Báo cáo khoa học: Optimization of an Escherichia coli system for cell-free synthesis of selectively 15N-labelled proteins for rapid analysis by NMR spectroscopy

doi:10.1111/j.1432-1033.2004.04346.x

Eur. J. Biochem. 271, 4084–4093 (2004) (cid:1) FEBS 2004

Optimization of an Escherichiacolisystem for cell-free synthesis of selectively 15N-labelled proteins for rapid analysis by NMR spectroscopy

Kiyoshi Ozawa, Madeleine J. Headlam, Patrick M. Schaeffer, Blair R. Henderson, Nicholas E. Dixon and Gottfried Otting

Research School of Chemistry, Australian National University, Canberra, Australia

fugation. Cross-peaks from metabolic by-products were evident in the 15N-HSQC spectra of 13 of the samples. All metabolites were found to be small molecules that could be separated readily from the labelled proteins by dialysis. No significant transamination activity was observed except for [15N]Asp, where an enzyme in the cell extract efficiently converted Asp fi Asn. This activity was suppressed by replacing the normally high levels of potassium glutamate in the reaction mixture with ammonium or potassium acetate. In addition, the activity of peptide deformylase appeared to be generally reduced in the cell-free expression system.

Keywords: cell-free protein synthesis; human cyclophilin A; 15N-HSQC spectrum; NMR; selective stable isotope labe- ling.

Cell-free protein synthesis offers rapid access to proteins that are selectively labelled with [15N]amino acids and suitable for analysis by NMR spectroscopy without chromatographic purification. A system based on an Escherichia coli cell ex- tract was optimized with regard to protein yield and minimal usage of 15N-labelled amino acid, and examined for the presence of metabolic by-products which could interfere with the NMR analysis. Yields of up to 1.8 mg of human cyclophilin A per mL of reaction medium were obtained by expression of a synthetic gene. Equivalent yields were obtained using transcription directed by either T7 or tandem phage k pR and pL promoters, when the reactions were supplemented with purified phage T7 or E. coli RNA polymerase. Nineteen samples, each selectively labelled with a different 15N-enriched amino acid, were produced and analysed directly by NMR spectroscopy after ultracentri-

[2,10,15,16]. At these yields, protein concentrations are sufficiently high to allow the recording of NMR spectra without further concentration of the reaction medium. In particular, 15N-HSQC spectra of selectively 15N-labelled proteins can be recorded without purification of the protein, because only signals from 15N-labelled amide groups are detected [16]. 15N-HSQC spectra present well-resolved, highly sensitive fingerprint information that is uniquely suited to assess the NMR properties of a protein prior to structure determination [17,18].

High-yield, cell-free protein synthesis systems are now available that allow efficient high-throughput production of protein samples for structural genomics and other applications [1–7]. They are also useful for the production of toxic proteins that interfere with cell replication. An attractive application for this method is in production of selectively isotope-labelled samples, as in vitro expression uses much smaller volumes and therefore requires corres- pondingly smaller quantities of expensive isotope-labelled amino acids than conventional in vivo systems. This has been exploited both for the production of stable-isotope labelled proteins for NMR spectroscopy [2,4,7–11], as well as for the incorporation of amino acid analogues [12] such as selenomethionine [13] and 3-iodo-L-tyrosine [14] for X-ray crystallography.

Protein yields of up to 6 mgÆmL)1 have been reported using expression systems based on Escherichia coli extracts

Correspondence to G. Otting, Australian National University, Research School of Chemistry, Canberra, ACT 0200, Australia. Fax: +61 2 61250750, Tel.: +61 2 61256507, E-mail: gottfried.otting@anu.edu.au Abbreviations: aaRS, amino-acyl tRNA synthetase; hCypA, human cyclophilin A; RNAP, RNA polymerase; s-CYPA, synthetic gene encoding hCypA. (Received 24 June 2004, revised 9 August 2004, accepted 27 August 2004)

The present study focussed on optimization of a cell-free E. coli expression system with regard to residue-selective 15N-labelling of proteins. The expression system has been shown previously to be suitable for direct NMR analysis of the reaction mixture, resulting in minimal sample handling [16]. However, the NMR spectroscopic analysis can be complicated by the formation of undesired by-products which originate from metabolic reactions of the 15N-labelled amino acids due to the presence of a broad range of enzymes in the E. coli cell extract used in these reactions [16]. While the amino groups from unincorporated [15N]amino acids do not give rise to cross-peaks in 15N-HSQC spectra due to the rapid proton exchange with the water, some of by-products seem to engage the labelled amino groups in amide bonds.

Here we present a spectral catalogue of metabolic by- products obtained for amino acids with [15N]amino[aN] groups to provide a reference for the NMR analysis of

Cell-free synthesis of 15N-labelled proteins (Eur. J. Biochem. 271) 4085

(cid:1) FEBS 2004

Characterization of proteins

Protein purity was assessed from SDS/PAGE gels that were stained with Coomassie blue. Except where specified, protein concentrations were determined by the method of Bradford [23] using bovine serum albumin as the standard. The molecular masses of purified proteins were confirmed by ESI mass spectrometry, using a VG Quattro II mass spectrometer (VG Biotech, Altrincham, UK). The proteins were extensively dialyzed into 0.1% (v/v) formic acid prior to mass spectrometric analysis.

