Interactions among drugs play a critical role in the killing efficacy of multi-drug treatments. Recent advances in theory and experiment for three-drug interactions enable the search for emergent interactions-ones not predictable from pairwise interactions. Previous work has shown it is easier to detect synergies and antagonisms among pairwise interactions when a rescaling method is applied to the interaction metric. However, no study has carefully examined whether new types of normalization might be needed for emergence. Here, we propose several rescaling methods for enhancing the classification of the higher order drug interactions based on our conceptual framework. To choose the rescaling that best separates synergism, antagonism and additivity, we conducted bacterial growth experiments in the presence of single, pairwise and triple-drug combinations among 14 antibiotics. We found one of our rescaling methods is far better at distinguishing synergistic and antagonistic emergent interactions than any of the other methods. Using our new method, we find around 50% of emergent interactions are additive, much less than previous reports of greater than 90% additivity. We conclude that higher order emergent interactions are much more common than previously believed, and we argue these findings for drugs suggest that appropriate rescaling is crucial to infer higher order interactions.

Background

In drug-drug interactions, there are surprising cases in which the growth inhibition of bacteria by a single antibiotic decreases when a second antibiotic is added. These interactions are termed suppressive and have been argued to have the potential to limit the evolution of resistance. Nevertheless, little attention has been given to suppressive interactions because clinical studies typically search for increases in killing efficiency and because suppressive interactions are believed to be rare based on pairwise studies.

Results

Here, we quantify the effects of single-, double-, and triple-drug combinations from a set of 14 antibiotics and 3 bacteria strains, totaling 364 unique three-drug combinations per bacteria strain. We find that increasing the number of drugs can increase the prevalence of suppressive interactions: 17% of three-drug combinations are suppressive compared to 5% of two-drug combinations in this study. Most cases of suppression we find (97%) are "hidden" cases for which the triple-drug bacterial growth is less than the single-drug treatments but exceeds that of a pairwise combination.

Conclusions

We find a surprising number of suppressive interactions in higher-order drug combinations. Without examining lower-order (pairwise) bacterial growth, emergent suppressive effects would be missed, potentially affecting our understanding of evolution of resistance and treatment strategies for resistant pathogens. These findings suggest that careful examination of the full factorial of drug combinations is needed to uncover suppressive interactions in higher-order combinations.

Interactions and emergent processes are essential for research on complex systems involving many components. Most studies focus solely on pairwise interactions and ignore higher-order interactions among three or more components. To gain deeper insights into higher-order interactions and complex environments, we study antibiotic combinations applied to pathogenic Escherichia coli and obtain unprecedented amounts of detailed data (251 two-drug combinations, 1512 three-drug combinations, 5670 four-drug combinations, and 13608 five-drug combinations). Directly opposite to previous assumptions and reports, we find higher-order interactions increase in frequency with the number of drugs in the bacteria's environment. Specifically, as more drugs are added, we observe an elevated frequency of net synergy (effect greater than expected based on independent individual effects) and also increased instances of emergent antagonism (effect less than expected based on lower-order interaction effects). These findings have implications for the potential efficacy of drug combinations and are crucial for better navigating problems associated with the combinatorial complexity of multi-component systems.

Rationale

Respiratory viral infections can result in the establishment of chronic lung diseases. Understanding the early innate immune mechanisms that participate in the development of chronic postviral lung disease may reveal new targets for therapeutic intervention. The intracellular viral sensor protein melanoma differentiation-associated protein 5 (MDA5) sustains the acute immune response to Sendai virus, a mouse pathogen that causes chronic lung inflammation, but its role in the development of postviral chronic lung disease is unknown.

Objectives

To establish the role of MDA5 in the development of chronic lung disease.

Methods

MDA5-deficient or control mice were infected with Sendai virus. The acute inflammatory response was evaluated by profiling chemokine and cytokine expression and by characterizing the composition of the cellular infiltrate. The impact of MDA5 on chronic lung pathology and function was evaluated through histological studies, degree of oxygen saturation, and responsiveness to carbachol.

Measurements and main results

MDA5 deficiency resulted in normal virus replication and in a distinct profile of chemokines and cytokines that associated with acute lung neutropenia and enhanced accumulation of alternatively activated macrophages. Diminished expression of neutrophil-recruiting chemokines was also observed in cells infected with influenza virus, suggesting a key role of MDA5 in driving the early accumulation of neutrophils at the infection site. The biased acute inflammatory response of MDA5-deficient mice led to an enhanced chronic lung inflammation, epithelial cell hyperplasia, airway hyperreactivity, and diminished blood oxygen saturation.

