Complementing substitutions at the bottom of the barrel influence catalysis and stability of ribulose-bisphosphate carboxylase/oxygenase.

The temperature-conditional photosynthesis-deficient mutant 68-4PP of Chlamydomonas reinhardtii results from a Leu-290 to Phe substitution in the chloroplast-encoded large subunit of ribulose-1,5-bisphosphate carboxylase/oxygenase (EC 4.1.1.39). Although this substitution occurs relatively far from the active site, the mutant enzyme has a reduced ratio of carboxylation to oxygenation in addition to reduced thermal stability in vivo and in vitro. In an attempt to understand the role of this region in catalysis, photosynthesis-competent revertants were selected. Two revertants, named R96-4C and R96-8E, were found to arise from second-site mutations that cause V262L and A222T substitutions, respectively. These intragenic suppressor mutations increase the CO2/O2 specificity and carboxylation Vmax back to wild-type values. Based on the crystal structure of the spinach holoenzyme, Leu-290 is not in van der Waals contact with either Val-262 or Ala-222. However, all three residues are located at the bottom of the α/β-barrel active site and may interact with residues of the nuclear encoded small subunits. It appears that amino acid residues at the interface of large and small subunits can influence both stability and catalysis.

Ribulose-1,5-bisphosphate carboxylase/oxygenase (EC 4.1.1.39, Rubisco) 1 initiates both photosynthetic carbon assimilation and photorespiration (see Refs. 1 and 2 for review). The enzyme generates either two molecules of phosphoglycerate by carboxylation of RuBP or one molecule each of phosphoglycerate and phosphoglycolate by oxygenation of RuBP. Because phosphoglycolate enters the fruitless photorespiratory pathway and leads to the loss of CO 2 (see Ref. 3 for review), an increase in carboxylation or a decrease in oxygenation would likely improve plant productivity (see Ref. 1 for review). The ratio of carboxylation to oxygenation at any given concentrations of CO 2 and O 2 is defined by the CO 2 /O 2 specificity factor, ⍀ ϭ V c K o /V o K c , where V c and V o are the V max values for carboxylation and oxygenation, and K c and K o are the Michaelis constants for CO 2 and O 2 , respectively (4). Because CO 2 and O 2 are mutually competitive at the same active site, the differ-ential stabilization of the carboxylation and oxygenation transition states ultimately determines ⍀ (5,6).
The Rubisco holoenzyme in the chloroplasts of plants and green algae is composed of eight copies each of large and small subunits (see Refs. 1 and 2 for review). A family of nuclear rbcS genes encodes the 15-kDa small subunits, whereas the chloroplast rbcL gene encodes the 55-kDa large subunits. The small subunit is synthesized as a 21-kDa precursor in the cytosol and processed to mature form upon entry into chloroplasts (see Refs. 7 and 8 for review). Holoenzyme assembly is then facilitated by the action of chloroplast chaperonin 60 (see Ref. 9 for review). The active site of Rubisco is formed at the interface between the carboxyl-terminal ␣/␤-barrel domain of one large subunit and the amino-terminal domain of a second large subunit (10 -12). Rubisco also requires an additional nuclear encoded protein, named Rubisco activase, for its activation in vivo (see Ref. 13 for review).
In contrast to directed mutagenesis (2,14,15), random screening for chloroplast rbcL mutations in the green alga Chlamydomonas reinhardtii, followed by genetic selection, has identified a number of complementing large subunit substitutions that influence ⍀ (16 -19). Because these substitutions are most often found in the secondary structure elements that comprise the cores of the ␣/␤-barrel or amino-terminal domains their influence on ⍀ is not readily deduced from the existing x-ray crystal structures (11,12,20). In particular, one temperature-conditional mutant strain, named 68-4PP, results from an L290F substitution at the end of ␤-strand 5 at the bottom of the ␣/␤ barrel (21,22). This substitution reduces ⍀ by 13%, and the mutant enzyme has decreased thermal stability at 35°C in vivo and in vitro (22,23). Genetic selection previously identified a nuclear suppressor mutation, named S52-2B, that restores the thermal stability and ⍀ value of the 68-4PP enzyme back to wild-type levels (23,24). The S52-2B mutation does not reside in either of the two rbcS genes (24), and its gene product affects Rubisco at a posttranslational step (23,25). Otherwise, the molecular basis for the S52-2B mutation is not yet known.
We reasoned that if suppressor mutations could arise in other nuclear genes or elsewhere in the 68-4PP rbcL gene, analysis of such suppressors might help to understand the mode of action of the 68-4PP L290F substitution and, perhaps, the means by which S52-2B acts as a suppressor. Therefore, additional photosynthesis-competent revertants were selected from mutant 68-4PP and analyzed. Two of these new revertants arose from rbcL second-site mutations that change residues at the interface between large and small subunits at the bottom of the ␣/␤ barrel.

