Cell Biology of Cnidarian-Dinoflagellate Symbiosis

Recognition.It is worthwhile to briefly clarify the definitions of recognition and specificity, two terms that permeate the literature and often occur together. Recognition is molecular signaling that takes place between the host cnidarian and algae destined to be symbionts, most often during the onset of the association. Specificity is the taxonomic range of partners with which an organism associates (92). The focus here will be on recognition, i.e., relatively early events in the winnowing process that can contribute to the formation of a specific partnership. Other reviews are devoted to the topic of specificity in cnidarian-dinoflagellate associations and include coverage of studies that describe successional ecological competition for the host intracellular niche that occurs downstream of the initial recognition events (16, 17). It is important to emphasize here that the temporal window of recognition events in relation to engulfment and subsequent persistence of symbionts in host tissues, described below, is unclear. For example microbe-associated molecular pattern (MAMP)-pattern recognition receptor (PRR) interactions (Fig. 2) might be transient, occurring at the beginning of the interaction, or they might also be required for the maintenance of a stable association.

Algal entry into hosts.Perhaps surprisingly, the process of algal entry into hosts has received the least amount of attention of any stage of symbiosis establishment. Overall there is general agreement throughout studies of both the green Hydra and cnidarian-dinoflagellate symbioses that phagocytosis is the prevalent mode of entry. However, whether or not phagocytosis is selective or differential, for example, between food particles and algae or between different algal strains, is less clear. Green Hydra studies describe differential rates of uptake and differently shaped engulfing phagosomes (phagocytic profiles), visualized by scanning electron microscopy (SEM), in hosts when engulfing latex spheres versus heat-killed versus intact freshly isolated Chlorella (234) and when latex spheres were coated with anions and cations (235). However, others found no evidence that phagocytosis is specific to particle type or is anything more than a general feeding phenomenon (178, 236). These authors attributed the differential profiles to a confounding factor of contaminating host material that was present with freshly isolated algae.

In cnidarian-dinoflagellate symbioses, phagocytic profiles of host gastrodermal cells engulfing Symbiodinium spp. have been described for C. xamachana scyphistomae (56) and for larvae of F. scutaria (325). In experiments measuring rates of algal uptake, competition with inert carmine particles resulted in a decreased infection rate, suggesting that phagocytosis was nonspecific (56). In contrast, confocal imaging of infection in F. scutaria larvae revealed evidence of phagocytosis as a selective process (317). Larvae were challenged with both Symbiodinium type C1f freshly isolated from adult F. scutaria and type C31 from cooccurring Montipora capitata, which cannot successfully colonize F. scutaria larvae (401). Rates of incorporation of algae into gastrodermal cells from the gastrovascular cavity were lower for Symbiodinium C31 than for C1f. In addition, larvae demonstrated spatial selectivity, whereby C1f dinoflagellates were taken up preferentially around the equator of the larva compared to latex spheres and C31, which were taken up throughout the gastrodermis.

Both studies detailed above and all of the green Hydra work that examined algal engulfment used experimental manipulation to enhance infection: injecting algae into polyps in the case of C. xamachana and green Hydra and inducing a feeding response by the addition of food in the case of F. scutaria larvae. Very few studies have examined algal uptake under conditions that more closely approximate natural uptake from the surrounding environment. Hirose and coworkers (158) described dinoflagellate acquisition in juvenile polyps of Acropora spp. after coincubation with freshly isolated dinoflagellates. After several days of coincubation, dinoflagellates were still free in the gastrovascular cavity and associated with host cellular debris. After this, some dinoflagellates appeared to be trapped by cytoplasmic processes and/or elongate cilia extending from host gastrodermal cells before later appearing in host gastrodermal cells. In a study of F. scutaria embryonic development, Marlow and Martindale (215) found symbionts in the endoderm and ectoderm prior to the formation of a mouth. This indicates that there are modes of invasion other than phagocytosis of algae after entry into the gastrovascular cavity. Mechanisms of symbiont sequestration into germ cells or developing embryos that occurs in cnidarian species that undergo vertical symbiont transmission is also very poorly understood (24, 79). Symbiodinium acquisition by or invasion of cnidarian hosts is an obvious area that needs substantial further study.

Arrest of phagosomal maturation.The question of how invading algae manage to avoid intracellular attack in host phagocytes has been of interest for decades. Simply put, how is a phagosome, destined to destroy its occupant, converted to a symbiosome that tolerates algae and allows them to persist? The problem was first addressed with the ciliate Paramecium bursaria, which is colonized by a species of Chlorella. Both pioneering (182, 183) and recent work (reviewed in reference 187) found evidence that Chlorella inhibits phagosome-lysosome fusion. Similar examinations describing phagosome-lysosome fusion followed with green Hydra and with the jellyfish C. xamachana, which harbors S. microadriaticum. Aposymbiotic H. viridis fed healthy Chlorella showed the lysosomal markers acid phosphatase, ferritin, and thorium surrounding the phagocytosed heat-killed algae or other particles but showed no label around healthy algae (167, 280). Likewise, in C. xamachana, Fitt and Trench (110) found that both ferritin and acid phosphatase colocalized to symbiosomes with heat-killed ingested S. microadriaticum, while symbiosomes containing healthy dinoflagellates remained free of either marker.

