Intracellular pH and its response to CO₂-driven seawater acidification in symbiotic versus non-symbiotic coral cells

Reef-building corals depend on a symbiotic association with photosynthetic dinoflagellates of the genus Symbiodinium for survival. This intimate partnership evolved in the mid-Triassic period (Muscatine et al., 2005) and coral reefs have prospered in tropical oceans, particularly in areas characterised by high degrees of environmental stability (Hoegh-Guldberg, 1999). Consequently, corals have adapted to live within narrow physiological limits, and are highly sensitive to fluctuations in the surrounding environment (Jones et al., 1998). Rising sea surface temperatures (Hoegh-Guldberg et al., 2007) and the increasing acidity of the ocean (Orr et al., 2005) are threatening the stability of coral–dinoflagellate symbioses, leading to dire projections for the future of coral reefs (Pandolfi et al., 2011; Silverman et al., 2009). At present, our ability to accurately predict the response of corals to global climate change is severely hampered by our limited understanding of the cellular mechanisms that underpin coral–dinoflagellate symbiosis (Fabry et al., 2008; Weis et al., 2008; Davy et al., 2012), which ultimately frame how corals respond to environmental stress.

Intracellular pH (pH_i) is crucial for virtually all elements of cellular homeostasis (Smith and Raven, 1979), directly influencing protein structure, enzymatic rates and membrane solubility (Madshus, 1988). The maintenance of pH within an optimal functional range therefore plays a critical role in determining the metabolic activity of the cell, and is specific to the metabolic pathway in question. Thus, eukaryotic cells have evolved compartmentalised organelles to provide sites with specific pH conditions within the cell for different metabolic activities to occur (Casey et al., 2010). Disruption of pH_i has serious physiological consequences (Pörtner et al., 2004; Fabry et al., 2008; Hofmann et al., 2013). Indeed, a drop as small as 0.1–0.2 pH_i units can induce metabolic depression (Reipschläger and Pörtner, 1996), so it is not surprising that changes in pH_i are stringently avoided (Casey et al., 2010). In general, eukaryotic cells are protected from fluctuations in their pH_i at two levels. Acute, localised changes in pH_i, such as those arising from metabolic reactions, are neutralised by manipulating the various weak acids and bases in the cytosol (Casey et al., 2010). Longer-term changes are buffered by more permanent mechanisms such as transmembrane exchangers (Boron, 2004). The mechanisms involved in pH regulation are not well understood in corals, and this is partly due to the intrinsic complexity associated with their endosymbiosis. By virtue of their intracellular location, the Symbiodinium cells can essentially be regarded as heavily fortified organelles belonging to the coral host cell. However, unlike other organelles, Symbiodinium cells, through their photosynthetic activity, are able to exert significant control over the pH of the host cell (Venn et al., 2009; Laurent et al., 2013). Consequently, the response of corals to a change in ambient CO₂/pH is likely to be influenced by their own physiological capacity and that of their symbionts (McCulloch et al., 2012).

Here, we investigated how the presence/absence of symbionts affects the response of pH_i to CO₂-driven seawater acidification in cells isolated from the Hawaiian reef coral Pocillopora damicornis (Linnaeus 1758). Using the fluorescent dye BCECF-AM, in conjunction with live cell imaging, we characterised the response of pH_i (NBS scale) in Symbiodinium cells freshly isolated from coral hosts, isolated non-symbiotic coral cells, and isolated coral host cells with their symbionts enclosed. The cells were exposed to control seawater (pH 7.8) and CO₂-acidified seawater, designed to expose the cells to a gradient of declining external pH (pH_e 7.8–6.8) over a 105 min period, to mimic diurnal changes in pH in reef water due to reef photosynthesis, respiration and calcification (Hofmann et al., 2011; Price et al., 2012), which are particularly strong in Kaneohe Bay, Hawaii (Putnam, 2012). In both treatments, the cells were exposed to saturating white light in the presence and absence of the photosynthetic inhibitor DCMU, with measurements of both pH_i and pH_e taken every 15 min. Our findings demonstrate that CO₂ addition initiates a very different response in the pH_i of the symbiont compared with that of the coral host cell. Crucially, we show that the photosynthetic activity of the symbiont plays a key role in determining the intracellular buffering capacity of its coral host cell to changes in pH_e.

