Directed mutagenesis of chloroplast ribulose-1,5-bisphosphate carboxylase/oxygenase. Loop 6 substitutions complement for structural stability but decrease catalytic efficiency.

The structure of active-site loop 6 plays a role in determining the CO2/O2 specificity of chloroplast ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco, EC 4.1.1.39). Rubisco from the green alga Chlamydomonas reinhardtii differs from higher plant Rubisco within the loop 6 region, and the C. reinhardtii enzyme has a CO2/O2 specificity 25% lower than that of higher plant enzymes. To examine whether differences in sequence may account for differences in catalytic efficiency, we focused on a conserved pair of residues that are in van der Waals contact at the base of loop 6. C. reinhardtii Rubisco contains Leu-326 and Met-349, whereas higher plant enzymes contain Ile-326 and Leu-349. By employing in vitro mutagenesis and chloroplast transformation, L326I and M349L substitutions were created within the Rubisco large subunit of C. reinhardtii. M349L had little effect, but L326I destabilized the holoenzyme in vivo and in vitro. When present together, the M349L substitution partially alleviated the instability resulting from the L326I substitution, but caused a 21% decrease in CO2/O2 specificity and a 74% decrease in the Vmax of carboxylation. Interactions between loop 6 and other structural regions are likely to be responsible for both holoenzyme stability and catalytic efficiency in higher plant Rubisco enzymes.

bound CO 2 and O 2 for enol-RuBP (4). ⍀ is ultimately determined by the differential stabilization of the carboxylation and oxygenation transition states for these latter partial reactions (5,6). Carboxylation produces 3-phosphoglycerate, which enters the photosynthetic carbon reduction cycle and forms net carbohydrate. Oxygenation produces one molecule of 3-phosphoglycerate and one molecule of 2-phosphoglycolate. Phosphoglycolate enters the nonessential photorespiratory pathway and leads to the loss of CO 2 . If one could engineer Rubisco to increase carboxylation or decrease oxygenation, an increase in net photosynthetic CO 2 fixation would be expected (see Ref. 1).
Most prokaryotic and all eukaryotic Rubisco holoenzymes are composed of eight copies of two different subunits (see Refs. 1 and 2, for review). In higher plants and green algae, the 55-kDa large subunit is encoded by the rbcL chloroplast gene, whereas the 15-kDa small subunit is encoded by a family of rbcS nuclear genes. X-ray crystallography has revealed that the large subunit contains a carboxyl-terminal ␣/␤-barrel domain (7,8). The active site is formed between the ␣/␤-barrel domain of one large subunit and the amino-terminal domain of a second large subunit (7,9). The function of the Rubisco small subunits remains unknown.
Genetic screening and selection in the green alga Chlamydomonas reinhardtii have shown that complementing substitutions in ␣/␤-barrel loop 6 can influence ⍀ (10,11). Directed mutagenesis of prokaryotic Rubisco enzymes expressed in Escherichia coli has further confirmed that ⍀ can be altered by several loop 6 amino acid substitutions (12)(13)(14)(15)(16). A highly conserved Lys-334 residue within loop 6 is known to be essential for catalysis (17)(18)(19). It hydrogen bonds with the 2-carboxy group of the transition state analog CABP (7,8). Loop 6 substitutions that influence the differential stabilization of the carboxylation and oxygenation transition states (and change ⍀ (5, 6)), probably do so by altering the structure of loop 6 and changing the position of Lys-334 within the active site (10,11,18). Comparisons of Rubisco x-ray crystal structures with and without bound CABP have now promoted the view that loop 6 is a flexible flap that provides induced fit for catalysis (7,8). It is interesting to consider whether further genetic engineering of loop 6 might substantially improve Rubisco catalytic efficiency.
Although chloroplast genes are difficult to manipulate by standard genetic methods (see Ref. 1), it is now possible to routinely perform directed mutagenesis and chloroplast transformation with the rbcL gene of C. reinhardtii (20). The C. reinhardtii large subunit has ϳ87% sequence identity with the Rubisco large subunits of spinach and maize (which have 89% sequence identity with each other) (21)(22)(23), but the ⍀ value for C. reinhardtii Rubisco (⍀ Ϸ 60) is substantially lower than the ⍀ values of the higher plant enzymes (⍀ Ϸ 80) (24). Rubisco from C. reinhardtii differs from higher plant Rubisco at three residues within the loop 6 region (Fig. 1). In particular, a pair of residues (Leu-326 and Met-349) that are likely to be in van der Waals contact at the base of the loop are replaced by a different pair (Ile-326 and Leu-349) in higher plant Rubisco enzymes (Fig. 2). To learn more about the significance of this conserved pair of residues with respect to Rubisco catalytic efficiency, L326I and M349L substitutions were created separately and together.

