RasV12 Expression in Microglia Promotes Retinal Inammatory Circuits and Impairment of Photoreceptor

Background Microglial activation is commonly observed in neurodegenerative diseases. However, its role in the complex inammatory network remains unclear. In this study, we generated a microglial activation mouse model by inducing the expression of a constitutively active form of Ras in microglia. Methods The double transgenic lines CAG-LSL-RasV12-IRES-EGFP; Cx3cr1 CreER (Cx3cr1-RasV12 mice) and CAG-LSL-EGFP;Cx3cr1 CreER (control mice) were generated. The expression of RasV12 was induced in microglia by tamoxifen administration. Effects of the expression of RasV12 in the retina were examined by immunohistochemistry of frozen sections, RT-qPCR, and live imaging. Results RasV12 expression in retinal microglial cells promoted cell proliferation, cytokine expression, and phagocytosis. RasV12-expressing microglia migrated towards the apical and basal regions of the retina. Examination of GFAP expression revealed activation of Müller glia in the retina. We also observed loss of the photoreceptors in the outer nuclear layer (ONL) in close proximity to microglial cells. However, no neurodegeneration was observed in the inner nuclear layer (INL) or ganglion cell layer (GCL). Furthermore, we found that RasV12-expressing microglia in the ONL were smaller in size and engulfed photoreceptors. In contrast, the morphology of RasV12-expressing microglia in the GCL and INL resembled resting microglia. In conclusion, RasV12-induced microglial activation impaired rod photoreceptor survival in the ONL, but not in other regions of the retina. The expression of RasV12 is sucient to activate microglia, and our results suggested that microenvironment cues may modulate the microglial phenotypic features and effects of microglial activation.

onset and progression of retinal degeneration, by exerting neurotoxic effects upon activation. In the rd10 mouse model, microglial activation promoted photoreceptor apoptosis in an IL-1β-dependent manner, aggravating retinal degeneration [6]. Consistently, administration of minocycline ameliorated retinal degeneration in the rd10 mouse model [7]. However, the elucidation of the role of microglia in photoreceptor degeneration remains challenging due to the complexity of the retinal structure and immunological circuits involved in retinal diseases. Rod photoreceptors secrete alarmins, such as EDN2 [8]. Alarmins and other cytokines activate Müller glial cells (MGCs), which upon activation, secrete various cytokines, including leukemia inhibitory factor (LIF) [9,10]. Furthermore, the retinal pigment epithelium (RPE) is often involved in immune responses under pathological conditions [11], further increasing the complexity of the mechanisms underlying retinal in ammation.
Aberrant Ras activation promotes oncogenesis, and Ras mutations have been identi ed in a wide variety of human cancers [12]. Substitution of G12 with V in H-Ras (H-RasV12) results in a constitutively active form, which was rst described in epithelial cancer. The serine/threonine kinase BRAF is one of the oncoproteins that are activated by Ras. A recent study showed that microglial expression of a constitutively active form of BRAF, BRAFV600E, promoted microglia expansion and late-onset neurodegeneration in the brain [16].
The aim of this study was to investigate the effects of microglial RasV12 expression in the retina. We found that RasV12 expression promoted microglial cell proliferation and migration to all regions of the retina, accompanied by aberrant MGC activation. We also found that microglial cell migration to the ONL was accompanied by photoreceptor degeneration.

Methods
All experiments adhere with the declaration of Helsinki. All animal experiments were approved by the Animal Care Committee of the Institute of Medical Science, University of Tokyo and conducted in accordance with the Guidelines laid down by the NIH in the US regarding the care and use of animals for experimental procedures and the ARVO (Association for Research in Vision and Ophthalmology) statement for the use of animals in ophthalmic and vision research.
Hematoxylin and Eosin (HE) staining Brain and retina were isolated after perfusion xation using 4% PFA in PBS and xed overnight in the same solution at 4 °C. The samples were embedded into para n, and sectioned (4 µm). After depara nization, the sections were stained with Hematoxylin solution (FUJIFILM Wako Pure Chemical) for 10 min and washed by water for 30 min and stained with Eosin (MUTO PURE CHEMICALS). The sections were washed by ethanol for 20 min and then xylene for 15 min and mounted using Malinol (MUTO PURE CHEMICALS).

