SHORT COMMUNICATION doi:10.1111/j.1365-2052.2009.01907.x located on a SSC17 meat quality QTL region are associated with growth in outbred pig populations

A. M. Ramos*, J. W. M. Bastiaansen†, G. S. Plastow‡ and M. F. Rothschild* *Department of Animal Science and Center for Integrated Animal Genomics, Iowa State University, Ames, IA 50011, USA. †Animal Breeding and Genomics Centre, Animal Breeding and Genetics Group, Wageningen University, 6709 PG Wageningen, The Netherlands. ‡Department of Agricultural, Food and Nutritional Science, University of Alberta, Edmonton, Alberta T6G 2P5, Canada

Summary The objective of this study was to evaluate the effect of markers developed in eight genes, located in a previously detected meat quality QTL region on SSC17, on growth, fat and meat quality traits collected in commercial pig populations of different genetic backgrounds. The genes had been previously mapped to SSC17 as part of a fine-mapping effort. Association analyses were conducted between each marker and the available phenotypic traits. Results showed that three genes (CTSZ, CSTF1 and C20orf43) were significantly associated with the growth traits. In addition, CTSZ also impacted on meat colour, with the less favourable genotype for growth being associated with darker meat. The differences observed between genotypes were substantial and may be of economic importance to pig producers. These markers may be useful for selecting for faster growth or improved meat quality.

Keywords commercial pig populations, genetic markers, growth.

In the past three decades, traditional selection methods linkage map, including markers in the CYP24A1 (cyto- based on quantitative genetics have been used to select for chrome P450, family 24, subfamily A, polypeptide 1), DOK5 pigs with faster growth and increased percentage of lean (docking 5), CSTF1 (cleavage stimulation factor, 3¢ pre- muscle. This strategy is believed to have caused an overall RNA, subunit 1, 50 kDa), C20orf43 ( 20 open decrease in several meat quality parameters. Genetic reading frame 43), SPO11 (SPO11 meiotic protein covalently markers associated with all these traits are of interest to the bound to DSB-like (Saccharomyces cerevisiae)), RAE1 [RAE1 pig industry because, when used in combination with per- RNA export 1 homolog (Schizosaccharomyces pombe )], GNAS formance data, they may allow faster improvement of the (GNAS complex locus) and CTSZ (cathepsin Z) genes. The traits of economic importance without decline in muscle SSC17 position (in cM) of these genes on the BY population quality. linkage map is given in parenthesis and was CYP24A1 Several QTL for meat quality traits were previously (85.3), DOK5 (88.3), CSTF1 (92.4), C20orf43 (92.6), identified on pig chromosome 17 (SSC17) (Malek et al. SPO11 (97.2), RAE1 (98.5), GNAS (107.3) and CTSZ 2001), using a Berkshire · Yorkshire (BY) resource popu- (108.2). lation. These QTL were for loin meat colour (subjective score To investigate the effects of these markers in outbred pig and 48-h Hunter and Minolta L values), average lactate and populations, several growth, fat and meat quality pheno- average glycolytic potential. A detailed explanation types, as well as DNA samples, were collected in four regarding these phenotypic measurements is provided by commercial pig lines. The measured phenotypes were Malek et al. (2001). Parent of origin QTL for early growth weight at end of test period (kg), days to market weight, traits was also identified in the same SSC17 region lifetime daily gain (g/day), daily gain during test period (Thomsen et al. 2004) and QTL for other growth traits were (g/day), backfat thickness at the P2 position (mm), detected in different SSC17 regions (Pierzchala et al. 2003). Hennessy probe backfat thickness (mm), pH, and Minolta An effort to fine map these QTL was subsequently under- and Japanese colour score (JPCS) measurements. The pH taken by adding several markers to the SSC17 genetic and colour measurements were taken 24 h post-mortem in longissimus dorsi (pH, JPCS) or semimembranosus (ham Address for correspondence Minolta) muscle samples. The Japanese colour score is a M. F. Rothschild, Department of Animal Science and Center for Integ- method used to assess meat colour using a scale from 1 rated Animal Genomics, Iowa State University, Ames, IA 50011, USA. (very pale, light pink meat) to 6 (very dark red meat). The E-mail: [email protected] breed composition of these lines included purebred Landrace Accepted for publication 23 March 2009 (LR) and Large White (LW) as well as crossbred

