Contribution of the eye and of opn4xa function to circadian photoentrainment in the diurnal zebrafish

Abstract

The eye is instrumental for controlling circadian rhythms in mice and human. Here, we address the conservation of this function in the zebrafish, a diurnal vertebrate. Using lakritz (lak) mutant larvae, which lack retinal ganglion cells (RGCs), we show that while a functional eye contributes to masking, it is largely dispensable for the establishment of circadian rhythms of locomotor activity. Furthermore, the eye is dispensable for the induction of a phase delay following a pulse of white light at CT 16 but contributes to the induction of a phase advance upon a pulse of white light at CT21. Melanopsin photopigments are important mediators of photoentrainment, as shown in nocturnal mammals. One of the zebrafish melanopsin genes, opn4xa, is expressed in RGCs but also in photosensitive projection neurons in the pineal gland. Pineal opn4xa+ projection neurons function in a LIGHT ON manner in contrast to other projection neurons which function in a LIGHT OFF mode. We generated an opn4xa mutant in which the pineal LIGHT ON response is impaired. This mutation has no effect on masking and circadian rhythms of locomotor activity, or for the induction of phase shifts, but slightly modifies period length when larvae are subjected to constant light. Finally, analysis of opn4xa;lak double mutant larvae did not reveal redundancy between the function of the eye and opn4xa in the pineal for the control of phase shifts after light pulses. Our results support the idea that the eye is not the sole mediator of light influences on circadian rhythms of locomotor activity and highlight differences in the circadian system and photoentrainment of behaviour between different animal models.

Fig 1. Locomotor activity of larvae devoid of RGCs in LD, DD and LL.

A) Experimental design of LD, DD and LL experiments. White rectangles represent the day period, while black rectangles represent the night period, light grey rectangles represent the subjective day period and dark grey rectangles the subjective night. For each experiment, larvae are entrained for 5 LD cycles and their locomotor activity is tracked either in LD (LD), constant darkness (DD) or constant light (LL) the larvae are therefore 5dpf at the beginning of locomotor activity measurements. B) Average distance moved (mm/min) In LD. Merged data from 3 independent experiments represented in 10 min bins. Error bars represent SE. The distance moved is lower in lak larvae than control sibling larvae during the 1st (p = 0.008) and 2nd days (p = 0.005) but not during the 3rd day (p = 0.13) nor during the night (p = 0.42, p = 0.51 and p = 0.57 for the 1st, 2nd and 3rd nights. The lack of a phenotype of lak larvae in the third light phase could be linked to the overall state of the larvae as they are not fed during the experiment; Mann-Whitney two-tailed test), see S1 Table. * P<0.05, **p<0.01, ***p<0.005. C) Average distance moved in DD. Merged data from 3 independent experiments represented in 10 min bins. Error bars represent SE. No differences are detected between the distance moved of siblings versus lak larvae using a Mann-Whitney two-tailed test for each subjective night or day, see S2 Table. D) Estimation of the periods using the FFT-NLLS method. Calculations were made on four complete cycles in DD. The mean period is not significantly different between sibling and lak larvae in DD (control: 25.08 ± 1.59 hours (n = 114), lak: 24.95 ± 1.44 hours (n = 83); mean ± S.D; p = 0.32; Mann-Whitney two-tailed test, sibling vs lak larvae). Each grey point represents a single larva. E) Average distance moved in LL. Merged data from 3 independent experiments represented in 10 min bins. Error bars represent SE. F) Estimation of the periods using the FFT-NLLS method. Calculations were made on three complete cycles in LL. Mean± sd (in hours) is represented. Each grey point represents a larva, n = 66 siblings, n = 68 lak larvae.

Fig 2. Larvae devoid of RGCs still photoentrain to pulses of white light at CT16 and CT21.

