Light Entrained Rhythmic Gene Expression in the Sea Anemone Nematostella vectensis: The Evolution of the Animal Circadian Clock

Most organisms exhibit daily physiological and behavioral rhythms that are regulated by molecular circadian clocks. Although light is the most common signal entraining these rhythms, other environmental signals such as food availability and temperature can also drive oscillations in gene expression through the same molecular clock [1], [2]. Characterization of the genes composing clocks in diverse organisms (e.g., bacteria [3], plants [4], fungi [5], animals [6]) has suggested that the circadian clock has evolved independently many times [7].

In animals, the molecular components of the circadian clock have been exquisitely characterized in mammals and diverse insects [8], [9]. From these studies, it is clear that many of the core clock genes are conserved in these two disparate animal groups [10], suggesting that this molecular clock dates back to at least the ancestor of deuterostomes and protostomes. Both mammalian and insect clocks are based on two interacting molecular feedback loops: a positive loop driven by expression of the bHLH-PAS transcription factors Clock and Cycle (Bmal1/Mop3 in mammals) that then heterodimerize to regulate downstream gene expression through E-box motifs, and a negative loop where dimerization of PERIOD and cryptochrome paralogs (mammals) or PERIOD and TIMELESS (Drosophila) inhibit the CLOCK:CYCLE heterodimer [10]. Despite broad similarities, studies of additional insect species have revealed a number of variations of the negative loop involving cryptochromes and Timeless[11], [12]. The divergence of molecular components and their function between mammals and insects and within insects can be explained, in part, by gene duplication and loss [12], [13].

Several studies comparing gene families between bilaterians and cnidarians have suggested that their most recent common ancestor had a complex genome [14], [15]. Most of these studies have focused on transcription factors involved in embryogenesis and morphogenesis, leaving unknown to what extent the molecular components that regulate physiology and behavior are conserved. Because the core genes that regulate circadian cycling are largely conserved within the Bilateria, we hypothesized that the circadian machinery would operate similarly in cnidarians. Understanding the mechanics of the cnidarian clock will reveal which genes likely composed the clock of the ancestor to the Cnidaria and Bilateria, which was certainly a marine species living in an environment dominated by blue light [16].

Cnidarian physiology and behavior strongly vary on diel cycles. In some cnidarians, skeletal growth and energetic metabolism are strongly coupled with photosynthesis by algal endosymbionts [17]. However, other behaviors, such as spawning, tentacle extension and feeding, may also be dependent on regular cycling of light:dark periods and variations in subjective day length [18], [19]. In the only study of circadian-like gene expression in a cnidarian, Levy et al. [20] identified two cryptochromes from the coral Acropora millepora that displayed rhythmic gene expression in response to a light:dark treatment. Notably, one of these cryptochromes was upregulated during full moon, prior to spawning. Indeed, corals are extremely sensitive to low levels of blue light, consistent with detection of lunar irradiation [21]. Together, the behavioral and molecular studies suggest that cnidarians may share conserved molecular elements in the circadian clock with insects and mammals.

In this study, we have identified key components of the circadian clock in a model cnidarian, the starlet sea anemone, Nematostella vectensis. For this work, N. vectensis has several advantages including a sequenced genome, ease of laboratory culture, a simple, transparent body column, and absence of algal symbionts, which allows for more direct interpretation of cnidarian-specific responses [14], [18], [22]. Previous research has shown that gametogenesis by N. vectensis can be strongly influenced by light:dark cycling [23] and that N. vectensis contains likely homologs to the bilaterian clockwork [24]. Through bioinformatic searches and phylogenetic analyses we identified conserved circadian genes from N. vectensis. We then quantified the expression of circadian genes in response to light:dark cycles as well as in response to different portions of the light spectrum. We also tested for conserved protein-protein interactions between N. vectensis' orthologs to Clock and Cycle and identified canonical E-box motifs in the promoters of NvClock and one cryptochrome, two genes that showed rhythmic gene expression. Together, our data show that two components of the animal circadian clock, cyclic expression of orthologous circadian genes and heterodimerization of key bHLH-PAS proteins, were in place at the divergence of the Cnidaria and Bilateria.

