Cell-type specific binocular interactions in mouse visual thalamus

Summary

To determine whether intrinsic circuits of the dorsal lateral geniculate nucleus (dLGN) mediate excitatory and/or inhibitory binocular interactions, in vitro dual color optogenetics was used to examine convergence of eye-specific retinal inputs to thalamocortical (relay) cells and GABAergic interneurons. Both relay cells and interneurons were found to receive direct binocular retinogeniculate input, but the incidence was higher for interneurons. Moreover, the distribution of binocular relay cells was limited to regions in and around the ipsilateral patch of retinogeniculate terminals, while binocular interneurons were distributed throughout the dLGN. Accordingly, retinogeniculate-evoked inhibitory interactions were common. Retinal inputs arising from either eye inhibited the majority of relay cells via the activation of intrinsic interneuron circuits. In addition, the majority of interneurons were found to be interconnected, receiving both monocular and binocular retinogeniculate-evoked inhibition. These results reveal that intrinsic interneurons provide the primary source of binocular interactions within the dLGN.

Graphical abstract

Highlights

• •dLGN interneurons receive more binocular retinal input than relay cells
• •Most dLGN relay cells receive binocular retinogeniculate-evoked inhibition
• •Binocular relay cells are more spatially restricted than binocular interneurons
• •dLGN interneurons are interconnected

Molecular biology; Neuroscience

Introduction

Projections from each eye are segregated in separate domains within the dorsal lateral geniculate nucleus (dLGN).^1,2,3,4,5,6 Yet, in vivo studies indicate that the activity of single dLGN neurons can be influenced by visual stimuli presented to either eye.^{7,8,9,10,11,12,13} In this study, we explored whether intrinsic circuits mediate binocular interactions in the mouse dLGN. We employed dual color optogenetics in vitro to selectively activate input from each eye and recorded synaptic responses in thalamocortical (relay) cells as well as inhibitory interneurons, which have extensive dendritic arbors that are not confined to eye specific domains.^{14,15,16,17,18} While most relay cells received monocular retinal input, most interneurons received binocular retinal input, and most relay cells received binocular retinogeniculate-evoked inhibition. In recordings from closely spaced pairs of relay cells and interneurons, the most common relationship observed was retinogeniculate excitation of interneurons paired with inhibition of relay cells via activation of inputs from the same eye(s). Moreover, relay cells that received binocular inhibition were located in and around the ipsilateral patch of retinogeniculate terminals, suggesting that interneuron output is spatially related to retinal input. Finally, we found that dLGN interneurons are interconnected, displaying both monocular and binocular inhibition in response to retinal activation. In sum, our results indicate that geniculate interneurons provide one of the first locations where signals from the two eyes can be compared, integrated, and adjusted before being transmitted to cortex, shedding new light on the role of the thalamus in binocular vision.

Results and discussion

Dual optogenetic activation of ipsilateral and contralateral retinogeniculate terminals

To examine the degree to which geniculocortical (relay) cells and interneurons receive input from ipsilateral or contralateral retinogeniculate terminals, we carried out in vitro experiments using a dual optogenetic activation approach¹⁹ in dLGN slices obtained from GAD67-GFP mice. We injected a virus in one eye to induce the expression of the red-shifted opsin Chrimson (ChRred) and a virus in the other eye to induce the expression of the blue-light activated opsins Chronos or Channelrhodopsin (ChRblue; Figure 1A). Two weeks later, slices of the dLGN were prepared (Figures 1B and 1C) and whole-cell recordings were targeted to interneurons (which contain GFP in the GAD67-GFP line; Figure 1D, i) or relay cells (which do not contain GFP; Figure 1D, r). The location of the patch pipette was photographed at the conclusion of each recording, and biocytin was included in the internal solution so that the neuron identity (interneuron, Figure 1E, or relay cells; Figure 1F) could be confirmed and the location of all recorded neurons (n = 123 relay cells, n = 110 interneurons) could be plotted on a template of the dLGN. Recorded neurons were distributed across the dLGN but were concentrated in the binocular zones (dorsomedial half, indicated by the dotted line in Figures 1G and 1H).

Figure 1.

Experimental protocol

(A) Schematic showing the experimental design and methods. In GAD67-GFP mice, a virus (pAAV9-Syn-Chrimson-tdT) was injected in one eye to induce the expression of the red-shifted opsin Chrimson (ChRred), a second virus (pAAV9-Syn-Chronos-GFP or AAV9-hSyn-hCHR2(H134R)-EYFP) was injected in the other eye to induce the expression of the blue-light activated opsins Chronos or Channelrhodopsin (ChRblue), and in vitro whole-cell recordings were obtained from acutely prepared coronal thalamic slices. Retinal terminals were photo-stimulated (using 460 nm, and/or 630 nm light pulses) and light evoked synaptic responses were obtained from relay neurons and interneurons using pipettes containing K⁺ or Cs⁺⁺ based internal solutions.

