Replacement of Trp28 in Escherichia coli thioredoxin by site-directed mutagenesis affects thermodynamic stability but not function.

Escherichia coli thioredoxin contains two tryptophan residues (Trp and Trp) situated close to the active site disulfide/dithiol. In order to probe the structural and functional roles of tryptophan in the mechanism of E. coli thioredoxin (Trx), we have replaced Trp with alanine using site-directed mutagenesis and expressed the mutant protein W28A in E. coli. Changes in the behavior of the mutant protein compared with the wild-type protein have been monitored by a number of physical and spectroscopic techniques and enzyme assays. As expected, removal of a tryptophan residue causes profound changes in the fluorescence spectrum of thioredoxin, particularly for the reduced protein (Trx-(SH)), and to a lesser extent for the oxidized protein (Trx-S). These results show that the major contribution to the strongly quenched fluorescence of Trx-S in both wild-type and mutant proteins is from Trp, whereas the higher fluorescence quantum yield of Trx-(SH) in the wild-type protein is dominated by the emission from Trp. The fluorescence, CD, and 1H NMR spectra are all indicative that the mutant protein is fully folded at pH 7 and room temperature, and, despite the significance of the change, from a tryptophan in close proximity to the active site to an alanine, the functions of the protein appear to be largely intact. W28A Trx-S is a good substrate for thioredoxin reductase, and W28A Trx-(SH) is as efficient as wild-type protein in reduction of insulin disulfides. DNA polymerase activity exhibited by the complex of phage T7 gene 5 protein and Trx-(SH) is affected only marginally by the W28A substitution, consistent with the buried position of Trp in the protein. However, the thermodynamic stability of the molecule appears to have been greatly reduced by the mutation: guanidine hydrochloride unfolds the protein at a significantly lower concentration for the mutant than for wild type, and the thermal stability is reduced by about 10°C in each case. The stability of each form of the protein appears to be reduced by the same amount, an indication that the effect of the mutation is identical in both forms of the protein. Thus, despite its close proximity to the active site, the Trp residue of thioredoxin is not apparently essential to the electron transfer mechanism, but rather contributes to the stability of the protein fold in the active site region.

Thioredoxin (Trx), 1 a 12-kDa, heat-stable, redox-active protein, has been characterized in many functions from a wide variety of prokaryotic and eukaryotic cells (Holmgren, 1985;Gleason and Holmgren, 1988;Holmgren, 1989). All thioredoxins contain an active site cysteine disulfide/dithiol in a conserved sequence Trp-Cys-Gly-Pro-Cys. The disulfide of oxidized thioredoxin, Trx-S 2 , can be reduced to the dithiol by NADPH and a specific flavoprotein enzyme, thioredoxin reductase, while reduced thioredoxin, Trx-(SH) 2 , participates in a number of redox reactions mostly linked to reduction of protein disulfides. Thioredoxin has been most extensively studied in Escherichia coli. The three-dimensional structure of E. coli Trx-S 2 has been refined to 1.7 Å by x-ray crystallography (Katti et al., 1990), and high resolution structures of both Trx-S 2 and Trx-(SH) 2 have recently been calculated from multidimensional NMR data (Jeng et al., 1994). The protein contains 108 amino acid residues, including two tryptophans; one, Trp 31 , is conserved throughout known thioredoxin sequences, while the other, Trp 28 , is conserved throughout prokaryotes, being replaced by serine in eukaryotic sequences (Eklund et al., 1991). The active site disulfide is located on a protrusion of the molecule and is close to the two tryptophans both in the sequence (Trp 28 -Ala-Glu-Trp 31 -Cys 32 -Gly-Pro-Cys 35 -) and in the threedimensional structure (Katti et al., 1990;Jeng et al., 1994). The close proximity of the two tryptophans to the active site provides a probe for detecting structural changes between the oxidized and reduced molecule, since the tryptophan fluorescence is affected by the change of oxidation state.
