Adult N. vectensis (measuring between 2 and 5 cm in length) were kept under a 12-hour:12-hour light–dark cycle (lights on at 7:00 h (Hawaii standard time); Zeitgeber Time ZT = 0; lights off at 19:00 h; Zeitgeber Time ZT = 12 - under full spectrum lights, (Corallife 50/50 bulb, approximately 3,600 lux), in glass bowls with 13 parts per thousand (‰) filtered local seawater (diluted with distilled water, referred to as ‘1/3X seawater’) in a temperature-controlled incubator (17°C). Water was changed every day (ZT = 0), and animals were fed twice weekly with freshly hatched brine shrimp. A second group of adult individuals were kept in constant darkness for 20 days under the same conditions, with water changes also occurring at ZT = 0. For a second experiment, to measure dissipation of transcriptional oscillations in animals removed from light cycling, groups of animals were kept in constant darkness for 24 h, 48 h and 96 h and otherwise with the same conditions.

For the melatonin treatments, adult animals were kept in constant darkness for 5 days and then exposed to 0.1 μM melatonin (Sigma Chemical Co., St. Louis, MO, USA) for the subsequent 15 consecutive days in constant darkness. Exogenous melatonin addition was administered at the time of the transition to the subjective night (ZT = 12), corresponding to a time of peak melatonin production (see Results). After 12 hours of incubation, the animals were rinsed in fresh 1/3X seawater and placed in new dishes for the remainder of the experiment.

To determine the temporal production of melatonin in adults, embryos (48-hour post-fertilization), and juveniles (1-week and 2-week post-fertilization), animals were individually collected from their respective treatment every 3 hours and placed in 1.5 ml microcentrifuge tubes and flash frozen with liquid nitrogen. Embryos and juveniles were also kept from egg stage under the same condition of the adults in 12-hour:12-hour light–dark cycle. Eggs were fertilized in the ZT = 6. Individuals for RNA extractions were individually placed into a 1.5 ml tube with TriPure Reagent (Roche Inc, USA), then flash frozen.

In situ hybridizations were performed as previously described 40 . Embryos were collected from specific time points post-fertilization (24, 48, 96, 144 and 168 hours - polyps were not fed during the stages). Eggs were fertilized in the ZT = 6. The animal collection was also at ZT = 6, occurring 24, 48, 96, 144 or 168 hours after fertilization and corresponding with the developmental stages - blastula, late blastula, gastrula, planula, and late planula, respectively. All stages were fixed in fresh ice-cold 3.7% formaldehyde with 0.2% glutaraldehyde in 1/3X seawater for 60 seconds and then postfixed in 3.7% formaldehyde in 1/3X seawater at 4°C for 1 hour. Fixed embryos were rinsed five times in PBS buffer plus 0.1% Tween 20 (PTw) and once in deionized water, and transferred to 100% methanol for storage at -20°C. Early embryos were removed from the jelly of the egg mass by treating with freshly made 2% cysteine in 1/3X seawater (pH 7.4 to 7.6) for 10 minutes. Planula and late planula were relaxed in 7% MgCl2 in 1/3X seawater for 10 minutes prior to fixation. In situ hybridization using 1 to 2 kb digoxigenin-labeled riboprobes for Clock, Timeout, and Cryptochromes were performed to determine the spatial and temporal distribution of transcripts in each stage. Probe concentration ranged from 0.5 to 1.0 ng ml^-1, and hybridizations were performed at 70°C for 40 hours in 50% formamide. Probe detection was achieved by incubation with an antidigoxigenin antibody conjugated to alkaline phosphatase (Roche, Inc). Subsequently, the presence of alkaline phosphatase was detected by a colorimetric detection reaction using the substrate NBT-BCIP. Specimens were photographed on a Axioskop II with a Zeiss Zxiocam HRc.

