Light-Driven Behavior in Larval Zebrafish without Eyes or Pineal(A) Attraction of control and enucleated larvae to a phototaxis stimulus, measured by the percent of larvae observed on the illuminated side of the testing arena over time. Enucleated larvae exhibit a gradual shift to the illuminated side of the arena (symbols show one-sample t test to 50%; n = 4 groups of 15 larvae). Larval positions were recorded every second and then averaged over 10 s for each time point. Color along x axis indicates light condition.(B) Larval body orientation during exposure to a phototaxis stimulus. A significant proportion of control larvae exhibit a “head-on” orientation toward the spotlight (one-way ANOVA; F_{7, 24} = 51.21, p < 0.001; comparisons are Tukey’s post hoc), whereas enucleated larvae show no bias in body orientation (ANOVA; F_{7, 24} = 1.73, p = 0.15; n = 4 groups of 15 larvae). Data represent mean proportion of larvae oriented relative to the target light over 1 min.(C) Locomotor activity during dark-induced VMR. Arrow indicates O-bend spike observed only in controls. Inset: enucleated larvae significantly increase activity following light extinction (repeated-measures ANOVA; F_{2.5, 88.7} = 16.57, p < 0.001; n = 36 larvae). Data represent the mean activity for the preceding minute. Color along x axis indicates light condition. Pairwise comparisons are to the baseline time point at 5 min.(D and E) Kinematic analysis of VMR. Enucleated larvae (red) retain elevated R-turn initiation frequency (D; repeated-measures ANOVA; F_{2.1, 56.6} = 4.63, p = 0.013) and swim bout duration (E; repeated-measures ANOVA; F_{3, 40} = 6.41, p = 0.001) as seen in controls. Data represent the mean of observations during the first 16 s following each time point. Pairwise comparisons are to the baseline measurement at 5 min (empty circles) (control: n = 18 groups of 10 larvae; enucleated: n = 28 groups of 10 larvae).(F) Diving and climbing speed during VMR. In either response, enucleated larvae were not significantly different from controls (dive: t test, p = 0.27; climb: t test, p = 0.07; n = 6 groups of 5 larvae). Additionally, speed in all conditions is significantly different from 0 (one-sample t test, p < 0.005). Data represent mean vertical swim speed over the first 20 s of dive and ascent.(G) Nitroreductase-mediated ablation of the pineal. Dorsal views at 6 days postfertilization (dpf) are shown. (i–iv) Epifluorescence images of the pineal (arrow) in untreated and metronidazole (Met)-treated Tg(tph2:NfsB-mCherry)y227 larvae with anti-mCherry (red; i and ii) and anti-RET-P1 (green; iii and iv). Scale bar represents 100 μm. (v and vi) Confocal z projections (mCherry + RET-P1) showing concurrent nitroreductase and opsin expression in the pineal in untreated (v) and Met-treated (vi) larvae. Scale bar represents 25 μm.(H) VMR in enucleated, pineal-ablated larvae. Both pineal-ablated and pineal-ablated + enucleated larvae show a robust VMR following light extinction (repeated-measures ANOVA; F_3.8,131.3 = 32.44, p < 0.001) (control and pineal-ablated + enucleated: n = 36 larvae; pineal-ablated: n = 26 larvae). Data represent mean activity for the preceding minute. Pairwise comparisons are to baseline time point 5 min.(I) Diving and climbing speed of enucleated, pineal-ablated larvae during VMR. In either response, lesioned larvae were not significantly different from controls (t test; dive: p = 0.26; climb: p = 0.42). Mean speed of dive and ascent for both groups is significantly different from zero (one-sample t test, p < 0.005; n = 14 groups of 5 larvae). Data represent mean swim speed over the first 20 s of dive and ascent.For all panels, error bars show SEM; p < 0.05, p < 0.01. See also Figure S1.

