TUV93-0, a laboratory-adapted derivative of EDL933 (E. coli O157:H7) lacking the Shiga-toxin gene (stxA1), was used during this study (Campellone et al., 2002). For all studies, isolates were grown aerobically at 37°C on Luria-Bertani (LB) agar or broth as required.

Four disinfectants were selected, representing different chemical compositions commonly used on farms. Those selected contained the following active ingredients: peroxymonosulphate (PMS), quaternary ammonium compound (QAC) and glutaraldehyde, iodophor (IOD) and chlorocresol (CRE). In addition, a second disinfectant (QAC2), which contained quaternary ammonium compounds, was used for selected experiments. General Orders (GO) concentrations for each Defra-approved disinfectant are available.⁴ For products QAC, IOD, and CRE, a 4 x GO stock was prepared and a 2-fold dilution series was created to achieve 12 dilutions. The final concentration ranges were as follows: the QAC disinfectant ranged from 2.04% (v/v) to 0.001% (v/v), the CRE disinfectant from 2.17% (v/v) to 0.0011% (v/v) and the IOD disinfectant from 1.16% (v/v) to 0.0006% (v/v). For the PMS disinfectant, a 2x GO stock was prepared and a 2-fold dilution series was created to achieve 12 dilutions. The final concentration ranged from 0.5% (w/v) and 0.0002% (w/v). All disinfectants were diluted using WHO hard water (2 mM calcium chloride and 0.68 mM magnesium chloride).

The survival and growth kinetics of TUV93-0 were determined using the FLUOstar (BMG, Labtech). TUV93-0 was grown for 16 h at 37°C in LB broth, then diluted (1,100) into fresh broth, with 150 μL of the diluted cells aliquoted into the wells of a 96-well plate. The Cells were grown for 150 min at 37°C in the FLUOstar, with absorbance (600 nm) measured at 15 min intervals. After 150 min, 50 μL of the diluted disinfectant or WHO water (control) was added, and the growth was monitored for a further 21.5 h. Each experiment was performed in triplicate, and the growth was compared to that of the cells grown in the presence of WHO water. The area under the curve (AUC) values were calculated for each growth curve using GraphPad Prism software.

TUV93-0 was grown overnight in LB broth at 37°C, then diluted (1:100) into fresh broth and grown for 2.5 h at 37°C with shaking (150 rpm) until the disinfectant was added at sub-inhibitory concentrations and grown for a further 21.5 h. The following day, the cultures were diluted (1:100), and the exposure was repeated every 24 h for a total of 72 h. After each passage, a sample of the cells was frozen at-80°C for future analysis.

The frozen stocks of E. coli exposed to the disinfectant were diluted in PBS to achieve 100–500 colonies/plate. Then, they were plated onto LB agar and grown for 16 h at 37°C. The colonies were replica plated onto LB-agar supplemented with either 8 mg/L nalidixic acid (8Nal) or 4 mg/L tetracycline (4Tet) using replicating cloths and incubated. The resulting colonies were counted, and some were selected for further analysis.

For the selected mutants obtained following repeated disinfectant exposure, Minimum inhibitory concentration (MIC) values were determined for ampicillin (Amp), tetracycline (Tet), chloramphenicol (Chl) and nalidixic acid (Nal) using agar dilution methods described elsewhere (Andrews, 2001). Control strains ATCC 25922 and NTCC 10418 were included along with TUV93-0. The breakpoints applied to the MIC values were as follows: Amp >8 mg/L, Tet >4 mg/L, Chl >8 mg/L and Nal >16 mg/L.

The derivatives of TUV93-0 generated by disinfectant exposure and selected from 8Nal were assessed for mutations in gyrA, parC and parE. All three genes were amplified with previously described methods (Kim et al., 2009; Ling et al., 2003) and sequenced with ABI technology. The sequences were aligned using SeqMan Pro (DNAstar Lasergene), and single nucleotide polymorphisms (SNPs) were identified and compared to known mutations.

The selected isolates were whole genome sequenced (WGS) to identify the SNPs associated with antimicrobial resistance (AMR). DNA was extracted using the MagMAX CORE nucleic acid purification kit (ThermoFisher Scientific) and a KingFisher duo prime. The DNA was sequenced on the MiSeq platform (Illumina) following library preparation using Nextera XT. TUV93-0 was assembled with SPAdes 3.12 (Bankevich et al., 2012) and annotated using the Escherichia-specific database with Prokka 1.11 (Seemann, 2014) to produce a GenBank file. The mutants were mapped against the TUV93-0 GenBank file using Snippy 3.1 (Seemann, 2015) to identify the SNPs. Wherever indels or deletions were indicated by Snippy, the gene was translated to check for stop codons using the ExPASy translate tool.⁵ Valid SNPs or deletions were those with a depth of coverage >10 and 100% agreement for the mutated sequence across all reads.