(B and C) Examples of 300 μm thick coronal slices showing the viral labeling pattern of retinal terminals in the dorsal lateral geniculate nucleus (dLGN), ventral lateral geniculate nucleus (vLGN), and optic tract (OT) contralateral (B) and ipsilateral (C) to virus injections to express ChRblue (green), and contralateral (C) and ipsilateral (B) to virus injections to express ChRed (magenta). Interneurons labeled with GFP in the GAD67-GFP line are also shown in green.

(D) Confocal image showing GFP-expressing interneurons along with a biocytin filled interneuron and relay cell. Recordings were targeted to interneurons (which contain GFP in the GAD67-GFP line; i) or relay cells (which do not contain GFP; r) and biocytin was included in the internal solution to fill the dendritic arbors (magenta).

(E and F) Confocal images show a biocytin filled interneuron (E) and 3 biocytin filled relay cells (F).

(G and H) Template of the dLGN showing the approximate location of all recorded relay neurons (black dots) and interneurons (red dots). Each dot represents a single neuron. (G) Locations of recorded relay cells (n = 49) and interneurons (n = 52) recorded using K⁺ based internal solution and the GABA_A antagonist SR95531 in the bath. (H) Locations of recorded relay cells (n = 74) and interneurons (n = 58) recorded using Cs⁺⁺ based internal solutions and no GABA antagonist in the bath.

(I) Postsynaptic currents recorded in a relay cell that responded only to photoactivation of retinogeniculate terminals expressing ChRred. Traces below the response illustrate the timing, duration, and temporal frequency of blue and red light photoactivation. The ChRred was activated by 20 Hz red light pulses (I₁), 20 Hz blue light pulses (I₂) and to the initiation of a 1.5 s red light pulse which subsequently occluded responses to paired 20 Hz blue light pulses (I₃).

(J) Postsynaptic currents recorded in a relay cell that responded only to photoactivation retinogeniculate terminals expressing ChRblue. The ChRblue was not activated by 20Hz red light pulses (J₁) but was activated by 20 Hz blue light pulses (J₂) even when paired with a 1.5 s red light pulse (J₃). Scale bars: B, 100 μm, also applies to C; D, 50 μm; E, 50 μm and also applies to F.

To activate ipsilateral or contralateral ChRred-expressing retinogeniculate terminals, 20 Hz red light pulses (10 ms) were directed through the microscope objective (Figure 1I₁). Since ChRred-expressing terminals are also activated by blue light¹⁹ (Figure 1I₂), to activate ipsilateral or contralateral ChRblue-expressing retinogeniculate terminals, a 1.5 s red light pulse was paired with 20 Hz blue light pulses (1 ms duration, 10 pulses). The long red-light pulse served to occlude any responses evoked from activation of the ChRred-expressing terminals; after an initial release of neurotransmitter from the ChRred-expressing terminals, the continuous light stimulation resulted in no further neurotransmitter release²⁰ (Figure 1I₃). However, the red light had no effect on the ChRblue-expressing retinogeniculate terminals so that blue light pulses paired with red light could be used to exclusively activate the ChRblue-expressing terminals (Figure 1J₃). This paired light protocol was chosen to avoid spectral crosstalk that can occur with sequential light activation protocols²¹ and allowed us to activate retinogeniculate terminals expressing ChRred or ChRblue with short high intensity trains of light pulses. The light trains ensured the maximum activation of retinal terminals to examine excitatory responses as well as the subsequent activation of intrinsic circuits to examine retinogeniculate evoked inhibitory response. Multiple responses to each pulse in the train also provided confidence that a cell was indeed responsive.

Excitatory retinogeniculate responses in relay cells and interneurons

Using our dual optogenetic activation technique, we first compared the direct innervation of relay cells and interneurons by ipsilateral and/or contralateral retinogeniculate terminals. For these experiments, the GABA receptor antagonist SR95531 (20 μM) was included in the bath to block interneuron synaptic activity and neurons were held at −60 mV in voltage (Figures 2A and 2B) and current clamp (Figures 2C and 2D) mode using pipettes filled with a potassium-based internal solution. Under these conditions, monocular (Figure 2E) and binocular (Figure 2F) excitatory responses were recorded in both relay cells and interneurons. As previously described,²¹ input from one eye dominates the excitatory retinogeniculate responses of most relay cells, with 29/49 (59%) receiving monocular input. The remaining 20 (41%), received excitatory input from both eyes (Figure 2G), with dominant responses evoked by activation of retinogeniculate terminals expressing ChRBlue (10 cells) or ChRed (10 cells).

Figure 2.

Cell-type specific excitatory responses to photoactivation of ipsilateral and contralateral retinogeniculate terminals

(A) Excitatory postsynaptic currents (EPSCs) recorded in a relay cell that responded to photoactivation of input from both eyes.

(B) EPSCs recorded in an interneuron that responded to photoactivation of input from both eyes.

(C) EPSPs recorded in a binocular relay cell; action potentials were initiated in response to input from one eye only.