Serotonin (5HT) content was quantified using a sensitive enzyme-linked immunosorbent assay method (IBL International, Hamburg, Germany). The embryos samples (approximately 60 embryos per sample) were centrifuged (3,000 × g for 2 min) and the sea water removed. Samples were reconstituted in 120 μl of the assay buffer of the kit. To normalize the assays, 2 μl of each sample was used to protein measures. Briefly, the assay was performed as follows: 20 μl of samples and 5HT standard dilutions were applied to 96-well microtiter plates previously coated with goat anti-rabbit antibody followed by 50 μl biotin-labeled 5HT and 50 μl rabbit antibody against 5HT. After incubation overnight at 4°C and washing with wash buffer, 150 μl of fresh prepared enzyme conjugated was added. Samples from the developmental stages, standards, positive and negative controls were incubated for 1 hour at room temperature with gentle mixing, washed again, then 200 μl p-nitrophenylphosphate substrate was added, and the enzyme reaction was terminated after 60 minutes by addition of 50 μl of p-nitrophenylphosphate stop solution. Absorption was measured at 405 nm in a spectrophotometer and the concentration of 5HT was calculated from the reference curve. A previous test adding known concentrations of serotonin to the N. vectensis samples showed us that the recovery was approximately 90% (data not shown).

Total RNA was extracted from samples using TriPure Reagent (Roche Inc, USA) according to the manufacturer's specifications. Briefly, adult animals/embryos were lysed in 0.5 ml TriPure reagent and incubated for 5 min at room temperature. Two-hundred μl of 1-bromo-3-chloropropane (BCP) was added to the tubes and centrifuged at 12,000 × g for 15 minutes. The aqueous phase was transferred to a fresh tube, treated with DNAse (Ambion, Life Technologies, Grand Island, NY, USA) and total RNA was pelleted by precipitation with isopropyl alcohol and centrifugation (12,000 × g for 10 min). The RNA pellet was washed with ethanol (75%) and pelleted at 7,500 × g for 5 min and air-dried at room temperature. RNA pellets were reconstituted in RNase-free water. RNA was quantified spectrophotometrically at 260 nm with 260/280 ratios between 1.8 and 2.0. RNA quality was also checked by 1.4% agarose gel electrophoresis stained with 5 μg ml^-1 ethidium bromide. Complementary DNA (cDNA) was synthesized with the Advantage RT-for-PCR Kit protocol (Clontech Laboratories, Mountain View, CA, USA) following the supplier's instructions. cDNA was stored in water at -20°C.

Primers for quantitative PCR were designed for genes of interest to amplify gene fragments of length 75 to 150 bp [see Additional file 1] using MacVector (MacVector, Inc, North Carolina USA). The genes studied include presumptive circadian clock genes previously reported by Reitzel et al. 35 , as well as identified genes likely involved in the melatonin synthesis pathway 39 . Ribosomal protein P0 [XM_001626244.1] was used as a normalization gene for all experiments. Previous qPCR assays showed that this gene does not have variation in transcription, either between different time points within a treatment or between different treatments. The cycling parameters used were 10 seconds at 95°C, 20 seconds at specific annealing temperature for each primer [see Additional file 1], and 20 seconds at 72°C during 45 cycles. qPCR was conducted on a LightCycler™ 480 Real-Time PCR System (Roche, Inc.) using SYBR Green I Master mix (Roche, Inc.).

Tryptophan hydroxylase (TPH) activity was quantified as previously described 41 . Briefly, each tissue sample was sonicated in sodium phosphate buffer (2 mM, pH 7, 100 μL), and the following mixture was added to each sample: HEPES (50 mM, pH 7), catalase (100 μg ml^-1), tryptophan (50 μM), dithiothreitol (5 mM), Fe (NH₄)² (SO₄)² (10 μM), 6-MPH4 (500 μM) and 1 μL of [3^H] tryptophan (1 mCi ml^-1 - previously dried under nitrogen). The material was incubated at 37°C for 10 minutes. An activated charcoal solution was added (7.5% in 1 M HCl) to terminate the reaction. Two-hundred μL of the supernatant was transferred to scintillation vials, and radioactivity was determined by a Beckman LS6500 β counter (Beckman Coulter Inc., CA, USA). Positive (rat pineal glands) and negative (water) controls were included in each assay.

