Molecular Cloning of a cDNA Encoding Chicken T-protein of the Glycine Cleavage System and Expression of the Functional Protein in Escherichia coli EFFECT OF mRNA SECONDARY STRUCTURE IN THE TRANSLATIONAL INITIATION REGION ON EXPRESSION*

DNA clones encoding chicken T-protein of the gly- cine cleavage system were isolated from chicken liver XgtlO cDNA libraries. Three overlapping clones pro- vided an open reading frame of 11 76 nucleotides that predicts a polypeptide of 392 amino acids (Mr 42,056) comprised of a 16-residue mitochondrial targeting se- quence and a 376-residue mature protein (Mr 40,292). The amino acid sequence predicted for the mature pro- tein showed 67% identity with that of bovine T-pro-tein. A cDNA encoding mature T-protein was constructed, and the nucleotide sequence just downstream of the initiation codon was modified without amino acid substitution to reduce the free energy of formation for the folded mRNA. Expression plasmids containing these cDNA variants produced large amounts of T- protein in Escherichia coli, while very low expression was observed with a plasmid containing wild type cDNA. Enzymatically active T-protein was obtained when the expression was conducted at 30 OC with 25 WM isopropyl-1-thio-8-D-galactopyranoside. Under the full inducing condition (at 37 O C and 1 mM inducer), the expressed T-protein was recovered as insoluble and inactive protein. The recombinant T-protein was purified to near homogeneity with a yield of about 30%. Apparent molecular weight on sodium dodecyl sulfate-polyacrylamide gel electrophoresis is approximately 40,000, similar

ammonia from the intermediate attached to H-protein and the synthesis of 5,10-CH2-H4folate1 in the presence of H4folate. L-protein is a lipoamide dehydrogenase that catalyzes the reoxidation of the resulting dihydrolipoic acid on Hprotein. In eukaryotes, all of the components are synthesized cytoplasmically as precursor forms (6-9), transported into mitochondria, and assembled in the complex after processing. The primary structures of chicken P-protein and H-protein have been determined. Chicken P-protein is composed of 970 amino acid residues (8) and has pyridoxal phosphate at Lys-704 (10). Chicken H-protein is composed of 125 amino acids with a lipoic acid prosthetic group at Lys-59 (11). Recently we have cloned a full-length cDNA encoding bovine T-protein and a partial cDNA for chicken T-protein (9). The predicted primary sequence of bovine T-protein exhibited no sequence homology with other folate-requiring proteins. There is little information about the binding site for H4folate or the association site for H-protein and other components of the glycine cleavage system on T-protein. Since the structural studies require relatively large quantities of enzymatically active protein and since purification and the elucidation of the primary structure of the components of the system have been accomplished mainly with the chicken system, we decided to clone and overexpress chicken T-protein.
In this paper, we describe the cloning of overlapping cDNAs encoding chicken liver T-protein and the expression of mature chicken T-protein in Escherichia coli. We modified the nucleotide sequence flanking to the initiation codon of the constructed cDNA for mature T-protein without altering the amino acid sequence to reduce the free energy of formation for the folded mRNA. The plasmids constructed were successful in expressing soluble and enzymatically active Tprotein in E. coli when the expression was conducted at a relatively low growth temperature under low inducing conditions. The availability of large amounts of active T-protein will facilitate our structure/function studies on T-protein.
Inc. Radioactive compounds were either from Amersham Corp. or Du PontLNew England Nuclear. Oligonucleotides were synthesized on an Applied Biosystems 381A DNA synthesizer. Chicken liver T-protein was purified by the method described below for recombinant Tprotein and stored at -40 "C. All other reagents were obtained commercially.
Preparation of Antibody-A rabbit was immunized by the subcutaneous injection of the purified chicken liver T-protein in Freund's complete adjuvant. Injections (550 pg of protein each time) were given three times at bi-weekly intervals. The IgG fraction was obtained by ammonium sulfate fractionation followed by chromatography on DEAE-cellulose (12). The antibody was further purified by affinity chromatography on a column of Sepharose 4B coupled with chicken T-protein. The specificity of the antibody was checked by Western blot analysis.
