A combination treatment based on drug repurposing demonstrates mutation-agnostic efficacy in pre-clinical retinopathy models

Abstract

Inherited retinopathies are devastating diseases that in most cases lack treatment options. Disease-modifying therapies that mitigate pathophysiology regardless of the underlying genetic lesion are desirable due to the diversity of mutations found in such diseases. We tested a systems pharmacology-based strategy that suppresses intracellular cAMP and Ca2+ activity via G protein-coupled receptor (GPCR) modulation using tamsulosin, metoprolol, and bromocriptine coadministration. The treatment improves cone photoreceptor function and slows degeneration in Pde6βrd10 and RhoP23H/WT retinitis pigmentosa mice. Cone degeneration is modestly mitigated after a 7-month-long drug infusion in PDE6A-/- dogs. The treatment also improves rod pathway function in an Rpe65-/- mouse model of Leber congenital amaurosis but does not protect from cone degeneration. RNA-sequencing analyses indicate improved metabolic function in drug-treated Rpe65-/- and rd10 mice. Our data show that catecholaminergic GPCR drug combinations that modify second messenger levels via multiple receptor actions provide a potential disease-modifying therapy against retinal degeneration.

Fig. 1. Dietary TMB self-administration stabilizes the retinal transcriptome and mitigates the disease phenotype in rd10 mice.

A Study design. B Representative examples of drug doses via dietary tamsulosin, metoprolol, and bromocriptine (TMB) coadministration, obtained from the blood and the target tissues at 12:00 a.m. The blood level of bromocriptine did not reach the sensitivity limit of the assay, so it is labeled as “not detected” (n.d.). C Representative optical coherence tomography (OCT) images. The plus sign (+) indicates outer nuclear layer (ONL). ONH, optic nerve head. D ONL thickness and retinal detachment, as measured from OCT images. E, F Group-averaged electroretinogram (ERG) waveforms in response to green E and UV F stimuli under photopic conditions in a representative cohort. G, H M-cone- G and S(UV)-cone-dominant H ERG b-wave amplitudes. I Venn diagrams of upregulated and downregulated genes in retinal bulk RNA-sequencing. J, K Transcriptome heatmap of 5000 most-abundant nuclear genes J, and 10 most-abundant mitochondrial-encoded genes K, that show an expression change compared to WT in dark-reared (DR) rd10 mice (at P28, non-treated), or experimental rd10 mice (P37, changed from dark to vivarium/cyclic light-rearing [CLR] at P29) that were treated with vehicle (veh) or TMB. The asterisks signify statistical differences (adjusted P < 0.05) compared to WT expression level. L Heatmap of regulation in a set of genes associated with retinitis pigmentosa (RP). M Heatmap of regulation in genes encoding adrenergic and dopamine receptors, and major catecholamine synthetizing or degrading enzymes. The statistical analysis employed for graphs in D was the Mann–Whitney U-test (two-tailed), and for graphs G and H the Kruskal–Wallis test was followed by Dunn´s multiple comparisons test; The asterisks signify: ***P < 0.001, ****P < 0.0001. Bar graph data are presented as mean ± SD. Source data are provided as a Source Data file.

Fig. 2. Chronic TMB administration improves visual function and cone viability in rd10 mice.

Study design is depicted in Fig. 1A. This figure presents data after a two-week A–D or one-month-long E–M treatment period. A Group-averaged visual evoked potential (VEP) responses (WT, n = 7 mice, n = 21 electrodes; rd10-vehicle, n = 10 mice, n = 30 electrodes; rd10-TMB, n = 10 mice, n = 31 electrodes). The traces represent group´s mean responses. B Normalized VEP amplitudes and first negative component latencies from response to light onset. C Group-mean neural spiking rate from single V1 neurons in response to ON-OFF light stimulus. The total number of single cells recorded in WT mice was 217, in rd10-vehicle 638, and in rd10-TMB mice 642 cells. D Spiking during ON response peak (left), and background spiking during sustained light OFF (right). E Representative partial retinal whole mount, stained for cone opsins, of a dark-reared P30 rd10 mouse. Image is zoomed and centered at the optic nerve head (ONH). F–G Comparison of ONH-centered zoomed images of vehicle-treated (F) and TMB-treated (G) experimental rd10 mice. The circular cone counting windows (yellow, 600 µm diameter) were centered at 500 µm from the ONH. H Superior central M-cone and (I) inferior central S(UV)-cone counts. J M-cone and (K) S(UV)-cone ERG b-wave amplitudes. L Western blots of alpha-tubulin (TUBA1B, loading control), cone arrestin (ARR3), and rhodopsin (RHO). The graph shows expression differences between the vehicle and TMB groups. M Optomotor responses as a function of changing spatial-frequency stimuli. The insert shows extrapolated visual acuity from the OMR index versus spatial frequency graphs. The statistical analysis employed for graphs I, and K was the Mann–Whitney U-test (two-tailed); for graphs (B) and (D) the Kruskal-Wallis test was followed by Dunn´s multiple comparisons test; for graph M (insert) one-way ANOVA analysis was followed by Bonferroni posthoc tests; and for graphs (H, J) and (L), Welch´s t-test (two-tailed) was used. The asterisks signify: *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Bar graph data are presented as mean ± SD. Source data are provided as a Source Data file.