15N-labelled protein samples produced in high-throughput mode without further purification or concentration steps. In addition, the following issues were addressed: (a) how do yields compare, when transcription is performed by T7 RNA polymerase from a T7 promoter or by E. coli RNA polymerase from tandem phage k pR and pL promoters; (b) does cross-labelling among different amino acids occur due to amino- or amido-transferase activity and, if so, can it be suppressed; (c) which amino acids are prone to the formation of amide-containing by-products; (d) are all by- products sufficiently small to be separated from the protein product by dialysis or ultrafiltration; (e) what are the minimum concentrations of labelled amino acids required for good protein yields and (f) which buffer can be used to replace the large amount of glutamate present in the original medium described by Yokoyama and coworkers [2,15] and also in the buffer of a commercial rapid translation system [11], to enable selective labelling with [15N]Glu?

Cell-free protein synthesis

Materials and methods

L-[15N]cysteine,

The Km values of the 20 aminoacyl-tRNA synthetases (aaRS) for each cognate amino acid, from E. coli strains where possible, were compiled (using the website http:// brenda.bc.uni-koeln.de/) and the aaRSs were categorized into three major groups (Table 1). Groups I–III comprise aaRS with Km values for the appropriate amino acid < 10 lM, between 10 and 50 lM, and between 50 and 500 lM, respectively. Based on these Km values and the frequency of occurrence of each amino acid in the primary sequence of hCypA, the concentrations of [15N]amino acids chosen for hCypA labelling were 50 lM for [15N]Trp and [15N]Tyr, 150 lM for [15N]Ile, [15N]Thr and [15N]His, 0.35 mM for remaining Group II [15N]amino acids, and 1 mM for those in Group III. The S30 cell extract from E. coli strain A19 was prepared by the procedure of Pratt [24], followed by concentration with polyethylene glycol 8000 as described by Kigawa et al. [2].

contained 0.7 mL) In vitro synthesis of PpiB, using the k-promoter vector pKE874 and E. coli RNAP holoenzyme, was essentially as described previously [16]. For in vitro protein synthesis of hCypA, the inner chamber reaction mixtures (total volume 55 mM HEPES/KOH (pH 7.5), 1.7 mM dithiothreitol, 1.2 mM ATP, 0.8 mM

Table 1. Classification of the 20 aminoacyl-tRNA synthetases by Km values. Data compiled from http://brenda.bc.uni-koeln.de/. All values are for E. coli unless noted otherwise.

Enzyme Km (mM) Group Enzyme Km (mM) group Group

ThrRSa (cid:1) 0.002 TrpRSb ¼ 0.005 (cid:1) 0.005 IleRS TyrRS (cid:1) 0.008 HisRS (cid:1) 0.008 ArgRS ¼ 0.011 CysRSc ¼ 0.013 PheRSd (cid:1) 0.028 AsnRS (cid:1) 0.032 ProRSe

–

I I I I I II II II II II

LeuRS (cid:1) 0.05 GluRS (cid:1) 0.05 AspRS (cid:1) 0.06 SerRS ¼ 0.07 MetRS (cid:1) 0.08 ValRS (cid:1) 0.10 GlnRS (cid:1) 0.15 GlyRS (cid:1) 0.16 (cid:1) 0.34 AlaRS LysRSf –

III III III III III III III III III III

Materials L-[U-15N]Arginine.HCl, L-[15N]aspartic acid, L-[U-15N]aspa- L-[15N]glutamic ragine.H2O, acid, L-[15N]glutamine[aN], [15N]glycine, L-[15N]histidine[aN]. HCl, L-[15N]methionine, L-[15N]isoleucine, L-[15N]phenyl- alanine, L-[15N]serine, L-[U-15N]tryptophan, L-[15N]tyrosine and L-[15N]valine were purchased from Cambridge Isotope (Andover, MA, USA). L-[15N]Alanine, Laboratories L-[15N]leucine and L-[15N]threonine were obtained from Spectra Stable Isotopes (Columbia, MD, USA) and L-[15N]lysine[aN].2HCl was from Euriso-top (Saint-Aubin, France). Oligonucleotides were purchased from Auspep (Parkville, Australia). Vent DNA polymerase and RNase inhibitor were from Promega, and creatine kinase and E. coli total tRNA were from Roche. E. coli RNA poly- merase (RNAP) holoenzyme was purified as described previously [16]. Spectra/Por 2 dialysis tubing was purchased from Spectrum Laboratories Inc. (Rancho Dominguez, CA, USA).

a Value for Saccharomyces carlsbergensis, no data available for E. coli; b value for Lupinus luteus and bovine, no data available for E. coli; c value for Paracoccus denitrificans, no data available for E. coli. d value for Bacillus subtilis, no data available for E. coli; e no data available, assigned to group II based on side-chain rigidity; f no data available, assigned to group III based on simi- larity with Glu and Met.

E. coli strains A19 [16], BL21(DE3)/pLysS [19] and BL21(DE3)recA [20] were as described previously. The plasmid pKE874 [16] was used for production of the E. coli peptidyl-prolyl cis-trans isomerase PpiB under the control of tandem phage k pR and pL promoters. Plasmid pBH964 was used for cell-free expression of a synthetic gene (s-CYPA) that encodes human cyclophilin A (hCypA) under control of the phage T7 promoter in vector pETMCSI [21]. hCypA was produced in vivo in strain BL21(DE3)/pLysS/pBH964 as a standard for comparison with protein production using the cell-free system. The construction of the s-CYPA gene and plasmid pBH964, as well as the procedure for purification of hCypA, are described in detail in the Supplementary material.

Plasmid pKO1166 contains phage T7 gene 1 (which encodes T7 RNAP) under the transcriptional control of the phage k pL promoter in vector pMA200U [22]. T7 RNAP was produced in vivo using strain BL21(DE3)recA/ pKO1166. Construction of pKO1166, and the production and purification of the protein are also described in the Supplementary material.

