Skip to main content
  • Research article
  • Open access
  • Published:

Genome-wide association studies (GWAS) identify a QTL close to PRKAG3 affecting meat pH and colour in crossbred commercial pigs

Abstract

Background

Improving meat quality is a high priority for the pork industry to satisfy consumers’ preferences. GWAS have become a state-of-the-art approach to genetically improve economically important traits. However, GWAS focused on pork quality are still relatively rare.

Results

Six genomic regions were shown to affect loin pH and Minolta colour a* and b* on both loin and ham through GWAS in 1943 crossbred commercial pigs. Five of them, located on Sus scrofa chromosome (SSC) 1, SSC5, SSC9, SSC16 and SSCX, were associated with meat colour. However, the most promising region was detected on SSC15 spanning 133–134 Mb which explained 3.51% - 17.06% of genetic variance for five measurements of pH and colour. Three SNPs (ASGA0070625, MARC0083357 and MARC0039273) in very strong LD were considered most likely to account for the effects in this region. ASGA0070625 is located in intron 2 of ZNF142, and the other two markers are close to PRKAG3, STK36, TTLL7 and CDK5R2. After fitting MARC0083357 (the closest SNP to PRKAG3) as a fixed factor, six SNPs still remained significant for at least one trait. Four of them are intragenic with ARPC2, TMBIM1, NRAMP1 and VIL1, while the remaining two are close to RUFY4 and CDK5R2. The gene network constructed demonstrated strong connections of these genes with two major hubs of PRKAG3 and UBC in the super-pathways of cell-to-cell signaling and interaction, cellular function and maintenance. All these pathways play important roles in maintaining the integral architecture and functionality of muscle cells facing the dramatic changes that occur after exsanguination, which is in agreement with the GWAS results found in this study.

Conclusions

There may be other markers and/or genes in this region besides PRKAG3 that have an important effect on pH and colour. The potential markers and their interactions with PRKAG3 require further investigation

Background

Improving meat quality has become a high priority for the pork industry to satisfy consumers’ preferences for a better eating experience. This is vital for the swine industry to sustain profitability and enhance its competiveness in terms of both swine genetics and sales of pork (including export markets). Minolta colour and pH are two common industry measures of pork quality that are directly correlated with glycogen and glycolytic potential [1]. Colour and pH are also highly correlated with other pork quality measurements (e.g. drip loss, texture score) and carcass yield (e.g. carcass weight, loin depth, loin length) as reported in our recent work [2]. However, improving these traits by traditional breeding methods is a challenge due to the wide range of reported heritability (0.04 - 0.57) in different breeds [3], the high cost of measurement and the limited amount of data available post-mortem. In recent decades, nearly 900 QTLs (PigQTLdb, release 25, December 2014. http://www.animalgenome.org/cgi-bin/QTLdb/SS/index) and a few gene variants, such as PRKAG3 (Protein Kinase, AMP-Activated, Gamma 3 Non-Catalytic Subunit) [1,4,5], have been found to affect pork pH and colour, which provides a new breeding strategy to improve these traits on the basis of DNA markers.

Most of the QTLs identified were detected using linkage mapping and cover large regions of the genome. With the availability of high-density SNP chips and genome wide analysis methodology, GWAS have been increasingly used to identify more precisely the genomic regions and markers associated with quantitative traits, and have begun to be used in dairy cattle and poultry breeding programs as reviewed by Goddard and Hayes [6]. Recently, additional GWAS have been reported in pigs and several genomic regions have been identified for different traits [7-13]. However, because of the difficulty and expense of measurement for meat quality, GWAS focused on these traits have still been limited for now [14-18].

Therefore, the aims of this study were to detect genomic regions and identify potential candidate genes and markers affecting pork pH and colour by combining GWAS and post-GWAS bioinformatics analysis. The results will contribute useful information for marker based genomic improvement of these traits. Knowledge of the gene networks will provide new insights for the biological mechanisms underlying these traits.

Results

Genomic regions

Population stratification with Identity-by-State (IBS) showed that all animals from the two commercial populations could be classified into two clusters, which correspond to the populations from which the individuals were derived. Therefore, population was fitted as a fixed effect to optimize the association analysis model.

Genomic regions affecting both fresh and thawed loin pH, colour a* and b* on fresh ham, and colour b* on thawed loin muscle were detected by Bayes B 1 Mb window analysis. Accordingly, a larger proportion of genetic variance explained by the region indicates a stronger effect of the corresponding region on the target trait. The genomic regions that explained more than 1% of the total genetic variance for the traits studied are shown in Table 1. The most promising genomic region associated with the pH and colour was located on SSC15 between 133 - 134 Mb (15_133). A total of 26 SNPs in this region explained the major proportion of the total genetic variance for the five traits, with 17.06%, 8.68%, 12.91%, 9.26% and 3.51% for colour b* (QFCOL b*) and a* (QFCOL a*) measured on the quadriceps femoris on the ham, colour b* measured on the thawed loin (TMCOL b*), pH 24 h post-mortem measured on fresh loin (FpH24) and pH measured on thawed loin (TMpH), respectively. Other genomic regions on SSC1, SSC5, SSC9, SSC16 and SSCX also explained a relatively large proportion of genetic variance of colour. For example, the region on SSC5 (5_107) explained 4.44% of the genetic variance for QFCOL a*, and each of the others explained about 1% of the total genetic variance for QFCOL a* and TMCOL b*.

