Identification of MFRP and the secreted serine proteases PRSS56 and ADAMTS19 as part of a molecular network involved in ocular growth regulation

Summary

Precise regulation of ocular size is a critical determinant of normal visual acuity. Although it is generally accepted that ocular growth relies on a cascade of signaling events transmitted from the retina to the sclera, the factors and mechanism(s) involved are poorly understood. Recent studies have highlighted the importance of the retinal secreted serine protease PRSS56 and transmembrane glycoprotein MFRP, a factor predominantly expressed in the retinal pigment epithelium (RPE), in ocular size determination. Mutations in PRSS56 and MFRP constitute a major cause of nanophthalmos, a condition characterized by severe reduction in ocular axial length/extreme hyperopia. Interestingly, common variants of these genes have been implicated in myopia, a condition associated with ocular elongation. Consistent with these findings, mice with loss of function mutation in PRSS56 or MFRP exhibit a reduction in ocular axial length. However, the molecular network and cellular processes involved in PRSS56- and MFRP-mediated ocular axial growth remain elusive. Here, we show that Adamts19 expression is significantly upregulated in the retina of mice lacking either Prss56 or Mfrp. Importantly, using genetic mouse models, we demonstrate that while ADAMTS19 is not required for ocular growth during normal development, its inactivation exacerbates ocular axial length reduction in Prss56 and Mfrp mutant mice. These results suggest that the upregulation of retinal Adamts19 is part of an adaptive molecular response to counteract impaired ocular growth. Using a complementary genetic approach, we show that loss of PRSS56 or MFRP function prevents excessive ocular axial growth in a mouse model of early-onset myopia caused by a null mutation in Irbp, thus, demonstrating that PRSS56 and MFRP are also required for pathological ocular elongation. Collectively, our findings provide new insights into the molecular network involved in ocular axial growth and support a role for molecular crosstalk between the retina and RPE involved in refractive development.

Keywords:

Biometrics – Eyes – Gene expression – Genetically modified animals – Genetics – Mouse models – Retina

Introduction

Nanopthalmos is a rare developmental disorder characterized by significantly smaller but structurally normal eyes and extreme hyperopia resulting from compromised ocular growth [1]. In addition, nanophthalmic individuals are highly susceptible to developing blinding conditions including secondary angle-closure glaucoma, spontaneous choroidal effusions, cataracts, and retinal detachment [1]. Both sporadic and familial forms of nanophthalmos with autosomal dominant or recessive inheritance have been described [2]. To date, six genes (PRSS56, MFRP, TMEM98, CRB1, BEST1, and MYRF) have been implicated in familial forms of nanophthalmos, with PRSS56 and MFRP mutations being the most prevalent among multiple cohorts [1,3–10]. Notably, the eyes of nanophthalmic individuals with biallelic mutations in PRSS56 or MFRP were found to be significantly smaller as compared to those carrying dominant mutations in TMEM98 or MYRF. Interestingly, common PRSS56 and MFRP variants have also been found to be associated with myopia, a condition phenotypically opposite to nanophthalmos that is characterized by excessive ocular growth [11,12]. Consistent with human genetic studies, we have demonstrated previously that loss of PRSS56 function in the mouse leads to ocular axial length reduction and hyperopia [13]. Similarly, mice and zebrafish deficient for MFRP exhibit ocular axial length reduction [14–16]. Together, these findings underscore the importance of PRSS56 and MFRP in ocular growth [2].

Ocular growth can be broadly divided into two distinct phases that take place prenatally and postnatally, respectively [17]. Prenatal ocular growth occurs in the absence of visual experience and is primarily dictated by genetic factors [13]. In contrast, postnatal ocular growth also referred to as emmetropization, is a vision-guided process modulated by the refractive status of the eye to ensure that the axial length matches the optical power of the eye to achieve optimal focus and visual acuity. Nanophthalmos is generally attributed to impaired prenatal ocular growth as individuals with this condition are born hyperopic [1,18]. In contrast, common forms of myopia arise from excessive postnatal ocular axial growth. Thus, the association of PRSS56 and MFRP variants with both nanophthalmos and myopia supports a critical role for these factors during distinct phases of ocular growth and suggests that some of the factors and mechanisms regulating ocular growth are conserved between the prenatal and postnatal stages of ocular development.

It is generally accepted that postnatal ocular growth is regulated by a cascade of signaling events by which information is relayed from the retina to the sclera to ultimately induce scleral extracellular matrix (ECM) remodeling and promote ocular axial elongation [19,20]. Notably, PRSS56 is specifically expressed in the retina [13], which is consistent with a central role for the retina in ocular growth regulation. MFRP is predominantly expressed in the retinal pigment epithelium (RPE) and ciliary epithelium [18]. Importantly, the RPE lies between the retina and the sclera and has been suggested to act as an intermediate in transmitting signals to the sclera during ocular growth [21].

