Functional Hybrid Rubisco Enzymes with Plant Small Subunits and Algal Large Subunits

There has been much interest in the chloroplast-encoded large subunit of ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco) as a target for engineering an increase in net CO2 fixation in photosynthesis. Improvements in the enzyme would lead to an increase in the production of food, fiber, and renewable energy. Although the large subunit contains the active site, a family of rbcS nuclear genes encodes the Rubisco small subunits, which can also influence the carboxylation catalytic efficiency and CO2/O2 specificity of the enzyme. To further define the role of the small subunit in Rubisco function, small subunits from spinach, Arabidopsis, and sunflower were assembled with algal large subunits by transformation of a Chlamydomonas reinhardtii mutant that lacks the rbcS gene family. Foreign rbcS cDNAs were successfully expressed in Chlamydomonas by fusing them to a Chlamydomonas rbcS transit peptide sequence engineered to contain rbcS introns. Although plant Rubisco generally has greater CO2/O2 specificity but a lower carboxylation Vmax than Chlamydomonas Rubisco, the hybrid enzymes have 3–11% increases in CO2/O2 specificity and retain near normal Vmax values. Thus, small subunits may make a significant contribution to the overall catalytic performance of Rubisco. Despite having normal amounts of catalytically proficient Rubisco, the hybrid mutant strains display reduced levels of photosynthetic growth and lack chloroplast pyrenoids. It appears that small subunits contain the structural elements responsible for targeting Rubisco to the algal pyrenoid, which is the site where CO2 is concentrated for optimal photosynthesis.

Ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco, 2 EC 4.1.1.39) is the key photosynthetic enzyme responsible for CO 2 fixation. In plants and green algae, eight ϳ16-kDa nuclear encoded small subunits assemble with eight ϳ55-kDa large subunits in the chloroplast to form the functional holoenzyme (reviewed in Ref. 1). The chloroplast-encoded large subunit contains the ␣/␤-barrel active site at which CO 2 and O 2 com-pete for substrate ribulose-1,5-bisphosphate (RuBP) (reviewed in Refs. [1][2][3]. Photosynthetic CO 2 fixation depends on the Rubisco V max of carboxylation (V c ) and the K m for CO 2 (K c ), but net CO 2 fixation is decreased by competitive inhibition from O 2 and the loss of CO 2 in photorespiration, which are defined by the V max of oxygenation (V o ) and the K m for O 2 (4). The ratio of the catalytic efficiencies of carboxylation (V c /K c ) to oxygenation (V o /K o ) defines the CO 2 /O 2 specificity factor (⍀) (4,5). Because ⍀ is determined by the difference between the free energies of activation for carboxylation and oxygenation at the rate-determining step of catalysis (6), there is much interest in defining the structural basis for variation in this kinetic constant. The catalytic properties of Rubisco vary among divergent species (7)(8)(9), indicating that it might be possible to engineer improvements in Rubisco function (1). However, some plant and algal species contain CO 2 -concentrating mechanisms (CCMs) (reviewed in Refs. 10 -14), which make it difficult to tell whether any Rubisco enzyme in nature is better than any other with respect to the CO 2 and O 2 concentrations (and temperatures) that the enzyme encounters in vivo (1,15). Nonetheless, various mutant substitutions have been identified that can influence ⍀ (reviewed in Refs. 1 and 3). A deeper understanding of the structure-function relationships of Rubisco may prove useful for developing genetic engineering strategies aimed at either increasing carboxylation (and photosynthetic productivity) or decreasing non-essential oxygenation (and the loss of fixed carbon in photorespiration). An increase in net CO 2 fixation would lead to an increase in agricultural productivity for the production of food, fiber, and renewable energy (1).
Considering that the nuclear encoded small subunits of Rubisco are evolving faster than the large subunits (i.e. small subunits have greater sequence diversity), it would seem quite possible that small subunits provide unique structural and functional properties to the Rubisco enzymes of diverse species (reviewed in Ref. 16). Directed mutagenesis of Rubisco from the cyanobacterium Synechococcus expressed in Escherichia coli and directed mutagenesis and genetic selection in vivo in the green alga Chlamydomonas reinhardtii have identified several residues of the small subunit that can influence ⍀ (17)(18)(19)(20)(21). When the longer loop between ␤-strands A and B of the Chlamydomonas small subunit was replaced with the shorter loop of Synechococcus, a decrease in ⍀ was observed (15,22). This small subunit loop from the spinach enzyme, along with five substitutions in the large subunit, increased ⍀ and imparted land plant catalytic properties to the Chlamydomonas enzyme (15). Alanine-scanning mutagenesis has recently identified potential functional interactions between the most conserved small subunit residues and ␣-helix 8 of the large subunit ␣/␤-barrel (21). Perhaps there are other small subunit residues and regions that play a role in Rubisco function and that may, in part, contribute to the differences in catalytic properties observed among divergent Rubisco enzymes (15).
