A Polymorphism in the 5′-Untranslated Region of the Porcine Cholecystokinin Type A Receptor Gene Affects Feed Intake and Growth

Abstract

The location and utilization of quantitative trait loci (QTL) and candidate genes with significant effects on economically important traits are becoming increasingly important in livestock breeding programs. The porcine cholecystokinin type A receptor (CCKAR) is a candidate gene for performance traits, due to its known role in the physiological control of feed intake, satiety, and obesity. We investigated the association of CCKAR polymorphisms with feeding, growth, and efficiency traits in an F₂ population derived from a cross between Meishan and Large White founder animals and in lines of Large White pigs that had been divergently selected on the basis of lean growth efficiency traits. In the F₂ population, CCKAR genotype was significantly associated with daily feed intake and average daily gain. The effects of the polymorphisms were then assessed in a larger-scale analysis of segregating commercial lines. A newly discovered single-nucleotide polymorphism (SNP) within the 5′-untranslated region (5′-UTR) had highly significant effects on feed intake, average daily gain, and days to 110 kg, which were not seen for a previously reported SNP within the CCKAR gene. Furthermore, we provide evidence that the novel SNP disrupts the binding of the YY1 transcription factor, which raises the possibility that it is the causal variant. The 5′-UTR SNP could be utilized as a molecular genetic test for increased feed intake, faster lean growth, and reduced days to market weight in segregating commercial lines.

DEFINING the molecular genetic basis of economically important traits in agricultural species is an increasingly important goal. A large number of quantitative trait loci (QTL) and candidate gene analyses have revealed genomic regions and specific markers that are associated with traits such as growth, feed intake, reproductive efficiency, and meat quality in pigs (reviewed in Rothschild 2004). Favorable alleles at these QTL and candidate loci can be incorporated into animal breeding programs through marker-assisted selection to directly increase the quantity or quality of product or to reduce the cost at which the product can be produced. Thus, genetic markers with large and consistent effects on economically important traits can be of considerable commercial value to pig breeding companies. Although the utilization of such favorable loci has remained somewhat limited to date, this molecular genetic information is likely to be increasingly utilized in selection decisions over the next few years, in conjunction with traditional phenotype-based selection strategies (Van der Steen et al. 2005).

The candidate gene approach involves the analysis of a locus that may be expected to contain genetic variation underlying traits of interest. In this approach, the locus may be chosen on the basis of its map position in relation to a previously identified QTL or on the basis of knowledge of the function of the protein that it encodes, for example, if mutations within the locus have been demonstrated to have measurable phenotypic effects in other species. An example of this approach is the study by Kim et al. (2000) in which a single-nucleotide polymorphism (SNP) discovered within the porcine melanocortin-4 receptor (MC4R) gene was found to have significant effects on growth, fatness, and feed intake in an association analysis across a number of commercial lines. MC4R was chosen for study because it is known to be involved in obesity and the control of energy balance in humans and rodent models (Kim et al. 2000). The genetic test for the MC4R SNP is currently being used by the pig breeding industry to select pigs with higher genetic potential for feed intake, growth, and leanness (Plastow 2004).

The hormone cholecystokinin (CCK) is a gut-produced peptide that acts both centrally and peripherally to inhibit food intake and is therefore vital in the control of energy balance (Gibbs et al. 1973). Two CCK receptor subtypes exist and most experimental evidence suggests that the cholecystokinin type A receptor (CCKAR) is the most important in satiety regulation (e.g., Crawley et al. 1991; Asin and Bednarz 1992). Rats with a deletion in the CCKAR gene have a substantially higher food intake than wild-type control animals and, as a result, they become obese (Moran et al. 1998). Furthermore, immunization against CCK in piglets results in a substantial increase in feed intake and growth (Pekas and Trout 1990). Therefore, like MC4R, CCKAR is potentially an excellent candidate gene for performance traits in pigs. Previous work on the CCKAR locus in pigs has resulted in the discovery of a SNP that was used to map CCKAR to chromosome 8 (Clutter et al. 1998a), although to date no association with performance traits has been reported.

Most candidate gene analyses in livestock have focused on simple linkage disequilibrium-based associations between genotype and phenotype. Since lines of animals that are derived from divergent selection on the basis of a single trait will differ in phenotype due largely to alterations in allele frequency at loci that affect that trait (Falconer and MacKay 1996), comparing candidate gene allele frequencies in divergent selection lines may also demonstrate the association of genes with traits of interest. For example, allele frequencies at the aforementioned MC4R polymorphism have been shown to differ significantly in Large White lines divergently selected on the basis of lean growth and efficiency (Houston et al. 2004). The CCKAR locus may be particularly interesting to examine in divergent selection lines, since there is evidence that lines with divergent genetic potential for growth and feed intake may differ in their response to plasma CCK (Clutter et al. 1998b).