Isolation and Sequence Determination of cDNA Clones-Isolation of a partial cDNA clone for chicken T-protein, CT5C, was as described previously (9). To obtain the cDNA encoding the 5"terminal region of the mRNA, we constructed two chicken liver primerextended XgtlO cDNA libraries with the cDNA cloning system X g t l O thetic oligonucleotide, 5' GGTCAGGTCATCCGGCAGCCCGGC 3', (Amersham Corp.). Initially, a library was constructed with a syncorresponding to the 5'-terminal nucleotide sequence of CT6C as the primer. A 658-bp Sad-BglII fragment of BT5A, a cDNA clone for bovine T-protein (9), was labeled with the Multiprime DNA-labeling system (Amersham Corp.) and [cx-~'P]~CTP and used to screen about lo6 recombinant phages. Hybridization was performed under the conditions as described previously (9) except that the final wash of the blots was carried out in 1 X SSC (1 X SSC = 150 mM NaCl, 15 mM sodium citrate, pH 7.0). DNA from positive recombinant phages was isolated, and the cDNA inserts were subcloned into a plasmid pGEM-3Z (Promega) and sequenced by the dideoxy chain termination method (13). A clone named CTlO (Fig. l) was thus isolated. Another library was then constructed with the mixture of two synthetic oligonucleotides, 5' GGTGAGCAGCGTCAGGGT 3' and 5' CTCCAGCCGGCGAACGGCACCAT 3', derived from the nucleotide sequence of CTlO as the primer. The library was screened as above and clones CT5A and CT6A ( Fig. 1) were isolated. Northern Bot Analysis-Total RNA was extracted from chicken liver by the guanidinium isothiocyanate method (14). Poly(A)+ RNA was isolated using Oligotex-dT30 (Nippon Roche, Tokyo, Japan) according to the manufacturer's instruction. Five pg of poly(A)+ RNA was denatured with formaldehyde, electrophoresed through a 1% agarose gel in the presence of formaldehyde, and transferred to Gene-Screen Plus membrane (15). The blot was hybridized with CT5C labeled with [cx-~'P]~CTP by random priming under the condition described previously (9). The filter was washed twice for 5 min at room temperature in 2 X SSC, once for 30 min at 60 "C in 2 X SSC containing 1% SDS, and once for 30 min at room temperature in 0.1 X SSC containing 0.05% SDS. Autoradiography was performed at -80 "C for 5 h with an intensifying screen.
Construction of E. coli Expression Vector for Mature T-protein-The strategy for construction of the expression plasmid is shown in Fig. 2. Three cDNA clones, CTGA, CT10, and CT5C ( Fig. l), were used to assemble the entire coding sequence for mature T-protein.
The plasmid pGEMICT5C was constructed by cloning CT5C into  pGEM-3Z as described previously (9). pGEM/CT5C was then digested with Sac1 and BglII, and the plasmid containing a 545-bp 3' fragment of CT5C was isolated. The 250-bp Sad-SmaI fragment of CTGA, the 243-bp SnaI-PstI fragment of CT10, and the 119-bp PstI-BglII fragment of CT5C were isolated and ligated in a one-step reaction into the above mentioned SacI-BglII-digested pGEM/CT5C. The resulting plasmid encoding a full-length cDNA for mature chicken T-protein (pGEM/CT) was verified by DNA sequencing and restriction analysis, and then a 1176-bp Sad-EcoRI fragment was isolated from pGEM/CT and subcloned into a phagemid vector, pTZ18U, generating the plasmid pTZ/CT. An NdeI site carrying inframe initiator methionine codon at the 5'-end and a BarnHI site located downstream of the stop codon were introduced to the mature T-protein cDNA in pTZ/CT by oligonucleotide-directed mutagenesis according to the method of Kunkel et al. (16) with a mutagen phagemid in vitro kit from Bio-Rad. Two oligonucleotides containing several modifications (in boldface), 5' CCCCGGGTACATATG TCGGCGCCGGA 3' (the underlined section is the NdeI site) and 5' TGAGACCCGGATCCATTAAAGCCC 3' (the underlined section is the BamHI site), were synthesized and used for mutagenesis. The 1141-bp NdeI-BarnHI fragment was isolated from the resulting recombinant phagemid, pTZ/CTNB, and ligated into the NdeI-and BamHI-digested . This plasmid will be referred to as pET/MCT. The modified cDNAs in which the 5"coding sequence was altered without the amino acid substitution were also constructed cleotides 5' CATATGTCGGCACCAGAAGGACTGAAGCAGACG from pTZ/CTNB by site-directed mutagenesis employing oligonu-3' (for pET/MCT4) or 5' CATATGTCGGCACCAGAAGGACTGA-AGCAAACACCGCTGGACG 3' (for pET/MCT6) and inserted into PET-3a as above. All of the constructs were confirmed by sequencing.
