Somatic mutational landscape of hereditary hematopoietic malignancies caused by germline variants in RUNX1, GATA2, and DDX41.

Individuals with germ line variants associated with hereditary hematopoietic malignancies (HHMs) have a highly variable risk for leukemogenesis. Gaps in our understanding of premalignant states in HHMs have hampered efforts to design effective clinical surveillance programs, provide personalized preemptive treatments, and inform appropriate counseling for patients. We used the largest known comparative international cohort of germline RUNX1, GATA2, or DDX41 variant carriers without and with hematopoietic malignancies (HMs) to identify patterns of genetic drivers that are unique to each HHM syndrome before and after leukemogenesis. These patterns included striking heterogeneity in rates of early-onset clonal hematopoiesis (CH), with a high prevalence of CH in RUNX1 and GATA2 variant carriers who did not have malignancies (carriers-without HM). We observed a paucity of CH in DDX41 carriers-without HM. In RUNX1 carriers-without HM with CH, we detected variants in TET2, PHF6, and, most frequently, BCOR. These genes were recurrently mutated in RUNX1-driven malignancies, suggesting CH is a direct precursor to malignancy in RUNX1-driven HHMs. Leukemogenesis in RUNX1 and DDX41 carriers was often driven by second hits in RUNX1 and DDX41, respectively. This study may inform the development of HHM-specific clinical trials and gene-specific approaches to clinical monitoring. For example, trials investigating the potential benefits of monitoring DDX41 carriers-without HM for low-frequency second hits in DDX41 may now be beneficial. Similarly, trials monitoring carriers-without HM with RUNX1 germ line variants for the acquisition of somatic variants in BCOR, PHF6, and TET2 and second hits in RUNX1 are warranted.

RUNX1-mutated families show phenotype heterogeneity and a somatic mutation profile unique to germline predisposed AML.

First reported in 1999, germline runt-related transcription factor 1 (RUNX1) mutations are a well-established cause of familial platelet disorder with predisposition to myeloid malignancy (FPD-MM). We present the clinical phenotypes and genetic mutations detected in 10 novel RUNX1-mutated FPD-MM families. Genomic analyses on these families detected 2 partial gene deletions, 3 novel mutations, and 5 recurrent mutations as the germline RUNX1 alterations leading to FPD-MM. Combining genomic data from the families reported herein with aggregated published data sets resulted in 130 germline RUNX1 families, which allowed us to investigate whether specific germline mutation characteristics (type, location) could explain the large phenotypic heterogeneity between patients with familial platelet disorder and different HMs. Comparing the somatic mutational signatures between the available familial (n = 35) and published sporadic (n = 137) RUNX1-mutated AML patients showed enrichment for somatic mutations affecting the second RUNX1 allele and GATA2. Conversely, we observed a decreased number of somatic mutations affecting NRAS, SRSF2, and DNMT3A and the collective genes associated with CHIP and epigenetic regulation. This is the largest aggregation and analysis of germline RUNX1 mutations performed to date, providing a unique opportunity to examine the factors underlying phenotypic differences and disease progression from FPD to MM.

Metastatic tumor evolution and organoid modeling implicate TGFBR2 as a cancer driver in diffuse gastric cancer.

