Gene polymorphisms influencing yield, composition and technological properties of milk from Czech Simmental and Holstein cows

Objective The aim of the study was to evaluate the influence of polymorphic loci and other factors on milk performance and the technological properties of milk. Methods The analysis was performed on Simmental and Holstein cows in field conditions (n = 748). Milk yield in kg, fat and protein percentage and yield were evaluated. Technological properties were evaluated by milk fermentation ability, renneting, and an alcohol test. Polymorphisms in the acyl-CoA diacylgycerol transferase 1 (DGAT1), leptin (LEP), fatty acid synthase (FASN), stearoyl CoA desaturase 1 (SCD1), casein beta (CSN2), casein kappa (CSN3), and lactoglobulin beta genes were genotyped, and association analysis was performed. Results The DGAT1 AA genotype was associated with higher milk, protein and fat yields (p<0.05). The MM genotype in the LEP gene was associated with a lower protein percentage and the W allele with a higher protein percentage (p<0.05). In cows with the FASN GG genotype, the protein percentage was higher, but the A allele was associated with higher milk, protein and fat yields than the G allele. The TT genotype in SCD1 was associated with the lowest milk, protein and fat yields and with the highest milk protein percentage (p<0.01). The T allele had higher values than the C allele (p<0.05) except for fat percentage. The genotype CSN3 AA was associated with a significantly heightened milk yield; BB was associated with a high protein percentage. The effect of the alleles on the technological properties was not significant. The CSN2 BB genotype was associated with the best alcohol test (p<0.01), and the renneting order was inverse. Milk from cows with the CSN2 A1A1 genotype was best in the milk fermentation ability. CSN3 significantly affected the technological properties. Conclusion The findings revealed the potential of some polymorphic loci for use in dairy cattle breeding and for the management of milk quality. In field research, the pivotal role of farms in milk yield, composition and technological properties was confirmed.


INTRODUCTION
Milk yield and composition substantially impact the economics of dairy farms. For milk manufacturing, in addition to the percentages of fat and protein and the microbial quality, the technological properties of milk are important, as cheese has become a very important product in the dairy industry and enzymatic coagulation of milk is a crucial step in cheese making process [1]. The influence of diet on the cheesemaking properties of milk is often analyzed [2]. Other impacts on technological properties, such as the housing system or the stage of lactation, have also been the focus of studies [3,4].

Animals
All experiments were performed in accordance with relevant guidelines and regulations recommended by the Ministry of Agriculture of the Czech Republic. All animal experiments were under supervision of the Institutional Animal Care and Use Committee of the Faculty of Agriculture of the South Bohemia University, where the experiment was carried out, approval number 22036/2019MZE18134. DNA was extract ed noninvasively from milk samples.
The analysis was performed in cows of the Czech Simmen tal (part of the Simmental group) and Holstein breeds and their crosses; the cows were kept in the Czech Republic. As the crossbreds of Holstein and Simmental with small propor tions of Ayrshire are common in herds in the Czech Republic, they were included into our field research as well. The numbers of purebred and crossbred cows are given in Supplementary  Table S1. Cows were kept by five companies in free housing and fed with maize silage, grass silage, hay and feed concen trates yearround. The ratios differed among companies in terms of the share of constituents and their quality. The cows calved from 2015 through 2017. The 1st lactation was record ed for 748 cows, and the 2nd was also recorded for 660 of those cows. The mean milk yield was 8,036 kg in the 1st lac tation and 8,722 kg in the 2nd lactation. The fat percentages were 4.12 and 4.12, the crude protein percentages were 3.46 and 3.48, the fat yields were 329.8 kg and 358.0 kg, and the crude protein yields were 274.8 kg and 301.3 kg for the first and second lactations, respectively.
The technological properties (milk fermentation ability, renneting measured by classical procedure and by nephe lometry, ethanol test) of milk samples from 242 cows were examined. Of these cows, 81 were sampled once, 86 twice, 53 three times, 16 four times, and 6 five times. The cows were sampled throughout the course of the year.

Genotyping
Milk samples were individually collected, and DNA was iso lated using the DNA/RNA extractor MagCore HF16 Plus (RBC Bioscience, New Taipei, Taiwan). Isolation was per formed according to the manufacturer's instructions using the MagCore DNA Whole Blood Kit and MagCore Genomic DNA Tissue Kit (RBC Bioscience, Taiwan). The quality and quantity of the isolated DNA were verified by electrophoresis and spectrophotometry.
