C172S Substitution in the Chloroplast-encoded Large Subunit Affects Stability and Stress-induced Turnover of Ribulose-1,5-bisphosphate Carboxylase/Oxygenase*

Previous work has indicated that the turnover of chloroplast ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco; EC 4.1.1.39) may be controlled by the redox state of certain cysteine residues. To test this hypothesis, directed mutagenesis and chloroplast transformation were employed to create a C172S substitution in the Rubisco large subunit of the green alga Chlamydomonas reinhardtii. The C172S mutant strain was not substantially different from the wild type with respect to growth rate, and the purified mutant enzyme had a normal circular dichroism spectrum. However, the mutant enzyme was inactivated faster than the wild-type enzyme at 40 and 50 °C. In contrast, C172S mutant Rubisco was more resistant to sodium arsenite, which reacts with vicinal dithiols. The effect of arsenite may be directed to the cysteine 172/192 pair that is present in the wild-type enzyme, but absent in the mutant enzyme. The mutant enzyme was also more resistant to proteinase K in vitro at low redox potential. Furthermore, oxidative (hydrogen peroxide) or osmotic (mannitol) stress-induced degradation of Rubiscoin vivo was delayed in C172S mutant cells relative to wild-type cells. Thus, cysteine residues could play a role in regulating the degradation of Rubisco under in vivo stress conditions.

Ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco 1 ; EC 4.1.1.39) catalyzes the photosynthetic fixation of CO 2 through the Calvin cycle, the metabolic route that accounts for most of the carbon input into the biosphere (reviewed in Refs. 1 and 2). However, O 2 is mutually competitive with CO 2 at the same large-subunit active site, and oxygenation of RuBP leads to the loss of CO 2 through the seemingly wasteful photorespiratory pathway. In all plants and green algae, Rubisco is composed of eight 55-kDa large subunits (encoded by the chloroplast rbcL gene) and eight 15-kDa small subunits (encoded by a family of nuclear rbcS genes). Photosynthetic prokaryotes have plant-type hexadecameric holoenzymes, homodimeric en-zymes composed of large-subunit homologues, or both (reviewed in Ref. 3). The x-ray crystal structures of representative enzymes have been solved (4 -7), and in each case, the catalytic site has been found to reside at the interface between large subunits.
Because the difference between the velocities of carboxylation and oxygenation determines net photosynthetic CO 2 fixation (1,8), many studies have been devoted to understanding the structural basis for CO 2 /O 2 specificity by screening for chloroplast Rubisco mutants in the green alga Chlamydomonas reinhardtii (reviewed in Ref. 9) or by expressing directed mutant prokaryotic enzymes in Escherichia coli (reviewed in Ref. 2). Directed mutagenesis of eukaryotic Rubisco is hindered by the need to transform the highly polyploid chloroplast genome. However, the chloroplasts of Chlamydomonas and tobacco can be transformed (e.g. Refs. 10 and 11), and directed mutagenesis has already been used to investigate Chlamydomonas Rubisco (10,12,13).
In addition to its enzymatic function, Rubisco is a remarkably abundant protein storing a considerable amount of nitrogen. The reutilization of this nitrogen through the turnover of the enzyme and the mobilization of its amino acids during natural or stress-induced senescence plays a role in the nutritional balance of plants (reviewed in Ref. 14). Proteolysis of Rubisco is indeed a prominent feature in the stress responses of plants and algae, and several studies have shown that Rubisco is oxidatively modified prior to its degradation (15)(16)(17)(18)(19)(20). Moreover, oxidation of Cys residues in vitro has been proposed to switch Rubisco from a protease-resistant to protease-sensitive form (21)(22)(23). The critical residues have not yet been identified, but because oxidative modification of Rubisco is widespread among species (15)(16)(17)(18)(19)(20), it is likely that these residues are conserved. Redox regulation by oxidation/reduction of Cys thiol groups has been demonstrated for other chloroplast proteins, including carbohydrate metabolism enzymes (24), ATP synthase (25,26), components of photosynthetic membrane complexes (27), and translation factors (28). Nonetheless, there is as yet no evidence for a physiological role of such a regulatory mechanism acting on Rubisco stability/turnover in vivo. Thus, we have started to use directed mutagenesis and chloroplast transformation of Chlamydomonas to scan for the proposed regulatory properties of Cys residues. Accordingly, the highly conserved Cys-172 has been changed to Ser, and the mutant Rubisco has been studied both in vitro and in vivo.

