TARPγ2 Is Required for Normal AMPA Receptor Expression and Function in Direction-Selective Circuits of the Mammalian Retina

TARPγ2 expression is concentrated in SACs. A, B, Localization of TARPγ2 and ChAT in the IPL of the wild-type mouse retina. C, Average normalized fluorescence intensity profiles of TARPγ2 and ChAT in the mouse IPL (N = 6 mice). The peaks of TARPγ2 staining align with the dendrites of the OFF and ON-ChAT bands. Shading represents ± 1 SEM 0% is the outer and 100% is the inner border of the IPL, respectively. D, E, Localization of TARPγ2 and ChAT in the IPL of a human retina shows a similar pattern to mouse. Scale bar in E = 20 μm and applies to A,B,D,E. F, Relative expression of Cacng2 (TARPγ2) transcript in different mouse amacrine cell types. The ChAT-expressing amacrine cells (SACs, AC_17) show relatively higher levels of Cacng2 than other amacrine cell types. Raw data from Yan et al. (2020a,b). G, Relative expression of Cacng2 transcript across different human amacrine cell types shows higher levels of CACNG2 expression in SACs (Gaba5). H, Expression of AMPAR subunits in SACs (AC_17) relative to other mouse amacrine cell types (data not shown). Note that Gria1 levels are low in SACs, whereas transcript levels of Gria2, Gria3, and Gria4 are higher. Dot size indicates % expressing, dot color indicates relative expression level across rows.

TARPγ2 is associated with specific GluA receptor subunits in mouse retina. A–C, Confocal images of the mouse IPL showing double labeling of GluA1 (A), GluA2 (B), and GluA4 (C) subunits (top panels) with TARPγ2 (second row). Merge of GluA subunits and TARPγ2 in shown in third row. Square ROIs in merged images are shown enlarged below (inset). D–F, Normalized fluorescence intensity profiles of TARPγ2 (green) and GluA1 (D), GluA2 (E), and GluA4 (F) subunits (magenta) from projected confocal image stacks of the IPL. Scale bar in C merge applies to all main panels = 20 μm. Inset scale bars = 5 μm.

Absence of TARPγ2 reduces expression of specific AMPAR subunits in the mouse inner retina. A–D, Immunolocalization and intensity profiles of GluA subunits (A–D), PSD95 (E), and GABA_AR β2-β3 subunit (F), in the IPL of wt (left) and stg mutant mouse (center). ChAT or calretinin were used as reference markers in A–D and are shown in magenta. Right, Normalized average fluorescence intensity profiles from wt (black) and stg mice (red). The average values were obtained from six retinas from six independent animals for each genotype. Shading shows ± 1 SEM. Cyan lines are difference plots showing wt - stg. Pink shading shows position of ChAT bands in A, B, D, and outer CalR bands in C. Statistical comparisons are taken at the position of the OFF and ON ChAT bands using unpaired t tests with the Bonferroni correction for multiple comparisons (see also Extended Data Table 2-1). ns = not significant, **p < 0.01, ***p < 0.001, ****p < 0.0001. Scale bar in F = 20 μm applies to all images.

Absence of TARPγ2 reduces excitatory input to ON-SACs. A, Example of an ON-SAC filled with Alexa 488 hydrazide. B, Total membrane current during depolarizing voltage steps to holding potentials between −90 and +60 mV. Timing of a 175 μm diameter positive contrast stimulus spot shown beneath traces. Shading and error bars show ±1 SD C, Average current–voltage relations for the passive membrane conductance measured at the time points shown by the square symbols in B. D, Average net light-evoked EPSC amplitudes measured at the time points indicated by the circle symbols in B. Lines show the linear fits used to calculate the excitatory and inhibitory conductance components shown in E. E–F, Excitatory and inhibitory conductances were calculated for the data in B. Circles indicate the time points used to calculate peak conductances. The magenta traces show the difference between the het and stg traces. These show the net conductance that is lost in the stg retinas. Error bars and shading are ±1 s.d.