We analysed the relative change in pH_i (calculated as the pH_i after acidification minus the mean control pH_i) in three symbiotic states – Symbiodinium cells freshly isolated from coral hosts, isolated non-symbiotic coral cells, and isolated coral host cells with their symbionts enclosed (Fig. 1) – across a gradient of pH_e under light, in the presence and absence of the photosynthetic inhibitor DCMU (Fig. 2). Actual changes in pH are provided in supplementary material Fig. S1. Initial analysis of the dataset confirmed that there was a significant interaction between external pH × symbiotic state × DCMU treatment [repeated measures (rm)ANOVA, F_15.03,160.34=4.60, P<0.001], so subsequent analyses were carried out on the two separate treatments (no DCMU and DCMU added). The response of pH_i to pH_e was dependent on the symbiotic state of the cell in both the presence (rmANOVA, F_14.23,75.87=7.51, P<0.001) and absence of DCMU (rmANOVA, F_11.15,59.4=14.80, P<0.001). Subsequent post hoc analyses revealed where the differences in the response of pH_i to acidification lay (Table 1).

The disparities in pH_i were driven primarily by the opposing reactions of the host coral cells and the algal cells to acidification. Within 15 min of CO₂ addition, the pH_i of both the isolated symbiotic and non-symbiotic host coral cells decreased by 0.3– 0.4 pH units, irrespective of whether DCMU was present (Fig. 2). Following this initial drop, the response of pH_i to further acidification was dependent on the symbiotic state of the host cell, and the DCMU treatment it was exposed to. There was no change in the pH_i of the non-symbiotic host cell in either treatment (Fig. 2). Similarly, there was no change in the pH_i of the symbiotic host cell in the presence of DCMU (Fig. 2B). In contrast, without DCMU, the pH_i of the host cell increased over time when in symbiosis with its dinoflagellate partner, returning to control levels within 75 min (when the pH_e reached pH 7). The response of the pH_i of the dinoflagellate symbiont to the addition of CO₂ differed between the DCMU treatments (Fig. 2), irrespective of whether the alga was in isolation or in symbiosis (Table 1). In the presence of DCMU there was no change in the pH_i of either the isolated or symbiotic algae (Fig. 2B), whereas without DCMU both the isolated and symbiotic algae were able to increase their pH_i relative to control levels (Fig. 2A).

Understanding the coral–dinoflagellate symbiosis is pivotal to accurately predicting the susceptibility of coral reefs to climate change and ocean acidification (Weis et al., 2008; Davy et al., 2012). Recent ecological research suggests that the symbiotic state may play a critical role in determining a coral's capacity to tolerate changes in seawater chemistry (Ohki et al., 2013). It is well established that photosynthesis in the dinoflagellate symbiont enhances host coral calcification rates in the light (Allemand et al., 2004). At the cellular level, however, a more important role may be the ability of the symbiont to manipulate aspects of the host's physiology, such as cellular pH (Venn et al., 2009; Laurent et al., 2013). As yet, no study has investigated how this relationship will be impacted by acidification. We addressed this knowledge gap by simultaneously quantifying pH_i in both partners of the symbiosis in response to CO₂ addition. Our experimental design exposed isolated cells to a wide range of pH_e, at levels of CO₂ much greater than those predicted to arise from ocean acidification. These treatments were not designed to replicate climate change scenarios but rather to provide a means of determining how the symbiotic state influences the recovery of its coral host cell under induced cellular acidosis. Our results reveal that the photosynthetic activity of the symbiont increases the ability of the host cell to recover from cellular acidosis after exposure to high CO₂/low pH. The responses seen in the various symbiotic states are summarised in Fig. 3 and are discussed in more detail here.

