Introduction

SHOX located in the pseudoautosomal region 1 (PAR1) regulates skeletal development.1 SHOX haploinsufficiency underlies idiopathic short stature (ISS; MIM no. 300582), Leri–Weill dyschondrosteosis (LWD) characterized by Madelung deformity (MIM no. 249700) and less-specific skeletal changes including micrognathia.1, 2, 3, 4 SHOX transcripts include SHOXa encoded by exons 1–5 and 6a, and SHOXb encoded by exons 1–5 and 6b.1 SHOX expression requires multiple cis-acting elements in PAR1;5, 6, 7, 8, 9, 10, 11 previous studies have identified highly conserved non-coding DNA elements (CNEs) with cis-regulatory activity in the upstream (CNE-2, CNE-3 and CNE-5) and downstream regions (CNE4, CNE5, ECR1 and CNE9/ECS4) of SHOX (Supplementary Figure S1).7, 8, 10, 11

SHOX haploinsufficiency in cytogenetically normal individuals is mainly caused by microdeletions involving SHOX exons 1–6a and/or the CNEs.2, 3, 5, 6, 7, 8, 9 Recently, microduplications encompassing SHOX were also identified in patients with ISS and LWD.12, 13 However, the pathogenicity of these copy-number variations (CNVs) remains controversial, because SHOX overdosage has been associated with tall or normal stature.14, 15, 16, 17 Likewise, although a deletion in SHOX intron 6b was identified in an ISS family, its relevance to the disease phenotype remains unclear.18 Furthermore, the genomic origins of PAR1-linked CNVs have been poorly investigated.

Subjects and methods

Detailed methods are provided in the Supporting Information (Supplementary Methods). This study was approved by the Institutional Review Board Committee at the National Center for Child Health and Development and performed after obtaining informed consent. We searched for PAR1-linked CNVs in 245 patients with ISS or LWD and in 15 unaffected individuals. All patients had short stature with s.d. scores of <−2.0. Diagnosis of LWD was based on radiological findings of Madelung deformity.3 Individuals with chronic diseases or apparent chromosomal alterations were excluded. When possible, we analyzed participants’ parental samples. Genomic DNA samples were subjected to multiplex-ligation-dependent probe amplification. All CNVs, except for well-documented microdeletions involving SHOX and/or downstream CNEs,2, 3, 9 were further characterized by array-based comparative genomic hybridization. We also investigated the genomic structures of CNVs.

Results

Rare CNVs were identified in patients 1–5 and control 1 (Figure 1,Table 1, and Supplementary Figure S2). Patients 1–3 carried microduplications involving SHOX and three or four of the seven CNEs. Patient 4 and control 1 had microduplications in the SHOX 3′-region encompassing one and all downstream CNEs, respectively. Patient 5 had a microdeletion involving exon 6b and CNE4. The CNV in patient 1 was a 270 737 bp tandem duplication, whose breakpoints resided within Alu repeats and shared a 12 bp overlap (Figure 2a). Fusion junctions of other CNVs could not be determined. Fluorescence in situ hybridization analysis of patient 2 using a SHOX-containing cosmid generated a signal only at Xp22.3 (Figure 2b), indicating that the amplified DNA fragment was probably inserted into a genomic interval adjacent to the original position. Patients 1–5 manifested mild or moderate short stature (Table 1). None of the five patients had Madelung deformity, whereas patient 3 showed micrognathia. The phenotypically normal father of control 1 had the same CNV as the proband.

Figure 1
figure 1

(a)Array-based comparative genomic hybridization. The upper horizontal line indicates the physical distance from the Xp/Yp telomere (pter; hg19, Build 37). The purple box depicts SHOX. Orange and blue boxes denote highly evolutionarily conserved non-coding DNA elements (CNEs) in the SHOX upstream and downstream regions, respectively. The black, red and green dots denote normal, increased (log ratio⩾+0.5) and decreased (log ratio⩽–1.0) copy-numbers, respectively. The red and green arrows indicate duplicated and deleted regions, respectively. cen, centromere. (b) Schematic representation of the SHOX-flanking region. The upper horizontal line indicates the physical distance from the Xp/Yp telomere. The red and green lines denote the duplicated and deleted regions, respectively.

