*Review* **Harnessing Astrocytes and Müller Glial Cells in the Retina for Survival and Regeneration of Retinal Ganglion Cells**

**Hyung-Suk Yoo, Ushananthini Shanmugalingam and Patrice D. Smith \***

Department of Neuroscience, Carleton University, Ottawa, ON K1S 5B6, Canada; hyungyoo@cmail.carleton.ca (H.-S.Y.); UshananthiniShanmuga@cmail.carleton.ca (U.S.) **\*** Correspondence: patrice.smith@carleton.ca

**Abstract:** Astrocytes have been associated with the failure of axon regeneration in the central nervous system (CNS), as it undergoes reactive gliosis in response to damages to the CNS and functions as a chemical and physical barrier to axon regeneration. However, beneficial roles of astrocytes have been extensively studied in the spinal cord over the years, and a growing body of evidence now suggests that inducing astrocytes to become more growth-supportive can promote axon regeneration after spinal cord injury (SCI). In retina, astrocytes and Müller cells are known to undergo reactive gliosis after damage to retina and/or optic nerve and are hypothesized to be either detrimental or beneficial to survival and axon regeneration of retinal ganglion cells (RGCs). Whether they can be induced to become more growth-supportive after retinal and optic nerve injury has yet to be determined. In this review, we pinpoint the potential molecular pathways involved in the induction of growth-supportive astrocytes in the spinal cord and suggest that stimulating the activation of these pathways in the retina could represent a new therapeutic approach to promoting survival and axon regeneration of RGCs in retinal degenerative diseases.

**Keywords:** macroglia; astrocytes; Müller cells; optic nerve crush; retinal ganglion cells; spinal cord injury; signal transducer and activator of transcription 3; epidermal growth factor

### **1. Introduction**

The retina originates from the CNS during embryonic development [1]. The inner most layer of the retina harbors RGCs whose axons form the optic nerve that directly relays visual information to the brain [2]. RGCs can be considered CNS neurons because the optic nerve is myelinated by oligodendrocytes and do not regenerate spontaneously after injury [2,3]. Hence, the optic nerve crush (ONC) model has been widely used to determine the molecular mechanisms of neuronal survival and axon regeneration in the CNS [3–9]. While genetic and pharmacologic manipulations have been shown to activate neural repair mechanisms in RGCs after ONC, activation of macroglia in the retina, particularly astrocytes and Müller cells, has been shown to exert either detrimental or beneficial effects on survival and regeneration of RGCs after the injury [3,10–15].

Astrocytes in the CNS become activated in response to neuronal damage and neuroinflammation and form a dense network encapsulating the lesion site, known as the glial scar [15,16]. Although the glial scar functions as a physical and chemical barrier against further exposure to inflammatory agents, it also prevents growth of axons into the lesion site [16,17]. Astrocytes, therefore, have long been associated with the failure of axon regeneration in the CNS, and studies have attempted to eliminate or inhibit astrocytes to promote axon regeneration after CNS injury [16,18–20]. However, accumulating evidence now supports the concept that astrocytes are required for successful neuronal survival and axon regeneration in the CNS [12,13,16,19,21–25]. In this review, we will outline the evidence that reactive gliosis is required for successful neural repair in the CNS and suggest that harnessing the function of macroglia in the retina could promote survival and axon regeneration of RGCs.

**Citation:** Yoo, H.-S.; Shanmugalingam, U.; Smith, P.D. Harnessing Astrocytes and Müller Glial Cells in the Retina for Survival and Regeneration of Retinal Ganglion Cells. *Cells* **2021**, *10*, 1339. https:// doi.org/10.3390/cells10061339

Academic Editors: Maurice Ptito and Joseph Bouskila

Received: 27 April 2021 Accepted: 26 May 2021 Published: 28 May 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

#### **2. Macroglia in the Retina**

There are two types of macroglia in the retina: astrocytes and Müller cells. During retinal development, astrocytes from the brain enter the retina along the developing optic nerve [26,27]. In the mature retina, they are confined to the nerve fiber and ganglion cell layers [28]. On the contrary, Müller cells, the largest glial cell in the retina, originate from the retinal epithelium and span the entire retinal thickness [29,30]. The somata of Müller cells are located at the inner nuclear layer, and they extend their processes toward the outer and inner limiting membranes [31]. Müller cells ensheath retinal neurons and blood vessels in the plexiform and nerve fiber layers, allowing metabolic exchange between retinal vasculature and RGCs [32]. Astrocytes play a vital role in the development of the vascular system in the retina and contributes to the formation of the blood-retinal barrier [33,34]. Unlike Müller cells, astrocytes only envelop blood vessels in the nerve fiber and ganglion cell layers [32]. Astrocytes and Müller cells together maintain the integrity of the blood-retinal barrier by stabilizing tight junctions between endothelial cells and ensure the immune privilege of the eye [15,30]. They also provide essential nutrients, such as lactate and amino acids, from the circulation to neurons while participating in the retinal regulation of neurotransmitters, glucose metabolism and blood flow [10,15,30,35].

Astrocytes and Müller cells can provide RGCs with neurotrophic factors and antioxidants to maintain their viability [15,36–38]. Astrocytes can produce and release ciliary neurotrophic factor (CNTF), while Müller cells are known to be the major source of retinal brain derived neurotrophic factor (BDNF) and are capable of producing other well-known neurotrophic factors such as nerve growth factor (NGF) and glial cell line derived neurotrophic factor (GDNF) [14,39,40]. Increasing the retinal expression of these neurotrophic factors has been shown to promote survival and/or axon regeneration of RGCs. Moderate overexpression of BDNF in glaucomatous eye can result in long-term RGC survival while daily topical application of NGF can promote both survival and axon regeneration of RGCs after ONC [41,42]. Additionally, intravitreal co-administration of GDNF and CNTF can lead to survival and axon regeneration of RGCs after ONC possibly by directly binding to their respective receptors expressed by RGCs and/or inducing Müller cells to release additional neuroprotective factors, including BDNF and osteopontin [43]. Both macroglia can synthesize glutathione and counteract reactive oxygen species produced in the retina [11,15,36,37].

In response to the retinal injury, both astrocytes and Müller cells undergo reactive gliosis, up-regulating intermediate filaments, namely glial fibrillary acidic protein (GFAP), vimentin and nestin, and becoming more rigid [35]. This increased rigidness of both macroglia is mediated by signal transducer and activator of transcription 3 (STAT3), which is known as the master regulator of glial scar formation [18,35]. The rigidness allows for the glial scar formation, establishing the physical and chemical barrier to RGC axon regeneration [35]. Despite their proposed roles in inhibiting axon regeneration, studies show that these macroglia are rather essential in axon regeneration of RGCs. Astrocytes are known to release CNTF after lens injury and transform mature RGCs into a regenerative state, and the up-regulation of CNTF in astrocytes is also mediated by STAT3 [14,25]. Müller cells are also known to express CNTF after lens injury and hence may be involved in promoting the regeneration of RGCs in cooperation with astrocytes [44].

Since lens injury leads to intraocular inflammation, activation of inflammatory responses has been proposed to contribute to RGC axon regeneration [3,45]. Indeed, injecting the yeast wall extract zymosan can reproduce the regenerative effects of lens injury [3,14,44,45]. However, bacterial membrane component lipopolysaccharide (LPS) could not yield the same RGC axon regeneration, and this is due to zymosan's unique ability to stimulate dectin-1 receptors on leukocytes that invade the eye in response to the intraocular inflammation [46]. This finding suggests that infiltrating immune cells may secret neurotrophic factors that ultimately stimulate RGCs to regenerate after ONC [46]. Indeed, both macroglia and macrophages have been proposed to be the source of neurotrophic factors that promote RGC axon regeneration [44]. However, studies have shown that

depletion of macrophages from the eye does not reduce the regenerative effects of lens injury whereas reduced number of reactive macroglia compromised the beneficial effects of zymosan [44,47]. This suggests that macroglia may be the major mediators of the regenerative effects in response to the intraocular inflammation [44]. Considering the dual role of macroglia in RGC axon regeneration, they may exist in two reactive states: neurotoxic state and growth-supportive state. Since RGC axon regeneration could be induced specifically by lens injury, the reactive state of macroglia may depend on the type of injury.

#### **3. Two Distinct Reactive States of Astrocytes in the CNS**

Pioneering studies by Sofroniew and colleagues established the concept that reactive astrocytes are necessary for protecting neurons from further damages after CNS injury and helping them survive and regenerate [17,20,23,48–52]. Additionally, accumulating evidence now posits the idea of phenotypical heterogeneity among reactive astrocytes [18,53–56]. Zamanian et al. have identified two distinct reactive states of astrocytes in the CNS using a transcriptome; they genetically profiled astrocytes after a systemic injection of LPS or cerebral ischemia [55]. LPS-induced neuroinflammation resulted in astrocytes expressing the components of classical complement cascade that are hypothesized to drive the loss of synapses and subsequently neurodegeneration [55]. On the contrary, the ischemic injury induced astrocytes to express neurotrophic factors and cytokines that can promote neural repair in the CNS [55]. The two types of reactive astrocytes induced by neuroinflammation and ischemia are known as A1 and A2, respectively [56]. Studies showed that the A1 phenotype is neurotoxic while the A2 phenotype is beneficial for neuronal survival and axon regeneration [56–60]. Figure 1 depicts the two distinct molecular pathways leading to changes in phenotype and function of reactive astrocytes.

κ **Figure 1.** Neuroinflammation and ischemia lead to the generation of A1 and A2 astrocytes, respectively, in the spinal cord. Neuroinflammation may activate the NFκB pathway in reactive astrocytes and induce the A1 phenotype; on the other hand, ischemia may activate the STAT3 and/or PI3K/Akt pathways and induce the A2 phenotype. A1 and A2 astrocytes may exist on a phenotypic continuum. A1 astrocytes may be neurotoxic and promote neurodegeneration whereas A2 astrocytes may release neurotrophic factors and growth-supportive substrates and promote both survival and axon regeneration of CNS neurons. EGFR ligands may induce the generation of A2 astrocytes via activation of STAT3 and PI3K/Akt pathways (created with BioRender.com). Abbreviations: EGF, epidermal growth factor, EGFR, epidermal growth factor receptor, TGF-α, transforming growth factor-α, HB-EGF, heparin-binding EGF-like growth factor, NFκB, nuclear factor κB, STAT3, signal transducer and activator of transcription 3, PI3K/Akt, phosphoinositide 3-kinase/protein kinase B.

