Tripartite Loops Reverse Antibiotic Resistance

Abstract

Antibiotic resistance threatens to undo many of the advancements of modern medicine. A slow antibiotic development pipeline makes it impossible to outpace bacterial evolution, making alternative strategies essential to combat resistance. In this study, we introduce cyclic antibiotic regimens composed of 3 drugs or "tripartite loops" to contain resistance within a closed drug cycle. Through 424 discrete adaptive laboratory evolution experiments we show that as bacteria sequentially evolve resistance to the drugs in a loop, they continually trade their past resistance for fitness gains, reverting back to sensitivity. Through fitness and genomic analyses, we find that tripartite loops guide bacterial strains toward evolutionary paths that mitigate fitness costs and reverse resistance to component drugs in the loops and drive levels of resensitization not achievable through previously suggested pairwise regimens. We then apply this strategy to reproducibly resensitize or eradicate 4 drug-resistant clinical isolates over the course of 216 evolutionary experiments. Resensitization occurrs even when bacteria adapted through plasmid-bound mutations instead of chromosomal changes. Combined, these findings outline a sequential antibiotic regimen with high resensitization frequencies, which may improve the clinical longevity of existing antibiotics even in the face of antibiotic resistance.

Keywords: collateral sensitivity; cyclic therapy; epistasis; sequential antibiotic therapy; soft agar gradient evolution; tripartite loops.

Fig. 1.

Tripartite loops improve antibiotic resensitization. a) SAGE is used to study 3-drug cyclic regimens or tripartite loops. Bacteria were inoculated into soft agar containing antibiotic gradients to generate resistant mutants (n = 16). SAGE plates were incubated for a fixed duration of 7 d, after which mutants were harvested and passed through 3 “flat plates” containing the same antibiotic from the prior SAGE plate at a concentration = ½ the evolved MIC of the mutants. The incubation period for each flat passage is noted in the figure. MIC and CS profiles of mutants were determined after the end of the flat plates. b) GEN MICs of strains that passed through the GEN-PIP-NIT tripartite loop. The y axis denotes the GEN MICs and the x axis denotes the sequence of antibiotics against which the strains were evolved. For example, the GEN-PIP bar shows the GEN MICs of strains that were sequentially evolved to GEN and PIP (as shown in a). c) NIT MICs of strains that passed through the NIT-PIP-GEN loop. Resen, resensitization counts. Dotted red lines indicate the clinical breakpoint (EUCAST). Bars represent the median MICs. **P < 0.01, ****P < 0.0001, Kruskal–Wallis with uncorrected Dunn's test.

Fig. 2.

PIP aids resensitization in tripartite loops. a) GEN MICs of GEN-resensitized and b) GEN-resistant strains that passed through the GEN-PIP-NIT loop. c) NIT MICs of NIT resensitized and d) resistant strains that passed through the NIT-PIP-GEN loop. *P < 0.05, **P < 0.01, ***<P < 0.001, ****P < 0.0001, Kruskal–Wallis with uncorrected Dunn's test. e) GEN MIC of strains that passed through a PIP-GEN-NIT tripartite loop. MICs after the PIP step are not shown. For all graphs, the y axis denotes the MICs and the x axis denotes the sequence of antibiotics against which the strains were evolved before measuring the MICs. For example, the PIP-GEN-NIT bar shows the GEN MICs of strains that were sequentially evolved to PIP, GEN and NIT.**P < 0.01, Mann–Whitney test. Bars represent the median MICs.

Fig. 3.

