Interactions between ploidy and resource availability shape clonal interference at initiation and recurrence of glioblastoma.

Glioblastoma (GBM) is the most aggressive form of primary brain tumor. Complete surgical resection of GBM is almost impossible due to the infiltrative nature of the cancer. While no evidence for recent selection events have been found after diagnosis, the selective forces that govern gliomagenesis are strong, shaping the tumor's cell composition during the initial progression to malignancy with late consequences for invasiveness and therapy response. We present a mathematical model that simulates the growth and invasion of a glioma, given its ploidy level and the nature of its brain tissue micro-environment (TME), and use it to make inferences about GBM initiation and response to standard-of-care treatment. We approximate the spatial distribution of resource access in the TME through integration of in-silico modelling, multi-omics data and image analysis of primary and recurrent GBM. In the pre-malignant setting, our in-silico results suggest that low ploidy cancer cells are more resistant to starvation-induced cell death. In the malignant setting, between first and second surgery, simulated tumors with different ploidy compositions progressed at different rates. Whether higher ploidy predicted fast recurrence, however, depended on the TME. Historical data supports this dependence on TME resources, as shown by a significant correlation between the median glucose uptake rates in human tissues and the median ploidy of cancer types that arise in the respective tissues (Spearman r = -0.70; P = 0.026). Taken together our findings suggest that availability of metabolic substrates in the TME drives different cell fate decisions for cancer cells with different ploidy and shapes GBM disease initiation and relapse characteristics.

Mathematical Modeling of Clonal Interference by Density-Dependent Selection in Heterogeneous Cancer Cell Lines.

Many cancer cell lines are aneuploid and heterogeneous, with multiple karyotypes co-existing within the same cell line. Karyotype heterogeneity has been shown to manifest phenotypically, thus affecting how cells respond to drugs or to minor differences in culture media. Knowing how to interpret karyotype heterogeneity phenotypically would give insights into cellular phenotypes before they unfold temporally. Here, we re-analyzed single cell RNA (scRNA) and scDNA sequencing data from eight stomach cancer cell lines by placing gene expression programs into a phenotypic context. Using live cell imaging, we quantified differences in the growth rate and contact inhibition between the eight cell lines and used these differences to prioritize the transcriptomic biomarkers of the growth rate and carrying capacity. Using these biomarkers, we found significant differences in the predicted growth rate or carrying capacity between multiple karyotypes detected within the same cell line. We used these predictions to simulate how the clonal composition of a cell line would change depending on density conditions during in-vitro experiments. Once validated, these models can aid in the design of experiments that steer evolution with density-dependent selection.

Virtual alignment of pathology image series for multi-gigapixel whole slide images.

Interest in spatial omics is on the rise, but generation of highly multiplexed images remains challenging, due to cost, expertise, methodical constraints, and access to technology. An alternative approach is to register collections of whole slide images (WSI), generating spatially aligned datasets. WSI registration is a two-part problem, the first being the alignment itself and the second the application of transformations to huge multi-gigapixel images. To address both challenges, we developed Virtual Alignment of pathoLogy Image Series (VALIS), software which enables generation of highly multiplexed images by aligning any number of brightfield and/or immunofluorescent WSI, the results of which can be saved in the ome.tiff format. Benchmarking using publicly available datasets indicates VALIS provides state-of-the-art accuracy in WSI registration and 3D reconstruction. Leveraging existing open-source software tools, VALIS is written in Python, providing a free, fast, scalable, robust, and easy-to-use pipeline for registering multi-gigapixel WSI, facilitating downstream spatial analyses.

Intra-tumor heterogeneity, turnover rate and karyotype space shape susceptibility to missegregation-induced extinction.

