High-Fat Diet Induces Toll-Like Receptor 4-Dependent Macrophage/Microglial Cell Activation and Retinal Impairment

Abstract

Purpose.: The toll-like receptor 4 (TLR4) signaling pathway is involved in chronic inflammation and insulin resistance, which are associated with obesity and diabetes mellitus. In the present study, a model of high-fat diet (HFD) feeding of mice was used to investigate the role of TLR4 in overnutrition- and obesity-associated inflammation and infiltration of macrophages and microglia in the retina.

Methods.: Wild-type C57BL/6 and TLR4 knockout (TLR4KO; B6.B10ScN-Tlr4^lps-del/JthJ) mice were fed a HFD or control chow diet (CD) for 6 months. The TLR4 expression, the relative increase in macrophages/microglia (CD11b⁺ and CD45⁺ cells), the presence of markers of oxidative stress (gp91phox and malondialdehyde; MDA), and DNA damage (phosphorylated histone H2AX; γH2AX) were assessed by real-time PCR and immunofluorescence studies.

Results.: The HFD for 6 months showed increased obesity, glucose intolerance, and insulin resistance in mice. Toll-like receptor 4 expression was found in vascular pericytes at the inner retina. Increased CD11b⁺ and CD45⁺ cells, phosphorylated NF-κB, interleukin-6, gp91phox, MDA, and γH2AX were observed in the retina of mice fed a HFD compared to CD counterparts. TLR4KO mice did not show the adverse effects of HFD.

Conclusions.: Our results indicate that HFD-induced macrophage/microglial cell activation and retinal impairment were reduced in the absence of TLR4. The findings suggest that TLR4 is implicated in the pathogenesis of retinal diseases caused by metabolic disorders.

Overnutrition and reduced exercise due to the modern lifestyle are often associated with obesity and systemic diseases, including type 2 diabetes mellitus (T2DM), in human populations worldwide.¹ Obesity can lead to islet β-cell failure, increased insulin resistance, and subsequently T2DM.² Chronic and poorly controlled T2DM may result in multiple systemic vascular complications including diabetic retinopathy (DR), which is the primary cause of visual loss in diabetic patients between 20 and 64 years of age.³ 

The overexpression of proinflammatory cytokine in adipose tissue connects obesity with inflammation,⁴ leading to systemic insulin resistance and metabolic syndrome through the secretion of adipokines.⁵ In the central nervous system (CNS), obesity-induced metabolic alterations may synergize with age to impair brain function and accelerate age-related diseases of the nervous system.⁶ High-fat diet (HFD) feeding is an important model for insulin resistance and T2DM.⁷ A HFD elicits increases in tumor necrosis factor-alpha (TNF-α), which leads to the activation of microglia and macrophages in murine brains.⁸ In mouse retina, 12 weeks of HFD induces the differential expression of several stress-related genes,⁹ higher levels of inducible nitric oxide synthase (iNOS), and 4-hydroxynonenal and retinal degeneration.¹⁰ However, the mechanism underlying this effect has not been fully explored, particularly in the retina. 

The toll-like receptor (TLR) family, composed of 13 members in mammals, is the best-characterized class of pattern recognition receptors capable of detecting conserved molecular patterns present in a broad spectrum of pathogens and triggering innate immune responses.¹¹ Toll-like receptors are chronically activated in patients with T2DM.^12,13 Nutrient metabolism could elicit TLR-associated innate immune responses, triggering the release of inflammatory molecules from monocytes or B cells into the circulation.^14,15 Moreover, nutrients or other ligands may elicit TLR4 signaling to induce inflammasome activation in dendritic cells, leading to the release of cytokines previously implicated in insulin resistance.¹⁶ 

In this study, C57BL/6 mice fed a HFD exhibited impaired glucose tolerance and insulin sensitivity that was physiologically equivalent to T2DM. The effect of HFD on pathological changes within the retina was examined in both wild-type (WT) and TLR4 knockout (TLR4KO) mice in order to elucidate the involvement of TLR4 signaling in DR. 

