Publicreport_wp1


Public Report Of The Project's Results: WP1 - Retina Neurobiology

WP1: Retina Neurobiology

1. Dynamic spatio-temporal image processing in genetically identified retinal circuits

The brain incorporates neuronal circuits where graded changes in the input lead to distinct circuit computations and animal behaviors. Such discontinuous changes in the computation performed by a circuit could be achieved by switching between discrete circuit architectures. However, this has not been demonstrated experimentally. Here, by sliding through the light levels from starlight to daylight-like conditions, we identify retinal ganglion cell types that switch the visual feature they extract at a specific light level, while other ganglion cells maintain or smoothly adjust their response properties across the range of input
intensities. A combination of two-photon targeted patch clamp recordings, pharmacology, knock-out and monosynaptic viral tracing experiments revealed that the switch is located at a spiking wide field inhibitory neuron (amacrine cell) that provides direct input to the ganglion cell. We find little or no inhibition at light levels when only rods are active, while inhibition appears abruptly after the stimulus intensity crosses the light level that activates cones. The visual feature extracted by the ganglion cell can be toggled back and forth by changing the light levels around the cone threshold. The circuit architecture mediating the switch centers on the different types of excitatory synapses, electrical or chemical that drive the amacrine and ganglion cell, respectively. We postulate that weak excitatory input, via electrical synapses, together with the spiking threshold in amacrine cells act as a switch.

Does human vision show similar abrupt changes in behavior around the cone threshold? By pairing a color-discrimination and contrast sensitivity test across light levels we reveal a switch-like component in spatial integration properties of human visual function, which could be toggled on and off around cone threshold. This work uncovers the circuit elements and their interactions composing a circuit switch that controls the inhibitory input to a central neuron. By toggling the balance between excitation and inhibition one can alternate between two discrete visual features extracted by an individual visual channel.



Figure 1: Recordings from PV 1 cell at different light levels. The cells were stimulated with either small or large spot. The cell switches response properties at a specific light level.


2. Recordings from PV cells

We set up an experimental strategy to record from PV cells (Figure 2). The strategy had six steps. First, we took a two-photon z stack in a brightly labeled region. After obtaining the stack we could clearly see the stratification of the cells in the inner plexiform layer. Based on stratification we targeted a specific cell for recording. Second, we used cell-attached mode to record light evoked spikes from the cell. After these recordings were done, we changed electrodes and re-targeted the same cell. Third, we recorded the currents of the PV cell to steps of voltage in order to characterize its cellular input-output function. Fourth, we recorded excitation by clamping the cell to the reversal potential of chloride channels. Fifth, we recorded inhibition by clamping the cell to the reversal potential of the glutamate gated non-selective ion channels. Sixth, we filled the cell with neurobiotin for post hoc labeling and after the histochemical steps we reconstructed the detailed morphology of the cell under the confocal microscope.



Figure 2: Targeting a class of genetically-identified neurons in the Fluomouse retina. (a) YFP expression in a Parvalbumin-Cre x Thy1-Stop-EYFP (Fluomouse) retina (epifluorescence). Left, in vivo image through the lens of the eye. Right, ex vivo wholemount image. Filaments are axons extending to the optic disk (center). Scale bars: 250 µm. (b) Maximum intensity z-projection of a two-photon image series of a region of a wholemount PV retina. (c) Infrared image of the GCL (grayscale) of the same region in (b) with an overlay of the two-photon signal (green). (d) Light-evoked spiking from the YFP-positive PV cell in (b) and (c) (indicated by an asterisk). Visual stimulus (below): 200 µm black then white spot on a gray background, aligned to dendritic field center. Dashed lines indicate the onset or offset of the spot. Black bars: spike trains. Below, discrete firing-rate (in Hz, 25 ms time bins). (e) Voltage steps during whole cell voltage clamp of the same PV cell (left) and corresponding current-voltage (I-V) relations (right). Left, 10 mV steps from -110 to 30 at two time points during the recording. Right, mean current taken from light-gray region from voltage steps, plotted against the voltage at each step (I-V curve). Black, 0 min after break-in (QX314 from intracellular solution has not yet blocked the voltage-dependent sodium currents). Gray, 20 min after break-in. (f) Whole cell voltage clamp of same the PV cell measuring inward currents (excitation, black trace) and outward currents (inhibition, gray trace), respectively. Same stimulus as in (e). (g) Maximum intensity z-projections of confocal images in the fixed, immunostained Fluomouse retina. Left, the recorded PV cell, filled with neurobiotin (magenta). Axon shown in top left. Middle, region from (b) and (c) shown in rotated white box, with same PV cell marked with an asterisk; anti-GFP (green). Right, merge showing GFP-positive neurobiotin-filled PV cell. (h) Maximum intensity yz-projection of the IPL region for the PV cell. Left, dendrites of the PV cell in the IPL, filled with neurobiotin (magenta). Middle, ChAT marker antibody labelling two bands (cyan). Right, merge showing bistratified PV cell colocalising with the ChAT bands. Scale bars in (b), (c), (g), (h): 20 µm.

