Creation and evaluation of full-text literature-derived, feature-weighted disease models of genetically determined developmental disorders.

There are >2500 different genetically determined developmental disorders (DD), which, as a group, show very high levels of both locus and allelic heterogeneity. This has led to the wide-spread use of evidence-based filtering of genome-wide sequence data as a diagnostic tool in DD. Determining whether the association of a filtered variant at a specific locus is a plausible explanation of the phenotype in the proband is crucial and commonly requires extensive manual literature review by both clinical scientists and clinicians. Access to a database of weighted clinical features extracted from rigorously curated literature would increase the efficiency of this process and facilitate the development of robust phenotypic similarity metrics. However, given the large and rapidly increasing volume of published information, conventional biocuration approaches are becoming impractical. Here, we present a scalable, automated method for the extraction of categorical phenotypic descriptors from the full-text literature. Papers identified through literature review were downloaded and parsed using the Cadmus custom retrieval package. Human Phenotype Ontology terms were extracted using MetaMap, with 76-84% precision and 65-73% recall. Mean terms per paper increased from 9 in title + abstract, to 68 using full text. We demonstrate that these literature-derived disease models plausibly reflect true disease expressivity more accurately than widely used manually curated models, through comparison with prospectively gathered data from the Deciphering Developmental Disorders study. The area under the curve for receiver operating characteristic (ROC) curves increased by 5-10% through the use of literature-derived models. This work shows that scalable automated literature curation increases performance and adds weight to the need for this strategy to be integrated into informatic variant analysis pipelines. Database URL: https://doi.org/10.1093/database/baac038.

Clinically-relevant postzygotic mosaicism in parents and children with developmental disorders in trio exome sequencing data.

Mosaic genetic variants can have major clinical impact. We systematically analyse trio exome sequence data from 4,293 probands from the DDD Study with severe developmental disorders for pathogenic postzygotic mosaicism (PZM) in the child or a clinically-unaffected parent, and use ultrahigh-depth sequencing to validate candidate mosaic variants. We observe that levels of mosaicism for small genetic variants are usually equivalent in both saliva and blood and ~3% of causative de novo mutations exhibit PZM; this is an important observation, as the sibling recurrence risk is extremely low. We identify parental PZM in 21 trios (0.5% of trios), resulting in a substantially increased sibling recurrence risk in future pregnancies. Together, these forms of mosaicism account for 40 (1%) diagnoses in our cohort. Likely child-PZM mutations occur equally on both parental haplotypes, and the penetrance of detectable mosaic pathogenic variants overall is likely to be less than half that of constitutive variants.

FISH mapping of de novo apparently balanced chromosome rearrangements identifies characteristics associated with phenotypic abnormality.

We report fluorescence in situ hybridization (FISH) mapping of 152, mostly de novo, apparently balanced chromosomal rearrangement (ABCR) breakpoints in 76 individuals, 30 of whom had no obvious phenotypic abnormality (control group) and 46 of whom had an associated disease (case group). The aim of this study was to identify breakpoint characteristics that could discriminate between these groups and which might be of predictive value in de novo ABCR (DN-ABCR) cases detected antenatally. We found no difference in the proportion of breakpoints that interrupted a gene, although in three cases, direct interruption or deletion of known autosomal-dominant or X-linked recessive Mendelian disease genes was diagnostic. The only significant predictor of phenotypic abnormality in the group as a whole was the localization of one or both breakpoints to an R-positive (G-negative) band with estimated predictive values of 0.69 (95% CL 0.54-0.81) and 0.90 (95% CL 0.60-0.98), respectively. R-positive bands are known to contain more genes and have a higher guanine-cytosine (GC) content than do G-positive (R-negative) bands; however, whether a gene was interrupted by the breakpoint or the GC content in the 200 kB around the breakpoint had no discriminant ability. Our results suggest that the large-scale genomic context of the breakpoint has prognostic utility and that the pathological mechanism of mapping to an R-band cannot be accounted for by direct gene inactivation.

Chromosome analysis: what and when to request.

Chromosome abnormalities have long been recognised as an important cause of learning disability and multiple malformation syndromes; 0.8% of live born infants have numerical or structural chromosomal anomalies resulting in an abnormal phenotype. The identification of such anomalies is important, both clinically and for accurate genetic counselling. Recently, the human genome sequence has enabled higher resolution screens for chromosome anomalies using both molecular cytogenetic and array based techniques. This review suggests a simple algorithm for the targeted use of diagnostic cytogenetic tools in specific patient groups commonly seen in paediatric practice.

Hypermethylation of the 5' CpG island of the gene encoding the serine protease Testisin promotes its loss in testicular tumorigenesis.