4086 K. Ozawa et al. (Eur. J. Biochem. 271)

(cid:1) FEBS 2004

80 mM creatine

total

each of CTP, GTP and UTP, 0.64 mM 3¢,5¢-cyclic AMP, 68 lM folinic acid, 27.5 mM ammonium acetate, 208 mM potassium glutamate, phosphate, 250 lgÆmL)1 creatine kinase, a [15N]amino acid (at the concentration given above), 1 mM each of 19 unlabelled L-amino acids, 15 mM magnesium acetate, 175 lgÆmL)1 E. coli tRNA, 0.05% (w/v) NaN3, 168 lL of (containing 5.2 mg of concentrated S30 extract total protein), 16 lgÆmL)1 of supercoiled plasmid DNA (pBH964, as described above), 150 U of RNase inhibitor and 93 lgÆmL)1 of T7 RNAP. For labelling with [15N]Glu, ammonium or potassium acetate (200 mM) was used instead of potassium glutamate (208 mM).

Even with promoters recognized by E. coli RNAP, supplementation of the S30 extract with RNAP is important for good protein yields [16]. When using T7 promoters, addition of T7 RNAP was critically required. T7 RNAP, when purchased from commercial sources, is the most expensive component of our cell-free system. T7 RNAP is, however, relatively easy to produce and purify, as it is a much smaller and simpler protein than E. coli RNAP holoenzyme, which is composed of five different subunits (a2bb¢xr) [25]. We adapted published methods [26] to develop a simple procedure for the isolation of T7 RNAP from a strain containing a plasmid that directed overproduction of the protein under control of the heat- inducible k pL promoter. Up to 25 mg of pure T7 RNAP could be obtained within a few days from 5 g of cells. This is sufficient for many hundreds of in vitro reactions.

Our initial attempts to purify T7 RNAP were hindered by proteolytic cleavage. This was controlled by use of an ompT– E. coli host strain [BL21(DE3)recA], limitation of the time cultures were treated at 42 (cid:2)C during induction of synthesis of T7 RNAP to 30 min, followed by treatment for 2 h at 40 (cid:2)C [26], and use of a-toluenesulphonyl fluoride in the buffer during cell lysis.

The inner chamber reaction mixtures were dialyzed in Spectra/Por 2 tubing with a nominal size cutoff of 12–14 kDa for 10 or 12 h at 37 (cid:2)C with gentle shaking against the outer chamber solution (14 mL) that was changed after 3 and 6 h. The inner chamber assembly and the outer chamber buffer were housed within a 50-mL polypropylene tube. The outer chamber solution had the same composition as the inner chamber reaction mixture, except that S30 extract, tRNA, plasmid DNA, T7 RNAP, creatine kinase and RNase inhibitor were omitted, and the concentration of magnesium acetate was increased to 19.3 mM. For SDS/PAGE analysis, in vitro reaction samples were diluted two-fold with gel loading mix containing 2% (w/v) SDS and heated for 2 min at 90 (cid:2)C. The reaction mixtures containing hCypA were clarified by ultracentrifu- gation (100 000 g, 4 h) and stored at 4 (cid:2)C.

NMR spectroscopy

t2max ¼ 102 ms and total

E. coli PpiB [16] was produced equally well in vitro when using either T7 or k promoters (data not shown). Corre- spondingly similar yields were expected for hCypA, as PpiB and hCypA are functional homologues with similar three- dimensional structures and amino acid compositions, and the codon usage of the CYPA gene had been adjusted for the E. coli expression system by construction of an artificial gene. The protein yields obtained in vitro for hCypA with the T7 promoter system were about 1.5–1.8 mgÆmL)1 of cell-free reaction medium, and were indeed closely compar- able to those obtained for PpiB with either promoter system (Fig. 1, lane 8 and Fig. 2, lane 1). With transcription under the control of the k promoters, however, cell-free synthesis of hCypA was below the detection limit of SDS/PAGE with Coomassie blue staining, even though the same plasmid produced excellent yields in vivo (data not shown). More- over, PpiB was always produced in vitro as a fully soluble protein, whereas a portion (15–20%) of hCypA was invariably found in the insoluble fraction. This is probably due to the pI of hCypA being close to the pH of the reaction mixture (pH 7.5); the pI value of PpiB is about one unit lower.

All NMR spectra were recorded at 25 (cid:2)C using a Varian INOVA 600 MHz NMR spectrometer equipped with a probe operating at room temperature. 15N-HSQC spectra were recorded with 5 mm sample tubes using t1max ¼ 32 ms, recording times of 17–24 h. NMR measurements were made using the super- natant from the clarified reaction mixtures after addition of 10% (v/v) D2O to provide a lock signal. In addition, spectra were recorded after dialysis of the samples overnight at 4 (cid:2)C in Spectra/Por 2 tubing, against buffer comprising 10 mM sodium phosphate (pH 6.5), 100 mM NaNO3, 5 mM dithiothreitol and 50 lM NaN3. The dialyzed samples were concentrated to a final volume of about 0.6 mL using Millipore Ultra-4 centrifugal filters (MWCO 10 000) and D2O was added to a final concentration of 10% (v/v) before NMR measurement.