One other approach that may be useful for dissecting the role of the small subunit is to create hybrid enzymes composed of subunits from Rubisco enzymes that differ in their catalytic properties. When the large subunit of Synechococcus Rubisco (⍀ ϭ 41) and the small subunit of the marine diatom Cylindrotheca (⍀ ϭ 107) were expressed in E. coli, a holoenzyme was formed that had a 58% increase in the Synechococcus ⍀ value, but this beneficial increase in ⍀ occurred at the expense of a 95% decrease in V c (23). When hybrid enzymes composed of Synechococcus large subunits and the small subunits of other higher ⍀ Rubisco enzymes from prokaryotes, algae, and plants were expressed in vitro or in E. coli, they had the same or slightly lower ⍀ values relative to that of Synechococcus Rubisco (23)(24)(25)(26)(27)(28). It has been difficult to study the contribution of small subunits to the function of eukaryotic Rubisco enzymes because eukaryotic subunits cannot assemble into functional Rubisco when expressed in E. coli (29,30), and plant small subunits are coded by a family of rbcS genes in the nucleus that cannot be eliminated and replaced with foreign copies. However, pea small subunits were introduced into Arabidopsis thaliana to form a heterologous Rubisco enzyme that had a small decrease in carboxylation activity nearly equal to the percentage of foreign small subunits in the holoenzyme (31). The Rubisco large subunit of sunflower was also used to replace the chloroplast large subunit in tobacco to produce a hybrid enzyme composed of sunflower large subunits and tobacco small subunits (32). Although tobacco (⍀ ϭ 85-89) and sunflower (⍀ ϭ 98 -105) Rubisco were reported to have different ⍀ values (32,33), the hybrid enzyme was found to have an ⍀ value similar to tobacco Rubisco (32), indicating that small subunits may determine the value of ⍀. However, the hybrid enzyme tobacco plants had a substantial decrease in Rubisco V c and an increase in K c and required an elevated CO 2 concentration for growth (32). A more recent study of similar hybrid Rubisco tobacco plants, with Rubisco composed of tobacco small subunits and sunflower large subunits, found no major difference in Rubisco catalytic properties and concluded that the hybrid enzyme plants required elevated CO 2 for growth because of a deficiency in Rubisco expression (34).
Chlamydomonas has served as a useful genetic model for the study of Rubisco because host strains have been isolated that lack chloroplast rbcL genes and nuclear rbcS genes (35,36) and because Rubisco-deficient mutants can be maintained with acetate as an alternative source of carbon and energy (reviewed in Refs. 1, 5, and 16). Considering the apparent difficulty of eliminating the rbcS gene family of plants (31,37,38), it seemed reasonable to attempt investigations of hybrid Rubisco enzymes in Chlamydomonas. If photosynthesis-deficient mutant strains are recovered that express foreign small subunits, it might be possible to exploit in vivo genetic selection to identify small subunit residues or regions responsible for holoenzyme assem-bly or function (18,20). However, because Chlamydomonas rbcS genes contain introns required for small subunit expression, it was first necessary to develop a method for the expression of foreign rbcS cDNAs. This problem has now been solved by transferring the introns to the region of the rbcS gene that encodes the small subunit transit peptide. Considering that Chlamydomonas and plant small subunits share only ϳ45% sequence identity, it was remarkable to find that small subunits from spinach, Arabidopsis, and sunflower were readily recovered in functional hybrid holoenzymes in vivo. These hybrid Rubisco enzymes are expressed at wild-type levels and appear to have near normal rates of carboxylation and small increases in ⍀. Nonetheless, the hybrid enzyme mutants display reduced levels of photosynthetic growth and lack chloroplast pyrenoids, indicating that the small subunit may be responsible for targeting Rubisco to the algal pyrenoid, where CO 2 concentration is elevated.

