Identity-by-descent refines mapping of candidate regions for preaxial polydactyly II /III in a large Chinese pedigree

Preaxial polydactyly (PPD) is congenital hand malformation characterized by the duplication of digit. Herein, we scan the genome-wide SNPs for a large Chinese family with PPD-II/III. We employ the refined IBD algorithm to identify the identity-by-decent (IBD) segments and compare the frequency among the patients and normal relatives. A total of 72 markers of 0.01 percentile of the permutation are identified as the peak signals. Among of them, 57markers locate on chromosome 7q36 which is associated with PPD. Further analyses refine the mapping of candidate region in chromosome 7q36 into two 380 Kb fragments within LMBR1 and SHH respectively. IBD approach is a suitable method for mapping causal gene of human disease. Target-enrichment sequencing as well as functional experiments are required to illustrate the pathogenic mechanisms for PPD in the future.


Background
Preaxial polydactyly (PPD; OMIM#188740) is characterized as complete or partial duplication of the thumb [1]. It is one of the most common congenital deformities [2]. The worldwide incidence of PPD is 1 in 3000 births [3]. The prevalence rate of polydactyly in Chinese ranks third in birth defects after congenital heart diseases and central nervous system diseases [4]. Polydactyly has genetic and clinical heterogeneity [2]. The mainstream treatment is resection for excess digits.
A series of efforts have been performed to investigate the genetic basis for PPD. Zguricas et al. conducted linkage analysis for Dutch, British, Turkish, Cuban pedigrees and mapped the candidate region to a 1.9 cM interval between D7S550 and D7S2423 of 7q36 region [5]. Heus et al. further refined the candidate region to approximately 450 Kb including five genes: C7orf2 (i.e. LMBR1), C7orf3 (i.e. NOM1), C7orf4 (i.e. LINC00244), HLXB9 (i.e. MNX1) and RNF32 [6] by reconstructed a detailed physical map using a combination of exon trapping, cDNA selection, and EST mapping methods. Further evidence shows that PPD is caused by ectopic expression of SHH in mice, cats and humans [7]. The zone of polarizing activity regulatory sequence (ZRS), performs as the limbspecific cis-regulator, in controlling the expression of SHH. ZRS locates within intron 5 of the neighboring gene LMBR1, which is~1 Mb upstream from SHH [8]. In a number of cases, mutations of ZRS disturb the expression of SHH at the anterior limb bud margin and consequently caused PPD [8][9][10][11][12][13][14][15]. Homozygous deletion of ZRS can cause limb-specific absence of SHH expression in the acheiropodia [16]. It actually exists in the snake species and a limbless newt [17]. Duplication of ZRS results in Triphalangeal thumb-polysyndactyly syndrome (TPTPS; OMIM#174500), that is a subtype of PPD. It also can lead to syndactyly type IV (SD4; OMIM#186200) [18].
The common PPD only involves in hands/feet. In extreme and rare cases, PPD occur both in hands and feet. To investigate the genetic basis, Li et al. adopted a candidate gene approach to genotype nine microsatellite markers of 7q36 chromosomal region in a Chinese family with PPD both in hands and feet. By linkage analysis and haplotype construction, they located the linked region spanning 1.7 Mb between D7S2465 and D7D2423 [19]. It includes the 450 kb candidate region previously identified by Henus [6]. Nevertheless, the other parts of genome is not investigated yet. Herein, we genotyped genome-wide SNPs and employed the identity-by-descent (IBD) to refine the mapping of potential candidate loci for PPD in the same family.

Patients
This study has been approved by the internal review board of Kunming Institute of Zoology, Chinese Academy of Sciences (SMKX 2012013). The sixgeneration pedigree (including 21 patients and 24 normal relatives) involved in this study has been described previously in Li et al. [19]. All patients show hexadactyly of hands and feet. They have been diagnosed by physical examination & X-ray and assigned as isolated PPD-II on hand and isolated PPD-III on feet according to Temtamy and McKusick's classification [20]. PPD shows autosomal dominant inheritance in this pedigree.