Table 1 Promising chromosome regions for pH and colour detected by Bayes B analysis

Detailed analysis of the region on SSC15

The region of SSC15_133 spans 898,069 base pairs and includes a total of 28 SNPs on the 60K SNP panel. Two SNPs (1 and 17) with low genotype call rates were filtered out from the analysis. An overview of the distribution of all the SNPs and their relationships is shown in the LD haplotype map in Figure 1. Out of the three haplotype blocks identified, block 2 consisting of five SNPs (18 - 22) is the most interesting one as it harbors PRKAG3 which was previously reported as a candidate gene associated with meat quality. The LD between block 2 and SNP27 is also strong.

Figure 1
figure 1

Haplotype block pattern (r 2 -scheme) for the region SSC15_133 based on the LD (r 2 ) among the 28 SNPs within this region. The numbers on the top indicate the SNP order in this region; SNPs in the red circles are the most likely markers detected from GWAS; The green line indicates the approximate map position of the PRKAG3 gene; The SNPs grouped in each triangle box means they are grouped in one block based on LD information.

For further study, window (15_133) GEBVs for the traits were summed based on the marker effects of the 26 SNPs, and then regressed on the genotype covariate for each SNP (0/1/2). Three SNPs (20, 21 and 27) with almost complete LD captured an average of 98.2%, 99.2% and 99.0% variances of window GEBVs. These results indicated that any of the three SNPs could be affecting the traits. To further analyze these markers, SNP21 (MARC0083357) was chosen as a fixed effect in the animal model (ASREML) to estimate its allele substitution effect. The results (Table 2) showed that SNP21 had a significant effect (P < 0.0001) on all traits. Allele B has positive effects on QFCOL a*, QFCOL b* and TMCOL b*, and negative effects on both FpH24 and TMpH. Further analysis indicated that SNP21 can capture almost all of the phenotypic variance that is explained by this region, e.g. 1% for pH and 1.38% - 1.75% for colour. While, as shown in Table 2, the percentage of phenotypic variance explained by SNP21 (from ASREML) is even slightly higher than that explained by the region. This is because the proportions of phenotypic variance explained by the window were calculated based on the molecular heritability (the percentage of the whole phenotypic variance explained by the fitted SNPs). It is usually considered that the molecular heritability based on the genotypes is smaller than that estimated based on pedigree due to the missing values from the unconsidered markers.

Table 2 Effects and proportions of phenotypic variance explained by SNP21 (MARC0083357)

To investigate other SNPs in this region, each SNP was fitted as a fixed effect to test their statistical significance on the traits after the effect of SNP21 was accounted for in the model. The P_values for the significance test of all the SNPs are shown in Table 3. Only a few of the SNPs remained significant (P < 0.05) after accounting for the effect of SNP21, such as SNP12 for QFCOL b*; SNPs 9, 13, 19 and 28 for QFCOL a*; SNPs 16 and 18 for TMCOL b*; SNPs 14, 18, 19, 22, 23 and 24 for TMpH. To reduce the overestimation of single marker regression, all these significant SNPs, including SNP21, were fitted as fixed effects in the model simultaneously to retest their significance. The results showed that a reduced number of SNPs remained significant (P < 0.05). These were SNPs 12 and 21 for QFCOL b*; SNPs 9, 13 and 21 for QFCOL a*; SNPs 16 and 21 for TMCOL b*; SNPs 14, 21 and 24 for TMpH.

Table 3 Single marker statistical significance test ( P_values ) for other SNPs in the SSC15_133 region after accounting for the effect of SNP21 (MARC0083357)

Haplotypes were constructed for the significant SNPs identified in the multiple marker association analysis. For FpH24, none of the other SNPs was significant after fitting SNP21 in the model. The least squares means (LSMs) of the haplotypes for other traits are shown in Figure 2. For QFCOL b* (Figure 2 (A)), the B allele of SNP21 had a positive effect (BB > AB > AA). SNP12 showed an over-dominance effect. Therefore, the genotype BBAB (SNP 21–12) had the largest value for colour b* and AAAA animals the lowest value.

Figure 2
figure 2

Least squares mean (LSMs) for the haplotypes constructed for the significant SNPs from multiple-marker association. (A): LSMs of QFCOL b* for the haplotypes constructed by SNP21 and SNP12. AA, AB and BB are the three genotypes for each SNP. (B): LSMs of QFCOL a* for the haplotypes constructed by SNP21, SNP13 and SNP9. (C): LSMs of TMCOL b* for the haplotypes constructed by SNP21 and SNP16. (D): LSMs of TMpH for the haplotypes constructed by SNP21, SNP24 and SNP14.

For QFCOL a* (Figure 2 (B)), three SNPs (21, 9 and 13) remained significant and resulted in 16 major haplotypes accounting for 99% of the whole sample. For the haplotype SNP 21-9-13, the largest value was observed for BBABBB animals (n = 19) and the lowest value was for the animals with AAABAA (n = 23) or AAAAAB (n = 208). However, the small sample size for these groups might have resulted in a larger standard error. These results need further validation in a larger sample size.

For TMCOL b* (Figure 2 (C)), SNP21 and SNP16 remained significant and resulted in 6 haplotypes. Individuals with BBBB (n = 509) accounted for 26.2% of the population and had the highest b* value. Individuals with AAAA genotype (n = 50) showed the lowest value. However further study is encouraged due to the small number of AAAA animals in this sample.

For TMpH (Figure 2 (D)), three SNPs of 21, 14 and 24 remained significant and resulted in 10 haplotypes. Haplotype AAAAAB (SNP 21-14-24) (n = 257) had the highest pH compared with the haplotype BBAAAA (n = 87) with the lowest value. However, no significant difference was detected between other haplotypes, which indicated that the main effect seems to be due to SNP21.

