Multi-omic analysis of bat versus human fibroblasts reveals altered central metabolism.

Bats have unique characteristics compared to other mammals, including increased longevity and higher resistance to cancer and infectious disease. While previous studies have analyzed the metabolic requirements for flight, it is still unclear how bat metabolism supports these unique features, and no study has integrated metabolomics, transcriptomics, and proteomics to characterize bat metabolism. In this work, we performed a multi-omics data analysis using a computational model of metabolic fluxes to identify fundamental differences in central metabolism between primary lung fibroblast cell lines from the black flying fox fruit bat (Pteropus alecto) and human. Bat cells showed higher expression levels of Complex I components of electron transport chain (ETC), but, remarkably, a lower rate of oxygen consumption. Computational modeling interpreted these results as indicating that Complex II activity may be low or reversed, similar to an ischemic state. An ischemic-like state of bats was also supported by decreased levels of central metabolites and increased ratios of succinate to fumarate in bat cells. Ischemic states tend to produce reactive oxygen species (ROS), which would be incompatible with the longevity of bats. However, bat cells had higher antioxidant reservoirs (higher total glutathione and higher ratio of NADPH to NADP) despite higher mitochondrial ROS levels. In addition, bat cells were more resistant to glucose deprivation and had increased resistance to ferroptosis, one of the characteristics of which is oxidative stress. Thus, our studies revealed distinct differences in the ETC regulation and metabolic stress responses between human and bat cells.

A high-density microfluidic bioreactor for the automated manufacturing of CAR T cells.

The manufacturing of autologous chimaeric antigen receptor (CAR) T cells largely relies either on fed-batch and manual processes that often lack environmental monitoring and control or on bioreactors that cannot be easily scaled out to meet patient demands. Here we show that human primary T cells can be activated, transduced and expanded to high densities in a 2 ml automated closed-system microfluidic bioreactor to produce viable anti-CD19 CAR T cells (specifically, more than 60 million CAR T cells from donor cells derived from patients with lymphoma and more than 200 million CAR T cells from healthy donors). The in vitro secretion of cytokines, the short-term cytotoxic activity and the long-term persistence and proliferation of the cell products, as well as their in vivo anti-leukaemic activity, were comparable to those of T cells produced in a gas-permeable well. The manufacturing-process intensification enabled by the miniaturized perfusable bioreactor may facilitate the analysis of the growth and metabolic states of CAR T cells during ex vivo culture, the high-throughput optimization of cell-manufacturing processes and the scale out of cell-therapy manufacturing.

Computational modeling of light processing in the habenula and dorsal raphe based on laser ablation of functionally-defined cells.

BACKGROUND: The habenula is a major regulator of serotonergic neurons in the dorsal raphe, and thus of brain state. The functional connectivity between these regions is incompletely characterized. Here, we use the ability of changes in irradiance to trigger reproducible changes in activity in the habenula and dorsal raphe of zebrafish larvae, combined with two-photon laser ablation of specific neurons, to establish causal relationships. RESULTS: Neurons in the habenula can show an excitatory response to the onset or offset of light, while neurons in the anterior dorsal raphe display an inhibitory response to light, as assessed by calcium imaging. The raphe response changed in a complex way following ablations in the dorsal habenula (dHb) and ventral habenula (vHb). After ablation of the ON cells in the vHb (V-ON), the raphe displayed no response to light. After ablation of the OFF cells in the vHb (V-OFF), the raphe displayed an excitatory response to darkness. After ablation of the ON cells in the dHb (D-ON), the raphe displayed an excitatory response to light. We sought to develop in silico models that could recapitulate the response of raphe neurons as a function of the ON and OFF cells of the habenula. Early attempts at mechanistic modeling using ordinary differential equation (ODE) failed to capture observed raphe responses accurately. However, a simple two-layer fully connected neural network (NN) model was successful at recapitulating the diversity of observed phenotypes with root-mean-squared error values ranging from 0.012 to 0.043. The NN model also estimated the raphe response to ablation of D-off cells, which can be verified via future experiments. CONCLUSION: Lesioning specific cells in different regions of habenula led to qualitatively different responses to light in the dorsal raphe. A simple neural network is capable of mimicking experimental observations. This work illustrates the ability of computational modeling to integrate complex observations into a simple compact formalism for generating testable hypotheses, and for guiding the design of biological experiments.