A) Experimental design of phase delay (PD) experiments. White rectangles represent the day or light pulse period, black rectangles represent the night period and dark grey rectangles represent the subjective day. For each experiment, larvae are entrained for 5 LD cycles, the larvae are therefore 5dpf at the beginning of locomotor activity measurements. Locomotor activity is tracked either in constant darkness for 4 days (DD) or tracked in constant darkness for 4 days and subjected to a 2-hours pulse of light during the night of the 2nd day of constant darkness at CT16 (PD). The phase of locomotor activity is calculated for each larva before and after the timing of the pulse for DD and PD experiments and the Δphase (phase after the pulse–phase before the pulse) is calculated. B) Average distance moved by control larvae (mm/min over 10min) in DD and PD experiments. Mean ± SE. The Δphase calculated using the FFT-NLLS method of PD larvae is higher than the one of DD larvae (p<0.0001, Mann-Whitney two-tailed test), showing that the pulse of light induced a phase delay. C) Average distance moved merged from PD experiments represented in 10 min bins. Mean ± SE. The Δphase of control versus lak larvae calculated using the FFT-NLLS method is not significantly different (p = 0.24, Mann-Whitney two-tailed test). lak show lower levels of activity during the light pulse (p = 0.03, Mann-Whitney two-tailed test). D) Experimental design of phase advance (PA) experiments. The iconography is similar to A). PA-pulsed larvae were subjected to a one-hour pulse of light at CT21. E) Average distance moved of control larvae (mm/min over 10min) in DD and PD experiments Mean ± SE. The Δphase calculated using the FFT-NLLS method is negative in PA-pulsed larvae and statistically different from DD larvae (p<0.0001, Mann-Whitney two-tailed test), showing that the pulse of light induced a phase advance. F) Average distance moved merged from PA experiments represented in 10 min bins. Mean ± SE.

Fig 3

cry1a but not per2 expression is induced upon a pulse of white light at CT21 (A-G) Expression of per2 (A-D) or cry1a (E-G) at CT22 in 7 days old larvae. Before fixation for in situ at CT 22 the larvae were treated exactly like in the experiment described in Fig 2D, meaning that they were not depigmented before the light pulse which was administered from CT21 and CT22 (CT 22 pulse). In parallel, larvae from the same litter were maintained in the dark and fixed at CT22 (CT 22 dark). (A-D) The white ellipse identifies the pineal gland where no expression is observed. ctl DARK: n = 3, ctl PULSE n = 7, lak DARK: n = 4, lak PULSE: n = 7. Scale bar: 50 μm. (E-G) Three different expression patterns can be identified with the cry1a probe, The grey ellipse surrounds the pineal which expresses cry1a in the ‘mild’ and ‘high’ patterns, the two dark ellipses in G surrounds the habenulae which express cry1a in the ‘high’ pattern. Tel = Telencephalon, Tec = Optic tectum. Scale bar: 200 μm. (H) Countings of the repartition of Dark and Pulsed 7 days old larvae at CT22 stained with the cry1a probe, ctl DARK: n = 10, ctl PULSE n = 4, lak DARK: n = 5, lak PULSE: n = 7. Ctl DARK versus Ctl PULSE: p = 0.0010, lak DARK versus lak PULSE: p>0.9999, ctl DARK versus lak DARK: p>0.9999, ctl PULSE versus lak PULSE: p = 0.0182 using Fisher test.

Fig 4

Mutation in opn4xa abolishes light sensitivity in pineal opn4xa+ cells (A) Scheme showing the 5’ part of the opn4xa locus and in particular the second exon targeted by the CRISPR guide RNA (target sequence is highlighted in red) as well as the WT and mutant exon2 sequences. (B-C) Counts of the number of opn4xa+ in the RGC layer (B) and the interneuron layer of the retina (C) after in situ hybridization at 4 days (ZT0) (D) Expression of opn4xa in the pineal at 4 days (ZT0) (upper panel) and expression of fos at 3 days after 30 min of illumination (lower panel) Red arrowheads point to individual labelled pineal cells. Dorsal views are shown, anterior is up (E) Counts of the number of opn4xa+ pineal cells. Scale bars respectively represent 10 μm (B) and 5 μm (C-D). n.s: not significant, see results section for details.

Fig 5. Locomotor activity of larvae devoid of opn4xa-mediated photosensitivity (opn4xa-/-) in LD, DD and LL.

A) Experimental design of LD, DD and LL experiments. White rectangles represent the day period, black rectangles represent the night period, dark grey rectangles represent the subjective day period and light grey rectangles the subjective night. For each experiment, larvae are entrained for 5 LD cycles and their locomotor activity is tracked either in LD (LD), constant darkness (DD) or constant light (LL) the larvae are therefore 5dpf at the beginning of locomotor activity measurements. B) Average distance moved (mm/min over 10min) of 3 independent experiments in LD. Mean ± SE. The distance moved is not different in opn4xa+/+ and opn4xa-/- larvae during the 1st (p = 0.73), 2nd days (p = 0.50) and 3rd days (p = 0.07) nor during the 1st (p = 0.30), and 2nd nights (p = 0.27) (S4 Table). A lower level of activity is found in opn4xa-/- larvae during the 3rd night (p = 0.01) but is visually clear in only one of the 3 independent experiments (Mann-Whitney two-tailed test). C) Average distance moved (mm/min over 10min) of 3 independent experiments in DD. Mean ± SE. D) Estimation of the periods using the FFT-NLLS method calculated over four cycles. The mean period is not significantly different between control and opn4xa+/+ and opn4xa-/- larvae in DD (opn4xa+/+: 25.05 ± 1.43 hours (n = 64), opn4xa-/-: 25.35 ± 1.60 hours (n = 60); mean±SD; p = 0.29; Mann-Whitney two-tailed test). Mean± sd (in hours) is represented. Each grey point represents a larva. E) Average distance moved (mm/min over 10min) of 4 independent experiments in LL. Mean ± SE. opn4xa-/- are more active than controls during the first night (p = 0.02, see S6 Table). F) Estimation of the periods using the FFT-NLLS method calculated over three cycles. The mean period is significantly different between opn4xa+/+ and opn4xa-/- larvae in LL. Mean± sd (in hours) is represented. Each grey point represents a larva.

