The GTPase Activating Rap/RanGAP Domain-Like 1 Gene Is Associated with Chicken Reproductive Traits

Background Abundant evidence indicates that chicken reproduction is strictly regulated by the hypothalamic-pituitary-gonad (HPG) axis, and the genes included in the HPG axis have been studied extensively. However, the question remains as to whether any other genes outside of the HPG system are involved in regulating chicken reproduction. The present study was aimed to identify, on a genome-wide level, novel genes associated with chicken reproductive traits. Methodology/Principal Finding Suppressive subtractive hybridization (SSH), genome-wide association study (GWAS), and gene-centric GWAS were used to identify novel genes underlying chicken reproduction. Single marker-trait association analysis with a large population and allelic frequency spectrum analysis were used to confirm the effects of candidate genes. Using two full-sib Ningdu Sanhuang (NDH) chickens, GARNL1 was identified as a candidate gene involved in chicken broodiness by SSH analysis. Its expression levels in the hypothalamus and pituitary were significantly higher in brooding chickens than in non-brooding chickens. GWAS analysis with a NDH two tail sample showed that 2802 SNPs were significantly associated with egg number at 300 d of age (EN300). Among the 2802 SNPs, 2 SNPs composed a block overlapping the GARNL1 gene. The gene-centric GWAS analysis with another two tail sample of NDH showed that GARNL1 was strongly associated with EN300 and age at first egg (AFE). Single marker-trait association analysis in 1301 female NDH chickens confirmed that variation in this gene was related to EN300 and AFE. The allelic frequency spectrum of the SNP rs15700989 among 5 different populations supported the above associations. Western blotting, RT-PCR, and qPCR were used to analyze alternative splicing of the GARNL1 gene. RT-PCR detected 5 transcripts and revealed that the transcript, which has a 141 bp insertion, was expressed in a tissue-specific manner. Conclusions/Significance Our findings demonstrate that the GARNL1 gene contributes to chicken reproductive traits.


Introduction
Egg number at 300 d of age (EN300), age at the first egg (AFE), and brooding behavior are valuable indices of chicken reproductive ability. In female chickens, sexual maturity is usually expressed as AFE. The AFE trait has been under artificial selection to enhance egg production efficiency [1]. EN300 is another reproductive trait of economic importance, while incubation behavior also affects egg production, as it results in the cessation of egg laying [2]. Chicken reproduction is controlled by photoperiod [3]. Generally, the process of chicken egg production is strictly regulated by the hypothalamic-pituitary-gonad (HPG) axis [4]. Gonadotrophin releasing hormone (GnRH) and its receptor (GnRHR) start the cascade, and neuropeptide Y (NPY) is known to inhibit GnRH secretion via its receptor (Y1R) and to control ovulation [5]. Under photo-stimulation, GnRH is synthesized, secreted by the hypothalamus and binds to its receptor, which stimulates the pituitary gland to secret gonadotrophins that evoke steroid synthesis in the gonad, regulating ovarian follicle growth and ovulation in hens [6,7]. The hypothalamic vasoactive intestinal peptide (VIP) -pituitary prolactin (PRL) neuroendocrine pathway also controls reproductive cycles via dopaminergic neurotransmission in avian HPG system [8][9][10]. PRL is a key hormone that is absolutely necessary for egg laying and incubation behavior in poultry [11,12]. After stimulation by VIP, PRL inhibits the release of gonadotropins and thereby induces and maintains chicken incubation behavior [13][14][15].
Several approaches have been applied to identify the novel genes involved in chicken reproduction. A genome-wide scan is a powerful approach to understanding this complex trait. Quantitative trait loci (QTLs) for egg number, egg production rate, AFE and broodiness were identified through genome-wide scans [34][35][36][37][38][39][40][41][42]. Genome-wide association studies based on high density SNPs can be performed to detect QTLs that could not be detected by previous studies based on microsatellite genotyping [43][44][45][46][47]. A genome-wide association study attempts to obtain information on all variants, but a gene-centric SNP approach would be efficient enough to capture SNPs associated with particular traits [48,49].