Evaluation of an Evidence-Based Care Bundle for Preventing Hospital-Acquired Pressure Injuries in High-Risk Surgical Patients.

The aim of this study was to evaluate the effectiveness of an evidence-based care bundle to prevent perioperative pressure injuries. In a single facility, using a preintervention and postintervention quasi-experimental design, we compared the pressure injury incidence rate for two patient groups (ie, before and after care bundle implementation). The bundle included a variety of elements, such as educating patients, applying protection, controlling skin moisture, and using pressure-relieving devices according to the patient's risk. Before the intervention, patients received standard care before procedures that did not address risk for pressure injury development. The study involved a total of 944 patients, and the incidence of pressure injury was lower in the postintervention group than in the preintervention group (1.6% versus 4.8%; P < .001). However, the odds ratio was nonsignificant and therefore the clinical relevance of the bundle is unclear. Additional research with a control group and multiple sites is needed.

Myoglobin-derived iron causes wound enlargement and impaired regeneration in pressure injuries of muscle.

The reasons for poor healing of pressure injuries are poorly understood. Vascular ulcers are worsened by extracellular release of hemoglobin, so we examined the impact of myoglobin (Mb) iron in murine muscle pressure injuries (mPI). Tests used Mb-knockout or treatment with deferoxamine iron chelator (DFO). Unlike acute injuries from cardiotoxin, mPI regenerated poorly with a lack of viable immune cells, persistence of dead tissue (necro-slough), and abnormal deposition of iron. However, Mb-knockout or DFO-treated mPI displayed a reversal of the pathology: decreased tissue death, decreased iron deposition, decrease in markers of oxidative damage, and higher numbers of intact immune cells. Subsequently, DFO treatment improved myofiber regeneration and morphology. We conclude that myoglobin iron contributes to tissue death in mPI. Remarkably, a large fraction of muscle death in untreated mPI occurred later than, and was preventable by, DFO treatment, even though treatment started 12 hr after pressure was removed. This demonstrates an opportunity for post-pressure prevention to salvage tissue viability.

Deep learning for de-convolution of Smad2 versus Smad3 binding sites.

BACKGROUND: The transforming growth factor beta-1 (TGF ß-1) cytokine exerts both pro-tumor and anti-tumor effects in carcinogenesis. An increasing body of literature suggests that TGF ß-1 signaling outcome is partially dependent on the regulatory targets of downstream receptor-regulated Smad (R-Smad) proteins Smad2 and Smad3. However, the lack of Smad-specific antibodies for ChIP-seq hinders convenient identification of Smad-specific binding sites. RESULTS: In this study, we use localization and affinity purification (LAP) tags to identify Smad-specific binding sites in a cancer cell line. Using ChIP-seq data obtained from LAP-tagged Smad proteins, we develop a convolutional neural network with long-short term memory (CNN-LSTM) as a deep learning approach to classify a pool of Smad-bound sites as being Smad2- or Smad3-bound. Our data showed that this approach is able to accurately classify Smad2- versus Smad3-bound sites. We use our model to dissect the role of each R-Smad in the progression of breast cancer using a previously published dataset. CONCLUSIONS: Our results suggests that deep learning approaches can be used to dissect binding site specificity of closely related transcription factors.

The panniculus carnosus muscle: a missing link in the chronicity of heel pressure ulcers?

