A Role for Non-B DNA Forming Sequences in Mediating Microlesions Causing Human Inherited Disease.

Missense/nonsense mutations and microdeletions/microinsertions (<21 bp) represent ∼ 76% of all mutations causing human inherited disease, and their occurrence has been associated with sequence motifs (direct, inverted, and mirror repeats; G-quartets) capable of adopting non-B DNA structures. We found that a significant proportion (∼ 21%) of both microdeletions and microinsertions occur within direct repeats, and are explicable by slipped misalignment. A novel mutational mechanism, DNA triplex formation followed by DNA repair, may explain ∼ 5% of microdeletions and microinsertions at mirror repeats. Further, G-quartets, direct, and inverted repeats also appear to play a prominent role in mediating missense mutations, whereas only direct and inverted repeats mediate nonsense mutations. We suggest a mutational mechanism involving slipped strand mispairing, slipped structure formation, and DNA repair, to explain ∼ 15% of missense and ∼ 12% of nonsense mutations yielding perfect direct repeats from imperfect repeats, or the extension of existing direct repeats. Similar proportions of missense and nonsense mutations were explicable by hairpin/loop formation and DNA repair, yielding perfect inverted repeats from imperfect repeats. We also propose a model for single base-pair substitution based on one-electron oxidation reactions at G-quadruplex DNA. Overall, the proposed mechanisms provide support for a role for non-B DNA structures in human gene mutagenesis.

Remotely acting SMCHD1 gene regulatory elements: in silico prediction and identification of potential regulatory variants in patients with FSHD.

BACKGROUND: Facioscapulohumeral dystrophy (FSHD) is commonly associated with contraction of the D4Z4 macro-satellite repeat on chromosome 4q35 (FSHD1) or mutations in the SMCHD1 gene (FSHD2). Recent studies have shown that the clinical manifestation of FSHD1 can be modified by mutations in the SMCHD1 gene within a given family. The absence of either D4Z4 contraction or SMCHD1 mutations in a small cohort of patients suggests that the disease could also be due to disruption of gene regulation. In this study, we postulated that mutations responsible for exerting a modifier effect on FSHD might reside within remotely acting regulatory elements that have the potential to interact at a distance with their cognate gene promoter via chromatin looping. To explore this postulate, genome-wide Hi-C data were used to identify genomic fragments displaying the strongest interaction with the SMCHD1 gene. These fragments were then narrowed down to shorter regions using ENCODE and FANTOM data on transcription factor binding sites and epigenetic marks characteristic of promoters, enhancers and silencers. RESULTS: We identified two regions, located respectively ~14 and ~85 kb upstream of the SMCHD1 gene, which were then sequenced in 229 FSHD/FSHD-like patients (200 with D4Z4 repeat units <11). Three heterozygous sequence variants were found ~14 kb upstream of the SMCHD1 gene. One of these variants was found to be of potential functional significance based on DNA methylation analysis. Further functional ascertainment will be required in order to establish the clinical/functional significance of the variants found. CONCLUSIONS: In this study, we propose an improved approach to predict the possible locations of remotely acting regulatory elements that might influence the transcriptional regulation of their associated gene(s). It represents a new way to screen for disease-relevant mutations beyond the immediate vicinity of the specific disease gene. It promises to be useful for investigating disorders in which mutations could occur in remotely acting regulatory elements.

Complexity and robustness in hypernetwork models of metabolism.

Metabolic reaction data is commonly modelled using a complex network approach, whereby nodes represent the chemical species present within the organism of interest, and connections are formed between those nodes participating in the same chemical reaction. Unfortunately, such an approach provides an inadequate description of the metabolic process in general, as a typical chemical reaction will involve more than two nodes, thus risking oversimplification of the system of interest in a potentially significant way. In this paper, we employ a complex hypernetwork formalism to investigate the robustness of bacterial metabolic hypernetworks by extending the concept of a percolation process to hypernetworks. Importantly, this provides a novel method for determining the robustness of these systems and thus for quantifying their resilience to random attacks/errors. Moreover, we performed a site percolation analysis on a large cohort of bacterial metabolic networks and found that hypernetworks that evolved in more variable environments displayed increased levels of robustness and topological complexity.