343

It should be noted that ONC results in generation of A1 astrocytes [57]. Neutralizing the factors that induce the A1 phenotype, such as interleukin-1α, tumor necrosis factor α and complement component 1q, could prevent the A1 formation and RGC death up to 14 days after injury [57]. Currently, it is unknown whether lens injury can generate A2 astrocytes in the retina and whether this phenotypic division also exists among Müller cells. However, identifying growth factors and downstream effectors involved in the induction of the A2 phenotype would allow for developing therapeutic strategies for protecting RGCs and promoting axon regeneration. Considering that findings from the ONC model generally have been found to hold true for spinal cord injury (SCI) [3], molecular mechanisms of inducing the A2 phenotype in the spinal cord could be applied to the retina and may hold therapeutic potential for retinal degenerative diseases.

#### **4. Harnessing Astrocytes to Promote Neural Repair in the Spinal Cord**

Astrogliosis has been extensively studied in the spinal cord, and accumulating evidence now suggests that it may have both beneficial and detrimental roles in the pathophysiology of SCI [18,23,51,52,56,61]. Although the signaling pathways leading to A1 and A2 phenotypes following SCI have yet to be determined, the nuclear factor κB (NFκB) and STAT3 pathways are believed to transform astrocytes into A1 and A2, respectively, because the roles of these two pathways seem to coincide with the hypothetical functions of A1 and A2 astrocytes (Figure 1) [61]. The NFκB pathway has a pivotal role in inducing neuroinflammation, and the inactivation of the pathway can reduce the expression of proinflammatory cytokines and significantly enhance the function of the spinal cord after SCI [62,63]. On the contrary, the inactivation of STAT3 pathway results in widespread infiltration of inflammatory cells and demyelination after SCI, whereas the activation of this pathway leads to rapid migration of reactive astrocytes to the lesion site to establish a physical barrier against inflammatory cells and promote significant improvement in functional recovery [64]. In further support of the idea that the STAT3 pathway may be involved in the generation of A2 astrocytes, Su et al. recently showed that down-regulation of microRNA-21 (miR-21) in astrocytes leads to STAT3-mediated conversion of A1 to A2 while up-regulation of miR-21 in astrocytes suppresses STAT3 activation and reverses the conversion process [60]. They also showed that these A2 astrocytes can promote axonal growth of neurons through the STAT3 pathway in vitro, suggesting that they may be beneficial to axon regeneration [60]. In addition to the STAT3 pathway, the phosphoinositide 3-kinase/protein kinase B (PI3K/Akt) pathway may also contribute to the generation of A2 astrocytes, as Xu et al. have shown that up-regulation of PI3K/Akt pathway and down-regulation of NFκB pathway are involved in counteracting A1 formation and promoting A2 formation [65]. Considering the evidence that the activation of STAT3 and PI3K/Akt pathways may be involved in the generation of A2 astrocytes, pharmacological stimulation of these pathways may lead to the increased population of A2 astrocytes that can promote neuronal survival and axon regeneration after CNS injury. The members of epidermal growth factor (EGF) family are known to activate these pathways via activation of epidermal growth factor receptors (EGFRs), and there is growing evidence that EGFR signaling can harness astrocytes to promote neuronal survival and axon regeneration in the CNS [21,22,66–70].

#### **5. Manipulating Epidermal Growth Factor Signaling to Promote Neural Repair in the CNS**

The EGF family is a group of related growth factors that are involved in a wide range of developmental processes, including proliferation, differentiation, and migration; the most notable members are EGF, transforming growth factor-α (TGF-α) and heparin-binding EGF-like growth factor (HB-EGF) [70]. These ligands signal through EGFR and three other homologous receptors, ErbB2, ErbB3 and ErbB4 [71]. Upon ligand binding, EGFRs undergo either homodimerization or heterodimerization; as an example, EGF can induce EGFR-EGFR homodimerization or EGFR-ErbB2 heterodimerization [70]. After this dimerization

process, phosphorylated tyrosine residues function as docking sites for signaling protein complexes that are involved in PI3K/Akt and STAT3 pathways [70].

Members of the EGF and EGFR families are widely expressed in various regions of the CNS, including spinal cord, brainstem, cerebellum, diencephalon, telencephalon and hippocampus [70]. Their main function in the developing and adult CNS is to stimulate the proliferation and differentiation of neural progenitors; as an example, EGF and TGF-α stimulate both embryonic and adult striatal progenitors to proliferate and then differentiate into neurons and astrocytes [72–74]. The EGF and EGFR families also have been shown to be involved in neural repair after CNS injury. Studies showed not only that EGFR expression increases in subventricular zone (SVZ) after ischemic injury but also that intraventricular infusion of EGF promotes proliferation of neural stem cells in SVZ after cerebral ischemia and eventually leads to neuronal replacement in the injured striatum [75,76]. EGFR ligands can also exert neuroprotective effects against neurodegeneration; studies have shown that EGF and HB-EGF can promote dopaminergic neuronal survival in animal models of Parkinson's disease [77,78].

Although EGF and EGFR have been shown to promote neurite outgrowth of cultured CNS neurons, retinal studies have shown that EGFR signaling is activated by growthinhibitory molecules, and inhibition of the EGFR signaling can promote axon regeneration of RGCs [79–81]. Koprivica et al. showed that myelin-derived proteins Nogo-66 and oligodendrocyte myelin glycoprotein can trigger indirect phosphorylation of EGFR in cultured postnatal cerebellar granule cells by activating their common receptor complexes that consist of Nogo-66 receptor (NgR) and its co-receptors p75/TROY and Lingo-1 [81]. Blocking the NgR-induced phosphorylation of EGFR using irreversible EGFR inhibitor PD168393 promoted neurite growth in retinal explant and axon regeneration in the ONC model [81]. The authors suggested that the EGFR inhibitor acts directly on RGCs to block their growth-inhibitory responses to myelin-derived proteins [81]. However, Douglas et al. later showed that the EGFR inhibitor has no direct impact on RGCs and, surprisingly, EGFR [82]. They reported that EGFR is only activated in glial cells, such as astrocytes and oligodendrocytes, in the retina and optic nerve, 14 days after optic nerve injury [82]. Furthermore, siRNA-mediated knockdown of EGFR in retinal culture could not promote neurite growth, but addition of competitive EGFR inhibitor AG1478 could restore neurite growth to the siRNA-treated cultures, suggesting that EGFR itself does not mediate the inhibition of axon regeneration [82]. Although the authors could not pinpoint the exact target of the EGFR inhibitor, they provided in vitro evidence that the inhibitor stimulates the release of neurotrophins, such as BDNF and NGF, from RGCs and retinal glia, and increases cyclic adenosine monophosphate, a second messenger involved in axon regeneration [82].

Currently, it is unclear whether EGFR signaling is involved in inhibition of axon regeneration. However, in support of the possibility that EGFR signaling could support axon regeneration in the CNS, a couple of studies have shown that TGF-α can promote axon regeneration after SCI [21,22]. Based on previous evidence that astrocytes can promote neuroprotection and may support axonal growth after injury, White et al. hypothesized that endogenous astrocytes could be harnessed to support axon regeneration with proper stimulation after SCI, and they intrathecally administered TGF-α in adult mice for two weeks following the injury [21]. TGF-α was able to stimulate proliferation and migration of astrocytes toward the lesion center and promote axon regeneration within the lesion [21]. As EGFR immunoreactivity was the strongest in GFAP-positive cells, the authors suggested that TGF-α acts directly on astrocytes [21]. Interestingly, TGF-α treatment not only increased the expression of neurocan, which is associated with inhibition of axonal growth, but also increased the expression of laminin throughout the lesion site [21]. As laminin immunoreactivity was co-localized with axons, the authors suggested that astrocytes may contribute to the formation of basal lamina structures throughout the lesion and provide a permissive substrate for axon elongation [21]. White and colleagues later published another study showing not only that TGF-α treatment can transform astrocytes into a growth-supportive phenotype that supports robust neurite outgrowth of dorsal root

ganglion cells in vitro but also that overexpression of TGF-α in vivo by intraparenchymal adeno-associated virus injection adjacent to the injury site increased axon regeneration at the rostral lesion border [22]. In support of these findings, Sofroniew and colleagues included EGF as a part of the combinatorial pharmacological treatment for CNS axon regeneration, showing that EGF can increase the release of axon growth-supportive substrates, including laminin, fibronectin and collagen, and contribute to the overall axon regeneration after SCI [68]. They also noted that although EGF significantly increased astrocyte proliferation and density, axons were able to grow through and beyond glial scar formation [68]. Recently, Chen et al. showed that EGF can generate A2 astrocytes in vitro by down-regulating A1-like genes and up-regulating A2-like genes, further supporting the previous findings [66]. Overall, it is evident that manipulation of EGFR signaling can aid CNS axon regeneration by inducing the A2 phenotype that can provide axon growth-supportive substrates.

#### **6. Manipulating EGFR Signaling in the Retina**

Although EGFR signaling seems to induce the A2 phenotype in the spinal cord, it is currently unknown whether EGFR signaling can also generate A2 astrocytes in the retina and promote RGC survival and axon regeneration after injury. Ever since the discovery of the intrinsic ability of mature RGCs to regenerate their axons, the major focus of ONC studies has been on further deciphering how to activate the intrinsic growth ability of RGCs [4–9]. Despite the evidence that retinal astrocytes and Müller cells can provide neurotrophic support to RGCs and maintain their viability, the post-injury reparative roles of these retinal macroglia still need to be elucidated. Additionally, no studies have shown whether EGFR ligand-induced EGFR signaling can lead to survival and axon regeneration of RGCs. Figure 2 depicts three molecular pathways that may be induced by manipulating EGFR signaling in the retina after ONC.