Genotyping of all loci was performed by the polymerase chain reaction and restriction fragment length polymorphism (PCR/RFLP) method. DGAT1 gene alleles A (alanine) and K (lysine) were genotyped according to the methods of Kuhn et al [14]; LEP gene alleles M and W according to Buchanan et al [15]; FASN gene alleles A and G according to Roy et al [16]; CSN2 gene alleles A and B according to Medrano and Sharrow [17]; and alleles A 1 and A 2 according to Miluchová et al [18]. CSN3 gene alleles A, B, C, and E were analyzed according to the methodology of Barroso et al [19]; LGB alleles A and B according to Strzalkowska et al [20]; and SCD1 gene alleles C and T according to Inostroza et al [21]. The primer sequences are given in Supplementary Table S2.
The resulting genotypes were electrophoretically deter mined, and genotype and allelic frequencies were calculated. To evaluate the HardyWeinberg Equilibrium, the differences between the observed and expected frequencies of the geno types were tested using a χ 2 test with the significance level p<0.05 and p<0.01. Supplementary Table S3 gives the fre quencies of genotypes and alleles of all genotyped Simmental, Holstein and crossbred cows.

Milk performance, composition and analysis of technological qualities
Data on milk performance were collected from the milk re cording breeder´s database. Milk yield in kg, fat and crude protein percentage, and fat and crude protein yield in kg were evaluated. Milk composition (fat and crude protein contents) was determined in breeder milk laboratories of the Czech Moravia Breeders Association using infrared spectroscopy (Foss Electric, Foss A/S, Hilleroed, Denmark; and Bentley Instruments, Chaska, MN, USA) instrumentation. These laboratories are accredited to the ISO standard (CSN EN ISO/ IEC 17025) for official milk performance analysis in the Czech Republic and are working under the ICAR (International Committee for Animal Recording) umbrella (ICAR certificate to Czech Moravian Breeders' Corp, Hradistko; Accredited Milk Laboratory Bustehrad, Czech Republic, for identifica tion of dairy cattle, production recording in dairy cattle, genetic evaluation, milk laboratory operation, linear classification/ scoring, and data processing), regularly taking part in rele vant proficiency testing. Analytical instruments were regularly monthly calibrated according to the reference method results (extraction by the Röse-Gottlieb method for fat content [22] and distillation and titration according to the Kjeldahl method for crude protein content (total nitrogen content ×6.38) [23]). Technological properties were evaluated by a milk fermen tation ability test, renneting was measured subjectively and instrumentally, and an ethanol test was performed.
The milk ethanol stability was determined by milk titra tion (5 mL) with 96% ethanol until the first precipitation flakes of milk protein were visible and is reported as ml of alcohol. This procedure was modified according to Horne [24].
The milk fermentation ability of the yogurt test was carried out according to the Czech milk industry standards. A sam ple of raw milk (50 mL) was heated at 85°C for 5 min and cooled at 43°C±2°C. Subsequently, the sample was inoculated with 2 mL of the thermophilic lactic culture YC18040FLEX (Chr. Hansen, Horshholm, Denmark; Streptococcus thermophilus, Lactobacillus delbrűeckii subsp. lactis, and L. d. subsp. bulgaricus). The inoculated sample was incubated at 43°C for 3.5 hours. The result was expressed as the titration acidity of the yogurt in mL of 0.25 mol×L 1 NaOH×100 mL 1 (or the socalled SoxhletHenkel degree) [25,26].
Rennetability (classical procedure) was determined dur ing the tempering (35°C) of a defined milk volume after the addition of rennet (1% vol.) by measuring the time (rennet coagulation time RCT) until the first flakes of lactoproteins formed (beginning of coagulation). Rennetability was also determined by using nephelometry (turbidimetry measure ment) to assess the milk coagulation time (ML -2 analyzer). This is the use of the optical method (NEF, Nephelo turbi dimetric milk coagulation sensor ML -2) to evaluate the intensity of the socalled diffusely scattered Tyndall light on dispersed particles (coagulating lactoprotein flakes) [26,27].
The milk ethanol stability, milk fermentation ability and milk rennetability are not introduced by an official standard as technological property in world literature references, but they are known according to citations in the scientific litera ture. These procedures were modified according to literature sources cited.

