Congenital Cataract in Gpr161vl/vl Mice Is Modified by Proximal Chromosome 15

The morphology and severity of human congenital cataract varies even among individuals with the same mutation, suggesting that genetic background modifies phenotypic penetrance. The spontaneous mouse mutant, vacuolated lens (vl), arose on the C3H/HeSnJ background. The mutation disrupts secondary lens fiber development by E16.5, leading to full penetrance of congenital cataract. The vl locus was mapped to a frameshift deletion in the orphan G protein-coupled receptor, Gpr161, which is expressed in differentiating lens fiber cells. When Gpr161vl/vl C3H mice are crossed to MOLF/EiJ mice an unexpected rescue of cataract is observed, suggesting that MOLF modifiers affect cataract penetrance. Subsequent QTL analysis mapped three modifiers (Modvl3-5: Modifier of vl) and in this study we characterized Modvl4 (Chr15; LOD = 4.4). A Modvl4MOLF congenic was generated and is sufficient to rescue congenital cataract and the lens fiber defect at E16.5. Additional phenotypic analysis on three subcongenic lines narrowed down the interval from 55 to 15Mb. In total only 18 protein-coding genes and 2 micro-RNAs are in this region. Fifteen of the 20 genes show detectable expression in the E16.5 eye. Subsequent expression studies in Gpr161vl/vl and subcongenic E16.5 eyes, bioinformatics analysis of C3H/MOLF polymorphisms, and the biological relevancy of the genes in the interval identified three genes (Cdh6, Ank and Trio) that likely contribute to the rescue of the lens phenotype. These studies demonstrate that modification of the Gpr161vl/vl cataract phenotype is likely due to genetic variants in at least one of three closely linked candidate genes on proximal Chr15.


Introduction
Congenital cataract is the presence of an opacity in the lens at birth, affecting approximately 3 out of 10,000 live births in United States [1]. Similar to other birth defects, congenital cataract has a genetic basis with approximately 30-50% of the disorder being transmissible [2]. Several groups of genes are involved in congenital cataract, including lens crystallins (Crya, Cryb and Cryg), gap junction proteins (GJA3 and GJA8), membrane proteins (MIP), filament proteins (BFSP1 and BFSP2) and transcription factors (Hsf4 and Pax6) [3,4]. Human congenital cataract can be caused by autosomal dominant, autosomal recessive or X-linked mutations [5,6]. However, the same genetic mutation in different families can affect the penetrance or severity of the congenital cataract, suggesting that additional genes or environmental factors can modify the primary mutation [3,7,8].
Mouse mutant models of congenital cataract have significantly contributed to our understanding of the disorder at a molecular level. Several genes that cause congenital cataract in mice also contribute to human cataracts, including Cryg, Connexin, Foxe3 and Sox1 [9,10]. However, whether genetic variation can modify the penetrance and severity of mouse congenital cataract remains to be determined. In this study, we present the first multi-genic mouse model of congenital cataract.
The vl locus was positionally cloned to an 8bp deletion in the orphan G-protein coupled receptor, Gpr161. The deletion shifts the open reading frame, which truncates the receptor at the C-terminal intracellular domain. This truncation causes reduced steady state protein levels and a disruption of receptor mediated endocytosis, suggesting that the Gpr161 vl allele is a hypomorphic mutation. Consistent with the cataract phenotype, Gpr161 is expressed throughout lens development. At E12.5 and E14.5, Gpr161 transcripts are restricted to differentiating lens fiber cells and are absent from the proliferating anterior lens epithelium [11]. We previously determined that all the Gpr161 vl/vl embryos display normal lens phenotype up until E14.5. Starting at E14.5, when secondary lens fiber differentiation begins, a subset of Gpr161 vl/vl embryos exhibit abnormal lens fiber organization and vacuoles and by E16.5, 100% of the Gpr161 vl/vl embryos display a disorganized lens fiber phenotype.
Interestingly, genetic background modifies the Gpr161 vl/vl mutant phenotypes. When we previously crossed Gpr161 vl/vl C3H to the MOLF/EiJ genetic background, we found that the incidence of congenital cataract is decreased by 85%, along with a total rescue of embryonic lethality and 40% increased incidence of belly spot phenotype. These data suggested that unlinked modifiers could bypass the effect of the Gpr161 vl/vl mutation, establishing Gpr161 vl as a multi-genic mouse mutant model for congenital cataract.
In this study, we characterized Modvl4 (Chr15; LOD = 4.4). Genotyping and morphological analyses demonstrated that Modvl4 MOLF partially rescues Gpr161 vl/vl congenital cataract. To further delimit the modifying region, three subcongenic lines were generated which narrowed down the interval to 18 protein-coding genes and 2 miRNAs. RTPCR analysis identified fifteen genes with detectable expression in E16.5 eye. We further investigated the likely contribution of these fifteen genes to the Modvl4 MOLF modifying effect by their biological relevancy to lens development or Gpr161 signaling, the presence of likely functional C3H/MOLF polymorphisms using online resequencing data for the inbred lines, and measuring mRNA levels in Gpr161 vl/vl eyes to determine whether the congenic rescued any expression difference. These studies identified three genes in the interval, Cdh6 (Cadherin-6), Ank (pyrophosphate transporter) and Trio (Guanine exchange factor), that likely contribute to the Modvl4 MOLF cataract modifying effect.

