A structural UGDH variant associated with standard Munchkin cats

Background Munchkin cats were founded on a naturally occurring mutation segregating into long-legged and short-legged types. Short-legged cats showed disproportionate dwarfism (chondrodysplasia) in which all four legs are short and are referred as standard Munchkin cats. Long-legged animals are referred as non-standard Munchkin cats. A previous study using genome-wide single nucleotide polymorphisms (SNPs) for genome-wide association analysis identified a significantly associated region at 168–184 Mb on feline chromosome (FCA) B1. Results In this study, we validated the critical region on FCA B1 using a case-control study with 89 cats and 14 FCA B1-SNPs. A structural variant within UGDH (NC_018726.2:g.173294289_173297592delins108, Felis catus 8.0, equivalent to NC_018726.3:g.174882895_174886198delins108, Felis catus 9.0) on FCA B1 was perfectly associated with the phenotype of short-legged standard Munchkin cats. Conclusion This UGDH structural variant very likely causes the chondrodysplastic (standard) phenotype in Munchkin cats. The lack of homozygous mutant phenotypes and reduced litter sizes in standard Munchkin cats suggest an autosomal recessive lethal trait in the homozygote state. We propose an autosomal dominant mode of inheritance for the chondrodysplastic condition in Munchkin cats.


Background
Short-legged cats have been reported as early as 1944, describing individuals with unusually short forelimbs but long-legged hind limbs, resulting in forelimb micromelia [1][2][3]. Today's Munchkin cats presumably have their roots in a female cat with all four short legs on which Sandra Hockenedel [4] used to found the breed in 1983. This short-legged animal gave birth to a healthy litter, with half of the kittens showing short legs. This shortlegged female cat and one of its short-legged male kittens established today's Munchkin cat breed. To expand the Munchkin cats´gene pool, these two Munchkin cats were mated to long-legged cats. Through this outcross program, a variety of coat colors and patterns as well as different Munchkin cat subtypes were developed [4][5][6][7][8][9][10] (Additional file 1). Breeders refer to short-legged as standard Munchkin and long-legged as non-standard Munchkin cats [4]. Offspring from standard Munchkin cat parents can segregate into standard and nonstandard Munchkin cats. Munchkin cats have been recognized by The International Cat Association (TICA) as an independent breed since 1994. Breeding data collected by TICA suggest a dominant mode of inheritance [11,12] and lethality in the homozygous state at an early embryonic stage [5].
The Munchkin cat phenotype and genotype is not yet fully characterized. A genome-wide association study (GWAS) was performed for the standard Munchkin phenotype resulting in a highly significant critical region at 168-184 Mb on feline chromosome (FCA) B1 [13].
The present study characterized the standard Munchkin phenotype using computed tomography (CT) scans and comparative long bone measurements. Pedigree data of Munchkin cats were collated to establish the mode of inheritance. In order to unravel the responsible mutation for standard Munchkin cats, we analyzed whole genome sequencing (WGS) data of a Munchkin cat family and controls for Munchkin-associated variants. Candidate variants were validated in large cohorts of standard and non-standard Munchkin cats as well as in a cat breed panel with the most popular breeds. We identified a structural UDP-glucose 6-dehydrogenase (UGDH) variant perfectly segregating with the standard Munchkin cat phenotype. For this variant, we assumed dominance as homozygous mutants were not observed.

