Circadian and circalunar clock interactions in a marine annelid.

Life is controlled by multiple rhythms. Although the interaction of the daily (circadian) clock with environmental stimuli, such as light, is well documented, its relationship to endogenous clocks with other periods is little understood. We establish that the marine worm Platynereis dumerilii possesses endogenous circadian and circalunar (monthly) clocks and characterize their interactions. The RNAs of likely core circadian oscillator genes localize to a distinct nucleus of the worm's forebrain. The worm's forebrain also harbors a circalunar clock entrained by nocturnal light. This monthly clock regulates maturation and persists even when circadian clock oscillations are disrupted by the inhibition of casein kinase 1δ/ε. Both circadian and circalunar clocks converge on the regulation of transcript levels. Furthermore, the circalunar clock changes the period and power of circadian behavior, although the period length of the daily transcriptional oscillations remains unaltered. We conclude that a second endogenous noncircadian clock can influence circadian clock function.

Introduction

Most, if not all, organisms feed periodic changes in light conditions into molecular clockworks that allow them to anticipate rhythmic changes in their environment and to synchronize their behavior and physiology (Cold Spring Harbor Symposia on Quantitative Biology, 2007; Roenneberg and Merrow, 2005). Efforts to study the underlying molecular mechanisms have focused almost exclusively on circadian clocks (i.e., clocks anticipating daily cycles). One of the critical mechanisms driving animal circadian clocks are transcriptional/translational feedback loops formed by a set of regulatory genes. These genes are partially shared between insect and mammalian models, arguing for a common origin of animal circadian clocks. The feedback loops continue to run under constant conditions and are coordinated with the animal’s environment by entrainment (Cold Spring Harbor Symposia on Quantitative Biology, 2007; Roenneberg and Merrow, 2005).

However, many organisms also exhibit rhythms of longer and shorter period lengths (Aschoff, 1981; Naylor, 2010). In order to maximize the chance of finding mature mates, to avoid predators, and to have favorable environmental conditions, organisms ranging from brown algae and corals to worms and vertebrates synchronize their maturation and spawning to a particular moon phase, to particular times of the day, and/or to specific seasons within a year (Fox, 1924; Harrison et al., 1984; Korringa, 1947; Tessmar-Raible et al., 2011). As with circadian rhythms, such noncircadian (e.g., annual and monthly) rhythms are often driven by internal oscillators (circannual and circalunar clocks, respectively), which use light cues (photoperiod and moonlight, respectively) for the adjustment with the outer environmental conditions (Dupré and Loudon, 2007; Franke, 1985; Lincoln et al., 2006; Naylor, 2010; Hazlerigg and Lincoln, 2011; Kaiser et al., 2011).

Numerous studies have assessed the influence of additional light cues on the molecules and function of the circadian clock. Photoperiod influences circadian clock gene oscillations in insects (Koštál, 2011) and the waveform of circadian oscillations in mice (Ciarleglio et al., 2011), resulting in activity differences between animals raised under different photoperiods. Likewise, dim nocturnal light at moonlight intensity has been shown to influence circadian clock gene expression levels and timing in Drosophila, resulting in elevated nocturnal activity under laboratory conditions (Bachleitner et al., 2007). However, no elevated nocturnal activity was observed in corresponding moon phases under natural conditions, and the level of Period was differently altered (Vanin et al., 2012). Lunar light influences the levels of the putative light receptor and/or core clock gene cryptochrome in the circalunar spawning coral Acropora and the rabbit fish Siganus guttatus (Fukushiro et al., 2011; Levy et al., 2007). In Siganus, moonlight has also been shown to elevate levels of the circadian clock gene per2 (Sugama et al., 2008). However, a key issue that has remained obscure is if and how circadian and noncircadian internal oscillators interact molecularly to influence the behavior of an organism, independent of illumination effects.

A suitable model system to assess this question has to be a molecularly accessible, extant animal that at the same time possesses circadian and noncircadian timing mechanisms. The bristle worm Platynereis dumerilii offers this dual advantage. Platynereis was among the first species for which a circalunar spawning rhythm was scientifically documented (Fage and Legendre, 1927; Ranzi, 1931a, 1931b). In addition, Platynereis has emerged as a highly suitable model for molecular neurobiology (Arendt et al., 2004; Backfisch et al., 2013; Tessmar-Raible et al., 2007; Tomer et al., 2010).

Here, we establish that Platynereis dumerilii possesses both a circadian and a circalunar clock. Whereas the circalunar-clock-controlled reproductive timing rhythms are insensitive to the functional disruption of circadian clock gene oscillations, the circalunar clock affects the circadian clock on at least two levels. First, the period length and power of circadian-clock-controlled locomotor behaviors are significantly different between different phases of the circalunar clock, while the period length of the presumptive core circadian clock molecular oscillations remains unaffected. Second, clock, period, pdp1, and timeless transcript levels oscillate in specific brain nuclei of the worm’s forebrain not only over 24 hr, but also across different phases of the lunar month. This establishes changes in RNA levels as a direct or indirect output of the circalunar clock.

Results

Platynereis Possesses a Light-Entrained Circalunar Clock

The circalunar reproductive periodicity of Platynereis dumerilii (Figures 1A and 1B) has been extensively documented (Figure S1). Reproductive state, as measured by the number of animals reaching sexual maturity, is maximal shortly after new moon (NM) and minimal during periods of full moon (FM) (Figure S1).

Figure 1

Circalunar Reproductive Periodicity of Platynereis dumerilii Is Entrained by Light and Controlled by a Clock Mechanism

(A) Premature adult (>2 months of age) as used in subsequent molecular and behavioral experiments is shown.

(B) Mature male and female as counted for the quantification of mature worms during mating dance are shown.

(C) Schematization of illumination conditions is shown. Daylight, yellow bars; nights without moon (new moon [NM]), black bars; nights with dim light simulating full moon (FM), light yellow bar. For “lunar” free-running experiments, the dim nocturnal light signal is omitted (FR-FM, free-running full moon; FR-NM, free running new moon). Illumination conditions used on x axis encode for 1; number of days, 2; day/night (in vertical direction).

(D and E) Light-entrained lab cultures exhibit maturation peaks comparable to nature (Figure S1). Nocturnal illumination in phase (D) and out of phase (E) with the natural moon is shown.

(F) Maturation synchronization continues for several months under circalunar free-running conditions after entrainment with dim nocturnal light (see C); dashed line indicates decreasing amplitude.

(G) Fourier analysis of free-running full moon spawning data shown in (F) reveals a 30-day period length, corresponding to the length of one lunar month.

(H) Worms grown under constant light (same light intensity during day/night) or without nocturnal illumination show no synchronization in maturation.

Figure S1

Synchronized Spawning Behavior of Platynereis dumerilii, Related to Figure 1

Numbers of mature animals oscillate in accordance with the lunar cycle. Figure redrawn from Hauenschild (1955).

We first assessed if our Platynereis dumerilii culture also possesses a nocturnal-light-adjusted circalunar spawning cycle. Following the conditions used in classical experiments (Hauenschild, 1954, 1955, 1960), we subjected the culture to a circadian light regimen of 16 hr light and 8 hr darkness (Figure 1C). For eight consecutive nights of a lunar month, we exposed the worms to dim nocturnal light (termed “full moon period” [FM]). We refer to the middle week of the remaining period as NM. This monthly light cycle in the lab can be in phase (Figure 1D) or out of phase (Figure 1E) with the natural moon. In accordance with classical observations (Figure S1), the daily number of mature animals peaked at the time between the FM stimuli (Figures 1D and 1E). Irrespective of the natural moon phase, these peaks of maturing animals remained in phase with respect to the week of the nocturnal light stimulus (Figures 1D and 1E). This establishes that nocturnal light stimuli alone are sufficient to synchronize circalunar reproductive cycles in our Platynereis lab culture.

