ABSTRACT
Some chlorine-resistant Escherichia coli isolates harbor the locus of heat resistance (LHR), a genomic island conferring heat resistance. In this study, the protective effect of the LHR for cells challenged by chlorine and oxidative stress was quantified. Cloning of the LHR protected against NaClO (32 mM; 5 min), H2O2 (120 mM; 5 min), and peroxyacetic acid (105 mg/liter; 5 min) but not against 5.8 mM KIO4, 10 mM acrolein, or 75 mg/liter allyl isothiocyanate. The lethality of oxidizing treatments for LHR-negative strains of E. coli was about 2 log10 CFU/ml higher than that for LHR-positive strains of E. coli. The oxidation of cytoplasmic proteins and membrane lipids was quantified with the fusion probe roGFP2-Orp1 and the fluorescent probe BODIPY581/591, respectively. The fragment of the LHR coding for heat shock proteins protected cytoplasmic proteins but not membrane lipids against oxidation. The middle fragment of the LHR protected against the oxidation of membrane lipids but not of cytoplasmic proteins. The addition of H2O2, NaClO, and peroxyacetic acid also induced green fluorescent protein (GFP) expression in the oxidation-sensitive reporter strain E. coli O104:H4 Δstx2::gfp::amp. Cloning of pLHR reduced phage induction in E. coli O104:H4 Δstx2::gfp::amp after treatment with oxidizing chemicals. Screening of 160 strains of Shiga toxin-producing E. coli (STEC) revealed that none of them harbors the LHR, additionally suggesting that the LHR and Stx prophages are mutually exclusive. Taking our findings together, the contribution of the LHR to resistance to chlorine and oxidative stress is based on the protection of multiple cellular targets by different proteins encoded by the genetic island.
IMPORTANCE Chlorine treatments are used in water and wastewater sanitation; the resistance of Escherichia coli to chlorine is thus of concern to public health. We show that a genetic island termed the locus of heat resistance (LHR) protects E. coli not only against heat but also against chlorine and other oxidizing chemicals, adding to our knowledge of the tools used by E. coli to resist stress. Specific detection of the oxidation of different cellular targets in combination with the cloning of fragments of the LHR provided insight into mechanisms of protection and demonstrated that different fragments of the LHR protect different cellular targets. In E. coli, the presence of the LHR virtually always excluded other virulence factors. It is tempting to speculate that the LHR is maintained by strains of E. coli with an environmental lifestyle but is excluded by pathogenic strains that adapted to interact with vertebrate hosts.
INTRODUCTION
Escherichia coli is a common member of the microbiota of the gastrointestinal tracts of vertebrate animals. Most strains of E. coli are nonpathogenic, but pathogenic strains cause enteric and extraintestinal diseases (1, 2). An important source of exposure to pathogenic E. coli is contaminated water used for irrigation, drinking, and the processing of fruits and vegetables (3, 4). Water sanitation can be achieved by ozonation, UV light, and, most commonly, chlorination (5, 6). Hypochlorous acid (HOCl), the active compound of water chlorination, reacts with multiple cellular macromolecules, such as proteins, nucleic acids, and lipids (7). However, bacteria develop resistance to chlorination, and recovery of viability by bacterial cells following chlorination has been observed (8). In particular, strains of E. coli isolated from wastewater showed high resistance to chlorine (9).
Chlorine-specific resistance in E. coli involves three chlorine-sensitive transcription factors, hypT, rclR, and nemR, which are activated specifically by chlorine ion oxidation (10–12). The chlorine resistance of E. coli is also mediated through the RpoS-regulated general stress response (13), the oxidative stress regulons oxyR and soxR (14, 15), and heat shock proteins (16). An overview of the mechanisms affecting chlorine resistance in E. coli is shown in Fig. 1.
Cytoplasmic determinants of chlorine resistance in E. coli. (A) Chlorine-specific transcription factors. HypT is specifically activated by chlorine through methionine oxidation, and Cys4 becomes oxidized and inhibits DNA binding to avoid unnecessary regulation of target genes (10). Proteins encoded by rclR form a membrane-associated complex responsible for reducing cellular components specifically oxidized by chlorine (11). The NemR-mediated chlorine response relies on the reversible oxidative modification of the conserved Cys106 (12). (B) Oxidative stress response. The insufficiency of NADPH leads to SoxR reduction, and the oxidized Fe−S clusters trigger a conformational change of SoxR (59). Chlorine activates OxyR by the formation of a disulfide bond (14), and then OxyR is regenerated by the glutaredoxin–glutathione–glutathione reductase (Grx/GSH/Gor) system upon return to nonstress conditions (15). Oxidized OxyR activates the grxA and katG genes, encoding enzymes involved in chlorine resistance, and sufABC for the repair of damaged Fe–S clusters (60). (C) General stress response. RpoS regulates the transcription of katE (encoding catalase), dps (encoding a DNA-binding protein), tktB (encoding transketolase 2), and grxB (encoding glutaredoxin 2), which act against chlorine resistance (13). (D) Prevention of protein aggregation. The reduced Hsp33 in chlorine leads to the simultaneous formation of two intramolecular disulfide bonds and binds to unfolding substrate proteins, whereas DnaK is inactivated because of the low ATP level. When cellular ATP levels are restored, Hsp33 becomes reduced, and bound substrates are transferred to the DnaK system for refolding (16). Inorganic polyphosphate (polyP) forms stable complexes with unfolding proteins. After the relief of stress, polyphosphate can be reconverted to ATP, which can be used by DnaK to refold polyphosphate-protected proteins (61). RidA is N-chlorinated by chlorine and then binds to a wide range of unfolded client proteins, preventing their aggregation. Under nonstress condition, the chlorine is removed, leading to the release of the client protein and protein refolding (62).
