Alleles of cytosolic phosphoenolpyruvate carboxykinase (PEPCK): trait association and interaction with mitochondrial PEPCK in a strain of White Leghorn chickens

Alleles of Cytosolic Phosphoenolpyruvate Carboxykinase (PEPCK): Trait Association and Interaction with Mitochondrial PEPCK in a Strain of White Leghorn Chickens1 R. Parsanejad, A. Torkamanzehi, D. Zadworny, and U. Kuhnlein2 Department of Animal Science, McGill University, Ste. Anne de Bellevue, QC, Canada H9X 3V9 drial PEPCK (PEPCK-M) genotypes defined by a single RFLP. The latter enzyme catalyzes the same reaction but is located in the matrix of the mitochondria and is encoded by a different nuclear gene. Interaction was evident from an analysis of the egg weight and egg specific gravity in the early phase of egg laying. It was such that the effect of the variation in one gene depended entirely on the genotype of the second gene. In addition, significant genotypic disequilibria were observed between two of the three alleles of PEPCK-C and between one of these alleles and the two RFLP alleles of PEPCK-M. This finding indicates variations of genes in the gluconeogenesis pathway may affect feed utilization and egg production traits, as well as reproductive fitness.

(Key words: association analysis, feed efficiency, functional genomics, gene interaction phosphoenolpyruvate carboxykinase, residual feed consumption) 2003 Poultry Science 82:1708–1715

INTRODUCTION Phenotypes are determined by networks of interacting metabolites, proteins, RNA, and DNA. Such interactions occur between and within all levels. Enzymes regulate metabolic processes and are themselves subject to regulation by metabolites. Enzymes and proteins control their own expression by regulating processes such as transcription, translation, RNA splicing, and RNA editing. At a higher level, phenotypes determine the reproductive success of an individual in a breeding population. Hence they determine the repertoire of genetic variations present in a breeding population. The stability of such networks is provided by a myriad of regulatory loops and feedback controls. Due to the complexity of the relation between genotype and phenotype, the search for quantitative trait mutations

2003 Poultry Science Association, Inc. Received for publication December 30, 2002. Accepted for publication June 9, 2003. 1 Supported by grants from the Natural Sciences and Engineering Research Council of Canada, Agriculture and Agri-Food Canada, The Poultry Industry Council and Shaver Poultry Breeding Farms Ltd. 2 To whom correspondence should be addressed: urs.kuhnlein@ mcgill.ca.

has been shifted from segregation analyses to association studies. Association studies entail the matching of genetic patterns with phenotypic patterns. Such patterns, however, are difficult to characterize. At the phenotypic level, they may be classified by techniques such as cluster analysis or principal component analysis (Manly, 1994; Nagajara et al., 2000; Dopazo et al., 2001). At the genetic level, nucleotide variations may be grouped into haplotypes (i.e., arrays of alleles that are co-inherited). In addition, one may pinpoint sites of recombination and further refine the genotypic profiles. The major problem is to create a classification of phenotypes and genotypes that are meaningful, yet broad enough to contain a sufficient number of observations for statistical analysis. We have previously analyzed White Leghorn chickens for DNA polymorphism in the gene coding for the cytosolic form of phosphoenolpyruvate carboxykinase (PEPCK) (Parsanejad et al., 2002). PEPCK is a key regulatory enzyme of gluconeogenesis, catalyzing the formation

Abbreviation Key: AFE = age at first egg; EGM = egg mass; FE = feed efficiency; PEPCK= phosphoenolpyruvate carboxykinase; PEPCKC = cytosolic form of PEPCK; PEPCK-M = mitochondrial form of PEPCK; RFC = residual feed consumption; SNP = single nucleotide polymorphism; SPG = egg specific gravity.