Immunohistochemistry
Immunohistochemistry of frozen sectioned retinas was performed as previously described [20]. Brain was isolated after perfusion xation using 4% paraformaldehyde (PFA) and xed overnight in 4 °C in 4% PFA.

Ex vivo confocal time-lapse imaging
Isolated retinas of Cx3cr1-RasV12 mice were placed on Millicell chamber lters (Millipore) and stained with 0.01 mg/ml of Hoechst33342 (DOJINDO LABORATORIES) for 10 min at room temperature. Then, the retinas were embedded in 1% agarose (NIPPON GENE) in PBS at 37 °C, and after agarose has the property of being solid, the lters were carefully removed from embedded retinas, and then the samples were served to the time-lapse imaging. The images were taken at 1 min interval, 90 times by LSM710 confocal microscope (Carl Zeiss). Images were edited using ZEN 2009 and ZEN 2.3 lite software (Carl Zeiss).

Flow cytometry
Isolated retains were dissociated to single cell level as described previously [20]. Peripheral blood were collected from tail vein, and the erythrocytes were lysed using ACK lysing buffer (Thermo Fisher Scienti c). Cells were stained with PE-conjugated anti-CD11b antibody (Bio Legend), Alexa Fluor647conjugated anti-CD73 antibody (BD Biosciences), or Annexin V-PE (Bio Vison). Dead cell exclusion was done by using propidium iodide (Sigma-Aldrich). Cells were analyzed by FACS Calibur (BD Biosciences). Data analysis were performed using Flow Jo software (BD Biosciences).

RT-qPCR
Total RNA was puri ed from mouse whole retina using Sepasol (Nacalai Tesque), and cDNA was synthesized using ReverTra Ace qPCR RT Master Mix (TOYOBO). Quantitative PCR (qPCR) was performed using the THUNDERBIRD SYBR qPCR Mix (TOYOBO) and the Roche Light Cycler 96 (Roche Diagnostics). Actb and Sdha were used as reference genes.

Statistical analysis
P-values were calculated by unpaired t-test.

