Genome-wide association studies for conformation traits in the Turkish Holstein cattle population

Abstract

Background:

Conformation traits (CNTs) are part of the selection goals that significantly affect cow economic efficiency, health, welfare, and productive life in the dairy industry.

Aims:

This study focused on a genome-wide association study (GWAS) and genetic parameters estimation for 21 CNTs, including udder, leg-foot, body, type, and final classification traits in the Turkish Holstein (THol) dairy cattle population.

Methods:

A restricted maximum likelihood with a univariate model including the fixed effects of herd-year-season and days in milk was used. The total dataset consisted of CNTs records and Affymetrix BovineSNP54K data for 3,008 THol cows that calved from 2019 to 2022. The gene ontology and Kyoto Encyclopedia of genes and genome pathway databases were used to assign genes to functional categories. The biological pathways were performed in BioMart databases.

Results:

The heritability of these 21 CNTs ranged from 0.01 (udder index) to 0.133 (udder depth). A total of 16 significant single nucleotide polymorphisms (SNP) associated with 13 CNTs was identified. Significant SNP overlap in the candidate genes, which include ITGB1, TNN, and SEMA3D, have potential for researchers and breeders for CNTs in cattle breeding.

Conclusion:

These results provide valuable knowledge and contribute to the elucidation of the genetic factors responsible for conformation traits in dairy cattle populations.

Introduction

Conformation traits (CNTs), including udder, feet and legs, body, rump framework, and milk production traits, are used to select animals for higher production and longevity in dairy cows (Miglior et al., 2017). Selection for linear type traits can affect cow welfare, health, production, economic efficiency, reproduction, and longevity extension (Miglior et al., 2017; Gutiérrez-Reinoso et al., 2023). A selection index that combines CNTs based on their economic importance is applied in several countries (VanRaden, 2004; Miglior et al., 2017). A descriptive classification program that recorded these CNTs in the Turkish Holstein (THol) population became part of the selection objectives. Genomic selection in the THol population has been carried out since 2017 and implemented in applied breeding since 2023. A reference population consisting mainly of cows has been established and continuously updated. In the genomic evaluation of the THol population, milk performance, reproductive, and CNTs are mainly considered.

CNTs are closely associated with milk yield, reproduction, lameness, mastitis, and longevity in dairy cows (Long et al., 2024), and could be useful as phenotypes, since they help improve economic efficiency in cattle (Nazar et al., 2022). Heritability values based on CNTs records and pedigree kinship data reported low 0.10 (rear legs set-side view) to medium values of 0.32 (stature) in the Serbian Holstein by Djedovic et al. (2023), and from low (0.09) for udder texture and foot angle to high (0.38) for teat length in Holstein cows in Brazil by Campos et al. (2015). The genomic heritability of udder traits was reported at 0.04 to 0.49 (Nazar et al., 2022).

Genome-wide association studies (GWAS) have proven to be an efficient tool to identify markers for single nucleotide polymorphisms (SNP) in the genome that are associated with phenotypes (Erdoğan et al., 2024). The determination of quantitative trait loci (QTL) is a crucial process for understanding the genetic variations associated with the selected traits. Ashwell et al. (2005) identified QTLs affecting 22 CNTs and found 41 significant QTLs. Schrooten et al. (2004) performed a whole genome scan using microsatellite markers for QTLs that affect 18 CNTs. As a result, numerous GWAS results have been documented using SNP markers that specifically target CNTs and provide comprehensive insights into genetic improvement through gene-based selection (Cole et al., 2011).

In this study, GWAS was performed to investigate 21 CNTs within the contemporary first parity THol population. The SNP panel (Affymetrix-54K) was used to discover SNP markers and genes associated with CNTs useful for the THol cattle population.

Materials and Methods

Ethics statement

The whole process of blood sampling and data recording was completed according to the program of the Ministry of Agriculture and Forestry (MAF), and the Turkish Council on Animal Care. THol population genomic selection for higher milk production, milk fat, milk protein, conformation traits, and reproductive performance is conducted under the “National Cattle Genetic Selection Program”. This study was approved by the Institutional Animal Care and Ethics Committee of the International Center for Livestock Research and Training (approval No.: 2021-196).