The phenotypic efficacy of somatic copy number alterations (SCNAs) stems from their incidence per base pair of the genome, which is orders of magnitudes greater than that of point mutations. One mitotic event stands out in its potential to significantly change a cell's SCNA burden-a chromosome missegregation. A stochastic model of chromosome mis-segregations has been previously developed to describe the evolution of SCNAs of a single chromosome type. Building upon this work, we derive a general deterministic framework for modeling missegregations of multiple chromosome types. The framework offers flexibility to model intra-tumor heterogeneity in the SCNAs of all chromosomes, as well as in missegregation- and turnover rates. The model can be used to test how selection acts upon coexisting karyotypes over hundreds of generations. We use the model to calculate missegregation-induced population extinction (MIE) curves, that separate viable from non-viable populations as a function of their turnover- and missegregation rates. Turnover- and missegregation rates estimated from scRNA-seq data are then compared to theoretical predictions. We find convergence of theoretical and empirical results in both the location of MIE curves and the necessary conditions for MIE. When a dependency of missegregation rate on karyotype is introduced, karyotypes associated with low missegregation rates act as a stabilizing refuge, rendering MIE impossible unless turnover rates are exceedingly high. Intra-tumor heterogeneity, including heterogeneity in missegregation rates, increases as tumors progress, rendering MIE unlikely.

A deep learning model for molecular label transfer that enables cancer cell identification from histopathology images.

Deep-learning classification systems have the potential to improve cancer diagnosis. However, development of these computational approaches so far depends on prior pathological annotations and large training datasets. The manual annotation is low-resolution, time-consuming, highly variable and subject to observer variance. To address this issue, we developed a method, H&E Molecular neural network (HEMnet). HEMnet utilizes immunohistochemistry as an initial molecular label for cancer cells on a H&E image and trains a cancer classifier on the overlapping clinical histopathological images. Using this molecular transfer method, HEMnet successfully generated and labeled 21,939 tumor and 8782 normal tiles from ten whole-slide images for model training. After building the model, HEMnet accurately identified colorectal cancer regions, which achieved 0.84 and 0.73 of ROC AUC values compared to p53 staining and pathological annotations, respectively. Our validation study using histopathology images from TCGA samples accurately estimated tumor purity, which showed a significant correlation (regression coefficient of 0.8) with the estimation based on genomic sequencing data. Thus, HEMnet contributes to addressing two main challenges in cancer deep-learning analysis, namely the need to have a large number of images for training and the dependence on manual labeling by a pathologist. HEMnet also predicts cancer cells at a much higher resolution compared to manual histopathologic evaluation. Overall, our method provides a path towards a fully automated delineation of any type of tumor so long as there is a cancer-oriented molecular stain available for subsequent learning. Software, tutorials and interactive tools are available at: https://github.com/BiomedicalMachineLearning/HEMnet.

Tipping Cancer Cells Over the Edge: The Context-Dependent Cost of High Ploidy.

Tetraploidy is an aneuploidy-permissive condition that can fuel tumorgenesis. The tip-over hypothesis of cytotoxic therapy sensitivity proposes that therapy is effective if it pushes a cell's aneuploidy above a viable tipping point. But elevated aneuploidy alone may not account for this tipping point. Tissue microenvironments that lack sufficient resources to support tetraploid cells can explain the fitness cost of aneuploidy. Raw materials needed to generate deoxynucleotides, the building blocks of DNA, are candidate rate-limiting factors for the evolution of high-ploidy cancer cells. Understanding the resource cost of high ploidy is key to uncover its therapeutic vulnerabilities across tissue sites with versatile energy supplies.

Applications of Topological Data Analysis in Oncology.

The emergence of the information age in the last few decades brought with it an explosion of biomedical data. But with great power comes great responsibility: there is now a pressing need for new data analysis algorithms to be developed to make sense of the data and transform this information into knowledge which can be directly translated into the clinic. Topological data analysis (TDA) provides a promising path forward: using tools from the mathematical field of algebraic topology, TDA provides a framework to extract insights into the often high-dimensional, incomplete, and noisy nature of biomedical data. Nowhere is this more evident than in the field of oncology, where patient-specific data is routinely presented to clinicians in a variety of forms, from imaging to single cell genomic sequencing. In this review, we focus on applications involving persistent homology, one of the main tools of TDA. We describe some recent successes of TDA in oncology, specifically in predicting treatment responses and prognosis, tumor segmentation and computer-aided diagnosis, disease classification, and cellular architecture determination. We also provide suggestions on avenues for future research including utilizing TDA to analyze cancer time-series data such as gene expression changes during pathogenesis, investigation of the relation between angiogenic vessel structure and treatment efficacy from imaging data, and experimental confirmation that geometric and topological connectivity implies functional connectivity in the context of cancer.

Integrating Mathematical Modeling with High-Throughput Imaging Explains How Polyploid Populations Behave in Nutrient-Sparse Environments.