A high proportion of chlorine-resistant isolates of E. coli recovered from wastewater also harbor the locus of heat resistance (LHR) (9), a genomic island that mediates extreme heat resistance in E. coli (17). The genes on the LHR are predicted to encode proteins associated with responses to heat shock, cell envelope stress, and oxidative stress (18). The putative mechanisms of LHR-mediated heat resistance (18) overlap the chlorine resistance mechanisms (Fig. 1). The LHR-encoded heat shock proteins sHSP20, ClpKGI, and sHSPGI may prevent chlorine-mediated protein aggregation. The LHR additionally encodes a homologue of the oxidoreductase thioredoxin. The activities of the enzymes encoded in the LHR indicate a potential role in resistance to chlorine and oxidative stresses during water treatment, although this role has not been confirmed experimentally. Moreover, it remains unknown to what extent heat resistance and chlorine resistance overlap with the presence of E. coli virulence factors. Current knowledge is thus insufficient to assess the frequency of pathogenic strains of E. coli that use LHR-mediated resistance to oxidative water treatment. Of particular concern are Shiga toxin-producing E. coli (STEC) strains that cause foodborne and waterborne illness outbreaks (19–21). The definitive virulence factor of STEC is Shiga toxin (also termed verotoxin), and STEC strains may have high infectivity, with a 1 to 10% risk of infection upon exposure to a single cell (1, 22). In addition, a growing body of evidence suggests that uropathogenic E. coli (UPEC) appears to differentially survive wastewater treatments and may also be transmitted by contaminated water (23–26). UPEC strains do not share a defined set of virulence factors but are described on the basis of their ability to cause infections in the urinary tract and bladder. This study therefore aimed to investigate whether the LHR confers resistance to chlorine in E. coli and to determine the frequency of the LHR in strains of STEC and UPEC.
RESULTS
The LHR confers resistance to oxidizing chemicals on E. coli.A previous study found that 59% of 70 E. coli isolates from chlorinated sewage carried the LHR (9). In contrast, none of the STEC groundwater isolates tested in this study carried the LHR. To test the hypothesis that the LHR contributes to chlorine resistance, the heat and chlorine resistance of 10 LHR-positive and 10 LHR-negative E. coli strains isolated from wastewater was determined (Fig. 2). Ten LHR-positive wastewater isolates were randomly selected and matched with 10 randomly selected LHR-negative isolates. Two LHR-positive strains, E. coli AW1.3 and E. coli AW1.7, and the LHR-negative strain E. coli AW1.7ΔpHR1 served as controls. The reductions in the cell counts of E. coli AW1.7 after heat and chlorine treatments were about 6 and 2 log10 CFU/ml lower, respectively, than the reductions in the cell counts of E. coli AW1.7ΔpHR1, a heat-sensitive derivative of AW1.7. Similarly, the lethality of chlorine treatment against LHR-positive wastewater isolates ranged from 1 to 2 log(N0/N) [where log(N0/N) is the log-transformed ratio of cell counts before treatment to cell counts after treatment], while chlorine lethality against LHR-negative wastewater isolates ranged from 3.5 to 6 log(N0/N) (Fig. 2).
Lethality of treatments with heat or chlorine to 23 strains of E. coli. Treatment lethality is expressed as the log-transformed ratio of cell counts before treatment (N0) to cell counts after treatment (N). Filled bars represent cells treated at 60°C for 5 min; open bars represent cells treated with 15 mM NaClO for 25 min. E. coli AW1.7ΔpHR1 and FUA strains are LHR negative; the other 12 strains are LHR positive. Data are shown as means ± standard deviations for three independent experiments.
The contributions of genes encoded by the LHR to resistance to chlorine and other oxidants were confirmed by cloning fragments of the LHR into E. coli MG1655, followed by determination of the lethality of NaClO, H2O2, peroxyacetic acid (PAA), and KIO4. Acrolein and allyl isothiocyanate (AITC) were additionally used as oxidizing chemicals that E. coli may encounter in natural habitats. The LHR or LHR fragments were introduced into E. coli MG1655 after cloning into the low-copy-number vector pRK767. E. coli MG1655 transformed with pRK767 served as a vector control (Fig. 3). Cloning of the LHR protected against challenge with NaClO, H2O2, and PAA but provided no protection against KIO4, acrolein, or AITC. Protection against NaClO, H2O2, and PAA was provided by all plasmids encoding the full-length LHR or fragment 1 or 2; plasmid pRF3 was less effective (against NaClO and H2O2) or ineffective (against PAA). Remarkably, cloning of the LHR or any of its parts into LHR-negative E. coli increased sensitivity to AITC (Fig. 3).