Next, we tested if the observed circalunar spawning rhythm was controlled by an endogenous circalunar clock. As this point was debated previously (Hauenschild, 1960; Palmer, 1974), we performed lunar free-running experiments. After entrainment of animals for more than 2 months in the described circadian and circalunar light regimens, the FM stimulus was omitted, whereas the circadian light cycle remained unchanged (termed “free-running full moon” [FR-FM] in Figure 1C). The NM after this FR-FM is termed “free-running new moon” (FR-NM in Figure 1C). Worms continued to exhibit a monthly reproductive periodicity under these conditions (Figure 1F), with a period of 30 days (Figure 1G). Worms under constant light or raised without any nocturnal illumination did not show reproductive rhythms (Figure 1H). This establishes that circalunar reproductive periodicity in our culture is governed by an endogenous circalunar clock.

Platynereis Possesses the Full Complement of Drosophila and Mouse Core Circadian Oscillator Gene Orthologs

After we established that our worms possessed an endogenous circalunar clock, we tested for the presence of an endogenous circadian clock. For this, we determined the worms’ complement of core circadian clock genes and their expression dynamics. Bmal, period, and clock are present in the core circadian oscillator in vertebrates and flies (Young and Kay, 2001); cryptchrome (cry) acts as core clock component in vertebrates and nondrosophilid invertebrates (Chaves et al., 2011; Zhan et al., 2011; Zhu et al., 2005); timeless is crucial for the insect circadian clock (Myers et al., 1995), but the gene is absent from vertebrates (Gotter, 2006). Orthologs of timeout (also termed tim2) and cry are important for circadian clock entrainment in insects (Benna et al., 2010; Emery et al., 2000). Moreover, cry is part of the circadian oscillator in the fly peripheral clock (Ivanchenko et al., 2001; Krishnan et al., 2001; Levine et al., 2002). Pdp1 acts together with vrille in a modulatory feedback loop on the core transcription/translational feedback loop in insects (Blau and Young, 1999; Cyran et al., 2003; Glossop et al., 2003). Of these genes, only bmal had been identified in larval Platynereis (Arendt et al., 2004). By a combination of degenerated PCR and massive transcript sequencing, we identified Platynereis orthologs of period, clock, timeless, timeout, pdp1, and vrille, as well as two cry genes that we name L-cry (“L” indicating orthology to light-receptive Crys) and tr-cry (“tr” indicating orthology to Crys acting as transcriptional repressors) (Figure S2A).

Figure S2

Platynereis L-Cry and tr-Cry Represent Members of the Light Receptive and Transcriptional Repressor-type Cryptochrome Groups, Respectively, Related to Figure 2

(A) Platynereis L-Cry and tr-Cry represent members of the light receptive and transcriptional repressor-type Cryptochrome groups, respectively. Maximum likelihood and neighbor joining trees group Pdu-L-Cry and Pdu-tr-Cry into the different classes with high branch support. The topology of the ML tree is shown, support values are given as ML/NJ at critical branches. Branch colors: green- vertebrates, red- lophotrochozoa, dark blue- insects, gray- plants, black- bacteria.

(B–G) Temporal profiles of clock gene RNA expression from Platynereis heads; (B-D) NM under circadian light regime (E-G) NM under constant darkness. n = 5-15 (B-D); n = 3-6 (E-G); 4-5 heads/n. Values are means ± SEM p value determined by one-way ANOVA.

(H) Example of one of the western blots used for the quantification shown in (Figure 2L). 6h: 6hr light puls; c: dark control. Related to Figure 2.

As transcriptional oscillations are important for circadian clock function (Kadener et al., 2008; Padmanabhan et al., 2012), we next investigated messenger RNA (mRNA) dynamics of these genes. In addition, we focused our analyses on premature adult heads (Figure 1A). It had previously been shown that the maturation of Platynereis, which is the major event known to be synchronized by circadian and circalunar clocks, is controlled by the brain (Hauenschild, 1964, 1966; Hofmann, 1975). In order to ensure that any observed changes were due to the experimental conditions, but not due to developmental stage differences of the worms, we carefully staged the worms based on segment numbers, appendage shape, pigment appearance, and eye and body size.

We first investigated the temporal expression profiles of bmal, period, clock, tr-cry, timeless, vrille, pdp1, and timeout using quantitative PCR (qPCR). This would also allow us to obtain an understanding of how the different circadian clock genes might relate to each other in terms of their regulation. We sampled heads during different circadian points at the NM phase under light-dark (LD) conditions. In these experiments, bmal, period, clock, tr-cry, and timeless (Figures 2A–2E) and vrille, pdp1, and timeout (Figures S2B–S2D) exhibited robust circadian cycles. With the exception of timeless and timeout, this cycling was maintained during constant darkness (DD) (Figures 2F–2J; Figures S2E–S2G), consistent with the notion that bmal, period, clock, pdp1, vrille, and tr-cry are components of a core circadian oscillator in Platynereis heads. Clock and bmal transcripts cycled in phase with each other (Figures 2A, 2C, 2F, and 2H), consistent with a possible heterodimer formation known from flies to mammals (Darlington et al., 1998; Gekakis et al., 1998). Period, pdp1, and timeout transcript oscillations (Figures 2B and 2G; Figures S2C, S2D, S2F, and S2G) were in antiphase with bmal/clock expression. In contrast to Drosophila, where vrille RNA levels peak prior to pdp1 levels (Cyran et al., 2003), vrille RNA peaks followed those of pdp1 in Platynereis (Figures S2B, S2C, S2E, and S2F). tr-cry and timeless RNA level changes were neither directly in phase nor directly in antiphase with bmal and clock. They showed either high levels in the morning or during the evening/night (Figures 2D, 2E, 2I, and 2J). Furthermore, timeless transcripts displayed significantly lower levels under DD, as well as less pronounced and strongly phase-shifted circadian oscillations (Figures 2E and 2J), suggesting that the changes in its RNA level are predominantly directly controlled by light. Finally, transcriptional fluctuations of L-cry did not follow a clear circadian periodicity (Figure 2K).

Figure 2

Platynereis Circadian Clock Gene Orthologs Show Circadian Oscillations on the RNA Level

(A–J) Temporal profiles of clock gene RNA expression in Platynereis heads sampled under NM (A–E) circadian light regimen and constant darkness (F–J) are shown. Values are means ± SEM, n = 5–16 (A–E), n = 6 (F–J); four to five heads/n. The p value was determined by one-way ANOVA. See Figures S2B–S2G for additional circadian clock genes.

(K) Platynereis L-cry transcript levels fluctuate, but do not show regular cycling patterns over 4 days (n = 2).

(L) Light decreases Pdu-L-Cry, but not Pdu-tr-Cry, levels in S2 cells. Dp-Cry1 and Dp-Cry2 serve as positive and negative controls, respectively. V5 epitope-tagged Pdu-L-Cry, Pdu-tr-Cry, Dp-Cry1, Dp-Cry2 was coexpressed with GFP. After a 6 hr light pulse (gray bars) or constant darkness (black bars), cell extracts were collected, western blotted, and probed with anti-V5 and anti-GFP (see Figure S2H). CRY levels were quantified by densitometry of antibody staining after normalization with GFP. The dark value for each CRY was plotted as 100%. Data are means ± SEM; n = 3 independent transfections. Significant differences were assessed by Student’s t test (∗∗p < 0.01; ∗∗∗∗p < 0.0001).

(M) Platynereis tr-Cry, but not the closely related Pdu-L-Cry or Pdu-6-4-photolyase, strongly inhibit Pdu-CLK:Pdu-BMAL-mediated transcription in a luciferase reporter gene assays. The monarch butterfly per E-box-containing enhancer (DpPer4Ep-Luc) was used in the absence (control) or presence of Pdu-clock/Pdu-bmal plasmids (350 ng each). Dp-cry1 and Dp-cry2 serve as positive and negative controls, respectively. Data are means ± SEM; n = 4–8 independent transfections. Significant differences were determined by Student’s t test (∗∗∗∗p < 0.0001).