Hydroxyindol-O-methyltransferase (HIOMT) activity was assayed as previously described 22 . Samples were sonicated in phosphate buffer (0.05 M, pH 7.9, 50 μL). One hundred fifty μL of a solution containing ¹⁴C-S-adenosyl-L-methionine (specific activity 43.8 mCi - Sigma Chemical Co., St. Louis, MO, USA) and N-acetylserotonin (1 mM) was then added. The homogenates were incubated for 30 minutes at 37°C. The reaction was terminated by adding 200 μL of sodium borate buffer (12.5 mM, pH 10) and 1 mL of buffer saturated chloroform. The tubes were centrifuged at 13,000 x g for 5 minutes at 4°C. The ¹⁴[C] melatonin product was extracted in 800 μL of chloroform, air-dried, and the radioactivity was determined by a Beckman LS6500 β counter (Beckman Coulter Inc., CA, USA). Positive (rat pineal glands) and negative (water) were included in each assay. To normalize the TPH and HIOMT assays, 2 μl of each sample were used quantify protein concentration to the TPH and to the HIOMT experiments. The results are presented as the ratio of activity per milligram of protein.

Melatonin was successfully detected in adults and embryos using the ELISA kit. High performance liquid chromatography (HPLC) with electrochemical detection assay confirmed the presence of melatonin, with the same retention time for the indol in the standards and in all samples [see Additional file 2]. For the adults in the light–dark treatment, we observed peak expression at 4 hours before the transition to the dark phase (that is, ZT = 8, Figure 1a, light-gray line) that was sustained until the ZT = 20 (4 hours before the transition to subjective day phase). In order to evaluate if these oscillations in melatonin production were a response to the presence or absence of light, we repeated the experiment using animals that were kept in constant darkness for 20 days. After 20 days of constant darkness, melatonin still showed significantly higher concentration in the transition from the subjective day to the subjective night (ZT = 12, Figure 1a, black line, one-way ANOVA, P = 0.0124). However, the night peak of melatonin was not sustained during the subjective night where we measured significantly lower concentrations of melatonin in ZTs = 16 and 20. This observation is confirmed by rhythmic analysis using the cosinor procedures, which showed that melatonin levels cycled in the light–dark cycle, but not in constant darkness (Figure 1b).

For N. vectensis embryos, we measured a peak expression of melatonin 6 hours after lights on (ZT = 6) in 48-hour embryos (Figure 1c) as well as juvenile polyp stages (1-week post-fertilization, Figure 1d; 2-weeks post-fertilization, Figure 1e). The peak amount of melatonin was similar in the 48-hour (14.11 ± 1.04 pg mg^-1 f protein) and 1-week juveniles (14.04 ± 0.88 pg mg^-1 of protein) but was almost the half of the amount observed in the 2-week juveniles (25.41 ± 0.54 pg mg^-1 of protein). The lowest melatonin concentration was observed in the middle of the dark period (ZT = 18) for all the groups (4.34 ± 0.86 pg mg^-1 of protein for the 48-hour; 5.81 ± 0.62 pg mg^-1 of protein for 1-week and 8.64 ± 0.701 pg mg^-1 of protein for 2-week juveniles).

To better understand the dynamics of the melatonin production in the embryos of N. vectensis, we also made serotonin measures on the same developmental time points. Serotonin was successfully detected in all samples. The pattern of production replicates the one showed in the melatonin assays. Again, there was a peak expression 6 hours after lights on (ZT = 6) in the 48-hour embryos (Figure 2a), 1-week (Figure 2b) and 2-week-old polyps (Figure 2c). The lowest serotonin peak was observed in the 48-hour embryos (14.73 ± 0.91 pg mg^-1 of protein). In the 1-week juveniles, the peak was 61.43 ± 1.02 pg mg^-1 of protein, which was higher than the 2-week juveniles (53.25 ± 1.30 pg mg^-1 of protein). This difference between the juvenile stages potentially indicates a higher amount of serotonin is being converted in melatonin in these older juveniles.