EXPERIMENTAL PROCEDURES
Strains and Culture Conditions-C. reinhardtii 2137 mt ϩ is the wild-type strain (29). Rubisco mutant 18-7G mt ϩ was used as the host for transformation. Mutant 18-7G lacks Rubisco holoenzyme and requires acetate for growth (30). It results from an rbcL nonsense mutation that changes large-subunit Trp-66 to amber (31). All strains were maintained at 25°C in the dark with medium containing 10 mM sodium acetate and 1.5% Bacto-agar (29). For biochemical analysis, photosynthesis-competent strains were grown in acetate liquid medium (without agar) in either 500-ml or 2.5-liter half-filled Erlenmeyer flasks under continuous white light (15 microeinsteins/m 2 /s) on a rotary shaker at 28°C. For Rubisco extraction, cells were harvested at stationary phase (ϳ10 7 cells/ml). Growth curves were determined in 50-ml cultures with acetate or minimal (without acetate) medium at 15 microeinsteins/m 2 /s. Cells were counted with a hemocytometer.
Directed Mutagenesis and Chloroplast Transformation-A 2951-base pair HaeIII DNA fragment (bases Ϫ175 to 2776), containing the entire rbcL gene (bases 1-1428) (32), was cloned into SmaI-digested pBluescript SK ϩ (Stratagene). Site-directed mutagenesis was performed according to the method of Vandeyar et al. (33) by employing a T7-GEN kit from U. S. Biochemical Corp. as described previously (10). To produce the C172S mutation, the rbcL gene sequence TGT (bases 514 -516) was changed to TCT. This mutation abolished an RsaI site and expedited the initial screening of transformants. The mutation was confirmed by performing DNA sequencing with an IsoTherm DNA sequencing kit (Epicentre Technologies Corp.) and synthetic oligonucleotide primers (Genosys Biotechnologies, Inc.). The plasmid containing the C172S rbcL mutant gene was named pLS-C172S. Chloroplast transformation with this plasmid was performed via microprojectile bombardment (34) as described previously (10,13). Although the phenotype conferred by the rbcL mutant gene could not be predicted, transformation of the 18-7G rbcL mutant with pLS-C172S was found to yield photosynthesis-competent colonies. Successive rounds of single-colony isolation were performed to ensure homoplasmicity of the mutant genes (10,12). Total DNA was then purified (31); the rbcL gene was amplified by the polymerase chain reaction (35); and the entire coding region was sequenced. Only the intended mutation was present. The rbcL mutant strain created by directed mutagenesis and transformation was named C172S.
Rubisco Purification and Assay-Cells were collected by centrifugation at 2500 ϫ g for 5 min. Pellets were weighed, frozen with liquid N 2 , and stored at Ϫ80°C. Frozen cells were resuspended (0.25 g/ml) in 100 mM Tris-HCl, pH 7.8, 20 mM MgCl 2 , 5 mM 2-mercaptoethanol, and 2 mM phenylmethylsulfonyl fluoride and sonicated at 0°C. After adding 2% (w/v) polyvinylpolypynolidone and stirring for 5 min at 4°C, the crude extract was centrifuged at 35,000 ϫ g for 10 min. Rubisco was purified from the supernatant by ammonium sulfate precipitation, sucrose gradient fractionation, and DEAE-cellulose chromatography as previously detailed (22). The final protein preparation contained Ͼ90% Rubisco (ascertained by densitometry of Coomassie Blue-stained gels) and was essentially free of nucleic acids (A 280/260 nm Ͼ 1.7). Purified Rubisco was activated in 100 mM Tris-HCl, pH 8.2, 10 mM MgCl 2 , and 10 mM NaHCO 3 by gel filtration through a Sephadex G-25 column (Amersham Pharmacia Biotech PD-10) and further diluted to an absorbance of 0.31 at 280 nm (ϳ0.2 mg/ml). RuBP carboxylase activity was measured as the incorporation of acid-stable 14 C from NaH 14 CO 3 (22). The amount of total protein in the extracts was determined by the method of Lowry et al. (36).
