Stress and death of cnidarian host cells play a role in cnidarian bleaching

Many cnidarians, including corals, form a mutualistic symbiosis with the photosynthetic dinoflagellate Symbiodinium spp., where the algae are housed in host gastrodermal cells, thereby giving corals their typical brown coloration. Host corals provide the resident symbionts with inorganic nutrients, a high-light environment and refuge from herbivory. In return, the symbionts supply the host coral with photosynthetically fixed carbon (Davy et al., 2012; Muscatine and Weis, 1992). Corals are the trophic and structural centerpiece of coral reefs, one of the most diverse biomes on the planet. Worldwide, the survival of coral reefs is threatened by a number of factors including elevated seawater temperature, ocean acidification, rising sea levels, overfishing and pollution (Richmond and Wolanski, 2011). One biological consequence of some of these stressors, especially elevated temperature, is the phenomenon of coral bleaching. Bleaching is the breakdown of the cnidarian–dinoflagellate symbiosis leading to the loss of Symbiodinium cells from host tissues, and thus loss of color. Coral bleaching results in reduced host fitness, as evidenced by reduced growth rates and fecundity, and increased susceptibility to disease (Coles and Brown, 2003; Rosenberg et al., 2007). The ecological and socio-economic implications of bleaching are therefore immense (Hoegh-Guldberg et al., 2007).

The process of coral bleaching is complex and involves several inter-related mechanisms that vary depending on the type, duration and intensity of environmental stress. While it is well known that environmental disturbances, chiefly elevated temperature, initiate the onset of bleaching, the cellular cascade of events that leads to symbiont loss is less clear (Lesser, 2011; Weis, 2008). There is a partial picture emerging of the multiple cellular pathways involved, including host cell death either in a sudden and uncontrolled manner, or via a controlled process of programmed cell death. In addition, Symbiodinium cells can be expelled from host cells or be degraded in situ (Brown et al., 1995; Dunn et al., 2002a; Dunn et al., 2007; Dunn et al., 2004). Teasing apart these mechanisms is difficult because of the potential dual effects of stress on the host cells and their resident symbionts individually, in addition to the interaction between the two partners.

To date, most studies on bleaching have contributed to a storyline where the cascade of events leading to bleaching is initiated by the symbiont (reviewed in Lesser, 2011; Weis, 2008). Many studies have shown that heat stress damages the symbiont's photosynthetic apparatus, which results in production of excess reactive oxygen species (ROS) that in turn overwhelm existing oxygen-handling mechanisms (Franklin and Berges, 2004; Lesser, 1997). ROS diffuse into the host cell, thereby activating its apoptotic pathway, a highly ordered program of self-destruction mediated by the caspase family of proteases; this leads to host cell death (Lesser, 2006; Suggett et al., 2008). The fate of Symbiodinium cells is less clear. There is evidence that released Symbiodinium cells can be in a healthy or damaged state depending on the severity of the heat stress and the strength of the Symbiodinium cell's antioxidant defenses, but that apparently healthy cells begin to die after a few days (Hill and Ralph, 2007). Furthermore, if the released Symbiodinium cells are dead, it is unknown whether their death is due to the stress itself, to host cell apoptosis or to some other cellular pathway, such as in situ degradation via host autophagic digestion.

Very few studies have examined the effect of stress on the host cell alone and whether host-specific effects could be contributing to the bleaching process. For example, different coral species containing the same Symbiodinium clade have different responses to heat stress (Fitt et al., 2009; LaJeunesse, 2002). This could be, in part, a result of differences in host protective mechanisms between species. A variety of these protective mechanisms have been identified, such as the use of fluorescent proteins and mycosporine-like amino acids to manage high levels of ultraviolet and visible light, increased heat shock protein expression to mitigate a cellular stress response, and elevated production of antioxidants to manage high concentrations of ROS (reviewed in Baird et al., 2009). Differences in the susceptibility of corals to bleaching could also come from actions of the host as a result of stress directly impacting host cells. In studies on the model anemone Aiptasia pulchella, Nii and Muscatine (Nii and Muscatine, 1997) found that aposymbiotic anemones produced excess ROS under high heat and light conditions in the absence of symbionts. In other organisms, this is indicative of stress-induced mitochondrial dysfunction (Haak et al., 2009). Recently, Dunn et al. (Dunn et al., 2012) have shown that heat stress can induce mitochondrial damage in a host cell with healthy Symbiodinium cells. Together, this information raises the possibility that host stress and cell death alone could lead to stress and/or death of resident Symbiodinium cells.