Candidate genes and their functional network

A total of 29 genes including 17 of unknown function were identified in a 1 Mb region on SSC15. The nine significant SNPs (including 20, 21 and 27) were found to be intragenic or close to ten genes as shown in Table 4. SNP21 is close to ZNF142 and STK36, and about 62 kb upstream of PRKAG3. For SNP20 and SNP27, which have very high LD with SNP21, one is located in the intron of ZNF142 (SNP20) and the other is close to CDK5R2 (SNP27). For the other six SNPs, four are intragenic with ARPC2, TMBIM1, NRAMP1 and VIL1, and two are close to RUFY4 and CDK5R2.

Table 4 The candidate or nearest genes to the significant SNPs in the region on SSC15

A functional network (Figure 3) among these ten genes was constructed based on Gene Ontology (GO) and their functional pathways. NRAMP1 was replaced by the human orthologues of SLC11A1 during the process. Except for RUFY4, all the genes (coloured by pink and red) were strongly connected through the functional pathways which are involved in cell-to-cell signaling and interaction, cellular function and maintenance. Usually the genes with darker colour indicate they have more contributions to the pathway and have stronger connection to the network. For example, ZNF142 and PRKAG3 were coloured dark red indicating the greatest contribution to these pathways compared to the other genes that were coloured light red and pink. Gene annotation (Additional file 1: Table S1) indicated that most of these genes are related to the cellular component of cell-substrate junction, membrane and extracellular region, which play important roles in maintaining the internal architecture of muscle cells through signal transduction and a variety of biological processes. These gene products might play important roles in regulating post-mortem effects that impact muscle pH and colour.

Figure 3
figure 3

Gene network constructed based on the candidate and/or nearest genes to the significant SNPs in the SSC15_133 region using IPA software. The genes marked by red and pink means they are candidate genes or the closest ones to the significant SNPs detected in the region. The darker means the greater contribution involved in the functional pathway. The white ones were selected automatically by IPA procedure based on their functional annotations that are related to the pathways.

Analysis of other genomic regions

Candidate genes in the regions identified on other chromosomes are presented in Table 5. Thirteen genes were identified in the region on SSC1 impacting QFCOL a*, where the most likely marker was H3GA0001381 which explained 91.5% of the window GEBV variance. For SSC5, only one known gene (SYT1) was detected in this region where MARC0017490 was the most promising marker (RSQ = 0.975) for QFCOL a*. Three genes were found for SSC9_147 where three SNPs with high LD (r2 = 0.63 - 0.89) contributed the major effect (each of them explained more than 87% variance of the window GEBV). Only one gene was identified for the region on SSC16 and the most promising SNP was ALGA0091417. For the region on SSCX, three genes were identified and ASGA0083984 was the best marker explaining almost all the variance of window GEBV (RSQ = 0.998).

Table 5 Genes detected in other regions and the most likely SNPs in the regions

Discussion

Genomic regions for pork pH and colour

In this study, a 1 Mb region (133 – 134 Mb) on SSC15 explained the majority of the genetic variance for loin pH, loin and ham colour a* and b*. Previous linkage mapping studies have reported that SSC15 is enriched for QTLs affecting meat pH and colour. These include five QTLs between 56 cM to 119 cM for ham and loin pH 24 h post-mortem [1,19], and a large region of 88.5 - 122 cM [20] as well as peaks located at 42.21 cM and 44.80 cM [21] for colour b*. A recent GWAS study in the Finnish Yorkshire breed detected a relatively small region (133.64 - 134.01 Mb) on SSC15 associated with meat quality [18]. This region was also identified in this study. These results provide further support for the region of 133 - 134 Mb on SSC15 being a true QTL affecting meat pH and colour, especially as it is found in different populations. A positional candidate gene approach also showed that PRKAG3, located within this region, can be considered as one of the few major genes known to affect meat pH and colour. Four missense substitutions (T30N, G52S, I199V and R200Q) in this gene were found to significantly affect meat quality [1,4]. However, since none of these mutations are included in the 60K SNP panel, the contribution of PRKAG3 to the effect of the region on the traits remains unknown. Therefore, it is essential to extend the analysis of the regions flanking PRKAG3 and investigate the genomic architecture of this region affecting meat pH and colour.

Five other regions, on SSC1, SSC5, SSC9, SSC16 and SSCX, were also found to capture a relatively large proportion of genetic variance for meat colour in this study, indicating their important effects on these traits. However, the limited gene and map information in these regions restricted further analysis of these QTLs. For example, thirteen genes including 8 of unknown function were identified in the region on SSC1. Only a few (1 - 3) genes were detected in the other regions and none of them harbours SNPs on the 60K SNP panel. Previous GWAS studies in the pig have reported a few QTLs besides SSC15 to affect meat quality in different pig breeds/populations. Nonneman et al. [16] detected regions on SSC1 and SSC6 associated with multiple meat quality traits in a Landrace-Duroc-Yorkshire population. Sanchez et al. [17] identified seven QTLs on SSC1, SSC4, SSC8, SSC9 and SSC13 affecting meat pH and colour in French Large White (n = 385). Other chromosome regions affecting meat pH and/or colour include SSC3 [15], SSC2 and SSC6 [18], SSC4 in Swiss Large White Boars (n = 192) [14] and SSC6 in Pietrain boars [22]. These results indicate that meat pH and colour are determined by multiple genomic regions with relatively small effects, and that the genetic variants might be different in different pig breeds or populations. This complexity should be taken into account when designing a marker assisted selection program for specific breeds or populations.

