Skip to main content
  • Loading metrics

Gpr63 is a modifier of microcephaly in Ttc21b mouse mutants

  • John Snedeker,

    Roles Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration, Visualization, Writing – original draft, Writing – review & editing

    Affiliations Division of Human Genetics, Cincinnati Children’s Hospital Medical Center, Cincinnati, Ohio, United States of America, Neuroscience Graduate Program, University of Cincinnati College of Medicine, Cincinnati, Ohio, United States of America

  • William J. Gibbons Jr.,

    Roles Formal analysis, Investigation, Writing – review & editing

    Affiliation Division of Human Genetics, Cincinnati Children’s Hospital Medical Center, Cincinnati, Ohio, United States of America

  • David F. Paulding,

    Roles Data curation, Formal analysis, Investigation, Writing – review & editing

    Affiliation Division of Human Genetics, Cincinnati Children’s Hospital Medical Center, Cincinnati, Ohio, United States of America

  • Zakia Abdelhamed,

    Roles Conceptualization, Data curation, Investigation, Validation, Visualization, Writing – review & editing

    Affiliations Division of Human Genetics, Cincinnati Children’s Hospital Medical Center, Cincinnati, Ohio, United States of America, Department of Anatomy and Embryology, Faculty of Medicine (Girl’s Section), Al-Azhar University, Cairo, Egypt

  • Daniel R. Prows,

    Roles Conceptualization, Formal analysis, Funding acquisition, Investigation, Project administration, Visualization, Writing – original draft, Writing – review & editing

    Affiliations Division of Human Genetics, Cincinnati Children’s Hospital Medical Center, Cincinnati, Ohio, United States of America, Department of Pediatrics, University of Cincinnati College of Medicine, Cincinnati, Ohio, United States of America

  • Rolf W. Stottmann

    Roles Conceptualization, Data curation, Funding acquisition, Investigation, Methodology, Project administration, Supervision, Visualization, Writing – original draft, Writing – review & editing

    Affiliations Division of Human Genetics, Cincinnati Children’s Hospital Medical Center, Cincinnati, Ohio, United States of America, Neuroscience Graduate Program, University of Cincinnati College of Medicine, Cincinnati, Ohio, United States of America, Department of Pediatrics, University of Cincinnati College of Medicine, Cincinnati, Ohio, United States of America, Division of Developmental Biology, Cincinnati Children’s Hospital Medical Center, Cincinnati, Ohio, United States of America, Shriner’s Hospital for Children - Cincinnati, Cincinnati, Ohio, United States of America


The primary cilium is a signaling center critical for proper embryonic development. Previous studies have demonstrated that mice lacking Ttc21b have impaired retrograde trafficking within the cilium and multiple organogenesis phenotypes, including microcephaly. Interestingly, the severity of the microcephaly in Ttc21baln/aln homozygous null mutants is considerably affected by the genetic background and mutants on an FVB/NJ (FVB) background develop a forebrain significantly smaller than mutants on a C57BL/6J (B6) background. We performed a Quantitative Trait Locus (QTL) analysis to identify potential genetic modifiers and identified two regions linked to differential forebrain size: modifier of alien QTL1 (Moaq1) on chromosome 4 at 27.8 Mb and Moaq2 on chromosome 6 at 93.6 Mb. These QTLs were validated by constructing congenic strains. Further analysis of Moaq1 identified an orphan G-protein coupled receptor (GPCR), Gpr63, as a candidate gene. We identified a SNP that is polymorphic between the FVB and B6 strains in Gpr63 and creates a missense mutation predicted to be deleterious in the FVB protein. We used CRISPR-Cas9 genome editing to create two lines of FVB congenic mice: one with the B6 sequence of Gpr63 and the other with a deletion allele leading to a truncation of the GPR63 C-terminal tail. We then demonstrated that Gpr63 can localize to the cilium in vitro. These alleles affect ciliary localization of GPR63 in vitro and genetically interact with Ttc21baln/aln as Gpr63;Ttc21b double mutants show unique phenotypes including spina bifida aperta and earlier embryonic lethality. This validated Gpr63 as a modifier of multiple Ttc21b neural phenotypes and strongly supports Gpr63 as a causal gene (i.e., a quantitative trait gene, QTG) within the Moaq1 QTL.

Author summary

TTC21B in humans is a known ciliopathy gene and contributes to the pathophysiology of a number of ciliopathies. Mice homozygous for a null allele of Ttc21b also have a spectrum of ciliopathy phenotypes, including microcephaly (small brain). Further work has shown that the severity of the microcephaly significantly depends on the genetic background of the mouse model. The genetic mechanisms of the Ttc21b pathophysiology and the interacting gene network remain far from understood. As an initial attempt to understand the underlying mechanism(s) underlying the variable effects on brain size, we performed a quantitative trait locus (QTL) analysis and found two regions of genomic significance that correlated with smaller brain size. We confirmed both QTLs with congenic lines. One of the two regions was small enough that we considered candidate genes and hypothesized Gpr63 might be a contributing locus for a number of reasons. We evaluated this hypothesis directly with precise variant creation using genome editing and provide evidence that Ttc21b and Gpr63 do indeed genetically interact. Thus, we have been able to combine classical QTL analysis and genome editing to directly test the resulting hypothesis.


Primary cilia are microtubule-based organelles known to play essential roles in proper development and function of a number of organ systems including the central nervous system (CNS) [14]. Ciliopathies are a class of human diseases caused by pathogenic variants in genes encoding proteins responsible for the proper form and function of cilia [57]. The frequent presentation of cognitive impairment in ciliopathy patients, in addition to severely compromised brain development in a number of ciliary mutant mouse models, clearly displays the importance of primary cilia in CNS health and development [8, 9]. Primary cilia have been implicated in transducing and regulating several critical developmental pathways including SHH, WNT, PDGF, TGFβ/BMP, RTK, and Notch [3, 10]. An array of cell surface receptors, including G-protein coupled receptors (GPCRs), localize to the ciliary membrane to modulate these pathways and other signaling events [11]. Intraflagellar transport (IFT) proteins are responsible for the movement of cargo within the cilium with complex-B proteins regulating anterograde transport from the basal body to the distal tip of the cilium and complex-A proteins regulating retrograde transport [12]. IFT-A proteins have also been shown to specifically play a role in the trafficking of select GPCRs in and out of the cilium [1315].

Tetratricopeptide repeat domain 21b (Ttc21b; Thm1; Ift139; MGI 1920918) encodes an IFT-A protein necessary for the proper rate of retrograde trafficking within the primary cilium [16]. TTC21B is not part of the “core IFT-A complex” [15]. When TTC21B is reduced, a large portion of the remaining IFT-A protein subunits remains intact as this “core” complex and continues to interact with the dynein motor powering retrograde movement. Thus, TTC21B likely acts to facilitate binding of peripheral cargo proteins to the core proteins of the complex, thus assisting the transport of these cargo proteins through the cilium. Proper ciliary trafficking is known to be critical for the processing of GLI transcription factors, which are in turn essential for proper regulation of the SHH pathway [17, 18]. Mutant mice homozygous for the alien (aln) null allele of Ttc21b display impaired processing of GLI3, the primary repressor of SHH target genes, and consequently show an increase in SHH pathway activity in multiple tissues including the developing forebrain [16, 19].

The ciliopathies are only one example of a class of human diseases with broad phenotypic variability. The advent of genome sequencing is facilitating the study of genes and variants affecting disease penetrance and expressivity. The ability to specifically model such modifiers in experimental organisms like mice with genome editing should allow more in-depth studies of genetic interactions and their role in human disease. It is well known that genetic backgrounds of mouse inbred strains can affect phenotypic severity and penetrance of specific mouse mutations. These differences can be used to detect genetic interactions in complex traits and begin to understand the mechanisms leading to this variability [20, 21]. One example of genetic background influencing the expressivity of a phenotype is the forebrain malformations seen in Ttc21baln/aln mutants. Here we present a QTL analysis to further understand the underlying pathogenic molecular mechanism leading to the microcephaly. We demonstrate that Gpr63 interacts with Ttc21b to modify the size of the forebrain in Ttc21baln/aln null mutants and provide evidence that the Gpr63 mutation is the likely causal variant for the chromosome 4 QTL affecting brain size.


