Glucocorticoid receptor inhibits Müller glial galectin‐1 expression via DUSP1‐dependent and ‐independent deactivation of AP‐1 signalling

Abstract Galectin‐1/LGALS1 is a hypoxia‐induced angiogenic factor associated with diabetic retinopathy (DR). Recently, we elucidated a hypoxia‐independent pathway to produce galectin‐1 in Müller glial cells stimulated by interleukin (IL)‐1β. Here we revealed glucocorticoid receptor (GR)‐mediated inhibitory mechanisms for Müller glial galectin‐1/LGALS1 expression. Activator protein (AP)‐1 site in the LGALS1 enhancer region, to which activating transcription factor2, c‐Fos and c‐Jun bind, was shown to be essential for IL‐1β‐induced galectin‐1/LGALS1 expression in Müller cells. Ligand (dexamethasone or triamcinolone acetonide)‐activated GR induced dual specificity phosphatase (DUSP)1 expression via the glucocorticoid response element and attenuated IL‐1β‐induced galectin‐1/LGALS1 expression by reducing phosphorylation of these AP‐1 subunits following AKT and extracellular signal‐regulated kinase (ERK)1/2 deactivation. Moreover, activated GR also caused DUSP1‐independent down‐regulation of IL‐1β‐induced LGALS1 expression via its binding to AP‐1. Administration of glucocorticoids to mice attenuated diabetes‐induced retinal galectin‐1/Lgals1 expression together with AKT/AP‐1 and ERK/AP‐1 pathways. Supporting these in vitro and in vivo findings, immunofluorescence analyses showed co‐localization of galectin‐1 with GR and phosphorylated AP‐1 in DUSP1‐positive glial cells in fibrovascular tissues from patients with DR. Our present data demonstrated the inhibitory effects of glucocorticoids on glial galectin‐1 expression via DUSP1‐dependent and ‐independent deactivation of AP‐1 signalling (transactivation and transrepression), highlighting therapeutic implications for DR.

hypoxia-induced angiogenic factor related to cancer progression 2 as well as retinal neovascularization in rodents. 3 Our human sample data showed the significant association of galectin-1 with the neovascular pathogenesis of diabetic retinopathy (DR). 4 Müller glial cells were suggested as a cellular source of its production in the diabetic retina. 4 As there was no correlation between VEGF and galectin-1 protein levels in the eye, 4 VEGF-unrelated regulatory pathways were speculated to exist to induce galectin-1.
Indeed, we also reported that galectin-1 levels increased in DR eyes along with the progression of clinical stages from the preischaemic (background) stage with macular oedema. 5 As concerns the hypoxia-independent mechanism of galectin-1 expression, we revealed that advanced glycation end product (AGE)-induced interleukin (IL)-1β activated extracellular signal-regulated kinase (ERK)1/2 and phosphatidylinositol-3 kinase (PI3K)/AKT pathways in Müller glial cells, leading to galectin-1 expression via these diabetes-associated inflammatory cascades. 5 Importantly, IL-1β was expressed in AGE-positive macrophages migrating into the fibrovascular tissue from proliferative DR patients, and in vivo macrophage depletion significantly reduced diabetes-induced retinal Il1b and Lgals1 expression levels. 5 As the pathogenesis of DR harbours AGE-driven chronic inflammation, glucocorticoid drugs including dexamethasone and triamcinolone acetonide are currently used for the treatment of diabetic macular oedema as an anti-inflammatory strategy, in addition to the first-line anti-VEGF therapy. On top of its pro-angiogenic function, VEGF, originally reported as vascular permeability factor, 6 is also regarded as an inflammatory cytokine that causes oedematous and exudative lesions. Nowadays, the anti-VEGF drugs ranibizumab and aflibercept have expanded their applications from diabetic macular oedema to proliferative DR 7,8 and are established as a gold standard therapeutic strategy broadly covering the early to late clinical stages of DR. However, DR is a complex disorder associated with multiple molecules, and anti-VEGF refractory cases seen in clinical practice are likely to ensue from several other candidates responsive for the pathogenesis of DR. Of these, galectin-1, a hypoxia-induced angiogenic factor, is theorized to be another important therapeutic target covering the entire clinical stages of DR, given that galectin-1, as well as VEGF, contributes to the inflammatory pathogenesis of DR prior to the proliferative (angiogenic) stage. 5 Moreover, galectin-1 showed no correlation with VEGF in intraocular protein concentration 4 possibly because of its different induction pathway caused by AGE-driven inflammation rather than hypoxia, 5 highlighting its potential as an alternative and complementary target in the treatment of anti-VEGF refractory cases. Detailed investigation into mechanisms for inducing the expression of and blocking the function of galectin-1 would therefore be highly warranted in order to improve the long-term management of DR.
Aflibercept is a chimeric glycoprotein consisting of the VEGFbinding domains of VEGFR1 and VEGFR2 fused to the Fc portion of immunoglobulin (Ig)G. 9 Previously, we reported the neutralizing efficacy of aflibercept against galectin-1, which utilizes the binding affinity between sugar chains on the VEGFR2 portion and galectin-1. 4 On top of its blockade by aflibercept, we herein revealed a novel mechanism related to galectin-1 down-regulation, in which glucocorticoids inhibit IL-1β-induced galectin-1 expression in Müller glial cells via transactivation and transrepression. These findings were further supported by blocking experiments with diabetic animals and tissue localization of related proteins in human DR samples.