Although current evidence supports a key role for MFRP and PRSS56 in ocular size regulation, the underlying mechanisms remain elusive. In this study, we use Prss56 and Mfrp mutant mouse models and complementary genetic approaches to gain insights into the molecular network involved in ocular size regulation [3,13,22]. Importantly, we identified characteristic changes in the retinal gene expression profile triggered by impaired ocular growth. Specifically, we show that Adamts19 mRNA levels are significantly increased in the retina of Prss56 and Mfrp mutant mice and provide evidence that the upregulation of retinal Adamts19 expression is part of an adaptive molecular response to mitigate impaired ocular growth. Moreover, we demonstrate that loss of PRSS56 or MFRP function prevents excessive ocular axial elongation in a mouse model of early-onset myopia caused by a null mutation in Irbp [23], further establishing PRSS56 and MFRP as critical regulators of ocular growth. Collectively, our findings identify factors involved in the crosstalk between the retina and the RPE during ocular growth and provide new insight into the molecular network involved in ocular size regulation and refractive development.

Results

Adamts19 expression is upregulated in the retina of Prss56 mutant mice

To characterize the molecular processes underlying PRSS56-mediated ocular size regulation, we performed a high throughput gene expression analysis on the retina from Prss56^glcr4 mutant mice and their control littermates. We demonstrated recently that the Prss56^glcr4 mutation, which leads to a truncated PRSS56 protein, caused ocular size reduction via a loss of function mechanism, hence Prss56^glcr4/glcr4 mice will be referred to as Prss56^-/- throughout the manuscript [13]. Our transcriptome analysis (RNA-Seq) identified 396 differentially expressed genes, of which 196 were upregulated and 200 downregulated in Prss56^-/- retina (S1 Fig). Fold change analysis revealed that Prss56 and Adamts19 were the top two differentially expressed genes between Prss56 mutant (Prss56^-/-) and control (Prss56^+/-) retina (S1 Table and S1 Data). In addition to being the most significantly upregulated genes, Prss56 and Adamts19 both encode secreted serine proteases and might impact similar molecular pathways. Therefore, we focused on these two genes to begin elucidating the mechanisms implicated in ocular size regulation and related diseases. Consistent with the RNA-Seq results, qPCR analysis revealed that Prss56 and Adamts19 mRNA levels were significantly upregulated in the retina of Prss56^-/- mice compared to their Prss56^+/- and Prss56^+/+ littermates at both ages examined (postnatal day (P) 15 (p = 0.002) and P30 (p = 0.0003) (Fig 1A and 1B). Notably, Prss56 and Adamts19 retinal expression levels in Prss56^+/- mice were comparable to those detected in Prss56^+/+ mice, which is consistent with the absence of an ocular phenotype in Prss56^+/- mice [13]. Prss56^+/- mice were therefore used as controls for all experiments presented in this study. As described previously, we detected a progressive upregulation of Prss56 mRNA levels in Prss56^-/- retina from P15 to P60 (S2A Fig) [13]. The increase in Adamts19 retinal expression was found to precede that of Prss56 in Prss56^-/- retina, and was detected as early as P10 and gradually increased to reach a peak by P30 (S2B Fig). Upregulation of retinal Prss56 and Adamts19 expression was also observed in an alternate genetic mouse model carrying a null allele of Prss56 (Prss56^Cre) we described previously [13], further confirming that the increase in retinal Prss56 and Adamts19 expression results from loss of PRSS56 function (Fig 1C). To determine the spatial expression pattern of Adamts19, we next performed in situ hybridization on ocular sections from Prss56^-/- and Prss56^+/- mice. Despite using the highly sensitive QuantiGene ViewRNA assay, Adamts19 expression was only detected in Prss56^-/- retina, indicating that Adamts19 expression was below the threshold of detection in control Prss56^+/- retina (Fig 1D). In Prss56 mutant retina, Adamts19 expression was predominantly observed in the inner nuclear layer (INL), a region where the cell bodies of Müller glia are located and which coincides with the retinal expression pattern of Prss56 (Fig 1E). Collectively, these results demonstrate that in addition to causing ocular size reduction, loss of PRSS56 function leads to alterations in retinal gene expression characterized by increased Prss56 and Adamts19 mRNA levels.

Retinal Prss56 and Adamts19 mRNA levels are upregulated in response to ocular size reduction in Prss56 mutant mice

To determine whether the upregulation of retinal Adamts19 and Prss56 mRNA levels correlates with ocular size reduction in Prss56 mutant mice, we took advantage of the Egr1;Prss56 double mutant mouse model (Egr1^-/-;Prss56^-/-) we described previously [13]. EGR1 (Early Growth Response-1) is a major regulator of ocular growth and Egr1^-/- mice exhibit increased ocular axial length [13,24]. We have previously shown that Egr1 inactivation rescues the reduction in ocular axial length and vitreous chamber depth (VCD) in Prss56 mutant mice as the ocular size of Egr1^-/-;Prss56^-/- mice is comparable to that of control Egr1^+/-;Prss56^+/- mice[13]. Using qPCR analysis, we show that in addition to rescuing ocular axial length reduction, Egr1 inactivation also prevented the increase in retinal Prss56 and Adamts19 expression in Prss56 mutant mice (compare Egr1^-/-;Prss56^-/- to Egr1^+/+;Prss56^-/- in Fig 2). Notably, Prss56 and Adamts19 mRNA levels in Egr1^-/-;Prss56^-/- retina were comparable to those observed in control Egr1^+/+;Prss56^+/- retina (S3 Fig). These findings suggest that the upregulation of retinal Prss56 and Adamts19 expression resulted from the ocular size reduction caused by the loss of PRSS56 function. Using a conditional gene targeting approach, we found that retinal Adamts19 expression was also increased following conditional inactivation of Prss56 during the post eye-opening period (P21, Fig 2C). This suggests that Adamts19 upregulation might play a role at various stages of ocular growth in Prss56 mutant mice, including the developmental window when the eye is responsive to visual experiences.