Relatively few studies into the molecular genetic basis of performance traits in livestock have identified the causal variants that underlie the association between a marker and a trait. For the purposes of marker-assisted selection, there is a limited need to identify causal variants since selection based on a marker in strong linkage disequilibrium will achieve a similar response (Dekkers and Hospital 2002). However, with successive generations of selection, linkage disequilibrium will be eroded such that the marker–trait association will become weaker. Also, the phase relationship between the genetic marker and the causal variant may differ depending on the specific line or breed being tested. Therefore, identifying functional variants is at least preferable for the application of molecular genetic technology to animal breeding programs. Furthermore, it is interesting from a biological perspective to understand how genetic variation can cause functional changes and, ultimately, phenotypic effects.

In the current study, the CCKAR gene was examined as a candidate gene for traits related to feed intake, feeding behavior, growth, and efficiency in an F₂ population derived from a cross between Meishan and Large White pigs. To further investigate the association of the gene with energy balance-related traits, the allele frequencies were then analyzed in Large White lines, which had been divergently selected for seven generations on the basis of traits related to lean growth efficiency. Finally, to confirm the effects, a larger-scale association analysis of the CCKAR gene with performance traits within commercial lines was performed. In addition to the previously reported SNP in CCKAR (Clutter et al. 1998a), a novel SNP within the 5′-UTR was used in these association studies and also tested for putative effects on the binding of a transcription factor that may ultimately lead to phenotypic effects.

MATERIALS AND METHODS

Meishan × Large White F₂:

The F₂ population was derived from a cross between Meishan and Large White pigs at the Roslin Institute, Midlothian, United Kingdom. The animals for which feed intake had been recorded came from two separate batches—one born in 1995 and the other born in 1996. For both of these populations, F₁ parents were produced from reciprocal matings of Meishan and Large White founder animals, with F₂ families being produced by the intercrossing of unrelated F₁ parents (Lee et al. 2003). A total of 152 F₂ animals (86 males and 66 females) from 54 full-sib families were included in the study. These animals were performance tested in two batches over a weight range of ∼30–85 kg. During performance testing, the animals were housed in pens with solid floors and bedded with straw. The animals were housed in pens of 13 and given ad libitum access to a commercial pig fattening growth ration. Electronic feed intake recording equipment (FIRE; Osborne Europe, Newcastle-Upon-Tyne, United Kingdom) recorded the animal identification number, weight, feed intake, and time spent in the feeder for every visit. This information was summarized into a daily summary file of median weight measurement, total feed intake, number of visits, and total time in the feeder. These data were used to calculate average daily gain (ADG, grams), daily feed intake (DFI, kilograms), feed conversion ratio (FCR), number of visits per day (NVD), average feed intake per visit (AFV, grams), total time spent in the feeder per day (TTF, mins), and a measure of feeding rate (AFR, grams per minute) over the period of growth from 35 to 80 kg.

Divergent selection lines:

Large White pig populations were divergently selected over seven generations into four groups of high and low lines from a single base population. In three of the four groups, the selection criteria were applied when animals had ad libitum access to feed: daily feed intake (DFI), lean feed conversion (LFC), and lean growth rate (LGA). The fourth criterion was lean growth on a restricted feeding scale (LGS). There were ∼30 animals in each of the complete high and low lines and 32 control animals, which were not subjected to any artificial selection pressure, giving a total of 270 animals included in the analysis. Details of the response to selection for ad libitum feeding are given in Cameron and Curran (1994) and for restricted feeding in Cameron et al. (1994), while the family structure and performance test data for the completed divergent lines are described in Houston et al. (2004).

Commercial lines:

The lines analyzed were sampled from PIC (PIC USA, Franklin, KY) breeding populations and were raised under normal production conditions under the care of PIC employees in nucleus farms in the United States and Europe. Pigs were put on the performance test at ∼70 days of age and taken off the test after 13 weeks. At the end of the trial subcutaneous backfat was measured ultrasonically in real time (B mode) at the 10th rib 2 cm from the center line. Average daily gain over the test period was calculated as weight gained divided by days on test. Days to 110 kg market weight were estimated using standard procedures and feed intake was measured using electronic feed intake recording equipment (FIRE).

