Efficient synthesis of a disulfide-containing protein through a batch cell-free system from wheat germ Takayasu Kawasaki 1 , Mudeppa D. Gouda 1 , Tatsuya Sawasaki 1,2 , Kazuyuki Takai 1,2 and Yaeta Endo 1,2 1 Cell-free Science and Technology Research Center (CSTRC), and 2 Department of Applied Chemistry, Faculty of Engineering, Ehime University, Matsuyama, Japan We have developed a highly productive cell-free protein synthesis system from wheat germ, which is expected to become an important tool for postgenomic research. However, this system has not been optimized for the synthesis of disulfide-containing proteins. Thus, we sear- ched here for translation conditions under which a model protein, a single-chain antibody variable fragment (scFv), could be synthesized into its active form. Before the start of translation, the reducing agent dithiothreitol, which normally is added to the wheat germ extract but which inhibits disulfide formation during translation, was removed by gel filtration. When the scFv mRNA was incubated with this dithiothreitol-deficient extract, more than half of the synthesized polypeptide was recovered in the soluble fraction. By addition of protein disulfide isomerase in the translation solution, the solubility of the product was further improved, and nearly half of the soluble polypeptides strongly bound to the antigen immobilized on an agarose support. This strong binding component had a high affinity as shown by surface-plas- mon resonance analysis. These results show that the wheat germ cell-free system can produce a functional scFv with a simple change of the reaction ingredients. We also discuss protein folding in this system and suggest that the disulfide bridges are formed cotranslationally. Finally, we show that biotinylated scFv could be synthesized in simi- lar fashion and immobilized on a solid surface to which streptavidin is bound. SPR measurements for detection of antigens were also possible with the use of this immobi- lized surface. Keywords: cell-free synthesis; scFv; disulfide; wheat germ; biotinylation. As a consequence of the successes of the large-scale genome sequencing projects, structural and functional analyses of proteins are growing more and more important. We have developed a highly productive cell-free protein synthesis system from wheat germ, which enables the production of milligram quantities of cDNA-encoded proteins and highly parallel synthesis of many different proteins under unified reaction conditions [1,2]. The expression of a cDNA clone in this cell-free method can be performed with the 5¢-and 3¢-untranslated regions (UTRs) that were devised for efficient mRNA translation [3]. It has been pointed out that the parallel procedure, including cDNA amplification, transcription and translation, could be operated by a programmable liquid-handling machine. Therefore, the wheat germ protein synthesis system is expected to become a powerful tool for genome-wide high-throughput analyses of proteins, proteome-based diagnosis and other post- genomic applications [3]. On the other hand, it is clear that many proteins encoded in the genomes have disulfide bonds that play important roles in their function. However, no data have been reported that demonstrate the successful production by a wheat germ cell-free system of a disulfide-containing protein in active form. In fact, disulfide bonds are unlikely to form during cell-free synthesis because the synthesized polypeptides normally are released into a buffer containing dithiothreitol (DTT). DTT, or some other reducing agent, is required for preserving the protein synthesis activity of the wheat germ extract during storage and the translation reaction [1]. In addition, the cell-free system may lack those enzymes that would catalyze the correct disulfide formation in the lumen of endoplasmic reticulum in vivo, such as protein disulfide isomerase (PDI) [4,5]. In the present study, we searched for translation condi- tions under which a disulfide-containing protein could be efficiently produced in an active form. The model protein we chose for the study was a single-chain antibody variable fragment (scFv) against Salmonella O-antigen [6]. The scFvs