ADAMTS19 is not required for ocular growth during normal development

PRSS56 and ADAMTS19 are both secreted serine proteases, raising the possibility that they might have overlapping functions during ocular growth. To test this possibility, we first generated Adamts19 knockout mice by crossing the conditional Adamts19 mice with the ubiquitous β-actin-Cre line (S4 and S5 Figs). To determine if the loss of ADAMTS19 function leads to ocular defects, we performed spectral-domain optical coherence tomography (SD-OCT) to assess various ocular biometric parameters. We found that all the ocular parameters examined, including ocular axial length, VCD, lens diameter, and retinal thickness were indistinguishable between Adamts19^+/+, Adamts19^+/- and Adamts19^-/- mice (Figs 3 and S6). These results demonstrate that ADAMTS19 is not required for ocular growth during normal development.

Adamts19 inactivation exacerbates the reduction in ocular axial length in Prss56^-/- mice

In light of our findings, we hypothesized that the upregulation of retinal Adamts19 expression might be part of an adaptive molecular response to compensate for the loss of PRSS56 function and promote ocular axial growth. To examine this possibility, we tested the effect of Adamts19 inactivation in Prss56^-/- mice by crossing Prss56 mutant mice to the Adamts19 mutant line to generate Prss56^-/- mice that are wild-type, heterozygous, or homozygous for the Adamts19 null allele (Adamts19^+/+, Adam19^+/- or Adamts19^-/-). Since all ocular biometric parameters of Adamts19^+/-; Prss56^+/- mice were comparable to those of wild-type (Adamts19^+/+;Prss56^+/+) littermates (S7 Fig), they were used as controls. As expected, ocular axial length and VCD were significantly reduced in all three groups of mice lacking Prss56 (Prss56^-/-) compared to the Adamts19^+/-;Prss56^+/- control mice (Fig 4A and 4B). However, while the lens diameter was comparable between control mice and Prss56 mutant mice, irrespective of their Adamts19 genotype (S8A Fig), ocular axial length and VCD were significantly smaller in Adamts19;Prss56 double mutant mice (Adamts19^-/-;Prss56^-/-) compared to Adamts19^+/-;Prss56^-/- or Adamts19^+/+;Prss56^-/- at both age examined (p<0.05 and p<0.01 for axial length and VCD at P18 and p<0.05 for axial length and VCD at P30) (Fig 4B and 4C). As reported previously [13], ocular axial length reduction in Prss56^-/- mice was associated with an increase in retinal thickness compared to the Adamts19^+/-;Prss56^+/- control mice (S8B Fig). Notably, the retina from Adamts19^-/-;Prss56^-/- mice were modestly but significantly thicker than those from Adamts19^+/+; Prss56^-/- mice at P18 (p = 0.0057). In addition, we found that Adamts19 expression was significantly increased in the retina of both Adamts19^+/+;Prss56^-/- and Adamts19^+/-;Prss56^-/- mice compared to those of control Adamts19^+/-;Prss56^+/- mice (p<0.0001 and p = 0.0014, respectively), which is in agreement with the observation that Adamts19 can partially rescue the reduction in ocular size in Prss56 mutant mice (S8C Fig). Together, these results demonstrate that Adamts19 inactivation exacerbates ocular size reduction in Prss56^-/- mice and is consistent with the upregulation of retinal Adamts19 expression being part of an adaptive molecular response triggered by impaired ocular growth in Prss56 mutant mice.

Adamts19 inactivation exacerbates the reduction in ocular axial length in Mfrp mutant mice

Interestingly, elevated levels of retinal Prss56 expression have recently been reported in another mouse model of nanophthalmos caused by a mutation in the gene coding for membrane frizzed related-protein (Mfrp) [22]. Although increased Adamts19 expression was also reported in Mfrp^-/- eyes, the specific ocular tissue/cell type expressing Adamts19 was not addressed [22]. Since Adamts19 expression was specifically detected in the retina of Prss56^-/- mice, we performed a qPCR analysis to confirm that Prss56 and Adamts19 mRNA levels were upregulated in the retina from Mfrp^-/- mice compared to control Mfrp^+/- littermates (Fig 5A). To determine if Adamts19 inactivation also exacerbates the ocular size reduction caused by Mfrp deficiency, we crossed Mfrp mutant mice with the Adamts19 mutant line and conducted SD-OCT analyses on the progeny. Since the ocular biometric parameters of Adamts19^+/-;Mfrp^+/- mice were comparable to those of wild-type mice (Adamts19^+/+;Mfrp^+/+), they were used as controls (S9 Fig). As expected, ocular axial length and VCD were significantly reduced in Mfrp mutant mice (Adamts19^+/-;Mfrp^-/-) compared to Adamts19^+/-;Mfrp^+/- control littermates, and the lens diameter was comparable across the groups (Figs 5C–5E and S10A). Importantly, the ocular axial length and VCD were further reduced in Adamts19^-/-;Mfrp^-/- mice compared to Mfrp mutant mice (Adamts19^+/-;Mfrp^-/-) (p = 0.0388 and p = 0.0001, respectively) (Fig 5C–5E). Additionally, retinal thickness was significantly increased in Adamts19^+/-;Mfrp^-/- and Adamts19^-/-;Mfrp^-/- mice compared to control Adamts19^+/-; Mfrp^+/- mice (p<0.0001) (S10B Fig). These findings further support a role for the upregulation of retinal Adamts19 expression being part of a compensatory mechanism triggered by impaired ocular axial growth.

