Extracellular CahB1 from Sodalinema gerasimenkoae IPPAS B-353 Acts as a Functional Carboxysomal β-Carbonic Anhydrase in Synechocystis sp. PCC6803

Cyanobacteria mostly rely on the active uptake of hydrated CO2 (i.e., bicarbonate ions) from the surrounding media to fuel their inorganic carbon assimilation. The dehydration of bicarbonate in close vicinity of RuBisCO is achieved through the activity of carboxysomal carbonic anhydrase (CA) enzymes. Simultaneously, many cyanobacterial genomes encode extracellular α- and β-class CAs (EcaA, EcaB) whose exact physiological role remains largely unknown. To date, the CahB1 enzyme of Sodalinema gerasimenkoae (formerly Microcoleus/Coleofasciculus chthonoplastes) remains the sole described active extracellular β-CA in cyanobacteria, but its molecular features strongly suggest it to be a carboxysomal rather than a secreted protein. Upon expression of CahB1 in Synechocystis sp. PCC6803, we found that its expression complemented the loss of endogenous CcaA. Moreover, CahB1 was found to localize to a carboxysome-harboring and CA-active cell fraction. Our data suggest that CahB1 retains all crucial properties of a cellular carboxysomal CA and that the secretion mechanism and/or the machinations of the Sodalinema gerasimenkoae carboxysome are different from those of Synechocystis.


Introduction
Carbonic anhydrases (CAs; EC 4.2.1.1) form a polyphyletic group of enzymes currently comprising eight classes (α, β, γ, δ, ζ, θ, η, ι; [1]) and are ubiquitous throughout the tree of life. Catalysing the bi-directional interconversion of water-dissolved carbon dioxide (CO 2 /H 2 O) and bicarbonate (HCO 3 − /H + ), CAs are integral to carbon-based metabolism, especially by facilitating photosynthetic carbon fixation through release of CO 2 in the vicinity of RuBisCO in the context of cyanobacterial and also microalgal carbon-concentrating mechanisms (e.g., [2][3][4][5]). Moreover, when acting outside the cell, CAs are presumed to facilitate the uptake of inorganic carbon from the surrounding media as bicarbonate [6]. In the model cyanobacterium Synechocystis sp. PCC6803 (hereafter Synechocystis), the β-CA CcaA (Slr1347) and the γ-CA-like protein CcmM (Sll1031) constitute crucial components of a RuBisCO-associated bicarbonate dehydration complex within its β-type carboxysome [7]. While CcmM has been shown to be structurally important to bicarbonate dehydration complex organization [7], the enzymatic function of CcmM as an active CA has been lost in Synechocystis and various other beta-cyanobacteria [8,9]. Synechocystis CcaA, meanwhile, has been confirmed to be an active CA enzyme [10,11] which is constitutively expressed under a range of exogenous CO 2 supply conditions [12][13][14][15]. Under atmospheric CO 2 concentrations, CcaA has been shown to be indispensable for photoautotrophic growth (i.e., the ccaA knockout allele was found impossible to fully segregate), while elevated CO 2 (5% v/v) allows for full knockout segregation [16]. For β-cyanobacteria, CcaA is hypothesized to  (c) Schematic maps of the p∆ccaA knock-out and the pNS_cahB1 knock-in constructs used in this study. KanR, kanamycin resistance gene nptI; CmR, chloramphenicol resistance gene cat; PpsbA2, promoter of Synechocystis psbA2 gene encoding D1 subunit of photosystem II; bp, base pairs. Upstream/downstream regions used for homologous recombination with genomic target loci ccaA (slr1347) and neutral site (slr0168) are indicated as grey boxes. Primer-binding sites for genotyping PCR are indicated as half arrows. (d) Genotyping PCR of Synechocystis wildtype (WT), wildtype transformed with a CahB1 expression construct (WT + cahB1), mutant harbouring the ccaA deletion construct (∆ccaA KD ), and mutant harbouring the ccaA deletion construct transformed with cahB1 expression construct (∆ccaA + cahB1). Positions and sizes of the WT ccaA locus amplicon (1936 bp), the ∆ccaA locus amplicon (2048 bp), and the cahB1 expression construct amplicon (984 bp; PpsbA2:cahB1) are highlighted with arrowheads. Note the change from incomplete to complete segregation status of ∆ccaA upon introduction of the cahB1 gene.
To assess the functionality of CahB1 within Synechocystis, the ORF encoding CahB1 was cloned into an expression construct targeting the designated genomic neutral site slr0168. CahB1 was expressed under control of the strong psbA2 promoter and co-introduced with a chloramphenicol resistance gene to enable positive selection (Figure 1c). To generate knockout strains of the endogenous ccaA gene for complementation studies, the slr1347 ORF was targeted by homologous recombination and replaced with a kanamycin resistance gene (Figure 1c). Upon introduction into the ccaA-depleted mutants, the cahB1 expression construct was found to allow for segregation of the ∆ccaA knockout allele under atmospheric CO 2 while transformants harbouring only the ccaA KO construct do not segregate completely under the same conditions (resulting in effective knock-down of ccaA; hereafter ∆ccaA KD ) (Figure 1d). This indicates that some copies of the endogenous ccaA gene are indispensable under atmospheric CO 2 unless its function is complemented by a foreign gene, such as cahB1. This result is compatible with the previous reports [16,32].
The close resemblance of CahB1 to CcaA, as well as the capacity of cahB1 to genetically complement ∆ccaA, are highly suggestive of a role of CahB1 as carboxysomal β-CA rather than as a secreted β-CA in Synechocystis.