Results

In order to achieve maximal yields in the cell-free system, smaller mass amounts of T7 RNAP were required than E. coli RNAP, when the same proteins were produced under control of the T7 and k promoters, respectively (data not shown). A further advantage of the T7 system lies in its to temperature changes: better tolerance with respect lowering the reaction temperature from 37 to 30 (cid:2)C decreased protein yields insignificantly, whereas this tem- perature change resulted in decreased yields when the expression was under control of the k promoters. This may be because pKE874 also directs low-level synthesis of a thermolabile version of the k cI repressor, which may repress transcription by E. coli RNAP at 30 (cid:2)C. The optimized concentrations of T7 or E. coli RNAP used in the present work were found to be suitable for production of many different proteins. Cell-free protein synthesis enhanced by T7 RNA polymerase Prior to the preparation of 15N-labelled samples of hCypA and analysis by NMR spectroscopy, the performance of the cell-free expression system was explored with respect to a number of parameters. We first examined the yields obtained when transcription is conducted by T7 RNAP from the T7 promoter rather than by E. coli RNAP holoenzyme from phage k promoters as in our previous work [16].

Cell-free synthesis of 15N-labelled proteins (Eur. J. Biochem. 271) 4087

(cid:1) FEBS 2004

Optimization of other conditions for invitro protein synthesis

Several other parameters were assessed individually to maximize protein yields. In particular, the optimal quan- tity of concentrated S30 extract in the reaction mixture was determined for each new preparation, but results from several batches nevertheless gave similar results. In addition, the optimal concentrations of MgCl2 and template DNA were found to vary with different batches of S30 extract. Concentration of the S30 extract by dialysis against a solution of PEG 8000 [2] reduced the volume of the in vitro reaction mixture, but had little effect on protein yields.

Fig. 1. Cell-free synthesis of PpiB under control of phage k promoters. Identical volumes of reaction products were loaded into lanes of a 15% SDS/polyacrylamide gel, which were stained with Coomassie blue. Lanes 1, 3, 5 and 7: reaction mixtures before the start of in vitro synthesis of PpiB, with transcription by E. coli RNAP from tandem phage k promoters (0 h reactions). Lanes 2, 4, 6 and 8: corresponding mixtures after synthesis for 12 h at 37 (cid:2)C. Each amino acid was present at a concentration of 10 lM (lanes 1 and 2), 30 lM (lanes 3 and 4), 300 lM (lanes 5 and 6) or 1 mM (lanes 7 and 8). Mobilities of molecular mass markers (kDa) were as indicated.

The concentration of tRNA was found to affect the yields of proteins. For production of aspartyl-tRNA synthetase [16], for example, the optimal tRNA concentration was about 45 lgÆmL)1, whereas tRNA concentrations of 87 and 175 lgÆmL)1 worked equally well for PpiB and hCypA. Proteins produced and stored in the reaction mixture appeared to be stable with respect to proteolysis. After two months of storage at 4 (cid:2)C, hCypA was not significantly degraded, as evaluated by NMR measurements.

Cell-free synthesis of PpiB with increasing concentra- tions of amino acids showed that almost no protein was synthesized when each amino acid was present at 10 lM (Fig. 1, lane 2). This result confirmed that the extract is depleted in natural amino acids. Absence of free amino acids from the cell extract is a prerequisite for efficient incorporation of labelled amino acids into target proteins. Furthermore, the protein yields increased with increasing amino acid concentration (Fig. 1, lanes 1–8), indicating that the concentrations of amino acids limit protein synthesis at < 1 mM. After ultracentrifugation to remove ribosomes and ribosome-associated translation factors, the target proteins were found to be the most abundant protein in the reaction mixtures, provided the amino acids (Fig. 1). had been supplied at high concentrations Remarkably, reduction of the concentration of amino acids to 30 lM lowered the yield only by about 50% (Fig. 1, lanes 4 and 8).

Fig. 2. Cell-free synthesis of hCypA under control of the T7 promoter with minimized concentrations of labelled amino acids from the three different groups defined in Table 1. The samples were centrifuged for 4 h at 100 000 g before analysis by SDS/PAGE (15%). The gel was loaded with the soluble fractions (supernatants) of hCypA synthesized in vitro during 10 h at 37 (cid:2)C with different concentrations of 15N-labelled amino acids (all other amino acids were at 1 mM), and stained with Coomassie blue. Lane 1: 1 mM [15N]Glu; lane 2: 50 lM [15N]Tyr; lane 3: 150 lM [15N]Thr; lane 4; 350 lM [15N]Asn; lane 5; 1 mM [15N]His. These fractions were subjected to NMR measurements without further purification (Fig. 3). Mobilities of molecular mass markers (kDa) were as indicated. The [15N]Glu-labelled hCYPA sample in lane 1 was produced with 200 mM ammonium acetate in the reaction buffer, whereas the samples in the other lanes were produced with 208 mM potassium glutamate.

Invitrosynthesis of hCypA selectively labelled with [15N]amino acids

The achievable yields of proteins would be expected to depend on the concentrations of available amino-acylated tRNAs. As the loading efficiency of the amino-acyl tRNA synthetases depends on their Km values, amino acids processed by synthetases with low Km values are expected to be more readily available for protein synthesis than others. To limit the expense of use of labelled amino acids, their concentrations in the inner and outer chambers were adjusted according to their frequency in the primary structure of the protein and according to the Km values of their respective tRNA synthetases (Table 1). Our results confirmed that labelled amino acids from Groups I and II (Table 1) could be used at reduced concentrations (still several-fold above the respective Km values) without signi- ficantly affecting protein yields. Figure 2 shows a compar- ison of yields obtained for hCypA with substantially reduced concentrations of Tyr, Thr and Asn, compared to

4088 K. Ozawa et al. (Eur. J. Biochem. 271)

(cid:1) FEBS 2004

those obtained when all amino acids were at 1 mM. The similarity in protein production levels is corroborated by the similarity in cross-peak intensities observed in 15N-HSQC NMR spectra recorded of the reaction mixtures (Fig. 3).

than the average protein cross-peaks. In all cases, the cross- peaks from metabolic by-products were removed easily by dialysis of the sample prior to NMR analysis (data not shown). For proteins with a molecular mass similar to hCypA, dialysis can be achieved simply by transfer of the tubing containing the reaction mixture to a buffer suitable for NMR measurements. The Spectra/Por 2 dialysis tubing (12–14 kDa cut-off) was found to retain a 9.5-kDa globular protein under these conditions. Comparison of spectra obtained with selectively 15N-labelled PpiB and hCypA showed closely related sets of metabolite HSQC cross-peaks (data not shown).

