Intein mediated biotinylation of proteins and its application in protein microarray

We successfully demonstrated that, for the first time, intein-mediated protein biotinylation proceeded inside both bacterial and mammalian living cells, as well as in a cell-free protein

NATIONAL UNIVERSITY OF SINGAPORE

Acknowledgements

First, I would like to thank the National University of Singapore (NUS) and the Agency for Science, Technology and Research (A*STAR) of Singapore for the funding support I also thank the Department of Biological Sciences (NUS) for granting me the Research Scholarship that financially supported me through my post-graduate days I would also like to thank Joan and Reena for their advice on most of the administrative matters I really appreciate A/P Yao Shao Qin for his guidance and moral support As the supervisor for both my honors and master’s projects, he has always provided me with insightful discussion His continuing vision is the main key to the success of this project Special thanks to Grace for organizing the enjoyable lab outings Besides that, she has also given

me lots of technical advice and assistance on the project Thanks also to Dr Zhu Qing for synthesizing the cysteine-biotin probe Last but not the least, I would like to thank all the people working in the Functional Genomic Laboratory (FGL) and my fellow lab mates in the Department of Chemistry for their valuable friendship

2.1 Chemical synthesis of the cysteine-biotin 11

2.1.3 Purification and identification of cysteine-biotin 12

2.2 Cloning of target genes into pTYB1 & pTWIN expression vector 12

2.3 Site-directed mutagenesis of pTYB1-wtEGFP (Lys239)-intein 14

2.5 Affinity purification & C-terminal biotinylation of 15

recombinant proteins

2.7 In vivo protein biotinylation in E coli 16

2.8 In vivo protein biotinylation of in mammalian cells 18

2.10 Cell free synthesis and biotinylation of MBP 20

3.1 General features of pTYB expression vectors 21

3.2 Intein-Mediated Biotinylation of three model proteins 24

3.2.1 Cloning of target genes into pTYB1 expression vector 24

3.2.2 Expression and extraction of fusion proteins 25 3.2.3 Affinity purification and on-column biotinylation 26

3.4 Immobilization of biotinylated proteins onto 34 self-assembled monolayers (SAM) in SPR analysis

3.5 Influence of C-terminal residues on biotinylation 37

3.6 High-throughput expression and biotinylation of yeast proteins 43

3.6.1 Cloning of yeast gene into pTYB1 expression vector 43

3.6.2 Expression, purification & biotinylation of yeast proteins 45

3.7.2.1 Construction of mammalian expression plasmid, 51

pT-Rex-DEST30-EGFP-Sce VMA intein-CBD 3.7.2.2 Expression and in vivo biotinylation of EGFP 52

in HEK 293 cells 3.7.3 Protein microarray generation using crude bacterial cell lysate 57

3.8 Protein biotinlyation using different inteins 59 3.9 Protein biotinylation in a cell-free system 63

4 Conclusion 66

Summary

The post-genome era has led us to a new frontier of proteomics that requires us to gain information on the millions of proteins encoded by these identified genes The challenge ahead therefore lies in the development of protein microarray that would enable

us to unravel the biological function of proteins in a massively parallel fashion This high-throughput screening technique would allow thousands of functional molecules to

be analyzed simultaneously, possibly leading to a better understanding of how these molecules affect cellular functions It can be used for discovery of novel protein functions, screening of protein-protein interactions, detecting enzyme-substrate interactions and identifying protein targets of biologically active small molecules Beside basic protein expression studies, application of the protein microarray technology has also evolved to diagnostics, mutation analysis, and toxicology in recent years The idea of

a protein microchip is to immobilize tens of thousands of protein molecules (e.g antibodies, receptors, enzymes) onto a solid surface such as glass slides Each of these proteins is geared towards identifying and binding of specific targets, thus it is necessary

to immobilize them in its native conformation and correct orientation to preserves their functional sites There are several reported strategies of immobilizing proteins onto solid surfaces but many of these mode of attachments are unspecific, causing the molecules to

be immobilized in the ‘wrong’ orientation In this report, we present an intein-mediated approach for efficient and site-specific immobilization of proteins The reactive C-

terminal thioester generated from intein-assisted protein splicing, either in vitro or in live

cells, served as an attractive, as well as exclusive site for attaching cysteine-containing biotin Using this novel biotinylation strategy, we were able to biotinylate many proteins

were subsequently immobilized onto different avidin-functionalized solid surfaces for applications such as protein microarray and surface plasmon resonance (SPR) spectroscopy We highlighted the numerous advantages of using biotin over other tags (e.g GST, His tag etc) as the method of choice in protein purification/immobilization In addition, our intein-mediated strategies also provided critical advantages over other protein biotinylation strategies in a number of different ways We successfully demonstrated that, for the first time, intein-mediated protein biotinylation proceeded inside both bacterial and mammalian living cells, as well as in a cell-free protein synthesis system Taken together, our results indicate the versatility of these intein-mediated strategies, which should provide invaluable tools for potential high-throughput proteomics applications They may also serve as useful tools for various biochemical and