CahB1 Functionally Complements Loss of Synechocystis CcaA
Strains expressing the cahB1 gene in the WT background (i.e., harbouring both Synechocystis CcaA and CahB1; WT + cahB1) did grow to significantly higher final optical densities than parental WT (p = 2.3 × 10 −3 ), while ∆ccaA KD strains were significantly impaired in growth (p = 6.6 × 10 −10 ). This growth defect was alleviated in ∆ccaA + cahB1 strains (Figure 2a), which was in line with our genetic complementation data (Figure 1d). Based on whole-cell absorbance-spectra-derived estimates for all strains, cellular phycobiliprotein (PBP) contents were found largely unaltered with ∆ccaA + cahB1 levels being restored to WT levels, while WT + cahB1 displayed sightly increased (p = 5.4 × 10 −2 ) and ∆ccaA KD significantly decreased (p = 4.5 × 10 −4 ) cellular PBP levels, respectively ( Figure 2b). Correspondingly, the molar ratio of PBPs (phycocyanin and allophycocyanin) to chlorophyll a was significantly increased in ∆ccaA KD (Figure 2b; p = 4.0 × 10 −3 ). These findings parallel significant increases and reductions in cellular chlorophyll a content in WT + cahB1 (p = 2.3 × 10 −2 ) and ∆ccaA KD (p = 2.0 × 10 −2 ), respectively, while total cellular carotenoids were significantly reduced only in ∆ccaA KD mutants (p = 5.4 × 10 −6 ). Both chlorophyll a and carotenoid contents were recovered in ∆ccaA + cahB1 (Figure 2c). These findings indicate proportionate increases in cellular PBPs and chlorophyll a levels in WT + cahB1, and slightly stronger reductions in cellular chlorophyll a than in PBPs in ∆ccaA KD , while ∆ccaA + cahB1 broadly recapitulates WT-like growth and pigmentation with statistically insignificant deviations. Finally, a strong increase in minimum cellular fluorescence F o was observed in ∆ccaA KD relative to WT (p = 1.6 × 10 −5 ) which was fully recovered in ∆ccaA + cahB1 complementation strains (Figure 2d). A significant reduction in maximum photosystem II quantum yield approximate QY max observed in ∆ccaA KD (p = 2.6 × 10 −8 ) was partially recovered in ∆ccaA + cahB1 (p = 2.6 × 10 −5 ), while a slight yet statistically significant reduction in QY max was also induced by expressing cahB1 in the WT background (p = 4.7 × 10 −3 ) (Figure 2d).   Taken together, these data suggest that the cahB1 gene has a stimulating effect on photoautotrophic cell growth when added to WT cells and is capable of largely complementing the physiological defects caused by CcaA depletion. Thus, CahB1 is likely capable of functionally replacing carboxysomal β-CA in Synechocystis.