Transcriptome profiling can be also used to identify new genes associated with chicken reproductive traits. Although many studies on the genetic effects of candidate chicken reproduction genes have been reported, few studies have reported transcriptomic and proteomic changes. In previous studies, transcripts related to highegg production were identified by suppressive subtractive hybridization analysis (SSH), and several of the identified transcripts were further confirmed to be significantly increased in hens with higher egg production, though they were not part of the HPG axis [50][51][52]. Therefore, it is valuable to identify novel genes related to chicken reproduction.
The aim of the present study is to identify novel genes involved in chicken reproductive traits using SSH analysis, an Illumina 60K chicken Beadchip GWAS, and a gene-centric GWAS, with confirmation via analysis of single marker-traits, allelic frequency spectra, and alternative splicing.

Results
GARNL1 identified as a candidate gene underlying chicken broodiness by suppression subtractive hybridization A subtraction library was made by subtracting cDNA from the pituitary at the egg-laying stage. As shown in Figure 1, construction of the pituitary-subtracted cDNA libraries was successful. Genes differentially expressed between brooding and non-brooding chicken pituitary glands were enriched for and sequenced, and 57 annotation transcripts and 20 unknown transcripts were characterized (Table S1). Gene ontology (GO) analysis was performed to investigate the functions of the putatively differentially expressed transcripts. Biological process accounted for the major portion of GO annotations, compared with cellular component and molecular function. Among the category of biological process, genes were involved in processes such as eye photoreceptor cell development, ovarian follicle development, epinephrine biosynthesis, regulation of small GTPase mediated signal transduction, G-protein coupled receptor protein signaling pathways, and so forth (Table S2). On the basis of biological process annotations, 10 transcripts were selected to be validated by qPCR. Among the 10 transcripts, one was identified as belonging to the chicken GARNL1 gene ( Figure S1). The GARNL1 gene was differentially expressed between tissues ( Figure 2). Low levels of mRNA expression were detected in the ovary, oviduct, liver, spleen, lung, kidney, muscular stomach, sebum, abdomen fat, and duodenum; however, higher expression levels were observed in the cerebrum, cerebellum, hypothalamus, pituitary, heart, and glandular stomach. Gene expression levels in the cerebrum, cerebellum, hypothalamus, pituitary, ovary, oviduct and spleen were significantly higher in broody chickens than in non-broody chickens (P,0.05), with the levels in tissues from broody chickens 1.6 times to 4.3 times higher than those of nonbroody chickens. In contrast, GARNL1 expression in leg muscle was 2-fold higher in non-broody chickens.
GWAS indicates that SNPs associated with chicken reproductive traits are located in the GARNL1 gene Before GWAS analysis was carried out, stratification analysis was conducted in the two-tail sample. The IBS was not significantly different between two tails sample (P permu ,0.05). In all, 2802 SNPs were associated with EN300 in the NDH two tail sample at the 5% genome-wide level (validated by 10000 permutation tests), and of this total, 470 SNPs were at significant at the 1% level (Table S3). On chicken chromosome 5, 118 SNPs were associated with EN300 (Table S3). Among the 118 EN300associated SNPs, rs14533299 and GgaluGA282818 composed a haplotype block. The linkage distance in this block is 1691 kb, and the GARNL1 gene was observed to be located within this block ( Figure S2).
Gene-centric GWAS reveals an association of several SNPs in the GARNL1 gene with chicken EN300 and AFE Six SNPs were highly significantly associated with both EN300 and AFE (P permu ,0.05 and P permu ,0.01) ( Table 1). A SNP cluster located on chromosome 5 was associated with both AFE and EN300 in another NDH two tail sample. Among the SNP cluster, 5 SNPs were located in the GARNL1 gene.
