CO2 supply modulates lipid remodelling, photosynthetic and respiratory activities in Chlorella species

Abstract Microalgae represent a potential solution to reduce CO2 emission exploiting their photosynthetic activity. Here, the physiologic and metabolic responses at the base of CO2 assimilation were investigated in conditions of high or low CO2 availability in two of the most promising algae species for industrial cultivation, Chlorella sorokiniana and Chlorella vulgaris. In both species, high CO2 availability increased biomass accumulation with specific increase of triacylglycerols in C. vulgaris and polar lipids and proteins in C. sorokiniana. Moreover, high CO2 availability caused only in C. vulgaris a reduced NAD(P)H/NADP+ ratio and reduced mitochondrial respiration, suggesting a CO2 dependent increase of reducing power consumption in the chloroplast, which in turn influences the redox state of the mitochondria. Several rearrangements of the photosynthetic machinery were observed in both species, differing from those described for the model organism Chlamydomonas reinhardtii, where adaptation to carbon availability is mainly controlled by the translational repressor NAB1. NAB1 homologous protein could be identified only in C. vulgaris but lacked the regulation mechanisms previously described in C. reinhardtii. Acclimation strategies to cope with a fluctuating inorganic carbon supply are thus diverse among green microalgae, and these results suggest new biotechnological strategies to boost CO2 fixation.

rate and waste products as well as wastewater-derived effluent can be used as fertilizers for their cultivation (Lum, Kim, & Lei, 2013).
Light is harvested in the microalgal chloroplast by pigment binding protein complexes called Photosystem I (PSI) and II (PSII). These complexes are composed of a core complex, where photochemical reactions occur, and an external antenna system, which increases light harvesting efficiency and where several photoprotective reactions occur (Gao, Wang, Yuan, & Feng, 2018;Pan, Cao, Su, Liu, & Li, 2020).
In oxygenic photosynthetic organisms, as eukaryotic microalgae, PSI and PSII work in series to strip electrons from water and transfer them to NADP + producing NADPH. During this linear electron transport, protons are pumped from stroma to the lumen generating an electrochemical gradient used by ATPase to synthetize ATP. ATP and NADPH are then used by the Calvin Benson cycle to fix CO 2 into sugars. In parallel, another electron transport chain takes place in mitochondria, consuming oxygen and NADH and releasing NAD + and ATP. A constant balance between chloroplast and mitochondrial activity is fundamental for cell survival and for adaptation to fluctuating environmental conditions.
It is important to point out that CO 2 diffusion in the water environments, where microalgae live, is strongly reduced compared to CO 2 diffusion in air. CO 2 -limitation is known to reduce the consumption of ATP and NADPH by the Calvin Benson cycle leading to an over-reduced photosynthetic electron transport chain, which could potentially lead to oxidative stress (Y. Wang, Stessman, & Spalding, 2015). For this reason several microalgae species evolved an efficient system to enrich the CO 2 level inside the cell, that is, Carbon Concentrating Mechanism (CCM), a complex mechanism by which inorganic carbon is actively transported close to the enzyme responsible for its fixation, that is, the Ribulose-1,5-bisphosphate carboxylaseoxygenase (RUBISCO) enzyme (Y. Wang et al., 2015). The CCM mechanism is induced by low CO 2 concentrations (air level or lower) (Y. Wang et al., 2015). CO 2 availability plays a critical role in modulating photosynthetic efficiency and biomass accumulation in microalgal cultures. For example, in the model green alga Chlamydomonas reinhardtii, CO 2 has been reported to act as a molecular switch inducing a complex network of cell adaptation mechanisms including a translational up-regulation in the formation of PSII antenna complexes (Berger et al., 2014(Berger et al., , 2016Blifernez-Klassen et al., 2011;Mussgnug et al., 2005;Wobbe et al., 2009). In conditions of low CO 2 availability, accumulation of the cytosolic RNA-binding protein NAB1 is triggered by the transcription factor LCRF (Low Carbon dioxide Response Factor) (Blifernez-Klassen et al., 2011). NAB1 then represses the translation of transcripts encoding light-harvesting antenna proteins (Berger et al., 2014;Mussgnug et al., 2005). The translation repressor activity of NAB1 is controlled by two independent mechanisms related to the methylation of Arg90 and Arg92 residues (Blifernez, Wobbe, Niehaus, & Kruse, 2011) and to the redox state of Cys181 and Cys226 residues (Wobbe et al., 2009). NAB1 is highly active in the methylated state, while reduced Cys181 and reduced Cys226 are required for NAB1 RNA-binding activity. Nitrosylation of Cys181 and Cys226 has also been reported to inhibit the RNA binding activity of NAB1 (Berger et al., 2016). The truncation of the PSII antenna reduces the excitation pressure on the photosynthetic apparatus as a response to diminished CO 2 availability (Berger et al., 2014). Accumulation of NAB1 and its post-translational regulation have been demonstrated to be finely tuned as an acclimation mechanism to different environmental conditions, including varying CO 2 availability (Berger et al., 2014;Berger et al., 2016;Wobbe et al., 2009).
