Lunar Phase Modulates Circadian Gene Expression Cycles in the Broadcast Spawning Coral Acropora millepora
Abstract
Many broadcast spawning corals in multiple reef regions release their gametes with incredible temporal precision just once per year, using the lunar cycle to set the night of spawning. Moonlight, rather than tides or other lunar-regulated processes, is thought to be the proximate factor responsible for linking the night of spawning to the phase of the Moon. We compared patterns of gene expression among colonies of the broadcast spawning coral Acropora millepora at different phases of the lunar cycle, and when they were maintained under one of three experimentally simulated lunar lighting treatments: i) lunar lighting conditions matching those on the reef, or lunar patterns mimicking either ii) constant full Moon conditions, or iii) constant new Moon conditions. Normal lunar illumination was found to shift both the level and timing of clock gene transcription cycles between new and full moons, with the peak hour of expression for a number of genes occurring earlier in the evening under a new Moon when compared to a full Moon. When the normal lunar cycle is replaced with nighttime patterns equivalent to either a full Moon or a new Moon every evening, the normal monthlong changes in the level of expression are destroyed for most genes. In combination, these results indicate that daily changes in moonlight that occur over the lunar cycle are essential for maintaining normal lunar periodicity of clock gene transcription, and this may play a role in regulating spawn timing. These data also show that low levels of light pollution may have an impact on coral biological clocks.
Introduction
Broadcast spawning corals––despite their simple anatomy, which includes only a basic nerve net, but no central nervous system or eyes––are capable of extraordinarily accurate temporal patterns of spawning behavior in response to environmental light. Broadcast spawning typically occurs on just one or two nights per year (e.g.,Harrison et al., 1984; Willis et al., 1985; Babcock et al., 1986), and in some species, is predictable to within 15 minutes from year to year (Vize et al., 2005; Levitan et al., 2011). Different species often have unique time windows, or at least windows with limited overlap with the spawning times of sympatric species (e.g.,Babcock et al., 1986; Levitan et al., 2004, 2011; Vize et al., 2005). Three different environmental parameters are accurately integrated by corals to achieve this amazing feat of timekeeping. Seasonal cycles in weather and/or water temperatures set the month, the lunar cycle sets the evening (date), and the daily solar cycle sets the hour (Babcock et al., 1986; Wallace et al., 1986; Penland et al., 2004; van Woesik et al., 2006; Brady et al., 2009). Potential genetic mechanisms underlying this timekeeping are poorly understood in corals (Levitan et al., 2011).
Many cyclical metabolic processes are regulated by biological clocks (reviewed by Dunlap et al., 2004), giving rise to circadian rhythms of metabolic processes that differ between day and night on an approximately 24-hour cycle. Clocks are entrained to their local conditions by environmental cycles in input cues such as light or nutrition. The clocks sit on a molecular foundation of transcription, translation, and post-translational interactions between a small network of genes, which, if defective, impact clock function and the circadian rhythms they drive (Konopka and Benzer, 1971). When clocks are functioning, clock gene networks continue to cycle, even when the entraining signals are removed, potentially driving downstream metabolic and behavioral processes for considerable periods of time (Dunlap et al., 2004). These clock genes are conserved in corals and other cnidarians, and, in some cases, continue undergoing circadian cycles of transcription when kept in constant darkness (Vize, 2009; Reitzel et al., 2010; Brady et al., 2011; Hoadley et al., 2011; Shoguchi et al., 2013; Peres et al., 2014). For example, in the starlet anemone Nematostella many clock genes display entrained cycles of transcription under constant darkness for up to 48 h (Peres et al., 2014). One clock gene, Timeless, has been shown to respond to moonlight in Drosophila (Bachleitner et al., 2007), as do cryptochrome and period genes in some fishes (Ikegami et al., 2014).
An alternative environmental response timing system is known as an hourglass mechanism (Truman, 1971). Much like a traditional, sand-filled hourglass, these timekeeping systems are directly activated by an environmental signal and run at a constant rate until complete. They do not cycle and do not restart until they are reset by the regulating environmental signal. This type of timing system may function in corals to set the hour of spawning after sunset, as evidence has shown that release of gamete bundles occurs at a specific time after sunset; if the timing of sunset changes, the timing of spawning is changed correspondingly (Brady et al., 2009). A second potential example of an hourglass mechanism in corals may be gamete maturation rates, which appear to be governed by annual temperature cycles, as suggested by increased rates of gamete maturation at elevated temperatures (Crowder et al., 2014).
The Moon controls many marine processes through cycles in both tides and illumination. Tides dominate near-shore communities, and are known to regulate circatidal rhythms with a 12.4-h periodicity. This form of biological clock can be entrained by multiple factors, including temperature, salinity, and turbulence, and can free-run when the entraining signal is removed (reviewed in Tessmar-Raible et al., 2011). Some corals live in shallow seas, where tides may have a major impact on their metabolism, while others that live in deeper waters or tropical regions with very little tidal flux are unlikely to utilize circatidal systems. Interestingly, lunar control of larval release in brooding corals and of gamete release in broadcast spawning species seems to be solely under the control of lunar light; experimental corals that are insulated from tidal effects, but subjected to changes in lunar illumination patterns, can reset to new lunar light patterns (Jokiel et al., 1985). The wavelengths of lunar light that are important for setting the timing of spawning are not known. However, lunar light has a broad spectrum very similar to that of sunlight (Johnsen et al., 2006), except when near the horizon (Sweeney et al., 2011), and is unlikely to be a controlling factor. Key elements of the lunar cycle that are more likely to control spawning are its periodicity and the timing of moonrise and moonset (Boch et al., 2011).
Reproductive systems in many animals often function in a seasonal manner, with offspring consistently born at a similar time of the year. In one such well-studied system, the Soay sheep, day length regulates reproductive hormone cycles and, thus, fertility and seasonal timing of gametogenesis (Hanon et al., 2008). In mammals, increasing day length triggers the production of higher levels of TSH-beta and other hormones, in part by circadian clock genes that are expressed in the daytime rather than during the long winter nights, and because of overlap with melatonin cycles (Dardente et al., 2010; Saenz de Miera et al., 2014). However, exactly how the lunar light cycle interacts with potential circadian rhythms or hourglass mechanisms to set the date of spawning in corals following the full Moon is not known.
