Direct NMR observation of the thioredoxin-mediated reduction of the chloroplast NADP-malate dehydrogenase provides a structural basis for the relief of autoinhibition.

The chloroplastic NADP-dependent malate dehydrogenase (NADP-MDH) catalyzing the reduction of oxaloacetate into L-malate is regulated by light. Its activation results from the thioredoxin-mediated reduction of two disulfides, located, respectively, in N- and C-terminal sequence extensions typical of all NADP-dependent light-regulated forms. Site-directed mutagenesis studies and the resolution of the three-dimensional structure of the oxidized (inactive) Sorghum vulgare enzyme showed that the C-terminal Cys(365)-Cys(377) disulfide constrains the C-terminal extension to fold into the active site where it acts as an internal inhibitor. In the present study, two-dimensional proton NMR spectra of an engineered NADP-MDH rendered monomeric by a 33-amino acid deletion at the N terminus (38 kDa) revealed that a 15-amino acid-long C-terminal peptide (Ala(375) to C-terminal Val(389)) acquired an increased mobility upon reduction, allowing its direct sequence-specific NMR assignment. The location of the flexible peptide in the sequence suggests that the first part of the C-terminal peptide is still folded near the core of the enzyme, so that cysteines 365 and 377 remain in proximity to allow for an efficient reoxidation/inactivation of the enzyme.

(NADP-MDH (EC 1.1.1.82)) possesses the particularity to be regulated by light (1). The light activation is mediated by reduction of disulfide bridges through the ferredoxin-thioredoxin system (2). Compared with the NAD-dependent MDHs, the amino acid sequences of the chloroplastic MDHs exhibit specific extensions at their N and C termini (3). In the S. vulgare enzyme the length of these extensions is 40 and 17 amino acids, respectively. Previous chemical and site-directed mutagenesis studies (4 -7) identified two regulatory disulfide bridges, whose thioredoxin-specific reduction is necessary for enzyme activation. Both disulfides are located in the extensions: the Cys 24 -Cys 29 disulfide at the N terminus and the Cys 365 -Cys 377 disulfide at the C terminus. Recently, the structure of the inactive, oxidized chloroplastic NADP-MDH was solved and provided a structural basis for the regulatory role of the Cys 365 -Cys 377 disulfide (8,9). The C-terminal disulfide was shown to constrain the C-terminal peptide to fold into the active site, thereby obstructing substrate access. Biochemical data emphasized the role of the negative charge of the side chain of the penultimate glutamate in locking the extension inside the active site (10). It was suggested that the C-terminal inhibitory peptide would be released upon reduction, making the active site accessible (8). To better help understand the conformational rearrangements occurring during reduction, the structure of an active reduced enzyme would be most valuable. However, the need to use reduced thioredoxin to activate the enzyme makes crystallization of the active form difficult. As a first step toward a comprehensive structural representation of the activation process, NMR spectra of oxidized and reduced forms of NADP-MDH were recorded. For this purpose, we have used an engineered MDH truncated in 33 residues at the N terminus. This truncated form is a monomer of 38 kDa (⌬N-MDH) still functional and redox-regulated, even though its activation by reduced thioredoxin is very fast compared with the wild-type enzyme. NMR measurements clearly show that a C-terminal peptide of 15 amino acids is released from the active site upon reduction and acquires an increased flexibility yielding locally narrow NMR line widths despite the relatively high molecular weight of the recombinant enzyme. Interactions between the C-terminal extension residues and the core structure of the protein in the oxidized form as inferred from the crystal structure are discussed together with the NMR data obtained for the reduced form in order to explain the efficient reversible activation/inactivation redox control of the chloroplast MDH.