Data collection

The data of the THol population in this study included 3,008 first parity Holstein cows. All cows came from 49 dairy farms in the population. The CNTs were collected by experts from the Cattle Breeder Association of Turkey (CBAT) as part of the Holstein Genomic Selection Program. From 2019 to 2022, the phenotypic measurement of CNTs was completed for each cow. According to the linear classification system, 16 linear traits were scored from 1 to 9, and 5 composite traits were measured with an index ranging from 65 to 90. The 16 traits were fore udder attachment (FUA), udder support (US), udder depth (UD), front teat placement (FTP), front teat length (FTL), rear teat placement (FTP), rear udder height (RUH), hock quality (HQ), rear leg view (RLRW), rear leg angle (RLA), heel depth (HD), body depth (BD), chest width (CW), rump width (RW), and rump angle (RA). The five traits of the conformation index were udder (UD), leg-foot (LF), body (BO), type (TY), and general (GN). The scores for the five composite traits were calculated based on linear point value weights. Description of the linear-type traits and composite index traits were given in Table 1.

Table 1.

Description of the linear-type traits and composite index traits

Traits	Abrv.	Definitions	Score2 1/65	Score 9/90	Ideal score
Udder	UI 2	Udder index	Poor	Excellent	90
Fore udder attachment	FUA3	Angle formed between the fore udder and abdomen, side view	Weak	Strong	9
Udder support	US3	Depth of cleft at base of rear udder	Weak	Strong	9
Udder depth	UD3	Depth of udder from hock to udder floor	Deep	Shallow	5-6
Front teat placement	FTP3	The position of the front teat from center of quarter	Wide	Close	5
Front teat length	FTL3	Length of the front teats from side view	Short	Long	5
Reat teat placement	RTP3	Teat placement from center of quarter	Wide	Close	5-6
Rear udder height	RUH3	The distance between the vulva and the milk secreting tissue	Low	High	9
Dairy character	DC3	Angle of cidago bones	Coarse	Thin	9
Leg-Foot	LF 2	Linear leg and foot index	Poor	Excellent	90
Hock quality	HQ3	Quality of the hocks	Coarse	Flat	5-6
Rear leg rear view	RLRV3	Turn of hock when viewed from rear	Hocked-in	Straight	9
Rear leg angle	RLA3	Rear leg side view (rear leg hock angle)	Steep	Low	5
Heel depth	HD3	Depth of heel on outside claw	Shallow	Deep	7-8
Body	BO 2	Linear body index	Poor	Excellent	90
Body depth	BD3	Depth of body at the rear rib	Shallow	Deep	7
Chest width	CW3	Width of chest (narrow to wide)	Narrow	Wide	9
Rump width	RW3	Distance between inside pins	Narrow	Wide	8-9
Rump angle	RA3	Height of the pin bones relative to height of hook bones	High	Low	5-6
Type	TY 2	Body harmony	Poor	Excellent	90
General	GN 2	Total index: TY × 0.15 + BO × 0.20 + LF × 0.25 + UI × 0.40	Poor	Excellent	90

1: UI: Udder index, FUA: Fore udder attachment, US: Udder support, UD: Udder depth, FTP: Front teat placement, FTL: Front teat length, RTP: Rear teat placement, RUH: Rear udder height, DC: Dairy character, LF: Leg foot, HQ: Hock quality, RLRV: Rear leg rear view, RLA: Rear leg angle, HD: Heel depth, BO: Body, BD: Body depth, CW: Chest width, RW: Rump width, RA: Rump angle, TY: Type, and GN: General. 2: Traits calculated 65-90, and 3: Traits measured in scores 1-9

Genotyping data and quality control

Blood samples from the tail vein of 3008 cattle were collected for DNA extraction and genotyping. Genotyping (using the Axiom Bovine 54K SNP, Affymetrix) was performed in the Laboratory of the International Center for Livestock Research and Training (ICLRT) in Ankara, Turkey.

Quality control (QC) was carried out with the code Markers function of the statgen GWAS package in the R program (R core team, 2021). SNPs were not included in the analysis if the call rate was less than 90%, a call rate for animal genotypes of less than 90%, a minor allele frequency (MAF) of less than 1%, and an extreme deviation from Hardy-Weinberg equilibrium (HWE; 1E-13). In addition, single nucleotide polymorphisms (SNPs) mapped on sex-linked and mitochondrial DNA chromosomes, as well as SNPs with unclear positions (SNPs without information on chromosomal position), were excluded from further analysis.