biophysical studies of proteins both in vitro and in vivo

List of Tables

1 The influence of C-terminal residues on the in vivo cleavage of 39

EGFP-intein and on-column cleavage/biotinylation of EGFP

List of Figures

2 Biotin-tagging of protein via IPL reaction 8

3 Chemical structure of cysteine-biotin derivative 9

4 Three intein-mediated protein biotinylation strategies 10

5 Map and multiple cloning sites (MCS) of 22

pTYB1 & pTYB2 expression vector

6 Cloning of gene fragment into pTYB1 & pTYB2 23

9 Site-specific immobilization of biotinylated, functionally 29

active proteins onto avidin slides

10 Integrity of biotinylated proteins immobilized on 30

avidin-functionalized glass surface

11 Chemical structure of glutathione, natural ligand of GST 30

12 Biotinylated GST on an avidin slide treated with 32

different washing conditions

13 Overview of the on-column biotinylation strategy and 33 site-specific immobilization procedure

14 SPR data showing immobilization of biotinylated MBP on 35

avidin-functionalized sensor chip

15 SPR response of anti-MBP through the MBP-coated sensor chip 36

16 Influence of the C-terminal amino acid residue 40

17 Effect of an extra glycine residue on intein-mediated biotinylation 41

18 DNA fragments obtained from PCR amplification of the 44

19 DNA fragments obtained from NdeI and SapI 44

digestion of the TA plasmid

20 Cloning of yeast gene fragment into pTYB1 45

21 High-throughput expression and biotinylation of yeast proteins 47

22 Purification and biotinylation of a yeast protein (YAL012W) 47

23 Optimizing in vivo biotinylation conditions in bacterial cells 50

by anti-biotin blot

26 Construction of mammalian expression plasmid 54 using GatewayTM Technology

27 Expression of EGFP-Sce VMA intein-CBD in 55

different mammalian cell line

shown by anti-biotin blot

30 Site-specific immobilization of biotinylated proteins 57

onto avidin slides using bacterial cell lysate

31 Schematic representation of pTWIN vectors 60

32 Recovery of the intein fusion proteins from the cell extract 61

and its cleavage efficiency with MESNA and cysteine-biotin

33 Yield of EGFP from the different intein fusion 62

35 Protein biotinylation in a cell-free system 65

EDTA Ethylenediaminetetraacetic acid

EGFP Enhanced green fluorescent protein

FITC Fluorescein Isothiocyanate

IPL Intein-mediated protein ligation

IPTG Isopropyl thiogalactosidase

Kanr Kanamycin resistant

Ni-NTA Nickel nitrilotriacetic

NMR Nuclear magnetic resonance

PBS Phosphate buffer saline

PCR Polymerase chain reaction

Pro Proline

RTS Rapid Translation System

SDS-PAGE Sodium dodecyl sulfate polyacrylamide gel electrophoresis

TBTU O-(Benzotriazol-1-yl)-N,N,N',N'-tetramethyluronium

tetrafluoroborate TCEP Tris-(2-carboxyethyl)phosphine

1 Introduction

With the completion of the Human Genome Project, one may estimate that the number

of proteins in human could vary from approximately 40,000 to as many as 1,000,000.1This poses an even greater challenge ahead - to identify the structures and functions of all proteins encoded by the human genome Traditionally, the function of a protein is elucidated through its structure using NMR, X-ray crystallography and other related techniques Although structural features in a protein can help in determining it biochemical functions, they are not as useful in defining its biological functions (e.g its interacting partners and/or biological pathways) This calls for new methods that allow high-throughput determination of protein functions and/or interactions Protein microarray technologies satisfy many of these criteria, and in the past few years, have emerged as the uprising technology in the field of proteomics.1-5 Success stories from the DNA microarray in the last decade have propelled the rapid development of the protein microarray, providing a potential means for high-throughput identification and quantification of proteins from biological samples.6-13 Due to fundamental differences between proteins and DNA, however, the protein array technology is currently in its infancy Unlike DNA, which is highly stable and robust, proteins are known to lose its functional integrity upon immobilization onto a solid surface Furthermore, there is presently no known technique, which can effectively amplify proteins as in DNA Existing methods for protein expression have many limitations The inevitable chemical, physical and structural variation among different proteins results in their non-specific absorption to solid surfaces, thus creating further problems for their immobilization in a microarray Despite these technical hurdles, several research groups have successfully demonstrated

the functional use of protein microarrays using a wide variety of surface substrates and attachment chemistries.6,7,14-25

The simplest way to immobilize proteins on a solid support relies on non-covalent interactions such hydrophobic or van der Waals interactions, hydrogen bonding or electrostatic forces Examples of electrostatic immobilization include the use of materials such as nitrocellulose and poly-lysine- or aminopropyl silane-coated surfaces.14-17 Protein microarrays were also fabricated by means of physical adsorption onto porous gel pads.18-

22 A major advantage of these non-covalent immobilization concepts is their ease of use Usually no protein modification is needed prior to imprinting onto the surface The disadvantage is that proteins often get denatured on these fairly undefined surfaces due non-specific interactions between the protein and the surface material On top of that, physical adsorption of proteins onto surfaces may also lead to de-adsorption of proteins during biochemical assays, which can lead to signal loss Covalent attachment of proteins onto NHS-activated glass surface, via nucleophiliic groups (-NH2, -SH, -OH) located on protein surface, has been described by MacBeath and Schreiber.23 Other surfaces such as epoxide surfaces have also been used to capture proteins covalently.24 However, in these cases, the immobilization is random which may lead to deactivation of the protein molecules on the array Ideally, the proteins should be site-specifically immobilized on the surface to obtain a homogeneous orientation Oriented immobilization was first reported

by Zhu et al who expressed more than 90 % of proteins encoded by the Saccharomyces

cerevisiae ORFs using a double-tagging system The yeast proteins were expressed

laboriously in the form of fusion proteins containing both histidine and GST tags before affinity purification through the glutathione (GSH) column These proteins were

subsequently immobilized on a single 25 x 75 mm Ni-NTA coated glass slide to generate

a ‘yeast proteome chip’.25 Unfortunately, this strategy of immobilization is extremely tedious and time consuming, requiring multiple steps of sample processing Moreover, protein immobilization via His-tag/Ni-NTA interaction was shown to be neither strong nor robust enough to withstand harsh wash condition, thereby limiting the downstream application of the protein microarray Avidin-biotin technology has gained much prominence in research due to the remarkable affinity between avidin (or streptavidin-its

bacteria relative from streptomyces avidinii) and biotin (vitamin H, 0.24kDa).26 With a Kd

of 10-15 M, avidin/biotin binding is the strongest non-covalent interaction known in nature Consequently, avidin/biotin systems have been exploited for a variety of diverse applications in modern biology including peptide and protein microarray technology.26-32

In a recent example, Peluso et al were able to site-specifically immobilized biotinylated

antibodies and antibody fragments onto streptavidin surfaces leading to an increase in sensitivity over random attachment in a microarray assay.32

