2000)

239. [37] J.A. Francisco, C.F. Earhart, G. Georgiou, P Natl Acad Sci USA 89

(1992) 2713. [38] C. Yang, Q. Zhao, Z. Liu, Q. Li, C. Qiao, A. Mulchandani, W. Chen,

Environ. Sci. Technol. 42 (2008) 6105. [39] H.-M. Wan, B.-Y. Chang, S.-C. Lin, Biotechnol Bioeng 79 (2002) 457. [40] B. Stentebjerg-Olesen, L. Pallesen, L.B. Jensen, G. Christiansen, P.

Klemm, Microbiology 143 ( Pt 6) (1997) 2027. [41] K. Kjaergaard, J.K. Sørensen, M.A. Schembri, P. Klemm, Appl

Environ Microb 66 (2000) 10. [42] B. Westerlund-Wikström, J. Tanskanen, R. Virkola, J. Hacker, M.

Lindberg, M. Skurnik, T.K. Korhonen, Protein Eng 10 (1997) 1319. [43] P. Klemm, Microbiology (2000). [44] P.K. Wolber, C.A. Deininger, M.W. Southworth, J. Vandekerckhove,

M. van Montagu, G.J. Warren, P Natl Acad Sci USA 83 (1986) 7256. [45] H.C. Jung, J.H. Park, S.H. Park, J.M. Lebeault, J.G. Pan, Enzyme

Microb Tech 22 (1998) 348. [46] H.-C. Jung, S. Ko, S.-J. Ju, E.-J. Kim, M.-K. Kim, J.G. Pan, J Mol

Catal B: Enzym 26 (2003) 177. [47] J.S. Lee, K.S. Shin, J.G. Pan, C.J. Kim, Nat Biotechnol 18 (2000) 645. [48] S.-K. Yim, D.-H. Kim, H.-C. Jung, J.G. Pan, H.-S. Kang, T. Ahn, C.-

H. Yun, J. Microbiol. Biotechnol. 20 (2010) 712. [49] I.R. Henderson, F. Navarro-Garcia, M. Desvaux, R.C. Fernandez, D.

Ala'Aldeen, Microbiol Mol Biol R 68 (2004) 692. [50] J. Jose, R.M. Maas, M.G. Teese, J Biotechnol 161 (2012) 92. [51] W. Chen, G. Georgiou, Biotechnol Bioeng 79 (2002) 496. [52] U. Binder, G. Matschiner, I. Theobald, A. Skerra, J. Mol. Biol. {400}

(2010) {783. [53] J.A. Francisco, R. Campbell, B.L. Iverson, G. Georgiou, Proc Natl

59

Acad Sci USA 90 (1993) 10444. [54] J.J. Rice, P.S. Daugherty, Protein Eng. Des. Sel. 21 (2008) 435. [55] F. Fleetwood, N. Devoogdt, M. Pellis, U. Wernery, S. Muyldermans, S.

Ståhl, J. Lofblom, Cell Mol Life Sci 70 (2013) 1081. [56] J. Lofblom, H. Wernérus, S. Ståhl, FEMS Microbiol. Lett. 248 (2005)

189. [57] M.J. Feldhaus, R.W. Siegel, L.K. Opresko, J.R. Coleman, J.M.W.

Feldhaus, Y.A. Yeung, J.R. Cochran, P. Heinzelman, D. Colby, J. Swers, C. Graff, H.S. Wiley, K.D. Wittrup, Nat Biotechnol 21 (2003) 163.

[58] N. Gera, M. Hussain, B.M. Rao, Methods (2012). [59] A. Detmer, J. Glenting, Microb Cell Fact 5 (2006) 23. [60] M. Saleem, H. Brim, S. Hussain, M. Arshad, M.B. Leigh, Zia-ul-

Hassan, Biotechnol Adv 26 (2008) 151. [61] R.D. Richins, I. Kaneva, A. Mulchandani, W. Chen, Nat Biotechnol 15

(1997) 984. [62] D. Lovley, J. Coates, Curr Opin Biotechnol 8 (1997) 285. [63] G.M. Gadd, C. White, Trends Biotechnol 11 (1993) 353. [64] R. Biondo, F.A. da Silva, E.J. Vicente, J.E. Souza Sarkis, A.C.G.