SNP array
We genotyped 900,015 markers in 45 individual with HumanOmniZhongHua-8 BeadChip v1.0 (Illumina). We exported the chip data in accordance with the reference sequence GRCh37 into PLINK format via GenomeStudio (Illumina). The markers on mitochondrial DNA and sex chromosomes were disregarded. We adopted a series of quality control strategies [21] by using PLINK 1.9 [22]. Two individuals with call rate < 90% were removed. The SNPs with call rate < 90%, minor allele frequency < 1%, and deviation of Hardy-Weinberg equilibrium (P < 1e-6) were excluded. After filtering, a total of 595,534 autosomal SNPs for 43 individuals were utilized in subsequent

IBD detection
We used BEAGLE 4.0 [24] to phase and impute the genotype data referring to the pedigree information and the genetic map of HapMapII [25]. We detected the IBD segment with the refined IBD in BEAGLE 4.1 [26]. The IBD segment length shorter than 1 cM and the logarithm of odds (LOD) score under 3 were excluded before permutation [27]. The threshold of the genomewide significance was set to the 0.05 percentile of the distribution of the permutation p-value.

Results
The length distribution of detected IBD segments approximates a Pareto distribution (Additional file 1: Figure S1). The permutation result shows the significant segments distributing widely across genomes (Fig. 1). When considering the top 0.01% outliers of signals, we find the peak signals of 72 SNPs, of which 57 markers locate at 7q36 chromosomal region (Additional file 2: Table S1). We map the markers into the IBD fragments including LMBR1 and SHH ( Table 1). The minimal IBD segments within LMBR1 and SHH are around 380 Kb, respectively (Additional file 3: Table S2). The IBD segments are more frequently in patient-patient (ratio; percentage) than normal-normal (ratio; percentage) ( Table 2). We make annotation for the significant SNPs (Additional file 2: Table S1). All the SNPs are not haven't been reported to be associated with PPD before.

Discussion
Our IBD analyses refine the mapping of the candidate regions for PPD into two~380 Kb segments in 7q36 referring to LMBR1 and SHH genes, respectively (Additional file 3: Table S2). The segment for LMBR1 includes three genes (i.e. LMBR1, NOM1, and RNF32) and lies within the 450 kb candidate region identified before [6].Mutations in the ZRS is playing an important role in the pathogenesis of PPD (Additional file 4: Table S3). The duplication of ZRS can cause TPTPS and SD4 [18]. Its role in PPD-II /PPD-III is unclear. In the previous investigation of the same family, Li et al. detected no pathogenic mutation in ZRS as well as no duplication of ZRS [19]. Consequently, the etiology of this PPD family may be another limb-specific regulatory element of SHH gene exists in the noncoding region.
In addition to the segment of LMBR1, we also identified a segment of SHH. The SHH gene encodes sonic hedgehog, a secreted protein, which plays a key role in the limb development [28]. The ectopic expression of SHH in the anterior limb margin can cause PPD in human, in mouse [29], Hemingway cat [7] and chicken [30]. Recently, Petit et al. identified a 2 kb deletion occurring about 240 kb upstream from the SHH promoter in a large family with PPDhypertrichosis. They found the 2 kb deletion repress the transcriptional activity of the SHH promoter in vitro [31]. It raises a possibility that long range regulation may be an explanation for the PPD. Further target-enrichment sequencing and further functional experiments for LMBR1 and SHH are required to identify the pathogenic mutation(s).
In summary, we refine the mapping of the candidate regions for PPD based on high-density genomic SNPs. The potential candidate mutations are most likely to locate in LMBR1 and/or SHH gene. It is much improved compred with previous results [6,19]. Our study suggests that the IBD approach is a suitable method for mapping the causal genes of human diseases. Moreover, as disruptions of topological chromatin domains can result in limb malformations [32], more attention should be paid when studying PPD in the future on this aspect.

Additional files
Additional file 1: Figure S1. Plot of the distribution of the IBD segments.