Identification of polymorphisms influencing feed intake and efficiency in beef cattle

Transcription

1 doi:0./j x Identification of polymorphisms influencing feed intake and efficiency in beef cattle E. L. Sherman*, J. D. Nkrumah*,, B. M. Murdoch* and S. S. Moore* *Department of Agricultural, Food and Nutritional Science, University of Alberta, Edmonton, Alberta, Canada T6G 2P5. Igenity Livestock Production Business Unit, Merial Ltd, Edmonton, Alberta, Canada T6G 2P5 Summary Feed efficiency is an economically important trait in beef cattle. Net feed efficiency, measured as residual feed intake (RFI), is the difference between actual feed intake and the predicted feed intake required for maintenance and gain of the animal. SNPs that show associations with RFI may be useful quantitative trait nucleotides for marker-assisted selection. This study identified associations between SNPs underlying five RFI QTL on five bovine chromosomes (BTA2, 5, 0, 20 and 29) with measures of dry matter intake (DMI), RFI and feed conversion ratio (FCR) in beef cattle. Six SNPs were found to have effects on RFI (P < 5). The largest single SNP allele substitution effect for RFI was )0.25 kg/day located on BTA2. The combined effects of the SNPs found significant in this experiment explained 6.9% of the phenotypic variation of RFI. Not all the RFI SNPs showed associations with DMI and FCR even though these traits are highly correlated with RFI (r = 0.77 and r = 0.62 respectively). This shows that these SNPs may be affecting the underlying biological mechanisms of feed efficiency beyond feed intake control and weight gain efficiency. These SNPs can be used in marker-assisted selection but first it will be important to verify these effects in independent populations of cattle. Keywords beef cattle, feed efficiency, feed intake, polymorphisms, residual feed intake. Introduction In beef cattle production, feed intake and feed efficiency are important traits to study for both economic and environmental reasons. Feed is the highest variable cost in beef production. It has been shown that an improvement in feed efficiency can improve overall beef production system efficiency and reduce methane production (Archer et al. 999; Nkrumah et al. 2006; Hegarty et al. 2007). Feed efficiency is measured in several different ways. Traditionally, it is measured as feed conversion ratio (FCR), which is the ratio of feed to gain and is therefore highly correlated with the growth of the animal. Selection to improve FCR can be beneficial in younger, growing animals but can also lead to larger animals in the breeding herd that are more expensive to maintain (Archer et al. 999). An alternative measure of feed efficiency is residual feed intake (RFI) or net feed efficiency. It is the difference between an animalõs Address for correspondence S. S. Moore, Department of Agricultural, Food and Nutritional Science, University of Alberta, Edmonton, Alberta, Canada T6G 2P5. stephen.moore@ualberta.ca Accepted for publication 7 December 2007 actual feed intake and its predicted feed intake based on growth and weight of an animal over a time period (Koch et al. 963; Archer et al. 999). Positive aspects for using RFI as a measure of feed efficiency is that it is phenotypically independent of the production traits used to measure it (growth and weight; Kennedy et al. 993) and therefore may reveal variation in basic metabolic processes in the animals that determine efficiency (Kennedy et al. 993; Archer et al. 999; Arthur et al. 200a,b). The heritability of RFI in cattle is moderate (Liu et al. 2000; Arthur et al. 200a,b), suggesting that genetic variation exists in RFI, and therefore the identification of polymorphisms affecting this trait is possible (Archer et al. 999). Indeed, a recent whole-genome association study identified many SNPs throughout the bovine genome with effects on RFI (Barendse et al. 2007). The traditional method to identify genes and genetic markers affecting a trait such as RFI is to identify QTL in the genome followed by fine mapping and positional cloning. A primary genome scan for RFI QTL has been demonstrated (Nkrumah et al. 2005, 2007b) and the several identified cattle chromosomes with RFI QTL (BTA2, 5, 0, 20 and 29) have been fine mapped to even smaller confidence intervals (Moore et al. 2006). The objectives of the current study were to evaluate SNPs near these QTL for effects on feed 225