Breast cancer progresses in a multistep process from primary tumor growth and stroma invasion to metastasis. Nutrient-limiting environments promote chemotaxis with aggressive morphologies characteristic of invasion. It is unknown how coexisting cells differ in their response to nutrient limitations and how this impacts invasion of the metapopulation as a whole. In this study, we integrate mathematical modeling with microenvironmental perturbation data to investigate invasion in nutrient-limiting environments inhabited by one or two cancer cell subpopulations. Subpopulations were defined by their energy efficiency and chemotactic ability. Invasion distance traveled by a homogeneous population was estimated. For heterogeneous populations, results suggest that an imbalance between nutrient efficacy and chemotactic superiority accelerates invasion. Such imbalance will spatially segregate the two populations and only one type will dominate at the invasion front. Only if these two phenotypes are balanced, the two subpopulations compete for the same space, which decelerates invasion. We investigate ploidy as a candidate biomarker of this phenotypic heterogeneity and discuss its potential to inform the dose of mTOR inhibitors (mTOR-I) that can inhibit chemotaxis just enough to facilitate such competition. SIGNIFICANCE: This study identifies the double-edged sword of high ploidy as a prerequisite to personalize combination therapies with cytotoxic drugs and inhibitors of signal transduction pathways such as mTOR-Is. GRAPHICAL ABSTRACT: http://cancerres.aacrjournals.org/content/canres/80/22/5109/F1.large.jpg.

High School Internship Program in Integrated Mathematical Oncology (HIP IMO): Five-Year Experience at Moffitt Cancer Center.

Modern cancer research, and the wealth of data across multiple spatial and temporal scales, has created the need for researchers that are well versed in the life sciences (cancer biology, developmental biology, immunology), medical sciences (oncology) and natural sciences (mathematics, physics, engineering, computer sciences). College undergraduate education traditionally occurs in disciplinary silos, which creates a steep learning curve at the graduate and postdoctoral levels that increasingly bridge multiple disciplines. Numerous colleges have begun to embrace interdisciplinary curricula, but students who double major in mathematics (or other quantitative sciences) and biology (or medicine) remain scarce. We identified the need to educate junior and senior high school students about integrating mathematical and biological skills, through the lens of mathematical oncology, to better prepare students for future careers at the interdisciplinary interface. The High school Internship Program in Integrated Mathematical Oncology (HIP IMO) at Moffitt Cancer Center has so far trained 59 students between 2015 and 2019. We report here on the program structure, training deliverables, curriculum and outcomes. We hope to promote interdisciplinary educational activities early in a student's career.

Joint single cell DNA-seq and RNA-seq of gastric cancer cell lines reveals rules of in vitro evolution.

Cancer cell lines are not homogeneous nor are they static in their genetic state and biological properties. Genetic, transcriptional and phenotypic diversity within cell lines contributes to the lack of experimental reproducibility frequently observed in tissue-culture-based studies. While cancer cell line heterogeneity has been generally recognized, there are no studies which quantify the number of clones that coexist within cell lines and their distinguishing characteristics. We used a single-cell DNA sequencing approach to characterize the cellular diversity within nine gastric cancer cell lines and integrated this information with single-cell RNA sequencing. Overall, we sequenced the genomes of 8824 cells, identifying between 2 and 12 clones per cell line. Using the transcriptomes of more than 28 000 single cells from the same cell lines, we independently corroborated 88% of the clonal structure determined from single cell DNA analysis. For one of these cell lines, we identified cell surface markers that distinguished two subpopulations and used flow cytometry to sort these two clones. We identified substantial proportions of replicating cells in each cell line, assigned these cells to subclones detected among the G0/G1 population and used the proportion of replicating cells per subclone as a surrogate of each subclone's growth rate.

CD81 is a novel immunotherapeutic target for B cell lymphoma.

The tetraspanin CD81 was initially discovered by screening mAbs elicited against a human B cell lymphoma for their direct antiproliferative effects. We now show that 5A6, one of the mAbs that target CD81, has therapeutic potential. This antibody inhibits the growth of B cell lymphoma in a xenograft model as effectively as rituximab, which is a standard treatment for B cell lymphoma. Importantly, unlike rituximab, which depletes normal as well as malignant B cells, 5A6 selectively kills human lymphoma cells from fresh biopsy specimens while sparing the normal lymphoid cells in the tumor microenvironment. The 5A6 antibody showed a good safety profile when administered to a mouse transgenic for human CD81. Taken together, these data provide the rationale for the development of the 5A6 mAb and its humanized derivatives as a novel treatment against B cell lymphoma.

