Sirt6 protects retinal ganglion cells and optic nerve from degeneration during aging and glaucoma

Abstract

Glaucoma is characterized by the progressive degeneration of retinal ganglion cells (RGCs) and their axons, and its risk increases with aging. Yet comprehensive insights into the complex mechanisms are largely unknown. Here, we found that anti-aging molecule Sirt6 was highly expressed in RGCs. Deleting Sirt6 globally or specifically in RGCs led to progressive RGC loss and optic nerve degeneration during aging, despite normal intraocular pressure (IOP), resembling a phenotype of normal-tension glaucoma. These detrimental effects were potentially mediated by accelerated RGC senescence through Caveolin-1 upregulation and by the induction of mitochondrial dysfunction. In mouse models of high-tension glaucoma, Sirt6 level was decreased after IOP elevation. Genetic overexpression of Sirt6 globally or specifically in RGCs significantly attenuated high tension-induced degeneration of RGCs and their axons, whereas partial or RGC-specific Sirt6 deletion accelerated RGC loss. Importantly, therapeutically targeting Sirt6 with pharmacological activator or AAV2-mediated gene delivery ameliorated high IOP-induced RGC degeneration. Together, our studies reveal a critical role of Sirt6 in preventing RGC and optic nerve degeneration during aging and glaucoma, setting the stage for further exploration of Sirt6 activation as a potential therapy for glaucoma.

Graphical abstract

Xia and colleagues unraveled a novel role of anti-aging molecule Sirt6 in preventing degeneration of retinal ganglion cells and their axons during aging and glaucoma, and provided compelling evidence that therapeutically targeting Sirt6 with pharmacological activator or AAV2-mediated gene delivery ameliorated RGC degeneration in glaucoma.

Introduction

Glaucoma ranks second globally as a cause of irreversible vision loss.¹ Elevated intraocular pressure (IOP), aging, and genetic background are main risk factors for glaucoma. Among them, only IOP is well-studied and serves as the sole target for therapeutic interventions in glaucoma.² However, lowering IOP does not always stop the progression of glaucoma. Moreover, in the case of normal-tension glaucoma (NTG), where optic nerve damage occurs in the absence of elevated IOP, the underlying pathogenic mechanisms remain poorly understood.^3,4,5,6 Thus, there is an urgent need to identify other mechanisms contributing to neuropathy in glaucoma in order to facilitate the development of targeted therapeutic interventions.

Glaucoma is intricately associated with the aging process, yet comprehensive insights into the complex mechanisms of aging and their specific impact on retinal ganglion cells (RGCs) and their axons are largely unknown. Among the potential regulatory factors, Sirtuins (Sirts) emerge as noteworthy candidates. Sirts belong to a widely preserved protein family that has similarities with the yeast Sir2 protein, which operates as an NAD-dependent histone deacetylase, prolonging the yeast’s lifespan.^7,8 In mammals, seven different Sirts (Sirt1–7) have been identified. Sirt1, Sirt6, and Sirt7 are primarily located in the nucleus, regulating transcriptional activities by acting on transcription factors, co-factors, or histones. Conversely, Sirt3, Sirt4, and Sirt5 mainly exist within the mitochondria, playing roles in the modulation of mitochondrial energy metabolism. Sirt2 is mostly in the cytoplasm, serving as a tubulin deacetylase.⁹ Among these Sirts, Sirt6 is strikingly similar to yeast Sir2 when considering its intracellular location, functional features, and the phenotypic outcomes observed upon Sirt6 deletion in animal models.⁷ Notably, while both Sirt1 and Sirt6 exhibit anti-aging function, it is Sirt6 rather than Sirt1 that is proven to promote longevity, although Sirt1 overexpression has demonstrated improvements in various health measures.¹⁰

As a key anti-aging molecule, Sirt6 plays critical roles in aging-related diseases such as cancer, bone loss, cardiovascular disease, and nephropathy.^{8,11,12,13,14} Sirt6 is expressed in the retina, but its function is largely unknown, especially under pathophysiological conditions of the retina, including glaucoma. Considering the critical role that Sirt6 plays in age-related conditions, we assessed the role of Sirt6 in glaucoma in our novel Sirt6 transgenic mice, including global and conditional knockout/knockin strains, in conjunction with three well-established mouse models of ocular hypertension, collectively encompassing diverse aspects of glaucomatous pathology. We found that Sirt6 deficiency induced RGC death and optic nerve degeneration during aging. These detrimental effects were potentially mediated by accelerated RGC senescence through Caveolin-1 upregulation and by the induction of mitochondrial dysfunction. Furthermore, overexpression of Sirt6 globally or specifically in RGCs alleviated high IOP-induced glaucomatous neurodegeneration, whereas partial or RGC-specific Sirt6 deletion accelerated RGC loss. More importantly, pharmacological activation of Sirt6 or boosting Sirt6 expression with adeno-associated virus 2 (AAV2)-mediated gene therapy effectively countered high IOP-induced RGC loss, demonstrating the therapeutic promise of Sirt6 activation.

Results

Sirt6 deletion contributes to progressive RGC loss and optic nerve degeneration with aging

To determine whether Sirt6 has a role in maintaining the health of the retina, we examined the expression pattern of Sirt6 in the retina, together with Sirt1, which is another well-known anti-aging molecule in the sirtuin family (Figure S1A). We found that Sirt6 was abundantly expressed in cells within the ganglion cell layer (GCL), although weak immunoreactivity of Sirt6 was also noted in the inner nuclear layer (INL) and outer nuclear layer (ONL). In contrast, Sirt1 immunoreactivity was uniformly distributed across retinal cell types. The predominant expression of Sirt6 in RGCs suggests its potential involvement in the regulation of RGC function. Next, we generated Sirt6 global knockout (KO) (Sirt6^−/−) mice on a mixed genetic background of C57BL/6 and 129/SvJ. In contrast to the conventional Sirt6^−/− mice, which are reared on a pure C57BL/6 background and typically die within 4 weeks, the mixed-background Sirt6^−/− mice exhibited an extended lifespan of up to 1 year,¹¹ enabling the study of retinal alterations in matured animals. We first confirmed the deletion of Sirt6 in the retina by western blotting and immunostaining (Figures S1B and S1C). We also found that the acetylation levels of both H3K9 and H3K56, which are known physiological substrates of Sirt6, were increased in Sirt6^−/− retina, indicating functional loss of Sirt6 (Figure S1C). Moreover, Sirt6 deletion did not alter the expression of Sirt1 in the retina (Figure S1B).

To further evaluate the role of Sirt6 in the retina, retinal morphology was assessed at 2 months and 7 months of age. At 2 months of age, Sirt6^−/− mice exhibited a similar retinal structure (Figure S1D) and comparable RGC number (Figure 1A, upper panels) to their wild-type (WT) littermates, as determined by H&E staining and immunostaining for RGC marker Tuj1 in retinal flatmounts, respectively. However, at 7 months of age, Sirt6^−/− mice demonstrated significant decrease in RGC density revealed by the staining of RGC marker RBPMS (Figure 1A, lower panels). More strikingly, RGC axon bundles in Sirt6^−/− retina were thinner and shorter compared with those in WT retina (Figure 1B). In addition, in the Sirt6^−/− retinal surface, increased microglia were observed, acquiring morphological features indicative of activation (Figures 1C and 1D). To ascertain the evolution of axonal changes, we examined the optic nerves of WT and Sirt6^−/− mice at 7 months of age (Figures 1E–1L). We found that the optic nerve of Sirt6^−/− mice was 40% thinner than that from WT mice (Figures 1E, 1F, and 1M), along with enhanced glial activation as indicated by increased fluorescence intensity of glial fibrillary acidic protein (GFAP) staining (Figures 1F and 1G) and notable infiltration of inflammatory cells as indicated by DAPI and Iba1 staining (Figures 1H, 1I, 1N, and 1O). To examine the integrity of axons in the optic nerve, we performed immunostaining with neurofilament marker SMI-31 (for phosphorylated heavy neurofilament chain [pNFH]) and myelin basic protein (MBP, for myelin sheath) antibodies on the transverse sections of the optic nerve. We found that the immunoreactivity of both SMI-31 (Figures 1J and 1P) and MBP (Figures 1K and 1Q) was decreased in Sirt6-deficient optic nerves, as has been observed in glaucoma-related models,^15,16 implicating the loss of intact axons and myelinated axons. The ultrastructural changes in the optic nerve were further assessed by electron microscopy, showing that Sirt6 deficiency dramatically decreased the number of axons and the thickness of myelin sheath (Figure 1L). As RGC loss and optic nerve degeneration are often caused by elevated IOP, we measured IOP of Sirt6^−/− and WT littermates at 2 and 7 months of age but found that the mean IOP of Sirt6^−/− mice remained within the normal physiological range and indistinguishable from that of WT littermates at both ages (Figure 1R). Therefore, Sirt6 deletion-related degeneration of RGC and their axons occurs independently of IOP change.

