| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
ANIMAL GENETICS |
Institute of Animal Breeding and Genetics, University of Veterinary Medicine Hannover, Bünteweg 17 p, 30559 Hannover, Germany
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionImplicationsLiterature Cited |
|---|
Key Words: DNA Markers • Leukemia Inhibitory Factor • Litter Size • Pigs
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionImplicationsLiterature Cited |
|---|
for a PvuII-restriction fragment length polymorphism (RFLP) of the estrogen receptor (ESR) gene.
The leukemia inhibitory factor (LIF) gene encodes a pleiotropic cytokine and was chosen as a candidate gene for litter size because of its essential role in blastocyst growth and implantation in mice (Stewart, 1994; Savatier et al., 1996). Mice with a null mutation in the gene for LIF are infertile, owing to a failure of embryo implantation (Stewart et al., 1992). This implies that LIF also may serve a vital function in conceptus development and implantation—thereby litter size—in pigs (Geisert and Yelich, 1997). This implication is supported by the detection of LIF gene expression in porcine endometrium at the time of blastocyst attachment (Anegon et al., 1994; Modric et al., 2000) and the presence of LIF receptor (LIFR) mRNA in porcine peri-implantation conceptuses (Yelich et al., 1997; Modric et al., 2000). The LIFR is a specific LIF receptor subunit (Gearing et al., 1991) and a member of the cytokine-binding family of receptor subunits. Formation of a high-affinity signaling complex requires the association of the LIF-LIFR complex with another transmembrane signal-transducing molecule gp130 (Gearing et al., 1992a,b), which itself exhibits features of the cytokine family of receptors (Hibi et al., 1990).
The objective of the current study was to examine the effect of a porcine SNP-based RFLP marker in the LIF gene on litter size in a sample of 273 sows of a German synthetic pig line. To identify possible pleiotropic effects of this marker (Spötter et al., 2001) on growth and carcass traits, ADG and backfat thickness also were analyzed for an association with the marker in this population.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionImplicationsLiterature Cited |
|---|
All animals were reared on a single farm and were subjected to the same fertility management (e.g., estrous control) and insemination regimen. The population consisted of 273 sows belonging to a German synthetic line of Duroc and Large White origin. Back fat thickness (BF), scored by ultrasonic measurement at d 168, and ADG were recorded in the sows prior to the reproduction data. Daily gain was determined by dividing BW at d 168 by age in days. Number of piglets born alive (NBA) was recorded in 955 litters of sows farrowing up to 10 times. In Table 1, an overview is given of the number of animals genotyped, the available phenotypic records, and the means for NBA. Of the sows with multiple parities, there were 63 sows with two litters, 34 sows with three litters, 36 sows with four litters, 28 sows with five litters, 33 sows with six litters, and 25 sows with more than six litters. Table 1 also shows the mean values of BF and ADG of the performance-tested sows.
|
) were used to amplify a 407-bp fragment. The PCR amplification (20 µL final volume) was performed using 20 ng of genomic porcine DNA, 1x PCR buffer (Promega, Mannheim, Germany), 100 µM each dNTP, 4 pmol each primer, and 1 U Taq polymerase (Promega). Conditions were 94°C for 4 min, followed by 35 cycles of 94°C for 30 s, 58°C for 60 s, and 72°C for 30 s. Three microliters of the PCR product was digested with 3 U DraIII (N.E.B., Frankfurt/Main, Germany) and separated on a 2% Metaphor agarose gel (Cambrex Bio Science Rockland Inc., Rockland, ME). A 407-bp fragment was observed for the A allele, and 266- and 144-bp fragments for the B allele (Figure 1). The sum of the two cleaving products is greater than the original 407-bp fragment because of the 5' overhangs generated by DraIII digestion. This PCR-RFLP assay does not have an internal positive control to ensure that access to the restriction enzyme was complete. To control for incomplete or missing digestion, a sample of the previously mentioned 407-bp PCR product, including 16 of the 273 genotyped sows, was cloned into the polylinker of pCR 2.1-TOPO (Invitrogen, Groningen, The Netherlands). Recombinant plasmid DNA was sequenced with the ThermoSequenase kit (AmershamPharmacia, Freiburg, Germany) and a LICOR 4200 automated sequencer (LI-COR Inc., Lincoln, NE). Sequence data were analyzed with Sequencher 4.0.5 (GeneCodes, Ann Arbor, MI). A comparison of the sequences determined with the respective RFLP results showed complete consistency. Of the 16 animals genotyped using both methods, six had the AA genotype, and five each had the AB and BB genotypes.
