|
FIGURE 1. Method and comparison of swimming behavior in two different experimental conditions. (A) Free tadpole swimming was recorded by a camera placed above the tank. Then homemade imageJ macro was used to track eye position frame by frame and create body skeleton (light pink dot line). This macro provided XY coordinates of each point (color dots), used to calculate right/left angular excursion of the first tail section (1st sct, green line) and the body axis (body ax., pink line), and between each tail section. (B,C) Semi-intact preparations used to quantify the oculomotor behavior induced by swimming. To avoid any visuo-vestibular sensory inputs, optic cranial nerves were cut and the tadpole’s head was fixed on the Sylgard by pins closed to its otic capsules (otic c.). The tail was free to swim in a deeper compartment. Homemade software tracked lateral movements and measured lateral angle values for each eye (eye ROI) between eye axis (minor ax.) and head axis (head ax.). Coordinates of each tail section (green, gray and blue dot) were calculated from tail ROI. (D) Mean undulation frequency of the first tail section during swimming was significantly lower (Dunn’s multiple comparisons test, p ≤ 0.05) in head fixed (black bar, Fixed h., n = 7) preparations than in free swimming condition (dark gray bar, Free s., n = 5). Mean bursting frequency recorded in spinal rostral ventral root (light gray bar, in vitro, n = 11) was also lower than mean undulatory frequency of the first tail section measured in semi-intact or free swimming animals (Dunn’s multiple comparisons test, p < 0.001). (E) Averages and range of peak-to-peak tail amplitude movements were similar (Mann–Whitney test, ns) in both experimental conditions.
|
|
FIGURE 2. Temporal relationship between tail sections during larval swimming. (A) During free swimming (Ai) the sinewave-like undulatory movements of the tail were generated by alternative left and right excursion of 1st tail section (green trace, 1st sct) spread to next caudal sections. Tail sections were video-tracked frame by frame. Traces show angular excursion of each tail sct (in degree) over time. Cycle analysis (25 cycles) shows that the 4th tail section (orange trace, 4th sct) was totally out of phase with the 1st section (1st sct, green trace). (Aii) Mean latency between oscillations of the 1st and consecutive more caudal tail sections increased gradually and significantly (Kruskal–Wallis test, p < 0.01, n = 5). Therefore, the phase shift relative to the 1st section (green line) increased significantly (Watson–Williams test, p < 0.0001, n = 5. The 4th section was significantly out of phase (orange arrow) with the 1st tail section (V-test, p < 0.001, n = 5). (B) In semi-intact preparations, overlapping traces (Bi, 50 cycles) showed also that the 4th tail section (orange trace) was in phase opposition with the 1st tail section (green trace). (Bii) Mean latency between oscillations of the 1st and consecutive more caudal tail sections increased gradually and significantly (Kruskal–Wallis test, p < 0.01, n = 6). Therefore, the phase shift relative to the 1st section (green line) increased also significantly (Watson–Williams test, p < 0.0001, n = 5). The 4th section was significantly out of phase (orange arrow) with the 1st tail section (V-test, p < 0.001, n = 5). (C) Simultaneous extracellular recordings of spinal ventral roots (Vr) of brainstem-spinal cord preparations isolated in vitro showed a rostro-caudal delay typical for fictive swimming. The 20th Vr (orange trace) was consequently totally out of phase with the 5th Vr (green trace, 50 cycles). (Cii) The delay between the 5th root’s bursts and the next caudal Vrs (10, 15, and 20 successively) increased significantly (Kruskal–Wallis test, p < 0.01, 180 cycles) and consequently the phase shift of bursts recorded in Vrs 10, 15, and 20 relative to Vr-5 (green line) also increased significantly (Watson–Williams F-test, p < 0.0001, 180 cycles). Vr-20 (orange arrow) was significantly in phase opposition with Vr-5 (V-test, p < 0.0001, 180 cycles). (Ciii) On different in vitro preparations, latency and phase relationships of Vr-20 relative to Vr-5 bursts (Vr-20 vs. Vr-5) were calculated for fictive swimming frequencies ranging from 4 to 10 Hz. The latency didn’t change (Kruskal–Wallis test, ns, n = 7), whereas the phase relationship increased as the Vr-5 frequency was raised (Watson–Williams F-test, p < 0.01, n = 7).