(A) Schematic representation of the locus of heat resistance and putative functions encoded by genes located on the genomic island. Genes that are expressed in E. coli MG1655(pLHR) are framed and printed in boldface. Proteins are color coded based on their predicted functions: red, heat shock proteins; yellow, hypothetical proteins with a possible relationship to envelope stress; blue, proteins related to oxidative stress; orange, DegP, with possible relationships to signaling in the Cpx, EvgA, and σE pathways. Genes carry the subscript “GI” for genomic island if an orthologue of the same gene is present in E. coli genomes. Open reading frames are numbered if there is no known function associated with the genes. Predicted promoters are indicated by arrows; the gray caret on the right indicates a predicted terminator. The three fragments of the LHR that were used to assemble pRF1, pRF2, pRF3, and pRF1-2 are indicated below the diagram. Modified according to reference 18. (B) Lethality of treatment with different oxidants to cultures of E. coli MG1655 expressing the LHR or specific fragments of the LHR that are carried on pRF1, pRF2, pRF3, or pRF1-2. Treatment lethality is expressed as the log-transformed ratio of cell counts before treatment (N0) to cell counts after treatment (N). Cells were treated with 32 mM NaClO, 120 mM H2O2, 105 mg/liter peroxyacetic acid (PAA), 5.80 mM KIO4, 10 mM acrolein, or 75 μg/ml allyl isothiocyanate (AITC) for 5 min. Reactions were terminated by adding an equivalent volume of 10% Na2S2O3 as a reducing agent. Values for different plasmids within a treatment that do not have a common superscript are significantly different (P < 0.05). Data are shown as means ± standard deviations for three independent experiments.
The LHR prevents the oxidation of multiple cellular targets.The oxidation of cytoplasmic proteins and membrane lipids was assessed in E. coli MG1655 transformed with the complete LHR or LHR fragments. Oxidation of cytoplasmic proteins was assessed with the probe roGFP2_Orp1 and ratiometric fluorescence spectroscopy (27). Oxidation of the probe enhances green fluorescence when excited at 488 nm but not at 405 nm. For the nonoxidized probe, the ratio of the fluorescence intensity at an excitation wavelength of 405 nm to that at an excitation wavelength of 488 nm was equal to 1. Oxidation of the probe increases the green fluorescence and decreases the 405/488-nm fluorescence intensity ratio. In comparison to that for untreated E. coli MG1655, a decreased fluorescence ratio of roGFP2_Orp1 upon treatment of E. coli MG1655 with NaClO, H2O2, or PAA indicated that the probe was oxidized (Fig. 4). Insertion of the full LHR sequence reduced the oxidation of the probe, as indicated by a higher 405/488-nm fluorescence ratio (Fig. 4). pRF1 prevented probe oxidation as effectively as the full-length LHR, while pRF2 or pRF3 had a fluorescence ratio similar to that of the empty vector, pRK767, indicating that pRF2 and pRF3 had little or no effect on probe oxidation. The protective effects of the LHR against the oxidation of cytoplasmic proteins can thus be attributed to the heat shock proteins encoded by fragment 1 on pRF1.
Oxidation of roGFP2-based probes expressed in E. coli MG1655 with different plasmids (pRK767, pLHR, pRF1, pRF2, pRF3, pRF1-2) after exposure to different oxidants. The ratio of the fluorescence intensity at an excitation wavelength of 405 nm to the fluorescence intensity at an excitation wavelength of 488 nm was calculated to indicate the oxidation level in the cytoplasm. Cells either were left untreated (control) or were treated for 5 min with 32 mM NaClO, 120 mM H2O2, or 105 mg/liter peroxyacetic acid. Values for different plasmids within a treatment that do not have a common superscript are significantly different (P < 0.05). Data are means ± standard deviations for three independent experiments.
The oxidation of membrane lipids was determined with the fluorescent probe C11-BODIPY581/591, which is sensitive to lipid peroxides in membranes (28). The presence of the LHR consistently decreased the population of oxidized cells after treatment of E. coli MG1655 or AW1.7 with oxidizing chemicals; a corresponding increase in the population of unoxidized cells was observed after H2O2 or PAA treatment (Fig. 5; see also Table S2 in the supplemental material). A consistent effect of fragments of the LHR was observed only after treatment with H2O2; the presence of LHR fragment pRF3 or pRF1-2 resulted in a high number of unoxidized cells and a correspondingly reduced number of oxidized cells following treatment with H2O2.
Flow cytometric quantification of the oxidation of membrane lipids in E. coli MG1655 with different plasmids (pRK767, pLHR, pRF1, pRF2, pRF3, pRF1-2) by use of C11-BODIPY581/591 after different treatments. Shown are the levels of stained, unoxidized cells (A) or stained, oxidized cells (B) as percentages of the total cell population. Cells were treated for 5 min with 32 mM NaClO, 120 mM H2O2, or 105 mg/liter peroxyacetic acid. Values for different plasmids within a treatment that do not have a common superscript are significantly different (P < 0.05). Data are means ± standard deviations for 10 independent experiments.
The LHR reduces the peroxide-induced induction of the Stx prophage in E. coli O104:H4.The expression of the late genes in the Shiga toxin prophage in E. coli O104:H4 is induced by oxidative stress (29). Quantification of green fluorescent protein (GFP) fluorescence in the reporter strain E. coli O104:H4 Δstx2::gfp::amp is thus an indirect indication of cytoplasmic oxidative stress (29). The LHR or LHR fragments were cloned into E. coli O104:H4 Δstx2::gfp::amp in order to determine the effect of the LHR on the expression of the Shiga toxin prophage. H2O2, NaClO, and PAA induced GFP expression in E. coli O104:H4 Δstx2::gfp::amp (Fig. 6). Cloning of pLHR, pRF1, or pRF1-2 reduced the expression of the prophage after treatment with any of the three oxidizing chemicals but not after treatment with the positive control mitomycin C (MMC). Cloning of pRF3 reduced prophage induction after treatment with mitomycin C but not after treatment with NaClO or PAA; a modest reduction in the percentage of GFP-expressing cells was observed after treatment with H2O2 (Fig. 6). The LHR thus reduces the level of oxidative-stress-induced prophage expression, and this effect is predominantly attributable to fragment 1 on pRF1.