| Cell line and reagents
The human Müller glial cell line Moorfields/Institute of Ophthalmology-Müller 1 (MIO-M1) was provided from Dr G. Astrid Limb (UCL Institute of Ophthalmology). 10  were purchased from Integrated DNA Technologies and used at 10 nmol/L. Cells were transfected with siRNA using Lipofectamine RNAiMAX Reagent (Thermo Fisher Scientific) following the manufacturer's protocols.
The Pathway Profiling SEAP System (TAKARA BIO) was utilized to detect pathways in response to corticosteroids following the manufacturer's recommendations. After transfection, the medium was collected, and secreted alkaline phosphatase was measured using the chemiluminescent alkaline phosphatase detection kit Great EscAPe™ SEAP (TAKARA BIO) according to the manufacturer's procedures.

| Chromatin immunoprecipitation-quantitative PCR (ChIP-qPCR)
Assays were performed using the SimpleChIP Enzymatic Immunoprecipitation Chromatin IP Kit (Cell Signaling Technology) according to the manufacturer's protocols. Chromatin was immunoprecipitated with rabbit antibodies against activating transcription factor (ATF)2, c-Fos, FosB, c-Jun (Cell Signaling Technology) and glucocorticoid receptor (GR) (Thermo Fisher Scientific).
Normal rabbit IgG (Cell Signaling Technology) was used as control. Thereafter, chromatin immunoprecipitates were evaluated by real-time quantitative PCR (qPCR) using the primers specific for the previously described AP-1-binding site in the LGALS1 enhancer region 13 and GR-binding site in the DUSP1 promoter region, 14 together with 2% input DNA as reference samples. All primers are listed in Table S1. Real-time qPCR was performed using KOD SYBR qPCR Mix (TOYOBO, Tokyo, Japan) and StepOne Plus Systems (Thermo Fisher Scientific). ChIP-qPCR signals were calculated as percentage of input.

| Human surgical samples
During surgery (vitrectomy for traction retinal detachment), five fibrovascular tissues were excised from eyes of patients with proliferative DR and used for immunohistochemistry. This study was conducted in accordance with the tenets of the Declaration of Helsinki and after receiving approval from the institutional review board of Hokkaido University Hospital. Written informed consent was obtained from all patients after an explanation of the purpose and procedures of this study.

| Immunofluorescence microscopy
Immunofluorescence analyses were performed as described previously. 4

| Real-time quantitative PCR (qPCR)
Total RNA isolation and reverse transcription were performed from cells using SuperPrep Cell Lysis & RT Kit for qPCR (TOYOBO) and from tissues using PureLink RNA Mini Kit (Thermo Fisher Scientifc) and GoScrip Reverse Transcriptase (Promega), as previously described. 4,5,15 All primers are listed in Table S1. Real-time qPCR was performed using the GoTaq qPCR Master Mix (Promega) and StepOne Plus Systems (Thermo Fisher Scientific). Gene expression levels were calculated using the 2 −ddCt method.

