Trehalose Is a Chemical Attractant in the Establishment of Coral Symbiosis

Coral reefs have evolved with a crucial symbiosis between photosynthetic dinoflagellates (genus Symbiodinium) and their cnidarian hosts (Scleractinians). Most coral larvae take up Symbiodinium from their environment; however, the earliest steps in this process have been elusive. Here we demonstrate that the disaccharide trehalose may be an important signal from the symbiont to potential larval hosts. Symbiodinium freshly isolated from Fungia scutaria corals constantly released trehalose (but not sucrose, maltose or glucose) into seawater, and released glycerol only in the presence of coral tissue. Spawning Fungia adults increased symbiont number in their immediate area by excreting pellets of Symbiodinium, and when these naturally discharged Symbiodinium were cultured, they also released trehalose. In Y-maze experiments, coral larvae demonstrated chemoattractant and feeding behaviors only towards a chamber with trehalose or glycerol. Concomitantly, coral larvae and adult tissue, but not symbionts, had significant trehalase enzymatic activities, suggesting the capacity to utilize trehalose. Trehalase activity was developmentally regulated in F. scutaria larvae, rising as the time for symbiont uptake occurs. Consistent with the enzymatic assays, gene finding demonstrated the presence of a trehalase enzyme in the genome of a related coral, Acropora digitifera, and a likely trehalase in the transcriptome of F. scutaria. Taken together, these data suggest that adult F. scutaria seed the reef with Symbiodinium during spawning and the exuded Symbiodinium release trehalose into the environment, which acts as a chemoattractant for F. scutaria larvae and as an initiator of feeding behavior- the first stages toward establishing the coral-Symbiodinium relationship. Because trehalose is a fixed carbon compound, this cue would accurately demonstrate to the cnidarian larvae the photosynthetic ability of the potential symbiont in the ambient environment. To our knowledge, this is the first report of a chemical cue attracting the motile coral larvae to the symbiont.

Introduction depend on it, suggesting that the symbiont may have stimulated a feeding response but this was not investigated further [12]. Other studies on the sensation and detection of the partners focused on a larval-produced attractant for the symbiont. For example, data from Horiguchi et al. [20] and Hollingsworth et al. [21] support a "beacon hypothesis" where motile Symbiodinium specifically swim toward light generated by green fluorescent proteins on the mouthparts of cnidarian larvae. Other experiments describe chemical attractants [22], such as nitrogen-based cues [23]. In contrast, a demonstration of signaling from the symbiont to the animal partner comes from the freshwater system, where the symbiont of the cnidarian hydra produces maltose to attract the host [24], [25].
The objective of this study was to determine whether Symbiodinium synthesize and excrete chemoattractants sensed by cnidarian larvae. Two recent reports demonstrate that coral, crustacean, and cephalopod larvae as well as juvenile reef fish prefer water from healthy reefs, suggesting that chemical cues sensed by these organisms play a role in recruitment and settlement [26], [27]. We examined both the production and excretion of sugars from Symbiodinium, and sugar levels in adult coral from three species and developing larvae from one species. To determine whether the coral larvae would follow chemical cues, we tested their chemo-orienting and onset of feeding behaviors when placed in a Y-maze and presented with potential chemoattracting sugars. The final set of experiments focused on analysis of enzymes that might break down the sugar within the Symbiodinium and coral, as well as analysis of the developmental expression of these enzymes.
Three species of coral were used in these experiments, with the majority of the data collected from individual adult polyps of F. scutaria and their larvae. Comparative data was also collected from adult Porites compressa and Pocillopora damicornis fragments to determine whether members of different scleractinian families behaved similarly to Fungia. F. scutaria is a useful model organism for these studies, because its spawning, development [28] and symbiotic onset [11] are well described, and occur at least 4 times throughout the summer months (following the full moon) [28]. Uptake of symbionts in F. scutaria occurs around 3 to 5 days post-fertilization [12]. Within a few more days, the larvae change shape, becoming more spherical, and stop swimming and begin crawling before settling and undergoing metamorphosis (with approximately 50% of larvae having settled by day 7) [12]. The uptake of Symbiodinium may aid in metamorphosis and settlement, but this appears to be somewhat variable [12].
Our array of experiments focusing on defining the chemoattractants excreted by Symbiodinium and identifying the host enzymes responsible for sugar degradation, coupled with suggestions about how this might work on the reef, support the hypothesis that trehalose is a chemical attractant in the establishment of coral symbiosis.
for the described experiments were generated and reared in bowls cleaned daily, consisting of eggs from one individual with sperm from 5 to 7 males. For experiments, thousands of larvae were collected from at least 3 to 5 different bowls to accommodate any slight developmental differences in the bowls. Only one such pooled sample was collected on each developmental day, so the 'n' for larvae indicates the number of different pooled larval samples. For Symbiodinium, the 'n' indicates the number of individual corals the algae were extracted from or the number of pellets of naturally excreted Symbiodinium either from captive or wild coral. For adult coral, the 'n' indicates the number of discrete individuals, unless otherwise stated. The sample size for each experiment is defined in each figure legend or table.

