24th Int. Symp. “Animal Science Days”, Ptuj, Slovenia, Sept. 21st−23rd, 2016. Acta argiculturae Slovenica, Supplement 5, 55–59, Ljubljana 2016 COBISS: 1.08 Agris category code: L10 PLEIOTROPIC AND EPISTATIC INTERACTIONS BETWEEN STILLBIRTH AND CALVING EASE IN CATTLE Gábor MÉSZÁROS 1, Roland TAFERNER 2, Johann SÖLKNER 3 Pleiotropic and epistatic interactions between stillbirth and calving ease in cattle 1 University of Natural Resources and Life Sciences, Vienna (BOKU), Department of Sustainable Agricultural Systems, Division of Livestock Sciences, Gregor-Mendel- Strasse 33, 1180 Vienna, Austria, e-mail: gabor.meszaros@boku.ac.at 2 Same address as 1, e-mail: roland.taferner@students.boku.ac.at 3 Same address as 1, e-mail: johann.soelkner@boku.ac.at ABSTRACT Pleiotropic and epistatic interactions were analyzed for stillbirth and calving ease in Fleckvieh cattle, with der- egressed breeding values from 7384 bulls used as phenotypes. The genotype data consisted of 41,884 SNPs after quality control. A multivariate linear regression approach was used to obtain pleiotropic and epistatic effects using the CAPE package in R. Total of five regions on chromosomes 6, 14, 17 and 21 showed evidence of pleiotropic interaction. The genes in these significant regions were also relevant to the two phenotypes of interest, such as PLAG1 affecting growth and height, XKR4 affecting rump thickness and insulin levels, ELMOD2 connected to immune response, UCP1 con- nected to thermoregulation and UBE3A connected to the Prader-Willi syndrome. The results for epistatic effects were less convincing, although they identified connections to possible regulatory genes within chromosome 21. Key words: cattle, breeds, Fleckvieh cattle, functional traits, genetics, genetic interactions, SNP, genes 1 INTRODUCTION With the availability of dense molecular markers at a relatively cheap price we have now the possibility to analyze the genomic background of any qualitative or quantitative trait. Genome wide association studies can identify genomic regions and genes influencing a trait of interest. There are cases however, when this single gene single trait restriction is not descriptive enough. In some cases a single gene could affect multiple traits, a phenom- enon called pleiotropy, in other cases at least two genes need to interact for a single phenotype to show, also known as epistasis. Functional traits are good candidates for research on genetic interactions both between and within traits. Although absence of extremely influential genes on the phenotypes is suggested by their low heriatability, it is likely that an unseen network of interactions is present between the genes of small effect. Uncovering such re- lationships and interactions was also the motivation for our research. Therefore, the aim of this paper was to identify plei- otropic and epistatic effects for stillbirth and calving ease in cattle. 2 MATERIALS AND METHODS Single nucleotide polymorphism (SNP) data from 7384 Fleckvieh bulls genotyped with both versions of Il- lumina BovineSNP50 BeadChips and the Illumina HD chip were used in the study. For analysis only autoso- mal SNPs present on all chips were used, with mapped to the UMD3 bovine genome assembly. Standard quality control procedures were applied to remove genotyping errors using PLINK1.9 (Chang et al., 2015). In this step Acta agriculturae Slovenica, Supplement 5 – 201656 G. MÉSZÁROS et al. SNPs with more than 10 % missingness per SNP, less than 0.01 minor allele frequencies, with p-value less than 1E-9 for Hardy-Weinberg equilibrium threshold, and animals with more than 10 % missing SNPs were removed from the analysis. After all quality control steps a total of 7384 animals and 41,884 SNPs remained for analysis. Deregressed