Internal carbonic anhydrase activity in the tissue of scleractinian corals is sufficient to support proposed roles in photosynthesis and calcification

Active inorganic carbon (C_i) supply pathways are required to support the high rates of photosynthesis and calcification that have allowed reef-building corals to thrive and create the extensive reef structures that support coral reef ecosystems. These pathways must deliver C_i to the symbiotic dinoflagellates (of the genus Symbiodinium) that reside primarily within the upper endodermal layer of the coral (Furla et al., 2000b; Bertucci et al., 2013). The algae further concentrate C_i obtained from the coral using a CO₂ concentrating mechanism and ultimately convert the C_i to CO₂ for fixation in the Calvin cycle (Leggat et al., 1999). At the same time, C_i is routed to the calcifying space at the interface between the existing CaCO₃ skeleton and the ectodermal cells of the coral animal where the C_i is converted to CO₃²⁻ and combined with Ca²⁺ to form new skeleton (Tambutte et al., 2011). Much of the carbon fixed by the symbiotic algae is transported to and then respired by the coral. The coral recycles most of the respired CO₂, using it for both photosynthesis and calcification (Furla et al., 2000b). The full details of these C_i supply and recovery pathways are not fully understood, but it is clear that they are complicated and interconnected (Fig. 1; Tambutte et al., 2011; Bertucci et al., 2013).

A critical component of nearly all C_i supply pathways is the enzyme carbonic anhydrase (CA). This enzyme catalyzes the hydration of CO₂ and dehydration of HCO₃⁻ moving the CO₂/HCO₃⁻ system towards chemical equilibrium. CAs play diverse roles in organisms and are involved in CO₂ uptake, gas exchange, pH homeostasis and internal C_i transport pathways (Henry, 1996; Badger, 2003; Boron, 2004). In corals and related cnidarians, CAs are involved in supplying C_i for both photosynthesis and calcification. A surface-associated extracellular CA is involved in CO₂ uptake (Furla et al., 2000a; Tansik et al., in press), and internal CAs are likely used to convert newly imported CO₂ to HCO₃⁻, and for recovery of respired CO₂ as HCO₃⁻ (Fig. 1; Bertucci et al., 2013). These transformations serve to concentrate C_i as HCO₃⁻, which is the form of C_i typically concentrated by organisms because as a charged molecule it does not pass easily through membranes. In contrast, biological membranes are highly permeable to CO₂, a small, uncharged molecule (Gutknecht et al., 1977). Accumulated HCO₃⁻ can then be transported to the symbiotic algae, possibly via SLC4-type bicarbonate transporters, which have recently been identified in the genomes and transcriptomes of several coral species (Moya et al., 2012; Kenkel et al., 2013), or converted to CO₂ in the oral endoderm for uptake by Symbiodinium (Al-Moghrabi et al., 1996; Bertucci et al., 2010). Inhibition of CAs reduces the photosynthetic rates of corals, showing that this enzyme is essential for maintaining high rates of photosynthesis (Weis et al., 1989; Al-Moghrabi et al., 1996).

CAs are also involved in calcification (Goreau, 1959), probably as part of the C_i transport system in coral tissue and within the calcifying fluid to enhance calcification. C_i used for calcification is a mix of respired CO₂, HCO₃⁻ transported through coral tissue, and C_i obtained from direct seawater import to the calcifying fluid (Furla et al., 2000b; Cohen and McConnaughey, 2003; Gagnon et al., 2012). The C_i transport pathways for calcification are thus broadly similar to those used to transport C_i for photosynthesis, and so the likely roles of CA in these transport steps are analogous. In several corals (Stylophora pistillata, Acropora hebes and Tubastrea aurea), CA has been localized directly to the calicoblastic ectoderm, the cell layer that mediates calcification and is in direct contact with the calcifying fluid, or the skeletal matrix and it has been suggested that CA may be secreted into the calcifying fluid (Isa and Yamazato, 1984; Tambutte et al., 2007; Moya et al., 2008). CA in the calcifying fluid would convert respired CO₂ to HCO₃⁻ for subsequent conversion to CO₃²⁻ (Fig. 1; Cohen and McConnaughey, 2003).

