MATERIALS AND METHODS

The Reserva Natural El Hatico provided 94 adult animals (20 males and 74 females) used in this study. Inclusion criteria were the lowest degree of relationships and clinical health. El Hatico is located in the municipality of El Cerrito, (Valle del Cauca, Colombia). At an altitude of 1,000 meters above sea level (masl), with a mean annual temperature of 24 °C, 75% humidity and 750 mm of annual rainfall, corresponding to a tropical dry forest (bs-T), according to the Holdridge classification system (Molina et al. 2016). In addition, 10 animals of the Hartón del Valle (HDV) breed from the Agricultural Laboratory of the Mario González Aranda farm belonging to Universidad Nacional de Colombia, Palmira Headquarters, were used as a reference population.

RESULTS AND DISCUSSION

Of the four polymorphisms present in the CSN1S1 gene, the SNPs rs433385179 and rs132656458 were monomorphic. This meant that only the CSN1S1*B and *C alleles were found in both breeds, with the former being the most common (Table 1). All SNPs were in HWE. Genotypic frequencies did not differ significantly between sexes (Figure 1A) or generations (Figure 2A).

Table 1: Allele frequencies of SNPs located in the CSN1S1 gene and allele and genotypic frequencies of protein variants in the αS1-CN in the breeds studied.

Figure 1: Genotypic frequencies of CSN1S1 (A), CSN2 (B), CSN1S2 (C), CSN3 (D), and LGB (E) genes by sex in the Lucerna breed.

Figure 2: Genotypic frequencies of CSN1S1 (A), CSN2 (B), CSN1S2 (C), CSN3 (D), and LGB (E) genes by generation (generation 1: cattle born before 2010; generation 2: animals born between 2010 and 2015; generation 3: individuals born between 2016 and 2021) in the Lucerna breed.

Ten variants (*A, *B, *C, *D, *E, *F, *G, *H, *I and *J ) are recognized in the α _S1 -CN protein (Meier et al. 2019), from which only three (*B, *C and *D) can be reconstructed with the SNPs present in the GeneSeek® Genomic Profiler™ Bovine 150 K chip, despite the highly polymorphic nature of the gene (Ahmed et al. 2017), but this is explained by the breeds used to develop the chip. The CSN1S1*B allele is reported to be the most frequent in several Bos taurus breeds, including Angler (1.0), Eringer (0.77), German Red Pied (0.93), German Yellow (1.0), Highland Cattle (1.0), Hinterwälder (0.93), Hungarian Grey Steppe (0.73), Jersey (0.68), Limpurger (1.0), Pinzgauer (0.74), Sarabi (0.54), Shorthorn (1.0) and Vorderwälder (0.87) (Gallinat et al. 2013). In the Holstein breed, the frequency of this allele is reported to be higher than 0.95 (Ardicli et al. 2018; Meier et al. 2019) and in the Simmental breed, 0.90 (Čítek et al. 2023). In the Colombian creole breed Blanco Orejinegro (BON), the frequency of the *B allele reached 0.80 (Hernández-Herrera et al. 2024). On the other hand, in Bos indicus breeds, the CSN1S1*C allele is the most common (Butana=0.66, Gir=0.50, Golpayegani=0.48, and Sistani=0.88) (Gallinat et al. 2013).

In the LUC breed, the *BB genotype was the most common (0.774). In contrast, the *BC genotype was the most frequent in the HDV breed (0.500) (Table 1). It has been reported that animals with the homozygous *BB genotype have higher milk production and better milk quality, as it contains higher percentages of fat and protein (Ardicli et al. 2018; Mohan et al. 2021; Čítek et al. 2023).

