Age-related alterations in retinal neurovascular and inflammatory transcripts

Animals

Pathogen-free Fischer 344 x Brown Norway (F1) hybrid male rats, aged 3 months (Young-Adult), 12 months (Adult), and 26 months (Aged; n=5 per group) were obtained from the National Institute on Aging colony at Harlan Industries (Indianapolis, IN), as described previously [51]. All animals were singly housed in laminar flow cages (Polysulfone) in the Oklahoma University Health Sciences Center (OUHSC) Barrier Facility with free access to food and water (Purina Mills, Richmond, IN). The animal rooms were maintained on a 12h:12h light-dark cycle at constant temperature and humidity. The OUHSC animal facilities are fully accredited by the Association for Assessment and Accreditation of Laboratory Animal Care. All animal procedures were approved by the Institutional Animal Care and Use Committee in accordance with the Public Health Service Policy on Humane Care and Use of Laboratory Animals and the National Research Council’s Guide for the Care and Use of Laboratory Animals. Animals were sacrificed by decapitation without anesthesia and the retinas rapidly dissected, snap frozen in liquid nitrogen, and stored at −80 °C. Trunk blood was also collected at the time of sacrifice. At sacrifice, gross necropsy was performed and no visible tumors were evident in any of the animals. Visual inspection of the eye did not reveal cataracts in any of the animals.

Serum glucose quantitation

Trunk blood was collected into 15 ml conical tubes (VWR, West Chester, PA) and separated by centrifugation at 2,500 rpm for 20 min at 4 °C. Following serum isolation (upper phase), samples were stored at −80 °C until analysis. Serum samples were analyzed for glucose levels by colorimetric assay (Glucose Assay Kit; Abcam, Cambridge, MA). Each sample was analyzed in technical triplicates with a standard curve ranging from 0 to 20 nmol/μl using a SpectraMax M2 microplate reader (Molecular Devices, Sunnyvale, CA) at 570 nm absorbance.

RNA isolation and quantitation

Total RNA was isolated from snap-frozen retinas using standard extraction, precipitation, and purification methods, as previously described [45-48]. Briefly, retinas were homogenized in ice-cold Tri-Reagent (Sigma-Aldrich, St. Louis, MO) using an automated Retsch TissueLyser II bead mill (Qiagen, Inc., Germantown, MD) and stainless steel beads at 15 Hz for 1 min. 1-bromo-3-chloropropane (BCP; Molecular Research Center, Inc., Cincinnati, OH) was added in 1/10th volume to separate phases. Isopropanol was then added at 1/10th volume to precipitate RNA from the aqueous phase. Precipitated RNA was purified using Qiagen RNeasy spin columns (Qiagen, Inc., Valencia, CA) and subsequently resuspended in RNase-free water. Quality and quantity were evaluated using the RNA 6000 Nano LabChip with an Agilent 2100 Expert Bioanalyzer (Agilent, Palo Alto, CA) and NanoDrop ND100 (Nanodrop, Wilmington, DE), respectively. RNA quality, as measured by a Bioanalyzer, was determined using RNA integrity numbers, which are a measure of RNA degradation ranging from 10 (intact) to 2 (degraded). RNA purity and concentration, as measured by a NanoDrop spectrophotometer, was determined by absorbance at 260 and 280 nm.

Quantitative reverse transcription PCR

Following RNA isolation and quality control, cDNA was synthesized using the ABI High Capacity cDNA Reverse Transcription Kit (Applied Biosystems Inc., Foster City, CA) according to standard protocols [46,49]. Briefly, 2× reverse transcription (RT) master mix was prepared by mixing 2 μl 10× RT Buffer, 0.8 μl 25× 100 mM deoxynucleotide triphosphates (dNTPs), 2 μl 10× random primers, 1 μl MultiScribe Reverse Transcriptase, 1 μl RNase Inhibitor, and 3.2 μl nuclease-free water per reaction (total of 15 reactions) for a total of 10 μl master mix per reaction. Total RNA (1 μg/10 ul) was pipetted into each well followed by 10 μl of the prepared 2× master mix. Reverse transcription was performed using a GeneAmp PCR System 9700 (Applied Biosystems Inc.) with the following settings: Step 1) 25 °C – 10 min; 2) 37 °C – 2 h; 3) 85 °C – 5 s; 4) hold at 4 °C.

