Marine organisms exhibit 12.4-hour rhythms of gene expression, physiology and behavior synchronized by tidal cues. The mechanism underlying these circatidal rhythms, and its overlap with the circadian clockwork, has remained elusive. However, recent studies showed that the core circadian gene BMAL1 sustains circatidal behavior in crustaceans. Therefore, we mutagenized the other three core circadian clock genes (
Organisms adjust their metabolism, physiology and behavior to predictable cycles in their surrounding environment using biological clocks [1][2][3][4] . Among them, circadian clocks are by far the best understood. They are entrained by the alternance of day and night and freerun with a ~24-hour (h) period when held in constant environmental conditions. The circadian clock is highly conserved across animals 1,5 . Its core is a negative transcriptional feedback loop comprised of the activators CLOCK (CLK) and BMAL1 (called CYCLE in Drosophila) that heterodimerize to bind to E-boxes and activate the transcription of numerous genes, including those encoding their own repressors PERIOD (PER, CRYPTOCRHOME (CRY) and/or TIMELESS (TIM), depending on the species. Other CLK/BMAL1 targets include transcription factors that form interlocked loops that modulate and strengthen the core loop [5][6][7][8][9][10][11] .
In marine environments, especially in the intertidal zone, organisms are subjected to drastic environmental changes caused by tides. In response, besides circadian rhythms, they also exhibit ~12. 4-h rhythms that synchronize to tidal cues 12,13 . The molecular mechanism behind these circatidal rhythms remain poorly characterized. Early behavioral studies of different intertidal organisms led to three main hypotheses. The first posited that circadian and circatidal rhythms are generated by distinct clocks 14 . The second proposed that circadian and circatidal rhythms are produced by a single clock, which can run with a period of ~24 h or ~12.4 h, depending on the rhythmic cues an animal is exposed to 15 . The last hypothesis suggested the existence of two circalunidian (~24.8-h) clocks running in antiphase, thus generating circatidal (~12.4-h) outputs 16 . This third hypothesis could involve essentially the same mechanism as the circadian clock, with a minor adjustment of the oscillator's period (from 24 to 24.8h). A similar minor period adjustment could also permit the circadian clock mechanism to generate 12.4-h period rhythms by exploiting its two antiphasic interlocked transcriptional feedback loops, or by driving neuronal activity in opposite phase in two different neuronal groups, as observed in Drosophila in Morning and Evening activity promoting neurons 17,18 .
Several studies have thus tested whether circadian clock genes are required for circatidal rhythms. Knock downs of core circadian clock genes via RNA interference (RNAi) were performed in the mangrove cricket, Apteronemobius asahinai, and in the marine isopod Eurydice pulchra. Results suggested that the core circadian genes Clk, Per and Cry2 are not required to sustain circatidal outputs [19][20][21][22] . However, pharmacological inhibition of Casein Kinase Iε, which phosphorylates PER proteins and thus regulates circadian rhythms in flies and mammals 23 , lengthened circatidal behavioral rhythms, suggesting that this kinase is shared between the circadian and circatidal clocks 21 . More recent studies further point towards a mechanistic overlap between the two clocks. Indeed, CRISPR-Cas9 mediated knock out of PhBmal1 in the amphipod Parhyale hawaiensis eliminated circatidal behavior 24 . Moreover, RNAi directed at Bmal1 reduced circatidal rhythmicity in E. pulchra 22 . Recently, distinct circadian and circatidal clocks cells were identified in the brain of P. hawaiensis and E. pulchra 25 . Results were particularly striking in P. hawaiensis, as circatidal (12.4-h) rhythms of expression for both PhPer and PhCry2 mRNAs were observed in four out of the ~60 cells expressing core circadian genes. Moreover, these cells synchronize with a tidal cue (i.e. vibration), but not to the light/dark cycle. By contrast, circadian cells displayed 24-h period rhythms of PhPer and PhCry2 mRNAs that synchronized to the LD cycle, but not to vibration. These data suggest that PhPer and PhCry2 may contribute to circatidal rhythms in P.
hawaiensis.
In the present study, we used CRISPR-Cas9 genome editing to test the degree of mechanistic overlap between the circadian and the circatidal clocks in P. hawaiensis by generating knock-out lines for PhPer, PhClk and PhCry2 to complement the existing PhBmal1 knock-out animals. Interestingly, we found that all four core clock genes are necessary to sustain circatidal rhythms, indicating considerable mechanistic overlap between the two clock mechanisms. Nonetheless, our results also suggest that the transcriptional wiring between the four core clock genes differ between circadian and circatidal cells.
Based on genomic studies, the circadian clock of P. hawaiensis is predicted to comprise homologs of CLK, BMAL1, PER and repressor-type cryptochromes 26 . The genome does not appear to encode a photoreceptive CRY or a TIM homolog. We previously demonstrated that BMAL1 is required for both circadian and circatidal behavioral rhythms. We therefore decided to determine whether the circadian and circatidal clocks share additional common building blocks.
We first targeted the PhCry2 gene. Two knock-out lines were generated using CRISPR-Cas9 and a set of two guide RNAs (gRNA) targeting the photolyase domain (Figure 1A), which is required for cryptochromes' repressor function 27 . The first mutant line (PhCry2-mutation 1) harbors two deletions of 18 and 20 base pairs (bp), respectively, which cause a premature stop codon that truncates the PhCRY2 protein from 929 amino acids (AA) to 422 AA. The second line (PhCry2-mutation 2) harbors a single deletion of 16 bp, also causing a premature termination codon that truncates PhCRY2 to 428 amino acids.
To verify that the truncation affects PhCRY2 activity, we turned to a transcriptional assay in mammalian HEK-293T cells (Figure 1B). We previously showed that PhBMAL1 can activate E-bx mediated transcription with the help of mCLK 24 (there is still uncertainty on the exact sequence of PhCLK, see below). We thus tested if PhPER and PhCRY2 can repress PhBMAL1/mCLK, as observed in other animals 21,[28][29][30][31][32] . Indeed, we found that expression of both PhPER and PhCRY2 repressed PhBMAL1/mCLK activity in a dose-dependent manner. Repression was also observed with PhCRY2 alone, but it was not as efficient as with PhPER (Two-way ANOVA, *p=0.01). In addition, PhPER did not repress on its own. Thus, as in other animals, CRY2 appears to be the primary repressor, while PER plays a modulatory function 29,30,32,43 . No repression was observed with expression of the truncated PhCRY2 -/-, which is thus severely defective.
To assess whether the loss of PhCRY2 disrupts the circadian clock as expected, we measured the swimming activity of mutants and control animals under constant conditions after entrainment to a 12:12 light-dark (LD) cycle. We prioritized observing free running behavior under constant light (LL), as we previously observed that circadian behavioral rhythms are more robust under these conditions than under constant darkness (DD) 24 . This was also the case in the present study (Figure 1C and S1A), although rhythms were weaker than previously reported under both conditions. As expected, under LL, average activity traces of wild-type (WT) animals (PhCry2 +/+) showed lower activity during the subjective light phase and higher activity during the subjective dark phase (Figure 1C and S2A). The heterozygous animals (PhCry2 +/-) exhibited a similar pattern. In contrast, the average swimming activity of the homozygous animals (PhCry2 -/-) showed no obvious rhythmicity. This was reflected in the proportion of rhythmic animals (Figure 1D) and in the rhythm power (amplitude) which was quite weak in the rare rhythmic knockout animals (Figure 1E). We next determined the period of behavioral rhythms using periodogram analysis. After LDonly entrainment, most rhythmic animals exhibit statistically significant rhythm periodicities in the circadian (~24-h) range ("doublet"), but harmonics in the ~12-h range can also be observed ("singlets") 24 . The average circadian doublet period of WT animals was around 25.5 h as previously described under LL, since continuous light exposure lengthens circadian rhythms in P. hawaiensis 24 . While the average circadian period did not statistically differ between wild-type, heterozygous and the rare rhythmic homozygous mutants, individual periodicities were not clustered around 25.5 h in mutant animals, as in wild-type and heterozygous animals (Figure 1F). Actually, out of the four rhythmic PhCry2 -/-, only one had a period within the expected range. The singlet period of the homozygous was also very variable, and statistically significantly longer than WT (Figure 1G).
To further establish that PhCRY2 controls circadian rhythms, we tested LD-only entrained animals in constant darkness (DD). Again, circadian behavior was severely disrupted (Figure S1). Only two knockout animals out of 32 were rhythmic and their period was not in the expected range (Figure 1H to 1K). Therefore, our results show that in the absence of functional PhCRY2, circadian behavior is severely compromised. As previously observed with PhBmal1-/-mutant animals, residual rhythms of low amplitudes and poorly defined period emerge in a few animals when the circadian clock is disrupted 24 .
In Drosophila, ca. 240 neurons rhythmically express circadian clock genes 33 . These neurons drive different 24-h metabolic and behavioral rhythms, including the timing of the sleep/wake cycle. In addition, numerous glial cells express core clock genes [34][35][36] . In the Parhyale brain, only ~ 60 cells detectably express PhPer and PhCry2 25 . To assess the effect of the PhCry2 knock-out on the circadian clock at the molecular level, we turned to HCR-FISH to monitor PhPer and PhCry2 mRNA rhythms in putative circadian neurons, the medioposterior (MP) cells 25 . These cells were found to entrain to the LD cycle, showing 24-h mRNA abundance rhythms after entrainment to both a 12:12 LD cycle and a 12.4-h vibration cycle that mimics tides 25 . Using probes to the pan-neuronal marker elav, we established that most MP cells indeed are neurons, although 1/10 cell per hemisphere appeared elavnegative (Figure S3A). Since we use immersion cycles instead of vibration for tidal entrainment, we first verified that the MP neurons exhibit circadian mRNA rhythms in our hands. As expected, we observed that both PhCry2 and PhPer showed circadian expression in constant condition after entrainment to both LD and tides in WT (Figure S4A and B). As surprisingly observed by Oliphant et al., PhCry2 and PhPer mRNAs also oscillate with an opposite phase after exposure to our entrainment conditions. While the PhPer expression pattern was clearly circadian, PhCry2 showed a statistically significant 12-h period rhythm according to JTK cycle analysis, although the secondary peak was separated from the main peak by about 8 h. This is consistent with Oliphant et al., who could also observe a secondary PhCry2 mRNA peak when varying the phase relationship between the LD and vibration cycles
. In PhCry2 -/-animals, PhCry2 mRNA levels remained constantly low (Figures 1L and M).
