Establishment of Intein-Mediated Protein Ligation under Denaturing

conditions, of target proteins that are expressed insolubly as intein fusion proteins. ... antibody-intein fusion protein under denaturing conditions ...
0 downloads 0 Views 157KB Size
Download PDF
Bioconjugate Chem. 2002, 13, 707−712

707

Establishment of Intein-Mediated Protein Ligation under Denaturing Conditions: C-Terminal Labeling of a Single-Chain Antibody for Biochip Screening Jens R. Sydor, Maria Mariano, Steve Sideris, and Steffen Nock* Zyomyx, Inc., 26101 Research Road, Hayward, California 94545. Received March 26, 2002; Revised Manuscript Received June 3, 2002

Intein-mediated protein ligation is a recently developed method that enables the C-terminal labeling of proteins. This technique requires a correctly folded intein mutant that is fused to the C-terminus of a target protein to create a thioester, which allows the ligation of a peptide with an N-terminal cysteine (1, 2). Here we describe the establishment of this method for the labeling, under denaturing conditions, of target proteins that are expressed insolubly as intein fusion proteins. A GFPuv fusion protein with the Mycobacterium xenopi gyrA intein was expressed in inclusion bodies in Escherichia coli and initially used as a model protein to verify intein cleavage activity under different refolding conditions. The intein showed activity after refolding in nondenaturing and slightly denaturing conditions. A construct of the same intein with an anti-neutravidin single-chain antibody was also expressed in an insoluble form. The intein-mediated ligation was established for this single chain antibody-intein fusion protein under denaturing conditions in 4 M urea to prevent significant precipitation of the fusion protein during the first refolding step. Under optimized conditions, the single-chain antibody was labeled with a fluorescent peptide and used for antigen screening on a biochip after final refolding. This screening procedure allowed the determination of binding characteristics of the scFv for avidin proteins in a miniaturized format.

INTRODUCTION

Inteins are self-splicing protein elements that excise themselves out of a precursor protein, ligating the flanking polypeptide sequences together. Since their discovery in 1990 (3, 4), inteins have been used for a variety of protein engineering techniques including the C-terminal ligation of proteins to peptides or protein fragments (5-8). In this case, the C-terminus of a target protein is fused to an intein mutant that is unable to splice itself out of the precursor protein, but is still capable of generating a thioester at the C-terminus of the target protein. An N-terminal cysteine-containing peptide or protein fragment is then ligated to the Cterminal thioester of the target protein (Figure 1). This is usually achieved by the aid of another thiol compound that releases the target protein from the intein in a transthioesterification reaction. This C-terminal thioester is then reacted with a peptide containing an N-terminal cysteine, the thiol of which exchanges with the low molecular weight thiol compound. Finally, the thioester formed between the target protein and the peptide spontaneously rearranges to generate a native amide bond (1, 2). This technique has proven useful in protein engineering to incorporate unnatural modifications such as fluorophores or isotope labels into proteins (9-11). An important prerequisite for intein-mediated protein ligation (IPL) is that the fusion protein is in a correctly folded conformation to allow the intein-mediated formation of the thioester. Unfortunately, a significant number of proteins, including those fused to inteins, can only be obtained in insoluble form in inclusion bodies when * To whom correspondence should be addressed: Zyomyx, Inc., 26101 Research Road, Hayward, CA 94545. Tel: +1-510266-7504, fax: +1-510-266-7792, e-mail: [email protected].

Figure 1. Schematic description of the intein fusion constructs and the intein-mediated protein ligation.

expressed in E. coli. The protein isolation from inclusion bodies generally requires a denaturation and refolding step. So far, there are only a limited number of examples that describe the refolding of intein fusion proteins into active conformations. For example, the refolding of intein fusions into nondenaturing buffer has been performed, although with a low yield of refolded protein and intein cleavage activity (12). Additionally, low concentrations of denaturants such as guanidine hydrochloride have been added to increase the solubility of peptides during the ligation without affecting intein activity (10). A general method that could be applied to a variety of

10.1021/bc025534z CCC: $22.00 © 2002 American Chemical Society Published on Web 06/28/2002