Following the QC phase (stage), SNP imputation assignment was performed for the missing SNP genotypes. There are many algorithms among these imputation methods, and the imputation was performed randomly in the R program. After SNP imputation, the criterion of an MAF of less than 1% of loci with SNPs was also checked and removed. The HWE was then re-evaluated to identify genotyping errors. Loci with probability values below 1E-13 were excluded based on the χ-squared test. After all these evaluations, 31,944 SNPs belonging to 3,008 genotyped cows were selected for further analysis.

Genome-wide association analysis

The association analysis was performed with mixed model equations in an R environment statgen GWAS package with EMMA-based REML (Kang et al., 2008; Kang et al., 2010). SNP effects were estimated using the following statistical model:

 $\mathrm{y}=1\mathrm{\mu }+\mathrm{Za}+\mathrm{e }$Where,

y: A vector of phenotype adjusted for known environmental factors (heard-year-season of calving, days in milk)

1: A vector of ones

µ: The overall mean

Z: An incidence matrix, which holds the genotypes

a: A vector of random polygenic effects and assumed as normally distributed [a ~ N (0, G  ${\mathrm{\sigma }}_{\mathrm{a}}^2$), G is the genomic relationship matrix (GRM) (VanRaden, 2008) built from SNP genotypes,   is the additive genetic variance]

e: The vector of the random residual effect

To identify the significance of the SNPs, the p-values were calculated and transformed into a -log scale to create a Manhattan plot. The Manhattan plot visually represents the association between SNP markers and the observations to identify potential genomic regions associated with additive variation in the traits. The significance level and effect of each SNP were calculated with GLS solutions (Segura et al., 2012). The additive variance explanation rate (A-VER) or SNP heritability with low significance for each variant was calculated as follows (Kimura et al., 1970):

$\frac{\mathrm{var}\left(\mathrm{SNP}\right)}{\mathrm{var}\left(\mathrm{phenotype}\right)}=\frac{2{\mathrm{pq\beta }}^2}{\mathrm{var}\left(\mathrm{phenotype}\right)}$

Where,

p: The major allele frequency of each SNP

q: The minor allele frequency of each SNP

β: SNP effect to a related trait

The descriptive statistics of CNT were generated using SPSS for data analysis (mean, standard error, minimum and maximum values, standard deviation, and coefficient of variation in %).

Identification of SNP locations and gene annotation

Using the NCBI database (Genome Data Viewer, NCBI, www.ncbi.nlm.nih.gov), the SNP locations were compared with their positions in the UMD 3.1.1 bovine reference genome. The genes and QTLs closest to the significant SNPs were reported by the National Animal Genome Research Program (Animal QTLdb of Cattle QTLdb, www.animalgenome.org) and the National Center for Biotechnology Information (www.ncbi.nlm. nih.gov).

Enrichment of functional pathways and gene network analysis of candidate genes

In this study, to better understand the biological information between and among candidate genes, all candidate genes from the GWAS analysis were entered into Gene Ontology (GO), and a KEGG (Kyoto Encyclopedia of Genes and Genomes) analysis was performed using ShinyGO (v8.0) (http://bioinformatics. sdstate.edu/go/). In addition, a GSEA (gene-set enrichment analysis) was performed with BIOMART (https://www.ensembl.org/info/data/biomart/index.html).

Results

Descriptive analysis

The 21 CNT of 3,008 cattle, including udder index (UI), leg-foot index (LF), body index (BO), type index (TY), and general index (GN), are shown in Table 2. The variation values of the 16 linear scores for the body CNT ranged from 5.16 ± 0.027 (HQ) to 6.73 ± 0.024 (BD), as shown in Table 2. For the general index (GN) traits, the mean value of the total score was 80.81 ± 0.078, while the mean values of the composite index traits in the 5 components ranged from 78.91 ± 0.105 (LF) to 83.21 ± 0.127 (UI). The coefficient of variation (CV%) ranged from 5.27% to 32.52%.

Table 2.

Phenotypic means, standard errors (SE), coefficient of variation (CV) and ideal means for the evaluated traits

