Search

Supplementary Figure 1 from Coordinated Functions of E-Cadherin and Transforming Growth Factor β Receptor II In vitro and In vivo

In epithelial cells, E-cadherin plays a key role in cell-cell adhesion, and loss of E-cadherin is a hallmark of tumor progression fostering cancer cell invasion and metastasis. To examine E-cadherin loss in squamous cell cancers, we used primary human esophageal epithelial cells (keratinocytes) as a platform and retrovirally transduced wild-type and dominant-negative forms of E-cadherin into these cells. We found decreased cell adhesion in the cells expressing dominant-negative E-cadherin, thereby resulting in enhanced migration and invasion. To analyze which molecular pathway(s) may modulate these changes, we conducted microarray analysis and found up-regulation of transforming growth factor β receptor II (TβRII) in the wild-type E-cadherin-overexpressing cells, which was confirmed by real-time PCR and Western blot analyses. To investigate the in vivo relevance of this finding, we analyzed tissue microarrays of paired esophageal squamous cell carcinomas and adjacent normal esophagus, and we could show a coordinated loss of E-cadherin and TβRII in ∼80% of tumors. To determine if there may be an E-cadherin-dependent regulation of TβRII, we show the physical interaction of E-cadherin with TβRII and that this is mediated through the extracellular domains of E-cadherin and TβRII, respectively. In addition, TβRI is recruited to this complex. When placed in the context of three-dimensional cell culture, which reflects the physiologic microenvironment, TβRII-mediated cell signaling is dependent upon intact E-cadherin function. Our results, which suggest that E-cadherin regulates TβRII function, have important implications for epithelial carcinogenesis characterized through the frequent occurrence of E-cadherin and TβRII loss. (Cancer Res 2006; 66(20): 9878-85)

Genomic medicine holds great promise for improving health care, but integrating searchable and actionable genetic data into electronic health records (EHRs) remains a challenge. Here we describe Neptune, a system for managing the interaction between a clinical laboratory and an EHR system during the clinical reporting process. We developed Neptune and applied it to two clinical sequencing projects that required report customization, variant reanalysis, and EHR integration. Neptune has been applied for the generation and delivery of over 15,000 clinical genomic reports. This work spans two clinical tests based on targeted gene panels that contain 68 and 153 genes respectively. These projects demanded customizable clinical reports that contained a variety of genetic data types including single-nucleotide variants (SNVs), copy-number variants (CNVs), pharmacogenomics, and polygenic risk scores. Two variant reanalysis activities were also supported, highlighting this important workflow. Methods are needed for delivering structured genetic data to EHRs. This need extends beyond developing data formats to providing infrastructure that manages the reporting process itself. Neptune was successfully applied on two high-throughput clinical sequencing projects to build and deliver clinical reports to EHR systems. The software is open source and available at https://gitlab.com/bcm-hgsc/neptune .

The advancement of precision medicine requires new methods to coordinate and deliver genetic data from heterogeneous sources to physicians and patients. The eMERGE III Network enrolled >25,000 participants from biobank and prospective cohorts of predominantly healthy individuals for clinical genetic testing to determine clinically actionable findings. The network developed protocols linking together the 11 participant collection sites and 2 clinical genetic testing laboratories. DNA capture panels targeting 109 genes were used for testing of DNA and sample collection, data generation, interpretation, reporting, delivery, and storage were each harmonized. A compliant and secure network enabled ongoing review and reconciliation of clinical interpretations, while maintaining communication and data sharing between clinicians and investigators. A total of 202 individuals had positive diagnostic findings relevant to the indication for testing and 1,294 had additional/secondary findings of medical significance deemed to be returnable, establishing data return rates for other testing endeavors. This study accomplished integration of structured genomic results into multiple electronic health record (EHR) systems, setting the stage for clinical decision support to enable genomic medicine. Further, the established processes enable different sequencing sites to harmonize technical and interpretive aspects of sequencing tests, a critical achievement toward global standardization of genomic testing. The eMERGE protocols and tools are available for widespread dissemination.

When sequencing small RNA libraries derived from whole blood, the most abundant microRNAs (miRs) detected are often miR-486-5p, miR-451a, and miR-92a-3p. These highly expressed erythropoietic miRs are released into the sample from red blood cell hemolysis. Next-generation sequencing of these unwanted miRs leads to a waste in sequencing cost and diminished detection of lowly expressed miRNAs, including many potential miRNA biomarkers. Previous work has developed a method to reduce targeted miRNAs using oligonucleotides that bind their target miRNA and prevent its ligation during library construction, although the extent to which oligonucleotides can be multiplexed and their effect on larger cohorts has not been thoroughly explored. We present a method for suppressing detection of three highly abundant heme miRs in a single multiplexed blocking oligonucleotide reaction. In a small paired-sample pilot (n = 8) and a large cohort of samples (n = 901), multiplexed oligos reduced detection of their target miRNAs by approximately 70%, allowing for an approximately 10-fold increase in reads mapping to nonheme miRs and increased detection of very lowly expressed miRs, with minimal off-target effects. By removing all three highly expressed erythropoietic miRNAs from next-generational sequencing libraries, this commercially available multiplexed blocking oligonucleotide method allows for greater detection of lowly expressed biomarkers, improving the efficacy, cost-efficiency, and sensitivity of biomarker studies and diagnostic tests.