Studies of the anemone A. pulchella have generated further evidence, by tracking the location of Rab GTPases, that phagosomal maturation and endosomal trafficking (described in Appendix 1) are altered by the presence of Symbiodinium cells in phagosomes (see Appendix 1). First, cnidarian orthologs to human Rab5 and -7 were sequenced and shown in heterologous systems to be present in the same endosomal locations as their corresponding orthologs in vertebrates: Rab5 and Rab7 in early and late endosomes, respectively (47, 48). In immunofluorescence examinations of A. pulchella gastrodermal cell macerates, anti-human Rab5 appeared around healthy newly ingested and already-established Symbiodinium but was absent from around heat-killed or DCMU-treated newly ingested Symbiodinium. Conversely, anti-human Rab7 localized around heat-killed or DCMU-treated newly ingested Symbiodinium but was absent from untreated newly infected or already-established Symbiodinium. These studies suggest that the Symbiodinium cell somehow arrests phagosomal maturation in the early phagosome stage, as evidenced by the presence of Rab5 and absence of Rab7.

Subsequent work by this group has continued along these same lines to investigate the roles of Rab11 and -4 (49, 168), two controllers of endosomal membrane recycling (323). Symbiosomes containing healthy Symbiodinium were Rab11 negative and Rab4 positive. Rab4 was found to be immediately recruited to all early phagosomes but was retained only in those containing healthy symbionts, suggesting that Rab4 is essential to the generation of the symbiosome.

Venn and coworkers have recently measured the intracellular pH of macerated host gastrodermal cells with resident symbionts from the coral Stylophora pistillata using the pH-sensitive fluorophore SNARF-1 and confocal imaging (383; see below for more details). They estimated that the pH of the symbiosome is <6.0, a level consistent with the luminal pH of late phagosomes (see Appendix 1). Taken together these investigations indicate that, like Leishmania spp., Mycobacterium tuberculosis, and others (some are listed in Table A1 in Appendix 1), Symbiodinium survives in symbiosomes in part by manipulating endosomal trafficking. Studies examining the specific effectors of this arrest are a fundamental next step.

Evidence of apoptosis and autophagy during recognition.Following engulfment, there is a complex suite of downstream cellular responses to an invading microbe (Fig. 2; see Appendix 1). To date, this is an area of symbiosis establishment that has received relatively little attention. Dunn and Weis (101) found evidence that apoptosis plays a role in postphagocytic sorting processes. As described above, larvae of F. scutaria were challenged with both Symbiodinium C1f from F. scutaria and Symbiodinium C31 from M. capitata, which cannot colonize F. scutaria. Larvae challenged with C31 showed high caspase activity in the gastrodermis, measured by quantifying a caspase-specific fluorophore with confocal microscopy, compared to those with C1f symbionts. This activity was inhibited with the addition of caspase inhibitors. These data suggest that the host mounts an innate immune response against incompatible Symbiodinium. To expand on the winnowing process, this indicates that Symbiodinium C31 cells make it past the glycan-lectin stage but not past a subsequent step which in turn sets off the apoptotic response by the host.

The phenomenon of autophagy, the cellular process of removal and degradation of organelles, cytoplasmic contents, and microbial invaders (202, 203), is a microbial control mechanism that is yet to be investigated in cnidarian-dinoflagellate symbiosis recognition. Autophagy is of interest because of its links to other membrane trafficking pathways and to apoptosis. Furthermore, there is some evidence that it plays an active role in the elimination of symbionts during the bleaching response and could therefore also function in recognition (93, 99). Characterization of this highly conserved process in cnidarians and its potential involvement in recognition could be a fruitful area for future research.

Evidence of cell signaling processes in recognition.Two of the most highly upregulated genes in symbiotic anemones encode sym32, a protein described first in Anthopleura elegantissima (311, 326, 399) and more recently in Anemonia viridis (122), and calumenin (122). Ganot and coworkers (122) have developed a testable model for the role that these two proteins play in recognition and tolerance of symbionts. sym32 is a member of the fasciclin 1 family of cell adhesion and signaling proteins. A previous study localized sym32 to symbiosomes (326) and hypothesized that it serves an interpartner signaling function. Calumenin is a Ca²⁺-binding EF hand protein that participates in critical posttranslational γ-carboxylation of proteins, including fasciclin I orthologs in mammals (68). The model proposes that the presence of symbionts signals calumenin to promote γ-carboxylation maturation of sym32 on symbiosomes. How sym32 participates in cell signaling or recognition interactions between host and symbiont is a topic for future investigation.

In summary, the processes of recognition and phagocytosis are areas of cnidarian-dinoflagellate symbiosis cell biology under active investigation. Rapidly increasing genomic resources will continue to provide crucial information about both the innate immune repertoire and the genes involved in phagocytosis in the host. Complementary genomic resources for the dinoflagellate are far behind, in large part because the genome sizes of dinoflagellates are equal to or greater than the size of the human genome. Investment in these resources is important for the elucidation of pathways that symbionts use to induce host tolerance to invasion. The continuing development of techniques such as transformation and reverse genetics (97, 417) will allow for direct description of the functional mechanisms that underlie the establishment of symbiosis.