Circular Dichroism Spectroscopy-Far (195-240 nm) and near (250 -300) UV circular dichroism spectra were obtained in a CD-6 dicrograph (Jobin-Yvon) using quartz cells of 0.1-and 10-mm optical paths, respectively. Spectra were recorded at 0.5 nm/s with an integration constant of 1 s. After averaging three runs for each sample, solvent spectra were subtracted from those for purified Rubisco (0.2 mg/ml in 100 mM Tris-HCl, pH 8.2, 10 mM MgCl 2 , and 10 mM NaHCO 3 ). Molar ellipticity is expressed as degrees cm 2 /dmol, where the molar units refer to amino acid residues in the far-UV spectrum and to Rubisco heterodimer in the near-UV region.
Thermal Stability Measurements-Purified Rubisco samples (0.2 mg/ml in 100 mM Tris-HCl, pH 8.2, 10 mM MgCl 2 , and 10 mM NaHCO 3 ) were placed in Eppendorf tubes and submerged in a water bath at 40 or 50°C. At the indicated times, 20-l aliquots of the samples were mixed with 180 l of 100 mM Tris-HCl, pH 8.2, 10 mM MgCl 2 , and 10 mM NaHCO 3 in preincubated (30°C) plastic vials (Bio-vial, Beckman Instruments). The mixtures were further incubated at 30°C for 15 min before the RuBP carboxylase assay was started by the addition of RuBP and NaH 14 CO 3 (22).
Redox Buffers and Susceptibility to Proteolysis-Aliquots (200 l) of purified Rubisco (0.2 mg/ml) were combined with 100 l of cystamine/ cysteamine mixtures of constant monomeric concentration (120 mM in 100 mM Tris-HCl, pH 8.2, 10 mM MgCl 2 , and 10 mM NaHCO 3 ). Different cystamine/cysteamine ratios or other oxidative agents were added as specified. Samples were incubated at 30°C for 2 h, and 20-l aliquots were then assayed for RuBP carboxylase activity. To measure proteolysis, 80-l aliquots were incubated with 20 l of freshly prepared proteinase K (0.5 g/ml in the same buffer; Roche Molecular Biochemi-cals) at 30°C for 5 min. Proteolysis was stopped by adding 10 l of 22 mM phenylmethylsulfonyl fluoride in 100% ethanol and keeping the mixture on ice for 10 min. Then, 55 l of 0.25 M Tris-HCl, pH 6.8, 2 M 2-mercaptoethanol, 6.5% SDS, and 25% (w/v) glycerol were added, and the samples were boiled for 4 min. Aliquots (10 -20 l) were subjected to SDS-polyacrylamide gel electrophoresis, immunoblotting with rabbit antiserum directed against Euglena gracilis Rubisco, and blot densitometry as described previously (17,22).

RESULTS
Recovery and Phenotype of the C172S Mutant-The C172S mutant was recovered by transforming the 18-7G rbcL nonsense mutant with pLS-C172S DNA and selecting for photosynthetic ability on minimal medium in the light. Thus, it was clear that the C172S mutant Rubisco retained substantial catalytic activity. The C172S mutant strain was indistinguishable from the wild type when grown on acetate-supplemented medium, and no difference was observed when its rate of growth in liquid culture was compared with that of the wild type under photoautotrophic, photoheterotrophic, or heterotrophic conditions (Fig. 1). However, crude extracts of the C172S mutant cells had less RuBP carboxylase activity (ϳ50 or 25% less in extracts of autotrophically or photoheterotrophically grown cells, respectively) than wild-type extracts. This difference was due both to a decreased enzyme content of the C172S mutant cells and to a lower specific activity of the purified C172S enzyme (ϳ20% lower than that of the wild-type enzyme).
Structural Stability of Purified C172S Rubisco-Circular dichroism spectra obtained for the purified C172S and wild-type Rubisco enzymes were coincident, both in the far-and near-UV regions (Fig. 2), indicating that the single-atom substitution in C172S Rubisco did not substantially affect the secondary or tertiary structure. However, when incubated at elevated temperatures of 40 or 50°C, C172S Rubisco was inactivated faster than the wild-type enzyme (Fig. 3). These results indicate that the loss of a possible Cys-172-Cys-192 disulfide bond (5) might be responsible for a decrease in the thermal stability of C172S Rubisco. However, reduction of disulfide bonds with DTT caused similar increases in thermal inactivation for both the mutant and wild-type enzymes at 50°C (Fig. 3B).