CahB1 Localizes to the Carboxysome in Synechocystis
In Sodalinema gerasimenkoae, CahB1 has been described to localize to the outer membrane/cell envelope fraction exclusively [26]. By virtue of complementing the ∆ccaA mutation in Synechocystis, however, localization of CahB1 to the carboxysome had to be assumed. To assess the subcellular localization of CahB1 in Synechocystis, cells were fractionated and CahB1 protein accumulation was traced by immunoblot assay using a specific peptide antibody raised against CahB1. Cells were broken mechanically using glass beads and a swing mill, and separated into total (T), soluble (S) and pelletable (P) fractions by centrifugation. Membrane proteins of the P fraction were then solubilized using a mixture of n-dodecylα/β-D-maltosides to obtain solubilizable (P S ) and insolubilizable (P I ) subfractions of the pelletable fraction. The obtained cell fractions displayed pronounced differences in their coloration and UV/Vis absorbance spectra (Figure 3a,b), as well as relative PsbA (i.e., D1) thylakoid marker protein content (i.e., relative depletion in fractions S and P I , and relative enrichment in fractions P and P S ) ( Figure 3b). The CahB1 protein could only be detected in cell fractions containing insolubilizable, pelletable components (i.e., fractions T, P, P I ; Figure 3c). It was co-accumulated with a non-soluble and insolubilizable subpopulation of the RuBisCO large subunit protein RbcL (Figure 3c), suggesting co-localization of CahB1 with carboxysome-encapsulated cellular RuBisCO [33]. CO 2 hydration assays commonly used to assess CA activity [6] showed CA-inhibitorsensitive enhancements of CO 2 hydration rates in P I fractions of both WT and WT + cahB1, which was observed via proton release and a concordant drop in pH over a measuring period of five minutes (Figure 3d). CO 2 hydration rates in WT + cahB1 P I fractions were found to be increased above WT levels and were significantly higher than both the nosample control (p = 6.6 × 10 −3 ) and WT sample rates (p = 1.3 × 10 −2 ) starting after one minute, while the measured WT P I fraction activity was only found to be significantly different from that of the no-sample control after two minutes (p = 2.2 × 10 −3 ) and onward (Figure 3d middle). Meanwhile, for the entire measuring period of five minutes, no elevation above the uncatalyzed background rates was observed in intact cells assayed for extracellular CA activity (Figure 3d left) or acetazolamide-treated P I fraction samples (Figure 3d right). These findings suggest that in Synechocystis CahB1 localizes to a cellular subfraction that harbours carboxysomes, and that the enzymatic activity of CahB1 is restricted to the said fraction rather than to the cell surface as would be expected based on previous reports from Sodalinema gerasimenkoae [24][25][26].   and P I cell fractions of WT and WT + cahB1. P I CA activity was measured in absence (middle) and presence (right) of 500 µM acetazolamide (AZA). Averages (graphs) and standard deviations (shaded areas) of n = 3 biological replicates. Uppercase letters indicate statistically significant differences (p ≤ 0.05) in relative CO 2 hydration rates according to multiple simultaneous comparisons in post hoc Bonferroni-Holm-corrected Tukey HSD tests after statistically significant among-group differences were detected by one-way ANOVA (d Intact Cells) p = 0.194; (d P I 0 µM AZA) p = 0.005; (d P I 500 µM AZA) p = 0.319; n.s., not significant.

