Müller glia-derived PRSS56 is required to sustain ocular axial growth and prevent refractive error

A mismatch between optical power and ocular axial length results in refractive errors. Uncorrected refractive errors constitute the most common cause of vision loss and second leading cause of blindness worldwide. Although the retina is known to play a critical role in regulating ocular growth and refractive development, the precise factors and mechanisms involved are poorly defined. We have previously identified a role for the secreted serine protease PRSS56 in ocular size determination and PRSS56 variants have been implicated in the etiology of both hyperopia and myopia, highlighting its importance in refractive development. Here, we use a combination of genetic mouse models to demonstrate that Prss56 mutations leading to reduced ocular size and hyperopia act via a loss of function mechanism. Using a conditional gene targeting strategy, we show that PRSS56 derived from Müller glia contributes to ocular growth, implicating a new retinal cell type in ocular size determination. Importantly, we demonstrate that persistent activity of PRSS56 is required during distinct developmental stages spanning the pre- and post-eye opening periods to ensure optimal ocular growth. Thus, our mouse data provide evidence for the existence of a molecule contributing to both the prenatal and postnatal stages of human ocular growth. Finally, we demonstrate that genetic inactivation of Prss56 rescues axial elongation in a mouse model of myopia caused by a null mutation in Egr1. Overall, our findings identify PRSS56 as a potential therapeutic target for modulating ocular growth aimed at preventing or slowing down myopia, which is reaching epidemic proportions.

A mismatch between optical power and ocular axial length results in refractive errors. Uncorrected refractive errors constitute the most common cause of vision loss and second leading cause of blindness worldwide. Although the retina is known to play a critical role in regulating ocular growth and refractive development, the precise factors and mechanisms involved are poorly defined. We have previously identified a role for the secreted serine protease PRSS56 in ocular size determination and PRSS56 variants have been implicated in the etiology of both hyperopia and myopia, highlighting its importance in refractive development. Here, we use a combination of genetic mouse models to demonstrate that Prss56 mutations leading to reduced ocular size and hyperopia act via a loss of function mechanism. Using a conditional gene targeting strategy, we show that PRSS56 derived from Müller glia contributes to ocular growth, implicating a new retinal cell type in ocular size determination. Importantly, we demonstrate that persistent activity of PRSS56 is required during distinct developmental stages spanning the pre-and post-eye opening periods to ensure optimal ocular growth. Thus, our mouse data provide evidence for the existence of a molecule contributing to both the prenatal and postnatal stages of human ocular growth. Finally, we demonstrate that genetic inactivation of Prss56 rescues axial elongation in a mouse model of myopia caused by a null mutation in Egr1. Overall, our findings identify PRSS56 as a potential therapeutic target for modulating ocular growth aimed at preventing or slowing down myopia, which is reaching epidemic proportions. PLOS  Introduction A central feature of organogenesis is the intrinsic ability to faithfully determine final organ size and shape. This is particularly important in the case of the eye where size determination is a complex and tightly coordinated process critical for achieving optimal vision. During ocular refractive development, precise ocular growth regulation is essential to ensure that the eye's axial length matches the optical focal plane, enabling focused images to fall on the retina (emmetropization), resulting in clear vision. Alterations in ocular axial length constitute the major cause of refractive errors. Shortened or elongated axial length result in focused images falling behind or in front of the retina leading to hyperopia and myopia, respectively [1,2]. Uncorrected refractive error is the most common cause of vision loss, and the second leading cause of blindness after cataract [3]. Notably, the prevalence of myopia, the most common form of refractive errors, is rising rapidly, reaching epidemic proportions in some countries [4][5][6]. This has significant public health implications as individuals with high myopia are at disproportionately increased risk of developing irreversible blinding conditions including retinal detachment, myopic macular degeneration, cataract, and glaucoma [7]. Thus, there is an urgent need to identify interventions offering the promise of modulating ocular growth, restoring healthy refractive development, and preventing associated blinding conditions.
Although environmental factors have a strong influence in determining an individual's refractive status, genetic factors account for over 50% of the variability in refractive status within populations [8][9][10]. Notably, mutations in genes involved in ocular size determination have been implicated in a subset of refractive errors with a strong developmental basis (high myopia and high hyperopia/nanophthalmos) [11][12][13][14]. More common forms of myopia are thought to have a complex multifactorial etiology resulting from an intricate interplay between multiple genetic and environmental factors [8,9], and genome-wide association studies (GWAS) have led to the identification of several genes/loci linked to myopia as well as those accounting for natural variations in refraction in the general population [15,16]. An improved understanding of the contribution of individual genes, the pathways involved and their interactions will provide the necessary framework to define the mechanisms underlying refractive development and associated errors, and identify potential targets for therapeutic intervention [5].
Ocular growth broadly comprises two distinct phases. In humans, the first phase occurs prenatally and is primarily dictated by genetic factors [14], while the second phase takes place postnatally and relies on a complex interplay between environmental and genetic factors [17]. Notably, postnatal ocular growth takes place when the eyes are developmentally open and is highly dependent on visual experience, which modulates the rate of ocular growth. Consistent with a requirement for patterned visual stimulation in ocular size determination, studies in animal models have suggested an important role for retinal neurotransmitters and neuromodulators in postnatal ocular growth and normal refractive development [18][19][20]. Signals emanating from the retina are thought to be relayed to the sclera to promote remodeling of the scleral extracellular matrix (ECM), a key step necessary to support ocular axial growth [21].
