P38 inhibition reverses TGFβ1 and TNFα-induced contraction in a model of proliferative vitreoretinopathy

Abstract

Proliferative vitreoretinopathy (PVR) is a metaplasia in the vitreous of the eye manifested by the transformation of retinal pigment epithelial (RPE) cells and the development of contracting epiretinal membranes (ERM), which lead to retinal detachment and vision loss. While TGFβ1 and TNFα have been associated with PVR, here we show that these cytokines act synergistically to induce an aggressive contraction phenotype on adult human (ah)RPE. Connected RPE detach upon contraction and form motile membranes that recruit more cells. TGFβ1 and TNFα (TNT)-induced contracting membranes uniquely express muscle and extracellular rearrangement genes. Whole transcriptome RNA sequencing of patient-dissected PVR membranes showed activation of the p38-MAPK signaling pathway. Inhibition of p38 during TNT treatment blocks ahRPE transformation and membrane contraction. Furthermore, TNT-induced membrane contractility can be reversed by p38 inhibition after induction. Therefore, targeting the p38-MAPK pathway may have therapeutic benefits for patients with PVR even after the onset of contracting ERMs.

Keywords: Collective cell migration; Mechanisms of disease.

Fig. 1

TGFβ1 and TNFα work synergistically to induce transformation of RPE. RPE cultured in the presence of 10 ng/ml TGFβ1 and/or TNFα for 5 days then assayed morphologically and transcriptionally. a RPE dissection schematic. Scale bar = 50 μM. b Phase images of RPE 5 days after treatment with TGFβ1 and/or TNFα. Scale bar = 50 μM. c Time-lapse image frames highlighting stages of RPE transformation when stimulated with TGFβ1 and TNFα for 5 days. Scale bar = 300 μM. d Quantification of contractile membranes with varying concentrations of FBS after 5 days treatment with TGFβ1 and/or TNFα. e Coefficient of variation (CV) for total RPE lines (n = 11) and one RPE line. f RT-qPCR of RPE markers. Statistical significance was calculated using Student’s t-test, *P < 0.05; **P < 0.01; and ***P < 0.001

Fig. 2

The role of contractility in TGFβ1 and TNFα induced PVR-like membranes. a RT-qPCR of extracellular matrix and cell motility markers. b Anti-ACTG2 and TENASCIN C immunostaining. Scale bar = 75 μm. c Phase images of RPE 5 days after treatment with TNT, TNT + (s)-nitro-Blebbistatin, or TNT + W-7. Scale bar = 50 μm. d Quantification of contractile membranes per well after 5 days in control media, TNT media, TNT + (s)-nitro-Blebbistatin media, or TNT + W-7 media. Statistical significance was calculated using Student’s t-test, *P < 0.05; **P < 0.01; and ***P < 0.001

Fig. 3

Characterization of TNT-treated ahRPE compared to patient-dissected PVR membranes. a Patient-dissected PVR samples and ahRPE were treated with 10 ng/ml TGFβ1 and/or TNFα for 5 days then fixed. Immunostaining was performed with Anti-αSMA (scale bar = 100 μm), SNAIL, COL1A1, COL1A2 (scale bar = 50 μm), and LAMININ (scale bar = 50 μm) antibodies. b RT-qPCR of extracellular matrix and EMT specific genes. c RT-qPCR of extracellular matrix genes in TNT-treated ahRPE compared to patient-dissected PVR membranes. d Western blot analysis comparing E- and N-Cadherin expression in ahRPE treated with TNT-for 5 days. Statistical significance was calculated using Student’s t-test, *P < 0.05; **P < 0.01; and ***P < 0.001

