Introduction

American mink breeding is entering the genomic era through the availability of a high-quality chromosome-based genome assembly [1] and a genome-wide single-nucleotide polymorphisms (SNPs) array. Such technologies have paved the way for the precise identification of homozygous segments in livestock species [2]. Runs of homozygosity (ROH) are homozygous regions, which are composed of two identical haplotypes inherited from a common ancestor [3]. Characteristics of ROH in a population can be used as an indicator for estimation of inbreeding level in different species, such as cattle [4, 5], pigs [6, 7], chicken [8, 9], sheep [10, 11], goat [12, 13], and buffalo [14, 15].

Groups of several ROH within a specific region of the genome in a population are known as ROH islands [16]. It was reported that the analysis of ROH islands might reveal genomic regions under selection pressure, which in turn helps to identify candidate genes associated with traits of economic interest [17, 18]. Furthermore, several studies have suggested the feasibility of performing association analyses using ROH to detect homozygous genomic regions associated with complex traits in livestock [19]. Sanglard et al. [20] identified several regions of ROH significantly associated with antibody response to porcine reproductive and respiratory syndrome virus vaccination in pigs. In cattle, substantial numbers of ROH regions are reported to be associated with milk yield [21, 22], fertility [23, 24], and production traits [25], suggesting a complementary role of ROH in elucidating the genetic mechanisms underlying economically important traits.

Feed cost is the largest expense for mink production systems, and thereby, improving feed efficiency holds significant potential for increasing the profitability of mink farming through strategic breeding programs [26]. Several studies have reported moderate to high heritability for growth [27, 28] and feed efficiency traits [26, 29, 30] in American mink, which highlighted a substantial genetic basis and presented opportunities for improvement by genetic and genomic breeding programs.

To the best of our knowledge, there is no study that examined homozygous segments in the American mink genome and their potential association with growth and feed efficiency traits. Therefore, the main objectives of this study were to (1) reveal the distribution and pattern of ROH within the genome of American mink; (2) identify highly frequent consensus ROH (ROH islands) and investigate the candidate genes within these regions; and (3) assess their associations with growth and feed efficiency traits.

Materials and methods

Animals and traits

Mink were humanely euthanized using carbon monoxide sourced from a compressed gas cylinder, adhering to the protocols outlined in the Canada Mink Breeders Association’s Code of Practice for the Care and Handling of Farmed Mink (ISBN 978-1-988793-24-5) (https://www.nfacc.ca/codes-of-practice/farmed-mink). The procedure involved maintaining a minimum of 4% carbon monoxide concentration within the chamber to ensure a swift and irreversible onset of unconsciousness, leading to a quick and relatively painless death for the mink. Confirmation of the mink’s death was conducted through a thorough check for the cessation of vital signs, which included no movement, the absence of heartbeat and respiration, pupils that were fixed and dilated, and a lack of reflexive responses.

A detailed description of the phenotypic data used in this study can be found in Davoudi et al. [26]. Briefly, a total of 2,288 American mink with growth and feed efficiency records were available. These traits were collected according to Davoudi et al. [26]: final body weight (FBW), final body length (FBL), harvest weight (HW), harvest length (HL), daily feed intake (DFI), average daily gain (ADG), feed conversion ratio (FCR), Kleiber ratio (KR), residual feed intake (RFI), residual gain (RG), and residual intake and gain (RIG). Descriptive statistics for growth and feed efficiency in American mink are shown in Additional file 1: Table S1.

SNP genotyping and quality control

All mink were genotyped using the Affymetrix Mink 70K SNP array (Neogen, Lincoln, Nebraska, United States). Genotypes were pruned by PLINK 1.9 software based on the proportion of missing genotypes (> 0.95), individual call rate (> 0.90), and Hardy-Weinberg equilibrium (P > 10e-6). In addition, SNPs located on sex chromosomes were removed, resulting in a final data set of 49,268 SNPs for further analyses.