In this study, we examined the role of host cell stress and death in the bleaching response. We performed our experiments with the symbiotic anemone Aiptasia sp., a useful laboratory model for the study of corals (Weis et al., 2008). First we found that colchicine, a pharmacological agent that is a known elicitor of apoptosis in metazoans, induces cell death in host cells, but not in cultured Symbiodinium CCMP-830 cells or those freshly isolated from hosts (Gorman et al., 1999). Treatment of symbiotic Aiptasia with colchicine resulted in death of Symbiodinium in hospite; an indication that host cell apoptosis can contribute to Symbiodinium cell death. A similar increase in death of Symbiodinium in hospite was observed when anemones were subjected to heat stress. Live-cell confocal microscopy of colchicine-treated symbiotic host cells revealed a pattern of host cell death followed by death of the resident Symbiodinium cells in the majority of cases. This same pattern was repeated when the symbiotic cell isolates where subjected to heat stress. These studies do not negate the crucial role of symbiont stress in bleaching, but they do suggest that cell death initiated in host cells can contribute to cnidarian bleaching.

Symbiodinium cells (CCMP-830, sub-clade B1) were maintained in f/2 medium at 25±1°C on a 12 h:12 h light:dark photoperiod at an irradiance of 70 μmol quanta m⁻² s⁻¹, a light level at which there is no stress on the symbionts. For all experiments on cultured Symbiodinium, cells were washed and resuspended to a final concentration of 1×10⁶ cells ml⁻¹. Individual anemones (Aiptasia sp.) containing clade B Symbiodinium (data not shown) were maintained in 0.1 μm ceramic filtered artificial seawater (FASW) under temperature and light conditions identical to those described above. Anemones were fed twice a week with freshly hatched brine shrimp nauplii. Anemones were transferred to 6- or 24-well culture plates and starved for 2 to 3 days prior to an experiment.

The pharmacological agent colchicine was used to induce host cell death. Colchicine is known to induce apoptosis in animal cells including cnidarians (David et al., 2005; Dunn et al., 2004; Dunn and Weis, 2009; Pernice et al., 2011). Symbiotic individuals of Aiptasia were incubated in FASW alone or in 12.5 mmol l⁻¹ colchicine dissolved in FASW for 3 h prior to addition of 50 μmol l⁻¹ fluoroprobe Rhodamine 110 aspartic acid, a caspase substrate (Life Technologies, Grand Island, NY, USA); this concentration of colchicine has previously been shown to cause apoptosis in cnidarians (Dunn and Weis, 2009). Upon enzymatic cleavage via caspases, the non-fluorescent substrate is converted to the fluorescent Rhodamine 110. This method is described in detail in Detournay and Weis (Detournay and Weis, 2011). Images of anemone tissue were acquired with a Zeiss LSM 510 Meta microscope with a 40/0.8 water objective lens (Zeiss, Thornwood, NY, USA). Fluorescence excitation/emission for Rhodamine 110 was 488/505–530 nm. Relative caspase activity was quantified using ImageJ software (Schneider et al., 2012).

Dinoflagellate cell survival has been found to be unaffected by incubation in colchicine (Cho et al., 2011; Wong and Kwok, 2005). We wanted to test this both in Symbiodinium CCMP-830 in culture and in symbiont cells freshly isolated from hosts. Cell death was measured using Sytox Green, a fluorescent probe that can only cross plasma membranes of dying cells, where it binds to nucleic acids. Sytox Green therefore labels an early stage of cell death prior to cell lysis.