are engineered proteins that contain the heavy chain and light chain variable domains (V H and V L , respectively) of an antibody connected by a linker [7]. Each domain has an intradomain disulfide bond. The anti-(O-antigen) scFv (26 kDa) has been produced successfully through an Correspondence to Y. Endo, Department of Applied Chemistry, Faculty of Engineering, Ehime University, Matsuyama 790–8577, Japan. Fax: + 81 899279941, Tel.: + 10 81 899279936, E-mail: yendo@eng.ehime-u.ac.jp Enzymes: biotin ligase from E. coli (EC 6.3.4.15); protein disulfide isomerase (EC 5.3.41). Abbreviations: DTT, dithiothreitol; PBT, phosphate-buffered Tween 20; PDI, protein disulfide isomerase; scFv, single-chain anti- body variable fragment; SPR, surface plasmon resonance; V H ,heavy chain variable domain; V L , light chain variable domain; UTR, untranslated region. (Received 4 August 2003, revised 30 September 2003, accepted 20 October 2003) Eur. J. Biochem. 270, 4780–4786 (2003) Ó FEBS 2003 doi:10.1046/j.1432-1033.2003.03880.x Escherichia coli secretory expression system, and its three- dimensional structure has been solved [8]. We describe here suitable conditions for the cell-free synthesis of the active disulfide-containing scFv. Materials and methods Materials The gene encoding the scFv fragment of the Se155–4 IgG antibody, in the orientation VL-linker-VH (scFvLH) was a generous gift from M. N. Young [6]. LATaq PCR kit, RNA LA PCR kit, and protein disulfide isomerase (EC 5.3.41) from bovine liver were obtained from Takara Shuzo, Japan. pGEM-T Easy cloning kit, DNA ligase, RNasin, SP6 RNA polymerase and streptavidin magnetic beads were obtained from Promega. NICK columns, Sephadex G25 spin col- umn and epoxy-activated Sepharose 6B gel were from Amersham Pharmacia. Biotin ligase from E. coli (EC 6.3.4.15) was purchased from Avidity LLC. Restriction enzymes were obtained from New England Biolabs, Inc. Lipopolysaccharides from Salmonella typhimurium and E. coli O111, biotin and cycloheximide were purchased from Sigma. Amine-terminated magnetic beads were pur- chased from Polysciences, Inc. Construction of plasmids The DNA fragment coding the scFvLH (10 pg) was amplified with 0.4 l M each of the primers s1: 5¢-CTACC AGATCTGCCATGCAGATCGTTGTTACCCAGG-3¢ and a1: 5¢-GGCTAAGAGCTCACGGTCAGGCTCG-3¢ by using a LATaq PCR kit (50 lL). The sequences underlined are the BglII restriction site and initiation codon, and the bold sequence is an anti-stop codon. The PCR product was subcloned into pGEM-T Easy, and the resultant plasmid DNA was digested with BglII and NotI. This fragment was inserted into pEU [3] at the same restriction sites. The obtained plasmid pEU–scFvLH was purified by a standard cesium chloride density-gradient centrifugation method. pEU-scFvLBH, which had a biotin tag sequence coding 15 amino acids (GLNDIFEAQKI EWHE) [9] at the linker region was constructed by ampli- fying the whole plasmid pEU-scFvLH by inverse PCR with the primers s2: 5¢- CAAAAAATTGAATGGCATG AACCGCCGAGCTCCAAC-3¢ and a2: 5¢-AGCTTCAA AAATATCATTTAAACCCGACGGGCTGCTTTT-3¢ (the sequences for the biotin tag are underlined) followed by circularization with DNA ligase. Preparation of mRNAs The pEU plasmids were transcribed in vitro (400 lL) at 37 °Cfor2hwith80m M Hepes-KOH (pH 7.6), 16 m M Mg(OAc) 2 ,2m M spermidine, 10 m M DTT, 2.5 m M of each nucleotide triphosphate (ATP, UTP, GTP and CTP), 0.8 unitÆlL )1 RNasin (ribonuclease inhibitor), 20 lg plasmid DNA, and 1.0 unitÆlL )1 SP6 RNA polymerase. The reaction mixture was extracted with phenol/water, then with chloroform/water. These mRNAs were purified by gel filtration with NICK columns and finally by ethanol precipitation. Cell-free translation The wheat-germ extract was prepared by the reported method [1]. DTT in the extract (4 m M ) was excluded just prior to translation by gel-filtration with a Sephadex G25 spin column pre-equilibrated by a DTT-deficient buffer [40 m M Hepes/KOH (pH 7.8), 100 m M KOAc, 5 m M Mg(OAc) 2 and 0.3 m M each of the 20 amino acids]. The translation mixture (38 lL) containing 1.2 m M ATP, 0.25 m M GTP, 15 m M creatine phosphate, 0.4 m M spermi- dine, 28 m M Hepes/KOH (pH 7.8), 0.23 m M each of 20 amino acids (including