Novel BsuRI-c.930A>G-FSH Associated with Litter Size Traits on Large White X Landrace Crossbred Sows

The objective of this experiment was to identify the novel single nucleotide polymorphisms (SNPs) on porcine follicle stimulating hormone  subunit (FSH) genes. Moreover, their association with litter size traits in commercial pigs will be analyzed. 1,155 Large White x Landrace crossbred sows were bled and DNA was extracted. The records included total number of piglet born (TNB), number of piglet born alive (NBA), number of piglets stillbirth (SB) and number of piglets mummified (MM). The known sequence of porcine FSH (GenBank accession no. D00621.1) was screened homology to known sequences in term of express sequenced tags (ESTs) in public domain, GenBank. The primers were designed for amplified the novel BsuRI-c.930A>G-FSH fragment which confirmed by PCR-RFLP and nucleotide sequencing then genotyping. The favorable cut homozygous G/G allele was highly significant higher than A/G allele in terms of TNB and NBA. While, published marker HaeIII-g.5894A>G-FSH was not significant difference in any litter traits. For haplotype analysis, c.930A>G-g.5894A>G of FSHβ, unfavorable GG/AA haplotype was significantly lower TNB and NBA than other haplotypes. Conversely, this haplotype was significantly higher MM than others. The study concluded that BsuRI-c.930A>G and the haplotype of HaeIII-g.5894A>G BsuRI-c.930A>G of FSH may used for marker-assisted selection on pig breeding program.


Introduction
Reproductive performance, especially litter size in terms of TNB and NBA are one of the most economically important traits in pig production.Reproductive success in the pig viewed the number and quality of piglets produced (Kyriazakis and Whittemore, 2006).It is well known that litter size is affected by many factors such as genetics, farm, feeds, season mating boars and fertilization rate, which impact for a large part of the phenotype variance.From a genetic background, litter size might be controlled by numerous genes in complicated physiological metabolisms such as those affecting on ovulation rate, fertilization rate, embryo survival, uterine capacity (Foxcroft et al., 2006), fetal survival and pre-weaning losses (Distl, 2007).Nowadays, mean of total number of piglet born is 10.03+2.64 piglets/litter (Imboonta et al., 2007).Litter size is low heritability approximately 0.09 for NBA that affecting slowly changed in genetic drift (Rothschild, 1996;Distl, 2007) and difficult to measure phenotypically by traditional breeding method (Merks et al., 2000).Furthermore, litter size trait is sex-limited and is not measurable until sexual maturity (Spötter and Distl, 2006).This breeding method is the-state-of-the-art and setback over 3-5 years or up to 10 generation that gain of about 0.6 piglets within-line selection (Bolet et al., 2001).
Polymorphism of candidate genes and their association with litter size was intensively studied in many laboratories all over the world (Ernst et al., 2003).The classical molecular marker for litter size traits is estrogen receptor gene which SNP of estrogen receptor gene associated with TNB and NBA in pigs (Short et al., 1997a).Other candidate genes affected litter size traits were prolactin receptor (Drogemüller et al., 2001), retinol-binding protein4 (Linville et al., 2001), osteopontin (Short et al., 1997b), leukemia inhibitory factor (Spötter et al., 2005) folate binding protein (Valet et al., 2005a) and erythropoietin receptor (Vallet et al., 2005b).Bertani et al. (2004) identified transcriptional candidate genes for fertility in anterior pituitary gland of pigs.FSH was found 10% highly expressed between sows selected for ovulation rate and embryo survival compared with the control line.Moreover, FSH gene was chosen as a direct candidate gene because its functions in follicle maturation (Wang and Greenwald, 1993a, b;Simoni and Nieschlag, 1995).Also, in a direct candidate gene analysis, Li et al. (1998), Zhao et al. (1999) and Liu et al. (2009) found that the genetic polymorphism of intronic FSH was associated with litter size in Chinese breed pigs.Whereas, no association of these intronic polymorphism with litter size traits in Large White/Landrace composite (Linville et al., 2001) and Polish synthetic breed (Korwin-Kossakowska et al., 2003) was found.Hence, there must be another nucleotide change which is responsible for its function.The aims of this study were to identify the novel SNP on porcine FSH genes using in silico analysis.Moreover, their association with litter size traits in commercial pigs would be analyzed.

