Mechanisms of carbon dioxide acquisition and CO₂ sensing in marine diatoms: a gateway to carbon metabolism

Abstract

Diatoms are one of the most successful marine eukaryotic algal groups, responsible for up to 20% of the annual global CO₂ fixation. The evolution of a CO₂-concentrating mechanism (CCM) allowed diatoms to overcome a number of serious constraints on photosynthesis in the marine environment, particularly low [CO₂]_aq in seawater relative to concentrations required by the CO₂ fixing enzyme, ribulose-1,5-bisphosphate carboxylase/oxygenase (RubisCO), which is partly due to the slow diffusion rate of CO₂ in water and a limited CO₂ formation rate from in seawater. Diatoms use two alternative strategies to take up dissolved inorganic carbon (DIC) from the environment: one primarily relies on the direct uptake of through plasma-membrane type solute carrier (SLC) 4 family transporters and the other is more reliant on passive diffusion of CO₂ formed by an external carbonic anhydrase (CA). Bicarbonate taken up into the cytoplasm is most likely then actively transported into the chloroplast stroma by SLC4-type transporters on the chloroplast membrane system. Bicarbonate in the stroma is converted into CO₂ only in close proximity to RubisCO preventing unnecessary CO₂ leakage. CAs play significant roles in mobilizing DIC as it is progressively moved towards the site of fixation. However, the evolutionary types and subcellular locations of CAs are not conserved between different diatoms, strongly suggesting that this DIC mobilization strategy likely evolved multiple times with different origins. By contrast, the recent discovery of the thylakoid luminal θ-CA indicates that the strategy to supply CO₂ to RubisCO in the pyrenoid may be very similar to that of green algae, and strongly suggests convergent coevolution in CCM function of the thylakoid lumen not only among diatoms but among eukaryotic algae in general. In this review, both experimental and corresponding theoretical models of the diatom CCMs are discussed.

This article is part of the themed issue ‘The peculiar carbon metabolism in diatoms’.

1. Introduction

The carbon-dioxide-concentrating mechanism (CCM) is a major evolutionary innovation for aquatic photosynthetic organisms that helps them to overcome constraints on acquisition of the photosynthetic substrate, CO₂. These systems are of particular relevance in the marine environment, where CO₂ availability is consistently low because of the aqueous chemistry of seawater, most notably its alkaline pH, high pH buffer capacity and high salinity. The solubility of CO₂ rarely exceeds 25 µM and the rate of dehydration of to CO₂ is slow. Carbon dioxide as high as 25 µM is generally below the concentration needed to support fixation by ribulose-1,5-bisphosphate carboxylase/oxygenase (RubisCO) in most marine microalgae [1–4].

Marine diatoms are major primary producers in the oceans and are responsible for up to 20% of annual global primary production [5,6]. A long history of physiological experiments showed that diatoms take up both CO₂ and and concentrate dissolved inorganic carbon (DIC) in their cells [7–12], strongly suggesting the occurrence of a CCM. However, only a select number of molecular components of diatom CCMs had been elucidated until very recently. The first breakthrough in identifying the molecular components of diatom CCMs was the isolation and characterization of two pyrenoidal β-carbonic anhydrases (CAs), PtCA1 and PtCA2, in the marine pennate diatom Phaeodactylum tricornutum [13–17]. Studies on CA in the context of the CCM had also revealed interesting new subsets of CAs in the marine centric species Thalassiosira weissflogii and Thalassiosira pseudonana, reinforcing the notion that CAs have evolved many times through convergent evolution and leading to new marine type CA classes denoted δ-CAs, ζ-CAs and θ-CAs [18–22]. In the mid to late 2000s, the sequencing of diatom genomes and development of molecular tools for diatoms including genetic transformation systems enabled more rapid identification of CCM components. CAs have been identified and localized in the model diatoms P. tricornutum and T. pseudonana, revealing a considerable diversity of origins and subcellular localizations of these CAs [15,23]. In contrast with such divergent features of diatom CCMs, convergent aspects in the function of the pyrenoid have recently begun to be recognized [22,24].

While molecular-level analysis of diatom CAs advanced, diatom DIC transport mechanisms were far less understood until recently. Nakajima et al. [25] reported the first discovery of a diatom DIC transport mechanism when they characterized a plasma membrane solute carrier (SLC) 4 type transporter, PtSLC4-2 in P. tricornutum which was shown to be a Na⁺-dependent transporter allowing this species to access the abundant extracellular in seawater. Orthologous SLC4-type transporters are found in many diatom genomes and may also be located in chloroplast membranes and probably function as DIC transporters.