Results
A Modvl4 MOLF congenic was generated to test the sufficiency of this locus to rescue Gpr161 vl/vl phenotypes. C3H/MOLF hybrid mice were backcrossed to C3H isogenic mice for 8 generations and progeny were selected based on SSLP markers across the genome. The male mice with the most contribution to the C3H background but were still C3H/MOLF for the Modvl4 QTL 95% CI were selected for further backcrossing until the congenic (Modvl4 C/M ) was generated (S1 Fig,  panel 1B). In Gpr161 vl/vl Modvl4 C/C , all animals display bilateral or unilateral cataract whereas in Gpr161 vl/vl Modvl4 C/M and Gpr161 vl/vl Modvl4 M/M , about 30% of the animals display eyes that are grossly indistinguishable from wild type. To quantify this phenotype, each individual animal was assigned a value: 1 for bilateral cataract, 0.5 for unilateral cataract and 0 for normal. The average value represents the incidence of cataract for each genotype. Both Gpr161 vl/vl Modvl4 C/M and Gpr161 vl/vl Modvl4 M/M mice display significantly (~35%) lower incidence compared to Gpr161 vl/vl Modvl4 C/C (Fig 1D). We conclude that Modvl4 MOLF partially rescues Gpr161 vl/vlassociated cataract in a dominant manner.

Subcongenic analysis defines the Modvl4 cataract modifier interval
According to the mouse genome (GRCm38/mm10 Assembly), more than 350 genes and ESTs are located in the Modvl4 95% CI, making it difficult to determine which genes contribute to the rescue of the cataract phenotype. To narrow down the interval, Modvl4 C/M congenic was further backcrossed to C3H isogenic mice. Subcongenic progeny with recombination breakpoints within the Modvl4 95% CI was selected by genotyping 14 SSLP markers spanning Modvl4. Three subcongenic lines were generated: two of them (Sub1 and Sub2) contain the MOLF background for proximal Modvl4 but differ in the proximal breakpoint, whereas the  The effect of the subcongenics on the cataract phenotype was quantified using the same method mentioned above. For Gpr161 vl/vl Sub1 and Sub2 heterozygotes (Gpr161 vl/vl Sub1 C/M , Gpr161 vl/vl Sub2 C/M ) as well as Gpr161 vl/vl Sub3 heterozygotes and homozygotes (Gpr161 vl/vl Sub3 C/M and Gpr161 vl/vl Sub3 M/M ), no significant difference is observed when compared to the Gpr161 vl/vl Modvl4 C/C mice. However, for Gpr161 vl/vl Sub1 and Sub2 homozygotes (Gpr161 vl/vl Sub1 M/M and Gpr161 vl/vl Sub2 M/M ), the incidence of cataract is reduced by~60%, and is significantly lower than Gpr161 vl/vl Modvl4 C/C (Fig 1C and 1D).
Since both the Sub1 and Sub2 homozygotes can rescue the cataract phenotype while the Sub3 MOLF congenic has no effect, these results indicate that the MOLF background within either Sub1 or Sub2 is sufficient to partially rescue Gpr161 vl/vl -associated cataract in a recessive manner. Because the Sub1 MOLF congenic is contained within Sub2 MOLF (S1 Fig, panels C and D), Sub1 delimits the minimal region sufficient for Modvl4 MOLF cataract rescue. In addition, because the original Modvl4 MOLF congenic (which is equal to the sum of the subcongenics) rescues cataract in a dominant manner, it suggests that genetic modifiers in the Sub3 interact with proximal Sub1 and Sub2 modifiers to improve the rescuing effect from a recessive to dominant mode of inheritance.
While it would be interesting to identify the MOLF modifiers in Sub3 that contribute to the complex inheritance pattern of cataract in Gpr161 vl/vl mutants, this is difficult because 95% (more than 300) of the Modvl4 genes are located in the Sub3 MOLF interval. In addition Sub3 MOLF does not function independently and must interact with the proximal Sub1 MOLF or Sub2 MOLF region to rescue the cataract phenotype. Therefore, we decided to focus on the cataract repressors that are situated in the Sub1 MOLF and Sub2 MOLF region. Because Sub1 MOLF is smaller than Sub2 MOLF and has the same rescuing effect as Sub2 MOLF , we decided to focus all subsequent analyses on Sub1 MOLF . Sub1 MOLF partially rescues lens fiber defect during secondary lens fiber differentiation Previous analysis determined that Gpr161 vl/vl lens is normal until E14.5. Starting from E14.5, lens fiber disorganization is observed and by E16.5, 100% of the Gpr161 vl/vl lenses are abnormal with disorganizations specifically localized to the posterior medial and nasal bow regions [11]. To test whether Sub1 MOLF rescues these defects during development, a histological analysis was performed at E16.5.
A careful examination of all the serial transverse sections of E16.5 lenses (1675 sections in total from 14 Gpr161 +/+ Sub1 C/C , 13 Gpr161 vl/vl Sub1 C/C , 11 Gpr161 vl/vl Sub1 C/M and 16 Gpr161 vl/vl Sub1 M/M lenses) revealed that the abnormal phenotypes can be categorized into four groups: 1) sections with normal lens fiber, which have the typical packing of differentiated lens fiber cells (Fig 2A and 2B); 2) sections with mild lens fiber defects which display disorganized lens fibers that are restricted only to the posterior medial region (Fig 2C and 2D); 3) sections with moderate lens fiber defects which have disorganized lens fibers in both the posterior medial and nasal bow regions of the lens (Fig 2E and 2F) and 4) sections with severe lens fiber defects that display lens fiber disorganization in posterior medial and nasal bow regions, as well as vacuoles in the nasal bow region (Fig 2G and 2H). All 490 sections derived from 14 wild-type Gpr161 +/+ Sub1 C/C lenses display normal lens fiber phenotypes. The 325 sections generated from 13 mutant Gpr161 vl/vl Sub1 C/C lenses, however, displayed all four phenotypic categories (Fig 2I and for individual lens, refer to S1 Table). Because serial sections were generated and placed in an ordered fashion on the microscopic slides, their relative position along the anterior-posterior axis can be estimated. We found that sections with more severe lens fiber defects are from the equatorial region of the lens, whereas sections with less severe defects are anterior or posterior of the equatorial region (see Fig 2K for illustration). These results suggest that there is an area of disorganized lens fibers in the center of the Gpr161 vl/vl lens, which could potentially be the precursor of the postnatal cataract phenotype.
The 359 sections from 11 Gpr161 vl/vl Sub1 C/M lenses and 501 sections from 16 Gpr161 vl/vl Sub1 M/M lenses were examined using the same criteria. Interestingly, no severe lens fiber defect is observed in either Gpr161 vl/vl Sub1 C/M or Gpr161 vl/vl Sub1 M/M lenses. Only normal, mild, and moderate phenotypes are detected. In addition, when compared to Gpr161 vl/vl Sub1 C/C , a higher percentage of normal sections is observed (Fig 2I). In the congenics, sections with a moderate defect occupy the equatorial zone instead of sections with severe phenotypes as observed in Gpr161 vl/vl Sub1 C/C . In addition, the moderate phenotype was sandwiched by sections with mild phenotype (Fig 2K). These results indicate that on the Sub1 MOLF congenic background, there is a less severe lens fiber defect, suggesting a partial rescue by Sub1 MOLF modifiers.
To investigate if the above differences among Gpr161 vl/vl Sub1 C/C , Gpr161 vl/vl Sub1 C/M and Gpr161 vl/vl Sub1 M/M lenses are statistically significant, the frequencies for each of the four phenotypes were determined for each genotype. Pairwise comparisons among genotypes were made using Mantel-Haenszel test with non-parametric (rank) score [14]. Statistical significances were observed for all three comparison pairs (Gpr161 vl/vl Sub1 C/C vs Gpr161 vl/vl Sub1 C/M ; Gpr161 vl/vl Sub1 C/M vs Gpr161 vl/vl Sub1 M/M and Gpr161 vl/vl Sub1 C/C vs Gpr161 vl/vl Sub1 M/M ) ( Fig  2J), indicating that Sub1 MOLF can partially suppress the severity of Gpr161 vl/vl -associated lens fiber defect in a semi-dominant manner, confirming Modvl4 MOLF partially rescues lens defect in Gpr161 vl/vl during development.