An essential prerequisite for the success of avidin-biotin technology is the incorporation of biotin moiety into experimental system Historically, biotinylation of proteins has been carried out by standard bioconjugate techniques using biotin-containing chemicals This leads to random biotinylation of proteins and in many cases, the subsequent inactivation of some protein biological activities.33 Alternative techniques have been developed which allows for site-specific labeling of proteins with biotin 34,35 A stretch of amino acids sequences has been identified by Cronan for site-specific tagging of biotin to proteins.35 The covalent attachment of biotin to specific lysine residue on the tag

sequence was catalyzed by biotin ligase (EC 6.3.4.10), a 35.5 kDa monomeric enzyme

encoded by the birA gene36, in a 2 step reaction as follows;

Step 1: Biotin + ATP ↔ Biotinoyl-AMP + PPi

These sequences, however, are typically quite large (> 63 AAs) and thus may interfere with the biological activity of the proteins they are fused to Further optimization of these sequence tags revealed that smaller tags (15-30 AAs) may be used.37 In general, proteins

fused with these tag sequences could be biotinylated either in vitro or in vivo by biotin

ligase.38,39 Unfortunately, in vivo biotinylation of proteins catalyzed by biotin ligase is

often inefficient and cell toxicity is likely to occur due to decrease biotinylation of important endogenous proteins within the host cells.39 Recent advances in the field however, have partially rectified this problem, and at the same time unequivocally demonstrated numerous advantages associated with protein biotinylation in live cells.40 In

vitro biotinylation is used when in vivo expression of the soluble fusion protein is

insufficient This however, also faces with problems such as proteolytic degradation of tag sequences and inhibitory effects of commonly used reagents towards biotin ligase.38 To overcome these drawbacks, we have developed an intein-based system to incorporate biotin moiety exclusively at the C-terminus of protein

According to the central dogma of gene expression, genetic information flows from DNA to RNA through transcription, and is then translated and expressed as protein

However, more genetic information appears to be present within the chromosomal DNA other then the gene that actually encodes for the protein product These excess genetic information are known to be introns, which get excised post-transcriptionally during RNA splicing In 1990, two groups independently reported the existence of protein splicing elements that is capable of excising themselves post-translationally through a process analogous to RNA splicing.41,42 These protein “introns”, known as intein, are found within genes of other proteins and translated as a single polypeptide chain After translation, the intein initiates an autocatalytic event to excise itself and join the flanking host segments with a new peptide bond to form the final protein product (Figure 1).43 To date, over 100 inteins have been discovered in unicellular organisms from all three domains of life.44,45

They can be divided into four basic classes: (1) the bifunctional/maxi-inteins, which contain an endonuclease domain inserted into the splicing domain46,47; (2) the mini-inteins, which lack the endonuclease domain48-52; (3) the trans-splicing inteins, which is

splitted in the splicing domain and each precursor fragment is present as a different primary translation product53-57; and (4) the newly discovered alanine inteins, which contain a naturally occurring N-terminal alanine residue58

Figure 1 Mechanism of protein splicing In the initial step of protein splicing, a linear

ester/thioester intermediate is formed by an N-O or N-S acyl rearrangement at Ser1/Cys1

of intein Next, trans-thioesterification involving nucleophilic attack of the hydroxyl/thiol group of Ser1/Cys1 on the linear ester/thioester bond results in the formation of a branched intermediate Excision of the intein occurs by peptide bond cleavage coupled to

succinimide formation of the intein C-terminal asparagines The ligated exteins undergoes

a spontaneous O-N or S-N acyl rearrangement to form a stable peptide bond

The discovery of inteins and elucidation of its self-splicing mechanism has triggered the research of new applications and techniques for protein chemistry and engineering For example, through identification of the residues directly participating in the breakage and

peptide bond formation, Chong et al were able to engineer inteins with controllable

cleavage at single splice junctions.59 By fusing the chitin binding domain (CBD, 5kD) of

Bacillus circulans60 to one terminus of the intein, they developed an intein-mediated

affinity purification system that eliminates the use of protease, which may further complicate the downstream purification process Protein purification occurs within a single chitin beads packed column due the self-catalytic activity of the fused intein.61,62The ability of intein to generate C-terminal thioester and N-terminal cysteine protein during bond breakage also greatly expands the utility of native chemical ligation chemistry in protein engineering Intein-mediated protein ligation (IPL) is an extremely useful method for protein synthesis with a variety of peripheral applications.63-67 It has been used to incorporate noncoded amino acids into a protein sequences64, purify cytotoxic proteins65, study protein structure/function relationship by segmental isotopic labeling of proteins for NMR analysis66, and introduce fluorescent probes into a protein sequence67 Intein fusion system has also been employed to generate both complementary reactive groups on the same protein resulting in either inter- or intramolecular ligation, leading to multimeric or cyclic protein species, respectively.68,69 Trans-splicing between two foreign protein sequences enable in vitro fusion of two protein sequences via a simple

peptide bond thus creating a protein chimera with new added properties.70 The utility of IPL can be expanded to a wide range of proteomic application by a variety of functionalities, depending on the experimental requirements.71,72 Herein, we described a

novel and highly efficient approach for site-specific biotin-tagging of proteins using IPL (Figure 2)

H2N

HS SO3

-NH2HS

O

Intein

N-S acyl shift Thiol induced cleavage

Chemoselective reaction S-N acyl shift

Figure 2 Biotin-tagging of protein via IPL reaction Engineered intein, fused with the

protein of interest, catalyzed the formation of thioester The intein gets cleaved off in a thioester exchange reaction with a thiol compounds, such as 2-mercaptoethanesulfonic (MESNA), generating proteins with an active C-terminal thioester IPL occurs between the thioester protein and the cysteine-containing biotin tag resulting in a native peptide bond formation