Quantitative PCR analysis was performed using a 7900HT Sequence Detection System (Applied Biosystems Inc.), 384-well optical plates, and Assay-On-Demand gene expression assays containing the gene-specific primers and probes listed in Table 1 (Applied Biosystems Inc.) [46,49]. All samples were assayed in technical triplicates. Briefly, the PCR reaction mix was prepared by mixing 0.125 μl 20× gene expression assay and 5 μl 2× Gene Expression Master Mix (Applied Biosystems Inc.) per reaction. Then, 50 ng cDNA (5 μl of 10 ng/μl stock solution) was pipetted into each well, followed by 5 μl of the prepared master mix, yielding a 10 μl reaction. The prepared plate was incubated at 50 °C for 2 min and at 95 °C for 10 min, followed by 40 PCR with the following specifications: 1) 95 °C – 15 s; 2) 60 °C – 1 min. Relative amounts of gene transcript were determined using SDS 2.2.2 software (Applied Biosystems Inc.) and the 2^-ΔΔCt method with β-actin (Actb) as the endogenous control [52,53].

Statistical analyses

Serum glucose concentrations and transcript expression data were analyzed by one-way ANOVA (ANOVA) with Student–Newman–Keuls (SNK) post-hoc testing with α<0.05. For the gene expression analysis, a Benjamini-Hochberg multiple testing correction was applied to the one-way ANOVA across the 36 genes examined to control for Type I false positive errors [54]. For any data that failed normality testing, a Kruskal–Wallis one-way ANOVA on ranks was performed using SigmaStat 3.5 (Systat Software, San Jose, CA). Potential overrepresentation of genes commonly regulated with aging and diabetes was tested using a binomial distribution, assuming the independence of each gene and equal distribution of increases and decreases in gene expression by random chance. Principal component analysis was used to visualize the relationship between groups. Individual animal data for Young, Adult, and Aged rats from all 36 genes quantitated were normalized to mean Young values and imported into GeneSpring GX 11.0. Likewise, previously reported expression data for 3-month diabetic and nondiabetic control rats [45,46,49] were scaled to mean control values for comparison with age-group data. The first two principal components were mean centered and scaled, and plotted with individual animal data compressed to represent each group as a single point. Potential common transcriptional regulators were assessed by Ingenuity Pathway Analysis (Ingenuity Systems, Redwood City, CA).

Serum glucose

Serum glucose levels at sacrifice were assessed to confirm normoglycemia. The average serum glucose level in Young animals was 5.24 nM±0.7 (94 mg/dl), 6.72 nM±0.8 (121 mg/dl) in Adult animals, and 5.60 nM±0.5 (101 mg/dl) in Old animals (Figure 1). All groups were normoglycemic and no statistically significant differences were found between the groups. No individual animals were hyperglycemic (>250 mg/dl).

Retinal mRNA expression

To evaluate potential similarities between retinal transcript expression changes with aging and diabetes, 36 gene targets from our previously described analyses of DR [45-48] were quantitated in Young, Adult, and Aged rats by quantitative PCR. Our previous analyses of retinal gene expression with diabetes have shown that these transcripts are differentially expressed in a highly consistent manner across multiple experiments [46]. Of the 36 genes examined, 14 were differentially expressed (one-way ANOVA, Benjamini-Hochberg multiple testing correction, p<0.05). Eleven genes were increased similarly to what we have previously demonstrated with diabetes. Three genes were regulated in a direction opposite to that seen in diabetes. More genes were regulated in the same direction and magnitude with aging and diabetes than would be expected by chance (p=0.028), assuming a random distribution. To test for the potential coregulation of these genes, Ingenuity Pathway Analysis was performed and no transcription factors or other transcriptional regulators were identified as common to more than three of the differentially expressed genes. However, more complex regulatory networks that are not currently understood may be commonly engaged by aging and diabetes, which would affect the assumption of a random distribution. Although these genes have multiple functions, they can be separated into three distinct groups according to our previous analyses [45]: inflammation-related genes, microvascular-related genes, and neuronal function–related genes.

