Uropathogenic Escherichia coli employs both evasion and resistance to subvert innate immune-mediated zinc toxicity for dissemination

Significance Uropathogenic Escherichia coli (UPEC) is responsible for most urinary tract infections and is also a frequent cause of sepsis, thus necessitating an understanding of UPEC-mediated subversion of innate immunity. The role of zinc in the innate immune response against UPEC infection, and whether this pathogen counters this response, has not been examined. Here we demonstrate, both in vitro and in vivo, that UPEC both evades and resists innate immune-mediated zinc toxicity to persist and disseminate within the host. Moreover, we have defined the set of UPEC genes conferring zinc resistance and have developed highly selective E. coli reporter systems to track zinc toxicity. These innovative approaches substantially enhance our understanding of immune-mediated metal ion toxicity and bacterial pathogenesis.

GFP under the control of the S. Typhimurium rpsM promoter, as well as mCherry under the control of the E. coli zntA promoter, was generated based on the vector pFCcGi (2). DNA manipulation and cloning were performed by Epoch Life Science (USA). Plasmids were transformed into E. coli MG1655 or EC958. An overview of plasmids pGcCe and pGcCzntAp is presented in SI Appendix, Fig. S9.
TraDIS. Approximately 2x10 8 cells from a previously constructed miniTn5 mutant library of EC958 (3) were inoculated into 100 mL of either LB (control) or LB supplemented with 1.0 mM ZnSO4 (test) and incubated for 18 h at 37°C. Subsequently, genomic DNA was extracted from 5 ml of cultures using the Qiagen genomic DNA purification kit. The screening assays were performed in duplicate. TraDIS was performed as previously described (4) The raw, de-multiplexed fastq files from MiSeq runs were filtered to capture reads containing the 12-bp Tn5-specific barcode (5'-TATAAGAGACAG-3'), allowing for 2 mismatches (fastx_barcode_splitter.pl, FASTX-Toolkit v.0.0.13). These reads were trimmed to remove the 12-bp barcode and 58-bp at the 3' end (fastx_trimmer, FASTX-Toolkit v.0.0.13), resulting in high quality sequence reads of 30-bp in length that were mapped to the EC958 chromosome (gb|HG941718) by Maq version 0.7.1 (5). Subsequent analysis steps were carried out using an in-house Perl script, as previously described (3) Fig. 1A). For analysis of combinatorial effects of zinc and reactive oxygen species on bacterial growth, growth curves were performed at 37 °C over a 12 h time course in the presence or absence of paraquat (0.1 mM) and ZnSO4 (0.1 or 0.2 mM). A starting OD600 of 0.1 was used, and readings were taken every 20 min, following a 30 s rotation (100 rpm), using a POLARstar Omega spectrophotometer (BMG Labtech).

Quantitative real-time PCR (qRT-PCR). Infected cells and/or bacteria were lysed in TRIzol
(Invitrogen) and total RNA was extracted using Direct-zol RNA Miniprep Kit (Zymo), following the manufacturer's instructions. One µg of RNA was treated with Turbo DNA-free (Invitrogen) and then reverse transcribed to cDNA using random hexamers and Superscript III (Life Technologies). Levels of zntA mRNA (relative to the control genes gapA, ifhB or rrsA) were quantified by qRT-PCR utilising the Applied Biosystems 7900HT fast RT-PCR system and the DCt method. Primers used for qRT-PCR (Sigma, Australia) are listed in SI Appendix, Table S5.    Fig. S1. Controls for regulation of zntA. (A) Growth kinetics and CFU data for EC958 grown in complete RPMI ± 0.5 mM CuSO4. OD 600 was recorded using a FLUOstar OPTIMA plate reader over a 12 h period. CFU counts were determined at the start of the experiment and after 6 h incubation (the same conditions used to determine the zntA mRNA levels reported in Fig. 1A). Data (n=3, mean ± SEM) are combined from 3 experiments. (B) Growth kinetics of EC958 and MG1655 in LB containing the indicated concentrations of CuSO4, with OD 600 recorded for 12 h as above. Data (n=3, mean ± SEM) are combined from 3 experiments. (C-D) HMDM were infected with EC958 (MOI 100) for 1 h, whereupon the media was removed and extracellular bacteria removed by gentamicin exclusion. Cells were lysed at 2, 6 or 24 h post-infection, after which RNA was extracted. Levels of zntA mRNA, relative to either ifhB (C) or rrsA (D), were determined by qPCR. Bacteria cultured for 2 h in complete RPMI media alone served as a control. Data (mean + range, n=2) are combined from 2 independent experiments, and are displayed as fold change relative to the control.

