Figure 1.
Pharmacological isolation of mouse ipRGCs
. Representative raw voltage traces from a wildtype mouse retina (C57Bl6) recorded simultaneously from 15 electrodes of a multielectrode array. Most electrodes recorded spikes originating from more than one cell. Timing of full-field light stimulus is indicated by the step above the traces and by the time scales below the traces. Left column: Spike responses recorded in control Ames' medium to a relatively dim scotopic flash (2.3 Rh*/rod/s; 500 nm; 1 second) that should excite rods but not cones or melanopsin[59]. Multiunit light responses are detectable on every channel, and exhibit various forms (ON, OFF or ON-OFF; transient or sustained). Right column: Light-evoked spike activity recorded on the same electrode in the presence of drugs blocking rod and cone signaling to ganglion cells. To permit activation of melanopsin, the stimulus was longer and >10,000-fold brighter than that at left (∼2×105 Rh*/rod/s sampled at 500 nm; 10 sec). This evoked responses on a minority of channels (arrows) and these exhibited the long onset latency and persistent post-stimulus discharge typical of melanopsin-mediated intrinsic responses (note the slower time scale).
Figure 2.
Input from the most sensitive rod pathway to ipRGCs.
A: Raster plots of synaptically driven spike responses of a single typical ipRGC in response to a series of flashes increasing in intensity (500 nm; 1 sec). Recordings were extracellular, obtained on a multielectrode array in a dark-adapted wildtype (C57Bl6) mouse retina. B: Irradiance-response curve plotted in semi-log coordinates and based on data averaged from 23 wildtype mouse ipRGCs. For each cell, response amplitude at a given intensity was expressed as the firing rate during that stimulus normalized to the maximal light-evoked firing rate at any intensity. These values were averaged across all cells for each intensity and plotted, with error bars representing the standard error of the mean. The smooth curve represents the Michaelis-Menten fit to these points and the single triangle along the abscissa indicates the response threshold (5% of maximum, calculated from the fit). This irradiance-response profile closely resembles those of ON type RGCs receiving input from the primary rod pathway [50], [51]. C: Distribution of the thresholds of the ON responses in ipRGCs (left; n = 8) in comparison with those of non-ipRGCs (right; n = 22) recorded in the same piece of retina. The thresholds of ipRGCs all located in the lower part of this plot, indicating they are among the most light-sensitive mouse RGCs.
Figure 3.
Light-evoked synaptic inputs to ipRGCs are less sensitive in Cx36 knockout mice.
A: Raster displays comparing responses of a single representative ipRGC in a Cx36 knockout mouse (right column) with those of an ipRGC in a wildtype littermate (left column). Responses of each cell are shown for a series of 500 nm flashes of increasing intensity. The dimmest stimulus evoking a clear response was approximately three log units higher in the Cx36 KO mouse (∼22.7 Rh*/rod/s) than it was in the wildtype control (0.02 Rh*/rod/s; see also the data for the C57Bl76 control mice; Fig. 2). B: Group data comparing irradiance-response curves for ipRGCs recorded in the Cx36 knockout (black squares; n = 31) and wildtype animals (open squares; n = 24). Data come from 4 pairs of littermates consisting of one wildtype and one knockout animal. The triangles near the abscissa indicate the response thresholds (5% of maximum; open triangle, Cx36+/+; filled triangle, Cx36−/−), which are about 3 orders of magnitude higher in the knockouts than in the wildtype mice. This reduction in sensitivity presumably reflects the loss of input from the primary rod pathway.
Figure 4.
Pharmacological blockade of gap junctions mimics Cx36 knockout, reducing sensitivity of synaptic input to ipRGCs.