Next-generation sequencing (NGS) is now routinely used to interrogate large sets of genes in a diagnostic setting. Regions of high sequence homology continue to be a major challenge for short-read technologies and can lead to false-positive and false-negative diagnostic errors. At the scale of whole-exome sequencing (WES), laboratories may be limited in their knowledge of genes and regions that pose technical hurdles due to high homology. We have created an exome-wide resource that catalogs highly homologous regions that is tailored toward diagnostic applications. This resource was developed using a mappability-based approach tailored to current Sanger and NGS protocols. Gene-level and exon-level lists delineate regions that are difficult or impossible to analyze via standard NGS. These regions are ranked by degree of affectedness, annotated for medical relevance, and classified by the type of homology (within-gene, different functional gene, known pseudogene, uncharacterized noncoding region). Additionally, we provide a list of exons that cannot be analyzed by short-amplicon Sanger sequencing. This resource can help guide clinical test design, supplemental assay implementation, and results interpretation in the context of high homology. Genet Med 18 12, 1282–1289.

Diagnostic next-generation sequencing (NGS)-based gene panels are increasingly used for prevalent disorders with genetic and clinical heterogeneity. Clinical development, validation, and quality management of these panels ideally includes reference samples containing prevalent pathogenic variants; however, clinical domain expertise to select appropriate variants may not be present, samples are often not publicly available, and their inclusion is associated with added cost. Expert-designed, multiplexed controls can remedy some of these challenges. One approach relies on spiking biosynthetic fragments carrying desired variants into human genomic DNA. We piloted the utility of this approach for hypertrophic cardiomyopathy. Data from >3000 previously sequenced probands were used to select 10 common pathogenic and/or technically challenging variants in the top hypertrophic cardiomyopathy genes. Multiplexed controls were constructed across a range of ideal and realistic allelic fractions for heterozygous germline variants. NGS was performed in quadruplicate, and results were compared with diagnostic NGS data for the source patient samples. Overall, results were indistinguishable from patient-derived data with variants being detected at or reasonably close to the targeted allelic fraction ratios. The exception was a common 25-bp deletion in MYBPC3, underscoring the importance of including such variants in test development. These controls may be an attractive addition to the repertoire of materials for development, validation, and quality monitoring of clinical NGS assays.

With next generation sequencing technology improvement and cost reductions, it has become technically feasible to sequence a large number of genes in one diagnostic test. This is especially relevant for diseases with large genetic and/or phenotypic heterogeneity, such as hearing loss. However, variant interpretation remains the major bottleneck. This is further exacerbated by the lack in the clinical genetics community of consensus criteria for defining the evidence necessary to include genes on targeted disease panels or in genomic reports, and the consequent risk of reporting variants in genes with no relevance to disease. We describe a systematic evidence-based approach for assessing gene–disease associations and for curating relevant genes for different disease aspects, including mode of inheritance, phenotypic severity, and mutation spectrum. By applying this approach to clinically available hearing loss gene panels with a total of 163 genes, we show that a significant number (45%) of genes lack sufficient evidence of association with disease and thus are expected to increase uncertainty and patient anxiety, in addition to intensifying the interpretation burden. Information about all curated genes is summarized. Our retrospective analysis of 539 hearing loss cases tested by our previous OtoGenomeV2 panel demonstrates the impact of including genes with weak disease association in laboratory wet-bench and interpretation processes. Our study is, to our knowledge, the first to highlight the urgent need for defining the clinical validity of gene–disease relationships for more efficient and accurate clinical testing and reporting. Genet Med 18 6, 545–553.

Background Diagnostic testing for genetic cardiomyopathies has undergone dramatic changes in the last decade with next generation sequencing (NGS) expanding the number of genes that can be interrogated simultaneously. Exon resolution copy number analysis is increasingly incorporated into routine diagnostic testing via cytogenomic arrays and more recently via NGS. While NGS is an attractive option for laboratories that have no access to array platforms, its higher false positive rate requires weighing the added cost incurred by orthogonal confirmation against the magnitude of the increase in diagnostic yield. Although copy number variants (CNVs) have been reported in various cardiomyopathy genes, their contribution has not been systematically studied. Methods We performed single exon resolution NGS‐based deletion/duplication analysis for up to 46 cardiomyopathy genes in >1400 individuals with cardiomyopathies including HCM, DCM, ARVC, RCM, and LVNC. Results and Conclusion Clinically significant deletions and duplications were identified in only 9 of 1425 (0.63%) individuals. The majority of those (6/9) represented intragenic events. We conclude that the added benefit of exon level deletion/duplication analysis is low for currently known cardiomyopathy genes and may not outweigh the increased cost and complexity of incorporating it into routine diagnostic testing for these disorders.