Animals and data
A totally 1,155 commercial sows (Large White X Landrace or its reciprocal) including 4,162 parities, were bled and recorded in terms of TNB which was calculated as NBA plus SB and MM on 1 st parity to 5 th parity for association study.Primary data were plotted and 20 sows were selected form extremely tails of population into 2 groups; 10 High-TNB (>15 piglets/litter) and 10 low-TNB sows (<8 piglets/litter) for screening SNPs.Associated study was analyzed between candidate gene and phenotypic traits in term of TNB and NBA and mortality traits including SB and MM.The descriptive statistics was shown as table 1.
Blood samples were collected from jugular vein with 0.5M EDTA anti-coagulant and extracted by using Chelex®.Briefly, 20 l of blood was lysed twice, centrifuge at 2,000 rpm 1 min, added Chelex® solution over the pellet and incubated overnight.The solution samples were boiled at 95C for 5 minutes and stored at 4C until analyzed (Walsh et al., 1991).

Cloning and sequencing
PCR product would be cloned into pGEM-T Easy Vector system.The cloned PCR fragment compared with PCR product obtained from genomic DNA.Three clones with the same size of inserted fragment would be sequenced on the CEQ 8000 Genetic Analysis System, Beckman Coulter by using GenomeLab TM DTCS Quick Start Kit (Beckman Coulter, USA).

PCR-RFLP
In silico SNP analysis was potentially technique that was based on BLAST software which screened homology to known sequences in term of express sequenced tags (ESTs) in public domain, GenBank.Specific primers were designed based on published sequence information of porcine FSH (GenBank accession no.D00621.1).Similarly, standard PCR was performed in a final reaction volume of 20 μl by using 50.0 ng of genomic DNA sample, 1X NH 4 SO 4 buffer, 0.4 mol of primer; f-acagttttttacaggcctta & r-ctggctgggtccttgtat, 0.5 l of 0.2 mmol dNTPs, 1.5 mmol MgCl 2 , 13.4 l dH 2 O and 0.25 unit Taq DNA polymerase (Fermentas).The PCR product was checked on 6% polyacrylamide gel electrophoresis.A 2.5 l of the PCR product of c.930A>G-FSH was digested with 0.25 unit of BsuRI by manufacturing instruction.As well as, PCR of FSH published marker (Li et al., 1998) was performed as above, the primer; f-gtataccaggtcctaag & rgtctcgtacaccagctcctt and digested with 0.25 unit of HaeIII by manufacturing instruction.The PCR-RFLP fragments of each candidate genes were ran on 6% polyacrylamide gel electrophoresis and stained with standardized silver staining protocol.

Statistic analysis
Allele and genotype frequencies of each candidate genes were determined by the total count of an allele and genotype divided by twice the number of observations.Association of SNP markers with litter size traits were analyzed with general linear model included fixed effect of genotype, parity number and residual as following: Where  denoted the average normalized of population; genotype i denoted the fixed effect of genotype i; parity j denoted the fixed effect of parity j; and e ij denoted the residual effects of genotype i, and parity j.Moreover, additive effect (a) would be calculated as by comparison of the means of the traits value for homozygote (a = ½ (BB-AA)).The dominance effect (d) = alleles A and B would be calculated from the means for three genotypes as follows: d = AB-1/2(AA+BB).The estimated effects would be tested by t-test on significant deviation from zero.

Results
In this experiment, in silico analysis for screening SNPs that used BLAST software based analysis was used.The FSH sequence (GenBank accession no.D00621.1) was screened homology to known sequences in public database with BLAST software (Altschul et al., 1997).Totally, 231 SNPs were found comprising 8 SNPs in 5'-UTR, 23 SNPs in exon1, 32 SNPs in exon2, 45 SNPs in exon3 and 123 SNPs in 3'-UTR (Table 2).Only exon3 c.930A>G SNP of FSH gene was potentially found.Consequently, according to this in silico SNP was proved by using PCR-RFLP technique.
Interestingly, BsuRI-FSH was discovered by in silico analysis and proved by PCR-RFLP that did not know about the nucleotide change of this SNP between polymorphic patterns of homozygous uncut and homozygous cut.Consequently, homozygous uncut and homozygous cut of BsuRI-FSH was sequenced and aligned with BLAST software.Transition of c.930A>G of FSH was found that confirmed in silico analysis.