708 Bioconjugate Chem., Vol. 13, No. 4, 2002

proteins expressed as insoluble intein fusions would, therefore, expand the applicability of the IPL. One application of this technique could be the Cterminal site-specific labeling of single-chain antibodies, which are very important binding reagents used for the detection of biomolecules (13-15). The expression and isolation of active single-chain antibodies is usually problematic, however, due in part to the reducing environment in the cytoplasm, which inhibits the formation of the disulfide bonds within these antibody fragments. Production of active scFv’s1 is usually accomplished by expressing them in the periplasm, which is an oxidizing environment, or in inclusion bodies, with a subsequent refolding and oxidation step (16-19). The amount of scFv’s isolated from the periplasm is usually low, however, and refolding of scFv’s from inclusion bodies is usually unsuccessful. To overcome these limitations, we describe a method that results in a high yield of functional scFv’s that have been labeled by IPL. The basis of this method is to first express the scFv as an insoluble intein-fusion protein. The protein is then fully resolubilized and denatured, after which it is transferred into a mildly denaturing buffer in which the intein is functional, but where the denaturant allows the partially folded scFv to remain soluble. Correctly folded fusion proteins then undergo IPL, becoming labeled in the process. IPL under mildly denaturing conditions was tested on two unrelated proteins fused to the gyrase A intein from Mycobacterium xenopi. We also describe the utilization of scFv’s with a C-terminal fluorescent tag for antigen screening in a biochip format. This strategy also offers a variety of other possibilities for the C-terminal labeling of scFv’s, such as the introduction of other unnatural modifications for oriented capture agent immobilization on surfaces. EXPERIMENTAL PROCEDURES

Materials. For the cloning procedures, oligonucleotides were supplied by MWG (High Point, NC) and enzymes and chitin beads by New England Biolabs (Beverly, MA). Denaturants, urea and guanidine-hydrochloride, were purchased from Fisher Scientific (Pittsburgh, PA). Roche Molecular Biochemicals was the supplier of the 1,4-dithiothreitol (Inidianapolis, IN). Ethanethiol was purchased from Fluka (Milwaukee, WI). 2-Mercaptoethanesulfonic acid, sodium salt (MESNA), and all other inorganic salts were obtained from Sigma Chemical Company (St. Louis, MO). Pierce Chemical Company (Rockford, IL) was the supplier of neutravidin, avidin, and streptavidin. Cloning and Expression of Intein Fusion Constructs. The DNA sequence of the target proteins was cloned into pTXB1 (New England Biolabs, Beverly, MA) that was modified by the introduction of a His-tag 1 Abbreviations: Boc, tert-butyloxycarbonyl; BSA, bovine serum albumin; CBD, chitin-binding domain; DMF, dimethylformamide; EDTA, ethylenediaminetetraacetic acid; ELISA, enzyme-linked immunosorbent assay; Fmoc, 9-fluorenylmethoxycarbonyl; GFPuv, variant of the green fluorescent protein optimized for higher bacterial expression and higher fluorescence when excited by UV light; GuHCl, guanidine hydrochloride; His6, Histidine-tag consisting of six histidines; HRP, horseradish peroxidase; IPL, intein-mediated protein ligation; MESNA, 2-mercaptoethanesulfonic acid; OPD, O-phenylenediamine; PBS, phosphate-buffered saline; PLL-g-PEG-B30%, poly(L-lysine)-grafted-poly(ethylene glycol) with 30% biotin content; scFv, single-chain variable fragment of an antibody; TAMRA, 5-(and 6)-carboxytetramethylrhodamine.

Sydor et al.

