Tectal microcircuit generating visual selection commands on gaze-controlling neurons

Abstract

The optic tectum (called superior colliculus in mammals) is critical for eye-head gaze shifts as we navigate in the terrain and need to adapt our movements to the visual scene. The neuronal mechanisms underlying the tectal contribution to stimulus selection and gaze reorientation remains, however, unclear at the microcircuit level. To analyze this complex--yet phylogenetically conserved--sensorimotor system, we developed a novel in vitro preparation in the lamprey that maintains the eye and midbrain intact and allows for whole-cell recordings from prelabeled tectal gaze-controlling cells in the deep layer, while visual stimuli are delivered. We found that receptive field activation of these cells provide monosynaptic retinal excitation followed by local GABAergic inhibition (feedforward). The entire remaining retina, on the other hand, elicits only inhibition (surround inhibition). If two stimuli are delivered simultaneously, one inside and one outside the receptive field, the former excitatory response is suppressed. When local inhibition is pharmacologically blocked, the suppression induced by competing stimuli is canceled. We suggest that this rivalry between visual areas across the tectal map is triggered through long-range inhibitory tectal connections. Selection commands conveyed via gaze-controlling neurons in the optic tectum are, thus, formed through synaptic integration of local retinotopic excitation and global tectal inhibition. We anticipate that this mechanism not only exists in lamprey but is also conserved throughout vertebrate evolution.

Keywords: GABAergic inhibition; evolution; gaze control; optic tectum; superior colliculus.

Fig. 1.

Tectoreticular pathway formed by gaze-controlling neurons in lamprey. (A, Left) Schematic illustrating a unilateral dextran tracer injection into the MRRN used for retrograde labeling of tectal output cells. Brainstem projecting tectal output cells were classified into two categories, into ipsilateral brainstem projecting (iBP) and contralateral brainstem projecting (coBP), by comparing cell soma location with the side of the injection site. (Right) Schematic of optic tectum divided into a two-by-two grid spanning rostrocaudally and mediolaterally. Color coding shows cell count percentage averaged from iBP and coBP cells in 13 animals. iBP cells are homogenously distributed all over tectum, whereas approximately two-thirds of the coBP cells were located rostrally. Overall, iBP cells outnumbered coBP by roughly 10 times with a total cell count of 1,555 and 145, respectively. (B, Left) Schematic illustrating the dorsal view of the lamprey brain preparation used to test the effect of electrical stimulation of tectal output cells on reticulospinal (RS) cells in the medial rhombencephalic reticular nucleus (MRRN) of the lower brainstem. Abbreviations: Rec, intracellular recording electrode; Stim, extracellular stimulation electrode. Shaded region illustrates the optic tectum. (Right) Simultaneous intracellular recordings of RS cells in response to unilateral electrical stimulation of the optic tectum. Monosynaptic EPSPs are recorded in both ipsilateral (Top) and contralateral RS cells (Bottom) in MRRN. Averaged traces (in black) are obtained from 20 trials. (C) Representative examples of three reconstructed iBP (Top) and three coBP (Bottom) tectal output cells intracellularly filled with Neurobiotin. Arrows indicate example cells (an iBP and a coBP cell) that are shown as microscopic images in Fig. S1. (D) Two-dimensional contour maps of dendritic fields of tectal output cells. Color coding refers to line-density percentage (100% maximum line-density). Data based on six iBP and six coBP cells. Abbreviations: DL, deep layer; IntL, intermediate layer; SL, superficial layer. (E) Phase-plane plots illustrating the different components of action potentials generated by tectal output cells. CoBP (n = 9) have an 8 mV lower threshold than iBP cells (n = 11; P < 0.001; Table S1). The rate of change in membrane potential (slope; y axis) is plotted against the membrane potential (x axis) during the discharge of one action potential in response to intracellular ramp-current injection. Here, we illustrate the phase plots of 6 iBP and 6 coBP cells out of the total of 20 recorded cells. Each portrayed cycle is obtained from an individual neuron. Horizontal dotted line shows 10-mV⋅ms⁻¹ level used to locate intersection points for spike-threshold determination. (Inset) Voltage trace of a coBP cell (V_m) in response to a 500-ms ramp current-injection (I_inj).

