The evolution of inorganic carbon concentrating mechanisms in photosynthesis

Abstract

Inorganic carbon concentrating mechanisms (CCMs) catalyse the accumulation of CO₂ around rubisco in all cyanobacteria, most algae and aquatic plants and in C₄ and crassulacean acid metabolism (CAM) vascular plants. CCMs are polyphyletic (more than one evolutionary origin) and involve active transport of ${\text{HCO}}_3^{-}$, CO₂ and/or H⁺, or an energized biochemical mechanism as in C₄ and CAM plants. While the CCM in almost all C₄ plants and many CAM plants is constitutive, many CCMs show acclimatory responses to variations in the supply of not only CO₂ but also photosynthetically active radiation, nitrogen, phosphorus and iron. The evolution of CCMs is generally considered in the context of decreased CO₂ availability, with only a secondary role for increasing O₂. However, the earliest CCMs may have evolved in oxygenic cyanobacteria before the atmosphere became oxygenated in stromatolites with diffusion barriers around the cells related to UV screening. This would decrease CO₂ availability to cells and increase the O₂ concentration within them, inhibiting rubisco and generating reactive oxygen species, including O₃.

1. Introduction

Photosynthesis by O₂-producing organisms invariably involves the carboxylase rubisco (ribulose bisphosphate carboxylase–oxygenase) as the core enzyme involved in producing reduced carbon (Tabita et al. 2008). The different molecular forms of this enzyme, with different kinetics with respect to substrate affinities and maximum specific reaction rates, appear to function optimally within the constraints of the overall mechanism of the enzyme (Tcherkez et al. 2006). However, no known rubisco has a half-saturation value for CO₂ that saturates the carboxylase activity at the present atmospheric CO₂ concentration and at the temperature that the enzyme functions in vivo. Furthermore, all known rubiscos have an oxygenase activity competitive with the carboxylase activity that adds resource costs to net C assimilation in the present atmosphere with diffusive CO₂ entry, and they all have high molecular masses and relatively low CO₂-saturated specific reaction rates so that offsetting the relatively low CO₂ affinity by increasing the enzyme quota on a biomass basis is very costly in terms of energy and nitrogen inputs.

These considerations can help to rationalize, in evolutionary terms, why about half of the total inorganic C fixation on Earth, almost all by oxygenic photosynthesis, involves inorganic carbon concentrating mechanisms (CCMs). These CCMs increase the CO₂ concentration, and the CO₂/O₂ ratio, at the site of rubisco activity. This increases the rate of net C assimilation per unit rubisco and decreases the resource costs associated with rubisco oxygenase activity, although involving additional resource costs for the synthesis, maintenance and operation of the mechanisms that bring about the active accumulation of CO₂. An additional resource cost comes from the need for additional pumping of inorganic C to make up for any inorganic carbon leakage that occurs from the internal pool (Raven 1991a; Sültemeyer & Rinast 1996). Since any barriers to leakage of CO₂ (the most likely species of inorganic carbon to leak) are also likely to act as barriers to the efflux of photosynthetically produced O₂, there is also likely to be the additional problem of higher steady-state O₂ concentrations around the photosynthetic apparatus with implications for rubisco catalysis and the generation of reactive oxygen species (Raven & Larkum 2007).

Here we outline the mechanism and environmental and phylogenetic distribution of CCMs in extant organisms, the times at which they evolved, and their probable future in relation to atmospheric composition. We first consider the CCMs in cyanobacteria and algae, which account for well over half of the 50 Pg or so of carbon assimilated each year in the oceans, and then with the C₄ and crassulacean acid metabolism (CAM) found in terrestrial vascular plants that accounts for 18–30% of the 60 Pg or so of C assimilated each year on land (Berry 1994; Ehleringer et al. 1997). The cyanobacterial and algal CCMs, like those of many submerged vascular plants, involve energized transport of CO₂, ${\text{HCO}}_3^{-}$ and/or H⁺ across one or more membranes. C₄ and CAM involve a C₃/C₄ carboxylation–decarboxylation cycle preceding rubisco. In C₄, this cycle involves decarboxylation at different sites, but within seconds or tens of seconds, of the carboxylation, while in CAM the carboxylation and decarboxylation parts of the cycle are very close together spatially but are approximately 12 hours out of phase over the diel light–dark cycle.

2. The range of organisms with CCMs

(a) Background

CCMs occur today in all the cyanobacteria examined, in all algae apart from a few members of the Chlorophyta and Rhodophyta as well as most or all members of the Chrysophyceae and Synurophyceae, as well as in some hornworts, lycopsids and ferns, and in a significant number of flowering plants (Winter & Smith 1995; Badger et al. 1998, 2002; Sage & Monson 1998; Colman et al. 2002; Keeley & Rundel 2003; Price & Badger 2003; Edwards et al. 2004; Giordano et al. 2005; Raven et al. 2005a–c; Price et al. 2007; Roberts et al. 2007a,b; table 1).

