Deletion of Tgfβ signal in activated microglia prolongs hypoxia‐induced retinal neovascularization enhancing Igf1 expression and retinal leukostasis

Abstract Retinal neovascularization (NV) is the major cause of severe visual impairment in patients with ischemic eye diseases. While it is known that retinal microglia contribute to both physiological and pathological angiogenesis, the molecular mechanisms by which these glia regulate pathological NV have not been fully elucidated. In this study, we utilized a retinal microglia‐specific Transforming Growth Factor‐β (Tgfβ) receptor knock out mouse model and human iPSC‐derived microglia to examine the role of Tgfβ signaling in activated microglia during retinal NV. Using a tamoxifen‐inducible, microglia‐specific Tgfβ receptor type 2 (Tgfβr2) knockout mouse [Tgfβr2 KO (ΔMG)] we show that Tgfβ signaling in microglia actively represses leukostasis in retinal vessels. Furthermore, we show that Tgfβ signaling represses expression of the pro‐angiogenic factor, Insulin‐like growth factor 1 (Igf1), independent of Vegf regulation. Using the mouse model of oxygen‐induced retinopathy (OIR) we show that Tgfβ signaling in activated microglia plays a role in hypoxia‐induced NV where a loss in Tgfβ signaling microglia exacerbates and prolongs retinal NV in OIR. Using human iPSC‐derived microglia cells in an in vitro assay, we validate the role of Transforming Growth Factor‐β1 (Tgfβ1) in regulating Igf1 expression in hypoxic conditions. Finally, we show that Tgfβ signaling in microglia is essential for microglial homeostasis and that the disruption of Tgfβ signaling in microglia exacerbates retinal NV in OIR by promoting leukostasis and Igf1 expression.


| INTRODUCTION
Retinal neovascularization (NV) is the major cause of severe visual impairment in patients with ischemic or inflammatory ocular diseases such as diabetic retinopathy, retinal vein occlusion, uveitis and retinopathy of prematurity (Usui et al., 2015). Vascular Endothelial Growth Factor (VEGF) plays a pivotal role in the development of pathological NV; drugs that inhibit this pro-angiogenic cytokine have been widely used to treat retinal neovascular diseases (Fogli et al., 2018;Witmer et al., 2003). While intravitreal anti-VEGF injection represents a major breakthrough for the treatment of retinal neovascular diseases, not all patients respond to anti-VEGF agents (Ashraf et al., 2016;Ip et al., 2015; Writing Committee for the Diabetic Retinopathy Clinical Research et al., 2015). Moreover, there are safety issues associated with repeated intravitreal administration of anti-VEGF agents, especially for at-risk patients with diabetes, cardio-and cerebrovascular diseases, or premature babies who are vulnerable to modulation of crucial trophic factors (Falavarjani & Nguyen, 2013;Usui-Ouchi & Friedlander, 2019). Understanding VEGF-independent mechanisms of retinal NV and their role in retinal neovascular disease is critical for developing additional complimentary or alternative therapeutic strategies.
Microglia are the resident immune cells in the retina localized to the outer and inner plexiform layers and superficial plexus, engaging in surveillance and maintenance of retinal synapses (Lee et al., 2008;Wang et al., 2016). Under pathological conditions such as retinal degeneration, neovascularization and aging, microglia are activated and migrate into the affected sites where they respond to inflammation by up-regulationg phagocytic activity and expression of inflammatory cytokines (Ma et al., 2009;Usui-Ouchi et al., 2020;Zhao et al., 2015). Activated and mis-localized retinal microglia are a common hallmark of various retinal degenerative, inflammatory and angiogenic diseases (Silverman & Wong, 2018). In ischemic retinopathy, activated microglia are found in the central avascular zone prior to neovascularization (Fischer et al., 2011;Vessey et al., 2011). In the OIR mouse model of ischemic retinopathy, ablation of microglia can rescue NV (Kubota et al., 2009) suggesting a key role in pathological NV. However, their role appears to be independent of VEGF activation (Boeck et al., 2020). The mechanism by which microglia are activated and promote NV is not understood.
Tgfβ1 has previously been shown to be a potent immunoregulatory factor for microglia in vivo and in vitro where the loss of Tgfβ signaling results in the increase of microglia activation (Brionne et al., 2003;Butovsky et al., 2014;Ma et al., 2019;Spittau et al., 2013;Zoller et al., 2018). Tgfβ-signaling is propagated by binding of Tgfβ to Tgfβ receptor type 2 (Trfbr2) that phosphorylates the Tgfβ receptor type 1 (Tgfbr1) (Wrana et al., 1994;Yamashita et al., 1994). Pan-ocular deletion of Tgfβ signaling can also cause common changes observed in proliferative diabetic retinopathy including pericyte loss, microaneurysms, leaky capillaries, and retinal hemorrhages (Braunger et al., 2015). Targeted ablation of Tgfbr2 in retinal microglia promotes activation causing a neuroinflammatory response and choroidal NV (W. Ma et al., 2019). However, the role of Tgfβ signaling in microglia on retinal NV and the mechanisms that regulate microglia function during ischemia are not understood.
Here, we investigate the role of Tgfβ signaling in the activation of microglia and demonstrate that it can induce VEGF-independent NV pathways through regulation of Igf1. Using the OIR mouse model of ischemic retinopathy, targeted ablation of Tgfbr2 in microglia, and human iPSC derived microglia we demonstrate that hypoxia in the retina regulates microglial activation, the expression of chemoattractant chemokines, leukostasis, and Igf1 dependent NV through the repression of Tgfβ signaling in microglia. These results provide insight into the mechanism of microglial activation under ischemic retina and the role they play in the formation of pathological NV.

