Substitutions at the Asp-473 latch residue of chlamydomonas ribulosebisphosphate carboxylase/oxygenase cause decreases in carboxylation efficiency and CO(2)/O(2) specificity.

The loop between alpha-helix 6 and beta-strand 6 in the alpha/beta-barrel active site of ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco, EC 4.1.1.39) plays a key role in discriminating between gaseous substrates CO(2) and O(2). Based on numerous x-ray crystal structures, loop 6 is either closed or open depending on the presence or absence, respectively, of substrate ligands. The carboxyl terminus folds over loop 6 in the closed conformation, prompting speculation that it may trigger or latch loop 6 closure. Because an x-ray crystal structure of tobacco Rubisco revealed that phosphate is located at a site in the open form that is occupied by the carboxyl group of Asp-473 in the closed form, it was proposed that Asp-473 may serve as the latch that holds the carboxyl terminus over loop 6. To assess the essentiality of Asp-473 in catalysis, we used directed mutagenesis and chloroplast transformation of the green alga Chlamydomonas reinhardtii to create D473A and D473E mutant enzymes. The D473A and D473E mutant strains can grow photoautotrophically, indicating that Asp-473 is not essential for catalysis. However, both substitutions caused 87% decreases in carboxylation catalytic efficiency (V(max)/K(m)) and approximately 16% decreases in CO(2)/O(2) specificity. If the carboxyl terminus is required for stabilizing loop 6 in the closed conformation, there must be additional residues at the carboxyl terminus/loop 6 interface that contribute to this mechanism. Considering that substitutions at residue 473 can influence CO(2)/O(2) specificity, further study of interactions between loop 6 and the carboxyl terminus may provide clues for engineering an improved Rubisco.

Like many ␣/␤-barrel-domain enzymes (1), the large subunit of ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco, 1 EC 4.1.1.30) has a loop (i.e. between ␤-strand 6 and ␣-helix 6) that folds over substrate during catalysis (reviewed in Refs. [2][3][4]. Numerous studies have indicated that loop 6 plays a major role in discriminating between CO 2 and O 2 in the competing RuBP carboxylation and oxygenation reactions of Rubisco (reviewed in Refs. 2 and 3), and Lys-334 at the apex of loop 6 interacts with the C-2 carboxyl group of the transition state analog CABP in various Rubisco x-ray crystal structures (5)(6)(7)(8)(9). However, Rubisco may be unique among ␣/␤-barrel enzymes in that the carboxyl terminus folds over loop 6 and appears to stabilize its closed conformation. This arrangement of loop 6 and the carboxyl terminus is also observed in the crystal structure of unactivated Rubisco that contains RuBP in the active site (10). In vivo, Rubisco activase is responsible for facilitating the opening of the closed structure to release this RuBP (reviewed in Refs. 2 and 11), thereby allowing spontaneous carbamylation of an active site Lys-201 and introduction of Mg 2ϩ to produce the active form of the enzyme (reviewed in Ref. 12). The change from closed to open conformation in the carboxyl-terminal, ␣/␤-barrel domain of the large subunit is accompanied by movement of the amino-terminal domain of a neighboring large subunit (reviewed in Ref. 4), which also contributes residues to the active site and may contain the primary site of interaction with Rubisco activase (13,14). In the absence of RuBP or CABP, the carboxyl terminus is disordered, and loop 6 is usually disordered or misfolded into an open conformation depending on the source or treatment of the Rubisco used for crystallization (15)(16)(17)(18)(19). An open crystal structure of tobacco Rubisco was solved in which phosphate was observed to reside at a site normally occupied by the carboxyl group of Asp-473 in the closed structure (18). This prompted speculation that Asp-473 may serve as a latch that holds the carboxyl terminus over loop 6 in the closed conformation (18).
