Genomic and Transcriptomic Characterization of Atypical Recurrent Flank Alopecia in the Cesky Fousek

Abstract

Non-inflammatory alopecia is a frequent skin problem in dogs, causing damaged coat integrity and compromised appearance of affected individuals. In this study, we examined the Cesky Fousek breed, which displays atypical recurrent flank alopecia (aRFA) at a high frequency. This type of alopecia can be quite severe and is characterized by seasonal episodes of well demarcated alopecic areas without hyperpigmentation. The genetic component responsible for aRFA remains unknown. Thus, here we aimed to identify variants involved in aRFA using a combination of histological, genomic, and transcriptomic data. We showed that aRFA is histologically similar to recurrent flank alopecia, characterized by a lack of anagen hair follicles and the presence of severely shortened telogen or kenogen hair follicles. We performed a genome-wide association study (GWAS) using 216 dogs phenotyped for aRFA and identified associations on chromosomes 19, 8, 30, 36, and 21, highlighting 144 candidate genes, which suggests a polygenic basis for aRFA. By comparing the skin cell transcription pattern of six aRFA and five control dogs, we identified 236 strongly differentially expressed genes (DEGs). We showed that the GWAS genes associated with aRFA are often predicted to interact with DEGs, suggesting their joint contribution to the development of the disease. Together, these genes affect four major metabolic pathways connected to aRFA: collagen formation, muscle structure/contraction, lipid metabolism, and the immune system.

Keywords: Cesky Fousek; GWAS; RNA-seq; atypical recurrent flank alopecia; canine alopecia; differential gene expression; dog; skin biopsies.

Figure 1

Cesky Fousek individuals affected by aRFA. (A) unusual manifestation on the head; (B) level 1 aRFA—loss of hair on ears only; (C) level 2 aRFA—loss of hair on the body sides up to approximately 10 × 10 cm; (D) level 3 aRFA—loss of hair on the body sides up to approximately 10 × 25 cm; (E) level 4 aRFA—loss of hair of the body sides up to approximately 10 × 40 cm. Pictures A and E were taken prior to the hair-loss peak in these individuals; alopecia worsened in the weeks after the pictures were taken.

Figure 2

Manhattan and QQ plot for case/control GWAS. The chromosomes of the nine most significant SNPs are shown in green. The significance threshold (shown as a purple line) was set based on Bonferroni correction (cut-off = 1.16 × 10⁻⁶). The lambda value is shown in the QQ plot.

Figure 3

Histological representation of biopsy samples from control skin from unaffected dogs (A) and affected dogs (B–F). Note numerous anagen hair follicles in A identified by the presence of numerous hair bulbs (black arrows). In aRFA, infundibuli (gray arrows) are moderately to severely dilated (B,C,E) and are filled with abundant keratin, which extends into the openings of the secondary follicles, resulting in a “witch’s feet”-like appearance (E). The follicular parts proximal to the infundibuli are shortened and limited to the dermis (B–E). A few telogen follicles (C, white arrow) or kenogen follicles (F, white arrow) can be identified. Follicular atrophy may be seen (F, black cross). A mild distortion of the HFs is observed (B–F). All samples are stained with hematoxylin and eosin (H&E) and the scale bars represent 200 microns.

Figure 4

Principal component analysis (PCA) of samples demonstrating clustering based on expression profiles plotted against the two most variable components (PC1 and PC2). Samples from control animals (red) and normal skin from affected animals (green) cluster together, whereas samples from alopecic skin from affected animals (blue) are clearly separated from the clusters representing normal skin but show a higher inter-group variability.

Figure 5

Interactions of GWAS candidate genes (green) and STRING-associated strongly differentially expressed genes colored by their level of expression. We used only medium confidence associations and higher (increasing thickness of lines connecting genes indicates greater confidence). Colorful bubbles represent the metabolic pathways common for each cluster of genes.