PhPer expression was also significantly disrupted as mRNA levels were significantly higher than in WT, except at CT18. A weak, but significant decrease in PhPer2 mRNA level was observed at CT12 and 18, but the phase of this weak apparent rhythm was completely distinct from that of the robust rhythms observed in WT animals. A similar progressive decrease in mRNA level was observed in mCry1 -/-mCry2 -/-double-mutant mice, perhaps as a result of prior mPer1 and mPer2 photic induction 37 . Taken together, our results show that PhCRY2 is indeed part of the molecular circadian clock and is thus necessary for circadian behavior.
PhCry2 is an essential element of the circatidal clock Next, we determined the role of PhCRY2 in the circatidal clock. We entrained animals to both LD and tidal (10.3 h high tide: 2.1 h low tide) cycles and monitored their behavior under DD and constant high tide (HT) conditions. As expected, WT and heterozygous animals showed strong rhythms of activity following the expected tides (Figure 2A and S2B). With both genotypes, low activity was observed at subjective low tide and activity peaked during the subjective high tide. In striking contrast, the homozygous animals did not show any rhythms of activity. All PhCry2 -/-animal scored as arrhythmic (Figure 2B). A slight gene dosage effect was observed in heterozygous animals compared to WT, as the period of their circatidal behavior was significantly longer (Figure 2C, D and E). These results show that PhCRY2 is necessary to generate circatidal behavior.
To assess the implication of PhCry2 on molecular circatidal rhythms, we returned to HCR-FISH. The expression of PhCry2 and PhPer was this time measured in the dorsal-lateral (DL) cells, which we found to be neurons as well (Figure S3B). The DL neurons entrain to vibration, exhibit ca. 12-h mRNA rhythms and are thus candidate circatidal oscillators 25 . In WT, both genes showed robust 12-h rhythm of expression with low expression around the expected low tide (Figure S4C and S4D). In PhCry2 -/-, as observed in circadian neurons, the expression of PhCry2 remained constantly low (Figure 2F and G). Also, similarly to circadian neurons, PhPer expression was high and decreased over time, but did not show any sign of residual circatidal expression. In summary, our behavioral and molecular results demonstrate that PhCRY2 is an essential circatidal protein and is thus shared between the circadian and the circatidal systems.
PhPER contains two Per-ARNt-Sim (PAS) domains and a Per-ARNt-Carboxy-terminal (PAC) domain. Three knock-out mutant lines were generated using CRISPR-Cas9 and a set of two gRNAs targeting an exon preceding those encoding the key functional domains (Figure 3A). The first mutant line (PhPer -mutation 1) harbored an 8 bp deletion combined with 2 added bp at the site targeted by the 5' gRNA, and a 25-bp deletion with 4 added bp at the site of the 3' gRNA. These indels thus caused a premature stop codon truncating PER from 1258 AA to 333 AA, and eliminating all key functional domains. The second mutant line (PhPer -mutation 2) had a premature stop codon following the deletion of 10 bp and the addition of 5 bp at the 5' gRNA site, plus the deletion of 2 bp at the 3' gRNA site, thus truncating PhPER to 317 AA. The last mutant line (PhPer -mutation 3), similarly to the others, had a premature stop codon and a truncated protein of 355 AA after a deletion of 9 bp with 3 extra bp at 5' gRNA site, and the deletion of 7 other bp at the 3' gRNA site. The absence of functional domains is predictive of a complete loss of function for all three mutations. Again, we first measured circadian behavior in LL. Both WT (PhPer +/+) and heterozygous animals (PhPer +/-) showed rhythmic behavior at the population level with higher activity during the subjective dark phase (Figure 3B and S5A). The homozygous mutant (PhPer -/-) population showed no clear rhythmic activity pattern. Accordingly, the percentage of rhythmic mutant animals was significantly lower than in PhPer +/+ and PhPer +/-populations (Figure 3C). Rhythm power of mutant animals was very low, but significance was not reached compared to wild-type and heterozygous, with only two animals showing rhythmicity (Figure 3D). Out of the two rhythmic PhPer -/-animals, only one had a period within the expected range (Figures 3E and F). When recorded in DD, homozygous animals were significantly less rhythmic than the WT with only four rhythmic animals out of 72 (Figures 3G to J and S6). Altogether, these results show that PhPer is critical to sustain circadian behavioral rhythms.
We then determined the effect of PhPer knock-out on the molecular circadian clock, with HCR-FISH staining on mutant animals entrained to LD and tidal cycles. As expected, both PhCry2 and PhPer rhythms were disrupted in the circadian MP neurons, significantly decreasing over time instead of showing peaks of expression in the evening or night for PhCry2 and PhPer, respectively (Figures 3K and L), reminiscent of the decay observed for PhPer mRNA in PhCry2 -/-animals. As mentioned above, this could be because mRNA expression responds directly to light, and thus decreases progressively in DD. Altogether, these data show that loss of PhPER disrupts both molecular and behavioral circadian rhythms, as expected.
We then determined the role of PhPer in the circatidal clock. After entrainment to both LD cycle and tides (last high tide [LHT] 13:20), both WT and heterozygous animals showed lower average activity around the subjective low tides and higher activity around subjective high tides (Figure 4A and S5B). The average activity of animals lacking PhPer also showed this rhythmic pattern for 3 cycles, but it was much noisier and rhythm amplitude decreased rapidly. To determine if this residual rhythmicity was circatidal in nature, animals were entrained to a 6h-shifted tidal cycle (LHT 19h20) (Figure 4B and S5C). Again, PhPer mutants showed lower activity around the subjective low tides for 2, perhaps 3 cycles. Importantly, the phase of these residual rhythms was shifted by 6 hours, as in WT animals, confirming that they are entrained by tides (Figure 4C). However, the proportion of animals scoring as rhythmic was much lower in PhPer -/-mutants compared to WT and heterozygous animals (Figure 4D). For the few rhythmic mutant animals, power and period did not significantly differ from WT and heterozygotes ( Figures 4E, F and G). These results show that PhPER sustains circatidal behavioral rhythms, although it might not be as critical as PhCRY2.
We then determined the impact of the loss of PhPER on molecular circatidal rhythms in circatidal DL neurons. Neither PhPer nor PhCry2 mRNAs showed statistically significant time-dependent variations in abundance (Figures 4H and I). Circatidal molecular rhythms are therefore clearly disrupted. However, we noticed slightly higher levels of both circadian mRNAs at the time at which expression is high in WT animals. Weak residual molecular rhythms might thus be present in PhPER-/-animals, which could explain the weak circatidal behavior rhythms that were observed. Overall, however, our data show that PhPer is part of the circatidal clock mechanism, although PhPer mutant animals can maintain weak circatidal behavioral rhythms in its absence, contrary to PhCry2 -/-animals.
PhClk is also required for circadian rhythms CLK proteins are composed of a Beta-Helix-Loop-Helix (BHLH) domain required to bind to the E-Box and two PAS domains needed to bind to BMAL1 (Figure 5A). A previous study failed to identify CLK's PAS-A coding sequences in the P. hawainesis genome or transcriptome. However, we did find two putative exons containing PAS-A coding sequences (Figure S7), indicating that PhCLK contains all three canonical domains. However, in the absence of transcriptomics support, we still cannot predict the full PhCLK sequence. To knock-out PhClk using CRISPR-Cas9, two gRNA were designed targeting the bHLH domain. Two mutant lines lacking all functional domains due to an early stop-codon in the BHLH domain were generated. The first line (PhClk -mutation 1) had a deletion of 27 bp with 4 extra bp inserted, generating a frameshift. The predicted truncated protein was 61 AA long instead of 2065 AA. The second line (PhClk -mutation 2) had a deletion of 80 nucleotides and generated truncated 42 AA protein.
Both WT and heterozygous animals showed the expected circadian behavior pattern after LD-only entrainment, with higher activity during the expected dark phase (Figure 5B and S8A). By contrast, PhClk -/-animals did not show a rhythmic behavior pattern at the population level. Also, we observed fewer rhythmic animals in PhClk -/-compared to WT and heterozygous animals, although the difference did not reach significance (Figure 5C). Similarly, rhythm power was lower in PhClk -/-animals, though not significantly so (Figure 5D). Importantly, periods in the few weakly rhythmic animals were broadly distributed rather than clustered around 25.5 h, as previously observed in weakly rhythmic PhBmal1 mutants 24 (Figures 5E and F). Circadian behavior thus appears to be disrupted. Moreover, PhClk was clearly required for circadian behavior under DD, as none of the knock-out animals were rhythmic under these conditions (Figures 5G to J and S9). Furthermore, HCR-FISH revealed that the molecular circadian clock is severely disrupted in the MP neurons of PhClk mutants. Neither PhPer nor PhCry2 mRNA levels oscillated (Figures, 5K and L), with extremely low expression level of PhCry2. In summary, PhClk is required for the functioning of the circadian clock in P. hawaiensis and ensures proper circadian swimming activity.
We also probed the impact of loss of PhCLK on circatidal rhythms at the behavioral and molecular levels. In constant conditions following entrainment to both LD cycles and tides, both PhClk +/+ and PhClk +/-showed robust rhythmic behavior (Figures 6A and B and S8B). Animals lacking PhClk displayed a weakly rhythmic pattern for about 2 circatidal cycles that appeared entrained to tides, reminiscent of our observations with PhPer mutants. The proportion of rhythmic animals and the power of the rhythm were significantly lower in PhClk -/-compared to the other genotypes (Figures 6C and D). No significant difference was observed between genotypes for the singlet period (Figure 6E), although the doublet period of PhClk -/-was significantly different than WT and heterozygous animals (Figure 6F). Circadian phase was only quantified for the LHT 13h20 as only one PhClk -/-individual was rhythmic with the shifted tidal cycle (Figure 6G). No significant difference of phase was observed between mutants and WT, suggesting tidal entrainment of the weak residual rhythms. These results indicate that PhClk is necessary to sustain circatidal swimming activity.