Quantification of stx expression in the reporter strain E. coli O104:H4 Δstx2::gfp::amp (FUA1302) with different plasmids (pRK767, pLHR, pRF1, pRF2, pRF3, pRF1-2) after exposure to different inducers. Exponential-phase E. coli O104:H4 Δstx2::gfp::amp was incubated at 37°C with the addition of either mitomycin C (MMC; 0.5 mg/liter) for 3 h, NaClO (10 mM) for 1 h, H2O2 (2.5 mM) for 1 h, or peroxyacetic acid (PAA; 50 mg/liter) for 1 h. GFP fluorescence was quantified by flow cytometry. Values for different plasmids within each treatment that do not have a common superscript are significantly different (P < 0.05). Data are means ± standard deviations for at least three independent experiments.
The inhibition of prophage induction by the LHR was confirmed by quantification of gfp expression by reverse transcription-quantitative PCR (RT-qPCR). The expression of gfp after induction by mitomycin C or H2O2 relative to that for uninduced controls was quantified in E. coli O104:H4 Δstx2::gfp::amp(pLHR) and O104:H4 Δstx2::gfp::amp(pRK767). The LHR reduced the expression of gfp after H2O2 treatment but not after mitomycin C treatment (Fig. 7A), a finding consistent with the data obtained by flow cytometry (Fig. 6). To determine whether the effect of the LHR on prophage induction relates to the RecA-dependent SOS response (29), the expression of recA was also quantified. The LHR did not affect the expression of recA after induction with mitomycin C or H2O2 (Fig. 7B).
Expression of gfp (A) and recA (B) in E. coli O104:H4 Δstx2::gfp::amp after mitomycin C or H2O2 treatment. Relative gene expression was quantified by RT-qPCR with gapA as the housekeeping gene and untreated exponential-phase cultures as reference conditions. Exponential-phase cultures were treated with LB broth containing mitomycin C (MMC; 0.5 mg/liter) or with H2O2 (2.5 mM) for 40 min. Values for different plasmids within a treatment highlighted with an asterisk are significantly different (P < 0.05). Data are means ± standard errors for three independent experiments.
The LHR does not interfere with the induction of some but not all of the Stx prophages in E. coli.The influence of the LHR on the expression of the Shiga toxin prophage in E. coli O104:H4 implies that the genetic island may interfere with the conversion of the prophage to the lytic cycle. The low prevalence of the LHR observed in STEC may indicate that the presence of the LHR selects against Shiga toxin prophages, or vice versa. About 2% of E. coli strains harbor the LHR (17); however, only 0.5% of 615 clinical isolates of STEC and/or E. coli O157 have been reported to harbor the LHR (30). Screening of 100 STEC strains revealed that none of them carried the LHR or parts of the LHR and that none of them exhibited a level of heat resistance equivalent to that of LHR-positive strains (31) (see Table S3 in the supplemental material). The sequence diversity of the late promoter pR′ region of Stx prophages affects the efficiency of Stx expression in different strains of STEC (32). Therefore, the effects of the LHR on gene expression from different pR′ regions in native STEC and in heterologous hosts were compared. To investigate if the LHR inhibits the expression of the same pR′ region in different strains, pLHR and pRK767 were introduced into E. coli O157:H7 CO6CE900 (FUA1399), E. coli O157:H7 1935 (FUA1303), and E. coli O45:H2 05-6545 (FUA1311). These strains were additionally transformed with Pp1302::rfp::chl, which carries the rfp gene, coding for red fluorescent protein (RFP), under the control of the late promoter of the E. coli O104:H4 11-3088 prophage (see Tables 3 and 4) (32). Prophage expression was induced with H2O2, and RFP expression was quantified by flow cytometry. The presence of the LHR resulted in a reduced proportion of cells expressing RFP in all three strains, but the extent of inhibition differed among strains (Fig. 8). The effect of the LHR on the expression of the pR′ region was further evaluated with RFP under the control of promoter pR′ regions derived from E. coli FUA1399, FUA1303, and FUA1311. Reporter plasmids were cloned into homologous and heterologous hosts. The presence of the LHR decreased the proportion of E. coli FUA1399 cells that expressed RFP from promoters p1399-28 and p1399-79, but the proportion of cells that expressed RFP remained unchanged with p1303-s1 and increased with p1303-2a in E. coli FUA1303 and with p1311 in E. coli FUA1311. These results demonstrate that the induction of the Shiga toxin promoter in the presence of the LHR was diverse, due to the sequence diversity of the pR′ region and prophage-encoded regulatory proteins in different strains.
