Identification of residues of spinach thioredoxin f that influence interactions with target enzymes.

The necessity for two types of thioredoxins (Trx f and m) within chloroplasts of higher plants that mediate the same redox chemistry with various target enzymes is not well understood. To approach this complex issue, we have applied site-directed mutagenesis to the identification of residues of Trx f that affect its binding to and selectivity for target enzymes. Based upon amino acid sequence alignments and the three-dimensional structure of Escherichia coli thioredoxin, putative key residues of Trx f were replaced with residues found at corresponding positions of Trx m to generate the mutants K58E, Q75D, N74D, and deletion mutants ΔAsn-74 and ΔAsn-77. Kinetics of activation of oxidized recombinant sorghum leaf NADP-dependent malate dehydrogenase and oxidized spinach chloroplastic fructose-1,6-bisphosphatase by wild-type Trx f, wild-type Trx m, and Trx f mutants were compared. All of the mutants are less efficient than wild-type Trx f in the activation of fructose-1,6-bisphosphatase and are altered in both S0.5 and Vmax. In contrast to literature reports, the activation of NADP-dependent malate dehydrogenase does not display rate saturation kinetics with respect to the concentration of Trx f, thereby signifying very weak interactions between the two proteins. The mutants of Trx f likewise interact only weakly with NADP-dependent malate dehydrogenase, but the apparent second-order rate constants for activation are increased compared to that with wild-type Trx f. Thus, Lys-58, Asn-74, Gln-75, and Asn-77 of Trx f contribute to its interaction with target enzymes and influence target protein selectivity.

The necessity for two types of thioredoxins (Trx f and m) within chloroplasts of higher plants that mediate the same redox chemistry with various target enzymes is not well understood. To approach this complex issue, we have applied site-directed mutagenesis to the identification of residues of Trx f that affect its binding to and selectivity for target enzymes. Based upon amino acid sequence alignments and the three-dimensional structure of Escherichia coli thioredoxin, putative key residues of Trx f were replaced with residues found at corresponding positions of Trx m to generate the mutants K58E, Q75D, N74D, and deletion mutants ⌬Asn-74 and ⌬Asn-77. Kinetics of activation of oxidized recombinant sorghum leaf NADP-dependent malate dehydrogenase and oxidized spinach chloroplastic fructose-1,6-bisphosphatase by wild-type Trx f, wild-type Trx m, and Trx f mutants were compared. All of the mutants are less efficient than wild-type Trx f in the activation of fructose-1,6-bisphosphatase and are altered in both S 0.5 and V max . In contrast to literature reports, the activation of NADPdependent malate dehydrogenase does not display rate saturation kinetics with respect to the concentration of Trx f, thereby signifying very weak interactions between the two proteins. The mutants of Trx f likewise interact only weakly with NADP-dependent malate dehydrogenase, but the apparent second-order rate constants for activation are increased compared to that with wild-type Trx f. Thus, Lys-58, Asn-74, Gln-75, and Asn-77 of Trx f contribute to its interaction with target enzymes and influence target protein selectivity.
Thioredoxins are ubiquitous, redox-active proteins that mediate thiol-disulfide exchanges with target enzymes (for reviews see Refs. [1][2][3][4]. Two chloroplastic thioredoxins have been identified, designated f and m, which can be reduced by ferredoxin as catalyzed by ferredoxin-thioredoxin reductase (for review, see Ref. 5). Both of these thioredoxins are crucial to the light regulation of CO 2 assimilation through the modulation of the redox status of various enzymes. Thioredoxins activate (or inactivate) target enzymes by a step-wise, thiol-disulfide exchange in which the enzyme disulfide undergoes nucleophilic attack by an active-site sulfhydryl of the Trx 1 (Cys-46 in spinach Trx f) (6) to form a protein-protein mixed disulfide. The proximal active-site sulfhydryl (Cys-49 in Trx f) then intramolecularly attacks the mixed disulfide, generating reduced enzyme and oxidized Trx. Trx f has been reported to selectively activate fructose-1,6-bisphosphatase (FBPase), sedoheptulose-1,7-bisphosphatase, phosphoribulokinase, and NADP-linked glyceraldehyde-3-phosphate dehydrogenase (1,2). On the contrary, Trx m appears selective for NADP-dependent malate dehydrogenase (MDH) and chloroplastic glucose-6-phosphate dehydrogenase (7)(8)(9). A comparison of the amino acid sequences between the two thioredoxins shows only 29% identity but complete conservation of the active-site sequence Cys-Gly-Pro-Cys (10).
