Metabolic reprogramming of the neovascular niche promotes regenerative angiogenesis in proliferative retinopathy

Cagnone, Gael; Pundir, Sheetal; Betus, Charlotte; Agnihotri, Tapan; Ren, Anli; -Kim, Jin Sung; Harvey, Noémie-Rose; Heckel, Emilie; Chen, Mei Xi; Situ, Anu; Gaub, Perrine; Kim, Nicholas; Das, Ashim; Leclerc, Severine; Wünnemann, Florian; Berillon, Louis; Andelfinger, Gregor; Crespo-Garcia, Sergio; Dubrac, Alexandre; Rezende, Flavio A.; Clish, Clary B.; Maranda, Bruno; Rivera, José Carlos; Smith, Lois E. H.; Sapieha, Przemyslaw; Joyal, Jean-Sébastien

doi:10.1038/s41467-025-60061-4

Download PDF

Article
Open access
Published: 25 June 2025

Metabolic reprogramming of the neovascular niche promotes regenerative angiogenesis in proliferative retinopathy

Gael Cagnone^1,2^na1,
Sheetal Pundir^2,3^na1,
Charlotte Betus^1,2,
Tapan Agnihotri^2,3,
Anli Ren^2,4,
Jin Sung -Kim ORCID: orcid.org/0000-0002-7823-3563^2,3,
Noémie-Rose Harvey⁵,
Emilie Heckel^1,2,
Mei Xi Chen²,
Anu Situ²,
Perrine Gaub²,
Nicholas Kim^2,3,
Ashim Das^1,2,
Severine Leclerc ORCID: orcid.org/0000-0002-1485-4897²,
Florian Wünnemann^2,6,
Louis Berillon^1,2,
Gregor Andelfinger^2,7,
Sergio Crespo-Garcia⁸,
Alexandre Dubrac ORCID: orcid.org/0000-0002-3829-2577^1,2,9,10,
Flavio A. Rezende^9,11,
Clary B. Clish ORCID: orcid.org/0000-0001-8259-9245¹²,
Bruno Maranda¹³,
José Carlos Rivera^2,9,
Lois E. H. Smith¹⁴,
Przemyslaw Sapieha ORCID: orcid.org/0000-0002-9171-2825^9,11 &
…
Jean-Sébastien Joyal^1,2,3,7,9

Nature Communications volume 16, Article number: 5377 (2025) Cite this article

1078 Accesses
1 Altmetric
Metrics details

Subjects

Abstract

Healthy blood vessels supply neurons to preserve metabolic function. In blinding proliferative retinopathies (PRs), pathological neovascular tufts often emerge in lieu of needed physiological revascularization. Here we show that metabolic shifts in the neovascular niche define angiogenic fate. Fatty acid oxidation (FAO) metabolites accumulated in human and murine retinopathy samples. Neovascular tufts with a distinct single-cell transcriptional signature highly expressed FAO enzymes. The deletion of Sirt3, an FAO regulator, shifted the neovascular niche metabolism from FAO to glycolysis and suppressed tuft formation. This metabolic transition increased Vegf expression in astrocytes and reprogrammed pathological neovessels to a physiological phenotype, hastening vascular regeneration of the ischemic retina and improving vision. Hence, strategies to change the metabolic environment of vessels could promote a regenerative phenotype in vascular diseases.

Relapse of pathological angiogenesis: functional role of the basement membrane and potential treatment strategies

Article Open access 15 February 2021

Retinal neovascularization as self-organized criticality on ultra-widefield fluorescein angiography imaging of diabetic retinopathy

Article 13 February 2023

Microenvironmental determinants of endothelial cell heterogeneity

Article 28 January 2025

Introduction

Proliferative retinopathy (PR) is a leading cause of blindness in children with retinopathy of prematurity and working-age adults with diabetic retinopathy¹. The initial blood vessel loss in PR and the neuroretinal ischemia that follows trigger the excessive and misplaced production of vascular endothelial growth factor (VEGF), driving pathological angiogenesis in an attempt to reinstate metabolic homeostasis^1,2. While inhibiting VEGF to block pathological neovascular tuft formation has been the focus of therapeutic interventions to date³, the regenerative forces driving physiological revascularization of the ischemic neuroretina have been less explored. Discerning unique molecular fingerprints differentiating pathological and physiological neovessels, along with their associated vascular niche, could help reveal the drivers of healthy revascularization and neuronal survival.

Neuroglial cells of the vascular unit secrete growth factors and guidance cues that shape vascular architecture to meet their metabolic requirements^4,5. During vascular development, astrocytes^6,7 and retinal ganglion cells⁸ (RGCs) secrete VEGF that guides the advancing migratory tip endothelial cells (EC) of the vascular front. During PR, when the retina has fewer blood vessels and is hypoxic and nutrient-starved, excessive VEGF and repulsive guidance cues are secreted by ischemic neuroglial cells^8,9,10,11. However, dosed amounts of VEGF are critical for the survival of astrocytes and the physiological revascularization of the ischemic retina¹². While many secreted factors regulating pathological angiogenesis have been identified, it is unclear whether metabolic shifts could influence angiogenic fate, be it healthy or diseased.

During developmental angiogenesis, endothelial metabolism relies primarily on glycolysis to produce energy¹³. Glycolysis contributes to EC proliferation and migration required for vessel growth. Glycolytic enzymes are localized in lamellipodia and leading membrane ruffles of EC, powering the migration of endothelial tip cells¹⁴ that guide physiological vascular growth. Since glucose diffuses further from vessels than oxygen^14,15,16, EC can rely on anaerobic glycolysis when sprouting in avascular tissues. Pharmacological inhibition of glycolysis prevents sprouting angiogenesis^14,17. Hence, glycolysis may be a metabolic trait of EC that would be advantageous to revascularizing the ischemic retina.

Fatty acid oxidation (FAO) also promotes EC proliferation¹³. In addition to energy production, fatty acids are an essential carbon source to produce de novo nucleotides for DNA replication in EC¹⁸. This biosynthetic function contributes to EC proliferation, but unlike glycolysis, it is not associated with increased EC migration, a characteristic of tip cells driving vascular regeneration¹⁴. Inhibition of carnitine palmitoyl-transferase 1 (CPT1), shuttling long-chain fatty acids inside mitochondria for oxidation, decreases pathological tufts in murine PR¹⁸, but its impact on physiological revascularisation is unknown.

Sirtuins (SIRT) are NAD + -dependent deacetylases pivotal for mitochondrial function¹⁹. SIRT3 is a master regulator of FAO and oxidative energy metabolism within mitochondria^20,21,22. Since both FAO and glycolysis fuel many critical biological functions, including angiogenesis, we reasoned that SIRT3 could be a metabolic switch able to modulate the relative balance of energy utilization within the vascular niche without the deleterious effects of completely abrogating one of these essential energy sources. We combined unbiased large-scale metabolomics and single-cell transcriptomics data to define distinct metabolic and transcriptional signatures for physiological and pathological EC. While glycolysis is essential to the migratory and regenerative tip EC, we show that FAO is a hallmark of misguided yet actively proliferating pathological neovascular EC. Metabolic reprogramming of the retinal microenvironment caused by Sirt3 deletion, primarily expressed in astrocytes, shifts neovascular metabolism from FAO to glycolysis and enhances physiological vascular regeneration of the retina, improving neuroretinal function.

Results

Fatty-acid oxidation is a metabolic hallmark of pathological angiogenesis

We first examined the metabolite profile of vitreous humor from subjects with proliferative diabetic retinopathy (Supplementary Table 1, Supplementary Fig. 1a–c). Vitreous samples were collected adjacent to leaky pathological neovascular tufts that characterize PR (Fig. 1a). Amongst the top 25 most significantly dysregulated metabolites, we observed an accumulation of acylcarnitines in human neovascular PR samples compared to control subjects with epiretinal membranes (Fig. 1a, red font). Fatty acylcarnitines are by-products of fatty-acid β-oxidation (FAO), the process of oxidizing fatty acids in mitochondria to produce acetyl-CoA (Fig. 1b). Using Metaboanalyst²³, a metabolomics analysis tool, 5 of the 6 most enriched metabolic pathways in human PR pertained to FA metabolism, including long and short-chain saturated FAO pathways (Fig. 1b). Hence, the accumulation of FAO metabolites is a prominent characteristic of human proliferative retinopathy.

**Fig. 1: Fatty-acid oxidation is a hallmark of human proliferative diabetic retinopathy.**

Vitreous metabolites are often used as surrogate markers of adjacent retinal metabolism^24,25. However, decerning the precise cellular origin of these metabolites in human retinas could impair residual vision. Fibrovascular membranes (FVM) resected from diabetic retinopathy patients have been analyzed through single-cell RNAseq²⁶, revealing some tip-like endothelial cells (FVM EC 2, ESM1), yet these lesions typically lack active neovascular tufts (Supplementary Fig. 1d–f). We, therefore, used the oxygen-induced retinopathy (OIR) mouse model to study the formation of pathological neovascular tufts characteristic of PR (Fig. 2a)²⁷. In this model, mice pups are exposed to high oxygen concentration for 5 days (post-natal day (P) 7 to 12), causing vaso-obliteration (VO) and subsequent retinal ischemia. After returning to room air (P12), the ischemic neuroretina triggers both its physiological revascularization and the formation of pathological neovascular tufts that invade the vitreous²⁷. At the start of neovascularization in OIR (P14), we observed comparable enrichments of FAO metabolites to human PR (Fig. 2b and Supplementary Fig. 2a–c). Interestingly, acyl-carnitine fatty acid levels across all chain lengths (C2–C18) were elevated in both PR models despite their distinct etiologies, indicating a global disruption in FAO rather than a specific enzymatic blockage. Thus, the metabolomic profile in mice with PR closely resembled the human condition, establishing a relevant model to investigate the influence of FAO on neovascular disease.