Genotyping:

The Meishan × Large White pedigree and the divergent selection lines were genotyped for the polymorphism in the CCKAR locus described by Clutter et al. (1998a). This polymorphism is a guanine/cytosine polymorphism at position +471 of the CCKAR gene and hereafter is denoted +471C/G. PCR reactions were performed with 25 ng porcine DNA, 1 mm of each primer, and 5 μl RedTaq mix (Sigma) in a 10-μl final reaction volume. The reaction was carried out in a Geneamp PCR system 9700 thermal cycler (Applied Biosystems) and PCR conditions were 2 min at 94°; 35 cycles of 30 sec at 94°, 1 min at 55°, and 1 min 30 sec at 72°; and a 15-min extension at 72°. The PCR product was then incubated at 37° with the Fnu4HI restriction enzyme and its supplied buffer (New England Biolabs, Beverly, MA) for at least 2 hr. The products were loaded onto a 1.5% agarose gel containing ethidium bromide and run at 120 V for ∼1 hr 30 min. Allele A (+471C) is defined as fragments of ∼500, 300, and 100 bp and allele B (+471G) is defined as fragments of ∼300, 280, 220, and 100 bp.

The divergent selection lines and the grandparent founder animals in the F₂ population were then genotyped for a novel adenine/guanine SNP at position +179 within the 5′-UTR of the CCKAR gene (described below and hereafter referred to as +179A/G). The following modifications to the protocol described above were used. PCR primers F: CTTGGGAGACTCTGCAGTCC and R: GGGCTGATCCAAACAGAAAA were used to amplify the gene fragment. The restriction enzyme Hpy8I was used to reveal the polymorphism. Standard acrylamide gels (6% polyacrylamide) were run to separate and analyze the restricted product.The DNA fragments were 110 and 90 bp (allele A, +179A) or 90, 55, and 55 bp (allele B, +179G).

The commercial lines were genotyped by Kbiosciences (Hoddeston, UK) using their in-house genotyping system. Initially 10 lines were screened for segregation of both the +471C/G SNP and the +179A/G SNP. On the basis of their intermediate allele frequencies for the two SNPs, 2 lines were chosen for the association analysis, namely lines E and G: a synthetic dam line derived from Duroc and Large White-based lines and a Pietrain-based line, respectively. A third line that had intermediate allele frequency for +471C/G, but was fixed for +179G was chosen for analysis to assess the possibility of a direct effect of the +471C/G SNP. This line was named line F and was a Hampshire-based line. There were 547 animals from line E, 539 animals from line F, and 545 animals from line G genotyped for both SNPs.

Sequencing and binding-domain analysis:

The PCR product amplified by the primers described by Clutter et al. (1998a) was excised from the gel, purified using a gel extraction kit (Promega), and sequenced forward and reverse using a CEQ 8000 DNA sequencer (Beckman Coulter). A nucleotide–nucleotide BLAST analysis (http://www.ncbi.nlm.nih.gov/BLAST/) was performed and this revealed through cross-species comparison that the sequence was predominantly intron 1 of the CCKAR gene. To amplify promoter regions and exon 1 of the gene, primers were designed on the basis of cross-species homology (F: TCC AAA GCT AAA AAG GCA CTA AT and R: CCC CAC CAC CCC AAC TG), and the PCR product was sequenced as above. To identify polymorphisms in this promoter/exon 1 region, two animals of each +471C/G genotype from the divergent selection were sequenced and compared. Three SNPs were located, the A/G substitution at position 179 within the 5′-UTR, a T/C substitution at position 386 within the first intron, and the previously described C/G change at position 471 within the first intron. The Matinspector program (Geneomatrix software, Germany) was then used to assess whether the located SNPs resulted in the creation or destruction of a binding site for putative regulatory proteins.

Electrophoretic mobility shift assay:

To test the hypothesis that the +179A/G polymorphism resulted in an in vitro disruption of YY1 binding, an electrophoretic mobility shift assay (EMSA) was performed. Hindbrain tissue was harvested from a recently slaughtered Large White boar (Scotch Premier Meat, Inverurie, UK) and nuclear extracts were prepared according to the protocol described by Scarpace et al. (2001). Single-stranded DNA oligonucleotides (TAG Newcastle) were designed to incorporate the region of the porcine CCKAR sequence containing the putative YY1 binding site, with two oligonucleotides differing only by the 5′-UTR A/G single-base-pair change. These oligonucleotides had the sequence 5′-CCG AGC AGG AGC C(A/G)T TCA CCA GCT C-3′ (hereafter referred to as oligonucleotide A and oligonucleotide G). A consensus YY1-binding oligonucleotide was also synthesized, for use in competition assays, and had the sequence 5′-CGC TCC CCG GCC ATC TTG GCG GCT GGT-3′ (hereafter referred to as oligonucleotide Y). This sequence is known to bind with a high specificity to the YY1 protein across different species in EMSA experiments (Santa Cruz Biotechnology, product information). The exact reverse complement sequence to oligos A, G, and Y was also synthesized and each pair of forward and reverse oligos was annealed at a 2-μm concentration, except oligonucleotide Y, which was annealed at a 20 μm concentration. The forward and reverse oligonucleotides were added to an annealing buffer (200 mm Tris pH 8.0, 10 mm EDTA pH 8.0, 500 mm NaCl) and annealed by heating to 90° in a water bath and allowing them to cool to room temperature to form the double-stranded, unlabeled probe. To verify that oligonucleotides were annealed correctly, 2 μl of each double-stranded oligonucleotide (and 2 μl of single-stranded oligonucleotide as a control) were run according to a standard agarose gel protocol.