In this study, we examined the transcription of circadian clock genes in the broadcast spawning coral Acropora millepora (Ehrenberg, 1834), focusing on six different clock genes that are molecular markers of clock function: Clock, Cryptochromes 1 and 2 (Cry1, Cry2), Cycle/Bmal, Timeless, and Eyes Absent (Eya). These markers represent both entrained (Clock, Cry2, Cycle/Bmal) and non-entrained (Cry1, Eya, Timeless) genes. Our results show that corals have significant differences in clock gene transcription over the course of a synodic lunar cycle, and that the stability of these changes requires a normal pattern of lunar illumination. It is possible that lunar shifts in transcription play a role in intersecting with the post-sunset timing system to orchestrate the precise temporal accuracy of coral spawning behavior.
Materials and Methods
Corals
Lunar light experiments were conducted at Orpheus Island Research Station (OIRS) in Queensland, Australia, in November, 2009. Three experimental treatments were set up in a temperature-controlled room (27 °C), from 3 November until 6 December. Six medium-sized colonies of Acropora millepora (greatest diameter = 26.67 ± 1.89 cm (mean ± SEM)) were collected from Cattle Bay at a depth of 2 to 3 m, transported to the OIRS field station, and placed in 55-l plastic tanks supplied with flow-through, unfiltered seawater. We chose an experimental setup for our gene expression study for many reasons, including the capacity for greater control over diurnal light cycles given weather and cloud variation in the field, greater precision in control over temperature and other environmental parameters (e.g., wave action, water flow rate, oxygenation), and the ability to alter the lunar cycle in comparative experimental treatments (see below). All field work was carried out under permit (G09/31214.1) from the Great Barrier Reef Marine Park Authority (GBRMPA).
Each of the three tanks housed two colonies of A. millepora. Tanks were aerated using an air stone, and additional circulation was achieved via a small aquarium pump. Daylight conditions were constant throughout each tank, with lights on at 0530 and off at 1830 Australian Eastern Standard Time (AEST), using a timer, producing a normal 13:11 day:night cycle similar to the local daily solar cycle. To mimic light conditions on the reef, 1 Sylvania Coral-Arc 150W bulb housed in a Sylvania Oracle light fixture (Osram Sylvania, Wilmington, MA), was placed 10–15 cm above each tank. Photosynthetically active radiation (PAR) light conditions underwater on the reef were measured to be 750 μE m−2 s−1, while in-tank light levels were, at maximum, 450 μE m−2 s−1. Such light levels are sufficient for Symbiodinium to photosynthesize, thereby providing a food source in addition to any prey captured by the coral polyps from the unfiltered seawater supplied. Illuminance provided by the lamps over the corals was approximately 21,680 lux, while illuminance of the local sun was 100,000 lux at midday.
To determine the effect of lunar light on patterns of gene expression, one of three lunar light treatments was applied to each tank: 1) a normal lunar cycle, following the natural patterns of lunar periodicity in luminance and in times of moonrise and moonset; 2) a constant lunar cycle, repeating the full Moon time period experienced on November 3, 2009, on every evening; and 3) the absence of a lunar cycle, that is, no lunar light supplied. The normal moonlight cycle treatment consisted of 1 150W Sylvania Coral-Arc bulb, covered with several layers of 70% light-reducing shade cloth to reduce illuminance levels to lunar light measured in the field (0.22 lux at full Moon). The timing and duration of the lunar light period were adjusted each day to follow the moonrise and moonset that occurred at Orpheus Island Research Station, as calculated from the U.S. Naval Observatory’s website (http://aa.usno.navy.mil/data/), using the “Complete Sun and Moon Data for One Day” option. As lunar light levels and moonrise and moonset times varied throughout the lunar cycle, illuminance levels in the “normal lunar light treatment” were reduced to 0.08 lux from the last quarter Moon until the first quarter Moon, with no lunar light present during the new Moon. The “constant lunar cycle treatment” also used 1 150W Sylvania Coral-Arc bulb for the 2 colonies, with lunar lights on at 1900 and off at 0615 AEST time each day, and illuminance set at 0.22 lux. For the “absence of lunar light treatment,” no lunar lights were on during the night. Previous studies using experimental lunar light used only 0.01 μE m−2 s−1 (Franke, 1986) or 0.03 lux (Bachleitner et al., 2007).
To capture changes in both circalunar and circadian gene expression patterns, 2 sets of samples were collected over 2 24-h periods that corresponded to full Moon and new Moon phases for experimental corals held under normal lunar illumination cycles. Samples were collected every 4 h over the 2 24-h periods. Six time points in total were targeted: 3 during the night at 1 h post-sunset (1930, AEST), at midnight (0030), and at 1 h prior to sunrise (0430); and 3 time points during the day: 0730, 1130, and 1530. These sample sets were collected during the day of the full Moon and on the day of the new Moon.
A different set of sample dates and times was used to follow changes at midnight over the lunar month under all three lunar light schedules (normal lunar cycle, constant full Moon, and no lunar light). These samples were collected at midnight on the November full Moon, the last quarter Moon, the new Moon, the first quarter Moon, and the following December full Moon (Table 1). Because moonrise and moonset change every night, there is no optimal single time point for this type of analysis. Midnight was selected because it is close to the nighttime midpoint, it marks the end of the coral spawning window, and it is relatively convenient experimentally.
| Full Moon* | Third quarter Moon | New Moon | First quarter Moon | Full Moon† | |
|---|---|---|---|---|---|
| Tank midnight | ✓ | ✓ | ✓ | ✓ | ✓ |
| Tank 6 × 4 h | ✓ | ✓ | |||
| Field 2130 | ✓ | ✓ | ✓ |
A third set of samples was collected to determine if gene expression in experimental corals mimicked that of corals under natural field conditions. For this purpose, 2 fragments were sampled at 2130 from buoy-marked colonies on the reef flat, adjacent to the OIRS water intake pipe, at a depth of 2 m, on each of the last quarter Moon, new Moon, and full Moon (Table 1). It was not possible to collect at later times and time points because of limited access to field sites at night and poor weather. The experimental and in situ colonies sampled were all sexually mature and of a similar diameter to minimize size or maturity factors that might influence transcription patterns. In total, over 75 time points were sampled. While a larger number of biological replicates would have been optimal, field station space limitations, collection permitting, time, reagent limitations, and sample shipping, among other constraints, made this impossible.