Through bioinformatic queries and phylogenetic analyses we identified a number of Nematostella vectensis orthologs to core circadian clock genes shared by vertebrates and insects. From the bHLH-PAS family, N. vectensis has strongly supported orthologs to Clock and Cycle[as previously reported in 25], as well as a number of other bHLH-PAS transcription factors (Figure 1A). There are no orthologs to Period in N. vectensis or the coral A. millepora, despite an earlier report suggesting these two cnidarians may have these genes [[24], see Figure S1]. Indeed, Period genes have not been identified in any non-bilaterian phylum [25], suggesting this subfamily of transcription factors evolved within the Bilateria. We identified a single gene with high similarity to Timeout from human and Timeless/Timeout from diverse insects (Figure 2a). We also identified three cryptochromes, two of which are orthologs to previously reported genes from the coral A. millepora (Figure 3a, [20]). Additionally, we previously showed that N. vectensis lacks orthologs to the nuclear receptors Rev-Erb (NR1D1) and ROR (NR1F1) [26], important components of the mammalian clock.

Clock and Cycle proteins form the core of the positive regulatory loop of the circadian clock in both mammals and insects. However, the expression of these transcription factors during a light cycle differs between these taxa [reviewed in 10]. NvClock showed strong diurnal expression with peak expression in the mid- to late subjective day (Figure 1B, zeitgeber time (ZT) = 6–12). The magnitude and timing of peak expression were significantly influenced by the light quality, with higher expression and a slightly shifted peak in the two light treatments with a high proportion of blue wavelengths. NvCycle expression showed small differences between the full spectrum and dark treatments over the sampled time points (Figure 1C). Together, the pattern of rhythmic expression for these two bHLH-PAS transcripts was more similar to Drosophila, where DmClock exhibits peak expression in the light portion of a light:dark cycle, albeit at an earlier portion of this period (ZT = 0–4) [27], [28]. In contrast, Clock expression in mouse does not vary during a light:dark rhythm [29].

A shared protein-level interaction of mammalian and insect circadian clocks is the heterodimerization of CLOCK and CYCLE that together act as transcriptional regulators. In mammals, the CLOCK:CYCLE heterodimer upregulates expression of Period paralogs, among other genes. In Drosophila, as well as some other insects, the CLOCK:CYCLE dimer upregulates Period and Timeless, as well as other transcriptional targets. For N. vectensis, co-immunoprecipitation of NvCYCLE and NvCLOCK revealed that these two genes specifically dimerize in vitro (Figure 4). No non-specific dimerization was observed between NvCLOCK and NvARNT, a bHLH-PAS transcription factor closely related to Cycle (see Figure 1A). These data support the hypothesis that the dimerization of CLOCK and CYCLE dates back to at least the cnidarian-bilaterian ancestor and is likely an ancient component of the animal circadian clock.

The Timeless-Timeout family contains two paralogs in Drosophila and other insects but only one gene in diverse vertebrates, invertebrates, and non-animal species [30], [31]. Members of the Timeless-Timeout family in vertebrates are orthologous to Timeout from insects but have been frequently referred to as “Timeless”, likely due to their description prior to the identification of the paralogs in Drosophila. Early descriptions of the taxonomic distribution of genes in the Timeless-Timeout family suggested an insect-specific gene duplication [31]. However, Rubin et al. [30] identified both Timeless and Timeout in the sea urchin Stronglyocentrotus purpuratus indicating that this gene duplication event occurred prior to the divergence of protostomes and deuterostomes. We recovered Timeless and Timeout orthologs from additional protostomes (Figure 2A), supporting a hypothesis these paralogs are present broadly in bilaterian animals, with Timeless independently lost in diverse lineages (e.g., chordates, nematodes). We recovered only one gene from each surveyed cnidarian and placozoan, further supporting a conclusion that the gene duplication event occurred after the cnidarian-bilaterian divergence but before the protostome-deuterostome split. The genes from these early diverging phyla, including the gene from N. vectensis, shared greater sequence similarity to Timeout and grouped with these genes in the phylogenetic analysis (Figure 2A).