The use of genetically tractable animal models such as the mouse has recently gained popularity to study ocular size determination and emmetropization [22]. The mouse eye responds to form deprivation and lens defocus suggesting the existence of a functional emmetropization mechanism similar to that operating in humans [22][23][24], and recent technological advances have allowed precise measurement of refraction and ocular size in mice [25][26][27]. Importantly, studies in the mouse have guided the identification and validation of genes, pathways and mechanisms involved in ocular growth, refractive development and associated errors [22,[28][29][30][31][32][33][34][35]. For instance, mice with a null mutation of the transcriptional factor Egr1 recapitulate the ocular axial length elongation and refractive shift characteristic of myopia and constitute a useful animal model to dissect the mechanisms involved in ocular growth and myopia [31]. In addition, we have previously demonstrated that mice carrying a mutation in the gene coding for the secreted trypsin-like serine protease PRSS56 (Prss56 glcr4 ) exhibit reduced ocular size [36]. Mice homozygous for the Prss56 glcr4 mutation have reduced ocular axial length, primarily caused by a decrease in posterior segment size, without any gross morphological changes in ocular tissues [36]. Furthermore, we and other groups have shown that PRSS56 mutations lead to nanophthalmos (posterior microphthalmia) and extreme hyperopia characterized by significant reduction in ocular axial length in humans [36][37][38][39]. Interestingly, GWAS from multiple independent groups found an association between PRSS56 and myopia, suggesting that common PRSS56 variants may also participate in the manifestation of more complex and common forms of refractive errors [16,40]. The implication of PRSS56 variants in both hyperopia and myopia suggests a critical role for PRSS56 in ocular axial growth regulation and refractive development. The availability of Prss56 mutant mice exhibiting reduced ocular axial length and recapitulating hallmark features of human nanophthalmos constitutes a unique resource to mechanistically dissect the molecular pathways contributing to ocular size determination.
Here we have used a combination of genetic mouse models to perform a detailed molecular and cellular characterization of the role of PRSS56 in ocular size determination. Our results establish a previously unknown role of Müller glia in the regulation of ocular axial growth and refractive development. Significantly, we demonstrate a requirement for PRSS56 activity to support ocular growth during distinct developmental stages spanning the pre-and post-eye opening periods, suggesting that at least some molecules contribute to both the prenatal and postnatal stages of ocular growth in humans. Importantly, we demonstrate that genetic inactivation of Prss56 can rescue axial elongation in a mouse model of myopia caused by loss of EGR1 function. the presence of a functional catalytic domain, we have previously demonstrated intact trypsinlike serine protease activity of the recombinant PRSS56 mutant protein [36]. Similarly, human PRSS56 mutations identified by independent groups are not predicted to completely disrupt PRSS56 catalytic activity [37][38][39]. Thus, it is currently unknown whether PRSS56 mutations causing ocular size reduction act via a loss or gain of function mechanism.
To determine if the reduction in ocular size observed in Prss56 mutant mice (Prss56 glcr4/glcr4 ) results from loss of PRSS56 function, we characterized the ocular phenotypes of a mouse strain carrying a null allele of Prss56 (Prss56 Cre also referred to as Prss56in the manuscript). This strain was generated by replacing exon1 of Prss56 with a sequence coding for CRE recombinase resulting in a null allele and Prss56 promoter-driven CRE expression [41]. As the ocular size of mice heterozygous for the Prss56 Cre allele (Prss56 +/-) was indistinguishable from that of their wild-type littermates (S1A and S1B Fig), Prss56 +/mice were used as controls for most of our experiments unless otherwise specified. Slit-lamp eye examination did not reveal any difference between mice homozygous for the null allele (Prss56 -/-) and control mice ( Fig 1A). Next, we performed a detailed ocular biometric analysis to assess various parameters, including axial length, equatorial diameter, vitreous chamber depth (VCD), anterior chamber depth (ACD), and lens thickness (as seen in OCT images, Fig 1B). Biometric analysis revealed reduced ocular size in Prss56 -/mice, in which both ocular axial length and equatorial diameter were significantly smaller compared to their Prss56 +/littermates at all ages examined (postnatal day (P) 15, P25 and P60) (Fig 1B-1D, S1C-S1G Fig). The VCD of Prss56 -/eyes was significantly smaller compared to their Prss56 +/littermates ( Fig 1E). Interestingly, the ACD was marginally larger in Prss56 -/compared to control eyes (Fig 1E), and no significant difference was observed in lens thickness (S1H Fig). These findings suggest that the reduction in post-equatorial segment primarily accounts for the decreased ocular axial length observed in Prss56 -/mice. Consistent with their reduced ocular axial length, Prss56 -/mice also exhibited hyperopic refraction (Fig 1F). In addition, retinal thickness was increased in Prss56 -/eyes compared to control eyes (Figs 1G and S1F). A detailed summary of ocular biometric measurements is presented in S1 Table. Together these findings show that Prss56 -/mice have reduced ocular size, develop hyperopia, and recapitulate the ocular phenotypes observed in the previously characterized ENU-induced Prss56 glcr4 mutant mice [36], demonstrating that loss of PRSS56 function leads to reduced ocular size. We next performed the ocular biometry (A-and B-scan) on an individual exhibiting nanophthalmos caused by a homozygous PRSS56 mutation (missense variant, p.G320R) [38]. Consistent with ocular findings in Prss56 -/mice, the human PRSS56 mutation led to a substantial reduction in VCD compared to a normal emmetropic eye (Fig 1H and 1I and Table 1), suggesting that PRSS56 mutations leading to nanophthalmos and extreme hyperopia in humans likely act via a loss of function mechanism.
Prss56 ocular expression is restricted to the neural retina and is first detected in a pool of late retinal progenitor cells. As a first step in addressing the role of PRSS56 in ocular size determination, we performed a lineage tracing experiment to determine the identity and fate of cells expressing Prss56 in the developing eye. To this end, Prss56 Cre/+ mice expressing CRE recombinase under the control of the Prss56 promoter were bred to the inducible R26 tdTomato reporter mice that express tdTomato in presence of CRE (Fig 2A). In the resulting offspring, the tdTomato reporter gene will be expressed in Prss56-expressing cells and their derivatives, thereby allowing lineage tracing. Assessment of tdTomato fluorescence on ocular sections revealed that ocular Prss56 expression is restricted to the retina during both embryonic and postnatal ages. Prss56 expression is first detected in sparse cells in the outer neuroblastic layer (ONBL) at embryonic day (E) 16.5 ( Fig 2B). The number of tdTomato positive retinal cells increases over time as shown at E18.5 and P2 (Fig 2C and 2D). As the retina matures and cells differentiate, intense tdTomato expression is detected in the inner nuclear layer (INL), and Role of PRSS56 in refractive development weaker signal in the photoreceptor layer (Fig 2E and 2F). Of note, tdTomato-positive cells are enriched in the peripheral retina relative to the central retina (Figs 2F, S2A and S2C) and no tdTomato signal was detected in other ocular tissues, including the sclera, choroid, cornea, lens, ciliary body and iridocorneal angle (S2C and S2D Fig).