Assessment of runs of homozygosity

We used PLINK 1.9 software [31] to identify homozygous segments across autosomes of each individual’s genome. The ROH were discovered based on the sliding window approach with the following parameters: (1) sliding window of 50 SNPs across the genome; (2) a minimum ROH length of 1,000 kb; (3) the minimum SNP density was set to 50 kb/SNP; (4) maximum gap between consecutive homozygous SNPs was 1,000 kb; (5) only one heterozygous and one missing genotype were allowed; and (6) a minimum of 57 consecutive SNPs were included in an ROH, which was determined according to the formula proposed by Lencz et al. [32], to control the false positive rate of the identified ROH:

$$The\, minimal\, number\, of\, SNPs\, in\, an\, ROH=\frac{{log}_{e}\frac{\alpha }{{n}_{a}{n}_{s}}}{{log}_{e}(1-het)}$$

,

where \(\alpha\) is the percentage of false positive ROH (set to 0.05), \({n}_{s}\) is the number of genotyped SNPs per individual, \({n}_{a}\) is the number of individuals, and \(het\) is the proportion of heterozygosity across all SNPs.

Consensus regions and ROH islands

The ‘homozyg-group’ function of the PLINK 1.9 software [31] was applied to merge the individual ROH into different ROH groups in a pool containing the overlapping regions between all the individual ROH in the group i.e. the consensus homozygous region [25, 33]. We retained the consensus ROH with a minimum of five SNPs and a frequency of more than 5% for association analyses. In addition, to investigate the genomic regions with a high frequency of ROH in the population (ROH islands), a threshold of higher than 80% was defined for consensus ROH [34]. The overlapped genes within ROH islands were annotated from the American mink reference genome annotation file [1] through the ‘intersect’ function in Bedtools version 2.30.0 [35].

Association analyses between consensus ROH and phenotypes

According to the model described by Sanglard et al. [20], we evaluated the association between consensus ROH with growth and feed efficiency traits using the linear model as follows:

$$y=\mu +Xb+Zu+e,$$

where \(\varvec{y}\) is the vector of phenotypic observation, \(\mu\) is the grand mean, \(\varvec{b}\) is the vector of fixed effects, \(\varvec{X}\) and \(\varvec{Z}\) are the incidence matrices that relate the fixed and random effects with the dependent variable, respectively; \(\varvec{u}\) is the vector of random animal genetic effects and e is the vector of random residual effects. The random effects \(\varvec{u}\) and \(\varvec{e}\) were distributed as: \(\varvec{u} \sim N(0, \varvec{G}{\sigma }_{u}^{2})\) and \(\varvec{e} \sim N(0,\varvec{I}{\sigma }_{e}^{2})\), where \({\sigma }_{u}^{2}\) and \({\sigma }_{e}^{2}\) are the additive genetic and residual variances, respectively, G is the genomic relationship matrix, which was constructed by ASRgenomics package [36] using the VanRaden Eq. [37], and\(\varvec{I}\)is an identity matrix. The consensus ROH (n=196) were simultaneously fitted in the model as categorical fixed effects, coding as “yes” if the individual contained the ROH segment, or “no” otherwise. The other fixed effects, as described by [26], were farm (two farms), sex (male and female), color type (dark, demi, mahogany, pastel, and stardust), row-year (year: 2018 and 2019; row: 1, 4, 5, 7, 8, and 11). We included color type in our analyses because it significantly impacts growth parameters in American mink, likely due to the pleiotropic effects of genes controlling both feed efficiency and color type [26]. The age of animals (in days; with a minimum and maximum of 184 and 229, respectively) was included as a covariate in the model. The fixed effects and covariate were statistically tested (P < 0.01) using univariate models in ASReml 4.0 [38]. The associations between each consensus ROH and studied traits were tested through linear mixed model analysis in ASReml 4.0 [38] with a statistical significance level (P < 0.01).