In order to characterize the melatonin production in N. vectensis, we identified and measured the expression pattern of transcripts coding for enzymes in the classic pathway of melatonin production. By searching the N. vectensis gene models and EST library (Joint Genome Institute Nematostella genome browser, http://genome.jgi-psf.org/Nemve1/Nemve1.home.html), we identified candidates for the first and the last enzymes in the classical melatonin pathway, tryptophan hydroxylase [XM_001623795.1] and hydroxyindol-O-methyltransferase [XM_001627179.1]. HIOMT has been previously described in N. vectensis 39 . Both genes were amplified, cloned and sequenced, with the same nucleotide sequence present in the EST library [see Additional file 3 and Additional file 4]. The sequences had high similarity to sequences previously described in other species. Phylogenetic analyses show that the candidate N. vectensis TPH gene groups with PaH (phenylalanine hydroxylases), the sister group to TPH, with no clear TPH ortholog in the genome [Additional file 5]. Because the N. vectensis TPH-like gene groups with PaH, we refer to it as TPH/PaH, indicating its co-orthologous relationship to both TPH and PaH from bilaterians. N. vectensis HIOMT groups with previously reported HIOMT proteins among various animal species. The phylogenetic analysis of the N. vectensis HIOMT gene clearly supports orthology with other animal HIOMT genes to the exclusion of non-animal genes (o-methyltransferases). Human and zebrafish HIOMT-like genes are supported as more recent duplication events in the deuterostome or chordate lineage [see Additional file 6].

NvTPH/PaH showed little change in expression in adult animals exposed to light–dark conditions. There was a moderate, but measurable, increase in transcription before the transition from the light to the dark period (Figure 3a, light-gray line), which correlates with the pattern observed in the melatonin production. There was significant variation in expression of TPH/PaH in the animals in constant darkness in a pattern with two peaks that does not resemble the one observed for melatonin (Figure 3a, black line). The Pearson correlation test between the TPH/PaH expression and melatonin content was positive for the animals in the light–dark cycle (r = 0.871, P = 0.0053). There was no correlation for the animals in constant darkness (r value = 0.286, P = 0.267). Expression of TPH/PaH in the 2-week embryos showed a pattern that again correlated with the serotonin and melatonin production profile (Figure 3c). The Pearson correlation test between the TPH/PaH expression and serotonin content was positive for the embryos (r = 0.926, P = 0.0044). A similar result with lower expression was observed for the 48-hour embryos and 1-week juveniles (data not shown).

NvHIOMT had increased expression at the beginning of the light period and had significantly lower expression during the dark period for adult animals (Figure 3b, light-gray line). This pattern correlated with the pattern of melatonin production observed, with a 4-hour advance of the HIOMT gene expression relative to the melatonin peak determined with ELISA (Pearson correlation test r value = 0.696, P = 0.042). For animals in constant darkness, HIOMT expression was near constant during the 24-hour period (Figure 3b, black line). Expression of the HIOMT gene in the 2-week-old juveniles also was similar to that pattern of serotonin and melatonin production (Figure 3d), with a peak 6 hours after lights on (ZT = 6). The Pearson correlation test between the HIOMT expression and melatonin content was positive for the embryos (r = 0.931, P = 0.0108). A similar result was observed for the 48-hour embryos and 1-week juvenile (data not shown).