Single-cell transcriptome analysis identifies distinct cell types and niche signaling in a primary gastric organoid model.

The diverse cellular milieu of the gastric tissue microenvironment plays a critical role in normal tissue homeostasis and tumor development. However, few cell culture model can recapitulate the tissue microenvironment and intercellular signaling in vitro. We used a primary tissue culture system to generate a murine p53 null gastric tissue model containing both epithelium and mesenchymal stroma. To characterize the microenvironment and niche signaling, we used single cell RNA sequencing (scRNA-Seq) to determine the transcriptomes of 4,391 individual cells. Based on specific markers, we identified epithelial cells, fibroblasts and macrophages in initial tissue explants during organoid formation. The majority of macrophages were polarized towards wound healing and tumor promotion M2-type. During the course of time, the organoids maintained both epithelial and fibroblast lineages with the features of immature mouse gastric stomach. We detected a subset of cells in both lineages expressing Lgr5, one of the stem cell markers. We examined the lineage-specific Wnt signaling activation, and identified that Rspo3 was specifically expressed in the fibroblast lineage, providing an endogenous source of the R-spondin to activate Wnt signaling. Our studies demonstrate that this primary tissue culture system enables one to study gastric tissue niche signaling and immune response in vitro.

Exploring the spatiotemporal genetic heterogeneity in metastatic lung adenocarcinoma using a nuclei flow-sorting approach.

Variable tumor cellularity can limit sensitivity and precision in comparative genomics because differences in tumor content can result in misclassifying truncal mutations as region-specific private mutations in stroma-rich regions, especially when studying tissue specimens of mediocre tumor cellularity such as lung adenocarcinomas (LUADs). To address this issue, we refined a nuclei flow-sorting approach by sorting nuclei based on ploidy and the LUAD lineage marker thyroid transcription factor 1 and applied this method to investigate genome-wide somatic copy number aberrations (SCNAs) and mutations of 409 cancer genes in 39 tumor populations obtained from 16 primary tumors and 21 matched metastases. This approach increased the mean tumor purity from 54% (range 7-89%) of unsorted material to 92% (range 79-99%) after sorting. Despite this rise in tumor purity, we detected limited genetic heterogeneity between primary tumors and their metastases. In fact, 88% of SCNAs and 80% of mutations were propagated from primary tumors to metastases and low allele frequency mutations accounted for much of the mutational heterogeneity. Even though the presence of SCNAs indicated a history of chromosomal instability (CIN) in all tumors, metastases did not have more SCNAs than primary tumors. Moreover, tumors with biallelic TP53 or ATM mutations had high numbers of SCNAs, yet they were associated with a low interlesional genetic heterogeneity. The results of our study thus provide evidence that most macroevolutionary events occur in primary tumors before metastatic dissemination and advocate for a limited degree of CIN over time and space in this cohort of LUADs. Sampling of primary tumors thus may suffice to detect most mutations and SCNAs. In addition, metastases but not primary tumors had seeded additional metastases in three of four patients; this provides a genomic rational for surgical treatment of such oligometastatic LUADs. Copyright © 2018 Pathological Society of Great Britain and Ireland. Published by John Wiley & Sons, Ltd.

Single-cell RNA-Seq of follicular lymphoma reveals malignant B-cell types and coexpression of T-cell immune checkpoints.