Munchkin cat phenotype
The standard Munchkin cats of the subtype Genetta showed a disproportionate short stature with shortened fore-and hind limbs compared to a female non-standard Munchkin cat (Fig. 1). Subtype Genetta resulted from the outcrossing program of standard Munchkin with Bengal cats and this subtype is recognized by TICA as an experimental breed. The upper-and forearms of the front limbs as well as the upper-and forelegs had nearly the same length in the standard Munchkin cats. The size of paws was proportional to the body. The head of both types, standard and non-standard Munchkin cats showed no signs of chondrodysplastic changes such as brachygnathia superior or a spherical appearance.
CT of the forelimbs of the 4-year old standard Munchkin tomcat showed a shortening of all distal und proximal long bones including humerus, radius, ulna and metacarpalia ( Fig. 2 and Additional file 2). Lengths of the humerus, radius, ulna, metacarpalia, femur and tibia of the standard Munchkin cat were reduced by 71, 58, 64, 84, 74, and 70%, respectively, in comparison to the domestic cat. Metatarsalia had normal length. All bones of the front and hind limbs exhibited higher average diaphyseal diameters, in particular humerus (+ 14%) and femur (+ 29%). Furthermore, the humerus showed a slight internal rotation along its longitudinal axis, resulting in a moderate incongruity in the elbow joint with axial deviation. The humerus compacta in the middle segment revealed a moderate degree of thickening of 2.4 mm (0.09 in). Furthermore, the radius was too short in relation to the ulna, as well as the ulna was slightly medially rotated, whereas the radius was bent to a high degree in the longitudinal axis of about 43°. This resulted in an incongruity in the ulnocarpal and radiocarpal joint. The hind limbs did not show any rotation and structural effects.

Pedigree analysis
Pedigree analysis of the Munchkin cat family of the subtype Genetta demonstrated that three repeated matings between a standard Munchkin tomcat and a standard Munchkin queen resulted in seven standard Munchkin and two non-standard Munchkin cats (Additional file 3). As reported by breeders, matings among standard Munchkin cats resulted in reduced litter sizes.

Whole genome sequencing and variant detection
A mean read coverage of 11.99X (non-standard Munchkin kitten), 12.67X (standard Munchkin kitten), 12.90X (standard Munchkin sire) and 8.49X (standard Munchkin dam) was obtained in WGS data. A filter analysis for exclusively heterozygous genotypes in the standard Munchkin cats and homozygous wild genotypes in 16 control cats did not reveal functional variants on FCA B1, B4 and X.
Thus, an obvious protein modifying candidate variant was not identified in the genome-wide associated region on FCA B1 at 168 to 184 Mb or within the genomic regions with the next highest peaks on FCA B4 and X from the previous GWAS [13].

Mutation analysis
We did not detect variants in wild type cats for the cDNA sequence of exon 10 and 11 (ENSFCAT00000009602.6) or exon 9 and 10 (ENSFCAT00000055794.2) and the proximal parts of the 3'UTR using primer pairs within exon 10 and 3'UTR (PCR-type 3 and PCR-type 1, Additional file 7). Amplicons of cDNA from wild type cats were of an expected size of 2514 bp (PCR-type 1) or 185 bp (PCR-type 3). Sanger sequencing of wild type allele cDNA for a region including exon 10 (ENSFCAT00000009602.6) or exon 9 (ENSFCAT00000055794.2) and 3'UTR (PCR-type 4-7) confirmed the 3'UTR region reference genome sequence (Ensembl database).
PCR-type 2 was designed to show whether the 108 bp insert (see above) is transcribed into mRNA and in this case, should give an amplicon size of 110 bp only for the mutant allele in standard Munchkin cats. In standard Munchkin cats heterozygous for the NC_018726.3: g.174882895_174886198delins108 (Felis catus 9.0) mutation, PCR-type 1 should yield two amplicons with sizes of 402 bp (insertion not transcribed) or 510 bp (insertion transcribed) and 2514 bp for the mutant and wild type allele, respectively. The shorter fragment size of 402 bp was calculated based on genomic DNA analysis assuming deletion of exon 11 (ENSFCAT00000009602.6) or exon 10 (ENSFCAT00000055794.2) and 2001 bp of the proximal part of the 3'UTR. However, PCR-type 1 yielded an amplicon size of 493 bp for the mutant allele in standard Munchkin cats. This amplicon contained 73 bp of exon 10 (ENSFCAT00000009602.6) or exon 9 (ENSFCAT00000055794.2) and 309 bp from intron 10-11 (ENSFCAT00000009602.6) or intron 9-10 (ENSFCAT00000055794.2) and in addition, 111 bp of the 3'UTR ( Fig. 3 and Additional file 8). We found an alternative splice site in the intronic sequence with the motif TACACAatttaaaa explaining the partial intron retention of 309 bp. PCR-type 2 failed to generate an amplicon and thus, suggests that the 108 bp insert is not transcribed into mRNA.