breeding values for stillbirth rate and direct calving ease with minimum reliability of 0.3 pro- vided by ZuchtData EDV-Dienstleistungen GbmH were used as phenotypes. The CAPE package (Tyler et al., 2013) in R was used for combined detection of pleiotropy and epistasis. This software maximizes the linear independence of pheno- types by performing singular value decomposition on the traits of interest, deriving so called eigentraits (ET). Commonly the two ET accounting for the most vari- ance explained i.e. with the largest contributions to the phenotypes were selected for further analysis. A genome wide single variant scan was performed to identify the 100 most influential loci for follow up analysis. To avoid effects driven purely by LD, marker pairs with Pearson correlations above 0.6 were skipped before marker selec- tion. Pair-scans were performed for each ET by multivar- iate linear regression, with an intercept, main effects and interaction term for each pair of markers. The significant main effect coefficients for variant-to-phenotype influ- ences denote pleiotropy, while interaction coefficients denote epistatic effects. The Bonferroni-Holm approach (Holm, 1979) was used to correct for multiple testing. In CAPE the epistatic effects are interpreted as the degree to which the alternate allele at the source locus changes the phenotypic effects of the alternate allele at the target locus. For example, there is a cross between two strains “A” (reference allele) and “B” (alternate allele). Let allele “B” at locus 1 have a positive main effect on the phenotype. Let there be epistasis between locus 1 and locus 2. According to CAPE when allele “B” is present at locus 2, the allele suppresses the positive effect of the “B” allele at locus 1, such that it now the allele at locus 1 has a negative effect on the phenotype. So when only locus 1 has the “B” allele, that locus has a positive main effect on the phenotype, but when locus 2 also has a “B” allele, the “B” allele at locus 1 is changed through epistasis to have a negative effect on the phenotype. According to this approach to epistasis the direction of such effect can be derived. The coefficient shows the extent of enhancement or suppression of the phenotypic effect of the target locus, given the presence of the alter- nate allele at the source locus. The derived model fits all phenotypes. So if locus 2 suppresses the effect of locus 1, it will suppress its effect across all phenotypes for which locus 1 has a main effect. 3 RESULTS AND DISCUSSION The correlations between calving ease and stillbirth deregressed breeding values used as phenotypes were in the range of 0.8, which is the upper boundary for analy- sis of pleiotropic effects. Similarly high correlations from 0.78 to 0.84 between the two calving traits were found by (Steinbock et al., 2003). Yet it has been reported that only 43 % of the stillbirths were caused by difficult calving and 32 % were born without any signs of difficulties during the birth process (Berglund et al., 2003). This emphasizes that calving difficulties were not the only rea- son for stillbirth, thus supporting our efforts to find epistatic regions influencing both traits. Five genomic regions on four chromo- somes showed significant pleiotropic effects for the stillbirth and calving ease trait pair, with their center around 4.3 Mb on BTA6, 47.0 Mb on BTA6, 17.5 Mb on BTA17, and 2.1 Mb on BTA21 and a wider region around 23–24  Mb on BTA14 (Fig. 1). QTL for direct still birth on BTA6 with a pleiotropic effect on the trait combination still- birth and calving ease was reported by (Kühn et al., 2003). These results are supported by (Schrooten et al., 2000), who found indication for a QTL for the trait calving ease at the same