Generation of mice with inducible RasV12 expression in microglia
The double transgenic lines CAG-LSL-RasV12-IRES-EGFP; Cx3cr1 CreER (Cx3cr1-RasV12 mice) and CAG-LSL-EGFP;Cx3cr1 CreER (control mice) were used in this study. The chemokine receptor CX3CR1 is speci cally expressed in monocytes and tissue-resident macrophages, including microglia [21]; hence, Cx3cr1 CreER knock-in mice expressed tamoxifen-inducible Cre recombinase in microglial cells. EGFP was also expressed after tamoxifen administration in both Cx3cr1-RasV12 mice and control mice, allowing for lineage tracing.
Control and Cx3cr1-RasV12 mice were born at the expected Mendelian ratio (Table 1). Tamoxifen was administered to control and Cx3cr1-RasV12 mice at postnatal day 14 (P14). Death of unknown causes was observed in Cx3cr1-RasV12 mice 9-10 days after tamoxifen injection. Since Cx3cr1 is also expressed in circulating monocytes [21,22], we evaluated the presence of monocytes with Cre-mediated recombination in the peripheral blood by ow cytometry (Supplementary Fig. S1a). In Cx3cr1-RasV12 mice, EGFP expression was detected in approximately 1% of CD11b-positive cells ( Supplementary Fig.  S1b). Therefore, we could con dently conclude that EGFP-positive cells in the retina and brain were mostly microglia and that the representation of macrophages in this population is negligible.
Characterization of RasV12-expressing microglia in the retina and brain We then evaluated the presence of RasV12-expressing microglia in the retina and the brain. Retinas harvested at day 7 were frozen-sectioned for histological examination. The number of EGFP-positive cells was considerably higher in retinas from Cx3cr1-RasV12 mice compared with control retinas (Fig. 1a, c).
Moreover, the vast majority of EGFP-expressing cells also expressed Iba1, which is a marker of microglia and macrophages (Fig. 1b). We then examined the time course of EGFP expression at days 1, 3, 5, and 7 after tamoxifen administration. EGFP was expressed from day 3 in the retinas of control and Cx3cr1-RasV12 mice (Fig. 1e, arrowheads). EGFP-positive cells also expressed the microglia-speci c marker TMEM119 (Fig. 1e) [23]. EGFP + TMEM119 + cells accounted for nearly 90% of the total EGFP-positive cell population in control and Cx3cr1-RasV12 retinas at days 3, 5, and 7 after tamoxifen administration (Fig. 1F). The number of EGFP + TMEM119 + cells increased until day 7 in Cx3cr1-RasV12 retinas, whereas only a slight increase was observed in control retinas (Fig. 1e, g). Furthermore, immunohistochemical analysis revealed the presence of phosphorylated ERK1/2 in EGFP-positive microglia in Cx3cr1-RasV12 retinas (Fig. 1h, arrowheads), con rming Ras signaling pathway activation in these microglial cells.
In the brains of Cx3cr1-RasV12 mice, we observed prominent EGFP expression in the ventral orbital cortex (VO), lateral orbital cortex (LO), and ventral tenia tecta (VTT) at day 7 after tamoxifen administration ( Supplementary Fig. Sa). Hematoxylin and eosin staining con rmed strong nuclear staining in these regions of the brain (Supplementary Fig. S2b). Similar to what we observed in the retina, EGFP expression was apparent from day 3 in the orbital cortex region of the brain ( Supplementary Fig. S2c), and EGFPpositive cells also expressed Iba1 ( Supplementary Fig. S2c, arrowheads d, e). The number of EGFP + Iba1 + cells continuously increased in the brain of Cx3cr1-RasV12 mice ( Supplementary Fig. S2c, d). Moreover, ERK1/2 phosphorylation was detected in EGFP-positive cells in the brain of Cx3cr1-RasV12 mice ( Supplementary Fig. S2f, arrowheads).

Migration and proliferation of RasV12-expressing microglia
Resting microglia are often found in the IPL and OPL of the retina [24]. In control retinas, EGFP + TMEM119 + cells were primarily found in the IPL and OPL at day 7 after tamoxifen administration (Fig. 1e, 2a). In contrast, RasV12-expressing microglia were observed in the ganglion cell layer (GCL), ONL, and SR (Fig. 1e, 2a), suggesting microglial cell migration to the apical and basal sides of the retina. At day 7 after tamoxifen administration, retinas from Cx3cr1-RasV12 mice contained a signi cantly higher number of EGFP + TMEM119 + cells compared with control retinas (Fig. 1g). Additionally, retinas from Cx3cr1-RasV12 mice contained a signi cantly higher number of proliferating microglial cells (Ki67 + EGFP + ; Fig. 2b, arrow heads). Ki67 + EGFP + cells comprised nearly 70% of the total EGFP-positive cell population in retinas from Cx3cr1-RasV12 mice (Fig. 2c). Although Ki67 + EGFP + cells were present in all retinal layers, notably high numbers were observed in the apical and basal regions (Fig. 2d).