Tryptophan fluorescence is highly quenched in Trx-S 2 . A 3.5-fold increase in fluorescence is observed upon reduction. This is thought to be due mainly to the removal of the fluorescence quenching provided by the disulfide bond (Holmgren, 1972), as well as to slight shifts in the positions of the tryptophan rings accompanying a local conformational change which may include a change in hydrogen bonding interactions of the tryptophan ring (Dyson et al., 1990;Jeng et al., 1994). A number of studies, including chemical modification of Trp 31 (Holmgren, 1981), comparison of yeast and calf thymus thioredoxins (Mérola et al., 1989), and studies of several Trp 31 mutants , indicated that this in-crease in fluorescence is due to Trp 28 . Oxidized Trp 31 mutant proteins have a very low tryptophan fluorescence emission from the remaining Trp 28 , but reduction results in a large (up to 11-fold) increase of fluorescence intensity . The Trp 28 aromatic ring is in close proximity to the disulfide in wt Trx-S 2 (Katti et al., 1990;Jeng et al., 1994), and changes in NMR chemical shifts upon reduction indicate that its position differs slightly in Trx-(SH) 2 (Dyson et al., 1989). Characterization of mutant thioredoxin with Trp 28 replaced by a nonaromatic residue can be used to determine the role of this residue in the changes of fluorescence emission intensity, as well as to elucidate any involvement in the oxidoreduction mechanism of thioredoxin. Also, since Trp 28 is a conserved residue in prokaryotes, it might be expected to be of structural importance. In this study, we have chosen to examine the mutant where Trp 28 is substituted by the smallest hydrophobic residue, alanine. While other substitutions, such as the insertion of serine, would be of some interest, since the mammalian thioredoxins have serine at this position, we considered that the disruption of structure caused by the insertion of a hydrophilic side chain deep into the hydrophobic pocket behind the active site might have a number of effects that could not be analyzed with any certainty. The W28A mutation is the simplest change, with fewest competing effects.
The present study also examines the role of Trp 28 in bacteriophage systems. The reduced form of E. coli thioredoxin is essential for an assembly of filamentous phages f1 and M13 (Russel and Model, 1985) and for phage T7 DNA replication (Mark and Richardson, 1976). T7 DNA polymerase consists of the T7 phage-encoded gene 5 protein (g5p) and reduced E. coli thioredoxin. Isolated g5p has a high single-stranded exonuclease activity, but only a very low 5Ј-3Ј polymerase and 3Ј-5Ј double-stranded exonuclease activities. Addition of Trx-(SH) 2 to g5p in vitro reconstitutes highly processive DNA polymerase and double-stranded exonuclease activities. To date, little is known about how thioredoxin acts to give T7 DNA polymerase high processivity. The role of Trx in T7 DNA polymerase is apparently different from its well known redox functions and is probably linked to a structural interaction which stabilizes the binding of g5p to a primer-template (Tabor et al., 1987;Slaby and Holmgren, 1989). Site-directed mutagenesis of E. coli thioredoxin with subsequent DNA polymerase assay has been successfully used for probing the participation of specific amino acid residues in the binding interaction with the gene 5 protein (Huber et al., 1986;.
M13-vector DNA, ␣-35 S-dATP, and other molecular biology materials were purchased from Boehringer Mannheim, Pharmacia Biotech Inc., and Amersham Corp. DNA sequencing was done by the dideoxy method using T7 DNA polymerase kit and equipment from Pharmacia Biotech Inc.
DNA Cloning Techniques-Mutagenesis primer was purchased from Scandinavian Gene Synthesis AB and phosphorylated at the 5Ј end by T4 polynucleotide kinase. The procedure of Vandeyar et al. (1988) for generation of mutants was used. Phage DNA of the subclone in M13mp19, containing antisense orientation of the trxA gene, served as the template. Synthesis of complementary mutated DNA strand, removal of parental strand, transfection of competent E. coli CU9276 strain, and the subcloning into the expression plasmid pUC118 were performed as described previously .
Most of the DNA work was done according to standard procedures (Sambrook et al., 1989). Screening for mutants was performed directly by sequencing M13 DNA purified by the method of Vieira and Messing (1987).
Preparation of Mutant Thioredoxin-A 2-liter culture of the JF521/ pUC118-trxA-W28 E. coli strain was grown in a fortified broth with 50 g/ml ampicillin to about A 550 nm ϳ 1.0. The cells were then induced with 0.1 mM final concentration of isopropyl-␤-D-thiogalactopyranoside and the incubation continued another 8 h with vigorous shaking before harvesting. The purification of W28A thioredoxin was carried out by DEAE-cellulose and Sephadex G-50 chromatography as described .
