An Investigation of Polymorphism in SMO and LMF 1 Genes and Their Association with Body Size in White Fulani and Muturu Cattle Breeds

Cattle production is integral to the people of Africa and her economy. To improve cattle productivity, there is a need to inculcate molecular marker-assisted selection into current breeding practices. In this study, we investigated, in White Fulani and Muturu cattle breeds, polymorphism of two candidate genes (Smoothened—SMO and Lipase Maturation Factor 1—LMF1) that have been reported to influence body size in cattle. The SMO gene encodes for the SMO (smoothened) receptor protein; a membrane signal transducer that plays a role in the control of bone formation and adipogenesis. LMF1 encodes for the LMF1 protein which regulates post-translational maturation of vascular lipases. Desired regions of the genes were amplified and genotyped by direct DNA sequencing or Polymerase chain reaction—Restriction Fragment Length Polymorphism. For association study, linear body measurements traits (BMTs) that can be used to predict body size were measured on each animal and values of BMTs measured were compared between both breeds. Very significant (P < 0.001) differences, in favour of White Fulani, were observed for all the BMTs compared. Six (6) previously reported SNP (G21234C, C22424T and C22481T, T22939C, C23329T and T23458G) on monomorphic the (T 4 mutation site. Association analysis that polymorphism on G21234C, C22424T, T22939C and loci significantly affected BMTs (P < 0.05 or 0.01) in both cattle breeds. The outcome of this study showed that SMO gene could be a useful candidate gene to facilitate marker assisted selection for body size while LMF1 is monomorphic in the cattle breeds.


Introduction
Cattle production is integral to the people of Africa and her economy. It contributes 33% of the agricultural GDP in developing countries [1]. According to [2], the industry contributed 3.2% to 4% of Nigeria's overall yearly Gross Domestic Product (GDP) between 1970 to 2005. Cattle are a valuable source of protein to the people, and it represents a means of livelihood for the small-scale rural cattle owners and those involved in the cattle production value chain [3].
Over the years, researchers have used genomic tools to identify genes influencing physical traits in cattle and genetic variations that underline the differences observed between individual cattle. The SMO gene has been marked as having a contributory role in skeletal tissue development and adipogenesis in both vertebrates and invertebrate [4]. It encodes for the SMO (smoothened) receptor protein; a membrane signal transducer that plays a role in the control of bone formation and adipogenesis through the hedgehog signalling pathway [4,5]. The gene is mapped to bovine chromosome 4 with 12 exons encoding 780 amino acids [6]. The hypothesis of SMO gene polymorphism affecting economically important traits in cattle was first tested by [7]. Using PCR-Restriction Fragment Length Polymorphism (RFLP), Zhang et al. [7] detected three single nucleotide polymorphisms (SNP) on the SMO genes of 562 Qinchuan cattle and association studies further revealed that two out of the three polymorphic sites (G21234C and C22424T) are significantly associated with body measurement and meat quality traits. In 2015, [6] also conducted a study that confirmed the hypothesis that variation in bovine SMO gene influences body size in cattle. By direct sequencing of the coding region and 3'UTR region of SMO gene in 520 Qinchuan beef cattle, they identified eight single nucleotide polymorphisms (SNP) and linkage disequilibrium analysis revealed five haplotypic groups. Collectively, SMO genes could be a useful candidate gene to be included in marker-assisted selection for cattle with superior body size during breeding programs. To the best of our knowledge, there is no information about polymorphism of this gene in African cattle.
Another candidate gene of influence on body size is the Lipase maturation factor 1 (LMF1). LMF1 encodes for the LMF1 protein. Paterniti et al. [8] were the first to identify the protein in mouse, where a mutation that affected LMF1 protein resulted in hyperglycaemia in mice carrying double mutant allele. This was as a result of very low levels of plasma lipases in those mice. Such abnormality has also been identified in human [9]. LMF1 is located on bovine chromosome 25 [10]. Metabolism of plasma lipid is under the control of vascular lipas-es-lipoprotein, hepatic, and endothelial lipases. Through their action on triglycerides, they regulate the level of plasma lipid and control tissue uptake; thereby protecting the animal against cardiovascular diseases [11]. For these lipases to attain active enzymatic function, post-translational maturation of their embryonic polypeptide chain is required [12]; a process dependent on lipase maturation factor 1 (LMF1) [13]. All these processes take place in the endoplasmic reticulum of cells within lipase expression tissues such as adipose muscle and heart [14]. In cattle, [10] screened four Chinese cattle breeds-Nanyang, Qinchuan, Jiaxian, and Chinese Holstein-for a missense mutation in LMFI. Using PCR-SSCP, direct sequencing, and PCR-RFLP (AvaI endonuclease), they reported a SNP (T > C) in LMF1 exon 4 in three of the four cattle breeds. The polymorphism was associated with growth traits in Nanyang cattle. The T > C mutation resulted in Trp > Arg substitution in LMF1 protein. Ren [15]. The Muturu, a West African taurine, is a small-bodied trypanotolerant cattle [16]. Because routine measurements of cattle body weight can be labour intensive and not sustainable, linear body measurements are usually used as predictive measures for body weight. Traits such as heart girth and hip height have been reported to have high phenotypic and genetic correlations with body weight [17]. Ige et al. [18] reported that heart girth, horn length, body length, and hip height could be used to predict body weight in White Fulani cattle with 0.92, 0.82, 0.83 and 0.69 as their respective coefficients of determination. Dim et al. [19] reported a similar result in the same breed, emphasized that heart girth is the most reliable parameter for bodyweight prediction. Daikwo et al. [20] obtained a multiple regression equation to predict body weight of Muturu cattle from linear body measurement. Heart girth and body length accounted for 90% of the disparity in body weight; a proof that bodyweight could be predicted from the two parameters with high accuracy. were taken using measuring stick/tape as described by Gilbert et al. [21] and done by one person to minimize error. Using the PG-100 collection kit from PERFORMAGENE, a nasal swab was collected from each animal for DNA extraction. The study was carried out between January, 2019-September, 2019.