774 2009 The Authors, Journal compilation 2009 Stichting International Foundation for Animal Genetics, Animal Genetics, 40, 774–778 Genes associated with growth 775

Duroc · Large White (D · LW) and synthetic (SYN) lines. ing were performed using an FDR approach as implemented The number of animals investigated in each line was dif- in the package q-value (Storey & Tibshirani 2003) in R ferent and varied from trait to trait, as some of the studied (http://www.r-project.org). animals did not have phenotypic information for all traits. Significant (P < 0.05) associations with the available The number of animals considered per line varied from 292 growth traits were detected for markers in the CTSZ, CSTF1 to 527 in LR, 163 to 344 in LW, 394 to 629 in D · LW and and C20orf43 genes (Tables 1 and 2), as well as additional 84 to 169 in SYN. Prior to genotyping the entire dataset, associations with meat colour and pH. For each of these each marker was genotyped in a smaller sample of unre- markers, a single favourable genotype associated with faster lated individuals from each line to determine if the marker growth was identified. The results regarding the other was polymorphic in that specific line. markers tested provided no evidence of any significant Association analyses with the available phenotypes were associations with the traits analysed. conducted for all genes using the PCR-RFLP tests developed Animals carrying the CTSZ genotype g.557AA displayed for each . Details regarding the PCR-RFLP tests were higher weight at end of test period (P < 0.01) and higher previously described (Ramos et al. 2006) or are provided in average daily gain on test (P < 0.01) and consequently they Table S1. Nomenclature for genotype identification was spent fewer days to reach market weight (P < 0.01). These determined by consulting the adequate rules (http:// results were obtained when data from the lines where the www.hgvs.org/mutnomen/). The details regarding the marker was polymorphic (LR, LW and SYN) were analysed naming of the SNPs according to the official nomenclature together (Table 1). Similar results were observed when each rules are indicated in Table S2. Data were analysed with a line was analysed individually (Table S4), even though the mixed model that included slaughter date and marker associations were less significant. This may possibly be genotype as fixed effects and sire as random effect. Additive explained by the smaller sample size of the dataset available and dominance effects for each marker and trait combina- for each individual line, which decreased the statistical tion were calculated using the mixed model mentioned power to detect associations between this marker and the previously, which also included the additive and dominance traits analysed. Ideally, studies within-line using larger coefficients and are shown in Table S3. The model for datasets should be conducted to investigate the effect of the analyses that combined data from markers polymorphic in CTSZ marker on each individual line. No evidence was more than one pig population also included line as an found for a significant interaction between lines. Recently, a additional fixed effect. Significant differences were declared similar effect of the CTSZ gene on porcine growth traits was when the marker genotype effect was a significant described in an Italian Large White population (Russo et al. (P < 0.05) overall source of variation and/or the P-value for 2008). Moreover, the SNP used by Russo et al. (2008) the difference between the least squares means for each was the same SNP used in this study, allowing a direct marker genotype was <0.05. Corrections for multiple test- comparison of the SNP effect in different studies and

Table 1 Least squares means, standard errors P-values and q-values for the association analysis of CTSZ with growth and meat colour phenotypes in outbred pig populations.

