Transport and Use of Bicarbonate in Plants: Current Knowledge and Challenges Ahead

Bicarbonate plays a fundamental role in the cell pH status in all organisms. In autotrophs, HCO3− may further contribute to carbon concentration mechanisms (CCM). This is especially relevant in the CO2-poor habitats of cyanobacteria, aquatic microalgae, and macrophytes. Photosynthesis of terrestrial plants can also benefit from CCM as evidenced by the evolution of C4 and Crassulacean Acid Metabolism (CAM). The presence of HCO3− in all organisms leads to more questions regarding the mechanisms of uptake and membrane transport in these different biological systems. This review aims to provide an overview of the transport and metabolic processes related to HCO3− in microalgae, macroalgae, seagrasses, and terrestrial plants. HCO3− transport in cyanobacteria and human cells is much better documented and is included for comparison. We further comment on the metabolic roles of HCO3− in plants by focusing on the diversity and functions of carbonic anhydrases and PEP carboxylases as well as on the signaling role of CO2/HCO3− in stomatal guard cells. Plant responses to excess soil HCO3− is briefly addressed. In conclusion, there are still considerable gaps in our knowledge of HCO3− uptake and transport in plants that hamper the development of breeding strategies for both more efficient CCM and better HCO3− tolerance in crop plants.


Introduction
Life on Earth is based on the photosynthetic transformation of inorganic carbon (C inorg ) and water into energy-rich organic carbon (C org ) compounds. In turn, these are oxidized by heterotrophs to obtain cellular energy, releasing again C inorg in the form of CO 2 into the atmosphere. Atmospheric CO 2 is the main form of C inorg assimilated by the terrestrial photosynthetic organisms. Dissolution of CO 2 in water provides carbonic acid, which dissociates into bicarbonate (HCO 3 − ) and carbonate (CO 3 2− ).

Figure 1.
Mechanisms of HCO3 − transport by solute carriers (SLC) in humans and mammals drawn with information from [37,40].
(ii) SbtA is a low Cinorg-inducible, high affinity (K0.5 2-5 µ M), plasma membrane HCO3 − transporter that uses Na + as a driving ion with a half saturation constant around 1 mM for this ion [47]. Although initially considered a single unit-type transporter, it has a bigger complex size, which suggests that, in its functional form in the plasma membrane, this transporter is a tetramer [48]. It has been suggested that SbtA is activated by a serine-threonine protein kinase [49] that also depends on Na + [50]. SbtA homologs seem to be present in many ß-cyanobacteria although this has only been confirmed in Synechocystis PCC6803 [47] and Synechococcus PCC7002 [50].
(iii) BicA HCO3 − transporters are also dependent on Na + . Their affinity for HCO3 − transport with a Km ranging from 74 µ M to 353 µ M (1.7 mM for Na + ) is lower than that of SbtA. Nonetheless, BicA is able to maintain a high flow of Cinorg for photosynthesis. BicA transporters are expressed at low levels under conditions of high CO2 but they are highly inducible under low CO2. They have been discovered in the coastal marine cyanobacterium Synechococcus PCC7002 [51] and they are present in both α-cyanobacteria and ß-cyanobacteria. BicA transporters belong to the large family of prokaryotic and eukaryotic transporters often described as sulphate, SulP family transporters. The C-terminus includes a hydrophilic STAS domain (see also Section 2.1) involved in the allosteric regulation that has also been found in A. thaliana sulfate transporters [52].
(iv) NDH-I4 is a constitutive protein complex located in the plasma membrane that accelerates CO2 uptake. The passive entry of CO2 is followed by the conversion (NDH-I mediated) to HCO3 − [53,54].
(v) NDH-I3 is a second, complex, low CO2 -inducible system involved in CO2 uptake located in the thylakoid membrane. It works in a similar manner to NDH-I4 [53,54].

Cinorg Transport in Microalgae
C. reindhardtii uptake of Cinorg has been associated with the activity of an ATP-binding cassette transporter, HLA3, and the homolog of a formate-nitrite transporter LCIA that is also called NAR1.2 [30]. HLA3 is located in the plasma membrane and LCIA in the chloroplast envelope. The absence of LCIA decreases the amount of HLA3 mRNA, which indicates a regulation by the chloroplast-encoded LCIA of the expression of HL3 encoded in the nuclear genome [31]. While the HCO3 − transport mechanism of HL3 seems to be clear, LCIA has been proposed to be an HCO3 − channel [22,30,31]. If so, HCO3 − ions would be transported through such a channel downhill and could not accumulate HCO3 − over the equilibrium prediction. However, the addition of mM concentrations of HCO3 − to

