Cloning, Expression, and Characterization of a NovelEscherichia coli Thioredoxin*

Thioredoxin (Trx) is a small ubiquitous protein that displays different functions mainly via redox-mediated processes. We here report the cloning of a gene (trxC) coding for a novel thioredoxin in Escherichia coli as well as the expression and characterization of its product. The gene encodes a protein of 139 amino acids (Trx2) with a calculated molecular mass of 15.5 kDa. Trx2 contains two distinct domains: an N-terminal domain of 32 amino acids including two CXXC motifs and a C-terminal domain, with the conserved active site, Trp-Cys-Gly-Pro-Cys, showing high homology to the prokaryotic thioredoxins. Trx2 together with thioredoxin reductase and NADPH is an efficient electron donor for the essential enzyme ribonucleotide reductase and is also able to reduce the interchain disulfide bridges of insulin. The apparent K m value of Trx2 for thioredoxin reductase is similar to that of the previously characterized E. coli thioredoxin (Trx1). The enzymatic activity of Trx2 as a protein-disulfide reductase is increased by preincubation with dithiothreitol, suggesting that oxidation of cysteine residues other than the ones in the active site might regulate its activity. A truncated form of the protein, lacking the N-terminal domain, is insensitive to the presence of dithiothreitol, further confirming the involvement of the additional cysteine residues in modulating Trx2 activity. In addition, the presence of the N-terminal domain appears to confer heat sensitivity to Trx2, unlike Trx1. Finally, Trx2 is present normally in growing E. coli cells as shown by Western blot analysis.


REACTION 1
Trx-(SH) 2 ϩ Protein-S 2 7 Trx-S 2 ϩ Protein-(SH) 2 REACTION 2 Thioredoxin is present in all living organisms and has been isolated and characterized from a wide variety of prokaryotic and eukaryotic cells (1). In Escherichia coli, thioredoxin was first identified as an electron donor for ribonucleotide reductase (RNR), the enzyme that reduces ribonucleotides to deoxyribonucleotides for DNA synthesis and repair (2). E. coli thioredoxin can also function as a hydrogen donor for 3Ј-phosphoadenosine 5Ј-phosphosulfate reductase in the sulfate assimilation pathway as well as methionine sulfoxide reductase (3,4). Apart from these functions, E. coli thioredoxin is necessary for the life cycle of some bacteriophages such as T7, M13, and f1 (5)(6)(7). In eukaryotic cells, thioredoxin can also function as a hydrogen donor for RNR, 3Ј-phosphoadenosine 5Ј-phosphosulfate, and methionine sulfoxide reductases, similar to the prokaryotic thioredoxin (1). In addition, thioredoxin can (a) facilitate refolding of disulfide-containing proteins (8); (b) activate the interleukin-2 receptor (9); (c) modulate the DNA binding activity of some transcription factors, e.g. NF-B (10); and (d) stimulate proliferation of lymphoid cells and a variety of human solid tumors (11,12). Furthermore, thioredoxin is an efficient antioxidant able to reduce hydrogen peroxide (13), scavenge free radicals (14), and protect cells against oxidative stress (15). In photosynthetic organisms, three types of thioredoxins have been identified, two forms in chloroplasts (f and m) involved in regulatory systems in oxygen photosynthesis and one form in cytosol and endoplasmic reticulum (h) (16).
E. coli Trx is a well studied enzyme, and its three-dimensional structure has been determined by NMR for both oxidized and reduced forms, as well as by x-ray crystallography (17,18). E. coli thioredoxin contains 109 amino acids and has a central core of five strands of twisted ␤-pleated sheet flanked by four ␣-helices and the active site located in a protrusion of the protein (19).