EXPERIMENTAL PROCEDURES
Strains and Culture Conditions-C. reinhardtii wild-type 2137 mt ϩ (26), mutant 68-4PP mtϩ (21,22), and revertant strains are maintained at 25°C in darkness with 10 mM acetate medium containing 1.5% Bacto-agar (Difco) (26). The temperature-conditional 68-4PP mutant dies on minimal medium in the light at 35°C but survives on minimal medium at 25°C in the light or on acetate medium at either 25 or 35°C in the dark (21). This temperature-conditional photosynthesis deficiency results from an L290F substitution in the Rubisco large subunit (22,23). For biochemical analysis, cells were grown on a rotary shaker with 500 ml of liquid acetate medium in darkness.
Revertant Selection and Genetic Analysis-Independent clones of mutant 68-4PP were used for reversion experiments to ensure the genetic independence of the revertants (27). Dark-grown mutant cells were plated on solid minimal medium at a density of 2 ϫ 10 6 cells/ 100-mm Petri plate at 35°C with a light intensity of 68 microeinsteins m Ϫ2 s Ϫ1 (21,27). In several reversion experiments, cells were treated with 5-fluoro-2Ј-deoxyuridine and/or methyl methanesulfonate prior to selection to increase the recovery of chloroplast mutations (26,27). Crosses were performed as described previously (21,26), and temperature-conditional, acetate-requiring progeny were scored by replicaplating dark-grown (25°C) tetrads to minimal medium in the light at 35°C.
Gene Cloning and Sequencing-Isolation of total DNA, amplification of rbcL by the polymerase chain reaction, subsequent gene cloning of the amplified products, and DNA sequencing were performed as described previously (18). Several independently cloned rbcL genes were partially sequenced to confirm the assignment of mutations, and at least one complete rbcL gene from each revertant was sequenced to ensure that only the expected mutations were present.
Biochemical Analysis-About 2 ϫ 10 9 cells were harvested by centrifugation and sonicated in 1 mM dithiothreitol, 10 mM MgCl 2 , 10 mM NaHCO 3 , and 50 mM N,N-bis(2-hydroxyethyl)glycine, pH 8.0, at 0°C for 3 min. Protein was quantified by the method of Bradford (28). The cell extracts were fractionated by SDS-polyacrylamide gel electrophoresis (29), and Western analysis was performed as described previously (25). Rubisco holoenzyme was purified from cell extracts by sucrose gradient centrifugation (30). RuBP carboxylase activity was measured as the incorporation of acid-stable 14 C from NaH 14 CO 3 . ⍀ was determined by the simultaneous measurement of the carboxylase and oxygenase activities of purified Rubisco (20 g/reaction) with 200 M [1-3 H]RuBP (7.2 Ci/mol) and 2 mM NaH 14 CO 3 (5.0 Ci/mol) in 30-min reactions at 25°C (27,31). Synthesis and purification of [1-3 H]RuBP were performed according to the methods of Kuehn and Hsu (32). Other kinetic constants of the purified enzymes were determined as described previously (22). (24), photosynthesis-competent revertants were recovered at a frequency of 7 ϫ 10 Ϫ9 cells by directly plating rbcL mutant 68-4PP mt ϩ on minimal medium in the light at 35°C. In the present study, revertants of 68-4PP mt ϩ were recovered spontaneously (at a frequency of 5 ϫ 10 Ϫ9 ) or they were recovered after methyl methanesulfonate or 5-fluoro-2Ј-deoxyuridine/methyl methanesulfonate treatment (at a frequency of 1 ϫ 10 Ϫ8 ) as a means for increasing the number of potential suppressor mutations (17,26,30). Genetic analysis of 11 revertants revealed that two of them, named R96-4C and R96-8E, were inherited in a uniparental pattern indicative of mutations in chloroplast DNA. Both of these strains were recovered following methyl methanesulfonate treatment. When the rbcL genes from wild type, R96-4C, and R96-8E were cloned, sequenced, and compared, both of the revertants were found to arise from intragenic suppression. In addition to the original 68-4PP (L290F) mutation, the revertant R96-4C rbcL gene also contained a G to T transversion mutation that would change Val (GTA) to Leu (TTA) at large subunit residue 262. Because this mutation eliminated an RsaI restriction site, the assignment of the mutation was further verified by restriction analysis of revertant R96-4C DNA. DNA sequencing of the R96-8E rbcL gene revealed a G to A transition mutation that would change Ala (GCT) to Thr (ACT) at large subunit residue 222. Thus, the L290F substitution caused by the original mutation can be complemented by either a V262L or A222T substitution. Because either substitution could restore photosynthetic growth of the 68-4PP mutant at the 35°C restrictive temperature, we assumed that both sub-stitutions must affect Rubisco catalysis or stability.

