Thylakoid localized bestrophin-like proteins are essential for the CO2 concentrating mechanism of Chlamydomonas reinhardtii

Significance Models of the CO2 concentrating mechanism (CCM) of green algae and diatoms postulate that chloroplast CO2 is generated from HCO3− brought into the acidic thylakoid lumen and converted to CO2 by specific thylakoid carbonic anhydrases. However, the identity of the transporter required for thylakoid HCO3− uptake has remained elusive. In this work, 3 bestrophin-like proteins, BST1–3, located on the thylakoid membrane have been found to be essential to the CCM of Chlamydomonas. Reduction in expression of BST1–3 markedly reduced the inorganic carbon affinity of the alga. These proteins are prime candidates to be thylakoid HCO3− transporters, a critical currently missing step of the CCM required for future engineering efforts of the Chlamydomonas CCM into plants to improve photosynthesis.

A quatic photosynthetic organisms, which account for close to 50% of the world's carbon fixation (1), face several challenges in carrying out efficient photosynthesis. Limitations include the slow diffusive rate of gases in water, fluctuations in pH, and the slow interconversion of inorganic carbon (C i ) forms. Thus, most aquatic autotrophs have developed an adaptation called the CO 2 concentrating mechanism (CCM) that increases the concentration of CO 2 around Ribulose bisphosphate carboxylase/oxygenase (Rubisco) to increase its carboxylase activity. Aside from Rubisco's slow rate of catalysis, O 2 can compete with CO 2 for the active site of the enzyme, resulting in the wasteful process of photorespiration (2). Since CO 2 and O 2 are competitive substrates, the CCM reduces photorespiration and increases photosynthetic efficiency.
The CCM of the unicellular green alga Chlamydomonas reinhardtii (hereafter referred to as Chlamydomonas) has a number of bicarbonate (HCO 3 − ) transporters that help increase the HCO 3 − concentration in the chloroplast stroma relative to the external HCO 3 − concentration. These transporters are located on the plasma membrane (LCI1 and HLA3) as well as the chloroplast envelope (NAR1.2/LCIA). Loss of any one of these transporters reduces the ability of the cell to accumulate HCO 3 − at high external pH (3,4). In addition, Rubisco is tightly packaged in a microcompartment of the chloroplast called the pyrenoid (5)(6)(7). Finally, carbonic anhydrase 3 (CAH3), located in the lumen of pyrenoid-traversing thylakoids, converts the accumulated HCO 3 − to CO 2 near the site of Rubisco (8,9), increasing photosynthetic and growth rates at otherwise growthlimiting CO 2 levels.
Carbonic anhydrases play an essential role in the Chlamydomonas CCM (10). The loss of CAH3 results in cells that cannot grow on air levels of CO 2 , even though these mutants tend to overaccumulate HCO 3 − (11). Chlamydomonas CCM models propose that mutants missing CAH3 accumulate the HCO 3 − brought into the chloroplast by the transport proteins but cannot convert that HCO 3 − to CO 2 , the actual substrate of Rubisco (12,13). These CCM models postulate that the pH gradient across the thylakoid membrane in the light helps drive the conversion of HCO 3 − to CO 2 . The apparent acid dissociation constant (pK a ) of the interconversion of HCO 3 − and CO 2 is about 6.4, with the chloroplast stoma having a pH close to 8 in the light and the thylakoid lumen having a pH close to 5.7 under low CO 2 concentrations (14). Therefore, as HCO 3 − is brought from the stroma to the thylakoid lumen, it goes from an environment favoring HCO 3 − to one favoring CO 2 . Therefore, the Significance Models of the CO 2 concentrating mechanism (CCM) of green algae and diatoms postulate that chloroplast CO 2 is generated from HCO 3 − brought into the acidic thylakoid lumen and converted to CO 2 by specific thylakoid carbonic anhydrases. However, the identity of the transporter required for thylakoid HCO 3 − uptake has remained elusive. In this work, 3 bestrophinlike proteins, BST1-3, located on the thylakoid membrane have been found to be essential to the CCM of Chlamydomonas. Reduction in expression of BST1-3 markedly reduced the inorganic carbon affinity of the alga. These proteins are prime candidates to be thylakoid HCO 3 − transporters, a critical currently missing step of the CCM required for future engineering efforts of the Chlamydomonas CCM into plants to improve photosynthesis.
acidification of the thylakoid lumen is important to the functioning of the CCM.