EXPERIMENTAL PROCEDURES
Strains, Plants, and Growth Conditions-C. reinhardtii strain SS1 serves as the wild type (21). It was created by transforming the rbcS⌬-T60-3 deletion strain with a plasmid (pSS1) containing only rbcS1 (35). Cell wall-less strain rbcS⌬-T60-3 was used as the host for rbcS nuclear gene transformation. It lacks photosynthesis and requires acetate for growth due to deletion of the 13-kb locus that contains the rbcS1 and rbcS2 gene family (35). All Chlamydomonas strains are maintained at 25°C in darkness on medium containing 10 mM acetate and 1.5% Bacto agar (39). For biochemical analysis, Chlamydomonas cells were grown in 250 -500 ml of liquid acetate medium at 25°C on a rotary shaker at 120 rpm. Arabidopsis (A. thaliana cv. Columbia) and sunflower (Helianthus annuus cv. Sunspot) were grown in soil at 27°C and 65% humidity with a photoperiod (900 microeinsteins/m 2 /s) of 12 h. Arabidopsis was harvested 30 days after germination. Sunflower was harvested between 7 and 15 days after germination. Spinach (Spinacia oleracea) was purchased locally.
Construction of Plasmids-To remove the introns from rbcS1, plasmid pSS1 was used as the starting material (35) (see Fig. 1). It contains an ϳ5-kb EcoRI fragment of Chlamydomonas nuclear DNA, which contains rbcS1 (40) (supplemental Fig. S1), cloned at the EcoRI site of pUC19 (41). This EcoRI fragment apparently contains all the necessary sequence elements for rbcS1 expression (35), none of which were altered during subsequent rbcS1 engineering. Intron 3 was removed from rbcS1 by replacing the StuI fragment with the StuI fragment from plasmid pCS2.1, which contains the Chlamydomonas rbcS2 cDNA sequence (40). The rbcS1 and rbcS2 protein-coding sequences are identical in this region. This new plasmid was named pSS1-⌬I3 (see Fig. 1). Using pSS1-⌬I3 as the template, oligonucleotides with overlapping sequences flanking intron 2 were used to synthesize fragments by PCR before and after intron 2. Both fragments were combined and extended into a longer product (42), which was used to replace the rbcS1 sequence between NcoI and BlpI (see Fig. 1). The resultant plasmid was named pSS1-⌬I23. Intron 1 was deleted from pSS1-⌬I23 in a similar way, replacing the NcoI-BlpI fragment of rbcS1 (42). The result-ant plasmid, which lacks all three introns in rbcS1, was named pSS1-cDNA (see Fig. 1).
To insert introns 1 and 2 into the rbcS1 region of pSS1-cDNA that encodes the small subunit transit peptide, plasmid pSS1 was used as the template. Synthetic oligonucleotides with overlapping sequences were used to synthesize intron 1 from pSS1 and two fragments upstream and downstream of the insertion point in the encoded transit peptide region. The three fragments were combined and extended into a longer product, which was digested and cloned between the EagI and BlpI restriction sites of pSS1-cDNA (see Fig. 1). This new plasmid was named pSS1-I1TP. Intron 2 was inserted into the rbcS1encoded transit peptide region of pSS1-I1TP in a similar way by exchanging an NcoI-BlpI restriction fragment (see Fig. 1). This new plasmid was named pSS1-ITP. Sites for the insertion of introns into the encoded transit peptide region were chosen based on a comparison of splicing sites for other Chlamydomonas nuclear genes (43). The wild-type sequence and features of the Chlamydomonas EcoRI restriction fragment cloned in pSS1 (35)  Coding sequences for the mature small subunits of spinach (44), Arabidopsis (45), and sunflower (46) were synthesized to encode codons common to the Chlamydomonas rbcS1 mRNA using overlapping synthetic oligonucleotides (47). When necessary, directed mutagenesis was performed with synthetic oligonucleotides and a QuikChange kit from Stratagene (48). The final sequences are presented in supplemental Fig. S3. In each case, the plant rbcS sequence was changed to encode the Met-2 common to the Chlamydomonas mature small subunit sequence. The foreign rbcS coding region sequences along with their exact Chlamydomonas flanking sequences were digested with NcoI and BlpI and used to replace the NcoI-BlpI fragment in the rbcS1 gene of pSS1-ITP (see Fig.  1). These new plasmids were named pSS1-SSSO, pSS1-SSAT, and pSS1-SSHA (see Fig. 1).