Modvl4 MOLF does not rescue Gpr161 vl/vl -associated lethality
In addition to congenital cataract, we also examined the lethality phenotype in the Modvl4 MOLF congenic and subcongenics. We have previously determined that~36% of the Gpr161 vl/vl mice on the C3H isogenic background die before weaning and other Modvl modifiers rescue the vlassociated lethality [13]. To test if Modvl4 MOLF also rescues Gpr161 vl/vl -associated lethality, the Gpr161 +/vl Modvl4 C/C x Gpr161 +/vl Modvl4 C/M cross was performed and 291 adult progeny were genotyped. Taking into account the 36% lethality for the Gpr161 vl/vl mutation, Gpr161 vl/vl Modvl4 C/C and Gpr161 vl/vl Modvl4 C/M mice are expected to account for 8% of the total progeny. Chi-square test determined that for both genotypes, the observed numbers do not significantly deviate from the expected numbers (S2 Table, top section), consistent with the Modvl4 C/M congenic not rescuing lethality. To investigate whether being homozygous for the congenic affects shown for Gpr161 +/+ Sub1 C/C , Gpr161 vl/vl Sub1 C/C , Gpr161 vl/vl Sub1 C/M and Gpr161 vl/vl Sub1 M/M . (J) Pairwise comparisons between genotypes were performed using Mantel-Haenszel test with rank score. (K) The distribution of lens fiber phenotypes along the A-P axis is shown by representative illustrations for each of the four genotypes (green: normal; yellow: mild defect; orange: moderate defect and red: severe defect). In Gpr161 vl/vl mutant background, transverse sections that are closer to the lens equatorial region display more severe defects. In the Modvl4 MOLF congenic background, a partial rescue of the defect is observed by the reduction in severe lens fiber defect, and the expansion of normal and mild phenotypes.  Table, bottom section). Similar mating strategies and statistical methods were also used to determine the lethality incidences on the three subcongenic backgrounds, and again no rescuing effect was observed (See S3 Table for details). Taken together, these results demonstrate that the Modvl4 MOLF congenic and subcongenics do not rescue the Gpr161 vl/vl -associated lethality.