leucine), 1.8 mgÆmL )1 creatine kinase, 53 m M KOAc, 1.6 m M Mg(OAc) 2 ,0.6m M CaCl 2 , 0.4 unitÆlL )1 RNasin, 200 lgÆmL )1 mRNA, and the DTT- deficient wheat germ extract (A 260 ¼ 42) was incubated at 26 °C for 4 h. To label the synthesized proteins, 4 lCiÆmL )1 [ 14 C]Leu was also included (nonradioactive leucine was not omitted, and the final concentration of leucine thus adds up to 0.24 m M ). For biotinylation, the mRNA from pEU– scFvLBH was translated in the same way, except that 19.5 l M biotin and 19.5 lgÆmL )1 biotin ligase were added. The amounts of the synthesized polypeptides were deter- mined from the 14 C radioactivity of the trichloroacetic acid-precipitated materials. The soluble fractions of the synthesized polypeptides were obtained by centrifugation (20 000 g,10min),andtheÔsolubilityÕ was determined as the amount of the soluble fraction divided by the total amount. The concentration of the scFv synthesized without [ 14 C]Leu was estimated from that of the polypeptide synthesized with the radioisotope in parallel. Affinity chromatography The antigen-coupled agarose gel was prepared by a reported method [10]. Lipopolysaccharide (15 mg) from Salmonella typhimurium was deacetylated, oxidized, and aminated to be approximately 10 mg of freeze-dried powder containing 25 lmol amino group, determined with picrylsulfonic acid. The aminated polysaccharide was mixed with the epoxy- activated Sepharose 6B gel (1.5 g) in 20 mL of 0.2 M NaOH/ KCl (pH 12.5). The coupling yield was 1.3 lmol as deter- mined by a standard phenol/sulfuric acid procedure. The soluble fraction containing an scFv polypeptide (20 lL) was diluted by the same volume of 50 m M Tris/HCl, pH 8.0, and loaded on the antigen column (120 lL) pre-equilibrated with the same buffer. The column was washed with 280 lL of the buffer (fractions 2–8) and then 200 lL of the buffer containing 0.15 M NaCl (9–13). The antigen-bound poly- peptide was eluted (14–22) with 360 lLof4%(w/v)SDS (for analysis) or 0.1 M glycine/HCl, pH 2.3 (for purifica- tion). The amount of polypeptide in each 40 lLfractionand the ratio of the antigen-binding fraction to the total loaded polypeptide were determined by measuring the radioactivity. SPR analysis For the quantitative binding studies, an IAsys biosensor (Affinity Sensors) was used. The oxidized Salmonella polysaccharide (0.3 lmol) was dissolved in 50 lLof 20 m M sodium borate buffer, pH 9.0, and then added onto the amino cuvette supplied with the instrument at the level of 233 arcÆsecond (6.2 ng). To avoid the nonspecific binding Ó FEBS 2003 Cell-free synthesis of a disulfide-containing protein (Eur. J. Biochem. 270) 4781 of protein, the surface of the cuvette was prewashed with the same volume of the wheat-germ extract. The scFv was synthesized under the DTT-deficient condition in a 76 lL reaction without a radioisotope and purified by the above method. The resulting solution (50 lL) was neutralized by 200 m M phosphate-buffered Tween 20 (PBT), pH 8.0, desalted by G25 spin column pre-equilibrated with 10 m M PBT, pH 8.0, and then added to the antigen-coated cuvette at various concentrations. The antigen–antibody association curves were measured for 5 min at 25 °C, and the bound polypeptide was dissociated from the surface by the addition of 10 m M PBT, pH 8.0. The dissociation rate constant k dissoc (s )1 ) and the association rate constant k assoc ( M )1 Æs )1 )were obtained by using the FASTFIT software supplied with the instrument. The dissociation constant K D ( M ) was calculated according to the equation: K D ¼ k dissoc /k assoc . Affinity capture of the nascent polypeptide–mRNA complex The linear DNA fragment (910 bp) including the scFvLH sequence was amplified on the pEU–scFvLH using SP6 primer (5¢-ATTTAGGTGACACTATAG-3¢)andanti-pri- mer (5¢-ATGGCGCCAGCTGCAGGCTA-3¢, anti-stop codon in bold), and transcribed in the same way as above. Translation was carried out with the purified mRNA (75 lgÆmL )1 )for30minat26°C as mentioned above, and the reaction was stopped by addition of cycloheximide (2.3 l M ). The soluble fraction was diluted two-fold with 50 m M Tris/HCl, pH 7.5, containing 10 m M MgCl 2 and gel- filtrated on a G25 spin column