Identification and characterization of single nucleotide polymorphisms in 12 chicken growth-correlated genes by denaturing high performance liquid chromatography

The genes that are part of the somatotropic axis play a crucial role in the regulation of growth and development of chickens. The identification of genetic polymorphisms in these genes will enable the scientist to evaluate the biological relevance of such polymorphisms and to gain a better understanding of quantitative traits like growth. In the present study, 75 pairs of primers were designed and four chicken breeds, significantly differing in growth and reproduction characteristics, were used to identify single nucleotide polymorphisms (SNP) using the denaturing high performance liquid chromatography (DHPLC) technology. A total of 283 SNP were discovered in 31 897 base pairs (bp) from 12 genes of the growth hormone (GH), growth hormone receptor (GHR), ghrelin, growth hormone secretagogue receptor (GHSR), insulin-like growth factor I and II (IGF-I and -II), insulin-like growth factor binding protein 2 (IGFBP-2), insulin, leptin receptor (LEPR), pituitary-specific transcription factor-1 (PIT-1), somatostatin (SS), thyroid-stimulating hormone beta subunit (TSH-β). The observed average distances in bp between the SNP in the 5'UTR, coding regions (non- and synonymous), introns and 3'UTR were 172, 151 (473 and 222), 89 and 141 respectively. Fifteen non-synonymous SNP altered the translated precursors or mature proteins of GH, GHR, ghrelin, IGFBP-2, PIT-1 and SS. Fifteen indels of no less than 2 bps and 2 poly (A) polymorphisms were also observed in 9 genes. Fifty-nine PCR-RFLP markers were found in 11 genes. The SNP discovered in this study provided suitable markers for association studies of candidate genes for growth related traits in chickens.


INTRODUCTION
Several quantitative traits for production such as growth, egg laying, feed conversion, carcass weight and body weight at different day-ages are important in domestic animals. These traits are controlled by genetic factors, also called quantitative trait loci (QTL). Progress has been made in mapping QTL for production traits by using microsatellite markers [29-31, 36, 38, 39], but fine mapping of QTL requires a much higher density of informative genetic markers. Due to the apparent lower complexity of the chicken, as compared to mammalian genomes, there seems to be lower numbers of microsatellite DNA markers present in the genome.
SNP are a new type of DNA polymorphism, mostly bi-allelic, but widely distributed along the chicken genome [40]. In humans, several high resolution SNP maps have been created for several chromosomes or even the whole genome, providing useful resources for studies on haplotypes associated with human diseases [2,23,28]. Furthermore, an SNP map of porcine chromosome 2 has been reported [18], however such studies have not been performed in the chicken yet. Nevertheless the results of the Chicken Genome Project, which ended in February of 2004, (http://genome.wustl.edu/projects/chicken/) enable the utilization of the draft sequence to identify SNP.
The candidate gene approach is an interesting way to study QTL affecting traits in chickens. As in mammals, the growth and development of chickens are primarily regulated by the somatotropic axis. The somatotropic axis, also named neurocrine axis or hypothalamus-pituitary growth axis, consists of essential compounds such as growth hormone (GH), growth hormone releasing hormone (GHRH), insulin-like growth factors (IGF-I and -II), somatostatin (SS), their associated carrier proteins and receptors, and other hormones like insulin, leptin and glucocorticoids or thyroid hormones [7,26]. SNP markers in genes for this network could function as candidate genes for the evaluation of their effects on chicken growth traits [5].
Previous studies have shown that some SNP of the somatotropic axis genes indeed affected (economic) traits or diseases either in domestic animals or in humans [7,26]. In chickens, certain SNP of GH [11], GHR [11,12], IGF-I and -II genes [3,41] have been reported to be associated with chicken growth, feeding and egg laying traits. The SNP in the porcine pituitaryspecific transcription factor-1 (PIT-1) gene are also significantly related to carcass traits [33]. In humans, point mutations in ghrelin, PIT-1 and thyroidstimulating hormone beta subunit (TSH-β) genes have significant relationships with obesity [37], congenital hypothyroidism or pituitary dwarfism [4,27], and TSH-deficiency hypothyroidism [9], respectively. Until now, only limited SNP have been identified in these and other important genes of the chicken somatotropic axis. In part because the sequence of these genes was unknown, and since few efficient methods are available to identify SNP in chromosomal regions spanning 100 kb or even 1 Mb.
The present study was conducted to identify SNP in the complete sequences of 12 chicken genes of the somatotropic axis in four chicken populations that were significantly different in growth and reproduction characteristics. The 12 selected genes are GH, GHR, ghrelin, growth hormone secretagogue receptor (GHSR), IGF-I and -II, insulin-like growth factor binding protein 2 (IGFBP-2), insulin, leptin receptor (LEPR), PIT-1, SS, TSH-β. The sequences were obtained from Genbank [25] and were used to design gene specific primers for the identification of SNP. Denaturing high-performance liquid chromatography (DHPLC) was used to identify SNP because it is an efficient way for screening sequence variation. The SNP identified with DHPLC were also confirmed by direct sequencing. In addition, the possible effects of these SNP on growth and laying traits were analysed. Potential PCR-RFLP markers were also deduced when looking for restriction sites within sequences explored for SNP.