| Mice and animal experimental procedures
All animal protocols were approved by the IACUC committee at The Scripps Research Institute, La Jolla, California. All animals received food and water ad libitum. C57BL6 mice and Balb/c mice were obtained from The Scripps Research Institute animal facility. Chemokine (C-X3-C motif) receptor 1 (Cx3cr1) CreÀERT mice expressing tamoxifen-inducible Cre recombinase (The Jackson Laboratory, #021160) (Parkhurst et al., 2013) were crossed with mice possessing loxP sites that flank exon 4 of the Tgfbr2 (Tgfbr2 flox/flox , The Jackson Laboratory, #012603) (Leveen et al., 2002) to generate Cx3cr1 CreÀERT ; Tgfbr2 flox/flox . To induce Cx3cr1-Cre recombination, 100 μg of tamoxifen (Sigma-Aldrich, T5648)/ cone oil solution was administered to Cx3cr1 CreÀERT ; Tgfbr2 flox/flox and control littermates, Tgfbr2 flox/flox (Control) subcutaneously once a day from P9 to P14 and for OIR P4 to P6 and P12 to P14 to avoid the oxygen level fluctuation in chamber from P7 to P12. Oxygen-induced retinopathy (OIR) was induced as previously described (Murinello et al., 2019;Smith et al., 1994). Postnatal day 7 (P7) pups and their mothers were exposed to 75% oxygen in a hyperoxia chamber (BioSpherix ProOx P110) for 5 days and returned to room air at P12.
Mice were euthanized by cervical dislocation at varying time points, as indicated in the results and figure legends.

| Cell and cell culture
The human induced pluripotent stem cell (hiPSC) line used was derived from peripheral blood mononuclear cells from a female.
Reprogramming was performed by the Harvard iPS core facility using sendai virus for reprogramming factor delivery. All cell lines were obtained with verified normal karyotype and contamination-free. hiPSC were maintained on Matrigel (BD Biosciences) coated plates with mTeSR1 medium (STEMCELL Technologies). Cells were passaged every 3-4 days at approximately 80% confluence. Colonies containing clearly visible differentiated cells were marked and mechanically removed before passaging. Microglia precursors were generated as previously described (Haenseler et al., 2017;van Wilgenburg et al., 2013). The embryoid bodies (EBs) are formed using Aggrewells (STEMCELL Technologies), cultured with bone morphogenetic protein 4 (BMP4), vascular endothelial growth factor (VEGF), and stem cell factor (SCF), then plated into T175 flasks with Interleukin-3 and macrophage colony-stimulating factor (M-CSF). After 4 weeks, microglia precursors emerged into the supernatant. It was previously revealed that their ontogeny is MYB-independent primitive myeloid cells, which is same ontogeny as microglia . hiPS derived Microglia precursors (pMG) were plated into 12 well plates containing X-VIVO15 with 100 ng/ml M-CSF, 2 mM Glutamax, 100 U/ml penicillin, and 100 μg/ml streptomycin for further in vitro assays. The cells were stimulated with human recombinant Tgfβ1 (Peprotech, 100-21), 10 μM of SB525334 (Selleckchem, S1476), 200 μM of DMOG (Millipore sigma, D3695), then cell culture supernatant and cells were stored at À80 C for following qPCR and ELISA assays.