Figure 1.

Sirt6 deletion induces RGC loss and optic nerve degeneration

(A) RGCs were stained with anti-Tuj1 or RBPMS, and images were taken in the peripheral retinas of WT and Sirt6^−/− mice at 2 and 7 months of age. The number of RGCs was quantified. n = 6 for 2-month-old and n = 9 for 7-month-old mice. (B) Representative images of RGC axons in the retinas of 7-month-old WT and Sirt6^−/− mice. Squares in the left panels are zoomed in to show the details of axons. (C and D) Microglia were labeled with anti-Iba1 in the retinas of WT and Sirt6^−/− mice at 7 months of age, and microglial number was quantified. n = 3. (E–Q) Optic nerves were collected from 7-month-old WT and Sirt6^−/− mice for analysis. Representative images of optic nerve (E). Staining on the cross-sections of the optic nerves with anti-GFAP for glial activation (F at lower magnification and G at higher magnification), DAPI for the nuclei (H), anti-Iba1 for microglia (I), anti-pNFH (J), and MBP (K) for axons. Images of optic nerves by electron microscopy (L). Bar graphs represent the quantifications of relative cross-sectional area (M), nuclei (N), microglial number (O), and pNFH⁺ and MBP⁺ axons (P and Q). n = 3–5. (R) IOP of 2- and 7-month-old WT and Sirt6^−/− littermate mice was measured. n = 10–12. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001. Scale bar, 50 μm, except (L) where scale bar, 0.5 μm for upper panels and 200 nm for lower panels. Image size of (A): 159.73 μm × 159.73 μm; image size of upper panels of (C): 400 μm × 500 μm; image size of (G)–(K): 143.36 μm × 143.36 μm.

To rule out the possibility that the observed degeneration of RGC and their axons in Sirt6 global KO mice is due to systemic dysfunction, we further assessed the specific role of Sirt6 in the retina, particularly in RGCs, using AAV2-mediated gene deletion. We intravitreally injected AAV2 carrying the Cre recombinase to Sirt6^flox/flox mice that are on a pure C57BL/6 background and exhibit indistinguishable retinal structure as WT mice (Figure S2). Our prior work, involving intravitreal injection of AAV2-GFP, demonstrated that AAV2 could deliver genes into RGCs efficiently and specifically,¹⁷ which was consistent with previous studies showing a specific tropism of AAV2 for RGCs.^18,19 In Sirt6^flox/flox mice receiving AAV2-Cre, Sirt6 protein expression was decreased in the retina, in particular RGCs, compared with those receiving AAV2-Null (Figures 2A and 2B). Consistent with the results from Sirt6^−/− mice, AAV2-Cre-induced Sirt6 specific deletion in RGCs led to significant RGC loss (16.5%) in the retina at 8 months after AAV2 injection (Figures 2C and 2J). Moreover, optic nerve thinning, glial activation, inflammatory cell infiltration, axonal degeneration, and loss of myelin sheaths were observed on the cross-sections of the optic nerves at 8 months after AAV2 injection (Figures 2D–2O), which resembled pathological changes in the optic nerve of Sirt6 global KO mice. With aging, Sirt6^flox/flox mice with AAV2-Cre injection exhibited further increases in RGC loss (28.8% loss at 12 months, and 49.6% loss at 20 months, respectively) (Figure 2P), concurrent with axon thinning (Figure 2Q), microglial activation (Figure 2R), and optic nerve degeneration (Figure 2S) at 12 and 20 months after AAV2 injection.

Figure 2.

Sirt6 deletion in RGCs progressively induces RGC loss and optic nerve degeneration with aging

Two-month-old Sirt6^flox/flox mice were intravitreally injected with AAV2-Null as control or AAV2-Cre for Sirt6 deletion. (A–O) At 8 months after injection, retinas, eyeballs, and optic nerves were collected for analysis. Sirt6 deletion in the retina was validated by western blot (A) and immunostaining (B). n = 3. RGCs were stained with anti-RBPMS and images taken in the peripheral retinas are shown (C). Staining on the cross-sections of the optic nerves with anti-GFAP for glial activation at lower magnification (D) and at higher magnification (E), DAPI for the nuclei (F), anti-Iba1 for microglia (G), anti-pNFH (H), and MBP (I) for axons. Bar graphs represent the quantifications of RGC number (J), relative cross-sectional area (K), nuclei (L), microglia number (M), and pNFH⁺ and MBP⁺ axons (N and O). n = 4–5. (P–S) At 12 months or 20 months after AAV2 injection, eyeballs were collected for analysis. (P) Representative images of RGCs labeled with anti-RBPMS in the peripheral retinas are shown, and the number of RGCs was quantified. n = 4–6. (Q) RGC axons in the retina. n = 3. (R) Microglia were stained with anti-Iba1 in retinal flatmounts, and the number of microglia was quantified. n = 3–7. (S) Representative images of optic nerve. n = 3. ∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001. Scale bar, 50 μm. Image size of (C) and (P): 319.45 μm × 319.45 μm; image size of (E)–(I): 143.36 μm × 143.36 μm; image size of (R): 159.73 μm × 159.73 μm. GCL, ganglion cell layer; INL, inner nuclear layer; ONL, outer nuclear layer.

Taken together, these data collectively indicate that Sirt6 that is highly expressed in RGCs plays an important role in the maintenance of RGCs and their axons during aging, and the loss of Sirt6 leads to retinal pathology resembling glaucomatous neurodegeneration.

Loss of Sirt6 accelerates RGC senescence by upregulating Caveolin-1

To understand mechanisms by which Sirt6 preserves RGCs and their axons during aging, we investigated senescence-associated phenotype after Sirt6 deletion, given that Sirt6 is an anti-aging molecule. Senescent cells, regardless of being caused by replicative senescence or stress-induced premature senescence (SIPS), exhibit several common features, including a positive reaction for senescence-associated β-galactosidase (SA-β-gal) activity and the release of numerous biologically active factors, known as the senescence-associated secretory phenotype (SASP).²⁰ The SASP comprises many pro-inflammatory cytokines, chemokines, proteases, and growth factors that can trigger chronic inflammation and has the potential to promote age-related degenerative pathology.²⁰ Therefore, we stained retinas from 7-month-old Sirt6^−/− mice using the CellEvent Senescence Green Detection Kit, which enables the image-based detection of senescent cells, together with anti-Tuj1 antibody to identify RGCs. We found that there were more SA-β-gal-positive cells in the retinas from Sirt6^−/− mice than those from WT littermate mice, and SA-β-gal-positive cells were co-localized with Tuj1-positive cells (Figure 3A). In AAV2-Cre-infected Sirt6^flox/flox retinas, the SASP, as indicated by vascular inflammation (leukocyte attachment to retinal vessels ‒ leukostasis), was also significantly increased after Sirt6 was deleted (Figure 3B). These data provide compelling evidence that loss of Sirt6 accelerates the process of RGC senescence.

Figure 3.

