1. Introduction
Milk and dairy products are essential for human nutrition. Goat milk is rich in minerals, vitamins, and bioactive components, is easily digestible, and contains fewer allergic proteins than cow milk. These characteristics also suggest the possible use of goat milk for therapeutic purposes [
1,
2]. In Europe, the small ruminant milk industry is not widespread because of the low number of animals and insufficient milk volume in goats compared to cows. However, the number of goats used for milk production is growing due to expanding demand. In many countries, somatic cell count (SCC), as an indicator of milk quality, plays a role in the milk industry [
3,
4,
5]. A high SCC also negatively affects some flavour characteristics of cheese and ice cream due to a more intense caprine flavour [
5,
6,
7].
Many studies have investigated the relationship between genetic background and milk performance [
8,
9,
10,
11,
12]. Like cows, the heritabilities of milk yield in goats have been found to be low to moderate. E.g., a study in New Zealand (64,604 lactation records from 23,583 does distributed in 21 flocks) estimates heritabilities of 0.25 for milk yield, 0.24 for fat yield, 0.24 for protein yield, and 0.21 for SCS, suggesting the presence of useful heritable variation [
8]. Multiple trait selections for these traits could improve the milk revenue produced from successive generations of New Zealand dairy goats. To a similar extent, a study in Germany estimated heritabilities of 0.15–0.31 for milk yield, 0.21–0.34 for fat content, 0.26–0.50 for protein content, and 0.10–0.17 for the persistence of milk yield [
9]. The phenotypic and additive genetic correlations between the milk yield persistence and milk yield in kg were highly positive (0.52–0.72); similarly, the correlations between the protein and fat content were 0.45–0.55. The phenotypic correlations between the fat and protein content and the milk yield were negative, −0.13 to −0.26 and −0.21 to −0.36, respectively (n = 16,579 goats, 42,973 lactations). Others reported milk yield heritabilities of 0.10 to 0.29 (n = 529 goats, 15,509 milk yield test-day records) and a direct heritability for protein percentage of 0.441 (518 phenotypic records from the progeny of 48 sires and 131 dams) [
10,
11]. The genetic correlation between milk production and SCS can range broadly from −0.16 to 0.43, with a large standard error [
12]. Non-genetic factors have been analysed as well. In Alpine goats in Croatia, the herd explained 24% of the variability in daily milk yield, 12% of that in fat content, and 9% of that in protein content; values for the herd test day were 17%, 29% and 30%, and those for the permanent environment were 16%, 3% and 5%, respectively [
13].
The effect of major genes is the main research focus, similar to studies in other milking species. The consideration of the αS1-casein genotype may improve the model’s efficiency, translating into more accurate genetic parameters and breeding values [
14]. In another study, 48 SNPs in αS1, αS2, β, and κ casein genes were included in the evaluation of genetic parameters [
15]. Including genetic effects and relationships among these heritable biomarkers may improve the model efficiency, genetic parameters, and breeding values for milk yield and composition; this inclusion could also help optimise selection practices and profitability for components where technological application may be especially relevant for the cheese-making dairy sector [
15]. In addition to the study of candidate loci, the genomic approach in the genome-wide association study has also been applied for the detection of genetic regions of interest [
16]. They found a total of 43 genome-wide significant SNPs for lactation yields of milk (MY), fat (FY), protein (PY), and somatic cell score (SCS). A cluster of variants on chromosome 19 associated with MY, FY, and PY was identified, and another cluster on chromosome 29 associated with SCS. The most significant genomic window was located on chromosome 19, explaining up to 9.6% of the phenotypic variation for MY, 8.1% for FY, 9.1% for PY, and 1% for SCS [
16].
