MicroRNA‐150 and its target ETS‐domain transcription factor 1 contribute to inflammation in diabetic photoreceptors

Abstract Obesity‐associated type 2 diabetes (T2D) is on the rise in the United States due to the obesity epidemic, and 60% of T2D patients develop diabetic retinopathy (DR) in their lifetime. Chronic inflammation is a hallmark of obesity and T2D and a well‐accepted major contributor to DR, and retinal photoreceptors are a major source of intraocular inflammation and directly contribute to vascular abnormalities in diabetes. However, how diabetic insults cause photoreceptor inflammation is not well known. In this study, we used a high‐fat diet (HFD)‐induced T2D mouse model and cultured photoreceptors treated with palmitic acid (PA) to decipher major players that mediate high‐fat‐induced photoreceptor inflammation. We found that PA‐elicited microRNA‐150 (miR‐150) decreases with a consistent upregulation of ETS‐domain transcription factor 1 (Elk1), a downstream target of miR‐150, in PA‐elicited photoreceptor inflammation. We compared wild‐type (WT) and miR‐150 null (miR‐150−/−) mice fed with an HFD and found that deletion of miR‐150 exacerbated HFD‐induced photoreceptor inflammation in conjunction with upregulated ELK1. We further delineated the critical cellular localization of phosphorylated ELK1 at serine 383 (pELK1S383) and found that decreased miR‐150 exacerbated the T2D‐induced inflammation in photoreceptors by upregulating ELK1 and pELK1S383, and knockdown of ELK1 alleviated PA‐elicited photoreceptor inflammation.


| INTRODUC TI ON
The incidence of diabetes is projected to rise to 33% of the US population by 2050 owing to the obesity epidemic, 1 of which 95% of diabetic patients will have type 2 diabetes (T2D). 2 More than 85% of T2D patients have diabetes-related eye disorders, and 60% develop diabetic retinopathy (DR), the leading cause of blindness in US adults age 20-74. 2,3 While anti-vascular endothelial growth factor (VEGF) treatments significantly improve the outcomes of DR, nearly 30% of patients do not respond to anti-VEGFs, 4,5 making the development of new treatment strategies imperative. Chronic inflammation is a hallmark of obesity and T2D 6,7 and a well-accepted major contributor to DR, 8,9 but numerous studies have indicated that intraocular rather than systemic inflammation is more closely associated with the vascular complications in DR. 10,11 Interestingly, diabetic patients who also have retinitis pigmentosa (RP), a congenital blindness with initial degeneration of rod photoreceptors, rarely develop DR, even though they have other non-retinal diabetic vascular complications. 12,13 Thus, there is a clear inverse correlation between RP and DR. 12,13 In mice, genetic deletion of rod photoreceptors or pharmacological inhibition of photoreceptors reduces retinal inflammation and alleviates progression of DR. 14,15 Therefore, retinal photoreceptors are a major source of intraocular inflammation and directly contribute to vascular abnormalities in diabetes. However, how photoreceptors contribute to intraocular inflammation and vascular complications under T2D is still not well elucidated.
MicroRNAs (miRs) are small non-coding RNAs containing 22~25 nucleotides, and their seed sequences (nucleotides 2-8) pair with the complementary 3' untranslated region (UTR) of their target mRNAs and mediate the translational inhibition or degradation of mRNAs. 16 MicroRNAs represent a set of modulators that regulate metabolism, inflammation and angiogenesis, [17][18][19] and changes of various miR levels observed in diabetic patients have been linked to the pathogenesis of DR. 20,21 Among them, there is a strong correlation between miR-150 downregulation and patients with diabetes and DR. Retinal miR-150 is decreased in the eyes under ischaemic insults 22 and in patients with proliferative DR. 23 Serum miR-150 is decreased in patients with obesity, 24 T1D 25,26 or T2D, 27,28 in association with increased inflammation and upregulation of angiogenic factors. We and others have reported that miR-150 is significantly decreased in the blood, heart, and retina in animals with high-fat diet (HFD)-induced T2D 29 or streptozotocin (STZ)-induced T1D. 20,30 Deletion of miR-150 in mice (miR-150 −/− ) further exacerbates obesity-associated T2D and T2DR compared with the wild-type (WT) mice, including significantly elevated systemic insulin resistance, glucose intolerance, inflammation, 19 worsen retinal light responses, 29 and more severe retinal microvascular degeneration and leakage. 23 Thus, downregulation of miR-150 correlates with the progression of diabetes and DR.
In addition, miR-150 is an intrinsic suppressor of inflammation. 19 Overexpression of miR-150 downregulates TNFα and nuclear factor kappa B (NF-ĸB) induced by lipopolysaccharide (LPS) in endothelial cells. 31 Deletion of miR-150 (miR-150 −/− ) exacerbates the increase of TNFα and IL-1β in mice with an HFD-induced T2D. 19 We previously showed that miR-150 −/− -HFD mice have more severe inflammation in photoreceptors and exacerbated vascular degeneration compared with the WT-HFD mice. 23 Overexpression of miR-150 protects the retinal vasculature from degeneration induced by oxygen-induced retinopathy (OIR), a well-established model for pathological angiogenesis. 32 Moreover, overexpressing miR-150 restores endothelial cell functions including proliferation and migration. 33 Therefore, miR-150 could restrain the development of DR by mitigating inflammation in the neural retina especially in photoreceptors. However, how miR-150 and its downstream targets contribute to diabetesinduced inflammation in photoreceptors remains unclear.
The biological processes mediated by miRs and their targets are often tissue-and cell-type-dependent. 16 Previously, Dr. B. Zhou's laboratory screened the top 30 predicted target genes of miR-150 by combining computational analyses, transcriptome profiles, and reporter assays, 19,34,35 and identified new bona fide targets that are pro-inflammatory. 19 Among them, ETS-domain transcription factor 1 (Elk1), eukaryotic translation termination factor 1 (Etf1), early growth response 1 (Egr1) and MYB proto-oncogene (c-Myb) are expressed in photoreceptors and retinal endothelial cells. In this study, we used a functional assay with cultured photoreceptors treated with palmitic acid (PA) to generate a high-fat environment for cells. We found that PA elicited decreased miR-150 with a consistent upregulation of Elk1 but not others, so we focussed on Elk1 and its associated signalling in promoting retinal inflammation in T2DR with in vitro and in vivo assays. We compared WT and miR-150 −/− mice fed with an HFD and determined the role of miR-150 and Elk1 in mediating inflammation in photoreceptors under T2D. We further used cultured 661W cells, a mouse photoreceptor cell line, 36 to decipher the relationship between miR-150, Elk1, ELK1 and inflammation in photoreceptors. We delineated the critical cellular localization of phosphorylated ELK1 at serine 383 (pELK1 S383 ) and HFD-or PA-associated inflammation in photoreceptors. Our data indicate that decreased miR-150 exacerbates the T2D-induced inflammation in photoreceptors by upregulating ELK1 and pELK1 S383 , and knocking down ELK1 alleviates the inflammation and reduces pELK1 S383 .