Trait1	n	Min	Max	Mean	SE	SD	CV (%)
Udder
UI2	3008	65	90	83.21	0.127	6.981	8.39
FUA3	3008	1	9	5.86	0.028	1.540	26.28
US3	3008	1	9	6.25	0.026	1.433	22.92
UD3	3008	1	9	5.22	0.031	1.698	32.52
FTP3	3008	1	9	5.30	0.027	1.499	28.30
FTL3	3008	1	9	5.54	0.020	1.113	20.07
RTP3	3008	1	9	6.33	.027	1.503	23.74
RUH3	3008	1	9	6.12	0.027	1.457	23.79
DC3	3008	1	9	6.19	0.029	1.588	25.63
Leg-foot
LF2	3008	65	90	78.91	0.098	5.387	6.83
HQ3	3008	1	8	5.16	0.027	1.474	28.56
RLRV3	3008	1	9	5.29	0.025	1.345	25.44
RLA3	3008	1	8	5.50	0.020	1.110	20.18
HD3	3008	1	9	5.30	0.025	1.365	25.76
Body
BO	3008	65	90	79.52	0.164	9.006	11.33
BD3	3008	1	9	6.73	0.024	1.321	19.63
CW3	3008	2	9	6.21	0.024	1.302	20.95
RW3	3008	1	9	6.20	0.023	1.255	20.24
RA3	3008	1	9	5.20	0.030	1.651	31.75
Type
TY	3008	65	90	79.31	0.144	7.889	9.95
General
GN	3008	65	89	80.81	0.078	4.260	5.27

1: UI: Udder index, FUA: Fore udder attachment, US: Udder support, UD: Udder depth, FTP: Front teat placement, FTL: Front teat lenght, RTP: Rear teat placement, RUH: Rear udder height, DC: Dairy character, LF: Leg foot, HQ: Hock quality, RLRV: Rear leg rear view, RLA: Rear leg angle, HD: Heel depth, BO: Body, BD: Body depth, CW: Chest width, RW: Rump width, RA: Rump angle, TY: Type, and GN: General. 2: Traits calculated 65-90, and 3: Traits measured in scores 1-9

Genomic variation and estimation of heritability

The genomic, residual, and phenotypic variances, as well as the SNP heritabilities of these traits, are shown in Table 3. The estimated heritabilities were low for all traits and ranged from 0.001 to 0.133, 0.014 to 0.064, 0.030 to 0.111, 0.078, and 0.036 for udder, leg-foot, body, type, and general traits, respectively.

Table 3.

Estimates of variance components and heritabilities for linear type traits

Traits	σ2g	σ2e	σ2p	h2
Udder
UI	0.041	39.943	39.984	0.001
FUA	0.144	1.710	1.853	0.078
US	0.059	1.779	1.838	0.032
UD	0.265	1.729	1.994	0.133
FTP	0.095	0.924	1.019	0.093
FTL	0.167	1.623	1.789	0.093
RTP	0.175	1.927	2.102	0.083
RUH	0.111	1.318	1.429	0.078
DC	0.097	1.413	1.510	0.064
Leg-foot
LF	0.279	20.071	20.350	0.014
HQ	0.082	1.186	1.268	0.064
RLRV	0.047	1.010	1.056	0.044
RLA	0.014	0.990	1.003	0.014
HD	0.029	0.940	0.969	0.030
Body
BO	3.831	30.558	34.390	0.111
BD	0.042	0.921	0.964	0.044
CW	0.034	1.116	1.150	0.030
RW	0.089	0.982	1.071	0.083
RA	0.163	1.935	2.098	0.078
Type
TY	2.952	35.128	38.080	0.078
General
GN	0.445	11.786	12.231	0.036

1: UI: Udder index, FUA: Fore udder attachment, US: Udder support, UD: Udder depth, FTP: Front teat placement, FTL: Front teat lenght, RTP: Rear teat placement, RUH: Rear udder height, DC: Dairy character, LF: Leg foot, HQ: Hock quality, RLRV: Rear leg rear view, RLA: Rear leg angle, HD: Heel depth, BO: Body, BD: Body depth, CW: Chest width, RW: Rump width, RA: Rump angle, TY: Type, and GN: General. 2: Traits calculated 65-90, and 3: Traits measured in scores 1-9

Genome-wide association study

Quantile-quantile plots (QQ plots) indicated that this study’s GWAS model was appropriate (Supplementary Figure 1 (SF1)). The lambda values (λ) were all close to 1.0, and the Manhattan plots show the results of GWAS significance levels (-log10 of the p-value of each SNP) by chromosomal position (Supplementary Figure 1 (SF1)). The single SNP regression analysis shows significantly associated SNP at the Bonferroni-corrected level of 5% for the CNTs (Table 4). FTP, FTL, RUH, HQ, BD, and RW shared a common SNP (rs109459144) that was 94 kb away from LOC100139826 on Bos taurus autosome (BTA11). Twelve SNPs were identified in previously documented genes or QTL, linked to conformation-related traits (Table 4). More than one CNT was influenced by the three significant SNP (rs41653166, BTA11: rs109459144, and BTA18: rs1163262).