chromosomal region. Furthermore, Schroot- en et al. (2000) detected a QTL for size of the animals and the dairy character in this region. Both of these traits might influence the delivery Figure 1: Network graph showing the combination of the traits stillbirth and calving ease Acta agriculturae Slovenica, Supplement 5 – 2016 57 PLEIOTROPIC AND EPISTATIC INTERACTIONS BETWEEN STILLBIRTH AND CALVING EASE IN CATTLE of the calf. A QTL for birth weight was also identified in the same chromosomal region of BTA6 by (Casas et al., 2000). Similarly to our results, the genomic regions on BTA 14 and BTA 21 were also highlighted in a genome wide association study of calving ease and growth related traits in Fleckvieh cattle (Pausch et al., 2011). The region on BTA14 was associated with growth related traits, influ- encing the body size of the calf and eventually the calving ease. The region at BTA21 contained genes whose lack of expression leads to Prader–Willi syndrome. One of the sympthomatic effects of this disorder are fetal growth anomalies, which influence the stillbirth rate. The presence and function of genes was also inves- tigated within the ± 0.5 Mb region from the pleiotropic SNP. There were several genes in each identified region, although their connectedness to stillbirth or calving ease was not always clear. The most important genes were PLAG1 on BTA14 affecting growth and height (Utsu- nomiya et al., 2013), XKR4 on BTA14 affecting rump thickness and insulin levels (Bolormaa et al., 2011), ELMOD2 on BTA17 connected to immune response (Thompson-Crispi et al., 2014), UCP1 on BTA17 con- nected to thermoregulation (Sonstegard and Kappes, 1999) and UBE3A on BTA21 connected to the Prader- Willi syndrome (Horsthemke and Wagstaff, 2008). Table 1 shows the epistatic interactions between genomic regions as interpreted by the CAPE software. They seem to be centered on BTA21 for most SNP pairs. Although all of these SNP pairs were shown to have a highly significant interaction, but the standard errors of effect sizes were also very high (results not shown). For this reason caution has to be exercised when interpreting the results. The 2.1–2.5 Mb region of BTA 21 contains a large number of uncharacterized, possibly protein coding genes. The two known genes in the region are UBE3A and ATP10A connected to Angelman and Prader-Willi syndromes. The epistatic interactions from these re- gions are pointing to different directions, such as BTA6 (4.22  Mb), BTA14 (23.48  Mb), BTA17 (64.84  Mb), BTA21 (36.62 Mb) and BTA21 (44.72 Mb). The region in BTA6 contains no obvious candidates, although the NDNF might be connected to craniofacial development (Melvin et al., 2013). The 23–24 Mb region of BTA14 contains growth and size related genes as men- tioned above. The identified region on BTA17 is extreme- ly rich in genes, thus we have no clear indication about the true target. The 36.62Mb region of BTA21 contains multiple uncharacterized genes and the MIR2888-1, a non-coding RNA gene with regulatory function. It was shown to be expressed in bovine ovaries (Hossain et al., 2009). Inter- estingly the 44.72  Mb region of BTA21 also contains a similar regulatory gene MIR6522, which was also found to be expressed in testicular and ovarian tissues of cat- tle (Huang et al., 2011). Although there is no clear con- nection of gonadal functions to either of our traits, the authors suggest that it might be involved in different bio- logical pathways, which encourages further research. An additional epistatic interaction was identified between the 24 Mb region of BTA14 and 2.93 Mb region of BTA29. The only gene close to the identified SNP on BTA29 is FAT3, that was shown to participate in mor- phogenetic processes of animal development (Nagae