2 226 Sherman et al. efficiency and feed intake by exploiting potential linkage disequilibrium (LD) between the QTL and nearby underlying markers. Materials and methods Animals and data collection The animals used in this study were 464 steers sired by Angus, Charolais or Alberta Hybrid bulls crossed to Hybrid dams and are further described in Nkrumah et al. (2007a,b). Briefly, the Hybrid animals were from three composite lines: one line was composed of 33% each of Angus and Charolais, 20% Galloway and the remainder from other beef breeds; another line was composed of about 60% Hereford and 40% other beef breeds and the last line was composed of approximately 60% dairy breeds (Holstein, Brown Swiss or Simmental) and 40% beef breeds (mainly Angus and Charolais). Animals were weighed every 2 weeks and feed intake measurements were collected over six different 70-day feedlot tests spanning 3 years using the GrowSafe system (GrowSafe Systems Ltd). The test procedures and diet are further described by Nkrumah et al. (2007a,b). Briefly, the diet in year was composed of 80% dry-rolled corn, 3.5% alfalfa hay pellet, 5% feedlot supplement and.5% canola oil, which was 88.9% DM and supplied 2.90 Mcal/kg of ME and % CP. The test diet in years 2 and 3 contained 64.5% barley grain, 20% oat grain, 9.0% alfalfa hay pellet, 5.0% beef feedlot supplement (32% CP beef mineral supplement containing 440 mg/kg of monensin, trace minerals and vitamins) and.5% canola oil, which was 90.5% DM and supplied 4.0% CP and 2.9 Mcal/kg of ME. The diet in year was different because of a barley grain shortage. Traits Feed intake measurements from the 70-day test period were used to calculate the average daily dry matter intake (DMI). FCR was calculated as the ratio of DMI to average daily gain (ADG), which was derived using linear regression of weight measurements taken throughout the test period. Residual feed intake was calculated as the difference between the DMI of each animal and its predicted feed intake, which was calculated either using a phenotypic regression () or genetic regression () of weight (metabolic body weight) and growth (ADG) on DMI (Arthur et al. 200a,b; Crews 2005). These traits are further described in Nkrumah et al. (2007a,b). Genotypes and analysis Genotyping of the SNPs was carried out using the Illumina GoldenGate assay (Oliphant et al. 2002) on the BeadStation system (Illumina, Inc.). For each RFI QTL on BTA2, 5, 0, 20 and 29 (Nkrumah et al. 2005; Moore et al. 2006) 4 8 SNPs were analysed in a region surrounding the QTL for associations with,, FCR and DMI. In total, 78 SNPs and were analysed in these five regions. The locations of these SNPs are from McKay et al. (2007). Linkage disequilibrium was tested in a pairwise fashion for SNPs on the same chromosome using the measure r 2 (Devlin & Risch 995), which was calculated using an EM algorithm in SAS v. 9. (SAS Institute). For the SNPs that were found to have significant effects on RFI, the locations on each chromosome in centimorgans, the GenBank accession number, base pair positions and genotype frequencies are shown in Table 2. The SNP analysis was carried out using a general mixed model in SAS (PROC MIXED). The model included random effects of sire and dam, fixed effects of test group and sire breed and a linear covariate of age of animal on test