C inorg Transporters in Cyanobacteria
Five modes of C inorg transport have been described in cyanobacteria. (i) BCT1 is an inducible high affinity (K 0.5 for HCO 3 − ≈ 15 µM) transporter located in the plasma membrane that belongs to the ATP binding cassette (ABC) transporter family [42] although transport energization by ATP consumption has not been proven [43]. BCT1 is a multi-meric complex composed by four subunits. CmpA is located in the periplasmic space and binds HCO 3 − with a very low K 0.5 of 5 µM [44] and also binds Ca 2+ as a cofactor [45]. CmpB is a dimer within the plasma membrane and CmpC and CmpD are extrinsic proteins that share binding sites for ATP. CmpC has an extra domain involved in the allosteric regulation of BCT1 similar to the NrtC protein of NRT1 transporter. In this later case, the domain of the NrtC protein is involved in the inhibition of transport in the presence of NH 4 + [46]. BCT1 is found in ß-cyanobacteria but absent in marine cyanobacteria. However, it is present in the α-cyanobacteria Synechococcus WH5701, which can live in a wide range of C inorg concentrations and salinities [43].
(ii) SbtA is a low C inorg -inducible, high affinity (K 0.5 2-5 µM), plasma membrane HCO 3 − transporter that uses Na + as a driving ion with a half saturation constant around 1 mM for this ion [47]. Although initially considered a single unit-type transporter, it has a bigger complex size, which suggests that, in its functional form in the plasma membrane, this transporter is a tetramer [48]. It has been suggested that SbtA is activated by a serine-threonine protein kinase [49] that also depends on Na + [50]. SbtA homologs seem to be present in many ß-cyanobacteria although this has only been confirmed in Synechocystis PCC6803 [47] and Synechococcus PCC7002 [50].
(iii) BicA HCO 3 − transporters are also dependent on Na + . Their affinity for HCO 3 − transport with a K m ranging from 74 µM to 353 µM (1.7 mM for Na + ) is lower than that of SbtA. Nonetheless, BicA is able to maintain a high flow of C inorg for photosynthesis. BicA transporters are expressed at low levels under conditions of high CO 2 but they are highly inducible under low CO 2 . They have been discovered in the coastal marine cyanobacterium Synechococcus PCC7002 [51] and they are present in both α-cyanobacteria and ß-cyanobacteria. BicA transporters belong to the large family of prokaryotic and eukaryotic transporters often described as sulphate, SulP family transporters. The C-terminus includes a hydrophilic STAS domain (see also Section 2.1) involved in the allosteric regulation that has also been found in A. thaliana sulfate transporters [52].
(iv) NDH-I4 is a constitutive protein complex located in the plasma membrane that accelerates CO 2 uptake. The passive entry of CO 2 is followed by the conversion (NDH-I mediated) to HCO 3 − [53,54].
(v) NDH-I3 is a second, complex, low CO 2 -inducible system involved in CO 2 uptake located in the thylakoid membrane. It works in a similar manner to NDH-I4 [53,54].

C inorg Transport in Microalgae
C. reindhardtii uptake of C inorg has been associated with the activity of an ATP-binding cassette transporter, HLA3, and the homolog of a formate-nitrite transporter LCIA that is also called NAR1.2 [30]. HLA3 is located in the plasma membrane and LCIA in the chloroplast envelope. The absence of LCIA decreases the amount of HLA3 mRNA, which indicates a regulation by the chloroplast-encoded LCIA of the expression of HL3 encoded in the nuclear genome [31]. While the HCO 3 − transport mechanism of HL3 seems to be clear, LCIA has been proposed to be an HCO 3 − channel [22,30,31]. If so, HCO 3 − ions would be transported through such a channel downhill and could not accumulate HCO 3 − over the equilibrium prediction. However, the addition of mM concentrations of HCO 3 − to Xenopus laevis oocytes expressing NAR1.2 evokes a membrane depolarization as does the addition of mM concentrations of NO 2 − , which suggests an HCO 3 − transport into the chloroplast by H + symport instead of the transport through a channel. This mechanism would also be consistent with the need to overcome the electrochemical gradient for HCO 3 − in the stroma relative to the cytosol [55]. The ycf10 is also related to C inorg transport. Disruption of the plastid ycf10 inhibits the C inorg accumulation in the chloroplast. Its gene product known as the protein CemA was originally proposed as a C inorg transporter, but its similarities with the cyanobacterial PxcA involved in Na + -dependent H + extrusion suggest that CemA may play a similar role in the energization of the chloroplast envelope [55]. The HCO 3 − uphill transport through the plasma membrane and the chloroplast envelope agrees with the early observation of a vanadate sensitive C inorg transport at both levels. A second C inorg transporter proposed for the plasma membrane in C. reindhardtii is LCI1 [56,57]. The overexpression of this protein increases the affinity for C inorg and enhances HCO 3 − uptake. The protein is encoded by an orphan gene [55] and does not have any known functional motif. The proteins CCP1/2 have also been proposed to take part in C inorg uptake by the chloroplasts. They show similarities with the mitochondrial carrier proteins superfamily, but knock-outs of CCP1/2 do not show defects in photosynthesis [58]. Thus the specific role of CCP1/2 proteins in C inorg transport has yet to be clarified. The active uptake of HCO 3 − was described for natural populations of marine phytoplankton dominated by large diatoms [59]. However, the HCO 3 − transport mechanisms at the molecular level have been studied in the model diatom species Phaeodactylum tricornutum and Thalassiosira pseudonana [60,61]. In P. tricornutum, ten putative HCO 3 − transporters have been identified. They are similar to the unrelated SLC4 and SLC26 mammalian protein families (see Section 2.1). SLC4 has been characterized as a HCO 3 − transporter in the plasma membrane of P. tricornutum and SLC4 homologs have also been found in T. pseudonana [25]. Photosynthesis in the diatom species is sensitive to 4,4 -diisothiocyanatostilbene-2, 2 -disulfonic acid (DIDS), which is an inhibitor of anion exchange, and depends on the presence of Na + in the medium (K 0.5 28 mM, saturation at 100 mM Na + ). This suggests the existence of an HCO 3 − uptake mechanism based on Na + symport or on a Na + dependent Cl − /HCO 3 − anti-port [25]. A different group of SLC4 transporters located in the chloroplast envelope have been proposed for transporting HCO 3 − to the chloroplast stroma [25,60,61].
The active transport rate of dissolved inorganic carbon through the chloroplast envelope is ten-fold that of HCO 3 − transport across the plasmalemma [54]. However, further investigations are required to elucidate the molecular identity of the protein and the transport mechanism in the context of the complex four-layer chloroplast envelope of diatoms [62].
In micro-algal species, genetic tools are still not available and HCO 3 − uptake has been revealed by physiological methods that include the photosynthetic sensitivity to inhibitors of external CA, pH buffers to dissipate electrochemical H + gradients, and inhibitors of anion exchangers. Therefore, a direct entry of HCO 3 − has been proposed for the marine eustigmatophycean Nannochloropsis gaditana [63][64][65]. The absence of external CA and the sensitivity to DIDS suggest an anion exchange mechanism for HCO 3 − transport. A DIDS and 4-acetamido-4 -isothiocyanato-stilbene-2, 2 -disulfonic acid (SITS) sensitive photosynthesis has been described in Eminliania huxleyi [66]. SITS is the putative inhibitor of the anion exchanger 1 (AE1), which works as a Cl − /HCO 3 − antiporter in red blood cells [67] (see Figure 1). In contrast, a DIDS/SITS insensitive HCO 3 − transport has been described for Dunaliella tertiolecta [68].