Genotype and allele frequencies for the investigated SNP for FSH gene
Frequencies for the FSH gene in 1,155 parent stock of commercial sows including 4,162 parity records were shown in Table 3. Frequency of A and G allele of BsuRI-c.930A>G-FSH was 0.81 and 0.19, respectively, while frequency of HaeIII-g.5894A>G-FSH(according to Li et al., 1998) was 0.70 and 0.30, respectively.The allele frequencies of these two SNPs were not distributed according to Hardy-Weinberg equilibrium proportion.

Association study and haplotype analysis
The association between FSH genotype and litter size traits in terms of TNB, NBA, SB and MM piglets was analyzed as shown as Table 4.For BsuRI-c.930A>G-FSH, G/G genotype of FSH was highly significant difference than those A/A and A/G genotype in TNB and NBA.Significantly dominance effect of BsuRI-c.930A>G-FSH on TNB and NBA was found.While, HaeIII-g.5894A>G-FSHgene, according to Li et al. (1998), all litter size traits including TNB, NBA, SB and MM were not significant difference in this population.
The haplotype analysis between c.930A>G-FSH and HaeIII-g.5894A>G-FSHon litter size traits was shown in table 5. Higher litter trait was expected, adversely, mortality trait was nil.However, the unfavorable haplotype GG/AA was significant different than others haplotype in TNB, NBA and MM, the GG/AA was highly significant lower TNB and NBA than other haplotypes, otherwise, was highly significant higher MM than other haplotypes.
In present study, 2006-2008, 1,155 Large White x Landrace crossbred sows showed 11.24+0.05and 9.91+0.05piglets/litter in TNB and NBA, respectively.These parameters exhibited the genetic progress in Thai's swine commercial that tended to be favorable for white breeds and widely used by the commercial to produced F 1 crossbred gilts resulting from imported nucleus herd.Moreover, crossbred gilts tended to younger at puberty, weighted more at farrowing, and produced larger, heavier litters to weaning than the purebred gilts (Cassady et al., 2001).In addition, the main direction of selection conventional method for this founder pigs was the meat deposition in carcass and growth rate, but was not considered in the litter size traits and its related traits.

SNPs discovery techniques
There are two different strategies are employed to detect markers for piglet number and its related traits; linkage analyses and direct candidate gene.These approaches were employed to detect genes influencing piglet number and its related traits.Candidate gene approaches identified via their physiological or functional role in the reproductive system in the pig, while linkage analysis detected genomic regions that responded the trait of interest as known as positional candidate (Distl, 2007).All of them were focus on SNPs that associated with economically traits as well as piglet number and its related traits.Normally, SNPs detection referred to the identification of allelic variants of single or low-copy-number gene sequences.Direct sequencing of specific sequence regions is still the most reliable and accurate method of SNPs detection, but it is expensive and labor intensive.PCR-based technique, however, was used to analyze that offers the powerful advantage of allowing one to conveniently pre-screen large numbers of unknown sample of which only implicated variants would need to be directly sequenced (Armersham, 1998).
In silico SNP analysis was potentially technique that is based on BLAST software that screened homology to known sequences in term of express sequenced tags (ESTs) in public domain, GenBank (Marth et al., 1999;Savage et al., 2005 andKerstens et al., 2009).The main advantages of using EST source were that markers closely associated with or directly in the coding region, could be identified that possible leaded to amino acid substitutions or perhaps leaded to functional differences which could be associated with phenotypic effect in the population.Approximately 40% of this approach was newly candidate SNPs of which cost effectiveness (Picoult Newberg et al., 1999).Consequently, candidate SNPs was validated with RFLP and sequencing approach, respectively, that confirmed true positive results (Marth et al., 1999).However, simulation by Kerstens et al. (2009) in 1.2 Gb of pig genome sequence, 98,151 SNPs were identified in which one of the sequences in the alignment represented the polymorphism and 6,374 SNPs in which two sequences represent an identical polymorphism.To benchmark the SNP identification method, 163 SNPs, in which the polymorphism was represented twice in the sequence alignment, were selected and tested on a panel of three purebred boar lines and wild boar.Of these 163 in silico identified SNPs, 134 were shown to be polymorphic in animal panel.In addition, an economically example of in silico method was the variation in POU1F1 gene that found 23 polymorphic sites of porcine POU1F1 within the intron (Song et al., 2007).Moreover, SNP of Secreted phosphoprotein 1 (SPP1) or Osteopontin (OPN) at position g.3836A>G represented an interesting DNA-marker to study phenotypic effects (Murani et al., 2009).