Pdu-L-Cryptochrome and Pdu-tr-Cryptochrome Can Function as a Light Receptor and Transcriptional Repressor, Respectively

In order to test if the investigated Platynereis circadian clock genes can indeed function in the conventional clock mechanism, we employed two assays previously used to validate the activity of presumptive core circadian clock genes of the monarch butterfly (Zhu et al., 2005, 2008). Cryptochromes functioning as light receptors undergo a light-dependent reduction in protein levels in S2 cells because of proteasome-mediated degradation (Lin et al., 2001; Zhu et al., 2005). If Pdu-L-CRY can indeed function as light receptor, we should be able to observe such a reduction. We assessed the effect of a 6 hr light pulse to promote Pdu-L-CRY, as well as Pdu-tr-CRY degradation, and compared the responses to those of the two monarch butterfly Cryptochrome proteins as positive and negative controls, respectively. We found that the Platynereis L-Cryptochrome, most closely related to Dp-Cry1 and dCry, was strongly degraded under a 6 hr light pulse, whereas Pdu-tr-Cry was not affected (Figure 2L; Figure S2H). This suggests that Pdu-L-Cry can function as a light receptor, like its orthologs in the fruit fly and the monarch butterfly.

We further asked if Pdu-bmal and Pdu-clock are able to activate transcription from an E-box-containing construct. We constructed a luciferase construct, based on previous work in the monarch (Zhu et al., 2005; Yuan et al., 2007), containing two consensus E-boxes of the 5′ flanking region of Dp-per. Cotransfection of this construct with Pdu-bmal and Pdu-clock into S2 cells led to a strong activation of luciferase activity (Figure 2M). Additional transfection of Pdu-tr-cry strongly and highly significantly reduced this activation, similar to our positive control, the monarch butterfly’s tr-cry ortholog, cry2 (Figure 2M). Addition of Pdu-L-cry or its monarch ortholog cry1 did not reduce Pdu-Bmal/Pdu-Clock-mediated luciferase expression in comparable levels (Figure 2M). Similarly, Pdu-6-4photolyase, a gene most closely related to Pdu-tr-cry, but whose orthologs function in UV-induced DNA repair (Sancar, 2008), did not show obvious transcriptional repressor activity (Figure 2M). We thus conclude that bmal, clock, and tr-cry likely function in a core circadian clock positive/negative transcriptional loop in Platynereis dumerilii, like their orthologs in other species.

Platynereis Core Circadian Clock Gene Orthologs Are Confined to Specific Domains in the Medial Forebrain

In order to determine if the uncovered circadian clock gene orthologs localize to a centralized structure or are broadly expressed, we performed whole-mount in situ hybridizations (WMISH).

All genes tested were specifically expressed in the posterior medial brain (Figures 3A–3E; Figures S3A–S3C), particularly in paired oval-shaped structures (arrows and magnifications in Figures 3A-3D; Figure S3A; compare Figures S3D and S3E for sense controls and Figure S3F for expression examples of two noncircadian transcription factors). These brain regions were already noted by Retzius as distinct nuclei in the brain of Nereis, a close relative of Platynereis (Retzius, 1895), hence representing nuclei conserved among nereidid worms. The brain morphology of Platynereis changes little during development from larvae to premature adults (Tomer et al., 2010). By position and relation to the axonal scaffold and the prominent cilia of the ciliary photoreceptor cells (arrows in Figures S3G–S3J), these distinct nuclei arise from the area demarcated by bmal in the 2-day-old larval medial forebrain (Arendt et al., 2004). Our findings of a medial forebrain nucleus harboring the core circadian clock genes are hence also consistent with our previous analyses in Platynereis larvae (Arendt et al., 2004).

Figure 3

Platynereis Circadian Clock Gene Orthologs Are Confined to a Specific Brain Nucleus

(A–D) Whole-mount in situ hybridization of circadian clock genes on premature adult Platynereis heads is shown. Arrows point at the morphologically visible border of the medial brain nuclei expressing the genes. See also magnified view as indicated by the box; dorsal view, anterior to the top. For additional circadian clock genes, sense controls and expression of nonclock genes, see Figures S3A–S3F. Arrowheads indicate expression in eyes. Scale bar represents 50 μm, and asterisk indicates the position of major brain neuropil.

(E) Scheme of worm head indicating area is shown. Circadian clock gene expressing brain nuclei are indicated as blue ovals. e, adult eyes.

Figure S3

Specific Expression of Platynereis Circadian Clock Gene Orthologs in the Medial Forebrain, Related to Figure 3

(A–C) Whole mount in situ hybridization (WMISH) of period, L-cry and pdp1 on premature adult Platynereis heads. Timeline for period expression anti-correlates with Pdu-bmal expression timeline (see Figure 3A). Dotted oval outlines circadian clock gene expressing, oval-shaped nucleus. Dorsal views, anterior up. Compare to Figures 3A–3E.

(D and E) WMISHs of period and bmal sense controls. Arrows point at oval-shaped brain nucleus expressing circadian clock genes.

(F) WMISHs of non-circadian transcription factors pax6 and nk2.1 show that the co-expression of circadian clock genes in the oval-shaped nucleus is specific.

(G–J) Co-localization of circadian clock gene expression with ciliary photoreceptor cells. Arrows point at the large cilia of the ciliary photoreceptors, counterstained by the anti-acetylated tubulin antibody, visualizing stabilized microtubules (Arendt et al., 2004).

(G) DIC image.

(H) Fluorescent image visualizing the anti-acetylated tubulin antibody staining (red).

(I) Overlay of N and O. (J) scheme of P. (K) Expression of bmal in cells of the adult eye as revealed by WMISH on a Platynereis eye color mutant. (L) WMISH of ck1 delta/epsilon shows co-expression with Platynereis circadian clock genes in the oval-shaped nucleus (see magnified view). In addition ck1 delta/epsilon expressing cells are located anterior and posterior to the neuropil (arrow). Dorsal views, anterior up, asterisk- position of major brain neuropil. e- adult eye. Scale bar 50μm. Related to Figure 3.

The observed coexpression of the Platynereis clock gene orthologs is consistent with them acting together in a positive-negative feedback loop, as typical for the core circadian oscillators of all animals analyzed to date. Likewise, the expression of L-cry in the same oval-shaped posterior medial forebrain domains (Figure S3B; compare to Figures 3A–3E), along with the presented functional data, are consistent with L-Cry serving as a possible light sensor for the Platynereis circadian clock. This is also coherent with the fact that light should be able to reach these cells, as the worm’s brain is relatively small and the cuticle largely transparent.

In addition to these nuclei, we also noted circadian clock gene expression in the area of the eyes (arrowhead in Figures 3A and 3B). Again, this staining was not present in sense controls, nor was it typically present when other genes were stained (Figures S3D–S3F). In order to analyze the exact position and extent of this expression, we performed WMISH on a Platynereis eye pigment mutant (Fischer, 1969). As in Drosophila (Hunter-Ensor et al., 1996), cells in the eyes also exhibited circadian clock gene expression (Figure S3K), albeit in general less than in the posterior medial forebrain domain.

We next analyzed if our WMISH confirms the daily transcriptional oscillations observed by qPCR. For this, we focused on the two most strongly expressed clock gene orthologs, Pdu-bmal and Pdu-period. When analyzed at different circadian time points, the expression of both genes showed circadian fluctuations within the described two medial brain nuclei (Figure 3A; Figure S3A), suggesting that these are the major circadian clock centers of Platynereis.

Platynereis Locomotor Activity Is under Circadian Clock Control

Given that Platynereis exhibits molecular circadian oscillations in paired medial forebrain nuclei, we next asked if the worms also displayed circadian behavior. We therefore recorded worms in a box over several days using an infrared camera and categorized their behavior into active (searching, fighting) and inactive (no movements, undulatory fanning movements) types (Figures S4A–S4C). We first analyzed if the worms showed any consistent activity patterns over multiples of 24 hr during NM/LD. The activity data were analyzed using ActogramJ for chronobiological analyses for rhythmicity and period lengths (Schmid et al., 2011). Under NM/LD conditions, the worms displayed primarily nocturnal activity (Figures 4A and 4B) with an average period length of 24.2 hr (±0.2) (Figures 4C-4E and 4J; Figures S4D and S4J). These data are consistent with the fact that the nuptial dance of Platynereis only occurs during few hours of the night (Korringa, 1947), and with previous observations in the related nereidid Nereis (Last, 2003; Last and Olive, 2004).

