Machine learning multi-omics analysis reveals cancer driver dysregulation in pan-cancer cell lines compared to primary tumors.

Cancer cell lines have been widely used for decades to study biological processes driving cancer development, and to identify biomarkers of response to therapeutic agents. Advances in genomic sequencing have made possible large-scale genomic characterizations of collections of cancer cell lines and primary tumors, such as the Cancer Cell Line Encyclopedia (CCLE) and The Cancer Genome Atlas (TCGA). These studies allow for the first time a comprehensive evaluation of the comparability of cancer cell lines and primary tumors on the genomic and proteomic level. Here we employ bulk mRNA and micro-RNA sequencing data from thousands of samples in CCLE and TCGA, and proteomic data from partner studies in the MD Anderson Cell Line Project (MCLP) and The Cancer Proteome Atlas (TCPA), to characterize the extent to which cancer cell lines recapitulate tumors. We identify dysregulation of a long non-coding RNA and microRNA regulatory network in cancer cell lines, associated with differential expression between cell lines and primary tumors in four key cancer driver pathways: KRAS signaling, NFKB signaling, IL2/STAT5 signaling and TP53 signaling. Our results emphasize the necessity for careful interpretation of cancer cell line experiments, particularly with respect to therapeutic treatments targeting these important cancer pathways.

TPP1 promoter mutations cooperate with TERT promoter mutations to lengthen telomeres in melanoma.

Overcoming replicative senescence is an essential step during oncogenesis, and the reactivation of TERT through promoter mutations is a common mechanism. TERT promoter mutations are acquired in about 75% of melanomas but are not sufficient to maintain telomeres, suggesting that additional mutations are required. We identified a cluster of variants in the promoter of ACD encoding the shelterin component TPP1. ACD promoter variants are present in about 5% of cutaneous melanoma and co-occur with TERT promoter mutations. The two most common somatic variants create or modify binding sites for E-twenty-six (ETS) transcription factors, similar to mutations in the TERT promoter. The variants increase the expression of TPP1 and function together with TERT to synergistically lengthen telomeres. Our findings suggest that TPP1 promoter variants collaborate with TERT activation to enhance telomere maintenance and immortalization in melanoma.

A Functional Precision Medicine Pipeline Combines Comparative Transcriptomics and Tumor Organoid Modeling to Identify Bespoke Treatment Strategies for Glioblastoma.

Li Fraumeni syndrome (LFS) is a hereditary cancer predisposition syndrome caused by germline mutations in TP53. TP53 is the most common mutated gene in human cancer, occurring in 30-50% of glioblastomas (GBM). Here, we highlight a precision medicine platform to identify potential targets for a GBM patient with LFS. We used a comparative transcriptomics approach to identify genes that are uniquely overexpressed in the LFS GBM patient relative to a cancer compendium of 12,747 tumor RNA sequencing data sets, including 200 GBMs. STAT1 and STAT2 were identified as being significantly overexpressed in the LFS patient, indicating ruxolitinib, a Janus kinase 1 and 2 inhibitors, as a potential therapy. The LFS patient had the highest level of STAT1 and STAT2 expression in an institutional high-grade glioma cohort of 45 patients, further supporting the cancer compendium results. To empirically validate the comparative transcriptomics pipeline, we used a combination of adherent and organoid cell culture techniques, including ex vivo patient-derived organoids (PDOs) from four patient-derived cell lines, including the LFS patient. STAT1 and STAT2 expression levels in the four patient-derived cells correlated with levels identified in the respective parent tumors. In both adherent and organoid cultures, cells from the LFS patient were among the most sensitive to ruxolitinib compared to patient-derived cells with lower STAT1 and STAT2 expression levels. A spheroid-based drug screening assay (3D-PREDICT) was performed and used to identify further therapeutic targets. Two targeted therapies were selected for the patient of interest and resulted in radiographic disease stability. This manuscript supports the use of comparative transcriptomics to identify personalized therapeutic targets in a functional precision medicine platform for malignant brain tumors.

The case for using mapped exonic non-duplicate reads when reporting RNA-sequencing depth: examples from pediatric cancer datasets.