pre-equilibrated with the same buffer. Antigen-coated magnetic beads were prepared by mixing the oxidized polysaccharide with the amine terminated particles. The antigen from Salmonella typhimu- rium or E. coli O111 (0.32 lmol and 0.42 lmol, respectively) was reacted with 100 lL of the beads in 20 m M sodium borate buffer, pH 9.0, at room temperature for 6 h (total 200 lL). The coupling yield was 0.14 lmol for either antigen as determined by a standard phenol–sulfuric acid procedure. The filtrated solution (30 lL) containing the mRNA– polypeptide complex was mixed with the antigen beads (5 lL) for 10 min at room temperature. The beads were washed five times with 30 lLof50m M Tris/HCl (pH 7.5) containing 10 m M MgCl 2 and 0.15 M NaCl, then finally mixed with the same volume of 0.1 M glycine/HCl (pH 2.3) for 15 min. The eluted mRNA was recovered by precipita- tion with ethanol and reversely transcribed and amplified by using RNA LA PCR kit with the above primers. Results Plasmid construction The cDNA for the scFv was first inserted into the pEU vector [3] that was designed for the specific purpose of protein expression in the cell-free system. pEU contains the sequences for the SP6 promoter, translation enhancer region independent of the cap structure at the 5¢-end, restriction enzyme sites, and a 3¢-untranslated region required for the high translation efficiencies (Fig. 1A). The transcript from a cDNA inserted into this plasmid was used directly for the cell-free translation. Synthetic condition allowing efficient disulfide formation The reducing agent DTT has been conventionally inclu- dedinthewheat-germextractatahighconcentrationfor the sake of stability during storage [1]. Thus, in order to reduce the DTT concentration for the translation reaction, Fig. 1. Syntheses of the scFv. (A) Transcription unit in the pEU plasmid containing the scFvLH. P SP6, SP6 promoter; 5¢-UTR, 5¢-untranslated region; 3¢-UTR, 3¢-untranslated region. The initiation codon and the stop codon were introduced by the PCR primers (s1 and a1). (B) Incorporation of [ 14 C]Leu into polypeptides during the cell-free translation of the scFvLH mRNA with various additives. (a) 2 m M external DTT (e); (b) 2 m M external DTT + 0.5 l M PDI (h); (c) 0 m M external DTT (n); (d) 0 m M external DTT + 0.5 l M PDI (·); (e) translation under the conventional condition (2.5 m M DTT) with the untreated extract (s). (C) Autoradiogram of SDS/ PAGE separating the scFvLH polypeptides synthesized under the same conditions as indicated in (B). Total (t) and soluble (s) fractions were denatured without 2-mercaptoethanol before loading onto the gel. 4782 T. Kawasaki et al. (Eur. J. Biochem. 270) Ó FEBS 2003 the extract was passed through a gel filtration column just before the start of reaction. With the use of this low-DTT extract, the scFvLH mRNA was translated in the presence of [ 14 C]Leu, and the productivity was monitored by measuring the radioactivity of the trichloroacetic acid- insoluble precipitates (Fig. 1B). The polypeptide produc- tion under the DTT-deficient condition (Fig. 1Bc) was more than half that under the DTT-rich condition (Fig. 1Ba). The translation activity of the wheat-germ extract was not affected by the gel filtration, as the productivity under the DTT-rich condition using the gel- filtrated extract (Fig. 1Ba) was almost the same as that under the conventional condition with the untreated extract (Fig. 1Be). We also tested if PDI would affect the productivity. Although the synthesis efficiency was slightly less in the presence of PDI (Fig. 1Bb,d) than in the absence (Fig. 1Ba,c), which might be due to the phosphate that buffers the PDI stock solution, the enzyme did not largely interfere with the translation machinery at the low level (0.5 l M ). The polypeptide productivity of the DTT-deficient, PDI-containing reaction was approxi- mately 40 lgÆmL )1 . Then, in order to estimate the extent of disulfide formation, the synthesized polypeptide was analyzed on a nonreducing SDS gel after 4 h translation (Fig. 1C). In this gel, the oxidized form runs faster than the reduced form [11]. The result showed that the polypeptides from the DTT- deficient reactions (Fig. 1Cc,d) migrated