Background Chromosome 17p13.3 contains extensive repetitive sequences and is a recognised region of genomic instability. Haploinsufficiency of PAFAH1B1 (encoding LIS1) causes either isolated lissencephaly sequence or Miller–Dieker syndrome, depending on the size of the deletion. More recently, both microdeletions and microduplications mapping to the Miller–Dieker syndrome telomeric critical region have been identified and associated with distinct but overlapping phenotypes.
Methods Genome-wide microarray screening was performed on 7678 patients referred with unexplained learning difficulties and/or autism, with or without other congenital abnormalities. Eight and five unrelated individuals, respectively, were identified with microdeletions and microduplications in 17p13.3.
Results Comparisons with six previously reported microdeletion cases identified a 258 kb critical region, encompassing six genes including CRK (encoding Crk) and YWHAE (encoding 14-3-3ε). Clinical features included growth retardation, facial dysmorphism and developmental delay. Notably, one individual with only subtle facial features and an interstitial deletion involving CRK but not YWHAE suggested that a genomic region spanning 109 kb, encompassing two genes (TUSC5 and YWHAE), is responsible for the main facial dysmorphism phenotype. Only the microduplication phenotype included autism. The microduplication minimal region of overlap for the new and previously reported cases spans 72 kb encompassing a single gene, YWHAE. These genomic rearrangements were not associated with low-copy repeats and are probably due to diverse molecular mechanisms.
Conclusions The authors further characterise the 17p13.3 microdeletion and microduplication phenotypic spectrum and describe a smaller critical genomic region allowing identification of candidate genes for the distinctive facial dysmorphism (microdeletions) and autism (microduplications) manifestations.
- clinical genetics
- molecular genetics
Statistics from Altmetric.com
The advent of microarray analysis for the detection of chromosomal abnormalities in patients with developmental problems has heralded a new era of cytogenetic testing, which in turn is impacting significantly on clinical practice. It has been shown that pathogenic, submicroscopic copy-number alterations are detectable in 10–15% of patients with idiopathic mental retardation and/or multiple congenital abnormalities.1–5 A significant outcome of this testing has been the identification of several ‘new’ microdeletion/duplication syndromes through so-called ‘reverse phenotypics’—that is, using a genotype-to-phenotype approach.6–11
Novel co-locating microdeletions and microduplications in chromosome 17p13.3 distinct from isolated lissencephaly sequence (ILS) and Miller–Dieker syndrome (MDS) have recently been described in six and 10 unrelated individuals, respectively.12–15 Prior to these, Cardoso et al16 described two cases of derivative chromosomes involving similar deletions in 17p13.3 but with concurrent large duplications on other chromosomes. Here we report an additional eight individuals with the microdeletion and five with the microduplication. None of these microdeletions involve the PAFAH1B1 gene, encoding LIS1, whose haploinsufficiency causes either ILS (OMIM 607432) or MDS (OMIM 247200), depending on the size of the deletion. The thirteen 17p13.3 copy-number variations (CNVs) were detected by genome-wide microarray screening of a cohort of 7678 patients referred to six independent clinical centres for genetic testing. All the patients tested had unexplained learning difficulties and/or autism, with or without other congenital abnormalities.
Clinical samples (n=7678) were obtained from: Melbourne, Australia (n=349); Stockholm, Sweden (n=1289); Groningen, The Netherlands (n=2107); Antwerp, Belgium (n=100); Atlanta, USA (n=2000); and Nijmegen, The Netherlands (n=1833). We used 1171 parental samples as healthy controls to assess the extent of normal (non-pathogenic) CNV in this region of 17p13.3.
The control samples were tested using the 105K Agilent array (n=631), the 250K Nsp (n=240) or the 6.0 Affymetrix array (n=300). Detection of CNVs, pathogenic and non-pathogenic, in the patient samples used the following array platforms: 250K Nsp (n=1650) and 6.0 (n=81) array (Affymetrix, Santa Clara, California, USA); 370K (n=100) and HumanCytoSNP-12 300K (n=65) array (Illumina, San Diego, California, USA); 33K (n=30) and 38K (n=79) BAC array (Swegene, Lund, Sweden)17; 32K (n=386) BAC array (Microarray Facility, Nijmegen, the Netherlands), 244K catalogue oligonucleotide array (n=858), 180K catalogue oligonucleotide array (n=322) and 105K custom-designed oligonucleotide array (n=1292) (design ID 019015; Agilent Technologies, Santa Clara, California, USA); subtelomere BAC array containing 500 clones18 (n=345), 1Mb BAC array containing 6465 BAC clones (including two BACs covering YWHAE and one BAC for CRK)19 (n=470), and a custom-designed 44K oligonucleotide array20 (n=2000).