(D) EPSPs recorded in a binocular interneuron; action potentials were initiated in response to input from both eyes.

(E) Locations of recorded relay cells (black dots, n = 29) and interneurons (red dots, n = 19) that responded to activation of retinogeniculate terminals originating from one eye only. Monocular relay cells and interneurons are dispersed across the dLGN.

(F) Locations of recorded relay cells (n = 20) and interneurons (n = 33) that responded to activation of retinogeniculate terminals originating from both eyes. Binocular relay cells are located in and around the ipsilateral patch (small oval) while binocular interneurons were more widely dispersed.

(G) Histogram illustrates the percentages of relay cells (black bars, n = 46) and interneurons (red bars, n = 52) with monocular or binocular responses. The incidence of binocular responses was higher among interneurons, with 33/52 (63%) responding to input from both eyes (22% higher than relay cells, 95% CI: 4%, 42%).

(H) The probability of retinogeniculate inputs activating action potentials was greater in interneurons (30/33) 91% of interneurons versus (12/20) 60% of relay cells (31% higher for interneurons, 95% CI: 7.3%, 55%). Spikes were initiated in relay cells only in response to dominant eye input, while (8/33) 24% of interneurons fired in response to activation of input from the dominant or non-dominant eyes (24% higher than relay cells, 95% CI: 9.6%, 39%.

(I) Plot of the contralateral versus ipsilateral EPSC amplitudes recorded in binocular relay cells (black dots or squares, n = 20) and interneurons (red dots or squares, n = 33). Squares indicate cells in which the light was reduced to eliminate spikes in the dominant responses. Maximum EPSC amplitudes in relay cells were all larger in response to activation of input from the contralateral eye (R_c), relative to the ipsilateral eye (R_i, paired t test, t = 4.047, df = 19, p < 0.0007).

(J) The ocular dominance indexes (ODI, calculated as (R_c – R_i)/(R_c + R_i))) of relay cells (black dots or squares, n = 20) and interneurons (red dots or squares, n = 33) are significantly different (unpaired two-tailed t test, t = 2.42, df = 51, p = 0.0194). Mean and 95% confidence interval indicated.

Although ipsilateral and contralateral retinogeniculate terminals originate from different subsets of ganglion cells,²² their synaptic terminals do not differ in size or in the proportion of contacts on relay cells and interneurons.²³ However, since interneurons make up only 6% of neurons in the mouse dLGN,¹⁷ each interneuron is likely innervated by at least twice the number of retinal terminals that innervate each relay cell. This is supported by both anatomical and in vitro studies which have revealed a high level of retinal convergence on interneurons.^{18,24,25,26,27,28} Accordingly, we found that the incidence of binocular responses was higher among interneurons, with 33/52 (63%) responding to input from both eyes (Figure 2G; 22% higher than relay cells, 95% confidence interval [CI]: 4%, 42%).

Retinogeniculate terminal photoactivation was also more likely to initiate action potentials in interneurons (Figure 2D) than in relay cells (Figure 2C). During current clamp recordings, photostimulation of retinogeniculate terminals initiated action potentials in (30/33) 91% of interneurons versus (12/20) 60% of relay cells (31% higher for interneurons, 95% CI: 7.3%, 55%). Moreover, spikes were initiated in relay cells only in response to dominant eye input, while (8/33) 24% of interneurons fired in response to activation of input from the dominant or non-dominant eyes (Figure 2H, 24% higher than relay cells, 95% CI: 9.6%, 39%). A likely explanation for this difference in the incidence of firing is that the average input resistance of interneurons (645 ± 166 mΩ, n = 32) is significantly greater than that of relay cells (289 ± 111 mΩ, n = 36; p ≤ 0.0001, t = 10.53, df = 66, unpaired two tailed t test, see also Jager et al.²⁹). However, there was no significant difference in the input resistance of interneurons that received monocular input (608 ± 234 mΩ, n = 9) or binocular input (659 ± 140 mΩ, n = 23, p = 0.4363, t = 0.7890, df = 30, unpaired two-tailed t test) or relay cells that received monocular input (294 ± 130 mΩ, n = 18) or binocular input (284 ± 91 mΩ, n = 18, p = 0.7866, t = 0.2729, df = 34, unpaired two-tailed t test, Figure S1).

For binocular neurons, we compared the amplitudes of excitatory postsynaptic currents (EPSCs) elicited by photoactivation of ipsilateral or contralateral retinogeniculate terminals (Figure 2I). For some of these cells (2 relay cells and 4 interneurons, designated by squares in Figures 2I, 2J, and S2), we were unable to effectively clamp action potentials evoked by the dominant eye. For these cells the light intensity was subsequently decreased to values just below the levels that evoked spiking in the postsynaptic neurons, potentially leading to an underestimation of the dominant eye EPSC amplitudes. Nevertheless, maximum EPSC amplitudes in relay cells were all larger in response to activation of input from the contralateral eye (R_c), relative to the ipsilateral eye (R_i, Figure 2I, paired t test, t = 4.047, df = 19, p < 0.0007), and the ocular dominance index (ODI, calculated as (R_c – R_i)/(R_c + R_i)) of each relay neuron was a positive value (Figure 2J). Similar to relay cells, the binocular responses of interneurons were dominated by one eye (Figure S2). However, dominant responses could be elicited by activation of ipsilateral or contralateral retinogeniculate terminals (Figure 2I) and the ODI values of interneurons (mean 0.25) were significantly lower than those of relay cells (mean 0.61, Figure 2J, p = 0.0194, t = 2.42, df = 51, unpaired two-tailed t test).