Methods

Animals

Six- to eight-week old male mice with body weights of 16 to 20 g at the beginning of the study were used. Toll-like receptor 4 knockout, Tlr4^−/− (B6.B10ScN-Tlr4^lps-del/JthJ), mice were purchased from Jackson Laboratory (Bar Harbor, ME, USA). Wild-type C57BL/6 mice (C57BL/6JNarl) were purchased from the National Laboratory Animal Center, Taiwan. Both groups of mice showed no Rd8 mutation in the Crb1 gene (Supplementary Fig. S3).¹⁷ Housing and all surgical procedures, analgesia, and assessments were implemented in the animal center of Kaohsiung Chang Gung Memorial Hospital, an Association for Assessment and Accreditation of Laboratory Animal Care-accredited, specific pathogen-free (SPF) facility following national and institutional guidelines. The study was approved by the Institutional Animal Care and Use Committee in our hospital. All animals were managed and experiments were conducted according to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

HFD Feeding Induced Obesity, Glucose Intolerance, and Insulin Resistance

Male WT and TLR4KO mice were maintained in a temperature-controlled room with a 12-hour light-dark cycle for 6 months. They were allowed access to water and diet ad libitum. The experimental group was fed a HFD (D12331; Research Diets, Inc., New Brunswick, NJ, USA), and the control group was fed a chow diet (CD). The HFD (total 5.56 kcal/gm; 58 kcal% fat) was composed of protein 23.0 gm%, carbohydrate 35.5 gm%, and fat 35.8 gm%. The CD (total 4.07 kcal/gm; 11 kcal% fat) was composed of protein 16.8 gm%, carbohydrate 74.3 gm%, and fat 4.8 gm%. The diet contained approximately 20% protein and met the American Institute of Nutrition requirements for mice with regard to mineral and vitamin content. Blood tests were conducted for plasma sugar levels as well as for the insulin resistance index by using an intraperitoneal glucose tolerance test (IPGTT) and an intraperitoneal insulin tolerance test (IPITT). Animals were euthanized after 6 months of feeding. The eyes were extracted for studies. 

Animal Glucose and Insulin Tolerance Tests

Intraperitoneal glucose tolerance tests and IPITT were performed (n = 3 of each group) after the animals had fasted for 16 hours. Mice were placed in restrainers, and blood samples were obtained by tail bleeding and analyzed by glucose meter (Optium Xceed XCN 289-2337; Abbott Taiwan, Taipei, Taiwan) immediately before (0 minutes) and at 60 and 120 minutes after intraperitoneal glucose (1 g/kg in 0.9% NaCl) or 30, 60, and 90 minutes after insulin (0.75 U/kg) injection. 

Preparation of Mouse Retina

Eye samples were prepared for immunofluorescence by modifying a published procedure.¹⁸ Briefly, after extraction, the eyeballs were fixed in 4% paraformaldehyde in 0.1 M phosphate buffer for 1 hour, soaked in 10% sucrose in 0.1 M phosphate buffer overnight, and then shifted to 20% sucrose in 0.1 M phosphate buffer for 8 hours. The tissues were embedded with optimal cutting temperature compound (Sakura Finetek, Torrance, CA, USA) and stored at −80°C until further use. 

Immunofluorescence Study

For the immunofluorescence study, mouse retinas (n = 4) were cut as 8-μm-thick sections in the region of 400 μm from the center of the eyeballs. Sections were blocked at room temperature with phosphate-buffered saline (PBS) containing 5% normal goat serum for 1 hour and mouse on mouse blocking reagent (Vector, Burlingame, CA, USA) for 1 hour, when mouse-derived primary antibody was used. The sections were then incubated overnight at 4°C with the following primary antibodies diluted in PBS: TLR4 (Santa Cruz Biotechnology, Santa Cruz, CA, USA), isolectin GS-IB₄ conjugate (Invitrogen, Carlsbad, CA, USA), α-smooth muscle actin (α-SMA; Abcam, Cambridge, UK), BRN3a (Abcam), monocyte chemotactic protein-1 (MCP-1; Santa Cruz Biotechnology), CD11b (clone M1/70; BioLegend, San Diego, CA, USA), CD45 (BioLegend), phospho-H2A histone family member X (γH2AX; Cell Signaling, Danvers, MA, USA), 8-hydroxy-2′-deoxyguanosine (8-OHdG; JaICA, Shizuoka, Japan), malondialdehyde (MDA; JaICA), nitrotryptophan (JaICA), and glutamine synthetase (GS; Abcam). Sections were next incubated with secondary antibody conjugated with Alexa Fluor 488 or 594 (Invitrogen) and with 4′,6-diamidino-2-phenylindole (DAPI) for nuclear staining (Invitrogen). Coverslips were mounted on slides using fluorescence mounting medium (Dako, Glostrup, Denmark), and fluorescence images were captured and analyzed using the Leica DMI3000 B fluorescent microscope (Leica, Singapore) at ×200 magnification unless otherwise specified. To quantify the numbers of macrophage/microglia in retina, cells with CD11b- and CD45-positive cells were counted in three nonoverlapping views of 250- × 100-μm size within the central one-third of the retina from each section by two independent observers. 