Among microalgae species discovered, Trebouxiophyceae represent an evolutionary defined class of green algae (Chlorophyta) comprising the green freshwater algae of the Chlorella genus, one of the first microalgae to be cultured on a large scale due to their easy cultivation and high resistance to stresses (Borowitzka, 2018;B. Yang, Liu, Jiang, & Chen, 2016). Species belonging to the Trebouxiophyceae class are evolutionarily separated from the model species of green alga, C. reinhardtii, belonging to the Chlorophyceae class. Chlorella species are interesting for industrial cultivation, being reported to rapidly accumulate biomass containing high lipid, protein, carotenoid and vitamin amounts (Camacho, Macedo, & Malcata, 2019;Cecchin et al., 2019;Li, Zheng, Yu, & Chen, 2013;Lum et al., 2013;Sarayloo et al., 2017;Treves et al., 2013). However, the lack of genetic resources and the low efficiency of transformation methods have limited the development of genetic engineering in these species (Cecchin et al., 2019;Lin, Tan, Hsiang, Sung, & Ng, 2019).
In this work two Chlorella species, Chlorella sorokiniana and Chlorella vulgaris, were investigated for their physiologic and metabolic responses to a fluctuating CO 2 availability. Chlorella sorokiniana and C. vulgaris were chosen among other Chlorella species being already selected as those with the highest biomass accumulation potential (Kobayashi et al., 2015). Our results highlighted contrasting metabolic responses to CO 2 availability among green algae.
2 | RESULTS 2.1 | Species of the genus chlorella adjust their carbon flow differentially in response to high CO 2 availability Chlorella sorokiniana and C. vulgaris cells were grown in 80 ml batch airlift photobioreactors bubbled with air (CO 2 concentration of 0.04%, defined as AIR condition throughout the manuscript) or air enriched with 3% of CO 2 (defined as CO 2 condition throughout the manuscript). Chlorella strains were cultivated at 300 μmol photons m À2 s À1 until the saturation phase was reached. Growth kinetics were followed by measuring the optical density (OD) at 720 nm and fitted with sigmoidal function as showed in Figure 1a,b. In both species, the increased CO 2 concentration induced a faster growth rate, as highlighted by the first derivative of the growth curves (Figure 1c,d).
Total biomass production was increased in CO 2 compared to AIR condition by 253% ± 52% in C. sorokiniana and 269% ± 13% in C. vulgaris (Figure 1e,f). Moreover, $4-fold increase in maximal daily productivities was observed for both species in CO 2 condition. These data confirmed that in the cultivation conditions applied carbon fixation processes were stimulated in Chlorella species by high CO 2 suggesting a possible carbon-limitation in AIR condition. Interestingly a reduction of dry biomass per cell was evident in both C. sorokiniana and C. vulgaris in CO 2 condition, with a stronger decrease compared to the AIR condition in the latter species ( Figure S1).
Biomass composition at the end of the growth curves was evaluated, and this revealed a significant increase in total lipids in both species, while different effects were observed on protein and starch content per dry weight ( Figure 2). Indeed, CO 2 condition triggered an accumulation of proteins per dry weight in C. sorokiniana with the amount of starch not being altered. In C. vulgaris, protein amounts remained stable, but a decrease in starch content per dry weight was observed. On a cell basis, despite the reduction of dry biomass per cell in CO 2 conditions compared to AIR in both C. sorokiniana and C. vulgaris ( Figure S1), a slight increase of protein and lipid accumulation were measured in C. sorokiniana cells grown at high CO 2 availability, while only lipids remained constant in C. vulgaris in both AIR or CO 2 conditions ( Figure S2). This suggests a different behaviour of the two species in macromolecule accumulation upon increased carbon availability underlying potential differences in metabolic rearrangement.