Location of binocular relay cells and interneurons

While binocular interneurons were widely distributed across the dLGN, binocular relay cells were located in and around the ipsilateral patch of retinogeniculate terminals (Figure 2F). This may be related to the fact that relay cells primarily receive retinal input on their proximal dendrites, restricted to regions of approximately 100 μm,^3,25,26 while interneurons receive retinal input across their full dendritic arbors, spanning regions of up to 500 μm.^3,14,24 Thus, our study likely underestimates the differences in the retinal innervation of relay cell versus interneurons, since the dendritic fields of interneurons are more likely to be truncated in the 300 μm thick slices used for this study. In addition, our recordings from somata, as well as light that did not illuminate the entire dLGN, may also have also resulted in underestimates of the impact of retinal input (or interneuron input described in the following text) to the more distal dendrites of interneurons.

Retinogeniculate evoked inhibitory responses in relay cells and interneurons

We next examined the inhibitory responses in relay cells (n = 74) and interneurons (n = 58) evoked by photoactivation of ipsilateral or contralateral retinogeniculate terminals (Figure 3A). For these experiments, we again made use of the dual opsin approach but did not add GABA receptor antagonists to the bath. Additionally, pipettes were filled with a cesium-based internal solution and neurons were recorded in voltage clamp mode at holding potentials of −60 mV and 0 mV. This approach allowed us to identify both excitatory and/or inhibitory currents evoked by photostimulation of retinal terminals arising from either eye.

Figure 3.

Cell-type specific inhibitory responses evoked by photoactivation of ipsilateral and contralateral retinogeniculate terminals

(A) Inhibitory postsynaptic currents (IPSCs) recorded in a relay cell in response to photoactivation of input from either eye.

(B) Histogram illustrating the percentage of relay cells (black bars, n = 74) and interneurons (red bars, n = 58) with monocular, binocular, or no inhibitory responses following activation of retinogeniculate terminals.

(C) Histogram depicting the direct (monosynaptic) excitatory retinogeniculate responses recorded in relay cells (n = 41) and interneurons (n = 9) that received binocular inhibition (i.e., multi-synaptic IPSCs following activation of retinogeniculate input from either eye).

(D) Locations of recorded relay cells (black dots, n = 32) and interneurons (red dots, n = 26) that responded with IPSCs to activation of retinogeniculate terminals originating from one eye only. Relay cells and interneurons with monocular IPSCs are dispersed across the dLGN.

(E) Locations of recorded relay cells (n = 41) and interneurons (n = 9) that responded with IPSCs to activation of retinogeniculate terminals originating from both eyes. Binocular relay and interneurons are primarily located in and around the ipsilateral patch.

In these experiments, nearly all relay cells (73 of 74, 99%) received some form of retinogeniculate-evoked inhibition. In fact, the majority (41 of 74, 55%) of relay cells exhibited inhibitory responses following activation of input originating from either eye (Figure 3B), potentially reflecting the direct binocular innervation of interneurons (Figure 2G). Of the 41 relay cells that received binocular inhibition, only 15 (37%) were directly innervated by both ipsilateral and contralateral retinogeniculate terminals (Figure 3C). Moreover, 6 relay cells that received monocular inhibition did not exhibit direct excitatory responses to photoactivation of retinogeniculate terminals in the slice. Thus, despite the fact that a minority of relay cells receive direct excitatory input from both eyes, interneuron connections allow for additional binocular interactions within the dLGN.

The majority of interneurons (35/58, 60%) also responded with inhibitory currents following activation of retinogeniculate input, indicating that interneurons are interconnected with each other (as supported by connectomic data²⁴). Most of these interneurons received monocular inhibition (26, 45%; Figure 3B), but 9 (15%) responded with inhibitory currents following photoactivation of input from either eye. However, unlike relay cells, all of the interneurons that received binocular inhibition were also directly innervated by both ipsilateral and contralateral retinogeniculate terminals (Figure 3C). Furthermore, while relay cells and interneurons that received monocular inhibition were located throughout the dLGN (Figure 3D), those that received binocular inhibition were primarily located in and around the ipsilateral patch of retinogeniculate terminals (Figure 3E).

