Metabolic interactions between algal symbionts and invertebrate hosts

ABSTRACT

Some invertebrates have enlisted autotrophic unicellular algae to provide a competitive metabolic advantage in nutritionally demanding habitats. These symbioses exist primarily but not exclusively in shallow tropical oceanic waters where clear water and low nutrient levels provide maximal advantage to the association. Mostly, the endosymbiotic algae are localized in host cells surrounded by a host-derived membrane (symbiosome). This anatomy has required adaptation of the host biochemistry to allow transport of the normally excreted inorganic nutrients (CO₂, NH₃ and PO₄³⁻) to the alga. In return, the symbiont supplies photosynthetic products to the host to meet its energy demands. Most attention has focused on the metabolism of CO₂ and nitrogen sources. Carbon-concentrating mechanisms are a feature of all algae, but the products exported to the host following photosynthetic CO₂ fixation vary. Identification of the stimulus for release of algal photosynthate in hospite remains elusive. Nitrogen assimilation within the symbiosis is an essential element in the host's control over the alga. Recent studies have concentrated on cnidarians because of the impact of global climate change resulting in coral bleaching. The loss of the algal symbiont and its metabolic contribution to the host has the potential to result in the transition from a coral-dominated to an algal-dominated ecosystem.

INTRODUCTION

Unicellular algae are responsible for approximately 45% of the carbon fixation on our planet (Field et al. 1998). A small percentage of this is fixed by algae that live in symbiosis with invertebrates (= host) (Fig. 1), many of which are found in clear tropical waters, with low nutrient concentrations and consequently low plankton density (Muscatine & Porter 1977). In this environment, the autotrophic algal symbionts provide a significant competitive advantage to the host largely supported by nutrient recycling between the symbiotic partners (for reviews, see Trench 1979; Smith & Douglas 1987).

Figure 1. A variety of successful invertebrate–algal symbioses exist including those between marine invertebrates and the dinoflagellate Symbiodinium which includes (a) scleractinian corals and (b) the giant clams (Tridacna sp.). Other symbioses include those between (c) the green alga Chlorella and the cnidarian hydra, and (d) the prochlorophyte Prochloron and ascidians. [Panel (d) was taken by A.W.D. Larkum].

Algal–invertebrate symbioses are predominantly endosymbiotic and the alga generally, although not exclusively, is associated, in the broadest of terms, with the host's digestive apparatus. For instance, in cnidarians the algae are found in host-derived vacuoles (symbiosomes) in the endodermal layer of the organism (Wakefield & Kempf 2001), while in tridacnid clams the symbionts are found in an extension (digestive diverticulum) of the stomach (Leggat et al. 2002). Can these algae live independent of their host? While some have resisted attempts to culture them, the implication is that some at least must be free-living because their hosts acquire them from the water column, although many are maternally inherited (Smith & Douglas 1987). While some symbionts can be found living in phytoplankton communities (Coffroth et al. 2006), they do not appear to be a major component.

Over the past 20 years, hermatypic corals have been a major focus for research on algal–invertebrate symbioses because of their fundamental role in the formation and maintenance of coral reefs. Historically, these habitats have provided a subsistence livelihood for people living in coastal tropical communities, a livelihood that is now under threat from anthropogenic pollution and global warming (Hughes et al. 2003). With the tourism and fishing industries in a number of countries dependent on the existence of corals, a concerted effort has been made to study the biology of this symbiosis along with other reefal algal symbioses such as tridacnid clams, anemones, ascidians and sponges (for a recent review of sponges, see Taylor et al. 2007). Other symbioses including the hydra–Chlorella association, radiolarians and foraminiferans symbiotic with dinoflagellates as well as the sea slug–chloroplast relationship have also drawn some attention.

The original deBary definition of the term symbiosis is the living together of differently named organisms. Moreover, symbiosis is frequently used to describe a relationship in which it is thought that all the partners derive mutual benefit from the association (i.e. mutualism as opposed to, e.g. parasitism). However, it is difficult to determine a clear benefit for the symbiont within an association (Douglas & Smith 1989), and these associations can be a shifting continuum which, as has been proposed recently, can under specific conditions shift into parasitism (Thrall et al. 2007). Irrespective of the precise definition, there are clearly metabolic interactions between the partners, and these interactions have dominated investigations of algal–invertebrate symbioses. The alga requires nutrients that are derived from the host or from the environment surrounding the host, and the host acquires photosynthetic products from the autotrophic alga. That these activities normally occur within the intact association presents significant experimental difficulties for the study of these metabolic relationships.

Symbiotic associations are not simply the sum of the organisms. Their metabolic activities are integrated with each other and, while a metabolic knowledge of each is invaluable, the sum may well be quite different and must entail the study of the intact association. These associations are dynamic entities, particularly at the metabolic level, and require both a qualitative and quantitative understanding of the metabolic interaction. Major developments in our understanding over recent years have come from the application of a number of approaches each with its own limitations.

The study of the individual organisms is one approach that can enhance our understanding of the symbiosis. However, the culture of both symbionts in the absence of the other has only been shown in a couple of instances. The best example is anemones where viable aposymbiotic animals can be prepared in isolation from their dinoflagellate symbionts (Muscatine, Grossman & Doino 1991). The dinoflagellates can be cultured separately. This is a valuable tool, but extrapolation to the symbiosis relies on confirmatory studies on the intact association as the host and dinoflagellates are known to modify their metabolism following separation (Weis & Levine 1996; Leggat, Badger & Yellowlees 1999).

Another approach is the study of the intact symbiosis. However, this has been fraught with problems. The separation of the symbionts or symbiotic fractions is difficult to achieve in a metabolically meaningful time-frame and without cross-contamination of the partners. Identification of metabolites and their site of synthesis is therefore problematic. Finally, the study of the algal symbiont freshly isolated from the invertebrate host is difficult and proving its ‘purity’ has proved impossible. Even the extraction of uncontaminated mRNA from one of the symbionts has proved impossible (Forêt et al. 2007; Leggat et al. 2007).

COMMON METABOLIC THEMES

The major metabolic theme in algal–invertebrate symbioses is the transport of inorganic nutrients via the heterotrophic invertebrate host to the photoautotrophic algal symbiont, which following assimilation and export, supplements the host's energy and nutrient demands. Host metabolism then allows for their recycling back to the algal symbiont. The extent to which this process is under the control of either host and/or symbiont is not resolved (Table 1).