Figure S4

Platynereis Circadian Locomotor Activity under Laboratory Conditions, Related to Figure 4

(A) Types of activities scored as either active (1) or inactive (0) for analyses of Platynereis circadian locomotor behavior.

(B) Recording set-up closed (large image), open (small image) showing LED infrared array (white box), camera and the control recording device for light and temperature ( = HOBO temperature tight 28000DP logger), arrow).

(C) Spectral analysis of white light LED's (OSRAM 24V) used in the black box set up (visible light spectrum).

(D–G) Mean locomotor activity (hourly average ± SEM) over 3 consecutive days containing rhythmic and arrhythmic animals, (D) N = 14, (E) N = 17, (F) N = 22, (G) N = 24. Compare to Figures 4A and 4F–4H.

(H) Average periodogram for NM/DD conditions using Lomp-Scargle analysis. NM/DD: worms were entrained normally with circadian and circalunar illumination conditions. Recordings were performed during NM in complete darkness. Compare to Figures 4F, 4J, and 4L.

(I) Frequency power was calculated for all rhythmic animals. Worms under NM LD/DD versus FR-FM LD/DD show significant differences in strength of the rhythm. Error bars represent ±SEM. Student’s t test. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

(J–M) Individual Lomp-Scargle periodograms arranged according to increased period length for (J) NM/LD, (K) NM/DD, (L) FR-FM/LD, (M) FR-FM/DD, see D–G for illumination. Red line indicates significant p-level = 0.05. Period length > 29 were excluded from the 3 day analysis (indicated by dotted line) R; rhythmic, WR; weakly rhythmic, AR; arrhythmic. Related to Figure 4.

Figure 4

The Circalunar Clock Affects Circadian-Clock-Controlled Activity Rhythms

(A) Mean locomotor activity (hourly average ± SEM) shows higher nocturnal activity in Platynereis in NM under 16:8LD circadian illumination over the course of 3 days (N = 12 rhythmic animals). Active behaviors were counted as 1, inactive as 0. See Figures S4A–S4C for details on active versus inactive behaviors and recoding setup.

(B) Quantification of average locomotor activity per hour of day hours (yellow bar) versus night hours (black bar) of 3 consecutive days is shown. Error bars represent ±SEM. Significant differences were determined by Student’s t test (∗∗∗∗p < 0.0001).

(C) Percentage of present period length of individual worms under NM/LD conditions is shown. See individual periodograms in Figure S4J.

(D) Average periodogram (N = 12) for NM/LD conditions shows a dominant period of 24 hr and an additional 12 hr peak. The red line indicates the significant p level = 0.05.

(E) Actograms and their corresponding periodogram of 3 individual worms recorded under NM/LD conditions are shown.

(F) Platynereis locomotor activity cycles continue in NM under complete darkness (DD) over at least 3 consecutive days (N = 10 rhythmic animals) showing a higher nocturnal activity. NM/DD: worms were entrained normally with circadian and circalunar illumination conditions. Recordings were performed during NM in complete darkness. See (A) for scoring details and Figure S4E for activity cycles including arrhythmic animals and Figures S4H and S4K for periodogram analysis.

(G) Mean locomotor activity cycles continue in FR-FM under normal light/dark (LD) conditions showing an increase in daily locomotor activity (N = 18 rhythmic animals). See (A) for scoring details and Figure 1C for details on illumination. See Figures S4F and S4L for activity cycles including arrhythmic animals and periodogram analysis, respectively.

(H) Platynereis daily locomotor activity in FR-FM under complete darkness (DD) is flattened and displays a shorter period of about 18 hr (N = 15 rhythmic animals). See Figures S4G and S4M for activity cycles including arrhythmic animals and periodogram analysis, respectively.

(I) Quantification of average locomotor activity per hour of day hours (yellow bar) versus night hours (black bar) comparing NM/LD versus FR-FM/LD versus FR-FM/DD is shown. Worms under FR-FM/LD are nocturnal, but exhibit higher daily locomotor activity than during NM/LD. Worms in FR-FM under complete darkness (DD) show no nocturnal activity anymore, but an increase in daily activity. Error bars represent ±SEM. Significant differences were determined by Student’s t test (∗∗∗p < 0.001; ∗∗∗∗p < 0.0001).

(J) Summary of Lomb-Scargle periodogram analyses of time series of locomotor activity observed under different circadian and circalunar conditions over the course of 3 days (see Figure 1C) is shown. Period and Power were calculated for all rhythmic worms. N, number of worms analyzed; R, rhythmic; WR, weakly rhythmic; AR, arrhythmic; see Experimental Procedures for classification. Data from three independent NM, DD, FR-FM experiments and from two independent FR-FM/DD were pooled, respectively.

(K) Worms in NM versus FR-FM show significant differences in circadian activity period length. Error bars represent ±SEM. Significant differences were determined by Student’s t test (∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001).

(L–N) Percentage of present period lengths of individual worms is shown. (L) Under the NM/DD condition, the circadian period is reduced to 40%. Worms show additional longer and shorter periods. (M) Under the FR-FM/LD condition, worms display additional periods of about 9 hr and 18 hr, which are not present in NM under LD or DD (compare C and L). (N) In the FR-FM under DD condition, worms show an increase in period lengths of about 18 hr and 9 hr, decreasing the percentage of other periods (compare C, L, and M).

In addition, a dominant ∼12 hr period was observed in 8% of the analyzed individuals (n = 14, Figures 4C and 4E). This 12 hr activity rhythm does not appear to be crepuscular, possibly rather resembling a circatidal rhythm (see example worm 3 in Figure 4E). Under NM/DD conditions, the worms continued to show a circadian periodicity (23.6 ± 1.5 hr) over at least 3 days (Figures 4F, 4J, and 4L; Figures S4E, S4H, and S4K), evidencing that the worm’s locomotor activity is under circadian clock control. The generally still relatively high level of variability in the period lengths of our periodogram analyses might be due to the representative, yet still relatively short analyses timeframe and small sample size.

The Circalunar Clock Impacts Circadian Behavior

Having established that both circadian and circalunar clocks exist in Platynereis, we next investigated how these two clocks interact with each other.

We started by comparing the Platynereis circadian locomotor activity cycles between two different phases of the circalunar clock (NM versus FR-FM; see Figure 1C). Compared to NM, the worms were less rhythmic in their locomotor behavior under FR-FM in both circadian LD and DD conditions (Figures 4F–4H and 4J). Their activity during the day significantly increased (Figures 4G–4I), while the average period length significantly shortened to 18.2 hr (±1.5 hr) for FR-FM/LD and 15.9 hr (±1.7 hr) for FR-FM/DD, the power of the rhythm decreased to 18.9 (±1.6) and 16.9 (±1.9), respectively (Figures 4J and 4K; Figures S4F, S4G, and S4I). Analyses of the periodograms of individual animals revealed that such shorter period length occurred indeed on individual bases, but can vary from 8 hr to 18 hr (Figures 4M and 4N; Figures S4L and S4M). Occasionally, worms already showed rhythms of shorter period lengths during NM (LD and DD) (Figures 4C and 4L; Figures S4J and S4K). However, the number of worms exhibiting behavioral rhythms with periods clearly different from 24 hr was strongly increased in FR-FM (LD and DD) compared with NM (LD and DD) (Figures 4C and 4L–4N). This provides strong evidence that the circalunar clock affects circadian behavior.