At the molecular level, both PhPer and PhCry2 were detected in circatidal DL neurons of PhClk mutants (Figure 6H and I) but showed no time-dependent changes in expression.
PhPer was at a constant high level compared to WT, while PhCry2 was constantly low. This shows that PhClk is also required for the functioning of the circatidal molecular clock.
PhBmal1 also controls circadian and circatidal expression, but PhBMAl1 and PhCLK functions differ in circatidal neurons Our previous work showed that PhBmal1 is required to maintain both circadian and circatidal activity in P. hawaiensis, but its impact on the molecular circadian and circatidal clock was not tested 24 . We thus performed FISH-HCR on PhBmal1 knock-out animals. In the circadian MP neurons, the circadian expression of PhPer and PhCry2 was deeply disrupted (Figures 7A and B). The expression of PhPer decreased over time whereas PhCry2 expression was low at all time points. Nevertheless, PhCry2 showed a weak rhythm with a completely abnormal phase. We therefore conclude that PhBMAL1 is necessary for proper circadian molecular rhythms. In the circatidal DL neurons, neither repressor genes showed circatidal expression (Figures 7C and D). This confirms that BMAL1 is part of the circadian and the circatidal clocks systems in P. hawaiensis.
As expected, since BMAL1 is a transcriptional activator, PhPer and PhCry2 mRNAs remained constantly at a low range in both circadian and circatidal neurons, never reaching WT peak expression. This was also the case in brain circadian neurons of PhClk mutants (Figure 5K and L), as expected since CLK and BMAl1 work as dimers. In stark contrast, while PhCry2 mRNA was as expected also low, PhPer mRNA was constantly high in circatidal neurons of PhClk mutant animals (Figure 6H and I). PhCLK thus represses, rather than activate, PhPer expression specifically in circatidal neurons. This reveals a surprising difference in the transcriptional wiring of core clock genes between circadian and circatidal neurons, and a novel PhBMAL1-independent function for PhCLK (Figure 7E).
We previously showed that PhBmal1 is necessary for circatidal behavior in P. hawaiensis, in addition to its canonical role in the circadian clock 24 . With the present study, we find that there is considerable overlap between circadian and circatidal clock mechanisms. Using CRISPR-Cas9 genome editing, we demonstrate that the three other core circadian genes are required to sustain circatidal behavior. This is in stark contrast to previous studies in E. pulchra and A. asahinai [19][20][21][22] . It is possible that the absence of RNAi phenotypes when targeting Per, Clk and Cry2 is the result of incomplete suppression of gene expression. In support of this idea, RNAi to EpBmal1 decreased the percentage of rhythmic animals, but this phenotype was much weaker than the complete disruption observed in PhBmal1 knock-out animals 24 . Another possibility is that circatidal timekeeping mechanisms differ across marine organisms, resulting from convergent evolution or divergent evolution of a common proto-circatidal clock. The only previous study made at the cellular level gives some support to these possibilities: in putative circatidal cells of E. pulchra, only EpTim showed circatidal expression, whereas EpPer and EpCry2 were either arrhythmic or displaying a circadian expression pattern 25 . In contrast, in P. hawaiensis, which lacks Tim, both PhPer and PhCry2 showed circatidal expression. Also, a recent study in which a CRISPR-Cas9 Per knockout was generated in A. asahinai concluded that PER plays no role in circatidal rhythms in this insect. However, this conclusion was based on a single rhythmic animal out of 10 monitored, and no molecular rhythms were measured 38 .
In P. hawaiensis, we found that loss of both PhCRY2 and PhBMAL1 completely disrupted molecular and behavioral circatidal rhythms 24 . On the other hand, in both PhPer -/-and PhClk -/-animals, residual circatidal behavioral rhythms are observed, even though molecular rhythms are severely disrupted. This demonstrates that PhPER and PhCLK are important cogs of the circatidal clocks, although we surmise that residual molecular rhythmicity is present in their absence, at least in the few animals that exhibited circatidal behavior. It is possible that another bHLH/PAS transcription factor can weakly substitute for PhCLK. PhHIFα, in particular, comes to mind, since in mammals it can dimerize with BMAL1 and has been implicated in the entrainment of circadian clock by O 2 levels [39][40][41][42] . Weak rhythms might be possible in the absence of PhPER, because repressive cryptochromes are the dominant circadian repressors in most species, while PER have supporting functions, such as promoting CRY nuclear entry 29,43 . This was confirmed by our luciferase assay, where PhCRY2 alone is able to repress PhBMAL1:mCLK transactivation, while PER was unable to do so.
A recent study revealed that while PhPer and PhCry2 mRNAs cycle in phase inside circatidal neurons, their oscillations are unexpectedly antiphasic in circadian neurons 25 . We confirmed these observations under our tidal entrainment protocol. This suggests that the transcriptional network linking different clock genes is wired differently in circadian and circatidal neurons. We therefore examined closely the impact of core clock gene mutations on PhPer and PhCry2 mRNA levels in both circadian and circatidal neurons. In both neuron types, the loss of PhPer led to moderately high mRNA levels of PhCry2, and the loss of PhCry2 to moderately high PhPer mRNA levels. This is consistent with a repressive role for both CRY2 and PER. Note that the loss of core circadian repressors does not result in peak mRNA levels in Drosophila, because interlocked feedback loops are also disrupted by the loss of core clock repressors, thus reducing DmCLK levels 44,45 . Low PhPer mRNA levels were observed in PhPer -/-mutants, and low PhCry2 mRNA levels in PhCry2 knock-outs. Since our mutations create premature stop codons, these low mRNA levels are probably the result of nonsense-mediated mRNA decay 46 .
Our FISH-HCR and cell culture assays thus support a rather standard repressive role for PhPER and PhCRY2 in both circadian and circatidal clocks. They also support the expected activator role of PhBmal1, since PhCry2 and PhPer mRNA levels are low in both clock neuron types of PhBmal1 mutants. In circadian neurons, the loss of PhClk also led to low mRNA levels, supporting its canonical role as a circadian transcriptional activator along with BMAL1. However, PhCry2 levels were particularly low in MP neurons, essentially undetectable. This was not linked to a technical issue, since we detected PhPer mRNAs in these circadian neurons and PhCry2 mRNAs in the circatidal DL neurons of the same brains. This suggests that PhBMAL1 might be able work with a different partner than PhCLK to weakly activate and therefore maintain constant low transcript levels of PhPer, but is unable to do so for PhCry2 in circadian MP neurons. As mentioned above, PhHIF1α could substitute for PhCLK. Also, HIF1α has been hypothesized to bind to the non-canonical EBox2 of mPeriod2 47,48 . Therefore, we propose that PhPer and PhCry2 genes are activated by different sets of transcription factors in circadian neurons (Figure 7E). This could contribute to the phase difference in expression of the two repressors. Other possibilities include differences in mRNA stability, and PhCLK/PhBMAl1 controlling an intermediate positive PhCry2 regulator (Figure 7E), which would work similarly to RORα or PDP1ε in promoting antiphasic expression of mammalian Bmal1 and DmClk, respectively 6,8,9,49,50 .
In circatidal neurons missing PhClk, PhPer levels were constantly at peak levels while PhCry2 levels were constantly low. This is rather unexpected. It indicates that PhClk represses PhPer expression independently of PhBmal1 (Figure 7E). This is a striking deviation from its regular function in the circadian clock mechanism. It will be interesting to determine whether PhCLk represses PhPer as a homodimer, or with a novel partner, and whether its impact on PhPer expression in circatidal neurons is direct or mediated by an intermediate repressor. Critically, PhCLK represses PhPer expression specifically in circatidal neurons.
In conclusion, transcriptional control within the core circadian and circatidal clocks is wired differently (Figure 7E). However, based on the surprisingly antiphasic oscillations of PhPer and PhCry2 mRNAs in circadian neurons, one would have expected the major deviation from the canonical function of core clock genes to occur in these neurons, rather than in circatidal neurons. Obviously, deeper studies will be necessary to resolve this apparent paradox. However, the identification of a novel function for PhClk, specific to circatidal neurons, could be critical to understanding how circadian and circatidal clocks oscillate with such different periods. Among the three historical models for circatidal rhythms, our observations, combined with those of Oliphant et al. 25 , support the model proposed by Naylor 14 , of separate circatidal and circadian clocks, even though they use the same core clock genes. However, because of their overlapping mechanisms, we cannot exclude entirely the Enright model of a plastic clock 15 . It is possible that during development, clock neurons adopt different transcriptional wiring based on the sensory neurons they are coupled with, and thus the environmental input they receive: tidal or diurnal. We note that the period of core clock gene expression changes as a function of the presence of tides in oysters, but the mechanism underlying this plasticity is not known 51 . We also note that the same species of freshwater snail can adopt more circadian or circatidal phenotypes (behaviorally and molecularly), depending on whether the animals live within or outside of the tidal zone of the Kiso River in Japan 52,53 . In P. hawaiensis, it will be very interesting to determine what dictates the periodic fate of clock neurons, and whether this fate could depend on environmental factors. Regardless, our results reveal that in P. hawaiensis, the evolutionary solution to the conundrum of simultaneously tracking tidal and diurnal environmental oscillations that are almost, but not exactly, harmonics, is to use the same set of core clock genes in two spatially distinct neuronal clocks, each having different autoregulatory feedback loops to set their specific periodicity. This spatial separation of the circadian and circatidal clocks likely underlies P. hawaiensis ability to integrate diurnal and tidal cues, rendering it capable of adapting its behavior and physiology to the different phase relationships of the two environmental cycles. It also likely permits plastic behavioral responses to different tidal regimens. Indeed, P. hawaiensis lives in a broad range of locations that experience different tidal patterns 54 : semi-diurnal (the most common 12.4h tidal cycle), mixed (two tides per 24.8h, of unequal duration and amplitude) and diurnal (24.8-h tidal cycle). This work improves our understanding of the complex mechanisms by which an organism copes with complex environmental periodicities. The extent to which this mechanism applies across marine organisms remains to be determined. Given the conservation of the circadian clock, these rules could prove universal.