BACKGROUND: The reproducibility of gene expression measured by RNA sequencing (RNA-Seq) is dependent on the sequencing depth. While unmapped or non-exonic reads do not contribute to gene expression quantification, duplicate reads contribute to the quantification but are not informative for reproducibility. We show that mapped, exonic, non-duplicate (MEND) reads are a useful measure of reproducibility of RNA-Seq datasets used for gene expression analysis. FINDINGS: In bulk RNA-Seq datasets from 2,179 tumors in 48 cohorts, the fraction of reads that contribute to the reproducibility of gene expression analysis varies greatly. Unmapped reads constitute 1-77% of all reads (median [IQR], 3% [3-6%]); duplicate reads constitute 3-100% of mapped reads (median [IQR], 27% [13-43%]); and non-exonic reads constitute 4-97% of mapped, non-duplicate reads (median [IQR], 25% [16-37%]). MEND reads constitute 0-79% of total reads (median [IQR], 50% [30-61%]). CONCLUSIONS: Because not all reads in an RNA-Seq dataset are informative for reproducibility of gene expression measurements and the fraction of reads that are informative varies, we propose reporting a dataset's sequencing depth in MEND reads, which definitively inform the reproducibility of gene expression, rather than total, mapped, or exonic reads. We provide a Docker image containing (i) the existing required tools (RSeQC, sambamba, and samblaster) and (ii) a custom script to calculate MEND reads from RNA-Seq data files. We recommend that all RNA-Seq gene expression experiments, sensitivity studies, and depth recommendations use MEND units for sequencing depth.

Identification of a differentiation stall in epithelial mesenchymal transition in histone H3-mutant diffuse midline glioma.

BACKGROUND: Diffuse midline gliomas with histone H3 K27M (H3K27M) mutations occur in early childhood and are marked by an invasive phenotype and global decrease in H3K27me3, an epigenetic mark that regulates differentiation and development. H3K27M mutation timing and effect on early embryonic brain development are not fully characterized. RESULTS: We analyzed multiple publicly available RNA sequencing datasets to identify differentially expressed genes between H3K27M and non-K27M pediatric gliomas. We found that genes involved in the epithelial-mesenchymal transition (EMT) were significantly overrepresented among differentially expressed genes. Overall, the expression of pre-EMT genes was increased in the H3K27M tumors as compared to non-K27M tumors, while the expression of post-EMT genes was decreased. We hypothesized that H3K27M may contribute to gliomagenesis by stalling an EMT required for early brain development, and evaluated this hypothesis by using another publicly available dataset of single-cell and bulk RNA sequencing data from developing cerebral organoids. This analysis revealed similarities between H3K27M tumors and pre-EMT normal brain cells. Finally, a previously published single-cell RNA sequencing dataset of H3K27M and non-K27M gliomas revealed subgroups of cells at different stages of EMT. In particular, H3.1K27M tumors resemble a later EMT stage compared to H3.3K27M tumors. CONCLUSIONS: Our data analyses indicate that this mutation may be associated with a differentiation stall evident from the failure to proceed through the EMT-like developmental processes, and that H3K27M cells preferentially exist in a pre-EMT cell phenotype. This study demonstrates how novel biological insights could be derived from combined analysis of several previously published datasets, highlighting the importance of making genomic data available to the community in a timely manner.

Hydra: A mixture modeling framework for subtyping pediatric cancer cohorts using multimodal gene expression signatures.

Precision oncology has primarily relied on coding mutations as biomarkers of response to therapies. While transcriptome analysis can provide valuable information, incorporation into workflows has been difficult. For example, the relative rather than absolute gene expression level needs to be considered, requiring differential expression analysis across samples. However, expression programs related to the cell-of-origin and tumor microenvironment effects confound the search for cancer-specific expression changes. To address these challenges, we developed an unsupervised clustering approach for discovering differential pathway expression within cancer cohorts using gene expression measurements. The hydra approach uses a Dirichlet process mixture model to automatically detect multimodally distributed genes and expression signatures without the need for matched normal tissue. We demonstrate that the hydra approach is more sensitive than widely-used gene set enrichment approaches for detecting multimodal expression signatures. Application of the hydra analysis framework to small blue round cell tumors (including rhabdomyosarcoma, synovial sarcoma, neuroblastoma, Ewing sarcoma, and osteosarcoma) identified expression signatures associated with changes in the tumor microenvironment. The hydra approach also identified an association between ATRX deletions and elevated immune marker expression in high-risk neuroblastoma. Notably, hydra analysis of all small blue round cell tumors revealed similar subtypes, characterized by changes to infiltrating immune and stromal expression signatures.