Validation
The UGDH structural variant comprising a deletion and an insertion on FCA B1 was found in all 42 standard Munchkin cats whereas the 15 non-standard Munchkin cats and all 203 cats of other breeds were homozygous wild type. Perfect segregation of the UGDH structural variant with the standard and non-standard Munchkin cats was shown for the Munchkin cat family of the subtype Genetta (Additional file 3).

Discussion
This is the first study identifying a mutation in UGDH on FCA B1 perfectly associated with standard Munchkin cats exhibiting a disproportionate stature with shortened legs. In our study, we confirmed the significantly associated genome region on FCA B1 at 168-184 Mb [13] in a case-control and haplotype analysis and identified a potential causative structural variant in UGDH. UGDH is involved in heparan sulfate proteoglycan (HSPGs) synthesis through catalyzing conversion of UDP-glucose into UDP-glucuronic acid (UDP-GlcA) [14]. Proteoglycans regulate growth factors and morphogens such as fibroblast growth factor (Fgf), hedgehog (Hh), transforming growth factor-ß (TGFß) and wingless (Wg, Wnt family) [14][15][16][17]. Studies on growth related signaling molecules showed Wnt and Fgf proteins necessary for limb structure, ensuring proliferation, differentiation and cell survival [18,19]. Various mutations in Wnt proteins (Wnt-3a and Wnt-7a) were found to result in abnormalities during limb formation including irregular limb size [17,20]. Furthermore, a disruption of fgf10 in mutant zebrafish (daedalus) was shown to provoke an absence of fgf10 activity and thus, a severely compromised fin bud development [19]. Whether the standard Munchkin cat fits into a model with disturbed proteoglycan signaling and aberrant signals for morphogens requires further studies. Furthermore, the finding that no homozygous mutant individual was present in the study population supported the previous suggestion of lethality of homozygous mutant individuals at an early embryonic stage [5].
Our investigations of heterozygous mutant standard Munchkin cats compared to homozygous wild type cats showed a partial retention of the last intron in UGDH resulting in a loss of an acceptor site and a new alternative donor site. cDNA sequencing results revealed a partial intron retention of 309 bp as well as a predicted UGDH protein with a truncation of 12 aa and a modification of the aa sequence. Previous studies showed that intron-retaining transcripts are prone to produce novel protein isoforms with modified biological functions resulting in detrimental effects on cells and tissues [21][22][23]. This was also observed in Miniature Zebu, in which an aggrecan (ACAN)-associated mutation caused a recessive lethal chondrodysplasia due to a truncated and modified protein [24]. In Dexter cattle, a functionally null variant of ACAN was found in a homozygous state in spontaneously aborted fetuses and non-viable calves with a bulldog-like phenotype [25]. In our study, we propose the predicted truncation and modification of UGDH as the reason for an early embryonic death in a homozygous mutant state. Studies for UGDH in mice and drosophila (sugarless or kiwi) mutant embryos deficient for UGDH activity showed also embryonic lethality Fig. 3 Schematic representation of cDNA in wild type (wt) and standard Munchkin cats (sMc) for PCR-type 1. PCR across the deletion from (a) exon 10 (ENSFCAT00000009602.6) or (b) exon 9 (ENSFCAT00000055794.2) to the 3′-UTR (non-deleted region) showed a 493 bp amplicon. Sanger sequencing of this 493 bp amplicon showed that the 108 bp insert is not transcribed but this PCR-product consists of 73 bp of the exon 10 (ENSFCAT00000009602.6) or exon 9 (ENSFCAT00000055794.2), plus a partially retained intron 10-11 (ENSFCAT00000009602.6) or intron 9-10 (ENSFCAT00000055794.2) of 309 bp and 115 bp of the 3'UTR [26][27][28][29][30]. The zebrafish ugdh mutant jekyll was found to be lethal due to severe defects during organogenesis of the heart [28]. Heterozygous standard Munchkin exhibit reduced leg length without impaired health, as found in Dexter cattle [25].