Variant interpretation depends on accurate annotations using biologically relevant transcripts. We have developed a systematic strategy for designating primary transcripts, and applied it to 109 hearing loss-associated genes that were divided into 3 categories. Category 1 genes (n=38) had a single transcript, Category 2 genes (n=32) had multiple transcripts, but a single transcript was sufficient to represent all exons, and Category 3 genes (n=38) had multiple transcripts with unique exons. Transcripts were curated with respect to gene expression reported in the literature and the Genotype-Tissue Expression Project. In addition, high frequency loss of function variants in the Genome Aggregation Database, and disease-causing variants in ClinVar and the Human Gene Mutation Database across the 109 genes were queried. These data were used to classify exons as "clinically relevant", "uncertain significance", or "clinically insignificant". Interestingly, 7% of all exons, containing >124 "clinically significant" variants, were of “uncertain significance”. Finally, we used exon-level next generation sequencing quality metrics generated at two clinical labs, and identified a total of 43 technically challenging exons in 20 different genes that had inadequate coverage and/or homology issues which might lead to false variant calls. We have demonstrated that transcript analysis plays a critical role in accurate clinical variant interpretation.

Variant interpretation depends on accurate annotations using biologically relevant transcripts. We have developed a systematic strategy for designating primary transcripts and have applied it to 109 hearing loss–associated genes that were divided into three categories. Category 1 genes (n = 38) had a single transcript; category 2 genes (n = 33) had multiple transcripts, but a single transcript was sufficient to represent all exons; and category 3 genes (n = 38) had multiple transcripts with unique exons. Transcripts were curated with respect to gene expression reported in the literature and the Genotype-Tissue Expression Project. In addition, high-frequency loss-of-function variants in the Genome Aggregation Database and disease-causing variants in ClinVar and the Human Gene Mutation Database across the 109 genes were queried. These data were used to classify exons as clinically significant, insignificant, or of uncertain significance. Interestingly, 6% of all exons, containing 124 reportedly disease-causing variants, were of uncertain significance. Finally, we used exon-level next-generation sequencing quality metrics generated at two clinical laboratories and identified a total of 43 technically challenging exons in 20 different genes that had inadequate coverage and/or homology issues that might lead to false-variant calls. We have demonstrated that transcript analysis plays a critical role in accurate clinical variant interpretation.

Pathogenic variants at the DFNB1 locus encompassing the GJB2 and GJB6 genes account for 50% of autosomal-recessive, congenital nonsyndromic hearing loss in the United States. Most cases are caused by sequence variants within the GJB2 gene, but a significant number of DFNB1 patients carry a large deletion (GJB6-D13S1830) in trans with a GJB2 variant. This deletion lies upstream of GJB2 and was shown to reduce GJB2 expression by disrupting unidentified regulatory elements. First-tier genetic testing for hearing loss includes GJB2 sequence and GJB6-D13S1830 deletion analysis; however, several other deletions in this locus, each with distinct breakpoints, have been reported in DFNB1 patients and are missed by current panels. Here, we report the development of a targeted droplet digital polymerase chain reaction-based assay for comprehensive copy-number analysis at the DFNB1 locus that detects all deletions reported to date. This assay increased detection rates in a multiethnic cohort of 87 hearing loss patients with only one identified pathogenic GJB2 variant. We identify two deletions, one of which is novel, in two patients (2/87 or 2.3%), suggesting that other pathogenic deletions at the DFNB1 locus may be missed. Mapping the assayed DFNB1 deletions also revealed a ∼ 95 kb critical region, which may harbor the GJB2 regulatory element(s).

Next-generation sequencing (NGS) technologies have revolutionized genetic testing by enabling simultaneous analysis of unprecedented numbers of genes. However, genes with high-sequence homology pose challenges to current NGS technologies. Because diagnostic sequencing is moving toward exome analysis, knowledge of these homologous genes is essential to avoid false positive and negative results. An example is the STRC gene, one of >70 genes known to contribute to the genetic basis of hearing loss. STRC is 99.6% identical to a pseudogene (pSTRC) and therefore inaccessible to standard NGS methodologies. The STRC locus is also known to be a common site for large deletions. Comprehensive diagnostic testing for inherited hearing loss therefore necessitates a combination of several approaches to avoid pseudogene interference. We have developed a clinical test that combines standard NGS and NGS-based copy number assessment supplemented with a long-range PCR-based Sanger or MiSeq assay to eliminate pseudogene contamination. By using this combination of assays we could identify biallelic STRC variants in 14% (95% CI, 8%–24%) of individuals with isolated nonsyndromic hearing loss who had previously tested negative on our 70-gene hearing loss panel, corresponding to a detection rate of 11.2% (95% CI, 6%–19%) for previously untested patients. This approach has broad applicability because medically significant genes for many disease areas include genes with high-sequence homology.