Fig. 3. Retinal protection in dark-reared rd10 mice is associated with decreased lipid peroxidation.

A Study design. Note, graphs B and E-K represent data for mice after one month on treatments; graphs L–Q represent data for mice after three months on treatments. B Group-averaged scotopic ERG waveforms in a representative cohort. C, D Scotopic ERG a-wave C and b-wave D amplitudes at different stages of disease progression in the same mice (repeated measures design). E, F Representative OCT images. G ONL-thickness assessment. H Whole retinal 4-HNE contents, as measured by ELISA (WT, n = 8; rd10-vehicle, n = 9; rd10-TMB, n = 10). I–J Quantification of expression of catalase (CAT) and GFAP from immunoblots (WT, n = 5; rd10-vehicle, n = 12; rd10-TMB, n = 9). K Representative immunoblots from cohort 1. A total of two cohorts (individual experiments) were performed and the data pooled for the analyzes shown in panels (I, J). L, M Representative retinal flat mounts stained with antibodies against M-cone opsin or S(UV)-cone opsin. N, O M-cone and S(UV)-cone counts from retinal flat mounts. P, Q M-cone- and S(UV)-cone-dominant photopic ERG amplitudes. The three-month-long experiments were performed in female rd10 mice. Welch´s t-test (two-tailed) was used to analyze data in G, P and Q. ERG data in (C, D) (age as within-subjects and treatment as between-subjects factor) and cone count data in (N, O) (retinal location as within-subjects and treatment as between-subjects factor) were analyzed by two-way RM ANOVA with the Geisser-Greenhouse correction. The statistical analysis performed for H was one-way ANOVA and Bonferroni´s post hoc test, whereas I, J were analyzed by Welch´s ANOVA followed by Dunnett´s T3 tests. The pound signs indicate an ANOVA-significant main effect between treatments: ^#P < 0.05, ^##P < 0.01^. The asterisks mark significant post hoc test effects: *P < 0.05, **P < 0.01, ***P < 0.001, **** P < 0.0001. Data are presented as mean ± SD. Source data are provided as a Source Data file.

Fig. 4. Single-cell RNA sequencing reveals a decreased cell stress- and an improved metabolic status in Müller glia by chronic TMB therapy in dark-reared female rd10 mice.

A Group-consolidated UMAPs by cluster. The insert shows cluster size breakout as a pie chart. Single-cell suspensions were prepared from three rd10 mice per group, whereas the WT group had four mice. Analysis was performed from 4478 cells from WT mice, 5952 cells from vehicle-treated rd10, and 4812 cells from TMB-treated rd10 female mice. B Table showing the number of differentially expressed (DE) genes for all clusters and for all groupwise comparisons. The cutoff for DE genes was set at P < 0.05 after adjustment for multiple comparisons. C, D Gene interaction network (generated by String database version 12.0) of downregulated (C) and upregulated (D) DE genes in vehicle-treated rd10 versus TMB-treated rd10 mice in the Müller cell(0) cluster. E, F DE genes in vehicle-treated rd10 versus TMB-treated rd10 mice in the ON-CBC cluster 3 (E) and cone cluster 4 (F), shown as heatmaps where regulation is compared to WT-gene expression. AC amacrine cell, HC horizontal cell, ON-CBC ON cone bipolar cell, OFF-CBC, OFF cone bipolar cell, RBC rod bipolar cell, RGC retinal ganglion cell, RPE retinal pigment epithelium.

Fig. 5. Monotherapies with tamsulosin, metoprolol, or bromocriptine are not effective in delaying retinal degeneration in rd10 mice.