PCR Amplification
For the two genes considered in this study, primers listed in the literature to amplify desired gene regions were ordered. Amplified regions and the expected SNP within each gene region, as reported in the literature, are presented in Table 1. The primers were first mixed with their respective quantities of 1xTE buffer to make it up to 100 µM as prescribed the primer-design company. The forward and reverse primers were then mixed and diluted to a working concentration of 10 µM.
Polymerase chain reaction (PCR) was performed in a 15 µL reaction mixture that contained 2 µL genomic DNA, 1 µM primer mix, 7.5 µL PCR ImmoMix (contained magnesium chloride, DNA polymerase, and dNTPs-Bioline Ltd, UK) and 4.5 µL dd H 2 O. The PCR cycling was performed in a G-Storm GS1 thermal cycler (Akribis Scientific Limited, UK) with the following general protocols: 98˚C for 10 mins (initial denaturation), followed by 35 cycles of denaturation at 98˚C for 30 s, 50˚C annealing for 30 s, and a final extension at 72˚C for 60 s. The actual annealing temperature used for amplifying each gene region is presented in Table 1. PCR products were checked on 1% agarose gel.

Genotyping by PCR-Restriction Fragment Length Polymorphism
SMO exon 9 amplicon was genotyped by restriction fragment length polymorphism (RFLP) using Sau96I (New England, Biolabs) restriction enzymes. The digestion mixture-which contained aliquots of 5 µL PCR product, 0.5 µL restriction enzyme, 2 µL Cut smart buffer, and 12.5 µL dd H 2 O-was incubated overnight at 37˚C. Digested products were electrophoresed on 2% agarose gel following the same procedure described above, except that the quantity of agarose is 3 grams.

Amplicon Clean up
PCR products from gene regions to be directly sequenced were first purified so

Association Study and Statistical Analyses
The body measurement traits were linearly regressed against animals' age (19 -24 months) in order to account for the effect of age differences on body size. Using the regression coefficient value from the linear regression equation, data obtained from each animal was transformed to its predicted 24-month value.
The original data obtained from the field and the 24-month adjusted data is presented in Supplementary Table S6. The adopted linear regression model adopted was where; Y i = Linear body measurement, a = intercept of the regression line on the Y-axis, b = regression coefficient, X = age, e i = random error about the regression line.
Data from Muturu and White Fulani were compared using the t-test. Allelic and genotypic frequencies for each SNP were calculated by direct counting. The genotypic frequencies were tested for deviation from Hardy-Weinberg equilibrium using Chi-square (χ 2 ) analysis. Single-factor multivariate ANOVA was used to determine the association between SNP marker and body measurement traits using SPSS software; version 24. General linear models (GLM) procedure was followed with the following model: where; Y i = trait measured on each individual; μ = overall mean; G i = fixed effect of the ith genotype; and ε i = random error.

