Photoreceptor deletion of pyruvate dehydrogenase E1 subunit α1 induces retinal degeneration and reprograms retinal metabolism

Abstract

Rod and cone photoreceptors are among the most energy-demanding cells in the body, exhibiting a high rate of ATP consumption. Their primary energy source is glucose, which is metabolized through both glycolysis and mitochondrial pyruvate oxidative phosphorylation. The pyruvate dehydrogenase E1 subunit α1 is a critical component of the pyruvate dehydrogenase, which catalyzes the conversion of pyruvate to acetyl-CoA, thereby regulating mitochondrial pyruvate metabolism. To determine the significance of mitochondrial pyruvate metabolism in these cells, we investigated the impact of photoreceptor-specific Pdha1 deletion in the mouse retina. Rod- or cone-specific Pdha1 knockout mice at 2–5 months were used. These mice were evaluated across multiple modalities, including retinal structure and integrity (morphometry), retinal function (electroretinogram), photoreceptor ultrastructure (transmission electron microscopy), retinal metabolic profiles (mass spectrometry), gene expression (RT-PCR), and retinal stress response (glial activation analysis). Mice with rod- or cone-specific Pdha1 deletion exhibited retinal degeneration phenotype, manifested by impaired retinal morphology and light responses and significant retinal glial activation. Mechanistically, these retinas displayed profound metabolism reprogramming, evidenced by changes in key glycolysis and decreased tricarboxylic acid (TCA) cycle intermediates, carbohydrates, amino acids, nucleotides and their derivatives. This metabolic remodeling was further supported by enhanced glycolysis and decreased TCA cycle gene expression and was accompanied by impaired mitochondrial morphology. Our findings demonstrate that PDHA1 is essential for photoreceptor energy metabolism and for maintaining both their structural and functional integrity, thus highlighting the critical importance of proper mitochondrial glucose metabolism for photoreceptor health.

HIGHLIGHTS

• •Pyruvate dehydrogenase E1 subunit α1 encoded by the PDHA1 gene is the key regulator of mitochondrial pyruvate metabolism.
• •Photoreceptor-specific Pdha1 deletion in mice leads to photoreceptor degeneration and retinal glial activation.
• •Photoreceptor-specific Pdha1 deletion reprograms retinal metabolism and impairs mitochondrial morphology.
• •PDHA1 is essential for photoreceptor energy metabolism and for maintaining both their structural and functional integrity.

1. Introduction

Rod and cone photoreceptors are responsible for light sensation, vision, and visual acuity. They are among the most energy-demanding cells in the body, requiring a high rate of adenosine 5′-triphosphate (ATP) consumption [[1], [2], [3], [4], [5]]. The substantial ATP requirement is predominantly dedicated to active ion transport, a process fundamental for maintaining cellular ion homeostasis, electrical potential, phototransduction, and synaptic transmission. Significantly, metabolic impairment within these photoreceptors is increasingly recognized as a critical cause of retinal degeneration [[6], [7], [8]]. This impairment ultimately contributes to the etiology of various pathologic conditions, including retinitis pigmentosa, diabetic retinopathy, and age-related macular degeneration.

Glucose serves as the primary energy source for photoreceptors, metabolized through both glycolysis and mitochondrial oxidative phosphorylation. Notably, 80–90% of the glucose taken up is converted to lactate via aerobic glycolysis (a phenomenon known as the “Warburg effect” in the retina), while only 10–20% is metabolized through oxidative phosphorylation [9,10]. The Warburg effect is critical for maintaining photoreceptor structural and functional integrity and promoting glucose transport from the retinal pigment epithelium [[11], [12], [13]]. While photoreceptors only oxidize a small fraction of glucose, it nonetheless maintains a high dependence on oxidative phosphorylation [5,14,15]. Consequently, inhibiting this process, for example, by blocking the transport of pyruvate (a glucose metabolite) into the mitochondria in mice, significantly impairs both photoreceptor structural integrity and overall retinal function [16].

Mitochondrial pyruvate metabolism is the crucial process that links glycolysis to the tricarboxylic acid (TCA) cycle and subsequent oxidative phosphorylation within the mitochondria, generating the majority of the cell's ATP. The process starts with the transport of pyruvate from the cytosol into the mitochondrial matrix, facilitated by the mitochondrial pyruvate carrier (MPC) complex across the inner mitochondrial membrane. Inside the matrix, pyruvate is converted to acetyl-CoA, which then enters the TCA cycle for further oxidation. This decarboxylation reaction, catalyzed by the pyruvate dehydrogenase (PDH) complex, is an essential and rate-limiting step for mitochondrial pyruvate metabolism. The PDH complex, localized in the mitochondrial matrix, consists of three catalytic component enzymes: pyruvate dehydrogenase (E₁), dihydrolipoamide transacetylase (E₂), and dihydrolipoamide dehydrogenase (E₃). The pyruvate dehydrogenase is the primary site of regulation for the entire complex. The pyruvate dehydrogenase is a heterotetramer of two α and two β subunits. The X-linked PDHA1 gene codes for the α subunit found in most cells, and the PDHA2 gene codes for a testis-specific subunit involved in sperm production. Mutations in the PDHA1 gene are the leading cause of primary pyruvate dehydrogenase complex deficiency (PDCD), a potentially fatal metabolic disorder [[17], [18], [19], [20]]. This deficiency, a frequent manifestation of metabolic disease in children, is characterized by lactic acidosis and neurological disorders due to the buildup of pyruvate and its conversion to lactate. PDHA1 mutations account for approximately 80% of primary PDCD cases, resulting in primary lactic acidosis and hyperpyruvatemia [[17], [18], [19], [20]]. Mouse models with Pdha1 deficiency mimic the phenotypes in human patients with PDHA1 mutations [18,21,22].

