Biology subjects

Aittokallio, T.

Publications and source records attributed to Aittokallio, T..

At least 37 records · Page 2Linked to original sources

bioRxiv2024

Breast Cancer Remodels Lymphatics in Sentinel Lymph Nodes

Cancer metastasis to sentinel lymph nodes (LNs) is often the first marker of potential disease progression. Although it is recognized that tumor-induced lymphangiogenesis facilitates metastasis into LNs in murine models, tumor-induced alterations in human lymphatic vessels remain obscure. We used single-cell RNA sequencing to profile lymphatic endothelial cell (LEC) subsets in paired metastatic and non-metastatic LNs obtained from patients with treatment-naive breast cancer. Tumor metastasis decreases immunoregulatory LEC subsets, such as PD-L1+ subcapsular sinus LECs, while inducing an increase in capillary-like CD200+ HEY1+ LECs. Matrix Gla protein (MGP) was the most upregulated gene in metastatic LN LECs, and its expression on LECs was TGF-{beta} and VEGF dependent. Upregulated MGP promotes cancer cell adhesion to LN lymphatics. Thus, breast cancer cell metastasis to LNs remodels LEC subsets in human LNs and escalates MGP expression, potentially facilitating cancer cell dissemination through the lymphatic system.

immunology↗

bioRxiv2024

Scaling up drug combination surface prediction

Drug combinations are required to treat advanced cancers and other complex diseases. Compared to monotherapy, combination treatments can enhance efficacy and reduce toxicity by lowering the doses of single drugs - and there especially synergistic combinations are of interest. Since drug combination screening experiments are costly and time consuming, reliable machine learning models are needed for prioritizing potential combinations for further studies. Most of the current machine learning models are based on scalar-valued approaches, which predict individual response values or synergy scores for drug combinations. We take a functional output prediction approach, in which full, continuous dose-response combination surfaces are predicted for each drug combination on the cell lines. We investigate the predictive power of the recently proposed comboKR method, which is based on a powerful input-output kernel regression technique and functional modelling of the response surface. In this work, we develop a scaled-up formulation of the comboKR, that also implements improved modeling choices: 1) we incorporate new modeling choices for the output drug combination response surfaces to the comboKR framework, and 2) propose a projected gradient descent method to solve the challenging pre-image problem that traditionally is solved with simple candidate set approaches. We provide thorough experimental analysis of comboKR 2.0 with three real-word datasets within various challenging experimental settings, including cases where drugs or cell lines have not been encountered in the training data. Our comparison with synergy score prediction methods further highlights the relevance of dose-response prediction approaches, instead of relying on simple scoring methods.

bioinformatics↗

bioRxiv2024

Predicting drug combination response surfaces

Prediction of drug combination responses is a research question of growing importance for cancer and other complex diseases. Current machine learning approaches generally consider predicting either drug combination synergy summaries or single combination dose-response values, which fail to appropriately model the continuous nature of the underlying dose-response combination surface and can lead to inconsistencies when a synergy score or a dose-response matrix is reconstructed from separate predictions. We propose a structured prediction method, comboKR, that directly predicts the drug combination response surface for a drug combination. The method is based on a powerful input-output kernel regression technique and functional modeling of the response surface. As an important part of our approach, we develop a novel normalisation between response surfaces that standardizes the heterogeneous experimental designs used to measure the dose-responses, and thus allows training the method with data measured in different laboratories. Our experiments on two predictive scenarios highlight the suitability of the proposed approach especially in the traditionally challenging setting of predicting combination responses for new drugs not available in the training data.

bioinformatics↗

bioRxiv2024

Attention-based approach to predict drug-target interactions across seven target superfamilies