The Sodalinema gerasimenkoae Genome Lacks Genes for Carboxysomal CAs Other Than CahB1
A genome sequence of the recently re-assigned Sodalinema gerasimenkoae IPPAS B-353 [30] has previously been published under the alias of Microcoleus chthonoplastes strain IPPAS B-353 (alias Microcoleus chthonoplastes strain IPPAS B-270) [31]. This genome sequence raises several additional questions as it encodes all structural subunits commonly forming a β-carboxysome [31,34], with very few atypical features to be observed in the predicted protein-coding gene sequences (Figure 4a). Unlike in Synechocystis where CcmK proteins are C-terminally truncated by 40-50 amino acid residues, the Sodalinema gerasimenkoae shell proteins CcmK1 and CcmK2 display typical lengths of 103 and 112 amino acid residues [35], respectively. Two near-identical copies of ccmK4 are found within the ccmK3/4 operon (designated ccmK4-1) and as a standalone satellite locus (designated ccmK4-2), which likewise display canonical lengths of 119 and 120 amino acid residues, respectively. The ccmK3 gene meanwhile appears to be subject to a frameshift mutation resulting in a premature stop codon at position 44. As mentioned previously, the genome lacks any other ccaA homologue besides CahB1 that could serve as carboxysomal β-CA. While some betacyanobacteria are known not to possess ccaA homologues, in exchange their ccmM gene appears to encode an enzymatically active γ-CA [8]. CcmM of Sodalinema gerasimenkoae is highly homologous to Synechocystis CcmM and phylogenetically clusters with the latter rather than any other CcmM proteins, however (Figure 4b). Similar to in Synechocystis, CcmM in Sodalinema gerasimenkoae does not conserve several amino acid residues that are presumably essential to the CcmM's γ-CA activity ( Figure 4c); therefore, it is unlikely that it represents an active γ-CA enzyme.
In summary, these findings lead us to conclude that the Sodalinema gerasimenkoae βcarboxysome may function in an unusual way, presumably lacking an active CA enzyme, while the genome encodes a CcaA-homologous and β-carboxysome-compatible CA, CahB1. The mode of CahB1 secretion to the extracellular space in Sodalinema gerasimenkoae and the reasons for its apparent limitations in Synechocystis remain to be elucidated, just like the functional properties of the tentatively CA-less carboxysome in Sodalinema gerasimenkoae. gene appears to encode an enzymatically active γ-CA [8]. CcmM of Sodalinema gera menkoae is highly homologous to Synechocystis CcmM and phylogenetically clusters w the latter rather than any other CcmM proteins, however (Figure 4b). Similar to in Sy echocystis, CcmM in Sodalinema gerasimenkoae does not conserve several amino acid re dues that are presumably essential to the CcmM's γ-CA activity ( Figure 4c); therefore is unlikely that it represents an active γ-CA enzyme. (c) Protein sequence alignment of two regions tentatively crucial to CcmM γ-carbonic anhydrase activity. Four presumably functionally important residues (marked by •; [8]) are indicated: a structurally important tryptophane (W) and asparagine (N) residue are highlighted in orange; a pair of presumably catalytic cysteine (C) residues is highlighted in yellow. Conserved positions (black), highly similar positions (>80% similarity/identity; dark grey), and similar positions (>60% similarity/identity; light grey) are highlighted.