RNA Expression Analysis Using Real-Time PCR

Individual retinas (n = 6 from each group) were obtained from eyeballs of mice and homogenized with sonication. Total RNA in retina was extracted using the Qiagen RNeasy Plus Universal Mini Kit (Venlo, The Netherlands). After quantification, 2 μg total RNA was used to synthesize cDNA by reverse transcription with the PrimeScript RT Reagent Kit (Takara Bio, Otsu, Japan) according to the manufacturer's recommendations. Real-time quantitative PCR (RT-qPCR) was performed using the Applied Biosystems 7500 Real-Time PCR system (Life Technologies, Carlsbad, CA, USA) with SYBR green assays. We used the following primers in RT-qPCR: β-actin forward 5′-AGGCCCCTCTGAAACCTAAG-3′, reverse 5′-CAACACAGCCTGGATGGCTAC-3′; TLR4 forward 5′-ATGGCATGGCTTACACCACC-3′, reverse 5′-GAGGCCAATTTTGTCTCCACA-3′; IL-6 forward 5′-TAGTCCTTCCTACCCCAATTTCC-3′, reverse 5′-TTGGTCCTTAGCCACTCCTTC-3′; F4/80 forward 5′-CTTCTGGGCCTGCTGTTCA-3′, reverse 5′-CCAGCCTACTCATTGGGATCA-3′¹⁹; gp91phox forward 5′-TTGGGTCAGCACTGGCTCTG-3′, reverse TGGCGGTGTGCAGTGCTATC-3′. 

Statistical Analysis

Statistical analyses were performed using the Statistical Package for Social Science program (SPSS for Windows, version 13.0; IBM, Armonk, NY, USA). Statistical comparisons were performed using one-way ANOVA with post hoc Bonferroni test. Values were considered significant at a P value of <0.05, and the data are presented as means ± SEM. 

Results

Animal Model

Wild-type and TLR4KO mice fed a HFD for 6 months exhibited systemic changes, including increased body weight (average increase of 45% in WT; 40% in TLR4KO) as well as impaired responses to IPGTT and IPITT when compared to mice fed control CD (Table). Interestingly, TLR4KO mice displayed higher overall body weights than the WT mice regardless of diet (19% and 15% increase with CD and HFD, respectively). No significant differences in response to IPGTT and IPITT were observed between WT and TLR4KO mice fed either CD or HFD. 

Table

 

Comparison of Changes in Body Weight, Fasting Blood Glucose, Glucose Tolerance, and Insulin Tolerance in C57BL/6 Mice Fed Either Control Chow Diet (CD) or High-Fat Diet (HFD) for 6 Months

 

HFD Induced TLR4-Dependent Nuclear Factor κ-Light-Chain-Enhancer of Activated B Cells (NF-κB) Activation and IL-6 Expression in the Retina