Fatty acid compositional and lipid class analyses revealed genusspecific response to varying CO 2 availability (Figure 3a-f). An increase in galactolipids under CO 2 -replete conditions was observed in C. sorokiniana while triacylglycerol (TAG) was only increased in C. vulgaris. Differently, a decrease in phospholipids in CO 2 could be noted in both genera ( Figure 3, Figure S3). Specifically, a strong increase in monogalactosyldiacylglycerol (MGDG) and digalactosyldiacylglycerol (a) (c)   responsible for CO 2 fixation in organic molecules, was first quantified.
RUBISCO amount was similar in both species in the two conditions analysed on a cell basis ( Figure S4), suggesting that RUBISCO accumulation was not tuned by CO 2 concentration in Chlorella. This result is consistent with previous findings in soybean leaves exposed to different CO 2 concentrations (Campbell, Allen, & Bowes, 1988). We then analysed PSII maximum quantum yield by measuring the fluorescence parameter F v /fm, which is often used as a general indicator of the fitness of the culture. As reported in Figure  This adaptation in C. sorokiniana is consistent with results previously reported for the model alga C. reinhardtii, where a similar strong reduction of Chl content per cell was observed at high CO 2 concentration (Polukhina, Fristedt, Dinc, Cardol, & Croce, 2016). In contrast, this was not the case for C. vulgaris, where Chl/cell content was not significantly different in CO 2 compared to AIR condition ( Figure 4a).
To further investigate remodelling of the components of photosynthetic complexes, PSI/PSII ratio and LHCII/PSII ratio was evaluated by immunoblot analysis (Figure 5a,b). In C. sorokiniana, both PSI/PSII and LHCII/PSII ratio were decreased in CO 2 compared to AIR condition, while no significant differences were observed in C. vulgaris respectively, the conditions with minimum or maximum migration of LHCII to PSI, can be induced by consuming or increasing the reducing power in the chloroplast as described in (Fleischmann et al., 1999). In particular, S1 or S2 state can be measured on whole cells by measuring Chl fluorescence emission at 77 K, where both PSI and PSII emission is detectable. As reported in Figure 6, both C. vulgaris and C. sorokiniana were able to undergo state transitions in both AIR and CO 2 with an increased PSI contribution in S2 compared to S1. However, C. sorokiniana cells grown in CO 2 exhibited an increased capacity for state transitions compared to cells grown in AIR. When 77 K fluorescence emission in S1 or S2 were compared to 77 K fluorescence of cells directly harvested in their growing conditions, it was possible to observe that both C. vulgaris and C. sorokiniana were essentially growing in S2 state in AIR condition. A different behaviour was instead observed in CO 2 condition between the two species herein analysed: while C. vulgaris cells were characterized by an intermediate state between S1 and S2, C. sorokiniana was essentially in S1. These results indicate that the reduction in C. sorokiniana grown in CO 2 of LHCII/ PSII ratio was related to a decrease in LHCII preferentially connected in AIR to PSI, suggesting a compensatory regulation between antenna regulation and state transitions.
Photosynthetic electron transport is coupled to the formation of a proton gradient across the thylakoid membrane, exploited by ATPase as proton motif force to produce ATP (Tikhonov, 2013). ATPase content on Chl basis and proton-motive force (pmf ) upon exposure to different light intensity was evaluated in CO 2 and AIR grown cells ( Figure 7). The pmf can be estimated by measuring the light dependent-electrochromic shift of carotenoid absorption (Bailleul, Cardol, Breyton, & Finazzi, 2010). In this case, the behaviour of the two Chlorella species was similar with a reduced pmf in CO 2 compared to AIR condition. At the same time, an increase in ATPase content under CO 2 condition was detected for both C. vulgaris and C. sorokiniana.
Likely, the higher level of ATPase in CO 2 condition improved proton movement back to the stroma resulting in reduced pmf and higher ATP production in CO 2 condition. Furthermore, we investigated the influ- PSI is a plastocyanin-ferredoxin oxidoreductase that reduces NADP + to NADPH by a ferredoxin-NADP + reductase (FNR) enzyme.
