Genotypic and phenotypic characterization of the Sdccag8Tn(sb-Tyr)2161B.CA1C2Ove mouse model

Nephronophthisis-related ciliopathies (NPHP-RC) are a group of disorders that present with end-stage renal failure in childhood/adolescence, kidney cysts, retinal degeneration, and cerebellar hypoplasia. One disorder that shares clinical features with NPHP-RC is Bardet-Biedl Syndrome (BBS). Serologically defined colon cancer antigen 8 (SDCCAG8; also known as NPHP10 and BBS16) is an NPHP gene that is also associated with BBS. To better understand the patho-mechanisms of NPHP and BBS caused by loss of SDCCAG8 function, we characterized an SDCCAG8 mouse model (Sdccag8Tn(sb-Tyr)2161B.CA1C2Ove) generated by Sleeping Beauty Transposon (SBT)-mediated insertion mutagenesis. Consistent with the previously reported, independent SDCCAG8 mouse models, our mutant mice display pre-axial polydactyly in their hind limbs. In addition, we report patterning defects in the secondary palate, brain abnormalities, as well as neonatal lethality associated with developmental defects in the lung in our mouse model. The neonatal lethality phenotype is genetic background dependent and rescued by introducing 129S6/SvEvTac background. Genetic modifier(s) responsible for this effect were mapped to a region between SNPs rs3714172 and rs3141832 on chromosome 11. While determining the precise genetic lesion in our mouse model, we found that SBT insertion resulted in a deletion of multiple exons from both Sdccag8 and its neighboring gene Akt3. We ascribe the patterning defects in the limb and the secondary palate as well as lung abnormalities to loss of SDCCAG8, while the developmental defects in the brain are likely due to the loss of AKT3. This mouse model may be useful to study features not observed in other SDCCAG8 models but cautions are needed in interpreting data.


Introduction
The primary cilium, in association with the centrosomes, acts as a sensory organelle in most mammalian cells. Studies during the past two decades have uncovered that mutations disrupting ciliary and centrosomal proteins cause a set of human genetic diseases termed ciliopathies [1,2]. Nephronophthisis (NPHP) is a leading cause of end-stage renal failure in children and a1111111111 a1111111111 a1111111111 a1111111111 a1111111111

PCR genotyping
Genotypes of mice were determined using primer sequences suggested by the Jackson Laboratory and purchased from Integrated DNA Technology (Coralville, IA) (S1 Table). Wild type forward and reverse primers span the SBT insertion site in the intron region between exons 12 and 13 of Sdccag8. Mutant forward and reverse primers are specific to the 3' region of the gene-trap cassette. PCR amplification of Akt3 and Sdccag8 exons was performed using standard protocols with primer sequences listed in S2 Table. Bone staining and histology Post-natal day 0 (P0) mice were euthanized by decapitation with a surgical scissors per guidelines from the Office of Animal Care and Use (OACU) of the NIH. Mice were dissected to isolate the secondary palate and the fore and hind limbs. Skin and fat was manually removed from the secondary palate. Limbs were submerged in 70˚C water for 30 seconds to facilitate removal of the epidermal layer. Specimens were kept in all following solutions slowly rocking at room temperature. Specimens were fixed in 95% ethanol for 12-48 hours. Ethanol was replaced with Alcian blue staining solution (0.03% Alcian blue (g/ml), 80% ethanol, 20% acetic acid) for 1-3 days. One to two days were long enough for limbs but the secondary palate usually required 3 days for sufficient Alcian blue staining to occur. Alcian blue staining solution was then replaced with 95% ethanol for 6 hours. Ethanol solution was replaced with 2% KOH solution for 12-24 hours. Specimens were then stained in Alizarin red solution (0.03% Alizarin Red (g/ml), 1% KOH, water) for 12-24 hours. Skeletons were cleared in 1% KOH/20% glycerol solution and imaged by Olympus Stereoscope SZX12.
Lung, brain, and kidney were collected and immersed in a solution of 4% paraformaldehyde in PBS. Tissues were fixed overnight at 4˚C then embedded in paraffin at the University of Iowa Central Microscopy Research Facility following standard protocol for embryos. Microtome sections were collected at a thickness of 5 μm. Paraffin was melted at 37˚C before sections were stained following the standard hematoxylin and eosin (H&E) staining protocol.