One advantage of using A. millepora is ease of sampling its branching morphology. At each sampling time point, one peripheral branch from each colony was broken off using bone cutters, and immediately immersed in 1 ml of Trizol. The entire branch and Trizol were ground into a slurry using a mortar and pestle, then transferred into a cryogenic tube and stored in a –80 °C freezer. All samples were transported back to Canada with an export permit from CITES (Convention on International Trade in Endangered Species of Wild Fauna and Flora; Permit No. 2009-AU-563189) on dry ice and stored long-term at –80 °C.
qPCR
RNA was isolated from one peripheral branch of each colony at each time point, and then shipped to the laboratory for analysis. Each sample was tested in triplicate, and expression was normalized against the endogenous reference gene RNA polymerase 2, which we have shown to be stable over a 24-h diel cycle in Acropora millepora in RNA sequencing (RNA-seq) and quantitative polymerase chain reaction (qPCR) studies (Brady et al., 2011). Further evidence of its stability, as manifested by no differential expression between day and night over a whole lunar cycle, has also recently been confirmed via additional RNA-seq in a related species, Acropora humilis (M. Moldach and P.D. Vize, unpubl. data).
Primers were developed using Primer3 software (Rozen and Skaletsky, 2000) and transcriptome data from Meyer et al. (2009); see Appendix Table A1 for primer sequences. Primer3 parameters were selected to generate 100–150 bp (base pair) amplicons with a target primer length of 20 nucleotides and melting temperatures of 55–61 °C.
For each reverse transcribed sample, triplicate PCR reactions were performed using the SYBR green kit from Quantace (London, UK) and a Bio-Rad iCycler PCR system (Bio-Rad Laboratories, Hercules, CA). Serial dilutions of complementary DNA (cDNA) were performed to ensure linear responses, melt curves were checked to ensure that there were no primer-dimer issues, and positive and negative controls were performed for each plate. Cycle parameters were 95 °C for 10 min, then 95, 55, and 72 °C for 30 s each for 47 cycles, with data collection occurring in the 72 °C steps.
All other qPCR results were analyzed according to the double delta CT Method (Livak and Schmittgen, 2001), in which the calculated value represents the fold change in gene expression normalized to an endogenous reference gene. For the samples comparing 24-h gene expression profiles on 2 different lunar periods, the calibrator time point was arbitrarily chosen for 10 h after lights on, simply reflecting when initial sampling began. For the samples that investigated changes in gene expression over a complete lunar cycle with varying lunar light treatments, the calibrator time point was also arbitrarily chosen to be the first full Moon, as sampling also began on this day.
To use the double delta CT analysis method, the amplification efficiency of the gene of interest and the reference gene must be approximately equal. Comparable amplification efficiencies were tested using the methods described in Livak and Schmittgen (2001). Using the two-fold serial dilutions, a plot of log cDNA dilution versus ΔCT (CT,Gene minus CT,Ref) was made for each gene of interest to ensure that the absolute value of the slope was ≤ 0.1. All expression levels represent differences relative to the RNA polymerase 2 control gene, rather than absolute expression levels.
Data normalization
The qPCR results for each gene were tested for normal distributions using the Shapiro-Wilk W test for normality (Shapiro and Wilk, 1965). Cry1 and Timeless were found to have normal distributions, while the remaining genes of interest were transformed to meet this requirement. The resulting data met the normal distribution requirement for further analyses (see Appendix Table A2 for P-values, and W values from the Shapiro-Wilk W test and Box-Cox transformations). Variances for all genes were also tested on the normally distributed data (see Appendix Table A3 for P-values from the Bartlett’s Test of Homogeneity of Variances (Bartlett, 1937)).
Statistical methods
All statistical analyses examined effects on the ratio of RNA abundance for sample i (si) relative to the mean RNA abundance for a reference condition (r̄). For the analysis of diurnal changes at the new Moon and full Moon, the reference condition was samples collected at 1530; for the analysis of removal of a normal lunar cycle, the reference condition was samples collected during the initial full Moon. These ratios are asymmetrical because all ratios for si ≤ r̄ lie between 0 and 1, whereas all ratios for r̄ ≤ si lie between 1 and infinity (∞). Therefore, to treat these ranges of possibilities equally, our analyses used generalized linear models that considered the lognormal distribution (possible range –∞ to ∞, centered on 0 when si = r̄). All analyses were implemented using the GLIMMIX (generalized linear mixed models) procedure of SAS/STAT 13.2 (SAS Institute, 2014).
The statistical analyses of both main experiments reflected their respective designs. The analysis of diel expression profiles involved two colonies, with replicate samples from each colony subjected to all possible combinations of lunar phase and time of day. Given this three-factor, completely randomized design, the associated generalized linear mixed model considered all three main effects, all two-factor interactions, and the three-factor interaction. The analysis of midnight expression profiles over a lunar month involved a nested design; two colonies were assigned to each culture condition (six colonies total) and subject to five successive lunar phases. Therefore, the generalized linear model considered the main effects of lunar phase, treatment, and replicate nested within treatment (replicate[treatment]), as well as the lunar phase × treatment and lunar phase × replicate (treatment) interactions. As fragments of only a few colonies were available for the laboratory experiments, replication of among-colony variation was limited. Thus, interpretation cannot reasonably be extrapolated to the population as a whole. Instead, we included colony (replicate) as a fixed factor in the analyses so that interpretations apply only to the material examined during our experiments.
We focused our interpretation primarily on the interactions of lunar phase with time of day or treatment rather than the interactions with replicate, because the former related most directly to our hypotheses. Specifically, we used pairwise comparisons with the Dunn-Šidák Type I error rate correction to compare means among treatments or lunar phases.