Insulin was used to determine the protein disulfide reductase activity of thioredoxin coupled to NADPH oxidation via thioredoxin reductase (Holmgren, 1979). The reactions, containing 0.5-1.5 M thioredoxin, 0.4 mM NADPH, 160 M insulin, 2 mM EDTA in 0.1 M phosphate buffer, pH 7.0, were started by addition of 200 nM thioredoxin reductase. The rate of NADPH oxidation was calculated from the decrease in absorbance at 340 nm, using a molar extinction coefficient for NADPH of 6,200 M Ϫ1 ⅐cm Ϫ1 .
DNA Polymerase Assay-Stimulation of DNA polymerase activity of gene 5 protein (3 nM final concentration) by wild-type (wt) or mutant thioredoxins of varying concentrations was measured by standard assay of polymerase activity on denatured salmon sperm DNA at 37°C (Slaby et al., 1984). Thioredoxin was added directly to the assay mixtures containing gene 5 protein followed by at least 10-min preincubation at 4°C. The reactions were started by a transfer to 37°C water bath. One unit of DNA polymerase catalyzes the incorporation of 10 nmol of total nucleotide into an acid-insoluble product in 30 min at 37°C (Huber et al., 1986).
Fluorescence Measurements-Protein fluorescence was measured with a thermostatted Shimadzu RF-510LC spectrofluorimeter at 18°C. Excitation of fluorescence was at 283 nm and emission spectra from 300 to 400 nm were recorded. The samples contained 1 M concentrations of thioredoxin in potassium phosphate buffer, pH 7.0. Oxidized thioredoxin was reduced by dithiothreitol (DTT) in 1 mM final concentration. The denaturation experiments were carried out by addition of stock 8 M guanidine hydrochloride to the samples.
NMR Measurements-All spectra were acquired on a Bruker AMX500 spectrometer operating at 500 MHz for protons as described previously (Dyson et al., 1988(Dyson et al., , 1989. 2QF COSY (Rance et al., 1983) and 2Q (Braunschweiler et al., 1983) spectra were acquired at 298 K for both the oxidized and reduced forms of the W28A mutant. Samples were approximately 4 mM in 100 mM potassium phosphate buffer, pH 6.0. Spectral widths were commonly 7042 Hz with 4,000 complex points in 2 and 14085 Hz with 512 complex points in 1 . Spectra were Fourier transformed on a Sun workstation using the FTNMR software of Dennis Hare to give total matrix size of 2,000 ϫ 1,000 real points. Spectra were referenced to an internal standard of dioxane at 3.75 ppm.
Circular Dichroism Measurements-Circular dichroism spectra were recorded on JASCO J700 or Aviv 61DS spectropolarimeters using a 1-mm path length at 24°C. Protein samples in 50 mM phosphate buffer, pH 7.0, and varying concentrations of guanidine HCl were prepared by dilution of stock solutions to a final thioredoxin concentration of 50 M. The samples were reduced with DTT at a final concentration of 1 mM and equilibrated for 15 min before CD spectra from 250 to 215 nm were recorded. The fraction of unfolded thioredoxin at each concentration of guanidine HCl was determined from the percent ellipticity change at 222 nm relative to the values at 0 and 5 M guanidine HCl.
Analytical Methods-Protein concentrations were determined from the absorbance at 280 nm using molar extinction coefficients of 13,700 To determine the amino acid composition, protein samples were lyophilized and hydrolyzed with 6 M HCl, 0.5% phenol for 24 h at 110°C in vacuo and analysis was performed with Beckman 121M amino acid analyzer. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis was run on 8 -12% gradient gels using the Phast System from Pharmacia.