Effect of the LHR on the expression of RFP under the control of pR′ promoters derived from Shiga toxin-producing E. coli. Data are the percentages of cells expressing RFP after induction with H2O2. RFP fluorescence was quantified by flow cytometry. The graph compares STEC strains carrying pLHR with STEC strains carrying pRK767 as a control. Promoter activity was assessed with pR′::rfp::chl constructs cloned into three strains of STEC. Plasmid pR′::rfp::chl, containing the pR′ promoter derived from E. coli O104:H7 (FUA1302), was cloned into all strains. In addition, each strain was transformed with a plasmid harboring a fusion of pR′::rfp::chl with the pR′ promoter(s) present in that strain, i.e., p1399-28 and p1399-79 in E. coli O157:H7 CO6CE900 (FUA1399), p1303-s1 and p1303-2a in E. coli O157:H7 1935 (FUA1303), and p1311 in E. coli O45:H2 05-6545 (FUA1311). Significant differences (P < 0.05) in promoter activity between strains with pLHR and strains with pRK767 are indicated by asterisks. Data are means ± standard deviations for three independent experiments.
The LHR is correlated with the absence of most UPEC virulence factors.Seventy E. coli wastewater isolates were screened for the LHR, a wastewater marker (IS30), and five virulence factors of UPEC to identify LHR-positive strains of UPEC (Table 1). Among the 70 isolates, 22.9% (16/70) were LHR positive, and 40% (28/70) carried virulence factors that are typical for UPEC. Two of 16 LHR-positive strains carried fyuA, which encodes the yersiniabactin receptor (33). The other 14 LHR-positive strains excluded UPEC virulence factors. Conversely, 26 of the 28 strains with UPEC virulence factors, and in particular all strains with multiple virulence factors, were LHR negative (Table 1). Both IS30-positive strains carried the LHR, matching prior results (9).
Screening of 70 E. coli wastewater strains for the locus of heat resistance, IS30, and virulence genesa
DISCUSSION
LHR-positive strains of E. coli have been observed at increased frequency in wastewater after chlorination, pointing to a contribution of the LHR to chlorine resistance (9). This study documented, with multiple and complementary experimental methods (27, 28, 34), that the LHR protects against oxidative stress. The use of multiple methods to quantify the survival and oxidation of cellular components not only confirmed the protective effect of the LHR but also provided information on the mechanisms of protection.
The LHR and stress resistance in E. coli.The LHR protected against NaClO, H2O2, and peroxyacetic acid but not against KIO4, acrolein, or AITC. Acrolein is generated in vivo by the chemical conversion of β-hydroxypropionaldehyde, a metabolite of intestinal bacteria including Salmonella enterica and Lactobacillus reuteri (35, 36). AITC is formed from glucosinolates in plants of the Brassicaceae family upon cellular injury and reacts with sulfhydryl groups and disulfide bonds (37). KIO4 has been suggested to contribute to electrochemical inactivation of E. coli, but the mechanisms of activity remain unclear (38). The selective protective effect of the LHR against oxidizing organic and inorganic chemicals may relate to differences in the chemical reactivity and permeance of the compounds for the cytoplasmic targets. The presence of the LHR consistently increased resistance to chlorine and peroxides, which are a mainstay in the sanitation of water and in food-processing plants; therefore, further experiments focused on these chemicals.
The LHR encodes proteins that are involved in protein folding or disaggregation and may protect against oxidative stress (Fig. 1) (17). The heat shock proteins encoded by the protein homeostasis module (fragment 1), sHSP20, ClpKGI, and sHSPGI, prevent protein aggregation or disaggregate proteins (18, 39). In agreement with the function of these proteins, transformation of E. coli with this portion of the LHR provided the greatest protection of cytoplasmic proteins from oxidation but was less protective of membrane lipids. LHR fragment 3 did not protect against chlorine or the oxidation of cytoplasmic components but decreased the oxidation of membrane lipids (Fig. 3 to 5). The role of the oxidation of membrane lipids in chlorine resistance is poorly documented (Fig. 1). The oxidative stress module (fragment 3) encodes thioredoxin, KefB, and DegP. Thioredoxin protects against oxidative stress through thiol-disulfide exchange reactions (40). KefB may have a function related to oxidative stress; it is a K+/H+ antiporter that protects against two electrophiles, methylglyoxal and N-ethylmaleimide (41). The serine protease DegP is a periplasmic chaperone that prevents or reduces the iron-induced oxidation of membrane proteins (42). Fragment 2 encodes YfdX1GI and YfdX2, two hypothetical proteins of unknown function, and HdeDGI, a stress protein whose relation to oxidative stress is unknown (43). These genes protected against chlorine and provided limited protection against the oxidation of membrane lipids but did not prevent the oxidation of cytoplasmic proteins. These findings not only confirm the previous designation of fragment 2 as an “envelope stress module” (18) but also suggest additional, undescribed mechanisms of protection. Taking these findings together, the contribution of LHR to chlorine resistance is based on the protection of multiple cellular targets by different parts of the LHR. The multifaceted mechanisms of protection provided by the LHR also explain its contribution to resistance to multiple stressors, including heat, chlorine, and peroxides (17, 18, 39, 44).
Two distinct LHR variants in E. coli confer equivalent heat resistance (45). LHR2 is a 19-kb genomic island that contains five open reading frames (ORFs) that are not carried by the LHR but lacks two proteins encoded by the LHR (17, 45). Proteins that are present in both versions of the genomic island include sHSP20, ClpKGI, sHSPGI, YfdX1GI, YfdX2, KefB, orf15, and DegP; these proteins appear to be essential for the role of the genomic island in bacterial stress resistance.
