**4. Discussion**

Here, we show that activated microglia are a major component of the ERMs from RD and PDR patient samples. Microglial cells are involved in neurotrophic and neurodegenerative disease mechanisms in the retina under stress conditions. In mature retina, microglia reside in the inner and outer plexiform layers and exhibit an abundantly ramified morphology spanning the complete nuclear layers with their long protrusions [21]. Their most important role is to maintain a constant active surveillance of retinal homeostasis where they are indispensable for the immune response and synaptic

pruning and transmission [22,23]. Under injury or ischemic stress, they ge<sup>t</sup> activated, proliferate and migrate to the site of damage, while releasing pro-inflammatory cytokines and ROS to counter the damage [24]. We recently demonstrated the role of microglial mediated aberrant complement activation in the pathogenesis of DR [25]. That study showed that a gradual increase in the expression of CFH (inhibitor of alternate complement pathway) and CD11b (activated microglial marker) in the retina (at an early stage) culminated in epiretinal membranes changes in DR patients further supporting for a major role of microglia and the alternative complement pathway in disease progression.

While there were not many di fferences in the cell types present in the membranes from di fferent pathologies, higher number of GFAP positive cells in PDR and RD confirmed the presence of astrocytes and Müller glia and the glial activation (reactive gliosis). Reactive gliosis is known to cause inflammation, which leads to neurodegeneration [26]. The presence of GFAP positive cells could be due to phagocytosis and gliotic responses of Müller glia in the retina. Astrocytes are predominantly localized to the subretinal space near the ILM, where they activate and proliferate during vascular injury and neovascularization. Along with endothelial cells and pericytes, they are also involved in the formation of epiretinal vessels under ischemic condition [27,28]. Besides Müller glia and astrocytes, a large number of activated microglia was also detected in these membranes (RD and PDR). The presence of these cells indicates their role in ERM formation besides underlying gliosis. Such mechanisms may underlie the extensive gliotic changes, which pose surgical challenges.

Activated microglia secrete di fferent types of cytokines [29]. Underlying gliosis causes the induction of inflammation, which in turn promotes oxidative stress. This is supported by our results of immunohistochemistry using anti-OXR1, an oxidative stress responsive gene. The fibrocellular membranes from RD and PDR showed significantly higher number of anti-OXR1 positive cells as compared to MH. Higher expression of CD11b along with significant increase of oxidative stress markers (*HIF1-* α and *Nrf2*), inflammatory markers (*IL1-*β and *MMP9*) and vascular endothelial growth factor (VEGF) in RD and PDR further suggested oxidative stress activates the resident microglia in the retina and adopt inflammatory phenotype. The activated ramified microglia migrate and secrete inflammatory cytokines and growth factors that promote cellular proliferation, neuronal apoptosis and phagocytosis. Inflammatory cytokines could serve as chemoattractants for invading macrophages. Higher expression of VEGF increases the endothelial cell permeability that would leads to further immune cell recruitment and disease manifestation. A study on histopathological analysis of eyes from patients with non-PDR and PDR showed increased numbers of hypertrophic microglia which correlated with disease severity [30].

Activated microglia also express activated C3a and C3b receptors on cell surface and in turn, induce astrocytes to adopt a reactive and proinflammatory phenotype. Such phenomena compromise the ability of neurons to form synapses and phagocytose dead neurons in the central nervous system. Increased deposition of dead neurons could further induce neuroinflammation and neovascularization [31]. We therefore, propose that activated microglia and astrocytes in the ERMs from RD and PDR could result from the proinflammatory phenotype of activated astrocytes in the retina. Further studies are needed to test this hypothesis.

In retinal pigment epithelium cells, Notch2 has been shown to be the major Notch receptor, and its inhibition significantly reduces the intracellular ROS production and cellular apoptosis upon ultraviolet B-induced damage. On the other hand, hypoxic stress to retinal ganglion cells induces Notch1 expression and activation. A study by Jiao et al. 2018, examined the e ffects of *NOX4* mediated Notch signaling under hyperglycemic (HG) stress. The study observed that HG upregulates Nox4 expression via the activation of Notch signaling, resulting in increased ROS production and cell death in HRECs and thus the inhibition of Notch signaling or Nox4 expression was suggested to be potential therapeutic strategy for the treatment of DR [32].

Several theories have been suggested based on cellular evidences for the processes involved in the membrane formation. Some studies reported that PVR exposes the retina to blood derived fibroblast and other cytokines that causes RPE cells to undergo epithelial to mesenchymal transition and forming proliferative spindle like cells that initiate retinal remodelling and membrane formation over the ILM. Epithelial to Mesenchymal Transition (EMT) facilitates the adoption of a mesenchymal phenotype by an epithelial cell by undergoing multiple biochemical changes. The process of EMT has been well-documented in di fferent types of cancers, leading to tumor progression and increased tumor invasiveness. Besides the other know mechanisms, release of inflammatory cytokines by immune cells has also been shown to cause EMT. Recently, a link between MMP9 and NOTCH1 signalling was also demonstrated [33]. Overexpression of MMP9 acts as a strong activator of NOTCH1 signalling. An earlier study by our group demonstrated increased expression of MMP 9 in PDR vitreous humor that is contributed by activated microglia. The present study also demonstrated an increased expression of CDllb, MMP9, IL-1β and Notch in the membranes as seen by heatmap and co expression using STRING analysis. All together these observations definitely provide a link between increased activation of microglia with RPE undergoing EMT thereby leading to membrane formation as seen in PDR and RD cases by activating Notch 1 signaling and thus inhibiting Notch 1/MMP9 signaling can be a potential therapy to prevent DR severity.