The CCM models also proposed the presence of a thylakoid HCO 3 − transporter that brings in HCO 3 − from the stroma to the lumen for dehydration by CAH3 (12,13). In a recent interactome study, the CCM complex LCIB/LCIC is shown to interact with the bestrophin-like proteins encoded by Cre16.g662600 and Cre16.g663450 (15). These proteins were also shown to interact with each other and another bestrophin-like protein encoded by Cre16.g663400 (15). All 3 genes were found to be up-regulated in low CO 2 conditions in a transcriptomic study showing they belonged to a cluster of genes that had increased expression in low CO 2 and were controlled by CIA5 (16). Bestrophins are typically chloride channels, including the Arabidopsis bestrophinlike protein AtVCCN1 (17). However, they have also been shown to transport a range of anions, with some showing high HCO 3 − permeability (18). The interactome study also putatively localizes these bestrophin-like proteins to the thylakoid membrane, which makes them promising candidates to be the thylakoid HCO 3 − transporter in the CCM of Chlamydomonas. In the present study, we investigate the role of these 3 proteins using an RNA interference (RNAi) approach to knock down the expression of all 3 genes. This approach was feasible as the 3 genes are extremely similar at the DNA sequence level. Knockdown mutants with low expression of all 3 genes grow poorly in limiting CO 2 conditions, exhibit a poor affinity for external C i , and have a severely reduced ability to accumulate HCO 3 − . This study sheds light on the intracellular location and function of these bestrophinlike proteins in the CCM of Chlamydomonas.

Results
Chlamydomonas Has 3 Very Similar Bestrophin-Like Proteins on the Thylakoid Membrane. BST1 (Cre16.g662600), BST2 (Cre16.g663400), and BST3 (Cre16.g663450) (collectively BST1-3) are paralogous bestrophin-like genes located within a 130-kilobase pair (kbp) region on the 16th chromosome of Chlamydomonas. Phylogenetic analyses revealed that bestrophin-like proteins are found in a diverse variety of photosynthetic organisms (Fig. 1), including vascular plants, nonvascular plants, and diatoms, with the homologs with the highest sequence identity to BST1-3 found in algae. The amino acid sequences encoded by these genes were analyzed in TMHMM, which predicted that BST1-3 are membrane proteins having 4 predicted transmembrane domains each. Further analysis using PredAlgo predicted that each BST protein had a chloroplast transit peptide and was likely to be a chloroplast membrane protein. BST1 was annotated as a bestrophin-like protein in Phytozome (version 12.1), and BST2 and BST3 were previously reported as LCI11 by Fang et al. (16). An alignment between the 3 Chlamydomonas bestrophin-like proteins showed that the proteins are >80% identical to one another (SI Appendix, Fig. S1). There are 7 more genes annotated as encoding bestrophin-like proteins in the Chlamydomonas genome, but they share less than 50% identity to BST1-3. Sequence alignment of BST1-3 with human Bestrophin 1 (BEST1) showed low sequence identity between BEST1 and BST1-3 (21 to 23%; SI Appendix, Fig. S1). The most similar protein in terrestrial plants, the thylakoid localized AtVCCN1 protein of Arabidopsis (17), has approximately a 30% sequence identity with BST1-3. To further explore the potential structure and function of BST1-3, we did homology modeling using SWISS-MODEL (19). Structural studies show that human and Klebsiella pneumoniae bestrophins are pentameric, and modeling of BST1 in a pentameric assembly is of high confidence (SI Appendix, Fig. S2A). The highest ranking template identified by SWISS-MODEL for BST1-3 was K. pneumoniae bestrophin. BST1-3 contain nonpolar residues along their selective pore that are conserved in proteins of the bestrophin family and are involved in anion transport (20) (SI Appendix, Fig. S2B). The entry pocket of BST1 has a predominantly neutral/negative electrostatic potential, and the selective pore is positively charged, supporting the hypothesis that BST1-3 transport negatively charged ions (19,21) (SI Appendix, Fig. S2 C and D), as does AtVCCN1 in Arabidopsis (17).