faster than those from the DTT-rich reactions (Fig. 1Ca,b) around 30 kDa. Both sets of samples had the same mobility on a reducing gel (data not shown). The solubility of the synthesized scFv polypeptide was calculated to be 50% in (Fig. 1Ca), 50% in (Fig. 1Cb), 65% in (Fig. 1Cc), and 85% in (Fig. 1Cd), respectively. Therefore, the DTT-deficient condition in the presence of PDI was effective for the disulfide formation and increased the soluble polypeptides. Although we have not determined the exact concentration of DTT in the translation mixture, the above results were highly reprodu- cible. Fig. 2. Functional analyses. (A) Elution profiles of the soluble scFvLH polypeptides on the agarose-gel coupled with Salmonella polysaccha- ride. Solvents: 50 m M Tris/HCl buffer, pH 8.0 (numbers 2–8); the same buffer containing 0.15 M NaCl (numbers 9–13); 4% SDS (numbers 14–22). e,2m M external DTT; h,2m M external DTT + 0.5 l M PDI; n,0m M external DTT; ·,0m M external DTT + 0.5 l M PDI. The inserted autoradiogram showed the nonreducing SDS/PAGE. (B) SPR analysis using IAsys instrument. The concentrations of the purified scFv polypeptides were 41, 21 and 11 n M (from top to bot- tom). (C) Post-translational effect of PDI on the active fraction of the scFv. The enzyme was added in the DTT-deficient synthesis solution after 3 h translation, and the reaction was further continued for the same number of hours. The soluble material of the product was sep- arated on the affinity column described in (A). (D) Agarose gel (1.8%) electrophoreses of cDNAs from the nascent polypeptides–mRNA complexes. Translations were performed under the two different con- ditions, and each solution was mixed with the cognate or noncognate antigen beads. The captured mRNA was dissociated from the complex with the acidic buffer, recovered with ethanol precipitation, reversely transcribed, and amplified. Lane 0, 100 bp DNA marker; lane 1, DTT- deficient condition (0 m M external DTT) with Salmonella antigen; lane 2, DTT-deficient condition with E. coli antigen; lane 3, DTT-rich condition (2 m M external DTT) with Salmonella antigen; lane 4, DTT- rich condition with E. coli antigen. Ó FEBS 2003 Cell-free synthesis of a disulfide-containing protein (Eur. J. Biochem. 270) 4783 Antigen-binding analyses We then examined the antigen binding activity. The soluble material from each reaction was loaded on an antigen column and was separated into the unbound (numbers 2–8), washed (numbers 9–13), and eluted (numbers 14–22) fractions (Fig. 2A) according essentially to the method described for the Se155–4 IgG antibody [10]. The third fraction that eluted with SDS was judged to include the strong-binding polypeptide. The results clearly showed that the only products from the DTT-deficient media (Fig. 2Ac,d) contained significant amounts of the active molecules, whereas most of the products from the DTT-rich media (Fig. 2Aa,b) were recovered in the inactive unbound fractions. The total amount of the active fractions (numbers 16–18) in Fig. 2Ad was 37% of the loaded soluble polypeptides. The sample from fraction number 17 gave a radioactive band on the nonreducing SDS gel correspond- ing to the oxidized form. Therefore, PDI increased the fraction of functional scFv during translation under DTT- deficient conditions. Next, this active polypeptide was eluted by the acidic buffer instead of SDS solution and applied to the SPR analysis (Fig. 2B). The antigen–antibody association curves were measured on addition of various concentrations of the purified polypeptide onto the antigen-coated surface. K D was calculated by using the linear equation, k on ¼ k assoc [ligate] + k dissoc (K D ¼ k dissoc /k assoc ), where k on (s )1 ) is the on-rate constant defined as a slope of the exponential association curve, and [ligate] is the concentration of the added polypeptide [12]. The resulted K D for scFvLH was found to be in the order of 10 )8 M , which indicated that the synthesized polypeptide had a high affinity for the immo- bilized antigen (Table 1). These results showed that approximately 13 lgofa functional scFv was produced in a 1-mL batch reaction under the DTT-deficient, PDI-containing