Screening in silico predicted remotely acting NF1 gene regulatory elements for mutations in patients with neurofibromatosis type 1.

Neurofibromatosis type 1 (NF1), a neuroectodermal disorder, is caused by germline mutations in the NF1 gene. NF1 affects approximately 1/3,000 individuals worldwide, with about 50% of cases representing de novo mutations. Although the NF1 gene was identified in 1990, the underlying gene mutations still remain undetected in a small but obdurate minority of NF1 patients. We postulated that in these patients, hitherto undetected pathogenic mutations might occur in regulatory elements far upstream of the NF1 gene. In an attempt to identify such remotely acting regulatory elements, we reasoned that some of them might reside within DNA sequences that (1) have the potential to interact at distance with the NF1 gene and (2) lie within a histone H3K27ac-enriched region, a characteristic of active enhancers. Combining Hi-C data, obtained by means of the chromosome conformation capture technique, with data on the location and level of histone H3K27ac enrichment upstream of the NF1 gene, we predicted in silico the presence of two remotely acting regulatory regions, located, respectively, approximately 600 kb and approximately 42 kb upstream of the NF1 gene. These regions were then sequenced in 47 NF1 patients in whom no mutations had been found in either the NF1 or SPRED1 gene regions. Five patients were found to harbour DNA sequence variants in the distal H3K27ac-enriched region. Although these variants are of uncertain pathological significance and still remain to be functionally characterized, this approach promises to be of general utility for the detection of mutations underlying other inherited disorders that may be caused by mutations in remotely acting regulatory elements.

Gains of glycosylation comprise an unexpectedly large group of pathogenic mutations.

Mutations involving gains of glycosylation have been considered rare, and the pathogenic role of the new carbohydrate chains has never been formally established. We identified three children with mendelian susceptibility to mycobacterial disease who were homozygous with respect to a missense mutation in IFNGR2 creating a new N-glycosylation site in the IFNgammaR2 chain. The resulting additional carbohydrate moiety was both necessary and sufficient to abolish the cellular response to IFNgamma. We then searched the Human Gene Mutation Database for potential gain-of-N-glycosylation missense mutations; of 10,047 mutations in 577 genes encoding proteins trafficked through the secretory pathway, we identified 142 candidate mutations ( approximately 1.4%) in 77 genes ( approximately 13.3%). Six mutant proteins bore new N-linked carbohydrate moieties. Thus, an unexpectedly high proportion of mutations that cause human genetic disease might lead to the creation of new N-glycosylation sites. Their pathogenic effects may be a direct consequence of the addition of N-linked carbohydrate.

Genotype-phenotype associations in neurofibromatosis type 1 (NF1): an increased risk of tumor complications in patients with NF1 splice-site mutations?

Neurofibromatosis type 1 (NF1) is a complex neurocutaneous disorder with an increased susceptibility to develop both benign and malignant tumors but with a wide spectrum of inter and intrafamilial clinical variability. The establishment of genotype-phenotype associations in NF1 is potentially useful for targeted therapeutic intervention but has generally been unsuccessful, apart from small subsets of molecularly defined patients. The objective of this study was to evaluate the clinical phenotype associated with the specific types of NF1 mutation in a retrospectively recorded clinical dataset comprising 149 NF1 mutation-known individuals from unrelated families. Each patient was assessed for ten NF1-related clinical features, including the number of café-au-lait spots, cutaneous and subcutaneous neurofibromas and the presence/absence of intertriginous skin freckling, Lisch nodules, plexiform and spinal neurofibromas, optic gliomas, other neoplasms (in particular CNS gliomas, malignant peripheral nerve sheath tumors (MPNSTs), juvenile myelomonocytic leukemia, rhabdomyosarcoma, phaechromocytoma, gastrointestinal stromal tumors, juvenile xanthogranuloma, and lipoma) and evidence of learning difficulties. Gender and age at examination were also recorded. Patients were subcategorized according to their associated NF1 germ line mutations: frame shift deletions (52), splice-site mutations (23), nonsense mutations (36), missense mutations (32) and other types of mutation (6). A significant association was apparent between possession of a splice-site mutation and the presence of brain gliomas and MPNSTs (p = 0.006). If confirmed, these findings are likely to be clinically important since up to a third of NF1 patients harbor splice-site mutations. A significant influence of gender was also observed on the number of subcutaneous neurofibromas (females, p = 0.009) and preschool learning difficulties (females, p = 0.022).