Dilated cardiomyopathy is characterized by substantial locus, allelic, and clinical heterogeneity that necessitates testing of many genes across clinically overlapping diseases. Few studies have sequenced sufficient individuals; thus, the contributions of individual genes and the pathogenic variant spectrum are still poorly defined. We analyzed 766 dilated cardiomyopathy patients tested over 5 years in our molecular diagnostics laboratory. Patients were tested using gene panels of increasing size from 5 to 46 genes, including 121 cases tested with a multiple-cardiomyopathy next-generation panel covering 46 genes. All variants were reassessed using our current clinical-grade scoring system to eliminate false-positive disease associations that afflict many older analyses. Up to 37% of dilated cardiomyopathy cases carry a clinically relevant variant in one of 20 genes, titin (TTN) being the largest contributor (up to 14%). Desmoplakin (DSP), an arrhythmogenic right ventricular cardiomyopathy gene, contributed 2.4%, illustrating the utility of multidisease testing. The clinical sensitivity increased from 10 to 37% as gene panel sizes increased. However, the number of inconclusive cases also increased from 4.6 to 51%. Our data illustrate the utility of broad gene panels for genetically and clinically heterogeneous diseases but also highlight challenges as molecular diagnostics moves toward genome-wide testing. Genet Med 16 8, 601–608.

Parental imprinting of differentially methylated regions (DMRs) contributes to appropriate expression of several developmentally important genes from paternally or maternally derived chromosomes. Rhabdomyosarcoma (RMS) is the most common soft-tissue sarcoma in children and is associated with altered expression of certain parentally imprinted genes. As previously reported, RMS cells display loss of imprinting (LOI) of the DMR at the IGF2-H19 locus, resulting in insulin-like growth factor 2 (IGF2) transcription from both paternally and maternally inherited chromosomes, and overall IGF2 overexpression. As the DLK1-GTL2 locus is structurally similar to the IGF2-H19 locus, the status of parental imprinting of the DLK1-GTL2 locus was studied in RMS. We observed that while both embryonal and alveolar rhabdomyosarcomas (ERMS and ARMS, respectively) show LOI of the DMR at the IGF2-H19 locus, imprinting of the DMR at the DLK1-GTL2 locus varies in association with the histological subtype of RMS. We found that, while ERMS tumors consistently show LOI of the DMR at the DLK1-GTL2 locus, ARMS tumors have erasure of imprinting (EOI) at this locus. These changes in imprinting status of the DLK1-GTL2 locus result in a higher GTL2/DLK1 mRNA ratio in ARMS as compared to ERMS. This difference in imprinting elucidates a novel genetic difference between these two RMS subtypes and may provide a potential diagnostic tool to distinguish between these subtypes.

Diagnostic next-generation sequencing (NGS)-based gene panels are increasingly used for prevalent disorders with genetic and clinical heterogeneity. Clinical development, validation, and quality management of these panels ideally includes reference samples containing prevalent pathogenic variants; however, clinical domain expertise to select appropriate variants may not be present, samples are often not publicly available, and their inclusion is associated with added cost. Expert-designed, multiplexed controls can remedy some of these challenges. One approach relies on spiking biosynthetic fragments carrying desired variants into human genomic DNA. We piloted the utility of this approach for hypertrophic cardiomyopathy. Data from3000 previously sequenced probands were used to select 10 common pathogenic and/or technically challenging variants in the top hypertrophic cardiomyopathy genes. Multiplexed controls were constructed across a range of ideal and realistic allelic fractions for heterozygous germline variants. NGS was performed in quadruplicate, and results were compared with diagnostic NGS data for the source patient samples. Overall, results were indistinguishable from patient-derived data with variants being detected at or reasonably close to the targeted allelic fraction ratios. The exception was a common 25-bp deletion in MYBPC3, underscoring the importance of including such variants in test development. These controls may be an attractive addition to the repertoire of materials for development, validation, and quality monitoring of clinical NGS assays.

