Zinc blocks SOS-induced antibiotic resistance via inhibition of RecA in Escherichia coli

Abstract

Zinc inhibits the virulence of diarrheagenic E. coli by inducing the envelope stress response and inhibiting the SOS response. The SOS response is triggered by damage to bacterial DNA. In Shiga-toxigenic E. coli, the SOS response strongly induces the production of Shiga toxins (Stx) and of the bacteriophages that encode the Stx genes. In E. coli, induction of the SOS response is accompanied by a higher mutation rate, called the mutator response, caused by a shift to error-prone DNA polymerases when DNA damage is too severe to be repaired by canonical DNA polymerases. Since zinc inhibited the other aspects of the SOS response, we hypothesized that zinc would also inhibit the mutator response, also known as hypermutation. We explored various different experimental paradigms to induce hypermutation triggered by the SOS response, and found that hypermutation was induced not just by classical inducers such as mitomycin C and the quinolone antibiotics, but also by antiviral drugs such as zidovudine and anti-cancer drugs such as 5-fluorouracil, 6-mercaptopurine, and azacytidine. Zinc salts inhibited the SOS response and the hypermutator phenomenon in E. coli as well as in Klebsiella pneumoniae, and was more effective in inhibiting the SOS response than other metals. We then attempted to determine the mechanism by which zinc, applied externally in the medium, inhibits hypermutation. Our results show that zinc interferes with the actions of RecA, and protects LexA from RecA-mediated cleavage, an early step in initiation of the SOS response. The SOS response may play a role in the development of antibiotic resistance and the effect of zinc suggests ways to prevent it.

Fig 1. Activators and inhibitors of the SOS Response in E. coli.

Panels A and B, induction and inhibition of recA in STEC strains by qRT-PCR. Panel A, effect of mitomycin C on recA expression in STEC EDL933. Panel B, effects of ciprofloxacin and zinc acetate on recA expression in STEC TSA14. Panels C-H, effects of inducers and inhibitors of recA as measured using the Miller Assay and reporter strain JLM281. For all panels the activators and inhibitors were added 1 h after beginning the subculture in DMEM broth. *, significant by ANOVA compared to ciprofloxacin alone. In panels E-F the inhibitory effect of zinc on zidovudine-induced recA was significant for 0.2 mM and higher, despite the lack of asterisks. Panel H, the vertical dotted lines represent the IC₅₀ of zinc pyrithione (ZPT) and zinc acetate, and show that zinc pyrithione was 89 times more potent than zinc acetate in inhibition of zidovudine-induced recA.

Fig 2. Induction and inhibition of hypermutation by zinc in various strains of bacteria.

For each panel, the bacterial strain indicated was treated for 3 hours with and without the concentration of ciprofloxacin indicated, as guided by the ciprofloxacin MIC for each strain, and ± 0.2 mM zinc acetate. Then serial dilutions were performed, and plated on plain LB agar to determine the total number of bacteria, and on LB + rifampin, to determine the number of rifampin resistant colonies per mL. The rifampin resistance frequency was calculated for each condition. The rifampin concentrations, for each strain were, in μg/mL: Popeye-1, 12; B171-8, 8; CP9, 8; and Kpneu_707, 45. Paired ANOVA was by Kruskal- Wallis for non-parametric data due to skewing.

Fig 3. Assaying for hypermutation using ß-glucuronidase assay on MUG Selective agar.

MUG selective agar was formulated using methyl-umbelliferyl-glucuronide (MUG) as stated in Materials and Methods. STEC Popeye-1 was treated with zidovudine as indicated for 3 h, followed by serial dilutions and plating on MUG Selective agar as well as on plain LB plates to determine total counts. Panel A, plate of untreated Popeye-1 on MUG agar, showing that although faint colonies are visible they do not fluoresce. Panel B, a heavy inoculum of zidovudine-treated Popeye-1 formed a lawn of growth, within which several brightly fluorescent colonies are visible. Panel C, dose-response of zidovudine on the frequency of glucuronidase-positive colonies, in the absence and presence of zinc.

Fig 4. Attempts to develop more rapid screening methods for hypermutation in E. coli.