Chicken populations
Four chicken breeds with different growth-rates, morphological characteristics, and laying were used in this study: Leghorn (L), White Recessive Rock (WRR), Taihe Silkies (TS) and Xinghua (X). Genomic DNA of 10 animals per breed were isolated from the blood. The Leghorn is a layer breed and has been bred as a laying-type for dozens of years, whereas WRR is a fast-growing broiler line that has also been bred as a meat-type for many generations. Both TS and X chickens are Chinese native breeds with the characteristics of being slow-growing, and having lower reproduction and favorable meat quality. They have not been subjected to dedicated or intensive breeding programs.

Primer design and PCR amplification
The sequences of the 12 chicken candidate genes of the somatotropic axis are obtained from Genbank (http://www.ncbi.nlm.nih.org). The accession numbers are given in Table I. Primers were designed using the GENETOOL program (http://www.biologysoft.com/).    The twenty-five µL PCR reaction mixture contained 50 ng of chicken genomic DNA, 1 × PCR buffer, 12.5 pmol of each primer, 100 µM dNTP (each), 1.5 mM MgCl 2 and 1.0 Units Taq DNA polymerase (all reagents were from the Sangon Biological Engineering Technology Company; Shanghai, China). The PCR conditions were 3 min at 94 • C, followed by 35 cycles of 30 s at 94 • C, 45 s at certain annealing temperatures (ranged from 55 • C to 68 • C for each primer), 1 min at 72 • C, and a final extension of 5 min at 72 • C in a Mastercycler gradient (Eppendorf Limited, Hamburg, Germany). The PCR products were analyzed on a 1% agarose gel to assess the correct size and quality of the fragments.

SNP identification with the DHPLC method and sequencing confirmation
Mutation analysis was conducted with the DHPLC method on a WAVE  DNA Fragment Analysis System (Transgenomic Company, Santa Clara, USA). Eight µL PCR products from each pair of primers were loaded on a SaraSep DNASep column, and the samples were eluted from the column using a linear acetonitrile gradient in a 0.1 M triethylamine acetate buffer (TEAA), pH = 7, at a constant flow rate of 0.9 mL per min. The melting profile for each DNA fragment, the respective elution profiles and column temperatures were determined using the software WAVE Maker (Transgenomic Company, Santa Clara, USA). Chromatograms were recorded with a fluorescence detector at an emission wavelength of 535 nm (excitation at 505 nm) followed by a UV detector at 260 nm. The lag time between fluorescence and UV detection was 0.2 min.
According to the DHPLC profiles, the representative PCR products with different mutation types were purified and sequenced forward and reverse by BioAsia Biotechnology Co. Ltd (Shanghai, China). The sequences obtained were analyzed using the DNASTAR program (http://www.biologysoft.com/) for SNP confirmation.