CO₂ and H₂O exist in equilibrium with carbonic acid (H₂CO₃), so when CO₂ is removed for photosynthesis it causes an increase in the conversion of bicarbonate (HCO₃⁻) to H₂CO₃, a process that consumes protons (H⁺) (Allemand et al., 1998). Conversely, an increase in CO₂ will reverse the process, resulting in the production of H⁺ and leading to cellular acidosis. The pH_i of freshly isolated Symbiodinium cells is therefore strongly influenced by the availability of CO₂ in the surrounding seawater (Nimer et al., 1999) (Fig. 3A). Our results clearly show that, upon CO₂ addition, Symbiodinium cells are able to increase their pH_i relative to the control, by up to 0.5 pH units. This alkalinisation demonstrates the capacity of Symbiodinium cells to strongly buffer the external environmental pH signal, probably due to a fertilising effect on photosynthesis in these normally CO₂-limited algae (Nimer et al., 1999). An increase in photosynthetic productivity after CO₂ addition has also been observed in other symbiotic associations, most notably in the temperate sea anemones Anemonia viridis (Suggett et al., 2012) and Anthopleura elegantissima (Towanda and Thuesen, 2012), and the benthic foraminiferan Marginopora vertebralis (Uthicke and Fabricius, 2012). The application of DCMU (a photosynthetic inhibitor) reversed the increase in pH_i, confirming that the change was a direct consequence of photosynthesis, as the inhibited photosynthetic machinery of the symbionts is not able to ameliorate the increasing H⁺ concentration (Fig. 3A).

CO₂ supplementation initiated a markedly different response in the host coral cells compared with the symbiont. In the host cells, acidosis of the cytoplasm was observed within 15 min of CO₂ addition, with a decline in pH_e of 0.2 pH units causing the pH_i to fall 0.3–0.4 pH units below the usual pH of the cell, irrespective of the presence or absence of DCMU and the symbiotic state of the cell (Fig. 3B). This strongly suggests that the host's intrinsic buffering capacity is, initially at least, overwhelmed by the accumulation of protons resulting from the passive diffusion of CO₂, and active transport of HCO₃⁻ into the cell (Furla et al., 2000), which drives the aforementioned equilibrium reaction. The decline in pH_i, however, was halted after 15 min in both the symbiotic and non-symbiotic host cells. It is likely that this represents a time lag between the onset of acidosis and the activation of the regulatory membrane transporters. Indeed, such activity could explain the subsequent stability (~pH 6.6) of pH_i in non-symbiotic host cells, which was achieved irrespective of the surrounding seawater being subject to further acidification. Nevertheless, the pH_i of these non-symbiotic host cells never recovered to pre-acidosis levels. In contrast, the pH_i of symbiotic host cells showed a full recovery to control pH levels within 105 min of CO₂-addition (Fig. 3C). Again, this recovery was negated in the presence of DCMU, confirming that the photosynthetic removal of CO₂ by the symbiont (and hence the consumption of protons) was responsible for the increase in the host's pH_i.

These results confirm that the symbiont is able to exert a significant level of control over its host's cellular pH, corroborating the findings of previous research on reef corals (Laurent et al., 2013). Furthermore, they demonstrate that the photosynthetic activity of the symbionts plays a key role, at least in the short term, in regulating the cellular response of their host to external CO₂-driven acidification. Perhaps more importantly, however, the inability of non-symbiotic or photosynthetically compromised symbiotic host cells to recover from cellular acidosis in the short term (at least under the experimental conditions used here) suggests that they may be more susceptible than are host cells that contain fully functional Symbiodinium cells. However, whether the cells have the potential for recovery in the longer term, especially under constant pH, warrants further investigation. While some caution needs to be exercised when interpreting the results for the non-symbiotic coral cells, as we cannot be sure of their precise origin (i.e. ectodermal or endodermal), it is notable that the symbiotic endodermal host cells responded in the same manner when photosynthesis was inhibited. This limited capacity to withstand acidosis in the absence of a functional symbiont could therefore be a general response. If this is the case, it raises the possibility that the number of Symbiodinium cells that a coral host cell contains may also influence the cellular response to acidification. In Stylophora pistillata, for example, the majority of host cells (~60%) typically contain one algal cell, while fewer contain two (~35%) or more (<5%) (Houlbrèque et al., 2004). Given that we focused on host cells that harboured two algal symbionts only, there is a need for future studies that clarify the influence of symbiont number on pH_i, and the overall response at the organismal level.