Statistical analysis
Statistical analyses were performed using SAS (SAS 9.3, SAS Institute, Cary, NC, USA). Descriptive statistics for milk yield in kg, protein and fat percentages and protein and fat yield in kg during the first and second lactations are given in Sup plementary Table S4. Descriptive statistics for the indicators of the technological quality of milk, i.e., the milk fermentation ability, rennetability assessed subjectively and instrumentally, and alcohol test are given in Supplementary Table S5. For the descriptive statistics and genotype frequencies in Supplemen tary Table S4, each record was assessed as a separate entry; when two lactations were recorded for the same cow, it was included twice. Similarly, for Supplementary Table S5, when a cow was measured repeatedly, the genotype was included repeatedly as well.
The data set contained repeated measurements per cow. Repeated measurements were obtained for the first and sec ond lactation for milk performance traits. For technological quality, measurements were obtained several times over the course of two consecutive lactations. To analyze the influence of polymorphisms on milk yield and technological quality, the MIXED procedure of the SAS system with repeated mea surements and the least squared mean method were used to compare genotypes. The models were developed as follows.
For milk performance traits, the following mathematical model was used: Y ijk = µ+gen i +lac j +anim k +e ijk www.animbiosci.org 5 Čítek et al (2021) Anim Biosci 34 :2-11 where Y ijk = milk performance trait; μ = population mean; gen i = fixed effect of the genotype (class effect i = 1, 2, 3); lac j = fixed effect of the lactation order (class effect j = 1, 2); anim k = random effect of the animal; and e ijk = random residual effect.
Different mathematical models were used to determine the technological quality of milk: Yogurt ijklmn = µ+gen i +farm j +protein k +casein l +lacs m +anim n +e ijklmn where Yogurt ijklmn = yogurt test values; μ = population mean; gen i = fixed effect of genotype (class effect i = 1, 2, 3); farm j = fixed effect of farm (class effect j = 1, 2, 3, 4, 5); protein k = fixed effect of protein percentage content in milk; casein l = fixed effect of casein content in milk; lacs m = fixed effect of lacta tion stage in days; anim n = random effect of the animal; and e ijklmn = random residual effect.
Rennetability ijklmn = μ+gen i +farm j +protein k +NFS l +season m +anim n +e ijklmn where Rennetability ijklmn = rennetability assessed subjectively or instrumentally; μ = population mean; gen i = fixed effect of genotype (class effect i = 1, 2, 3); farm j = fixed effect of farm (class effect j = 1, 2, 3, 4, 5); protein k = fixed effect of protein percentage content in milk; NFS l = fixed effect of not fat solids content in milk; season m = fixed effect of season (class effect m = 1, 2, 3, 4)*; anim n = random effect of the animal; and e ijklmn = random residual effect. * The fixed effect of season was created as a combination of three months according to natural weather conditions, temperature, pasture quality, etc. A year was divided into four seasons: 1 = December, January, Feb ruary; 2 = March, April, May; 3 = June, July, August; and 4 = September, October, November.
The effect of alleles on milk production traits and the tech nological quality of milk was computed using the following mathematical model: where Y ij = observed trait; μ = population mean; allele i = fixed effect of allele (class effect i = 1,2); anim j = random effect of the animal; and e ij = random residual effect.
For post hoc comparisons, the TukeyKramer test was used [28].

Milk yield and composition
In the DGAT1 gene, the genotype AA and allele A, which codes for alanine, had a higher frequency than the genotype KA and the K allele, which codes for lysine (Supplementary  Table S3); the homozygous genotype KK was not found at all. Other researchers found similarly unbalanced frequencies.
In Israeli Holstein cows, the frequency of the K allele was reported to be 0.09 overall and 0.16 in sires [29]; in another study, the A allele had the highest frequency in dairy breeds, with the exception of Jersey [30]. Similarly, low frequencies of the K allele and KK homozygous genotype were found in Simmentals [31]. The A allele was confirmed repeatedly to be associated with higher milk, fat and protein yields, and its frequency in intensively selected populations increases due to indirect selection [14,32]. In our analysis, cows with the AA genotype outperform the heterozygous ones signifi cantly in milk, protein and fat yields (Table 1). Additionally, when the effects of alleles are evaluated, A is advantageous but not significantly so ( Table 2). This result is generally in agreement with previous findings and confirms our previous finding in German Holsteins regarding the trend of increas ing frequency of the alanine variant [32,33].
Additionally, for the LEP gene, the MM genotype domi nated (Supplementary Table S3). In Holstein cows, a reverse order of genotypes was also published [34]. MM homozy gous cows had a lower protein percentage, and the difference between MM and MW was significant (Table 1). Allele W positively and significantly influenced the protein percentage ( Table 2); differences in other indicators of milk performance were nonsignificant.
