Enhanced emergence of antibiotic-resistant pathogenic bacteria after in vitro induction with cancer chemotherapy drugs

Abstract

Background

Infections with antibiotic-resistant pathogens in cancer patients are a leading cause of mortality. Cancer patients are treated with compounds that can damage bacterial DNA, potentially triggering the SOS response, which in turn enhances the bacterial mutation rate. Antibiotic resistance readily occurs after mutation of bacterial core genes. Thus, we tested whether cancer chemotherapy drugs enhance the emergence of resistant mutants in commensal bacteria.

Methods

Induction of the SOS response was tested after the incubation of Escherichia coli biosensors with 39 chemotherapeutic drugs at therapeutic concentrations. The mutation frequency was assessed after induction with the SOS-inducing chemotherapeutic drugs. We then tested the ability of the three most highly inducing drugs to drive the emergence of resistant mutants of major bacterial pathogens to first-line antibiotics.

Results

Ten chemotherapeutic drugs activated the SOS response. Among them, eight accelerated the evolution of the major commensal E. coli, mostly through activation of the SOS response, with dacarbazine, azacitidine and streptozotocin enhancing the mutation rate 21.3-fold (P < 0.001), 101.7-fold (P = 0.01) and 1158.7-fold (P = 0.02), respectively. These three compounds also spurred the emergence of imipenem-resistant Pseudomonas aeruginosa (up to 6.21-fold; P = 0.05), ciprofloxacin-resistant Staphylococcus aureus (up to 57.72-fold; P = 0.016) and cefotaxime-resistant Enterobacteria cloacae (up to 4.57-fold; P = 0.018).

Conclusions

Our results suggest that chemotherapy could accelerate evolution of the microbiota and drive the emergence of antibiotic-resistant mutants from bacterial commensals in patients. This reveals an additional level of complexity of the interactions between cancer, chemotherapy and the gut microbiota.

Introduction

Bloodstream infections are a major threat to the lives of cancer patients, causing death in ∼10%.¹ Nosocomial outbreaks of antibiotic-resistant pathogens in cancer patients can increase the mortality to much higher rates.¹ Cancer patients are treated with compounds that directly or indirectly damage the DNA of fast-growing cancer cells, but they also unintentionally damage the DNA of bacteria. Hence, low concentrations of chemotherapeutic drugs, such as streptozotocin, bleomycin, mitomycin C and daunorubicin, damage bacterial DNA, subsequently triggering the SOS response, a conserved regulatory network.^2–4 During the SOS response, single-stranded DNA-bound RecA stimulates the cleavage of the LexA repressor. This results in the derepression of LexA-controlled genes, including the error-prone DNA polymerases PolII, PolIV and PolV, increasing the bacterial mutation rate.⁵ Bacterial resistance to antibiotics occurs not only after acquisition of foreign genes encoding resistance mechanisms but also after the mutation of core genes. Thus, we tested whether chemotherapeutic drugs can enhance the emergence of antibiotic-resistant mutants from bacterial commensals. We assessed the ability of 39 chemotherapeutic drugs to enhance bacterial mutagenesis through activation of the bacterial SOS response, potentially driving the emergence of mutants resistant to major antibiotics and complicating the treatment of infections in patients treated for cancer.

Methods

Bacterial strains

We used the strain Escherichia coli MG1655 PrecN∼gfp (hereafter called E. coli WT). RecN belongs to the RecA regulon. Thus, the fluorescence of the strain at 508 nm is proportional to the activation of the SOS response. E. coli MG1655 lexAind PrecN∼gfp (hereafter called E. coli ΔSOS) is derived from E. coli WT, but has an uncleavable LexA protein that permanently represses the SOS genes and blocks the SOS response.⁴ Both E. coli strains were resistant to chloramphenicol. The emergence of resistant mutants was also assessed from the WT reference strains Pseudomonas aeruginosa ATCC 27853, Staphylococcus aureus ATCC 25923 and Enterobacter cloacae ATCC 13047.

Chemotherapeutic drugs and antibiotics

Figure 1(a) lists the 39 chemotherapeutic drugs tested and their classification. We also tested the monoclonal antibody trastuzumab as a control. All compounds were provided by the pharmacy of the University Hospital of Besançon and used extemporaneously for microbiological experiments. MICs of all compounds were determined using the CLSI agar dilution method.⁶ All antibiotics tested were provided by Sigma–Aldrich.

Figure 1.