Bayesian Framework for Detecting Gene Expression Outliers in Individual Samples.

PURPOSE: Many antineoplastics are designed to target upregulated genes, but quantifying upregulation in a single patient sample requires an appropriate set of samples for comparison. In cancer, the most natural comparison set is unaffected samples from the matching tissue, but there are often too few available unaffected samples to overcome high intersample variance. Moreover, some cancer samples have misidentified tissues of origin or even composite-tissue phenotypes. Even if an appropriate comparison set can be identified, most differential expression tools are not designed to accommodate comparisons to a single patient sample. METHODS: We propose a Bayesian statistical framework for gene expression outlier detection in single samples. Our method uses all available data to produce a consensus background distribution for each gene of interest without requiring the researcher to manually select a comparison set. The consensus distribution can then be used to quantify over- and underexpression. RESULTS: We demonstrate this method on both simulated and real gene expression data. We show that it can robustly quantify overexpression, even when the set of comparison samples lacks ideally matched tissue samples. Furthermore, our results show that the method can identify appropriate comparison sets from samples of mixed lineage and rediscover numerous known gene-cancer expression patterns. CONCLUSION: This exploratory method is suitable for identifying expression outliers from comparative RNA sequencing (RNA-seq) analysis for individual samples, and Treehouse, a pediatric precision medicine group that leverages RNA-seq to identify potential therapeutic leads for patients, plans to explore this method for processing its pediatric cohort.

Comparative RNA-seq analysis aids in diagnosis of a rare pediatric tumor.

Gliomatosis peritonei is a rare pathologic finding that is associated with ovarian teratomas and malignant mixed germ cell tumors. The occurrence of gliomatosis as a mature glial implant can impart an improved prognosis to patients with immature ovarian teratoma, making prompt and accurate diagnosis important. We describe a case of recurrent immature teratoma in a 10-yr-old female patient, in which comparative analysis of the RNA sequencing gene expression data from the patient's tumor was used effectively to aid in the diagnosis of gliomatosis peritonei.

Whole genome sequencing in cats, identifies new models for blindness in AIPL1 and somite segmentation in HES7.

BACKGROUND: The reduced cost and improved efficiency of whole genome sequencing (WGS) is drastically improving the development of cats as biomedical models. Persian cats are models for Leber's congenital amaurosis (LCA), the most severe and earliest onset form of visual impairment in humans. Cats with innocuous breed-defining traits, such as a bobbed tail, can also be models for somite segmentation and vertebral column development. METHODS: The first WGS in cats was conducted on a trio segregating for LCA and the bobbed tail abnormality. Variants were identified using FreeBayes and effects predicted using SnpEff. Variants within a known haplotype block for cat LCA and specific candidate genes for both phenotypes were prioritized by the predicted variant effect on the proteins and concordant segregation within the trio. The efficiency of WGS of a single trio of domestic cats was evaluated. RESULTS: A stop gain was identified at position c.577C > T in cat AIPL1, a predicted p.Arg193*. A c.5A > G variant causing a p.V2A was identified in HES7. The variants segregated concordantly in a Persian - Japanese bobtail pedigree. Over 1700 cats from 40 different breeds and populations were genotyped for the AIPL1 variant, defining an allelic frequency in only Persian -related breeds of 1.15%. A sub-set of cats was genotyped for the HES7 variant, supporting the variant as private to the Japanese bobtail breed. Approximately 18 million SNPs were identified for application in cat research. The cat AIPL1 variant would have been considered a high priority variant for evaluation, regardless of a priori knowledge from previous genetic studies. CONCLUSIONS: This study represents the first effort of the 99 Lives Cat Genome Sequencing Initiative to identify disease--causing variants in the domestic cat using WGS. The current cat reference assembly is efficient for gene and variant identification. However, as the feline variant database improves, development of cats as biomedical models for human disease will be more efficient, providing an alternative, large animal model for drug and gene therapy trials. Undiagnosed human patients with early-onset blindness should be screened for this AIPL1 variant. The HES7 variant should further calibrate the somite segmentation clock.