A: Raster plots of the response of a single ipRGC in a wildtype mouse (C57Bl6) to 500 nm light steps recorded before (left), during (middle) and after (right) bath application of MFA, a gap junction blocker. Stimuli increased in intensity by ∼1 log unit with each trial, shown in successive rows of the raster. MFA application abolished the light response to dim stimuli and elevated threshold by about 3 log units; spontaneous activity was also much reduced. Sensitivity largely recovered upon washout of the drug. B: Group data comparing irradiance-response functions ipRGCs recorded in MFA-treated wildtype mouse retina (n = 26) with those recorded in Cx36 knockout mice (n = 41). The triangles near the abscissa indicate the response thresholds (open triangle, C57Bl6, MFA; filled triangle, Cx36−/−). The dashed curve, is reproduced from Fig. 2B, permits comparison with C57Bl6 mice under control conditions. The reduction in light sensitivity was comparable in these two manipulations, supporting the view that the effect is due in both cases to loss of synaptic circuits that are dependent upon gap junctions rather than to off-target genetic or drug effects.
Figure 5.
High-threshold light-evoked synaptic responses survive interruption of primary and secondary rod pathways by Cx36 knockout.
A: Raster plots of light responses in a representative ipRGC from a Cx36 KO mouse recorded first with synapses functional (left) and then with synapses blocked (right). In synaptic blockade, the light response exhibited the long latency and prominent poststimulus persistence of spiking characteristic of melanopsin-mediated responses. With synaptic input intact (left), responses were brisker and more sensitive than the melanopsin-driven response. Because both primary and secondary rod ON pathways were disrupted by the genetic deletion of Cx36, these brisk, synaptically mediated responses were presumably driven by cones. Note that stimulus intensities are not matched in the left and right columns of panel A. Also, stimulus duration was increased 10-fold under synaptic blockade (right) because a 1 sec stimulus was too short to evoke an intrinsic photoresponse at most intensities tested. B: Group irradiance-response data of the sort illustrated in panel A for ten ipRGCs recorded from the same piece of Cx36 KO mouse retina. The triangles along the abscissa indicate the response thresholds (5% of maximum; open triangle, synapses functional; filled triangle, synapses blocked). The presumptive cone-driven response is at least 2 log units more sensitive than the melanopsin drive; the difference would have been greater if stimulus duration had not been increased under synaptic blockade to enhance the intrinsic response. Qualitatively similar results were obtained in a total of 8 retinas (data not shown).
Figure 6.
Evidence for ON channel input to ipRGCs from both M-cones and S-cones.
A, B: Raster plots of responses to spectrally narrowband stimuli for ten ipRGCs recorded simultaneously in a Cx36 knockout mouse retina. These brisk ON responses were presumably exclusively cone-driven because Cx36 KO mice lack the primary and secondary rod ON pathways and because the flashes were subthreshold for activation by intrinsic melanopsin phototransduction. The response to a 400 nm stimulus (A) closely resembled that to a 500 nm stimulus (B), despite the fact that the former would be more effective in stimulating S-cones and the latter much more effective for M-cones. The two stimuli were approximately matched for photon flux density (400 nm: ∼5×1011 photons cm−2 s−1; 500 nm: ∼7×1011 photons cm−2 s−1). C: synaptic blockade eliminated these responses (in this case, the 400 nm response), confirming their origin in the outer retina. D: intrinsic, melanopsin-driven responses revealed in the same cells under synaptic blockade by increasing the intensity, spectral bandwidth and duration of the light. E, F: Group data summarizing the irradiance-response behavior at 400nm and 500nm for these cells (n = 10), with response amplitudes measured either in terms of the number of evoked spikes (E) or the mean firing rate during the stimulus (F). The curves for the two wavelengths are nearly identical. This is inconsistent with a pure M-cone input, which would predict sensitivity at 500 nm ∼1 log unit greater than that at 400 nm. It is also inconsistent with a pure S-cone input, which would predict >4 log unit greater sensitivity at 400 nm than at 500 nm. We confirmed in a larger sample (25 ipRGCs from three Cx36 KO retinas) that 400nm and 500nm stimuli of matched irradiance evoked comparable responses. G: Whole-cell voltage clamp recordings from a representative example of an M1 cell (top panels) and an M2 cell (bottom panels) from the dorsal retina, where middle and long-wavelength cone opsins are expressed in distinct cone types. Responses are shown to a 1 s step of 360 nm (left) or 500 nm (right) light (horizontal black line). Different colored traces represent responses to different light intensities (in photons•cm−2•s−1) for the M1 cell: 3×1011−2×1014 (360 nm) and 2×1012−6×1014 (500 nm); and for the M2 cell: 7×1010−5×1013 (360 nm) and 1×1011−2×1014 (500 nm). Similar results were obtained in all M1 and M2 cells recorded, whether in the dorsal retina (six M2 cells and one M1 cell) or the ventral retina (four M2 cells and one M1 cell).