MotivationDrug-target interactions (DTIs) hold pivotal role in drug repurposing and elucidation of drug mechanisms of action. While single-targeted drugs have demonstrated clinical success, they often exhibit limited efficacy against complex diseases, such as cancers, whose development and treatment is dependent on several biological processes. Therefore, a comprehensive understanding of primary, secondary, and even inactive targets becomes essential in the quest for effective and safe treatments for cancer and other indications. The human proteome offers over a thousand druggable targets. Yet, most FDA-approved drugs bind with only a small fraction of disease targets. ResultsThis study introduces an attention-based method to predict drug-target bioactivities for all human proteins across seven superfamilies. Nine different descriptor sets were meticulously examined to identify optimal signature descriptors for predicting novel DTIs. Our testing results demonstrated Spearman correlations exceeding 0.72 (P<0.001) for six out of seven superfamilies. The proposed method outperformed nine state-of-the-art deep learning and graph-based methods, and importantly, maintained relatively high performance for most target superfamilies when tested in independent sources of bioactivity data. We computationally validated 185,676 drug-target pairs from ChEMBL-V33, that were not available in model training, with a reasonable performance of Spearman correlation > 0.57 (P<0.001) for most superfamilies. This justifies the robustness of the proposed method for predicting novel DTIs. Finally, we applied our method to predict missing activities among 3,492 approved molecules in ChEMBL-V33, offering a valuable tool for advancing drug mechanism discovery and repurposing existing drugs for new indications. Availability and implementationThe codes and training datasets to reproduce and extend the results are available at https://github.com/AronSchulman/MMAtt-DTA.

systems biology↗

bioRxiv2023

Small molecule STAT3/5 inhibitors exhibit therapeutic potential in acute myeloid leukemia and extra-nodal natural killer/T cell lymphoma

The oncogenic transcription factors STAT3, STAT5A and STAT5B are essential to steer hematopoiesis and immunity, but their enhanced expression and activation drives the development or progression of blood cancers. Current therapeutic strategies focus on blocking upstream tyrosine kinases, but frequently occurring resistance often leads to disease relapse, emphasizing the need for more targeted therapies. Here we evaluate JPX-0700 and JPX-0750, which are STAT3/5-specific covalent cysteine binders that lead to growth arrest of acute myeloid leukemia (AML) and natural killer/T cell lymphoma (NKCL) cell lines in vitro and in vivo, as well as reduce cell viability of primary AML blasts ex vivo. Our non-PROTAC small molecular weight degraders selectively reduce STAT3/5 activation and total protein levels, as well as downstream target oncogene expression, exhibiting nanomolar to low micromolar efficacy. We found that both AML and NKCL cells hijack STAT3/5 signaling through either upstream activating mutations in tyrosine kinases, activating gain-of-function mutations in STAT3, mutational loss of negative STAT regulators, or genetic gains in anti-apoptotic, pro-proliferative or epigenetic-modifying STAT3/5 targets. Moreover, we have shown synergistic inhibitory action of JPX-0700 and JPX-0750 upon combinatorial use with approved chemotherapeutics (doxorubicin, daunorubicin, cytarabine), epigenetic enzyme blocker vorinostat, tyrosine kinase inhibitor cabozantinib or BCL-2 inhibitor venetoclax. Importantly, JPX-0700 or JPX-0750 treatment reduced leukemic cell growth in human AML/NKCL xenograft mouse models without adverse side effects. These potent small molecule degraders of STAT3/5 could propel further clinical development for use in AML and NKCL patients.

cancer biology↗

bioRxiv2023

Single-cell transcriptomes identify patient-tailored therapies for selective co-inhibition of cancer clones

Intratumoral cellular heterogeneity necessitates multi-targeting therapies for improved clinical benefits in patients with advanced malignancies. However, systematic identification of patient-specific treatments that selectively co-inhibit cancerous cell populations poses a combinatorial challenge, since the number of possible drug-dose combinations vastly exceeds what could be tested in scarce patient cells. Here, we developed scTherapy, a machine learning model that leverages single-cell transcriptomic profiles to prioritize multi-targeting treatment options for individual patients with hematological cancers or solid tumors.