Regression Equation
Presented in Table 2 is the coefficient of determination (R 2 ) values for each trait regressed against age. The values were positive and above 80% except for rump length and chest depth in White Fulani, which were 73% and 64% respectively.

t-Test Comparing Values from Muturu and White Fulani
Values from both breeds were compared using t-test (Table 3). Very significant (P < 0.001) difference, in favour of White Fulani, was observed for all the traits compared.

SMO Exon 9
A 163-bp region was amplified in the exon 9 of bovine SMO gene. When incubated with the restriction enzyme, Sau96I, to detect the G > C mutation observed at the 21,234 bp-locus in Qinchuan cattle, the PCR product produced three banding patterns representing CC, GC, and GG ( Figure 1 and Figure 2). Commensurate with the result of [7], the CC is represented by one fragment (163 bp), GG by two fragments (64 and 99 bp), and GC by three fragments (64, 99 and 163 bp).
Allelic and genotypic frequencies were calculated for each breed (Table 4). In White Fulani, individuals with heterozygous GC genotype occurred most, followed by CC and GG, respectively. In Muturu, the heterozygous (GC) genotype,    even though it still has the highest frequency, only occurred in one more animal when compared to the GG genotype in the population. Genotype CC had the lowest frequency of occurrence. All genotypic frequencies did not deviate from Hardy-Weinberg equilibrium.

Effect of the SMO Exon 9 G21234C Polymorphism on Body Measurement Traits
In both breeds, significant differences existed among the genotypes for all traits measured, except for rump length in White Fulani (Table 5).
1) White Fulani Mean values for heart girth, wither height, and chest depth were significantly (p < 0.05) higher in genotypes GG and GC when compared with CC in White Fulani. Significant (P < 0.001) differences existed among all the three genotypes for body length increasing from CC to GC and GG, respectively. Mean values for rump length were not significantly (P > 0.05) different across the genotypes.
2) Muturu In Muturu population, mean values for body length, heart girth and wither height increased significantly (P < 0.001) and progressively from CC to GC and GG genotypes. Unlike in White Fulani where rump length was not significantly different across genotypes, GG genotype had significantly (P < 0.001) higher rump length than GC and CC. Mean values GC and CC were not significantly (P < 0.05) different for rump length. For chest depth, GG genotype had significantly (P < 0.05) higher values than CC, but GG and GC were not significantly different.

SMO Exon 11
319-bp of SMO exon 11 gene was amplified and sequenced for 77 cattle. Geneious Prime (version 2019.2) software [22] was used to analyse the sequence map.
The two SNP-C22424T and C22481T-reported by Zhang et al. [7] that caused synonymous substitution of glycine and histidine, respectively, were identified. The software allowed identification of homozygous and heterozygous genotype carriers.
For the C22424T mutation, two genotypes (CC and TT) were found in the two cattle breeds under examination (Figure 3), whereas for the C22481T mutation, White Fulani individuals had two genotypes (CC and CT) while all Muturu cattle were homozygous for the C allele ( Figure 4).
Allelic and genotypic frequencies were calculated for each breed (Table 6). There were similarities in the allelic and genotypic distribution in both cattle breeds for C22424T polymorphism. 76% and 77% of the cattle population had the CC genotype in White Fulani and Muturu breeds, respectively. For the C22481T SNP, 86% of the White Fulani cattle had CC genotype, while all Muturu cattle had the CC genotype ( Table 6). All genotypic frequencies did not deviate from Hardy-Weinberg equilibrium.

Effect of the SMO Exon 11 C22424T SNP on Body Measurement
Traits 1) White Fulani Significant difference (P < 0.05) was observed between the two genotypes-CC and TT-for all body measurement traits measured except body length (Supplementary Table S1). The difference was in favour of the CT genotype. Advances in Bioscience and Biotechnology   2) Muturu Significant differences (P < 0.05) was only observed in body length and chest depth (Supplementary Table S1). Animals with CT genotype had significantly higher mean body length and chest depth compared to those with CC genotype.