|
|
FIGURE 3. Propulsive swimming behavior triggered locomotor-induced eye movement. (A) Tadpole swimming (freely behaving) captured on camera during strong propulsive swimming which provoked angular head movements (head trajectory, dark dot line) (Ai) or during weak swimming without oscillatory head movements (Aii). During strong propulsion (Ai), the 1st tail section (green trace, 1st sct) undulated in phase opposition with the 4th tail section (orange trace, 4th sct). In contrast, during slow swimming (Aii), the more rostral (1st) tail section showed no oscillation. (B) When the 1st tail section (green trace) undulated in head fixed condition (dark gray area), related to strong swimming, angular excursions of the 1st tail section were always propagated to 4th tail section, and appeared also associated with conjugated eye movements (red and blue traces, respectively, left and right eyes) with a weak shift phase (see left overlapping traces, 50 cycles). In contrast, locomotor-induced eye movements didn’t occur when the 4th tail section undulated alone (see light gray area and right overlapping traces, 50 cycles). (Bii) The magnitude of lateral eye angular excursion was linearly correlated with the amplitude of the 1st tail section undulatory movements with an average gain close to 0.5 (green circle and line, r2 = 0.46; s = 0.51 ± 0.08) during strong propulsive swimming. Therefore, the correlation between the amplitude of eye angular excursion and the amplitude of the 4th tail section movements was also linear with a gain highly similar (orange square and line, r2 = 0.21; s = 0.47 ± 0.16) in the same condition. In contrast, the amplitude of eye movements was not correlated with the amplitude of 4th tail section movements when tadpole generated slow swimming (gray square, r2 = 0.02; s = –0.02 ± 0.02). (C) Simultaneous recordings of the right lateral rectus (RLR), the 5th contralateral spinal ventral root (LVr-5) and the 20th left spinal ventral root (LVr-20) activities from an isolated in vitro preparation (Ci), during a strong (dark gray area) or weak (light gray area) swimming. Integrated traces (dark lines) were superimposed on raw traces (light traces). Bursting activities recorded on RLR (blue trace) occurred in phase with LVr-20 (orange trace) and in phase opposition with LVr-5 (green trace) (see mean overlapped traces on left, 120 cycles). (Cii) Burst areas of LR were strongly linearly correlated (r2 = 0.80; s = 0.83 ± 0.04) with Vr-5 burst area during propulsive swimming (green circle and line). Conversely during weak swimming, only LVr-20 discharged (mean overlapped traces on right, Ci, 41 cycles) and thus LR burst areas were weakly correlated with the Vr-20 bursts area (orange square and line, r2 = 0.18; s = 0.11 ± 0.01).