The spatio-temporal pattern of tdTomato labeling in the retina is suggestive of Prss56 being expressed in late retinal progenitor cells (RPCs) and its derivatives. To confirm this, we performed a series of immunolabeling studies. Supporting expression of Prss56 by RPCs, we show that tdTomato positive cells express Ki67, a marker of dividing cells, in P0 retina ( Fig 2G). In addition, we found that tdTomato positive cells located in the INL express the Müller glia markers vimentin and SOX2 (Figs S3A and 3A), as well as PKCα, a marker of bipolar cells ( Fig  3B), demonstrating that tdTomato also labels late RPC derivatives. The reporter line used in our experiments showed an inherent variability in tdTomato fluorescence intensity between various retinal cell types. While intense tdTomato signal was detected in SOX2 and vimentin immunopositive Müller cells, weaker tdTomato signal was observed in PKCα immunopositive bipolar cells (Fig 3B, arrow). Using flow cytometry, we further demonstrate that Müller glial cells (immunolabeled for glutamate synthetase (GS)) exhibit more intense tdTomato signal, while rod photoreceptors (immunolabeled for Rhodopsin) exhibit weaker tdTomato signal ( Fig 3C). Together, these findings demonstrate that Prss56 is expressed in late RPCs fated to give rise to subsets of rod photoreceptor, bipolar and Müller cells.
Prss56 expression is restricted to a subset of Müller glia following retinal cell differentiation. The reporter (tdTomato) is not only expressed in cells actively transcribing Prss56 but also in cells derived from parent cells expressing Prss56. Therefore, we performed in situ hybridization to determine the expression pattern of Prss56 at critical time points during and following complete maturation of retinal cells. We show that Prss56 expression is selectively detected in the INL of the retina during (P10) and following (P15) maturation of retinal cell types, and that Prss56 expression is enriched in the peripheral retina relative to the central retina (S3B and S3C Fig). Dual immunofluorescent labeling using antisense probes for GS (glial cell marker) and Prss56, revealed colocalization of Prss56 (red) and GS (green) expression in the INL of adult retina, while no signal was detected using the sense probes ( Fig 4A). Interestingly, in situ hybridization revealed a substantial increase in the number of Prss56 expressing cells in Prss56 mutant retina (Prss56 glcr4/glcr4 , ENU-induced mutation with C-terminal truncation) compared to control retina (Prss56 glcr4/+ ), suggesting increased Prss56 expression in mutant retina (Fig 4A). Quantitative PCR analysis further demonstrated a significant upregulation in the levels of Prss56 mRNA in mutant retina (Prss56 glcr4/glcr4 ) compared to their control counterparts from P15 onwards ( Fig 4B). Collectively, these findings demonstrate that Prss56 is selectively expressed by a subset of Müller glia following retinal maturation, and suggest the existence of a feedback regulatory loop modulating retinal Prss56 expression in response to alteration in ocular size.
Early requirement for PRSS56 during ocular axial growth. Having demonstrated a role for PRSS56 in ocular growth and its expression pattern during retinal development, we next assessed the temporal requirement for PRSS56 activity during ocular size determination. Using spectral-domain optical coherence tomography (SD-OCT), we detected a slight but significant reduction in axial length in Prss56 mutant eyes compared to control eyes as early as P6 ( Fig 5A). Moreover, Prss56 mutant retina was significantly thicker than control retina at P17 and 2 months (Figs 5B and S1F and S1 Table). Histological analysis of Prss56 mutant retina revealed no overt morphological defect with the exception of an increased number of nuclear stacks in both the outer and inner nuclear layers, which could contribute to the observed increase in retinal thickness (Fig 5C and 5D). PRSS56 derived from Müller glia contributes to ocular axial growth. Ocular size reduction in Prss56 mutant mice is detected as early as P6 when a significant pool of undifferentiated late RPCs is still present in the retina. Since Prss56 is expressed by late RPCs, it raises the question of whether PRSS56 derived from either committed or differentiated Müller glia has any role to play in ocular axial growth. To address this, we generated a conditional Prss56 mutant allele by flanking exons 2 to 4 with LoxP sites (Prss56 F ) that will give rise to a catalytically inactive PRSS56 protein following CRE-mediated excision (S4 Fig). Following validation of the conditional allele (S5 Fig), we ablated Prss56 from Müller glia using the Rax-Cre ER T2 which specifically express CRE recombinase in Müller cells following tamoxifen induction [42]. Selective inactivation of Prss56 in Müller glia was induced by tamoxifen injection at P8, a time point preceding the developmental stage at which the majority of RPCs have differentiated into their respective retinal cell types, including Müller cells (~P10). Based on slit lamp examination, eyes from tamoxifen-injected Prss56 F/F ; Rax-Cre ER T2 and control Prss56 F/+ ; Rax-Cre ER T2 mice were indistinguishable ( Fig 6A). Ocular biometric analysis revealed a significant reduction in ocular axial length and VCD and an increase in retinal thickness in tamoxifen injected Prss56 F/F ; Rax-Cre ER T2 mice compared to both tamoxifen injected and uninjected controls (Fig 6B-6E). Similar to what we observed in the Prss56 glcr4/glcr4 retina, we detected a significant upregulation of retinal Prss56 mRNA levels following conditional ablation of Prss56 from Müller glia compared to control eyes ( Fig 6F). These findings suggest that PRSS56 derived from differentiated Müller cells contributes to ocular axial growth.