1708

Downloaded from http://ps.oxfordjournals.org/ at NERL on May 26, 2015

ABSTRACT White Leghorn chickens from a nonselected closed population were typed for two RFLP located in the 3′ end of the gene coding for cytosolic phosphoenolpyruvate carboxykinase (PEPCK-C), a major control gene of gluconeogenesis. The two RFLP gave rise to three alleles (or haplotype classes), thus defining six genotypes. Feed efficiency (FE) and residual feed consumption (RFC) varied significantly among the genotypes and indicated that all three haplotypes differed from each other. FE is the ratio between feed consumption and egg mass produced, whereas RFC is the feed consumption after correcting for BW and egg production. There was significant interaction between PEPCK-C genotypes and mitochon-

1709

TRAIT ASSOCIATION OF PEPCK IN WHITE LEGHORNS 1

TABLE 1. Definition and frequency of PEPCK-C haplotypes

particularly interested in whether this analysis coupled with the knowledge of the gene tree would enable us to identify the branches of the gene tree that harbored the mutation responsible for trait differences. In addition, we examined whether there was genotypic interaction with a marker in the gene coding for the isozyme PEPCK-M.

Haplotype frequencies RFLP alleles Haplotype A B C

AciI

BstEII

Female (n = 350)

Males (n = 119)

+ + −

+ − −

0.29 0.16 0.55

0.29 0.21 0.50

MATERIALS AND METHODS

1

Cytosolic form of phosphoenolypyruvate carboxykinase.

Experimental Chickens and Traits Chickens were from a White Leghorn strain (strain 7) that had been established from North American commercial stocks in 1958. The strain had been propagated by pedigreed random mating using 100 sires mated to two females each. The estimated effective population size was 457 (Gowe et al., 1993). Genotypic data were collected from 350 females, and phenotypic data were from a subset of 340 females. Each sire family was represented in the phenotypic data set by one to six half sibs. The BW was measured prior to onset of egg laying (140 d of age, housing BW) and at 265 d of age (mature BW). The age at first egg (AFE) was recorded. Egg weight and specific gravity (SPG) were measured on eggs laid during 5 consecutive d in three periods starting from 240 d (period 1), 350 d (period 2), and 450 d (period 3) and were averaged for each period. SPG is an indirect measure of eggshell calcium, which amounts to a daily output of about 10% of the total body calcium (Soares, 1984). The rate of egg

TABLE 2. Comparison of the trait medians of the PEPCK-C1 genotype classes2 PEPCK-C genotype3 Trait BW (g) 130 d 265 d Age at first egg (d) Egg weight (g) Period 1 Period 2 Period 3 Egg specific gravity Period 1 Period 2 Period 3 Rate of egg laying (%) Period 1 Period 2 Period 3 Feed consumption (g) Feed efficiency4 RFC5 (g) 1

AA (21–19)

AB (29–23)

AC (125–114)

BB (10–7)

BC (62–48)

CC (88–74)

Significance (P-value)

1,310 1,781 169.0

1,265 1,772 168.0

1,270 1,745 167.0

1,220 1,700 168.0

1,310 1,764 163.5

1,305 1,777 165.5

0.20 0.15 0.54

52.6 57.2 61.5

51.4 56.6 61.0

52.3 57.7 60.0

53.0 59.4 62.5

52.7 58.2 59.8

52.4 58.2 61.2

0.70 0.59 0.50

86.0 82.0 80.0

86.0 82.0 78.0

84.7 82.0 78.5

86.8 80.7 76.0

85.0 82.0 79.3

85.2 82.0 80.0

0.27 0.73 0.61

86.7 71.3 65.0 117 2.48 4.28

84.1 72.5 60.0 117 2.56 3.49

84.2 72.5 60.0 113 2.44 −0.38

89.1 71.2 50.6 111 2.27 −5.75

84.2 70.0 55.0 111 2.50 −0.61

84.6 71.3 57.5 116 2.42 −1.36

0.28 0.84 0.37 0.25 0.026 0.028

Cytosolic form of phosphoenolypyruvate carboxykinase. Significance was tested using the Kruskal-Wallis analysis of variance by rank, because some traits were not normally distributed. Parametric analysis of the traits whose values were close to normally distributed gave the same results. 3 The number of observations in each class is indicated below the genotype. Variations within a genotypic class are due to incomplete observations. 4 Significant contrasts were AB > AC, BB, CC; AC > BB and BB < BC, CC (Kruskal-Wallis multiple comparison test). P-values < 0.5 are indicated in bold type. 5 Deviation of the expected values from the values predicted from the regression of feed consumption on mature BW and egg mass. Significant contrasts were AA > BB; AB > AC, BB, BC, CC, and AC > BB. 2