The Promega gel-shift assay system was used for the radiolabeling of oligonucleotides and binding reactions and the full protocol for this system is given on the Promega website (http://www.promega.com/tbs/tb110/tb110.pdf). Briefly, to radiolabel the oligonucleotides, 2 μl of double-stranded unlabeled oligo, 1 μl T4 polynucleotide kinase 10× buffer, 1 μl of [γ-³²P]ATP (3000 Ci/mmol at 10 mCi/ml; Amersham, Little Chalfont, UK), 5 μl of nuclease-free water, and 1 μl polynucleotide kinase (10 μl final volume) were added to a 200-μl PCR tube. This mixture was incubated at 37° for 10 min and the reaction was stopped by adding 1 μl of 0.5 m EDTA, before the addition of 89 μl of 0.5 m TE buffer.

For a standard incubation reaction, 1 μl of labeled oligonucleotide was incubated with 2 μl of brain nuclear extract, 2 μl of 5× gel shift binding buffer, and 5 μl of nuclease-free water. Negative control reactions did not contain nuclear extract. Competition assays contained the unlabeled oligonucleotide Y as a specific competitor at molar excesses ranging from 2× to 50×. The nonspecific competitor oligonucleotide was an AP-1 consensus oligonucleotide. The mixture was incubated for 10 min at room temperature, the labeled oligonucleotide was added, and then the mixture was incubated for a further 20 min at room temperature. One microliter of gel-loading 10× buffer was added to each reaction and the reaction mixture was then loaded onto a 20 × 20-cm 6% polyacrylamide gel and run for ∼1.5 hr at 240 V. The gel was placed on Whatman 3MM filter paper, covered with cling film, and exposed to X-ray film overnight at −70° before development.

Each binding assay (standard and competition) was repeated twice. The level of binding for each assay was quantified by scanning the gels using a high-resolution scanner and using the optical density (OD) readings associated with the bands on the gel, as obtained by the Scion (Frederick, MD) Image Version 4.0.2 program. The percentage of labeled oligonucleotide bound to the nuclear protein was calculated by measuring the OD reading for the peak corresponding to the protein–oligonucleotide complex and calculating this value as a percentage of the sum of the peaks associated with the protein–oligonucleotide complex plus the unbound labeled oligonucleotide.

Statistical analysis:

To assess the effect of CCKAR genotype on the FIRE feeder data in the F₂ population, the genotype and phenotype information was analyzed using the restricted maximum-likelihood (REML) method of Genstat 6.1 (Genstat Committee 2002). The model included sex, batch (year of birth), and candidate genotype as fixed effects with litter fitted as a random effect, and age at 35 kg fitted as a covariate. The Wald statistic associated with dropping the genotype term from the model was used to determine the significance of the genotype effect. The Genstat software was also used to obtain the predicted genotype means and their associated standard errors.

The differences in CCKAR allele frequency between the divergent selection lines were assessed by Fisher's exact test with 2 d.f. using Microsoft Excel, with the “analyze-it” add in (http://www.analyse-it.com). The data were analyzed to test the hypothesis that there had been a significant divergence in allele frequency between the high and low lines and between the selection and control lines. To give an indication of the contribution of genetic drift to the differences in allele frequency between the lines, the standard error of the difference between lines was estimated according to the method described in Ollivier et al. (1997).

The association between CCKAR genotype and the measured performance traits in the commercial lines was then assessed using REML. In the first analysis, the effects were assessed within line. All animals within each line were male and from the same test farm, so the model included the fixed effect of CCKAR genotype, with the random effect of dam and the covariate of “age at 35 kg” included for traits measured during the test period only (35–90 kg). In the second analysis, lines E and G were pooled to look at the overall effects of both SNPs across lines. The model was as above, except that line was included as a fixed effect.