The regulation of pH_i in corals is an important area for future research as cellular acidosis has serious physiological repercussions for the fitness of the individual. At a biochemical level, acidosis disrupts ion transport, nutrient trafficking and carbon acquisition (Fabry et al., 2008), causing metabolic suppression (Pörtner et al., 2004) and inevitably leading to shortfalls in the energy available for other cellular processes (Reipschläger and Pörtner, 1996). In addition to these fundamental cellular attributes, regulating pH_i is particularly important in reef-building corals for calcification. In the recently proposed ‘proton flux hypothesis’, Jokiel presented the first organism-scale model of acid–base balance in corals (Jokiel, 2011). The hydroxide ions (OH⁻) produced as an indirect by-product of CO₂ removal (Furla et al., 2000) are proposed to play a critical role in calcification, neutralising protons released by H⁺/Ca²⁺ ion exchangers in the gastrovascular cavity, and thus facilitating a pH gradient high enough for the precipitation of CaCO₃ (Jokiel, 2011; Comeau et al., 2013). Acidosis of the host coral cells could therefore place further chemical and energetic constraints on corals by affecting their ability to calcify (Jokiel, 2011; Venn et al., 2013). It remains to be seen whether a non-symbiotic host cell can reverse the changes in pH_i over a longer time frame, though this might depend, in part, on its ability to upregulate the expression of membrane transporters (Kaniewska et al., 2012), as this will afford greater control over pH_i. Indeed, several organisms are able to reverse initial decreases in pH_i (Michaelidis et al., 2005; Stumpp et al., 2012).

The ability of corals to respond to external pH change is likely to depend on the response of both the symbionts and coral host (McCulloch et al., 2012). There is considerable diversity within the genus Symbiodinium (Pochon and Gates, 2010), which translates into substantial genotypic differences in physiology (Tchernov et al., 2004; Hennige et al., 2009; Brading et al., 2011; Gibbin and Davy, 2013). Our experiment used genetically identical fragments from a single colony. This had two advantages: firstly, it minimised the baseline variation in host cellular activity; and secondly, it reduced the chances of sampling corals that contained very different Symbiodinium clades. However, future research should aim to determine whether the host and symbiont responses, and their relative abilities to buffer the effects of CO₂-driven acidification, vary according to host and symbiont genotype. In addition, it will be important to determine whether the patterns we observed in isolated coral cells are replicated in intact coral tissues, and whether longer-term incubations at more moderate levels of acidification (i.e. predicted climate change scenarios) induce similar responses to those reported here. Nevertheless, our model system elucidates the interrelationship between symbiont photosynthesis and the capacity of host coral cells to withstand CO₂-driven acidification. Moreover, our findings highlight the possibility that bleached corals may be more sensitive to cellular acidosis than are their non-bleached counterparts; this is an especially interesting topic for future research.

One large normally pigmented adult P. damicornis colony was collected 3 weeks prior to experimentation (in February 2013) from a shallow fringing reef (<3 m) in Kaneohe Bay, Hawaii. This single colony was cut into 50 genetically identical fragments (4×2 cm) that were secured to 3×3 cm plastic tiles with underwater expoxy (Z-spar, Splash Zone compound) and placed in a 50 l holding tank supplied with flowing seawater from Kaneohe Bay. Seawater chemistry was monitored frequently according to the recommended best practices for ocean acidification research and reporting (Riebesell et al., 2010), with daily measurements of salinity (psu) as well as pH (NBS scale), taken via the m-Cresol dye method stipulated in SOP 6B (Dickson et al., 2007); total alkalinity (TA) was measured on a weekly basis (Dickson et al., 2007). These characteristics were stable for the duration of the experiment, with an average salinity of 35.5±0.1 ppt, pH of 7.8±0.1 and a TA of 2166±25 (means ± s.e.m., N=5). It is important to note that the P_CO2 of seawater in Kaneohe Bay is markedly higher than average oceanic conditions (Drupp et al., 2011), resulting in a lower ambient pH. An ambient seasonal temperature of 22.6±0.3°C was maintained by a dual-stage temperature controller (Aqualogic, TR115DN), while tanks were illuminated on a 12 h:12 h light/dark cycle by metal halide lights (Ice CapMetal Halide lights, 250 W DE 14K bulbs, 250 W double-ended pendants), which were mounted on motorised light rails and provided irradiances ranging between 3.85 and 328.85 μmol photons m⁻² s⁻¹, corresponding to a mean irradiance of 125±10 μmol photons m⁻² s⁻¹ over each coral fragment.