Results
Comparison of gene expression patterns between in situ and experimental corals
Most of the experiments described below were performed on tank-kept corals that were maintained with daytime and nighttime light cycles matching conditions on the reef. To ensure that our tank-kept corals had transcription patterns similar to corals in their natural environment, samples were collected from both a sexually mature Acropora millepora colony exposed to the natural lunar cycle on the reef immediately offshore from the Orpheus Island Research Station, and from tank-kept colonies that had an artificial light cycle mimicking field samples. The latter setup was chosen as the experimental setup for many reasons: it gave us more control over the light cycle due to weather/cloud variations, it allowed for precise control over temperature and other environmental parameters (e.g., wave action, water flow rate, oxygenation), and it enabled alteration of the lunar cycle (see next section). qPCR analysis of the two-field samples collected under natural lighting versus the two tank samples with a constructed normal lunar cycle were compared, and are presented in Figure 1. A t-test comparing qPCR results between corals sampled in the field (n = 2) versus from the experimentally simulated normal lunar cycle treatment (n = 2) confirmed no significant differences in gene expression between field and tank corals, except in the case of Timeless, and then only at one sampling time, the new Moon (Fig. 1). Although levels of gene expression differed significantly at different stages in the lunar cycle, the consistency in patterns of expression between field and experimental corals for all genes––except for one gene at one time point––indicates that the experimental corals were good proxies for our study of changes in gene expression in response to the lunar cycle.
Comparison of quantitative polymerase chain reaction (qPCR) cycle threshold values for field versus tank-kept samples at each lunar period. Comparisons were performed using a student’s t test (*P < 0.05; Timeless (Tim) during the new Moon (NM): P = 0.008). Bars indicate standard error of the mean (SEM). Genes: Cry1, Cr2, Cryptochromes 1 and 2; Clk, Clock; Cyc, Cycle/Bmal; Tim, Timeless; Eya, Eyes Absent.
Diurnal gene expression patterns at different phases of the Moon
In order to compare gene expression during the full Moon and new Moon phases of the lunar cycles, samples were collected every 4 h over each 24-h period from two adult colonies maintained under the artificial light regimen that matched the normal lunar cycle. RNA was isolated from one peripheral branch of each clonal colony at each time point, then shipped to the laboratory for analysis. The results comparing expression of six circadian genes, Cry1, Cry2, Clock, Cycle/bmal, Timeless, and Eya by qPCR are shown in Figure 2. Statistical comparisons examining the effect of lunar phase, time of day, and individual colony are shown in Table 2. The results depict the mean level of expression of both colonies (each tested in triplicate) normalized against the control gene, RNAP2, which serves as a good control in measuring diel expression cycles via both RNA-seq and qPCR (Brady et al., 2011).
Change in mean (± SE) RNA expression relative to the mean for samples collected at 1530 during the daily cycle under new Moon and full Moon for fragments from two coral colonies. Symbols without error bars represent means for colony 1 (up-pointing triangles) and colony 2 (down-pointing triangles). Asterisks along the upper end of a plot identify times for which RNA expression differed significantly, based on Dunn-Šidák multiple comparisons. Note the logarithmic scaling of the ordinate. Genes: Cry1, Cr2, Cryptochromes 1 and 2; Clk, Clock; Cyc, Cycle/Bmal; Tim, Timeless; Eya, Eyes Absent.
| Effect | Gene | |||||
|---|---|---|---|---|---|---|
| Cry1 | Cry2 | Clk | Cyc | Tim | Eya | |
| Replicate (R) | F1,40 = 2.27 | F1,70 = 15.35* | F1,70 < 0.001 | F1,100 = 10.69† | F1,40 = 168.4* | F1,40 = 84.47* |
| Lunar phase (L) | F1,40 = 11.93† | F1,70 = 37.95* | F1,70 = 20.63* | F1,100 = 3.26 | F1,40 = 273.5* | F1,40 = 2.69 |
| R × L | F1,40 = 29.09* | F1,70 = 69.14* | F1,70 = 7.20† | F1,100 = 9.23† | F1,40 = 52.90* | F1,40 = 83.77* |
| Time (T) | F4,40 = 338.5* | F4,70 = 7.96* | F4,70 = 1.82 | F4,100 = 0.43 | F4,40 = 21.09* | F4,40 = 1.76 |
| R × T | F4,40 = 1.83 | F4,70 = 13.69* | F4,70 = 15.69* | F4,100 = 1.07 | F4,40 = 63.95* | F4,40 = 13.31* |
| L × T | F4,40 = 18.37* | F4,70 = 4.86† | F4,70 = 4.65† | F4,100 = 0.45 | F4,40 = 44.12* | F4,40 = 18.18* |
| R × L × T | F4,40 = 2.87‡ | F4,70 = 2.89‡ | F4,70 = 3.47‡ | F4,100 = 0.52 | F4,40 = 33.97* | F4,40 = 2.40 |
Diurnal expression patterns were characterized by a range of responses for the six circadian genes, Cry1, Cry2, Clock, Cycle/bmal, Timeless, and Eya, when samples collected during the full Moon were compared with those collected during the new Moon phase in the simulated normal lunar cycle treatment (Fig. 2). Three genes, Clock, Cycle/bmal, and Eya, did not vary significantly over the 24-h cycle, and of these, Cycle/bmal and Eya also showed no difference with the Moon phase (Table 2). However, three genes, Cry1, Cry2, and Timeless, showed strong diurnal cycling, and all three of these diurnal patterns changed with the phase of the Moon (Fig. 2, Table 2). All three genes showed a lesser difference between daytime and nighttime expression under a new Moon than under a full Moon. Table 2 also shows that for four genes, Cry2, Cycle/bmal, Timeless, and Eya, the data showed strong differences in expression between individual replicate colonies.
Cycles over the lunar month
In addition to the new Moon and full Moon diel cycles, expression of the circadian gene set was also examined during the last and first quarter Moon phases. Additional samples were collected at midnight on each quarter of a lunar cycle, and qPCR analysis was performed (Fig. 3). Two colonies maintained in tanks with a normal lunar light cycle were sampled and analyzed in triplicate. The results highlighted the higher expression of some circadian genes at midnight during the new Moon (Cry2, Clock), and also showed that the first quarter Moon had even higher levels of midnight transcription of Cry1 and Cycle/bmal. Eya had lower expression at the full Moon than at all other phases. The general monthly trend for midnight transcription is an increase over the course of the lunar month for most genes, with a rapid decrease back to baseline levels at the full Moon. The biggest difference between the full Moon and the last quarter Moon––the time window in which most broadcast spawners release gametes––was observed in the Eya gene, followed by Cry2 (Fig. 3). These results showed that the expression levels changed for Cry1, Cry2, Cycle/bmal, and Eya at midnight over the course of a lunar cycle, and that different genes have different lunar cycle peak times.