In parallel, the mitochondrial respiratory electron transport chain oxidase NADH releasing NAD + . Chloroplasts and mitochondria communicate to balance the NAD(P) + /NAD(P)H pool (Dang et al., 2014;Johnson & Alric, 2013;Uhmeyer, Cecchin, Ballottari, & Wobbe, 2017 2.3 | The response of mitochondrial respiratory pathways to CO 2 availability Mitochondrial respiration is a fundamental process that allows producing ATP while releasing NAD + that can return to the chloroplast. The mitochondrial electron transport chain, also called cytochrome pathway, includes an ATP synthase complex, called also complex V, and four oxidoreductase complexes that oxidize the reducing power and produce ATP thank to the electrochemical gradient that is formed across the membrane. In addition, an alternative oxidase (AOX) might operate directly coupling ubiquinol oxidation with the reduction of O 2 to H 2 O serving as an alternative route bypassing the electron transport chain thus dramatically reducing the energy (ATP) yield. AOX was reported having a role in the protection mechanism for the respiratory chain (Boekema & Braun, 2007;Vanlerberghe, 2013).
The contribution of cytochrome and alternative pathways ( Figure 9) was investigated by measuring the dark respiration in the presence of two specific inhibitor: SHAM (salicylhydroxamic acid) that inhibits AOX and so the alternative pathway, and myxothiazol that locks the complex III, therefore, blocking the cytochrome pathway (Dang et al., 2014). We observed that the total dark respiration on a cell basis is essentially unaffected in C. sorokiniana, while a strong reduction was reported for C. vulgaris in CO 2 condition. In both species a reduction of the fraction of dark respiration operating through AOX was evident, leading to an increased efficiency of ATP production by NADH oxidation through the cytochrome pathway.
2.4 | NAB1-like proteins in C. vulgaris and C. sorokiniana Acclimation to different carbon availability has been reported in C. reinhardtii to involve the translational repressor NAB1. NAB1 acts as a molecular switch triggered by the redox state of the cell, which is F I G U R E 6 State transitions in AIR versus CO 2 . State transition analysis by 77 K fluorescence emission spectra in state 1 (S1, red lines) or state 2 (S2, blue lines) conditions in C. sorokiniana (a-c) and C. vulgaris (d-f) in AIR (A, D) or CO 2 (b, e). S1 was induced by shaking vigorously cells in a low light (⁓5 μmol m 2 s À1 ) with 10 μm of DCMU for at least 15 min to oxidize the plastoquinone pool while S2 was induced by adding 250 μm sodium azide to inhibit mitochondrial respiration and to reduce the plastoquinone pool as described in Fleischmann et al. (1999). Black lines are related to cells harvested in the different growing conditions (AIR or CO 2 ) and dark adapted for 1 min before freezing at 77 K. maximum capacities for state transitions were then quantified from the maximum fluorescence emission at 720 nm as (F S2 À F S1 )/F S2 . Data reported are means of three biological replicates with standard deviation shown. Significant difference in CO 2 versus AIR are indicated with different letters (a, b, or c, p < .05) in turn strongly influenced by the carbon availability, repressing the translation of specific transcripts, including those for LHC subunits.
To evaluate the possible conservation of NAB1 in the C. vulgaris and C. sorokiniana species, BLAST search was performed using the  Figure S6). Consistent with the bioinformatic analysis, immunoblot analysis using anti-NAB1 antibody revealed a band at the expected molecular weight (27 kDa) in both C. reinhardtii and C. vulgaris but not in C. sorokiniana ( Figure S7). However, while the accumulation of NAB1 was increased in AIR compared to CO 2 in C. reinhardtii condition, as previously described, a similar level of NAB1-like protein was observed in C. vulgaris cells grown in AIR or CO 2 ( Figure S7). Increased accumulation of NAB1 in C. reinhardtii at low CO 2 availability has been recently reported as being related to a transcriptional activation mediated by the transcription factor LCRF, which belongs to the Squamosa promoter binding protein ( were conserved, while Cys226 or Arg90 are substituted by a valine and a serine residue, respectively, in the NAB1-like subunit in C. vulgaris ( Figure S8). It is important to note that the substitution of Cys226 in C. reinhardtii NAB1 arrested the protein in its active state and abolished the redox control mechanism leading to a pale green phenotype in strains exclusively expressing NAB1 Cys226Ser (Wobbe et al., 2009). As reported in Figure S5, when comparing the functional antenna size of C. reinhardtii, C. vulgaris and C. sorokiniana grown in AIR or CO 2 conditions, a smaller light harvesting capacity was evident in the Chlorella species compared to C. reinhardtii in CO 2 , while similar PSII antenna size was observed in AIR. This result suggests that NAB1 dependent adaptation of PSII antenna size to different CO 2 concentration is not conserved in C. vulgaris and C. sorokiniana, because either the protein is not present (C. sorokiniana) or the redox control is not functional (C. vulgaris). As another important difference, the NAB1 homolog in C. vulgaris does not accumulate upon CO 2 limitation.