The basis for the existence of two types of thioredoxins within the chloroplasts of higher plants that preferentially mediate the same redox chemistry of different target enzymes is unknown. As one approach to this issue, we are exploring, by site-directed mutagenesis, the roles of residues that appear unique to Trx f relative to other thioredoxins. One such residue is Lys-58, which is replaced by glutamate in spinach Trx m, Escherichia coli Trx, and indeed the majority of thioredoxins in other organisms ( Fig. 1) (10). Based on the structure of E. coli Trx (11)(12)(13), Lys-58 is positioned on the surface, is salt-bridged with residue 110 (Asp in Trx f; Lys in E. coli Trx and Trx m), and is located near the active site. Hence, we were prompted to construct and characterize the K58E mutant of Trx f as reported herein.
Residues 74, 75, and 77 were also examined. These residues are located near a conserved, putative hydrophobic contact surface encompassing Gly-33, Pro-34, Ile-75, Pro-76, Val-91, Gly-92, and Ala-93 of E. coli Trx (10,14,15). Specifically, the region of 74 -77 appears to be on the same general face of the molecule as both the hydrophobic contact site and the active site (based upon visual inspection of the E. coli Trx structures reported by Dyson et al. (11) and Katti et al. (12)). One sequence alignment (16,17) (Fig. 1A) suggests that the Asn-74 of Trx f aligns with an Asp in Trx m and Asp-61 of E. coli Trx, resulting in an insertion of an asparagine at position 77. Supportive of this alignment, a D61N E. coli Trx mutant generated by Jacquot and colleagues showed an increase in efficiency of spinach FBPase activation (16). An alternative sequence alignment (10), however, indicates that Asn-74 represents an insertion, with Gln-75 aligning with aspartic acid in Trx m (Fig. 1B). Given this uncertainty, two deletion mutants (⌬N74 and ⌬N77), as well as two single-substitution mutants (N74D and Q75D), were constructed. The kinetic parameters of these four mutants and the K58E mutant of Trx f in the activation of both FBPase and MDH were determined and compared to the corresponding parameters of wild-type Trx f and Trx m. Bacto-tryptone and bacto-yeast extract were obtained from Difco. Dithiothreitol, N,NЈ-bis-(2-hydroxyethyl)glycine, HEPES, and Tris (free base) were purchased from Research Organics. Assay coupling enzymes, NAD, NADPH, oxaloacetate, EDTA, and bovine serum albumin were from Sigma. FBP was from Boehringer Mannheim.
Site-directed Mutagenesis-Site-directed mutants of spinach Trx f were generated according to the procedure of Kunkel (18) using the plasmid pFL404 (6). Single-stranded dU-substituted template was generated in the E. coli strain CJ236, which had been infected with M13K07 helper phage. Oligonucleotide primers used to encode the denoted structural changes were as follows: pCGATTGTAACCAG-GAAAAGACATTAGCAAAGG for ⌬N77; pAAATACGAAGAGCTAGCA-GAG for K58E; pCTCGATTGTGACCAGGAAAAC for N74D; pGCTCG-ATTGTCAGGAAAACA for ⌬N74; and pGATTGTAACGATGAAAAC-AAG for Q75D. The primer extension reaction products were electroporated into competent E. coli MV1190 cells with the Bio-Rad Gene Pulser. Following selection on LB medium containing ampicillin, candidate mutants were screened by dideoxy-terminator cycle sequencing (19) using U.S. Biochemical Corp. Sequenase version 2.0 and instructions therein.
A single colony of E. coli strain MV1190 containing the desired expression plasmid was used to inoculate 20 ml of 2 ϫ YT (20) medium containing 50 g/ml ampicillin and 1% (v/v) glycerol. By shaking at 37°C overnight, a stationary culture was generated. This culture was then diluted 1:50 into 500 ml of the same medium in a 2-liter baffled flask and shaken at 250 rpm at 37°C for 4 h (A 600 nm ϭ 2.3). Isopropylthio-␤-D-galactopyranoside was then added to a final concentration of 100 M. After shaking for an additional 3 h, the cells were harvested by centrifugation, washed once with 20 mM N,NЈ-bis-(2-hydroxyethyl)glycine-NaOH (pH 8.0) and 1 mM EDTA, and stored at Ϫ80°C.