In plant-like Rubisco enzymes, comprised of eight ϳ55-kDa large and eight ϳ16-kDa small subunits, Asp-473 is 100% conserved, but the large subunit carboxyl terminus is quite variable in length and sequence identity (Fig. 1). Some bacterial Rubisco enzymes exist as only large subunit dimers (reviewed in Ref. 20), and without electron density for the carboxyl terminus (21), it is more difficult to assign a residue comparable with Asp-473 in their divergent sequences (i.e. Rhodospirillum rubrum, Fig. 1). In previous studies, the role of the large subunit carboxyl terminus in Rubisco function was investigated by using biochemical or genetic methods that deleted or added residues (22)(23)(24)(25)(26) or by using directed mutagenesis to replace cyanobacterial carboxyl-terminal residues with those characteristic of land plants (25). Although some alterations were observed in carboxylation catalytic efficiency or holoenzyme stability (22)(23)(24)(25)(26), only the deletion of as many as 10 residues from the carboxyl terminus of the Synechococcus enzyme caused a substantial reduction in CO 2 /O 2 specificity (Fig.  1, residues 466 -475) (25). If the carboxyl terminus plays a role in stabilizing loop 6 during catalysis, and loop 6 determines the differential stabilization of the carboxylation and oxygenation transition states (27,28), one might expect that specific carboxyl-terminal residues may influence CO 2 /O 2 specificity.
However, the function of individual residues was not assessed by directed-mutagenesis substitutions in previous studies.
The green alga Chlamydomonas reinhardtii serves as a useful model organism for the study of Rubisco because photosynthesis-deficient mutants can be maintained with acetate as an alternative source of carbon and energy. Mutant strains that lack functional genes for the large (chloroplast rbcL) or small (nucleus rbcS) subunit can be complemented by transformation (reviewed in Ref. 29), and the wild-type rbcL gene can be replaced with a mutant rbcL gene regardless of whether the mutant large subunit can produce a functional holoenzyme (e.g. Ref. 30). Furthermore, because mutant genes exist in vivo, genetic selection can be used to recover second-site suppressor mutations as a means for identifying complementing structural interactions (29,31). With this in mind, and considering the evolutionary conservation and potential critical function of Asp-473, we decided to examine the essentiality of Asp-473 by directed mutagenesis and Chlamydomonas chloroplast transformation. D473A and D473E substitutions did not prevent the photosynthetic growth of the Chlamydomonas transformants, indicating that Asp-473 is not essential for Rubisco function. However, both substitutions caused substantial decreases in catalytic efficiency and CO 2 /O 2 specificity.

EXPERIMENTAL PROCEDURES
Strains and Culture Conditions-C. reinhardtii 2137 mt ϩ is the wild-type strain (32). Mutant MX3312, which lacks the rbcL gene that encodes the Rubisco large subunit, was obtained from Dr. Genhai Zhu (Maxygen, Inc., Redwood City, CA) and used as the host for chloroplast transformation. This mutant was created via chloroplast transformation of wild-type 2137 mt ϩ with the 786-bp coding region of the bacterial aadA gene (33) flanked precisely by the Chlamydomonas rbcL 5Ј-and 3Ј-untranslated regions (34). Because chloroplast transformation occurs by homologous recombination in Chlamydomonas, mutant MX3312 has a photosynthesis-deficient, acetate-requiring phenotype due to the complete loss of the rbcL coding sequence, but it is resistant to aminoglycosides such as spectinomycin and streptomycin due to the expression of aadA (35). All strains are maintained at 25°C in darkness with 10 mM acetate medium containing 1.5% Bacto-agar (32). For biochemical analysis, cells were grown with 250 -500 ml of liquid acetate medium at 25°C on a rotary shaker (220 rpm) in darkness.