Inactivation of Prss56 or Mfrp prevents pathological ocular axial elongation in Irbp mutant mice

PRSS56 and MFRP are required for ocular axial growth during normal development [3,6,13,18]. To determine if these factors are also implicated in pathological ocular elongation, we tested the effects of Prss56 and Mfrp inactivation in a mouse model of early-onset myopia associated with excessive ocular axial growth that is caused by a null mutation in the gene coding for IRBP (Interphotoreceptor retinoid-binding protein) [23]. To this end, each of the Prss56 and Mfrp mutant lines were crossed to Irbp mutant mice and biometric ocular assessment was conducted on the progeny. Since the ocular biometric parameters of double heterozygous mice (Irbp^+/-;Prss56^+/- and Irbp^+/-;Mfrp^+/-) were comparable to those of wild-type mice (Irbp^+/+;Prss56^+/+ and Irbp^+/+;Mfrp^+/+), they were used as controls (S11 Fig). As expected, SD-OCT analyses revealed that both the ocular axial length and VCD were significantly increased in Irbp single mutant mice (Irbp^-/-;Prss56^+/- or Irbp^-/-;Mfrp^+/-) (p<0.0001) and significantly reduced in Prss56 or Mfrp single mutant mice (Irbp^+/-;Prss56^-/- or Irbp^+/-;Mfrp^-/-) (p<0.001 for axial length and VCD in Prss56 single mutants and p = 0.01 and p<0.001 for axial length and VCD in Mfrp single mutants) compared to their respective controls (Irbp^+/-;Prss56^+/- and Irbp^+/-;Mfrp^+/- mice) (Figs 6A–6D, S12A and S13A). Inactivation of either Prss56 or Mfrp prevented ocular axial elongation in Irbp mutant mice (Irbp^-/-;Prss56^-/- and Irbp^-/-;Mfrp^-/-, respectively). Notably, ocular axial length and VCD were significantly reduced in both double mutant lines (Irbp^-/-;Prss56^-/- and Irbp^-/-;Mfrp^-/-) (p<0.001) compared to their respective control littermates (Irbp^+/-;Prss56^+/- and Irbp^+/-;Mfrp^+/-) and were comparable to those observed in Prss56 and Mfrp single mutant mice (Irbp^+/-;Prss56^-/- and Irbp^+/-;Mfrp^-/-, respectively). In addition, while retinal thickness was increased in both Prss56 and Mfrp single mutant mice (Irbp^+/-;Prss56^-/- and Irbp^+/-;Mfrp^-/-), it was significantly reduced in Irbp single mutant mice (Irbp^-/-;Prss56^+/- or Irbp^-/-;Mfrp^+/-) compared to control littermates (Irbp^+/-;Prss56^+/- and Irbp^+/-;Mfrp^+/-, respectively) (p<0.0001) (S12B and S13B Figs). Consistent with the observed reduction in their ocular size, the retinal thickness of Irbp^-/-;Prss56^-/- and Irbp^-/-;Mfrp^-/- mice was comparable to that of their respective Prss56 and Mfrp single mutant littermates (Irbp^+/-;Prss56^-/- and Irbp^+/-;Mfrp^-/-). Together, these findings demonstrate that the excessive ocular elongation observed in Irbp^-/- mice is dependent on PRSS56 and MFRP functions and suggest that IRPB, PRSS56, and MFRP are likely part of the same pathway contributing to ocular growth.

To further investigate the relationship between IRBP and PRSS56 during ocular elongation, we next assessed the retinal expression of Adamts19 in Irbp mutant mice carrying various Prss56 genotypes. Notably, the qPCR analysis revealed that the retinal expression of Adamts19 in double heterozygous (Irbp^+/-;Prss56^+/-) mice was comparable to that of wild-type mice (Irbp^+/+;Prss56^+/+, S14 Fig), which is consistent with the absence of an ocular phenotype in Irbp^+/-;Prss56^+/- mice (Figs 6 and S11). As expected, Adamts19 mRNA levels were significantly upregulated in Prss56 mutant retina (Irbp^+/-;Prss56^-/-) compared to Irbp^+/+;Prss56^+/+ and Irbp^+/-;Prss56^+/- control retinas. Interestingly, while no change in retinal Adamts19 expression was detected in Irbp mutant mice carrying two wild-type Prss56 alleles (Irbp^-/-;Prss56^+/+), it was upregulated in Irbp^-/-;Prss56^+/- mice, although to a lesser extent of what is observed in Prss56 mutant mice. These results suggest that the upregulation of retinal Adamts19 expression might be an attempt to compensate for the loss of one functional copy of Prss56 and might contribute to the excessive ocular elongation in Irbp^-/-;Prss56^+/- mice. Furthermore, retinal expression of Adamts19 was significantly higher in double mutants (Irbp^-/-;Prss56^-/-) compared to Irbp or Prss56 single mutant mice (Irbp^-/-;Prss56^+/- and Irbp^+/-;Prss56^-/-, respectively) suggesting that Adamts19 upregulation induced by loss of PRSS56 function is exacerbated by Irbp inactivation. Overall, our data suggest that increased Adamts19 upregulation is achieved through distinct mechanisms to mitigate the effect of molecular defects leading to impaired ocular axial growth (Fig 7A and 7B).