Downloaded from http://ps.oxfordjournals.org/ at NERL on May 26, 2015

of phosphoenolpyruvate by decarboxylation of oxaloactetate while hydrolyzing guanosine triphosphate (GTP) (Hanson and Reshef, 1997). There are two isozymes encoded by two different nuclear genes (Savon et al., 1993). One isozyme cystolic PEPCK (PEPCK-C) is located in the cytosol, and the other (PEPCK-M) is in the mitochondrial matrix. Sequence analysis of a 2,000-bp fragment of PEPCK-C in White Leghorns revealed the presence of 19 single nucleotide polymorphisms (SNP) that could be grouped into six different haplotypes (combinations of SNP alleles). Two of the SNP coincided with RFLP, which enabled us to further subdivide the six haplotype classes into three classes, representing different branches of the gene tree (Parsanejad et al., 2002). Here we investigated the association of the nine PEPCK-C genotypes defined by two RFLP with production traits in a strain of White Leghorn chickens. We were

1710

PARSANEJAD ET AL.

laying was measured between AFE and d 275 (period 1), d 276 to 356 (period 2), and d 357 to 457 (period 3). Feed consumption and egg mass (EGM) were measured between 247 and 268 d of age. Feed efficiency (FE) is the feed consumption divided by the EGM. The residual feed consumption (RFC) values are residuals of the regression of feed consumption on the mature BW and EGM (Luiting and Urff, 1991). No phenotypic data were collected from male offspring, but blood samples of 119 individuals were collected and genotyped.

RFLP Assays and Statistical Analyses For the PEPCK-C AciI RFLP the primer sequences were GTCTCTCCCAACGAACCCA ACATG (forward) and CCTCTTCTGACATCCAGCGACC (reverse); for the PEPCK-C BstEII RFLP they were GCTGGGACTGAATGGAAGAGGAG (forward) and CTGT TGAGTCGGATGGGTGTCAG (reverse); and for the PEPCK-M AccI RFLP they were CCTTCGCCATGAGCCCCTTTTTC (forward) and CAGCTCCGCCATGACATCCCT (reverse). The PCR conditions were 5 min presoaking at 95°C , followed by 35 cycles of 60 s at 94°C, 80 s at 62°C (PEPCK-C) or 60°C

(PEPCK-M), and 90 s at 72°C. The RFLP locations for PEPCK-C (sequence numbering of Sato et al., 1997) were position 1,808 for AciI and −664 for BstEII. The PEPCKM AccI RFLP was located at position 1,578 (sequence numbering of Weldon et al., 1990). For statistical analyses and graphics, we used the NSSC program (Hintze, 1997). Unless otherwise stated we used rank based statistics to avoid the necessity to specify biologically unrealistic models and to avoid problems with distributions, that deviated from normality. However, parametric tests (GLM procedures, ANOVA, or t-tests) gave the same result, even when traits deviated from normality or subgroups had significantly different variances. Four outliers with AFE > 200 d and one with henday rate of egg production of 1 < 50% were removed from the data set. Composite genotypic disequilibria between the three alleles of PEPCK-C and the two alleles of PEPCK-M were analyzed using the program of Lewis and Zaykin (2001). In this program, the frequency of each allele is estimated and used to calculate the expected frequencies of the allelic combinations. The disequilibrium coefficients are the deviations of the expected from the observed frequen-

Downloaded from http://ps.oxfordjournals.org/ at NERL on May 26, 2015

FIGURE 1. Quantile plots of the distributions of residual feed consumption, BW at 260 d of age, egg mass produced, and feed consumed. The distribution of the genotype AA is not shown for clarity. It followed the distribution of AC, BC, and CC below the median and the distribution of AB above the median. PEPCK = cytosolic form of phosphoenolpyruvate carboxykinase.

