Plant Carbonic Anhydrases: Structures, Locations, Evolution, and Physiological Roles

1Carbonic anhydrases (CAs) play essential roles in all photosynthetic organisms. In cyanobacteria, CAs located in the carboxysome are required for the conversion of accumulated HCO3− to CO2 for fixation by ribulose 1,5-bisphosphate carboxylase/oxygenase (Rubisco). Loss of this carboxysomal CA results in abnormal growth of cyanobacteria when they are grown at ambient levels of CO2. In Chlamydomonas reinhardtii, a thylakoidal CA is necessary for the functioning of the CO2 concentrating mechanism (CCM), and in the diatom, Pheodactylum tricorutum, a CA in the pyrenoid is required for the CCM of this species. In these organisms, the conversion of HCO3− to CO2 for Rubisco is needed in a very specific location in the cell. In addition, the CCM of C4 plants requires CA activity specifically in the mesophyll (M) cell cytosol. While the correct inter- and intracellular location of CAs is essential for efficient physiological functioning of photosynthetic organisms, it is also important that CA activity is not present in certain organelles or cell types. demonstrated this when they transformed cyanobacteria with a gene encoding a human CA. This CA was expressed in the cytoplasm of the Synechocystis cells, effectively short circuiting the CCM. Similarly, a defective C4 CCM resulted when a cytosolic CA was expressed in bundle-sheath (BS) cells of the C4 plant Flaveria bidentis. Recently, a number of research initiatives have been working to improve photosynthesis in plants by introducing CCM components from cyanobacteria, algae, or C4 plants into terrestrial C3 plants. While the introduction of active transporters and enzymes is required for these initiatives to work, it is also necessary to know where the endogenous CAs are active within the recipient plant, as introducing CA activity in the wrong location could short circuit attempts to improve photosynthesis.

2This review focus on what is known about the genes encoding CA, and the locations of the CA isoforms in both C3 and C4 plants. Up-to-date research on the physiological roles of the different CA isoforms is also covered, as well as our current understanding of the molecular changes that were responsible for the evolution of the genes encoding C4-associated CAs from their ancestral C3 orthologs.

3All CAs are zinc metalloenzymes that catalyze the interconversion of CO2 and HCO3−. The enzymes are ubiquitous in nature and are an example of convergent evolution, as multiple, structurally and sequentially distinct families of CA have been discovered. Plants have three types of CA: α-, β-, and γ-type CAs. The α-type CA (αCA) was first found in erythrocytes and was the first CA family discovered. The majority of the enzyme is composed of 10 β strands that create a large central β sheet, which is surrounded by seven α helices on the periphery of the protein (Figure 1A; ). The zinc at the αCA active site is coordinated by three His residues and one water molecule organized in a tetrahedral conformation, and is located in the central part of the protein, at the bottom of a cone-shaped crevice. While most αCAs are monomers, multimeric αCAs have been discovered as well as αCAs containing extra domains. However, even in multimeric αCAs, the zinc ion is always coordinated by His residues from a single polypeptide.

4The β-type CA (βCA) was first discovered in plants, and its protein sequence and structure are very different from that of the αCAs. In βCAs, the zinc ion is coordinated by two Cys residues, one His residue, and a water molecule (Figure 1B; ). The structure of a βCA monomer is mostly composed of α helices that surround a β sheet consisting of four parallel β strands. There is also a fifth, C-terminal β strand involved in the oligomerization of βCA. The functional unit of the βCA is a dimer, although the most common βCA oligomerization is a tetramer. The βCA dimer is formed via extensive interactions created by two N-terminal α helices of one monomer wrapping around the second monomer and by minor hydrogen bonding between the second β strand of each monomer. Tetramers are formed by interactions made primarily by the fifth, C-terminal β strand. In pea, the chloroplastic βCA forms an octamer. For some βCAs, dicots have a unique C-terminal extension of the fifth β strand, whereas monocots do not. Octamers are formed via slightly different interactions with these fifth β-strand extensions.

5The γ-type CA (γCA) was first discovered in archaea but has since been found in photosynthetic bacteria and in plants. The first crystal structure of γCA from Methanosarcina thermophila was reported by Kisker and colleagues in 1996 (Figure 1C). Much like the active site of αCA, the active site of γCA also contains a zinc atom coordinated by three His and a water molecule. However, unlike the structure of αCAs, which are monomers, the functional unit of γCA is a trimer, with three active sites spanning the monomer-monomer interfaces. The zinc ion is coordinated by His residues provided by two different subunits. A β-strand region dominates the structure of γCA and consists of seven complete turns creating a left-handed β helix. Each full turn contains three β strands making the β helix look like an equilateral triangle from the top view. In photosynthetic organisms, γCA may contain extra domains as seen in the cyanobacterial CcmM proteins of cyanobacteria that have two or three repeated C-terminal domains with high similarity to the small subunit of Rubisco. In cyanobacteria, CcmM sometimes functions as an active CA, but some CcmM proteins do not have activity. However, CcmM is thought to organize the packing of Rubisco in the carboxysome even when it does not have CA activity.