Reduction of VMR in otpa Mutants and Lack of Dopaminergic Contribution(A) Locomotor activity during dark-induced VMR of intact and enucleated otpa mutants and sibling larvae. Intact mutants show a response to light extinction (repeated-measures ANOVA; F_3.0,175 = 8.5, p < 0.01; n = 59 larvae) that is greatly reduced relative to controls (intact siblings: n = 47 larvae; enucleated siblings: n = 97 larvae). Without eyes, mutants lose any response to light extinction (repeated-measures ANOVA; F_3.2,317 = 1.72, p = 0.16; n = 101 larvae). Data represent mean activity for the preceding minute. Color along x axis indicates light condition. Pairwise comparisons to baseline time point at 0 min: p < 0.05, p < 0.01.(B and C) Kinematic analysis of photokinesis in intact otpa mutants. Otpa mutants retain O-bend responses to light extinction (B) (repeated-measures ANOVA; F_3,9 = 53.7; p < 0.001; n = 4 groups of 10 larvae) but do not show characteristic increases in R-turn initiation (C) (mutants, red; siblings, black) (F_3,9 = 2.1; p = 0.17; n = 4 groups of 10 larvae). #p < 0.05, p < 0.01 for pairwise comparisons to baseline at 5 min (empty circles). Data represent the mean and SEM of observations during the first 16 s following each time point.(D) Diving and climbing speed of intact otpa mutant larvae. Compared to siblings, otpa mutants exhibit significantly reduced diving speed during and climbing speed following a 60 s dark flash (t test; dive: p < 0.001; climb: p < 0.005; n = 3 groups of 8 larvae). Data represent mean and SEM swim speed over first 20 s of dive and ascent.(E) Nitroreductase-mediated ablation of dopaminergic (DA) neurons in Tg(BACth:Gal4VP16)m1233; Tg(UAS:EGFPCAAX); Tg(UAS-E1b:NfsB-mCherry) triple-transgenic larvae. Asterisk indicates mCherry aggregates remaining from ablated cells. Arrow indicates GFP-expressing nonablated cells. The following abbreviations are used: DC2 and DC4, Otp-dependent dopaminergic groups 2 and 4; PT, posterior tuberculum; cH, caudal hypothalamus. Dorsal view is shown. Scale bar represents 50 μm.(F) VMR in DA neuron-ablated larvae. Control Tg(th:Gal4VP16); Tg(UAS:EGFPCAAX) (black line) and DA neuron-ablated Tg(th:Gal4VP16); Tg(UAS-E1b:NfsB-mCherry) (red line) larvae show similar, robust VMR following light extinction (repeated-measures ANOVA; F_{12, 564} = 148.29, p < 0.001; n = 48 larvae). Data represent mean and SEM activity as in (A). Pairwise comparisons to baseline time point at 5 min: p < 0.05, p < 0.01.See also Figure S2.

opn4a Expression Depends on Otp Activity in Areas of Coexpression with otpa(A) Analysis of expression domains of opn4a, otpa, and TH (anti-TH) in 3 dpf wild-type larvae. Top row shows z projections of combined channels (see labels) of confocal stacks recorded from lateral (left) and dorsal (right) views of the brain. Anterior is at left; dorsal is at top for lateral view. Bottom row shows z projections of channel combinations of a confocal stack showing dorsal views of the brain. Scale bars represent 50 μm. opn4a is coexpressed with otpa in the anterior preoptic area (aPO) (asterisk) and posterior tuberculum (PT) (arrowhead). The following abbreviations are also used in (A), (B), and (D): Th, thalamus; pPO, posterior preoptic area; vH, ventral hypothalamus; DC2, dopaminergic group 2; DC4, dopaminergic group 4; ac, anterior commissure; oc, optic chiasm; PI, pineal; PT DA, posterior tuberculum dopaminergic neurons.(B) Expression of opn4a in otpa and otpb mutants. Whole-mount in situ hybridization reveals loss of opn4a expression in the aPO (asterisk) and PT (arrowhead) of otpa and otpa, otpb double mutants (3 dpf). otpb mutants alone did not significantly affect opn4a expression, which is in line with the previously reported compensation of otpb knockdown by otpa activity in A11 DA neuron differentiation [20]. Anterior is at left; dorsal is up. Scale bar represents 50 μm.(C) Increase in activity following decrements in light intensity in enucleated Tg(otpb.A:Gal4)zc67; Tg(UAS:GFP-v2a-opn4)y233 larvae. Data show the difference in mean activity between 2 min after light change and 1 min prior to light change (mean displacement at t2  mean displacement at t1). Enucleated opn4-overexpressing larvae show an increased response to decrements in light intensity (repeated-measures two-way ANOVA; F_1,100 = 8.84, p < 0.005; nonexpressing control: 28 larvae; GFP positive in otpb domain: n = 42 larvae). p < 0.001.(D) Schematic summarizing our results suggesting that preoptic opn4a-expressing neurons are deep brain photoreceptors driving dark photokinesis. We eliminated all depicted photoreceptive regions except the PO and found that the VMR response remained intact. Because otpa mutants lack aPO but not pPO opn4a expression, and because mutants without eyes do not show a VMR, the behavior must originate in the opn4a-positive cells in the aPO (pink). The illustrated DA domain (red) only comprises the diencephalic groups 2–6.See also Figure S3.

