Litter Size of Sheep (Ovis aries): Inbreeding Depression and Homozygous Regions

Ovine litter size (LS) is an important trait showing variability within breeds. It remains largely unknown whether inbreeding depression on LS exists based on genomic homozygous regions, and whether the homozygous regions resulted from inbreeding are significantly associated with LS in sheep. We here reanalyze a set of single nucleotide polymorphism (SNP) chip of six breeds to characterize the patterns of runs of homozygosity (ROH), to evaluate inbreeding levels and inbreeding depressions on LS, and to identify candidate homozygous regions responsible for LS. Consequently, unique ROH patterns were observed among six sheep populations. Inbreeding depression on LS was only found in Hu sheep, where a significant reduction of 0.016, 0.02, and 0.02 per 1% elevated inbreeding FROH4–8, FROH > 8 and the total inbreeding measure was observed, respectively. Nine significantly homozygous regions were found for LS in Hu sheep, where some promising genes for LS possibly via regulation of the development of oocytes (NGF, AKT1, and SYCP1), fertilization (SPAG17, MORC1, TDRD9, ZFYVE21, ADGRB3, and CKB), embryo implantation (PPP1R13B, INF2, and VANGL1) and development (DPPA2, DPPA4, CDCA4, CSDE1, and ADSSL1), and reproductive health (NRG3, BAG5, CKB, and XRCC3) were identified. These results from the present study would provide insights into the genetic management and complementary understandings of LS in sheep.


Introduction
Litter size (LS), defined as the number of lambs born per ewe lambing, is of zoological and economical importance given the roles in survival of species and supply of products in sheep. LS is various within sheep breeds. Importantly, individual LS also varies among different parities. For instance, LS of 1-6 was recorded for Finnsheep, 1-5 for Romanov, 1-4 for Wadi, and 1-3 for Hu, Icelandic, and Texel sheep [1]. One possible explanation for the variability is the presence of inbreeding [2].
Inbreeding means the mating of related individuals, which is unfavorable in livestock due to the accompanying decrease on productive performances [3]. Inbreeding always extends the runs of homozygosity (referred to as ROH, homozygous segments) at the DNA level. ROH has been proposed to be used to calculate genomic inbreeding coefficients [4]. This measure was prevalent to evaluate inbreeding due to the consensus that it was more accurate using genomics data than only incorporating pedigree information [5]. Controlling inbreeding and inbreeding depression is a general goal in the management of livestock. However, the fact was that most inbreeding and inbreeding depression were evaluated only using pedigrees in sheep [6,7]. Hence, it would be interesting and necessary to use genomic information to evaluate the inbreeding coefficient and its effects on LS in different sheep breeds.
When the inbreeding depression on LS exists, the question then is, which homozygous regions resulted from inbreeding are associated with LS? As is well-known, single nucleotide polymorphisms (SNPs) have been widely used to identify quantitative trait loci Genes 2021, 12, 109 2 of 11 (QTLs) and candidate genes from a genome-wide landscape [8]. The ability of homozygous regions to expose deleterious variants makes them a potential agency used in association analyses [9], which would make the best of the massive SNP chip for previous genomewide association study (GWAS) based on SNPs. Recently, a growing body of interest was poured into the application of ROH in association with phenotypes in livestock [10,11]. Although some major genes such as BMPR1B, GDF9, and BMP15 have been identified by GWAS or QTL mapping in some sheep populations [12], the relationship between genomic homozygous regions and LS remains largely unknown. Thereupon, genetic findings on homozygous regions related to inbreeding may provide complementary understandings and shed light on the mechanisms underlying LS in sheep.
Under the hypothesis that some homozygous regions, resulted from inbreeding and contributing to inbreeding depression on LS, may be significantly associated with LS, we reanalyzed a set of SNP chip from a previous GWAS to characterize the patterns of ROH, to evaluate inbreeding levels and inbreeding depressions on LS, and to identify candidate homozygous segments responsible for LS in six sheep breeds. These breeds excluding Texel are potential genetic resources to improve fecundity of other populations. These results would provide insights into genetic management and complementary understandings of LS in sheep.