Table 1.  Characteristics of various invertebrate–algal symbioses

Algal symbiont	Host	Localization	Algal export products	Evidence for host factors	CCM in alga
CCM, CO2-concentrating mechanism.
Dinoflagellate	Coral	Symbiosome	Glycerol, glucose, amino acids, lipids, C4 acids	✓	✓
	Sea anemone	Symbiosome	Glycerol	✓	✓
	Tridacnid clam	Extracellular	Glucose		✓
Chlorella	Hydra	Symbiosome	Maltose	pH	✓
Prochloron	Ascidian	Extracellular	?		✓

Nutrient acquisition and uptake by the alga

The capacity for photoautotrophy provided by the algal symbiont is the prime benefit to the host. This necessitates the transport of inorganic carbon (Ci), a suitable source of nitrogen (NH₃, NO₃^- or N₂), phosphate and other inorganic nutrients through the host tissues. As such, the alga can be considered to be providing a service to the host by removing both respiratory CO₂ and other metabolic breakdown products. While this recycles carbon and other elements, it does not provide for growth. For photoautotrophy to contribute to growth of the host, it must support the import of both Ci and nutrients from an exogenous source and transport them through the host to the alga. Whether the host has modified its transport capability and/or metabolism to accommodate the demands of its algal symbionts is still unclear because all animals rely on clearing these metabolic end-products from the cell because of their potential toxicity.

Given the low dissolved CO₂ concentration (∼ 12 µm) in seawater and the need for a high concentration surrounding the enzyme ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco), symbiotic algae need to increase the availability of CO₂ in the chloroplast to enable productive carbon fixation. This is a regular metabolic theme in both prokaryotic and eukaryotic algae (Badger & Price 1992), which is obviated by the presence of CO₂-concentrating mechanisms (CCMs). These have evolved in many algae, including those symbiotic with invertebrates, overcoming this limitation (Raven 2003). A CCM has been demonstrated for dinoflagellates symbiotic with tridacnid clams and corals (Leggat et al. 1999). While the presence of CCMs in other symbiotic algae has not been investigated, the presence of carboxysomes in Prochloron (Griffiths 2006), cyanobacteria symbiotic with sponges (Usher et al. 2004) and pyrenoids in some Chlorella species (Moroney & Chen 1998) would suggest that these symbionts also possess a mechanism to increase the concentration of CO₂ at the site of Rubisco. Whether the presence of CCMs overcomes possible carbon limitations on the intact association proposed by Al-Moghrabi et al. (1996) and Goiran et al. (1996) is still not resolved and may depend upon the species examined and the adaptations of the host.

Another recurring theme in invertebrate symbioses is the presence of carbonic anhydrase (CA) activity in both the host and algal symbiont. All algal CCMs require at least one chloroplast CA for CO₂ supply to Rubisco (Badger et al. 1998). However, evidence suggests that other CAs in both algae and the invertebrate host may be essential in the acquisition of Ci. In anemone–dinoflagellate symbioses, a symbiosis-inducible CA has been demonstrated (Weis 1991; Weis & Reynolds 1999), while in tridacnid clams loss of zooxanthellae is coincident with a decrease in CA activity in the host gills (Leggat et al. 2003).

While some algal invertebrate symbioses show reduced photosynthetic productivity with depth and changes in the expression of some of the proteins involved in photosynthesis (McCloskey & Muscatine 1984), there have been no studies on the effect of depth on CA in these symbioses. However, it has been demonstrated in the coral symbiosis that both CA and Rubisco activity are modified by water flow rates (Lesser et al. 1994). Rubisco activity in the dinoflagellate symbiont is directly proportional to water flow, while CA activity in the host is inversely proportional. The logical interpretation is that water flow reduces the cell boundary layer resulting in increased CO₂ diffusion, a consequent decrease in the requirement for CA and an increase in Rubisco activity resulting from an elevation in CO₂.

While Ci acquisition is of critical importance in supplying CO₂ for photosynthesis, a nitrogen source is also essential for normal cell maintenance and the production of amino acids by the alga. Both ammonium and nitrate are N sources for endosymbiotic algae, but dinitrogen (N₂) is also a potential source for nitrogenase-containing prokaryotes such as Prochloron (Paerl 1984; Kline & Lewin 1999). Of recent interest has been the discovery of N₂ fixation by symbiotic cyanobacteria within corals (Lesser et al. 2004), although how widespread this is within coral species, and their efficacy in providing ammonium to the dinoflagellate symbionts requires further investigation.

Compared with carbon and nitrogen acquisition, phosphate has received little attention despite its recognized potential to limit growth of algae. Dinoflagellates symbiotic with corals have a high-affinity phosphate transporter which is demand driven and light dependent (Jackson & Yellowlees 1990). The symbionts can store polyphosphate when phosphate is replete, but under ambient nutrient conditions for corals it is not detected (Jackson, Miller & Yellowlees 1989). Apart from carbon and nitrogen, phosphate is often considered to be a limiting nutrient in algae. Deviations from the Redfield C : N : P atomic ratio of 106:16:1, which is derived from heterogenous assemblages of marine phytoplankton, are often taken to infer a nutrient limitation of the relevant element (Atkinson & Smith 1983). While phosphate is often regarded as a limiting nutrient, virtually nothing is known of its acquisition in alga–invertebrate symbioses.

Diel cycles

The presence of a photoautotrophic symbiont within the host's tissues means that these animals must respond to profound changes during a normal light/dark cycle that would not normally be faced by non-symbiotic animals, including large daily fluctuations in O₂, CO₂ and NH₄⁺ tensions and pH driven by algal photosynthesis and metabolism. Levy et al. (2006) demonstrated that coincident with an increase in tissue oxygen concentrations, resulting from a change in the balance of respiration to photosynthesis in the coral symbiosis, there is an increase in host superoxide dismutase (SOD) and catalase, thus confirming the pioneering work of Dykens & Shick (1984) on SOD. These protective enzymes are essential in minimizing the effect of oxidative damage to cells. The correlation with symbiont SOD and catalase was not as pronounced, although this may be because of the small sample size. The light/dark cycle and its influence on photosynthetic activity also impose a diel cycle on algal cell division (Hoegh-Guldberg 1994; Fitt 2000).