Nonetheless, the role of mitochondrial pyruvate metabolism and the specific impact of PDHA1 deficiency in photoreceptors remains largely unexplored. This knowledge gap is primarily due to the lethality, diagnostic complexity, and clinical overlap of deficiency with other metabolic disorders. In this study, we investigated the consequences of photoreceptor-specific Pdha1 deletion in the mouse retina. Specific Pdha1 deletion in rod and cone photoreceptors caused significant defects, including structural and functional deterioration, mitochondrial morphology impairment, metabolomic alterations (which were characterized by alterations in key intermediates of the TCA cycles and glycolysis), altered expression of glycolysis and TCA cycle genes, and retinal glial activation. This investigation is the first to examine the role of PDHA1 in photoreceptors. Our findings demonstrate that PDHA1 is essential for photoreceptor energy metabolism and for maintaining photoreceptor structural and functional integrity, highlighting that proper mitochondrial glucose metabolism is critical for photoreceptor normality and health.

2. Methods

2.1. Mice and reagents

The C57BL/6J mouse line was obtained from The Jackson Laboratory. The Pdha1^fl/fl line was purchased from Jackson Laboratory (https://www.jax.org/strain/017443) [18], the cone-specific Cre (Hrgp^Cre) and the rod-specific Cre (LMOP^Cre) lines were generated as reported previously [23,24]. Mice expressing LMOP^Cre or Hrgp^Cre exhibit no retinal abnormalities for up to the maximum age studied (8 months). These mice maintain normal electroretinography (ERG) responses, morphology (including ONL thickness), and cone density/distribution, rendering them indistinguishable from wild-type mice [[23], [24], [25]]. Consequently, these strains have been widely utilized in studies requiring photoreceptor-specific gene deletion [[26], [27], [28], [29], [30], [31]]. The Pdha1^fl/fl/Hrgp^Cre and Pdha1^fl/fl/LMOP^Cre lines were generated by crossbreeding. Experimental animals were selected based on genotyping with PCR diagnosis [23,24]. The Rd1 and Rd8 mutation, originally on an FVB/6N-derived transgenic mouse, was screened following previously described [32,33]. Mice were housed under a 12-hour light/dark cycle with cage illumination at 7 foot-candles during the light phase. All animal procedures were approved by the Institutional Animal Care and Use Committee at the University of Oklahoma Health Campus and adhered to the guidelines of the Society for Neuroscience and the Association for Research in Vision and Ophthalmology. Both male and female mice were used in experiments, with littermates randomly assigned to experimental groups. Antibodies and reagents used in the experiments are listed in Supplementary Table 1.

2.2. Eye preparation, immunofluorescence labeling, confocal microscopy, and morphometric analysis

Retinal whole mounts or cross sections were prepared for immunofluorescence labeling, as described previously [34], [35], [36]. For retinal whole-mount preparations, eyes were enucleated, marked at the superior pole with a green dye, and fixed in 4% paraformaldehyde (PFA, Thermo Scientific) for 30 min at room temperature, followed by removal of the cornea and lens. The eyes were then fixed in 4% PFA for 4–6 h at room temperature, retinas were isolated, and the superior portion was marked for orientation with a small cut. For retinal cross sections, mouse eyes were enucleated (the superior portion of the cornea was marked with green dye before enucleation) and fixed in Prefer (Anatech) for 3–4 h at room temperature before being transferred into 70% ethanol. Paraffin sections (5 μm thickness) passing vertically through the retina (along the vertical meridian passing through the optic nerve head) were prepared using a microtome (Leica Biosystems). Cryosections were cut at a thickness of 10 µm using a Cryostar NX70 cryostat (Thermo Scientific).

Immunofluorescence labeling from either paraffin-embedded or cryosections was performed as described previously [34,35,37]. Briefly, retinal whole mounts were blocked with HBSS containing 5% bovine serum albumin (BSA) and 0.5% Triton X-100 overnight at 4 °C. PNA immunohistochemistry was performed using biotinylated PNA and then streptavidin-Cy3 at room temperature for 1 h. For immunofluorescence staining on sections, after deparaffinization and rehydration steps, antigen retrieval was performed in 10 mM sodium citrate buffer, pH 6.0, in a 70 °C water bath for 30 min. After blocking with 10% FBS containing 0.5% Triton X-100 for 2 h, primary antibody was performed overnight at 4 °C. Slides were mounted and cover-slipped after fluorescence-conjugated secondary antibody incubation and wash steps.

Immunofluorescence was imaged using a confocal laser-scanning microscope (FV1000, Olympus) and FluoView imaging software (Olympus). Evaluation of GFAP fluorescence density was conducted as described previously [38]. Briefly, images from the central, middle, and peripheral retinal regions were taken, and confocal images of 10 layers of each region were stacked with the z-stack function in the ImageJ software (https://imagej.nih.gov/ij/) to obtain a maximal immunofluorescence density. Fluorescence density levels of the immunolabeling in the center, middle, and peripheral regions were measured after the removal of the background, and the average fluorescence density from the three regions was used for statistical analysis. For retinal morphometric analysis, retinal cross sections stained with hematoxylin and eosin (H&E) were used to evaluate outer nuclear layer (ONL) thickness/rod survival, as described previously [34,35].