Conclusions
In conclusion, we identified a structural UGDH variant perfectly associated with the standard Munchkin phenotype. No homozygous mutant cats have been observed, supporting the suggestion that the mutation has an autosomal dominant mode of inheritance for the chondrodysplastic condition in Munchkin cats. A diagnostic test is now available for cat breeders to prove if their animals carry the standard Munchkin cat mutation in order to get recognition by TICA.

Clinical examination
For the clinical investigation, members of the segregating Munchkin cat family of the subtype Genetta were available. A cat breeder provided a 4-year old standard Munchkin tomcat and three of his progeny including two male standard and a female non-standard Munchkin kitten for a detailed examination. The 4-year old standard Munchkin tomcat underwent a CT scan in sternal recumbency using a multislice helical CT scanner (Brilliance 64-CT, Philips Medical Systems, Best, The Netherlands). A slice thickness for the head of 0.67 mm (0.026 in) and settings of 120 kV/ 250 mAs and for the body a slice thickness of 0.67 mm (0.026 in) and settings of 140 kV/ 250 mAs were set for analysis. A 1024 × 1024 matrix was used.
In addition, absolute total length, diaphyseal length and the diaphyseal diameter of humerus, radius, ulna, femur, tibia, metacarpalia and metatarsalia bones were measured according to Jones and Jolly [31]. The relative length of individual bones was determined through the ratio of length of the respective individual bone to the length measured from the proximal end of the humerus to the distal end of metarcarpus III.

Haplotype analysis
To refine the critical genome-wide associated region on FCA B1, Kompetitive Allele Specific PCR (KASP) assay was used (LGC Genomics, Middlesex, UK). We selected 14 SNPs (PRJEB30080) of which four SNPs were located distally to the critical region and ten SNPs within the genome-wide associated region at a distance among each other of about 430 to 7600 kb on FCA B1. These 14 SNPs were genotyped in 9 standard Munchkin cats, one non-standard Munchkin cat and 79 controls. These animals were unrelated. KASP genotyping was performed according to standard protocols (LGC) using an ABI7300 real-time system for 96 well plates (Additional file 13). A case-control analysis was carried out using SAS/Genetics, version 9.4 (Statistical Analysis System, Cary, NC, USA). Haplotype analysis was done with the procedure HAPLOTYPE in SAS.

Whole genome sequencing
Animals for WGS were from the Munchkin cat family of the subtype Genetta including the standard Munchkin parents (sire and dam) and two members of their litter with a standard Munchkin cat male and a non-standard Munchkin cat female (Additional file 3). Libraries were prepared using the NEBNext Ultra II DNA Library Prep Kit for Illumina (New England BioLabs, Ipswich, MA, USA) and run on an Illumina NextSeq500 in a 2 × 150 bp paired-end mode. Quality control was performed using fastqc 0.11.5 [32] and reads were trimmed using PRINSEQ (V 0.20.4) [33].
A read depth of 8-999 and quality score values > 20 were applied for variant detection. Those variants, which were heterozygous in the three standard Munchkin cats, homozygous wild type in the non-standard Munchkin cat and homozygous wild type in all controls and concordant to an autosomal monogenic dominant inheritance, were filtered using SAS (version 9.4, Statistical Analysis System, Cary, NC, USA) for further investigation. Only variants with high or moderate effects according to SNPEff version 4.3 t (2017-11-24, SNPEff database Felis catus 8.0) [38] were considered for further analysis. A candidate gene list from previous studies about chondrodysplasia was used to find candidate genes in the filtered variant list [24].

Structural variant detection
For structural variant detection, LUMPY software, version 0.2.13 [39], integrating multiple structural variation signals jointly across multiple samples, was run for Bam files of all four Munchkin cats and 16 controls. The generated VCF file was filtered for structural variants on FCA B1, heterozygous in the standard Munchkin cats and homozygous wild type in the non-standard Munchkin cat and the 16 controls.