Caveolin-1 is involved in Sirt6-regulated RGC senescence

(A) Retinal flatmounts from 7-month-old WT and Sirt6^−/− mice were stained for SA-β-gal (green) for senescence and anti-Tuj1 (red) for RGCs. n = 3. (B) Sirt6^flox/flox mice were intravitreally injected with AAV2-Null as control or AAV2-Cre for Sirt6 deletion. At 8 months after injection, leukostasis was performed and representative images of leukostasis in the central retina are shown. Green: Con A-labeled retinal vasculature and adherent leukocytes. Rectangles in the upper panels are zoomed in and arrows indicate stationary leukocytes adherent to the vascular endothelium. Adherent leukocytes in the whole retina were quantified. n = 6. (C and D) Sirt6^flox/flox mice were intravitreally injected with AAV2-Null as control or AAV2-Cre for Sirt6 deletion. Eyeballs were collected at 6 weeks after injection. Retinal sections were stained with anti-Caveolin-1 (Cav1) (red) and RBPMS (green) for RGCs. Arrowheads indicate Cav1 staining in vessels. Cav1 staining in the GCL was quantified. n = 3. (E) Differentiated R28 cells were transfected with control siRNA (C) or Sirt6 siRNA (S) for 48 h. Expressions of Cav1 and Sirt6 were assessed by western blot. Bar graph represents the relative expression of Cav1 from three independent experiments. (F and G) Differentiated R28 cells were transfected with control siRNA, Cav1 siRNA, Sirt6 siRNA, or Sirt6 siRNA + Cav1 siRNA for 72 h. The expressions of Cav1 and Sirt6 were examined by western blot (F). SA-β-gal activity was detected by Senescence Green Detection Kit with bar graph representing the relative intensity of SA-β-gal (G). Each experiment was repeated three times. ∗p < 0.05 and ∗∗∗∗p < 0.0001. Scale bar, 50 μm. GCL, ganglion cell layer; INL, inner nuclear layer; ONL, outer nuclear layer.

To delineate the potential molecular mechanisms by which loss of Sirt6 accelerates RGC senescence, we performed reverse phase protein array (RPPA) analysis, a high-throughput antibody-based proteomic approach, for retinal lysates from 2- and 7-month-old Sirt6^−/− and WT littermate mice. Through this analysis, we found that Annexin 1, Caveolin-1 (Cav1), FSP1, p-Stat3, and Stat3 were consistently and significantly upregulated in the retinas from 2- and 7-month-old Sirt6^−/− mice, associated with downregulation of the PI3K/Akt, FAK, and PIAS1 pathways that are critical for cell survival (Figure S3).^21,22 Given that Cav1 is recognized as a master regulator of cellular senescence,²³ we further examined its expression in AAV2-Null- or AAV2-Cre-infected Sirt6^flox/flox retinas by immunostaining and found that deleting Sirt6 in RGCs led to robust increase in Cav1 expression in RGCs (Figures 3C and 3D). To further explore the involvement of Cav1 in neuronal cell senescence induced by Sirt6 deletion, we knocked down Sirt6 with siRNA in differentiated R28 neuronal cells (Figure S4) that have been used as an in vitro model for retinal neurons including RGCs.^24,25,26 Consistently, knockdown of Sirt6 significantly increased Cav1 protein level (Figure 3E). We then knocked down Cav1 with small interfering RNA (siRNA) and assessed its effects on Sirt6 deletion-induced senescence of R28 cells (Figures 3F and 3G). Similarly, we found that SA-β-gal-positive staining in cells with Sirt6 knockdown was prominently increased. However, such increase was significantly attenuated by knockdown of Cav1, and the deletion of Cav1 itself did not induce cell senescence. These data indicate that Cav1 is a mediator of RGC senescence induced by Sirt6 deficiency.

Loss of Sirt6 induces mitochondrial dysfunction

Mitochondrial dysfunction is among the earliest changes observed in RGCs and plays a key role in RGC loss during the development of glaucoma.²⁷ Therefore, we reasoned that Sirt6 deletion-induced retinal pathological changes may also involve its regulation of mitochondrial function in the retina. We found that the expressions of a number of genes related to mitochondrial function, including Cox4, Cycs, and Pgc1α, were significantly reduced in the retinas from 2-month-old Sirt6^−/− mice compared with those from WT mice (Figure 4A). To examine the morphological changes of mitochondria in RGCs, we intravitreally injected AAV2-carrying dsRed2-Mito to AAV2-Null- or AAV2-Cre-infected Sirt6^flox/flox retinas to label mitochondria. In comparison with AAV2-Null-infected retinas, deleting Sirt6 in RGCs with AAV2-Cre infection resulted in progressive reduction of the overall density of mitochondria within RGC axons during aging (Figures 4B and S5). Consistently, mitochondrial dysfunction was observed in differentiated R28 neuronal cells with Sirt6 deficiency. Mitochondrial membrane potential in differentiated R28 cells transfected with either control or Sirt6 siRNA were evaluated by TMRE and JC-1 staining (Figure 4C).²⁸ After Sirt6 was knocked down, the fluorescence intensity of TMRE and the fluorescence ratio (red/green) of JC-1 were dramatically diminished in R28 cells, suggesting reduced mitochondrial membrane potential. Furthermore, analysis of mitochondrial function by measuring oxygen consumption rate (OCR) with a Seahorse XFe96 Analyzer revealed that R28 cells with Sirt6 knockdown displayed significant decreases in basal respiration rate and maximal respiration rate compared with control cells (Figure 4D). Collectively, these findings demonstrate that Sirt6 deletion leads to mitochondrial dysfunction in RGCs.

Figure 4.

Sirt6 deletion impairs mitochondrial function

(A) The mRNA levels of mitochondrial genes in the retinas from 2-month-old WT and Sirt6^−/− mice. n = 6–8. (B) Sirt6^flox/flox mice were intravitreally injected with AAV2-Null as control or AAV2-Cre for Sirt6 deletion and AAV2-carrying dsRed2-Mito (red) to label mitochondria. Eyeballs were collected at 12 months after injection, and representative images of mitochondria in the retinas are shown. Squares in the upper panels are zoomed in. n = 3–4. Scale bar, 10 μm. (C) Differentiated R28 cells were transfected with control or Sirt6 siRNA. At 72 h after transfection, mitochondrial membrane potential was assessed by TMRE and JC-1 staining. Graphs represent the quantification of the fluorescence intensity of TMRE and JC-1 (red/green) staining. Three independent experiments were performed and 6–7 images from each set were quantified. Scale bar, 20 μm. (D) R28 cells were transfected with control or Sirt6 siRNA. At 72 h after transfection, mitochondrial OCR was measured by a Seahorse XFe analyzer, and basal respiration and maximal respiration were calculated. n = 43 wells from three independent experiments. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001.

Sirt6 overexpression prevents RGC loss in high-tension glaucoma

Since the loss of Sirt6 resulted in signs of glaucoma including progressive RGC loss and optic nerve degeneration during aging (Figures 1 and 2), we reasoned that Sirt6 might have a neuroprotective effect on RGCs in high-tension glaucoma (HTG), where elevated IOP may compromise the integrity of this system to induce RGC and optic nerve degeneration. To test this possibility, we employed a mouse model of retinal ischemia-reperfusion (IR) by inducing a temporary IOP elevation. This model has been widely used to investigate the mechanisms behind the demise of RGCs in retinopathies including acute glaucoma.^29,30,31,32 We found that Sirt6 expression was significantly reduced while acetylated H3K56 was increased at 12 h after IOP elevation compared with control eyes (Figures 5A and 5B).

Figure 5.