Fine-mapping and annotation of the 15 Mb minimal region
Next we more precisely determined the recombination breakpoints for the minimal chromosomal interval that rescues the cataract phenotypes. Based on the previous genotyping results (S1 Fig, panel D), we expect the proximal border (marked by proximal border of Sub1) of the interval to be between D15MIT51 (Chr15: 12280730) and D15MIT81 (Chr15: 15365366), and the distal border (marked by proximal border of Sub3) to be between D15MIT94 (Chr15: 27443957) and D15MIT204 (Chr15: 32994622) (Fig 3A). To more precisely define the position of the recombination breakpoints, genomic DNA of the two subcongenic lines was genotyped for 11 additional SSLP and SNP markers. This analysis demonstrated that the proximal border is between Chr15: 13062568-13100445 while the distal border is between Chr15: 27443957-28269407. Therefore, the modifier interval that rescues the cataract phenotype is delimited to a 15 Mb region between Chr15:13062568 and Chr15:28269407 ( Fig 3B).
The interval 15:13062568-28269407 was searched in UCSC Genome Browser (http:// genome.ucsc.edu/cgi-bin/hgGateway; GRCm38/mm10 assembly) and Ensembl Genome Browser (http://www.ensembl.org; Build 75). Interestingly, a large part of the 15 Mb interval falls into a gene desert and only 20 genes are annotated for both genome browsers (Fig 3C; Also see Table 1 for descriptions of each gene).