Of the 14 SNPs evaluated in the CSN2 gene, 11 were fixed in the LUC breed and 13 in the HDV breed. This meant the presence of *A ¹, *A ², *B, and *F variants of the β-CN protein in the LUC breed and only *A ¹ and *A ² variants in the HDV breed (Table 2) 15 variants (*A ¹, *A ², *A ³, *B, *C, *D, *E, *F, *G, *H ¹, *H ², *I, *J, *K, and *L) have been reported for this protein (Meier et al. 2019). The frequency of the reference allele for this gene (CSN2*A ¹) was low, highlighting the high frequency of the CSN2*A ² variant in both breeds. This is in agreement with that reported for the breeds Eringer (0.54), German Yellow (0.52), Highland Cattle (0.81), Hinterwälder (0.79), Hungarian Grey Steppe (0. 67), Jersey (0.63), Limpurger (0.77), Pinzgauer (0.65), Sarabi (0.65), Hinterwälder (0.79), Hungarian Grey Steppe (0.63), Limpurger (0.77), Pinzgauer (0.65), Sarabi (0.75) and Shorthorn (0.57) (Gallinat et al. 2013), Holstein (0.55) and Jersey (0.73) (Chessa et al. 2020), Karan Fries (0.60) (Mohan et al. 2021) and in the BON breed (0.51) (Hernández-Herrera et al. 2024). However, this variant is more common in B. indicus breeds (Ahmed et al. 2017; Gallinat et al. 2013). The CSN2*I allele has been reported at low frequency in the Black Pied (0.017) and Holstein (0.059) breeds from Germany (Meier et al. 2019), and, as reported in this study, the *B and *F alleles were found at low frequency in the BON breed (Hernández-Herrera et al. 2024). On the other hand, Padilla-Doval et al. (2021), who studied the Holstein breed in Colombia, reported that the *B allele was the most common.

Table 2: Allele frequencies of SNPs located in the CSN2 gene and allele and genotypic frequencies of protein variants in the β-CN in the breeds studied.

All SNPs analysed in the CSN2 gene were in HWE. No differences in genotypic frequencies were found between the sexes (Figure 1B), but differences were found for the *A ² A ² genotype in the third generation (P<0.05), which shows an increase in frequency with each generation (Figure 2B). The same trend of increase has been reported in the BON breed (Hernández-Herrera et al. 2024). The importance of the high frequency of this variant (β-CN*A ²) is understood when it is considered that the intestinal digestion of milk protein releases the bioactive peptide β-casomorphin-7 in animals with the β-CN*A ¹ genotype. This peptide has been implicated in the development of several diseases (Chessa et al. 2020; Villalobos-Cortés et al. 2023) and in the increase of both milk allergy and lactose intolerance (Muntean et al. 2022). In contrast, antithrombotic, antihypertensive, immunomodulatory, anticancer, and antimicrobial activities are recognized, as well as bioactive peptides resulting from protein digestion of the β-CN*A ², *A ³, *D, *E, *H2, and *I variants (Muntean et al. 2022). All of these variants have proline at position 67 of the protein instead of histidine (Caroli et al. 2009), producing what are known as A2-type milks.

There are at least 23 different brands of A2-type milk in markets around the world (Dantas et al. 2023). In Italy, for example, there has been selection for this allele since 1990, with a 10% increase by 2017 (Chessa et al. 2020). This research shows a significant increase in the *A ² A ² genotype between generations 1 (β-CN*A ² A ²=0.25) and 3 (β-CN*A ² A ²=0.58). It is important to emphasize two points: first, the high frequency of this genotype in the LUC breed is likely due to its origin in the HDV breed; and second, there is no direct selection for this genotype in the herd under study. In addition, A2-type milk is not currently marketed in Colombia. Therefore, the increase in this genotype is likely due to the associated effects of this genotype, such as an increase in milk production, higher fat and protein content, and better yield in curd production (Mohan et al. 2021). Demand for this type of product may increase in the country as consumer awareness of functional foods increases (Dantas et al. 2023).

Five variants are known in the α _S2 -CN protein, including *A, *B, *C, *D, and *E (Meier et al. 2019). However, the genotyping chip only has two SNPs (rs441966828 and rs463985801) in the CSN1S2 gene that allow analysis of the *A and *D variants. In the HDV breed, the SNPs were monomorphic and only the SNP CSN1S2_2 varied in LUC (Table 3); therefore, the reference variant CSN1S2*A and the *AA genotype were fixed in HDV and with a very high frequency in LUC.

Table 3: Allele frequencies of SNPs located in the CSN1S2 gene and allele and genotypic frequencies of protein variants in the αS2-CN in the breeds studied.

These results are similar to those reported in the BON breed (CSN1S2*A=0.997) (Hernández-Herrera et al. 2024) and in 20 other breeds studied (Gallinat et al. 2013; Ahmed et al. 2017; Meier et al. 2019). In contrast, the *D allele has a high frequency (0.98) in the Holstein breed (Ardicli et al. 2018) and values between 2 and 10% in German breeds such as Angler, German Yellow, Hinterwälder, Limpurger, and Vorderwälder (Gallinat et al. 2013). In contrast, the CSN1S2*E variant was only found in the Sarabi breed (0.02) (Gallinat et al. 2013) and the CSN1S2*C variant in Angus with a frequency of 0.075. The positive effect of this gene on milk production and milk quality was attributed to the *AA genotype (Gallinat et al. 2013). This research shows that there is no obvious selection in favor of the genotypes found in the CSN1S2 gene, as the genotypic frequencies did not vary by sex (Figure 1C) or generation (Figure 2C) and were all in HWE.