C inorg Transport in Macroalgae
One of the first examples for the use of HCO 3 − in macro-algae was described in the giant inter-nodal cells of Characeae living in alkaline media [69]. The active efflux of H + through the putative H + -ATPase causes a local acidification of the apoplast in about two pH units [69]. The presence of CA activity in the acidic zones accelerates the conversion of HCO 3 − to CO 2 that diffuses across the plasmalemma [70]. The cytosolic pH homeostasis requires the presence of alkaline areas between the acid zones, which produces the spectacular banding observed in these organisms under the light [71]. An alternative mechanism for HCO 3 − use and hence for banding was given by Lucas et al. [72].  [22,73]. The use of HCO 3 − as a source of inorganic carbon for photosynthesis has been described for the majority of marine macro-algae and seagrasses [74][75][76]. The most common mechanism of HCO 3 − use is the apoplastic conversion to CO 2 , which is shown in Condrus chrispus [77], Porphyra leucosticta [78], a series of red macroalgae [79], and Phyllariopsis puspurascens [80]. More information is available in References [33,75,81,82]. Alternatively, other algal species have been described as direct HCO 3 − users. Most of the evidence for a direct uptake of HCO 3 − ions comes from experiments in which the inhibitors of anion exchanger, mainly DIDS and SITS, are used to inhibit HCO 3 − transport and, therefore, photosynthesis [82]. Larsson and Axelson [83] examined 11 green, 5 red, and 11 brown macro algae. Photosynthesis was DIDS-sensitive only in Chaetomorpha, Monostroma, and ulvaceans (Ulva and Enteromorpha), but not in the rest of green, red, or brown algae tested. More information is available in Reference [84]. Fernández et al. [26] show a DIDS-sensitive anion exchanger as the main mechanism for HCO 3 − uptake in the giant kelp Macrocystis pyrifera. DIDS-sensitivity has also been reported in the red algae Eucheuma denticulatum [85] while a residual DIDS-sensitive photosynthetic activity was found in Gracilaria gaditana [86]. Calculations made from photosynthetic conductance were used to suggest direct HCO 3 − uptake in Laurencia pinnatifida [87].

C inorg Transport in Seagrasses
Seagrasses have been described as HCO 3 − users [82,[88][89][90][91]. Based on the lack of photosynthesis inhibition in seagrasses by DIDS and SITS, Larkum et al. [91] hold that HCO 3 − influx through anion exchangers does not take place in the leaves of seagrasses. These substances inhibit AE1 that are present in algae but in angiosperms (including marine) DIDS and SITS have been described as inhibitors of anion channels [29,92] that may have a distinct role in plasma membrane anion transport [87].
This mechanism is proposed for plants that show AZ-sensitive TRIS-insensitive photosynthesis, but the ubiquitous presence of plasmalemma H + -ATPase cannot be ignored [99]. Second, the catalyzed apoplastic dehydration of HCO 3 − to CO 2 in acid regions generated by the activity of the H + -ATPases.
This mechanism would be sensitive to AZ and TRIS. Third, the direct uptake of HCO 3 − ions by symport with H + . In this case, the electrochemical gradient for H + generated by the activity of the plasmalemma H + -ATPases drives the direct HCO 3 − transport. This mechanism would be AZ-insensitive and TRIS-sensitive. The two first mechanisms involve apoplastic accumulation of OH − and CO 2 diffusion across the plasmalemma and the third one implies accumulation of HCO 3 − and likely OH − in the cytosol (see Table 1). In contrast to humans (see Figure 1) and cyanobacteria (Section 2.2), no Na + -dependent HCO 3 − uptake system has been reported in plants. The only example for a Na + -driven ion transport system is the high affinity transporter for NO 3 − and P i in the seagrass Zostera marina [100].
In that case the electrochemical gradient for Na + is maintained because of very low membrane permeability for Na + and the action of a Na + /H + antiporter, which is similar to the SOS1 present in terrestrial vascular plants [101]. Table 1. C inorg uptake mechanisms proposed for several seagrass species based on their photosynthetic sensitivity to TRIS and AZ. Question mark (?) denotes that the mechanism is partially supported by available evidences.
The availability of the genome of Zostera marina [103] allows the in silico search for genes potentially involved in C inorg transport. Using the web application Phytozome (http://www.phytozome.net), which is a comparative platform for green plant genomics [104], we searched for genes with homologies with the HLA3 transporter of C. reinhardtii and SLC4 transporter of P. tricornutum. The search for homologies in the genome of Z. marina with ChHLA3 sequence results in six genes with high homology, all of them listed as ABC transporters. In contrast, the search for homologies with PhSLC4 results in five sequences of medium-to-low homologies with genes encoding boron transporters and anion exchangers. The public availability of the genome of seagrasses will be a valuable tool for the future investigation of the exact molecular identities of C inorg transporters, cellular location, mechanism, kinetic properties, and regulation.