6Plants have a large number of genes encoding CA. There are 17 distinct genes in total encoding all α, β, and γ, isoforms in Arabidopsis, including two γ-like CAs (Table 1). A similar number of genes are present in the genomes of other plant species, including mosses, monocots, and dicots (Table 1). Since plants can be polyploids or paleopolyploids, having undergone genome duplication in the past, the number of CA genes may be much higher. For example, in soybean, a diploid plant thought to have undergone genome duplication relatively recently, the total number of genes coding for CA is in excess of 25. Genes encoding CA are expressed in almost all tissues of the plant and CA isoforms can be found in most intracellular compartments, such as chloroplasts, mitochondria, the plasma membrane, and the cytoplasm.

7It is likely that the numbers and types of CA are quite ancient in the plant lineage as the unicellular green alga Chlamydomonas reinhardtii also has multiple genes encoding α-, β- and γCAs. C. reinhardtii has three genes encoding αCA, six encoding βCA, and three encoding γCA and γCA-like proteins. The CA isoforms of C. reinhardtii are found throughout the algal cell: in the periplasmic space (cell wall), chloroplast, cytoplasm, and mitochondria. As seen in higher plants, the C. reinhardtii γCAs and γCA-like proteins are mitochondrial with evidence suggesting they are part of Complex I of the mitochondrial electron transport chain, and the βCAs are found in similar intracellular locations as in higher plants, with isoforms in the mitochondria, chloroplast, and cytoplasm. However, two of the C. reinhardtii βCAs have hydrophobic C-terminal extensions, this is not observed so far in terrestrial plants. In addition, in C. reinhardtii the αCAs seem to play different physiological roles. In terrestrial plant species, only β- and γCAs have been implicated in CCMs, whereas two αCAs play important roles in the CCM of C. reinhardtii: CAH1, which is located in the periplasmic space, and CAH3, which is found in the thylakoid lumen. For more information on Chlamydomonas CAs, see.

8αCAs are the largest CA gene family in most plants, but they are also the least studied. The scarcity of published work on αCAs is most likely because the proteins are not highly abundant in leaves and roots. In Arabidopsis, complete expressed sequence tags (ESTs) exist for only three of the eight αCA genes, and RNA-seq data also only poorly cover the other five annotated genes. Arabidopsis αCA8 is clearly a pseudogene as it encodes in-frame stop codons. Limited expression information for some αCA genes is available through RNA-seq data on genome sites. Interestingly, in sorghum, the αCA Sb5G039000 is expressed specifically in anthers (Figure 2; ), while in Medicago trunculata, the αCAs Mt1g059900 and Mt1g059940 are expressed in root nodules. In Arabidopsis, αCA2 is expressed in trichomes of the leaf. Clearly at least some αCA genes show quite specific organ or tissue expression patterns. To date, there are no reports of plants where one or more αCA genes have been disrupted.

9There is very little information on the intracellular location of αCA isoforms. Two reports suggest that Arabidopsis αCA1 is a chloroplastic protein. The targeting of this αCA1 is unusual as it moves through the endoplasmic reticulum and is glycosylated. However, the location of αCA1 in the chloroplast has not yet been confirmed by proteomic studies. Possibly the amount of αCA in leaves is low, or perhaps the glycosylation obscures its detection. There are no reports on the subcellular location or function of αCAs from other plants at this time.

10Genes encoding γCAs and γCA-like proteins have been found in all plants. In fact, every species appears to have at least two genes encoding γCAs and at least one encoding a γCA-like protein (Table 1). For example, Arabidopsis has three γCA genes and two genes encoding γCA-like proteins. γCAs are well conserved in photosynthetic organisms, from green algae, to mosses, monocots, and dicots. While no higher plant γCA with CA activity has been identified, the proteins have the active-site residues found in γCAs from archaebacteria and cyanobacteria. In contrast, the γCA-like proteins do not have the required Zn coordinating amino acid residues. While γCAs are encoded by the nucleus, they are mitochondrial proteins. They have been shown to be part of the mitochondrial Complex I (NADH-ubiquinone oxidoreductase), and make up an extrinsic domain known as the carbonic anhydrase domain of the oxidoreductase, which is composed of three subunits: two γCA subunits and one γCA-like subunit. The γCA and γCA-like proteins are part of nine plant-lineage-specific subunits.

11The expression level of genes coding for γCA and γCA-like isoforms is average or above in almost all tissues for which expression data are available (Figure 2). This is not surprising for a subunit of Complex I as the mitochondrial electron transport chain is found in most plant tissues and cell types. If either AtγCA1 or AtγCA2 is knocked out, there is a small reduction in Complex I; however, if both AtγCA1 and AtγCA2 are knocked out, the plant is profoundly and adversely affected. The γca1γca2 mutants lack Complex I altogether, and do not produce viable seed, having to be maintained using an embryo rescue method, which involves supplying the embryos with sucrose in the growth medium. The double mutants also exhibit high levels of Complexes II and IV (succinate dehydrogenase and cytochrome oxidase, respectively), and the alternative oxidase, and in contrast, reduced levels of photosynthetic proteins.