Table 1.

The distribution of CCMs among extant organisms.

mechanism	location within organism	phylogenetic distribution	references
a. Passive CO2 entry, energized conversion to ${\text{HCO}}_3^{-}$	plasmalemma, thylakoid; ${\text{HCO}}_3^{-}$→CO2 and assimilation by rubisco in carboxysomes	cyanobacteria	Badger et al. (1998, 2002), Price & Badger (2003), Giordano et al. (2005) and Price et al. (2007)
b. Energized entry of ${\text{HCO}}_3^{-}$	plasmalemma, carboxysomes	cyanobacteria	Badger et al. (1998, 2002), Price & Badger (2003), Giordano et al. (2005) and Price et al. (2007)
c. Energized entry of CO2	plasmalemma? plastid envelope? rubisco in stroma/pyrenoid	many algae, and either CO2 or ${\text{HCO}}_3^{-}$ in hornworts with CCMs and some of the aquatic vascular plants with CCMs (see e and f)	Smith & Griffiths (1996), Colman et al. (2002), Maberly & Madsen (2002), Giordano et al. (2005), Raven et al. (2005b), Kevekordes et al. (2006), Burey et al. (2007), Moroney & Ynalvez (2007), Roberts et al. (2007a) and Spalding (2007)
d. Energized entry of ${\text{HCO}}_3^{-}$	plasmalemma? plastid envelope? rubisco in stroma/pyrenoid	many algae, and either CO2 or ${\text{HCO}}_3^{-}$ in hornworts with CCMs and some of the aquatic vascular plants with CCMs (see d and f)	Smith & Griffiths (1996), Colman et al. (2002), Maberly & Madsen (2002), Giordano et al. (2005), Raven et al. (2005b), Kevekordes et al. (2006), Burey et al. (2007), Moroney & Ynalvez (2007), Roberts et al. (2007a) and Spalding (2007)
e. Energized flux of H+ to cell wall, conversion of ${\text{HCO}}_3^{-}$ to CO2	plasmalemma, CO2 flux to rubisco in stroma	some algae, including characean green algae, and either CO2 or ${\text{HCO}}_3^{-}$ in hornworts with CCMs and some of the aquatic vascular plants with CCMs (see d and e)	Walker et al. (1980), Beer et al. (2002), Maberly & Madsen (2002), Helblom & Axelsson (2003), Uku et al. (2005)
f. Energized flux of H+ to thylakoid lumen, conversion of ${\text{HCO}}_3^{-}$ to CO2	thylakoids, CO2 flux to rubisco in pyrenoid	freshwater green microalga Chlamydomonas	Pronina & Semenenko (1992), Raven (1997a), Giordano et al. (2005) and Moroney & Ynalvez (2007)
g. C4 metabolism in single-cell type	inorganic C+C3 acid→C4 acid in cytosol, C4 acid→C3 acid + CO2 in chloroplast stroma (or nearby), rubisco in stroma	marine green acellular macroalga Udotea, possibly a marine diatom, a few terrestrial and submerged flowering plants	Beardall et al. (1976), Reiskind et al. (1988), Raven (1997a), Sage & Monson (1998), Reinfelder et al. (2000), Keeley & Rundel (2003), Edwards et al. (2004), Giordano et al. (2005), Osborne & Beerling (2006), Roberts et al. (2007a,b)
h. C4 metabolism in two-cell types	inorganic C+C3 acid from bundle sheath (bs) cell→C4 acid in cytosol of mesophyll (mes) cell, C4 acid→bs cell, decarboxylated to C3 acid (→mes) and CO2 fixed by rubisco in stroma of bs chloroplasts.	most C4 terrestrial flowering plants, a few amphibious/aquatic flowering plants	Sage & Monson (1998), Keeley & Rundel (2003) and Osborne & Beerling (2006)
i. C3–C4 intermediate	combinations of improved internal recycling of photorespiratory CO2 and partial two-cell C4 photosynthesis in flowering plants; in one cell in a diatom?	a few terrestrial flowering plants, and a diatom	Sage & Monson (1998) and Roberts et al. (2007a,b)
j. Crassulacean acid metabolism	inorganic C+C3 organic acid in dark in cytosol of mes→C4 acid stored for approximately 12 hours in vacuole; in light C4 acid→cytosol→C3 acid (→carbohydrate) and CO2 fixed by rubisco in stroma	some aquatic/amphibious lycophytes, a few terrestrial ferns and gymnosperms, some terrestrial and aquatic vascular plants	Winter & Smith (1995), Keeley (1998) and Keeley & Rundel (2003)