RESULTS AND DISCUSSION
Site-directed Mutagenesis and Expression of W28A Thioredoxin-To obtain a template for mutagenesis, the trxA gene from pUC118-trxA (Lim et al., 1985) was inserted into M13mp19 via the EcoRI and HindIII sites. A primer with the sequence 5Ј-CTCTGCCGCGAAATCGACG-3Ј containing a 2-base mismatch in its central region was hybridized to the single-stranded M13mp19-trxA DNA. Mutant thioredoxin genes were identified by sequencing at a frequency of about 10%. The replicative form of M13mp19 carrying the trxA mutation was cleaved and recloned to the pUC118 expression vector utilizing the EcoRI and HindIII sites.
The host E. coli strain JF521 lacking the chromosomal wildtype trxA gene was used for expression of the pUC118-trxAencoded W28A thioredoxin. The expression was inducible by isopropyl-␤-D-thiogalactopyranoside, since pUC118 contains the lac promoter just in front of the trxA gene. The insertion of the fragment into the pUC118 polylinker region introduces a TAA stop codon between lacZ and trxA, thus excluding a protein fusion .
Purification and Physical Properties-A yield of about 30 -50 mg of mutant thioredoxin/liter of culture was obtained from the JF521/pUC118-trxA-W28A cells. The W28A protein has the same chromatographic properties during purification as wildtype thioredoxin, and electrophoretic properties are also similar: W28A thioredoxin appears as a single band in the same position as wt thioredoxin in sodium dodecyl sulfate gels (data not shown). The amino acid composition indicates that W28A thioredoxin contains only one tryptophan residue (data not shown), and this is confirmed by analysis of the ultraviolet and fluorescence spectra. The ultraviolet spectra of the mutant protein show the expected decrease of tryptophan absorbance at 280 mm (data not shown), and the corresponding lower molar extinction coefficient is in good agreement with the disulfide content of W28 thioredoxin as determined by NADPH oxidation catalyzed by thioredoxin reductase.
NMR Spectra of W28A-Diagnostic 2Q and 2QF COSY NMR spectra of the mutant thioredoxin confirm that the protein is folded in solution and that its structure is very similar to that of the wt protein. A portion of the 1 H 2Q spectrum of Trx-(SH) 2 is shown in Fig. 1, and several of the resonance assignments, obtained by analogy with those of wt (Dyson et al., 1989) are indicated.
Activity of W28A Thioredoxin in Oxidoreduction Catalysis-The activity of wt and mutant Trx with thioredoxin reductase was assayed with DTNB as a substrate (Moore et al., 1964). At pH 8.0, DTNB reduction rates with relatively high concentrations of thioredoxin reductase (100 nM) are 63 M⅐min Ϫ1 for wt and 73 M⅐min Ϫ1 for W28A. The rates of reduction of insulin disulfides at pH 7.0 are 26 M Trx-S 2 ⅐min Ϫ1 for wt and 30 M Trx-S 2 ⅐min Ϫ1 for the mutant. No decrease in the rates of reaction is evident: indeed W28A thioredoxin appears to be a marginally better substrate for thioredoxin reductase and slightly more efficient in the reduction of insulin disulfides than the wild-type protein.
Interaction with the Bacteriophage T7 Gene 5 Protein-The interactions of mutant and wt thioredoxins with g5p were tested in a reconstitution assay in vitro with pure proteins. The results obtained under identical conditions with wt thioredoxin and W28A mutant are shown in Fig. 2, where polymerase activity is plotted against the logarithm of increasing thioredoxin concentration at a constant concentration of g5p of 3 nM. A quantitative evaluation of the reconstitution assays was achieved by using Scatchard plot analysis to determine an observed equilibrium dissociation constant, K obs (Huber et al., 1986). This constant depends on the initial concentration of g5p and corresponds to the thioredoxin concentration yielding halfmaximum enzyme activity, assuming that the polymerase ac-tivity is proportional to the concentration of the g5p-thioredoxin complex. In our experiment, K obs of 4 nM for wt and 6 nM for W28A thioredoxin were calculated. The maximal polymerase activities of the reconstituted complexes were similar with both thioredoxins (1,800 units/nmol).