et al., 2007). Figure 2 shows an example of a previously identified epistatic interaction from Table 1. The alternate allele in Source SNP2 BTA Source Position Source 1 Target SNP 2 BTA Target Position Target 1 1 BTB.01281052 6 4.22 ARS.BFGL.NGS.53975 21 2.15 2 BTB.01281052 6 4.22 ARS.BFGL.NGS.108925 21 2.46 3 BTB.00557532 14 24.64 BTB.00998633 29 2.93 4 ARS.BFGL.NGS.53975 21 2.15 Hapmap39377.BTA.113470 21 36.62 5 ARS.BFGL.NGS.53975 21 2.15 BTB.00820789 21 44.72 6 ARS.BFGL.NGS.108925 21 2.46 BTB.01281052 6 4.22 7 ARS.BFGL.NGS.108925 21 2.46 ARS.BFGL.BAC.36626 17 64.84 8 ARS.BFGL.NGS.108925 21 2.46 UA.IFASA.7382 14 23.48 9 Hapmap39377.BTA.113470 21 36.62 ARS.BFGL.NGS.53975 21 2.15 10 BTB.00820789 21 44.72 ARS.BFGL.NGS.53975 21 2.15 11 BTB.00820789 21 44.72 ARS.BFGL.NGS.108925 21 2.46 Table 1: Epistatic interactions for stillbirth and calving ease 1 Positions shown in megabases; 2 All significant interactions are shown, one in each direction in case both of them were significant. Acta agriculturae Slovenica, Supplement 5 – 201658 G. MÉSZÁROS et al. locus 2 (UA.IFASA.7382, BTA14) decreases the pheno- type value if the reference allele is homozygous in locus 1 (ARS.BFGL.NGS.108925, BTA21), but seems to have the opposite effect when locus 1 is homozygous for its alternate allele. 4 CONCLUSIONS Pleiotropic and epistatic regions influencing still- birth and calving ease were identified in our study. Pleiotropic effects were shown on BTA6, 14, 17 and 21, containing genes related to body size, growth, immune response and thermoregulation. The epistatic interac- tions were also found on BTA29 in addition to the chro- mosomes involved in pleiotropy. Two of these pleiotropic interactions within BTA21 were involving regulatory genes involved in molecular pathways of gonadal func- tions, which also merits further research. 5 ACKNOWLEDGEMENTS The authors would express their gratitude to Dr. Anna Tyler for her support and helpful comments with the interpretation of CAPE results. The SNP chip and phenotype data were kindly pro- vided by AGÖF (Arbeitsgemeinschaft österreichischer Fleckviehzüchter) and ASR (Arbeitsgemeinschaft süd- deutscher Rinderzüchter), LfL (Bayerische Landesanstalt für Landwirtschaft) and ZuchtData EDV-Dienstleistun- gen GbmH. The provision of data without any costs is gratefully acknowledged. 6 REFERENCES Berglund, B., Steinbock, L., Elvander, M. (2003). Causes of still- birth and time of death in Swedish Holstein calves exam- ined post mortem. Acta Veterinaria Scandinavica, 44(3–4), 111–120. Bolormaa, S., Neto, L. R. P., Zhang, Y. D., Bunch, R. J., Harrison, B. E., Goddard, M. E., Barendse, W. (2011). A genome-wide association study of meat and carcass traits in Australian cattle. Journal of Animal Science, 89(8), 2297–2309. Re- trieved from http://doi.org/10.2527/jas.2010-3138. Casas, E., Shackelford, S. D., Keele, J. W., Stone, R. T., Kappes, S. M., Koohmaraie, M. (2000). Quantitative trait loci affecting growth and carcass composition of cattle segregating alter- nate forms of myostatin. Journal of Animal Science, 78(3), 560–569. Chang, C. C., Chow, C. C., Tellier, L. C., Vattikuti, S., Purcell, S. M., Lee, J. J. (2015). Second-generation PLINK: rising to the challenge of larger and richer datasets. GigaScience, 4, 7. Retrieved from http://doi.org/10.1186/s13742-015-0047-8. Figure 2: Deregressed breeding values (dEBV) for stillbirth and calving ease according to genotype combinations (0 – homozygous for the reference allele, 1 – heterozygous, 2 – homozygous for the alternate allele). The values on the X axis denote genotype combinations for SNPs ARS.BFGL.NGS.108925 (BTA21, before dash) and UA.IFASA.7382 (BTA14, after dash) Acta agriculturae Slovenica, Supplement 5 – 2016 59 PLEIOTROPIC AND EPISTATIC