A pyrenoid-localized protein SAGA1 is necessary for Ca2+-binding protein CAS-dependent expression of nuclear genes encoding inorganic carbon transporters in Chlamydomonas reinhardtii

Microalgae induce a CO2-concentrating mechanism (CCM) to maintain photosynthetic affinity for dissolved inorganic carbon (Ci) under CO2-limiting conditions. In the model alga Chlamydomonas reinhardtii, the pyrenoid-localized Ca2+-binding protein CAS is required to express genes encoding the Ci-transporters, high-light activated 3 (HLA3), and low-CO2-inducible protein A (LCIA). To identify new factors related to the regulation or components of the CCM, we isolated CO2-requiring mutants KO-60 and KO-62. These mutants had insertions of a hygromycin-resistant cartridge in the StArch Granules Abnormal 1 (SAGA1) gene, which is necessary to maintain the number of pyrenoids and the structure of pyrenoid tubules in the chloroplast. In both KO-60 and the previously identified saga1 mutant, expression levels of 532 genes were significantly reduced. Among them, 10 CAS-dependent genes, including HLA3 and LCIA, were not expressed in the saga1 mutants. While CAS was expressed normally at the protein levels, the localization of CAS was dispersed through the chloroplast rather than in the pyrenoid, even under CO2-limiting conditions. These results suggest that SAGA1 is necessary not only for maintenance of the pyrenoid structure but also for regulation of the nuclear genes encoding Ci-transporters through CAS-dependent retrograde signaling under CO2-limiting stress.


