Oxidative stress induces mitochondrial dysfunction and a protective unfolded protein response in RPE cells

Abstract

How cells degenerate from oxidative stress in aging-related disease is incompletely understood. This study's intent was to identify key cytoprotective pathways activated by oxidative stress and determine the extent of their protection. Using an unbiased strategy with microarray analysis, we found that retinal pigmented epithelial (RPE) cells treated with cigarette smoke extract (CSE) had overrepresented genes involved in the antioxidant and unfolded protein response (UPR). Differentially expressed antioxidant genes were predominantly located in the cytoplasm, with no induction of genes that neutralize superoxide and H2O2 in the mitochondria, resulting in accumulation of superoxide and decreased ATP production. Simultaneously, CSE induced the UPR sensors IRE1α, p-PERK, and ATP6, including CHOP, which was cytoprotective because CHOP knockdown decreased cell viability. In mice given intravitreal CSE, the RPE had increased IRE1α and decreased ATP and developed epithelial-mesenchymal transition, as suggested by decreased LRAT abundance, altered ZO-1 immunolabeling, and dysmorphic cell shape. Mildly degenerated RPE from early age-related macular degeneration (AMD) samples had prominent IRE1α, but minimal mitochondrial TOM20 immunolabeling. Although oxidative stress is thought to induce an antioxidant response with cooperation between the mitochondria and the ER, herein we show that mitochondria become impaired sufficiently to induce epithelial-mesenchymal transition despite a protective UPR. With similar responses in early AMD samples, these results suggest that mitochondria are vulnerable to oxidative stress despite a protective UPR during the early phases of aging-related disease.

Keywords: Aging-related disease; ER stress; Epithelial–mesenchymal transition; Free radicals; Mitochondria; Oxidative stress.

Figure 1

ARPE-19 cells remain viable after exposure to CSE, except at 500ug/ml CSE using the Propidium Iodide assay. ***p<0.001; n=3 independent experiments.

Figure 2

Principal component analysis of differentially expressed genes shows transcriptional separation between cells exposed to CSE 100ug/ml (blue dots) and 250ug/ml (green dots) from vehicle control (red dots).

Figure 3

A. ARPE-19 cells exposed to CSE produce superoxide in a dose-dependent manner, as assessed by the DHE assay. *p<0.05; n=3 independent experiments. B. ARPE-19 cells exposed to CSE accumulate carbonylated proteins. Arrows were used to quantify the relative abundance of carbonylated proteins of different size. Lanes to the right of each condition are negative controls where proteins were not derived to expose carbonyl groups. C. Graphs of relative carbonylated protein at the arrows in B. *p<0.05, **p<0.01. D. ARPE-19 cells exposed to CSE developed in a dose-dependent manner, early apoptosis (annexin V+, PI-; green) at viable doses, and then late apoptosis (annexin V+, PI+; yellow), and necrosis (annexin V-, PI+; red). ***p<0.001; n=3 independent experiments.

Figure 4

CSE induces the UPR in ARPE-19 cells. A. Representative western blots of A) IRE1α, B) p-PERK, C) ATF6, and D) CHOP using b-Actin as a loading control, and graph quantifying the relative abundance from 3 independent experiments for each protein after ARPE-19 cells were exposed to CSE. *p<0.05, **p<0.01. Note that because ATF6 is proteolyzed in response to ER stress, the full length protein decreases. E. CHOP is protective after ARPE-19 cells are exposed to CSE because CHOP knockdown using an siRNA to CHOP decreased cell viability compared to scrambled control siRNA. **p<0.01; n=3 independent experiments.

Figure 5

CSE impairs the proteasome. A. CSE impairs proteasome activity using the proteasome SDS-activated activity assay at all doses of CSE compared to vehicle control treated ARPE-19 cells. **p<0.01; n=3 independent experiments. B. Non-toxic doses of proteasome inhibitor MG-132 alone or combined with 100ug/ml CSE, did not alter cell viability using the Propidium Iodide assay, but when combined with higher doses of CSE, decreased viability compared to controls. *p<0.05; n=3 independent experiments.

Figure 6

CSE induces mitochondrial injury. A. A dose dependent increase in mitochondrially generated superoxide, as measured by the Mitosox assay, and B. a decrease in ATP production was observed in ARPE-19 cells treated with CSE. *p<0.05, **p<0.01; ***p<0.001; n= 3 independent experiments.

Figure 7

The UPR is induced and mitochondria are injured in the RPE/choroid 24 hours after mice were given an intravitreal injection of CSE. A. Western blot of IRE1α after mice received intravitreal injection of 250ug/ml CSE in one eye and vehicle in the other eye. B. Graph showing increased IRE1α protein in the RPE/choroid after intravitreal injection of 250ug/ml CSE compared to vehicle control. *p<0.05; n=3 eyes per condition. C. Graph showing decreased ATP production in the RPE/choroid of eyes that received an intravitreal injection of 250ug/ml CSE compared to vehicle control. *p<0.05; n=6 eyes per condition.