Tissue preparations
Fungia scutaria coral larvae. Coral larvae (without symbionts) were concentrated in 40-μm cell sorting baskets (Falcon, Fisher). Larvae used for enzyme and internal sugar assays (methods described below) were blotted through the mesh to remove excess seawater and the larval concentrate (20,000 to 40,000 larvae/tube) was transferred into 1.5 ml microcentrifuge tubes and rapidly frozen in liquid nitrogen to −196°C, before storage at −80°C until further processing.
Adult coral tissue. Enzyme levels and internal sugars were measured from small coral pieces (~2cm square, n = 5 individuals from each species) from three species, F. scutaria, P. damicornis and P. compressa. Pieces were flash frozen in liquid nitrogen and airbrushed to remove tissue from the skeleton into a Ziploc bag, and then squeezed into a preweighed microcentrifuge tube.
Extracted and naturally excreted Symbiodinium. Symbiodinium were extracted from adult coral tissue with the following methods. Adult coral tissue (tissue/Symbiodinium complex) was dissociated from the skeleton with a water pik following the methods of Hagedorn et al. [29]. Adult tissue was collected into a large beaker, moved to 50 ml plastic centrifuge tubes, and then centrifuged at 3,200 × g for 15 minutes to remove excess water. The tissue was homogenized using a glass homogenizer, and resuspended into filtered seawater. Samples were then centrifuged at 9,800 × g for 10 minutes, causing Symbiodinium to pellet at the tube bottom. In the first cleaning, the coral tissue layer was removed from the top of the Symbiodinium pellet. In some experiments when a cleaned coral tissue stimulus was needed, this layer was saved. Otherwise it was discarded and the Symbiodinium were centrifuged 3 to 4 times to further clean them and then used in various experiments.
Naturally excreted Symbiodinium were isolated from F. scutaria pellets produced around spawning from both captive and wild animals. Captive animals were observed for 3 to 4 days before, during and after spawning. The wild animals were observed on the reef off of Coconut Island for 7 consecutive days in August 2013 during and after the spawn. Pellets were collected using 5 ml transfer pipettes and placed into 1.5 ml microcentrifuge tubes, removing excess seawater and topping to 1000 μl with FSW (filtered seawater) and homogenized using a microcentrifuge sample pestle to disperse the pellet without shearing the symbionts. Following the methods of Hagedorn et al. [29], a Walz Junior pulse amplitude modulation (PAM) Chlorophyll Fluorometer was used to measure the functionality of photosystem II in the expelled Symbiodinium as a indicator to ensure there were live symbionts within the pellet. This was performed with all pellets isolated from the captive animals. However, because the samples collected on the reef were contaminated with sand and algal debris from wave action, PAM values were not acquired beyond the first night of collection for pellets obtained in the wild. Symbiodinium in each sample were counted with a hemocytometer on an Olympus BX41 at 200x under both white and fluorescent lighting, using a 450 to 480 nm band pass filter. If the sample demonstrated glowing red cells of the correct size and shape to be Symbiodinium under fluorescent lighting and a PAM quantum yield above 0.15, the sample was scored as containing live Symbiodinium, and processed as described above.
Weight and dry weight measurements and calculations. The percent wet and dry weight was measured for adult coral tissue and coral larvae and Symbiodinium to allow us to normalize values for sugars and enzymes across tissues and species for more accurate comparisons. Some measurements were taken from previous work, as noted in the table or figure. The wet weight of adult coral tissue was determined from small pieces of coral (~2cm square, n = 5 individuals from each species), of F. scutaria, P. damicornis and P. compressa that were flash frozen in liquid nitrogen. Each piece was then airbrushed to remove tissue from the skeleton into a Ziploc bag, squeezed into a preweighed microcentrifuge tube, reweighed to determine the wet weight, and processed as described below. The wet weight of developing coral larvae was measured from thousands of concentrated larvae collected in 40-μm cell sorting baskets (as described above) on each developmental day. Then, the larvae were placed into preweighed 1.5 ml microcentrifuge tubes (n = 4 to 5 tubes/developmental day). The tube was reweighed to determine the wet weight and processed as described below. The wet weight for the Symbiodinium from each adult coral species was determined from extracted symbionts (n = 5 individuals from each species of the 3 species) that were placed in preweighed microcentrifuge tube, filled with cells (~2 × 10 6 /ml), centrifuged at 8,100 × g, excess water decanted, the pellet dried by blotting with a Kimwipe, reweighed to determine the wet weight and processed as described below.