RasV12 expression in microglia promotes photoreceptor degeneration
Histological analysis indicated the loss of photoreceptors in Cx3cr1-RasV12 retinas (Fig. 3a, arrowheads). Immunohistochemical staining revealed that microglial cells were localized close to the degenerating regions of the ONL (Fig. 3b). Moreover, differential interference contrast (DIC) imaging showed a reduction in the thickness of the outer layer of Cx3cr1-RasV12 retinas (Fig. 3b, c). We also performed Annexin V staining to measure apoptosis [29] and found an increased number of apoptotic cells in Cx3cr1-RasV12 retinas. Importantly, the number of apoptotic cells was higher in CD73-positive rod cells than other retinal cells (Fig. 3d, e). In Cx3cr1-RasV12 retinas, we also observed upregulation of Edn2 and Fgf2 (Fig. 3f), which are commonly induced upon photoreceptor damage [8].
Retinal degeneration is often accompanied by reactive gliosis, which is characterized by strong GFAP expression [30]. Consistently, GFAP was upregulated in Cx3cr1-RasV12 retinas (Fig. 3g, h), and microglia were localized adjacent to the GFAP-positive MGCs (Fig. 3f, enlarged panels). We found no differences in the numbers of RBPMS-positive retinal ganglion cells ( Supplementary Fig. S4a), AP-2α-positive amacrine cells ( Supplementary Fig. S4b), Chx10-positive bipolar cells ( Supplementary Fig. S4c), and Calbindinpositive horizontal cells (Supplementary Fig. S4d) in RasV12-expressing and control retinas, suggesting that survival and maintenance of these cell types was not affected by RasV12 expression in microglia.
RasV12-expressing microglia in the ONL, but not in the INL, mediates rod photoreceptor phagocytosis Confocal microscopy revealed that RasV12-expressing microglia engulfed recoverin-positive rod photoreceptors in the damaged ONL regions (Fig. 4a); hence, we measured phagocytosis related molecules in Cx3cr1-RasV12 retinas. In Cx3cr1-RasV12 mice, EGFP-positive cells strongly expressed CD68 (Fig. 4b, arrow heads), which is a lysosomal protein highly expressed in activated microglia [7,31]. Interestingly, the number of CD68-expressing cells was higher in the ONL and SR than in the inner layers of the retina (Fig. 4c). Furthermore, population of the EGFP + CD68 + TMEM119 + cells in EGFP and Tmem119 double positive cells was considerably higher in the ONL and SR than in other retinal regions (Fig. 4d). Microglial iNOS expression was demonstrated in rd8 mice [28,32], and an optic nerve injury mouse model [33]. Immunohistochemical staining revealed strong iNOS expression in EGFP-positive microglial cells in Cx3cr1-RasV12 retinas (Fig. 4e). EGFP + iNOS + cells were also observed in the inner retinal layers (Fig. 4e, f); nevertheless, this double-positive cell population was higher in the ONL and SR than in other regions of the retina (Fig. 4g). Furthermore, the phagocytosis-related genes C3 and Cd68 were upregulated in Cx3cr1-RasV12 retinas (Fig. 4h).
The behavior of RasV12-expressing microglia differs in the IPL and ONL We next examined the behavior of RasV12-expressing microglia in different regions of the retina. We prepared retinal explants from Cx3cr1-RasV12 mice and labeled the nuclei of rod photoreceptors with Hoechst33342. Time-lapse imaging revealed that RasV12-expressing microglia in the ONL engulfed photoreceptors by extending their pseudopodia. Moreover, multiple nuclei were observed in individual microglial cells, suggesting rod photoreceptor engulfment by microglia (Fig. 5, upper panel, Supporting information movie 1). However, we found no evidence of cells engulfment by RasV12-expressing microglia in the IPL (Fig. 5, lower panel, Supporting information movie 2).

