The dorsal hair ridge in Rhodesian and Thai Ridgeback dogs is caused by a dominant mutation that also predisposes to the congenital developmental disorder dermoid sinus. Here we show that the causative mutation is a 133-kb duplication involving three fibroblast growth factor (FGF) genes. FGFs play a crucial role in development, suggesting that the ridge and dermoid sinus are caused by dysregulation of one or more of the three FGF genes during development.

Dogs with a characteristic dorsal hair ridge seem to have been present in both Africa and Asia long before European colonization (Fig. 1). The Rhodesian Ridgeback dog (Fig. 1a), first registered in South Africa in 1924, is most likely a blend of European dogs (brought to Africa by early colonizers) and an extinct indigenous breed of Africa, the ridged Hottentot Khoi dog¹. The Thai Ridgeback (Fig. 1b) and the Vietnamese Phu Quoc dog are two Asian breeds with a dorsal hair ridge closely resembling the one found in Rhodesian Ridgeback dogs.

(a,b) The dorsal hair ridge of Rhodesian Ridgeback (a) and Thai Ridgeback dogs (b). (c) A ridgeless Rhodesian Ridgeback dog. (d) Light micrograph of ridge skin taken from the dorsal median plane with cross-sectioned, laterally oriented hair follicles (arrow) and sebaceous glands in the dermis (objective lens × 4). (e) Corresponding cross-section from a ridgeless dog. Note that the hair follicles are caudally oriented in the median plane (arrow) (objective lens × 4). (f) The full extension of a dermoid sinus, extending from the upper dermis to the nuchal ligament, which overlays the cervical spinous process. The dermoid sinus–affected Rhodesian Ridgeback was 6 weeks old at the time of euthanasia. (g) Light micrograph of a cross-sectioned dermoid sinus with hair and keratin debris in the lumen covered by a stratified squamous keratinized epithelium. In the upper portion, appendages such as a hair follicle (filled arrow) and sebaceous glands (^*) are seen; in the right corner, a hair follicle with multiple hair shafts is indicated (open arrow); (objective lens × 20). At lower magnification, the whole dermoid sinus is seen with the surrounding loose connective tissue (inset, objective lens × 10).

Full size image

Histology of the skin from a ridged dog, taken strictly from the dorsal median plane, showed cross-sectioned appendages (that is, hair follicles and sebaceous glands) of normal appearance but lateral orientation (Fig. 1d). In contrast, skin from the median plane of a ridgeless dog showed caudally oriented hair follicles (Fig. 1e). Ridgeback dogs are affected by the congenital malformation dermoid sinus (Fig. 1f), which closely resembles a neural-tube defect in humans that is usually termed dermal sinus². In dogs, anatomical locations of dermoid sinus, anterior and posterior to the ridge, correspond with those found in humans. The dermoid sinus was characterized by a tubular indentation of the skin, with keratin and hair in the lumen; appendages such as sebaceous glands and aberrant hair follicles with multiple hair shafts were also observed (Fig. 1g). Dermoid sinus is closely associated with the ridge phenotype, and no ridgeless dogs with dermoid sinus have been reported³. Approximately 5–6% of Rhodesian Ridgebacks born in Sweden are ridgeless, and around 8–10% of ridged offspring have dermoid sinus^3,4. Both ridgeless dogs and dogs with dermoid sinus are excluded from breeding.

We assumed a genetic model in which (i) ridgeless dogs are homozygous (r/r) for the wild-type allele, (ii) ridged dogs without dermoid sinus are heterozygous or homozygous for the Ridge allele (R/r or R/R) and (iii) ridged dogs with dermoid sinus are homozygous R/R. We were able to assign the Ridge locus to a 750-kb region on chromosome 18 using only nine ridgeless and 12 ridged dogs, of which 11 had dermoid sinus, in a genome-wide association analysis⁵. The results indicated that 10 of 11 dogs with dermoid sinus were homozygous for a haplotype not present among ridgeless dogs. Notably, all Rhodesian ridgebacks with dermoid sinus were heterozygous for a SNP (SNP_51,399,353) within the 750-kb region associated with the phenotype. Further analysis showed that 43 of 45 ridged Rhodesian Ridgebacks were heterozygous for this SNP. This significant deviation (P < 0.001) from Hardy-Weinberg proportions suggested that the SNP is part of a duplicated region, and that the two copies tend to carry different SNP variants.