In our work with 3 model proteins, namely MBP (Maltose Binding Protein), EGFP (Enhanced Green Fluorescent Protein) and GST, we demonstrated that site-specific biotinylation of proteins could be efficiently carried out by applying a cysteine-containing biotin tag (Figure 3) to the intein-fused protein purified and bound onto a chitin column

carried out in a single column, the eluted proteins can therefore be immobilized directly onto an avidin-coated glass slide to generate a protein microarray, without further purification step To further validate the feasibility of this biotinylation strategy for protein array and other high-throughput proteomic applications, we went on to examine the versatility of this biotin-tagging approach for many other different classes of yeast

proteins In addition, we also demonstrate for the first time, that this intein-mediated biotinylation strategy could be successfully implemented in both bacterial and mammalian

cells to generate in vivo biotinylated proteins The cell lysate containing the biotin-tagged

proteins were subsequently used to generate corresponding protein array in a single step

without further downstream processing (Method A in Figure 4) Beside cell expressed

recombinant proteins, intein-mediated biotinylation strategy may also be extended to

biotinylate proteins synthesized in a cell-free synthesis system (Method C in Figure 4).31

Figure 4 summarizes the 3 intein-mediated protein biotinylation strategies described

H2N

N H

H N SH

O

HN

NH O

Figure 3 Chemical structure of cysteine-biotin derivative

N N S

O

N H

H2N O HS

H2N

HS

=

Cell-free protein synthesis

Cytochemical localization studies.

Chitin column

Gene

Expression

Figure 4 Three intein-mediated protein biotinylation strategies: (A) in vivo biotinylation

in live cells; (B) in vitro biotinylation of column-bound proteins; & (C) cell-free

biotinylation of proteins

2 Materials and Methods

2.1 Chemical synthesis of the cysteine-biotin

Cysteine-biotin derivatives (Figure 3) can be synthesized with either with (1) protected, or (2) Fmoc-protected cysteine:

Boc-2.1.1 Using Boc-protected cysteine

N- -t-Boc-S-trityl-L-cysteine (1.2 g, 2.6 mmol), TBTU (1.0 g, 3.10 mmol), and HOBt (0.60 g, 3.9 mmol) were dissolved in 50 ml of dry DMF This mixture was stirred under argon for 20 min at room temperature before addition of 4-methyl morpholine (0.75 g, 7.8 mmol) and biotinylethylenediamine (0.75 g, 2.6 mmol) The reaction was stirred further

for 3 h, followed by evaporation in vacuo The crude product was dissolved in 200 ml of

CH2Cl2, extracted with 3 x 200 ml of H2O, dried over MgSO4, and concentrated in vacuo

Further purification was done by flash chromatography (4-8% MeOH in CH2Cl2) to give

the protected form of 1, which was deprotected by first stirring in a solution containing

trifluoroacetic acid (50 ml), H2O (1.6 ml), and triisopropylsilane (1.2 g, 7.8 mmol) for 30

min, and then evaporation in vacuo The resulting residue was taken in a mixture of 1:1

H2O/CH2Cl2 (200 ml), and the aqueous layer was extracted with 3 x 100 ml of CH2Cl2

before evaporation to dryness

2.1.2 Using Fmoc-protected cysteine

N-Fmoc-S-Trityl-L-cysteine (0.996 g, 1.7 mmol), TBTU (0.674 g, 2.1 mmol) and HOBt (0.3989 g, 2.6 mmol) were dissolved in 17 ml of DMF After stirring for 30 minutes

at room temperature, biotinylethylenediamine (0.5 g, 1.7 mmol) and triethylamine (0.515

g, 5.1 mmol) were added The reaction was carried out under nitrogen for 3 hours at room

temperature, followed by concentration in vacuo The resulting residue was dissolved in

ethyl acetate (50 ml), and extracted with 1.0 M HCl (50 ml), 10% Na2CO3 (50 ml), saturated NaCl (50 ml), dried over MgSO4, and then evaporated to dryness A solution of 20% piperdine in DMF (15 ml) was added to the resulting residue and stirred for 30 minutes at room temperature Following evaporation, the residue was dissolved in ethyl acetate and washed with 2 x 10% Na2CO3 (50 ml), saturated NaCl (50ml), dried over MgSO4, and then evaporated to dryness The residue was taken in 15 ml of TFA/EDT/H2O (9/0.5/0.5), stirred for 1 hour, and then evaporated to dryness The residue was taken in 100 ml of 1:1 DCM/H2O and insoluble solid was removed by filtration

2.1.3 Purification and identification of cysteine-biotin

Final purification of the product from both syntheses was done using HPLC with a C18 reverse-phase column to give the final product as a white solid (39% overall yield)

1H NMR (400 MHz, D2O) 4.57 (dd, 1H, J = 7.8, 5.0), 4.39 (dd, 1H, J = 7.8, 5.0), 4.12 (t, 1H, J = 5.4), 3.45 (m, 1H), 3.33-3.24 (m, 4H), 3.03 (dd, 1H, J = 14.9, 5.4), 3.00-2.93 (m, 2H), 2.74 (d, 1H, J = 13.2), 2.22 (t, 2H, J = 7.3), 1.72-1.50 (m, 4H), 1.48-1.31 (m, 2H);

13C NMR 179.62, 170.46, 64.53, 62.70, 57.79, 56.01, 42.16, 42.12, 41.45, 37.96, 30.39, 30.12, 27.50, 27.30; ESI 390.2 (MH+)

2.2 Cloning of target genes into pTYB1 & pTWIN expression vector

To construct intein fusion proteins, target gene fragments were first PCR amplified from pEGFP (CLONTECH), pGEX-4T1 (Pharmacia Biotech), and yeast ex-clones (Invitrogen) respectively PCR amplification for both EGFP and GST gene fragments

utilized upstream primers (5’-GGC GGC CAT ATG GTG AGC AAG GGC GAG-3’) & (5’-GGC GGC CAT ATG TCC CCT ATA CTA GGT-3’) containing an NdeI site with a translation initiation codon (ATG), and downstream primers (5’-GGC GGC TGC TCT