Table 4.

Genome-wide significant SNPs for linear conformation traits

Traits1	SNP	Chr.	Pos.	MAF	P-value	Nearest gene	Distance (bp)	A_VAR
Udder
UI	rs41653166*	6	95988438	0.19	2.48 × 10-7	LOC784058	79,442	0.007
FTP	rs41567590	6	84475210	0.17	6.71 × 10-8	-	-	0.009
FTP	rs109459144*	11	65353177	0.19	5.31 × 10-9	*	*	0.014
FTL	rs109459144*	11	65353177	0.19	7.90 × 10-7	*	*	0.010
FUA	rs41586703	1	11603310	0.08	3.24 × 10-7	-	-	0.008
RUH	rs109459144*	11	65353177	0.19	5.71 × 10-8	*	*	0.010
Leg-foot
RLRV	rs109552830	3	117929861	0.49	9.28 × 10-8	RAMP1	Within	0.007
RLRV	rs438343186	4	114451421	0.31	1.28 × 10-6	FASTK	Within	0.005
HQ	rs109459144*	11	65353177	0.19	6.81 × 10-9	LOC100139826	94,823	0.010
HQ	rs108964424	15	75821904	0.41	8.71 × 10-7	TSPAN18	Within	0.006
RLA	rs109184865	13	20282525	0.19	1.22 × 10-6	ITGB1	Within	0.007
RLA	rs41632062*	18	25067681	0.17	2.93 × 10-7	LOC104974792 LOC104974793	1,287 14,849	0.008
Body
BO	rs109089868	16	57762562	0.37	3.10 × 10-7	MRPS14 TNN	31,415 13,862	0.008
BD	rs109459144*	11	65353177	0.19	4.39 × 10-8	*	*	0.007
RW	rs109417275	5	63113033	0.08	1.22 × 10-7	IKBIP	within	0.007
RW	rs41602734	5	80277740	0.15	2.94 × 10-7	TMTC1	within	0.014
RW	rs109459144*	11	65353177	0.19	5.10 × 10-10	*	*	0.014
RW	rs41666756	12	5413602	0.40	8.06 × 10-7	PCDH17	210,339	0.006
RW	rs41632062*	18	25067681	0.17	2.25 × 10-8	*	*	0.008
RW	rs41583184	26	37621755	0.27	3.72 × 10-8	SHTN1	within	0.008
RA	rs437402990	4	114446132	0.03	9.62 × 10-7	SLC4A2	within	0.010
General
GN	rs109547262	4	35605434	0.23	3.30 × 10-7	SEMA3D	within	0.006
GN	rs41653166*	6	95988438	0.19	3.14 × 10-8	ANTXR2	358294	0.007

1: UI: Udder index, FTP: Front teat placement, FTL: Front teat length, FUA: Front udder attachment, RUH: Rear udder height, RLRW: Rear leg rear view, HQ: Hock quality, RLA: Rear leg angle, BO: Body, BD: Body depth, RW: Rump width, RA: Rump angle, and GN: General. * Same SNps have found significant for different traits

Udder traits

A total of 4 SNPs (on BTA1, BTA6, and BTA11) were significant for five udder traits. One significant SNP on BTA11: 65353177 (rs109459144) was significant for RUH, FTP, and FTL. For UI, one significant SNP was determined on BTA6: 95988438 (rs41653166), 79 kb away from gene LOC784058. For FTP, two significant SNPs were identified on BTA6: 84475210 (rs41567590), and BTA11: 65353177 (rs109459144). For FUA, only one SNP on BTA1 was significant: 11603310 (rs41586703).

Leg and foot traits

A total of six significant SNPs were identified for leg and foot traits. Two important SNPs for RLRW were found in the RAMP1 and FASTK genes. These are BTA3: 117929861 (rs109552830) and BTA4: 114451421 (rs438343186). Within the TSPAN18 gene, 94 kb of LOC100139826 (shared with RUH, FTL, FTP, BD, and RW), BTA11: 65353177 (rs109459144), and BTA15: 75821904 (rs108964424), two significant SNP for HQ were identified. Two significant SNPs were identified in the ITGB1 gene and close to the LOC104974792 (1,287 bp) and LOC104974793 (14,849 bp) genes on BTA13: 20282525 (rs109184865) and BTA18: 25067681 (rs41632062).