Directed Mutagenesis and Chloroplast Transformation-Using a plasmid containing the Chlamydomonas rbcL gene (36), directed mutagenesis was performed using synthetic oligonucleotides and a QuikChange mutagenesis kit from Stratagene (37). To create the D473A substitution, the sequence GAC was changed to GCT, which also introduced a CviRI restriction site useful for screening. The D473E substitution was created by changing GAC to GAA. In this case, the mutant oligonucleotide also contained a silent mutation (ACA to ACT) at the codon for Thr-471 that introduced an MfeI site useful for screening. The resultant mutant plasmids, named pLS-D473A and pLS-D473E, were transformed into the chloroplast of rbcL mutant MX3312 by using a helium-driven biolistic device (38), and photosynthesis-competent transformants were selected on minimal medium in the light (80 microeinsteins/m 2 /s) by standard methods (36,39). Successive rounds of selection, single-colony isolation, and restriction enzyme analysis were performed to ensure the homoplasmicity of the D473A and D473E mutant genes (40). The mutant rbcL genes were PCR-amplified and completely sequenced to confirm that only the intended mutations were present (30,36,39). The mutant Chlamydomonas strains were named D473A and D473E.
Gel Electrophoresis and Western Analysis-Soluble cell protein was isolated from dark-grown cells as described previously (40) and quantified (41). Cell extracts were subjected to SDS-polyacrylamide gel electrophoresis with a 7.5-15% polyacrylamide gradient in the running gel (42). After electrophoresis, proteins were either stained with Coomassie Blue or transferred to nitrocellulose membrane, probed with rabbit anti-tobacco Rubisco immunoglobulin G (0.5 g/ml), and detected by enhanced chemiluminescence (Amersham Biosciences) (43).
RuBP carboxylase and oxygenase activities were measured via the incorporation of acid-stable 14 C from NaH 14 CO 3 (45). The ratio of carboxylation to oxygenation at any given concentrations of CO 2 (47,49). K m (RuBP) was determined by adding 1-5 g of activated Rubisco to 1-ml reaction mix-  To measure the binding affinity for the transition state analog CABP (27,30,39), purified Rubisco (200 g) was incubated at 25°C for 2.5 h in 50 mM Bicine (pH 8.0), 10 mM NaHCO 3 , 10 mM MgCl 2 , 1 mM dithiothreitol, and 50 M [ 14 C]CABP (50 Ci/mol). The mixture was loaded on a Sephadex G-75 column (1 ϫ 30 cm) that had been equilibrated at 25°C with 50 mM Bicine (pH 8.0), 10 mM NaHCO 3 , 10 mM MgCl 2 , and 50 mM KCl. The column was eluted with the same buffer, and 1-ml fractions were collected and assayed for protein and 14 C dpm (39). The peak protein fraction was mixed with a 1,000-fold molar excess of unlabeled CABP. After incubation at 25°C for 38 h, the mixture was fractionated again. CABP and [ 14 C]CABP were synthesized by standard methods (50,51).

Directed Mutagenesis, Transformation, and Mutant Pheno-
types-To test the functional significance of Asp-473, directed mutagenesis was used to eliminate its charged side group by replacement with Ala or to alter its size without affecting charge by replacement with Glu. When rbcL mutant plasmids pLS-D473A and pLS-D473E were transformed into the chloroplast of the rbcL deletion mutant MX3312, photosynthesiscompetent transformants were recovered on minimal medium in the light at frequencies similar to that of transformation with the wild-type rbcL gene (36,39). Furthermore, phenotypic analysis of a number of independent transformants via spot tests (32) at 25 or 35°C on minimal medium or acetate medium (in the light or in the darkness) revealed no obvious differences in growth relative to that of wild type. Such analysis has previously proved useful for identifying rbcL mutants with various defects in Rubisco catalysis, assembly, or stability (reviewed in Ref. 29). Thus, despite the critical role proposed for Asp-473 in Rubisco catalysis, substitution with Ala or Glu caused no major alterations in holoenzyme function or stability in vivo.