Further analysis of the major QTL on SSC15

Further analysis of the SSC15 region showed that three SNPs (20, 21 and 27) in very strong LD captured nearly all of the genetic variance of this region. SNP21 (MARC0083357), one of the closest SNPs to PRKAG3 in the 60K panel, showed significant effects on all of the traits studied. PRKAG3 was previously reported as a candidate gene affecting meat quality. The most important substitution (R200Q) in the gene caused a 70% increase in muscle glycogen in RN homozygous and heterozygous pigs that resulted in lower ultimate pH and water-holding capacity [4]. Another three mutations (T30N, G52S, I199V) in PRKAG3 also showed significant effects on pH, and Minolta L* and b* for both loin and ham [1,5]. The most recent study [23] on PRKAG3 detected several novel mutations by sequencing the coding and promoter regions, and the haplotype (g.-157C - g.-58A - 24E - 199I) was found to have a positive effect on meat quality in pigs. In our study, the most significant effects of SSC15_133 region reported here are on a*, b* and pH, but not lightness (L*) (Table 1). However, we also found evidence for effects of this region (SNP21) on other pork quality traits with P < 0.01 (see Additional file 1: Table S2), such as cooking loss, drip loss, shear force, colour a* for both fresh and thawed loin, colour b* on fresh loin, and colour (L*, a*, b*) on ham gluteus medius.

The most likely hypothesis is that the main effect of the QTL detected on SSC15 is due to the effect of polymorphism in PRKAG3. The three SNPs (SNPs 20, 21 and 27) might be in high LD with some of the mutations in PRKAG3 that affect pH and colour. SNP20 (ASGA0070625), in complete LD with SNP21 in this study, was previously reported to affect meat pH in Yorkshire (n = 220) and Finnish Landrace (n = 230) [18]. However, further analysis [23] showed that ASGA0070625 was in very weak LD with the mutations of T30N, G53S and I199V in PRKAG3, but was in complete LD with some novel mutations such as g.-157C > G, g.-58A > G and K24E. These results indicate that there may be other polymorphisms and/or genes associated with variation in meat quality traits. Additional genes in this region may contribute to variation of meat quality independently or by interacting with PRKAG3. Three other known genes (TTLL7, STK36 and ZNF142) are much closer to SNP21 than PRKAG3, and furthermore, SNP20 (ASGA0070625) is located in the intron 2 of ZNF142 (zinc finger protein 142). ZNF142 was reported as a putative candidate gene for both developmental and malignant disorders in human [24]. ZNF is one of the most abundant classes of transcriptional factors that regulate cell growth and differentiation, and ZNF142 is involved in calcium ion binding and cell membrane integrity. The STK36 (Serine/Threonine Kinase 36) gene product is related to the hedgehog signaling pathway and activation of cAMP-dependent PKA. TTLL7 is Member 7 of the Tubulin Tyrosine Ligase-Like Family. SNP27 (MARC0039273) is close to CDK5R2 (Cyclin-Dependent Kinase 5, Regulatory Subunit 2 (P39)), a neuron-specific activator of CDK5 kinase which is associated with cyclin-dependent kinase activating proteins. All these functions are associated with signal transduction, energy supply or cellular maintenance, which may play important roles on the activity of muscle cells post-mortem.

When SNP21 was fitted in the model, another 6 SNPs in this region were still significant for at least one of the studied traits. When the phenotypic LSMs were compared among the different haplotypes constructed with SNP21 and these SNPs, the differences increased compared with the difference among the three genotypes of SNP21. These results indicate that the other SNPs in this region contributed some effect on the traits, and these effects might be due to other genes in this region. Four of the six SNPs are intragenic with ARPC2, TMBIM1, NRAMP1 and VIL1, and two are close to RUFY4 and CDK5R2. ARPC2 (actin related protein 2/3 complex, subunit 2) encodes one of seven subunits of the human Arp2/3 protein complex, which has been implicated in the control of actin polymerization in cells. Actin polymerization and actin/myosin interaction are strongly related to the toughening of muscle during the first 24 h post-mortem [25], consequently affecting meat pH and colour. The orthologue of NRAMP1 (natural resistance-associated macrophage protein one) in human and other species is SLC11A1, which is a member of the solute carrier family 11 and encodes a multi-pass membrane protein that regulates intracellular pathogen proliferation and macrophage inflammatory responses [26]. VIL1 (Villin 1) encodes a member of a family of calcium-regulated actin-binding proteins, it is a major actin-modifying protein associated with the microvillar actin filaments, and regulates epithelial cell morphology, actin reorganization, and cell motility through the functions in the capping, severing, and bundling of actin filaments [27]. These activities are very important to regulate cell death and enhance motility of actin-based proteins to maintain cellular structure and functionality. RUFY4 (RUN and FYVE domain containing 4) is affiliated with the lncRNA class of proteins with roles in metal ion binding. Therefore, the main functions of these genes are related to membrane components and functionality, transporter activities and calcium ion binding, which play important roles in muscle activity related to structural integrity. These may therefore be involved in the dramatic micro-environmental changes that occur post-mortem [28].

Gene network for this region of SSC15

The gene network constructed for the genes identified in the region on SSC15 showed strong interactions among them. PRKAG3 is one of the major hubs in the network connecting with several of the genes found in this region of SSC15 such as TTLL7, VIL1, and SLC11A1. The protein encoded by PRKAG3 is a regulatory subunit of adenosine monophosphate kinase (AMPK) which is an important energy-sensing enzyme that monitors cellular energy status and responses to cellular metabolic stresses. Activated AMPK is expected to inhibit glycogen synthesis and stimulate glycogen degradation. Studies in pig suggested that PRKAG3 may play a key role in the regulation of energy metabolism in skeletal muscle [29], which is directly associated with the shift of the glycolysis between aerobic and anaerobic after exsanguination and consequently impact meat pH, colour and water holding capability [28]. Other genes in the network are directly or indirectly associated with a second major hub, UBC (ubiquitin C). The important biological processes associated with UBC include activation of mitogen-activated protein kinase, cellular response to hypoxia, energy homeostasis and ion transmembrane transport, which are strongly associated with the muscle cell activities that are likely to be impacted post-mortem. These two major hubs are linked by APP (amyloid beta (A4) precursor protein) through AMPK subunits (e.g. PRKAB1 and PRKAB2) and kinesin (e.g. KLC1 and KLC2), also through some mediators associated with energy homeostasis like Ca2+ and ATP. These findings indicate that the genes identified in this region of SSC15 have strong interactions with PRKAG3 and UBC. Based on their annotations, they may work together to impact meat quality through processes related to the regulation of actin filament polymerization, protein/ion/ATP binding, kinase activities and maintaining the cytoskeleton (Additional file 1: Table S1). All these processes are essential to maintain muscle membrane and myofibrillar protein integrity in the face of dramatic metabolic changes after exsanguination, such as the shift from aerobic to anaerobic glycolysis with anoxia and high body temperature, depletion of ATP, accumulation of lactate and hydrogen ions, and the consequent lowering of pH and lightening of meat colour [28,30].

