Fig 1.
Operon eck1864-66Δ mutant serves as a synergy response marker.
(A) Outline of the overlap2 method (O2M), which we first presented in Brown et al. [18]. O2M requires a chemical-genetic dataset. To generate these datasets, a collection of mutants is grown in the presence of a number of different small molecules. Using colony size as a stand-in for growth, we calculated a quantitative growth score for each combination of mutant + small molecule. From these data, we generate a chemical-genetic signature for each small molecule. This signature includes the score for each mutant in the collection when grown on a particular small molecule. In the heat maps (middle) of “starting drug” versus “genetic mutants”, each vertical line represents a different mutant. A blue line represents small colony size compared to wild-type cells, or a sensitive mutant; a yellow line represents larger colony size, or a resistant mutant. We compare the genetic signatures for starting drugs (e.g., trimethoprim) and known synergistic molecules (e.g., sulfamethizole) computationally. From this analysis, we identify genes whose knockout mutants show the significant growth scores to the starting drug and all its known synergistic partners (outlined by red boxes). These represent the putative synergy prediction mutants. Since our starting drug and its known synergizers induce significant phenotypes from these mutants, we hypothesize that other small molecules that induce significant phenotypes will also synergize with the starting drug. We reanalyze the chemical-genetic dataset to identify these small molecules, then test them in checkerboard analyses. (B) The folate biosynthesis pathway, with trimethoprim and sulfamethizole targets marked. (C) Checkerboard results from trimethoprim + predicted synergistic small molecules (green labels), known synergizer (purple label), and negative control small molecules (blue labels) that are not predicted to synergize with trimethoprim. The fractional inhibitory concentration index (FICI) cutoff for synergy is ≤0.5 (red line), and synergistic FICI values are marked with yellow bars on the graph. Nonsynergistic values are colored blue. Average FICI scores are shown. Individual FICI scores are shown in S2 Table. (D) Predicted synergizers with sulfamethizole. The color scheme is the same as in part C. P values were calculated using a Fisher’s exact test. Individual FICI scores are shown in S3 Table.
Table 1.
Minimum inhibitory concentration (90% inhibition) of small molecules used in Fig 1.
Fig 2.
High-throughput screen with synergy response markers.
(A) Screen format to identify molecules that synergize with trimethoprim (TMP). (B) Fractional inhibitory concentration index (FICI) of screen hits. Small molecules predicted to synergize with trimethoprim are labeled green. Negative control small molecules, which were part of the Microsource Spectrum collection but not predicted to synergize with trimethoprim, are labeled with blue text. Synergistic FICI values (≤0.5) are marked with yellow bars, and nonsynergistic FICI values are marked with blue bars. Data from this graph are shown in S4 Table. (C) Bliss independence scores for predicted trimethoprim synergizers that do not inhibit E. coli growth and thus cannot be tested in checkerboard assays. Small molecules were tested at either 10 μM (grey labels) or 100 μM (green labels) in combination with trimethoprim. Small molecules are considered synergistic if they exhibit a negative score at both concentrations (yellow bars). Bars representing data for nonsynergistic small molecules are colored with blue. Data from this graph are shown in S5 Table.
Table 2.
Predicted trimethoprim synergizers from Microsource Spectrum collection screen.
Table 3.
Minimum inhibitory concentration (90% inhibition) of trimethoprim, sulfamethizole, AZT, and floxuridine against clinical isolates.
Fig 3.
Trimethoprim and azidothymidine (AZT) act synergistically in clinical strains that do not respond to trimethoprim and sulfamethizole.
(A) Trimethoprim + sulfamethizole. (B) Trimethoprim + AZT. Synergistic fractional inhibitory concentration index (FICI) values (≤0.5) are marked with yellow bars, and nonsynergistic (FICI > 0.5) FICI values are marked with blue bars. Trimethoprim/sulfamethizole-resistant isolates are labeled with red text and sensitive isolates are labeled with black text. Average FICI scores are shown in the graph. Individual FICI scores are shown in S6 Table.
Fig 4.
The SOS response is induced by trimethoprim and azidothymidine (AZT).