Holoenzyme Levels and Thermal Stability in Vitro-When extracts of mutants D473A and D473E were fractionated on sucrose gradients, no significant differences in the amount of Rubisco holoenzyme were observed relative to that of wild type (data not shown). SDS-polyacrylamide gel electrophoresis and Western analysis also showed that the mutants have wild-type levels of Rubisco subunits (Fig. 2). To further confirm that the D473A and D473E mutants do not affect the structural stability of Rubisco, thermal inactivation experiments were performed (36,44). As shown in Fig. 3, the D473A and D473E mutant enzymes are not different from wild-type Rubisco with respect to thermal stability in vitro. Thus, removing the Asp-473 side group or increasing its size has no obvious effect on Rubisco holoenzyme stability in vivo or in vitro.
Catalytic Properties-The purified D473A and D473E mutant enzymes were found to have 17 and 14% decreases in ⍀, respectively (Table I). Despite ϳ2-fold beneficial increases in K o /K c , the 52-59% decreases in V c /V o and 87% decreases in carboxylation catalytic efficiency (V c /K c ) are ultimately responsible for the decreased ⍀ values (Table I). The D473A and D473E mutant enzymes also have decreases in the binding affinity for CABP (Table II), which is anticipated from the decreased carboxylation V c /K c and ⍀ values, the latter of which is indicative of a change in the relative stabilities of the carboxylation and oxygenation transition states (27,28). However, K m (RuBP) is increased ϳ2-fold for the mutant enzymes (Table I). This indicates that the conversion of RuBP to the 2,3-enediol(ate), a step common to both carboxylation and oxygenation, is also altered by the mutant substitutions.

DISCUSSION
Rubisco catalyzes the rate-determining step of photosynthesis, but it has a carboxylation k cat of only a few per second and is competitively inhibited by O 2 . Oxygenation of RuBP generates phosphoglycolate, which is the first intermediate in the nonessential photorespiratory pathway that leads to the loss of CO 2 . These properties make Rubisco an obvious target for genetic engineering aimed at improving agricultural productivity (reviewed in Refs. 2 and 3). However, despite the variation in the V c and ⍀ values of Rubisco enzymes from divergent species (52), Rubisco catalysis depends on a conserved set of active site residues (reviewed in Refs. 2,4,12). Thus, attempts to design a better Rubisco will likely depend on our depth of understanding of the structure-function relationships of the enzyme some distance from the active site. With 16 subunits, ϳ4,800 amino acids, and ϳ76,000 atoms arranged in strikingly similar x-ray crystal structures (reviewed in Ref. 4), it is a daunting task to identify regions of structure far from the active site that may influence catalysis. Only by genetic screening and selection in vivo (29), random or scanning mutagenesis (53)(54)(55), bioinformatic analysis of sequence divergence (56), and/or detailed comparative analysis of divergent x-ray crystal structures (18) may it be possible to identify regions worthy of experimental analysis. Based on an extensive comparison of Rubisco crystal structures, Duff et al. (18) proposed that Asp-473 in the carboxyl-terminal region of the large subunit may serve as a critical latch residue that accounts for the folding of the carboxyl terminus over loop 6 during catalysis. Because the large subunit carboxyl terminus is one of the most variable regions of Rubisco structure with respect to both length and sequence identity (4, 26) (Fig. 1), Asp-473 might serve as a critical connection by which the variation in the carboxyl terminus could influence loop 6 catalysis.