The Circalunar Clock Impacts Transcript Levels of clock, period, pdp1, and timeless

Changes in circadian behavior have been directly connected to changes in circadian clock gene levels in Drosophila and mice (Antoch et al., 1997; Benito et al., 2007; Kadener et al., 2008). We therefore next investigated if the oscillation and levels of circadian clock gene orthologs were also affected by the circalunar clock by comparing RNA levels between NM and FR-FM (cf. Figure 1C). For the genes pdp1, clock, period, and timeless, the circadian expression dynamics (period lengths and phase, represented as shape of the graphs) under FR-FM were not detectably different to NM conditions (Figures 5A-5C; Figure S5A, pink graphs). However, their overall transcript levels were significantly elevated at FR-FM compared to NM (Figures 5A–5C and 5A′–5C′; Figures S5A and S5A′). In contrast, expression levels and circadian dynamics of bmal (Figures 5D and 5D′), tr-cry, vrille, and timeout (Figures S5B–S5D and S5B′–S5D′) did not differ between FR-FM and NM conditions (pink versus blue graphs).

Figure 5

The Circalunar Clock Influences Circadian Clock Gene Expression

(A–D) Temporal profiles of clock gene RNA expression in Platynereis heads sampled during NM (blue) and FR-FM (pink) at the indicated Zeitgeber time point (ZT) are shown. See Figure 1C for detailed information on the circalunar-light regimen. Values are means ± SEM, NM n = 5–16, FR-FM n = 3–10; four to five heads per n. The p value was determined by one-way ANOVA.

(A′–D′) Overall daily transcript levels calculated as area under the curve (AUC) based on 24 hr expression data shown in (A)–(D) are shown. Values are means ± SEM; NM n = 6–16, FR-FM n = 3–10. The p value was determined by one-way ANOVA. Significant differences were determined by Wilcoxon signed rank test (∗p < 0.05; ∗∗∗p < 0.001); four to five heads per n. (E) Whole mount in situ hybridization shows an increase of pdp1, clock, and period levels at FR-FM versus NM in the oval shaped circadian clock gene expressing forebrain domain (compare Figures 3A–3E). See Figure S5 for analyses of additional circadian clock genes.

Figure S5

Circadian Transcriptional Regulation of timeless Is under Circalunar Clock Control, Related to Figure 5

(A–D) Temporal profiles of clock gene RNA expression measured by qPCR under NM and FR-FM conditions from Platynereis heads. See Figure 1C for detailed information on circalunar-light regime. Values are means ± SEM, NM n = 5-15, FR-FM n = 3-11; 4-5 heads/n. p-value determined by one-way ANOVA.

(A′–D′) Overall daily transcript levels calculated as area under the curve (AUC) based on 24h expression data shown in (A–D). Significant differences by Wilcoxon signed rank test. ∗p < 0.05.

(E) Expression of timeless increases in the circadian core brain domains in FR-FM compared to NM. Scale bar 50μm. Oval- outline of circadian clock brain area (also compare to Figures 3A–3E).

(F–H) Temporal profiles of clock gene mRNA expression under FR/NM (gray graph) resemble that of NM and FR-FM. Values are means ± SEM, FR-NM n = 3. Data for NM and FR-FM re-plotted from Figures 5B–5D.

(F′–H′) Area under the curve (AUC) based on 24h data shown in (F–H). Platynereis overall expression dynamics of clock and period are significantly decreased in FR-NM (gray bar) compared to FR-FM and are similar to NM (FR-NM n = 3; 4-5 heads/n). Data for NM and FR-FM re-plotted from Figure 5B'-D'. Significant differences by Wilcoxon signed rank test. ∗p < 0.05, ∗∗∗p < 0.001. Related to Figure 5.

We hence conclude that the overall transcript levels of clock, period, pdp1, and timeless are directly or indirectly modulated by the circalunar clock.

If this is indeed the case, the transcript levels at the next NM under circalunar free-running conditions (FR-NM; see Figure 1C) should return to the levels observed under normal NM. This is indeed the case for the three genes tested representatively. Circadian oscillations and transcript levels of clock, period, and bmal in FR-NM resembled that of NM (Figures S5F–S5H and S5F′–S5H′).

Analyses of premature adult brains using WMISH revealed that the elevation of clock, period, pdp1, and timeless transcripts during FR-FM was not due to additional brain domains expressing these genes, but that the same cells in the two core circadian brain nuclei now express at higher levels (Figure 5E; Figure S5E).

These results predict that at least one of the circadian clock genes clock, period, pdp1, or timeless function either downstream of the circalunar oscillator, or as part of it, and establish the regulation of mRNA levels as an output of the circalunar clock.

Circadian Clock Gene Oscillations Are Not Required for Circalunar Clock Function

We next asked if the circadian clock affects, or is part of, the worm’s circalunar clock. Different hypotheses have been put forward to explain rhythms with a semilunar or lunar period length. Many of these models involve circadian oscillators. One model relies on the interaction of the circadian oscillator with an oscillator running with a circalunidian or tidal period (i.e., 24.8 hr or 12.4 hr) so that both only coincide once per lunar or semilunar month (Figure 6A; Soong and Chang, 2012). Alternatively, the counting of circadian cycles has been proposed in the frequency demultiplication hypothesis to lead to a circalunar rhythmic output (Soong and Chang, 2012).

Figure 6

The Circalunar Clock Is Independent of Circadian Clock Oscillations

(A) A dual oscillator model could explain circalunar clock function. A circadian (24 hr, length of the solar day) and circalunidian (24.8 hr, length of a lunar day) oscillator function together to generate monthly (29.5 days) periods.

(B and C) Circadian clock gene transcriptional oscillations are severely affected under PF-670462 treatment compared to nontreated controls (dashed line). Values are means ± SEM; n = 3; four to five heads per n. See Figure S6 for additional circadian clock genes.

(D) Behavioral analyses (one behavioral score per minute of a 10 min interval per hour) as described in Figures S4A and S4B from one representative example of untreated controls (active behavior, indicated by arrows, mainly restricted to the dark phase) versus PF-670462-treated worms (active behavior distributed).

(E) PF-670462 abolishes rhythmic circadian locomotor activity in Platynereis. Worms were recorded under 16:8LD circadian illumination (see Figure 4A for a nontreated comparison).

(F and G) Periodogram analyses of individual worms show that PF-670462 treated animals are in majority arrhythmic (AR). No worm was rhythmic (R), and few worms remaining weakly rhythmic (WR) showed a strongly altered period length of 17 hr.

(H and I) Circalunar spawning cycles are maintained in control (H) and under PF-670462 treatment (I). Collection data from five independent experiments were pooled.

We thus next tested if circadian clock gene oscillations were required for circalunar clock function in Platynereis. For this, we interfered with the Platynereis circadian clock and assessed the effects of this interference on circalunar spawning peaks. Mammalian casein kinase 1δ/ε and its Drosophila ortholog Double time (DBT) are crucial for normal circadian clock function (Lee et al., 2009). Their best-documented function is Period phosphorylation, which serves to enhance Period degradation in both systems (Gallego and Virshup, 2007). PF-670462 and other CK1δ/ε inhibitors severely affect the circadian period in mammalian cells (Eide et al., 2005; Walton et al., 2009).

The Platynereis ck1δ/ε ortholog is widely expressed, including in areas of the medial forebrain and the oval- shaped core circadian clock brain nuclei (Figure S3L). Upon PF-670462 treatment, the amplitudes of bmal, clock, tr-cry, timeout, timeless, and pdp1 transcriptional cycling were flattened to a level that no clear oscillations were observable anymore in Platynereis (Figures 6B and 6C; Figures S6A–S6D), while period transcription showed irregular fluctuations (Figure S6E).

Figure S6

PF-670462 Treatment Disrupts Circadian Transcript Oscillations, Related to Figure 6

(A–E) Clock gene transcriptional oscillations are severely affected under PF-670462 treatment compared to non-treated controls (dashed line). Values are means ± SEM; n = 3; 4-5 heads/n. Related to Figure 6.