ROH Calling and Inbreeding
For each breed, ROH was detected using Plink v1.90 [13] with an observational genotype-counting method. ROH was defined as a segment of (i) at least 1 Mb length, (ii) at least 100 consecutive homozygous SNPs, (iii) at least 1 SNP per 50 kb, (iv) at most 1 Mb of consecutive homozygous SNPs, and (v) at most one heterozygote and five missing calls.
In order to characterize the pattern of ROH within breeds, we calculated some descriptive statistics at individual level, including the number of ROH, and average and total length. We also calculated the genomic inbreeding coefficient F ROH , which equaled to the proportion of total ROH length on autosome 2655.71 Mb used in the present study. To further describe the effects of inbreeding on prolificacy, we performed simple linear regressions using F ROH as an explanatory variable and ALS as our dependent variable with the linear model ALS = β 0 + β 1 F ROH + ε where β 0 was the intercept, β 1 was the slope, and ε was the error. F ROH was classified into three categories: (i) F ROH1-4 based on ROH of 1-4 Mb, (ii) F ROH4-8 based on ROH of 4-8 Mb, (iii) F ROH > 8 based on ROH with length of >8 Mb. Given the limited sample size, the descriptive statistics and inbreeding depression were performed for all individuals, not separately for case and control groups within each breed.

Genome-Wide ROH Hotspots Association Analysis
To explore the association between consensus homozygous segments and LS, we focused on the overlapping and potentially matching segments covering at least 10 SNPs and shared by at least 2 animals within each breed (here referred to as ROH hotspots). For simplicity, Fisher exact test was employed respectively to perform genome-wide casecontrol analysis (the presence and absence of ROH hotspots versus LS viz. high-yield or low-yield), followed by a Bonferroni correction (p-value = 0.05/the number of ROH hotspots tested) to control the false positive rate. Candidate ROH hotspots were mapped to the reference genome Oar_v.4.0 using ANNOVAR [14]. Metascape [15] was used to perform functional analyses for genes of interest, accepting a significance with a threshold of 0.05. Fisher exact test was also implemented to investigate the relationship between ALS and the variations within candidate ROH hotspots corrected by Bonferroni method using Plink v1.90 [13].

Quality Control and Population Structure
A total of 308 individuals passed the quality control, including 100 Wadi (77 cases vs. 23 controls), 71 Hu (58 cases vs. 13 controls), 23 Icelandic (8 cases vs. 15 controls), 37 Finnsheep (28 cases vs. 9 controls), 38 Romanov (29 cases vs. 9 controls), and 39 Texel sheep (28 cases vs. 11 controls). Significant differences of LS within each breed were shown in Table 1. The number of markers that remained for ROH detection was over 440,000 for each breed ( Table 1). The results of PCA showed the presence of slight population stratification ( Figure S1). We repeated the following analyses without the outlying animals and got very similar results. Thus, we did not rule them out considering the small simple size and low average pairwise IBD (Wadi, 0.004; Hu, 0.002; Icelandic, 0.028; Finnsheep, 0.018; Romanov, 0.069; and Texel, 0.035) ( Figure S1).

Pattern of ROH and Inbreeding Level
ROH features were characterized by number, and total and average length ( Table 2). The number of ROH of Hu and Wadi sheep is much lower than that of other breeds (Icelandic, Finnsheep, Romanov and Texel). For the total length of ROH, which is in line with individual inbreeding level, Icelandic has the maximum (306.98 Mb), followed by Texel (258.17 Mb), whereas Wadi possesses the minimum (84.97 Mb). However, for average ROH length Hu features the longest (4.89 Mb) and Romanov the shortest (2.91 Mb). Collectively, unique ROH patterns were observed among these six breeds. We also estimated the inbreeding coefficients for each breed (Table S1 and Table 3, Figure 1a). The extreme values (max 59.51% and min 0.10%) were in Hu population. Out of these ewes with F ROH > 20% (22 animals), a large portion (16 animals, 73%) were Hu, followed by Wadi (four animals, 18%), Icelandic (one animal, 4.5%), and Finnsheep (one animal, 4.5%) (Figure 1). Among breeds, low inbreeding levels and a similar trend in line with total ROH length were observed. The regression analyses show that most effects are not significantly different from zero in the breeds excluding Hu sheep, and even the effect of F ROH1-4 is favorable in Romanov sheep (Table 3). However, a reduction of 0.16, 0.02, and 0.02 for ALS occurs as a result of per 1% elevated F ROH4-8 , F ROH > 8 and the total inbreeding measure in Hu sheep, respectively (Table 3). These data imply the presence of inbreeding and inbreeding depression on ALS in Hu sheep.