Cultured Symbiodinium CCMP-830 cells were co-incubated in 5 μmol l⁻¹ Sytox Green and 0, 6, 12.5 or 25 mmol l⁻¹ colchicine. There was an additional treatment in Sytox Green and 10 mmol l⁻¹ H₂O₂, an ROS and therefore a positive control for oxidative stress. A high concentration of H₂O₂ was used to ensure that cell death occurred rather than to mimic physiological conditions. Sytox Green fluorescence was measured every 30 min for 10 h. Total relative Sytox Green fluorescence was measured from triplicate experiments using a SPECTRAmax GEMINI Spectrofluorometer (Molecular Devices, Sunnyvale, CA, USA) set for high resolution and sensitivity, and an excitation/emission of 448/523 nm with a 515 nm cut-off.

To prepare freshly isolated Symbiodinium cells, two medium-sized Aiptasia (approximate oral disc diameter 4 mm) were homogenized for 30 s in 3 ml FASW using a glass tissue grinder. Homogenates were centrifuged for 1 min at 735 g, the supernatant was removed and the algal pellet was resuspended in 1 ml FASW. This process was repeated three times to remove cellular debris. Between each centrifugation, the suspension was gently syringed 10 times with a 23-gauge needle to isolate the Symbiodinium cells from host cells. Symbiodinium cells were allowed to recover for 3 h at room temperature before experimental treatment and imaging.

Cells were incubated in a seawater control, 10 mmol l⁻¹ H₂O₂ (positive control) or 12.5 mmol l⁻¹ colchicine for 3 or 24 h. Thirty minutes prior to imaging, Sytox Green was added to a final concentration of 5 μmol l⁻¹ to cell suspensions in a glass-bottom dish (MatTek Corporation, Ashland, MA, USA). Images were captured with a Zeiss LSM 510 META confocal microscope through a Plan-APOCHROMAT 63×/1.4 oil DIC objective. Fluorescence excitation/emission was 488/505–530 nm for Sytox Green and 688/679–754 nm for Symbiodinium chlorophyll autofluorescence. Experiments were performed in triplicate and, for each sample, 10 random images were taken. Acquired images were then randomized and a naïve counter quantified the total number of Symbiodinium cells and the fraction containing Sytox Green fluorescence, allowing for the calculation of a ratio of Sytox:control that accounted for any death induced by the isolation process.

Specimens of symbiotic Aiptasia were incubated in 12.5 mmol l⁻¹ colchicine for 3 or 24 h. Symbiodinium cells were isolated and cell mortality was quantified using confocal microscopy as described above.

Symbiodinium CCMP-830 cell suspensions were incubated in a range of heat-stress conditions: 25 (ambient), 27, 30 or 33°C, for 3 or 24 h in glass-bottom dishes. Thirty minutes prior to imaging, Sytox Green was added to a final concentration of 5 μmol l⁻¹. The mortality of Symbiodinium cells at each temperature was quantified with confocal microscopy, as described above.

Specimens of symbiotic Aiptasia were incubated at the same set of temperatures and for the same duration as were the cultured Symbiodinium cells. Symbiodinium cells were subsequently isolated and incubated with 5 μmol l⁻¹ Sytox Green, and their mortality was quantified using confocal microscopy as described above.

To isolate individual host gastrodermal cells from symbiotic Aiptasia, a method modified from Gates and Muscatine (Gates and Muscatine, 1992) was utilized. Animals were placed in relaxing solution (1:1, 370 mmol l⁻¹ MgCl₂:FASW) 30 min prior to snipping off tentacles with small scissors and placing them into Ca²⁺-free seawater. Tentacles were cut into 1 mm barrels with a scalpel and then gently pelleted by centrifugation for 1 min at 735 g. Under aseptic conditions, the tentacle bits were resuspended in 0.05% collagenase (Sigma-Aldrich, St Louis, MO, USA) in 0.1 mol l⁻¹ Na₃PO₄ buffer containing 3% NaCl and 0.004% CaCl (pH 7.4), and allowed to dissociate during a 1 h incubation at room temperature on a shaker table at 150 r.p.m. The cellular suspension was passed once through an 18-gauge needle and then centrifuged for 1 min at 735 g. Cell macerates were resuspended in culture medium and plated onto poly-l-lysine-coated (Sigma-Aldrich) glass dishes. Several different culture media were tested, and a medium composed of 25% Dulbecco's modified Eagle medium (GIBCO, Life Technologies) in FASW resulted in the lowest level of mortality over time (data not shown). In all conditions, it was found that the rate of death decreased after a 3 h recovery period and thus cells were allowed to acclimatize for 3 h at room temperature before further experimental manipulation.

