Dairy production depends on the frequency at which cows conceive; thus, female fertility is essential for this industry. To improve overall female fertility, the dairy industries in certain countries have combined several reproductive components into one overall index: the fertility selection index (SI) 1234. In Japan, the SI consists of the estimated breeding values (EBVs) for days open (DO), the number of inseminations per lactation (NI), success after first insemination (SFI), and pregnancy within 70 d (P70), 90 d (P90), and 110 d (P110) after parity 5. The first two traits are continuous traits, whereas the other four traits are binary traits that show either success or failure after insemination. The genetic correlations among these reproductive traits are not necessarily ± 1.0 13. Therefore, a principal components analysis is used to estimate weights for each trait and combine them into the SI to allow for the selection of these six fertility traits simultaneously.

Several groups have already conducted genome-wide association studies (GWASs) for SI using Danish Jersey 4, Nordic Red 6, and Italian Holstein 7. These studies revealed some single nucleotide polymorphisms (SNPs) associated with both SI and individual traits included SI, suggesting that important fertility traits for selection in the population would generate the signal. Thus, studying the genetics of the SI using a GWAS may be helpful for understanding the underlying biological mechanisms and improving the general fertility level of the population.

Our GWAS identified BovineHD2800009796 as the most significantly associated SNP on chromosome 28 (Figure 1B), and there were no more SNP that may be impacting SI as shown in Graph A in the Supplemental data. The region associated with this SNP included 13 genes (Figure 2A). To detect possible causative polymorphisms in this region, we sequenced all exons and the 5’ and 3’ UTRs of these 13 genes and found 42 novel SNPs (Figure 2A). The positions, the minor allele frequencies, and the genotype counts in the GWAS sample for the novel 42 SNPs in Table E in the Supplemental data. A reanalysis of the newly sequenced SNPs demonstrated that FAM213A (A31G: I11V) was the most significant (Figure 2A). Moreover, we genotyped FAM213A (A31G: I11V) in 2,682 cows and 4,165 bulls and found that the allele substitution effect of FAM213A on the dEBV for the SI was 0.18 (Figure 2B). Cows with the A/A genotype, which codes for Ile/Ile, exhibited a 0.18 higher SI than those with the A/G genotype, which codes for Ile/Val, whereas cows with the G/G genotype, which codes for Val/Val, exhibited a 0.18 lower SI than those with the A/G genotype. FAM213A also had favorable effects on the dEBV for the traits that compose the SI (DO, NI, SFI, P90, and P110). The effects of FAM213A were similar to those of PKP2 (chromosome 5: 82,452,366 bp), CTTNBP2NL (chromosome 3: 33,171,753 bp), SETD6 (chromosome 18: 26,204,767 bp), CACNB2 (chromosome 13: 32,851,153 bp), and UNC5C (chromosome 6: 31,142,900 bp), which have previously been identified to be associated with the CR in the Japanese dairy cow population 812 (Figure 2B). Therefore, FAM213A (A31G: I11V) was the most promising causative SNP on chromosome 28 and had a similar impact on the SI as the five genes previously identified to be associated with CR.

FAM213A protein is a CXXC motif-containing peroxiredoxin (PRX) 2-like protein 16. PRXs and thioredoxins (TRXs) are redox regulatory proteins. PRXs play a protective antioxidant role in cells through their peroxidase activity, whereas TRXs alter the redox state of target proteins by catalyzing the reduction of their disulfide bonds through a CXXC motif. FAM213A has a similar sequence to PRXs and contains a CXXC motif that is shared by many enzymes with redox regulatory functions. Thus our identified two types of FAM213A might modulate reactive oxygen species (ROS) differently. To assess its redox activity, we used BEnEpCs derived from bovine uterine tissue because FAM213A is expressed in several bovine tissues including the uterus (Figure 3A). Cell-viability assays using BEnEpCs transfected with FAM213A revealed that oxidative stress induced by H₂O₂ treatment decreased cell viability in a dose-dependent manner and indicated that cells with FAM213A^Val had increased viability relative to that of FAM213A^Ile in the presence of 0.2 mM H₂O₂ (Figure 3B). Consistent with the results of the cell-viability assay, the GSH/GSSG ratio, a proxy for the cellular redox status, was increased in BEnEpCs transfected with FAM213A^Val to a greater extent than in BEnEpCs transfected with FAM213A^Ile (Figure 3C). Furthermore, the GOR prediction for a protein’s secondary structure 17 revealed that FAM213A^Val would likely have a shorter b-sheet than FAM213A^Ile (Figure 3D). These results suggest that the activity of FAM213A^Val shifts conditions towards a more reduced environment than does FAM213A^Ile and indicate that this single amino acid substitution has an impact on the function of this protein.

ROS are important factors for fertility. In the uterus, the establishment of a maternal-fetal connection is associated with an increase in ROS 18. Although ruminants have different types of placentation relative to humans or mice, ROS fluxes occur before implantation in all of these mammals 19. Increases of ROS production by peripheral bovine lymphocytes have been reported as pregnancy progresses 20. ROS triggers the NF-κB pathway, which enhances COX-2 expression 2122, suggesting that the more reduced environment created by FAM213A^Val might decrease COX-2 expression. To explore this possibility, we conducted real-time PCR using RNA extracted from BEnEpCs transfected with FAM213A. As expected, BEnEpCs transfected with FAM213A^Val expressed less COX-2 than cells transfected with FAM213A^Ile (Figure 4A). Moreover, pretreatment of BEnEpCs with a selective inhibitor of NF-κB, 10 µM PDTC, significantly repressed COX-2 expression (Figure 4A). We found the same result in the level of protein expression (Figure 4B). Thus, the more reduced environment created by FAM213A^Val appears to decrease COX-2 expression through effects on the NF-κB pathway.