The standard reaction mixture [2,15,16] contains a high concentration of potassium glutamate, which makes it unsuitable for selective labelling of Glu residues in target proteins. A buffer with 208 mM potassium D-glutamate performed equally well as one with L-glutamate, suggesting that glutamate served as an osmolyte rather than playing a more specific role. However, this buffer was still unsuitable for selective labelling with Glu, presumably because of the presence of glutamate racemase in the S30 extract. The glutamate could not be substituted by 200 mM betaine, which had an inhibitory effect on the synthesis of PpiB and hCypA. In contrast, reaction mixtures with ammonium or potassium acetate instead of potassium glutamate were found to perform well. Maximal yields of PpiB and hCypA were obtained with either acetate salt at concentrations in the range of 200–230 mM (e.g. Figure 2, lane 1). Further- more, this alternate medium suppressed amino- or amido- transferase activities which otherwise incorporated the a-amino-nitrogen of Asp into the a- and side-chain NH2 groups of Asn (see below).

There was no evidence for transaminase activity except for the [15N]Asp-labelled sample produced in the conven- tional way in the presence of potassium glutamate. Under those conditions, cross-peaks of the backbone and side- chain amides of the [15N]Asn-labelled sample were also observed in the NMR spectrum of[15N]Asp-labelled hCypA (fourth panel of Fig. 3). The intensities of the undesired [15N]Asn cross-peaks were about two thirds of those of the [15N]Asp cross-peaks, indicating highly efficient transami- nation/transamidation. This suggests that the labelled amino group is not released in the form of ammonia because it would have been diluted by the presence of 27.5 mM ammonium present in the reaction buffer. The E. coli asparagine synthetases A and B (asnA and asnB gene products) synthesize Asn from Asp and may be responsible for this activity in the S30 extract. Remarkably, no evidence of transamination was observed in the [15N]Asn sample. (The additional cross-peaks in Fig. 3 are due to the side- chain amides which were labelled in the [U-15N]Asn substrate). We further noted that the amido/aminotrans- ferase activity could be suppressed by replacing potassium glutamate in the buffer by ammonium or potassium acetate (last panel of Fig. 3). This new buffer did not affect the production levels of the proteins tested (hCypA, PpiB and ubiquitin) to a significant degree (e.g. see lane 1 in Fig. 2). We also compared the buffers with 200 mM potassium acetate or 200 mM ammonium acetate for the synthesis of [15N]Asp- and [15N]Glu-hCypA with regard to the number and positions of HSQC cross-peaks from metabolites. No significant differences were observed (data not shown).

15N-HSQC NMR spectra of selectively 15N-labelled hCypA 15N-HSQC spectra were recorded of hCypA samples produced in vitro with 19 different 15N-labelled amino acids. Spectra with acceptable sensitivity could be recorded at 25 (cid:2)C using the reaction mixtures at pH 7.5 (Fig. 3). Sample handling was kept to a minimum to explore the potential of this methodology for high-throughput protein analysis. Although spectra recorded before and after ultracentrifugation were not significantly different, all data presented in Fig. 3 were recorded after the ribosomes and other macromolecular assemblies had been removed by ultracentrifugation to avoid the formation of a precipitate during data acquisition. The NMR chemical shifts of purified hCypA are known [27], allowing the identification of individual cross-peaks from the protein and detection of any additional cross-peaks due to metabolic by-products. The spectra recorded for hCypA produced with 15N-labelled Asn, Gln, Ile, Leu, Phe and Tyr contained only cross-peaks from the protein. Samples with 15N-labelled Arg, Asp, His, Lys, Met, Thr and Val contained a few additional cross- peaks, due to limited metabolic conversion of the labelled less amino acids. The additional peaks were, however, intense than the average of the protein cross-peaks. Samples with 15N-labelled Ala, Asp, Cys, Glu, Gly, Ser and Trp contained additional cross-peaks that were more intense

Fig. 3. 15N-HSQC spectra of hCypA selectively labelled with 15N-amino acids. The spectra were recorded at 25 (cid:2)C and pH 7.5 using the in vitro reaction mixture after centrifugation (100 000 g, 4 h) and addition of 10% D2O. The assignments of the backbone amide cross-peaks are indicated by the one-letter amino acid symbols and the sequence numbers. Squares identify the cross peaks which could be assigned according to the previously published assignment [27]. Circles identify the cross-peaks from metabolites. Dotted squares mark the positions of cross-peaks which were assigned at pH 6.5 [27] but are absent from the present spectra or observable only at lower plot levels. The spectrum recorded with [l5N]Asn also contains the cross-peaks from the side-chain amide groups (backbone and side-chain NH groups were labelled in the amino acid used in the reaction). The cross-peak from the side-chain NH of W121 is labelled W121e. Question marks in the spectra of [15N]Met and [15N]Val hCypA identify tentative new assignments of cross-peaks which had not been assigned previously [27].

Quite generally, the cell-free expression system allows the use of high concentrations of unlabelled amino acids in the reaction mixtures to dilute the effects of incorporation of amino acids that are 15N-labelled by transamination. In this situation, the main effect of transamination is a reduction of the degree of 15N-labelling in the amino acid of interest. In the case of [15N]Gly, much of the label appeared to have been trapped in an amide group of a low molecular mass metabolite (Fig. 3).