The 5"coding sequences of mRNAs derived from these cDNAs are shown in Fig. 3.
Expression of T-protein in E. coli-Expression plasmids pET-Ba, pET/MCT, pET/MCT4, and pET/MCT6 were transfected into E.
coli strain BL21(DE3)pLysS, which contained the T7 RNA polymerase gene integrated in the chromosome under the control of the lacUV5 promoter and T7 lysozyme expression plasmid, pLysS (17). The transformants were grown at 37 "C in 10 ml of M9H medium, which is composed of the components of both M9 and H medium (1 g of NH4CI, 3 g of KHzP04, 15 g of Na2HPO4. 12Hz0, 4 g of glucose, 1 ml of 1 M MgS04, 10 g of Bacto Tryptone (Difco), and 5 g of NaCl in 1 liter of water) in the presence of ampicillin (20 pg/ml) and chloramphenicol (30 pg/ml). When the culture reached an Am of 0.8, the T7 RNA polymerase was induced by adding IPTG to 1 mM, and the cell growth was continued for 3 h at 37 "C. Preincubation was omitted and the concentration of IPTG was reduced to 25 p~ when cells were cultured at 30 "C for 24 h. Aliquots of 0.25 ml of the culture were withdrawn, and the cells were pelleted by centrifugation, boiled in 50 pl of SDS-PAGE sample buffer (18), and saved (referred to as whole cell lysate). The cells in remaining cultures were harvested and resuspended in ice-cold lysis buffer (50 mM Tris-HC1, pH 7.5, 2 mM EDTA, 1 mM DTT, 10 WMp-amidinophenylmethanesulfonyl fluoride) of 10% of the original culture volume and incubated at 0 "C for 30 min. After freezing at -80 "C, the cells were thawed and sonicated for 30 s twice at 0 "C in a Branson Sonifier 250. The lysate was centrifuged at 16,000 X g for 30 min at 4 "C, and the supernatant was collected (referred to as soluble fraction). The pellet was washed once with lysis buffer and extracted with 8 M urea in lysis buffer of the same volume of the soluble fraction (referred to as insoluble fraction). The three fractions were subjected to SDS-PAGE and Western blot analysis.