Introduction
Photosynthetic organisms are capable of sensing and responding to various environmental stresses. Guard cells of terrestrial plants, for example, sense the concentrations of extracellular carbon dioxide (CO 2 ), the substrate for photosynthesis, and control the opening and closing of stomata to take it in (Hashimoto et al. 2006). In aquatic environments, CO 2 exists as inorganic carbon (Ci; CO 2 and HCO 3 -), and the diffusion rate of CO 2 is substantially slower than in the atmosphere. To maintain efficient photosynthetic activity under such CO 2 -limiting conditions, most microalgae induce a CO 2 -concentrating mechanism (CCM) by actively taking up extracellular Ci and increasing the CO 2 concentration around the CO 2 -fixation enzyme ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco) (Fukuzawa et al. 2012).
Chlamydomonas can adapt to at least two CO 2 -limiting conditions (Vance and Spalding 2005;Yamano et al. 2022): very low-CO 2 (VLC; < 7 μM CO 2 ) and low-CO 2 (LC; 7-70 μM CO 2 ). Under VLC conditions, active HCO 3 uptake is driven by high-light activated 3 (HLA3) (Duanmu et al. 2009), an ABC-type transporter located in the plasma membrane, low-CO 2 -inducible protein A (LCIA), an anion channel located in the chloroplast envelope, and bestrophin-like proteins (BST1-3) located in the thylakoid membranes. Insertional mutants of HLA3 and LCIA show a reduced affinity for Ci, especially under alkaline conditions where HCO 3 is the predominant form of Ci (Yamano et al. 2015), and RNAi mutants with reduced expression of BST1-3 are unable to grow in CO 2 -limiting conditions, exhibit a reduced affinity for Ci, and show reduced Ci-uptake (Mukherjee et al. 2019). Moreover, a mutant of low-CO 2 -inducible protein 1 (LCI1) shows reduced Ci-affinity in acidic conditions, where CO 2 is the predominant form of Ci, suggesting that LCI1 is 1 3 involved in CO 2 uptake (Kono and Spalding 2020). LCI1 has also been reported to localize to the plasma membrane (Ohnishi et al. 2010) and to interact with HLA3 ), but the physiological significance of the interaction remains unclear.
In Chlamydomonas, a pyrenoid, a liquid-liquid phaseseparated organelle in the chloroplast, plays an essential role in photosynthetic CO 2 fixation (Freeman Rosenzweig et al. 2017). The pyrenoid matrix aggregated with Rubisco is surrounded by a starch sheath and penetrated by thylakoid membranes called pyrenoid tubules (Engel et al. 2015). Recently, two Chlamydomonas proteins required for pyrenoid formation have been identified. One is Essential Pyrenoid Component 1 (EPYC1), the Rubisco-binding protein that localizes in the pyrenoid matrix . Considering that EPYC1 is required for a development of pyrenoid structure and that Rubisco and EPYC1 are both necessary and sufficient to phase separate and form liquid droplets in vitro, EPYC1 is shown to function as a linker protein of Rubisco to form a phase-separated pyrenoid (Wunder et al. 2018). The other is StArch Granules Abnormal 1 (SAGA1), the Rubisco-and starch-binding protein that localizes between the starch sheaths and the pyrenoid matrix (Itakura et al. 2019). The saga1 mutant forms multiple pyrenoids which lack pyrenoid tubules, which causes a decrease in photosynthetic Ci-affinity. Therefore, SAGA1 could be a linker between the starch sheath and Rubisco that functions to maintain pyrenoid morphology. Moreover, based on the analysis of the Rubisco-binding motif derived from SAGA1, a common motif that is necessary and sufficient for targeting the pyrenoid was revealed and novel pyrenoid proteins such as SAGA2, RBMP1, RBMP2, and CSP41A have been identified (Meyer et al. 2020).
Although the genes related to the pyrenoid formation tend to be constitutively expressed, the expression of genes that encode Ci-transporters and several carbonic anhydrases (CAs) are induced under CO 2 -limiting conditions by CCM1/ CIA5, a master regulator of the CCM (Fukuzawa et al. 2001;Miura et al. 2004;Fang et al. 2012). A recent study revealed that the basic leucine zipper transcription factor 8 (BLZ8) is involved in the expression of HLA3, CAH7, and CAH8 in response to oxidative stress conditions (Choi et al. 2022); however, the genetic relationship between CCM1/CIA5 and BLZ8 during CCM induction is not known. Additionally, the mutant strain H82 which lacks the Ca 2+ -binding protein CAS demonstrates reduced growth, Ci-affinity, and more than four-fold lower transcriptional levels for 13 genes (Wang et al. 2016), including HCO 3 transporters (HLA3 and LCIA), mitochondrial-localized CAs (CAH4 and CAH5), mitochondrial membrane transporters (CCP1 and CCP2), chaperone proteins (DNJ31), proteins involved in non-photochemical quenching (LHCSR3.1 and LHCSR3.2), phosphoprotein phosphatase 2C-related (PPP30), and unknown proteins (LCID, Cre12.g541550, and Cre26.g756747). These genes transiently induce mRNA levels at 0.3 h after LC induction in H82, but their mRNA levels decrease at 2 h (Wang et al. 2016). This suggests that CAS may be required to maintain mRNA levels after the initial induction of CCM1/CIA5-dependent genes. Considering that CAS is dispersed throughout the chloroplast under high-CO 2 (HC; > 70 μM CO 2 ) conditions but migrates into the pyrenoid through the pyrenoid tubules under CO 2 -limiting conditions (Yamano et al. 2018), the CAS-mediated retrograde signal from pyrenoid to nucleus could play an important role in regulating the CCM.
Another essential CCM component is LC-inducible protein B (LCIB). LCIB has a structure like β-type CA (Jin et al. 2016) and is dispersed throughout the chloroplast under HC conditions but localizes around the starch sheath of the pyrenoid under VLC conditions (Yamano et al. 2010. The starch sheath is required for LCIB localization around the pyrenoid and for maintaining the Ci-affinity (Toyokawa et al. 2020). These findings suggest that LCIB has a role in converting CO 2 leaking from the gaps in the starch sheath to HCO 3 − under VLC conditions. Thus, it is becoming clear that the pyrenoid structure is not simply a site of CO 2 fixation but is necessary to the precise localization of CAS and LCIB and to efficiently transport Ci while maintaining relatively complex structures such as starch sheaths and pyrenoid tubules. However, the global impact of pyrenoid morphological abnormalities on gene expression has remained unclear. It is also important to know whether the reduced photosynthetic affinities for Ci of saga1 mutants were caused solely by structural disruption of the pyrenoid. In this study, we focused on the subcellular relocation of the CAS, which regulates nuclear genes encoding Ci-transporters involved in the CCM. It is interesting to know whether SAGA1-dependent pyrenoid structure contributes to the localization of CAS or not. We provide the physiological and gene expression data using comparative RNA-sequencing analysis of the SAGA1 allele mutants isolated from a random insertional mutant library and discuss the impact of pyrenoid morphological changes on nuclear gene expression.