Symbiont expulsion.The extrusion of pellets containing symbiotic dinoflagellates was reported in the early 1970s for a range of coral, sea anemone, and zoanthid species (307, 338, 339, 372). Steele (339) was the first to consider the significance of these pellets in detail. He conducted light microscopic examinations of pellets released by 11 tropical sea anemone species, noting the life history stage of the dinoflagellates and their state of health, and found that these pellets consisted largely of algal debris held together by mucus but also contained “a very small number” of intact symbiont cells. Steele concluded that the expulsion of these pellets must be a common occurrence and that they help to control the symbiont population by removing degraded and, to a lesser extent, healthy symbiont cells; the expulsion of healthy symbionts (with the potential to infect new hosts) has more recently been supported by assessments of their photosynthetic health (29, 295).

A wide range of expulsion rates has now been reported, suggesting that the importance of this regulatory mechanism varies between symbioses. Very low rates of symbiont expulsion (∼0.1 to 1.0% of the symbiont standing stock per day) have been reported for a range of tropical cnidarians, including the scleractinian corals Stylophora pistillata, Seriatopora hystrix, Pocillopora damicornis, and Acropora formosa (159, 160, 179, 271, 346), the soft corals Xenia macrospiculata and Heteroxenia fuscescens, and the fire coral Millepora dichotoma (159). These expulsion rates are not constant throughout the day, however, with no clear diel pattern across all species (159, 179, 346). The lowest rates of ∼0.1%, observed in various Red Sea corals, equated to just 4% of the rate at which cells were added to the symbiont population (159). This, together with the relative constancy of expulsion under steady environmental conditions (332) and the stimulation of expulsion only under conditions of stress (123, 159, 160, 332, 345), led Smith and Muscatine (332) to propose that expulsion is associated with host cell turnover, consistent with observations of symbiotic dinoflagellates being expelled within intact host cells (123, 179).

A more dominant role for expulsion has been suggested elsewhere. Expulsion was suggested as the primary process by which the symbiont density is maintained in the temperate sea anemone Anthopleura elegantissima, given that expulsion can exceed daily growth of the symbionts at irradiances of less than full sunlight (232), while in the temperate Atlantic coral Astrangia poculata, relatively high rates of expulsion allow the coral to sometimes persist in a stable, near-nonsymbiotic state (86). Similarly, Baghdasarian and Muscatine (15) concluded that expulsion is one of the primary regulators of the symbiont density in some (though not all) tropical cnidarian-dinoflagellate symbioses. These authors measured a daily expulsion rate of 4.6% in Aiptasia pulchella under steady-state conditions, with those symbionts undergoing division (as determined by ³H incorporation) being preferentially expelled; they also observed preferential expulsion in the coral Pocillopora damicornis but not in the corals Montipora verrucosa (= Montipora capitata), Porites compressa, and Fungia scutaria.

Despite these various observations, the cellular mechanisms that lead to symbiont expulsion as a part of host-symbiont biomass homeostasis remain completely unexplored in the cnidarian-dinoflagellate symbiosis. In contrast, symbiont loss as a part of the coral bleaching process has been examined and reviewed extensively elsewhere (201, 402), with processes shown to cause symbiont expulsion including exocytosis of symbionts, host cell apoptosis, and host cell detachment. These same mechanisms could be at play in a more modulated fashion, as a part of symbiont population maintenance. Indeed, a recent histological study of the green Hydra symbiosis showed that immediately following feeding under normal conditions, the symbiotic algae migrate from their usual basal location to the apex of the cell and are then expelled into the gastrovascular cavity via either “pinching off” of aposomes or active exocytosis (109). Establishing the role that such mechanisms play in the cnidarian-dinoflagellate symbiosis is an obvious area for future research.

Symbiont degradation.Degraded symbiotic dinoflagellates tend to show a loss of circular symmetry, uneven and relatively dark coloration, an accumulation of unidentified droplets and starch-like globules, and a loss of cellular integrity (e.g., cell wall damage) before degenerating into smaller fragments that contain “accumulation bodies,” unpacked thylakoids, and starch (113, 364, 372). Degradation appears to be an intracellular process, regulated by the activities of the host's gastrodermal cells. Symbiont degradation by a coral (Astrangia danae) was first reported by Boschma (36), who noted dinoflagellate cells in various stages of degradation in the gastrovascular cavity of the host animal. Degraded symbionts and cellular debris were subsequently observed to form a substantial component of the pellets extruded by various tropical sea anemones (339, 340), and moribund or dead dinoflagellate cells have now been observed in the tissues of numerous species of reef corals (364–366) and other anthozoans (340, 372), as well as the hydroid Myrionema amboinense (112, 113). In the coral Stylophora pistillata, the frequency of degrading symbionts was found to be 5 to 6 times and 30 times greater in the digestive cells of the mesenteries than in those of the connecting sheet and tentacles, respectively (364). This finding supported the previous suggestion of Trench (372) and Steele and Goreau (340) that the symbiont cells are expelled into the gastrovascular cavity and then incorporated into the mesenteries by phagocytosis, where they undergo degradation; the unassimilated cell fragments are then released back into the gastrovascular cavity before being expelled into the ambient environment as a pellet. However, Trench (372) described the degradation of only senescent cells in the zoanthid Zoanthus sociatus, whereas Titlyanov et al. (364) also observed the degradation of apparently healthy symbionts.