Fig. 2.

The visuomotor pathway: excitation and inhibition integrated by tectal gaze-controlling cells. (A) Transverse section of optic tectum showing retinal afferents (red) and cell somatas of iBP cells (green) in the superficial and deep layers, respectively. (B) Schematic illustrating the relative positions of the stimulation electrode (stim) in the superficial layer and the recording electrode (rec) in the deep layer for whole-cell recordings from identified tectal output neurons. (C) Subthreshold (blue) and suprathreshold (black) response of an iBP neuron after electrical microstimulation of the superficial layer at 20 and 30 μA with short latencies (4.1 ± 1.9 ms; n = 17, 8 coBPs, 9 iBPs), respectively. (D) Activation of a coBP cell after stimulation of the superficial layer in control solution (gray; 10 superimposed traces) and after bath application of 10 μM gabazine (GABA_A antagonist; red; 10 traces; n = 13). (E) The excitatory signal transmission between the superficial and the deep layer is glutamatergic. It consists of a NMDA-mediated component (blocked with 50 μM APV; NMDA receptor antagonist) and an AMPA-mediated component (blocked with 10 μM NBQX; AMPA receptor antagonist). (F) EPSPs evoked in an iBP cell, held at −65 mV, to repetitive stimulation of the superficial layer at 10 Hz in the presence of 10 μM gabazine and 3 mM QX-314 in the recording pipette. (Bottom) Average of EPSP amplitudes for each pulse normalized to the first EPSP (n = 6, 2 coBPs, 4 iBPs). (G) IPSPs evoked in a coBP cell, held at −20 mV, to 10-Hz stimulation of the SL with excitation blocked by 2 mM kynurenic acid (nonselective glutamate block; n = 5; 2 coBPs, 3 iBPs). (Bottom) Similarly to F, average of amplitudes normalized to first IPSP for each pulse. Means ± SEMs are shown. (H) To examine whether this excitation from the superficial layer is monosynaptic, we recorded postsynaptic responses using high divalent ion concentrations in the bath solution (4 mM Ca²⁺ and 8 mM Mg²⁺) to reduce the probability of neurotransmitter release in postsynaptic cells (n = 4, 2 iBPs, 2 coBPs). (Bottom) EPSPs were recorded from an iBP cell, held at −65 mV, in response to repetitive stimulation of the SL at 10 Hz in control solution (black) and during high divalent ion concentration in the perfused aCSF (red). (Top) This increase in EPSP amplitude was accompanied by a decrease in IPSP amplitude when holding at −20 mV and using QX-314 in the recording pipette. Means ± SEMs are shown.

Fig. 3.

Gaze-controlling cells driven directly by retinal afferents and tectal GABAergic input. (A, Left) A triple-labeled confocal image of the recorded iBP cell in Fig. 2H. iBP cells in the deep layer retrogradely labeled with dextran (red), an iBP cell intracellularly stained with Alexa Fluor 488 hydrazide (green; the recorded cell appears yellow due to the strong prelabeling near the soma), and anterogradely labeled retinal afferent fibers (blue) in the superficial layer following Neurobiotin injection in the optic nerve. (Scale bar: 20 μm.) (Right) Confocal optical section (1-μm thickness; i–iii corresponding to boxed areas in Middle) showing the dendrites of the recorded iBP cell in the SL (green; intracellular label) along with the close apposition of retinal afferent varicosities (blue), suggesting points of putative synaptic contacts. (Scale bar: 2 μm.) (B, Top) Transverse section of optic tectum illustrating labeled retinal afferents in the superficial layers, after a bidirectional tracer injection (Neurobiotin) into the optic nerve (in green). GABA-immunoreactive neurons (in red) are mainly located within the stratum opticum of the superficial layer—the layer of retinotectal fiber entry. (Scale bar: 150 μm.) (Bottom) Enlarged image of region within the box showing retinal afferent fibers in close apposition to GABAergic cell bodies. White arrows indicate location of GABA-positive superficial layer cells. (Scale bar: 40 μm.)