CCMs are highly responsive to environmental conditions, especially CO₂ levels, and the regulatory mechanisms that mediate these responses are of great interest. In P. tricornutum, transcriptional level regulation of the nuclear-encoded pyrenoidal β-CA genes, Ptca1 and Ptca2, have been investigated extensively as representative genes of the diatom CCM. The role of the promoter regions of these genes (PPtca1 and PPtca2) in responding to changes in CO₂ levels has been studied using the β-glucuronidase (GUS) reporter assay. These studies showed that transcriptional regulation of PtCA1 and PtCA2 are governed by the second messenger cAMP via interactions of a basic-ZIP (bZIP) type transcription factor, PtbZIP11, with its counterpart cis-element in the promoter regions, which were denoted CO₂–cAMP-responsive elements (CCREs) [26,27]. Through transcriptomic analysis, CCRE-bZIP was implicated in CO₂-based regulation of the CCM and photorespiration in T. pseudonana [28], strongly suggesting CCRE-bZIP CO₂ response systems are a general feature in marine diatoms. Recent work on the CCRE-bZIP-based regulation system revealed that light intensity signals are also integrated into this signalling pathway [29,30], indicating a pivotal function of the cAMP-mediated CO₂ signal transduction system as a point of cross-talk between CO₂ and light signals.

Taking this molecular information on CCM components and their regulation into account, mathematical modelling of the diatom CCM and its dynamics are also being refined [16,31,32]. However, the functions of the four-layered chloroplast envelopes of diatoms and the pyrenoid are still largely unknown at the molecular level, knowledge which is essential for further refinement of diatom CCM models.

In this review, recent progress in understanding the diatom CCM is updated and perspectives on future directions of the field are discussed. Finally, it should be noted that some diatom species are reported to incorporate a C₄-like biochemical CCM system. For brevity, however, this review focuses on the biophysical CCM.

2. Membrane dissolved inorganic carbon transport systems

(a). Plasma membrane transport systems

The uptake of DIC across the plasma membrane represents the critical first step in supplying DIC for photosynthesis, and is especially challenging because environmental conditions are variable and in some cases unpredictable. Diatoms have been shown to actively take up and/or CO₂, which has been demonstrated using a range of physiological approaches [7–12,33–35]. However, until recently, the molecular mechanisms responsible for DIC uptake and internal transport, and the regulatory mechanisms controlling these fluxes were not known.

In a recent study, it was demonstrated that a plasma membrane-localized transporter in the marine diatom P. tricornutum homologous to the mammalian SLC4 family, PtSLC4-2, functions as a major uptake mechanism for the acquisition of extracellular DIC under low-CO₂ conditions [25] (figure 1). PtSLC4-2 specifically transports in the presence of a high concentration of sodium ions with a saturation level of about 100 mM Na⁺. The uptake rate of this protein reached its maximum level at pH 8.2, which is the typical pH of seawater [25]. Like PtSLC4-2, two other closely related putative transporters, PtSLC4-1 and PtSLC4-4, are induced specifically under low CO₂ conditions, suggesting that these transporters also make a significant contribution to influx into the cell in low CO₂ environments like seawater [25,31], although their localizations and functional details have yet to be determined.

Figure 1.

Localization of exogenously introduced PtSLC4-2:green fluorescent protein fusion in diatom cells. (a,b) Localization of PtSLC4-2:GFP fusion in P. tricornutum: (a) light image; (b) merged image of PtSLC4-2:GFP fusion (green), Hoechst-stained nucleus (blue) and auto-fluorescence of the chloroplast (red). (c–e) Z-stacked images with GFP signals (green) and chlorophyll auto-fluorescence. (d) The cross-section at the orange line in (c). (e) The cross-section of tiled image c at the middle part of the cell. Scale bar,10 µm.

Interestingly, SLC4-type transporters in diatoms form a diatom-specific cluster and this group is phylogenetically close to members of the human SLC4 family with high bootstrap support [25]. In addition, transporters already identified in cyanobacteria and the green alga Chlamydomonas reinhardtii do not share homology with the SLC4 transporters identified in diatoms. This strongly indicates that eukaryotic algae acquired transporters independently from various ancestral eukaryotic hosts. Indeed, it has been suggested that diatom transporters share a common origin with those in human cells [25]. However, more detailed investigation is needed to substantiate this claim because several scenarios have been postulated for the evolution of the Heterokontphyta [36] and consequently, the origin of diatom SLCs may be complex. There has been no molecular work on plasma-membrane-type transporters in freshwater diatoms. Physiological studies on the freshwater diatom Navicula pelliculosa have revealed that this species is able to take up without external CA activity [37], suggesting it uses a specific transporter. However, the freshwater environment does not fulfil the Na⁺ requirement needed by the SLC4 studied in the marine diatom P. tricornutum, strongly suggesting the occurrence of a different type of plasma membrane transporter.

(b). Bicarbonate transport in the plastidic membrane system

The four-layered chloroplast membranes represent a series of barriers that prevent DIC imported into the cytosol from making its way to the chloroplast for fixation. Therefore, it has been suggested that transporters are located at each four-layered chloroplast membrane [31] and that these transporters, in conjunction with CAs densely packed in the spaces between chloroplast membranes, control the permeation of DIC [15,23]. However, chloroplast-membrane-type transporters have not yet been identified in diatoms.