C inorg Transport in Higher Land Plants
While HCO 3 − transporters are already quite well-characterized in cyanobacteria, algae, and mammals, the information on higher vascular land plants is scarce. Seven loci of genes coding for transporters of the HCO 3 − family are listed in the gene databases of the genetically well-characterized A. thaliana. The best studied protein is BOR1. This protein belongs to the solute carrier family type SLC4 and presents homology to SLC4A1, which is the band 3 transporter highly abundant in erythrocytes. As SLC4A1, BOR1 has a gate and a core domain and acts with an elevator mechanism. However, BOR1 has an inward rotated core domain providing an occluded state, which suggests that it may undergo structural transitions allowing access from either side of the membrane [105]. Bicarbonate transport by Band 3 is a unidirectional pathway out of the erythrocyte. A further substantial difference is that BOR1 is an efflux-type borate transporter responsible for root-to-shoot transport of this essential plant nutrient. BOR1 is located in the xylem parenchyma cells and loads borate into the xylem, which is then transported to shoots by the transpiration stream [106]. The other six genes code for BOR2 to BOR7 [107]. All seem to be involved in the transport of borate or boric acid rather than in HCO 3 − transport.
Although no selective HCO 3 − transporters or channels have so far been characterized in higher land plants, the possibility of membrane transport by specific or unspecific anion transporting proteins cannot be excluded. Several studies provide indirect support for HCO 3 − uptake by plant roots. Under exposure to high HCO 3 − (5 mM to 20 mM), a strong inhibition of nitrate, sulphate, and phosphate uptake by roots has been observed [108]. Such inhibition could be caused, at least in part, by competition between HCO 3 − and other anions for transport mechanisms with low anion specificity. An electrophysiological approach to ion selectivity of a voltage-dependent anion channel in A. thaliana hypocotyls revealed low but reproducible HCO 3 − currents. A permeability ranking of [28]. More recently, such a channel with permeability for several anions has been identified as QUAC1/ALMT12, which is a channel that releases anions from guard cells [109]. In fact, anion channel currents in plants have mainly been studied in guard cells where they contribute to the mechanisms for controlling stomatal resistance (see Section 3.3). Slow Anion Channels (SLACs) and Quick Anion Channels (QUAC) are involved in the transport of NO 3 − and Cl − (SLAC1), NO 3 − (SLAH3), or malate (QUAC1/ALMT1) [110].
Anion channels in roots are less characterized. In A. thaliana roots, a slah3-1 mediated Cl − and possibly NO 3 − efflux in response to ABA has been shown [112]. Recently, Canales et al. [113] reported comparative root expression profiles at a cell resolution level for anion channels in A. thaliana. Nitrate channel SLAH3 was strongly expressed in the mature root zone while the Voltage Dependent Anion Channel (VDAC1) was localized to the meristem zone. VDACs 2 and 4 have been reported expressed in all plant organs [114]. At the subcellular level, VDACs are localized at the outer mitochondrial membrane and in small vesicles located in the cell periphery [115]. The ion selectivity of VDACs depends on ionic strength. Higher selectivity for Cl − is achieved with lower ionic strength [116]. It has been stated that, on limestone soils, HCO 3 − can passively enter into plant roots. Then it is long-distant transported via xylem vessels to the leaves where, after transformation by CA anhydrase, the resulting CO 2 can be assimilated along with the atmospheric CO 2 [11]. The apoplastic, passive radial transport pathway in the roots is disrupted at the endodermal level due to the hydrophobic Casparian strip. Therefore, to reach the vascular cylinder, a substance has to first pass through the plasma membrane into the symplasm. This implies either a still unidentified HCO 3 − membrane transport system or the conversion of HCO 3 − into CO 2 , which may easily diffuse into the stele.
Apoplastic by flow, either through the young root tips where the Casparian strip has still not fully developed or at sites where lateral root emergence from the pericycle disrupts this hydrophobic barrier, may be another way HCO 3 − enters the stele. Contribution of this apoplastic bypass is relatively small in the case of NaCl [117] or Cd [118]. We could not find specific data for HCO 3 − .
Early investigations using 11 C or 14 C isotopes as markers for HCO 3 − provided evidence for uptake of HCO 3 − by roots and transport to the shoots [119][120][121][122]. However, the contribution of C inorg taken up by roots may be less than 1% taken up by leaves [123]. The 14 C from labelled H 14 CO 3 − supplied through the roots was found to be incorporated into sugar, starch, and proteins of leaves [124]. As plants can acquire C inorg from different sources including atmospheric CO 2 and respiratory CO 2 , the experimental design is critical. Solution pH used for supplying labelled HCO 3 − to the plants deserves special attention. At pH 8, most of the labelled C inorg is in the form of HCO 3 − but a small percentage of labelled CO 2 can be present and CO 2 diffusion into the root cells may occur, which will be followed by transformation of this CO 2 into labelled HCO 3 − by CA. This transformation can be even more relevant considering that the pH of cell walls and xylem sap are usually in the acid range. The pH of the leaf apoplast of sunflowers remained stable around 6.4 to 6.5 even if roots were exposed to 10 mM HCO 3 − [125]. However, it has to be taken into account that apoplast alkalinization is a general response to stress in plants [126]. Enhanced Cl − supply under stress causes alkalization of the root apoplast due to the symport of 2 H + per 1 Cl − [127]. Increasing the external pH of the root bathing solution also increases pH of both A. thaliana root cell walls [128] and xylem sap [129]. This favors HCO 3 − over CO 2 formation. Nonetheless, even under severe stress conditions, such as drought or fungal infection with a strong alkalinization effect in the apoplast, the increased pH values remain nearly neutral [130]. Therefore, in plants with their roots exposed to HCO 3 − , the proportion of HCO 3 − over CO 2 during radial transport of C inorg from soil to the stele and within the xylem sap up to the leaves may be considerably lower than in the soil solution surrounding the plant roots. The direct use of root-derived HCO 3 − by CA in chloroplasts to supply CO 2 for Rubisco is unlikely when taking into account the low chloroplast permeability of HCO 3 − (1 × 10 −8 m s −1 ) in comparison to CO 2 (range from 2.3 × 10 −4 to 8 × 10 −4 m s −1 ), which was recently shown by mass inlet mass spectrometry (MIMS) using 18 O labelled C inorg [131].

Formation and Use of Bicarbonate in Plants
As seen in higher plants, no selective HCO 3 − transporter or channel has been characterized at the molecular level. Membrane transport of HCO 3 − in these organisms is still unclear. Contrastingly, the contribution of HCO 3 − to essential metabolic pathways and the total assimilation of C inorg in  [132]. Malonyl-CoA is the precursor for fatty acid formation and elongation. Moreover, it participates in the biosynthesis of ring A of flavonoids through the polycetic pathway and in the biosynthesis of malonylated aminocyclopropane-1-carboxylic acid (MACC), which is involved in the down-regulation of ethylene production in plants (see Figure 2). Other biotin-containing carboxylases operating with HCO 3 − are 3-methylcrotonyl-CoA carboxylase, which is involved in the mitochondrial pathway of leucine catabolism. Geranyl-CoA carboxylase likely works in the metabolism of cyclic terpenes [133,134].