Titlyanov and coworkers (364) estimated that, in eight different coral species, 1 to 6% of symbionts per day are in a state of degradation, a value that is notably similar to the symbiont division rate. The same authors reported a diel periodicity in the appearance of degrading symbionts, with a peak in the middle of the night and the whole degradation process lasting for about 6 h. A nighttime peak was similarly reported for M. amboinense, which in the morning was surrounded by expelled pellets composed largely of “unhealthy” (i.e., asymmetrical and dark) dinoflagellate cells (113); this peak equated to ∼1 unhealthy symbiont per host digestive cell out of an average of 2.67 symbionts per cell. Titlyanov et al. (364) considered this degradative process as “digestion,” based on four criteria: (i) its occurrence inside the host's gastrodermal cells; (ii) the observation of a diel rhythm of degradation; (iii) the absence of nuclei, proteinaceous structures, lipid droplets, and chlorophyll c in the expelled cell debris; and (iv) intensified degradation during periods of host starvation. Older or “weaker” symbionts were hypothesized to be more susceptible to lysosomal attack in the hydroid M. amboinense (112); in contrast, Titlyanov et al. (364) did not observe this mechanism in reef-building corals.

As with the process of expulsion, the mechanisms of symbiont degradation have been examined only in the context of stress and bleaching and not as part of a homeostatic mechanism. Processes that could result in degradation of symbionts include programmed cell death of the symbiont (98), a reengagement of the phagosomal maturation process in the host (see Appendix 1), and autophagic digestion of the symbiont by the host cell. Indeed, the observation of increased symbiont degradation during starvation is consistent with an onset of autophagy, which is commonly induced by nutrient limitation (see, e.g., reference 33). Moreover, Dunn and coworkers (99) found significantly increased bleaching in samples of A. pallida incubated at ambient temperatures in the presence of rapamycin, a promoter of autophagy. This suggests that autophagy could play a role in the regulation of the symbiont population even in the absence of stress. We are still a long way, though, from resolving the mechanisms underlying symbiont degradation and their relative importance in the regulation of the cnidarian-dinoflagellate symbiosis.

Studies of green Hydra.When green Hydra is starved, the symbiont size slowly increases until the cells accumulate at the size typical for division (6 to 9 μm in diameter), yet the frequency of division is reduced substantially (91, 221, 225 [but see reference 286]). Furthermore, in the dark, the symbiont cells divide at a smaller size than in the light (222), suggesting that in the light they grow beyond their critical size for division. Taken together, these observations indicate that algal proliferation in the green Hydra symbiosis is regulated, with potential restrictions on the cell cycle that limit progression through interphase and into the M phase.

Four different mechanisms for the control of the symbiont cell cycle have been suggested for the green Hydra symbiosis: (i) pH-stimulated release of photosynthate (maltose) to the host and concomitant reduction in symbiont growth (91, 223); (ii) host-regulated supply of a metabolite(s), a so-called “division factor” (219, 221, 222, 225, 265); (iii) host-regulated supply of inorganic nutrients to the symbiotic algae (31, 219, 223, 265, 266, 277, 299, 301); and (iv) production of a density-dependent inhibitor by the algae themselves (219, 265). With the exception of self-inhibition by the algae, these mechanisms have been explored experimentally.

Isolated Chlorella cells have been shown to release maltose in response to acidification, with maximal release at about pH 4.5 (43, 239, 258). Moreover, Douglas and Smith (91) demonstrated that those Chlorella strains that release large amounts of maltose do not grow at low pH in vitro, whereas those strains that release less maltose are able to do so. Together, these observations provide a mechanism by which a small, temporary shift in the pH of the symbiosome vacuole (i.e., the space between the symbiont and host) could modify the amount of photosynthate released to the host. This diversion of carbon could then cause carbon starvation and so hinder progression through the cell cycle either directly (91) or indirectly, by reducing the capacity for nitrogen assimilation and protein synthesis (90, 223, 229, 231, 300, 304). Direct evidence for the operation of such a mechanism in hospite is, however, lacking; indeed, immunocytochemical analysis of the symbiosome vacuole in green Hydra did not support the view that it is acidic (296).

As the algal symbionts continue to divide and increase in density, albeit slowly, in the digestive cells of starved green Hydra (91, 221) and excised and unfed regenerating heads and peduncles (217, 219), it was speculated that symbiont division may be limited (except during periods of host cell division) by restricted access to a pool or pools of metabolites (a “division factor”) derived from host digestion (221, 222, 228). However, the identity of this putative division factor has never been explored in detail.

In contrast, inorganic nutrients are well known to influence the symbiont division rate in green Hydra. In one study, the addition of a complex mixture of nutrients even caused the symbionts to overgrow their host (266). This suggests that the host might regulate symbiont growth by controlling the supply of inorganic nutrients arising directly from host digestion. In particular, while there appears to be little capacity for the host to withhold phosphate (411) and sulfate (60), there is evidence that the control of nitrogen delivery (especially excretory ammonium) to the symbionts is important (224–226, 299–304). Indeed, McAuley and Muscatine (225) showed that when Hydra is starved, the Chlorella cells can slowly progress through S phase of the cell cycle, but feeding is required before mitosis can occur and the cell cycle is completed; this observation suggests arrest of the cell cycle at the G₂/M checkpoint (Appendix 2), but how the host controls this is unknown. Enzymatic assays showed that Hydra has the potential to assimilate ammonium and withhold it from its symbionts (299). This ultimately led to a model in which the host maintains a low concentration of ammonium in the symbiosome vacuole except for when the algal symbionts divide; this flux of ammonium not only would provide nitrogen for amino acid synthesis but also would increase the pH of the symbiosome vacuole, thus reducing the rate of maltose release and leaving more carbon for symbiont growth (300, 305). Support for this model was provided by Rees (303), who reported rates of uptake of the ammonium analog methylammonium by freshly isolated Chlorella cells that were consistent with a low-ammonium environment in the symbiosome vacuole. However, whether this proposed mechanism actually operates in hospite has yet to be confirmed, while we still cannot be certain how it relates to the other mechanisms discussed, which could be interrelated.