After recovery, host cell macerates were placed in FASW at 25°C, 12.5 mmol l⁻¹ colchicine in FASW at 25°C or FASW at 33°C, and followed for 8 h. For heat treatments, a CTI-controller 3700 humidified chamber (Zeiss) was utilized. Confocal imaging began 1–2 h after initiation of experimental treatments. For each of three replicates, images were acquired every 2 h from 16 different locations in the glass-bottomed dish. Each location contained a minimum of two symbiotic host cells housing multiple Symbiodinium cells. Confocal images were captured as described above. DIC-transmitted light images were captured simultaneously with fluorescence images, using the transmitted light detector. After acquisition, images were randomized and a naïve counter scored host cells and resident symbionts as either containing or missing Sytox Green labeling. Subsequent tabulation of all images for each host cell over the duration of the experiment resulted in four patterns of Sytox Green labeling, as shown in Fig. 1: (0) no label; (1) label in host cell only; (2) label in both partners; and (3) label in host cell first followed by appearance of label in symbiont. Because Sytox Green cannot penetrate the membrane of live host cells, we were unable to detect the presence of live host cells containing dead Symbiodinium cells. Therefore, healthy host cells containing both live and/or dying Symbiodinium cells showed no label. In addition, when label was detected in both host and Symbiodinium cells, we were unable to determine whether this occurred simultaneously or whether mortality of Symbiodinium cells occurred prior to host cell death.

All data were tested for normality and heteroscedasticity before being analyzed by two-way ANOVA, which applied treatment (chemical or elevated temperature versus control) and time as effects. T-tests with a Bonferroni correction were employed when multiple comparisons were made (GraphPad, La Jolla, CA, USA). Data are presented as means ± s.e.m. All observations are derived from at least three independent experiments. Significance was set at P<0.05.

To determine the fate of the Symbiodinium cells in hospite after host cell death, the pharmacological agent colchicine was used to induce host cell but not Symbiodinium cell mortality. Incubation of symbiotic Aiptasia in 12.5 mmol l⁻¹ colchicine for 3 h elicited apoptotic host cell death, as evidenced by the presence of Rhodamine 110 fluorescence, a measure of caspase activity, in whole-mount Aiptasia (Fig. 2). Levels of fluorescence increased in both the epidermis and gastrodermis from relative fluorescence of 957 and 307 in control animals to 4012 and 1814 in colchicine-incubated animals, an increase of fourfold (P<0.0001) and sixfold (P<0.01), respectively (Fig. 2B).

Other investigators have found that colchicine does not affect dinoflagellate cell survival (Cho et al., 2011; Wong and Kwok, 2005). We wanted to test this both in Symbiodinium CCMP-830 cells in culture and in cells freshly isolated from their host. Control CCMP-830 cells displayed no increase in fluorescence throughout the duration of the experiment (Fig. 3A). In contrast, cells incubated in H₂O₂ displayed a steady increase in fluorescence throughout the experiment to a final treatment:control ratio of 5 (Fig. 3A). However, there was no significant difference in Sytox Green fluorescence in either the 6 or 12.5 mmol l⁻¹ colchicine treatments compared with controls (P>0.05). Between 1 and 2.5 h, cells incubated in 25 mmol l⁻¹ colchicine exhibited a slight increase in fluorescence to ~1.6 times the control value (P<0.0001) and subsequently remained at this level for the duration of the experiment.