1711

TRAIT ASSOCIATION OF PEPCK IN WHITE LEGHORNS

we will be interested in the trigenic disequilibrium DABB. It can be shown that DABB = -1/2 DABb = DAbb = −DaBB = 1/2 DaBb = −Dabb. Hence, if the allele combination ABB is in excess, the combinations Abb and aBb are also in excess, whereas the other three combinations are at a deficit. FIGURE 2. Partial regression lines describing the relationship between predicted feed consumption and egg mass and BW for the genotypes AB, BB, AC, BC, and CC. The predicted values obtained by regressing feed consumption on BW and egg mass in each genotypic class were plotted against BW. Only the regression lines for the genotypes AB and BB are marked.

cies. We used the program that screens pairs of alleles, one at each locus. The disequilibrium is called genotypic, because it does not assume Hardy-Weinberg equilibrium and composite, as the two types of double heterozygote are grouped together (Weir, 1990). The program yields six disequilibrium coefficients. For two loci with alleles A, a and B, b, respectively, the two coefficients DA and DB are the deviations from Hardy-Weinberg equilibrium of the two loci. DAB describes the deviation of the digenic combinations of the A and B alleles at the two loci; DABB and DAAB the deviations of the two possible combinations of three alleles; and DAABB the deviation of the combination of all four alleles. Each disequilibrium coefficient is corrected by the preceding lower level disequilibria. The program lists the disequilibrium coefficients for the combinations of one of the two alleles at the two loci. The disequilibrium coefficients for other combinations of the four alleles A, a, B, and b can be deduced from the coefficients of the allelic combinations of A and B. In our case

RESULTS AND DISCUSSION Frequencies and Trait Association of PEPCK-C Alleles A total of 350 females and 119 males from a White Leghorn strain (strain 7) were typed for the RFLP at the AciI site and the RFLP at the BstEII site (Parsanejad et al., 2002). The absence or presence of restriction sites at the two loci defined three haplotype classes: haplotype A TABLE 3. Genotypes frequencies of PEPCK-C1 and deviation from Hardy-Weinberg equilibrium

Genotype A/A B/B C/C A/B A/C B/C 1

Observed frequency

Deviation from expected frequencies

Significance (chi-square, df = 1)2

0.060 0.032 0.266 0.086 0.375 0.181

−0.024 +0.004 −0.030 −0.010 +0.059 +0.001

9.13** 0.67 4.74* 0.57 7.74** 0.01

Cytosolic form of phosphoenolypyruvate carboxykinase. Determined as described by Hernandez and Weir (1991). **P < 0.01; *P < 0.05. 2

Downloaded from http://ps.oxfordjournals.org/ at NERL on May 26, 2015

FIGURE 3. Normality plot of the juvenile BW for the genotypic classes that deviate from Hardy-Weinberg disequilibrium. The intercept with the ordinate at 0 equals the median and the slope of the curve the standard deviation. The frequencies of AA and CC are lower and the frequency of AC is higher than expected from Hardy-Weinberg equilibrium. The variances between AC vs. AA and CC combined differed at a P-level of 0.026 (variance-ratio equal-variance test) or 0.054 (modifiedLevene equal-variance test).

1712

PARSANEJAD ET AL.

[AciI+ BstEII+], haplotype B [AciI+ BstEII−], and haplotype C [AciI− BstEII−] (Table 1). The fourth combination [AciI− BstEII+] was absent, as expected, due to parsimony during evolution and the lack of recent recombination or gene conversion events. Based on previous analysis of the mutation spectrum (Parsanejad et al., 2002), haplotypes A and C represent two different branches of the gene tree and are genetically most distant. Haplotype B comprised haplotypes closer to the putative ancestral PEPCK-C gene [A ← B → C ]. For convenience we subsequently refer to the three haplotype classes as alleles A, B, and C. The frequency of the alleles as well as of the genotypes did not differ significantly between males and females. Six different genotypes arising from the three alleles were analyzed for association with production traits (Ta-

ble 2). Nonparametric tests between the genotypic classes revealed significant differences in the median values for FE and RFC, but not for any of the other traits. Pairwise contrasts indicated that none of the three alleles of the PEPCK-C genotypes were equivalent. Specifically, the median for RFC was lower for B homozygotes than A and C homozygotes, indicating that the more ancient allele B was different from A and C. Further, alleles A and C were not equivalent, as the median values for the heterozygotes AB and BC differed significantly (Table 2). The median is a measure of the central tendency of the distribution. However, other aspects of the distributions of RFC values were also affected (Figure 1). Quantile plots indicate differences between distributions for genotype BB and genotypes AB and BC. It again reflected that