NvTimeout expression did not vary consistently in response to light treatment (Figure 2B). The role of Timeless in circadian gene regulation was originally reported in Drosophila, where expression follows a 24-hr cycle [32] through interaction with the CLOCK:CYCLE heterodimers [28], [32]. Timeless is part of the circadian clock for some other insects besides Drosophila[33], [34], but other insects lack Timeless completely [30]. In addition, Timeout has recently been shown to play dual roles in Drosophila as part of the circadian clock and in chromosome stability [35]. Despite conflicting data showing an inconclusive role for Timeout in the mammalian circadian clock [36], recent work has provided evidence that Timeout indeed plays a critical role in regulating the mouse clock [37]. Similar to Drosophila, mammalian Timeout has a second function independent of the circadian clock as one component of the DNA replication fork complex [38]. Because Timeout and its binding partner Tipin are well conserved in both animals and fungi, the ancestral role of these proteins may have been in regulation of DNA replication rather than circadian cycling [38].

The role, if any, for Timeout from N. vectensis in the circadian clock is presently uncertain. NvTimeout, like vertebrate and insect Timeout genes, has poorly conserved first Period-interacting and cytoplasmic localization domains, and lacks a conserved nuclear localization signal (Figure S2). In addition, NvTimeout lacks the DEDD portion of the cytoplasmic localization domain, which is conserved in vertebrates and insects. Through bioinformatic searches, we identified a homolog of Tipin from N. vectensis (XP_001625500↗), suggesting that NvTIMEOUT may interact with TIPIN to regulate DNA replication. In addition, because N. vectensis and other early diverging animals lack Period genes, we have no hypothesis for a binding partner for TIMEOUT in a potential repressor loop of the circadian clock. Future studies testing whether NvTIMEOUT is degraded in a light-dependent manner, as shown in Drosophila[39], would be particularly informative.

We identified three cryptochromes from N. vectensis. Two are supported as a cnidarian-specific duplication of an ancestral gene that groups between Type 1 cryptochromes and the DNA repair enzyme 6-4 photolyase, and the third groups with vertebrate and insect Type 2 cryptochromes (Figure 3A). In N. vectensis, expression of these three cryptochromes varied in response to the light treatments and time in the light cycle. NvCry1a expression significantly increased in response to all three light treatments with the highest expression in those treatments with blue wavelengths (Figure 3B, i.e., full spectrum and blue-only). Transcription increased immediately after the light period started, was sustained while light was present, and decreased rapidly once light was removed. Anemones cultured in continuous darkness showed no cycling, suggesting that the response was a result of exposure to light, not an unmeasured variable or an endogenous signal. NvCry1b, an ortholog of AmCry2 previously reported from the coral A. millepora, had significant diel cycling only in the blue-light only treatment (Figure 3C). Transcription increased midway through the light period and then decreased once light was removed. In A. millepora, this cryptochrome showed a similar rhythmic transcriptional response. Interestingly, AmCry2 was shown to be responsive to moonlight, which is particularly enriched with blue light. Taken together, these data suggest that this cnidarian-specific cryptochrome may have evolved sensitivity to a narrow band of the light spectrum early in the anthozoan lineage.

The third N. vectensis cryptochrome, NvCry2, showed no strong response to diurnal light cycling for any light treatment (Figure 3D). These results were of interest for two reasons. First, NvCry2 expression in a light:dark cycle was strikingly different from that of the closely related cryptochrome AmCry1 from A. millepora. The coral cryptochrome had strong diurnal expression that peaked during the daylight, similar to AmCry2. Thus, the coral and anemone orthologs, despite similar evolutionary history, have strongly divergent expression in response to light treatment. Secondly, NvCry2 and AmCry1 are most similar to previously characterized vertebrate and insect Type 2 cryptochromes. Vertebrate and insect Type 2 cryptochromes are derived from 6-4 photolyase and act as repressors in insects and mammals [12], [34], [40], [41]. In mammals, cryptochrome expression is offset from Cycle[41]. Unlike the mammalian pattern, NvCry2 expression was not offset from the putative positive regulators NvCycle and NvClock. NvCry2 expression was more similar to the pattern observed in some insects (e.g., monarch and mosquito), in which expression of these insect Cry2 genes does not significantly vary over a light:dark period [12], [42]. However, these insect proteins still act as a critical part of the negative circadian regulatory loop. It is plausible that NvCry2 could similarly act as a repressor (discussed below).