Characterization of VMR, Related to Figure 1
(A-C) The VMR persists in blind larvae. (A) Maternal zygotic headless mutants (MZHdl, 7 dpf) and (B) retinal ganglion cell ablated larvae Tg(Atoh7:Gal4)s1992t ; Tg(UAS-E1b:NfsBmCherry) (6-7 dpf) and (C) chokh mutants (5 dpf): (i) allele: t25327 and (ii) allele s399. Hdl: repeated measures ANOVA: F7, 21 = 4.26, p = 0.005. Sibling control: n = 3; Hdl: n = 4 groups of 20 larvae. Tg(Atoh7:Gal4)s1992t ; Tg(UAS-E1b:NfsB-mCherry: ANOVA: F12, 276 =14.893, p < 0.001. Control (Tg(Atoh7:Gal4)s1992t ; Tg(UAS:Dendra-Kras) [1]) (n = 24 larvae). Chokh^t25327: repeated measures ANOVA: F12,259 = 30.64, p < 0.001, n = 12 larvae. Sibling control n = 20 ; chokht25327 n = 20 larvae. Chokh^s399 : repeated measures ANOVA: F12,129 = 15.50, p < 0.001. Sibling control: n = 16 larvae; chokh: n = 10 larvae. Timepoints show mean activity for the preceding minute (mean ± SEM). For pairwise comparisons to baseline time point -5 min. # p < 0.05, *p < 0.01. Inset in (A) shows MZHdl larvae and siblings and in (B) shows metronidazole treated Tg(Atoh7:Gal4)^s1992t larvae with visual background adaptation phenotype after nitroreductase ablation of retinal ganglion cells (6 dpf), and (C) chokh mutant larvae prescreened for locomotor responsiveness prior to behavioral experiments and siblings (5dpf).
(D) Confocal z-projection showing a sagittal view of the pineal in Tg(tph2:NfsB-mCherry)y227 (2 dpf). Anti-mCherry (red) shows cell bodies of photoreceptor neurons at the pineal base and anti-RET-P1 (green) highlights photosensitive outer segments. Scale bar 25 μm.
(E) Kinematic analysis of VMR in control and enucleated larvae (7 dpf). Enucleated larvae do not perform O-bends in response to light extinction as is seen in control larvae (control: n = 18 groups of 10 fish ; enucleated: n = 28 groups of 10 fish). Statistical comparisons are to the baseline measurement at -5 min. Data represents the mean of observations during the first 16 s following each time point (mean ± SEM).
(F-H) Kinematic analysis of wild-type larvae during VMR and subsequent dark adaptation. Following light extinction, R-turn initiation frequency (F) peaks at approximately 5 min. (VMR) then declines to a low baseline (dark adaptation) (ANOVA; F3.2,28.3 = 24.37, p < 0.001). In contrast, both R-turn swim bout duration (G) and swim tail beat magnitude (H) continue to steadily increase over 30 min (bout duration: ANOVA; F1.9,13.3 = 13.27, p = 0.001); swim magnitude: ANOVA; F7,49 = 4.46, p = 0.001). Statistical comparisons are to the baseline measurement at -5 min. Data represents the mean (± SEM) of observations during the first 16 s following each time point. For pairwise comparisons to -5 min *p < 0.05, **p < 0.01 (n = 10 groups of 25 larvae).