Next, we assessed if loss of PRSS56 function affects the structural or morphological organization of Müller glia. To this end, we took advantage of our Prss56 Cre ; R26 tdTomato reporter line, which specifically labels the subset of Müller glia derived from Prss56-expressing RPCs and performed a morphological analysis of retinal sections and whole-mounts at P6. This specific time point was selected for three main reasons: 1) it corresponds to the developmental time point when Müller glia committed progenitors begin to express markers of mature Müller glia [43], 2) it marks the initiation of endfeet formation and precedes complex endfeet elaboration in the inner limiting membrane (ILM) [44], and 3) it coincides  (Fig 6G and 6H). We found that the proportion of tdTomato positive Müller glia endfeet exhibiting a more elaborate morphology was significantly greater in Prss56 mutant compared to control retinal whole mounts (59.26 ± 8.04% in mutant vs 22.72 ± 5.33% in control, p<0.001). These findings suggest premature branching and maturation of Müller glia endfeet in Prss56 mutant mice and raise the possibility that altered structural organization of Müller glia may contribute to the ocular size reduction observed in Prss56 mutant mice (Fig 6G-6I).  Sustained PRSS56 activity is required for ocular axial growth. Our results show that Prss56 is actively expressed in both the developing and adult retina, raising the possibility that PRSS56 might regulate ocular growth during distinct development stages spanning the pre and post-eye opening periods. To determine the temporal requirements of PRSS56 in modulating ocular axial growth, we bred conditional Prss56 mutant mice to the inducible ubiquitous Ubc-Cre line to ablate Prss56 at distinct stages of ocular development. Unlike the human eye, the mouse eye remains closed during the early postnatal period until the eyelids open around P13 ( Fig 7A). Therefore, in this regard, postnatal time points preceding P13 in the mouse are analogous to late prenatal stages of human ocular growth when the eyes are closed. Mice carrying the Prss56 F allele and the inducible Ubc-Cre transgene (Prss56 F/F ; Ubc-Cre ER T2 and control: Prss56 F/ + ; Ubc-Cre ER T2 ) were injected with tamoxifen at P6 or P8 (stages at which the eyes are closed) and compared to uninjected and injected control groups. Ablation of Prss56 at both time points caused a significant decrease in ocular axial length and VCD compared to control groups (measured at P17; Fig 7B-7D). Administration of tamoxifen at P6 caused a greater reduction in ocular axial length (compare P6 to P8 in Fig 7B), suggesting a continuous requirement for PRSS56 activity during ocular growth. Next, we injected tamoxifen at P13 to determine whether PRSS56 function is required after eye opening. Ablation of Prss56 following tamoxifen injection at P13 caused a modest but significant decrease in ocular axial length compared to control groups (measured at P30 and P45, Fig 7E). The decrease in ocular size was accompanied by a significant decrease in VCD and increase in retinal thickness (Fig 7F and 7G).
To test for a potential role of PRSS56 at stages when eyes are responsive to patterned visual stimulation [45], we injected Prss56 F/F ; Ubc-Cre ER T2 and control mice (Prss56 F/+ ; Ubc-Cre ER T2 ) with tamoxifen at P18 and performed ocular biometry at 3 different time points (P30, P45, and P60). Although the axial length of Prss56 F/F ; Ubc-Cre ER T2 eyes tended to be slightly shorter compared to that of control eyes, the difference was not statistically significant at any of the ages examined (P = 0.056 at P60 measurement, Fig 7H). In contrast, a significant decrease in VCD was detected in Prss56 F/F ; Ubc-Cre ER T2 eyes compared to control eyes at all ages examined (Fig 7J). Notably, the decrease in VCD following conditional Prss56 ablation was progressive and age-dependent. Since we did not observe a significant reduction in ocular axial length, we also measured other ocular layers contributing to ocular axial length, including the lens and anterior chamber. While the lens thickness was comparable between the Prss56 ablated and control eyes (S6A Fig), the ACD was marginally increased following Prss56 ablation (S6B Fig), adding to the difficulty of detecting small changes in ocular axial length. Additionally, we assessed ocular refraction at 3 months of age following tamoxifen injection at P18 and found that Prss56 F/F ; Ubc-Cre ER T2 mice are hyperopic compared to control mice ( Fig  7M). Furthermore, a significant upregulation of retinal Prss56 mRNA levels was also observed following tamoxifen injection at P13 and P18 (Fig 7L). A detailed summary of ocular biometric injection at P8. Prss56 F/F ; Rax-Cre ERT2 eyes were indistinguishable from control Prss56 F/+ ; Rax-Cre ERT2 eyes at 2 months of age. (B, C) Following tamoxifen injection (TAM +; horizontal axis) at P8, Prss56 F/F Rax-Cre ERT2 mice display a significant decrease in ocular axial length (B) and increase in retinal thickness (C) compared to control eyes (Prss56 F/+ ; Rax-Cre ERT2 or Prss56 F/F mice without Rax-Cre ERT2 ), N = 6 to 8 per group. (D) Following tamoxifen injection at P8, a significant reduction in VCD was detected in Prss56 F/F ; Rax-Cre ERT2 eyes compared to control Prss56 F/+ ; Rax-Cre ERT2 eyes, N = 6 per group. Ocular axial length, retinal thickness, and VCD in uninjected Prss56 F/F ; Rax-Cre ERT2   measurements, sample size, and statistical significance following conditional ablation of Prss56 using Rax-Cre and Ubc-Cre is presented in S2 Table. Collectively, our findings indicate that sustained PRSS56 activity is required throughout distinct stages of ocular development to regulate axial growth. Significantly, they demonstrate that PRSS56 is a molecular factor that operates during both the pre-and post-eye opening developmental stages of ocular axial growth.