The spiking output of the PV cells to spots of light are shown on Figure 3 together with an example morphology. We found that each of the 8 morphological types of PV cells (PV0-PV7) has a very distinct firing pattern, representing a different spatio-temporal filter. We found 2 ON cells, 4 OFF cells, and 2 ON-OFF cells. PV 0 was directional selective, representing a class of retinal ganglion cells called direction selective retinal ganglion cells.

Figure 4 and figure 5 summarize the recordings of excitation and inhibition from all of the PV cells together with the firing rate. It is clearly seen that each PV cell type receives a specific pattern of inhibitory and excitatory inputs. These input differ in their dynamics as well as spatial weightings. Importantly what is specific to each of the PV cells is their relative timing and phase (ON,OFF or ON-OFF) of inhibition and excitation. We plotted at each time point following a flash of light the mean (for each PV cell type) magnitude of inhibition and excitation on a phase plot. These phase plots (Figure 6 and Figure 7) show the different dynamic motifs of inhibition-excitation interactions.



Figure 3: Spiking responses of PV cells to spots of different sizes. For each cell type an example morphology is shown on the right. Recordings from n>10 cells of the same type is shown. Individual spikes are labeled by vertical lines are shown in red and black alternating colors for each recorded cell. Repetitions of recordings from the same cell are shown in different rows. On the top of each cell type-recordings, the average spike frequency of the cells of the same type is shown. – sign refers to contrast decrement stimulus, + sign refers to contrast increment stimulus.



Figure 4: PV0-PV3: Inhibition, excitation and spiking rate. Top row: light onset (grey to spot stimulus). Bottom row: light offset (spot stimulus to gray). First 600 ms after stimulus event is shown for six spot sizes. Red traces: discrete firing rate in 25 ms bins, normalized to peak rate (value below each column). Black and gray traces: excitation ± SEM (dark grey band) and inhibition ± SEM (light gray band), normalized to peak value (amplitude in pA below each column). Excitation has been flipped for comparison with the relative timing of inhibition.



Figure 5: PV4-PV7: Inhibition, excitation and spiking rate. Top row: light onset (grey to spot stimulus). Bottom row: light offset (spot stimulus to gray). First 600 ms after stimulus event is shown for six spot sizes. Red traces: discrete firing rate in 25 ms bins, normalized to peak rate (value below each column). Black and gray traces: excitation ± SEM (dark grey band) and inhibition ± SEM (light gray band), normalized to peak value (amplitude in pA below each column). Excitation has been flipped for comparison with the relative timing of inhibition.



Figure 6: PV1-PV4: Inhibitory-excitatory interactions during visual stimulation. At each time point during response mean (for a cell type) inhibition (y-axis, pA) plotted against mean excitation (x-axis, pA). PV-1 and PV-2 are shown for white spot stimuli: PV-3 and PV-4 are shown for black spot stimuli. Black points represent inhibitory-excitatory interactions at light OFF (offset of white, or onset of black); white points represent interactions at light ON (onset of white, or offset of black). A ring indicates the first time point after the stimulus change. Interactions are plotted for the first 1000 time points (each representing 1 msec). Unity line artificially divides regions of excitatory and inhibitory dominance.



Figure 7: PV5-PV7 and PV0: Inhibitory-excitatory interactions during visual stimulation. At each time point during response mean (for a cell type) inhibition (y-axis, pA) plotted against mean excitation (x-axis, pA). All are shown for black spot stimuli. Black points represent inhibitory-excitatory interactions at light OFF (offset of white, or onset of black); white points represent interactions at light ON (onset of white, or offset of black). A ring indicates the first time point after the stimulus change. Interactions are plotted for the first 1000 time points (each representing 1 msec). Unity line artificially divides regions of excitatory and inhibitory dominance.


3. Recordings from PV ganglion cells using natural movie stimuli.

We developed an experimental approach to compare the responses of PV ganglion cells to natural movies. We recorded responses to natural movies in 26 PV cells. The key design step that allowed us to compare responses was that we projected the center of the natural movie to the center of the dendritic field of each recorded PV cell. We found that PV cells of the same type have highly reproducible patterns of responses to the same stimulus and PV cells of different types have very different response patterns.