EGFR signaling is involved in proliferation of retinal progenitor cells and macroglia during retina development, and retinal EGFR expression has been shown to decrease as the retina matures and loses its mitogenic response to EGFR ligands [83]. However, retinal injury can increase EGFR expression in adult retina, suggesting that the retina becomes more responsive to EGFR ligands after injury [83]. Whether intravitreal injection of EGFR ligands after ONC can induce A2 formation and promote RGC survival and axon regeneration has yet to be investigated. However, as Harder et al. recently suggested that astrocytes may exert protective effects on RGCs through EGFR signaling [84], it may be possible that activation of EGFR signaling in astrocytes may at least promote RGC survival (Figure 2). Interestingly, they also showed that the up-regulation of complement C3 in astrocytes is mediated by EGFR signaling and responsible for protecting RGCs against high intraocular pressure [84]. Because C3 is a known marker of A1 astrocytes [57], this finding does not support the idea that A1 phenotype is neurotoxic and raises the question whether so-called neurotoxic astrocytes are indeed detrimental to neurons. However, this could also imply that astrocytes perhaps exist as a continuum, with a heterogeneous population of A1 and A2, as previously suggested by Liddelow and Barres (Figure 1) [56]. Therefore, it is possible that those neuro-supportive astrocytes with increased C3 expression could have exhibited a genetic profile that may lie somewhere in the middle of the phenotypic continuum. Overall, future research could consider investigating: (1) whether EGFR ligands can promote RGC survival and/or axon regeneration after injury via A2 formation, (2) whether downstream targets of EGFR signaling are involved in the potential neuroprotective and regenerative effects of EGFR ligands, and (3) the phenotypic ratio between retinal A1 and A2 astrocytes after injury and after post-injury treatment with EGFR ligands.

Since the retina also contains Müller cells that undergo reactive gliosis along with astrocytes, dissecting the reparative roles of retinal gliosis would require understanding the post-injury molecular changes in Müller cells. Unlike reactive astrocytes, the phenotypic dichotomy of reactive Müller cells has yet to be defined although studies seem to suggest that Müller cells may also exhibit a similar continuum of reactive states [35,85]. Because these two macroglia both express EGFR and respond to EGFR ligands [83,86,87], they may undergo similar molecular changes after injury (Figure 2). However, what makes Müller cells unique is that they can dedifferentiate into retinal progenitor-like cells and become pluripotent in response to EGFR signaling (Figure 2) [35,87]. Although how this dedifferentiation of Müller cells may contribute to the overall reactive gliosis in the retina remains unknown, studies have suggested that inducing Müller cells to become pluripotent may be a therapeutic strategy for retina regeneration in retinal degenerative diseases, such as age-related macular degeneration and glaucoma [35,87–89]. Future studies could try to determine the continuum of reactive states of Müller cells and investigate: (1) whether EGFR ligand treatments can harness Müller cells to become like A2 astrocytes and/or dedifferentiate into progenitor-like cells that may also contribute to survival and axon regeneration of RGCs, (2) the potential interaction between Müller cell gliosis and Müller cell-derived progenitor-like cells, and (3) how this interaction may affect retinal astrogliosis and the overall survival and axon regeneration of RGCs.

**Figure 2.** Intravitreal injection of EGFR ligands after ONC may lead to generation of A2 astrocytes and promote survival and axon regeneration of RGCs. It may also induce Müller cells to become more growth-supportive and help maintaining survival of RGCs and promoting optic nerve regeneration in cooperation with A2 astrocytes. Alternatively, Müller cells may dedifferentiate into retinal progenitor-like cells and contribute to survival and axon regeneration of RGCs (created with Biorender.com). Abbreviations: BDNF, brain derived neurotrophic factor, CNTF, ciliary neurotrophic factor, NGF, nerve growth factor, RGC, retinal ganglion cell, EGFR, epidermal growth factor.

#### **7. Conclusions**

Despite having an overall negative connotation for over a decade, reactive gliosis is now being recognized as an essential contributor to CNS repair [20,23,68]. Although

its positive contributions to neuronal survival and axon regeneration in the spinal cord has been established, its potential roles in promoting RGC survival and axon regeneration have yet to be determined. However, a growing body of evidence suggests that retinal macroglia may be essential in aiding post-injury survival and axon regeneration of RGCs [14,15,44,45,47,84]. Considering that EGFR signaling can contribute to axon regeneration in the spinal cord and activate the STAT3 pathway, which is critical to generation of A2 astrocytes, future research should investigate whether EGFR ligand-mediated EGFR signaling can induce A2 formation and promote survival and axon regeneration of RGCs after injury. It is also critical to understand whether EGFR signaling can induce Müller cells to become like A2 astrocytes, as they undergo reactive gliosis along with retinal astrocytes in response to injury. Additionally, there has been increasing attention to the possibility to deliver neurotrophic factors by eye drops, and clinical studies suggest that this could represent a safe and effective strategy for treating retinal degenerative diseases [42,90–94]. Hence, future research should evaluate the efficacy of EGFR ligand-based eye drops after ONC and investigate whether the topical delivery of EGFR ligands could generate growthsupportive phenotypes of retinal astrocytes and Müller cells. Harnessing retinal macroglia to promote survival and axon regeneration of RGCs would contribute to designing and improving therapeutic strategies for degenerative retinal diseases.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

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

#### **References**


### *Article* **Loss of Motor Protein MYO1C Causes Rhodopsin Mislocalization and Results in Impaired Visual Function**

**Ashish K. Solanki 1,† , Manas R. Biswal 2,† , Stephen Walterhouse 1,†, René Martin <sup>3</sup> , Altaf A. Kondkar <sup>4</sup> , Hans-Joachim Knölker <sup>3</sup> , Bushra Rahman <sup>1</sup> , Ehtesham Arif <sup>1</sup> , Shahid Husain <sup>5</sup> , Sandra R. Montezuma <sup>6</sup> , Deepak Nihalani 7,\* and Glenn Prazere Lobo 1,5,8,\***


**Abstract:** Unconventional myosins, linked to deafness, are also proposed to play a role in retinal cell physiology. However, their direct role in photoreceptor function remains unclear. We demonstrate that systemic loss of the unconventional myosin MYO1C in mice, specifically causes rhodopsin mislocalization, leading to impaired visual function. Electroretinogram analysis of *Myo1c* knockout (*Myo1c*-KO) mice showed a progressive loss of photoreceptor function. Immunohistochemistry and binding assays demonstrated MYO1C localization to photoreceptor inner and outer segments (OS) and identified a direct interaction of rhodopsin with MYO1C. In *Myo1c*-KO retinas, rhodopsin mislocalized to rod inner segments (IS) and cell bodies, while cone opsins in OS showed punctate staining. In aged mice, the histological and ultrastructural examination of the phenotype of *Myo1c*-KO retinas showed progressively shorter photoreceptor OS. These results demonstrate that MYO1C is important for rhodopsin localization to the photoreceptor OS, and for normal visual function.

**Keywords:** motor protein; myosin 1C; photoreceptor; rhodopsin; retina; outer segments; visual function

### **1. Introduction**

Protein trafficking and proper localization within the photoreceptors must occur efficiently and at high fidelity for photoreception, photoreceptor structural maintenance, and overall retinal cell homeostasis. Additionally, it is well-known that proper opsin localization is tightly coupled to photoreceptor cell survival and function [1–9]. However, the cellular events that participate in retinal injuries due to improper signalling and protein localization to the photoreceptor outer segments (OS) are not yet fully understood. While many proteins are known to play essential roles in retinal cell development and function, the involvement of motor proteins in eye biology is less understood. Identification of genetic mutations in the *Myo7a* gene, associated with retinal degeneration in Usher syndrome, suggests that unconventional myosins play a critical role in retinal pigmented

**Citation:** Solanki, A.K.; Biswal, M.R.; Walterhouse, S.; Martin, R.; Kondkar, A.A.; Knölker, H.-J.; Rahman, B.; Arif, E.; Husain, S.; Montezuma, S.R.; et al. Loss of Motor Protein MYO1C Causes Rhodopsin Mislocalization and Results in Impaired Visual Function. *Cells* **2021**, *10*, 1322. https://doi.org/ 10.3390/cells10061322

Academic Editors: Maurice Ptito and Joseph Bouskila

Received: 4 May 2021 Accepted: 25 May 2021 Published: 26 May 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

epithelium (RPE) and photoreceptor cell function [10,11]. Unconventional myosins are motor proteins that are proposed to transport membranous organelles along the actin filaments in an adenosine triphosphate (ATP)-dependent manner, and additional roles are currently being discovered [11–13]. The loss of *Myo7a* primarily affects RPE and OS phagocytosis, leading to retinal cell degeneration [10,11]. However, it is believed that other yet unidentified class I myosins may participate more directly in photoreceptor cell function. Here, we present compelling evidence for another unconventional actin-binding motor protein, MYO1C, which plays an important role in retinal cell structure and function via opsin localization to the photoreceptor OS.

Rhodopsin and cone pigments in photoreceptor OS mediate scotopic and photopic vision, respectively. The visual pigment rhodopsin is a prototypical G-protein-coupled receptor (GPCR), expressed by retinal rods for photon absorption. Light sensitivity is conferred by 11-*cis* retinaldehyde, a chromophore that is covalently linked to the K296 residue of the opsin protein [14–18]. Photon absorption causes a cis-to-trans conformational shift in the retinaldehyde, leading to structural changes in the opsin protein moiety [6,15]. This initiates a GPCR signalling pathway/phototransduction cascade, signalling the presence of light. Each photoreceptor cell contains an OS housing the phototransduction machinery, an inner segment (IS) where proteins are biosynthesized, and a synaptic terminal for signal transmission. One of the fundamental steps in vision is the proper assembly of signal-transducing membranes, including the transport and sorting of protein components. A major cause of neurodegenerative and other inherited retinal disorders is the improper localization of proteins. Mislocalization of the dim-light photoreceptor protein, rhodopsin, is a phenotype observed in many forms of blinding diseases, including retinitis pigmentosa (RP) [3,16]. The proteins that participate in phototransduction (including rhodopsin, transducin, phosphodiesterase (PDE6), or the cyclic nucleotide-gated channels (CNG)) are synthesized in the IS and must be transported through the connecting cilium to the OS. These proteins are either transmembrane or peripherally associated membrane, which are attached to the membrane surface [1–9]. How the transmembrane proteins (e.g., rhodopsin and CNG) and peripherally associated proteins (e.g., transducin and PDE6) traffic through the IS to incorporate eventually in the nascent disc membrane, or the photoreceptor outer membrane, is not fully understood and constitutes an area of intense research, as the mislocalization of these proteins causes retinal cell degeneration and can lead to blindness [1–9].