between the intein and the chitin-binding domain as well as SfiI and NotI restriction sites in the MCS for cloning of the scFv or GFPuv in frame with the Mxe gyrA intein (Figure 1). The sequence for GFPuv was obtained from Clontech (Palo Alto, CA), and the sequence for the scFv was published elsewhere (20). Transformation of this plasmid into E. coli BL21(DE3) was done by electroporation according to the manufacturer’s instructions (Stratagene, La Jolla, CA). BL21(DE3) cells contain a chromosomal copy of the T7 RNA polymerase gene under the control of the IPTG-inducible lacUV5 promoter. Expression was induced by the addition of 0.1 mM IPTG. Six hours after induction and incubation at 30 °C, cells were harvested by centrifugation at 4 °C, which yielded ∼8 to10 g of wet cell pellet per liter cell culture. The cells were stored at -80 °C until further use. Synthesis of the myc Peptide with and without Rhodamine Label. Solid-phase peptide synthesis (21) was performed manually or on a CS536 peptide synthesizer from CSBio (San Carlos, CA), using in situ neutralization/2-(1H-benzotriazol-1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate activation protocols for stepwise Boc-chemistry chain elongation on an MBHA resin (22). The rhodamine derivative 5-(and 6)-carboxytetramethylrhodamine (TAMRA) was coupled to the lysine side chain with the following procedure. The -amino group of the lysine side chain was protected with an Fmoc group. After chain elongation, the amino group was deprotected by two 5 min incubations of the resin with 20% piperidine in DMF and subsequent washing with DMF. TAMRA was then activated and coupled like other amino acids, described in the general protocol (22). After HF cleavage, the peptides were purified with a C18column from Vydac (Hesperia, CA) by using linear gradients of buffer B (acetonitrile, 0.1% trifluoroacetic acid) in buffer A (H2O, 0.1% trifluoroacetic acid) and UV detection at 215 nm. Isolation and Solubilization of Inclusion Bodies. Cells were resupended in cell lysis buffer (200 mM NaCl, 20 mM sodium phosphate, pH 7.3, 1 mM EDTA) and lysed in a microfluidizer (Microfluidics International Corporation, Newton, MA). The crude inclusion body (IB) fraction was isolated by centrifugation at 31000 g for 10 min at 4 °C and resuspended in 0.1 M Tris, pH 7, 1mM EDTA. A 0.5 volume of 60 mM EDTA, 6% Triton X-100, 1.5 M NaCl, pH 7, was added and the suspension incubated at 4°C for 30 min with stirring. The IB fraction was isolated by centrifugation as described above and resuspended in wash buffer (50 mM sodium phosphate, pH 7.3, 1 mM EDTA). Centrifugation was repeated to isolate the IB pellets, which were subsequently stored at -20 °C (2 weeks) or -80 °C (>2 weeks). From a 1L culture, 1.5-2 g of IB were typically obtained. Inclusion bodies were solubilized in wash buffer plus 6 M GuHCl and 100 mM DTT for 2 h at RT with constant stirring. The pH was adjusted to 3-4 by the dropwise addition of 1 N HCl. Finally, the sample was centrifuged at 10000 g for 30 min at 4 °C to remove the insoluble cellular debris from the solubilized protein, which was then filtered. Partial Refolding and IPL under Denaturing Conditions. The solubilized protein sample (10 mg/mL) in 6 M GuHCl was rapidly diluted at a rate of 0.1 mL/ min into refolding buffer (wash buffer plus 4 M urea) at RT to give a final 1:10 dilution. The diluted sample was dialyzed against refolding buffer overnight at RT and then centrifuged at 10000 g for 30 min at 4 °C. The target protein-MxegyrA-His6-CBD fusion in the supernatant was purified by binding to a chitin resin at RT and

Intein-Mediated Ligation under Denaturing Conditions

washing with 5 column volumes of refolding buffer. The IPL was induced by the addition of 1 volume of 200 mM MESNA and 4 mM of peptide to 1 volume of the chitin beads containing the intein fusion protein in the same buffer and incubated for 36 h at RT. The supernatant was isolated, and the beads were washed with the refolding buffer. The wash fractions were combined with the supernatant, and the sample was dialyzed using one of the following methods: (1) directly into PBS with 10% glycerol, (2) directly into PBS, 10% glycerol, 3 mM/0.3 mM reduced/oxidized glutathione and then into PBS, 10% glycerol, or (3) stepwise dialysis into 2 M urea, 50 mM sodium phosphate, pH 7.3, 1 mM EDTA, 3 mM/0.3 mM reduced/oxidized glutathione, and then into 1 M urea in the same buffer and finally into PBS, 10% glycerol. Determination of scFv Activity by ELISA. The wells of a Nunc-Immuno Plate Maxisorp Surface (Nalge Nunc International, Denmark) were coated with 100 µL of 2 µg/mL neutravidin (or 5 µg/mL BSA as a negative control) in 0.1 M sodium carbonate, pH 9.6, overnight at 4 °C. The wells were then blocked with 200 µL of 3% w/v nonfat dried milk in PBS, 0.1% v/v Tween-20, followed by the addition of 100 µL of scFv-myc in PBS, 0.1% v/v Tween-20, and incubated for 1 h at 37 °C. The wells were washed three times with PBS, 0.1% v/v Tween-20, followed by the addition of 100 µL of anti-myc antibody at a 1:1000 dilution in 3% w/v nonfat dried milk in PBS, 0.1% v/v Tween-20, with incubation for 1 h at 37 °C. The wash step described earlier was repeated, and an HRPconjugated goat anti-mouse IgG (Jackson ImmunoResearch Laboratories, Inc., West Grove, PA) was added at a 1:2000 dilution in 3% w/v nonfat dried milk in PBS, 0.1% v/v Tween-20, and incubated at 37 °C for 1 h. After repeating the same washing step, OPD peroxidase substrate (SigmaFast o-phenylenediamine dihydrochloride tablets; Sigma, St. Louis, MO) was added first and then 3 N HCl to deactivate the enzyme. The optical density readings at 490 nm were recorded as relative units. Antigen Screening on Protein Biochips. The preparation of the protein microarray surfaces was performed with an in-house microarrayer as described elsewhere (23). Briefly, a silicon chip with a 20 nm layer of TiO2 was incubated with a solution of PLL-g-PEG-B30% for 1 h at room temperature to form the biotin surface. Neutravidin, avidin, and streptavidin (Pierce, Rockford, IL) were arrayed at varying concentrations onto the biotin-activated chip. After being washed, a microfluidic channel array was positioned onto the TiO2 surface to create individual channels on the chip. The surface in the channels was then incubated with 100 nM scFvE9myc-TAMRA in PBS, 0.05% Tween-20, 1% (w/v) BSA, and washed with PBS and then with H2O. The chips were scanned in a confocal fluorescence scanner (ScanArray 5000; GSI Lumonics, Billerica, MA) with a 543 nm excitation laser and a 578 nm emission filter. Signal intensities for given spot sizes were determined with the program QuantArray Version 3.0 (Packard BioScience, Billerica, MA). RESULTS AND DISCUSSION

