Recurrent Reciprocal Genomic Rearrangements of 17q12 Are Associated with Renal Disease, Diabetes, and Epilepsy

DNA Samples

DNA samples were obtained from prenatal autopsy specimens (n=155) and from individuals with renal disease and/or MODY5, with the appropriate institutional-review-board approval. Fetal liver tissue was obtained from Children’s Hospital and Regional Medical Center. All samples were from fetuses that underwent autopsy after elective termination or fetal demise between 1995 and 2006. DNA was extracted using Gentra PureGene DNA extraction kit. The renal pediatric and MODY5 cases represent patients who were analyzed previously for TCF2 molecular abnormalities.^10–12 Two control groups were used to assess the extent of normal copy-number variation. The first control group consisted of 316 unrelated individuals (HapMap and Diversity population) who had been tested using the same BAC duplication microarray.^8,9 A second control population, comprising 960 unrelated white adults (aged 40–70 years) from the United States, was genotyped using the HumanHap300 Genotyping BeadChips (Illumina), which consists of ∼317,000 HapMap SNPs spread throughout the genome. Each individual was enrolled in the Pharmacogenomics and Risk of Cardiovascular Disease (PARC) study, which aims to identify genetic contributors to the variable efficacy of statin drugs for cardiovascular disease (PharmGKB: PARC Profile). Hybridizations, data analysis, and copy-number analysis, with particular reference to chromosome 17q12 (137 probes within the critical region), were performed according to published protocols.¹³

Array-Based Comparative Genomic Hybridization (Array CGH)

DNA was hybridized to a custom BAC array consisting of 2,007 clones targeted to regions of the genome flanked by segmental duplications, as described elsewhere.⁹ This array includes all regions associated with known genomic disorders and an additional ∼105 regions with similar genomic architecture. Because of the targeted nature of this array, we will not detect rearrangements that are not mediated by segmental duplications. Regions were scored as copy-number variant if the log₂ ratio of two or more consecutive clones each exceeded twice the SD of the autosomal clones in dye-swap replicate experiments.⁷ We used a whole-genome tiling path array containing 32,433 BAC clones¹⁴ or a custom oligonucleotide array (NimbleGen Systems), consisting of 385,000 isothermal probes (length 45–75 bp) covering several chromosomal regions, including a 3-Mb region of chromosome 17q12 (55,888 probes; mean density 1 probe per 53 bp) to refine the breakpoints. Hybridizations were performed as described elsewhere,¹⁵ and a single unaffected male (GM15724 [Coriell]) was used as reference. Oligonucleotide array data are available at the Human Genome Structural Variation Project Web site.

Quantitative PCR

Oligonucleotides for real-time quantitative PCR assays were selected using Primer3 software.¹⁶ Primer sequences (5′→3′) for exon 1 primers are as follows: forward primer ATTTCCTGGTGCGAGTTTTG and reverse primer CAGGGGATGACTCCTGAAGA. PCRs were performed in an LC480 machine (Roche) by use of SYBR Green PCR Master Mix (Applied Biosystems). PCR conditions were 95°C for 5 min, followed by 40 cycles at 95°C for 15 s, 55°C for 20 s, and 72°C for 20 s. Reactions were performed in triplicate, with a final reaction volume of 10 μl containing 10 ng DNA, 1.6 μM primer, and 5 μl SYBR Green Master Mix. Control primers were placed in the albumin gene.

FISH

Metaphase spreads and nuclei were obtained from phytohemagglutinin-stimulated peripheral lymphocytes of individuals 6498 and 6840 by standard procedures. FISH experiments were performed using fosmid WIBR2-6022g13 (NCBI May 2004 assembly [hg17] coordinates chr17:32364308–32399464), directly labeled by nick-translation with Cy3-dCTP (Perkin-Elmer), as described elsewhere,¹⁷ with minor modifications. Fosmids WIBR2-3001A12 (hg17 coordinates chr17:32309158–32347025; not duplicated in either case) and WIBR2-946N02 (hg17 coordinates chr17:33156592–33197819; duplicated in both cases) were used in control experiments.

Fetal Anomalies and Genomic Rearrangements

We evaluated 155 fetuses with one or more congenital anomalies and no known cytogenetic abnormalities in this study (128 with normal karyotype, 12 of whom also had normal FISH results for 22q11 deletion, and 27 of whom had no karyotype information reported). To minimize false-positive results, we focused on microdeletions and microduplications defined by two or more adjacent BAC clones on our targeted array that were not observed as copy-number polymorphisms in >400 control individuals either on our BAC array platform^8,9 or in studies of full-tiling path microarrays.^18,19 We identified nine individuals (6%) with detectable deletions or duplications, eight of which may be pathogenic (table 1). Analysis using whole-genome tiling-path BAC arrays or high-density oligonucleotide arrays allowed refinement of the breakpoints. Three of the alterations are microdeletions with breakpoints in flanking segmental duplications: a 1.8-Mb deletion at 17q12 in a fetus with multicystic renal dysplasia (FA-275 in fig. 1), a 2.5-Mb deletion at 15q25.2 in a fetus with congenital diaphragmatic hernia and mild hydrocephalus (fig. 2b), and a 2-Mb deletion at 16p13.11 in a fetus with posthemorrhagic hydrocephalus, cleft lip, preauricular tags, and two cleft vertebrae (fig. 2c). In each case, the regions flanking the microdeletions are polymorphic in copy number (on the basis of an analysis of 270 unaffected individuals). The smallest rearrangement we detected was a 157-kb duplication (case FA-142 in fig. 2a).