Chronic and recurring pressure ulcers (PUs) create an unmet need for predictive biomarkers. In this work, we examine the panniculus carnosus, a thin cutaneous muscle, traditionally considered vestigial in humans, and ask whether the panniculus may play a role in the chronicity and reinjury of heel PUs. To determine whether humans have a panniculus muscle layer at the heel, we dissected eight cadavers. To assess the influence of the panniculus layer on PU, we performed computational simulations of supine weight bearing. Finally, we assessed panniculus regeneration in fluorescent mice. Results show a panniculus layer present in all cadavers examined. Simulations show a thin layer of panniculus muscle causes a dramatic decrease in the volume of soft tissue experiencing high strain and stress, compared to a heel without a panniculus. Importantly, in the mouse model, the panniculus fails to regenerate after PU, even when other cutaneous layers had fully regenerated. Our work shows that the panniculus is able to redistribute load around the heel bone, which might allow it to prevent PUs. Moreover, it is highly susceptible to incomplete regeneration after PU. Poor panniculus regeneration after PU might be a predictive anatomical biomarker for recurrence, and this biomarker should be evaluated prospectively in future clinical trials.

Functional coding haplotypes and machine-learning feature elimination identifies predictors of Methotrexate Response in Rheumatoid Arthritis patients.

BACKGROUND: Major challenges in large scale genetic association studies include not only the identification of causative single nucleotide polymorphisms (SNPs), but also accounting for SNP-SNP interactions. This study thus proposes a novel feature engineering approach integrating potentially functional coding haplotypes (pfcHap) with machine-learning (ML) feature selection to identify biologically meaningful, possibly causative genetic factors, that take into consideration potential SNP-SNP interactions within the pfcHap, to best predict for methotrexate (MTX) response in rheumatoid arthritis (RA) patients. METHODS: Exome sequencing from 349 RA patients were analysed, of which they were split into training and unseen test set. Inferred pfcHaps were combined with 30 non-genetic features to undergo ML recursive feature elimination with cross-validation using the training set. Predictive capacity and robustness of the selected features were assessed using six popular machine learning models through a train set cross-validation and evaluated in an unseen test set. FINDINGS: Significantly, 100 features (95 pfcHaps, 5 non-genetic factors) were identified to have good predictive performance (AUC: 0.776-0.828; Sensitivity: 0.656-0.813; Specificity: 0.684-0.868) across all six ML models in an unseen test dataset for the prediction of MTX response in RA patients. INTERPRETATION: Majority of the predictive pfcHap SNPs were predicted to be potentially functional and some of the genes in which the pfcHap resides in were identified to be associated with previously reported MTX/RA pathways. FUNDING: Singapore Ministry of Health's National Medical Research Council (NMRC) [NMRC/CBRG/0095/2015; CG12Aug17; CGAug16M012; NMRC/CG/017/2013]; National Cancer Center Research Fund and block funding Duke-NUS Medical School.; Singapore Ministry of Education Academic Research Fund Tier 2 grant MOE2019-T2-1-138.

Machine learning using genetic and clinical data identifies a signature that robustly predicts methotrexate response in rheumatoid arthritis.

OBJECTIVE: To develop a hypothesis-free model that best predicts response to MTX drug in RA patients utilizing biologically meaningful genetic feature selection of potentially functional single nucleotide polymorphisms (pfSNPs) through robust machine learning (ML) feature selection methods. METHODS: MTX-treated RA patients with known response were divided in a 4:1 ratio into training and test sets. From the patients' exomes, potential features for classifier prediction were identified from pfSNPs and non-genetic factors through ML using recursive feature elimination with cross-validation incorporating the random forest classifier. Feature selection was repeated on random subsets of the training cohort, and consensus features were assembled into the final feature set. This feature set was evaluated for predictive potential using six ML classifiers, first by cross-validation within the training set, and finally by analysing its performance with the unseen test set. RESULTS: The final feature set contains 56 pfSNPs and five non-genetic factors. The majority of these pfSNPs are located in pathways related to RA pathogenesis or MTX action and are predicted to modulate gene expression. When used for training in six ML classifiers, performance was good in both the training set (area under the curve: 0.855-0.916; sensitivity: 0.715-0.892; and specificity: 0.733-0.862) and the unseen test set (area under the curve: 0.751-0.826; sensitivity: 0.581-0.839; and specificity: 0.641-0.923). CONCLUSION: Sensitive and specific predictors of MTX response in RA patients were identified in this study through a novel strategy combining biologically meaningful and machine learning feature selection and training. These predictors may facilitate better treatment decision-making in RA management.