12Plants have a moderate number of βCA genes, usually between four and seven (Table 1). There have been a number of studies on the βCAs as they are highly expressed in leaf tissue (Figure 2). βCAs have been found in chloroplasts, mitochondria, the cytosol, and the plasma membrane of Arabidopsis (Table 2), and in the cytosol and chloroplasts of many plants. When the predicted amino acid sequences of βCAs from monocots and dicots are aligned, the isoforms can be divided roughly into three groups (Figure 3). One group was found in all monocots and dicots considered (Figure 3) and is represented by AtβCA5 and AtβCA6, which localize to the chloroplast and mitochondria, respectively. The second group was found only in dicots, and the Arabidopsis AtβCA1, AtβCA2, AtβCA3, and AtβCA4 proteins are in this group. All dicots examined had at least two βCA proteins in this group (Figure 3). The third group contains only monocot CAs, and these proteins are known to localize to the chloroplast and cytosol. The length of the C termini of the proteins was an indicator for the group in which an isoform clusters. The monocot-specific proteins were the shortest, with the dicot-specific isoforms about 10 amino acids longer and the AtβCA5/AtβCA6-related proteins, found in all species examined, about 20 amino acids longer (Figure 3).

13As the βCAs have been the most intensely studied family of plant CAs, the remainder of this review focuses on this group.

14Alternative splicing can result in a single gene coding for multiple proteins that may show tissue-specific expression patterns and/or be targeted to different organelles of the cell. Deposited ESTs in TAIR, as well as RNA-seq data, indicate that transcription of the AtβCA1 gene may result in two different mRNAs (Figure 4). The RNA variants arise from the splicing of the ninth and tenth exons, where one variant has all 10 exons, and the other has an extended ninth exon, making proteins that differ slightly at their C termini. Two different transcription start sites for AtβCA2 (Figure 4) are predicted to encode two AtβCA2 isoforms with different N termini, resulting in the two proteins having different projected destinations in the plant cell. AtβCA4 is another example of a gene that can produce multiple mRNA forms (Figure 4; ), due to different transcription start sites (Figure 4). The shorter AtβCA4 mRNA lacks the first two exons, encoding a different N terminus relative to the longer form. Interestingly, RNA-seq data for AtβCA4 indicate that the shorter mRNA has a unique first exon that is not present in the longer transcript and is expressed in both roots and leaves. In contrast, the longer AtβCA4 mRNA is expressed specifically in leaves of Arabidopsis.

15In Neurachne munroi leaves, four βCA transcripts are expressed that are derived from two genes by way of alternative splicing (Figure 4; ). For each gene, the alternatively spliced transcripts encode polypeptides that share the same active-site residues but have distinct N termini, thereby influencing the subcellular location of each isoform. In N. munroi, the CA1 gene encodes NmCA1a and NmCA1b. Exon 2 is not present in NmCA1a transcripts, such that in CA1a and CA1b, the initiating AUG and N-terminal regions are encoded in different frames (Figure 4; ). In the case of the NmCA2 gene, NmCA2a transcripts do not contain exons 2 and 3, while exon 1 is not present in the NmCA2b transcript, again resulting in each splice form encoding a distinct N-terminal region (Figure 4; ). The same gene-transcript relationship was observed in all other Neurachne species for which transcript and genomic DNA sequences have been obtained (data not shown). While experimental evidence for alternative splicing of other monocot βCA genes has not been reported, parsing of EST databases and comparison with genome sequences suggest that alternative splice forms do exist in other species.

16Linking CA to photosynthesis has been a major focus in the plant CA field. Studies using reverse transcription (RT)–PCR, microarrays, and RNA-seq have shown that all six βCA genes are expressed in leaves of Arabidopsis; however, subcellular localization studies using green fluorescent protein (GFP) fusion constructs showed that not all of the proteins localize to the chloroplast. AtβCA1 was shown to have a long chloroplast transit peptide over 100 amino acid residues in length, and was later confirmed to be located in the chloroplast via GFP studies (Table 2;, ). AtβCA2 and AtβCA3 are cytosolic βCAs (Table 2;, ) although AtβCA2 is expressed at a much higher level than AtβCA3 in Arabidopsis leaves. The long form of AtβCA4, AtβCA4.1, localizes to the plasma membrane while the short form, AtβCA4.2, is cytosolic, as it no longer has a secretory transit peptide (Table 2;, ). Studies using GFP fusions also showed that AtβCA5 localizes to the chloroplast, and AtβCA6 is a mitochondrial form of the enzyme (Table 2;, ).