Recordings from closely spaced relay cell and interneuron pairs

Most of the experiments described previously (using cesium-based electrodes and no GABA antagonists in the bath) were carried out by sequentially recording from closely spaced relay cell and interneuron pairs (somata 16.7 ± 10 μm apart). This enabled the direct comparison of the retinogeniculate responses of neurons surrounded by the same sets of ChRred- and ChRblue-expressing retinogeniculate arbors. Given the more extensive dendritic arbors of interneurons, the dendrites of adjacent relay cells and interneurons overlapped to some degree, but often occupied distinct regions of the dLGN (Figures 4A–4C).

Figure 4.

Excitatory and inhibitory responses to photoactivation of ipsilateral and contralateral retinogeniculate terminals recorded in relay cell and interneuron pairs

(A–C) Examples of closely spaced (somata 16.7 ± 10 μm apart) pairs (n = 39) of biocytin-filled relay cells (black dot on soma) and interneurons (red dot on soma). The inset in panel A illustrates the location of the 2 pairs (red) in relation to GFP-labeled interneurons in the dLGN (green). Each pair exhibits overlapping and nonoverlapping dendritic domains.

(D) Binocular IPSCs recorded in a relay cell.

(E) Binocular EPSCs recorded in an adjacent interneuron.

(F) Distribution of pairs (n = 20) that displayed matching monocular interneuron excitation and monocular relay cell inhibition.

(G) Distribution of pairs (n = 10) that displayed matching binocular interneuron excitation and binocular relay cell inhibition. Black outlines indicate pairs illustrated in (B) and in (C)–(E).

(H) Model of local interneuron inhibition. Interneuron (Int) dendritic terminals directly innervated by retinal terminals (R) release GABA to provide local inhibition of relay cell dendrites.

(I) Model of global interneuron inhibition. Action potentials generated at the somata of interneurons initiate GABA release from axon terminals and backpropagate to initiate global GABA release from all dendritic terminals.

(J) Model supported by the interneuron firing incidence (Figure 2H) and distribution of binocular pairs illustrated in (G). Action potentials generated at the soma backpropagate to enhance local GABA release from dendritic terminals directly innervated by ipsilateral (I) and contralateral (C) retinal terminals. If axons are present, they may also be activated (indicated by dotted line). Scale in A, 50 μm (inset 100 μm). Scale in B, 50 μm and also applies to C. Scale for cell plots (F) and (G), 50 μm.

We recorded a total of 39 pairs of relay cells and interneurons that were closely spaced and found that they rarely displayed identical responses. Only 7 pairs displayed identical patterns of excitation and inhibition. A more common relationship was complementary interneuron excitation and relay cell inhibition (Figures 4D and 4E). In 20 pairs, the relay cell received monocular inhibition, and the interneuron was excited by input from the same eye; these pairs were distributed throughout the dLGN (Figure 4F). In 10 pairs, the relay cell received binocular inhibition, and the interneuron was excited by input from both eyes; these pairs were all distributed in and around the ipsilateral patch of terminals (Figure 4G).

Thalamic interneurons are unique in that they have two forms of output, both axonal and dendritic.^30,31 Previous studies suggest that when retinal terminals activate the dendritic terminals of interneurons, this can provide fast local inhibition of relay cells (schematically indicated in Figure 4H). In addition, when interneurons fire action potentials, they may provide a more global inhibition of relay cells via axon outputs as well multiple dendritic terminals that can be activated by backpropagation^32,33 (schematically indicated in Figure 4I). Since retinogeniculate photoactivation initiated action potentials in 91% of recorded interneurons, our results presumably reflect the global output of interneurons.

However, relay cells that received binocular inhibition were only located in and around the ipsilateral patch. This distribution indicates that even though binocular interneurons are located throughout the dLGN, their inhibitory output is spatially related to the distribution of retinogeniculate terminals. In other words, with our activation protocol, interneurons apparently inhibit relay cells via the release of GABA from the dendritic sites where they receive direct retinal innervation (schematically indicated in Figure 4J). In this case, the backpropagation of sodium spikes from interneuron somata may serve to amplify dendritic calcium responses,^28,32 but the direct retinal activation of local dendritic terminals is still necessary to elicit neurotransmitter release.

Finally, because monocular and binocular inhibitory currents were recorded in interneurons, relay cells may be disinhibited via binocular interactions. Such a scenario is difficult to detect using our dual optogenetic activation techniques but may partially account for changes in the amplitudes of inhibitory currents with repetitive optogenetic activation of retinal terminals (Figure 3, Figure 4A and 4D). In other words, strong initial inhibition of relay cells may subsequently be reduced when interneurons inhibit each other.