Chlamydomonas Transformation-Transformation of cell wall-less strain rbcS⌬-T60-3 was performed by electroporation (22,49), and photosynthesis-competent colonies were selected on minimal medium (without acetate) in the light (80 microeinsteins/m 2 /s). The engineered rbcS1 genes and appropriate flanking regions were PCR-amplified and completely sequenced at the University of Nebraska DNA sequencing facility to verify gene sequence and accuracy of construction. A wild-type strain recovered by transformation with the pSS1-ITP plasmid was named SS1-ITP. The rbcS1 hybrid mutant strains were named SSSO, SSAT, and SSHA.
The kinetic constants of purified and activated Rubisco were measured by the incorporation of acid-stable 14 C from NaH 14 CO 3 (55). The value of ⍀ (V c K o /V o K c ) was determined as the ratio of the rates of carboxylase (v c ) and oxygenase (v o ) activities measured simultaneously with 88 M [1-3 H]RuBP (15.8 Ci/mol) and 2 mM NaH 14 CO 3 (0.5 Ci/mol) in 30-min reactions at 25°C according to the (56,57). The [1-3 H]RuBP and phosphoglycolate phosphatase used in the assays were synthesized/purified by standard methods (56,58).

RESULTS
Expression of Chlamydomonas rbcS cDNA-To investigate the influence of foreign small subunits on Rubisco holoenzyme function, it seemed sensible to use cDNA sequences rather than contend with potential problems that may arise from the processing of divergent introns. However, numerous attempts to transform the rbcS⌬-T60-3 deletion mutant with Chlamydomonas rbcS1 cDNA were unsuccessful. Of the three introns present in rbcS1 (Fig. 1), only intron 3 is not required for expression (22). In a previous study of Chlamydomonas gene expression (59), engineering Chlamydomonas rbcS2 intron 1 into either the flanking region or the coding region of a heterologous selectable marker gene increased transformation frequency and expression. Furthermore, when transit peptides were first discovered (60,61), it was shown that the transit peptide of the Rubisco small subunit may have species specificity for Chlamydomonas chloroplast targeting (61,62). Therefore, we decided to move rbcS1 introns 1 and 2 from the mature protein-coding region into the transit peptide-coding region (Fig. 1). In this way, the introns would still be present to facilitate gene expression, and the Chlamydomonas transit peptide would allow targeting of foreign small subunits to the Chlamydomonas chloroplast. When this rbcS1-ITP gene was created (in plasmid pSS1-ITP; Fig. 1) and transformed into the photosynthesis-deficient rbcS⌬-T60-3 strain, photosynthesis-competent colonies were recovered at the same frequency as when transformed with the wild-type rbcS1 gene (ϳ5 ϫ 10 Ϫ6 /cell). One of these transformants, named SS1-ITP, was analyzed further. As shown in Fig. 2, the SS1-ITP strain has a photosynthetic growth phenotype nearly identical to the SS1 wild-type strain, which was recovered by transformation with the wild-type rbcS1 gene (21). Furthermore, SS1-ITP expresses the same level of Rubisco holoenzyme as wild-type SS1 (Fig. 3). Thus, transferring introns 1 and 2 to the transit peptide-encoding region of rbcS1 has little if any negative effect on the expression of the Rubisco small subunit, and pSS1-ITP appears to be a suitable vector for ultimately expressing foreign rbcS cDNAs in Chlamydomonas.
Expression of Foreign Rubisco Small Subunits in Chlamydomonas-Because small subunits are coded by families of rbcS genes (16), decisions had to be made about which sequences would be used for genetic engineering. Chlamydomonas contains two rbcS genes that encode mature small subunit proteins that differ by four amino acid residues. The rbcS1 gene is routinely used for experiments solely because it was observed to give a higher frequency of transformation than rbcS2 (35). Although Chlamydomonas cells containing only one rbcS gene have about half as much Rubisco as wild-type cells (35), Rubisco enzymes containing either small subunit 1 or small subunit 2 do not differ with respect to level of expression, catalytic properties, or thermal stability (19,20,35). Only one rbcS sequence is   known for sunflower (32,46). Four rbcS genes are present in Arabidopsis that differ from each other by as many as nine amino acids in the encoded mature proteins (45). We chose rbcS1B because it is neither the most conserved nor the most divergent among the four homologs. Spinach small subunit sequences have been determined by x-ray crystallography (63,64) and direct protein sequencing (65). However, in both cases, the sequences likely represent a mixture of different small subunits. Therefore, we used a sequence deduced from a cloned spinach rbcS gene (44), which differs from the high resolution crystal structure of spinach Rubisco (63) by only seven residues near the small subunit protein termini.