Cell-free synthesis of 15N-labelled proteins (Eur. J. Biochem. 271) 4089

(cid:1) FEBS 2004

4090 K. Ozawa et al. (Eur. J. Biochem. 271)

(cid:1) FEBS 2004

Fig. 3. (Continued).

Cell-free synthesis of 15N-labelled proteins (Eur. J. Biochem. 271) 4091

(cid:1) FEBS 2004

the relative complexity of targets for favourable folding characteristics than PpiB. Alternat- the E. coli RNAP ively, holoenzyme may offer more inhibitory protein–protein interactions.

The cell-free expression system lends itself to the addition of specific enzyme inhibitors, such as the alanine racemase inhibitor, b-chloro-L-alanine [28]. When we tested the effect of adding b-chloro-L-alanine (0.5 mM) to the reaction [15N]Ala labelled PpiB, mixture for the synthesis of however, only a single cross-peak belonging to a minor metabolite was removed from the set of metabolite signals (data not shown). The metabolic enzymes and pathways in E. coli are not sufficiently well understood to suppress all side reactions in this way.

Whereas the present study required significant NMR measurement time to record each 15N-HSQC spectrum, this is no longer prohibitive with the increased sensitivity available on high-field NMR spectrometers equipped with cryogenic probeheads [16]. We thus envisage that parallel production of a large number of selectively labelled protein samples followed by the recording of 15N- or 13C-HSQC spectra will provide a practical approach to support resonance assignments, provided that sample handling can be kept to a minimum. The present study shows that this latter condition is easily fulfilled. For most amino acids, purification of the produced 15N-labelled protein is not required for identification of the HSQC cross-peaks from the protein. In the few cases where cross-peaks from metabolites and protein could overlap, a simple dialysis step is sufficient to remove the metabolites. The easy removal of the interfering signals from metabolites is a clear advantage of the cell-free expression system over in-cell NMR analyses [29,30]. Interestingly, only low-mass metabolites appear to be produced also when [U-15N]protein is synthesized in vivo using ammonium chloride [31].

(18 040 Da) and (cid:1) 10% had lost Single additional peaks present in the 15N-HSQC spec- trum of [15N]Met and [15N]Val labelled hCypA were still present after dialysis and were attributed to the amide protons of Met1 and Val2. These had not been reported in the original assignment of hCypA [27], in agreement with the frequent observation that the HN resonances of amino- terminal residues are broadened beyond detection by fast base-catalyzed exchange with the water. In the case of our cell-free expression system, the amino-terminal Met of PpiB has been observed (by proteolysis and mass spectrometry) to remain predominantly N-formylated due to saturation of the activity of peptide deformylase (D. Mouradov and T. Huber, unpublished data). In the present case of hCypA, retention of the N-formyl group enables the observation of the cross-peaks from the amino-terminal amide protons. Interestingly, ESI mass spectrometric analysis of hCypA produced in vivo in E. coli also showed N-terminal hetero- geneity. Although (cid:1) 70% of the protein started with a deformylated Met (18 012 Da), (cid:1) 20% had N-terminal N-formyl-Met the N-terminal Met (17 881 Da).

Most importantly, the transfer of the 15N-label to other amino acids is insignificant for 18 of the amino acids and can be suppressed for [15N]Asp by use of a modified medium in which glutamate is replaced by acetate. Notably, these results were achieved without preparation of extracts from auxotrophic E. coli strains. Replacement of glutamate by acetate has recently also been described by Klammt et al. [7].

Cell-free expression kits for large-scale protein produc- tion have become commercially available, making cell-free expression a readily accessible technology [7,11,32]. Our strategy extends the applicability of the system with regard to protein analysis by NMR spectroscopy. The catalogue of 15N-HSQC spectra in Fig. 3 provides the basis for the straightforward identification of metabolite cross-peaks by visual inspection, as their generation is insensitive to the identity of the target protein. In our experiments, weak cross-peaks with broad line shapes were difficult to detect. Hence not all signals assigned previously [27] could be observed. As the original assign- ments had been reported for pH 6.5, we measured the 15N-HSQC spectra after dialysis at this pH. The lower pH value enhanced some of the signals as expected. For example, the cross-peak of Asp9 was observed at pH 6.5 (data not shown), whereas it was missing from the spectrum recorded at pH 7.5 (Fig. 3). In contrast to the previously reported assignment [27], there was no evidence for more than a single conformation of His70, neither at pH 7.5 nor at pH 6.5. Ultimately,

Discussion

In this study, several aspects of a cell-free protein synthesis system based on an E. coli cell extract were investigated and optimized. One point is the observation that T7 RNA polymerase performs as well as the E. coli holoenzyme in the in vitro coupled transcription/transla- tion system. As many modern plasmid vectors used for protein overproduction are based on transcription from the T7 promoter, it is convenient if the same vectors can be used in the cell-free system. In the case of hCypA, good protein yields were obtained with transcription by T7 RNAP from the T7 promoter, whereas no production could be detected in vitro with a phage k promoter and transcription by the E. coli RNAP holoenzyme. However, no difference between the two systems was observed for the closely related protein PpiB. This phenomenon may be explained in different ways. Possibly, hCypA has less the formation of metabolites could be avoided by the use of a completely reconstituted cell-free system comprising only the minimum set of enzymes necessary for translation [33]. Currently achievable yields, however, do not justify the effort associated with the purification of the large number of enzymes required. If all metabolites generated by the E. coli S30 extract could be identified, it might be possible to suppress their production by the use of appropriate enzyme inhibitors. This would provide the benefit of less isotopic dilution and thereby improved labelling efficiency and enhanced sensitivity of the NMR analysis. In the absence of enzyme inhibitors, increased yields may be obtained by providing the rapidly metabolized amino acids in excess, beyond the quantities suggested by the Km values of the aminoacyl-tRNA synthetases (Table 1) and the number of amino acids present in the protein. The data of Fig. 3 provide a guideline for a corresponding rebalancing of amino acid concentra- tions. They apply equally to the in-cell NMR analysis [30] of selectively labelled proteins.