Conclusions

The most significant genomic region affecting both fresh and thawed loin pH, colour a* and b* for fresh ham and colour b* measured on thawed loin muscle has been identified on SSC15 spanning 133 - 134 Mb. Three SNPs (ASGA0070625, MARC0083357 and MARC0039273) with almost complete LD were the markers most strongly associated with the effect in this region and could be considered for marker assisted selection for meat quality improvement. An additional six SNPs also showed significant effects. Ten known genes have been identified to contain or be close to these markers, including the previously reported candidate gene PRKAG3 along with ZNF142, TMBIM1, NRAMP1, VIL1 and ARPC2. These genes appear to be networked through super-pathways of cell-to-cell signaling and interaction, cellular function and maintenance. They may play important roles in maintaining the integral architecture and functionality of muscle cells during the dramatic changes in the microenvironment post-mortem. These findings significantly contribute to our knowledge of the genomic architecture of this region and will potentially lead to new insights of the molecular mechanisms regulating meat pH and colour. Re-sequencing of this region may help resolve the molecular architecture of the region and identify the causative mutations.

Methods

Animals and phenotypic data

All animal procedures related to this project were reviewed and approved by the Canada Animal Care and Use Committee.

A total of 1943 crossbred commercial pigs from two populations (961 samples from Hypor and 982 samples from Genesus) were obtained for this study. All animals were from a typical Canadian three-way cross consisting of a Duroc sire line on an F1 hybrid female (Landrace * Large White). Pigs in both populations were fed ad libitum and harvested from 2010 to 2012. All animals were sent to the same processing plant when the body weight was close to 115 kg. Carcasses were processed within 24 h post-mortem and fresh meat measurements taken at that time, and pork loins (4th - 7th ribs) were harvested from each pig, vacuum packaged, and chilled (2 - 4°C) within 24 hours after exsanguination. All the packaged loins in the same batch were simultaneously frozen (−20°C) within 24 hours after exsanguination and maintained frozen until meat quality measurement. Prior to testing, the pork loins were thawed for 72 hours at 4°C. All samples were treated in the same way.

Fresh meat pH 24 h post-mortem (FpH24) was measured at two different locations on the surface of the loin muscle at the 10th rib. The average was used for the final statistical analysis. Minolta colours L*, a* and b* were measured at 24 h after exsanguination on the longissimus dorsi muscle (FMCOL) and subcutaneous fat above the longissimus dorsi muscle (FCOL) on the fresh loin, and on the muscle of gluteus medius (GMCOL), quadriceps femoris (QFCOL) and iliopsoas (ICOL) on the fresh ham face. Thawed muscle pH (TMpH) and colours (TMCOL L*, a*, b*) were measured on the thawed loin muscle. pH was measured using a portable pH meter (Fisher Scientific Company, Toronto, Ontario) equipped with a glass electrode (Hanna HI 98121, Hanna instruments, Canada) calibrated at room temperature using standards of pH 4.01 and 7.01. The pH probe was inserted about 5 cm into the centre of the loin three times through a small slit in three different locations within 2.5 cm from the posterior of the longissimus loin. The average was used as the final value for the analysis. Minolta colours were taken from three different locations on the surface of the muscle, and the average value was used for the final analysis. All the colour assessment was made using the colour system values specified by the Commission Internationale de L’eclairage (CIE) L* (lightness), a* (redness) and b* (yellowness), using a Konica Minolta Chroma-meter CR-400 (Konica Minolta Sensing Inc., Japan). Other traits including drip loss on both fresh (FMDL) and thawed (TMDL) loin muscle, cooking loss (CL) and shear force (SF) on thawed muscle were also measured in this project. The details of the measurement were described previously [31].

Population stratification and LD estimation

Population stratification was analyzed using an identical-by-state (IBS) distance clustering method using the PLINK program [32]. Linkage disequilibrium (LD) of the pairwise SNPs was measured using r2 by proc allele program in SAS 9.3. The haplotype block pattern was built using haploView software [33].

SNP array genotyping and filtering

Genomic DNA was isolated from tissue samples following the DNA Extraction instruction manual (Thermo Fisher Scientific Ltd., Ottawa, ON, Canada). Genotyping was conducted by Delta Genomics (Edmonton, AB, Canada) using Illumina PorcineSNP60 V2 Genotyping Beadchip according to the Illumina Infinium Assay (Illumina, Inc., San Diego, CA, USA).

Quality control for genotypes was performed with the criteria of genotyping call rate ≥ 95%, Chi-square of Hardy-Weinberg equilibrium test < 600, and minor allele frequency (MAF) ≥ 5%. After filtering, 1943 individuals and 42456 SNPs remained in the model for the final analysis.

SNP positions were updated according to the newest release from Ensembl (Sscrofa 10.2 genome version). Map information was adjusted based on the LD decay for each chromosome. Only the SNPs with reliable positions were used in this study. The missing genotypes were imputed by Fimpute v2.2 [34].