Although we had intended to exploit chloroplast transforma-   tion and genetic selection in Chlamydomonas to identify critical residues that could compensate for the loss of Asp-473, it was surprising to find that D473A and D473E substitutions did not eliminate the ability of Chlamydomonas to grow photoautotrophically. This indicated that Asp-473 is not essential for catalysis, and further analysis showed that neither substitution altered Rubisco stability in vivo (Fig. 2) or in vitro (Fig. 3). However, both mutant enzymes displayed 87% decreases in carboxylation catalytic efficiency and ϳ16% decreases in CO 2 /O 2 specificity (Table I, V c /K c and ⍀). In a previous study that relied on the expression of Synechococcus Rubisco in Escherichia coli (25), deletion of the last 8 carboxyl-terminal residues (which includes Asp-473, Fig. 1) caused a 99% decrease in V c /K c but did not change ⍀. This indicates that the decreases in ⍀ resulting from the D473A and D473E substitutions in the Chlamydomonas enzyme may be due to secondary alterations in structure rather than simply the loss of the Asp side chain. Both substitutions, by either lengthening or eliminating a charged side chain, may prevent a close association of the carboxyl terminus with loop 6, thereby introducing unfavorable interactions between other residues (Fig. 4). Because the biochemical properties of the D473A and D473E enzymes are quite similar (Table I), one might also assume that they cause similar structural alterations. The idea that other interactions between the carboxyl terminus and loop 6 may play a significant role in Rubisco function is borne out by the previous study with the Synechococcus enzyme (25). Deletion of two additional residues, Lys-466 and Phe-467 (Fig. 1), nearly eliminated carboxylation and caused a 38% decrease in ⍀ (25). Thus, the conserved Phe-467 residue, which shields Lys-334 from solvent (Fig. 4), also plays an important role in the interactions between the carboxyl terminus and loop 6. In the 1.4-Å x-ray crystal structure of Chlamydomonas Rubisco, numerous interactions can be observed between the carboxyl terminus and loop 6 in the closed form of the enzyme (Fig. 4). Because Asp-473 is in van der Waals contact with Glu-338 and Val-341, and its amide nitrogen forms a hydrogen bond with the side group of Glu-336, alterations in the interactions with these loop 6 residues may be sufficient to account for the observed decreases in ⍀ that arise from the D473A and D473E substitutions. However, changing Chlamydomonas Val-341 to the Ile residue common to land plant Rubisco had little or no effect on catalysis (39). Asp-473 is also in van der Waals contact with Asp-302, Arg-303, and His-310 and forms an ionic bond with Arg-134 (Fig. 4). All of these residues are conserved, and all but Arg-134 are in van der Waals contact with loop 6 residues. Thus, changes in Asp-473 might also be transmitted to loop 6 via alterations in such intervening residues. In a previous study with the Synechococcus enzyme (57), substitutions at Lys-128 also caused decreases in ⍀. The ⑀-amino group of Lys-128 forms hydrogen bonds with the carbonyl groups of both carboxyl-terminal Phe-467 and loop 6 Val-331 (Fig. 4), and a V331A substitution in Chlamydomonas Rubisco was shown previously to cause a 37% decrease in ⍀ (58). Based on these observations, it seems possible that a number of carboxylterminal residues might influence Rubisco CO 2 /O 2 specificity via interactions with residues that also interact with loop 6.
We are particularly interested in the value of ⍀ because it is a measure of the differential stabilization of the carboxylation and oxygenation transition states at the rate-determining step of both reactions (27,28). Thus, any substitutions that influence ⍀ may be potential targets for genetic engineering aimed at improving Rubisco. As may be expected from the decreased ⍀ values, the D473E and D473A mutant enzymes do not bind the carboxylation transition state analog CABP as tightly as does the wild-type enzyme (Table II). However, the mutant enzymes also have 2-fold increases in K m (RuBP) ( Table I), and a similar increase was observed for the Synechococcus carboxyl-terminal deletion mutant that caused a decrease in ⍀ (25). Although loop 6 is not essential for the enolization of RuBP in the first partial reaction common to both carboxylation and oxygenation (59), these observations agree with the structural analysis of substrate binding (18), indicating that loop 6 and the carboxyl terminus are folded in a closed state prior to the addition of CO 2 or O 2 in the second, rate-limiting step of carboxylation or oxygenation. This may raise a question as to the path by which CO 2 and O 2 enter the active site. Further study of the interactions between the carboxyl terminus and loop 6 may define the precise role of this structural arrangement in catalysis and may also provide useful information leading to the design of an improved Rubisco enzyme.