| DISCUSSION
Atmospheric CO 2 concentration has significantly increased over the last 100 years and is continuing rising at an unprecedented speed. This greenhouse gas strongly contributes to climate change and global warming leading to a potential severe environmental crisis. Microalgae are promising platforms to capture CO 2 , possibly integrating microalgae cultivation with CO 2 recovery from flue gases, thus reducing industry derived CO 2 emission and carbon footprint Collet et al., 2011;Dineshkumar, Chauhan, & Sen, 2020;Garcia-Cubero, Moreno-Fernandez, & Garcia-Gonzalez, 2018). For this reason, understanding the cell acclimation process involved in CO 2 metabolism is crucial to develop new strategies for improving the ability of microalgae to acquire and accumulate carbon. In this work we focused on two of the most promising species for microalgae cultivation at industrial level, C. sorokiniana and C. vulgaris (Bernaerts, Gheysen, Foubert, Hendrickx, & Van Loey, 2019;Camacho et al., 2019;Kobayashi et al., 2015;Li et al., 2013;Niccolai, Zittelli, Rodolfi, Biondi, & Tredici, 2019;Sun, Chen, & Du, 2016). Moreover, high quality annotated genome and transcriptomes are available for these species (Cecchin et al., 2019;Hovde et al., 2018), paving the way for possible future biotechnological manipulation on the base of the results obtained. Chlorella sorokiniana and C. vulgaris were grown in airlift photobioreactors under atmospheric level of CO 2 (~0.04% CO 2 , AIR condition) or 3% CO 2 (CO 2 condition). We discuss the metabolic consequence of the different photosynthetic and respiratory responses to high or low CO 2 levels in two Chlorella genus. It worth to note that CO 2 concentration in flue gases can reach higher concentration than 3%, being usually in the 10-20% range (X. Wang & Song, 2020). Previous work demonstrated that increase CO 2 availability in the range of 1-8% saturated biomass productivity in C. vulgaris, while higher CO 2 concentration caused a strong reduction in biomass yield (Garcia-Cubero et al., 2018). Nevertheless, cultivation of different Chlorella species at higher CO 2 concentration (up to 20%) was described in some reports Freitas, Morais, & Costa, 2017;Garcia-Cubero et al., 2018;Zhang et al., 2019). The metabolic changes herein described upon cultivation of C. vulgaris and C. sorokiniana providing 3% CO 2 compared to the AIR condition could thus be further modulated upon cultivation of these species at industrial level for partially capturing CO 2 in flue gases. Nevertheless, the highlighted adaptation responses in CO 2 conditions compared to the AIR case, provide useful insights to understand the different cell strategies evolved in these Chlorella species to manage high carbon availability.