Algal strains and cultural conditions
Chlamydomonas reinhardtii strains C9-1 and C9-3 share the same origin of the wild-type (WT) strain C9 originally provided from IAM Culture Collection at the University of Tokyo as described previously (Tsuji et al. 2022). In this study, C9-3 was used as the WT strain and is described as C9, which is available from Chlamydomonas Resource Center as CC-5098. Strains CC-5325, CC-5420 (saga1), and CC-5422 (SAGA1-Venus/saga1) were obtained from the Chlamydomonas resource center. For physiological and biochemical experiments, cells were precultured in TAP medium as described previously (Wang et al. 2016). Then, cells were resuspended in 50 ml of modified HSM supplemented with MOPS and grown in air containing 5% (v/v) CO 2 with illumination at the intensity of 120 μmol photons·m -2 ·s -1 until cultures reached mid-log phase as described previously (Wang et al. 2016). For induction to VLC condition, HC-acclimated cells were centrifuged, resuspended in 50 ml of fresh HSM medium, and further cultured with bubbling of ordinary air containing 0.04% (v/v) CO 2 at 120 μmol photons·m -2 ·s -1 for 2 or 12 h.

Transformation of cells
A DNA cassette containing the hygromycin (hyg)-resistant gene aphVII driven by the β2-tubulin promoter was amplified by PCR from plasmid pHyg3 (Berthold et al. 2002) using PrimeSTAR GXL DNA Polymerase (Takara Bio). The PCR product was purified using a PCR purification kit (QIA-GEN), and the concentration was adjusted to 100 ng μl −1 . The gene cassette was used to transform C9 cells by electroporation using a NEPA-21 electroporator (NEPAGENE), as described previously (Yamano et al. 2013). The transformants were incubated at 25 °C for 24 h with gentle agitation under illumination of 1.5 μmol photons m −2 ·s −1 and then screened on TAP plates containing 30 μg ml −1 hyg at 25 °C under illumination of 80 μmol photons m −2 ·s −1 . After 4 days, colonies of transformants appeared and were subjected to the screening process of HC-requiring mutants.

Screening of HC-requiring mutants
Cells were replicated into 96-well microtiter plates containing liquid TAP medium. For the first screening, cells were spotted on agar plates at pH 6.2 and 8.4 and grown in 0.04% [v/v] CO 2 chambers for 4 days. Transformants that did not grow under 0.04% CO 2 conditions were selected as candidates for HC-requiring mutants. For the second screening, the candidate mutants grown in liquid TAP medium were diluted to an optical density at 700 nm (OD 700 ) of 0.15, 0.07, or 0.03, and 3 μl of each cell suspension was spotted onto agar plates at pH 6.2, 7.0, 7.8, and 9.0 and grown for 4 days in different chambers supplied with air containing 5% [v/v] CO 2 , 0.04% CO 2 , and 0.01%, respectively.

Generation of ccm1 mutant by the CRISPR-Cas9 method
For CRISPR-Cas9-mediated genome editing, the guide RNAs was designed using CRISPOR (Concordet and Haeussler 2018) and shown in the supplemental Fig. S1a. Introduction of ribonucleoprotein complex and the aphVII cassette into cells were performed according to Tsuji et al. (2022). After transformation cells were incubated at 25 °C for 24 h in dim light and then screened on TAP plates containing 30 μg ml −1 hyg at 25 °C under 80 μmol photons m −2 ·s −1 of light for 7 days.

RNA-Seq analysis
Total RNA was extracted from cells cultured in LC conditions for 2 h using a RNeasy Plant Mini Kit (QIAGEN) according to the manufacturer's protocol. After RNA purification, total RNA was analyzed by GENEWIZ (https:// www. genew iz. com/) using Illumina Novaseq. The resulting reads were aligned to version 5.5 of the Chlamydomonas reinhardtii genome annotation (downloaded from https:// phyto zome-next. jgi. doe. gov/). Alignments were performed using the HISAT2 program. Counts of the number of reads mapping to each gene were calculated by HTSeq. The resulting expression counts were normalized with the trimmed mean of the M-values method as implemented in the edgeR package. Differentially expressed genes with FDR < 0.01 between strains were plotted as volcano plots.