A computational model of mutual antagonism in the mechano-signaling network of RhoA and nitric oxide.

BACKGROUND: RhoA is a master regulator of cytoskeletal contractility, while nitric oxide (NO) is a master regulator of relaxation, e.g., vasodilation. There are multiple forms of cross-talk between the RhoA/ROCK pathway and the eNOS/NO/cGMP pathway, but previous work has not studied their interplay at a systems level. Literature review suggests that the majority of their cross-talk interactions are antagonistic, which motivates us to ask whether the RhoA and NO pathways exhibit mutual antagonism in vitro, and if so, to seek the theoretical implications of their mutual antagonism. RESULTS: Experiments found mutual antagonism between RhoA and NO in epithelial cells. Since mutual antagonism is a common motif for bistability, we sought to explore through theoretical simulations whether the RhoA-NO network is capable of bistability. Qualitative modeling showed that there are parameters that can cause bistable switching in the RhoA-NO network, and that the robustness of the bistability would be increased by positive feedback between RhoA and mechanical tension. CONCLUSIONS: We conclude that the RhoA-NO bistability is robust enough in silico to warrant the investment of further experimental testing. Tension-dependent bistability has the potential to create sharp concentration gradients, which could contribute to the localization and self-organization of signaling domains during cytoskeletal remodeling and cell migration.

Immune cell phenotypes associated with disease severity and long-term neutralizing antibody titers after natural dengue virus infection.

Prior immunological exposure to dengue virus can be both protective and disease-enhancing during subsequent infections with different dengue virus serotypes. We provide here a systematic, longitudinal analysis of B cell, T cell, and antibody responses in the same patients. Antibody responses as well as T and B cell activation differentiate primary from secondary responses. Hospitalization is associated with lower frequencies of activated, terminally differentiated T cells and higher percentages of effector memory CD4 T cells. Patients with more severe disease tend to have higher percentages of plasmablasts. This does not translate into long-term antibody titers, since neutralizing titers after 6 months correlate with percentages of specific memory B cells, but not with acute plasmablast activation. Overall, our unbiased analysis reveals associations between cellular profiles and disease severity, opening opportunities to study immunopathology in dengue disease and the potential predictive value of these parameters.

Computational modeling suggests binding-induced expansion of Epsin disordered regions upon association with AP2.

Intrinsically disordered regions (IDRs) are prevalent in the eukaryotic proteome. Common functional roles of IDRs include forming flexible linkers or undergoing allosteric folding-upon-binding. Recent studies have suggested an additional functional role for IDRs: generating steric pressure on the plasma membrane during endocytosis, via molecular crowding. However, in order to accomplish useful functions, such crowding needs to be regulated in space (e.g., endocytic hotspots) and time (e.g., during vesicle formation). In this work, we explore binding-induced regulation of IDR steric volume. We simulate the IDRs of two proteins from Clathrin-mediated endocytosis (CME) to see if their conformational spaces are regulated via binding-induced expansion. Using Monte-Carlo computational modeling of excluded volumes, we generate large conformational ensembles (3 million) for the IDRs of Epsin and Eps15 and dock the conformers to the alpha subunit of Adaptor Protein 2 (AP2α), their CME binding partner. Our results show that as more molecules of AP2α are bound, the Epsin-derived ensemble shows a significant increase in global dimensions, measured as the radius of Gyration (RG) and the end-to-end distance (EED). Unlike Epsin, Eps15-derived conformers that permit AP2α binding at one motif were found to be more likely to accommodate binding of AP2α at other motifs, suggesting a tendency toward co-accessibility of binding motifs. Co-accessibility was not observed for any pair of binding motifs in Epsin. Thus, we speculate that the disordered regions of Epsin and Eps15 perform different roles during CME, with accessibility in Eps15 allowing it to act as a recruiter of AP2α molecules, while binding-induced expansion of the Epsin disordered region could impose steric pressure and remodel the plasma membrane during vesicle formation.