BST1-3 Are Up-Regulated under Low CO 2 Growth Conditions and
Localized to the Thylakoid. Semiquantitative RT-PCR ( Fig. 2A) was performed using complementary DNA isolated from strains D66 and cia5 grown under high CO 2 or ambient CO 2 conditions. For this work, we have used 5% CO 2 (vol/vol) in air as high CO 2 , 0.04% as ambient CO 2 , and <0.02% as low CO 2. D66 is the wildtype strain for these studies, and cia5 is missing the CCM1 protein, which is required for the induction of the CCM in Chlamydomonas (22). This work demonstrated that all 3 BST genes were up-regulated under ambient CO 2 growth conditions in D66 and that this up-regulation was not observed in cia5 (SI Appendix, Fig. S3A). In addition, the cia5 mutant exhibited severely reduced expression of BST1 and BST3 under both CO 2 conditions, a transcriptional pattern observed with other CCM genes. BST2 transcript levels in cia5 cells showed reduced induction in ambient CO 2 when compared with D66 cells, where BST2 transcript levels increase in ambient CO 2 conditions. A time course study of the expression of these 3 genes during induction of the CCM was done by transferring high CO 2 -grown cells to ambient CO 2 levels for 2 to 12 h ( Fig. 2B and SI Appendix, Fig. S3B) . All 3 genes had increased transcript levels within 2 h after the switch to low CO 2 , and these elevated levels of expression continued until at least 12 h after induction. BST1 had a lower level of expression than BST2 or BST3 (Fig. 2).
To determine the localization of these 3 BST-like proteins in Chlamydomonas, fluorescent protein fusions were constructed linking Venus to the C terminus of each BST protein. All 3 BSTlike proteins localized to the thylakoid membranes of the chloroplast (Fig. 3A), and this localization visibly extended into the thylakoid tubules of the pyrenoid (Fig. 3B)  studies visually showed that BST1, BST2, and BST3 were preferentially concentrated near the pyrenoid (Fig. 3B). To confirm that expression using the constitutive PSAD promoter was not affecting localization or pyrenoid periphery enrichment, we constructed a BST3-Venus line with the BST3 gene under its native promoter. This line showed the same localization pattern as BST3 under the constitutive PSAD promoter (Fig. 3C), and quantification of enrichment showed a 1.46-fold enrichment (P < 0.01, Student's paired t test) around the pyrenoid relative to the rest of the chloroplast. Thus, BST1, BST2, and BST3 are thylakoid localized anion transporters enriched at the pyrenoid periphery that are expressed coordinately with the expression of other Chlamydomonas CCM proteins.

Reduction of BST1-3 Expression Results in Cells that Grow Slowly
under Low CO 2 Conditions. A BST3 knockout (bst3) was obtained from the Chlamydomonas Library Project (CLiP) mutant collection (23) with a paromomycin insert in the last exon of the bst3 gene (SI Appendix, Fig. S4A). The BST3 transcript was not detected in bst3 (SI Appendix, Fig. S4B), and the BST3 protein was absent (SI Appendix, Fig. S4C). We observed a weak growth difference for this strain as compared with wild-type cells under ambient CO 2 (SI Appendix, Fig. S5 A and B), but no clear phenotype on plates at pH 7 or pH 8.4 at 100 μmol of photons per m −2 ·s −1 at low CO 2 (SI Appendix, Fig. S5C). However, there was no significant difference in C i affinity between wild type and bst3 grown at ambient CO 2 (SI Appendix, Fig. S5D), and C i uptake by bst3 was only slightly lower than wild type (SI Appendix, Fig. S5 E and F). This led us to think that BST1 or BST2 function might be redundant with BST3 and that the expression of all 3 genes must be reduced to determine their physiological role(s). Therefore, to elucidate the function of BST1-3, RNAi constructs complementary to regions of identity among BST1-3 were designed (SI Appendix, Table S1). The D66 strain was transformed with these constructs, and colonies were kept at high CO 2 . Colonies were then screened for growth on high CO 2 versus low CO 2 , and BST1-3 expression was quantified using RT-qPCR. Three independent colonies from 2 different transformations were chosen for further study and designated as bsti-1, bsti-2, and bsti-3 (BST RNAi triple-knockdown lines 1, 2, and 3).