Tridacnid clams also demonstrate clearly the effect of photosynthesis on host physiology and biochemistry. Photosynthesis directly modifies the animal's blood pH, being more alkaline at high light levels and tending to be acidic in the dark (Fitt, Rees & Yellowlees 1995). Coincident with the increase in pH is an increase in glucose (Rees et al. 1993) in the blood, indicating fixation of CO₂ by the algal symbiont and its subsequent export to the host as glucose. These results have been confirmed using ¹⁴CO₂ where the glucose in the blood supply was labelled as was glycogen stored in the mantle and adductor muscle of the clam (Yellowlees, unpublished results).

Other change driven by diel cycles may be less obvious but of no less importance. For example, the presence of mycosporine-like amino acids in coral tissue, which are thought to be produced by Symbiodinium, varies on a diel cycle (Yakovleva & Hidaka 2004). This group of compounds are an integral component in coral's response to harmful light levels where they are implicated in the absorption of UV and as antioxidants (Dunlap & Yamamoto 1995).

Host–symbiont architecture

In corals, sea anemones and green hydra symbioses among others, the symbiont is surrounded by a host-derived membrane (symbiosome membrane, Fig. 2). This is derived during the acquisition and division of the algal symbionts, and is analogous to the symbiosome in legumes where the plant membrane encloses the symbiotic Rhizobium cells. All nutrient trafficking between symbionts must occur through the symbiosome membrane, which is critical to metabolic interaction between symbionts and host.

Figure 2. The subcellular arrangement of the dinoflagellate Symbiodinium when in symbiosis with scleractinian corals. The dinoflagellate cell contents are surrounded by the algal plasma membrane and cell wall; this is in turn surrounded by multiple-layer host-derived membrane complex (the symbiosome), the coral endodermal cell and finally the host plasma membrane. This multimembrane architecture necessitates that metabolites that pass between the alga and host must pass through multiple cell membranes. (This image was provided by R. Quinnell).

The symbiosome, while well characterized in legumes, is still poorly understood in invertebrate symbioses. This is primarily because of the difficulty in preparing both intact symbiosomes and isolated symbiosomal membranes. Progress has been made on structure (Wakefield, Farmer & Kempf 2000) and isolation (Kazandjian et al. 2008), but this is limited to associations with Symbiodinium. Microscopy has revealed that apart from the host-derived symbiosome membrane, there are thecal vesicles surrounding the symbiont in dinoflagellate symbioses (Wakefield et al. 2000; Kazandjian et al. 2008). These provide additional layers of membranes between the host and the cytosol of the dinoflagellates through which nutrients must pass. While this describes membrane structure, molecular composition has proven largely intractable despite being crucial to an understanding of the symbiosis.

DINOFLAGELLATE–CNIDARIAN SYMBIOSES

The diversity of symbioses

Dinoflagellates form symbioses with a variety of cnidarians including corals, sea anemones and jellyfish. In all cases, the symbionts are found within a host-derived vacuole within the gastrodermal cell layer. This cellular arrangement presumably allows the cnidarian host some control over metabolite transfer to the alga. Dinoflagellate cell numbers can range from one symbiont per host cell (Muscatine et al. 1998) to over 60 in some hydroids (Fitt 2000). Although these symbiotic dinoflagellates are generally referred to as either Symbiodinium sp., or the common name of zooxanthellae, they actually represent a large species complex. To date, sexual reproduction has not been observed in this group, although it has been inferred (LaJeunesse 2001; Santos et al. 2003), and taxonomy is generally based upon sequence variation of taxonomic markers (e.g. ITS1, ITS2, 28s) (LaJeunesse 2001; Baker 2003; Coffroth & Santos 2005), with latest estimates suggesting that at least 160 distinct Symbiodinium types can be identified (LaJeunesse, personal communication), each of which may represent distinct species. Despite this diversity of both cnidarian hosts and symbionts, there is high host–symbiont fidelity (LaJeunesse et al. 2003, 2004; Santos et al. 2004; Thornhill et al. 2006). However, we know very little about how this specificity is maintained. Symbionts may be passed ‘maternally’ from one generation to the next through the eggs, termed maternal or vertical transmission, or each generation may acquire symbionts from their surrounding environment, so-called open or horizontal transmission. Despite the fact that those cnidarians with an open system of symbiont transmission should be susceptible to infection by all the available symbionts, there is a great degree of specificity (e.g. LaJeunesse 2002). In open transmission, symbionts are indiscriminately ingested in a manner similar to any other food particle. However, those vacuoles containing the correct symbionts avoid lysosome fusion (Colley & Trench 1983), possibly because of Rab protein interactions between the host and symbiont (Chen et al. 2003, 2004), and begin to proliferate. A variety of different symbionts can be taken up via this process and are observed in the early juvenile stages, which can lead to different host growth rates (Little, van Oppen & Willis 2004). However, generally, the adult is dominated by one symbiont type (del Gómez-Cabrera et al. 2007) with other symbionts maintaining a much smaller, residual population.

The evolution of symbiosis requires a significant alteration of the metabolism of both partners to coordinate and facilitate the symbiotic life history strategy. A key question then is: how large are these changes, and are there novel symbiotic genes? Intraspecific comparisons of symbiotic versus non-symbiotic anemones demonstrated significantly different protein profiles (Weis & Levine 1996); however, subsequently microarray analysis has failed to identify any ‘symbiosis-specific’ genes. This suggests that existing protein pathways have been seconded for a role in symbiosis and that differences in expression may be extremely subtle (Rodriguez-Lanetty, Phillips & Weis 2006; deBoer, Krupp & Weis 2007). Recently, there have been a number of studies that have indicated a role for lectin/glycan interaction in symbiont recognition and persistence (Jimbo et al. 2000; Lin, Wang & Fang 2000; Koike et al. 2004; Wood-Charlson et al. 2006) following the initial identification of strain-specific difference of symbiont surface antigens (Kinzie & Chee 1982). This is analogous to the roles for lectin–glycan interactions found in other well-characterized symbioses (Nyholm et al. 2000; Rodriguez-Navarro, Dardanelli & Ruiz-Sainz 2007). A number of α-mannose/α-glucose and α-galactose glycans on the dinoflagellate cell wall have been identified as potential targets for cnidarian lectins, with modification of these glycans leading to reduced dinoflagellate infection in larvae (Wood-Charlson et al. 2006) and adults (Lin et al. 2000). The potential for a sophisticated interaction between cnidarians and their symbionts can be illustrated by the surprising fact that cnidarians have a gene complement which contains many of those pathways involved in immunity and cell recognition, which were previously thought to be restricted to higher animals (Kortschak et al. 2003; Kuo et al. 2004; Miller, Ball & Technau 2005; Technau et al. 2005; Miller et al. 2007). In addition, the Symbiodinium genome is at least as large as that of humans (LaJeunesse et al. 2005) with a wide variety of unique genes, including a number which appear to be most closely related to that of bacteria (Leggat et al. 2007).