There are seven PtSLC4 genes and three PtSLC26 genes in the genome of P. tricornutum [25]. A sequence alignment of seven PtSLC4s is given in the electronic supplementary material, figure S1, in comparison with human SLC4s, hsSLC4A1 and hsSLC4A4, and SLC4s in T. pseudonana, TpSLC4-1, TpSLC4-2 and TpSLC4-3. Eleven to 12 membrane-spanning helices of human SLC4A1 were highly conserved from PtSLC4-1 to PtSLC4-5 with the exception of the first transmembrane domain unique to PtSLC4-2 (electronic supplementary material, figure S1, TM2–TM13 of PtSLC4-1–PtSLC4-5). By contrast, TM2, 4, 5, 8 and 9 were relatively well conserved throughout SLCs as compared in the electronic supplementary material, figure S1, but there were significant variations in TM 6, 7 and 10–14 in PtSLC4-6, PtSLC4-7 and three TpSLC4s. Three of the PtSLC4 genes, PtSLC4-1, PtSLC4-2 and PtSLC4-4, are CO₂ responsive as mentioned earlier. Two other PtSLC4 genes, PtSLC4-6 and PtSLC4-7, encode proteins previously predicted to localize in the four-layered chloroplast membrane systems based on the presence of targeting sequences [38]. In addition, the N-terminal transit peptide sequence in PtSLC4-6 (GSA-FTS), PtSLC4-7 (SAA-FHT), TpSLC4-2 (SFS-FAP) and TpSLC4-3 (VNA-FPT) includes both an endoplasmic reticulum (ER) signal and a plastid-transit sequence at the predicted cleavage site of ER signal, corresponding to one of the variants of the ASA-FAP motif [39] (the upper N-terminal sequence including these transit sequences is not shown in the electronic supplementary material, figure S1). Interestingly, PtSLC4-6 and PtSLC4-7 cluster phylogentically with heterokont genes which are related to the human/diatom cluster, as mentioned above. The expression levels of PtSLC4-6 and PtSLC4-7 genes were constitutive under high CO₂ and low CO₂ conditions [25], suggesting that PtSLC4-6 and PtSLC4-7 perhaps constantly regulate DIC flow from the cytosol to the plastid regardless of ambient CO₂ concentrations. Currently, very little is known about the intracellular localizations and functions of either PtSLC4-6 and PtSLC4-7 as DIC transporters. The driving forces for transporters within the four-layered chloroplast membranes are also not known. Most probably, these transporters would work with pH and/or ionic gradients across these membranes. In fact, there are some studies reporting the periplastidal compartment (PPC) to be an acidified compartment [40,41]. However, pH and ion regulation systems within the four-layered chloroplast membranes in secondary endosymbionts have not been identified. These are clearly enticing targets for future research.

(c). Energy to drive transport

Bicarbonate uptake is an energy-dependent active transport process, but how energy consumption is coupled with transport remains elusive. In P. tricornutum, uptake across the plasma membrane is mediated by PtSLC4-2, and this activity is dependent on the Na⁺ concentration. Most likely PtSLC4-2 is a secondary active transporter, co-transporting Na⁺ and , making use of a transmembrane Na⁺ gradient [25], which can be made quite large due to the high salinity of seawater. As Na⁺ continuously flows into the cell with , an efflux of Na⁺ back out of the cell is required to maintain a [Na⁺] gradient across the plasma membrane. This would require an ATP-dependent primary transporter to export Na⁺. While the molecular identity of this putative Na⁺ efflux pump has not been identified, there is some suggestive evidence for the occurrence of Na⁺/K⁺-ATPase and involvement of K⁺ in DIC acquisition in some diatoms [42–44]. However, orthologous genes encoding known Na⁺/Ka⁺-ATPases were not found in diatom genomes. Alternatively, there is a possibility that a secondary Na⁺/H⁺ antiporter maintains the Na⁺ gradient as suggested in the cyanobacterial CCM [45]. Further study is required to reveal the molecular mechanism of Na⁺-dependent uptake. There are several possible mechanisms to generate the ATP ultimately required for uptake including photophosphorylation, cyclic electron flow (CEF) around photosystem I (PSI) and respiration. In the case of cyanobacterial and green algal CCMs, the involvement of CEF in active DIC uptake has been suggested [46,47]. By contrast, in diatoms, the rate of CEF is reported to be negligibly low relative to total electron transport activity [48], making CEF an unlikely ATP source. Most probably, ATP is generated by linear electron transport through the photosynthetic electron transport chain or respiration, although further study is needed to distinguish between the two.

(d). Diffusive CO₂ uptake

In addition to the uptake facilitated by plasma membrane SLC4, diatoms take up CO₂ from the external environment to support photosynthesis [7,10,11]. Because lipid bilayer membranes are permeable to CO₂ [49], CO₂ uptake cannot proceed through a typical membrane-embedded transporter mechanism. Instead, organisms take up CO₂ through a diffusive mechanism by generating a CO₂ deficit inside the cell, which then draws CO₂ in from the external environment. In cyanobacteria, this deficit is generated by active conversion of CO₂ to in the cytoplasm through the action of NADPH dehydrogenase (NDH-1) complexes [50] with ferredoxin likely acting as the electron donor to NDH-1 [51,52]. In diatoms, this deficit was instead proposed to be generated by the active transport of out of the cytoplasm and into the chloroplast resulting in a low concentration in the cytoplasm [16,31]. The cytoplasmic CO₂ concentration is then lowered through the action of a cytoplasmic CA, which when the concentration is below equilibrium with CO₂ will drive a net hydration of CO₂ to . The CO₂ gradient passively draws CO₂ into the cell across the plasma membrane, and continued export of from the cytoplasm maintains a constant cytoplasmic CO₂ deficit resulting in sustained CO₂ uptake. The activity of the transporter exporting from the cytoplasm must match or exceed the rates of CO₂ and influx in order to maintain the inward CO₂ gradient.

