Opposing effects of target overexpression reveal drug mechanisms

Abstract

Overexpression of a drug's molecular target often increases drug resistance, offering a pathway for adaptive evolution and a tool for target identification. It is unclear though why this phenomenon applies to some drugs but not others. Here we gradually overexpressed antibiotic targets in Escherichia coli and found that drug resistance can increase, remain unchanged, decrease or even change non-monotonically. Even a single target can produce opposing responses to its different inhibitors. We explain these contradicting effects with quantitative models of enzyme inhibition that account for fitness costs and the biochemical activity or inactivity of drug-enzyme complexes. Thus, target overexpression confers resistance or sensitivity as a predictable property of drug mechanism, explaining its variable presence in nature as a resistance mechanism. Though overexpression screens may fail at identifying unknown targets, overexpressing known or putative targets provides a systematic approach to distinguish between simple inhibition and complex mechanisms of drug action.

Figure 1. Overexpression of a drug's target gene can increase, decrease, or have no effect on drug resistance

a, E. coli strains were constructed with IPTG adjustable overexpression of drug target genes. b, For each drug-gene pair, bacterial growth rates (heatmap) were measured over gradients of drug dose (vertical axis) and IPTG-induced gene dose (horizontal axis). Transcript overexpression (E_add) was quantified in Miller Units (MU) from kinetic beta-galactosidase assays of a strain whose plasmid encoded lacZ instead of a drug target, grown in the absence of drug (Supplementary Fig. 1). At each E_add level, the IC50 (drug concentration that inhibits growth to 50% of uninhibited wildtype growth) is shown (white dots and trend line) and compared to the IC50 of the wildtype (thin flat white line; WT denotes strain with empty plasmid). c, Drug target genes incur variable fitness costs when overexpressed, even in the absence of drugs.

Figure 2. Drug mechanism of action and fitness costs of target overexpression define the diverse changes in resistance when drug targets are overexpressed

a, Mass-action models of enzyme inhibition quantified the relation between drug concentration, growth inhibition, and enzyme expression (wildtype expression = E_wt, additional expression = E_add; arbitrary units). Model variants explored different mechanisms of drug action: inhibition of a beneficial reaction, wasting of an essential substrate, and enzyme-drug complexes that actively exert a toxic effect. Also modeled was the effect of some drug targets to incur fitness costs when overexpressed (Figure 1c, Supplementary Equations). b, The following models use a cost function (green line) based on the experimentally observed costs of PBP2 overexpression (circles; 1 model unit of E_add = 50 MU). c, Simple models that consider different mechanisms of drug action produce diverse changes in resistance as drug targets are overexpressed, thus rationalizing the conflicting experimental observations (data from Figure 1b), the variable presence in nature of drug resistance by target overexpression, and presenting a tool to investigate drug mechanism of action.

Figure 3. Two different modes of folate synthesis inhibition demonstrate that target overexpression confers resistance to a drug that inhibits metabolic flux but not to a drug that diverts metabolic flux

Drug action on folate synthesis is shown in blue. Both trimethoprim and sulfonamides are considered ‘inhibitors’ of enzymes in the folate synthesis pathway, but while trimethoprim inhibits catalysis by competing with dihydrofolic acid for binding to Dihydrofolate reductase (DHFR), sulfonamides compete with para-aminobenzoic acid for binding to Dihydropteroate synthase (DHPS) and, rather than preventing catalysis, are covalently linked by DHPS to the substrate pteridine diphosphate. Our data (Figure 1b) and mathematical analysis (Supplementary Equations) show that this mechanistic distinction between flux inhibition or flux diversion leads to resistance by target overexpression for trimethoprim but not for sulfonamides. HPPK: 6-hydroxymethyl-7,8-dihydropterin pyrophosphokinase; ADCL: Aminodeoxy-chorismate lyase; FPGS: Folylpolyglutamate synthetase.