BACKGROUND: Gastric cancer is the second-leading cause of global cancer deaths, with metastatic disease representing the primary cause of mortality. To identify candidate drivers involved in oncogenesis and tumor evolution, we conduct an extensive genome sequencing analysis of metastatic progression in a diffuse gastric cancer. This involves a comparison between a primary tumor from a hereditary diffuse gastric cancer syndrome proband and its recurrence as an ovarian metastasis. RESULTS: Both the primary tumor and ovarian metastasis have common biallelic loss-of-function of both the CDH1 and TP53 tumor suppressors, indicating a common genetic origin. While the primary tumor exhibits amplification of the Fibroblast growth factor receptor 2 (FGFR2) gene, the metastasis notably lacks FGFR2 amplification but rather possesses unique biallelic alterations of Transforming growth factor-beta receptor 2 (TGFBR2), indicating the divergent in vivo evolution of a TGFBR2-mutant metastatic clonal population in this patient. As TGFBR2 mutations have not previously been functionally validated in gastric cancer, we modeled the metastatic potential of TGFBR2 loss in a murine three-dimensional primary gastric organoid culture. The Tgfbr2 shRNA knockdown within Cdh1-/-; Tp53-/- organoids generates invasion in vitro and robust metastatic tumorigenicity in vivo, confirming Tgfbr2 metastasis suppressor activity. CONCLUSIONS: We document the metastatic differentiation and genetic heterogeneity of diffuse gastric cancer and reveal the potential metastatic role of TGFBR2 loss-of-function. In support of this study, we apply a murine primary organoid culture method capable of recapitulating in vivo metastatic gastric cancer. Overall, we describe an integrated approach to identify and functionally validate putative cancer drivers involved in metastasis.

A programmable method for massively parallel targeted sequencing.

We have developed a targeted resequencing approach referred to as Oligonucleotide-Selective Sequencing. In this study, we report a series of significant improvements and novel applications of this method whereby the surface of a sequencing flow cell is modified in situ to capture specific genomic regions of interest from a sample and then sequenced. These improvements include a fully automated targeted sequencing platform through the use of a standard Illumina cBot fluidics station. Targeting optimization increased the yield of total on-target sequencing data 2-fold compared to the previous iteration, while simultaneously increasing the percentage of reads that could be mapped to the human genome. The described assays cover up to 1421 genes with a total coverage of 5.5 Megabases (Mb). We demonstrate a 10-fold abundance uniformity of greater than 90% in 1 log distance from the median and a targeting rate of up to 95%. We also sequenced continuous genomic loci up to 1.5 Mb while simultaneously genotyping SNPs and genes. Variants with low minor allele fraction were sensitively detected at levels of 5%. Finally, we determined the exact breakpoint sequence of cancer rearrangements. Overall, this approach has high performance for selective sequencing of genome targets, configuration flexibility and variant calling accuracy.

Identification of Insertion Deletion Mutations from Deep Targeted Resequencing.

Taking advantage of the deep targeted sequencing capabilities of next generation sequencers, we have developed a novel two step insertion deletion (indel) detection algorithm (IDA) that can determine indels from single read sequences with high computational efficiency and sensitivity when indels are fractionally less compared to wild type reference sequence. First, it identifies candidate indel positions utilizing specific sequence alignment artifacts produced by rapid alignment programs. Second, it confirms the location of the candidate indel by using the Smith-Waterman (SW) algorithm on a restricted subset of Sequence reads. We demonstrate that IDA is applicable to indels of varying sizes from deep targeted sequencing data at low fractions where the indel is diluted by wild type sequence. Our algorithm is useful in detecting indel variants present at variable allelic frequencies such as may occur in heterozygotes and mixed normal-tumor tissue.

The Human OligoGenome Resource: a database of oligonucleotide capture probes for resequencing target regions across the human genome.

Recent exponential growth in the throughput of next-generation DNA sequencing platforms has dramatically spurred the use of accessible and scalable targeted resequencing approaches. This includes candidate region diagnostic resequencing and novel variant validation from whole genome or exome sequencing analysis. We have previously demonstrated that selective genomic circularization is a robust in-solution approach for capturing and resequencing thousands of target human genome loci such as exons and regulatory sequences. To facilitate the design and production of customized capture assays for any given region in the human genome, we developed the Human OligoGenome Resource (http://oligogenome.stanford.edu/). This online database contains over 21 million capture oligonucleotide sequences. It enables one to create customized and highly multiplexed resequencing assays of target regions across the human genome and is not restricted to coding regions. In total, this resource provides 92.1% in silico coverage of the human genome. The online server allows researchers to download a complete repository of oligonucleotide probes and design customized capture assays to target multiple regions throughout the human genome. The website has query tools for selecting and evaluating capture oligonucleotides from specified genomic regions.