The growth of bsti-1, bsti-2, and bsti-3 on high and low CO 2 was compared with D66 and the CAH3 knockout mutant, cia3 (Fig.  4A). In low CO 2 , bsti-1 showed severely reduced growth that was further exacerbated at high pH, resembling the growth of cia3 (Fig. 4A). The bsti-2 and bsti-3 also grew more slowly than wildtype cells, but better than bsti-1. However, at high CO 2 , the growth of all 3 strains was comparable to wild type. RT-qPCR showed that bsti-1 had significantly reduced expression of BST1, BST2, and BST3 compared with D66 (Fig. 4B), and bsti-2 and bsti-3 had a more moderate knockdown of expression of the 3 genes. To see if reduced transcript levels resulted in decreased protein abundance, we checked BST3 protein levels in the knockdown lines. All showed reduced levels relative to D66, although this was only significant for bsti-2 and bsti-3 (P < 0.05, Student's t test; SI Appendix, Fig. S6A). Thus, the BSTs are required for wild-type-like growth of Chlamydomonas under low CO 2 conditions.

Reduction of BST1-3 Expression Also Results in Cells that Have a
Reduced Capacity to Accumulate Inorganic Carbon. Two characteristics of algal cells with a CCM are a very high affinity for C i and the ability to accumulate C i to levels higher than can be obtained by diffusion. The bsti-1, bsti-2, and bsti-3 acclimated to ambient CO 2 exhibited a lower affinity for C i as judged by their measured C i concentration needed for half-maximum oxygen evolution [K 1/2 (C i )] (Fig. 5). When grown at high CO 2 , bsti1-3 and D66 exhibited similar C i affinities (SI Appendix, Fig. S6B). These results indicate that the expression of BST1-3 is required for optimal C i affinity when cells are grown on ambient levels of CO 2 . At pH 8.4, the K 1/2 (C i ) values for bsti1-3 are elevated in sharp contrast to a low K 1/2 (C i ) for D66 (Fig. 5 A and B). At the higher pH of 8.4, the predominant C i species in the medium would be HCO 3 − . Thus, the increased affinity of the cells for C i reflects their ability to actively take up and utilize HCO 3 − . For bsti-1, where the expression of all 3 BST genes is between 60 and 90% reduced, there is a reduced C i affinity at both pH 8.4 (Fig. 5 A and B) and pH 7.8 ( Fig. 5 C and D). In contrast, bst3, the mutant missing only BST3, the difference in C i affinity with wild type (SI Appendix, Fig. S2B) is much smaller. Thus, we can conclude that BST1-3 are necessary components of the CCM of Chlamydomonas.
C i uptake activity was measured in D66, bsti-1, bsti-2, and bsti-3 to evaluate the importance of BST1-3 in accumulation and  fixation of C i . Ambient CO 2 -acclimated bsti-1 had a notably lower accumulation and fixation of 14 C i compared with D66 at pH 8.4 (Fig. 6), and bsti-2 and bsti-3 also had inhibited 14 C uptake and fixation, although not as reduced as bsti-1 (Fig. 6). The most severely affected mutant, bsti-1, accumulated 14 C i to only 30 to 40% of the levels observed in D66 cells. These results indicate that BST1-3 play an important role in C i uptake and fixation in low CO 2 conditions in Chlamydomonas.
