Background
Acid treatment is commonly used for the control or elimination of pathogenic microorganisms on the surface of medical devices or in environments, as well as in treatment of wastewater and food, as most microorganisms, including pathogenic bacteria, grow optimally at a pH range of 5–9 [
1‐
3]. For instance, in pharmaceutical and medical environments, hypochlorous acid is used as an antimicrobial agent against a wide range of microorganisms causing wound infections [
1]. A nitrous acid as a disinfectant for wastewater for 48 h treatment [
4], and a hydrochloric acid (HCl) and organic acids composite was commercially used to spray on meat [
5], and a 4 log reduction in
E. coli O157:H7 and
Listeria monocytogenes abundance on the lettuce leaf combined with chlorinated water adjusted to pH 2.5 [
6] has a synergistic effect.
However, improper or sub-lethal application of acids can induce acid tolerance response (ATR), which contributes to the survival of infectious foodborne pathogens such as enterohaemorrhagic
Escherichia coli (EHEC),
Salmonella, Shigella spp., and
Listeria monocytogenes in acidic environments, including the human gastrointestinal [
2]. The risk of infection, including foodborne illness, may increase if the pathogen survives under extreme acid stress (pH 2.0–3.0), as is found in the gastric fluid or following acid treatment of medical devices [
7]. In our previous study, 22–33% of commensal
E. coli food isolates survived in gastric pH conditions of the Korean population, and thus the antimicrobial resistance gene can be transferred from the surviving cells to resident microbiota in the human gut [
8]. However, ATR can cause many side-effects other than pathogen survival in acidic environments. For instance, that cross-protective properties can develop in acid-resistant cells is a typical example: ATR cells were reported to have increased resistance to several stresses including heat, salt, crystal violet, and antimicrobials [
9]. ATR can also affect the ability of pathogens to bind to surfaces and form biofilms, by increasing cell-surface hydrophobicity, which correlates with pathogenic adhesion to various surfaces in medical and food environments [
10,
11].
These biological variabilities provide a mechanism for foodborne pathogens to survive in changing environmental conditions, and thus, are critical targets for mitigating the risk of infectious disease [
12]. The relationship between differential gene expression and phenotypic variability, which is often determined by various omics technology, including genome-wide sequencing and transcript analysis, provides increasingly detailed insights into cellular responses to changing environments [
13]. Moreover, analysis of transcriptome changes during the adaptation of foodborne pathogens following exposure to acidic environments can provide useful information for developing management strategies to reduce the risk of infectious diseases. Although ATR of foodborne pathogens is an important issue for public health, studies on gene expression profiles and cross-protection from antimicrobials in the context of ATR of foodborne pathogens are limited.
EHEC strains can cause a spectrum of human diseases, ranging from watery diarrhoea and bloody stool to serious extraintestinal complications such as haemolytic uremic syndrome. Among the two identified strains harbouring the Shiga toxin-producing gene, a clinical isolate of ATCC 43889 was selected herein, for analysing the transcriptome changes during ATR, as this isolate was originally acid-sensitive, however, was reported to adapt to acidic conditions [
14‐
16]. In contrast, the well-known EDL933 (ATCC 43895) strain is already acid-resistant [
2,
14,
15]. Considering the importance of ATR in foodborne pathogens for public health, analysis of the transcriptome and phenotypic changes during the adaptation of such pathogens to acidic environments can provide information that can be useful for developing intervention technologies and mitigating the risk owing to ATR pathogens. Therefore, in the current study, we aimed to develop an acid-resistant ATCC 43889 strain via cell adaptation in a sub-lethal acidic environment for 100 h. We conducted RNA sequencing (RNA-seq) and quantitative reverse transcription-polymerase chain reaction (qRT-PCR) to compare differentially expressed transcripts between the acid-adapted and non-adapted bacteria. We also performed de novo whole genome sequencing (WGS) of ATCC 43889, which has not been reported previously. Finally, antimicrobial resistance and biofilm formation of the ATR strains were investigated as changes in phenotypic characteristics related to cross-protection from antimicrobials of the ATR strain.
Discussion
Exposure of ATCC 43889 cells to acidic environments induces an acid adaptation response. A previous study showed that acid adaptation at approximately pH 5 was required to obtain increased resistance to a sub-lethal pH (pH 3.0) [
14,
18]. Furthermore, acid adaptation may result in enhanced protection against lethal heat, a process often referred to as cross-protection owing to exposure of bacteria to acidic conditions. However, information regarding transcriptomic changes in ATCC 43889 adapted under sub-lethal pH conditions is limited. In this study, transcriptomic changes were observed in
E. coli adapted at the extremely acidic pH of 2.75, which is found in some food items such as vinegar [
19]. As the gastric pH in humans ranges from 2.0 to 4.8 depending on the food buffering capacity [
20],
E. coli adapted to pH 2.75 can survive in the human stomach, the first barrier against pathogens that cause foodborne illness, or transfer virulence genes, as well as antimicrobial resistance genes, to gut microbiota or to bacteria in the external environment such as food and/or biofilms [
8,
21,
22].