Mutation analysis
To test whether the 108 bp insert is transcribed or not transcribed into mRNA, we isolated RNA from hair roots stabilized in RNAlater reagent (Qiagen, Hilden, Germany) of two standard Munchkin cats of the subtype Genetta and two domestic shorthair cats. All samples were transcribed into cDNA according to standard protocols (Maxima First Strand cDNA Synthesis Kit, Thermo Fisher). Primer pairs were designed using Primer3 tool (version 0.4.0, http://bioinfo.ut.ee/primer3-0.4.0/). One primer pair was designed to produce a PCR-product in mutant and wild type allele (PCR-type 1), one primer pair to produce a PCR-product only if the mutant allele is present and the insertion is transcribed (PCR-type 2), and one control primer pair to produce a PCR-product in the presence of wild type allele (PCR-type 3, Additional file 14). The first primer pair (PCR-type 1), located across the deletion from exon 10 (ENSFCAT00000009602.6) or exon 9 (ENSFCAT00000055794.2) to the 3′-UTR within the nondeleted region, yields an expected amplicon size of 510 bp for the mutant allele if the insertion is transcribed and 402 bp for the mutant allele if the insertion is not transcribed, as well as 2514 bp for the wild type allele (Additional file 7). The second primer pair (PCR-type 2) from the proximal region of exon 10 (ENSFCAT00000009602.6) or exon 9 (ENSFCAT00000055794.2), to a region within the 108 bp insert produces an expected amplicon of 110 bp only in the mutant allele when the insertion is transcribed. This PCRproduct is not present in the wild type or mutant allele when the insertion is not transcribed. The third primer pair (PCR-type 3) was designed to bridge exon 10 and 11 (ENSFCAT00000009602.6) or exon 9 and 10 (ENSFCAT00000055794.2) with an expected amplicon size of 185 bp for the wild type allele in standard Munchkin cats and control cats with wild type alleles on both chromosomes. Furthermore, we designed five primer pairs (PCRtype 4-7) for cDNA sequencing the region spanning exon 10 (ENSFCAT00000009602.6) or exon 9 (ENSFCAT00000055794.2) and those parts of the 3'UTR present in the mutant and wild type allele (Additional file 15). PCR included 0.25 μM of the respective primers, 20 ng cDNA, 12.5 μl UCP HiFidelity MasterMix (Qiagen, Hilden, Germany) and was prepared as recommended by Qiagen, Hilden, Germany. Sanger sequencing was done for the standard Munchkin dam from the Munchkin cat family of the subtype Genetta (Additional file 3) for PCR-type 1 and in one domestic shorthair cat for PCR-type 4-7.

Validation
A structural variant composed of a 3303 bp deletion (ss5015497294) and a 108 bp insertion (ss5015497295) within the critical region on FCA B1 was Sanger sequenced and visualized using a duplex PCR in individuals including standard Munchkin, non-standard Munchkin and control cats. Duplex PCR was performed in 260 individuals including 15 non-standard Munchkin, 42 standard Munchkin and 203 controls from breeds other than Munchkin cats. Sanger sequencing was done in the standard Munchkin sire, dam and the standard Munchkin kitten. A forward primer proximally of the detected structural variant and two different reverse primers located within the deleted region and distal of this deletion were designed using Primer3 tool (version 0.4.0, http:// bioinfo.ut.ee/primer3-0.4.0/) (Additional file 16 and Additional file 17). Duplex PCRs were performed in 22-μl reaction volumes containing 2 μl DNA, 1.5 mM deoxyribonucleoside triphosphates, 5 pmol of primers MK_del_R and MK_wt_R, 10 pmol of primer MK_wt_F, 1.5 U of Taq polymerase in the reaction buffer supplied by the manufacturer (MP Biomedicals, Eschwege, Germany) and 4.2 μl enhancer reagent (MP Biomedicals). After a 5 min initial denaturation at 95°C, 40 cycles of 30 s at 94°C, 30 s at 59°C, and 45 s at 72°C were run on a Thermocycler TProfessional 96 (Biometra, Göttingen, Germany). All samples were visually evaluated on a 1% agarose gel using Gel iX20 Imager (Intas Science Imaging Instruments, Göttingen, Germany).
Break point validation was performed in the three standard Munchkin cats of the subtype Genetta with WGS data using the primers MK_wt_F and MK_del_R accoding to the duplex PCR protocol. Evaluation of Sanger sequences was performed using Sequencher 4.8 software (GeneCodes, Ann Arbor, MI, USA).