Initial Refolding Studies for the Mxe gyrA Intein with GFPuv as a Model Protein. In a first set of experiments, we chose GFPuv as a model protein to test the refolding capability of the Mxe gyrA intein in a target protein-intein-His6-CBD fusion (Figure 1). This GFPuvintein construct can be expressed as a soluble or insoluble intein fusion protein in E. coli, depending on the growth

Bioconjugate Chem., Vol. 13, No. 4, 2002 709

Figure 2. SDS-PAGE and fluorescence analysis of expression and intein-mediated cleavage of the GFPuvK238Y-MxegyrA-H6CBD construct. The fusion protein was isolated from inclusion bodies after expression and refolded in nondenaturing PBS buffer containing 10% glycerol. Cleavage was performed after binding to chitin beads for 36 h at 4 °C with 2% ethanethiol. A, SDS-PAGE analysis. Lane 1: Protein marker (molecular weight in kDa). Lane 2: Cell lysate before induction. Lane 3: Cell lysate after induction with 0.1 mM IPTG and expression for 6 h at 30 °C. Lane 4: Insoluble fraction after lysis with a microfluidizer. Lane 5: Soluble fraction after lysis. Lane 6: Protein fraction bound to the chitin beads before cleavage. Lane 7: Protein fraction bound to the chitin beads after 36 h incubation with 2% ethanethiol at 4 °C. Lane 8: Supernatant fraction after 36 h incubation with 2% ethanethiol at 4 °C. Lane 9: Protein fraction bound to the chitin beads in the control incubation without ethanethiol. Lane 10: Supernatant fraction of the control incubation without ethanethiol. B, Fluorescence analysis. UV irradiation of the microcentrifuge tubes containing the intein-mediated cleavage samples after centrifugation. Tube 1 contains the reaction with 2% ethanethiol, tube 2 contains the control reaction without ethanethiol. Lane 7-10 in the SDS-PAGE analysis represent the chitin beads fractions and supernatants of these reaction tubes.

conditions, and its refolding can be monitored by the intrinsic fluorescence properties. We made the observation in our experiments that, in contrast to expression at 15 °C, expression of a GFPuvK238Y-MxegyrA-His6CBD intein fusion protein at 30 °C in E. coli gave mostly insoluble fusion protein (Figure 2). This construct gave us the opportunity to test whether we could regain activity of this particular intein from Mycobacterium xenopi after refolding. The inclusion bodies were isolated and solubilized in 6 M GuHCl by a procedure which is similar to the one published by Rudolph et al. (24, 25). The GFPuv-intein fusion protein was refolded by dialysis into nondenaturing PBS buffer containing 10% glycerol. The solubilized protein in denaturing buffer showed no fluorescence upon irradiation with UV light whereas the refolded sample clearly showed GFPuv fluorescence which indicated a correct folding of the soluble fraction. There was, however, also a significant amount of precipitated protein in the dialyzed sample. After centrifugation, the supernatant was loaded onto chitin beads and incubated with 2% ethanethiol for 36 h at 4 °C. The results presented in Figure 2 show that a fraction of the gyrase A intein mutant is able to generate a thioester at its N-terminus after refolding, which leads

