Integrative Biology

Signaling pathways transmit and process information, enabling cells to respond accurately to external cues. Disease states like cancer can corrupt signal transmission, though the magnitude to which they reduce information capacity has not been quantified. Here we apply pseudo-random pulsatile optogenetic stimulation, live-cell imaging, and information theory to compare the information capacity of receptor tyrosine kinase (RTK) signaling pathways in EML4-ALK-driven lung cancer cells (STE-1) and non-transformed lung epithelial cells (BEAS-2B). The information rate through the RTK/ERK pathway in STE-1 cells was below 0.5 bit/hour but increased to 3 bit/hour after oncogene inhibition. Information was transmitted by only 50-70% of cells, whose channel capacity (maximum information rate) was estimated through in silico protocol optimization. Although oncogene inhibition increased the capacity of the RTK/ERK pathway in STE-1 cells (6 bit/hour), capacity remained lower than in BEAS-2B (11 bit/hour). The capacity of the parallel RTK/calcineurin pathway in BEAS-2B exceeded 15 bit/hour. This study highlights information capacity as a sensitive metric for identifying disease-associated dysfunction and evaluating effects of targeted interventions.

Antibody-drug conjugates (ADCs) have had remarkable clinical success in recent years with multiple new approvals. However, for some ADCs, the response rates dont closely correlate with clinical target expression. One particular ADC targeting HER2, trastuzumab deruxtecan or T-DXd, is notable due to its success at expression levels ranging from high to low and ultralow. This raises the question of the relative contributions of target-independent mechanisms on ADC efficacy in the clinic, and several such mechanisms have been proposed. However, in vitro and preclinical data have different doses and exposures, making it challenging to quantitatively extrapolate preclinical data to the clinic. In this work, we use our computational hybrid agent-based model, SimADC, to simulate target-dependent and -independent mechanisms, scaling from mice to humans. We first demonstrate that CD8+ T cells can significantly contribute to tumor regression, especially when the ADC further activates the immune cells. Next, we test target-independent payload-driven mechanisms including: 1) Fc-mediated internalization of ADC by intratumoral macrophages and payload release to neighboring cancer cells, 2) free payload circulating in the blood and re-entering the tumor, and 3) extracellular linker cleavage and payload release due to an abundance of proteases in the tumor. We find that free payload in the blood and extracellular linker cleavage had low and moderate impacts, respectively, while macrophage uptake and payload release resulted in high levels of efficacy. This is due to the macrophages ability to sustain free payload in the tumor. Moderate and high HER2 expression were more efficacious than target-independent mechanisms. Overall, our simulations demonstrate that moderate to high HER2 expression, immune activation, or macrophage uptake and payload release are sufficient for T-DXd tumor regression. Additionally, SimADC provides a robust framework for modeling both target-dependent and target-independent mechanisms for any ADC, providing the opportunity to engineer more effective therapeutic agents. Author SummaryCancer is one of the most prevalent diseases in the world, impacting the lives of millions of people every year. Antibody-drug conjugates (ADCs) are a form of targeted therapy that can deliver cytotoxic drugs directly to cancer cells, increasing efficacy. However, ADCs are complex to design and test, as each part of the ADC (targeting antibody, cytotoxic payload, and linker) must be optimally selected for delivery for each target and type of patient. Here, we studied ADCs using a computational model, which allowed us to simulate ADCs in varying cancer environments efficiently and economically. We validated our model using preclinical data to incorporate patient immune responses, target-independent payload release, and systemic payload uptake, allowing us to make accurate predictions in mice and extrapolate to human tumors. We compared multiple mechanisms by which ADCs can kill cancer cells to help identify the most effective methods. Besides high target expression, immune stimulation and target-independent release in the microenvironment can contribute to tumor regression. Investigating these mechanisms enables the design of ADCs and treatment regimens that maximize efficacy across a range of tumor types and target expression.