EXPERIMENTAL PROCEDURES
Strains and Culture Conditions-Wild-type C. reinhardtii 2137 mt ϩ (31) and rbcL insertion-mutant 25B1 mt ϩ (32) were used as hosts for chloroplast gene transformation. The 25B1 strain contains a 480-base pair yeast DNA insertion at a PstI site within the 3Ј-region of rbcL (21), which eliminates Rubisco holoenzyme and produces a photosynthesisdeficient acetate-requiring phenotype (32). All strains are maintained in darkness at 25°C with 10 mM acetate medium containing 1.5% Bacto-agar (31). For biochemical analysis, cells were grown on a rotary shaker with 50 to 500 ml of liquid acetate medium.
Chloroplast Transformation-The mutant rbcL plasmids were transformed into the chloroplast by employing a helium-driven biolistic device (34) as described previously (20). Because the phenotypes conferred by the mutant rbcL genes could not be predicted, transformants were either selected based upon photosynthetic ability in 25B1 mt ϩ , or they were recovered by screening for a photosynthesis-deficient, acetaterequiring phenotype in 2137 mt ϩ (20,31). Successive rounds of singlecolony isolation were performed to ensure homoplasmicity of the mutant genes (20). Total DNA was then purified from potential transformants (11), and a 1917-base pair fragment (bases Ϫ161 to 1756) containing the entire rbcL gene (bases 1-1428) was amplified by the polymerase chain reaction (35). A Sau3A subfragment (bases Ϫ33 to 1703), containing the complete rbcL gene, was cloned into pUC19 (36) and transformed into E. coli SMD (U. S. Biochemical Corp.). At least six independent E. coli transformants were screened for the mutant rbcL genes by plasmid isolation, restriction enzyme analysis, and DNA sequencing (20). The expected mutations were found in all cases. A representative gene from each mutant was then sequenced in its entirety to confirm that only the intended mutations were present. The rbcL mutant strains created by directed mutagenesis and transformation were named L326I, V341I, M349L, and L326I/M349L.
Biochemical Analysis-Rubisco holoenzyme was purified from cell extracts by sucrose-gradient centrifugation (37), and quantified by the dye-binding method (38). RuBP carboxylase activity was measured as the incorporation of acid-stable 14 C from NaH 14 CO 3 . ⍀ of purified and activated Rubisco (30 g/reaction) was determined by assaying carboxylase and oxygenase activities simultaneously with [1-3 H]RuBP (8.6 Ci/mol) and NaH 14 CO 3 (5.0 Ci/mol) in 1-h reactions at 25°C (39,40). [1-3 H]RuBP was synthesized and purified according to the methods of Kuehn and Hsu (41). Other kinetic constants were determined as described previously (42). Binding affinity of the carboxylation transition state analog CABP was determined by incubating fully activated, sucrose gradient-purified Rubisco (600 g) with 0.3 mM [ 14 C]CABP (50 Ci/mol) at 25°C for 2 h (20,45). The unbound [ 14 C]CABP was then separated from Rubisco by desalting on an Econo-Pac 10DG (Bio-Gel P-6) column (Bio-Rad). After adding a 1,500-fold molar excess of unlabeled CABP, aliquots of the mixture were refractionated on the column after various times at 25°C to determine the rate of [ 14 C]CABP exchange. CABP and [ 14 C]CABP were synthesized from RuBP and KCN (or K 14 CN) according to standard methods (43,44).

RESULTS
Mutant Phenotypes-C. reinhardtii Rubisco differs from higher plant Rubisco at three residues within the loop 6 primary structure (Fig. 1). Residue 341 is either isoleucine or methionine in higher plant Rubisco, but C. reinhardtii Rubisco contains valine. In addition, a conserved pair of residues differs at the base of the loop (Fig. 2). Higher plant Rubisco enzymes contain Ile-326 and Leu-349 whereas C. reinhardtii Rubisco contains Leu-326 and Met-349. By employing directed mutagenesis and chloroplast transformation, each of the three C. reinhardtii residues was changed to the corresponding residue commonly found in higher plants. These mutant strains (L326I, V341I, and M349L), as well as the L326I/M349L double mutant, were recovered by transforming the 25B1 rbcL insertion mutant and selecting for photosynthetic ability on minimal medium in the light. Thus, it was readily apparent that the mutant Rubisco enzymes must maintain a substantial level of activity in vivo. Mutants V341I and M349L are indistinguishable from wild-type when grown on minimal medium in the light, but mutants L326I and L326I/M349L grow somewhat slower. Because all of the mutants can be maintained under photosynthetic conditions, no essential function can be attributed to any of the manipulated residues.