Ultrasensitive detection of rare mutations using next-generation targeted resequencing.

With next-generation DNA sequencing technologies, one can interrogate a specific genomic region of interest at very high depth of coverage and identify less prevalent, rare mutations in heterogeneous clinical samples. However, the mutation detection levels are limited by the error rate of the sequencing technology as well as by the availability of variant-calling algorithms with high statistical power and low false positive rates. We demonstrate that we can robustly detect mutations at 0.1% fractional representation. This represents accurate detection of one mutant per every 1000 wild-type alleles. To achieve this sensitive level of mutation detection, we integrate a high accuracy indexing strategy and reference replication for estimating sequencing error variance. We employ a statistical model to estimate the error rate at each position of the reference and to quantify the fraction of variant base in the sample. Our method is highly specific (99%) and sensitive (100%) when applied to a known 0.1% sample fraction admixture of two synthetic DNA samples to validate our method. As a clinical application of this method, we analyzed nine clinical samples of H1N1 influenza A and detected an oseltamivir (antiviral therapy) resistance mutation in the H1N1 neuraminidase gene at a sample fraction of 0.18%.

A cross-sample statistical model for SNP detection in short-read sequencing data.

Highly multiplex DNA sequencers have greatly expanded our ability to survey human genomes for previously unknown single nucleotide polymorphisms (SNPs). However, sequencing and mapping errors, though rare, contribute substantially to the number of false discoveries in current SNP callers. We demonstrate that we can significantly reduce the number of false positive SNP calls by pooling information across samples. Although many studies prepare and sequence multiple samples with the same protocol, most existing SNP callers ignore cross-sample information. In contrast, we propose an empirical Bayes method that uses cross-sample information to learn the error properties of the data. This error information lets us call SNPs with a lower false discovery rate than existing methods.

Performance comparison of whole-genome sequencing platforms.

Whole-genome sequencing is becoming commonplace, but the accuracy and completeness of variant calling by the most widely used platforms from Illumina and Complete Genomics have not been reported. Here we sequenced the genome of an individual with both technologies to a high average coverage of ∼76×, and compared their performance with respect to sequence coverage and calling of single-nucleotide variants (SNVs), insertions and deletions (indels). Although 88.1% of the ∼3.7 million unique SNVs were concordant between platforms, there were tens of thousands of platform-specific calls located in genes and other genomic regions. In contrast, 26.5% of indels were concordant between platforms. Target enrichment validated 92.7% of the concordant SNVs, whereas validation by genotyping array revealed a sensitivity of 99.3%. The validation experiments also suggested that >60% of the platform-specific variants were indeed present in the genome. Our results have important implications for understanding the accuracy and completeness of the genome sequencing platforms.

Targeted sequencing library preparation by genomic DNA circularization.

BACKGROUND: For next generation DNA sequencing, we have developed a rapid and simple approach for preparing DNA libraries of targeted DNA content. Current protocols for preparing DNA for next-generation targeted sequencing are labor-intensive, require large amounts of starting material, and are prone to artifacts that result from necessary PCR amplification of sequencing libraries. Typically, sample preparation for targeted NGS is a two-step process where (1) the desired regions are selectively captured and (2) the ends of the DNA molecules are modified to render them compatible with any given NGS sequencing platform. RESULTS: In this proof-of-concept study, we present an integrated approach that combines these two separate steps into one. Our method involves circularization of a specific genomic DNA molecule that directly incorporates the necessary components for conducting sequencing in a single assay and requires only one PCR amplification step. We also show that specific regions of the genome can be targeted and sequenced without any PCR amplification. CONCLUSION: We anticipate that these rapid targeted libraries will be useful for validation of variants and may have diagnostic application.

Efficient targeted resequencing of human germline and cancer genomes by oligonucleotide-selective sequencing.