A bestrophin-like protein recently discovered in Arabidopsis, AtVCCN1, is a Cl − channel that helps regulate the proton motive force (pmf) in the Arabidopsis thylakoid (17). Elimination of AtVCCN1 results in plants that have an increased pmf, altering how the plant regulates nonphotochemical quenching and the ΔpH across the thylakoid membrane. It is possible that the reduction of these BST proteins in Chlamydomonas could render cells less able to regulate the membrane potential (Δψ) and ΔpH components of the pmf, leading to photodamage or to an adenosine 5′-triphosphate (ATP)/NADPH imbalance. To investigate if BST1-3, in addition to being critical for C i affinity and accumulation, have a role in regulating pmf similar to AtVCCN1, we measured electrochromic shift to estimate the pmf in the knockdown lines under HCO 3 − -depleted conditions (SI Appendix, Fig. S7). We found a small reduction of the pmf in the bsti mutants (SI Appendix, Fig. S7  A and B), which is opposite to what is seen in Arabidopsis. In addition, the pmf decayed slightly faster in the bsti-1 and bsti-3 mutants than in wild type (SI Appendix, Fig. S7C). We also measured the yield of variable chlorophyll a fluorescence to estimate photosystem II function in the mutants and found that F v /F m was the same in mutant and wild-type cells (SI Appendix, Fig. S7D). The fact that the bsti1-3 mutants grew normally at relatively high light levels (Fig. 4A) indicates that reducing BST1-3 does not cause severe photodamage.
In conclusion, the localization, the C i affinity, and the C i accumulation phenotypes of the bsti triple-knockdown mutants support an essential role for BST1-3 in the CCM.

Discussion
We present evidence here that BST1-3 are chloroplast thylakoid localized anion transporters that are important components of the Chlamydomonas CCM. Cells that have reduced BST1-3 transcript levels fail to grow on low CO 2 (Fig. 4), have a lower affinity for C i (Fig. 5), and have a reduced ability to accumulate added 14 C i (Fig. 6). A key aspect of current Chlamydomonas CCM models is that accumulated HCO 3 − is converted to CO 2 by CAH3, a carbonic anhydrase located in the thylakoid lumen (11)(12)(13). This feature of algal CCMs may extend to other algal types, notably diatoms, where Kikutani et al. (24) recently discovered a θ-type carbonic anhydrase within the thylakoid of Phaeodactylum tricornutum that was required for CCM function. These CCM models predict that a thylakoid HCO 3 − transporter is required to deliver HCO 3 − from the chloroplast stroma to the thylakoid lumen. We propose that BST1-3 are the transporters that bring HCO 3 − to CAH3 inside the thylakoid. Members of the human bestrophin family transport both HCO 3 − and Cl − ions (18). The homology modeling presented here supports the function of BST1-3 as anion transporters, with BST1-3 having predicted structural and conserved transport residue similarities to chicken and bacterial bestrophins (SI Appendix, Fig. S2).
The expression of all of the CCM transporters discovered previously is induced by ambient or lower CO 2 conditions, and their expression is controlled by the transcription factor CIA5/ CCM1 (22,25). We have observed that all 3 BST genes are induced when Chlamydomonas is grown under ambient CO 2 conditions (Fig. 2) and that this induction is absent in the cia5 mutant ( Fig. 2A). In addition, LCIB and LCIC, possible θ-carbonic anhydrases (24,26) essential to the CCM (4) that interact with BST1-3 (15), have the same expression pattern Error bars indicate SD. (16). Thus, the expression of the BST1-3 genes is consistent with these proteins playing a role in the uptake and accumulation of C i when Chlamydomonas is exposed to low CO 2 conditions. An alternative hypothesis is that the 3 BST proteins have a function similar to AtVCCN1 (20) and are involved in Cl − transport to regulate the pmf across the thylakoid. The presence of AtVCCN1 decreases pmf in Arabidopsis, but the presence of the 3 BST proteins increases pmf in Chlamydomonas. This result, in combination with our genetic and physiology data, suggests that the function of the BST proteins in Chlamydomonas is not the same as VCCN1 in Arabidopsis. A further understanding of this interconnection and the balancing/regulation of pmf within the context of the CCM is critical.