17To date, three transcripts encoding distinct βCA isoforms (CA1, CA2, and CA3) have been found in leaves of Flaveria species. Chloroplast import assays showed in both F. bidentis (a C4 species) and F. pringlei (a C3 species) that CA1 was imported into chloroplasts, while CA2 was not (Table 2;, ). However, the subcellular location of CA3 was not conserved between the two species; FpCA3 was found to be chloroplast targeted, while FbCA3 was not (Table 2;, ). Comparison of the predicted amino acid sequences of the CA3 polypeptides revealed that FbCA3 lacks 71 amino acids at the N terminus, including the chloroplast transit peptide, when compared with FpCA3, a situation that was proposed to be important to the molecular evolution of C4 photosynthesis in F. bidentis.

18The only monocot species so far in which the subcellular location of βCA isoforms has been experimentally determined are from the genus Neurachne. GFP fusion constructs indicated that N. munroi CA2b is imported into tobacco chloroplasts or mitochondria, while NmCA1a, NmCA1b, and NmCA2a localize to the cytosol (Table 2; ). As in Flaveria, the location of the isoforms is not conserved between C3 and C4 Neurachne species. While the GFP localization experiments showed that NaCA1b from the C3 species N. alopecuroidea is cytosolic as predicted, NaCA1a is imported into the chloroplasts, in contrast with the cytosolic location of CA1a from N. munroi (Table 2; ). Analyses of the predicted proteins indicated that 11 amino acids present in the N-terminal region of the NaCA1a polypeptide, but absent in NmCA1a, is important for chloroplast targeting.

19Numerous algorithms exist for predicting protein subcellular location. Predicted βCA amino acid sequences from various monocot and dicot species were analyzed using four algorithms: Predotar, ChloroP, TargetP, and MultiLoc (Supplemental Table 1;,,, ). The results indicate that each species contains at least one isoform that is chloroplast targeted and at least one isoform that is likely cytosolic, and that many also contain a mitochondrial βCA (Supplemental Table 1). Comparison with actual experimental results for Arabidopsis, Flaveria, and Neurachne (Table 2) indicates that these predictions are often but not always correct (cf. Table 2 and Supplemental Table 1).

20The above results indicate that all predictions must be experimentally tested, keeping in mind that the proteins resulting from different splice forms of a single gene may have different subcellular locations (Table 2;, ). In addition, predicting βCA subcellular location based on the products of orthologous genes in closely related species may not assist in assigning location since it has been shown that locations are not conserved between closely related C3 and C4 species.

21To date, the specific expression patterns of all identified βCA isoforms have been reported in the scientific literature for only three species: Arabidopsis thaliana, F. bidentis, and F. pringlei. An in-depth CA expression study of Arabidopsis rosette leaves found that AtβCA1, AtβCA2, and AtβCA4 are the most highly expressed CA genes in M cells, and AtβCA1, AtβCA4, and AtβCA6 are the most highly expressed CA genes in guard cells. Many studies also report CA gene expression in Arabidopsis roots, with RNA-seq data showing that AtβCA2 and AtβCA3 are the two lowest expressed βCA genes in Arabidopsis roots, whereas AtβCA4 and AtβCA5 are the two most highly expressed genes. Microarray analyses indicated all AtβCA genes are expressed in roots of Arabidopsis, albeit AtβCA1, AtβCA2, and AtβCA3 are expressed at very low levels. Results of an RT–PCR analysis showed that transcripts from most of the AtβCAs genes are found in Arabidopsis roots with the exception of AtβCA2, whereas another RT–PCR experiment found that AtβCA3 and AtβCA6 show the highest expression in roots. Both RT–PCR studies found that Arabidopsis genes encoding βCA are expressed in stems and floral tissues, and all six AtβCA genes are expressed in Arabidopsis stem and floral tissues according to microarray analyses. Interestingly, the microarray data indicate that all six AtβCA genes are expressed in Arabidopsis seeds although, with the exception of AtβCA5 and AtβCA6, their expression diminishes as the seeds develop.

22In F. pringlei, RT–quantitative (q)PCR assays indicated that transcripts encoding FpCA1 and FpCA3 were primarily expressed in leaves, whereas FpCA2 was expressed in leaves, roots, and flowers. By contrast, transcripts encoding CA1, CA2, and CA3 were detected in F. bidentis leaves, roots, and flowers, but FbCA1 and FbCA3 transcripts were most abundant in leaves, whereas FbCA2 mRNA levels were consistent among the three tissues.

23While numerous reports have shown that the majority of CA activity in leaves of C4 species is in the M cells, the location of multiple βCA isoforms within a leaf has not been comprehensively examined in any species. In F. bidentis, immunocytochemical experiments showed that CA is expressed predominantly in M cells and is undetectable in BS cells. Presumably this is the FbCA3isoform previously shown to be essential for C4 photosynthesis. The tissue/cell type-specific expression patterns of FbCA1 and FbCA2 were not examined.