Effect of the SMO Exon 11 C22481T SNP on Body Measurement
Traits In White Fulani breed, no significant (P > 0.05) difference was observed between the two genotypes in all body measurement traits measured in this study (Supplementary Table S2). All Muturu cattle carry same CC genotype. SNP [6] was identified. All the animals were genotyped for the polymorphism by direct sequencing and three genotypes-CC, CT, and TT-were found in both cattle breeds ( Figure 5).

SMO Exon 12
Allelic and genotypic frequencies were calculated for each breed (Table 7). In White Fulani, 70% of the cattle had CC genotype, 27% had heterozygous CT while only one animal had the homozygous TT. Contrarily, 28% of the Muturu cattle population carry the homozygous TT, 32% carry CC genotype while 40% had heterozygous CT. Therefore, the T-allele was more frequent in the Muturu breed (48%) compared to White Fulani (17%). All genotypic frequencies did not deviate from Hardy-Weinberg equilibrium.

Effect of the SMO Exon 12 T22939C SNP on Body Measurement
Traits 1) White Fulani Significant difference (P < 0.05) was observed among the genotypes for body length and wither height. Mean values for body length and wither height in CT genotype was significantly (P < 0.05) higher compared to CC genotype. However, TT and CT genotypes were not significantly different (Supplementary Table   S3). No significant difference was observed between the three genotypes for all other traits measured.

2) Muturu
For all the traits except chest depth, mean values for cattle with TT genotype was highest (P ≤ 0.001), followed by CT and the least mean value was found in cattle carrying the CC genotype (Supplementary Table S3).

SMO 3' Untranslated Region (3'UTR)
3' UTR region of the SMO gene was amplified and sequenced for 76 cattle. Sequence mapping and genotype identification were done with Geneious Prime software (version 2019.2) [22]. Two SNP-C23329T and T23458G-reported by [6] were identified in the cattle population under examination.
Three genotypes were found for each SNP. CC, CT and TT; and TT, TG, and GG genotypes were present for C23329T and T23458G SNP respectively ( Figure  6 and Figure 7).
In both breeds, allelic and genotypic distribution were similar for the two SMO 3'UTR SNP. In C23329T, C-allele predominates, occurring at a frequency of 92% and 91% in White Fulani and Muturu respectively. Only one animal per breed carried the TT homozygous (Table 8). For the T23458G SNP, T and G-alleles were prominent in both breeds; T-allele occurred in 53% and 57% of the White Fulani and Muturu cattle respectively (Table 8). All genotypic frequencies did not deviate from Hardy-Weinberg equilibrium.

Effect of the SMO 3'UTR C23329T SNP on Body Measurement Traits
In both breeds, no significant (P > 0.05) difference was found between the three genotypes for all the body measurement traits (Supplementary Table S4).

Effect of the SMO 3'UTR T23458G SNP on Body Measurement
Traits Contrary to the no-effect obtained in C23329T SNP, significant difference existed between the genotypes for the T23458G SNP.

1) White Fulani
Mean values for body length and wither height were significantly (P < 0.01 or P < 0.05) different across the genotypes. Values for TT and TG were similar but significantly greater than values for GG in both body length and wither height (Supplementary Table S5). No significant difference was found between the three genotypes in all other traits measured.

2) Muturu
There was a significant difference (P < 0.01) between the genotypes for all body measurement traits except chest depth. Cattle with TT genotype had significantly (P < 0.01) higher mean values for body length, heart girth, wither height, and rump length when compared to those with GG genotype (Supplementary Table S5).

Effect of the LMF1 Exon 4 T27154C SNP on Body Measurement Traits
Using primers listed above, 307-bp of LMF1 exon 4 covering the T27154C polymorphic site reported by Ren et al. [10] in Chinese cattle was amplified and sequenced for 74 African cattle. Sequence analysis was done with Geneious software (version 2019.2). At the T27154C mutation point, only homozygous T-allele were found (Figure 8). In the population of White Fulani (38) and Muturu (36) examined in this study, all animals had a fixed T-allele, and genotypic frequency for TT was 1.0.