2.3. Scotopic and photopic ERG recordings

Full-field ERG recordings were conducted as described previously [35,38]. Briefly, after overnight dark adaptation, mice were anesthetized by intraperitoneal injection of 85 mg/kg ketamine and 14 mg/kg xylazine. ERGs were recorded using the Espion Visual Electrophysiology System (Diagnosys) with the ColorDome Advanced Performance Ganzfeld Dome system (Diagnosys). Potentials were recorded using a gold-wire electrode to contact the corneal surface through a layer of 2.5% hypromellose (Gonak, Akorn Pharmaceuticals). Bilateral gold electrodes were placed on the corneal surface to record the light-induced retinal potentials. The reference and ground electrodes were placed subcutaneously in the forehead and tail, respectively. For the assessment of scotopic responses, a stimulus intensity of 1.89 log cd · s m−2 was presented to dark-adapted dilated mouse eyes. To evaluate photopic responses, mice were adapted to a 1.48 log cd · s m−2 light for 7 min, and then a light intensity of 1.89 log cd · s m−2 was given. Responses were differentially amplified, averaged, and analyzed using Espion 100 software (Diagnosys).

2.4. Metabolomics analysis

Metabolites were extracted from 10 μl of medium samples prepared from frozen retinas, dried with frozen dryer, and analyzed with liquid chromatography mass spectrometry (LC MS) and gas chromatography mass spectrometry (GC MS), as described previously [39,40]. An ACQUITY UPLC BEH Amide analytic column (2.1 × 50 mm, 1.7 μm, Waters) was used for chromatographic separation. A Shimadzu LC Nexera X2 UHPLC coupled with a QTRAP 5500 LC MS (AB Sciex, Hong Kong), and an Agilent 7890B/5977 B GC MS (Agilent Technologies, Santa Clara, CA. USA) were used for metabolite analysis. Protein concentrations from extraction pellets were measured for data normalization [41]. A total of 185 metabolites that cover major metabolic pathways were quantified. The data were analyzed by MultiQuant 3.0.2 (AB Sciex) and Agilent MassHunter Quantitative Analysis Software [42]. Multivariate analysis was performed with a supervised classification model partial least-squares discriminant analysis (PLS-DA) using MetaboAnalyst 5.0. The comparison of specific metabolites was analyzed with Volcano plot with a P of less than 0.05 and fold changes of more than 1.2 or less than −1.2 for all figures which were generated from website (https://huygens.science.uva.nl/VolcaNoseR/) and the heatmaps were generated from website (http://www.heatmapper.ca/expression/).

2.5. PCR array

Total RNA preparation from mouse retina and reverse transcription were performed as described previously [43]. Mouse Glucose Metabolism RT2 Profiler™ PCR Array Kit (Qiagen, PAMM-006ZD, Lot# CE001667) was used to analyze gene expression. The resulting data were submitted to the Qiagen web server (https://geneglobe.qiagen.com/us/my-geneglobe/analysis) for gene expression analysis.

2.6. Transmission electron microscopy

Tissue fixation and embedding for transmission electron microscopy (TEM) were performed as previously described [44]. Mitochondrial diameter measurements were conducted using Viking software. Mitochondrial measurements were conducted using Viking software. For each mitochondrion, three independent diameter measurements were taken along its length and averaged to generate a single representative value. Measurements were originally collected in micrometers (μm) in Viking and subsequently converted to nanometers (nm) for improved numerical clarity and standardization. For each mouse (n = 4 per genotype), 30 mitochondria were analyzed, totaling 120 mitochondria per genotype. These values were used for both group-level analysis (individual mitochondria) and mouse-level analysis (average diameter per mouse). Mitochondrial diameter variance was also assessed by measuring the smallest and largest diameters within a single mitochondrion and calculating the difference between them. Variance analysis was conducted at the group level only. The same EM images were used to evaluate the connecting cilium structure in WT and Pdha1 deletion photoreceptors.

2.7. Statistical analysis

The results are expressed as the mean ± SD of the number of mice or the number of assays. One-way ANOVA was used for significance within sets of data, followed by Dunnett's multiple-comparisons test. Unpaired Student's t-test was used for differences between two groups of data. Differences were considered statistically significant at p < 0.05. Data were analyzed and graphed using GraphPad Prism 10.5 (GraphPad Software, San Diego, CA). In the EM analysis, for group-level analysis, all individual mitochondria were treated as independent data points. For mouse-level analysis, the average mitochondrial diameter from each mouse was used.

3. Results

3.1. Rod and cone degeneration in Pdha1 deletion mice

We generated mouse lines with rod-specific deletion of Pdha1 (Pdha1^fl/+/LMOP^Cre and Pdha1^fl/fl/LMOP^Cre) and cone-specific deletion of Pdha1 (Pdha1^fl/+/Hrgp^Cre and Pdha1^fl/fl/Hrgp^Cre) to evaluate the impact of PDHA1 deficiency. Mice of both genders were used in this evaluation. For this X-linked gene, the heterozygous group consisted of females only. The homozygous group included both hemizygous males and homozygous females. We first assessed retinal structure and photoreceptor integrity in these mice using morphometric analyses and cone density evaluations. Morphometric analysis of H&E-stained retinal sections from 3-month-old rod-specific Pdha1 deletion mice revealed a significant reduction in outer nuclear layer (ONL) thickness in these mice compared to wild-type (WT) controls (Figure 1A–B). There was no significant difference in ONL thickness between the heterozygous and homozygous mice (Figure 1A–B).