To determine the dry weight of all tissue types, samples were placed into a drying oven at 60°C for 3-7 days. At the end of the drying period, the microcentrifuge tubes were closed and immediately weighed again. Once the weight of the tube was subtracted, the percent dry weight was calculated as (g dry weight/ g wet weight) × 100. Microcentrifuge tubes without any tissue were also weighed, left in the oven and reweighed as control for changes in the tubes during the drying period. No change in weight was noted in the empty tubes during this process. For calculations of enzyme activities and solute contents per dry weight, the actual grams dry weight were used because the amount of water held in different tissues varied.

Assessment of potential sugar chemoattractants
Manipulations of Symbiodinium. Excreted sugars were measured from extracted or naturally expelled Symbiodinium as described [31] with some modification as noted below. Symbiodinium were diluted to 2×10 6 cells/ml in 1.5 ml plastic microcentrifuge tubes, and then mixed 1:1 with either filtered seawater (FSW), FSW and coral tissue (~50 μl fresh or previously boiled for 10 min at 100°C tissue [32], 7 vials/treatment) or FSW with~1,000 live Day 1 F. scutaria larvae (n = 4), resulting in a one ml final volume. Thousands of larvae from each sample were concentrated in a 40-μm cell-sorting basket before use. We used Day 1 larvae to reduce the possibility that the larvae might consume the trehalose. Samples were incubated in microcentrifuge tubes at 24 to 25°C in a flowing seawater table to maintain temperature under natural sunlight for 1 h. The Symbiodinium samples were spun at 9,800 × g for 10 minutes. The supernatant was removed and the remaining pellet and separate supernatant were rapidly frozen in liquid nitrogen. To analyze physiological processes, Muscatine [33] used radioactive bicarbonate and a 100% ethanol extraction followed by two-dimensional chromatography for analysis, instead we used a 70% ethanol extraction and analyzed the symbionts' seawater for trehalose and glycerol using enzymatic assays [34] (see also below).
Measurement of sugars present in coral, coral larvae, Symbiodinium and seawater. Frozen tissue aliquots were homogenized in 70% ice-cold ethanol, with proteins allowed to precipitate overnight; then the homogenates were centrifuged at 15,000 rpm for 30 min and the supernatants were run overnight in a rotary evaporator to remove the ethanol. The evaporates were dissolved in 600 μl pure water.
Samples were analyzed by spectrophotometry utilizing enzymatic assays. Kits and reagents were all from Sigma-Aldrich and used according to the manufacturer's directions. Glycerol was measured with reagent #F6428, sucrose with kit #SCA-20 and maltose with kit #MAK018 containing acid alpha-glucosidase (a maltase). Trehalose was analyzed with pig-kidney trehalase following the methods of Sigma bulletin SSTREH01. Trehalose and maltose are both disaccharides composed of two glucose molecules. Therefore, the product of enzymatic digestion of one of these sugar molecules is two glucose molecules. The glucose produced by the trehalase and maltase enzymatic assays was measured using a hexokinase glucose assay (Sigma-Aldrich #G3293) with the amount of trehalose or maltose present originally corresponding to half the glucose found after cleavage. To test for glucose already present in the tissues, extracts were measured without the addition of the hydrolyzing enzymes (trehalase, etc.).
Chemotaxis assays. Chemo-orienting behavior of coral larvae to a variety of sugars was measured in a Y-maze. A chamber was created using a glass slide with two stacked pieces of electrical tape (~600 μm total finished thickness) with a Y-shape cut out in the center of the tape. A total volume of 250 μl of 0.2 μm filtered seawater was added to the maze, then 20 μl of seawater containing larvae (100 to 500/run, pooled from at least 5 bowls) were added to the base of the Y, and finally 0.5 μl volume of a solution (0.1 M sucrose, trehalose, maltose, glycerol or glucose or seawater) was added to one arm and 0.5 μl of 0.2 μm filtered seawater was added to the other. To determine whether the larvae could respond to lower concentrations of trehalose in the maze, 0.01 and 0.001M solutions were also tested. The arms containing the test solutions were randomly changed. All test solutions, as well as the control seawater, contained 1 mg/ml methylene blue (Sigma, St Louis, MO) to visualize the movement of the solute front.