The Chicago F-strain Parhyale hawaiensis strain was used for our study. The animals were maintained in Pyrex aquariums with lids containing 30 psu artificial sea water (ASW) and ~1 cm of crushed coral substrate. The animals were reared at a temperature of 25°C under 12:12 h LD cycles with ZT0 occurring at 08:00 in the morning. Wild type (WT) and PhBmal1 mutants 24 were fed baby carrots. PhCry2, PhPer and PhClk mutants were fed with a mix of ground TetraMin PLUS Tropical Flakes (Tetra, Melle, Germany), ground OSI Spirulina Flake Fish Food (Ocean Star International, Snowville, UT), ground Hikari Wheat Germ Floating Pellets for Pets (Hikari, Japan), ground Tubifex (Hagen Group, Quebec, Canada), and Kelp granules (Starwest Botanicals, Sacramento, CA). The dry food was mixed in 50 mL falcon tubes containing ASW supplemented with 50 µL of American Marine Selcon Vitamin Supplement (American Marine Inc., Ridgefield, CT) and 100 µl of KENT Marine Zoe Marine Vitamin (KENT Marine, Franklin, WI).
Mating pairs were isolated a day prior to embryo removal. Females were then lightly anesthetized using 0.02% clove oil in ASW. The collection of single-cell fertilized eggs was achieved as previously described 24 . Fertilized eggs were stored in a tissue culture dish containing filtered ASW water prior to injection. Approximately 40-60 picoliters of injection mixture was injected into one-cell embryos or both cells of two-cell embryos, as previously described 55 . The injection mixture contained 100 ng/µL or 162 ng/µL of each chemically protected guide RNA, 333 ng/µL of NLS-Cas9 protein (Synthego Co., Redwood City, CA), and 0.05% phenol red (Sigma-Aldrich). Both injection mixture concentrations generated knockouts. For each gene of interest, two guide RNAs were designed to target a functional domain or a region prior to a key functional domain. Injected embryos were then transferred to 60 mm culture dishes filled with filter-sterilized ASW and incubated at 28°C under a 12:12 h LD cycles. Embryos were transferred on a daily basis to a new culture dish containing ASW. After hatching, juvenile animals were reared individually in a custom multi-compartment insert placed inside a lidded plastic container filled with ASW. Each compartment had a mesh-covered opening that allowed water exchange with the shared bath while preventing mixing and aggression/cannibalism between individuals. Once injected animals reached sexual maturity, they were paired with other injected or wild-type individuals of the opposite sex to generate heterozygous offspring (F1) with fixed germline mutations. Heterozygotes F1 were then reared in a separate multi-compartment tank until sexual maturity. After the screening of mutations (see below), F1 males having targeted genotype were individually selected and transferred individually to a larger aquarium (20x10x10 cm) with at least 4 wild-type females. These tanks contained ~2 cm of crushed coral substrate and were fed with the food mix described above.
For the screening of mutations, adult animals were anesthetized using clove oil in ASW, after which two to three pleopods were removed and digested in 250 µL of lysis buffer (100 mM Tris, 5 mM EDTA, 200 mM NaCl, 0.2% SDS) containing 100 µg/mL of proteinase K at 50°C for 48 h. The tubes were centrifuged at 13000 rpm at 4°C for 10 min to pellet the debris. DNA was precipitated by adding 250 µL of isopropanol to the supernatant followed by centrifugation for 3 minutes at room temperature. The pellet was then washed with 500 µL of 70% ethanol, dried on a heat block at 55°Cand resuspended in 20 µL of milliQ water. PCRs were performed using Standard Taq DNA polymerase (New England BioLabs) as indicated by the manufacturer. PCR primers are listed in Table S1. The amplification program consisted of 35 cycles of denaturation at 95°C for 30 s, 30s annealing (54°C for PhCry2 primers, and 56°C for PhPer PhClk primers) and elongation at 68°C for 20 s. PCR products were run on a 3% agarose gel and the genotype was confirmed by gel purification of the products followed by sanger sequencing.
Tidal entrainment was performed as previously described 24 with cycles of 10.3 hr of high tide and 2.1 hr of low tide. Low tide was simulated when water was transferred to a reservoir using a draining pump. For high tides, water was pumped back into the aquarium containing the animals. Animals were entrained to tides for a minimum period of 30 days prior to the recording of behavior. After at least five days of undisturbed entrainment (i.e. no water changes or feeding), males and females were individually loaded into polystyrene vials containing ~2 mm of crushed coral substrate and 30 psu ASW. The water height was adjusted to equal their original aquaria level, thereby maintaining a constant high tide. ParaFilm was used to close the vial and limit evaporation. Drosophila Activity Monitors (DAM) (LAM25, TriKinetics, Waltham MA) placed in a temperature and light controlled incubator (I36LL, Percival Scientific, Iowa) were used for behavioral recordings. Following a day of acclimation, swimming and roaming activity were recorded for a duration of five days with recordings conducted at one-minute intervals. After LD-only entrainment, behavior was recorded under either constant light (LL) or constant darkness conditions (DD). The behavior after tidal and LD entrainment was recorded in DD. A minimum of three sets of experiments were conducted for each condition and for each genotype.
For most behavioral studies, the mutant colonies were of mixed genotypes, containing wild type (WT), heterozygous and homozygous mutant animals. Their genotypes were determined after recording as described above for mutant screening. For DD experiments and FISH-HCR experiments (see below), mutant colonies containing almost exclusively homozygous animals were used. These animals were nonetheless genotyped after behavioral monitoring or prior to brain dissection.
The recorded one-minute beam-crossing was averaged over 30-minute bins using DAMFileScan114 software (TriKinetics, Waltham MA). The rhythm analysis was performed with the FaasX software (https://neuropsi.cnrs.fr/en/cnn-home/francois-rouyer/faas-software/, courtesy of F. Rouyer, Centre National de la Recherche Scientifique, Gif-sur-Yvette, France) 56 . Autocorrelation and χ 2 periodogram analysis ("power" ≥ 10 and "width" ≥ 1h) were used to determine whether an animal was rhythmic. The filter for high frequencies was on. As previously described, both "vertical swimming" and "roaming" behaviors were monitored 24 . Animals were scored as rhythmic if vertical swimming and/or roaming were significantly rhythmic. For each animal, the period and rhythm power of the most robustly rhythmic of these two behaviors was used for subsequent analysis. For the population activity plots, behavior values were normalized for each animal by dividing each 30 minutes bin by its mean activity. Plots were generated using GraphPad Prism 10. Statistical comparisons of the power and the period were made using the same software. Outliers were identified and removed using a ROUT test with Q=1%. The normality was assessed using Shapiro-Wilk test. If the normality was respected, one-way ANOVA test was used followed by Tukey's for multiple comparisons. Otherwise, the Kruskal-Wallis test was used, followed by Dunn's multiple comparisons test. In instances where the comparison was between two groups, normal dataset was analyzed using Welch's test for normal distribution. Otherwise, Mann-Whiney test was used. Statistical significance is demonstrated as ***p < 0.001; **p < 0.01; *p < 0.05.
Animals were entrained to tides under LD cycles as described above. After LHT 12h10, five animals per sampling point were loaded separately in vials in the DAM system as described before (their behaviors are shown on figure S10). The animals were in DD after the light was switched off at 20h00. The following day, the sampling process was initiated at 8h00. First WT animals were sampled every two hours over 24 hours (13 sampling points). For the mutants, four sampling points spaced by six hours were chosen based on WT results. The following strains were used: PhCry2 mutation 1, PhPer mutations 1 and 3 and mutation 1 for PhClk. For the brain sampling, the animals were sacrificed by immersion in 4°C ASW for 20 minutes on ice. Heads were removed and placed in 4% PFA prepared in 3X PBS for 4 hours with rotation at 4°C. The heads were then washed with PBS-Tween 0.1% for 5 minutes three times. Brains were dissected in PBST using sharpened forceps. The brains were then dehydrated using a methanol dilution series (50%, 70% and 90% methanol in PBST) for five minutes each wash, followed by two brief washes in 100% methanol. Then brains were stored for a minimum of 24 hours at -20°C before staining.
HCR TM RNA-FISH Prior to staining, the brains were rehydrated using serial dilutions of methanol (5 min in 70%, 50% and 25%) followed by 1X10min and 2X5min washes in PBST. They were then incubated overnight in 300 µL of detergent solution 57 (50mM Tris, 1mM EDTA, 150mM NaCl, 0.5% Tween, 1% SDS) at 4°C. The following day, the brains were pre-hybridized in 200µl hybridization buffer (Molecular Instruments Inc) at 37°C for 3 hours. The brains were then incubated in 100 µl of hybridization buffer containing 20nM PhCry2, PhPer and PhClk probes at 37°C. To determine neuronal identity, cells were co-stained using PhPer and PhElav probes (Figure S3). HCR TM probes V3.0 were designed and synthesized by Molecular Instruments Inc, USA. After 48 h of incubation, the brains were washed 1x5min, 1x10min and 3x15min with 500µL of wash buffer (Molecular Instruments Inc) at 37°C and then 3X5min in 5XSSC supplemented by 0.1% Tween-20 at room temperature (RT). Pre-amplification consisted of 2x15min washes in 500µL of amplification buffer (Molecular Instruments Inc) at RT. Fluorescent hairpins were snap-cooled and added to 100µl of amplification buffer at a concentration of 60 nM. The brains were incubated overnight in the hairpin preparation at RT. Washes consisted of 2X5min, 2X30min and 1X5min in 500µL of 5XSSCT. The brains were then incubated in 50% glycerol in PBS containing 2µg/ml of DAPI for 1 hour at RT followed by 20 minutes in 70% glycerol. Brains were mounted in VectaShield Antifade Media (Vector Laboratories, USA) between two coverslips spaced using a Grace Bio-Labs SecureSeal TM imaging spacer. Images were acquired with a LSM 900 Zeiss confocal microscope at 1µm z-stack intervals.