RESULTS

The initial test of CCKAR as a candidate gene was performed on the +471C/G SNP in the Meishan × Large White F₂ population. The +471G homozygotes were found to have significantly greater ADG and DFI than +471C homozygotes during the growth period from 35 to 80 kg (Table 1). Sequencing of the PCR product (GenBank accession no. DQ496228) revealed that the +471C/G SNP was intronic and was not predicted to alter any known binding domains. The sequencing of the surrounding regions was performed to search for nearby linked polymorphisms that may be postulated to have functional effects (GenBank accession no. 796827, Figure 1). The +179A/G SNP was discovered in the 5′-UTR of the CCKAR gene and was predicted to alter the core binding sequence of the transcription factor YY1, which binds to promoter and 5′-UTR regions of mammalian genes to regulate gene expression and, therefore, can have downstream effects on phenotype (Romey et al. 2000; Cook et al. 2001; Costa et al. 2001; Hines et al. 2003; Kalayci et al. 2003). In all further association analyses, both the +179A/G SNP and the +471C/G SNP were analyzed.

TABLE 1.

Predicted means (standard errors) of feeding and growth traits in the Meishan × Large White F₂ population

	CCKAR genotype
	+471C/+471C (n = 22)	+471C/+471G (n = 62)	+471G/+471G (n = 49)
Average daily gain (g)	550 (28.3)	603 (22.9)	615 (20.8)	P < 0.04
Daily feed intake (kg)	1.92 (0.06)	2.01 (0.04)	2.11 (0.04)	P < 0.03
Feed conversion ratio	3.53 (0.11)	3.37 (0.10)	3.47 (0.09)	P < 0.22
No. of visits per day	6.11 (0.44)	6.94 (0.28)	6.94 (0.30)	P < 0.22
Total time in feeder (min)	61.3 (2.5)	56.0 (1.6)	57.9 (1.7)	P < 0.20
Average feed per visit (g)	343 (21)	310 (13)	336 (14)	P < 0.24
Feeding rate (g/min)	33.5 (1.6)	36.0 (1.0)	37.3 (1.1)	P < 0.15

Figure 1.—

Schematic diagram of the putative structure of the porcine CCKAR gene (based on cross-species comparisons, adapted from Funakoshi et al. 2000) and a section of the porcine sequence data. The sequence data between the dashed lines show the exact locations of the +471C/G SNP and the +179A/G SNP. The A/G change is within the putative 5′-UTR, while the C/G change is between exons 1 and 2. Within the sequence data, italics show (putative) transcribed regions, underlines show (putative) translated regions, and (A/G) shows a polymorphic site.

The Meishan founder animals were fixed for +179G and +471G while the Large White founders were segregating for both SNPs. Despite the small distance between the two SNPs within the CCKAR gene, there was a recombinant haplotype in six Large White founder animals, whereby cytosine at +471 was inherited with guanine at +179 (+179G/+471C). The frequency of the +179G/+471C haplotype was 0.38, compared to 0.62 for the +179A/+471C haplotype. When the offspring from the Large White founders with the recombinant haplotype were removed from the analysis, the size of the effect remained large, although the effects were no longer statistically significant with the smaller numbers (data not shown).

Assessing CCKAR allele frequencies associated with divergent selection showed that in the low lean growth lines (LGA and LGS), the +179A and +471C alleles were significantly higher in frequency than in their complementary selection lines and the control line, thus indicating that this allele is associated with an effect on reduced lean growth rate. However, selection for low lean feed conversion was associated with fixation for the +179G and +471G alleles (Table 2), suggesting that the guanine allele is associated with lower conversion of feed to lean mass. In the daily feed intake lines, the frequency of +179A and +471C was higher in the low line than in the high line or the control line, but the difference was not significant (P > 0.05). The +471C/G SNP and the +179A/G SNP are in strong linkage disequilibrium in the divergent selection lines. However, the +179G/+471A haplotype also appears in these lines, and the allele frequencies for both SNPs are shown in Table 2. In the LGA lines, the +179G allele is significantly higher in frequency in the high line than the low line or control, suggesting an association of this allele with high lean growth, which was not seen for the +471G allele. To give an indication of the contribution of genetic drift to the differences in the allele frequencies between the selection lines, the estimated standard errors adjusted for drift were 0.29 for the +179A/G SNP and 0.30 for the +471 SNP.

TABLE 2.

Allele frequencies for the CCKAR locus in the divergent selection lines

Selection criteria	Direction	Frequency of +179A allele	Frequency of +471C allele
Lean feed conversion (LFC)	High	0.42a	0.42a
	Low	0.00ab	0.00ab
Lean growth ad libitum (LGA)	High	0.17ab	0.43a
	Low	0.61a	0.70a
Lean growth restricted feeding (LGS)	High	0.40a	0.40a
	Low	0.93ab	0.93ab
Daily feed intake (DFI)	High	0.43	0.43
	Low	0.60	0.60
Control	—	0.43	0.45

^aSignificantly different from complementary selection line (P < 0.01).