Three-dimensional (3D) cell culture systems with tumor spheroids are being adopted for research on the antitumor activity of drug treatments and cytotoxic T cells. Analysis of the cytotoxic effect on 3D tumor cultures within a 3D scaffold, such as collagen, is challenging. Image-based approaches often use confocal microscopy, which greatly limits the sample size of tumor spheroids that can be assayed. We explored a system where tumor spheroids growing in a collagen gel within a microfluidics chip can be treated with drugs or co-cultured with T cells. We attempted to adapt the system to measure the death of cells in the tumor spheroids directly in the microfluidics chip via automated widefield fluorescence microscopy. We were able to successfully measure drug-induced cytotoxicity in tumor spheroids, but had difficulties extending the system to measure T cell-mediated tumor killing. O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=193 SRC="FIGDIR/small/842039v1_ufig1.gif" ALT="Figure 1"> View larger version (44K): org.highwire.dtl.DTLVardef@c95725org.highwire.dtl.DTLVardef@785728org.highwire.dtl.DTLVardef@a23dceorg.highwire.dtl.DTLVardef@187bd74_HPS_FORMAT_FIGEXP M_FIG C_FIG

The identification of prognostic biomarkers fuels personalized medicine. Here we tested two underlying, but often overlooked assumptions: 1) measurements at the steady state are sufficient for predicting the response to drug action, and 2) specifically, measurements of molecule abundances are sufficient. It is not clear that these are justified, as 1) the response results from non-linear molecular relationships, and 2) the steady state is defined by both abundance and orthogonal flux information. An experimentally validated mathematical model of the cellular response to the anti-cancer agent TRAIL was our test case. We developed a mathematical representation in which abundances and fluxes (static and kinetic network features) are largely independent, and simulated heterogeneous drug responses. Machine learning revealed predictive power, but that kinetic, not static network features were most informative. Analytical treatment of the underlying network motif identified kinetic buffering as the relevant circuit design principle. Our work suggests that network topology considerations ought to guide biomarker discovery efforts. Graphic abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=199 SRC="FIGDIR/small/452900v1_ufig1.gif" ALT="Figure 1"> View larger version (56K): org.highwire.dtl.DTLVardef@188edd2org.highwire.dtl.DTLVardef@b5b661org.highwire.dtl.DTLVardef@1d8af28org.highwire.dtl.DTLVardef@d384a9_HPS_FORMAT_FIGEXP M_FIG C_FIG Highlights- Biomarkers are usually molecule abundances but underlying networks are dynamic - Our method allows separate consideration of heterogeneous abundances and fluxes - For the TRAIL cell death network machine learning reveals fluxes as more predictive - Network motif analyses could render biomarker discovery efforts more productive eTOC blurbPrecision medicine relies on discovering which measurements of the steady state predict therapeutic outcome. Loriaux et al show - using a new analytical approach - that depending on the underlying molecular network, synthesis and degradation fluxes of regulatory molecules may be more predictive than their abundances. This finding reveals a flaw in an implicit but hitherto untested assumption of biomarker discovery efforts and suggests that dynamical systems modeling is useful for directing future clinical studies in precision medicine.

Background The effectiveness of many targeted therapies is limited by toxicity and the rise of drug resistance. A growing appreciation of the inherent redundancies of cancer signaling has led to a rise in the number of combination therapies under development, but a better understanding of the overall cancer network topology would provide a conceptual framework for choosing effective combination partners. In this work, we explore the scale-free nature of cancer protein-protein interaction networks in 14 indications. Scale-free networks, characterized by a power-law degree distribution, are known to be resilient to random attack on their nodes, yet vulnerable to directed attacks on their hubs (their most highly connected nodes).Results Consistent with the properties of scale-free networks, we find that lethal genes are associated with ∼5-fold higher protein connectivity partners than non-lethal genes. This provides a biological rationale for a hub-centered combination attack. Our simulations show that combinations targeting hubs can efficiently disrupt 50% of network integrity by inhibiting less than 1% of the connected proteins, whereas a random attack can require inhibition of more than 30% of the connected proteins.Conclusions We find that the scale-free nature of cancer networks makes them vulnerable to focused attack on their highly connected protein hubs. Thus, we propose a new strategy for designing combination therapies by targeting hubs in cancer networks that are not associated with relevant toxicity networks.Competing Interest StatementThe authors have declared no competing interest.View Full Text