Ten chemotherapeutic drugs induce the bacterial SOS response. (a) Classification of the 39 compounds tested. (b) The first column shows the result of the screening of SOS-response induction in E. coli WT (black box, positive; light grey box, negative). The second column shows the intensity of the induction, which is the ratio of the mean fluorescence of E. coli WT induced (at the concentration shown in the third column) by a chemotherapeutic drug versus those that were not. n ≥ 3; two-sided Student’s t-tests using 1.0 as comparator. The fourth column shows the drug concentrations used for experiments with pathogenic species. The fifth column shows the range of therapeutic concentrations, retrieved from the literature (Table 1). The monoclonal antibody trastuzumab was used as a negative control (data not shown).

Assessment of SOS-response induction

We first screened the ability of each chemotherapeutic compound to induce the bacterial SOS response using 6 mm paper discs loaded with appropriate dilutions of each drug on E. coli WT cultures on Mueller–Hinton agar (MHA), as previously described.⁷ SOS induction was indicated by fluorescent halos in the areas of bacterial growth closest to the discs. The intensity of SOS induction was further measured for each compound that tested positive. Briefly, 200 μL of Mueller–Hinton broth (MHB), containing each compound at therapeutic concentrations (i.e. the concentration found in the blood or other body fluid, Figure 1b and Table 1), was inoculated with 10⁷ bacteria and incubated in 96-well plates at 37°C under agitation (360 rpm) for 4 h, during which the fluorescence at 508 nm was measured every 15 min (Synergy H1, BioTek Instruments, Inc.). We calculated the ratio of the fluorescence at 508 nm between the E. coli WT strain exposed to each chemotherapeutic drug and the unexposed E. coli WT strain. All assays were independently performed at least three times. Compounds that induced the fluorescence by at least 2.0-fold were retained for further testing.

Table 1.

Antibacterial activity, concentrations tested and therapeutic concentrations of the 13 chemotherapeutic drugs that trigger the bacterial SOS response in E. coli

Chemotherapeutic drugs	MIC (mg/L)		Concentration (mg/L) tested in induction experiments			Therapeutic concentrations in vivo
	E. coli WT	E. coli ΔSOS	E. coli WT (ratio with MIC)	E. coli ΔSOS (ratio with MIC)	P. aeruginosa, E. cloacae, S. aureus	range (mg/L; body fluid)	ref
Azacitidine	8	2	2 (0.25)	0.5 (0.25)	2	0.1–1.0 (plasma, patients)	22
Bleomycin	2	1	1 (0.5)	0.5 (0.5)	–	4–10 (plasma, mice)	23
Cisplatin	4096	512	4 (0.5)	0.5 (0.001)	–	0.5–19.9 (peritoneal fluid, mice)	24
Dacarbazine	64	16	10 (0.16)	2.5 (0.16)	40	14–45 (plasma, patients)	25
Daunorubicin	1024	256	50 (0.04)	12.5 (0.04)	–	25 (plasma, patients)	26
Epirubicin	1024	256	50 (0.05)	12.5 (0.05)	–	57–148 (plasma, patients)	27
Etoposide	–	–	10	–	–	0.4–22 (plasma, patients)	28
Fotemustine	–	–	4	–	–	0.7–2.4 (plasma, patients)	29
Gemcitabine	512	128	30 (0.06)	7.5 (0.06)	–	7–10 (plasma, patients)	30
Mitomycin C	4	0.5	0.1 (0.025)	0.013 (0.025)	–	0.01–1 (plasma and bile, patients)	31
Oxaliplatin	4096	512	4 (0.001)	0.5 (0.001)	–	1.1–1.7 (blood, patients)	32
Streptozotocin	256	2	25 (0.1)	0.195 (0.1)	25	400–610 (peritoneal cavity, patients) 5.3–10.6 (plasma, patients)	33
Temozolomide	1024	256	30 (0.03)	7.5 (0.03)	–	0.1–10 (serum, patients)	34

Emergence of antibiotic-resistant mutants after incubation with SOS-inducing chemotherapeutic drugs

Bacteria were grown in MHB for 8 h at 37°C under agitation and then diluted 1:100. Each half of the cultures was then exposed to the drugs for 18 h. Bacteria were incubated at concentrations resulting in the highest level of induction compatible with those found in vivo (Figure 1b and Table 1). E. coli ΔSOS was incubated with each chemotherapeutic drug at the equivalent inhibitory concentration (Table 1). Cultures were collected, washed twice, appropriately diluted in saline solution and plated on MHA plates ± rifampicin at 100 mg/L, ciprofloxacin at 2 mg/L, cefotaxime at 32 mg/L or imipenem at 2 mg/L.^8,9 The frequency of resistant mutants under stress was then calculated as the number of resistant colonies divided by the number of plated cells. The results are expressed as the ratios of the mutant frequency with and without chemotherapeutic drug. All assays were independently performed at least three times.