Ttc21baln microcephaly is background dependent

The phenotypes observed in Ttc21baln/aln mutants were first identified as part of an ENU mutagenesis experiment to identify recessive developmental phenotypes in the perinatal mouse [22]. A/J mice were treated with the ENU mutagen and the subsequent breeding and mapping of the causative mutation involved an outcross to the FVB/NJ (FVB) strain [16, 22]. Ttc21baln/aln homozygous mutants on this mixed A/J;FVB background exhibited multiple forebrain and craniofacial phenotypes (Fig 1A–1D) [16, 19]. While the microcephaly phenotype was completely penetrant, craniofacial phenotypes such as cleft lip and palate were only present in a minority of embryos. To increase the frequency of craniofacial phenotypes and facilitate future molecular analyses, the Ttc21baln allele was serially backcrossed onto the C57BL/6J (B6) strain. The B6 genetic background has previously been shown to be more susceptible to craniofacial phenotypes in a number of studies [2326]. This initial, exploratory backcross was done informally and was not genotyped. After the Ttc21baln allele was serially backcrossed to the B6 background for at least five generations, we noticed the relative size of the forebrain tissues in B6.Cg-Ttc21baln/aln mutants increased substantially as compared to the mutants previously analyzed, although they remained smaller in all Ttc21baln/aln mutants than control brains and still lacked olfactory bulbs (Fig 1E and 1F). To confirm this was indeed an effect of varying the genetic background, the Ttc21baln allele was again serially backcrossed to the FVB background for at least three generations. Embryonic analysis showed the severity of microcephaly initially seen was recovered in the new FVB.Cg-Ttc21baln/aln mutants (Fig 1C and 1D). We measured the forebrain surface area of Ttc21baln/aln mutants on both FVB and B6 genetic backgrounds (Fig 1G) and saw a 60% decrease in forebrain size in FVB.Cg-Ttc21baln/aln animals as compared to littermate controls (n = 7 wt and n = 4 aln mutants; p<0.0001), but only a 30% decrease in B6.Cg-Ttc21baln/aln animals (n = 4 wt and n = 3 aln mutants; p = 0.00033). The interaction between genetic background and the Ttc21baln allele was analyzed with a 2-way ANOVA (p = 0.0053).

Fig 1. Genetic background affects the Ttc21baln/aln microcephaly phenotype.

(A-F) Embryos and whole mount brains viewed from the dorsal aspect from control (A,B, E17.5), FVB;Cg-Ttc21baln/aln mutants (C,D, E18.5), and B6;Cg-Ttc21baln/aln mutants (E,F, E17.5). (G) Relative forebrain dorsal surface areas between wild-type and Ttc21baln/aln mutants from each genetic background. Data shown are mean ± 95% confidence interval. n = 7, 4, 4, 3 sample points. Scale bars in A-F indicate 1mm.

Given all of this preliminary data, we hypothesized the genetic background effect on Ttc21baln/aln mutant forebrain size would allow a Quantitative Trait Locus (QTL) analysis to generate insight into the underlying molecular mechanism(s). In order to generate mice as genetically uniform as practical prior to performing a QTL analysis, we further backcrossed both strains of Ttc21baln mice. After at least three further generations of backcrossing onto FVB (FVB;B6-Ttc21baln/aln) and four generations onto B6 (B6;FVB-Ttc21baln/aln), we performed a genome-wide high-density MegaMUGA SNP scan and confirmed that each strain was approximately 90% pure (S1 Table). We further purified the FVB and B6 advanced backcross lines with at least two additional backcrosses combined with targeted microsatellite marker screening of chromosomal areas not yet homozygous for FVB or B6 as identified in the initial genome scan. This allowed us to create FVB.Cg-Ttc21baln/wt animals that were >99% FVB and B6.Cg-Ttc21baln/wt animals purified to ~97% B6 background, as assayed by a GigaMUGA genome SNP scan providing strain-specific sequence information at over 143,000 SNPs [27] (S1 Table).

As we performed this breeding, we continued to note an incompletely penetrant exencephaly phenotype in Ttc21baln/aln animals, but the frequency did not appear to significantly differ between backgrounds (Table 1). We calculated the incidence of exencephaly and found that maintaining the Ttc21baln allele on either the FVB or B6 genetic backgrounds yielded exencephaly in ~40% of Ttc21baln/aln mutants (24/58, 41% in FVB;B6-Ttc21baln/aln and 30/65, 46% in B6;FVB-Ttc21baln/aln). Note these embryos were collected for other experimental purposes as the incipient congenics were being further purified and are on a mixed genetic background.

We first generated F2 Ttc21baln/aln mutant progeny from a B6.FVB F1 (Ttc21baln/wt) intercross for the QTL analysis. Fifty-four F2 litters were generated, which identified 148 Ttc21baln/aln mutant embryos from 607 total embryos. We excluded 35 mutants (24%) with exencephaly, which precluded a measurement of forebrain area. Ninety-six brains from the remaining F2 mutants were micro-dissected and the dorsal surface area of the forebrain was measured. Whole genome SNP genotyping of each F2 recombinant was carried out using a GigaMUGA panel and QTL mapping was performed using R/qtl [28] to identify chromosomal regions linked to differential brain sizes in Ttc21baln/aln mutants. The dataset used in the QTL analysis is presented in S2 Table and an overview of the QTLs identified and their effects on the phenotype is provided in Table 2.

A significant QTL (genomewide p < 0.05; LOD 6.1) was mapped to chromosome 4 with its peak at 27.8 Mb (marker UNC6953268) and a 95% confidence interval spanning 19.5 Mb to 46 Mb (Fig 2A and 2B). This QTL explained 25.1% of the phenotypic variance (Table 2). We have named this chromosome 4 locus modifier of alien QTL 1 (Moaq1). A second QTL was identified with a LOD score suggesting significance (genomewide p < 0.63; LOD 4.7) and mapped to chromosome 6 at 93.6 Mb (UNCHS018175) with a 95% confidence interval from 0 to 117.5 Mb (Fig 2A and 2C). This chromosome 6 QTL explained 20.1% of the trait variance and was named modifier of alien QTL 2 (Moaq2). A third peak was found on chromosome 2 at the Ttc21b locus, which likely represents the A/J passenger background on which the initial mutagenesis was performed (Fig 2A). To assess the contribution of each locus to the trait, we determined the change in LOD score and percent variance explained for the three possible 2-loci models and the 3-loci model (Table 2). A significant increase in LOD score (>+3) was found for all comparisons, indicating that all three loci (aln and the 2 modifiers) significantly added to the phenotype, and together they explained 44.4% of the phenotypic variance. To assess whether the three loci interacted with each other and/or were additive, we performed a scantwo analysis in R/qtl (Table 2), using 1,000 permutations to establish the threshold levels for significance. Results indicated that none of the loci interacted ( for all 3 pairings, not significant). However, the main effects of both Moaq1 and Moaq2 were independently additive with aln. The additive effect of Moaq1 and aln was highly significant (lod.add = 9.13) and that for Moaq2 and aln was significant (lod.add = 8.47). Interestingly, main effects of Moaq1 and Moaq2 were not additive (lod.add = 5.44; significance threshold = 8.29).

Fig 2. QTL analysis.

(A) QTL LOD plots with dotted lines displaying genomewide significance scores of p = 0.05 and p = 0.63. (a) marks the significant QTL peak on chromosome 4, Moaq1, and (b) marks the suggestive Moaq2 peak on chromosome 6. (B) Chromosome 4 QTL LOD plot and 95% confidence interval shown in red. (C) Chromosome 6 LOD plot and 95% confidence interval. (D) Analysis of brain size in Ttc21baln/aln homozygous for FVB genome identity, heterozygous for FVB/B6 and homozygous for B6 genome identity as genotyped by the peak marker for Moaq1. n = 24, 51, 21 sample points. (E) Analysis of brain size only among littermates in litters with both Ttc21baln/aln-Moaq1FVB/FVB and Ttc21baln/aln-Moaq1FVB/B6 or B6/B6 embryos (n = 15 litters). (F) Analysis of brain size in animals homozygous for FVB genome identity, heterozygous for FVB/B6 and homozygous for B6 genome as genotyped by the peak marker for Moaq2. n = 24, 51, 21 sample points. (G) Analysis of brain size only among littermates in litters with both Ttc21baln/aln-Moaq2FVB/FVB and Ttc21baln/aln-Moaq2FVB/B6 or B6/B6 embryos (n = 6 litters).