EXPERIMENTAL PROCEDURES
Materials-Restriction endonucleases, DNA modification enzymes, T4 DNA ligase, and T4 DNA polymerase were obtained from Appligene. DEAE-Sephacel and Matrex red A chromatographic supports were, respectively, from Amersham Pharmacia Biotech and Millipore. Chemicals (purchased from Sigma, Roche Molecular Biochemicals, or Prolabo) were of analytical grade. Oligonucleotides were purchased from Eurogentec and Life Technologies, Inc. Radiolabels were from Amersham Pharmacia Biotech Escherichia coli strain XL1 blue (CLON-TECH) was used to produce high yields of plasmids and M13 singlestranded DNA. E. coli strain RZ1032 (Amersham Pharmacia Biotech) was used to produce dU-substituted M13 single-stranded DNA for site-directed mutagenesis. E. coli strain BL21 (DE3) (11) was used for the production of mutated NADP-MDHs encoded by recombinant pET vectors. M13 mp19 phage (Amersham Pharmacia Biotech) was used for site-directed mutagenesis and pET-8c (11) for the production of recombinant NADP-MDHs. Bacteria were grown at 37°C on Luria broth medium; ampicillin at 50 g/ml was added when the bacteria carried * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
§ Recipients of a Ph.D. fellowship from the French Ministère de l'Education Nationale de la Recherche et de la Technologie.
plasmids conferring drug resistance.
Engineering, Expression, and Purification of the Recombinant Proteins-MDH cDNA encoding a protein deleted in 33 amino acids at the N terminus (⌬N-MDH) was engineered by site-directed mutagenesis as described previously (5). A mutant where Cys 207 was replaced by an Ala (C207A ⌬N-MDH) was engineered as described in Ref. 12, by introducing a cassette bearing the mutation C207A into ⌬N-MDH. Both mutant proteins were produced with high yields using the pET/BL21 E. coli production system and purified using previously described procedures (5) combining ammonium sulfate fractionation, DEAE-Sephacel ion exchange chromatography, and affinity chromatography on Matrex red A. The purity of the preparations was checked by SDS-polyacrylamide gel electrophoresis.
Activation and Enzyme Activity Assay-The activation kinetics of the MDH mutants were measured by preincubating the enzymes with 10 mM (Ϯ)-dithiothreitol and 20 M recombinant thioredoxin (either from E. coli or from Chlamydomonas reinhardtii) in 100 mM Tris-HCl buffer, pH 7.9. Samples were taken out at regular time intervals and injected into a spectrophotometer cuvette containing the reaction medium composed of 140 M NADPH, 780 M oxaloacetate in 100 mM Tris-HCl buffer, pH 7.9. The activity was recorded as a decrease in absorbance at 340 nm due to the oxidation of NADPH.
NMR Experiments-All the NMR spectra were recorded using a Bruker Avance DRX 500 ( 1 H ϭ 500 MHz) spectrometer using a 5-mm ( 1 H, 13 C, 15 N) triple-resonance probe head, equipped with a supplementary self-shielded z gradient coil. Data were processed using the Bruker XwinNMR or GIFA V.4.22 (14) softwares. All homonuclear two-dimensional experiments were recorded with 512 (t 1 ) ϫ 1024 (t 2 ) complex data points (32 scans per t 1 increments). The data were apodized with shifted sine bell and gaussian window functions in both F 1 and F 2 dimensions after zero-filling in the t 1 dimension. The same window functions were applied for both redox states of the proteins. DQF-COSY (15), TOCSY (Hartmann-Hann spectroscopy) (16,17), and NOESY (18, 19) experi-ments were recorded at 303 K. 1 H chemical shifts were quoted relative to the solvent (H 2 O) chemical shift at the temperature of the study (4.725 ppm at 303 K). The solvent signal was suppressed with the WATERGATE sequence using a 3-9-19 pulse sequence with z gradients (20,21). For DQF-COSY an additional very low power presaturation of the water resonance during the relaxation delay to minimize the radiation damping effect. The quadrature detection in the t 1 dimension was achieved using the States-TPPI method (22). The spectral width of all experiments was 10 ppm (5000 Hz) with a carrier frequency on-resonance with the water resonance. For TOCSY experiments, a MLEV pulse sequence was used for the isotropic mixing, arranged according to the clean-TOCSY method (23) to minimize the signals arising from the rotating frame Overhauser effect. In the NOESY experiment, a water flip back pulse was used and gradients were added during t 1 to minimize the radiation damping effect (24). The total mixing times were 40 ms for TOCSY spectra and 75 and 150 ms for NOESY spectra.