RNA-sequencing (RNA-seq) was used to assess the effect of disinfectant exposure on gene expression. RNA was isolated from the TUV93-0 cells exposed to the PMS disinfectant and QAC disinfectant at final concentrations of 0.125% w/v and 0.002% v/v, respectively, for 24 h at 37°C. The RNA was stabilised using a solution of 95% Ethanol and 5% phenol for a minimum of 30 min and then frozen at-80°C until the RNA was extracted. The RNA was extracted from the pelleted cells using the SV total RNA isolation kit (Promega).

Ribosomal RNA was removed using the Ribo-Zero bacterial rRNA depletion kit (Illumina) with 2.5 μg of total RNA, following the manufacturer’s protocol. The remaining RNA was purified using magnetic beads and a modified RNAeasy MinElute clean-up method (Promega). The RNA was then converted to cDNA by reverse transcription using random hexamers (Roche or similar). The RNA samples were mixed with 400 μM of the random hexamers and incubated at 70°C for 10 min. The cDNA reaction mix contained AMV-reverse transcription buffer (1x), 100 mM DTT, 10 mM dNTPs, 12.5 U of Protector RNase inhibitor and 25 U of AMV reverse transcriptase and was incubated at 25°C for 10 min followed by 42°C for 60 min. The cDNA synthesis system (Roche or similar) was used to generate the second strand of the cDNA. The samples were purified using magnetic beads (AMPure, Agencourt) according to the manufacturer’s instructions before the ligation of the sequencing primer. The Nextera XT kit was used to generate the sequencing libraries before sequencing on the NextSeq (Illumina).

Reads in the fastq files were trimmed using trimmomatric (Version 0.39; Bolger et al., 2014) with a sliding window of 10 bases, a quality cutoff of 20 and a minimum read length of 80. The reads were mapped to the E.coli O157:H7 EDL933 genome (accession number: AE005174) using the aligner Smalt (version 0.7.6; Institute S, 2014) and samtools (version 0.1.19; Li et al., 2009). FeatureCounts (Liao et al., 2014) was used to count the number of reads mapped to each gene, and the R package Deseq2 (Love et al., 2014) was used to carry out four different comparisons (Table 1) and generate a summary of the number of differentially expressed genes (DEGs; adjusted p < 0.05, Log2 Fold Change >2 and both filters combined). Volcano plots for the six contrasts are shown in Appendix 5. The gene ontology of the DEGs was determined using DAVID⁶ and Biocyc.⁷

The sub-lethal concentration was determined for each disinfectant so that these levels could be used to assess transient and permanent genomic changes. For the O157:H7 strain TUV93-0, the lowest lethal concentrations were 0.5x (0.58% (v/v)) and 0.25x (0.25% (w/v)) GO strength for iodophor disinfectant (IOD) and peroxymonosulphate disinfectant (PMS), respectively. The chlorocresol-containing disinfectant (CRE) and glutaraldehyde and quaternary ammonium compound (QAC) inhibited the growth at much lower concentrations of 0.033x GO (0.067% (v/v)) and 0.002x GO (0.004% (v/v)), respectively. No growth inhibition was observed in the water control. Therefore, sub-lethal was defined as the highest concentration of the disinfectants at which growth was observed. Although the maximum OD 600 nm achieved after 24 h growth was lower for some disinfectants compared to the water control (Supplementary Data 1B).

Transcriptional changes were assessed following the exposure of TUV93-0 to the sub-lethal disinfectant concentrations using RNA-seq. In the presence of the PMS disinfectant, 442 genes were differentially expressed with a p-value <0.05 and log₂ fold change >1 or < −1 (Table 1) when compared to the water control (Supplementary Data 1). The exposure of E. coli O157:H7 to the PMS disinfectant resulted in the upregulation of the expression of 341 genes. The differentially expressed genes (DEGs) were divided into functional groups using DAVID (Figure 1). The DEGs had a range of functions and included genes associated with transcription and translation, metabolic pathways, respiration and enzymes (Figure 1; Supplementary Data 2). The DEGs with the greatest fold change increase (>4 fold) were the ribosomal proteins (rplC, rplW and rpsJ) and several uncharacterised loci (Z5140, Z5881 and Z3062). A total of 215 genes (181 upregulated and 34 downregulated) were hypothetical proteins without functional classification.