|
Allele and genotype frequencies of the LIF marker were calculated from the genotypes of the 273 sows. Hardy-Weinberg equilibrium in the studied population was tested by comparing expected and observed genotype frequencies using a 
2 test.
An animal model with the additive genetic relationship matrix for the sows, including pedigree information on 488 animals up to 15 generations of the synthetic line, was employed for the association analysis between genotypes of the RFLP marker and the different phenotypic traits. Additionally, the mates of the 273 genotyped sows (88 boars) were considered as a random permanent environmental effect. A multivariate analysis was performed simultaneously for the records of the first and second parities and for the records of third to 10th parities of the sows. The litter size trait NBA was multivariately analyzed using PEST (Groeneveld, 1990) and the following linear animal Models I and II for first, second, and third to 10th parity records, as well as for all parities.
First parity or second parity records (Model I):
 |
For records from third to 10th parity ((Model II):
 |
Year-season-classes (YSF) for farrowing and marker genotypes (GT) were treated as fixed effects. Random effects included the additive genetic (a; n = 1 to 488) effect of the sow, a random permanent environmental boar effect as the mating partner of the sow (peb; m = 1 to 88), and a random residual effect (e). For the analyses of the records from third to 10th parities and all parities of the sows, the model was extended to include parity number (PN) as a fixed effect and the random permanent environmental effect of the sow (pes; l = 1 to 273).
The number of sows per genotype and year of birth between 1993 and 1997 ranged from one to six (average = 3.6) for AA sows, from 5 to 34 (average = 22.2) for AB sows and from 8 to 44 (average = 28.8) for BB sows.
The number of litters per genotype and year season of farrowing between 1995 and 1998 ranged from two to seven (average = 4.1) for AA sows, from 11 to 43 (average = 27.0) for AB sows, and from 15 to 46 (average = 32.7) for BB sows (Table 2).
|
 |
where with 
2pes = variance of the permanent environmental effect of the sow, 
= additive genetic variance of the sow, 2peb = variance of the permanent environmental effect of the boar, and 
= mean error variance.
For the analysis of BF and ADG, the following linear animal model was used:
 |
Year and season of birth and the marker GT were regarded as fixed effects, and the additive genetic effect (a) of the sows was random. To test for the effect of contemporary group, which is a major factor causing variation of growth rate and BF, we included in this analysis the fixed effect of birth times within year (two groups of birth times per year) instead of a simple year of birth effect for sows born between 1995 and 1997, but not for the birth years 1993 and 1994 because of low numbers of sows.
Additive genetic effects were estimated by pairwise comparisons of the least squares means of the two homozygous genotypes, whereas the dominance effects were calculated as the deviations of the least squares means of the heterozygotes from the average of the two homozygous genotypes. The additive and dominance effects estimated for the sows with different parities were tested for significance by using F-tests. Additionally, each of the genetic effects was jointly tested for significance of the first and second parity records and records from the third to 10th parity.