Velo-cardio-facial syndrome/DiGeorge syndrome, also known as 22q11.2 deletion syndrome (22q11DS) is the most common microdeletion syndrome, with an estimated incidence of 1/2,000 – 1/4,000 live births. Approximately 9–11% of patients with this disorder have an overt cleft palate (CP), but the genetic factors responsible for CP in the 22q11DS subset are unknown. The TBX1 gene, a member of the T-box transcription factor gene family, lies within the 22q11.2 region that is hemizygous in patients with 22q11DS. Inactivation of one allele of Tbx1 in the mouse does not result in CP, but inactivation of both alleles does. Based on these data, we hypothesized that DNA variants in the remaining allele of TBX1 may confer risk to CP in patients with 22q11DS. To test the hypothesis, we evaluated TBX1 exon sequencing (n = 360) and genotyping data (n = 737) with respect to presence (n = 54) or absence (n = 683) of CP in patients with 22q11DS. Two upstream SNPs (rs4819835 and rs5748410) showed individual evidence for association but they were not significant after correction for multiple testing. Associations were not identified between DNA variants and haplotypes in 22q11DS patients with CP. Overall, this study indicates that common DNA variants in TBX1 may be nominally causative for CP in patients with 22q11DS. This raises the possibility that genes elsewhere on the remaining allele of 22q11.2 or in the genome could be relevant.

Haploinsufficiency of TBX1, encoding a T-box transcription factor, is largely responsible for the physical malformations in velo-cardio-facial /DiGeorge/22q11.2 deletion syndrome (22q11DS) patients. Cardiovascular malformations in these patients are highly variable, raising the question as to whether DNA variations in the TBX1 locus on the remaining allele of 22q11.2 could be responsible. To test this, a large sample size is needed. The TBX1 gene was sequenced in 360 consecutive 22q11DS patients. Rare and common variations were identified. We did not detect enrichment in rare SNP (single nucleotide polymorphism) number in those with or without a congenital heart defect. One exception was that there was increased number of very rare SNPs between those with normal heart anatomy compared to those with right-sided aortic arch or persistent truncus arteriosus, suggesting potentially protective roles in the SNPs for these phenotype-enrichment groups. Nine common SNPs (minor allele frequency, MAF > 0.05) were chosen and used to genotype the entire cohort of 1,022 22q11DS subjects. We did not find a correlation between common SNPs or haplotypes and cardiovascular phenotype. This work demonstrates that common DNA variations in TBX1 do not explain variable cardiovascular expression in 22q11DS patients, implicating existence of modifiers in other genes on 22q11.2 or elsewhere in the genome. Hum Mutat 32:1278–1289, 2011. ©2011 Wiley Periodicals, Inc.

Insulin-like growth factor 2 mRNA-binding protein-1 (IMP-1) is an oncofetal protein that binds directly to and stabilizes oncogenic c-Myc and regulates, in turn, its posttranscriptional expression and translation. In contrast to normal adult tissue, IMP-1 is reexpressed and/or overexpressed in human cancers. We show that knockdown of c-Myc in human colon cancer cell lines increases the expression of mature let-7 miRNA family members and downregulates several of its mRNA targets: IMP-1, Cdc34, and K-Ras. We further show that loss of IMP-1 inhibits Cdc34, Lin-28B, and K-Ras, suppresses SW-480 cell proliferation and anchorage-independent growth, and promotes caspase- and lamin-mediated cell death. We also found that IMP-1 binds to the coding region and 3′UTR of K-Ras mRNA. RNA microarray profiling and validation by reverse transcription PCR reveals that the p53-inducible proapoptotic protein CYFIP2 is upregulated in IMP-1 knockdown SW480 cells, a novel finding. We also show that overexpression of IMP-1 increases c-Myc and K-Ras expression and LIM2405 cell proliferation. Furthermore, we show that loss of IMP-1 induces Caspase-3- and PARP-mediated apoptosis, and inhibits K-Ras expression in SW480 cells, which is rescued by CYFIP2 knockdown. Importantly, analysis of 228 patients with colon cancers reveals that IMP-1 is significantly upregulated in differentiated colon tumors (P ≤ 0.0001) and correlates with K-Ras expression (r = 0.35, P ≤ 0.0001) relative to adjacent normal mucosa. These findings indicate that IMP-1, interrelated with c-Myc, acts upstream of K-Ras to promote survival through a novel mechanism that may be important in colon cancer pathogenesis. Cancer Res; 71(6); 2172–82. ©2011 AACR.

Purpose: To develop and validate VisCap, a software program targeted to clinical laboratories for inference and visualization of germ-line copy-number variants (CNVs) from targeted next-generation sequencing data. Genet Med 18 7, 712–719. Methods: VisCap calculates the fraction of overall sequence coverage assigned to genomic intervals and computes log2 ratios of these values to the median of reference samples profiled using the same test configuration. Candidate CNVs are called when log2 ratios exceed user-defined thresholds. Genet Med 18 7, 712–719. Results: We optimized VisCap using 14 cases with known CNVs, followed by prospective analysis of 1,104 cases referred for diagnostic DNA sequencing. To verify calls in the prospective cohort, we used droplet digital polymerase chain reaction (PCR) to confirm 10/27 candidate CNVs and 72/72 copy-neutral genomic regions scored by VisCap. We also used a genome-wide bead array to confirm the absence of CNV calls across panels applied to 10 cases. To improve specificity, we instituted a visual scoring system that enabled experienced reviewers to differentiate true-positive from false-positive calls with minimal impact on laboratory workflow. Genet Med 18 7, 712–719. Conclusions: VisCap is a sensitive method for inferring CNVs from targeted sequence data from targeted gene panels. Visual scoring of data underlying CNV calls is a critical step to reduce false-positive calls for follow-up testing. Genet Med 18 7, 712–719.