Gene conversion: mechanisms, evolution and human disease.

Gene conversion, one of the two mechanisms of homologous recombination, involves the unidirectional transfer of genetic material from a 'donor' sequence to a highly homologous 'acceptor'. Considerable progress has been made in understanding the molecular mechanisms that underlie gene conversion, its formative role in human genome evolution and its implications for human inherited disease. Here we assess current thinking about how gene conversion occurs, explore the key part it has played in fashioning extant human genes, and carry out a meta-analysis of gene-conversion events that are known to have caused human genetic disease.

Towards In Silico Identification of Genes Contributing to Similarity of Patients' Multi-Omics Profiles: A Case Study of Acute Myeloid Leukemia.

We propose a computational framework for selecting biologically plausible genes identified by clustering of multi-omics data that reveal patients' similarity, thus giving researchers a more comprehensive view on any given disease. We employ spectral clustering of a similarity network created by fusion of three similarity networks, based on mRNA expression of immune genes, miRNA expression and DNA methylation data, using SNF_v2.1 software. For each cluster, we rank multi-omics features, ensuring the best separation between clusters, and select the top-ranked features that preserve clustering. To find genes targeted by DNA methylation and miRNAs found in the top-ranked features, we use chromosome-conformation capture data and miRNet2.0 software, respectively. To identify informative genes, these combined sets of target genes are analyzed in terms of their enrichment in somatic/germline mutations, GO biological processes/pathways terms and known sets of genes considered to be important in relation to a given disease, as recorded in the Molecular Signature Database from GSEA. The protein-protein interaction (PPI) networks were analyzed to identify genes that are hubs of PPI networks. We used data recorded in The Cancer Genome Atlas for patients with acute myeloid leukemia to demonstrate our approach, and discuss our findings in the context of results in the literature.

Identification of recurrent type-2 NF1 microdeletions reveals a mitotic nonallelic homologous recombination hotspot underlying a human genomic disorder.

Nonallelic homologous recombination (NAHR) is one of the major mechanisms underlying copy number variation in the human genome. Although several disease-associated meiotic NAHR breakpoints have been analyzed in great detail, hotspots for mitotic NAHR are not well characterized. Type-2 NF1 microdeletions, which are predominantly of postzygotic origin, constitute a highly informative model with which to investigate the features of mitotic NAHR. Here, a custom-designed MLPA- and PCR-based approach was used to identify 23 novel NAHR-mediated type-2 NF1 deletions. Breakpoint analysis of these 23 type-2 deletions, together with 17 NAHR-mediated type-2 deletions identified previously, revealed that the breakpoints are nonuniformly distributed within the paralogous SUZ12 and SUZ12P sequences. Further, the analysis of this large group of type-2 deletions revealed breakpoint recurrence within short segments (ranging in size from 57 to 253-bp) as well as the existence of a novel NAHR hotspot of 1.9-kb (termed PRS4). This hotspot harbored 20% (8/40) of the type-2 deletion breakpoints and contains the 253-bp recurrent breakpoint region BR6 in which four independent type-2 deletion breakpoints were identified. Our findings indicate that a combination of an open chromatin conformation and short non-B DNA-forming repeats may predispose to recurrent mitotic NAHR events between SUZ12 and its pseudogene.

Characterization of the nonallelic homologous recombination hotspot PRS3 associated with type-3 NF1 deletions.