Discussion
The data presented in this study strongly suggest that Sodalinema gerasimenkoae CahB1 acts as a functional carboxysomal β-CA in Synechocystis, which can genetically and physiologically compensate for the loss of the endogenous ccaA/CcaA (Figures 1-3). This is in line with the close sequence similarity that CahB1 bears to Synechocystis CcaA (Figure 1; [26]) and multiple reports of its functionality as an active and efficient CA enzyme [26][27][28]. Depletion of CcaA in the ∆ccaA KD mutant results in a pronounced increase in minimal fluorescence, F o (Figure 2d), which likely indicates a limitation of the acceptor side upon impaired carbon fixation [36][37][38], although the convolution of fluorescence from chlorophyll a with that from phycobiliprotein impair direct interpretation [39]. While ∆ccaA KD shows overall reduced cellular pigment contents, cellular chlorophyll a levels appear to be disproportionately reduced compared to cellular phycobiliproteins (Figure 2b,c). This suggests the increase in F o in ∆ccaA KD may in part be attributed to a larger contribution of (allo-)phycocyanin fluorescence. Still, as both cellular chlorophyll a and phycobiliprotein levels are reduced, the observed increase in F o is likely to underestimate the actual photosynthetic electron sink limitation caused by CcaA depletion. This limitation could be almost completely alleviated by the expression of CahB1, as indicated by the restored culture growth (Figure 2a Our findings in the heterologous Synechocystis test system contrast sharply with the proposed location of CahB1 outside of the Sodalinema gerasimenkoae carboxysome and outside the cytosol altogether [25,26]. While no extracellular CA activity could be observed upon expression of CahB1 in Synechocystis, a slight yet statistically significant enhancement of CA activity in the carboxysome-enriched cell fraction P I of WT + cahB1 cells as compared to the parental WT cells (Figure 3d) may account for the enhanced growth observed in the said expression strains (Figure 2b). As no enhanced growth beyond WT levels was observed in ∆ccaA + cahB1 complementation strains, growth enhancement is unlikely to stem from the presence of CahB1 alone and may either result from synergy effects in a cellular hybrid population of the two carboxysomal β-CAs, or from an overall increased level of cellular carboxysomal β-CA. In CahB1-expressing strains, the slight reduction in photosystem II maximum quantum efficiency estimate QYmax we observed (Figure 2d) may simply be attributed to the use of chloramphenicol as a positive selection marker. Chloramphenicol is a potent protein biosynthesis inhibitor commonly used to suppress the reparation of photodamaged D1 protein [40] and may therefore affect QYmax even in the presence of chloramphenicol acetyl transferase enzyme. Given the enhanced growth phenotype of WT + cahB1 strains, however, we consider this reduction in QYmax to be largely inconsequential. Future experiments will be required to dissect the underlying mechanism of enhanced growth in cahB1 and whether this feature is applicable to other model systems and cyanobacterial production strains.
Isolation of intact carboxysomes was attempted using a gentle approach employing glass beads and relatively weak detergents instead of the conventional French-press/Triton-Percoll methods [41], as the latter have been found to disrupt the carboxysome shell by removing large proportions of the CcmK2 shell component [42], thus potentially releasing any carboxysome-encapsulated CahB1 if it associates with the carboxysomal shell similar to CcaA [7]. Successful enrichment of carboxysomes in the P I fraction is indicated by the complete transfer of RbcL from the P fraction into the P I fraction (Figure 3c), with a large subpopulation of cytosolic free-standing RuBisCO complexes and assembly intermediates, as indicated by the RbcL signal in the S fraction being well in line with the previous reports [33]. The low catalytic activity observed in CO 2 hydration assays (Figure 3d) is similar to what has been observed in intact carboxysomes, whose CA catalytic activity is much lower than that of ruptured carboxysomes [43] or purified CcaA [7]. This is likely due to the intact microcompartment shell acting as a strong diffusive barrier which, under physiological conditions, is hypothesized to prevent leakage of internally generated CO 2 from carboxysomes while simultaneously protecting encapsulated RuBisCO from cytosolic O 2 [44]. The complete inhibition of P I -fraction CA activity by acetazolamide (AZA) is in line with previous reports of CahB1 being susceptible to AZA [27]. To our knowledge, susceptibility of Synechocystis CcaA to inhibitors has not been investigated in detail so far but recombinant CcaA is enzymatically active [7,11], and AZA-tolerant Synechocystis mutants have been described and hypothesized to harbour mutations in CA [45], even prior to the initial cloning of the ccaA gene [10]. Moreover, the high level of homology between CcaA and CahB1 (Figure 1; [26]), and the fact that CcaA is the only essential CA identified in Synechocystis so far [16,23], strongly suggest that CcaA would also be susceptible to AZA inhibition. Thus, the P I fraction is considered to likely contain enriched and intact carboxysomes with whom active CahB1 enzyme co-precipitates in both WT + cahB1 and ∆ccaA + cahB1 (Figure 3c). We hence conclude that heterologously-expressed CahB1 is likely to play the structural and physiological roles of the Synechocystis β-carboxysome, which is further supported by our observation of the cahB1 gene allowing for complete segregation of ∆ccaA KD mutants (Figure 1c).
The Sodalinema gerasimenkoae genome appears to encode a full set of functional βcarboxysomal genes [30,31]. While the ccmK3 gene is likely functionally disrupted, the shell protein CcmK3 has been shown to be dispensable, and knockout mutations do not severely affect the growth in Synechococcus sp. PCC7942 [46,47]. Still, its absence may result in assembled carboxysome shells with altered properties, e.g., considering the pHdependent metabolite permeability of the carboxysome shell [47]. Much more intriguingly, the Sodalinema gerasimenkoae genome lacks potential gene(s) encoding carboxysomal CA besides cahB1, (Figure 4; [30,31]). In both αand β-type carboxysomes, a functional CA enzyme is widely considered functionally integral [48,49], although Synechocystis carboxysomes have been observed to structurally assemble even in the absence of CcaA [16]. This raises the question of how the Sodalinema gerasimenkoae carboxysome functions and whether there is an unknown component that can substitute the β-CA CahB1 and/or the γ-CA CcmM [8]. Thus, the Sodalinema gerasimenkoae carboxysome appears highly atypical in this regard and may invite various lines of future functional investigation.
Lastly, the mode of secretion employed to target CahB1 to the outer membrane in Sodalinema gerasimenkoae remains elusive. Synechocystis sp. PCC6803 seemingly sequesters active CahB1 to the carboxysome, yet comparative analyses of the N-terminal region of CahB1/CcaA may hint at an underlying mechanism. The CahB1 N-terminus, like all other investigated homologues except for Synechocystis CcaA, starts with an MKK tripeptide. Unlike its homologues, the first 20 amino acid residues do not contain any negatively charged residues, however (Figure 1b), making it more similar to the positively charged n-region and the non-polar h-region of other cyanobacterial CA secretion peptides, such as that of Synechococcus EcaA [19]. While an MKK-tripeptide close to the N-terminus has been described to facilitate protein secretion in a Bacillus-derived synthetic secretion signal [50], negatively charged amino-acid residues are commonly absent from pre-cleavage site secretion signal sequences in eukaryotes, Gram-negative, and Gram-positive bacteria [51]. Thus, the CahB1 N-terminus may serve as a kind of proto-secretion signal that is sufficient to elicit protein secretion in the original Sodalinema gerasimenkoae system, but not in Synechocystis. CahB1 may thus serve as an interesting case study on how secretion signals can evolve out of previously non-secreted N-terminal sequences. On the other hand, as CcaA is hypothesized to be a later acquisition by horizontal gene transfer [8], the originally acquired beta-CA may have indeed been a secreted enzyme that evolved not to be secreted but rather to be sequestered to the carboxysome in β-cyanobacteria. In this case, CahB1 may represent a relic CcaA isoform whose secretion has not (yet) been evolutionarily suppressed in Sodalinema gerasimenkoae.