| CO 2 availability boosts biomass accumulation
Increasing CO 2 supply boosted (~260% increase) in biomass yield in both species. It is worth to note that the increased biomass productivity observed was accompanied in both species with a decrease in dry weight per cell at high CO 2 availability with a consequent strong increase in cell density ( Figure S1). Likely, high carbon flux in CO 2 conditions caused increased cell divisions, reducing the cell size. Moreover, it is possible that the higher production of macromolecules per dry weight, as reported in Figure 2, allow the cells grown at high CO 2 to support cell replication for longer even when nutrient in the medium become limiting. Interestingly, differential response was observed between the two strains in terms of biomass composition (i.e., protein, starch and lipid amount). In C. vulgaris the fractions of dry biomass related to starch and lipids were, respectively, reduced and increased, with in particular a strong increase of TAG accumulation either on dry weight or on a cell basis. This suggests a redirection of the energy reserves from starch to TAG accumulation, a class of macromolecules with a higher energy content per gram, indicating an improved light energy conversion. In C. sorokiniana not only lipids, but also protein content increased, the latter being an additional carbon sink in cells grown at high CO 2 concentration. Comparing the productivity of the two Chlorella species herein investigated, C. vulgaris was characterized by an increased biomass productivity at high CO 2 availability, making this species the best candidate for industrial application. The high protein content observed in C. vulgaris suggests a possible food application of this organism, which is already widely considered a novel food (Bernaerts et al., 2019;Camacho et al., 2019;Niccolai et al., 2019). The carbon storage as TAG observed in CO 2 condition in C. vulgaris is also a desirable trait for possible use of this species for biodiesel production, even if the feasibility of biofuels production by microalgae cultivation is highly debated (Koyande et al., 2019). Finally, the high starch content in C. sorokiniana suggests a possible biotechnological application of this species to convert CO 2 into sugars to be used for different purposes as for fuelling fermentative process by different microorganisms' communities to produce bioethanol, biomethane or other high value metabolic compounds.

| Lipid remodelling and CO 2 availability
Both Chlorella genus showed an increase in total lipid composition of dry biomass accumulated under high CO 2 , and they further remodelled their fatty acid as well as lipid class composition albeit in different ways. Phospholipids were reduced whereas the betaine lipid DGTS were increased in both species by high CO 2 level. Betaine lipid, a non-phospholipid, has been observed often increased during phosphate limitation, presumably replacing the function of phospholipids in cell membranes (Hidayati et al., 2019;Murakami, Nobusawa, Hori, Shimojima, & Ohta, 2018;Riekhof, Naik, Bertrand, Benning, & Voelker, 2014). For still unknown reasons, a similar response between P limitation and high CO 2 was observed. We can speculate that increased CO 2 availability required phosphate reallocation to the different macromolecules produced, requiring partial phospholipids substitution with betaine lipid DGTS.
In C. vulgaris, the increase in lipid content per dry weight was mostly due to an increase in TAG, whereas, in C. sorokiniana the increase in total lipids was mainly due to an increase in the two galactolipids (MGDG and DGDG), the major lipids of photosynthetic membranes . The differential response in lipid classes in the two strains to high CO 2 level is further supported by fatty acid compositional alterations. Among other reasons, lipid compositional changes are mostly likely results of altered redox status brought about by differential chloroplast and mitochondrial energetic activities in response to varying CO 2 availability in the two Chlorella genus (Figures 6 and 8) (Burlacot, Peltier, & Li-Beisson, 2019).

| Thylakoid reorganization and photosynthetic activity under varying CO 2
The increase in DGDG and MGDG (Figure 3, Figure S3), surprisingly, was observed together with a reduction in Chl content per cell in C. sorokiniana under high CO 2 . Nevertheless, the observed reduction of Chl content per cell in C. sorokiniana grown in CO 2 condition compared to AIR was in line with the results obtained previously in C. reinhardtii (Polukhina et al., 2016), while this was not the case for C. vulgaris, where Chl content per cell is independent from CO 2 availability. Again, similar to C. reinhardtii, C. sorokiniana grown in high CO 2 was characterized by reduction of LHCII/PSII content and a reduction in PSI/PSII ratio, while these adaptations were not observed in C. vulgaris. The reduced LHCII/PSII content observed in C. sorokiniana grown in CO 2 condition did not affect the functional antenna size of PSII: differently from previous observation in C. reinhardtii, functional antenna size of PSII was not influenced by CO 2 availability in both Chlorella species herein investigated (Figure 5f, Figure S5). In C. reinhardtii it was indeed reported that high CO 2 availability caused an increase in the functional antenna size of PSII, due to suppressed accumulation at high CO 2 concentration of NAB1, the translation repressor specific for LHCII encoding mRNAs (Berger et al., 2014).