Formation and Use of Bicarbonate in Plants
As seen in higher plants, no selective HCO3 − transporter or channel has been characterized at the molecular level. Membrane transport of HCO3 − in these organisms is still unclear. Contrastingly, the contribution of HCO3 − to essential metabolic pathways and the total assimilation of Cinorg in plants is reliably documented.
The CA-generated HCO3 − serves as a substrate for different carboxylases among others acetyl-CoA carboxylase (ACCase, EC.6.4.1.2) and phosphoenolpyruvate carboxylase (PEPC, EC 4.1.1.31). ACCase contains a biotin carboxylase, a biotin carboxyl carrier protein, and a carboxyl transferase. It catalyzes the carboxylation of acetyl-CoA to malonyl-CoA in the chloroplast and the cytosol [132]. Malonyl-CoA is the precursor for fatty acid formation and elongation. Moreover, it participates in the biosynthesis of ring A of flavonoids through the polycetic pathway and in the biosynthesis of malonylated aminocyclopropane-1-carboxylic acid (MACC), which is involved in the downregulation of ethylene production in plants (see Figure 2). Other biotin-containing carboxylases operating with HCO3 − are 3-methylcrotonyl-CoA carboxylase, which is involved in the mitochondrial pathway of leucine catabolism. Geranyl-CoA carboxylase likely works in the metabolism of cyclic terpenes [133,134].
PEPC plays a major role in the carbon assimilation processes in plants. This enzyme in the presence of Mg 2+ or Mn 2+ ions catalyzes the -carboxylation of phosphoenolpyruvate (PEP) yielding oxalacetate (OAA) and inorganic phosphate (Pi) in an irreversible reaction (see Figure 2).  PEPC plays a major role in the carbon assimilation processes in plants. This enzyme in the presence of Mg 2+ or Mn 2+ ions catalyzes the β-carboxylation of phosphoenolpyruvate (PEP) yielding oxalacetate (OAA) and inorganic phosphate (P i ) in an irreversible reaction (see Figure 2).
The relative importance of the contribution of HCO 3 − to the total plant C org as well as the assimilation mechanisms and their consequences for plant adaptation to different environmental conditions depend on the plant species and the characteristics of the habitat (see Sections 3.2 and 4). In all cases, the cooperation of the two enzymes, CA and, in higher plants, PEPC, is essential.

Plant Carbonic Anhydrase and Phosphoenolpyruvate Carboxylase
Carbonic anhydrases (CAs, EC 4.2.1.1) are metallo-enzymes that catalyze the reversible hydration of CO 2 forming HCO 3 − . Zinc is the required metal at the catalytic site for CA activity. Some exceptions are several coastal diatoms with cadmium-containing CA (CDCA). The Cd 2+ at the catalytic site is fully exchangeable for Zn 2+ [135]. The natural use of Cd 2+ in this ζ-CA class enzyme is considered an evolutionary adaptation to low Zn 2+ availability in marine habitats [136]. CA enzymes are ubiquitous in nature (animals, plants, archaebacteria, and eubacteria) and are an example of convergent evolution. Based on sequence comparison, CA proteins are grouped into seven distinct classes: α, β, γ, δ, ζ, η, and θ-CAs [137][138][139][140]. In higher land plants, only α, β, γ CAs are found. The δ and ζ classes are restricted to marine diatoms and η-CA so far has only been reported in Plasmodium falsiparum [141]. θ-CA seems more widely distributed in algae and cyanobacteria [142] and it has been reported critical for photosynthesis in the diatom Phaedactylum tricornutum [143]. The ubiquity of the distribution of CAs implies that they play diverse and essential roles in many biological processes. They have been related to respiration and transport of CO 2 /HCO 3 − between tissues, pH and CO 2 homeostasis, electrolyte secretion in a variety of tissues/organs, various biosynthetic reactions, and CO 2 fixation [142,144]. In addition, CA is a plausible source of hydrogen sulphide (H 2 S) within plant leaves by catalyzing the conversion of carbonyl sulphide (COS) to CO 2 and H 2 S [145].
Higher plants contain three evolutionarily distinct CA families including αCAs, βCAs, and γ CAs where each family is represented by multiple isoforms in all species [142,146,147]. Alternative splicing of CA transcripts is common. Consequently, the number of functional CA isoforms in a species may exceed the number of genes [147]. CAs are expressed in numerous plant tissues and in different cellular locations. The most prevalent CAs are those in the chloroplast, cytosol, and mitochondria. CAs have been found in the thylakoid lumen of Chlamydomonas and Phaeodactylum. They are an important component of the CCM in these species and, therefore, essential for photosynthesis and growth [143,148]. This diversity in location is paralleled in the many physiological and biochemical roles that CAs play in plants [142,147,149,150]. As in humans and animals, many of these roles are related to the CA-driven regulation of cell pH, which, in turn, can participate in multiple regulatory processes through electrical signals, changes in cytosolic Ca 2+ concentrations, and plant hormones [150][151][152] among others.

Plant α-Carbonic Anhydrases (αCA)
Arabidopsis thaliana contain eight αCA (AtαCA1-8) [153]. Genes for αCa1, αCA2, and αCA3 are expressed in green and reproductive tissue (stems, rosette leaves, caulinar leaves, and flowers). Only αCA2 presents root expression. While expression of αCA1 is independent of the level of CO 2 , the expressions of αCA2 and αCA3 are induced under conditions of low CO 2 concentrations [149]. αCa1 is expressed in chloroplasts and αCA2 is expressed in the plasma membrane. α-CA4 is implicated in the processes leading to energy dissipation in the PSII antenna [154]. Arabidopsis αCA8 is clearly a pseudogene since it encodes in-frame stop codons [147]. Tissue-specific expression has also been reported for other species. In sorghum, the αCA Sb5G039000 is expressed specifically in anthers while, in the legume species Medicago truncatula αCAs Mt1g059900 and Mt1g059940, are expressed in root nodules [147]. There is increasing evidence that αCAs can play an important role in photosynthesis [150]. Under conditions of increasing light intensity, the expression of αCA2 decreases while the expression of αCA4 increases. Knock-out mutants of these chloroplast-located αCAs exhibit contrasting responses in comparison of the wild type. Both the quantum yield at photosystem 2 (PS2) and the electron transfer to O 2 decreased while non-photochemical quenching (NPQ) and CO 2 assimilation were enhanced in plants lacking αCA2. The opposite was observed in αCA4 knock-outs [155]. The authors hypothesize that these αCAs may participate in the regulation of H + flux into the PS2 protein PsbS, which regulates qE-type NPQ.

