Circadian clock disruption promotes retinal photoreceptor degeneration

Abstract

Daily rhythms are a central hallmark of vision, in particular by adapting retinal physiology and light response to the day-night cycle. These cyclic processes are regulated by retinal circadian clocks, molecular machineries regulating gene expression across the 24-h cycle. Although hundreds of genes associated with genetic retinal disorders have been identified, no direct link has been established with the clock. Hence, we investigated the hypothesis that a poorly functioning circadian clock aggravates retinal photoreceptor disease. We performed this study in the P23H rhodopsin-mutated mouse model (P23H Rho) that mimics one major cause of human autosomal dominant retinitis pigmentosa. We also used the rod-specific knockout (rod-Bmal1KO) of Bmal1, a key clock component. More specifically, we used either heterozygous P23H Rho mice or rod-Bmal1KO alone, as well as double mutants of these strains and control mice. We showed by structural (histology, immunohistochemistry) and functional (electroretinography: ERG) analyses that the retinitis pigmentosa phenotype is exacerbated in the double mutant line compared to the P23H Rho mutation alone. Indeed, we observed marked ERG amplitude reduction and more photoreceptor cell loss in double mutants with respect to simple P23H Rho mutants. These observations were further corroborated by transcriptome analysis revealing major gene expression differences between these genotypes. In this data, we identified unique gene expression sets implicating neurogenesis, phototransduction cascade, and metabolism, associated with enhanced photoreceptor degeneration. Thus, our results establish a link between clock dysfunction and retinal degeneration and suggest underlying molecular mechanisms, together providing new concepts for understanding and managing blinding diseases.

Keywords: Bmal1; ERG; P23H; RNA‐seq; circadian clock; retinitis pigmentosa; rod.

FIGURE 1

Dark‐adapted ERG analysis reveals synergistic effect between the simple P23H mutation of rhodopsin and the rod‐specific Bmal1 KO. Amplitudes of a‐ and b‐waves are presented according to intensity of the light stimulus in Ctrl (gray), rod‐Bmal1KO (green), P23H Rho (blue), and DM (red) groups at P40 (A, E), P80 (B, F), P112 (C, G) and P180 (D, H). There is a genotype effect (p‐value on the graphs) at all ages, on a‐ and b‐wave amplitudes (Mixed‐effects model analysis). (A–D) a‐wave in DM displayed a tendency for reduced amplitude with respect to littermate P23H Rho mice at all ages and was significantly lower at P40 (p < .05, A). (E–H) b‐wave amplitudes in DM mice appear also reduced with respect to P23H Rho. This decrease is significant at P80 (*p < .05, F) and P112 (**p < .01, G). No more difference is detectable between DM and P23H Rho mice at P180 (D, H), with nearly undetectable visual responses in both genotypes. Comparison between genotypes was performed by Mixed‐effects model analysis and Holm‐Sidak's post hoc testing (n = 6–10, n = 6–8, n = 5–8, and n = 5–6/genotype at P40, 80, 112, and 180, respectively). (I–L) Analysis of oscillatory potentials (OP) amplitudes (sum of amplitudes from the four OPs) shows a significant decrease in the light‐response signal originating from the inner retina in the DM already at P40 (I), progressively declining until P180 (I–L). Comparison between genotypes was performed by One way analysis of variance (n = 4 to 8). Data are means ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001.