We used the recently developed multiple ligation–dependent genome amplification (MLGA) technique⁶ to test whether the Ridge allele is associated with a duplication (Supplementary Methods online). We designed MLGA probes with 100-kb spacing over a 2-Mb region (Supplementary Table 1 online). This analysis showed conclusively that ridged Rhodesian and Thai Ridgebacks are heterozygous or homozygous for a large duplication (>100 kb) (Fig. 2a). We defined the precise location of the duplication breakpoints using additional interspersed probes in the critical intervals. Using PCR, we then amplified and sequenced the duplication breakpoint between the two tandem copies. The 133.4-kb duplication spans from nucleotide position 51,398,518 to position 51,531,941 in the CanFam2.0 genome assembly, with a single-nucleotide insertion at the breakpoint (Fig. 2b and Supplementary Fig. 1 online). Notably, sequencing of ∼1.7 kb flanking the breakpoint from Rhodesian and Thai Ridgebacks revealed identical breakpoints, although mitochondrial DNA analysis did not reveal a close relationship between the breeds (Supplementary Note, Supplementary Tables 2 and 3 and Supplementary Fig. 2 online). This strongly suggests that the duplication is the causative mutation for the ridge. Furthermore, it implies that either the Ridge allele has introgressed from one population to the other, or the mutation is old enough to have been present in the early domestic dog population.

(a) A 133-kb duplication on chromosome 18 in Ridgeback dogs detected using MLGA. Relative probe signals in two comparisons of a ridged dog and a ridgeless (r/r) dog; a value of ∼1 indicates the same copy number in ridged and ridgeless dogs. One of the ridged dogs was interpreted as homozygous for the duplication (R/R, black staples) and the other was assumed to be heterozygous (R/r, unfilled staples). Error bars, s.d. (b) Genome annotation of the duplicated region on dog chromosome 18. Genome-wide mapping SNPs (pink bars) and MLGA probes (blue bars) identify a duplicated region that contains four complete genes (FGF3, FGF4, FGF19 and ORAOV1) and the 3′ end of CCND1, but does not include conserved elements (yellow) and an enhancer region⁷ (red) upstream of FGF3. Dog, human, mouse and rat sequence conservation from the University of California, Santa Cruz genome browser (http://www.genome.ucsc.edu) is shown in black.

Full size image

The nucleotide sequence within 3 kb of the breakpoint did not contain any known repetitive elements that could have acted as target sites for the recombination event creating the duplication. Duplicated loci often show copy-number instability, but the MLGA analysis gave no clear indication of Ridge haplotypes with more than two copies of the duplicated region. However, this needs to be further investigated in a larger sample of ridged dogs, as it may be relevant for the incidence of dermoid sinus.

We detected a complete association between the duplication and the ridge phenotype in more than 50 Rhodesian and Thai Ridgebacks (Table 1). This is consistent with a fully penetrant, dominant inheritance of the ridge. Furthermore, 13 of 15 Ridgebacks with dermoid sinus were homozygous for the duplication. Thus, the mutation predisposes to dermoid sinus with low penetrance in duplication heterozygotes and with high penetrance in homozygotes. Data indicate that about 80% of healthy ridged dogs in the Swedish Rhodesian Ridgeback population are heterozygous for the duplication, suggesting that about 16% of newborn dogs ought to be ridgeless (0.8² × 1/4). This is threefold higher than the estimated rate of 5–6%^3,4, suggesting either that the frequency of the undesired ridgeless phenotype is under-reported, or that R/R dogs without dermoid sinus are over-represented in the breeding population.

The duplication includes FGF3, FGF4, FGF19, ORAOV1 and the 3′ end of CCND1 encoding cyclin D1 (Fig. 2b). The duplication may act as a regulatory mutation, either through a dosage effect or by disrupting the location of coding sequences in relation to important regulatory elements. The proximal duplication breakpoint is ∼10 kb upstream of FGF3 but downstream of an enhancer critical for embryonic expression of FGF3 (Fig. 2b)⁷. The next gene resides ∼200 kb upstream, suggesting that multiple regulatory elements essential for normal expression of these FGFs may be located in this region. Both the hair ridge and dermoid sinus are most likely caused by dysregulated gene expression during early embryogenesis. CCND1 is unlikely to contribute to these phenotypes, because only the 3′ end of the gene is duplicated. ORAOV1 (oral cancer overexpressed 1) is a poorly characterized proto-oncogene whose function during development is unknown⁸. In contrast, it is well established that tight regulation of FGF expression is crucial during embryonic development, including hair follicle morphogenesis⁹. Several of the 22 known mammalian FGF genes have been implicated in regulation of both hair growth and skin development¹⁰. Ridgeback dogs apparently have a mild defect of the planar cell polarity system, which is required both for a normal orientation of hair follicles¹¹ and for neural-tube closure¹². The dorsal hair ridge in Ridgeback dogs resembles the global disorganization of hair orientation in frizzled 6 (Fzd6) knockout mice¹¹. Notably, Fzd3 Fzd6 double-knockout mice also show severe defects in neural tube closure¹³. Thus, we propose that dysregulated FGF expression along the dorsal midline during embryonic development leads to disorganized hair follicles and an increased risk of dermoid sinus in Ridgeback dogs. We did not have access to embryonic tissue to test this hypothesis, but expression of ORAOV1 mRNA was about twofold higher in postnatal tissue from homozygous ridged dogs compared with ridgeless dogs, consistent with a chromosome duplication (Supplementary Note and Supplementary Fig. 3 online).