As predicted, the RNA-seq results showed that genes associated with stress regulation were upregulated in acid-adapted cells. In our study,
kdpA was highly upregulated (log
2 3.23-fold) in the acid-adapted cells, compared to that in the non-adapted cells, thus supporting the results of a previous study [
23].
kdpA encodes a potassium-binding subunit of a potassium-transporting ATPase, which functions to bind and transport potassium ions across the cytoplasmic membrane [
24]. Responses to acidic stress involve different cellular mechanisms, such as proton pumps [
25]. Martirosov et al. described the H
+-K
+ exchanging system, which involves a dicyclohexylcarbodiimide-sensitive exchange of 2 H
+ from cells for 1 K
+, and it is carried out through a proton pump [
26]. Furthermore, the KdpFABC complex comprises an ion channel, an ion pump, and an ABC transporter. The KdpA protein corresponds structurally to an ion channel [
27]. This is essential for the survival of
E. coli as it blocks H
+ ions from crossing the cell membrane. Moreover, upregulated
kdpA was detected in acid-adapted
Salmonella Typhimurium [
28]. In this study,
kdpA was upregulated in
E. coli that survived in acidic environments for more than 100 h. Additionally, upregulation of
bhsA (previously
ycfR)
, which encodes a putative outer membrane protein, was also observed. During chlorine treatment,
bhsA was the most significantly upregulated gene, which encodes a protein directly involved in the cellular transport of metabolites [
29,
30].
bhsA is also known to significantly induce biofilm formation in
E. coli, although no significant increase in biofilm formation was observed in the acid-adapted cells in this study.
The growth rate of the acid-adapted cells was lower than that of non-adapted cells under the same conditions, which may result from fitness cost as resistance to environmental changes is often related to reduced bacterial fitness (the ability to survive and reproduce) [
31]. Alternatively, reduced growth rate of the acid-adapted cells may be related to downregulation of
lamB,
malK, and
malE, which are associated with maltose metabolism. According to Nuoffer et al.,
E. coli produces glucose intracellularly via phosphorylation of maltose [
32]. A previous report showed that the genes regulating maltose metabolism (
malE, lamB) were highly upregulated when glucose was limited as a nutrient [
33]. We observed a decrease in the expression of maltose genes in acid-adapted cells, which is supported by the results of a previous study which reported that expression of
lamb, encoding the maltoporin precursor, was strongly reduced in various K-12 strains, as observed using SDS-PAGE, when the pH of the growth medium was decreased [
34]. Moreover, the expression of
malE, a gene associated with maltose-binding periplasmic protein maltose receptor, decreased during recovery from acid stress [
35].
malE is known to be related to alkaline induction [
36]. Therefore, downregulation of
lamB and
malE may be a hallmark of acid adaptation.
Interestingly,
gadA/B was not identified in our differential gene expression analysis, which may have been caused by two factors. First, M9 media does not contain glutamate. Previous studies have shown that the glutamate-dependent AR pathway consists of the glutamate decarboxylases
gadA/B and the glutamate/γ-aminobutyric acid antiporter
gadC, which showed low activity in the absence of glutamate [
37]. Second,
gadA/B undergoes a stepwise conformational change to its inactive form when the pH increases back to neutral, and the optimal pH for
gadABCEWX and
ybaS gene expression is pH 5.5 [
38‐
41]. Gene expression patterns differ following long- and short-term acid adaptation; for example, gene expression of
E. coli K-12 in glucose-limited media after short-term adaptation was 12.6-fold higher than that after long-term adaptation [
42]. These results suggest that gene expression is reduced after long-term adaptation.