Downloaded from http://ps.oxfordjournals.org/ at NERL on May 26, 2015

FIGURE 4. Interactive effects between phosphoenolpyruvate carboxykinase (PEPCK) cytosolic form (-C) and mitochondrial form (-M) on egg weight and egg specific gravity. Upper panels show egg weight distributions (240 d) for different PEPCK-M genotypes in combination with the PEPCK-C genotypes B−/− (panel A) and B+/− (panel B), respectively. The symbols 䊊, 䊐, and 䉭 designate PEPCK-M genotypes AccI−/−, AccI+/ −, and AccI+/+, respectively. Lower panels show egg specific gravity distributions (240 d) for different PEPCK-C genotypes in combination with PEPCK-M genotypes AccI+/− (panel C) and AccI+/+ (panel D). The symbols 䊊, 䊐, and 䉭 designate PEPCK-C genotypes C−/−, C+/−, and C+/ +, respectively. Genotypes −/− are designated as 0; genotypes +/− are designated 1, and genotypes +/+ are designated 2.

1713

TRAIT ASSOCIATION OF PEPCK IN WHITE LEGHORNS 1

TABLE 4. PEPCK-M and PEPCK-C interaction: genotype combinations and associations with production traits PEPCK-M genotype PEPCK-C Genotype A+/+ A+/− A−/− B+/+ B+/− B−/− C+/+ C+/− C−/−

AccI+/+

AccI+/−

AccI−/−

n

Median2

n

Median

n

Median

10 61 59 3 30 97 40 63 27

− − − − 50.5a g 52.4b g 85.0ab 83.3a 87.0*

9 71 87 7 48 112 45 93 29

− − − − 52.6b g 52.8b g 86.0b 85.2ab 85.6ab

2 22 19 1 14 28 7 29 6

− − − − 52.4b g 51.5ab g − 86.0ab −

Trait with significant association3 None Egg weight Period 1 (P = 0.017) Egg specific gravity Period 1 (P = 0.022)

alleles A, B, and C were not equivalent. In contrast, the distributions for the genotypes AC, BC, and CC were nearly superimposable, indicating that allele C had a dominant effect on trait distribution. The distribution of BW and EGM, independent variables used to determine the predicted feed consumption, did not vary among the different genotypic classes (Figure 1). However, there were some differences in feed consumption. Similar to RFC, the AB genotypes tended to have higher feed consumption, whereas the BB genotypes tended to have a lower feed consumption. The distribution of RFC values indicated how well the data of the different genotypic classes fit the regression equation of feed consumption on EGM and BW derived from the entire strain. It indirectly tested for differences in the elevation and the slopes of the regression equations derived separately for each of the genetic groups. A graphic presentation of the partial regression lines of the feed consumption on EGM and BW is shown in Figure 2. The elevations reflect the feed efficiencies for production of EGM and BW, respectively, whereas the slopes (regression coefficients) are linear approximations of the change of these efficiencies with increasing production traits. For EGM, BB appeared to be more efficient than the other genotypes throughout the entire range. For the BW maintenance, BB was only more efficient for individuals with low BW but not with high BW. The genotypic class AB, which also markedly differed in the distribution of RFC values, had a lower elevation and higher slope for EGM. However, in this case the feed consumption was higher throughout the entire range of EGM values. The partial regression of feed consumption on BW of AB was similar to the majority of the genotypes.

Deviation from Hardy-Weinberg Equilibrium The PEPCK-C genotypes defined by the three alleles deviated from Hardy-Weinberg equilibrium (Hernandez