bioinformatics↗

bioRxiv2022

STAT5 Gain-of-Function Variants Promote Precursor T-Cell Receptor Activation to Drive T-Cell Acute Lymphoblastic Leukemia

T-cell acute lymphoblastic leukemia (T-ALL) is an aggressive immature T-cell cancer. Hotspot mutations in JAK-STAT pathway members IL7R, JAK1 and JAK3 were analyzed in depth. However, the role of STAT5A or STAT5B mutations promoting their hyperactivation is poorly understood in the context of T-cell cancer initiation and acute leukemia progression. Importantly, the driver mutation STAT5BN642H encodes the most frequent activating STAT5 variant in T-ALL associated with poor prognosis. Here, we show that hyperactive STAT5 promotes early T-cell progenitor (ETP)-ALL-like cancer in mice and upregulated genes involved in T-cell receptor signaling (TCR), even in absence of surface TCR promoting. Importantly, these genes were also overexpressed in human T-ALL and other STAT5-dependent T-cell cancers. Moreover, human T-ALL cells were sensitive to pharmacologic inhibition by dual STAT3/5 degraders or ZAP70 tyrosine kinase blockers. Thus, we define STAT5 target genes in T-ALL that promote pre-TCR signaling mimicry. We propose therapeutic targeting using selective ZAP70 or STAT3/5 inhibitors in a subgroup of T-ALL patients with prominent IL-7R-JAK1/3-STAT5 activity. SignificanceWe provide detailed functional characterizations of hyperactive STAT5A or STAT5B in thymic T-cell development and transformation. We found that hyperactive STAT5 transcribes T-cell-specific kinases or pre-TCR signaling hubs to promote T-ALL. Biomolecular and next-generation-sequencing methods, transgenesis and pharmacologic interference revealed that hyperactive STAT5 is a key oncogenic driver that can be targeted in T-ALL using STAT3/5 or SYK family member tyrosine kinase inhibitors. Conflict of interestThe authors declare no potential conflicts of interest.

cancer biology↗

bioRxiv2022

A synthetic lethal dependency on casein kinase 2 in response to replication-perturbing drugs in RB1-deficient ovarian and breast cancer cells

Treatment of patients with high-grade serous ovarian carcinoma (HGSOC) and triple-negative breast cancer (TNBC) includes platinum-based drugs, gemcitabine, and PARP inhibitors. However, resistance to these therapies develops in most cases, highlighting the need for novel therapeutic approaches and biomarkers to guide the optimal treatment choice. Using a CRISPR loss-of-function screen for carboplatin sensitizers in the HGSOC cell line OVCAR8, we identified CSNK2A2, the gene encoding for the alpha () catalytic subunit of casein kinase 2 (CK2). Expanding on this finding, we confirmed that the CK2 inhibitors silmitasertib and SGC-CK2-1 sensitized many, but not all, TNBC and HGSOC cell lines to the drugs that perturb DNA replication, including platinum drugs, gemcitabine, and PARP inhibitors. We identified RB1 tumor suppressor deficiency as a prerequisite context for the CK2 inhibition-mediated sensitization to these therapeutics. In RB1-deficient cells, CK2 inhibition resulted in accumulation of cells in S phase of the cell cycle, associated with micronuclei formation, and accelerated PARP inhibitor-induced aneuploidy and mitotic cell death. Patient HGSOC organoids that lacked RB1 expression displayed an enhanced long-term response to carboplatin and PARP inhibitor niraparib when combined with silmitasertib, suggesting RB1-stratified efficacy in patients. As RB1 deficiency affects up to 25% of HGSOC and 40% of TNBC cases, CK2 inhibition, proven safe from previous clinical exploration with silmitasertib, is a promising approach to overcome resistance to standard therapeutics in large strata of patients.

cancer biology↗

bioRxiv2022

OSCAR: Optimal subset cardinality regression using the L0-pseudonorm with applications to prognostic modelling of prostate cancer

