Intrinsic ON responses of the retinal OFF pathway are suppressed by the ON pathway

Abstract

Parallel ON and OFF pathways conduct visual signals from bipolar cells in the retina to higher centers in the brain. ON responses are thought to originate by exclusive use of metabotropic glutamate receptor 6 (mGluR6) expressed in retinal ON bipolar cells. Paradoxically, we find ON responses in retinal ganglion cells of mGluR6-null mice, but they occur at long latency. The long-latency ON responses are not blocked by metabotropic glutamate or cholinergic receptor antagonists and are not produced by activation of receptive field surrounds. We show that these longer-latency ON responses are initiated in the OFF pathways. Our results expose a previously unrecognized intrinsic property of OFF retinal pathways that generates responses to light onset. In mGluR6-null mice, long-latency ON responses are observed in the visual cortex, indicating that they can be conducted reliably to higher visual areas. In wild-type (WT) mice, APB (DL-2-amino-4-phosphonobutyric acid), an mGluR6 agonist, blocks normal, short-latency ON responses but unmasks longer-latency ones. We find that these potentially confusing ON responses in the OFF pathway are actively suppressed in WT mice via two pharmacologically separable retinal circuits that are activated by the ON system in the retina. Consequently, we propose that a major function of the signaling of the ON pathway to the OFF pathway is suppression of these mistimed, and therefore inappropriate, light-evoked responses.

Figure 1.

Neurons in the visual cortex of mGluR6 homozygous null mice respond to the onset of light stimuli with a long-latency response. Multiunit, extracellular electrode recordings of responses to 2 s, full-field light presentations were made from the visual cortices of WT and mGluR6-null mice anesthetized with urethane. A, Peristimulus time histograms (PSTHs) of two recordings from neurons in WT visual cortex. B, PSTHs of two recordings from neurons in visual cortex of the mGluR6 homozygous null mouse. Bin width, 50 ms. The light/dark bar below the histograms represents the light stimulus, which originated from a computer monitor. C, Summary of visual cortical neuron latencies to peak frequency after light onset. Light bars show WT latencies; black bars show mGluR6-null latencies. Neurons in the mGluR6 null have a longer latency on average compared with that of WT. Frequency is in impulses (spikes) per second (imp/sec).

Figure 2.

RGCs of the mGluR6 homozygous null retina have a long-latency ON response. Responses of RGCs were recorded from WT and mGluR6-null retinas using a multielectrode array. These responses were sorted into individual units representing spiking responses from individual RGCs. A, Example raster plots of spiking responses from WT RGCs to individual stimulus presentations and their resulting peristimulus time histograms (PSTHs) are shown. Short-latency, transient ON, sustained ON, ON–OFF, and OFF responses are all recorded by the array. B, Example raster plots and corresponding PSTHs for mGluR6-null neurons with prominent long-latency ON responses are shown. No transient or sustained ON responses with short latency were recorded in these retinas. For all retinal PSTHs, bin width is 10 ms. The light/dark bar below the histograms represents the light stimulus, which originated from a computer monitor and was projected onto the retinal surface. Frequency is in impulses (spikes) per second (imp/sec). C, A large percentage of RGCs from mGluR6-null retinas had a long-latency ON response and an OFF response. Values indicate the number of neurons recorded from 32 null retinas.

Figure 3.

mGluR antagonists do not prevent the long-latency ON response in the mGluR6 homozygous null retina. Antagonists to group I, II, and III metabotropic glutamate receptors did not block the long-latency ON response in the mGluR6-null retina. Control, Untreated; LY341495 (50 μm), antagonist for mGluR2, -3, -7a, -8, and partially for mGluR4a at this concentration; LY367385, mGluR1a antagonist; MPEP, mGluR5 antagonist; CPPG, mGluR4 and mGluR6 antagonist. The percentage of total units with a long-latency ON response was compared before and after treatment of retinas with antagonists. In contrast, cadmium (100 μm), which blocks synaptic transmission, blocked all light responses. Error bar indicates SEM.

Figure 4.

The electroretinogram of the mGluR6 homozygous null mouse does not have a delayed b-wave during 2 s light stimulus presentations. Electroretinograms in response to 2 s light stimuli from a WT mouse (gray curve) and from an mGluR6-null mouse (black curve) are shown. The a-, b-, c-, and d-waves are indicated. The b-wave in the WT results from ON bipolar cell activity. No b-wave activity was apparent in the mGluR6-null mouse. The light/dark bar below the traces represents the light stimulus.

Figure 5.