Molecular Genetics of Photosynthesis-competent Revertants-In a previous study
Rubisco Holoenzyme Stability-When extracts of 35°Cgrown wild-type, mutant, and revertant cells were fractionated on sucrose gradients, the revertants were found to have a modest increase in the amount of Rubisco holoenzyme relative to that of the 68-4PP mutant (Table I). However, the revertants had only about 30% of the wild-type level of holoenzyme when grown at 35°C, and when grown at 25°C they had less holoenzyme than the 68-4PP mutant (Table I). To confirm that the levels of extractable holoenzyme represented the levels of holoenzyme in vivo, extracts were fractionated on SDS-polyacrylamide gels and analyzed by immunoblotting (data not shown). The V262L and A222T substitutions may complement the L290F mutant substitution by partially restoring Rubisco thermal stability in vivo at 35°C, but this "improvement" is associated with a decrease in holoenzyme stability at the 25°C permissive temperature (Table I).
Catalytic Efficiency-The ratio between Rubisco carboxylase activities measured at limiting CO 2 (0.53 mM NaHCO 3 ) under 100% N 2 and 100% O 2 is a function of the K c and K o kinetic constants (22,24). When these ratios were determined for purified Rubisco, wild-type and 68-4PP (L290F) mutant enzymes had values of 3.0 and 2.4, respectively. The R96-4C (L290F/V262L) and R96-8E (L290F/A222T) revertant enzymes had N 2 /O 2 ratio values of 2.2 and 2.8, respectively, indicating that these enzymes had kinetic properties different from either the wild-type or 68-4PP mutant enzyme.
The 68-4PP (L290F) mutant enzyme is known to have an ⍀ value lower than that of the wild-type enzyme (22,24). Detailed biochemical analysis of purified Rubisco confirmed that the 68-4PP enzyme has a 17% decrease in ⍀ and revealed that the revertant enzymes have ⍀ values restored to the wild-type value (Table II). The improved ⍀ values of both revertant enzymes arise from increases in V c , V c /K c , and K o /K c relative to the original 68-4PP mutant enzyme (Table II). However, neither enzyme has a V c or V c /K c value as good as that of the wild-type enzyme. With regard to the revertant R96-4C (L290F/V262L) enzyme, the improved ⍀ occurs despite an increase in K c and decrease in V c /V o relative to the values of the 68-4PP (L290F) mutant enzyme.

DISCUSSION
Mutant 68-4PP Rubisco has a decreased ⍀ (22) and reduced thermal stability in vivo and in vitro (22,23). These enzyme defects arise from an L290F substitution in the chloroplastencoded Rubisco large subunit (22,24). According to the crystal structure of spinach Rubisco (10,12), Leu-290 is the first residue of ␤-strand 5 at the bottom of the ␣/␤ barrel. As such, it is relatively far from the active site residues that coordinate with the transition-state analog carboxyarabinitol 1,5-bisphosphate (10,12). It was previously suggested (1) that the L290F substitution may disrupt the hydrophobic core of the ␣/␤ barrel by affecting a hydrogen bond network that extends from Glu-168 (at the bottom of the ␣/␤ barrel) to active-site His-327 (at the base of flexible loop 6) (10).
Because the 68-4PP (L290F) mutation gives rise to a temperature-conditional, acetate-requiring phenotype in vivo (21,22), it was possible to select for complementing mutations at the 35°C restrictive temperature. Past genetic studies with Chlamydomonas Rubisco have identified compensatory substitutions at residues that are often in van der Waals contact with the original mutant residues (16 -18). However, second-site substitutions that complement L290F were found to be relatively distant. Either a V262L or A222T substitution produces a moderate increase in the amount of L290F mutant Rubisco at 35°C (Table I) and, more significantly, restores ⍀ to the wildtype value (Table II). Because the V262L and A222T substitutions both increase the volume of the affected residues, it seems unlikely that they compensate for the increase in size caused by the original L290F mutant substitution. Furthermore, based on the x-ray crystal structure of spinach Rubisco (10,12), neither of these phylogenetically conserved residues resides in the Rubisco active site. Ala-222 (in the middle of ␣-helix 2) is more than 10 Å away from the atoms of Leu-290. Whereas Val-262 (below ␤-strand 4) appears to be close to Leu-290 (bottom of ␤-strand 5), the side chains of these two residues are oriented away from each other. Instead, Val-262 is in van der Waals contact with Ala-222 (10, 12). These "long-distance" interactions relative to residue 290 are interesting, especially considering that residues 222, 262, and 290 are close to residues of the Rubisco small subunit (10,12).