Figure 8: Parallel processing of natural movies in by different PV cell types. (a) Two-photon maximum intensity z-projections of a region of the PV
retina at two different levels: GCL (ganglion cell layer), ON IPL (innermost inner plexiform layer), and below the full z-projection (GCL with full IPL). Numbers indicate each targeted PV cell. Scale bar, 20 µm. (b) Light-evoked spiking activity of PV cells from (a), for increasing-size spot stimulus. Left of each recording is a two-photon overlay (green) with the patch pipette on the PV cell in the GCL (greyscale). (c) Area-response profiles for PV cells 1-6. PV cells 5 and 6 have a higher firing rate in the surround (large spots). Right, PV cell type for each recorded cell, identified by two-photon imaging (morphology-based) and the spiking profile from the spot stimuli (physiology-based). (d) Three natural scenes presented to all 7 targeted cells (scenes were centered about the soma). Each line represents a single recording. Horizontal bars represent the onset and offset of each movie, with uniform gray as background before and after. Each point is a spike, and each cell is given a different color. Some movies were presented more times to some cells than others.


4. Description of the presynaptic cell types of three selected PV ganglion cells.

We used monosynaptically restricted viral tracing from PV1, PV6 and PV0 cells to determine the presynaptic circuitry of these cells. Moreover in PV0 cells and in general direction selective cells we used, the first time, transsynaptic tracers, which express genetically encoded calcium sensors. These sensors express in a single PV cell (starter cell) and in many other cells that are synaptically connected to the specific PV cell. We could then image the concerted activity of the circuit elements of a single PV cell while the circuit is responding to light. Here we discuss findings about each circuit.

1. Circuit of PV1 cells (Figure 9). Among others we found that a specific type of wide field amacrine cell which monostratified in the same retinal stratum as the PV1 cell is connected to PV1 cells. Using approaches of electrophysiology, pharmacology and anatomy we described the function of the wide filed amacrine cell, which provided lateral inhibition to PV1 cells. Interestingly the lateral inhibition was switched on at a specific light level exactly when cones are activated.



Figure 9: Camera lucida reconstruction of circuit elements of a PV1 cell. PV1 cell is shown blue. Two wide field amacrine cells with specific morphology. These cells are synaptically connected to the PV1 cell is shown in green and brown. Scale bar is 100 µm. Note the very similar morphology of the two amacrine cells, having 8-9 long processes radiating from the cell body.

2. Circuit of PV6 cells (Figure 10). Similar to PV1 cell we found that a specific type of wide field amacrine cell which monostratified in the same retinal stratum as the PV6 cell is connected to PV6 cells. Using approaches of electrophysiology, pharmacology and anatomy we described the function of the wide filed amacrine cell, which provided lateral inhibition to PV6 cells. As in PV1 cells the lateral inhibition was switched on at a specific light level exactly when cones are activated. It appears that the PV1 cell, an ON cell, and the PV6 cell, an OFF cell, have very similar function and circuit structure.



Figure 10: Camera lucida reconstruction of a circuit element of a PV6 cell. PV6 cell is shown red. A wide field amacrine cell with specific morphology (blue). Scale bar is 100 µm.

3. Circuit of directional selective cells (PV0 is a member of this group, Figure 11 and Figure 12). Here we specifically discuss the so-called ON-direction selective (ON-DS) cells. We found however that the other members have the same circuit elements. We found that ON-DS cells are connected by a single type of bipolar cell, called type 5. Furthermore we confirmed that starburst cells are connected to ON-DS cells. Using transsynaptic viruses that express a genetically encoded calcium sensor we found that while ON-DS cells are direction selective, the bipolar cell terminals that provide excitatory input to the ON-DS cells are not selective.



Figure 11: Image of a retina from a mouse line in which a single type of ON-DS cell (upward preferring) is labeled with GFP (green). Here the transsynaptic virus expresses RFP (magenta) was initiated from a brain area where only ON-DS cells project. The axons of individual ON-DS cells are visible in magenta. Circuit elements of ON-DS cells can be seen as magenta dots that cluster around ON-DS cells.



Figure 12: Bipolar cell types that are connected to single ON-DS cells. On this figure the virus labeled cells are shown in green since they express the GCaMP3 a green fluorescent calcium sensor. (a) Confocal images of a retina in which an ON DS cell (*) and starburst amacrine cells (magenta) are infected with trnassynaptic virus expressing GCaMP3 (green). (b) Side view (left) and top view (right) of an example of a labeled type-5 bipolar cell. The image for the top view was taken at the depth of the proximal ChAT-labeled layer. (c) Nine more examples of labeled type-5 bipolar cells.