In epithelial cells, E-cadherin plays a key role in cell-cell adhesion, and loss of E-cadherin is a hallmark of tumor progression fostering cancer cell invasion and metastasis. To examine E-cadherin loss in squamous cell cancers, we used primary human esophageal epithelial cells (keratinocytes) as a platform and retrovirally transduced wild-type and dominant-negative forms of E-cadherin into these cells. We found decreased cell adhesion in the cells expressing dominant-negative E-cadherin, thereby resulting in enhanced migration and invasion. To analyze which molecular pathway(s) may modulate these changes, we conducted microarray analysis and found up-regulation of transforming growth factor β receptor II (TβRII) in the wild-type E-cadherin-overexpressing cells, which was confirmed by real-time PCR and Western blot analyses. To investigate the in vivo relevance of this finding, we analyzed tissue microarrays of paired esophageal squamous cell carcinomas and adjacent normal esophagus, and we could show a coordinated loss of E-cadherin and TβRII in ∼80% of tumors. To determine if there may be an E-cadherin-dependent regulation of TβRII, we show the physical interaction of E-cadherin with TβRII and that this is mediated through the extracellular domains of E-cadherin and TβRII, respectively. In addition, TβRI is recruited to this complex. When placed in the context of three-dimensional cell culture, which reflects the physiologic microenvironment, TβRII-mediated cell signaling is dependent upon intact E-cadherin function. Our results, which suggest that E-cadherin regulates TβRII function, have important implications for epithelial carcinogenesis characterized through the frequent occurrence of E-cadherin and TβRII loss. (Cancer Res 2006; 66(20): 9878-85)

The genetic basis of pancreatic ductal adenocarcinoma, which constitutes the most common type of pancreatic malignancy, involves the sequential activation of oncogenes and inactivation of tumor suppressor genes. Among the pivotal genetic alterations are Ki-RAS oncogene activation and p53 tumor suppressor gene inactivation. We explain that the combination of these genetic events facilitates pancreatic carcinogenesis as revealed in novel three-dimensional cell (spheroid cyst) culture and in vivo subcutaneous and orthotopic xenotransplantation models. N-cadherin, a member of the classic cadherins important in the regulation of cell-cell adhesion, is induced in the presence of Ki-RAS mutation but subsequently downregulated with the acquisition of p53 mutation as revealed by gene microarrays and corroborated by reverse transcription-PCR and Western blotting. N-cadherin modulates the capacity of pancreatic ductal cells to migrate and invade, in part via complex formation with keratinocyte growth factor receptor and neural cell adhesion molecule and in part via interaction with p120-catenin. However, modulation of these complexes by Ki-RAS and p53 leads to enhanced cell migration and invasion. This preferentially induces the downstream effector AKT over mitogen-activated protein kinase to execute changes in cellular behavior. Thus, we are able to define molecules that in part are directly affected by Ki-RAS and p53 during pancreatic ductal carcinogenesis, and this provides a platform for potential new molecularly based therapeutic interventions.

Loss of heterozygosity on chromosome 22q13.31 is a frequent event during human breast and colorectal carcinogenesis. Herein we characterize a novel gene at chromosome 22q13.31 designated PRR5 . Alternative promoter usage and splicing converge to generate five PRR5 transcript variants with maximum mRNA expression in kidney. In vitro transcription/translation demonstrated that the five variants generate three protein isoforms differing in their N-terminal length. Mutational analysis of PRR5 in human breast and colorectal tumors did not reveal somatic mutations. However, mRNA expression analyses revealed PRR5 overexpression in a majority of colorectal tumors but substantial downregulation of PRR5 expression in a subset of breast tumors and reduced expression in two breast cancer cell lines. Treatment with trichostatin A increased PRR5 mRNA levels in BT549 and MDA-MB-231 cells, whereas 5′-aza-2′-deoxycytidine induced expression in MDA-MB-231 cells only. Thus, PRR5 may represent a potential candidate tumor suppressor gene in breast cancer.