The Chlamydomonas rbcS1 sequence that encodes the mature small subunit in plasmid pSS1-ITP was first replaced exactly with the codon-optimized rbcS sequence for the spinach small subunit. After numerous attempts to transform the rbcS⌬-T60-3 strain with this engineered spinach rbcS gene, only one slow growing, photosynthesis-competent colony was recovered. When Rubisco was purified from this strain and analyzed by SDS-PAGE, the small subunit was found to be about 2 kDa larger than the native spinach small subunit (data not shown). This observation was quite similar to that previously observed when small subunit transit peptide processing was blocked by directed mutant substitutions flanking the transit peptide-processing site (66). Because transit peptide processing in Chlamydomonas occurs in a two-step fashion, only 21 residues of the 45-residue transit peptide remained attached to the amino terminus of the mature small subunit sequence (66). Furthermore, transit peptide processing in Chlamydomonas may depend on residues next to those at the primary processing site (67). Taking these observations into account, we wondered whether the nature of residue 2 in the mature small subunit sequence might influence transit peptide processing. This residue is Met in the Chlamydomonas small subunit but Lys in the spinach small subunit. When directed mutagenesis was used to change the Lys residue to Met, the resultant plasmid, named pSS1-SSSO, yielded photosynthesis-competent transformant colonies at the same frequency as the wild-type pSS1-ITP plasmid. Arabidopsis and sunflower rbcS genes were synthesized to also encode Met-2 (supplemental Fig. S3). These plasmids, named pSS1-SSAT and pSS1-SSHA, also yielded photosynthesis-competent colonies at wild-type frequencies when transformed into the rbcS⌬-T60-3 strain. Growth phenotypes of the transformants were compared, and one representative strain obtained from each transforming plasmid was designated for further study. These hybrid Rubisco rbcS1 mutant strains were named SSSO, SSAT, and SSHA.
Phenotypes of Hybrid Rubisco Mutants-Although the SSSO, SSAT, and SSHA hybrid enzyme mutants grew as well as the wild-type SS1 and SS1-ITP strains in darkness (data not shown), they displayed reduced levels of photosynthetic growth on minimal medium at either 25°C or 35°C (Fig. 2). Growth is routinely checked at the restrictive temperature of 35°C because temperature-conditional mutants may be useful for selecting second site suppressor substitutions (18,20,68). When photosynthetic growth was examined at a saturating level of CO 2 (5% in air), it was somewhat surprising to find that the hybrid enzyme mutants grew as well as the wild-type SS1 and SS1-ITP strains at both 25°C and 35°C (Fig. 2). If photosynthetic growth at air levels of CO 2 (.03%) was decreased due to a decreased amount of Rubisco holoenzyme, one would not expect to see a wild-type level of growth at elevated CO 2 .
Protein Levels, Processing, and Stability-When extracts of cells grown at 25 and 35°C in darkness were subjected to SDS-PAGE and Western analysis (Fig. 3A), the SSSO, SSAT, and SSHA hybrid enzyme mutants were found to have levels of Rubisco large subunits equal to or greater than those of the wild-type SS1 and SS1-ITP strains. Small subunit bands varied in intensity because the antibody used for detection was affinity-purified with Chlamydomonas Rubisco (22), and small subunits are less conserved in sequence than are large subunits (16). Furthermore, when Rubisco was purified from 25°Cgrown mutant strains on sucrose gradients, the amount of Rubisco in the hybrid enzyme mutants per total cell protein was equal to that in the SS1 and SS1-ITP strains (data not shown).