TCC GCA CTT GTA CAG CTC-3’) & (5’-GGC GGC TGC TCT TCC GCA GTC ACG

ATG CGG-3’) containing a SapI site, respectively A common upstream primer (5’-GGC GGC CAT ATG GAA TTC CAG CTG ACC ACC-3’) and individual downstream primers (5’-GGC GGC TGC TCT TCC GCA ACC ACC N15-18-3’) were used to amplify the yeast gene fragments from the Yeast ExClonesTM, and at the same time introduce 2 extra Gly residues to the C-terminus of the yeast gene A standard PCR mixture contained 1x HotStarTaq DNA polymerase buffer (Qiagen), 0.2 mM of each dNTPs (NEB), 0.5 µM of each primer, 100 ng of plasmid DNA template and 2 units of HotStarTaq DNA polymerase (Qiagen) Amplification was carried out with a DNA Engine™ thermal cycler (MJ Research) at 94 °C for 45 sec, 65 °C for 45 sec and 72 °C for 1 min, for 25 cycles for the EGFP and GST gene fragments, and at 94°C for 45 sec, 55°C for 45 sec and 72°C for

2 min, for 25 cycles for the yeast gene fragments The PCR products were then cloned into pCR2.1-TOPO using TOPO TA cloning kit (Invitrogen) prior to double digestion with

NdeI and SapI (NEB) Digested EGFP, GST and yeast gene fragments of correct sizes

were gel-purified and cloned into either pTYB1 or pTWIN expression vector (NEB,

USA), via NdeI and SapI sites to yield the intein-fused constructs The C-terminal residue

of GST in pTYB1-GST-intein was site-mutagenized from Cys to Gly using using QuickChange™ XL Site–Directed Mutagenesis Kit (Stratagene) with upstream primer

(5'-CGG CCG CAT CGT GGG TGC TTT GCC AA-3’) and downstream primer (5'-TT GGC AAA GCA CCC ACG ATG CGG CCG-3'); Gly is underlined in the primers The

pTYB1 construct containing the MBP gene, pMYB5, is commercially available (NEB) The resulting T7-driven expression plasmids, shown to be free of mutation by automated

DNA sequencing (Applied Biosystems), were then transformed into E.coli ER2566 host

(NEB) for protein expression

2.3 Site-directed mutagenesis of pTYB1-wtEGFP (Lys239 )-intein

The C-terminal residue of wtEGFP in pTYB1-wtEGFP (Lys239)-intein was mutagenized from the original Lys239 to the other 19 amino acids using QuickChange™

site-XL Site–Directed Mutagenesis Kit 19 sets of primers, each containing a primer (5'-GAC GAG CTG TAC NNN TGC TTT GCC AA-3’) and a complementary primer (5'-TT GGC AAA GCA N’N’N’ GTA CAG CTC GTC-3'), were used, in which NNN (and N’N’N’) in each set of primers represents an amino acid to which Lys239 in pTYB1-wtEGFP (Lys239)-intein was replaced Upon confirmation by DNA sequencing, the mutated plasmids (e.g

pTYB1-mutEGFP (AA239)-intein, where AA represents a corresponding mutated amino

acid) were transformed into ER2566 E coli EGFP (Asp239)-intein and EGFP (Cys239

)-intein were also cloned into pTYB-2 vector via NdeI and SmaI site based on ImpactTM-CN protocols (NEB)

2.4 Expression of intein-fused proteins

The transformed E coli host was grown in Luria Bertani (LB) medium supplemented

with 100 µg/ml ampicillin at 37 °C in a 250 rpm shaker to an OD600 of about ~0.5 Protein expression was induced overnight at room temperature using 0.3 mM isopropyl thiogalactosidase (IPTG) Upon harvest (4000 rpm, 15 min, 4°C), cells were resuspended

in lysis buffer (20 mM Tris-HCl pH 8.0, 0.5 M NaCl, 1 mM EDTA, 1 % CHAPS, 1 mM TCEP and 1 mM PMSF) and lysed by glass beads (Sigma) The cell debris was pelleted down by centrifugation (20,000 × g, 30 min, 4 °C) to give a clear lysate ready for loading onto a column packed with chitin affinity resin (NEB, USA) for purification and biotinylation

2.5 Affinity purification & C-terminal biotinylation of recombinant proteins

Microspin columns were pre-packed with 100 µl of chitin resin and pre-equilibrated with 1 ml of column buffer (20 mM Tris-HCl pH 8.0, 500 mM NaCl and 1 mM EDTA)

To purify the fusion protein, the clarified cell lysate was incubated on the column for 30 min at 4 ˚C with gentle agitation to ensure maximum protein binding Unbound impurities were then washed away with 2 ml of column buffer To biotinylate recombinant proteins,

200 µl of the column buffer containing 50 mM MESNA (Sigma) and 5 mM biotin was passed through the column to distribute it evenly throughout the resin before the flow was stopped and the column was incubated at 4°C overnight The resulting biotinylated protein was eluted with 100 µl of column buffer, and analyzed by 12 - 15% SDS-PAGE gel Resin-bound proteins were analyzed by first boiling the resin with DTT-free SDS-PAGE loading buffer, then separated by SDS-PAGE Silver or coomassie

cysteine-staining of the gel was done to visualize the separate proteins bands Premature in vivo

cleavage and on-column cleavage of the intein-fusion was determined from the stained SDS-PAGE gel To determine the ratio between the biotinylated and the non-biotinylated protein in the eluted fraction, an absorption experiment with streptavidin beads was performed The eluted fraction was first incubated with excessive Streptavidin

MagneSphere® Paramagnetic Particles (Promega) for 1 h at 4 oC to ensure all biotinylated proteins were absorbed onto the beads Both eluents, before and after streptavidin adsorption, were then analyzed by SDS-PAGE Western blots with horseradish peroxidase (HRP)-conjugated anti-biotin antibody (NEB) and the Enhanced ChemiLuminescent (ECL) Plus kit (Amersham) were performed to confirm the presence

of biotin-tagged proteins

2.6 SPR analysis

All SPR experiments were performed with a BIAcore X instrument (Biacore) Biotinylated MBP was prepared as described above Surface activation of the CM5 sensor chip (Biacore) was done using standard amino-coupling procedures according to manufacture’s instructions 1.75 µg of avidin in 10 nM acetate (pH 4.5) and 0.125 M NaCl was passed over the activated chip surface Excess reactive groups were then deactivated with 1 M ethanolamine hydrochloride (pH 8.5) before injection of 35 µl biotinylated MBP (10 µg/ml) to the avidin-functionalized surface Subsequently, 10 µl of anti-MBP antibody (0.1 mg/ml) was injected at a flow rate of 1 µl/min to confirm the immobilization of MBP onto the chip surface 10 mM HCl was used to regenerate the chip surface before subsequent rounds of antibody injections The Kd of the anti-MBP/MBP binding was determined by BioEvaluation software installed on the BIAcore X