Identifying candidate QTGs and QTNs for the Modvl4 MOLF minimal region
Next we wanted to identify the candidate genes (QTGs) and genetic variations (QTNs) from the 20 genes that could be responsible for rescuing the lens fiber and cataract phenotypes. Four criteria were used: Expression analysis. We would expect cataract QTGs to be expressed in the developing eye so the expression of the 20 genes were checked by RT-PCR using E16.5 eye cDNA from Gpr161 +/+ Sub1 C/C . Fifteen of the 20 genes show detectable expression ( Fig 4A). No expression was observed for Acot10, Fam134b, Mir7212, Mir7117 and Dnahc5, suggesting that they are not involved in the Sub1 MOLF rescue of the lens defects (data not shown).
Biological relevancy to lens development and Gpr161 signaling. Another criterion to identify candidate QTGs is the biological function of the 15 genes that display positive expression. Published literature was searched for their biological relevancy to lens development and pathways known to be downstream of Gpr161, including Shh, retinoic acid, and canonical Wnt signaling [13,15]. Four of the genes regulate canonical Wnt signaling: Cdh6, Basp1, Ank and Fam105b. [16][17][18][19][20]. In addition, Trio is required for lens pit invagination [21] and is a guanine nucleotide exchange factor known to be downstream of GPCR signaling [22]. None of the 15 genes were found to be relevant to Shh or retinoic acid pathways.  The proximal extent of the minimal region that rescues the Gpr161 vl/vl cataract phenotype is defined by the proximal border of Sub1 whereas the distal extent is defined by the proximal border of Sub3 (blue dashed arrows; also refer to Fig 1C and 1D). (B) To determine the proximal and distal breakpoints of the 15 Mb cataract modifying region, SSLP and SNP genotyping was performed using genomic DNA sample from Sub1 and Sub3 congenic mice. Left: the proximal border was previously determined to be between D15MIT51 and D15MIT81 (blue dashed arrow). Further analysis delimited the proximal border to be between D15MIT265 and D15MIT53. Only one gene Cadherin 6 (Cdh6) is within this interval. Additional SNP genotyping determined that the proximal border is distal to rs45839473 (Chr15:13062568; between exon 3 and 4 of Cdh6). Right: the distal border was previously determined to be between D15MIT94 and D15MIT204 (blue dashed arrow). Additional SSLP and SNP markers further determined the identified (S4 Table). 10 SNPs and 1 in-frame deletion are non-synonymous variations that affect amino acid sequences in five genes (Cdh12, Basp1, Myo10, March11 and Ank).
To evaluate the functional impact of the 10 nsSNPs, online software was used. Polyphen (http://genetics.bwh.harvard.edu/pph2/) and SIFT (http://sift.jcvi.org/) were employed to border to be proximal to rs32933300 (Chr15:28269407; in exon 18 of Dnahc5). (C) The Rs45839473 SNP sits between Exon 3 and 4 of Cdh6 while the Rs32933300 is within Exon 18 of Dnahc5 (red arrows). In total 18 protein coding genes and 2 miRNAs are within the 15 Mb region flanked by the two SNPs (Chr15: 13062568-28269407). The genes are aligned across the chromosome based on their genetic loci. All labels of base pair information are based on mouse genome assembly GRCm38/mm10. predict impact of variants on protein function based on evolutionary conservation and available 3D crystal structures. Only V201A in Ank displayed positive results using both online tools ( Table 2). Ank encodes a multipass transmembrane transporter that regulates intra-and extracellular concentrations of pyrophosphate (PPi) [23]. The protein is highly conserved across species with only 8 out of the 492 amino acids being different between human and mouse. The Valine allele in MOLF background is conserved from zebrafish to human, as well as among 24 out of the 28 inbred mouse strains, consistent with it being important in protein function. To determine the impact of the 15bp inframe deletion in Myo10, online software PROVEAN (http://provean.jcvi.org/index.php) was employed [24]. The five-amino acid deletion, SELAE/-, is predicted to be tolerant to protein structure (PROVEAN score = -0.137; Scores below -2.5 are considered deleterious) and consistently, it sits within a region that is only conserved among rodents. In conclusion, only Ank has one coding polymorphism that is likely to affect the structure and the function of the encoded protein.
To model the structural and potential functional impact of V201A on Ank, we predicted its structure based on homology to known proteins using TMHMM server. Ank is projected to contain eight transmembrane-spanning helices (S3 Fig, panel A). The V201A substitution is situated within the fourth transmembrane helix, and interestingly sits in a groove that is known to mediate helix-helix interactions in the membrane (S3 Fig, panel B) [25][26][27]. Typically these motifs are of the form AxxxBC where A and B are small amino acids (Gly, Ser, Ala) that are flanked by C, a beta-branched amino acid (Val, Ile) [25]. The Val to Ala substitution at 201 alters the beta-branched nature of this flanking sidechain, potentially affecting Ank folding or protein-protein interactions between Ank and other binding partners in the membrane. Sub1 qRTPCR analysis. Finally we also quantified the expression levels of the 15 genes in E16.5 eyes from Gpr161 +/+ Sub1 C/C , Gpr161 vl/vl Sub1 C/C and Gpr161 vl/vl Sub1 M/M . Ten of the 15 genes display significantly reduced expression in Gpr161 vl/vl Sub1 C/C compared to Gpr161 +/+ Sub1 C/C , coinciding with the lens fiber defects appeared at this stage of development ( Fig 4B). Interestingly, when the expression level of those ten genes were further measured in Gpr161 vl/vl Sub1 M/M , which partially rescues lens fiber phenotypes, five genes display restored expression levels with 4 of them (Cdh6, Cdh12, Fbxl7 and Ank) show wild type-like expression and for Trio, a partial restoration of expression is observed (Fig 4C). We conclude that the expression levels of Cdh6, Cdh12, Fbxl7, Ank and Trio are correlated to the presence of cataract in Gpr161 vl/vl Sub1 C/C , and the rescue of cataract in the congenic background (Table 3 second  column).
Next, we investigated if genetic variation exists in the non-coding regions of these five genes that could explain the expression differences between Gpr161 vl/vl Sub1 C/C and Gpr161 vl/vl Sub1 M/M . Flanking sequences (up to 50kb 5' and 3' of the gene) and intronic sequences were searched for all five genes (http://www.sanger.ac.uk/sanger/Mouse_SnpViewer/rel-1410) and a total of 129 Indels and SNVs were identified (S5 Table). We then investigated if any of these variants fall within predicted transcription factor binding sites by using online resources (http://www.gene-regulation.com/pub/programs/alibaba2/index.html). By comparing the results between C3H and MOLF, all genes were found to have multiple variants (89 total) that are predicted to affect transcription factor binding. We conclude that MOLF variants exist within 50kb of Cdh6, Cdh12, Fbxl7, Ank and Trio that could explain the expression differences between Gpr161 vl/vl Sub1 C/C and Gpr161 vl/vl Sub1 M/M . Lastly, we also measured Gpr161 expression using E16.5 eye cDNA. The expression level dropped by 70% in Gpr161 vl/vl Sub1 C/C but is not rescued in Gpr161 vl/vl Sub1 M/M (Fig 4D). This result indicates that Sub1 MOLF is a bypass allele that rescues cataract by either acting through a parallel pathway, or restoring downstream signaling activities of Gpr161, compensating for the loss of Gpr161 function in Gpr161 vl/vl . Combining all criteria to identify candidate QTGs. The Complex Trait Consortium (CTC) has established eight criteria for identification of candidate QTG [28]. The four criteria we described above belong to three of those eight criteria: 1) expression in the appropriate target tissue(s) or cell type(s); 2) polymorphisms in coding or regulatory regions and 3) published in vitro/in vivo functional studies. Generally, CTC requests at least two criteria to be fulfilled in order to consider a gene as QTG [28]. Among all 15 genes that are expressed in developing eyes, a total of 3 genes (Cdh6, Ank and Trio) have fulfilled at least two criteria (Table 3). Although this does not rule out the possibility that the other 12 genes may also contribute to the rescue of lens phenotypes by Sub1 MOLF , we conclude that these three genes have the strongest evidence for contributing to the rescue of Gpr161 vl/vl cataract phenotype.