Purification of Recombinant Chicken T-protein-All purification steps were carried out at 4 "C if not otherwise specified. A 500-ml culture of BLSl(DE3)pLysS transfected with pET/MCT4 was harvested after 18 h of growth at 30 "C in the presence of 25 p M IPTG. The cells were suspended in 50 ml of ice-cold buffer A (20 mM Tris-HCI, pH 8.3, 2 mM EDTA, 1 mM DTT, 10 p~ p-amidinophenylmethanesulfonyl fluoride, 10% glycerol) and held for 30 min on ice. After freezing at -40 "C overnight, the frozen suspension was thawed and sonicated for 5 X 30 s using a Branson Sonifier 250 with a %inch disrupter horn. The sonicate was centrifuged for 60 min at 105,000 X g, and the supernatant was applied to a column of DEAE-Sepharose CL-GB (2.5 X 10 cm, Pharmacia LKB Biotechnology Inc.) equilibrated with buffer A. The column was washed with 200 ml of buffer A at a flow rate of 31 ml/h, and pass-through fractions containing T-protein activity were collected. The fraction (49 ml) was applied to a column of SP-Sephadex C-50 (2 X 5 cm, Pharmacia LKB Biotechnology Inc.) equilibrated with buffer B (20 mM Tris-HCI, pH 8.0, 2 mM EDTA, 1 mM DTT, 10 ~L M p-amidinophenylmethanesulfonyl fluoride, 10% glycerol). After washing the column with 100 ml of buffer B containing 50 mM NaCl, bound proteins were eluted with a linear gradient of NaCl(50-150 mM in 320 ml) in buffer B. The flow rate was maintained at 22 ml/h. The peak fractions (80-105 mM NaCl) were pooled, dialyzed against buffer C (20 mM potassium phosphate buffer, pH 7.4, 1 mM DTT, 10% glycerol), and loaded onto a column of hydroxyapatite (1.5 X 5 cm) equilibrated with buffer C. After washing with 50 ml of 50 mM potassium phosphate buffer, pH 7.4, containing 1 mM DTT and 10% glycerol, the bound proteins were eluted with a linear gradient of potassium phosphate (50-150 mM, pH 7.4, in 180 ml) containing 1 mM DTT and 10% glycerol. The flow rate was kept at 15 ml/h. The peak fractions (70-85 mM potassium phosphate) were pooled and subjected to gel filtration on a column of Sephadex G-100 (2.5 X 90 cm, Pharmacia). The column was equilibrated and eluted with buffer C at a flow rate of 15 ml/h.
Western Blot Analysis-The whole cell lysate and the soluble and insoluble fractions described above were boiled for 5 min in SDS-PAGE sample buffer, subjected to 10% SDS-PAGE (181, and transferred to Immobilon-P (Millipore) using a semidry electroblot apparatus (Sartoblot 11, Sartorius) according to the protocol given by the manufacturer. The filter was blocked for 1 h at room temperature in blocking solution (1% gelatin in TTBS, which is composed of 50 mM Tris-HC1, pH 7.5,150 mM NaCl, and 0.05% Tween 20), incubated for 3 h with anti-chicken T-protein antibody (2 pg/ml in blocking solution), washed with TTBS, and then incubated for 1 h in blocking solution containing 4.6 kBq/ml '251-protein A. After removal of excess protein A, the filter was subjected to autoradiography and quantitative analysis using a Bioimaging Analyzer BAS2000 (Fuji Photo Film, Tokyo, Japan) with the purified recombinant T-protein as a standard. Peptide Mapping-About 200 pg of purified recombinant T-protein was carboxymethylated and digested with lysylendopeptidase (Wako Pure Chemical Industries, Osaka, Japan) at an enzyme/substrate ratio of 1:150 (mol/mol) in 0.6 ml of 10 mM Tris-HCI, pH 9.0, containing 3 M urea at 30 "C overnight. The resulting peptides were fractionated on an ODs-12OT column (4.6 X 250 mm, Tosoh, Tokyo, Japan) with acetonitrile gradient in 0.1% trifluoroacetic acid at a flow rate of 1.0 ml/min. Elution profile was monitored at 220 nm and compared with that of the authentic chicken liver T-protein processed as above.
Assay of T-protein and Kinetic Analysis-T-protein was routinely assayed in the reverse direction essentially as described previously (5). The reaction mixture (0.5 ml) contained 50 units of P-protein from Arthrobacter globiformis (19), 5 units of chicken H-protein (l), 2.5 pg of diaphorase (Boehringer Mannheim) as L-protein, and Tprotein. Kinetic experiments were performed according to the "COUpled assay" in a previous report (5) (20) with bovine serum albumin as a standard. The concentration of chicken H-protein was determined as described previously (1). Amino acid sequences were analyzed by automated Edman degradation in an Applied Biosystems 470 sequencer.