In many real-world applications, such as those based on patient electronic health records, prognostic prediction of patient survival is based on heterogeneous sets of clinical laboratory measurements. To address the trade-off between the predictive accuracy of a prognostic model and the costs related to its clinical implementation, we propose an optimized L0-pseudonorm approach to learn sparse solutions in multivariable regression. The model sparsity is maintained by restricting the number of nonzero coefficients in the model with a cardinality constraint, which makes the optimization problem NP-hard. In addition, we generalize the cardinality constraint for grouped feature selection, hence making it possible to identify key sets of predictors that may be measured together in a kit in clinical practice. We demonstrate the operation of our cardinality constraint-based feature subset selection method, named OSCAR, in the context of prognostic modelling of prostate cancer, where it enabled one to determine the key explanatory predictors at different levels of model sparsity, and to explore how the model sparsity affects the model accuracy and implementation cost. Author summaryFeature selection has become a crucial part in building biomedical models, due to the abundance of available predictors in many applications, yet there remains an uncertainty of their importance and generalization ability. Regularized regression methods have become popular approaches to tackle this challenge by balancing the model goodness-of-fit against the increasing complexity of the model in terms of coefficients that deviate from zero. Regularization norms are pivotal in formulating the model complexity, and currently L1 (LASSO), L2 (Ridge Regression) and their hybrid (Elastic Net) norms dominate the field. In this paper, we present a novel methodology using the L0-pseudonorm, also known as the best subset selection, which has largely gone overlooked due to its challenging discrete nature. Our methodology makes use of a continuous transformation of the discrete optimization problem, and provides effective solvers implemented in a user friendly R software package. We exemplify the use of oscar-package in the context of prostate cancer prognostic prediction using both real-world hospital registry and clinical cohort data. By benchmarking the methodology against related regularization methods, we illustrate the advantages of the L0-pseudonorm for better clinical applicability and selection of grouped features.

bioinformatics↗

bioRxiv2022

Discovery of host-directed modulators of virus infection by probing the SARSCoV-2-host protein-protein interaction network

The ongoing coronavirus disease 2019 (COVID-19) pandemic has highlighted the need to better understand virus-host interactions. We developed a network-based algorithm that expands the SARS-CoV-2-host protein interaction network and identifies host targets that modulate viral infection. To disrupt the SARS-CoV-2 interactome, we systematically probed for potent compounds that selectively target the identified host proteins with high expression in cells relevant to COVID-19. We experimentally tested seven chemical inhibitors of the identified host proteins for modulation of SARS-CoV-2 infection in human cells that express ACE2 and TMPRSS2. Inhibition of the epigenetic regulators bromodomain-containing protein 4 (BRD4) and histone deacetylase 2 (HDAC2), along with ubiquitin specific peptidase (USP10), enhanced SARS-CoV-2 infection. Such proviral effect was observed upon treatment with compounds JQ1, vorinostat, romidepsin, and spautin-1, when measured by cytopathic effect and validated by viral RNA assays, suggesting that HDAC2, BRD4 and USP10 host proteins have antiviral functions. Mycophenolic acid and merimepodib, two inhibitors of inosine monophosphate dehydrogenase (IMPDH 1 and IMPDH 2), showed modest antiviral effects with no toxicity in mock-infected control cells. The network-based approach enables systematic identification of host-targets that selectively modulate the SARS-CoV-2 interactome, as well as reveal novel chemical tools to probe virus-host interactions that regulate virus infection. Synopsis O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=80 SRC="FIGDIR/small/494640v1_ufig1.gif" ALT="Figure 1"> View larger version (31K): org.highwire.dtl.DTLVardef@653d14org.highwire.dtl.DTLVardef@8d1234org.highwire.dtl.DTLVardef@1a632faorg.highwire.dtl.DTLVardef@5ce4f6_HPS_FORMAT_FIGEXP M_FIG C_FIG Viruses exploit host machinery and therefore it is important to understand the virus-host dependencies to gain better insight of the key regulators of viral infection. O_LIUsing a context-specific SARS-COV-2 PPI network, a computational framework was developed to identify host modulators of viral infection. C_LIO_LIChromatin modifying host proteins HDAC2 and BRD4, along with deubiquitinating enzyme USP10, act as antiviral proteins. C_LIO_LIIMPDH inhibitors mycophenolic acid and merimipodib showed modest antiviral response to SARS-COV-2 infection, and no toxic effects. C_LIO_LICell context specificity is a critical factor when identifying selective modulators of viral infection and potential antiviral therapeutics. C_LIO_LITopology-based network models cannot distinguish between host-proteins, the inhibition of which leads to either virus suppressive or enhancing effects. C_LI