Two types of runs were examined: 1) test-runs had one arm with a sugar solution added (n = 10 to 20 runs/solution) and the other arm with seawater added, and 2) control-runs had seawater in both arms (n = 10 to 20 runs/solution). The larvae were allowed to swim for 3 min, photographed, and then the larvae present in each arm were counted. This area was chosen because it was the limit of visual detection of the dye after 3 min. To compare the responses, the percent difference was calculated as (a−b)/((a+b)/2)Ã100, where a and b are the numbers in each arm, and then the numbers were translated with a constant and then log (10) transformed.
Measurement of enzymes present in coral, coral larvae and Symbiodinium. Activities of trehalase, sucrase and maltase were measured in weighed aliquots (~0.2 g) of frozen aposymbiotic larvae, adult tissue (with symbionts within the tissue) or extracted symbionts. All tissue was homogenized in 0.01M potassium phosphate buffer pH 7.4 (with a 1:4 ratio of coral tissue to phosphate buffer). The reaction mixtures contained 600 μl of 0.1M potassium phosphate buffer (pH 7.4) containing 20 mM trehalose, sucrose, or maltose with 150 μl tissue homogenate, and were incubated at 30°C for 60 min. The tube was then boiled for 5 min to denature enzyme activity, cooled, and centrifuged for 10 min at 15,000 rpm. Then, 100 μl of the supernatant was assayed for glucose in 1.00 ml of a spectrophotometric hexokinase glucose assay (Sigma-Aldrich #G3293). We report the specific activity, expressed in units of μg glucose generated/h/g dry weight. To test for glucose already present in the tissues, reaction mixtures were incubated without added disaccharides, boiled, and then tested for glucose; those values were subtracted from the glucose measured with the disaccharides. Disaccharides were also added to extracts after boiling to test for glucose contamination in the disaccharide reagents, which was found to be undetectable.
Identification of potential trehalase-encoding genes in a coral. Sequences from Drosophila melanogaster (D.m.AAM68193.1) and Saccharomyces cerevisiae (S.c. YDR001C) were utilized as queries to search Acropora digitifera genome sequences hosted at http:// marinegenomics.oist.jp/genomes/gallery?project_id=0 using tblastn. A. digitifera was used as it represents the most complete coral genome available and, like F. scutaria acquires its Symbiodinium from the environment [35], [36]. The A. digitifera sequence with significant homology to the query sequences was examined using Prosite (prosite.expasy.org).
Statistics. All data analyses in this study were performed using Instat 3.0 or Prism 5.0 (Graphpad, San Diego, CA) and Excel (Microsoft, Redmond, WA). For multiple group comparisons, normality was tested graphically and ANOVAs were done with a Tukey's Multiple Comparison Test or Student-Newman-Keuls post-test. All percentage data was log transformed prior to ANOVA analysis. These tests are identified in the results when reporting the P-value, and those values 0.05 were considered significant.

Extracted and naturally-expelled Symbiodinium excrete trehalose
It seemed most likely that a fixed carbon compound would be a chemoattractant synthesized by Symbiodinium. Such a compound would be feasible for the symbiont to make and would reflect photosynthetic ability under the ambient environmental conditions. Therefore, to determine what sugars Symbiodinium synthesized and excreted as possible attractants, we repeated and extended Muscatine's 1967 [33] experiments. He found that about 40% of the Symbiodinium's photosynthate (produced in 1 h) was liberated as glycerol in response to the presence of the host coral tissue, but found no glycerol release when the tissue was absent or boiled. We saw a similar change in glycerol excretion when tissue-extracted Symbiodinium were incubated with host coral tissue. Strikingly, however, we found that trehalose was released constitutively, without the need for co-incubation with the host tissue (Fig. 1). Our analysis shows lower levels of trehalose in the presence of coral larvae compared to other samples, but it might be that trehalose levels in these samples are somewhat decreased by the uptake and utilization of trehalose by the larvae in those samples.
We believe that no one has previously reported trehalose as important in the Symbiodinium/cnidarian symbiosis previously because: i) the levels released from Symbiodinium in 1 h incubations are relatively low; ii) it may not be produced from radioactive carbonate during photosynthesis during incubations using tracers in 1 h; and iii) 100% ethanol was used for paper chromatography of sugars in Muscatine's 1967 experiments [33], and trehalose can crystallize out of ethanol solutions at 80% or higher concentration [37], thus removing it from consideration by this process.