710 Bioconjugate Chem., Vol. 13, No. 4, 2002

Figure 3. SDS-PAGE analysis of expression and optimized intein-mediated protein ligation under denaturing conditions with an scFv-intein fusion protein. A, Lane 1: Protein marker (molecular weight in kDa). Lane 2: Cell lysate before induction. Lane 3: Cell lysate after induction with 0.1 mM IPTG and expression for 6 h at 30 °C. Lane 4: Soluble fraction after lysis with a microfluidizer. Lane 5: Insoluble fraction after lysis. B, Lane 1: Protein marker (molecular weight in kDa). Lane 2: Protein fraction bound to the chitin beads before cleavage. Lane 3: Protein fraction bound to the chitin beads after 36 h incubation with 2% ethanethiol at RT. Lane 4: Supernatant fraction after 36 h incubation with 2% ethanethiol at RT. Lane 5: Protein fraction bound to the chitin beads after 36 h incubation with 2% ethanethiol and 2 mM myc peptide at RT. Lane 6: Supernatant fraction after 36 h incubation with 2% ethanethiol and 2 mM myc peptide at RT. Lane 7: Protein fraction bound to the chitin beads after 36 h incubation with 100 mM MESNA at RT. Lane 8: Supernatant fraction after 36 h incubation with 100 mM MESNA at RT. Lane 9: Protein fraction bound to the chitin beads after 36 h incubation with 100 mM MESNA and 2 mM myc peptide at RT. Lane 10: Supernatant fraction after 36 h incubation with 100 mM MESNA and 2 mM myc peptide at RT.

to the cleavage of GFPuv from the beads by an attack of the thioester with ethanethiol. GFPuv accumulates in the supernatant of the reaction with ethanethiol whereas there is only a minor cleavage of GFPuv in the control reaction without ethanethiol. This might be due to some hydrolysis of the thioester (Figure 2). About 50% of the refolded construct contains an active intein, estimated from the relative protein concentrations in the SDSPAGE, which gives reasonable amounts of target protein in the supernatant. In this case, several milligrams of GFPuv could be obtained from 1 L of cell culture by this procedure, which proves that a Mxe gyrA intein fusion protein can be successfully refolded to an active intein conformation. Intein-Mediated Ligation under Denaturing Conditions for the C-Terminal Labeling of an scFv. As mentioned above, the refolding of the GFPuv-intein fusion protein led to significant amounts of protein loss due to precipitation. Also, unlike GFPuv, which is relatively easy to refold, more challenging proteins, such as scFv’s, largely precipitated as intein fusion proteins during this refolding step. For this reason, we wished to establish a refolding and cleavage procedure that could be applied for target proteins that are difficult to refold into active conformations. As a model system to develop such a procedure, we constructed a vector which directs the cytoplasmic expression of a fusion construct of the Mxe gyrA intein with a single-chain Fv against neutravidin (20). This fusion protein, like that of the GFPuv construct, was expressed in an insoluble form in E. coli (Figure 3A). The inclusion body fraction containing the fusion protein was isolated and solubilized in 6 M GuHCl. Instead of a direct transfer of the fusion protein into physiological buffer conditions, we first diluted it into an intermediate buffer that is mildly denaturing. This was

Sydor et al.