(b) Cyanobacteria

In cyanobacteria, the CCM is based on accumulation of ${\text{HCO}}_3^{-}$ in the cytosol (table 1a,b), movement of this ${\text{HCO}}_3^{-}$ through their proteinaceous shells into the carboxysomes containing rubisco, and conversion to CO₂ using one or more carbonic anhydrase (CA) enzymes (Badger et al. 2002; Price et al. 2007). The open ocean cyanobacteria with form IA rubiscos have a restricted suite of ${\text{HCO}}_3^{-}$ accumulation processes and, probably, little capacity to acclimatize to decreased inorganic C availability (an unlikely event in the pelagic ocean). Coastal and freshwater cyanobacteria with form IB rubisco have a greater range of ${\text{HCO}}_3^{-}$ accumulation mechanisms and an ability to acclimatize to decreased external inorganic carbon supply by expressing high-affinity ${\text{HCO}}_3^{-}$ accumulation processes, yielding a greater whole-cell affinity for ${\text{HCO}}_3^{-}$ and CO₂ (Price & Badger 2003; Price et al. 2007).

A lot is known about the components of the cyanobacterial CCMs at the molecular genetic level. It is known that regulation involves sensing the size of the intracellular ${\text{HCO}}_3^{-}$ pool, with an auxiliary role for the concentration of O₂ (Price et al. 2007).

(c) Algae

There is apparently greater diversity of CCMs in algae than in cyanobacteria (table 1c–g,i) and less is known about the mechanisms and molecular genetics of the CCMs (Badger et al. 1998; Colman et al. 2002; Giordano et al. 2005). Pyrenoids, which are within the chloroplast stroma and contain rubisco, occur in many algae expressing CCMs. Pyrenoids are conjectured to be the analogues of cyanobacterial carboxysomes, although they are not present in all algae with CCMs (Badger et al. 1998; Giordano et al. 2005; Kevekordes et al. 2006). The most detailed knowledge is for the green freshwater alga Chlamydomonas reinhardtii (Moroney & Ynalvez 2007; Spalding 2007; cf. Im et al. 2003). Rather less is known for the biogeochemically very important diatoms (Roberts et al. 2007a,b), although there are now two completed diatom genome sequences. While very small (picoplanktonic) cells could, in principle, rely on diffusive CO₂ entry when similar, larger cells could not, the evidence for eukaryotic algae (and cyanobacteria) shows that even the smallest cells have CCMs (Raven 1991b; Iglesias-Rodriguez et al. 1998; Giordano et al. 2005).

Much research is based on the assumption that the algal CCM is based on the active transport of one or more of CO₂, ${\text{HCO}}_3^{-}$ and/or H⁺ and that the first acid-stable product of inorganic carbon assimilation is 3-phosphoglycerate, the immediate product of the carboxylase reaction of rubisco, that is, the photosynthetic biochemistry is C₃. However, there are tracer carbon and other data suggesting C₄ photosynthetic biochemistry in a green alga and possibly in a diatom (table 1g). However, more recent tracer carbon data (table 1i) suggest that the diatom Thalassiosira weissflogii has C₃–C₄ intermediate biochemistry while T. pseudonana has C₃ biochemistry; both have CCMs, presumably based on mechanisms in table 1b–d (Roberts et al. 2007a,b; cf. McGinn & Morel 2008).

Keeley (in Winter & Smith 1995) and Keeley (1998) concludes that there is little evidence of a significant contribution of CAM to the carbon budget of algae.

(d) Symbioses with cyanobacteria and/or algae as the partner performing photosynthesis

Raven (2003) reviews the available evidence; this shows that all of the cyanobacteria and dinoflagellates investigated, and some of the green algae examined, express CCMs when they are contributing photosynthesis to a symbiosis.

(e) Embryophyta: anthocerotes, lycopsids, pteropsids and coniferopsids

The hornworts are terrestrial; some of them have pyrenoids in their chloroplasts and have CCMs (Smith & Griffiths 1996; table 1c,d). The lycopsid Isoetes (an embryophyte clade at the fern level of organization) often lives submerged in freshwater and obtains most of its inorganic carbon through the roots and assimilates CO₂ by CAM in the leaves (Keeley 1998; table 1j). This is also true of amphibious species (e.g. Isoetes andicola, formerly Stylites andicola) that, even when their leaves are exposed, do not form functional stomata and behave ecophysiologically as if they are still submerged (Keeley 1998). A number of epiphytic ferns (pteropsids) have CAM (Keeley & Rundel 2005; table 1j). The coniferopsid cycadophyte Dioon edule (Vovides et al. 2002) and gnetophyte Welwitschia mirabilis have some features of CAM, which could constitute a CCM (Keeley & Rundel 2003).

(f) Embryophyta: magnoliophytes with CAM

CAM was first known from terrestrial dicotyledonous plants where it has been associated with restricted inorganic carbon supply as a result of restricted stomatal opening in the light when evaporative demand is highest in habitats of limited or fluctuating water availability, for example, epiphytes and inhabitants of some arid zones (Winter & Smith 1995; Keeley & Rundel 2003). CAM in freshwater aquatic plants was recognized much later, and has also been related to restricted inorganic carbon availability although not, in aquatics, as a result of water deficiency (Keeley 1998; Maberly & Madsen 2002). Inorganic carbon deficiency in freshwaters is not strongly coupled to atmospheric variations, being more dependent on autogenous factors in the water body (Keeley 1998; Keeley & Rundel 2003).