Inorganic carbon uptake and fixation

The success of the cnidarian–dinoflagellate symbioses is demonstrated by the dominance of coral reefs in the shallow waters of the tropics, with estimates of over 700 coral species found in symbiosis with dinoflagellates. A large amount of effort has gone into understanding the metabolic exchange between these two partners, in particular the transfer of carbon (both organic and inorganic), nitrogen and phosphate. The presence of photosynthetic symbionts within the gastrodermal layer in cnidarians necessitates the host playing a key role in the transport of Ci from the surrounding seawater to the photosynthetic dinoflagellate. This transport of Ci from an external environment to a sink within the tissue (the alga) contrasts with other animals where Ci is excreted from the body to the external media. This necessitates a number of physiological changes including localization of a variety of enzymes. The currently accepted model for Ci uptake by the host involves an H⁺-ATPase acidifying the boundary layer where bicarbonate is converted to CO₂ by an external CA isoform. This CO₂ subsequently diffuses into the host where it is trapped by conversion to bicarbonate and is then transported to the symbiosome by an unknown pathway (for reviews, see Furla, Allemand & Orsenigo 2000a; Furla et al. 2000b, 2005; Leggat et al. 2002).

Once in the symbiosome, the Ci is taken up by the Symbiodinium CCM (Goiran et al. 1996; Leggat et al. 1999) and fixed by a form II Rubisco (Whitney, Shaw & Yellowlees 1995; Whitney & Yellowlees 1995; Rowan et al. 1996). This Rubisco isoform has a CO₂/O₂ specificity which, although much higher than other form II Rubiscos, is significantly less than the form I Rubisco found in all other eukaryotic photoautotrophs (Whitney & Andrews 1998). Despite this limitation, photosynthetic CO₂ fixation is sufficient to meet not only the dinoflagellates' requirements for growth and respiration, but can also meet up to 95% of the host's energy requirements (Falkowski et al. 1993). These high photosynthetic rates generate significant diel variation in oxygen tensions within the coral tissue. At midday, during periods of peak photosynthesis, oxygen concentrations can be up to 250% air saturation and decrease to less than 5% during darkness (Kühl et al. 1995). As with other photosynthetic organisms, CO₂ fixation in cnidarians follows a diurnal hysteresis. However, unlike the majority of other organisms, higher photosynthetic rates occur in the afternoon rather than the morning at the same PAR levels, although this may vary significantly between species (Levy et al. 2004).

Understanding Ci transport and use in scleractinian corals is further complicated by the calcification process which produces the calcium carbonate skeletons of coral reefs. This also utilizes the coral intracellular Ci pool, and the relationship between Ci utilized for photosynthesis and calcification is still unknown. Recent analysis of the calcium carbonate skeleton of corals demonstrates that the coral skeleton is one of the most efficient collectors of solar radiation in nature and, by increasing the length of the internal light path, is able to increase the internal light field up to fivefold (Enriquez, Mendez & Iglesias-Prieto 2005). As with photosynthesis, Ci for calcification can be derived from host/algal respiration (Al-Horani, Al-Moghrabi & de Beer 2003; Al-Horani et al. 2005) or the surrounding seawater Ci pool. However, the exact pathway by which Ci is transported to the aboral ectoderm, where skeletogenesis occurs, is unknown (Gattuso, Allemand & Frankignoulle 1999). In addition to Ci, calcium must also be transported to the aboral ectoderm, and a putative plasma membrane calcium pump has been characterized and localized to the site of calcification (Zoccola et al. 2004). Elucidation of the exact mechanisms of Ci movement for calcification has taken on a greater urgency given the increasing acidification seen in the world's oceans as anthropogenically generated CO₂ enters the oceans, reducing pH and subsequently leading to reduced calcification rates (Hoegh-Guldberg 2005). Moreover, we know little about the effect of increased CO₂ levels (hypercapnia) on the physiology of these associations.

In cnidarian symbioses, photosynthetically fixed carbon appears to be transported to the host in the form of glycerol, glucose, amino acids or lipids (Muscatine 1967; Trench 1971a; Kellogg & Patton 1983; Patton & Burris 1983; Whitehead & Douglas 2003). Although no transporters have yet been characterized in the dinoflagellate or host, it has been suggested that the process is controlled by cnidarian host factors or host release factors (HRFs), which can stimulate the release of over 40% of the photosynthetically fixed carbon from the dinoflagellate (Trench 1971b; Muscatine, Pool & Cernichiari 1972; Sutton & Hoegh-Guldberg 1990). The exact nature of these HRFs is unclear, although some evidence points to them being low-molecular weight compounds, possibly amino acids (Gates et al. 1995; Cook & Davy 2001). In addition, there is evidence for a host-synthesized photosynthetic inhibitory factor which decreases the photosynthetic rates of the alga (Grant et al. 2001). If confirmed, this ability of the host to modulate photosynthate release may be one way in which it controls the intracellular algal population, as intracellular algal growth rates in culture, where doubling times are approximately 3 d (Fitt & Trench 1983), are significantly higher than that found when in symbiosis, where doubling times can range from 6 to 74 d (Muscatine et al. 1984; Wilkerson, Kobayashi & Muscatine 1988). Recently, it was reported that the export of photosynthate from different Symbiodinium strains varies significantly when in symbiosis with the same host (Loram, Trapido-Rosenthal & Douglas 2007), indicating that different algal strains may display significantly different behaviour under similar conditions.

