Control of late off-center cone bipolar cell differentiation and visual signaling by the homeobox gene Vsx1

Abstract

Retinal bipolar cells are interneurons that transmit visual signals from photoreceptors to ganglion cells. Although the visual pathways mediated by bipolar cells have been well characterized, the genes that regulate their development and function are largely unknown. To determine the role in bipolar cell development of the homeobox gene Vsx1, whose retinal expression is restricted to a major subset of differentiating and mature cone bipolar (CB) cells, we targeted the gene in mice. Bipolar cell fate was not altered in the absence of Vsx1 function, because the pan-bipolar markers Chx10 and Ret-B1 continued to be expressed in inner nuclear layer neurons labeled by the Vsx1-targeting reporter gene, tauLacZ. The specification, number, and gross morphology of the subset of on-center and off-center (OFF)-CB cells defined by tauLacZ expression from the Vsx1 locus were also normal in Vsx1(tauLacZ)/Vsx1(tauLacZ) mice. However, the terminal differentiation of OFF-CB cells in the retina of Vsx1(tauLacZ)/Vsx1(tauLacZ) mice was incomplete, as demonstrated by a substantial reduction in the expression of at least four markers (recoverin, NK3R, Neto1, and CaB5) for these interneurons. These molecular abnormalities were associated with defects in retinal function and documented by electroretinography and in vitro ganglion cell recordings specific to cone visual signaling. In particular, there was a general reduction in the light-mediated activity of OFF, but not on-center, ganglion cells. Thus, Vsx1 is required for the late differentiation and function of OFF-CB cells and is associated with a heritable OFF visual pathway-specific retinal defect.

Figures

Fig. 1.

CB-like cells are specified in the Vsx1-null retina. Representative confocal micrographs of β-gal staining on frozen retinal sections of 1-yr-old littermates: Vsx1+/+ (A), Vsx1τLacZ/Vsx1τLacZ (B), and Vsx1τLacZ/+ (C). PR, photoreceptors; INL, inner nuclear layer; GC, ganglion cells. β-Gal staining in the IPL of the Vsx1τLacZ/Vsx1τLacZ retina was typical of the pattern normally seen for OFF (B, arrow) and ON (B, arrowhead) CB cells. In 5-mo-old Vsx1τLacZ/+ mice, β-gal-positive cells with both ON-CB morphology (D, arrowhead) and OFF-CB morphology (D–G, arrow) were present (in D, a and b indicate sublaminar regions bounded by dashed lines). Two types of OFF-CB types were observed (D–G): one type with a slightly longer axon (solid lines indicate axon length in E vs. F) and a less elaborate arbor (D and E, arrow) similar to the B2 OFF-CB cell (6), and a second type (F and G, arrow) with a shorter axon and a more elaborate arbor, similar to the B1 OFF-CB cell previously described in the mouse retina (6). β-Gal and Vsx1 staining colocalized in the Vsx1τLacZ/+ retina (H–J, arrows), although not all Vsx1-positive cells were colabeled with β-gal (H–J, arrowheads). The number of Chx10-positive cells (expressed as a percentage of 4′,6-diamidino-2-phenylindole-stained nuclei in the INL) was not altered in the Vsx1 heterozygotes (+/-)or Vsx1 homozygotes (-/-) (L); three sets of littermates (5–12 mo old) were examined (two with the τLacZ allele and one with the Δ1–4 allele). (M) The number of β-gal-positive cells in the mature (5–12 mo old) Vsx1τLacZ/+ retina (+/-, yellow bar) was significantly different (indicated by*, Student's t test, P < 0.05) from the number of Vsx1-positive cells (green bars) in the Vsx1 wild-type (+/+) and Vsx1τLacZ/+ (+/-) retina and from the number of β-gal-positive cells in the Vsx-τLacZ null (-/-) retina (yellow bar); the number above the bars is the total of Chx10-positive cells in the INL examined. β-Gal- and Vsx1-positive cells in M are expressed as a percentage of Chx10-positive cells. Each bar in M represents data from three mice except for the last bar (-/-), which is derived from four Vsx1τLacZ/Vsx1τLacZ mice and one Vsx1Δ1–4/Vsx1τLacZ mouse (the Vsx1Δ1–4/Vsx1τLacZ value is indicated on the bar by an arrowhead). [Bar = 85 μm(A–C); 70 μm(D–G); 61 μm(H–J); 75 μm(K).]