| Immunohistochemistry of whole-mount retinas
Enucleated eyes were placed in 4% paraformaldehyde (PFA) for 1 h. After fixation, the cornea, the lenses, the sclera, choroid, and the vitreous were removed and the retinas were laid flat with four radial relaxing incisions. Retinas were incubated in blocking buffer (PBS with 10% fetal bovine serum, 10% normal goat serum, and 0.2% Triton X-100) for 2 h at 4 C, following by an overnight incubation with primary antibodies in blocking

| Retinal microglia isolation by flow cytometry
A postnatal neural dissociation kit (Miltenyi, 130-092-628) was used to prepare a single cell suspension from mouse retinas. Cells were centrifuged at 150g for 5 min at 4 C. The digested tissue was resuspended in 100 μl of 4% FBS in PBS containing an FITC antibody to CD11b (1:100; BioLegend, 101206) and PE antibody to Gr-1 (1:100; BD Biosciences, 553128) and incubated for 20 min on ice.
We used clone RB6-8C5 for Gr-1 antibody because it reacts with a common epitope on Ly6-G and Ly6-C to eliminate blood born monocytes and granulocytes. We did not use CD45 antibody to detect CD45 low fraction as microglia population because CD45 expression in Tgfbr2-ablated microglia is upregulated transforming to activated form as previously shown (Ma et al., 2019). Labeled retinal microglia

| RNA isolation and real-time PCR
For whole retina and culture cells, single retinas were collected in 500 μl of Trizol and total RNA was isolated using a PureLink RNA Mini Kit (Thermo Fisher Scientific) according to manufacturer's instructions.
Seven hundred and fifty nanograms of RNA was used for RT-qPCR using a high-capacity cDNA reverse transcription kit (Thermo Fisher Scientific).
For flow-sorted cells, total RNA was isolated from sorted cells using the RNeasy Micro Kit (QIAGEN) and reverse transcribed using Maxima First Strand cDNA Synthesis Kit for RT-qPCR (Thermo Scientific). qPCR was performed using Power-up SYBR™ Green PCR Master Mix (Thermo Fisher Scientific) and primers on a Quantstudio 5 Real-Time PCR System (Thermo Fisher Scientific). β-actin (Actb) was used as the reference gene for all experiments. Levels of mRNA expression were normalized to those in controls as determined using the comparative CT (ΔΔCT) method. Primer sequences are listed in Table S1.

| Enzyme-linked immunosorbent assay (ELISA)
Forty-eight hours after Tgfβ1 supplementation to hiPS derived pMG, cell culture supernatants were assayed for ELISA assay to detect the protein level of IGF1 using the Human IGF-1 Quantikine ELISA kit (R&D systems) according to the manufacturer's protocol.

| Lectin labelling of adherent retinal leukocytes
The retinal vasculature and adherent leukocytes were imaged by perfusion labeling with TRITC-conjugated Concanavalin A (Con A) lectin (Vector Laboratories), as described previously (Joussen et al., 2001;Okunuki et al., 2019). Briefly, after deep anesthesia, the chest cavity was opened and a 27-gauge cannula was inserted into the left ventricle. Mice were then perfused through the left ventricle first using 5 ml of PBS, followed by fixation with 1% PFA (5 ml), 5 ml of TRITCconjugated Con A (20 μg/ml in PBS), and 5 ml of PBS. The eyes were then fixed in 4% PFA for an hour, and the retinas were flat-mounted.
The total number of TRITC positive adherent leukocytes in the retinal vessels was counted.