Interestingly, in this study, the polymyxin resistance gene
arnA was upregulated in acid-adapted
E. coli O157:H7, and the antimicrobial resistance against polymyxin B and colistin confirmed the phenotypic changes in these cells. There is similar report on the development of antimicrobial resistance in the foodborne pathogens (
E. coli,
S. Typhimurium, and
S. aureus) exposed to acidic conditions, i.e., resistance against amikacin, ceftriaxone and trimethoprim for gram negative strains and gentamicin and erythromycin for gram positive strains were developed when the cells are contact with sublethal pH conditions for 24 h [
43]. However, the upregulated
arnA did not affect the MIC of polymyxins in acid-adapted cells. According to Sinel et al., a 14.3-fold upregulation of the gene encoding the quinolone resistant protein resulted in a 2- to 6-fold increase in MIC for
Enterococcus faecium [
44]. Owing to the prevalence of antibiotic resistance and lack of novel classes of antibiotics in the development pipeline, clinical use of polymyxins has significantly increased over the past decades [
45]. The most common mechanism of acquired resistance to polymyxins is modification of the bacterial outer membrane lipopolysaccharide. Global epidemiological surveillance studies have reported the occurrence of polymyxin resistance to be common in Enterobacteriaceae, specifically in
Enterobacter species and in
Acinetobacter baumannii [
46]. Considering the limited number of agents available for treating infections caused by multidrug-resistant gram-negative organisms [
46], the emergence of polymyxin resistance, particularly from the cross-protection of acid-adapted EHEC, is of considerable concern. In this study, increased biofilm formation was observed at sub-MIC concentrations of polymyxins, which provides additional protection against environmental stresses including antimicrobials. Similar results on increased biofilm formation in
E. coli were observed under sub-MIC concentrations of glycopeptide, cyclic peptide, fluoroquinolone, and β-lactam [
47].
This study has some limitations; acid adaptation procedure was not a typical, but considering practices in the food industries or medical environments. It is more realistic to alternate exposure to acidic and neutral environments. We only observed phenotypic changes and DEG of acid-adapted E. coli O157:H7 but molecular basis was not explored. Thus, studies on the phenotypic changes and DEGs of other types of pathogenic E. coli and molecular basis (e.g. gene deletion and/or protein study) of the acid-adapted cells are needed in the future work. However, phenotypic changes in ATR of pathogenic bacteria with relation to gene expression can be incorporated in safety management of infectious pathogens to ensure public health.
Materials and methods
Preparation of bacterial strain and acid adaptation
The
E. coli O157:H7 (Migula) Castellani and Chalmers ATCC® 43889™ strain was obtained from the National Culture Collection for Pathogens (Osong, Korea). The commensal
E. coli ATCC® 10536™, obtained from the Korean Type Culture Collection (Daejeon, Korea), was used as negative control. Bacteria were inoculated into 5 mL of tryptic soy broth (TSB) (Merck, Darmstadt, Germany) in 17 × 100-mm glass culture tubes at 37 °C for 24 h in a shaking incubator at 140 rpm (SI-600R, Jeio Tech, Korea). A volume of 50 μL of the broth cultures of a stationary phase was subsequently transferred to 5 mL of M9 minimal media containing 0.4% glucose supplemented (MB Cell, Los Angeles, CA, USA) at 37 °C for 24 h with agitation. The cells were harvested, washed, and suspended in phosphate buffered saline (PBS) (Gibco, Rockville, MD) to obtain a stationary phase-adjusted concentration of approximately 10
7 CFU/mL. For the acid-adapted ATCC 43889 strain, a 200-μL aliquot of the cultures was transferred to 10 mL M9 media without supplement at pH 4. The pH of M9 (control, pH 7.4) was adjusted by adding 5 N hydrochloric acid (Kanto Chemical, Tokyo, Japan; PubChem CID: 313) before inoculation. Next, the inoculated acidic M9 was incubated for 24, 50, 75, and 100 h with agitation without medium change. After incubation, the cultures were serially diluted and the adapted cells were recovered in triplicate on Luria Bertani agar (LB) (Merck) and sorbitol-MacConkey agar (SMAC) (Merck) at 37 °C for 24 h. Following 100 h of incubation in acidic growth media, the surviving bacteria were recovered after incubation on agar plates for 24 h and the colonies were counted. If colonies were recovered on both agar plates, a single colony was picked from SMAC agar and the process was repeated in pH reduced by 0.25 until no colonies were recovered. The pH of M9 media immediately preceding the pH at which no colonies were obtained was considered the final acid-adapted pH. At the final pH, a 200-μL aliquot of acid-adapted and non-adapted cells were transferred to 10 mL of acidic M9 media without supplement and incubated for 200 h with agitation without medium change to analyse survival as presented in Fig.
1. Survival of
E. coli in acidic M9 media were performed in duplicate and the data presented were obtained from at least three independent experiments.
DNA isolation, WGS, assembly, and functional annotation
The experimental procedures are described in detail in the Additional file
1.
Total RNA isolation and RNA-seq
RNA-seq analysis was used to compare gene expression between acid-adapted and non-adapted cells. Total RNAs were isolated from acid-adapted cells that survived at pH 2.75 of M9 medium for 150 h and the corresponding non-adapted cells using a Qiazol lysis reagent (Qiagen GmbH, Hilden, Germany) according to the manufacturer’s instructions. The DNA in each sample was removed using an on-column digestion using RNase-free DNase I (Qiagen).