We confirmed a correlation between the genetic identity of Moaq1 and Moaq2 and an effect on forebrain size by comparing the genotype of the peak SNP marker for each QTL with forebrain size (Fig 2D and 2F). We first independently assessed F2 mice for genotype at the peak marker of the Moaq1 locus and found an incremental decrease in brain size among Ttc21baln/aln-Moaq1FVB/FVB animals as compared to Ttc21baln/aln-Moaq1FVB/B6 or Ttc21baln/aln-Moaq1B6/B6 animals (Fig 2D). An overall ANOVA was performed to identify significant differences among these groups (p<0.0001) followed by a subsequent Tukey’s multiple comparison test analysis for comparison of each group (Ttc21baln/aln-Moaq2FVB/FVB, Ttc21baln/aln-Moaq2FVB/B6 and Ttc21baln/aln-Moaq2B6/B6) with individual p-values ranging from 0.0145 to <0.0001 (Fig 2D). To control for a potential litter-based difference, we also compared only Ttc21baln/aln-Moaq1FVB/FVB mutants from litters with Ttc21baln/aln-Moaq1FVB/B6 or Ttc21baln/aln-Moaq1B6/B6 littermates (Fig 2E). Although this analysis necessarily used a smaller sample size than the non-litter matched comparison (Fig 2D), we again observed a 1.94mm2 reduction in Ttc21baln/aln-Moaq1FVB/FVB animals (p = 0.012).

A similar analysis testing just the Moaq2 locus alone showed a significant variation among Ttc21baln/aln-Moaq2FVB/FVB, Ttc21baln/aln-Moaq2FVB/B6 and Ttc21baln/aln-Moaq2B6/B6 animals (Fig 2F, ANOVA p = 0.0243). However, the subsequent analysis identified less striking differences between the groups (Fig 2F; p = 0.0255, 0.0547, 0.8654). We again complemented this with a matched littermate analysis and found Ttc21baln/aln-Moaq2FVB/FVB animals had approximately 20% smaller forebrains than littermates (Fig 2G; p = 0.0101).

Congenic lines

We produced independent congenic lines to replicate the Moaq1 and Moaq2 loci. We noted in our original F2 population of 96 mice for the QTL analysis that no Ttc21baln/aln mutants were homozygous for B6 SNPs over a 10-Mb region on chromosome 14. We hypothesized this represented a region of the B6 genome that, in combination with homozygosity for the Ttc21baln allele, may lead to either early embryonic lethality and/or exencephaly. Thus, we attempted to avoid this complication and create FVB congenic lines with Moaq1 and Moaq2 genomic regions derived from the B6 strain with the hypothesis that the B6 genomic identity of Moaq1 and Moaq2 would increase the size of the Ttc21baln/aln over the size of the congenic FVB Ttc21baln/aln mutants. B6;FVB-Ttc21baln/wt F1 mice, (obligate heterozygotes across their genome), were repetitively backcrossed with FVB mice to produce a congenic strain that approached inbred FVB except for the heterozygous Moaq1 or Moaq2 B6 QTLs (i.e., FVB.Cg-Moaq1FVB/B6 and FVB.Cg-Moaq2FVB/B6). Mice from each congenic line were then intercrossed to determine if the B6-derived Moaq1 and/or Moaq2 QTLs could influence forebrain size on their own. During the process of creating these congenics, we noted that successive generations of mating onto the FVB background was leading to lethality of Ttc21baln/aln mutants prior to E17.5 or exencephaly. We observed Ttc21baln/aln mutants on a >99% FVB background and 21/22 either died prior to E17.5 or had exencephaly precluding measurement of brain size (Table 1). Note that while we had previously created mice that were largely enriched for FVB, even as much as 90%, (e.g., Fig 1), the incipient congenics created here for homozygote analysis are the purest FVB-Ttc21baln/wt population created to date. (While creating the mice used to perform the initial QTL analysis, we had not been intercrossing and dissecting embryos to create Ttc21baln/aln homozygotes so had not previously identified this effect on the phenotype.) This suggests a possibility for future work identifying a separate exencephaly modifier from this population. Given that the increasing incidence of exencephaly precluded an analysis of forebrain size in the congenic lines (N5 and greater), we analyzed animals from the earlier backcross generations. Littermate brains were analyzed using paired t-tests to account for potentially confounding variables, such as length of gestation and/or background differences based on the generation of backcross. The Moaq1 QTL was found to affect brain size in FVB;B6-Ttc21baln/aln mutants in the N2 to N5 generations. Ttc21baln/aln-Moaq1FVB/B6 and Ttc21baln/aln-Moaq1B6/B6 animals had larger brains as compared to Ttc21baln/aln-Moaq1FVB/FVB (Fig 3A, p = 0.0177), with an average decrease of 2.24mm2 (Fig 3B). The Moaq2 QTL was also validated using this strategy and Ttc21baln/aln-Moaq1FVB/B6 and Ttc21baln/aln-Moaq1B6/B6 animals again had larger brains than Ttc21baln/aln-Moaq1FVB/FVB (Fig 3C, p = 0.0201), with an average decrease of 3.40mm2 (Fig 3D). Thus, we conclude the QTL analysis of just 96 B6;FVB-Ttc21baln/aln F2 forebrains revealed two validated, novel QTLs that modify the effect of the Ttc21baln/aln mutation on forebrain size.

Fig 3. Congenic line validation of Moaq1 and Moaq2.

(A) Forebrain size from Ttc21baln/aln-Moaq1FVB/FVB (FVB/FVB) and Ttc21baln/aln-Moaq1FVB/B6 or B6/B6 (FVB/B6 or B6/B6). (B) The amount of forebrain reduction in (FVB/B6 or B6/B6)–(FVB/FVB) brain sizes shows an average decrease of 2.24mm2. (C,D) Parallel analyses for Moaq2 animals are shown.

Candidate genes for Moaq1 QTL

In order to identify potential candidate causal genes within the ~26.5 Mb Moaq1 QTL, the 253 genes within the 95% confidence interval were analyzed (Fig 4; S3A Table). We initially focused our analysis on strain-specific missense coding variants. We recognize that this initial analysis of missense variants, rather than a parallel approach first examining the strain-specific insertions and/or deletions, introduces some ascertainment bias towards different candidate genes. We also could include RNA expression analysis of mutants from both background and look for non-coding regulatory variants within the QTL. However, we initially reasoned that missense coding variants might be best suited to act as modifiers than more severe protein perturbations resulting from insertion/deletion events. We first identified 33 genes with missense polymorphisms when comparing B6 and FVB genomic sequences (S3B Table). The list was further refined to 21 (S3C Table) by selecting only those genes in which the A/J and FVB SNP were identical, given the phenotypic similarities in Ttc21baln/aln mutants on those genetic backgrounds. The remaining missense mutations were analyzed with the SIFT prediction algorithm to assess pathogenicity [29], as well as information on known phenotypes in mouse models or human disease (OMIM), expression data and other information from the literature (S3C Table). Four genes were found with predicted deleterious missense mutations (<0.05 SIFT value). We then performed a literature review of these four remaining genes (Table 3, S3D Table) to identify those previously known to have a role in developmental neurobiology and narrowed the list to 3 candidates: gamma-glutamyl hydrolase (Ggh, MGI 1329035), G-protein coupled receptor 63 (Gpr63, MGI 2135884), and origin recognition complex subunit 3 (Orc3, MGI 1354944). Ggh is a lysosomal enzyme with a role in folate metabolism, which is provocative given the exencephaly noted in Ttc21baln/aln mutants [30]. Furthermore, Ggh is the only gene in the interval with 2 potentially deleterious missense mutations. Orc3 is a nuclear localized component of the Origin Recognition Complex [31, 32]. While conditional ablation of Orc3 in the brain revealed a requirement in proper neural progenitor and radial glial development, germline loss of Orc3 led to early embryonic lethality [33]. We next considered any evidence for these remaining candidate genes within the Moaq1 QTL to have a role in primary cilia biology. Unlike Ggh or Orc3, Gpr63 is a membrane protein, making it the more attractive candidate to be a modifier of a mutation in a plasma membrane-based organelle, the primary cilium. An siRNA screen to identify novel regulators of ciliogenesis in vitro did provide some evidence Gpr63 may have a role in cilia biology [34]. Gpr63 has been reported to have brain-specific expression and to be enriched in the forebrain [35, 36]. The predicted deleterious polymorphic SNP within Gpr63 (rs13477613; NM_030733 c.G1635T; NP_109658 p.R407S; SIFT score of 0.01) is a missense variant altering the coding of amino acid 407 from Arginine, an amino acid with an electrically charged side chain, in the B6 reference genome to Serine, an amino acid with a polar uncharged side chain, in the A/J and FVB genomes (Table 3). This Arginine residue is conserved from frogs to chimpanzees (Fig 5A), although humans have a Lysine at the orthologous residue in GPR63, an amino acid with an electrically charged side chain like Arginine. The loss of Arginine from the mouse GPR63 cytoplasmic tail alters a predicted RxR endoplasmic reticulum (ER) retention signal (Fig 5) [37]. The C-terminus of GPCRs is understood to play critical roles in protein-protein interactions and regulation of the receptor [38]. As shown in Fig 6A, the c.G1635T SNP changes the amino acid sequence of the coat protein complex I (COPI) binding sequence. COPI is a coatomer responsible for the retrograde transport of vesicles from the trans-Golgi network to the cis-Golgi and from the cis-Golgi to the ER [39, 40]. COPI has been implicated in the retrieval of proteins with exposed RxR motifs [41, 42]. These motifs function to retain proteins in the ER until the RxR containing protein is completely processed and ready to be transported out through the Golgi [37]. Often these proteins escape the RxR retrieval mechanism by masking the motif through interaction with another protein, a critical step in preparing the protein for its function which may include assembling a hetero-multimeric complex [37]. There are candidate motifs for protein domain interactions on the GPR63 C-tail which could serve in this masking role including: LPRLPGH, a predicted SH3 binding motif of DLG4, and EHRTV, a predicted PDZ binding domain of RGS3 [4346]. We also note a BBSome binding motif which may serve as a potential ciliary targeting sequence [47].