Functional implications

In mice, binocular vision is essential for depth perception and acuity,^34,35 and loss of binocular vision severely impacts fundamental behaviors such as prey capture.²² While cortical circuits are certainly required for creating a unified percept of visual signals originating from both eyes,^36,37 and the emergence of binocular depth perception,³⁴ our study indicates that subcortical circuits, particularly those involving dLGN interneurons, may also be important components of this process. In fact, recent studies demonstrate that cortical ocular dominance plasticity is absent in adult mice lacking thalamic synaptic inhibition.^38,39

When one or both eyes are stimulated in awake mice, geniculate neurons exhibit dominant/non-dominant eye interactions that include both enhancement and suppression.¹³ While these interactions could be mediated by extrinsic sources (e.g., cortex,³⁸ thalamic reticular nucleus,^40,41 superior colliculus,⁴² pretectum,^43,44 and parabigeminal nucleus^23,45), our study indicates that geniculate interneurons provide one of the first locations where signals from the two eyes can be compared, integrated, and adjusted before being transmitted to cortex. Whether these interneuron-mediated binocular interactions translate to species where retinogeniculate inputs from the two eyes are more highly segregated and arranged in a laminar fashion requires additional investigation. In carnivores and primates, geniculate neurons exhibit binocular suppression^7,8,9,10 and in both species there is evidence for interneurons with interlaminar connections.^46,47,48 However, binocular modulation is rare in the primate relative to the carnivore dLGN, at least in anesthetized animals.⁴⁹ Thus, future studies are needed to determine whether similar or different circuits underly binocular integration in nonrodent species. Nevertheless, our findings shed new light on the role of the thalamus in binocular vision, providing information which may be of critical importance for the understanding and treatment of binocular vision dysfunction.⁵⁰

Five critical dLGN circuit interactions were revealed by our experiments: (1) both relay cells and interneurons receive direct binocular excitatory input from the retina, but this feature is more prevalent among interneurons, and retinogeniculate terminal photoactivation is more likely to initiate action potentials in interneurons; (2) binocular relay cells are located in and around the ipsilateral patch whereas binocular interneurons are located throughout the dLGN; (3) both cell types receive binocular retinogeniculate evoked inhibition but it is far more prevalent among relay cells than interneurons; (4) in recordings from adjacent neurons, the most common relationship observed was excitation of interneurons paired with inhibition of adjacent relay cells, and (5) dLGN interneurons are interconnected, displaying both monocular and binocular inhibition in response to retinal activation. Collectively, these results indicate that although dLGN relay cells are considered monocular in terms of their direct responses to input from either eye, their responses can readily be modified by input from either eye via inhibition from interneurons and/or disinhibition brought about by interneuron-to-interneuron connections.

Limitations of the study

While in vitro slice recordings allow for the interrogation of thalamic circuits on a cellular level, there are some notable limitations to this approach. For example, thalamic slices cut in the coronal plane could potentially disrupt some aspects of retinogeniculate connectivity which in turn could lead to an underestimation of the number of functional convergent inputs. Moreover, because the expansive dendritic fields of interneurons are more likely to be truncated in a coronal slice preparation, our results may underestimate differences in the retinal innervation of relay cells versus interneurons. Another contributing factor that could lead to an underestimation of convergent retinal input is the complex dendritic architecture of interneurons. Because interneurons have long, small-diameter dendrites, distal synaptic events are prone to electrotonic isolation resulting in attenuation of signals recorded at the soma. Thus, some retina or interneuron inputs may remain undetected in somatic recordings. Finally, while these experiments reveal new patterns of retinogeniculate and interneuron connectivity, future studies are needed to evaluate how these circuits contribute to the generation of binocular receptive field properties at both the thalamic and cortical levels.

Resource availability

Lead contact

Microscopy data and any additional information required to reanalyze the data reported in this paper are available from the lead contact, Martha E. Bickford (martha.bickford@louisville.edu), upon request.

Materials availability

This study did not generate new unique materials.

Data and code availability

Data

The datasets used and/or analyzed during the current study have been deposited in Mendeley data and are available as of the date of publication. The access link is listed in the key resources table.

Code

The code used to analyze the data has been deposited in GitHub and is available as of the date of publication. The access link is listed in the key resources table.

Acknowledgments

The authors thank Arkadiusz Slusarczyk and Barbara O’Steen for their excellent technical assistance. This work was funded by the National Eye Institute (grants EY035523, EY031322, and EY12716).

Author contributions

S.P.M., G.G., W.G., and M.E.B. carried out the experiments. S.P.M., W.G., and M.E.B. analyzed the data. W.G. and M.E.B. wrote the manuscript. The final version of the manuscript was reviewed and approved by all authors.

Declaration of interests

The authors declare no competing interests.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Viruses
pAAV9-Syn-Chronos-GFP	Addgene	Addgene Item # 59171-AA9
or AAV9-hSyn-hCHR2(H134R)-EYFP	Addgene	Addgene Item # 26973-AAV9
pAAV9-Syn-Chrimson-tdT	Addgene	Addgene Item # 59171-AAV9
Chemicals
SR95531	Tocris Bioscience	Tocris Catalogue #1262
Biocytin	Sigma-Aldrich	Sigma Cat. #B4261
Strepavidin-633	Thermo Fisher Scientific	S-21375 RRID: AB_2313500
Software
Clampfit 11.1	Molecular Devices	pClamp (RRID:SCR_011323)
Photoshop	Adobe	Adobe Photoshop RRID:SCR_014199
Prism 10.4	GraphPad	GraphPad Prism (RRID:SCR_002798)
Experimental Models
GAD67-GFP mice	The Jackson Laboratory	Jax Stock # 007677, G42 line). RRID: IMSR_JAX:007677
Deposited Data
Raw and analyzed data	Deposited in Mendeley Data	https://data.mendeley.com/datasets/2jdyyyd8hx/1
Original Code
processABF 12-22-19	Available in GitHub	https://github.com/UofL-Bickford-lab/processABF