Activation of FBPase by Trx-Rates of activation of oxidized FBPase (40 g/ml) as a function of Trx concentration were determined at 25°C in 50 mM HEPES-KOH (pH 7.8), 0.1 mM EDTA, and 5 mM dithiothreitol. Periodically, 40-l aliquots of the activation mixture were withdrawn and assayed for FBPase activity. Assay solutions (1 ml) consisted of 50 mM HEPES-KOH (pH 7.8), 50 M EDTA, 1 mM MgSO 4 , 4 mM FBP, 2 units of phosphoglucose isomerase, 3 units of glucose-6-phosphate dehydrogenase, and 1 mM NAD ϩ . The formation of NADH was followed spectrophotometrically at 340 nm. One unit of FBPase activity is represented by an absorbance change of 6.22/min. Activation of MDH by Trx-Rates of activation of oxidized MDH (2.6 g/ml) as a function of Trx concentration were determined at 4°C in 100 mM Tris-HCl (pH 7.9), 10 mM dithiothreitol, 1 mM EDTA, and 1 mg/ml bovine serum albumin. At various times, 20-l aliquots of the activation mixture were removed and assayed for MDH activity. The assays, in a 100-l volume, were performed at 25°C in 100 mM Tris-HCl (pH 7.9), 1 mM EDTA, 1 mg/ml bovine serum albumin, 1.5 mM oxaloacetate, and 0.3 mM NADPH. The oxidation of NADPH was followed spectrophotometrically at 340 nm. One unit of MDH activity corresponds to an absorbance change of 6.22/min. The rates reported are normalized to an assay volume of 1 ml (i.e. the same volume as in the case of the FBPase assays). properties of the wild-type thioredoxins. Consistent with prior reports (9, 23), wild-type Trx f is far superior to wild-type Trx m in the activation of FBPase, displaying a 35-fold higher affinity and a 25-fold increased V max (Fig. 3A and Table I). All of the mutants analyzed are impaired relative to wild-type Trx f but are still superior to Trx m when analyzed with FBPase ( Fig. 3B and Table I). The largest deviation from the S 0.5 value of 0.9 M determined with wild-type Trx f occurs with the Asn-77 deletion mutant, in which case the affinity is lowered about 15-fold. Among the mutants analyzed, replacement of Lys-58 by a glutamyl residue has the largest impact on V max with a decrease of 3-fold.

RESULTS
As anticipated, Trx m is far more effective as an activator for MDH than for FBPase. Indeed, the rates of activation of MDH by Trx m are so rapid that reliable monitoring could only be achieved at 4°C as opposed to room temperature. Even without correction for this differential in temperature, the V max for Trx m in the activation of MDH is 21 mU/min (Fig. 4) compared to 2 mU/min in the activation of FBPase (Fig. 3A and Table I); however, the S 0.5 values for the interaction of Trx m with the two enzymes are both approximately 30 M. Despite the efficiency of activation of MDH by Trx m, a comparison of the activation profiles for MDH by Trx f and Trx m argues against preferential regulation by the latter (Fig. 4). Even though rate saturation with respect to Trx f concentrations up to 50 M cannot be demonstrated, signifying very weak association between Trx f and MDH, the rates of activation of MDH by Trx f are more rapid than those with Trx m throughout the concentration range examined.
The activation kinetics of MDH by the mutants of Trx f are complex, exhibiting pronounced sigmoidicity as seen with wildtype Trx m (Fig. 5). However, analogous to wild-type Trx f, the mutants do not show rate saturation in activation of MDH. Thus, the relative effectiveness of Trx f mutants as activators of MDH are expressed as apparent second-order rate constants (Table II) calculated from the linear region of the activation curves (Fig. 5). Based on this criterion, all mutants are more effective than wild-type Trx f in the activation of MDH. DISCUSSION Despite both qualitative and quantitative variations among published data, several laboratories have confirmed differential selectivities of Trx f and Trx m for target enzymes as discovered in pioneering studies by Buchanan and colleagues (3,9). As the redox potentials of these two thioredoxins are identical (Ϫ210 mV) (24), their selectivities must reflect differential affinities for target enzymes and/or differential propensities for sulfhydryl-disulfide exchanges within the Trx-enzyme complexes. The present study was predicated on the supposition that recognition and selectivity determinants of Trx f and Trx m could be determined by site-directed mutagenesis as guided by clearly discernible structural differences between the two thioredoxins. Satisfyingly, the residue (Lys-58) and segment (positions 74 -77) of Trx f chosen for change are shown to influence its efficiency (V max /S 0.5 ) and selectivity in the activa-  tion of target enzymes. The impaired efficiency of K58E, relative to wild-type Trx f, in the activation of FBPase is due primarily to a lower V max , indicative of a slower rate of thiol-disulfide exchange within the productive complex. Slower exchange could reflect misalignment of the exchanging groups or altered pK a values of the active-site sulfhydryls of Trx f. Certainly, the functional group of residue 58 is too far removed, 15-20 Å as extrapolated from the three-dimensional structures of E. coli Trx (11,12), human Trx (25), and Anabaena Trx-2 (26), from the active-site sulfhydryls to directly influence their reactivities. However, either suggested outcome could result from conformational perturbations brought about by the lysyl to glutamyl substitution. Any such perturbations must be slight as judged by the mere 2-fold increase in S 0.5 for activation of FBPase.