4092 K. Ozawa et al. (Eur. J. Biochem. 271)

(cid:1) FEBS 2004

Acknowledgements

extract for highly productive cell-free protein expression. J. Struct. Funct. Genomics 5, 63–68.

16. Guignard, L., Ozawa, K., Pursglove, S.E., Otting, G. & Dixon, N.E. (2002) NMR analysis of in vitro-synthesized proteins without purification: a high-throughput approach. FEBS Lett. 524, 159–162.

We thank Dr Simon Bennett for measurements of ESI mass spectra. K.O. and G.O. thank the Australian Research Council for a CSIRO- Australian Postdoctoral and Federation Fellowships, respectively. Financial support by the Australian Research Council is gratefully acknowledged.

References

17. Yamazaki, T., Yoshida, M., Kanaya, S., Nakamura, H. & Nagayama, K. (1991) Assignments of backbone 1H, 13C, and 15N resonances and secondary structure of ribonuclease H from Escherichia coli by heteronuclear three-dimensional NMR spec- troscopy. Biochemistry 30, 6036–6047.

1. Spirin, A.S., Baranov, V.I., Ryabova, L.A., Ovodov, S.Y. & Alakhov, Y.B. (1988) A continuous cell-free translation system capable of producing polypeptides in high yield. Science 242, 1162–1164.

2. Kigawa, T., Yabuki, T., Yoshida, Y., Tsutsui, M., Ito, Y., Shi- bata, T. & Yokoyama, S. (1999) Cell-free production and stable- isotope labeling of milligram quantities of proteins. FEBS Lett. 442, 15–19.

18. Yee, A., Chang, X., Pineda-Lucena, A., Wu, B., Semesi, A., Le, B., Ramelot, T., Lee, G.M., Bhattacharyya, S., Gutierrez, P., Deni- sov, A., Lee, C.-H., Cort, J.R., Kozlov, G., Liao, J., Finak, G., Chen, L., Wishart, D., Lee, W., McIntosh, L.P., Gehring, K., Kennedy, M.A., Edwards, A.M. & Arrowsmith, C.H. (2002) An NMR approach to structural genomics. Proc. Natl Acad. Sci. USA 99, 1825–1830.

19. Studier, F.W., Rosenberg, A.H., Dunn, J.J. & Dubendorff, J.W. (1990) Use of T7 RNA polymerase to direct expression of cloned genes. Methods Enzymol. 185, 60–89.

3. Yokoyama, S. (2003) Protein expression systems for structural genomics and proteomics. Curr. Opin. Chem. Biol. 7, 39–43. 4. Shi, J., Pelton, J.G., Cho, H.S. & Wemmer, D.E. (2004) Protein signal assignments using specific labeling and cell-free synthesis. J. Biomol. NMR 28, 235–247.

5. Swartz, J.R., Jewett, M.C. & Woodrow, K.A. (2004) Cell-free protein synthesis with prokaryotic combined transcription-trans- lation. Methods Mol. Biol. 267, 169–182.

20. Williams, N.K., Prosselkov, P., Liepinsh, E., Line, I., Sharipo, A., Littler, D.R., Curmi, P.M.G., Otting, G. & Dixon, N.E. (2002) In vivo protein cyclization promoted by a circularly permuted Synechocystis sp. PCC6803 DnaB mini-intein. J. Biol. Chem. 277, 7790–7798.

6. Jewett, M.C. & Swartz, J.R. (2004) Mimicking the Escherichia coli cytoplasmic environment activates long-lived and efficient cell-free protein synthesis. Biotech. Bioeng. 86, 19–26.

21. Neylon, C., Brown, S.E., Kralicek, A.V., Miles, C.S., Love, C.A. & Dixon, N.E. (2000) Interaction of the Escherichia coli replica- tion terminator protein (Tus) with DNA: a model derived from DNA-binding studies of mutant proteins by surface plasmon resonance. Biochemistry 39, 11989–11999.

7. Klammt, C., Lo¨ hr, F., Scha¨ fer, B., Haase, W., Do¨ tsch, V., Ru¨ terjans, H., Glaubitz, C. & Bernhard, F. (2004) High-level cell- free expression and specific labeling of integral membrane pro- teins. Eur. J. Biochem. 271, 568–580.

22. Elvin, C.M., Thompson, P.R., Argall, M.E., Hendry, P., Stam- ford, N.P.J., Lilley, P.E. & Dixon, N.E. (1990) Modified bacter- iophage lambda promoter vectors for overproduction of proteins in Escherichia coli. Gene 87, 123–126.

8. Kigawa, T., Muto, Y. & Yokoyama, S. (1995) Cell-free synthesis and amino acid-selective stable isotope labeling of proteins for NMR analysis. J. Biomol. NMR 6, 129–134.

In Transcription

systems.

cell-free

9. Yabuki, T., Kigawa, T., Dohmae, N., Takio, K., Terada, T., Ito, Y., Laue, E.D., Cooper, J.A., Kainosho, M. & Yokoyama, S. (1998) Dual amino acid-selective and site-directed stable-isotope labeling of the human c-Ha-Ras protein by cell-free synthesis. J. Biomol. NMR 11, 295–306.