RESULTS AND DISCUSSION
Biochemical Activity of the ⌬N-MDH-The chloroplastic NADP-dependent malate dehydrogenase is a homodimeric enzyme activated by the reduction of two disulfide bridges located at the N and C termini of each monomer. Its activation rate is slow, with a distinct lag phase. The truncated ⌬N-MDH is a monomer devoid of the N-terminal disulfide but still totally inactive in the oxidized form. After activation, its catalytic properties are identical to those of the WT enzyme (5). It constitutes a simplified system where the effect of the C-terminal extension can be studied independently from the interaction with the N-terminal extension. In addition, the 38kDa molecular mass of the monomeric ⌬N-MDH makes it potentially accessible to high resolution NMR in solution (25), allowing more detailed future structural studies.
It has been shown previously that this mutant exhibits fast activation kinetics, suggesting that the reduction of the Nterminal disulfide was the rate-limiting step of the activation. This N-terminal extension is also necessary for the dimerization (6). For the WT dimeric enzyme, site-directed mutagenesis The sequence-specific assignment in the reduced enzyme is indicated. The amino acid numbering refers to the wild-type MDH from sorghum (8).

NMR Observation of the Chloroplast MDH Activation 34540
and crystallographic studies clearly demonstrated that the Cterminal extension was bound inside the active site, acting as an internal inhibitor in the oxidized enzyme (Fig. 2) (8, 10). It could be expected that upon reduction by thioredoxin, the Cterminal inhibitory extension should be released from the active site. The fast activation rate of the mutants missing the N-terminal disulfide bridge suggested that a local shift of the C-terminal extension could be sufficient to unblock the active site without complex conformational changes.
Recent experiments have shown that in the presence of thiol oxidants the ⌬N-MDH could form dimers through an intersubunit disulfide bridge linking the Cys 207 of both subunits (12). Therefore, a C207A mutant of ⌬N-MDH, unable to form covalent dimers, was also studied, in order to avoid potential dimerization problems. This mutant is a good alternative to the use of the ⌬N-MDH, having similar activation properties and catalytic parameters (data not shown).
NMR Spectra of Reduced and Oxidized Forms of ⌬N-MDH and the C207A ⌬N-MDH Mutant-From the crystallographic structure of the oxidized MDH (Fig. 2), it was proposed that the reduction of the C-terminal Cys 365 -Cys 377 disulfide bridge would release the last two C-terminal residues Glu 388 and Val 389 from the active site, making it accessible to the substrate. This unblocking of the MDH active site could be the consequence of an increased mobility of the C-terminal extension upon reduction of the Cys 365 -Cys 377 disulfide bridge by thioredoxin. Such a mobility difference could be characterized by the comparison of both redox states of the chloroplast MDH by solution NMR which is sensitive to molecular motions.
The NMR spectra of the oxidized form of the ⌬N-MDH and the ⌬N-MDH C207A mutant indicate no COSY cross-peaks and only a few peaks in the TOCSY and NOESY experiments of which no sequence-specific assignment was possible (Fig. 1). This result is not surprising due to the molecular weight of the proteins that induces fast transverse relaxation rates and broad NMR lines, making ineffective the transfer of magnetization. For the reduced MDH forms, several narrower NMR signals could be easily discerned from the rest of the broad amide proton resonances in the region 7.6 -8.6 ppm of the one-dimensional spectra (data not shown). These narrower signals gave rise to a number of cross-peaks in the different two-dimensional spectra in the same 7.6 -8.6-ppm region as shown in Fig. 1. Thirteen spin systems were detected in the TOCSY spectra of the reduced ⌬N-MDH and ⌬N-MDH C207A mutant, with magnetization transfer from H N to H ␤ side chains proton resonances. The COSY and TOCSY correlations in the aliphatic-aliphatic part of the spectra allowed the recognition of specific amino acid spin systems. Eleven of the thirteen H N to H ␣ TOCSY cross-peaks were also detected in the DQF-COSY spectrum, as illustrated with boxes in Fig. 1. The NOEs observed ( Fig. 1) indicated that the 13 amino acids are sequentially linked in both proteins.
