Lack of Pnpla2 Accelerates Progression of AMD-Like Features in Mice

Abstract

Purpose

Lipid accumulation in the retinal pigment epithelium (RPE) contributes to cellular stress and progression of age-related macular degeneration (AMD). However, the regulation of lipid homeostasis in AMD development is not fully elucidated. The study investigates the effects of Pnpla2 deletion, a gene involved in lipid regulation, on key markers of RPE senescence and aging with potential relevance to AMD.

Methods

RPE flat mounts and retinal cryosections were analyzed from Pnpla2^−/− and Pnpla2^+/+ mice aged 3 months. Senescence-associated β-galactosidase (SA-β-gal) activity was assessed in flat mounts. DAPI was used to quantify RPE cells with single or multiple nuclei. Immunohistofluorescence was carried out to assess RPE tight junctions and expression of senescence and AMD markers using antibodies to zonula occludens 1 (ZO-1), phospho-histone (P-γ-H2AX), apolipoprotein E (ApoE), and high mobility group box 1 (HMGB1). Fundus imaging was acquired, and electroretinography (ERG) assessed visual function.

Results

Pnpla2^−/− RPE exhibited increased SA-β-gal activity, multinucleation of the population, and the translocation of HMGB1 from nucleus to cytoplasm, indicative of cellular senescence. Tight junctions were disrupted. The number of P-γ-H2AX–positive RPE cells increased by 50%, suggesting increased DNA damage. ApoE levels were elevated in Bruch's membrane and subretinal regions. At 3 months of age, attenuation of ERG c-wave amplitude was observed in both Pnpla2^−/− and Pnpla2^+/− mice. By 7 months of age, Pnpla2^+/− mice exhibited continued attenuation of ERG c-wave amplitude and developed white spots.

Conclusions

Pnpla2 deficiency accelerates cellular features of RPE aging and generates AMD-like features. These findings underscore the importance of PNPLA2 in mitigating AMD progression and highlight its significance in retinal health and degeneration.

Lipid accumulation in the retinal pigment epithelium (RPE)–choroid contributes to cellular stress and the progression of age-related macular degeneration (AMD).^1–4 Variants in multiple genes related to lipid regulation/metabolism are important in increasing or decreasing the risk of late AMD.⁵ However, the mechanisms regulating lipid homeostasis in AMD remain inadequately understood. Lipases and phospholipases are critical mediators of lipid metabolism, playing essential roles in maintaining the structure and function of the RPE and retina.^6–8 Among these, pigment epithelium-derived factor (PEDF) and its receptor, PEDF-R (encoded by the PNPLA2 gene, and also known as ATGL, desnutrin, PLA2ζ), are key regulators in lipid remodeling, with PEDF stimulating the lipase activity of PEDF-R to reduce lipid deposits.⁹ Both PEDF and PEDF-R are involved in retinal survival.^10,11 Notably, our recent findings reveal that PEDF-R depletion in the RPE leads to lipid accumulation and phagocytosis impairment, while PEDF depletion exacerbates RPE senescence and phagocytic dysfunction.^12,13 However, the specific consequences of PEDF-R depletion on the processes of senescence and aging remain unexplored.

RPE aging and senescence, central to AMD progression, are characterized by markers such as senescence-associated β-galactosidase (SA-β-gal) activity, DNA double-strand breaks (DSBs), inflammation, and the accumulation of apolipoprotein E (ApoE) and lipofuscin. The integrity of the blood–retinal barrier (BRB), maintained by the RPE, is essential for RPE function, photoreceptor nourishment, and overall retinal health.¹⁴ BRB breakdown, commonly observed in physiological aging and AMD, is often attributed to weakened intercellular tight junctions, where zonula occludens 1 (ZO-1) serves as a critical structural protein.¹⁵ Senescent multinucleated giant cells, frequently observed in age-related diseases, serve as markers of cellular aging.^16,17 DNA DSBs are hallmarks of senescence and are implicated in AMD pathogenesis.^18–20 High mobility group box 1 (HMGB1) is another marker of senescence, reflecting a transition from homeostasis to stress, inflammation, and irreversible growth arrest.²¹ In senescent cells, HMGB1 translocates from the nucleus to the cytoplasm and is secreted extracellularly, signaling stress-related changes. Additionally, senescent cells secrete inflammatory cytokines, which activate microglia in the retina.²² Moreover, ApoE, a protein expressed in the healthy retina, has been identified as a key marker of AMD, when its levels or distribution are altered, suggesting a role in disease pathology.^23,24 Its accumulation in drusen (heterogeneous extracellular deposits characteristic of AMD) and basal laminar deposits, a thickened basal lamina of the RPE, further implicates it in AMD pathophysiology.^25,26 Given the accumulation of lipofuscin and drusen as hallmark features of RPE dysfunction in AMD, fundus autofluorescence (FAF) imaging serves as a noninvasive method to visualize these changes in vivo.^27,28 It is commonly used to monitor disease progression, evaluate RPE integrity, and assess treatment responses that may reflect metabolic stress in the RPE.

Given the age-related nature of AMD, its link to metabolic stress, and the emerging role of lipid dysregulation in disease, there is increasing interest in PNPLA2, a gene crucial for lipid metabolism and retina survival, and its role in RPE senescence and aging. In this study, we investigate how Pnpla2 depletion in mice affects key markers of RPE senescence, aging, and AMD progression, offering new insights into the importance of lipid homeostasis in retinal health.