New evidence suggests this model for CO₂ uptake needs to be modified because there is no known CA localized at the cytoplasm in P. tricornutum [15,23] and cytoplasmic CA localized in T. pseudonana is a γ type [23], which is confirmed to be a CA enzyme only in bacteria and archaea [53,54] but not in eukaryotes [55]. Even under alkaline pH of diatom cytoplasm at around 7.6 [8,56], it would not be possible without CA activity to maintain an inward CO₂ gradient between the cytoplasm and the external environment, because the uncatalysed hydration rate of CO₂ is slow. Instead, the CO₂ deficit may be generated in one of the membrane-bound compartments surrounding the chloroplast (the chloroplastic ER or the periplastidal space) where CAs are definitely present rather than the cytoplasm. Modelling studies of this modified CO₂ uptake mechanism indicate that it would be functionally equivalent to the older mechanism involving the cytoplasm [31].

An additional emerging complexity in the CO₂ uptake pathway involves the possible role that aquaporins play in enhancing membrane permeability to CO₂. While aquaporins were originally identified as water channels, they are also believed to facilitate CO₂ diffusion in cyanobacteria [47,57], higher plant mesophyll cells [57,58] and red blood cells [59,60] and are found in diatom genomes. While lipid bilayers are inherently permeable to CO₂, they do present some resistance to diffusion that could be reduced by the presence of aquaporins. CO₂ permeation through diatom membranes is very rapid [16], and this high permeability may be in part due to the presence of channels such as aquaporins.

3. Carbonic anhydrase as a mobilizer and an insulator for dissolved inorganic carbon movement

(a). Carbonic anhydrases and their locations in diatom cells

CA is typically an extremely fast enzyme and catalyses CO₂ hydration and dehydration moving the CO₂/ system towards equilibrium [61]. As CA interconverts CO₂ and , and these inorganic carbon species have contrasting membrane permeability, CA must have significant roles in controlling the direction and magnitude of DIC fluxes, as will be described in the next section. In the marine diatom P. tricornutum, at least 10 putative CA genes from four families have been identified in its genome. The most prominent CA family is the α-CA family with five members, but there are also two β-CAs, two γ-CAs and one CA from the newly discovered θ family [15,22]. These 10 CAs have all been localized and interestingly, they display specific localizations based upon subtype: α-CAs are located at the four-layered chloroplastic membrane, β-CAs in the pyrenoid, γ-CAs in the mitochondria and the θ-CA in the thylakoid lumen [15,22]. No cytosolic CAs have been identified so far in P. tricornutum. Of these 10 CAs, CA activity has only been confirmed with two β-CAs and one θ-CA, but the Zn binding sites in most of the other CAs are intact, suggesting they are functional [17,22]. Most of the CAs are not regulated transcriptionally by CO₂, but the two pyrenoidal β-CAs, PtCA1 and PtCA2, are highly CO₂ responsive at the transcriptional level via signal transduction involving a second messenger cAMP [26,27,29]. Transcription of Ptca1 and Ptca2 is also controlled by light using the same signal transduction pathway as that used to respond to CO₂ [29], illustrating that there is a cross-talk between light and CO₂ signals, two factors that have a major effect on the need to generate CO₂ for photosynthesis [29]. Interestingly, these two CAs are also regulated at the post-translational level by the redox state of the chloroplast through the counteraction of thioredoxins (Trxs) and molecular oxygen [17]. The reduced forms of PtCA1 and PtCA2 showed higher activity than oxidized forms [17]. Oxygen from PSII and reduced Trxs would competitively modulate CA activity, suggesting the occurrence of a system to fine tune pyrenoidal CA activities during photosynthesis. In P. tricornutum, there are no external, cytosolic nor free stromal CAs.

In T. pseudonana, at least 13 putative CA genes have been identified [23]. In sharp contrast with the case of P. tricornutum, subcellular locations of T. pseudonana CAs are not related to the CA subtypes [15,23]. Also in contrast with P. tricornutum, there is no identified pyrenoidal CA, but there is a stromal α-CA, a cytosolic γ-CA, and two external CAs, one δ-CA and one ζ-CA [23]. There are also three γ-CAs and one δ-CA in the mitochondria, and one δ-CA in the four-layered chloroplast membrane system [23]. Of these CAs, the activity of δ-CA is confirmed [62,63] and also ζ-CA activity has been confirmed in T. weissflogii [21]. Among these putative CAs in T. pseudonana, transcript levels of two external CAs (Tpδ-CA1 and Tpζ-CA1) and a CA in the PPC (Tpδ-CA1) were greatly increased in air-grown cells relative to those in high CO₂-grown cell [23]. The result strongly suggests contributions of these putative CAs to DIC acquisition and/or recapturing leaked CO₂ under CO₂-limited conditions.

Interestingly, in addition to these 13 potential CAs, the occurrence of the new subtype θ-CA family with chloroplast and thylakoid targeting motifs at its N-terminus was discovered recently in the T. pseudonana genome [20], suggesting the generality of occurrence of this type of CA in the lumen of diatom thylakoid.