In Chlamydomonas, there seems to be a built-in redundancy of C i transporter functions. For example, both LCI1 and HLA3 are present on the plasma membrane, and loss of only one of the proteins fails to cause an extreme growth phenotype at low CO 2 (3). However, when more than 1 transporter is knocked down, a significant change in C i uptake and growth is observed (4, 5). BST1-3 also appear to have redundant or overlapping functions. This is demonstrated in this study, as knocking out BST3 by itself did not cause a drastic change in growth or reduction in C i affinity at ambient levels of CO 2 (SI Appendix, Figs. S4 and S5). However, when the expression of all 3 genes is decreased in bsti1-3, cells could not grow at low CO 2 and C i uptake was severely compromised. In addition, BST1-3 transcript levels in the RNAi strains correlated to C i affinity and C i uptake, supporting their C i transport role and functional redundancy. This redundancy likely explains why BST1-3 were not identified in earlier mutant screens, as these screens typically knock out only 1 gene at a time. The 3 BST proteins do, however, have sequence differences, particularly at their C termini. Therefore, they might have specific (or slightly different) physiological roles that cannot be differentiated under the conditions employed in this present study. Fig. 7 shows a refined model for the Chlamydomonas CCM, which now includes our proposed function for BST1-3. In this model, HLA3 and LCI1 transport HCO 3 − across the plasma membrane, bringing HCO 3 − into the cell (12,13). At the chloroplast envelope, NAR1.2 (LCIA) transports HCO 3 − into the chloroplast stroma. Then, BST1-3 on the thylakoid bring HCO 3 − into the thylakoid lumen, where CAH3 in the pyrenoidal thylakoid tubules converts HCO 3 − to CO 2 to be fixed by Rubisco. Since BST1-3 are found throughout the thylakoid, a potential futile cycle is possible. However, for the futile cycle to take place, CAH3 needs to be present and active in the thylakoid membranes away from the pyrenoid. There is published work that CAH3 is preferentially located and activated in the pyrenoid tubules under low (<0.02%) CO 2 conditions (8,9). The location of BST1-3 is also likely to be important in the recapture of CO 2 that is generated by the CCM (Fig. 7). Any CO 2 in the pyrenoid not fixed by Rubisco has the potential to simply diffuse out of the cell (4,(27)(28)(29). The LCIB/C complex is thought to help recapture this CO 2 (27) by directionally driving CO 2 to HCO 3 − or by acting as a tightly regulated carbonic anhydrase (30) at the pyrenoid periphery (28). This is interesting because there are data supporting the interaction of LCIB/C with BST1 and BST3 (15). Our model adds BST1-3 to this hypothesized recapture system (Fig. 7). Having BST1-3 throughout the thylakoid (Fig. 3) would increase the surface area for the reuptake of HCO 3 − in the stroma. As such, we propose that the recapture of C i is a 2-step process, with leaked CO 2 from the pyrenoid converted to HCO 3 − by LCIB/C and BST1-3 transporting the HCO 3 − back into the thylakoid, creating an overall cyclic recapture mechanism. Loss of either BST1-3 or LCIB/C results in cells that cannot accumulate C i to normal levels, which agrees with experimental observations.