Fig. S2. Growth curves and competition assays with mutant strains for genes identified by TraDIS. (A)
Growth kinetics of wild-type (blue line) versus mutants (red line) in LB-ZnSO4 conditions. Data (n=3) are combined from 3 independent experiments, presented as mean (solid line) ± SEM (grey shades). (B) Competition assays were performed in triplicate in 96-well plates with the mutant of interest mixed in a 1:1 ratio with EC958 Δlac reference strain at a starting OD of 0.05. Each competitive pair was grown in LB and LB supplemented with 1 mM zinc, at 37°C and 200 rpm. Viable counts were performed at 0 and 18 h by serial dilution and plating on MacConkey agar. Competitive Index (CI) of each mutant compared to wild-type EC958 was calculated using the formula CI = (CFUmutant at t18/CFUmutant at t0) / (CFUEC958∆lac at t18/CFUEC958∆lac at t0), and Student's t-tests were performed to determine any significant fitness difference. * denotes p<0.05.  In E. coli, the transcription of zntA is controlled by ZntR, which binds and represses the promoter in the absence of zinc. Upon binding zinc, ZntR is converted into a transcriptional activator and activates the zntA promoter. As shown in (B), a chromosomal zinc-reporter strain was generated whereby the sequence for a fluorophore and chloramphenicol resistance was inserted directly downstream of zntA. (C) Depicts the sequence that was inserted into the pQF50 plasmid, being the promotor of zntA, fluorophore and chloramphenicol resistance. For both (B) and (C), increasing zinc concentrations induce the expression of zntA, and also result in the co-expression of a fluorophore that can be detected by flow cytometry or fluorescence microscopy.

Fig. S5. Generation and testing of chromosomal insert and plasmid-based GFP zinc-stress reporter strains. (A)
An EC958 strain constitutively expressing mKate and containing the gfp gene inserted downstream of zntA on the chromosome (EC958-zntA-GFP) was incubated in complete RPMI media containing 500 µM ZnSO4 over an 8 h time course, before median GFP and mKate fluorescence was determined by flow cytometry. Data (mean + SEM, n=3) are combined from 3 independent experiments. (B) EC958-zntA-GFP was incubated in complete RPMI media containing 500 µM ZnSO4, CuSO4, or FeSO4, with water used as a control. Data (mean + SEM, n=3) are combined from 3 independent experiments. For both A-B, statistical analysis was performed using a two-way ANOVA with Sidaks's multiple comparisons test, comparing the mean fluorescence of each time point with the control mean (** denotes p<0.01, **** p<0.0001, and all other comparisons were not significant). (C) Wild-type EC958, EC958-GFP (constitutively expresses GFP) and the two zinc stress reporter strains EC958 pQF50-zntA-GFP (plasmid-encoded zntA-GFP reporter) and EC958-zntA-GFP (chromosomal zntA-GFP reporter) were incubated in LB plus 500 µM zinc sulfate over an 8 h time-course, before GFP fluorescence was assessed by flow cytometry. In (D), EC958-zntA-GFP and EC958 pQF50-zntA-GFP were incubated in complete RPMI media containing increasing concentrations of zinc sulfate (0 to 500 µM), before being assessed by flow cytometry. For both C-D, data are combined from 2 independent experiments (n=2, mean + range). nd = not determined.   S7. Zinc concentrations within control and E. coli-infected HMDM. HMDM were plated at a density of 5 x 10 6 cells/10 mL on a 10 cm dish overnight, after which they were infected with MG1655 or EC958 (MOI 100) or left uninfected as a control. At 1 h p.i., gentamicin exclusion was performed, and at 8 h p.i., cells were washed twice with HBSS, before being lysed in 5 mL lysis buffer (0.1% SDS in MilliQ water). Zinc concentrations were determined by ICP-OES, and values obtained for media in an empty dish that had been treated identically were subtracted from the experimental samples. Data (mean + range, n=2) are combined from 2 independent experiments (each experiment contained 2 replicates) and are presented as fold change relative to the uninfected control cells. The values above each bar represent the estimated mean intramacrophage zinc concentration from the 2 experiments.  To generate these plasmids, first the bla gene in the pFCcGi vector was swapped for the cat gene to enable Cm selection in strains derived from EC958; then the location of GFP and mCherry in pFCcGi were swapped. The araC-pBAD fragment in pFCcGi was also replaced with terminators and a multiple cloning site, resulting in the vector pGcCe (A). The promoter region of zntA (zntAp) was then cloned into pGcCe to generate pGcCzntAp (B).