Allelic frequency spectrum of the chicken GARNL1 gene Allelic frequencies of rs15700989 were different among the 5 populations. The frequency of rs15700989 was 1.0 in Leghorn layers (Table 4), with a highly significant difference between Leghorn layer and the other 4 native Chinese chickens. The chisquare test values for the genotype distribution of rs15700989 showed significant difference between Leghorn layer and the other 4 Chinese native chickens (P,0.01) ( Table 5), in accordance with their egg-production performance. Alternative splicing of the chicken GARNL1 gene The chicken GARNL1 gene is predicted to be located on chromosome 5 and to span positions 38,617,769-38,729,036 on the reverse strand, with a total gene size of 111,268 bp. Four, six, and seven isoforms from the pituitary, ovary, and oviduct, respectively, could be detected by Western blotting. The molecular weights of these isoforms ranged from 150 KDa to 250 KDa ( Figure 3).
The variant GANRL1-v2 (deduced to encode a 2134 AA peptide) skips exon 40 and includes a 141 bp intron sequence between the exon 16 and exon 17. RT-PCR showed that the 141 bp intron inclusion was tissue specific, being observed only in the cerebrum, cerebellum, hypothalamus, heart, pectoral muscle, and leg muscle. Its mRNA expression level was higher than the other isoforms without 141 bp intron inclusion ( Figure 4). Similarly, the GARNL1-v4 transcript contained a 201 bp fragment of intron 19, and a single amino acid change, N (Asn) to D (Asp), occurs at the new exon-exon junction. GARNL1-v4 mRNA with the 201 bp intron fragment was present at very low levels (data not shown). The mRNA expression levels of transcripts with the 141 bp intron inclusion sequence ( Figure 5) in the cerebrum, cerebellum, and hypothalamus were almost the same between brooding and nonbrooding chickens. However, its expression levels in heart and pectoral muscles of the brooding chickens were 1.5 and 2 times greater than those of the non-brooding chickens, respectively. In leg muscle, the expression was 8-fold higher in the non-brooding chickens than in brooding chickens.

Discussion
In this study, data from a SSH analyses, a GWAS, and a genecentric GWAS indicate that the GARNL1 gene is involved in reproduction and that some GARNL1 variants are associated with chicken reproductive traits.
The SSH analysis indicated that the GARNL1 gene was involved in chicken brooding behavior. Comparing to the digital gene expression methods, such as RNA-seq, SSH is not a prevailing experimental method for detecting differentially expressed genes.  SSH have several limitations, relatively low throughput, highly false positives, and generally not statistical significance. But the following qPCR validation would help to get some good results with the following experiment validation [53][54][55]. This result was consistent with previous findings. Chicken GARNL1 was identified as being potentially related to high egg production in Taiwan Country chickens [50], and higher GARNL1 expression levels have been observed in high egg producing strains [51]. Furthermore, the mRNA level of the GARNL1 gene was specifically associated with total egg number at 500 d of age or egg rate after the first egg [52]. The cerebellum was found to have the highest expression level of human GARNL1 gene among the brain tissues, corresponding to its influence on 14q13-linked neurological phenotypes [56]. In zebrafish, GARNL1 was a strong candidate gene for brain developmental delay [57]. In our study, the expression of chicken GARNL1 gene varied at different stages. We found GARNL1 to be predominantly expressed in the brain, and the levels of the GARNL1 gene were consistently higher in the hypothalamus, pituitary, ovary, and oviduct of broody hens. The expression level of transcripts that included the 141 bp intron sequence suggested that the cerebellum may be an important action region in chickens and the variants of GARNL1 do not impair their function on chicken reproductive traits. In conclusion, the expression levels of the GARNL1 gene could reflect its functions in chicken reproduction. Two tail samples were used to detect SNPs associated with broodiness and EN300 in this study. The first QTLs for broodiness were recently detected in a region within 95 cM of GGA5 [42], where the GARNL1 gene is located. In the present study, the haplotype block between rs14533299 and GgaluGA282818 was also shown to be related to EN300 (data not showed). The GARNL1 gene is located in this region. Among all 25 proteincoding genes located on this region, the GARNL1 gene was the only one that has been reported to be related to reproductive traits in mRNA level in chickens [50][51][52]. Therefore, the GARNL1 gene may be associated with reproduction in chickens.  