The genotyping chip used detected 12 SNPs in the CSN3 gene, five of which were fixed in the LUC strain and eight in the HDV strain. From these polymorphisms it is possible to reconstruct the *A, *A ¹, *B, *E, *H and *I variants of Κ-CN of the 14 reported (*A, *A ¹, *B, *B ² , *C, *D, *E, *F ¹, *F ², *G ¹, *G ² , *H, *I and *J) (Caroliet a. 2009; Gallinat et al. 2013; Meier et al. 2019). In both breeds studied, the *A variant of Κ-CN had the highest frequency, followed by the Κ-CN*B variant in LUC and by Κ-CN*I in HDV (Table 4). The other distinguishable alleles (Κ-CN*A ¹, *E, and *H) showed a low frequency in LUC and were absent in HDV.

Table 4: Allele frequencies of SNPs located in the CSN3 gene and allele and genotypic frequencies of protein variants in the Κ-CN in the breeds studied.

The frequency found for the *A allele in the LUC breed is higher than what was reported in many breeds around the world, including Angler (0.55), Butana (0.47), Eringer (0.50), German Red Pied (0.71), German Yellow (0.45), Gir (0.07), Hinterwälder (0.75), Holstein (0.52), Hungarian Grey Steppe (0.55), Jersey (0. 42), Karan Fries (0.62), Limpurger (0.80), Pinzgauer (0.59), Retinta (0.13), Sarabi (0.30), Shorthorn (0.60), Sistani (0.28) and Vorderwälder (0.78) (Gallinat et al. 2013; Ahmed et al. 2017; Meier et al. 2019; Chessa et al. 2020; Mohan et al. 2021); only slightly surpassed by Black Pied (0.83) (Meier et al. 2019). This gene has been studied in other breeds in Colombia. The frequency of the Κ-CN*A variant has been reported as 0.68 in the BON breed (Hernández-Herrera et al. 2024), 0.68 in the HDV breed (Naranjo et al. 2007), and 0.80 in the Holstein breed (Padilla-Doval et al. 2021). In contrast, the Κ-CN*B variant is reported at a high frequency in the Jersey breed (0.9) (Chessa et al. 2020), while the Κ-CN*H variant is more common in B. indicus breeds such as Gir (0.5) and Sistani (0.5) (Gallinat et al. 2013).

The *BB genotype was not found in the breeds studied, but the *AB genotype was found (Table 4). Both genotypes are associated with higher milk production (Mahmoudi et al. 2020). In terms of quality, milk from animals with the *AA genotype has a lower protein content (P<0.05) (Mahmoudi et al. 2020). In contrast, milk from animals with genotype *BB has higher protein, fat, calcium, and phosphorus content (Mahmoudi et al. 2020; Mohan et al. 2021). It also has a better fatty acid content and a higher yield in milk-to-cheese processing, understood by a higher structural stability of the micelles and a shorter coagulation time (Poulsen et al. 2017; Zepeda-Batista et al. 2017). IgE-binding epitopes common to the *A and *E variants are also known in the Κ-CN, which could potentially be more allergenic than the *B variant. However, bioactive peptides with antithrombotic, anti-inflammatory, and immune response modulating activity have also been described in this protein (Vargas-Bello-Pérez et al. 2019).

Due to the above-mentioned advantages of the CSN3*B variant, breeding programs have been initiated to increase its frequency. For example, in the Holstein breed in Italy, the frequency has increased from 0.18 to 0.35 in the last 17 years (Chessa et al. 2020). This does not seem to be the case in the LUC breed, where frequencies have not changed over time (Figure 2D), nor in the BON breed, where a tendency to decrease over time has been reported (Hernández-Herrera et al. 2024). None of the SNPs studied in the CSN3 gene showed significant HWE deviations. The comparison of frequencies concerning sex in the LUC breed showed no differences (Figure 1D), contrary to those previously reported in the BON breed, where the *B variant is more frequent in females (Hernández-Herrera et al. 2024).