(A) SOS response measured by a green fluorescent protein (GFP) reporter gene under control of the sulA reporter. Small molecules were added at 50% minimum inhibitory concentration (MIC). Expected sulA induction (green) is either the induction by the second molecule or, if the sulA reporter is repressed by the second molecule, no induction or repression. Trimethoprim does not induce the sulA reporter, so it is considered to not have any contribution to the expected value, and we do not simply sum the induction of trimethoprim + molecule #2. The observed sulA induction (purple) is significantly higher than expected in the trimethoprim and AZT combination but not nonsynergistic combinations, such as trimethoprim + hydroxyurea or trimethoprim + rifampicin. Since the trimethoprim + sulfamethizole combination does not induce sulA, the molecular mechanisms underlying trimethoprim + sulfamethizole synergy likely differ from trimethoprim + AZT synergy. Significance was calculated using a Mann-Whitney test. Error bars represent the standard deviation. In all cases when we observed a significant difference between expected and observed, we also found a significant difference between induction by molecule #2 and induction in the combination. The data for these graphs are in S7 Table. (B) Fluctuation assay measures the mutation rate following small-molecule treatment. Small molecules alone (solid colors) do not significantly increase mutation rate. Trimethoprim combined (striped bars) with synergistic partners AZT or mitomycin C increases mutation rate. P values were calculated using Fisher’s exact test. Error bars represent the 95% confidence interval. The data for these graphs are in S8 Table.
Fig 5.
The combination of disrupted nucleotide balance and premature DNA chain termination act synergistically to inhibit E. coli growth.
Bars for nonsynergistic combinations (fractional inhibitory concentration index [FICI] > 0.5) are colored blue, and bars for synergistic combinations (FICI ≤ 0.5) are colored yellow. Checkerboard assays for (A) nucleotide homeostasis disruptors + azidothymidine (AZT) or trimethoprim and (B) Nucleoside analogs + AZT or trimethoprim show synergy between nucleotide homeostasis inhibitors and AZT or nucleoside analogs and trimethoprim but not the reverse. (C) Growth of wild-type and mutant cells on AZT, trimethoprim (TMP) or rifampicin (RIF). These data are the average of 3 replicates. The data for parts A and B are in S9 Table. The data for part C are in S10 Table.
Fig 6.
Trimethoprim replacement molecules act synergistically with azidothymidine (AZT) against clinical isolates.
Synergistic fractional inhibitory concentration index (FICI) values (≤0.5) are marked with yellow bars, and nonsynergistic (FICI > 0.5) FICI values are marked with blue bars. Trimethoprim/sulfamethizole-resistant isolates are labeled with red text, and sensitive isolates are labeled with black text. (A) Hydroxyurea (HU) + AZT or (B) floxuridine + AZT. Individual FICI scores for this graph are listed in S11 Table. Example checkerboard data are shown in S3 Fig.
Table 4.
Drug doses in zebrafish infection experiment.
Fig 7.
The rationally designed synergistic combination of floxuridine + azidothymidine (AZT) improves treatment of infected zebrafish embryos.
(A) We injected a trimethoprim/sulfamethizole-resistant E. coli strain (blood isolate #8, or BEC8) into zebrafish embryos, then treated them with drugs starting at 3 hours postinoculation (hpi). At 24 hpi, we euthanized embryos and determined bacterial burden (colony-forming units [CFU]) in whole fish. Each symbol (blue circles for floxuridine + AZT, orange squares for trimethoprim [TMP] + sulfamethizole [SFZ], and grey triangles for vehicle control) represents a single fish. N ≥ 30 each condition. Data from 3 separate experiments are shown. Inoculum levels are shown in S4 Fig. Black lines represent median bacterial burden for each condition. P value was calculated using a Mann-Whitney test. Grey arcs show the percent of individual embryos within each population group. For example, 57% of the floxuridine + AZT-treated group has a bacterial burden between 1 CFU and 103 CFU. (B) Zebrafish embryos infected with E. coli strain F11, which is sensitive to trimethoprim/sulfamethizole. The color scheme and symbols are the same as in part A. Bacterial burden for each individual embryo are listed in S12 Table. Fractional inhibitory concentration index (FICI) information for strain F11 is in S13 Table.
Fig 8.
Inhibition of nucleotide homeostasis amplifies the consequences of DNA damage, increasing the toxicity of DNA-damaging agents.
Trimethoprim treatment (orange graphics) blocks DNA repair, which is induced by azidothymidine (AZT) treatment (blue graphics). The combined effects result in increased growth inhibition relative to single-agent treatment.