The LHR and the virulence of E. coli.The LHR is a mobile genetic element that occurs in diverse members of the Gammaproteobacteria and Betaproteobacteria. The LHR has been identified in one Salmonella strain and in several isolates of Cronobacter spp. and Klebsiella spp. (45, 46). The frequency of the LHR in STEC is very low compared to that in the general population of E. coli (30; this study). Genes coding for Shiga toxin production are invariably found in the late regions of prophages. Shiga toxin production by E. coli has been proposed to protect against predatory protozoa, which are significant E. coli predators in the rumen but are less prevalent in other ecosystems that are relevant for the evolution of E. coli (47). Expression of the Shiga toxin is upregulated in the lytic cycle of the phage, which is lethal to the host cell, but it has been hypothesized that Shiga toxin production by engulfed cells will kill the predator, reducing predation on the remainder of the population. Accordingly, STEC strains are highly associated with diverse ruminant hosts (48). One explanation of the low overlap of the LHR and Shiga toxin production in E. coli is ecological incompatibility; i.e., ecosystems that select for maintenance of the Shiga toxin select against the LHR, and vice versa. Molecular incompatibility is an alternative explanation; i.e., the LHR interferes with the expression of Shiga toxin prophages and hence reduces or abolishes the ecological advantage conferred by lysogeny.
The lytic cycle of the Shiga toxin prophage is regulated by the DNA repair protein RecA and is induced in response to DNA damage and/or oxidative stress. If the LHR protects against oxidative stress, it also reduces Shiga toxin production in response to predation by protozoa and may reduce the selective pressure to maintain the Shiga toxin prophages in STEC. Reduced expression of GFP as an indicator of Stx expression was observed in E. coli O104:H4 but not for all Shiga toxin prophages in other STEC strains. Different Shiga toxin prophages respond differently to inducing agents (32, 49). In this study, it was observed that all the prophage promoters investigated responded to H2O2, but not all responded to mitomycin C, which is routinely used to induce lambdoid prophages, including Shiga toxin prophages. In summary, the LHR interfered with the induction of some, but not all, Shiga toxin prophages. Thus, it does not appear that the LHR is generally incompatible with the contribution of Shiga toxin to the ecological fitness of E. coli in response to protozoan predation. The assumption that the LHR increases fitness in ecological niches in which Shiga toxin prophages do not provide a selective advantage is a more likely explanation for the low cooccurrence of these two mobile genetic elements in E. coli.
Naturalized strains of E. coli found in wastewater possess the LHR, are resistant to chlorine, and carry as many as 44 different virulence genes associated with UPEC (9, 50). UPEC strains appear to differentially survive wastewater treatment processes, including chlorination and UV-C irradiation (24, 25, 51). Interestingly, the proportion of antibiotic-resistant UPEC cells has been reported to be higher in water after treatment with chlorine (52, 53). Collectively, these studies led us to hypothesize that antibiotic resistance correlates with the increased likelihood of UPEC surviving wastewater treatment (chlorination and oxidation). LHR-positive E. coli and UPEC strains were found to coexist in wastewater, and the prevalence of the LHR in wastewater isolates of E. coli, 23%, is about 10-fold higher than that in the general population of E. coli, 2%. Surprisingly, this study found low cooccurrence of the LHR and UPEC virulence factors in 70 wastewater isolates. This finding should be confirmed by screening a larger number of strains or genomes, but it suggests that the LHR may not account for the resiliency of UPEC in surviving wastewater treatment processes and attests to the diverse mechanisms by which microbes evolve resistance to wastewater treatment processes.
In conclusion, the LHR confers resistance to heat and oxidative stress on E. coli. The protein homeostasis module provided protection to the contents of the cytoplasm, while the oxidative stress module provided protection to membrane lipids. In addition, the reduction effect of the LHR on Shiga toxin expression was specific to the late promoter regions and regulatory proteins. The low prevalence of the LHR in STEC and UPEC strains indicates that the selective pressure for maintenance of the LHR in E. coli is different from the selective pressure that maintains Shiga toxin prophages and UPEC virulence factors. This study also shows that water chlorination selects for LHR-positive E. coli strains that are heat and chlorine resistant.
MATERIALS AND METHODS
Collection and screening of E. coli from wastewater and groundwater.Ten LHR-positive strains and 10 LHR-negative strains of E. coli were selected from a total of 70 strains that had been isolated from wastewater previously (9). In addition, 70 E. coli isolates collected from wastewater effluents during routine monitoring programs were provided by a municipal water treatment plant in Alberta, Canada. These samples were collected from undigested sludge, digested sludge, and biosolids from a lagoon prior to agricultural land application. Strains of E. coli were subcultured from most-probable-number tubes in EC broth (Oxoid) after 24 h of incubation. The methods used to confirm the identification of E. coli and to detect specific virulence or resistance markers are outlined below. Sixty-five groundwater isolates positive for stx1, stx2, or both were collected from routine screening of well water samples submitted to the Alberta Provincial Laboratory for Public Health. Isolates were screened for the presence of the LHR with primers targeting three fragments of the LHR in E. coli AW1.7. The primers used in this study are listed in Table 2; a list of isolates used in the study is provided in Table S1 in the supplemental material.
