Emergent Gene Expression Responses to Drug Combinations Predict Higher-Order Drug Interactions

Elucidating Compound Mechanism of Action and Polypharmacology with a Large-scale Perturbational Profile Compendium

The Mechanism of Action (MoA) of a drug is generally represented as a small, non-tissue-specific repertoire of high-affinity binding targets. Yet, drug activity and polypharmacology are increasingly associated with a broad range of off-target and tissue-specific effector proteins. To address this challenge, we have leveraged a microfluidics-based Plate-Seq technology to survey drug perturbational profiles representing >700 FDA-approved and experimental oncology drugs, in cell lines selected as high-fidelity models of 23 aggressive tumor subtypes. Built on this dataset, we implemented an efficient computational framework to define a tissue-specific protein activity landscape of these drugs and reported almost 50 million differential protein activities derived from drug perturbations vs. vehicle controls. These analyses revealed thousands of highly reproducible and novel, drug-mediated modulation of tissue-specific targets, leading to generation of a proteome-wide drug functional network, characterization of MoA-related drug clusters and off-target effects, dramatical expansion of druggable human proteome, and identification and experimental validation of novel, tissue-specific inhibitors of undruggable oncoproteins, most never reported before. The drug perturbation profile resource described here represents the first, large-scale, whole-genome-wide, RNA-Seq based dataset assembled to date, with the proposed framework, which is easily extended to elucidating the MoA of novel small-molecule libraries, facilitates mechanistic exploration of drug functions, supports systematic and quantitative approaches to precision oncology, and serves as a rich data foundation for drug discovery.

A hybrid deep forest-based method for predicting synergistic drug combinations

Combination therapy is a promising approach in treating multiple complex diseases. However, the large search space of available drug combinations exacerbates challenge for experimental screening. To predict synergistic drug combinations in different cancer cell lines, we propose an improved deep forest-based method, ForSyn, and design two forest types embedded in ForSyn. ForSyn handles imbalanced and high-dimensional data in medium-/small-scale datasets, which are inherent characteristics of drug combination datasets. Compared with 12 state-of-the-art methods, ForSyn ranks first on four metrics for eight datasets with different feature combinations. We conduct a systematic analysis to identify the most appropriate configuration parameters. We validate the predictive value of ForSyn with cell-based experiments on several previously unexplored drug combinations. Finally, a systematic analysis of feature importance is performed on the top contributing features extracted by ForSyn. The resulting key genes may play key roles on corresponding cancers.

Sample-efficient identification of high-dimensional antibiotic synergy with a normalized diagonal sampling design

Antibiotic resistance is an important public health problem. One potential solution is the development of synergistic antibiotic combinations, in which the combination is more effective than the component drugs. However, experimental progress in this direction is severely limited by the number of samples required to exhaustively test for synergy, which grows exponentially with the number of drugs combined. We introduce a new metric for antibiotic synergy, motivated by the popular Fractional Inhibitory Concentration Index and the Highest Single Agent model. We also propose a new experimental design that samples along all appropriately normalized diagonals in concentration space, and prove that this design identifies all synergies among a set of drugs while only sampling a small fraction of the possible combinations. We applied our method to screen two- through eight-way combinations of eight antibiotics at 10 concentrations each, which requires sampling only 2,560 unique combinations of antibiotic concentrations.

Antibiotic resistance is a growing public health concern, and there is an increasing need for methods to combat it. One potential approach is the development of synergistic antibiotic combinations, in which a mixture of drugs is more effective than any individual component. Unfortunately, the search for clinically beneficial drug combinations is severely restricted by the pace at which drugs can be screened. To date, most studies of combination therapies have been limited to testing only pairs or triples of drugs. These studies have identified primarily antagonistic drug interactions, in which the combination is less effective than the individual components. There is an acute need for methodologies that enable screening of higher-order drug combinations, both to identify synergies among many drugs and to understand the behavior of higher-order combinations. In this work we introduce a new paradigm for combination testing, the normalized diagonal sampling design, that makes identifying interactions among eight or more drugs feasible for the first time. Screening d drugs at m different combinations requires m ⋅ 2^d samples under our design as opposed to m^d under exhaustive screening, while provably identifying all synergies under mild assumptions about antibiotic behavior. Scientists can use our design to quickly screen for antibiotic interactions, accelerating the pace of combination therapy development.

Expanding the search for small-molecule antibacterials by multidimensional profiling

New techniques for systematic profiling of small-molecule effects can enhance traditional growth inhibition screens for antibiotic discovery and change how we search for new antibacterial agents. Computational models that integrate physicochemical compound properties with their phenotypic and molecular downstream effects can not only predict efficacy of molecules yet to be tested, but also reveal unprecedented insights on compound modes of action (MoAs). The unbiased characterization of compounds that themselves are not growth inhibitory but exhibit diverse MoAs, can expand antibacterial strategies beyond direct inhibition of core essential functions. Early and systematic functional annotation of compound libraries thus paves the way to new models in the selection of lead antimicrobial compounds. In this Review, we discuss how multidimensional small-molecule profiling and the ever-increasing computing power are accelerating the discovery of unconventional antibacterials capable of bypassing resistance and exploiting synergies with established antibacterial treatments and with protective host mechanisms.