Discussion
Phenotypic variability is commonly observed for human disease, affecting the penetrance and severity of most disorders [29][30][31][32][33]. For congenital cataract, studies have shown that the same mutation can result in differences in the morphology, location, color and density of cataracts. These results suggest that unlinked genetic variants can contribute to the modification of the primary mutation. While the influence of genetic background on the penetrance and expressivity of human congenital cataract has been postulated, how this may occur has not been investigated previously [3,7,8]. In this study, we took advantage of the natural genetic variations among different inbred mouse strains to model how unlinked modifiers could affect a primary mutation and affect the penetrance of congenital cataract. Gpr161 vl is a unique polygenic mouse model for congenital cataract. We previously determined that the cataract incidence in Gpr161 vl/vl is variable among different inbred strains, with 100% penetrance on the C3H/HeSnJ background but only a 15% penetrance when crossed to the MOLF/EiJ background [11,12]. In this study, we focused on the Modvl4 MOLF locus and by generating congenic and subcongenics for Modvl4 MOLF , we identified a 15Mb interval on proximal Chromosome 15 that is sufficient to partially restore normal lens fiber development. Three genes (Cdh6, Ank and Trio) were identified as candidate QTGs with coding and noncoding QTNs that are predicted to be functional. Our studies suggest that at least one of those QTGs and QTNs can affect the penetrance of the Gpr161 vl mutation, providing insight into how human congenital cataract may be modified by genetic background. In addition, since some of these genes have not been implicated in lens development previously, their further characterization may also provide new insight into the molecular basis of lens development and cataract.

Modvl4 MOLF as a modifier of congenital cataract
Our previous congenic analysis for a different QTL, Modvl5 MOLF (Chr18; LOD = 5.0), determined that Modvl5 MOLF specifically rescues the lethality and neural tube defects (NTDs) associated with the Gpr161 vl/vl mutation, but did not affect congenic cataract [13]. Interestingly, we show in this study that Modvl4 MOLF has no effect on the lethality but instead partially rescues the cataract and lens fiber phenotypes. While it remains to be determined whether Modvl4 also plays a role in modifying the NTD phenotypes, these two studies indicate that the pleiotropic effects of the Gpr161 vl/vl mutation on lethality and lens development are mediated by both Gpr161 and unlinked genes situated on different chromosomes. By generating a Modvl4-Mod-vl5 MOLF double congenic, it will be interesting to investigate whether Modvl4 and 5 act independently or play synergistic roles in regulating these Gpr161 vl/vl mutant phenotypes.
To narrow down the modifying interval of Modvl4, we generated three subcongenic lines for morphological analysis. Interestingly, our subcongenic studies revealed an unexpected, complicated inheritance pattern of the cataract modifying effect. We determined that the whole region of Modvl4 MOLF represses cataract in a dominant manner, whereas Sub1 and Sub2, which have the MOLF background in the proximal portion of Modvl4, repress cataract in a recessive manner. Sub3, which has MOLF background in the distal portion of Modvl4, has no effect. These results indicate that while the QTGs in proximal Modvl4 (Sub1 and Sub2) are sufficient to rescue cataract, additional QTGs in distal Modvl4 (Sub3) genetically interact with proximal QTGs to improve the rescuing effect from a recessive to dominant mode of inheritance.
Within the scope of this study, we focused on proximal Modvl4 that overlap in large part with the Sub1 interval. Future work beyond the scope of this manuscript will investigate QTGs from the distal region that contributes to the complex inheritance pattern of the cataract modifying effect. As distal Modvl4 interacts with proximal Modvl4, one hypothesis would be QTGs from distal region share similar functions as proximal QTGs to regulate lens development. For example, as our analysis identified the cell adhesion molecule Cdh6 as a candidate QTG from Sub1, we might expect some QTGs in the distal Modvl4 to be involved in cell adhesion. One potential candidate in the distal region is Cadherin-Associated Protein, Delta 2 (Ctnnd2), which forms a complex with cadherins to mediate intercellular adhesion [34,35].

E16.5 lens histological analysis revealed a partial rescue of lens fiber defect
Interestingly, although 30-50% of the adult eyes for the Modvl4 congenic and subcongenic animals have no visible opacity, our histological analysis on E16.5 lens sections revealed that almost all (26 out of the 27 lens analyzed, refer to S1 Table) of the Gpr161 vl/vl Sub1 C/M and Gpr161 vl/vl Sub1 M/M lens display different severities of abnormal lens fiber phenotypes. The severity can be scored based on the size of affected area and the presence/absence of vacuoles in the nasal bow region. A careful inspection of all transverse sections along the A-P axis of each individual lens demonstrates that Gpr161 vl/vl Sub1 C/M lens has a decreased number of sections with severe and moderate defects and a greater number of sections with mild and normal phenotypes, compared to the Gpr161 vl/vl Sub1 C/C littermates. In addition, Gpr161 vl/vl Sub1 M/M embryos show a further shift from severe to normal phenotypes, compared to Gpr161 vl/vl Sub1 C/M embryos. These results indicate that the congenic mice display a smaller area of lens fiber abnormality compared to the C3H isogenic mutant mice.
Adult lens fiber disorganization is associated with lens opacity, reduced transmission of light through the lens, and decreased sight. In the congenic background, a less severe lens fiber defect diminishes the incidence of opacity. This would improve the transmission of light and the ability to see, although it remains to be determined if actual improvement of visual ability is achieved in the congenic. In addition, being heterozygous or homozygous for the Sub1 MOLF interval can repress the lens fiber defect during development, but only the Sub1 M/M subcongenic rescues congenital cataract in the adult, suggesting that there is a threshold in the lens fiber defect that determines the presence of lens opacity.