Tryptophan Fluorescence Measurements-The fluorescence emission spectra between 300 and 400 nm with excitation at 283 nm are shown in Fig. 3 for wt and mutant thioredoxins W28A and W31A. The fluorescence behavior of the three proteins differs significantly. The quantum yield of tryptophan fluorescence in wt Trx-S 2 is very low (Q ϭ 0.02). Reduction of the disulfide bond by DTT causes a 3.5-fold increase in fluorescence at pH 7.0 (Holmgren, 1972). The oxidized form of the W31A mutant shows a decrease of approximately 50% in tryptophan fluorescence compared to wt Trx-S 2 and a large increase in fluorescence (10-fold) upon reduction to give a fluorescence significantly greater than that wild-type Trx-S 2 . The fluorescence quantum yield of W28A Trx-S 2 is about 20% lower than wt Trx-S 2 and increases only   FIG. 4. A and B, tryptophan fluorescence spectra of wt Trx (A) and W28A Trx (B) in guanidine hydrochloride. E, native Trx-S 2 ; q, native Trx-(SH) 2 ; Ⅺ, Trx-S 2 in 1 M GdnHCl; f, Trx-(SH) 2 in 1 M GdnHCl; Ç, Trx-S 2 in 5 M GdnHCl; å, Trx-(SH) 2 in 5 M GdnHCl. C, denaturation of thioredoxins by GdnHCl: variation in tryptophan fluorescence emission at 350 nm with increasing concentrations of GdnHCl for wt Trx-S 2 (E), wt Trx-(SH) 2 (q), W28A Trx-S 2 (Ⅺ), and W28A Trx-(SH) 2 (f).
2-fold after reduction to give a fluorescence for W28A Trx-(SH) 2 only slightly greater than that of wt Trx-S 2 . These results imply that the strongly quenched fluorescence observed for Trx-S 2 is mainly contributed by Trp 31 , whereas the high fluorescence intensity observed in wt Trx-(SH) 2 is dominated by the emission from Trp 28 . The striking change in quantum yield of fluorescence upon reduction of all three proteins is not accompanied by a shift in the wavelength of the emission maximum, indicating that any change in conformation upon reduction is small and localized to the active site region. This is confirmed by comparison of the solution structures of wt Trx-S 2 and Trx-(SH) 2 , which show only small differences in structure upon change of oxidation state (Jeng et al., 1994).
Thermodynamic Stability of wt and W28A Thioredoxin-Denaturing agents cause major changes in the fluorescence spectra of both oxidized and reduced thioredoxin (Holmgren, 1972). The effect of the W28A mutation on the structural stability of thioredoxin was investigated by fluorescence measurements at different concentrations of guanidine hydrochloride (GdnHCl). The effect of GdnHCl on the fluorescence spectrum of wild-type thioredoxin is shown in Fig. 4A. Major changes in the fluorescence of Trx-S 2 are not observed until the concentration of GdnHCl exceeds 2 M. An increase of the GdnHCl concentration to 5 M results in a 5-fold increase in fluorescence with a shift in the wavelength of the emission maximum from 345 to 356 nm, indicating unfolding of the protein (Holmgren, 1972). The fluorescence of wt Trx-(SH) 2 is increased about 2-fold in 1 M GdnHCl with no change in the emission maximum wavelength; a slight increase in fluorescence intensity is observed between 1 and 5 M GdnHCl, but the emission maximum shifts to 356 nm, the same as for unfolded Trx-S 2 . The effect of GdnHCl on the fluorescence spectra of the W28A mutant is different (Fig. 4B). Although the fluorescence intensity increases between 1 and 5 M GdnHCl, the fluorescence of both oxidized and reduced W28A is already significantly increased in 1 M GdnHCl (Fig. 4C) and the emission maximum is already shifted to 356 nm. These results indicate that the overall structure W28A Trx is markedly changed even at a low concentration of GdnHCl.
Circular Dichroism Spectra-The CD spectrum provides a measure of the integrity of the secondary structure of the protein. Since thioredoxin contains a significant amount of helical structure, the negative ellipticity at 222 nm can be used as a measure of the extent of unfolding of the molecule, although contributions from the ellipticity of the five-stranded ␤-sheet ( max ϭ 215 nm) may also be present. Differential stability of the helix and sheet portions of thioredoxin have also been inferred from hydrogen exchange studies (Jeng and Dyson, 1995).