Nitrogen utilization

Another way in which cnidarian hosts may control their algal populations is through nutrient limitation. It is well known that addition of nitrogen and/or phosphate sources leads to increased dinoflagellate growth rates, population densities and photosynthetic rates within the symbiosis (Taylor 1978; Marubini & Davies 1996; Koop et al. 2001). Indeed, nitrogen limitation has been shown to prolong the progression from G₁ to S phase in dinoflagellates including Symbiodinium (Olson & Chisholm 1986; Smith & Muscatine 1999). Two possible conclusions from this are that the host may be restricting the availability of nutrients, in particular nitrogen, to limit algal growth rates while maintaining high photosynthetic rates (Gunnerson, Yellowlees & Miller 1988; Rees 1991; Falkowski et al. 1993), or alternatively available seawater nitrogen concentrations may be limiting. Under N-limited conditions, photosynthetically fixed carbon that exceeds the alga's reduced carbon requirement, because of the high C : N ratio, must then be exported from the cell, where it is utilized by the host.

In cnidarian symbioses, both the host and the alga are capable of ammonium assimilation, with both possessing the enzymes glutamine synthetase (GS) and glutamate dehydrogenase (Anderson & Burris 1987; Catmull, Yellowlees & Miller 1987; Rahav et al. 1989; Yellowlees, Rees & Fitt 1994). The algal symbiont also possesses glutamine:2-oxoglutarate aminotransferase activity (Rahav et al. 1989; Roberts, Fixter & Davies 2001) which is required for the regeneration of glutamate for ammonium assimilation by GS to proceed. This ability to assimilate ammonium from the surrounding seawater, and that generated by host and algal metabolism, is thought to be a prime reason for the success of the symbiosis in the nutrient-poor tropical oceans. A variety of ammonium transporters have recently been found in Symbiodinium with similarity to bacterial transporters (Leggat et al. 2007).

Given that both host and alga are capable of ammonium assimilation, one of the main questions is: where is the primary site of ammonium assimilation in the symbiosis, the host or alga? D'Elia, Domotor & Webb (1983) proposed the depletion–diffusion hypothesis where the alga was responsible for the majority of ammonium assimilation, thereby creating a concentration gradient of ammonium from the seawater into the animal host and subsequently to the alga. In contrast, a number of studies have suggested that the host is limiting ammonium availability to the alga as a way of population control. If so, it would be expected that the majority of ammonium is assimilated in the animal fraction by GS which has a high affinity for ammonia. Early work found ammonium assimilation rates in the intact symbioses were enhanced in the light (Muscatine & D'Elia 1978) and that aposymbiotic cnidarians excreted ammonium supported the depletion–diffusion hypothesis. However, Wang & Douglas (1998), in a set of elegant experiments, found that dark ammonium assimilation, and protein and free amino acid pool sizes were the same as in the light following the addition of organic carbon (e.g. α-ketoglutarate) to the symbiosis. This supports the proposition that host ammonium assimilation is high and requires a constant supply of carbon skeletons, presumably from photosynthesis, for ammonium assimilation, which in the dark can be replaced by exogenous organic carbon. This evidence suggests that host control of ammonium concentrations is possible. Despite this, there is evidence that some nitrogen is assimilated by Symbiodinium and transported back to the host, in particular the essential amino acids histidine, isoleucine, leucine, lysine, phenylalanine, tyrosine and valine (Wang & Douglas 1999). This interplay between host and algal ammonium assimilation is yet to be clearly elucidated. Moreover, it is likely to vary significantly under different nitrogen regimes and between different host–algal combinations.

Symbiodinium is also capable of utilizing nitrate as a nitrogen source with long-term field studies showing that zooxanthellae densities are positively correlated to seawater nitrate concentrations (Fagoonee et al. 1999). Transporters for both nitrate and nitrite have been found among expressed sequence tags from Symbiodinium (Leggat et al. 2007), and the intact cnidarian symbiosis has been shown to remove nitrate from the water column (Crossland & Barnes 1977; Bythell 1990) which can lead to increased algal cell densities within the symbiosis (Marubini & Davies 1996, although see Ferrier-Pages et al. 2001) and mitotic indices (Fitt & Cook 2001). Nitrate is converted to nitrite through the action of the enzyme nitrate reductase whose cDNA sequence has been found (Leggat et al. 2007). Nitrite reductase subsequently reduces nitrite to ammonium. As would be expected, given that only the alga is capable of nitrate utilization, the majority of ¹⁵N in stable isotope studies is found in the algal fraction (Grover et al. 2003; Tanaka et al. 2006), but this nitrogen can be quickly translocated to the host (Tanaka et al. 2006) presumably as amino acids. Given that both nitrate and ammonium can serve as nitrogen sources, and are present at varying levels in seawater, a number of studies have examined the interaction between these two sources. Generally, nitrate levels in the oligotrophic waters that surround coral reefs, which are normally in the low micromolar range, are higher than ammonium, which is quickly assimilated by a variety of organisms and is found in sub-micromolar concentrations. Therefore, it is not surprising that cnidarian symbioses are capable of reasonably high rates of nitrate uptake. However, once ammonium, the preferred nitrogen source, is available, nitrate uptake rates decrease significantly (Grover et al. 2003). Elevated levels of either nitrate or ammonium through eutrophication of near-shore waters have the potential to severely disrupt the metabolism of both host and algal symbionts (Fabricius 2005).

In addition, ammonium and nitrate cnidarian symbioses are also capable of N₂ fixation. The ability of some corals to fix nitrogen has only recently been discovered and is attributed to intracellular cyanobacterial symbionts and not the cnidarian host or alga (Lesser et al. 2004, 2007). However, these bacterial symbionts are not always present. Nitrogen fixation occurs in a diurnal fashion with peak fixation in the early morning and late evening, thereby avoiding the high O₂ tensions in the coral tissue. The majority of the fixed nitrogen is utilized by Symbiodinium (Lesser et al. 2007). A variety of other genes associated with nitrogen fixation and metabolism have also recently been characterized from a metagenome study of the microbial associates of corals (Wegley et al. 2007), suggesting that a variety of coral-associated bacteria may be capable of nitrogen fixation.

Perhaps the least well-understood nutrient in cnidarian symbioses is phosphate. It has been suggested by a number of studies that cnidarian–Symbiodinium symbioses are phosphate limited (Jackson et al. 1989; Jackson and Yellowlees 1990) with increased coral growth and calcification rates seen in the presence of elevated phosphate (although see Renegar & Riegl 2005) in addition to higher algal densities (Koop et al. 2001). However, there are few studies that have extensively examined phosphate effects. As with nitrogen, phosphate uptake does not occur in aposymbiotic cnidarians, while uptake rates by the intact symbiosis are greater in the light than in the dark (D'Elia 1977). Phosphate uptake rates reduce after exposure to phosphate after as little as 1 h (Todd, Thornhill & Fitt 2006), suggesting that phosphate uptake becomes saturated or phosphate transporters are down-regulated. However, the exact mechanism is still unknown.