(g) Embryophyta: magnoliophytes with C₄ photosynthesis

A range of terrestrial flowering plants in a number of families has polyphyletically evolved the C₄ pathway of photosynthesis, which usually involves the division of biochemical functions between two cell types in ‘Kranz anatomy’, which could be a palaeontological indicator of this form of C₄ photosynthesis (table 1h), but sometimes the whole reaction sequence occurs in single cells (Sage & Monson 1998; Edwards et al. 2004; table 1g). A few submerged freshwater aquatic flowering plants also have C₄ photosynthesis (Maberly & Madsen 2002; Keeley & Rundel 2003). Sage et al. (2002) discuss why the evolution of C₄ and CAM in a single genus cannot occur, with the possible exception of Portulaca.

(h) Embryophyta: magnoliophytes with C₃–C₄ intermediate photosynthesis

This syndrome, initially defined on the basis of physiology, involves a range of metabolic and structural features in different species, is apparently limited to terrestrial flowering plants (and a diatom!) and is polyphyletic.

(i) Embryophyta: aquatic magnoliophytes

Some submerged freshwater aquatic magnoliophytes have constitutive or facultative CAM, while others have constitutive or facultative C₄ photosynthesis. However, other freshwater flowering plants, and seagrasses, have C₃ photosynthetic biochemistry, and many of these have CCMs based on active transport of ${\text{HCO}}_3^{-}$, CO₂ and/or H⁺ (table 1c–e,g,h,j).

3. Regulation of CCMs

C₄ (and C₃–C₄ intermediate) photosynthesis in terrestrial plants (and, as far as it has been tested, in a diatom) is constitutive, as is CAM in many terrestrial CAM plants (Keeley & Rundel 2003; Roberts et al. 2007a,b). The expression of CAM in some terrestrial plants is facultative, with the expression when the organism has a limited supply of water relative to evaporative demand (Keeley & Rundel 2003). CAM in submerged vascular plants of the isoetid life form is facultative in some species, which revert to C₃ physiology when emersed, but other species of isoetids, with thick cuticles on their shoots and no openable stomata, retain CAM using CO₂ taken up through the roots even when the shoots are emersed. Other submerged CAM plants and those with C₄ that evolved after they returned to living in water show variability in expression of these pathways as a function of temperature and inorganic carbon availability. Cyanobacteria and algae with CCMs typically show constitutive expression of the CCMs within the range of inorganic carbon concentrations and speciation that they normally encounter in nature, although in many cases there is increased affinity for inorganic carbon of the overall CCM when the organism is grown at lower inorganic carbon concentrations (Giordano et al. 2005). The adaptation, and the acclimation mechanisms, of inorganic C acquisition mechanisms will have been influenced by the glacial–interglacial cycles of atmospheric CO₂ (180–280 ppm); this variation is not within the anthropogenic range of CO₂ increase, currently at 380 ppm.

The expression of CCMs in algae and cyanobacteria is not just regulated by the supply of inorganic carbon. The levels of PAR (photosynthetically active radiation), which are lower than those yielding the maximum rate of growth lead to a decreased affinity for inorganic carbon (Giordano et al. 2005), with the probable consequence that the reduced photosynthetic energy supply reduces the selective significance of CCMs relative to diffusive CO₂ entry. This is because, in the case of CCMs, a constant leakage rate of CO₂ needs to be offset, so that the energy cost per unit net carbon assimilation increases, whereas in the case of diffusive CO₂ entry, the energy loss due to photorespiration declines in proportion to carboxylation (Giordano et al. 2005). There are also effects of inorganic carbon supply for growth on acclimation to variations in PAR (MacKenzie et al. 2002; Burns et al. 2006).

The other aspect of electromagnetic radiation that has been investigated for its effect on CCMs is the effect of UV-B. The available data are for three species of eukaryotic algae, and range from showing that the CCM is less sensitive to UV-B than are other reactions of photosynthesis to finding that the CCM is more sensitive than the rest of photosynthesis (Beardall et al. 2002; Sobrino et al. 2004). The data from terrestrial flowering plants (Johnson & Day 2002; Correia et al. 2005) do not permit a decision as to whether the C₄ CCM shows specific sensitivity to UV. No work seems to have been performed on the effect of UV-A when energizing photosynthesis (Gao et al. 2007) on the functioning of CCMs.