intended to allow partial refolding of the intein and the chitin-binding domain to active conformations, thus allowing for protein immobilization and intein-mediated protein ligation. At the same time, it was intended to prevent the precipitation of the fusion protein that was observed when dialyzing directly into physiological buffer conditions. We first tried to perform the IPL in buffers with varying concentrations of GuHCl. Although the CBD was still able to bind to the chitin beads to some extent at low GuHCl concentrations, the intein showed no cleavage activity (data not shown). Ayers et al. have shown that intein cleavage activity in a soluble fusion protein can be mostly retained upon addition of up to 2 M GuHCl (10). In our case, however, the fusion protein had already been insoluble and needed to be solubilized in 6 M GuHCl before a transfer to lower GuHCl concentrations, and so the conditions are not directly comparable. In our next experiments, we transferred the fusion protein from 6 M GuHCl to a buffer containing the milder denaturant urea in order to prevent precipitation of the fusion protein and regain activity of the intein by the same time. A variety of different transfer protocols were compared, and the best results regarding intein activity were achieved by a stepwise dilution approach into 4 M urea (24, 25). In this case, 10 mL of a 10 mg/mL solution of the scFv-intein-CBD fusion protein was slowly added to 90 mL of a 4 M urea buffer at room temperature with a flow rate of 0.1 mL/min with constant stirring. This stepwise dilution approach lowers the concentration of the denatured protein during the refolding process and therefore prevents aggregation of folding intermediates (24, 25). The dilution was followed by another dialysis step into 4 M urea overnight at room temperature to ensure the removal of GuHCl. The protein sample was then bound to 1 mL chitin beads for the purification of the intein fusion protein. We then added the thiol additive and the peptide with the N-terminal cysteine. The IPL releases the target protein coupled to the modifying peptide into the supernatant while retaining the intein-CBD fusion protein on the beads. The ligation was carried out for 36 h at RT with a 50% chitin beads suspension in 4 M urea containing 100 mM MESNA and 2 mM peptide. The SDS-PAGE analysis of this reaction, performed with the scFv against neutravidin and a myc peptide with an N-terminal cysteine (NH2-CGGEQKLISEEDLNCONH2), showed results that were consistently seen in our investigations (Figure 3). The use of MESNA as a thiol additive improves the cleavage of the target protein from the intein fusion on the chitin beads in comparison to ethanethiol. The ligations at room temperature gave slightly higher amounts of product than ligations at 4 °C, without affecting the activity of the scFv (as measured by ELISA, see below). The addition of the peptide slightly increased the fraction of the fusion protein that cleaves from the intein, but also increased the amount of precipitated byproduct, as can be seen in the chitin bead fraction, which appears to contain some insoluble product. This approach, however, yielded significant amounts of ligated scFv in the supernatant. The yield of the intein cleavage was up to 20% under these conditions, based on the estimation of protein concentrations, using the protein standards of the SDS-PAGE analysis. The overall yield was approximately 2 mg of labeled scFv per liter of culture. In comparison, the amounts of scFv obtained from an E. coli periplasmic expression system is usually in the microgram range. The additional label-

Intein-Mediated Ligation under Denaturing Conditions

Bioconjugate Chem., Vol. 13, No. 4, 2002 711

Figure 4. ELISA with the scFv E9 against the antigen neutravidin after different final refolding conditions. ELISA signals for the (2) myc peptide labeled scFv E9 after stepwise dialysis from 4 M urea to PBS, 10% glycerol buffer with an oxido-shuffling system, (b) direct dialysis in PBS, 10% glycerol buffer with an oxido-shuffling system, and (9) direct dialysis in PBS, 10% glycerol buffer. (1) ELISA signal of the same scFv expressed with a C-terminal myc tag in a soluble form in the periplasm of E. coli as a positive control and ([) ELISA signal for the negative control with BSA as an antigen.

ing of the antibody fragment makes this procedure particularly useful. In the last step, we used different final dialysis procedures to transfer the labeled scFv into nondenaturing PBS buffer and compared these products by ELISA to the same scFv that was solubly expressed in the E. coli periplasm with the same C-terminal myc tag. This ensures that a given dialysis protocol leads to a correctly folded and site-specifically labeled scFv with a specific activity, which is comparable to an scFv obtained from soluble periplasmic expression. We used the following three dialysis protocols to transfer the IPL products from the 4 M urea buffer into physiological buffer: (1) directly into PBS, 10% glycerol with reduced/oxidized glutathione as an oxido-shuffling system to ensure proper formation of disulfide bonds. (2) PBS, 10% glycerol without an oxido-shuffling system. (3) Stepwise dialysis into 2 M urea in PBS, and then into 1 M urea in PBS, and finally into PBS, 10% glycerol with reduced/oxidized glutathione. The ELISA assay was performed on neutravidin-coated plates (BSA as a negative control) in which the myctagged scFv’s were detected by a mouse anti-myc antibody, followed by an HRP-conjugated goat anti-mouse IgG. The final transfer protocol of the IPL-labeled antibody fragment does not have a significant effect on the specific activity, as the three different dialysis conditions led to about the same activity in the ELISA assay (Figure 4). The comparison with the myc-tagged scFv that was solubly expressed in the E. coli periplasm shows that the activity is in the same range. The slightly lower signal for the myc-tagged scFv’s that were obtained from the IPL under denaturing conditions is probably due to the fact that the conditions of the IPL usually lead to some cleavage of the scFv without ligation of the peptide, so the total signal resulting from the peptide label of the scFv is slightly reduced. The periplasmic expression of the scFv-myc construct, however, led to much lower overall yields and does not allow the labeling of the antibody fragment with unnatural modifications such as fluorophores. The expression and isolation of scFv’s in a soluble form usually shows limited success, and the yields are very low. On the other hand, the expression in inclusion bodies