Consistent with the abolished molecular circadian clock oscillations, we also found that 70% of PF-670462-treated worms were arrhythmic in their daily activity when tested under NM(LD) conditions (Figures 6D–6G). The remaining 30% showed weak rhythmicity, but their period length was severely altered to about 17 hr. Despite their severely disrupted circadian rhythmicity, PF-670462-treated worms were still capable of displaying all types of normal behaviors (Figure 6D). This is apparent from the mean analysis (Figure 6E), but also from individual worms (Figures 6D–6G), attesting to the notion that PF-670462 treatment leads to a disruption of the circadian core clock in Platynereis in the majority of the population. Despite these significant changes in circadian clock gene dynamics, however, PF-670462 treatment did not affect the circalunar spawning periodicity of Platynereis when compared to controls in free-running experiments (Figures 6H and 6I; compare to Figure 1H for arrhythmic spawning).

We tested several concentrations of PF-670462 and performed the circalunar spawning assays with the lowest concentration still exhibiting robust effects on circadian clock molecular oscillations. Whereas we cannot exclude that PF-670462 also affects other targets at the given concentration, we can conclude that none of these effects, including the one on the circadian clock, shows an obvious impact on the circalunar clock.

Based on these results, we conclude that the circalunar clock in Platynereis is independent of the oscillations of the circadian transcriptional clock (Figure 7).

Figure 7

Circadian and Circalunar Clock Model in Platynereis

Proposed interaction of separate circadian and circalunar oscillators in Platynereis dumerilii is shown. Solid blue line indicates impact of the circalunar oscillator on the transcriptional regulation of circadian clock gene expression resulting in elevated levels of pdp1, period, clock, and timeless. The impact of the circalunar clock on the circadian clock genes can be direct or indirect on one or all of these genes.

Discussion

Life with More Than One Type of Clock

Here, we show that the bristle worm Platynereis dumerilii harbors two endogenous clocks, with a circadian and a circalunar period length, respectively. The coexistence of multiple clocks in one organism is likely a rather common phenomenon, yet most clearly displayed outside of the group of the conventional molecular animal model species (Naylor, 2010; Tessmar-Raible et al., 2011). Consequently, the interactions of such clocks have only been investigated to a very limited extent (Takekata et al., 2012). We provide evidence that the oscillatory mechanisms of both clocks are distinct, but that they both converge on the regulation of transcript levels and behavior.

Whereas our behavioral analyses focused on premature adult Platynereis worms, we propose that the observed modulation of circadian behavior by the circalunar clock also underlies the regulation of other behaviors, such as the nuptial dance of mature animals. This mating behavior is known to be synchronized both to particular days of the month and to specific hours of the night (Korringa, 1947). Synchronized mating likely increases the reproductive success of externally fertilizing animals, especially when they occur in large populations, as for instance in reef corals (Harrison et al., 1984).

The biological implication of the changes in behavioral period length of the premature adult worms might only be understandable when we will know more about the natural conditions the worms have to adapt to outside of the time of mating.

Parallel to our study, work on Eurydice pulchra revealed the coexistence of molecularly independent circatidal and circadian clocks in this crustacean (Zhang et al., 2013). A possible coordination of these two clocks might occur by their coregulation by CK1δ/ε, as PF-670462 incubation led to an increased period length of both circadian and circatidal clocks (Zhang et al., 2013). An effect of PF-670462 on the period length of the Platynereis circalunar clock is possible, but as Platynereis only spawns once, our current analyses rely on scoring large populations, making this question technically very difficult to test. Live readouts of the circalunar clock in individual worms will be helpful to answer such and further questions on circalunar and circadian clock interactions in the future.

The Effect of the Circalunar Clock on Circadian Period Length

Our results show that on the behavioral level, the period length and strength of the circadian rhythm are significantly modulated by the circalunar clock. The change in behavioral period length contrasts with the seemingly unaltered period length of the molecular oscillations of the core circadian clock genes. We currently see two possibilities to explain this discrepancy.

On the one hand, the period length of the worm’s locomotor rhythms could be modulated independently of the core circadian clock, albeit still also under circadian clock control. In such a model, the circalunar clock would directly target genes (downstream or independently of the circadian clock) that can regulate behavior. It is for instance conceivable that the circalunar clock affects the levels of hormone precursors, processing enzymes or neurotransmitters. By changing thresholds, these changes (in combination with the circadian clock control) could subsequently result in the observed behavioral phenotype. To exemplify, if lowering the overall levels of a suppressor, a transmitter affecting behavioral activity could reach critical levels high enough to elicit activity more often (e.g., twice per day instead of once per day).

On the other hand, it could be possible that the elevation of clock, period, pdp1, and timeless mRNA levels during FR-FM causes (at least partly) the behavioral changes. A possible scenario how this could be the case is outlined below.

It is well-established that 12 hr rhythms occur in the expression of approximately 1% of all genes in mouse liver, although the circadian clock is unaltered (Hughes et al., 2009, 2012; Vollmers et al., 2009). In addition, 8 hr rhythms in gene expression also occur naturally (Hughes et al., 2009). It seems plausible that what happens in the liver might also happen to cells in other tissues, such as neurons in the brain. In addition, it is also plausible that changes in locomotor activity rhythms can be controlled by changes in gene activity of genes affecting behavior, such as hormonal precursors, processing enzymes, or neurotransmitters. Thus, gene activity cycling with 12 hr or 8 hr rhythms could generate 12 hr or 8 hr behavioral activity cycles in the background of a normally functioning circadian clock.

A recent theoretical work provides a mathematical model explaining the generation of such naturally occurring 12 hr gene expression cycles based on changes in the binding of circadian transcription factors to separate (noncompetitive) binding sites (Westermark and Herzel, 2013). More specifically, two points of the mathematical model might help to explain the findings described in our work. (1) The oscillation amplitudes of the core circadian transcription factors have an impact on the circadian term of the equation (i.e., if they are equal, the circadian term will vanish). In other words, 12 hr cycles can occur based on changes in the amplitude of the core circadian transcription factors that themselves still cycle with a 24 hr periodicity. This could explain, how the changes in transcript levels we observe for some core circadian transcription factors could finally lead to changes in locomotor activity cycles. (2) A combination of less 24 hr and more 12 hr periods in transcription factor rhythms can produce 8 hr fluctuations. In both FR-FM (DD and LD) conditions, we observe such a decrease of 24 hr behavioral periods, combined with an increase in 12 hr periods. Thus, our observed combination might “automatically” lead to the appearance of ∼8 hr rhythms, which is what we indeed observed.

Furthermore, besides the mathematical-model-based considerations, there is also functional evidence that slight changes in gene levels can influence the period length of locomotor activity. The introduction of one or more additional copies of the clk genomic region significantly alters the circadian locomotor activity period in Drosophila (Kadener et al., 2008). This effect is thought to be caused by the increased transcriptional levels of clk’s direct target genes per, pdp1, and tim (Kadener et al., 2008). Remarkably, we see the same genes upregulated by the circalunar clock in Platynereis, raising the possibility that, in analogy to Drosophila, an increase in RNA levels of Platynereis clock can account for the upregulation of period, pdp1, and timeless transcript levels, and in consequence for the significant shortening of the circadian behavioral period length of the worm. One additional piece of evidence that changes in circadian clock gene mRNA levels can manifest themselves in changes in locomotor output rhythms stems from a study of the pdp1 gene (Benito et al., 2007).

Finally, it should also be taken into consideration that the changes in locomotor period length are differently prominent in different individual animals. While we can observe individual difference on behavioral levels, the observation of gene activity in individual animal heads (and not in pools of animal heads) over time is currently technically not feasible. This could blur smaller changes in the period length of the molecular oscillations.

Possible Circalunar Clock Models in Platynereis dumerilii

Our study shows that circalunar clock function is not affected even when the transcriptional oscillations of putative circadian clock genes are severely impaired, arguing against any circalunar clock model involving the classical circadian clock. It is, however, conceivable that the maintained daily light-dark cycle is sufficient to drive circalunar rhythms, in absence of circadian clock oscillations. Finally, our data do not test if the classical circadian clock might still be required for the entrainment of the circalunar oscillator.