Genetic inactivation of PRSS56 rescues ocular axial elongation in a mouse model of myopia. Our data shows that lack of PRSS56 activity cause ocular size reduction. To explore the translational potential of this finding, we tested the hypothesis that inactivation of PRSS56-mediated pathway(s) could slowdown axial elongation linked to myopia. To this end, we performed a genetic study to determine the effect of Prss56 ablation in a mouse model of myopia caused by a null mutation in Egr1. EGR1 is a known regulator of ocular axial growth, and Egr1 expression is up-or down-regulated in animal models with experimentally-induced reduction or increase in ocular axial length, respectively [46,47] Importantly, mice lacking EGR1 (Egr1 -/-) recapitulate the characteristic hallmarks of myopia including increased ocular axial length and a myopic refractive shift [31]. Because ocular biometric parameters were indistinguishable between Prss56 +/-; Egr1 +/and Prss56 +/+ ; Egr1 +/+ mice, we used Prss56 +/-; Egr1 +/mice as controls for our experiments (S7 Fig). The ocular axial elongation resulting from Egr1 deficiency was detected as early as P10, but was more prominent at later time points (compare Prss56 +/-; Egr1 -/to Prss56 +/-; Egr1 +/mice, Fig 8A and 8B). Importantly, a concomitant increase in VCD, but no significant change in retinal thickness was observed at all ages examined in Egr1 deficient mice compared to control mice (Fig 8D and 8E). As expected, Prss56 mutant eyes (Prss56 -/-; Egr1 +/-) exhibit a significant reduction in axial length and VCD compared to control eyes (Prss56 +/-; Egr1 +/-, Fig 8B-8D). Interestingly, the axial length and VCD of double mutant (Prss56 -/-; Egr1 -/-) eyes were not significantly different from that of the control eyes, (Fig 8A, 8B and 8D). Consistent with our ocular size findings, refraction was significantly rescued in double mutants compared to single Egr1 or Prss56 mutants, exhibiting myopia or hyperopia, respectively (Fig 8C). Interestingly, despite the ocular axial length of double mutants being comparable to that of control mice, they exhibit a significantly thicker To determine the temporal window critical for the PRSS56-mediated effect on ocular axial growth, Prss56 conditional mutant mice (Prss56 F/F ) were crossed to mice expressing the ubiquitous inducible Ubc-Cre recombinase (Ubc-Cre ERT2 ). (A) Schematic of tamoxifen treatment at distinct developmental stages preceding and following the opening of the eyes. Tamoxifen injection at two different time points, P6 and P8, was performed to ablate Prss56 after the earliest detectable effect of mutant Prss56 on ocular axial length. (B-D) OCT-based ocular biometry demonstrates that following tamoxifen injection, Prss56 F/F ; Ubc-Cre ERT2 mice display a significantly reduced ocular axial length (B) and increased retinal thickness (C) and a significant decrease in VCD (D) compared to the control Prss56 F/+ ; Ubc-Cre ERT2 mice (measured at P17). The ocular axial length, retinal thickness, and VCD of uninjected Prss56 F/F ; Ubc-Cre ERT2 and Prss56 F/+ ; Ubc-Cre ERT2 mice were indistinguishable. Administration of tamoxifen at P6 caused a greater decrease in ocular axial length compared to administration at P8 suggesting a requirement for continuous PRSS56 activity during ocular development to sustain normal ocular growth, N = 6 to 8 per group for A and B. (E-G) OCT measurements demonstrate that following tamoxifen injection at P13 (a time point when the eyes are open), Prss56 F/F ; Ubc-Cre ERT2 mice display a slight but significant decrease in ocular axial length (E) and increase in retinal thickness (F) at P30 and P45. Reduced ocular axial length was associated with a significant decrease in VCD in Prss56 F/F ; Ubc-Cre ERT2 mice compared to Prss56 F/+ ; Ubc-Cre ERT2 and uninjected controls at P30 and P45, N = 5 to 10 per group (G). (H-J) Ocular biometry following tamoxifen injection at the beginning of a critical emmetropization period (P18) shows that the ocular axial length is not significantly different between the Prss56 F/F ; Ubc-Cre ERT2 and control mice at any of the three ages examined (P30, P45, and P60). However, Prss56 F/F ; Ubc-Cre ERT2 mice display a slightly thicker retina and significantly reduced VCD compared to the control groups. (K) Following Prss56 ablation at P13 and P18, the eyes display a decrease in the combined value of retinal thickness and VCD. (L) qPCR analysis revealed elevated Prss56 mRNA levels in Prss56 F/F ; Ubc-Cre ERT2 retina compared to Prss56 F/+ ; Ubc-Cre ERT2 retina following tamoxifen injection at P8, P13, or P18 (shown are data from mice harvested at P17 and P60, respectively). (M) Following Prss56 ablation at P18, the eyes display a hyperopic shift in refraction compared to control eyes at 3 months (N = 6 per group). Values are presented as mean ± SD (or mean ± SEM in L); Ã p<0.05, ÃÃ p<0.01, ÃÃÃ p<0.001, t-test.
https://doi.org/10.1371/journal.pgen.1007244.g007  (Fig 8E). A detailed summary of ocular biometric measurements from Egr1 and Prss56 single and double mutants, sample size, and their statistical significance are presented in S3 Table. These findings show that Prss56 inactivation rescues axial elongation/myopia resulting from EGR1 deficiency.

Discussion
The molecular and cellular mechanisms involved in ocular size regulation and refractive development are poorly understood. Here, we have used a combination of genetic mouse models to elucidate the role of PRSS56 in ocular axial growth. We show that loss of PRSS56 function causes ocular axial length reduction and hyperopia. Moreover, utilizing a lineage tracing strategy and a combination of molecular approaches, we demonstrate that Prss56 ocular expression is first detected in a pool of late RPCs and then in a subset of Müller glia following retinal cell differentiation. Importantly, our findings demonstrate that PRSS56 derived from Müller glia contributes to ocular axial length elongation, uncovering a previously unrecognized role for Müller glia in ocular growth. Furthermore, we show that continuous PRSS56 activity is required to sustain ocular growth throughout distinct stages of ocular development spanning the pre-and post-eye opening periods. Thus, findings from our mouse models suggest that at least some of the factors guiding ocular growth are conserved across the prenatal and postnatal stages of ocular development.