CAs are clearly important in corals and there has been substantial progress in identifying, localizing and in some case characterizing individual CAs. Weis et al. (1989) found that corals contain substantial amounts of CA activity and that zooxanthellate species had significantly more CA than species lacking symbiotic algae. While this work showed that corals have high levels of CA activity, the approach used was not quantitative in an absolute sense, that would, for example, allow the measured rates to be incorporated into models of photosynthesis and calcification (e.g. Hohn and Merico, 2012; Nakamura et al., 2013). Here, we adapted a method that relies on the rate of ¹⁸O-removal from labeled C_i to obtain quantitative measurements of internal CA (iCA) activity in corals – in this case Orbicella faveolata (Ellis and Solander 1786), Porites astreoides Lamarck 1816 and Siderastrea radians (Pallas 1766). As the ¹⁸O-removal measurement is made on coral homogenate, the method provides a measure of the average iCA activity throughout the corals tissue layers including the calcifying fluid. It is highly unlikely that CA is in fact distributed homogeneously throughout the tissue, and the need for CA localization is explored using a spatially resolved model of C_i processing.

Coral iCA activity was determined by quantitative analysis of ¹⁸O-removal catalyzed by purified coral tissue homogenate using a model that includes coral iCA and external (e)CA activity, and Symbiodinium iCA activity (see Materials and methods). The method relies on the fact that all CA isoforms accelerate the hydration of CO₂ and dehydration of HCO₃⁻, which in turn accelerates the exchange of the ¹⁸O label in C_i for ¹⁶O from water (Silverman, 1982). In this analysis, CA activity in the purified tissue homogenate is partitioned into various experimentally separable fractions. eCA activity is first measured on intact coral fragments (Tansik et al., in press), and then the coral tissue is removed and homogenized (Fig. 2A). Symbiodinium cells are removed from the homogenate by centrifugation and their CA activity is measured. Finally, the purified tissue homogenate is assayed for CA activity, and the iCA activity of the coral is determined by subtracting the contributions of coral eCA and residual Symbiodinium iCA from the total CA activity in the purified homogenate.

Determination of coral iCA activity using this approach first requires measurement of coral eCA activity and the algal iCA activity and mass transfer coefficients. The coral eCA activity was measured by an ¹⁸O-removal assay on intact coral fragments. These data are presented elsewhere (Tansik et al., in press) and the eCA activities are listed in Table 1. eCA activity varied substantially between taxa with O. faveolata having the highest eCA activity, approximately one order of magnitude greater than that of S. radians and P. astreoides. Algal iCA activity and mass transfer coefficients were measured using an ¹⁸O-removal assay on cells freshly isolated from coral tissue. Algal iCA activity also differed among taxa, although activities in all taxa were very high compared with those in other microalgae (Tu et al., 1986; Hopkinson et al., 2013). Symbiodinium from S. radians had the highest iCA activity, followed by those from P. astreoides, and then Symbiodinium from O. faveolata. CO₂ and HCO₃⁻ mass transfer coefficients were similar among the taxa, and showed that the cell membranes are highly permeable to CO₂ but impermeable to HCO₃⁻ like other microalgae (Table 1).

After obtaining coral eCA and algal parameters, coral iCA activity was determined from analysis of the effect of purified coral tissue homogenate on ¹⁸O-removal rates. The coral homogenate substantially accelerated the rate of ¹⁸O-removal (Fig. 2, Table 2). Addition of the CA inhibitor acetazolamide (AZ) slowed ¹⁸O-removal rates to background rates (or even slightly below background rates), verifying that CA activity was responsible for the acceleration (Table 2). Although the potency of AZ differs with CA isoform, the 100 μmol l⁻¹ AZ added should be sufficient to fully inhibit most CA isoforms as typical inhibitory constants are <50 nmol l⁻¹ (Moya et al., 2008). The fact that ¹⁸O-removal rates returned to background levels after addition of AZ confirms that CA activity was effectively inhibited for these corals. Analysis of these data using the ¹⁸O-removal model presented in Eqns 3–6 (see Materials and methods) shows that the model fits the data well for all taxa, allowing accurate determination of coral iCA activity. The coral iCA activity can either be normalized by the coral tissue volume to obtain a first-order rate constant, or normalized by coral surface area. The advantage of the first-order rate constant is that it is directly relevant to internal C_i fluxes, can be applied to models of C_i processing in corals and is intuitive. However, normalization by volume treats the iCA as being homogeneously distributed throughout the internal volume of the coral, which is not likely to be the case. When normalized to coral tissue volume, S. radians has approximately 10-fold higher iCA activity than both O. faveolata and P. astreoides (t-test, P<0.05), but the iCA activities of O. faveolata and P. astreoides are not significantly different from one another (Fig. 3).