Previous studies demonstrated that the biochemical pathways in mammary glands related to the biosynthesis and secretion of lipids, lactose, and proteins are regulated by complex gene networks [
17,
18]. Due to the significance of fatty acid biosynthesis, polymorphisms in the following genes were the focus of this study: butyrophilin (
BTN1A1); acetyl-CoA carboxylase α (
ACACA); lipoprotein lipase (
LPL); and stearoyl-CoA desaturase (
SCD). BTN1A1 is a milk fat globule membrane protein that plays a crucial role in lipid secretion and milk production [
19]. The
BTN1A1 gene is located at chromosome 23, has eight exons and seven coding exons, respectively, and a transcription length of 3256 bps. The gene codes for 526 amino acid (AA) residues (NCBI gene 100860762). In the transcript, 219 variant alleles were described. ACACA is the primary regulatory enzyme of fatty acid biosynthesis; it catalyses acetyl-CoA conversion to malonyl-CoA [
18,
19]. Fatty acids are essential for forming cell membranes and are used to synthesize fat for storage in adipose tissue or secretion into milk by the mammary gland [
20]. The
ACACA gene is located at chromosome 19, consists of 52 exons, the transcription length is 6990 bps, and codes for 2329 AA residues (NCBI gene 100861224). In all, 3555 variant alleles were found in the transcript.
The
SCD gene, which is located on goat chromosome 26, has an important role in the cellular biosynthesis of monounsaturated fatty acids (MUFAs), because most of the conjugated linoleic acids are synthesised in the mammary gland by the action of SCD in circularizing vaccenic acid [
21,
22]. The
SCD mRNA has been identified by Bernhard et al., 2001 [
23] and Yahyaoui et al., 2003 [
24]. There are many SNPs that have been described and identified in exon 3, intron 3, intron 4, exon 5 and 6, and a deletion of a nucleotide triplet in the 3′UTR [
23,
24,
25,
26]. The
LPL gene is involved in the hydrolysation of triglycerides to glycerol and free fatty acids and in lipoprotein transportation [
27,
28]. It is synthesised in the mammary gland’s epithelial cells and influences the release of fatty acids in the mammary tissue [
29]. LPL enzymes are encoded by the
LPL gene, which consists of nine exons and eight introns, for a total of 3555 nucleotides. This gene encodes a protein containing 478 amino acids, JQ670882. A few SNPs have been described in goats: a missense mutation responsible for a Ser17Thr amino acid substitution at position 17 of the signal peptide (DQ370053:c.G50C), a C2094T (DQ370053) substitution in the 3′UTR of the gene, and a substitution ss522928251:C > T in intron 7 [
21,
30].
Our aim was to perform an association analysis of SNPs in the ACACA, BTN1A1, LPL, and SCD genes with milk production traits: daily milk yield (DMY), protein, and fat percentage (PP, FP), and SCC of goats on organic farms in the Czech Republic.
2. Materials and Methods
2.1. Ethical Approval
The experiment was carried out under Directive 2010/63/EU of the European Parliament and European Council of 22 September 2010, on the protection of animals used for scientific purposes.
2.2. Animals
In this study, a total of 590 animals were included. All individuals belonged to the two Czech national goat breeds: White Shorthaired (WSH) and Brown Shorthaired (BSH) goats or their crosses with other breeds. Thus, three breed groups were determined: purebred WSH, purebred BSH, and crossbreds of both breeds, where the proportion of WSH or BSH was 50% or higher. The experiment was performed on two farms that were located in the Czech Republic. Both farms specialize in organic goat milk production, which means that goats grazed and were fed only organic feed; no hormones, antibiotics, or similar substances were applied (except a form of veterinary treatment for individuals), no genetically modified organisms were included, and animals were kept under welfare conditions according to legislation in the European Commission Regulation 889/2008, and the European Commission Regulation 834/2007. The winter feed ration consisted of haylage at approximately 2 kg a day, hay ad libitum, and a grain mix, which was dosed during milking in the milking parlour with a total amount of 300 g a day. The summer feed consisted of grass at approximately 2 kg a day (loaded in the stable), hay ad libitum, and grain mix at 300 g a day. Goats were machine-milked twice a day.
2.3. Performance-Testing Database
Phenotype data of genotyped animals were obtained from the performance-testing database of the Czech-Moravian Breeders’ Association. In this study, the daily milk yield (DMY) in kg, milk protein (PP) and fat (FP) per cent, and somatic cell count (SCC) were considered. The analysis of DMY, PP, and FP was performed in the group of 590 goats belonging to the White Shorthair breed (n = 490), Brown Shorthair breed (n = 76), and crossbreeds between WSH, BSH, and Saanen (n = 24). Purebred individuals were approximately 75–100% pure. The goats were sired by 37 bucks; the average number of daughters per buck was 8 (minimum 1, maximum 35). They were on the first to eleventh lactation. In all, 8640 milking records from two farms were analysed. At farm A, 2241 repeated-milk records from 279 dairy goats in 2013, 2014, and 2016 were collected. Milk records from farm B included 311 dairy goats with 6399 milk records between 2010 and 2017. Milk samples were collected repeatedly during the milking seasons. There were 2 to 49 repeated records per goat, on average, there were 14 records (approximately four milk controls per year). Milk samples for DMY, PP and FP analysis were collected throughout the whole year as follows: January (n = 571), February (n = 533), March (n = 735), April (n = 1061), May (n = 1161), June (n = 1203), July (n = 1035), August (n = 1069), September (n = 950), October (n = 74), November (n = 57), and December (n = 191).