Experimental model and study participant details

All breeding and experimental procedures were approved by the University of Louisville Institutional Animal Care and Use Committee (protocol #24386). Experiments were carried out using mice, of either sex, postnatal day 14-42, in a line in which neurons that contain the 67KD isoform of glutamic acid decarboxylase (GAD) express green fluorescent protein (GFP; GAD67-GFP; Jax Stock No: 007677, G42 line). Full litters of mouse pups received eye injections and the sex of each was not recorded. Therefore, our study cannot address any potential sex differences in the circuitry of the dLGN.

Method details

Virus injections

To induce the expression of the opsins and fluorescent proteins in retinogeniculate axons and terminals, GAD67-GFP pups (p14-18) received intravitreal virus injections in each eye. Each pup was anesthetized with isoflurane via a small nose cone, the sclera was pierced with a sharp tipped glass pipette, and excess vitreous was drained. Additional pipettes, filled with AAV solutions and attached to a picospritzer, were inserted into the holes made by the first pipette and volumes of approximately 1μl of AAV solution were injected into each eye. The nose cone used to administer isoflurane was then removed and once alert the pup was returned to the cage containing the dam and littermates. Intravitreal injections of pAAV9-Syn-Chrimson-tdT (Addgene) were paced in one eye to induce the expression of the red-shifted opsin Chrimson and the red fluorescent protein TdTomato in retinogeniculate terminals. In the other eye, intravitreal injections of pAAV9-Syn-Chronos-GFP or AAV9-hSyn-hCHR2(H134R)-EYFP (Addgene) were used to induce the expression of blue-light activated opsins and GFP. Initially robust expression was achieved with the first lot of pAAV9-Syn-Chronos-GFP (first 6 animals) but expression could not be achieved with subsequent lots. Therefore AAV9-hSyn-hCHR2(H134R)-EYFP was used for the final 13 animals.

Slice preparation and optogenetic stimulation

Ten-28 (average 14) day following intravitreal virus injections, mice were deeply anesthetized with isoflurane and decapitated. At this point, the mice were at least 4 weeks of age, a time point when retinogeniculate and inhibitory circuits have matured.^12,18,51 The brain was removed from the head, chilled in cold slicing solution (in mM: 2.5 KCl, 26 NaHCO3, 2.5 KCl, 1.25 NaH2PO4, 10 MgCl2, 2 CaCl2, 234 sucrose, and 11 glucose) for 2 min, and quickly transferred into a Petri dish with room temperature slicing solution to block the brain for subsequent sectioning. Coronal slices (300 μm) through the dorsal lateral geniculate nucleus (dLGN) were cut in cold slicing solution using a vibratome (Leica VT1000 S). Slices were then transferred into a room temperature incubation solution of oxygenated (95% O2/5% CO2) artificial cerebrospinal fluid (ACSF) containing the following (in mM: 126 NaCl, 26 NaHCO3, 2.5 KCl, 1.25 NaH2PO4, 2 MgCl2, 2 CaCl2, and 10 glucose) for 30 min to 6 h. Individual slices were transferred into a recording chamber, which was maintained at 32°C by an inline heater and continuously perfused with room temperature oxygenated ACSF (2.5 ml/min, 95% O2/5% CO2). Slices were stabilized by a slice anchor or harp (Warner Instruments, Hamden, CT, United States). Neurons were visualized on upright microscopes (Olympus, BX51WI) equipped with both differential interference contrast optics and filter sets to detect fluorescence in the sections using 4X or 60X water-immersion objectives (Olympus, Center Valley, PA, United States) and a CCD camera. In GAD67-GFP mice, geniculocortical cells were identified as neurons that did not contain GFP, while interneurons were identified as cells that contained GFP.

Recording electrodes were pulled from borosilicate glass capillaries (World Precision Instruments, Sarasota, FL, United States) by using a Model P-97 puller (Sutter Instruments, Novato, CA, United States). To record excitatory postsynaptic potentials (EPSPs) or excitatory postsynaptic currents (EPSCs) in geniculate neurons, electrodes were filled with a potassium-based intracellular solution containing the following (in mM): 117 K-gluconate, 13.0 KCl, 1 MgCl2, 0.07 CaCl2, 0.1 EGTA, 10 HEPES, 2 Na2-ATP, and 0.4 Na2-GTP, with pH adjusted to 7.3 using KOH and osmolarity 290 –295 mOsm. To record inhibitory post-synaptic currents (IPSCs) as well as EPSCs in geniculate neurons, electrodes were filled with a cesium-based internal solution containing (in mM): 117 Cs-gluconate, 11 CsCl, 1 MgCl2, 1 CaCl2, 0.1 EGTA, 10 HEPES, 2 Na2-ATP, 0.4 Na2-GTP, with pH adjusted to 7.3 using CsOH and osmolarity of 290 –295 mOsm. Biocytin (0.5%) was added to both intracellular solutions to allow subsequent examination of the morphology and location of the recorded neurons.