Materials and Methods

Animals

CRISPR Pnpla2 (C57BL/6J) mice were generated, as previously described,¹⁰ with genotypes verified using the genotyping service of Transnetyx (Cordova, TN, USA) and the KO model validated as previously reported.¹⁰ Experimental Pnpla2^−/− (homozygous) mice were used at 3 months of age, while Pnpla2^+/+ (wild-type [WT] C57BL/6J) and Pnpla2^+/− (heterozygous) mice were used at 3 and 7 months of age, with mixed sexes. Since homozygous mice do not survive beyond 3 months, 7-month-old heterozygous Pnpla2^+/− mice were used for aging experiments.⁵ Animals were maintained in the animal facility of the National Institutes of Health (Bethesda, MD, USA), fed a normal chow diet, and kept at a 12-hour light/dark cycle. All experimental procedures were approved by the National Eye Institute Animal Care and Use Committee and performed as per guidelines of the ARVO statement for the Use of Animals in Ophthalmic and Vision Research and in accordance with the Animal Research: Reporting of In Vivo Experiments guidelines.

SA-β-Gal Activity Assay

SA-β-gal enzymatic assays were performed in RPE/choroid flat mounts and as described before.¹² Briefly, eyecups were dissected from 3-month-old mice. To start the enzymatic reaction, the β-galactosidase substrate C₁₂FDG (5-dodecanoylaminofluorescein di-β-D-galactopyranoside; cat. D2893; Thermo Fisher Scientific, Carlsbad, CA, USA) at 50 mM was added to culture medium for RPE/choroid flat mounts, which contains Dulbecco's modified Eagle’s medium/F-12 (cat. 11330032; Thermo Fisher Scientific) supplemented with 10% fetal bovine serum (cat. 10082-147; Thermo Fisher Scientific), and incubated for 2 hours in an incubator. Then the flat mounts were fixed with 4% paraformaldehyde (PFA; cat. 28906; Thermo Fisher Scientific), followed by staining with phalloidin probe and DAPI at 0.1 mg/mL (stock diluted 1:50) in antifade mounting medium (cat. H-2000-10; Vector Laboratories, Newark, CA, USA). Imaging was performed using a Zeiss LSM880 (Oberkochen, Baden-Württemberg, Germany) confocal microscope.

Cryostat Sections and Immunofluorescent Staining

Eyeballs were fixed in 4% PFA at 4°C overnight, followed by dehydration through a glucose (D-(+)-Glucose; cat. G5767; Millipore Sigma, Rockville, MD, USA) gradient (10%, 20%, and 30%) glucose in PBS (cat. 10010-023; Thermo Fisher Scientific), with each concentration applied for 4 hours. The dehydrated eyeballs were then embedded in optimal cutting temperature compound (Tissue-Tek O.C.T. Compound, cat. 4583; SAKURA, Torrance, CA, USA) for sectioning with a Leica CM1860 Cryostat (Wetzlar, Hesse, Germany). The sections were mounted on microscope slides and rehydrated in PBS (three washes, 5 minutes each). Permeabilization was performed using 0.4% Triton-X 100 (cat. T8787; Millipore Sigma, Rockville, MD, USA) in PBS for 10 minutes, followed by blocking with 5% bovine serum albumin (cat. 9048-46-8; Goldbio, St. Louis, MO, USA) in PBS for 1 hour at room temperature.

Immunoreaction with primary antibodies diluted in blocking buffer (see Table) was performed by incubating the tissue at 4°C for 16 hours. Then the tissues on the slides were washed with PBS (three washes, for 5 minutes each), followed by incubation with secondary antibodies for 2 hours at room temperature (RT) and washed with PBS (as above). Finally, the slides were mounted with antifade reagent containing DAPI (cat. H-2000-10; Vector Laboratories). Imaging was conducted using a Zeiss LSM880 confocal microscope.

Table.

List of Probes and Antibodies

Antibody or Probe	Dilution	Company	Catalog Number
Anti–apolipoprotein E antibody, goat polyclonal	1:200	Millipore Sigma	Ab947
Anti–IBA-1 antibody, goat polyclonal	1:200	Abcam	ab5076
Alexa Fluor 555 phalloidin	1:500	Thermo Fisher Scientific	A34055
Phospho-histone H2A.X (Ser139) (20E3) rabbit mAb	1:200	Cell Signaling Technology	9718S
ZO-1, rabbit polyclonal antibody	1:200	Thermo Fisher Scientific	40-2200
Donkey anti-rabbit IgG (H+L) highly cross-adsorbed secondary antibody, Alexa Fluor 555	1:400	Thermo Fisher Scientific	A-31572
Donkey anti-goat IgG (H+L) cross-adsorbed secondary antibody, Alexa Fluor 488	1:400	Thermo Fisher Scientific	A-11055
DAPI	0.1 mg/mL	Thermo Fisher Scientific	D3571

RPE/Choroid Flat Mount Immunofluorescent Staining

RPE/choroid complexes were obtained by removing the cornea and neural retina. Four cuts were made on the RPE/choroid complex and transferred to a dish as a drop. The RPE/choroid complex was flattened by aspirating PBS with a pipette, fixed in 4% PFA for 30 minutes at RT, permeabilized in 0.4% Triton X-100 at RT for 10 minutes, and incubated in blocking solution of 5% BSA, followed by incubation in a solution of primary antibody diluted 1:200 in 5% BSA in a 1.5-mL microcentrifuge tube at 4°C for 16 hours, followed by three washes in 1 mL PBS for 5 minutes each time. Then, the complexes were incubated with a solution with secondary antibody for 1.5 hours at RT, followed by washing three times in 1 mL PBS for 5 minutes each, and then transferred to glass slides (cat. 1255015; Fisher Scientific, Pittsburgh, PA, USA), incubated with antifade solution containing DAPI for 1 minute, and finally covered with a coverslip. Images were acquired with a Zeiss LSM880 confocal microscope.