Experimental Procedures

Worm Culture and Light Conditions

Worms were maintained as described previously (Hauenschild and Fischer, 1969). See the Extended Experimental Procedures for further detail. Worms of the following inbred strains were used: PIN-mix, VIO-mix, and ORA-mix. All animal work was conducted according to Austrian and European guidelines for animal research.

Gene Identification

Fragments of Platynereis sequences described in this study were identified by high-throughput sequencing of normalized complementary DNA (cDNA) using 454 technology. These fragments were subsequently expanded by rapid amplification of cDNA end (RACE) PCR, using Clontech’s Smart RACE cDNA amplification kit. Primers and program are listed in the Extended Experimental Procedures.

Phylogenetic Analyses

Sequences were aligned using the MAFFT alignment algorithm. (http://www.ebi.ac.uk/Tools/mafft/index.html). The resulting alignments were subsequently used to generate NJ and ML trees. See the Extended Experimental Procedures for further detail.

Total RNA Extraction and RT-PCR

Total RNA was extracted from heads of premature adult worms using the RNeasy Mini Kit (QIAGEN). Reverse transcription was carried out using 0.4 μg of total RNA as template (QuantiTect Reverse Transcription kit, QIAGEN). RT-PCR analyses were performed using a Step-One-Plus cycler. The expression of each test gene was normalized by the amount of the internal control gene cdc5. Using rps9 as reference genes made no significant difference. The relative expression was calculated using the following formula: 1/2ΔCt. Overall levels of expression (area under the curve) were calculated using the trapezoid rule on the relative expression profile of any given gene over 24 hr. All data are shown as the mean ± SEM. See Extended Experimental Procedures for primers and program.

Behavioral Observations and Analyses

Animals entrained under circadian and circalunar light regimes for at least 2 month were transferred into a box (20 × 20 cm, 15–20 animals) containing saltwater (depth 1 cm). Animals were fed prior to the recording to eliminate any behavioral changes in response to feeding. Locomotor behavior was recorded within a black box (white light light-emitting diodes [LEDs]: COINlight CM01E, 150 lux; see spectral analysis in Figure S4C) under given light regimen (LD, DD) using a Chameleon USB 2.0 digital video camera. Light intensity was measured with a USB2000+ spectrometer (Ocean Optics). In order to visualize the worms under dark conditions, an infrared-light LED array (Roschwege GmbH) (990 nm) was placed inside the black box and an infrared (IR) high-pass filter restricted to the detection of IR light into to the camera. Video images were taken continuously over several days and used to evaluate the behavior according to the specified types of behavior (active = 1, inactive = 0).

Behavior was analyzed manually every 1 min of a 10 min interval per hour and the data were imported into ActogramJ Software (University of Wuerzburg) for circadian analysis (Schmid et al., 2011). Locomotor activity was calculated as the number of active behavior events occurring every 1 hr. Periodograms were generated using Lomb-Scargle analysis. Periodicities were confirmed using Fourier transform analysis (FFT) and chi-square analysis. The significant p level was set to 0.05. Worms with a power ≥ 15 were designated as rhythmic (R), worms with a power ≤ 15 were designated as weakly rhythmic (WR). Worms with a power ≤ p level in periodogram analysis were defined as arrhythmic (AR). For time-point analysis, a t test was performed using GraphPad Prism version 6.00 for Windows.

Light-Induced Degradation Assay

Full-length Pdu-tr-cry and Pdu-L-cry sequences were codon optimized for insect codon usage and subcloned at the NotI/XbaI restriction sites into pAC5.1/V5-HisA by Entelechon. As the positive/negative controls, monarch butterfly Danaus plexippus Dp-Cry1 and 2 were used (Zhu et al., 2005). Dp-Cry1 and 2 subcloned into pAC5.1/V5-HisA were kindly supplied by Dr. Reppert. pAct-EGFP, in which Drosophila actin promoter, EGFP, and SV40 polyA sequences were subcloned into pBlueScript (Invitrogen), was used for internal control of transfection. S2 cells were maintained at 25°C in Schneider’s Drosophila medium (Invitrogen) supplemented with 10% heat-inactivated fetal bovine serum (Biological Industries). S2 cells (1.5 × 106) were seeded in six-well plates and next-day transfection was performed using Cellfectin reagent (Invitrogen). Each transfection had 4 μg of each Pdu-tr-cry, Pdu-L-cry, Dp-Cry1, or Dp-Cry2, and 1 μg of pAct-EGFP was added. Then 48 hr after transfection, light treatment was performed as described previously (Yuan et al., 2007). Light treatment involved placing S2 cell culture plate under fluorescent light (3,000–4,500 lux) for 6 hr at 24°C. Dark control plate was wrapped with aluminum foil and incubated beside the light-treated plate. Western blotting was performed using a monoclonal mouse anti-V5 immunoglobulin G (IgG) (Nacalai Tesque) and a monoclonal mouse anti-GFP IgG (Roche Diagnostics). Bands intensity was measured by LAS1000 (FUJIFILM). The cryptochrome’s (V5) band intensity was normalized by each GFP band’s intensity.

Transcription Repression Assay

Full-length Pdu-clock, Pdu-bmal and Pdu-6-4photolyase sequences were PCR amplified from cDNA, subcloned into pJet2.1, sequence verified and subsequently subcloned into pAC5.1/V5-HisA, generating pAct-Pdu-clock, pAct-Pdu-bmal and pAct-Pdu-6-4photolyase.

To generate the reporter construct, a 120 bp segment of the 5′ flanking region of monarch butterfly, Danaus plexippus, per2 gene (NCBI accession number AY364479, bases 1,177–1,296), which contains two E-boxes, was synthesized and cloned in the pGL3-Basic vector (Promega), generating plasmid pDpPer2 (E-box)-luc. S2 cells (6 × 105) were seeded in 12-well plates and transfected the next day with Cellfectin (Invitrogen). Each transfection had 350 ng each of pAct-Pdu-clock, pAct-Pdu-bmal, and various amounts of pAct-Pdu-tr-cry or pAct-Pdu-L-cry or pAct-Pdu-6-4photolyase or 350 ng of pAct-Dpcry1 or 2 (Kume et al., 1999). In each transfection experiment, the reporter plasmid pDpPer2-luc (10 ng) and the pRL-SV40 vector (Promega) (25 ng) were added (Kobayashi et al., 2000). The total DNA per well was adjusted to1.05 μg by adding pAC5.1/V5-HisA as carrier. Then 48 hr after transfection, cells were harvested and their firefly and Renilla luciferase activities determined by luminometry. The reporter luciferase activity was normalized for each sample by determining the firefly:Renillaluciferase activity ratios. In each experiment, the luciferase activity of the PduClk:PduBMAL1-containing sample was taken as 100%. All experiments were repeated at least three times.

Statistical Analysis

Statistical analysis of real-time data was performed using the nonparametric Wilcoxon signed rank test using R Software: A Language and Environment for Statistical Computing, providing a conservative test for significant differences between two sample types (http://www.R-project.org) (Hollander and Wolfe, 1973).

For the Wilcoxon signed rank test, a paired, one-tailed significance interval of 0.05 was used (∗p < 0.05, ∗∗p < 0.005, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001).

One-way ANOVA test and Student’s t test was performed using GraphPad Prism version 6.00 for Windows (R Development Core Team, 2005).

Treatment of Worms with PF-670462

Premature adult worms of mixed ages were incubated in 800 μM PF-670462 (Tocris, #3316) and grown as the rest of the worm culture. Water was changed and new drug added every week. After 2 weeks of continuous treatment, worms were incubated repeatedly for 5 days in 800 μM PF-670462 and in fresh seawater for 2 days to avoid possible side effects. Treatment was always continuous during the FR-FM phase. PF-670462 is dissolved in water. Thus, control animals were cultured under the same conditions (same room, light cycle, moon cycle, water change, feeding), but not incubated in PF-670462.