Identifying candidate QTGs/QTNs from the 15 Mb interval of proximal Chr15
Analysis of our three subcongenic lines narrowed down the modifying interval from a 55Mb region with more than 350 genes/ESTs to a 15Mb region flanked by Rs45839473 and Rs32933300 that contains only 20 genes. Our expression analysis identified 15 genes with detectable mRNA in the developing eye. While it remains formally possible that the other 5 unexpressed genes may affect the Gpr161 vl/vl lens phenotype through a non-cell autonomous mechanism or by being expressed earlier in development, we decided to focus on the 15 expressed genes. To further narrow down the 15 genes, three criteria were used: 1) whether the expression level of the gene at E16.5 eye is disrupted in Gpr161 vl/vl and restored by the congenic, consistent with flanking/intronic C3H/MOLF variants regulating mRNA levels; 2) whether the gene has coding nonsynonymous variations that are predicted to affect the protein structure and 3) whether published manuscripts have determined that the gene has relevancy to lens development or Gpr161 downstream signaling. In total, we identified three genes (Cdh6, Ank and Trio) that fulfilled at least two criteria, and are considered as candidate QTGs of Modvl4 according to the rationale established by Complex Trait Consortium [28]. This analysis has also identified candidate genes that could function during lens development and Gpr161 signaling. For instance, Trio is a Guanine Exchange Factor (GEF) so it may regulate G protein signaling downstream of Gpr161 in the lens. Interestingly, Trio also regulates lens pit invagination during development [21]. Cadherins are important for many different steps in lens development. For instance a switch from E-to N-cadherin expression is needed for secondary lens fiber differentiation, which is the stage when the Gpr161 vl/vl mutation affects lens development [36]. Cdh6 also regulates axon-targeting in the visual circuit, and differentiation of retinal ganglion cells, amacrine cells, and photoreceptors, were disrupted in Cdh6 zebrafish morphant [37,38]. The identification of Cdh6 as a Modvl4 MOLF QTG suggests that this cadherin may also be important for lens development. Finally, Ank is a membrane transporter for pyrophosphate (PPi). Pyrophosphate is generated during the hydrolysis of ATP to AMP. Maintenance of extracellular and intracellular concentrations of pyrophosphate is needed for normal articular cartilage cellular function [39]. In addition, PPi has also been shown to be a signaling molecule important for the generation of inositol pyrophosphates which regulate numerous processes including metabolic homeostasis and apoptosis [40]. The identification of Ank as a candidate QTG for Modvl4 suggests that PPi levels and/or signaling through PPi may be important for lens development.
In summary, using Modvl4 MOLF congenic as a model, we studied the multigenic basis of Gpr161 vl/vl -associated congenital cataract. We demonstrated that a 15Mb interval in proximal Modvl4 MOLF genetically interacts with Gpr161 to partially rescue the lens fiber orientation defect and congenital cataract. Among all 20 genes situated in this region, we determined three genes (Cdh6, Ank and Trio) as candidate QTGs by multiple criteria. This study provides new insight into the multigenic basis of congenital cataract and identified novel candidate genes for future investigation.

Modvl4 MOLF congenic analysis
This study was carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. All the rodents used were approved by the IACUC of Rutgers University-Robert Wood Johnson Medical School (Protocol number is I12-113). Pregnant females and adult mice of either sex were anesthetized by intraperitoneal injection of Euthasol (Nembutal: Phenytoin solution; 200mg/kg), followed by cervical dislocation.
Modvl4 MOLF congenic mice were generated by backcrossing C3H/MOLF hybrid to C3H isogenic background for 8 generations (for detailed description, refer to S1 Fig, panel A). Progeny of each generation were genotyped for 65 SSLP markers that span the genome as described in [11].
The Gpr161 +/vl Modvl4 C/M x Gpr161 +/vl Modvl4 C/C and Gpr161 +/vl Modvl4 M/M x Gpr161 +/vl Modvl4 M/M matings were performed to generate adult progeny. Tail samples of post-weaning adults were collected and processed with Wizard SV genomic DNA purification kit (Promega). The DNA samples were genotyped for Gpr161 and Modvl4 95% CI (PCR condition: 94˚C 15 s, 51˚C 15 s, 72˚C 45 s-30 cycles). For Gpr161 vl/vl mutation, a primer pair flanking the 8-bp deletion as described in [11] was used. For Modvl4, 14 SSLP MIT markers that are evenly distributed within the Modvl4 55 Mb interval (from proximal to distal: D15MIT51, D15MIT81, D15MIT130, D15MIT252, D15MIT267, D15MIT94, D15MIT204, D15MIT49, D15MIT24, D15MIT86, D15MIT59, D15MIT195, D15MIT27 and D15MIT257) were genotyped (primer sequences are listed at genome.ucsc.edu). To test the effect of the Modvl4 MOLF on Gpr161 vl/vlassociated lethality, the number of progeny of different genotypes were determined and compared to expected number using a Chi-square test as described in [13]. To test the effect of the Modvl4 MOLF congenic on congenital cataracts, adult Gpr161 vl/vl progeny derived from the two matings listed above were inspected for the presence or absence of opacity in the eye. Mice were sacrificed by cervical dislocation and placed under a stereomicroscope (Nikon SMZ800) for phenotypic inspection (5X magnification). Pictures of the left and right eyes were then taken by digital camera (SPOT RT Color).