systems biology↗

bioRxiv2022

Pharmacological PP2A reactivation overcomes multikinase inhibitor tolerance across brain tumor cell models

BackgroundGlioblastoma is characterized by hyperactivation of kinase signaling pathways. Regardless, most glioblastoma clinical trials targeting kinase signaling have failed. We hypothesized that overcoming the glioblastoma kinase inhibitor tolerance requires efficient shut-down of phosphorylation-dependent signaling rewiring by simultaneous inhibition of multiple critical kinases combined with reactivation of Protein Phosphatase 2A (PP2A). MethodsLive-cell imaging and colony growth assays were used to determine long-term impact of therapy effects on ten brain tumor cell models. Immunoblotting, MS-phosphoproteomics, and Seahorse metabolic assay were used for analysis of therapy-induced signaling rewiring. BH3 profiling was used to understand the mitochondrial apoptosis mechanisms. Medulloblastoma models were used to expand the importance to other brain cancer. Intracranial xenografts were used to validate the in vivo therapeutic impact of the triplet therapy. ResultsCollectively all tested ten glioblastoma and medulloblastoma cell models were effectively eradicated by the newly discovered triplet therapy combining inhibition of AKT and PDK1-4 kinases with pharmacological PP2A reactivation. Mechanistically, the brain tumor cell selective lethality of the triplet therapy could be explained by its combinatorial effects on therapy-induced signaling rewiring, OXPHOS, and apoptosis priming. The brain-penetrant triplet combination had a significant in vivo efficacy in intracranial glioblastoma and medulloblastoma models. ConclusionThe results confirm highly heterogenous responses of brain cancer cells to mono - and doublet combination therapies targeting phosphorylation-dependent signaling. However, the brain cancer cells cannot escape the triplet therapy targeting of AKT, PDK1-4, and PP2A. The results encourage evaluation of brain tumor PP2A status for design of future kinase inhibitor combination trials. Key PointsO_LIDevelopment of triplet kinase-phosphatase targeting therapy strategy for overcoming therapy tolerance across brain tumor models. C_LIO_LIIdentification of interplay between therapy-induced signaling rewiring, OXPHOS, and BH3 protein-mediated apoptosis priming as a cause for kinase inhibitor tolerance in brain cancers. C_LIO_LIValidation of the results in intracranial in vivo models with orally bioavailable and brain penetrant triplet therapy combination. C_LI Importance of the StudyBased on current genetic knowledge, glioblastoma should be particularly suitable target for kinase inhibitor therapies, However, in glioblastoma alone over 180 clinical trials with kinase inhibitors have failed. In this manuscript, we recapitulate this clinical observation by demonstrating broad tolerance of brain cancer cell models to kinase inhibitors even when combined with reactivation of PP2A. However, we discover that the therapy-induced signaling rewiring, and therapy tolerance, can be overcome by triplet targeting of AKT, PDK1-4 and PP2A. We provide strong evidence for the translatability of the findings by orally dosed brain penetrant triplet therapy combination in intracranial brain cancer models. The results encourage biomarker profiling of brain tumors for their PP2A status for clinical trials with combination of AKT and PDK1-4 inhibitors. Further, the results indicate that rapidly developing PP2A reactivation therapies will constitute an attractive future therapy option for brain tumors when combined with multi-kinase inhibition.