Bile canaliculi contract autonomously by releasing calcium into hepatocytes via mechanosensitive calcium channel.

Drug-induced hepatocellular cholestasis leads to altered bile flow. Bile is propelled along the bile canaliculi (BC) by actomyosin contractility, triggered by increased intracellular calcium (Ca2+). However, the source of increased intracellular Ca2+ and its relationship to transporter activity remains elusive. We identify the source of the intracellular Ca2+ involved in triggering BC contractions, and we elucidate how biliary pressure regulates Ca2+ homeostasis and associated BC contractions. Primary rat hepatocytes were cultured in collagen sandwich. Intra-canalicular Ca2+ was measured with fluo-8; and intra-cellular Ca2+ was measured with GCaMP. Pharmacological modulators of canonical Ca2+-channels were used to study the Ca2+-mediated regulation of BC contraction. BC contraction correlates with cyclic transfer of Ca2+ from BC to adjacent hepatocytes, and not with endoplasmic reticulum Ca2+. A mechanosensitive Ca2+ channel (MCC), Piezo-1, is preferentially localized at BC membranes. The Piezo-1 inhibitor GsMTx-4 blocks the Ca2+ transfer, resulting in cholestatic generation of BC-derived vesicles whereas Piezo-1 hyper-activation by Yoda1 increases the frequency of Ca2+ transfer and BC contraction cycles. Yoda1 can recover normal BC contractility in drug-induced hepatocellular cholestasis, supporting that Piezo-1 regulates BC contraction cycles. Finally, we show that hyper-activating Piezo-1 can be exploited to normalize bile flow in drug-induced hepatocellular cholestasis.

Mitochondrial peptide BRAWNIN is essential for vertebrate respiratory complex III assembly.

The emergence of small open reading frame (sORF)-encoded peptides (SEPs) is rapidly expanding the known proteome at the lower end of the size distribution. Here, we show that the mitochondrial proteome, particularly the respiratory chain, is enriched for small proteins. Using a prediction and validation pipeline for SEPs, we report the discovery of 16 endogenous nuclear encoded, mitochondrial-localized SEPs (mito-SEPs). Through functional prediction, proteomics, metabolomics and metabolic flux modeling, we demonstrate that BRAWNIN, a 71 a.a. peptide encoded by C12orf73, is essential for respiratory chain complex III (CIII) assembly. In human cells, BRAWNIN is induced by the energy-sensing AMPK pathway, and its depletion impairs mitochondrial ATP production. In zebrafish, Brawnin deletion causes complete CIII loss, resulting in severe growth retardation, lactic acidosis and early death. Our findings demonstrate that BRAWNIN is essential for vertebrate oxidative phosphorylation. We propose that mito-SEPs are an untapped resource for essential regulators of oxidative metabolism.

Transcompp: understanding phenotypic plasticity by estimating Markov transition rates for cell state transitions.