**Fig. 2: Fatty-acid oxidation is enriched in the neovascular unit in mouse proliferative retinopathy.**

To determine the retinal source of FAO intermediates in PR, we compared the transcriptome of individual retinal cells using single-cell RNA sequencing data from mice exposed to OIR and their normoxic littermate controls at the instigation of retinal revascularization (at P14) and the peak of pathological angiogenesis (at P17)²⁸. Retinal cells were annotated, as previously described²⁸ (Fig. 2c). Gene set variation analysis (GSVA) for FAO pathways showed that cell types forming the vascular unit (EC, pericytes, astrocytes, and Müller glia) were the most enriched in transcripts coding for saturated FAO enzymes at both time points in the OIR hypoxic retina (Fig. 2d). EC (P14) and astrocytes (P17) showed strong transcriptional expression of mitochondrial FAO genes (Acadl, Acadvl, Hadh, Hadha, and Hadhb) compared to other cell types (Fig. 2e and Supplementary Fig. 2d). These results indicated that cell types comprising the neovascular unit (Fig. 2f), particularly EC and astrocytes, may contribute to the accumulation of FAO intermediates in ischemic PR.

Single-cell RNAseq identifies a unique transcriptional signature for neovascular tufts

Uncovering the specific molecular fingerprints of pathological neovessels has been a critical objective in vascular biology. To examine EC heterogeneity in PR, we enriched retinal single-cell suspensions for EC (using magnetic beads coated with PECAM1/CD31, Supplementary Fig. 3a-b) and performed a single-cell transcriptomic analysis of OIR and normoxic retinas. The unbiased assessment of uniform manifold approximation and projection (UMAP) results uncovered six retinal EC subclusters (Fig. 3a) comprising five known EC subtypes according to vascular zonation (arterial, vein, capillary, proliferative, tip cells²⁹) and one previously unidentified EC subtype (Aqp1 + EC, Fig. 3b). Normoxic control retinas were composed primarily of capillary, vein, and arterial EC, with relatively few tip cells (1-3%) and proliferating EC (0–5%). In contrast, OIR retinas at P14, following the initial vaso-obliterative phase, contained more proliferative (24%) and endothelial tip cells (25%) to revascularize the ischemic retina and less mature arteries (2%) and veins (32%) (Fig. 3c). Later, at P17, once further retinal revascularization was achieved, arterial (11%) and vein EC (57%) were more abundant, and fewer tip (10%) and proliferative EC (5%) were recovered.

Exclusively in OIR retinas, we detected a small EC cluster (1 to 2 % of total EC) defined by the specific expression of Aquaporin 1 (Aqp1), a water channel that controls cellular osmotic homeostasis. These unique Aqp1+ cells co-expressed bona fide EC genes (Cdh5, Cldn5, Pecam1) as well as vein (Rpl18a, Tmsb10, Rps28) and tip cell (Esm1, Mcam, Prss23) markers. Still, they were unique in their expression of genes involved in cell permeability (Aqp1) and matrix remodeling (Col18a1, Col15a1), amongst other pathways (Fig. 3b and Supplementary Fig. 3c, d). We also confirmed the presence of Aqp1+ endothelial subclusters in publicly available scRNAseq datasets from mouse OIR retinas (P14) (Supplementary Fig. 3e,f)³⁰. Hierarchical clustering based on highly variable genes showed that the Aqp1 + EC were the most divergent compared to other endothelial subtypes and closest to the tip cell subcluster (Fig. 3d). Based on these transcriptional characteristics and the absence of Aqp1 + EC in normoxic retinas (devoid of tufts), we postulated that Aqp1 + EC might correspond anatomically to pathological neovascular tuft EC. Indeed, we localized AQP1-expressing EC to pathological tufts in OIR retinas using immunofluorescence but not in healthy normoxic vessels (Fig. 3e) nor other retinal cell types, contrarily to previous reports³¹ (Supplementary Fig. 3g, h). We also confirmed the expression of tip cell marker ESM1 in both tip and tuft EC (Fig. 3f), in line with previous evidence that tuft EC might be misguided tip cells⁹. Hence, we identified a unique transcriptional signature for pathological neovascular tuft EC in the murine OIR model, characterized by AQP1 expression.

FAO and glycolysis dichotomize neovascular tufts and physiological tip cells

We then compared the newly obtained transcriptional signature of pathological tuft EC to the physiological tip EC required for regenerative angiogenesis. We identified 449 differentially expressed genes (DEGs) between tuft and tip EC (P value < 0.05, log2(FC) > 0.5, Fig. 3g). Up-regulated genes in tip EC were enriched for extracellular matrix-receptor interaction, focal adhesion, and glycolysis pathways. In contrast, tuft EC genes were significantly enriched for ferroptosis, fatty-acid degradation, tyrosine, and nicotinamide metabolism pathways (Fig. 3h). To better visualize the metabolic dichotomy between tuft and tip EC, we performed a gene set variation analysis (GSVA) to rank EC subtypes based on their relative expression of genes implicated in glycolysis and FAO pathways (Fig. 3i). In OIR, tip EC were highly glycolytic and relied less on FAO (P17). Inversely, tuft EC relied preferentially on FAO and least on glycolysis (P17).

Focusing on the most abundant FAO and glycolysis transcripts, tip cells showed greater expression of essential glycolic genes (Pkm, Aldoa, Tpi1, Gpi1), particularly at P17, while tuft EC expressed higher levels of FAO genes (Hadha, Acadvl) (Fig. 3j). Notably, tuft EC strongly expressed Hadha (Hydroxyacyl-CoA Dehydrogenase Trifunctional Multienzyme Complex Subunit Alpha), which encodes a subunit of the mitochondrial trifunctional protein essential for FAO. Importantly, Hadha expression colocalized with pathological Aqp1+ tufts by RNAscope in situ hybridization on retinal cryosections (Fig. 3k). Thus, we discerned distinct transcriptional metabolic identities for healthy and diseased EC in murine PR, favoring glycolysis in the case of tip EC and FAO in the case of pathological tuft EC (Fig. 3l, top). This metabolic dichotomy was also evident in the vascular niche, with astrocytes over-expressing FAO pathways, while Müller cells were more glycolytic in OIR (Fig. 3l, bottom). Importantly, our findings provide an endothelial atlas from the ischemic phase of proliferative retinopathy (P14/P17), enabling deeper insights into the divergent transcriptomic programs separating pathological and physiological angiogenesis (Fig. 3m).

Targeting FAO curbs pathological neovascular tufts

We then explored the role of FAO in neovascular tuft formation. We utilized etomoxir to inhibit FAO, blocking CPT1a, which is essential for transporting lipids into mitochondria¹⁸ (Fig. 4a). Etomoxir reduced pathological neovascularization (NV) and vaso-obliteration (VO) in WT retinas exposed to OIR (Fig. 4b), in line with a contribution of FAO to neovascular tuft formation. However, etomoxir caused substantial toxicity and fatalities during the hypoxic phase of OIR (P12 to P17), highlighting the vital role of FAO metabolism systemically. We, therefore, sought to identify a more targeted metabolic regulator capable of yielding therapeutic gains without harmful systemic side effects.

Mitochondrial SIRT3 is an interesting metabolic target due to its central role in regulating cellular energy homeostasis without entirely inhibiting FAO. SIRT3 enhances the activity of essential enzymes of FAO²², the Krebs cycle^32,33, and the respiratory chain³⁴ (Fig. 4a). SIRT3 directly deacetylates HADHA and ACADVL³⁵ (Supplementary Fig. 4a), two FAO enzymes predominantly enriched in the neovascular unit (Fig. 2e). During OIR, Sirt3 expression increased significantly in astrocytes (12.5% of cells, P = 0.008), whereas its expression was limited in EC (2%) and absent in tuft EC by scRNAseq (Fig. 4c). Accordingly, conditional EC deletion of Sirt3 (Cdh5-CreERT2, Sirt3^lox/lox) did not impact the vascular phenotype of transgenic mice exposed to OIR (Supplementary Fig. 4b–d). Interestingly, global Sirt3 deletion (Sirt3^−/−) markedly reduced pathological neovascular tuft (NV) and vaso-obliteration (VO) compared to wild-type (WT) retinas (Fig. 4d), akin to the effects of etomoxir but without increased mortality or apparent deleterious consequences. Hence, Sirt3 depletion of the neurovascular niche, but not EC alone, reduced pathological angiogenesis.

Next, we examined the metabolic effects of Sirt3 depletion. Using metabolic cages, Sirt3^−/− mice displayed a higher exchange ratio (vCO₂/vO₂) during their active nocturnal phase, indicating a preference for carbohydrate metabolism over lipid reliance (Fig. 4e). We also observed increased glucose uptakes in Sirt3^−/− retinas compared to WT, using trace amounts of radioactively labeled 2-deoxyglucose (DG) in mice exposed to OIR (Fig. 4f); 2-DG is a glucose analog that is not metabolized by the retina. Their acyl-carnitine profiles revealed that Sirt3 deletion prevented the increase in total acylcarnitines noted in WT retinas during OIR (Supplementary Fig. 2c). Indeed, reduced levels of fatty acylcarnitine metabolites were measured in mutant retinas compared to WT during OIR (Fig. 4g, h), underscoring the importance of SIRT3 in mitochondrial FAO.

To define the transcriptional effects of Sirt3 deletion during pathological angiogenesis, we repeated a single-cell RNAseq of EC-enriched retinal cell suspensions. The enrichment strategy improved EC yields, but other retinal cell types were also detected, except for retinal ganglion cells (Fig. 4i). To identify which retinal cell types were most impacted by the global deletion of Sirt3 at a transcriptional level, we employed a machine-learning approach using a Random Forest Classifier termed Augur³⁶. Based on Augur results, EC and macroglial cells, comprising Müller cells and astrocytes, were the most impacted cell types in our model (Fig. 4j). In Sirt3 depleted retinas, the EC composition transitioned rapidly from a more proliferative neovascular profile at P14 compared to WT (tuft, tip, and proliferative EC: 43.7% vs. 38.2%) to a more quiescent EC composition at P17 (16.1% vs 23.7%), resembling the normoxic retina (Fig. 4k and Supplementary fig. 4e). At P17, tuft EC were barely detected (0.1 % vs 0.7%) in Sirt3^−/− retinas (Fig. 4l). This reduction in pathological angiogenesis in Sirt3-depleted retinas (Fig. 4d, l) was associated with a metabolic shift from FAO to glycolysis within the vascular niche, particularly in tuft EC and astrocytes (Fig. 4m).