Quantitative reverse transcription-PCR (RT-qPCR)
Total RNAs from the brain, kidney, and lung were extracted using TRIzol Reagent (Invitrogen) following manufacturer instructions. Quantitative PCR was performed as previously described [24,25]. Briefly, 1 μg of total RNA was used for cDNA synthesis using SuperScript III reverse transcriptase (Invitrogen). Quantitative real-time PCR was performed with iQ SYBR Green Supermix (Bio-Rad) and CFX96 Real-Time PCR Detection System (Bio-Rad). Rpl19 mRNA levels were used for normalization and the ΔΔCt method [26] was used to calculate changes in gene expression. PCR products were confirmed by melt-curve analysis and sequencing. PCR primer sequences are shown in S3 Table. Genetic modifier screen Sdccag8 +/SBT FVB/NJ were crossed to Sdccag8 +/+ 129S6/SvEvTac to generate Sdccag8 +/SBT 129S6/FVB F1 mice. These mixed background heterozygous mice were then intercrossed to generate Sdccag8 SBT/SBT 129S6/FVB F2 mice. Mice were classified as "fatalities" if they expired on post-natal day 0 within 8 hours from the point at which they were first discovered. Breeding pairs were checked every morning at 8 AM and again at 4 PM. Mice were classified as "survivors" if they lived to post-natal day 21 or later. Homozygotic mutant mice that died at some point in between P0 and P21 were excluded from this study. Genotypes of mice were determined as described above. The 129SvEv/Tac strain does not have publicly available SNP sequences. Therefore, to identify SNPs around breakpoints, we used the SNP Query tool available from Jackson Lab and compared the known SNPs in FVB/NJ to two available 129 strains: 129S1/SvImJ and 129X1/SvJ. SNPs were preferentially selected when they showed a difference in both 129 strains compared to FVB/NJ. Primers were selected to flank predicted SNPs for PCR amplification (S4 Table). Primers were first tested on both parental strains to ensure viability and SNP differences. Some primer sets allowed for direct Sanger sequencing submission while others required agarose gel clean up followed by Sanger sequencing.