Thioredoxin-negative mutants (trxA Ϫ ) of E. coli are viable (5), and analysis of these mutants led to the identification of a novel cofactor, glutaredoxin-1 (Grx1), as an efficient substitute of thioredoxin for RNR and 3Ј-phosphoadenosine 5Ј-phosphosulfate reductase enzymatic activity (20,21). However, Grx1 could neither substitute for thioredoxin in methionine sulfoxide reduction nor in bacteriophage growth or assembly (5)(6)(7)22), which thus remained typical phenotypes of E. coli thioredoxin mutants. The isolation of an E. coli double mutant in thiore-doxin/glutaredoxin-1 allowed the identification of two novel glutaredoxins, Grx2 and Grx3, but only Grx3 is able to serve as a hydrogen donor for RNR (23,24). Thus, it seemed that only one thioredoxin did exist in E. coli. However, the existence of another thioredoxin has been suggested to be necessary for the maintenance of the reducing environment in E. coli cytoplasm (25). In addition, a triple Trx, Grx1, and Grx3 mutant was viable (26), indicating the presence of an alternative protein capable of reducing RNR in vivo.
We report here the cloning of a DNA sequence coding for a novel E. coli thioredoxin (Trx2) based upon biological activity data and protein homology. We also present evidence that the protein is normally expressed in E. coli cells and that the N-terminal sequence of the protein contains a novel domain with four cysteine residues that partly regulates its enzymatic activity as protein-disulfide reductase.

Strains and Media
E. coli K-12 was a stock from our laboratory. K38 (wild type) and A179 (trxA::kan) (27) strains were a kind gift from Prof. Arne Holmgren (Karolinska Institutet, Stockholm). Cells were grown in LB medium supplemented (when necessary) with 50 g/ml ampicillin or kanamycin.

Cloning of the E. coli Thioredoxin 2
A thioredoxin-like sequence (Trx2) containing an open reading frame coding for a protein of 139 amino acids (GenBank TM accession number 1788936), 2 was used to design the specific mutagenic primers EcTrx2-NdeI, 5Ј-TCCCGAGGTTACATATGAATACCGTTTG-3Ј (forward), and EcTrx2-BamHI, 5Ј-CAAGATGGGATCCGGTAAGATTAAAGAGATT-C-3Ј (reverse), that introduce an NdeI site and a BamHI site at the N terminus and C terminus of the coding sequence, respectively. These primers were used to amplify E. coli K-12 genomic DNA by polymerase chain reaction (PCR) with the Expand TM Long Template PCR System (Boehringer Mannheim) (30 cycles at 94°C for 20 s, 58°C for 30 s, and 68°C for 2 min, linked to a 68°C for 30 min cycle). The PCR product was cloned into the pGEM-T Easy Vector System I (Promega) and sequenced.

Protein Expression and Purification
The insert of Trx2 cloned into the pGEM-T vector was digested with NdeI and BamHI and cloned into the NdeI/BamHI sites of the pET-15b expression vector (AMS Biotechnology). The recombinant plasmid was designated pET-Trx2. E. coli BL21 (DE3) was transformed with the pET-Trx2 construct, and a single positive colony was inoculated in 1 liter of LB medium plus ampicillin and grown at 37°C until A 600 ϭ 0.5. Then fusion protein was induced by the addition of 0.5 mM isopropyl-1-thio-␤-D-galactopyranoside, and growth was continued for another 3.5 h. The cells were harvested by centrifugation at 10,000 ϫ g for 10 min, and the pellet was resuspended in 50 ml of 20 mM Tris-HCl, pH 8.0, 100 mM NaCl, and 1 mM phenylmethylsulfonyl fluoride. Lysozyme was added to a final concentration of 0.5 mg/ml with stirring for 30 min on ice. Subsequently, MgCl 2 (10 mM), MnCl 2 (1 mM), DNase I (10 g/ml), and RNase (10 g/ml) were added, and the incubation was continued for another 45 min on ice. The cells were sonicated, and the supernatant was cleared by centrifugation at 15,000 ϫ g for 30 min and loaded onto a Talon Resin Column (CLONTECH). The His-Trx2 protein was eluted with 50 mM imidazole, dialyzed against 20 mM HEPES, pH 8.0, and concentrated with Centricon concentrators (Amicon Inc.), and the size and purity of His-Trx2 was determined by SDS-polyacrylamide gel electrophoresis. When indicated, the His tag was removed by thrombin incubation. Protein concentrations were determined from the absorbance at 280 nm using a molar extinction coefficient of 14,180 M Ϫ1 cm Ϫ1 for E. coli Trx1 and 17,790 M Ϫ1 cm Ϫ1 for E. coli Trx2.