The myosin-1 family of molecular motors consists of eight different isoforms that participate in a wide range of cell biological processes that require generation or regulation of membrane tension, angiogenesis, formation of cell adhesions, and changes in the actin architecture [19–22]. Additionally, myosin-1 motors affect intracellular trafficking; function as tension-sensitive docks, phagocytosis, or tethers; and power membrane deformation [19–22]. Unconventional myosins are also proposed to be involved in the lightinduced translocation of mitochondria in photoreceptors and in human non-syndromic deafness [23–28]. Genetic mutations in myosins that lead to hearing loss have also been associated with retinal degeneration [29–33]. Some of the essential genes involved in either or both of these functions belong to a family of unconventional motor proteins and include MYO3A [29], MYO7A, MYO6, MYO15 [29–31], and MYO5. Recently, it was reported that mutations affected the nucleotide-binding pocket and calcium binding ability of another unconventional myosin, MYO1C, and these were associated with deafness [32,33]. Importantly, MYO1C was identified in proteomic analysis of the retina and vitreous fluid as part of a protein hub involved in oxidative stress [34]. MYO1C is an actin-binding motor protein that is widely expressed in multiple cell types. It participates in a variety of cellular functions, including protein trafficking and translocation [12,35–37]. As MYO1C has low tissue specificity based on mRNA and protein expression, it remains unclear which cell type is most dependent on MYO1C function and is affected by the loss of MYO1C.

In this study, we systematically analysed the function of the unconventional motor protein, MYO1C, in proper protein localization in photoreceptors. We found that a global genetic deletion of *Myo1c* resulted in a retinal phenotype only, which manifested as a progressive mislocalization of opsins to the OS. Using retinal lysate from wild-type (WT) mice in co-immunoprecipitation assays, we showed that MYO1C and rhodopsin directly interact, indicating that opsin is a cargo for MYO1C. Loss of MYO1C promoted a progressive shortening of OS that was concomitant with a reduction in photoreceptor function, suggesting that MYO1C is critical for maintenance of photoreceptor cell structure and for visual function. Our findings have significant clinical implications for degenerative rod and cone diseases, as mutations in MYO1C or its interacting partners are predicted to affect retinal health and visual function by altering opsin localization to the photoreceptor OS, a fundamental step for maintaining visual function in humans.

#### **2. Experimental Procedures**

#### *2.1. Materials*

All chemicals, unless stated otherwise, were purchased from Sigma-Aldrich (St. Louis, MO, USA) and were of molecular or cell culture grade quality.

#### *2.2. Myo1c-Knockout (Myo1c-KO) Mouse Model*

Mice were kept with ad libitum access to food and water at 24 ◦C in a 12:12 h light– dark cycle. All mice experiments were approved by the Institutional Animal Care and Use Committee (IACUC protocol #00780; G.P.L.) of the Medical University of South Carolina and performed in compliance with ARVO Statement for the use of Animals in Ophthalmic and Vision Research. We previously generated *Myo1c* transgenic mice (*Myo1cfl/fl*) in C57BL/6N-derived embryonic stem cells, flanking exons 5 to 13 of the mouse *Myo1c* gene, which allowed us to specifically delete all *Myo1c* isoforms in a cell-specific manner [28]. Here, a complete *Myo1c*-knockout was generated by crossing *Myo1cfl/fl* mice with an Factin Cre mouse strain (B6N.FVB-Tmem163Tg (ACTB-cre)2Mrt/CjDswJ) obtained from Jackson Labs. We refer to the *Myo1cfl/fl* × f-actin Cre cross as *Myo1c* knockout (*Myo1c*-KO) mice. Since the role of Myo1c has not been investigated, the F-actin Cre+ mice gave us an opportunity to study MYO1C function in an un-biased fashion in various cell types/tissues. For this study, the *Myo1c*-KO mice were crossed onto a C57BL/6J background to avoid potential problems with the *Rd8* mutation (found in C57BL/6N lines), and all breeding pairs were sequenced and were negative for *Rd8* and *Rd1* mutations [38]. Equal numbers of male and female mice (50:50 ratio) were used per group and timepoint.

#### *2.3. Immunohistochemistry and Fluorescence Imaging*

Light-adapted mice were euthanized, and their eyes were immediately enucleated. The eyes were fixed in 4% paraformaldehyde and buffered with 1X PBS for 2 h at 4 ◦C, using established protocols [39]. After fixation, samples were washed in 1X PBS and embedded in paraffin and processed (MUSC Histology core facility). Sections (10 µm) were cut and transferred onto frost-free slides. Slide edges were lined with a hydrophobic marker (PAP pen), deparaffinized using xylene, and processed through ethanol washes before blocking for 1–2 h at RT. Blocking solution (1% BSA, 5% normal goat serum, 0.2% Triton-X-100, 0.1% Tween-20 in 1X PBS) was applied for 2 h in a humidified chamber. Primary antibodies were diluted in blocking solution as follows: anti-rhodopsin 1D4 (1:500, Abcam, Cambridge, MA, USA), anti-Myo1c M2 (1:100) [40], cone-arrestin (1:250, Millipore-Sigma, St. Louis, MO, USA), conjugated PNA-488 (1:2000, Molecular Probes, Eugene, OR), anti-red/ green cone opsin (M-opsin; 1:500; Millipore, St. Louis, MO, USA), anti S-opsin (1:500, Millipore-Sigma, St. Louis, MO), ZO1 (1:2000, Invitrogen, Waltham, MA, USA), Pde6b (1:300, ThermoFisher, Waltham, MA, USA), CNGA1 (1:250, Abcam), rod arrestin (1:250, Invitrogen), Stra6 (1:250, Millipore-Sigma), CRALBP (1:100, Invitrogen), rod transducin (1:250, Santa Cruz, Dallas, TX, USA), and 4′ ,6-diamidino-2-phenylendole (DAPI; 1:5000, Invitrogen) or Hoechst (1:10,000, Invitrogen), which were used to label nuclei. All secondary

antibodies (Alexa 488 or Alexa 594) were used at 1:5000 concentrations (Molecular Probes, Eugene, OR, USA). Optical sections were obtained with a Leica SP8 confocal microscope (Leica, Wetzlar, Germany) and processed with the Leica Viewer software, or using a Keyence BZ-X800 scope. All fluorescently labelled retinal sections on slides were analysed by the BioQuant NOVA Prime Software (R & M Biometrics, Nashville, TN, USA) and fluorescence within individual retinal layers quantified using Image *J* or Fiji (NIH).

#### *2.4. Measurement of Photoreceptor ONL Thickness and OS Lengths*

The lengths of the photoreceptor OS in WT and *Myo1c*-KO animals (from H&E sections of retinas) were imaged (Keyence BZ-X800 microscope) and measured at 12 consecutive points (at 150 µm distances) from the optic nerve (ON). The OS length was measured from the base of the OS to the inner side of the retinal pigment epithelium. The total number of layers of nuclei in the ONL of retinal sections through the optic nerve (ON) was imaged (Keyence BZ-X800 microscope) and measured at 12 locations around the retina, six each in the superior and inferior hemispheres, starting at 150 µm from the ON. Retinal sections (*n* = 5–7 retinal sections per eye) from *n* = 8 mice for each genotype and timepoint were analysed. Two-way ANOVA with Bonferroni post-tests compared *Myo1c*-KO to WT mice at each segment measured.

#### *2.5. ERG Analysis*

Dark-adapted WT and *Myo1c*-KO mice (50:50 male–female ratio; *n* = 8 each genotype) at 2 months of age (young mice; early timepoint), and 6 months of age (end timepoint) were anesthetized by intraperitoneal injection of a ketamine/xylene anaesthetic cocktail (100 mg/kg and 20 mg/kg, respectively), and their pupils were dilated with 1% tropicamide and 2.5% phenylephrine HCl. ERGs were performed under dim red-light in the ERG rooms in the morning (8 a.m.–12 noon). Scotopic ERGs were recorded with a computerized system (UTASE-3000; LKC Technologies, Inc., Gaithersburg, MD, USA), as previously described [39,41,42].

#### *2.6. TEM Analysis of Retinas*

Eyecups at the indicated timepoints were harvested and fixed overnight at 4 ◦C in a solution containing 2% paraformaldehyde/2.5% glutaraldehyde (buffered in 0.1 M cacodylate buffer). Samples were rinsed in the buffer (0.1 M cacodylate buffer) and then placed in a post-fixative of 2% OsO4/0.2 M cacodylate buffer for 1 h at 4 ◦C, followed by a 0.1 M cacodylate buffer wash. The samples were dehydrated through a graded ethanol series and then embedded in Epon (EMbed 812; EM Sciences, Hatfield, PA, USA). For TEM analysis, each eye (*n* = 6 individual eyes from *n* = 6 animals of each genotype) was cut in half before embedding in Epon blocks. Sections were parallel to the dorsoventral meridian and near the optic nerve (ON). The cured blocks were sectioned at 0.5 microns (semi-thin plastic sections) and stained with 1% toluidine blue to orient the blocks to the required specific cell types. The blocks were trimmed to the precise size needed for ultrathin sectioning. The blocks were cut at 70 nm and gathered on one-micron grids. The grids were air-dried, stained with uranyl acetate for 15 min and lead citrate for 5 min, and rinsed between each stain. They were allowed to dry and imaged with a JEOL 1010. Images were acquired with a Hamamatsu camera and software. All samples were processed by the Electron Microscopy Resource Laboratory at the Medical University of South Carolina, as previously described [39].