SBT-mediated gene-trap cassette insertion resulted in a recombination deletion of the tail end of Sdccag8 and neighboring gene Akt3
The Sdccag8 SBT mutant mouse line was generated by SBT-mediated insertion of the pT2-BART3 transposon transgene [27]. It was reported to the Jackson Laboratory that the gene-trap transgene was inserted between exons 12 and 13 of Sdccag8 (Fig 1; top), generating a truncated Sdccag8 mRNA (http://www.jax.org). PCR genotyping with primers spanning the insertion site or primers specific to the transgene show the presence of a normal Sdccag8 allele in wild-type (WT) and the insertion of the transgene in heterozygotes and homozygotic mutants (Fig 2A; S1 Fig). To determine whether full-length mRNAs encoding SDCCAG8 are produced in mutant mice, primers specific to the upstream (spanning the exon 10-11 boundary) and downstream (spanning the exon 14-15 boundary) (S3 Table) regions of the insertion site were designed, and cDNAs from the brain, kidney, and lung were analyzed by quantitative PCR (qPCR). Our qPCR results indicate that Sdccag8 transcript levels containing the 5' region are normal but undetectable at the 3' end ( Fig 2B). The lack of full-length SDCCAG8 was also confirmed by immunoblotting at the protein level (Fig 2C-2E; S2 Fig) and no truncated proteins were detected (S2 Fig). These data indicate that Sdccag8 mRNAs transcribed from the gene-trap allele do not undergo nonsense-mediated decay but no full-length protein is produced in mutant animals.
The complete absence of transcripts containing exons 14 and 15 is surprising given the levels of transcripts containing exons 5' of the SBT insertion in mutant animals. To determine whether there were additional genetic lesions in these exons, we designed primers to amplify and sequence exons 13-18 of Sdccag8 (S2 Table). However, while WT and heterozygous samples showed amplification of these 6 exons, the mutant sample showed no such amplification (Fig 3, S3 Fig), suggesting that Sdccag8 exons 13-18 are in fact absent in Sdccag8 SBT/SBT mice. To map the boundaries of this genomic deletion, we looked farther downstream of the Sdccag8 locus. We began with amplification and sequencing of the exons of the downstream neighboring gene, Akt3. We found that exons 2-13 of Akt3 (which is tail-to-tail with Sdccag8) also failed to amplify in the mutant samples, but showed amplification in WT and heterozygous samples.
The lack of Akt3 transcripts in Sdccag8 SBT/SBT mice was also confirmed by RT-qPCR assays targeting the junction between exons 3-4 of Akt3 ( Fig 3B).
These results indicate that, contrary to the original annotation, the transgene did not simply insert into Sdccag8 but resulted in a deletion of the region between exon 12 of Sdccag8 and exon 1 of Akt3 (Fig 1; bottom). The complete absence of Sdccag8 mRNAs containing the 3' region ( Fig 2B; exon [14][15], which is unusual for gene-trap alleles, is also consistent with the deletion of those exons on the chromosome.
The most plausible explanation for this large deletion is that instead of a simple insertion occurring, the gene-trap cassette likely inserted in two places, one between exons 12 and 13 of Sdccag8 and the other between exons 1 and 2 of Akt3. Following this double insertion, a recombination event occurred between these two sites that then resulted in a deletion of 200-250 kb fragment.

Sdccag8 loss causes patterning defects in the hind limb and the secondary palate
As mentioned in the Introduction, two independent mouse models of SDCCAG8 have been described (Sdccag8 Gt(OST40418)Tigm [16,17] and Sdccag8 tm1e(EUCOMM)Wtsi [18]) and considerable variations were observed between these two models. We characterized our mouse model and compared its phenotype with those previously reported, as well as the available Akt3 -/mouse ( Table 1). As for the cleft palate, we were not able to find obvious external clefting in Sdccag8 SBT/SBT mice (n = 47). To investigate whether there are mild patterning defects that are not externally visible, we performed Alcian blue and Alizarin red staining on the secondary palate of P0 mice. Isolation of the secondary palate allowed us to look at the bone and cartilage structure of the premaxilla region, usually associated with visible clefting of the palate. Staining of the secondary palate showed abnormal patterning in mutant mice most noticeable in the basisphenoid, presphenoid, and the premaxilla (Fig 4A; n = 4). However, this abnormal patterning was restricted to minor changes in the pattern of the two sides of the palate coming together. It did not include a failure of the two sides to meeting, and thus did not result in a cleft.
Pre-axial polydactyly was reported in both Sdccag8 Gt(OST40418)Tigm and Sdccag8 tm1e(EUCOMM)Wtsi models with variable penetrance [16][17][18]. Consistent with the prior studies, polydactyly was observed only in the hind limbs but not in the forelimbs (Fig 4B and 4C). The penetrance of pre-axial polydactyly in our mouse model was 100% (n = 47) and in all cases polydactyly was bilateral. Therefore, our model was closer to the Sdccag8 tm1e(EUCOMM)Wtsi mouse (94% display polydactyly, limbs affected and laterality not reported) with respect to the penetrance [18].
The increased penetrance seen in Sdccag8 SBT/SBT mice may be due to the additional Akt3 gene deletion. However, to our knowledge, Akt3 does not have a role in limb bud development and Akt3 knockout mice do not exhibit polydactyly [28,29]. In addition, the penetrance of polydactyly in the Sdccag8 tm1e(EUCOMM)Wtsi mouse model, in which the Akt3 locus is intact, is nearly complete. This suggests that Akt3 is not likely a contributing factor to the increased penetrance seen in Sdccag8 SBT/SBT mice. A more plausible explanation is that the deletion of 6 coding exons in the Sdccag8 SBT allele makes it a null allele, while Sdccag8 tm1e(EUCOMM)Wtsi and    Sdccag8 Gt(OST40418)Tigm are strong hypomorphic alleles that allow a low level expression of normal Sdccag8 gene.