Regression Equation
The coefficient of determination (R 2 ) values for all the traits regressed against age were positive and above 80% except for rump length and chest depth in White Fulani, which were 73% and 64% respectively. This suggests that, unlike in Muturu breed, age difference only accounted for a relatively smaller component of the variations in rump length and chest depth between White Fulani cattle. Such distinctions observed between cattle breeds might be as a result of breed to breed differences in the shape of measured traits. Berge et al. [23] noted that curvature of the first rib, around where chest depth is measured, varies across breeds. Variations in the shape of rump between the breeds have been reported. While the taurine breed has a well-muscled and more horizontal rump [24], White Fulani rump gently slopes downward from hook to pin bone [25]. In general, the regression equations can be used to make predictions for values of body measurement traits with a moderate to high degree of confidence. This agrees with the findings of [26] where age of cattle significantly accounted for the differences in linear body measurement values (body length, height at wither, heart girth) in Muturu and White Fulani cattle. Bene [27] also reported a moderate correlation (up to 0.46) between age of animal and body measurement traits in nine cattle breeds.

t-Test Comparing Values of White Fulani and Muturu
Mean values for all body measurement traits were significantly higher in White Fulani compared to Muturu. This is not unexpected as the highly varied body size of both cattle breeds is well documented. Udeh [28] reported that at P < 0.01, body measurement traits (heart girth, withers height, and body length) were significantly greater in zebu breed than Muturu. In the research conducted by [26], body measurements-live weight, head length, head width, body length, height at wither and heart girth-were significantly (P < 0.05) greater in White Fulani compared to Muturu.

SMO Polymorphisms and Their Association with Body Measurement Traits (BMTs)
As a result of its association with BMTs in other cattle breeds [6,7], the SMO gene was investigated here in White Fulani and Muturu breed. No study has tried to investigate this.

SMO9 Polymorphism and Its Association with BMTs
Although with different allelic and genotype distribution, the G21234C SNP in bovine SMO exon 9 was found in both breeds under investigation. The G-allele had a higher occurrence in Muturu (56%) compared to 34% in White Fulani. In agreement with the findings of [7], association analysis revealed that the G21234C polymorphism affected body length, heart girth, wither height, and chest depth in both breeds. It was also associated with rump length in Muturu, but not in White Fulani cattle. The non-significant effect of SMO9 on rump length observed in White Fulani might not be unconnected to the relatively low coefficient of regression obtained when age was regressed on rump length, as described above. From the association result, the G allele and GG homozygous genotype are favourable for selection to improve body measurement traits in both breeds. In both breeds, GG genotype was superior to CC in certain body measurement traits (BL, HG, WH and CD). In Muturu, homozygous GG genotype had significantly (P < 0.05) higher mean values than the heterozygous GC genotype for all traits except chest depth, unlike in White Fulani where mean values for the GG and GC genotypes were not significantly different for those traits. The results for White Fulani closely match the outcome observed in Qinchuan cattle where homozygous GG cattle had the same mean BMTs as those with heterozygous GC 7 . It is equally important to note that the GG genotype only appeared twice in the White Fulani population under study. This low GG genotype might be as a result of the small sample size used for this study. In all, this finding revealed that genetic improvement program should focus on using selection to increase G allele in the population. As reported by [7], the G21234C SNP results in synonymous substitution of arginine, thereby not altering the amino acid sequence of the translated protein. This is not expected to alter the function of the SMO protein. However, influence of the SNP on body measurement traits observed can be explained by codon bias. This change in codon (G/C), although does not affect amino acid sequence, could affect the timing of translation from gene to protein [29,30], thereby altering protein function and structure. This could be the scenario affecting mRNA translation into SMO protein, which then influenced the transmission of hedgehog signals. It has also been reported that through interaction with RNA-binding proteins, such silent mutation can have a regulatory effect on post-translational gene expression [31,32]. Furthermore, non-synonymous SNP can also influence DNA methylation in epigenetic regulation [33].

G222424T
White Fulani and Muturu cattle population studied had two genotypes-CC and CT. Homozygous TT was absent in both breeds. In Qinchuan cattle, [7] found all the possible three genotypes. Absence of TT genotype in this study is likely to be due to the small sample size. Statistical analysis revealed that mean values for four BMTs in White Fulani (heart girth, wither height, rump length and chest depth) and two BMTs in Muturu (body length and chest depth) were significantly (P < 0.05) greater in cattle with CT genotype compared to those with TT genotype. This finding agrees with what was reported by [7] where cattle with CT genotype had significantly higher mean values for body measurement traits relative to those with CC and TT genotype. Although the C > T SNP only resulted in synonymous glycine mutation in the SMO protein, studies have re-vealed that such mutation can also affect protein function and structure as previously explained above in SMO exon 9 polymorphism. It appears that the heterozygous CT genotype is favourable for superior body measurement traits in the breeds. Therefore, conscious selection and breeding program should be put in place to increase the frequency of this genotype in the Muturu and White Fulani cattle population.