and Weir, 1989). There was an excess of heterozygotes AC and a deficit of the homozygotes AA and CC (Table 3). In contrast, the heterozygotes AB and BC occurred at the expected frequencies. Deviations from equilibrium were not observed for other genes, such as GH, ODC, IGF-1, and PEPCK-M (data not shown). Because the strain was propagated at an estimated effective population size of 457, we assumed that the deviation from equilibrium was due to differential fitness. A comparison between the genotypic class AC (excess) with the combined genotypic classes AA and CC (deficient) revealed similar medians for the traits listed in Table 2, with the exception of the housing BW (t-test, P = 0.024). In addition to the median, the variance in class AC was also lower than in the combined classes AA and CC, with the net effect that the difference between the two BW distributions increased with increasing BW (Figure 3). Hence, PEPCK-C may be rate limiting in chickens with high BW but not with low BW. It is unlikely that the reduction in housing BW is directly associated with reduced reproductive fitness. Rather, PEPCK-C in conjunction with other genes may affect sperm count, fertility, or hatchability via other pathways. Divergent selection for RFC, the major trait affected by PEPCK-C, has been reported to reduce reproduction and alter sperm characteristics (Morisson et al., 1997).

Interaction with the PEPCK-M The PEPCK-M carries out the same reaction as the cellular form of PEPCK but is encoded by a different nuclear gene. In order to test for interaction between the two genes, we made use of an AccI RFLP marker in the PEPCK-M gene. The three genotypes defined by this RFLP were not significantly associated with differences in the medians or means of the production traits listed in Table 2 (data not shown). The three PEPCK-M genotypes,

Downloaded from http://ps.oxfordjournals.org/ at NERL on May 26, 2015

a,b Contrasts were analyzed for traits that showed significant association with genotype combinations at the two loci. Different superscripts indicate significance (Kruskal-Wallis multiple-comparison Z-value test). 1 PEPCK = phosphoenolpyruvate carboxykinase; −M = mitochondrial form; −C = cytosolic form. 2 Median of traits that significantly varied among genotypes. 3 Only the genotype combinations with >10 observations were included in the analysis. Signifiance was assessed using the Kruskal-Wallis one-way ANOVA on ranks. Traits analyzed are those listed in Table 2. Trait associations with P > 0.1 were considered as not significant. *The combination PEPCK-C: C −/−, PEPCK-M: AccI +/+ had significantly higher median egg specific gravity (period 1) than any other genotype combination.

1714

PARSANEJAD ET AL.

Analysis of contrasts indicated that the PEPCK-M genotype affected the egg weight among the PEPCK-C genotypes B+/− but not B−/−. The genotypes B+/+ were too rare for meaningful analysis. Analyses of the combinations of PEPCK-M genotypes with genotypes of PEPCKC defined by the C-allele indicated significant differences for the SPG. The PEPCK-C genotypes C+/+, C+/−, and C−/− affected SPG in combination with the PEPCK-M genotype AccI+/+ but not AccI+/−.

Linkage Disequilibrium Between PEPCK-C and PEPCK-M

together with the six PEPCK-C genotypes gave rise to 18 different combinations. To reduce the number of combinations and increase the number of observations in each class, we analyzed the combination of the genotypes defined by one PEPCK-C allele at a time (A+/+, A+/−, A−/ −, B+/+, B+/−, B−/−, C+/+, C+/−, and C−/−). Analysis yielded nine combinations of PEPCK-C/PEPCK-M genotypes for each PEPCK-C allele. In addition, genotype combinations for which the number of observations was less than 5% were omitted from the analysis. A comparison of the trait distributions of the different PEPCK-C/PEPCK-M combinations is shown in Table 4 and Figure 4. Analysis of the combinations of PEPCK-M genotypes with genotypes defined by the A-allele revealed similar median values for the production traits listed in Table 2. For allele B significant differences were observed for the egg weights measured at 240 d of age.

TABLE 5. Genotypic disequilibrium between PEPCK-M1 and PEPCK-C PEPCK-C allele A+ B+ C+ 1

Significance of linkage disequilibrium (chi-square values2) D 13

D2

D12

D112

D122

D1122

−1.16 −1.16 −1.16

−4.19 +0.33 −4.22

+0.27 −1.79 +0.39

+1.41 −0.63 −0.19

+0.47 +0.14 +6.414

−0.43 −0.43 −0.13

Cytosolic form of phosphoenolypyruvate carboxykinase. The degree of freedom is one. The sign in front of chi-square indicates whether the observed frequency of an allele combination is smaller or larger than the expected frequency. Significant chi-square values are in bold script. 3 The numeric subscript refers to the gene and locus. The subscript 1 stands for the AccI+ allele of the PEPCKM gene, and the subscript 2 stands for the PEPCK-C allele listed in the first column. 4 A significant positive trigenic disequilibrium D122 means that the observed frequency of the allelic combination p122 (i.e., PEPCK-M AccI+, PEPCK-C A+/+) is higher than the expected frequency that is the product of the estimated allelic frequencies (p1ª p2ª p2) corrected for the lower-order disequilibria. 2