| Quantification and statistical analysis
For OIR, the percentage of the area of NV and vaso-obliteration (VO) in OIR retinas was automatically quantified using deep learning segmentation software available at http://oirseg.org (Xiao et al., 2017). All statistical tests were performed in GraphPad Prism v8 (GraphPad Software, Inc). Data comparisons between two groups were performed using unpaired two-tailed Student t-tests.
Data comparisons between multiple groups were performed with one-way ANOVA with Tukey's correction. Statistical tests used for each experiment are specified in the figure legends. Data are represented as mean ± SEM. A p value of p < .05 was considered significant.

| Study approval
All animal protocols were approved by the IACUC committee at The Scripps Research Institute, La Jolla, California, and all federal animal experimentation guidelines were adhered to.

| Tgfβbr2 deficient microglia transform to activated status
Tgfβ1 has previously been shown to be a potent immunoregulatory factor for microglia in vivo and in vitro where the loss of Tgfβ signaling results in the increase of microglia activation (Brionne et al., 2003;Butovsky et al., 2014;Ma et al., 2019;Spittau et al., 2013;Zoller et al., 2018). To investigate the role of Tgfβ signaling in microglia activation and NV we used microglia specific  3.4 | The expression of Tgfβ receptors was downregulated and the expression of Igf1 was upregulated specifically in hypoxic microglia of wild type C57BL6 OIR Numerous studies have shown that Igf1 is associated with pathological NV in proliferative diabetic retinopathy or retinopathy of prematurity (Boulton et al., 1997;Haurigot et al., 2009;Hellstrom et al., 2001;Kondo et al., 2003;Meyer-Schwickerath et al., 1993;Ruberte et al., 2004;Smith et al., 1999;Wilkinson-Berka et al., 2006). Since we have found that Tgfβ signaling regulates Igf1 expression in microglia, we next examined the potential regulation of Igf1 in  were also expressed in pMG ( Figure S1c).
We next validated the suppression of Igf1 by Tgfβ signaling in human pMGs using Tgfβ ligand and Tgfβ signaling inhibitors. Twenty four hours following the addition of human recombinant Tgfβ1 to pMG culture media, the expression of Igf1 mRNA in pMG was significantly suppressed as expected (Figure 6a). We subsequently confirmed that IGF1 protein secreted into media was also suppressed by human recombinant Tgfβ1 (Figure 6b). Conversely, addition of the Tgfβbr2 inhibitor, SB525334, to pMGs increased Igf1 expression and was sufficient to rescue Igf1 suppression in the presence of Tgfβ1

Control OIR Tgfbr2 KO (ΔMG) OIR
Temporary NV Legend on next page.
following 24 h of DMOG treatment and only showed a significant increase in expression at 48 h indicating that it is not directly induced by HIF-1α ( Figure 6e). Next, we looked at the interaction between hypoxia and Tgfβ signaling on Igf1 regulation and found that human recombinant Tgfβ1 rescued hypoxia induced Igf1 upregulation ( Figure 6g), but had no effect on hypoxia induced Vegfa upregulation (Figure 6h).