Sirt6 overexpression is neuroprotective in the IR model

(A and B) Sirt6 expression (red) and acetylated H3K56 (green) in WT retinal sections at 12 h after IR injury. n = 4. (C and D) Validation of Sirt6 overexpression in the retinas of Sirt6-Tg mice. Retinal lysates from WT and Sirt6-Tg mice were blotted with anti-human and mouse Sirt6, respectively. Tubulin was used as internal control (C). Retinal sections from WT and Sirt6-Tg mice were stained with anti-Sirt6 (green) and propidium iodide (PI) (red) (D). (E and F) Apoptotic cells were detected by TUNEL assay on retinal sections of WT and Sirt6-Tg mice at 12 h after IR. n = 5–6. (G and H) Representative images of leukostasis in the peripheral retina of WT and Sirt6-Tg mice at 24 h after IR. Green: Con A-labeled retinal vasculature and adherent leukocytes. Rectangles in the upper panels are zoomed in and arrows indicate stationary leukocytes adherent to the vascular endothelium. Bar graph represents the number of adherent leukocytes in the whole retina. n = 6. (I and J) Microglia were stained with anti-Iba1 in retinal flatmounts of WT and Sirt6-Tg mice at 7 days after IR. Images were taken in the peripheral retina and microglia were counted. n = 4. (K and L) RGCs were stained with anti-Tuj1 at 7 days after IR and the number of RGCs was quantified. n = 8 for control retinas, and n = 13 for injured retinas. ∗p < 0.05 and ∗∗∗∗p < 0.0001. Scale bar, 50 μm. Image size of (E): 250 μm × 400 μm; image size of (I) and (K): 450.56 μm × 450.56 μm. GCL, ganglion cell layer; INL, inner nuclear layer; ONL, outer nuclear layer.

To investigate whether the decrease in Sirt6 expression is a cause of RGC injury, we crossed C57BL/6-Rosa26-lox-Sirt6 mice³³ with EIIa-Cre mice and generated C57BL/6-Sirt6-Tg mice that overexpress Sirt6 in all tissues including the retina (Figures 5C and 5D). EIIa-Cre has been removed by backcrossing with C57BL/6-WT mice. We then subjected 2-month-old Sirt6-Tg and WT mice to IR. After 12 h of IR, apoptotic cells were examined by TUNEL assay, which revealed that Sirt6 overexpression dramatically reduced IR-induced TUNEL⁺ cells (Figures 5E and 5F). Moreover, Sirt6 overexpression remarkably reduced IR-induced inflammatory reactions as demonstrated by decreased leukostasis (Figures 5G, 5H, and S6A) and microglial recruitment/activation (Figures 5I, 5J, and S6B). At 7 days after ischemia, WT mice showed a 50.5% reduction in RGC number, whereas Sirt6-Tg mice exhibited a smaller decline, only losing 27.6% of their RGCs (Figures 5K and 5L). Of note, no difference in the overall RGC distribution was observed in naive retinas between WT and Sirt6-Tg mice at this age (Figures 5K and 5L).

Given that Sirt6 is highly expressed in RGCs, we reasoned that the protective effects in Sirt6-Tg mice might be mainly mediated by Sirt6 overexpressed in RGCs. To explore this potential, we crossed Rosa26-lox-Sirt6 mice (Sirt6-Tg^fl-STOP-fl) with Syn1-Cre mice³² to specifically overexpress Sirt6 in retinal neurons, particularly in RGCs (Sirt6^RGC−Tg: Sirt6-Tg^fl-STOP-fl; Syn1-Cre). The specificity of Sirt6 overexpression in RGCs was confirmed by immunostaining with antibodies against Sirt6 and RGC marker Tuj1 (Figure 6A). Consistent with the result from global Sirt6 overexpression in mice, Sirt6-specific overexpression in RGCs also alleviated RGC loss after ischemia injury (Figure 6B). Moreover, while RGC loss and retinal inflammation were increased in very aged WT mice (30 months old), retinas from Sirt6^RGC−Tg mice exhibited decreased RGC loss, leukostasis, and Iba⁺ microglial activation (Figure S7), further demonstrating a neuroprotective role of Sirt6 expressed in RGCs.

Figure 6.

Sirt6-specific overexpression and deletion in RGCs exert opposing effects on IR-induced RGC loss

(A) Confirmation of Sirt6 overexpression in RGCs of Sirt6^RGC−Tg mice. Retinal flatmounts of Sirt6-Tg^fl-STOP-fl control mice and Sirt6^RGC−Tg mice were stained with anti-Sirt6 (red) and Tuj1 (green). (B) Two-month-old Sirt6-Tg^fl-STOP-fl control mice and Sirt6^RGC−Tg littermates were subjected to IR. Retinas were collected at 7 days after IR and stained with anti-RBPMS for RGCs. Bar graph represents the quantification of RGCs. n = 5–6. (C) Confirmation of Sirt6 deletion in RGCs of Sirt6^RGC−KO mice. Retinal flatmounts from Sirt6^flox/flox control mice and Sirt6^RGC−KO mice were stained with anti-Sirt6 (red) and anti-Tuj1 (green). (D) Two-month-old Sirt6^flox/flox control and Sirt6^RGC−KO mice were subjected to IR. RGCs in retinal flatmounts were labeled with anti-Tuj1 (green) at 7 days after IR, and representative images were taken at peripheral area of the retina. Graph represents the number of RGCs. n = 7–9. ∗p < 0.05 and ∗∗∗∗p < 0.0001. Scale bar, 50 μm. Image size of (B) and (D): 319.45 μm × 319.45 μm.

To further investigate the role of Sirt6 in RGC protection, we subjected Sirt6 heterozygous mice (Sir6^+/−) and their WT littermates on a C57BL/6 background to IR. In contrast to the effect of Sirt6 overexpression, partial loss of Sirt6 resulted in increased RGC loss following IR (Figure S8). Furthermore, we crossed Sirt6^flox/flox mice with Syn1-Cre mice to generate RGC-specific deletion of Sirt6 mice (Sirt6^RGC−KO: Sirt6^flox/flox; Syn1-Cre) (Figure 6C), and subjected 2-month-old Sirt6^flox/flox and Sirt6^RGC−KO mice to retinal ischemic injury. Although RGC numbers were comparable in Sirt6^flox/flox and Sirt6^RGC−KO mice at this stage, deletion of Sirt6 in RGCs significantly increased RGC loss at 7 days after ischemia (Figure 6D).

Next, we used a microbeads-induced chronic glaucoma model to further explore the involvement of Sirt6 in HTG (Figure 7). In this model, microbeads obstruct the drainage canals, leading to decreased aqueous outflow, sustained IOP increase, and progressive degeneration of RGCs and their axons similar to those in chronic glaucoma (open-angle glaucoma).³² At 7 days after microbeads injection, Sirt6 expression was found noticeably diminished in RGCs with increased H3K56 acetylation, indicating the reduction of Sirt6 activity (Figure 7A). We then subjected both Sirt6-Tg and age-matched WT mice to this model (Figures 7B–7F). At 6 weeks after microbeads injection, RGC loss in the retina and axonal damage in the optic nerve were assessed by immunostaining for RBPMS in retinal flatmounts and for SMI-32, which is an axonal marker and recognizes non-phosphorylated neurofilaments, on optic nerve sections. Although boosting Sirt6 expression did not change IOP level (Figure 7B), Sirt6 overexpression markedly prevented RGC loss from high IOP insult (Figures 7C and 7E) and alleviated axonal degeneration (Figures 7D and 7F). Altogether, using gain- and loss-of-function approaches, we have provided strong evidence that Sirt6 acts as a neuroprotective molecule in both acute and chronic glaucoma.

Figure 7.

Sirt6 overexpression is neuroprotective in microbeads-induced glaucoma

(A) Sirt6 expression and acetylated H3K56 (green) in WT retinal sections at 7 days after beads injection. n = 4. Scale bar, 20 μm. (B) IOP measurement after beads injection in WT and Sirt6-Tg mice. n = 8–15. ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001 for WT control vs. WT beads; ####p < 0.0001 for Sirt6-Tg control vs. Sirt6-Tg beads. (C) RGCs were labeled with anti-RBPMS (red) at 6 weeks after beads injection and representative images were taken at the peripheral area of the retina. Scale bar, 50 μm. (D) Optic nerve sections were stained with anti-NFH (SMI-32) for axons. Scale bar, 20 μm. (E and F) Graphs represent the relative number of RGCs in the retina (n = 5–6 for control eyes, and n = 12–14 for beads-injected eyes) and axons in the optic nerve (n = 5). ∗∗∗∗p < 0.0001. Image size of (C): 319.45 μm × 319.45 μm. Image size of (D): 101.41 μm × 101.41 μm. GCL, ganglion cell layer; INL, inner nuclear layer; ONL, outer nuclear layer.

