Diabetic retinopathy is a ceramidopathy reversible by anti-ceramide immunotherapy

Abstract

Diabetic retinopathy is a microvascular disease that causes blindness. Using acid sphingomyelinase knockout mice, we reported that ceramide generation is critical for diabetic retinopathy development. Here, in patients with proliferative diabetic retinopathy, we identify vitreous ceramide imbalance with pathologic long-chain C16-ceramides increasing and protective very long-chain C26-ceramides decreasing. C16-ceramides generate pro-inflammatory/pro-apoptotic ceramide-rich platforms on endothelial surfaces. To geo-localize ceramide-rich platforms, we invented a three-dimensional confocal assay and showed that retinopathy-producing cytokines TNFα and IL-1β induce ceramide-rich platform formation on retinal endothelial cells within seconds, with volumes increasing 2-logs, yielding apoptotic death. Anti-ceramide antibodies abolish these events. Furthermore, intravitreal and systemic anti-ceramide antibodies protect from diabetic retinopathy in standardized rodent ischemia reperfusion and streptozotocin models. These data support (1) retinal endothelial ceramide as a diabetic retinopathy treatment target, (2) early-stage therapy of non-proliferative diabetic retinopathy to prevent progression, and (3) systemic diabetic retinopathy treatment; and they characterize diabetic retinopathy as a "ceramidopathy" reversible by anti-ceramide immunotherapy.

Keywords: IL-1β; TNFα; apoptosis; ceramide; diabetes; diabetic retinopathy; endothelial cell; inflammation; retina; sphingomyelinase.

Figure 1:. Ceramide levels are tightly-regulated via four interconnected, yet distinct metabolic pathways.

Abbreviations: ASM, acid sphingomyelinase; CerS, ceramide synthase; SphK, sphingosine kinase; S1P lyase, sphingosine 1-phosphate lyase; SPT, serine palmitoyl transferase; KDSR, 3-ketodihydrosphingosine reductase; DES1, dihydroceramide reductase 1; GCS, glucosylceramide synthase; ASAH1/2, acid ceramidase.

Figure 2:. Vitreous samples of patients with PDR show imbalances in long vs. very long chain ceramides, but not in total ceramide levels compared to patients without diabetes.

Ceramide quantification was performed using nano-electrospray high resolution mass spectrometry. Vitreous samples were obtained from patients diagnosed with PDR or non-diabetic controls diagnosed with macula hole. The total level of intravitreal ceramide was unchanged between groups (inset) whereas intravitreal ceramide species were altered between cohorts as follows: d18:1/16:0-, d18:1/C24:0- and d18:1/C24:1-ceramides were increased, while d18:1/C26:0-ceramides were reduced in patients with PDR. Data represent mean ± SD, n = 7 PDR and non-diabetic patient samples, respectively. Unpaired T-test compared patients with PDR with non-diabetic controls.

Figure 3:. TNFα induces time- and dose-dependent CRP formation in BRECs.

(A) Representative 2D images of TNFα (20 ng/ml)-induced CRP formation over time (0–120 seconds). The cell nucleus is stained with DAPI (blue), whereas ceramide is stained with mouse anti-ceramide MID 15B4 (1:50 dilution) followed by Cy3-conjugated anti-mouse IgM (red stained signal). (B) Quantitation of 2D images in (A) shows the dose-dependent incidence of TNFα-induced CRP formation over time (0–300 sec). (C) Representative 3D-stills of TNFα (20 ng/mL)-induced CRP formation over time. Merged images each show a single cell displaying a cell surface CRP (red) and the cell nucleus (blue). (D) CRP volumes (μm³) were quantified from “movies” of the time-dependence of TNFα (20 ng/mL)-induced CRP formation. Data mean ± SEM are derived from n=50 individual cells per time point. **p<0.01, ***p<0.001 from 1 of 2 independent experiments. (E) 2D image of anti-ceramide 6B5 scFv (10 μg/mL) inhibition of TNFα (20 ng/mL)-induced CRP formation at 45 seconds, the peak of CRP formation. (F) Quantitation of the data from E. (G) 3D still images of TNFα-induced CRP formation at 45 seconds of 20 ng/mL TNFα treatment with and without anti-ceramide scFv, performed as in C. (H) Quantitation of 3D data (mean ± SEM, n=50 individual cells) derived from “movies” from 1 of 2 independent experiments performed as in G. **p<0.01, ***p<0.001. (I) Dose- and time-dependence of TNFα-induced apoptosis. Data represent best fit of 1 of 3 independent experiments. (J) 6B5 anti-ceramide scFv inhibits TNFα (20 ng/mL)-induced apoptosis quantified by bisbenzimide staining. Data (mean ± SEM) are collated from 3 experiments. **p<0.01 vs. control.

