Localization of age-related macular degeneration-associated ARMS2 in cytosol, not mitochondria

Abstract

Purpose: To analyze the relationship between ARMS2 and HTRA1 in the association with age-related macular degeneration (AMD) in an independent case-control dataset and to investigate the subcellular localization of the ARMS2 protein in an in vitro system.

Methods: Two SNPs in ARMS2 and HTRA1 were genotyped in 685 cases and 269 controls by a genotyping assay. Allelic association was tested by a chi(2) test. A likelihood ratio test (LRT) of full versus reduced models was used to analyze the interaction between ARMS2 and smoking and HTRA1 and smoking, after adjustment for CFH and age. Immunofluorescence and immunoblot were applied to localize ARMS2 in retinal epithelial ARPE-19 cells and COS7 cell transfected by ARMS2 constructs.

Results: Both significantly associated SNP rs10490924 and rs11200638 (P < 0.0001) are in strong linkage disequilibrium (LD; D' = 0.97, r(2) = 0.93) that generates virtually identical association test and odds ratios. In separate logistic regression models, the interaction effect for both smoking with ARMS2 and with HTRA1 was not statistically significant. Immunofluorescence and immunoblot show that both endogenous and exogenous ARMS2 are mainly distributed in the cytosol, not the mitochondria. Compared with the wild-type, ARMS2 A69S is more likely to be associated with the cytoskeleton in COS7 cells.

Conclusions: The significant associations in ARMS2 and HTRA1 are with polymorphisms in strong LD that confer virtually identical risks, preventing differentiation at the statistical level. ARMS2 was mainly distributed in the cytosol, not in the mitochondrial outer membrane as previously reported, suggesting that ARMS2 may not confer risk to AMD through the mitochondrial pathway.

Fig 1

Immunostaining of ARMS2 in ARPE-19 cells. ARMS2 is not localized within the mitochondria shown by mitoTracker. Furthermore, ARMS2 is not co-localized with the markers for other subcellular organelles and cytoskeleton either. These include golgi apparatus (anti-golgin), endoplasmic reticulum (anti-calnexin), lysosome (anti-LAMP1), microtubule (anti-γ-tubulin), F-actin (phalloidin) and nucleus (DAPI). Most ARMS2 is mainly localized in the perinuclear region and no specific organelle is found to host ARMS2. Preimmune serum and blocking peptide attenuating anti-ARMS2 antibody show negative staining.

Fig 2

Exogenous ARMS2 is not found in any specific organelles. COS7 cells were transfected with N-terminal or C-terminal GFP (green fluorescence protein) tagged, His-tagged ARMS2 constructs and GFP control construct. Most GFP-ARMS2 and His-ARMS2 are mainly localized in the perinuclear region, while GFP control is distributed uniformly within the cell. Co-immunostaining with markers of different cellular organelles shows that no specific organelle is found to host ARMS2. Immunostaining with ARMS2 antibody is largely co-localized with GFP or His-tag staining.

Fig 3

ARPE-19 cells were homogenized and mitochondria were separated from the cytosolic fraction by centrifugation. The ARMS2 antibody was probed against ARPE-19 cell lysate on a western blot and showed a band of the expected size only in the cytosol, not in mitochondrial fraction. The blot was reprobed with porin, which constitutes a mitochondrial membrane protein, and β-actin.

Fig 4

ARMS2 A69S is co-localized with microtubule (anti-tubulin) and actin (stained by phalloidin) while it does not co-localize with mitochondria.