Samples from only one farm were analysed for somatic cell count, n = 146 goats, that is, White Shorthair goats n = 100, Brown Shorthair goats n = 38, and crossbreeds n = 8. The goats were sired by 27 bucks, and the average number of daughters was 5 (minimum 1, maximum 19). Milk samples were taken during the third to eleventh lactation. The number of repeated records for SCC was 857. Data were collected during 2016 (n = 728) and 2017 (n = 129). The analysis was conducted repeatedly throughout lactation. There were 2–11 repeated records per goat, on average 5 samples per goat, with approximately three controls per year. Milk samples for SCC analysis were collected throughout the milking season as follows: March (n = 26), April (n = 148), May (n = 148), June (n = 128), July (n = 163), August (n = 124), and September (n = 120).
2.4. DNA Extraction and SNP Genotyping
Blood samples from 590 animals (5 mL of each) were collected from the jugular vein and preserved in 0.5 mM EDTA (pH 8.0). Genomic DNA was extracted from blood using GeneAll, Exgene
TM, and a Clinic SV mini isolation kit (GeneAll Biotechnology cp., Ltd., Seoul, Korea; Bohemia Genetic Ltd., Prague, Czech Republic) according to the manufacturer’s recommendations. The SNPs analysed in our study are described in
Table 1. SNPs in the
BTN1A1 and
LPL genes were genotyped according to previously described methods [
30,
31,
32].
Individual SNPs of the
ACACA and
SCD loci were detected by primer extension analysis with the SNaPshot Commercial Kit (Applied Biosystem, Foster City, CA, USA). Primers used for the PCR, extension analysis and electropherogram of the SNaPshot product along with the GeneScan
TM 120LIZTM size standard are given in
Supplementary Table S1.
For the PCR reaction (512 bp fragment) of 3 SNPs in the 5′UTR [
33], of
ACACA locus, we used the set of primers, designed on the basis of the ovine sequence AJ292286 [
34]. The PCR assay was performed in a 10 µL reaction mixture, consisting of 2 µL genomic DNA (10–100 ng), 1× PPP Master Mix (Top Bio Ltd., Prague, Czech Republic), and 0.5 µL (10 pmol/µL; 0.01 mM) of each primer (GENERI Biotech Ltd., Prague, Czech Republic), and sterile water up to volume. Thermal cycling conditions are presented in
Supplementary Table S2.
A 536bp fragment of the
SCD locus, at region exon3 and intron 3, was amplified by PCR with the following set of primers: F: 5′-TCCTAAgCTTATTCCAgCCCC-3′and R: 5′-gCCAgTCACTCAgAAgTACCC-3′, designed on the basis of the GenBank goat sequence (GenBank AH011188.2; AF422168.1) using Primer 3 software [
35]. PCR assay was performed in a 20 µL reaction mixture consisting of 2 µL genomic DNA (10–100 ng), 1× PPP Master Mix (Top Bio Ltd., Prague, Czech Republic), of each primer (GENERI Biotech Ltd., Prague, Czech Republic) and sterile water up to volume. Thermal cycling conditions are presented in
Supplementary Table S2.
The presence of fragments obtained in this phase was confirmed by gel-electrophoresis stained with ethidium bromide. The obtained PCR products were purified by using 1 unit of FastAP Thermosensitive Alkaline Phosphatase and Exonuclease I (Fermentas, Ltd., Prague, Czech Republic) to remove unwanted subproducts and incubated at 37 °C for 60 min, followed by 15 min at 85 °C.
