GENETIC POLYMORPHISM OF KAPPA CASEIN AND CASEIN MICELLE SIZE IN THE BULGARIAN RHODOPEAN CATTLE BREED

The present study aimed to compare the size of casein micelle in cow milk sample in function of kappa casein (CSN3) genetic polymorphism. Sixteen cows from Bulgarian Rhodopean cattle breed were genotyped by PCRRFLP analysis. Milk samples from the three found CSN3 genotypes (AB, AA and BB) were employed for the determination of casein micelles size by Dynamic Light Scattering (DLS). The results showed differences in the size and polydispersity of the casein micelles between the milks of cows with different genotypes. Hydrodynamic radii of micelles at a scattering angle of 90 °C varied from 80 to 120 nm and polydispersity varied from 0.15 to 0.37. In conclusion casein micelle size of CSN3 AA cows (~ 120 nm) exceed with about 60% cows with AB (~ 80 nm) and BB genotype (~ 70 nm). These results could be useful for improving technological properties of the milk.

Since milk protein genes' polymorphism in genus Bos have been discovered and characterized its development is associated mainly to milk practice to clarify association between genetic variants and milk quantitative and qualitative traits (Tsiaras et al., 2005). This gives opportunity to usage some allelic variants as markers for milk composition and manufacturer properties of milk. Also researches of milk proteins polymorphism are focused to clarify origin, domestication and biogeography of modern cattle breeds (Jann et al., 2004) Genetic variants of milk proteins can be detected by various identification methods. These techniques include, e.g., acrylamide electrophoresis in denaturing (SDS PAGE) or native conditions, isoelectric focusing (IEF), HPLC chromatography, Cryo-scanning electron microscopy etc. (Hallen et al., 2009;Ren et al., 2013). An inexpensive and fast method to determine the size distribution profiles of small particles in suspension or solution is Dynamic Light Scattering (DLS) (Gebhardt et al., 2006;de Kruif et al., 2012). DLS is an optical detection method that can directly measure some important structural parameters of biomacromolecules, such as hydrodynamic radius (R h ) and diffusion coefficient in solution. It exhibits many advantages in the analysis process, including sensing very small amounts of samples without destruction; real-time monitoring of the specimens under different conditions (temperature, pH etc.) and in addition is relatively simple and convenient for operating. Thus, DLS has been widely applied to the structural researches of proteins, polysaccharides and other biomacromolecules (Chu et al., 1995;O'Connell et al., 2001).
In the present work, the correlation between CSN3 different genotypes and the hydrodynamic radius (Rh) of casein micelles from individual milk samples are investigated.

Animals and sample collection
The experiments were performed on nasal swab and milk samples from pure breed cows of Bulgarian Rhodopean cattle (BRC). That breed originated from autochthonous Shorthorn cattle, upgraded mainly with Jersey cow.
From a herd of 80 Bulgarian Rhodopean cows a total of 16 unrelated cows were selected; number of lactations (2-5), age of the cows (3−8 years). All animals were genotyped for the CSN3 gene by PCR-RFLP analysis as described previously (Hristov et al., 2013). CSN3 genotyping showed three genotypes: AA (four cows), AB genotype (eight cows) and BB genotype (four cows). Milk production of each animal was recorded monthly for 305-d lactation period and protein and fat content were determined with MilkoScan 133-B (Foss Electric, Denmark). For DLS analyses milk samples from each cow in mid lactation (100-130 days) were taken after morning milking and sodium azide (0.02%, m/m) was added to all tubes to prevent microbial growth.

Dynamic light scattering measurements
DLS measurements on milk solutions were carried out on a Brookhaven Instruments 90Plus (Brookhaven Instruments Corporation, NY, USA) apparatus at 22.0 °C and scattering angle of 90 °C (wavelength 657 nm and 35 mW). Time dependent fluctuations in the scattered intensity were measured using an avalanche photo detector (APD) and a digital correlator. To check for sedimentation or aggregation data collections were performed in triplicate as 2 minutes co-added runs (total time of 6 min). NIST traceable polystyrene solutions 3020 A, 22 nm ± 1.8 and 3090 A, 92 ± 2 nm (Thermo Scientific) and a blank, 0.02 μm filtered ultrapure water, were used as standards. The used buffer solution (50 mM TBS, pH 7.2) was filtered through 0.44 μm filter and also examined by DLS to account for eventual "dust" particles. Prior to DLS data collection and in order to remove major aggregates (fat fraction and unspecific precipitates) individual milk samples were centrifuged at 2 000 x g for 3 minutes at 4 °C. Then the "skim milk" fraction was diluted 100 times with 50 mM TBS (pH 7.2) and filtered through 0.44 µm syringe filters (de Kruif and Huppertz, 2012). DLS employs the Stokes−Einstein relationship between the diffusion coefficient (D) and the hydrodynamic radius where η is the viscosity. For obtaining size distributions the autocorrelation functions were deconvoluted using the non-negatively constrained least squares fit (multiple pass NNLS) algorithm. In addition, the intensity of scattered light is proportional to the particle size to the sixth power resulting in a higher scattered intensity for larger particles. Thus the intensity weight distributions measured by DLS were converted to number weighted distributions using the analysis software provided by Brookhaven (Brookhaven Instruments Corporation, NY, USA).