Schenberg, Environ. Sci. Technol. 46 (2012) 8325. [65] K. Kuroda, M. Ueda, Appl Microbiol Biot 87 (2010) 53. [66] J. Jose, S. von Schwichow, ChemBioChem 5 (2004) 491. [67] M. Washida, S. Takahashi, M. Ueda, A. Tanaka, Appl Microbiol Biot

56 (2001) 681. [68] Y. Lei, P. Mulchandani, W. Chen, J. Wang, A. Mulchandani,

Biotechnol Bioeng 85 (2004) 706. [69] Y. Fujita, S. Takahashi, M. Ueda, A. Tanaka, H. Okada, Y. Morikawa,

T. Kawaguchi, M. Arai, H. Fukuda, A. Kondo, Appl Environ Microb 68 (2002) 5136.

[70] Y. Zheng, Z. Pan, R. Zhang, International Journal of Agricultural and Biological Engineering 2 (2009) 51.

[71] C.M.G.A. Fontes, H.J. Gilbert, Annu. Rev. Biochem. 79 (2010) 655. [72] Y. Fujita, J. Ito, M. Ueda, H. Fukuda, A. Kondo, Appl Environ

Microb 70 (2004) 1207. [73] J. Pohlner, R. Halter, K. Beyreuther, T.F. Meyer, Nature 325 (1987)

458. [74] R.D. Finn, J. Mistry, J. Tate, P. Coggill, A. Heger, J.E. Pollington, O.L.

Gavin, P. Gunasekaran, G. Ceric, K. Forslund, L. Holm, E.L.L. Sonnhammer, S.R. Eddy, A. Bateman, Nucleic Acids Res 38 (2009) D211.

[75] M.J. Pallen, R.R. Chaudhuri, I.R. Henderson, Curr Opin Microbiol 6 (2003) 519.

[76] A. Economou, P.J. Christie, R.C. Fernandez, T. Palmer, G.V. Plano, A.P. Pugsley, Mol Microbiol 62 (2006) 308.

[77] D.L. Leyton, A.E. Rossiter, I.R. Henderson, Nat Rev Microbiol 10 (2012) 213.

[78] I.R. Henderson, J.P. Nataro, J.B. Kaper, T.F. Meyer, S.K. Farrand,

60

D.L. Burns, B.B. Finlay, J.W. St Geme, Trends Microbiol 8 (2000) 352. [79] J. Maurer, J. Jose, T.F. Meyer, J. Bacteriol. 181 (1999) 7014. [80] T. Klauser, J. Kramer, K. Otzelberger, J. Pohlner, T.F. Meyer, J. Mol.

Biol. 234 (1993) 579. [81] C.J. Oomen, P. Van Ulsen, P. Van Gelder, M. Feijen, J. Tommassen,

P. Gros, Embo J 23 (2004) 1257. [82] T.J. Barnard, N. Dautin, P. Lukacik, H.D. Bernstein, S.K. Buchanan,

Nat. Struct. Mol. Biol. 14 (2007) 1214. [83] B. van den Berg, J. Mol. Biol. 396 (2010) 627. [84] Y. Zhai, K. Zhang, Y. Huo, Y. Zhu, Q. Zhou, J. Lu, I. Black, X. Pang,

A.W. Roszak, X. Zhang, N.W. Isaacs, F. Sun, Biochem J 435 (2011) 577.