Impaired processing of PRSS56's endogenous substrate(s) potentially contributes to ocular size reduction. Using mice homozygous for a null allele of Prss56, we demonstrate that loss of PRSS56 function causes a reduction in ocular size and hyperopia (Fig 1A-1F), suggesting that impaired processing of PRSS56 protease's endogenous substrate(s) may underlie ocular size reduction. Since the loss of PRSS56 function leads to reduced ocular size, PRSS56 variants associated with myopia (identified by GWAS) likely act in an opposite manner, i.e. via a gain of function mechanism, to induce ocular axial elongation [16,40,48]. Our current efforts are geared towards identifying PRSS56 endogenous substrates to gain further mechanistic insight into factors contributing to ocular growth.
Identification of a novel subpopulation of Müller glia. Our results demonstrate that Prss56 ocular expression is first detected embryonically in a pool of late RPCs and later in Müller glia following retinal cell differentiation (Figs 2 and 3). Müller glia has traditionally been viewed as a homogeneous cell population. Although previous studies have reported heterogeneity in gene expression among individual Müller cells, evidence supporting Müller glia being a heterogeneous group of cells is very limited [49,50]. Here, we have identified a unique subset of Müller glia that expresses Prss56 and are enriched in the peripheral region of the retina (Figs 2, S3B and S3C). It is tempting to speculate that the heterogeneity in the molecular profile and distribution of Müller cells may have evolved to support their diverse retinal functions.
Identification of a novel role of Müller glia in ocular axial growth. In this study, we used conditional Prss56 mutant mice to demonstrate that PRSS56 derived from differentiated Müller glia contributes to ocular axial growth (Fig 6A-6F). However, ocular size reduction in Prss56 mutant mice is detected before complete differentiation of Müller glia, suggesting that both glial committed progenitor cells and differentiated Müller cells constitute important sources of PRSS56 to support ocular growth. Being a secreted protease, PRSS56 could modulate the extracellular milieu of the retina by processing its endogenous substrate(s), such as retinal or neighboring ECM components or cell surface receptors. Notably, PRSS56 was recently shown to be localized in the retinal ILM (the retinal basement membrane) [52]. Thus, it is possible that loss of PRSS56 activity could alter the composition of the ILM, which in turn could lead to the premature elaboration of Müller glia endfeet observed in Prss56 mutant retina ( Fig  6G-6I). Defective ILM has been linked to abnormal ocular growth and mutations in genes encoding ECM proteins present in the ILM have been implicated in axial elongation and myopia [53]. Henceforth, the ILM has been proposed as a critical structure providing the mechanical strength regulating the pressure within the vitreous chamber, which is subsequently transferred to the sclera of the developing eye to modulate ocular growth [53]. It is tempting to speculate that failure to generate sufficient tension due to altered ILM composition may underlie ocular size reduction resulting from loss of PRSS56 function. Alternatively, as reduced VCD was the most conspicuous ocular manifestation detected in mice and a patient with mutant PRSS56 (Fig 1E and 1I), a failure to maintain normal vitreous volume may account for the altered ocular axial length. Müller glial cells have been suggested to be important regulators of intraretinal water flow into the vitreous cavity [54][55][56]. Thus, another possible mechanism by which loss of PRSS56 function may contribute to ocular size reduction is by failure to regulate fluid flow and maintain normal vitreous volume.
Findings from our mouse study suggest a role for PRSS56 in pre-and post-natal human ocular growth. Our results suggest that continuous PRSS56 activity is required for normal ocular growth. Developmental stages preceding and following eye opening in the mouse are analogous to the prenatal and postnatal stages (when the eyes are exposed to visual experience) of human ocular growth, respectively. Thus, the impact of PRSS56 mutations on human ocular growth is likely to begin prenatally before patterned visual activity modulates the rate of ocular growth. These findings are in agreement with previous suggestions that genetic alterations leading to nanophthalmos interfere with prenatal ocular growth [14]. Importantly, ablation of Prss56 following eye opening in the mouse (tamoxifen injection at P13 and P18) resulted in a modest reduction in ocular dimension and a hyperopic shift in refraction (Fig 7E, 7J and 7M), suggesting that the role of PRSS56 in ocular growth regulation may extend beyond the prenatal developmental window in humans. These findings establish that persistent PRSS56 activity is required during distinct stages of ocular development to support ocular growth and normal refractive development. Although, visual input plays a critical role in the regulation of postnatal ocular growth, both vision-adjusted and unadjusted ocular growth are likely to operate in concert to determine ocular size [17,57]. Here, we have discovered that PRSS56 contributes to ocular growth during a window when the eyes are subject to patterned visual stimulation ( Fig  7E-7J). The specific contribution of PRSS56 in vision-adjusted ocular growth, if any, is yet to be determined. However, it is plausible that the fundamental processes controlling ocular growth are conserved throughout the various stages of ocular development [17], with an additional level of regulation during the vision-adjusted phase where the rate of ocular growth is dependent on the refractive status of the eye. Future studies aimed at investigating the specific role of PRSS56 in emmetropization will utilize experimental paradigms to examine the effect of stage-specific ablation of Prss56 on vision-guided ocular growth.
Identification of PRSS56 as a potential target for therapeutic intervention to slowdown myopia progression. The most common forms of myopia result from aberrant postnatal ocular growth. Given the role of PRSS56 in supporting ocular growth at stages following opening of the eyelid, targeting PRSS56 appears a viable therapeutic strategy to slowdown axial elongation underlying myopia. Our genetic experiment suggests that opposing effects of Prss56 and Egr1 mutations are canceled in double mutant animals, causing their eyes to attain a size that is indistinguishable from that of control eyes. Thus, it is likely that parallel reciprocal pathways driven by PRSS56 and EGR1 regulate ocular axial growth. Interestingly, EGR1 is detected in cells of the retinal INL, including amacrine cells [47], raising the possibility that PRSS56 and EGR1 may act through two different retinal cell types to regulate ocular size. Overall, these results demonstrate that Prss56 inactivation rescues axial elongation/myopia resulting from EGR1 deficiency, and thereby establish PRSS56 as a potential therapeutic target for interventions aimed at preventing myopia.