At a constant inorganic C supply, the affinity for inorganic carbon increases when nitrogen (as ${\text{NO}}_3^{-}$) is limiting growth; this could be related to higher carbon assimilation per unit catalytic nitrogen when a CCM is operating (Raven 1991a,b; Raven et al. 2005c). However, when nitrogen is supplied as ${\text{NH}}_4^{+}$ there is a decreased affinity for inorganic carbon when nitrogen is limiting (Raven et al. 2005c). The effects of phosphorus deficiency on CCM expression are not consistent between the two datasets available (Raven et al. 2005c). Iron deficiency increases inorganic carbon affinity, perhaps, by a similar argument to that used for nitrate deficiency, because greater CCM expression increases net carbon assimilation per unit catalytic iron (Raven 1990, 1991a,b; Raven et al. 2005c). The inorganic C assimilation rate per unit catalytic Zn could be increased if the need for CA (which usually requires Zn) for photosynthesis is decreased or abolished by switching from C₃ to C₄ biochemistry if ${\text{HCO}}_3^{-}$ enters the cells (Reinfelder et al. 2000; Raven et al. 2005c; cf. Roberts et al. 2007a,b).

The findings for increased inorganic C affinity related to ${\text{NO}}_3^{-}$ and Fe limitations of growth, and the connection to predictions from biochemical modelling as to reductions in the N and Fe cell quotas needed for growth at a given rate as a function of increased engagement of CCMs, clearly need further examination by whole-cell elemental, and quantitative proteomic, analyses (cf. Finkel et al. 2006).

4. Implications of CCMs for natural abundance of C isotopes

In terrestrial plants that lack CCMs, that is, embryophytes and lichens with diffusive CO₂ entry associated with C₃ physiology and biochemistry, the δ¹³C values are in the range −25‰ to −30‰. By contrast, those with CCMs, i.e. C₄ biochemistry, CAM, and those based on active transport across membranes in some anthocerotes and lichens, have higher, i.e. less negative, δ¹³C values, up to −10‰. The values for C₃–C₄ plants are more variable and frequently near the δ¹³C values of those with C₃ physiology. For aquatic organisms, the picture is less clear (Raven et al. 2002a,b; Kevekordes et al. 2006), even in oceans where the δ¹³C values of the source inorganic carbon are somewhat more variable than that in the atmosphere on yearly to decadal time scales, but much less variable than among freshwater bodies. The most negative values of organism δ¹³C are generally correlated with diffusive CO₂ entry, while more positive values relate to either presence of CCMs, or diffusive limitation of organic supply to the organism, or a combination of the two (Raven et al. 2002a,b; Kevekordes et al. 2006). However, there are possibilities of CCMs which can also bring about 1:1 exchange fluxes of inorganic carbon across membranes, which would yield very negative δ¹³C values (Raven et al. 2002a). The use of δ¹³C values in detecting the occurrence of CCMs in the past requires a knowledge of the source inorganic C δ¹³C value, and is also influenced by temperature and by inorganic carbon concentrations as well as the rate of photosynthesis and, especially in aquatic organisms, cell or organ size and boundary-layer thickness (Korb et al. 1996; Finkel et al. 2005; Katz et al. 2005).

5. Evolutionary origins of CCMs: whence?

The genes for many biochemical components of C₄ and CAM occur as parts of anaplerotic pathways and storage mechanisms in their C₃ ancestors (Winter & Smith 1995; Sage & Monson 1998; Keeley & Rundel 2003). This is not to say that the gene duplications, alterations in regulation and/or localization of the products of the duplicate (C₄ or CAM) genes, and other developmental changes involved in the evolution of C₄ and CAM are trivial, but it is clear that these evolutionary changes have occurred independently in C₄ approximately 30 times, and 20 times for CAM.

Raven (1980) pointed out that CCMs based on active transport processes were analogous to those accumulating other nutrient solutes (e.g. ${\text{NH}}_4^{+}$, ${\text{NO}}_3^{-}$, urea, amino acids, ${\text{H}}_2{\text{PO}}_4^{-}$, ${\text{SO}}_{\text{4}}^{2^{-}}$ and Si(OH)₄) prior to their assimilation. Some of these accumulation mechanisms could have evolved in relation to episodes of nutrient shortage over geological time (Raven et al. 2005a; cf. Konhauser et al. 2007). Raven et al. (2008) suggest that the CCMs could generally be interpreted in terms of the recruitment of many components from pre-existing transporters and, in the case of CO₂ acquisition by cyanobacteria, NAD(P)H dehydrogenase. The argument of Raven et al. (2008) could be further tested with the phylogenies of gene families containing CCM components when more is known of the genetic background of CCMs, especially for eukaryotes. The case for such bricolage is most obvious for secondarily aquatic flowering plants, which evolved less than 120 Myr ago (Raven et al. 2008). As Raven et al. (2008) point out, each stage in CCM evolution must have had selective significance, though identification of these stages and their significance is difficult.