For the FASN gene, the protein content was slightly but significantly higher in GG homozygous cows. The A allele had significantly higher milk yield than the G allele, which resulted in significantly higher protein and fat yield. The dif ferences in fat and protein percentages between alleles were negligible and nonsignificant ( Table 2). The frequencies of allele G were markedly higher than those of A, which does not correspond fully with the differences between alleles in terms of performance. However, considering both genotypes and alleles, the performance differences were low, which may explain the differences in frequency.
The TT homozygous genotype in the SCD1 gene was sig nificantly associated with the lowest milk, protein and fat yields and with the highest protein contents ( Table 1). The analysis of allele associations showed superiority of the T allele in all characteristics except fat content. The differences were not high but significant ( Table 2). The differences among geno types hint at intermediate heredity.
For the CSN2 gene, the differences among genotypes were not significant. The B allele had significantly higher milk yield and therefore protein yield than A. In fat yield, the pvalue was near the significance threshold. The differences in con tents were low and nonsignificant (Tables 1, 2). Similarly, for the A 2 and A 1 genotypes, the effect was nonsignificant. Allele A 2 was significantly better in terms of milk, protein and fat yields. The results of Ozdemir et al [35] indicated that none of the CSN2 variants provide an advantage.
Genotype AA in the CSN3 gene was significantly associ LGB, lactoglobulin beta. * Significant at p < 0.05; ** significant at p < 0.01. a,b Different letters between genotypes in the same column represent significant differences at p < 0.05. A,B Different letters between genotypes in the same column represent significant differences at p < 0.01. X Differences between CSN3 genotypes AA and BE in the protein percentage are significant at p < 0.05. Y Differences between CSN3 genotypes EE on the one hand and AB, AE, BE on the other hand in the protein percentage are significant at p < 0.05. LGB, lactoglobulin beta. * Significant at p < 0.05; ** significant at p < 0.01. ated with high milk yield. BB homozygous cows had milk with a significantly higher protein percentage. Although the highest value was associated with the EE genotype, there were only two cows with a total of three lactations with this geno type, making it a minor consideration; similarly, for the BC genotype, there were two cows with four lactations. However, there were 21 cows of the BE genotype with 40 lactations, and they had significantly higher protein percentages comparing with the AA genotype ( Table 1). The lowest protein content was found in AA homozygous cows, and the difference was significant. Thus, the positive influence of the B variant on the protein content was repeatedly shown. This was confirmed when evaluating the effect of alleles, specifically, the following significant effect was observed: E>B and B>A. However, the CSN3 genotypes were not significantly associated with pro tein yield. Additionally, BB and B performed significantly better than AA and A in fat percentage. The differences in fat yield were not significant. Our findings on the prevalence of the AA genotype and its effects on milk yield agree with other results found in Sim mentals [36]. The authors also found the BB genotype to be associated with the highest protein percentage, but the fat percentage and yield were highest in milk from AA cows. In Czech Simmental cows, significant differences were reported among genotypes in daily milk yield, but the differences in protein and fat percentages were nonsignificant [37]. Ozdemir et al [35] conclude their review and metaanalysis by stating that the CSN3 genotypes are ranked BB>AB>AA in terms of protein content and that the B allele could be considered a marker to improve milk protein content. The prevalence of the BB genotype in relation to protein yield was not always obvious. Additionally, for fat content, the BB genotype was better. They report that the associations of genotypes and al leles with milk yields were not significant. These findings are in general agreement with our results.
The AA genotype of the CSN3 gene is usually the most fre quent in both BlackandWhite and Simmental cattle [20,36]. The frequencies of the BB genotype and B allele in our cows, both Holstein and Simmental, were rather low, which is also consistent with the frequencies found by other authors in Czech Simmental [6]. However, one other group of authors also found that, in Czech Simmental, the most frequent gen otype was AB (0.487), and the frequency of the B allele was high (0.418) [37]. The frequency of the E allele (0.030) was the same as in our Simmental group (0.036). Apparently, there is leeway for breeding, and many Czech breeding companies report the genotype of the CSN3 gene for the sires in their catalogues. However, changing genotype frequencies is a long distance run.
For the LGB gene, the AB genotype was significantly as sociated with higher milk yield than in both homozygous genotypes, resulting in higher protein and fat yields. The rank of genotypes may indicate the effect of heterosis. Allele B out performed A in milk, protein and fat yields, but the differences in the contents were not significant (Tables 1, 2). Additionally, other authors found significantly higher milk, protein yield and fat contents in Simmental cattle and higher fat yields (which were nonsignificant) in AB heterozygous cows [36].