Fig. 2.

Cells lacking Vsx1 function express pan-bipolar cell markers. In the Vsx1τLacZ/Vsx1τLacZ retina of 5-mo-old mice, β-gal-labeled cells (green) colocalized with a subset of cells (red) expressing two pan-bipolar markers, Chx10 (A) and Ret-B1 (B) (Chx10 and Ret-B1 colocalized with β-gal in the cell body region indicated by arrowheads in A and B). β-Gal staining did not colocalize with the RB marker PKCα in the Vsx1τLacZ/Vsx1τLacZ retina (C, arrowheads indicate two of the β-gal-positive cells). (Bar = 50 μm.)

Fig. 3.

Vsx1 is essential for late CB differentiation. Confocal micrographs of retinal expression of CB markers in 1-yr-(A–D) and 5-mo-old (E–L) littermates. Recoverin expression in OFF-CB cells is absent from the Vsx1τLacZ/Vsx1τLacZ retina (B) compared to the wild-type Vsx1+/+ retina (A), whereas photoreceptor recoverin staining (A and B, staining at top) is unchanged. Arrow in B indicates a single weakly labeled recoverin-positive cell. NK3R CB staining is greatly reduced in the Vsx1τLacZ/Vsx1τLacZ retina (D) compared to the Vsx1+/+ retina (C). In the Vsx1+/+ retina, punctate Neto1 staining is observed in both the outer plexiform layer (E, arrowheads) and IPL (E, between dotted lines) of the Vsx1+/+ retina. In contrast to the Vsx1+/+ retina, Neto1 immunofluorescence is greatly reduced in both the outer plexiform layer of the Vsx1τLacZ/Vsx1τLacZ retina (F, arrow indicates a single patch of Neto1-positive staining in this layer) and the IPL of the Vsx1τLacZ/Vsx1τLacZ retina (F, the bracketed region indicates an isolated region of faint Neto1 staining). CaB5 staining (G–J) was moderately diminished in the Vsx1Δ1–4/Vsx1Δ1–4 retina in the OFF-CB axon terminal region (G–J, bracketed region) but was normal in the ON-CB axon terminal region (H and J, arrowhead) and RB axonal terminal regions (H and J, arrow). The area between the ON-CB axonal terminal region and the RB axonal terminal region that is normally free of CaB5 staining was absent in the Vsx1Δ1–4/Vsx1Δ1–4 retina (H and J, compare region between dashed lines on left). PMCA1 staining (K and L) was not diminished in the Vsx1Δ1–4/Vsx1Δ1–4 retina in the OFF-CB axon terminal region (K and L, arrow) or in the ON-CB axon terminal region (K and L, arrowhead). Frozen sections, A–D; paraffin sections, E–L. [Bar = 50 μm(A–D, K, and L); 40 μm(E and F); 95 μm(G and I); 34 μm(H and J).

Fig. 4.