Indirect immunofluorescence assay
Cells cultured in VLC conditions were collected by centrifugation and rinsed with phosphate-buffered saline (PBS) buffer twice. Immunofluorescence staining by purified antibodies against CAS and LCIB was performed as described previously (Wang et al. 2016). The primary antibodies against CAS and LCIB were used at dilutions of 1:500 and 1:200, respectively. Alexa Fluor 488 goat anti-rabbit IgG (Life Technologies) was used at a 1:500 dilution.

Measurement of CO 2 concentration in the culture medium
Cell suspensions were centrifuged to remove cells from the supernatant. Subsequently, aliquots of the supernatant were directly injected, and total Ci concentrations were measured using a gas chromatograph GC-8A (Shimadzu) with a methanizer MTN-1 (Shimadzu) as described previously (Toyokawa et al. 2020). CO 2 concentrations were also calculated as described previously (Toyokawa et al. 2020).

Isolation of CO 2 -requiring mutants KO-60 and KO-62
To obtain mutants defective in the induction of the CCM, we constructed a mutant library and isolated CO 2 -requiring mutants, such as the cas mutant H82, which could not grow under CO 2 -limiting conditions. First, we constructed a random insertional mutant library by transforming a wild-type (WT) strain, C9, with an aphVII cassette conferring hygresistance, yielding ~ 72,000 transformants. Next, because the mutant H82 showed a significantly reduced growth rate under LC conditions with air containing 0.04% (v/v) CO 2 (Wang et al. 2016), we evaluated the growth rate of the mutants on agar plates under LC conditions as a first screening and selected two strains, KO-60 and KO-62.
To evaluate the induction of the CCM under LC conditions, the photosynthetic affinity for Ci was evaluated by measuring the rates of photosynthetic oxygen (O 2 ) evolution. The K 0.5 (Ci) values, the Ci concentrations required for half of the maximal rate of O 2 -evolving activity (V max ), in KO-60 (554 ± 130 µM) and KO-62 (471 ± 40 μM) cells were Fig. 1 The growth and O 2 -evolving activity of HC-requiring mutants (a). Cells were diluted to the indicated optical density (OD 700 = 0.15, 0.07, or 0.03), 3 μl of the cell suspensions were spotted on agar plates (pH 6.2, 7.0, 7.8, and 9.0) and incubated for 4 days under 5% [v/v] CO 2 , 0.04% [v/v] CO 2 , or 0.01% [v/v] CO 2 conditions with continuous light at 120 μmol photons m −2 s −1 . b K 0.5 (Ci) values and maximum rates of O 2 -evolving activity (V max ) of C9, KO-60, and KO-62 cells grown in LC conditions for 12 h at pH 6.2, 7.0, 7.8, and 9.0. The K 0.5 (Ci) values were calculated as the Ci concentration required for half-maximum oxygen-evolving activity. Data from all experiments show mean values ± SD from three biological replicates. *P < 0.05 by Student's t-test 4.4 and 3.7 times higher than that of C9 cells (126 ± 34 µM) at pH 9.0, respectively (Fig. 1b). Even at pH 6.2, 7.0, and 7.8, the K 0.5 (Ci) values of KO-60 and KO-62 were over 1.5 times higher than that of C9 cells. The K 0.5 (Ci) values of H82 cells were also higher than those of C9 cells at all tested pH conditions (Wang et al. 2016). At pH 6.2, 7.0, and 7.8, the V max of KO-60 and KO-62 were 69-84% of that of C9 (Fig. 1c). At pH 9.0, the V max of KO-60 and KO-62 were 47-61% of that of C9. These results showed that both KO-60 and KO-62 had reduced Ci-affinity and V max . On the other hand, the V max of H82 cells is not significantly different from C9 at any pH condition (Wang et al. 2016). These results showed that KO-60 and KO-62 had reduced Ci-affinity regardless of pH conditions.
After shifting conditions from HC to LC, C9 induced the accumulation of HLA3, LCIA, LCI1, LCIB, and CAH1, while ccm1-2 showed little or markedly reduced accumulation. On the other hand, KO-60 and KO-62 showed little accumulation of HLA3, LCIA, and LCI1 as in the case of H82, even though CAS was accumulated at the same level as in C9 (Fig. 2a).
To determine the insertion site of the aphVII cassette in KO-60 and KO-62, thermal asymmetric interlaced (TAIL)-PCR was performed. In KO-60 and KO-62, the aphVII cassette was inserted in the 10th and 15th intron of the SAGA1 gene, respectively (Supplemental Fig. S1d and e). The growth retardation, increased K 0.5 values and decreased V max observed in KO-60 and KO-62 grown under CO 2 -limiting condition at pH 7.0 and 7.8 were similar in the original saga1 mutant at pH 7.3 (Itakura et al. 2019). To confirm that SAGA1 is defective in KO-60 and KO-62, we examined the number of pyrenoids in C9, KO-60, and KO-62 cells by introducing a plasmid expressing Venus-tagged Rubisco small subunit 1 (RBCS1-Venus). Although a single fluorescent signal was observed in the C9 background, multiple fluorescent signals were observed in KO-60 and KO-62 background cells (Supplemental Fig. S1f), as in the case of the original saga1 mutant (Itakura et al. 2019), indicating that KO-60 and KO-62 were loss-of-function mutants of SAGA1.
Next, to further investigate whether the reduced accumulation of Ci-transporters was due to the SAGA1 disruption, we also examined the accumulation of the above CCMrelated proteins in the original saga1 mutant (Itakura et al. 2019). As in the case of KO-60 and KO-62, HLA3, LCIA, and LCI1 were not accumulated in saga1 compared to the parental strain, CC-5325. Furthermore, the complemented . Cells were first grown under HC conditions for 24 h and shifted to HC or LC conditions for 12 h. Histone H3 was used as a loading control. Asterisks indicate the non-specific protein bands strain (SAGA1-Venus/saga1) recovered the accumulation of these Ci-transporters (Fig. 2b), suggesting that changes in pyrenoid morphology due to the disruption of SAGA1 are responsible for the reduced accumulation of Ci-transporters. The accumulation level of CAH1 was reduced in saga1 compared to CC-5325 and was not fully recovered in SAGA1-Venus/saga1. The accumulation level of CAH1 is not reduced in KO-60 and KO-62, suggesting that this contradictory phenotype is due to mutations to different genes in saga1 or to differences in parental strains, not to SAGA1 mutation.