Nonallelic homologous recombination (NAHR) is the major mechanism underlying recurrent genomic rearrangements, including the large deletions at 17q11.2 that cause neurofibromatosis type 1 (NF1). Here, we identify a novel NAHR hotspot, responsible for type-3 NF1 deletions that span 1.0 Mb. Breakpoint clustering within this 1-kb hotspot, termed PRS3, was noted in 10 of 11 known type-3 NF1 deletions. PRS3 is located within the LRRC37B pseudogene of the NF1-REPb and NF1-REPc low-copy repeats. In contrast to other previously characterized NAHR hotspots, PRS3 has not developed on a preexisting allelic homologous recombination hotspot. Furthermore, the variation pattern of PRS3 and its flanking regions is unusual since only NF1-REPc (and not NF1-REPb) is characterized by a high single nucleotide polymorphism (SNP) frequency, suggestive of unidirectional sequence transfer via nonallelic homologous gene conversion (NAHGC). By contrast, the previously described intense NAHR hotspots within the CMT1A-REPs, and the PRS1 and PRS2 hotspots underlying type-1 NF1 deletions, experience frequent bidirectional sequence transfer. PRS3 within NF1-REPc was also found to be involved in NAHGC with the LRRC37B gene, the progenitor locus of the LRRC37B-P duplicons, as indicated by the presence of shared SNPs between these loci. PRS3 therefore represents a weak (and probably evolutionarily rather young) NAHR hotspot with unique properties.

Non-B DNA-forming sequences and WRN deficiency independently increase the frequency of base substitution in human cells.

Although alternative DNA secondary structures (non-B DNA) can induce genomic rearrangements, their associated mutational spectra remain largely unknown. The helicase activity of WRN, which is absent in the human progeroid Werner syndrome, is thought to counteract this genomic instability. We determined non-B DNA-induced mutation frequencies and spectra in human U2OS osteosarcoma cells and assessed the role of WRN in isogenic knockdown (WRN-KD) cells using a supF gene mutation reporter system flanked by triplex- or Z-DNA-forming sequences. Although both non-B DNA and WRN-KD served to increase the mutation frequency, the increase afforded by WRN-KD was independent of DNA structure despite the fact that purified WRN helicase was found to resolve these structures in vitro. In U2OS cells, ∼70% of mutations comprised single-base substitutions, mostly at G·C base-pairs, with the remaining ∼30% being microdeletions. The number of mutations at G·C base-pairs in the context of NGNN/NNCN sequences correlated well with predicted free energies of base stacking and ionization potentials, suggesting a possible origin via oxidation reactions involving electron loss and subsequent electron transfer (hole migration) between neighboring bases. A set of ∼40,000 somatic mutations at G·C base pairs identified in a lung cancer genome exhibited similar correlations, implying that hole migration may also be involved. We conclude that alternative DNA conformations, WRN deficiency and lung tumorigenesis may all serve to increase the mutation rate by promoting, through diverse pathways, oxidation reactions that perturb the electron orbitals of neighboring bases. It follows that such "hole migration" is likely to play a much more widespread role in mutagenesis than previously anticipated.

A meta-analysis of single base-pair substitutions in translational termination codons ('nonstop' mutations) that cause human inherited disease.

'Nonstop' mutations are single base-pair substitutions that occur within translational termination (stop) codons and which can lead to the continued and inappropriate translation of the mRNA into the 3'-untranslated region. We have performed a meta-analysis of the 119 nonstop mutations (in 87 different genes) known to cause human inherited disease, examining the sequence context of the mutated stop codons and the average distance to the next alternative in-frame stop codon downstream, in comparison with their counterparts from control (non-mutated) gene sequences. A paucity of alternative in-frame stop codons was noted in the immediate vicinity (0-49 nucleotides downstream) of the mutated stop codons as compared with their control counterparts (p = 7.81 × 10-4). This implies that at least some nonstop mutations with alternative stop codons in close proximity will not have come to clinical attention, possibly because they will have given rise to stable mRNAs (not subject to nonstop mRNA decay) that are translatable into proteins of near-normal length and biological function. A significant excess of downstream in-frame stop codons was, however, noted in the range 150-199 nucleotides from the mutated stop codon (p = 8.55 × 10-4). We speculate that recruitment of an alternative stop codon at greater distance from the mutated stop codon may trigger nonstop mRNA decay, thereby decreasing the amount of protein product and yielding a readily discernible clinical phenotype. Confirmation or otherwise of this postulate must await the emergence of a clearer understanding of the mechanism of nonstop mRNA decay in mammalian cells.