Derived variants at six genes explain nearly half of size reduction in dog breeds.

Selective breeding of dogs by humans has generated extraordinary diversity in body size. A number of multibreed analyses have been undertaken to identify the genetic basis of this diversity. We analyzed four loci discovered in a previous genome-wide association study that used 60,968 SNPs to identify size-associated genomic intervals, which were too large to assign causative roles to genes. First, we performed fine-mapping to define critical intervals that included the candidate genes GHR, HMGA2, SMAD2, and STC2, identifying five highly associated markers at the four loci. We hypothesize that three of the variants are likely to be causative. We then genotyped each marker, together with previously reported size-associated variants in the IGF1 and IGF1R genes, on a panel of 500 domestic dogs from 93 breeds, and identified the ancestral allele by genotyping the same markers on 30 wild canids. We observed that the derived alleles at all markers correlated with reduced body size, and smaller dogs are more likely to carry derived alleles at multiple markers. However, breeds are not generally fixed at all markers; multiple combinations of genotypes are found within most breeds. Finally, we show that 46%-52.5% of the variance in body size of dog breeds can be explained by seven markers in proximity to exceptional candidate genes. Among breeds with standard weights <41 kg (90 lb), the genotypes accounted for 64.3% of variance in weight. This work advances our understanding of mammalian growth by describing genetic contributions to canine size determination in non-giant dog breeds.

Variation of BMP3 contributes to dog breed skull diversity.

Since the beginnings of domestication, the craniofacial architecture of the domestic dog has morphed and radiated to human whims. By beginning to define the genetic underpinnings of breed skull shapes, we can elucidate mechanisms of morphological diversification while presenting a framework for understanding human cephalic disorders. Using intrabreed association mapping with museum specimen measurements, we show that skull shape is regulated by at least five quantitative trait loci (QTLs). Our detailed analysis using whole-genome sequencing uncovers a missense mutation in BMP3. Validation studies in zebrafish show that Bmp3 function in cranial development is ancient. Our study reveals the causal variant for a canine QTL contributing to a major morphologic trait.

Leading the way: finding genes for neurologic disease in dogs using genome-wide mRNA sequencing.

Because of dogs' unique population structure, human-like disease biology, and advantageous genomic features, the canine system has risen dramatically in popularity as a tool for discovering disease alleles that have been difficult to find by studying human families or populations. To date, disease studies in dogs have primarily employed either linkage analysis, leveraging the typically large family size, or genome-wide association, which requires only modest-sized case and control groups in dogs. Both have been successful but, like most techniques, each requires a specific combination of time and money, and there are inherent problems associated with each. Here we review the first report of mRNA-Seq in the dog, a study that provides insights into the potential value of applying high-throughput sequencing to the study of genetic diseases in dogs. Forman and colleagues apply high-throughput sequencing to a single case of canine neonatal cerebellar cortical degeneration. This implementation of whole genome mRNA sequencing, the first reported in dog, is additionally unusual due to the analysis: the data was used not to examine transcript levels or annotate genes, but as a form of target capture that revealed the sequence of transcripts of genes associated with ataxia in humans. This approach entails risks. It would fail if, for example, the relevant transcripts were not sufficiently expressed for genotyping or were not associated with ataxia in humans. But here it pays off handsomely, identifying a single frameshift mutation that segregates with the disease. This work sets the stage for similar studies that take advantage of recent advances in genomics while exploiting the historical background of dog breeds to identify disease-causing mutations.

Integration of biochemical signalling in spines.

Short-term and long-term changes in the strength of synapses in neural networks underlie working memory and long-term memory storage in the brain. These changes are regulated by many biochemical signalling pathways in the postsynaptic spines of excitatory synapses. Recent findings about the roles and regulation of the small GTPases Ras, Rap and Rac in spines provide new insights into the coordination and cooperation of different pathways to effect synaptic plasticity. Here, we present an initial working representation of the interactions of five signalling cascades that are usually studied individually. We discuss their integrated function in the regulation of postsynaptic plasticity.