Pharmacologic activation of Sirt6 or AAV2-mediated delivery of Sirt6 cDNA alleviates RGC injury in HTG

Encouraged by our findings demonstrating the neuroprotective effect of Sirt6 in glaucoma-relevant conditions, we sought to investigate whether boosting Sirt6 could be used to treat glaucoma. Sirt6 activator ML-800 at 100 mg/kg, which is an effective dose for in vivo study,³⁴ or vehicle solution was intraperitoneally injected into WT mice 6 h before the induction of IR model and once a day for 6 days afterward. We observed that the activation of Sirt6 with ML-800 dramatically prevented RGC loss at 7 days after IR (Figure 8A, upper panels). Concurrently, the treatment with Sirt6 activator resulted in significant reduction in microglial recruitment/activation (Figure 8A, lower panels; Figure S9).

Figure 8.

Sirt6 activator or AAV2-mediated Sirt6 overexpression is neuroprotective in high tension-induced glaucoma models

(A) WT mice were subjected to IR and administered with vehicle or Sirt6 activator ML-800 intraperitoneally once a day. RGCs and microglia in retinal flatmounts were labeled with anti-RBPMS (green) or anti-Iba1 (purple) at 7 days after IR, and representative images were taken at the peripheral area of the retina. Graphs represent the relative number of RGCs (n = 5 for control retinas, and n = 8–10 for injured retinas) and microglia (n = 5–6). (B–D) Female DBA/2J mice were injected with AAV2-Null, AAV2-GFP, or AAV2-Sirt6 intravitreally at 6 months of age. Retinas were collected at 11–12 months of age and stained with anti-RBPMS (red) for RGC soma and anti-Tuj1 (green) for RGC axons (B). The number of RGCs was quantified (C). PERG was measured before sample collection (D). n = 6–10. (E–G) Male DBA/2J mice were injected with AAV2-Null, AAV2-GFP, or AAV2-Sirt6 intravitreally at 6 months of age. Retinas were collected at 13–15 months of age and stained with anti-RBPMS (red) for RGC soma and anti-Tuj1 (green) for RGC axons (E). The number of RGCs was quantified (F). PERG was measured before sample collection (G). n = 7–10. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001. Scale bar, 50 μm. Image size of (A) and upper panels of (B) and (E): 319.45 μm × 319.45 μm.

AAV is widely used in clinical trials because of its safety and high efficiency for gene delivery and can maintain long-term transgene expression when delivered to neurons,³⁵ which is a preferred feature when treating chronic diseases including open-angle glaucoma. We asked whether delivery of AAV2-mediated overexpression of Sirt6 in RGCs could be used to treat chronic glaucoma. To address this question, we used the DBA/2J mouse strain because it develops glaucoma resembling inherited, age-related, and chronic human glaucoma³⁶ and exhibits more severe glaucomatous neurodegeneration than the microbeads-induced chronic glaucoma model, allowing better assessment of the translational value of interventions. In our DBA/2J colony, IOP significantly elevated by 8 months of age (Figures S10A and S10E), and RGC loss was around 50% at 10 months of age, then gradually increased to around 80% at 11–12 months of age in female DBA/2J mice compared with control DBA/2J-Gpnmb+ female mice (Figures S10B and S10C), whereas RGC loss was only around 50% loss even at 13–15 months of age in male DBA/2J mice (Figures S10F and S10G). Analysis of RGC function with pattern ERG (PERG) revealed amplitude decline about 70% and 50% in females and males, respectively (Figures S10D and S10H). This observation is consistent with previously documented data that female DBA/2J mice tend to develop elevated IOP earlier and more severe nerve damage than age-matched males.³⁷

Next, we generated and intravitreally delivered AAV2 carrying Sirt6 cDNA (AAV2-Sirt6) into the retina and found that AAV2 efficiently delivered Sirt6 genes into RGCs (Figure S11). Due to aforementioned sex-specific difference, we injected AAV2-Sirt6 or control virus (AAV2-Null or AAV2-GFP) intravitreally into DBA/2J mice at 6 months of age but analyzed RGC loss at 11–12 months of age for female mice and at 13–15 months of age for male mice. RGCs and their axons were largely maintained along with their functions in AAV2-Sirt6-treated retinas of female DBA/2J mice, exhibiting almost 7-fold more in cell number, denser axons, and higher PERG amplitude compared with AAV2-Null- and AAV2-GFP-treated retinas of female mice (Figures 8B–8D). Although male DBA/2J mice had less severe neural damages, AAV2-Sirt6 injection still effectively protected RGC number, axons, and function (Figures 8E–8G). These findings indicate that AAV2-mediated delivery of Sirt6 cDNA could potentially be used to alleviate RGC degeneration in chronic glaucoma. Altogether, enhancing Sirt6 expression/activity offers a promising approach for achieving neuroprotection in glaucoma.

Discussion

Glaucoma is a well-known age-related disease; nevertheless, the precise mechanisms of aging influence on glaucoma are still largely unknown, let alone any anti-aging therapeutic interventions. In this study, we demonstrated that Sirt6, an anti-aging molecule, was critically involved in glaucomatous neurodegeneration. We found that Sirt6 was highly expressed in RGCs and deleting Sirt6 globally or locally in RGCs led to progressive RGC loss and optic nerve degeneration during aging as observed in glaucoma, even without IOP elevation. In contrast, overexpressing Sirt6 attenuated age-induced RGC loss. In three established experimental mouse models of ocular hypertension, commonly employed to study and characterize the mechanisms of RGC injury in acute and chronic glaucoma and investigate strategies for neuroprotection, we discovered that Sirt6 level was decreased following IOP elevation while boosting Sirt6 expression/activity led to significant reduction of high IOP-induced degeneration of RGCs and their axons, whereas deleting Sirt6 accelerated high IOP-induced RGC loss. These findings demonstrate that Sirt6 plays a critical role in regulating RGC homeostasis during aging and ocular hypertension hijacks and compromises this system, leading to progressive RGC loss and optic nerve degradation as observed in various models of HTG.

Considering that glaucoma leads to irreversible vision loss due to RGC and optic nerve degeneration and that reducing IOP, the only established therapeutic option, does not always prevent visual loss in all glaucoma patients, there is a pressing need for novel interventions that can rescue RGCs and preserve vision. Our study provides compelling evidence that Sirt6 is an excellent target that could be modulated for glaucoma intervention. This possibility was supported by our promising findings demonstrating that two clinical feasible therapeutic approaches, including the application of pharmacological Sirt6 activator and AAV2-mediated Sirt6 gene therapy, significantly attenuated RGC loss in established mouse models of ocular hypertension. As small-molecule Sirt6 activators that belong to different classes have been developed,^34,38 further investigation of the therapeutic benefits of these activators in preclinical and clinical settings is warranted, which may bring novel management for glaucomatous neurodegeneration.