Body traits

For trait BO, one SNP on BTA16 was significant: 57762562 (rs109089868). For BD, one SNP on BTA11: 65353177 (rs109459144) was significant, together with traits HQ, RW, and udder. For the traits of RW and RA, a total of seven SNPs were found to be significant. Three SNPs were located within the IKBIP, TMTC1, and SHTN1 genes for RW, and one SNP within the SEMA3D gene. The significant SNP for RW and RA were not significant for other CNTs.

General traits

For GN traits, two SNP were significant on BTA4: 35605434 (rs109547262) and BTA6: 95988438 (rs41653166). One significant SNP (rs109547262) was located within the SEMA3D gene. One SNP (rs41653166) was also significant for UI.

The gene enrichment (GO, KEGG, and GSEA) analyses

Eleven bovine genes associated with CNTs entered into the ShinyGO (v. 0.80) program are listed in Table 5. The gene ontology (GO) enrichment analysis showed that 49.7% of the analyzed genes were involved in the interaction between the extracellular matrix (ECM) and the receptors. 22.5% of the genes were involved in axon guidance, while 20.7% were associated with focal adhesion pathways (Tables 5 and 6 and Fig. 1).

Table 5.

The bovine genes related to conformation traits

Gene symbol	Ensembl gene ID	Entrez	Type	Chr.	Position (Mbp)	Description
TSPAN18	ENSBTAG00000013320	506550	Coding	3	99.8959	Tetraspanin 1
RAMP1	ENSBTAG00000011534	617017	Coding	3	117.2795	Receptor activity modifying protein 1
SEMA3D	ENSBTAG00000024394	536417	Coding	4	35.3902	Semaphorin 3D
SLC4A2	ENSBTAG00000011226	404084	Coding	4	113.6364	Solute carrier family 4 member 2
FASTK	ENSBTAG00000011228	509781	Coding	4	113.6515	Fas activated serine/threonine kinase
IKBIP	ENSBTAG00000021660	540640	Coding	5	62.7575	IKBKB interacting protein
TMTC1	ENSBTAG00000005370	533191	Coding	5	79.8077	Transmembrane O-mannosyltransferase targeting cadherins 1
ANTXR2	ENSBTAG00000014324	510080	Coding	6	94.5799	ANTXR cell adhesion molecule 2
ITGB1	ENSBTAG00000015910	281876	Coding	13	19.9716	Integrin subunit beta 1
TNN	ENSBTAG00000015650	517433	Coding	16	56.3254	Tenascin N
SHTN1	ENSBTAG00000007578	532603	Coding	26	37.2709	Shootin 1

Table 6.

Top pathways among the submitted genes influencing conformational traits

Enrichment FDR	nGenes	Pathway genes	Fold enrichment	Pathways
2.1E-02	2	80	49.7	ECM-receptor interaction
4.0E-02	2	177	22.5	Axon guidance
4.0E-02	2	192	20.7	Focal adhesion

Fig. 1.

A bar plot of genes associated biological pathways

The significant pathways based on the KEGG analysis are shown in the Supplementary Figures 2-4 (SF2-SF4). According to these results, genes play a role in cell biological processes (BP), such as ECM-receptor interaction, axon guidance, and focal adhesion pathways.

Gene enrichment analysis of eleven bovine genes was performed using Biomart software (https://www. ensembl.org/info/data/biomart/index.html). Based on the software analyses, all genes, except for the SLC4A2 gene, showed no phenotype description. The SLC4A2 gene is responsible for the phenotypic manifestation of osteopetrosis, a rare disease characterized by abnormal bone growth and excessive bone density (www.ncbi. nih.gov). Only three of the eleven genes showed an interaction in biological processes in cells. As shown in Table 7, the FDR values represent the p-values (P≤0.05) and gene interactions in the pathways. The genes ITGB1 and SEMA3A from the SEMA3 gene group were found to interact in the repulsion and attraction of axons. Accordingly, the genes are ITG1, SEMA3D, and TNN on BTA4, and the ITGB1 gene was found to interact with both SEMA3D and TNN genes, resulting in a statistically significant molecular function (P≤0.05). The ITG1 and TNN genes function within cells and play a role in the interaction with extracellular matrix (ECM) receptors. The pathway of ECM receptor interaction influences whether some related genes are transcribed more or less at the beginning of lactation (P<0.01). In addition, the ITG1 and TNN genes together have a focal adhesion function. It is hypothesized that they form large, dynamic protein complexes whose scaffold is associated with the ECM (P<0.05). It was found that the genes ITG1 and SEMA3D play a role in the emission of axons (nerve cell processes or fibers) that bring the neurons to their proper destinations (P<0.05).