| Animals
Four-week-old male C57BL/6J mice (wild type, WT) were purchased from the Jackson Laboratory. B6(C)-Mir150 tm1Rsky /J (miR-150 −/− ) mice were originally purchased from the Jackson Laboratory, and a colony was bred and maintained at Texas A&M University. Only male mice were used in this study. All animal experiments were approved Non-fasting blood glucose levels and glucose tolerance were measured monthly by taking blood from the tail vein. Glucose levels were measured using a Clarity BG1000 blood glucose monitoring system (Clarity Diagnostics).

| Lipofectamine transfection
Transfection was conducted using the Lipofectamine 3000 kit (#L3000015, Invitrogen/ThermoFisher) according to the manufacturer's instruction. Briefly, the 661W cells were seeded at 30% confluency and allowed to grow for 24 h to reach 50% confluency. For

| Quantitative real-time RT-PCR (qPCR)
After cells were collected, total RNA from each sample was prepared by using a commercially available purification kit (miRNeasy mini kit; #217004, Qiagen). From each sample, 500 ng of total RNA was used to quantify miR-150 or mRNAs by qPCR using a TaqMan For each individual experiment, a standard curve was generated with known quantities of RNAs loaded in curved dilutions (ie 2x, 1x, 1/2, 1/4, 1/8, 1/16 and 1/32). The cycle values, corresponding to the log values of the standard curve quantities, were used to generate a linear regression formula. The amplification efficiency of the qPCR reactions (90%-100%) was calculated using the standard curve.
The quantification of sample RNA was calculated by the 2 (−ΔΔCt) method 37 using U6 (for miR-150) or β-actin (for other genes) as the internal control.

| Western blot
Samples for Western blots were collected, prepared and analysed as described previously. 38

| Immunofluorescent staining (retina and cultured cells)
Mouse eyes were collected, fixed with 4% paraformaldehyde and processed for paraffin-embedded sectioning after 24 weeks of the diet regimen. Paraffin sections (4 μm) of mouse eyes from all four experimental groups were mounted on the same glass slide. The retina sections were deparaffinized by heating at 57°C followed by washing with xylene and serial dilutions of ethanol. The antigen retrieval for retinal sections was carried out in a sodium citrate buffer (10 mM sodium citrate, 0.05% Tween 20, pH 6.0) at 80°C for 1 h. The 661 cells cultured on coverslips were fixed with 4% paraformaldehyde at room temperature for 1 h and permeabilized with 0.1% Triton X-100 in 0.1% sodium citrate at 4°C for 10 min.
Eye sections or coverslips were then blocked with 10% goat serum in phosphate-buffered saline (PBS) for 2 h at room temperature and incubated with primary antibodies overnight at 4°C. After washing with PBS, sections or coverslips were incubated with secondary antibodies for 2 h at room temperature and mounted with ProLong Gold antifade mountant with 4′,6′-diamidino-2-phenylindole (DAPI) (#P36935, ThermoFisher). Images were obtained using a Zeiss Axiovert 200 M microscope (Carl Zeiss AG). All fluorescent images were taken under identical settings including light intensity, exposure time and magnification. 23,38 The fluorescent intensity was measured in the inner and outer