Despite decades of research, we still know very little about the more complex aspects of metabolic interplay between cnidarian–algal partners. The uptake and assimilation rates of various nutrients have been characterized; however, their ultimate metabolic destinations are often still unclear and will probably vary depending upon the relative nutrient concentrations. For example, up to half of the photosynthetically fixed carbon exported to the host is released from the host as mucus (Wild et al. 2004). This may change when nitrogen is not limiting. We still know little about the flux of metabolites between symbiotic partners and how they alter over a variety of time-scales (hours to days to months). Perhaps the most clear gap in our current knowledge is about the communication that exists between the symbiotic partners under various conditions. An example of this is the osmoregulation of both the host and alga (Mayfield & Gates 2007). Given their evolutionary success, there must be a complex conversation which coordinates the metabolism of host and symbiont. What form that takes and the molecular mechanism involved are still to be resolved. This may be of particular importance given that the greatest threat to the survival of coral reefs is the disassociation of the symbiosis, so-called coral bleaching, because of anthropogenic factors such as global warming (Hughes et al. 2003).

OTHER DINOFLAGELLATE SYMBIOSES

Tridacnid clams

Unlike corals, the dinoflagellate symbionts (Symbiodinium sp.) in tridacnid clams are not intracellular. They occur in a myriad of fine tubules emanating from the digestive diverticulum of the stomach. These tubules proliferate throughout the mantle of the clam and are pervaded by the open blood system (haemolymph) of the host (see Leggat et al. 2002). This provides a ready avenue for the export of photosynthetic products from autotroph to clam tissue. This, in contrast to corals, facilitates the design of real-time experiments on metabolic parameters in this symbiosis (Rees et al. 1993).

The algal symbionts can acquire nutrients either through the haemolymph or via the epithelium on the mantle surface. The haemolymph in turn is able to exchange solutes with the seawater through the gills of the clam. During photosynthesis by the symbionts, there is a significant decrease in Ci (>50%) in the haemolymph coincident with an increase in pH from ∼7.2 to >8.0 (Leggat, Rees & Yellowlees 2000). During photosynthesis, Symbiodinium exports glucose to the clam (Streamer, Griffiths & Thinh 1988; Rees et al. 1993); this in turn has been detected in ¹⁴C-labelled glycogen in both muscle and mantle tissue of the host (Yellowlees, unpublished results). The haemolymph glucose concentration is inversely proportional to the Ci concentration with a 3.5-fold increase in glucose concentrations at peak light levels over those present in the dark (Rees et al. 1993). While there is no evidence for glycerol export by the symbiont to the host, there is evidence for lipids (Johnston, Yellowlees & Gilmour 1995) and amino acids (Leggat et al. 2002) being released.

Carbonic anhydrase has been demonstrated to play a major role in the acquisition of the Ci in the host. Two CA isoforms have been located in the mantle and gills. One is a unique membrane-attached α-CA with dual catalytic domains (Leggat et al. 2005) and the other is a 32 kDa cytosolic α-CA. The expression of CA is correlated with the presence of symbionts (Yellowlees et al. 1993). While the 70 kDa CA did not appear to change with bleaching (loss of symbionts), the 32 kDa species decreased during bleaching in both mantle and gill tissue. This is consistent with the localization of this CA in the tubules which house the algal symbionts in the mantle (Leggat et al. 2003).

The concentration of ammonium in the haemolymph is a function of pH which is under the control of light. Ammonium concentration in Tridacna gigas is inversely proportional to light levels and haemolymph pH with the highest concentrations being in the dark (Fitt et al. 1995). This may affect the availability of ammonium to the algal symbiont, thus forcing its metabolism towards carbohydrate production rather than the production of proteins and nucleic acids required for algal division (which peaks prior to dawn). This is consistent with the demonstration that increased availability of ammonium reduces the starch sheath surrounding the pyrenoid in Symbiodinium (Ambariyanto & Hoegh-Guldberg 1996). Increased seawater ammonium concentrations have also been shown to down-regulate GS in clam mantle and gill tissue, thus decreasing the capacity of the clam to modify the ammonium concentrations available to the algal symbiont (Rees, Fitt & Yellowlees 1994). This explains the increased mitotic index in clams exposed to elevated concentrations of ammonium (Belda, Lucas & Yellowlees 1993). Even though all the symbionts potentially have access to this elevated ammonium, they will be in different stages of the cell cycle, and not all of them will divide after experiencing a period of elevated ammonium.

NON-DINOFLAGELLATE SYMBIOSES

Chlorella

Green hydra harbours the microalga Chlorella in its gastrodermal cells. Each algal cell is surrounded by a membrane of host origin (perialgal membrane), and there are 15–25 algae in each gastrodermal cell. Early work (Muscatine & Lenhoff 1963) demonstrated that photosynthetically fixed ¹⁴C is released by symbionts to the host. The only photosynthetic product that is released in appreciable quantities by symbiotic Chlorella is the disaccharide maltose, and the only condition known to promote release is acidic pH (Muscatine 1965; Cernichiari, Muscatine & Smith 1969; Mews 1980; Douglas & Smith 1984). At pH 4.0, 40–50% of carbon fixed is released as maltose (Cernichiari et al. 1969).

In common with other alga–invertebrate associations, aposymbiotic animals of the green hydra association release ammonium, whereas animals with symbionts do not (Rees 1986). The conventional interpretation of these observations is that the symbionts are responsible for uptake and assimilation of both internally derived and externally supplied ammonium, with the assimilated nitrogen being recycled back to the host in the form of an amino acid(s) (Muscatine & Porter 1977). However, there is no direct evidence for such recycling in the green hydra symbiosis. Another interpretation is that the animal host has low rates of deamination (Rees 1986), and that this largely explains the absence of ammonium release; carbon for host respiration is provided by photosynthate released by symbionts rather than amino acids derived from holozoic feeding (Rees & Ellard 1989). That levels of all measured 18 amino acids are greater in symbiotic than aposymbiotic animals (McAuley 1991) is consistent with this hypothesis. A further consequence of this assertion is that although symbionts are deprived of nitrogen by the host, they are not nitrogen deficient in the conventional sense of this term (Rees 1991). The release of photosynthate prevents accumulation of starch, and many properties of nitrogen-deficient algal cells appear to be a consequence of this accumulation of carbon (Rees 1991). Other characteristics of nitrogen-deficient cells appear to be specifically related to the machinery for acquiring the nutrient rather than carbon accumulation per se (Rees 1991). Specific predictions of this are outlined in Table 2, together with experimental evidence.