Follicular lymphoma (FL) is a low-grade B-cell malignancy that transforms into a highly aggressive and lethal disease at a rate of 2% per year. Perfect isolation of the malignant B-cell population from a surgical biopsy is a significant challenge, masking important FL biology, such as immune checkpoint coexpression patterns. To resolve the underlying transcriptional networks of follicular B-cell lymphomas, we analyzed the transcriptomes of 34 188 cells derived from 6 primary FL tumors. For each tumor, we identified normal immune subpopulations and malignant B cells, based on gene expression. We used multicolor flow cytometry analysis of the same tumors to confirm our assignments of cellular lineages and validate our predictions of expressed proteins. Comparison of gene expression between matched malignant and normal B cells from the same patient revealed tumor-specific features. Malignant B cells exhibited restricted immunoglobulin (Ig) light chain expression (either Igκ or Igλ), as well the expected upregulation of the BCL2 gene, but also downregulation of the FCER2, CD52, and major histocompatibility complex class II genes. By analyzing thousands of individual cells per patient tumor, we identified the mosaic of malignant B-cell subclones that coexist within a FL and examined the characteristics of tumor-infiltrating T cells. We identified genes coexpressed with immune checkpoint molecules, such as CEBPA and B2M in regulatory T (Treg) cells, providing a better understanding of the gene networks involved in immune regulation. In summary, parallel measurement of single-cell expression in thousands of tumor cells and tumor-infiltrating lymphocytes can be used to obtain a systems-level view of the tumor microenvironment and identify new avenues for therapeutic development.

Lgl1 controls NG2 endocytic pathway to regulate oligodendrocyte differentiation and asymmetric cell division and gliomagenesis.

Oligodendrocyte progenitor cells (OPC) undergo asymmetric cell division (ACD) to generate one OPC and one differentiating oligodendrocyte (OL) progeny. Loss of pro-mitotic proteoglycan and OPC marker NG2 in the OL progeny is the earliest immunophenotypic change of unknown mechanism that indicates differentiation commitment. Here, we report that expression of the mouse homolog of Drosophila tumor suppressor Lethal giant larvae 1 (Lgl1) is induced during OL differentiation. Lgl1 conditional knockout OPC progeny retain NG2 and show reduced OL differentiation, while undergoing more symmetric self-renewing divisions at the expense of asymmetric divisions. Moreover, Lgl1 and hemizygous Ink4a/Arf knockouts in OPC synergistically induce gliomagenesis. Time lapse and total internal reflection microscopy reveals a critical role for Lgl1 in NG2 endocytic routing and links aberrant NG2 recycling to failed differentiation. These data establish Lgl1 as a suppressor of gliomagenesis and positive regulator of asymmetric division and differentiation in the healthy and demyelinated murine brain.

SVEngine: an efficient and versatile simulator of genome structural variations with features of cancer clonal evolution.

Background: Simulating genome sequence data with variant features facilitates the development and benchmarking of structural variant analysis programs. However, there are only a few data simulators that provide structural variants in silico and even fewer that provide variants with different allelic fraction and haplotypes. Findings: We developed SVEngine, an open-source tool to address this need. SVEngine simulates next-generation sequencing data with embedded structural variations. As input, SVEngine takes template haploid sequences (FASTA) and an external variant file, a variant distribution file, and/or a clonal phylogeny tree file (NEWICK) as input. Subsequently, it simulates and outputs sequence contigs (FASTAs), sequence reads (FASTQs), and/or post-alignment files (BAMs). All of the files contain the desired variants, along with BED files containing the ground truth. SVEngine's flexible design process enables one to specify size, position, and allelic fraction for deletions, insertions, duplications, inversions, and translocations. Finally, SVEngine simulates sequence data that replicate the characteristics of a sequencing library with mixed sizes of DNA insert molecules. To improve the compute speed, SVEngine is highly parallelized to reduce the simulation time. Conclusions: We demonstrated the versatile features of SVEngine and its improved runtime comparisons with other available simulators. SVEngine's features include the simulation of locus-specific variant frequency designed to mimic the phylogeny of cancer clonal evolution. We validated SVEngine's accuracy by simulating genome-wide structural variants of NA12878 and a heterogeneous cancer genome. Our evaluation included checking various sequencing mapping features such as coverage change, read clipping, insert size shift, and neighboring hanging read pairs for representative variant types. Structural variant callers Lumpy and Manta and tumor heterogeneity estimator THetA2 were able to perform realistically on the simulated data. SVEngine is implemented as a standard Python package and is freely available for academic use .

Genomic Instability in Cancer: Teetering on the Limit of Tolerance.