Discussion

The molecular and cellular mechanisms involved in ocular axial growth and emmetropization are poorly understood. Previous studies have identified PRSS56 and MFRP mutations as a major cause of nanophthalmos, a condition characterized by severe ocular size reduction and extreme hyperopia, suggesting that these factors play a critical role in ocular growth [3–6]. Consistent with this, Prss56 and Mfrp mutant mice recapitulate the cardinal features of nanophthalmos, i.e. reduced ocular axial length and hyperopia [3,13,16]. Here, we use complementary genetic strategies in Prss56 and Mfrp mutant mouse models as a first step to elucidate the molecular and cellular networks implicated in ocular size determination. Notably, we identified ADAMTS19 as a novel factor involved in ocular size regulation and demonstrate that the upregulation of retinal Adamts19 expression is part of a protective molecular response to impaired ocular growth. Furthermore, we show that inactivation of Prss56 or Mfrp prevent excessive ocular elongation in a mouse model of developmental myopia caused by a null mutation in Irbp, demonstrating that PRSS56 and MFRP are not only essential for ocular growth during normal development but also required to induce ocular elongation associated with early-onset/developmental myopia.

Using two distinct mouse models of nanophthalmos and a combination of genetic approaches, we identified a retinal gene expression signature characterized by the upregulation of the genes coding for PRSS56 and ADAMTS19 that is associated with the reduced ocular size. Notably, RNA-Seq analysis revealed that Prss56 and Adamts19 were the top two differentially expressed genes between Prss56 mutant (Prss56^-/-) and control (Prss56^+/-) retina. Besides Prss56, Tmem98 was the only other differentially expressed gene that has previously been implicated in nanophthalmos [8,25] and was found to be significantly downregulated in the retina from Prss56^-/- mice compared to control Prss56^+/- mice (S1 Table). Interestingly, TMEM98, which encodes a single transmembrane protein of unknown function, is also one of the several loci associated with high myopia in genome-wide association study (GWAS) [12]. Notably, conditional ablation of Tmem98 from the RPE, leads to significant increase in ocular size and axial length [8]. Thus, it is possible that the reduced Tmem98 expression detected in Prss56^-/- mice might have resulted from the differential expression in the fraction of RPE attached to the isolated retina for RNA-Seq analysis. In which case Tmem98 downregulation could be part of a molecular response resulted to counteract impaired ocular elongation in Prss56^-/- mice.

Importantly, we show that although ADAMTS19 is not required for ocular axial growth during normal development, its inactivation exacerbates the reduction in ocular axial length and VCD in both Prss56 and Mfrp mutant mouse models. Of note, the presence of a single wild-type Adamts19 allele (Adamts19^+/-) is sufficient to confer a protective effect against axial length reduction in Prss56 and Mfrp mutant mice. In addition to ocular axial length reduction, Mfrp mutant mice also develop retinal degeneration by 8 to 10 weeks of age [26]. Since ocular axial length reduction is detected before any noticeable retinal morphological changes are observed, it is likely that ocular size reduction and retinal degeneration are independent insults. Thus, these results suggest that the upregulation of retinal Adamts19 expression constitutes a protective molecular response to overcome impaired ocular axial growth.

Our in situ hybridization analysis revealed that Adamts19 expression was predominantly observed in the retinal INL in Prss56^-/- eyes and was below the threshold of detection in wild-type eyes, which is in agreement with the absence of ocular phenotypes in Adamts19 mutant. Interestingly, we previously demonstrated that Prss56 is specifically expressed in Müller cells whose cell bodies are found in the INL [13]. Together, these observations are consistent with the hypothesis of retinal Adamts19 upregulation being part of a compensatory response to altered ocular growth. Furthermore, they suggest that Müller glia constitute a key cellular component of the ocular growth regulatory network. Interestingly, Müller glia has been postulated to act as sensors of biomechanical alterations in the retina, responding to mechanical tension by detecting subtle changes in retinal organization due to stretching of their long processes or side branches [27,28]. It is possible that impaired ocular growth (reduced ocular size) alters the structural and mechanical properties of the retina that are sensed by Müller glia triggering transcriptional activation of factors participating in the regulation of ocular axial growth [28]. Accumulating evidence indicates that mechanical cues, such as physical forces (e.g. tension, compression, or shear stress) or changes in cellular shape and size can trigger molecular responses [29]. Interestingly, the cyclic mechanical strain has been reported to induce extensive changes in the gene expression in rat Müller cells [30]. Future studies will focus on characterization of the Prss56 promoter and the factors involved in regulating Prss56 expression in the retina.