Rd10 mice were dark-reared between P0-P28, and treatments were started at P28. One day later (P29), mice were transferred to normal laboratory housing conditions, and treatment effect was tested at P36. The drug effect was evaluated using four different parameters: photopic M- or S(UV)-cone targeted ERG b-wave amplitudes, ONL thickness, or retinal detachment from the RPE. A–D Tamsulosin (T) monotherapy data. E–H Metoprolol (M) monotherapy data. I–L Bromocriptine (B) monotherapy data. M–P Tamsulosin/bromocriptine (TB) dual-treatment data contrasted with TMB triple-treatment. Q–T Tamsulosin and metoprolol (TM) dual-treatment data contrasted with TMB triple-treatment. U–X Metoprolol and bromocriptine (MB) dual-treatment data contrasted with TMB triple-treatment. The ONL thickness and retinal detachment data were statistically analyzed by the non-parametric Mann–Whitney U-test (two comparisons; two-tailed) or the Kruskal-Wallis test (three comparisons) followed by Dunn´s multiple comparisons tests. All ERG data were analyzed using repeated measures two-way ANOVA with Geisser-Greenhouse correction and followed by Bonferroni post hoc tests. Asterisks illustrate significant Mann-Whitney or Bonferroni test results: *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. The pound signs illustrate significant between-subjects ANOVA main effects: ^#P < 0.05, ^##P < 0.01, ^###P < 0.001, ^####P < 0.0001. Data are presented as mean ± SD. Source data are provided as a Source Data file.

Fig. 6. Dietary TMB administration slows retinal degeneration in female Rho^P23H/WT mice.

A Study design. B–C Representative OCT images at 3-months of age. D Representative group-averaged (thick lines) scotopic ERG waveforms (log cd·s/m² = 1.7). Thin traces represent individual mouse responses. E–F Scotopic ERG a-wave E and b-wave F amplitudes. G Longitudinal ONL thickness follow-up throughout the study in one treatment cohort (repeated measures study design). Data analyzed from OCT images. H Representative group-averaged (thick lines) photopic ERG waveforms at 6-months of age. Stimulus was elicited with a green LED that stimulates primarily the M cones (log cd·s/m² = 2.5). I M-cone- J and S(UV)-cone-dominant photopic ERG b-wave-amplitudes. A repeated cross-sectional study. K, L Optic nerve head (ONH)-centered (* shows ONH) central retina images showing M-cone K and S(UV)-cone L populations in a vehicle-treated retina. Dashed-line yellow rectangles indicate 0.75 mm sampling site location. M, N ONH-centered central retina images showing M-cone M and S(UV)-cone N populations in a TMB-treated retina. O–P Representative M-cone O and P UV/S-cone images in superior and inferior retina, respectively, at different distances from the ONH. Q–R Inferior retina S(UV)-cone count analysis Q and superior retina M-cone count analysis R. Counting windows sizes were: width 360 µm, height 270 µm. S, T Representative OCT images. Statistical analysis performed for data in graphs E, and F was by Welch´s t-test (two-tailed). Data in graphs G, Q and R were analyzed by two-way RM ANOVA with the Geisser-Greenhouse correction. Data in graphs I and J were analyzed by ordinary two-way ANOVA. ANOVAs were followed by Bonferroni posthoc tests. T-test and Bonferroni post hoc results: *P < 0.05, **P < 0.01. ANOVA between-subjects main effects: ^##P < 0.01, ^####P < 0.0001. Data are presented as mean ± SD. Source data are provided as a Source Data file.

Fig. 7. TMB administration stabilizes retinal transcriptome and improves visual function in Rpe65^−/− mice.