Decrease in the mRNA levels of CCM1-dependent genes in the saga1 mutants
To examine the effect of pyrenoid morphological changes caused by the SAGA1 mutation on nuclear gene expression, C9, KO-60, CC-5325, saga1, and complementary strain cells were cultured while bubbling with ordinary air containing 0.04% (v/v) CO 2 for 2 h, and the respective transcriptome profiles were compared by RNA-seq analyses. Comparison of the transcriptomes of C9 and KO-60, and CC-5325 and saga1, showed that the expression levels of 2288 and 3633 genes, respectively, were significantly reduced (FDR [false discovery rate] < 0.01; Figs. 3 and S2a). Of the genes down-regulated as a result of SAGA1 mutations in KO-60 and saga1, 532 were recovered in the saga1 complemented strain (Supplemental Dataset S1).
To assess whether changes in pyrenoid morphology in saga1 mutants affect the expression of CCM-related genes, we compared the 532 genes down-regulated in KO-60 and saga1 but recovered in the complemented strain with those that are induced under CO 2 -limiting conditions and regulated by CCM1/CIA5 (Fang et al. 2012). The expression levels of 41 CCM1-dependent genes were decreased in KO-60 and saga1 (Table 1). In particular, among the 13 genes regulated by CAS (Wang et al. 2016), the expression levels of 10 genes (HLA3, LCIA, LHCSR3.1, LHCSR3.2, CAH4, CAH5, CCP1, CCP2, LCID, and DNJ31) decreased in both KO-60 and saga1 and recovered in the saga1 complemented strain (Fig. 3). The expression of LCI1 was not significantly reduced in KO-60 but was reduced in saga1 and recovered in the complemented strain. So, the LCI1 protein may be degraded after transcription in KO-60. It was not concluded whether changes in pyrenoid morphology by SAGA1 mutations affect the gene expression of LCI1. These results indicate that the changes in pyrenoid morphology caused by the SAGA1 mutation affect the expression of most CAS-dependent genes.
In addition to CAS-dependent genes, CCM1-regulated LCIE and DNJ15 were also downregulated more than fourfold. LCIE is a homolog of LCIB (Yamano et al. 2010) and DNJ15 has a domain that acts as a chaperone that may interact with LCIB and LCIC ).
Next, focusing on genes with upregulated expression in the saga1 mutants, we compared the transcriptomes of C9 and KO-60, and CC-5325 and saga1, and found that the expression levels of 2352 and 3725 genes were significantly induced in KO-60 and saga1, respectively (Fig. S2b). Of the genes upregulated in the saga1 mutants KO-60 and saga1, 570 were recovered in the saga1 complemented strain (Supplemental Dataset S2). When comparing those 570 genes scriptomes of KO-60 with C9 and of saga1 with CC-5325 and SAGA1-Venus/saga1. The X-and Y-axes represent log 2 (FC) andlog 10 (FDR), respectively. The blue and red dots represent downregulated and upregulated DEGs with FDR < 0.01. The gray dots represent no significant difference in transcriptomes with CCM1/CIA5-dependent genes, the expression levels of STA2, BST1, and BST2 were increased in KO-60 and saga1 (Table 2). STA2 is localized in the starch sheath , and pyrenoid morphology is altered in the sta2 mutant (Delrue et al. 1992). BST1 and BST2 are anion channels localized in the thylakoid membrane close to the pyrenoid (Mukherjee et al. 2019).