In contrast, when normalized to coral surface area, O. faveolata and S. radians have similar iCA activity, while P. astreoides has approximately half the iCA activity of the other two species. However, only the difference between S. radians and P. astreoides is statistically significant (t-test, P<0.05). The overall range in coral iCA activity between taxa is lower when normalized to surface area (2.8×) than when normalized by volume (9×), and it is notable that photosynthetic rates per unit area are very similar among the coral species (Fig. 3).

Obtaining the coral iCA activity requires accounting for the effects of coral eCA and algal iCA on ¹⁸O-removal catalyzed by the purified tissue homogenate (see Materials and methods; supplementary material Fig. S1). The greater the contribution of these additional components to ¹⁸O-removal, the more difficult it is to accurately determine coral iCA activity. We assessed the effects of these additional components in two ways: first, graphically, showing how the ¹⁸O-removal model predicts that ¹⁸O-CO₂ isotopologues would evolve as components are added, and second, by determining how neglecting the additional components in the model would affect inferred coral iCA activity.

We ran the ¹⁸O-removal model in a forward (predictive) sense first, considering how the ¹⁸O-CO₂ isotopologues would evolve in the absence of any CA activity, and then sequentially adding the effects of algal iCA, coral eCA and coral iCA. Fig. 4 shows representative results for each taxon, but similar results were obtained for each sample within a taxon. The effect of residual Symbiodinium on ¹⁸O-removal was quite small in all taxa. Centrifugation removes more than 98% of the Symbiodinium cells from the crude homogenate, which is sufficient to effectively eliminate any interference for determination of coral iCA activity. In general, the effects of coral eCA were also small, though in O. faveolata, which had the highest eCA activity, coral eCA could have a substantial effect on ¹⁸O-removal. This analysis indicates that accounting for the effect of coral eCA activity is important, but that residual Symbiodinium contribute little to ¹⁸O-removal as long as the centrifugation is effective.

To make a more quantitative assessment, we examined the effects of neglecting coral eCA and algal iCA on the inferred coral iCA activity. When these components are neglected in the model, their contribution to ¹⁸O-removal will be attributed to coral iCA activity, the unknown model parameter, increasing the inferred coral iCA activity. As shown in Fig. 5, neglecting algal iCA resulted in only minor increases in the inferred coral iCA activity (<7%), consistent with the graphical analysis showing that residual algal cells have negligible effects on ¹⁸O-removal. For P. astreoides and S. radians, neglecting coral eCA activity led to at most moderate increases in inferred coral iCA activity (<15%), but for O. faveolata the effects were much more significant, especially for one specimen that had low iCA activity.

A one-dimensional model of coral C_i processing that includes active HCO₃⁻ transport, passive CO₂ transport, CA-catalyzed interconversion of CO₂ and HCO₃⁻, photosynthesis and calcification was used to help interpret the measured iCA activities. C_i processing pathways in corals are still under study and are in some cases controversial. For example, electron microscopy of the tissue–skeleton interface suggests that there is little to no calcifying fluid (Clode and Marshall, 2002), but more sizable reservoirs of calcifying fluid have been detected as high pH regions in confocal microscopic imaging (Venn et al., 2011). The C_i pathways in the model were chosen from among proposed routes in a way that maximizes iCA requirements (using iCA in all the reactions shown in Fig. 1). The model is schematic and is used here only to assess the suitability of measured iCA activity to perform proposed functions and to explore iCA requirements. It is not a general mechanistic model of coral C_i processing. Most notably, the metabolic fluxes (photosynthesis, respiration, calcification, HCO₃⁻ transport) are specified based on measured rates or to achieve mass balance, rather than being determined mechanistically using enzyme kinetics.