A Study design. B Group-averaged scotopic ERGs. Thin lines represent responses from individual mice. C Scotopic ERG a- and b-wave amplitudes. D–I Ex vivo ERG. The perfusion medium was supplemented with synaptic blockers (vehicle, n = 6; TMB, n = 7 mice), or not (vehicle, n = 6; TMB, n = 6 mice). H–I Averaged intensity-response functions for ERG a-wave H or b-wave I recorded in the presence or absence of synaptic blockers, respectively. J Optomotor responses at different pattern contrasts. K Outer retina thickness as measured from OCT images. L Whole dorsal retina M-cone counts from retinal whole mounts. M Venn diagram of differentially expressed genes in retinal bulk RNA-seq. Comparison shown between WT, Rpe65^-/--vehicle, and Rpe65^-/--TMB groups. N Expression heatmap of 10 most highly expressing mitochondrial-encoded genes. The asterisks in heatmaps indicate significant difference (adjusted for multiple comparisons) compared to expression in WT mice. O Expression heatmap of 5000 most highly expressing nuclear-encoded genes. For clarity of presentation outlier genes Gm4735 and Eno1b were removed. P, Q Volcano plots of DE genes between WT and vehicle-treated Rpe65^-/- mouse MG cells P, or WT and TMB-treated Rpe65^-/- mouse MG cells Q, from scRNA-seq data. Each Rpe65^-/- group data was derived from cell suspensions combining seven retinas, whereas the WT group consisted of three retinas. Analysis was performed from 4,135 cells of the WT mice; 3979 cells from vehicle-treated; and 5,901 cells from TMB-treated Rpe65^−/− mice. R DE gene comparison of Rpe65^-/--vehicle and Rpe65^-/--TMB groups in MG cells. Selected enriched KEGG pathways are shown. Data in graphs C, H, I and J were analyzed using RM two-way ANOVA with the Geisser-Greenhouse correction, whereas data in graph K were analyzed by 1-way ANOVA followed by the Bonferroni post hoc test: ****P < 0.0001. Data in graph L were analyzed by Welch´s t-test. The pound signs indicate an ANOVA-significant main effect between treatments: ^#P < 0.05, ^####P < 0.0001. Data are presented as mean ± SD. Source data are provided as a Source Data file.

Fig. 8. Tamsulosin, metoprolol, or bromocriptine monotherapies are not effective in improving retinal function in Rpe65^−/− mice.

Treatments were started at P21, and treatment effect was tested after 3–4 weeks of drug therapy. The drug effect was evaluated using two different methods: Scotopic ERG b-wave and optomotor response. The same vehicle-group data are used for comparison with each of the data sets for all treatments. Data in M is duplicated from Fig. 6J. A, B Tamsulosin (T) monotherapy data. C, D Metoprolol (M) monotherapy data. E, F Bromocriptine (B) monotherapy data. G, H Tamsulosin and bromocriptine (TB) dual-treatment data. I, J Tamsulosin and metoprolol (TM) dual-treatment data. K, L Metoprolol and bromocriptine (MB) dual-treatment data. M, N Tamsulosin, metoprolol, and bromocriptine (TMB) triple-treatment data. Statistical analysis for all data was performed using repeated measures two-ANOVA with the Geisser-Greenhouse correction followed by Bonferroni posthoc tests: *P < 0.05, **P < 0.01, ****P < 0.0001. Pound signs denote ANOVA between subject main effect significance levels: ^##P < 0.01, ^###P < 0.001, ^####P < 0.0001. Data are presented as mean ± SD. Source data are provided as a Source Data file.

Fig. 9. Subcutaneous TMB infusion improves cone viability in PDE6A^−/− dogs.

A Study design. B Drug serum levels during the first 7 weeks of study in three dogs. Dots and lines represent data from individual dogs. Gray dashed line shows linear regression. Outliers were removed using the ROUT method with FDR set at 0.5%. C Drug serum level follow-up throughout the study in subject #3. D Group-averaged photopic ERG waveforms of each monthly recording. The ERGs are contrasted to PDE6A^+/--carrier dogs without RP phenotype. Black, vehicle n = 10; red, TMB n = 9; gray, unaffected ctrl (PDE6A^+/-, n = 3). E, F Photopic ERG b-wave peak time E and amplitude F follow-up. G Representative dog retina whole mounts stained with a cone-marker peanut agglutinin (PNA). H, I PNA puncta counting windows (width 240 µm, height 180 µm) in the superior middle retina H, upper row controls and lower row TMB) and I in the inferior middle retina in litters 6-8 (note, these cohorts were vehicle-controlled). J PNA puncta count. Semi-automated counting was performed by a blinded observer. All litters 1-8 were included in the analysis. Due to distinctly differential degeneration in inferior versus superior retina, PNA puncta count in inferior and superior retinal sides were analyzed separately. Vehicle, n = 10; TMB, n = 9. K Representative area centralis (AC) images that were used as PNA puncta counting windows (150 µm, height 110 µm). Images from dogs in litters 6–8. L PNA puncta count in AC. Counting in this region was performed manually by a blinded observer. Litters 1–8 were included in the Mann-Whitney U-test (two-tailed) analysis (ctrl, n = 9; TMB, n = 9; n.s.). Statistical analysis in E, F and J was performed by two-way repeated measures (E, F, J at superior retina) or mixed-model (J at inferior retina) ANOVA with Geisser-Greenhouse correction and followed by Bonferroni post hoc tests. The pound signs signify a significant between-subjects main effect: ^#P < 0.05. The asterisk * signify P < 0.05 in Bonferroni test. Data are presented as mean ± SD. Source data are provided as a Source Data file.