MOTIVATION: Gradual population-level changes in tissues can be driven by stochastic plasticity, meaning rare stochastic transitions of single-cell phenotype. Quantifying the rates of these stochastic transitions requires time-intensive experiments, and analysis is generally confounded by simultaneous bidirectional transitions and asymmetric proliferation kinetics. To quantify cellular plasticity, we developed Transcompp (Transition Rate ANalysis of Single Cells to Observe and Measure Phenotypic Plasticity), a Markov modeling algorithm that uses optimization and resampling to compute best-fit rates and statistical intervals for stochastic cell-state transitions. RESULTS: We applied Transcompp to time-series datasets in which purified subpopulations of stem-like or non-stem cancer cells were exposed to various cell culture environments, and allowed to re-equilibrate spontaneously over time. Results revealed that commonly used cell culture reagents hydrocortisone and cholera toxin shifted the cell population equilibrium toward stem-like or non-stem states, respectively, in the basal-like breast cancer cell line MCF10CA1a. In addition, applying Transcompp to patient-derived cells showed that transition rates computed from short-term experiments could predict long-term trajectories and equilibrium convergence of the cultured cell population. AVAILABILITY AND IMPLEMENTATION: Freely available for download at http://github.com/nsuhasj/Transcompp. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Superoxide induced inhibition of death receptor signaling is mediated via induced expression of apoptosis inhibitory protein cFLIP.

The death inhibitory proteins, cFLIP and Bcl-2, canonically act at different steps to regulate receptor-mediated apoptosis in cancer cells. Here we report that pharmacological or genetic means to effect an increase in intracellular superoxide result in cFLIP upregulation. Interestingly, Bcl-2 overexpression is associated with a concomitant increase in cFLIP, and reducing superoxide sensitizes Bcl-2 overexpressing cancer cells to receptor-mediated apoptosis via downregulation of cFLIP. Moreover, inhibiting glycolytic flux overcomes apoptosis resistance by superoxide-dependent downregulation of cFLIP. Superoxide-induced upregulation of cFLIP is a function of enhanced transcription, as evidenced by increases in cFLIP promoter activity and mRNA abundance. The positive effect of superoxide on cFLIP is mediated through its reaction with nitric oxide to generate peroxynitrite. Corroborating these findings in cell lines, subjecting primary cells derived from lymphoma patients to glucose deprivation ex vivo, as a means to decrease superoxide, not only reduced cFLIP expression but also significantly enhanced death receptor sensitivity. Based on this novel mechanistic insight into the redox regulation of cancer cell fate, modulation of intracellular superoxide could have potential therapeutic implications in cancers in which these two death inhibitory proteins present a therapeutic challenge.

De novo ensemble modeling suggests that AP2-binding to disordered regions can increase steric volume of Epsin but not Eps15.

Proteins with intrinsically disordered regions (IDRs) have large hydrodynamic radii, compared with globular proteins of equivalent weight. Recent experiments showed that IDRs with large radii can create steric pressure to drive membrane curvature during Clathrin-mediated endocytosis (CME). Epsin and Eps15 are two CME proteins with IDRs that contain multiple motifs for binding the adaptor protein AP2, but the impact of AP2-binding on these IDRs is unknown. Some IDRs acquire binding-induced function by forming a folded quaternary structure, but we hypothesize that the IDRs of Epsin and/or Eps15 acquire binding-induced function by increasing their steric volume. We explore this hypothesis in silico by generating conformational ensembles of the IDRs of Epsin (4 million structures) or Eps15 (3 million structures), then estimating the impact of AP2-binding on Radius of Gyration (RG). Results show that the ensemble of Epsin IDR conformations that accommodate AP2 binding has a right-shifted distribution of RG (larger radii) than the unbound Epsin ensemble. In contrast, the ensemble of Eps15 IDR conformations has comparable RG distribution between AP2-bound and unbound. We speculate that AP2 triggers the Epsin IDR to function through binding-induced-expansion, which could increase steric pressure and membrane bending during CME.

An expectation-maximization algorithm enables accurate ecological modeling using longitudinal microbiome sequencing data.