Table 2.  Comparison of characteristics associated with N-deficient free-living microalgal cells with those predicted for symbiotic Chlorella (Rees 1991)

Characteristic associated with N-deficient algal cells	Prediction for symbiotic algae	Experimental evidence	Reference
Supporting experimental evidence is given for the presence or absence of these characteristics in either (a) freshly isolated symbionts or (b) cultured high maltose-releasing strain (3N813A) of Chlorella grown at pH 5.0 (at which there is little or no growth because of maltose release).
Characteristics associated with carbon accumulation
Increased C : N (atomic) ratio	No	Partial (a), No (b)	McAuley 1992
Increased cellular starch content	No	No (b)	Dorling, McAuley & Hodge 1997
Increased rate of dark nitrogen assimilation (as a percentage of rate in light)	No	No (a)	Rees 1989
Increased rate of respiration on addition of nitrogen	No	No (b)	McAuley, Dorling & Hodge 1996
Decreased rate of photosynthesis	No	Small (b)	Dorling et al. 1997
Decreased rate of dark respiration	No	No (b)	Dorling et al. 1997
Decreased cellular chlorophyll content	No	Yes (a), no (b)	McAuley 1992
Decreased cellular protein content	No	No (a), no (b)	McAuley 1992
Characteristics associated with nutrient acquisition
Increased ratio of cytosolic:chloroplastic isoforms of glutamine synthetase	Yes	Yes	Rees, Shah & Stewart 1989
Increased rates of transport for nitrogenous compounds	Yes	Yes for amino acids; no for methylammonium	McAuley 1986; Rees 1990

Comparisons of rates of ammonium uptake by cultured 3N813A (which was originally isolated from Paramecium bursaria and releases maltose in culture at low pH) and freshly isolated symbionts of the association E/3N8 (a heterologous association between the European strain of hydra and 3N813A) show that the symbionts are similar to nitrogen-replete cells (Rees 1989; McAuley et al. 1996), suggesting that ammonium assimilation at low pH is restricted by the availability of carbon skeletons (McAuley et al. 1996). Freshly isolated symbionts have very low rates of uptake of the ammonium analogue ¹⁴C-methylammonium (Rees 1990). In contrast, both nitrogen-replete and nitrogen-deprived cells of 3N813A have high rates of methylammonium uptake. However, maintaining cells of 3N813A at low pH (4.0–4.5) in the presence of low ammonium concentrations (6–27 µm) results in rates of ammonium and methylammonium uptake that are similar to those of freshly isolated symbionts from E/3N8 (Rees 1990).

The ratio of cytosolic : chloroplastic isoforms of GS in the green hydra symbiosis is high (2.9), as it is in nitrogen-deficient cultured Chlorella (2.7–6.1), but not nitrogen-replete cells (0.6–0.7) (Rees et al. 1989). The glutamine : glutamate ratio is 0.29 in freshly isolated symbionts from E/3N8 (McAuley 1992) and the ratio is similar (0.22) in cultured 3N813A cells maintained at pH 4.0 (McAuley et al. 1996). Although externally supplied ammonium is taken up by the intact association (McAuley 1990), there is no effect of this ammonium on the glutamine:glutamate ratio of the symbionts (McAuley 1995), suggesting that access to this ammonium by symbionts is limited. At acidic pH, there is also a significant decrease in cellular levels of 2-oxoglutarate and phosphoenolpyruvate in cultured 3N813A (McAuley et al. 1996), presumably because of release of photosynthate, which would further restrict their ability to assimilate ammonium.

Freshly isolated symbionts can take up proline, serine, alanine, glycine, arginine, lysine (McAuley 1986, 1987a), leucine (McAuley 1988), histidine, methionine, threonine and glutamine (McAuley 1991). When brine shrimp containing ³H-amino acids are fed to green hydra, 0.8% (lysine) to 7.8% (arginine) of the radiolabel is recovered from the symbionts (McAuley 1987b, 1991). The C : N (atomic) ratio in symbiotic algae is intermediate between values for N-replete and N-deficient cells. The C : N ratio in N-replete 3N813A is 6.3 and 13.1 in N-deficient cells, which compares with a value of 8.6 in freshly isolated symbionts (McAuley 1992).

Indirect evidence that the space between the alga and the perialgal membrane is acidic has been summarized by Rees (1989). There is a reciprocal relationship between maltose release at low pH and growth rate (Douglas & Smith 1984). Given that the growth rate of symbionts in the green hydra symbiosis is markedly lower than in culture, maintaining the symbionts at low pH would provide a simple mechanism to ensure slow growth and release of photosynthate within the symbiotic association. However, use of immunocytochemical location of a weak base failed to detect any evidence for the perisymbiont space being acidic (Rands et al. 1992). This suggests that either: (1) photosynthate release by symbiotic Chlorella occurs via a molecule other than maltose; and/or (2) there is another unknown mechanism that promotes release of photosynthate by symbiotic Chlorella.

Prochloron

The prochlorophyte (which are a polyphyletic group within the cyanobacteria) Prochloron is found almost exclusively in symbiosis with ascidians. While present in phagocytes, the vast majority are extracellular and present in the test and cloacae of the host (Kühl & Larkum 2002). Their immediate environment is ill-defined, although exposure to low pH and high concentrations of vanadium is possible. Research is further inhibited by Prochloron's resistance to culture and the lack of aposymbiotic hosts.

CO₂ uptake is facilitated by the presence of CA activity associated with the cell surface of Prochloron. A much lower level of CA is present within the cell based upon patterns of CA inhibition by membrane permeable/impermeable inhibitors (Dionisio-Sese et al. 1993). Some of the internal CA activity will be associated with carboxysomes (Griffiths 2006) which are ubiquitous in prokaryotic carbon-concentrating mechanisms and are known to contain CA (So & Espie 2005). Rubisco is exclusively localized in the Prochloron carboxysomes (Swift & Leser 1989).