Here, we report the identification of a NAB1-like protein in C. vulgaris only, which however, is not differently accumulated in AIR versus CO 2 conditions. Accordingly, homologous protein of the transcription factor LCRF, the transcriptional regulator of NAB1 recently identified in C. reinhardtii (Blifernez-Klassen et al., 2011) could not be found in either C. vulgaris or C. sorokiniana. Moreover, in the C. vulgaris NAB1-like subunit only Cys181 and Arg92 residues are conserved among the two cysteine (Cys181, Cys226) and two arginine (Arg90, Arg92) residues reported in C. reinhardtii NAB1 to be involved in its activity as translational repressor. Taken together, the NAB1-dependent regulation of PSII antenna size at different CO 2 concentration in C. reinhardtii is absent in both C. sorokiniana and C. vulgaris, where NAB1 homolog was, respectively, not identified or is missing the crucial redox control of its translational inhibition activity ( Figures S6 and S7). Indeed, the PSII antenna size of both C. sorokiniana and C. vulgaris was not affected by different CO 2 availability, being similar to the PSII antenna size measured in the case of C. reinhardtii grown in AIR ( Figure S5). Regulation of PSII antenna size by NAB1 is thus an acclimation mechanism finely controlled by the redox state of the cell, which is not conserved among Chlorophyta.
It is important to note that the effect of CO 2 availability on C. reinhardtii LHCII/PSII ratio is still under debate, with Polukhina and coworker reporting a general reduction of LHCII/PSII content in C. reinhardtii grown in high CO 2 , in parallel with a decrease in PSI/PSII ratio (Polukhina et al., 2016). Considering the possibility of LHCII proteins to function as PSI antenna, it was not excluded by Polukhina and coworkers that the decrease of LHCII content per PSII observed at high CO 2 might be mainly related to the amount of LHCII proteins acting as PSI antenna. Here we report a similar acclimation mechanism only in the case of C. sorokiniana, with the difference that PSII antenna size was not modulated by CO 2 availability.

| Cellular redox balance and CO 2 availability
Cellular reducing power is crucial for the overall carbon flow and cell metabolism: catabolic process generate reducing power, which can be used by oxidative phsophorylation to generate ATP or by the anabolic pathways. In photosynthetic organisms, the NADP + /NADPH balance influence both the light phase of photosynthesis and the carbon fixation reactions. In general it is possible to hypothesize that the increased capacity of Calvin Benson cycle to regenerate NADP + and ADP, thanks to the increased CO 2 availiability, trigger the light phase of photosynthesis in order to keep the NADPH/NADP + ratio similar to the AIR condition, as reported in Figure 8: this occurs by increasing the total amount of PSII compared to PSI and relatively redistributing the excitation pressure among PSI and PSII reducing the excitation pressure at the level of the former reducing the LHCII content bound to PSI. This acclimation process would explain the reduced LHCII/PSII content despite the similar PSII antenna size observed. The absence of such acclimation mechanisms in the case of C. vulgaris could be at the base of the strong reduction of NAD(P)H/NAD(P) + ratio observed at high CO 2 concentration, as a consequence of increased NADPH consumption by the Calvin Benson cycle. Alteration of RUBISCO content was not detected ( Figure S4), suggesting an enhanced RUBISCO activity due to the higher availability of substrate rather than an upregulation of enzyme to exploit the higher CO 2 availability. CO 2 fixation requires both ATP and NADPH: in both Chlorella species a decrease of pmf and an increased ATPase content was measured in CO 2 condition ( Figure 6). Likely, the higher level of ATPase in CO 2 condition prevents the accumulation of the electrochemical gradient, suggesting a higher ATP production.
Interestingly, dark respiration is differentially regulated in the two In conclusion, high CO 2 availability caused an increased biomass accumulation in both C. vulgaris and C. sorokiniana, likely related to increased photosynthates production. The increased carbon assimilation in high CO 2 redirected the metabolism toward biosynthesis of lipids (TAG) in C. vulgaris, and proteins in C. sorokiniana, respectively.
Increased carbon fixation at high CO 2 concentration requires an increased NADPH and ATP availability: while increased ATPase content and reduced pmf suggesting indeed an increased ATP regeneration in both C. vulgaris and C. sorokiniana, the increased NADPH requirement was differently satisfied in C. vulgaris and C. sorokiniana in CO 2 . In the case of C. vulgaris, chloroplast acted as a sink for reducing power, inducing consequently a reduced NADH availability for mitochondrial respiration, reducing in particular the relative contribution of alternative pathways, not related to ATP biosynthesis. In C. sorokiniana a reduced PSI/PSII ratio and reduced LHCII binding to PSI was observed in CO 2 , allowing an increased electron flow toward NADP + to NADPH regeneration and ensuring a similar NAD(P)H/ NAD(P) + ratio in both AIR or CO 2 conditions. A summary of the adaptation to CO 2 condition is shown in Figure S9.