Image analysis was conducted using ImageJ software. Prior to quantification, the images were smoothed and each cell group was isolated. The number of FISH dots, which reflect the expression of the gene, was counted using the RS-FISH ImageJ plugin 58 . The normality of the dataset was assessed using a Shapiro-Wilk test. If a normal distribution was observed, a oneway ANOVA was used to determine if significant variations occurred though time, otherwise a Kruskal-Wallis test was used. When a statistically significant effect was observed, rhythm analyses were conducted in R (v.4.4.1) 59 using JTK CYCLE in the package MetaCycle 60 for 24 h periods. P-values were adjusted using the Bonferroni method. For 12-h rhythms, the statistical difference between peaks and trough were calculated using post-hoc tests.
The TK-E54 plasmid 61 harboring 3 E-Boxes driving the expression of the firefly luciferase reporter, the CMV β-galactosidase plasmid (Clontech: PT2004-5) 28 containing the LacZ gene under the control of the CMV promoter and the mClk 62 cDNA under the CMV promoter in pcDNA3.1 were all provided by D. Weaver (University of Massachusetts Chan Medical School). The PhBmal1 plasmid has been previously described 24 . PhPer and PhCry2 were produced by cutting mClk plasmid using Xhol and EcoRI. gBlocks (Twist Biosciences) were designed to contain a start codon, a Flag tag (PhPer) or V5 tag (PhCry2) and the whole PhPer or PhCry2 sequences flanked by Xhol and EcoRI. Sequences used were based on Hunt et al. in silico reconstruction 26 . PhCry2-/-plasmid was generated by integrating an early stop-codon into PhCry2 plasmid to mimic the mutations generated in vivo. To do so, the PhCry2 plasmid was amplified through PCR using primers designed to introduce premature stop codon. As part of this amplification, the downstream portion of the gene was removed, and a homologous overhang was included for later re-ligation of the plasmid. The resulting PhCry2 -/-is missing a part of the photolyase domain and the FAD binding domain, as the KO lines used in this paper. Following the amplification, Dpn1 digestion (New England Biology) was achieved following the protocol outlined by manufacturers. Following digestion, a DNA cleanup step was performed using the QIAquick PCR purification kit (Quiagen). The cleaned PhCry2-/-pcDNA3.1 amplicon was then religated through homologous assembly using the NEB Gibson Assembly kit (New England Biology) generating the finalized construct.
HEK-293T cells were plated at 5.10 5 cells per well in 6 well plates (Thermo Fisher Scientific) and were transfected 24-h later using Lipofectamin 2000 Transfection Reagent from Invitrogen (Waltham, MA). A total of 1600 ng of DNA was transfected per well (30 ng of TK-E54, 100 ng of PhBmal1, mClk and CMV β-gal and different increasing concentration of repressors (i.e. 0, 10, 50, 200 and 500 ng of PhCry2 and/or PhPer or PhCry2-/-), supplemented with empty pcDNA3.1 as needed. After transfection, plates were incubated 48 h at 37°C and 5% CO 2 . They were then washed with ice-cold PBS and incubated 15 min in 350 µL of Reporter Lysis Buffer (Promega). Cell lysates were removed from the well and spun down at 13000 rpm. The supernatant was used immediately for assays or stored at -80°C.
Following the manufacturer protocol for the β-Galactosidase assay, 50 µL of the lysate was mixed with 50 µl Assay Buffer (Promega) in a 96 well plate (Corning) and incubated for 30 minutes at 37°C or until a faint yellow color appeared. The reaction was stopped using 150 µl of 1M Sodium Carbonate and the absorbance was read at 420 nm using a SpectraMax iD5 microplate reader (Molecular Devices, San Jose, CA). For the luciferase assay, 20 µl of the lysate was added to 20 µl of D-luciferin mix (0.132mg/mL D-luciferin (Sigma), 20mM Tricine, 2.67mM MgSO4, 0.1mM EDTA, 33.3mM DTT, 530uM ATP (Sigma), 270uM acetyl coenzyme A (Sigma), 265uM 4MgCO3Mg(OH)2) and the luminescence was measured using the same microplate reader. The β-Galactosidase signal was used to normalize the firefly luciferase signal. A) Schematics of PhCry2 mutagenesis. Blue and red boxes indicate the gene sequences targeted by the gRNAs used for CRISPR/Cas9 mutagenesis. Two knock-out mutant lines with premature stop codons were obtained. The resulting mutant PhCRY2 proteins are very similar, with slightly different C-terminal tails (see boxed amino acid sequences). Both are missing most of the photolyase domain (pink) and the Flavin Adenine Dinucleotide (FAD) binding domain (orange). Thus, behavior data from both strains were combined on this figure and on figure 2. B) Reconstruction of the circadian clock mechanism in HEK-293T cells with luciferasebased transcriptional assay. Co-transfection of increasing concentrations (in ng) of PhPer and PhCry2 increase the repression of mClk:PhBmal1-mediated transcription (Kruskal-Wallis test, ***p<0.001; subsequent Dunn's multiple comparison, 0 ng vs 200 ng, *p=0.023, 0 ng vs 500 ng, ***p=0.001; N=5). The transfection of PhCry2 alone can repress as well (Kruskal-Wallis test, **p=0.009; subsequent Dunn's multiple comparison, 0 ng vs 500 ng, *p=0.023; N=4). PhPer alone does not repress (Kruskal-Wallis test, p=0.002; subsequent Dunn's multiple comparison p = ns; N=4). The mutated version of PhCry2 (-/-) does not repress transcription (Kruskal-Wallis test, *p=0.018; subsequent Dunn's multiple comparison shows no significant differences compared to control; N=3). C) Average relative swimming activity of wild-type (PhCry2+/+), heterozygous (PhCry2+/-) and homozygous (PhCry2-/-) animals recorded in constant light (LL) after entrainment to a LD cycle. Gray shading represents the subjective night. Standard error to the mean (SEM) is represented with the orange shading. Vertical swimming 24 traces are shown in all main figures. See supplemental figures for roaming behavior 24. D) Percentage of rhythmic animals in LL (Chi-square test, **p = 0.01; subsequent pairwise Chi-square test, +/+ vs +/-, p = 0.94, +/+ vs -/-, **p = 0.004 and +/-vs -/-, **p = 0.005). F) Doublet period averages under LL (Kruskal-Wallis test, p = 0.670). G) Singlet period averages under LL (Kruskal-Wallis test, *p = 0.031; subsequent Dunn's multiple comparison, +/+ vs +/-, p = 0.421, +/+ vs -/-, *p = 0.026 and +/-vs -/-, p = 0.345). H) Percentage of rhythmic animals under DD after LD entrainment (Chi-square test, ***p < 0.001; subsequent pairwise Chi-square test, ***p < 0.001) Note that the same control animals were used in Figures 3 and 5. Therefore statistical analysis for the DD datasets were performed with multiple-comparisons tests that included all 4 genotypes (Wild-type, PhCry2, PhPer and PhClk mutants). I) Rhythm power averages under DD (Mann-Whitney test, p = 0.231). J) Doublet period averages under DD (Mann-Whitney test, p = 0.730). K) Singlet period averages under DD. L) HCR-FISH of PhPer and PhCry2 mRNAs in circadian medioposterior clock neurons exposed to constant conditions after entrainment to 10.3:2.1 tidal and 12:12 LD cycles in PhCry2 -/-animals (PhCry2 -mutation 1). A z-stack for each mRNA visualized in the left hemisphere is shown across circadian time (CT). Scale bars represent 10µm. M) Quantification of the number of HCR-FISH spots per hemisphere across CT in PhCry2 -/-animals. Each dot represents one hemisphere of one individual and the solid line the mean. Grey shading represents the subjective dark phase and blue shading the subjective high tide. Grey trace represents the WT and black trace the mutant. After testing the effect of time on gene expression (PhPer, Kruskal-Wallis, **p = 0.0072, PhCry2, One-way ANOVA, p = 0.205), rhythm was tested using the JTK-cycle algorithm for 24-h periodicity for PhPer (See Table S2 for exact p-value). Statistical differences between genotypes are indicated with * signs (Tukey's multiple comparison, See Table S3 for exact p-value). Significance thresholds: *p<0.05, **p<0.01 and ***p<0.001. Figure 2: PhCRY2 is required to maintain circatidal rhythms in P. hawaiensis. A) Average relative swimming activity of wild-type (PhCry2+/+), heterozygous (PhCry2+/-) and homozygous (PhCry2-/-) animals recorded in DD and high tide after entrainment to a LD and tidal (10.3 h high tide: 2.1 h low tide) cycles. Blue shading represents the expected high tide. Standard error to the mean (SEM) is represented with the orange shading. B) Percentage of rhythmic animals under DD and high tide (Chi-square test, ***p <0.001; subsequent Chi-square test, +/+ vs +/-, p = 0.86, +/+ vs -/-, ***p <0.001 and +/-vs -/-, ***p <0.001). C) Rhythm power averages under DD and high tide (Welch's t test, p = 0.416). D) Singlet period averages under DD and high tide (Welch's t test, **p = 0.007). E) Doublet period averages under DD and high tide (Welch's t test, **p = 0.005). F) HCR-FISH of PhPer and PhCry2 mRNAs in circatidal dorsal-lateral clock neurons exposed to constant conditions after entrainment to 10.3:2.1 tides and 12:12 LD cycles in PhCry2-/-animals (PhCry2 -mutation 1), as represented in Figure 1.
G) As represented in Figure 1, quantification of the number of HCR-FISH spots in PhCry2 -/-animals. In the mutant, there was an effect of time on gene expression for PhPer, (Kruskal-Wallis, *p = 0.0159), but not for PhCry2 (Kruskal-Wallis, p = 0.545) (See Table S4 for post-hoc test p-value). Statistical differences between genotypes are indicated with * signs (Tukey's multiple comparison, See Table S5 for exact p-value).