The ellipticity at 222 nm as a function of GdnHCl concentration is shown in Fig. 5 for wt and W28A thioredoxin. For both proteins, the oxidized form is significantly more stable than the reduced form to GdnHCl denaturation, consistent with the greater thermal stability of Trx-S 2 (see below). Interestingly, the concentration of GdnHCl at which half of the protein is denatured differs by exactly the same amount between the two forms of W28A Trx compared with wt, indicating that changing Trp 28 does not affect the relationship between the structures of the two oxidation states. This observation is entirely consistent with the results of the enzymatic assays, which indicate no significant difference in activity for the mutant.
However, it is clear from Fig. 5 that the stability of W28A in either oxidation state is significantly decreased relative to the wild-type protein. The GdnHCl concentration for half of the protein to be denatured is decreased from 2.38 M to 1.73 M for Trx-S 2 and from 1.60 M to 1.00 M for Trx-(SH) 2 . A similar plot (not shown) for the fluorescence intensity results shown in Fig.  3 is more complex due to the complexity of the fluorescence changes, including changes in emission wavelength, upon reduction of either wt or W28A mutant Trx, but still shows the same overall behavior. These results indicate that although the protein is correctly folded under native conditions, as indicated by the similarity of the mutant and wt NMR spectra (Fig. 1), the stability of the W28A mutant protein has been significantly decreased by the change of tryptophan for alanine at position 28.
The thermal stability of the oxidized and reduced forms of the mutant and wild-type thioredoxins was estimated by following the ellipticity at 222 nm as a function of temperature (data not shown). The transition temperatures of the mutant proteins are significantly lowered from those of the wild-type proteins (80°C compared with 88°C for Trx-S 2 and 70°C compared with 80°C for Trx-(SH) 2 ). Once again, the relative difference between the two oxidation states has been preserved, consistent with the functional similarity of the mutant to the wild-type and with the behavior of the proteins toward denaturation in GdnHCl.

CONCLUSIONS
The two tryptophan residues are highly conserved in thioredoxins (Eklund et al., 1991). Trp 31 is conserved throughout prokaryotic and eukaryotic proteins and its replacement with alanine causes significant changes in function . Consistent with this, Trp 31 is situated in an unusual position on the surface of the molecule (Katti et al., 1990;Jeng et al., 1994). By contrast, Trp 28 is conserved only in prokaryotic thioredoxins (Eklund et al., 1991) and is largely buried in the hydrophobic core of the molecule (Jeng et al., 1994). The residue that replaces Trp 28 in eukaryotes is serine, which is not even remotely homologous to tryptophan; this argues that tryptophan at this position is probably not necessary for the function of thioredoxin. Its conservation among such a large and varied group as the prokaryotes argues that it is necessary for some other reason, possibly for stabilizing local structure in the vicinity of the active site cysteine residues.
We have shown that the replacement of Trp 28 with alanine results in no significant change in the functional properties of thioredoxin. The NMR, fluorescence, and CD spectra of the mutant protein under native conditions all indicate that the FIG. 5. Guanidine HCl unfolding of wt and W28A thioredoxin measured by circular dichroism at pH 7.0. The fraction denatured was determined from the percent ellipticity change relative to the ellipticity difference between fully native (without GdnHCl) and fully denatured (5 M GdnHCl) thioredoxin at 222 nm. E, wt Trx-S 2 ; q, wt Trx-(SH) 2 ; Ç, W28A Trx-S 2 ; å, W28A Trx-(SH) 2 .
protein is correctly folded, consistent with its correct function both in assays that require reduction or oxidation of the cysteines and in assays such as the T7 DNA polymerase reaction that measure tightness and specificity of binding. However, both fluorescence and CD measurements as a function of added guanidine hydrochloride and temperature clearly show that the stability of the W28A protein is significantly decreased relative to wild type. Furthermore, the effects of the mutation on the stability of the protein are the same for both Trx-S 2 and Trx-(SH) 2 , consistent with the integrity of function in this mutant. It is perhaps somewhat surprising that such a significant mutation as the replacement of a tryptophan by an alanine in close proximity to the active site should have as little effect on the function of the protein as it does in this case. These results are an indication of the robust nature of thioredoxin and of its stability of function toward a wide variety of mutational challenges.