The truncated form of Trx2 (⌬Trx2) lacking the first 32 amino acids at the N-terminal part of the protein was amplified using the primer Ec⌬Trx2-NdeI 5Ј-GCGGTCACGACCATATGGACGGAGAGGTG-3Ј as forward primer and EcTrx2-BamHI as a reverse primer. The cloning, overexpression, and purification of this truncated form was identical to that described above for the full-length one. Protein concentration was determined from the absorbance at 280 nm using a molar extinction coefficient of 13,310 M Ϫ1 cm Ϫ1 for ⌬Trx2.

Antibodies and Immunoblotting Analysis
For antibody production, we used a modification of the method described by Song et al. (28). Female chickens were injected (once every 15 days during 6 weeks) subcutaneously at multiple sites with 100 g (in 0.25 ml) of His-Trx2 in complete adjuvant. After the second injection, eggs were collected daily, and when a suitable number of eggs were obtained antibodies were purified as follows. Egg yolks were separated from whites by dropping the egg contents on a funnel placed in a graduated cylinder. An equal volume of buffer S (10 mM phosphate, pH 7.5, 0.1 M NaCl, containing 0.01% sodium azide) was added to the yolks and stirred. Next, 10.5% PEG 8000 (Sigma) in buffer S was added to a final concentration of 3.5%. The mixture was stirred for 30 min at room temperature and centrifuged at 12,000 ϫ g for 20 min. The supernatant was filtered through two layers of 3MM Whatman chromatography paper, and 42% PEG 8000 in buffer S was added to a final concentration of 12%. The mixture was stirred thoroughly and centrifuged at 12,000 ϫ g for 20 min. The supernatant was discharged, and the pellet was redissolved in 12% PEG in buffer S to the original yolk volume. After centrifugation, the pellet was dissolved in 30 ml of buffer S and dialyzed overnight against buffer S without NaCl. Affinity-purified antibodies were prepared using a cyanogen bromide-activated Sepharose 4B column where 1 mg of Trx2, in which the His tag had been removed by thrombin, had been coupled following the procedure recommended by the manufacturer (Pharmacia Biotech Inc.). The specificity of the antibodies was tested by Western blotting using recombinant Trx2 and total cell extracts.

Preparation of Cellular Fractions from E. coli
Lysozyme Treatment-Cells were harvested at A 600 ϭ 0.5 and washed three times in 50 mM Tris, pH 8.0. The pellet was resuspended in the same buffer plus 2 mM MgCl 2 and 2 mg/ml lysozyme at an A 600 ϭ 40, incubated for 30 min on ice, and centrifuged at 14,000 rpm at 4°C for 1 min. The supernatant after centrifugation was collected as the periplasmic lysate fraction. The pellet was resuspended in the same volume of 50 mM Tris, pH 8.0, 2 mM MgCl 2 and kept as the cytosolic fraction.
Freeze-Thaw Treatment-Cells were harvested and washed as in the previous treatment. The pellet was resuspended in 50 mM Tris, pH 8.0, 2 mM MgCl 2 at the same A 600 and quickly frozen in a dry ice/ethanol bath. The frozen suspension was slowly thawed in an ice bucket, and this cycle was repeated three times. The suspension was centrifuged, and periplasmic and cytosolic fractions were collected as described above. For crude bacteria extracts, exponentially growing cells were harvested by centrifugation, and the pellet was resuspended in 50 mM Tris-HCl, pH 8.0, 1 mM EDTA. Cells were sonicated, and the supernatant was cleared by centrifugation for 15,000 ϫ g at 4°C for 30 min. Heated extracts were prepared by heating crude extracts for 5 min, placing them on ice and spinning as described above.