Results

Assessment of runs of homozygosity

A total of 298,313 runs of homozygosity (ROH) were identified in the entire mink population studied. Detailed information on detected ROH in all individuals is provided in Additional file 1: Table S2. The results showed that the average number of ROH segments per individual was 99.90, spanning from 30 to 134, respectively, and the length of ROH segments ranged from 1.02 to 55.44 Mb, with an average of 4.16 Mb (Table 1). We classified ROH segments into five different length categories, including 1–2 Mb, 2–4 Mb, 4–8 Mb, 8–16 Mb, and > 16 Mb (Fig. 1A). The majority of detected ROH were classified as 2–4 Mb, representing 46.99% of all ROH (n = 140,178), followed by the length of 4–8 Mb and 1–2 Mb with 26.3% (n = 78,451) and 17.42% (n = 51,967), respectively. The percentage of ROH segments higher than 16 Mb was only 1.08% of all detected ROH (n = 3,236). The distribution of ROH lengths across the genome is represented in Fig. 1B. The largest ROH was located on chromosome 1 (55.44 Mb with 1563 SNPs), and the shortest ROH was identified on chromosome 3 (1.02 Mb with 79 SNPs). Further, the number of ROH segments varied across chromosomes, ranging from the lowest in chromosome 9 (n = 5,728) to the highest in chromosome 1 (n = 52,311). As shown in Fig. 1C, the total length of the genome covered by ROH among individuals ranged from 84.78 Mb to 683.16 Mb, with an average of 414.81 Mb.

Table 1 Descriptive statistics of runs of homozygosity (ROH) number and length by ROH length class
Full size table
Fig. 1
figure 1

Characteristics of runs of homozygosity in American mink: (A) Frequency distribution of the average number of ROH in different length classes (Mb) in each chromosome; (B) Length distribution of ROH; (C) Relationship between ROH number per animal and total length of the genome covered by them. Each point represents one individual

Full size image

Consensus regions and ROH islands

To provide the shared homozygous regions for the association analyses, initially, 6,980 consensus groups were formed using ‘–homozyg-group’ function in PLINK 1.9 software, of which a total of 196 consensus ROH fulfilled the criteria of presenting in more than 5% of individuals with a minimum of five SNPs (Additional file 1: Table S3). The chromosomal distribution map of identified ROH across mink autosomes and consensus ROH shared among individuals is shown in Fig. 2.

Fig. 2
figure 2

Chromosome ideograms showing the position of identified ROH and consensus ROH shared between individuals. The color scale within each chromosome represents the number of identified ROH, changing the gradient with more ROH detected in an area. The position of consensus ROH is marked with the green triangle next to the chromosome

Full size image

The ROH islands were determined as regions where the consensus ROH was presented in more than 80% of animals, with the aim of pinpointing the genes they encompass. The implementation of this approach resulted in the detection of ten ROH islands spanning 14 autosomes, most of which were located on chromosome six with seven ROH islands. These specific regions harbored 12 annotated genes, some with known effects on immune systems processes such as DTX3L, PARP9, PARP14, CD86, and HCLS1 (Table 2). Notably, the three ROH islands on other chromosomes did not contain any known annotated genes.

Table 2 Characterization of consensus ROH shared by more than 80% of the mink population and the annotated genes in the corresponding regions
Full size table

Association analyses between consensus ROH and phenotypes

The association analysis revealed 13 consensus regions that were significantly (P < 0.01) associated with growth and feed efficiency traits, of which four ROH affected more than one trait. The physical position of significant consensus ROH across the mink autosomes is shown in Fig. 3. The frequency of the associated consensus regions ranged from 6.6 to 81.9% across all individuals. The average length of significant consensus ROH was 147.46 kb, ranging from 8.62 to 327.85 kb. Chromosome one exhibited the highest number of significant regions (n = 5), followed by two significant regions on chromosome 13, and one significant region on chromosomes 2, 4, 5, 8, and 9. Detailed information regarding the consensus ROH significantly associated (P < 0.01) with the studied traits, along with their annotated candidate genes can be found in Table 3.