Genome-Wide ROH Hotspots Association Analysis
Genome-wide association analyses based on ROH hotspots were performed for each breed to identify homozygous regions linked to LS (Figure 2). Unfortunately, no outliers were statistically observed in five populations including Wadi, Icelandic, Finnsheep, Romanov, and Texel sheep. However, nine ROH hotspots were identified for Hu sheep according to a threshold of 6.17 × 10 −5 (Table 4). Accordingly, a total of 56 genes were found partly or entirely on these regions (Table S2). These genes were functionally grouped into four pathways, including the PID PI3K PLC TRK pathway, apoptotic signaling pathway, response to calcium ion, and cellular process involved in reproduction in multicellular organism ( Figure 3 and Table S3). The relevance of candidate genes on reproduction was determined by a literature search. Consequently, straightforward connections between LS and two candidate genes (PPP1R13B and NGF) have been reported [16,17]. Additionally, the potential reproductive roles of candidate genes in development of oocytes and sperm, zygotic transcriptional program, embryo implantation and development, and reproductive diseases were uncovered (Figure 4).    (Table S3). To investigate promising variations at above regions, Fisher exact tests were implemented to explore the relationship between SNPs and ALS. Manhattan plots for each ROH hotspots of interest were shown in Figure S2. Only one intergenic marker on S325 was statistically significant, which was in the proximity of DPPA4 and DPPA2 ( Figure S2).

ROH and Inbreeding
In this study, we characterized the patterns of homozygosity in six sheep breeds, including two Chinese breeds and four non-Chinese breeds. Unique ROH patterns were observed, which would be attributed to their population history and selective breeding. A low genomic inbreeding coefficient (0.065) of Finnsheep was evaluated in the present study, in line with an evaluation from a previous study (0.060) [18]. F ROH of 0.093 was estimated for Hu sheep breed, which was larger than the previous value (0.048) using the same method [18]. One possible explanation for this discrepancy is different sample size and density of SNP chip. Interestingly, only the inbreeding level of Wadi sheep (0.032, equals to the average of 41 Chinese breeds) is lower compared with the global average (0.046) derived from 1910 animals in 97 populations [18]. For Wadi sheep, the low inbreeding means an effective genetic management. However, for other five breeds measures should be taken to control inbreeding.
Unfavorable effects of inbreeding on LS were observed in Hu sheep, which coincides with the inbreeding depressions of Danish and Czech populations based on pedigree [7,19]. It is important to note that Hu sheep take a large portion (16/22) of these ewes with a high inbreeding level (F ROH > 0.2), of which the maxima is about 0.6. This coincides with the history of mother-son mating as long as at least 900 years in Hu sheep [20]. Specifically, short ROH (1-4 Mb) can represent ancient inbreeding, and the unfavorable effect of F ROH1-4 on ALS was supported by the history of mother-son mating. On the other hand, the recent inbreeding resulted in long ROH (>8 Mb) and the significant decrease of ALS suggested recently ineffective genetic management in Hu sheep. The ancient and recent inbreeding highlighted the urgence of measures to maintain high fecundity and limit inbreeding in Hu sheep. However, these results are from a small herd and more animals are needed in future studies.