First, the model was run with an iCA activity of 50 s⁻¹, approximately what was measured in O. faveolata and P. astreoides, distributed homogeneously throughout the coral tissue layers and calcifying fluid (Fig. 6). This iCA activity was able to support rates of photosynthesis, calcification and respiration measured in O. faveolata, but in the case of calcification the iCA activity was only just sufficient. In the oral ectoderm, iCA catalyzes the conversion of newly imported and respired CO₂ to HCO₃⁻. Although CO₂ concentrations in this layer are lower than in seawater, driving the CO₂ influx, they are in fact above equilibrium concentrations with HCO₃⁻ because HCO₃⁻ concentrations are low as a result of active export to the underlying oral endoderm. The active transport of HCO₃⁻ across membranes would be mediated by transporters such as the SLC4-type HCO₃⁻ transporters recently identified in corals (Moya et al., 2012; Kenkel et al., 2013). In the oral endoderm, HCO₃⁻ is imported and converted to CO₂ by iCA. The CO₂ is then consumed by the Symbiodinium that reside in the oral endoderm layer, lowering CO₂ concentrations below equilibrium values. CO₂ and HCO₃⁻ are in equilibrium in the coelenteron where no metabolic processes occur. In the aboral tissue layer, respiration produces CO₂, part of which is converted to HCO₃⁻ by CA and the remainder of which diffuses into the calcifying fluid where it is then converted to HCO₃⁻ by CA for use in calcification.

The iCA activity in the calcifying fluid was only just sufficient to support the measured rates of calcification for O. faveolata. This can be most clearly seen from the low HCO₃⁻ concentration in the calcifying fluid, indicating high rates of removal relative to input, and in the nearly 100% deviation of CO₂ concentrations from equilibrium with HCO₃⁻ (Fig. 6C). Greater deviations from equilibrium (either positive or negative) indicate that maximal CO₂ hydration or HCO₃⁻ dehydration fluxes are being reached because the maximal net fluxes occur when the product concentration (HCO₃⁻ for hydration, CO₂ for dehydration) is zero. Consequently, the deviation of CO₂ from equilibrium with HCO₃⁻ is used as an index of the effectiveness of the C_i processing system, with moderate deviation (<25%) being acceptable, but further deviation indicating the system is near its maximal capacity. We note that in our simplistic model, meant only to explore the role of iCA, there is no dependence of the calcification rate on HCO₃⁻ or CO₃²⁻ concentrations. The low HCO₃⁻ and CO₃²⁻ concentrations in this model would probably not provide a high enough aragonite saturation state to obtain the observed calcification rates (e.g. Burton and Walter, 1987).

Running the model with higher (500 s⁻¹) and lower (5 s⁻¹) iCA activities shows that the observed iCA activity is nearly optimal for supporting the required C_i fluxes and keeping CO₂ and HCO₃⁻ concentrations close to equilibrium (Fig. 6). Decreasing iCA 10-fold leads to significant deviations of CO₂ from equilibrium with HCO₃⁻ and the model is only able to support calcification at 40% of the observed rate. This results in a lower HCO₃⁻ concentration in the calcifying fluid compared with that in models with higher iCA activity. Increasing iCA 10-fold does lead to nearly complete equilibrium between CO₂ and HCO₃⁻ throughout the coral, but there are no performance gains directly associated with this increase.

The C_i processing model was used to explore the effect of changes in photosynthetic rate and tissue thickness on required iCA activity. The model with a homogeneous iCA activity similar to that measured in the reef corals (50 s⁻¹) and metabolic rates of O. faveolata was used as a reference point. First, the net photosynthetic rate was varied (with the respiration rate being set equal to the net photosynthetic rate) and the iCA activity was optimized to obtain the same average CO₂ deviations in each tissue layer as in the base model. The average CO₂ deviations are used here as a measure of the effectiveness of CA at supplying C_i for transport or metabolism. The required iCA activity scaled approximately linearly with the photosynthetic rate (Fig. 7A). Next, the tissue thickness was varied from the reference point (2 mm), keeping the metabolic rates per unit surface area constant, and the iCA activity was optimized to match CO₂ deviations to the base model. The required iCA activity per unit volume increased rapidly as the tissue thickness decreased and conversely decreased as the tissue thickness increased (Fig. 7B). However, the iCA activity per unit surface area was nearly constant, increasing slightly with increasing tissue thickness.