Whole-cell recordings were concentrated in the binocular regions (dorsomedial half) of the central dLGN (i.e. where the most prominent patch of labeled terminals labeled from the ipsilateral eye could be detected). Recordings were carried out on both sides of the brain so that each opsin was expressed in either the ipsilateral or contralateral dLGN during recordings. Recordings were obtained with an Axon Instruments multiclamp 700B amplifier (Molecular Devices), and a Digidata 1440A was used to acquire electrophysiological signals. The stimulation trigger was controlled by Clampex 11.03 software (Molecular Devices). The signals were sampled at 20 kHz, and data were analyzed offline using Matlab. For current-clamp recordings, voltage signals were obtained from cells with resting potentials of −50 to −65 mV. For voltage-clamp recordings, currents were recorded at 0 mV or −60 mV.

For photoactivation of retinogeniculate terminals, light from a blue (Prizmatix UHP 460) and/or red (Prizmatix UHP 630) light-emitting diode was reflected into a 60X water-immersion objective. This produced spots of light onto the submerged slice with diameters of ∼0.3 mm (at full power, blue light 107.2 mW/mm², red light 228.7 mW/mm²). Pulse duration and frequency were under computer control. To activate Chrimson-expressing and/or ChR2/Chronos-expressing retinogeniculate terminals the following light activation protocol was used: 500ms of 20Hz red light pulses (10ms duration), followed by 5 seconds of no light, followed by 500ms of 20Hz blue light pulses (1ms duration), followed by 3.5ms of no light, followed by 1.5 seconds of continuous red light stimulation with simultaneous 500 ms of 20Hz blue light pulses (1ms duration) during the last 500ms of the continuous red light stimulation. Initial recordings were carried out with the blue and red light at full power to evaluate the presence or absence or postsynaptic responses to photoactivation of retinogeniculate terminals. An optically-evoked postsynaptic response was accessed by determining if the amplitude exceeded twice the root mean square of the baseline.^40,51 If light activation of retinogeniculate terminals induced spikes in the postsynaptic neurons, the light intensity was reduced to record postsynaptic responses just below the threshold for action potential activation. The location of the patch pipette was photographed at the conclusion of each recording. To block GABAergic transmission pharmacologically, in some experiments GABA receptors (GABA_A) were blocked via bath application of the antagonist 2-(3-carboxypropyl)-3-amino-6-(4-methoxyphenyl)-pyridazinium bromide (SR95531, 20 μM; Tocris Bioscience, catalog #1262).

Following recording, slices were placed in a fixative solution of 4% paraformaldehyde in 0.1 M, pH 7.4 phosphate buffer (PB) for at least 24 hours. The sections were then rinsed in PB and incubated overnight in a 1:1000 dilution of streptavidin conjugated to AlexaFluor-633 (Invitrogen) in PB containing 1% Triton X-100. The following day, the sections were rinsed in PB and mounted on slides to be imaged with a confocal microscope (Olympus FV1200BX61). Confocal images of each slice were aligned with an outline of the central dLGN and biocytin-filled interneurons and relay cells were plotted.

Quantification and statistical analysis

Ipsilateral versus contralateral EPSC amplitudes, the ocular dominance index (ODI, calculated as (R_c – R_i)/(R_c + R_i), and the membrane resistance of relay cells and interneurons were compared using paired t-tests. The confidence interval for comparing differences in proportions was calculated using the normal distribution approximation with the formula $C\mathrm{I}=\left({\hat{\mathrm{p}}}_1-{\hat{\mathrm{p}}}_2\right)\pm Z_{1-\raisebox{1ex}{$\alpha $}\!\left/ \!\raisebox{-1ex}{$2$}\right.}\sqrt{\frac{{\hat{\mathrm{p}}}_1\left(1-{\hat{\mathrm{p}}}_1\right)}{{\mathrm{n}}_1}+\frac{{\hat{\mathrm{p}}}_2\left(1-{\hat{\mathrm{p}}}_2\right)}{{\mathrm{n}}_2}}$ with ${\hat{\mathrm{p}}}_1,{\hat{\mathrm{p}}}_2$ being the proportions of the compared groups, and $n_1,n_2$ being their sample sizes, and $Z_{1-\raisebox{1ex}{$\alpha $}\!\left/ \!\raisebox{-1ex}{$2$}\right.}$ the z-score (critical value) corresponding to $\alpha$ significance level. The significance level was set to 5%. The results of all statistical tests can be found in the results text as well as the legends for Figure 2 and Figure S1. In all cases, n = number of recorded cells.

Published: June 19, 2025