FSH polymorphisms discovery and association
FSH was chosen as the candidate gene because FSH was physiological role on the reproductive in pigs.The function of FSH included follicular development and sex steroid production necessary for fertility in female (Layman and McDonough, 2000).These functions were supported by studies involving knockout of the FSH gene ligand and the FSH receptor, as well as by human gene mutations (Layman et al., 2002).
Porcine FSH genomic DNA sequence composted of 10,172 bp comprising 3 exons where exon1 was nucleotide position 5,664-5,696 exon2 at 6,614-6,810 and exon3 at 6,811-8,366 that encoded 129 amino acids at nucleotide position 6,652-6,810 and 8,367-8,597 (accession no.D00621.1).Known D00621.1 sequence was aligned with ESTs from GenBank (Altschul et al., 1997).Exon2 of FSH was aligned with 16 ESTs that found 32 polymorphic sites.31 polymorphic sites were only 1 of 16 ESTs, while 1 polymorphic site was 6 of 16 ESTs that PvuII-c.128T>C-FSH was found.Moreover, exon3 was aligned with 16 ESTs that found 45 polymorphic sites.The BsuRI-c.930A>G-FSH polymorphism was found 6 of 16 ESTs.These polymorphic sites were validated and confirmed by using PCR-RFLP and with DNA sequencing, respectively.In this study, the novel polymorphism of BsuRI-c.930A>G-FSH was found that associated with TNB and NBA in sows.However, the allele frequency of BsuRI-c.930A>G-FSH in our study could not be compared with results published by other authors (Ernst et al., 2003).Classical polymorphism of FSH was discovered by Ellegren et al. (1994) that microsatellite FSH located on SSC2.Rohrer et al. (1994) reported a PCR-RFLP marker located within the first intron of FSH gene.Li et al. (1998) reported an association of FSHgene in the first intron and litter size.Zhao et al. (1998) found the insertion is a retroposon of 292 bp siting in intron I at the site between + 809 and + 810 base.Li et al. (2000) identified FSH allele in Chinese pigs that had an 11 bp deletion in 3'-UTR in anterior pituitary glands.Recent study, Li et al. (2009) reported microsatellite FSH associated with the number of litter per year, but not yet with litter size and its related traits and Liu et al. (2009) fund insertion in the first intron of FSH.
The results reported in this study provide important evidence in favor of the presence of new alleles of the FSH gene associated with litter size traits.We observed a low number of sows with G/G genotype in the BsuRI-c.930A>G-FSHgene or generally low presence of G allele in our population that was mutant type.The favorable G allele of novel BsuRI-c.930A>G-FSH in 3'-UTR was exhibited in the high-piglet number sows as well as TNB and NBA in Large White x Landrace crossbred sows.G/G homozygote sows produced on average 0.76 and 0.87 piglets more than did A/G sows for TNB and NBA, respectively.Adversely, the published marker, HaeIII-g.5894A>G-FSH(Li et al., 1998), found no associative effect between genotype and litter traits in this population.Moreover, no additive and dominance effect was found.Although large differences in allele frequencies were found for genetic marker within BsuRI-c.930A>G-FSH locus, influences of potential genetic drift were confounded with those results because the additive and dominance effects of the allele did not differ from zero for any trait (Bertani et al., 2004).The results of HaeIII-g.5894A>G-FSHfailed to confirm population wide linkage disequilibrium between the HaeIII-g.5894A>G-FSHpolymorphism and litter size traits suggesting that the degree of linkage disequilibrium probably varies between populations (Gibson et al., 2002).Zhao et al. (1998) found a retroposon in intron 1 of FSH that non-retroposon homozygote sows produced 2.53 piglets more than retroposon homozygote sows for TNB on the 1 st parity.On the contrary, recent study reported inversely association between piglet number and HaeIII-FSH locus in Polish pig line (Korwin-Kossakowsak et al., 2003), in Large White (Wang et al., 2006;Humpolicek et al., 2007).Microsatellite 5'-flanking FSH affected number at weaning, litter weight at weaning, average individual weight at weaning in Large White x Chinese Meishan F 2 (Li et al. 2008).The fluctuated result of association study on porcine FSH gene may be allele effects differed between lines or populations or the genetic background, it can be explained by a high number of low-effect genes influencing the litter size (Drogemüller et al., 2001).The results can be explained by a high number of low-effect genes influencing the piglet number.Moreover, the animal genome is a complex set that interacted on different levels of biological organization, it seems difficult to find direct marker that proofed these effects (Matousek et al., 2005).