Plant β-Carbonic Anhydrases
βCAs are most abundant in land plants where they participate in photosynthesis [147]. Arabidopsis thaliana has six βCAs [147]. βCAs genes are highly expressed in leaf tissue. Expressed sequence tag experiments revealed that βCA1 to βCA6 are expressed in rosette leaves, caulinar leaves, and flowers.
βCA3 is also strongly expressed in reproductive tissue while βCA4, βCA5, and βCA6 are expressed in all tissues including roots. βCAs have been found in chloroplasts, mitochondria, the cytosol, and the plasma membrane [144,149]. Targeting analysis using green fluorescent protein fusion proteins confirmed the subcellular localization of plant βCAs: βCA1 and βCA5 are expressed in chloroplasts while βCA2 and βCA3 are cytosolic. Isoforms βCA4, βCA4.1, are localized in the plasma membrane while the short form, βCA4.2, is cytosolic. βCA5 and βCA6 are localized in the chloroplast and mitochondria, respectively [149].
The role of βCAs in photosynthesis of land plants seems especially relevant in grasses with C 4 -type photosynthesis [156] or for plants under limited C inorg supply (see Section 3.2). Carbonic anhydrases could be versatile. They may be involved not only in photosynthesis and responses to CO 2 and light but also in seed germination, morphogenesis, nodule development, and responses to abiotic stress [157,158]. The tobacco salicylic acid-binding protein 3 (SABP3) is a chloroplast βCA that exhibits antioxidant activity and plays a role in the hypersensitive defense response [159]. Furthermore, βCA1 is related to ethylene signaling responses, photosynthetic performance of cotyledons, and Arabidopsis seedling survival [160].

Plant γ-Carbonic Anhydrases
Plant γCAs are codified in the nucleus but localized in mitochondria [139]. So far, no higher plant γCA with CA activity has been identified. Nonetheless, plant proteins with the active-site residues found in γCAs from archaebacteria and cyanobacteria have been found. In A. thaliana, five γCArelated genes have been reported including three γCA genes and two genes encoding γCA-like proteins. In contrast to γCA proteins, the γCA-like proteins do not have the required Zn-coordinating amino acid residues. Plant γCA genes encode for a part of the mitochondrial Complex I (NADH-ubiquinone oxidoreductase). Complex I knock-out lines present adverse effects: non-viable seeds, high levels of mitochondrial Complexes II and IV, and the alternative oxidase. However, this is in contrast with reduced levels of photosynthetic proteins [161]. A proteomic approach has recently found enhanced γCA root levels during the induction phase of Al-tolerance in the hyper-resistant grass Urochloa decumbens. This increase occurred along with higher adenylate kinase activity and supports a role for γCA in the maintenance of ATP-production during the Al tolerance response [162].

Plant PEP Carboxylases
Phosphoenolpyruvate carboxylases (PEPC) are located in the cytosol and catalyze the β-carboxylation of PEP to oxaloacetate using HCO 3 − in an irreversible process. The OAA can then be reduced through NADH or NADPH-dependent malate dehydrogenase to malate in a reversible process. PEPCs are present in bacteria, algae, and plants. The typical plant PEPC (class 1 PEPC) has four identical subunits of 107 kDa. Multiple isoforms have been identified in leaves of C 3 , C 4 , and CAM plants [163][164][165]. In Sorghum bicolor, which is a plant with C 4 -type photosynthesis, five PEPC genes (PEPC1-5) have been identified. The plant PEPC is highly regulated. Phosphorylation through PEP carboxykinase (PEPCK) at the N-terminal phosphorylation domain [166] and allosteric regulation by glycine and glucose-6-P enhances the activity. Inhibition is achieved by both allosteric regulation especially by malate and by ubiquitination [167]. In addition to the typical plant-type PEPC, plants also contain a bacterial-type PEPC (BPEPC) of 118 kDa [153]. The BPEPC is highly expressed in floral tissues as well as in seeds and fruits. It has recently been shown that high BPEPC occurs in tissues that accumulate high malate concentrations [168]. There is a tight interaction between PTPC and BTPC, which forms the class 2 PEP. This is an enzyme complex that, in contrast to class 1 PEPC, is mostly insensitive to allosteric inhibition by high malate concentrations [169,170]. While class 1 PEPCs are constitutively expressed in the cytosol, the BPEPC is associated with the outer mitochondrial surface. This location is in line with a central role of this enzyme in collaboration with CA in the efficient fixation of respiratory CO 2 and the anaplerotic supply of organic acids to the Krebs cycle [171]. This is especially relevant in developing seeds that store fatty acids such as castor bean seeds [172]. PEPC activity also plays a central role in symbiotic N 2 fixation in root nodules (see Figure 2) where it provides OAA for nitrogen assimilation and malate for the bacteroids [173].