| Statistical analysis
All data are presented as mean ± standard error of the mean (SEM).
Student's t test or one-way analysis of variance (ANOVA) followed by Tukey's post hoc tests were used for statistical analyses among groups. Throughout, p < 0.05 was regarded as significant. Origin 9.0 (OriginLab) was used for statistical analyses.

| Deletion of miR-150 (miR-150 −/− ) exacerbates inflammation in the obesity-associated T2D retina
Inflammation is a major culprit in the pathogenesis of DR. 9, 39 We previously showed that mice fed with an HFD develop obesityassociated T2D, in which inflammation is detected in the vitreous and neural retina, and phosphorylated NFĸB P65 (pP65; a biomarker for inflammation) but not the total P65 is significantly increased in the whole retina. 40,41 Using immunostaining, we found that deletion of miR-150 (miR-150 −/− ) further exacerbates retinal inflammation in obesity-associated T2DR 23

| MiR-150 knockout exacerbates palmitic acid (PA)-elicited inflammation in cultured 661W cells
As miR-150 is decreased in the blood and retina in diabetic patients 23,25,26,42 and animals with streptozotocin-induced T1D or HFD-associated T2D, 23 there is a correlation between decreased miR-150 and diabetes. We next tested whether decreased miR-150 directly triggered inflammatory responses, or miR-150 is a medium linking diabetic insults and inflammation using cultured 661W cells. 36 The 661W cells were originally derived from a mouse retinal tumour and characterized as a cone-photoreceptor cell line for expressing opsins, transducin and arrestin, 36 and they are widely used in photoreceptor research. We found that cultured cells treated with palmitic acid (PA, 100 µM) had significantly increased levels

| Elk1, but not c-Myb, Etf1 or Egr1, is the direct target of miR-150 in response to PA treatments in 661W cells
There are several bona fide targets of miR-150 known to be proinflammatory including c-Myb, Etf1, Egr1 and Elk1. 19 In cultured adipose B lymphocytes, lipopolysaccharides (LPS) induces inflammatory responses that correspond with decreased miR-150 and upregulated c-Myb, Etf1, Egr1 and Elk1, and knockdown of miR-150 further increases the expression of these genes and escalates LPS-elicited inflammation. 19 Because miRs and their downstream targets have tissue-and cell type-specific bioactivities, 16,43 we set forth to determine the downstream target(s) of miR-150 responsible for diabetesassociated inflammation in retinal photoreceptors.
Using cultured 661W cells treated with 100 µM PA, we found that at 3 h these cells had transient increases of c-Myb, Etf1 and Egr1 ( Figure 3A-C), even though miR-150 was decreased consistently in the presence of PA (3-24 h; Figure 3D). Elk1 was the only downstream target that was consistently increased in the presence of PA (3-24 h; Figure 3E). In addition, ELK1 protein was decreased in cells transfected with a miR-150 mimic (150m) but increased in cells transfected with the miR-150 inhibitor (150in; Figure 3F). These results confirmed that Elk1 is a direct target of miR-150 in 661W cells in responding to PA.

| Knocking down Elk1 alleviates PA-induced inflammation in 661W cells
We then tested whether Elk1 could regulate PA-induced inflammation in 661W cells. The 661W cells were first transfected with the siRNA of Elk1 (siElk1) to knock down the ELK1 protein ( Figure 4A

| Knocking down miR-150 increases cytoplasmic pELK1 S383 , while PA treatments increase nuclear pELK1 S383 in 661W cells
To verify that deletion of miR-150 further upregulated the HFDinduced increase of pELK1 S383 , and whether overexpression of miR-150 might prevent HFD-induced increase of pELK1 S383 , we next examined the differential effects of PA and miR-150 in regulating pELK1 S383 (nucleus versus cytoplasm) in 661W cells. After is not sufficient to downregulate pELK1 S383 (Figure 6) or overcome PA-induced inflammation ( Figure 2B).