^bSignificantly different from control line (P < 0.01).

The data from the experimental populations indicated that genetic variation at the CCKAR locus may be having an effect on performance traits and, therefore, the CCKAR polymorphisms were investigated for effects within commercial lines. The overall results show that the +179A allele is associated with lower feed intake (P < 0.001), average daily gain (P < 0.004), and a higher number of days to 180 kg (P < 0.001) than heterozygotes or +179G homozygotes. There was substantially less evidence for an effect of the +471C/G SNP, as this SNP showed a significant association only with days to 180 kg (P < 0.03) in the overall analysis. Within the individual lines, the effects were most significant in line G, with line E showing a nonsignificant trend in the same direction. However, the interaction term between line and genotype in the model was not significant, indicating that there is no significant evidence for a difference in the effects of the +179A/G SNP between the two lines. In line F, which was fixed for +179G but segregating for the +471C/G SNP, there was no evidence for an effect on any of the measured feed intake or growth traits (Table 3). There was evidence for an effect on backfat in this line (P < 0.05), but when analyzed across all lines this was not significant (data not shown).

TABLE 3.

Summary of the association between the CCKAR SNPs and the performance traits in lines E (n = 547), G (n = 545), and F (n = 539)

Genotypea	Overall: +179A/G	Overall: +471C/G	Line E: +179A/G	Line E: +471C/G	Line G: +179A/G	Line G: +471C/G	Line F: +471C/G
Daily feed intake (kg)
AA	2.04 (0.02)	2.10 (0.02)	2.28 (0.03)	2.33 (0.02)	1.81 (0.03)	1.85 (0.03)	2.30 (0.03)
AB	2.15 (0.01)	2.15 (0.01)	2.36 (0.02)	2.37 (0.02)	1.94 (0.02)	1.93 (0.02)	2.30 (0.02)
BB	2.15 (0.02)	2.12 (0.02)	2.37 (0.02)	2.34 (0.03)	1.92 (0.02)	1.90 (0.03)	2.33 (0.03)
	P < 0.001	P < 0.06	P < 0.06	P < 0.48	P < 0.001	P < 0.04	P < 0.71
Days to 110 kg
AA	168.0 (1.0)	165.9 (0.8)	157.3 (1.3)	155.6 (0.9)	178.3 (1.5)	177.0 (1.4)	153.7 (0.9)
AB	163.9 (0.6)	163.9 (0.6)	154.9 (0.7)	154.7 (0.8)	172.9 (1.0)	173.1 (0.9)	153.0 (0.7)
BB	163.5 (0.7)	163.1 (0.8)	154.9 (0.7)	155.1 (1.2)	172.0 (1.2)	171.3 (1.2)	153.5 (0.9)
	P < 0.001	P < 0.03	P < 0.25	P < 0.71	P < 0.003	P < 0.008	P < 0.78
Lifetime daily gain (g)
AA	658.0 (5.2)	667.1 (4.1)	720.4 (8.1)	726.5 (5.3)	599.5 (6.6)	604.1 (6.2)	735.7 (6.3)
AB	677.2 (3.1)	677.1 (3.1)	731.3 (4.4)	731.7 (4.8)	623.7 (4.2)	622.4 (4.0)	740.3 (4.4)
BB	677.7 (3.8)	679.6 (4.5)	727.6 (5.2)	730.0 (7.5)	626.9 (5.2)	628.4 (5.4)	736.1 (6.0)
	P < 0.004	P < 0.08	P < 0.52	P < 0.76	P < 0.003	P < 0.01	P < 0.78
Test daily gain (g)
AA	844.8 (8.0)	855.5 (6.2)	933.0 (12.5)	944.5 (8.1)	757.5 (10.0)	758.7 (9.5)	944.2 (9.7)
AB	870.5 (4.7)	870.4 (3.0)	949.4 (6.8)	950.1 (7.2)	791.6 (6.3)	791.5 (6.2)	1001 (6.7)
BB	869.6 (5.8)	870.7 (6.9)	945.0 (8.1)	943.8 (11.4)	796.0 (8.0)	795.7 (8.3)	989.6 (9.2)
	P < 0.02	P < 0.13	P < 0.52	P < 0.84	P < 0.006	P < 0.006	P < 0.61
Feed conversion ratio
AA	2.43 (0.03)	2.46 (0.02)	2.45 (0.03)	2.48 (0.02)	2.41 (0.04)	2.46 (0.04)	2.32 (0.03)
AB	2.48 (0.01)	2.48 (0.02)	2.50 (0.02)	2.51 (0.02)	2.46 (0.02)	2.44 (0.02)	2.31 (0.02)
BB	2.49 (0.02)	2.46 (0.02)	2.52 (0.02)	2.49 (0.03)	2.45 (0.03)	2.42 (0.03)	2.36 (0.02)
	P < 0.14	P < 0.70	P < 0.23	P < 0.63	P < 0.56	P < 0.71	P < 0.19
Fat depth (mm)
AA	10.2 (0.2)	10.5 (0.2)	10.7 (0.3)	11.0 (0.2)	9.8 (0.3)	9.8 (0.3)	11.3 (0.3)
AB	10.5 (0.1)	10.5 (0.1)	10.9 (0.2)	11.0 (0.2)	10.0 (0.2)	10.0 (0.2)	11.6 (0.2)
BB	10.5 (0.2)	10.4 (0.2)	10.8 (0.2)	10.4 (0.3)	10.3 (0.2)	10.3 (0.2)	10.8 (0.2)
	P < 0.58	P < 0.94	P < 0.78	P < 0.20	P < 0.38	P < 0.41	P < 0.04
Muscle depth (mm)
AA	62.7 (0.5)	63.1 (0.4)	59.8 (0.8)	60.2 (0.5)	65.9 (0.7)	65.7 (0.7)	65.6 (0.7)
AB	63.9 (0.3)	63.8 (0.3)	60.4 (0.4)	60.3 (0.5)	67.3 (0.5)	67.4 (0.5)	65.7 (0.5)
BB	63.1 (0.4)	62.7 (0.5)	59.4 (0.5)	58.5 (0.7)	66.7 (0.6)	66.6 (0.6)	65.2 (0.7)
	P < 0.09	P < 0.09	P < 0.33	P < 0.10	P < 0.24	P < 0.10	P < 0.86