Figure 1.

Rod and cone degeneration in Pdha1 deletion mice. Pdha1^fl/+/LMOP^Cre, Pdha1^fl/fl/LMOP^Cre, Pdha1^fl/+/Hrgp^Cre, Pdha1^fl/fl/Hrgp^Cre, and WT mice were evaluated for retinal integrity by morphometric analysis. A-B. Shown are representative microscopic images of H&E-stained retinal sections and corresponding quantitative analysis of ONL thickness in mice at 3 months. C–F. Shown are representative confocal images of PNA labeling on retinal whole mounts and corresponding quantitative analysis of PNA-positive cells in mice at 3 (C-D) and 5 (E-F) months. ONL, outer nuclear layer; INL, Inner nuclear layer; GCL, ganglion cell layer. Data are represented as the mean ± SD for 3–13 mice per group. Data were analyzed by one-way ANOVA, followed by Dunnett's multiple-comparisons test for D and F, and unpaired Student's t-test for B (∗p < 0.05, ∗∗∗p < 0.001).

We evaluated the cone density in retinal whole mounts prepared from cone-specific Pdha1 deletion mice at 3 and 5 months using peanut agglutinin (PNA) staining. This analysis demonstrated a significant decrease in cone density across both the dorsal and ventral retinal regions of the Pdha1 deletion mice when compared to WT controls (Figure 1C–F). Of note, the cone density in the dorsal region of 3-month-old heterozygous mice was not statistically different from WT controls (Figure 1D), but it was significantly reduced at 5 months of age (Figure 1F).

PDHA1 expression in the retina was examined by immunofluorescence labeling using an antibody against mouse PDHA1. The signal was detected throughout all retinal cell layers. In the photoreceptor layer, PDHA1 labeling was primarily observed in the inner segment (IS) and outer plexiform layer (OPL), with some perinuclear staining, which is consistent with its known mitochondrial localization. This immunofluorescence signal was drastically reduced in the photoreceptor layer of Pdha1^fl/fl/LMOP^Cre mice. Conversely, labeling in Pdha1^fl/fl/Hrgp^Cre mice was comparable to that in WT mice, which is consistent with the sparse number of cones in the retina (Fig S1).

3.2. Reduced retinal light response in Pdha1 deletion mice

To evaluate the functional consequences of Pdha1 deletion, we performed electroretinography (ERG) analysis on the mice. In rod-specific Pdha1 deletion heterozygous mice, both scotopic a- and b-wave amplitudes were significantly reduced at 2 and 4 months compared to age-matched WT controls, while photopic b-wave amplitudes remained unaffected (Figure 2A–B). The homozygous mice showed a comparable reduction in scotopic a- and b-wave amplitudes to that observed in the heterozygous mice (Figure 2A–B). In cone-specific Pdha1 deletion mice, photopic b-wave amplitudes were significantly reduced in both heterozygous and homozygous mice aged 4 months compared to age-matched WT controls, but this effect was not observed in 2-month-old mice (Figure 2C–D). The scotopic a- and b-wave amplitudes were initially unaffected at 2 months of age. However, the scotopic a-wave amplitudes became significantly reduced in the cone-specific Pdha1 deletion mice at 4 months old (Figure 2C–D).

Figure 2.

Reduced retinal light response in Pdha1 deletion mice.Pdha1^fl/+/LMOP^Cre, Pdha1^fl/fl/LMOP^Cre, Pdha1^fl/+/Hrgp^Cre, Pdha1^fl/fl/Hrgp^Cre, and WT mice were evaluated for retinal function using ERG. A-B. Shown are quantitative results of scotopic and photopic amplitudes in rod-specific Pdha1 deletion mice at 2 and 4 months. C-D. Shown are quantitative results of scotopic and photopic amplitudes in cone-specific Pdha1 deletion mice at 2 and 4 months. Data are represented as the mean ± SD for 4–15 mice per group. Data were analyzed by one-way ANOVA, followed by Dunnett's multiple-comparisons test for B-D, and unpaired Student's t-test for A (∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001).

3.3. Altered retinal metabolomic profiles in Pdha1 deletion mice

To study the impact of Pdha1 deletion on retinal metabolism, we analyzed the abundance of metabolites in the retinas. Due to the rod-dominant nature of murine retina (cones represent only 3–5% of the total photoreceptor population), we focused on rods for metabolomic studies. Because previous studies suggest sex differences in retinal metabolism [45], we therefore used female mice for the metabolomic study. This choice also allowed us to analyze the heterozygous Pdha1 deletion, a condition unique to females given the gene's X-linked nature. Retinas of female rod-specific Pdha1 deletion and WT mice aged 2–4 months were collected and subjected to liquid chromatography mass spectrometry (LC MS) and gas chromatography mass spectrometry (GC MS) analysis, followed by multivariate analyses with partial least squares discriminant analysis (PLS-DA).