Figure 5. Biochip screening of the fluorescently labeled scFv E9 with different surface-immobilized antigens. A, Microarray surface with 50 µm spots containing different concentrations of the analytes and additional spots with no analyte as a control. Column 1: no analyte; columns 2-8: neutravidin spotted at concentrations of 500, 250, 125, 65, 31, 16, and 8 µg/mL; column 9: no analyte; column 10-16: avidin spotted at concentrations of 500, 250, 125, 65, 31, 16, and 8 µg/mL; column 17: no analyte; column 18-24: streptavidin spotted at concentrations of 500, 250, 125, 65, 31, 16, and 8 µg/mL; column 25: no analyte. B, Averaged fluorescence signals from the five spots with the highest concentration of the antigens (indicated by arrows), NA ) neutravidin, AV ) avidin, SA ) streptavidin. Error bars represent the standard deviations of the averaged signals from the five replicates in each column. The averaged signals were not background subtracted.

results in higher expression yields, but the refolding to an active conformation is generally very difficult and also of limited success. Our technique circumvents these problems and adds the additional opportunity of Cterminal labeling of a single-chain antibody with unnatural modifications such as fluorophores. Antigen Screening with a Fluorescently Labeled scFv in a Biochip Format. This advantage was further demonstrated by site-specific intein-mediated labeling of the scFv with a rhodamine-containing myc peptide (NH2CGGEQKLISEEDLNGGK(TAMRA)-CONH2) using an identical protocol to that described above. To test the utility of the resulting fluorescent scFv, we used a biochip assay. The preparation of the surface was published elsewhere (23). The analyte neutravidin and two control proteins, avidin and streptavidin, were microarrayed on a TiO2 chip containing a biotin-derivatized poly(L-lysine)poly(ethylene glycol) copolymer (PLL-g-PEG-B30%), and the chips were then incubated with a 100 nM solution of the scFv-myc-TAMRA by interfacing the chip with a microfluidic channel array. After the chips were washed, the fluorescence on the chips was analyzed with a fluorescence scanner (Figure 5). The fluorescence intensity was quantified and averaged for five spots in each

712 Bioconjugate Chem., Vol. 13, No. 4, 2002

row containing the antigens at varying concentrations. The results clearly show the binding properties of the fluorescently labeled scFv in a chip assay format. The specific binding of neutravidin compared to streptavidin and the weak cross-reactivity with avidin are in good correlation with BIAcore experiments that we performed (data not shown). In contrast to ELISA’s, chip-based assays can be performed in a more miniaturized format to ensure higher throughput and smaller sample consumption (23). The microfluidic channels interfacing the microarray have a volume of about 10 µL and contain 250 features with a 50 µm diameter. Therefore, the assay can be performed on a significantly smaller scale compared to other immunoassay formats. In conclusion, the site-specific intein-mediated labeling of single-chain antibodies under denaturing conditions for biochip assays offers new possibilities in the work with antibody fragments. This technique can produce large amounts of active and site-specifically labeled scFv for different applications. The antibody fragment can initially be labeled with a fluorophore for antigen screening in a chip assay format. The same fusion protein can then also be used to introduce other modifications for an oriented immobilization of the antibody fragment on surfaces. ACKNOWLEDGMENT

We thank Walter Khu and Zhenni Ding for the synthesis of the peptides. In addition, we thank David Wilson and Hans Beernink for critical reading of the manuscript. LITERATURE CITED (1) Muir, T. W., Sondhi, D., and Cole, P. A. (1998) Expressed protein ligation: A general method for protein engineering. Proc. Natl. Acad. Sci. U.S.A. 95, 6705-6710. (2) Severinov, K., and Muir, T. W. (1998) Expressed Protein Ligation, a Novel Method for Studying Protein-Protein Interactions in Transcription. J. Biol. Chem. 273, 1620516209. (3) Kane, P. M., Yamashiro, C. T., Wolczyk, D. F., Neff, N., Goebl, M., and Stevens, T. H. (1990) Protein Splicing Converts the Yeast TFP1 Gene Product to the 69-kD Subunit of the Vacuolar H+-Adenosine Triphosphatase. Science 250, 651657. (4) Hirata, R., Ohsumi, Y., Nakano, A., Kawasaki, H., Suzuki, K., and Anraku, Y. (1990) Molecular Structure of a Gene, VMA1, Encoding the Catalytic Subunit of H+- Translocating Adenosine Triphosphatase from Vacuolar Membranes of Saccharomyces cerevisiae. J. Biol. Chem. 265, 6726-6733. (5) Noren, C. J., Wang, J., and Perler, F. B. (2000) Dissecting the Chemistry of Protein Splicing and Its Applications. Angew. Chem., Int. Ed. 39, 450-466. (6) Perler, F. B., and Adam, E. (2000) Protein splicing and its applications. Curr. Opin. Biotechnol. 11, 377-383. (7) Blaschke, U. K., Silberstein, J., and Muir, T. W. (2000) Protein Engineering by Expressed Protein Ligation. Methods Enzymol. 328, 478-496. (8) Giriat, I., Muir, T. W., and Perler, F. B. (2001) Protein Splicing and Its Applications. Genet. Eng. 23, 171-199.