Taxonomy
The organism from which the cahB1 gene was originally isolated has undergone multiple taxonomic re-assignments. Originally being referred to as Microcoleus chthonoplastes IPPAS B-270 [25,26], synonymous to Microcoleus chthonoplastes IPPAS B-353 [31], the species was re-assigned to the genus Coleofasciculus [52] and then re-named Sodalinema gerasimenkoae IPPAS B-353 [21,30]. For the scope of this study, we refer to it as per the latest designation, i.e., Sodalinema gerasimenkoae IPPAS B-353.

Synechocystis cultivation and mutant generation
Cyanobacterial cells were routinely grown on BG11 media [59]. Stock cultures were grown on BG11 agar (0.8% w/v) plates containing 10 mM TES-KOH and 4 g L −1 sodium thiosulfate. Assay cultures were grown for seven days in liquid BG11 media at 25 • C under continuous illumination of 50 µmol photons m −2 s −1 white fluorescent light, continuous orbital agitation (65 rpm), and under exclusively atmospheric CO 2 supply. Glucose-tolerant non-motile Synechocystis sp. PCC6803 wildtype cells were obtained from the Leister lab (LMU Munich). Mutant strains were generated by natural transformation and homologous recombination as described previously [60].
Sodalinema gerasimenkoae cahB1 was cloned from a genomic library fragment provided by Elena Kupriyanova [26]. Synechocystis genes, promoters, and sequences for homologous recombination were amplified from genomic DNA extracts using HS VeriFi TM high-fidelity polymerase (PCR Biosystems Ltd., London, UK). Constructs for expression of CahB1 (pNS_cahB1_CmR) and KO of Synechocystis ccaA (p∆ccaA-KanR) were cloned by Gibson assembly [61] (New England Biolabs Inc., Ipswich, MA, USA). For DNA vector sequences, see Supplementary Material File S2.
Transformants were selected and segregated on incrementally increasing antibiotic concentrations in solid media, reaching final working concentrations of 100 µg mL −1 for kanamycin, and 15 µg mL −1 for chloramphenicol. Genotyping PCR was performed using Phire Plant Direct PCR Master Mix (ThermoFisher Scientific Inc., Waltham, MA, USA).