Toll-like receptor 4-expressing cells were found from the outer plexiform layer (OPL) to the retinal ganglion cell layer (RGCL) of the inner retina by immunofluorescence (Fig. 1). Multiple cell types with TLR4 expression were observed in retina. The association between TLR4-expressing cells and large vascular structures in the RGCL of WT mice fed the HFD was determined using isolectin GS-IB4 (Supplementary Fig. S1A). The TLR4-expressing cells that encircled a lumen in the RGCL and colocalized with αSMA, a vascular pericyte marker, were observed only in HFD-fed WT mice (Supplementary Fig. S1B). Additionally, relatively weak TLR4 expression was also detected in the BRN3a⁺ cells, presumed to be neuronal cells in the RGCL, and was less commonly found in the inner nuclear layer of both CD- and HFD-fed mice (Supplementary Fig. S1C). Conversely, no TLR4 expression was detected in the glutamine synthetase-positive Müller cells (Supplementary Fig. S1D). The cells in RGCL with phosphorylated NF-κB, an indicator of NF-κB activation, were significantly higher in HFD-fed WT mice as compared to those fed a CD (Fig. 2A). However, the difference in cells with phosphorylated NF-κB was insignificant between CD and HFD-fed TLR4KO mice, whereas a significant reduction of HFD-induced phosphorylated NF-κB was observed in HFD-fed TLR4KO mice when compared to the WT counterparts. Furthermore, WT mice fed a HFD exhibited increased mRNA expression of interleukin-6 (IL-6) in the retina when compared to CD controls, yet no significant change in IL-6 expression was observed between CD- and HFD-fed TLR4KO mice (Fig. 2B). The IL-6-expressing cells located in the RGCL colocalized with TLR4 in WT mice fed a HFD (Fig. 2C). 

Figure 1

 

TLR4 expression was assessed in the inner retina of wild-type (WT) C57BL/6 mice. (A) Immunofluorescence study showed TLR4⁺ cells in the inner retina between outer plexiform layer (OPL) and retinal ganglion cell layer (RGCL) of the WT mice fed the chow diet (CD) and high-fat diet (HFD) but not in TLR4-knockout (TLR4KO) mice (scale bar: 50 μm). (B) In the magnified image of RGCL in HFD-fed WT mice, note multiple types of TLR4-expressing cells including the well-aligned small cells (arrow) and scattered large cells (scale bar: 20 μm). (C) TLR4 mRNA expression in HFD-fed WT mice was higher than in CD-fed controls.

Figure 2

 

TLR4 is associated with proinflammatory markers in the retinas of mice fed the HFD. (A) At RGCL of retina, the percentage of cells with phosphorylated NF-κB (p-NFκB) in HFD-fed mice was higher than that in CD-fed WT mice; however, this difference was insignificant between CD- and HFD-fed TLR4KO mice. (B) HFD induced an increased IL-6 mRNA expression in the retina of WT, but not TLR4KO mice. (C) The white arrow shows the colocalization of IL-6 with TLR4 in large cells of the RGCL in HFD-fed WT mice (scale bar: 50 μm).

HFD Induced TLR4-Dependent Macrophages and/or Microglial Activation in the Retina

The HFD-fed WT mice showed a 103% (P = 0.001) and 91% (P = 0.011) mean increase of CD11b⁺ and CD45⁺ cells, respectively, when quantified in the RGCL, inner plexiform layer (IPL), and OPL of the retina as compared to CD-fed controls. In contrast, the HFD induced only a 20% increase (P = 0.404) in CD11b⁺ and a 33% increase (P = 1.000) in CD45⁺ cells in TLR4KO mice (Fig. 3). Despite same HFD feeding, the prevalence of CD11b⁺ and CD45⁺ cells in the retinas of TLR4KO mice was only 68.3% (5.0 vs. 3.4, P = 0.028) and 41.0% (6.5 vs. 2.7, P = 0.001), respectively, of that observed in WT counterparts. This change in the macrophage and/or microglial cells was also verified with the expression of F4/80 mRNA that increased significantly following HFD in WT, but not the TLR4KO mice (Fig. 3C). Dual immunostaining of the retina from HFD-fed WT mice study showed ramified or amoeboid CD11b⁺ or CD45⁺ cells colocalized or closely associated with TLR4-expressing cells in RGCL of HFD-fed WT mice (Supplementary Fig. S2). 

Figure 3

 

HFD induced a TLR4-dependent increase in macrophages and/or microglia in the retinas of HFD-fed mice. In a fixed area, the numbers of CD11b⁺ (A) and CD45⁺ cells (B) in the inner retina of HFD-fed WT mice were significantly higher than in CD-fed WT mice. Additionally, HFD did not induce significant difference in the numbers of CD11b⁺ and CD45⁺ cells between CD and HFD-fed TLR4 KO mice. (C) HFD elicited increased F4/80 mRNA expression in the retinas of WT, but not the TLR4KO mice (scale bar: 50 μm).