(I and J) Validation of automated kinematic measurements (by Flote software) of larval R-turn swim bouts via comparison with manual observations.
(I) Trace shows body curvature of a single larvae during an R-turn (blue line = raw measurement, black line = smoothed data). Green diamonds show time points during the swim bout identified by manual inspection of the video. Red diamonds show the same points identified by software. C1 and C2 refer to the initial bend and counterbend (for consistency with terminology during startle responses) and S 1/2 refer to the first two tail bends during subsequent swimming. Swim bout duration is marked, and swim magnitude is the mean of the change in larval curvature from C1-S1 and S1-S2 (mean of two marked amplitudes). Tail beat frequency is calculated from the time taken for C2-S2, which represents a single complete tail movement.
(J) Comparison of automated and manually identified time points during swim bouts for 50 Rturns. The largest discrepancy is in identification of swim bout termination, which Flote underestimates by about 10 ms due to the presence of subtle tail undulations at the end of some R-turn swims. The mean swim bout duration for R-turns did not differ significantly between automatically and manually recorded observations (Flote: 120.3 ± 3.6 ms; Manual: 113.9 ± 4.3 ms ; F1,98 =1.28, p = 0.26).
(K) Vertical movement of control larvae (6-7 dpf) during VMR in a 30 mm deep column of water (E3). Larvae maintain position close to the water surface under lit conditions. After light extinction, larvae dive downward. Mean vertical position of larvae in groups of 10 per trial was recorded every second (shaded line represents ± SEM)(n = 5 groups of 5 larvae).

VMR in Larvae with Ablated Ventral Diencephalic Otp-Expressing Neurons and DA Neuron Ablation Efficiency, Related to Figure 2
(A) Nitroreductase-mediated ablation of a population of preoptic, posterior tubercular and hypothalamic Otp-expressing neurons in Tg(otpb.A:Gal4)zc67 ; Tg(UAS:Nfsb-mCherry) larvae (control Tg(otpb:Gal4); Tg(UAS:NfsB-mCherry) no metronidazole (5dpf). Dorsal view, anterior at left. Scale bar is 50 μm.
(B) NTR-mediated ablation of opn4a cells in Tg(otpb.A:Gal4)zc67 ; Tg(UAS:Nfsb-mCherry) transgenic larvae (5dpf). Top row shows Tg(otpb:Gal4); (UAS:Nfsb-mCherry) control larvae (no metronidazole), mCherry is co-expressed with a subset of opn4a in the anterior preoptic area (aPO) and posterior tuberculum (PT). Bottom row shows ablated fish. In ablated larvae reduction of opn4a cells in the aPO and PT region is visible (arrowhead). Dorsal view, anterior to the left. Scale bar is 50 μm.
(C) VMR in Tg(otpb.A:Gal4)zc67 ; Tg(UAS:Nfsb-mCherry) neuron ablated larvae (6-7dpf). Ablated larvae (red lines) show reduced VMR compared to controls (black line) (ANOVA; F12,276 =17.42, p < 0.001) (n = 24 larvae). Larval activity was recorded every 10 seconds and averaged over 1 minute intervals. For pairwise comparisons, *p < 0.05, **p < 0.01.
(D) Metronidazole-mediated ablation of posterior tubercular dopaminergic neurons in 7 dpf Tg(BAC th:Gal4VP16)^m1233 BAC transgenic line. Dopaminergic neuron groups 2 and 4 are strongly reduced in ablated larvae when compared to control larvae (Tg(BAC th:Gal4VP16) ^m1233; Tg(UAS: EGFPCAAX)^m1230 ) as seen by anti-Tyrosine hydroxylase immunostaining. Groups 3, 5, 6 and 7 dopaminergic neurons, arch associated neurons, area postrema, very few amacrine cells and a subset of Locus coeruleus neurons are also labeled by the transgene and therefore also ablated by using this line. Other TH neurons are not ablated. Dorsal view, anterior to the left. Scale bar is 50 μm.