Independent confirmation of deletions/duplications in the 17p13.3 region was performed by multiplex ligation-dependent probe amplification (MLPA), fluorescence in situ hybridisation (FISH) and/or array comparative genomic hybridisation (aCGH) according to previously described methods. Rearrangements of 17p13.3 were further analysed with the use of a custom oligonucleotide array (see High-resolution breakpoint mapping below).
Parental analyses were performed to investigate whether the finding (gain/loss) was de novo or inherited. This was performed by FISH, MLPA or microarray analysis. Six sequence-tagged site (STS) markers spanning the maximum deleted/duplicated segment of 17p13.3 were typed to determine the parental origin of de novo rearrangements (where possible). Primer details are available upon request. Amplification was carried out with a 5′-FAM(carboxyfluorescein)-labelled primer, and PCR products were separated by capillary electrophoresis using a MegaBACE sequencer (GE Healthcare, Milwaukee, Wisconsin, USA). The results were analysed with Fragment Profiler V1.2 (GE Healthcare).
High-resolution breakpoint mapping
The breakpoint regions were further refined in 11 of the 12 cases, in 10 cases using a custom high-density CGH array (ie, CGH 2105K format (Agilent Technologies) (cases 1, 3–6 and 8–12), and in one case (case 2) by MLPA analysis (data not shown). The custom array consisted of 24 469 60-mer catalogue probes spanning the distal 3.4 Mb of 17pter and a further 23 235 tiling oligonucleotide probes targeting the ‘apparent’ breakpoint regions. The effective resolution was 60–1000 bp across this 3.4 Mb region. The custom array was designed using Agilent's web portal eArray (https://earray.chem.agilent.com/earray/) with a very high density of probes in the first 3 352 600 bp of 17 p. Firstly, we created a probe set with probes selected from the HD CGH Database within eArray (total of 24 469 probes). We then designed custom-made probes to tile across the previously defined breakpoint regions filtering out repeat-masked regions (23 235 probes). The two probe sets (47 704 probes) were combined into an array with an average density of one probe every 43 bp in the breakpoint regions and one probe every 130 bp in the intervening regions. Genome background coverage was achieved by adding the entire Agilent 44 K Whole Human Genome CGH probe set. Hybridisations were performed as previously described21 with minor modifications. Briefly, patient DNA and sex-matched controls (Promega, Madison, Wisconsin, USA) were double-digested with RsaI and AluI (Promega). One microgram of each digested sample was labelled by random priming (Enzo Life Sciences, New York, USA) with Cy3-dUTP or Cy5-dUTP, and labelled products were column purified with the Qiaquick PCR purification kit (Qiagen AB, Solna, Sweden). After probe denaturation and pre-annealing with 50 μg Cot-1 DNA (Invitrogen, Carlsbad, California, USA), hybridisation was performed at 65°C with rotation for 40 h. The array was analysed with the Agilent scanner and Feature Extraction software (V10.2). A graphical overview was obtained using the DNA analytics software (V4.0). Genomic start and stop positions of the deletions and duplications were determined by visual inspection of the numerical normalised log2 ratio values in the table view of the DNA analytics software package. The genomic architecture at the breakpoints was assessed by referring to the Human Genome Build 36.1 reference sequence (University of California Santa Cruz (UCSC) genome browser), and shared sequences between the telomeric and centromeric breakpoints were identified using BLAST2 (http://www.ncbi.nlm.nih.gov/BLAST/).
Junction fragment analysis
Long-range PCR primers were designed from the high-resolution custom-aCGH coordinates of CNV breakpoints for cases 1, 9 and 12. PCR products spanning the junctions were obtained and sequenced using the ET SEQ Mix (Applied Biosystems, Scoresby, Victoria, Australia) on a MegaBACE 1000 (GE Healthcare). Sequences were aligned using Mutation Surveyor (GE Healthcare). Junction sequences were further analysed using BLAST2 and the Human Genome Build 36.1 reference sequence in order to identify shared sequences between the telomeric and centromeric breakpoint sequences.
In total, 13 co-locating CNVs were detected in chromosome 17 band p13.3. This included seven simple microdeletions (cases 1–7), which are genomically distinct from the known microdeletions that cause ILS and MDS, five interstitial microduplications (cases 9–13) and one case of a complex rearrangement involving a concurrent deletion–duplication (case 8).