6. Evolutionary origin of CCMs: why?

The replacement of rubisco and the photosynthetic carbon reduction cycle in O₂ evolvers by one of the other inorganic carbon assimilation pathways found in autotrophs is apparently prevented by, among other things, oxygen damage to one or more of the enzymes involved in these alternative pathways (Berg et al. 2007; Raven et al. 2008).

The fact that O₂ evolvers cannot replace rubisco means that the ‘obvious’ role of CCMs is in improving the working conditions for rubisco in an environment in which diffusive gas exchange would yield significantly subsaturating concentrations of CO₂ at the active site of rubisco and/or a significant restriction on net carbon assimilation by the occurrence of rubisco oxygenase activity (Giordano et al. 2005; Raven & Larkum 2007). However, it must be remembered that not all photosynthetic O₂ evolvers have CCMs (Raven et al. 2005a,b); optimality criteria that compare the costs and benefits of the trait should be employed to investigate the distribution of CCMs (Rosen 1967; Raven et al. 2008). As will be seen under consideration of when CCMs evolved, a low CO₂ concentration, either globally or locally, is frequently considered as a selective factor in the evolution of CCMs (Keeley & Rundel 2003; Giordano et al. 2005; Osborne & Beerling 2006), if not always in their spread (Keeley & Rundel 2005). The role of O₂ in the evolution of CCMs is not clear (Giordano et al. 2005; Riding 2006). While not claiming that the environmental signal for inducible CCMs reflects the selective factors involved in their evolution, in cyanobacteria increased expression of CCMs involves decreased intracellular inorganic carbon concentration with a secondary role for increased oxygen (Price et al. 2007), while the eukaryotic algae tested respond to extracellular CO₂ with no observable effect of O₂ (Giordano et al. 2005; Vance & Spalding 2005). There are other factors that influence the expression of CCMs in vascular plants, for example, facultative CAM plants on land and water availability, and temperature and submerged C₄ (Keeley & Rundel 2003).

7. Evolutionary origin of CCMs: when?

The CCM with the most recent origin that can be relatively accurately dated is C₄ in land plants, with fossil anatomical and stable isotope evidence from a grass at ca 12.5 Myr ago and from soil carbonate and herbivore teeth and skeleton carbon isotope ratios indicating sufficient local C₄ photosynthesis to substantially alter this proxy for plant organic carbon (Keeley & Rundel 2003). Molecular clock data suggest an origin 20–30 Myr ago (Keeley & Rundel 2003). For CAM, the pathway in terrestrial and aquatic flowering plants and Welwitschia evolved less than 120 Myr ago subsequent to the origin of the gnetophytes and flowering plants, the ancestral condition in these clades being C₃ physiology (Keeley & Rundel 2003). Carbon isotope evidence for CAM 200 Myr ago could be related to the occurrence of CAM cycling in an extant cycad (Decker & de Wit 2006; see Griffiths et al. (2007) for the interpretation of carbon isotope data in CAM). Extant ferns with CAM are in a clade that originated in the Cretaceous (120–65 Myr ago). The lycopsid Isoetes, expressing CAM when submerged (or effectively submerged), has a fossil record from palustrine (marsh or wetland) habitats from 230 Myr ago (Keeley & Rundel 2003). Inorganic carbon availability in submerged palustrine habitats is not, as Keeley & Rundel (2003) point out, closely coupled to the atmospheric CO₂ level. There are very few carbon isotope data consistent with terrestrial CCMs during the low CO₂ episode (Royer et al. 2007) in the Carboniferous some 300 Myr ago (Keeley & Rundel 2003; Osborne & Beerling 2006).

Badger et al. (2002), following Raven (1997b), suggest that CCMs in cyanobacteria and algae evolved in the low CO₂ (and high O₂) environment of the Carboniferous. There is also the possibility of CCMs evolving in ice house, and presumably low CO₂, episodes in the Proterozoic at approximately 2.4, 0.75 and 0.6 Gyr ago (Giordano et al. 2005; Harland 2007). Although there seem to be no carbon isotope data in support, we have seen that high δ¹³C values are not inevitable consequences of the operation of CCMs. A major problem is the maintenance of CCMs in the high CO₂ intervals, up to hundreds of millions of years, between low CO₂ episodes (Giordano et al. 2005; Sheldon 2006; Came et al. 2007; Royer et al. 2007). There were clearly cyanobacteria at the earliest ice house time mentioned above, eukaryotic oxygen evolvers for the two later ice houses (Brocks & Pearson 2005; Giordano et al. 2005), although the biomarker and structural fossil evidence do not permit conclusions as to the occurrence of CCMs.