See also Figure S2 and Table S4, S5 and S6 proteins are very similar, with slightly different C-terminal tails (see boxed amino acid sequences). All lines are missing key functional domains: the two Per-ARNt-Sim (PAS) (yellow) and the Per-ARNt-Carboxy-terminal (PAC) domains (blue). Thus, behavior data from the three strains were combined on this figure and on Figure 4. B) Average relative swimming activity of wild-type (PhPer+/+), heterozygous (PhPer+/-) and homozygous (PhPer-/-) animals recorded in constant light (LL) after entrainment to a LD cycle. Gray shading represents the subjective night. Standard error to the mean (SEM) is represented with the orange shading. C) Percentage of rhythmic animals in LL (Chi-square test, *p = 0.049; subsequent Chisquare test, +/+ vs +/-, p = 0.93, +/+ vs -/-, *p = 0.023 and +/-vs -/-, *p = 0.016). D) Rhythm power averages under LL (Kruskal-Wallis test, p = 0.137). E) Doublet period averages under LL (Kruskal-Wallis test, p = 0.544). F) Singlet period averages under LL (Kruskal-Wallis test, p = 0.746). G) Percentage of rhythmic animals under DD after LD entrainment (Chi-square test, ***p < 0.001; subsequent pairwise Chi-square test, ***p < 0.001). H) Rhythm power averages under DD (Mann-Whitney Smirnov test, p = 0.492). I) Doublet period averages under DD (Welch's test, *p = 0.043).
J) Singlet period averages under DD (Welch's test, *p = 0.045).
K) HCR-FISH to PhPer and PhCry2 mRNAs in circadian medioposterior clock neurons exposed to constant conditions after entrainment to 10.3:2.1 tidal and 12:12 LD cycles in PhPer -/-animals (PhPer -mutation 1 and 3) as represented in Figure 1.
L) As represented in Figure 1, quantification of the number of HCR-FISH spots in PhPer -/-animals. After testing the effect of time on the gene expression (PhPer, Kruskal-Wallis, ***p < 0.001, PhCry2, Kruskal-Wallis, ***p < 0.001), rhythm was tested using the JTK-cycle algorithm for 24-h periodicity (See Table S2 for exact p-value).
Statistical differences between genotypes are indicated with * signs (Tukey's multiple comparison, See Table S3 for exact p-value). differences between genotypes are indicated with * signs (Tukey's multiple comparison, See Table JTK-cycle algorithm for 24-h periodicity (See Table S2 for exact p-value). Statistical differences between genotypes are indicated with * signs (Tukey's multiple comparison, See Table S3 for exact p-value).
See also Figures S7, S8, S9 Table S2, S3 and S6 S5 for exact p-value).
See also Figure S8 and Table S4, S5 and S6 A) HCR-FISH to PhPer and PhCry2 mRNAs in circadian medioposterior clock neurons exposed to constant conditions after entrainment to 10.3:2.1 tidal and 12:12 LD cycles in PhBmal1 -/-animals as represented in Figure 1.
B) As represented in Figure 1, quantification of the number of HCR-FISH spots in PhBmal1 -/-animals. After testing the effect of time on the gene expression (PhPer, One-way ANOVA, ***p < 0.001, PhCry2, One-way ANOVA, **p = 0.002), rhythm was tested using the JTK-cycle algorithm for 24 h (See Table S2 for exact p-value).
Statistical differences between genotypes are indicated with * signs (Tukey's multiple comparison, See Table S3 for exact p-value).
PhPer and PhCry2 mRNAs in circatidal dorsal-lateral clock neurons exposed to constant conditions after entrainment to 10.3:2.1 tides and 12:12 LD cycles in PhBmal1 -/-animals as represented in Figure 1. D) As represented in Figure 1, quantification of HCR-FISH spots in PhBmal1-/-animals. (PhPer, One-way ANOVA, p =0.376, PhCry2, One-way ANOVA, p =0.098). Statistical differences between genotypes are indicated with * signs (Tukey's multiple comparison, See Table S5 for exact p-value). E) Circadian and circatidal clock models for P. hawaiensis. Activators are in green, and repressors in red. Based on our mutant analysis, we propose that the two clocks are wired differently, transcriptionally. A key difference is PhCLK's role in PhPer regulation. In the circadian clock, PhCLK promotes PhPer transcription. However, in the circatidal clock, it is predominantly a PhPer repressor. It might do so through direct binding to the PhPer promoter, or by promoting the expression of an unknown transcriptional repressor. PhBmal1 promotes PhPer expression in circatidal neurons, either with the help of PhClk, or a different partner. In the circadian neurons, PhPer and PhCry2 mRNAs cycle in antiphase. This could be because PhCry2 is indirectly regulated by PhBmal1 and PhClk, through posttranscriptional regulation, or perhaps because PhBMAL1 also directly or indirectly regulates PhPer with an unknown partner in addition to PhClk. See main text for more discussion. PhCry2 (Cry2-/-) PhPer (Cry2-/-) -0 12 24 36 48 60 72 84 96 108 120 132 144 0 2 4 6 Relative Average Activity (30 min bin) 0 12 24 36 48 60 72 84 108 120 132 144 0 2 4 Relative Average Activity (30 min bin) PhCry2 exon structure +/+ CTGTTGGGACGAGACTCCCTGAGGTAGTAATTGACGCTCTGTAAAGCGACGTATATGTAGGACCTGGGCACCAAGCGACCTAGAAGGTTACAG gRNA/PAM site 1 gRNA/PAM site 2 PhCry2 +/+ ; nt1252 to nt1345 PhCRY2 +/+ Protein Structure (929 AA) HWFRRGLRLHDNPALRDSIINCETFRCIYILDPWFAGSSNVGVNKW GACACC------------------ATCATTAACTGCGAGACATTTCGCTGCATATACATCCTGGACCCGTG--------------------TC PhCry2-muta�on 1 ; nt1252 to nt1345 PhCRY2-muta�on 1 ; Protein Structure (422 AA) HWFRRGLRLHDNHH* GACAACCCTGCTCTGAGGGACTCCATCATTAACTGCGAGACATTTCGCTGCATATACATCCTGGA----------------TCTTCCAATGTC -muta�on 2 ; nt1252 to nt1345 PhCRY2-muta�on 2 ; Protein Structure (428 AA) HWFRRGLRLHDNPASEGLHH* PhCry2 A B CT0 CT6 CT12 CT18 PhPer PhCry2 0 6 12 18 0 500 1000 1500 2000 Number of spots CT * * * * * * * JTK: ** (24h) 0 6 12 18 0 500 1000 1500 2000 2500 3000 Number of spots CT * * * * * * * ANOVA: ns 0 12 24 36 48 60 72 84 96 108 120 132 144 0 2 4 6 Elapsed time (h) Relative Activity min bin) * * * M +/+ -/-*** +/+ -/-Singlet Doublet J K -/-+/+ -/-I H LL DD + + + + + + + + + + + + + + mClk + + + + + + + + + + + + + + PhBmal1 + + + + + + + + + + + + + PhCry2+ ------PhPer 10 10 200 500 ---------PhCry2 -/-E-Box-Luc mClk PhBmal1 PhCry2+ PhPer PhCry2 -/-E-Box-Luc mClk PhBmal1 PhCry2+ PhPer PhCry2 -/------------PhCry2 -/------50 200 500 50 10 200 500 50 10 200 500 50 E-Box-Luc + + + + + mClk + + + + + PhBmal1 + + + + + PhCry2+ -----PhPer 10 200 500 -50 Figure 1 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint this version posted March 2, 2026. ; https://doi.org/10.64898/2026.02.27.708297 doi: bioRxiv preprint PhCry2 +/+, n = 43 PhCry2 +/-, n = 41 PhCry2 -/-, n = 21 +/+ +/--/-+/+ +/-+/+ +/-+/+ +/-Singlet Doublet A B C D E 0 12 24 36 48 60 72 84 96 108 120 132 144 0 2 4 6 Relative average activity (30min bins) 0 12 24 36 48 60 72 84 96 108 120 132 144 0 2 4 6 Relative average activity (30min bins) 0 12 24 36 48 60 72 84 96 108 120 132 144 0 2 4 6 Time elapsed (h) Relative average activity (30min bins) PhPer PhCry2 CT0 CT6 CT12 CT18 G PhCry2 (Cry2-/-) 0 6 12 18 0 100 200 300 400 CT Number of spots * * * * PhPer (Cry2-/-) 0 6 12 18 0 200 400 600 800 CT Number of spots * * K-W: ns K-W: * * * * ** * * * * Dorsal-lateral cells F Figure 2 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint this version posted March 2, 2026. ; https://doi.org/10.64898/2026.02.27.708297 doi: bioRxiv preprint Singlet Singlet Doublet F Medioposterior cells L B A gRNA/PAM site 1 gRNA/PAM site 2 PhPer +/+ ; nt946 to nt1013 ACTTCGTATGCGGTAGCGAGCACCGGTGCCGTTAAGCGTGAGCCACCGGATCACCCCACACTCATCTA TGAAGCATACGCCATCGCTCGTGGCCACGGCAATTCGCACTCGGTGGCCTAGTGGGGTGTGAGTAGAT PhPer exon structure +/+ PhPER +/+ Protein Structure (1258 AA) TSYAVASTGAVKREPPDHPTLIYTQALNYILRIKESF PhPer -muta�on 1; nt946 to nt1013 (333 AA) ACTTCGTATGCGGTATG------CGGTGCGGT------------------- (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint this version posted March 2, 2026. ; https://doi.org/10.64898/2026.02.27.708297 doi: bioRxiv preprint PhPer +/+, n = 38 +/-, n = 45 -/-, n = 21 PhPer PhPer A B PhPer +/+, n = 42 +/-, n = 52 PhPer -/-, n = 31 PhPer D F H +/+ +/--/-C +/+ +/ --/-E PhClk (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint this version posted March 2, 2026. ; https://doi.org/10.64898/2026.02.27.708297 doi: bioRxiv preprint PhClk +/+, n = 75 +/-, n = 44 -/-, n = 23 A B +/+, n = 49 +/-, n = 60 -/-, n = 19 D F PhClk PhClk PhClk PhClk PhClk +/+ +/--/-C +/+ +/ --/-E +/+ +/ --/-+/+ +/ --/-Singlet Doublet G -/ --/ -+ / + + / + PhPer PhCry2 CT0 CT6 CT12 CT18 ANOVA: ns K-W: ns PhPer (Clock-/-) PhCry2 (Clock-/-) 0 6 12 18 0 100 200 300 400 Number of spots CT 0 6 12 18 0 200 400 600 800 CT Number of spots