BACKGROUND: The dynamics of microbial communities is driven by a range of interactions from symbiosis to predator-prey relationships, the majority of which are poorly understood. With the increasing availability of high-throughput microbiome taxonomic profiling data, it is now conceivable to directly learn the ecological models that explicitly define microbial interactions and explain community dynamics. The applicability of these approaches is severely limited by the lack of accurate absolute cell density measurements (biomass). METHODS: We present a new computational approach that resolves this key limitation in the inference of generalized Lotka-Volterra models (gLVMs) by coupling biomass estimation and model inference with an expectation-maximization algorithm (BEEM). RESULTS: BEEM outperforms the state-of-the-art methods for inferring gLVMs, while simultaneously eliminating the need for additional experimental biomass data as input. BEEM's application to previously inaccessible public datasets (due to the lack of biomass data) allowed us to construct ecological models of microbial communities in the human gut on a per-individual basis, revealing personalized dynamics and keystone species. CONCLUSIONS: BEEM addresses a key bottleneck in "systems analysis" of microbiomes by enabling accurate inference of ecological models from high throughput sequencing data without the need for experimental biomass measurements.

A prediction tool for hospital-acquired pressure ulcers among surgical patients: Surgical pressure ulcer risk score.

Surgical patients are prone to developing hospital-acquired pressure ulcers (HAPU). Therefore, a better prediction tool is needed to predict risk using preoperative data. This study aimed to determine, from previously published HAPU risk factors, which factors are significant among our surgical population and to develop a prediction tool that identifies pressure ulcer risk before the operation. A literature review was first performed to elicit all the published HAPU risk factors before conducting a retrospective case-control study using medical records. The known HAPU risks were compared between patients with HAPU and without HAPU who underwent operations during the same period (July 2015-December 2016). A total of 80 HAPU cases and 189 controls were analysed. Multivariate logistic regression analyses identified eight significant risk factors: age ≥ 75 years, female gender, American Society of Anaesthesiologists ≥ 3, body mass index < 23, preoperative Braden score ≤ 14, anaemia, respiratory disease, and hypertension. The model had bootstrap-corrected c-statistic 0.78 indicating good discrimination. A cut-off score of ≥6 is strongly predictive, with a positive predictive value of 73.2% (confidence interval [CI]: 59.7%-84.2%) and a negative predictive value of 80.7% (CI: 74.3%-86.1%). SPURS contributes to the preoperative identification of pressure ulcer risk that could help nurses implement preventive measures earlier.

Combination Therapy and the Evolution of Resistance: The Theoretical Merits of Synergism and Antagonism in Cancer.

The search for effective combination therapies for cancer has focused heavily on synergistic combinations because they exhibit enhanced therapeutic efficacy at lower doses. Although synergism is intuitively attractive, therapeutic success often depends on whether drug resistance develops. The impact of synergistic combinations (vs. antagonistic or additive combinations) on the process of drug-resistance evolution has not been investigated. In this study, we use a simplified computational model of cancer cell numbers in a population of drug-sensitive, singly-resistant, and fully-resistant cells to simulate the dynamics of resistance evolution in the presence of two-drug combinations. When we compared combination therapies administered at the same combination of effective doses, simulations showed synergistic combinations most effective at delaying onset of resistance. Paradoxically, when the therapies were compared using dose combinations with equal initial efficacy, antagonistic combinations were most successful at suppressing expansion of resistant subclones. These findings suggest that, although synergistic combinations could suppress resistance through early decimation of cell numbers (making them "proefficacy" strategies), they are inherently fragile toward the development of single resistance. In contrast, antagonistic combinations suppressed the clonal expansion of singly-resistant cells, making them "antiresistance" strategies. The distinction between synergism and antagonism was intrinsically connected to the distinction between offensive and defensive strategies, where offensive strategies inflicted early casualties and defensive strategies established protection against anticipated future threats. Our findings question the exclusive focus on synergistic combinations and motivate further consideration of nonsynergistic combinations for cancer therapy.Significance: Computational simulations show that if different combination therapies have similar initial efficacy in cancers, then nonsynergistic drug combinations are more likely than synergistic drug combinations to provide a long-term defense against the evolution of therapeutic resistance. Cancer Res; 78(9); 2419-31. ©2018 AACR.