(E) Quantification of loss of posterior tubercular dopaminergic groups 2 and 4 in metronidazolemediated cell ablation of 7 dpf Tg(BAC th:Gal4VP16) ^m1233 transgenic line. Control Tg(BAC th:Gal4VP16) ^m1233; Tg(UAS:EGFPCAAX)^m1230 (dark blue, n = 33 larvae) and DA neuron ablated larvae Tg(BAC th:Gal4VP16) ^m1233; Tg(UAS-E1b:NfsB-mCherry)) (light blue, n = 34 larvae). Quantification reveals a strong reduction of dopaminergic groups 2 and 4 in more than 80% of the ablated larvae. Average percentage and 95% confidence intervals are shown.

Expression of Nonvisual Opsins in the otpa Domain and Expression of opn4a in Wild-Type and otp Mutant Embryos, Related to Figure 3
(A) Expression on opn4a in wild-type and otpa mutant embryos. opn4a expression was analyzed in 3dpf larvae by whole-mount in situ hybridization. Top panels: dorsal views of individual focal planes of the larval head in dorsal to ventral order. The otp mutant genotypes are indicated at top. otpa and otpa otpb double mutant larvae reveal loss of opn4a expression in the aPO (anterior preoptic region) and PT (posterior tuberculum). pPO-posterior preoptic region, vHventral hypothalamus. Dorsal views. Scale bar is 50 μm. Bottom panels: frontal sections of a wild-type larval head in rostral to caudal order reveal the anatomical position of opn4a expressing neurons (see also [2]). ac-anterior commissure; aPO-anterior preoptic region; cHcaudal hypothalamus; DA DC2 / DA DC4 - Otp dependent posterior tubercular dopaminergic groups DC2 (rostral) and DC4 (caudal); PT-posterior tuberculum; vH-ventral hypothalamus. Frontal views, dorsal at top. Scale bar is 50 μm.
(B) Presence of neurohormone producing cells in the preoptic region in wild-type and otpa mutants. Expression of crh, vsnp, itnp, trh, and sst1 at 3dpf in wild-type embryos and otpa^m866 mutants. Whole-mount in situ hybridization (3dpf dpf) reveals that otpa mutants do not show reduced formation of neurohormone producing cells in the preoptic region. Dorsal view. Scale bar is 50 μm.
(C) Expression of “nonvisual” opsins in wild-type and otpa mutants. Expression of tmtopsb, valopa, opn3-like, opn3 at 3dpf in wild-type embryos and otpa^m866 mutants. Whole-mount in situ hybridization (3 dpf) reveals that otpa mutants do not show reduced expression of the analyzed nonvisual opsins. Dorsal view. Scale bar is 50 μm.
(D) tmtopsa expression in preoptic regions does not depend on otpa activity. Expression of tmtopsa in the preoptic region and hindbrain. Whole-mount in situ hybridization (3 dpf) reveals otpa mutants do not show reduced tmtopsa expression. Dorsal view. Scale bar is 50 μm.
(E) tmtopsa expression does not co-localize with otpa domain. Analysis of expression domains of tmtopsa and otpa in 3 dpf wild-type larvae. z-projections of a confocal stack of individual and combined channels (see labels), showing dorsal view of brain (anterior at left, scale bar in left panel is 50 μm). tmtopsa is expressed in cells near the otpa domain but do not overlap.
(F) Change in activity (visual motor response) following decrements in light intensity in intact (black line) and enucleated (red line) wild-type larvae (7 dpf). Timepoints show difference in mean activity between 2nd min after light change and 1 min prior to light change (mean displacement at t2 – mean displacement at t-1). Intact and enucleated larvae show similar response curves with increasing responses to greater decrements in light intensity (n = 17 larvae). Enucleated: F1,16 = 6.9, p < 0.05; intact: F1,16 = 4.62, p < 0.05. Pairwise comparisons to -3 fold; open circles indicate significance: p < 0.01.
(G) opn4a-expressing cells in the aPO region are not affected in chokh mutant embryos. Analysis of expression domains of opn4a, otpa (WISH) and TH (anti-TH) in 3 dpf chokh^s399 mutant larvae. z-projections of confocal data stacks showing dorsal views of the ventral diencephalon (anterior at left; scale bar 50 μm). Both the opn4a and otpa coexpressing cell groups in the anterior preoptic area (aPO) and posterior tuberculum (PT) are still present in chokh mutant embryos.