The smallest deletion (case 3) was 328 kb in size, containing only six genes (TUSC5, YWHAE, CRK, MYO1C, SKIP and exons 1–4 of PITPNA) (figure 1a). All the deletions, except in case 6, involved the YWHAE gene. The complex rearrangement (case 8) showed a terminal deletion extending to SMG6 followed by a tandem duplication extending 777 kb beyond PAFAH1B1. The breakpoints were scattered across the distal 3.4 Mb of 17p13.3 and all were unique (figure 1 and table 1). With the exception of two cases (cases 8 and 10), the PAFAH1B1 gene was not involved. The smallest duplication (case 13) was 59–88 kb in size. It contained the entire TUSC5 gene, but did not involve YWHAE or CRK.
Parental studies using FISH, MLPA or microarray analysis showed that eight of the genomic imbalances (cases 1–5 and 8–10) were de novo and three were inherited (cases 6, 12 and 13). In one case, the parents were deceased and no material was available for testing (case 7), and in another case the father was unavailable for testing (case 11). The deletion in case 6a was present in a similarly affected brother (case 6b) and their less severely affected mother (case 6c). Analysis of grandparental samples revealed that the deletion had occurred de novo in the mother. Both brothers had mild developmental delay and pronounced postnatal growth retardation, while their mother had growth retardation but normal cognition. Notably, these individuals (cases 6a–c) showed only discrete facial dysmorphism. The patient in case 12 was more severely affected than the other duplication cases; she had a global delay with severe mental retardation and autism (tables 2 and 3). The duplication in case 12 was inherited from her healthy mother and was also present in her healthy sister. Therefore this duplication may represent an incidental finding that is either not causal or only partly contributes to her phenotype. Similarly, the patient in case 13 was more severely affected than the other duplication cases. This duplication involved a single gene, TUSC5, and was inherited from the patient's healthy mother. Parent-of-origin analysis using six polymorphic STS markers, located in the terminal 3 Mb of 17p13.3, was performed for seven of the cases (table 4). Of these, four were of paternal origin (cases 1, 2, 5 and 10) and two of maternal origin (cases 8 and 9). The genotyping results for case 11, for which a paternal sample was unavailable, showed that the duplication was not maternal in origin.
We are able to review and refine the molecular and clinical features of the recently described novel microdeletions13 15 and microduplications12 14 in chromosome 17p13.3 by describing eight and six new cases (including one sib pair), respectively. Comparing all 14 deletions, the delineated critical region spans approximately 258 kb (chr17: 1 136 270–1 394 633) and includes six genes (exons 2-3 of TUSC5, YWHAE, CRK, MYO1C, SKIP and exons 1–4 of PITPNA) (figure 1a). Owing to its function in the central nervous system,12 22 YWHAE is very likely to play a role in the phenotypes of the deletion patients. CRK is the likely candidate for growth restriction; notably, case 1 described by Sreenath Nagamani et al15 and the single case described by Mignon-Ravix et al13 showed neither growth restriction nor deletion of CRK (figure 1a and tables 2, 3 and 5). Benign CNVs within the deletion minimal overlapping region (MRO) have not been reported in a series of 891 healthy individuals (see Methods) nor in the Database of Genomic Variants (http://projects.tcag.ca/variation/) or the CHOP CNV Database (http://cnv.chop.edu/).
We describe a recognisable but variable phenotype in eight individuals with microdeletions (table 5 and figure 3a) in 17p13.3 consistent with that recently described by Sreenath Nagamani et al.15 All individuals with 17p13.3 microdeletions had developmental delay of varying degree, except the patient in case 5, who had only mild learning difficulties, and the mother (6c) in cases 6a and 6b who, however, experienced transient speech problems as a young child. None displayed behavioural problems. Postnatal growth retardation varied from mild to pronounced and was present in all cases except case 5. Interestingly, two of the individuals (cases 1 and 4) were treated with growth hormone, and both showed a good to excellent catch-up response. Some facial features were shared in individuals with 17p13.3 microdeletion, albeit very discrete in family 6 (cases 6a–c) (figure 3). The most common facial features were laterally extended eyebrows, infraorbital folds, broad nasal tip, maxillary prominence and prominent upper and/or lower lip. The features lacking in family 6 were high/prominent forehead, maxillary prominence and micro/retrognathia, suggesting this aspect of the phenotype maps to the region (chr17: 1 136 270–1 245 560) spanning 109 kb.