along with the SNP genotypes. Significance thresholds for P-values were determined using permutations (Churchill & Doerge 994). Single SNP associations are shown for the SNPs with P < 5 in Table 2, and association profiles of all the SNPs analysed in the study are illustrated in Fig. for each of the five QTL regions. Results and discussion In previous studies RFI QTL were detected on several chromosomes (Nkrumah et al. 2005, 2007b), and these were subsequently fine-mapped with the addition of 08 more SNPs with a spacing of approximately cm across the five chromosomes BTA2, 5, 0, 20 and 29 (Moore et al. 2006). Briefly, these QTL were mapped using half-sib regression interval mapping, which was implemented in the QTL EXPRESS software (Seaton et al. 2002). Families were first analysed to see if they were segregating for the QTL and then only the families that were segregating were reanalysed together (Schnabel et al. 2005). Half-sib families had an average size of 23 progeny. Chromosomewise thresholds were calculated from 2000 permutations. Details of the QTL are shown in Table. To further analyse these QTL, available SNPs surrounding each QTL were tested for associations with RFI, FCR and DMI. An average of 5.8 SNPs were analysed on each chromosome with an average spacing of 0.76 cm or 290 kb (physical SNP locations based on the Btau 3. genome build). To evaluate the extent of the coverage of the QTL areas by these SNPs, the LD between all SNP pairs on the same chromosome were calculated using r 2 (Devlin & Risch 995). Table 2 shows all pairs of SNPs with an r 2 value > 0.5. Only 3 SNP pairs were found to have an r 2 > 0.5, of which four had r 2 > Barendse et al. (2007) found that LD was low beyond SNPs separated by <30 kb and estimated that a distance of 30 kb from a SNP associated with RFI was required to detect other SNPs in LD that would also show associations with RFI. Considering the low LD found between the SNPs used in our study and the

3 Feed efficiency in beef cattle 227 (a) 3.5 L og(/p) Log(/ P) (c) Log(/P) (e) BTA2 SNP Location (cm) BTA0 SNP Location (cm) BTA29 SNP Location (cm) (b) Log(/P).5.0 (d) 3.0 Log(/P) BTA5 SNP Location (cm) BTA20 SNP Location (cm) 5 Figure Association profiles of SNPs near five residual feed intake QTL. Associations were plotted as log(/p), where P is the P-value from the allele substitution effect. SNP locations are in centimorgans. SNP 2 on BTA20 is GHR SNP AY643807:g.300A>G and SNP 6 on BTA5 is rs290230:a>g as identified by Barendse et al. (2007). (a) BTA2, (b) BTA5, (c) BTA0, (d) BTA20 and (e) BTA29. average spacing of 290 kb, it is likely that more SNPs are needed to fully explore these regions and it is still possible that some associations were not detected. The association profiles of all the SNPs analysed are illustrated in Fig. for each chromosome. The locations and genotype frequencies for the most significant SNP on each chromosome are listed in Table 3 and the allele-substitution effects and P-values are shown in Table 4 for RFI, FCR and DMI. The largest allele-substitution effect on RFI from rs :c>t was )0.25 kg/day (P =009). This SNP also had the largest effect on FCR ()0.26 kg DMI/kg gain) but had no significant effect on DMI. The largest effect on DMI was 0.3 kg/day from rs290296:a>t (P =53) on BTA0 but the most significant effect due to a smaller SE was from rs :c>t on BTA5 at 9.6 cm ()0.22 kg/day). Both rs290296:a>t and rs29020:c>t were associated with all the traits, RFI, FCR and DMI, while rs :a>g on BTA29 only affected and but neither FCR nor DMI. There were significant correlations between RFI and DMI (r = 0.73) as well as between RFI and FCR (r = 0.62) (Nkrumah et al. 2004). Even though these traits are correlated, not all SNPs showed associations with all traits, demonstrating that genetic markers may or may not have pleiotropic effects. It has been suggested