Tuberous sclerosis complex, an autosomal dominant disease caused by mutations in either TSC1 or TSC2, is characterized by the development of hamartomas in a variety of organs. The proteins encoded by TSC1 and TSC2, hamartin and tuberin, respectively, associate with each other forming a tight complex. Here we show that hamartin binds the neurofilament light chain and it is possible to recover the hamartin-tuberin complex over the neurofilament light chain rod domain spanning amino acids 93-156 by affinity precipitation. Homologous rod domains in other intermediate filaments such as neurofilament medium chain, alpha-internexin, vimentin, and desmin are not able to bind hamartin. In cultured cortical neurons, hamartin and tuberin co-localize with neurofilament light chain preferentially in the proximal to central growth cone region. Interestingly, in the distal part of the growth cone hamartin overlaps with the ezrin-radixin-moesin family of actin binding proteins, and we have validated the interaction of hamartin with moesin. These results demonstrate that hamartin may anchor neuronal intermediate filaments to the actin cytoskeleton, which may be critical for some of the CNS functions of the hamartin-tuberin complex, and abolishing this through mutations in TSC1 or TSC2 may lead to certain neurological manifestations associated with the disease.

We established a bioinformatics procedure at our laboratory that processes sequencing data, identifies single nucleotide variants (SNVs) and small insertions/deletions (indels), annotates the variants based on the curated variant databases and algorithms, and filters variants according to custom defined filtration schemas that are selected by geneticists specifically for each case. Since we are operating in the clinical settings, rigorous validation procedures are applied at each step to ensure the accuracy and quality of the data. Here we describe how we implemented our current clinical exome pipeline, and our approaches for validation.

Next-generation sequencing (NGS) technologies have revolutionized genetic testing by enabling simultaneous analysis of unprecedented numbers of genes. However, genes with high-sequence homology pose challenges to current NGS technologies. Because diagnostic sequencing is moving toward exome analysis, knowledge of these homologous genes is essential to avoid false positive and negative results. An example is the STRC gene, one of70 genes known to contribute to the genetic basis of hearing loss. STRC is 99.6% identical to a pseudogene (pSTRC) and therefore inaccessible to standard NGS methodologies. The STRC locus is also known to be a common site for large deletions. Comprehensive diagnostic testing for inherited hearing loss therefore necessitates a combination of several approaches to avoid pseudogene interference. We have developed a clinical test that combines standard NGS and NGS-based copy number assessment supplemented with a long-range PCR-based Sanger or MiSeq assay to eliminate pseudogene contamination. By using this combination of assays we could identify biallelic STRC variants in 14% (95% CI, 8%-24%) of individuals with isolated nonsyndromic hearing loss who had previously tested negative on our 70-gene hearing loss panel, corresponding to a detection rate of 11.2% (95% CI, 6%-19%) for previously untested patients. This approach has broad applicability because medically significant genes for many disease areas include genes with high-sequence homology.

Palatal anomalies are one of the identifying features of 22q11.2 deletion syndrome (22q11.2DS) affecting about one third of patients. To identify genetic variants that increase the risk of cleft or palatal anomalies in 22q11.2DS patients, we performed a candidate gene association study in 101 patients with 22q11.2DS genotyped with the Affymetrix genome-wide human SNP array 6.0.Patients from Children's Hospital of Philadelphia, USA and Wilhelmina Children's Hospital Utrecht, The Netherlands were stratified based on palatal phenotype (overt cleft, submucosal cleft, bifid uvula). SNPs in 21 candidate genes for cleft palate were analyzed for genotype-phenotype association. In addition, TBX1 sequencing was carried out. Quality control and association analyses were conducted using the software package PLINK.Genotype and phenotype data of 101 unrelated patients (63 non-cleft subjects (62.4%), 38 cleft subjects (37.6%)) were analyzed. A Total of 39 SNPs on 10 genes demonstrated a p-value ≤0.05 prior to correction. The most significant SNPs were found on FGF10. However none of the SNPs remained significant after correcting for multiple testing.Although these results are promising, analysis of additional samples will be required to confirm that variants in these regions influence risk for cleft palate or palatal anomalies in 22q11.2DS patients.

Parvin-beta is a focal adhesion protein downregulated in human breast cancer cells. Loss of Parvin-beta contributes to increased integrin-linked kinase activity, cell-matrix adhesion, and invasion through the extracellular matrix in vitro. The effect of ectopic Parvin-beta expression on the transcriptional profile of MDA-MB-231 breast cancer cells, which normally do not express Parvin-beta, was evaluated. Particular emphasis was placed upon propagating MDA-MB-231 breast cancer cells in three-dimensional culture matrices. Interestingly, Parvin-beta reexpression in MDA-MB-231 cells increased the mRNA expression, serine 82 phosphorylation (mediated by CDK9), and activity of the nuclear hormone receptor peroxisome proliferator-activated receptor gamma (PPARgamma), and there was a concomitant increase in lipogenic gene expression as a downstream effector of PPARgamma. Importantly, Parvin-beta suppressed breast cancer growth in vivo, with associated decreased proliferation. These data suggest that Parvin-beta might influence breast cancer progression.