Table 7.

The gene-set enrichment (GSEA) analysis of the candidate genes

Enrichment FDR	nGenes in interactions	Pathway (gene functions)	Interactions between genes
0.01752	2	BTA04512 ECM-receptor interaction	ITGB1 TNN (these two gene function in ECM-receptor interaction)
0.03297	2	BTA04360 Axon guidance	ITGB1 SEMA3D (these two genes function in Axon guidance)
0.03297	2	BTA04510 Focal adhesion	ITGB1 TNN (these two gene function in focal adhesion)

Discussion

Accurate heritability estimates are critical for predicting expected selection responses and estimating breeding values. In this study, the heritability of the CNTs was low, ranging from 0.04 to 0.13 (Table 3). Long et al. (2024) recently reviewed CNT studies and reported that the heritability of CNTs generally ranges from low to moderate. Several studies based on the pedigree kinship heritabilities of the CNTs reported low to medium, ranging from 0.09 to 0.38 by Campos et al. (2015), 0.10 to 0.32 by Djedovic et al. (2023), 0.04 to 0.23 by Olasege et al. (2019). A study by Roveglia et al. (2019) on the Italian Jersey breed estimated heritabilities ranging from low 0.04 (legs and locomotion) and 0.07 (foot angle), whereas stature shows moderate heritability at 0.32. Wu et al. (2013) reported a GWAS study heritability ranging from 0.07 (rump width) to 0.37 (stature). Pedigree-based heritability covers the whole genome and is expected to be higher than genomic heritability. Efforts have been made to enhance CNTs in THol dairy populations, but the low heritability of this trait may indicate smaller genetic gains.

Because of the research carried out to date on this topic, many researchers have investigated QTL on BTA6, indicating the presence of multiple QTLs on this chromosome. In a study conducted by Schrooten et al. (2000), two significant single nucleotide polymorphisms (SNPs) (rs41653166) were discovered on BTA6 at position 95988438. One is located 79 kb away from LOC784058 and is linked to UI and GN. The other is located in a QTL (10286) that has been associated with udder height (Schrooten et al., 2000), milking speed (Boichard et al., 2003), clinical mastitis (Lund et al., 2008), carcass weight (McClure et al., 2010), and teat length (Ashwell et al., 2005). Ashwell et al. (2005) identified another important SNP for FTP on BTA6: 84475210. This had a peak of 106.4 centiMorgan (cM) in a QTL associated with teat length (1569). SNPs and QTL affecting udder and general index traits were both significant in this region. The region in question was associated with candidate genes, suggesting that it warrants further investigation.

The two significant SNPs associated with RLRV (BTA3: 117929861 and BTA4: 114451421) are located in the RAMP1 and FASTK genes, respectively. The RAMP1 gene encodes a protein called receptor activity modifying protein 1, while the FASTK gene encodes a serine/threonine kinase that is activated by Fas. Cells with RAMP1 gene over-expression show typical calcium deposits in the bone matrix structure (Zhao et al., 2013). In addition, several QTLs in this specific genomic region have been reported to have effects on body weight (McClure et al., 2010). These genes are proposed as functional candidate genes for RLRV.

In this study, six significant SNPs for RW were found on BTA5, BTA11, BTA12, BTA18, and BTA26. A significant relationship between aspects of fertility and rump traits has been shown in studies. Animals with wide pins, long sloping rumps, and a low slope from the thurl to the pin bone are favored for easy calving (Ali et al., 1984; Cue et al., 1990). Similarly, Lu et al. (2021) reported that the rump angle is closely related to the reproductive performance of dairy cows. Cows with high and narrow pin bones had an increased risk of retained placenta (Van Dorp et al., 1998). Makgahlela et al. (2009) reported longer calving intervals associatied with deep angular bodies and steep rump angles. In this study, two notable SNPs were discovered for RW on BTA5: 63113033 and BTA5: 80277740, located in the IKBIP and TMTC1 genes, respectively. Weller et al. (2018) reported OTL for thurl width on BTA5. Boichar et al. (2003) also identified a significant quantitative trait locus (QTL) on BTA5 at position 103.1 (cM) for traits related to RW and rump length. Rump breeding values indicated that animals with an average rump angle (from 4 to 6 on a scale of 1 to 9) had a lower culling rate than animals at the extremes (high and low pin bones) (Wall et al., 2005). The relative risk of unwanted culling is smallest at intermediate rump angles (Caraviello et al., 2004). The use of RW can help overcome management issues that may be present in fertility performance. The RA can provide additional information for the estimation of the breeding value of fertility in dairy cattle breeding. Further studies for RW and RA may be beneficial for estimating genomic breeding values. Therefore, this study has shown there is also some potential in the RW and RA traits in helping to estimate genomic breeding values of fertility.

Several QTLs for the rump angle and body weight were detected on BTA11 (Boichard et al., 2003; McClure et al., 2010). McClure et al. (2010) found an important QTL for calving ease on the BTA12 peak located at 6.04 (cM), which covers the locus where we detected a significant SNP (rs41666756; BTA12: 5413602) for RW. The significant SNP (rs41666756) at BTA12: 5413602 for RW is 210,339 bp away from the PCDH17 gene. McClure et al. (2010) found a QTL (10914) on BTA12 peak 6.04 (cM), affecting calving ease. Müller et al. (2017) found an important QTL (126847) on BTA5 peak at 65.85 (cM) for calving ease and stillbirths. Consistently, a very prominent QTL for different fertility traits was identified on BTA5 in one region. In this study, a significant SNP (rs109459144) on BTA11: 65353177 affected five CNTs (HQ, RW, BD, RUH, FTL, and FUA) 98 kb near LOC100139826, an uncharacterized protein-coding gene. This SNP within the QTL region is associated with rump angle (Boichard et al., 2003) and body weight (Michenet et al., 2016). In this study, a notable SNP (rs41583184) was identified on BTA26: 37621755 for RW, which is consistent with the results of Thomasen et al. (2008), who reported a prominent quantitative trait locus (QTL) on BTA26 affecting stillbirths. The RLA and RW traits share a common SNP (rs41632062) on BTA18: 25067681, which is probably the most important conformation trait for calving ease (Dadati et al., 1985). The results of the study indicate that the major SNP for RW is within QTL associated with calving ease, stillbirth, RW, and rump angle phenotypes (Boichard et al., 2003; Thomasen et al., 2008; McClure et al., 2010). Therefore, this study provides additional evidence for a QTL or several closely related QTLs.

Gene set enrichment analysis has been a successful extension of genome-wide association analysis (Abdalla et al., 2016). Considering GO and GSEA analyses, ITGB1 and TNN genes were found to play a role in extracellular matrix (ECM) receptor interactions, ITGB1 and SEMA3D in axon guidance interactions, and ITGB1 and TNN genes in focal adhesion interactions (Tables 6 and 7). The ECM is crucial for the development of animal tissues and organs. Enzymes involved in the ECM signaling pathway promote cell proliferation, cell differentiation, skeletal development, and morphogenesis (Jeong et al., 2017). The ECM has been found to promote the functional recovery of several tissue components, including musculoskeletal tissues (Zhao and Bass, 2018). In addition to the interactions, the influence of these genes on CNTs was also determined. In this regard, ITGB1, TNN, and SEMA3D genes were found to influence the leg-foot criterion, the body trait, and the general CNTs, respectively (Table 4). These genomic markers overlap with potential candidate genes that are involved in biological pathways such as skeletal development, morphogenesis, and adipogenesis. This study confirmed previous results from GWAS of CNTs and also identified additional regions in the cattle genome associated with these economically important traits. These results can be used to increase genetic progress in breeding programs.

Efforts have been made to improve CNTs in THol dairy populations, but the low heritability of this trait may indicate smaller genetic gains. A reference population of reasonable size is crucial for genomic selection; the amount of both genotyped individuals of CNTs is smaller. Our results showed that the accuracy of genomic prediction for CNTs could be improved by GWAS. Therefore, both genotyped and phenotype animals in the reference population need to be enhanced. The present GWAS identified 16 significant SNPs associated with 13 CNTs in the THol population. The significant SNP markers associated with the ITGB1, TNN, and SEMA3D genes discovered in this research could have potential for researchers and breeders in the genomic selection of CNTs in dairy cattle breeding. These results contribute to the elucidation of the genetic factors responsible for conformation traits in the THol cattle population.

Conflict of interest

The authors declare that they have no conflict of interest.

Supporting Online Material

Refer to web version on PubMed Central® (PMC) for Supplementary Material.