cancer biology↗

bioRxiv2022

Functional testing of PI3K inhibitors stratifies responders to idelalisib and identifies treatment vulnerabilities in idelalisib-refractory/intolerant chronic lymphocytic leukemia

PurposePhosphatidylinositol 3-kinase inhibitors (PI3Ki) are approved for relapsed chronic lymphocytic leukemia (CLL). While patients may show an initial response, development of treatment intolerance or resistance remains a clinically challenging. Prediction of individual treatment responses based on clinically actionable biomarkers is needed to overcome these challenges. Here, we investigated whether ex vivo functional responses to targeted therapies can stratify responders to idelalisib and guide precision medicine in CLL. Experimental designCLL cells from treatment naive, idelalisib-responding, and idelalisib-refractory/intolerant patients (n=33 in total) were profiled against ten PI3Ki and the Bcl-2 antagonist venetoclax. Cell signaling and immune phenotypes were analyzed by flow cytometry. Cell viability was monitored by detection of cleaved caspase-3 and the CellTiter-Glo assay. ResultsAmong the ten PI3Ki studied, pan-PI3Ki were most effective at inhibiting PI3K signaling and cell viability, and they showed activity also in CLL cells from idelalisib-refractory/intolerant patients. The pan-PI3Ki copanlisib, but not the p110{delta} inhibitor idelalisib, inhibited PI3K signaling in CD4+ and CD8+ T cells in addition to CD19+ B cells, while it did not significantly affect T cell numbers. Combination treatment with a PI3Ki and venetoclax resulted in synergistic induction of apoptosis. Based on ex vivo drug sensitivity testing, a relapsed CLL patient was treated with idelalisib plus venetoclax, and the patient achieved a partial response. A more systematic analysis revealed that CLL cells from patients with a long-term response to idelalisib showed significantly higher drug sensitivities to 73 drug combinations at baseline compared to short-term responders. ConclusionsOur findings suggest novel treatment vulnerabilities in idelalisib-refractory/intolerant CLL, and demonstrate that ex vivo functional profiling may guide precision medicine and predict treatment responses of individual CLL patients. TRANSLATIONAL RELEVANCEThe phosphatidylinositol 3-kinase inhibitors (PI3Ki) idelalisib and duvelisib are approved for relapsed chronic lymphocytic leukemia (CLL), but their use has been limited by severe toxicity and acquired resistance. Identification of biomarkers that predict individual treatment responses, as well as alternative treatment vulnerabilities in PI3Ki refractory/intolerant patients, is needed to optimally tailor CLL therapy. We performed functional analyses of CLL cells from treatment naive, idelalisib-responding and idelalisib-refractory/intolerant patients to identify clinically actionable biomarkers. We show that CLL cells from idelalisib-refractory/intolerant patients remain sensitive to pan-PI3Ki and PI3Ki plus venetoclax combinations. Ex vivo drug sensitivity testing was used to guide treatment of a relapsed CLL patient who obtained a partial response after idelalisib plus venetoclax therapy. A systematic analysis of drug sensitivities to 73 drug combinations stratified responders to idelalisib using baseline samples from short-term and long-term responders to idelalisib. Our study demonstrates the power of functional precision medicine in relapsed CLL.

cancer biology↗

bioRxiv2022

Triplet kinase-phosphatase targeting to overcome kinase inhibitor tolerance in brain tumor cells