To further validate the serotonin and melatonin assays, we performed enzymatic assays for tryptophan hydroxylase and hydroxyindol-O-methyltransferase for adults from ZT = 8 and ZT = 20. These time periods showed activity for each enzyme at both time points (Table 1). Highest activity was measured at the ZT = 8, the same point of the melatonin peak. Enzymatic assays were also performed with samples from embryos from ZT = 6. The measured activity of tryptophan hydroxylase was 38.94 ± 7.32 picomoles per 200 mg of protein per hour (n = 4) and 61.51 ± 0.74 picomoles per 200 mg of protein per hour (n = 4) for hydroxyindol-O-methyltransferase.

Values are mean ± SEM for four samples per assay. Activity expressed in picomoles per 200 mg of protein per hour. ZT20 values are statistically different from ZT = 8; P <0.05.

To characterize the potential role of N. vectensis clock genes in embryonic development, we examined the spatial and temporal expression dynamics of those genes by whole mount in situ hybridization. Figure 4 shows the expression patterns of NvClock, NvTimeout, NvCry1a and NvCry1b (NvCycle reported in 43 ). Diffuse expression of NvClock begins at blastula stage (Figure 4a). Expression is concentrated in the oral pole during late blastula (Figure 4b) and then remains restricted to the endoderm during the gastrula (Figure 4c) and late planula (Figure 4d-e) stages of development. Expression appeared to be uniform in all cells of the endoderm and did not show the ‘salt and pepper’ pattern characteristic of cells in the endodermal nervous system 44 .

NvTimeout showed a pattern similar to NvClock, weak expression at the blastula stage (Figure 4f) with expression intensifying at the oral pole during invagination (Figure 4g). After this stage, expression is then restricted to the endoderm, during early (Figure 4h) and late planula (Figure 4i-j).

The cryptochrome genes NvCry1a and NvCry1b show similar patterns. The expression begins at the blastula stage diffusely in future ectoderm (Figure 4k,p). During invagination of prospective endoderm at gastrulation, expression begins to be expand to the endoderm for NvCry1a (Figure 4l), and over time restricts its expression to become more restricted to the endoderm in later stages (Figure 4m-o). A similar pattern of expression is observed for NvCry1b; predominantly ectodermal during gastrulation (Figure 4q) but switching to endoderm during later gastrula stages (Figure 4r). In the planula stages, the expression is more concentrated in the endoderm, but faint expression was evident in ectodermal tissues (Figure 4s-t).

We performed qPCR assays on the clock genes to investigate the generation of the rhythmicity during N. vectensis embryogenesis. NvClock, NvTimeout, NvCry1a and NvCry1b, at six hour time points over a 24-hour period (that is, 6, 12, 18, and 24-hours post-fertilization (blastula)), 48-hours (late blastula), 1-week (late planula) and 2-weeks (early polyp with tentacle buds) post-fertilization.

Up to the 48-hours post-fertilization, there was no clear rhythm of expression of NvClock (Figure 5a). One-way ANOVA test showed no difference in expression levels between any of the time points (P = 0.4179). At 1-week post-fertilization, there was evidence of a rhythm (Figure 5b), with a peak in the transition light–dark (ZT = 12), that was confirmed by the one-way ANOVA (P <0.0001). However, the cosinor test was not valid, indicating that even that there was variation between time points that did not follow a 24-hour period. In the 2-week polyps, however, there was a clear circadian pattern of expression, with variation confirmed by the one-way ANOVA test (P <0.0001), and the cosinor test was valid, (P = 0.048), indicating that this gene had acquired a circadian pattern of expression (Figure 5c).

For the NvTimeout, again there was no rhythmicity in expression in 48-hour embryos (Figure 5d), confirmed by the one-way ANOVA (P = 0.7247). In the 1-week polyps, however, it was evidence of rhythmic expression (Figure 5e). The one-way ANOVA showed that expression was significantly different (P = 0.0008) and the cosinor test was valid (P = 0.045), revealing that the gene already had a circadian pattern of expression (graph not shown). Rhythmic expression was maintained in the 2-week polyps (Figure 5f), where the cosinor model was also valid (P = 0.025).