Calculations
In order to obtain an estimate of nucleotide diversity, the normalized numbers of variant sites (θ) was calculated as the number of observed nucleotide changes (K) divided by the total sequence length in base pairs (L) and corrected for sample size (n), as described by Cargill et al. [8]. The formula is as follows:

Locating genes on chromosomes
The chicken genome sequence draft could be obtained from http://genome.ucsc.edu/cgi-bin/hgBlat and http://genome.wustl.edu/projects/ chicken/. By BLAST analysis, the locations of all 12 genes in the chromosomes were made clear, which was consistent with the original mapping results of some genes [10,16,32,34,42].

Characterizations of the primers
Ninety-two primer pairs were tested in this study, of which seventy-five successfully amplified specific fragments. There were 9 primer pairs for GH, 11 for GHR, 7 for ghrelin, 7 for GHSR, 10 for IGF-I, 3 for IGF-II, 9 for IGFBP-2, 4 for insulin, 2 for LEPR, 7 for PIT-1, 2 for SS and 4 for the TSH-β gene. The details of these 75 primers, including their nucleotide constituents, length of PCR products, annealing temperature for PCR and column temperature for DHPLC, are shown in Table I

PCR amplification, DHPLC profiles and sequencing confirmation
In 40 animals from the four divergent breeds used for SNP identification, good quality PCR products were obtained using each of these 75 pairs of primers. After PCR products were analyzed with the WAVE  DNA Fragment Analysis System, different DHPLC profiles were observed among 40 individuals (example shown in Fig. 1). Different nucleotides among individuals with different DHPLC profiles were identified, and their sites and nucleotide mutations were determined by direct sequencing (Fig. 1). In addition, three genotypes in each SNP can also be easily determined by direct sequencing (Fig. 1).

Single nucleotide polymorphisms in 12 chicken candidate genes
In total, 283 SNP were identified in 31 897 bp of sequence within the 12 selected genes. The SNP markers are summarized in Table II. Considering the 12 genes as a whole, every 113 bps generated one SNP on average, giving rise to its corresponding θ value of 2.07 × 10 −3 . The average spread in bps per SNP and per gene region is presented in Table III. The 283 SNP identified contained 74.2% of transitions (210 SNP), 11.3% of transversions (15), and 1.8% of indel (5). All SNP obtained were bi-allelic  polymorphisms except in two cases: a tri-allelic SNP was observed in the insulin gene (T/C/A, nt 1295 of AY 438372) and the other in the LEPR gene (T/G/A, nt 885 of AF 222783). For these two tri-allelic SNP, sequencing artefacts were excluded by performing repetitive sequencing for several individuals with different genotypes.

Non-synonymous SNP
Fifteen non-synonymous SNP were identified in the present study, most of which (12 of 15) affected the translated mature proteins (Tab. IV). In the GH gene, G1494A and G2075A changed the signal peptide (A13T) and mature protein (R59H) respectively. Five SNP of G1359A (A442T), G1475C (Q480H), G1507T (S491I), A1512T (T493S) and G1599C (E522Q) all occurred in the intracellular region of the GHR gene, but they had no influence on the conserved features of 5 cysteine residues in this domain. A1071T and C3833T altered the mature protein of the GHSR gene with the amino acid changes of N227Y and A323V. Transitions A499G and A761G in the PIT-1 gene led to the changes of M167V and N254S, however, the conserved POU domain was not affected. A2355G was located in the coding region of preproghrelin. A275G (Q79R) and A370G (K111E) of the SS gene changed the precursor and mature somatostatin-14 (or -28) respectively.