Statistical analysis

Two-sided Student’s t-tests were used to determine statistical significance for comparisons of fluorescence (between drug-induced and non-induced E. coli WT; Figure 1b), frequencies of the emergence of rifampicin-resistant mutants, with or without induction by chemotherapeutic drug in E. coli WT and E. coli ΔSOS (Figure 2), and frequencies of the emergence of antibiotic-resistant P.aeruginosa, S.aureus and E.cloacae mutants, with and without induction by chemotherapeutic drugs (Figure 3). Data were log transformed. The emergence of antibiotic-resistant mutants (Figures 2 and 3) is expressed on a log scale. The chosen significance cut-off was 0.05 for all tests.

Figure 2.

Eight chemotherapeutic drugs enhance the frequency of mutation in E. coli, mostly through the activation of the SOS response. The frequency of emergence of rifampicin-resistant mutants is shown from E. coli WT and E. coli ΔSOS after induction with chemotherapeutic drugs at therapeutic concentrations (from Figure 1b). Ratios of the geometric means of the frequency of resistant mutants with error bars representing the 95% CI are shown. n ≥ 3; two-sided Student’s t-tests using 1.0 (broken line) as comparator. The monoclonal antibody trastuzumab was used as a negative control (data not shown).

Figure 3.

Therapeutic concentrations of dacarbazine, azacitidine and streptozotocin enhance the emergence of antibiotic-resistant mutants of bacterial pathogens. The value shown on the y-axis is the ratio of the frequencies of emergence of imipenem-resistant P. aeruginosa, ciprofloxacin-resistant S. aureus and cefotaxime-resistant E. cloacae, with and without induction with chemotherapeutic drugs at the concentrations given in Figure 1(b). Ratios of the geometric means of the frequency of resistant mutants with error bars representing the 95% CI are shown. n ≥ 3; two-sided Student’s t-tests using 1.0 (broken line) as comparator.

Results

We tested the ability of 39 chemotherapeutic drugs (Figure 1a) to activate the SOS response of E.coli and found that 10 compounds activated the bacterial stress response at therapeutic concentrations (Figure 1b), thus potentially increasing the rate of mutagenesis. We next measured the emergence of rifampicin-resistant mutants of E. coli after induction with the SOS-inducing chemotherapeutic drugs at concentrations found in patients.⁸ The emergence of rifampicin-resistant mutants was: (i) weakly enhanced by daunorubicin [ratio mean (R) = 1.69; 95% CI = 1.50–1.89; P = 0.01], epirubicin (R = 2.23; 95% CI = 1.91–2.56; P = 0.008), mitomycin C (R = 2.30; 95% CI = 1.81–2.80; P = 0.02) and gemcitabine (R = 6.22; 95% CI = 4.82–7.62; P = 0.004); (ii) moderately enhanced by bleomycin (R = 16.75; 95% CI = 13.60–19.90; P = 0.02) and dacarbazine (R = 21.31; 95% CI = 17.61–25.02; P < 0.001); and (iii) strongly enhanced by azacitidine (R = 101.70; 95% CI = 79.41–123.99; P = 0.01) and streptozotocin (R = 1558.73; 95% CI = 873.15–2244.32; P = 0.02) (Figure 2). We failed to show that weak SOS inducers (i.e. cisplatin and oxaliplatin) enhanced the frequency of mutation in our model (Figure 2). The SOS response was generally involved in the increased rate of mutagenesis since its inactivation in the strain E. coli ΔSOS markedly reduced the emergence of mutants, except for azacitidine (Figure 2). High levels of resistance to major antibiotics, such as carbapenems, fluoroquinolones and third-generation cephalosporins, readily occur after mutations of the bacterial core genes encoding porins, DNA gyrases or regulators of the AmpC cephalosporinase, respectively.¹⁰ We tested whether therapeutic concentrations of chemotherapeutic drugs drive the emergence of such resistant mutants and subsequently tested the top three inducers (dacarbazine, azacitidine and streptozotocin). The frequency of emergence of imipenem-resistant P. aeruginosa was enhanced after incubation with dacarbazine (R = 6.21; 95% CI = 1.99–10.42; P = 0.05) and azacitidine (R = 3.47; 95% CI = 2.84–4.11; P = 0.005) (Figure 3). The frequency of emergence of ciprofloxacin-resistant S. aureus was enhanced by dacarbazine (R = 57.72; 95% CI = 2.69–112.76; P = 0.016), azacitidine (R = 6.83; 95% CI = 5.11–8.54; P = 0.05) and streptozotocin (R = 7.42; 95% CI = 4.04–10.80; P = 0.01) (Figure 3). The frequency of emergence of cefotaxime-resistant E. cloacae was enhanced by dacarbazine (R = 4.57; 95% CI = 2.96–6.18; P = 0.018) and streptozotocin (R = 4.46; 95% CI = 2.35–6.57; P = 0.03) (Figure 3).