Fig. 3
figure 3

Physical position of significant consensus ROH across the mink autosomes. FBL: Final body length, ADG: Average daily gain, RG: Residual gain, HW: Harvest weight, KR: Kleiber ratio, FBW: Final body weight, RIG: Residual intake and gain, RFI: Residual feed intake, DFI: Daily feed intake

Full size image
Table 3 Regions of runs of homozygosity (ROH) significantly associated with growth and feed efficiency traits in American mink
Full size table

Discussion

In this study, the mean number of ROH per individual was 99.9, which was in agreement with Karimi et al. [1] who reported an average of 102 per animal using whole-genome sequencing data of 100 American mink. Yet, both studies reported higher numbers of ROH counts compared to the study of Karimi et al. [39], which identified 82 ROH segments per individual solely based on scaffolds. This discrepancy indicates that the recent chromosome-based reference genome in American mink has facilitated our capacity to detect homozygous segments. The distribution of detected ROH revealed that approximately more than 90% of ROH were shorter than 8 Mb, which was consistent with the results reported in other species, such as cattle [17, 40], pigs [7, 41], chicken [42, 43], sheep [44, 45], and buffalo [14, 46]. It is well-established that the large ROH (~ 10 Mb) represents recent inbreeding (up to five generations ago), whereas short ROH (~ 1 Mb) indicates more distant ancestral effects (up to 50 generations ago) [47, 48]. Considering the predominant of ROH with a length of 1 to 8 Mb, it is reasonable to hypothesize that the inbreeding events in American mink occurred approximately 5 to 50 generations ago. This timeline corresponds with the findings of Hu et al. (2023), who reported the rapid decline in the effective population size in American mink from 5 to 50 generations ago.

In recent years, the identification of ROH islands across the genome has gained popularity due to their capacity to reveal selection footprint in livestock species [49]. The Aleutian disease, the most significant health concern for global mink farming, is an immune complex disease that causes autoimmune disorders in mink [50]. Despite efforts to detect and eliminate infected animals using various immunological tests, these strategies have largely failed due to the high persistence nature of Aleutian disease in the breeding environment [51, 52]. Intriguingly, our study uncovered several genes within ROH islands known to affect immune system processes, including DTX3L, PARP9, PARP14, CD86, and HCLS1. This implies that natural selection plausibly acts on immune-related genes in American mink.

The DTX3L gene, also known as BBAP (B-lymphoma and BAL-associated protein), plays regulatory functions on DNA damage signaling, tumor cell growth, and IFN signaling and antiviral response [53,54,55]. Interestingly, Hong et al. [56] reported that inhibiting the DTX3L gene restrained the cell invasion and secretion of inflammatory factors, suggesting its potential as a therapeutic target for rheumatoid arthritis, a complex autoimmune disease characterized by chronic synovitis of the joints in humans. The PARP9 and PARP14 genes, located within the ROH island on chromosome 6 (121,883,426:122,139,161 bp), belong to the PARP superfamily that regulate diverse biological processes such as DNA damage repair, cellular stress response, and antiviral innate immunity [57]. Research has demonstrated that PARP9 gene, highly expressed in glioma, is correlated with checkpoint molecules involved in inflammatory and immune responses [58]. Moreover, study has shown that knockdown of PARP9 gene in human or mouse dendritic cells and macrophages resulted in substantial reduction of type I IFN production (IFN-α and IFN-β), highlighting its critical role in the antiviral immunity system [59]. Similarly, PARP14 knockout has shown therapeutic effects on tumors and allergic inflammation through mediating T-cell differentiation and action of cytokines [60, 61]. Other genes of interest were CD86 and HCLS1 located within two different ROH islands on chromosome six (122,500,609:122,510,002 bp and 122,908,246:122,958,392 bp, respectively). Several lines of evidence indicated that CD86, which is one of the essential co-stimulatory molecules expressed on antigen presenting cells, plays a regulatory role in the immune response by mediating the activation of T-cells, B-lymphocytes, and macrophages [62, 63]. It was indicated that the HCLS1 gene, which is expressed only in cells with lymphohematopoietic origin, plays a functional role in the regulation of T-cell immune synapses [64].