Detailed multi-omics analysis would be required to investigate the molecular mechanisms at the base of the different responses of

| Biomass composition analysis
Lipid, starch and protein content of the biomass harvested at the end of the exponential phase were analysed as previously reported in Cecchin et al. (2019).

| Photosynthetic parameters and pigments extraction
Pigments were extracted with 100% DMSO at 60 C in dark conditions and measured with Jasco V-550 UV/VIS spectrophotometer.
Proton motive force upon exposure to different light intensities was measured by ECS with MultispeQ v2.0 (PhotosynQ) according to (Kuhlgert et al., 2016) and normalized to the Chl content of the sample. PSII activity was analysed by fluorescence measurements on whole cells using a Dual-PAM 100 instrument (WALZ). 77 K fluorescence emission spectra were acquired with a charge-coupled device spectrophotometer (JBeamBio) as previously described (Allorent et al., 2013). State transitions were measured on whole cells induced to state 1 or state 2 as described in (Fleischmann et al., 1999). PSII functional antenna size was measured from fast Chl induction kinetics induced with a red light of 11 μmol photons m À2 s À1 on dark-adapted cells incubated with 50 μM DCMU (Malkin et al., 1981). The reciprocal of time corresponding to two-thirds of the fluorescence rise (τ 2/3 ) was taken as a measure of the PSII functional antenna size (Malkin et al., 1981). P700 activity was measured with the DUAL-PAM-100 (Heinz-Walz) following the transient absorption at 830 nm upon exposure to actinic light. Maximum P700 activity was measured after a pulse of saturating light in whole cells treated with DCMU (3-[3,4-dichlorophenyl]-1,1-dimethylurea), ascorbate and methylviologen, as described in Bonente et al. (2012). The formation rate of NADPH was determined with the NADPH/9-AA module of the Dual-PAM 101 (Schreiber & Klughammer, 2009). Cells were harvested and resuspended in BG11 medium with 10% of Ficoll to reduce cells precipitation. Measurement was performed as described in (Schreiber & Klughammer, 2009) at the same light intensity of growth (300 μmol photons m À2 s À1 ). The slope during the light phase, between 60 and 120 s, was used to determine the rate of NADPH formation.

| Mitochondrial respiration
Samples in the exponential phase were subjected to respiratory rate measurements in the dark using a Clark-type O 2 electrode (Oxygraph Plus; Hansatech Instruments). Respiratory rates were normalized to cells number obtained by Countless ® II FL automated cell counter (Thermo Fisher). To discriminate between the individual contributions of the alternative and the cytochrome pathway dark respiration measurements were conducted as follows: cell samples (5*10 7 cell/ml) were transferred to the measurement chamber of the Clark electrode, respiration rates were recorded for 3 min prior to the addition of the first inhibitor, then respiration rates were recorded for three additional min finally the second inhibitor was added and measurements were continued for another 3 min. Alternative respiration was inhibited by adding 2 mM SHAM (salicylhydroxamic acid), while the cytochrome pathway (complex III) was inhibited by adding 5 μM myxothiazol. To assess the relative contribution of the cytochrome pathway, respiration was first measured in the absence of inhibitors (total dark respiration) before alternative respiration was inhibited by adding SHAM. Cytochrome dependent respiration was then inhibited using myxothiazol and the residual respiration determined in relation to the uninhibited state. The contribution of alternative respiration was determined by reversing the order of inhibitor addition (myxothiazol followed by SHAM) (Bailleul et al., 2015).

| Experimental replication and statistical treatment
All the experiments herein reported were performed at least three times. Errors are reported as standard deviations. Statistical significance was tested by Tukey's test.

ACKNOWLEDGMENTS
The research was supported by the ERC Starting Grant SOLENALGAE (679814) to Matteo Ballottari. The lipid analyses were performed at the HelioBiotec platform (CEA Cadarache).