Using the β-LG*B allele as a reference, 11 variants (*A, *B, *C, *D, *E, *F, *G, *W, *I, *J, and *W) have been reported (Caroli et al. 2009). The 11 SNPs included in the genotyping chip, located in the LGB gene, allowed the reconstruction of the variants *A, *B, *D, *G, and *I of β-LG, of which the last three showed low frequency in the LUC breed and absence in the HDV breed (Table 5). In both breeds, the *B allele of beta-lactoglobulin was the most common, followed by the *A allele. In the breeds Butana (0.46) (Ahmed et al. 2017), BON (0.67) (Hernández-Herrera et al. 2024), Holstein (0.70), and Simmental (0.59) (Čítek et al. 2023), the *B allele was the most frequent. In contrast, in a previous study, the β-LG*A variant was reported to be the most frequent in BON (0.51) (Rosero et al. 2012), Gyr and Jersey breeds (Gai et al. 2021).

Table 5: Allele frequencies of SNPs located in the LGB gene and allele and genotypic frequencies of protein variants in the β-LG in the breeds studied.

Other reports suggest that animals with the β-LG*AA genotype produce more milk. On the contrary, the β-LG*BB genotype produces higher protein, fat, and shorter rennet formation time (Gai et al. 2021; Čítek et al. 2023). This genotype showed a frequency of around 30% in the breeds analyzed (Table 5). At least two interesting bioactive peptides (LQKW 58-61 and WYSLAMAASDI 19-29) have been reported in β-LG, which, through their antioxidant activity, reduce stress granule formation, scavenge 2,2&apos;-azino-bis-acid radical cations, and increase myotube differentiation in skeletal muscle, resulting in an anti-aging effect (Kim et al. 2019; Čítek et al. 2023). However, these beneficial effects have not been associated with any particular genetic variant.

In this gene, the variant LGB_X14710_5263 (rs109625649) showed significant deviations from HWE. However, this did not imply differences in the genotype frequencies between sexes (Figure 1E) and between generations (Figure 2E), although the *BB genotype was more common in males and the third generation. As mentioned above, there was no selection for this genotype in the herd studied, but it is possible that there was indirect selection due to the milk traits of these animals.

Regarding the population structure in the LUC breed, the wFST values obtained were low, considering all SNPs and as structuring factors sex (wFST=0.0043) and generation (wFST=0.026). In the BON breed, using the herd as a structuring factor, the reported F _ST is 0.035, 0.003 by sex, and 0.001 by year (Hernández-Herrera et al. 2024). The wFST value for the variant associated with the β-CN*A ² protein was 0.0512 (P<0.05). While the estimated F _IT value was -0.0086 (P>0.05), which shows that there is no inbreeding, evaluated from the milk genes.

It has been proposed that, due to the physical proximity of the casein genes, they are more likely to be inherited as a linkage group rather than independently (Villalobos-Cortés et al. 2023). To test this, the 32 SNPs analyzed in caseins and the 11 SNPs in beta-lactoglobulin were pruned. Finally, 14 SNPs were informative in the casein genes and five in the LGB gene. Figures 3A and 3B show the heat maps of the LODs obtained. The highest LOD value was found between the SNPs CSN3_AY380228_13165 (Κ-CN*A) and CSN3_AY380228_13068 (Κ-CN*B), followed by another pair of SNPs also determining the *A and *B variants of the Κ-CN gene (CSN3_AY380228_13165 and CSN3_AY380228_13104_1) and the SNPs determining the Κ-CN*B variant (CSN3_AY380228_13068 and CSN3_AY380228_13104_1). The CSN2S2_1 SNPs, which determine the *A variant of the α _S2 -CN gene, showed a high LOD value with the CSN3_AY38022228_13165 variant, which determines the Κ-CN*A variant. Among the SNPs of the β-CN gene, the LOD values were low (LOD<0.07) and presented an average value between ΚCN and β-CN. The SNP CSN2_7, which determines the most studied variant of β-CN (*A2), did not show LOD values higher than 0.15 with other polymorphisms (Figure 3A). Then, the SNP alphaS1Casein26181 of the α _S2 -CN*C gene showed mean LD values with the SNP determining the *B variant of the same gene, with the SNP CSN2S2_1 determining the *A variant of the α_S2 -CN gene, and the variant CSN3_AY380228_13065 coding for the *H allele of the Κ-CN. The linkage values between the informative SNPs of the LGB gene were low (LOD<0.03), except between LGB_X14710_3982 and LGB_X14710_5174 (LOD=0.50), which both determine the *A allele of the β-LG (Figure 3B). These results show that the linkage between casein genes is low, and the highest linkage is found between SNPs of the same gene. A similar conclusion is presented by Hernández-Herrera et al. (2024) in BON cattle.