The translocation of photosynthetic products from Prochloron to supplement the ascidian's respiration appears to be host dependent. It represents between 12 and 56% of reduced carbon for host respiration (Olson & Porter 1985; Alberte, Cheng & Lewin 1987). The molecular composition of this has been suggested as early photosynthetic products (Kremer, Pardy & Lewin 1982).

While no specific metabolic information has been published, there is convincing evidence that in addition to carbon, nitrogen is also recycled in the symbiosis (Koike, Yamamuro & Pollard 1993). Ammonium is the major nitrogenous waste of the host (Goodbody 1974; Kühl & Larkum 2002), and this is readily assimilated by the algal symbiont (Parry 1985). The potential also exists for nitrogenase to contribute to the symbiosis nitrogen requirements. Nitrogenase has been detected in Prochloron present in the ascidian Lissoclinum patella (Paerl 1984). While disputed by Parry (1985), its presence is supported by nitrogen isotope ratios in host and Prochloron, which are consistent with nitrogenase activity (Kline & Lewin 1999).

ALGAL CHLOROPLAST SYMBIOSES

While not strictly an algal symbiosis, the kleptoplastic association between some invertebrates and algal chloroplasts is of growing interest. These associations have been studied in sacoglossan molluscs (see Rumpho et al. 2006), but also occur in marine ciliates (Stoecker 1991) and foraminifera (Chai & Lee 2000). These invertebrates actively farm the chloroplasts of multicellular algae and can acquire products of photosynthetic CO₂ fixation from these kleptoplasts. The key point of interest with this symbiosis is the strong possibility that a mixture of chloroplast protein stability and lateral gene transfer is crucial to the stability and productivity of these associations (Rumpho et al. 2006).

The majority of studies have been conducted on sacoglossan molluscs of the genus Elysia. Associations occur between the mollusc and a range of green, red and chromophytic algae. These can last for any period up to several months with the longest recorded (9 months) being between the chromophytic alga Vaucheria litorea and the host Elysia chlorotica (Green et al. 2000).

Like intracellular algal endosymbiotic relationships, the ciliate Strombidium capitatum has a host membrane surrounding the acquired chloroplast (Laval-Peuto 1991). However, in E. chlorotica which acquires chloroplasts from V. litorea, not only does the chloroplast appear to have no host membrane, but it appears to have lost the outer two membranes of the four-membrane-thick algal chloroplast (Rumpho, Summer & Manhart 2000).

The transfer of photosynthetically fixed CO₂ to the host and its subsequent metabolism have been demonstrated by a number of workers (e.g. Trench 1969; Kremer 1976). While conclusions cannot be clearly drawn as to the metabolite(s) transferred, results would suggest they are low molecular weight. Estimates of % photosynthate translocated to the host, based on ¹⁴C-labelling and δ¹³C values indicate variability between associations but may be up to 60% (Trench, Boyle & Smith 1973; Raven et al. 2001).

FUTURE RESEARCH DEVELOPMENTS IN ALGA–INVERTEBRATE SYMBIOSES

The study of alga–invertebrate symbiosis has only recently entered the genomic age, with either large-scale expressed tag sequencing and genomic data available for both symbiotic and non-symbiotic cnidarians (for review, see Forêt et al. 2007), the dinoflagellate Symbiodinium (Leggat et al. 2007), the intact coral holobiont (including bacteria) (Wegley et al. 2007) and the symbiotic cyanobacteria Prochloron. In addition, a variety of microarray studies have been published recently examining cnidarian symbioses (Edge et al. 2005; Rodriguez-Lanetty et al. 2006). However, studies of these symbioses have yet to truly utilize the power of proteomics and, in particular, metabolomics (Fiehn 2002). Many of the key questions about the metabolism and metabolic integration of alga–invertebrate symbiosis that were raised by researchers in the 1970s and early 1980 are still to be answered [see the seminal work of Muscatine, Trench and others (e.g. Muscatine & Hand 1958; Trench 1971b; Muscatine, Pool & Trench 1975; Muscatine et al. 1984; for a review of Len Muscatine's career, see Hoegh-Guldberg et al. 2007)]; however, it appears that these new technologies may provide the avenue to answer many of the fundamental questions still unresolved. Microarray and quantitative PCR analyses are now possible with the recent EST sequencing projects, although, as with other non-model systems, a significant amount of effort must be invested to ensure the identification of suitable housekeeping genes. A number of possible candidate genes have now been identified (Rodriguez-Lanetty et al. 2008; Seneca et al., unpublished results) and used (Levy et al. 2007) for studies on cnidarian hosts; however, no studies have published housekeeping genes for the photosynthetic symbionts. Another problem faced when using these techniques has been the difficulty in quantifying the highly variable cross-contamination by the two partners which may introduce variability between samples. A new method, based upon the ‘universal reference gene’ protocol, involving simultaneous extraction of RNA and DNA (which can be used for relative quantification) and RNA spiking may overcome some of these limitations (Mayfield et al., in review). It is hoped that overcoming these problems will allow researchers for the first time to examine the gene regulation of the partners, while in symbiosis under a variety of conditions, such as increased temperatures which can lead to disassociation of the symbiosis, and the expression of transporters that are used to move metabolites between the symbionts. The ability to move beyond ¹⁴C and stable isotope tracer studies with high throughput large-scale proteomic and metabolomic studies will further elucidate the complex metabolic interactions between the symbiotic partners. Furthermore, when linked to transcriptional analysis, studies of these types will provide a way to elucidate the function of the variety of genes of unknown function, which in some cases are up to 50% of the transcriptome, in the symbiotic partners. These technologies will now allow us to begin to unravel the intricacies of these symbioses.

ACKNOWLEDGMENTS

This review is dedicated to Bob Trench and the late Len Muscatine who have contributed so much to research in this field. We are grateful to A.W.D. Larkum for the image of Prochloron (Fig. 1d) and Rosanne Quinnell for providing the EM image of the dinoflagellate symbiosome (Fig. 2). We would like to thank Rosanne Quinnell ([e-mail]), Francois Seneca ([e-mail]) and Anderson Mayfield ([e-mail]) for access to manuscripts in preparation. We acknowledge financial support from ARC through the ARC Centre of Excellence for Coral Reef Studies (W.L. and D.Y.).