In spinach Rubisco, a small subunit hairpin loop (residues 46 -67 flanked by ␤-strands A and B) is in close contact with large subunit residues at the bottom of the ␣/␤ barrel (10). Cyanobacterial Rubisco lacks 12 residues of this loop (residues 52-63), whereas Chlamydomonas Rubisco contains 6 additional residues (see Ref. 1 for review). Because ⍀ values also diverge among these species, it is interesting to consider whether the small subunit hairpin loop might contribute to the enhanced catalytic efficiency of eukaryotic Rubisco (10,20). With regard to the crystal structure of spinach Rubisco (10,12) and as shown in Fig. 1, the L290F substitution could place residue 290 in van der Waals contact with small subunit residues Gly-60 and Tyr-62 (Leu-66 and Tyr-68 in Chlamydomonas Rubisco, respectively). Furthermore, the backbone atoms of large subunit Val-262 are likely to be in van der Waals contact with the C ␣ , C ␤ , and C ␥ atoms of the spinach small subunit Pro-59 (Cys-65 in the Chlamydomonas enzyme), which is located at the tip of the small subunit hairpin loop (10) (Fig. 1). The C ␤ atom of large subunit Ala-222 is also in van der Waals contact with one of the C ⑀ atoms of spinach small subunit Tyr-61 (Tyr-67 in the Chlamydomonas enzyme). However, this tyrosyl residue resides in a second neighboring small subunit (10) (Fig. 1). In conclusion, it seems possible that the V262L and A222T large subunit substitutions could complement the L290F substitution via interactions transmitted through the Rubisco small subunit (Fig. 1). Perhaps these interactions at the interface between large and small subunits contribute to both the structural stability (Table I) and catalytic efficiency (Table II) of Rubisco.
Directed mutagenesis of pea rbcS followed by in vitro synthesis and transport into isolated chloroplasts indicated that Arg-53 in the small subunit hairpin loop is required for holoenzyme assembly (33). In spinach Rubisco (10,12) this arginyl residue (Arg-59 in the Chlamydomonas enzyme) hydrogen bonds with large subunit Tyr-226, which is also in van der Waals contact with the side chain atoms of Ala-222 and Val-262. Such observations further support the idea that substitutions at Ala-222 and Val-262 may also influence the interactions between large and small subunits.
When a hybrid Rubisco enzyme comprised of cyanobacterial large subunits and diatom small subunits was expressed in Escherichia coli, it was found to have an ⍀ value intermediate to the ⍀ values of the native cyanobacterial and diatom holoenzymes (34). Thus, even though the hybrid enzyme had substantial decreases in V c and V c /K c (34), these results indicate that small subunits can contribute to the catalytic efficiency of Rubisco. Considering that substitutions at large subunit residues 290, 222, and 262 can influence ⍀ (Table II), perhaps the nearby small subunit residues (Fig. 1) could also play a role in determining the catalytic efficiency of Rubisco.
Most of the revertants recovered from mutant 68-4PP in this and a previous study (24) arose from nuclear suppressor mutations. One of these revertants has been analyzed in detail (23,24), but the nuclear suppressor mutation does not reside in either of the two rbcS genes (24). Nonetheless, the idea that small subunits can act as a bridge between L290F and the complementing V262L and A222T substitutions would imply  that substitutions in the small subunit may also complement the original 68-4PP (L290F) mutation. Further study of the existing nuclear suppressors will test this hypothesis. Because the L290F, A222T, and V262L substitutions can influence Rubisco ⍀ (Table II) and may do so by influencing the structure of the large subunit/small subunit interface (Fig. 1), it would be interesting to examine the effect of nearby small subunit substitutions on Rubisco catalysis. However, it has not been possible to create such substitutions via directed mutagenesis because prokaryotic Rubisco enzymes lack either the small subunit or the small subunit hairpin loop (see Ref. 1 for review), and eukaryotes contain a family of rbcS genes that precludes transformation of mutant genes (see Refs. 1 and 8 for review). Only recently has a mutant strain of Chlamydomonas been recovered that lacks the rbcS gene family (35). Because the rbcS deletion mutant can be transformed with a single rbcS gene (35), it may soon be possible to directly investigate the potential role of small subunit residues 65, 66, 67, and 68 in Rubisco catalysis.