Four experimental treatments were designed to investigate how CO₂ addition influenced pH_i (NBS scale) in cells isolated from the coral P. damicornis. Three symbiotic states were tested: (1) isolated Symbiodinium cells; (2) isolated non-symbiotic host coral cells; and (3) isolated host coral cells containing their symbiotic algae (Fig. 1). Visual inspection (by both light and confocal microscopy) failed to establish conclusively whether isolated Symbiodinium cells were surrounded by an intact symbiosome membrane. Non-symbiotic host coral cells were classified as intact host cells not containing an algal symbiont. Typically 10 μm in diameter and spherical, it is not known whether these cells were ectodermal or endodermal in origin (see Discussion). Finally, only symbiotic cells containing two algal cells were used for calculating the pH_i change in symbiotic host coral cells. These were deemed the most suitable choice for two major reasons: (1) the host cell region of interest (ROI) is much larger in doublet cells than in host cells containing a single alga, making pH_i measurements much easier (Venn et al., 2009); and (2) this cell type is found in much greater abundance than triplet cells (Houlbrèque et al., 2004). Furthermore, standardisation of symbiont number allowed for the possibility that this parameter influences host pH_i (see Discussion).

The cell and dye loading procedure was repeated five times for each treatment (N=5) to achieve independent replicate cell preparations. Individual cells were then imaged every 15 min for 105 min, with pH_i calculated from the images taken. At each experimental time point, 1.5 ml of seawater was carefully removed by pipette so as not to disrupt the cells and analysed for pH_e (NBS scale, N=5) via the m-Cresol dye method described in SOP 6B (Dickson et al., 2007). The treatments were carried out under an external white light source, provided by a variable-irradiance fibre optic cable (Halogen Reflector lamp, 150 W GX5.3 21V 1CT bulb, Philips, Somerset, NJ, USA) that produced a saturating irradiance of ~400 μmol photons m⁻² s⁻¹ photosynthetically active radiation. Previous studies have shown that photosynthetic activity in the symbiont modifies the pH_i of the host under normal CO₂ conditions (Venn et al., 2009), so a preliminary measurement was performed with no additional CO₂ added, in order to establish the control or baseline change in pH_i caused by photosynthesis. This experimental run was then repeated with CO₂ addition (5.0% setting on the incubation unit of the LSM 710 confocal microscope, Carl Zeiss, Oberkochen, Germany), resulting in a decreasing gradient of pH_e over time. The negative controls were run in the light, as before, under zero and high CO₂ conditions, but in the presence of 100 μmol l⁻¹ 3-(3-4-dichlorophenyl)-1,1-dimethylurea (DCMU) in 0.1% acetone, a photosynthetic inhibitor that blocks the plastiquinone binding site of photosystem II and thus prevents the transfer of electrons and formation of ATP. DCMU has been used extensively to study photosynthesis in Symbiodinium, as it effectively blocks photosynthesis without impairing cellular functioning (Iglesias-Prieto et al., 1992).

Coral fragments were selected randomly from the acclimation tank before the start of each experimental run. Cells were isolated immediately by gently brushing the tip of a partially submerged fragment in 50 ml of 0.22 μm filtered seawater (FSW) using a soft bristle toothbrush. The resulting slurry was centrifuged for 5 min at 1700 g. The supernatant was discarded and the pellet re-suspended in a further 50 ml FSW. An additional centrifugation step was introduced to wash the pellet and remove any residual host-generated mucus. This time, when the supernatant was discarded, the pellet was resuspended in 1 ml FSW containing 10 μmol l⁻¹ BCECF-AM ester and 0.01% Pluronic F-127 with/without 100 μmol l⁻¹ DCMU and 0.1% acetone depending on the treatment in question, to give a final concentration of ~1×10⁶ cells ml⁻¹. The dye-loaded cell suspension was then transferred to a 35 mm poly-d-lysine-coated Petri dish (MatTek Corporation, Ashland, MA, USA) and placed on the stage of a confocal microscope (LSM 710), where the cells were left to settle for 30 min in the dark at 22°C. After dye loading, the cells were carefully washed twice in 1 ml FSW to remove any residual dye without dislodging cells from the surface of the dish. Finally, 6 ml FSW was added and the dish was placed in a closed-exchange live-cell chamber (PeCon, Erbach, Germany) before experimental treatments were carried out. All dyes used for microscopy were purchased from Invitrogen (Grand Island, NY, USA).