Whole-Mount In Situ Hybridization

Platynereis WMISH was performed according to Tessmar-Raible et al. (2005), with the modifications for adult heads outlined in Backfisch et al. (2013).

Immunocytochemistry

Monoclonal anti-mouse anti-acetylated α-tubulin (clone no. 6-11B-1; Sigma-Aldrich. T6793) was used in a 1:200 dilution as previously described (Arendt et al., 2004).

Mounting and Microscopy

Platynereis adult heads were mounted in 90% glycerol. See Extended Experimental Procedures for details.

Worm Culture and Light Conditions

For circadian and circalunar experiments, worms were housed in separated shelving systems using LED lights at 8000K. Nocturnal light intensities mimicking “full moon” were between 1-10lux according to previously published conditions (Hauenschild, 1960). Animals were exposed to a circadian light regime of 16 hr light and 8 hr darkness. Daylight was switched on at ZT0 and switched of at ZT16. NM and FR-FM: no light from ZT16 to ZT24; FM: dim nocturnal light from ZT16 to ZT24.

Gene Identification

Fragments of Pdu-pdp1, Pdu-vrille, Pdu-timeless, Pdu-ck1δ/ε sequences and full fragments of Pdu-clock,Pdu-bmal, Pdu-tr-cry and Pdu-4-6-photolyase sequences were identified based on BAC and EST sequences. Pdu-clock: upper primer: 5′ TGTGGTCTATAATGGCTTGGATT 3′, lower primer: 5′-AAACGTCAGAACTGTGATTAGGC-3′ Tm: 53°C (full sequence).

Pdu-bmal: upper primer: 5′-ATGGGTCGGACTCCGATCGTTTCG 3′, lower primer: 5′ CTATAACGGCCAAGGAAAGTCAG-3′ Tm: 61°C (full sequence).

Pdu-tr-cry: upper primer: 5′-GGGTGGATAATCAGGCTTGTATTTCAAC 3′, lower primer: 5′ GTATCAATCGGTTTCTGTGACTC 3′ Tm: 58°C (full sequence).

Pdu-4-6-photolyase: upper primer: 5′ TTGCACAGCTTTGCAGAGAATCCAC 3′, lower primer: 5′ GATCTAATAACAACAATGAACACATAG 3′ Tm: 58°C (full sequence).

Pdu-pdp1: upper primer: 5′ ACATTATGACGTGTAAGTATTTATGC 3′, lower primer: 5′ ACACTCAATTCTGCAGGTACACC 3′ Tm: 55°C.

Pdu-vrille: upper primer: 5′ ATGGACCCCGCTTGCAGAACGAACAC 3′, lower primer: 5′CTCCGGCTCATCCGGGTGCTTC 5′ Tm: 63°C.

Pdu-timeless: upper primer: 5′ TTGGTCTACTTGGACGAGGAAGAAG 3′, lower primer: 5′ CCCTAGGCAGATTTATGCAAGTCTC 3′ Tm: 58°C.

Pdu-ck1δ/ε: upper primer: 5′ GAATTCCCGGGATGTAAATAATGGCG 3′, lower primer: 5′ TAGGCCTCGGAGTTGGTCATGACAC 3′ Tm: 61°C.

Fragments of all other genes were first identified by degenerated PCR, and (if necessary) expanded by RACE PCR using QIAGEN Hot Star Taq (203205). Primers were used at 3μM.

Degenerated PCR program:

15min 95°C /5 cycles with 1min 94°C, 2min Tm −5°C, 1min 94°C / 35 cycles with 1min 94°C, 2min Tm, 4min 72°C/ final elongation of 10min 72°C.

Pdu-L-cry: upper primer: 5′GAYGGNGARGTIGCIGGIAC 3′, lower primer: 5′ACCCACATCCARTTICCIGCRCAIAC 3′ Tm:49°C.

Pdu-timeout: upper primer: 5′ YTNHTNGTXAAYYTXACXCARCC 3′, lower primer: 5′ YTCNARCATYTTXAVRAAXAVRTG 3′ and nested upper primer: 5′ GAYGAYGCXWSXHTXCAYGAYCA 3′, Tm: 48°C; nested lower primer: 5′ NGTNADCATCATYTCRTARTA 3′, Tm: 41°C.

Pdu-period: upper primer: 5′ ATHYTNGGITAYCCIAARGAY 3′, lower primer: 5′ RTTNARYTGNARIGGISWISWRCA 3′ and nested upper primer: 5′ ATHYTNGGI TAYCCXAARGAY 3′, Tm: 45°C; nested lower primer: 5′ CRTTRTGNGCXCKCA TICKRTAIGG 3′, Tm: 44.3°C.

Pdu-period Race: The 3′-end of the coding region was obtained by rapid amplification of cDNA ends. 1st primer: 5′ ATCACCGTTCTTCGCCCTCAT 3′, 2nd (nested) primer: 5′ GAACGTCTGTCCAGGCATCT 3′, UPM of the Clontech kit was used as counter-primer.

Phylogenetic Analyses

NJ trees were generated using ClustalX software (Larkin et al., 2007), correcting for multiple substitutions and bootstrapped by 1000 repetitions. For ML analysis, consensus trees were calculated, using TREE-PUZZLE (Schmidt et al., 2002) (v5.3), from 100 trees generated with IQPNNI (Vinh and Von Haeseler, 2004) (v3.3.2) using the BLOSUM62 substitution matrix and 4 categories of gamma-distributed substitution rates.

Primers Used in RT-PCR

PCR primers were designed using the ProbeFinder v.2.45 (Roche). The amplification reactions (20μl) contained 2x SYBR Green Master Mix (Applied Biosystems), 0.5μM of each primer and 5μl of the cDNA template. Each sample was run in duplicate along with control reactions that did not include reverse transcriptase. Amplification thermal profile was for 10 min at 95°C (for 15 s at 95°C, for 1 min at 60°C) x 40 cycles.

Pdu-cdc5: upper primer: 5′ CCTATTGACATGGACGAAGATG 3′, lower primer: 5′ TTCCCTGTGTGTTCGCAAG 3′

Pdu-bmal: upper primer: 5′ tccgatttatctccacgagaa 3′, lower primer: 5′ tcc gtctttacaggcagc a 3′

Pdu-period: upper primer: 5′ GGTCAACATGAAGTCGTACAGG 3′, lower primer: 5′ CACTGGTTTTCGGCTCAA G 3′

Pdu-clock: upper primer: 5′ tgagcgtatgaaagtatacaatgttct 3′, lower primer: 5′ tgagctggttctcatccttttc 3′

Pdu-tr-cry: upper primer: 5′ TGTTCAAATTCCATGGGACA 3′, lower primer: 5′- TGT TTTAGCCTCAGCCCATT −3′

Pdu-L-cry: upper primer: 5′ AGGACCTTGATGACAGCTTGA 3′, lower primer: 5′ TCTTGTCACTCCCCATTCCT 3′

Pdu-pdp1: upper primer: 5′ GAAACTGCGTCTCGAAATGA 3′, lower primer: 5′ TCCTCCTTAACTTCCACAGTCTTC 3′

Pdu-vrille: upper primer: 5′ GAAAGAAACTTCTCCGCAAAGCA 3′, lower primer: 5′ CTCATCCGGGTGCTTCAT 3′

Pdu-timeless: upper primer: 5′ ATGTCGCAGGAGACGTCAA 3′, lower primer: 5′ AGACCTTCAAGATGGGCAAC 3′

Pdu-timeout: upper primer: 5′ TGATGCATGCTGTGAAGGAT 3′, lower primer: 5′ GCCCACAGGTAGTAGGTCTCA 3′

Mounting and Microscopy

Fluorescently stained animals were mounted in DABCO-glycerol. Pictures were taken on a Zeiss Axioplan2 microscope using 20X Plan-Neofluar objective (dry) or 40X Neofluar objective (oil immersion). Images were recorded with Zeiss AxioCam MRC 5 camera using AxioVison 4.8 software.