The PEA assay utilises internal unlabelled primers which bind to a complementary PCR-generated template in the presence of AmpliTaq DNA Polymerase and fluorescently labelled ddNTPs. The polymerase extends the primer one nucleotide, adding a single ddNTP to its 3′ end. Primers were designed to allow size and colour discrimination between the different alleles (
Table S1) and were optimised to be used simultaneously.
The single-base extension (SBE) reaction for ACACA locus was performed in a reaction mixture with final volume of 5.0 µL, containing 1.5 µL of purified multiplex PCR product, 1× extension primer mixture (0.01 mM concentrations): K-ACACA (1206) = 0.5 µL, K-ACACA (1255) = 0.5 µL, K-ACACA (1322) = 0.5 µL, 1.5 µL deionized water, and 2.0 µL of SNaPshot Multiplex Ready Reaction Mix (Applied Biosystems, Foster City, CA, USA). The single-base extension (SBE) reaction for SCD locus in positions EX3_15G > A, IVS3+46 C > T, IVS3+55A > G, EX3_68A > G and IVS3+105A > G was performed in a reaction mixture with a final volume of 6.0 µL, containing 1.5 µL of purified multiplex PCR product, 1× extension primer mixture (0.01 mM concentrations): K-EX3_15G > A = 0.5 µL, K-EX3_68A > G = 0.5 µL, K- IVS3+46C > T = 1.0 µL, K-IVS3 + 55A > G = 0.5 µL K-IVS3 + 105A > G = 0.5 µL, 0.7 µL deionized water, and 2.3 µL of SNaPshot Multiplex Ready Reaction Mix (Applied Biosystems, Foster City, CA, USA). Thermal cycling consisted of 25 cycles of denaturation at 96 °C for 10 s, primer annealing at 50 °C for 5 s, and primer extension at 60 °C for 30 s (Biometra Thermoblock: 050-801 TGradient 96, Biometra, Goettingen, Germany).
For electrophoretic detection, 0.5 µL of purified multiplex SBE reaction was mixed with 0.5 µL of GeneScan-120 LIZ size standard (Applied Biosystems, Foster City, CA, USA) and 9.0 µL of Hi–DiTMFormamide (Applied Biosystems, Foster City, CA, USA), following denaturation step at 95 °C for 5 min and analysed by capillary electrophoresis using the Applied Biosystems
® 3130 Genetic Analyzer, an E5-Matrix Standard Set DS-02, a 36 cm capillary, and POP7 polymer (Applied Biosystems, Foster City, CA, USA). The results of genotyping were analysed and evaluated using GeneMapper v 3.5 software (Applied Biosystems, Foster City, CA, USA). The dye colour of the fragment was used to identify the nucleotide of interest (
Figure S1).
2.5. Statistical Analysis
The dataset was edited, and unreliable data were removed. SCCs less than 13 and more than 9998 were removed from the analysis. To achieve an approximately normal distribution, the SCC was log-transformed into somatic cell score (SCS). The transformation was performed as follows:
where SCC is somatic cell count, which is expressed in thousands per 1 mL of milk.
Hardy–Weinberg equilibrium was tested by the χ
2 test. The effect of gene polymorphisms on milk performance traits and SCS was analysed using the PROC MIXED procedure of SAS with repeated measurements [
36]. Tested effects were considered statistically significant at
p < 0.05, but biological importance was also considered. The following linear model was used for all traits (DMY, PP, FP, SCS):
where Y
ijklmno = DMY, FP, PP, SCS; G
i = fixed effect of the genotype (class effect i = 1, 2, 3); HY
j = combined fixed effect of herd-year (class effect j = 1, …, 11); month
k = fixed effect of the month of the year of milking (class effect l = 1, …, 12); lac
l = fixed effect of the lactation order (class effect l = 1, …, 9 for DMY, FP and PP or l =1, …, 9 for SCS); breed
m = fixed effect of the breed (class effect m = 1, …, 3); sire
n = random effect of the father of the goat; goat
o= permanent environment of the goat (repeated measurement); and e
ijklmno = random residual effect.
The post hoc comparison was performed by Scheffe´s method. A Bonferroni correction for multiple comparisons was applied to all significant associations. The correction factor was derived from the number of SNPs tested. The significance threshold (p < 0.05 and p < 0.01) was divided by the number of tests. Thus, Bonferroni-corrected significance levels of 0.05/13 = 0.004 and 0.01/13 = 0.0008 were applied.