Sirt3 depletion alters the metabolic and angiogenic landscape of the neovascular niche

Since Sirt3 was mainly expressed in astrocytes (Fig. 4c), we characterized the metabolic impact of Sirt3 depletion in human retinal astrocytes exposed to hypoxia to mimic the ischemic retinal environment of PR. Using a Seahorse analyzer, Sirt3-depleted astrocytes (siSIRT3) exhibited higher glycolysis and glycolytic capacity than controls (scrRNA, Fig. 5a, b). In contrast, their basal oxygen consumption rates (OCR) and ATP-linked respiration were not significantly impacted (Fig. 5c), suggesting a Warburg-like effect (Fig. 5d). In turn, Sirt3 deletion altered the interactome of the vascular niche with tip and tuft EC (Fig. 5e), dampening inflammatory and proliferative signals observed in WT retinas during OIR (Supplementary Fig. 5a–d). Notably, astrocytes of Sirt3-depleted retinas expressed more Vegfa (5-fold increase) and other important angiogenic factors (Fig. 5f, g) to levels comparable during retinal development (P6-P10) (Supplementary Fig. 5e–g). This was also observed in vitro, where silencing Sirt3 further increased hypoxia-induced Vegfa expression in human cultured astrocytes (Supplementary Fig. 5h).

To confirm the critical role of Sirt3 in retinal astrocytes, we infected Sirt3^lox/lox retinas with AAV2 viruses carrying a GFAP-driven Cre recombinase, which explicitly targets neuroglial cells³⁷ (Supplementary Fig. 5i). The virus was delivered by intravitreal injection at P7, before OIR. Similar to Sirt3^−/− OIR retinas, selective depletion of Sirt3 in astrocytes and activated Müller cells (AAV2-GFAP-Cre) reduced vaso-obliteration (VO) compared to control contralateral retinas injected with AAV2-GFAP-GFP or PBS (Fig. 5h), indicating accelerated retinal revascularization. However, tuft formation (NV) remained unchanged, suggesting the involvement of additional neovascular niche cells. Astrocytes classically form a physical path and secrete VEGF, guiding and enabling physiological vascular growth^4,7. In OIR, the shift in the vascular niche metabolism and the higher VEGF secretion from astrocytes were associated with more tip and tuft EC in Sirt3^−/− retinas at P14 (Fig. 4l). At this early stage in retinal revascularization (P14), neovascular cells are not yet fully committed, and tufts are barely visible histologically. Surprisingly, fewer tuft EC were observed at P17 in Sirt3^−/− retinas, raising the possibility that early tuft EC fate was altered due to their microenvironment (Fig. 5i).

Sirt3 depletion reprograms early neovascular tufts for regenerative angiogenesis

Given that EC behavior adapts to their environment, we explored whether the angiogenic shift noted in the Sirt3^−/− neovascular niche could prompt tuft EC to adopt a tip-like endothelial fate. Indeed, despite an early increase in tuft EC at P14, the ratio of tip to tuft EC significantly increased at P17 in Sirt3^−/− retinas (Fig. 6a), suggesting a transition from pathological to physiological angiogenesis in the mutant retinas. Moreover, analysis of transcriptional trajectory (using Monocle 3 and RNA velocity) showed that in contrast to WT, tuft EC were still connected to tip cells in Sirt3^−/− retinas (Fig. 6b and Supplementary Fig. 6a), raising the possibility that they could be reprogrammed to take on a physiological fate.

**Fig. 6: Sirt3 deletion reprograms early neovascular tufts for regenerative angiogenesis.**

Considering that FAO distinctly characterizes tuft from tip EC (Fig. 3), we examined the expression of FAO genes in tuft EC of Sirt3^−/− retinas. Using GSVA and differential quantitative set analysis for gene expression (QuSAGE³⁸), we identified a marked reduction in FAO pathways in Sirt3^−/− tuft EC, primarily due to reduced gene expression of Acadl, Acadm, and Hadha (Supplementary Fig. 6b,c). Given the predisposition of tuft EC to proliferate, possibly aided by FAO¹⁸, we compared the expression of cell cycle genes coding for proliferation (G2/M) or cell cycle arrest (G1) in both tuft and tip EC. We observed that, at P14, tuft EC in Sirt3^−/− retinas were less proliferative than WT (30% vs. 80%), akin to the reduced proliferation index of migratory tip cells (~30%, Supplementary Fig. 6d). Hence, Sirt3 deletion prevented tuft EC from acquiring defining traits of tuft identity.

Finally, we investigated if the initial reduction in FAO and proliferation of tuft EC in Sirt3^−/− retinas might correlate with a comprehensive transcriptional shift towards a tip-cell-like identity. Using GSVA and a pre-established tip cell gene signature (Fig. 3g), we found that early tuft EC (P14) exhibited a transcriptional profile resembling physiological tip cells, more so than their WT counterparts (Fig. 6c). We then analyzed pathways that define tip cell identity at P14 (Fig. 3h), a critical time point for EC fate specification in OIR. Tip cells in WT and Sirt3^−/− retinas shared analogous expression profiles (Fig. 6d). However, tuft EC of Sirt3^−/− retina began expressing tip cell markers, focal adhesion, hypoxia-regulated, and glycolytic genes (Fig. 6d) to levels comparable to tip cells. The findings corroborated with the elevation of hypoxia, glycolysis, and sprouting angiogenesis pathways in Sirt3^−/− retinas, in contrast to WT (Fig. 4m). These pathways are characteristic transcriptional traits of tip cells³⁹. Correspondingly, Sirt3 deletion markedly increased the number of early tip cell filopodia (P14; Fig. 6e) and accelerated revascularization of the ischemic neuroretina in OIR at P14 and P17 without significantly affecting maximal vaso-obliteration at P12 (Fig. 6f). More importantly, prompt revascularization of Sirt3^−/− retinas improved later inner neuroretinal and RGC function (at P21) assessed by the photopic negative response⁴⁰ (PhNR) of full-field electroretinogram (ERG, Fig. 6g). Hence, Sirt3 deletion within the neovascular niche shifted the transcriptional identity of tuft EC to resemble tip cells, thereby accelerating the physiological vascular regeneration of the ischemic neuroretina in PR and improving vision.

Discussion

Recognizing the unique attributes of healthy and diseased blood vessels and their microenvironment could pave the way for more targeted and efficacious therapies against vascular diseases. The lack of a suitable murine model that accurately replicates human proliferative retinopathies (PR) is a challenge in the field. Poor glucose control in diabetics and extreme prematurity in neonates correlate with neovascular retinal disease in humans but not in mice. The most commonly used murine model of PR is the oxygen-induced retinopathy (OIR) model²⁷, giving rise to the predictable formation of pathognomonic tufts that resemble human retinal proliferative neovessels. Interestingly, we observed comparable FAO metabolite signatures in both human diabetic PR and murine OIR samples, suggesting a shared metabolic dysregulation in neovascular disease despite their distinct etiologies. This study uncovered a distinct transcriptional signature for pathological tuft EC, a defining feature of PR and a leading cause of blindness. Unlike healthy tip cells, these pathological EC exhibited a preferential metabolic dependency on FAO. Notably, shifting the metabolic program of the vascular niche from FAO to glycolysis in Sirt3^−/− retinas reduced pathological neovascular tufts and promoted healthy retinal revascularization.

Pathological neovascular tufts are portrayed as misdirected tip cells invading the vitreous⁹ that are more proliferative and leaky, resulting in retinal edema⁴¹. Through single-cell transcriptomics, we found a novel subset of EC that, while displaying multiple tip cell markers such as ESM1, were less prone to migrate and more inclined to proliferate. The new EC sub-cluster expressed high levels of aquaporin 1 (AQP1) which we localized to pathological neovascular tufts. In tumor models, Aqp1 directly contributes to EC migration, sprouting, and tubulogenesis⁴², implying a direct role in pathological angiogenesis⁴³. Hence, AQP1 uniquely labels pathological tuft EC in PR and could be a potential therapeutic target to curb pathological neovascularization in proliferative vascular diseases.

Using their distinct transcriptional signature, we showed that unique metabolic profiles defined tip and tuft EC. In both human neonates and mice, aberrant vascular growth is ushered by an increase in the neuroretina’s metabolic demands, pointing to the pivotal role that retinal metabolism plays in determining EC fate⁵. Unbiasedly comparing tip and tuft EC’s most enriched transcriptional pathways revealed divergent metabolic preferences: tip cells favored glycolysis, whereas tuft cells leaned towards FAO. Notably, glycolysis is less dependent on oxygen for energy production and cellular growth, a phenomenon known as the Warburg effect, while FAO necessitates mitochondrial respiration⁴⁴. This metabolic specialization aligns anatomically with the ___location of tip EC at the leading edge of the oxygen-poor, avascular neuroretina. In contrast, neovascular tufts are found at the confluence of venous capillaries⁴⁵, an area richer in oxygen. FAO is a fuel source for the neuroretina⁴⁶ and serves as a biosynthetic pathway promoting EC proliferation¹⁸. Unlike glycolysis, FAO is not associated with the increased migration typical of tip cells¹⁴. Hence, distinct metabolic pathways define the behavior of glycolytic tip cells, essential to the regenerative process, in contrast to misguided yet actively proliferating tuft EC that rely on FAO. We speculate that using FA as an alternative fuel substrate in better-perfused regions could limit glucose uptake in tufts and surrounding cells, thereby preserving glucose for critical revascularization and diffusion into the avascular retina to nourish hypoxic neurons. Hence, metabolic heterogeneity could help sustain the regenerating retina and is not necessarily in contradiction with an overall reduction in mitochondrial function in PR avascular regions. Crucially, these distinct metabolic profiles for healthy and pathological angiogenesis become apparent at specific stages of disease development, presenting a potential therapeutic window for targeted interventions.