We recently identified Pam (for protein associated with c-Myc), as a binding partner for the tuberous sclerosis complex (TSC) protein tuberin in brain. The highly conserved Pam homologs in Drosophila and C. elegans are neuron-specific proteins that regulate synaptic growth. The Pam gene contains 83 exons and encodes a 4,641-amino-acid polypeptide with a predicted molecular weight of approximately 510 kDa. In a previous study, we demonstrated that Pam is expressed as two forms, approximately 450 kDa in rat embryonic and a approximately 350 kDa in rat adult brain. Here we have extended that work to show the approximately 450 kDa form is expressed in rat embryonic kidney, heart, and lung and in rat cell lines, and the approximately 350 kDa form is expressed in adult rat tissues as well as in human and mouse brain and human and mouse cell lines. To understand the size difference, we investigated alternative splicing of Pam in brain and detected six isoforms in the Myc-binding region resulting from splicing of exon 53, and three new exons, 52A, 56, and 56A. We also demonstrate that the presence of exon 52A in Pam significantly enhances binding to Myc, suggesting functional importance of this alternative splicing. The presence of Pam in many cellular compartments, its spliced variants, as well as its multiple binding partners, including tuberin, make it a complex, yet intriguing protein in the nervous system.

The epidermal growth factor receptor (EGFR) activates several signaling cascades in response to epidermal growth factor stimulation. One of these signaling events involves tyrosine phosphorylation of signal transducer and activator of transcription (STAT), whereas another involves activation of the phosphatidylinositol 3-OH kinase pathway. Two possibilities for STAT activation exist: a janus kinase (JAK)-dependent and a JAK-independent mechanism. Herein, we demonstrate that EGFR overexpression in primary esophageal keratinocytes activates STAT in a JAK-dependent fashion with the functional consequence of enhanced cell migration, which can be abolished by use of a JAK-specific inhibitor, AG-490. We determined the mechanisms underlying the signal transduction pathway responsible for increased cell migration. Stimulation of EGFR induces Tyr701 phosphorylation of STAT1 and initiates complex formation of STAT1 and STAT3 with JAK1 and JAK2. Thereafter, the STATs translocate to the nucleus within 15 min. In addition, we found that activation of this signaling pathway results in matrix metalloproteinase-1 (MMP-1) activity. By contrast, Akt activation does not impact the EGFR-STATs-JAKs complex formation and nuclear translocation of the STATs with subsequent MMP-1 activity, although Akt activation may contribute to cell migration through an independent mechanism. Taken together, we find that the recruitment of the STAT-JAK complex by EGFR is responsible for keratinocyte migration that, in turn, might be mediated by MMP-1 activation.

Expression of fetal globin is silenced normally in adult life; however, determinants linked and/or unlinked to the globin-gene clusters could modify Hb F expression so it persists into adults. Increased expression in adults offers hope as a cure for sickle cell disease (SCD) and beta thalassemia, since formation of FS hybrids in SCD inhibits deoxy Hb S polymerization while increased fetal chain expression compensates partially for decreased adult beta-globin chains in beta thalassemia. Characterization and controlled manipulation of high Hb F determinants is critical to decreasing clinical severity of these life-threatening genetic diseases, which result in high morbidity and mortality worldwide. We report on analysis of a unique beta-thalassemia cohort from Sardinia who present with either a mild, non-transfusion-dependent (NTD) form expressing high Hb F, or with a severe, transfusion-dependent (TD) form expressing low Hb F. Both groups are homozygous for the beta39 chain-termination mutation and lack adult beta globin. Genome-wide DNA arrays were run on 42 TD and 33 NTD patients using the Affymetrix 500K (500,568 SNPs in a total of 23 individuals) or 6.0 (909,622 SNPs in a total of 52 individuals) SNP chip platforms. The average sample call rates were 96.3% for the 6.0 chip and 94.3% for the 500K chip. Data from the two chips were merged and a total of 482,243 SNPs that were present in both chips were considered for genetic association analysis using a case/control design (NTD versus TD). Quality control filters removed 42,166 SNPs (8.7%) with genotype call rates < 90%. An additional 54,683 SNPs (12.4%) were removed because they either failed a test of Hardy-Weinberg equilibrium at a p< 0.0001 or had a minor allele frequency (MAF) of < 0.01. The genotype call rate for the remaining 385,394 SNPs was 97.3%. Preliminary analysis of difference in allele frequency distributions in the two groups of patients did not reveal any signal of association that remained statistically significant when corrected for multiple tests. The smallest p-value observed was 2x10-6 for SNP rs8028098 on chromosome 15. However, several SNPs in two candidate regions on chromosomes 2p16 and 6q23 were nominally significant (p