Enzyme Assays of E. coli Thioredoxin 2
The activity of E. coli Trx2 was determined by the DTNB and insulin assays. The DTNB assay was performed essentially as described elsewhere (29). Briefly, the reaction mix contained 200 mM phosphate buffer, pH 7.0, 2 mM EDTA, 0.1 mg/ml BSA, 1 mM DTNB, and 0.5 mM NADPH. The reactions containing 0.5-8 M Trx1, Trx2, or ⌬Trx2 were started by the addition of 10 nM E. coli thioredoxin reductase (IMCO, Sweden). The reaction was followed at 412 nm against a blank containing thioredoxin reductase in a SpectraMax TM 250 Microplate Spectrophotometer (Molecular Devices Corp.) for 7 min at 25°C in a final volume of 100 l. Insulin was used to determine the protein-disulfide reductase activity of thioredoxin as described previously (29). The rate of DTNB reduction was calculated from the increase in A412 using a molar extinction coefficient of 27,200 M Ϫ1 cm Ϫ1 , since reduction of DTNB by 1 mol of Trx-(SH) 2 yields 2 mol of 3-carboxy-4-nitrobenzenethiol each with a molar extinction coefficient of 13,600 M Ϫ1 cm Ϫ1 (29). 2 The sequence is available on the World Wide Web at the server for the E. coli data base collection (http://susi.bio.uni-giessen.de/ecdc.html).
Values of ⌬A412 were multiplied by a factor of 4.3 to give the ⌬A412 of a cuvette with a path length of 1 cm. Reduction by DTT was carried out by preincubation of aliquots of Trx2 at 37°C for 20 min with 2 l of 50 mM HEPES, pH 7.6, 100 g/ml BSA, and 2 mM DTT.

Ribonucleotide Reductase Assay
The ability of E. coli Trx1 and Trx2 to serve as hydrogen donor for NrdAB and EF reductases was determined using the standard ribonucleotide reductase assay (30). Stoichiometric amounts (1 M final concentration) of R1A/R2B and R1E (1.8 g)/R2F (0.6 g) were incubated for 20 min at 37°C with Trx1 (IMCO) or Trx2 in a final volume of 50 l containing 0.5 mM [ 3 H]CDP (19,100 cpm/nmol), 10 mM MgCl 2 , 50 mM Tris-HCl, pH 8.0, 1 mM NADPH, 0.1 M E. coli thioredoxin reductase, and 0.3 mM dATP (for NrdEF) or 1.5 mM ATP (for NrdAB). The reaction was stopped with 0.5 ml of HClO 4 , and the [ 3 H]dCDP formed was determined by liquid scintillation after ion exchange chromatography on Dowex 50 columns. One enzyme unit is the activity that produces 1 nmol of dCDP during 1 min under these conditions.

RESULTS
Cloning of E. coli Trx2-The identification of a novel thioredoxin sequence in the E. coli genome appeared serendipitously when using the recently described mammalian Trx2 (31) cDNA as probe in a Northern blot analysis with E. coli genomic DNA as a negative control. The strong signal obtained (data not shown) prompted us to search for the tentative E. coli homologue of the mammalian Trx2. We identified a DNA sequence in the E. coli genome (not completed at the time of this search) that displayed high homology with mammalian Trx2, but when translated to protein in the three possible reading frames only short peptide sequences were obtained. These short peptide sequences could be due to a possible sequencing mistake (an additional deoxycytidine) that, when removed, resulted in an open reading frame of 417 base pairs encoding a protein of 139 amino acids with the typical thioredoxin active site WCGPC. To further confirm the existence of this sequence coding for a new thioredoxin in the E. coli genome, we used two primers, Ec-Trx2-NdeI and EcTrx2-BamHI (see "Materials and Methods") to amplify a fragment of 420 base pairs from genomic E. coli K-12 DNA. This PCR fragment was further cloned into pGEM-T Easy vector, and sequencing confirmed the absence of the additional deoxycytidine.
