Naturally occurring combinations of receptors from single cell transcriptomics in endothelial cells.

VEGF inhibitor drugs are part of standard care in oncology and ophthalmology, but not all patients respond to them. Combinations of drugs are likely to be needed for more effective therapies of angiogenesis-related diseases. In this paper we describe naturally occurring combinations of receptors in endothelial cells that might help to understand how cells communicate and to identify targets for drug combinations. We also develop and share a new software tool called DECNEO to identify them. Single-cell gene expression data are used to identify a set of co-expressed endothelial cell receptors, conserved among species (mice and humans) and enriched, within a network, of connections to up-regulated genes. This set includes several receptors previously shown to play a role in angiogenesis. Multiple statistical tests from large datasets, including an independent validation set, support the reproducibility, evolutionary conservation and role in angiogenesis of these naturally occurring combinations of receptors. We also show tissue-specific combinations and, in the case of choroid endothelial cells, consistency with both well-established and recent experimental findings, presented in a separate paper. The results and methods presented here advance the understanding of signaling to endothelial cells. The methods are generally applicable to the decoding of intercellular combinations of signals.

LIF, a mitogen for choroidal endothelial cells, protects the choriocapillaris: implications for prevention of geographic atrophy.

In the course of our studies aiming to discover vascular bed-specific endothelial cell (EC) mitogens, we identified leukemia inhibitory factor (LIF) as a mitogen for bovine choroidal EC (BCE), although LIF has been mainly characterized as an EC growth inhibitor and an anti-angiogenic molecule. LIF stimulated growth of BCE while it inhibited, as previously reported, bovine aortic EC (BAE) growth. The JAK-STAT3 pathway mediated LIF actions in both BCE and BAE cells, but a caspase-independent proapoptotic signal mediated by cathepsins was triggered in BAE but not in BCE. LIF administration directly promoted activation of STAT3 and increased blood vessel density in mouse eyes. LIF also had protective effects on the choriocapillaris in a model of oxidative retinal injury. Analysis of available single-cell transcriptomic datasets shows strong expression of the specific LIF receptor in mouse and human choroidal EC. Our data suggest that LIF administration may be an innovative approach to prevent atrophy associated with AMD, through protection of the choriocapillaris.

Digital Cell Sorter (DCS): a cell type identification, anomaly detection, and Hopfield landscapes toolkit for single-cell transcriptomics.

MOTIVATION: Analysis of singe cell RNA sequencing (scRNA-seq) typically consists of different steps including quality control, batch correction, clustering, cell identification and characterization, and visualization. The amount of scRNA-seq data is growing extremely fast, and novel algorithmic approaches improving these steps are key to extract more biological information. Here, we introduce: (i) two methods for automatic cell type identification (i.e., without expert curator) based on a voting algorithm and a Hopfield classifier, (ii) a method for cell anomaly quantification based on isolation forest, and (iii) a tool for the visualization of cell phenotypic landscapes based on Hopfield energy-like functions. These new approaches are integrated in a software platform that includes many other state-of-the-art methodologies and provides a self-contained toolkit for scRNA-seq analysis. RESULTS: We present a suite of software elements for the analysis of scRNA-seq data. This Python-based open source software, Digital Cell Sorter (DCS), consists in an extensive toolkit of methods for scRNA-seq analysis. We illustrate the capability of the software using data from large datasets of peripheral blood mononuclear cells (PBMC), as well as plasma cells of bone marrow samples from healthy donors and multiple myeloma patients. We test the novel algorithms by evaluating their ability to deconvolve cell mixtures and detect small numbers of anomalous cells in PBMC data. AVAILABILITY: The DCS toolkit is available for download and installation through the Python Package Index (PyPI). The software can be deployed using the Python import function following installation. Source code is also available for download on Zenodo: DOI 10.5281/zenodo.2533377. SUPPLEMENTARY INFORMATION: Supplemental Materials are available at PeerJ online.

Polled Digital Cell Sorter (p-DCS): Automatic identification of hematological cell types from single cell RNA-sequencing clusters.