To obtain a more detailed profile of sugars released from Symbiodinium in a variety of different subclades, we used freshly isolated Symbiodinium from F. scutaria, P. compressa and P. damicornis. These symbionts are all Symbiodinium clade C, but belong to different subclades [38]. Symbiodinium isolated from the cnidarian host (10 ml at 10 6 or 10 7 cells/ml in FSW) were incubated in sealed vials at 24 to 25°C under natural sunlight for 6 h (F. scutaria symbionts) or 24 h (all others). During this incubation, F. scutaria Symbiodinium released trehalose, which accumulated to 650 μM (Table 1). In contrast, far less sucrose, maltose, and glucose were released (Table 1). While the numbers of symbionts and the time of incubation are different in these experiments, the same pattern was found for Symbiodinium isolated from the two other coral species (Table 1), suggesting this may be a broad physiological trait of cnidarian symbionts. Using the measurements from F. scutaria's Symbiodinium (e.g., 650 μM from 10 7 cells/ml in 6 h; Table 1), we estimated that each Symbiodinium cell released~1 ×10 −14 moles of trehalose/h.
Having established that Symbiodinium can release trehalose into the environment when freshly isolated from host tissue, we next asked whether Symbiodinium naturally released by F. scutaria during their reproductive phase also excreted trehalose. To answer this, we cultured naturally released pellets (n = 5) in 1.0 ml FSW for 6 h in sunlight. Their trehalose production, analyzed in the FSW after the cells were removed by gentle centrifugation, was 12.2 ± 10.7 μM (SD). On average 429,000 cells/ml (± 249,000 SD) within the pellets were scored as live Symbiodinium (defined in Methods). Using the value from Table 1 (650 μM trehalose from 10 7 cells/ml over 6 h), the average 429,000 live cells in the pellets should have produced an estimate of 29 μM trehalose over 6 h.  To understand more fully the sources and levels of trehalose and other sugars in the symbiont/coral complex, we analyzed symbionts, larvae and adult tissues for internal disaccharides, as well as glycerol and glucose. F. scutaria symbionts contained a wide variety of sugars within their cells, including trehalose, glucose, and glycerol (Table 1 and Yancey et al. [34]). Aposymbiotic F. scutaria larvae contained trehalose and glucose in relatively constant, low amounts throughout the developmental period studied, and adults contained similar levels of glucose, but higher levels of trehalose and glycerol (Table 2). We do not know whether the trehalose found in the symbiotic adult tissue is made by the symbionts, cnidarians or associated bacteria. Similarly, the low levels of trehalose in the aposymbiotic larvae could be synthesized by the larvae, maternally provisioned, or from bacteria. Sequence analysis of various cnidarian databases did not identify trehalose synthetases that were clearly of animal origin, as the sequences identified were most closely related to protist sequences and may be due to contamination with Symbiodinium DNA (data not shown).

Coral larvae are attracted to trehalose
Given that Symbiodinium excrete trehalose, we asked whether it might play a role in attracting coral larvae for the initiation of symbiosis. To test this, we placed Day 3 F. scutaria larvae without symbionts in a Y-maze and tested whether they were attracted to specific sugars ( Fig. 2A). The larvae showed distinct orienting responses only to glycerol and trehalose (P <0.05, ANOVA, Tukey's Multiple Comparison test) while the other solutions were neutral (P >0.05, ANOVA, Tukey's Multiple Comparison test), and a 0.1M trehalose solution was the lowest concentration that produced this response (Fig. 2B). In these experiments with a 0.1 M trehalose solution, there was an obvious accumulation of larvae in the arms where trehalose or glycerol was added. Their selective behavior is demonstrated in supplementary video clips. Specifically, for sugars other than trehalose or glycerol, and for control (FSW) treatments, the larvae swam constantly with no clear pattern to their behavior, and they did not change their morphology during the trial (S1 Fig.). However, when the larvae discovered the trehalose or glycerol solutions, they stopped swimming, changed their morphology from elongated to more spherical, settled to the bottom of the maze, and began feeding behavior (S2 Fig.). The spot where the larvae stopped swimming, sank to the bottom and started feeding was at a point 2 mm from the end of the maze, indicated by the shaded area of inset in Fig. 2B. This volume is approximately 10 μl. Thus, if 0.5 μl of a 0.1M trehalose solution were to fully mix in this volume in 3 min, a rough upper estimate of the coral larvae's sensitivity to trehalose would be 5 mM.
Because the Symbiodinium released glycerol only when in contact with host tissue, it is unlikely to function as a chemoattractant during the initiation phase of symbiosis. In contrast, Symbiodinium release trehalose constitutively and coral larvae are attracted to it, suggesting it might play a role in the initiation of symbiosis.

Trehalase is present in corals and developmentally regulated in coral larvae
We measured activities of trehalase and other disaccharidases in the developing F. scutaria larvae, as well as in adult coral tissues and in isolated symbionts to determine whether these organisms had the capacity to metabolize these sugars. To compare activities among the extracts, which varied widely in water content, we normalized the data with tissue dry weights following the methods of Hagedorn et al. [39], [29] and Yancey et al. [34].