In mammals, there exist seven Sirts characterized by their unique tissue and subcellular distributions as well as diverse biochemical activities.¹⁰ Among them, Sirt1 has been extensively studied in pathophysiological conditions including retinal diseases. Overexpressing or activating Sirt1 has been shown to confer neuroprotection in mouse models of IR, diabetic retinopathy, optic nerve crush, and optic neuritis, while deleting Sirt1 in retinal neurons impedes revascularization in a mouse model of oxygen-induced retinopathy.^{39,40,41,42,43} However, the function of Sirt2-7 in retinal diseases is largely unknown and none of them have been studied in glaucoma-relevant models. Our work represents a comprehensive study that used multiple glaucoma-relevant models with gain- and loss-of-functions to elucidate a critical role of Sirt6 in glaucomatous neurodegeneration. Compared with a uniform expression pattern of Sirt1 in the retina, Sirt6 is dominantly expressed in RGCs, which highlights a role of Sirt6 in RGC function and homeostasis. This notion was supported by our data showing that Sirt6 deficiency induced RGC and optic nerve degeneration, while boosting Sirt6 attenuates glaucomatous neurodegeneration in three ocular hypertension models. It has previously been reported that as the retina ages, NAD⁺ level decreases but administrating NAD⁺ precursor nicotinamide results in the protection of RGCs in DBA/2J mice.⁴⁴ Our study suggests that Sirt6 is likely to be one of the major mediators for the beneficial effects of NAD⁺ since overexpressing Sirt6 rendered RGC more resilient to ocular hypertension-induced stress in DBA/2J mice although such overexpression could further decrease NAD⁺ availability for other NAD⁺ consuming enzymes. This feature makes it possible to use Sirt6 activators to protect RGCs in glaucoma even with lower NAD⁺ levels because our recent study shows that Sirt6 activators could still boost Sirt6 activity at less optimal NAD⁺ concentration.³⁸ Certainly, considering Sirt1 also has neuroprotective function, future studies are needed to understand whether Sirt6 and Sirt1 exhibit neuroprotective functions in the retina through similar, partially overlapped or rather distinct mechanisms.

Specific mechanisms underlying the neuroprotective role of Sirt6 in glaucomatous neurodegeneration remain to be elucidated. Aging is a natural progression in all living mammals, and as a result, primary mammalian cells have a finite proliferative lifespan. Upon reaching a certain replication threshold, cells can reach a state of continuous cell cycle halt, known as replicative senescence, which has been tied to a decrease in telomerase function and the reduction of telomere length.⁴⁵ In addition to replicative senescence, a variety of stresses or stimuli can induce a senescence response called SIPS that does not necessitate telomere reduction and excessive cell splitting and can occur in any cell type including non-dividing neurons.^{46,47,48,49,50} As Sirt6 is an anti-aging molecule and the only one among the seven mammalian Sirts whose overexpression promotes longevity, we examined RGC senescence and found that loss of Sirt6 accelerated RGC senescence including increased SA-β-gal activity and the SASP (as indicated by vascular inflammation), whereas overexpressing Sirt6 attenuated SASP in IR. Our data are consistent with a previous study showing the presence of SA-β-gal-positive RGCs in the glaucomatous retina, coinciding with the SASP.³¹ Given that prompt elimination of senescent cells shields unaffected RGCs from both senescence and programmed cell death and prevents the loss of retinal functions in IR,⁵⁰ it is possible that Sirt6 maintains RGC homeostasis and prevents glaucomatous neurodegeneration by preventing RGC senescence via modulating multiple pathways involved in senescence and cell survival, in particular Cav1. Cav1 is a master regulator of cellular senescence.²³ Although loss of Cav1 impairs retinal neuronal function due to the disturbance of subretinal microenvironment⁵¹ and causes blood-retinal barrier breakdown,⁵² Cav-1 ablation increases TrkB phosphorylation, reduces endoplasmic reticulum stress, and partially protects the inner retinal function in mouse models of glaucoma⁵³ and neuroretina-derived Cav1 has been shown to promote endotoxin-induced retinal inflammation.⁵⁴ Therefore, whether Cav1 exerts protective vs. deleterious effects depends on the specific contexts such as cell types and stresses. Additionally, our in vivo and in vitro findings provided important evidence that loss of Sirt6 led to mitochondrial dysfunction in RGCs. Since RGCs have numerous mitochondria and the intraretinal portions of their axons rapidly turn over mitochondria to meet high metabolic activity and energy demand to exert their function,²⁷ and mitochondrial dysfunction plays a key role in RGC loss during glaucoma,²⁷ it is reasonable that the neuroprotective function of Sirt6 in glaucoma also involves its maintenance of mitochondrial functions. Although Sirt6 is not localized in mitochondria, it has been shown that Sirt6 can regulate mitochondrial homeostasis and function by activating AMP-activated protein kinase and the transcription of the Nrf2-dependent genes including Sirt3 in other tissues.^55,56,57 However, considering that Sirt6 can catalyze lysine deacetylation and deacylation and ADP-ribosylation,⁸ it is likely that other mechanisms may also participate in this process. Further studies with single-cell genomics and proteomics analysis are needed to further explore the pleiotropic effects of Sirt6 in the retina and to improve the current understanding of the cellular and molecular mechanisms by which Sirt6 maintains RGC homeostasis and prevents its degeneration in glaucoma.

Of note, mice with Sirt6 deletion developed progressive degeneration of RGCs and their axons without IOP elevation. While a simple explanation could be that loss of Sirt6 causes an acceleration of normal age-related RGC death, it is noteworthy that these phenotypes resemble NTG in which chronic degeneration of RGCs and their axons occurs as in primary open-angle glaucoma, but with IOP within the normal range. In clinical practice, the primary treatment approach for NTG still aims to reduce IOP as studies indicate that a 25%–30% reduction in IOP from the baseline slows down the progression of the disease.⁴ However, it has been reported that even with a 30% reduction in IOP, a significant proportion of cases still experience glaucomatous progression.⁴ There is an urgent need to understand the pathological mechanisms of NTG and develop alternative treatments independent of IOP regulation.⁴ Nonetheless, studies on NTG are rare and the knowledge on NTG is limited. From an epidemiological perspective, age is the primary risk factor for NTG. While HTG usually occurs in patients older than 50 years, the mean age of NTG patients is older than 60 years,⁶ suggesting a closer relationship between NTG and aging. Current, only a few genes are linked to NTG.⁵⁸ Given that Sirt6 is an anti-aging molecule and its deletion led to phenotypes similar to NTG while Sirt6 overexpression attenuated RGC loss in aged mice, our data warrant further exploring the potential connection between Sirt6 and NTG.

We intravitreally delivered AAV2 carrying Cre recombinase or Sirt6 to delete or overexpress Sirt6 in RGCs. AAV2 has a preferential tropism for RGCs,^18,19 but the magnitude and cell-type specificity of transgene expression delivered by AAV2 are highly influenced by promoters.⁵⁹ The chicken β-actin (CBA) and cytomegalovirus (CMV) promoters poorly drive transgene expression in RGCs while the short CMV early enhancer/chicken β-actin/short β-globin intron (sCAG) and mouse phosphoglycerate kinase (PGK) promoters drive transgene expression in a significant portion of non-RGCs in the retina. We used a full-length CAG promoter to drive the expressions of GFP, Cre recombinase, or Sirt6, and observed that AAV2 efficiently delivered these genes into RGCs with good specificity. Moreover, using integrated approaches including AAV2-mediated gene delivery and transgenic mice with RGC-specific Sirt6 deletion and overexpression, our data strongly support the conclusion that Sirt6 plays a critical role in preventing the degeneration of RGCs and optic nerve during aging and glaucoma. Nonetheless, it remains possible that observations using AAV2 to delete or overexpress Sirt6 might involve some non-RGC effects. Future studies employing an RGC-specific promoter would be necessary to further explore this possibility.

In summary, Sirt6 is critically involved in maintaining RGC homeostasis. Its deficiency leads to RGC and optic nerve degeneration during aging, while enhancing Sirt6 expression/activity is neuroprotective in aged mice and in mouse models of HTG. This protection is potentially achieved by Sirt6-mediated suppression of RGC senescence and mitochondrial dysfunction. The simultaneous involvements of Sirt6 in RGC protection during aging and HTG suggest that Sirt6 may be an excellent target for therapeutic intervention as well as a potential marker for assessing the level of a patient’s resilience/defense to glaucomatous damage. More broadly, our work provides a framework for better understanding aging in glaucoma and warrants further exploration of other anti-aging molecules in glaucomatous neurodegeneration.