Figure 8

Intravitreal CSE induces epithelial-mesenchymal transition. Mice were given intravitreal injection of 250ug/ml CSE or vehicle control. A. Western blot of TOM20 using b-Actin as a loading control. B. Graph showing decreased TOM20 in the RPE/choroid 7 days after CSE treatment; n=4 eyes per condition. *p<0.05. C. Western blot of LRAT using b-Actin as a loading control. The 25kDa band represents the monomer and the 60kDa band represents the dimer. D. Graph showing decreased LRAT (monomer and dimer) in the RPE/choroid 7 days after CSE treatment; n=3 eyes per condition. *P<0.05. E. Representative confocal micrograph of an RPE flatmount of a vehicle control injected eye 10 days after CSE treatment. Note the regular cobblestone shape of the RPE and the regular ZO-1 immunostaining at the cell periphery. F. Representative RPE flatmount of CSE injected eye showing enlarged, irregularly shaped RPE with variable ZO-1 immunostaining compared to E, which was imaged at the same magnification. n=4 eyes per group. Bar=25um.

Figure 9

The UPR is induced and mitochondria are injured in the RPE/choroid 24 hours after mice were given an intravenous injection of CSE. A. Western blot of IRE1α after mice received intravenous injection of 125ug/ml CSE (in 50ul volume) or vehicle. B. Graph showing increased IRE1α protein in the RPE/choroid after intravenous injection of CSE compared to vehicle control. *p<0.05; n=3 eyes per condition. C. Graph showing decreased ATP production in the RPE/choroid of eyes from mice that received intravenous CSE compared to vehicle control. *p<0.05; n=3 eyes per condition.

Figure 10

Intravenous CSE induces epithelial-mesenchymal transition. Mice were given intravenous injection of 125ug/ml CSE (in 50ul volume) or vehicle every other day for 3 doses. A. Western blot of TOM20 using b-Actin as a loading control after 10 days. B. Graph showing decreased TOM20 in the RPE/choroid 7 days after CSE treatment; n=4 eyes per condition. *p<0.05. C. Western blot of LRAT using b-Actin as a loading control. The 25kDa band represents the monomer and the 60kDa band represents the dimer. D. Graph showing decreased LRAT (monomer and dimer) in the RPE/choroid 7 days after CSE treatment; n=3 eyes per condition. *P<0.05. E. Representative confocal micrograph of an RPE flatmount of a vehicle control injected eye 10 days after CSE treatment. Note the regular cobblestone shape of the RPE and the regular ZO-1 immunostaining at the cell periphery. F. Representative RPE flatmount after intravenous CSE showing enlarged, irregularly shaped RPE with variable ZO-1 immunostaining compared to E, which was imaged at the same magnification. n=4 eyes per group. Bar=25um.

Figure 11

IRE1α is decreased in the RPE overlying large drusen. A. Macular section from a 54 yo unaffected M. IRE1α immunolabeling is consistent and prominent in morphologically normal RPE overlying unthickened Bruch’s membrane. Bar=25um. Ch, choroid. B. IRE1α labeling is obvious in the same section after Nuance software has subtracted melanin pigment. C. Macular section from a 61 yo M with early AMD. IRE1α immunolabeling remains prominent in the RPE with thinned morphology that is overlying a small druse (*). D. IRE1α labeling in the RPE after Nuance subtraction of melanin from the section in C. E. Macular section from an 87 yo M with early AMD has a large druse (***). The thinned RPE overlying the druse and the adjacent RPE have decreased IRE1α labeling. F. Same section as E after Nuance subtraction of melanin. G. IgG control. Druse (**) H. IgG control after Nuance subtraction of melanin.

Figure 12

Decreased mitochondrial TOM20 immunolabeling in early AMD. A. Macular section from a 54 yo unaffected M. RPE morphology is preserved and TOM20 immunolabeling is basally located. Bar = 25um. B. Nuance subtraction of melanin. C. Macular section from a 68 yo F with early AMD. The RPE over a small druse (*) shows decreased TOM20 labeling compared to adjacent RPE. D. Nuance subtraction of melanin. E. Macular section from the same donor as C, D (68 yo F with early AMD). The RPE overlying a large druse (*) has minimal or absent TOM20 labeling. Residual TOM20 labeling is seen in RPE adjacent to druse. F. Nuance subtraction of melanin. G. IgG control. Ch, choroid. H. IgG control after Nuance subtraction of melanin.