Differential regulation of the genes associated with the bacterial membrane was observed (Figure 1) and included 12 genes associated with LPS synthesis (wbdR, wbdO, wbdN, wbdQ, wbdP, waaD, wzy, waaI, waaL, waaY, waaJ and per/rfbE), which were upregulated between 1.9 and 3.4-fold. In addition, one of the major outer membrane porins, ompC, known to be essential for the survival of E. coli in low pH conditions (Wang et al., 2021), was upregulated in response to the PMS disinfectant. Nineteen other membrane-associated genes were downregulated in TUV93-0 in response to the PMS disinfectant and included the osmotic shock proteins osmB and osmE, which maintain cell integrity and encode putative lipoproteins involved in the stress response (Brauer et al., 2023; Figure 1). Transporters can play a role in the response to disinfectant exposure as they facilitate transport across the cell membrane. Seven genes associated with transport across the cell membrane were differentially expressed, with ygjU and argT downregulated (1.2 to 1.8-fold) and yaeT, yaeC, livJ, rbsB and malE upregulated (1.3 to 2.6-fold; Supplementary Data 2). A stress response is associated with exposure to harsh environmental conditions, and eight stress-associated DEGs were identified in PMS-exposed TUV93-0 (Figure 1), six of which were surprisingly downregulated in the presence of the disinfectant.

Several virulence-associated genes of E. coli O157 were upregulated in response to the PMS disinfectant and included the type III secretion genes (escC, escJ, escD, escN, escR, sepQ, sepZ, sepL, escT and espD) and genes required for cell adherence to the gut epithelium (tir, eae; Figure 1). In addition, a 2-fold upregulation of the virulence-associated haemolysin E gene (hlyE) was observed.

Twelve DEGs were identified when TUV93-0 (p < 0.05 and log₂ fold change >1 or < −1) was exposed to the QAC disinfectant and compared to the control (TUV93-0 exposed to water), four of which were downregulated and eight were upregulated under these conditions (Table 2). Five genes associated with the phage shock protein operon (pspABCDE) were upregulated >5-fold in the presence of the disinfectant. Two genes associated with heat shock were also upregulated (mopB and htpX), along with one hypothetical protein (yjbO). The genes downregulated during the disinfectant exposure included gadA and gadC (xasA), which are involved in response to acidic conditions, as well as cyoC and a putative transporter protein (Z3658).

Repeated exposure to sub-lethal concentrations of biocides has been shown to induce mutations in Salmonella, leading to antibiotic and biocide resistance. The repeated exposure of TUV93-0 to sub-lethal disinfectant levels was assessed for four disinfectants and water (control) for up to three passages. Approximately 3.2×10⁴ (1.67×10⁴ screened for tetracycline resistance and 1.59×10⁴ screened for nalidixic acid resistance) colonies were screened for resistance to antimicrobials acquired through chromosomal mutations for all disinfectants and controls. Twenty-five E. coli colonies (0.15% of the total colonies were screened for tetracycline resistance) generated from exposure to any disinfectant or control grew on 4Tet. Seven hundred and one E. coli colonies (4.4% of the total colonies screened for nalidixic acid resistance) generated from exposure to all disinfectants and controls grew on 8Nal (Supplementary Data 1A). Eighty-five per cent of the Nal^R colonies were generated following QAC exposure.

Nal MICs were determined for selected passaged isolates, with the levels increasing from 2 mg/L (TUV93-0 parent strain) to between 16 and 64 mg/L (Table 3). Sequencing of the topoisomerase genes (gyA, gyrB, parC and parE) in selected mutants, either through sequence analysis of PCR-generate amplicons (gyrA, parC and parE) or through whole genome sequencing (gyrB), identified several non-synonymous single nucleotide polymorphisms (SNP) in gyrA (G74, D82 and D87) and gyrB (K447; Table 4). The mutation of the amino acid D87 in gyrA was associated with the increased MIC of 64 mg/L. Mutations in the gyrA and gyrB genes had no effect on resistance to Amp, Tet and Chl.

The derivatives of TUV93-0, generated following exposure to the sub-lethal QAC disinfectant, developed elevated Chl MIC (16–32 mg/L) and Tet MIC (2–8 mg/L; Table 3). This represented an increase between 3 and 4 doubling dilutions compared to TUV93-0. The analysis of the growth kinetics of QAC4 and QAC55 in the presence of the QAC disinfectant (Appendix 3) revealed that both isolates could grow in the disinfectant at a concentration of 0.004% (v/v), while TUV93-0 (parent strain) and QAC21 were unable to grow at this concentration. The isolates QAC4 and QAC55 were also able to grow in the presence of higher concentrations of a second QAC disinfectant (QAC2) compared to TUV93-0 or QAC21 (Supplementary Data 1B). The increase in tolerance to the disinfectant shown by QAC4 and QAC55 was not observed in the presence of the PMS disinfectant (data not shown).