Quality control

Traits of TMpH and FMDL did not exhibit normal distribution and were transformed using cosine and natural logarithm functions, respectively. All the records were adjusted within population using GLM procedure of SAS 9.3. A contemporary group consisting of slaughter batch, year and group during the test (~70 – 115 kg) was treated as a fixed effect in the model. There were 57 contemporary groups for Genesus and 46 for Hypor. Fixed effects of sex and contemporary group were fitted in the model below. For the Hypor population, the room and pen are fitted in the model as well, as they are available and have significant effect on the traits. The interactions among these factors were not significant and ignored in the model.

$$ {y}_{ij\;k\;l}=\mu +{s}_i+{\mathit{\mathsf{g}}}_j+{r}_k+{p}_l+{e}_{ij\;k\;l} $$

where y ijkl is phenotypic observation, μ is overall mean, s i is the effect of the i th sex (1 = male and 2 = female), g j is the effect of the j th contemporary group (j = 1 - 57 for Genesus and 58 - 103 for Hypor), r k is effect of the k th room (k =1 - 10 for Hypor), p l is the effect of the l th pen (l = 1 - 13 for Hypor), and e ijkl is the random residual error \( \Big[{e}_{ijkl} \sim N\left(0,\ {\sigma}_e^2\Big)\right] \). The adjusted phenotype calculated as residual plus mean was used for the subsequent association studies.

Genome-wide association analyses

The adjusted phenotype for each population were combined together and considered as a new phenotype for the genome-wide association. Population (Hypor and Genesus) was fitted as a fixed effect in the model. The marker effects were estimated by fitting all SNPs simultaneously using Bayesian methods as follow, as implemented in the online software GenSel.

$$ y=Xb+{\displaystyle \sum_j^k{z}_j{\alpha}_j{\delta}_j+\varepsilon } $$

where, y = vector of adjusted phenotypic observations, X = incidence matrix relating fixed factors to phenotypes, b = vector of fixed effects (mean and population), z j  = vector of the genotype covariate for SNP j (j = 1 to k) based on the number of B alleles using Illumina’s (San Diego, California) genotype calling (coded 0, 1, 2), α j = allele substitution effect for SNP j, and δ j  = indicator for whether SNP j was included (δ j  = 1) or excluded (δ j  = 0) in the model for a given iteration of the Markov chain Monte Carlo (MCMC). A total of 50 000 iterations were run for each analysis, with the first 5000 iterations discarded as burn-in. The probability of δ i  = 0 was set equal to π = 0.99. The Bayesian model was implemented using method Bayes B. Genomic regions associated with traits were identified using 1 Mb, non-overlapping windows using Ensembl build 10.2 of the swine genome.

Further marker effect and haplotype analysis

Further analysis on the SNPs within a 1 Mb window was implemented by a mixed animal model in ASREML [35]. The model equation was: Y = Xb + Za + e, where Y is the vector of adjusted phenotypes, b is the vector of fixed effects including mean, population and fitted SNPs, a is the random polygenic effects with \( \Big[u \sim N\left(0,\ \boldsymbol{G}{\sigma}_{\mathrm{a}}^2\right] \), where G is the relationship matrix that was constructed based on pedigree information, and \( {\sigma}_a^2 \) is the polygenetic additive variance, e is the vector of residual errors with a distribution of \( \Big[e\sim N\left(0,\ I{\sigma}_e^2\Big)\right] \), where I is the identity matrix and \( {\sigma}_e^2 \) is the residual variance. X and Z are the incidence matrices associated with b and a.

Window genomic estimated breeding value (GEBV) was calculated by sum of the marker effects of the SNPs within the window. And then the window GEBV was regressed on each SNP variant (0/1/2) within that window. The R-squared (RSQ) for the regression model, defined as the percentage of the GEBV variance explained by the SNP, was used to detect the most likely SNP in that window. The marker effect for the most likely SNP was estimated by fitting the SNP as fixed factor in the above animal model in ASREML. Then each of the remaining SNP was included one at a time in the model to test their significance after accounting for the effect of the most likely SNP. Finally, all the significant SNPs were simultaneously fitted in the model to retest their significance for the traits. Only the significant SNPs from the multiple marker association were used to construct the haplotypes. The least squares mean (LSMs) of the phenotype for each haplotype were estimated and contrasted by GLM procedure in SAS9.3. The significant threshold was set at P < 0.05.

Post-GWAS bioinformatics analysis

Candidate gene identification and functional annotation for the significant SNPs were obtained using Ensembl annotation of Sscrofa 10.2 genome version (http://www.ensembl.org/biomart/martview). Gene networks were explored by the online software Ingenuity Pathways Analysis (IPA) (http://www.ingenuity.com/products/ipa). Human genes were used as background in pathway and gene network investigation, because some genes have not been characterized in pigs and translational gene aspects are of high interest. The functions and related information about candidate genes were summarized using the database of GENECARDS the human gene compendium (http://www.genecards.org/).

Availability of supporting data

The original data sets supporting the results of this article are available. They are not public, but are able to be accessed by request for result verification only.

Abbreviations

GWAS:

Genome-wide association studies

LD:

Linkage disequilibrium

IPA:

Ingenuity pathways analysis

LSMs:

Least squares mean

SNP:

Single nucleotide polymorphism

SSC:

Sus scrofa chromosome

QTL:

Quantitative trait loci

IBS:

Identity-by-State

GEBV:

Genomic estimated breeding value

GO:

Gene ontology

RSQ:

R-squared

References

  1. Ciobanu D, Bastiaansen J, Malek M, Helm J, Woollard J, Plastow G, et al. Evidence for new alleles in the protein kinase adenosine monophosphate-activated gamma(3)-subunit gene associated with low glycogen content in pig skeletal muscle and improved meat quality. Genetics. 2001;159(3):1151–62.