Antibody-drug conjugates (ADCs) are novel therapeutics combining two molecules linked together: an antibody that provides targeting to specific cells, and a cytotoxic drug (warhead) that can deconjugate from the antibody and kill those cells. The warhead by itself would be too nonspecific and toxic; the antibody by itself would be insufficiently effective at killing the cells. As more and more ADCs enter the drug development pipeline, understanding the mechanistic reasons behind efficacy and toxicity is critical to evaluating both successful and failed trials. Here, we have developed a mechanistic computational model of ADCs, and specifically parameterize it using data for MEDI2228, an anti-BCMA antibody conjugated to pyrrolobenzodiazepine (PBD) warheads. We build the model to track not only the concentrations of ADC and its released warhead drug inside and outside the cell, but also to track the recent history of the warhead, so that we can distinguish between pathways for on-target and bystander (nontargeted) cell killing. We show that this effect is predicted to be small under in vitro conditions due to dilution in large volumes of media, but likely to form a significant part of both targeted and nontargeted cell killing in vivo, where the extracellular volume has less of a dilution effect. We also explore the impact of key design parameters of ADCs, including drug to antibody ratio (DAR), warhead potency, and lipophilicity; this analysis demonstrates the balance needed between killing of targeted and nontargeted cells. Using this quantitative systems pharmacology model, we can generate insights for optimization of ADC design and determine which factors are most critical to efficacy and toxicity, leading to more informed and rational development of cancer therapies.

As part of the first line of defense against pathogens, macrophages possess the ability to differentiate into divergent phenotypes with varying functions. The process by which these cells change their characteristics, commonly referred to as macrophage polarization, allows them to change into broadly pro-inflammatory (M1) or anti-inflammatory (M2) subtypes, and depends on the polarizing stimuli. Deregulation of macrophage phenotypes can result in different pathologies or affect the nature of some diseases, such as cancer and atherosclerosis. For this reason, it is necessary to better understand macrophage phenotype conversion in relevant models. However, there are few existing probes to track macrophage changes in multicellular environments. In this study, we generated an eGFP reporter cell line based on inducible nitric oxide synthase (iNos) promoter activity in RAW264.7 cells (RAW:iNos-eGFP). iNos is associated with macrophage activation to pro-inflammatory states, and decreases in immune-suppressing ones. We validated the fidelity of the reporter for iNos, including following cytokine-mediated polarization, and confirmed that reporter and parental cells behaved similarly. RAW:iNos-eGFP cells were then used to track macrophage responses in different in vitro breast cancer models, and their re-education from anti- to pro-inflammatory phenotypes via a previously reported pyrimido(5,4-b)indole small molecule, PBI1. Using two mouse mammary carcinoma cell lines, 4T1 and EMT6, effects on macrophages were assessed via conditioned media, two-dimensional/monolayer co-culture, and three-dimensional spheroid models. While conditioned media derived from 4T1 or EMT6 cells and monolayer co-cultures of each with RAW:iNos-eGFP cells all resulted in decreased fluorescence, the trends and extents of effects differed. We also observed decreases in iNos-eGFP signal in the macrophages in co-culture assays with 4T1- or EMT6-based spheroids. We then showed that we are able to enhance iNos production in the context of these cancer models using PBI1, tracking increased fluorescence. Taken together, we demonstrate that this reporter-based approach provides a facile means to study macrophage responses in complex, multicomponent environments. Beyond the initial studies presented here, this platform can be used with a variety of in vitro models and extended to in vivo applications with intravital imaging.

Anticancer drug resistance is challenging to overcome because it can arise through both intrinsic and acquired mechanisms, each driven by distinct cellular machinery. In particular, there is a sharp need for therapies that target hormone-insensitive prostate tumors due to the growing incidence of castration-resistant prostate cancer. Optimizing the pathways that regulate apoptosis in prostate cancer offers a promising strategy to induce apoptosis and inhibit tumor progression, since these mechanisms do not depend on hormonal signaling. Here, we identified strategies to enhance apoptosis in prostate cancer cells. We used several computational tools (including sensitivity analysis, particle swarm optimization, and ImageJ) to design an ordinary differential equation model of caspase-mediated prostate cancer apoptosis signaling. We apply the model to identify key modalities that increase the propensity toward apoptosis across three separate pro-apoptotic drugs (Tocopheryloxybutyrate, Narciclasine, and Celecoxib). Overall, we demonstrate that apoptosis dynamics can be accurately captured in response to each of the three drugs and identify which features of the model represent viable targets for overcoming intrinsic drug resistance.