Localization of the metabolic and angiogenic signals shapes regenerative angiogenesis. Sirt3 expression was only significantly increased in astrocytes in mouse PR and was not detected in tuft EC. In the absence of Sirt3, astrocytes exhibited a pronounced metabolic shift towards glycolysis and elevated angiogenic signaling, with a 5-fold increase in Vegfa expression, comparable to levels seen in the developing retina (P6). Given that astrocytes act as a scaffold to guide migrating EC during retinal development, they could provide localized signals to guide beneficial revascularization in PR. PR treatments aim to lower VEGF concentrations after neovascular tuft formation has taken place, irrespective of its source. Our findings highlight the importance of the cellular source of growth factors, including VEGF, to guide healthy retinal revascularization, not merely its concentration. VEGF has multiple neuroprotective roles that could be leveraged to enhance neuronal function⁴⁷. Timely revascularization and the redistribution of VEGF expression in supporting neuroglial cells may contribute to restoring inner retinal function, as indicated by ERG. Therefore, modifying the metabolic topography of the vascular niche, especially in astrocytes, could help steer EC toward a more regenerative path, thereby improving vision.

Maintaining lower oxygen saturation levels is the mainstay of retinopathy of prematurity (ROP) prevention⁴⁸, which might contribute to EC reprogramming. Within limits, however, since more severe hypoxia, although beneficial to ROP, is associated with increased mortality⁴⁹. Hypoxia blocks mitochondrial respiration, forcing tissues to rely on glycolysis instead^50,51, as we observed in SIRT3-depleted mice. In mouse models of mitochondrial diseases, exposure to chronic hypoxia is protective by diverting energy metabolism away from mitochondrial oxidative phosphorylation and reducing oxidative stress⁵⁰. In the OIR model, pharmacological stabilization of hypoxia-inducible factor (HIF) also shifts the retina towards aerobic glycolysis, improving retinopathy⁵². Similarly, Sirt3 null mice shifted their metabolism from FAO towards glycolysis⁵³, recreating a pseudo-hypoxic state that could pre-condition mice for ischemic stress in proliferative retinopathy. Thus, if administered locally in the eye, targeting SIRT3 could favor a regenerative phenotype while avoiding some of the harmful consequences of severe hypoxia. Then again, SIRT3 was reported to have protective antioxidant roles in photoreceptors⁵⁴. Fortunately, SIRT3 was not significantly expressed in photoreceptors during PR, and Sirt3 depletion alone did not negatively impact vision in other models, suggesting protective effects of other mitochondrial sirtuins, such as SIRT5^55,56.

Our current therapies target pathological neovessels that emerge in the later phases of PR. In contrast, our research aims to enable early physiological vascular regeneration to minimize neuroretinal ischemia and improve vision. Targeting SIRT3 shifted the metabolic environment of the vascular niche, steering EC fate towards a regenerative vascular phenotype. This work expands our understanding of the distinct metabolic and transcriptomic landscapes of healthy and diseased vessels and their microenvironment, offering insights for developing targeted therapies against vascular diseases^57,58. Additionally, our findings have implications for other neuro-ischemic conditions like stroke, where enhancing regenerative angiogenesis could preserve neuronal function⁵⁹.

Methods

Human subjects

The study conforms to the tenets of the Declaration of Helsinki, and approval of the human clinical protocol and informed consent were obtained from the Maisonneuve-Rosemont Hospital Ethics Committee (CER: 10059). All patients previously diagnosed with proliferative diabetic retinopathy or epiretinal membrane were followed clinically, and surgery was performed by a single vitreoretinal surgeon when indicated by ‘standard-of-care’ guidelines (Supplementary Table 1). The collection was performed prior to 2017, and informed written consent was obtained. Patients with epiretinal membranes represent appropriate controls because they don’t exhibit proliferative retinopathy but still need surgery, allowing the collection of vitreous fluid under informed consent. Undiluted vitreous samples were aspirated in the region adjacent to neovascular tufts or epiretinal membrane (control) and were frozen on dry ice immediately after biopsy and stored at –80 °C. All patients received their first anti-VEGF injection after vitreous sampling; they did not have prior anti-VEGF or photocoagulation treatment. Medication details are provided in Supplementary Table 1, except for four deceased diabetic patients whose records were no longer available (NA). Sex and gender factors were not analyzed due to insufficient statistical power when disaggregated.

Animal care

Animal procedures complied with the Animal Care Committee of CHU Sainte-Justine (CIBPAR Nagano 2024-5421#712), following the principles of the Guide for the Care and Use of Experimental Animals developed by the Canadian Council on Animal Care. All in vivo work adhered to the Association for Research in Vision and Ophthalmology Statement for the Use of Animals in Ophthalmic and Vision Research. S129 (#002448), mutant mice with Sirt3 deletion (129-Sirt3^tm1.1Fwa/J, #012755), CAG-tdTomato reporter mice (#007914), and Sirt3^lox/lox mice (#031201) were obtained from The Jackson Laboratory. Cdh5-CreERT2 mice were kindly provided by Dr. Ralf Adams (Max Planck Institute, Germany). Mutant mice did not carry Pde6b (rd1) or Crb1 (rd8) mutations, and a single-nucleotide polymorphism panel analysis confirmed that S129 (#002448) and Sirt3^−/− (#012755) had comparable genetic backgrounds. Mice were housed under a standard 12-hour light-dark cycle and had ad libitum access to water and chow. Animals were monitored daily for general health status, body weight, hydration, and signs of distress or discomfort. Humane endpoints, including sustained weight loss exceeding 15%, lethargy, respiratory distress, or severe morbidity, were established. Animals meeting these endpoints were promptly euthanized using CO₂ asphyxiation followed by cervical dislocation, per institutional guidelines. Pups weighing less than 6 grams or more than 9 grams on post-natal (P) day 14 or 17 (P14, P17) were excluded. Both female and male littermates were used, but sex information was not collected. All animals were sacrificed by lethal intraperitoneal (i.p.) injections of pentobarbital at P12, P14, or P17 for analyses.

Oxygen-induced retinopathy model

Mice pups and their adoptive lactating S129 mother were exposed to 75% oxygen from P7 to P12, as first described²⁷. Mice were returned to room air at P12 until used for experimentation. All animals were euthanized before ocular enucleation. Littermate controls, as well as age and weight-matched mice, were injected intraperitoneally twice daily with etomoxir (27 mg/kg per day) or vehicle (0.9% saline) from P12 to P17 and euthanized at P17. 4-OH tamoxifen (5 μg) was injected intraperitoneally daily for 3 days (P12-P14) and euthanized at P17.

Human vitreous metabolomics

Human vitreous metabolite extracts were analyzed using two liquid chromatography-tandem mass spectrometry (LC-MS) methods to measure polar metabolites, as described previously⁶⁰. From vitreous samples, negative ion mode profiling samples were prepared from 30 µl extracted with 120 µl of 80% methanol containing inosine-15N4, thymine-d4, and glycocholate-d4 internal standards (Cambridge Isotope Laboratories, MA). In comparison, samples for positive ion mode profiling were prepared by extracting 10 µL with 90 µl of acetonitrile/methanol/formic acid (74.9:24.9:0.2 v:v:v) containing valine-d8 and phenylalanine-d8 internal standards (Cambridge Isotope Laboratories, MA). Negative ion mode data were acquired by injecting extracts (10 µL) onto a 150 × 2.0 mm Luna NH2 column (Phenomenex, CA). The column was eluted at a flow rate of 400 µl/min with initial conditions of 10% mobile phase A (20 mM ammonium acetate and 20 mM ammonium hydroxide in water) and 90% mobile phase B (10 mM ammonium hydroxide in 75:25 v/v acetonitrile/methanol) followed by a 10 min linear gradient to 100% mobile phase A. MS analyses were carried out using electrospray ionization in the negative ion mode using full scan analysis over m/z 70–750 at 70,000 resolution and 3 Hz data acquisition rate. Additional MS settings were ion spray voltage, −3.0 kV; capillary temperature, 350 °C; probe heater temperature, 325 °C; sheath gas, 55; auxiliary gas, 10; and S-lens RF level 50. Positive ion mode data were acquired by injecting extracts (10 µl) onto a 150 × 2 mm, 3 µm Atlantis HILIC column (Waters, MA). The column was eluted isocratically at a flow rate of 250 µl/min with 5% mobile phase A (10 mM ammonium formate and 0.1% formic acid in water) for 0.5 minutes, followed by a linear gradient to 40% mobile phase B (acetonitrile with 0.1% formic acid) over 10 minutes. MS analyses were carried out using electrospray ionization in the positive ion mode using full scan analysis over 70–800 m/z at 70,000 resolution and 3 Hz data acquisition rate. Other MS settings were: sheath gas 40, sweep gas 2, spray voltage 3.5 kV, capillary temperature 350 °C, S-lens RF 40, heater temperature 300 °C, microscans 1, automatic gain control target 1e6, and maximum ion time 250 ms. Raw data were processed using TraceFinder 3.3 and 4.1 software (Thermo Scientific, MA) and Progenesis QI (Nonlinear Dynamics, UK). For each method, metabolite identities were confirmed using authentic reference standards.