Sequence Analysis and Comparison-The sequence of the Trx2 is positioned at min 58.5 in the E. coli chromosome map downstream of the uracyl DNA glycosylase gene. The nucleo-tide sequence of the E. coli Trx2 gene and flanking regions is given in Fig. 1 together with the deduced amino acid sequence. The open reading frame extends from the methionine codon, ATG, at coordinate ϩ1 to the termination codon, TAA, at coordinate 417. This region codes for a protein of 139 amino acids with a estimated molecular mass of 15.5 kDa, containing the classical active site of thioredoxins WCGPC. The methionine ϩ1 codon is preceded by a potential ribosome binding site at position Ϫ9 including six out of nine bases of the consensus ribosome binding sequence (32) and an upstream in frame TAA stop codon. There is a putative Ϫ10 region centered at position Ϫ73 that conforms quite well to the consensus Pribnow box of E. coli. If we assume a standard 16 -18-base pair spacing region, a putative Ϫ35 region, TTGTCT, can be defined (Fig. 1). The DNA sequence following the coding part of the Trx2 gene contains a putative transcriptional Rho-independent terminator detected as an inverted repeat of 10 nucleotides followed by a stretch of T bases (33). This is followed by another open reading frame of 723 nucleotides coding for a putative 27-kDa protein with no clear homology with any other protein in the data base.
The main difference in Trx2 protein sequence with respect to the well known Trx1 is the presence of an extra stretch of 32 amino acids at the N terminus. The C-terminal half of the protein contains the active site found in all thioredoxins (Fig.  2). Recently, Lim et al. (34) have described a third thioredoxin in Corynebacterium nephridii, which displays a similar structure to E. coli Trx2 with an extra N-terminal domain of 32 amino acids and a C-terminal domain homologous to the rest of the prokaryotic thioredoxins (Fig. 2). E. coli Trx2 displays a 29% identity with E. coli Trx1 and 37% with C. nephridii Trx3 and is less similar to the mammalian Trx1 and Trx2. A phylogenetic analysis of several thioredoxins places E. coli Trx2 and C. nephridii Trx3 in the same branch of the tree between E. coli Trx1 and mammalian Trx2 (data not shown).
Enzymatic Activities of E. coli Thioredoxin 2-To establish that the putative Trx2 does in fact encode a protein with thioredoxin activity, we cloned the Trx2 gene into the pET-15b expression vector under the control of a T7 promoter. The resulting plasmid pET-Trx2 was transformed in E. coli BL21 (DE3), and the expression of His-Trx2 was induced by the addition of isopropyl-1-thio-␤-D-galactopyranoside for 3.5 h. The recombinant protein was expressed to levels of approximately 20% of the total soluble protein and was purified almost to homogeneity by affinity chromatography with a Talon column (Fig. 3, inset). Recombinant Trx2 (with or without the His tag) was used to examine the reduction of insulin, a classical assay in which thioredoxin catalyzes disulfide reduction of insulin in a coupled reaction with NADPH in the presence of E. coli thioredoxin reductase.
As shown in Fig. 3, oxidized Trx2 was active as a disulfide reductase; however, it exhibited approximately 5-fold lower activity than Trx1 at concentrations between 0.2 and 0.5 M and 1.5-2-fold lower activity between 0.5 and 2 M. At these higher thioredoxin concentrations, the efficiency of thioredoxin reductase is the rate-limiting step, since the reactivity of Trx-(SH 2 ) and insulin is known to be very fast (35,36).
To understand the role of the extra cysteines present at the N terminus of Trx2, we preincubated Trx2 with DTT. As shown in Fig. 4, the ability of Trx2 to reduce insulin was increased to levels similar to that of Trx1 (compare with Fig. 3). The fact that Trx2 activity was enhanced after reduction of the protein suggested that cysteine residues in Trx2 other than those in the active site could be involved in regulating its enzymatic activity. We used the truncated form of Trx2 (⌬Trx2) lacking the N-terminal portion, including the four cysteine residues, to further test this possibility. Fig. 4 shows that ⌬Trx2 has thioredoxin activity independent of DTT reduction, similar to Trx1. Furthermore, the activity is similar to the reduced Trx2 and higher than the one displayed by the full-length oxidized protein, indicating that the N-terminal portion of the protein partly regulates the activity of Trx2.
DTNB is an artificial disulfide substrate and a fast oxidant of Trx-(SH) 2 , which keeps the concentration of Trx-S 2 constant. DTNB is often used in the assay of thioredoxin reductase, where one molecule of DTNB is split into two 5Ј-thionitrobenzoic acid molecules by reduced Trx1 (1). The DTNB assay and low concentrations (10 nM) of thioredoxin reductase to obtain saturation Michaelis-Menten kinetics were used to calculate the K m of Trx2 for E. coli thioredoxin reductase at pH 7.0 and 25°C. The values obtained were used to calculate the k cat as well as the k cat /K m or apparent second order rate constant for the reaction between thioredoxin reductase and Trx-S 2 . The K m value for Trx2 and the k cat /K m were similar to Trx1. The ⌬Trx2 has a lower K m value, and the k cat is approximately halved. Table I shows the comparison of Trx1 and Trx2 kinetic parameters.