The metabolite analyses revealed a clear separation between heterozygous or homozygous Pdha1 deletion and WT mice (Figure 3A, B, F, G). Heat maps further demonstrated distinct clustering patterns separating the WT mice from Pdha1 deletion mice (Figure 3C, H). The significantly altered metabolites were categorized into several groups, including glycolysis, the TCA cycle, amino acids and their derivatives, carbohydrates and their derivatives, cofactors/vitamins, nucleotides and nucleosides, organic acids and their derivatives, and miscellaneous groups (Figure 3D–E, I-J). Specifically, several metabolites in glycolysis were increased in Pdha1 deletion mice, including lactate, pyruvate, phosphoenolpyruvic acid, xylulose-5-phosphate, dihydroxyacetone phosphate, phosphocreatine, xylulose 5-phosphate, glyceraldehyde 3-phosphate, and d-ribulose 5-phosphate (Figure 3D, I). Many amino acid derivatives that are associated with the TCA cycle, including aspartate, oxalic acid, isocitrate, l-argininosuccinic acid, l-pyroglutamic acid, pipecolic acid, l-proline, l-selenomethionine, and l-valine, were increased in Pdha1 deletion mice, while α-ketoglutarate and itaconic acid were reduced (Figure 3D,I). Notably, the heterozygous mice showed more pronounced metabolite alterations than the homozygous mice. Specifically, there were 42 and 14 significantly altered metabolites in the heterozygous and homozygous mice, respectively, when compared to WT mice, and only 5 metabolites showed changes common to both genotypes (Figure 3C, H).

Figure 3.

Altered retinal metabolomic profiles in Pdha1 deletion mice. Female rod-specific Pdha1 deletion (Pdha1^fl/+/LMOP^Cre and Pdha1^fl/fl/LMOP^Cre) and WT mice at 3 months were evaluated for metabolomic profiles using LC MS and GC MS. A-E. Altered retinal metabolomic profiles in heterozygous Pdha1 deletion mice. F-J. Altered retinal metabolomic profiles in homozygous Pdha1 deletion mice. A, F. PLS-DA plots of retinal metabolomic profiles in heterozygous (A) and homozygous (F) Pdha1 deletion mice, relative to WT controls. B, G. Volcano plot of retinal metabolites altered in heterozygous (B) and homozygous (G) Pdha1 deletion mice, relative to WT controls. C, H. Heatmap of differentially expressed metabolites in heterozygous (C) and homozygous (H) Pdha1 deletion mice, relative to WT controls. D, I. Fold change of retinal metabolites of glycolysis and TCA cycle in heterozygous (D) and homozygous (I) Pdha1 deletion mice, relative to WT controls. E, J. Fold change of retinal metabolites of amino acids and their derivatives, carbohydrates and their derivatives, cofactors/vitamins, nucleotides and nucleosides, and organic acids and their derivatives in heterozygous (E) and homozygous (J) Pdha1 deletion mice, relative to WT controls.

We further compared metabolomic differences between the heterozygous and female homozygous mice. Principal component analysis (PCA) partially distinguished the heterozygous from the homozygous retinas (Fig S2A) and identified 22 significantly altered metabolites between these two groups (Fig S2B-C). Specifically, the TCA cycle metabolites isocitrate, malate, and oxaloacetate were significantly increased in the heterozygous mice compared to the homozygous mice (Fig S2B-C).

In a separate experiment, we analyzed metabolite alterations in the retinas of female homozygous and male hemizygous mice to evaluate the effects of Pdha1 deficiency across different genders. PCA revealed a metabolite separation of homozygous from hemizygous mice (Fig S3A). This analysis identified 10 significantly altered metabolites when comparing the hemizygous to the homozygous deletion mice (Fig S3B-C). Of the 10 metabolites, 8 were increased and 2 were decreased in the hemizygous mice. Specifically, the TCA cycle metabolites isocitrate and aspartate were significantly increased in the hemizygous mice compared to the homozygous mice.

3.4. Altered expression of glucose metabolic genes in Pdha1 deletion mice

To evaluate the impact of Pdha1 deletion on retinal glucose metabolism at the transcriptional level, we analyzed the gene expression profiles in Pdha1 deletion mice. Retinas were collected from 3-month-old female Pdha1^fl/+/LMOP^Cre, Pdha1^fl/fl/LMOP^Cre, and WT mice, prepared for mRNA, and subjected to a PCR array using the Mouse Glucose Metabolism RT² Profiler™ PCR Array kit. These analyses revealed that Pdha1 deletion induced widespread, pronounced alterations in the expression of genes involved in glucose metabolism, particularly within the glycolysis and TCA cycle pathways. The homozygous mice exhibited broader, more extensive changes compared to the heterozygous mice (Figure 4A). Of the 84 genes analyzed, 19 (23%) were significantly altered in the heterozygous mice, while a substantially higher number, 45 (54%), were altered in the homozygous mice (Figure 4A, B–C). Furthermore, the homozygous mice showed greater magnitudes of change; upregulated genes exhibited higher fold increases and downregulated genes displayed stronger suppression compared to the heterozygous group (Figure 4D). Among the differentially expressed genes (DEGs), 7 of 19 genes in the heterozygous group and 25 of the 45 genes in the homozygous group were associated with glycolysis and the TCA cycle (Figure 4, B-D). The examples of these glycolysis and TCA cycle genes include Aldob, Pgk2, Gck, Hk3, Pklr, Pgam2, and Mdh1b. The remaining 20 DEGs in the homozygous group were related to other metabolism pathways, such as the pentose phosphate pathway, glucose metabolism regulation, gluconeogenesis, and glycogen synthesis/degradation (Figure 4C–D).