Fig. 4.

Retinal excitation and intratectal inhibition. (A) Schematic of a whole-mount brain preparation that enables electrical microstimulation of the optic while performing whole-cell recordings from tectal output neurons or from cells in the superficial layer of the optic tectum. View of transverse section of the midbrain through the optic tectum, with the patch pipette targeting tectal output cells (Methods). The stimulation electrode (stim) is positioned on the optic nerve directly at the exit from the center of the retina. (B) Stereotypical synaptic pattern of a tectal output neuron (an iBP is shown here) in response to a single pulse delivered into the optic nerve as illustrated in A. This EPSP–IPSP sequence is revealed when holding the recorded tectal output cells (n = 26) at two holding potentials: at −65 mV (bottom trace) and at −20 mV (top trace), while using QX-314 in the intracellular solution to block action potentials. See Fig. S2B for quantification. (C) Voltage traces as those shown in B were deconvolved into an excitatory (G_e, red) and an inhibitory (G_i, blue) conductance throughout the duration of the synaptic response. Three membrane potential measurements were used to estimate the relative conductances (holding at −65, −45, and −20 mV) and by assuming that the reversal for glutamate-mediated synaptic transmission was at 0 mV and the reversal for GABA_A-mediated inhibition at −65 mV (experimentally determined; Fig. S5). A balance between excitation and inhibition was always achieved. However, inhibition was estimated to be always stronger than excitatory with a time lag to excitation of ∼8 ms on average (amplitudes and latencies for both G_e and G_i are shown in Fig. S2 C and D, respectively). Black lines are averaged from 20 trials. (D) Schematic of proposed interlaminar tectal circuitry. Excitation (red); Inhibition (blue). GPi, globus pallidus interna; IntL, intermediate layer; RS, reticulospinal cell; SNr, substantia nigra pars reticulata. For simplicity, the basal ganglia output nuclei—GPi and SNr—are shown together.

Fig. 5.

Visually evoked synaptic inhibition with local excitation in tectal gaze-controlling cells. (A) Schematic of the isolated in vitro eye–brain preparation. View of transverse section of the midbrain through the optic tectum, with the patch pipette targeting tectal output cells. Light stimuli (L1, L2, and L3) from LED sources targeting the retina at different angles. (B) Example of visually evoked synaptic responses of a iBP tectal output cell during local (L1; magenta) and global (L1–L2–L3; green) light stimulation (20 traces superimposed). (C) Membrane potential recordings of an iBP cell held at −20 mV (Top) and at −65 mV (Bottom; at reversal potential for GABA_A; see SI Methods) using QX-314 to block action potentials during local (Left) and global visual input (Right). (D) Excitatory (in red) and inhibitory (in blue) synaptic conductances during local (Top) and global (Bottom) light stimulation. Arrows indicate the delayed onset of inhibition relative to excitation (n = 12, for both). (E) Peak synaptic conductances for excitation and inhibition (for local, mean difference: 0.18 nS, ***P < 0.001, n = 12; for global, mean difference: 0.10 nS, **P < 0.01, n = 12; paired t test). (Bottom) Fractional peak conductance for inhibition. Balance between inhibition and excitation remains similar across conditions [means: 0.73 for local; 0.76 for global; no statistical significance (ns)].

Fig. 6.