Transcription levels at midnight for six circadian genes over a lunar month. Bars indicate standard error of the mean. FM, full Moon; LQ, last quarter Moon; NM, new Moon; FQ, first quarter Moon. Genes: Cry1, Cr2, Cryptochromes 1 and 2; Clk, Clock; Cyc, Cycle/Bmal; Tim, Timeless; Eya, Eyes Absent.
It is important to note that moonrise and moonset times change by about an hour a day, and both intensity of lunar light levels and length of exposure time change nightly. The results of the midnight sampling were also plotted as level of expression versus length of time of lunar illumination and against the intensity of lunar illumination (in lux) prior to sampling. Neither of these parameters showed any systematic association with levels of gene expression (data not shown).
Transcriptional response to altering the lunar cycle
Lunar cycles were altered in tank-kept corals for an entire lunar month, where changes were made only to the lunar lighting pattern. The three conditions tested were a normal lunar cycle (NLC) in terms of light intensity and moonrise and moonset times, full Moon conditions (FMC) on every evening, and no lunar illumination at all, corresponding to a whole cycle of new Moon conditions (NMC). In addition to the difference in level of illumination in terms of photons, these altered cycles also differ in that the nightly illumination patterns are constant from night to night, whereas the normal lunar cycle sees moonrise and moonset times advancing by about one hour per evening (Boch et al., 2011). All samples were collected at midnight, and the level of expression was compared to that of the full Moon sample representing the first sampling time point. As in Figure 3, all genes tested, with one exception (Timeless), showed differences in the level of midnight transcription when maintained under different lunar cycles (Fig. 4). Removal of the normal lunar cycle and replacement by either constant full Moon or constant new Moon conditions on every evening collapses the normal monthly midnight change for all genes that showed differences over the lunar month (Table 3). Some of the genes that showed the biggest differences between the normal lunar cycle and the full and new Moon cycles were those that showed the weakest differences in diel patterns in a normal lunar cycle between the full and new moons. For example, Cycle/bmal and Eya displayed much lower levels of expression and abnormal cycles under both a constant full Moon and constant new Moon, but did not show any significant differences between diel cycles under the full and new moons. These data show that normal patterns of lunar illumination are critical for normal patterns of transcriptional cycling. They also indicate that long-term blocking of lunar light by shading or cloud cover, or long-term exposure to extraneous lighting equivalent in strength to a full Moon, can interrupt normal clock gene expression cycles.
Altering lunar illumination distorts periodicity in circalunar transcription patterns. Change in mean (± SE) RNA expression relative to the mean for samples during the initial full Moon during a lunar cycle in response to new Moon (black circles), full Moon (white circles), and normal lunar cycles (gray circles) are shown. Triplets of symbols and associated lines along the upper end of a plot indicate specific significant differences among means for the corresponding lunar phase, based on Dunn-Šidák multiple comparisons: symbols not underscored by a common line differ significantly (P < 0.05). Note the logarithmic scaling of the ordinate. The results for the normal lunar cycle are the same as those displayed in Figure 3. FM, full Moon; LQ, last quarter Moon; NM, new Moon; FQ, first quarter Moon. Genes: Cry1, Cr2, Cryptochromes 1 and 2; Clk, Clock; Cyc, Cycle/Bmal; Tim, Timeless; Eya, Eyes Absent.
| Effect | Gene | |||||
|---|---|---|---|---|---|---|
| Cry1 | Cry2 | Clk | Cyc | Tim | Eya | |
| Lunar phase (L) | F3,40 = 10.68 | F3,40 = 13.53 | F3,40 = 31.84 | F3,40 = 45.86 | F3,40 = 1.90 | F3,40 = 42.11 |
| Treatment (T) | F2,40 = 133.8 | F2,40 = 40.52 | F2,40 = 77.35 | F2,40 = 882.4 | F2,40 = 122.9 | F2,40 = 453.9 |
| L × T | F6,40 = 57.07 | F6,40 = 8.68 | F6,40 = 33.22 | F6,40 = 133.0 | F6,40 = 7.16 | F6,40 = 41.17 |
| Replicate(Treatment) | F3,40 = 312.2 | F3,40 = 164.7 | F3,40 = 697.1 | F3,40 = 550.3 | F3,40 = 28.46 | F3,40 = 246.7 |
| L × R(T) | F9,40 = 21.33 | F9,40 = 4.92 | F9,40 = 23.34 | F9,40 = 137.5 | F9,40 = 6.89 | F9,40 = 17.03 |
Discussion
The results presented in this report show that circadian genes have different cycles of transcription at different phases of the Moon. As there were significant individual effects and only two biological replicates at each sampling point, a greater number of replicates will be required to determine exactly how much change occurs, and whether it represents changes in the phase or in the amplitude of clock gene expression cycles (Fig. 5). The data presented in Figure 2 illustrate examples that resemble both types of change. For example, the Cry1 gene had very similar times of maximal and minimal expression at both the new Moon and the full Moon, but the levels of expression differed. This finding may represent a change in amplitude only. The Cry2, Clock, and Timeless genes, however, showed similar ranges in expression levels at different lunar phases, but the maximal level of expression occurred earlier in the day during a new Moon than a full Moon. This may represent a change in the phase of the diurnal oscillation. The maximal level of expression was reached 12 h earlier during the new Moon for both Cry2 and Clock. If this shift is confirmed, it would mean that the peak of expression changes for these genes by approximately one hour per night––in spite of the fact that they are entrained genes and they continue to cycle in larvae under constant darkness (Brady et al., 2011; Hoadley et al., 2011; Peres et al., 2014).