Statistical analysis
Descriptive statistics was used concerning the milk productivity and qualitative milk traits data. The calculated mean values (shown as mean value ± SEM) for milk productivity and qualitative traits were compared within different genotypes and evaluated by Student's t-test. These statistical assays were performed with GraphPad Prism version 5.04 (GraphPad software).

Results and Discussion
Effect of κ-CN genotypes on milk quantitative and qualitative traits PCR-RFLP analysis showed three genotypes AB (eight cows), AA (four cows) and BB (four cows) for the 16 selected cows. To determine milk production, butter milk, fat and protein contents during the lactation period (305 d) a total of ten milk samples were collected on a 30 days basis from each animal. The average composition of milk (butter milk, fat and protein) for the different κ-CN genotypes is shown in Table 1. The results clearly demonstrate correlation between genotypes and milk production (AB > AA > BB). Milk production of heterozygous AB animals significantly exceeds that of homozygous BB cows, with 12% or about 500 L (P < 0.01) and with 5% (about 200 L, P < 0.01) that of animals homozygous by A allele of the gene. Regarding fat and protein contents there are only slight differences amongst the three genotypes (Table 1). Values express as means ± standard deviation; values within the same row not sharing a common letter differ significantly, P < 0.01.
Our results support the data by (Bovenhuis et al., 1992) suggesting a 15 % decrease of the milk production of the BB homozygous cows compared to the AB heterozygous cows. Some studies claim that the BB genotype is associated with higher (Van Eenennaam et al., 1991) or lower (Bovenhuis et al., 1992) milk yield whereas other studies indicated no effect (Comin et al., 2008). One should be very careful when crosschecking the results of different studies as in most cases they are not comparable due to differences in population size, breed of cows, frequency of occurrence of specific genetic variants under consideration, methods of expressing traits (whether test day or lactation averages) and the effect of other genetic variants.

DLS measurement of milk samples
Previous studies has linked the size distribution of casein micelle to the lactating stage (de Kruif and Huppertz, 2012), have investigatd the influence of the feeding regimes and investigated the micelle size of native and heated milk samples (Devold et al., 2000). This study focuses on the correlation between κ-CN genotypes and casein micelle size in individual milk samples. The resulting data from DLS measurements (hydrodynamic radius, polydispersity and multimodal distribution) is presented in Table 2. Hydrodynamic radii (R h ) of micelles at a scattering angle of 90 °C varied from 40 to 120 nm. These values are in accordance with micelle size for Norwegian Red cattle (Devold et al., 2000) and Holstein-Friesian cows (de Kruif and Huppertz, 2012). DLS measurments were performed on 16 milk samples with distinct CSN3 genotypes (AA/AB/BB) and the variation of the micelle size in function of genotype is shown on Figure 1. One can see that the micelle size for AA and BB genotypes is not fluctuating a lot. In contrast the micelle size of cows with AB genotype varies a lot. An over simplification of the data interpretation, links the highest observed values of casein micelle size (over 110 nm) to AA genotype. This finding is in agreement with data reported by Bijl et al. (2014) for skimmed milk of Holstein-Friesian cows. According to the data from Table 1 one can eventually suggest a correlation only to the highest protein and fat content in milk for CSN3 AA genotype (3.76 ± 0.04 %; 4.78 ± 0.07 %, respectively). However, there is no clear correlation between casein micelle size and observed milk quantitative and qualitative traits.

Figure 1. Genotypes vs. particle size (radius) distributions (normalized) of micelles obtained with dynamic light scattering
A major complication of light scattering studies is due to the presence of dust particles in the sample, therefore careful filtering procedures have to be applied. In the case of casein micelles an efficient filtering is not always possible, since dust particles and micelles are of similar size. It is thus essential to work with relatively concentrated solutions (resulting in higher polydispersity index (PDI)).
Habitually polydispersity values below 0.1 are suited for DLS experiments while more elevated values of polydispersity are linked to either more concentrated samples or a multi modal distribution. As can be seen from Table 2, polydispersity of casein micelles varied from 0.15 to 0.37 related to a bimodal distribution. As one can see from the multimodal distribution "number vs. diameter" (Figure 2) the number average reveals that the dominant species has smaller size, consistent with the lower average intensity (Peak 1 in Table 1) while the contribution of aggregates with bigger size is minimal.

Conclusion
This study reveals for the first time the correlation between κ-CN genotypes and casein micelle size in individual milk samples. CSN3 AB genotype showing distinct variations of micelle size. DLS data suggest that there is a correspondence with CSN3 genotype e.g. AA genotype shows bigger size of casein micelle. In contrast, protein and fat content in milk cannot be correlated to casein micelles size.