Cancer genomic instability contributes to the phenomenon of intratumoral genetic heterogeneity, provides the genetic diversity required for natural selection, and enables the extensive phenotypic diversity that is frequently observed among patients. Genomic instability has previously been associated with poor prognosis. However, we have evidence that for solid tumors of epithelial origin, extreme levels of genomic instability, where more than 75% of the genome is subject to somatic copy number alterations, are associated with a potentially better prognosis compared with intermediate levels under this threshold. This has been observed in clonal subpopulations of larger size, especially when genomic instability is shared among a limited number of clones. We hypothesize that cancers with extreme levels of genomic instability may be teetering on the brink of a threshold where so much of their genome is adversely altered that cells rarely replicate successfully. Another possibility is that tumors with high levels of genomic instability are more immunogenic than other cancers with a less extensive burden of genetic aberrations. Regardless of the exact mechanism, but hinging on our ability to quantify how a tumor's burden of genetic aberrations is distributed among coexisting clones, genomic instability has important therapeutic implications. Herein, we explore the possibility that a high genomic instability could be the basis for a tumor's sensitivity to DNA-damaging therapies. We primarily focus on studies of epithelial-derived solid tumors. Cancer Res; 77(9); 2179-85. ©2017 AACR.

Pan-cancer analysis of the extent and consequences of intratumor heterogeneity.

Intratumor heterogeneity (ITH) drives neoplastic progression and therapeutic resistance. We used the bioinformatics tools 'expanding ploidy and allele frequency on nested subpopulations' (EXPANDS) and PyClone to detect clones that are present at a ≥10% frequency in 1,165 exome sequences from tumors in The Cancer Genome Atlas. 86% of tumors across 12 cancer types had at least two clones. ITH in the morphology of nuclei was associated with genetic ITH (Spearman's correlation coefficient, ρ = 0.24-0.41; P < 0.001). Mutation of a driver gene that typically appears in smaller clones was a survival risk factor (hazard ratio (HR) = 2.15, 95% confidence interval (CI): 1.71-2.69). The risk of mortality also increased when >2 clones coexisted in the same tumor sample (HR = 1.49, 95% CI: 1.20-1.87). In two independent data sets, copy-number alterations affecting either <25% or >75% of a tumor's genome predicted reduced risk (HR = 0.15, 95% CI: 0.08-0.29). Mortality risk also declined when >4 clones coexisted in the sample, suggesting a trade-off between the costs and benefits of genomic instability. ITH and genomic instability thus have the potential to be useful measures that can universally be applied to all cancers.

Pharmacologic inhibition of histone demethylation as a therapy for pediatric brainstem glioma.

Pediatric brainstem gliomas often harbor oncogenic K27M mutation of histone H3.3. Here we show that GSKJ4 pharmacologic inhibition of K27 demethylase JMJD3 increases cellular H3K27 methylation in K27M tumor cells and demonstrate potent antitumor activity both in vitro against K27M cells and in vivo against K27M xenografts. Our results demonstrate that increasing H3K27 methylation by inhibiting K27 demethylase is a valid therapeutic strategy for treating K27M-expressing brainstem glioma.

EXPANDS: expanding ploidy and allele frequency on nested subpopulations.

MOTIVATION: Several cancer types consist of multiple genetically and phenotypically distinct subpopulations. The underlying mechanism for this intra-tumoral heterogeneity can be explained by the clonal evolution model, whereby growth advantageous mutations cause the expansion of cancer cell subclones. The recurrent phenotype of many cancers may be a consequence of these coexisting subpopulations responding unequally to therapies. Methods to computationally infer tumor evolution and subpopulation diversity are emerging and they hold the promise to improve the understanding of genetic and molecular determinants of recurrence. RESULTS: To address cellular subpopulation dynamics within human tumors, we developed a bioinformatic method, EXPANDS. It estimates the proportion of cells harboring specific mutations in a tumor. By modeling cellular frequencies as probability distributions, EXPANDS predicts mutations that accumulate in a cell before its clonal expansion. We assessed the performance of EXPANDS on one whole genome sequenced breast cancer and performed SP analyses on 118 glioblastoma multiforme samples obtained from TCGA. Our results inform about the extent of subclonal diversity in primary glioblastoma, subpopulation dynamics during recurrence and provide a set of candidate genes mutated in the most well-adapted subpopulations. In summary, EXPANDS predicts tumor purity and subclonal composition from sequencing data. AVAILABILITY AND IMPLEMENTATION: EXPANDS is available for download at http://code.google.com/p/expands (matlab version--used in this manuscript) and http://cran.r-project.org/web/packages/expands (R version).