Models of 24-h transcription cycle change over a lunar month. The left panel illustrates a phase shift of diel transcription cycles for a model gene over the lunar month. In this model, the nightly maxima and minima of a 24-h cycle are the same, but they occur at different times of the day, at different stages of the lunar cycle. This pattern matches what is observed in Figure 2 for the Cry2 gene. Arrows show that the relative level of expression of this gene at midnight is at different phases of the Moon. In the panel on the right, the diel level of expression for a model gene at a specific time point changes over the course of a lunar cycle. While the level of expression at a specific time point changes over the cycle, the phase (timing of the maximum and minimum per day) stays the same. This model resembles the data shown in Figure 2 for the Cry1 gene that did not show a phase shift, but did show different expression levels at midnight at different points of the lunar month.
Other circadian clocks phase shift by one hour or more per night; for example, there is the jet lag response in humans (Eastman and Burgess, 2009) or wheel running in rodents under constant darkness (Dunlap et al., 2004), and both moonrise and moonset times shift by a similar amount of time. Drosophila shifts peaks of two proteins, period and timeless, by two hours when grown under quarter Moon versus new Moon lighting (Bachleitner et al., 2007). This result was obtained with treatments that were long-term growth under the two different lunar light conditions, not cycling intensity and time windows as occur in the natural lunar cycle. Interestingly, these results cast one of our previous conclusions into doubt––the direct control of spawn timing by sunset time. In these earlier experiments (Brady et al., 2009), we shifted sunset time by one hour and observed that spawning also shifted by one hour. However, the results in Figure 2 show that entrained genes can phase shift in response to the 1-h difference in moonrise times, so that even an entrained behavior might still have been able to shift by one hour. A better experimental design would have been to shift by 60 and 30 min and -30 min (30 min earlier); we will perform this experiment when the opportunity becomes available.
Lunar gene cycles have also been studied in the polychaete worm Platynereis dumerilii, which displays increased locomotor activity under new Moon phases and less activity under full Moon conditions (Zantke et al., 2013). As reported here for Cry1 in Acropora, lunar light also alters the amplitude of transcription profiles of circadian genes in Platynereis. In the Platynereis study, only two different light conditions were used, new Moon and full Moon-type conditions; quarter Moon levels were not described. Under those two lunar light options, the greatest difference was observed for clock.Zantke et al. (2013) also reported that Platynereis’ Timeless continued to show different nighttime transcription over a lunar month under a constant full Moon versus a new Moon. Our result in our Acropora experiment, showing that Timeless along with Clock had elevated expression under a constant full Moon (Fig. 4), may indicate that the Platynereis result is due to this effect and not an entrained circalunar clock. All of our data indicate that the transcription of circadian genes is not under the control of an endogenous circalunar timer (Fig. 4), and support the behavioral observations of Boch et al., who showed that Acropora humilis could be reprogrammed to spawn by artificial moonlight over a short time period (Boch et al., 2011). Our data also support the recent results of Kaniewska et al. (2015), who blocked spawning during only eight evenings of either full Moon or new Moon conditions preceding spawning night.
Our lunar cycles in midnight transcription levels showed some differences to previously published results in Acropora millepora. Using A. millepora, Levy et al. found increased expression of Cry2 at midnight on a full Moon evening compared to a new Moon (Levy et al., 2007), but observed no differences for Cry1. This same group recently published an RNA-seq analysis which found that Cry1 levels did change with the lunar cycle, as we observed herein (Fig. 2), but they found that midnight levels of Cry1 were higher in the full Moon (Kaniewska et al.,2015). In contrast, in our analysis Cry1 was higher at the new Moon and peaked at the first quarter Moon. Our data show that Cry2 midnight transcription peaks during the new Moon (Figs. 3 and 4). Our result is supported by Hoadley et al., who studied the stony coral Favia fragum (Hoadley et al., 2011). They found that Cry2 showed a significant increase in expression 10 days after the full Moon. Unfortunately, this later study did not continue to sample through the next new Moon, so it is not possible to determine exactly how similar the cycles were between these two coral species. Furthermore, Levy et al. (2007) described only two time points, the full Moon and new Moon. We also showed previously that Cry2 transcription was strongly entrained, and continued to oscillate transcription cycles under constant darkness; a recent study in Nematostella vectensis showed similar results (Peres et al., 2014). However, Levy et al. did not find these results for Cry2 transcription. We also noted that Cry1 peaked during the first quarter Moon, a time window not measured by Levy et al. (2007). The patterns described herein are also similar to our previous results using these primers on adult tissue of Acropora millepora (Brady et al., 2011), although there are some differences. As the data in Figure 2 show significant individual effects (Table 2), differences to previous studies may simply be due to differences between individual coral colonies or differences in the locations from which corals were collected. The data presented here show that in the same two individual colonies, patterns were significantly different at different phases of the Moon, and that these cycles are, in most instances, similar to patterns in other studies.
Replacing the normal lunar cycle with either full Moon conditions on every evening or new Moon conditions every night destroyed the normal Acropora lunar phase cycle of midnight transcription peaking at the first quarter Moon for all genes tested in at least one phase of the Moon, and for most genes, at two or more phases. In all of these cases, expression collapsed without the normal lunar cycle. For one gene, Eya, full Moon conditions had no effect relative to the normal lunar cycle, but this treatment affected at least one lunar phase for the other five genes. These data also support previous observations that showed that the influence of the Moon on coral timing processes was mediated by lunar light (Jokiel et al., 1985; Boch et al., 2011; Kaniewska et al., 2015) and not other lunar regulated processes such as tides. The only variable that differed between tanks in our experiments was light.
The results presented herein show that circadian gene expression patterns change over the course of a lunar month. These changes require a normal lunar light illumination cycle; changing nightly illumination to either full Moon or new Moon strength every evening destroys the normal monthly cycles. Not only do the levels of transcription of these genes change at midnight over the lunar month, but also the amplitude and possibly the phase of diel expression cycles change. Showing that transcription levels and phase change with the lunar cycle is a starting point at which to begin to explore the role of these genes in circalunar processes. These results also indicate that extraneous light of the intensity of the full Moon is sufficient to disrupt normal transcriptional cycles in clock genes. This implies that light pollution could have serious impacts on coral clock-driven processes. The absence of moonlight also disrupts normal transcription cycles, and changing weather patterns with increased cloud cover may have similarly detrimental effects on coral clock-driven systems, potentially including spawn timing.