This genomic segment includes the two genes TUSC5 and YWHAE (figure 1a, aqua box), further stressing the predominant role of YWHAE in the phenotype. Brain MRI performed in five individuals (table 5 and figure 3) showed no evidence of lissencephaly, but rather identified mild structural anomalies in the white matter (wide perivascular spaces, white matter hyper intensities), in four of them. In one of these (case 2), the frontal and upper parietal lobes were predominantly affected. Subtle hand/foot abnormalities were observed in six individuals (this study) with a microdeletion (table 5). Hence, the main characteristics of the microdeletion syndrome are significant postnatal growth retardation, mild to moderate mental retardation and facial dysmorphic manifestations.
We suggest that there are two classes of co-locating microduplications in 17p13.3. Class I duplications (six cases) involve YWHAE (encoding 14-3-3ε), but notably not PAFAH1B1.12 Class II duplications (seven cases) always involve PAFAH1B1 and may also include the genomic region encompassing the CRK and YWHAE genes.12 14 Class I show autistic manifestations and other behavioural symptoms, speech and motor delay, subtle dysmorphic facial features, subtle hand/foot malformations, and a tendency to postnatal overgrowth (table 2). Class II microduplications have recently been shown to be associated with moderate to mild developmental and psychomotor delay and hypotonia. Some dysmorphic features, such as prominent forehead and pointed chin, are shared with the class I duplications, while overgrowth, behavioural problems and hand/foot abnormalities are less often noted (table 3). Notably, individuals with duplication of PAFAH1B1 but not YWHAE or CRK (cases 5 and 6 from Bi et al12) show microcephaly and severe growth restriction, but are not particularly dysmorphic.
Collating the data for all class I microduplications, including cases 12 and 13, which were considered non-pathogenic, we identify a minimal region of overlap spanning 72 kb (chr17: 1 182 563–1 255 000). This region contains a single gene, YWHAE (figure 1b). All class II microduplications described include PAFAH1B, and five of the seven cases also include the class I minimal region of overlap.
All three cases with a ‘pathogenic’ duplication (tables 2 and 3) displayed different neurobehavioural symptoms. Two patients had autism with normal cognitive development, whereas the third patient had mild developmental delay. All had motor delay, while two patients also showed speech delay. Prenatal and postnatal growth was normal in two (cases 9 and 10), while the patient in case 11 had overgrowth. Dysmorphic manifestations were subtle and included a full tip of the nose, prominent cupid bow and pointed chin. Finally, subtle foot abnormalities were present in all three individuals. The phenotypes of individuals with 17p13.3 microduplications, including those recently reported by Bi et al12 and Roos et al,14 are summarised in tables 2 and 3. In conclusion, the main phenotypic characteristics of the patients with 17p13.3 microduplication are autistic manifestations, behavioural symptoms, speech delay, subtle dysmorphic facial manifestations, and subtle hand/foot malformations.
All thirteen 17p13.3 microdeletions/duplications were non-recurrent, with all of the breakpoints distinct from each other. Six of the 11 rearrangements that were characterised by high-resolution aCGH/MLPA (cases 3, 5, 6, 8, 10 and 12) had one of the breakpoints that mapped to known Alu elements, while the other was located within a unique sequence with no repetitive elements (figure 2 and table 6). Alu elements, part of the short interspersed nucleotide elements family of transposable elements, have a well-established role in both benign and pathogenic CNV formation.3 23–27 The centromeric breakpoint in case 2, with a terminal deletion, was also located within a unique sequence. In three cases (cases 1, 4 and 9), the centromeric and telomeric breakpoints were located within repetitive elements, and comparison of the breakpoint sequences revealed significant homology with the primary sequences of overlap mapping to Alu elements that showed identical genomic orientations (table 6). These data suggest that non-allelic homologous recombination is the likely mechanism responsible for the microdeletions (cases 1 and 4) and microduplication (case 9) in these individuals. Within the limits of high-density oligonucleotide aCGH analysis, no homology was observed between the breakpoints in cases 2, 3, 5, 6, 8, 10, 11 and 12, which is consistent with non-homologous end joining or Fork Stalling and Template Switching.25
Thus, high-density aCGH/MLPA analysis precisely located the breakpoint regions to either Alu elements (n=12) or unique sequences (n=11). Sequence analysis of the breakpoint junctions was carried out for three cases (cases 1, 9 and 12). Junction fragments were obtained for all three (data not shown). However, sequencing of deletion/duplication junctions was possible for only one of these (case 1). The inability to sequence the other junction fragments is most likely due to the complexity of the genomic sequence at these breakpoints (table 6). Interestingly, rearrangements within 17p13.3, such as those causing the co-locating duplication syndromes, ILS and MDS, have breakpoints that are not associated with the typical paired low-copy repeats.12 28 Instead, these non-recurrent rearrangements, including the 11 characterised here, appear to result from diverse molecular mechanisms (ie, non-allelic homologous recombination, non-homologous end joining or Fork Stalling and Template Switching).