[85] B.J. Loveless, M.H. Saier, Mol Membr Biol 14 (1997) 113. [86] D.C. Oliver, G. Huang, R.C. Fernandez, J. Bacteriol. 185 (2003) 489. [87] T. Suzuki, M.C. Lett, C. Sasakawa, J. Biol. Chem. 270 (1995) 30874. [88] N. Dautin, H.D. Bernstein, Annu. Rev. Microbiol. 61 (2007) 89. [89] I. Benz, M.A. Schmidt, Int J Med Microbiol 301 (2011) 461. [90] I. Benz, M.A. Schmidt, Infect Immun 60 (1992) 13. [91] L. Zhao, N.T. Nguyen, R.C. Fernandez, M.E.P. Murphy, Acta

Crystallogr F Struct Biol Cryst Commun 65 (2009) 608. [92] B.R. Otto, R. Sijbrandi, J. Luirink, B. Oudega, J.G. Heddle, K.

Mizutani, S.-Y. Park, J.R.H. Tame, J. Biol. Chem. 280 (2005) 17339. [93] L.D. Brandon, M.B. Goldberg, J. Bacteriol. 183 (2001) 951. [94] P. Emsley, I.G. Charles, N.F. Fairweather, N.W. Isaacs, Nature 381

(1996) 90. [95] C. Eslava, F. Navarro-Garcia, J.R. Czeczulin, I.R. Henderson, A.

Cravioto, J.P. Nataro, Infect Immun 66 (1998) 3155. [96] G. Meng, N. Spahich, R. Kenjale, G. Waksman, J.W. St Geme, Embo

J 30 (2011) 3864. [97] T.A. Johnson, J. Qiu, A.G. Plaut, T. Holyoak, J. Mol. Biol. 389 (2009)

559. [98] K.A. Gangwer, D.J. Mushrush, D.L. Stauff, B. Spiller, M.S. McClain,

T.L. Cover, D.B. Lacy, P Natl Acad Sci USA 104 (2007) 16293. [99] M. Junker, C.C. Schuster, A.V. McDonnell, K.A. Sorg, M.C. Finn, B.

Berger, P.L. Clark, P Natl Acad Sci USA 103 (2006) 4918. [100] A.V. Kajava, A.C. Steven, J. Struct. Biol. 155 (2006) 306. [101] Y. Ohnishi, M. Nishiyama, S. Horinouchi, T. Beppu, J. Biol. Chem.

269 (1994) 32800. [102] J.P. Renn, P.L. Clark, Biopolymers 89 (2008) 420. [103] D.C. Oliver, G. Huang, E. Nodel, S. Pleasance, R.C. Fernandez, Mol

Microbiol 47 (2003) 1367. [104] M. Desvaux, L.M. Cooper, N.A. Filenko, A. Scott-Tucker, S.M.

Turner, J.A. Cole, I.R. Henderson, FEMS Microbiol. Lett. 264 (2006) 22.

[105] R. Sijbrandi, J Biol Chem 278 (2002) 4654. [106] J.H. Peterson, J Biol Chem 278 (2003) 46155. [107] N. Chevalier, M. Moser, H.-G. Koch, K.-L. Schimz, E. Willery, C.

61

Locht, F.C.O. Jacob-Dubuisson, M. M uuml ller, J. Mol. Microbiol. Biotechnol. 8 (2004) 7.

[108] J.H. Peterson, R.L. Szabady, H.D. Bernstein, J. Biol. Chem. 281 (2006) 9038.

[109] L.D. Brandon, N. Goehring, A. Janakiraman, A.W. Yan, T. Wu, J. Beckwith, M.B. Goldberg, Mol Microbiol 50 (2003) 45.

[110] N. Dautin, T.J. Barnard, D.E. Anderson, H.D. Bernstein, Embo J 26 (2007) 1942.

[111] M. Junker, R.N. Besingi, P.L. Clark, Mol Microbiol 71 (2009) 1323. [112] J. Jose, R. Bernhardt, F. Hannemann, ChemBioChem 2 (2001) 695. [113] J. Jose, J. Kramer, T. Klauser, J. Pohlner, T.F. Meyer, Gene 178 (1996)

107. [114] D.L. Leyton, Y.R. Sevastsyanovich, D.F. Browning, A.E. Rossiter, T.J.