FIGURE 2

Alteration of the photoreceptor layer in the DM retina at P45. (A) Representative images of retinal sections stained with H&E from DM mice and their distinct controls (Ctrl, rod‐Bmal1KO, and P23H Rho) at P45. No abnormalities were observed in Ctrl and rod‐Bmal1KO retinas. Loss of photoreceptor cells was visible in the P23H Rho retina sections and a more marked loss was observed in DM mice. OS, outer segments; IS, inner segments; ONL, outer nuclear layer; INL, inner nuclear layer; GCL, ganglion cell layer. (B) Morphometric quantification of histological retinal sections from Ctrl (gray), rod‐Bmal1KO (green), P23H Rho (blue), and DM (red) mice: Photoreceptor nuclei were counted within a 100 μm‐width rectangle displaced along the retina, on both sides of the optic nerve head. A significant reduction in number of photoreceptor nuclei in both P23H Rho and DM mice was observed, compared with either Ctrl or rod‐Bmal1KO retinas (Two‐way repeated measures ANOVA, Holm‐Sidak's post hoc test; p < .001). Data are presented as mean ± SEM; n = 3–4 mice/genotype. Scale bars: 50 μm.

FIGURE 3

DM retinas show enhanced degenerative process. Representative images of immunohistochemical staining from Ctrl (A, E, I, M), rod‐Bmal1KO (B, F, J, N), P23H Rho (C, G, K, O), and DM (D, H, L, P) P120 retina sections (n = 4 per genotype). Samples were stained for: Rhodopsin (A–D), cone arrestin (E–H), GFAP (I–L), PKCα (M–P). Rhodopsin is expressed in the photoreceptor outer segments in Ctrl (A) and rod‐Bmal1KO (B) mice. Rhodopsin expression is decreased in P23H Rho retinas (C) and mostly disappeared in DM retinas (D). The cone photoreceptor staining with cone arrestin was severely reduced in P23H Rho retina (G) and almost lost in DM retina (H), in which only few cells were labeled at synaptic terminals. P23H Rho and DM retinas showed activated Müller glial cells, detected by increased level of GFAP (K, L). The dendritic processes of bipolar cells (labeled with anti‐PKCα) also appeared affected in these genotype groups (O, P). DAPI staining is shown in blue. GCL, ganglion cell layer; INL, inner nuclear layer; ONL, outer nuclear layer; OS, outer segments. Scale bar: 50 μm.

FIGURE 4

Transcriptional profiling of P23H Rho single mutant and DM double mutant mouse retinas with respect to Ctrl at P120 shows extensive alteration of gene expression in DM. (A) Histogram shows differentially expressed genes in P23H Rho vs. Ctrl and DM vs. Ctrl (Adj. p < .05) mouse retinas (up‐regulated and down‐regulated; marked in red, and blue, respectively). (B, C) Functional annotation using gProfiler of differentially expressed genes (DEG) (up‐regulated in red, down‐regulated in blue) in P23H Rho vs. Ctrl (B) and DM vs. Ctrl (C). The 5 most significant biological pathways from the Gene Ontology Biological processes (BP) and (when relevant) WikiPathways (WP) databases are shown. Detailed lists of DEG in P23H Rho vs. Ctrl and DM vs. Ctrl with additional cut offs of fold change ≤ −2 or ≥2 are shown in Tables S5 and S6 respectively. Green line: Significance level cut off (Adj. p = .05).

FIGURE 5

Analysis of uniquely differentially expressed genes between DM and P23H Rho retinas. (A) Venn diagram showing the overlap between genes that are significantly downregulated or upregulated in DM retinas as compared to P23H Rho retinas and differentially expressed between P23H Rho and Ctrl retinas, in P120 mice (n = 4/genotype). Numbers of DEG specifically between DM and P23H Rho are highlighted (red, up; blue, down). (B) Most significant biological pathways from the Gene Ontology Biological processes (BP) and Cellular component (CC) categories for genes that are significantly up (116) and downregulated (262) in DM retinas vs. P23H Rho (red, up; blue, down) (detailed lists of pathways in Table S11). (C, D) STRING network analysis of functional associations (edges) of protein products (nodes) for gene lists highlighted in (A). The figure specifically highlights the protein network related to the Glycolysis/Neoglucogenesis KEGG pathway (blue nodes) among the 262 downregulated genes (C; interaction confidence score 0.7) and to the Neurogenesis BP (red nodes) among the 116 upregulated genes (D; interaction confidence score 0.3).