Quantification of Nucleation Status

RPE cells in C57BL/6J mice exhibit regional variation in nuclear number—cells in the mid-peripheral and peripapillary regions are frequently multinucleated, while those in the peripheral third are more consistently mononucleated. To reduce variability and ensure consistency in quantification, we focused our analysis on the peripheral third of the RPE. RPE cells with continuous tight junctions, as evidenced by ZO-1 expression, to ensure individual cell boundaries and clearly defined nuclei, were analyzed for the percentage of single-nucleated and multinucleated RPE populations. RPE cells were categorized according to the number of nuclei in a single RPE cell. The percentage of multinucleated RPE cells was calculated by subtracting the percentage of mononucleated cells from 100%.

Quantification of Fluorescence Intensity and Cell Size

Fluorescence intensity and cell size were quantified using ImageJ software (v1.54g; National Institutes of Health). For general quantification, fluorescence channels were separated, and the relevant channel was thresholded to define regions of interest (ROIs). Mean intensity and area within each ROI were measured, and total fluorescence intensity was calculated as the product of area and mean intensity.

For HMGB1, both nuclear and cytoplasmic fluorescence intensities were analyzed. The DAPI channel was used to define nuclear ROIs, and the HMGB1 channel was used to measure corresponding signal intensity. Six contiguous RPE cell nuclei were selected per image, and nuclear fluorescence was measured after clearing all nonnuclear signals. Cytoplasmic intensity was measured by masking nuclear signals and selecting the surrounding cytoplasmic region of the same cells. Total intensity for both compartments was calculated as area × mean intensity.

Cell size was assessed by tracing the phalloidin-labeled borders of RPE cells. Images were scaled using the embedded scale bar, and the area of individual cells was measured using freehand selection. Eight cells were selected randomly and analyzed per image, across three images per mouse and three mice per group. Graphs were generated using GraphPad Prism 10 (GraphPad Software, La Jolla, CA, USA).

RPE Extracts

Protein extracts were prepared from dissected RPE tissue by incubating it in RIPA Lysis Extraction Buffer (cat. #89900; Thermo Fisher Scientific, Waltham, MA, USA) supplemented with protease inhibitors (Pierce Protease Inhibitor Tablets, cat. A32963; Thermo Fisher Scientific) at a volume of 80 µL per RPE, as previously described.²⁹ Protein concentrations were determined using a NanoDrop spectrophotometer (Thermo Fisher, Waltham, MA, USA) at 280 nm, calibrated against a standard curve generated with pure BSA (0–4 mg/mL), and prepared in RIPA Lysis and Extraction Buffer.

ELISA

IL-6 concentrations in the RPE extracts were measured using a Quantikine Mouse Interleukin (IL-6) ELISA kit according to the manufacturer's instructions (cat. D6050; R&D Systems, Minneapolis, MN, USA). Freshly prepared soluble RPE extracts (30 µL per sample) were loaded in duplicate into each well of a 96-well plate. Optical density was measured at 540 nm using a SpectraMax iD5 microplate reader (Molecular Devices, San Jose, CA, USA). IL-6 levels were determined by comparing the sample optical densities to a standard curve generated with pure IL-6 (0–300 pg/mL). Final IL-6 concentrations were normalized to total protein content and expressed as pg IL-6 per milligram of protein.

PEDF concentrations in the RPE extracts were also measured using a Mouse PEDF ELISA (cat. XPEM1280; XpressBio, Frederick, MD, USA) in RPE extracts because it is a ligand for PEDF-R (Supplementary Fig. S1).

BODIPY Staining in the RPE of Pnpla2 Knockout Mouse Line

Eyecups were fixed in 4% paraformaldehyde for 1 hour, followed by incubation in 10% goat serum in TBS-T for 1 hour. Samples were then incubated with BODIPY-493/503 (cat. D2191; Invitrogen, Grand Island, NY, USA) and phalloidin-568 (1:500, cat. A34054; Thermo Fisher Scientific) for 1 hour. After staining, eyecups were cut into petal-like sections and placed on a slide with a cover slip. Images were acquired using an Olympus FV-1000 (Hachioji-shi, Tokyo, Japan) confocal microscope. Images from different mice were acquired with identical magnification, gain and exposure settings, and the same RPE region.

Fundus Autofluorescence

Fundus imaging was performed using the Heidelberg Retina Angiography 2 (HRA2) system (Heidelberg Engineering, Heidelberg, Germany). To prepare for imaging, mice were anesthetized as previously described.^10,30 Their pupils were dilated with 1% tropicamide (cat. NDC 17,478-101-12; Alcon, Fort Worth, TX, USA) for 5 minutes. GenTeal eye gel with 0.3% Hypomellose (cat. NDC 0078-0429-57; Alcon) was applied liberally to the cornea throughout the procedure to prevent corneal dryness. A single scan was acquired using an excitation wavelength of 488 nm (blue light), while a detection range of 787 to 830 nm was used to stimulate autofluorescent molecules in the RPE and other retinal layers.