The discovery of CCM components on the thylakoid (BST1-3) and inside the thylakoid lumen (CAH3) also indicates how light energy may be used to energize the CCM. The apparent pK a of the interconversion of HCO 3 − to CO 2 is about 6.4. The pH of the chloroplast stroma, thought to be near 8.0, is well above the pK a , while the pH of the thylakoid lumen is thought to be close to 5.7 (14), below the pK a . When HCO 3 − moves from the chloroplast stroma to the thylakoid lumen, it moves from an environment that favors HCO 3 − to one that favors CO 2 . This effectively allows the algal cells to increase the CO 2 concentration to levels higher than could be obtained by the action of carbonic anhydrase alone. Thus, a transthylakoid pH gradient is necessary for this proposed "CO 2 pump," and this pH gradient is set up by the photosystems and requires light. To date, all experimental data available indicate that light and the activity of the photosystems are required for the Chlamydomonas CCM to function. In fact, some of the earliest work in the field indicated that electron transport inhibitors and mutations that disrupt electron transport also inhibited the Chlamydomonas CCM (31,32). One potential problem with this CO 2 pump model is that it would partially reduce the pmf across the thylakoid membrane, thus reducing ATP biosynthesis. However, it should be pointed out that only a single H + would be consumed per CO 2 generated, which is the   6. C i uptake of D66 and bsti1-3. C i uptake and C i accumulation were measured in D66 and bsti1-3 using the silicone oil uptake method (Materials and Methods). Cells were grown in high CO 2 and then acclimated to ambient CO 2 for 12 h prior to the assays. Cells were harvested and depleted of endogenous C i before running the assays. A time course of intracellular C i accumulation (A) and CO 2 fixation (B) is shown for pH 8  Dashed lines indicate the proposed leakage-reducing pathway that involves recycling CO 2 by LCIB/C back to HCO 3 − . The dotted black line represents the light-driven establishment of a proton gradient across the thylakoid membrane by PSII and the cytochrome b6f complex of the photosynthetic electron transport chain (PETC). equivalent of less than one-third of an ATP per CO 2 generated. This cost is far less than the 2 additional ATPs required for C 4 photosynthesis, and C 4 photosynthesis has been shown to be energetically competitive with C 3 photosynthesis once the costs of photorespiration are considered (33). In conclusion, BST1-3 are bestrophin-like, thylakoid localized membrane proteins that are synthesized in coordination with other CCM components, and their predicted structures fit well with functionally characterized bestrophins. As such, they are excellent candidates to be the HCO 3 − transporters that not only bring HCO 3 − into the thylakoid lumen for CO 2 generation but may also play a role in C i recapture as well.

Materials and Methods
Cell Cultures, Growth, and Photosynthetic Assays. C. reinhardtii culture conditions were set according to the conditions used previously (34). The D66 strain (nit2 − , cw15, mt + ) was obtained from Rogene Schnell (University of Arkansas, Fayetteville, AR), and CMJ030 (CC-4533; cw15, mt − ) and bst3 (BST3 knockout LMJ.RY0402.089365) were obtained from the CLiP collection at the Chlamydomonas culture collection (23,35). For acclimation experiments, Tris-acetate-phosphate-grown cells were switched to minimal media and bubbled with high CO 2 (5% [vol/vol] CO 2 in air) to reach an optical density at 730 nm between 0.2 and 0.3 (∼2 to 3 × 10 6 cells per milliliter). This was followed by CCM induction when the cells were transferred to ambient CO 2 (0.04% CO 2 ) bubbling. For photosynthetic assays, cells acclimated to 5% or 0.04% CO 2 were resuspended in C i -depleted buffer at pH 7.8 or pH 8.4, and O 2 evolution was measured at different Ci concentrations. K 1/2 (C i ) was calculated as the C i concentration needed for the half-maximal rate of oxygen evolution.
Fluorescence Protein Tagging and Confocal Microscopy. The BST1-3 genes driven by the constitutive PSAD promoter were cloned as reported by Mackinder et al. (15). Briefly, the open reading frames of BST1-3 genes were PCR-amplified from genomic DNA and cloned into pLM005 with C-terminal Venus-3xFLAG and a PSAD promoter through Gibson assembly. BST3 driven by its native promoter was cloned using recombineering based on methods reported by Sarov et al. (36). Transformation of these genes into Chlamydomonas and selection of colonies are described in SI Appendix, SI Materials and Methods. Images were captured with a laser-scanning microscope (LSM880; Zeiss) equipped with an Airyscan module using a 63× objective with a 1.4 numerical aperture. Argon lasers at 514 nm and 561 nm were used for excitation of Venus and chlorophyll, respectively. Filters were set at 525 to 550 nm for the Venus emission and at 620 to 670 nm for chlorophyll emission.
Additional details of materials and methods are provided in SI Appendix, SI Materials and Methods