Although not necessarily indicating the time of origin of algal CCMs, there is carbon isotope evidence (not a result of changed atmospheric δ¹³CO₂) from marine sediments in the latter part of the Cenozoic (the last 15 Myr) that CCMs became more significant in the ocean as well as on land as the CO₂ content of the atmosphere (and the ocean) decreased (Katz et al. 2005). The increasing δ¹³C through the latter part of the Cenozoic could also be explained in terms of diffusive CO₂ entry if there was an increasing cell size, based on the allometry of photosynthetic rate with the surface area per unit volume and the increasing restriction on photosynthesis by diffusion boundary layers around the cell (Korb et al. 1996; Finkel et al. 2005; Katz et al. 2005). However, the available data show that for diatoms, the dominant contributors to marine primary productivity at that time, the mean cell size decreased rather than increased with time in the Cenozoic (Finkel et al. 2005; Katz et al. 2005).

8. Evolution of CCMs: hypotheses about stromatolites

Giordano et al. (2005) and Raven & Larkum (2007) show that cyanobacteria in stromatolites could have low intracellular CO₂ concentrations and high O₂ concentrations, despite high external CO₂ and negligible O₂ (see Anbar et al. 2007; Kaufman et al. 2007) in the medium some 2.3–2.5 Gyr ago. The reasoning here is that the cyanobacteria in a stromatolite are separated from the medium by a millimetre or centimetre of unstirred interstitial water in a mineral and organic polymer, which adds tortuosity to the aqueous diffusion path (although the possibility of photomotion by the cyanobacteria, altering their distribution on the illuminated relative to the shaded side, must be acknowledged; Batchellor et al. 2004). Here we explore the implications of UV screening as a factor favouring the occurrence of cyanobacteria in stromatolites and other dense benthic populations. We discuss the indirect and direct effects of high UV flux on the intracellular environment and thus the influence on the potential environment that early CCMs would be exposed to.

Our suggestion is based on the observations that stromatolites occur, and occurred, intertidally or in shallow water, where fluxes of UV relative to PAR incident on the Earth's atmosphere were higher more than 2.4 Gyr ago, and that there was no stratospheric O₃ layer that would have attenuated UVC>UVB>UVA>PAR. The high UV flux at the surface of the ocean, and the absence of a great depth of seawater over the stromatolites even with plausible levels of organic compounds dissolved in the seawater, would not significantly attenuate UV relative to PAR at the stromatolite surface relative to the water surface. The minerals and UV-screening organic compounds produced by the cyanobacteria between the surface of the stromatolites and the cyanobacteria act as not only a UV screen (not a new suggestion!) but also, and incidentally, a barrier to diffusion of inorganic carbon and of oxygen (Giordano et al. 2005; Raven & Larkum 2007).

The restriction on inorganic C supply to the cyanobacteria could, as suggested by Giordano et al. (2005) and Raven et al. (2008), have implications for the evolution of CCMs as well as of form IA and form IB rubiscos. The corresponding build-up of O₂ has received rather less attention, except in relation to rubisco oxygenase activity (Giordano et al. 2005; Raven & Larkum 2007; Raven et al. 2008). Such O₂ build-up might be the first encounter that organisms had with high O₂ concentrations, and hence with high rates of production of reactive oxygen species such as ¹O₂, ^·OH, ^·${\text{O}}_2^{-}$ and H₂O₂. To these can be added O₃, since any UV not attenuated by the extracellular inorganic and organic screening compounds can generate O₃ from photosynthetically produced O₂ and from cyanobacterial mats in an initially anoxic atmosphere.

The production of ozone from photosynthetically produced oxygen was demonstrated in experiments subjecting cyanobacterial mats to a simulated early Earth atmosphere (10% CO₂/90% N₂) and a short-wavelength UV source (see electronic supplementary material; C. S. Cockell 2002, unpublished data). The mats were incubated in quartz tubes purged with the simulated atmosphere and exposed to a 185 nm source modulated to deliver the same biologically effective irradiance as that expected on the surface of the Archaean Earth in the absence of any atmospheric UV screens apart from carbon dioxide.

After 10 min of exposure, tubes containing no mats or mats that had been heat killed did not exhibit significant production of ozone (approx. 0.2 ppm O₃). The present-day (21% O₂) atmosphere supported the formation of 1.4 ppm O₃. Tubes with live cyanobacterial mats and the simulated early Earth atmosphere showed a net production of 0.7 ppm O₃, formed from the interaction of low-wavelength UV radiation with the photosynthetically produced oxygen. This value was significantly higher than all the controls assessed by Student's t-test (p<0.01).