RESULTS
Isolation and Characterization of cDNA Clones-We have described the isolation of a cDNA clone (CT5C) containing a 675-bp insert encoding the COOH-terminal half of chicken T-protein from a Clontech library (9). Since we failed to isolate clones extending to the 5' end of the mRNA from a second library (9), we constructed two chicken liver primerextended cDNA libraries with an oligonucleotide corresponding to nucleotides 544-567 (Fig. 4) and a mixture of two oligonucleotides corresponding to nucleotides 309-327 and 187-207 (Fig. 4), respectively, as primers. These libraries were screened with a nucleotide fragment of 658 bp containing the 5"terminal sequence of bovine T-protein. Overlapping cDNA clones designated as CT5A, CTGA, and CTlO were isolated and characterized. The sequence at the 5'-end of the chicken T-protein mRNA, however, was only in a single cDNA clone numbered from the presumed start codon, and the amino acid sequence is numbered from the initiation methionine. Amino acid sequences confirmed by Edman analysis of the purified chicken liver T-protein are underlined. Nucleotide sequences used for the synthesis of primers for the construction of cDNA libraries are boxed, and the polyadenylation signal is doubly underlined. (Fig. 1). The combined nucleotide sequence obtained from two of the above clones (CT5A and CT10) and the previously obtained CT5C is 1228-bp-long and consists of a 19-bp 5'untranslated region followed by a 1176-bp open reading frame, thus coding for 392 amino acid residues of M, 42,056 and a 3'-noncoding region of 33 bp (Fig. 4). The sequence surrounding the first ATG matches the consensus sequence for the initiator methionine codon (21). A polyadenylation signal ATTAAA is found 19 nucleotides upstream from the poly(A) tail. Northern blot analysis of chicken liver poly(A)+ RNA probed with CT5C indicated that the mRNA is present as a single species of about 1.5 kilobase pairs (Fig. 5). The size matches that of the composite cDNA. The underlined sequences in Fig. 4 exactly correspond to peptide sequences obtained by direct protein sequencing, confirming that the cDNA encodes chicken T-protein. The NH2-terminal amino acid sequence determined for the intact chicken liver Tprotein is present in positions 17-43 in the deduced amino acid sequence (Fig. 4). The sequence predicted for 1-16 is enriched for arginine and leucine and is consistent with identity as a mitochondrial targeting sequence. The removal of these initial 16 residues yields a mature protein of M, 40,292, which is in fair agreement with the value for the purified chicken liver enzyme estimated by SDS-PAGE (2).

Expression of Recombinant Mature T-protein in E. coli
Cells-Construction of the expression vectors as described under "Experimental Procedures'' is illustrated in Fig. 2. The plasmid pET/MCT, which harbored the native sequence of mature T-protein, was used to transform E. coli strain BLZl(DE3)pLysS to test for expression. Bacteria containing the recombinant T-protein cDNA or a control plasmid were grown at 37 "C in the presence of 1 mM IPTG for 3 h, and a portion of harvested cells was boiled in SDS-PAGE sample buffer. Analysis of the whole cell lysate by SDS-PAGE and Western blotting indicated a very low level of expression (Fig.  6, A and B ) . Longer induction up to 24 h could not increase the level of expression significantly, while the degradation products were not observed during the period. To test the instability of the target plasmid in the host cells, the transformants were titrated on the plates, which have ampicillin, IPTG, both, or neither added to the top agar according to the protocol described by Studier et al. (17). The results were consistent with the observation described by Studier et al. (17) for cells containing stable plasmids (not shown). The low yield of T-protein was, therefore, due to neither the instability of the plasmid pET/MCT nor the rapid degradation of Tprotein in the cell. Several investigators have reported the involvement of mRNA secondary structure of the translational initiation region as the major determinant of the translation efficiency in E. coli (see Ref. 22). Therefore, we attempted to change the nucleotide sequence downstream from the start codon of mature T-protein by site-directed mutagenesis to obtain expression plasmids capable of expressing high levels of T-protein. The translation initiation region of PET/ MCT is enriched for cytosine and guanine nucleotides and is predicted to form a stem-and-loop structure with a free energy of -20.8 kcal/mol according to Zuker and Stiegler (23) (Fig.  3). This predicted hairpin structure was destabilized either by substitutions of G + A at positions +9, +12, +15, and +18 (MCT4) or at positions +27 and +30 in addition to the aforementioned substitutions (MCT6) (Fig. 3). The substitutions were designed to retain the original amino acid sequence of the region. The modified mature T-protein DNA variants have calculated free energies of -10.8 (MCT4) and -14.  c Activity of bacterial T-protein was subtracted. processed for SDS-PAGE and Western blot as described under "Experimental Procedures." Panel A, total cellular proteins and soluble and insoluble fractions of E. coli induced by 1 mM IPTG for 3 h a t 37 "C were analyzed by 10% SDS-PAGE. Samples from 25 pl of each cell culture and 1 pg of purified chicken liver Tprotein were loaded. S and I denote soluble and insoluble fraction, respectively. Panel 23, the same samples as above but from 0.2 pl of cell culture and 5 ng of purified recombinant T-protein were separated by SDS-PAGE, electrotransferred onto Immobilon-P, and immunostained with anti-T-protein antibody. Panel C, samples from 0.2 p1 of each cell culture induced by 25 p~ IPTG for 24 h a t 30 "C and 5 ng of purified recombinant T-protein were treated as above.