Figure 7.
Surprisingly persistent cone-driven response in ipRGCs.
Data are from a Cx36 knockout mouse retina, in which the primary and secondary rods paths are eliminated and ON channel synaptic input to ipRGCs comes mainly or exclusively from cones. A, B: Raster plots (above) and peri-stimulus time histograms (below) of responses of a single ipRGC (A) and of multiunit responses of conventional RGCs (B) to 10 sec light steps of various intensities (500 nm narrowband; n = 7 cells in A; n = 14 multiunit recording sites in B). Firing rates of each cell (or recording site) were first normalized to the maximal firing rate at any intensity, and then averaged over cells (or recording sites) to plot the histograms. Light intensities are listed at the left (in Rh*/rod/sec ). Light responses are always much more sustained in the ipRGCs than in the conventional ganglion cells, even at the two lower intensities, which are below the threshold for intrinsic, melanopsin-mediated phototransduction. Timing of light stimulus is indicated by the marker above the traces. C–F: Plots comparing of steady-state firing during the stimulus (right point) to the prestimulus spontaneous rate (left point) for individual ipRGCs (C, E) and for multiunit recordings of conventional RGCs (D, F). Spontaneous rate was assessed during the one-second interval preceding the stimulus, and steady state rate during the last second of the 10 sec stimulus. All data in C-F obtained simultaneously from a single Cx36 KO retina. Light intensities (in Rh*/rod/sec) are 22.7 in C and D; and 227 in E and F. Both of the intensities are below melanopsin threshold. Note that all ipRGCs showed elevated steady-state firing during prolonged stimulation, whereas nearly all conventional RGCs returned to baseline firing rates during the stimulus.
Figure 8.
ipRGCs with distinct types of intrinsic light response exhibit similar synaptically driven photoresponses.
Upper panel: raster plots of two representative ipRGCs with distinct kinetics of melanopsin-driven intrinsic photoresponse, apparently corresponding to two types defined by Tu et al., (2005) in adult mice. Type II cell (blue) has long onset latency and shorter post-stimulus discharge; Type III has short latency and prolonged afterdischarge. Stimulus was a 60 sec light step (yellow bar; 480 nm, ∼6×1012 photons/cm2·s). Recordings were performed under synaptic blockade. Lower panel: intensity-response series of the same two cells, but without synaptic blockade. The light stimuli were one second flashes 500 nm flashes ascending in intensity within the series. Both cells showed the same irradiance-response properties, with thresholds approximately 0.02 Rh*/rod/s.
Figure 9.
Schematic summary of pathways by which rod signals could reach ipRGCs.
The ipRGC is a hybrid of known melanopsin-expressing ganglion cell types, some of which stratify exclusively in the uppermost OFF sublayer (site 3), others in the ON sublayer (e.g., site 2), and still others in both (see text). Jagged lines represent gap junctional contacts between AII amacrine cells and ON cone bipolar terminals (site 1) and between rods and cones (site 4). The primary rod pathway is shown in red, the secondary rod pathway in green, and a novel pathway from rods to ON cone bipolar cells in blue [35]. All of these pathways have been considered to relay to ganglion cells in the ON sublayer (site2), but they may also pass through ectopic ON cone bipolar terminals in the OFF sublayer to the dendrites of some ipRGCs (site 3; see text). A pathway directly linking rod bipolar cells to ipRGCs, proposed by Ostergaard et al. (2007) [17], is shown in gold. R: rod; C: cone; RB: rod bipolar cell; CBon: ON cone bipolar cell; AII: AII amacrine cell; ipRGC: intrinsically photosensitive retinal ganglion cell. Horizontal gray lines indicate the ON and OFF sublayers of the inner plexiform layer.