Resensitization does not correlate with CS but mitigates fitness loss. a) Contingency table for the 11 strains, which evolved NIT resistance through the GEN-PIP-NIT loop, showing no associations between CS and GEN resensitizations. Fisher's exact test. b) First column: NIT CS of the GEN and PIP evolved mutants from the GEN-PIP-NIT loop. Second column: GEN and PIP CS of WT bacteria evolved to NIT. CS interactions are reported on a log2 scale. c) DOX MICs of an 8 strain subset of the GEN and PIP evolved mutants from the GEN-PIP-NIT loop. CS interactions are reported on a log2 scale. The y axis denotes the ID of the strains that were picked for DOX MIC testing. d) GEN MICs of the subset that passed through the GEN-PIP-DOX loop. The x axis denotes the sequence of antibiotics against which the strains were evolved before measuring GEN MICs. For example, the GEN-PIP-DOX bar shows the GEN MICs of strains that were sequentially evolved to GEN, PIP, and DOX. Dotted line indicates the clinical breakpoint. Bars represent the median MICs. *P < 0.05, Mann–Whitney test. e and f) AUCs of strains before and after NIT evolution for GEN-resensitized and GEN-resistant strains, respectively. The x axis denotes the sequence of antibiotics against which the strains were evolved before measuring AUCs. GEN-PIP = before NIT evolution, GEN-PIP-NIT = after NIT evolution. ΔAUC is the average of the difference between post- and pre-NIT AUCs. For the GEN resistant group, we considered every strain that did not meet our resensitization criteria as resistant. This resulted in the inclusion of 1 strain that was below the GEN resistant breakpoint but did not reach our resensitization standard. Arrows indicate the strains that were sequenced. g) ΔAUC of individual strains plotted, grouped by resensitized and resistant. Horizontal lines represent the mean. **P < 0.01, unpaired t-test.

Fig. 4.

Tracking genomic changes through the GEN-PIP-NIT loop. a-d) Venn diagrams show overlapping and unique mutations in the GEN-resensitized and GEN-resistant strains from the 3 strains sequenced. The label on the left denotes when the strains were sequenced, with the most recent evolution step highlighted. Mutations that appeared in one step were carried forward to the next step, but are only displayed the first time they appeared in this figure. Strikethroughs denote mutations that appeared in a prior step but were not present in the current step. Underlined mutation in D, strain 3 represents a newly acquired mutation absent from strain 3 in (c). e) Venn diagram showing all overlapping and unique mutations between the GEN resensitized and GEN resistant group, pooled from every step (GEN-PIP-NIT only). f and g) GO term enrichment analysis of unique mutations in the GEN resensitized and GEN resistant groups.

Fig. 4.

Tracking genomic changes through the GEN-PIP-NIT loop. a-d) Venn diagrams show overlapping and unique mutations in the GEN-resensitized and GEN-resistant strains from the 3 strains sequenced. The label on the left denotes when the strains were sequenced, with the most recent evolution step highlighted. Mutations that appeared in one step were carried forward to the next step, but are only displayed the first time they appeared in this figure. Strikethroughs denote mutations that appeared in a prior step but were not present in the current step. Underlined mutation in D, strain 3 represents a newly acquired mutation absent from strain 3 in (c). e) Venn diagram showing all overlapping and unique mutations between the GEN resensitized and GEN resistant group, pooled from every step (GEN-PIP-NIT only). f and g) GO term enrichment analysis of unique mutations in the GEN resensitized and GEN resistant groups.

Fig. 4.

Tracking genomic changes through the GEN-PIP-NIT loop. a-d) Venn diagrams show overlapping and unique mutations in the GEN-resensitized and GEN-resistant strains from the 3 strains sequenced. The label on the left denotes when the strains were sequenced, with the most recent evolution step highlighted. Mutations that appeared in one step were carried forward to the next step, but are only displayed the first time they appeared in this figure. Strikethroughs denote mutations that appeared in a prior step but were not present in the current step. Underlined mutation in D, strain 3 represents a newly acquired mutation absent from strain 3 in (c). e) Venn diagram showing all overlapping and unique mutations between the GEN resensitized and GEN resistant group, pooled from every step (GEN-PIP-NIT only). f and g) GO term enrichment analysis of unique mutations in the GEN resensitized and GEN resistant groups.

Fig. 5.

The NIT-PIP-GEN loop reduces clinically acquired NIT resistance. a) Four uropathogenic clinical E. coli strains resistant to NIT were used to start sequential PIT and GEN evolution. Each strain started 8 replicates in SAGE. The rest of the experimental evolution design remained identical to the one used for the laboratory strain. b–e) NIT MICs of the clinical replicates post-GEN adaptation. The dotted line represents NIT MICs of the parental strain pre-SAGE adaptation. Labels on the x axis denote the parental strain of the replicates for which the MICs are displayed. **P < 0.01, ****P < 0.0001, 1 sample t-test. f) PIP, PIP and PAV MICs of 5 clinical replicates after PIT exposure. The y axis denotes the strain ID where A5 = the 5th replicate from strain A.