TLR4 Is Associated With Increased Oxidative Stress in the Inner Retinas of HFD-Fed Mice

Following 6 months of HFD, the mRNA expression of gp91phox—a subunit of superoxide-generating nicotinamide adenine dinucleotide phosphate (NADPH) oxidase—in retina was significantly higher than that of CD WT mice. This finding was notably absent in TLR4KO mice (Fig. 4A). Furthermore, an increase in markers of oxidative stress, including MDA and 8OH-dG, was observed in the RGCL, IPL, and OPL, indicating that increased oxidative DNA and lipid peroxidation injury were present in HFD WT animals compared to CD-fed controls. In contrast, HFD TLR4KO mice were relatively spared from this HFD-induced oxidative stress (Figs. 4B, 4D). Interestingly, the CD11b⁺ cells were closely associated with cells expressing MDA and 8OH-dG in the inner retina of WT mice fed the HFD (Figs. 4C, 4E). 

Figure 4

 

TLR4KO inhibited oxidative damage in retinal cells following 6 months of HFD. (A) HFD induced an increase in gp91phox (NOX2) mRNA expression in the retinas of WT, but not TLR4KO mice. (B, D) HFD induced an increase of malondialdehyde (MDA [B]) and 8OH-dG (D) in the inner retina of WT mice, but not TLR4KO. (C, E) Close spatial correlations between the CD11b and MDA (C) and 8OH-dG (E) were observed in the inner retina (×400 images) (scale bar: 50 μm [A, C]; 20 μm [D, E]).

TLR4 Is Associated With Increased DNA Damage in the Inner Retinas of HFD-Fed Mice

DNA damage incurred in the retinal cells was assessed using an antibody against phosphorylated histone H2AX (γH2AX), a marker of DNA double-strand breaks. When compared to those with CD, mice fed HFD showed an increased number of γH2AX⁺ cells primarily in the RGCL and occasionally in the INL. Notably, the percentage of γH2AX⁺ cells in RGCL was 11.8% in CD and 27.9% in HFD WT mice (P = 0.009). In contrast, no significant difference in the percentage of γH2AX⁺ cells was observed in TLR4KO mice (Fig. 5A), indicating that TLR4 is required for the DNA damage occurring in cells in the RGCL. As previously noted, the CD11b⁺ cells also showed a close spatial relationship with γH2AX⁺ cells in the RGCL (Fig. 5B). 

Figure 5

 

TLR4 expression was associated with DNA damage in cells of retina after HFD feeding. (A) A significant increase in the number of cells with phosphorylated histone H2AX (γH2AX, red) was observed in the RGCL of HFD-fed WT mice, but not TLR4KO mice (scale bar: 50 μm). (B) The ramified CD11b⁺ cells (green) were spatially associated with γH2AX⁺ cells (red) in HFD-fed WT mice (×400 images; scale bar: 20 μm).

Discussion

Animals fed a HFD have served as models for T2DM, metabolic syndrome, and obesity for many years.⁷ In this study, C57BL/6 mice fed a HFD for 6 months exhibited increases in body weight and systemic manifestations resembling T2DM in humans, including a higher fasting blood glucose concentration, glucose intolerance, and insulin resistance. Furthermore, the number of CD11b⁺ or CD45⁺ cells in the inner retina, from the OPL to the RGCL, was significantly higher in the HFD-fed mice than in the CD-fed controls. The activation of CD11b⁺ and CD45⁺ macrophages and/or microglia in the retina has been reported in diabetic patients and in animal models of diabetes mellitus that do not use HFD feeding.^20–22 Toll-like receptor 4 could play a critical role in the activation of macrophages and/or microglia in the retina since the difference in diet-induced increase of CD11b⁺ and CD45⁺ cells, as well as the associated oxidative stress and DNA damage, was not observed in TLR4KO mice. This increase of CD11b⁺ and CD45b⁺ cells in retina could be associated with oxidative stress and cell injury (Fig. 6). 