The localization of CAs in these two model diatoms are updated in figure 2. Even within diatoms, the subtype of CAs found in each species and their localizations are extremely diverse, strongly suggesting that CA genes were acquired by diatoms from diverse origins after they had undergone substantial diversification and that these CAs may be involved in diverse strategies for DIC flux control depending upon the diatom species.

Figure 2.

Subtypes and localization of CAs in P. tricornutum (a) and T. pseudonana (b). Black colouring indicates that the CA has been localized, but activity has not been confirmed, whereas red colouring indicates the CA has been both localized and confirmed to have CA activity. CAs shown in grey have been identified in the genome, but have not been localized or tested for CA activity. The compartmental abbreviations are: periplasmic space (PPS), cytoplasm (Cyt), chloroplastic endoplasmic reticulum (CER), mitochondria (Mito), periplastidal compartment (PPC), chloroplast envelope (CEV), stroma (Str), pyrenoid (Pyr) and pyrenoid-penetrating thylakoid (PPT).

(b). Function of carbonic anhydrases in diatom CO₂-concentrating mechanisms

CAs may have multiple critical functions in diatom CCMs, including roles in CO₂ uptake, generation of CO₂ in the chloroplast for fixation and recovery of CO₂ leaking out of the chloroplast. As discussed above, CO₂ uptake proceeds by the generation of an internal CO₂ deficit, proximally generated by the CA-catalysed hydration of CO₂ to HCO⁻₃ in the cytoplasm or other compartment [16]. External CAs, located in the periplasmic space [15,23], facilitate CO₂ uptake by generating CO₂ from at the cell surface [64]. These external CAs are apparently very important for diatom CCMs as they occur in a number of diatoms species and are expressed at very high levels, in contrast with other microalgae where they are less commonly found [11,65,66].

Bicarbonate imported into the cell must be converted to CO₂ for fixation by RubisCO and CAs are necessary to catalyse this conversion, because the intrinsic dehydration rate of bicarbonate is slow. In P. tricornutum, a newly discovered route to produce CO₂ involves import of into the pyrenoid-penetrating thylakoid lumen where the low pH and action of a θ-CA rapidly converts to CO₂ for RubisCO [22]. β-CAs inside the pyrenoid also convert bicarbonate that diffuses in the stroma into CO₂, elevating the CO₂ concentration around RubisCO [14,15]. A portion of the CO₂ supplied to RubisCO is fixed, but the CCM is not perfectly efficient, and a significant fraction of the CO₂ leaks out of the chloroplast [16]. This leaked CO₂ is recovered by CA-catalysed conversion to back in the cytoplasm or other outlying compartment, and the regenerated can be transported back into the chloroplast. Like many other aspects of the diatom CCM, such CA-based CO₂ recovery systems seem to be highly diverse. In T. pseudonana, either the cytoplasmic, chloroplast envelope or stromal CAs could function in CO₂ recovery. In sharp contrast, P. tricornutum has numerous chloroplast envelope CAs but lacks cytosolic and stromal CAs [15,23], strongly suggesting that the main recovery points are in the four-layered chloroplastic envelope. It should also be noted that pyrenoidal β-CAs in P. tricornutum may be a part of such a recovery system of leaking CO₂, which will be discussed in the next section. The detailed molecular role of these CAs in diatoms however has yet to be determined and await further reverse genetics approaches to be confirmed.

(c). Pyrenoid functions with carbonic anhydrase

Apart from the significant divergence in the origins of chloroplast-based CAs, examples of convergence can also be found in the chloroplast and pyrenoid. The recent discovery of θ-CA (Pt43233) in the lumen of the thylakoid membrane suggests dramatic convergent coevolution of CCM function in eukaryotes. In the green alga C. reinhardtii, it has long been known that the final conversion of to CO₂ occurs within a portion of the thylakoid that penetrates the pyrenoid. is imported into the thylakoid lumen where the low pH and action of CA convert it to CO₂, which then diffuses out into the pyrenoid [67]. The newly discovered θ-CA in P. tricornutum is also localized specifically to the region of the thylakoid that passes through the pyrenoid and it is critical for CCM function, suggesting this final CO₂ generation step works quite similarly in the disparate diatom and green algal groups. Notably, the θ-CA sequence is not related to that of the C. reinhardtii thylakoid CA (an α-CA).

The P. tricornutum θ-CA possesses a Cys-Gly-His rich (CGHR) domain with an N-terminal chloroplast-thylakoid targeting motif and similar putative thylakoid targeted CGHR family proteins also exist in T. pseudonana, strongly suggesting this new class θ-CA occurs throughout the diatoms [22]. The RNAi suppression of this protein in P. tricornutum resulted in inhibition of growth and reduced photosynthetic DIC affinity [22], indicating the pivotal function of the thylakoid luminal CA for the diatom CCM. Interestingly, CGHR-containing proteins also occur in C. reinhardtii (denoted as LCIB, C, D and E). An LCIB/C hexamer complex was localized at the peripheral pyrenoid and is critical for CCM function [64,65]. CA activity of the LCIB/C complex in C. reinhardtii has not yet been tested, but the occurrence of proteins homologous to diatom luminal θ-CA as a pyrenoid-localized component critical to the CCM in the distant green algae [68,69] indicates that this CA class is commonly involved in CCM functions associated with the pyrenoid.