CTSZ genotypic least squares means

Trait g.557AA g.557AG g.557GG P-value q-value

Weight at end test period (kg) 112.4 ± 0.58a,e 111.4 ± 0.40b,c 110.3 ± 0.49f,d 0.005 0.18 (216)* (580) (294) Days until market weight 155.5 ± 1.03e 157.0 ± 0.77a 158.6 ± 0.90f,b 0.02 0.24 (139) (336) (181) Life time daily gain (g/day) 667.0 ± 3.21c 662.2 ± 2.08 657.8 ± 2.64d 0.06 0.36 (216) (580) (294) Daily gain on test period (g/day) 887.4 ± 5.57a,e 877.0 ± 3.54b 869.6 ± 4.52f 0.04 0.36 (189) (517) (267) Ham Minolta L value** 47.33 ± 0.44 47.70 ± 0.31e 46.62 ± 0.37f 0.02 0.24 (112) (279) (157) Japanese colour score*** 3.30 ± 0.07a 3.41 ± 0.05b 3.38 ± 0.06 0.22 0.66 (136) (365) (190)

Data derived from Landrace, Large White and one synthetic line of pigs were jointly analysed. Significance levels for the differences between genotypic means: a, b = P < 0.1; c, d = P < 0.05; e, f = P < 0.01. *Number of animals; **Minolta lightness (L*) score, light reflection measurement taken on the surface of meat (lower values indicate darker meat); ***Subjective score of pork colour (six classes, higher values indicate darker meat)

2009 The Authors, Journal compilation 2009 Stichting International Foundation for Animal Genetics, Animal Genetics, 40, 774–778 776 Ramos et al.

Table 2 Least squares means, standard errors, P-values and q-values for the association analysis of CSTF1 and C20orf43 with growth phenotypes in a Duroc · Large White pig population.

CSTF1 genotypic least squares means

Trait g.108TT g.108TC g.108CC P-value q-value

Weight at end test period (kg) 110.9 ± 0.54c 111.0 ± 0.53c 113.0 ± 0.96d 0.1 0.37 (236)* (265) (69) Days until market weight 159.6 ± 0.78c 159.3 ± 0.75c 156.2 ± 1.39d 0.07 0.37 (163) (184) (47) Life time daily gain (g/day) 669.7 ± 3.42a 668.6 ± 3.35c 682.6 ± 6.06b,d 0.09 0.37 (236) (265) (69) Daily gain on test period (g/day) 857.6 ± 6.19a 850.4 ± 6.03c 878.9 ± 10.4b,d 0.04 0.37 (166) (196) (54)

C20orf43 genotypic least squares means

Trait g.653_674del/653_674del g.653_674del/653 g.653/653 P-value q-value

Weight at end test period (kg) 112.3 ± 0.85a 111.8 ± 0.50a 110.5 ± 0.52b 0.08 0.37 (86) (283) (260) Days until market weight 157.1 ± 1.22c 158.2 ± 0.70a 159.9 ± 0.73d,b 0.06 0.37 (57) (200) (178) Life time daily gain (g/day) 680.1 ± 5.36c 674.2 ± 3.18 667.6 ± 3.31d 0.08 0.37 (86) (283) (260) Daily gain on test period (g/day) 870.4 ± 9.30a 859.6 ± 5.65 852.9 ± 5.83b 0.24 0.65 (64) (195) (186) 24-h loin pH 5.72 ± 0.02 5.72 ± 0.01c 5.69 ± 0.01d 0.04 0.37 (80) (266) (251)

Significance levels for the differences between genotypic means: a, b = P < 0.1; c, d = P < 0.05; e, f = P < 0.01. *Number of animals.