Conclusions

MDA5 modulates the development of chronic lung inflammation by regulating the early inflammatory response in the lung.

Understanding how multiple stressors interact is needed to predict the dynamical outcomes of diverse biological systems, ranging from drug-resistant pathogens that are combated and treated with combination drug therapies to ecosystems impacted by environmental toxicants or disturbances. Nevertheless, extensive studies of higher-order (more than two component) interactions have been lacking. Here, we conduct experiments using 20 three-drug combinations and their effects on the bacterial growth of Escherichia coli We report our measurements of growth rates in single, pairwise and triple-drug combinations. To uncover emergent interactions, we derive a simple framework to calculate expectations for three-way interactions based on the measured impact of each individual stressor and of each pairwise interaction. Using our framework, we find that (i) emergent antagonisms are more common than emergent synergies and (ii) emergent antagonisms are more common and emergent synergies are more rare than would be inferred from measures of net effects that do not disentangle pairwise interactions from three-way interactions.

Understanding how multiple stressors interact is needed to predict the dynamical outcomes of diverse biological systems, ranging from drug-resistant pathogens that are combated and treated with combination drug therapies to ecosystems impacted by environmental toxicants or disturbances. Nevertheless, extensive studies of higher-order (more than two component) interactions have been lacking. Here, we conduct experiments using 20 three-drug combinations and their effects on the bacterial growth of Escherichia coli We report our measurements of growth rates in single, pairwise and triple-drug combinations. To uncover emergent interactions, we derive a simple framework to calculate expectations for three-way interactions based on the measured impact of each individual stressor and of each pairwise interaction. Using our framework, we find that (i) emergent antagonisms are more common than emergent synergies and (ii) emergent antagonisms are more common and emergent synergies are more rare than would be inferred from measures of net effects that do not disentangle pairwise interactions from three-way interactions.

In the version of this Article originally published, in Fig. 1a, all cells in the top schematic were missing, and in the bottom-left schematic showing multiple pattern shapes, two cells were missing in the bottom-right corner. This figure has now been updated in all versions of the Article.

As cells with aberrant force-generating phenotypes can directly lead to disease, cellular force-generation mechanisms are high-value targets for new therapies. Here, we show that single-cell force sensors embedded in elastomers enable single-cell force measurements with ~100-fold improvement in throughput than was previously possible. The microtechnology is scalable and seamlessly integrates with the multi-well plate format, enabling highly parallelized time-course studies. In this regard, we show that airway smooth muscle cells isolated from fatally asthmatic patients have innately greater and faster force-generation capacity in response to stimulation than healthy control cells. By simultaneously tracing agonist-induced calcium flux and contractility in the same cell, we show that the calcium level is ultimately a poor quantitative predictor of cellular force generation. Finally, by quantifying phagocytic forces in thousands of individual human macrophages, we show that force initiation is a digital response (rather than a proportional one) to the proper immunogen. By combining mechanobiology at the single-cell level with high-throughput capabilities, this microtechnology can support drug-discovery efforts for clinical conditions associated with aberrant cellular force generation.

N-methyl-D-aspartate (NMDA) receptors are widely expressed in the central nervous system. However, their presence and function at extraneuronal sites is less well characterized. In the present study, we examined the expression of NMDA receptor subunit mRNA and protein in human pulmonary artery (HPA) by quantitative polymerase chain reaction (PCR), immunohistochemistry and immunoblotting. We demonstrate that both GluN1 and GluN2 subunit mRNAs are expressed in HPA. In addition, GluN1 and GluN2 (A-D) subunit proteins are expressed by human pulmonary artery smooth muscle cells (HPASMCs) in vitro and in vivo. These subunits localize on the surface of HPASMCs and form functional ion channels as evidenced by whole-cell patch-clamp electrophysiology and reduced phenylephrine-induced contractile responsiveness of human pulmonary artery by the NMDA receptor antagonist MK801 under hypoxic condition. HPASMCs also express high levels of serine racemase and vesicular glutamate transporter 1, suggesting a potential source of endogenous agonists for NMDA receptor activation. Our findings show HPASMCs express functional NMDA receptors in line with their effect on pulmonary vasoconstriction, and thereby suggest a novel therapeutic target for pharmacological modulations in settings associated with pulmonary vascular dysfunction.