Because Symbiodinium cells in hospite display a different phenotype than those in culture, and could potentially be differentially susceptible to colchicine, we tested the effect of colchicine on Symbiodinium cells freshly isolated from Aiptasia and incubated in a FASW control, 10 mmol l⁻¹ H₂O₂ (positive control) or 12.5 mmol l⁻¹ colchicine for 3 or 24 h (Fig. 3B). Whereas Symbiodinium cells incubated in H₂O₂ exhibited six (P<0.0001) and 12 times (P<0.0001) greater mortality than controls at 3 and 24 h, respectively, the level of mortality in Symbiodinium cells incubated in colchicine was not significantly different from that of control incubations, regardless of incubation time (P>0.05). Together, these results suggest that colchicine does not induce Symbiodinium cell mortality under experimental conditions.

Next, we were interested in determining whether induction of cell death in hosts could lead to cell mortality of Symbiodinium cells in hospite. Treatment of Aiptasia for 3 h in colchicine led to a 2.3-fold increase (P<0.05) in Symbiodinium cell mortality compared with controls, and further increased to 4.7-fold (P<0.001) after 24 h of colchicine treatment (Fig. 4). Because colchicine does not induce Symbiodinium cell mortality, as demonstrated in Fig. 3, these data suggest that apoptosis of host gastrodermal cells may play a role in mortality of Symbiodinium cells in hospite.

With an understanding that host cell death could be contributing to symbiont cell mortality, the next step was to test the hypothesis that heat stress causes host cell death, which in turn can contribute to symbiont cell mortality. First, to assess the mortality levels in Symbiodinium CCMP-830 cells due to elevated temperature, the cultured dinoflagellates were incubated under a range of heat stress conditions [25 (ambient), 27, 30 or 33°C for 3 or 24 h] and mortality was quantified with Sytox Green fluorescence and confocal microscopy (Fig. 5A). Cells sampled at 3 h increased from treatment:control ratios of 1 at 25°C to a high of 2.5 at 33°C, and those sampled at 24 h increased from ratios of 1 at 25°C to a high of 5.2 at 33°C. Cells incubated at 30°C (3 h P<0.01, 24 h P<0.0001) and 33°C (3 and 24 h both P<0.0001) displayed significantly increased rates of mortality compared with controls. The level of mortality in Symbiodinium cells incubated at 33°C for 24 h was 2.5-fold higher than the level in Symbiodinium cells incubated for 3 h (P<0.0001). These findings reflect an increase in mortality that is both dose and time dependent.

A similar experiment was carried out to determine whether levels of mortality due to heat stress were different in Symbiodinium cells in hospite compared with those in culture. As in treatments of cultured Symbiodinium cells, mortality of algae in hospite increased with rising temperature and prolonged duration of treatment (Fig. 5B). Cells sampled at 3 h increased from treatment:control ratios of 1 at 25°C to a high of 4.6 at 33°C, and those sampled at 24 h increased from ratios of 1 at 25°C to a high of 7.7 at 33°C. Values at 30°C (3 h, P<0.05; 24 h, P<0.0001) and 33°C (3 h, P<0.0001; 24 h, P<0.0001) were significantly higher than those at ambient temperatures, with the rate of symbiont cell mortality significantly elevated between 3 and 24 h (30°C, P<0.01; 33°C, P<0.0001).

Increases in Symbiodinium cell mortality appeared more pronounced in cells that had been in hospite than in culture: the dose- and time-dependent responses were more dramatic in symbiont cells resident in the host. The level of mortality in Symbiodinium cells in hospite incubated for 24 h at 33°C was nearly eightfold higher (P<0.0001) than controls, compared with 5.5-fold higher (P<0.0001) than controls in Symbiodinium cells in culture. These data suggest that the health of the host cell has a direct impact on the health of the endosymbiont during stress.