Mouse retinal lipidomics

Metabolic profiling was performed by LC-MS on samples pooled from both retinas of P14 mice; samples were analyzed for fatty acylcarnitine species and beta-oxidation intermediate metabolites. Retinas were harvested after sacrificing P14 mice and immediately snap-frozen in liquid nitrogen. The acylcarnitines were extracted by protein precipitation (PP) and derivatized with butanol-HCl for analysis by LC/MS/MS. The samples were diluted with 400 µl solution of methanol with eight ISTD. After 1.5 minutes of homogenization in a bead Ruptor12 from Omni International Inc., samples were centrifuged for 5 minutes at 12,000 rpm. The supernatant was transferred in a 2 ml polypropylene tube for evaporation under nitrogen. When dry, 100 µl of butanol-HCl (3 N) is used for butanolysis performed by heating at 55 °C for 20 min. After evaporation under nitrogen, the residue was reconstituted in 100 µl of mobile phase (ACN:H2O 80:20 with 0.05% of formic acid). Twenty microliters were injected in flow injection onto an Alliance 2795 LC coupled to a Quattro micro MS/MS (Waters). Data were recorded in positive electrospray ionization and analyzed with Neolynx from Waters Corp. The average of two technical replicates (2 LC/MS/MS reads/sample) was recorded.

Metabolomics data analysis

Human metabolomics profiles were analyzed using MetaboAnalyst²³. Data were filtered based on the interquartile range for further statistical comparisons. We used a Random Forest (RF) classifier to identify outliers based on proximities during tree construction, utilizing the randomForest R package and PlotRF. Glucuronate and urate levels contributed the most to classifier accuracy. Outliers exhibited elevated levels of cotinine, hydroxycotinine (PDR 153), ceramide (PDR 126), caffeine (PDR 144), and bilirubin (ERM 136). Outliers detected by the Random Forest algorithm were removed, and PCA, t-test, AUC test, and Metabolite Set Enrichment Analysis were performed on auto-scaled data previously normalized on internal standard (Phe-d8). Mouse acylcarnitine profiling was performed on unfiltered, auto-scaled values from MetaboAnalyst²³ followed by PCA and hierarchical clustering (dendrogram and heatmap).

Calorimetry/Metabolic chambers

Mice (5-6 months old) were acclimatized to the metabolic cages for the first 24 hours and evaluated for the next 5 days. Mice had free access to food and water. The consumption of oxygen (vO₂), the production of carbon dioxide (vCO₂), the ratio of gaseous exchange (RER), the heat production, and the physical activity were repeatedly collected with the Oxymax/clams calorimetry system (Colombus Instruments). The weight of every animal was measured before and after the stay in metabolic cages.

Droplet Sequencing (Drop-seq)

Drop-seq procedure was done on endothelial enriched cell suspension isolated from mouse retinas. Briefly, single-cell suspensions were prepared from P14 and P17 retina, as reported⁶¹, through successive steps of digestion (using papain solution; Worthington, LK003150), trituration, and filtration to obtain a final concentration of 120 cells/μl. According to the manufacturer’s protocol, endothelial cells were enriched using CD31 microbeads (Miltenyi Biotech, 130-097-418). Droplet generation and cDNA libraries were performed as described in the Drop-seq procedure⁶¹, and sequencing was done on Illumina NextSeq 500.

Unique molecular identifier (UMI) counts for WT and/or Sirt3^–/– scRNAseq replicates were merged into one single Digital Gene Expression (DGE) matrix and processed using the Seurat package⁶². Cells expressing less than 100 genes and more than 10% of mitochondrial genes were filtered out. Single-cell transcriptomes were normalized by dividing by the total number of UMIs per cell and then multiplying by 10,000. All calculations and data were then performed in log space. PCA analysis on the most variable genes in the DGE matrix identified 20 significant PCs, which served as input for Uniform Manifold Approximation and Projection (UMAP). We used a density clustering approach to identify putative cell types on the embedded map and computed average gene expression for each identified cluster based on Euclidean distances. We then compared each cluster to identify marker genes differentially expressed across clusters.

Cell type-specific transcriptomic differences between conditions, time points, and genotypes were statistically compared using a negative binomial model. Seurat visualization tools included Ridge Plot, Dot Plot, and UMAP Plot. Single-cell gene expression profiles from each separate cell type identified by scRNAseq were further analyzed using ggplot2⁶³, Gene Set Variation Analysis⁶⁴, EnrichR⁶⁵, AUGUR³⁶, Psupertime⁶⁶, NicheNet⁶⁷, SCENIC⁶⁸, Monocle3⁶⁹ and Qusage³⁸. For RNA velocity analysis, Drop-seq fastq files were processed with kb-python⁷⁰ to obtain spliced and unspliced matrices that were further analyzed with Seurat RunVelocity function based on Velocyto⁷¹.

RNAscope in situ hybridization and Immunohistochemistry

For RNAscope analysis, experiments were performed per manufacturer instructions on fixed retinal cryosections (12 μm) using the RNAscope 2.5 chromogenic duplex assay (ACD bio-techne). WT and Sirt3^–/– eyes at P17 were fixed in paraformaldehyde (PFA) 4% (24 h) at 4 °C. Eyes were then immersed in a sucrose gradient (10 to 30%), frozen in optimal cutting temperature compound (Surgipath FSC22 Clear, Leica), and stored at −80 °C until use. Aqp1 and Hadha mRNA expression were targeted using probes designed by ACD Bio-techne. For cryosection, fixed and cryoprotected samples were embedded in OCT compound (Fisher Scientific; Cat # 23-730-571). Retinal cryosections (10 μm) were blocked with 1% bovine serum albumin (Multicell; Cat #800-095-EG), 0.1% Triton X- 100 (Sigma; Cat #X100)) and 0.05% Tween-20 (Fisher; Cat # BP337-500) for 1 h and labeled overnight at 4 °C with primary antibodies against GFAP (1:100; Millipore-Sigma Cat. #SAB2500462), glutamine synthetase (1:250, Sigma Cat.#MAB302) and AQP1 (1:150, Alomone, AQP-001), followed by secondary antibodies in blocking solution for 2 hours RT. RNAscope and IF retina were counterstained with isolectin B4 (Alexa Fluor 594, I21413; Molecular Probes) or DAPI (Invitrogen; Cat #D21490). Retinal tissues were washed three times for 5 minutes in PBS before a cover slip was attached with Fluoromount Aqueous Mounting Medium (EMS; Cat # 17985). Retinas were then imaged with a Leica DME-6 microscope using a 20X dry objective (HC PL APO 20X/0.75 CS2) and analyzed with ImageJ (NIH). Labeled retinas were examined with a laser scanning confocal microscope (Leica, SP8 DLS).

Intravitreal injection of AAV

AAV-GFAP-Cre plasmid was obtained from Addgene (Plasmid #105550). AAV2/ShH10-GFAP-GFP and AAV2/ShH10-GFAP-Cre vectors were produced and highly purified from an established cell line (HEK293) at Boston Children’s Hospital Viral Core. The ShH10 variant of the AAV-derived capsid has been shown to preferentially infect glial cells and produce sustained transgene expression following intravitreal injection³⁷. The titer of AAV2/SHH10-GFAP-GFP was 1.49 × 10¹³ gc/ml and AAV2/SHH10-GFAP-Cre was 1.55 × 1012 gc/ml. Mice were anesthetized with 3% isoflurane, and intravitreal injections were performed on deeply sedated animals as we described⁷². Intravitreal injections of PBS, AAV2/ShH10-GFAP-GFP or AAV2/ShH10-GFAP-Cre vectors (1 μl each) were performed through the eyelid using a 50-gauge glass capillary attached to a micropipette (Hamilton; Cat # 7681C15) that was inserted 1–2 mm behind the ora serrata into the vitreous cavity to avoid damaging the lens. Once the animals had fully recovered from anesthesia, they were subjected to the OIR model. The animals were euthanized 10 days after intravitreal injection with AAV vectors.

Quantification of neovascularization (NV) and vaso-obliterated (VO) areas

Eyes were fixed for 1 hour in 4% paraformaldehyde at room temperature (RT). Retinas were dissected and stained overnight at RT with fluoresceinated isolectin B4 (Alexa Fluor 594, I21413; Molecular Probes) in 1 mM CaCl₂ in PBS. Lectin-stained retinas were whole-mounted onto Superfrost Plus microscope slides (Fisher Scientific) with the photoreceptor side down and embedded in SlowFade Antifade Reagent (Invitrogen). Stained retinas were flat-mounted, and the pictures were taken with an epifluorescence slide scanner, Zeiss Axio Scan.Z1 (Zeiss) at 10x magnification. The neovascular (NV) and vaso-obliterated (VO) areas were quantified using the SWIFT_NV program⁷³.

Cell culture

Human retinal astrocytes (HRAs) (1870, ScienCell, USA) free of pathogens were cultured in Astrocyte Medium (1801, ScienCell, USA), supplemented with 10% fetal bovine serum (FBS) and 1% antibiotic-antimycotic solution. Cells were maintained at 37 °C in a 5% CO₂, and only passages 5–10 were used for the experiments. HRAs were transfected with either siRNA targeting SIRT3 (20 nM, L-004827-01-0005, Dharmacon, Canada) or scrambled RNA control (scrRNA) using HiPerFect transfection reagent (301704, Qiagen, Germany) on the second day after seeding. Maximum gene depletion was achieved 72 hours post-transfection. Cells were then exposed to hypoxia (5% O₂) or normoxia for 6 hours before RNA extraction (2 × 105 cells/well) or 4 hours before Seahorse assays (3.5 × 104 cells/well).

Quantitative RT-PCR

Total RNA was extracted using the RNeasy Mini Kit (74106, Qiagen, Germany) according to the manufacturer’s instructions. For RT-qPCR, 1 μg of total RNA was reverse-transcribed using the QuantiTect Reverse Transcription Kit (205314, Qiagen). Quantitative PCR was performed with iTaq Universal SYBR Green SuperMix (1725124, Bio-Rad, USA) on a LightCycler® 480 System (Roche, Switzerland). Melt curve analysis was conducted to confirm the specificity of the PCR products. Gene expression fold-changes were calculated to GAPDH from normoxic control (norm scrRNA) using the ΔΔCt method.

Primer sequences were:

Sirt3_human_F = ACCCAGTGGCATTCCAGAC, Sirt3_human_R = GGCTTGGGGTTGTGAAAGAAG, Vegfa_Human_F = AGGGCAGAATCATCACGAAGT, Vegfa_Human_R = AGGGTCTCGATTGGATGGCA, Gapdh_human_F = GAAGGTGAAGGTCGGAGT, Gapdh_human_R = GAAGATGGTGATGGGATTTC.