populations. However, despite the significant associations faster growth or improved meat quality. The q-values detected with average daily gain, weight of lean cuts, ham obtained for the CTSZ marker (Table 1) varied from 0.16 weight and feed to gain ratio, the direction of the effects to 0.36, except for JPCS, which had a q-value of 0.66. The detected in the study by Russo et al. (2008) differed from the q-values derived from the FDR analysis provide a conser- results reported here. In fact, while in our study CTSZ allele vative estimate of the proportion of results that are falsely g.557A was associated with faster growth, the opposite was positive. Ideally, these q-values should be small, but for the observed in the Italian Large White population analysed by CTSZ marker, some of the values were somewhat larger. Russo et al. (2008), where CTSZ allele g.557G was the Assuming a q-value of 0.2 as a threshold for significance preferred allele for average daily gain. Nevertheless, the would render a single raw P-value as significant. Conse- similar findings detected in two independent studies rein- quently, the possibility that some of the associations for the force the notion that the CTSZ gene is involved in the reg- CTSZ marker are false positives can not be ruled out. ulation of porcine growth traits, even though the effects The results obtained for the CSTF1 and C20orf43 genes may differ between different populations. were similar (Table 2). As these two genes are located close In our study, additional associations with several meat to each other on SSC17 (Hart et al. 2007), linkage dis- quality traits were also identified, with results showing that equilibrium cannot be ruled out as a possibility to explain genotype g.557GG, associated with slower growth, dis- this result. These genes were polymorphic only in the played darker meat colour. This fits the commonly held D · LW line and for both one genotype significantly asso- belief that (selection for) higher growth is associated with ciated with faster growth was identified (CSTF genotype paler meat, partly due to a greater proportion of fast muscle g.108CC and C20orf43 genotype g.653_674del/ fibres. Moreover, in the pig lines where this marker was g.653_674del). For the CSTF1 marker, the results detected polymorphic, the frequency of allele g.557A was interme- were in agreement with the results previously observed in diate (35–63%), indicating that the unfavourable allele for the BY population, where allele g.108C was also associated growth is still present at relatively high frequencies. These with faster growth (Table S5). In the D · LW line analysis results suggest that this CTSZ marker could be used in pig presented here, genotype g.108CC is preferred for all selection programmes to differentiate lines towards either growth traits (P < 0.05). Near identical results were