As the GARNL1 gene might be involved in chicken reproduction, its polymorphisms could be related to chicken reproductive traits. However, no studies on the association between the mutations of GARNL1 gene and chicken reproductive traits were carried out. In humans, the GARNL1 gene was an important candidate gene for human 14q13 deletion phenotypes, and two mutations in GARNL1 were identified in a family with idiopathic basal ganglia calcification [56]. Polymorphisms of the GARNL1 gene were associated with both EN300 and AFE in a two tail sample in our gene-centric association analysis. This result confirmed our previous SSH findings and was validated in a large population. An analysis of the allelic frequency spectra of GARNL1 SNPs further supported the association. The frequencies of the rs15700989 were associated with EN300 associated with divergent egg production performance, and the frequency of predominant alleles of rs15700989 was 1.0 in Leghorn layer and was descending in Leghorn, BEH, NDH, XH, and RJF. The predominant allele of rs15700989 was related to higher EN300 trait in NDH population. Thus, the allelic frequency data supports the conclusion that the chicken GARNL1 gene contributes to chicken reproduction. The block composed of rs14532787 and rs14532779 was significantly associated with EN300 traits. Although both of them were not showed significantly relationship with EN300 after corrected by SLIDE in single marker association, the CC genotype of rs14532787 resulted in a higher EN300 and an earlier AFE than did the other two genotypes (Table S4). However, the genotype CC can be only observed in the NDH population. Compared to the variance of total egg number at 40 week between early sexual mature group and later sexual mature group in Leghorn layer [58], rs14532787 might undergo artificial selection in NDH population, aiming at the increase of egg production by early sexual mature and shortening the interval of oviposition. These two SNP may contribute to EN300 by interacting each other.
Further analysis of the organization, tissue expression, and alternative splicing of the chicken GARNL1 gene was conducted.
Using Western blotting, 5 alternatively spliced transcripts of the GARNL1 gene were isolated from chickens in this study. Note that none of the alternative splicing isoforms had impaired protein domains. Chicken GARNL1 is conserved with mammals, but it has some unique features. A variant of the human GARNL1 lacking exon 40, has been found and corresponded to GARNL1-v1 in chicken [56]. Chicken GARNL1 has lost the N-terminal coiled coil domain and subsequently the ability to bind to other proteins. In mice, GARNL1 plays a crucial role during brain formation and maintenance. A partial murine GARNL1 product identified as GRIPE (GAP-related interacting protein to E12) binds to the helix-loop-helix domain of transcription factor E12 and regulates E12-dependent target gene transcription [59]. Similar to the murine GRIPE, the region responsible for binding to HLH domains was present in all isoforms of chicken GARNL1. The Rap/Ran-GAP domain is widely distributed in signaling proteins [60][61][62], and two arginine residues in Rap/Ran-GAP domain are important for the GAP activity of GRIPE in mice [59]. Two arginine residues were found in the Rap/Ran-GAP domain of chicken GARNL1.
In conclusion, we reveal that the chicken GARNL1 gene has an important effect on chicken reproductive traits, as determined from the data from SSH analyses, GWAS, and gene-centric GWAS. This effect was validated by analysis of allele frequency spectra, and further characterization of several aspects of the gene and its expression.

Ethics Statement
The study was approved by the Animal Care Committee of South China Agricultural University (Guangzhou, People's Republic of China) with approval number SCAU#0011. Animals involved in this study were humanely sacrificed as necessary to ameliorate their suffering. Identifying candidate genes underlying chicken broodiness by SSH analysis A pair of full-sib female NDH chickens was used for the suppression subtractive hybridization (SSH) experiment. One individual was broody, and its brooding lasted for more than 7 d, and the other one was continuously laying. The pituitary gland and 17 other tissues were collected after the brooding individual had been incubating for 10 d. At that time, the chicken's incubation behavior was quite typical, and both ovary and oviducts were atrophied. The laying chickens' tissues were also collected at the same time.