Electroretinography C-wave

Electroretinograms (ERGs) were recorded using an Espion E2 system with ColorDome (Diagnosys LLC, Lowell, MA, USA) with a heated surface. Mice were dark-adapted overnight as previously described.¹⁰ Briefly, pupils were dilated with 1% tropicamide for 10 minutes, and mice were anesthetized with an intraperitoneal injection of Zetamine (ketamine hydrochloride injection, USP; cat. NDC 13985-584-10; VetOne, Boise, ID, USA) at 92.5 mg/kg and AnaSed (xylazine injection; cat. NDC 59399-110-20; Akorn Pharmaceuticals, Lake Forest, IL, USA) at 5.5 mg/kg, according to previously published methodology.¹⁰ GenTeal eye gel with 0.3% Hypomellose was applied throughout the procedure, and ERG c-wave was recorded. Using a standard, bright flash stimulus under scotopic conditions to evoke the c-wave, luminances were set from 0.01 to 10.00 cd·s/m², with 15 flashes per stimuli.³¹ The values for the c-wave for both eyes were exported to Excel (Microsoft, Redmond, WA, USA), and both eyes were averaged to obtain amplitude values for each mouse.

Statistical Analysis

A minimum of three animals per genotype were used in each experiment. Data were analyzed using GraphPad Prism (version 10.2.2; GraphPad Software). Comparisons between experimental groups were performed using either one-way ANOVA with Dunnett's multiple comparisons test or unpaired t-test. Results are presented as mean ± SD, and P values less than 0.05 were considered statistically significant.

Results

Deletion of the Pnpla2 Gene Increased SA-β-Gal Activity in the RPE

SA-β-gal activity is widely recognized as a marker of cellular senescence, in both whole mounts and cryosections.^32,33 To assess this activity, RPE/choroid flat mounts from 3-month-old Pnpla2^−/− mice and their Pnpla2^+/+ wild-type littermates (three mice per group) were incubated in media containing C₁₂FDG, a substrate for SA-β-gal, to initiate the enzymatic activity reaction. Following incubation, the reaction products (C12-fluorescein) in the RPE flat mounts were examined, where green fluorescein detected the β-gal activity. Additionally, red phalloidin was used to highlight the RPE structure. In wild-type mouse RPE, only a few small green fluorescent dots were observed, whereas the Pnpla2^−/− mouse RPE exhibited more intense green fluorescein dots across the three mice assessed per group. Figure 1 presents representative ROIs from the stained flat mounts and quantification of the C12-fluorescein intensity in the ROIs. Fluorescence intensity was approximately ninefold higher in Pnpla2^−/− RPE compared to Pnpla2^+/+ RPE. These findings demonstrate that Pnpla2 gene deficiency increases SA-β-gal activity in the RPE, suggesting a potential role in promoting RPE senescence.

Figure 1.

SA-β-gal activity in Pnpla2^−/− RPE flat mounts. (A) SA-β-gal activity (green) was assessed in RPE/choroid flat mounts of Pnpla2^+/+ (WT) (top row) and Pnpla2^−/− (bottom row) mice at 3 months of age using the substrate of β-galactosidase C₁₂FDG to produce C12-fluorescein. Three individual mice per genotype were assayed. A representative region from each genotype is shown with the β-galactosidase product (green), phalloidin (red) to label RPE, and DAPI (blue) to label cell nuclei. Scale bar: 10 µm. (B) Quantification of C12-fluorescein intensity per ROI is shown. Three ROIs per eyecup and three eyecups from three different mice per genotype were analyzed (n = 3). Each data point corresponds to the average of three ROIs per eyecup. Statistical significance was calculated using an unpaired t-test for the two groups (WT and Pnpla2^−/− mice). ****P < 0.0001.

Pnpla2 Depletion in the RPE Leads to Tight Junction Breakdown and Increases in Multinucleated RPE Cell Generation

Breakdown of the BRB, often linked to physiological aging and AMD, frequently results from weakened intercellular tight junctions, with ZO-1 acting as a key structural protein.^34,35 To assess tight junction integrity in RPE, we performed ZO-1 immunocytochemistry on RPE/choroid flat mounts from 3-month-old Pnpla2^−/− and Pnpla2^+/+ mice (three mice per group). As shown in Figure 2A, ZO-1 staining appears continuous along cell borders in wild-type RPE, reflecting intact tight junctions, while in Pnpla2^−/−, RPE exhibits disrupted ZO-1 continuity, indicating tight junction degradation.

Figure 2.

Tight junctions and multinucleated cells in Pnpla2^−/− RPE. (A) Immunostaining of ZO-1 (green) was performed to label RPE cells in flat mounts of Pnpla2^+/+ (WT) and Pnpla2^−/− mice at 3 months of age. DAPI was used to stain the nuclei (blue). ROIs were acquired from the peripheral one-third area of RPE flat mounts (see scheme of flat mount with asterisk to illustrate this area). The right column shows a representative region co-labeled with ZO-1 and DAPI and merged from a WT mouse. The left column shows a representative region co-labeled with ZO-1 and DAPI and merged from Pnpla2^−/− mouse. Three animals were used, and three ROIs were acquired per eye (n = 3, 3 ROIs per sample). Scale bar: 20 µm. (B) Plots of numbers of multiple nuclei that were recorded are shown. Each data point corresponds to the average measurement of three ROIs per eyecup. Three eyecups, one from each WT, and three eyecups, one from each Pnpla2^−/− mouse, were used for statistical significance analysis calculated using an unpaired t-test for the two groups of WT and Pnpla2^−/− mice. ****P < 0.0001. (C) Plots of RPE size of WT and Pnpla2^−/− mice. Images of ZO-1–stained RPE flat mounts were analyzed for RPE size. Eight RPE cells with intact tight junctions labeled with ZO-1 per ROI were selected to determine cell size, with three ROIs per eyecup and three eyecups from different mice per genotype (n = 3). Statistical significance was calculated using an unpaired t-test for the two groups (WT and Pnpla2^−/− mice). **P < 0.01.