The other factors potentially affecting milk performance were evaluated. Farm, breed and lactation order were tested in a general linear mixed model. The milk yield, fat yield and protein percentage were significantly influenced by all the fac tors. Fat content was influenced by farm, and protein yield was influenced by farm and lactation order. Thus, the impor tance of the effect of farms, i.e., specific stable, management, nutrition, veterinary care, milking, etc. was emphasized, even if some polymorphisms showed a significant association with milk performance.

Milk technological characteristics
The testing of milk fermentation ability, renneting and ethanol stability was the final goal of our analysis. The topic is rele vant because genetic background exerts a strong influence on the cheesemaking properties of milk, largely due to ge netic polymorphisms in the major milk protein genes [38]. In our analysis, the effect of alleles was not significant with the exception of CSN2, with the B allele outperforming the A allele (p<0.01) in terms of milk fermentation ability (yo gurt test) ( Table 2). The milk of KA heterozygous cows in the DGAT1 gene significantly exceeded that of homozygous AA cows in milk fermentation ability ( Table 3). The LEP gene had no significant effect, but the differences among geno types in the alcohol test showed rising values in the order MM>MW>WW with significance differences of MM and MW vs WW at p<0.05. The differences among FASN geno types were not significant. For the SCD1 gene, the milk of TT cows was associated with significantly poorer performance in terms of renneting.
Certainly, the effects of polymorphous variants of milk protein genes are in focus. For the CSN2 gene, the BB geno type had the best heat stability of milk as measured by the ethanol test (p<0.01), which is important for ultrahigh tem perature milk production. However, with regard to renneting, the order was reversed. The A 1 A 1 genotype was significantly associated with the best milk fermentation ability and was not significantly associated with renneting. Poulsen et al [39] refers to the negative association of A 2 with coagulation. Ac cording to our results, it is difficult to describe the preferable genotype or allele in the CSN2 gene.
Kappacasein is the gene most often examined, as its influ ence on technological properties has been confirmed repeatedly. In our analysis, the rennetability measured instrumentally was significantly affected (p<0.05). Genotype BC was asso ciated with the best milk, but only four measurements were performed; because of the high standard errors, this associa tion is of limited importance. The BB genotype was associated with significantly better renneting than the AA genotype but not the AB genotype (Table 3). Genotypes with the A and E alleles (AE, AA, BE) were associated with the poorest ren netability. A probable explanation for the differences in milk protein coagulation is the changes in the primary amino acid sequence between the A and B variants of the kappa casein as a protective factor for raw milk casein micelles. The B vari ant of kappa casein differs from the A variant by amino acid substitutions at two positions: 136th, replacing threonine with isoleucine, and 148th, replacing asparagine with ala nine. These changes in the amino acid sequence of the B variant may interact positively with the action of the rennet enzyme that starts cleavage of the casein molecule between the 105th and 106th amino acids of the peptide chain, i.e., not far from the changed amino acids.
For the milk fermentation ability, the pvalue was close to the significance threshold. The ethanol test resistance was the best in milk from the cows with CSN3 BB genotypes. The dif ferences in these scores between the BB genotype on one hand and AA and AB genotypes on the other hand were significant at p<0.01. The difference BB>AE was significant at p<0.05. Overall, the advantage of the CSN3 BB genotype was con firmed, but the B allele was not significantly better than the others (Table 2). Better properties for the BB genotype have also been reported by other authors [39,40], who stated a positive effect of CSN2 B and CSN3 B. They hinted at the ad ditive genetic variation of milk coagulation and the possibility of selective breeding for variants associated with superior milk coagulation.
Lactoglobulin beta is the most important whey protein. In our analysis, the effect of genotype on ethanol number was significant at p<0.05, and BB was associated with a better value than AB at p<0.01.
When evaluating other factors, the farm was found to sig nificantly affect all parameters of technological quality. Protein percentage was associated with the milk fermentation ability and rennetability scores, while fat percentage was associated with ethanol test performance. Nonfat solid content was as sociated with renneting, while somatic cell count was associated with renneting measured subjectively. The content of casein, LGB, lactoglobulin beta. * Significant at p < 0.05; ** significant at p < 0.01. a,b Differences between genotypes with different letters in the same column are significant at p < 0.05. A,B Different letters between genotypes in the same column represent significant differences at p < 0.01. X Differences between CSN3 genotypes AB on the one hand and AA and BE on the other hand in renneting assessed instrumentally are significant at p < 0.05. Y Differences between CSN3 genotypes BE and BC in renneting assessed instrumentally are significant at p < 0.05.