Ethical statement

All procedures were in compliance with the Guide for the Care and Use of Agricultural Animals in Research and Teaching 2010 and approved by the Texas A&M University Animal Care and Use Committee AUP 2002–315.

Animals and collected samples

One hundred and eighty-nine animals were selected from a population of 1,370 Brahman calves (628 male and 742 female) born between 2002 and 2017 at the Texas A&M AgriLife Research and Extension Center in Overton, TX. The management of the Brahman cattle resource herd and data recording were previously reported by [6]. All temperament data:

• Exit Velocity (EV), an objective measure, the rate at which the animals exited the working chute and traveled a distance of 1.83m, measured with an infrared sensor (FarmTek Inc., North Wylie, TX, USA) [10, 11].
• Pen Score (PS), a subjective measure, based on individual visual assessments of animal behavior while confined to a pen in groups of five animals, where a score of 1 is calm and 5 aggressive [12]; and,
• Temperament Score (TS) were calculated by averaging the PS and EV [TS = (PS + EV)/2] [13], determined at the time that the calves were weaned from their dams.

Ear notches obtained at the time of weaning were snap frozen and were stored at -80°C until used to determine genotypes.

In this study a selective genotyping approach was applied, this strategy has been used in the analysis of other traits in cattle to facilitate the detection of causative alleles, due to an enrichment of these alleles among phenotypically extreme individuals [14]. Unlike other temperament evaluations, the EV and PS have been shown to be positively correlated (r ≥ 0.35, P <0.005) (11). On the basis of EV heritability calculated in the Brahman population used in this work (h² = 0.27) [6] and considering EV as an objective measure that offers higher reliability between evaluators than subjective methods (PS), the selective genotyping was achieved following the strategy of tailed EV deviations; the animals were classified into calm and temperamental, based on the sex-group average values (EV) ± 1/2 standard deviation (SD). Of the total population, we identified two groups of temperament-contrasting animals, the calmest and most temperamental. Females were calm if their EV was within the range of 0.16–1.82 m/s (n = 50) and males were calm if their EV was within the range of 0.4–1.56 m/s (n = 48). Females were temperamental if their EV was within the range of 3.13–7.66 m/s (n = 46) and males were temperamental if their EV was within the range of 3.05–10.83 m/s (n = 45).

Genotyping and GWAs

The DNA extraction from ear notch samples and genotyping of the 189 animals were done using the GeneSeek Genomic Profiler HD 150K chip (GGP-HD-150K) (Neogen, Lincoln, NE) by the NEOGEN company. The GWAS analyses were performed with PLINK 1.9 software [15] using the case-control method for unrelated animals, considering calm and temperamental animals as case and control, respectively. The algorithm considers a single marker association analyses through frequency comparison using a chi-square test. As part of quality control, the following SNP were excluded: a) those with an unknown genomic position; b) those located on a sex chromosome; c) monomorphic SNPs; d) SNPs with minor allele frequency (MAF) below 0.01, and, a call rate below 90%. Individuals with more than 10% missing genotypes were also excluded. After quality control, 172 genotyped animals [85 males and 87 females; 89 calm (46% males; 54% females) and 83 temperamental (53% males; 47% females)] were used. SNPs with P-values < 8 × 10−5 were considered significant and associated with EV, and therefore to bovine temperament. From the coordinates provided by the chip information, the physical position of significant SNPs was obtained based on the last available Bos taurus genome (ARS-UCD 1.2) using the Genome Data Viewer software available at the National Center for Biotechnology Information (NCBI). Candidate genes were selected based on their location in relation with the significant SNPs, through the genome navigator. All genes around 300Kb downstream and upstream, were selected and analysed according to literature reports.

Effects of significant SNPs with temperament traits

To assess the effect of significant SNPs on EV, PS and TS, a mixed model was fitted including the fixed effect of sex of calf, individual effect of fourteen loci (rs135340276, rs134895560, rs110190635, rs42949831, rs135982573, rs109393235, rs109531929, rs135087545, rs41839733, rs42486577, rs136661522, rs110882543, rs110864071, and rs109722627). Year of birth was considered as random effect, to capture and remove actual trend effects of this factor on response variables and avoid the use of unnecessary degrees of freedom. The model was fitted using the MIXED procedure. Genotypes with frequency lower than 0.05 were excluded from the analysis. Least square means of the genotypes were estimated and compared by a t-test with a Bonferroni adjustment using the PDIFF statement. All statistical analyses were performed using SAS 9.0 (SAS Inst. Inc., Cary, NC, USA).