The gene NvCry1a had a pattern similar to NvClock: no rhythm of expression in the 48-hour embryos (Figure 5g, P = 0.3745), but with a difference between the time points in the 1-week embryos (P <0.0001), without validation from the cosinor (Figure 5h). At 2 weeks, the polyps had a circadian pattern of expression of the gene (Figure 5i), with validation of the one-way ANOVA (P <0.0001) and from the cosinor (P = 0.003). The pattern for NvCry1bwas similar to NvTimeout. In 48-hour embryos, there was no rhythmicity in expression (Figure 5j), which was confirmed by the one-way ANOVA, which showed no difference in expression between time points (P = 0.928). In the 1-week polyps, however, we detected a rhythm (Figure 5k) with significant differences in expression over time (P = 0.0004) and a significant cosinor result (P = 0.0072, graph not shown). This circadian pattern of expression was continued in the 2-week-old polyps (Figure 5l), where again the one-way ANOVA showed difference between the points (P <0.0001) and the cosinor model was valid (P = 0.0052).

We analyzed transcription of clock genes in adults to determine how their expression was altered by removal of a light cycle and investigate if the addition of exogenous melatonin alters the expression of these genes in the absence of a light cue. Previous work showed that 30 days of constant darkness appears to result in the loss of rhythmicity in the same clock genes when assayed with qPCR 35 .

First, in order to investigate how rapidly the N. vectensis clock genes lost their rhythmicity in adults removed from the entraining light cue, we moved animals entrained in the normal 12:12 light–dark cycle to constant darkness and sampled individuals at 24, 48 and 96-hours. NvClock, lost rhythm in the first 24 hours of constant darkness, with no difference in the points along the period (Figure 6a, one-way ANOVA, P = 0.1880). NvTimeout, NvCry1a and NvCry2 maintained rhythmic expression for longer than NvClock, but by 96 hours all showed expression profiles lacking oscillations (one-way ANOVA, P = 0.2532 for NvTimeout, 0.1081 for NvCry1a and 0.4186 for NvCry2, Figure 6b,c and d). NvCry1b lost evidence of rhythmic expression at 48 hours (Figure 6e, one-way ANOVA, P = 0.6111).

Next, we measured expression of these genes in N. vectensis adults cultured in darkness for 20 days (approximately 16 days after loss of rhythmic gene expression) and supplemented with melatonin at ZT = 12. Melatonin treatment did not alter the expression pattern of NvClock; gene expression lacked oscillations in expression similar to animals in constant darkness, which contrasts with increased expression of NvClock during light periods when animals are cultured on a diel cycle (Figure 7a) 38 . The cosinor test showed that the variation in the expression of NvClock had a 24-hour period only in animals in a normal light–dark cycle (Figure 8a).

The result observed for the gene NvCycle was interesting because altering the light cycle did not alter substantially the pattern of expression of the gene (Figure 7b). The time points were significantly different both between and within treatments (one-way ANOVA, P <0.005), which enabled us to do a cosinor test that was valid for all groups (Figure 8b).

NvTimeout showed a significantly different pattern of expression in the animals throughout the light–dark cycle, with a peak 3 hours after the transition to the dark period (Figure 7c). The one-way ANOVA test showed that the expression levels were different during the 24-hour period (P = 0.0006) for this group and for the melatonin-treated animals (P <0.0001) but not significantly different for the animals in constant darkness (P = 0.7985). The cosinor test showed that the gene had a circadian pattern of expression, but with a difference in the acrophase of almost 13 hours between the light–dark and the melatonin-treated animals (Figure 8c, Table 2).

‘LD’ refers to animals kept in a light–dark cycle. ‘Melatonin’ refers to animals kept in a constant darkness cycle with melatonin (0.1 μM) supplementation. Values are mean ± SEM for five samples per assay. Bold indicates significant difference.