INTERACTIONS BETWEEN STILLBIRTH AND CALVING EASE IN CATTLE Holm, S. (1979). A simple sequentially rejective multiple test procedure. Scandinavian Journal of Statistics, 6, 65–70. Horsthemke, B., Wagstaff, J. (2008). Mechanisms of imprinting of the Prader-Willi/Angelman region. American Journal of Medical Genetics. Part A, 146A(16), 2041–2052. Retrieved from http://doi.org/10.1002/ajmg.a.32364. Hossain, M. M., Ghanem, N., Hoelker, M., Rings, F., Phatsara, C., Tholen, E., Schellander K., et al. (2009). Identification and characterization of miRNAs expressed in the bovine ovary. BMC Genomics, 10, 443. Retrieved from http://doi. org/10.1186/1471-2164-10-443. Huang, J., Ju, Z., Li, Q., Hou, Q., Wang, C., Li, J., … Zhong, J. First 7 authors then et al.(2011). Solexa sequencing of novel and differentially expressed microRNAs in testicular and ovarian tissues in Holstein cattle. International Journal of Biological Sciences, 7(7), 1016–1026. Kühn, C., Bennewitz, J., Reinsch, N., Xu, N., Thomsen, H., Looft, C., … Kalm, E. Some comment as above. (2003). Quantitative trait loci mapping of functional traits in the German Holstein cattle population. Journal of Dairy Sci- ence, 86(1), 360–368. Melvin, V. S., Feng, W., Hernandez-Lagunas, L., Artinger, K. B., Williams, T. (2013). A morpholino-based screen to identify novel genes involved in craniofacial morphogenesis. Devel- opmental Dynamics: An Official Publication of the American Association of Anatomists, 242(7), 817–831. Retrieved from http://doi.org/10.1002/dvdy.23969. Nagae, S., Tanoue, T., Takeichi, M. (2007). Temporal and spa- tial expression profiles of the Fat3 protein, a giant cadherin molecule, during mouse development. Developmental Dy- namics: An Official Publication of the American Association of Anatomists, 236(2), 534–543. Retrieved from http://doi. org/10.1002/dvdy.21030. Pausch, H., Flisikowski, K., Jung, S., Emmerling, R., Edel, C., Götz, K.-U., Fries, R. (2011). Genome-Wide Association Study Identifies Two Major Loci Affecting Calving Ease and Growth-Related Traits in Cattle. Genetics, 187(1), 289–297. Retrieved from http://doi.org/10.1534/genet- ics.110.124057. Schrooten, C., Bovenhuis, H., Coppieters, W., Van Arendonk, J. A. M. (2000). Whole Genome Scan to Detect Quantitative Trait Loci for Conformation and Functional Traits in Dairy Cattle. Journal of Dairy Science, 83(4), 795–806. Retrieved from http://doi.org/10.3168/jds.S0022-0302(00)74942-3. Sonstegard, T. S., Kappes, S. M. (1999). Mapping of the UCP1 locus to bovine chromosome 17. Animal Genetics, 30(6), 472. Steinbock, L., Näsholm, A., Berglund, B., Johansson, K., Philipsson, J. (2003). Genetic effects on stillbirth and calv- ing difficulty in Swedish Holsteins at first and second calv- ing. Journal of Dairy Science, 86(6), 2228–2235. Retrieved from http://doi.org/10.3168/jds.S0022-0302(03)73813-2. Thompson-Crispi, K. A., Sargolzaei, M., Ventura, R., Abo-Is- mail, M., Miglior, F., Schenkel, F., Mallard, B. A. (2014). A genome-wide association study of immune response traits in Canadian Holstein cattle. BMC Genomics, 15, 559. Re- trieved from http://doi.org/10.1186/1471-2164-15-559. Tyler, A. L., Lu, W., Hendrick, J. J., Philip, V. M., Carter, G. W. (2013). CAPE: An R Package for Combined Analysis of Pleiotropy and Epistasis. PLOS Computational Biology, 9(10), e1003270. Retrieved from http://doi.org/10.1371/ journal.pcbi.1003270. Utsunomiya, Y. T., Carmo, A. S. do, Carvalheiro, R., Neves, H. H., Matos, M. C., Zavarez, L. B., … Garcia, J. F. See comment above (2013). Genome-wide association study for birth weight in Nellore cattle points to previously de- scribed orthologous genes affecting human and bovine height. BMC Genetics, 14(1), 52. Retrieved from http://doi. org/10.1186/1471-2156-14-52.