that RFI will be a better indicator of feed efficiency than traditional measures, such as FCR, and that selection for RFI will benefit the whole production system, not just the young growing animals in the feedlot sector (Archer et al. 999). Because the SNPs identified for RFI were not always associated with FCR and DMI, it is possible that they were picking up the differences between these traits and reflect feed efficiency mechanisms not involved in feed intake control or growth efficiency alone. Indeed, use of these markers in markerassisted selection may be beneficial to the whole production system. For each QTL region on the five chromosomes, one SNP was found with an allele substitution effect on RFI with at

4 228 Sherman et al. Table Results, QTL locations and number of families used in half-sib regression interval QTL mapping on BTA2, 5, 0, 20 and 29 for RFI. BTA Trait 2 (cm) 3 F-value 4 families 5 Location Number of ** ** * *** *** * ** 69 Number of animals *P < 5, **P <, ***P < 0, P <. Results from Moore et al. (2006). 2, phenotypic residual feed intake (kg/day);, genetic residual feed intake (kg/day). 3 Map locations based on McKay et al. (2007). 4 F-statistic significance based on 2000 permutations. 5 Only families found to be segregating for QTL were used in the QTL analysis. least P < 5, except on BTA0 where the most significant SNP (rs290296:a>t) had P=5 for. This SNP had a low MAF and the rs290296:aa genotype class could not be included in the analysis because only two animals in the population contained the rs290296:aa genotype and therefore could skew the estimates. The locations of these SNPs correspond well with the respective QTL locations, all of them within 3 cm of the QTL location. Therefore, these SNPs provide a practical reference point to search the regions for genes or other regulatory elements such as promoter sequences, regulatory sequences or micro-rnas that could be the causative effectors of the QTL that were detected. Although these SNPs are not located in coding regions of genes, three of the SNPs are located in introns. The SNP rs :c>t on BTA2 is in the gene for ribosomal protein S6 kinase, 90 kda polypeptide, rs :c>t on BTA5 is in the hypothetical LOC56544 gene and rs :a>g on BTA29 is in the gene for G-protein-coupled receptor 37. It is possible that these SNPs are in LD with other causative SNPs within these genes, different nearby genes or regulatory sequence elements. The QTL identified on BTA20 were located at 49 and 56 cm for and respectively. This is close to the growth hormone receptor (GHR) gene located at 42 cm. The SNP AY643807:g.300A>G, which is located in intron 4 of GHR, has been previously identified to show associations with RFI (P = 38; Sherman et al. 2008). Two other SNPs close to this QTL (rs29020:c>t at 42.7 cm and rs290464:a>c at 55.2 cm) also had allele-substitution effects on RFI. No LD was seen between these three SNPs even though AY643807:g.300A>G and rs29020:c>t are < cm apart (although according to the bovine genome build Btau 3. these SNPs are kb apart and therefore LD is unlikely). It remains unclear which SNPs may reflect the effects of the QTL on this chromosome as the two other SNPs are not in any genes and the GHR SNP is in an intron. Further evaluation of this region is needed with a higher density of SNPs. BTA SNP pair 2 r 2 SNP position SNP2 position SNP interval 3 cm kb cm kb cm kb Table 2 SNP location and distance between SNPs for SNP pairs in LD (only SNP pairs with r 2 > 0.5 are shown) Positions of the two SNPs of the SNP pair tested for LD. Positions are in centimorgans (McKay et al. 2007) and in kilobase pair (genome build Btau_3.). 2 All SNPs on the same chromosomes were tested in two locus pairs for LD using r 2. SNP numbers of the pair refer to the SNP numbers as labelled in Fig. for each chromosome. 3 Distance between two SNPs of the SNP pair.