BACKGROUND: Single cell RNA sequencing (scRNA-seq) brings unprecedented opportunities for mapping the heterogeneity of complex cellular environments such as bone marrow, and provides insight into many cellular processes. Single cell RNA-seq has a far larger fraction of missing data reported as zeros (dropouts) than traditional bulk RNA-seq, and unsupervised clustering combined with Principal Component Analysis (PCA) can be used to overcome this limitation. After clustering, however, one has to interpret the average expression of markers on each cluster to identify the corresponding cell types, and this is normally done by hand by an expert curator. RESULTS: We present a computational tool for processing single cell RNA-seq data that uses a voting algorithm to automatically identify cells based on approval votes received by known molecular markers. Using a stochastic procedure that accounts for imbalances in the number of known molecular signatures for different cell types, the method computes the statistical significance of the final approval score and automatically assigns a cell type to clusters without an expert curator. We demonstrate the utility of the tool in the analysis of eight samples of bone marrow from the Human Cell Atlas. The tool provides a systematic identification of cell types in bone marrow based on a list of markers of immune cell types, and incorporates a suite of visualization tools that can be overlaid on a t-SNE representation. The software is freely available as a Python package at https://github.com/sdomanskyi/DigitalCellSorter . CONCLUSIONS: This methodology assures that extensive marker to cell type matching information is taken into account in a systematic way when assigning cell clusters to cell types. Moreover, the method allows for a high throughput processing of multiple scRNA-seq datasets, since it does not involve an expert curator, and it can be applied recursively to obtain cell sub-types. The software is designed to allow the user to substitute the marker to cell type matching information and apply the methodology to different cellular environments.

Modeling disease progression in Multiple Myeloma with Hopfield networks and single-cell RNA-seq.

Associative memories in Hopfield's neural networks are mapped to gene expression pattern to model different paths of disease progression towards Multiple Myeloma (MM). The model is built using single cell RNA-seq data from bone marrow aspirates of MM patients as well as patients diagnosed with Monoclonal Gammopathy of Undetermined Significance (MGUS) and Smoldering Multiple Myeloma (SMM), two medical conditions that often progress to full MM. Results: We identify different clusters of MGUS, SMM, and MM cells, map them to Hopfield associative memory patterns, and model the dynamics of transition between the different patterns. The model is then used to identify genes that are differentialy expressed across different MM stages and whose simultaneous inhibition is associated to a delayed disease progression.

Cell cycle time series gene expression data encoded as cyclic attractors in Hopfield systems.

Modern time series gene expression and other omics data sets have enabled unprecedented resolution of the dynamics of cellular processes such as cell cycle and response to pharmaceutical compounds. In anticipation of the proliferation of time series data sets in the near future, we use the Hopfield model, a recurrent neural network based on spin glasses, to model the dynamics of cell cycle in HeLa (human cervical cancer) and S. cerevisiae cells. We study some of the rich dynamical properties of these cyclic Hopfield systems, including the ability of populations of simulated cells to recreate experimental expression data and the effects of noise on the dynamics. Next, we use a genetic algorithm to identify sets of genes which, when selectively inhibited by local external fields representing gene silencing compounds such as kinase inhibitors, disrupt the encoded cell cycle. We find, for example, that inhibiting the set of four kinases AURKB, NEK1, TTK, and WEE1 causes simulated HeLa cells to accumulate in the M phase. Finally, we suggest possible improvements and extensions to our model.

Evolutionary and Topological Properties of Genes and Community Structures in Human Gene Regulatory Networks.

The diverse, specialized genes present in today's lifeforms evolved from a common core of ancient, elementary genes. However, these genes did not evolve individually: gene expression is controlled by a complex network of interactions, and alterations in one gene may drive reciprocal changes in its proteins' binding partners. Like many complex networks, these gene regulatory networks (GRNs) are composed of communities, or clusters of genes with relatively high connectivity. A deep understanding of the relationship between the evolutionary history of single genes and the topological properties of the underlying GRN is integral to evolutionary genetics. Here, we show that the topological properties of an acute myeloid leukemia GRN and a general human GRN are strongly coupled with its genes' evolutionary properties. Slowly evolving ("cold"), old genes tend to interact with each other, as do rapidly evolving ("hot"), young genes. This naturally causes genes to segregate into community structures with relatively homogeneous evolutionary histories. We argue that gene duplication placed old, cold genes and communities at the center of the networks, and young, hot genes and communities at the periphery. We demonstrate this with single-node centrality measures and two new measures of efficiency, the set efficiency and the interset efficiency. We conclude that these methods for studying the relationships between a GRN's community structures and its genes' evolutionary properties provide new perspectives for understanding evolutionary genetics.