In summary, we demonstrate that loss of PRSS56 function leads to ocular size reduction and hyperopia and identify a novel role for Müller glia in ocular axial growth. Future studies will aim at determining whether PRSS56 regulate ocular growth by directly influencing Müller glia function. Our findings also show that PRSS56 activity is required at distinct developmental stages spanning the pre-and post-eye opening periods. To the best of our knowledge, this is the first study documenting the existence of a genetic factor whose activity is required continuously through various stages of ocular development to support normal refractive development. Finally, we demonstrate that Prss56 inactivation rescues axial elongation in a mouse model of myopia. Since PRSS56 variants have been implicated in both human hyperopia and myopia, our findings have direct relevance to human ocular refractive development. Given the steep rise in the prevalence of myopia, there is an urgent public health need to identify therapeutic targets to prevent or slow down myopia. This study demonstrates that interventions aimed at regulating PRSS56 activity have the potential to modulate ocular growth, restore healthy refractive development and prevent associated blinding conditions.

Ethics statement
All experiments were conducted in accordance with the Association for Research in Vision and Ophthalmology's statement on the use of animals in ophthalmic research. Mouse studies were performed in compliance with protocols approved by the Institutional Animal Care and Use Committee at University of California San Francisco (Approval numbers: AN153083 and AN120008). Animals were given access to food and water ad libitum and housed under controlled conditions including a 12-h light/dark cycle in accordance with the National Institutes of Health guidelines. For some of the experiments mice were anesthetized with ketamine/xylazine (100 mg/kg and 5mg/kg, respectively).
Egr1 mutant mice: C57BL/6.Egr1 tm1Jmi /J, the targeted mutation by insertion of a PGK-neo cassette introduces stop codon resulting in protein truncation upstream of the DNA-binding domain [58].
Reporter line. Gt(ROSA)26Sor tm14(CAG-tdTomato)Hze : Cre reporter mice harbor a LoxPflanked STOP cassette preventing transcription of a CAG promoter-driven tdTomato [59]. The targeted mutation was inserted into the Gt(ROSA)26Sor locus. TdTomato expression is dependent on the presence of CRE recombinase.
No gender effect has been observed on ocular size reduction resulting from Prss56 mutations in our previous and current studies. Both males and females were included in all experiments described in this manuscript. For most experiments, littermates were used for comparisons between genotypes. PCR genotyping was performed on genomic DNA obtained from tail biopsies digested with Proteinase K (Sigma, St. Louis, MO, USA) using the primers indicated in S4 Table. Clinical slit lamp examination Ocular anterior segment examinations were performed on 1-5 months old mutant mice and control littermates using a slit lamp biomicroscope (Topcon SL-D7; Topcon Medical Systems, Oakland, NJ, USA) attached to a digital SLR camera (Nikon D200; Nikon, Melville, NY, USA). Observers were masked to mouse genotypes while evaluating clinical phenotypes. Phenotypic evaluation included considerations for iris structure, pupillary abnormalities, cataracts and the overall dimensions of the anterior chamber.

Ocular biometry
Ocular biometry was performed using optical coherence tomography or a digital Vernier caliper. Envisu R4300 spectral-domain optical coherence tomography (SD-OCT, Leica/Bioptigen Inc., Research Triangle Park, NC, USA) was employed to measure the ocular axial length, retinal thickness, vitreous chamber depth (VCD) and anterior chamber depth (ACD) as previously described with minor modifications [25]. Briefly, mice were anesthetized with ketamine/xylazine (100 mg/kg and 5mg/kg, respectively; intraperitoneal) and their eyes dilated before placing the animal in a cylindrical holder. The eye was hydrated with Genteal (Alcon, Fort Worth, TX, USA) and positioned in front of the OCT light source. Correct alignment of the eye was achieved by placing the Purkinje image in the center of the pupil. The images were acquired in rectangular volume and radial volume scans to capture the retinal thickness and axial length measurements, respectively. The axial length was calculated by measuring the distance from the corneal surface to the RPE/choroid interface. The distance between the innermost layer of the retina and the lens was used to calculate the VCD. ACD is the distance between the innermost cornea layer and lens. For biometric analyses performed before eye opening (before P13), mice were anesthetized and their eyelid carefully slit open using fine scissors. Thereafter, the eye was gently protruded using a Q-tip and aligned to the light source as described above. Digital Vernier caliper (Fowler Ultra-Cal Mark III) was used to measure the equatorial diameter as described previously [36]. Eyes were enucleated and magnified under a dissecting scope. The Vernier caliper was positioned along the nasal and temporal plane at a point of maximum diameter. Ocular biometry was performed on both the left and right eyes of a given mouse. S1-S3 Tables summarize the details about sample size, body weight, and biometric measurements of mice in each experimental cohort. To minimize the possible effect of body weight on ocular size, we ensured that body weight of littermates was within a narrow range in each of the comparative groups.

Refraction measurement
Ocular refractions were acquired using an automated infrared photorefractor as described previously with some minor modifications [28]. Refraction was measured following treatment of mouse eyes with cyclopentolate (Alcon, Fort Worth, Tx, USA) to temporarily paralyze the ciliary body (cycloplegic refraction). Mice were placed on a pedestal with their eyes facing the photorefractor. The photorefractor was maintained at a distance and maneuvered to obtain a clear focused image of the eye. The photorefractor registers a successful refraction measurement only when the Purkinje image is positioned in the center of the pupil as detected and marked by a green LED flash. Centering of the Purkinje image ensures the infrared rays pass along the optical axis. In a typical recording, around 30-50 refraction measurements are acquired, which are then used to calculate the mean OD. Refraction measurements were performed on both the left and right eyes of a given mouse. A total of at least 6 eyes per experimental group were used for refraction measurements and experimental mice of both sexes were used for refraction measurements.