PCR primers for amplification used in this study and melting temperatures for multiplex HRM analysis-qPCR detection of target genes
Determination of chlorine resistance.Chlorine resistance was determined by mixing 200 μl of overnight cultures with 6.5 μl of a 3% (wt/wt) sodium hypochlorite solution (Sigma-Aldrich, St. Louis, MO) to a concentration of 15 mM NaClO, followed by incubation for 25 min at 20°C. The reaction was terminated by the addition of 6.5 μl of 10% Na2S2O3 (Sigma-Aldrich, St. Louis, MO). Treatment conditions were selected to achieve a reduction in cell counts ranging from about 1 to 7 log CFU/ml. To benchmark the effect of the LHR on chlorine resistance against the previously described contribution of the LHR to heat resistance, the heat resistance of overnight cultures of E. coli was determined as described previously (17). Cell counts of cultures before and after treatment were determined by surface plating on LB agar and incubation at 37°C for 24 h. Results are expressed as log-transformed ratios of cell counts before treatment to cell counts after treatment [log(N0/N)].
Effect of the LHR on resistance to oxidizing chemicals.To assess the contributions of different regions of the LHR to survival under oxidative stress, E. coli MG1655 was transformed with plasmids pRK767, pLHR, pRF1, pRF2, pRF3, and pRF1-2 (17) (Tables 3 and 4). The pRK767 plasmid was used as a vector control for MG1655. Oxidative stress was induced by treatment of the transformants for 5 min with 32 mM NaClO, 120 mM H2O2 (30% [wt/wt] in H2O; Sigma-Aldrich, St. Louis, MO), 105 mg/liter peroxyacetic acid (PAA; 32% [wt/wt] in acetic acid; Sigma-Aldrich, St. Louis, MO), 5.80 mM KIO4 (Thermo Fisher Scientific, Waltham, MA, USA), 10 mM acrolein (Thermo Fisher Scientific, Waltham, MA, USA), or 75 mg/liter allyl isothiocyanate (AITC; Alfa Aesar Co., Inc.). Reactions were terminated by adding 10% Na2S2O3 to achieve a final concentration of 16 to 63 mM. The cell counts were determined as described above.
Plasmids used in this study and antibiotics used for plasmid maintenance
Strains used in this study
Measurement of cytoplasmic oxidation by a roGFP2-based probe.The fusion protein roGFP2-Orp1 was designed to measure H2O2 in biological systems (27). A plasmid encoding roGFP2-Orp1 was transformed into E. coli MG1655, along with plasmids carrying the whole LHR or part of the LHR (pRK767, pLHR, pRF1, pRF2, pRF3, pRF1-2) (Tables 3 and 4). Ampicillin (100 mg/liter) and tetracycline (15 mg/liter) were added to the cultivation media to maintain both plasmids. Exponential-phase cultures of transformants were incubated with 100 μM IPTG (isopropyl β-d-1-thiogalactopyranoside; Thermo Fisher Scientific, USA) at 37°C overnight to induce the expression of reduction-oxidation-sensitive green fluorescent protein 2 (roGFP2). Cells were washed twice in phosphate-buffered saline (PBS) buffer (pH 7.4) and were treated with NaClO (32 mM), H2O2 (120 mM), or PAA (105 mg/liter) for 5 min. Nontreated cells served as controls. Reactions were terminated by adding 10% Na2S2O3. Cultures (100 μl) were placed in the wells of a black, clear-bottom 96-well plate (Corning; Corning, NY, USA), and fluorescence was measured at excitation wavelengths of 405 and 488 nm and at the emission wavelength of 530 nm. The ratio of the fluorescence intensity obtained at the excitation wavelength of 405 nm to that obtained at 488 nm was used to evaluate the oxidation level of roGFP2 (27).
Determination of membrane lipid oxidation by C11-BODIPY581/591.E. coli MG1655 transformed with different plasmids (pRK767, pLHR, pRF1, pRF2, pRF3, pRF1-2) (Tables 3 and 4) was treated with NaClO (32 mM), H2O2 (120 mM), or PAA (105 mg/liter) for 5 min. Oxidized E. coli served as a nonstained control. Nonoxidized E. coli prepared with C11-BODIPY581/591 (Thermo Fisher Scientific, Waltham, MA, USA) served as a nontreatment control. In brief, 1.5 ml of each culture was washed with 2 ml ice-cold 50 mM Tris-HCl (pH 8.0) containing 20% (wt/vol) sucrose and was resuspended in 2 ml Tris-HCl buffer. The outer membrane was disrupted by the addition of 0.2 ml lysozyme solution (5 mg/ml lysozyme in 0.25 M Tris-HCl [pH 8.0]) and 0.4 ml EDTA (0.25 M; pH 8.0) (54). After incubation at 37°C for 30 min, cell pellets were suspended with 10 mM citrate buffer (pH 7.0) with the addition of 10 μM C11-BODIPY581/591, followed by incubation in the dark for 30 min at 37°C and 200 rpm. Flow cytometry was performed using a BD LSRFortessa X-20 system (BD Biosciences, San Jose, CA, USA) equipped with 488-nm excitation from a blue air laser at 50 mW and 561-nm excitation from a yellow air laser at 50 mW to excite green (530 ± 30 nm) and red (586 ± 15 nm) fluorescence. Single cells were quantified by forward scatter and side scatter gating on the flow cytometer. In brief, samples were centrifuged, resuspended, and diluted with 1 ml of PBS (pH 7.4) to keep the detected cell number per second (e/s) in the range of 300 to ∼3,000 events. Sample injection and acquisition were started simultaneously and continued until 10,000 events were recorded. Data were recorded by BD FACSDiva software and were analyzed by FlowJo (both from BD Biosciences, San Jose, CA, USA). The single-cell population was defined by selecting the cell population located along the diagonal of the “FSC-A; FSC-H” dot plot. The population was divided into four subpopulations by red and green fluorescence reference lines. The reference lines were determined from untreated samples, where at least 96% of the population was negative for red and green fluorescence.