Following RNA extraction and DNase I treatment, ribosomal RNA (rRNA) was removed using the Ribo-Zero rRNA removal kit (Bacteria) (Illumina, San Diego, CA, USA) and cDNA was synthesised using DNA polymerase I and RNase H (Illumina) according to the manufacturer’s instructions. The quantity and quality of the total RNA was analysed using NanoVue Plus (GE Healthcare, Buckinghamshire, UK) and a 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA) with an RNA integrity number (RIN) ≥ 8. The purified expression libraries were sequenced (100 base pair (bp) × 2) using a HiSeq 2000 (Illumina) platform. The presented RNA-sequencing data were obtained from an independent experiment and analysed in duplicate.
Quality control and data analysis of RNA-seq
Sequence read data were analysed with FastQC (v0.10), which provides a modular set of analyses and can rapidly reveal sequence quality [
48]. In addition, Trimmomatic (v0.32) was used to trim and crop Illumina (FASTQ) data and remove adapters [
49], whereas Bowtie was used to align sequences [
50]. DEGs were analysed using edgeR and the p-value was obtained from total count normalisation [
51].
An in-house script was used to calculate the reads per kilobase per million (RPKM) for individual transcripts [
52]. At least one sample with an RPKM value of zero was excluded from the analysis. The quintile method of normalisation was then applied to reduce systematic bias [
53]. Genes with differential expression indicated by an absolute log
2 (fold change) ≥ 2 were selected.
qRT-PCR validation
qRT-PCR was performed to confirm the DEGs identified using RNA-seq. The primer sequences of candidate genes/transcripts and a housekeeping gene (16S rRNA) were designed using NCBI Primer BLAST (Additional file
1: Table S5). cDNA was synthesised using the Maxima first strand cDNA synthesis kit for qRT-PCR (Thermo Scientific, Waltham, MA, USA) according to the manufacturer’s instructions. SYBR Green (Thermo Scientific) PCR (20 μL) was performed in duplicate for each sample using 250 ng cDNA and 300 nM each of the forward and reverse primers. Forty cycles of amplification and data acquisition were performed on a PikoReal 96 real-time PCR system (Thermo Scientific) (Additional file
1: Table S5). The 2
−ΔΔCt method was used to evaluate the expression levels of each target gene compared to that of the 16S rRNA internal control [
54]. All experiments were performed in duplicate and the data presented were obtained from at least three independent experiments.
Phenotyping for growth rate and antimicrobial resistance
Strains that show differential expression of genes under different conditions may show phenotypic variations. To compare growth rate, the acid-adapted and non-adapted cells were incubated in 5 mL TSB at 37 °C for 24 h with agitation, and the absorbance of the broth cultures was evaluated using a BioTek synergy Mx (BioTek Instruments, Winooski, VT, USA) at 600 nm.
To determine the antimicrobial resistance of acid-adapted and non-adapted cells, the MIC of polymyxins was determined by broth microdilution according to a previous study and recommendation of the joint CLSI-EUCAST polymyxin breakpoints, cation-adjusted Mueller–Hinton Broth (Merck) using concentrations of 400,000 U/L polymyxin B sulfate salt (Sigma; PubChem CID: 9833652) and 8 mg/L colistin sulfate salt (Sigma; PubChem CID: 73090) and serially diluted without surfactants (i.e. polysorbate-80) [
55,
56]. Polymyxin agar test was also performed [
17]. Muller Hinton agar (MHA) plates were prepared with 12,500 U/L polymyxin B sulfate salt and 1 mg/L colistin sulfate salt, which were then inoculated with 100 μL of 0.5 McFarland suspension and incubated for 20 h at 37 °C. Plates with visible growth of microorganisms was read as positive.
Survival of acid-adapted and non-adapted cells in SGF with pH adjusted to 2.75 was determined. A 100 μL bacterial suspension was inoculated into 10 mL of SGF prepared as previously described [
57]. The culture was incubated at 37 °C with agitation for 180 min, then 100 μL of the bacterial suspension was plated in triplicate onto SMAC.
Biofilms were formed in the presence of polymyxin B (0, 6250, 12,500, and 25,000 U/L) and colistin (0, 0.125, 0.25, 0.5, 1.0, and 2.0 mg/L) for 1, 2, and 7 days using the procedure developed by Zhu and Mekalanos [
58].
All the above experiments were performed in duplicate and the data presented were obtained from at least three independent experiments.
Statistical analyses of experimental data
The generated data were analysed for statistical significance using the paired two-tailed Student’s t-test of GraphPad Prism 8 (GraphPad Software, Inc., San Diego, CA). P values < 0.05 were considered statistically significant.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.