24Recently, advances in laser-capture microdissection and next-generation sequencing have enabled M and BS cell transcriptomes to be obtained. Comparative transcriptome analyses of leaf M and BS cells of Panicum virgatum, Setaria viridis, Gynandropsis gynandra (formerly Cleome gynandra; ), and Zea mays showed that particular βCA transcripts are enriched in M cells. Specifically, Pavir.J08788 and Pavir.J05107 transcripts are approximately three times more abundant in M cells than in BS cells in P. virgatum, and a similar fold difference is also observed for two transcripts encoding βCA in G. gynandra, GgβCA1 and GgβCA2 (At3g01500 and At5g14740 orthologs; ), one transcript in S. viridis (Si03061m.g ortholog; ), and one in Z. mays (GRMZM2G414528;, ). However, in the latter two species, there are several βCA transcripts that show much higher M-specific abundance, with at least a 20-fold enrichment compared with BS cells. These include Si003885m.g in S. viridis, and GRMZM2G121878, GRMZM2G348512, and GRMZM2G094165 in Z. mays. Interestingly, the maize GRMZM2G145101 transcript is the only βCA transcript identified so far that shows higher abundance in BS cells than M cells, while no βCA mRNAs from S. viridis, G. gynandra, and P. virgatum show this pattern. Taken together, the current findings suggest that within a C4 species, genes encoding the different βCA isoforms show different tissue- and cell-specific expression patterns, with some isoforms showing preferential expression in the leaf M cells.

25Total leaf CA activity within herbaceous dicotyledonous plants ranges from 2- to 10-fold (,,,,,, supplementary material), whereas the leaves of some monocotyledons reportedly contain 1000 times more CA activity than other monocot species (,,,,,,, supplementary material; ). Total leaf CA activity in C3 monocots can be 500 times higher than that of C4 monocots (,,,,, supplementary material), while leaves of herbaceous C4 dicots demonstrate total CA activities that fall within the range of values for C3 dicot leaves (,,,,, supplementary material).

26Very few studies have looked at CA activity in both isolated M and BS cells from C4 plant leaves; however, depending on the comparison being made, this is important. As it is the cytosolic CA in M cells that is associated with the C4 CCM, total leaf CA activity measurements may be misleading. Two forms of CA were isolated from Amaranthus cruentus leaves. One form was found associated with the chloroplasts of the BS cells and was responsible for 8% of total leaf CA activity. In contrast, the other form was found in the M cell cytoplasmic fraction, where it represented 62% of the total CA activity in amaranth leaves. In another C4 dicot, Flaveria bidentis, BS cell CA activity was found to contribute 0.5% of total leaf CA activity.

27also found low CA activity in BS cells of species representing the three C4 subtypes: NADP-malic enzyme (NADP-ME), NAD-malic enzyme (NAD-ME), and phosphoenolpyruvate carboxykinase (PCK). In the two NADP-ME-subtype species examined, sorghum and maize, BS cell CA activity was 1.8% and 1.6%, respectively, of total leaf CA activity. The activity of CA in the BS of NAD-ME-type (P. miliaceum and Atriplex spongiosa) and PCK-type species (Urochloa panicoides and Chloris gayana) was even lower, representing just 0.5%–0.8% of total leaf CA activity. These results suggested low CA activity in the BS was a requisite for efficient functioning of the C4 pathway. This idea was later supported by a transgenic approach in which wild-type (WT) plants of the C4 species F. bidentis were transformed with the sequence encoding mature tobacco CA (i.e., no chloroplast transit peptide) that was under the control of a constitutive promoter. This allowed tobacco CA expression in the cytosol of all cells, including leaf BS cells. The transformants showed increased BS leakiness to inorganic carbon (Ci), reduced rates of photosynthesis, and an impaired CCM. Together these results support the idea that in C4 plants demonstrating Kranz leaf anatomy, the strict M and BS cell compartmentalization of CA is essential for the proper functioning of the C4 CCM.

28Progress has been made in our understanding of cis elements and chromatin marks that control the preferential accumulation of transcripts encoding βCA in M cells of C4 species; however, the associated trans-acting factors remain elusive. Studies suggest several regulatory mechanisms were already present in ancestral C3 genes coding for βCA but were modified through recruitment of posttranscriptional pathways or the binding of different transcription factors during the evolution of C4 photosynthesis.

29In leaves of the C4 species Gynandropsis gynandra, transcripts encoding the homolog of the Arabidopsis plasma-membrane-associated CA (AtβCA4;,, ), showed abundances similar to the levels of mRNAs coding for other C4-associated proteins. Elements in either the 5′-untranslated region (UTR) or 3′-UTR of the G. gynandra gene encoding βCA4 were found to be sufficient for M-cell-specific expression using GUS fusion constructs. Similar sequences in the 5′- and 3′-UTRs of AtβCA4 were also shown to independently direct M-cell-specific expression when they were used to transform G. gynandra.