Seahorse assay

The oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) were measured using a Seahorse XFe 96 Flux Analyzer. Human retinal astrocytes (HRAs) transfected with siSIRT3 (20 nM) or scrRNA were seeded in 96-well plates at a density of 3.5 × 10⁴ cells per well. Two days post-transfection, cells were exposed to hypoxia (FiO2 5%) or normoxia for four hours. One hour before the assay, cells were incubated in either Mito Stress Test or Glycolysis Stress Test media under their respective oxygen conditions. The Mito Stress Test medium contained DMEM 5030 supplemented with 10 mM D-glucose (GLU501.5, BioShop Canada), 1 mM sodium pyruvate (P2256, Sigma-Aldrich), 2 mM L-glutamine (609-065-EL, WISENT, INC), and 5 mM HEPES (600-032-CG, WISENT). The Glycolysis Stress Test medium contained DMEM 5030 (Sigma-Aldrich) supplemented with 2 mM glutamine. Following equilibration, OCR and ECAR were measured. The Seahorse Mito Stress Assay (Agilent) consisted of sequential injections of oligomycin (3 µM, O4876, Sigma-Aldrich, USA), FCCP (2 µM, ab120081, Abcam, USA), and rotenone (1 µM, R887, Sigma-Aldrich, USA)–antimycin A (1 µM, A8674, Sigma-Aldrich, USA) (RAA). For the Glycolysis Stress Assay, cells were treated with 10 mM D-glucose, 3 µM oligomycin, and 100 µM 2-deoxy-D-glucose (SC-202010A, Santa Cruz, USA). These assays were used to assess basal respiration and ATP-linked respiration in the Mito Stress Test, as well as glycolysis and glycolytic capacity in the Glycolysis Stress Test. Results were normalized to the number of cells per well (quantified with DAPI).

Retinal glucose uptake assay

Consumption of glucose was measured using 2-[³H]deoxyglucose (DG) as we reported⁴⁶. Mice received a daily intraperitoneal injection of 0.5 µL of 2-[³H]DG (1 µCi/µL) from P12 to P17. Animals were sacrificed at P17, 60 minutes after the last injection, to ensure cellular incorporation of the radiotracer. Retinas were collected and homogenized in a scintillation cocktail (Ecolite +; MP Biomedicals), and beta counts were measured using an LS6500 Multipurpose Scintillation Counter (Beckman). Disintegrations per minute (DPM) counts were normalized to the total injected dose and the weight of each animal.

Electroretinogram (ERG) analysis

Experiments were performed on P21 Sirt3^−/− and WT S129 mice after OIR according to ISCEV extended protocol for PhNR ERG⁴⁰. A computerized system with an electronically controlled Ganzfeld dome was used (Espion E2 with Colordome; Diagnosys LLC, USA). ERGs were recorded from the corneal surface of each eye after pupil dilation with ophthalmic solution (1% Isopto®, Atropine Sulfate and 2.5% Mydfrin™, phenylephrine hydrochloride) using a DTL plus fiber electrode™ (Diagnosys LLC, USA). For all the recordings, ground and reference needle electrodes were placed subcutaneously in the tail and in the front of the head, respectively. Impedance was maintained at <5 kΩ. P21 mice were anesthetized by intraperitoneal injection of ketamine (10 mg/kg) and xylazine (20 mg/kg), and the mice were placed on a temperature-regulated heating pad during recording periods. The photopic negative response (PhNR) is a component of the full-field photopic electroretinogram (ERG). The PhNR was elicited by a 10 and 25 cd·s/m² white flash presented on a 100 cd/m² cones-saturating green background 1-Hz and 25 sweeps were averaged for each intensity. All ERG responses were filtered at 0.312–45 Hz, and signal averaging was applied. The PhNR amplitude was measured from the baseline to the maximum amplitude in the trough of the negative wave that comes after the b-wave, and also measured from the peak of the b-wave to the maximum amplitude in the same trough (Global PhNR). The template file that allowed the analysis of the electroretinogram data is available at https://github.com/Aghilusius/Template-ERG-Analysis.

Quantification and statistical analysis

Statistical analysis was performed using GraphPad Prism (10.0.3) for Mac (GraphPad Software) or R statistical programming language (4.4.0, R-project.org). Exact n, statistical test, and significance are reported in the Figure Legends. Data are presented as means \(\pm\)± the standard error of the mean (s.e.m.). Between-subject statistical tests were completed with two-tailed independent Student’s t-tests. Comparisons between groups were made using 1-way or 2-way ANOVA followed by the post hoc Tukey or Bonferroni multiple comparisons test among means. The D’Agostino-Pearson or Kolmogorov-Smirnov normality tests were used to confirm a normal distribution. Animals were not randomized, but quantification of the data was performed in a blinded fashion when possible. Outlier results were identified and removed using the ROUT method (Q5%) in GraphPad Prism. Results are presented as mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. P < 0.05 was considered statistically significant.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

Data supporting the findings of this study are available in the article, its Supplementary information, and the Source Data file. The single-cell RNA-seq data have been deposited in GEO (GSE216676 https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE216676). Publicly available data (GSE174400 https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE174400, GSE150703 https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE150703, GSE148077 https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE148077, GSE165784 https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE165784) were also used in this study. The human metabolomics data has been deposited in Metabolomics Workbench under Project ID: PR002423; Project DOI: doi: 10.21228/M8FC13. Additional information is available from the corresponding authors upon request. Source data are provided with this paper.

Code availability

Custom-written scripts used in this study are available from the corresponding author upon reasonable request. Any additional information required to reanalyze data reported in this paper is available from the lead contact and on github https://github.com/gaelcge/OIRDropseq.