Figure 4.

Altered expression of glucose metabolic genes in Pdha1 deletion mice. Female rod-specific Pdha1 deletion (Pdha1^fl/+/LMOP^Cre and Pdha1^fl/fl/LMOP^Cre) and WT mice at 3 months were evaluated for expression of glucose metabolism genes using the mouse glucose metabolism RT² Profiler™ Array. A. Heatmap of differentially expressed genes in heterozygous and homozygous mice relative to WT controls. B–C. Volcano plot of glucose metabolism gene expression in heterozygous (B) and homozygous (C) mice relative to WT controls. Significantly differentially expressed glycolysis and TCA genes are highlighted. D. Comparison of significantly differentially expressed genes in heterozygous and homozygous mice relative to WT controls.

3.5. Impaired photoreceptor mitochondrial morphology in Pdha1 deletion mice

To assess whether Pdha1 deletion affects photoreceptor ultrastructure, we conducted transmission electron microscopy (EM) on retinal sections prepared from Pdha1^fl/fl/LMOP^Cre and WT mice. The overall appearance of the rod photoreceptors in the Pdha1^fl/fl/LMOP^Cre mice was largely normal. Specifically, we evaluated the shape and diameter of the mitochondria. These analyses revealed a significant difference in mitochondrial diameter between the two genotypes. On average, mitochondrial diameter was greater in Pdha1^fl/fl/LMOP^Cre mice than in the WT controls (Figure 5A–B). Quantitative analysis confirmed this finding, showing a significant increase in both individual and mouse-averaged mitochondrial diameters in Pdha1^fl/fl/LMOP^Cre rods (Figure 5C, left and middle panels) compared to the WT controls. Conversely, the variation in diameter did not differ significantly between the two genotypes (Figure 5C, right panel). In this work, we also examined the connecting cilium integrity and did not find evidence showing connecting cilium structure being impaired (Fig S4).

Figure 5.

Impaired photoreceptor mitochondrial morphology in Pdha1 deletion mice.Pdha1^fl/fl/LMOP^Cre, Pdha1^fl/+/LMOP^Cre (male), and WT mice at 3 months were evaluated for photoreceptor ultrastructure using transmission EM. A-B. EM images of photoreceptor mitochondria in WT and Pdha1 mutant mice. Images in A and B are shown at different scales. Scale bars: 0.5 μm. C. Quantification of mitochondrial diameter and diameter variation. Left: Individual mitochondrial diameters are significantly increased in Pdha1 mutant photoreceptors (p < 0.0001). Middle: Mouse-averaged mitochondrial diameters are significantly higher in Pdha1 mutant mice (p = 0.0424). Right: Diameter variation is not significantly different between genotypes (p = 0.7633). Data are presented as individual mitochondria (dots, n = 120 per group), individual mice (color, n = 4 per group), mean, and standard deviation. Blue and purple indicate female mice; orange and green indicate male mice. Two-tailed Student's t-tests were performed for both individual and mouse-level comparisons.

3.6. Retinal glial activation in Pdha1 deletion mice

The presence and upregulation of glial fibrillary acidic protein (GFAP) in the intermediate filaments of retinal glial cells (Müller cells and astrocytes) are widely used as a sensitive and reliable marker for retinal glial response following injury, stress, or disease. In this study, we examined the effects of Pdha1 deletion on retinal glial activation. Retinal cross-sections prepared from 2-month-old WT, Pdha1^fl/fl/LMOP^Cre and male Pdha1^fl/+/LMOP^Cre, and Pdha1^fl/fl/Hrgp^Cre mice were analyzed for expression of GFAP by immunofluorescence labeling. This assay showed that the immunofluorescence labeling of GFAP increased in Pdha1 deletion mice compared to WT controls (Figure 6).

Figure 6.

Retinal glial activation in Pdha1 deletion mice.Pdha1^fl/+/LMOP^Cre, Pdha1^fl/fl/LMOP^Cre, Pdha1^fl/fl/Hrgp^Cre, and WT mice at 3 months were evaluated for retinal glial cell activation by GFAP staining. A-B. Shown are representative confocal images of GFAP labeling on retinal sections and corresponding quantitative analysis in rod-specific Pdha1 deletion mice. C-D. Shown are representative confocal images of GFAP labeling on retinal sections and corresponding quantitative analysis in cone-specific Pdha1 deletion mice. ONL, outer nuclear layer; INL, Inner nuclear layer; GCL, ganglion cell layer. Data are represented as the mean ± SD for 5–10 mice per group. Data were analyzed by one-way ANOVA, followed by Dunnett's multiple-comparisons test for B, and by unpaired Student's t-test for D (∗∗p < 0.01, ∗∗∗p < 0.001).