30A more recent study showed the 3′-UTR of a second G. gynandra CA gene, GgCA2, for which high transcript levels are found in leaves, and the homologous region from AtβCA2 also direct preferential accumulation of GUS in M cells. A common, nine-nucleotide motif in these CA2 3′-UTRs, as well as in the 5′- and 3′-UTRs of AtβCA4 and GgCA4, were identified as sufficient to direct M cell-specific expression and was designated MEM2 for mesophyll expression module 2. This study also showed that the MEM2 element does not control the level of GgCA4 gene expression but instead works post-transcriptionally through a mechanism that increases the amount of CA4 protein made in M cells relative to the BS. The high levels of GgCA4 transcripts in G. gynandra M cells appear to result from the loss of elements in the promoter region and introns from the ancestral CA4 gene that repress its expression in C3 species.

31Epigenetic marks have been identified that contribute to M cell-specific expression of genes encoding CA isoforms important in C4 photosynthesis. Trimethylation of the Lys residue at position 4 on histone H3 (H3K4me3) is associated with transcriptionally active genes and is enriched in the 5′-region of the transcribed sequence. found the gene encoding one of the C4-associated CAs in maize, GRMZM2G121878, showed a high ratio of H3K4me3 to the dimethylated form (H3K4me2) in M cells. This is consistent with the methylation state of histone H3K4 found at analogous positions in genes coding for the C4-associated forms of PEPC and pyruvate phosphodikinase (PPDK), which also show preferential accumulation of transcripts in M cells.

32The total number of genes encoding CA is similar in dicots and monocots, and in plants using C3 and C4 photosynthesis (Table 1; ), with some of these genes encoding isoforms that likely perform the same function in all species. However, as highlighted above, differences in total CA activity and control of βCA expression patterns have been detected within and between these plant groups, and are responsible for the very specific physiological roles exhibited by some of the enzymes.

33In the leaves of C3 plants, the majority of CA activity localizes to M cell chloroplasts, where the enzyme can make up 1%–2% of total leaf protein. Although a major component of the C3 leaf proteome, the actual role of βCAs in C3 photosynthesis remains ambiguous (Figure 5). Initial suggestions included the conversion of HCO3− to CO2 to ensure maximum rates of fixation by Rubisco, facilitating the diffusion of CO2 across the chloroplast membranes, buffering short-term changes in pH in the chloroplast stroma induced by changing light conditions, and the hydration of compounds other than CO2. However, to date, none of these proposed functions has strong empirical support.

34To determine the physiological function of C3 plant chloroplast βCA, WT tobacco plants were transformed with antisense constructs directed against transcripts encoding the tobacco chloroplastic βCA isoform. Primary transformants with 1%–2% of the CA activity of WT plants showed no significant differences in CO2 assimilation rates, Rubisco activity, and chlorophyll content relative to WT plants.

35More recently, Arabidopsis antisense transformants and knockout lines of βCA1 were examined, and these plants did demonstrate an obvious phenotype. Both lines of transformants showed reduced seedling survival that could be rescued by including sucrose in the growth medium, or growing the seedlings in elevated CO2. The cotyledons were found to have compromised CO2 assimilation rates that resulted in the observed reduced seedling establishment before development of the first true leaves (Figure 5). However, when the transformants did survive, the mature plants showed no phenotypic differences from WT plants, strongly suggesting that AtβCA1 plays no direct role in photosynthesis of mature Arabidopsis plants.

36In contrast to C3 plants, most βCA activity in C4 plants is found in the cytosol of M cells, where it catalyzes the first reaction in the C4 CCM (Figure 5; ), the conversion of atmospheric CO2 to HCO3−.

37Transgenic approaches have been used to test the suggestion that only enough CA is present in the M cytosol of C4 plants to not limit photosynthesis. In one study, WT plants of the C4 dicot F. bidentis were transformed with an antisense construct directed against transcripts encoding the C4-associated CA3. A decrease in CO2 assimilation rates was seen only when the transformants contained less than 20% of WT CA activity, and transformants exhibiting less than 10% of WT activity had very reduced rates of photosynthesis, about 8% of WT plants, and required a high CO2 environment to survive. In addition, this study also showed that the hydration rate of CO2 was about 58 times the photosynthetic rate. Taken together, these results indicate that CA is not limiting photosynthesis in F. bidentis; however, the CO2 response curves of the transformants indicated that a cytosolic CA is essential for an efficient C4 CCM in this dicot species.