Fig 6. opn4xa -/- larvae show subtle modifications of a few clock genes in LL: RTqPCR performed on pools of 15 larvae for the gene indicated at the top of the figures.

Mean expression relative to beta actin ± s.d. Three pools of larvae were used for each time point. ‘wt’ refers to pool of larvae from crosses of opn4xa+/+ animals (siblings of the opn4xa-/- fishes used for the opn4xa-/-points). Larvae were exposed to LD cycles (until d6 21h) followed by a LL cycle (from d6 21h to d7 18h). The grey rectangles represent the night phase, the yellow rectangles represent the subjective night in the LL cycle. The data were analysed using two-way ANOVAs which revealed time-genotype interaction for bmal1a and cry1a, as well as statistical differences between genotypes for specific time points using Bonferroni post-hoc tests. p< 0.05; ** p< 0.001; ***p< 0.0005.

Fig 7. Larvae devoid of opn4xa-mediated photosensitivity (opn4xa-/-) still photoentrain to pulses of light at CT16 and CT21.

A) Experimental design of phase shift experiments. White rectangles represent the day or light pulse period, black rectangles represent the night period and dark grey rectangles represent the subjective day. For each experiment, larvae are entrained for 5 LD cycles the larvae are therefore 5dpf at the beginning of locomotor activity measurements. Locomotor activity is tracked either in constant darkness for 4 days (DD) or tracked in constant darkness for 4 days and subjected to a 2-hours pulse of light during the night of the 2nd day of constant darkness (PD). The phase of locomotor activity is calculated for each larva before and after the timing of the pulse for DD and PS experiments and the Δphase (phase after the pulse–phase before the pulse) is calculated. B) Average distance moved (mm/min over 10min) of 3 independent PD experiments. Mean ± SE. The Δphase of opn4xa+/+ and opn4xa-/- larvae calculated with the FFT-NLLS method is not significantly different. opn4xa+/+ and opn4xa-/-show similar levels of activity during the light pulse (p = 0.56, Mann-Whitney two-tailed test). C) Experimental design of phase advance (PA) experiments. The iconography is similar to A). PA-pulsed larvae were subjected to a one-hour pulse of light at CT21. D) Average distance moved (mm/min over 10min) of 3 independent PA experiments. Mean ± SE. The Δphase of opn4xa+/+ and opn4xa-/- larvae calculated with the FFT-NLLS method is not significantly different (Mann-Whitney two-tailed test).

Fig 8. Larvae devoid of RGCs and opn4xa-mediated photosensitivity still entrain to pulses of light at CT16 and CT21.

A) Experimental design of phase shift experiments. White rectangles represent the day or light pulse period, black rectangles represent the night period and dark grey rectangles represent the subjective day. For each experiment, larvae are entrained for 5 LD cycles the larvae are therefore 5dpf at the beginning of locomotor activity measurements. Locomotor activity is tracked either in constant darkness for 4 days (DD) or tracked in constant darkness for 4 days and subjected to a 2-hours pulse of light during the night of the 2nd day of constant darkness (PD). The phase of locomotor activity is calculated for each larva before and after the timing of the pulse for DD and PS experiments and the Δphase (phase after the pulse–phase before the pulse) is calculated. B) Average distance moved (mm/min over 10min) of 3 independent PD experiments (n = 27 for lak -/- referred as lak and n = 27 lak-/-; opn4xa-/- larvae referred as ‘double’). Mean ± SE. The Δphase of lak and double larvae calculated with the FFT-NLLS method are not significantly different (see Table 3). C) Experimental design of phase advance (PA) experiments. The iconography is similar to A). After 5 training cycles in LD, PA-pulsed larvae were subjected to a one-hour pulse of light at CT21. D) Average distance moved (mm/min over 10min) of 3 independent PA experiments (n = 27 for lak and n = 27 double larvae). Error bars represent SE. The Δphase of lak and double larvae calculated with the FFT-NLLS method are not significantly different (see Table 3).