Fig 4. An outline of the analysis leading to the hypothesis Gpr63 is a causal gene in the Moaq1 interval.

Fig 5. Gpr63 is a conserved gene.

Amino acid sequence of GPR63 from multiple species and three inbred strains of mice. Invariant amino acids are indicated with *. Shading indicates amino acids composing four putative functional domains in the cytoplasmic tail of GPR63.

Fig 6. CRISPR-Cas9 alleles of Gpr63.

(A) Nucleotide sequence and Sanger sequencing of Gpr63 in the B6/reference strain and FVB strain with the c.G1635T SNP. Gpr63Arg(B6) is the FVB mouse line with the B6 Thymidine at Gpr63 c.1635. Gpr63Del is an 8-bp deletion in the same region of Gpr63. (B) Amino acid sequence of the above lines showing the R407S polymorphism between B6 (Arg) and FVB (Ser) reverted to B6 (Arg) in the CRISPR transgenic, and the missense mutation and premature truncation in the Gpr63Del allele.

CRISPR-Cas9 genome editing of Gpr63

A standard approach to reduce the number of candidate genes within a QTL is to narrow the genetic interval carrying the trait of interest using meiotic recombination. For this, QTL-interval specific recombinants are identified and overlapping sub-congenics produced and tested to determine the minimal region of effect. This process can be repeated as needed to critically narrow the candidate interval. We initially pursued this strategy, but the results above show that the combination of exencephaly and early lethality of congenic FVB.Cg-Ttc21baln/aln mice will preclude successful implementation unless we could also identify the region contributing to the exencephaly.

An alternative approach is to directly test hypotheses about candidate sequence variant(s) with genome editing to create novel transgenic models. We chose to pursue this and targeted FVB mice to change the rs13477613 polymorphism from FVB sequence (c.1635-T; coding for Gpr63Ser(FVB)) to the reference, non-pathogenic, B6 sequence (c.1635-G; coding for Gpr63Arg(B6)). This allele Gpr63em1Rstot, is hereafter referred to as FVB-Gpr63Arg(B6). An added benefit of CRISPR-Cas9 genome editing is the potential to recover an allelic series from the initial set of transgenic founders. We recovered and maintained a second allele Gpr63em2Rstot (Gpr63Del), with an 8-bp deletion resulting in a frameshift mutation that leads to the translation of 5 missense amino acids and an early stop, thus preventing translation of the last 21 amino acids of the GPR63 C-terminus (Fig 6A). This Gpr63Del mutation may disrupt both an ER retention signal and an RGS3 PDZ binding motif, creating a potentially more deleterious mutation than Gpr63Ser(FVB) [37, 4346].

We first determined that the Gpr63Arg/Ser polymorphism did not affect survival or brain size at E17.5 (Fig 7A, ANOVA p = 0.1743). We also did not see an effect at postnatal day (P) 42 on either forebrain dorsal surface, brain weight or total body weight (Fig 7B–7D, ANOVA p = 0.9639, 0.7753, 0.9420, respectively). This was expected, given that these are sequence polymorphisms found between two commonly maintained inbred strains. We hypothesized the Gpr63Del allele may be more deleterious than the Gpr63Ser(FVB) polymorphism alone. We first determined if this allele affects survival and created animals homozygous for the Gpr63Del allele. We recovered Gpr63Del/Del mutants in approximately Mendelian ratios at weaning (8/35) and they survived to adulthood (n = 8) with no discernible phenotypes seen to date. We measured forebrain dorsal surface, brain weight and body weight at both E17.5 and P42 and found no change in the Gpr63Del/Del mutants (Fig 7E–7L; ANOVA p = 0.2569, 0.6846, 0.6344, 0.6655, respectively). Consistent with the predicted deleterious effects of the SNP at amino acid 407 seen in both viable and fertile FVB and A/J animals, this suggests that Gpr63 does not have an essential requirement in development.

Fig 7. Gpr63 alleles do not affect survival or overall brain size and morphology.

(A-D) Animals with Gpr63407Arg/Arg, Gpr63407Arg/Ser, or Gpr63407Ser/Ser do not vary in forebrain area at E17.5 (A) or at P42 in forebrain area (B), brain weight (C) or overall body weight (D). (E-L) Animals homozygous for the cytoplasmic deletion of Gpr63 (Gpr63Del/Del) also have morphologically normal brains at E17.5 (E,F) and P42 (G,H). (I) Forebrain area at E17.5 is unaffected. At P42, forebrain area (J), brain weight (K) and overall body weight (L) are unaffected by the deletion allele. Scale bars indicate 2mm in E,F and 5mm in G,H.

Gpr63 expression in vitro shows ciliary localization

We directly tested the hypothesis that Gpr63 might, at least partially, localize to the primary cilium by measuring in vitro expression of a myc-flag-epitope tagged construct in NIH 3T3 cells (pCMV6-GPR63-myc). Indeed, we found colocalization of the GPR63 signal with Arl13b as a marker of the ciliary axoneme (Fig 8A–8C). We were not able to identify a completely reliable antibody to verify these findings, but saw similar results from an independent GPR63 expression construct (GPR63-TANGO-FLAG [48], S2 Fig). Site-directed mutagenesis was used to generate constructs recapitulating the variants seen in vivo (Fig 8D–8I). Overexpression of either polymorphism of Gpr63 or the Gpr63-deletion construct did not affect the proportion of ciliated cells (Fig 8M, ANOVA p = 0.067). However, when we expressed either the myc-Gpr63Ser or the myc-Gpr63Del construct we saw a reduction in the number of transfected cells with GPR63 in the primary cilia (Fig 8N, ANOVA p = 0.004). We saw a 65% decrease with Gpr63Ser407 as compared with Gpr63Arg407 (p = 0.011) and a 68% decrease with the Gpr63Del construct (p = 0.008). We conclude that the different alleles changing the protein composition of the GPR63 cytoplasmic tail can alter the subcellular (i.e., ciliary) localization of GPR63.

Fig 8. Ciliary localization is altered by overexpression of Gpr63 variants.