#### *2.7. Western Blot Analysis and Densitometry*

Total proteins from cells or mouse tissues (*n* = 3 per genotype) were extracted using the M-PER protein lysis buffer (ThermoScientific, Beverly, MA, USA) containing protease inhibitors (Roche, Indianapolis, IN, USA). Approximately 25 µg of total protein was electrophoresed on 4–12% SDS-PAGE gels and transferred to PVDF membranes. Membranes were probed with primary antibodies against anti-*Myo1c* (1:250), CRALBP (1:100, Invitrogen), rod transducin (1:250, Santa Cruz), PKCα (1:500, Novus Biologicals, Centennial, CO, USA), and β-Actin or Gapdh (1:10,000, Sigma) in antibody buffer (0.2% Triton X-100, 2% BSA, 1X PBS) [39,43]. HRP-conjugated secondary antibodies (BioRad, Hercules, CA, USA) were used at 1:10,000 dilution. Protein expression was detected using a LI-COR Odyssey system, and relative intensities of each band were quantified (densitometry) using Image *J* software version 1.49 and normalized to their respective loading controls. Each Western blot analysis was repeated thrice.

#### *2.8. Co-Immunoprecipitation (Co-IP) Assays*

Co-immunoprecipitation of endogenously expressed proteins (MYO1C and rhodopsin) was performed using mouse retinal extracts. Six retinas of each genotype (*n* = 3 animals of WT and *Myo1c*-KO) were used for extraction of retinal proteins in 250 µL of RIPA buffer (phosphate-buffered saline (PBS) containing 0.1% sodium dodecyl sulphate (SDS), 1% Nonidet P-40, 0.5% sodium deoxycholate, and 100 mM potassium iodide) with EDTAfree proteinase inhibitor mixture (Roche Molecular Biochemicals). Lysates were cleared by centrifugation at 10,000 rpm for 10 min at 4 ◦C. The prepared lysates were further incubated with anti-rhodopsin and mouse/rabbit IgG overnight at 4 ◦C and further with protein G-coupled agarose beads (ROCHE) for 1–2 h. Beads were then collected by centrifugation at 3000 rpm for 5 min at 4 ◦C, extensively washed in 1X PBS, and resuspended in SDS gel loading buffer. The proteins were separated on a 10% SDS-PAGE, transferred to a nitrocellulose membrane, and analysed by immunoblotting with the corresponding antibodies.

#### *2.9. Overlay Direct Binding Assay*

Rhodopsin protein was overexpressed in HEK293 cells using transient transfection (pcDNA3 rod opsin construct, a gift from Robert Lucas (Addgene plasmid #109361, http: //n2t.net/addgene:109361, accessed on 25 May 2021; RRID:Addgene\_109361) [44], and cell lysate with overexpressed rhodopsin was subjected to SDS-PAGE gel and transferred to PVDF membrane. The membrane was then probed by overlaying it with 5 µg of baculovirus-produced and purified recombinant full-length MYO1C FL [13] protein by incubating at 4 ◦C for 4 h. Following incubation, the membrane was Western blotted with MYO1C antibody to detect the direct binding of MYO1C to the rhodopsin bands. The location of rhodopsin on the membranes was marked by separately probing these membranes with an anti-rhodopsin (1:500, Millipore Sigma) antibody (Figure 7B).

#### *2.10. ELISA*

ELISA was performed as described previously, with minor modifications [44]. In total, 100 ng of mammalian-expressed and purified rhodopsin was coated on individual wells of a 96-well Maxisorp Immunoplate (Nunc, Rochester, NY, USA) and incubated at 4 ◦C overnight. The wells were blocked with 5% BSA (Sigma) in PBS for 4 h at 37 ◦C, and then washed with 1X PBS, 0.1% Tween 20 solution (PBS-T). The wells in the plates were incubated with 200 ng of MYO1C protein for 4 h at 37 ◦C. Following incubation, the wells were washed with PBS-T solution and incubated with MYO1C antibody for 4 h. Post incubation, secondary antibody (HRP-conjugated) against the Fc region of human IgG1 mAbs at a dilution of 1:5000 in PBS containing 5% BSA was added, and the plates were kept for 1 h at room temperature. The plates were then washed three times with PBS-T and twice with PBS and developed by adding 100 µL of substrate (3,3,5,5-tetramethylbenzidine) solution (Pierce, Hägersten, Sweden). Incubation was conducted at room temperature, the reaction was stopped as the colour developed by adding 100 µL of 2 M H2SO4, and absorbance at 450 nm was measured on a microplate reader (Biotek, Winuschi, VT, USA).

#### *2.11. Quantitative Real-Time PCR*

RNA was isolated from the retinas of WT and *Myo1c*-KO animals using Trizol reagent and processed as described previously [43]. One microgram of total RNA was reverse transcribed using the SuperScript II cDNA Synthesis Kit (Invitrogen, Eugene, OR, USA). Quantitative real-time PCR (qRT-PCR) was carried out using SYBR green 1 chemistry (BioRad, Hercules, CA, USA). Samples for qRT-PCR experiments were assayed in triplicate, using the BioRad CFX96 Q-PCR machine. Each experiment was repeated twice (*n* = 6 reactions for each gene), using newly synthesized cDNA.

#### *2.12. Liver Function Tests Using Alanine Aminotransferase (ALT) Assays*

To extract total protein, liver tissues from WT or *Myo1c*-KO mice (pooled livers *n* = 4 mice per genotype) were homogenized in RIPA buffer on ice and then centrifuged at 14,000 rpm at 4 ◦C for 10 min. Supernatant was collected, and the protein concentration was estimated using the Bio-Rad Protein Assay Dye Reagent (Sigma). A total of 10 µL of liver lysate was transferred to 96-well plate, and ALT was measured using a microplatebased ALT activity assay kit (Pointe Scientific, Cat. A7526, Irvine, CA, USA). Five biological replicates were used in the assay.

#### *2.13. Heart Function Tests Using Echocardiographic (ECHO) Analyses*

Echocardiographic (ECHO) analysis was performed on adult wild-type (WT) and *Myo1c*-KO animals (*n* = 4 per genotype) at the MUSC Cardiology Core Facility. For ECHO experiments, mutant and wild-type littermate controls were anesthetized in an induction chamber with 5% isoflurane in 100% oxygen. They were removed and placed on a warming table where anaesthesia was maintained via nose cone delivery of isoflurane (1% in 100% oxygen). They were placed in the supine position, and the thoracic area was shaved. The limbs were taped to the platform to restrict animal movement during echocardiography acquisition. This also provided a connection to ECG leads embedded in the platform. Sonography gel was applied to the chest and echocardiographic measurements of the peristernal long axis and short axis of the heart were acquired to derive the systolic and diastolic parameters of heart function. ECHO measurements were estimated using Vevo 2100 instrumentation.

#### *2.14. Statistical Analysis*

Data are expressed as means ± standard deviation by ANOVA in the Statistica 12 software (StatSoft Inc., Tulsa, OK, USA). Differences between means were assessed by Tukey's honestly significant difference (HSD) test. *P*-values below 0.05 (*p* < 0.05) were considered statistically significant. For Western blot analysis, relative intensities of each band were quantified (densitometry) using the Image *J* software version 1.49 and normalized to the loading control β-actin. The qRT-PCR analysis was normalized to 18S RNA, and the ∆∆Ct method was employed to calculate fold changes. Data of qRT-PCR are expressed as mean ± standard error of mean (SEM). Statistical analysis was carried out using PRISM 8 software-GraphPad.

#### **3. Results**

#### *3.1. Construction and Validation of Myo1c Null Mice*

We previously generated *Myo1c* floxed mice using the standard knockout strategy [45] (Figure S1A). Systemic deletion of *Myo1c* was achieved by crossing *Myo1c* floxed (*Myo1cfl/fl*) mice with Actin Cre+ (ActCre+; JAX labs) mice to generate *Myo1cfl/fl-ActCre+/*− knockout mice (referred to as *Myo1c*-KO mice in this manuscript). Western blotting of protein lysates from various tissues, including kidney, heart, and liver of *Myo1c*-KO mice showed complete loss of MYO1C, thus confirming the systemic deletion of *Myo1c* (Figure S1B). Additionally, immunofluorescence expression analysis of these tissues further confirmed loss of MYO1C protein in *Myo1c*-KO mice (Figure S2A–C).

#### *3.2. Genetic Deletion of Myo1c Induced Visual Impairment in Mice*

**2021**, , x 7 of 19

Immunofluorescence analysis showed that MYO1C was enriched in the rod photoreceptor outer (OS) and inner segments (IS) (Figure 1A), as well as in the cone photoreceptor OS of wild-type (WT) mice (Figure 1B), but it was absent in the photoreceptors of *Myo1c*-KO animals (Figure 1A,C). Western blot analysis further confirmed that MYO1C protein was absent in the retinas of *Myo1c*-KO mice (Figure 1D). Since mutations or deletion of the motor protein, MYO7A, were associated with retinal degeneration in Usher syndrome and its animal model, we were prompted to investigate the effect of *Myo1c* in retinal function. Using electroretinograms (ERGs) [46,47], we tested photoreceptor cell function of *Myo1c*-KO and WT mice (*n* = 8 mice per genotype and age-group; 50:50 male–female ratio) under dark-adapted scotopic conditions. In contrast to WT animals, we observed reduced ERGs for *Myo1c*-KO mice at different ages. Two-month-old *Myo1c*-KO mice showed a significant reduction in the *a*-wave amplitudes, but not in *b*-wave amplitudes (*p* < 0.0068 and *p* < 0.098, respectively) (Figure 2A,C). Strikingly, ERG analysis of adult six-month-old *Myo1c*-KO mice showed loss of retinal function, in which a significant reduction in both *a*and *b*-waves was observed (38–45% lower than WT animals (\*\* *p* < 0.005; Figure 2B,D).