Materials and methods

Animal care and use

Sirt6 global knockout (Sirt6^−/−) mice and their control mice, on a C57BL/6–129/SvJ mixed background, were generated by crossing Sirt6 heterozygotes and their genotypes were determined by PCR as described previously.¹⁴ Sirt6^+/− on a C57BL/6 background and Sirt6^flox/flox mice on a 129/SvJ background, originally developed by Dr. Chuxia Deng, were a generous gift from Dr. Jordan D. Miller at Mayo Clinic. C57BL/6J WT (strain#000664), Syn1-Cre (strain#003966), DBA/2J-Gpnmb+ (strain#007048), and DBA/2J (strain#000671) mice were originally purchased from the Jackson Laboratory (Bar Harbor, ME). Sirt6-Tg^fl-STOP-fl mice were generated as described previously.³³ Sirt6-Tg mice were generated by crossing Sirt6-Tg^fl-STOP-fl mice with EIIa-Cre mice (strain#003724) (Jackson Laboratory) followed by backcrossing with C57BL/6J-WT mice to remove EIIa-Cre. All mice were bred and housed in the animal facility with a dark/light cycle of 12 h/12 h at the University of Texas Medical Branch. All transgenic mice, except for Sirt6^−/− and their control mice on a C57BL/6–129/SvJ mixed background, DBA/2J-Gpnmb+, and DBA/2J mice, were fully backcrossed into C57BL/6J background before using to generate tissue-specific knockout/overexpression mice. The offspring of transgenic mice were genotyped by performing PCR of the genomic DNA derived from tail tip. All mice were used with appropriate age- and sex-matched controls, and examiners were blinded to mice genotype. All experimental procedures were conducted in accordance with the Association of Research in Vision and Ophthalmology (ARVO) Statement for the Use of Animals in Ophthalmic Vision and Research and approved by the Institutional Animal Care and Use Committee of the University of Texas Medical Branch.

Cell culture

R28 cells (EUR201), a retinal precursor cell line, were purchased from Kerafast, Inc (Boston, MA) where authentication and quality-control tests for distributed cell lines were comprehensively performed. The R28 cell line was initially propagated to generate clearly labeled frozen stocks of early passages, and one frozen stock vial was sent to IDEXX BioAnalytics (Columbia, MO) for cell check. The stock vial was confirmed to be mycoplasma negative and rat origin with matched short tandem repeat markers. For experiments, R28 cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% calf serum, 2 mM Glutamine, 1% MEM non-essential amino acids, 1% MEM Vitamin and 100 μg/mL gentamicin in a humidified 5% CO₂ incubator at 37°C. Cells with low passage numbers were used and cell morphology and doubling time were carefully checked to ensure they were healthy before treatment. R28 cells were seeded at a density of 1.5 × 10⁵ cells/well in 10 μg/mL laminin-coated 6-well plates and differentiated by 250 μM pCPT-cAMP (C3912; MilliporeSigma, Burlington, MA). After 24 h of incubation, differentiated R28 cells were transfected with control siRNA, Caveolin-1 siRNA, Sirt6 siRNA, or Sirt6+Caveolin-1 siRNA using Lipofectamine RNAiMAX reagent (13778-150; ThermoFisher Scientific, Waltham, MA). The final working concentration was 5 nM for control siRNA and Sirt6 siRNA, and 1.5 nM for Caveolin-1 siRNA. After 24 h of transfection, cells were detached and reseeded at a density of 3 × 10⁴ cells/well onto 10 μg/mL laminin-coated coverslips in 24-well plates for 48 h in the presence of pCPT-cAMP. Last, cells were subjected to Invitrogen CellEvent Senescence Green Detection Kit (C10850; ThermoFisher Scientific) according to manufacturer’s instruction. Images were taken by confocal microscope (LSM 800; Carl Zeiss, Inc., Thornwood, NY).

Evaluation of cellular mitochondrial membrane potential and function

Differentiated R28 cells were transfected with control siRNA or Sirt6 siRNA for 48 h. Next, to analyze mitochondrial membrane potential, R28 cells were reseeded into glass bottom culture dishes. After 24 h of culture, JC-1 (AG-CR1-3568; Adipogen, San Diego, CA) or TMRE (HY-D0985A; MedChemExpress, Monmouth Junction, NJ) was added to serum-free DMEM at a final concentration of 2 μM or 200 nM, respectively. Cells were incubated in the dark at 37°C for 15 min, washed with PBS, and immediately imaged by a confocal microscope. To measure mitochondrial respiration, R28 cells were reseeded into a precoated Seahorse XFe96 cell culture microplate (Seahorse Bioscience, Billerica, MA) at a density of 1 × 10⁴ cells/well in complete medium containing 250 μM pCPT-cAMP. After allowing attachment for 24 h, the culture medium was switched to Seahorse XF assay medium (pH 7.4) provided by Seahorse Bioscience, supplemented with 10 mM glucose, 1 mM sodium pyruvate, and 2 mM glutamine. Subsequently, cells were kept in a non-CO₂ incubator at 37°C for an hour. OCR was measured at basal level and after subsequent injections of 1.5 μM oligomycin, 2 μM FCCP, and 0.5 μM Rotenone/0.5 μM Antimycin A (Rot/AA) by Agilent Certified Pre-Owned Seahorse XFe96 Analyzer (Seahorse Bioscience). Basal respiration was determined by subtracting non-mitochondrial respiration (lowest measurement of rate post Rot/AA injection) from the basal rate measurement taken before introducing oligomycin. Maximal respiration was determined by subtracting non-mitochondrial respiration from the highest measurement of rate after FCCP injection.

IOP measurement

IOP was measured in the afternoons at approximately the same time using a rebound tonometer for rodents (TonoLab; Colonial Medical Supply, Franconia, NH) after mice were anesthetized by isoflurane. Five measurements were taken for each eye and averaged by the tonometer automatically for further analysis.

Mouse model of acute glaucoma

Retina IR model was induced as previously reported.^30,32 Two-month-old mice were anesthetized with intraperitoneal injection of 75 mg/kg pentobarbital; 0.5% proparacaine hydrochloride was applied for topical anesthesia. Following pupil dilation with 1% tropicamide and 2.5% phenylephrine hydrochloride, the IOP of the right eye was elevated to 110 mm Hg via inserting a 30G infusion needle connected to a standard saline reservoir into the anterior chamber for 50 min. The contralateral left eye without pressure elevation served as control. Sirt6 activator (MDL-800, 100 mg/kg; 565267; MedKoo Biosciences, Morrisville, NC)³⁴ or vehicle (5% DMSO and 30% PEG-300 in saline, pH 7–8) was administered through intraperitoneal injection 6 h prior to initiating IR, followed by daily injection for 6 consecutive days. Eyes or retinas were collected at the selected time points after IR.

Microbeads-induced glaucoma model

Microbeads-induced glaucoma was performed as previously reported.^32,60 In brief, 2-month-old mice were anesthetized with intraperitoneal injection of a mixture of 100 mg/kg ketamine hydrochloride and 10 mg/kg xylazine hydrochloride. Topical anesthesia was administered with 0.5% proparacaine hydrochloride. Subsequent to pupil dilation with 1% tropicamide and 2.5% phenylephrine hydrochloride, 2 μL of 1-μm-diameter polystyrene microbead suspension (containing 3.0 × 10⁷ beads) followed by an air bubble, 2 μL of 6-μm-diameter polystyrene microbead suspension (containing 6.3 × 10⁶ beads) (Polysciences, Inc., Warrington, PA), and 1 μL of PBS with 30% Healon (Abbott Laboratories, Abbott Park, IL) were injected into the anterior chamber using a Hamilton syringe with a 32G needle (Hamilton Company, Reno, NV) through a corneal tunnel pre-made by a 27G needle.³² IOP was measured in the afternoons at approximately the same time using a tonometer at 3 days after injection and thereafter once a week until 6 weeks after injection, and those with IOP exceeding 21 mm Hg were deemed to be the successfully induced glaucomatous model.