Acknowledgments
Funding for this work was provided by the National Science and Engineering Research Council (NSERC) via a Discovery Award to PDV, by an NSERC PGS award to AKB, and an Alberta Innovates Scholarship to AKB. The authors would like to thank the managers of the OIRS, Kylie and Rob Eddie, for their help in setting up and running the experiments described herein.
Literature Cited
Babcock, R. C., G. D. Bull, P. L. Harrison, A. J. Heyward, J. K. Oliver, C. C. Wallace, and B. L. Willis. 1986. Synchronous spawnings of 105 scleractinian coral species on the Great Barrier Reef. Mar. Biol. 90: 379–394. Bachleitner, W., L. Kempinger, C. Wülbeck, D. Rieger, and C. Helfrich-Förster. 2007. Moonlight shifts the endogenous clock of Drosophila melanogaster. Proc. Natl. Acad. Sci. USA 104: 3538–3543. Bartlett, M. S. 1937. Properties of sufficiency and statistical tests. Philos. Trans. R. Soc. Lond. A Math. Phys. Sci. 160: 268–282. Boch, C. A., B. Ananthasubramaniam, A. M. Sweeney, F. J. Doyle III, and D. E. Morse. 2011. Effects of light dynamics on coral spawning synchrony. Biol. Bull. 220: 161–173. Brady, A. K., J. D. Hilton, and P. D. Vize. 2009. Coral spawn timing is a direct response to solar light cycles and is not an entrained circadian response. Coral Reefs 28: 277–280. Brady, A. K., K. A. Snyder, and P. D. Vize. 2011. Circadian cycles of gene expression in the coral, Acropora millepora. PLoS One 6: e25072. Crowder, C. M., W.-L. Liang, V. M. Weis, and T.-Y. Fan. 2014. Elevated temperature alters the lunar timing of planulation in the brooding coral Pocillopora damicornis. PLoS One 9: e107906. Dardente, H., C. A. Wyse, M. J. Birnie, S. M. Dupré, A. S. I. Loudon, G. A. Lincoln, and D. G. Hazlerigg. 2010. A molecular switch for photoperiod responsiveness in mammals. Curr. Biol. 20: 2193–2198. Dunlap, J. C., J. L. Loros, and P. J. DeCoursey. 2004. Chronobiology: Biological Timekeeping. Sinauer, Sunderland, MA. Eastman, C. I., and H. J. Burgess. 2009. How to travel the world without jet lag. Sleep Med. Clin. 4: 241–255. Franke, H.-D. 1986. Resetting a circalunar reproduction rhythm with artificial moonlight signals: phase-response and ‘moon-off’ effect. J. Comp. Physiol. A 159: 569–576. Hanon, E. A., G. A. Lincoln, J.-M. Fustin, H. Dardente, M. Masson-Pévet, P. J. Morgan, and D. G. Hazlerigg. 2008. Ancestral TSH mechanism signals summer in a photoperiodic mammal. Curr. Biol. 18: 1147–1152. Harrison, P. L., R. C. Babcock, G. D. Bull, J. K. Oliver, C. C. Wallace, and B. L. Willis. 1984. Mass spawning in tropical reef corals. Science 223: 1186–1189. Hoadley, K. D., A. M. Szmant, and S. J. Pyott. 2011. Circadian clock gene expression in the coral Favia fragum over diel and lunar reproductive cycles. PLoS One 6: e19755. Ikegami, T., Y. Takeuchi, S.-P. Hur, and A. Takemura. 2014. Impacts of moonlight on fish reproduction. Mar. Genomics 14: 59–66. Johnsen, S., A. Kelber, E. Warrant, A. M. Sweeney, E. A. Widder, R. L. J. Lee Jr., and J. Hernández-Andrés. 2006. Crepuscular and nocturnal illumination and its effects on color perception by the nocturnal hawkmoth Deilephila elpenor. J. Exp. Biol. 209: 789–800. Jokiel, P. L., R. Y. Ito, and P. M. Liu. 1985. Night irradiance and synchronization of lunar release of planula larvae in the reef coral Pocillopora damicornis. Mar. Biol. 88: 167–174. Kaniewska, P., S. Alon, S. Karako-Lampert, O. Hoegh-Guldberg, and O. Levy. 2015. Spawning cascades and the importance of moonlight in coral broadcast mass spawning. eLife 4: e09991. Konopka, R. J., and S. Benzer. 1971. Clock mutants of Drosophila melanogaster. Proc. Natl. Acad. Sci. USA 68: 2112–2116. Levitan, D. R., H. Fukami, J. Jara, D. Kline, T. M. McGovern, K. E. McGhee, C. A. Swanson, and N. Knowlton. 2004. Mechanisms of reproductive isolation among sympatric broadcast-spawning corals of the Montastraea annularis species complex. Evolution 58: 308–323. Levitan, D. R., N. D. Fogarty, J. Jara, K. E. Lotterhos, and N. Knowlton. 2011. Genetic, spatial, and temporal components of precise spawning synchrony in reef building corals of the Montastraea annularis species complex. Evolution 65: 1254–1270. Levy, O., L. Appelbaum, W. Leggat, Y. Gothlif, D. C. Hayward, D. J. Miller, and O. Hoegh-Guldberg. 2007. Light–responsive cryptochromes from a simple multicellular animal, the coral Acropora millepora. Science 318: 467–470. Livak, K. J., and T. D. Schmittgen. 2001. Analysis of relative gene expression data using real–time quantitative PCR and the 2−ΔΔCT method. Methods 25: 402–408. Meyer, E., G. V. Aglyamova, S. Wang, J. Buchanan–Carter, D. Abrego, J. K. Colbourne, B. L. Willis, and M. V. Matz. 2009. Sequencing and de novo analysis of a coral larval transcriptome using 454 GSFlx. BMC Genomics 10: 219. doi: https://doi.org/10.1186/1471-2164-10-219. Penland, L., J. Kloulechad, D. Idip, and R. van Woesik. 