A scalable method for molecular network reconstruction identifies properties of targets and mutations in acute myeloid leukemia.

A key aim of systems biology is the reconstruction of molecular networks. We do not yet, however, have networks that integrate information from all datasets available for a particular clinical condition. This is in part due to the limited scalability, in terms of required computational time and power, of existing algorithms. Network reconstruction methods should also be scalable in the sense of allowing scientists from different backgrounds to efficiently integrate additional data. We present a network model of acute myeloid leukemia (AML). In the current version (AML 2.1), we have used gene expression data (both microarray and RNA-seq) from 5 different studies comprising a total of 771 AML samples and a protein-protein interactions dataset. Our scalable network reconstruction method is in part based on the well-known property of gene expression correlation among interacting molecules. The difficulty of distinguishing between direct and indirect interactions is addressed by optimizing the coefficient of variation of gene expression, using a validated gold-standard dataset of direct interactions. Computational time is much reduced compared to other network reconstruction methods. A key feature is the study of the reproducibility of interactions found in independent clinical datasets. An analysis of the most significant clusters, and of the network properties (intraset efficiency, degree, betweenness centrality, and PageRank) of common AML mutations demonstrated the biological significance of the network. A statistical analysis of the response of blast cells from 11 AML patients to a library of kinase inhibitors provided an experimental validation of the network. A combination of network and experimental data identified CDK1, CDK2, CDK4, and CDK6 and other kinases as potential therapeutic targets in AML.

Control of asymmetric Hopfield networks and application to cancer attractors.

The asymmetric Hopfield model is used to simulate signaling dynamics in gene regulatory networks. The model allows for a direct mapping of a gene expression pattern into attractor states. We analyze different control strategies aimed at disrupting attractor patterns using selective local fields representing therapeutic interventions. The control strategies are based on the identification of signaling bottlenecks, which are single nodes or strongly connected clusters of nodes that have a large impact on the signaling. We provide a theorem with bounds on the minimum number of nodes that guarantee control of bottlenecks consisting of strongly connected components. The control strategies are applied to the identification of sets of proteins that, when inhibited, selectively disrupt the signaling of cancer cells while preserving the signaling of normal cells. We use an experimentally validated non-specific and an algorithmically-assembled specific B cell gene regulatory network reconstructed from gene expression data to model cancer signaling in lung and B cells, respectively. Among the potential targets identified here are TP53, FOXM1, BCL6 and SRC. This model could help in the rational design of novel robust therapeutic interventions based on our increasing knowledge of complex gene signaling networks.

Identification of drug combinations containing imatinib for treatment of BCR-ABL+ leukemias.

The BCR-ABL translocation is found in chronic myeloid leukemia (CML) and in Ph+ acute lymphoblastic leukemia (ALL) patients. Although imatinib and its analogues have been used as front-line therapy to target this mutation and control the disease for over a decade, resistance to the therapy is still observed and most patients are not cured but need to continue the therapy indefinitely. It is therefore of great importance to find new therapies, possibly as drug combinations, which can overcome drug resistance. In this study, we identified eleven candidate anti-leukemic drugs that might be combined with imatinib, using three approaches: a kinase inhibitor library screen, a gene expression correlation analysis, and literature analysis. We then used an experimental search algorithm to efficiently explore the large space of possible drug and dose combinations and identified drug combinations that selectively kill a BCR-ABL+ leukemic cell line (K562) over a normal fibroblast cell line (IMR-90). Only six iterations of the algorithm were needed to identify very selective drug combinations. The efficacy of the top forty-nine combinations was further confirmed using Ph+ and Ph- ALL patient cells, including imatinib-resistant cells. Collectively, the drug combinations and methods we describe might be a first step towards more effective interventions for leukemia patients, especially those with the BCR-ABL translocation.

Prediction of kinase inhibitor response using activity profiling, in vitro screening, and elastic net regression.