Candidate Genes Likely Functioning in Prolificacy
Significant homozygous regions were found for LS in Hu sheep rather than the others in the present study, which supported our hypothesis that some homozygous regions contributing to inbreeding depression on LS, may be significantly associated with LS. From the reproductive standpoint of ewes, LS is the outcome of a series of physiological processes, including the development of oocytes, fertilization, embryo implantation and development [21,22]. Hence, more or less effects of genes associated with these processes were expected on the ending of ovine delivery, supported (at least partly) by these candidate genes identified in Hu sheep. Although previous prolificacy genes did not emerge from the present study, some function-associated genes were identified (Figure 4).
Straightforward evidence was well-documented that PPP1R13B and NGF influenced LS. Loss of PPP1R13B, also known as p53, female mice decreased significantly embryonic implantation, pregnancy rate and LS via regulating leukaemia inhibitory factor, which was a cytokine critical for implantation [16,23]. In pigs, 17β-estradiol, one important reproductive hormone, was found to promote transcription of p53 [24]. Interestingly, the polymorphism of NGF, playing a role in augmenting folliculogenesis, has been reported to be associated with LS in both sheep and goats [17,[25][26][27].
AKT1, one isoform of AKT (a serine/threonine kinase), has been extensively studied in reproductive processes through phosphoinositol 1,3 kinase/protein kinase B (PI3K/AKT) signaling, including follicular development and embryo implantation [28][29][30]. The roles of PI3K/AKT signaling in follicular development consist of (i) the initiation of primordial follicle growth via suppresses FOXO3 (a transcriptional factor keeping primordial follicle dormancy) actions, (ii) the regulation in granulosa cell differentiation of antral follicles via FSH (follicle-stimulating hormone), and (iii) the participation in oocyte maturation of preovulatory follicles [31][32][33]. In addition, SYCP1 is required for the formation of crossovers during meiotic prophase [34], indicating its role in the development of oocytes.
Embryo implantation and development is another important reproductive event influencing LS. Zygotic genome activation launches the expression of parental genomes, during which any wrongdoing may terminate embryo development [35]. DPPA2 and DPPA4 were involved in zygotic genome activation by regulating Dux and LINE-1 retrotransposons [36,37]. Protein inhibitor of activated STAT 4 (PIAS4) and histone methyltransferase SETDB1 could negatively modulate DPPA2 protein activity [35,38]. Late embryonic/perinatal death was also observed in Dppa4-deficient mice [39]. Furthermore, Cdca4 was identified as a target gene of miR-154 differently expressed between 2-cell and 4-cell mouse embryos [40]. The placenta is an organ linking baby and mom during pregnancy, on which proliferation, differentiation, and invasion of trophoblast cells are vital to healthy pregnancy. Both maternal and fetal phenotypes of placental insufficiency were observed in the Inf2-lacking mice [41], highlighting the essential roles of Inf2 during pregnancy. Vangl1, a principal component of Wnt/PCP signaling pathway, is required for embryonic stem cells (ESC) differentiation, early embryo development and embryo implantation [42,43]. Highly expressed in human ESC, CSDE1 was known as a central post-transcriptional regulator of ESC identity and neurogenesis [44]. ADSSL1 was identified as a candidate gene for fetal akinesia [45]. Collectively, these candidate genes play crucial roles in embryo implantation and development.

Conclusions
The present study grounded on an SNP chip explored genome-wide homozygosity in six sheep flocks and pinpointed the presence of inbreeding depression for LS in Hu sheep. Additionally, based on genome-wide ROH hotspots associated analysis, this study identified some promising genes for LS possibly via regulation of the development of oocytes (NGF, AKT1, and SYCP1), fertilization (SPAG17, MORC1, TDRD9, ZFYVE21, AD-GRB3, and CKB), embryo implantation (PPP1R13B, INF2, and VANGL1) and development (DPPA2, DPPA4, CDCA4, CSDE1, and ADSSL1), and reproductive health (NRG3, BAG5, CKB, and XRCC3). These results would provide insights into the genetic management and complementary understandings of LS in sheep.