Modvl4 MOLF subcongenic analysis
To generate subcongenic mice, the Gpr161 +/vl Modvl4 C/M x Gpr161 +/vl Modvl4 C/C mating was performed. Progeny were genotyped for the 14 SSLP markers described above as well as for the Gpr161 vl mutation. Gpr161 +/+ or Gpr161 +/vl animals with at least one recombination breakpoint within Modvl4 MOLF were used as founders to generate subcongenic lines (S1 Fig, panels  C and D). Lethality and cataract phenotypes were tested in the subcongenic lines as described above.
Histological analysis on the developing lens Heads from E16.5 embryos were fixed overnight with Bouin's Solution (Sigma) and stored in 70% EtOH at 4˚C. The tissue was then dehydrated, embedded in paraffin, and transversely sectioned by rotary microtome (Leica RM2135, 10 μm in thickness). Serial paraffin sections were sequentially aligned on glass slides according to their location along the A-P axis. The paraffin sections were then processed using a standard Nissl staining protocol. Sections were inspected under the microscope (Nikon Eclipse E600) for quantification of the lens fiber defects. To determine the statistical significance of the rescuing effect of the congenics, the number of sections with different severities of lens fiber defects was compared in pairs using Mantel-Haenszel test with non-parametric (rank) score [14].
After the break points were mapped between the two adjacent SSLP markers, SNPs were selected within the region for additional fine mapping. Genomic sequences spanning the SNP loci were PCR amplified and sequenced to determine whether C3H or MOLF allele of the SNP is present. The following primers were used: Rs48108000: F: GGAGAACCCCTCACGGAATAG TG; R: CCAAAGCCCCCAGTCTGATTG.Rs45839473: F: TATCGTGCTGGGACTTGAGACG;R: TTTGTTTTGGCGTGGGCTG. Rs47611653: F: AGGTGAATGAGAGAGGAGAGGAAAC; R: GGAAGGTGACAAATGATAGTTGGG. Rs31630379: F: GAGAGGAGGACTTTGTTACAGAGGC; R: GGCGTTTGGATTTGAACCG. Rs32933300: F: TGTGCTCTTTCTGCTCTTCCTGAC; R: AAATGACATCTCCCCCTCACCC.

Gene expression analysis
RNA was extracted from E16.5 eyes using standard phenol-chloroform extraction method. The pellet was resuspended into 10ul nuclease-free water. 20ul of cDNA was generated from 4ul of RNA using SuperScript™ II Reverse Transcriptase (Invitrogen). For RT-PCR, 2ul of 1:10
In our modeling using TMHMM, the V201A mutation lies in the center of the fourth predicted membrane-spanning domain. The length of membrane spanning domains (~20 residues) is consistent with a polytopic alpha-helical protein. The fourth domain (residues 191-213) was modeled as an alpha-helix using standard amino acid conformations in the molecular visualization platform pyMol (Version 1.7.0.3 Schrödinger, LLC).

Identification of coding and non-coding genetic variants of candidate genes
The Mouse Genome Project Team at Wellcome Trust Sanger Institute has used next-generation sequencing to sequence key laboratory mouse strains including C3H/HeJ and MOLF/EiJ, and the data were uploaded to a query website (http://www.sanger.ac.uk/sanger/Mouse_SnpViewer/ rel-1410). For identification of coding genetic variants, the minimal cataract modifying interval (Chr15: 13062568-28269407) as determined by subcongenic analysis was searched in the query site. Results were exported as a spreadsheet and C3H/MOLF genetic variants for the fifteen genes that are expressed in E16.5 eyes were selected (258 SNPs and 37 INDELs). 10 SNPs and 1 Indel were found to affect the amino-acid sequence of five genes (Cdh12, Basp1, Myo10, March11 and Ank), which were then selected for additional bioinformatics analyses.
C3H/MOLF genetic variants in the flanking and intronic sequences were also identified for five genes (Cdh6, Cdh12, Fbxl7, Ank and Trio) that display reduced expression in Gpr161 vl/vl Sub1 C/C but partial or full restoration of mRNA levels in Gpr161 vl/vl Sub1 M/M . The transcription start sites (TSSs) were first determined by searching Genome browser (GRCm18/mm10). Using the same query site mentioned above, the 5kb upstream promoter regions for all five genes were then searched (TSS location: Cdh6: 13173639; Cdh12: 21111452; Fbxl7: 26895564; Ank: 27466677 and Trio: 28025848). A total of 88 SNPs and 15 Indels were identified for three of the five genes, Cdh6, Cdh12 and Ank. Because no variations were identified within the 5 kb upstream region for Fbxl7 and Trio, the search was extended to 50kb upstream promoter region for Fbxl7. For Trio, both upstream and downstream 50kb sequences as well as the intronic region were searched. 14 SNPs and 12 Indels were identified for the two genes (S5 Table). In total 129 genetic variants were identified for additional transcription factor binding prediction analysis.