b-wave defects in the electroretinograms of Vsx1 mutants. (A) Representative rod full-field ERGs from a Vsx1-null mouse (solid lines). Shown are responses from a Vsx1-null mouse to a series of retinal illuminances spanning over 7 log units at ≈0.6 log unit intervals. Responses to high-energy flashes used for a-wave modeling are shown as broken curves. (Inset) Expanded view of a-waves to high-energy flashes. Solid lines show Vsx1-null responses, and dashed lines show best fit of the Lamb and Pugh model (37). (B) Relative loss of b-wave response in mixed rod-and-cone-mediated response in Vsx1-null mice. Each trace was derived from six mice, except for Vsx1+/+, which was based on five. Responses to 20-ms white flashes were normalized so that a-wave amplitudes are comparable. The double-headed arrow between the lines indicates the amplitude difference between Vsx1+/+ (upper line) and Vsx1-/- mice (lower line). (C) Rod a/b-wave amplitudes in Vsx1+/- and Vsx1-/- mice were comparable to wild-type mice. Cone a-waves are too small to measure in mice, but cone b-wave amplitudes were not significantly different among Vsx1+/+, Vsx1+/-, and Vsx1-/- animals. (D)b/a-wave amplitude ratios were comparable for the rod response in Vsx1+/+, Vsx1+/-, and Vsx1-/- mice. The b/a-wave amplitude ratio of the mixed rod-and-cone-mediated response was significantly smaller in Vsx1-/- mice. Consistent with a Vsx1-gene dosage effect, b/a-wave amplitude ratios from Vsx1+/- mice were observed to lie between Vsx1+/+ and Vsx-/- values. The differences among Vsx1+/+ (n = 6 mice), Vsx1+/- (n = 6; combined results from three Vsx1τLacZ/Vsx1τLacZ and three Vsx1Δ1–4/Vsx1Δ1–4 mice), and Vsx1-/- groups (n = 6; combined results from three Vsx1τLacZ/Vsx1τLacZ and three Vsx1Δ1–4/Vsx1Δ1–4 mice) were significant [ANOVA, F (2, 15) = 5.07; 0 = 0.02].

Fig. 5.

OFF ganglion cell response defects in the Vsx1-null retina. (A) Histogram showing response thresholds of high- and low-sensitivity ganglion cells in the Vsx1-null retina, which provide an index of rod-and-cone photoreceptor thresholds, respectively. The thresholds in the Vsx1-nulls (asterisks) matched those reported previously in the wild-type mouse (20). (B) Representative extracellular recordings from ON and OFF ganglion cells in the Vsx1-/- and Vsx1+/+ mouse retinas to three stimulus intensities. Because cross referencing of different ganglion cell types was not performed in this study, a direct comparison of the responses between Vsx-/- and Vsx1+/+ ganglion cells cannot be made. The OFF ganglion cell response to higher light intensities showed reduced spike frequency in the Vsx1-null mutant compared to wild-type, whereas the responses of ON ganglion cells in the wild-type and Vsx1-/- retinas were similar. Bars at bottom of trace indicate onset and offset of the light stimulus. (Bars = 100 μV/500 ms.) (C) Scatter plot showing the average light-evoked spike count to full-field light stimulation over a 6-log unit range. There was no difference between the responses of ON cells in the Vsx1-/- and Vsx1+/+ mice; t test, P > 0.1 for all light intensities. (D) Scatter plot comparing the light-evoked spike count of OFF ganglion cells in Vsx1-/- and Vsx1+/+ mice. There was no difference between wild-type and Vsx1-null mice in the responses of the ganglion cells to low (scotopic) stimulus intensities, but the responses of ganglion cells in Vsx1-null mice were markedly reduced to stimulus intensities above the cone threshold; t test, P < 0.05 for intensities above 100 Rh*/rod per second (Rh*/rod, time-average rate of photoisomerizations per rod). Data points indicate averages and standard errors of 48 ON/37 OFF ganglion cell recordings from Vsx1-null retinas and 48 ON/24 OFF ganglion cell recordings from wild-type retinas.

Fig. 6.

A model of the role of Vsx1 function in retinal bipolar interneuron development. We have established an essential role for Vsx1 function in the late differentiation of OFF-CB cells. Analysis of reporter β-gal staining in the Vsx1τLacZ/Vsx1τLacZ retina indicated that the number, specification, and gross morphology of OFF-CB cells is unaffected by the absence of Vsx1 function. Early roles in pan-bipolar specification and differentiation have previously been established for the Vsx1 homologue Chx10 (14, 15) and the basic helix–loop–helix transcription factor genes Mash1 and Math3 (16). The dashed line and question mark indicate it is unclear whether bipolar cells are first specified as a homogeneous group, which is then further specified into the various bipolar types, or whether each bipolar type is specified independently and directly from a retinal progenitor. Asterisks indicate that multiple OFF- and ON-CB cell types are present: at least three OFF-CB types and one ON type have been identified in mouse (6, 24).