Search strategy is regulated by somatostatin signaling and deep brain photoreceptors in zebrafish

Abstract

Background: Animals use sensory cues to efficiently locate resources, but when sensory information is insufficient, they may rely on internally coded search strategies. Despite the importance of search behavior, there is limited understanding of the underlying neural mechanisms in vertebrates.

Results: Here, we report that loss of illumination initiates sophisticated light-search behavior in larval zebrafish. Using three-dimensional tracking, we show that at the onset of darkness larvae swim in a helical trajectory that is spatially restricted in the horizontal plane, before gradually transitioning to an outward movement profile. Local and outward swim patterns display characteristic features of area-restricted and roaming search strategies, differentially enhancing phototaxis to nearby and remote sources of light. Retinal signaling is only required to initiate area-restricted search, implying that photoreceptors within the brain drive the transition to the roaming search state. Supporting this, orthopediaA mutant larvae manifest impaired transition to roaming search, a phenotype which is recapitulated by loss of the non-visual opsin opn4a and somatostatin signaling.

Conclusion: These findings define distinct neuronal pathways for area-restricted and roaming search behaviors and clarify how internal drives promote goal-directed activity.

Keywords: CRISPR; Goal-directed behavior; Melanopsin; Motivation; Non-visual photoreceptor; Orthopedia; Search; Somatostatin; Zebrafish; opn4a; sst1.1.

Fig. 1

Changes in 3D swim trajectories after loss of illumination. a Diagram of 3D recording set up. A mirror was positioned adjacent to arena in order to simultaneously view XY and Z planes. b Representative 30-s path trajectories of larvae (red traces) during illuminated baseline conditions (left panel), immediately (middle panel) or 3–5 min after loss of light (right panel). Black traces show 2D projections in YZ and XY planes. Chamber size: 85 × 85 × 75 mm. c Mean displacement per 30-s recording period in XY and Z axes during baseline swimming (N = 10), during the first minute (T₀; N = 25) and 3–5 min after loss of illumination (T_3–5; N = 13). * P < 0.05 versus baseline. d Mean position of larvae in 80-mm deep chamber during baseline illumination (yellow background) and sustained darkness (grey background). N = 30 groups of five larvae. * P < 0.05 versus baseline mean

Fig. 2

Locally restricted swimming after loss of illumination. a Representative path trajectories for a larva during full field illumination (left panel) and during the first 10 min after loss of illumination (right panel). Arrowheads denote starting position and numbers indicate time (min). Color scale indicates fractal dimension in 30 s windows. Main scale bar 10 mm, chamber is 200 × 200 mm. Arrowhead length equivalent to larva length. b–e Displacement, meander, fractal dimension, and trajectory bias for larvae measured during 10 min of full field illumination (yellow), followed by 10 min darkness (black). N = 32. * P < 0.05, paired t-test compared to corresponding baseline time-point. See Methods for description of measurements

Fig. 3

Local search is generated by increased utilization of same direction R-turn maneuvers. a Representative 10-s path trajectories of individual larva during baseline (left, yellow) and dark (right, grey) conditions. Arrows indicate path direction of the larva. Scoot (black circle) and R-turn (red circle) maneuvers are indicated along each path. Lock index (LI) for each recording noted at bottom. Scale bar 2 mm. Frequency (b, d) and LI (c, e) of maneuver pairs during full-field illumination (b, c; baseline) and during the first 10 s after loss of illumination (c, d; T₀). Baseline N = 39, T₀ N = 55. Maneuver frequency analysis excludes O-bends triggered by the sudden reduction in illumination and infrequent fast burst swims. c Baseline maneuver pair LIs are significantly increased over 0 for scoot-scoot (one-sample t-test vs. 0, t₃₉ = 8.36, P < 0.001), turn-turn (t₂₄ = 2.44, P = 0.022), and turn-scoot pairs (t₃₉ = 4.51, P < 0.001). Scoot-turn maneuver pair LI was not significantly different from 0 (t₃₈ = 1.01, P = 0.32). f Frequency of R-turn initiations (percentage of larvae that execute an R-turn per 400 ms analysis window). N = 17 groups of 10 larvae each. * P < 0.001. g LI for sequential R-turns during sustained loss of illumination. N = 51 larvae (baseline), 24 (5 s), 22 (60 s), 15 (300 s), 10 (600 s). * P < 0.05 compared to baseline time-point

Fig. 4

Light reverses changes in locomotor profiles during local and outward swimming. a Rate of re-orienting (meander) for larvae measured for 5 min after loss of illumination, either in constant darkness (black circles, N = 25), or when illumination was restored after 30 s darkness (orange triangles, N = 22). * P < 0.05 for corresponding time-points in sustained dark and after re-activation of the light. b Meander for larvae measured for 10 min after loss of illumination, with darkness maintained (black triangles) except during 1 min, when the light was re-activated (orange triangles). N = 19. * P < 0.05 compared to constant illumination. Dashed lines in (a, b) indicate mean meander for larvae measured during constant illumination