Two valine spin systems, one alanine, one threonine, one leucine, one glycine, one proline together with four AMX spin systems could be unambiguously be identified. The three remaining spin systems were attributed to either a glutamic acid, a glutamine, or a methionine spin system. Moreover, one of the AMX spin systems indicated strong NOESY connectivities with aromatic resonances characteristic of a tyrosine residue and was therefore tentatively assigned to this amino acid. In the NOESY spectra, only sequential (i, i ϩ 1) NOEs were detected. NOE d ␤N correlations were observed in some cases, together with very weak d NN NOE correlations. Four amino acid sequences could be identified as indicated in Table I. The possibilities to find the peptide sequences in the protein from residues 1 to 364 or in the C-terminal extension of the protein (residues 365-389) were examined. The results indicate that the peptide exhibiting the assigned TOCSY and NOESY cross peaks in the reduced form of the enzyme can be found only in  I Peptide sequences identified from the two-dimensional spectra of the reduced ⌬N-MDH AMX refers to an "AMX" NMR spin system (26).  NMR Observation of the Chloroplast MDH Activation 34541 C-terminal peptide Ala 375 -Val 389 , the last 15 residues of the protein (Tables I and II and Fig. 2). Therefore, it is clear that the C-terminal peptide of the enzyme is no longer bound to the core structure of the protein when the C-terminal Cys 365 -Cys 377 disulfide is reduced. The relative intensities of the sequential d ␣N and d NN NOEs and the absence of any (i, i ϩ 2) or longer distance NOEs, correlate with a random-coil conformation for the C-terminal peptide of the enzyme. The transverse relaxation times T 2 of the protons located in the C-terminal peptide were gained significantly by the actual additional motion, leading to observable magnetization transfers in the different two-dimensional NMR experiments. Structural Rearrangements of the C-terminal Part during the Activation Process-The assignment of the mobile peptide provides information about the structural rearrangements occurring upon reduction. The first part of the C-terminal peptide from Cys 365 to Asn 374 remains bound to the core structure as the corresponding residues keep a correlation time similar to that of the rest of the protein. The peptide is mobile from Ala 375 , i.e. only two residues before the second cysteine of the regulatory C-terminal disulfide Cys 365 -Cys 377 . This is probably a favorable feature to maintain the Cys 377 not too far away from Cys 365 , and the rest of the inhibitory C-terminal peptide not too far away from the whole protein. By this means, the inhibitory C-terminal peptide will easily find its way back to the active site when the photosynthetic electron transfer is stopped in the dark. This will allow MDH to readily reoxidize and consequently self-inactivate.
The conformational analysis of the C-terminal extension in the inactive, oxidized crystalline form of the enzyme (8)  . This indicates that residues 371-374 are more tightly bound to the protein than the rest of the C-terminal peptide. A number of interactions constrain the rest of the C-terminal extension to fold near the core structure of the protein in addition to the covalent interaction due to the disulfide bridge. In particular, Val 379 , Met 384 , and Leu 385 make hydrophobic interactions with residues of the rest of the protein and a number of hydrogen bonds are found between the peptide and the protein. We should assume that these interactions are not sufficient to hold the C terminus in the active site in the reduced form. However, these interactions undoubtedly stabilize the binding to the core structure when the C-terminal disulfide is oxidized.
Our work provides the first direct experimental observation of the in vitro thioredoxin-mediated activation of a thiol-regulated plant MDH. It shows that the activation is the result of the unobstruction of the active site by the acquisition of an additional mobility of the most C-terminal 15-residue stretch.