BACKGROUND: Many kinase inhibitors have been approved as cancer therapies. Recently, libraries of kinase inhibitors have been extensively profiled, thus providing a map of the strength of action of each compound on a large number of its targets. These profiled libraries define drug-kinase networks that can predict the effectiveness of untested drugs and elucidate the roles of specific kinases in different cellular systems. Predictions of drug effectiveness based on a comprehensive network model of cellular signalling are difficult, due to our partial knowledge of the complex biological processes downstream of the targeted kinases. RESULTS: We have developed the Kinase Inhibitors Elastic Net (KIEN) method, which integrates information contained in drug-kinase networks with in vitro screening. The method uses the in vitro cell response of single drugs and drug pair combinations as a training set to build linear and nonlinear regression models. Besides predicting the effectiveness of untested drugs, the KIEN method identifies sets of kinases that are statistically associated to drug sensitivity in a given cell line. We compared different versions of the method, which is based on a regression technique known as elastic net. Data from two-drug combinations led to predictive models, and we found that predictivity can be improved by applying logarithmic transformation to the data. The method was applied to the A549 lung cancer cell line, and we identified specific kinases known to have an important role in this type of cancer (TGFBR2, EGFR, PHKG1 and CDK4). A pathway enrichment analysis of the set of kinases identified by the method showed that axon guidance, activation of Rac, and semaphorin interactions pathways are associated to a selective response to therapeutic intervention in this cell line. CONCLUSIONS: We have proposed an integrated experimental and computational methodology, called KIEN, that identifies the role of specific kinases in the drug response of a given cell line. The method will facilitate the design of new kinase inhibitors and the development of therapeutic interventions with combinations of many inhibitors.

Metabolomics of the tumor microenvironment in pediatric acute lymphoblastic leukemia.

The tumor microenvironment is emerging as an important therapeutic target. Most studies, however, are focused on the protein components, and relatively little is known of how the microenvironmental metabolome might influence tumor survival. In this study, we examined the metabolic profiles of paired bone marrow (BM) and peripheral blood (PB) samples from 10 children with acute lymphoblastic leukemia (ALL). BM and PB samples from the same patient were collected at the time of diagnosis and after 29 days of induction therapy, at which point all patients were in remission. We employed two analytical platforms, high-resolution magnetic resonance spectroscopy and gas chromatography-mass spectrometry, to identify and quantify 102 metabolites in the BM and PB. Standard ALL therapy, which includes l-asparaginase, completely removed circulating asparagine, but not glutamine. Statistical analyses of metabolite correlations and network reconstructions showed that the untreated BM microenvironment was characterized by a significant network-level signature: a cluster of highly correlated lipids and metabolites involved in lipid metabolism (p<0.006). In contrast, the strongest correlations in the BM upon remission were observed among amino acid metabolites and derivatives (p<9.2 × 10(-10)). This study provides evidence that metabolic characterization of the cancer niche could generate new hypotheses for the development of cancer therapies.

Selective control of the apoptosis signaling network in heterogeneous cell populations.

BACKGROUND: Selective control in a population is the ability to control a member of the population while leaving the other members relatively unaffected. The concept of selective control is developed using cell death or apoptosis in heterogeneous cell populations as an example. Control of apoptosis is essential in a variety of therapeutic environments, including cancer where cancer cell death is a desired outcome and Alzheimer's disease where neuron survival is the desired outcome. However, in both cases these responses must occur with minimal response in other cells exposed to treatment; that is, the response must be selective. METHODOLOGY AND PRINCIPAL FINDINGS: Apoptosis signaling in heterogeneous cells is described by an ensemble of gene networks with identical topology but different link strengths. Selective control depends on the statistics of signaling in the ensemble of networks, and we analyze the effects of superposition, non-linearity and feedback on these statistics. Parallel pathways promote normal statistics while series pathways promote skew distributions, which in the most extreme cases become log-normal. We also show that feedback and non-linearity can produce bimodal signaling statistics, as can discreteness and non-linearity. Two methods for optimizing selective control are presented. The first is an exhaustive search method and the second is a linear programming based approach. Though control of a single gene in the signaling network yields little selectivity, control of a few genes typically yields higher levels of selectivity. The statistics of gene combinations susceptible to selective control in heterogeneous apoptosis networks is studied and is used to identify general control strategies. CONCLUSIONS AND SIGNIFICANCE: We have explored two methods for the study of selectivity in cell populations. The first is an exhaustive search method limited to three node perturbations. The second is an effective linear model, based on interpolation of single node sensitivity, in which the selective combinations can be found by linear programming optimization. We found that selectivity is promoted by acting on the least sensitive nodes in the case of weak populations, while selective control of robust populations is optimized through perturbations of more sensitive nodes. High throughput experiments with heterogeneous cell lines could be designed in an analogous manner, with the further possibility of incorporating the selectivity optimization process into a closed-loop control system.