Therapeutic resistance to kinase inhibitors constitutes a major unresolved clinical challenge in cancer and especially in glioblastoma. Multi-kinase inhibitors may be used for simultaneous targeting of multiple target kinase and thereby potentially overcome kinase inhibitor resistance. However, in most cases identification of the target kinases mediating therapeutic effects of multi-kinase inhibitors has been challenging. To tackle this important problem, we developed an Actionable Targets of Multi-kinase Inhibitors (AToMI) strategy and used it for characterization of glioblastoma target kinases of staurosporine derivatives displaying synergy with protein phosphatase 2A (PP2A) reactivation. AToMI consists of interchangeable modules combining drug-kinase interaction assay, siRNA high-throughput screening, bioinformatics analysis and validation screening with more selective target kinase inhibitors. As a result, AToMI analysis revealed AKT and mitochondrial pyruvate dehydrogenase kinase PDK1 and PDK4 as kinase targets of staurosporine derivatives UCN-01, CEP-701, and K252a that synergized with PP2A activation across heterogeneous glioblastoma cells. Based on these proof-of-principle results we propose that application and further development of AToMI for clinically applicable multi-kinase inhibitors could provide significant benefits in overcoming the challenge of lack of knowledge of target specificity of multi-kinase inhibitors.

cancer biology↗

bioRxiv2021

Modeling drug combination effects via latent tensor reconstruction

AO_SCPLOWBSTRACTC_SCPLOWO_ST_ABSMotivationC_ST_ABSCombination therapies have emerged as a powerful treatment modality to overcome drug resistance and improve treatment efficacy. However, the number of possible drug combinations increases very rapidly with the number of individual drugs in consideration which makes the comprehensive experimental screening infeasible in practice. Machine learning models offer time- and cost-efficient means to aid this process by prioritising the most effective drug combinations for further pre-clinical and clinical validation. However, the complexity of the underlying interaction patterns across multiple drug doses and in different cellular contexts poses challenges to the predictive modelling of drug combination effects. ResultsWe introduce comboLTR, highly time-efficient method for learning complex, nonlinear target functions for describing the responses of therapeutic agent combinations in various doses and cancer cell-contexts. The method is based on a polynomial regression via powerful latent tensor reconstruction. It uses a combination of recommender system-style features indexing the data tensor of response values in different contexts, and chemical and multi-omics features as inputs. We demonstrate that comboLTR outperforms state-of-the-art methods in terms of predictive performance and running time, and produces highly accurate results even in the challenging and practical inference scenario where full dose-response matrices are predicted for completely new drug combinations with no available combination and monotherapy response measurements in any training cell line. Availability and implementationcomboLTR code is available at https://github.com/aalto-ics-kepaco/ComboLTR Contacttianduanyi.wang@aalto.fi; juho.rousu@aalto.fi

bioinformatics↗

bioRxiv2020

Optimized chemogenomic library design strategies for precision oncology

Designing a focused compound screening library of bioactive small molecules is a challenging task since many compounds modulate their effects through multiple protein targets with various degrees of potency and selectivity. We describe here several analytic procedures with adjustable cut-off parameters that enable one to design anticancer target-focused compound libraries optimized for library size, cellular activity, biological and chemical diversity and target selectivity. Even though our focus was on designing compound libraries to enable a comprehensive investigation of the target biology of glioblastoma (GBM), the compound collections cover a wide range of protein targets and biological pathways implicated in various types of cancers, making the libraries widely applicable in precision oncology studies. We published the final screening set library, called the Comprehensive anti-Cancer small-Compound Library, or C3L.We hope these general library design principles and the current, widely annotated small molecule libraries will prove useful for the community in various phenotypic screening experiments in GBM and other cancers.

cancer biology↗

bioRxiv2020

comboFM: leveraging multi-way interactions for systematic prediction of drug combination effects