Fig. 5

Local and outward movement strategies differentially facilitate locating local and remote light. a Schematic of covert phototaxis assay. The light spot (14 mm diameter) was projected 7 mm directly behind a freely swimming larva. We then measured the time for the larva to enter the light spot. b Time to light spot perimeter when activated either 2 s (T₀) or 3–5 min (T₄) (mean 248 ± 19.6 s, depending on when larvae entered the motion trigger region of interest (ROI)) after loss of full field illumination. Controls: no target light. N = 33 (T₀ control), 40 (T₀ with spot), 18 (T₄ control), 20 (T₄ with spot). c Turn maneuver trajectories for larvae oriented with the light spot to their right, when tested 2 s (T₀) or 3–5 min (T₄) after loss of full-field illumination. d R-turn direction bias during phototaxis relative to a static light spot. Light spot (9 mm diameter, intensity of 20 μW/cm²) was illuminated either 2 s (grey; N = 22 groups of 15 larvae) or 3 min (black; N = 20 groups) after loss of full field illumination. Orientation of larvae relative to the target spot is indicated. * P < 0.05 between T₀ and T₄ orientation-matched groups. Bias is the proportion of R-turns directed toward the target spot, normalized between –100 (consistently away from target) to +100 (always toward target). e Phototaxis in a large area (200 × 200 mm) using a 55-μW light spot. Representative swim trajectories for larvae tested when the light spot was activated 1 s (T₀, grey traces) or 3–5 min (T₄, black traces) after loss of illumination. Arrowheads indicate start positions. Box plot shows closest approach to the light spot for larvae tested 1 s (T_0, grey) or 3–5 min (mean 247.7 ± 21 s; T_4, black) after loss of illumination, and for trials where the light spot was not activated (No spot). N = 11 larvae (dark, T₀), 13 (dark, T₄). f Representative swim trajectories for larvae tested when the light spot was activated 1 s (T_0, grey traces) or 3–5 minute (T_4, black traces) after loss of illumination. Arrowheads indicate start positions. g Quantification of concealed target test. Closest approach to the light spot for larvae tested 1 s (T_0, grey) or 3–5 min (mean 236 ± 13 s; T_4, black) after loss of illumination with either no target light (No spot), or targets of the indicated intensities. Horizontal black line represents position of barrier. N = 12 larvae (no spot, T₀), 12 (no spot, T₄), 10 (15 μW, T₀), 11 (15 μW, T₄), 12 (55 μW, T₀), and 9 (55 μW, T₄). # P < 0.05, * P < 0.001 between T₀ and T₄ groups. Experiment was performed as for (e) except with an interior barrier

Fig. 6

Retinal input is required to initiate local but not outward movement. a Lock index for R-turns during baseline and immediately following loss of light (T₀) during 10 s recordings, for sham operated (ctrl) and enucleated (enuc) larvae. N = 41 larvae (ctrl, baseline), 53 (ctrl, T₀), 44 (enuc, baseline), 59 (enuc, T₀). * P < 0.001. b, c Fractal dimension (b) and meander (c) of path trajectories during dark response for control (black circles, N = 29) and enucleated larvae (grey circles, N = 34). * P < 0.05 for control compared to enucleated larvae. Dashed lines show mean values for enucleated larvae under full-field illumination. Inset: Representative traces of first 2 min of dark for control (left) and enucleated (right) larvae. Scale bar 20 mm. d Mean meander for enucleated larvae during full-field illumination (base) compared to the first 2 min after loss of illumination (0–2) and from 3–10 min after loss of illumination (3–10). * P < 0.001, paired t-test compared to baseline

Fig. 7

Loss of somatostatinergic neurons and opn4a expressing deep brain photoreceptors potentiates local search. a R-turn lock index for otpa wildtype siblings and mutants during full-field illumination (base) and immediately following loss of illumination (T₀). N = 25 larvae (sibs, base), 27 (sibs, T₀), 47 (mutants, base), 38 (mutants, T₀). b Representative path trajectories for an otpa mutant larva during 10-min recording periods during baseline (orange trace) and after loss of illumination (black trace). Arrowheads denote starting positions. Chamber: 200 × 200 mm. Scale bar 2 mm. Color represents fractal dimension. c Fractal dimension of path trajectories for otpa wildtype siblings (black circles, N = 29) and mutants (grey circles, N = 31). * P < 0.05 for mutants versus siblings. Dashed line shows mean for mutants under full-field illumination. d Schematic diagram showing neuronal cell types within Orthopedia expression domain. Labeled neuron types correspond to the neuronal markers with reduced expression in the otpa mutant background. e Mean fractal dimension during 3–5 min following loss of illumination. Control group was injected with sgRNA against GFP in Tg(vglut2a:EGFP) transgenic larvae. Controls N = 26; otpa N = 31; mi174 N = 13; trh N = 37; sst1.1  N = 22; opn4a N = 36; valopa N = 6. * P < 0.05 for mutant groups versus control group. f Representative path trajectories of individual larva for a 10 minute duration following loss of illumination. Top left: CRISPR injected control; Top right: trh; Bottom left: sst1.1; and Bottom right: opn4a. Arrowheads note starting position at time of light extinction. Chamber: 200 × 200 mm. Color represents fractal dimension, with scale as for (b)

Fig. 8

Model for induction of light search behavior by retinal and deep brain photosensory systems. a Loss of light detected via the retina drives an initial strong local search (red). Simultaneous stimulation of opn4a and sst1.1 signaling drives outward locomotor patterns for remote light sources (blue). Local search activity initially masks extended search locomotor features. b In enucleated larvae, lack of retinal drive allows remote light-search patterns to emerge immediately after loss of illumination. c In the absence of opn4a and sst1.1 to promote outward search in otpa mutants, retinal signaling continues to drive local search patterns for a longer period of time, as outward search drive is absent