Senescent multinucleated giant cells are commonly associated with various age-related diseases,¹⁶ and the formation of multinucleated cells is often used as an indicator of cellular senescence.³⁶ To investigate the multinucleated RPE cell population in Pnpla2^−/− mice, we examined multinucleated RPE cell generation in 3-month-old Pnpla2^−/− and Pnpla2^+/+ mice (three mice per group). Our analysis focused on the RPE peripheral one-third area of the eye cup, where cells with continuous ZO-1 expression were categorized based on the presence of single or multiple nuclei. As illustrated in Figure 2B, 97% of RPE cells in wild-type mice contained a single nucleus, while only 3% were multinucleated. In contrast, RPE of Pnpla2^−/− mice exhibited a significant increase in multinucleated RPE cells, with 73% of cells containing multiple nuclei and 27% containing a single nucleus. In addition, comparison of RPE cell size between WT and Pnpla2^−/− mice (Fig. 2C) showed an increase in size for Pnpla2^−/− RPE relative to WT, providing further support for promotion of RPE senescence upon Pnpla2 depletion.

These results demonstrate that Pnpla2 depletion in RPE cells leads to tight junction breakdown and a marked increase in multinucleated cells, indicating its potential contribution to RPE aging and senescence.

Pnpla2 Deficiency Increases DNA Double-Strand Breaks in RPE

DNA double-strand breaks are key indicators of cellular senescence and are implicated in AMD pathogenesis.^37,38 To assess DNA damage, we performed immunofluorescent staining on frozen retinal sections from 3-month-old Pnpla2^+/+ and Pnpla2^−/− mice (three mice per group) using an antibody against P-γ-H2AX, a marker of double-strand DNA breaks. In Pnpla2^+/+ mice, no P-γ-H2AX–positive RPE cells were detected. In contrast, Pnpla2^−/− mice showed P-γ-H2AX–positive cells in the RPE. Quantification revealed a significant increase in P-γ-H2AX–positive nuclei in Pnpla2^−/− samples, with 53% of RPE nuclei marked positive, whereas no P-γ-H2AX–positive nuclei were detected in wild-type mice (Fig. 3). These findings indicate that Pnpla2 deficiency leads to increased DNA damage in the RPE, suggesting a role of Pnpla2 in promoting DNA integrity, thus potentially mitigating AMD progression.

Figure 3.

DNA double-strand breaks in Pnpla2^−/− retinal sections. Immunohistostaining of P-γ-H2AX (red) was performed to label DNA double-strand breaks in retinal cross sections from Pnpla2^+/+ (WT) and Pnpla2^−/− mice at 3 months old. DAPI stain was used to label nuclei (blue). (A) The differential interference contrast (DIC) image of a representative region (top) co-labeled with P-γ-H2AX antibody and DAPI (two in the middle) and merged (bottom). BM, basal membrane. (B) Quantification of P-γ-H2AX immunostaining. Each data point corresponds to the average measurement of three ROIs per eyecup, with three eyecups per genotype. The y-axis corresponds to the percentage of P-γ-H2AX–positive nuclei. Three eyecups from each WT and Pnpla2^−/− mouse were used for statistical significance analysis, which was calculated using an unpaired t-test for the two groups (WT and Pnpla2^−/− mice). ****P < 0.0001. Scale bar: 20 µm.

Loss of Pnpla2 Leads to HMGB1 Translocation

The subcellular localization of HMGB1 serves as a marker for cell senescence, reflecting the transition from normal homeostasis to a state of stress, inflammation, and irreversible growth arrest.²¹ Under normal physiological conditions, HMGB1 is predominantly localized in the nucleus, where it functions as a chromatin-binding protein involved in DNA repair, transcriptional regulation, and genome stability. In senescent cells, HMGB1 often translocates from the nucleus to the cytoplasm and is subsequently secreted into the extracellular space, a redistribution indicative of cellular stress and senescence-related changes.

To determine whether Pnpla2 knockout alters HMGB1 distribution in RPE cells, we stained retinal cryosections of 3-month-old wild-type and Pnpla2^−/− mice using anti-HMGB1 antibody. As shown in Figure 4, HMGB1 was primarily localized in the nucleus in RPE cells from wild-type mice. In contrast, in Pnpla2^−/− RPE, HMGB1 translocated from the nucleus to the cytoplasm. Quantification of HMGB1 intensity in nuclear and cytoplasmic compartments showed a marked redistribution in Pnpla2^−/− RPE compared to wild-type cells (Fig. 4B). In wild-type RPE, 79% of HMGB1 fluorescence was nuclear and 21% cytoplasmic, whereas in Pnpla2^−/− RPE, 36% was nuclear and 64% cytoplasmic. These values imply a 43% translocation of nuclear HMGB1 to the cytoplasm in Pnpla2^−/− cells. Nuclear HMGB1 intensity was reduced by about half, while cytoplasmic HMGB1 showed a threefold increase in Pnpla2^−/− RPE relative to wild-type RPE. These findings suggest that Pnpla2 deficiency promotes HMGB1 redistribution to the cytoplasm of RPE cells, indicating that Pnpla2 is essential for maintaining normal homeostasis in these cells. Its loss likely drives the cells into a state of cellular senescence.