Finally, to further test the hypothesis that host cell stress and death can contribute to symbiont mortality, the dynamics of cell death in individual host cells with resident symbionts was temporally monitored. The health of individual symbiotic host cells obtained by maceration was monitored by live-cell imaging over an 8 h period. The progression of host cell death and Symbiodinium mortality was then classified as one of three patterns or modes of cell death based on the presence of Sytox Green fluorescence labeling (Fig. 6A): (1) death of host cell only with healthy Symbiodinium cells; (2) death of host cell and Symbiodinium cell mortality observed simultaneously; and (3) death of host cell observed first followed by Symbiodinium cell mortality. As mentioned in the Materials and methods (Fig. 1), data from pattern 2 are difficult to interpret because the technique is unable to detect dying symbiont cells in live host cells. Pattern 2 could include live host cells where Symbiodinium mortality preceded host cell death. This approach, therefore, cannot quantify a pattern of symbiont cell mortality preceding host cell death.

Under control conditions, 28% of symbiotic host cells died during the 8 h incubation (Fig. 6B). Of these cells, 92% died according to the first pattern (P<0.0001), with 2 and 6% displaying patterns 2 and 3, respectively. In contrast, incubation of symbiotic host cell macerates in 12.5 mmol l⁻¹ colchicine led to 98% host cell death (P<0.0001). Of these dead cells, the majority (67%; P<0.0001) displayed death according to pattern 3, with 38% (P<0.05) and 5% (P>0.05) exhibiting patterns 1 and 2, respectively. Taken together, these results suggest that induction of host cell apoptosis via colchicine incubation leads primarily to a pattern of host cell death followed by Symbiodinium mortality.

Finally, to determine whether a similar pattern of death is observed under heat stress conditions, symbiotic host cells were subjected to 33°C temperature stress for 8 h. All host cells subjected to heat stress died. A slight majority of macerates (52%; P<0.0001) exhibited death according to pattern 3. A further 27% (P<0.001) and 21% (P<0.001) died by patterns 1 and 2, respectively. These data suggest that, in many cells, heat stress induces apoptosis in host cells that in turn could be contributing to death of the symbionts.

Multiple studies have shown that programmed cell death in the host and possibly in the symbiont plays an important role in high-temperature-induced cnidarian bleaching (Dunn et al., 2002a; Franklin et al., 2004; Pernice et al., 2011; Strychar and Sammarco, 2009). However, details such as the order in which each process occurs, in which partner, and the effect that one partner's stress has on the other remain difficult to dissect out from the complexity of the endosymbiosis. To aid our attempts to tease apart the effects of stress on each partner individually, in addition to the interaction between the two partners, we wanted to identify a pharmacological agent that would cause programmed cell death of the host cell but not the symbiont cell within the time frame of our experiments. Building on previous studies in Aiptasia and other animals, and in dinoflagellates, we focused on colchicine. Colchicine, a commonly used anti-cancer drug, is a microtubule-disrupting agent. In mammals, it also induces apoptosis through poorly understood mechanisms that cause release of cytochrome c from mitochondria and subsequent caspase activation (Gorman et al., 1999). Researchers studying apoptosis and/or bleaching in cnidarians have shown that colchicine causes apoptosis in a variety of cnidarians including Hydra, the corals Acropora millepora and Fungia scutaria, and in Aiptasia sp. (David et al., 2005; Dunn et al., 2004; Dunn and Weis, 2009; Pernice et al., 2011). In contrast, studies to date on dinoflagellates have shown that colchicine causes cell cycle arrest in G1 but no programmed cell death (Cho et al., 2011). Our experiments confirm these previous findings (Figs 2, 3). Incubation of aposymbiotic anemones in colchicine caused apoptosis, whereas no cell death was observed in cultured or freshly isolated Symbiodinium cells incubated in colchicine.