Inflammatory transcripts

For the genes associated with inflammatory processes, most of the changes were upregulations with age. Retinal transcript expression of complement component 1 s subcomponent (C1s), glial fibrillary acidic protein (Gfap), lipopolysaccharide-induced TNF factor (Litaf), chitinase 3-like 1 (Chi3L1), janus kinase 3 (Jak3), and signal transducer and activator of transcription 3 (Stat3) was significantly higher in Adult animals when compared to Young animals. Additionally, these transcripts showed significantly higher expression in Aged versus both Young and Adult animals (Figure 2). Complement 1-inhibitor I (Serping1) and tumor necrosis factor receptor subfamily member 12a (Tnfrsf12a) demonstrated a similar pattern of expression with significantly higher expression levels in Aged versus both Young and Adult animals. Guanylate nucleotide binding protein 2 (Gbp2) expression was significantly elevated in Aged and Adult versus Young animals. Pigment epithelium-derived growth factor (Pedf) showed significantly lower expression levels in both Adult and Aged animals compared to the Young group. Additional genes examined (X-linked inhibitor of apoptosis [Birc4], calcium regulated heat stable protein 1 [Carhsp1], chemokine CC motif receptor 5 [Ccr5], elongation of very long chain fatty acids [Elovl1], heat shock 27 kDa protein 1 [Hspb1], lectin galactoside-binding soluble 3 [Lgals3], lectin galactoside-binding soluble 3 binding protein [Lgals3bp], methylcrotonoyl-CoA carboxylase 1 [Mccc1], nuclear receptor subfamily 3 group C member 1 [Nr3c1], signal transducer and activator of transcription 1 [Stat1], and tissue inhibitor of metalloproteinase 1 [Timp1]) did not differ in expression between groups (Table 2). With the exception of Pedf, all of the age-related gene expression changes were induced in a similar fashion to what was seen in diabetes (Table 3).

Microvascular transcripts

Of the microvascular-related transcripts measured, endothelin 2 (Edn2) and intracellular adhesion molecule 1 (Icam1) were both found to significantly increase in expression in the retina with age (Figure 3). In particular, Edn2 showed significantly higher expression levels in the Aged versus Young and Adult rats, as well as in Adult versus Young rats (Kruskal–Wallis one-way ANOVA on ranks, p<0.05). Similarly, Icam1 showed significantly higher expression in Aged versus Young and Adult versus Young animal comparisons (SNK, p<0.001). No differences in expression of endothelin receptor B (Ednrb), natriuretic peptide precursor A (Nppa), or vascular endothelial growth factor A (Vegfa) were observed between age groups (Table 2). Similar to the age-related gene expression changes, both Edn2 and Icam1 showed significantly higher expression in rats with diabetes compared to nondiabetic controls (Table 3).

Neuronal function transcripts

Two genes associated with neuronal function were also found to be significantly regulated with age in the rat retina (Figure 4). Polycomb group ring finger 1 (Pcgf1) demonstrated significantly higher transcript levels when comparing Aged rats to Young ones (SNK, p<0.005). Glutamate receptor ionotropic NMDA 2A (Grin2a) also showed significantly higher levels of gene expression in Aged than Young rats (Student–Newman–Keuls [SNK], p=0.007) or Adult animals (SNK, p<0.001). However, Grin2a levels significantly decreased in Young as compared to Adult animals (SNK, p=0.002). Bardet-Biedl syndrome 2 (Bbs2), doublecortin-like kinase 1 (Dcamkl1), enolase 2 (Eno2), GABA transporter 3 (Gat3), isk-related voltage-gated K+ channel 2 (Kcne2), and neurofilament heavy polypeptide (Nefh) showed no significant differences between age groups (Table 2). In contrast to these results, Pcgf1 and Grin2a showed significantly decreased expression with diabetes as compared to nondiabetic controls (Table 3).

To visualize the relationship of gene expression profiles between age groups and our previously published characterizations of the rat retina after three months of diabetes, a principal component analysis was performed. The principal component analysis was used to combine all of the data into one visualization showing the relationship of the expression patterns from the different groups. For the purposes of this study, each gene was plotted on its own axis to form a multidimensional space from which the principal components were identified in order of decreasing variance. This permits visualization of the most significant relationships between groups in a standard two-dimensional plot, when the first two components are plotted. Retinal gene expression data was scaled to the mean of the Young group for each gene. Data from diabetic and nondiabetic controls were scaled to the mean of the controls on a gene-by-gene basis. The first component accounted for 74% of variance between groups, progressively separating the Young and Nondiabetic Control from the Adult, Aged, and Diabetic rats along the x-axis (Figure 5); this was interpreted as gene expression changes with diabetes. The second component (26% of variance) separated the Young, Nondiabetic Control, and Diabetic groups from the Adult and Aged groups along the y-axis. This component was interpreted as representing the age-specific changes in gene expression. In total, this visualization demonstrates that while there are some commonalities in gene expression patterns with diabetes and aging, they are not superimposable molecular phenotypes.