Studies have shown that ERMs impair oxygen and nutrient supply to the retina under ischemic condition [34]. It may also cause tractional retinal detachment by pulling on the neurosensory retina. Our study demonstrated higher level of oxidative stress along with inflammation in RD. Hence, the removal of ERMs may help in improving the oxygenation, nutrient supply and ion transport for retinal potassium bu ffering [35]. Interestingly, the surgical treatment ILM peeling is found to significantly depend on the type of ERM adherence and area where ERM attaches to the retina. ILM peeling causes mechanical trauma to the retinal nerve fiber layer (RNFL) [36], making this procedure controversial [37].

Our study shows overexpression of oxidative stress-related pathways genes in the fibrocellular membranes (Figure 8). We sugges<sup>t</sup> that modulating the levels of oxidative stress and inflammation in PDR and RD could assist in predicting the need of simultaneous ERM/ILM peeling and may be helpful in improving surgical outcomes.

**Figure 8.** Schematic representation of proposed microglial activity in the epiretinal membrane of the retina. In healthy retina, resident microglia are involved in immune surveillance of all the layers of the retina and maintain retinal homeostatis by synaptic pruning, regulation of neurogenesis and axonal growth. With their phenotypic change of shape, they phagocytose cellular debris. Di fferent kinds of stress/insults lead to abnormal functioning of retinal neurons, microglia, astrocyte, Müller glia and RPE cells. The resident/resting microglia in the retina ge<sup>t</sup> activated by transforming into an amoeboid shape and further interact with other neighboring retinal cells, causing abnormal functioning and trans-di fferentiation into proliferative cells types [10,11]. It also induces the secretion of several chemokines, proinflammatory cytokines and growth factors etc (results from the present study) that could aid in membrane formation by remodeling of the extracellular matrix proteins in retina and thereby contribute to disease pathogenesis.

#### **5. Conclusions and Future Scope**

We have shown the involvement of microglial cells, inflammation and oxidative stress in fibrocellular membranes (ERM/ILM) of different pathological conditions. Under oxidative stress microglia cells ge<sup>t</sup> activated and interact with other glial cells and retinal cells to cause neuroinflammation and neurodegeneration. Reducing oxidative stress and resultant inflammation may help in reducing the risk of ERM formation. Since ILM peeling also has detrimental role on the retina, these findings not only expand our knowledge about ERM pathogenesis but are helpful in identifying newer potential therapies to clinically manage blinding diseases like RD and PDR.

Further investigations on role of microglia and immunomodulation of its phenotype are underway to expand our understanding of its role in DR pathogenesis and better managemen<sup>t</sup> of this disease and related retinopathies.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2076-3921/9/8/654/s1, Table S1: List of antibodies, Table S2: List of qPCR primers, Figure S1: Representative images of ERMs after H&E at 40X showing the glial cells with clear spindle shaped structure of the cells and their long thin processes. Macrophages were also seen with round darkly stained structure. Pigmentation was also observed in the membranes. Figure S2: Characterization of Müller glia cells using AldH1L1 (a specific marker for Müller glia) in MH, iERM, PDR and RD membranes.

**Author Contributions:** Conceptualization, I.K. and J.C.; Data curation, I.K., S.V., R.K.G., S.J., M.T., R.R.P. and J.C.; Formal analysis, I.K., S.V., R.K.G., S.J. and H.K.; Funding acquisition, I.K., H.K. and J.C.; Investigation, S.V., S.J., M.T., R.R.P., K.R. and J.C.; Methodology, I.K., S.V., R.K.G., G.H. and M.R.V.; Project administration, I.K. and J.C.; Resources, I.K., M.T., R.R.P., K.R., G.H., M.R.V. and J.C.; Software, I.K., S.V., R.K.G. and H.K.; Supervision, I.K. and J.C.; Validation, I.K., S.V., R.K.G., S.J., M.T., R.R.P. and H.K.; Visualization, I.K., S.V. and M.R.V.; Writing –original draft, I.K., S.V. and J.C.; Writing–review & editing, I.K., R.K.G., S.J., K.R., G.H., M.R.V. and H.K. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by grants from the Department of Biotechnology, Government of India (BT/01/COE/06/02/10), Hyderabad Eye Research Foundation and the Department of Biotechnology (BT/PR/16582/BID/667/2016) to IK, Department of Science and Technology, Government of India (SB/YS/LS-183/2014) to JC and from the National Institutes of Health: SI0OD021580 (to GH) and R01-EY022372 to HK.

**Acknowledgments:** We thank the patients and their families for their participation in this study. We also thank Subhabrata Chakrabarti and Shahna Shahulhamed at LVPEI for his help during the course of this study and Wei Zhang at UMMS for help with tissue procurement for TEM analyses. We are thankful to Sreedhar and Naidu, department of Pathology at LVPEI for their support in procuring frozen tissue sections. We sincerely thank the fellow doctors Chetan Videkar, Hitesh Agrawal, Mahima Jhingan, and Komal Agrawal for their support incollecting the membranes.

**Conflicts of Interest:** The authors declare no conflict of interest.