With the tool of a host-only apoptosis-inducer in hand, we found that induction of apoptosis in hosts when incubated in colchicine also results in increased cell death of symbionts in hospite (Fig. 4). In addition, we provide evidence from colchicine-incubated host cell macerates that, in the majority of dying cells, host cell apoptosis occurs first, followed by death of the resident symbionts (Fig. 6A,B). These results suggest that host cell apoptosis can contribute to cell death of symbionts without prior signaling from stressed or dying symbionts. This response exhibits characteristics of a classic innate immune response, such as the removal and/or destruction of foreign bodies, so well characterized in other animals (Postigo and Ferrer, 2009; Wilson et al., 2009). A similar phenomenon has also been described in aposymbiotic larvae of the coral F. scutaria challenged with Symbiodinium sub-clade C31, which is unable to colonize this coral species (Dunn and Weis, 2009). Larvae challenged with C31 cells exhibited high numbers of apoptotic cells in the gastroderm, suggesting that hosts use apoptosis to eliminate inappropriate symbionts from their tissues.

Moving from simple studies using a single pharmacological agent to those that use heat stress, which causes a myriad of effects, could we show evidence that temperature stress on host cells contributes to cell death of symbionts in hospite? Results shown in Fig. 5 indicate that Symbiodinium cells subjected to elevated temperature die at a higher rate in hospite than they do in culture, suggesting that there is a negative host effect on Symbiodinium cell survival during heat stress. Furthermore, in the majority of cases, the host cell dies first, followed by mortality of the symbiont (Fig. 6B). However, the percentage of cases exhibiting simultaneous death of host and symbiont cells was much higher in heat-stress treatments compared with the colchicine treatment, suggesting that multiple effects are at play in the observed cell death patterns. Previous studies in Aiptasia sp. using quantitative ultrastructural techniques (Dunn et al., 2002; Dunn et al., 2000) have described a variety of host and symbiont cell death scenarios [cf. fig. 11 in Dunn et al. (Dunn et al., 2002)] depending on the duration and degree of the stress. Included in this were death profiles similar to what we describe in this study: death of both the host and the symbionts. Other studies have found evidence of elevated temperature stress in host tissues. ROS are produced in aposymbiotic anemones subjected to elevated temperature (Nii and Muscatine, 1997) or light (Dykens et al., 1992). Host cells from Aiptasia containing apparently healthy symbionts show mitochondrial damage (and presumably subsequent ROS generation) after heat stress (Dunn et al., 2012). Together with our results from colchicine treatment, these findings suggest that host stress plays a role in the bleaching response.

The present study has several limitations that constrain our ability to fully interpret the results. First, we were not able to observe Symbiodinium mortality in live host cells (see Fig. 1, pattern 0) and therefore we were unable to assess a potential pattern of symbionts dying in healthy host cells. However, if this pattern were observable, it would decrease the percentage of live host cells bearing healthy symbionts observed, but would not affect patterns 1–3 described in Fig. 6. Second, we were not able to assess a potential pattern of symbiont cells dying first followed by death of the host cells (see Fig. 1, pattern 2). If this pattern were observable, it would decrease the percentage of pattern 2. Finally, in the heat stress experiments, it is possible that initiation of host cell death was not caused directly by the host cell response to heat stress, but instead was a result of stress signals, such as ROS, from the Symbiodinium cell that are released prior to symbiont cell mortality; these in turn would launch a host innate immune response. Despite these limitations, the results provided in this study allow us to add to existing cellular models for cnidarian bleaching proposed in other studies and reviews (Fig. 7) (Lesser, 2006; Lesser, 2011; Weis, 2008). Specifically, with an elevated temperature stress, ROS generated by symbionts leak from the symbiont into host tissues and contribute to a host cell death response (pathway A). Based on data from this study and others (Dunn et al., 2012; Dykens et al., 1992; Nii and Muscatine, 1997), elevated temperature will stress host cells, causing host cell death (pathway B). Finally, our study suggests that host cell death can contribute to subsequent symbiont cell mortality (pathway C), a process that appears similar to a host innate immune response.

We thank Sheila Kitchen for help setting up the maceration protocol, Jessica Flesher and Kelsey Lance for their work counting and classifying cells, Riana Wernick for spell-checking and editing early drafts of the manuscript, and members of the Weis Laboratory for comments on the manuscript. We also wish to acknowledge the Confocal Microscopy Facility of the Center for Genomic Research and Biocomputing at Oregon State University.