“Overall” refers to effects across lines E and G.

^aAllele A refers to +179A and +471C, and allele B refers to +179G and +471G.

The results from the EMSA show that oligonucleotides A and G (corresponding to +179A and +179G) are clearly binding with a nuclear factor (Figure 2a). The darker band associated with the oligonucleotide A–nuclear factor complex suggests that this oligonucleotide has a stronger binding affinity for the protein, as would be expected if the protein is YY1 since oligonucleotide A represents the intact YY1 core binding site. By calculating the average OD reading for the bands associated with A and G over two runs, the binding of A to the nuclear protein was estimated to be 44% higher than that of G (Figure 2b). Also, the inclusion of oligonucleotide Y (the specific competitor oligonucleotide) at 50-fold molar excess reduced the intensity of the oligonucleotide G–protein complex by ∼90% (as measured by comparisons of the OD readings with and without oligonucleotide Y—repeated twice) and reduced the intensity of the band associated with the oligonucleotide A–protein complex by an estimated 60%, whereas a nonspecific competitor oligonucleotide did not (Figure 2b). Furthermore, the addition of a nonspecific competitor oligonucleotide had little impact on the binding of oligonucleotide A and oligonucleotide G to the nuclear protein (Figure 2a, right side), which implies that the nuclear protein is YY1 as hypothesized.

Figure 2.—

(a) Left, a comparison of the effects of increasing YY1 competitor oligonucleotide (oligonucleotide Y) on the nuclear protein–DNA complex with oligonucleotide A (1–7) and oligonucleotide G (8–14). Lanes 1 and 8 are negative controls without nuclear protein extract. Lanes 2 and 9 do not contain competitor oligo. The competitor oligonucleotide is added at molar excesses of 2× (lanes 3 and 10), 5× (lanes 4 and 11), 10× (lanes 5 and 12), 20× (lanes 6 and 13), and 50× (lanes 7 and 14). Right, a comparison of the addition of a 50-fold excess of the specific competitor oligonucleotide (oligonucleotide Y, lanes 1 and 2) and a nonspecific (AP1 consensus) oligonucleotide (lanes 3 and 4) for oligonucleotide A (lanes 1 and 3) and oligonucleotide G (lanes 2 and 4). (b) Histogram representing the optical density (OD) readings associated with the protein–oligonucleotide complex, expressed as a percentage of the sum of the protein–oligonucleotide complex and the unbound oligonucleotide, for oligonucleotide A and oligonucleotide G, with increasing concentration of competitor oligonucleotide Y (each data point is the average reading from two gels).