It is hypothesized that the LCIB/C complex in C. reinhardtii functions as a part of the system for recapturing CO₂ leaking out of the pyrenoid by hydrating it to [69–71]. If C. reinhardtii LCIB/C serves as CA, CO₂ hydration occurs via LCIB/C itself in the pyrenoid, which potentially competes with the reaction of RubisCO for CO₂. This consideration also applies to the case of P. tricornutum pyrenoid, which possesses pyrenoidal PtCAs outside the thylakoid membrane and θ-CA at the lumen [22]. It is noteworthy that, in C. reinhardtii and P. tricornutum, the LCIB/C complex and PtCAs revealed a clumped distribution to highly localized parts in the pyrenoid, while RubisCO disperses over the pyrenoid [15,70,72], suggesting that pyrenoidal CAs localize differentially from RubisCO. Detailed localization of these components is needed to clarify their functions. Recently, a novel pyrenoidal protein EPYC1/LCI5 was identified in the green alga C. reinharditii. EPYC1/LCI5 is essential to the formation of a dense aggregation of RubisCO in the pyrenoid, suggesting EPYC1/LCI5 is an essential structural component for pyrenoid formation [24]. A putative structural analogue of EPYC1/LCI5 occurs in the diatom genome [24], suggesting the involvement of such structural proteins in the arrangement of diatom pyrenoidal proteins.

4. CO₂ sensing at the ocean surface

(a). CO₂/light response

Like most algal CCMs, diatom CCMs are highly responsive to the ambient CO₂ concentration [73]. In general, components are more highly expressed at low CO₂ when the CCM is most needed to maintain high CO₂ concentrations around RubisCO [15,23,25,26,74,75]. Changes in the partial CO₂ pressure in the atmosphere result in a suite of changes in the concentrations of aqueous DIC species, including increased CO₂ and and decreased concentrations of in the range of pH relevant to marine systems. Typically, it is unclear which DIC species are eliciting the observed modifications of the algal CCM. However, a few quantitative investigations have been carried out on this topic. In cyanobacteria, the critical factor governing CCM expression is known to be the total DIC concentration, and photorespiratory metabolism is involved in the CO₂ signal transduction process [76–78]. By contrast, in freshwater green algae, Chlorella and Chlamydomonas, dissolved CO₂ is known to be a critical DIC species controlling CCM expression [79–81], strongly suggesting the involvement of some direct sensing systems in this process in eukaryotic algae. Similarly in diatoms, several physiological studies have shown that CO₂ is the critical determinant of the extent of CCM expression [8,10,11,34,73].

The molecular mechanisms mediating responses to CO₂ in diatoms are most well studied with regard to the transcriptional control system of two pyrenoidal β-CAs in P. tricornutum. It has been clearly demonstrated that transcription of both Ptca1 and Ptca2 are CO₂ and light responsive; i.e. the expression of these genes are stimulated by low (at least atmospheric level) CO₂ and largely repressed in enriched (1–5%) CO₂, and this expressional control requires light [26,74,75]. Interestingly, such stimulation (or de-repression) of these CA genes under low CO₂ conditions are efficiently suppressed by cAMP analogues and/or cAMP phosphodiesterase inhibitors [26,27,29], indicating an involvement of the cAMP second messenger system downstream of the CO₂ sensing mechanism (figure 3). A detailed GUS reporter assay targeting the region 1.3 kbp upstream of the transcription-start sites of both Ptca1 and Ptca2 revealed new cis-elements critical for the CO₂ response, termed CO₂/cAMP-responsive elements (CCREs: ACGTCA/G) [27,29]. Three CCRE sequences in the Ptca1 promoter region reside within −90 bp relative to the transcription-start site and these sequences run in opposite directions to each other with 15–18 bp intervals [27]. Gel shift assays showed that these CCREs are a target of a group 4 basic-zipper (bZIP) transcription factor, PtbZIP11 in P. tricornutum [27,83]. The promoter region of the Ptca2 gene also possesses three CCRE sequences with a different arrangement from that of the Ptca1 gene promoter, and these sequences were also shown to be critical cis-elements mediating the CO₂ response of this gene [29]. A further transcriptomics study using T. pseudonana demonstrated that the transcriptional response of the CCM and photorespiratory gene clusters are responsive to CO₂ using CCRE sequences for their regulation [28], indicating a common mechanism for CO₂ signalling in diatoms.

Figure 3.

The CO₂-responsive elements in the Ptca1 and Ptca2 promoters and the putative CO₂ signalling pathway. (a) Structures of the core-regulatory region of the Ptca1 and the Ptca2 promoter. (b) A suggested model of the cAMP-mediated CO₂ signalling pathway based upon the mammalian cAMP signalling cascade. (Redrawn from Matsuda & Kroth [82]). This model describes the signalling route under high CO₂ conditions. PDE, cAMP phosphodiesterase; PKA, protein kinase A; PKAC, C subunit of PKA; PP, protein phosphatase; CBP, CREB (cAMP binding protein) binding protein; DARPP, dopamine- and cAMP-regulated phosphoprotein. Parentheses indicate the number of candidate genes in the P. tricornutum genome. No functional analogue of DARPP has been identified so far.