| Statistical analyses
All the results are expressed as the mean ± SEM (standard error of the mean). Student's t test was used for statistical comparison between groups, and one-way analysis of variance (ANOVA) followed by the Tukey-Kramer method as a post hoc test was used for multiple comparison procedures. Differences between means were considered statistically significant when P values were <.05.

| Requirement of AKT-and ERK1/2-dependent AP-1 activity in IL-1β-induced LGALS1 expression in Müller glial cells
In addition to galectin-1 expression regulated by hypoxia, 4,5,16 we recently revealed IL-1β-induced LGALS1 gene expression via PI3K/AKT and ERK1/2 mitogen-activated protein kinase (MAPK) pathways in human Müller glial cells. 5 To further elucidate the regulatory mechanisms of IL-1β-induced LGALS1 gene expression, we analysed the promoter region of LGALS1. Given that the 0.5 kbp upstream region from the transcription start site of LGALS1 promoter plays a critical role in LGALS1 gene expression, 11,12 we first generated a luciferase vector driven by LGALS1 promoter (−500 bp to +67 bp, pGal) ( Figure 1A).
However, there was no significant increase in luciferase activity after IL-1β stimulation ( Figure 1B). Previous reports showed that LGALS1 expression in classical Hodgkin's lymphoma was mediated in part by a highly conserved (in humans and rodents) AP-1-dependent LGALS1 enhancer, located in the enhancer region (+450 bp to +1750 bp) 12,13 ( Figure 1A). To determine the role of AP-1-dependent LGALS1 enhancer in Müller glial cells, we next studied whether LGALS1 enhancer is selectively active because of IL-1β stimulation. The construct including the enhancer region (pGal + AP-1) did not change luciferase activity without IL-1β stimulation, as compared to the construct without the enhancer region (pGal); however, the administration of IL-1β significantly increased luciferase activity in the enhancer-containing construct (pGal + AP-1). Additionally, we observed that the construct lacking AP-1 site (+1594 bp to +1600 bp, pGalΔAP-1) exhibited significantly lower luciferase activity ( Figure 1B

| Glucocorticoid-mediated suppression of IL-1βinduced galectin-1/LGALS1 expression with AKT/ AP-1 and ERK/AP-1 activation in Müller glial cells
Glucocorticoid drugs, such as dexamethasone and triamcinolone acetonide, bind to GR, which belongs to the nuclear receptor superfamily of ligand-dependent transcription factors, and exert activation and repression of specific target genes. Next, we hypothesized that glucocorticoids act as a potential suppressors of IL-1β-induced galectin-1/LGALS1 expression. Although application with neither dexamethasone nor triamcinolone acetonide to Müller glial cells affected basal LGALS1 mRNA levels, IL-1β-induced LGALS1 mRNA levels were significantly reduced by these glucocorticoids, but not by the mineralocorticoid aldosterone ( Figure 2A).

Moreover, the suppressive effects of the glucocorticoids on
LGALS1 mRNA expression were cancelled by pre-treatment with the GR antagonist RU486 ( Figure 2B). We also confirmed that the glucocorticoids reduced IL-1β-induced elevation of phosphoryl- ( Figure 3B). Consistently, immunoblotting demonstrated the impact of the glucocorticoids on DUSP1 protein levels as well ( Figure 3B). ChIP-qPCR revealed that GR binding to GRE in the DUSP1 promoter region significantly increased after stimulation with dexamethasone and triamcinolone acetonide ( Figure 3C), in accordance with glucocorticoid-induced GRE reporter activity ( Figure 3A) and DUSP1 expression ( Figure 3B).
To confirm the involvement of DUSP1 in the suppression of IL-1βinduced galectin-1/LGALS1 expression, DUSP1 mRNA was knocked down using siRNA in Müller glial cells. Preliminary results confirmed the potent inhibition of glucocorticoid-activated DUSP1 mRNA and protein expression levels by siRNA ( Figure S1A). Interestingly, the siRNA-based depletion of DUSP1 mRNA reversed glucocorticoid-mediated down-regulation of LGALS1 transcript ( Figure 3D).
Supporting the gene expression data, immunoblotting also showed that DUSP1 knockdown cancelled glucocorticoid-mediated suppression of galectin-1 production with AKT and ERK1/2 phosphorylation in Müller glial cells ( Figure 3D).