Quantitative real-time polymerase chain reaction (qRT-PCR)
Eyes were enucleated and retinas were immediately dissected. Total RNA was isolated from retina using Qiagen RNeasy Mini Kit with on-column DNase I treatment (Qiagen, Valencia, CA, USA) and reverse transcribed using iScript cDNA Synthesis Kit (Bio-Rad, Hercules, CA, USA). qPCR was performed on a Bio-Rad C1000 Thermal Cycler/CF96 Real-Time System using SsoAdvanced TM SYBR Green 1 Supermix (Bio-Rad, Hercules, CA, USA), and primer sets listed in S5 Table. Briefly, 15ng of cDNA and 0.25 μM primers were used per reaction in a final volume of 10 μl of Supermix. Each cycle consisted of denaturation at 95˚C for 5s, followed by annealing and extension at 60˚C for 25s. Each reaction was run as technical duplicates and a minimum of 4 biological replicates was used per group. The relative expression level of each gene was normalized to housekeeping genes (Actb, Hprt1, and/or Mapk1) and analyzed using the CFX manager software (Bio-Rad, Hercules, CA, USA).

In situ hybridization
Mice were transcardially perfused with ice-cold RNase-free PBS followed by 4% PFA (in RNase-free PBS). Enucleated eyes were post-fixed in RNAse-free 4% PFA, cryoprotected in 20% sucrose, and embedded in OCT and sectioned within 24 hours for in situ hybridization. QuantiGene View RNA (Affymetrix, Santa Clara, CA, USA) in situ hybridization assay was performed according to the manufacturer protocol. Briefly, 12μm cryosections were fixed overnight in 4% PFA, dehydrated through a graded series of ethanol, were subjected to 2X protease digestion for 10 minutes, postfixed with 4% PFA and hybridized with probe sets against the gene of interest for 3 hours at 40˚C using a ThermoBrite system (Abbott Molecular, Des Plaines, IL, USA). Cryosections were then washed and subject to signal amplification and detection using fast red substrate, counterstained and mounted for subsequent imaging. For dual fluorescent in situ hybridization, Digoxigenin-and Fluorescein-labeled riboprobes were synthesized from full-length cDNA clones (MGC Mouse glutamine synthetase cDNA Clone Id:4224865). The hybridized mRNA was detected using the TSA-FITC/ TSA-CY5 Tyramide Signal Amplification System (PerkinElmer, Waltham, MA, USA).

Histology
Mice were euthanized and eyes enucleated and immediately immersed in cold fixative (1% PFA, 2% glutaraldehyde, and 0.1 M cacodylate buffer) for 24 hours, after which they were transferred to cold 0.1 M cacodylate buffer solution for an additional 24 hours. Samples were embedded in glycol methacrylate, and serial sagittal sections (2μm) passing through the optic nerve were cut and stained with hematoxylin and eosin (H&E).

Retinal whole mount preparation and endfeet assessment
Retina was dissected from Prss56 Cre/+ ; R26 tdTomato (control) or Prss56 Cre/glcr4 ; R26 tdTomato (mutant) mice and four radial incisions made and mounted on a slide. tdTomato positive terminal endings of Müller glia projection (endfeet) were visualized and images captured using a Confocal mircroscope (Carl Zeiss LSM700). Eight equivalent areas of the retina were consistently selected for each whole mount. Müller glia endfeet were classified into two groups based on their morphology: 1) Müller glia endfeet showing simple cohesive arrangement and occupying smaller area of the retina. 2) Müller glia endfeet exhibiting more spread-out morphology and covering a larger area, suggestive of increased branching and elaboration. Two independent observers masked to genotypes manually quantified the relative distribution of the two types of endfeet.

Microscopy
Bright-field images were captured using AxioVision software and an AxioImager M1 microscope equipped with an AxioCam ICc3 digital camera (Carl ZeissMicroscopy, LLC, Germany). Fluorescent images were acquired using AxioImager M1 microscope equipped with an MRm digital camera and AxioVision software, with an LSM700 confocal microscope and a Zen software (Carl Zeiss Microscopy, LLC, Germany). Amira software was used for 3D visualization and analysis.

Human ocular biometry
We obtained illustrative biometric data from a nanophthalmic patient affected with a homozygous missense variant (p.G320R) in PRSS56 previously identified by one of the authors (ACO) [38], and also a representative normal volunteer for comparison. Approval for this study was obtained from the Research Ethics Board of the Nova Scotia Health Authority, Halifax, Nova Scotia, Canada. In brief, ocular dimensions measured via ultrasound biomicroscopy A-and Bscan techniques revealed a very small globe bilaterally featuring crystalline lenses that were normally positioned, but large in size relative to that of the eye. The choroid was also observed to be diffusely thickened.

Statistical analysis
Statistical comparisons between control and mutant samples were performed by a two-tailed unpaired Student's t-test using Prism version 6.0f software. p values of <0.05 were considered significant. Power for a two-tailed two-sample t-tests was calculated using a range of means and standard deviation values of axial length that could be reasonably expected based on published data and our initial assessment [26]. Although the difference between the two group means and within-group standard deviation are statistically independent parameters, in many biological data sets they show various levels of collinearity. Therefore, for small, medium and large μ1-μ 2 values, we based our power calculation on correspondingly increasing expectations of SD. The detection of differences with 80% power is possible in these scenarios with sample sizes 7. For example, with respect to axial length, a mean difference of 50 μm, SD = ± 30, the effect size is 1.67 requiring a sample size of 6. Where logistically possible, and partly as a precaution against the possibility of some failed experiments, we collected data from more samples. Number of eyes, mean and standard deviation of all measurements in each group are presented in S1-S3 Tables. . PCR reactions using the F1R1 primer pair gives a product that is about 34 bp longer in mice carrying the Prss56 F allele compared to wild-type mice.

Supporting information
In the presence of Cre recombinase, deletion of exon 3 and 4 from the Prss56 F allele gives no PCR product. PCR reactions using the F2R2 primer pair lead to product sizes of 280bp for the wild-type allele and 350bp for the Prss56 F allele and no PCR product for the Prss56 F allele following Cre activation. PCR reactions using the primer pair F1R2 give rise to product sizes of 820bp and 920bp for the wild-type and Prss56 F alleles, respectively. Additionally, primer pair F1R2 confirmed excision of exons 3 and 4 from the Prss56 F allele following CRE activation as shown by the presence of a shorter PCR product size of 290bp.