    PubMed Central  CAS  PubMed  Google Scholar 

  2. Miar Y, Plastow GS, Moore SS, Manafiazar G, Charagu P, Kemp RA, et al. Genetic and phenotypic parameters for carcass and meat quality traits in commercial crossbred pigs. J Anim Sci. 2014;92(7):2869–84.

    Article  CAS  PubMed  Google Scholar 

  3. Rothschild MF, Ruvinsky A. Genetics of meat quality and carcass traits. In: The genetics of the pig. 2nd ed. CABI: 2011. p. 355–390.

  4. Milan D, Jeon J-T, Looft C, Amarger V, Robic A, Thelander M, et al. A mutation in PRKAG3 associated with excess glycogen content in pig skeletal muscle. Science. 2000;288(5469):1248–51.

    Article  CAS  PubMed  Google Scholar 

  5. Otto G, Roehe R, Looft H, Thoelking L, Knap PW, Rothschild MF, et al. Associations of DNA markers with meat quality traits in pigs with emphasis on drip loss. Meat Sci. 2007;75(2):185–95.

    Article  CAS  PubMed  Google Scholar 

  6. Goddard ME, Hayes BJ. Mapping genes for complex traits in domestic animals and their use in breeding programmes. Nat Rev Genet. 2009;10(6):381–91.

    Article  CAS  PubMed  Google Scholar 

  7. Boddicker NJ, Bjorkquist A, Rowland RR, Lunney JK, Reecy JM, Dekkers JC. Genome-wide association and genomic prediction for host response to porcine reproductive and respiratory syndrome virus infection. Genet Sel Evol. 2014;46:18.

    Article  PubMed Central  PubMed  Google Scholar 

  8. Do DN, Strathe AB, Ostersen T, Jensen J, Mark T, Kadarmideen HN. Genome-wide association study reveals genetic architecture of eating behavior in pigs and its implications for humans obesity by comparative mapping. PLoS One. 2013;8(8):e71509.

    Article  PubMed Central  CAS  PubMed  Google Scholar 

  9. Onteru SK, Fan B, Du ZQ, Garrick DJ, Stalder KJ, Rothschild MF. A whole-genome association study for pig reproductive traits. Anim Genet. 2012;43(1):18–26.

    Article  CAS  PubMed  Google Scholar 

  10. Onteru SK, Fan B, Nikkila MT, Garrick DJ, Stalder KJ, Rothschild MF. Whole-genome association analyses for lifetime reproductive traits in the pig. J Anim Sci. 2011;89(4):988–95.

    Article  CAS  PubMed  Google Scholar 

  11. Onteru SK, Gorbach DM, Young JM, Garrick DJ, Dekkers JC, Rothschild MF. Whole genome association studies of residual feed intake and related traits in the pig. PLoS One. 2013;8(6):e61756.

    Article  PubMed Central  CAS  PubMed  Google Scholar 

  12. Sahana G, Kadlecova V, Hornshoj H, Nielsen B, Christensen OF. A genome-wide association scan in pig identifies novel regions associated with feed efficiency trait. J Anim Sci. 2013;91(3):1041–50.

    Article  CAS  PubMed  Google Scholar 

  13. Schneider JF, Rempel LA, Snelling WM, Wiedmann RT, Nonneman DJ, Rohrer GA. Genome-wide association study of swine farrowing traits. Part II: Bayesian analysis of marker data. J Anim Sci. 2012;90(10):3360–7.

    Article  CAS  PubMed  Google Scholar 

  14. Becker D, Wimmers K, Luther H, Hofer A, Leeb T. A genome-wide association study to detect QTL for commercially important traits in Swiss Large White boars. PLoS One. 2013;8(2):e55951.

    Article  PubMed Central  CAS  PubMed  Google Scholar 

  15. Ma J, Yang J, Zhou L, Zhang Z, Ma H, Xie X, et al. Genome-wide association study of meat quality traits in a White DurocxErhualian F2 intercross and Chinese Sutai pigs. PLoS One. 2013;8(5):e64047.

    Article  PubMed Central  CAS  PubMed  Google Scholar 

  16. Nonneman DJ, Shackelford SD, King DA, Wheeler TL, Wiedmann RT, Snelling WM, et al. Genome-wide association of meat quality traits and tenderness in swine. J Anim Sci. 2013;91(9):4043–50.

    Article  CAS  PubMed  Google Scholar 

  17. Sanchez MP, Tribout T, Iannuccelli N, Bouffaud M, Servin B, Tenghe A, et al. A genome-wide association study of production traits in a commercial population of Large White pigs: evidence of haplotypes affecting meat quality. Genet Sel Evol. 2014;46:12.

    Article  PubMed Central  PubMed  Google Scholar 

  18. Uimari P, Sironen A, Sevon-Aimonen ML. Evidence for three highly significant QTL for meat quality traits in the Finnish Yorkshire pig breed. J Anim Sci. 2013;91(5):2001–11.

    Article  CAS  PubMed  Google Scholar 

  19. Malek M, Dekkers JC, Lee HK, Baas TJ, Prusa K, Huff-Lonergan E, et al. A molecular genome scan analysis to identify chromosomal regions influencing economic traits in the pig. II. Meat and muscle composition. Mamm Genome. 2001;12(8):637–45.

    Article  CAS  PubMed  Google Scholar 

  20. Gallardo D, Pena RN, Quintanilla R, Ramírez O, Almuzara D, Noguera JL, et al. Quantitative trait loci analysis of a Duroc commercial population highlights differences in the genetic determination of meat quality traits at two different muscles. Anim Genet. 2012;43(6):800–4.