NIH 3T3 cells were transfected to overexpress Gpr63Arg407 (A-C), Gpr63Ser407 (D-F) or Gpr63Del (G-I) constructs. ARL13B was used as a marker of the ciliary axoneme. (J-L) An untransfected immunocytochemistry control is shown. Small panels show higher magnification of cells highlighted by white boxes to show cilia with robust, no, or little GPR63 immunoreactivity. (M) Expression of Gpr63 did not appear to affect the number of cells producing cilia. (N) Gpr63Ser407 and Gpr63Del constructs result in reduced ciliary accumulation of the epitope-tagged GPR63. Scale bars represent 20μm in A-L and 5μm in the inset panels.

Alleles of Gpr63 modify Ttc21baln/aln mutant forebrain phenotype

The FVB-Gpr63Arg(B6) animals were crossed with FVB.Cg-Ttc21baln/wt (which will have the FVB allele of Gpr63: Gpr63Ser(FVB)) mice to produce FVB.Cg-Ttc21baln/wt;Gpr63Arg(B6)/Ser(FVB) double heterozygotes. This cross was designed to directly test the hypothesis that the predicted pathogenic polymorphism of Gpr63 (i.e., Ser407) contributes to the more severe microcephaly seen in FVB.Cg-Ttc21baln/aln homozygous mutants. However, we were unable to directly test this hypothesis. As this experiment was done concurrently with the backcrosses to produce FVB congenic lines described above, we also found that both FVB.Cg-Ttc21baln/aln;FVB.Gpr63Arg(B6)/Arg(B6) and FVB.Cg-Ttc21baln/aln;FVB.Gpr63Ser(FVB)/Ser(FVB) mice have exencephaly at a high frequency precluding measurement of forebrain size in these animals.

In order to test for a genetic interaction between Gpr63 and Ttc21b, we created FVB.Cg-Ttc21baln/wt-Gpr63Del/Ser(FVB) mice and crossed these with B6.Cg-Ttc21baln/wt (with no modification of the Gpr63 locus, e.g., B6.Cg-Ttc21baln/wt-Gpr63Arg(B6)/Arg(B6). We hypothesized that reduced GPR63 function in the resulting F1 FVB;B6-Ttc21baln/aln;Gpr63Del/Arg(B6) embryos as compared to F1 FVB;B6-Ttc21baln/aln-Gpr63 Ser(FVB)/Arg(B6) will lead to a smaller forebrain in the F1 FVB;B6-Ttc21baln/aln-Gpr63Del/Arg(B6) animals (i.e., deletion of the GPR63 cytoplasmic tail is more deleterious than a single Ser/Arg amino acid change). Indeed, F1 FVB;B6-Ttc21baln/aln-Gpr63Del/Arg(B6) mice had an approximate 25% decrease in forebrain area (Fig 9; -2.30 mm2; n = 15 and n = 6; p = 0.0019). This experiment demonstrates that alleles of Gpr63 can modify the Ttc21baln/aln forebrain phenotype and supports the conclusion that Gpr63 genetically interacts with Ttc21b.

Fig 9. Gpr63 alleles affect forebrain area in Ttc21baln/aln mutants.

Forebrain area in FVB;B6-Ttc21baln/aln;Gpr63Arg(B6)/Ser(FVB) E17.5 embryos (n = 16) is larger than FVB;B6-Ttc21baln/aln-Gpr63Del/Arg(B6) animals (n = 6), p = 0.0019.

To further address a possible interaction between Ttc21b and Gpr63, we aimed to create Ttc21baln/aln;Gpr63Del/Del double homozygous mutant animals. Note this experiment was done in a way that necessitated crosses with differing mixtures of the FVB and B6 background. Therefore, the Gpr63;Ttc21b double mutant analysis is best considered on a hybrid background with Gpr63 alleles being the only relevant variable. We recovered 8 Ttc21baln/aln;Gpr63Del/Del embryos at E17.5 and E18.5 from this cross and found that 75% (6/8) were embryonic lethal and had died around E12.5. The remaining two had exencephaly precluding measurement of forebrain area.

We further explored this phenotype at ~E12.5 (E10.5–14.5) and recovered 9 Ttc21baln/aln-Gpr63Del/Del double homozygous null embryos (Fig 10F). We saw many of the phenotypes previously noted in our studies of Ttc21baln/aln mutants. These included 89% (8/9) with exencephaly, and 100% with polydactyly in both the forelimb and hindlimb (9/9). Surprisingly, we also noted phenotypes never seen in any Ttc21baln/aln mutant. One of the double mutant embryos had quite dramatic spina bifida aperta (Fig 10G), a severe neural tube defect. 62.5% (5/8) of the remaining embryos had morphology consistent with the considerably milder spina bifida occulta. Both of the previously mentioned ~E17.5 double mutant embryos also had obvious spina bifida occulta (Fig 10J). We then rigorously examined all the embryos recovered from this cross for spinal neural tube defects (NTD) as this is not a phenotype we had seen in previous analyses of the Ttc21baln/aln mutants [16, 19]. We did not note any spinal NTDs in Gpr63Del/Del; Ttc21bwt/wt (Fig 10B; n = 5) or Gpr63Del/Del; Ttc21baln/wt embryos (Fig 10D; n = 13). In Gpr63wt/wt; Ttc21baln/aln embryos, we did note a previously unappreciated thickened neural tissue in one mutant from E10-E14 (Fig 10C; n = 1/9), and a disruption of the overlying ectoderm possibly consistent with spina bifida occulta from E15-17 (Fig 10H; n = 1/4). Based on our previous studies with the Ttc21baln/aln mutant where the NTDs were not noticed in studies performed on similar genetic backgrounds, we believe these to be relatively uncommon phenotypes in mice with just the Ttc21baln allele [16, 19]. Gpr63Del/wt; Ttc21baln/aln embryos with an additional deleterious Gpr63 allele however, showed a significantly increased incidence of NTDs (Fig 10E and 10I, n = 7/14). However, we never see the spina bifida aperta phenotype in embryos other than Gpr63Del/Del; Ttc21baln/aln double mutants. Even in this genotype, we have seen it once in this study. We therefore suggest it is a relatively rare phenotype, but is still restricted to the Gpr63Del/Del; Ttc21baln/aln double mutant genotype.

Fig 10. Gpr63del/del;Ttc21baln/aln double mutants have increased incidence of neural tube defects.

Embryos of multiple genotypes from a Gpr63del/wt;Ttc21baln/wt intercross are shown at stages from E10-14 (A-G, K) or E15-17 (H-J). Gpr63del/del;Ttc21baln/aln double mutants do not survive to term and show multiple phenotypes which are more severe than either single mutation including spina bifida aperta (G). (L) Incidence of neural tube closure defects of any severity in the embryos resulting from the Gpr63del/wt;Ttc21baln/wt intercross. Note the dramatic increase in penetrance in double mutants (8/11) as compared to any other genotype.

Thus, we conclude that an allele of Gpr63 with a truncation of the cytoplasmic tail genetically interacts with Ttc21b, such that double mutants are mostly embryonic lethal at stages prior to any lethality seen in either single mutant. The remaining animals have an increased incidence of NTDs unique from either single mutant. The mechanism(s) leading to this lethality will be the subject of future study. We further suggest that Gpr63 is likely to be a causal gene within the Moaq1 QTL and reduced function of GPR63 in the FVB and A/J genetic backgrounds leads to the more severe forebrain phenotype of the Ttc21baln/aln homozygous mutants on these backgrounds, as compared to those recovered on a B6 inbred background.


In this study, we sought to understand the genetic basis for the influence of mouse inbred strain background on the severity of the Ttc21baln/aln homozygous mutant microcephaly phenotype. Our QTL analysis identified two loci (Moaq1, Moaq2) which interact with Ttc21b to control brain size in Ttc21baln/aln mutants. These were validated by constructing QTL-containing congenic lines for each. The 95% confidence interval of Moaq1 was small enough to allow a candidate gene approach, and Gpr63 was selected for further study. Experiments directly testing Gpr63 function with novel mouse alleles generated with CRISPR-Cas9 genome editing support our hypothesis that reduced Gpr63 function further compromises embryos lacking functional Ttc21b. We further demonstrated that Gpr63;Ttc21b double mutants had earlier lethality than either single mutant as well as novel spina bifida phenotypes.