Redox Modulation of Purified Rubisco-Exposure of purified Rubisco to different thiol-oxidizing agents resulted in enzyme inactivation to a variable extent (Table I). Cystamine and 5,5Јdithiobis(2-nitrobenzoate) substantially reduced RuBP carboxylase activity, whereas copper ions and oxidized glutathione had less effect. The only difference between the C172S and wild-type Rubisco enzymes was observed with arsenite (Table I). Arsenite, a reagent for vicinal dithiols (37), decreased wild-type RuBP carboxylase activity by 26%, but the effect on the mutant enzyme was negligible. In wild-type Rubisco, arsenite may link the thiol groups of the closely associated Cys-172 and Cys-192 residues (5), one of which is absent in the C172S mutant enzyme.
Incubation of the purified C172S and wild-type Rubisco enzymes with different ratios of cystamine and cysteamine showed that similar amounts of oxidative inactivation took place at similar redox potentials (Fig. 4). A cystamine/cysteamine ratio of 2, which is close to the value of 1.5 found for Euglena Rubisco (22), caused a 50% inactivation of both the C172S and wild-type Rubisco enzymes (Fig. 4). The proteolytic sensitivity of the Rubisco large subunit also increased with increasing ratios of cystamine to cysteamine (Fig. 5), and this has also been observed for the Euglena enzyme (22). However, C172S mutant Rubisco was more resistant to proteolysis under moderately oxidative conditions than was the wild-type enzyme. For example, a disulfide/thiol ratio of 4.5 was required to obtain the same amount of mutant Rubisco proteolysis as observed for the wild-type enzyme at a ratio of 3.5 (Fig. 5).
In Vivo Turnover of the C172S and Wild-type Rubisco Enzymes in Stressed Cells-Exposure of exponentially growing cells to strong osmotic (0.45 M mannitol) or oxidative (10 mM H 2 O 2 ) stress induced the degradation of Rubisco (Fig. 6), but the loss of holoenzyme was slower in the C172S mutant cells than in the wild-type cells under both stress conditions. Despite the reduced structural stability of C172S mutant Rubisco in vitro (Fig. 3), its stress-induced half-life in vivo was 2-3 times greater than that of wild-type Rubisco (Fig. 6). This held true regardless of whether the amount of Rubisco was calculated as a percentage of initial Rubisco content (Fig. 6) or as a percentage of total protein (data not shown), indicating that the observed difference in turnover rate was selective for Rubisco. DISCUSSION Thirty years ago, Sugiyama et al. (38) discovered that chemical modification of Rubisco thiols with p-chloromercuribenzoate inactivated the enzyme, disassembled its subunits, and facilitated its proteolysis. More recently, several studies have shown that Rubisco is oxidatively modified and degraded in vivo under stress conditions (16,17,39). The discovery that oxidative modification of Cys residues also sensitizes purified Rubisco to proteolysis (21) leads one to consider that Cys oxidation in vivo may mark the enzyme for degradation during natural senescence or under stress conditions that are known to produce an oxidative environment in the chloroplast (23). Despite indirect evidence such as finding that some Rubisco Cys residues are indeed oxidized at the onset of stress (17), no direct connection between Cys oxidation and Rubisco degradation in vivo has been demonstrated. Now that it is possible to use directed mutagenesis and chloroplast transformation of Chlamydomonas as a means for examining eukaryotic Rubisco (9), we decided to examine the potential role of Cys oxidation in the regulated degradation of Rubisco.