| DISCUSSION
We have demonstrated that Tgfβ signal in microglia played a role in regulating microglia homeostasis; inhibiting it resulted in exacerbating pathological NV through the upregulation of Igf1 in a mouse model of ischemic retinopathy.
Microglia are resident immune cells in the central nervous system.
Under healthy conditions, retinal microglia are mainly localized in the outer and inner plexiform layer and superficial plexus where they engage in surveillance and maintenance of retinal synapses (Lee et al., 2008;Wang et al., 2016). However, under pathological conditions such as retinal degeneration, neovascularization and aging, microglia activate and migrate into the sites where those pathological changes occur. They also localize to damaged photoreceptors, RPE, and retinal or choroidal NV (Ma et al., 2009;Usui-Ouchi et al., 2020;Zhao et al., 2015). Given this association between activated/primed microglia and neurovasculodegenerative diseases, it is necessary to better understand activation mechanisms in microglia.
Tgfβ signaling has pleiotropic effects in various tissues on cell survival and inflammation (Travis & Sheppard, 2014). In mammals, the Tgfβ family consists of three members, Tgfβ1, Tgfβ2, and Tgfβ3, and all isoforms can be detected in various types of ocular cells including retinal neurons, retinal pigment epithelium, blood vessels, and microglia (Tosi et al., 2018). Tgfβ receptors are also broadly expressed in different retinal cell types, specifically, in retinal microglia and endothelial cells (Ma et al., 2019;Obata et al., 1999). Deletion of panocular Tgfβ signaling leads to proliferative diabetic retinopathic changes such as pericyte loss, the formation of abundant microaneurysms, leaky capillaries, and retinal hemorrhages (Braunger et al., 2015). Tgfβ1 has previously been described as a potent immunoregulatory factor for cerebral microglia in vivo and in vitro (Brionne et al., 2003;Makwana et al., 2007;Spittau et al., 2013;Zoller et al., 2018 We also report that Tgfbr2 ablation in microglia exacerbated pathological NV in OIR. We found two remarkable changes that could exacerbate pathological NV: (1) an increase in retinal leukostasis in retinal capillaries through production of chemoattractant factors; and (2) an increase in Igf1 expression in microglia. Retinal leukostasis indicates an increase in leukocyte recruitment and adhesion to the retinal capillary endothelium. Retinal leukostasis can lead to blood-retinal barrier breakdown, capillary occlusion, and amplification of the inflammatory response in various retinal diseases such as diabetic retinopathy, ischemic retinopathy, and uveitis (Eshaq et al., 2017;Tarr et al., 2013;Tsujikawa & Ogura, 2012). Dysregulated microglia caused by Tgfbr2 ablation were closely adherent to retinal vessels and produced chemoattractant factors such as Ccl2 and Ccl8, chemokines that can lead to the recruitment of circulating monocytes, promoting pathological NV.
For example, IGF1 is required for maximal VEGF-dependent NV via the IGF1 receptor and MAPK activation in OIR, and Igf1 knockout mice had impaired retinal vascular growth despite normal VEGF level (Smith et al., 1999). Jacobo et al demonstrated that IGF1 stabilizes endothelial cell tubes and retinal neovessels that form in response to VEGF, mediating prolonged activation of Erk, which antagonizes lysophosphatidic acid (LPA)-driven regression (Jacobo & Kazlauskas, 2015). Endothelial cell specific IGF1 receptor KO mice show reduced retinal NV in OIR (Kondo et al., 2003). On the other hand, overexpression of IGF1 in the retina results in changes similar to those of diabetic retinopathy (pericyte loss, capillary basement membrane thickness, inner retinal microaneurysms, and neovessels) (Ruberte et al., 2004). Patients with proliferative diabetic retinopathy have increased vitreous levels of IGF1 (Boulton et al., 1997;Grant et al., 1986;Meyer-Schwickerath et al., 1993). Thus, although IGF1 plays a key role in pathological NV in retina, the origin of IGF1 and interaction between endothelial cell and microglia in the retinal microenvironment has not been well defined. Our results support the idea that IGF1 in diabetic retinopathy or retinopathy of prematurity may be derived from retinal microglia. We further suggest that abundant IGF1 produced by activated microglia stabilizes VEGF-driven retinal neovessels resulting in the exacerbation and prolonged pathological NV in Tgfβr2 KO (ΔMG) OIR.
We have also shown that Tgfβ1-regulated Igf1 expression using hiPS derived microglia cells and Tgfβ1 rescued Igf1 upregulation under chemically induced hypoxia. We surmised Igf1 upregulation in hypoxic microglia induces retinal NVs synergistically with hypoxia induced Vegfa upregulation. Igf1 upregulation after hypoxia occurred at alter time than Vegfa upregulation, suggesting that Igf1 released by hypoxic microglia plays an important role in stabilizing and prolonging pathological NV induced by Vegfa under hypoxia.
Collectively, these results demonstrate that Tgfβ signaling in retinal microglia is critical for maintaining their homeostatic function and regulation of their hypoxic response in ischemic retinopathy.
Targeting Tgfβ signaling in microglia may be a potential therapeutic target to treat pathological NVs in ischemic retinopathy.