Loss of TARPγ2 produces circuit-specific reductions in inhibition. A, Total membrane current during depolarizing voltage steps to holding potentials between −90 and +60 mV recorded in control (black) and in the presence of SR95531 (SR; 100 μM, cyan). Timing of the 175 μm diameter positive contrast stimulus spot is shown beneath the traces. Shading and error bars show ±1 SD. B, Excitatory conductance calculated for the data in A. C, D, Inhibitory conductance calculated for the data in A. Left panels in A–D are from het SACs, right panels from stg SACs. E, Net GABA_AR-mediated conductance (control−SR) in het (black) and stg (red) SACs. F, Comparison of the inhibitory conductance missing in stg SACs (Fig. 5F, het - stg, magenta) with the GABA_AR-mediated conductance missing in stg SACs (Fig. 6E, het - stg, cyan). G, The traces are replotted from C, D (cyan traces) to compare the non-GABAergic (SR-resistant) inhibition in the het and stg amacrine cells. H, Inhibitory conductance averaged from 12 ON-SACs in het retinas in control (black) and after addition of 1 μM strychnine (strych, green). The thick green trace shows the net conductance blocked by strychnine (control minus strychnine). Error bars and shading are ±1 s.d.

Absence of TARPγ2 does not affect ON-SAC receptive field size. A, Average light-evoked EPSCs in ON-SACs from het (n = 28 cells) and stg retinas (n = 24 cells). EPSCs were elicited by centered light spots (diameters (μm) are shown above traces). B, C, Amplitudes of EPSCs versus stimulus diameter for the time points shown by the corresponding symbols in A. Smooth lines show fits to a difference-of-Gaussians function.Error bars and shading are ±1 s.d.

Spontaneous EPSCs (sEPSCs) are altered in the absence of TARPγ2. A, Sample current records from a het (black) and stg retina (red). Lighter traces at 0 pA show the derivative of the current records that were used to threshold and detect spontaneous events. Lower traces show detail for sample segments on an expanded time-base. B, Average sEPSCs generated from 580 events in five het SACs and 638 events in five stg SACs. The detection threshold was 5 SD (see Materials and Methods). Lower records show double exponential fits to the decay of the average sEPSCs. C, The amplitude of the fast component of the sEPSCs was unchanged. The amplitudes of the slow component for the stg SACs was much less variable than in het SACs, but the difference in means was not statistically significant. The fast decay component was significantly faster in stg SACs, but the slow decay time-constant was unchanged. p-values for an unpaired t test are shown above the parameters in the two panels (see also Extended Data Table 2-1). Error bars and shading are ±1 s.d.

Direction selectivity is largely unaffected by the absence of TARPγ2. Extracellular spikes were elicited by a dark bar moving in 12 directions through the receptive fields of ON-OFF direction-selective ganglion cells (DSGCs). The stimulus bar was 1 mm long, 200 μm wide and moved at 1 mm/s. The approximate dimension of the DSGC receptive field (blue circle) relative to the stimulus bar is shown in the schematic. A, Sample extracellular spike recordings from ON-OFF DSGCs in a het (black) and stg mutant retina (red) for preferred and null direction stimuli. Lower panels show peristimulus spike-time histograms (PSTHs) accumulated from 40 trials in each cell. B, Directional tuning for the OFF (leading edge) and ON (trailing edge) responses shown in A. Distance from the origin represents the peak of the respective PSTHs. Solid lines show fits to the von Mises function (see Materials and Methods). C, Distributions of the direction-selectivity index (DSI), here defined as the normalized length of the vector sum of the tuning function illustrated in B. D, Amplitude distributions for the tuning functions calculated from the von Mises fits to the data in B.

Directional tuning as a function of speed is unaffected by the absence of TARPγ2. PSTHs calculated for spikes elicited by narrow and wide bright bars moving through the receptive fields of ON-OFF direction-selective ganglion cells (DSGCs) in the preferred and null directions. The approximate dimension of the DSGC receptive field (blue circle) relative to the stimulus bars is shown in the schematic. The bars move, stop, and then disappear. A, PSTHs of extracellular spike recordings from ON-OFF DSGCs elicited by a narrow bar moving at a range of stimulus speeds (μm/s, shown in the legend) for both preferred (pref) and null (null) directions. A transient OFF-response is seen when the bar disappears at the end of the motion. B, DSIs, (Pref-Null)/(Pref+Null), as a function of stimulus speed for the data in A. C, D, Same format as A, B but for a wide-bar stimulus. Error bars and error shading indicate ±1 s.d.