Unlike the other reactive oxygen species, which are generated within the cells, O₃ can be generated outside the cells as well. There would have been opposing gradients of O₂ and of UV, with the concentration of O₂ highest within the cells and the UV flux lowest within the cells. Knowing the relationship between UV flux and O₂ concentration in producing O₃ would allow a quantitative model to be produced of the O₃ production rate along the gradient from the cells to the surface assuming an exponentially increasing UV flux and linearly decreasing O₂ concentration, further assuming no O₂ consumption by non-photosynthetic biota or inorganic reductants between the cyanobacteria and the surface of the structure. A problem with assessing the biological impact of the O₃ generated in and around the cyanobacterial cells is uncertainty as to the solubility of O₃ and the rate at which it would be consumed by chemical reactions with the organic and inorganic compounds outside the cells and, indeed, simply by being in aqueous solution. A further complication in assessing the role of reactive oxygen species and ozone in the evolutionary context is the possible role of the cyanobacterial sheath. A thick sheath will retard the diffusion of oxygen out of the cell, making it more likely that biologically detrimental concentrations of free radical species and ozone would be achieved intracellularly. However, a thick sheath will concomitantly protect the cell from free radicals and ozone in the intercellular environment that results from other cells' oxygen production. Conversely, thin sheaths provide less protection from the intercellular toxic species, but allow for faster diffusion of these species from the intracellular environment.

The effects of O₃ on aquatic photosynthetic organisms have not been investigated to the same extent as for their terrestrial counterparts, presumably owing to the uncertainties that have just been mentioned and the assumption that the effects of dissolved O₃ would be mediated largely by the other reactive oxygen species resulting from its reactions in solution. However, ozone readily mixes with water and its oxidizing abilities, more powerful than even those of chlorine, are used today to decontaminate drinking water and disinfect surfaces (Moore et al. 2000). At a concentration of 2 mg l⁻¹ (equivalent to 2 ppm by weight), 90% inactivation of some bacterial species is achieved in less than 20 s (Lezcano et al. 1999). Concentrations as low as 0.33 mg l⁻¹ can cause complete inactivation of Staphylococcus species in 7 min (Lezcano et al. 2000).

The scheme we have just outlined for cyanobacterial microbialites involves the addition of damage by reactive oxygen species (and effects of O₂ per se; Raven & Larkum 2007) to that from UV, with some of the load of reactive oxygen species resulting from absorbed UV radiation. This scheme presents the earliest occurrence in microcosm of the later global occurrence of damage from reactive oxygen species as well as of that from UV for organisms living in the surface ocean or on land. The UV screens in the microbialites (and then in other organisms) were supplemented, with atmospheric oxygenation, by the UV screening involving O₃.

Can we learn anything from the response of extant organisms to UV and reactive oxygen species that bear on the plausibility of the scheme that we have proposed on UV screening and the origin of CCMs? For the effects of UV on CCMs, we have no information on cyanobacteria and only a few data from algae, yielding responses varying from CCMs being less sensitive to being more sensitive to UV than is the rest of photosynthetic metabolism (Beardall et al. 2002; Sobrino et al. 2004). Data for terrestrial flowering plants do not permit definitive conclusions as to the UV sensitivity of the C₄ CCM (Johnson & Day 2002; Correia et al. 2005). For evidence as to any increased production of reactive oxygen species as a result of operation of CCMs and the related higher intracellular O₂ concentration (Raven & Larkum 2007) we can turn, in the absence of direct evidence, to changes in gene expression as the CCM expression is increased when cultures of the green alga C. reinhardtii and the glaucocystophyte alga Cyanophora paradoxa grown in high CO₂ are transferred to low CO₂. Here there is increased transcription of genes associated with increased production of oxidants (Im et al. 2003; Burey et al. 2007). Further investigation is needed.

9. CCMs in the future

Given the current and marked anthropogenically induced rise in atmospheric CO₂ concentrations, a widespread view is that CCMs will become less competitive relative to organisms relying on diffusive CO₂ influx to C₃ biochemistry (see Keeley & Rundel 2003; Giordano et al. 2005). However, as shown in §3, other interrelated environmental changes, such as temperature increase, changes in the depth of the ocean upper mixed layer and variations in the supply of resources other than C, will interact with the increase in CO₂.

In the much more distant future, the temporal trajectory of the Sun is such that the trend of increasing output of electromagnetic radiation seen over the Sun's existence will continue so that the Earth will be too hot to be habitable well before the Sun expands to a radius similar to that of the Earth's orbit as a supernova before it becomes a red dwarf (Franck et al. 2006). Thermal death of life on Earth can be delayed if greenhouse gas levels in the atmosphere are minimized, although such a reduction would, for CO₂, have implications for photosynthesis (Rampino & Caldeira 1994). This point was first considered by Lovelock & Whitfield (1982) in the context of C₃ physiology of photosynthesis; they suggested that CO₂ would be down to the lower limit for photosynthesis in 100 Myr, if this gas was acting as a major component of Earth's temperature regulation (cf. Berner & Kothavala 2001). Caldeira & Kasting (1992) revisited this question, and pointed out that C₄ photosynthesis would, through a higher affinity for CO₂ and a lower CO₂ compensation concentration, permit photosynthetically powered ecosystems to persist for an additional several hundreds of millions of years to 0.9–1.5 Gyr. Such ‘salvation through CCMs’ is independent of the means by which the decreased biosphere CO₂ concentration is achieved.

Supplementary Material

Supplementary methods

Methods related to experiments on ozone production by cyanobacteria