Cancer immunotherapy using engineered cytotoxic effector cells has demonstrated significant potential. The limited spatial complexity of existing in vitro models, however, poses a challenge to mechanistic studies attempting to approve existing approaches of effector cell-mediated cytotoxicity within a three-dimensional, solid tumor-like environment. To gain additional experimental control, we developed an approach for constructing three-dimensional (3D) culture models using smart polymers that form temperature responsive hydrogels. By embedding cells in these hydrogels, we constructed 3D models to organize multiple cell populations at specified ratios on- demand and gently position them by exploiting the hydrogel phase transition. These systems were amenable to imaging at low- and high-resolution to evaluate cell-to-cell interactions, as well as to dissociation to allow for single cell analyses. We have called this approach "thermal collapse of strata" (TheCOS) and demonstrated its use in creating complex cell assemblies on demand in both layers and spheroids. As an application, we utilized TheCOS to evaluate the impact of directionality of degranulation of natural killer (NK) cell lytic granules. Blocking lytic granule convergence and polarization by inhibiting dynein has been shown to induce bystander killing in single cell suspensions. Using TheCOS we showed that lytic granule dispersion induced by dynein inhibition can be sustained in 3D and results in a multi-directional killing including that of non-triggering bystander cells. By imaging TheCOS experiments, we were able to map a "kill zone" associated with multi-directional degranulation in simulated solid tumor environments. TheCOS should allow for the testing of approaches to alter the mechanics of cytotoxicity as well as to generate a wide-array of human tumor microenvironments to assist in the acceleration of tumor immunotherapy.

Angiogenesis, the formation of new vessels from existing vessels, is mediated by vascular endothelial growth factor (VEGF) and platelet-derived growth factor (PDGF). Despite discoveries supporting the cross-family interactions between VEGF and PDGF families, sharing the binding partners between them makes it challenging to identify growth factors that predominantly affect angiogenesis. Systems biology offers promises to untangle this complexity. Thus, in this study, we developed a mass-action kinetics-based computational model for cross-family interactions between VEGFs (VEGF-A, VEGF-B, and PlGF) and PDGFs (PDGF-AA, PDGF-AB, and PDGF-BB) with their receptors (VEGFR1, VEGFR2, NRP1, PDGFR, and PDGFR{beta}). The model, parametrized with our literature mining and surface resonance plasmon assays, was validated by comparing the concentration of VEGFR1 complexes with a previously constructed angiogenesis model. The model predictions include five outcomes: 1) the percentage of free or bound ligands and 2) receptors, 3) the concentration of free ligands, 4) the percentage of ligands occupying each receptor, and 5) the concentration of ligands that is bound to each receptor. We found that at equimolar ligand concentrations (1 nM), PlGF and VEGF-A were the main binding partners of VEGFR1 and VEGFR2, respectively. Varying the density of receptors resulted in the following five outcomes: 1) Increasing VEGFR1 density depletes the free PlGF concentration, 2) increasing VEGFR2 density decreases PDGF:PDGFR complexes, 3) increased NRP1 density generates a biphasic concentration of the free PlGF, 4) increased PDGFR density increases PDGFs:PDGFR binding, and 5) increasing PDGFR{beta} density increases VEGF-A:PDGFR{beta}. Our model offers a reproducible, fundamental framework for exploring cross-family interactions that can be extended to the tissue level or intracellular molecular level. Also, our model may help develop therapeutic strategies in pathological angiogenesis by identifying the dominant complex in the cell signaling. Author summaryNew blood vessel formation from existing ones is essential for growth, healing, and reproduction. However, when this process is disrupted--either too much or too little--it can contribute to diseases such as cancer and peripheral arterial disease. Two key families of proteins, vascular endothelial growth factors (VEGFs) and platelet-derived growth factors (PDGFs), regulate this process. Traditionally, scientists believed that VEGFs only bind to VEGF receptors and PDGFs to PDGF receptors. However, recent findings show that these proteins can interact with each others receptors, making it more challenging to understand and control blood vessel formation. To clarify these complex interactions, we combined computer modeling with biological data to map out which proteins bind to which receptors and to what extent. Our findings show that when VEGFs and PDGFs are present in equal amounts, VEGFs are the primary binding partners for VEGF receptors. We also explored how changes in receptor levels affect these interactions in disease-like conditions. This work provides a foundational computational model for studying cross-family interactions, which can be expanded to investigate tissue-level effects and processes inside cells. Ultimately, our model may help develop better treatments for diseases linked to abnormal blood vessel growth by identifying key protein-receptor interactions.