Wells, R.E. Fitzpatrick, M. Overduin, A.F. Cunningham, I.R. Henderson, J. Biol. Chem. 286 (2011) 42283.

[115] M. Junker, P.L. Clark, Proteins 78 (2010) 812. [116] J.P. Renn, M. Junker, R.N. Besingi, E. Braselmann, P.L. Clark, Chem.

Biol. 19 (2012) 287. [117] A. Sauri, Z. Soprova, D. Wickström, J.-W. de Gier, R.C. Van der

Schors, A.B. Smit, W.S.P. Jong, J. Luirink, Microbiology 155 (2009) 3982.

[118] R. Ieva, H.D. Bernstein, P Natl Acad Sci USA 106 (2009) 19120. [119] J.H. Peterson, P. Tian, R. Ieva, N. Dautin, H.D. Bernstein, P Natl

Acad Sci USA 107 (2010) 17739. [120] F. Ruiz-Perez, I.R. Henderson, D.L. Leyton, A.E. Rossiter, Y. Zhang,

J.P. Nataro, J. Bacteriol. 191 (2009) 6571. [121] J.E. Mogensen, D. Tapadar, M.A. Schmidt, D.E. Otzen, Biochemistry

44 (2005) 4533. [122] J. Qu, C. Mayer, S. Behrens, O. Holst, J.H. Kleinschmidt, J. Mol. Biol.

374 (2007) 91. [123] T. Klauser, J. Pohlner, T.F. Meyer, Embo J 9 (1990) 1991. [124] J. Maurer, J. Jose, T. Meyer, J. Bacteriol. 179 (1997) 794. [125] W.S. Jong, Z. Soprova, K. de Punder, C.M. ten Hagen-Jongman, S.

Wagner, D. Wickström, J.-W. de Gier, P. Andersen, N.N. van der Wel, J. Luirink, Microb Cell Fact 11 (2012) 85.

[126] B. Ramesh, V.G. Sendra, P.C. Cirino, N. Varadarajan, J. Biol. Chem. 287 (2012) 38580.

[127] M. Lum, R. Morona, Microb Cell Fact 11 (2012) 20. [128] I. Muñoz-Gutiérrez, R. Oropeza, G. Gosset, A. Martínez, J Ind

Microbiol Biotechnol 39 (2012) 1141. [129] J. Jose, D. Zangen, Biochem Biophys Res Commun 333 (2005) 1218. [130] J. Jose, D. Betscheider, D. Zangen, Anal Biochem 346 (2005) 258. [131] K. Petermann, S. Vordenbäumen, J.-C. Pyun, A. Braukmann, E. Bleck,

M. Schneider, J. Jose, Anal Biochem 407 (2010) 72. [132] M. Suhr, I. Benz, M.A. Schmidt, Mol Microbiol 22 (1996) 31. [133] M.-È. Charbonneau, J. Janvore, M. Mourez, J. Biol. Chem. 284 (2009)

17340.

62

[134] M.-È. Charbonneau, F. Berthiaume, M. Mourez, J. Bacteriol. 188 (2006) 8504.

[135] I. Benz, M.A. Schmidt, Mol Microbiol 40 (2001) 1403. [136] M.-È. Charbonneau, M. Mourez, Res Microbiol 159 (2008) 537. [137] I. Benz, T. van Alen, J. Bolte, M.E. Wörmann, M.A. Schmidt,

Microbiology 156 (2010) 1155. [138] L.A. Kelley, M.J.E. Sternberg, Nat Protoc 4 (2009) 363. [139] B. Nilsson, T. Moks, B. Jansson, L. Abrahmsen, A. Elmblad, E.

Holmgren, C. Henrichson, T.A. Jones, M. Uhlén, Protein Eng 1 (1987) 107.

[140] C.R. Braden, Clin. Infect. Dis. 43 (2006) 512. [141] J. Guard Petter, Environ Microbiol 3 (2001) 421. [142] A. Hogue, P. White, J. Guard-Petter, W. Schlosser, R. Gast, E. Ebel, J.