To test whether the data given had enough power to detect meaningful differences between genotypes, a simulation study was conducted. Given the total number of observations and the observed genotype frequencies, repeated samples were created by simulating phenotypic values for each observation. These phenotypic values were the sums of genetic values and random errors. The genetic values for the three different genotypes were based on different predefined values for the dominance effect d (Falconer, 1960) ranging from 0.36 to –1.09 piglets. The random errors were drawn from a normal distribution with a mean equal to zero and a variance equal to the residual variance observed in our data set. Two different cases were tested with respect to the genotype frequencies. In the first case, the frequencies were simulated as found in the data, whereas in the second case, the two alleles had the same frequency; thus, the simulated genotype frequencies were 0.25, 0.50, and 0.25 for the three genotypes. The total number of observations was 273, 546, or 819. The simulated sample was analyzed using a generalized linear model (GLM; SAS Inst., Inc., Cary, NC), testing the hypothesis that the dominant effects were different from zero. The simulation and analysis of the data were repeated 1,000 times for each combination of dominance effect d, and the observed P-values were stored. The power was defined as percentage of evaluations in which the observed P-values were smaller than the pre-defined error rate. Therefore, the power was a function of the number of observations, the genotype frequencies, the d-value, the residual variance and the error rate 
.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionImplicationsLiterature Cited |
|---|
). Polymerase chain reaction on genomic porcine DNA with the previously mentioned primers LIF3-SNPa and LIF3SNPb generated a 407-bp fragment. After a DraIII digest, uncleaved fragments were designated allele A, and cleaved fragments were designated allele B. The frequency of the LIF allele A was 0.27, and that of the allele B was 0.73. All three genotypes (AA, AB, and BB) occurred in the genotyped population, which was found to be in Hardy-Weinberg equilibrium for the genotyped locus (2 = 0.30; P = 0.86). The genotype frequencies for AA, AB, and BB were 0.07, 0.40, and 0.53, respectively.
The repeatability for litter size was 0.15 (h2 = 0.07). The total phenotypic and additive genetic variance of NBA were 
= 6.57 and = 0.48, respectively, in the multivariate model including the genotypic effect for the SNP within the LIF gene.
Additive and dominance effects of the genotypes are shown in Table 3. A significant dominance effect of –0.73 ± 0.36 (P = 0.047) was detected in the first parity, and a notable, but nonsignificant, trend for the dominance effect of –0.77 ± 0.42 (P = 0.067) was detected in the second parity. For third to 10th parity, a nonsignificant dominance effect of –0.44 ± 0.35 (P = 0.22) was estimated. A simultaneous test of the dominance effects of the first, second, and third to 10th parity records gave an error probability of P = 0.044 (2 = 6.24). There was no significant additive effect of LIF on litter size in these data, neither for the records of the first parities nor for the records of the second and third to 10th parities (Table 3), and a simultaneous test of the additive effects for the records of the first parities, second parities, and third to 10th parities resulted in a P-value of 0.099 (2 = 10.70); however, there is a trend for animals carrying the A allele to have increased numbers of piglets born alive across all parities (Table 3).
|
= 4.50; = 0.05), the power analysis showed that such an effect would be detected with a chance of 59.8% (Table 4). Even under an error rate of = 0.01, the power was still 33.4%, which indicated that this sample contained sufficient information. Therefore, the result is probably not a false positive one.
|
shows the results of the statistical evaluation for BF and ADG. This analysis was accomplished to identify possible pleiotropic effects of the employed LIF marker. No significant effects on BF and ADG were estimated in this population, even though the heterozygotes tended to have lower BF and ADG values than both of the homozygotes. This applies also to the test for the effect of contemporary group. Here, as well, no association between the LIF gene region and BF and ADG was discovered. |
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionImplicationsLiterature Cited |
|---|
). This technology seems to be especially promising for fertility traits such as litter size because of the low heritability and the existence of appropriate genetic markers. To our knowledge, this is the first study to investigate an association between the trait litter size and the porcine LIF gene.
The LIF-associated dominance effect of –0.73 ± 0.36 (P = 0.047) observed for first-parity NBA is supported by the significant result of the 2 test performed simultaneously for the dominance effects of the records from the first and second parity and the third to 10th parities. The suitability of the data structure for the analysis is demonstrated by Table 2. There was no bias in the distribution of genotypes in year-season classes. The differences in litter size between the genotypes in this study can be explained by an advantageous effect of a recessive A allele over a dominant B allele. In other words, the positive effect of the A allele on litter size is evident only in AA homozygotes. In this context, it is remarkable that, despite the low number of AA animals used in this study, the effect of the A allele could be detected.
Nonetheless, the results presented here need to be verified by association studies with larger sample sizes, particularly with larger numbers of AA genotypes. Verification also is necessary for the nonsignificant trend for animals with the A allele to have increased NBA across all parities (Table 3).