PRSS56 and ADAMTS19 both belong to the family of secreted serine proteases raising the possibility that they could have overlapping or redundant function(s) and act on the same substrate(s), which would explain the compensatory effect of ADAMTS19 on ocular elongation in Prss56 deficient mice. Previous studies have reported that PRSS56 is present in the retinal inner limiting membrane (ILM) [31], which consists of Müller glia endfeet and a basement membrane rich in type IV collagen and laminins that serves as a boundary between the retina and vitreous body. It is possible that secreted PRSS56 and ADAMTS19 may promote ECM remodeling of the retinal ILM, which in turn could facilitate ocular axial growth. Notably, changes in the ILM have previously been suggested to influence ocular size [32]. Proteolytic substrates of PRSS56 and ADAMTS19 are unknown and our current efforts are focused on identifying PRSS56 substrate(s).

Using a complementary genetic approach, we show that besides promoting ocular growth during normal development, PRSS56 and MFRP are also required for pathological ocular elongation observed in a mouse model of severe and early-onset myopia caused by a null mutation in Irbp [23]. Indeed, inactivation of either Prss56 or Mfrp prevented excessive ocular elongation in Irpb mutant mice, demonstrating that Prss56 and Mfrp are epistatic to Irbp and suggesting that PRSS56 and MFRP are likely part of the same pathway modulating ocular growth. It is generally accepted that ocular elongation is driven by signals originating from the retina that are relayed to the RPE and choroid before reaching the sclera to induce scleral ECM remodeling and ocular axial growth [19,20]. The observation that inactivation of Mfrp, a gene expressed in the RPE and ciliary epithelium but not in the neural retina, leads to increased retinal expression of Prss56 and Adamts19 further suggests the existence of molecular crosstalk between the retina and the RPE in ocular axial growth [13,18]. Müller glia spans the entire length of the retina and thus, represents an ideal cell type for the transmission of retinal signals during ocular elongation. Interestingly, IRBP is primarily found in the interphotoreceptor matrix of the retina, which is located between the photoreceptor cells and the RPE [23]. It is tempting to speculate that PRSS56, IRBP, and MFRP may be part of a molecular network connecting the retina and RPE and facilitating the flow of information during ocular growth (Fig 7C). Interestingly, while no change in retinal Adamts19 expression was detected in Irbp mutant mice (Irbp^-/-;Prss56^+/+), it was elevated in Irbp mutant mice that are heterozygous for mutant Prss56 (Irbp^-/-;Prss56^+/-) compared to control mice. This finding suggests that Adamts19 upregulation potentially compensates for the loss of one functional Prss56 allele and contributes to the excessive ocular elongation in Irbp mutant mice and further supports the notion that ADAMTS19 and PRSS56 may have an overlapping function.

The findings from the genetic mouse models used in our study are primarily relevant to ocular axial growth that is not dependent on visual experience which corresponds to prenatal and possibly postnatal vision-unadjusted ocular growth in humans. However, genetic variants of PRSS56 and MFRP are also associated with common forms of myopia [11,12], raising the possibility that these factors may also play a role during emmetropization. Further supporting a potential role of ADAMTS19 in postnatal ocular growth and possibly refractive development, we show that conditional ablation of Prss56 at P21 when the eyes are exposed to visual cues [33] also causes upregulation of retinal Adamts19 expression in the mouse. Thus, it is tempting to speculate that a molecular link between Müller glia and the RPE, mediated at least in part by PRSS56 and MFRP, serves a broader role, contributing to prenatal as well as postnatal vision-unadjusted and -adjusted ocular growth. Accordingly, any functional disruption of the molecular linkage between Müller glia and RPE could contribute to refractive errors. Furthermore, since the loss of PRSS56 or MFRP function causes a reduction in ocular axial length [6,13], it is likely that PRSS56 and MFRP variants associated with myopia act via a gain of function or overexpression mechanism(s) leading to an opposite phenotype characterized by an increase in ocular axial length. Consistent with this notion, a recent study has shown that optical defocus-induced axial elongation/myopia in marmoset is accompanied by increased expression of Prss56 [34].

Our findings raises a question as to whether the significance of Adamts19 upregulation is limited only to severe ocular axial as observed in PRSS56 and MFRP mutations or they have broader relevance in context of the axial length changes observed in general population. A recent GWAS has suggested a role for ADAMTS19 in refractive development [35]. Their findings suggest that ADAMTS19 may be part of a genetic pathway that influences both corneal curvature and ocular axial growth. Corneal curvature is closely linked to the refractive status of the eye, with flatter and steeper corneal curvature being associated with hyperopia and myopia, respectively [35]. Interestingly, although the ADAMTS19 variant was found to influence corneal curvature and axial length, it was not associated with refractive errors. Notably, the ADAMTS19 variant was associated with a flatter cornea and increased axial length raising the possibility that its effects on ocular elongation could overcome the hyperopia induced by a flattened cornea and suggesting a compensatory pleiotropic role for ADAMTS19 in the regulation of corneal curvature and ocular elongation during refractive development. In light of the GWAS data and findings from our mouse studies, it is tempting to speculate that human ADAMTS19 variant acts by increasing the retinal expression of ADAMTS19 and thereby, contributing to ocular axial length regulation in the general population.