Total RNA was extracted from the tissues using Trizol reagent (Invitrogen, California, USA) according to the manufacturer's protocol. Total RNA was treated with RNase-free DNaseI (Takara, Osaka, Japan) for 45 min at 37uC to ensure that it was free of DNA contamination. RNA quantity and quality were assessed using a Thermo Scientific Nanodrop1000 spectrophotometer (Nanodrop Technologies, Wilmington, Delaware, USA) and by formaldehyde denaturation agarose gel electrophoresis.
Suppression subtractive hybridization was performed with an equal amount of tester mRNA (2 mg) from the brooding stage, as well as the driver mRNA from the egg-laying stage. Following the manufacturer's protocol for the PCR-Select cDNA Subtraction Kit (Clontech, Palo Alto, CA, USA), after two subtraction hybridizations and two suppression PCRs, the subtraction efficiency was evaluated by PCR using primers for the chicken house-keeping gene G3PDH (P#1 , Table S5). cDNAs were cloned and inserted into the pMD20-T vector (Takara, Osaka, Japan) and were then transferred into chemically competent E. coli (JM109) cells to generate SSH libraries.
Subtractive products longer than 300 base pairs were picked for sequencing by Invitrigen Co. Ltd (Shanghai, China). The vector nucleotide sequences were removed, and the remaining sequences clustered into contigs using DNAstar software. The basic local alignment search tool (BLAST) http://blast.ncbi.nlm.nih.gov/ was used for identifying and annotating genes.
Quantitative real-time PCR (qPCR) was performed with the Agilent Stratagene Mx QPCR Instrumentation (Agilent Technologies, Wilmington, DE, USA) for follow-up of candidate genes, using the SsoFast EvaGreen Supermix (Bio Rad Laboratories, Hercules, CA, USA). The four individual cDNAs were used as templates for qPCR amplification. The primers used for the qPCR were designed using Primer Express 2.0 software (Applied Biosystems, Foster City, CA, USA). A housekeeping gene, the chicken beta-actin gene (accession: L08165), was used as internal control. Therefore, two sets of primers (P#2 and P#3, Table S5) were designed and used for the qPCR amplification. Each reaction mixture contained 10 mL of Eva Green PCR Master Mix, 1 mL of each primer (10 mM), 7 mL of RNase-free water and 1 mL of cDNA in a final volume of 20 mL. Standard amplification conditions were as follows: 95uC for 30 s, 40 cycles of 95uC for 5 s, 60uC for 30 s. Fluorescent signal was collected after an extension at 65uC in each cycle. Chicken GARNL1 gene relative expression was calculated by 2 2DCt method, and DCt = Ct target gene 2Ct b-actin .

GWAS for a two tail sample using 60K chips
The age of the first egg (AFE) and the total number of eggs at 300 d of age (EN300) were recorded in a breeding population kept in Guangdong Wens' Foodstuff Co. Ltd (Guangdong, China). Twenty Ningdu Sanhuang (NDH) female chickens from the above population were divided into 2 groups. Group 1 was composed of the 10 individuals with the highest EN300 values (an average of 145 eggs) and no observable incubation behavior, and group 2 was composed of the 10 individuals that had the lowest EN300 values (an average of 66 eggs) as well as an average duration of broodiness of 51 d. Twenty Illumina 60K chicken chips were used for the two-tailed association study.