Figure 4.

HMGB1 translocation in Pnpla2^−/− RPE. Immunohistostaining was performed to determine the subcellular distribution pattern of HMGB1 in retinal cryosections from 3-month-old WT (left) and Pnpla2^−/− (right) mice. Retinal sections from three mice were analyzed, and a representative region is shown for each genotype (n = 3). (A) HMGB1 was labeled using red fluorescence (a and e). DAPI was labeled in blue (b and f). Merged images of red and blue fluorescence channels are shown in c and g. The differential interference contrast (DIC) imaging, merged with images of two channels, is displayed in d and h. The inserts in the center are images at higher magnification from c and g and shown in boxes 1 and 2, respectively. Scale bar: 20 µm. (B) Quantification of HMGB1 immunostaining in nuclear and cytoplasmic compartments. Three mice per genotype were analyzed. One image of a retinal cryosection was acquired per mouse, and three ROIs, each containing six contiguous nuclei, were selected per image. HMGB1 intensity was measured separately in the nucleus and cytoplasm. The intensity of each compartment was expressed as a percentage of the total HMGB1 signal within the ROI (set to 100%). Each data point represents the average nuclear and cytoplasmic HMGB1 intensity per eyecup. Three eyecups per genotype were analyzed. Data were analyzed using an unpaired t-test. **P < 0.01, ****P < 0.0001.

Pnpla2 Knockout Induces Inflammation in the RPE

Senescent cells release inflammatory cytokines that recruit microglia in the retina, a process linked to AMD progression and commonly used to assess cellular senescence.³⁹ To investigate this phenomenon, we performed fluorescent immunostaining against IBA-1 to detect microglia/macrophages in RPE/choroid flat mounts, using phalloidin to label RPE cells. In wild-type mice, IBA-1–positive signals were very low. In contrast, Pnpla2^−/− RPE exhibited a sevenfold increase in the accumulation of IBA-1–positive cells, which were localized amid disorganized RPE cells (Fig. 5). To establish a relationship between immune cells migration, inflammation, and IL-6 levels, an ELISA was performed on RPE tissue from Pnpla2^+/+ and Pnpla2^−/− mice at 3 months of age. IL-6 is a key inflammatory marker in the RPE, and elevated IL-6 levels have been reported in patients with AMD compared to controls.⁴⁰ Our results showed a sixfold increase in IL-6 levels in the RPE of Pnpla2^−/− mice (Fig. 5C), suggesting a strong link between immune cells migration and the inflammatory response in the RPE.

Figure 5.

IBA-1 positive cells in Pnpla2^−/− RPE flat mounts. RPE flat mounts of 3-month-old mice were immunostained for IBA-1 (green) and stained with phalloidin (red) and DAPI (blue). (A) Representative images of RPE regions from each WT (top) and Pnpla2^−/− mouse (bottom) per row labeled for IBA-1 (left), phalloidin and DAPI (middle), and merged (right). Scale bar: 50 µm. The ROIs were mainly acquired from areas away from one-third of the optic nerve and inside one-fourth of the limbus (see scheme of flat mount with asterisk to illustrate this area). (B) Quantification of the mean fluorescent intensity of IBA-1 immunostaining per ROI is shown. Three ROIs were chosen from one eyecup, and three eyecups per genotype were analyzed (n = 3). Statistical significance was calculated using an unpaired t-test for the two groups (WT and Pnpla2^−/− mice). **P < 0.01. (C) The plot shows the IL-6 amount in the RPE of Pnpla2^+/+ and Pnpla2^−/− mice at 3 months of age. Each data point corresponds to the average of IL-6 expressed in picograms per milligram of protein in RPE extracts from duplicate determinations per extract using three different mice per genotype (n = 3). Each individual data point represents the mean ± SD of IL-6, with statistical significance determined by an unpaired t-test. ****P < 0.0001.

These findings suggest that Pnpla2 acts as a regulator of inflammation, and its absence promotes microglia/macrophage recruitment and accumulation in the RPE. This, in turn, may drive retinal inflammation, exacerbate cellular senescence, and potentially accelerate AMD progression.

Increased ApoE Levels in Pnpla2^−/− Retinae

ApoE is normally expressed in the retina and plays critical roles in lipid regulation and inflammation; however, altered expression levels and extracellular deposition of ApoE have been associated with AMD pathology.^41,42 In aging eyes, ApoE accumulates in drusen and basal laminar deposits.⁴³ We examined and compared ApoE expression in frozen sections of Pnpla2^+/+ and Pnpla2^−/− mouse retinas, using an antibody against ApoE for staining. As illustrated in Figure 6, very low levels of ApoE were detected in the apical RPE and subretinal areas of wild-type Pnpla2^+/+ controls. In contrast, ApoE was prominently distributed in the subretinal space and Bruch's membrane of Pnpla2^−/− mice. ApoE fluorescence intensity was quantified, revealing an approximately 47-fold increase in Pnpla2^−/− mouse retinas compared to wild-type controls. These findings demonstrate an elevation of ApoE levels in retinas lacking Pnpla2, suggesting an association between Pnpla2 loss and AMD-related aging features in the retina.

Figure 6.