BackgroundEpithelial-mesenchymal transition (EMT) and its reverse process Mesenchymal-Epithelial Transition (MET) are crucial during metastasis and therapy resistance. While the dynamics and master regulators of EMT are well-studied, the transcription factors that can prevent EMT or promote MET are relatively less understood. ResultsHere, by integrating bulk and spatial transcriptomic data analysis from cell lines and patient samples, with mechanism-based dynamical modelling, we identify IRF6 as a factor that strongly associates with an epithelial phenotype and is often inhibited during EMT. In vitro experiments in multiple cancer cell lines demonstrate the progression to a mesenchymal phenotype upon IRF6 knock-down, suggesting a role as an inhibitor of EMT. Finally, we observe that IRF6 expression levels correlates with worse patient survival in a subset of solid tumour types. ConclusionOur integrated computational-experimental systems-level analysis suggests that IRF6 is frequently downregulated during EMT and can also prevent the progression towards a complete EMT, underscoring its role as an MET stabilizing factor.

Cytotoxic T lymphocytes (CTLs) are important in controlling some viral infections, and therapies involving transfer of large numbers of cancer-specific CTLs have been successfully used to treat several types of cancers in humans. While molecular mechanisms of how CTLs kill their targets are relatively well understood we still lack solid quantitative understanding of the kinetics and efficiency at which CTLs kill their targets in different conditions. Collagen-fibrin gel-based assays provide a tissue-like environment for the migration of CTLs, making them an attractive system to study the cytotoxicity in vitro. Budhu et al. [1] systematically varied the number of peptide (SIINFEKL)- pulsed B16 melanoma cells and SIINFEKL-specific CTLs (OT-1) and measured remaining targets at different times after target and CTL co-inoculation into collagen-fibrin gels. The authors proposed that their data were consistent with a simple model in which tumors grow exponentially and are killed by CTLs at a per capita rate proportional to the CTL density in the gel. By fitting several alternative mathematical models to these data we found that this simple "exponential-growth-mass-action-killing" model does not precisely fit the data. However, determining the best fit model proved difficult because the best performing model was dependent on the specific dataset chosen for the analysis. When considering all data that include biologically realistic CTL concentrations (E [≤] 107 cell/ml) the model in which tumors grow exponentially and CTLs suppress tumors growth non-lytically and kill tumors according to the mass-action law (SiGMA model) fitted the data with best quality. Results of power analysis suggested that longer experiments ([~] 3 - 4 days) with 4 measurements of B16 tumor cell concentrations for a range of CTL concentrations would best allow to discriminate between alternative models. Taken together, our results suggest that interactions between tumors and CTLs in collagen-fibrin gels are more complex than a simple exponential-growth- mass-action killing model and provide support for the hypothesis that CTLs impact on tumors may go beyond direct cytotoxicity.