Microvasculature

Previous investigations of microvascular dysfunction with retinal aging have demonstrated declining RPE cell integrity [58], microvascular lesions [56,59], increased vessel leakage, and vascular cell apoptosis [22]. In the aged retina, decreased occludin expression is associated with increased vascular permeability [20] and RPE dysfunction, although the specific biochemical mechanisms of declining vascular integrity remain to be fully determined. We have previously demonstrated that Edn2, Ednrb, Icam1, Nppa, and Vegfa showed significantly altered expression patterns in the diabetic rat retina when compared to nondiabetic controls [45]. Icam1 and Edn2 are increased with diabetes and show similar magnitude increases in expression with age. Icam1 is elevated in patients diagnosed with proliferative DR, and contributes to blood-retinal breakdown and DR development [60-62]. Our findings agree with a previous report of increased retinal Icam1 expression with advanced age [62]. Icam1 has been implicated in leukocyte cell adhesion to the retinal vascular endothelium in the diabetic retina, promoting leukostasis and therefore retinal endothelial cell injury and death [19,63]. Increased Icam1 expression with aging could work additively in aging diabetics to increase the risk of microvascular dysfunction. Edn2 is a strong vasoconstrictor and regulator of ocular blood flow [64]. As blood flow abnormalities are observed with both diabetes and aging, the common upregulation of Edn2 may represent a common causative factor. Edn2 expression has been localized to photoreceptors and signals through the Ednrb receptor on Müller cells [65]. This signaling may serve as an integration point between microvascular dysfunction, microglial activation, and inflammation, which are observed in both diabetes and aging.

One interesting finding was that there was no change in Vegf with age in the retina. Previously, we have consistently identified a modest but significant decrease in Vegf mRNA in a streptozotocin (STZ)-induced rat model of diabetes [45,46]. While induction of retinal Vegf protein in late-stage DR patients has been described, data on the exact time course of Vegf induction are conflicting, with reports of Vegf expression being unchanged at 3 months [66], increased at 6 months [67], and decreased at 6 months [68] of STZ-induced diabetes in the rat.

Neuronal function

Neuronal dysfunction and vision loss with aging are well characterized at the psychophysical, electrophysiological, and biochemical levels, and suggest broad impairment of the neural retina that affects photoreceptors, interneurons, and ganglion cells [9,10,13]. Our previous studies have shown marked alterations in the expression of Bbs2, Dcamkl1, Eno2, Gat3, Grin2a, Kcne2, Nefh, and Pcgf1 in the diabetic retina (Table 3). Of these genes, two (Pcgf1 and Grin2a) were found to be regulated with age. Pcgf1 increased significantly with age while Grin2a showed decreased expression in Adults when compared to Young animals, yet Aged animals showed significantly increased expression when compared to both Young and Adult animals. These results are interesting, as both of these genes show a general upregulation with age while with diabetes, they were found to show significant downregulation.

Pcgf1, also known as Nspc1 (nervous system polycomb 1), is a novel transcriptional repressor that is highly expressed in the developing nervous system. This gene has been demonstrated to be responsive to protein kinase C (PKC) signaling and has also been shown to be associated with epigenetic mechanisms of gene transcription regulation [69,70]. Upregulation of this gene in aged animals could alter transcriptional control of genes regulated by Pcgf1, either directly (functioning as a transcriptional repressor) or indirectly (functioning as an epigenetic modifier). Further gene-specific studies of Pcgf1 are needed to determine the specific target genes regulated by Pcgf1.

Grin2a (glutamate receptor, ionotropic, N-methyl D-aspartate 2A) encodes the glutamatergic NMDAR2a receptor responsible for excitatory neurotransmission. Although the function of this receptor is unknown in the retina, the heteromeric subunit composition of the NMDA receptor is associated with variations in excitotoxicity [71]. Grin2a demonstrated a unique gene expression profile across ages, decreasing in adulthood as compared to Young animals, while seeming to overcompensate and show the highest expression in Aged animals. This could be a maladaptive compensatory response to other changes in neuronal gene expression and diminished neurotransmission that occur with advanced age, given the sharp increase in expression found specifically in the Aged animals. The functional outcome of increased Grin2a could facilitate neurotransmission, but also result in excitotoxicity and the induction of downstream inflammatory processes. The hypothesis seems to be plausible given the higher Grin2a expression in both the Young and Aged animals.