CA is a widely distributed enzyme having many isoforms and diverse roles in organisms. It has long been recognized to be important in marine organisms, some of the earliest work being in mollusks (Freeman and Wilbur, 1948; Wilbur and Jodrey, 1955), and more recently its roles in corals and other cnidarians have been investigated (Bertucci et al., 2013). We have developed a method to determine the iCA activity of corals by measuring the acceleration of ¹⁸O-removal from labeled C_i catalyzed by purified coral tissue homogenate (Fig. 2, Table 2). iCA activity was measured in three species of Caribbean corals. We found that the two species that inhabit offshore reefs (O. faveolata, P. astreoides) had similar iCA activity when normalized to tissue volume, whereas a species from Florida Bay (S. radians) had ∼10-fold higher iCA activity (Fig. 3). In contrast, when normalized to coral surface area, O. faveolata and S. radians had similar iCA activities, while the iCA activity of P. astreoides was lower. Determination of coral iCA activity required removing contributions from residual Symbiodinum cells and coral eCA from the total CA activity of the purified coral tissue homogenate. Residual Symbiodinium cell numbers were always low and did not significantly affect determination of coral iCA activity (Figs 4, 5). In P. astreoides and S. radians, eCA activity was also not a major issue, but in O. faveolata, which had very high eCA activity, accounting for eCA activity was essential for accurate determination of coral iCA activity.

The coral iCA activity measurement includes all CAs in the cytoplasm, on internal membranes, in the coelenteron and in the calcifying fluid. Membrane-bound CAs have been localized near the symbiosome membrane presumably to facilitate C_i delivery to Symbiodinium (Weis, 1993; Al-Moghrabi et al., 1996), and could also be facing the coelentron, or attached to the calicoblastic cells to aid calcification (Moya et al., 2008). iCAs have also been identified within the cytoplasm of the oral and aboral tissue (Grasso et al., 2008; Bertucci et al., 2011), and within calicoblastic cells (Moya et al., 2008), but there are many more CAs in coral genomes that have yet to be localized.

As the iCA measurements required homogenizing the tissue, the measured iCA activity may not precisely match in vivo activity. This is always a concern with enzyme assays, but is somewhat less of an issue for CA because it does not require an energy source, though its activity can be regulated by cellular conditions such as redox status (Kikutani et al., 2012). The in vivo iCA activity provides complementary rate information to molecular work on coral CAs that has provided more detailed information about CA localization and hence the potential role of individual CA isoforms. While molecular approaches can provide quantitative information about CA activity, this is quite time consuming and difficult, requiring over-expression of active CA proteins followed by assessment of CA activity, and is only rarely done (Moya et al., 2008; Bertucci et al., 2011).

The net CO₂ or HCO₃⁻ generation rate will depend on the CO₂ and HCO₃⁻ concentrations and pH in the coral tissue, and the rates and net direction of the reaction will likely vary in different regions of the coral. Nonetheless, the gross generation rates put limits on the net generation rates.

Assuming the CO₂ concentration in the coral is 10 μmol l⁻¹, similar to seawater concentrations, the gross iCA-catalyzed HCO₃⁻ generation rate for O. faveolata would be 3.4×10⁻⁴ mol cm⁻² h⁻¹. This rate is many orders of magnitude higher than the net photosynthetic rate we measured for O. faveolata (5×10⁻⁷ mol cm⁻² h⁻¹), and as calcification rates are generally comparable to or less than the net photosynthetic rate, the gross HCO₃⁻ generation rate also greatly exceeds the calcification rate. Similarly, the gross HCO₃⁻ generation rates for P. astreoides and S. radians greatly exceed net photosynthetic rates. Thus, there is in principle sufficient iCA activity to convert CO₂ imported for photosynthesis to HCO₃⁻ (Fig. 1, arrow 2), and to convert respired CO₂ to HCO₃⁻ for calcification (Fig. 1, arrows 4 and 5). The gross CO₂ generation rate, calculated assuming that the average internal pH is 7.4 (Venn et al., 2009,, 2011) and that the internal HCO₃⁻ concentration is 1.8 mmol l⁻¹, similar to seawater, is 1.6×10⁻³ mol cm⁻² h⁻¹. Again, the gross CO₂ production rate is orders of magnitude higher than the net photosynthetic rate, showing there is sufficient iCA activity to produce CO₂ for uptake by symbiotic algae (Fig. 1, arrow 3). These gross production rates are of course maximal net production rates, and use estimates for internal CO₂, HCO₃⁻ and H⁺ concentrations. But the fact that they are many orders of magnitude higher than the required net production rates suggests that there should be enough iCA capacity to fulfill any of these roles.