Comparative analysis of germline and somatic microlesion mutational spectra in 17 human tumor suppressor genes.

Mutations associated with tumorigenesis may either arise somatically or can be inherited through the germline. We performed a comparison of somatic, germline, shared (found in both soma and germline) and somatic recurrent mutational spectra for 17 human tumor suppressor genes, which focused upon missense single base-pair substitutions and microdeletions/microinsertions. Somatic and germline mutational spectra were similar in relation to C.G>T.A transitions but differed with respect to the frequency of A.T>G.C, A.T>T.A, and C.G>A.T substitutions. Shared missense mutations were characterized by higher mutability rates, greater physicochemical differences between wild-type and mutant residues, and a tendency to occur in evolutionarily conserved residues and within CpG/CpHpG oligonucleotides. Mononucleotide runs (≥4 bp) were identified as hotspots for shared microdeletions/microinsertions. Both germline and somatic microdeletions/microinsertions were found to be significantly overrepresented within the "indel-hotspot" motif, GTAAGT. Using a naïve Bayes' classifier trained to discriminate between five missense mutation groups, 63% of mutations in our dataset were on average correctly recognized. Applying this classifier to an independent dataset of probable driver mutations, we concluded that ∼50% of these somatic missense mutations possess features consistent with their being either shared or recurrent, suggesting that a disproportionate number of such lesions are likely to be drivers of tumorigenesis.

In Silico identification of pathogenic strains of Cronobacter from Biochemical data reveals association of inositol fermentation with pathogenicity.

BACKGROUND: Cronobacter, formerly known as Enterobacter sakazakii, is a food-borne pathogen known to cause neonatal meningitis, septicaemia and death. Current diagnostic tests for identification of Cronobacter do not differentiate between species, necessitating time consuming 16S rDNA gene sequencing or multilocus sequence typing (MLST). The organism is ubiquitous, being found in the environment and in a wide range of foods, although there is variation in pathogenicity between Cronobacter isolates and between species. Therefore to be able to differentiate between the pathogenic and non-pathogenic strains is of interest to the food industry and regulators. RESULTS: Here we report the use of Expectation Maximization clustering to categorise 98 strains of Cronobacter as pathogenic or non-pathogenic based on biochemical test results from standard diagnostic test kits. Pathogenicity of a strain was postulated on the basis of either pathogenic symptoms associated with strain source or corresponding MLST sequence types, allowing the clusters to be labelled as containing either pathogenic or non-pathogenic strains. The resulting clusters gave good differentiation of strains into pathogenic and non-pathogenic groups, corresponding well to isolate source and MLST sequence type. The results also revealed a potential association between pathogenicity and inositol fermentation. An investigation of the genomes of Cronobacter sakazakii and C. turicensis revealed the gene for inositol monophosphatase is associated with putative virulence factors in pathogenic strains of Cronobacter. CONCLUSIONS: We demonstrated a computational approach allowing existing diagnostic kits to be used to identify pathogenic strains of Cronobacter. The resulting clusters correlated well with MLST sequence types and revealed new information about the pathogenicity of Cronobacter species.

Gain-of-glycosylation mutations.

Disease-causing missense (and other in-frame) mutations can exert their deleterious effects at the cellular level through multiple mechanisms. A pathogenic mechanism involves the addition of a novel N-linked glycan. Up to 1.4% of known disease-causing missense mutations are predicted to give rise to gains-of-glycosylation. For some of these mutations, the novel glycans have been shown to be both necessary and sufficient to account for the deleterious impact of the mutation. The chemical complementation of cells from patients in vitro with various modifiers of glycosylation has been demonstrated and raises the possibility of specific chemical treatments for patients bearing gain-of-glycosylation mutations.