2.7 In vivo protein biotinylation in E coli

For in vivo biotinylation of proteins in E coli., pMYB5 and pTYB-1 constructs

containing two yeast proteins (YAL012W & YGR152C) were used Liquid cultures of

ER2566 carrying the intein-fusion construct were grown to OD600 of ~0.6 in LB medium supplemented with 100 µg/ml of ampicillin Expression of MBP and yeast protein fusions was induced with 0.3 mM IPTG at room temperature overnight MESNA and cysteine-biotin were subsequently added to final concentrations of 10 mM and 5 mM, respectively Other concentrations of MESNA/cysteine-biotin were also tested but these conditions

gave the best in vivo bintinylation efficiency while maintaining viability of the cells In

vivo biotinylation was allowed to proceed overnight at 4˚C with gentle agitation Cells

were harvested and washed thoroughly with PBS to remove excess biotin before lysed with glass beads Clear lysates containing the desired biotinylated proteins were collected by centrifugation, and used without further purifications The

MESNA/cysteine-entire process was monitored by SDS-PAGE and western blots with anti-MBP In vivo

protein biotinylation was unambiguously confirmed with HRP-conjugated anti-biotin

antibody Additionally, to confirm the affinity of the in vivo biotinylated protein towards

avidin/streptavidin and to determine the ratio of the biotinylated/non-biotinylated proteins

generated in vivo, an absorption experiment with streptavidin beads was performed Clear

cell lysates were incubated with Streptavidin MagneSphere® Paramagnetic Particles (Promega) at 4°C for 30 min The beads were then thoroughly washed with PBS to remove unbound proteins, and subsequently analyzed by boiling in SDS-PAGE loading buffer, then resolved on a 12% SDS-PAGE gel, followed by immunoblotting with HRP-conjugated anti-biotin antibody Cell lysates before and after streptavidin absorption were also separated on a 12% SDS-PAGE gel followed by western blots with anti-MBP and anti-biotin antibodies

2.8 In vivo protein biotinylation of in mammalian cells

EGFP-intein was cloned into pT-Rex-DEST30 (Invitrogen) mammalian expression

vector by Gateway™ cloning technology EGFP-Sce VMA intein-CBD was amplified from pTYB1-wtEGFP-intein using upstream primer (5’-GGGG ACA AGT TTG TAC

AAA AAA GCA GGC TTC GAA GGA GAT AGA ACC ATG GTG AGC AAG GGC GAG GAG-3’) and downstream primer (5’-GGG GAC CAC TTT GTA CAA GAA AGC TGG GTC TCA TTG AAG CTG CCA CAA GGC -3’), where the underline nucleotides

represent the attB recombination sites The amplified EGFP-Sce VMA intein-CBD gene

was first cloned into the pDONRTM 201 donor vector then to the final pT-Rex-DEST30 destination vector using BP and LR ClonasesTM Mix (Invitrogen) The final expression plasmid was transfected into HEK 293 cells, grown in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum, penicillin (100 units/ml) and streptomycin (100 µg/ml), using PolyFect Transfection Reagent (Qiagen) The mammalian cells were seeded at 2.4 x 106 cells per 100 mm tissue culture plate the day before transfection After 48 h of transient expression, the culture medium was changed

to DMEM containing 10 mM MESNA and 1 mM cysteine-biotin, and further incubated at

37 °C overnight These biotinylation conditions were optimized to ensure cell viability and maximum biotinylation efficiency Mammalian cells were then harvested, washed thoroughly with PBS to remove excess biotin, and lysed by glass beads The entire biotinylation process was monitored by SDS-PAGE and western blots with anti-EGFP The biotinylated protein in the mammalian cell lysates was purified using Streptavidin MagneSphere® Paramagnetic Particles before unambiguously confirmed by immunoblotting using HRP-conjugated anti-biotin antibody as described earlier

2.9 Generation of protein microarray

Glass slides were cleaned in a piranha solution and derivatized with a 1% solution of 3-glyicidoxypropyltrimethoxisilane (95 % ethanol, 16 mM acetic acid) for 1 hr and cured

at 150 °C for 2 hours The epoxy slides were reacted with a solution of 1 mg/ml avidin in

10 mM NaHCO3 for 30 minutes, washed with water, air dried, and the remaining epoxides were quenched with a solution of 2 mM aspartic acid in a 0.5 M NaHCO3 buffer, pH 9 Trace amount of cysteine-biotin in the eluted protein sample, from on-column

biotinylation, and the clarified cell lysate, from in vivo biotinylation, did not seem to affect

the spotting quality, as NAP-5 treated protein samples did not seem to improve the array quality Therefore, protein samples from both sources can be directly spotted onto the avidin-functionalized slides using an ESI SMA arrayer (Toronto), without any additional purification step No incubation was necessary before the slide were further processed by washing with PBS and drying in air Sequence specific monoclonal antibodies, anti-EGFP (Clontech) and anti-MBP (Santa Cruz Biotechnology), were labeled with Cy3-NHS (λEx = 548 nm; λEm = 562 nm) and Cy5-NHS (λEx = 646 nm; λEm =

664 nm)(Amersham Biosciences) respectively The antibody was reacted with the dye for one hour in 0.1 M NaHCO3, pH 9, according to manufacturer’s protocols and purified with a NAP5 column (Amersham Pharmacia) The anti-GST was purchased as a FITC-conjugate (λEx = 490 nm; λEm = 528 nm)(Molecular Probes) The spotted slides were incubated with the labeled antibody (or mixture of antibodies) for 1 hour, washed 4 times, each time for 15 min with PBST (PBS + 0.1 % Tween 20), dried and scanned with an ArrayWoRx microarray scanner (Applied Precision) To show selective binding of glutathione to GST on the protein array, the N-terminal amine group of glutathione was

first labeled with Cy3-NHS by reacting the molecule overnight with the dye in sodium phosphate buffer at pH 7 The reaction was subsequently quenched with ethanolamine for