Panels A and B, using recA-lacZ reporter strain JLM281 on plates containing LB + 150 μg/mL X-gal. Panel A, compared to ciprofloxacin (positive control), the herbicide paraquat also induced recA expression; 10 μl of a 50 mg/mL paraquat solution was spotted onto a sterile blank test disk. Panels B- D, testing for hypermutation in EPEC strain B171-8 on LB + 5 μg/mL rifampin using antibiotic test disks. Plates were inoculated with a 1:5 dilution of an overnight culture of B171-8 using a sterile cotton swab and a criss-cross pattern over the entire plate. Panel B, 10 μl of 1 μg/mL ciprofloxacin was spotted on the disk. A ring of rifampin-resistant colonies grew up in the vicinity of the ciprofloxacin. Panel C, same as Panel B, except that 10 μL of 40 mM arsenic trioxide was spotted onto the blank disk. Panel D, same as Panels B and C, but using 10 mM 6-mercaptopurine. Panels E and F, attempt to create a semi-quantitative screening method for hypermutation. STEC Popeye-1 was grown in the absence or presence of 10 ng/mL ciprofloxacin ± 0.2 mM zinc acetate, then diluted into sterile saline to achieve an OD₆₀₀ of 0.2 for each culture. The diluted cultures were spread using a sterile cotton swab and a criss-cross pattern to cover the entire plate, then a trimethoprim E-test strip was applied to each plate. Each condition was plated in triplicate. Panel E, the ciprofloxacin-treated cultures (middle Petri dish) showed many more inlier colonies within trimethoprim’s zone of inhibition than in the control (left) or the ciprofloxacin + zinc condition (right). Panel F, the number of inlier colonies were counted in triplicate for each condition and are shown. *, significantly more than control; **, significantly fewer than ciprofloxacin alone, both by ANOVA.

Fig 5. Regulation LexA by SOS activators and by zinc in E. coli.

Panel A, immunoblot for LexA in whole-cell extracts of cultures of E. coli CP9 after a 3 h exposure to ciprofloxacin with and without zinc. Uncleaved LexA appeared to migrate in the form of a LexA dimer in these blots; while the cleaved LexA product ran at ~ 15 kDa. Panel B, densitometry scan of blot in Panel A, raw (left) and corrected for the effects of treatment on growth (right panel). Panel C, densitometry scan of a LexA blot (not shown) after a 1 h exposure to ciprofloxacin in Popeye-1. Panels D- H, RecA-mediated LexA cleavage assays in vitro, showing immunoblots against LexA. Purified LexA and RecA were incubated in vitro in the presence of absence of necessary cofactors, such as ssDNA and ATP or ATP- γ -S as described in the Methods section Panel D, RecA-mediated cleavage of LexA. An unlabeled lane to the left of lane 1 contained RecA alone, showing that the antibody does not cross-react between the two proteins. All the labeled lanes in Panel D received RecA, LexA, and a 38-mer oligonucleotide. Lane 1, no ATP; Lanes 2 and 3 also received 0.3 mM ATP; Lanes 4 and 5 also received 0.3 mM ATP-γ-S. Faint LexA cleavage products were visible in lanes 2–5 in the original blots, arrows; Lanes 6 and 7, plus ATP-γ-S and 1 μM zinc acetate; Lanes 8 and 9, plus ATP-γ-S and 1 μM MnCl₂; Lane 10 received 0.3 mM GTP, which does not support RecA activation, as an additional control. Panel E, densitometry scan of the chemiluminescence signal from the blot shown in Panel D. Panel F, dose-response relationship of ATP-γ-S concentration vs. LexA cleavage in the absence and presence of 1 μM zinc acetate, showing protection by zinc against LexA cleavage at 0.1 to 0.3 mM ATP-γ-S. Panel G, combined results of 4 separate experiments testing for the effect of zinc acetate, and four experiments with MnCl₂ on LexA cleavage, with results normalized to the no- ATP-γ-S control so that separate experiments could be compared. Panel H, lack of protection by zinc on LexA auto-cleavage induced by incubation at pH 9. Control lanes 1 and 2 show LexA kept at pH 7.8; Lanes 3–6 show LexA protein incubated for 15 min at pH 9, 37°. Lanes 7–10 show samples incubated at pH 9 for 30 min, 37°. Lanes 5–6 and 9–10 also received 1 μM zinc acetate.