Loss of Sdccag8 does not result in the early development of kidney cysts
It was reported to the Jackson Laboratory (http://www.jax.org) that Sdccag8 SBT/SBT mice have cystic kidneys at P0. While kidney phenotypes were not reported in the Sdccag8 tm1e(EUCOMM)Wtsi model, which also expires at P0, Airik et al. reported cyst development in the kidney, first noticeable at P100, in Sdccag8 Gt(OST40418)Tigm mice [16]. In our Sdccag8 SBT colony, we were not able to find cystic kidneys at P0 (Fig 4D). Mutant kidneys were comparable to their wild type and heterozygous littermates. Due to the neonatal lethality, we were not able to examine kidney abnormalities in older animals.

Sdccag8 SBT/SBT mice display brain abnormalities consistent with Akt3 loss
Brain abnormalities were reported in both SDCCAG8 deficient (Sdccag8 tm1e(EUCOMM)Wtsi ) and Akt3 knockout mice [18,28,29]. The Sdccag8 tm1e(EUCOMM)Wtsi mouse demonstrates abnormal neuronal migration in the developing cortex. However, no obvious microcephaly is seen in the brains of mutant animals, suggesting that cortical neurogenesis is near normal in these animals [18]. While little is known about the role of Akt3, it has been shown that Akt3 knockout mice show a 20-25% reduction in the brain size in relation to whole body size due to a decrease in cell size and number [28,29]. They also show a thinning of the white matter fiber connections in the corpus callosum, as well as a loss of distinction between the corpus callosum and the surrounding grey matter [28]. We found that the corpus callosum and anterior commissures were thinner and there was a deficiency of white matter in Sdccag8 SBT/SBT mice (Fig 4E). These neurological phenotypes are very similar to those seen in Akt3 -/mice. Therefore, we conclude that the brain abnormalities observed in Sdccag8 SBT/SBT mice are at least partly due to the loss of AKT3 function. The contribution of SDCCAG8 loss to these phenotypes is currently unclear.

Loss of SDCCAG8 function results in neonatal death in the FVB background, associated with abnormal lung development
One notable difference between the Sdccag8 Gt(OST40418)Tigm and the Sdccag8 tm1e(EUCOMM)Wtsi model is the neonatal lethality phenotype of mutant pups. Sdccag8 Gt(OST40418)Tigm mice are born with a Mendelian ratio and survive to the weaning age [16]. In contrast, although the precise time of death was not reported, homozygous Sdccag8 tm1e(EUCOMM)Wtsi mutant mice die shortly after birth [18]. Sdccag8 SBT/SBT mutant mice are also born at Mendelian ratios (~25%). However, all of them (n> 47) die within 12 hours following birth, indicating that loss of Sdccag8 is neonatal lethal in the Sdccag8 SBT/SBT mouse model.
Most Sdccag8 SBT/SBT pups die within 8 hours after birth and all within 12 hours. Failure of suckling is often considered as a cause of neonatal lethality when cleft palate is present. However, despite the presence of some defects in the secondary palate, Sdccag8 SBT/SBT mice do not display obvious clefting externally. Instead, we found that Sdccag8 SBT/SBT mice are cyanotic and gasping, indicative of oxygen deprivation (Fig 5A). Airik et al. reported the expression of Sdccag8 in the prospective ciliated cells of the developing bronchi and bronchioles [16]. Prompted by these findings, we examined the histology of the lung at P0. H&E staining of the lung sections revealed that, compared to littermate controls, the interstitium of Sdccag8 SBT mutant lungs remain thick. Reduction of alveolar airspaces and accumulation of blood in the alveolar sacs was also observed in Sdccag8 SBT/SBT mice (Fig 5B). These findings suggest that the progression from the canalicular stage to the terminal saccular stage during lung development [30] is halted in Sdccag8 SBT/SBT mice. Therefore, although failure of suckling could contribute, our data indicate that developmental defects in the lung and consequent failure of normal gas exchange and oxygen deprivation are likely the main cause of neonatal lethality in Sdccag8 SBT/SBT mice.
Interestingly, human SDCCAG8 patients diagnosed with BBS have been reported to have recurrent pulmonary infections and poor respiratory function [4,13,14]. Although Akt3 is highly expressed in the lung, absence of Akt3 is not lethal and no lung phenotypes have been observed in Akt3 -/animals [29]. This suggests that Akt3 does not have an essential role in lung development. Based on the known expression pattern of Sdccag8 in the lung [16], neonatal lethality observed in another SDCCAG8 mouse model (Sdccag8 tm1e(EUCOMM)Wtsi ) [18], and the lack of a pulmonary phenotype in Akt3 -/mice [29], we conclude that the developmental defects in the Sdccag8 SBT/SBT mutant lung are due to a loss of Sdccag8. Characterizing the Sdccag8 Tn(sb-Tyr)2161B.CA1C2Ove mouse