For a more detailed assessment, a spatially resolved model of C_i processing in the coral tissue was developed. Because C_i processing pathways have not been fully defined in corals, the model is necessarily schematic, and iCAs are used in as many roles as possible to maximize the need for iCA activity. The model was first run with an iCA activity similar to that measured in O. faveolata and P. astreoides (50 s⁻¹) distributed homogeneously throughout the coral tissue. This iCA activity and distribution are sufficient to support realistic rates of photosynthesis and calcification in the model, though they are only just sufficient to support a realistic calcification rate (Fig. 6). The ability of the measured iCA activity to catalyze the multiple hydrations and dehydrations needed to move C_i from seawater to Symbiodinium and to the calcifying fluid is consistent with the simpler calculations showing that the potential CO₂ and HCO₃⁻ generation rates from iCA greatly exceed those required for individual steps in the C_i processing pathways. The measured iCA activity is neither deficient nor greatly excessive as shown by model runs with 10-fold higher or lower iCA activity. Higher iCA activity (500 s⁻¹) does not offer any direct benefit in the model, and the lower iCA activity (5 s⁻¹) is only barely sufficient to meet the C_i requirements of photosynthesis and does not supply enough C_i to support measured rates of calcification.

The fact that a homogeneous distribution of iCA is capable of meeting C_i needs does not imply that iCA is or should be homogeneously distributed. The calcifying fluid and oral endoderm would benefit from greater CA activity as shown by the significant deviation of CO₂ from equilibrium with HCO₃⁻ in these model layers (Fig. 6). The observed localization of CA in the calicoblastic cells and potentially in the calcifying fluid (Isa and Yamazato, 1984; Tambutte et al., 2007; Moya et al., 2008) may be particularly helpful as CaCO₃ is being rapidly precipitated from a very thin layer of calcifying fluid (∼3 µm thick; Venn et al., 2011), necessitating high rates of CO₂ conversion to HCO₃⁻ within a very small volume of fluid. It may also be detrimental to have CA activity in all parts of the tissue. For example, cyanobacteria lack CA in the cytoplasm, allowing them to accumulate HCO₃⁻, and if CA is artificially expressed in the cytoplasm, HCO₃⁻ is rapidly converted to CO₂, which leaks out of the cell (Price and Badger, 1989). Storage of HCO₃⁻ in a tissue layer or vacuole within the coral may require CA to be absent from that space.

Coral iCA activities were determined in three species of coral, O. faveolata and P. astreoides, which inhabit offshore reefs, S. radians, which is found in the shallow, soft-bottomed environment of Florida Bay. When normalized by tissue volume, the two reef species show similar iCA activity, while S. radians has about 10-fold higher iCA activity (Fig. 3). The slow circulation in the bay may limit the transfer of C_i between bulk seawater and the surface of S. radians, requiring increased iCA activity to help facilitate C_i acquisition. Lesser et al. (1994) showed that total CA activity of the coral Pocillopora damicornis increased when exposed to low flow in a controlled flume experiment. However, S. radians also has the thinnest effective tissue thickness (tissue volume per unit surface area; Table 1) of the corals surveyed and when coral iCA activity is instead normalized by surface area we find that S. radians and O. faveolata have similar iCA activities with P. astreoides having ∼50% lower iCA activity (Fig. 3).