We describe an approach for targeted genome resequencing, called oligonucleotide-selective sequencing (OS-Seq), in which we modify the immobilized lawn of oligonucleotide primers of a next-generation DNA sequencer to function as both a capture and sequencing substrate. We apply OS-Seq to resequence the exons of either 10 or 344 cancer genes from human DNA samples. In our assessment of capture performance, >87% of the captured sequence originated from the intended target region with sequencing coverage falling within a tenfold range for a majority of all targets. Single nucleotide variants (SNVs) called from OS-Seq data agreed with >95% of variants obtained from whole-genome sequencing of the same individual. We also demonstrate mutation discovery from a colorectal cancer tumor sample matched with normal tissue. Overall, we show the robust performance and utility of OS-Seq for the resequencing analysis of human germline and cancer genomes.

A flexible approach for highly multiplexed candidate gene targeted resequencing.

We have developed an integrated strategy for targeted resequencing and analysis of gene subsets from the human exome for variants. Our capture technology is geared towards resequencing gene subsets substantially larger than can be done efficiently with simplex or multiplex PCR but smaller in scale than exome sequencing. We describe all the steps from the initial capture assay to single nucleotide variant (SNV) discovery. The capture methodology uses in-solution 80-mer oligonucleotides. To provide optimal flexibility in choosing human gene targets, we designed an in silico set of oligonucleotides, the Human OligoExome, that covers the gene exons annotated by the Consensus Coding Sequencing Project (CCDS). This resource is openly available as an Internet accessible database where one can download capture oligonucleotides sequences for any CCDS gene and design custom capture assays. Using this resource, we demonstrated the flexibility of this assay by custom designing capture assays ranging from 10 to over 100 gene targets with total capture sizes from over 100 Kilobases to nearly one Megabase. We established a method to reduce capture variability and incorporated indexing schemes to increase sample throughput. Our approach has multiple applications that include but are not limited to population targeted resequencing studies of specific gene subsets, validation of variants discovered in whole genome sequencing surveys and possible diagnostic analysis of disease gene subsets. We also present a cost analysis demonstrating its cost-effectiveness for large population studies.

The liver pharmacological and xenobiotic gene response repertoire.

We have used a supervised classification approach to systematically mine a large microarray database derived from livers of compound-treated rats. Thirty-four distinct signatures (classifiers) for pharmacological and toxicological end points can be identified. Just 200 genes are sufficient to classify these end points. Signatures were enriched in xenobiotic and immune response genes and contain un-annotated genes, indicating that not all key genes in the liver xenobiotic responses have been characterized. Many signatures with equal classification capabilities but with no gene in common can be derived for the same phenotypic end point. The analysis of the union of all genes present in these signatures can reveal the underlying biology of that end point as illustrated here using liver fibrosis signatures. Our approach using the whole genome and a diverse set of compounds allows a comprehensive view of most pharmacological and toxicological questions and is applicable to other situations such as disease and development.

A gene expression signature that predicts the future onset of drug-induced renal tubular toxicity.

One application of genomics in drug safety assessment is the identification of biomarkers to predict compound toxicity before it is detected using traditional approaches, such as histopathology. However, many genomic approaches have failed to demonstrate superiority to traditional methods, have not been appropriately validated on external samples, or have been derived using small data sets, thus raising concerns of their general applicability. Using kidney gene expression profiles from male SD rats treated with 64 nephrotoxic or non-nephrotoxic compound treatments, a gene signature consisting of only 35 genes was derived to predict the future development of renal tubular degeneration weeks before it appears histologically following short-term test compound administration. By comparison, histopathology or clinical chemistry fails to predict the future development of tubular degeneration, thus demonstrating the enhanced sensitivity of gene expression relative to traditional approaches. In addition, the performance of the signature was validated on 21 independent compound treatments structurally distinct from the training set. The signature correctly predicted the ability of test compounds to induce tubular degeneration 76% of the time, far better than traditional approaches. This study demonstrates that genomic data can be more sensitive than traditional methods for the early prediction of compound-induced pathology in the kidney.