Despite the close proximity (1.1 Mb) of these novel microdeletions to PAFAH1B1, the absence of lissencephaly implies that PAFAH1B1 expression is not affected. This is notable in the light of reports of disturbance of gene expression within 2–6 Mb of a genomic copy number abnormality.29 Although, cerebral MRI did not identify abnormal gyral patterns, mild structural changes with a global distribution affecting white matter were observed, resulting in wide perivascular spaces in our deletion patients. Hence, deletions of YWHAE and CRK may affect white matter and myelinisation. This is in line with the notion that the combined deletion of YWHAE, CRK and PAFAH1B1 results in a more severe brain phenotype than the deletion of PAFAH1B1 alone, and the fact that both genes are involved in neuronal migration.22 In addition, CRK has been shown to control neuronal positioning in the developing brain, both dependent on the Reelin pathway and independently of it: Crk knockout mice are smaller and have smaller brains.30 CRK is also an attractive candidate for the subtle malformations of the hands and/or feet observed in six of the eight individuals with a microdeletion, and six of the seven individuals with co-locating microduplications (tables 2, 3 and 5). CRK is known to interact with FLNA, a gene involved in limb development.31 Notably, case 3 (in Bi et al12) had no duplication of CRK and no limb abnormalities.
A single gene (YWHAE) maps within the duplication MRO (figure 1b). Duplication of YWHAE might have an effect on neuronal network development and maturation, as all seven cases with duplications involving this gene show delay in motor function and symptoms within the autism spectrum or developmental delay (table 2). This gene is thus likely to contribute to the neurodevelopmental symptoms in these individuals, although it has not previously been identified as a candidate gene for autism.32–34 As the duplications in cases 12 and 13 are less likely to have a major phenotypic effect in these patients, it also indicates that the genes proximal to the duplication MRO are less likely to be involved in the duplication phenotype.
Interestingly, duplication of PAFAH1B1 with concomitant duplication (case 10) or deletion (case 8) of YWHAE and CRK appears to result in a milder neurological phenotype than duplication of PAFAH1B1 alone. These findings provide further evidence to support genetic interactions between CRK, YWHAE and PAFAH1B1 in key molecular pathways controlling neuronal migration and cortical development.22 35
Our results refine the molecular and clinical description of patients carrying the recently described microdeletions and microduplications in 17p13.3. Review of all published cases indicates that deletions of YWHAE and CRK, not including PAFAH1B1, represent a clinically variable phenotype consisting of growth retardation, facial dysmorphism and developmental delay, but notably without lissencephaly. Autism and a tendency to overgrowth appear to be notable features of the co-locating class I microduplications, being less often present in the class II microduplication phenotype. YWHAE and CRK, in particular, have been identified as attractive candidate genes for autism (duplications) and facial dysmorphology manifestations (deletions), and the growth restriction (deletions) and limb malformations (deletions and duplications), respectively.
We are grateful to all individuals and parents who participated in this study. This work was supported by grants from The Swedish Research Council (BMA), The Karolinska Institute foundation and Stockholm County Council (to JS, BMA), Perpetual Trustees Australia (to HRS), and the EU-funded AnEUploidy Project (to BBAdV and BvB) and the Netherlands Organisation for Health Research and Development (to BBAdV). We also thank Z Bowman, M Lagerberg, J Wincent and C Ngo for technical assistance and J Senior for critically reading the manuscript.
DLB and B-MA contributed equally to this work.
Funding Swedish Research Council, Karolinska Institute Foundation, Stockholm County Council, Perpetual Trustees Australia, EU-funded AnEUploidy Project, The Netherlands Organisation for Health Research and Development.
Competing interests None.
Patient consent Obtained.
Ethics approval This study was conducted with the approval of the Karolinska Institute, Stockholm, Sweden.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.