|
|
FIGURE 4. Conjugate eye movements generated by propulsive swimming compensate mid-caudal tail undulation. (A) In a head fixed preparation, the left and right eyes (respectively, red, and blue traces) rotated simultaneously in phase with undulation of the 1st tail section and in phase opposition with the 4th tail section. Latency between the peak of eye movements and the 1st tail section (green bar) was significantly lower (Mann–Whitney test, p < 0.05, n = 5) than latency between movements of the eyes and the 4th section of the tail (orange bar), resulting in a significant difference in phase relationship between the 1st and the 4th tail section relative to eye movement (polar plot). (B) Simultaneous recordings of right eye (Reye, blue trace) movements and activity in the left lateral rectus (LLR, red trace) nerve and the 5th (RVr-5) and 20th (RVr-20) right spinal ventral root during in vitro fictive swimming. (Bi) Real (blue) and fictive (red) leftward excursions of the eyes were in phase lag with bursting discharge in the 5th Vr and in reduced phase lead with the 20th Vr, compatible with a compensatory eyes movement during mid-caudal tail undulation (Vr-20). (Bii) Absolute average latency between the Reye movement (blue bar) or RVr-20 bursts (orange bar) and LLR firing were comparable, but significantly lower (Dunn’s multiple comparisons test, p < 0.0001, 61 cycles) than average latency between RVr-5 and LLR bursting (green bar). A significant difference is observed between the phase lead of the 5th ventral root bursting relative to that of the LLR and the phase lag of Reye movement and RVr-20 bursting relative to LLR firing (Watson–Williams F-test, p < 0.001, 61 cycles). (Biii) On different in vitro preparations, the absolute mean latency between RVr-20 and LLR was weak (12.93 ± 4.31 ms) and significantly lower (Mann–Whitney test, p < 0.01, n = 8) than the absolute mean latency between RVr-5 and LLR (37.11 ± 2.74 ms). Therefore, bursting discharges of LLR and RVr-20 (orange arrow) were nearly in phase (37.92°, 0.689, p < 0.001) while bursting discharges of RVr-5 (green arrow) were in phase lead with the LLR (274.41°, 0.65, p < 0.001). These temporal relationships were significantly different (Watson–Williams F-test, p < 0.0001, n = 8). (Biv) The latency between LLR and RVr-5 bursts (left curve) remained constant for fictive swimming frequencies ranging from 4 to 10 Hz (Kruskal–Wallis test, ns, n = 8). According to those latency results, phase shift decreased significantly with increasing fictive swimming frequency (Watson–Williams F-test, p < 0.05, n = 8).
|
|
FIGURE 5. Temporal relationship of spino-extraocular motor coupling is only adjusted in rostral spinal CPG. (Ai) Schematic of the in vitro preparation showing the recorded nerves and the experimental condition. (Aii-Control) Recording of spontaneous coordinated bursting discharge from left lateral rectus (LLR, red traces), and two contralateral spinal ventral roots (the 5th: RVr-5, green traces and the 20th: RVr-20 orange traces) during fictive swimming. Each raw trace was integrated and the result superimposed on the corresponding raw trace. Application of 10% sucrose on spinal segments more caudal than the 12th spinal ventral root (see schematic) blocked their activities as revealed by the Vr-20 recording without changing the temporal relationships between RVr-5 and LLR. (Aiii) Absolute latency average between LLR and RVr-5 did not differ (two-way ANOVA, ns, n = 3) in control and in sucrose condition on a range of fictive swimming frequency (4–10 Hz). Consequently, for all swimming frequencies, the phase relationship between the RVr-5 and the LLR was not significantly different in control compared to sucrose condition.
|
|
FIGURE S1, Supplemental data for Figure 4 | Histograms and Polar plots showing the time relationships (latency and phase) between the 1st tail section (sct) and the Leye and 4th section in semi-intact preparations (A); between the RVr-5 and Reye, LLR and RVr-20 (B) recorded in vitro; between RVr-5 and LLR or RVr-20 (C) and between RVr-5 and LLR (D) The four panels are, respectively, equivalent to Aii, Bii, Biii, and Biv of Figure 4 but by taking either the 1st sct (A) or the RVr-5 (B–D) as the phase marker. A; Mann Whitney test, p < 0,01, n = 5. B; histogram statistics: Dunn’s multiple comparisons test, p < 0.01 and p < 0.0001; polar plot statistic: Watson–Williams F-test, p < 0.001; 61 cycles. C; histogram statistics: Mann–Whitney test, p < 0,05, n = 10 and n = 7; polar plot statistic: Watson–Williams F-test, p < 0.001. D; left side plot: Kruskal–Wallis test, ns, n = 8; right side plot: Watson–Williams F-test, p < 0.05, n = 8.
|