The WGS analysis of QAC4, QAC7, QAC33 and QAC55 identified deletions and SNPs in the genes, including marR and toxB (Table 4). QAC4 and QAC7 also had identical SNPs in a hypothetical protein 01831. Frameshift deletions in marR and the gene encoding 01831 resulted in truncated proteins, while the changes in toxB were silent. In contrast to QAC55, marR carried an insertion that resulted in a frameshift, unlike the mutations observed in other derivatives. Mutations in the hypothetical protein 01831 and toxB were identical for all sequenced isolates. QAC55 carried three additional mutations not seen in the other isolates, which were non-synonymous SNPs in kup, lon and lacA (Table 4). Mutations in marR, toxB and the hypothetical protein 01831 were also observed in the derivatives of TUV93-0 with reduced susceptibility to Nal (QAC8 and QAC13; Table 4).

Two isolates (QAC4 and QAC8) generated following passage in the QAC disinfectant were selected for transcriptome analysis using RNA-seq. These isolates were selected because QAC4exhibited a high MIC for Chl and Tet () and QAC8 exhibited an elevated MIC for Nal without topoisomerase mutations (). Twenty-six genes were upregulated in TUV93-0 compared to QAC4, while 22 genes were downregulated (Figure 2A; Supplementary Data 2). These genes were functionally classified using DAVID and Biocyc based on biological processes, wherever possible. Exposure of QAC4 to the QAC disinfectant led to the overexpression of ribosomal genes (rplC, rplD, rplV, rpsC, rpsF, rpsJ and rpsS), genes related to pH regulation and hypothetical prophage-related genes (Z1835, Z1842, Z1847, Z1849 and Z1854). There was a 3 to 5-fold upregulation in expression the pH response genes gadABC, gadE (yhiE) and hdeABD in QAC4 (Figure 2A). In addition, the genes yhiM, yhiF and slp, known to be associated with glutamate-dependant acid resistance, were upregulated (1.9 to 3.4-fold) in QAC4 in response to the disinfectant.

The genes associated with metabolism (ftsJ, glnA, glpD, miaA), stress response (htpX, htrA) regulation (mtlR), transport (mtr), phage shock (pspABCDE and yjbO) and six hypothetical proteins were upregulated in TUV93-0 compared to QAC4 (Figure 2A). The phage shock-associated genes demonstrated a 5-fold increase in the expression within TUV93-0 (Figure 2A).

Sixteen genes were upregulated in TUV93-0 compared to QAC8, while four genes were downregulated in response to the QAC disinfectant (Figure 2B). Six genes associated with phage shock (pspABCDE and yjbO) were upregulated between 2-and 4-fold in TUV93-0 compared to QAC8. In addition, the genes associated with metabolism (ynhG, glpD and cmk), gene regulation (mtlR), transport (ynfM) and stress (htpX) were upregulated in TUV93-0 compared to QAC8. Four DEGs that encoded the hypothetical proteins (Z1638, Z1198, ynhG and yejG) were also upregulated in TUV93-0. The genes upregulated in QAC8 were involved in aerobic respiration (sucB, sucD and nuoG) and included a membrane protein (wbdP). A comparison of the adapted isolates in the presence of the disinfectant and water (control) was performed but not included in this study. For QAC4, no significant DEGs were detected between the QAC disinfectant-and water-treated cells, suggesting that the changes elicited by the disinfectant were unaffected by further disinfectant exposure. For QAC8, six DEGs were identified, including the psp genes.

All sequence data from this study are deposited in the European Nucleotide Archive (ENA) within project PRJEB34679. The mutants generated by exposure to disinfectant are stored under the following accession numbers: SAMEA7566848, SAMEA7566847, SAMEA7566852, SAMEA7566851, SAMEA7566846, SAMEA7566850 and SAMEA7566849. The data from the transcriptome analysis are stored under the following sample accession numbers: SAMEA5986189, SAMEA5986181, SAMEA5986175, SAMEA7566633, SAMEA5986184, SAMEA5986177, SAMEA5986173, SAMEA7566628, SAMEA7566631, SAMEA5986182, SAMEA7566624, SAMEA7566626, SAMEA5986179, SAMEA5986186, SAMEA7566629, SAMEA7566630, SAMEA7566632, SAMEA7566622, SAMEA7566623, SAMEA7566625 and SAMEA7566627.