kcal/mol (MCT6). SDA-PAGE and Western blot analysis indicated that the cells transfected with plasmids containing these modified DNAs expressed approximately 40-fold of Tprotein compared with the native codon plasmid (Fig. 6, A  and 23). The recombinant T-protein in E. coli was expressed in a highly insoluble form, and less than 30% was recovered in the soluble fraction. The insoluble T-protein could be solubilized with 8 M urea, but we failed to find the proper conditions for restoring enzymatic activity. We found that the reduction of both growth temperature and concentration of IPTG with prolonged incubation facilitates the recovery of soluble T-protein. When induction was carried out in the presence of 25 p~ IPTG at 30 "C for 24 h, about 80% of the expressed T-protein remained soluble as shown in Fig. 6C.
Purification and Properties of the Recombinant T-protein-E. coli BLSl(DE3)pLysS cells transformed with the plasmid pET/MCT4 were grown at 30 "C with 25 p~ IPTG. Recombinant T-protein was purified as described under "Experimental Procedures," and the result of a typical purification was summarized in Table I yielded approximately 4 mg of pure T-protein from 500 ml of the culture. Fig. 7 shows the Coomassie Brilliant Blue staining of the polypeptide patterns in each purification step. The purified protein (lane 6 ) had a similar electrophoretic mobility to that of chicken liver T-protein (lane 7) and was recognized by the specific antibody against chicken liver T-protein in immunoblot (not shown). The molecular weight of recombinant T-protein was estimated as 40,000 by SDS-PAGE. The value is in good agreement with that of native chicken Tprotein (2). E. coli BL21(DE3)pLysS contains its own glycine cleavage system. The bacterial T-protein was retained on a column of DEAE-Sepharose CL-GB at pH 8.3 and eluted with 0.2 M NaCl, while recombinant chicken T-protein was recovered in the pass-through fraction. The bacterial T-protein represented about 3% of the total activity of T-protein found in cells harboring pET/MCT4. The NH2-terminal sequence of recombinant T-protein was determined as Ser-Ala-Pro-Glu-Leu-Lys-Gln-Thr-Pro-, indicating that removal of the initiator methionine by E. coli aminopeptidase. The peptide mapping pattern of the lysylendopeptidase-digested recombinant T-protein on HPLC was comparable with that of native T-protein (Fig. 8).
To test whether the expressed T-protein was fully active, the kinetic properties of recombinant T-protein were compared with those of native chicken T-protein.
The T-protein-  (Table 11), yielding kyf values of 3.9 and 3.3 s-l, respectively. Previously, we reported a kcat value for the purified chicken T-protein ( Table 1 in Ref. 5), but the value was mistyped inadvertently during the copying of the data. The correct value is 5.8 s-'. Overall, the results of these analyses demonstrated that the cloned cDNA directs the synthesis of T-protein that is structurally and functionally equivalent to chicken T-protein.