**Figure 1.** MYO1C localizes to photoreceptors in mouse retina. Eyes from adult wild-type (WT) and *Myo1c*-KO mice (*n* = 8 mice per genotype; 50:50 male–female ratio) were harvested, and retina sections (*n* = 5–7 sections per eye) were immunostained with an anti-MYO1C antibody (**A**–**C**) and M-opsin antibody (**B**,**C**), followed by secondary (Alexa 488 or Alexa 594) antibody staining. MYO1C (green fluorescence), M-Opsin (red fluorescence), and DAPI or Hoechst (blue fluorescence). Figures in (**A**–**C**) are representative of retinal sections (*n* = 5–7 sections per eye) imaged from *n* = 8 animals per genotype. (**B**,**C**) Merge (orange) represents co-localization of MYO1C-488 (green) with M-Opsin-594 (red). White arrows in *B* show cones. *RPE*, retinal pigmented epithelium; *OS*, outer segments; *IS*, inner segments; *ONL*, outer nuclear layer. (**A**–**C**) Scale bar = 50 µm. (**D**) Total proteins isolated from WT (*n* = 4) and *Myo1c*-KO (*n* = 4) mouse retinas were pooled sequentially and subjected to SDS-PAGE. Two different concentrations of protein (10 µg and 20 µg) were used. Blots were then probed with anti-Myo1c and Gapdh antibodies. Western blot analysis were repeated thrice. Arrows indicate MYO1C protein band in retinal lysates of WT mice.

− − − − − − − − **Figure 2.** Genetic deletion of *Myo1c* in mice results in a decreased visual function. Dark-adapted scotopic ERGs were recorded in response to increasing light intensities in cohorts of control wildtype (WT) (blue bars, blue-traces) and *Myo1c*-KO (red bars, red-traces) mice, aged two months old (**A**,**C**), and six months old (**B**,**D**). Two-month-old *Myo1c*-KO mice had lower dark-adapted *a*- and *b*-wave amplitudes compared with those of controls (post-hoc ANOVA: *a*-waves, *p* < 0.0068; *b*-waves, *p* < 0.0098, n.s. not significant.), in particular at higher light intensities (−40, −30, −20, −10, 0 dB). Six-month-old *Myo1c*-knockout mice had lower dark-adapted *a*- and *b*-wave amplitudes compared with those of controls (post-hoc ANOVA: *a*-waves, \*\* *p* < 0.005; *b*-waves, \*\* *p* < 0.005), in particular at higher light intensities (−40, −30, −20, −10, 0 dB). Photoreceptor cell responses (*a*-waves), which drive the *b*-waves, were equally affected in 6-month-old *Myo1c*-KO animals (both reduced on average between 38 and 45% of WT animals). Data are expressed as mean ± S.E. (*Myo1c*-KO mice and WT mice, *n* = 8 per genotype and age-group; 50:50 male-female ratio).

#### *3.3. Localization of Rod and Cone Visual Pigments in Myo1c-KO Mice*

Since the phototransduction protein rhodopsin constitutes 85–90% of photoreceptor OS protein content [48,49], and as the ERG responses were impaired in *Myo1c*-KO mice, we hypothesized that the loss of MYO1C might have affected proper opsin localization to the photoreceptor OS. To test this hypothesis, we analysed retinal sections from WT and *Myo1c*-KO mice (at 2 and 6 months of age; 5–7 retinal sections per eye from *n* = 8 mice per genotype and age-group; 50:50 male–female ratio), probing for rhodopsin, two types of cone opsins, medium wavelength R/G opsin (M-opsin) and short wavelength S-opsin, rod-specific phosphodiesterase 6b (Pde6b), rod-specific CNGA1, rod arrestin (ARR1), rod transducin (G-protein), and the general cone marker PNA lectin. In WT mice at 2 and 6 months of age, rhodopsin localized exclusively to the rod OS (Figure 3A). While the majority of rhodopsin trafficked to the OS in two-month-old *Myo1c*-KO mouse retinas, some mislocalization to the base of the rod IS and the cell bodies in the outer nuclear layer (ONL) was noted (Figure 3A; white arrows; rhodopsin levels within individual retinal layers were quantified and shown in Figure S3A–C). This suggested incomplete opsin localization to photoreceptor OS in the absence of MYO1C. An even more severe mislocalization of rhodopsin to the rod IS and within the ONL was observed in the 6 month-old *Myo1c*-KO mice, suggesting a progressive retinal phenotype in the absence of MYO1C (Figure 3A; rhodopsin expression within individual retinal layers were quantified and shown in Figure S3D,E). Staining for the two cone opsins showed that the cone OS were shorter and mis-shaped by two months, and this abnormality increased by six months of age (Figure 3B,C). Retinas stained for PNA lectin showed progressively shorter and

mis-shaped cone OS, indicating that the cone OS structure was compromised in the absence of MYO1C as these mice aged (Figure 3D). Cone visual arrestin in WT mice retina typically outlines the entire cell, OS, IS, cell body, axon, and cone pedicle. Staining for cone arrestin in *Myo1c*-KO animals (2 months of age) confirmed the short and mis-shaped appearance of the cone OS compared to WT retinas at similar ages (Figure 4A, white arrows). In our case, this antibody did not reveal staining to other cone structures, except for the cone OS, so we could not distinguish changes (if any) in cone IS, cell body, axon, or pedicle among WT and Myo1c-KO animals. By contrast, staining for Pde6b, a lipidated rod-specific protein that traffics to the OS independently of rhodopsin [2], showed normal localization to the rod OS in both WT and *Myo1c*-KO retinas at 2 months of age (Figure 4B).

**Figure 3.** Immunohistochemical analysis of wild-type (WT) and *Myo1c*-knockout mice retinas shows rhodopsin localization defects: (**A**) Levels and localization of rhodopsin (Rho); (**B**) red/green medium wavelength cone opsin (M-opsin); (**C**) short wavelength cone opsin (S-opsin); (**D**) PNA-488 analysed in two- and six-month-old WT and *Myo1c*-KO mice retinas. *Arrows* in panel *A* highlight rhodopsin mislocalization to IS and cell bodies in *Myo1c*-knockout mouse retinas. Images in panels (**A**–**D**) are representative of immunostained retinal sections (*n* = 5–7 sections per eye) imaged from *n* = 8 animals per genotype and age group (50:50 male–female ratio). Scale bars = 75 µm and 25 µm (**A**, two months old and six months old, respectively); scale bar = 50 µm (**B**–**D**). *OS*, outer segments; *IS*, inner segments; *ONL*, outer nuclear layer; *OPL*, outer plexiform layer; *INL*, inner nuclear layer.

The CNG channels are also important mediators in the photoreceptor transduction pathways, and they require proper localization to the OS for normal photoreceptor cell function [5,49]. Additionally, the absence of CNGA1 or CNGB1 in mice led to decreased ERG responses and progressive rod and cone photoreceptor cell death [5]. Therefore, to rule out alternate mechanisms for the observed functional phenotypes in *Myo1c*-KO retinas, the retinas of WT and *Myo1c*-KO mice (3–4 months of age; 5–7 retinal sections per eye from *n* = 8 mice per genotype; 50:50 male–female ratio) were stained with the CNGA1 antibody. This analysis showed that even in the absence of MYO1C, both young and adult mice retinas showed no defects in the proper localization of CNGA1 protein to OS (Figure 4C; CNGA1 protein distribution in photoreceptor layer quantified and shown in Figure 4F). **2021**, , x 10 of 19

**Figure 4.** Immunohistochemical analysis of protein localization in photoreceptors of wild-type (WT) and *Myo1c*-knockout mice retinas: Levels and localization of (**A**) cone arrestin (ARR), (**B**) Pde6b; (**C**) CNGA1; (**D**) rod Arrestin (ARR1); and (**E**) *Gprotein* (*transducin*) were analysed in WT and *Myo1c*-KO mice retinas to evaluate proper protein localization to photoreceptor OS. Red Arrows in panel **A** highlight cone photoreceptor nuclei and OS in WT mouse retinas that were significantly reduced or shorter, respectively, in *Myo1c*-KO animals (white arrows in **A**). Images in panels **A**–**E** are representative of immunostained retinal sections (*n* = 5–7 sections per eye) imaged from *n* = 8 animals per genotype and age group (50:50 male–female ratio). Panels (**A**,**B**), mice were 2–3 months of age. Panels **C**–**E**, mice were 3–4 months of age. (**F**) Protein distribution (in %) of CNGA1, rod ARR1, and transducin within the photoreceptor OS and IS in light-adapted mice. For quantification of protein distribution within retinal layers, 5–7 retinal sections from each eye (*n* = 8 animals for each genotype) were analysed using Image *J*. (**G**) Representative Western blot (*n* = 3 repeats) images of retinal proteins from 3–4-month-old WT and *Myo1c*-KO mice (*n* = 2 animals per genotype) showed no significant differences in protein expression of key retinal genes among genotypes. *OS*, outer segments; *IS*, inner segments; *ONL*, outer nuclear layer; *INL*, inner nuclear layer; *OPL*, outer plexiform layer; *IPL*, inner plexiform layer.