As coral iCA is involved in C_i supply for photosynthesis and calcification, the similar rates of photosynthesis (and so presumably of calcification) per unit surface area for these corals would suggest that they need similar amounts of iCA on an areal basis, an intuition that we explored with the spatially resolved model of C_i processing. When photosynthetic rates were varied, the iCA activity required to maintain constant effectiveness of the C_i processing system in the model, as indicated by CO₂ deviations from equilibrium, increased or decreased approximately linearly with the photosynthetic rate (Fig. 7A). Next, photosynthetic rates were kept constant, while coral tissue thickness was varied. The iCA activity per unit surface area required to maintain constant CO₂ deviations from equilibrium remained essentially constant, though there was a slight increase in iCA activity per unit surface area as tissue thickness increased, a result of its role in facilitated diffusion of C_i through the tissue layers (Fig. 7B; Enns, 1967). In contrast, the iCA activity per unit volume must increase dramatically as the metabolic rates are squeezed into a thinner tissue layer and can decrease as the tissue expands. This analysis suggests that the primary reason iCA activity per unit volume is higher in S. radians is because of the thinner effective tissue thickness in this organism and not because of environmental differences, whereas the iCA activity per unit surface area is more similar among species because the metabolic processes iCAs are involved in are relatively constant on an areal basis.

In summary, we have developed a method to measure the absolute iCA activities of corals and have applied this method to three species of scleractinian corals found in the Florida Keys. A key advantage of these absolute measurements is that they allow us to assess the ability of measured iCA activity to fulfill proposed roles in photosynthesis and calcification. Using simple back of the envelope calculations and a more detailed spatially resolved model, we showed that the iCA activities are capable of sustaining the rapid rates of CO₂/HCO₃⁻ interconversion required to deliver C_i for photosynthesis and calcification.

Fragments of Orbicella (previously Montastraea) faveolata, Porites astreoides and Siderastrea radians were collected from Little Grecian reef and Florida Bay in Key Largo, FL, USA, in August of 2013 as permitted by the Florida Keys National Marine Sanctuary (FKNMS-2011-093, FKNMS-2014-015). Exposed skeleton was covered with modeling clay, and the colonies were maintained in closed circulation tanks filled with reef seawater, allowing at least 2 days of recovery after collection prior to experimentation. The tank was exposed to a natural light regime, with shading added at midday to keep solar irradiance below 600 μmol photons m⁻² s⁻¹. Coral surface area was measured using an aluminium foil method (Marsh, 1970). Coral tissue thickness was measured on freshly fragmented corals with an ocular micrometer on a stereomicroscope (Leica WILD M3Z, Wetzlar, Germany). The thickness varied between polyps and the coenosarc and multiple measurements across the colony were averaged to obtain an average tissue thickness (L_c) for each species (Table 1). The tissue of P. astreoides and S. radians is intermingled with the skeleton and the fractional occupancy of tissue (φ, porosity of the skeleton) was estimated from cross-sectional images of the fragmented coral colonies. The images were analyzed with ImageJ using color and brightness characteristics to segment the image between tissue and skeletal fractions. The effective tissue thickness, the thickness of the tissue if it was not intermingled with skeleton, was calculated as φL_c. Algal symbiont clades as determined by denaturing gradient gel electrophoresis (DGGE) analysis of the ITS2 sequence were: O. faveolata – B1; P. astreoides – A4a; S. radians – B5 (LaJeunesse et al., 2003).

eCA activity was measured on intact coral fragments using an ¹⁸O-removal technique described in Tansik et al. (in press). Briefly, a water-jacketed chamber is interfaced to a membrane inlet mass spectrometer (MIMS; QMS 220M2, Pfeiffer Vacuum, Asslar, Germany) that allows continuous monitoring of CO₂ isotopes. The chamber is filled with C_i-free artificial seawater (ASW), buffered with 20 mmol l⁻¹ Tris at pH 8.0, and ¹³C-¹⁸O-labeled C_i is added to the chamber. After monitoring ¹⁸O-removal for approximately 15 min to determine the background rate of removal, a coral fragment is added to the chamber and ¹⁸O-removal catalyzed by the coral is monitored for a further 10 min. Finally 100 μmol l⁻¹ of the eCA inhibitor dextran-bound AZ is added to the chamber and ¹⁸O-removal is monitored for 10 min. The data are fitted to a model of ¹⁸O-removal that accounts for background removal rates, iCA activity and eCA activity, with the measurement of interest being eCA activity.