There is evidence for the existence of both LIF-SNP alleles in different populations (Spötter et al., 2001). Associations between the marker and the trait may vary across populations, lines, or families. This was shown in several studies with diallelic DNA markers for reproductive traits. The effect of the B allele of a diallelic marker at the ESR locus differed from 0.6 to 2 piglets more per litter (Short et al., 1997). Another study showed no significant effect of the ESR genotype on litter size in 59 sows from a hyperprolific Large White line and a control Large White line (Legault et al., 1996). Vincent et al. (1998) showed that the A allele of a diallelic marker at the prolactin receptor locus is significantly associated with increased litter size in three of five commercial lines involving Meishan, Large White, Landrace, and Duroc. In contrast, Drögemüller et al. (2001) reported an additive effect of the B allele of this marker on NBA across all parities in a Duroc population. The previously mentioned studies demonstrate the difficulties in confirming previously published candidate gene effects in different genetic groups and show the need for studies of marker effects in different lines because of allele effects that differ between lines or populations. The observed differences between the lines may be explained through variations in the genetic background or different linkage phases between the markers and a causal mutation caused by recombination. In addition, still unknown QTL with effect on litter size could be linked to these gene-associated markers.
An association between the effect of the RFLP marker tested in this study on NBA and both of the tested growth and carcass traits (ADG and BF) was not ascertained in the genotyped population. The estimated effects did not reach the significance level of P < 0.05 (Table 3).
The possibility that a gene is really involved in a trait of interest is enhanced by coincidence between the chromosomal localizations of a QTL and a newly mapped candidate gene when there is a congruency between the affected QTL-linked reproductive trait and the physiological role the candidate gene takes in reproduction. The main reasons to choose the porcine LIF gene as a candidate gene for litter size were its expression in peri-implantation pig conceptuses and the failure of embryo implantation in mice when LIF is not present (Stewart et al., 1992). These findings imply that LIF also may serve a vital function in conceptus development and implantation—and thus litter size—in pigs (Geisert and Yelich, 1997). The LIF gene was physically mapped to SSC14q2.1-q2.2 using FISH and radiation hybrid mapping (Spötter et al., 2001). Based on porcine comparative cytogenetic, genetic, and radiation hybrid maps (Milan et al., 2000), the corresponding position on a genetic map was determined to be between S0162 at 38.5 cM and SW6 at 39.4 cM (Rohrer et al., 1996). The only litter size-related QTL identified on SSC14 to date is for total number of born piglets (de Koning et al., 2001) at 62 cM. In other QTL studies, no QTL for litter size or its component traits were detected (Rathje et al., 1997; Rohrer et al., 1999; Wilkie et al., 1999; Cassady et al., 2001). Thus, LIF does not map within a region reported to contain putative QTL for litter size or other reproductive traits in pigs.
The murine Lif gene maps to MMU11 at 0.25 cM. In a search for murine QTL affecting litter size, Peripato et al. (2004) found an additive-by-additive epistatic interaction between the loci D11Mit333 on Chromosome 11 and D14Mit5 on Chromosome 14; however, there is no correspondence between D11Mit333 on MMU 11 at 66 cM and the chromosomal localization of the murine Lif gene, which were assigned to opposite ends of the chromosome.
This lack of correspondence is perhaps a consequence of the sample sizes employed in most QTL studies for litter size, limiting the power of the methods used to detect QTL of modest effect (Kirkpatrick, 2002). Furthermore, lack of correspondence is no reason to exclude strong physical candidate genes, such as the porcine LIF, from examinations of their effects on litter size.
|
Implications |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionImplicationsLiterature Cited |
|---|
|
Footnotes |
|---|
2 Correspondence—phone: 49-511-953-8877; fax: 49-511-953-8582; e-mail: andreas.spoetter{at}tiho-hannover.de (or ottmar.distl{at}tiho-hannover.de.)
Received for publication February 7, 2005. Accepted for publication June 20, 2005.
|
Literature Cited |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionImplicationsLiterature Cited |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