Meiotic recombination or genome editing

Traditionally, the identification of a QTL is followed by a laborious mapping process and the creation of congenic strains to validate the QTL and begin to hunt for the truly causal sequence difference [49, 50]. The combination of deeply sequenced inbred mouse strains, combined with the capabilities conferred by new genome editing tools such as CRISPR-Cas9 to rapidly generate sequence specific alterations in novel mouse lines, can potentially greatly facilitate the identification of causal genetic variants within a QTL. In this case, we identified a potentially deleterious SNP in Gpr63 on the FVB and A/J genetic backgrounds which may cause reduced function as compared to the canonical sequence seen in B6 mice. We did pursue a traditional congenic backcross approach by continued backcrossing to FVB towards creating a true congenic line which would be “pure” FVB with only dilution by the Ttc21baln locus itself and the B6 Moaq1 region. However, this was confounded by the increased incidence of exencephaly and/or early embryonic lethality seen in Ttc21baln/aln mutants when maintained on the >90% FVB genetic background. Presumably this increased frequency in exencephaly and/or death was due to the remaining 10% of the heterozygous genome that was further purified to FVB during this backcrossing. This is an intriguing finding that could potentially facilitate mapping those modifiers independently. In parallel, we attempted to bypass the meiotic recombination and QTL refinement entirely by recreating the specific Gpr63 SNP and determining if it affected brain size in the Ttc21b mutant background. In the F2 recombinants generated for the QTL analysis, we did not recover embryos both without exencephaly and homozygosity for B6 at a region on chromosome 14. This suggested to us that the Ttc21baln allele on a pure B6 background may be early embryonic lethal; therefore, we chose to test our Gpr63 hypothesis by creating the B6 SNP on an FVB background (FVB Gpr63Arg(B6)/Ser(FVB)). Further study has shown this lack of homozygosity on chromosome 14 was a statistical anomaly with no apparent biological significance as we have since recovered Ttc21baln/aln mutants with B6 identity across this region. In contrast, the consequences of creating a pure inbred FVB strain with the Ttc21b allele was a strain with a high incidence of exencephaly preventing us from testing this Gpr63 hypothesis directly. We are now carrying out a further validation of the Moaq1 and Moaq2 QTL by constructing pure B6 congenics with FVB Moaq1 and Moaq2 intervals.

Gpr63 can localize to the primary cilia but is not essential for development

Our data suggest that the predicted deleterious Gpr63407Arg(B6)/Ser(FVB) SNP is not essential for development. This is consistent with the fact this SNP is present in the reference sequence for A/J and FVB mice. Furthermore, the deletion of the extreme C-terminus of the GPR63 GPCR also did not seem to have an effect on survival (Fig 7). This is consistent with data from the International Mouse Phenotyping Consortium and Knockout Mouse Project that Gpr63 deletion alleles do not affect survival (

We did use two different epitope-tagged expression constructs to demonstrate GPR63 at least partially localizes to the primary cilium (Fig 8). This is consistent with the BBSome binding motif located in the C-terminus. We further studied the ciliary localization of different alleles of Gpr63 in these overexpression experiments and have provided evidence that the Arg/Ser transition and the cytoplasmic truncation reduced ciliary localization of GPR63 (Fig 8). Neither of these residues affect the presumed BBSome binding motif, suggesting that other residues contribute to the ciliary targeting of GPR63 through currently unknown mechanisms. The detailed analysis of this trafficking motif and potential physical interaction with TTC21B should be the subject of future study.

Gpr63 and Ttc21b genetic interactions

The alleles we generated with CRISPR-Cas9 editing did allow us to demonstrate a genetic interaction between Gpr63 and Ttc21b. The confounding effects of FVB genetic purity on exencephaly and the apparent chromosome 14 lethality interval did hamper some of the initially designed experiments. However, the truncated Gpr63 allele we recovered from our CRISPR-Cas9 studies allowed us to address this hypothesis and we do see that Ttc21baln/aln;Gpr63Del/Arg(B6) embryos have smaller forebrain tissues when compared to Ttc21baln/aln;Gpr63 Ser(FVB)/Arg(B6 (Fig 9). The best evidence for a genetic interaction between Gpr63 and Ttc21b, however, is the novel phenotypes seen in the Ttc21baln/aln;Gpr63Del/Del embryos. The lethality of these embryos is earlier than what we observed in Ttc21baln/aln mutants and we have never previously seen some of the spina bifida phenotypes observed in the Ttc21baln/aln;Gpr63Del/Del embryos.

Here we identify regions of the genome linked to the differential effect of inbred strain background on the severity of the microcephaly phenotype seen in Ttc21baln/aln mutants. We follow this up with a new allelic series of Gpr63 and experiments in vitro and in vivo to demonstrate these polymorphisms do affect Ttc21b phenotypes and Gpr63 function. Although loss of Gpr63 alone does not seem to cause a significant phenotype, this work suggests Gpr63 may be a risk factor underlying some of the variability seen in the ciliopathies. The approach taken here may be quite useful in understanding modifier effects in any number of structural birth defects.

Materials and methods

Ethics statement

All animals were housed under a protocol approved by the CCHMC Institutional Animal Care and Use Committee (#2016–0098) in standard conditions. All euthanasia and subsequent embryo or organ harvests were preceded by Isoflurane sedation. Euthanasia was accomplished via dislocation of the cervical vertebrae.

Animal husbandry

For embryo collections, noon of the day of vaginal plug detection was designated as E0.5. The Ttc21baln allele used in this study has been previously published [16]. Ttc21baln mice were serially backcrossed to C57BL/6J (JAX:000664) and FVB/NJ (JAX:001800) mice to produce strain specific Ttc21baln mice. Genotyping was performed by PCR, Sanger Sequencing, or Taqman assays (S4 Table).

Forebrain surface area measurement

Brains were microdissected from embryos and photographed prior to fixation from the dorsal aspect on a Zeiss Discovery.V8 dissecting microscope. Zeiss Axiovision software was used to mark the boundaries of the cortical surface area for all brain examined, as shown in S1 Fig. Cortical surface area was quantified by the Zeiss Axiovision software and recorded. All images used to quantify surface area for the QTL analysis were taken on the same microscope at the same magnification.

QTL analysis

GigaMUGA and MegaMUGA genotyping of recombinants was performed by GeneSeek (Neogen; Lincoln, NE). Quality and intensity normalization of the SNP data was checked using routines provided in the R-package Argyle [51]. Linkage mapping, including scanone and scantwo analyses, was performed on informative SNP markers and brain sizes (mm2) of ninety-six F2 recombinants using the R/qtl software package (Broman 2003; Broman and Sen 2009). Threshold levels of significance were established using at least 1,000 permutations of the respective dataset.

Coding differences of candidate genes were compared using the Mouse Genomes Project website from Wellcome Sanger Trust Institute (, NCBIm37).

CRISPR-Cas9 genome editing

Donor oligonucleotide and guide RNA vector sequences were designed using Benchling (Benchling, 2016). sgRNAs were validated using Benchling and CRISPRscan with the following scores: Benchling-48.5 and CRISPRscan-49 (Benchling, 2016) (Moreno-Mateos, et al., 2015). The donor oligonucleotide was supplied by Integrated DNA Technologies (sequence with lower case letters representing coding change: CCATAATCGCTCGATATGTTTCAAGAGTTCGGTATTCACAACACCGTCCGATGTTCCCCACACACGTAGACGGCACTAGGGCGAATgCGTCGcCTTGTaTGcCCgGGGAGCCGTGGCAAGAACT. The sgRNA (sequence: TCCCTGGTCACACAAGTCGA) was synthesized and injected into FVB mouse zygotes together with donor oligonucleotide by the CCHMC Transgenic Animal and Genome Editing Core. Gpr63 mice were made by injection into FVB zygotes. The male founder mice were genotyped and those with evidence of edited GPR63 loci were bred to FVB-Ttc21baln/wt females. Offspring were analyzed by PCR and Sanger sequencing of approximately 1000 base pairs of the Gpr63 locus around the sgRNA recognition sequence. Initial crosses were designed to rapidly assess the hypothesized interaction between Gpr63 and Ttc21b. Carriers for the SNP allele (Gpr63Arg(B6)) were maintained by further breeding to FVB-Ttc21baln/wt females to immediately ascertain an effect on brain size. These were then further maintained on FVB to characterize the Gpr63 phenotype in isolation. Carriers for the Gpr63Del allele were also bred to FVB-Ttc21baln/wt (for at least two generations). In order to compare Gpr63Arg with the Gpr63Del allele, Gpr63Del;FVB-Ttc21baln/wt mice heterozygous for both alleles were crossed to B6 wild-type mice.