Two of the polymorphisms, BsuRI-c.930A>G-FSH and HaeIII-g.5894A>G-FSH,were considered to be the most interesting because of an associated with litter characteristics and because these two polymorphisms, assembled together in haplotypes, specified the nine individual haplotypes revealed in the FSH gene in the Large White x Landrace sows.Individual effects were analyzed and significant differences were detected for BsuRI-c.930A>G-FSH.However, because of the significant linkage disequilibrium between FSH polymorphisms, the haplotypes were tested instead of the individual polymorphisms and found to have a highly significance effect on TNB and NBA (Ciobanu et al., 2004).Moreover, increasing the number of marker genes or haplotype raised the positive detection rate more than single marker gene alone.These results proved that using multiple markers or haplotype indeed improved the positive detection rate.The detection rate indicated the sensitivity of using the multiple markers for detection (Sher et al., 2005).Three particular haplotypes; AA/AA, AA/AG and AG/AG, within the FSH gene are highly frequent in the population and do not differ in their effects on MM, even though they carry different alleles.This might indicate the effect of other SNP linked to the two SNPs considered in this study or some degree of epistasis among the SNP within the same chromosome (Schenkel et al., 2005).
A QTL mapping approach normally uses families created form crosses based on divergent lines.Indeed, this approach has been used in relation to litter size traits using segregation analysis (de Vries et al., 2000).A number of pig populations are now being used created for the purposes of searching for litter traits QTLs including crosses between the following breeds: wild boar, Large White, Meishan, Landrace and Duroc.The QTL scan for plasma FSH level in pigs identified several chromosome regions, independent of the FSH, influencing this trait that located on chromosome 3, 8, 10 and X (Linville et al., 2001 andRohrer et al., 2001).A major genetic effect of litter size has been located on SSC 7 (Milan et al., 1998), SSC 8 (Wise et al., 2001) and on SSC 11 (Cassady et al., 2001).Furthermore, mRNA differential display of granulosa cell culture with FSH treatment has been also mapped on porcine that localized on various porcine chromosomes (Clouscard-Martinato et al., 1998).ESR was located on SSC 1 (Rothschild, 1996) and LEPR was located on SSC 6 (Wilkie et al., 1999) those have an influence on litter size in pigs.FSH was located to 2p12-16 by physical mapping (Mellink et al., 1995) and mapped on SSC2 by linkage mapping (Rohrer et al., 1994).The genome scans for regions where significantly affected plasma FSH in boars showed that located on chromosome 3, 10 and X, while no evidence has been found that any loci located on chromosome 2 affecting FSH level (Rohrer et al., 2001).Rathje et al. (1997) stated the QTL for ovulation rate on chromosome 8.Short et al. (1997b) reported association between microsatellite and OPN gene that located in the same region of the chromosome 8 and litter traits.Recent study, a 5% genome-wide significant QTL was detected at 88 cM on SSC 15 for NBA, which also showed suggestive effect on TNB, four suggestive QTL were detected on SSC 6, 7, 8 and 15 for TNB, NBA or SB, and no QTL was found for MM (Li et al., 2009).Moreover, mRNA differential display of FSH-regulated gene was localized on SSC 7 where the QTL related to litter size was found (Milan et al., 1998;Clouscard-Martinato et al., 1998).Thus, an unknown gene must stimulate changes in FSH mRNA profiles.Pomp et al. (2001) proposed that FSH locus itself has not yet been implicated as a QTL.
This FSH candidate gene study concluded that polymorphism of novel BsuRI-c.930A>G-FSH was found that associated with TNB and NBA.However, animal genome is a complete set realized through a branch network of the hormonal interactions of hypothalamic-pituitary-ovarian-uterine axis.The polymorphisms in the gene studied may not directly affect the trait.These polymorphisms could be markers linked with causative mutation within the gene or a closely linked gene.It would be of interest to examine the association study and haplotype analysis in other populations and to extend research with a large number of animals to confirm the results of this study.Moreover, it should be focused at the analysis of high number of genes in order to identify gene interaction network and to describe metabolism of gene activities.

Table 1 .
Descriptive statistics of commercial sow population

Table 2 .
Distribution of nucleotide substitution types of the FSH SNPs

Table 3 .
Genotype and allele frequency of porcine FSH

Table 4 .
Piglet number related trait in parent stock of commercial pigs on parity 1-5 of FSH gene