DISCUSSION
We have isolated several partial cDNAs from a commercial chicken liver cDNA library and two primer-extended chicken liver X g t l O cDNA libraries. The composite cDNA sequence is 1228-bp-long and contains an open reading frame that encodes 392 amino acids of the precursor form of chicken Tprotein. The deduced amino acid sequence of mature chicken T-protein is compared with that of bovine T-protein (Fig. 9). The overall homology is 67%. Chicken T-protein is a basic protein with a PI value of 9.8 (2) and contains more arginine residues than bovine T-protein in the NHn-terminal half. Consecutive arrays of basic amino acids are found in the COOH-terminal half of both T-proteins. The regions rich in positive charges may contribute to the binding of HIfolate and/or acidic H-protein.
We used the T7 expression system described by Studier et al. (17) to express mature T-protein in E. coli. The PET-3a vector carries a strong T7 promoter (410 promoter) and the first 11 codons for the gene 10 protein. We joined the coding sequence for mature T-protein directly to the gene 10 initiation codon as an NdeI site that includes the initiation ATG. The resulting plasmid was expected to produce mature Tprotein under the direction of the upstream translation signals of gene 10. The result obtained with the plasmid thus constructed (pET/MCT) was unsatisfactory (Fig. 6A). Gross et al. (22) reported that RNA secondary structure in the translational initiation region controls the expression of interferonfi in E. coli using the temperature-sensitive X bacteriophage pL-pPR promoter. The rate of expression was enhanced when a stem-loop structure located downstream of the initiation codon within the initial binding site of the 30 S ribosomal subunit was modified to a form with reduced free energy. The region just downstream of the initiation codon of the nucleotide sequence for mature T-protein is rich in guanine and cytosine nucleotides. The region is, therefore, likely to form a stable secondary structure (Fig. 3). We substituted several of the guanine nucleotides with adenine nucleotides to reduce the free energy, as estimated according to Zuker and Stiegler (23), with the aid of a computer program. The modified Tprotein gene variants MCT4 and MCTG produced mature Tprotein efficiently in E. coli (Fig. 6). Although the correlation between the stability predicted by the algorithm of Zuker and Stiegler (23) and the rate of expression has not been examined extensively, the introduction of downstream secondary structure with the relatively low free energy should be considered in the case where the production of the desired protein is disappointingly small and there are no clear reasons (such as the instability of the expressed protein) found for the low expression.
Another problem with the expression of foreign proteins in E. coli is the production of aggregated inactive protein that is segregated in inclusion bodies (24). This was the case in the present study, where about 70% of the expressed T-protein at 37 "C and 1 mM IPTG was insoluble and inactive even after it was solubilized with 8 M urea (Fig. 6B). To optimize the production of the soluble T-protein, lower growth temperatures at reduced IPTG concentrations were tested as discussed for the expression of rabbit muscle phosphorylase (25). Higher yields of the soluble, enzymatically active Tprotein were obtained when the plasmid pET/MCT4 or PET/ MCTG was expressed at 30 "C with 25 PM IPTG. Under this condition, activity of T-protein continued to increase for some time after the immunologically detectable T-protein reached a constant level (not shown). For this reason, longer growth time is necessary for maximum accumulation of active Tprotein. Proper folding of the mature form of T-protein, like other mitochondrial proteins (26,27), may be facilitated by chaperonin. Formation of inactive aggregates of T-protein under the full inducing condition may be the result of the limitation of E. coli chaperonin to cope with a large amount of T-protein produced rapidly. It is also possible that chicken T-protein cannot associate with E. coli chaperonin, and the folding to the native structure without chaperonin is a slow process that takes place at a relatively low temperature and low concentration of T-protein.
The physical and kinetic properties of the pure recombinant T-protein were determined in this study. Recombinant Tprotein has the same molecular weight and the same NH,terminal amino acid sequence as native chicken T-protein.
The absence of a methionine residue at the NH2 terminus is in full agreement with the predictive rule for the methionine removal from proteins in E, coli described by Hire1 et al. (28).