Overall, our study identifies characteristic changes in retinal gene expression in response to impaired ocular growth and demonstrates that upregulation of retinal Adamts19 expression is part of a compensatory molecular pathway to promote ocular elongation. Furthermore, our results highlight the critical roles of PRSS56 and MFRP in promoting ocular growth both during normal development and pathological processes underlying excessive ocular elongation in Irbp mutant mouse model of early-onset myopia, thus suggest a potential role for Müller glia in mediating a retina-RPE cross-talk guiding axial growth. Collectively, these findings provide important insight into the molecular mechanisms underlying ocular growth, which could have implications for the development of therapeutic interventions to prevent or slow the progression of refractive errors.

Materials and methods

Ethics statement

All experiments were conducted in compliance with protocols approved by the Institutional Animal Care and Use Committee at the University of California San Francisco (IACUC) (Protocols # AN181358-01D) and per the guidelines from the Association for Research in Vision and Ophthalmology’s statement on the use of animals in ophthalmic research. Animals were given access to food and water ad libitum and housed under controlled conditions including a 12-h light/dark cycle per the National Institutes of Health guidelines. Both male and female mice were used in all experiments and no differences were observed between sexes all comparisons were made between littermates to minimalize variability.

Mouse lines

• Prss56^-/-: (C57BL/6J.Cg.Prss56^glcr4/Sj)- Mice carrying an ENU-induced mutation in Prss56 causing truncation of PRSS56 protein at its C-terminal region[3].

• Prss56^Cre/Cre:(C57BL/6J.Cg-Prss56tm(cre))- Mice carrying a null Prss56 allele in which the exon1 of Prss56 was replaced by the CRE recombinase sequence[13,36].

• Egr1^-/-: (C57BL/6J.Egr1^tm1Jmi)- Egr1 mutant mice generated through targeted mutation by insertion of a PGK-neo cassette introduces stop codon resulting in a null allele [37].

• Adamts19^-/-: (C57BL/6J.Adamts19^{tm4a(EUCOMM)Wtsi)})- A conditional knockout mouse with LoxP sites flanking exon3 of Adamts19. Excision of the LoxP sites by the ubiquitously expressed CRE recombinase driven by beta-actin promoter leads to the generation of a knockout allele of Adamts19 (S4A and S4B Fig) [38].

The Adamts19 conditional mutant mice originally generated on a C57BL/6N background were backcrossed to C57BL/6J mice for a minimum of 6 generations. The Adamts19 conditional mutant mice maintained on the original C56BL/6N background carry the retinal degeneration 8 (rd8) mutation in the Crb1 gene. However, Adamts19 mutant mice backcrossed to the C57BL/6J background used in this study do not carry the rd8 mutation (S4C Fig).

• 5

  Mfrp^-/-: (C57BL/6J.C3Ga-Mfrp^rd6)- The mouse strain is homozygous for rd6 exhibiting retinal degeneration around four weeks during the retinal developmental phase [22].

• 6

  Irbp^-/-: (C57BL/6J.Irbp^tm1Gil)- A Irbp knockout mouse model [23]. This mouse line carries a targeted mutation for the Irbp gene, also known as Rbp3, where the promoter and exon 1 have been replaced by a NEO selection cassette rendering IRBP protein inactive.

• 7

  Prss56^F/F: C57BL/6J.Cg-Prss56tm1/Sj, Prss56 conditional knockout mice carrying LoxP sites flanking exons 2 to 4 of Prss56. Excision of exons 2–4 results in a catalytically inactive form of PRSS56 [13].

• 8

  UBC-Cre ER^T2: C57BL/6J.Cg-Tg(UBC-Cre/ERT2)1Ejb, Cre-ERT2 transgenic mouse line carrying a human ubiquitin C (UBC) promoter sequence upstream of a Cre-ERT2 fusion gene.

All strains were maintained on a C57BL/6J background. PCR genotyping of all mouse strains was performed on genomic DNA obtained from tail biopsies digested with Proteinase K (Sigma, St. Louis, MO, USA) using primers listed in S2 Table. Allele-specific PCR for Crb1 mutation (rd8) was done as described previously [39]. With the exceptions of a few rare cases when an eye or retina was damaged, both eyes and retina from each experimental mouse were used for SD-OCT and qPCR analyses, respectively.