Stratification analysis was performed to detect the IBS of the two-tail sample before GWAS studies were carried out. SNP quality control metrics were analyzed using GenomeStudio software (version 2009.1). A SNP was removed if its call rate was less than 100%, or its minor allele frequency (MAF) was less than 5%, and its Hardy-Weinberg equilibrium (HWE) p-value was too low (P,0.00001). As a result, 54,424 SNPs were selected for use in the GWAS. PLINK was used for the single-marker association analysis. PLINK single marker basic allelic association (X 1 2 ) tests (the -assoc option) were performed for each of the post-  Length of the open reading frame (ORF). 2 The number of the amino acid coded by the chicken GARNL1 gene. 3 The type of alternative splicing observed.

Gene-centric GWAS
Ninety-six NDH chickens were selected on the basis of AFE and EN300 measurements and genotyped for 384 SNPs using the Illumina GoldenGate TM iSelect Array genotyping platform (BGI, Shenzhen, China) via a commercial service. The 384 SNPs were located in an AFE QTL region or in 20 novel candidate genes selected according to previous reports [18,[22][23][24][25][26]35,36,38,51,52,63] (Table S6). The DNA sample set included 3 replication pairs. Twenty-four early sexual mature individuals with an AFE of 91 to 95 d and 24 late sexual mature individuals with an AFE of 160 to 179 d, and 24 low production individuals with an EN300 of 1 to 22 eggs and 24 high production individuals with an EN300 of 140 to 163 eggs were used in the gene-centric genome-wide association study. The average AFE value in the early group was 92.5 d, whereas in the late group it was 166.4 d. The average EN300 value in the low group was 28 eggs, and in the high group, 146 eggs. The gene-centric GWAS association analysis was performed by quantitative trait association analysis using PLINK with 10000 permutations.

Marker-trait association analysis in NDH female population
A total of 5 site-specific primers were designed by Assay design and synthesis by the Sangon Biotech Company (Shanghai, China), a commercial service. We validated the SNPs for which associations were found using the Sequenom genotyping platform. The effects of an additional 12 SNPs on EN300 or AFE were investigated by an association study using the PCR-RFLP method in the NDH population (P#5 to P#16, Table S5). PCR was performed in a 10 mL reaction mixture containing 1 mL of Taq polymerase (Dongsheng Co., Guangzhou, China), 5 mL of the 26PCR buffer supplied by the manufacturer, 1 mM of each primer, 50 ng genomic DNA, with ddH 2 O added to a total volume of 10 mL. The PCR program used was 3 min at 94uC, followed by 32 cycles of 30 s at 94uC, 30 s at 59uC, 45 s at 72uC, and a final extension of 5 min at 72uC in a Bio-Rad Mycycles (Bio-Rad Laboratories, Hercules, CA, USA). PCR products were digested in a 37uC or 65uC water bath overnight with MspI, HindIII (2 sites), DraI (2 sites), SacII, KpnI, StuI, PvuII, TaqI, Csp6I, and BsuRI. Genotypes were determined by electrophoresis after restriction digestion.
PLINK (version 1.07) was used to perform analysis for evaluating genetic effect of 17 sites of GARNL1 gene on EN300 and AFE. Sliding-window method for Locally Inter-correlated markers with asymptotic Distribution Errors corrected (SLIDE) program [64] was used to correct the P value, and all the pointwise P value were corrected based on 10000 sampling. A P value #0.05 was considered significant in the analyses. Multiple comparisons were conducted with least squares means using Fisher's least significant difference method.
The Hardy-Weinberg equilibrium and haplotype structure were analyzed using Haploview 4.1 software [65]. Haplotypes were constructed on the basis of genotype data using PHASE 2.0 software (http://en.wikipedia.org/wiki/Phase).The minimum haplotype frequency was set to 1%. Association analyses of haplotypes with the EN300 and AFE were carried out using SAS GLM procedure (SAS Institute Inc., Cary, NC, USA) with the following model

Y~mzHzSze,
Where Y is a trait observation, m is the overall population mean, H is the effect of haplotype, S is the fixed effect of sire, and e is the residual error.