    Article  CAS  PubMed  Google Scholar 

  21. Rohrer G, Nonneman D, Miller R, Zerby H, Moeller S. Association of single nucleotide polymorphism (SNP) markers in candidate genes and QTL regions with pork quality traits in commercial pigs. Meat Sci. 2012;92(4):511–8.

    Article  CAS  PubMed  Google Scholar 

  22. Stratz P, Wellmann R, Preuss S, Wimmers K, Bennewitz J. Genome-wide association analysis for growth, muscularity and meat quality in Pietrain pigs. Anim Genet. 2014;45(3):350–6.

    Article  CAS  PubMed  Google Scholar 

  23. Uimari P, Sironen A. A combination of two variants in PRKAG3 is needed for a positive effect on meat quality in pigs. BMC Genet. 2014;15(1):29.

    Article  PubMed Central  PubMed  Google Scholar 

  24. Tommerup N, Vissing H. Isolation and fine mapping of 16 novel human zinc finger-encoding cDNAs identify putative candidate genes for developmental and malignant disorders. Genomics. 1995;27(2):259–64.

    Article  CAS  PubMed  Google Scholar 

  25. Koohmaraie M, Doumit M, Wheeler T. Meat toughening does not occur when rigor shortening is prevented. J Anim Sci. 1996;74(12):2935–42.

    CAS  PubMed  Google Scholar 

  26. Barton CH, Biggs TE, Baker ST, Bowen H, Atkinson P. Nramp1: a link between intracellular iron transport and innate resistance to intracellular pathogens. J Leukoc Biol. 1999;66(5):757–62.

    CAS  PubMed  Google Scholar 

  27. Wang Y, Srinivasan K, Siddiqui MR, George SP, Tomar A, Khurana S. A novel role for villin in intestinal epithelial cell survival and homeostasis. J Biol Chem. 2008;283(14):9454–64.

    Article  CAS  PubMed  Google Scholar 

  28. Huff-Lonergan E, Lonergan SM. Mechanisms of water-holding capacity of meat: the role of postmortem biochemical and structural changes. Meat Sci. 2005;71(1):194–204.

    Article  CAS  PubMed  Google Scholar 

  29. Hardie DG, Carling D, Carlson M. The AMP-activated/SNF1 protein kinase subfamily: metabolic sensors of the eukaryotic cell? Annu Rev Biochem. 1998;67(1):821–55.

    Article  CAS  PubMed  Google Scholar 

  30. Barnes BR, Marklund S, Steiler TL, Walter M, Hjalm G, Amarger V, et al. The 5′-AMP-activated protein kinase gamma3 isoform has a key role in carbohydrate and lipid metabolism in glycolytic skeletal muscle. J Biol Chem. 2004;279(37):38441–7.

    Article  CAS  PubMed  Google Scholar 

  31. Zhang CY, Wang Z, Bruce HL, Janz J, Goddard E, Moore S, et al. Associations between single nucleotide polymorphisms in 33 candidate genes and meat quality traits in commercial pigs. Anim Genet. 2014;45(4):508–16.

    Article  CAS  PubMed  Google Scholar 

  32. Purcell S, Neale B, Todd-Brown K, Thomas L, Ferreira MA, Bender D, et al. PLINK: a tool set for whole-genome association and population-based linkage analyses. Am J Hum Genet. 2007;81(3):559–75.

    Article  PubMed Central  CAS  PubMed  Google Scholar 

  33. Barrett JC, Fry B, Maller J, Daly MJ. Haploview: analysis and visualization of LD and haplotype maps. Bioinformatics. 2005;21:263–5.

    Article  CAS  PubMed  Google Scholar 

  34. Sargolzaei M, Chesnais J, Schenkel F. A new approach for efficient genotype imputation using information from relatives. BMC Genomics. 2014;15:478.

    Article  PubMed Central  PubMed  Google Scholar 

  35. Gilmour AR, Gogel B, Cullis B, Thompson R. ASReml user guide release 3.0. Hemel Hempstead, UK: VSN International Ltd; 2009.

    Google Scholar 

Download references

Acknowledgments

Thanks are given to all the staff and technicians who collected the samples and phenotypes. The authors are grateful to Dr Mehdi Sargolzaei at the University of Guelph who provided ASREML and FImpute 2.0 programs, Dr Dorian Garrick (Iowa State University) and Dr Nicholas Boddicker (Genesus Inc.) who provided help on GenSel online software. This work was financially supported by grants from the Natural Sciences and Engineering Research Council of Canada, MITACS Elevate postdoctoral fellowship and the industry partners Hypor and Genesus.

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Graham Plastow.

Additional information

Competing interests

The authors declare that they have no competing interests.

Authors’ contributions

Conceived and designed the experiments: GP; Performed the experiments: BH, CZ, YM; Analyzed the data: CZ, ZW, TY; Contributed reagents/materials/analysis tools: BK, PC; Wrote the paper: CZ, GP. All authors read and approved the final manuscript.

Additional file

Additional file 1: Table S1.

Functional annotation of the candidate and/or nearest genes to the significant SNPs in the region on SSC15. Table S2. Significant effects of SNP21 (MARC0083357) on other meat quality traits.

Rights and permissions

Open Access  This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made.

The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.

To view a copy of this licence, visit https://creativecommons.org/licenses/by/4.0/.

The Creative Commons Public Domain Dedication waiver (https://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Zhang, C., Wang, Z., Bruce, H. et al. Genome-wide association studies (GWAS) identify a QTL close to PRKAG3 affecting meat pH and colour in crossbred commercial pigs. BMC Genet 16, 33 (2015). https://doi.org/10.1186/s12863-015-0192-1

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s12863-015-0192-1

Keywords