When purified Rubisco was separated by SDS-PAGE and detected with Coomassie Blue staining (Fig. 3B), the stoichiometry of large and small subunits was found to be the same among all the samples. The observed molecular masses of the small subunits in the SSSO, SSAT, and SSHA hybrid enzyme mutants closely matched those of the small subunits in Rubisco holoenzymes purified from spinach, Arabidopsis, and sunflower, respectively (Fig.  3B). However, it is apparent that there are multiple small subunits in Arabidopsis and sunflower that differ in size or charge. To ensure that transit peptides were processed correctly for the SSSO, SSAT, and SSHA hybrid enzyme small subunits, small subunits were excised from SDS-polyacrylamide gels, digested with trypsin, and analyzed by mass spectrometry in the University of Nebraska Center for Proteomics. All of the small subunits begin with N-methyl-Met (data not shown) as expected for the mature small subunits of plants and green algae (69,70).
To determine whether the hybrid Rubisco enzymes may have an associated structural instability, thermal inactivation experiments were performed (19,54). As shown in Fig. 4, the SSSO, SSAT, and SSHA hybrid enzymes were somewhat more sensitive to thermal inactivation than wild-type SS1 Rubisco. In contrast, plant Rubisco enzymes were more thermally stable than the Chlamydomonas enzyme (Fig. 4). Nonetheless, because thermal inactivation occurs at temperatures far above growth or enzyme assay temperatures, any minor instability of the mutant holoenzymes would not be expected to influence the accuracy of catalytic constant measurements.
Enzyme Kinetics-Previous studies had indicated that sunflower Rubisco might have a higher ⍀ value than other plant Rubisco enzymes (32,33). However, when the Rubisco enzymes of spinach, Arabidopsis, and sunflower were analyzed with the highly accurate, dual labeling assay (56,57), their ⍀ values were found to be quite similar (Table 1). In fact, the ⍀ value of spinach Rubisco may be 4% greater than the ⍀ values of Arabidopsis and sunflower Rubisco. Further analysis of kinetic constants indicated that sunflower Rubisco does have a higher K o /K c value than that of spinach and Arabidopsis Rubisco (Table 1). This beneficial increase results from both an increase in K o and a decrease in K c . However, the value of ⍀ is not increased relative to the ⍀ values of spinach and Arabidopsis Rubisco because the increase in K o /K c would be offset by a decrease in V c /V o (Table 1).
When the hybrid mutant enzymes were analyzed with respect to catalytic properties, all were found to have small increases in ⍀ relative to Chlamydomonas wild-type SS1 Rubisco (Table 1). The greatest increase was observed for mutant SSAT Rubisco, which contains Arabidopsis small subunits. The ⍀ value is 11% greater than the ⍀ value of Chlamydomonas Rubisco but 12% lower than the ⍀ value of Arabidopsis Rubisco (Table 1). It is difficult to decide with confidence whether any of the other kinetic constants differ between Chlamydomonas SS1 Rubisco and the hybrid Rubisco enzymes, but small increases in K o could account for the increases in ⍀ (Table 1). With an increase in ⍀ and no substantial decrease in V c (Table 1), the hybrid enzymes may be somewhat better than the wild-type Chlamydomonas enzyme with regard to net CO 2 fixation (1, 4). However, although the SSSO, SSAT, and SSHA hybrid enzymes are no worse than wild-type SS1 Rubisco with regard to catalytic properties ( Table 1) or levels of expression (Fig. 3A), the mutant cells do not grow as well as wild-type SS1 under photosynthetic conditions at air levels of CO 2 (Fig. 2). One might consider that interactions between Rubisco and Rubisco activase have been altered in the hybrid enzyme mutants (1). However, a mutant of Chlamydomonas that lacks a functional activase gene grows as well as wild type at air levels of CO 2 (71), and the species specificity of Rubisco activase interactions has been shown to occur via the Rubisco large subunit (72,73).