A recent study further demonstrated that Ptca1 and Ptca2 genes are also light responsive and this response is governed by the same set of CCREs that are involved in the CO₂ response [29] (figure 3), strongly suggesting that the light signal is integrated with the CO₂/cAMP signal by some type of cross-talk mechanism [48]. Interestingly, a very weak dose of 2,6-dichlorophenol-indophenol, which oxidizes the acceptor side of PSI, efficiently suppressed the transcriptions of Ptca1 and Ptca2 under low-CO₂ and illuminated conditions [29]. This strongly suggests that light may generate a retrograde signal at the acceptor side of PSI as a part of an electron sorting system from ferredoxin, which is one of the first examples of an involvement of PSI in a CO₂/light retrograde signal which ultimately manipulates nuclear gene expression.

(b). CO₂ with diel light and Fe signals

The micronutrient iron has a major role in diatom metabolism, particularly in the light reaction of photosynthesis and in nitrate assimilation. For both of these metabolic pathways, light levels greatly influence the regulation, and thus potentially Fe requirements. A recently published study examined the global transcriptional response of P. tricornutum to diel light cycles at three different Fe concentrations [84], and we have examined the expression of the CCM components detailed in this review. As part of that study, individual gene expression profiles were determined to be statistically responsive to either light or iron levels using multiple methods.

Of the various CCM components, PtCA2 is by far the most highly expressed component, with PtCA1 being expressed below statistical significance. PtCA2 had a statistically similar expression profile to PtSLC4-1, PtSLC4-5, the θ-CA and PtbZIP11. Each of these components are upregulated in light conditions relative to dark conditions and are downregulated at lower Fe concentrations. PtbZIP11 is induced by both light conditions but also by low Fe concentrations. Curiously, both PtSLC4-3 and PtSLC4-52bd are upregulated in the dark relative to the light, suggesting different functional roles to the other putative outermembrane-localized PtSLC4. PtSLC4-7 shows constitutive expression, while PtSLC4-6 is mildly upregulated in light conditions. The names of the CCM components in P. tricornutum referred to in this paper are listed with their corresponding Protein ID and updated annotations in table 1.

Table 1.

CCM-associated proteins encoded in the genome of P. tricornutum and T. pseudonana, and role of each protein shown in this review. PM and CM indicate plasma membrane and chloroplast membrane, respectively.

annotation name	Protein ID	role
PtbZIP11	49205a	regulating
PtCA1	45433a	pyrenoidal carbonic anhydrase
PtCA2	51305a	pyrenoidal carbonic anhydrase
θ-CA	43233a	pyrenoid-penetrating thylakoidal lumen carbonic anhydrase
PtSLC4-1	1677a	PM-type bicarbonate transporter?
PtSLC4-2	bd1806b	PM-type bicarbonate transporter
PtSLC4-3	bd1743b	PM-type bicarbonate transporter?
PtSLC4-4	bd714b	PM-type bicarbonate transporter?
PtSLC4-5	54405a	PM-type bicarbonate transporter?
PtSLC4-52bd	bd52b	PM-type bicarbonate transporter?
PtSLC4-6	43194a	CM-type bicarbonate transporter?
PtSLC4-7	45656a	CM-type bicarbonate transporter?
PtSLC26-1	42555a	PM-type bicarbonate transporter?
PtSLC26-2	42556a	PM-type bicarbonate transporter?
PtSLC26-3	50075a	PM-type bicarbonate transporter?
TpSLC4-1	266801c	PM-type bicarbonate transporter?
TpSLC4-2	1403c	CM-type bicarbonate transporter?
TpSLC4-3	267979c	CM-type bicarbonate transporter?

^aProtein IDs refer to JGI genome database (http://genome.jgi.doe.gov/Phatr2/Phatr2.home.html).

^bProtein IDs refer to JGI genome database (http://genome.jgi.doe.gov/Phatr2_bd/Phatr2_bd.home.html).

^cProtein IDs refer to JGI genome database (http://genome.jgi.doe.gov/Thaps3/Thaps3.home.html).

5. Models of dissolved inorganic carbon flux and its dynamic control

The basic components used to create CCMs are few in number— transporters, CAs, RubisCO—and it is the careful spatial arrangement of these components and manipulation of CO₂/ equilibrium and gradients that combine to form efficient CCMs. Consequently, numerical models of CCMs have been especially helpful in establishing the essential functional characteristics of CCMs in diverse microalgae [1,85,86]. In the case of diatoms, the most extensive numerical modelling of the CCM has involved the diatom P. tricornutum. Here, model-data comparisons were used to infer quantitative rates of inorganic carbon fluxes into the cell and within the major cellular compartments [16,31]. These models suggested that a high rate of active transport out of the cytoplasm and into the chloroplast (the ‘chloroplast pump’) was the major driver of the CCM in P. tricornutum leading to accumulation in the chloroplast stroma and the CO₂ deficit in the cytoplasm. Furthermore, according to the transcriptional analysis of some CCM components in response to CO₂ (and partially to light), acclimation responses of the CCM to CO₂ concentration seemed to take place primarily in the DIC acquisition system from the bulk medium, while by contrast, putative components of a ‘chloroplast pump’ moving DIC from the cytoplasm to RubisCO were not affected by CO₂ availability [15,22,23,25]. The latest iteration of these P. tricornutum CCM models [31] emphasized the coherence between molecular and physiological data when incorporated into a modelling framework, and suggested that the system was largely understood. However, the recent discovery of the role of the thylakoid and its associated θ-CA in CO₂ supply require a reassessment of this conclusion. There is also the need to develop CCM models for other diatoms, and this should be feasible for additional species such as T. pseudonana as more molecular and physiological data on these species become available.