The efficiency of activation of FBPase by Trx f is very sensitive to changes in residues at positions 74 -77 with greater than 9-fold declines among three of the five mutants examined. In contrast to K58E, these mutants are less efficient than wildtype Trx f due primarily to weakened interactions with FBPase as gauged by S 0.5 values. Therefore, we conclude that residues 74 -77 serve as an important recognition site for target enzymes. The earlier finding that replacement of Asp-61 of E. coli Trx with Asn, as located at the corresponding position 74 in Trx f improves the activation parameters with FBPase (16), is consistent with this conclusion. Our conclusion is further supported by structural information derived from both human and Anabaena Trx. With human Trx, long range nuclear Over-hauser-effect contacts between Ala-29 (Thr-43 of spinach Trx f) and Asp-60 (Asn-74 or Gln-75 of Trx f) and also between Trp-31 (Trp-45 of Trx f) and Asp-60 are observed in the reduced but not the oxidized forms of the Trx (25). This suggests that localized conformational changes in these regions occur upon binding and subsequent activation of target enzymes. Furthermore, the solution structure of human Trx complexed with a peptide from the transcription factor NFB shows that the guanidinium group of Arg-57 of NFB forms a salt bridge with the side chain carboxylates of Asp-58 and Asp-61 of the Trx (corresponding to Asp-72 and Gln-75/Glu-76 of Trx f) (27). Functionally, Trx-2 from Anabaena resembles Trx f (28), and its overall threedimensional structure is very similar to that of E. coli Trx (26). However, the regions that correspond to residues 74/75-87 of Trx f differ considerably and thus appear poised structurally for preferential contact with target enzymes. Specifically, Saarinen et al. (26) state that differences in secondary structures of these regions and that amino acid substitutions in the hydrophobic cores are the causes of relative altered positions of these residues.
Whereas all of the engineered structural changes of Trx f were counterproductive with respect to activation of FBPase, they were beneficial with respect to activation of MDH, thereby validating the conversion of Trx f to a more "m-like" thioredoxin. Although V max and S 0.5 values for the activation of MDH by the Trx mutants could not be determined due to the absence of rate saturation even at 1000-fold molar excess of the given thioredoxin, the observed activation rates were greater with the mutants than with wild-type Trx f at concentrations exceeding 30 M. The Trx f mutants were also more akin to Trx m with respect to kinetics of MDH activations, displaying sigmoidicity as opposed to linearity with wild-type Trx f.
In addition to identifying functional and selectivity elements of Trx f, the present study partially addresses some of the published conflicts about relative target enzyme selectivities of Trx f and Trx m. Although all accounts agree that Trx f is superior to Trx m in the activation of FBPase (9,23,29), some allege that this enzyme is totally refractive to Trx m (30,31). Given the inefficiency of Trx m in the activation of FBPase, 550-fold lower than that of Trx f, we surmise that negative reports merely reflect insufficient concentrations and reaction times.
Beyond these readily reconcilable differences, reports concerning activation of MDH range from a striking preferential effectiveness of Trx m (9,32) to superiority of Trx f (30, 33) as we observe. These disparities could be due to differences in conditions for activation or differences in approaches to determining kinetic parameters. Shortly after the discovery of Trx f and Trx m (9), preferential activation of MDH by the latter was reported to be dependent on the presence of high concentrations of chloride (32). Although we observe stimulation of thioredoxin-dependent activation of MDH by 250 mM NaCl, the  Fig. 4, each data point indicates the initial rate of activation by the specified concentration of Trx at 4°C. Curve fittings were performed as described in Fig. 4. impact is similar for both Trx m and Trx f. We thus believe that methods, rather than conditions, are the more likely basis of variable results among different laboratories. In most prior publications, the level of activation at a given Trx concentration is based on a single, fixed time point assay; the rate of activation is then assumed to be linear throughout the time period selected. Thus, to some extent, equilibria are being determined rather than rates at which equilibria are attained. The kinetic patterns presently reported are based on true initial rates of activation, whereby an incubation mixture of Trx and target enzyme is sampled repeatedly in order to reveal the linear phase of time-dependent activation. Prior oversights to clearly distinguish rates and equilibria can readily account for the wide range of apparent affinities of spinach Trx f for spinach FBPase; e.g. S 0.5 values of 1.6 nM (23), 0.19 M (34), and 1 M (6, 16) have been reported. Given this enormous variation of apparent affinities with one enzyme system, reports of inverse selectivities of Trx f and Trx m are understandable. Although two prior studies include true rates of activation of Trx-regulated enzymes, quantitative comparisons with our data are precluded. One report was restricted to an examination of the Trx m/MDH system and did not entail a complete profile of the rate dependence on Trx m concentration (35). In the other report, the activation of FBPase by Trx f was examined at molar ratios that gave rise to first-order kinetics with respect to Trx f concentration (i.e. absence of rate saturation) (36).
Based on our finding of clear-cut kinetic superiority of Trx f relative to Trx m in the in vitro activation of MDH, we suggest that the physiological role of Trx m should be reassessed.