Other sequence variations identified
Seventeen DNA sequence variations, other than SNP, were identified in 9 genes: GH, GHR, ghrelin, GHSR, IGFBP-2, insulin, PIT-1, SS and TSH-β. These changes included 15 cases of indel polymorphisms of no less than 2 bp and 2 cases of polymorphic numbers of continuous A nucleotide in the present study. Most of these variations were polymorphisms with minor allelic frequencies over 1% (Tab. V). These variations occurred in non-coding regions of each functional gene, and did not change the terminal products of translated precursors.

PCR-RFLP DNA markers
From the 283 SNP and 17 other variations, 58 SNP and one case of a 6 bp indel polymorphism, led to the presence or absence of some restriction sites. As a result, 59 PCR-RFLP markers were developed, but they were not validated experimentally. The numbers of markers developed for the 12 genes are summarized in Table VI. All these PCR-RFLP markers were located in either coding regions (synonymous and non-synonymous) or non-coding regions such as 5'-flanking, 5'UTR, intron and 3'UTR. Furthermore, the choice of a PCR-RFLP marker was also based on the cost of the restriction enzyme. 1 Refer to Genbank accession number. 2 Indicates minor allele frequencies. 3 Means some results were proven by previous studies; AB075215 and AY299454 are Genbank accession numbers; "SSC 1(+) 86752736∼86752792" refer to the inserted 57 bp sequences were nt 86752736∼86752792 of chromosome 1(+) published by the Chicken Genome Project (http://genome.wustl.edu/projects/chicken/).