12 hours to degrade the remaining Cy3-NHS, and any glutathione labeled at its cysteinyl thiol Avidin slides, immobilized with biotinylated GST as described earlier, were incubated with the Cy3-labeled glutathione for 1 hour, and washed with PBST Finally, the slides were dried and specific binding between GST and glutathione was visualized with the microarray scanner

2.10 Cell free synthesis and biotinylation of MBP

The pMYB5 plasmid was used as the DNA template in the Rapid Translation System

(RTS) 100 E coli HY kit (Roche) for cell-free protein synthesis Based on the

manufacturer’s protocol, the reaction was performed at 30˚C for 4 h, based on the manufacturer’s protocol in a 25 µl reaction with 500 ng DNA as the template At the end

of protein synthesis, MESNA and cysteine-biotin were added to the lysate to final concentrations of 50 mM and 5 mM, respectively, to induce cleavage/biotinylation of MBP at 4°C overnight Cell lysates were precipitated with acetone and analyzed by SDS-PAGE Biotinylation of MBP was unambiguously confirmed by western blots with HRP-

conjugated anti-biotin antibody

3 Results and Discussion

3.1 General features of pTYB expression vectors

pTYB vectors are commercially available from New England Biolabs for expression and isolation of proteins processing a C-terminal thioester The target gene is inserted into the polylinker region of each vector such that the C terminus of the target protein is fused

in-frame to the N terminus of the Sce VMA intein (from Saccharomyces cerevisiae

VMA1 gene) Transcription of the fusion gene is initiated from the pTYB T7 promoter73

under the tight control of a lac operon Binding of the lac repressor (encoded by the lac I gene in the same vector) to the lac operator sequences immediately downstream of the T7

promoter, suppresses basal expression of the fusion gene in the absence of IPTG induction A T7 transcription terminator is located downstream of the CBD to prevent continued transcription pTYB vectors also carries an ampicillin resistance gene (Ampr)

for selection of transformed host strain Both pTYB1 and pTYB2 contain an NdeI site for cloning the 5’ end of a target gene The ATG codon of the NdeI site is used to initiate

translation of the fusion protein The only difference between the two vector lies within the 3’ end restriction site, just before the start of the intein gene pTYB1 and pTYB2 contains

SapI and SmaI sites at their 3’ ends, respectively (Figure 5) The use of SapI site in

pTYB1 allows the C-terminus of the target protein to be fused directly next to the intein

cleavage site, whilst the use of SmaI site in pTYB2 adds an extra glycine residue to the terminus of the target proteins (Figure 6A & B)

pTYB 2

7474 bp

Figure 5 Map and multiple cloning site (MCS) of pTYB1 & pTYB2 expression vectors

The only difference between the two expression vectors is highlighted in red

A

CAT ATG

GTA TAC

Sap I Nde I

TGC GGA AGA GCA ACG CCT TCT CGT

Gene of Interest

Cloned into pTYB1

via NdeI and SapI sites

Gene of Interest TGC TTT ACG AAA

GTA T AC

CysMet

Start site for translation

of intein-fusion proteins

Sce intein tag

Cloning of gene fragment into pTYB1 expression vector

CAT ATG GTA TAC

Nde I

Gene of Interest

Cloned into pTYB2

via NdeI and SmaI sites

Start site for translation

of intein-fusion proteins

Sce intein tag

Cloning of gene fragment into pTYB2 expression vector

Gly

B

Figure 6 Cloning of gene fragment into pTYB1 & pTYB2 (A) After digestion with NdeI

and SapI, the target gene fragment and pTYB1 (nucleotide sequences in blue) were ligated

to regenerate the NdeI site with the translation initiation codon (ATG), and codon for Cys1 (TGC) at the intein N-terminus SapI site is not regenerated after cloning No extra

vector-derived amino acid residue is added to the native sequences of the target protein

after intein cleavage (B) To clone into SmaI site of pTYB2, the target gene need not have

a SmaI site at its 3’end PCR with certain proofreading polymerase would generate a 3’ blunt end which can be ligated to SmaI-digested (blunt end) pTYB2 (nucleotide sequences

in pink) This results in an extra glycine residue added to the C-terminus of the target protein indicates the intein cleavage site

3.2 Intein-Mediated Biotinylation of three model proteins

3.2.1 Cloning of target genes into pTYB1 expression vector

In the proof-of-concept experiment, the gene fragments of three model proteins, namely MBP, EGFP and GST were cloned into pTYB1 to generate thioester-tagged proteins for biotinylation Restriction enzyme digestion of PCR product is often less efficient than releasing a fragment from a vector and may result in lower cloning efficiency Consequently, the PCR fragment of target gene was first cloned into a T-vector

to facilitate the cloning process.74 To add a single deoxyadenosine to the 3’ end of PCR

product, the target gene sequence had to be PCR-amplified using HotStar Taq polymerase

with proofreading ability The PCR product was then ligated to the corresponding vector containing 3’ deoxythymidine overhangs The choice of restriction sites in the primers determines the extra amino acids residues that may be attached to the target protein after intein cleavage Therefore to obtain target protein with no extra vector-

T-dervied residues, we decided to clone the target gene between the NdeI and SapI sites in pTYB1 (Figure 6A) NdeI and SapI restriction sites, absent in the target gene, are

incorporated into the forward and reverse primers, respectively The TA clone containing

the PCR fragment was double digested with NdeI and SapI and the target gene fragment, isolated by agarose gel electrophoresis, was then ligated to the NdeI/SapI digested pTYB1

Clones containing the target gene insert were identified by restriction digestion and colony PCR The final expression plasmid, verified by DNA sequencing, was transferred to

ER2566 for protein expression This E.coli stain carries a chromosomal copy of the T7 RNA polymerase gene, under the control of the lac promoter In the presence of IPTG,

expression of T7 RNA polymerase is activated which in turn initiate the transcription of the fusion gene