RNA-Seq analysis

Total RNA was isolated from retinas obtained from P15 Prss56^-/- and Prss56^+/- using the Qiagen RNeasy Mini Kit according to the manufacturer’s instructions. Total RNA was isolated from 3 sets of mice for each genotype (Prss56^-/- or Prss56^+/-) subsequent RNA-Seq analysis (n = 3/genotype, in each set the left and right retina, were pooled). The RNA samples were submitted to the Novogene Corporation Inc. (Sacramento, USA) passed the quality control (RNA integrity number >8). A total amount of 1 μg RNA per sample was used as input material for the RNA sample preparations. Sequencing libraries were generated using NEBNextUltra RNA Library Prep Kit for Illumina (NEB, USA) following the manufacturer’s recommendations. The quality of the prepared libraries was assessed on the Agilent Bioanalyzer 2100 system. The clustering of the index-coded samples was performed on a cBot Cluster Generation System using PE Cluster Kit cBot-HS (Illumina) according to the manufacturer’s instructions. After cluster generation, the library preparations were sequenced on an Illumina platform and paired-end reads were generated. The raw FASTQ files for the RNA-Seq libraries were deposited to the NCBI Sequence Read Archive (SRA) SRP295848 and have been assigned BioProject accession PRJNA682538 along with GEO Submission (GSE162647). Outputs from the transcripts were deposited to the NCBI Gene Expression Omnibus (GEO) with accession number GSM4956165- GSM4956170. Supplemental file GSE162647_MUTANTvsCONTROL.DEG.all.txt.gz contains the genome, counts, and normalized counts used to generate the statistical plots.

Ocular biometry

Ocular biometry was performed using Envisu R4300 spectral-domain optical coherence tomography (SD-OCT, Leica/Bioptigen Inc., Research Triangle Park, NC, USA). Measurements of various ocular parameters including axial length (AL), vitreous chamber depth (VCD), anterior chamber depth (ACD), lens diameter, and retinal thickness were performed on mice anesthetized with ketamine/xylazine (100 mg/kg and 5mg/kg, respectively; intraperitoneal (IP) following pupil dilation as described previously[13].

Quantitative polymerase chain reaction (qPCR)

For qPCR analysis of gene expression, eyes were enucleated and retinas were immediately dissected and total RNA was extracted from mouse retinal tissue using Qiagen RNeasy Mini Kit as per manufacturers protocol (Qiagen, Valencia, CA, USA) and reverse-transcribed using iScript cDNA Synthesis Kit (Bio-Rad, Hercules, CA, USA) and primer sets listed in S3 Table. qPCR was performed on Bio-Rad C1000 Thermal Cycler/CF96 Real-Time System using SSOAdvanced SYBR Green Supermix (Bio-Rad, Hercules, CA, USA). Briefly, 100ng of cDNA and 10uM primers were used per reaction in a final volume of 20ul. Each cycle consisted of denaturation at 95°C for 15s, followed by annealing at 60°C for 15s, extension 72°C for 30s for a total of 39 cycles. All the experiments were run as technical duplicates and a minimum of three biological replicates were used per group. The relative expression level of each gene was normalized to housekeeping genes (Actinβ, Hprt1, and Mapk1) and analyzed using the CFX Maestro 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). Eyes were enucleated post-fixed in RNAse-free 4% PFA, cryoprotected in 20% sucrose, and embedded in OCT compound and sectioned within 24 hours for in situ hybridization. QuantiGene View RNA (Affymetrix, Santa Clara, CA, USA). In situ hybridization 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, fixed and hybridized with probe sets against Adamts19 (NM_175506 (Adamts19), TYPE1, high sensitivity with 40~50 bp DNAs) for 3 hours at 40°C using a Thermo Brite system (Abbott Molecular, Des Plaines, IL, USA). Cryosections were then washed and subject to signal amplification and detection using a Fast Red substrate, counterstained with DAPI (blue) to visualize the nuclei, and mounted for subsequent imaging. Fluorescent images were acquired using an AxioImager M1 microscope equipped with an MRm digital camera and AxioVision software, with an LSM700 confocal microscope and Zen software (Carl Zeiss Microscopy, LLC, Germany).

Statistical analyses

Statistical comparisons between mutant and control groups or between multiple experimental groups at a given age were performed using two-tailed unpaired Student’s t-test and one-way ANOVA, respectively, using Prism statistical software (version 6.02, GraphPad Software, San Diego, CA). A p-value < 0.05 was considered significant.

Supporting information

S1 Fig [padj]
Differential gene expression between mutant and control retinas.

S2 Fig [a]
and expression in retina across ages.

S3 Fig [a]
inactivation normalizes the retinal expression of and expression in mutant mice.

S4 Fig [a]
Generation of mutant mice.

S5 Fig [a]
The mutant allele gives rise to a catalytically inactive truncated variant.

S6 Fig [a]
Ocular biometric analysis in mutant mice.

S7 Fig [a]
Ocular biometric parameters are indistinguishable between and wild-type mice.

S8 Fig [a]
Ocular biometric analysis of mice.

S9 Fig [a]
Ocular biometric parameters are indistinguishable between and wild-type mice.

S10 Fig [a]
Ocular biometric analysis of mice.

S11 Fig [tif]
Ocular biometric parameters are indistinguishable between or and wild-type mice.

S12 Fig [a]
Ocular biometric analysis of mutant mice following inactivation.

S13 Fig [a]
Ocular biometric analysis of mutant mice following inactivation.

S14 Fig [tif]
expression in mutant and mice.

S1 Table [xlsx]
List of genes differentially expressed between and control retina.

S2 Table [xlsx]
List of genotyping primers.

S3 Table [xlsx]
List of qPCR primers.

S1 Data [xlsx]
Complete list of differentially expressed genes between and control retina.