CAS and LCIB were dispersed throughout the chloroplast in the saga1 mutants
Given that CAS-dependent gene expression was impaired in the saga1 mutants and that CAS re-localizes into the pyrenoid upon CCM induction, abnormal pyrenoid formation may affect CAS localization. To test this hypothesis, we examined CAS localization in the saga1 mutants using an indirect immunofluorescence assay. Previous studies showed that CAS dispersed throughout the chloroplast stroma moves to the pyrenoid along the pyrenoid tubules under CO 2 -limiting conditions (Yamano et al. 2018). In C9, CC-5325, and the complemented strain, CAS was localized inside the pyrenoid as a distinct wheel-like structure, whereas in KO-60 and saga1, CAS was dispersed in the chloroplast (Figs. 4a and S3).
To assess the effect of abnormal pyrenoid formation on another pyrenoid peripheral protein, we also examined the localization of LCIB. Because the localization changes of LCIB are strictly controlled by CO 2 concentration with ∼7 µM as the boundary , the CO 2 concentration in the medium was calculated with the observation of LCIB localization. Although LCIB was localized around the pyrenoid in C9, CC-5325, and the complemented strain during aeration with air containing 0.04% (v/v) CO 2 for 12 h, it was dispersed in the chloroplast in KO-60 and saga1 under VLC conditions (Figs. 4b and S4). On the other hand, "blob-like" structures were observed in the chloroplasts of b,S2 and S3). These structures were also found in C9 (C9-3) cells, which were used as the parental strain for the mutant library (Supplemental Fig. S5). It is suggested that C9-3 had acquired particular characteristics during long-term subculturing such as faster chlorosis upon Table 2 Genes encoding pyrenoid-localized proteins upregulated in KO-60 and saga1 in LC condition Genes encoding pyrenoid-localized proteins upregulated in KO-60 and saga1 compared with those in C9, CC-5325, and SAGA1-Venus/saga1 from RNA-seq analysis with FDR < 0.01 in LC conditions for 2 h are shown  saga1, and SAGA1-Venus/saga1 was assessed using an indirect immunofluorescence assay with an anti-CAS and anti-LCIB antibody. Cells were grown for 2 h (a) or 12 h (b), aerating with air containing 0.04% (v/v) CO 2 . Scale bars = 5 µm N-deficiency (Tsuji et al. 2022). Original C9 (C9-1) cells freshly recovered from cryopreservation does not have these structures (Supplemental Fig. S5), suggesting that they are the characteristics of parental C9-3. Since KO-60 exhibited blob-like structures without immunofluorescence, these structures may obscure the residual accumulation of CAS and LCIB. However, CAS and LCIB were also dispersed in saga1 without these structures, suggesting that morphological changes in pyrenoids affect the localization of CAS and LCIB. These results suggest that the abnormal pyrenoid morphology caused by SAGA1 mutations impairs the localization of CAS under CO 2 -limiting conditions, resulting in defective CAS-dependent expression of CCM-related genes and a reduced Ci-affinity.

