Connectivity of inhibitory neurons implicated in the laterality of the escape response.

(A) The blue region marked on the drawing of the fish contains the two Mauthner cells and neuronal candidates for their control, which may determine whether the animal escapes to the left or right. Right side shows some known contacts in this network, with sensory inputs from the eighth nerve exciting the ipsilateral M-cell as well as inhibitory interneurons that inhibit both M-cells. (B1) A cross section through the hindbrain in the region of the Mauthner cell from a transgenic line with glycinergic neurons labeled with GFP (green). The two Mauthner cells (blue) and a filled feedforward (FF) inhibitory interneuron (red) were labeled via patch pipettes. The FF neuron lies at the bottom of the most lateral of three columns (arrowheads) of glycinergic neurons, near the lateral dendrite of the M-cell. (B2) A horizontal view of the region shows that red processes of the FF neuron are located in the vicinity of both the ipsilateral and contralateral M-cells. Optical sections through the region of the contralateral (B3) and ipsilateral M-cells (B4) show swellings (arrowheads) of the processes from the inhibitory FF cell adjacent to both M-cells. (B5) A plot of the number of boutons adjacent to ipsilateral versus contralateral M-cells from 14 fish (4 days old) in which individual FF cells and both M-cells were labeled. In every case, the number of boutons apposed to the contralateral cell exceeded the number apposed to the ipsilateral one, with a highly significant difference between the two sides (p<0.0001). (C1) Triple patch recordings from an FF neuron and the two M-cells show that firing the FF cell (asterisk) by current injection produces larger IPSPs in the contralateral M-cell than the ipsilateral one (arrowheads). IPSPs from four sweeps are shown at potentials above and below resting potential by the injection of plus or minus 1 nA of current into the M-cells. This electrophysiology is from the neurons whose morphology is shown in B. (C2) A plot of the amplitude of the IPSPs (for negative 1 nA injection) in the ipsilateral versus contralateral M-cell from 19 triple patch experiments like the one in C1. Seven of these were from 6 days old nacre fish and 12 from 4 days old relaxed fish. In 18 of 19 experiments, the IPSPs were larger in the contralateral M-cell, with a highly significant difference between the two sides (p<0.001).

Connections between FF inhibitory neurons on the two sides.

(A) Patching of pairs of FF neurons on opposite sides of the brain. In this case, firing the right cell (Right FF) led to an IPSP in the FF on the left side (Left FF) (middle panel), but the left cell did not produce a response in the right one (right panel). B1 and B2 show horizontal and cross sections respectively, of the neurons recorded in A after filling them with dye. Green processes from the right neuron give rise to swellings apposed to the left neuron (arrowheads on insets in B1,B2, which show the soma at higher magnification), consistent with the physiological connection. (C) Paired patch recordings of two reciprocally connected FF neurons. Firing the right neuron led to a depolarizing response in the left cell and vice versa (dark traces). Both responses were blocked by strychnine (1 µM, gray traces), consistent with the PSP being a glycinergic inhibition with a reversal potential just above resting potential in our recording conditions. (D1–2) The locations of the neurons recorded in C in confocal images of the dye filled neurons after recordings. (E1) Patching of 17 bilateral pairs of FF neurons from either 4 day (relaxed) or 6 day (nacre) old fish revealed that most were connected to each other in one or both directions. Numbers in the histograms indicate the number of pairs with connections in neither direction (none), one direction (1-way), or reciprocal connections (2-way). (E2) Box plot of the strengths of the connections from left to right and right to left (only measured in cases where there was a connection in a given direction). The connection strengths were not significantly different in the two directions.

Specificity of laser targeting of inhibitory neurons.

(A) FF neurons were backfilled with Texas red in a transgenic line with neurons labeled with nuclear targeted GCaMP6s. (B1) Image prior to laser targeting showing the cell to be targeted marked by the white arrow head. (B2) GCaMP6s fluorescence from the targeted cell following laser illumination for ablation shows a massive increase in fluorescence, also visible in images of the targeted neuron in horizontal (C1) and cross section (C2) images taken after illumination. Note the bright green signal is evident only in the targeted cell and not in surrounding neurons next to and above it, showing the specificity of the targeting.

Examples of Laser ablations.