Figure 6

 

Hypothetical diagram of the role of TLR4 in retinal injuries after HFD feeding. TLR4 signaling triggers the expression of inflammatory cytokines that impart a chemotactic effect on macrophages from the bloodstream or residential microglia in retina. These cells could be associated with the oxidative injury of blood vessels and DNA damage of neuronal cells observed in the retina.

The CD11b⁺ or CD45⁺ cells in the retina could be residential immune microglial cells or bone marrow (BM)-derived macrophages. The majority of CD45⁺ retinal cells also express CD11b and are phenotypically similar to CNS microglia.²³ Interestingly, another study investigating the turnover of monocyte-derived cells in the retina, using cells isolated from enhanced green fluorescent protein (EGFP) transgenic mice, found that the EGFP⁺ cells within the ganglion layer were amoeboid in shape and expressed high levels of CD45.²⁴ Therefore, the possibility also exists that HFD promotes migration of BM-derived monocyte precursor cells across the blood–retinal barrier to replace residential microglial cells. The results of these studies indicate that two distinct populations of monocyte-derived cells, the parenchymal microglial cells and perivascular macrophages, may exist and play specific roles in retinal homeostasis, neuroinflammation, and injury. 

In our study, we found that αSMA⁺ cells, which are presumably vascular pericytes, expressed TLR4 in the retinas of mice fed with HFD. Pericytes are adult pluripotent cells that surround capillaries and function as first-line immune defense for the CNS vasculature.^25,26 In our study, TLR4-expressing cells in RGCL may play a critical role in the induction of retinal microglia migration through TLR4-associated NFκ-B activation and the induction of cytokines including IL-6. Previous reports have shown that IL-6-stimulated microglia may impart a cytotoxic effect on neurons in the CNS,²⁷ and pericyte loss is an early event in experimental DR.²⁸ This interaction between inflammatory cells and vascular pericytes could be associated with the loss of the latter in the early stage of DR. 

Toll-like receptor 4 signaling is involved in diabetes mellitus-associated chronic inflammation and is also a molecular link between nutrition, lipids, and inflammation in adipocytes and macrophages.²⁹ Obese and type 2 diabetic subjects have significantly elevated TLR4 gene and protein expression in muscle.³⁰ Toll-like receptor 4 activation leads to increased transcription of proinflammatory genes as well as increases in cytokines, chemokines, reactive oxygen species, and eicosanoid levels that promote further insulin desensitization in both the target and adjacent cells via autocrine, paracrine, and systemic effects.³¹ In type 1 diabetes mellitus animal models, the increase of inflammatory cytokines in serum and macrophages was significantly attenuated in TLR4KO mice.³² Furthermore, increased TLR4 expression was also found in the renal tubules of humans with diabetic nephropathy and directly correlated with interstitial macrophage infiltration and hemoglobin A1c level, whereas an inverse effect was observed with estimated glomerular filtration rate.³³ The effect of TLR4 deficiency in protecting HFD-induced insulin resistance at 16 weeks of experimentation has been reported to be due to reduced inflammatory gene expression in liver and fat.^34,35 However, in our study, insulin resistance and glucose intolerance were instead observed in TLR4KO mice after 6 months of HFD. Dissimilarity in study design and duration may explain the difference between the results of these two studies, since it is possible that the protective effect of TLR4 deficiency on insulin resistance decreases with longer duration. Another possible explanation is that after the activation of receptors for advanced glycation end products (RAGE) and TLR2 induced by HFD in adipocytes, the systemic insulin response may also change.^36–38 Our data show that the TLR4KO mice insidiously gained more weight than the WT C57BL/6 after 4 months of HFD. These findings suggest that the lack of TLR4 may initially protect the mice from insulin resistance, but eventually the excessive weight of TLR4KO mice induces systemic metabolic disorder via other signaling pathways such as RAGE and TLR2.^36–38 