* * * * * * * (which was not certified by peer review) is the author/funder. All rights reserved. No allowed without The copyright holder for this preprint this version posted March 2, 2026. ; https://doi.org/10.64898/2026.02.27.708297 doi: bioRxiv preprint PhPer PhCry2 CT0 CT6 CT12 CT18 D E Dorsal-lateral cells (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint this version posted March 2, 2026. ; https://doi.org/10.64898/2026.02.27.708297 doi: bioRxiv preprint Supplemental information: Core circadian clock genes control molecular and behavioral circatidal rhythms in Parhyale hawaiensis Figure S1: Behavior of control and PhCry2 knock-out animals recorded in constant darkness after entrainment to LD cycles (related to Figure 1) A) Average relative vertical swimming activity of wild-type (PhCry2+/+) and homozygous mutant (PhCry2-/-) animals recorded in constant darkness (DD) after entrainment to a LD cycle. Light gray shading represents the subjective day and dark gray shading represents the subjective night. Standard error to the mean (SEM) is represented with the orange shading. B) Average relative roaming activity of wild-type (PhCry2+/+) and homozygous (PhCry2-/-) animals recorded in DD after entrainment to a LD cycle. Standard error to the mean (SEM) is represented with the blue shading. A) Average relative roaming activity of wild-type (PhCry2+/+), heterozygous (PhCry2+/-) and homozygous (PhCry2-/-) animals recorded in constant light (LL) after entrainment to a LD cycle. Gray shading represents the subjective night. Standard error to the mean (SEM) is represented with the blue shading. B) Average relative roaming activity of wild-type (PhCry2+/+), heterozygous (PhCry2+/-) and homozygous (PhCry2-/-) animals recorded in constant dark (DD) and high tide after entrainment to LD and tidal (10.3 h high tide: 2.1 h low tide) cycles. Blue shading represents the expected high tide. Standard error to the mean (SEM) is represented with the blue shading. A) Co-staining was observed in most Medioposterior cells, which are thus neurons. Two of them, however, appeared to be Elav negative and might thus be glial cells (*). N= 3 brains B) All four dorsal-lateral cells showed co-staining and are thus neurons. * * (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint this version posted March 2, 2026. ; https://doi.org/10.64898/2026.02.27.708297 doi: bioRxiv preprint Figure S4: PhPer and PhCry2 expression in circadian and circatidal neurons of wild type animals A) HCR-FISH of PhPer and PhCry2 mRNAs in circadian medioposterior clock neurons exposed to constant conditions after entrainment to 10.3:2.1 tidal and 12:12 LD cycles in wild type (WT) animals. A z-stack for each mRNA visualized in the left hemisphere is shown across circadian time (CT). Scale bars represent 10µm. B) Quantification of the number of HCR-FISH spots per hemisphere across CT in WT animals. Each dot represents one hemisphere of one individual and the solid line the mean. Grey shading represents the subjective dark phase and blue shading the subjective high tide. After testing the effect of time on gene expression (PhPer, Kruskal-Wallis, ***p <0.001, PhCry2, Kruskal-Wallis, ***p <0.001), rhythm was tested using the JTK-cycle algorithm for 24 h and 12 h PhPer mRNA rhythms (See Table S2 for exact p-value). Table S4 for exact p-value). (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint this version posted March 2, 2026. ; https://doi.org/10.64898/2026.02.27.708297 doi: bioRxiv preprint A) Average relative roaming activity of wild-type (PhPer+/+), heterozygous (PhPer+/-) and homozygous (PhPer-/-) animals recorded in constant light (LL) after entrainment to a LD cycle. Gray shading represents the subjective night. Standard error to the mean (SEM) is represented with the blue shading. B) Average relative roaming activity of wild-type (PhPer+/+), heterozygous (PhPer+/-) and homozygous (PhPer-/-) animals recorded in constant dark (DD) and high tide after entrainment to LD and tidal (10.3 h high tide: 2.1 h low tide) cycles. Blue shading represents the expected high tide. Standard error to the mean (SEM) is represented with the blue shading. The last high tide (LHT) was experienced at 13:20. which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint this version posted March 2, 2026. ; https://doi.org/10.64898/2026.02.27.708297 doi: bioRxiv preprint C) Average relative roaming activity of wild-type (PhPer+/+), heterozygous (PhPer+/-) and homozygous (PhPer-/-) recorded in DD and high tide after entrainment to LD and tidal (10.3 h high tide: 2.1 h low tide) cycles as represented in B). The last high tide (LHT) was experienced at 19:20. (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint this version posted March 2, 2026. ; https://doi.org/10.64898/2026.02.27.708297 doi: bioRxiv preprint A) Average relative vertical swimming activity of wild-type (PhPer+/+) and homozygous (PhPer-/-) animals recorded in constant darkness (DD) after entrainment to a LD cycle. Light gray shading represents the subjective day and dark gray shading represents the subjective night. Standard error to the mean (SEM) is represented with the orange shading. B) Average relative roaming activity of wild-type (PhPer+/+) and homozygous (PhPer-/-) animals recorded in DD after entrainment to a LD cycle. Standard error to the mean (SEM) is represented with the blue shading. Two putative exons that would encode the PAS-A domain of PhCLK were identified through sequence homology. In red, predicted splicing donor and acceptor sites. The donor site of the second exon cannot be predicted in the absence of transcriptomics information, because the spacer region separating the PAS-A and PAS-B domain is poorly conserved. In grey, short transcript sequences present in the P. hawaiensis sequence read archive (NCBI, Exon 1: SRA:SRR17898727.161596.1, Exon 2: SRA:SRR14908235.12534108.1 and SRA:SRR060813.29331.2).
In yellow, sequence showing weak homology to the region separating the PAS-A and PAS-B domain in other species.
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint this version posted March 2, 2026. ; https://doi.org/10.64898/2026.02.27.708297 doi: bioRxiv preprint B) Average relative roaming activity of wild-type (PhClk+/+), heterozygous (PhClk+/-) and homozygous (PhClk-/-) animals recorded in DD and high tide after entrainment to a LD cycle and tidal (10.3 h high tide: 2.1 h low tide) cycles. Blue shading represents the expected high tide. Standard error to the mean (SEM) is represented with the blue shading. The last high tide (LHT) was experienced at 13:20. C) Average relative roaming activity of wild-type (PhClk+/+), heterozygous (PhClk+/-) and homozygous (PhClk-/-) animals recorded in DD and high tide after entrainment to a LD cycle and tidal (10.3 h high tide: 2.1 h low tide) cycles as represented in B). The last high tide (LHT) was experienced at 19:20. A) Average relative swimming activity wild-type (PhClk+/+) and homozygous (PhClk-/-) animals recorded in constant darkness (DD) after entrainment to a LD cycle. Light gray shading represents the subjective day and dark gray shading represents the subjective night. Standard error to the mean (SEM) is represented with the orange shading. B) Average relative roaming activity wild-type (PhClk+/+) and homozygous (PhClk-/-) animals recorded in DD after entrainment to a LD cycle. Standard error to the mean (SEM) is