MODY5: A Recurrent 17q12 Microdeletion Disorder

Case FA-275 represents a fetus with bilateral multicystic renal dysplasia, bilateral ureteropelvic junction stenosis, atretic right ureter, and hypoplastic bladder (fig. 1a). We detected a deletion of seven BACs spanning nearly 2 Mb of sequence on our targeted BAC array (fig. 1b). Fine mapping with a customized oligonucleotide array (average probe spacing 53 bp) showed that the deleted region spans 1.8 Mb and involves 19 known genes, including CCL3L1 (MIM 601395), CCL4L1 (MIM 603782), and TBC1D3 (MIM 607741), which are present in multiple copies (fig. 1c). Because of copy-number polymorphism in the flanking regions (figs. 3 and 4), it is difficult to predict the exact breakpoints of the deletion; however, both proximal and distal breakpoints are within segmental-duplication clusters with multiple regions of high identity—the most significant has >99% identity across 76 kb (hg17 coordinates chr17:31808172–31883983 and chr17:33529232–33604211).

The deleted region in case FA-275 includes the TCF2 gene (MIM 189907), mutations of which are known to cause MODY5,²⁰ as well as both pediatric^11,21 and prenatally detectable¹⁰ cystic renal disease. One third of a series of MODY5-affected individuals with TCF2 alterations have deletions of the entire TCF2 gene and surrounding sequence.¹² Similar deletions have been found in pediatric cystic renal disease.^10,11,21 Because the deletion in case FA-275 appeared to overlap the deletions reported in patients with MODY5, we used oligonucleotide arrays to refine the deletion breakpoints in five patients with pediatric renal disease without diabetes and three patients with MODY5 (diabetes and renal disease) who had previously been shown to have deletions encompassing the TCF2 gene (table 2).

Our results show that four of the five pediatric patients and all three of the patients with MODY5 have deletions that are nearly identical to that of our fetal case, with breakpoints in all cases mapping to flanking segmental-duplication blocks (fig. 1c). Case FR09 provides the greatest breakpoint resolution, with a proximal breakpoint at 31,835,000 ± 1 kb and a distal breakpoint at 33,357,000 ± 1 kb defining the common minimal deletion region. The extensive and variable copy-number polymorphism in the flanking regions (figs. 3 and 4) and the presence of a sequence assembly gap make it difficult to define the breakpoints more precisely in the other three cases, but they all occur within the large segmental duplication blocks (proximal: 31,500,000–31,900,000 bp; distal: 33,300,000–33,600,000 bp). One case (FR35) is atypical, with a larger de novo deletion at least 2.1 Mb in size. The distal breakpoint is in unique sequence ∼33 kb distal to the TCF2 gene (hg17 coordinate ∼33212500), suggesting this deletion arose from a mechanism other than NAHR. These results show that the 17q12 deletion is a recurrent genomic disorder with breakpoints in flanking segmental duplications that results in renal disease and diabetes.

Reciprocal 17q12 Duplication

We previously identified a duplication of the 17q12 region in a patient with idiopathic mental retardation (IMR379) and an affected sibling.⁷ We have identified two additional patients with mild-to-moderate mental retardation, epilepsy, and focal cortical dysplasia who have overlapping duplications on 17q12 by BAC array CGH (table 2). Oligonucleotide microarray CGH analysis of these three individuals shows that two individuals, IMR379 and 5812, have duplications of the entire 17q12 region, with breakpoints mapping to the same duplication blocks associated with renal disease and MODY5 deletions (fig. 5a). These duplication events, therefore, represent the expected reciprocal duplication to the common 17q12 deletion and provide further evidence that these rearrangements are mediated by NAHR.²² Pedigree analysis of these families shows that dup17q12 can be stably transmitted meiotically (fig. 6). The duplication is present in all affected individuals in family 1 who we were able to test. However, apparently unaffected individuals in the family were also found to carry the duplication, so the pathological significance is unclear. It is possible that the duplication is a benign copy-number variant segregating with disease in this family. Alternatively, there may be incomplete penetrance of the seizure phenotype.

To assess the frequency of the 17q12 duplication in the general population, we analyzed a larger control group for copy-number variation of this locus (n=960 unrelated white individuals who had been genotyped with a high-density SNP Genotyping BeadChip for variable efficacy of statin-drug response by cardiovascular disease [R. Krauss and D. Nickerson, personal communication]). We identified a single individual with a duplication corresponding to the 1.5-Mb region (fig. 7), whereas no individuals with the deletion were observed. Combined with our previous analysis of 270 HapMap individuals, this suggests a slight enrichment of the 17q12 duplication among patients with mental retardation and/or epilepsy (2 of 380 affected individuals; 1 of 1,276 control individuals) but does not reach statistical significance (P=.13, by Fisher’s exact test).

Interestingly, the third patient (6498), who has mental retardation, focal cortical dysplasia, and focal seizures, has a smaller and apparently more complex duplication in which there is a 27-kb duplication encompassing the LHX1 gene (MIM 601999; hg17 coordinates chr17:32360000–32387000) and a 259-kb duplication corresponding to the TCF2 gene (hg17 coordinates chr17:33122000–33381000) (fig. 5a). The mother of patient 6498 is unaffected and also carries a duplication of the TCF2 gene. However, both oligonucleotide array CGH and real-time quantitative PCR of the 27-kb region encompassing LHX1 suggested that the mother does not carry the LHX1 duplication (fig. 5). Interphase FISH with peripheral blood lymphocytes from both the mother and daughter showed that the mother is mosaic for the duplication (28% of 135 nuclei), whereas the daughter has the LHX1 duplication in virtually all cells (98% of 138 nuclei) (fig. 5c).