6. Conclusion

The molecular components, their origin, their localization and some of their putative functions in the CCM described in this review illustrate the diversity of diatom CCMs. In particular, putative DIC flux control systems between the external medium and the cytosol seem to be divided into at least two low-CO₂ inducible strategies: one uses active pumps across plasma membranes and the other relies more significantly on passive CO₂ entry aided by external CAs, and these likely use contrasting diffusion barrier systems against CO₂ leakage (figure 4). However, in both cases, as a convergent aspect, we assume a strong constitutive pumping action at the chloroplast envelope, which allows an efficient removal of CO₂ from the cytosol and an efficient accumulation of in the stroma. Furthermore, both systems most likely use the pyrenoid as a hub to generate ample CO₂ flux from accumulated aided by thylakoid luminal θ-CA to support fixation by RubisCO. In this process, pyrenoidal CA activity may function as a critical part of the CCM in concert with luminal CA, which appears to be a general feature of algal CCM, but details of this mechanism in diatoms requires further fine structural analysis of the pyrenoid.

Figure 4.

Models of DIC transport in marine diatoms. (a) In P. tricornutum, plasma-membrane-located-SLC4s (PtSLC4-2, possibly PtSLC4-1 and PtSLC4-4) transport from seawater in a Na⁺-dependent manner, most probably Na⁺– cotransport. This would require an unidentified ATP-dependent primary Na⁺ pump or a proton-dependent secondary pump to maintain a [Na⁺] gradient across the plasma membrane. The mechanism of DIC transport from the cytosol to the stroma is unknown. Unidentified transporters on CER and CE might be involved in this process in combination with CAs in intermembrane spaces. In the pyrenoid, CO₂ is generated by luminal θ-CA in the pyrenoid-penetrating thylakoid (PPT). CO₂ released from the lumen is fixed by RubisCO, or otherwise recaptured at specialized pyrenoidal loci by β-type CAs, PtCA1 and/or PtCA2 by which is regenerated to be transported back to the thylakoid lumen. (b) In T. pseudonana, extracellular CA accelerates the dehydration of , resulting in high diffusive permeation of CO₂. Active transport into the chloroplast would lower cytoplasmic [], and then CO₂ that diffuses into the cell is readily converted into by cytosolic CA (alternatively this could occur in the PPC). As in P. tricornutum, transporters on CER and CE have not been identified. In T. pseudonana, a putative θ-CA is predicted in the PPT, suggesting a role for θ-CA in CO₂ generation as proposed in P. tricornutum. There is no known CA localized in the pyrenoid. (c) In C. reinharditii, is delivered into the stroma by plasma-membrane-located HLA3 and chloroplast-envelope-located LCIA. LCIA is shown to be on the chloroplast envelope, but its exact localization, whether inner or outer membrane, is unknown. Here, we assumed that LCIA is on the inner envelope because the outer membrane is generally permeable to low molecular weight compounds which most probably allows spontaneous permeation. LCI1 on plasma membrane was strongly suggested to be a DIC transporter, but the inorganic carbon species transported by LCI1 have not been determined. CAH3, a luminal CA in the thylakoid tubule, generates CO₂ from . Unlike P. tricornutum, the luminal CAH3 is α-type CA, suggesting convergent evolution of a mechanism to supply CO₂ by using luminal acidification. Similarly, the LCIB-C complex is a putative θ-CA, another example of convergent evolution of pyrenoidal CA which may play a role in recapturing leaking out CO₂. CE, chloroplast envelopes; CER, chloroplast ER; PM, plasma membrane; PPC, periplastidal compartment; PPS, periplasmic space.

Data accessibility

This article has no data.

Competing interests

We declare we have no competing interests.

Funding

This work was supported by Grant-in-Aid for Scientific Research B (grant no. 24310015 to Y.M.), by Grant-in-Aid for Scientific Research on Innovative Areas (grant no. 16H06557 to Y.M.), by Grant-in-Aid for Young Scientists B (grant no. 26870750 to K.N., and grant nos 15K16156 and 17K15326 to Y.T.) from Japan Society for the Promotion of Science (JSPS), by MEXT-Supported program for the Strategic Research Foundation for the Advancement of Environmental Protection Technology and for Development of Intelligent Self-Organized Biomaterials and by the Science Research Promotion Fund of the Promotion and Mutual Aid Corporation for Private Schools of Japan (to Y.M.). B.M.H. was supported by the US National Science Foundation (EF 1041023, MCB 1129326) and the Alfred P. Sloan Foundation (BR2014-049).