Trx2 as an Electron Donor for Ribonucleotide Reductase-E. coli contains genetic information for three different ribonucleotide reductases. The NrdAB is active during aerobiosis. NrdDG is active during anaerobiosis and uses formate as an electron donor. NrdEF is a cryptic enzyme that uses Grx1 but not Trx1 as an electron donor (37)(38)(39). The viability of a triple mutant Trx1, Grx1 and Grx3, the three known electron donors for the NrdAB protein when growing in aerobiosis, prompted us to assay Trx2 as a tentative reductant for this enzyme. As shown in Fig. 5, Trx2 was found to be a functional electron donor for the NrdAB enzyme. However, compared with Trx1, Trx2 was almost 2.5-fold less efficient as an electron donor than Trx1. Similar to Trx1, Trx2 was not an electron donor for NrdEF (40).
Expression of E. coli Thioredoxin 2-To assay for the presence of Trx2 activity in E. coli, we used protein extracts from K38 (wild type) and A179 (trxA::kan) strains (27). To decrease the background in the insulin assay due to interaction of DTNB with -SH groups, we used heated extracts (85°C, 5 min). Thioredoxin activity was detected in E. coli A179 (trxA Ϫ ), although about 15-fold lower than the one displayed by the wild type (data not shown).
Affinity-purified polyclonal antibodies were used to further analyze the presence of Trx2 in E. coli extracts. Fig. 6A shows that the antibodies reacted with one band at the expected position of Trx2 in a total crude extract of E. coli (lane 3), indicating that the protein is normally expressed. These antibodies also reacted with E. coli Trx1, being detectable in the wild type strain but not in the mutant (Fig. 6A, compare  with lane 4). However, antibodies raised against E. coli Trx1 did not cross-react with recombinant Trx2 (data not shown). The affinity of the anti-Trx2 antibodies for Trx1 was 10-fold lower than for Trx2 (Fig. 6B). From the immunoblot and densitometric analysis, the contents of Trx2 and Trx1 in the total crude extract of E. coli were estimated to be about 0.33 and 1.7 g/mg protein, respectively. The calculated value for Trx1 is in good agreement with the amount of Trx1 (1.91 g/mg protein) obtained by enzyme-linked immunosorbent assay (41).
Thus, the levels of Trx2 versus Trx1 were different when calculated from the Western blots (5-fold lower) than from the activity assays (15-fold lower). One possible explanation is the inactivation of Trx2 during the preparation of the heated bacteria extracts. Trx1 is a heat-stable protein, and heated extracts at 85°C have been used previously to measure thioredoxin activity. We compared the heat stability of Trx2 versus Trx1 (Fig. 7). As expected, the activity of Trx1 was not affected after 5 min of treatment at increasing temperatures from 37 to 85°C. In contrast, the activity of Trx2 was more dependent on temperature treatment, and after 5 min at 85°C 40% of the insulin-reducing activity was lost. When the N-terminal part of Trx2 was deleted, the heat stability of Trx2 was similar to Trx1, indicating that the N-terminal part is responsible for the partial heat inactivation. Treatment with DTT after the heat treatment seems to restore protein activity.