Downloaded from http://ps.oxfordjournals.org/ at NERL on May 26, 2015

FIGURE 5. Distribution of specific gravity among genotypes grouped on the basis of linkage disequilibrium between the cytosolic form of phosphoenolpyruvate carboxykinase (PEPCK) and the mitochondrial form of PEPCK (PEPCK-M). P = pool of genotypes that contain only alleles that are at an excess, and M = genotypes with alleles that are at a deficit.

Similar to the deviation from Hardy-Weinberg equilibrium that we observed for PEPCK-C alleles, we expected interaction of that gene with PEPCK-M to lead to genotypic disequilibria. Genotypic disequilibria refer to the deviation of the observed from expected allelic combinations. In a two locus per two allele system, genotypic disequilibria can be characterized by six parameters, two single gene disequilibrium coefficients (Hardy-Weinberg disequilibrium), a composite digenic disequilibrium coefficient, two trigenic coefficients, and a composite quadrigenic disequilibrium coefficient. In our case, genetic variation at PEPCK-C was characterized by three alleles and gave rise to three different sets of parameters. The significance (chi-square values) of the parameters for the PEPCK-M/PEPCK-C interaction are given in Table 5. The single genotypic disequilibria (Hardy-Weinberg disequilibrium) were significant for the PEPCK-C homozygotes A +/+ and C +/+ and indicated a deficiency as already indicated in Table 3. Among the higher-order disequilibria, one of the trigenic disequilibria was positive (P = 0.011) for the PEPCK-C allele C, indicating an excess of the allele combination (PEPCK-M, AccI+; PEPCK-C, C+/C+). The disequilibria listed in Table 5 are for the + alleles. They are interrelated with disequilibria of other combinations of + and − alleles (see Materials and Methods). Among the nine genotype combinations of the C-alleles of PEPCK-C and AccI alleles of PEPCK-M (see Table 4), three contain the triallelic combinations that are in excess

TRAIT ASSOCIATION OF PEPCK IN WHITE LEGHORNS

and three the triallelic combinations that are at a deficit. Genotypes with allelic combinations at an excess were as follows: (PEPCK-M: AccI+/+, PEPCK-C: C+/+), (PEPCKM: AccI+/+, PEPCK-C: C−/−), and (PEPCK-M: AccI−/−, PEPCK-C: C+/−); genotypes with allelic combinations at a deficit were (PEPCK-M: AccI+/+, PEPCK-C: C+/−), (PEPCK-M: AccI−/−, PEPCK-C: C+/−), and (PEPCK-M: AccI−/−, PEPCK-C: C+/+). A comparison of the pools of these two classes revealed that genotypes with allelic combinations in excess had a higher SPG in period 1 than those with allelic combinations in deficit (Figure 5). The median values were 86 and 84 units, respectively (P = 0.009). No differences were observed in period 2 or period 3.

CONCLUSIONS

increases the number of different groups while decreasing the number of observations in each group.

ACKNOWLEDGMENTS The authors thank D. Praslickova for excellent technical help and the staff of Agriculture and Agri-Food Canada for providing production data and DNA from strain 7.