| Knocking down Elk1 decreases cytoplasmic pELK1 S383 and prevents PA-elicited increase of nuclear pELK1 S383 in 661W cells
We showed that knocking down Elk1 alleviated PA-induced inflammation in 661W cells (Figure 4). In order to understand whether the suppression of inflammation is correlated with the cellular distribution of pELK1 S383 , we used siRNA to knock down Elk1 (siElk1) to determine cytoplasmic versus nuclear pELK1 S383 levels in 661W cells. While PA treatments elicited significant increases of nuclear pELK1 S383 , transfection with siElk1 significantly decreased cytoplasmic pELK1 S383 and attenuated the PA-induced increase of nuclear pELK1 S383 compared to cells transfected with the negative control (siNC; Figure 7). Therefore, knocking down Elk1 decreased PA-elicited increases of pELK1 S383 (Figure 7), which implies that downregulation of pELK1 S383 could dampen PA-induced inflammation ( Figure 4).

| DISCUSS ION
Chronic inflammation is a manifestation of diabetic retinas. 23,40,46 Diabetic conditions such as hyperlipidaemia and hyperglycaemia cause increases of pro-inflammatory molecules in retinal neurons including photoreceptors. 47,48 Photoreceptors are major contributors to inflammation in the diabetic retina, 11 Using cultured 661W cells, we identified that Elk1 is the direct target of miR-150 that mediates PA-elicited inflammation. The activation and specific function of ELK1 are phosphorylation sitedependent, 51,52 as phosphorylation on S383 (pELK1 S383 ) is known to promote inflammation. 44,45 Phosphorylation site-dependent functions for a protein are not unique to ELK1. For example, phosphorylation of AMP-activated protein kinase (AMPK) at Ser485/491 facilitates cardiac hypertrophy, while phosphorylated AMPK at Thr172 mediates the antihypertrophic response. 53 Phosphorylated ELK1 at S383 (pELK1 S383 ) translocates from the cytoplasm to the nucleus, and nuclear pELK1 S383 further transactivates its downstream genes to promote inflammation. 44,45 In this study, we found that pELK1 S383 increased in the retinal ONL of the HFD mice  1 and 2). Our results suggest that HFD/PA may induce inflammation in photoreceptors by increasing nuclear pELK1 S383 , and the upregulated cytoplasmic pELK1 S383 could further exacerbate the inflammation. While phosphorylated ELK1 promotes the transcription of downstream genes, SUMOylated ELK1 represses the F I G U R E 7 Knocking down Elk1 decreases cytoplasmic pELK1 S383 and prevents PA-elicited increase of nuclear pELK1 S383 in 661W cells. 661W cells were first transfected with siRNA-negative control (siNC) or Elk1 siRNA (siElk1) and then treated with culture medium (Ctrl), BSA or 100 µM PA (PA) for 24 h. Cells were fixed, processed, and immunostained with pELK1 S383 (green) and DAPI (blue). The fluorescent intensities of pELK1 S383 in the cell nucleus and cytoplasm were measured using ImageJ, and the nuclear/cytoplasmic ratio pELK1 S383 was calculated. siRNC-Ctrl: open diamond; siRNC-BSA: grey diamond; siNRC-PA: dark diamond; siElk1-Ctrl: open triangle; siElk1-BSA: grey triangle; siELK1-PA: dark triangle. Nuclear pELK1 S383 : # indicates a statistical significance when comparing PA-treated cells to BSA-treated cells. * indicates a statistical significance between siNC-PA and siElk1-PA. Cytoplasmic pELK1 S383 : * indicates that all siElk1 groups are significantly different from all siNC groups. Nuclear/Cytoplasmic ELK1 S383 : # indicates a statistical significance when comparing PA-treated to BSA-treated cells. * indicates a statistical significance between siNC-PA and siElk1-PA. p < 0.05, one-way ANOVA. Scale bar: 30 µm transactivation activity. 54,55 Stress elevates the phosphorylation of ELK1 but removing the SUMOylation. 56,57 It is possible that decreased SUMOylated ELK1 in the cytoplasm partially contributes to the upregulation of cytoplasmic pELK1 S383 (Figures 5 and 6), which may explain the exacerbated inflammation in miR-150-knockdown cells.
The nuclear content of pELK1 S383 relative to that in the cytoplasm (nuclear/cytoplasmic ratio; N/C) represents the translocation of pELK1 S383 from the cytoplasm to the nucleus, in which an increased N/C value of pELK1 S383 implies an increased inflammation at the cellular level. 44,45 We showed that PA treatments increased the N/C value of pELK1 S383 in 661W cells (Figure 6), and knocking down Elk1 (siElk1) decreased nuclear pELK1 S383 and the N/C value in PA-treated 661W cells (Figure 7), which correlated with the downregulation of pP65 ( Figure 4B). In addition, cytoplasmic pELK1 S383 was decreased in the siElk1 cells, which may further repress the activity of ELK1. The results imply that knocking down

DATA AVA I L A B I L I T Y S TAT E M E N T
The data presented in this study are available from Fei Yu and Dr.
Gladys Ko upon reasonable request.