Discussion

We found that 10 chemotherapeutic drugs spurred bacterial evolution at therapeutic concentrations, mostly through the SOS response. Furthermore, the top three inducing drugs (dacarbazine, azacitidine and streptozotocin) drove the emergence of mutants of the commensals S. aureus, P. aeruginosa and E. cloacae resistant to first-line antibiotics. This suggests that resistant mutants readily emerge in the intestinal microbiota of patients treated with chemotherapeutic drugs.

One limitation of the study is the reliance of the results on the drug concentrations tested for induction. The concentrations of the chemotherapeutic drugs in the faeces are unknown. However, we assessed induction of the SOS response at concentrations in the range of those found in the blood (Table 1) and that are likely to occur in the intestinal epithelium, in contact with the gut microbiota.

Many antibiotics have been shown to activate the bacterial SOS response and the subsequent emergence of mutations, leading to drug resistance. Although the intensity of the induction depends on the bacterial species and the nature and concentration of the antibiotic, the highest mutation rate has been observed with subinhibitory concentrations of ciprofloxacin, which enhanced the emergence of rifampicin-resistant mutants of Acinetobacter baumannii, S. aureus and Vibrio cholerae by 27- to 100-fold.^4,11 These rates are far below those obtained with subinhibitory concentrations of the alkylating agent streptozotocin (1558.7-fold increase; Figure 2). Most (8 of 10) of the drugs that drive the emergence of resistant mutants directly act on DNA (Figure 1a and b). Only two compounds that indirectly damage the DNA, the cytidine analogues azacitidine and gemcitabine, were able to enhance the mutagenesis of E. coli. Inactivation of the SOS response did not reduce the frequency of mutants after induction with azacitidine (P = 0.4; Figure 2). This confirms that base analogues (e.g. azacitidine and gemcitabine) can also accelerate the mutation rate independently of the error-prone DNA polymerase PolIV, although they also induce the SOS response (Figure 1b).¹² Similarly, it has been shown that DNA alkylation increases the mutation frequency of the naturally occurring Ada-deficient subpopulation of bacterial cells, independent of the SOS response.¹³ This is consistent with the observed enhanced frequency of mutants of E. coli ΔSOS incubated with the alkylating agent streptozotocin (R = 109.82; 95% CI = 66.91–152.72; P = 0.05) (Figure 2).

These results suggest that the combination of cancer chemotherapy and antibiotic use may increase the risk of bloodstream infection with antibiotic-resistant pathogens in cancer patients. Hence, antibiotics frequently used for prophylaxis (e.g. sulfamethoxazole/trimethoprim and ciprofloxacin) and chemotherapy promote antibiotic resistance mutations de novo via the SOS response (Figure 2).^4,11 Resistant pathogens are then further selected by the antibiotics given for treatment. Their translocation from the gut lumen into the bloodstream may be increased by chemotherapy, which damages the intestinal barrier and reduces host defences.¹⁴ The unravelling of the intricate relationship between cancer, the microbiota and cancer therapeutics is an area of intense interest. The microbiota has been shown to amplify or mitigate carcinogenesis, and an optimal response to cancer therapy requires an intact microbiota.^10,15,16 Chemotherapeutic drugs are directly toxic for commensal bacteria, with relatively low MICs (Table 1). Hence, chemotherapy can modify the balance of the gut microbiota and has been associated with a decrease in commensal anaerobes and the expansion of potentially pathogenic bacteria.^17–21 In addition to the dysbiosis reported by others, we have shown that antimitotic therapies can accelerate bacterial evolution in treated patients, revealing an additional level of complexity of the interactions between cancer, chemotherapy and the gut microbiota.

Funding

This work was supported by the University Hospital of Besançon, France (Grant RAPACE - APICHU 2016). The study sponsor had no role in the design of the study, the collection, analysis and interpretation of the data, the writing of the manuscript or the decision to submit the manuscript for publication.

Transparency declarations

None to declare.

Author contributions

Conceived and designed the experiments: D. H. and D. M. Performed the experiments: D. H., A. M. and M. R. Analysed the data: D. H., A. M., O. A., V. N., C. F-L., D. M. and X. B. Wrote the paper: D. H., O. A., V. N. and X. B.