References

Sapieha, P. et al. Retinopathy of prematurity: understanding ischemic retinal vasculopathies at an extreme of life. J. Clin. Invest. 120, 3022–3032 (2010).
Article CAS PubMed PubMed Central Google Scholar
Manavski, Y. et al. Clonal expansion of endothelial cells contributes to ischemia-induced neovascularization. Circ. Res. 122, 670–677 (2018).
Article CAS PubMed Google Scholar
Duh, E. J., Sun, J. K. & Stitt, A. W. Diabetic retinopathy: current understanding, mechanisms, and treatment strategies. JCI Insight 2, e93751 (2017).
Article PubMed PubMed Central Google Scholar
Selvam, S., Kumar, T. & Fruttiger, M. Retinal vasculature development in health and disease. Prog. Retin. Eye Res. 63, 1–19 (2018).
Article CAS PubMed Google Scholar
Joyal, J. S., Gantner, M. L. & Smith, L. E. H. Retinal energy demands control vascular supply of the retina in development and disease: The role of neuronal lipid and glucose metabolism. Prog. Retin Eye Res. 64, 131–156 (2018).
Article CAS PubMed Google Scholar
Stone, J. et al. Development of retinal vasculature is mediated by hypoxia-induced vascular endothelial growth factor (VEGF) expression by neuroglia. J. Neurosci. J. Soc. Neurosci. 15, 4738–4747 (1995).
Article CAS Google Scholar
Rattner, A., Williams, J. & Nathans, J. Roles of HIFs and VEGF in angiogenesis in the retina and brain. J. Clin. Investig. 129, 3807–3820 (2019).
Article PubMed PubMed Central Google Scholar
Sapieha, P. et al. The succinate receptor GPR91 in neurons has a major role in retinal angiogenesis. Nat. Med. 14, 1067–1076 (2008).
Article CAS PubMed Google Scholar
Joyal, J.-S. et al. Ischemic neurons prevent vascular regeneration of neural tissue by secreting semaphorin 3A. Blood 117, 6024–6035 (2011).
Article CAS PubMed PubMed Central Google Scholar
Fukushima, Y. et al. Sema3E-PlexinD1 signaling selectively suppresses disoriented angiogenesis in ischemic retinopathy in mice. J. Clin. Investig. 121, 1974–1985 (2011).
Article CAS PubMed PubMed Central Google Scholar
Pierce, E. A., Avery, R. L., Foley, E. D., Aiello, L. P. & Smith, L. E. Vascular endothelial growth factor/vascular permeability factor expression in a mouse model of retinal neovascularization. Proc. Natl. Acad. Sci. USA 92, 905–909 (1995).
Article ADS CAS PubMed PubMed Central Google Scholar
Dorrell, M. I. et al. Maintaining retinal astrocytes normalizes revascularization and prevents vascular pathology associated with oxygen-induced retinopathy. Glia 58, 43–54 (2010).
Article PubMed PubMed Central Google Scholar
Eelen, G., de Zeeuw, P., Simons, M. & Carmeliet, P. Endothelial cell metabolism in normal and diseased vasculature. Circ. Res. 116, 1231–1244 (2015).
Article CAS PubMed PubMed Central Google Scholar
De Bock, K. et al. Role of PFKFB3-driven glycolysis in vessel sprouting. Cell 154, 651–663 (2013).
Article PubMed Google Scholar
Buchwald, P. A local glucose-and oxygen concentration-based insulin secretion model for pancreatic islets. Theor. Biol. Med. Model 8, 20 (2011).
Article CAS PubMed PubMed Central Google Scholar
Gatenby, R. A. & Gillies, R. J. Why do cancers have high aerobic glycolysis?. Nat. Rev. Cancer 4, 891–899 (2004).
Article CAS PubMed Google Scholar
Schoors, S. et al. Partial and transient reduction of glycolysis by PFKFB3 blockade reduces pathological angiogenesis. Cell Metab. 19, 37–48 (2014).
Article CAS PubMed Google Scholar
Schoors, S. et al. Fatty acid carbon is essential for dNTP synthesis in endothelial cells. Nature 520, 192–197 (2015).
Article ADS CAS PubMed PubMed Central Google Scholar
Houtkooper, R. H., Pirinen, E. & Auwerx, J. Sirtuins as regulators of metabolism and healthspan. Nat. Rev. Mol. Cell Biol. 13, 225–238 (2012).
Article CAS PubMed PubMed Central Google Scholar
Ahn, B.-H. et al. A role for the mitochondrial deacetylase Sirt3 in regulating energy homeostasis. Proc. Natl. Acad. Sci. USA 105, 14447–14452 (2008).
Article ADS CAS PubMed PubMed Central Google Scholar
He, W., Newman, J. C., Wang, M. Z., Ho, L. & Verdin, E. Mitochondrial sirtuins: regulators of protein acylation and metabolism. Trends Endocrinol. Metab. 23, 467–476 (2012).
Article CAS PubMed Google Scholar
Hirschey, M. D. et al. SIRT3 regulates mitochondrial fatty-acid oxidation by reversible enzyme deacetylation. Nature 464, 121–125 (2010).
Article ADS CAS PubMed PubMed Central Google Scholar
Xia, J. & Wishart, D. S. Using metaboAnalyst 3.0 for comprehensive metabolomics data analysis. Curr. Protoc. Bioinforma. 55, 14.10.11–14.10.91 (2016).
Google Scholar
Paris, L. P. et al. Global metabolomics reveals metabolic dysregulation in ischemic retinopathy. Metabolomics 12, 15 (2016).
Article PubMed Google Scholar
Nawaz, I. M. et al. Human vitreous in proliferative diabetic retinopathy: Characterization and translational implications. Prog. Retin Eye Res. 72, 100756 (2019).
Article CAS PubMed Google Scholar
Hu, Z. et al. Single-cell transcriptomics reveals novel role of microglia in fibrovascular membrane of proliferative diabetic retinopathy. Diabetes 71, 762–773 (2022).
Article CAS PubMed Google Scholar
Smith, L. E. et al. Oxygen-induced retinopathy in the mouse. Investig. Ophthalmol. Vis. Sci. 35, 101–111 (1994).
CAS Google Scholar
Binet, F. et al. Neutrophil extracellular traps target senescent vasculature for tissue remodeling in retinopathy. Science 369, eaay5356 (2020).
Article CAS PubMed Google Scholar
Zarkada, G. et al. Specialized endothelial tip cells guide neuroretina vascularization and blood-retina-barrier formation. Dev. Cell 56, 2237–2251.e2236 (2021).
Article CAS PubMed PubMed Central Google Scholar
Smith, T. L. et al. eNOS controls angiogenic sprouting and retinal neovascularization through the regulation of endothelial cell polarity. Cell. Mol. Life Sci. 79, 37 (2022).
Article CAS Google Scholar
Schey, K. L., Wang, Z., L. Wenke, J. & Qi, Y. Aquaporins in the eye: Expression, function, and roles in ocular disease. Biochim. et. Biophys. Acta (BBA) - Gen. Subj. 1840, 1513–1523 (2014).
Article CAS Google Scholar
Yu, W., Dittenhafer-Reed, K. E. & Denu, J. M. SIRT3 protein deacetylates isocitrate dehydrogenase 2 (IDH2) and regulates mitochondrial redox status. J. Biol. Chem. 287, 14078–14086 (2012).
Article CAS PubMed PubMed Central Google Scholar
Cimen, H. et al. Regulation of succinate dehydrogenase activity by SIRT3 in mammalian mitochondria. Biochemistry 49, 304–311 (2010).
Article CAS PubMed Google Scholar
Vassilopoulos, A. et al. SIRT3 deacetylates ATP synthase F1 complex proteins in response to nutrient- and exercise-induced stress. Antioxid. Redox Signal 21, 551–564 (2014).
Article CAS PubMed PubMed Central Google Scholar
Yang, W. et al. Mitochondrial Sirtuin network reveals dynamic SIRT3-dependent deacetylation in response to membrane depolarization. Cell 167, 985–1000.e1021 (2016).
Article CAS PubMed PubMed Central Google Scholar
Skinnider, M. A. et al. Cell type prioritization in single-cell data. Nat. Biotechnol. 39, 30–34 (2021).
Article CAS PubMed Google Scholar
O’Carroll, S. J., Cook, W. H. & Young, D. AAV targeting of glial cell types in the central and peripheral nervous system and relevance to human gene therapy. Front. Mol. Neurosci. 13, 618020 (2021).
Article PubMed PubMed Central Google Scholar
Meng, H., Yaari, G., Bolen, C. R., Avey, S. & Kleinstein, S. H. Gene set meta-analysis with Quantitative Set Analysis for Gene Expression (QuSAGE). PLoS Comput. Biol. 15, e1006899 (2019).
Article ADS PubMed PubMed Central Google Scholar
Helker, C. S. et al. Apelin signaling drives vascular endothelial cells toward a pro-angiogenic state. Elife 9, e55589 (2020).
Frishman, L. et al. ISCEV extended protocol for the photopic negative response (PhNR) of the full-field electroretinogram. Doc. Ophthalmol. 136, 207–211 (2018).
Article PubMed PubMed Central Google Scholar
Claesson-Welsh, L., Dejana, E. & McDonald, D. M. Permeability of the endothelial barrier: identifying and reconciling controversies. Trends Mol. Med. 27, 314–331 (2021).
Article CAS PubMed Google Scholar
Zou, L. B. et al. Aquaporin-1 plays a crucial role in estrogen-induced tubulogenesis of vascular endothelial cells. J. Clin. Endocrinol. Metab. 98, E672–E682 (2013).
Article CAS PubMed Google Scholar
Saadoun, S., Papadopoulos, M. C., Hara-Chikuma, M. & Verkman, A. S. Impairment of angiogenesis and cell migration by targeted aquaporin-1 gene disruption. Nature 434, 786–792 (2005).
Article ADS CAS PubMed Google Scholar
Li, X., Sun, X. & Carmeliet, P. Hallmarks of endothelial cell metabolism in health and disease. Cell Metab. 30, 414–433 (2019).
Article CAS PubMed Google Scholar
Lee, H. W. et al. Role of venous endothelial cells in developmental and pathologic angiogenesis. Circulation 144, 1308–1322 (2021).
Article CAS PubMed PubMed Central Google Scholar
Joyal, J.-S. et al. Retinal lipid and glucose metabolism dictates angiogenesis through the lipid sensor Ffar1. Nat. Med. 22, 439–445 (2016).
Article CAS PubMed PubMed Central Google Scholar
Mackenzie, F. & Ruhrberg, C. Diverse roles for VEGF-A in the nervous system. Development 139, 1371–1380 (2012).
Article CAS PubMed Google Scholar
The STOP-ROP Multicenter Study Group. Supplemental Therapeutic Oxygen for Prethreshold Retinopathy of Prematurity (STOP-ROP), a randomized, controlled trial. I: primary outcomes. Pediatrics 105, 295–310 (2000).
Network, S. et al. Target ranges of oxygen saturation in extremely preterm infants. N. Engl. J. Med. 362, 1959–1969 (2010).
Article Google Scholar
Jain, I. H. et al. Hypoxia as a therapy for mitochondrial disease. Science 352, 54–61 (2016).
Article ADS CAS PubMed PubMed Central Google Scholar
Bertout, J. A., Patel, S. A. & Simon, M. C. The impact of O₂ availability on human cancer. Nat. Rev. Cancer 8, 967–975 (2008).
Article CAS PubMed PubMed Central Google Scholar
Hoppe, G. et al. Comparative systems pharmacology of HIF stabilization in the prevention of retinopathy of prematurity. Proc. Natl Acad. Sci. 113, E2516–2525 (2016).
Finley, L. W. S. et al. SIRT3 opposes reprogramming of cancer cell metabolism through HIF1α destabilization. Cancer Cell 19, 416–428 (2011).
Article CAS PubMed PubMed Central Google Scholar
Ban, N. et al. Neuroprotective role of retinal SIRT3 against acute photo-stress. NPJ Aging Mech. Dis. 3, 19 (2017).
Lin, J. B. et al. NAMPT-mediated NAD + biosynthesis is essential for vision in mice. Cell Rep. 17, 69–85 (2016).
Article CAS PubMed PubMed Central Google Scholar
Lin, J. B., Lin, J. B., Chen, H. C., Chen, T. & Apte, R. S. Combined SIRT3 and SIRT5 deletion is associated with inner retinal dysfunction in a mouse model of type 1 diabetes. Sci. Rep. 9, 3799 (2019).
Article ADS PubMed PubMed Central Google Scholar
Kane, A. E. & Sinclair, D. A. Sirtuins and NAD(+) in the development and treatment of metabolic and cardiovascular diseases. Circ. Res. 123, 868–885 (2018).
Article CAS PubMed PubMed Central Google Scholar
Oellerich, M. F. & Potente, M. FOXOs and sirtuins in vascular growth, maintenance, and aging. Circ. Res. 110, 1238–1251 (2012).
Article CAS PubMed Google Scholar
Williamson, M. R., Fuertes, C. J. A., Dunn, A. K., Drew, M. R. & Jones, T. A. Reactive astrocytes facilitate vascular repair and remodeling after stroke. Cell Rep. 35, 109048 (2021).
Article CAS PubMed PubMed Central Google Scholar
Mills, E. L. et al. Accumulation of succinate controls activation of adipose tissue thermogenesis. Nature 560, 102–106 (2018).
Article ADS CAS PubMed PubMed Central Google Scholar
Macosko, E. Z. et al. Highly parallel genome-wide expression profiling of individual cells using nanoliter droplets. Cell 161, 1202–1214 (2015).
Article CAS PubMed PubMed Central Google Scholar
Satija, R., Farrell, J. A., Gennert, D., Schier, A. F. & Regev, A. Spatial reconstruction of single-cell gene expression data. Nat. Biotechnol. 33, 495–502 (2015).
Article CAS PubMed PubMed Central Google Scholar
Villanueva, R. A. M. & Chen, Z. J. ggplot2: Elegant graphics for data analysis, 2nd edition. Meas. -Interdiscip. Res. 17, 160–167 (2019).
Google Scholar
Hänzelmann, S., Castelo, R. & Guinney, J. GSVA: gene set variation analysis for microarray and RNA-Seq data. BMC Bioinform. 14, 7 (2013).
Kuleshov, M. V. et al. Enrichr: a comprehensive gene set enrichment analysis web server 2016 update. Nucleic Acids Res. 44, W90–W97 (2016).
Article CAS PubMed PubMed Central Google Scholar
Macnair, W., Gupta, R. & Claassen, M. psupertime: supervised pseudotime analysis for time-series single-cell RNA-seq data. Bioinformatics 38, i290–i298 (2022).
Article PubMed PubMed Central Google Scholar
Browaeys, R., Saelens, W. & Saeys, Y. NicheNet: modeling intercellular communication by linking ligands to target genes. Nat. Methods 17, 159–162 (2020).
Article CAS PubMed Google Scholar
Aibar, S. et al. SCENIC: single-cell regulatory network inference and clustering. Nat. Methods 14, 1083–1086 (2017).
Article CAS PubMed PubMed Central Google Scholar
Qiu, X. et al. Reversed graph embedding resolves complex single-cell trajectories. Nat. Methods 14, 979–982 (2017).
Article CAS PubMed PubMed Central Google Scholar
Melsted, P. et al. Modular, efficient and constant-memory single-cell RNA-seq preprocessing. Nat. Biotechnol. 39, 813–818 (2021).
Article CAS PubMed Google Scholar
La Manno, G. et al. RNA velocity of single cells. Nature 560, 494–498 (2018).
Article ADS PubMed PubMed Central Google Scholar
Joyal, J.-S. et al. Subcellular localization of coagulation factor II receptor-like 1 in neurons governs angiogenesis. Nat. Med. 20, 1165–1173 (2014).
Stahl, A. et al. Computer-aided quantification of retinal neovascularization. Angiogenesis 12, 297–301 (2009).
Article CAS PubMed PubMed Central Google Scholar