| D ISCUSS I ON
The present study is the first to show several important data on GR- Alternatively, activated GR suppresses inflammatory gene transcription by interacting with transcription factors (ie direct protein-protein binding) in the regulation of transcriptional response to glucocorticoids, which is referred to as transrepression. 19 In consistence with our ChIP-qPCR data (Figure 4), GR can be tethered via DNA-bound transcription factors to the DNA sequences (eg AP-1 site) that do not contain GR's recognition site. 35 This regulatory mechanism of action involves the interaction between activated GR and inflammation-related transcription factors including AP-1, NF-κB and STAT (signal transduction and transcription proteins), 20 allowing GR to down-regulate the expression of their downstream molecules such as metalloproteinases. 36 Glucocorticoids are known to rapidly increase the transcription of DUSP1 in lung adenocarcinoma cells, 37 consistent with our data on Müller glial cells ( Figure 4); however, it was additionally found that glucocorticoids suppressed IL-1β-induced LGALS1 gene expression even before phosphorylated AKT and ERK1/2 protein levels did not decrease (ie DUSP1 protein did not increase). During the rapid phase, it was theorized from these results that ligand-activated GR interacted with phosphorylated AP-1 bound with AP-1 site in the LGALS1 gene, leading to DUSP1-independent repression of LGALS1 transcription in Müller glial cells.
Once the DUSP1 transactivation process is completed, however, this transrepression pathway is likely to be ceased from DUSP1-mediated lack of phosphorylated AP-1. This is consistent with our current data that glucocorticoid-mediated suppression of galectin-1 at 24 hours was almost completely dependent on DUSP-1 induced in Müller glial cells (Figure 3). Similarly, our in vivo results on retinal galectin-1 at 24 hours after glucocorticoid application were thought to stem from DUSP1 transactivation and subsequent dephosphorylation of AKT, ERK1/2 and AP-1 signalling molecules ( Figure 5). Transrepression would therefore be regarded as an anti-inflammatory action more rapidly exerted until transactivation is instead established. This is supported by our frequent observations in clinical practice that diabetic macular oedema resolves within a few hours after glucocorticoid injection to the eye. 38,39 The much faster effect of glucocorticoid drugs than anti-VEGF agents is also attributable in part to GR-unrelated non-genomic mechanisms (ie non-specific interactions of glucocorticoids with cellular membranes). 20 Nevertheless, GR-mediated transrepression is theorized to be faster in efficacy than transactivation by at least several hours of duration needed for acquiring the MAPK phosphatase activity, which was actually true of glucocorticoid-induced down-regulation of galectin-1 in Müller glial cells (Figures 3 and 4). Our present data demonstrated the inhibitory effects of glucocorticoids on retinal glial galectin-1 expression via DUSP1-dependent and -independent deactivation of AP-1 signalling (ie GR-mediated transactivation and transrepression, respectively), highlighting therapeutic implications for the management of DR, especially if patients suffer from vision-threatening macular oedema.

ACK N OWLED G EM ENTS
We thank Shiho Yoshida and Miyuki Murata (Hokkaido University) for their skilled technical assistance. This work was supported in part by Bayer Yakuhin Ltd, Bayer AG, the Uehara Memorial Foundation, the Japan National Society for The Prevention of Blindness and MEXT KAKENHI Grant Number 19K09944 (to AK).

CO N FLI C T O F I NTE R E S T
The authors declare no competing financial interests.

AUTH O R CO NTR I B UTI O N
AK designed research; IH, AK and KN performed the experiments; IH and AK analysed the data; AK and SI wrote the paper; and all authors approved the final version submitted for publication.

DATA AVA I L A B I L I T Y S TAT E M E N T
All data generated or analysed during this study are included in this published article and its supplementary information files.