As first suggested by Pardy (283, 284), the host has the potential to regulate symbiont division so that it is closely linked to the division of its own cells. When green Hydra is fed, the diel pattern of host cell division becomes synchronized, with a peak MI of about 2% that occurs 10 to 12 h after feeding (219, 266). Symbiont division closely matches this synchronous pattern (219), though the synchrony of host and symbiont division can be disrupted by starvation (221). A different pattern is seen in excised regenerating peduncles, where first the algae and then the host cells divide (217, 219), causing an initial increase and then a decrease in the symbiont density (285). In this case it was suggested that nondividing host cells with a full complement of symbionts inhibit algal division but that in recently divided host cells with fewer symbionts, inhibition is partially or wholly removed (217). This same density-dependent relationship has been suggested as the mechanism by which symbionts partitioned randomly and unevenly between host daughter cells at division ultimately attain similar densities in each digestive cell (227, 228). Competition between symbionts for their host cell's pools of inorganic nutrients or a “division factor” is consistent with this density dependence (228).

Studies of cnidarian-dinoflagellate symbiosis. (i) Resource limitation.There are parallels between the regulatory mechanisms suggested for the green Hydra symbiosis and those that operate in the cnidarian-dinoflagellate symbiosis, most obviously with respect to resource limitation, i.e., the restricted supply of inorganic nutrients and the loss of photosynthetic carbon to the host. Symbiotic dinoflagellates are typically nitrogen limited in hospite, as evidenced by their enhanced growth rate, population density, photosynthetic performance, and nitrogen status when supplied with an exogenous source of particulate or dissolved nitrogen (see, e.g., references 61, 64, 80, 161, 189, 230, 245, 250, 269, and 332). Similarly, there is evidence for phosphorus limitation, based on phosphatase activities (171, 172) (see below) and phosphate fluxes (249), and for carbon limitation, based on the enhancement of cell-specific photosynthesis by the addition of dissolved inorganic carbon (C_i) (200, 398) or when the in hospite density of symbionts is experimentally reduced (78). These limitations could explain the rapid growth of the symbiotic dinoflagellates upon release into the surrounding environment (232, 349) (but see reference 15) and during the early stages of symbiosis establishment, when the dinoflagellates can grow as rapidly as they do in culture (26, 322, 371). Moreover, the respiratory rate of Symbiodinium cells is higher in culture than when the dinoflagellates are freshly isolated from their host, providing further evidence that symbiont metabolism is inhibited when in hospite (134).

Given this evidence, alongside the host's capacity to regulate the delivery of carbon, nitrogen, and phosphorus to the symbiont (see below), it seems reasonable to suppose that, as in the green Hydra symbiosis, inorganic nutrients and/or metabolites play an important role in the control of symbiont proliferation in the cnidarian-dinoflagellate symbiosis. Indeed, Rahav et al. (294) calculated that in the coral Stylophora pistillata, unrestricted access to the host's waste ammonium would still only be enough to support about one-third of the maximal symbiont growth rate seen in culture. In addition, the leakage of photosynthetic carbon compounds to the host, perhaps as a result of a stimulatory “host release factor” (HRF) (see below), would further hinder the symbionts from achieving a state of balanced growth (106, 332); this unproven mechanism would be analogous to the influence of low pH in the green Hydra symbiosis.

How much photosynthetically fixed carbon is translocated?Quantification of the amount of photosynthetically fixed carbon translocated from the symbiont to host must take account of both the amount of carbon fixed in photosynthesis and the percentage of this production that is transported out of the symbiont cell. This quantification has largely relied on one of two methods, each with its own drawbacks. Indeed, it would be fair to say that carbon translocation has never been quantified conclusively. The earliest and most widely employed method involves labeling with [¹⁴C]bicarbonate in the light and measurement of the proportion of label found in the host's tissues. Using this method, translocation has been measured as anywhere between 5% and 60% of photosynthetically fixed carbon for a wide range of coral- and other cnidarian-dinoflagellate symbioses (51, 76, 107, 128, 166, 259, 268, 337, 351, 370, 386). Moreover, visual examination of autoradiographs from the temperate sea anemone Anemonia sulcata (= Anemonia viridis) suggested that >60% of photosynthate was translocated (360). The use of ¹⁴C has its limitations, however, especially because translocated products might be rapidly metabolized, thus liberating radiolabel from the system as ¹⁴CO₂ and leading to an underestimation of the percent translocation. Alternatively, ¹⁴CO₂ could be refixed by photosynthesis, thus confounding any translocation measurements, while there are also potential problems associated with the ¹⁴C-to-¹²C ratio disequilibrium (20, 21, 270, 333).