We present comboFM, a machine learning framework for predicting the responses of drug combinations in preclinical studies, such as those based on cell lines or patient-derived cells. comboFM models the cell context-specific drug interactions through higher-order tensors, and efficiently learns latent factors of the tensor using powerful factorization machines. The approach enables comboFM to leverage information from previous experiments performed on similar drugs and cells when predicting responses of new combinations in so far untested cells; thereby, it achieves highly accurate predictions despite sparsely populated data tensors. We demonstrated high predictive performance of comboFM in various prediction scenarios using data from cancer cell line drug screening. Subsequent experimental validation of a set of previously untested drug combinations further supported the practical and robust applicability of comboFM. For instance, we confirmed a novel synergy between anaplastic lymphoma kinase (ALK) inhibitor crizotinib and proteasome inhibitor bortezomib in lymphoma cells. Overall, our results demonstrate that comboFM provides an effective means for systematic pre-screening of drug combinations to support precision oncology applications.

bioinformatics↗

bioRxiv2020

Patient-tailored design of AML cell subpopulation-selective drug combinations

The extensive primary and secondary drug resistance in acute myeloid leukemia (AML) requires rational approaches to design personalized combinatorial treatments that exploit patient-specific therapeutic vulnerabilities to optimally target disease-driving AML cell subpopulations. However, the large number of AML-relevant drug combinations makes the testing impossible in scarce primary patient cells. This combinatorial problem is further exacerbated by the translational challenge of how to design such personalized and selective drug combinations that do not only show synergistic effect in overall AML cell killing but also result in minimal toxic side effects on non-malignant cells. To solve these challenges, we implemented a systematic computational-experimental approach for identifying potential drug combinations that have a desired synergy-efficacy-toxicity balance. Our mechanism-agnostic approach combines single-cell RNA-sequencing (scRNA-seq) with ex vivo single-agent viability testing in primary patient cells. The data integration and predictive modelling are carried out at a single-cell resolution by means of a machine learning model that makes use of compound-target interaction networks to narrow down the massive search space of potentially effective drug combinations. When applied to two diagnostic and two refractory AML patient cases, each having a different genetic background, our integrated approach predicted a number of patient-specific combinations that were shown to result not only in synergistic cancer cell inhibition but were also capable of targeting specific AML cell subpopulations that emerge in differing stages of disease pathogenesis or treatment regimens. Overall, 53% of the 59 predicted combinations were experimentally confirmed to show synergy, and 83% were non-antagonistic, as validated with viability assays, which is a significant improvement over the success rate of randomly guessing a synergistic drug combination (5%). Importantly, 67% of the predicted combinations showed low toxicity to non-malignant cells, as validated with flow-based population assays, suggesting their selective killing of AML cell populations. Our data-driven approach provides an unbiased means for systematic prioritization of patient-specific drug combinations that selectively inhibit AML cells and avoid co-inhibition of non-malignant cells, thereby increasing their likelihood for clinical translation. The approach uses only a limited number of patient primary cells, and it is widely applicable to hematological cancers that are accessible for scRNA-seq profiling and ex vivo compound testing.

systems biology↗

bioRxiv2020

Histopathological Landscape of Molecular Genetics and Clinical Determinants in MDS Patients

In myelodysplastic syndrome (MDS), bone marrow (BM) histopathology is visually assessed to identify dysplastic cellular morphology, cellularity, and blast excess. Yet, many morphological findings elude the human eye. Here, we extracted visual features of 236 MDS, 87 MDS/MPN, and 10 control BM biopsies with convolutional neural networks. Unsupervised analysis distinguished underlying correlations between tissue composition, leukocyte metrics, and clinical characteristics. We applied morphological features in elastic net-regularized regression models to predict genetic and cytogenetic aberrations, prognosis, and clinical variables. By parallelizing tile, pixel, and leukocyte-level image analysis, we deconvoluted each model to texture and cellular composition to dissect their pathobiological context. Model-based mutation predictions correlated with variant allele frequency and number of affected genes per pathway, demonstrating the models ability to identify relevant visual patterns. In summary, this study highlights the potential of deep histopathology in hematology by unveiling the fundamental association of BM morphology with genetic and clinical determinants.

bioinformatics↗