Fig. 4

RNA expression analysis reveals a central role for a p38 network in PVR. a Gene expression analysis of patient-dissected PVR membranes (PVR), control ahRPE (RPE), and TNT-treated ahRPE (TNT) by RNA-sequencing. a Using edgeR and DEseq2 pairwise comparisons of each condition were done and overlaps between significantly changing genes were identified. b The data structure was examined by singular value decomposition. SVD1 and SVD2 explain ~93% of the variance in the data. c A p38 gene network was constructed using STRINGdb and each of the pairwise comparisons between samples was intersected with this network and an enrichment was calculated using the BayesFactor package in R (Higher BF means more evidence for enrichment). Overlaps between comparisons were identified. d Using the genes identified by the intersection of control ahRPE vs. patient-dissected PVR membranes comparison and the p38 network, a sub network was created. The betweenness centrality of each gene and the betweenness of each edge of this sub network was calculated and determines the size and thickness of the nodes and lines, respectively. Communities in the network were identified by the GLay algorithm (identified by color), and the shape of each node indicates the group from the Venn diagram displayed in (c). e The top categories shared between patient-dissected PVR membranes and TNT-treated ahRPE by Ingenuity pathway analysis. Sig signaling, Int interaction, Maint maintanence, Org organization, Metab metabolism

Fig. 5

p38 inhibition in TNT-treated ahRPE. AhRPE were treated with TNT and TNT + SB 202190. a Phase images. Scale bar = 75 μm. b Anti-αSMA, SNAIL, COL1A1, COL1A2, and LAMININ immunostaining. Scale bar = 50 μm. c Western blot comparing phospho-p38 expression in TNT-treated ahRPE for 0 (control), 30, 60 and 120 min. d Western blot comparing phospho-p38 and phospho-HSP27 expression in ahRPE treated with TNT or TNT + SB 202190 for 30 min. e RT-qPCR of extracellular matrix and EMT genes in ahRPE treated with TNT and TNT + SB 202190. f RT-qPCR of extracellular matrix and EMT genes in ahRPE treated with TNT and TNT + BIRB 796. g RT-qPCR of p38 in cells transfected with or without p38 siRNA. h RT-qPCR of extracellular matrix and EMT genes in cells treated in the presence of TGFβ1 and TNFα co-supplemented basal media after transfection without p38 siRNA, and TGFβ1, TNFα co-supplemented basal media after ahRPE were transfected with p38 siRNA. Statistical significance was calculated using Wilcxon signed-rank test, *P < 0.05; **P < 0.01. i Quantification of wound healing scratch assay. j Phase images of wound healing scratch assay on ahRPE treated with basal media alone, TGFβ1 and TNFα supplemented media or TGFβ1 and TNFα and SB 202190 supplemented media for 72 h. Scale bar = 300 μm. White arrowheads indicate areas of preliminary mass formation. Statistical significance was calculated using Student’s t-test, *P < 0.05; **P < 0.01; and ***P < 0.001

Fig. 6

Inhibition of p38 reverses TNT-induced contractile mass formation. a Schematic of the experimental design to test contraction reversal. Scale bar = 100 μm. b Representative phase images of ahRPE treated with TNT for 72 h then media changed to TNT + SB 202190 for another 24 h. Scale bar = 100 μm. c Quantification of membrane formations between cultures of ahRPE in contraction reversal conditions described in 6a. d Western blot comparison of ACTG2 and COL1A2 between cultures of ahRPE in contraction reversal conditions. e RT-qPCR of extracellular matrix and EMT genes from ahRPE in contraction reversal conditions. (f) Anti-SNAIL, αSMA, COL1A2, and P38 immunostaining after ahRPE were placed in contraction reversal conditions. Scale bar = 75 μm. Student’s t-test, *P < 0.05; **P < 0.01; and ***P < 0.001

Fig. 7

Modeling the role of RPE in PVR. When TGFβ and TNFα are not present, p38 expression is low, RPE are a polarized epithelium expressing typical RPE identity markers. Upon binding of TGFβ and TNFα by their respective receptors, p38 is phosphorylated, which leads to HSP27 phosphorylation and p38 nuclear localization. Genes associated with EMT and contractility become up-regulated and RPE identity genes are down-regulated, which leads to expression of proteins associated with PVR and RPE contraction