Given the questions surrounding the use of ¹⁴C labeling, Muscatine et al. (267) introduced an alternative method for the quantification of translocation that was based on the assumption that all photosynthetic carbon not used for respiration or growth by the dinoflagellate symbionts is passed to the host. This approach therefore requires knowledge of the net photosynthetic production (i.e., the gross photosynthetic carbon fixation less the carbon used for dinoflagellate respiration) and the amount of carbon used for dinoflagellate cell growth, both of which rely on a combination of direct and indirect measures that are explained in detail elsewhere (see, e.g., references 75, 267, 268, and 342). Typically, this so-called “growth rate method” estimates that >90%, and sometimes as much as 99%, of photosynthetically fixed carbon is translocated to the host under well-lit conditions (74, 75, 81, 268, 342). Once again, however, this technique and the translocation rates estimated are questionable, for three key reasons: (i) no allowance is made for carbon storage by the dinoflagellate cell, which undoubtedly occurs, largely via starch accumulation (251); (ii) symbiont cell growth is estimated indirectly only, based on the percentage of dinoflagellates in a state of division (the mitotic index [MI]) and the duration of symbiont cytokinesis, which is often assumed as 11 h based on a derived estimate from just one symbiont type from the jellyfish Mastigias sp. (267, 412 [but see reference 232]); and (iii) no method has yet been developed for the accurate measurement of symbiont (and indeed host) respiration in the intact symbiosis, where the nutritional interplay between the two partners has stimulatory effects on cell physiology (both photosynthesis and respiration) that are not replicated when the partners are studied in isolation (102, 154, 330).

These various knowledge gaps severely hamper the quantification of carbon utilization and flux within the cnidarian-dinoflagellate symbiosis and hence our understanding of the supposed mutualistic nature of the association. However, a very recent, though currently expensive, methodological development may offer a solution: multi-isotope imaging mass spectrometry (MIMS). This method allows for the high-resolution tracking and quantification of molecules labeled simultaneously with a range of stable isotopes (e.g., ¹³C or ¹⁵N) or radioactive isotopes (e.g., ¹⁴C) (195, 196, 405). The locality of these isotopes can be imaged by nanoautography and the amount of an isotopic tracer present at the subcellular level accurately quantified by comparing the experimental isotopic ratio (e.g., ¹³C to ¹²C) with the natural abundance ratio. This method could in fact prove to be a very powerful tool in the study of cnidarian-dinoflagellate symbioses, given its capacity not only to quantify nutritional fluxes but also to visualize the fates of numerous metabolic compounds within the symbiosis.

In what forms is photosynthetically fixed carbon translocated?Not only are we uncertain about the amount of photosynthetic carbon translocated to the host, but we also lack comprehensive and reliable data about the identity of the compounds released. Muscatine (257) isolated symbiotic dinoflagellates from corals as well as giant clams, incubated them in homogenates of their host tissue in the presence of [¹⁴C]bicarbonate, and then identified both the intracellular and released products by two-dimensional (2-D) radiochromatography; he had previously developed this technique for identifying the photosynthetic products released in the green Hydra symbiosis (255). Analysis of the medium revealed that the photosynthate liberated by the dinoflagellates was largely (>90%) in the form of glycerol, with glucose and a ninhydrin-positive compound also being released. In contrast, glucose was the major photosynthetic product identified in intracellular extracts of the dinoflagellates, with other intracellular products being various photosynthetic intermediates and lipid. By applying this method, glycerol has subsequently been identified as the primary form (24.8 to 95.0%) in which photosynthetically fixed carbon is released in a number of different cnidarian-dinoflagellate symbioses, with other reported compounds including glucose, the amino acid alanine, and organic acids such as fumarate, succinate, malate, citrate, and glycolate (351, 370–372, 386). In addition to these various compounds, microscopic observation of isolated dinoflagellates from the coral Stylophora pistillata and the sea anemone Condylactis gigantea led to reports of fat droplets being released from the cells, suggesting that lipids are also translocated (185, 287). Muscatine et al. (272) later showed that these “blebs” stained positively with a DNA-specific fluorochrome and were in fact host cell nuclei, though this should not be taken as evidence for a lack of lipid translocation in hospite (see below).

A persistent problem, however, is that the isolated dinoflagellates do not necessarily behave in the same manner as they do in the intact symbiosis (see, e.g., reference 351). Furthermore, identifying the mobile compounds in hospite is problematic given the intracellular nature of the symbiotic dinoflagellates and the likelihood that translocated carbon is metabolized rapidly by the host. Nevertheless, some information has been gleaned from (i) patterns of incorporation of ¹⁴C-labeled compounds into the host animal's tissues (19, 20, 32, 259, 369), (ii) biochemical and/or microscopic analysis of cnidarian tissues in different symbiotic states and under different irradiance regimes (50, 130, 153, 209, 392), as well as more targeted proteomic and ultrastructural analysis of “lipid bodies” isolated from symbiotic coral cells (288), (iii) identification of compounds in the host's tissues that are specifically synthesized by the symbionts (282), (iv) measurement of respiratory quotients (127), and, very recently, (v) analysis of the host genome (328). Together, these various approaches again point to glycerol, amino acids (including seven essential amino acids [392]), and various lipids as major components of the translocated material. However, none of these approaches is direct and hence conclusive with regard to the identity of the mobile compounds. Lewis and Smith (204) applied the somewhat more direct “inhibition technique,” originally developed for the study of the lichen symbiosis (94), to a range of corals and other cnidarians. This method involves the addition of potential translocated compounds to the medium surrounding a ¹⁴C-labeled symbiosis, which thereby saturate all sites of uptake in the host and cause any corresponding ¹⁴C-labeled compounds that are released by the symbionts to leak out of the symbiosis altogether. Lewis and Smith again concluded that glycerol, glucose, and alanine were the dominant forms in which carbon was translocated; they did not consider lipids.