Discussion
In this study, we examined a broader range of phenotypes than previously thought to be caused by mutations in SAGA1, highlighting the importance of SAGA1 in the algal CCM. In WT cells, HLA3, LCI1, and LCIA take up Ci into the chloroplast stroma, and LCIB around the pyrenoid converts leaked CO 2 to HCO 3 − under VLC conditions. In addition, CAS migrates into the pyrenoid to induce the expression of CCM-related genes (Fig. 5a). However, in the saga1 mutants, not only is CO 2 recapture inhibited because LCIB is dispersed in the chloroplast, but the expression of CCM-related genes is impaired because CAS is also dispersed in the chloroplast (Fig. 5b). These results suggest that the reduced photosynthetic Ci-affinity exhibited by the saga1 mutants is not only due to abnormalities in pyrenoid formation but also to complex effects, including reduced expression of CCM-related genes and aberrant localization of LCIB and CAS.

Decreased accumulation of Ci-transporter and dispersion of LCIB in saga1 mutants contributes to the reduced Ci-affinity
Although the saga1 mutants showed multiple pyrenoids, elongated starch sheaths, and decreased pyrenoid tubules (Itakura et al. 2019), it was not clear how these various phenotypes related to reduced photosynthetic activity. Firstly, we showed that the accumulation levels of Ci-transporters, including HLA3, LCIA, and LCI1, were decreased in all saga1 allelic mutants (Fig. 5b). hla3 and lcia double mutants had reduced Ci-affinity, especially at pH 9.0 (Yamano et al. 2015), and the lci1 mutant had reduced Ci-affinity under pH 6.0 and 7.3 conditions (Kono and Spalding 2020). These phenotypes are one of the pieces of evidence that strongly suggest that HLA3/ LCIA and LCI1 are involved in HCO 3 and CO 2 uptake, respectively. The saga1 mutants showed two-times higher K 0.5 (Ci) values than the hla3 and lcia double mutants, suggesting that the pyrenoid abnormality in saga1 mutants may affect expression of CCM-related genes in addition to these two Ci-transporter genes. By RNA-seq analysis, we revealed that the expression levels of CAs CAH4 and CAH5, mitochondrial membrane proteins CCP1 and CCP2, and Ci-transporter genes were decreased in the saga1 mutants. Mutant strains of CAH4 and CAH5 have reduced Ci-affinity under aerated conditions Fig. 5 Model of Ci-uptake in wild type and saga1 mutant. a In wildtype, Ci-transporters and carbonic anhydrase work to concentrate inorganic carbon in the pyrenoid. The putative retrograde signal from CAS to the nucleus required for the expression of HLA3 and LCIA is shown as a red arrow. b In saga1 mutants, the accumulation of Citransporters is reduced, and the pyrenoid morphology is changed (Rai et al. 2021), and it was suggested that the mitochondrial CAs might play a role in the recapture of CO 2 leaking from chloroplasts. CCP1 and CCP2 are induced under LC conditions (Fang et al. 2012;Wang et al. 2016) and localize to the mitochondria (Atkinson et al. 2016). Although the growth of CCP1/CCP2 knockdown strains was reduced under LC conditions, their Ci-affinity was not affected (Pollock et al. 2004). Therefore, the reduced growth and Ci-affinity observed in saga1 mutants may partly be due to the reduced accumulation of Ci-transporters and mitochondria-localized CCM factors. Secondly, dispersion of LCIB in chloroplasts may also be responsible for the reduced inorganic carbon affinity. Since LCIB is thought to have CA activity (Jin et al. 2016), it is possible that more CO 2 is leaked from the pyrenoids in the saga1 mutant than in the wild type. Thirdly, many of the pyrenoids in the saga1 mutant do not have pyrenoid tubules (Itakura et al. 2019), suggesting that much of the transported Ci is unable to reach Rubisco within the pyrenoids. Therefore, the decreased V max observed in the saga1 mutant may be due to changes in pyrenoid morphology. The changes in photosynthetic properties observed in the saga1 mutant in previous studies and in this study could be due to these three effects.