Figure 4:. IL-1β induces time and dose-dependent CRP formation in BRECs.

(A) Representative 2D images of IL-1β (20 ng/ml)-induced CRP formation over time (0–120 seconds). Studies were performed as in Figure 2A. The cell nucleus is stained with DAPI (blue), whereas ceramide is stained with mouse anti-ceramide MID 15B4 (1:50 dilution) followed by Cy3-conjugated anti-mouse IgM (red stained signal). (B) Quantitation of 2D images in (A) shows the dose-dependent incidence of IL-1β-induced CRP formation over time (0–300 sec).(C) Representative 3D-stills of IL-1β (20 ng/mL)-induced CRP formation over time. Merged images each show a single cell displaying a cell surface CRP (red) and the cell nucleus (blue). (D) CRP volumes (μm³) were quantified from “movies” of the time-dependence of IL-1β (20 ng/mL)-induced CRP formation. Data (mean ± SEM) are derived from n=50 individual cells per time point. **p<0.01, ***p<0.001 from 1 of 2 independent experiments. (E) 2D image of anti-ceramide 6B5 scFv (10 μg/mL) inhibition of IL-1β (20 ng/mL)-induced CRP formation at 45 seconds, the peak of CRP formation. (F) Quantitation of the data from E. (G) 3D still images of IL-1β-induced CRP formation at 45 seconds of 20 ng/mL IL-1β treatment with and without anti-ceramide scFv, performed as in C. (H) Quantitation of 3D data (mean ± SEM, n=50 individual cells) derived from “movies” from 1 of 2 independent experiments performed as in G. **p<0.01, ***p<0.001. (I) Dose- and time-dependence of IL-1β-induced apoptosis. Data represent best fit of 1 of 3 independent experiments. (J) 6B5 anti-ceramide scFv inhibits IL-1β (20 ng/mL)-induced apoptosis quantified by bisbenzimide staining. Data (mean ± SEM) are collated from 3 experiments. **p<0.01 vs. control.

Figure 5:. Intravitreal anti-ceramide 6B5 scFv and 2A2 mAb alleviate inflammation and DR progression in a murine I/R model.

A single intravitreal injection of anti-ceramide 6B5 scFv or 2A2 mAb in mice protects against retinal inflammation [Tnfα, Il-1β, Il-6] and upregulation of cell adhesion molecules ICam-1 after I/R surgery measured by increased mRNA expression levels (A). Anti-ceramide scFv and mAb protect against DR progression in an I/R model assessed by vascular permeability. Respective images of retinal vasculature and leakage are shown for Control, I/R and I/R + scFv, and quantified in (B). Data represent mean ± SEM derived from n = 4–12 (Tnfα), n = 4–11 (Il-1β), n = 3–11 (Il-6), n = 4–11 (ICam-1) animals per group for retinal inflammation and n = 3–6 animals per group for vascular permeability collated from two independent experiments. Paired T-tests are compared with respective controls in A; Unpaired T-Test is used in B. Scale bars = 200 μm.

Figure 6:. Intravitreal anti-ceramide 6B5 scFv alleviates inflammation and DR progression in hyperglycemic rats.

Bodyweight and non-fasted blood glucose levels pre- and post-treatment with anti-ceramide 6B5 scFv (A). A single intravitreal injection of anti-ceramide scFv protects from chronic retinal inflammation in a hyperglycemic rat model assessed as retinal Tnfα mRNA expression levels (B) Anti-ceramide scFv protects from DR progression in a hyperglycemic rat model assessed as vascular permeability after 6 weeks of hyperglycemia (C) Respective images of retinal vasculature and leakage are shown for Control, STZ and STZ + scFv. Data represent mean ± SEM derived from n = 10–12 and n = 12–18 animals per group collated from two independent experiments. Statistics: Statistics were not shown to increase visibility (A); One-Way ANOVA followed by Dunnett’s multiple comparison test compared to hyperglycemic control (B, C). Scale bars = 100 μm.

Figure 7:. Systemic anti-ceramide 2A2 mAb treatment alleviates inflammation and vascular permeability in a murine I/R model.

A single subcutaneous injection of anti-ceramide 2A2 mAb protects from chronic retinal inflammation in a murine I/R model assessed as retinal Il-1β and MCP-1 (A) mRNA expression levels. Systemic anti-ceramide mAb protects from progression of DR assessed as vascular permeability 7 days post-I/R surgery (B). Respective images of retinal vasculature and leakage are shown for Isotype IgG treated Control and I/R retinas and mAb Control and I/R retinas, respectively. Data represent mean ± SEM derived from n = 4–5 animals per group. Paired T-test compared with respective control. Scale bars = 200 μm.