Subcellular Localization of E. coli Thioredoxin 2-The Nterminal part of Trx2 does not match any consensus sequence for protein translocation to the periplasmic space. To determine whether Trx2 is a cytosolic or periplasmic protein, we used two different treatments, lysozyme and freeze-thaw. The lysozyme treatment, a test for defining strict periplasmic localization, disrupts the cell's outer envelope, thus exposing the periplasmic space to the external environment. Freeze-thawing selectively releases cytosolic proteins attached to the inner surface of the cytoplasmic membrane into the periplasmic fraction (42). Cytosolic and periplasmic fractions obtained with freeze-thaw treatment (lanes 5 and 6, respectively) and cytosolic and periplasmic fractions obtained with lysozyme treatment (lanes 7 and 8, respectively) were subjected to immunoblotting analysis (Fig. 6A). Trx2 was identified in the same fraction as Trx1 in both treatments (lanes 6 and 7), suggesting not only its cytoplasmic localization but also its mainly peripheral association with the inner surface of the cytoplasmic membrane similarly to Trx1 (42). DISCUSSION The identification of a novel thioredoxin sequence in the E. coli genome addresses the question of whether the protein coded by this sequence is active as thioredoxin and if it is normally expressed. We have tried to answer these questions and have demonstrated that this newly identified sequence codes for a protein with thioredoxin activity in the insulin, DTNB, and RNR assay. Also, this protein is expressed normally in E. coli cells as shown by Western blot analysis and activity assays.
The existence of a second thioredoxin-like protein in E. coli has been proposed by Beckwith and co-workers (25,26). Thioredoxin reductase, the flavoenzyme that reduces thioredoxin via NADPH, is implicated in the maintenance of the reducing environment of the E. coli cytoplasm, since all mutants selected to allow disulfide bond formation in the cytoplasm mapped in the trxB gene (25), which codes for thioredoxin reductase. Nevertheless, E. coli thioredoxin reductase is a highly specific enzyme only capable of reducing thioredoxin and not any other disulfide bonds in cytoplasmic proteins. However, thioredoxin was not required for the maintenance of the reducing environment, since the single mutant trxA Ϫ showed no difference when compared with the wild type (25). These results indicated that  E. coli cytoplasm might contain another thioredoxin-like protein, reducible by thioredoxin reductase and responsible for the maintenance of the reducing environment. The novel E. coli Trx2 described here is an optimal candidate that could fulfill this function. Additional evidence suggesting the existence of another thioredoxin comes from a recent report that describes the viability of a triple mutant of all of the known electron donors for the essential enzyme RNR (26). Thus, a yet undis-covered disulfide-reducing protein able to reduce RNR has been proposed to exist in E. coli. Again, this novel thioredoxin could also fulfil this function. In fact, our results showed that Trx2 is an efficient electron donor for the NrdAB but not for the NrdEF protein.
Reduction of ribonucleotides to their corresponding deoxyribonucleotides is an essential step in all living organisms in which the building blocks for DNA synthesis are supplied. RNR, the enzyme responsible for this reaction is thus tightly regulated, and the electron supply for its catalytic activity must be guaranteed. If we consider the recently described NrdH protein (see below) as an efficient electron donor for RNR (43), E. coli Trx2 is the fifth electron donor identified for RNR. Whether all of these proteins can act as electron donors for RNR under physiological conditions or whether some of them support this function only when others are absent is a matter for further study based on the characterization of the respective mutants.
Several attempts in the past to identify any residual thioredoxin activity by biochemical methods in E. coli mutants lacking Trx1 have failed (20,26,41,44). By assaying heated extracts (85°C, 5 min) of K38 (wild type) and A179 (trxA Ϫ ) for activity with the insulin assay, we were able to identify a thioredoxin activity in the A179 extract that was approximately 15-fold lower than the one displayed by the wild type. Recently, Jordan et al. (43,45) reported on the existence of a glutaredoxin-like protein with thioredoxin-like activity in E. coli (NrdH). NrdH can be reduced by thioredoxin reductase and is almost as effective as Trx1 in reducing disulfide bonds in insulin. In addition, NrdH is an efficient electron donor for NdrAB and NdrEF enzymes with higher specificity for the latter. However, nrdH is located in the same poorly transcribed operon as nrdEF. Thus, the expression of NrdH is likely to be low, as in the case for NrdEF coded by a cryptic gene, and the protein could not be detected in E. coli. The fact that the levels of the thioredoxin activity in the trxA Ϫ mutant are in good agreement with the levels from the Western blot analysis also argues that this activity is due to Trx2.