Although the role of TLR4 in DR has been previously unexplored, TLR4 signaling is involved in ischemia-induced retinal damage and inflammation.³⁹ Toll-like receptor 4 has a pivotal role in the pathogenesis of ocular ischemic syndrome created by ligation of the unilateral external carotid artery and the pterygopalatine artery of mice.⁴⁰ Notably, TLR4-mediated microglial activation by endogenous photoreceptor proteins may exacerbate retinal cell death and represent an underlying common pathology in degenerative retinal disorders.⁴¹ In hyperglycemic mice, systemic treatment with TLR4 ligand was associated with pathological changes in early background DR.⁴² Consistent with the above studies, we found that TLR4 is important for macrophage and/or microglial activation in the retinas of mice fed a HFD for 6 months. In the CNS, reactive oxidative species (ROS) could be produced by the microglia, which express the components of NADPH oxidase, including gp91phox.⁴³ Accordingly, the expression of NADPH oxidase enzymes in the CNS and accompanying increase in ROS are associated with a variety of diseases including experimental brain ischemia, multiple sclerosis, and traumatic brain injury.^44–46 In an experimental study of stroke, NADPH oxidase (NOX) was shown to mediate oxidative stress through activation of NOX2 and NOX4, and superoxide within the ischemic core colocalized with the activated microglia and macrophages.⁴⁷ The oxidative burst generated by these cell types is associated with demyelination and free radical-mediated tissue injury characteristic of multiple sclerosis.⁴⁵ Altogether, the evidence supports that the increased levels of oxidative stress in the perivascular region could be the result of an overactivation of macrophages or microglia associated with TLR4 signaling. 

In addition to vascular damage, close contact between ramified CD11b⁺/CD45⁺ cells was also observed in the neuronal cells lining the RGCL in HFD-fed mice. We speculate that the observed DNA damage, demonstrated by the increased γH2AX phosphorylation, was the result of interactions between neuronal cells and macrophages or microglia. Microglia are known to be involved in both neuroprotection and neurotoxicity depending on the level of activation. Thus, microglia may be activated to clear unwanted or toxic cellular debris⁴⁸; however, these cells may have an innate immune memory, resulting in prolonged activation after acute injury and subsequent tissue degeneration.⁴⁹ Neuronal inflammation can induce the release of an “eat-me” signal exposure from otherwise viable neurons, leading to their death through phagocytosis by microglia.⁵⁰ Therefore, it is plausible that retinal microglia may initially respond to metabolic stress and clear the debris from neuronal cells, but begin to generate damaging effects in the presence of TLR4-dependent inflammatory signals. In contrast to the study by Ross et al.⁵¹ showing that the decrease in TLR4 expression and function in the RPE may contribute to retinal degeneration in the Ccl2^−/−/Cx3cr1^−/− mice, degeneration of photoreceptors was not observed in TLR4KO mice in our study. This suggested that knockout of TLR4 expression alone may not be able to duplicate the effect of double knockout of Ccl2/Cx3cr1 on the retina. Nevertheless, the lack of TLR4 and its role in the scavenging of waste products in retina by microglia, neuronal transduction, and visual function could be a topic for future study. 

In conclusion, 6 months of HFD resulted in obesity, impaired glucose tolerance, and insulin resistance in C57BL/6 mice, which was associated with simultaneous TLR4-dependent macrophage or microglia activation in the inner retina and the upregulation of oxidative stress markers at the perivascular zone. However, the involvement of multiple cell types and their complicated interactions within the retina make it difficult to clarify the exact effects resulting from TLR4 signaling on an individual cell layer basis, as well as the crosstalk between neuronal, glial, and perivascular cells and the subsequent retinal injury observed in the present study. Further studies are necessary to answer these questions. 

Acknowledgments

The fluorescence microscope was kindly provided by Chia-Wei Liou, MD, Department of Neurology, and real-time PCR was provided by Stem Cell Research Core Laboratory Facilities (CLRPG8B0052), Kaohsiung Chang Gung Memorial Hospital and Chang Gung University College of Medicine. 

Supported by Grant CMRPG8D0461 from Kaohsiung Chang Gung Memorial Hospital and Chang Gung University College of Medicine and NSC 101-2314-B-182A-123-MY3 from the National Science Council, Taiwan. 

Disclosure: J.-J. Lee, None; P.-W. Wang, None; I-H. Yang, None; H.-M. Huang, None; C.-S. Chang, None; C.-L. Wu, None; J.-H. Chuang, None