The soluble proteins, arrestin and transducin, exhibit light-dependent localization, where in response to light, arrestin migrates to rod OS and transducin translocates to rod IS [50]. To test whether the loss of MYO1C affected rod arrestin (ARR1) and rod G-protein (transducin) localization, we performed IHC staining for these proteins in retinas of light-adapted WT and *Myo1c*-KO mice (3–4 months of age; 5–7 retinal sections per eye from *n* = 8 mice per genotype; 50:50 male-female ratio). These analyses showed that in the presence of light, genetic loss of MYO1C had no negative effect on the localization of rod arrestin to the OS and G-protein to the IS and cell bodies in retinas of *Myo1c*-KO mice

α

(Figure 4D,E; rod ARR1 and transducin protein distribution in photoreceptor layer quantified and shown in Figure 4F). Using total protein lysates from retinas of WT and *Myo1c*-KO mice (3–4 months of age; four pooled retinas from *n* = 2 mice per genotype), we analysed protein expression of key retinal proteins in specific retinal cells: CRABLP1 (expressed in Müller cells), GNAT1 (expressed in photoreceptors), and PKCα (expressed in retinal bipolar cells). These analyses showed no significant differences in the expression of these genes in the inner or outer retinal layers of *Myo1c*-KO mice when compared to those of WT mice at 3–4 months of age (Figure 4G). Although MYO1C could not be detected by immunohistochemical analysis in mouse RPE, functional MYO1C and *Myo1C* mRNA were reported in human RPE cells [43] and mouse RPE [51], respectively. Since the elimination of the motor protein, *Myo7a*, in mice leads to alterations in protein localization in the RPE (RPE65) [52], we stained retinas of young and adult WT and *Myo1c*-KO mice (5–7 retinal sections per eye from *n* = 8 mice per genotype) with an anti-STRA6 antibody, another RPE-specific protein. This analysis showed that STRA6 expression and localization in the RPE was not affected in the absence of MYO1C (Figure S4). Since the motor protein MYO1C is proposed to have various functions, such as in protein trafficking, organization of F-actin, mitotic spindle regulation, and gene transcription [22,40], based on our observations above, we further investigated one of its roles in photoreceptor homeostasis. Our hypothesis was that its absence in photoreceptors of *Myo1c*-KO animals might contribute specifically to defective rhodopsin localization to the photoreceptor OS, which might result in retinal phenotypes.

#### *3.4. Native Cre+ Mice Showed No Retinal Phenotypes*

To rule out any Cre+-mediated effects on retinal phenotypes observed in the *Myo1c*-KO; Cre+ animals, the eyes from native Cre+ mice (3–4 months old; *n* = 3 animals) were harvested and subjected to similar histological and immunofluorescence analysis. As compared to age-matched WT mice retinas (*n* = 3 animals), the retinas of Cre+ mice showed no retinal pathology or mislocalization of opsins (Figure S5A vs. Figure S5B). These analyses support the view that genetic loss of MYO1C affects key components of phototransduction specifically, and this is further manifested in defects in visual function.

#### *3.5. Myo1c-KO Mice Demonstrated Photoreceptor OS Loss*

To further evaluate if rhodopsin mislocalization was associated with structural changes to the retina, histological and transmission electron microscopy (TEM) analyses of retinal sections of young and adult WT and *Myo1c*-KO mice were performed. In histological sections of retinas (5–7 retinal sections per eye from *n* = 8 mice per genotype and age), the progressive shortening of rod photoreceptor OS was observed. The OS of adult *Myo1c*-KO mice at 6 months of age were shorter than the OS of *Myo1c*-KO mice at 2 months of age, which in turn were shorter than those in WT mice at similar ages (Figure 5A,B; OS lengths quantified from H&E sections and represented using spider-plots in Figure 5C,D; \*\* *p* < 0.05). In comparison to WT mice, the photoreceptors in *Myo1c*-KO mice were less organized, especially in the 6-month-old mice (Figure 5B), suggesting that loss of MYO1C may progressively affect photoreceptor homeostasis. The retina outer nuclear layer (ONL) thickness between genotypes at both ages revealed no significant reduction in nuclear layers in *Myo1c*-KO animals compared to WT mice (ONL thickness quantified from H&E stained sections and represented using spider-plots in Figure 5E,F).

**Figure 5.** Histological analysis shows reduced photoreceptor OS lengths in *Myo1c*-KO mice retinas: (**A**,**B**) Retinas from 2- and 6-month-old WT and *Myo1c*-KO mice were sectioned, using an ultra-microtome, and semi-thin plastic sections were obtained to evaluate pathological consequences of MYO1C loss. Quantification of OS lengths from H and E sections (**C**), two- month-old mice; (**D**), six-month-old mice) and ONL thickness (**E**), two-month-old mice; (**F**), six month-old mice, using "spider graph" morphometry. The OS lengths and total number of layers of nuclei in the ONL from H and E sections through the optic nerve (ON; 0 µm distance from optic nerve and starting point) were measured at 12 locations around the retina, six each in the superior and inferior hemispheres, each equally at 150 µm distances. *RPE*, retinal pigmented epithelium; *OS*, outer segments; *IS*, inner segments; *ONL*, outer nuclear layer; *INL*, inner nuclear layer; *OPL*, outer plexiform layer. Retinal sections (*n* = 5–7 sections per eye) from *n* = 8 mice for each genotype and time point (50:50 male-female ratio) were analysed. Two-way ANOVA with Bonferroni post-tests compared *Myo1c*-KO mice with WT in all segments. \*\* *p* < 0.005, for OS length in only 6-month-old *Myo1c*-KO mice, compared to WT mice; and n.s. (not significant) for ONL thickness in both 2-month and 6-month-old *Myo1c*-KO animals, compared to WT mice). (**A**,**B**) Scale bar = 100 µm.

#### *3.6. Ultrastructural TEM Analysis Showed Shorter Photoreceptor OS in Myo1c-KO Mice*

To evaluate the structure of rod photoreceptors, ultrastructural analysis, using TEM, was performed (*n* = 6 retinal sections per eye from *n* = 8 mice per genotype and age). While the rod photoreceptor OS in the WT mice showed normal elongated morphology, they appeared slightly shorter in *Myo1c*-KO mice at two months of age (\* *p* < 0.05; Figure 6A; rod OS lengths quantified in Figure 6E). Specifically, comparing *Myo1c*-KO with WT mouse rod OS lengths at six months of age demonstrated that OS segment lengths in *Myo1c* retinas were significantly (36–45%) shorter than those of WT mice (\*\* *p* < 0.005; Figure 6B; rod OS lengths quantified in Figure 6E). Ultrastructurally, the cone OS in the *Myo1c*-KO mouse retina were shorter and had lost their typical cone shape (Figure 6C vs. Figure 6D; cone OS lengths quantified in Figure 6F), confirming the mis-shaped cone OS phenotype identified by immunohistochemistry (Figure 3B–D). These results suggest that the lack of MYO1C resulted in progressively severe opsin mislocalization (Figure 3A–D) and shorter photoreceptor OS (Figures 5 and 6), thus supporting the observed decrease in visual function by ERG (Figure 2).

**Figure 6.** Ultrastructural analysis of rods and cone photoreceptors using transmission electron microscopy (TEM): Representative TEM images of rod photoreceptors from two month (**A**) and six month (**B**) old WT and *Myo1c*-KO mice are presented. Representative images of cone photoreceptors from 2-month-old WT (**C**) and *Myo1c*-KO (**D**) mice. (**A**) Scale bar = 2 µm (**B**) Scale bar = 600 nm (**C**,**D**) Scale bar = 400 nm. Data are representative of *n* = 6 retinal sections per eye from *n* = 8 mice per genotype and timepoint. (**E**) Rod OS (*ROS*) lengths in WT animals were measured and compared to those of *Myo1c*-KO animals. (**F**) Cone OS (*COS*) lengths in WT animals were measured and compared to those of *Myo1c*-KO animals. \* *p* < 0.05; \*\* *p* < 0.005. RPE, retinal pigmented epithelium.

#### *3.7. MYO1C Directly Interacted with Rhodopsin*

Since the loss of MYO1C resulted in retinal function defects with significant alterations in the localization of opsins, we next evaluated whether MYO1C exerted this effect through a physical interaction with rhodopsin. Immunoprecipitation analysis, using WT and *Myo1c*-KO mice retinas (*n* = 6 retinas pooled from *n* = 3 animals per genotype, respectively), demonstrated that MYO1C was pulled down, using a rhodopsin antibody (Figure 7A; Co-IP). Using a baculovirus-produced purified recombinant mouse MYO1C protein in an overlay assay, we demonstrated that MYO1C directly interacted with rhodopsin, where opsin was overexpressed in HEK293 cells (transfected with pCDNA3 rod opsin). The cell lysate with overexpressed rhodopsin and control (cells transfected with empty pCDNA3 vector) was subjected to SDS PAGE and immobilized on nitrocellulose membrane, and probed with or without purified recombinant full-length MYO1C protein (Figure S6A schematic and Figure 7B) [13]. Post-incubation, the interaction of immobilized rhodopsin with MYO1C was probed using a MYO1C antibody. The immunoblot analysis of the over-layered MYO1C showed significant binding of MYO1C protein at the rhodopsin band, indicating a direct interaction between the two proteins (Figure 7B). Additionally, the direct interaction was also confirmed by ELISA, where mammalian-expressed and purified Flag rhodopsin was immobilized on individual wells of ELISA plate. The immobilized rhodopsin was then incubated with purified MYO1C protein, and the bound MYO1C was probed using MYO1C antibody (Figure S6A schematic and Figure S6B). These observations suggest that opsin is a cargo for MYO1C (arrows in Figure 7A,B).

**Figure 7.** Rhodopsin is a direct cargo for MYO1C: (**A**) Mice retinal protein lysates were isolated from *Myo1c*-KO and wildtype (WT) mice (6 retinas pooled from *n* = 3 mice per genotype) and subjected to co-immunoprecipitation analysis. MYO1C was co-immunoprecipitated with a rhodopsin antibody. (**B**) Lysate from HEK293 cells transfected with pCDNA and pCDNA rhodopsin plasmid was separated using SDS-PAGE and transferred to nitrocellulose membranes. The rhodopsin bound to nitrocellulose membrane was then incubated with 5 ug of purified recombinant active full-length MYO1C generated from a baculovirus expression system. To analyse whether MYO1C binds to immobilized rhodopsin, blots were washed and Western blotted with MYO1C antibody. A positive signal with MYO1C showed direct binding of MYO1C to rhodopsin.