Association of significant SNPs

The association of the fourteen SNPs on three temperament traits, i.e., EV, PS and TS was determined. As shown in Table 3, all of the SNPs evaluated were associated with EV (P<0.01) only rs42486577 was associated with PS. The rs42949831, rs135982573, rs41839733 and rs110882543, were associated with TS. The most significant association with EV was shown for rs135087545, rs110864071 and rs109722627, located in the SLC9A4. The three SNPs are in LD (r² = 0.266), and, it should be noted that the SNP in the EXOC4 gene had the greater significant difference between genotypic means for EV, with the carriers of GG genotype having EV values 2.64 m/s greater than the AA genotype (Table 3), while the homozygote GG genotype exhibited a TS that was 1.57 SD greater than those of the AA genotypes.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Least square means of genotype effects of SNPs associated with Brahman temperament traits using a mixed model fitted.

https://doi.org/10.1371/journal.pone.0237825.t003

Significant intergenic SNPs

The mapping of the regions around the intergenic SNPs identified in the GWAS allowed identification of seventeen candidate genes. Some of them such as ACER3 (BTA 15), VRK2 (BTA 11), FANCL (BTA 11), and SLCO3A1 (BTA 21) could be considered to be candidate genes, due their reported biological functions. These genes were located on BTA that according to the Cattle Quantitative Trait Locus (QTL) Database (https://www.animalgenome.org/cgi-bin/QTLdb) [17], harbour QTLs related to bovine temperament: a) BTA 11 (Social separation—Standing alert; tendency of an animal to stand alert upon separation from its pen mates); b), BTA 15 (Temperament; milking temperament, pen score and reactivity).

The ACER3 gene catalyzes the hydrolysis of C18:1, the major unsaturated long-chain ceramide in the brain to sphingosine, its phosphorylated form sphingosine-1-phosphate (S1P), which has been implicated in the survival of neurons, and its dyshomeostasis has been associated with different neurodegenerative disorders [18]. It has been demonstrated in mice that ACER3 plays an important protective role, since it controls the homeostasis of ceramides and their derivatives such as S1P, avoiding the appearance of neurological disorders such as cerebellar ataxia [18]. Little is known about the physiological function of the VRK2 gene. It is a serine-threonine which is potentially involved in neural proliferation and migration due to its interaction with multiple biological pathways (i.e., neurite initiation and axon outgrowth) [19]. The lack of VRK2 interferes with synaptic functions, knockout mice for this gene show changes in their social behavior [20]. In genome wide association studies in humans, VRK2 has been consistently associated with psychiatric and neurodegenerative disorders such as schizophrenia, major depressive disorder and genetic generalized epilepsy. The rs2312147 in the intron of this gene has been repeatedly associated with schizophrenia in large human populations of European and Asian individuals [19].

The FANCL gene codes for an ubiquitin ligase, which is found like VRK2 in a region associated with schizophrenia, due to the SNPs rs11682175 and rs75575209 being found in LD, they were identified by the Schizophrenia Working Group of the Psychiatric Genomics Consortium when studying 36,989 cases and 113,075 controls [21]. In the same way in this work we identified the candidate genes VRK2 and FANCL from SNPs rs134895560 and rs110882543 (r² = 0.82). Some studies have corroborated the relationship of genes identified through studies of GWAS with schizophrenia; such is the case of FANCL which has been identified by 2 studies of TWAS (Transcriptome-wide association study) as a gene with transcriptome-wide significant association from different samples of patients with schizophrenia [22, 23]. The SLCO3A1 gene is abundantly expressed in the postnatal brain of mice [24]. This gene participates in networks of genes involved in neurological and developmental disorders [25]. In humans through GWAS, one SNP has been identified within this gene, associated with Parkinson's disease [26]. Interestingly, here we identified two SNPs located at the SLC family at SLC18A2, which were associated with temperament traits in the studied population. Garza-Brenner et al. [27] also identified a SNP in this gene with an effect on Pen Score in Charolais cattle.

Intronic temperament-associated SNPs

Five SNPs located at introns of genes NRXN3, EXOC4, CACNG4 and SLC9A4 in Brahman cattle were identified. Although none of these genes have been previously identified with influence on bovine temperament, we could hypothesize their role on the basis of the reported genomic studies where they have been identified. In humans, these genes have been associated with different neurodegenerative disorders (i.e., Alzheimer and Parkinson’s disease) and neuropsychiatric disorders (i.e. autism and schizophrenia); for most of them, genetic variations and polymorphisms have been reported in association with a particular expression of each these diseases. Considering that the aforementioned disorders have as a common denominator, aspects related to the expression of behaviors as a response to stress and fear, and that temperament is a reflection of an animal´s stress response, innate fear or reaction to handling, as well as animal´s aggressiveness [6, 28], all of the genomic information about the genes and pathways previously related in humans could allow us to see patterns and consider the reported pathway interactions as candidates to also explain bovine temperament. Thus for example, the rs135982573 marker identified with significant effect for EV, is located at intron one of the NRXN3 genes; neurexins (NRXNs) act predominantly at the presynaptic terminal in neurons and play essential roles in neurotransmission and differentiation of synapses [29]. In humans, genetic variations in neuroxins have been associated with disorders affecting cognition and behavior via molecular mechanisms including cell adhesion, synaptic transmission and neurotransmitter secretion, [30]. In a bovine population, Qanbari et al. [31] found candidate selection regions harboring genes related to phenotypes relevant to domestication, such as neurobehavioral functioning. NRXN1 has been proposed as a candidate gene for behavioral traits such as a moderate temperament. Although the molecular basis of the effect of NRXN3 with temperament in cattle has not been directly determined, according to the STRING database (https://string-db.org/), this gene has interactions with the AMPA2 (GRIA2) gene, supported by text mining and co-expression. Lindholm-Perry et al. [32] reported that a SNP near GRIA2 was nominally associated with flight speed, a predictive indicator of bovine temperament. This interaction can be supported by the fact that both families of genes Neurexin and AMPA receptors are sorted by SorCS1 receptor, regulating neurexin and AMPAR surface trafficking [33]. Further studies are needed to investigate the effect of these interacting genes and their genetic variation influencing bovine temperament.