The fluorescent dye BCECF-AM, in conjunction with confocal microscopy, has been widely used to study pH_i in marine algae (Hervé et al., 2012) as its dual-excitation spectral properties allow fluorescence measurements to be taken that are not compounded by chlorophyll autofluorescence (>640 nm). In this study, confocal microscopy was conducted on a LSM 710 confocal microscope equipped with UV and visible laser lines. Cells loaded with BCECF-AM were sequentially excited, first at 458 nm then at 488 nm, both with laser strength set at 10% and pinhole set at 1.51 units using an X63-fold oil-immersion lens. Under both excitations, fluorescence emission was captured at 525±10 nm by imaging the z-stack profile of each cell 10 times in the x/y plane. BCECF-AM is able to enter both the host and the symbiont, and can therefore be used for imaging of both compartments of an intact symbiosis. However, the signal is much stronger in the host cell than in the symbiont, probably due to the thick cell wall that surrounds the symbiont. Therefore, prior optimisation of laser settings was essential to maximise the signal strength in the algae, without overexposing the host cell. In vivo calibration was carried out on each partner in the intact symbiosis to yield two separate calibration curves, one for the symbiont and one for the host (supplementary material Fig. S2). The calibration series is dependent on the calculation of the fluorescence intensity ratio (R; F₄₈₈/F₄₅₈) after cells are suspended in buffers of a known pH (pH 6–8.5) in the presence of 5 μmol l⁻¹ nigericin (Venn et al., 2009). This R value can then be linked to pH_i by the logarithmic equation pH=pK_a+log{[(R−R_A)/(R_B−R)]×(F_A,458/F_B,458)}, where pK_a represents the acid dissociation constant and A and B represent the acidic and basic end points of the titration. To check the level of background fluorescence in the sample, cells were loaded separately with 10 μmol l⁻¹ BCECF-free acid (in the presence of 0.1% DMSO and 0.01% Pluronic F-127), a membrane-impermeant form of the dye. Background fluorescence was negligible, with no accumulation in either the symbiont or host cells.

pH_e was manipulated via in vitro addition of 99% CO₂ in a fully adjustable CO₂/temperature-controlled live-cell chamber attached to an Axiovert 200 microscope. The amount of CO₂ that was added was controlled using Zen 2011 software (Carl Zeiss), with the pre-defined volume mixed in a CO₂ module and directly injected into the chamber. The final pH_e and the time taken for the pH_e to stabilise were dependent on the amount of CO₂ added (supplementary material Fig. S3). We selected the CO₂ injection setting that produced a gradient of pH_e that spanned one pH unit and stabilised below pH 7. This also provided an optimal time frame for monitoring pH, as experiments lasting longer than 2 h are often impacted by photo-bleaching or dye-leakage from the cells (Musgrove et al., 1986). The temperature of the microscope stage was maintained at a constant 22°C by a custom-made Plexiglas incubator encasing the entire system.

Data were analysed as the relative change in pH_i (the pH_i value after acidification minus the mean pH_i of the control at each time point). Actual pH values are provided in supplementary material Fig. S1. The effect of the symbiotic state (between-subject factor), external pH and external pH × symbiotic state (both within-subject factors) were analysed using rmANOVA. Initial analysis of the dataset confirmed that there was a significant interaction between external pH × symbiotic state × DCMU treatment (rmANOVA, F_15.03,160.34=4.60, P<0.001), so subsequent analyses were carried out on the two separate treatments (no DCMU and DCMU added). Likewise, post hoc analysis was carried out at the treatment level using Bonferroni-corrected paired t-test comparisons (α=0.001). The assumptions of normality were confirmed using the Kolgomorov–Smirnov test. The sphericity of the data was tested using Mauchly's sphericity test. Epsilon-adjusted univariate F-test (Greenhouse–Geisser) values are reported. All data were analysed using JMP 10.0.0 (SAS Institute Inc., USA).

We thank Shaun Wilkinson and Nyssa Silbiger for statistical advice.