In F. scutaria larvae, we found that the trehalase and sucrase enzymes were developmentally regulated and strikingly increased over time. The specific activity of trehalase, expressed in units of μg glucose generated/h/g dry weight (Fig. 3A), was low in Day 1 larvae (294 ± 146 SD units), but rose 11-fold to peak at Day 3 (3,291 ± 181 SD units). This is especially noteworthy because Day 3 has been previously identified to be the first time that F. scutaria larvae are capable of uptake of Symbiodinium [12]. Sucrase activity also increased during this same developmental period but had lower overall values during the entire period (38.9 ± 7.4 SD units at Day 1, peaking at 321 ± 80 SD units at Day 3). Nevertheless, both enzymes displayed a greater than 10-fold increase at Day 3 over their initial Day 1 values. While the apparent maltase activity was considerably higher than trehalase, it was without a dramatic rise during development (10,873 ± 254 SD units at Day 1; 15,747 ± 1305 SD units at Day 3), displaying only a 1.5-fold increase at Day 3 over their initial Day 1 values. It is important to note that the activity measured as "maltase" may be a less specific enzyme that also hydrolyzes large glucose polymers stored in the larvae. For example, mammalian acid maltase and alpha-glucosidase not only hydrolyze maltose but also larger glucose polymers [40] and the high levels of enzyme activity may reflect the catabolism of these polymers in the larvae.
Intact adult coral tissue (containing Symbiodinium) from all three species tested also contained substantial levels of trehalase activity, ranging from 1,159 ± 263 to 2,891± 752 SD units (Fig. 3B). F. scutaria adults had about 35% of the peak activity of the larvae.
However, when F. scutaria's Symbiodinium were separated from their coral host and examined, they had much lower levels of enzyme activity (407 ± 173 SD units per g dry weight; Fig. 3B), corresponding to less than 15% of the peak larval activity and 35% of the adult F. scutaria activity. Therefore, Symbiodinium cells contained and excreted substantial concentrations of trehalose but had low levels of the enzyme to break it down, suggesting that this sugar may be synthesized more for export than as a source of internal energy. . Coral larvae and adults contain trehalase activity. The mean specific trehalase activity was calculated by determining how much exogenously added trehalose was cleaved into glucose (in units of μg glucose/h/g dry or wet weight ± SD). A) Trehalase from F. scutaria larvae is per dry weight; a, b, c, d indicate statistical differences at P < 0.01 (ANOVA, Student-Newman-Keuls post-test, n = 3 to 5 with thousands of larvae/assay from each developmental time point). The highest activities correspond to the time point when coral larvae are first capable of symbiont uptake and may be actively seeking them, at Days 3-4. B) Adult coral tissue from 3 different coral species (black bars) and one Symbiodinium species (n = 3 individuals from each coral species, n = Symbiodinium from 5 F. scutaria individuals, grey bar) all contain trehalase activity; a, b, c indicate statistical differences (ANOVA). The highest trehalase activity was in P. damicornis, significantly higher (P < 0.05) than the other corals; and activity was found to be significantly lower in the Symbiodinium compared to all corals (P <0.01 to 0.05, Student-Newman-Keuls post-test), suggesting the dinoflagellates may only metabolize this sugar at low rates. Coral Symbiosis Cue-Trehalose

Trehalose as a Chemoattractant
Based on several lines of evidence presented here, we believe that trehalose produced by Symbiodinium may be a cue to help attract F. scutaria larvae for initiation of symbiosis. First, trehalose was produced and excreted by recently excised and wild Symbiodinium without the need of a host factor. Second, Fungia larvae followed both a glycerol and trehalose gradient, but as glycerol was only released by Symbiodinium in close proximity to live coral tissue, this cue is unlikely to play a role in the orienting events. Moreover, recent evidence indicates that glycerol release may be a stress response in Symbiodinium [41]. In contrast, it is unlikely that the excretion of trehalose into the environment by Symbiodinium is a symptom of stressed, lysed or dead cells. If the cells were lysed or dying the ratio of internal and external sugars should be similar. However, the ratio of excreted and internal trehalose and glucose in the Symbiodinium samples are quite dissimilar (Tables 1 and 2). Additionally, the Symbiodinium demonstrated a differential response to the presence of coral tissue for glycerol excretion (Fig. 1), which was not observed for trehalose excretion. Taken together, these data suggest that the Symbiodinium cells we have assayed have active metabolism and are regulating their physiology. Third, both adult and developing coral larvae produced the specific enzyme needed to break down trehalose, and the concentration of the enzyme increased during the developmental period when F. scutaria is taking up its symbionts (Days 3 and 4). Moreover, even though the symbionts produce trehalose, they produce very little of the trehalase enzyme.