Inflammation

Age-related increases in inflammatory processes are reported throughout the CNS. In the retina, vascular dysfunction likely contributes to proinflammatory signaling in a feed-forward cycle due to loss of barrier integrity, toxic drusen deposition, and monocyte infiltration [17,72]. Increased microglial activation, indicated by the presentation of MHC class II- and ED1-immunoreactive cells, has been reported in the retinal parenchyma in aged rats [20]. In aged mice (20 months), markers of para-inflammation observed in this study (C1s, Gpb2, Icam1, and Spering1) have been reported [62]. Furthermore, elevated circulating levels of the proinflammatory glycoprotein Chi3L1 have been demonstrated in aged humans (>80 years) compared with younger adults (18–30 years) [19]. Oxidative stress, the accumulation of photoreactive pigments, necrotic/apoptotic cells, and leukocyte infiltration have been implicated in the activation of inflammatory pathways involving these and other proinflammatory molecules.

A hallmark of DR is a marked increase in the expression of genes involved in the regulation of inflammatory processes. We have previously demonstrated that Birc4, Carhsp1, C1-Inh (Serping1), C1s, Ccr5, Cd44, Chi3L1, Elovl1, Gbp2, Gfap, Hspb1, Jak3, Lama5, Lgals3, Lgals3bp, Litaf, Mccc1, Nr3c1, Pedf, Stat1, Stat3, Timp1, and Tnfrsf12a significantly change in expression in the diabetic retina. Of these genes, 10 were found to be regulated with aging, specifically C1s, Chi3L1, Gbp2, Gfap, Jak3, Litaf, Pedf, Serping1, Stat3, and Tnfrsf12a. All of these genes increased in expression with DR, and several have been previously reported to change with aging in the CNS. Interestingly, similar increases in expression were seen with age in all genes except Pedf, which was actually found to significantly decrease in expression.

C1s (complement component 1, s subcomponent) encodes a serine protease that is a subunit of the complement subcomponent C1 complex, which is involved in activation of the complement system. Although the exact role of C1s upregulation in DR and aging is unknown, it has been shown to be expressed in retinal ganglion cells. Interestingly, elevated C1s expression was observed with retinal ganglion cell death in response to serum depravation, suggesting a role for the complement pathway in regulating neuronal cell death in the retina [73]. Therefore, elevated expression of C1s could be indicative of complement pathway activation and increased inflammation in response to retinal ganglion cell death. Furthermore, Chi3L1 (chitinase 3–like 1) is an inflammatory glycoprotein secreted by activated macrophages and is suggested to be involved in innate immune system activation, macrophage differentiation, extracellular matrix remodeling, and vascular cell migration [74]. Our tissue-specific findings agree with clinical reports of increased plasma Chi3L1 levels in Type 1 and Type 2 diabetic subjects [75,76]. Elevated Chi3L1 expression in the retina implies a general increase in retinal inflammation with both diabetes and aging. Although the exact role of Gbp2 (guanylate binding protein 2, interferon-inducible), an interferon-inducible GTPase, is unknown in the pathology of DR, our previous studies have shown a significant increase in expression in diabetic animals when compared to nondiabetic controls [46].

Gfap (glial fibrillary acidic protein) is an intermediate filament protein highly expressed in astrocytes that has been shown to be differentially expressed with DR. Increased Gfap expression in Müller cells with diabetes has been described previously [44,68], and we have observed the insulin therapy–resistant induction of Gfap mRNA with diabetes [49]. The step-wise induction of Gfap with increasing age is prototypical of increased para-inflammation with aging. Jak3 (janus kinase 3) is a proinflammatory component of the Jak-Stat signaling pathway that activates Stat3 (signal transducer and activator of transcription 3). This pathway regulates cytokine signaling and inflammatory responses [77]. Upregulation of Jak3 and Stat3 signaling with aging can induce a broad range of acute phase and other inflammatory mediators. The transcription factor Litaf (lipopolysaccharide-induced TNF factor) regulates the production of TNF-α, a potent and well characterized proinflammatory cytokine in the retina [78]. Our studies have demonstrated significant increase in Litaf expression with both DR and age, suggesting increased levels of inflammation. Tnfrsf12a (tumor necrosis factor receptor superfamily, member 12 A; aka Tweakr, Fn14) is the receptor for TNF-related weak inducer of apoptosis (Tweak) which is a proinflammatory cytokine. Previous studies have demonstrated that Tnfrsf12a activation via Tweak has proinflammatory effects in RPE cells [79] and produces more highly vascularized tumors in athymic mice [80] suggesting both proinflammatory and proangiogenic effects with increased expression.