More recently, Whitehead and Douglas (407) have used “metabolite” comparison: to elucidate the organic compounds translocated in the sea anemone A. viridis. This method compares ¹⁴C-labeled compounds in the host's tissues between treatments with [¹⁴C]bicarbonate in the light (i.e., photosynthesizing conditions) and when exogenous ¹⁴C-labeled substrates are provided in the absence of light. When the two treatments produce comparable labeling patterns in the host's tissues, then the exogenous compounds are thought to be similar to those translocated. This study suggested that glucose, succinate, and fumarate are all translocated, but crucially, there was no evidence for glycerol translocation. This finding was a major departure from previous reports and, interestingly, matched the situation in the intercellular giant clam-Symbiodinium symbiosis, where both temporal biochemical assays and radiotracer studies have suggested that glucose but not glycerol is translocated into the clam hemolymph (170, 306), while glycerol is released when the dinoflagellate symbionts are isolated and incubated in a host tissue homogenate (170). These studies highlight that there is still some uncertainty around the forms in which photosynthetic carbon is translocated to the host, with different symbioses not necessarily translocating the same compounds and isolated Symbiodinium cells responding differently to those in hospite, due to either the physiological impacts of isolation from the host (135) or the fact that homogenized host tissue is not representative of the environment to which symbiotic algae are exposed when inside their host. In addition, our steadily expanding knowledge of the Symbiodinium cell surface has revealed various glycoproteins that are likely exuded by the cell and hence form part of the translocated material (208, 212, 418) (see above). Such compounds are also exuded in Symbiodinium cultures (213), and their proteinaceous nature has enabled their transfer from symbiont to host to be confirmed in hospite (in the jellyfish Cassiopeia xamachana) via immunohistochemistry (214).

Currently, therefore, after several decades of research, we are still uncertain about the major compounds translocated and know little of the more minor compounds released. We also know virtually nothing about “reverse translocation” of carbon compounds, the transfer of metabolites from host to symbiont, though it is known that isolated symbionts can take up a range of organic and inorganic molecules from the surrounding medium (39, 40, 82, 89, 151, 344). The uptake of ³⁵S by symbiotic dinoflagellates from food ingested by Aiptasia sp. suggests that the reverse translocation of organic compounds does happen (58, 343), though the transfer of inorganic sulfate rather than organic sulfurous compounds cannot be excluded (343). Similar evidence via the supply of ³⁵S-, ³H-, and ¹⁴C-labeled food also exists for the green Hydra symbiosis (59, 256, 362). Importantly, however, two recent methodological advances should rapidly increase our knowledge of the metabolites translocated from the symbiont to host and vice versa.

What stimulates and controls the translocation of photosynthetically fixed carbon?The control of photosynthetic carbon translocation is one of the most controversial aspects of cnidarian-dinoflagellate symbiosis research. One hypothesis is that the host could limit symbiont growth, either by restricting the supply of nutrients or by actively “blocking” symbiont mitosis (see the previous section), thus generating a surplus of photosynthetic carbon for release (105, 267). It is important to note, however, that nutrient limitation may cause carbon storage rather than release in symbiotic dinoflagellates (78, 251), as it also does in free-living microalgae (375).

An alternative hypothesis has arisen from the observation that when isolated symbiotic dinoflagellates are incubated in homogenized host tissue and labeled with [¹⁴C]bicarbonate in the light, they release a substantial proportion of their photosynthetically fixed carbon (sometimes >50%) to the surrounding medium; these same dinoflagellates typically release <5% when incubated in seawater. In most cases an elevation in the cellular photosynthetic rate is also seen. Muscatine (257) was the first to observe this phenomenon (for corals and giant clams), and it has since been confirmed for a wide range of cnidarian-dinoflagellate symbioses (77, 139, 157, 260, 268, 351, 371) as well as a number of other alga-invertebrate symbioses (120, 261). Interestingly, however, no such activity is evident for the green Hydra symbiosis, where pH is instead thought to regulate translocation (43, 91). The identity of the chemical signal(s) responsible remains unknown, but there have been a number of attempts to characterize this so-called “host release factor” (HRF). It was originally suggested that HRF is proteinaceous due to its heat-labile properties (260, 351) and is >10 kDa in size (351). However, it has become increasingly apparent that HRF differs in different host species with respect to its thermal stability, host-symbiont specificity, and molecular weight (65, 125, 140, 157, 244, 260, 351), highlighting the dangers of generalizing across species. It has also become increasingly apparent that different cell signaling molecules may be responsible for the increased percent release of photosynthate and the control of photosynthetic performance (65, 141, 144, 390 [but see references 125 and 126), further complicating matters. Gates and coworkers (125, 126) suggested that HRF in the coral Pocillopora damicornis and the sea anemone A. pulchella consists of a suite of free amino acids (FAAs), as a synthetic HRF consisting of these FAAs both induced the release of photosynthetically fixed carbon and enhanced the rate of photosynthesis; it was proposed that these FAAs optimize the symbiont's intracellular environment, thus enabling it to fully realize its photosynthetic potential (30, 126). Evidence also suggested that mycosporine-like amino acids (MAAs) could be involved, as the active fraction (<4 kDa) had an absorbance peak at 320 nm and partially purified MAAs from other marine organisms enhanced the percentage of photosynthate released by dinoflagellates isolated from A. pulchella; MAAs did not elevate the photosynthetic rate, however, at least at the concentrations used (125).