The main difference between this novel Trx2 and the rest of the prokaryotic thioredoxins is the existence of 32 extra amino acids residues in its N terminus. This region of the protein does not match with any consensus sequence for translocation in E. coli, and the cytoplasmic localization of Trx2 was confirmed by Western blot analysis of periplasmic and cytosolic fractions. The co-localization of both thioredoxins suggests the attachment of Trx2 to the inner surface of the cytoplasmic membrane as it has been described previously for Trx1 (42). The levels of Trx2 were similar in the wild type and the trxA Ϫ mutant strain, indicating that the disruption of Trx1 does not significantly stimulate expression of Trx2. Also, the gene for Trx2 is positioned at 58.5 min, far away from the trxA gene, which is located at 84 min in the E. coli chromosome.
Interestingly, the N-terminal domain of E. coli Trx2 is homologous to the N terminus part of C. nephridii Trx3 (34). C. nephridii Trx3 could complement some thioredoxin-deficient E. coli mutant phenotypes (namely growth in minimal medium with methionine sulfoxide as the only methionine source and support of growth of bacteriophages T7 and M13 but not f1 growth), when cloned in a high expression vector. Thioredoxin-1-deficient mutants of E. coli have several phenotypes that should be expected to be complemented by the novel Trx2. The maintenance of these phenotypes in the trxA mutant indicates that either the function of Trx2 is different from that of Trx1 or the molecule number per cell of Trx2 is not enough to substitute Trx1. Expression of Trx2 from a high expression vector in a trxA Ϫ strain is necessary to answer this question. E. coli Trx2  7. Rate of insulin reduction as a function of temperature. 0.5 g of Trx1, Trx2, and ⌬Trx2 were incubated for 5 min at different temperatures and then analyzed with the insulin assay. Two separate measurements were made for each protein, and the average values are shown. q, Trx2; E, Trx2 plus DTT; f, ⌬Trx2; Ç, Trx1. and C. nephridii Trx3 are the only examples where prokaryotic thioredoxins have two extra pairs of CXXC motifs. We showed that the activity of Trx2 is dependent on the redox state of the protein, since preincubation with DTT increases insulin reduction. Moreover, elimination of the N-terminal region of the protein not only abolished the DTT dependence but also rendered the protein more resistant to heat inactivation. Sitedirected mutagenesis of these cysteine residues is needed to elucidate their role in this redox regulation.
It is also interesting to point out that the disposition of the four cysteine residues in the N terminus of Trx2 resembles the structure of a zinc finger. Zinc finger motifs have been described in various DNA-binding proteins, and the importance of the cysteine residues is supported by the observation that modification of these residues inhibited enzymatic activity (46). E. coli primase is an example of a bacterial protein with only one zinc finger domain that binds specific sequences in the DNA (47). In fact, the zinc finger domain of primase is located in the N terminus of the protein, similar to E. coli Trx2. We are currently exploring the possibility that E. coli Trx2 might contain a zinc finger motif.
The C-terminal part of the E. coli Trx2 is highly homologous to the rest of the prokaryotic thioredoxins with many amino acid residues conserved apart from those of the active site, including those necessary for protein-protein interactions or for maintenance of the native structure of the molecule. There is a Pro-108 at the position equivalent to Pro-76 in Trx1 that may serve to stabilize the active site. Ala-61, Trp-63, Asp-58, and Lys-89 (Ala-29, Trp-31, Asp-26, and Lys-57 in Trx1), which are invariant between thioredoxins, are also conserved in Trx2 (48). Prolines and glycines are common residues in or close to bends in proteins. Additionally, they are highly conserved in thioredoxins and fulfill this function (Gly-84, Gly-92, and Pro-64 in Trx1). There are also Gly residues in Trx2 (Gly-116, Gly-124); however, there is an Arg-96 at the corresponding position of Pro-64.
Trx2 is an atypical thioredoxin with an N-terminal extension whose potential functions still represent an open question. Although the biological function of Trx2 remains unknown, maintenance of the reducing environment in cytoplasm is likely to be such a function. Determination of the redox potential of Trx2 as well as disruption of its gene are necessary to answer these questions. Whether Trx2 is represented in higher organisms than prokaryotes will also be of great interest. We finally propose the nomenclature trxC for the gene and Trx2 for the protein corresponding to this novel thioredoxin in E. coli.