Accumulation of ApoE in Pnpla2^−/− mouse retinae. (A) Immunohistostaining of ApoE (green) was performed in retinal cryosections from Pnpla2^+/+ (WT) and Pnpla2^−/− mice at 3 months old. DAPI stain was used to label nuclei (blue). The differential interference contrast (DIC) imaging of a representative region (a and e) co-labeled with ApoE antibody (b and f) and DAPI (c and g) and merged (d and h) for each WT (a–d) and Pnpla2^−/− (e–h) mouse retinae is shown. BM, basal membrane. Scale bar: 20 µm. (B) Quantification of fluorescent intensity of ApoE. Three images per eyecup were acquired, and the ApoE signal was measured in three ROIs per image. Each data point represents the average ApoE intensity from three ROIs across three images per eyecup. Three independent mice per genotype were analyzed. Statistical significance between WT and Pnpla2^−/− mice was assessed using an unpaired t-test. **P < 0.01.

Loss of Pnpla2 Results in Lipid Accumulation in the RPE

Previous studies have shown that Pnpla2 deletion leads to lipid accumulation in the RPE, as well as in nonocular cells, including the liver and adipose cells.^10,13,44 We examined lipid accumulation in Pnpla2^−/− RPE using BODIPY dye, which has hydrophobic properties well suited for labeling lipids, membranes, and other lipophilic compounds. At 3 months of age, the Pnpla2^−/− RPE showed markedly increased intracellular lipid deposits—an 9.3-fold increase compared to age-matched Pnpla2^+/+ controls (Fig. 7). These findings support the role for Pnpla2 in maintaining lipid homeostasis in RPE cells.

Figure 7.

BODIPY staining in the RPE of the Pnpla2-knockout mouse line. RPE flat mounts from Pnpla2^+/+ and Pnpla2^−/− mice at 3 months of age were stained with BODIPY (green) and phalloidin (yellow). Two ROIs from each genotype are shown. One inset (red dotted rectangle) is magnified and displayed. Lipid deposits (LD) labeled by BODIPY were quantified from two ROIs per RPE flat mount using ImageJ. Each data point represents an ROI; data are presented as mean ± SD. Statistical significance was determined by an unpaired t-test. ****P < 0.0001.

Loss of Pnpla2 and Fundus Autofluorescence

Hyperreflective autofluorescent foci on FAF are commonly associated with lipid accumulation, as well as the presence of lipofuscin, drusen, or both, serving as markers of underlying retinal stress, inflammation, or degeneration.^24–30,38 To investigate the relationship between Pnpla2 expression and FAF signal intensity, we analyzed fundus photographs of 3-month-old Pnpla2^+/+ and Pnpla2^−/− mice. Given that the homozygous mice do not survive beyond 3 months of age,¹⁰ we also analyzed 7-month-old heterozygous Pnpla2^+/− mice. As shown in Figure 8A, wild type at either age showed a very low detection of fluorescent white dots. In contrast, Pnpla2^−/− mice exhibited a 5.3-fold increase in fluorescent white dots at 3 months of age, while Pnpla2^+/− mice showed a 10-fold increase. The number of fluorescent dots further increased in 7-month-old Pnpla2^+/− mice compared to both wild-type mice. Quantification revealed that 7-month-old heterozygous mice had about 49 times more hyperreflective autofluorescent foci than wild-type mice (Fig. 8B). These findings demonstrate an age-dependent inverse correlation between Pnpla2 expression and FAF signal intensity, resembling features observed in patients with retinal diseases.⁴⁵

Figure 8.

Fundus autofluorescence and ERG c-wave amplitudes of mice with Pnpla2 deficiency. (A) Representative FAF images of Pnpla2^+/+ (WT), Pnpla2^+/−, and Pnpla2^−/− mice at 3 months of age (top) and Pnpla2^+/+ and Pnpla2^+/− mice at 7 months of age (bottom) are shown. Punctate spots indicate hyperreflective autofluorescent foci. (B) Quantification of hyperreflective autofluorescent foci in Pnpla2^+/+, Pnpla2^+/−, and Pnpla2^−/− mice at 3 months of age and in Pnpla2^+/− at 7 months of age. Each data point represents the intensity of hyperreflective autofluorescence in the fundus of one eye; four eyes from individual mice were analyzed per genotype. Data are presented as bars showing the mean ± SD. Statistical analysis was performed using one-way ANOVA followed by Dunnett's multiple comparisons test. (C, D) ERG c-wave amplitudes recorded in response to stimuli ranging from 0 to 1000 cd·s/m² for each genotype and age group. Data are shown as the mean ± SD for each genotype. (C) At 3 months of age, data were collected from Pnpla2^+/+ (n = 10), Pnpla2^+/− (n = 10), and Pnpla2^−/− (n = 8) mice, analyzed using Dunnett's multiple comparison test following one-way ANOVA. (D) At 7 months, data were collected from Pnpla2^+/+ (n = 10) and Pnpla2^+/− (n = 10) mice and analyzed using an unpaired t-test (*P < 0.05, ***P < 0.001).

Pnpla2 Deletion Disrupts RPE Function

Our data reveal multiple changes in the RPE of Pnpla2^−/− mice that may compromise its function. RPE function is often evaluated by measuring ERG c-wave amplitude, which reflects the combined electrophysiological response of the RPE and Müller cells to light stimuli. The c-wave is primarily driven by the light-evoked hyperpolarization of the RPE, with a secondary contribution from Müller cells.^46,47 To investigate the role of Pnpla2 in RPE function, we recorded c-wave amplitudes in 3- and 7-month-old Pnpla2^+/+, Pnpla2^+/−, and Pnpla2^−/− mice. As shown in Figure 8C, heterozygous and homozygous mice exhibited reduced c-wave amplitudes compared to wild-type controls at 3 months of age. The c-wave amplitudes of 7-month-old heterozygous mice were further diminished relative to both wild-type and 3-month-old heterozygous counterparts (Fig. 8D). These results highlight the critical role of Pnpla2 in supporting normal RPE function. Its deletion leads to progressive functional deficits, particularly evident in heterozygous mice over time.