38Maize plants carrying mutations in genes encoding two isoforms of CA that have been correlated with C4 photosynthesis (ZmCa1 and ZmCa2; ) demonstrated that CA is not limiting for growth in this C4 monocot species. Unlike the F. bidentis CA3 antisense plants, however, both the ca1 single mutant and the ca1ca2 double mutant, which contained 3% of WT maize CA activity, showed no impairment in CO2 assimilation at ambient levels of CO2. It was not until the concentration of CO2 was sub-ambient that a decrease in CO2 assimilation was detected. It was concluded from gas exchange and carbon isotope data, and CA and PEPC activities, that the ca1 mutant contains only enough CA activity to supply PEPC with HCO3−, while the activity in the double mutant is below this level, and the plants rely, at least to some extent, on the uncatalyzed conversion of CO2 to HCO3−. Clearly, while CA in both WT maize and F. bidentis is not rate limiting for photosynthesis, and is necessary for efficient operation of the C4 pathway when CO2 availability to the leaf is limited, differences exist between these two C4 species with respect to the levels of CO2 that result in impaired CO2 assimilation. The basis for this discrepancy may be structural, enzymatic, or a combination of mechanisms.

39A mitochondrial βCA along with the γ and γ-like CAs associated with the mitochondrial Complex I are proposed to be part of a mechanism found in all plants that facilitates the fixation of mitochondrial respiratory CO2 in the chloroplasts (Figure 5; ). Relative to WT plants, Arabidopsis mutants lacking the mitochondrial AtβCA6 gene show a decrease in leaf area and overall biomass, inhibition of growth at low CO2, and a significant increase in respiration rates. In contrast, the overexpression of AtβCA6 in Arabidopsis resulted in larger plants with higher shoot fresh and dry weights, and decreased rates of respiration compared with WT plants. Interestingly, there appeared to be no significant difference in photosynthetic rates although the CO2 compensation point of the knockout lines was reportedly increased relative to WT values. From this work, the authors suggested that increasing expression levels of mitochondrial AtβCA6 affect cellular respiration, which impacts positively on biomass production.

40The finding that acetazolamide inhibited photosystem II (PS II) activity raised the possibility that CA activity was associated with PSII. Later, presented evidence that included the finding of low levels of CA activity with thylakoid preparations and even core PSII fractions. The discovery of CrCAH3, in the thylakoid lumen of C. reinhardtii, resulted in two different hypotheses as to its physiological role. In one proposal, CAH3 functions in light-driven generation of CO2 from accumulated HCO3−, taking advantage of the low pH of the thylakoid lumen to drive the reaction toward CO2 formation. A competing hypothesis was that CrCAH3 was required on the oxidizing side of PSII. Since CrCAH3 is an αCA, this hypothesis presented an attractive physiological role for αCAs in plants. However, evidence over the past 10–15 years strongly argues against a role for CA in PSII, with probably the most persuasive argument being the lack of CA in any of the crystal structures of PSII to date.

41Most cyanobacteria do not have an αCA, and for most cyanobacteria, the only CA in the cell is in the carboxysome as part of the CCM. In Arabidopsis, total chloroplast (intact chloroplasts) proteome studies indicate the only CAs in the chloroplasts are βCA1, βCA2, and βCA5. The other CA reported to be in the chloroplast, αCA1, has not been detected in proteome studies to date, and no other αCA has been found in chloroplasts. None of the Arabidopsis β-type CAs (AtβCA1, AtβCA2 and AtβCA5) have a leader sequence consistent with a thylakoid lumen location, and only the stromal AtβCA1 is present at high levels in chloroplasts of photosynthetically active cells. Finally, and found no CA activity in highly active PSII preparations using the very sensitive membrane inlet mass spectrometry assay. They convincingly argued that any CA activity associated with PSII was due to contamination. This is not surprising as stromal CA activity is extremely high and even a relatively low level of contamination by this protein could result in measurable activity in enriched PSII preparations.

42Increased transcript abundances are found for the genes encoding AtβCA1 and AtβCA4 in Arabidopsis guard cells ( and references therein). While single Atβca1 and Atβca4 T-DNA mutants demonstrated no CO2-sensitive phenotype compared with WT plants, the double ca1ca4 mutant showed impaired stomatal conductance in response to changing CO2 concentration as well as higher stomatal numbers and density. Consequently AtβCA1 and AtβCA4 were implicated in guard cell movement through a role in the early steps of the CO2 signaling pathway, and were suggested to function also in guard cell development. Recent work using reconstituted systems in Xenopus oocytes has suggested a model in which AtβCA4 functions alongside the aquaporin PIP2;1 at the guard cell plasma membrane, influencing intracellular CO2/HCO3− levels, which when elevated, enhance S-type anion channel activity and stomatal closure (Figure 5; ). As yet, no direct interactions between AtβCA4 and PIP2;1 have been reported.

43The Arabidopsis ca1ca4 mutants show an inverted response to CO2 relative to WT plants in that, at high CO2, they have increased stomatal numbers in cotyledons and mature leaves. Taking these results into account, as well as the earlier characterization of the double mutant, a preliminary model for the control of stomatal development has been constructed and involves an extracellular signaling pathway mediated by CA. Not all the components or steps in the model have been identified; however, it has been proposed that CA activity is necessary for the increased expression of the genes encoding the epidermal patterning factor EPF2, and the CO2-inducible protease that cleaves it, facilitating its binding to the receptor kinase ERECTA, which has been implicated in the regulation of stomatal development (Figure 5; ).