5 Feed efficiency in beef cattle 229 Table 3 Location, GenBank accession contig number, contig position, genotypes and frequencies for SNPs in 464 beef cattle. SNP BTA cm GenBank accession number 2 Base pair position 3 Genotype Frequency Count rs :c>t NW_ CC 26 2 CT TT rs :c>t NW_ AA AG GG rs290296:a>t 0 5. NW_ AA AT TT rs29020:c>t NW_ TT CT CC rs290464:a>c NW_ AA AC CC rs :a>g NW_ AA AG GG NCBI rssnp ID. 2 GenBank contig accession number. 3 Base pair position in GenBank contig. 4 These two animals were not used in the analysis of this SNP. Table 4 Estimates of allele-substitution effect for feed intake and feed efficiency for several SNP in beef cattle. SNP BTA Trait Allele substitution effect 2 Estimate SE P-value 3 rs :c>t 2 ) ) FCR ) rs :c>t 5 ) ) DMI ) rs290296:a>t FCR DMI rs29020:c>t FCR DMI rs290464:a>c FCR rs :a>g , phenotypic residual feed intake (kg/day);, genetic residual feed intake (kg/day); DMI, dry matter intake (kg/day); FCR, feed conversion ratio (kg DM/kg gain). 2 Allele-substitution effect is the effect of substituting one allele in the population with the other allele (Falconer & Mackay 996). For this analysis the SNPs were recoded as 2, and 0 and the effect was calculated as the parameter estimate for the SNP as a covariate during analysis. 3 P-value based on permutations. An additive model fitting all the SNPs simultaneously from each QTL region using multiple regression was attempted using the five most significant SNPs (rs :c>t, rs :c>t, rs29020:c>t, rs290296:a>t and rs :a>g). This model accounted for 6.9% of the variation in

6 230 Sherman et al. (P =3. 0 )5 ) and 5.7% of the variation in (P =.74 0 )5 ). A whole-genome association analysis for RFI (Barendse et al. 2007) with 8786 polymorphic SNPs also identified 6 SNPs of which 20 of these explained 76% of the genetic variance in RFI. This demonstrates that many SNPs throughout the genome will contribute to the variation seen in RFI in cattle, but they have not all been identified. Only one SNP overlapped between the current study and the RFI SNPs identified by Barendse et al. (2007). This was rs290230:a>g on BTA5 at 89.2 cm. Barendse et al. (2007) obtained P=035 for this SNP but no significant effect on RFI was seen in our study. This indicates that the SNP is not in LD with the QTL in our population, which could be due to differences in the cattle populations used for the two studies. This demonstrates the importance of validating SNPs in different cattle populations before a SNP can be applied to marker-assisted selection. This study provides a practical reference point to search the QTL regions for genes or other regulatory elements such as promoter sequences, regulatory sequences or micrornas that could be the causative effectors of the QTL. The identification of positional candidate genes that influence RFI may help elucidate the biological mechanisms of feed efficiency as well. Furthermore, the identification of these SNPs is an important step towards the improvement of RFI in beef cattle by providing polymorphisms in the bovine genome that can be used in marker-assisted selection. Before implementation of these SNPs in marker-assisted selection, it will be important to verify these effects in other independent populations of cattle as well as the effects on other economically important traits. Acknowledgements This work was supported by grants awarded to Stephen Moore from Canadian CattlemanÕs Association, Alberta Agricultural Research Institute, Alberta Beef Producers, Canada-Alberta Beef Industry Development Fund, Beef Cattle Research Council and through scholarships awarded to Laura Sherman from National Science and Engineering Research Council of Canada and Alberta Ingenuity Fund. References Archer J.A., Richardson E.C., Herd R.M. & Arthur P.F. (999) Potential for selection to improve feed efficiency of beef cattle: a review. Australian Journal of Agricultural Research 50, Arthur P., Archer J., Johnston D., Herd R., Richardson E. & Parnell P. (200a) Genetic and phenotypic variance and covariance components for feed intake, feed efficiency and other postweaning traits in Angus cattle. Journal of Animal Science 79, Arthur P., Renand G. & Krauss