Discussion

Our study demonstrates that Pnpla2 deficiency accelerates cellular features of RPE aging and promotes characteristics consistent with AMD, while recognizing that some of these changes may also manifest in other retinal diseases. This conclusion is supported by several lines of evidence, including increased SA-β-gal activity, disrupted tight junctions, multinucleated RPE cells, and increased cytoplasmic expression of HMGB1—all indicators of senescence, cellular dysfunction, and compromised blood–retinal barrier integrity. Additionally, Pnpla2 deficiency leads to genomic instability (elevated Pp-γ-H2AX), inflammation (increased IBA-1–positive microglia/macrophages and IL-6 levels), and altered lipid metabolism (elevated ApoE levels). Importantly, the age-associated increase in hyperreflective foci accompanying decreased Pnpla2 expression highlights its pivotal role in driving AMD-like retinal pathology.

AMD progression is influenced by genetic variation and retinal microenvironmental changes.^48–51 Additionally, DNA damage, a hallmark of aging, contributes to organ dysfunction and genomic instability, key factors in AMD pathogenesis. Our findings align with studies showing that reduced ERCC1-XPF expression induced RPE senescence, supporting the role of DNA damage in AMD.⁵² Another hallmark of AMD is the accumulation of drusen deposits between the RPE and Bruch's membrane, as well as subretinal drusenoid deposits.⁵³ In Pnpla2-deficient mice, we observed hyperreflective autofluorescent foci (Fig. 8A) and ApoE deposition in Bruch's membrane and subretinal space (Fig. 6), resembling drusen-, pseudodrusen-, or lipofuscin-like features seen with aging and in AMD.

Lipid metabolism is central to RPE homeostasis and AMD. Lipid accumulation promotes hypoxia and cellular stress, contributing to senescence, inflammation, and impaired phagocytic capacity—hallmarks of AMD and aging.^1,26 Lipofuscin builds up in RPE cells with age and is exacerbated in AMD.² Previous studies have demonstrated that lipid accumulation induces senescence,⁵⁴ impairs RPE phagocytosis,⁵⁵ and leads to cell death.^56,57 Extending these observations, our findings demonstrate that Pnpla2 deficiency promotes ApoE deposition (Fig. 6) and lipid accumulation in RPE cells^10,13,44 (Fig. 7), emphasizing Pnpla2’s role in lipid regulation. The impact of lipid droplet accumulation on RPE function is evident in both our study and previous reports⁵⁵ yet remains underexplored in Pnpla2 knockout models. Our findings help bridge this gap, demonstrating that Pnpla2-mediated lipid regulation intersects with cellular aging.

Although earlier studies demonstrated that global Pnpla2 deletion leads to photoreceptor degeneration and that RPE-specific loss impairs outer segment degradation, our results broaden the functional spectrum of Pnpla2. We now implicate it in maintaining genomic stability and mitigating cellular senescence via lipid control. Moreover, deletion of Serpinf1 (PEDF gene), which downregulates Pnpla2, also causes RPE senescence and phagocytic defects.¹² This is consistent with studies in Caenorhabditis elegans, showing that Pnpla2 overexpression extends life span, whereas its depletion shortens it.⁵⁸ Similarly, reduced Pnpla2 in aging adipocytes correlates with elevated inflammatory cytokines.⁵⁹ These observations support a broader role for Pnpla2 in age-related tissue dysfunction.

Pnpla2 encodes PEDF-R, a multifunctional lipase involved in hydrolyzing phospholipids, triglycerides, and retinyl esters.^44,60 Its loss leads to lipid and retinosome accumulation in RPE^10,44 (Fig. 7), contributing to oxidative stress and cell death. These lipids may also have a protective role in the cell.⁵⁵ PEDF-R overexpression protects against oxidative stress in ARPE-19, while its absence can paradoxically confer resistance, suggesting complex regulatory dynamics.⁶¹ Altered PEDF-R expression, either loss or overexpression, may disrupt fatty acid release and membrane remodeling, potentially compromising long-term cellular health.

PEDF-R also exerts phospholipase A2 activity that facilitates photoreceptor outer segment degradation during RPE phagocytosis.¹³ The phagocytic deficits observed in Pnpla2-KO align with age-related phagocytic decline.⁶² By hydrolyzing lipids, PEDF-R may reduce toxic accumulation and support retinoid recycling. Its retinyl ester hydrolase activity may also facilitate retinoid recycling by converting retinal esters to retinol within retinosomes,⁴⁴ thereby decreasing the availability of bisretinoid precursors involved in the visual cycle and subsequent lipofuscin formation. While all-trans retinyl ester is a necessary component of the visual cycle's storage and recycling process, its accumulation, such as when Pnpla2 is deficient, can indirectly indicate an imbalance in the system that ultimately leads to increased all-trans retinal, the direct precursor of A2E.⁶³ Consistent with this, Pnpla2-deficient mice exhibit hyperreflective autofluorescence foci, resembling lipofuscin.

In summary, our findings establish Pnpla2 as a key regulator of RPE function, preserving genomic stability, lipid homeostasis, and inflammation control. Its deficiency triggers molecular and structural defects leading to AMD-like pathology, particularly in aging heterozygous mice. These results position PNPLA2 as a compelling target for therapeutic intervention in AMD.