represented with the blue shading. A) Swimming behavior of wild type (WT) animals. Grey shading represents the expected dark phase and blue shading the expected high tide. B) As represented in A), swimming behavior of PhCry2-/animals. C) As represented in A), swimming behavior of PhPer-/animals. D) As represented in A), swimming behavior of PhClk-/animals. E) As represented in A), swimming behavior of PhBmal1-/animals. Table S1: PCR primers sequences used in this study for genotyping Target gene Forward Reverse Annealing T° PhCry2 5'-CACTGGTTCAGGAGAGGACTCC-3' 5'-CAAACAGGCGGAGTTGAGC-3' 56 PhPer 5'-GACTAGCAATAAGTTCTGCCCGC-3' 5'-ACACAAGACACTGCTGCTACG-3' 54 PhClk 5'-CCTGTCTGAGAAGAAGAGACG-3' 5'-CAGGCAAACTCCTAAGCCTGC-3' 54 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint this version posted March 2, 2026. ; https://doi.org/10.64898/2026.02.27.708297 doi: bioRxiv preprint Table S2: JTK-cycle outputs testing 24-h mRNA rhythms in Medioposterior neurons (Related to Figures 1, 3, 5 and 7) Genotype Target gene Variation through time p-value (Bonferroni corrected) Phase Amplitude WT (13 sampling points) PhPer KW ***p<0.001 3.668e-17 22 484.72 PhCry2 KW ***p<0.001 5.0561e-9 9 2411.54 WT (4 sampling points) PhPer ANOVA***p<0.001 0.00823 0 470.226 PhCry2 ANOVA***p<0.001 1.112e-7 12 876.105 PhCry2 -/-PhPer KW **p<0.007 0.003698 3 222.0315 PhCry2 ANOVA p = 0.205 PhPer -/-PhPer KW ***p<0.001 0.00447 3 236.174 PhCry2 ANOVA***p<0.001 0.03611 3 275.77 PhClk -/-PhPer ANOVA*p = 0.013 0.995 21 63.639 PhCry2 KW p = 0.528 PhBmal1 -/-PhPer ANOVA***p<0.001 0.0885 3 244.305 PhCry2 ANOVA**p = 0.002 0.00374 12 140.714 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint this version posted March 2, 2026. ; https://doi.org/10.64898/2026.02.27.708297 doi: bioRxiv preprint Table S3: Comparison of PhPer and PhCry2 mRNA levels in Medioposterior neurons as a function of genotype (Tukey's multiple comparisons test) (Related to Figures 1, 3, 5 and 7) Comparison PhPer PhCry2 PhPer PhCry2 Signif. pvalue Signif. pvalue Signif. pvalue Signif. pvalue ZT0 ZT6 WT vs PhCry2 -/-*** <0.001 ns 0.987 *** <0.001 *** <0.001 WT vs PhBmal1 -/ns 0.139 ns 0.142 ns 0.523 ns 0.320 WT vs PhClk -/ns 0.200 *** <0.001 ns 0.566 *** <0.001 WT vs PhPer -/-*** <0.001 *** <0.001 ns 0.373 ns 0.333 PhCry2 -/-vs PhBmal1 -/-** 0.002 ns 0.275 *** <0.001 *** <0.001 PhCry2 -/-vs PhClk -/-*** <0.001 *** <0.001 *** <0.001 ** 0.001 PhCry2 -/-vs PhPer -/ns 0.415 *** <0.001 *** <0.001 *** <0.001 PhBmal1 -/-vs PhClk -/-*** <0.001 ns 0.465 ns 1 *** <0.001 PhBmal1 -/-vs PhPer -/ns 0.205 *** <0.001 ns 1 ns 0.999 PhClk -/-vs PhPer -/-*** <0.001 *** <0.001 ns 0.999 *** <0.001 PhPer PhCry2 PhPer PhCry2 Signif. pvalue Signif. pvalue Signif. pvalue Signif. pvalue ZT12 ZT18 WT vs PhCry2 -/-*** <0.001 *** <0.001 * 0.014 ns 0.422 WT vs PhBmal1 -/-*** <0.001 *** <0.001 *** <0.001 ns 0.304 WT vs PhClk -/ns 0.269 *** <0.001 *** <0.001 *** <0.001 WT vs PhPer -/-** 0.005 *** <0.001 *** <0.001 * 0.013 PhCry2 -/-vs PhBmal1 -/ns 0.619 ns 0.999 *** <0.001 ns 1 PhCry2 -/-vs PhClk -/-*** <0.001 ** 0.007 * 0.036 ns 0.258 PhCry2 -/-vs PhPer -/-*** 0.001 *** <0.001 *** <0.001 *** <0.001 PhBmal1 -/-vs PhClk -/-* 0.011 * 0.04 *** <0.001 ns 0.095 PhBmal1 -/-vs PhPer -/ns 0.252 *** <0.001 ns 0.088 *** <0.001 PhClk -/-vs PhPer -/ns 0.541 *** <0.001 ns 0.125 *** <0.001 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint this version posted March 2, 2026. ; https://doi.org/10.64898/2026.02.27.708297 doi: bioRxiv preprint Table S4: Statistical analysis testing 12-h mRNA rhythms in Dorsal-lateral neurons (Related to Figures 2, 4, 6 and 7) Genotype Target gene Variation through time p-value (Bonferroni corrected) Phase Amplitude WT (13 sampling points) PhPer KW ***p<0.001 6.9578e-11 0 101.11627 PhCry2 KW ***p<0.001 1.057821e-6 1 146.01799 WT (4 sampling points) PhPer ANOVA***p<0.001 Tukey's multiple comparison test: **0vs6, 0vs12, *0vs18, **6vs12, 6vs18, *12vs18 PhCry2 KW ***p<0.001 Dunn's multiple comparison test: *0vs6, 0vs12, 0vs18, **6vs12, 6vs18, 12vs18 PhCry2 -/-PhPer KW *p = 0.016 Dunn's multiple comparison test: 0vs6, 0vs12, 0vs18, *6vs12, *6vs18, 12vs18 PhCry2 KW p = 0.545 PhPer -/-PhPer KW p = 0.174 PhCry2 KW p = 0.056 PhClk -/-PhPer ANOVA p = 0.370 PhCry2 KW p = 0.380 PhBmal1 -/-PhPer ANOVA p = 0.375 PhCry2 ANOVA p = 0.098 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint this version posted March 2, 2026. ; https://doi.org/10.64898/2026.02.27.708297 doi: bioRxiv preprint Table S5: Comparison of PhPer and PhCry2 mRNA levels in Dorsal-lateral neurons as a function of genotype (Tukey's multiple comparisons test) (Related to Figures 2, 4, 6 and 7). Comparison PhPer PhCry2 PhPer PhCry2 Signif. pvalue Signif. pvalue Signif. pvalue Signif. pvalue ZT0 ZT6 WT vs PhCry2 -/-ns 0.091 *** <0.001 *** <0.001 ns 0.148 WT vs PhBmal1 -/-*** <0.001 ns 0.849 ns 0.097 ns 0.593 WT vs PhClk -/ns 1 *** <0.001 *** <0.001 ns 0.679 WT vs PhPer -/-** 0.005 ns 0.933 ns 0.737 * 0.019 PhCry2 -/-vs PhBmal1 -/ns 0.256 *** <0.001 ns 0.270 ns 0.982 PhCry2 -/-vs PhClk -/ns 0.068 ns 0.927 ns 0.999 ns 0.796 PhCry2 -/-vs PhPer -/ns 0.560 ** 0.008 ** <0.001 ns 0.914 PhBmal1 -/-vs PhClk -/-*** <0.001 *** <0.001 ns 0.175 ns 0.996 PhBmal1 -/-vs PhPer -/ns 0.999 ns 1 ns 0.459 ns 0.724 PhClk -/-vs PhPer -/-** 0.004 *** <0.001 *** <0.001 ns 0.269 PhPer PhCry2 PhPer PhCry2 Signif. pvalue Signif. pvalue Signif. pvalue Signif. pvalue ZT12 ZT18 WT vs PhCry2 -/-** 0.009 *** <0.001 ns 0.928 ** 0.006 WT vs PhBmal1 -/ns 0.259 *** <0.001 ns 0.321 ns 0.467 WT vs PhClk -/ns 0.677 *** <0.001 * 0.042 ns 0.202 WT vs PhPer -/-*** <0.001 * 0.022 ns 0.941 ns 0.886 PhCry2 -/-vs PhBmal1 -/ns 0.935 ns 0.188 ns 0.804 ns 0.344 PhCry2 -/-vs PhClk -/ns 0.231 ns 0.921 ns 0.251 ns 0.658 PhCry2 -/-vs PhPer -/ns 0.867 *** <0.001 ns 0.523 ns 0.080 PhBmal1 -/-vs PhClk -/ns 0.892 ns 0.6332 ns 0.877 ns 0.986 PhBmal1 -/-vs PhPer -/ns 0.567 ns 0.248 ns 0.067 ns 0.949 PhClk -/-vs PhPer -/-* 0.037 *** <0.001 ** 0.005 ns 0.727 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint this version posted March 2, 2026. ; https://doi.org/10.64898/2026.02.27.708297 doi: bioRxiv preprint Table S6: Period and power of locomotor behavior. Mean and SEM are indicated. (Related to Figures 1, 2, 3, 4, 5 and 6) Genotype Power Period doublet (h) Period singlet (h) Entrainment to LD cycles recorded in LL PhCry2 +/+ 77.7 ± 9.62 25.4 ± 0.369 11.4 ± 1.40 PhCry2 +/-66.2 ± 6.93 25.2 ± 0.185 11.8 ± 1.50 PhCry2 -/-31.8 ± 6.63 27.4 ± 2.46 12.3 ± 4.30 PhPer +/+ 53.9 ± 9.71 25.6 ± 0.480 12.3 ± 0.051 PhPer +/-57.5 ± 6.36 24.7 ± 0.136 12.3 ± 0.102 PhPer -/-16 ± 4.3 25.6 ± 1.10 12.4 ± 0 PhClk +/+ 60.7 ± 7.46 25.0 ± 0.233 11.7 ± 1.1 PhClk +/-55.8 ± 8.27 24.9 ± 0.674 12.2 ± 0.9 PhClk -/-31.6 ± 6.43 27.6 ± 2.61 11.9 ± 3 PhCry2 +/+ 71.2 ± 8.33 24.8 ± 0.142 12.4 ± 0.101 Entrainment to LD cycles and tides (LHT 13 h20) recorded in DD and cst HT PhCry2 +/+ 74 ± 6.25 25.1 ± 0.071 12.6 ± 0.036 PhCry2 +/-83.5 ± 9.75 25.4 ± 0.089 12.7 ± 0.042 PhCry2 -/-ND ND ND PhPer +/+ 60.7 ± 4.91 25.06 ± 0.08 12.5 ± 0.05 PhPer +/-64.0 ± 4.46 24.8 ± 0.06 12.4 ± 0.03 PhPer -/-39.5 ± 7.60 25.5 ± 1.22 13.3 ± 0.97 PhClk +/+ 77.7 ± 7.12 25.0 ± 0.075 12.5 ± 0.032 PhClk +/-80.1 ± 8.87 24.8 ± 0.110 12.4 ±0.066 PhClk -/-37.4 ± 11.12 23.6 ± 0.862 13.9 ± 1.41 PhCry2 +/+ 58.7 ± 5.04 24.8 ± 0.09 12.5 ± 0.05 Entrainment to LD cycles recorded in DD WT 54.4 ± 4.72 24.5 ± 0.09 12.3 ± 0.051 PhCry2 -/-28.9 ± 11.8 26.6 ± 2.60 12.1 ± 0 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint this version posted March 2, 2026. ; https://doi.org/10.64898/2026.02.27.708297 doi: bioRxiv preprint PhPer -/-68.1 ± 15.7 23.3 ±0.382 11.8 ± 0.166 PhClk -/-ND ND ND Entrainment to LD cycles and tides (LHT 19 h20) recorded in DD and cst HT PhPer +/+ 64.3 ± 7.13 25.0 ± 0.118 12.5 ± 0.08 PhPer +/-62.0 ±6.45 25.2 ± 0.417 12.4 ± 0.05 PhPer -/-29.3 ± 10.9 26.3 ± 3.081 13.5 ± 1.35 PhClk +/+ 65.85 ± 6.86 25.3 ± 0.143 12.6 ± 0.06 PhClk +/-65.4 ± 10.2 24.9 ± 0.142 12.5 ± 0.06 PhClk -/-11.8 ± 0 26.3 ± 0