Development of a large-scale chemogenomics database to improve drug candidate selection and to understand mechanisms of chemical toxicity and action.

Successful drug discovery requires accurate decision making in order to advance the best candidates from initial lead identification to final approval. Chemogenomics, the use of genomic tools in pharmacology and toxicology, offers a promising enhancement to traditional methods of target identification/validation, lead identification, efficacy evaluation, and toxicity assessment. To realize the value of chemogenomics information, a contextual database is needed to relate the physiological outcomes induced by diverse compounds to the gene expression patterns measured in the same animals. Massively parallel gene expression characterization coupled with traditional assessments of drug candidates provides additional, important mechanistic information, and therefore a means to increase the accuracy of critical decisions. A large-scale chemogenomics database developed from in vivo treated rats provides the context and supporting data to enhance and accelerate accurate interpretation of mechanisms of toxicity and pharmacology of chemicals and drugs. To date, approximately 600 different compounds, including more than 400 FDA approved drugs, 60 drugs approved in Europe and Japan, 25 withdrawn drugs, and 100 toxicants, have been profiled in up to 7 different tissues of rats (representing over 3200 different drug-dose-time-tissue combinations). Accomplishing this task required evaluating and improving a number of in vivo and microarray protocols, including over 80 rigorous quality control steps. The utility of pairing clinical pathology assessments with gene expression data is illustrated using three anti-neoplastic drugs: carmustine, methotrexate, and thioguanine, which had similar effects on the blood compartment, but diverse effects on hepatotoxicity. We will demonstrate that gene expression events monitored in the liver can be used to predict pathological events occurring in that tissue as well as in hematopoietic tissues.

Classification of a large microarray data set: algorithm comparison and analysis of drug signatures.

A large gene expression database has been produced that characterizes the gene expression and physiological effects of hundreds of approved and withdrawn drugs, toxicants, and biochemical standards in various organs of live rats. In order to derive useful biological knowledge from this large database, a variety of supervised classification algorithms were compared using a 597-microarray subset of the data. Our studies show that several types of linear classifiers based on Support Vector Machines (SVMs) and Logistic Regression can be used to derive readily interpretable drug signatures with high classification performance. Both methods can be tuned to produce classifiers of drug treatments in the form of short, weighted gene lists which upon analysis reveal that some of the signature genes have a positive contribution (act as "rewards" for the class-of-interest) while others have a negative contribution (act as "penalties") to the classification decision. The combination of reward and penalty genes enhances performance by keeping the number of false positive treatments low. The results of these algorithms are combined with feature selection techniques that further reduce the length of the drug signatures, an important step towards the development of useful diagnostic biomarkers and low-cost assays. Multiple signatures with no genes in common can be generated for the same classification end-point. Comparison of these gene lists identifies biological processes characteristic of a given class.

Molecular basis for fungal selectivity of novel antimitotic compounds.

Compounds that selectively disrupt fungal mitosis have proven to be effective in controlling agricultural pests, but no specific mitotic inhibitor is available for the treatment of systemic mycoses in mammalian hosts. In an effort to identify novel mitotic inhibitors, we used a cell-based screening strategy that exploited the hypersensitivity of a yeast alpha-tubulin mutant strain to growth inhibition by antimitotic agents. The compounds identified inhibited yeast nuclear division and included one structural class of compounds shown to be fungus specific. MC-305904 and structural analogs inhibited fungal cell mitosis and inhibited the in vitro polymerization of fungal tubulin but did not block mammalian cell microtubule function or mammalian tubulin polymerization. Extensive analysis of yeast mutations that specifically alter sensitivity to MC-305904 structural analogs suggested that compounds in the series bind to a site on fungal beta-tubulin near amino acid 198. Features of the proposed binding site explain the observed fungal tubulin specificity of the series and are consistent with structure-activity relationships among a library of related compounds.