23. Bradford, M.M. (1976) A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding. Anal. Biochem. 72, 248–254. 24. Pratt, J.M. (1984) Coupled transcription-translation in prokary- and Translation otic (Hames, B.D. & Higgins, S.J., eds), pp. 179–209. IRL Press, Oxford, UK.

25. Young, B.A., Gruber, T.M. & Gross, C.A. (2002) Views of tran-

scription initiation. Cell 109, 417–420.

10. Kariya, E., Ohki, S.-Y., Hayano, T. & Kainosho, M. (2000) Backbone 1H, 13C, and 15N resonance assignments of an 18.2 kDa protein, E. coli peptidyl-prolyl cis-trans isomerase b (EPPIb). J. Biomol. NMR 18, 75–76.

11. Parker, M.J., Aulton-Jones, M., Hounslow, A.M. & Craven, C.J. (2004) A combinatorial selective labeling method for the assign- ment of backbone amide NMR resonances. J. Am. Chem. Soc. 126, 5020–5021.

26. Tabor, S. & Richardson, C.C. (1985) A bacteriophage T7 RNA polymerase/promoter system for controlled exclusive expres- sion of specific genes. Proc. Natl Acad. Sci. USA 82, 1074–1078. 27. Ottiger, M., Zerbe, O., Gu¨ ntert, P. & Wu¨ thrich, K. (1997) The NMR solution conformation of unligated human cyclophilin A. J. Mol. Biol. 272, 64–81.

12. Hendrickson, T.L. de Cre´ cy-Lagard V. & Schimmel, P. (2004) Incorporation of nonnatural amino acids into proteins. Annu. Rev. Biochem. 73, 147–176.

28. Kato, K., Matsunaga, C., Igarashi, T., Kim, H.-H., Odaka, A., Shimada, I. & Arata, Y. (1991) Complete assignment of the methionyl carbonyl carbon resonances in switch-variant anti-dansyl antibodies labeled with [1–13C]methionine. Biochem- istry 30, 270–278.

29. Ou, H.D., Lai, H.C., Serber, Z. & Do¨ tsch, V. (2001) Efficient identification of amino acid types for fast protein backbone assignments. J. Biomol. NMR 21, 269–273.

30. Serber, Z., Ledwidge, R., Miller, S.M. & Do¨ tsch, V. (2001) Eva- luation of parameters critical to observing proteins inside living Escherichia coli by in-cell NMR spectroscopy. J. Am. Chem. Soc. 123, 8895–8901.

13. Kigawa, T., Yamaguchi-Nunokawa, E., Kodama, K., Matsuda, T., Yabuki, T., Matsuda, N., Ishitani, R., Nureki, O. & Yokoyama, S. (2002) Selenomethionine incorporation into a protein by cell-free synthesis. J. Struct. Funct. Genomics 2, 29–35. 14. Kiga, D., Sakamoto, K., Kodama, K., Kigawa, T., Matsuda, T., Yabuki, T., Shirouzu, M., Harada, Y., Nakayama, H., Takio, K., Hasegawa, Y., Endo, Y., Hirao, I. & Yokoyama, S. (2002) An engineered Escherichia coli tyrosyl-tRNA synthetase for site- specific incorporation of an unnatural amino acid into proteins in eukaryotic translation and its application in a wheat germ cell-free system. Proc. Natl Acad. Sci. USA 99, 9715–9720.

15. Kigawa, T., Yabuki, T., Matsuda, N., Matsuda, T., Nakajima, R., Tanaka, A. & Yokoyama, S. (2004) Preparation of Escherichia coli

31. Gronenborn, A.M. & Clore, G.M. (1996) Rapid screening for structural integrity of expressed proteins by heteronuclear NMR spectroscopy. Protein Sci. 5, 174–177.

Cell-free synthesis of 15N-labelled proteins (Eur. J. Biochem. 271) 4093

(cid:1) FEBS 2004

32. Elbaz, Y., Steiner-Mordoch, S., Danieli, T. & Schuldiner, S. (2004) In vitro synthesis of fully-functional EmrE, a multidrug trans- porter, and study of its oligomeric state. Proc. Natl Acad. Sci. USA 101, 1519–1524.

33. Shimizu, Y., Inoue, A., Tomari, Y., Suzuki, T., Yokogawa, T., Nishikawa, K. & Ueda, T. (2001) Cell-free translation recon- stituted with purified components. Nat. Biotechnol. 19, 751–755.

Supplementary material

is available from http://www. The following material blackwellpublishing.com/products/journals/suppmat/EJB/ EJB4346/EJB4346sm.htm

the synthetic CYPA gene Fig. S1. Construction of (s-CYPA). The gene was constructed following manipula- tion of the nucleotide sequence to increase its identity with the E. coli ppiB gene, using a combination of ligation and recursive overlap extension of complementary synthetic oligonucleotides as described above. Oligonucleotides used for the construction of the s-CYPA are identified by arrows above the complete gene sequence. The NdeI, ApaI, FokI, XhoI, EcoRI and NcoI restriction endonuclease sites are boxed. The start codon (ATG) is within the NdeI site and the stop codon (TAA) is identified by a black box in the linker. Fig. S2. Plasmid pKO1166. This plasmid, which directs overproduction of T7 RNA polymerase, was constructed by insertion of a DNA fragment bearing T7 gene 1 under control of the bacteriophage k pL promoter into vector pMA200U [5]. Details of the construction of the synthetic s-CYPA gene, purification of cyclophilin A and expression and purification of T7 RNA polymerase are given and reference to the following supplementary figures is made: