Estimation of Genetic Parameters and Weighted Single-Step Genome-Wide Association Study for Indicators of Colostrum Quality in Chinese Holstein Cattle

Abstract

Colostrum is the milk harvested during the first few hours after calving, which contains high levels of immunoglobulins, antimicrobial peptides, and growth factors essential for the health of neonates. The primary objective of this study was to investigate the genetic background of colostrum quality traits (based on Brix percentage) in Holstein cows. Using phenotypic records of 58,338 Holstein cows from 37 dairy farms, we identified significant systematic effects influencing colostrum quality measured by digital Brix refractometer, estimated genetic parameters, and performed weighted single-step genome-wide association studies (WssGWAS) to identify genomic regions and candidate genes associated with these traits. The average (±SD) Brix percentage was 23.76 ± 3.25%. With heritability values ranging from 0.21 ± 0.03 (Brix in third parity) to 0.30 ± 0.02 (Brix in second parity), colostrum quality was determined to be moderately heritable. Genetic correlations between colostrum quality across parities ranged from 0.37 ± 0.14 to 0.81 ± 0.13. For colostrum quality from cows in the first, second, and third parities, WssGWAS enabled the identification of 30, 32, and 38 genomic regions explaining 4.18%, 4.42%, and 5.58% of the total additive genetic variance, respectively. Two immune-related genes (CNR1 and ZXDC) were identified as promising candidate genes for colostrum quality traits. In summary, colostrum quality measured in first parity cows should be evaluated as a different trait from measurements in later parities in breeding programs. These findings provide useful information for dairy breeders to genetically improve colostrum quality in dairy cattle populations.

1. Introduction

Colostrum is a special secretion of the mammary gland in mammals in the first milking after parturition. Compared to mature milk, bovine colostrum contains more protein, dry matter, immunoglobulin (Ig), including IgG, IgM, and IgA, hormones, peptides with antimicrobial action, and growth factors [1]. High quality colostrum is essential for the immune protection of calves through passive immunity, which depends on the rapid intake of good quality colostrum after birth [2]. As the predominant Ig in bovine colostrum, IgG content has been widely used to assess colostrum quality in previous studies (e.g., Gulliksen et al., Le Cozler et al. [3,4]). However, its measurement requires highly controlled environments and long processing times, which makes it difficult to be routinely performed on farms. Additionally, it can be costly to obtain IgG data from numerous cows. Colostrum collected within the first few hours after parturition has been shown to have high correlations between its Brix (% total solids in a solution) and IgG concentration (phenotypic correlations: 0.71–0.78, genetic correlation: 0.91) [5,6,7]. The IgG concentration of fresh colostrum and first milking can be characterized as >50 g/L when Brix > 22% [5]. Brix can be measured on farm using a handheld optical refractometer or digital refractometer, and it provides a rapid and cost-effective quality assessment of the colostrum quality in the field [8,9,10]. In the last decade, Brix has been used as a suitable method in dairy farms to evaluate colostrum quality [11,12,13].

The colostrum quality can be influenced by several factors such as breed, parity, cow health status, season, and length of the dry period [14]. Parity and colostrum quantity are related to colostrum quality, in which IgG content of later-parity cows tend to be higher [2], and cows that produce more colostrum typically have lower-quality colostrum because of the dilution effect [11]. The time between calving and colostrum collection is also associated with reduced amount of IgG in the milk [15]. According to earlier research, the heritability of colostrum quality parameters varied between 0.08 and 0.21 for IgG and 0.26 to 0.31 for Brix. These studies focused on Italian Holstein cows [7], Swedish dairy cows including Swedish Red, Holstein, and crossbreeds [16], Charolais cows [17], and Greek Holstein cows [18]. The latest colostrum collection time reported by those authors was 960 min postpartum [18]. Furthermore, to our knowledge, only two studies have performed genome-wide association studies (GWAS) for colostrum quality traits. Kiser et al. [19] identified a single SNP associated with Brix% in colostrum from 345 Jersey cows and Lin et al. [20] reported 75 significant SNPs and seven candidate genes associated with Ig concentrations (IgG, IgG1, and IgG2) in the colostrum of 588 Holstein cows. The candidate genes identified include FGFR4, FGFR2, NCF1, IKBKG, SORBS3, IGHV1S18, and KIT, which are related to important functions such as control of inflammation, NF-κB signaling pathway activation, and Ig heavy chain encoding [20].

Accurate estimates of genetic parameters are essential for designing or refining selection indexes and breeding programs to include novel traits. Although there are reports of heritability estimates for colostrum quality traits in dairy cattle populations, there is a lack of genetic correlation estimates between indicators of colostrum quality across parities. Furthermore, there are currently few known genomic areas and potential genes linked to colostrum quality. Therefore, the main objectives of this study were to (1) describe the phenotypic variability of colostrum quality in Chinese Holstein cattle; (2) estimate variance components and genetic parameters, including heritability and genetic correlations between colostrum quality traits in dairy cattle; and, (3) identify genomic regions and candidate genes associated with colostrum quality in Chinese Holstein cattle.

2. Materials and Methods

2.1. Ethics Committee Approval

The blood samples used for genotyping were collected by veterinarians along with the regular quarantine inspection of the farms and breeding stations. Therefore, no ethical approval was required for this study.

2.2. Data

2.2.1. Phenotypes

This study included data from 88,619 Chinese Holstein cows raised on 37 dairy farms located in nine Chinese provinces. Key variables such as cow ID, parity, birth date, colostrum records, Dairy Herd Improvement (DHI) data, and culling records were provided by the Beijing Sunlon Livestock Development Co., Ltd. (Beijing, China).

Colostrum records were collected from February 2016 to August 2022, which included the actual time of colostrum collection and colostrum quality (Brix, %). Only the colostrum samples harvested in the first 6 h post-calving were retained for subsequent analyses. Colostrum quality was tested using handheld ATC Brix refractometer or digital Brix refractometer (PAL-1, ATAGO Co., Ltd., Tokyo, Japan), which assessed the percentage of total solids in the milk samples. Records from cows with Brix values lower than 10% or higher than 40% (4% of the total number of records) and records without cow ID, parity, or calving date information (32%) were removed from further analyses. As the amount of data decreased in later parities, the records from parity 4 or later parities were not used in the present study. After quality control and data editing, 58,338 cows with 75,233 phenotypic records were retained in the dataset.

2.2.2. Pedigree

All the cows with phenotypic records were traced back as many generations as possible. A total of 51,605 cows with 67,206 colostrum records had pedigree records. The final pedigree included 120,433 females and 3971 males.

2.2.3. Genotypes

A total of 1885 female cows were genotyped using the Illumina 150K Bovine Bead Chip (Illumina Inc., San Diego, CA, USA). Genomic quality control was performed using the PLINK v1.90 beta software [21]. We excluded single nucleotide polymorphisms (SNPs) with minor allele frequency lower than 0.1 or SNPs with extreme departure from the Hardy–Weinberg equilibrium (p-value < 10⁻⁶). After genomic quality control, 109,619 SNPs were used in the study.

2.3. Statistical Model Development

We used the MIXED procedure from the SAS software (version 9.2, SAS Institute Inc., Cary, NC, USA) to identify systematic effects significantly influencing colostrum quality (Brix, %). The fixed effects tested included parity number (1, 2, and 3), interval between calving and colostrum collection (0–2 h, 2–6 h), farm area scale (farms in Beijing with more than 2000 records, farms in Beijing with less than 2000 records, farms out of the Beijing area with more than 20,000 records, farms outside of Beijing with 5000 to 10,000 records, farms outside of the Beijing area with less than 5000 records), calving year (2016, 2017, 2018, 2019, 2020, 2021, and 2022), and calving month (1 to 12).

2.4. Estimation of Genetic Parameters

The (co)variance components were estimated using AI-REML and EM-REML procedures implemented in the AIREMLF90 (v 1.148) from BLUPF90 (v 1.70) [22]. For colostrum quality traits, we divided the records into groups: Brix records from first-parity cows (Brix1), Brix records from second-parity cows (Brix2), Brix records from third-parity cows (Brix3).

Variance components and heritability for Brix1, Brix2, and Brix3 were estimated based on the model:

$$\mathit{y} \mathbf{=} \mathit{X}\beta \mathbf{+} \mathit{Z}\mathit{\alpha } \mathbf{+} \mathit{e}$$

(1)

Genetic correlations between colostrum quality traits in different parity were estimated using two-trait models as described below:

$$\left[\begin{array}{c}{\mathit{y}}_{\mathbf{1}}\\ {\mathit{y}}_{\mathbf{2}}\end{array}\right] \mathbf{=} \left[\begin{array}{cc}{\mathit{X}}_{\mathbf{1}}& \mathbf{0}\\ \mathbf{0}& {\mathit{X}}_{\mathbf{2}}\end{array}\right]\left[\begin{array}{c}{\mathit{\beta }}_{\mathbf{1}}\\ {\mathit{\beta }}_{\mathbf{2}}\end{array}\right] \mathbf{+} \left[\begin{array}{cc}{\mathit{Z}}_{\mathbf{1}}& \mathbf{0}\\ \mathbf{0}& {\mathit{Z}}_{\mathbf{2}}\end{array}\right]\left[\begin{array}{c}{\mathit{\alpha }}_{\mathbf{1}}\\ {\mathit{\alpha }}_{\mathbf{2}}\end{array}\right] \mathbf{+} \left[\begin{array}{c}{\mathit{e}}_{\mathbf{1}}\\ {\mathit{e}}_{\mathbf{2}}\end{array}\right]$$

(2)

In models 1 and 2, $\mathit{y}$, ${\mathit{y}}_{\mathbf{1}}$, and ${\mathit{y}}_{\mathbf{2}}$ are the vectors of phenotypic records. The subscripts 1 and 2 in model (2) indicate traits 1 and 2, respectively. $\mathit{\beta }$,$\mathbf{ }{\mathit{\beta }}_{\mathbf{1}}$, and ${\mathit{\beta }}_{\mathbf{2}}$ are the vector of fixed effects for colostrum quality traits, including year-season of calving (27 levels), interval between calving and colostrum collection (2 levels), farm area scale (5 levels); $\mathit{\alpha }$, ${\mathit{\alpha }}_{\mathbf{1}}$ and ${\mathit{\alpha }}_{\mathbf{2}}$ are the vector of random additive genetic effects, following $\mathit{\alpha }\mathbf{~}\mathit{N}\mathbf{(}\mathbf{0}\mathbf{,}\mathit{H}{\mathit{\sigma }}_{\mathit{a}}^{\mathbf{2}}\mathbf{)}$; $\mathit{e}$, ${\mathit{e}}_{\mathbf{1}}$ and ${\mathit{e}}_{\mathbf{2}}$ are the vectors of random residual effects following $\mathit{e}\mathbf{~}\mathit{N}\mathbf{(}\mathbf{0}\mathbf{,}{\mathit{I}\mathit{\sigma }}_{\mathit{e}}^{\mathbf{2}}\mathbf{)}$; $\mathit{X}$, ${\mathit{X}}_{\mathbf{1}}$, ${\mathit{X}}_{\mathbf{2}}$, $\mathit{Z}$, ${\mathit{Z}}_{\mathbf{1}}$, and ${\mathit{Z}}_{\mathbf{2}}$, are the corresponding incidence matrices; $\mathit{H}$ is the matrix of additive genetic relationships constructed from the pedigree and genotype; ${\mathit{\sigma }}_{\mathit{a}}^{\mathbf{2}}$ is the additive genetic variance; I is an identity matrix; and, ${\mathit{\sigma }}_{\mathit{e}}^{\mathbf{2}}$ is the residual variance. The inverse of the H matrix (H⁻¹) was calculated as follows [23],

$${\mathit{H}}^{\mathbf{-}\mathbf{1}} \mathbf{=} {\mathit{A}}^{\mathbf{-}\mathbf{1}} \mathbf{+} \left[\begin{array}{cc}\mathbf{0}& \mathbf{0}\\ \mathbf{0}& {\mathit{G}}^{\mathbf{-}\mathbf{1}} \mathbf{-} {\mathit{A}}_{\mathbf{22}}^{\mathbf{-}\mathbf{1}}\end{array}\right]$$

(3)

where ${\mathit{A}}^{\mathbf{-}\mathbf{1}}$ is the inverse of the pedigree-based relationship matrix; ${\mathit{A}}_{\mathbf{22}}^{\mathbf{-}\mathbf{1}}$ is the ${\mathit{A}}^{\mathbf{-}\mathbf{1}}$ for the genotyped animals; and ${\mathit{G}}^{\mathbf{-}\mathbf{1}}$ is the inverse of the genomic relationship matrix. The $\mathit{G}$ matrix was calculated as [24]:

$$\mathit{G} \mathbf{=} {\displaystyle \frac{{\mathit{Z}\mathit{D}\mathit{Z}}^{\mathbf{\prime }}}{\mathbf{2}{\sum }_{\mathit{i} \mathbf{=} \mathbf{1}}^{\mathit{M}}{\mathit{P}}_{\mathit{i}}\mathbf{(}\mathbf{1}\mathbf{-}{\mathit{P}}_{\mathit{i}}\mathbf{)}}}$$

(4)

where $\mathit{Z}$ is the matrix of genotypes adjusted for allele frequencies (0, 1, or 2 for aa, Aa, and AA, respectively); $\mathit{D}$ is a diagonal matrix of weights for SNP variances (initially $\mathit{D}$= $\mathit{I}$); $\mathit{M}$ is the number of SNPs; and ${\mathit{P}}_{\mathit{i}}$ is the minor allele frequency of the i_th SNP.

According to the one-tailed t-test, a genetic correlation was significantly different from zero when the absolute value of the genetic correlation was greater than 0 + standard error × 1.645.

2.5. Weighted Single-Step Genome-Wide Association Study (WssGWAS)

The estimates of SNP effects and weights for the WssGWAS analyses (five iterations) for colostrum quality traits were obtained according to Wang et al. [25]. The weight for each SNP was calculated as: ${\mathit{d}}_{\mathit{i}} \mathbf{=} {\mathbf{1.125}}^{{\displaystyle \frac{\left|{\widehat{\mathit{a}}}_{\mathit{i}}\right|}{\mathit{s}\mathit{d}\mathbf{\left(}{\widehat{\mathit{a}}}_{\mathit{i}}\mathbf{\right)}}}\mathbf{-}\mathbf{2}}\mathbf{ }$ [24], where i is the i_th SNP. The percentage of the total addictive genetic variance explained by the i_th region was calculated as:

$${\displaystyle \frac{\mathit{V}\mathit{a}\mathit{r}\mathbf{\left(}{\mathit{a}}_{\mathit{i}}\mathbf{\right)}}{{\mathit{\sigma }}_{\mathit{a}}^{\mathbf{2}}}} \mathbf{\times } \mathbf{100}\mathbf{\%} \mathbf{=} {\displaystyle \frac{\mathit{V}\mathit{a}\mathit{r}\mathbf{\left(}{\sum }_{\mathit{j} \mathbf{=} \mathbf{1}}^{\mathbf{10}}{\mathit{Z}}_{\mathit{j}}{\widehat{\mathit{u}}}_{\mathit{j}}\mathbf{\right)}}{{\mathit{\sigma }}_{\mathit{a}}^{\mathbf{2}}}} \mathbf{\times } \mathbf{100}\mathbf{\%}$$

(5)

where ${\mathit{a}}_{\mathit{i}}$ is genetic value of the i_th region that consists of 10 contiguous SNPs, ${\mathit{\sigma }}_{\mathit{a}}^{\mathbf{2}}$ is the total additive genetic variance,${\mathit{Z}}_{\mathit{j}}$ is a vector of gene content of the j_th SNP for all individuals, and ${\widehat{\mathit{u}}}_{\mathit{j}}$ is the marker effect of the j_th SNP within the i_th region.

The Manhattan plots showing the windows were created using the “ggplot2” (v 3.5.1) R package [26]. Non-overlapping contiguous genomic windows that explained 0.10% or more of the total additive genetic variance were considered to be associated with the studied traits. Candidate genes associated with colostrum quality were identified by examining genomic windows associated with colostrum quality traits based on the ARS-UCD1.2 [27]. The biological functions of these candidate genes, Gene Ontology (GO) terms [28,29], and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway [30,31,32] enrichment were identified using the R package “BiomaRt” (v 2.62.0) [33] and “clusterProfiler” (v 4.6.2) [34].

3. Results

3.1. Descriptive Statistics and Factors Influencing Colostrum Quality

Colostrum quality measured by Brix percentage in Chinese Holstein cows averaged 23.76 ± 3.25%. For parities 1 to 3, colostrum quality ranged from 23.50% to 23.98%, as shown in

Table 1. Descriptive statistics of colostrum quality in Chinese Holstein cattle.

Trait	Parity	No. of Records	Mean	SD	Min	Max	CV
Colostrum quality (Brix, %)	1	28,624	23.50	3.32	10.2	40	14.13%
	2	27,688	23.89	3.19	10	40	13.35%
	3	18,921	23.98	3.21	11	40	13.39%
	All	75,233	23.76	3.25	10	40	13.68%

. A total of 6.50% of the Brix data was below 20% and 7.62% exceeded 28%. Figure 1 displays the distribution of colostrum quality values. Colostrum quality increased with parity (Figure 1).

The factors significantly affecting colostrum quality (Brix, %) were year and month of calving, parity, interval between calving and colostrum collection, and farm area scale. The estimated least square means and multiple comparisons are shown in

Table 2. Effect of various factors on colostrum quality (Brix, %) in Chinese Holstein cattle.

Effects	Level	N 1	LSM ± SE 1
Calving year	2016	4943	23.85 ± 0.05 Ee
	2017	13,644	24.03 ± 0.04 Dd
	2018	10,970	24.12 ± 0.04 Dd
	2019	10,822	24.34 ± 0.04 Bb
	2020	7705	25.27 ± 0.04 Cc
	2021	16,524	25.51 ± 0.03 Aa
	2022	10,625	25.51 ± 0.04 Aa
Calving month	1	5974	24.59 ± 0.05 DEde
	2	5566	24.72 ± 0.05 BCDcd
	3	5337	24.91 ± 0.05 ABab
	4	4396	24.67 ± 0.05 CDEcd
	5	5223	24.79 ± 0.05 ABCDabcd
	6	6435	24.46 ± 0.04 Ee
	7	8664	24.23 ± 0.04 Ff
	8	9130	24.17 ± 0.04 Ff
	9	6333	24.73 ± 0.04 BCDbcd
	10	6323	24.85 ± 0.04 ACac
	11	5964	24.95 ± 0.05 Aa
	12	5888	24.86 ± 0.05 ABCabc
Parity	1	28,624	24.24 ± 0.03 Bb
	2	27,688	24.84 ± 0.03 Aa
	3	18,921	24.90 ± 0.03 Aa
Interval between calving and colostrum collection (in hours)	0–2	42,776	24.82 ± 0.03 Aa
	2–6	32,457	24.50 ± 0.03 Bb
Farm area scale 2	Group 1	13,859	23.59 ± 0.03 Dd
	Group 2	13,557	23.18 ± 0.03 Ee
	Group 3	24,310	25.36 ± 0.04 Bb
	Group 4	15,188	24.54 ± 0.04 Cc
	Group 5	8319	26.63 ± 0.04 Aa

^A–F Mean values in the same column with different superscripts differ (p < 0.01) for each effect factors. ^a–f Mean values in the same column with different superscripts differ (p < 0.05) for each effect factors. ¹ LSM = least squares mean; SE = stand error; N = number of records. ² Farm area scale was divided into five groups: Group 1 (farms in Beijing with more than 2000 data), Group 2 (farms in Beijing with less than 2000 data), Group 3 (farms outs of Beijing with more than 20,000 data), Group 4 (farms out of Beijing with 5000 to 10,000 data), Group 5 (farms out of Beijing with less than 5000 data).

. Colostrum quality increased gradually from 2016 to 2021, and the values of 2021 and 2022 were similar. This may be due to steady improvements in farm management practices over time. November was the month with the highest colostrum quality while August had the lowest Brix values. Colostrum quality increased with parity, and colostrum collected at 0–2 h was of higher quality than samples collected between 2 and 6 h post-calving. Significant differences existed between farms of various sizes and locations, which can be attributed to differences in weather, nutrition practices, and management.

3.2. Genetic Parameters of Colostrum Quality

3.2.1. Heritability Estimates

Variance components and heritability estimates for colostrum quality are shown in

Table 3. Variance components and heritability estimates for colostrum quality (Brix, %) traits in Chinese Holstein cattle.

Trait a	N b	${\mathit{\sigma }}_{\mathit{a}}^{\mathbf{2}}$ ± SE b	${\mathit{\sigma }}_{\mathit{e}}^{\mathbf{2}}$ ± SE b	${\mathbf{h}}^{\mathbf{2}}$ ± SE b
Brix1	24,108	2.68 ± 0.20	6.69 ± 0.16	0.29 ± 0.02
Brix2	25,685	2.54 ± 0.19	5.82 ± 0.17	0.30 ± 0.02
Brix3	17,413	1.80 ± 0.02	6.91 ± 0.24	0.21 ± 0.03

^a Brix1 = Brix for first-parity cow records; Brix2 = Brix for second-parity cow records; Brix3 = Brix for third-parity cow records; Brix = Brix for all records. ^b ${\sigma }_a^2$ = additive genetic variance; ${\sigma }_e^2$ = residual variance; ${\mathrm{h}}^2$ = heritability; SE = standard error; N = number of records.

. The heritability estimates of colostrum quality ranged from 0.21 ± 0.03 (Brix3) to 0.30 ± 0.02 (Brix2). The highest additive genetic variance estimates were observed for colostrum quality in the second parity.

3.2.2. Genetic Correlations and Repeatability Estimates

Genetic and phenotypic correlations between Brix1, Brix2, and Brix3 estimated using two-trait animal models are presented in

Table 4. Genetic (above diagonal) and phenotypic (below diagonal) correlations between colostrum quality traits in Chinese Holstein cattle.

Trait 1	Brix1	Brix2	Brix3
Brix1		0.57 ± 0.07 *	0.37 ± 0.14 *
Brix2	0.11 ± 0.01 *		0.81 ± 0.13 *
Brix3	0.04 ± 0.02 *	0.16 ± 0.01 *

* Significant genetic correlation (absolute value of genetic correlation > 0 + standard error × 1.645). ¹ Brix1 = Brix for first-parity cow records; Brix2 = Brix for second-parity cow records; Brix3 = Brix for third-parity cow records.

. Brix1 was not highly genetically correlated with Brix2 (0.57 ± 0.07) and Brix3 (0.37 ± 0.14), but all the genetic correlations were significant. Brix2 and Brix3 had a strong genetic correlation (0.81 ± 0.13). Additionally, there were only weak phenotypic associations amongst Brix traits (ranging from 0.04 ± 0.02 to 0.16 ± 0.01).

3.3. Weighted Single-Step Genome-Wide Association Studies

Since there was no strong estimated genetic correlation between Brix1 and other two colostrum quality traits, as well as insufficient data for fitting repeatability models, colostrum quality traits of different parities of the cows were considered as different traits. Figure 2 presents the genetic variance of each genomic window after performing WssGWAS. One hundred genomic regions in all that met the predetermined threshold point (0.10%) were found. For Brix1, Brix2, and Brix3, 30, 32, and 38 genomic regions were identified, accounting for 4.18%, 4.42%, and 5.58% of the total additive genetic variance, respectively. These genomic regions are located in all the Bos taurus autosomes (BTA) except BTA16 and BTA21. The detailed information about these genomic regions and candidate genes are presented in Table S1. A total of 189 protein encoding genes (44, 69, and 93 genes located based on the relevant genomic regions associated with Brix1, Brix2, and Brix3, respectively) were annotated in these genomic regions according to the Ensembl database. Among the three colostrum quality traits, the highest proportions of the total additive genetic variance explained by a single genomic window were 0.28 (Brix1, BTA10, 28.79–29 Mb), 0.35 (Brix2, BTA6, 38.23–38.25 Mb), and 0.27 (Brix3, BTA22, 60.79–61.00). However, no protein encoding gene was found in those regions.

Five overlapping genomic regions between two colostrum quality traits were discovered. As

Table 5. Shared genomic regions associated with colostrum Brix (%) traits in Chinese Holstein cattle.

Chromosome	Regions, Mb	Proportion of the Total Additive Genetic Variance Explained, %			Candidate Genes
		Trait 1
		Brix1	Brix2	Brix3
BTA6	38.23–38.25	0.24	0.35	-
BTA9	61.69–61.88	0.13	-	0.12	CNR1, SPACA1
BTA19	12.81–13.04	-	0.14	0.19	CA4, ZNHIT3, MYO19, PIGW, GGNBP2, DHRS11, MRM1
BTA19	13.07–13.26	-	0.13	0.14
BTA22	60.46–60.58	-	0.11	0.17	CHST13, UROC1, ZXDC, SLC41A3, ALDH1L1

¹ Brix1 = Brix for cows with first parity; Brix2 = Brix for cows with second parity; Brix3 = Brix for cows with third parity. “-” = additive genetic variance lower than 0.10%.

shows, these overlapping regions included BTA6, 38.23–38.25 Mb (Brix1 and Brix2), BTA9, 61.69–61.88 Mb (Brix1 and Brix3), BTA19, 12.81–13.04 Mb (Brix2 and Brix3), BTA19, 13.07–13.26 Mb (Brix2 and Brix3), and BTA22, 60.46–60.58 Mb (Brix2 and Brix3).

There was no protein encoding gene found in two regions including BTA6, 38.23–38.25 Mb and BTA19, 13.07–13.26 Mb. However, two, seven, and five genes were discovered in the other three regions, respectively. These genes included CNR1 (Cannabinoid Receptor 1), SPACA1 (Sperm Acrosome Associated 1), CA4 (Carbonic Anhydrase 4), ZNHIT3 (Zinc Finger HIT-Type Containing 3), MYO19 (Myosin XIX), PIGW (Phosphatidylinositol Glycan Anchor Biosynthesis Class W), GGNBP2 (Gametogenetin Binding Protein 2), DHRS11 (Dehydrogenase/Reductase 11), MRM1 (Mitochondrial RRNA Methyltransferase 1), CHST13 (Carbohydrate Sulfotransferase 13), UROC1 (Urocanate Hydratase 1), ZXDC (ZXD Family Zinc Finger C), SLC41A3 (Solute Carrier Family 41 Member 3), and ALDH1L1 (Aldehyde Dehydrogenase 1 Family Member L1).

To further interpret the candidate genes, we performed functional genomic enrichment analyses on these genes for each parity trait. Thirteen, three, and seven significant KEGG pathways, and 67, 47, and 13 significant GO terms were significantly (p-value < 0.05) enriched by Brix1, Brix2, and Brix3, respectively. The significant pathways for colostrum quality traits are shown in Table S2 and Figure 3. A total of 19, 20, and 27 genes, respectively, were enriched in significant KEGG pathways or GO terms.

Many enriched genes related to cell cycle regulation, intracellular protein transport, cytoskeletal, and transcriptional control were enriched. Some pathways directly associated with immune system, such as GO:0002228 (natural killer cell mediated immunity), GO:0045087 (innate immune response), and GO:0001909 (leukocyte mediated cytotoxicity), and we found LYST (Lysosomal Trafficking Regulator) from Brix1 in these pathways. Colostrum quality traits did not have any overlapping pathways, but they shared biological processes or functions such as cell cycle, transcription, and protein transport.

4. Discussion

Other investigations found slightly higher Brix levels than our findings. For example, an average Brix of 25.1% for first parity, 24.7% for second parity, and 27.6% for third and later parities have been reported for German Holstein cows [11]. Soufleri et al. [18] found an average Brix of 25.80% for Greek Holstein cows. The slightly lower values observed in Chinese Holsteins may be due to the differences in the genetic background of the population, milk production levels, and management practices.

Some reports have discussed the factors that affect the quality of colostrum. For instance, the average concentration of IgG has been reported to be lower in colostrum from cows exposed to a high Temperature–Humidity Index [35]. This is consistent with our results, as July and August are summer months in China. Heat stress can impair the transfer of IgG from the blood stream to the udder, which can contribute to lower colostrum quality [35,36]. In line with earlier research, colostrum quality steadily increased from the first to the third parity [2,3,37]. As the cow’s age increased, the number of transferrable antibodies tended to increase, and there may be an association between the number of transferrable antibodies and IgG concentration in colostrum, which could result in higher colostrum quality in later parities [38]. In addition, colostrum quality declined over time after calving, which is consistent with literature reports [15,39]. In our study, although colostrum quality from 2–6 h post-calving (24.50 ± 0.03) was significantly lower than that from 0–2 h (24.82 ± 0.03), both colostrum from 0–2 h and 2–6 h were higher than 22%, which implies that they were good quality colostrum [40].

Brix has been reported as a heritable trait in other Holstein cattle populations with estimates of 0.26 and 0.27 in Northern Italy and Greece populations [7,18], which are similar to our results. These findings support the colostrum quality’s moderate heritability. Because adding farm as an effect caused convergence issues, we used the farm area scale factor when building the models. This issue was probably due to the unbalanced data distribution across farms (hundreds to more than 20,000). As shown in Table 2, farm area scale significantly affected colostrum quality. Farm area scale eliminated some of the effects of farms because it took into consideration the effects of farm location and size.

The low estimated genetic correlation between Brix1 and other colostrum quality traits could be due to cows acquiring new antibodies throughout their lives [16]. Older cows might have greater and cumulative exposure to antigens, causing the Ig concentration in colostrum to vary across parities [7,16]. This phenomenon is similar to certain fertility traits, such as the interval from first to last insemination and the number of inseminations per conception, which show different performance between heifers and cows [41]. In fertility traits, the physiological status of cow changes considerably after first calving [42]. However, to our best knowledge, there are no reports on the causes of differences in colostrum quality between first-calving cows and cows with later parities, and this area requires further research. Because so few cows had records from multiple parities, the SE of genetic correlations was high. The majority of cows (51,605 cows with 67,206 entries) only had data from one parity. Colostrum quality as measured in the first parity might be regarded as an individual trait from Brix measured in subsequent parities, and it needs to be incorporated with later parity in selection indices aiming to improve lifetime colostrum quality. Although we observed a high genetic correlation between Brix2 and Brix3, this data format also made it challenging to build repeatability models. Due to this reason and many cows lacking repeatable colostrum records, we treated Brix1, Brix2, and Brix3 as distinct traits in subsequent analyses.

In previous studies, Kiser et al. [19] found only one SNP associated with Brix in BTA3 in a study with 345 Jersey cows in Texas. Moreover, seven promising candidate genes for Ig concentrations as colostrum quality traits (IgG, IgG1, and IgG2 in colostrum) in BTA11, BTA18, and BTA21 were reported by Lin et al. [20] in a study with 588 Chinese Holstein cows, but the p-value of the most significant SNP found by this research was 9.66 × 10⁻⁰⁶. There was no overlap with candidate genes discovered in this study and genes reported before, which may be due to the difference in sample size, statistical methods used, population characteristics, or the differences between Brix and IgG. However, neither ours nor previous studies identified highly significant SNPs, indicating the highly polygenic nature of colostrum quality traits in dairy cattle. In this context, WssGWAS may be able to more successfully identify genomic regions linked to the target traits [43,44].

Among overlapping genes, we considered CNR1 and ZXDC as promising candidate genes. These genes were only reported to be associated with disease and immunity traits in humans or mice. CNR1 encodes the cannabinoid receptor CB1 and was enriched by bta04723 (Retrograde endocannabinoid signaling). Cannabinoids (including endogenous and exogenous) are immunosuppressive. Lack of their receptors enhances cell-mediated immune responses in mice [45,46]. Through CB1, cannabinoid shows direct modulation of immune function in T cells, innate cells, and so on [47]. Endocannabinoids have been linked to both immunity and energy metabolism in dairy cows [48]. For example, endogenous cannabinoid levels in adipose tissue were elevated at the onset of lactation, as part of the metabolic adaptations in postpartum dairy cows [49]. ZXDC encodes a zinc-finger transcription factor modulating adaptive immunity. It is important for the transcription of the major histocompatibility complex class II molecules genes, which are involved with the adaptive immune system [50,51]. Since IgG determines the quality of colostrum, immune-related genes were chosen. As an Ig, IgG is produced by plasma cells, one of the key components of animal immune system. However, these genes are only the ones that affect immunity, and more research is needed to determine how they impact Ig and colostrum quality.

Our study used a large amount of data to initially reveal the inheritance patterns of colostrum quality traits in Chinese Holstein cows. Future research should additionally use direct IgG concentration data, as well as additional data from later parities (3+) or higher quality (28%+) and other potentially related traits such as health and longevity. Various candidate genes were identified to be associated with colostrum quality traits in dairy cows. However, there were no genes found for all studied traits. In addition, the candidate genes and genomic regions identified in this study still need to be validated in other independent populations and the biological mechanisms in which they are involved should be further investigated.

5. Conclusions

Colostrum quality is a moderately heritable trait in Chinese Holstein cattle, making it a possible target for genetic improvement. Colostrum quality in the first and later parities should be evaluated as distinct traits in breeding programs based on their low-to-moderate genetic correlations. Numerous genomic regions were identified on various chromosomes, but each explained a small proportion of the total additive genetic variance of colostrum quality, demonstrating that colostrum quality is a highly polygenic trait. Two genes (CNR1 and ZXDC) were considered as the main candidate genes influencing colostrum quality in Chinese Holstein cows and should be further investigated.