conditions, a simple alteration of the conventional translation condition. Affinity capture of the nascent polypeptide We also tested a post-translational effect of PDI on the active fraction of the mature scFv polypeptide. After the 3 h synthesis under the DTT-deficient conditions without PDI, the translation solution was incubated with the enzyme for the same number of hours. The soluble material was loaded on the affinity column (Fig. 2C). The active fraction that eluted with SDS made up only 12% of the total loaded polypeptides. This might mean that the disulfide isomeriza- tion proceeds more efficiently during elongation of the polypeptide chain, rather than after the release of mature polypeptide from the ribosomes. Thus, we examined whether the complex of the growing polypeptide of the correctly folded scFv and the mRNA connected via the ribosomes could be isolated. The polypeptide elongation was stopped by addition of cycloheximide, and the mRNA– polypeptide complexes were captured on antigen-coated magnetic beads. The isolated complex was dissociated with an acidic buffer, and the recovered mRNA was reversely transcribed and amplified (Fig. 2D). The cDNA band from the DTT-deficient reaction (lane 1) was more clearly seen than that from the DTT-rich reaction (lane 3). In addition, the cDNA bands from the reactions with the noncognate antigens (lane 2 and 4) were relatively weak. These results indicate that a larger fraction of the nascent polypeptide produced under the DTT-deficient conditions than under the DTT-rich conditions recognized the antigen and thus had the correct disulfide bonds. Functional biotinylation To show a possible application of the present method, we tested whether a biotinylated scFv could be synthesized by a modification of this method. We used a biotin ligation enzyme from E. coli for biotinylation [9]. A biotin tag sequence was introduced in the middle of the original linker of scFvLH by PCR (Fig. 3A), and the mRNA from this plasmid was translated in the presence of biotin and the enzyme. The efficiency of biotinylation was evaluated by measuring the fraction that could be trapped by the streptavidin-conjugated magnetic beads (Fig. 3B). The figure shows the autoradiogram of SDS/polyacryl- amide gel separating three materials [total solution before mixing with the beads (t), unbound solution (u) and bound solution (b) after mixing with the beads] of the synthesized polypeptides. The results clearly showed that much more polypeptide from the DTT-deficient reaction bound to the beads than that from the DTT-rich reaction. Therefore, the disulfide-formed polypeptides were prefer- entially biotinylated in this system. The biotinylated scFv was then purified with the use of the antigen column, and the affinity was measured in the same way as above. The antigen-binding activity was found to be as high as the wild type (Table 1), which showed that the present system was able to produce a biotinylated scFv without any loss of its function. Then, the crude synthesis solution containing the biotin- ylated scFv was immobilized on a streptavidin-coated cuvette for the IAsys instrument, and SPR measurements were carried out (Fig. 3C). When the cognate antigen was added (Fig. 3C A ), the response values increased as the concentrations increased. The affinity of the antigen to the immobilized scFv was determined to be 4.4 · 10 )6 M K D by using the same equation as above. The control reaction with noncognate antigen (Fig. 3C B ) gave low response levels. These results show that a sensitive antigen-detection system Table 1. Affinity parameters. K D ( M ) k dissoc (s )1 ) k assoc ( M )1 Æs )1 ) scFvLH 4.3 · 10 )8 (± 0.6) 0.7 · 10 )1 (± 0.07) 1.7 · 10 6 (± 0.06) Biotinylated scFvLH 5.4 · 10 )8 (± 1.3) 0.8 · 10 )1 (± 0.1) 1.5 · 10 6 (± 0.1) 4784 T. Kawasaki et al. (Eur. J. Biochem. 