Cells sense the stiffness of their extracellular matrix (ECM) and adapt their behavior accordingly. We investigated how ECM stiffness affects the spatial organization of talin1, a key mechanosensitive focal adhesion protein. Using polyacrylamide (PA) hydrogels with tunable stiffnesses (0.2-188 kPa), we analyzed cell morphology, migration, talin1 distribution, colocalization with tensin3, and fibronectin deposition. Softer substrates enhanced filopodia activity and altered migration behavior. On softer ECMs, talin1 displayed a more even intracellular distribution, whereas on stiffer matrices it localized to the cell periphery. PA gels supported elongated talin1-based adhesions, whose morphology showed minimal variation across the 3-188 kPa stiffness range. Talin1-tensin3 colocalization was maintained regardless of stiffness, indicating a stable interaction. Notably, cells deposited more fibronectin on softer substrates. While talin1 adhesion morphology varied little with stiffness, cell migration behavior changed markedly. Combined with prior studies, our data suggests that ECM stiffness regulates talin1 primarily through conformational changes rather than macroscopic adhesion remodeling. These findings highlight talin1s central role in translating mechanical cues into dynamic cellular responses. Summary statementTalin1 forms elongated adhesions and robustly colocalizes with tensin3 across varying matrix stiffnesses, showing that their spatial organization is largely insensitive to mechanical cues.

Fibrosis is involved in 45% of deaths in the United States, with no treatment options to reverse progression of the disease. Implantable devices (such as joint replacements, chips, stents, artificial organs, biosensors, catheters, heart valves, scaffolds for tissue engineering, etc.) can trigger a foreign body response, in which fibrotic tissue covers the implant and impedes its function. Myofibroblast are a key cellular component of scar tissue. To explore the relationship between extracellular matrix-based coatings and fibrosis, we coated tissue-culture surfaces using a library of extracellular matrix (ECM) proteins and then performed an in-vitro screen for myofibroblast differentiation on the coated surfaces as an indicator of fibrotic potential. The protein and proteoglycan components of cartilage (collagen II, biglycan, decorin, and chondroitin sulfate) were individually anti-fibrotic. Further, mixtures of collagen II, biglycan, decorin, and chondroitin sulfate inhibited myofibroblast differentiation to a greater degree than collagen II, biglycan, decorin, or chondroitin sulfate as individual coatings. Next, we performed an in-vivo model of a foreign-body response. Implanting an uncoated implant subcutaneously into mice resulted in a thicker layer of fibrotic scar tissue than implants coated with a cartilage-like ECM mixture. Our results indicate that the ECM microenvironment is key to the initiation, progression, and maintenance of fibrosis.

The interplay between the extracellular matrix (ECM) mechanical properties and the tumor microenvironment is increasingly recognized as a critical factor in cancer progression. Three-dimensional (3D) culture systems have emerged as essential platforms for in-vitro cell-based applications, offering microenvironments that are more physiologically relevant compared to traditional two-dimensional (2D) cultures. However, independently controlling the topological and mechanical features of 3D matrices remains challenging due to the interdependence of these parameters. In this study, we demonstrate a method for independently tuning pore size and stiffness in collagen I (Coll I) networks and examine their effects on breast cancer and epithelial cell morphology and cluster formation. Collagen concentration was used to modulate bulk stiffness, while polymerization temperature was adjusted to control pore size. Using this approach, we developed a 3D Coll I matrix with tuned stiffnesses from 80, 228 and 360 Pa while simultaneously holding pore size constant (2.5 {micro}m). Similarly, we developed a low- (1.5 mg/mL) and high- (3.5 mg/mL) concentration collagen hydrogel with varying pore sizes from 2.5 {micro}m to 3.1 {micro}m and 2.0 {micro}m to 2.4 {micro}m, respectively, without altering stiffness (80 Pa and 350 Pa). Integrating a breast epithelial cell line, MCF-10A, and metastatic breast cancer cell line, MDA-MB-231, we demonstrate matrix stiffness and pore size independently and differentially regulate cell morphology and cluster formation. Our results establish a robust method for decoupling stiffness and pore size in Coll I matrices enabling more precise investigations into how ECM mechanical properties influence metastatic and epithelial cell behavior. Statement of SignificanceThis study presents a robust method to independently tune stiffness and pore size in 3D collagen I matrices, overcoming a key challenge in extracellular matrix modeling. By decoupling these parameters through collagen concentration and polymerization temperature, the platform enables more accurate investigation of how ECM mechanical properties influence metastatic and epithelial cell behavior. Our finding reveals that matrix stiffness and pore size independently and differentially regulate cell morphology and cluster formation, demonstrating the distinct cellular responses to specific ECM properties and underscoring the importance of the tumor microenvironment in cancer biology and tissue engineering.