#### *3.8. Genetic Deletion of Myo1c Did Not Affect Systemic Organs in Mice*

Finally, to determine if the global deletion of *Myo1c* affected other organs, we harvested major systemic organs, including the liver, heart, and kidney of 2-month-old *Myo1c*-KO and WT mice (*n* = 4 per genotype), and performed histological analyses. Notably, *Myo1c*-KO mice developed and reproduced normally with no observable histological differences between the control and *Myo1c*-KO genotypes (Figure S7A–C). To further confirm that there were no functional defects in these systemic organs, we performed an echocardiogram (heart function), quantified protein/albumin levels in urine (kidney function), and measured alanine aminotransferase (ALT) enzyme levels (liver function) in *Myo1c*-KO mice (*n* = 4 mice per individual functional analysis) and compared these values to their WT littermates (*n* = 4 mice per individual functional analysis). All of these analyses showed no pathological defects in systemic organs of *Myo1c*-KO animals when compared to the age-matched WT littermates (Figures S7A'–C' and S8). Overall, these results indicate that except for the retinal phenotypes, *Myo1c*-KO animals retained normal physiology of the systemic organs examined.

#### **4. Discussion**

The proper localization of the G-protein coupled receptor (GPCR) type II opsins to the photoreceptor OS represents a critical event in the initiation of phototransduction for visual function in vertebrates [1–9]. Our work identified for the first time an unconventional motor protein, MYO1C, as a novel regulator of both rod and cone opsin localization to the photoreceptor OS in mice. In this study, based on MYO1C localization within the IS and OS of photoreceptors, and using a whole-body *Myo1c*-KO mouse model, we functionally

identified MYO1C as a novel component of retinal physiology, which was specifically found to be involved in photoreceptor cell function. Retinal analysis of *Myo1c*-KO mice identified opsins as novel cargo for MYO1C. In the absence of MYO1C, both young and adult *Myo1c*-KO mice showed impaired opsin localization, where rhodopsin was retained in the photoreceptor IS and the cell bodies. In contrast, cone opsins showed no retention in the cell body or mislocalization to other retinal cell layers, although staining patterns revealed deformed cone OS shapes. These two phenotypes manifested as a progressive decline in visual responses in the rod ERGs and shorter photoreceptor OS lengths as *Myo1c*-KO animals aged, indicating a progressive retinal phenotype. Interestingly, localization of other OS proteins (CNGA1, arrestin, and transducin) was largely unaffected in the absence of MYO1C. The genetic deletion of *Myo1c* only affected retina, and the other systemic organs examined, including the heart, liver, and kidney, remained unaffected. Overall, our data point to a novel mechanism by which MYO1C regulates opsin localization to the photoreceptor OS, a critical event for photoreceptor function and long-term photoreceptor cell homeostasis. Our study identifies an unconventional motor protein, MYO1C, as an essential component of mammalian photoreceptors, where it plays a canonical role in promoting opsin localization and maintaining normal visual function.

#### *4.1. MYO1C and Opsin Localization to Photoreceptor OS*

*Myo1c*-KO mice exhibited rhodopsin mislocalization defects similar to those of *Rpgr*−/−, *Myo7a*Sh1 , *Rp1*−/−, *Kinesin II*−/−, and *Tulp1*−/<sup>−</sup> mutant mice [1–9]. Since MYO1C, primarily localized to photoreceptor IS and OS, is proposed to be involved in protein trafficking (among other functions in different cell types), and uses actin as a track [32,40], we hypothesized that MYO1C likely participates in proper localization of opsins to the OS of photoreceptors. This hypothesis was supported by the observation that the rod opsins were mislocalized to IS and cell bodies. Defective assembly of cone OS in *Myo1c*-KO mice suggests that this phenotype is caused by an aberrant protein localization with OS degeneration as a secondary event. The normal ultrastructure of photoreceptors in our *Myo1c*-KO mice suggests that the retinal abnormalities in these animals were not due to structural defects in photoreceptors per se, but instead were induced by aberrant motor function leading to opsin mislocalization.

#### *4.2. MYO1C Contributed to Phototransduction and Retinal Homeostasis*

The opsin molecules and other phototransduction proteins are synthesized in the cell body of the photoreceptor [53,54]. They are then transported to the distal IS [55,56] and subsequently to the OS. Little is known about these transport processes and the molecular components involved in this process [1–9]. The localization of MYO1C in the rod photoreceptors' IS and OS, and in cone OS, suggested that opsins may utilize this molecular motor for transport to the OS. The immunohistochemical analysis of *Myo1c*-KO animals indicated that while rod and cone opsins trafficked to the OS, significant mislocalization was noted for rhodopsin in the IS and cell bodies in the ONL (Figure 2). Since they represent plasma membrane structural proteins, cone opsins presumably contribute to the cone OS stability and rhodopsin to the rod OS formation and stability [7]. Hence, photoreceptor OS shortening/degeneration in *Myo1c*-KO mice may be attributed, in a large part, to the mislocalization of opsins to the IS, or a progressive reduction of opsins in the OS membrane. Notably, the pattern of opsin mislocalization observed in *Myo1c*-KO mice closely resembled the retinal phenotype observed in our previously reported *Tulp1*-KO mice [4,56], *Cnga3*−/<sup>−</sup> mice [5], *Lrat*−/<sup>−</sup> and *Rpe65*−/<sup>−</sup> mice [9,57,58], and GC1-KO mice [1,9]. Importantly, in all of these studies, photoreceptor OS were unstable, and significant degeneration was noted. However, because 85–90% of OS protein is rhodopsin [59–61], the mislocalization of other less abundant proteins cannot be ruled out in the photoreceptors of *Myo1c*-KO mice.

#### *4.3. Contributions from Other Motor Proteins in Proper Opsin Localization*

Although this study demonstrates mislocalization of opsins due to a loss of MYO1C, the majority of opsin was still correctly localized, suggesting that contribution or compensation from other myosins cannot be ruled out. Nevertheless, the contributions from MYO1C were highly significant as its genetic deletion showed specific physiological defects in mouse retinas. It is likely that some redundancy exists among molecular motors, and several known candidates might compensate for the lack of MYO1C in photoreceptor function. However, the qPCR analysis of the retinas from WT and *Myo1c*-KO mice did not suggest compensation from other family myosin 1 members (Figure S9). Interestingly, the upregulation of Myo1f in our study was unable to rescue the Myo1c retinal phenotype, suggesting that Myo1f is unable to compensate for the functional loss of Myo1c in the retina (Figure S9). However, compensation by other motor proteins, including the members of kinesin superfamily [62,63], myosin VIIa, and conventional myosin (myosin II) [64,65], which have also been detected in the RPE and retina, cannot be ruled out and need further investigation. Additionally, Myo1C has been shown to be involved in several processes involving actin, such as actin–membrane interaction (by its PIP2 binding domain), in endocytosis, and in autophagosome–lysosome fusion. Therefore, the phenotypes observed upon the loss of Myo1c could also be caused by interfering with any of these processes, either in photoreceptors or in RPE or Mueller glia cells. To further understand the involvement of MYO1C in these retinal cell types, we are currently generating conditional knockout mice, using *Best1*-Cre+ (RPE), *Rho*-Cre+ (rod photoreceptors), and *HGRP*-Cre+ (cone photoreceptors) mice. Our findings have potential clinical implications for degenerative rod and cone diseases, as mutations in MYO1C, or its interacting partners, are predicted to affect retinal health and visual function by altering opsin localization to the photoreceptor OS, a fundamental step for maintaining visual function in humans. Overall, these results support a role for MYO1C in opsin localization in the photoreceptor OS and provide evidence that defective protein transport pathways are a pathologic mechanism, responsible for OS degeneration and decreased visual function in these mice.

**Supplementary Materials:** The following are available online at https://www.mdpi.com/article/ 10.3390/cells10061322/s1.

**Author Contributions:** G.P.L. and D.N. designed the research studies and wrote the manuscript. G.P.L., D.N., M.R.B., S.H., H.-J.K., and A.A.K. edited the manuscript. G.P.L., A.K.S., M.R.B., D.N., E.A., B.R., S.W., S.H., and S.R.M. conducted experiments and acquired data. A.K.S., G.P.L., A.A.K., M.R.B., S.W., S.H., and D.N. analyzed and interpreted the data. M.R.B. and S.H. performed ERG and interpreted the data. R.M., H.-J.K., M.R.B., S.R.M., and S.H. supplied reagents and software or provided equipment for data analysis. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by the National Institute of Health (NIH) grants, R21EY025034 and R01EY030889 to G.P.L.; 2R01DK087956-06A1, R56-DK116887-01A1, and 1R03TR003038-01 to D.N.; EY027013-02 to M.R.B.; and R01EY027355 to S.H. This project was also supported in part by a SCTR-NIH/NCATS grant (5UL1TR001450) to G.P.L.

**Institutional Review Board Statement:** All mice experiments were approved by the Institutional Animal Care and Use Committee (IACUC protocol #00780; G.P.L.) of the Medical University of South Carolina and performed in compliance with ARVO Statement for the use of Animals in Ophthalmic and Vision Research.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Not applicable.

**Acknowledgments:** This work was supported by the National Institute of Health (NIH) grants R21EY025034 and R01EY030889 to G.P.L.; 2R01DK087956-06A1, R56-DK116887-01A1, and 1R03TR003038- 01 to D.N.; EY027013-02 to M.R.B.; and R01EY027355 to S.H. This project was also supported in part by a DCI research grant (019898-001) and by a SCTR-NIH/NCATS grant (5UL1TR001450) to G.P.L. The pCDNA3 rod opsin construct was a gift from Robert Lucas (Addgene plasmid #109361; http://n2t.net/addgene:109361, accessed on 25 May 2021; RRID:Addgene\_109361). The authors thank George Robertson (Keyence Microscopes) for the use of the Keyence BZ-X800 scope for imaging of semi-thin plastic and H&E sections and immunofluorescence-stained slides. The authors thank Don Rockey (MUSC) and Seok-Hyung Kim (MUSC) for recommending suitable liver function tests and/or for providing the ALT liver function kit. We also thank Rupak D. Mukherjee (MUSC) and Russell A. Norris (MUSC) for performing the ECHO tests. We also thank Linda McLoon (Department of Ophthalmology, University of Minnesota) and Barb Rohrer (Department of Ophthalmology, Medical University of South Carolina) for critical review of the manuscript.

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

#### **References**