Intron-mediated induction of phenotypic heterogeneity

Intragenic regions that are removed during maturation of the RNA transcript—introns—are universally present in the nuclear genomes of eukaryotes¹. The budding yeast, an otherwise intron-poor species, preserves two sets of ribosomal protein genes that differ primarily in their introns^2,3. Although studies have shed light on the role of ribosomal protein introns under stress and starvation^4–6, understanding the contribution of introns to ribosome regulation remains challenging. Here, by combining isogrowth profiling⁷ with single-cell protein measurements⁸, we show that introns can mediate inducible phenotypic heterogeneity that confers a clear fitness advantage. Osmotic stress leads to bimodal expression of the small ribosomal subunit protein Rps22B, which is mediated by an intron in the 5′ untranslated region of its transcript. The two resulting yeast subpopulations differ in their ability to cope with starvation. Low levels of Rps22B protein result in prolonged survival under sustained starvation, whereas high levels of Rps22B enable cells to grow faster after transient starvation. Furthermore, yeasts growing at high concentrations of sugar, similar to those in ripe grapes, exhibit bimodal expression of Rps22B when approaching the stationary phase. Differential intron-mediated regulation of ribosomal protein genes thus provides a way to diversify the population when starvation threatens in natural environments. Our findings reveal a role for introns in inducing phenotypic heterogeneity in changing environments, and suggest that duplicated ribosomal protein genes in yeast contribute to resolving the evolutionary conflict between precise expression control and environmental responsiveness⁹.

Experiments in yeast show that introns have a role in inducing phenotypic heterogeneity and that intron-mediated regulation of ribosomal proteins confers a fitness advantage by enabling yeast populations to diversify under nutrient-scarce conditions.

An enhanced cascade-based deep forest model for drug combination prediction

Combination therapy has shown an obvious curative effect on complex diseases, whereas the search space of drug combinations is too large to be validated experimentally even with high-throughput screens. With the increase of the number of drugs, artificial intelligence techniques, especially machine learning methods, have become applicable for the discovery of synergistic drug combinations to significantly reduce the experimental workload. In this study, in order to predict novel synergistic drug combinations in various cancer cell lines, the cell line-specific drug-induced gene expression profile (GP) is added as a new feature type to capture the cellular response of drugs and reveal the biological mechanism of synergistic effect. Then, an enhanced cascade-based deep forest regressor (EC-DFR) is innovatively presented to apply the new small-scale drug combination dataset involving chemical, physical and biological (GP) properties of drugs and cells. Verified by the dataset, EC-DFR outperforms two state-of-the-art deep neural network-based methods and several advanced classical machine learning algorithms. Biological experimental validation performed subsequently on a set of previously untested drug combinations further confirms the performance of EC-DFR. What is more prominent is that EC-DFR can distinguish the most important features, making it more interpretable. By evaluating the contribution of each feature type, GP feature contributes 82.40%, showing the cellular responses of drugs may play crucial roles in synergism prediction. The analysis based on the top contributing genes in GP further demonstrates some potential relationships between the transcriptomic levels of key genes under drug regulation and the synergism of drug combinations.

Price equation captures the role of drug interactions and collateral effects in the evolution of multidrug resistance

Bacterial adaptation to antibiotic combinations depends on the joint inhibitory effects of the two drugs (drug interaction [DI]) and how resistance to one drug impacts resistance to the other (collateral effects [CE]). Here we model these evolutionary dynamics on two-dimensional phenotype spaces that leverage scaling relations between the drug-response surfaces of drug-sensitive (ancestral) and drug-resistant (mutant) populations. We show that evolved resistance to the component drugs - and in turn, the adaptation of growth rate - is governed by a Price equation whose covariance terms encode geometric features of both the two-drug-response surface (DI) in ancestral cells and the correlations between resistance levels to those drugs (CE). Within this framework, mean evolutionary trajectories reduce to a type of weighted gradient dynamics, with the drug interaction dictating the shape of the underlying landscape and the collateral effects constraining the motion on those landscapes. We also demonstrate how constraints on available mutational pathways can be incorporated into the framework, adding a third key driver of evolution. Our results clarify the complex relationship between drug interactions and collateral effects in multidrug environments and illustrate how specific dosage combinations can shift the weighting of these two effects, leading to different and temporally explicit selective outcomes.

Hidden suppressive interactions are common in higher-order drug combinations

The rapid increase of multi-drug resistant bacteria has led to a greater emphasis on multi-drug combination treatments. However, some combinations can be suppressive—that is, bacteria grow faster in some drug combinations than when treated with a single drug. Typically, when studying interactions, the overall effect of the combination is only compared with the single-drug effects. However, doing so could miss “hidden” cases of suppression, which occur when the highest order is suppressive compared with a lower-order combination but not to a single drug. We examined an extensive dataset of 5-drug combinations and all lower-order—single, 2-, 3-, and 4-drug—combinations. We found that a majority of all combinations—54%—contain hidden suppression. Examining hidden interactions is critical to understanding the architecture of higher-order interactions and can substantially affect our understanding and predictions of the evolution of antibiotic resistance under multi-drug treatments.