Characterisation of a functional intronic polymorphism in the human growth hormone (GH1) gene.

The +1169A allele of the A/T single nucleotide polymorphism (SNP; rs2665802), located within intron 4 of the human growth hormone 1 ( GH1 ) gene, has been associated with reduced levels of circulating GH and insulin-like growth factor 1, a reduced risk of colorectal cancer and a predisposition to osteoporosis. Whether this intronic SNP is itself the functional polymorphism responsible for exerting a direct effect on GH1 gene expression, however, or whether it is instead in linkage disequilibrium with the functional SNP, has been an open question. The evolutionary conservation of the +1169T allele (and the surrounding intronic sequence) in the bovine genome, as well as in primate genomes, is, however, suggestive of its functionality. Although a potential alternative splice site spans the location of the +1169 SNP, polymerase chain reaction-based assays failed to yield any evidence for alternative splicing associated with either allele. To determine whether the +1169 SNP, in different allelic combinations with SNPs at -278 (G/T), -57 (T/G) and +2103 (C/T), exerts a direct effect on gene expression and/or GH secretion, we performed a series of transfections of various GH1 haplotype-expressing constructs into rat GC (somatotroph) cells. The results obtained provided evidence to support the contention that the +1169A allele contributes directly to the observed reduction in both GH1 gene expression and GH secretion. Part of the apparent influence of the +1169A-bearing allele on GH1 gene expression and GH secretion may still, however, be attributable to alleles of additional SNPs in cis to +1169A and located within either the promoter or the 3'-flanking region.

Methylation-mediated deamination of 5-methylcytosine appears to give rise to mutations causing human inherited disease in CpNpG trinucleotides, as well as in CpG dinucleotides.

The cytosine-guanine (CpG) dinucleotide has long been known to be a hotspot for pathological mutation in the human genome. This hypermutability is related to its role as the major site of cytosine methylation with the attendant risk of spontaneous deamination of 5-methylcytosine (5mC) to yield thymine. Cytosine methylation, however, also occurs in the context of CpNpG sites in the human genome, an unsurprising finding since the intrinsic symmetry of CpNpG renders it capable of supporting a semi-conservative model of replication of the methylation pattern. Recently, it has become clear that significant DNA methylation occurs in a CpHpG context (where H = A, C or T) in a variety of human somatic tissues. If we assume that CpHpG methylation also occurs in the germline, and that 5mC deamination can occur within a CpHpG context, then we might surmise that methylated CpHpG sites could also constitute mutation hotspots causing human genetic disease. To test this postulate, 54,625 missense and nonsense mutations from 2,113 genes causing inherited disease were retrieved from the Human Gene Mutation Database (http://www.hgmd.org). Some 18.2 per cent of these pathological lesions were found to be C → T and G → A transitions located in CpG dinucleotides (compatible with a model of methylation-mediated deamination of 5mC), an approximately ten-fold higher proportion than would have been expected by chance alone. The corresponding proportion for the CpHpG trinucleotide was 9.9 per cent, an approximately two-fold higher proportion than would have been expected by chance. We therefore estimate that ∼5 per cent of missense/nonsense mutations causing human inherited disease may be attributable to methylation-mediated deamination of 5mC within a CpHpG context.

An isolated case of lissencephaly caused by the insertion of a mitochondrial genome-derived DNA sequence into the 5' untranslated region of the PAFAH1B1 (LIS1) gene.