These interesting findings stimulated a number of follow-up studies in the late 1990s, but these in turn raised considerable doubt about the universal role of FAAs as the HRF. For example, Wang and Douglas (390) suggested that the nonprotein amino acid taurine may be largely responsible for the increased translocation rate of dinoflagellates freshly isolated from A. pulchella, yet Withers et al. (415) could not replicate this response with dinoflagellates from the coral Plesiastrea versipora. Indeed, Withers and coworkers concluded that FAAs are not responsible at all for HRF activity in extracts of P. versipora, though the HRF of this coral appears to be of a correspondingly low molecular mass (<1 kDa) (140, 143). Similarly, Cook and Davy (65) found that the response (both percent release and photosynthetic rate) of symbiotic dinoflagellates to a <3-kDa extract of the coral Montastraea annularis was dose dependent but that there was no such relationship between the HRF response and the concentration of ninhydrin-positive FAAs present in tissue extracts; they did not measure taurine. These authors also found no effect of the amino acids glycine and glutamic acid on either photosynthate release or photosynthesis, though alanine did have an effect on the percent release only, particularly at high concentrations.

Outside this debate, only one research group (Hinde and coworkers, University of Sydney) has continued to focus intensively on the identity and mode of action of the HRF, focusing in particular on the coral Plesiastrea versipora (139–144, 157, 312, 313). To date, they have isolated two water-soluble signaling molecules of 1,000 kDa or less, one that stimulates photosynthate release but not the photosynthetic rate (HRF) and one that partially inhibits photosynthesis and hence was named “photosynthesis-inhibiting factor” (PIF). Homogenized P. versipora tissues induce similar stimulatory and inhibitory effects in dinoflagellates from a number of other cnidarians (142, 351), suggesting that the HRF and PIF in this coral might be “typical,” though photosynthetic inhibition does not occur in all cases (141, 142). Unfortunately, identification of the HRF and PIF has been severely hampered by their thermal lability and loss of activity when salt is removed during analysis (144), though the comparable effects of several synthetic compounds, such as clotrimazole, suggest that the HRF and PIF might act as calmodulin antagonists (144, 313). It is thought that the HRF from P. versipora stimulates translocation by either increasing glycerol synthesis or diverting it or other photosynthetic intermediates away from the symbiont's triglyceride stores (139). Moreover, the operation of both the HRF and PIF is unaffected by the presence of the symbiosome membrane complex around the isolated dinoflagellates (142).

Perhaps the biggest problem with all HRF studies is that they are conducted in vitro, with isolated dinoflagellates incubated in homogenized host tissue that is by no means representative of the environment in the intact symbiosis. Indeed, it is conceivable that HRF is an experimental artifact. This was suggested by different translocation patterns in response to host (A. pallida) starvation when the symbionts were either exposed in vitro to HRF or retained in hospite (77, 78). We therefore remain a long way from understanding the control of photosynthate release in cnidarian-dinoflagellate symbiosis. Hopefully, a better knowledge of the nutrient transporters associated with the symbiosome might provide some insight into their control, while knowledge of the host genome may facilitate the identification of candidate signal molecules whose function can be resolved by the likes of chemical genomics and reverse genetics (e.g., RNA interference), methods that are becoming widespread in more advanced fields such as plant science (reviewed in reference 347).

Light-enhanced calcification.Two groups of hypotheses, not mutually exclusive, have been proposed to explain the stimulation of calcification by the presence of the dinoflagellate symbionts.

Light regulation of skeletal morphology.While the morphology of the skeleton is genetically controlled, light, through the symbiotic relationship, can also modify it to some extent. These changes, known as phenotypic plasticity, can act both on the general morphology of the coral colony and at the level of skeletal microstructure and density (reviewed in references 357 and 367). These changes are adaptive, as they allow corals to optimize light capture by the symbiotic dinoflagellates. One can hypothesize that light may change the genetically controlled morphology by altering ion-pumping mechanisms and/or increasing OM synthesis and secretion toward the extracellular calcifying medium. However, the underlying mechanism is totally unknown. The diffusion of internal regulators such as the coral bone morphogenetic protein (coral BMP2/4) described in the calcifying epithelium (424) may be part of this mechanism.

Even if it is clear that the dinoflagellate symbionts are responsible for the LEC process, their exact role is still a matter of debate, and their action is probably mediated not by a single parameter but by several operating at the same time. The recent development of molecular methods such as transcriptomic analyses (145, 240, 310) will likely provide insight into the process by which the dinoflagellate symbionts enhance calcification, though it should be noted that the mechanism is metabolic rather than genetic, as the genes coding for proteins involved in coral calcification are not differentially expressed between light and dark conditions, while expression of photosynthetic genes is enhanced in light (247). Complete elucidation of this mechanism will therefore require the development of cell physiology and functional genomic (i.e., gene knockdown) approaches, while a comparative approach with nonsymbiotic corals will help to elucidate the mechanisms by which symbiotic dinoflagellates control calcification.