270) Ó FEBS 2003 can be conveniently constructed by using the crude cell-free protein synthesis solution. The present biotinylation technique may be useful for preparation of biotin-labeled proteins and also for non- radioisotope detection of cell-free synthesized proteins [13]. Discussion This report is the first to describe the synthesis by the wheat- germ cell-free system of a disulfide-containing protein. The functional and structural analyses revealed that a consider- able fraction of the synthesized polypeptides folded cor- rectly under the DTT-deficient condition with PDI present during peptide synthesis. The results show that the thiol/ disulfide interchange reaction proceeds efficiently in this cell- free system. The fact that the growing scFv polypeptide was captured by the affinity to the antigen means that the correct folding of the polypeptide catalyzed in part by PDI proceeds more or less before the release from the ribosomes. This seems consistent with the suggestion that de novo proteins fold cotranslationally during the eukaryotic protein synthesis [14], which might make the system suitable for the synthesis of multidomain proteins. In fact, our method does not require the addition of the chaperone molecules such as GroEL and GroES, whose addition appears to be required to support correct formation of disulfide bonds in E. coli cell-free protein synthesis systems [15,16]. The active frac- tion of the scFv produced in the E. coli cell-free system with chaperones was nearly half of the synthesized protein, which was almost the same as in our case. Therefore, the wheat- germ system might have a practical advantage over the bacterial cell-free system. The scFv against Salmonella O-antigen was for the first time prepared successfully through an E. coli expression system, in which the produced polypeptide was secreted into the periplasmic space of the bacteria [6]. The result of the antigen-binding analysis demonstrated that the scFv produced in the wheat-germ cell free system has a higher affinity than that produced in the E. coli system [8]. This difference in the activity of the synthesized polypeptide might be due to the difference in the intrinsic characteristics between the eukaryotic and prokaryotic translation systems. The synthesis method described in this paper may be more convenient than the bacterial expression system if only a small amount (1–20 lgÆmL )1 ) of the active protein is required. To improve the protein yield in the cell-free system, it might be necessary to add some reductive potential buffer to prolong the protein synthesis. Although we have already tried several other scFvs and observed that the method worked as far as these scFvs are concerned, the general applicability of this method for scFvs and other disulfide-containing proteins is yet to be determined. However, in the postgenomic researches, parallel produc- tion of many different proteins will be also needed. A protein synthesis machine that can perform PCR, tran- scription and translation automatically with a highly parallel operation is being developed in our laboratory. Fig. 3. Biotinylation of scFv. (A) Transcription unit in the pEU plas- mid containing the scFvLBH. The biotin-tag was introduced by PCR primers (s2 and a2). (B) Autoradiogram of a reducing SDS gel sep- arating the biotinylated scFv. The mRNA from pEU-scFvLBH was translated (38 lL) under the DTT-deficient (–) or the DTT-rich (+) condition in the presence of PDI for three hours. The soluble fractions were twice diluted with 50 m M Tris/HCl, pH 8.0 and loaded onto G25 spin columns pre-equilibrated with the same buffer to remove free biotin. The filtrate (30 lL) was mixed with 5 lL of streptavidin- magnetic beads at room temperature for 15 min and the bound polypeptide was eluted with 4% SDS. Three materials (filtrate before mixing with the beads (t), unbound solution (u) and bound solution (b) after mixing with the beads) were denatured with 2-mercaptoethanol before loading onto the gel. (C) Binding responses of the antigens to the biotinylated scFv on the streptavidin-coated cuvette of IAsys instrument. The biotinylated scFv was synthesized under the DTT- deficient condition (76 lL reaction without a radioisotope), and the mixture was desalted through G25 spin column pre-equilibrated with 10 m M PBT (pH 8.0). The biotin cuvette was first coated by strept- avidin (34 ng), then by the crude protein solution including the bio- tinylated scFv (50 lL). The responses were measured by the addition of free antigens A: Salmonella polysaccharide, at 9.7 l M ,4.9l M and 1.2 l M (upper, middle and bottom responses, respectively); B: E. coli O111 polysaccharide, 10 l M (upper), 5.0 and 2.5 l M (lower). Ó FEBS 2003 Cell-free synthesis of a disulfide-containing protein (Eur. J. Biochem. 