Sydor et al. (9) Cotton, G. J., and Muir, T. W. (2000) Generation of a duallabeled fluorescence biosensor for Crk-II phosphorylation using solid-phase expressed protein ligation. Chem. Biol. 7, 253-261. (10) Ayers, B., Blaschke, U. K., Camarero, J. A., Cotton, G. J., Holford, M., and Muir, T. W. (1999) Introduction of Unnatural Amino Acids into Proteins Using Expressed Protein Ligation. Biopolymers 51, 343-354. (11) Iakovenko, A., Rostkova, E., Merzlyak, E., Hillebrand, A. M., Thoma¨, N. H., Goody, R. S., and Alexandrov, K. (2000) Semi-synthetic Rab proteins as tools for studying intermolecular interactions. FEBS Lett. 468, 155-158. (12) Kawasaki, M., Makino, S.-I., Matsuzawa, H., Satow, Y., Ohya, Y., and Anraku, Y. (1996) Folding-Dependent in Vitro Protein Splicing of the Saccharomyces cerevisiae VMA1 Protozyme. Biochem. Biophys. Res. Comm. 222, 827-832. (13) Raag, R., and Whitlow, M. (1995) Single-chain Fvs. FASEB J. 9, 73-80. (14) Jackson, H., Bacon, L., Pedley, R. B., Derbyshire, E., Field, A., Osbourn, J., and Allen, D. (1998) Antigen specificity and tumor targeting efficiency of a human carcinoembryonic antigen-specific scFv and affinity-matured derivatives. Br. J. Cancer 78, 181-188. (15) Hanes, J., Schaffitzel, C., Knappik, A., and Plu¨ckthun, A. (2000) Picomolar affinity antibodies from a fully synthetic naive library selected and evolved by ribosome display. Nature Biotechnol. 18, 1287-1292. (16) Houston, J. S., Mudgett-Hunter, M., Tai, M.-S., McCartney, J., Warren, F., Haber, E., and Oppermann, H. (1991) Protein Engineering of Single-Chain Fv Analogues and Fusion Proteins. Methods Enzymol. 203, 46-88. (17) Glockshuber, R., Schmidt, T., and Plu¨ckthun, A. (1992) The Disulfide Bonds in Antibody Variable Domains: Effects on Stability, Folding in Vitro, and Functional Expression in Escherichia coli. Biochemistry 31, 1270-1279. (18) Skerra, A. (1993) Bacterial expression of immunoglobulin fragments. Curr. Opin. Immunol. 5, 256-262. (19) Kurucz, I., Titus, J. A., Jost, C. R., and Segal, D. M. (1995) Correct Disulfide Pairing and Efficient Refolding of DetergentSolubilized Single-Chain Fv Proteins From Bacterial Inclusion bodies. Mol. Immunol. 32, 1443-1452. (20) Phipps, M. L., Xu, X., Nock, S., and Kassner, P. D. (2000) Detection of antibody display phage without clearing of bacterial culture. Biotechniques 29, 739-740. (21) Merrifield, R. B. (1986) Solid-phase synthesis. Science 232, 341-347. (22) Schno¨lzer, M., Alewood, P., Jones, A., Alewood, D., and Kent, S. B. H. (1992) In situ neutralization in Boc-chemistry solid-phase peptide synthesis. Int. J. Pept. Protein Res. 40, 180-193. (23) Ruiz-Taylor, L. A., Martin, T. L., Zaugg, F. G., Witte, K., Indermuhle, P., Nock, S., and Wagner, P. (2001) Monolayers of derivatized poly(L-lysine)-grafted poly(ethylene glycol) on metal oxides as a class of biomolecular interfaces. Proc. Natl. Acad. Sci. U.S.A. 98, 852-857. (24) Rudolph, R., and Lilie, H. (1996) In vitro folding of inclusion body proteins. FASEB J. 10, 49-55. (25) Rudolph, R., Bo¨hm, G., Lilie, H., and Jaenicke, R. (1997) Folding proteins. Protein Function: A practical Approach (Creighton, T. E., Ed.) pp 57-99, IRL Press, Oxford, UK.

BC025534Z