To measure the iCA activity and mass transfer coefficients of the symbiotic algae, the algae were first isolated from the coral. Coral tissue was removed from the skeleton with an airbrush (Paasche, Chicago, IL, USA) using ASW with 20 mmol l⁻¹ Tris at pH 8.0 and containing a protease inhibitor cocktail (complete protease inhibitor cocktail mini tablets, 1 tablet per 20 ml). The algae were then separated from the coral homogenate by centrifugation at 2350 g for 5 min. iCA activity and the CO₂ and HCO₃⁻ mass transfer coefficients were measured on the algae using an ¹⁸O-removal method described previously (Tu et al., 1978; Hopkinson et al., 2011). Absolute iCA activity was inferred by fitting a model that treats the interior of the cell as a single homogeneous compartment to the ¹⁸O-removal data. The mass transfer coefficients describe controls on the flux of CO₂ and HCO₃⁻ into the cells, which is primarily controlled by the diffusive boundary layer and cell membrane permeability.

The CA activity of the coral tissue homogenate from which symbiotic algae had been removed (‘purified coral tissue homogenate’) was measured using an ¹⁸O-removal technique. ¹³C-¹⁸O-labeled C_i (2 mmol l⁻¹) was added to assay buffer (C_i-free ASW, 20 mmol l⁻¹ Tris at pH 8.0) in a small MIMS chamber that holds ∼1 ml of solution. Temperature in the chamber was maintained at 28°C with a water jacket. ¹⁸O-CO₂ species were monitored with the MIMS system for approximately 10 min to determine the background rate of removal due to CO₂ hydration/HCO₃⁻ dehydration. Then a sample of the purified homogenate was added to the chamber and the accelerated ¹⁸O-removal rate was monitored for approximately 10 min. In some cases a CA inhibitor, 10 μmol l⁻¹ AZ, was then added to confirm that the accelerated ¹⁸O-removal rate was due to CA activity. A sample of the purified homogenate was preserved in formalin to determine the residual concentration of symbiotic algae in the sample via microscopy.

The first two terms describe the chemical reactions: hydration of CO₂ and dehydration of HCO₃⁻, while the third term describes diffusion, and the final two terms represent transport (T) and the metabolic reactions (M): photosynthesis, respiration and calcification. k_f and k_r are the first-order rate constants for CO₂ hydration and HCO₃⁻ dehydration. D_c and D_b are the diffusion coefficients for CO₂ and HCO₃⁻, respectively, and z is the thickness of the model layer. pH in each tissue layer was set as shown in supplementary material Fig. S1, based on literature values (Venn et al., 2009,, 2011; Ries, 2011) except that the pH in the oral tissue layers had to be elevated somewhat to achieve a net CO₂ influx. HCO₃⁻ transport rates between the tissue layers were set to obtain steady-state concentrations in each layer, delivering C_i as needed for photosynthesis and calcification, and removing respired CO₂ (supplementary material Fig. S1).

Rates of gross photosynthesis (1×10⁻⁶ mol cm⁻² h⁻¹), respiration (5×10⁻⁷ mol cm⁻² h⁻¹) and calcification (1.7×10⁻⁷ mol cm⁻² h⁻¹) measured for O. faveolata were used as base rates in the model. The model was designed to maximize the need for iCA, employing iCA in as many consistent roles as possible as shown in supplementary material Fig. S1: conversion of imported and respired CO₂ to HCO₃⁻ in the oral ectoderm, conversion of HCO₃⁻ to CO₂ for uptake by Symbiodinium in the oral endoderm, and conversion of respired CO₂ to HCO₃⁻ in the aboral tissue and calcifying fluid. CO₂ and HCO₃⁻ concentrations were set at typical seawater values at the outer edge of the diffusive boundary layer and the model was then run until a steady state was reached.

We thank Dustin Kemp and Clinton Oakley for assistance with coral collection and maintenance, and Virginia Weis for advice on measurement of carbonic anhydrase activity in corals. Permission to collect corals was granted by the Florida Keys National Marine Sanctuary.