The two Type I and single Type II cryptochromes from N. vectensis showed differential responses NvCry1a showed a similar expression pattern as NvTimeout (Figure 7d) with the cosinor model showing that the light–dark and the melatonin-treated animals showed significant 24-h rhythms that were lost in the constant darkness group (Figure 8d). NvCry1b showed circadian expression for the light–dark and melatonin-treated animals, but in phase opposition to one another (Figure 7e). The cosinor analysis showed rhythmic expression for both those groups, showing again the ability of melatonin to alter the pattern of expression of genes likely associated with the anemone circadian clock (Figure 8e). The pattern of the gene NvCry2 was very similar to the one of NvCry1b, with phase opposition between the light–dark and the melatonin-treated animals and an expression pattern without oscillations for those in constant darkness (Figure 7f, 8f).

Previous research has shown variation in the seasonal and diel synthesis of melatonin in anthozoan cnidarians. Melatonin was first reported in the colonial anthozoan Renilla köllikeri 37 , where melatonin amounts fluctuated by season but not diel cycles. Melatonin was also at highest concentration in gametogenic regions. This work supported a role for melatonin in cnidarian seasonal reproduction but not circadian processes, which differs from bilaterians 8 14 . Subsequent studies in two other anthozoan species (Actinia equina and Nematostella vectensis) 38 39 showed that melatonin was also localized near or in gametogenic tissues. In addition, Roopin and Levy 38 showed that melatonin levels in A. equina vary on a diel cycle with peak concentration in early subjective night, but the oscillating expression dissipates upon exposure to constant darkness. These later results suggest that melatonin may serve a role in the circadian clock because cnidarians lose oscillations in gene expression when light cues are removed 35 43 45 . We show that melatonin concentration, as well as the genes in the melatonin synthesis pathway, have oscillating patterns in N. vectensis. The presence of nocturnal peaks of melatonin production in N. vectensis is similar to peak production of melatonin in diverse species. Nocturnal peaks are present in all the vertebrates investigated thus far 46 , as well as species of algae 8 47 , flatworms 10 and insects, including Drosophila 48 . Combined, these results suggest that melatonin has diel patterns in anthozoans, which are likely related to the circadian clock (see below).

We determined that TPH/PaH and HIOMT, central components of the melatonin pathway, are present and expressed in N. vectensis. Both enzymes have already been described in different groups of invertebrates, including flatworms 49 , sea urchins 50 and cnidarians 51 and the activity of the enzymes in mollusks 52 and arthropods 53 54 55 . We found a correlation between the rhythm of the transcription of TPH/PaH and production of serotonin (product of TPH in vertebrates) and HIOMT and production of melatonin (product of HIOMT), which provides consistent evidence that these enzymes participate in the melatonin pathway. The fact that we were also able to measure the activity of the enzymes gives addition support for the presence of a vertebrate-like melatonin pathway in cnidarians.

Our work presented here is the first to show melatonin synthesis in a cnidarian during invertebrate embryogenesis. There are few references in the literature of melatonin in embryos, showing its presence in fishes 29 , birds 30 and mammals 56 . The measurement of a rhythm in the melatonin production during development indicates a regulated production of the hormone in response to light environment, instead of maternal deposition into maturing oocytes. In addition, because developmental stages lack gametogenic tissue, the role for melatonin in cnidarians includes non-reproductive functions in addition to the hypothesized role in gametogenesis reported in previous studies 37 38 39 . The melatonin pattern of production in the embryos resembles the adults, with peaks during the second half of the light phase. These data suggest that a circadian pattern is initiated in N. vectensis prior to the adult stage. However, the melatonin content has higher variation between these life cycle stages, with a peak value close to 26 picograms per milligram of protein in the 2-week-old embryos and around 6 picograms per milligram in adults. This difference may indicate either a higher production in the embryos or a higher degradation of melatonin in the adults.