Carbon Concentration Mechanisms (CCM) in Terrestrial Plants
Under certain environmental conditions, CO 2 may become a limiting factor for photosynthesis not only in cyanobacteria, algae, and aquatic macrophytes where CCMs have been intensively studied but also in terrestrial higher plants. High-temperature favoring photorespiration and drought imposes an increase of stomatal resistance. These are the main factors limiting CO 2 availability for RuBisCo in land plants [174].
Long distance transport of HCO 3 − from roots to leaves usually makes only a small contribution to C inorg for photosynthesis (see Section 2.6). Exceptions are aquatic plants in the Lycophyta genus Isoetes and the non-stomatous land plant Stylites. They acquire all C inorg for photosynthesis from the soil through the roots and recycle carbon by CAM [175]. Other terrestrial plants take up most of the C inorg in the form of CO 2 through stomata of the leaves. This CO 2 diffuses into the chloroplast where it is assimilated by RuBisCO, which forms phopshoglycerate (PGA) as the first stable product of C inorg assimilation. After activation with ATP and reduction by NADPH provided by the light-driven chloroplastic electron transport, PGA forms phosphoglyceraldehyde, which is the first sugar molecule of the photosynthetic carbon metabolism. Most terrestrial plants fix CO 2 directly onto ribuslose-bis-phosphate. In contrast to plants with this so-called C 3 -type photosynthesis, plants with C 4 -type photosynthesis convert CO 2 entering through the stomata into the mesophyll cells to HCO 3 − using a cytosolic βCA. This C inorg in the form of HCO 3 − is initially fixed by PEP carboxylase in the cytosol of the outer mesophyll cells of the leaves. In this case, OAA is a first stable organic compound.
Oxalacetate is either reduced to malate or transformed by transamination to aspartate. Malate or aspartate are the molecules that transfer the newly fixed carbon to the inner bundle sheet cells of the leaves where decarboxylation provides CO 2 , which is the substrate for Rubisco [176]. While CA activity is high in C 3 chloroplasts where it facilitates the availability of CO 2 for RuBisCo, the absence of CA activity from bundle sheet cells seems essential for the C 4 mechanism [177]. This CCM around RuBisCo in C 4 plants is considered an evolutionary adaptation to reduce the oxygenase activity of RuBisCo, which inhibits photorespiration and is especially enhanced under high temperature in tropical or subtropical areas [178]. However, C 4 -type photosynthesis can also be induced in certain amphibious plant species such as Eleocharius vivipara [179] under conditions of leaf emergence under dry conditions. Extreme adaptation to drought is observed in many CAM plants, which capture CO 2 during the night when a lower temperature and a higher relative humidity in the atmosphere reduces transpiratory water loss. During the dark period, this C inorg is fixed in the form of HCO 3 − by PEP carboxylase and stored in large vacuoles mostly in the form of malate. The CO 2 for fixation with RuBisCo is obtained by decarboxylation of malate during the following day-light period [180]. Limitations of CCMs in higher plants, especially of the C 3 -type of photosynthesis, and advances in our understanding of CCMs in cyanobacteria and microalgae like C. rheinhardii have promoted genetic engineering approaches to introduce efficient CCM into crop plants for increasing yield. Different approaches include manipulation of photorespiration, C 3 to C 4 engineering, and introduction of CCMs of cyanobacteria of C. rheinharddii into C 3 crops. This well-known topic has recently been reviewed in detail by Mackinder [181] who identified gaps in our knowledge on bicarbonate transporter structure, functioning, and localization as important constraints that need priority attention for successful development of CCM engineered plants.

CO 2 /Bicarbonate Signalling in Stomatal Guard Cells
Regardless the type of photosynthesis, C 3 , C 4 , or CAM, the CO 2 flux from the atmosphere into the plants is regulated by the stomatal opening and closure due to turgor changes in the stomatal guard cells. These changes are strictly controlled by multiple external and internal factors. Among those, the binomial CO 2 /HCO 3 − plays a central role (see Figure 3). High CO 2 promotes stomatal closure, which is brought about by the activation of efflux anion channels: SLAC1 (S-type) facilitating Cl − or NO 3 − efflux and R-type (AtALMT12/QUAC1 in A. thaliana) for malate efflux (see Section 2.6).
The signal for stomatal closing in response to high CO 2 seems to be a combination of alkaline pH, high Ca 2+ , and high HCO 3 − in the cytosol [182]. The carbonic anhydrase double mutant ca1ca4 does not show any effect on stomatal conductance when CO 2 concentration is changed from 100 ppm to 80 ppm [183]. This points to HCO 3 − being the key signal. Abscisic acid (ABA) dependent and ABA-independent mechanisms seem to operate in stomatal closure under a high amount of CO 2 (see Figure 3). OST1 (Open Stomata 1) is a positive regulator of the anion efflux channels. In the ABA-independent signaling pathway, a high amount of HCO 3 − activates a MATE-like transporter protein (RHC1, Resistance to High CO 2 ), which acts as a positive regulator of OST 1 by inhibiting HT1 (High Leaf Temperature) known as a protein kinase that inactivates OST1 [183,184].
Regardless the type of photosynthesis, C3, C4, or CAM, the CO2 flux from the atmosphere into the plants is regulated by the stomatal opening and closure due to turgor changes in the stomatal guard cells. These changes are strictly controlled by multiple external and internal factors. Among those, the binomial CO2/HCO3 − plays a central role (see Figure 3). High CO2 promotes stomatal closure, which is brought about by the activation of efflux anion channels: SLAC1 (S-type) facilitating Cl − or NO3 − efflux and R-type (AtALMT12/QUAC1 in A. thaliana) for malate efflux (see Section 2.6). The signal for stomatal closing in response to high CO2 seems to be a combination of alkaline pH, high Ca 2+ , and high HCO3 − in the cytosol [182]. The carbonic anhydrase double mutant ca1ca4 does not show any effect on stomatal conductance when CO2 concentration is changed from 100 ppm to 80 ppm [183]. This points to HCO3 − being the key signal. Abscisic acid (ABA) dependent and ABAindependent mechanisms seem to operate in stomatal closure under a high amount of CO2 (see Figure  3). OST1 (Open Stomata 1) is a positive regulator of the anion efflux channels. In the ABAindependent signaling pathway, a high amount of HCO3 − activates a MATE-like transporter protein (RHC1, Resistance to High CO2), which acts as a positive regulator of OST 1 by inhibiting HT1 (High Leaf Temperature) known as a protein kinase that inactivates OST1 [183,184].