(A) Diagram showing the approach to ablations, with initial filling of the Mauthner cell with red dye by single cell electroporation in a line with GFP labeled glycinergic neurons, followed by imaging and ablations from above with a femtosecond laser. (B1–B3) Control fish exposed to laser power below threshold for lesions shown pre irradiation (B1), immediately post irradiation (B2) and between 1 and 2 days later (B3). Glycinergic neurons are green, with arrows pointing to the cluster of feedforward cells on either side of the red Mauthner cell. (C1–C3) As in B, but with laser ablation directed at the visible feedforward neurons (arrows in C1), most of which were removed (C3), while preserving the adjacent red dendrite of the M-cell. (D1–D3) Laser targeting of the lateral dendrite of the Mauthner cell (arrow) sometimes severs the lateral dendrite, while leaving the adjacent inhibitory neurons and the soma of the M-cell intact (D3). (E1–E3) Laser targeting of the lateral dendrite that led to death of the M-cell (soma marked by arrowhead in E1 is absent in E3), but preservation of adjacent inhibitory neurons.

Tail directed water pulse leads to activation of axons in the eighth cranial nerve.

(A) Diagram of the experiment in which extracellular recordings from the eighth cranial nerve in a paralyzed larvae were performed while applying a pulse of water directed at the tail. (B) Image of the glass recording electrode on the eighth nerve. (C) The water pulse led to extracellularly recorded activity in fibers of the eighth nerve, showing that a water pulse like that we used on freely swimming fish can produce an auditory response, even though tail directed.

Escape latency of control and lesioned fish in response to a unilaterally directed squirt of water.

(A) Example trial showing the escape elicited by a squirt of colored water from a needle on the left side of the frame. The bend begins in the second frame (*) and the fish performs a rapid escape bend away from the stimulus in subsequent frames, with the peak initial bend at 10 ms (**). (B.) All of the data for the latency of escape response from lesioned (red) and unlesioned (blue) sides in controls and in the different ablation conditions. Box and whisker plots represent median as well as first and third quartiles. Outlier points are solid. Asterisks mark significant differences between responses on intact and lesioned sides (*p<0.05, ***p<0.001, corrected for multiple comparisons). Sham ablations and ablation of medial glycinergic cells near the M-cell, but not implicated in escapes, did not affect the latency. Killing the Mauthner cell led to a significant increase in latency. Cutting dendrite of the M-cell also led to a trend toward a longer latency, although it was not significantly different from the intact side. Ablation of the lateral FF glycinergic neurons led to a significant reduction in the latency to respond.

Response laterality of control and lesioned fish in response to an omnidirectional stimulus.

(A) Diagram of the experiment in which a dish containing a freely swimming fish is vibrated and the left/right direction of the initial escape turn is monitored. (B) An example of the high speed video recordings of a trial in which the escape response occurred to the fish’s right side. Asterisks mark the beginning (*) and end (**) of the initial bend. (C) All of the data for response laterality for the control and lesion conditions, showing the percentage of responses initiated by the ablated side, which led to a turn away from that side and toward the intact side. Every lesion condition, except the ablation of medial glycinergic cells not implicated in escape, was significantly different from sham controls, with the magnitude of the change differing depending upon the type of lesion (*p<0.05, ***p<0.001, corrected for multiple comparisons).

Summary of the circuit and its relationship to hindbrain structure.

(A) Diagram of the circuit motif for the left right choice revealed and tested behaviorally in this paper. Relative synaptic strengths are indicated by contact size. Dotted line shows ipsilateral output of the FF. M: Mauthner cell, FF(red): Feedfoward inhibitory neuron. (B) Image of the disposition of the neurons in the hindbrain. The inhibitory neurons (red, FF) lie at the bottom the most lateral of three columns of glycinergic inhibitory neurons (green) marked by arrows. This lateral column contains commissural inhibitory neurons with outputs on both sides of the hindbrain, while more medial columns contain neurons with contralateral (middle column) or ipsilateral outputs (medial column). Scale bar=20 µm. (C) The column containing the FF neurons is one of a series of interleaved glutamatergic and glycinergic columns in the hindbrain. The variety of sensory inputs to the lateral glycinergic column and the shared morphology of the neurons in it raise the possibility that other neurons in the column might also contribute to the laterality of other behavioral responses to sensory inputs by implementing the same circuit motif.