Farrar, T. Gomez, J. Madden, M. Madison, A.M. McNamara, R. Morales, D. Parham, P. Sparling, W. Sutherlin, D. Swerdlow, Rev Sci Tech 16 (1997) 542.

[143] A.D. Ogunniyi, I. Kotlarski, R. Morona, P.A. Manning, Infect Immun 65 (1997) 708.

[144] J. Li, K. Nelson, A.C. McWhorter, T.S. Whittam, R.K. Selander, P Natl Acad Sci USA 91 (1994) 2552.

[145] R. Silverman, The Organic Chemistry of Enzyme-Catalysed Reactions, 2nd ed. Academic Press, London, 2000.

[146] M. Höhne, U.T. Bornscheuer, ChemCatChem 1 (2009) 42. [147] S. Pannuri, S.V. Kamat, A.R.M. Garcia, Thermostable Omega-

Transaminases, 2006. [148] K. Lerch, 600 (1995) 64. [149] M. Fairhead, L. Thöny-Meyer, N Biotechnol 29 (2012) 183. [150] S. Surwase, J. Jadhav, Amino Acids 41 (2011) 495. [151] H.M. El-Shora, M. Metwally, Biotechnology 7 (2008) 305. [152] B.X. Gu, C.X. Xu, G.P. Zhu, S.Q. Liu, L.Y. Chen, X.S. Li, J Phys

Chem B 113 (2009) 377. [153] V. Shuster, A. Fishman, J. Mol. Microbiol. Biotechnol. 17 (2009) 188. [154] N. Cabrera-Valladares, A. Martínez, S. Piñero, V.H. Lagunas-Muñoz,

R. Tinoco, R. de Anda, R. Vázquez-Duhalt, F. Bolívar, G. Gosset, Enzyme Microb Tech 38 (2006) 772.

[155] N. Casali, M. Konieczny, M.A. Schmidt, L.W. Riley, Infect Immun 70 (2002) 6846.

[156] M.O. Olsson, L.A. Isaksson, Mol Gen Genet 169 (1979) 251. [157] G.I. Evan, G.K. Lewis, G. Ramsay, J.M. Bishop, Mol Cell Biol 5

(1985) 3610. [158] B.L. Wanner, R. Kodaira, F.C. Neidhardt, J. Bacteriol. 130 (1977) 212. [159] I. Prytz, A.M. Sandén, T. Nyström, A. Farewell, Å. Wahlström, C.

Förberg, Z. Pragai, M. Barer, C. Harwood, G. Larsson, Biotechnol Bioeng 83 (2003) 595.

[160] S. Wagner, L. Baars, A.J. Ytterberg, A. Klussmeier, C.S. Wagner, O. Nord, P.-Å. Nygren, K.J. van Wijk, J.-W. de Gier, Mol. Cell Proteomics 6 (2007) 1527.

63

[161] N. Mizumoto, Y. Toyota-Hanatani, K. Sasai, H. Tani, T. Ekawa, H. Ohta, E. Baba, Vet. Microbiol. 99 (2004) 113.

[162] M. Fairhead, L. Thöny-Meyer, Febs J 277 (2010) 2083. [163] F. Baneyx, A. Ayling, T. Palumbo, D. Thomas, G. Georgiou, Appl

Microbiol Biot 36 (1991) 14. [164] F.C. Neidhardt, Escherichia Coli And Salmonella., 1996. [165] S.D. Schumacher, J. Jose, J Biotechnol 161 (2012) 113. [166] M.P. Konieczny, M. Suhr, A. Noll, I.B. Autenrieth, M. Alexander

Schmidt, FEMS Immunol. Med. Microbiol. 27 (2000) 321. [167] W.S.P. Jong, C.M. ten Hagen-Jongman, T. den Blaauwen, D.J.

Slotboom, J.R.H. Tame, D. Wickström, J.-W. de Gier, B.R. Otto, J. Luirink, Mol Microbiol 63 (2007) 1524.