•

Most instances of suppressive interactions are missed by standard methods

•

A majority (54%) of all antibiotic combinations tested contain hidden suppression

•

Identifying hidden suppression can affect what combinations should be used

Quantifying absolute gene expression profiles reveals distinct regulation of central carbon metabolism genes in yeast

In addition to controlled expression of genes by specific regulatory circuits, the abundance of proteins and transcripts can also be influenced by physiological states of the cell such as growth rate and metabolism. Here we examine the control of gene expression by growth rate and metabolism, by analyzing a multi-omics dataset consisting of absolute-quantitative abundances of the transcriptome, proteome, and amino acids in 22 steady-state yeast cultures. We find that transcription and translation are coordinately controlled by the cell growth rate via RNA polymerase II and ribosome abundance, but they are independently controlled by nitrogen metabolism via amino acid and nucleotide availabilities. Genes in central carbon metabolism, however, are distinctly regulated and do not respond to the cell growth rate or nitrogen metabolism as all other genes. Understanding these effects allows the confounding factors of growth rate and metabolism to be accounted for in gene expression profiling studies.

Extreme Antagonism Arising from Gene-Environment Interactions

Antagonistic interactions in biological systems, which occur when one perturbation blunts the effect of another, are typically interpreted as evidence that the two perturbations impact the same cellular pathway or function. Yet, this interpretation ignores extreme antagonistic interactions wherein an otherwise deleterious perturbation compensates for the function lost because of a prior perturbation. Here, we report on gene-environment interactions involving genetic mutations that are deleterious in a permissive environment but beneficial in a specific environment that restricts growth. These extreme antagonistic interactions constitute gene-environment analogs of synthetic rescues previously observed for gene-gene interactions. Our approach uses two independent adaptive evolution steps to address the lack of experimental methods to systematically identify such extreme interactions. We apply the approach to Escherichia coli by successively adapting it to defined glucose media without and with the antibiotic rifampicin. The approach identified multiple mutations that are beneficial in the presence of rifampicin and deleterious in its absence. The analysis of transcription shows that the antagonistic adaptive mutations repress a stringent response-like transcriptional program, whereas nonantagonistic mutations have an opposite transcriptional profile. Our approach represents a step toward the systematic characterization of extreme antagonistic gene-drug interactions, which can be used to identify targets to select against antibiotic resistance.

Universal Nature of Drug Treatment Responses in Drug-Tissue-Wide Model-Animal Experiments Using Tensor Decomposition-Based Unsupervised Feature Extraction

Gene expression profiles of tissues treated with drugs have recently been used to infer clinical outcomes. Although this method is often successful from the application point of view, gene expression altered by drugs is rarely analyzed in detail, because of the extremely large number of genes involved. Here, we applied tensor decomposition (TD)-based unsupervised feature extraction (FE) to the gene expression profiles of 24 mouse tissues treated with 15 drugs. TD-based unsupervised FE enabled identification of the common effects of 15 drugs including an interesting universal feature: these drugs affect genes in a gene-group-wide manner and were dependent on three tissue types (neuronal, muscular, and gastroenterological). For each tissue group, TD-based unsupervised FE enabled identification of a few tens to a few hundreds of genes affected by the drug treatment. These genes are distinctly expressed between drug treatments and controls as well as between tissues in individual tissue groups and other tissues. We also validated the assignment of genes to individual tissue groups using multiple enrichment analyses. We conclude that TD-based unsupervised FE is a promising method for integrated analysis of gene expression profiles from multiple tissues treated with multiple drugs in a completely unsupervised manner.

SynergyFinder 2.0: visual analytics of multi-drug combination synergies

SynergyFinder (https://synergyfinder.fimm.fi) is a stand-alone web-application for interactive analysis and visualization of drug combination screening data. Since its first release in 2017, SynergyFinder has become a widely used web-tool both for the discovery of novel synergistic drug combinations in pre-clinical model systems (e.g. cell lines or primary patient-derived cells), and for better understanding of mechanisms of combination treatment efficacy or resistance. Here, we describe the latest version of SynergyFinder (release 2.0), which has extensively been upgraded through the addition of novel features supporting especially higher-order combination data analytics and exploratory visualization of multi-drug synergy patterns, along with automated outlier detection procedure, extended curve-fitting functionality and statistical analysis of replicate measurements. A number of additional improvements were also implemented based on the user requests, including new visualization and export options, updated user interface, as well as enhanced stability and performance of the web-tool. With these improvements, SynergyFinder 2.0 is expected to greatly extend its potential applications in various areas of multi-drug combinatorial screening and precision medicine.