Discussion
In this study, we examined the role of Ras activation in retinal microglia by inducing the expression of constitutively active Ras (RasV12) in microglial cells. RasV12 expression promoted microglial activation, as well as activation of MGCs, which led to the degeneration of retinal photoreceptors. Previous studies highlighted the importance of FGF2, EDN2, LIF, and other in ammatory cytokines in retinal degeneration. Consistently, we found that Gfap, Edn2, and Fgf2 were upregulated in Cx3cr1-RasV12 retinas, suggesting that microglial activation results in cytokine secretion.
Most studies on the molecular mechanisms of microglia involved the use of photoreceptor degeneration models, suggesting that the initial event of in ammatory cascade is photoreceptor injury. Edn2 upregulation has been reported in injured photoreceptors, and EDN2 contributes to the activation of MGCs [9]; however, the mechanisms regulating its transcriptional activation remain elusive. Similarly, transcription of Lif is induced upon tissue damage, and its expression induces activation of MGCs, characterized by GFAP expression. In this study, we found evidence of Edn2 and Lif transcriptional activation in Cx3cr1-RasV12 retinas. Therefore, RasV12 expression in microglia is su cient to induce Edn2 expression in the absence of injury of photoreceptors, and in photoreceptor degeneration mouse models, Edn2 upregulation is likely induced at least partly by microglia. Alarmin is expressed at the initial phase of photoreceptor degeneration, and it may contribute to microglial activation early during degeneration. On the other hand, EDN2 is known to exerts protective effects on photoreceptors. Therefore, microglia may modulate retinal degeneration protectively through up-regulation of EDN2 on rod photoreceptors in certain situation. BRAFV600E mutation has been implicated in late-onset neurodegenerative diseases and histiocytosis [34,35]. Mosaic expression of BRAFV600E in the yolk-sac erythro-myeloid progenitors resulted in clonal expansion of microglia and late-onset neurodegeneration in the brain [16]. Our RasV12 microglial activation model shared some features with this BRAFV600E-expressing model, including the expansion of Iba1-positive cells, GFAP expression, and in ammatory cytokine expression. However, our RasV12expressing microglia mouse model exhibited early-onset microglial activation in the brain and neural degeneration in the retina. Furthermore, our mice did not survive more than nine days after tamoxifen administration. On the other hand, no description about early phenotype of BRAFV600E expressing model was available; therefore, we could not directly compare the ndings acquired from these two animal models. Considering that Ras acts upstream of BRAF and activates not only the MAPK pathway but also multiple other signaling pathways, it is not surprising that our RasV12-expressing microglia model has a more complex phenotype than the BRAFV600E-expressing model.
Interestingly, Ras activation in microglia resulted in cell migration towards the apical and basal regions of the retina. Because ERK1/2 activates various migration-related molecules, such as MLCK, Paxillin, FAK, and Calpain [36], the enhanced migration of RasV12-expressing microglia was anticipated; however, we could not explain the tropism observed in microglia migration patterns. In photoreceptor degeneration models, microglia migrated primarily in degeneration sites; however, the molecules promoting cell migration could not be identi ed. In this study, we found that RasV12-expressing microglia were localized close to MGCs; thus, it is likely that microglial cells migrated along the migration pathway of MGCs. However, no determinant of apical or basal direction is hypothesized currently, and diffused migration is rather plausible.
Nitric oxide has been identi ed as one of the mediators of the cytotoxic effects of activated microglia [37]. Consistently, we observed enhanced expression of iNOS and in ammatory cytokines in RasV12expressing microglia. These in ammatory stimuli may have contributed to rod photoreceptor apoptosis and subsequent phagocytosis by RasV12-expressing microglia. Interestingly, although iNOS upregulation was also observed in RasV12-expressing microglia in the GCL and IPL, we could not detect degeneration of non-photoreceptor cells or phosphatidylserine (PS) expression on CD73-negative cells in Cx3cr1-RasV12 retinas. Moreover, the cellular dynamics of RasV12-expressing microglia were profoundly different in the ONL and IPL. These results suggest that rod photoreceptors are targeted by activated microglia.

Conclusions
In summary, our data suggested that activation of Ras signaling in the microglia induced not only microglial proliferation, changing distribution, and increasing phagocytotic activity but also photoreceptor degeneration. Furthermore, cellular dynamics of RasV12-expressing microglia were different between in ONL and IPL. Therefore, constitutive microglial activation induced retinal degeneration and microenvironment cues may modulate the microglial phenotypic features and effects of microglial activation. Future studies are required to examine the role of the immunosuppressive microenvironment of the inner side of the retina. Availability of data and materials; The datasets used and/or analysed during the current study are available from the corresponding authors on reasonable request.