270) 4785 The present modification of the protein synthesis condition should be amenable to automation, because it involves only the alterations in the reaction ingredients. Acknowledgements We thank Dr N. M. Young for providing the scFvLH gene. This work was supported by Japan Society for the Promotion of Science Grant JSPS-RFTF 96100305 (to Y.E.). References 1. Madin, K., Sawasaki, T., Ogasawara, T. & Endo, Y. (2000) A highly efficient and robust cell-free protein synthesis system pre- pared from wheat embryos: plants apparently contain a suicide system directed at ribosomes. Proc. Natl Acad. Sci. USA 97, 559–564. 2. Sawasaki, T., Hasegawa, Y., Tsuchimochi, M., Kamura, N., Ogasawara, T., Kuroita, T. & Endo, Y. (2002) A bilayer cell-free protein synthesis system for high-throughput screening of gene products. FEBS Lett. 514, 102–105. 3. Sawasaki, T., Ogasawara, T., Morishita, R. & Endo, Y. (2002) A cell-free protein synthesis system for high-throughput proteomics. Proc.NatlAcad.Sci.USA99, 14652–14657. 4. Hillson, D.A., Lambert, N. & Freedman, R.B. (1984) Formation and isomerization of disulfide bonds in proteins: protein disulfide- isomerase. Methods Enzymol. 107, 281–294. 5. Gething, M.J. & Sambrook, J. (1992) Protein folding in the cell. Nature 355, 33–45. 6. Anand, N.N., Mandel, S., MacKenzie, C.R., Sadowska, J., Sig- urskjold, B., Young, N.M., Bundle, D.R. & Narang, S.A. (1991) Bacterial expression and secretion of various single-chain Fv genes encoding proteins specific for a Salmonella serotype B O-antigen. J. Biol. Chem. 266, 21874–21879. 7. Bird, R.E., Hardman, K.D., Jacobson, J.W., Johnson, S., Kauf- man, B.M., Lee, S., Lee, T., Pope, S.H., Riordan, G.S. & Whitlow, M. (1988) Single-chain antigen-binding proteins. Science 242, 423–426. 8. Zdanov,A.,Li,Y.,Bundle,D.R.,Deng,S.,MacKenzie,C.R., Narang, S.A., Young, N.M. & Cygler, M. (1994) Structure of a single-chain antibody variable domain (Fv) fragment complexed with a carbohydrate antigen at 1.7-A ˚ resolution. Proc. Natl Acad. Sci. USA 91, 6423–6427. 9. Schatz, P.J. (1993) Use of peptide libraries to map the substrate specificity of a peptide-modifying enzyme: a 13 residue consensus peptide specifies biotinylation in Escherichia coli. Biotechnology 11, 1138–1143. 10. Altman, E. & Bundle, D.R. (1994) Polysaccharide affinity columns for purification of lipopolysaccharide-specific murine monoclonal antibodies. Methods Enzymol. 117, 243–253. 11. Jurado, P., Ritz, D., Beckwith, J., Lorenzo, V. & Fernandez, L.A. (2002) Production of functional single-chain Fv antibodies in the cytoplasm of Escherichia coli. J. Mol. Biol. 320, 1–10. 12. Nygren, H., Werthen, M. & Stenberg, M. (1987) Kinetics of antibody binding to solid-phase-immobilised antigen. J. Immunol. Methods 101, 63–71. 13. Pavlickova, P., Knappik, A., Kambhampati, D., Ortigao, F. & Hug, H. (2003) Microarray of recombinant antibodies using a streptavidin sensor surface self-assembled onto a gold layer. Bio- techniques 34, 124–130. 14. Netzer, W.J. & Hartle, F.U. (1997) Recombination of protein domains facilitated by co-translational folding in eukaryotes. Nature 388, 343–349. 15. Ryabova, L.A., Desplancq, D., Spirin, A.S. & Pluckthun, A. (1997) Functional antibody production using cell-free translation: effects of protein disulfide isomerase and chaperones. Nat. Bio- technol. 15, 79–84. 16. Jiang, X., Ookubo, Y., Fujii, I., Nakano, H. & Yamane, T. (2002) Expression of Fab fragment of catalytic antibody 6D9 in an Escherichia coli in vitro coupled transcription/translation system. FEBS Lett. 514, 290–294. 4786 T. Kawasaki et al. (Eur. J. Biochem. 270) Ó FEBS 2003 . Efficient synthesis of a disulfide-containing protein through a batch cell-free system from wheat germ Takayasu Kawasaki 1 , Mudeppa D. Gouda 1 , Tatsuya. 5¢- CAAAAAATTGAATGGCATG AACCGCCGAGCTCCAAC-3¢ and a2 : 5¢-AGCTTCAA AAATATCATTTAAACCCGACGGGCTGCTTTT-3¢ (the sequences for the biotin tag are underlined) followed by circularization