Our results have implications for the breeding of Ridgeback dogs. Neither ridgeless dogs nor those with dermoid sinus are allowed for breeding by Rhodesian Ridgeback clubs. This leads to overdominance, because the heterozygote is the favored genotype: it expresses the ridge and has a low incidence of dermoid sinus. The problem with dermoid sinus could be virtually eliminated by allowing ridgeless dogs in breeding and by avoiding matings between ridged dogs.

In humans, dermal sinus is often associated with spina bifida occulta, but it may also occur independently¹⁴. Both disorders are collectively categorized as neural-tube defects, with a poorly understood genetic background¹⁵. An obvious topic for future research will be to test whether mutations in the region corresponding to the Ridge duplication contribute to human malformations.

URL. Information on the CanFam2.0 genome is available at http://www.genome.ucsc.edu.

Note: Supplementary information is available on the Nature Genetics website.

We thank C. Knudsen for access to dorsal skin samples from Rhodesian Ridgebacks, L. Gaffney for help with illustrations, D. Wright for statistical analysis, L. Spångberg for technical assistance and the Broad Institute of MIT and Harvard Genetic analysis Platform for SNP array genotyping. The work was supported by the Foundation for Strategic Research, the Swedish Research Council for Environment, Agricultural Sciences and Spatial Planning, the Swedish Research Council, the Swedish Kennel Club, the Thure F and Karin Forsbergs Foundation, the American Kennel Club/Canine Health Foundation (grant 373) and EU FP6 IP MolTools.

Affiliations

1. Department of Animal Breeding and Genetics, Swedish University of Agricultural Sciences (SLU), Biomedical Centre, Box 597, SE-751 24 Uppsala, Sweden.

   • Nicolette H C Salmon Hillbertz
   • , Ulla Gustafson
   • , Leif Andersson
   •  & Göran Andersson

2. Department of Genetics and Pathology, The Rudbeck Laboratory, Uppsala University, SE-751 85 Uppsala, Sweden.

   • Magnus Isaksson
   • , Eva Hellmén
   •  & Mats Nilsson

3. Broad Institute of Harvard and MIT, 7 Cambridge Center, Cambridge, Massachusetts 02142, USA.

   • Elinor K Karlsson
   • , Claire M Wade
   •  & Kerstin Lindblad-Toh

4. Bioinformatics Program, Boston University, 44 Cummington Street, Boston, Massachusetts 02215, USA.

   • Elinor K Karlsson

5. Department of Anatomy, Physiology and Biochemistry, SLU, Box 7011, SE-750 07 Uppsala, Sweden.

   • Eva Hellmén

6. Department of Medical Biochemistry and Microbiology, Uppsala University, Box 597, SE-751 24 Uppsala, Sweden.

   • Gerli Rosengren Pielberg
   • , Kerstin Lindblad-Toh
   •  & Leif Andersson

7. School of Biotechnology, Royal Institute of Technology, AlbaNova University Center, SE-106 91 Stockholm, Sweden.

   • Peter Savolainen

8. Center for Human Genetic Research, Massachusetts General Hospital, Boston, Massachusetts 02114, USA.

   • Claire M Wade

9. Department of Clinical Sciences, SLU, Box 7054, SE-750 07 Uppsala, Sweden.

   • Henrik von Euler
   •  & Åke Hedhammar

Contributions

N.H.C.S.H. was responsible for collection of field material, histopathological confirmation of dermoid sinus and autopsies together with E.H., phenotypic characterization of the field material, break-point analyses together with M.I., and real-time PCR analyses; M.I. and M.N. were responsible for the MLGA analyses; E.K.K. carried out the SNP association and bioinformatics analyses; E.H. carried out the histological analyses; G.R.P. did pyrosequencing, P.S. carried out the sequencing of mitochondrial DNA and performed phylogenetic analyses; C.M.W. contributed to the SNP screen, H.v.E. took part in the collection of tissue samples; U.G. provided technical assistance; Å.H. provided advice for the clinical part of the study; K.L.-T., L.A. and G.A. co-directed the study; and N.H.C.S.H., L.A., G.A., and K.L.-T. were responsible for preparation of the manuscript with input from the other authors.

Competing interests

A patent application concerning commercial testing for the presence of the Ridge mutation in dogs has been filed. N.H.C.S.-H., G.A., L.A., K.L.-T., M.N. and M.I. are listed as inventors on a provisional application to the US Patent Office (no. 60/928240) related to this work.

Corresponding authors

Correspondence to Leif Andersson or Göran Andersson.

To obtain permission to re-use content from this article visit RightsLink.

Publication history

DOI

https://doi.org/10.1038/ng.2007.4