BLAST analysis identified a likely trehalase enzyme from a scleractinian coral, supporting the enzymatic assay data described above demonstrating cnidarian trehalase activity. We used the fruit fly (Drosophila melanogaster) and budding yeast (Saccharomyces cerevisiae) trehalase sequences as queries. Sequences from these organisms were used because trehalase enzyme activity has been mapped to a specific genetic locus in both, demonstrating unequivocally these genes encode trehalases, and not related disaccharidases [42,43]. We searched the published Acropora digitifera genome (http://marinegenomics.oist.jp/genomes/gallery?project_id=0), another scleractinian coral that takes up symbionts at its larval stage [44]. A single match with significant homology to both D. melanogaster and S. cerevisiae was found in both organisms. The A. digitifera sequence was found on scaffold 6518:25988-27781 and was 37% identical and54% similar to the D. melanogaster protein throughout the entire length of the protein (S3 Fig.). Further analysis confirmed the A. digitifera sequence had 100% identity with the signature motifs of trehalase enzymes as defined by Prosite (http://prosite.expasy.org/ PDOC00717) (S3 Fig.). Additionally, a search of F. scutaria transcriptome data produced sequence with significant homology to trehalase enzyme in A. digitifera (S4 Fig.).
Taken together, these data show that coral larvae and corals have the capacity to utilize the sugar trehalose supplied by Symbiodinium at the developmental stage of the onset of symbiosis. In addition, this ability to metabolize trehalose persists into the adult stage and is present in at least three species of symbiotic cnidarians.
Importantly, trehalose has the characteristics of a reliable signal in this system in the sense that it is costly to the sender and transmits accurate information about its metabolic state. To generate the fixed carbon compound trehalose for use as a chemical cue, the symbiont's photosynthetic ability must be intact and producing energy above its own metabolic needs. Given that the provision of fixed carbon from the symbiont to the cnidarian host is essential for maintenance of this symbiosis, this signal may allow the cnidarian to identify healthy symbionts. Amongst fixed carbon compounds, trehalose may be an ideal chemo-attractant because it is a non-reducing disaccharide containing a very stable α,α-1,1-glucoside bond between two glucose molecules, producing one of the most thermodynamically and kinetically stable non-reducing sugars [45].

Chemo-attraction on the Reef
Understanding more about trehalose as a chemical cue on the reef is critical to understanding its biological relevance. In the laboratory, F. scutaria larvae demonstrated a clear preference for trehalose in the Y-maze assay. Moreover, the trehalose gradient stimulated the same changes in body morphology as when F. scutaria were taking up Artemia (brine shrimp) homogenate and Symbiodinium as described by Schwarz et al. [11]. While these Y-maze experiments do not mimic field conditions, as these types of perceptual experiments are difficult to conduct in complex environments, they indicate that coral larvae can specifically perceive trehalose and begin feeding, which is an essential step in the uptake of Symbiodinium.
There is some literature stating that larvae cannot use chemical cues for chemo-attraction over large distances because they cannot swim for long distances in a directional manner due to water flow in their environment, reviewed in [46]. However, recent studies disagree with these arguments. For example, Koehl and Hadfield have performed and reviewed a number of studies examining flow under field and laboratory conditions, and examined larval movement under various conditions and concluded that small larvae swim slowly but can alter their location in ambient water flow [46]. In addition, plannulae from three different Acropora species have recently been found to swim towards a preferred water source in a two-channel Atema plume [47] at a flow rate of 4.2 mm/s [26]. The overall flow rate in Kaneohe Bay is 10-20 cm/s [48] about 20 times the rate in the plume. However, the flow within the reef structure where the F. scutaria live would be much lower and conceivably within the flow rate measured in the Atema plume, suggesting that the larvae swimming toward a source would be well within the physical parameters of the system.
Other factors that support the hypothesis that F. scutaria can swim in a directional manner to follow chemical cues are related to the physiology of their eggs and the fine scale hydrodynamics of the reefs that F. scutaria inhabit in Kaneohe Bay. While the dispersal of F. scutaria larvae has not been described and the motility of adults prevents using their location as a proxy for larval settlement [49], it is clear that the adults live and breed in small holes and indentations on the reef, which provide a calm microenvironment, especially during reproduction in the summer months. F. scutaria eggs are negatively buoyant, which causes the developing embryos and resulting larvae to sink to the bottom in these calm microenvironments. Therefore, the larvae may not experience typical turbulent planktonic conditions. Additionally, water velocities within the reef have been shown to be relatively low with regions of both upward and downward flow, depending on the shape of the surface [50]. At sites other than Kaneohe Bay, F. scutaria has similarly been reported to be found in crevices on rubble substrata [49,51,52], which likely also give rise to calm microenvironments for the larvae formed after spawning. Taken together, these data suggest that there may be very low velocities in the immediate area of the F. scutaria larvae and the larvae may be able to detect cues in their environment and affect their position accordingly.