44Chloroplastic βCAs from C3 plants are part of a defense mechanism that is induced upon attack by various pathogens (Figure 5;,,,, ). In tobacco and Arabidopsis, the CAs have been identified as salicylic-acid-binding proteins that function in an antioxidant role during viral infections. Recombinant inbred lines of Arabidopsis with resistance to the insect herbivore, Plutella xylostella, had at least a 2-fold increase in abundance of AtβCA1 and AtβCA4 proteins.

45Salinity induces an increase in βCA transcript abundance in maize, and it was suggested that this response paralleled the antioxidant role seen with the biotic stressors described above (Figure 5; ). Both salinity and an osmotic stress treatment using polyethylene glycol led to an increase in rice seedling total CA enzyme activity, and the level of mRNA coding for a predicted chloroplastic CA isoform. Overexpression of the rice CA in Arabidopsis led to improved growth on media containing salt compared with WT Arabidopsis.

46Cytosolic CAs have been implicated in affecting amino acid biosynthesis levels (Figure 5; ). While PEPC, the primary carboxylase of C4 plants, uses HCO3− produced by cytosolic βCA activity to form C4 acids, as part of the C4 CCM, in C3 plants an estimated 50% of the free aspartate pool is created by PEPC activity. Arabidopsis double knockout mutants of the Atβca2 and Atβca4 genes, which code for cytosolic CAs, showed reduced growth rates and chlorosis of the younger leaves relative to WT plants when grown at 200 μL L−1 CO2. This phenotype was ameliorated when the plants were grown under high levels (1000 μL L−1) of CO2. The Atβca2ca4 double mutants also demonstrated reduced levels of aspartate, and a concomitant increase in glycine and serine levels. The low CO2 growth phenotype and amino acid profile could be mitigated by complementation of the double mutant with the AtβCA2 gene. The elevated amounts of glycine and serine in the double mutant were unanticipated and hints CA activity affecting other biochemical pathways.

47The nitrogen-fixing root nodules of numerous legumes contain relatively high CA activity, and transcripts encoding βCAs have been isolated from the nodules of several species. The location of these transcripts and the proteins they encode changes during maturation of the nodules, and this suggests that the role of the enzymes likely varies over the course of nodule development. The functions put forward involve the provision of HCO3− for processes such as amino acid and lipid biosynthesis and gluconeogenesis in the early developmental stages, and the release of CO2 generated from bacteroid respiration to the rhizosphere in mature nodules (Figure 5;, ). However, these have not been supported experimentally. The presence of additional forms of CA in nitrogen-fixing nodules complicates the identification of the precise role(s) of the βCA enzymes (,,; supplemental data;,, ).

48Fatty acid synthesis is a primary metabolic pathway in which acetyl-CoA carboxylase (ACC) uses HCO3− to carboxylate acetyl-CoA to produce malonyl-CoA, the building block of fatty acid chains. In plants, the production of acyl chains takes place in the chloroplast while their utilization occurs in essentially every cellular compartment. Since ACC requires HCO3−, and previous results have shown significant expression and activity of CA in cotton seedlings, examined the level of radiolabeled acetate incorporation into lipids in cottonseed embryos and tobacco cell suspensions. When embryos and suspension culture cells were incubated with [14C]acetate in the presence of the CA inhibitor, ethoxyzolamide, the rates of lipid synthesis were greatly decreased. Antisense tobacco lines with 5% of WT CA activity also showed lower levels of radiolabeled lipids, which is consistent with the transgenic plants demonstrating reduced rates of Ci entering the chloroplast. suggested that CA activity traps Ci within chloroplasts in the form of HCO3−, which is then used by ACC in fatty acid synthesis (Figure 5).

49Plants have many genes encoding α-, β- and γ-type CAs, which are found in most tissues and many intracellular compartments. In addition, alternative splicing and multiple transcription start sites have been shown in a number of βCAs, often leading to different proteins targeted to different organelles. Programs developed to predict protein targeting should be used with caution, particularly when working with monocot CAs or CAs predicted from gene models. The number of CA genes is relatively similar in monocot and dicots, and in plants using C3 or C4 photosynthesis, or CAM. Evidence is building that, during the evolution of the C4 pathway, C3 genes coding for CA were co-opted through changes in cis-regulatory sequences, modification of posttranscriptional controls, and/or recruitment of different transcription factors. CAs, while clearly important in photosynthesis, are also required for other metabolic pathways as well as signaling and developmental pathways.

50This work was supported by the University of Illinois as part of the Bill & Melinda Gates Foundation-funded Realizing Increased Photosynthetic Efficiency (RIPE) consortium, by NSF grant IOS-1146597 to J.V.M. and the Australian Research Council Discovery Projects DP130102243 and DP150101037 to M.L.

51R.J.D., H.C., A.M., M.L., and J.V.M. all contributed to the writing of the original draft and to the reviewing and editing of the manuscript.