Download references

Acknowledgements

We thank Marie-Josée Lacombe, Ferhat Djail, Zhongjie (Kira) Fu and Adriana Di Polo for their technical assistance. We thank Elke Küster-Schöck and the Platform for Imaging by Microscopy of CHU Sainte-Justine Research Center (CHUSJRC), supported by Leica Microsystems and funded by CHUSJRC, the Quebec government, CHUSJ Foundation and Canada Foundation for Innovation and the Vision Health Research Network for funding of the Single-cell Genomics Analysis Platform. GC was supported by a MITACS Elevate fellowship. J.-S.J. was supported by the Burroughs Wellcome Fund, the Canadian Institute of Health Research (CIHR; 390615, 479607, 506601), the National Sciences and Engineering Research Council of Canada (NSERC; 06743), and the Fonds de Recherche du Québec–Santé (FRQS). P.S. was supported by CIHR (353770), Heart & Stroke Foundation Canada (G-16-00014658), Foundation Fighting Blindness Canada, and the Canadian Diabetes Association (DI-3-18-5444-PS).

Author information

These authors contributed equally: Gael Cagnone, Sheetal Pundir.

Authors and Affiliations

Department of Pharmacology and Physiology, Université de Montréal, Montreal, QC, Canada
Gael Cagnone, Charlotte Betus, Emilie Heckel, Ashim Das, Louis Berillon, Alexandre Dubrac & Jean-Sébastien Joyal
CHU Sainte-Justine Azrieli Research Center, Montreal, QC, Canada
Gael Cagnone, Sheetal Pundir, Charlotte Betus, Tapan Agnihotri, Anli Ren, Jin Sung -Kim, Emilie Heckel, Mei Xi Chen, Anu Situ, Perrine Gaub, Nicholas Kim, Ashim Das, Severine Leclerc, Florian Wünnemann, Louis Berillon, Gregor Andelfinger, Alexandre Dubrac, José Carlos Rivera & Jean-Sébastien Joyal
Department of Pharmacology and Therapeutics, McGill University, Montreal, QC, Canada
Sheetal Pundir, Tapan Agnihotri, Jin Sung -Kim, Nicholas Kim & Jean-Sébastien Joyal
Department of Ophthalmology, Zhongnan Hospital of Wuhan University, Wuhan, China
Anli Ren
Department of General Surgery, Northern Ontario School Medicine, Sudbury, ON, Canada
Noémie-Rose Harvey
Heidelberg University, Faculty of Medicine, and Heidelberg University Hospital, Institute for Computational Biomedicine, Bioquant, Heidelberg, Germany
Florian Wünnemann
Department of Pediatrics, CHU Sainte-Justine, Université de Montréal, Montreal, QC, Canada
Gregor Andelfinger & Jean-Sébastien Joyal
Optometry School, Université de Montréal, Montreal, QC, Canada
Sergio Crespo-Garcia
Department of Ophthalmology, Université de Montréal, Montreal, QC, Canada
Alexandre Dubrac, Flavio A. Rezende, José Carlos Rivera, Przemyslaw Sapieha & Jean-Sébastien Joyal
Department of Pathology and Cell Biology, Université de Montréal, Montreal, Canada
Alexandre Dubrac
Hôpital Maisonneuve-Rosemont, Montreal, QC, Canada
Flavio A. Rezende & Przemyslaw Sapieha
Metabolomics Platform, The Broad Institute of MIT and Harvard University, Cambridge, MA, USA
Clary B. Clish
Department of Pediatrics, Université de Sherbrooke, Sherbrooke, QC, Canada
Bruno Maranda
Department of Ophthalmology, Boston Children’s Hospital, Harvard Medical School, Boston, MA, USA
Lois E. H. Smith

Authors

Gael Cagnone
View author publications
Search author on:PubMed Google Scholar
Sheetal Pundir
View author publications
Search author on:PubMed Google Scholar
Charlotte Betus
View author publications
Search author on:PubMed Google Scholar
Tapan Agnihotri
View author publications
Search author on:PubMed Google Scholar
Anli Ren
View author publications
Search author on:PubMed Google Scholar
Jin Sung -Kim
View author publications
Search author on:PubMed Google Scholar
Noémie-Rose Harvey
View author publications
Search author on:PubMed Google Scholar
Emilie Heckel
View author publications
Search author on:PubMed Google Scholar
Mei Xi Chen
View author publications
Search author on:PubMed Google Scholar
Anu Situ
View author publications
Search author on:PubMed Google Scholar
Perrine Gaub
View author publications
Search author on:PubMed Google Scholar
Nicholas Kim
View author publications
Search author on:PubMed Google Scholar
Ashim Das
View author publications
Search author on:PubMed Google Scholar
Severine Leclerc
View author publications
Search author on:PubMed Google Scholar
Florian Wünnemann
View author publications
Search author on:PubMed Google Scholar
Louis Berillon
View author publications
Search author on:PubMed Google Scholar
Gregor Andelfinger
View author publications
Search author on:PubMed Google Scholar
Sergio Crespo-Garcia
View author publications
Search author on:PubMed Google Scholar
Alexandre Dubrac
View author publications
Search author on:PubMed Google Scholar
Flavio A. Rezende
View author publications
Search author on:PubMed Google Scholar
Clary B. Clish
View author publications
Search author on:PubMed Google Scholar
Bruno Maranda
View author publications
Search author on:PubMed Google Scholar
José Carlos Rivera
View author publications
Search author on:PubMed Google Scholar
Lois E. H. Smith
View author publications
Search author on:PubMed Google Scholar
Przemyslaw Sapieha
View author publications
Search author on:PubMed Google Scholar
Jean-Sébastien Joyal
View author publications
Search author on:PubMed Google Scholar

Contributions

G.C., S.P., and J-S.J. conceived and designed all experiments and wrote the manuscript; F.R. and P.S. collected human samples; C.B.C. generated the metabolomic data of human samples; G.C. analyzed the human metabolomics data; B.M. and S.P. performed the carnitine quantification of mouse retinal sample, and G.C. analyzed the data; J.C.R., T.A., A.R., S.P. and P.G. performed the Seahorse experiments; G.C., F.W., S.L., and G.A. generated single-cell RNAseq (Drop-seq) data; G.C., A.S., and M.X.C. analyzed single-cell RNAseq data; S.P., J.S.K., N.R.H., A.Das, T.A., A.R. and C.B. performed OIR experiments and phenotypic quantification; S.P, J.S.K., E.H., N.R.H., T.A., J.C.R., N.K., L.B. and A.R. performed in vivo and in vitro metabolic experiments; J.C.R. performed intravitreal injection and immunohistochemistry experiments; A.R. performed in vitro culture and qPCR analysis on astrocytes; E.H. performed RNAscope experiments; C.B. performed ERG experiments; S.C.G. provided graphic design expertise; A.Dubrac., L.E.H.S. and P.S. offered valuable insight regarding retinal vascular biology; all authors analyzed the data.

Corresponding author

Correspondence to Jean-Sébastien Joyal.

Ethics declarations

Competing interests

The authors have no conflicts of interest to declare relevant to this article’s content. F.R. is an advisory board member for Bayer AG (Germany) and F. Hoffmann-La Roche AG (Switzerland). P.S. is the founder and Chief Scientific Officer of SemaThera Inc. (Montreal, Canada) and Chief Scientist of UNITY Biotechnology (California, US) and holds equity in both companies.

Peer review

Peer review information

Nature Communications thanks Rui Chen, and the other, anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary Information

Reporting Summary

Transparent Peer Review file

Source data

Source Data

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

About this article

Cite this article

Cagnone, G., Pundir, S., Betus, C. et al. Metabolic reprogramming of the neovascular niche promotes regenerative angiogenesis in proliferative retinopathy. Nat Commun 16, 5377 (2025). https://doi.org/10.1038/s41467-025-60061-4

Download citation

Received: 06 December 2023
Accepted: 13 May 2025
Published: 25 June 2025
DOI: https://doi.org/10.1038/s41467-025-60061-4