Cell culture and transfection

A pCMV6-GPR63-myc (Origene, Rockville, MD) or GPR63-TANGO-FLAG [48] ( expression plasmid was transfected into NIH 3T3 cells at ~80% confluency using Lipofectamine 3000 and incubated for 2–3 days. The pCMV6-GPR63-myc plasmid has a myc-DDK tag at the C-terminus of human GPR63 driven by a CMV promoter in the pCMV6 expression vector. GPR63-TANGO-FLAG is a plasmid designed to express human GPR63 with an N-terminal FLAG tag under the control of a CMV promoter from a Tango vector [48]. Mutations were made to the pCMV6-GPR63-myc expression plasmid to mimic the Gpr63Ser and Gpr63Del mouse alleles (Genscript, Piscataway, NJ). To induce ciliogenesis, cells were grown to confluency and serum starved. Immunocytochemistry was performed with a mouse anti-myc (Sigma M4439; 1:500), mouse anti-FLAG (Sigma F3165; 1:500), rabbit anti-Arl13b (Proteintech17711-1-AP; 1:500), and anti-mouse acetylated-tubulin (SIGMA T6793; 1:2000) antibodies to detect cilia and GPR63 localization. Secondary antibodies were Alexa Fluor-488, Alexa Fluor-594 Goat anti-mouse/rabbit (Invitrogen, A11001, A20980; 1:500). Confocal imaging was performed on a Nikon C2 system and analysis was performed with Nikon Elements software.

Analysis of localization of GPR63-myc was done with Imaris 6.2.1 software. First, a surface overlying cilia axonemes labeled with Arl13b was initiated in the 488 channel, then another surface was initiated in the 596 channel to detect the cilia labeled with myc and flag. Finally, the MATLAB plugin surface-surface colocalization was utilized to detect the areas of colocalization between Arl13b stained cilia and myc-flag tagged cilia axoneme. The number of double labeled cilia were normalized against the total number of Arl13b stained cilia per field of view. Statistical analysis and One-way ANOVA was performed using GraphPad Prism software. All samples were blinded to the investigator performing the analysis.

GPR63 protein domain analysis

Features of the GPR63 protein were identified through literature searches and the Modular Domain Peptide Interaction Server:

Statistical analysis

GraphPad Prism was used for all graphical analysis. Data are shown as mean ± 95% confidence interval unless otherwise noted. P values are explicitly stated in almost every instance, rather than asserting significance at a certain conventional threshold (e.g., p = 0.05). Fig 1G: Two-way ANOVA followed by unpaired t-test of each strain wild-type vs mutant; Figs 2D, 2F, 7A–7D, 7I–7L and 7O: One-way ANOVA followed by Tukey’s multiple comparisons when ANOVA indicates significance; Figs 2E, 2G and 3 –paired student t-test; Fig 9 –unpaired students t-test.

Supporting information

S1 Fig. Area measured to determine brain size.

Shaded area in B indicates area from brain in A measured as “forebrain area”.


S2 Fig. GPR63-TANGO-FLAG expression.

Transfection of a second GPR63-TANGO-FLAG expression plasmid [48] indicates ciliary localization similarly to the results with pCMV6-GPR63-myc.


S1 Table. Genome-wide SNP data.

Genome-wide SNP data to determine background purity; used for initial congenic FVB and B6 strains carrying Ttc21baln allele.


S2 Table. Genome survey and brain size of F2 mutants.

GigaMuga data and brain size of 96 F2 mutants used for initial QTL analysis.


S3 Table. Analysis of candidate genes within Moaq1 QTL.


S4 Table. Genotyping primers used in this study.