What trehalose concentration might be in the reef sediment and would it be high enough for the larvae to perceive? Free-living Symbiodinium reside only on reefs [55], [56], and, within the reef habitat, they are most concentrated within reef sediments (1000 to 4000 cells/ml) with lower densities (up to 80 cells/ml) in the water column [56]. Previous work has shown that larvae from the coral Acropora monticulosa acquire Symbiodinium earlier and at higher densities when they are incubated with reef sediment from the oxic layer compared to water from above the reef [53]. From our work on the concentration of trehalose released from Symbiodinium (e.g., 10 7 cells/ml in 6 h) into seawater, we estimated that each cell released 7 × 10 −14 moles of trehalose over that time. If we use the highest level of Symbiodinium/ml in the sediment, these numbers would produce 2.8 × 10 −7 M trehalose in 1 ml of sediment over a 6 h period, assuming no loss by diffusion or bulk transport. This concentration is well within the range of sensitivities of many marine organisms, such as lobsters that have 10 −8 M sensitivities [54]. However, there are several questions related to the robustness of this estimate. Diffusion is required to make the gradient and will lower the trehalose concentration in the sediment, but the rate of trehalose diffusion under these conditions is not known and likely varies substantially in different regions of the reef due to water flow and composition of sediment. In contrast, it is possible that this may be an underestimate of trehalose levels. It is not clear how deeply into the sediment Littman et al. sampled [56]. Takabayashi et al. [55] suggest that the oxic depth of the sediment is actually quite shallow (1-5mm), and given Littman et al.'s sampling methods, it is possible that the density of Symbiodinium in the oxic layer, where enough sunlight is available for photosynthesis, is actually much higher than reported. Furthermore, increases in local concentration of Symbiodinium from clumpy pellets released by sexually mature Fungia may play an important role in boosting levels of symbionts. Finally, it is not yet known when the trehalose might be perceived by the cnidarian larvae. Trehalose may act to orient the larvae towards the symbiont when they are in close proximity and then induce feeding behavior to facilitate uptake. Alternatively, it may be that trehalose plays a role in a more long-distance signaling cue of the type recently described [27], [28]. There may be only one role for trehalose, or the role may differ depending on the cnidarian species and the features of a particular reef.
Given the potentially small numbers of symbionts in the environment, the presence of sufficient adults containing healthy symbionts may be critical to seeding the reef during spawning to maintain and boost symbiont populations, thus allowing the symbionts and larvae to find one another as has been described in other systems [56,57]. Recent data demonstrates that some marine larvae demonstrate a preference towards water taken from live, healthy coral reefs [27], [28]. Taken together with our data, this suggests that there may be a positive feedback loop where healthy coral reefs recruit marine larvae which settle and sustain the reef. However, if the numbers of healthy animals on the reef decreases, there may be an insufficient signal for recruitment, leading to further degradation of the reef.

Conclusions
We have found that in the symbiotic relationship between corals and Symbiodinium, fixed carbon in the form of trehalose produced by the symbiont plays a key role in how these partners find each other. This signal may yield information about the nutritional status and potential fitness of the signal generator and may serve the critical role of attracting the partners towards one another, where further information, in direct cell-cell interactions, can then be exchanged. Given that a likely benefit of horizontal infection is the identification of symbionts best suited to a particular environment, the trehalose cue, reflecting photosynthetic capability, may be particularly important to a potential host.
Symbiosis between coral and intracellular dinoflagellates is critical for coral viability and reef ecosystem productivity. Most larval coral take up symbionts from their environment; but how these symbiotic partners signal and sense one another has been unknown. To help solve this mystery, we demonstrated in F. scutaria coral and their Symbiodinium: 1) a release of the disaccharide trehalose into the environment by the symbionts, 2) a coral larval chemo-attraction and feeding response in the presence of trehalose, 3) the presence of the enzyme, trehalase, in coral tissue, but not in the symbionts, and 4) adult release of live Symbiodinium, thus seeding the reef. To our knowledge, this is the first report of a specific chemical cue attracting the motile coral larvae to the symbiont.
Supporting Information S1 Fig. Chemo-orienting Behavior: F. scutaria larvae responding to FSW in both arms of the Y-maze. Larvae were added to the base of the Y-maze, 0.5 μl of FSW was added to at the tops of the arms, and then the larvae were allowed to swim around for 3 min. This video shows the last 10 sec of the run, demonstrating a lack of chemo-orienting behavior in either arm.