BackgroundDespite improved cancer outcomes with kinase inhibitors (KIs), their cardiotoxicity remains a significant clinical challenge. Current approaches to predict and prevent KI-induced cardiac adverse events (CAEs) are limited by an incomplete understanding of underlying mechanisms, including the contribution of off-target kinase engagement. ObjectivesTo establish links between kinase inhibition profiles and cardiotoxic phenotypes using empirical proteomic data, and to leverage these profiles in machine learning (ML) models capable of predicting KI cardiotoxicity. MethodsWe curated a database connecting kinome-wide target binding profiles of FDA-approved KIs (n=44) with documented incidence rates of six distinct CAEs. Binding profiles were derived from unbiased chemoproteomics and used to assess associations between KI selectivity, specific kinase targets, and CAEs. Profiles were further used to develop ML models to predict CAE risk, with SHAP-based model interpretation applied to identify cardiotoxicity-associated kinases. ResultsKI promiscuity was not a significant predictor of cardiotoxicity across all six CAEs. Frequency analysis revealed that kinases including RET, PDGFRB, and DDR1 are recurrently inhibited across CAE-linked compounds, with nearly all identified as off-targets not annotated by the FDA. Network and pathway enrichment analyses supported a systems-level model in which cardiotoxicity arises from coordinated disruption of cardiac-relevant signaling networks. ML models achieved 66-84% cross-validated accuracy (ROC-AUC 0.75-0.8) across CAE endpoints, with SHAP analysis identifying PDGFRB, EGFR, and MEK1/2 among the most predictive kinases. ConclusionsProteomic kinome profiling combined with machine learning provides a mechanistically grounded framework for predicting KI cardiotoxicity and supports off-target-aware drug design to minimize cardiovascular risk.

Although several cancer cells are shed into the circulation, most die due to the stresses they encounter. Detachment from the underlying extracellular matrix is one major stress; cell death due to matrix detachment is known as anoikis. Few cancer cells overcome this stress, become anoikis-resistant, and survive to seed metastasis. Autophagy, a cell survival mechanism during stressful conditions that promotes cellular homeostasis by recycling cellular components, is activated upon matrix detachment. On the other hand, apoptosis is a cellular mechanism that is responsible for programmed cell death which is also activated upon matrix detachment. It is unclear how matrix-deprived cancer cells maintain a balance between autophagy and apoptosis to decide the cell fate: whether the cell dies due to anoikis or acquires anoikis-resistance and survives to seed metastasis. Though multiple pathways contribute to cell fate decisions, we have shown experimentally that autophagy and apoptosis are influenced by Akt-AMPK axis and AMPK-ERK axis in matrix-deprived cancer cells. Since Akt, AMPK and ERK are in turn linked to each other it is essential to understand how these proteins simultaneously affect the cell signaling and survival/death outcomes of matrix-deprived cancer cells. To study the cumulative effect of Akt-AMPK-ERK activities on survival/death decisions and to understand the fine balance between apoptosis and autophagy that facilitates the survival/death of matrix-deprived cancer cells, we formulated a deterministic ordinary differential equation (ODE)-based protein interaction model of anoikis resistance. Model stability analysis and 3D-nullcline analysis depicted that the system has a unique steady state in matrix-attached and matrix-deprived condition. Parameter sensitivity analysis depicted that the model is highly robust, and the model variables are sensitive to only a few model parameters. By simulating differential activity of pAkt, pAMPK and pERK, the model predicted a heterogeneity in pERK levels: high/low levels pERK along with high pAMPK enable survival as long as levels of pAkt are maintained low. Additionally, the model predicted a heterogeneity in pAMPK: high/low levels of pAMPK along with low pERK determines the shift from survival to death when levels of pAkt are high. Such high levels of pAkt are obtained at critically low levels of pERK. Molecular perturbation revealed a hierarchy among proteins while deciding the cell fate: pAkt dominates over pAMPK which further dominates over pERK and intermediate to high levels of pAkt were sufficient for apoptosis to surpass autophagy in matrix-detached cells. The model also predicted that Akt impacts apoptosis more than autophagy and classified the cell fate decision into survival and death zones in matrix-deprived condition. Overall, this work provided multiple insights on the molecular interplay among key kinases Akt, AMPK and ERK and their effects on apoptosis and autophagy. This model also depicted that autophagosome formation is rather robust as compared to apoptosis which is more sensitive to molecular perturbations. Hence, apoptosis emerged as a deciding factor that influences the decision of cell fate of a matrix-deprived cancer cell.