Analysis of the Chloroplast Pyrenoid-We wondered whether the poor photosynthetic growth of the SSSO, SSAT, and SSHA hybrid enzyme mutants may result from a loss of the CCM (10). Chlamydomonas mutants that lack the CCM grow poorly, if at all, at air levels of CO 2 but are indistinguishable from wild type with regard to photosynthetic growth at elevated CO 2 , which is a condition when the CCM is normally repressed (74,75). The hybrid enzyme mutants have a similar CO 2 -requiring phenotype (Fig. 2). Furthermore, it was shown previously in cyanobacteria that mistargeting of carboxysomal carbonic anhydrase to the cytosol blocks or eliminates the CCM and produces a CO 2 -requiring phenotype (76,77). Because Rubisco is normally located in, and required for the formation of, the chloroplast pyrenoid in Chlamydomonas (78,79) and because the pyrenoid is an integral component of the CCM (10), it seemed possible that the hybrid Rubisco enzymes might not be correctly targeted for pyrenoid formation. Light microscopic observation of the SSSO, SSAT, and SSHA hybrid enzyme mutants indicated that they lacked the pyrenoid. This was confirmed by electron microscopy (Fig. 5). Because the probability of observing a pyrenoid depends on the plane of section, not  every wild-type cell was expected to display a pyrenoid (78). In single sections of ϳ3,000 cells of each strain observed at low magnification (ϫ1,200), pyrenoids were unambiguously identified in 66% of the wild-type SS1 cells. No pyrenoids were observed in the rbcS⌬-T60-3 strain or in any of the three hybrid enzyme mutant strains (Fig. 5). Thus, despite normal levels of catalytically proficient Rubisco ( Fig. 3 and Table 1), the loss of the pyrenoid and the CCM likely accounts for the poor photosynthetic growth of the SSSO, SSAT, and SSHA hybrid enzyme mutants in air (Fig. 2).

DISCUSSION
Our desire to express foreign rbcS cDNAs in Chlamydomonas led us to consider moving introns 1 and 2 of Chlamydomonas rbcS1 into the sequence encoding the transit peptide (59) (Fig. 1). This solved the problem of Chlamydomonas rbcS cDNA expression, and no difference was observed in the levels of Rubisco holoenzyme when either the Chlamydomonas rbcS1 cDNA (pSS1-ITP) or the genomic (pSS1) coding sequences were transformed into the rbcS⌬-T60-3 deletion strain (Figs. 2 and 3). However, similar levels of hybrid holoenzyme expression were achieved only after changing residue 2 of the spinach, Arabidopsis, and sunflower mature small subunits to the Met residue present in Chlamydomonas small subunits (Fig. 3). Based on previous studies with the Chlamydomonas small subunit (66,67), it is likely that the nature of residue 2 plays a role in efficient transit peptide processing. In the x-ray crystal structures of Chlamydomonas and spinach Rubisco (63,64,70), residue 2 is in contact with and shields large subunit residues Trp-411, Pro-415, Pro-453, and Glu-454 from solvent, but there are no notable differences between the two structures in this region. Thus, it seems unlikely that the identity of residue 2 would influence Rubisco function. The engineered rbcS1 transit peptide region containing introns 1 and 2 may be useful for expressing and targeting other cDNA-encoded proteins to the chloroplast of Chlamydomonas. However, because there are multiple transit peptide-processing peptidases in Chlamydomonas (62), additional engineering of the amino terminus of the mature protein may be necessary. Nonetheless, the idea of moving introns to transit peptide-encoding regions may serve as a common strategy for cDNA expression in Chlamydomonas.
The primary objective of our study was to investigate the influence of foreign small subunits on Rubisco function in Chlamydomonas. Previous studies had indicated that hybrid Rubisco enzymes expressed in Arabidopsis and tobacco had decreases in Rubisco carboxylation or expression (31,32,34), but because it is difficult to eliminate and replace the entire rbcS gene family in plants (37,38), combinations of subunits for analysis have been limited. When the Chlamydomonas small subunit was replaced with the small subunit of spinach, Arabidopsis, or sunflower, it was surprising to find that the hybrid Rubisco enzymes were expressed at normal levels ( Fig. 3A) and had near normal catalytic properties ( Table 1). As shown in Fig.  6, the Chlamydomonas small subunit shares only 44 -47% sequence identity with the plant small subunits, whereas plant small subunits share 71-76% identity with each other (44 -46, 80 -82). Despite this greater sequence identity, heterologous holoenzymes containing approximately one pea small subunit per Arabidopsis holoenzyme had only 85-88% of normal carboxylation activity, apparently due to the loss of one functional active site per holoenzyme (31). Arabidopsis plants containing higher levels of pea small subunits in their holoenzymes were not viable (31). Perhaps unfavorable interactions between divergent small subunits in the holoenzyme are more detrimental to function than are altered interactions between foreign small subunits and large subunits. However, hybrid holoenzymes composed entirely of tobacco small subunits and sunflower large subunits were reported to have a greater than 70% decrease in V c (32), and the hybrid enzyme tobacco plants required an elevated level of CO 2 for growth. In a more recent study, the tobacco-sunflower hybrid enzymes were reported to have normal catalytic properties when assayed in leaf extracts, and the hybrid enzyme plants appeared to require elevated CO 2 for growth because of a decrease in holoenzyme expression (34). Unresolved problems associated with the genetic engineering of Rubisco in plants may contribute to errors in the biochemical analysis of mutant enzymes. Problems of this nature related to expression of plant small subunits in Chlamydomonas have apparently been solved.