Synechocystis photosynthetic measurements
Minimal fluorescence F o and photosystem II maximum quantum yield QYmax were measured using a FluorCam 800 MF (Photon Systems Instruments, Drásov, Czech Republic) as previously described [62]. Culture samples grown as outlined above were harvested, washed with BG11 media, and adjusted to a final OD 730nm = 75. Aliquots of 10 µL were then dropped onto BG11 agar plates and incubated under previously applied growth conditions for three hours prior to measurement.

Synechocystis pigment extraction and quantification
Hydrophobic pigments were extracted and quantified as previously described [62,63]. Phycobiliprotein contents were estimated based on whole-cell UV/Vis absorbance spectra obtained at seven days past inoculation using a NanoDrop™ 2000c spectrophotometer (ThermoFisher Scientific Inc., Waltham, MA, USA) as previously described [64,65]. In preparation, whole-cell spectra were baseline corrected to absorbance values of 0.00 at λ = 750 nm (A 750 ) with A 550 being set to 0.2 × A 644 in accordance with established Synechocystis phycobilisome absorbance spectra [66] (see Supplementary Materials File S3).

CO 2 hydration assay
Carbonic anhydrase activity was assessed via observation of CO 2 -hydration-driven pH drop in 25 mM K 2 HPO 4 (pH 8.45) assay buffer as previously described [6]. For intact cell extracellular CA activity measurements, OD 730nm = 5.0 cell equivalents were harvested, washed, and resuspended in assay buffer prior to measurement. For P I fraction activity measurements, OD 730nm = 10.0 cell equivalents were fractionated as described above, and P I fractions were equilibrated for equal total protein content as approximated by Bradford protein assay (Nacalai Tesque Inc., Kyoto, Japan) with the lowest concentration sample being adjusted to 100 µL. An amount of 50 µL of P I fraction was then assayed for CA activity in absence and presence of 500 µM acetazolamide (Sigma-Aldrich, St. Louis, MO, USA). Assays were conducted on ice at 4 • C for 5 min, with a total assay buffer plus sample volume of 14.5 mL and 500 µL of ice-cold CO 2 -saturated H 2 O. Measurements were taken in 10 s intervals for the first 2 min, and in 60 s intervals for another three minutes.

Statistical Analyses
Charts were created using Microsoft Office Excel 365. For boxplots, internal datapoints are represented as dots. Horizontal lines represent the median, crosses represent average values, and boxes indicate the 25th and 75th percentiles. Whiskers extend 1.5-fold the interquartile range and outliers. Statistically significant among-group differences were tested for by one-way ANOVA, followed by post hoc Tukey HSD (honest significant differences) tests with Bonferroni-Holm p-value correction for multiple comparison. Significant differences according to multiple simultaneous post hoc comparisons are routinely indicated by uppercase/lowercase letters denoting the resultant groups of not significantly different (same letter) and significantly different (different letter) samples. Analyses were performed using the one-way ANOVA with post hoc test tool as implemented by Navendu Vasavada (https://astatsa.com/, accessed on 9 December 2022). All statistical analysis results are detailed in the corresponding source data sheets (see Supplementary Materials File S3).

Conclusions
The physiological role of extracellular carbonic anhydrases in cyanobacteria remains rather elusive. Sodalinema gerasimenkoae CahB1 represents an intriguing case study, as its activity may be crucial to efficient carbon uptake within alkaliphilic biofilms, but its mode of secretion and any functional compensation for its absence from the carboxysome are hard to reconcile given its genomic context and its demonstrated functionality as a carboxysomal β-CA in Synechocystis. Its close homology to and functional interchangeability with Synechocystis CcaA may hint at a relatively recent and subtle evolutionary divergence that allows for its re-localization from the carboxysome to the outer cell membrane. The exact mechanism underlying this evolutionary shift and its correlation with respect to the carboxysome function remain to be deciphered. This study provides the first stepping stone to elucidating the origin of this highly atypical and-to our knowledge-singular functional extracellular β-CA in cyanobacteria.