The analysis of the genes likely involved in the circadian clock in the adults showed that 4 days of constant darkness were adequate to abolish the circadian expression for virtually all of them. Our data are nearly identical to the pattern previously observed, although that study kept N. vectensis for 30 days in constant darkness 35 ; however, our results show that quenching of cyclical gene expression occurs within days of removal of light cues. Data from corals have similarly shown loss of the rhythmicity of some clock genes within 24 hours (Acropora millepora 57 ) or 72 hours (Favia fragum 45 ). Together, these data suggest that loss of rhythmic gene expression is characteristic of cnidarian clocks, in opposition to the classical description of the bilaterian clock, which is capable of maintaining rhythmicity even after several days in constant darkness 58 59 60 .

We tested whether melatonin could re-synchronize the clock genes in individuals cultured in total darkness because work in mammals has shown that melatonin can serve as a potent molecule for circadian clock resetting 61 62 . Previous studies in mammals showed that a short melatonin pulse (2 hours) is not enough to alter the expression of clock-related genes 63 . We observed that the addition of melatonin in 12-hour pulses results in oscillations in expression of N. vectensis circadian clock genes, supporting a hypothesis that the interaction of melatonin and clock genes appeared early in the evolution of the animals.

The differences in the timing of peak melatonin expression observed for clock genes between animals in the light:dark cultures and constant darkness with melatonin supplementation are likely due to the timing of melatonin introduction. Animals in a light–dark cycle showed increasing melatonin beginning at ZT = 8 that was sustained into subjective night. We added melatonin to cultures of animals in constant darkness at the subjective transition to the dark phase (that is, ZT = 12), which could explain the temporal shift in gene expression. In vivo melatonin is produced in a progressive way, with low concentrations in the beginning of the day, which increase during subjective day and peak during the night phase of the light:dark cycle. In our treatment, we added melatonin at the beginning of subjective night (ZT = 12) to coincide with the highest concentration of melatonin. Future experiments that manipulate the timing of melatonin supplementation would provide an additional empirical test for how melatonin additions may shift transcription of genes involved in the circadian clock and how the timing of addition relates to the endogenous melatonin levels.

Considering the potential interplay between melatonin and clock genes, it is difficult to postulate whether the changes we observed in the melatonin production in the animals in constant darkness are a product of the disruption of the rhythmicity of the clock genes, or vice versa. Manipulations to block the melatonin production with morpholinos to TPH/PaH or HIOMT, or to knockdown the expression of the clock genes themselves, could be useful to better understand the interaction between the two factors. These studies would test for conserved functions for the role of hormones in the transcription-translation feedback loop of the cnidarian circadian clock.

Quantitative and spatial analysis of clock gene expression during development indicates that their expression begins early in the development that precedes oscillating expression in response to environmental cues, similar to zebrafish. In situ expression analysis of these circadian genes showed that all genes are expressed broadly in the endoderm post-gastrulation. The two Type I cryptochrome paralogs Cry1a and Cry1b were expressed prior to gastrulation with general expression in the blastula. The lack of specific expression in any region of the endoderm precludes any clear hypotheses regarding potential developmental functions of these genes. Previous work describing expression of circadian clock genes in the development of zebrafish has suggested that these genes may function in cell cycle control prior to a role in circadian clock function in later stages. A similar function may be present in N. vectensis.

Two genes (Timeout and Cry1b) are transcribed in a circadian fashion in the 1-week-old embryos, and at two weeks, all the clock genes we evaluated were being expressed in a circadian rhythm, with a similar pattern to that observed in the adults. These results suggest that the circadian clock may not begin oscillating until the polyp stage. Interestingly, we observed melatonin concentrations fluctuated during embryogenesis, despite the lack of oscillating expression of the circadian clock. The difference in timing of oscillatory behavior of melatonin and circadian clock may indicate a non-circadian role for melatonin in development, potentially in responding to oxidative stress. Future research measuring behavioral entrainment of larval stages will be necessary to determine in N. vectensis larvae have diel activity patterns like those described in adult polyps 75 .