REFERENCES Dopazo, J., E. Zanders, I. Dragoni, G. Amphlet, and F. Falciani. 2001. Methods and approaches in the analysis of gene expression data. J. Immunol. Methods 250:93–112. Gowe, R. S., R. W. Fairfull, I. McMillan, and G. S. Schmidt. 1993. A strategy for maintaining high fertility and hatchability in a multi-trait egg-stock selection program. Poult. Sci. 72:1433–1448. Hanson, R. W., and L. Reshef. 1997. Regulation of phosphoenolpyruvate carboxykinase (GTP) gene expression. Annu. Rev. Biochem. 66:581–611. Hernandez, J. L., and B. S. Weir. 1989. A disequilibrium coefficient approach to Hardy-Weinberg testing. Biometrics 45:53–70. Hintze, J. L. 1997. NCSS Statistical system for windows. Kaysville, Utah. Lewis, P. O., and D. Zaykin. 2001. Genetic Data Analysis: Computer program for the analysis of allelic data. Version 1.0 (d16c). Free program, http://lewis.eeb.uconn.edu/lewishome/software.html. Accessed May, 2002. Luiting, P., and E. M. Urff. 1991. Residual feed consumption in laying hens. 1. Quantification of phenotypic variation and repeatabilities. Poult. Sci. 70:1655–1662. Morisson, M., A. Bordas, J. M. Petit, C. Jayat-Vignoles, R. Julien, and F. Minvielle. 1997. Associated effects of divergent selection for residual feed consumption on reproduction, sperm characteristics, and mitochondria of spermatozoa. Poult. Sci. 76:425–431. Manly, B. F. J. 1994. Multivariate Statistical Methods: A Primer. 2nd ed., Chapman Hall, London. Nagajara S. C., S. E. Aggrey, J. Yao, D. Zadworny, R. W. Fairfull, and U. Kuhnlein. 2000. Trait association of a genetic marker near the IGF-I gene in egg-laying chickens. J. Heredity 91:150–156. Parsanejad, R., D. Zadworny, and U. Kuhnlein. 2002. Genetic variability of the cytosolic phosphoenolpyruvate carboxykinase gene in White Leghorn chickens. Poult. Sci. 81:1668– 1670. Sato, A., H. Takahashi, K. Konishi, T. Suzuki, H. Kochi. 1997. Nucleotide sequence of the promotor region of chicken cytosolic phosphoenolpyruvate carboxykinase. J. Biochem. 121:711–716. Savon, S., P. Hakimi, and R. W. Hanson. 1993. Expression of the genes for the mitochondrial and cytosolic forms of phosphoenolpyruvate carboxykinase in avian liver during development. Biol. Neonate 64:62–68. Soares, J. H., Jr. 1984. Calcium metabolism and its control. Poult. Sci. 63:2075–2083. Weir, B. 1990. Genetic Data Analysis. Sinauer Assoc., Sunderland, MA. Weldon, S. L., A. Rando, A. S. Matath, Y. Hod, P. A. Kalovic, S. Savon, and J. S. Cook. 1990. Mitochondrial P-enolpyruvate carboxykinase from the chicken: Nucleotide sequence of the cDNA and the deduced amino acid sequence. J. Biol. Chem. 165:7308–7317.

Downloaded from http://ps.oxfordjournals.org/ at NERL on May 26, 2015

We have shown that PEPCK-C, a key regulatory gene in gluconeogenesis, affects RFC. The RFC is a measure of how efficiently feed is utilized to maintain BW and produce eggs. In the strain analyzed, two of six genotypes were associated with differences in the distribution of RFC. The two genotypes did not affect the distribution of egg production or BW, but there was a trend for the distribution of feed consumption to parallel that of the RFC values. It may, therefore, be that the genetic effect of PEPCK-C on RFC was at the level of feed digestion or nutrient absorption. However, regression analyses also indicate differences in partitioning of feed energy into body maintenance and egg production. The PEPCK-C variants had little effect on the distribution of other production traits. This finding indicates that within the breeding population, such traits are well buffered against variations in energy metabolism. However, this result does not mean that PEPCK-C has no effect on traits in individuals or subgroups of individuals. Analysis of the interaction with PEPCK-M showed significant effects of PEPCK-C on EWT and SPG with certain marker combinations. Although we have singled out PEPCK-M because it is an isozyme of PEPCK-C, we have observed similar effects with other genes (unpublished results). It supports the notion that networks involving many genes stabilize complex traits within breeding populations. The RFLP were chosen as markers because they can be conveniently assayed in a large number of individuals. They are a random sample of a series of other DNA polymorphisms present in these genes (Parsanejad et al., 2002). We were therefore surprised that each genotypic class defined by these markers had different genotypic properties. This finding indicates that the PEPCK-C gene does not segregate for a single quantitative trait mutation but for a number of variants that affect phenotypes. The multiplicity of alleles and consequently of genotypes is a formidable problem in association studies, because it

1715