DISCUSSION
In this study DHPLC was successfully used to discover SNP in functional chicken genes. As a highly sensitive and automated method, DHPLC is mainly based on the capability of ion-pair reverse-phase liquid chromatography  to resolve homoduplex from heteroduplex molecules under conditions of partial denaturation [15]. Currently, DHPLC seems to be limited in distinguishing different kinds of homoduplex and in genotyping individuals for each SNP, especially when several SNP are present in a DNA fragment [22]. For this reason, and due to the small sample size (10 individuals for each breed) used in this study, the allele frequency of each SNP in four chicken breeds was not calculated. Nevertheless allele frequency estimates would provide important information for a future evaluation of the potential effect of each SNP.
In the present study, 283 SNP were identified in a total length of 31 897 bp of DNA, covering the 12 chicken genes in the somatotropic axis. The results provide basic information on the distribution and characteristics of SNP in chicken genes. The average bps per SNP in the 12 selected genes was very low (113 bp), consequently the nucleotide diversity seems to be much higher in chickens even when this is adjusted for the small sample size studied (40 individuals or 80 chromosomes) (Tab. III). In human SNP screening studies, the SNP density reported is much lower, and one SNP is reported to occur in every 1000-2000 bases when two human chromosomes are compared [2,23,28]. Another study analysing SNP incidence in 106 human genes, provided a higher density of one SNP per 348 bp, and their θ values of synonymous and non-synonymous SNP in coding regions were 1.0 × 10 −3 and 1.96 × 10 −4 when corrected for sample size. These θ values were quite comparable to our results [8]. The lower SNP density reported in humans might be due to the fact that fewer intronic SNP were identified and sequences of less individuals were compared. On the contrary, the chicken genome is much more compact than that of humans, since their genome size were almost 3.2 and 1.1 billion respectively (http://genome.ucsc.edu/cgi-bin/hgBlat). The higher SNP incidence in chickens seemed to compensate for its small genome size and much lower repetitive DNA (including microsatellite sequences) occurrence. A forthcoming paper that focuses on millions of SNP in the chicken genome will be available soon in Nature. In the pig, a recently developed SNP map of chromosome 2 showed that the SNP density is much higher [18], which is in accordance with the present study.
Among 283 SNP, 278 were single-base substitutions and only 5 were single base indels. Furthermore, over 74% of the SNP (210 of 283) were transitions, similar to the ratio (75%) obtained from 10 human genes [14]. Although most SNP were bi-allele polymorphisms, two tri-allelic variations were observed in the insulin gene (T/C/A, nt 1295 of AY 438372) and the LEPR gene (T/G/A, nt 885 of AF 222783), respectively. Since expected introns had higher SNP densities than coding regions and up-or down stream regions because of selection pressure on exons and flanking regions, the latter is likely to be related to the control of expression levels.
In this study, most 283 SNP of the 12 candidate genes identified are from TS and X chickens, which seems to indicate that the two Chinese native chicken breeds are more diverse than the two commercial breeds. It has previously been shown that the level of heterozygosity in commercial broilers and layers is lower than that observed in Chinese native chicken breeds in allozymes, random amplified polymorphism DNA and microsatellite DNA polymorphisms [43]. The long-term and intense selection for growth and production traits has resulted in decreasing diversity of the Leghorn and WRR breeds. However, further study is needed on the effect of the observed variation and the differences in growth rate and egg production between these breeds. The SNP from the 12 candidate genes identified in the present study provides suitable genetic markers for the analysis of such differences.
The twelve functional genes studied are all key factors in the chicken somatotropic axis, and play crucial roles in growth and in the metabolism of the chicken. There might be certain underlying relationships between some of the SNP identified in these genes and quantitative traits like growth and carcass traits. The SNP or more specifically the 59 PCR-RFLP markers identified in this study provide a good opportunity to perform association studies for growth or reproduction related traits in the diverse breeds used.
A few SNP of these twelve genes have been reported previously, and some of them are related to growing, laying, meaty quality or disease-resistance traits. In the chicken GH gene, several SNP in introns have been identified and reported to be associated with growth, egg production and disease resistance [11,13,20]. Sex-linked dwarf chickens are just due to a mutation at an exon-intron splicing site of the GHR gene [17]. Another SNP that led to the presence or absence of a poly (A) signal in intron 2 was found to influence ages at first egg and egg production from 274 to 385 days [11,12]. Two SNP in the IGF-II gene were significantly related to growth and feeding traits [3].
Fifteen non-synonymous SNP changed the translated precursor of the chicken GH, GHR, ghrelin, IGFBP-2, PIT-1 and SS, and could affect the normal function of the mature proteins (Tab. IV). Other SNP in non-coding regions of 5'UTR, 3'UTR and introns, could also affect gene expression levels because of regulatory elements present in 5'UTR or 3'UTR regions [21]. These SNP with obviously different allelic frequencies between high reproduction (L) or fast-growing breeds (WRR) and slow-growing ones (TS and X) could contribute to their divergent growth performance (Tab. VI).
In seventeen other types of sequence variations (Tab. V), some of them were consistent with previous studies. A 50 bp deletion was reported to be present in the chicken GH gene of Chinese TS [24]. A 1773 bp deletion in exon 10 and 3'UTR of the GHR gene, however, have been proven to translate into a dysfunctional precursor and could explain the existence of sex-linked dwarf chickens [1]. For the PIT-1 gene, a 57 bp indel polymorphism in intron 2 was quite frequent both in Chinese native chickens (TS and X) and commercial lines (L and WRR). This indel was confirmed by a Genbank sequence (AY396150) and the released genome sequence (nt 86752736∼86752792 of Z chromosome) of the Chicken Genome Project (http://genome.wustl.edu/projects/chicken). For the SS gene, many variations have been described in several species, including chickens, however, insertion or deletion of dozens of bps has not been reported before [35]. Since the SS gene consists of two exons in nearly all species, the 81 bp insertion in the chicken SS gene in the present study is remarkable. This might mean that the SS gene of the chicken contains 27 additional amino acids. Further study at the functional level is needed to asses the biological effects of this large insertion.
In conclusion, 283 SNP and 17 other variations in 12 chicken growthcorrelated genes were identified in the present study. Some of these SNP could serve as useful markers for association studies for growth related traits, since there are indications that there are allele frequency differences among diverse chicken breeds.