Genetic contribution from 129S6/SvEvTac provides a protective effect against neonatal lethality
We have generated multiple BBS mouse models, including BBS1, BBS2, BBS4, and BBS6 [31][32][33][34]. We have noticed genetic background-dependent neonatal lethality in all of these mouse models; most BBS mutant mice in a 129 background survive normally to the weaning age and beyond, but those that reach weaning age are extremely rare in FVB or C57 backgrounds [35]. The Sdccag8 tm1e(EUCOMM)Wtsi model, which shows neonatal lethality, is on the C57 background [18], and the Sdccag8 Gt(OST40418)Tigm model, which does not show early lethality, is on a C57/ 129 mixed background [16]. Our Sdccag8 SBT mice are on the FVB background and show neonatal lethality. Based on the correlation between the genetic background and the early lethality phenotype in these mouse models, we hypothesized that there is a genetic modifier(s) that influences the survival of Sdccag8 mutant pups.
To test this hypothesis and map the modifier(s), we mated Sdccag8 +/SBT FVB/NJ mice with wild-type 129S6/SvEvTac mice and began an intercross of Sdccag8 +/SBT animals (F1) in a FVB/ 129S6 mixed background. The resulting F2 generation mice were used for phenotyping as well as genetic modifier mapping. While polydactyly was retained in the F2 Sdccag8 SBT/SBT mice at 100% penetrance, we found that adding the 129S6/SvEvTac background resulted in 78% (42 of 54) survival to P21 or later, with only 22% (12 of 54) of F2 pups expiring at P0 similar to their pure strain FVB/NJ counterparts, suggesting the presence of a 129 allele that allows for survival.
To map the modifier, DNA samples from a pilot set of F2 mice, 10 of which died at P0 (fatalities) and 10 of which survived to P21 or later (survivors), were sent to the Jackson lab and genotyped on a custom 150 SNP panel specific to the FVB/NJ and 129S6/SvEvTac lines. Using a Chi Square test of independence, we identified two adjacent SNPs on chromosome 11 which showed a significant genetic association (S5 Table). These SNPs were rs3023266 (p = 0.018) and rs3714172 (p = 0.03). Genotypes at these SNPs were confirmed using Sanger sequencing.
Further analysis was performed on additional SNPs throughout the region surrounding rs3023266 and rs3714172, using the original cohort and an additional 32 mice that survived. Statistical analysis of these SNPs was performed using Chi Square analysis with expected ratios of 1:2:1 FVB:Het:129 versus observed genotypes in the surviving mice. Since introduction of the 129S6 allele promoted survival, we strove to identify an interval lacking FVB/NJ homozygosity in mice that survived. This analysis identified a region of 8 Mb from rs3714172 to rs3141832 that showed a significant genetic skew towards decreased FVB homozygosity ( Table 2). There are 162 genes in this 8 Mb region (S6 Table).
We defined our region based upon the hypothesis that 129 provides a protective effect against expiration in survivors and thus we focus on minimizing the number of FVB homozygotes while deviating from expected distribution, rather than simply deviating from expected 1:2:1 distribution. While the rs27100337 SNP has the lowest p-value in this interval, it also has a greatly increased (33% increased from 6 to 8) number of FVB homozygotes.