Rubisco Holoenzyme Stability-When equal amounts of crude cell extracts were fractionated on sucrose gradients, mutant V341I was found to have a wild-type level of Rubisco holoenzyme (data not shown). In contrast, mutant M349L had about 75% of the wild-type level of holoenzyme, and mutant L326I had only 15% of the normal level (Fig. 3). Because  (25), Hordeum vulgare (26), Medicago sativa (27), Oryza sativa (28), the eukaryotic marine alga Cylindrotheca (29), and the cyanobacterium A. nidulans (30). Residues are numbered according to the structure of spinach Rubisco (7). residues 326 and 349 are in close proximity to each other (Fig.  2), and these residues differ as a pair between C. reinhardtii and higher plant Rubisco (Fig. 1), we reasoned that residues 326 and 349 might complement for structural stability or catalysis. In fact, the L326I/M349L double mutant was found to have about 50% of the wild-type level of Rubisco holoenzyme (Fig. 3), indicating that the M349L substitution can partially alleviate the reduction in holoenzyme caused by the L326I substitution. Because unassembled Rubisco subunits are rapidly degraded in vivo (45,46), we performed Western analysis (46) and confirmed that the reduced levels of holoenzymes observed on sucrose gradients (Fig. 3) accurately represented the levels of holoenzymes in vivo (data not shown).
To see whether the reduction in holoenzyme levels might be related to holoenzyme instability, thermal inactivation experiments were performed in vitro. As shown in Fig. 4, when purified enzymes were preincubated for 20 min at elevated temperatures, mutant L326I Rubisco retained much less RuBP carboxylase activity than did the wild-type enzyme. For example, mutant L326I Rubisco lost more than 75% of its initial activity after 20 min at 55°C, but wild-type Rubisco lost only 10% of its initial activity. Although mutant M349L Rubisco was indistinguishable from wild-type Rubisco with regard to thermal inactivation (data not shown), mutant L326I/M349L Rubisco also lost more activity than the wild-type enzyme. Nonetheless, the double-mutant enzyme retained more activity than the L326I mutant Rubisco at most temperatures. For example, mutant L326I/M349L Rubisco lost only 45% of its initial activity after preincubation for 20 min at 55°C (Fig. 4). Thus, the M349L substitution also alleviates the in vitro thermal instability caused by the L326I substitution.
Catalytic Efficiency-Biochemical analysis of the single-mutant enzymes revealed that all had near normal kinetic constants, with no major differences in the values for ⍀ (Table I). Thus, the only obvious significance that can be attributed to any of these individual residues would pertain to holoenzyme stability. The L326I substitution causes a major decrease in the amount of Rubisco (Fig. 3) without substantially affecting catalytic efficiency (Table I). At the opposite end of a spectrum of Rubisco stability-function relationships (42,46,47), other C. reinhardtii loop 6 substitutions do not affect Rubisco assembly or stability in vivo, but they do decrease catalytic efficiency (10,11). Although the L326I/M349L double-mutant enzyme displays an improvement in Rubisco stability with respect to the L326I enzyme in vivo (Fig. 3) and in vitro (Fig. 4), this improve-ment occurs at a significant cost to catalytic efficiency. As shown in Table I, L326I/M349L Rubisco was found to have a 21% decrease in ⍀ relative to the wild-type enzyme, primarily due to a 74% decrease in V c and a 23% decrease in V c /V o . Consistent with this reduced ⍀ value, only the double-mutant enzyme failed to bind the CABP transition state analog tightly (Fig. 5). DISCUSSION Previous directed mutagenesis studies of the plant-type Rubisco from the cyanobacterium Anacystis nidulans (⍀ Ϸ 50) were aimed, in part, at determining whether changing loop 6 to the higher plant sequence (Fig. 1) would increase ⍀ to the higher plant value (12,14). D338E and S341I substitutions had little or no effect on V c or ⍀, whereas an A340E substitution decreased V c by 12% and ⍀ by 17% (12). When all four residues (338 -341; Fig. 1) were changed simultaneously to the higher plant sequences EREI or ERDI, the mutant enzymes had 8% or 40% decreases in V c , respectively, but little or no change in ⍀ (12,14). In a similar study, A. nidulans residues 339 -342 were each changed separately to a residue corresponding to a conserved sequence for chloroplast Rubisco from a variety of marine algae (e.g. Cylindrotheca, Fig. 1) (16). Although these algal enzymes have ⍀ values higher than those of higher plant enzymes (⍀ Ϸ 110), the A. nidulans mutant enzymes were either not substantially altered (A340L, S341M), or they had decreases in V c (K339P) and ⍀ (T342I, T342V) (16).