Sprouting angiogenesis is orchestrated by an intricate balance of biochemical and mechanical cues in the local microenvironment. Interstitial flow has been established as a potent regulator of angiogenesis. Similarly, extracellular matrix (ECM) physical properties, such as stiffness and microarchitecture, have also emerged as important mediators of angiogenesis. Yet, the interplay between interstitial flow and ECM physical properties in the initiation and control of angiogenesis is poorly understood. Using a 3-D microfluidic tissue analogue of angiogenic sprouting with defined interstitial flow, we found that the addition of hyaluronan (HA) to collagen-based matrices significantly enhances sprouting induced by interstitial flow compared to responses in collagen-only hydrogels. We confirmed that both the stiffness and matrix pore size of collagen-only hydrogels were increased by the addition of HA. Interestingly, interstitial flow-potentiated sprouting responses in collagen/HA matrices were not affected when functionally blocking the HA receptor CD44. In contrast, enzymatic depletion of HA in collagen/HA matrices with hyaluronidase (HAdase) resulted in decreased stiffness, pore size, and interstitial flow-mediated sprouting to the levels observed in collagen-only matrices. Taken together, these results suggest that HA enhances interstitial flow-mediated angiogenic sprouting through its alterations to collagen ECM stiffness and pore size.

Tumor necrosis factor alpha (TNF), a pleiotropic cytokine, helps maintain a balance between proliferation and apoptosis in normal cells. This balance is often sacrificed in a diseased cell, such as that of a cancer, by preferring survival phenotype over apoptosis. Restoring this balance requires a detailed understanding of the causal intracellular mechanisms that govern TNF stimulated apoptotic response. In this study, we use a systems biology approach to unravel the interplay between the intracellular signaling markers that orchestrate apoptosis levels. Our approach deciphered the synergism between the early intracellular markers phosphorylated JNK (pJNK) and phosphorylated AKT (pAKT) that modulate the activation of Caspase3, an important apoptotic regulator. We demonstrate that this synergism depends critically on the survival pathway signaling mediated by NF{kappa}B which plays a dominant role in controlling the extent of the overall apoptotic response. By systematic inhibition of the signaling markers, we establish that the dynamic cross-talk between the pJNK and pAKT transients directs the apoptosis phenotype via accumulated Caspase3 response. Interestingly, superposition of the semi-quantitative correlation between apoptosis and Caspase3 transient levels on the proposed TNF network model permits quantification of the dynamic apoptotic response under different stimulation conditions. Thus, the predictive model can be leveraged towards arriving at useful insights that can identify potential targeted therapeutic strategies for altering apoptotic response.

Hypoxia, a low level of oxygen in the tissue, is a feature of most solid tumors. It arises due to an imbalance between the oxygen supply from the abnormal vasculature and oxygen demand by the large number of tumor and stromal cells. Hypoxia has been implicated in the development of aggressive tumors and tumor resistance to various therapies. This makes hypoxia a negative marker of patients survival. However, recent advances in designing new hypoxia-activated pro-drugs and adoptive T cell therapies provide an opportunity for exploiting hypoxia in order to improve cancer treatment. We used novel mathematical models of micro-pharmacology and computational optimization techniques for determining the most effective treatment protocols that take advantage of heterogeneous and dynamically changing oxygenation in in vivo tumors. These models were applied to design schedules for a combination of three therapeutic compounds in pancreatic cancers and determine the most effective adoptive T cell therapy protocols in melanomas.