Only three pairs of Cys residues are close enough within the Rubisco crystal structure to have the potential for forming disulfide bonds (5,40). The large-subunit Cys-247/Cys-247 (between subunits) and Cys-172/Cys-192 residues are conserved among eukaryotic Rubisco enzymes (5,7). Whereas the tobacco and Chlamydomonas Rubisco enzymes may also have a disulfide bond between Cys-449 and Cys-459 (5), spinach Rubisco contains Thr-449. Cys-172 and Cys-192 are particularly interesting because they appear to form a covalent bond in unactivated tobacco Rubisco, but are only in van der Waals contact in activated Rubisco (5,40). Affinity labeling of spinach Rubisco with N-bromoacetylethanolamine phosphate also revealed that Cys-172 and Cys-459 were chemically modified in unactivated Rubisco, but not in activated Rubisco (41). Although Cys residues do not play a direct role in catalysis (reviewed in Ref. 2), they may have a dynamic role in function, structure, or regulation.
When Cys-172 was replaced by Ser (substitution of only a single atom) in the Chlamydomonas large subunit, no signifi-cant effect on the growth of the mutant cells was observed (Fig.  1), and C172S mutant Rubisco assembled into a native structure (Fig. 2). Thus, the identity of Cys-172 and the proposed dynamics of Cys-172-Cys-192 disulfide bond formation (5,40) are not essential for holoenzyme assembly or catalysis. This is particularly surprising because previous analysis of Chlamydomonas rbcL mutations showed that substitutions G171D and T173I eliminate RuBP carboxylase activity without affecting the amount of Rubisco in vivo (42,43).
C172S Rubisco is less stable to elevated temperature than  the wild-type enzyme (Fig. 3). This might be expected if the C172S substitution disrupts a Cys-172-Cys-192 disulfide bond. However, activated mutant Rubisco was used in this study, and based on interpretation of the tobacco Rubisco crystal structures (5,40), Cys-172 and Cys-192 are not expected to form a disulfide bond under this condition. There are additional reasons for doubting that the C172S mutation exerts its influence by disrupting a disulfide bond.
Treatment of both the C172S mutant and wild-type Rubisco enzymes with the disulfide-exchanging thiol DTT decreased the thermal stability of both enzymes to the same extent (Fig.  3B). This indicates that the increase in thermal sensitivity is due to DTT acting on disulfides other than the Cys-172/Cys-192 pair. Nonetheless, one may argue that the putative Cys-172-Cys-192 disulfide bond is resistant to DTT, as is the Cys-247-Cys-247 disulfide bond that is buried within the hydrophobic core of spinach Rubisco (7,44).
In contrast to the wild-type enzyme, the RuBP carboxylase activity of C172S Rubisco was found to be resistant to arsenite inactivation (Table I). Because arsenite reacts exclusively with vicinal thiol groups (37), it is likely that arsenite bridges the only free thiols (Cys-172 and Cys-192) that are close together in wild-type Rubisco (5), thereby reducing the RuBP carboxylasespecific activity of the wild-type enzyme by 25% (Table I). The mutant enzyme would be resistant to arsenite because it lacks one of these free thiols (Cys-172). Thus, the results are consistent with the idea that Cys-172 and Cys-192 do not form a disulfide bond in activated wild-type Rubisco.
The C172S substitution does not affect the critical redox potential for enzyme inactivation (Fig. 4), further indicating that a Cys-172-Cys-192 disulfide bond is not required for the catalytic function of Rubisco. However, in comparison to wildtype Rubisco, the C172S mutant enzyme is somewhat more resistant to proteolysis in vitro at lower redox potentials (Fig. 5). Furthermore, mutant Rubisco degradation in vivo is slower than that of wild-type Rubisco under oxidative (hydrogen peroxide) or osmotic (mannitol) stress conditions (Fig. 6). This indicates a subtler role for Cys-172 than merely stabilizing Rubisco structure through a disulfide bond. In fact, the delayed degradation of Rubisco in stressed cells is opposite to what one would expect from the lower thermal stability of C172S Rubisco observed in vitro (Fig. 3). Perhaps the delayed degradation of C172S Rubisco observed in vivo results from alteration of an oxidative signal transduction pathway that triggers Rubisco turnover. Cys-172 would be a significant component of this pathway.
The absence of Cys-172 shifts the critical redox potential for Rubisco proteolysis to more oxidative conditions (Fig. 5) and protects Rubisco from degradation under stress conditions (Fig.  6). However, the C172S substitution does not completely prevent induced Rubisco turnover, indicating that more than one Cys residue may contribute to redox regulation of Rubisco degradation. Additional Cys substitutions may help to elucidate the redox signaling pathway.