In comparison with Chlamydomonas Rubisco, the plant Rubisco enzymes have lower V c values, but higher ⍀ values due to lower K c and higher K o /K c values ( Table 1). The hybrid enzymes, composed of plant small subunits and Chlamydomonas large subunits, have small increases in ⍀ relative to wildtype Chlamydomonas Rubisco. These increases, ranging from 3 to 11%, may arise from an increase in K o , but the SSSO spinach  (44 -46, 80, 81). Residues that comprise ␣-helixes A and B are colored dark gray, and those that comprise ␤-strands A through D are colored light gray. Residues identical to those in the Chlamydomonas small subunit are denoted as dots. Residues in the Chlamydomonas small subunit that are in contact with Rubisco large subunit residues are marked with asterisks below the sequences. Residues in the Chlamydomonas small subunit with the greatest solvent accessibility (82) are marked with carets below the sequences.

Hybrid Rubisco Mutants
hybrid enzyme may also have an increase in V c /V o ( Table 1). The V c and K c values of the hybrid enzymes remain more like those of Chlamydomonas Rubisco than like plant Rubisco (Table 1). Thus, the small subunits can influence the catalytic properties of Rubisco, but the active site large subunit must also contribute to the differences in catalysis observed between the plant and algal enzymes. In a previous study, replacing the longer loop between ␤-strands A and B of the Chlamydomonas small subunit with the shorter loop from the spinach small subunit (Fig. 6) caused a 26% decrease in K c (22). However, due to decreases in V c and K o , the value of ⍀ was not altered. There must be other regions outside the small subunit ␤A-␤B loop that are responsible for the higher ⍀ values of the hybrid enzymes, and these regions or residues must be different from those of the Chlamydomonas small subunit. As shown in Fig. 6, there are a few residues clustered in the amino-terminal region and in the loop between ␤-strands C and D that are in van der Waals contact with the large subunit and that differ between Chlamydomonas and plant small subunits. Perhaps these residues are responsible for the higher ⍀ values of the hybrid enzymes. In another study with Chlamydomonas Rubisco, when the spinach ␤A-␤B loop was combined with five large subunit direct mutant substitutions, there was a 12-17% increase in ⍀, and the other catalytic constants of the mutant enzyme were indistinguishable from those of spinach Rubisco (15). The small subunit ␤A-␤B loop does, in fact, play a role in increasing the value of ⍀, but as indicated by the analysis of the hybrid enzymes, there may be additional small subunit regions that can further increase ⍀. Because several residues in the amino-terminal region of the small subunit interact with the ␤A-␤B loop (22), this region may deserve further attention for defining the role of the small subunit in Rubisco function.
The SSSO, SSAT, and SSHA hybrid enzyme mutants have wild-type levels of functional Rubisco (Fig. 3 and Table 1), but they require an increased concentration of CO 2 to achieve wildtype levels of photosynthetic growth (Figs. 2 and 3). This phenotype appears to result from the loss of the chloroplast pyrenoid (Fig. 5), which is associated with the algal CCM (10,11,78,79). The mutants grow better in air than other Chlamydomonas CCM mutants (74,75), perhaps indicating that some components of the CCM are still functional but Rubisco is in a CO 2limited environment. In a previous study, sequence homology was observed between the Synechococcus small subunit and a carboxysomal CCM protein (83), indicating that small subunits may have evolved from a structural protein of the cyanobacterial carboxysome (16). If the eukaryotic small subunit plays a primary role in targeting Rubisco for pyrenoid formation (78), residues that differ between the small subunits of plants (which do not have pyrenoids) and Chlamydomonas may be most important for this function. As shown in Fig. 6, many of the Chlamydomonas small subunit residues that show the greatest level of solvent accessibility (as analyzed with the Protein Interfaces, Surfaces and Assemblies service at the European Bioinformatics Institute (82)) differ from the residues common to plant small subunits. Further dissection of foreign small subunits in Chlamydomonas may identify regions that are essential for the targeting of Rubisco to the algal pyrenoid.