A COX-2 deficiency in mice results in implantation defects 23, and one of the products of COX-2, PGE2, can correct these implantation defects in COX-2-deficient mice 24. Therefore, decreased COX-2 expression might reduce the production of PGE2 to levels that are inadequate for implantation. Our enzyme immunoassays revealed that BEnEpCs transfected with FAM213A^Val released less PGE2 compared with cells transfected with FAM213A^Ile (Figure 5A). Furthermore, pretreatment of BEnEpCs with a selective inhibitor of COX-2, 10 µM NS-398, or a selective inhibitor of NF-κB, 10 µM PDTC, significantly repressed PGE2production (Figure 5A). Additionally, the serum concentrations of PGE2 of cows carrying FAM213A^Ile/Ile were higher than those of cows carrying FAM213A^Val/Ile or FAM213A^Val/Val (Figure 5B). Therefore, the reduced uterine environment created by FAM213A^Val decreased PGE2production through effects on the NF-κB/COX-2 pathway, which might decrease the SI of cows carrying FAM213A^Val.

The present study is the first to demonstrate that FAM213A modulates fertility. Scanning the whole genome of 442 Holsteins and analyzing the results with EMMAX identified a novel mutation associated with the SI. Although we have previously identified several genes associated with the CR in the Japanese Holstein population using a principal component analysis approach 813, FAM213A is a novel gene associated with fertility, and it has only previously been reported to modulate osteoclast differentiation in vitro16. One of the reasons for this GWAS result might be that the SI follows a normal distribution, which makes it possible to compare samples belonging to the 5^th percentile with samples belonging to the 95^th percentile of the population (Figure 1A). A second reason might be that the EMMAX method could correct for severe stratification and reduce the effects of unusual clusters of cows that are significantly larger because of intensive genetic selection and the extensive use of artificial insemination in the dairy industry 25.

However, in the present GWAS, significant SNPs were not observed near the genes previously identified as associated with CR, although the allelic substitution effects of these genes on the SI were significant among the 2,682 cows and 4,165 bulls (Figure 2B) that were analyzed. The SI consists of reproductive traits that have low heritabilities ranging from 0.05 to 0.14 5. In dairy cattle, only DGAT1 and ABCG2 have been identified and verified by multiple studies to have significant effects on fat and protein concentrations, which have the highest heritabilities of all of the traits that are routinely analyzed 26. The heritabilities of these traits range from 0.20 to 0.61 27. Thus, using multiple GWASs to identify and verify the genes responsible for reproductive traits that have low heritabilities might be more difficult than for milk production traits that have high heritabilities.

We found that cows carrying FAM213A^Val have a lower SI than cows carrying FAM213A^Ile. This single amino acid substitution is located in the thioredoxin fold region and predicted to change the length of the b-sheet (Figure 3D) in this region. The thioredoxin fold region contains a b-a-b-a-b-a-b-b-a arrangement 28, and the length of the first b-sheet of the region might be critical for the redox regulatory function of this protein.

The identified single amino acid substitution decreases ROS, which reduces COX-2 expression through the NF-κB pathway (Figure 4) and reduces PGE2 production (Figure 5). It is worth noting that COX-2 expression and PGE2 production are also induced by interferon-tau 29, the pregnancy recognition signal produced by the trophoblast prior to implantation in ruminants, which has been shown to enhance fetal survival when administered to mice 3031. Along with the interferon-tau pathway, enhancing COX-2 expression and PGE2 production via the ROS/NF-κB pathway might be important for implantation.

Although we demonstrated effects using BEnEpCs in the present study, FAM213A might also have impacts on different tissues in cows because the SI consists of several reproductive components 5. In the ovary, oocyte maturation requires ROS 32, and FAM213A was found to be expressed in the ovary (Figure 3A). Ovulation is initiated by a surge in luteinizing hormone, which generates ROS 33. After ovulation, the corpus luteum is produced in the ovary, which also generates ROS 34. Moreover, COX-2-deficient mice have demonstrated multiple failures in female reproductive processes, including fertilization, ovulation, and implantation 23. Thus FAM213A might affect multiple reproductive processes through the ROS/NF-κB/COX-2/PGE2 pathway.

The present study investigated the whole genome in samples from 212 cows with a low SI and samples from 230 cows with a high SI, and a significant association was observed between the SI and FAM213A with a single amino acid polymorphism. Further functional studies revealed that this single amino acid polymorphism in FAM213A reduced ROS, thus leading to decreased COX-2 expression and lower PGE2 production. These results indicate that FAM213A plays a role in female fertility in cows.

The authors thank K. Maruyama for performing extensive genotyping and S. Sasaki for providing useful discussions.

Supplementary Data

Primers used to search for SNPs (Table A). Samples used for developing new SNPs (Table B). Primers used for real-time PCR (Table C). Primers used for generating the overexpression constructs (Table D). A quantile-quantile plot of the GWAS for the fertility selection index without BovineHD2800009796 (Graph A). The minor allele frequencies and the genotype counts in the GWAS sample for 42 novel SNPs in chromosome 28 (Table E). The standard error for the allelic substitution effects of FAM213A, PKP2, CTTNBP2NL, SETD6, CACNB2, and UNC5C on the deregressed estimated breeding value for the SI traits days open (DO), number of inseminations per lactation (NI), success after first insemination (SFI), pregnancy within 70 d (P70), 90 d (P90), and 110 d (P110) after parity or for the conception rate (CR) (Table F).