A 130 base pair (bp) insertion (g.-8delCins130) into the 5' untranslated region of the PAFAH1B1 (LIS1) gene, seven nucleotides upstream of the translational initiation site, was detected in an isolated case of lissencephaly. The inserted DNA sequence exhibited perfect homology to two non-contiguous regions of the mitochondrial genome (8479 to 8545 and 8775 to 8835, containing portions of two genes, ATP8 and ATP6 ), as well as near-perfect homology (1 bp mismatch) to a nuclear mitochondrial pseudogene (NUMT) sequence located on chromosome 1p36. This lesion was not evident on polymerase chain reaction (PCR) sequence analysis of either parent, indicating that the mutation had occurred de novo in the patient. Experiments designed to distinguish between a mitochondrial and a nuclear genomic origin for the inserted DNA sequence were, however, inconclusive. Mitochondrial genome sequences from both the patient and his parents were sequenced and found to be identical to the sequence inserted into the PAFAH1B1 gene. Analysis of parental PCR products from the chromosome 1-specific NUMT were also consistent with the interpretation that the inserted sequence had originated directly from the mitochondrial genome. The chromosome 1-specific NUMT in the patient proved to be refractory to PCR analysis, however, suggesting that this region of chromosome 1 could have been deleted or rearranged. Although it remains by far the most likely scenario, in the absence of DNA sequence information from the patient's own chromosome 1-specific NUMT, we cannot unequivocally confirm that the 130 bp insertion originated from mitochondrial genome rather than from the NUMT.

Intrachromosomal mitotic nonallelic homologous recombination is the major molecular mechanism underlying type-2 NF1 deletions.

Nonallelic homologous recombination (NAHR) is responsible for the recurrent rearrangements that give rise to genomic disorders. Although meiotic NAHR has been investigated in multiple contexts, much less is known about mitotic NAHR despite its importance for tumorigenesis. Because type-2 NF1 microdeletions frequently result from mitotic NAHR, they represent a good model in which to investigate the features of mitotic NAHR. We have used microsatellite analysis and SNP arrays to distinguish between the various alternative recombinational possibilities, thereby ascertaining that 17 of 18 type-2 NF1 deletions, with breakpoints in the SUZ12 gene and its highly homologous pseudogene, originated via intrachromosomal recombination. This high proportion of intrachromosomal NAHR causing somatic type-2 NF1 deletions contrasts with the interchromosomal origin of germline type-1 NF1 microdeletions, whose breakpoints are located within the NF1-REPs (low-copy repeats located adjacent to the SUZ12 sequences). Further, meiotic NAHR causing type-1 NF1 deletions occurs within recombination hotspots characterized by high GC-content and DNA duplex stability, whereas the type-2 breakpoints associated with the mitotic NAHR events investigated here do not cluster within hotspots and are located within regions of significantly lower GC-content and DNA stability. Our findings therefore point to fundamental mechanistic differences between the determinants of mitotic and meiotic NAHR.

A novel third type of recurrent NF1 microdeletion mediated by nonallelic homologous recombination between LRRC37B-containing low-copy repeats in 17q11.2.

Large microdeletions encompassing the neurofibromatosis type-1 (NF1) gene and its flanking regions at 17q11.2 belong to the group of genomic disorders caused by aberrant recombination between segmental duplications. The most common NF1 microdeletions (type-1) span 1.4-Mb and have breakpoints located within NF1-REPs A and C, low-copy repeats (LCRs) containing LRRC37-core duplicons. We have identified a novel type of recurrent NF1 deletion mediated by nonallelic homologous recombination (NAHR) between the highly homologous NF1-REPs B and C. The breakpoints of these approximately 1.0-Mb ("type-3") NF1 deletions were characterized at the DNA sequence level in three unrelated patients. Recombination regions, spanning 275, 180, and 109-bp, respectively, were identified within the LRRC37B-P paralogues of NF1-REPs B and C, and were found to contain sequences capable of non-B DNA formation. Both LCRs contain LRRC37-core duplicons, abundant and highly dynamic sequences in the human genome. NAHR between LRRC37-containing LCRs at 17q21.31 is known to have mediated the 970-kb polymorphic inversions of the MAPT-locus that occurred independently in different primate species, but also underlies the syndromes associated with recurrent 17q21.31 microdeletions and reciprocal microduplications. The novel NF1 microdeletions reported here provide further evidence for the unusually high recombinogenic potential of LRRC37-containing LCRs in the human genome.