Intravitreal injection

Adeno-associated virus 2 (AAV2) carrying green fluorescent protein (GFP) cDNA (AAV2-GFP, SL100816), AAV2-Null (SL100811), and AAV2-Cre (SL100884) were purchased from SignaGen Laboratories (Rockville, MD). AAV2-Sirt6 was produced using the custom recombinant AAV vector production service provided by SignaGen Laboratories using a human Sirt6 cDNA synthesized by Genscript (Piscataway, NJ) based on NCBI Reference Sequence NM_016539.4. The sequence of Sirt6 was validated by DNA sequencing. AAV2-DsRed2-Mito was also custom-made by SignaGen Laboratories. All cDNAs in AAV2 were driven by a full-length CAG promoter. AAV2-mediated gene transfer was performed as described.¹⁷ To study Sirt6 deletion-mediated RGC degeneration and mitochondrial changes, 1 μL AAV2-Null or AAV2-Cre (1 × 10¹² vector genomes [VG]/mL) was intravitreally injected into the eye of Sirt6^flox/flox mice at 2 months of age. Two months before sample collection, 1 μL AAV2-DsRed2-Mito (1 × 10¹² VG/mL) was injected intravitreally. To study the effect of Sirt6 gene therapy in glaucoma, 1 μL AAV2-Null, AAV2-GFP, or AAV2-Sirt6 (1 × 10¹³ VG/mL) was intravitreally injected into the eyes of DBA/2J mice at 6 months of age.

Retinal leukostasis assay

Retinal leukostasis assay was conducted to label adherent leukocytes as documented previously.³² Briefly, after mice were anesthetized, 10 mL PBS was perfused to the circulation system through the left ventricle to remove nonadherent blood cells, followed by perfusion with fluorescein isothiocyanate (FITC)-conjugated Con A lectin (40 μg/mL in PBS, pH 7.4) (RL-1002; Vector Laboratories, Burlingame, CA) to label adherent leukocytes and vasculature. Next, PBS was perfused to wash away nonbonding Con A lectin. Eyeballs were collected and fixed with 4% paraformaldehyde (PFA) at 4°C overnight. The next day, retinas were dissected and mounted. The total number of adherent leukocytes in the whole retina was counted.

Immunostaining of retinal flatmounts and cryosections

Immunostaining of retinal flatmounts was done as previously described.³² Eyeballs were fixed in 4% PFA at 4°C overnight. Fixed retinas were cautiously dissected, washed with PBS, and then blocked and permeabilized with PBS containing 5% normal goat serum and 0.3% Triton X-100 for 3 h. Following blocking, retinas were incubated with primary antibodies at 4°C overnight. The next day, retinas were incubated with corresponding Alexa Fluor 405, 488, 594, or 647-conjugated secondary antibodies (1:400, ThermoFisher Scientific) at 4°C for 4 h. Retinas were mounted, and images were taken by the confocal microscope. For RGC counting, eight non-overlapping images were taken at the peripheral region of each retinal flatmount and cells were manually counted and averaged for each sample. For retinal and optic nerve cryosections, sample preparation and immunostaining were performed as described previously.³² All series of images used for comparison were acquired with the same settings. Primary antibodies are listed in Table S1.

RPPA screening

Retinas from 2- and 7-month-old Sirt6^−/− and WT littermate mice were collected, homogenized, and lysed using modified Tissue Protein Extraction Reagent (TPER) (ThermoFisher Scientific) with a cocktail of protease and phosphatase inhibitors (Roche, Indianapolis, IN). RPPA was performed and analyzed by the Antibody-based Proteomics Core at Baylor College of Medicine as previous described^61,62 using probes with a set of 244 antibodies that target a wide range of total proteins and phosphoproteins associated with processes such as epithelial-mesenchymal transition, stem cells, apoptosis, DNA damage, autophagy, proliferation and cell cycle, growth factor receptors, cytokines/STATs, and nuclear receptors, as well as transcriptional and chromatin regulatory proteins.

Western blot

Retinas were homogenized and lysed for 30 min on ice in RIPA buffer (89901; ThermoFisher Scientific) supplemented with Complete Protease and Phosphatase Inhibitors (Roche), followed by centrifugation at 14,000 rpm and 4°C for 15 min. Protein concentration was determined by BCA Assay (23225; ThermoFisher Scientific) and equal amounts of protein were electrophoresed in SDS-PAGE gels, and electroblotted onto nitrocellulose or PVDF membranes. After blocking with 5% milk, the membranes were probed with primary antibodies as listed in Table S1 overnight at 4°C, followed by appropriate horseradish peroxidase (HRP)-conjugated secondary antibodies (Cytiva, Marlborough, MA) for 1 h at room temperature. After washing, proteins were detected by enhanced chemiluminescence (Pierce, Rockford, IL). The intensity of blots was analyzed densitometrically with ImageJ, and the expression of proteins was normalized to loading control tubulin and expressed as fold changes of the control.

Quantitative real-time PCR

Retinas were lysed in TriZol reagent (15596018; ThermoFisher Scientific) and RNA was isolated with RNAqueous-4PCR kit (AM1914; ThermoFisher Scientific). RNA concentration was measured using a NanoDrop (ThermoFisher Scientific) and cDNA was synthesized using the High-Capacity cDNA Reverse Transcription Kit (4368813; ThermoFisher Scientific). RT-qPCR was performed on a StepOnePlus PCR system (Applied Biosystems, Waltham, MA) using SYBR Green Master Mix (4368708; ThermoFisher Scientific). Data were calculated by the ΔΔCT method using Hprt as housekeeping reference gene and expressed as fold change. Primer sequences are listed in Table S2.

Electron microscopy of the optic nerve

Transmission electron microscopy was performed as previously described.¹⁴ Mice were anesthetized and perfused with Ringer’s solution and 2% PFA plus 2.5% glutaraldehyde in 0.15 M cacodylate buffer. Optic nerves were collected and stored in fixative at 4°C overnight, and then osmicated, soaked in uranyl acetate, and prepared for electron microscopy. Images were taken using a JEOL 2100 transmission electron microscope (JEOL Ltd.; Akishima, Tokyo, Japan).

TUNEL assay

TUNEL assay was performed on retinal cryosections using ApopTag Fluorescein In Situ Apoptosis Detection Kit (s7110; MilliporeSigma) according to the manufacturer’s instructions.³² After retinal cryosections were counterstained with propidium iodide (PI) to label nuclei, images were taken by an epifluorescence microscope, and TUNEL-positive cells were counted.

Pattern electroretinogram

After mice were anesthetized and placed on a platform without pupil dilation, two recording and one reference electrodes were inserted hypodermically between the ears, between the eyes, and at the caudal back. The electric signals (Figure S12) were recorded from both eyes by Jörvec instrument (Intelligent Hearing Systems, Miami, FL). PERG waveform consists of a positive wave (defined as P1) and a following negative wave with a broad trough (defined as N2), both automatically detected by the Jörvec software. PERG amplitudes were obtained by measuring the difference between the P1 peak and N2 trough. Each eye underwent 2–3 times independent measurements.

Hematoxylin and eosin staining

Hematoxylin and eosin (H&E) staining was performed as described.⁶³ Briefly, the eyes were enucleated, embedded in optimal cutting temperature compound, and frozen in dry ice/ethanol bath for cryosectioning; 10-μm-thick cryosections were stained with H&E and images were captured by a light microscope (Leica Camera Inc., Allendale, NJ). The thickness of total retina and individual retinal layers was measured with LAS EZ software (Leica Camera Inc).

Senescence-associated β-galactosidase staining of retinal flatmounts

CellEvent Senescence Green Detection Kit (C10850; ThermoFisher Scientific) was used to detect senescence-associated β-galactosidase (SA-β-gal) activity according to the manufacturer’s instructions. To label senescent cells in the retina, eyeballs were fixed in 4% PFA at 4°C overnight, and then retinas were dissected and incubated in working solution for 2 h at 37°C. Subsequently, retinas were incubated with anit-Tuj1 antibody to label RGCs. Retinas were finally washed, mounted, and imaged with the same settings by confocal microscopy.

Statistical analysis

Data are presented as means ± SEM. GraphPad Prism 9 (GraphPad Software, La Jolla, CA) was used to generate graphs and for statistical analyses. Student’s t test was used for two-group comparison, and one-way ANOVA was used for multiple comparisons. Values of p < 0.05 were considered statistically significant.

Data and code availability

Data presented are available within the article and the supplemental information. Additional information can be accessed from the corresponding author upon request.