Plant Response to Bicarbonate-Rich Soils
It is common knowledge that limestone soils containing high carbonate/bicarbonate concentrations restrict the performance of calcifuge plant species and limit yield especially in ironinefficient crops such as certain varieties of citrus, peach, pear, or soybeans suffering from limeinduced chlorosis [185,186]. Low pH leads to low availability of essential nutrients (especially Fe, Zn, and P) and high Ca soil concentrations are considered the main constraining factors. However, HCO3 − at concentrations occurring in the solution of limestone soils can inhibit root growth in sensitive plant species like the calcifuge grass Deschampsia caespitosa [187]. However, dicots like peas, beans, or sunflowers suffer more intense root growth inhibition due to CO2 and/or HCO3 − than the monocots barley and oats [123]. Recently soil carbonate has been identified as a main selection factor that drives local adaptation in natural populations of A. thaliana, which is a calcifuge species able to colonize soils with moderate carbonate contents [188]. to [183,184]

Plant Response to Bicarbonate-Rich Soils
It is common knowledge that limestone soils containing high carbonate/bicarbonate concentrations restrict the performance of calcifuge plant species and limit yield especially in iron-inefficient crops such as certain varieties of citrus, peach, pear, or soybeans suffering from lime-induced chlorosis [185,186]. Low pH leads to low availability of essential nutrients (especially Fe, Zn, and P) and high Ca soil concentrations are considered the main constraining factors. However, HCO 3 − at concentrations occurring in the solution of limestone soils can inhibit root growth in sensitive plant species like the calcifuge grass Deschampsia caespitosa [187]. However, dicots like peas, beans, or sunflowers suffer more intense root growth inhibition due to CO 2 and/or HCO 3 − than the monocots barley and oats [123]. Recently soil carbonate has been identified as a main selection factor that drives local adaptation in natural populations of A. thaliana, which is a calcifuge species able to colonize soils with moderate carbonate contents [188]. Currently, no HCO 3 − transporter has been characterized in higher plants (see Section 2.6).
Nonetheless, HCO 3 − -induced root growth inhibition is paralleled by enhanced root production of organic acids especially malate, succinate, and citrate [189]. This suggests that excess HCO 3 − enters the root and is metabolized by CA and PEP, which yields enhanced organic acid levels. However, CO 2 diffusing from the soil's atmosphere into the root can also be transformed inside the root into HCO 3 − by CA. Bicarbonate can be released again into the soil rhizosphere, which contributes to the plant's cation-anion balance. Especially under conditions of high nitrate uptake, enhanced HCO 3 − efflux from the roots has been claimed to contribute to the characteristic alkalinization of the rhizosphere when nitrate is the main N source for the plants [190,191]. In fact, in maize and a tomato, the sum of K + and NO 3 − uptake and the HCO 3 − efflux have been reported to be in electrical equilibrium [192]. However, no selective bicarbonate efflux transporters in plant roots have been reported and alkalinization can also be a consequence of either or both OH − release or H + uptake in cotransport with nitrate [193]. Actually, root supply of low HCO 3 − concentrations tended to increase rather than decrease root nitrate uptake in Populus canescens. Exposure to 1 mM external NaHCO 3 enhanced both nitrate reduction and assimilation as well as exported nitrogen to the shoots of poplar plants [194]. Higher HCO 3 − concentrations cause net K + and NO 3 − efflux as well as accumulation of organic acids, mainly malate, in the roots [195]. To what extent dark fixation of C inorg entering the roots plays a role in the carbon budget of terrestrial plants has been considered mainly in relation to lime-induced chlorosis in calcifuge plants. This type of chlorosis affects sensitive plant species when growing on carbonate-rich soil and may reflect an interference of HCO 3 − in the mechanisms of Fe acquisition and transport [196].
Key processes potentially impaired by HCO 3 − include the dicots' strategy 1 such as the acidification of the rhizosphere due to the strong buffer ability of HCO 3 − and the reduction of FeIII to FeII by ferric reductase, which operates optimally at acid pH [197]. The induction of root exudation of phenolic substances is not affected and is even stimulated by HCO 3 − . Induction of root accumulation and exudation of coumarin-type phenolics with high affinity for Fe has been reported as a response to Fe-deficiency under high pH conditions in A. thaliana [198]. An A. thaliana population which is naturally adapted to moderate soil carbonate had higher rates of coumarin root exudation than a sensitive population [188]. Furthermore, prevention of the imbalance of organic acid concentrations caused by dark fixation of HCO 3 − and shifting of C org into the shikimate pathway for the production of phenolic compounds has been reported as a mechanism of the extreme HCO 3 − tolerance in Parietaria difusa [199]. In the view of the multiple implications of HCO 3 − in plants' metabolism, breeding programs for better crop yield on carbonate-rich soil would greatly benefit from the characterization at the genetic and molecular level of the of bicarbonate uptake and efflux mechanisms in higher land plants.

Conclusions
During the last decade, there has been significant progress of our knowledge on the mechanisms of HCO 3 − transport and CCM in cyanobacteria, algae, and seagrass species due to improved genetic and molecular tools and electrophysiological approaches. In contrast, in higher land plants, no HCO 3 − transporter has been characterized so far. Advanced knowledge of the metabolic use of HCO 3 − in terrestrial plants has mainly been made in relation to C 4 and CAM metabolism including the genetic and molecular characterization of CAs and PEPC involved. However, there are still important gaps in our knowledge about the mechanisms of compartmentation and regulation especially regarding the complex interactions between light and dark fixation of C inorg , the recycling of respiratory and photorespiratory CO 2 , and the importance of anaplerotic supply of organic acids to the Krebs cycle. Filling these gaps is essential for progress in both genetic engineering approaches for transferring CCMs from cyanobacteria or microalgae to higher plants and breeding for bicarbonate tolerance in crops sensitive to lime-induced chlorosis.

Conflicts of Interest:
The authors declare no conflict of interest. The founding sponsors had no role in the design of the study, in the collection, analyses, or interpretation of data, in the writing of the manuscript, or in the decision to publish the results.