The Testisin gene (PRSS21) encodes a glycosylphosphatidylinositol (GPI)-linked serine protease that exhibits testis tissue-specific expression. Loss of Testisin has been implicated in testicular tumorigenesis, but its role in testis biology and tumorigenesis is not known. Here we have investigated the role of CpG methylation in Testisin gene inactivation and tested the hypothesis that Testisin may act as a tumour suppressor for testicular tumorigenesis. Using sequence analysis of bisulphite-treated genomic DNA, we find a strong relationship between hypermethylation of a 385 bp 5' CpG rich island of the Testisin gene, and silencing of the Testisin gene in a range of human tumour cell lines and in 100% (eight/eight) of testicular germ cell tumours. We show that treatment of Testisin-negative cell lines with demethylating agents and/or a histone deacetylase inhibitor results in reactivation of Testisin gene expression, implicating hypermethylation in Testisin gene silencing. Stable expression of Testisin in the Testisin-negative Tera-2 testicular cancer line suppressed tumorigenicity as revealed by inhibition of both anchorage-dependent cell growth and tumour formation in an SCID mouse model of testicular tumorigenesis. Together, these data show that loss of Testisin is caused, at least in part, by DNA hypermethylation and histone deacetylation, and suggest a tumour suppressor role for Testisin in testicular tumorigenesis.

A3243G mitochondrial mutation associated with polymicrogyria.

The mitochondrial transfer ribonucleic acid for leucine is encoded by nucleotides 3230-3304. A-to-G transition at nucleotide 3243 can cause maternally transmitted diabetes mellitus-deafness syndrome, and MELAS syndrome. MELAS syndrome is a rare disorder of mitochondrial energy production, and is an acronym for myopathy, encephalopathy, lactic acidosis, and stroke-like episodes. Cortical malformations are heterogeneous and result from abnormal cell proliferation/apoptosis, migration, and/or differentiation of neuroepithelial cells. They are an important and relatively common cause of intractable epilepsy and neurodevelopmental disorders. The association between these A3243G mutations and cortical malformation has never before been reported. Here a 14-year-old female with A3243G mutation and polymicrogyria is described and possible aetiologies of this association are discussed.

MLYCD mutation analysis: evidence for protein mistargeting as a cause of MLYCD deficiency.

Malonyl-CoA decarboxylase (MLYCD) deficiency is an autosomal recessive disorder characterized by malonic aciduria, developmental delay, seizure disorder, hypoglycemia, and cardiomyopathy. Genomic sequencing of MLYCD in nine unrelated patients identified 16 of 18 pathogenic alleles, which are documented in the newly created Human MLYCD Allelic Variant Database (http://mlycd.hgu.mrc.ac.uk/). Fibroblast cell lines were available from eight of these patients and two previously reported patients with homozygous MLYCD mutations. Western blot analysis using antisera raised to a C-terminal peptide detected a 66-kDa band that was absent in six patients and substantially reduced in three patients. One patient showed an increase in protein levels with a prominent smeary 68-l83-kDa band. Immunocytochemical analysis of MLYCD-expressing patient cell lines showed apparent intracellular mislocalization. An extreme N-terminal mutation c.8G>A (p.G3D) mislocalized to the plasma membrane, suggesting that a novel targeting signal may reside in a four-amino acid conserved N-terminal motif. A 25-base deletion between the putative mitochondrial and peroxisomal initiating codons (M1 and M40) and a point mutation ablating the second of these (c.119T>C, p.M40T) both showed punctate perinuclear staining. As none of the three mislocalizing mutations are predicted to alter the catalytic function of the peptide, it seems likely that correct subcellular localization of MLYCD is critical for it to function normally.

Symmetrical upper limb peromelia and lower limb phocomelia associated with a de novo apparently balanced reciprocal translocation: 46,XX,t(2; 12)(p25.1;q24.1).

We report a female fetus of 20 weeks gestation with severe symmetrical deformity affecting all four limbs. These deformities were unusual in that there was upper limb peromelia and lower limb phocomelia. No additional major malformations were identified on postmortem examination. In particular there was no evidence of splenogonadal fusion or micrognathia and hypoglossia. The limb malformations in this case are associated with a de novo apparently balanced reciprocal translocation 46,XX,t(2;12)(p25.1;q24.1). The cytogenetic features of Roberts-SC phocomelia syndrome were not detected. Unfortunately, the fibroblast line died and no FISH or DNA analysis could be carried out. In spite of this, the case is presented as it may be useful to other researchers in the selection of candidate genes for mendelian forms of peromelia and phocomelia.

A critical role for Dnmt1 and DNA methylation in T cell development, function, and survival.

The role of DNA methylation and of the maintenance DNA methyltransferase Dnmt1 in the epigenetic regulation of developmental stage- and cell lineage-specific gene expression in vivo is uncertain. This is addressed here through the generation of mice in which Dnmt1 was inactivated by Cre/loxP-mediated deletion at sequential stages of T cell development. Deletion of Dnmt1 in early double-negative thymocytes led to impaired survival of TCRalphabeta(+) cells and the generation of atypical CD8(+)TCRgammadelta(+) cells. Deletion of Dnmt1 in double-positive thymocytes impaired activation-induced proliferation but differentially enhanced cytokine mRNA expression by naive peripheral T cells. We conclude that Dnmt1 and DNA methylation are required for the proper expression of certain genes that define fate and determine function in T cells.