  1. 1. Guemez-Gamboa A, Coufal NG, Gleeson JG. Primary cilia in the developing and mature brain. Neuron. 2014;82(3):511–21. Epub 2014/05/09. pmid:24811376.
  2. 2. Valente EM, Rosti RO, Gibbs E, Gleeson JG. Primary cilia in neurodevelopmental disorders. Nat Rev Neurol. 2014;10(1):27–36. Epub 2013/12/04. pmid:24296655.
  3. 3. Goetz SC, Anderson KV. The primary cilium: a signalling centre during vertebrate development. Nat Rev Genet. 2010;11(5):331–44. Epub 2010/04/17. pmid:20395968.
  4. 4. Hildebrandt F, Benzing T, Katsanis N. Ciliopathies. N Engl J Med. 2011;364(16):1533–43. Epub 2011/04/22. pmid:21506742.
  5. 5. Wheway G, Mitchison HM. Opportunities and Challenges for Molecular Understanding of Ciliopathies-The 100,000 Genomes Project. Front Genet. 2019;10:127. Epub 2019/03/28. pmid:30915099.
  6. 6. Wang L, Dynlacht BD. The regulation of cilium assembly and disassembly in development and disease. Development. 2018;145(18). Epub 2018/09/19. pmid:30224385.
  7. 7. Badano JL, Mitsuma N, Beales PL, Katsanis N. The ciliopathies: an emerging class of human genetic disorders. Annu Rev Genomics Hum Genet. 2006;7:125–48. Epub 2006/05/26. pmid:16722803.
  8. 8. Han YG, Alvarez-Buylla A. Role of primary cilia in brain development and cancer. Curr Opin Neurobiol. 2010;20(1):58–67. Epub 2010/01/19. pmid:20080044.
  9. 9. Reiter JF, Leroux MR. Genes and molecular pathways underpinning ciliopathies. Nat Rev Mol Cell Biol. 2017;18(9):533–47. Epub 2017/07/13. pmid:28698599.
  10. 10. Wheway G, Nazlamova L, Hancock JT. Signaling through the Primary Cilium. Front Cell Dev Biol. 2018;6:8. Epub 2018/02/24. pmid:29473038.
  11. 11. Schou KB, Pedersen LB, Christensen ST. Ins and outs of GPCR signaling in primary cilia. EMBO Rep. 2015;16(9):1099–113. Epub 2015/08/25. pmid:26297609.
  12. 12. Prevo B, Scholey JM, Peterman EJG. Intraflagellar transport: mechanisms of motor action, cooperation, and cargo delivery. FEBS J. 2017;284(18):2905–31. Epub 2017/03/28. pmid:28342295.
  13. 13. Fu W, Wang L, Kim S, Li J, Dynlacht BD. Role for the IFT-A Complex in Selective Transport to the Primary Cilium. Cell Rep. 2016;17(6):1505–17. Epub 2016/11/03. pmid:27806291.
  14. 14. Hirano T, Katoh Y, Nakayama K. Intraflagellar transport-A complex mediates ciliary entry and retrograde trafficking of ciliary G protein-coupled receptors. Mol Biol Cell. 2017;28(3):429–39. Epub 2016/12/10. pmid:27932497.
  15. 15. Mukhopadhyay S, Wen X, Chih B, Nelson CD, Lane WS, Scales SJ, et al. TULP3 bridges the IFT-A complex and membrane phosphoinositides to promote trafficking of G protein-coupled receptors into primary cilia. Genes Dev. 2010;24(19):2180–93. Epub 2010/10/05. pmid:20889716.
  16. 16. Tran PV, Haycraft CJ, Besschetnova TY, Turbe-Doan A, Stottmann RW, Herron BJ, et al. THM1 negatively modulates mouse sonic hedgehog signal transduction and affects retrograde intraflagellar transport in cilia. Nat Genet. 2008;40(4):403–10. Epub 2008/03/11. pmid:18327258.
  17. 17. Huangfu D, Anderson KV. Cilia and Hedgehog responsiveness in the mouse. Proc Natl Acad Sci U S A. 2005;102(32):11325–30. Epub 2005/08/03. pmid:16061793.
  18. 18. Hui CC, Angers S. Gli proteins in development and disease. Annu Rev Cell Dev Biol. 2011;27:513–37. Epub 2011/08/02. pmid:21801010.
  19. 19. Stottmann RW, Tran PV, Turbe-Doan A, Beier DR. Ttc21b is required to restrict sonic hedgehog activity in the developing mouse forebrain. Dev Biol. 2009;335(1):166–78. Epub 2009/09/08. pmid:19732765.
  20. 20. Doetschman T. Influence of genetic background on genetically engineered mouse phenotypes. Methods Mol Biol. 2009;530:423–33. Epub 2009/03/07. pmid:19266333.
  21. 21. Mackay TF. Epistasis and quantitative traits: using model organisms to study gene-gene interactions. Nat Rev Genet. 2014;15(1):22–33. Epub 2013/12/04. pmid:24296533.
  22. 22. Herron BJ, Lu W, Rao C, Liu S, Peters H, Bronson RT, et al. Efficient generation and mapping of recessive developmental mutations using ENU mutagenesis. Nat Genet. 2002;30(2):185–9. Epub 2002/01/31. pmid:11818962.
  23. 23. Dixon J, Dixon MJ. Genetic background has a major effect on the penetrance and severity of craniofacial defects in mice heterozygous for the gene encoding the nucleolar protein Treacle. Dev Dyn. 2004;229(4):907–14. Epub 2004/03/26. pmid:15042714.
  24. 24. Hide T, Hatakeyama J, Kimura-Yoshida C, Tian E, Takeda N, Ushio Y, et al. Genetic modifiers of otocephalic phenotypes in Otx2 heterozygous mutant mice. Development. 2002;129(18):4347–57. Epub 2002/08/17. pmid:12183386.
  25. 25. Mukhopadhyay P, Brock G, Webb C, Pisano MM, Greene RM. Strain-specific modifier genes governing craniofacial phenotypes. Birth Defects Res A Clin Mol Teratol. 2012;94(3):162–75. Epub 2012/03/01. pmid:22371338.
  26. 26. Percival CJ, Marangoni P, Tapaltsyan V, Klein O, Hallgrimsson B. The Interaction of Genetic Background and Mutational Effects in Regulation of Mouse Craniofacial Shape. G3 (Bethesda). 2017;7(5):1439–50. Epub 2017/03/11. pmid:28280213.
  27. 27. Morgan AP, Fu CP, Kao CY, Welsh CE, Didion JP, Yadgary L, et al. The Mouse Universal Genotyping Array: From Substrains to Subspecies. G3 (Bethesda). 2015;6(2):263–79. Epub 2015/12/20. pmid:26684931.
  28. 28. Broman KW, Sen S. A guide to QTL mapping with R/qtl. Dordrecht: Springer; 2009. xv, 396 p. p.
  29. 29. Kumar P, Henikoff S, Ng PC. Predicting the effects of coding non-synonymous variants on protein function using the SIFT algorithm. Nat Protoc. 2009;4(7):1073–81. Epub 2009/06/30. pmid:19561590.
  30. 30. Greene ND, Stanier P, Copp AJ. Genetics of human neural tube defects. Hum Mol Genet. 2009;18(R2):R113–29. Epub 2009/10/08. pmid:19808787.
  31. 31. Bell SP. The origin recognition complex: from simple origins to complex functions. Genes Dev. 2002;16(6):659–72. Epub 2002/03/27. pmid:11914271.
  32. 32. Schneider E, Ryan TJ. Gamma-glutamyl hydrolase and drug resistance. Clin Chim Acta. 2006;374(1–2):25–32. Epub 2006/07/25. pmid:16859665.
  33. 33. Ma S, Kwon HJ, Johng H, Zang K, Huang Z. Radial glial neural progenitors regulate nascent brain vascular network stabilization via inhibition of Wnt signaling. PLoS Biol. 2013;11(1):e1001469. Epub 2013/01/26. pmid:23349620.
  34. 34. Wheway G, Schmidts M, Mans DA, Szymanska K, Nguyen TT, Racher H, et al. An siRNA-based functional genomics screen for the identification of regulators of ciliogenesis and ciliopathy genes. Nat Cell Biol. 2015;17(8):1074–87. Epub 2015/07/15. pmid:26167768.
  35. 35. Kawasawa Y, Kume K, Nakade S, Haga H, Izumi T, Shimizu T. Brain-specific expression of novel G-protein-coupled receptors, with homologies to Xenopus PSP24 and human GPR45. Biochem Biophys Res Commun. 2000;276(3):952–6. Epub 2000/10/12. pmid:11027574.
  36. 36. Lee DK, George SR, Cheng R, Nguyen T, Liu Y, Brown M, et al. Identification of four novel human G protein-coupled receptors expressed in the brain. Brain Res Mol Brain Res. 2001;86(1–2):13–22. Epub 2001/02/13. pmid:11165367.
  37. 37. Michelsen K, Yuan H, Schwappach B. Hide and run. Arginine-based endoplasmic-reticulum-sorting motifs in the assembly of heteromultimeric membrane proteins. EMBO Rep. 2005;6(8):717–22. Epub 2005/08/03. pmid:16065065.
  38. 38. Bockaert J, Marin P, Dumuis A, Fagni L. The ‘magic tail’ of G protein-coupled receptors: an anchorage for functional protein networks. FEBS Lett. 2003;546(1):65–72. Epub 2003/06/28. pmid:12829238.
  39. 39. Duden R. ER-to-Golgi transport: COP I and COP II function (Review). Mol Membr Biol. 2003;20(3):197–207. Epub 2003/08/02. pmid:12893528.
  40. 40. McMahon HT, Mills IG. COP and clathrin-coated vesicle budding: different pathways, common approaches. Curr Opin Cell Biol. 2004;16(4):379–91. Epub 2004/07/21. pmid:15261670.
  41. 41. Yuan H, Michelsen K, Schwappach B. 14-3-3 dimers probe the assembly status of multimeric membrane proteins. Curr Biol. 2003;13(8):638–46. Epub 2003/04/18. pmid:12699619.
  42. 42. Zerangue N, Malan MJ, Fried SR, Dazin PF, Jan YN, Jan LY, et al. Analysis of endoplasmic reticulum trafficking signals by combinatorial screening in mammalian cells. Proc Natl Acad Sci U S A. 2001;98(5):2431–6. Epub 2001/02/28. pmid:11226256.
  43. 43. Kundu K, Backofen R. Cluster based prediction of PDZ-peptide interactions. BMC Genomics. 2014;15 Suppl 1:S5. Epub 2014/02/26. pmid:24564547.
  44. 44. Kundu K, Costa F, Backofen R. A graph kernel approach for alignment-free domain-peptide interaction prediction with an application to human SH3 domains. Bioinformatics. 2013;29(13):i335–43. Epub 2013/07/03. pmid:23813002.
  45. 45. Kundu K, Mann M, Costa F, Backofen R. MoDPepInt: an interactive web server for prediction of modular domain-peptide interactions. Bioinformatics. 2014;30(18):2668–9. Epub 2014/05/30. pmid:24872426.
  46. 46. Saksela K, Permi P. SH3 domain ligand binding: What’s the consensus and where’s the specificity? FEBS Lett. 2012;586(17):2609–14. Epub 2012/06/20. pmid:22710157.
  47. 47. Klink BU, Zent E, Juneja P, Kuhlee A, Raunser S, Wittinghofer A. A recombinant BBSome core complex and how it interacts with ciliary cargo. Elife. 2017;6. Epub 2017/11/24. pmid:29168691.
  48. 48. Kroeze WK, Sassano MF, Huang XP, Lansu K, McCorvy JD, Giguere PM, et al. PRESTO-Tango as an open-source resource for interrogation of the druggable human GPCRome. Nat Struct Mol Biol. 2015;22(5):362–9. Epub 2015/04/22. pmid:25895059.
  49. 49. Drinkwater NR, Gould MN. The long path from QTL to gene. PLoS Genet. 2012;8(9):e1002975. Epub 2012/10/11. pmid:23049490.
  50. 50. Flint J, Valdar W, Shifman S, Mott R. Strategies for mapping and cloning quantitative trait genes in rodents. Nat Rev Genet. 2005;6(4):271–86. Epub 2005/04/02. pmid:15803197.
  51. 51. Morgan AP. argyle: An R Package for Analysis of Illumina Genotyping Arrays. G3 (Bethesda). 2015;6(2):281–6. Epub 2015/12/20. pmid:26684930.