D. (200b) Genetic and phenotypic relationships among different measures of growth and feed efficiency in young Charolais bulls. Livestock Production Science 68, 3 9. Barendse W., Reverter A., Bunch R.J., Harrison B.E., Barris W. & Thomas M.B. (2007) A validated whole-genome association study of efficient food conversion in cattle. Genetics 76, Churchill G.A. & Doerge R.W. (994) Empirical threshold values for quantitative trait mapping. Genetics 38, Crews D.H. Jr (2005) Genetics of efficient feed utilization and national cattle evaluation: a review. Genetic Molecular Research 4, Devlin B. & Risch N. (995) A comparison of linkage disequilibrium measures for fine-scale mapping. Genomics 29, Falconer D.S. & Mackay T.F.C. (996) Introduction to Quantitative Genetics, 4th edn. Longman Scientific and Technical, New York, NY. Hegarty R.S., Goopy J.P., Herd R.M. & McCorkell B. (2007) Cattle selected for lower residual feed intake have reduced daily methane production. Journal of Animal Science 85, Kennedy B.W., van der Werf J.H. & Meuwissen T.H. (993) Genetic and statistical properties of residual feed intake. Journal of Animal Science 7, Koch R.M., Swiger L.A., Chambers D. & Gregory K.E. (963) Efficiency of feed use in cattle. Journal of Animal Science 22, Liu M., Goonewardene L., Bailey D., Basarab J., Kemp R., Arthur P., Okine E. & Makarechian M. (2000) A study on the variation of feed efficiency in station-tested beef bulls. Canadian Journal of Animal Science 80, McKay S.D., Schnabel R.D., Murdoch B.M. et al. (2007) Construction of bovine whole-genome radiation hybrid and linkage maps using high-throughput genotyping. Animal Genetics 38, Moore S., Crews D.H. & Nkrumah D. (2006) Multiple and Candidate Gene Approaches to Genetic Evaluation of Feed Efficiency in Beef Cattle. 8th World Congress on Genetics Applied to Livestock Production, 3 8 August, Belo Horizonte, Brazil. Nkrumah J.D., Basarab J.A., Price M.A. et al. (2004) Different measures of energetic efficiency and their phenotypic relationships with growth, feed intake and ultrasound and carcass merit in hybrid cattle. Journal of Animal Science 82, Nkrumah D., Li C., Wang Z., Bartusiak R., Murdoch B., Basarab J., Crews D. & Moore S. (2005) Full genome scan of quantitative trait loci (QTL) for net feed efficiency in beef cattle. Journal of Animal Science 83(Suppl. ), 3. Nkrumah J.D., Okine E.K., Mathison G.W., Schmid K., Li C., Basarab J.A., Price M.A., Wang Z. & Moore S.S. (2006) Relationships of feedlot feed efficiency, performance and feeding behavior with metabolic rate, methane production and energy partitioning in beef cattle. Journal of Animal Science 84, Nkrumah J.D., Basarab J.A., Wang Z., Li C., Price M.A., Okine E.K., Crews D.H. & Moore S.S. (2007a) Genetic and phenotypic relationships of feed intake and different measures of feed efficiency with growth and carcass merit of beef cattle. Journal of Animal Science 85, Nkrumah J.D., Sherman E.L., Li C., Marques E., Crews D.H. Jr, Bartusiak R., Murdoch B., Wang Z., Basarab J.A. & Moore S.S. (2007b) Primary genome scan to identify putative QTL for feedlot growth rate, feed intake and feed efficiency of beef cattle. Journal of Animal Science 85,

7 Feed efficiency in beef cattle 23 Oliphant A., Barker D.L., Stuelpnagel J.R. & Chee M.S. (2002). BeadArray technology: enabling an accurate, cost-effective approach to high-throughput genotyping. BioTechniques 56(Suppl. 8), 60. Schnabel R.D., Kim J.J., Ashwell M.S., Sonstegard T.S., Van Tassell C.P., Connor E.E. & Taylor J.F. (2005) Fine-mapping milk production quantitative trait loci on BTA6: analysis of the bovine osteopontin gene. Proceedings of the National Academy of Sciences of the United States of America 02, Seaton G., Haley C.S., Knott S.A., Kearsey M. & Visscher P.M. (2002) QTL EXPRESS: mapping quantitative trait loci in simple and complex pedigrees. Bioinformatics 8, Sherman E.L., Nkrumah J.D., Murdoch B.M., Li C., Wang Z., Fu A. & Moore S.S. (2008) Polymorphisms and haplotypes in the bovine NPY, GHR, GHRL, IGF2, UCP2 and UCP3 genes and their associations with measures of growth, performance, feed efficiency and carcass merit in beef cattle. Journal of Animal Science 86, 6.