4. Discussion

4.1. Pdha1 is required for photoreceptor structural and functional integrity

This work reveals several characteristics of the mouse photoreceptor degeneration phenotype resulting from Pdha1 deletion. First, both rods and cones are sensitive to Pdha1 deletion, exhibiting comparable degeneration phenotypes. This observation highlights the importance of the PDHA1/pyruvate mitochondrial metabolism pathway in both photoreceptor types [14]. Second, the heterozygous mice also exhibit a retinal degeneration phenotype. This is consistent with the haplo-insufficient nature of the Pdha1 gene, and mirrors clinical observations in patients carrying heterozygous mutations [46]. Third, male (Pdha1 deletion hemizygous) and female (Pdha1 deletion homozygous) deficient mice show comparable retinal degeneration, which is in line with the X-linked nature. Fourth, the heterozygous mice show comparable photoreceptor degeneration phenotype to that in the homozygous/hemizygous mice. It is worth noting that in human, the phenotype in heterozygous females is highly variable, ranging from completely asymptomatic to a severe presentation similar to homozygous/hemizygous deficiency [47]. Fifth, photoreceptor-specific Pdha1 deletion induces retinal glial activation, as manifested by increased GFAP expression. At the ultrastructural level, Pdha1-deficient rods show an altered mitochondrial appearance, characterized by an increased organelle diameter. This increased mitochondrial diameter, often referred to as mitochondrial swelling or the formation of megamitochondria, is generally interpreted as a morphological sign of cellular stress and dysfunction [48,49].

The retinal phenotype observed in Pdha1 deletion mice is comparable to that reported in both Mpc knockout mice and Idh3a mutant mice, exhibiting an early onset, moderately progressive degeneration [16,50]. Idh3a codes for the α subunit of isocitrate dehydrogenase, which catalyzes the conversion of isocitrate to α-ketoglutarate in the TCA cycle. In contrast, the retinal phenotype in Pdha1 deletion mice is less severe than that observed in zebrafish mutants carrying naturally occurring PDH complex mutations. Specifically, the zebrafish Pdhb (coding for the E1b subunit) mutant [51] and the Dlat (coding for the PDHE2 subunit) mutant [52] are blind and exhibit a general loss of light perception. The pronounced retinal degeneration phenotype in these mutant fish has been attributed to the enzyme complex deficiency observed across all retinal cell types (i.e., global mutants).

4.2. Pdha1 is required for photoreceptor metabolic homeostasis

MS analysis reveals that photoreceptor-specific Pdha1 deletion disrupts metabolic homeostasis, inducing metabolic reprogramming. This is characterized by significant changes in many metabolites, similar to the pattern observed in Mpc knockout mice [16]. Specifically, glycolysis and TCA cycle metabolites are altered. Pdha1 deletion significantly elevates l-aspartic acid, asparagine, and N-acetylaspartic acid in the retina, while reducing α-ketoglutarate and itaconic acid. This shift likely stems from the disruption of pyruvate oxidation, resulting in aspartate accumulation and NADH/NAD⁺ imbalance [53]. Furthermore, the significant elevation of proline and its derivative 4-hydroxyproline reflects a metabolic shift toward glutamine-based anaplerosis and altered redox homeostasis in response to mitochondrial dysfunction [54]. Our finding of elevated proline levels in Pdha1 deletion mice aligns with the proposed use of the combined alanine-to-leucine and proline-to-leucine ratios as a screen for PDH complex deficiencies [55]. The analysis of the male and female mice suggests a more profound impact in the male Pdha1 deletion mice than in the female Pdha1 deletion mice.

Notably, the metabolomic alteration is more profound in the heterozygous mice than in the homozygous mice. This finding presents a significant paradox that currently lacks a clear explanation. The differential response could be explained by the following possibilities. 1) It may be related to compensatory reprogramming in homozygous retinas. The complete absence of PDH activity likely compels these cells to fully abandon pyruvate oxidation, rapidly initiating a robust and effective alternative metabolic pathway. 2) Partial PDH activity in heterozygous retinas may create chronic metabolic instability, where cells cannot efficiently use normal pathways but also fail to fully activate compensatory mechanisms, resulting in metabolic stress and pathway imbalance.

The metabolic remodeling is further supported by gene expression analysis. Pdha1 deletion induces downregulation of TCA cycle genes and upregulation of glycolysis genes. Furthermore, a significant number of genes involved in pentose phosphorylation and gluconeogenesis are also upregulated. These observations confirm the metabolic remodeling at the transcriptional level. Noteworthy, despite the differences in metabolomic profiles, homozygous Pdha1 deletion mice exhibit a greater number of dysregulated genes with higher fold changes in expression than the heterozygous mice. We propose that homozygous mice may develop stronger transcriptional changes as a compensatory response to the metabolic disruption. Consequently, the metabolic network may reach a new steady state, resulting in fewer detectable metabolite changes due to feedback regulatory mechanisms. In contrast, heterozygous mice, retaining partial PDH activity, may sustain greater metabolic flexibility, resulting in more diverse but less extreme gene expression alterations.

4.3. PDHA1 deficiency and ocular abnormalities in human

PDHA1 mutations account for approximately 80% of primary pyruvate dehydrogenase complex deficiency (PDCD) cases [[17], [18], [19], [20]]. PDCD is a rare, inherited metabolic disorder characterized by lactic acidosis, severe neurological dysfunction, and developmental delays. Onset can be prenatal or in infancy, with severe, often fatal cases, or later in childhood, which may be less severe than the early onset cases. Symptoms vary from fatal neonatal lactic acidosis to childhood onset involving seizures, hypotonia, ataxia, and developmental delay. Ocular abnormalities have been documented in patients with PDCD. The phenotypes include optic atrophy, cortical visual impairment, strabismus, abnormal ERG, pigmentary retinopathy, and ptosis [17,20,56]. PDCD is a primary genetic cause of Leigh syndrome, a rare and severe neurodegenerative mitochondrial disorder that typically manifests in early childhood [[57], [58], [59]]. Leigh syndrome is characterized by significant genetic heterogeneity, with mutations identified in over 75 different genes [60]. Among these, PDCD stands out as one of the most frequent contributors, accounting for approximately 10%–25% of all diagnosed cases [55,60]. The primary clinical manifestations of Leigh syndrome are often age-dependent, most commonly presenting as seizures, abnormal movements, and various ocular phenotypes [[61], [62], [63]]. Nonetheless, our understanding of the specific retinal and photoreceptor phenotypes associated with PDCD in humans remains limited. This knowledge gap stems primarily from the condition's early onset, diagnostic complexity, and its clinical overlap with other metabolic disorders.