Mutations in the 3beta-hydroxysterol Delta24-reductase gene cause desmosterolosis, an autosomal recessive disorder of cholesterol biosynthesis.

Desmosterolosis is a rare autosomal recessive disorder characterized by multiple congenital anomalies. Patients with desmosterolosis have elevated levels of the cholesterol precursor desmosterol, in plasma, tissue, and cultured cells; this abnormality suggests a deficiency of the enzyme 3beta-hydroxysterol Delta24-reductase (DHCR24), which, in cholesterol biosynthesis, catalyzes the reduction of the Delta24 double bond of sterol intermediates. We identified the human DHCR24 cDNA, by the similarity between the encoded protein and a recently characterized plant enzyme--DWF1/DIM, from Arabidopsis thaliana--catalyzing a different but partially similar reaction in steroid/sterol biosynthesis in plants. Heterologous expression, in the yeast Saccharomyces cerevisiae, of the DHCR24 cDNA, followed by enzyme-activity measurements, confirmed that it encodes DHCR24. The encoded DHCR24 protein has a calculated molecular weight of 60.1 kD, contains a potential N-terminal secretory-signal sequence as well as at least one putative transmembrane helix, and is a member of a recently defined family of flavin adenine dinucleotide (FAD)-dependent oxidoreductases. Conversion of desmosterol to cholesterol by DHCR24 in vitro is strictly dependent on reduced nicotinamide adenine dinucleotide phosphate and is increased twofold by the addition of FAD to the assay. The corresponding gene, DHCR24, was identified by database searching, spans approximately 46.4 kb, is localized to chromosome 1p31.1-p33, and comprises nine exons and eight introns. Sequence analysis of DHCR24 in two patients with desmosterolosis revealed four different missense mutations, which were shown, by functional expression, in yeast, of the patient alleles, to be disease causing. Our data demonstrate that desmosterolosis is a cholesterol-biosynthesis disorder caused by mutations in DHCR24.

Dengue virus binding to human leukocyte cell lines: receptor usage differs between cell types and virus strains.

Monocyte macrophages (Mphi) are thought to be the principal target cells for the dengue viruses (DV), the cause of dengue fever and hemorrhagic fever. Cell attachment is mediated by the virus envelope (E) protein, but the host-cell receptors remain elusive. Currently, candidate receptor molecules include proteins, Fc receptors, glycosaminoglycans (GAGs) and lipopolysaccharide binding CD14-associated molecules. Here, we show that in addition to Mphi, cells of the T- and B-cell lineages, and including cells lacking GAGs, can bind and become infected with DV. The level of virus binding varied widely between cell lines and, notably, between virus strains within a DV serotype. The latter difference may be ascribable to one or more amino acid differences in domain II of the E protein. Heparin had no significant effect on DV binding, while heparinase treatment of cells in all cases increased DV binding, further supporting the contention that GAGs are not required for DV binding and infection of human cells. In contrast to a recent report, we found that lipopolysaccharide (LPS) had either no effect or enhanced DV binding to, and infection of, various human leukocyte cell lines, while in all virus-cell combinations, depletion of Ca(2+)/Mg(2+) enhanced DV binding. This argues against involvement of beta(2) integrins in virus-host cell interactions, a conclusion in accord with the demonstration of three virus binding membrane proteins of < 75 kDa. Collectively, the results of this study question the purported exclusive importance of the E protein domain III in DV binding to host cells and point to a far more complex interaction between various target cells and, notably, individual DV strains.

HPP1: a transmembrane protein-encoding gene commonly methylated in colorectal polyps and cancers.

Adenomas are the precursors of most colorectal cancers. Hyperplastic polyps have been linked to the subset of colorectal cancers showing DNA microsatellite instability, but little is known of their underlying genetic etiology. Using a strategy that isolates differentially methylated sequences from hyperplastic polyps and normal mucosa, we identified a 370-bp sequence containing the 5' untranslated region and the first exon of a gene that we have called HPP1. Rapid amplification of cDNA ends was used to isolate HPP1 from normal mucosa. Using reverse transcription-PCR, HPP1 was expressed in 28 of 30 (93%) normal colonic samples but in only seven of 30 (23%) colorectal cancers (P < 0.001). The 5' region of HPP1 included a CpG island containing 49 CpG sites, of which 96% were found to be methylated by bisulfite sequencing of DNA from colonic tumor samples. By COBRA analysis, methylation was detected in six of nine (66%) adenomas, 17 of 27 (63%) hyperplastic polyps, and 46 of 55 (84%) colorectal cancers. There was an inverse relationship between methylation level and mRNA expression in cancers (r = -0.67; P < 0.001), and 5-aza-2-deoxycytidine treatment restored HPP1 expression in two colorectal cancer cell lines. In situ hybridization of HPP1 indicated that expression occurs in epithelial and stromal elements in normal mucosa but is silenced in both cell types in early colonic neoplasia. HPP1 is predicted to encode a transmembrane protein containing follistatin and epidermal growth factor-like domains. Silencing of HPP1 by methylation may increase the probability of neoplastic transformation.