The long-latency ON response of ganglion cell neurons in the mGluR6 homozygous null retina has center-surround characteristics. Receptive field organization of individual neurons was determined using nine light spots of increasing diameter presented three or five times each. Total spike counts for windows of time representing short (open triangles)- and long (open circles)-latency ON responses and OFF responses (X symbols) were normalized to the maximum count for each neuron. A, Examples of WT responses indicating typical organization of the center and antagonistic surround for an ON, OFF, and ON–OFF neuron are shown. The antagonistic surround depresses the spiking response during the presentation of the larger diameter spots. The previously published data of Stone and Pinto (1993) (dashed line) from intracellular recordings of whole-mounted mouse retina are presented in the left example. B, Examples of receptive field organization based on responses recorded from mGluR6-null retinas are shown. No short-latency ON responses were observed. Of the nine spot sizes presented, the optimal size for the long-latency response tended to be larger than that for the OFF responses in the same neurons and larger than the optimal WT spot size. Nonetheless, an antagonistic surround was observed, suppressing the responses to larger spot sizes. C, Raster plots of the spike responses to multiple presentations of spots of increasing diameter are displayed for the neuron from an mGluR6-null retina that is graphed on the right in B. The increase in latency associated with increasing spot diameter is attributable to increased brightness caused by scatter in the center as the spot size is made larger as well as activation of larger amounts of the center area.

Figure 6.

Blockade of ON bipolar cell function in the WT retina causes a long-latency ON response to emerge. Treatment of WT retinas with the mGluR6 agonist APB leads to generation of a long-latency ON response in many RGCs. A–D, Examples of raster plots (top graphs) and corresponding peristimulus time histograms (PSTHs) (bottom graphs) from strongly responding individual RGCs during 2 s light presentations before (top panel) and after (bottom panel) treatment with APB (100 μm) show blockade of short-latency ON responses, both transient and sustained, and the presence of long-latency ON responses in APB. In D, an OFF neuron in control saline has a long-latency ON response in APB. Light/dark bars below the panels indicate the light stimulus timing. Note that, in C and D, the first 60 ms of the response are not shown. For these RGCs, increased spontaneous dark activity in the presence of APB continued into the light onset period, giving the mistaken impression of a short-latency ON response during some of the stimulus presentations. The actual short-latency response, which has a latency >60 ms under these conditions, can be seen in the control responses in C. Frequency (Freq) is in impulses (spikes) per second (imp/sec).

Figure 7.

Blockade of inhibition in the WT retina causes a long-latency ON response to emerge. Treatment of WT retinas with antagonists to GABA_A (SR95531), GABA_B (CGP55845), GABA_C (TPMPA), and glycine (strychnine) receptors leads to generation of a long-latency ON response in many RGCs. A–C, The top panels show examples of neuronal responses (as trial spike rasters and their corresponding PSTHs) to 3 s light stimulus presentations in control saline, whereas the bottom panels show the responses of the same neurons while the retina was perfused with the antagonist mixture. Short-latency ON responses were maintained, and a long-latency ON response appeared. In C, the neuron also gained an OFF component to its response in the antagonists. Light/dark bars show the timing of the light stimulus. Frequency (Freq) is in impulses (spikes) per second (imp/sec).

Figure 8.

Blockade of ON bipolar cell function in combination with inhibitory antagonists in the WT retina can enhance the emergent long-latency ON response. APB, an mGluR6 agonist that blocks the short-latency ON response, can enhance the long-latency ON response that emerges in the WT retina when inhibitory receptors are antagonized. A–D, The top panels show examples of the responses of WT neurons to 3 s light stimulus presentations in control saline. The middle panels show the responses of these same neurons in the presence of GABA_A, GABA_B, GABA_C, and glycine receptor antagonists. The bottom panels show the responses of these same neurons in the presence of APB (100 μm) in addition to the inhibitory receptor antagonists. In all cases, APB eliminated the short-latency ON response. In D, the ON neuron gained an OFF component to its response in the presence of the inhibitory receptor antagonist cocktail only exhibited a long-latency ON response once APB was added. Light/dark bars show the timing of the light stimulus. Frequency (Freq) is in impulses (spikes) per second (imp/sec).

Figure 9.

Inhibitory antagonists with and without APB cause a long-latency ON response to emerge in WT retina. Retinas perfused with normal saline (ctrl) (n = 6 total retinas) or treated with a cocktail of inhibitory antagonists (inhib. antags.) (n = 6 of 6 retinas) (see Results) or the cocktail plus APB (n = 5 of 5 retinas) are compared. The bar for each condition shows the mean percentage ± SEM of total units recorded per retina that had a long-latency ON response.