2009 The Authors, Journal compilation 2009 Stichting International Foundation for Animal Genetics, Animal Genetics, 40, 774–778 Genes associated with growth 777 observed for C20orf43, for which genotype g.653_674del/ year and their market value is considered. Given the g.653_674del was found to be associated with higher remarkable progress observed in recent years in several weights and faster on-test growth (P < 0.1) and with faster fields of research in pig genetics and genomics, it is likely lifetime growth and days to market weight (P < 0.05). For that, in the future, the molecular dissection of these traits both markers, no other significant associations with any of will be performed with genomic selection using thousands of the fatness and meat quality traits were detected. This was genotypes per animal. in contrast with the results observed for CTSZ, where the best genotype for growth was associated with darker meat Acknowledgements and vice versa. The only exception occurred for the asso- ciation of C20orf43 with loin pH, where genotype Financial support for Antonio Marcos Ramos was provided g.653_674del/g.653_674del was associated with higher by FCT Fellowship BD/6877/2001. This work was also values of loin pH (P < 0.05), which is desirable. Hence, supported financially in part by Sygen International and the selection for faster growth using these two markers could be Iowa Agriculture and Home Economics Experimental Sta- considered. In addition, the best genotypes of CSTF1 and tion, State of Iowa and Hatch funds. C20orf43 for growth were the less frequent in this D · LW line, further increasing the potential for using these mark- References ers for selection. Interestingly the CSTF1 and C20orf43 al- leles associated with slower growth were either fixed or at a Hart E.A., Caccamo M., Harrow J.L., Humphray S., Gilbert J.G.R., low frequency in the LR, LW and SYN lines, which have Trevanton S., Hubbard T., Rogers J. & Rothschild M.F. (2007) been selected mainly for maternal (LR and LW) or meat Lessons learned from the initial sequencing of the pig genome: quality (SYN) traits. These results again reinforce the gen- comparative analysis of an 8 MB region of pig chromosome 17. eral negative correlation between growth and other traits, Genome Biology 8, R168. Malek M., Dekkers J.C., Lee H.K., Baas T.J., Prusa K., Huff-Lonergan which is managed through the use of multi-trait selection E. & Rothschild M.F. (2001) A molecular genome scan analysis to indices. For the traits shown in Table 2, q-values of 0.36 identify chromosomal regions influencing economic traits in the were obtained, except for the association between the pig. II. Meat and muscle composition. Mammalian Genome 12, marker on C20orf43 and daily gain on test period, which 637–45. presented a q-value of 0.65. The larger q-values for the Pierzchala M., Cieslak D., Reiner G., Bartenschlager H., Moser G. & associations detected for the CSTF1 and C20orf43 markers Geldermann H. (2003) Linkage and QTL mapping for Sus scrofa are all above 0.2, which is in line with the raw P-values chromosome 17. Journal of Animal Breeding and Genetics 120, observed, where some of the associations were only sug- 132–7. gestive (0.05 < P < 0.1). Hence, these results indicate that Ramos A.M., Helm J., Sherwood J., Rocha D. & Rothschild M.F. some of the associations detected for these two markers (2006) Mapping of 21 genetic markers to a QTL region for meat should be carefully interpreted as some of them may be false quality on pig chromosome 17. Animal Genetics 37, 296–7. Russo V., Fontanesi L., Scotti E., Beretti F., Davoli R., Nanni Costa positives. L., Virgili R. & Buttazzoni L. (2008) Single nucleotide polymor- In conclusion, three genes significantly affected the phisms in several porcine cathepsin genes are associated with growth traits analysed, with substantial differences growth, carcass and production traits in Italian Large White pigs. observed between genotypes. These results were obtained Journal of Animal Science 86, 3300–14. using a dataset containing a reasonable number of Storey J.D. & Tibshirani R. (2003) Statistical significance for individuals from outbred commercial pig lines of different genome-wide experiments. Proceeding of the National Academy of genetic background. The fact that genes located in a QTL Sciences 100, 9440–5. region for meat quality and early growth traits were Thomsen H., Lee H.K., Rothschild M.F., Malek M. & Dekkers J.C.M. significantly associated with growth may indicate that (2004) Characterization of quantitative trait loci for growth and selection for faster growth and increased leanness in the meat quality in a cross between commercial breeds of swine. commercial breeds and/or lines of pigs may be fixing Journal of Animal Science 82, 2213–28. favourable growth alleles, while less selected breeds (such as the Berkshire breed used to establish the BY population) Supporting information may still retain the favourable alleles for meat quality traits. Additional supporting information may be found in the For all markers, the favourable genotype displayed an online version of this article. advantage of 2 kg at the end of the test period, and 3 less Table S1 Information regarding the primers, PCR condi- days spent to reach market weight. Significant additive tions, position of SNPs, restriction enzymes used and PCR- effects on the four growth traits analysed were found for all RFLP fragment sizes for the markers in the eight SSC17 markers. These differences would likely be of economic genes analysed. importance and illustrate the potential benefit of using Table S2 Information regarding the naming of the markers molecular markers for the improvement of porcine growth according to official nomenclature rules, including reference traits, especially when the number of pigs slaughtered every

2009 The Authors, Journal compilation 2009 Stichting International Foundation for Animal Genetics, Animal Genetics, 40, 774–778 778 Ramos et al.

sequence used, allele codes and match with PCR-RFLP Table S5 Least squares means and standard errors for the alleles. association analysis of CSTF1 with daily gain on test period

Table S3 Information regarding the additive and dominance in an F2 experimental population of Berkshire · Yorkshire effects determined for the CTSZ, CSTF1 and C20orf43 mar- origin. ker-trait associations. Please note: Wiley-Blackwell is not responsible for the Table S4 Least squares means and standard errors for the content or functionality of any supporting information association analysis of CTSZ with growth and meat colour supplied by the authors. phenotypes in outbred pig populations.

2009 The Authors, Journal compilation 2009 Stichting International Foundation for Animal Genetics, Animal Genetics, 40, 774–778