Allelic frequency spectrum of the chicken GARNL1 gene Five populations, Red Jungle fowl (RJF) (n = 33), Baier Huang chicken (BEH) (n = 41), White Leghorn (LH) (n = 60), Xinghua chicken (XH) (n = 50), and Ningdu Sanhuang chicken (NDH) (n = 82), were genotyped at the reproduction-associated SNPs of the GARNL1 gene to obtain allele frequency spectra. RJF, XH and NDH showed low egg-production ability, with 60,130 eggs per year because of their intractable incubation behavior (100% incidence of broodiness in RJF, 70,80% incidence of broodiness in XH and 50,60% incidence of broodiness in NDH). BEH chickens had a 10,15% incidence of broodiness and an eggproduction of 180 per year. The Leghorn chicken is a famous layer breed with excellent egg-production ability and no incubation behavior. The primers used were the same as those used for genotyping the NDH population.
Chi-square (x 2 ) tests performed on a 263 (or n) contingency table. A P#0.05 was considered significant in all analyses.

Analysis of GARNL1 alternative splicing isoforms
Total protein was collected using Trizol reagent (Invitrogen, California, USA) according to the manufacturer's protocol. Proteins were separated on an 8% SDS-PAGE gel by electrophoresis and transferred to PVDF membranes (Millipore, Billerica, MA, USA). Membranes were blocked using TBST containing 5% nonfat milk and 0.05% (w/v) Tween 20 overnight at room temperature. Membranes were then incubated overnight at 4uC with rabbit polyclonal anti-GARNL1 (TULIP1) in a 1:1000 dilution (Santa Cruz Biotechnology, CA, USA). Membranes were washed three times in TBST containing 0.05% Tween 20 and incubated in HRP-conjugated secondary antibody for 1 h at 37uC. Membranes were washed as before and signals were detected using Super ECL Detection Reagent (Applygen, Beijing, China) and Kodak (Kodak Film, USA).
Using cDNA transcribed from RNA pool composed of 6 chicken tissues as a template, the GARNL1 ORF was amplified by primer pairs (P#4 , Table S5) on the basis of the predicted mRNA sequence (accession number: XM421244). PCR was performed in a 50 mL reaction mixture containing 1 mL of KOD FX polymerase (Toyobo, Osaka, Japan), 25 mL of the 26PCR buffer supplied by the manufacturer, and 10 mL of 2 mM dNTPs. The PCR conditions were 4 min at 94uC, followed by 30 cycles of 10 s at 98uC, 7 min at 68uC, and a final extension of 10 min at 68uC in a Bio-Rad S1000 (Bio-Rad Laboratories, Hercules, CA, USA). The amplified fragments were cloned into a pGEM-T Easy plasmid vector (Promega, Madison, USA) and then sequenced by Invitrogen Co. Ltd (Shanghai, China) via a commercial service. The obtained sequences were assembled to obtain the full-length GARNL1 cDNA sequence. After sequencing, some novel alterna- tive splice forms of the chicken GANRL1 gene were detected by Reverse Transcription-PCR.
Sequence databases were accessed and searched using the BLAST algorithm at the NCBI at http://www.ncbi.nlm.nih.gov/ BLAST and the UCSC Chicken Genome Project Working Draft at http://genome.ucsc.edu/. Protein domain predictions were obtained using the SMART program (Simple Modular Architecture Research Tool) from the European Molecular Biology Laboratory at http://smart.emblheidelberg.de/, and transmembrane helics motif and leucine zipper domain predictions were obtained using the SOSUI algorithm at http://bp.nuap.nagoya-u. ac.jp/sosui/sosui_submit.html and the PSORT II program at http://psort.ims.u-tokyo.ac.jp/form2.html. Multiple alignments of amino acid sequences were performed using ClustalW at http:// www.ebi.ac.uk/clustalw/ and DNAMAN software (Lynnon corporation, Quebec, Canada), with MEGA 4.1 software (http:// www.megasoftware.net/) used for homology analysis. Figure S1 qPCR results of ten putatively differentially expressed transcripts identified by SSH.