Full size image
Table 1 Clinical and molecular findings of individuals with pseudoautosomal copy-number variations
Full size table
Figure 2
figure 2

Characterization of copy-number variations (CNVs) in patients 1 and 2. (a) Upper panel: schematic representation of the duplication in patient 1. The purple boxes depict SHOX. Orange and blue boxes denote highly conserved non-coding DNA elements (CNEs) in the SHOX upstream and downstream regions, respectively. The red shaded area indicates the duplicated region. The genomic structure is not drawn to scale. Lower panel: DNA sequences at the fusion junction of the duplication in patient 1. The PCR product containing the junction was amplified using a primer pair, 5′-CCTCCAAAATAGCTGGCAATA-3′ and 5′-AGCATAAAATTCCCCATCTGA-3′. The CNV was a tandem duplication of 270 737 bp (chrX:486,700–757,437; hg19, Build 37). (b) Fluorescence in situ hybridization analysis of patient 2. The SHOX-containing cosmid probe (green) generated signals only at Xp22.3 (arrows). The red signals (arrowheads) indicate the control probe for the Xq/Yq telomere.

Full size image

Discussion

This study provides further evidence that SHOX-containing microduplications account for a small fraction of the etiology of ISS. Our findings contradict prior observations that trisomy of PAR1 owing to chromosomal rearrangements leads to tall or normal stature.13, 14, 15, 16, 17 These conflicting data can be explained by assuming that relatively large duplications containing all SHOX exons and cis-acting enhancers lead to gene overexpression and resultant tall stature, whereas small duplications encompassing only a part of these components attenuate SHOX expression by disrupting the cis-regulatory machinery. Consistent with this, the microduplication in patient 1 increases the distance between SHOX exons and the CNEs, and the duplicated DNA fragment in patient 2 was detected at a genomic region adjacent to the original position. On the other hand, the results of patient 4 and control 1 question the pathogenicity of duplications affecting only the CNEs. To our knowledge, no previous study has shown that enhancer overdosage impairs gene expression. Nevertheless, PAR1 microduplications apart from SHOX may disrupt the cis-regulatory machinery. Indeed, CNE-containing microduplications were identified in multiple ISS patients.12, 19 Furthermore, the pathogenicity of the exon 6b-containing microdeletion in patient 5 remains unclear. SHOX exon 6b has not been implicated in skeletal development; however, elimination of SHOXb may affect the function of SHOXa isoform, because SHOXb dimerizes with SHOXa.20 Alternatively, the ISS of patient 5 may be ascribed to CNE4 deletion. Moreover, as a deletion in intron 6b has been identified in an ISS family,18 exon 6b-flainking regions may contain another enhancer.

Our data imply that the duplication in patient 1 arose from Alu-mediated non-allelic homologous recombination (NAHR). Because PAR1 is enriched with Alu repeats4 and Alu-mediated NAHR has been implicated in the development of a microdeletion involving SHOX,9 it is possible that NAHR underlies various PAR1-linked CNVs. However, other mechanisms such as non-homologous end-joining may also underlie these CNVs.

Patients 1–5 showed no signs of Madelung deformity, indicating that PAR1-linked duplications and exon 6b-containing deletions result in relatively mild phenotypes. However, the mild phenotypes in our patients may be related to their young ages, because skeletal features in patients with SHOX abnormalities usually ameliorate after puberty.3 Notably, patient 3 with a duplication encompassing SHOX exons 1–3 manifested micrognathia, which may be associated with SHOX dysfunction,2 whereas patients 1–2 with duplications of all SHOX exons manifested no skeletal deformities. These findings are consistent with previous findings that partial SHOX duplications exert more significant effects on skeletal development than complete duplications.12 Alternatively, partial SHOX duplications may have a broad phenotypic spectrum, because these CNVs were identified in several unaffected relatives of the patients.12

Collectively, our results indicate that SHOX-containing microduplications cause ISS by disrupting the cis-regulatory machinery of SHOX and that Alu-mediated NAHR underlies at least some of these microduplications. Further studies are necessary to clarify the pathogenicity of other rare PAR1-linked CNVs.