Interlaminar inhibition generates visual stimulus suppression. (A) Global visual stimulation with and without tectal inhibition in an iBP tectal output cell. (Top) Visually evoked synaptic inhibition at −42 mV during control using normal artificial intracellular solution. (Middle) Global stimulation does not evoke significant excitation at −68 mV. (Bottom) When inhibition is blocked by bath application of 10 μM gabazine (directed over the dendritic arbor), the same global stimulus will induce a burst of action potentials (20 traces superimposed; n = 3). (B) Excitatory synaptic responses during local (Top) and global (Bottom) visual presentation when using QX-314. The recorded iBP output cell is responsive to local input (magenta), but nonresponsive during global visual input (green) during control conditions. However, after application of gabazine (red), EPSPs are drastically enhanced. EPSPs have similar magnitudes during both local and global stimuli (n = 4) but also during stimulus onset and offset. (C) An example of an iBP cell displaying the same effect after pressure injection of gabazine into the visualized dendritic field of the cell during local and global light stimulation (Methods). Spikes are truncated. Control traces display identical patterns of activity as those shown in Figs. 5 B and C, and 6 B and D, but are not plotted for clarity. (D) Recordings at holding potentials of −69 and −20 mV show that the recruited inhibition is reduced while the monosynaptic excitation is enhanced with both local and global light stimulation during perfusion of high divalent ion concentrations of calcium and magnesium (see Fig. 2 for microstimulation of retinal afferents and Fig. 3A). Solid lines are averaged traces.

Fig. 7.

Local excitation and global inhibition. (A) Schematic of the retina sectioned into four quadrants (A, anterior; D, dorsal; P, posterior; V, ventral) for electrical stimulation during whole-cell patch recordings of tectal output cells. (B, Top) Color coding indicating specific retinal area that is being stimulated. (Middle) Synaptic responses in an iBP neuron held at −65 mV (red traces; Top) and at −20 mV (blue traces; Bottom). This output cell receives intense excitation only from dorsal retina, but inhibition from all quadrants. (C) Mean amplitude of the evoked postsynaptic potential (PSP) plotted against the stimulated retinal quadrant. To avoid dependencies on the particular retinal area that gives rise to the excitatory component, quadrants are presented in descending order for simplicity without loss of generality. Amplitudes are normalized by the maximal response obtained for each recorded cell. Excitatory input is constrained to practically one quadrant. By contrast, inhibition is less tuned and maintained at ∼50% across the retina (n = 9). EPSP_max = 20.6 ± 8.8 mV; IPSP_max = 13.5 ± 6.9 mV. Means ± SEMs are shown.

Fig. 8.

Long-range monosynaptic inhibition onto tectal gaze-controlling neurons. (A) Schematic of the dorsal view of the lamprey brain with the injection site of the retrograde tracer used to label horizontal intratectal projections. (B) Long-range horizontal projecting tectal cells that have been retrogradely labeled after tracer injection (green) on the ipsilateral side of tectum as shown in A. Retinal afferents in the superficial layer also have been anterogradely labeled (green) by the tracer. GABA-immunoreactive neurons (in red) are located within the stratum opticum sublayer of the superficial layer. (Right panels) Enlarged region of areas A and B (dotted squares in Left) showing double-labeled long-range horizontal projecting cells that are GABAergic (Right; in box B) and non-GABAergic (Left; in box A). Top row shows GABA-positive labeling, and bottom row shows cells that have been retrogradely labeled that are indicative of long-range horizontally projecting neurons. (C, Top) Schematic illustrating the experimental design used for recording from tectal output neurons in the deep layer during stimulation of caudal parts of the superficial layer that were outside the receptive fields from the recorded cells. (Bottom) Inhibitory synaptic response measured at −20 mV from an iBP neuron after repetitive electrical microstimulation (at 10 Hz) of the caudal superficial layer (no excitation was elicited because no PSP was observed at −65 mV) in control solution (gray; 20 superimposed traces; n = 6) and during bath application of high divalent ion concentration in aCSF (red; 20 superimposed traces). The remaining IPSP (in red) is caused through ipsilateral and intratectal long-range monosynaptic inhibitory connections likely arising from GABAergic cells like those shown in B. Solid lines are averaged traces. (D) Summarizing circuit. Tectal output cell (red) and local inhibitory neurons (blue) that receive excitatory synaptic input from the various retinal areas. Dotted black line is the midline.