2004. Coral spawning in the western Pacific Ocean is related to solar insolation: evidence of multiple spawning events in Palau. Coral Reefs 23: 133–140. Peres, R., A. M. Reitzel, Y. Passamaneck, S. C. Afeche, J. Cipolla-Neto, A. C. Marques, and M. Q. Martindale. 2014. Developmental and light-entrained expression of melatonin and its relationship to the circadian clock in the sea anemone Nematostella vectensis. EvoDevo 5: 26. doi: https://doi.org/10.1186/2014-9139-5-26. Reitzel, A. M., L. Behrendt, and A. M. Tarrant. 2010. Light entrained rhythmic gene expression in the sea anemone Nematostella vectensis: the evolution of the animal circadian clock. PLoS One 5: e12805. Rozen, S., and H. Skaletsky. 2000. Primer3 on the WWW for general users and for biologist programmers. Methods Mol. Biol. 132: 365–386. Sáenz de Miera, C., S. Monecke, J. Bartzen-Sprauer, M.-P. Laran-Chich, P. Pévet, D. G. Hazlerigg, and V. Simonneaux. 2014. A circannual clock drives expression of genes central for seasonal reproduction. Curr. Biol. 24: 1500–1506. SAS Institute, Inc. 2014. SAS/STAT ® 13.2 User’s Guide. SAS Institute, Cary, NC. Shapiro, S. S., and M. B. Wilk. 1965. An analysis of variance test for normality (complete samples). Biometrika 52: 591–611. Shoguchi, E., M. Tanaka, C. Shinzato, T. Kawashima, and N. Satoh. 2013. A genome-wide survey of photoreceptor and circadian genes in the coral, Acropora digitifera. Gene 515: 426–431. Sweeney, A. M., C. A. Boch, S. Johnsen, and D. E. Morse. 2011. Twilight spectral dynamics and the coral reef invertebrate spawning response. J. Exp. Biol. 214: 770–777. Tessmar-Raible, K., F. Raible, and E. Arboleda. 2011. Another place, another timer: marine species and the rhythms of life. Bioessays 33: 165–172. Truman, J. W. 1971. Hour-glass behavior of the circadian clock controlling eclosion of the silkmoth Antheraea pernyi. Proc. Natl. Acad. Sci. USA 68: 595–599. van Woesik, R., F. Lacharmoise, and S. Kok̈sal. 2006. Annual cycles of solar insolation predict spawning times of Caribbean corals. Ecol. Lett. 9: 390–398. Vize, P. D. 2009. Transcriptome analysis of the circadian regulatory network in the coral Acropora millepora. Biol. Bull. 216: 131–137. Vize, P. D., J. A. Embesi, M. Nickell, D. P. Brown, and D. K. Hagman. 2005. Tight temporal consistency of coral mass spawning at the Flower Garden Banks, Gulf of Mexico, from 1997–2003. Gulf Mex. Sci. 23: 107–114. Wallace, C. C., R. C. Babcock, P. L. Harrison, J. K. Oliver, and B. L. Willis. 1986. Sex on the reef: mass spawning of corals. Oceanus 29: 38–42. Willis, B. L., R. C. Babcock, P. L. Harrison, J. K. Oliver, and C. C. Wallace. 1985. Patterns in the mass spawning of corals on the Great Barrier Reef from 1981 to 1984. Pp. 343–348 in Proceedings of the Fifth International Coral Reef Congress. Vol. 4: Symposia and Seminars (B). Antenne Museum-EPHE, Moorea, French Polynesia. Zantke, J., T. Ishikawa-Fujiwara, E. Arboleda, C. Lohs, K. Schipany, N. Hallay, A. D. Straw, T. Todo, and K. Tessmar-Raible. 2013. Circadian and circalunar clock interactions in a marine annelid. Cell. Rep. 5: 99–113.
Appendix
| Gene name | Abbreviation | Forward primer (5′ to 3′) | Reverse primer (5′ to 3′) |
|---|---|---|---|
| Candidate circadian genes | |||
| Cryptochrome 1 | Cry1 | ACTTAGCTCGCCATGCTGTT | GCCTGCATTCAGACTCCACT |
| Cryptochrome 2 | Cry2 | TGGCATTTAAGCTTGCCTCT | CATGCGCCAAACAGTTTTC |
| Clock | Clk | ACTTGGCAGCCGTCATTTAC | CGCTTTGAGAGGCAAACATA |
| Cycle/Bmal | Cyc | GCCTTACTCCTGATGTTTCG | GGACCGGAGTTATGGAGTCT |
| Timeless | Tim | AGTCCATTGTGCCATTTGAT | CAGCACTCTGTTGGTTCCTT |
| Eyes absent | Eya | CTTGGCTGACCTTGGCTTTA | AGTGGAAAAATTCCCCCAAG |
| Sine oculis-related homeobox | Six | ATAGAACGCCTCGGGAGATT | TGGTGAAAGGAAACCAAAGC |
| Reference gene | |||
| RNA Polymerase 2 | RPII | CCAAACTCCAATCCACCTTG | AAGACCTAAATAGTCATCCAT GAGG |
| Gene | Shapiro-Wilk W Test for normality | Shapiro-Wilk W Test for normality | |||
|---|---|---|---|---|---|
| W value | P-values | Transformation | W value | P-values | |
| Cry1 | 0.969901 | 0.1908 | |||
| Cry2 | (y−1.4-1)/−0.05 | 0.955539 | 0.0510 | ||
| Clk | (y−1-1)/−0.1 | 0.961924 | 0.0840 | ||
| Cyc | (y0.2-1)/0.5 | 0.980916 | 0.5404 | ||
| Tim | 0.961536 | 0.0807 | |||
| Eya | (y0.6-1)/0.6 | 0.978297 | 0.4309 | ||
| Gene | LQ | NM | FM |
|---|---|---|---|
| Cry1 | 0.6947 | 0.4509 | 0.5009 |
| Cry2 | 0.1819 | 0.1654 | 0.1343 |
| Clk | 0.3030 | 0.6696 | 0.2219 |
| Cyc | 0.0598 | 0.1047 | 0.2230 |
| Tim | 0.5875 | 0.3464 | 0.1304 |
| Eya | 0.7323 | 0.0995 | 0.0501 |