C22481T
In White Fulani, both CC and CT genotypes were found with CC occurring more frequently. TT genotype was absent just as it was in Qinchuan cattle [7]. Only the CC genotype was found in Muturu cattle. This represents a clear distinction between the White Fulani and Muturu cattle population studied. The absence of the T-allele in Muturu cattle might be due to the low sample size, or the C22481T mutation has not happened in the breed. The mutation resulted in a synonymous histidine substitution, as noted by [7]. Association studies in White Fulani revealed that no significant (P > 0.05) difference exist between CC and CT genotype for all the body measurement traits. This result tallies with the outcome from [7] that the C22481T polymorphism does not affect body measurement traits. A plausible explanation offered was that the effect of CC and CT genotype on mRNA translation to SMO protein, and on the protein structure, was of equal proportion. This SNP can therefore not be used for marker-assisted selection.

SMO Exon 12 T22939C Polymorphism and Its Association with BMTs
Three genotypes-CC, CT, and TT-were identified in both Muturu and White Fulani breeds for the C22939T SNP. The T-allele frequency was 17% and 48% in White Fulani and Muturu breed respectively. TT-genotypic frequency of 28% in Muturu compared to 3% in White Fulani exemplified the relative abundance of the T-allele in Muturu cattle population. Statistical analysis revealed that mean values for body length and wither height were significantly (P < 0.05) greater in White Fulani cattle carrying CT and TT genotypes when compared to those with the CC genotype. In Muturu, mean values for individuals with TT genotype was very significantly (P < 0.001) greater than those with CC and CT genotypes for all the body measurement traits except chest depth. Taken together, it appears that selection to increase the frequency of T-allele in the population will be beneficial to attain superior body measurement traits. In Qinchuan breed studied by [6], cattle with C-allele had significantly (P < 0.05) lower body size traits compared to those with the T-allele. Zhang et al. [6] reported that this T > C polymorphism resulted in a non-synonymous serine to proline amino acid substitution in the c-terminal of bovine SMO protein. Without mutation, activation of the SMO protein is triggered by phosphorylation of serine-threonine residue at the c-terminal end of the protein [34]. A substitution of serine to proline will not permit phosphorylation, which will invariably inactivate SMO protein. Therefore, this protein inactivation might be the basis for cattle with the C-allele hav-ing lower body measurement traits. The protein, which is a membrane signal transducer, plays a role in the control of bone formation and adipogenesis through the hedgehog signalling pathway [4]. and as such cannot be as molecular marker for cattle selection. The best practice is to examine the effect of a marker on a breed using a large population of that breed, as it is not unusual to find inconsistencies between different studies that utilized different breeds [35]. Therefore, a larger sample size is needed to confirm the veracity of this result.

T23458G
Contrary to the dominance of one allele found in the C23329T SNP, both T and  [37] showed that a mutation in the 3'UTR region of PPARA gene influenced transcript level of the gene, and subsequently had significant influence of adipose tissue accumulation in pig.

LMF1 Polymorphism and Its Association with BMTs
All animals in both breeds investigated for the T27154C SNP were monomorphic and homozygous for the T-allele. Ren et al. [10] had associated this SNP with body size in Chinese Nanyang cattle. It appears that Nigerian cattle-both taurine and indicine-are monomorphic at the mutation site or genetic drift has resulted in the loss of this variation. In Chinese Holstein cattle breed screened by [10], all the animals were also monomorphic and homozygous for the T-allele at the 27154 T > C site. Ren et al. [10] offered an explanation that C-allele of the

Conclusion
In this study, we've showed evidence of polymorphism of SMO gene in White Fulani and Muturu cattle breeds. In addition, we report that variations at G21234C, C22424T, T22939C and T23458G loci significantly affected BMTs in both cattle breeds. However, evidence suggests the absence of LMF1 polymorphism in the two African cattle breeds.