Association of the SNP rs109393235 located at intron 7 of the EXOC4 gene, was also found. This gene has been implicated in insulin processing and protein metabolism [34]. It has been associated with reproductive traits (age at first calving) in Canchim beef cattle using a GWAS approach [35]. Interesting, in humans, intronic polymorphisms and other genetic variations have been associated neurological disorders such as Alzheimer's, dementia, schizophrenia and personality disorders [36].

The rs136661522 is located at intron 1 of the CACNG4 gene. The CACNG4 gene in cattle has been associated with milk yield traits in Polish Holstein dairy cattle [37] and with sperm motility in cattle, humans and mice [38]. The Voltage-gated L-type calcium channels (VLCC) play an important role both in the nervous system (intracellular flow of Ca²⁺ by action potentials in the synapse) and in the cardiovascular system (regulates excitation-contraction coupling in the heart), because the CACNG1 to CACNG8 genes have been identified as regulators of VLCC function (activation and inactivation), it has been shown that CACNG4 interacts physically with the cardiac voltage-gated Ca²⁺ channel, regulating its function. However; its implications are yet unknown [39]. This is relevant because bovine temperament affects the activation of cardiovascular system. Because this gene is expressed in the human brain, it has been identified as a candidate gene to be involved in the risk of susceptibility to schizophrenia, since SNP near this gene showed a significant effect in a human population [40].

The SLC9A4 gene, which harbors 2 SNPs in introns 2 and 8, respectively, is part of the SLC family, which, as previously mentioned, transports various molecules, such as inorganic ions, fatty acids, saccharides and neurotransmitters [41]. SLC9A4 corresponds to a group of Na⁺/H⁺ exchangers, which are expressed in the stomach, small intestine, colon, skeletal muscle and brain and it has been associated with metabolic diseases [42]. However, given that in mice it is mostly expressed in the hippocampus of the brain, it could be expected that in some way, it is involved in the expression of temperament [42]. Some genes of the SCL9A family (SLC9A1) have been associated with neurological disorders, such as epilepsy and bovine genes of the SLC family, such as SLC18A2 and SLC6A4, have been identified as candidates associated with bovine temperament, in Charolais populations, SLC18A2 has been shown to have a significant effect on Pen Score [27].

At present, the GWAS approach is considered to be a viable strategy to identify disease- and trait-associated genetic variants, interestingly most of the variants discovered (~ 93%) using GWAS, are located in non-coding regions (i.e., introns, promoters, intergenic regions) [43]. In humans, for example, it has been reported that of a total of 920 GWAS studies for different diseases and traits (which included a large number of neurological and behavioural projects), only 4.9% represent genetic variants located in exons, 52.6% in intergenic regions (1Mb of nearest transcriptional start site) and 41% in intronic regions [44]. We used the contrasting group strategy and were able to identify 5 and 9 SNPs in the intronic and intergenic regions associated with EV, respectively. Further investigation is needed to identify candidate genes close to this genomic region, because until now no genes were mapped in this region.

An interesting result was the identification of associated SNP in introns of genes. It is known that the primary sequence of a gene is not limited to expression of a protein, but it also implies elements located outside the protein-coding regions, such as introns, since these may have both direct effects such as alternative splicing, enhanced gene expression or indirect effects, source of new genes, or may harbour several kinds of noncoding functional RNA genes [43]. In humans, neurological disorders such as schizophrenia are under the control of variants that affect gene expression, rather than variants that affect protein structure [21]. Similarly, for this same neurological disorder, it has been identified that there are relationships with intronic SNPs that upregulate the expression of splice variants in the human brain [45].

Further investigation will allow definition of the role of the five intronic SNPs identified in this study; they are potential candidates to address control of bovine temperament, as has been demonstrated for some neurological disorders in humans.