C. reinhardtii loop 6 differs from higher plant loop 6 at only one residue in the 338 -342 sequence (Fig. 1), but C. reinhardtii Rubisco has an ⍀ value (Ϸ60) about 25% lower than that of the higher plant enzymes (⍀ Ϸ 80) (24). When this Val-341 residue was "corrected" to the Ile-341 residue found in higher plants, the resultant V341I mutant Rubisco was indistinguishable from the wild-type C. reinhardtii enzyme with regard to both holoenzyme stability (data not shown) and catalytic efficiency (Table I). This result was not surprising, considering that conservative substitutions at this residue in the cyanobacterial enzyme also had little or no effect on catalysis (12,14,16). Furthermore, higher plant enzymes are not highly conserved at residue 341. They can contain either isoleucine or methionine without any obvious difference in Rubisco function (Fig. 1).
Because a number of studies have shown that Rubisco ⍀ can be influenced by residues relatively far from the active site (11,42,46,47), our attention was drawn deeper into the secondary structure core of loop 6. Both C. reinhardtii and A. nidulans Rubisco have a pair of residues at the base of loop 6 (Leu-326 and Met-349) that differs from a conserved pair of residues in the higher plant enzymes (Ile-326 and Leu-349) (Fig. 1). Because residues 326 and 349 are also likely to be in van der Waals contact (Fig. 2), it seemed reasonable to examine whether this pair of residues might play a significant role in determining the structure of loop 6 and the catalytic efficiency of Rubisco.
When the L326I and M349L substitutions were created separately and together, the mutant strains were found to retain photosynthetic ability. Kinetic analysis of L326I and M349L Rubisco further revealed that neither enzyme differs substantially from wild-type Rubisco (Table I). However, L326I causes a dramatic reduction in holoenzyme stability in vivo (Fig. 3) and in vitro (Fig. 4), and this instability can be alleviated by the additional M349L substitution (Figs. 3 and 4). Thus, these residues do, in fact, interact to maintain a stable holoenzyme. The cost of this interaction is quite severe. The L326I/M349L double-mutant enzyme has substantial decreases in both ⍀ and V c ( Table I). Regardless of the resultant structural alterations, the L326I/M349L double-mutant enzyme has a loop 6 primary structure more like that of higher-plant Rubisco, but this corrected enzyme has decreased structural stability (Figs. 3 and 4) and catalytic efficiency (Table I). There must be other residues within the higher plant enzymes that interact with Ile-326, Leu-349, or the structural regions in which they reside. These complementing residues must differ from those of the C. reinhardtii enzyme.
Directed mutagenesis of Rhodospirillum rubrum (48), A. nidulans (49), and C. reinhardtii (20) Rubisco enzymes has shown that an S379A substitution in loop 7 influences ⍀, and this residue is likely to be in contact with loop 6 residues (7, 18, 48). Because Ser-379 is conserved in all of these enzymes, it cannot directly account for the difference between C. reinhardtii and higher plant loop 6 structures. Nonetheless, other less obvious loop 7 residues might influence loop 6/loop 7 structural interactions. With respect to spinach Rubisco (7), the ␥ 2 C of Ile-326 in ␤-strand 6 contacts the ␥ 1 C of Val-374 at the beginning of ␤-strand 7. C. reinhardtii Rubisco has a methionyl residue at position 375 in ␤-strand 7, but higher plant enzymes generally contain either Leu-375 or Ile-375. Perhaps a change at this residue would complement the C. reinhardtii L326I/ M349L mutant enzyme. Residues 326 and 349 may also interact with other residues in ␣-helix 5 and ␤-strand 7 (7), but these residues are conserved between C. reinhardtii and higher plant Rubisco enzymes. Perhaps more distant residues that are less conserved could potentially influence the loop 6 interactions. Because the L326I/M349L mutant has reduced photosynthetic efficiency, it may be possible to identify such residues by employing genetic selection in vivo (10,11,46).