Figure 3: Linkage disequilibrium heat map of SNPs after pruning in the casein (A) and beta-lactoglobulin (B) genes in the Lucerne breed.

Twenty-two CSN1S1-CSN2-CSN1S2-CSN3 haplotypes were found, 19 in the LUC breed and eight in the HDV breed, of which five were shared between breeds and three were unique to the HDV (Table 6). Haplotype BB-A ² A ² -AA-AA had the highest frequency (0.312), followed by haplotypes BB-A ¹ A ² -AA-AA-AA (0.194) and BB-A ¹ A ² -AA-AB (0.118) in the LUC breed. The other haplotypes had frequencies of less than 6.5%. In the HDV breed, the most common haplotypes were BB-A ¹ A ² -AA-AA-AI and BC-A ¹ A ² -AA-AA-AB (0.2), which were rare in LUC. Coincidentally, the three most common haplotypes in the LUC breed were also common in the BON breed. In contrast, of the two most common haplotypes in the HDV breed, one was absent in BON, and the other had a frequency below 5% (Hernández-Herrera et al. 2024). Using haplotypes reconstructed from CSN1S1-CSN2-CSN1S2-CSN3 alleles-rather than from genotypes as in the present study-84 haplotypes were identified in the Butana breed, with the most frequent being C-A2-A-A (0.156) (Ahmed et al. 2017). On the other hand, using only CSN2-CSN3 genes, the most common haplotype was A ² -A in Holstein (0.474) and A ² -B (0.646) in the Jersey breed (Chessa et al. 2020).

Table 6: Haplotype frequencies of the CSN1S1-CSN2-CSN1S2-CSN3 genes in the Lucerna and Hartón del Valle breeds.

This research showed a low level of linkage between casein genes. Nevertheless, milk, being a composite of several proteins, could have synergistic relationships that affect its quality and technological performance. For this reason, the effects of CSN1S1-CSN2-CSN3 gene haplotypes on milk traits were studied. Thus, the most frequent haplotype BB-A ² A ² -AA (CSN1S1-CSN2-CSN3) in the LUC breed (31.2%) has contradictory reports. Perna et al. (2016) demonstrated that this haplotype is associated with low cheese yield when coagulation is induced by rennet. In contrast, when coagulation is induced by acidic methods, traits related to cheese formation are significantly improved (Ketto et al. 2017). Similarly, the second most common haplotype in LUC (BB-A ¹ A ² -AA, 19.4%) was associated with better curdling time, better curdling structure, and higher cheese yield (Ketto et al. 2017). On the other hand, the BB-A ² A ² -BB haplotype was associated with higher milk protein and fat content and better rennet formation (Perna et al. 2016). This haplotype had a low frequency in the LUC breed.

Finally, these results justify the efforts made by private and public bodies to conserve Colombian Creole breeds and to promote the development of current conservation programs into genetic breeding programs, not only to promote the use of adapted breeds, but also as breeds capable of producing high-quality milk.

CONCLUSION

Only 46 and 23% of the SNPs evaluated were found to be polymorphic in the LUC and HDV breeds, respectively. Nevertheless, it was possible to reconstruct the most significant variants (CSN1S1*B, β-CN*A ², Κ-CN*B, and β-LG*B) from the perspective of their association with milk production, nutritional quality, and technological performance. These variants exhibited high frequency in both breeds, with minor variations attributable to the limited number of HDV animals utilized and the geographical origin of the LUC breed in HDV. The linkage found between single-nucleotide polymorphisms located in the same gene was high. In contrast, the linkage between the casein genes was found to be low, suggesting that they are independently inherited. The high frequency of these favorable alleles in the Lucerna population-despite the absence of targeted selection-suggests a strategic opportunity to incorporate this information into genetic selection programs aimed at improving both productive efficiency and milk compositional and technological quality. Furthermore, the low genetic linkage among the loci studied suggests that independent selection for each gene is feasible, enabling greater genetic gain without constraints imposed by linkage disequilibrium. This genetic characterization provides a foundation for implementing marker-assisted selection (MAS), advancing towards genomic selection, and promoting the broader use of the Lucerna breed not only for its tropical adaptability but also for its potential to produce high-quality milk.

ETHICAL CONSIDERATIONS

All processes within the experimental analysis were carried out in accordance with the guidelines proposed in &apos;The international guiding principles for biomedical research involving animals&apos; (CIMOS and ICLAS 2012).

CONFLICT OF INTERESTS

The authors declare no competing interests.