4.4. Mitochondrial pyruvate metabolism deficiency in retinal disease

Accumulating evidence suggests that a deficiency in mitochondrial pyruvate metabolism is implicated in the progression of retinal degeneration across various pathological conditions. Mitochondrial pyruvate metabolism deficiency has been implicated in neovascularization retinal disease. The ischemia/reperfusion injury model is used to mimic retinal disease involving ischemic damage, such as diabetic retinopathy, retinopathy of prematurity, retinal vein occlusion, ischemic optic neuropathy, and glaucoma. Studies in rat and mouse ischemia/reperfusion injury models have demonstrated an elevation of lactate and pyruvate and a corresponding reduction in ATP in the retina [64,65]. Importantly, administration of a PDH kinase (PDK) inhibitor, which increases PDH activity, prevents retinal cell death and improves energy metabolism in these animals [64,65]. PDH activity is regulated by PDK, which phosphorylates and thereby suppresses the enzyme. Consequently, PDK inhibition is a common approach used to enhance PDH function and promote mitochondrial pyruvate metabolism. Furthermore, metabolic reprogramming, characterized by elevated lactate levels, is observed in the sera of both patients with neovascular age-related macular degeneration and mice subjected to laser-induced choroidal neovascularization, and treating the mouse model with a PDK inhibitor normalizes lactate levels and inhibits choroidal neovascularization formation [66].

Mitochondrial metabolism deficiency has also been implicated in autosomal recessive retinitis pigmentosa (arRP). Analysis of protein expression in liquid vitreous biopsies from arRP patients (with PDE6A mutations) and from arRP model mice (with Pde6α mutations) suggests affected molecular pathways involving mitochondrial metabolism [67]. Accordingly, treatment of arRP mice with metabolites, such as α-ketoglutarate or a ketogenic diet provides neuroprotection, delays photoreceptor cell loss, and enhances visual function [67]. Metabolic profiling of the neural retina in arRP mouse models (with Rhodopsin or Pde6α mutations) also reveals that TCA cycle intermediates are reduced during the course of the disease [68]. Accordingly, supplementation with metabolites that feed into the TCA cycle, including citrate, ɑ-ketoglutarate, glutamine, or succinate, prolongs photoreceptor cell survival, increases resilience against cell death, and prolongs visual function in these mice [68].

Importantly, PDH deficiency has been implicated in the pathogenesis of the retinal cancerous growth or retinoblastoma [69]. Tumor tissues from patients display strongly increased expression of PDK. Treatment with PDK inhibitors results in reduced cell growth and migration, and increased cancer cell death in patient-derived retinoblastoma cells.

5. Conclusion

This study investigates the impact of photoreceptor-specific Pdha1 deletion in mice. Deleting Pdha1 specifically in rods or cones leads to corresponding photoreceptor degeneration, mitochondrial malformation, metabolic reprogramming, and retinal glia activation. These findings demonstrate the necessity of the PDHA1 enzyme and mitochondrial pyruvate metabolism for maintaining photoreceptor and overall retinal health.

CRediT authorship contribution statement

Hongwei Ma: Writing – review & editing, Writing – original draft, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Lilliana R. York: Writing – review & editing, Methodology, Investigation, Data curation. Shujuan Li: Writing – review & editing, Methodology, Investigation, Data curation. Grayson Gagnon: Writing – review & editing, Methodology, Investigation, Data curation. Junhuang Zou: Writing – review & editing, Methodology, Investigation, Data curation. Haoran Yu: Writing – review & editing, Methodology, Investigation, Data curation. Jun Yang: Writing – review & editing, Supervision, Methodology, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization. Yun Le: Writing – review & editing, Resources, Investigation, Conceptualization. Mark Eminhizer: Writing – review & editing, Methodology, Investigation, Data curation. Isabella Mascari: Writing – review & editing, Methodology, Investigation, Data curation. Jianhai Du: Writing – review & editing, Supervision, Methodology, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization. Xi-Qin Ding: Writing – review & editing, Supervision, Methodology, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization.

Funding sources

This work was supported by multiple grants from NIH, including R01EY033841 (XQD), R01EY034524 and R01EY036877 (JY), and EY026030, EY031324, and EY032462 (JD). Additional NIH support was provided by core grants P30EY021725, P30GM122744, P30EY014800, and P20GM144230. Further funding was provided by the Presbyterian Health Foundation (XQD), the Retina Research Foundation (J.D.), and unrestricted challenge grants from Research to Prevent Blindness (RPB) to the Ophthalmology departments at WVU and the University of Utah.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following is the Supplementary data to this article.

Data availability

Data will be made available on request.