DISCUSSION

We describe the investigation of the porcine CCKAR gene for effects on performance traits in an experimental Meishan × Large White F₂ population and divergently selected Large White populations, followed by a larger-scale analysis of the effects in segregating commercial lines. In the F₂ population, the +471C/G SNP is significantly associated with ADG and DFI, with the B allele (the predominantly Meishan allele) corresponding to faster-growing animals with a higher daily feed consumption. However, since the level of linkage disequilibrium in this F₂ population will inevitably be substantial, it was entirely possible that a single marker–trait association could be due to a causal polymorphism elsewhere within the CCKAR gene or potentially some distance removed from the gene, perhaps up to tens of centimorgans.

Within the divergent selection lines, the differences in CCKAR allele frequency between the high, low, and control lines were clear for the lean growth traits. However, the effects of the individual alleles were not clear cut. For example, the A allele (for the +179A/G and the +471C/G SNPs) had been completely eliminated from the LFC low line with seven generations of selection, yet this allele was substantially higher in the LGA and LGS low lines than in their complementary selection lines. The selection objective of the LFC lines was to achieve equal correlated responses in carcass lean content and FCR, whereas the lean growth lines focused on carcass lean content and growth (Cameron 1994). Therefore, the CCKAR allele A may have a negative association with growth, which is supported by the association analyses in the other populations, but the B allele may negatively affect the ratio of feed conversion to lean mass in these lines. However, it is important to consider the possible contribution of genetic drift to the allele frequencies in the selection lines, given the large estimated standard errors associated with the differences between the lines after adjustment for drift.

There was some degree of recombination between the two SNPs in the Large White lines, and the data from the selection lines suggested that the +179G allele may increase lean growth under ad libitum feeding (frequency of 0.83 in high line), which was not seen with the +471G allele (frequency of 0.57 in the high line). A recombinant haplotype was observed in some of the Large White founders of the LW × MS F₂ population animals, but there were insufficient data to compare the effects of the two SNPs in this population. However, since it was clear that there were some performance trait effects associated with the CCKAR gene, a larger-scale study of commercial populations was undertaken. From these analyses, it became clear that either the +179A/G SNP was in closer linkage disequilibrium with the causal variant than the +471C/G SNP or indeed +179A/G may in fact be the causal variant given the putative effects on transcription factor binding discussed in more detail below.

The effects associated with the +179A/G SNP on the economically important traits of feed intake, daily gain, and days to 180 kg were highly significant in the Pietrain-based and Large White/Duroc-based commercial lines. Evidence for a putative direct effect of the +179A/G SNP, or a tightly linked variant, came from the analysis of the Hampshire-based line, which was fixed for +179G but segregating for +471C/G. Here there was no evidence for any effect on these performance traits. While this result alone could potentially be explained by the different genetic background of this line to the other lines, taken in conjunction with the lack of significance of the +471C/G SNP in the combined analysis of lines E and G, these data suggest that +179A/G is either the causal SNP or the better marker for the causal SNP. The +179G allele appears to be dominant, such that animals with at least one copy of this allele have on average ∼5% higher daily feed intake, 3% higher daily gain, and 3% less time to reach 180 kg, when compared to homozygotes for +179A.

The +179A/G SNP was found to alter a core binding site for the transcription factor YY1, which provides some evidence that it may be associated with functional effects. The binding of the nuclear factor to the oligonucleotide corresponding to +179G was clearly reduced compared to +179A. One hypothesis is that this may correspond to an in vivo effect of +179G leading to a reduced activity of the CCKAR gene, and thus reduced satiety effect and increased appetite and growth, which would link in with the findings that a lack of response to CCK in pigs is associated with higher feed intake and growth (Pekas and Trout 1990). Clearly, further experiments would be required to test this hypothesis, including a comparison of mRNA levels between the alternative genotypes, and perhaps a reporter assay comparing luciferase activity of promoters containing the +179 A/G variant. Nonetheless, it is interesting to note that the large effects of a regulatory mutation in the insulin-like growth factor 2 gene (IGF2) were probably mediated through the disruption of binding of a transcription factor (Van Laere et al. 2003).

Since the effect of the +179A/G SNP on the phenotypic traits is essentially dominant, the offspring of a +179G homozygous sire (or dam) would potentially see an improved growth performance irrespective of the other parental genotype. The +179G allele could also be expected to result in an increased appetite and daily feed intake. However, since there is no evidence for a negative effect on feed conversion in the commercial lines, this suggests that the extra ingested feed is converted to growth without a loss of efficiency. Furthermore, in the lines studied, there is no evidence for a negative influence of the increasing allele on backfat, which may have been expected given the feed intake and growth effect. Therefore, the novel +179A/G polymorphism in the 5′-UTR of the porcine CCKAR gene may have considerable value for segregating lines within a commercial breeding program as a simple genetic test for improved growth performance and reduced cost of production.