We searched N. vectensis promoter sequences for evidence of conserved motifs known to regulate circadian signaling. In mammals and insects, CLOCK:CYCLE heterodimers regulate transcription of downstream genes by binding to E-box motifs (CACGTG) [43], [44]. We searched for E-box motifs in the 2 kb upstream of the transcriptional start-sites of the six putative circadian genes we have identified in this study. We identified identical matches near the start site in promoters for two genes: one for NvClock (-225 bp) and two for NvCry1a (−228, −526 bp). These genes were the most strongly upregulated during subjective day. The only other gene in our survey that displayed rhythmic expression was NvCry1b, which has an E-box motif ∼1800 bp upstream of the start site. These data provide correlative evidence that the N. vectensis CLOCK:CYCLE heterodimer may regulate transcription of target genes through the same DNA motifs that have been identified in mammals and insects. Currently, there have been few studies assessing the conservation of bilaterian regulatory motifs in early diverging animals. One study has shown conserved specificity for DNA binding motifs for the transcription factor NF-κB from N. vectensis when compared with other animals [45]. Thus, it would not be unexpected that NvCLOCK:CYCLE heterodimers may regulate transcription of downstream genes through canonical E-box motifs.

One likely divergence in the N. vectensis clock from other animals is the composition of the negative regulatory loop. N. vectensis lacks Period genes, and NvTimout expression did not vary in response to a daily light cycle. The negative regulatory loop differs among mammals and insects, partially as a result of gene duplication and loss within each lineage (duplication of Period in mammals, loss of Timeless and CRY1 and 2 cryptochromes in some insects, see [12], [13]), so a novel loop in cnidarians may not be surprising. As discussed above, one likely component of the negative regulatory loop in N. vectensis is NvCry2. Cryptochromes play a critical portion of the negative regulatory loop in both vertebrates [40], [41] and nondrosophilid insects [12], [34], [42] by inhibiting CLOCK:CYCLE mediated transcription. Given that N. vectensis lacks Period genes, it is plausible that NvCry2 independently inhibits the positive loop, as has been shown in these other animals. The transcription of NvCry2 did not correlate with light exposure, similar to mCRY2 from mouse [41] and CRY2 from diverse insects [12], suggesting that this gene does not function as a photoreceptor. Future research studying whether NvCry2 protein suppresses CLOCK:CYCLE –mediated transcription would be particularly informative for understanding the potential role of this protein in the negative loop of the cnidarian circadian clock. Additionally, a repressive role by this anemone protein would further push back the date for the function of Type 2 cryptochromes in the animal circadian clock.

Our combined study of gene representation, rhythmic gene expression, and conserved protein-protein interactions suggest that the circadian clock from N. vectensis is similar to those reported from mammals and insects, and thus many molecular aspects of the circadian oscillator were present in the cnidarian-bilaterian ancestor. In addition, some molecular components of the circadian clock may be even more ancient. Orthologs to Clock and a presumptive ARNT/Cycle ancestral gene [25] as well as a cryptochrome [46] have been reported from sponges. Sponge behavior is also responsive to diel light cycles [e.g., 47]. Future work with sponges and other early diverging animals, including characterization of protein-protein interactions and DNA regulatory motifs, as we have reported here for N. vectensis, will further elucidate which components of the animal circadian clock were likely present at the origin of the animal kingdom.

We would like to thank Sibel Karchner (WHOI) for assistance with co-immunoprecipitation and Mark Hahn's group (WHOI) for discussions during the course of this project. We would also like to acknowledge technical assistance from Sam Laney and Dave Kulis (WHOI).