The presence of multiple pyrenoids inhibits LCIB migration
Under VLC conditions, LCIB migrates to the basal region of the chloroplast and localizes around pyrenoids ). This suggests that there is a mechanism by which LCIB recognizes the position of the pyrenoid and moves to the basal region of the chloroplast. In the isa1 mutant, which has lost its starch sheath, LCIB does not localize with the ring-like structure under VLC conditions but moves to the basal region of the chloroplast (Toyokawa et al. 2020). In contrast, LCIB migration is inhibited in the saga1 mutants, which have multiple pyrenoids surrounded by a thin starch sheath. This suggests that multiple pyrenoids in the chloroplast may prevent LCIB from recognizing the chloroplast basal region.
Another possibility is that morphological changes in pyrenoids, rather than just an increase in the number of pyrenoids, may affect LCIB localization. Considering that LCIB is localized as puncta around the pyrenoid periphery (Yamano et al. 2010) and BST3 localizes to the thylakoid membrane and interacts with LCIB and LCIC (Mukherjee et al. 2019;Mackinder et al. 2017), LCIB could localize at the site of thylakoid membrane penetration into the pyrenoid matrix. On the other hand, it has been suggested that SAGA1 also localize with pyrenoid tubules around the pyrenoid (Itakura et al. 2019). Therefore, it is possible that SAGA1 could bind to LCIB and contribute to the LCIB localization around the pyrenoid.

Morphological changes in pyrenoids affect CCM regulation
Although CAS is thought to migrate across the thylakoid membrane to the pyrenoid, in the saga1 mutant, CAS is dispersed throughout the chloroplast. The dispersion of CAS in the saga1 mutant suggests that SAGA1 may have a role in recruiting thylakoid-bound CAS into the pyrenoid. On the other hand, it is known that the morphology of thylakoid membranes changes during photoacclimation (Nagy et al. 2014). Because CAS is thought to localize on the stroma side of the thylakoid membrane via hydrophobic sequences (Wang et al. 2016), it is possible that thylakoid membrane remodeling is important for CAS localization changes. Since the number of pyrenoid tubules is reduced and the tubules is deformed in the saga1 mutant (Itakura et al. 2019), premature development of the morphology of the pyrenoid tubules may inhibit CAS migration into the center of pyrenoid via thylakoid membrane remodeling. In order to argue more clearly for a relationship between SAGA1 mutations and CAS localization, it is important to generate new saga1 mutants and CAS-Venus expressing strains that do not have blob structure.
Under CO 2 -limiting conditions, Ca 2+ concentration in the pyrenoid increases (Wang et al. 2016). Although the physiological significance of the increase is not yet clear, it is possible that the coexistence of CAS and high concentrations of Ca 2+ in the pyrenoid is important for the activation of CAS function, as CAS exhibits low-affinity and high-capacity Ca 2+ binding properties (Han et al. 2003). Thus, CAS dispersion and inhibition of CAS-Ca 2+ binding in saga1 mutants may be responsible for the suppression of gene expression. The CAS-dependent genes downregulated in saga1 mutants are also a subset of the CCM1-dependent genes (Table 1), so the migration of CAS into the pyrenoid tubules could be important for crosstalk between CCM1 and CAS. However, several CAS-dependent genes (PPP30, Cre12.g541550, and Cre26.g756747) were expressed in saga1 mutants, suggesting that the binding of dispersed CAS to Ca 2+ may induce the expression of these genes.
The relationship between CAS localization changes and gene expression is still speculative. Isolation and characterization of novel mutants with aberrant CAS localization without changing the structure of pyrenoid could reveal the importance of the localization of CAS visual screening using fluorescent proteins. On the other hand, it may be possible to evaluate the expression levels of CAS-dependent genes in the transgenic strain by forcing the CAS to migrate to the pyrenoid in the saga1 mutant by adding a Rubiscobinding motif necessary for pyrenoid localization (Meyer et al. 2020).