This region contains an interesting locus, a missense mutation in Matrix Metalloproteinase-28 (Mmp28) at rs6169904. The B6 reference strain and the FVB strain have a shared allele, whereas the 129S6 allele has a p.V342A (NP_536701.1) variant. Although this variation may not be a major change, it is within the 7th exon of Mmp28, which is subject to alternative splicing. Furthermore, this variant falls within a haemopexin domain found in isoforms 1 and 3 but not isoform 2 of Mmp28. The haemopexin domain can play a role in substrate recognition and it is thought that the alternative splicing of Mmp28 in this region may alter substrate binding capacity or preference [36]. However, the catalytic activity and substrate specificity of either alternative spice variant of Mmp28 towards a biologically relevant substrate has not been studied. Under normal conditions, Mmp28 is expressed at high levels in the lung and promotes epithelial cell survival [37]. It also regulates macrophage polarization and limits macrophage recruitment to tissues [38][39][40].
In addition to Mmp28, all 10 murine genes of the Schlafen family are found within the identified region of interest. One SNP studied, rs29432792, is found in the intronic region of Slfn5. While little is known about the expression patterns and functions of the individual members, the Schlafen (SLFN) family has been implicated in the control of cell proliferation, induction of immune responses, and the regulation of viral replication [41], along with an implication in DDK syndrome, a syndrome which causes implantation defects and early embroynic lethality when females of the DDK inbred mouse strain are mated to many non-DDK strain males [42]. More recently, it has been shown that Schlafen (SLFN) proteins are regulated by interferons and evidence suggests that SLFN proteins play an important role in the anti-neoplastic and growth inhibitory effects of interferons. SLFN1 and SLFN2 have shown upregulation in cystic fibrosis lungs [43] and SLFN5 has an anti-neoplastic effect in renal cell carcinoma [44]. While large changes to any of these proteins is not expected, the genetics of the Schlafen family on the FVB/NJ background may make them susceptible to a more extreme phenotype given the right conditions. The loss of SDCCAG8 may affect this family's ability to complete their normal duties, resulting in death on the FVB/NJ strain, with this extreme phenotype mitigated by having a 129S6/SvEvTac genome across this region.
While having a single copy of 129S6/SvEvTac in this interval greatly increased the odds of survival in Sdccag8 mice, it should be noted that some mice homozygous for FVB/NJ in this region did survive to P21 or later (Table 2). Therefore, although this region is associated with survival in the presence of the 129S6/SvEvTac background, there may be additional regions that also provide protection from early lethality.

Conclusions
We show that the Sdccag8 Tn(sb-Tyr)2161B.CA1C2Ove mouse displays early neonatal lethality due to abnormal lung development and patterning defects in the secondary palate and the hind limb. We also determined the precise genetic lesion present in this mouse model. Contrary to the original annotation that the pT2-BART3 gene-trap cassette is inserted between the exons 12 and 13 of Sdccag8, our study identified a large deletion that encompasses exons 13-18 of Sdccag8 and exons 2-13 of Akt3. We conclude that developmental defects of the secondary palate, hind limbs, and the lung are due to a loss of Sdccag8, whereas the brain phenotype is likely due to loss of Akt3. We suggest this mouse model may be useful to study the roles of SDCCAG8 in vivo but cautions are needed due to the confounding Akt3 deletion in this model. Finally, we identified a region of interest on chromosome 11 that provides a protective effect against neonatal lethality in Sdccag8 SBT/SBT mice when genomic contribution from the 129S6/SvEvTac background is present.