It is well-documented that American mink is one of the most highly susceptible non-human species to severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) infection, leading to massive culls of many millions of mink across the world [65,66,67]. Intriguingly, most of the aforementioned genes, one way or another, have been reported to be associated with SARS-CoV-2, the virus that causes coronavirus disease-2019 (COVID-19). It was indicated that in SARS-CoV-2 infection, the activation of macrodomain-sensitive ADP-ribosylation signal is mediated by PARP9/DTX3L complex, suggesting their critical role in interferon-mediated antiviral defence [68]. Similarly, it was reported that the PARP14 gene is essential for the optimal IFN expression, supporting the suggestion that PARP14 is involved in antiviral immune response in CoV-infected cells [69]. Several studies have shown that the expression of CD86 on monocytes and dendritic cells was substantially decreased in patients with severe COVID-19 [70,71,72,73]. These findings merit further exploration of the functional role of the ROH islands-harbored genes revealed in the current study on the Aleutian mink disease virus and COVID-19 infection in American mink.

In the present study, gene discovery performed on the 13 consensus regions that were significantly (P < 0.01) associated with growth and feed efficiency traits, highlighted several candidate genes (i.e. MEF2A, ADAMTS17, POU3F2, and TYRO3) with potential impacts on growth rate and feed efficiency as reported in previous studies. The MEF2A and ADAMTS17 were located within the consensus ROH on chromosome 13 (134,704,541:135,003,083 bp), which was significantly (P < 0.01) associated with RFI. The MEF2A gene, which plays an important role in vertebrate skeletal muscle development and differentiation by activation of numerous muscle-specific and growth factor-induced genes [74], is known to be the candidate gene for muscle development and body growth in livestock species [75,76,77]. Remarkably, research conducted by Foroutan et al. [78] revealed that MEF2A showed higher expression levels across all tested tissues (liver, muscle, and testis) in the offspring of low-RFI Angus bulls, as opposed to their high-RFI counterparts. The ADAMTS17 gene, which is a member of ADAMTS proteins with numerous biological functions [79], has been previously reported as one of the height-associated variants in several species, such as horse [80], cattle [81], dog [82, 83], and human [84,85,86]. Interestingly, the ADAMTS17 gene was reported as a selective signal associated with animal height in the Shetland pony [87], and Brazilian locally adapted taurine cattle [88], highlighting the potential impacts of ADAMTS17 gene on body size.

The POU3F2 gene located within a ROH on chromosome 11 (32,128,510: 32,316,405 bp), is associated with HW and FBW traits. The POU3F2 gene, which is widely expressed in the central nervous system, has been well-described to play a key role in diverse neuronal functions and hormonal regulation [89, 90]. Notably, Schönauer et al. [91] reported a negative correlation of POU3F2 gene expression with body mass index in humans, suggesting the critical role of POU3F2 in hyperphagic obesity in humans. The TYRO3 gene was found within the consensus ROH on chromosome 13 (85,787,981: 86,060,018 bp), significantly associated with the KR trait. The TYRO3 gene, which is expressed in neurons of the central nervous system, plays regulatory roles in cell proliferation and differentiation, associating with adipocyte size in moderately obese individuals [92]. A GWAS analysis by Sun et al. [93] reported that TYRO3 gene was associated with intramuscular fat content in the breast muscle of chicken. Interestingly, it was revealed that TYRO3 was significantly differentially expressed in muscle between low and high RFI pigs, indicating that TYRO3 might affect the body fat, and consequently increase feed efficiency in pigs [94].

Conclusion

We characterized the distribution of ROH and ROH islands, and the association between the consensus ROH with growth and feed efficiency traits in American mink. In total, we identified 13 consensus regions significantly associated with the studied traits, harboring several candidate genes that are known to be associated with growth and body size development, such as MEF2A, ADAMTS17, POU3F2, and TYRO3. In addition, ten ROH islands were identified across the genome, harboring genes related to immune systems processes such as DTX3L, PARP9, PARP14, CD86, and HCLS1. Overall, the results revealed the impact of homozygosity in the mink genome on growth and efficiency traits. These findings have important implications for the evaluation and selection of American mink in genetic improvement programs, offering valuable insights for enhancing the breeding and sustainability of this species.