3.2.2 Expression and extraction of fusion proteins

Expression level of fusion protein from the pTYB vector is greatly influence by: 1) bacterial cell line, 2) nature of the fusion protein and 3) induction condition (temperature,

duration and IPTG concentration) ER2566 is the E.coli strain supplied by NEB for

expression of fusion protein from a pTYB vector but other commercially available strain (e.g BL21) may be tested for optimal expression level of the fusion protein Different induction conditions (e.g 30°C for 3hrs, 20-25°C for 6-16hrs or 12-16°C for overnight) were tested out for the 3 fusion proteins (MBP-intein-CBD, EGFP-intein-CBD, GST-intein-CBD) to optimize expression of soluble fusion protein and minimize proteolysis After expression, the bacterial cells were harvested and lysed in simple lysis buffer,

containing Tris-HCL & NaCl & EDTA, using glass beads Beside glass beads, the E.coli

cells can also be broken either by sonication or french press or freeze-thawing method The type of mechanical lysis method used greatly depends on the amount of lysis buffer resuspending the induced bacterial cells Sonication and french press method is more efficient but requires large volume of suspension for lysis Lysozyme is not the preferred cell lysis method for extracting protein fused to an intein tag, since it is known to bind and digest chitin beads However, if no alternative method is available, low level of lysozyme

can still be used (incubate at 4 ºC for 1hr) for cell lysis

3.2.3 Affinity purification and on-column biotinylation

The intein-fused proteins were purified and biotinylated, in a single step, by first loading the clarified cell lysate onto a column pe-packed with chitin beads, then flushing the column with MESNA and cysteine-biotin, to obtain the C-terminally biotinylated proteins The affinity purification process was monitored by SDS-PAGE with coomassie and silver stain Figure 7 shows the SDS-PAGE result of MBP purification through the chitin column The full-length fusion precursor (97 kDa) and a small amount of the cleaved intein tag (55 kDa) were found to bind to the chitin resin after the lysate was passed through the column (Figure 7, lane 6) MBP was co-eluted with minute amount of contaminating proteins after the thiol-induced cleavage and its purity was estimated to be about 95 % (Figure 7, lane 8 - 9) The high affinity of CBD for the chitin beads has allowed a better recovery of the fusion protein from the crude extract and the use of stringent wash conditions (e.g high salt concentration and detergent) to reduce non-specific binding while increasing purity of the eluted MBP About 2 mg of MBP was yielded from a 200 ml of bacteria cell culture Less than 5 % of the intact fusion protein was found to remain bound on the chitin beads after the cleavage indicating the

effectiveness of on-column cleavage with MESNA (Figure 7, lane 10) Biotinylation of

the eluted MBP was unambiguously confirmed by western blotting as shown in Figure 8A

Immunoblot result indicates specific biotin-tagging of the affinity purified MBP in the presence of cysteine-biotin No biotinylation was observed for MBP eluted in the absence cysteine-biotin derivatives Streptavidin adsorption experiment was used to determine the on-column biotinylation efficiency with respect to the total amount of MBP eluted (Figure

8B) More than 95 % of the eluted MBP were adsorbed to the streptavidin matrix,

suggesting most eluted proteins were biotinylated following cysteine-biotin/MESNA

treatment The on-column biotin-tagging process is highly efficiency (> 95% efficiency), hence, we equate cysteine-biotin/MESNA-induced cleavage efficiency of a target protein

to its biotinylation efficiency for subsequent experiment Among the 3 fusion proteins, only a small amount of the intact GST(Asp719) -intein-CBD fusion protein was detected on the chitin beads after cell extraction Most of the fusion proteins were cleaved prematurely within the bacterial cells leaving mostly the intein-CBD tag to bind onto the affinity beads

during the purification process We eventually found out that the high in vivo cleavage of

GST(Asp719) -intein-CBD was mainly due to the C-terminal aspartic acid (Asp) of GST, which will be further explained in the later section of this report To minimize pre-mature cleavage of GST-intein-CBD inside the bacterial cells, we mutate the C-terminal Asp residue of GST to glycine (Gly) by PCR-based site-directed mutagenesis The mutant pTYB1-GST(Gly719)-intein construct was then transformed back to ER2566 for protein expression SDS-PAGE gel of the purification process showed higher amount of GST(Gly716)-intein-CBD binding to the chitin beads, resulting in higher yield of the biotinylated GST (data not shown) Lastly, to generate corresponding protein array, eluted protein fractions were spotted directly onto an avidin-functionalized slide, without further downstream processing

kDa

MBP Intein-CBD MBP-intein-CBD

Figure 7 Affinity purification of MBP Lane 1, Prestained protein marker (BioRad); Lane

2, uninduced cell extract; Lane 3, induced cell extract; Lane 4, flow through from the load, Lane 5, flow through from column wash; Lane 6, proteins bound to chitin column before MESNA cleavage Lane 7, flow through from quick MESNA flush; Lane 8-9, first two fraction of the elution after 4 °C overnight incubation with MESNA & cysteine-biotin; Lane 10, proteins bound to chitin column after MESNA cleavage SDS-PAGE gel was stained with silver nitrate

A

MBP MBP

Coomassie

t i Anti-biotin Blot

B

Coomassie

t i Anti-Biotin blot

MBP MBP

1 2 3

Figure 8 On-column biotinylation of MBP (A) MBP eluted from the chitin column was

subjected to SDS-PAGE and visualized by coomassie stain Biotinylation of the eluted MBP was confirmed by immunoblot using anti-biotin antibody Lane 1, MBP eluted with

MESNA only; Lane 2, MBP eluted with cysteine-biotin/MESNA (B) The eluted MBP

was incubated with streptavidin magnetic beads to assess the degree of biotinylation Samples before and after streptavidin adsorption was ran on a SDS-PAGE gel and quantitated via the coomassie blue staining intensity Anti-biotin blot was used to check for the presence of biotinylated MBP Lane 1, amount of MBP before streptavidin absorption; Lane 2, amount of MBP after streptavidin absorption; Lane 3, MBP bound on streptavidin beads