Contrary to the genes described above, Pedf (pigment epithelium derived factor) has anti-inflammatory properties and has been found to be enriched in the serum of patients with Type 1 diabetes with DR [81]. In one study, Pedf was able to inhibit the migration of cultured endothelial cells in a dose-dependent manner, suggesting its ability to decrease neovascularization [82]. Additionally, a recent report showed how Pedf was able to decrease advanced glycation end-product (AGE)-induced endothelial cell permeability, further suggesting that Pedf plays a role in counteracting angiogenesis associated with DR [83], and AGE-induced inflammation with aging and diabetes [55]. Decreased Pedf expression in the adult and aged retina would diminish a critical protective mechanism against AGE/AGE receptor (RAGE)-induced inflammation, leaving the retina more susceptible to a variety of physiologic insults. Serping 1 (serpin peptidase inhibitor, clade G [C1 inhibitor], member 1), also known as C1-Inh, is an inhibitor of the complement system activation. Patients with DR show increased expression of Serping1, which serves to protect the integrity of the vascular endothelium in the retina [84]. Unlike the diminished expression of Pedf, this protective mechanism is induced with age.

Taken together, the induction of proinflammatory genes (C1s, Chi3L1, Gbp2, Gfap, Jak3, Litaf, Stat3, and Tnfrsf12a) and reduction of the anti-inflammatory Pedf indicate a para-inflammatory state with increasing age in the retina [50]. This generalized proinflammatory state could help explain the increased vulnerability for retinal diseases such as DR and macular degeneration with increasing age. The potential mechanisms leading to the para-inflammatory state observed with increasing age include retinal stress through reactive oxygen species and AGEs. The commonalities in retinal gene expression changes occurring in young hyperglycemic and aged normoglycemic rats indicate that different stressors can lead to some, but not all, of the same inflammatory processes. Clearly, additional research is needed to build upon the existing knowledge related to retinal para-inflammation with aging [62].

These results demonstrate altered retinal gene expression across the adult rat lifespan in a set of genes previously demonstrated to be altered in expression with diabetes. While this analysis focused on a predetermined set of genes, these results provide the rationale for future genome-wide comparisons of retinal transcriptomic changes with aging and diabetes, as well as comparative analysis of retinal proteomic alterations. For both molecular and biochemical analyses, use of isolated retinal cell types, such as those previously reported (e.g., Müller [68], photoreceptors [23]), will provide a clearer picture of the altered pathways and processes with aging and diabetes in retinal cell types. While technically challenging, cell-specific analyses of in vivo models is needed to more clearly tie altered RNA/protein expression to dysfunction in specific cell types and to define gene expression regulatory networks that may be commonly engaged with aging and diabetes. Prospective studies across the lifespan, including induction of diabetes at different ages, will provide more specific insight into whether the effects of aging and diabetes are additive.

The specific inflammatory, neuronal function, and microvascular genes altered with aging provide potential molecular mechanisms for the well documented functional impairments and para-inflammatory state of the retina with advanced age. Additionally, these results support the hypothesis that diabetes and aging share some common molecular alterations. However, these results also demonstrate that the patterns of gene expression with advanced age and diabetes are not identical. Age-related changes in retinal gene expression indicative of para-inflammation may work synergistically with the same para-inflammatory processes induced by diabetes to accelerate DR pathogenesis. Given the growing incidence of diabetes and increasing longevity [85], further investigation of the contribution of age-related retinal gene expression alterations to retinal diseases common in the aged population is needed. The microvascular and inflammatory processes observed in this study may also serve as targets for development of therapies that prevent or treat age-related retinal dysfunction.