Flow cytometric determination of GFP fluorescence.E. coli O104:H4 Δstx2::gfp::amp is derived from a STEC strain, with in-frame replacement of the prophage-harbored stx2 gene with a gfp::amp cassette (29). The fusion of GFP in the Stx2 prophage provides a reporter for protein expression under the control of the Shiga toxin promoter. E. coli O104:H4 is tetracycline resistant, a characteristic that interferes with the antibiotic resistance of the other plasmids used in this study. Therefore, the tetracycline resistance gene on plasmids pRK767, pLHR, pRF1, pRF2, pRF3, and pRF1-2 was replaced with a chloramphenicol resistance gene from pKD3 (55). PCRs were carried out using Phusion High-Fidelity DNA polymerase (Thermo Scientific) according to the manufacturer’s guidelines. The gene encoding chloramphenicol resistance was amplified by the Priming site 1 and Priming site 2 primers (Table 2). Plasmids pRK767, pLHR, pRF1, pRF2, pRF3, and pRF1-2 were digested with HindIII to remove the tetracycline resistance gene, and the chloramphenicol resistance gene was then ligated into the plasmid as a HindIII/HindIII insert. The direction of the insert was confirmed by amplification with the primers targeting Priming site 1 and M13/pUC-F (Table 2). The recombinant plasmids were electroporated into E. coli O104:H4 Δstx2::gfp::amp (Tables 3 and 4). Exponential-phase cultures of these transformants were treated with mitomycin C (MMC; 0.5 mg/liter) for 3 h or with NaClO (10 mM), H2O2 (2.5 mM), or PAA (50 mg/liter) for 1 h. Control samples were incubated in the same manner without stressors. The method used for the detection of the population of fluorescent cells was similar to that described in reference 29. Samples were divided into two subpopulations, and the percentage values of GFP-positive cells were calculated.
Quantification of stx2 prophage expression in LHR-positive and LHR-negative strains in response to oxidative stress.Exponential-phase cultures of E. coli O104:H4 Δstx2::gfp::amp(pRK767) and E. coli O104:H4 Δstx2::gfp::amp(pLHR) were centrifuged and were resuspended in LB broth containing mitomycin C (0.5 mg/liter) or H2O2 (2.5 mM), followed by incubation at 37°C for 1 h. Corresponding treatment of cultures without any addition served as a control. After treatment, cells were harvested from samples; RNA was isolated using the RNAprotect Bacteria reagent and the RNeasy minikit (Qiagen) and was reverse transcribed to cDNA with a QuantiTect reverse transcription kit (Qiagen) according to the manufacturer’s protocols. The expression of gfp and recA was quantified by using SYBR green reagent (Qiagen) and a 7500 Fast PCR system (Applied Biosystems, Foster City, CA, USA). Negative controls included DNase-treated RNA and nontemplate controls. The gene coding for glyceraldehyde-3-phosphate dehydrogenase A (gapA) served as the reference gene. The ratios of expression of the gfp and recA genes in E. coli O104:H4 Δstx2::gfp::amp(pRK767) to their expression in E. coli O104:H4 Δstx2::gfp::amp(pLHR) under induced and control conditions were calculated according to the method of Pfaffl (57). The primers used for the quantification of gene expression are listed in Table 2.
Determination of the effects of the LHR on prophage induction in different STEC strains.The construction of the pR′::rfp::chl reporter system has been described previously (32). To measure the effect of the LHR on the induction of Shiga toxin prophages with different promoters, the tetracycline resistance gene in plasmids pLHR and pRK767 was replaced with the kanamycin resistance gene derived from pKD4 (55) as described above, and the resulting plasmids were electroporated into E. coli FUA1303, E. coli FUA1311, and E. coli FUA1399. Strains were additionally transformed with pR′::rfp::chl plasmids to obtain transformants where red fluorescent protein (RFP) expression is controlled by the native phage promoter that also controls the expression of the chromosomally encoded prophage, or to obtain RFP expression by alternate phage promoters that are not encoded on the chromosome of the host (Tables 3 and 4). The expression of rfp was measured by flow cytometry as described elsewhere (32).
Detection of the uspC-IS30-flhDC marker and virulence genes in E. coli wastewater isolates and detection of the LHR in STEC.PCR screening of 102 STEC strains for the three LHR fragments was performed using multiplex PCR with the same primers and protocol that were used for the wastewater and groundwater strains (Table 2).
Seventy E. coli wastewater isolates were screened for the presence of the uspC-IS30-flhDC marker (9, 50), as an indicator for a naturalized global lineage of wastewater isolates, and for five virulence genes associated with UPEC (papC, iroN, fyuA, ibeA, and sfa/foc). The genomic DNA of E. coli strains was extracted from bacterial cultures by using a DNeasy blood and tissue kit (Qiagen, Hilden, Germany) according to the manufacturer’s instructions. Lists of genes and primers are provided in Table 2. High-resolution melting (HRM) analysis-qPCR was conducted on a Rotor-Gene Q (Qiagen) system using a Type-it HRM PCR kit (Qiagen) (58) to detect the target genes listed in Table 2 with group-specific primers according to the manufacturer’s protocols. Based on the primer annealing temperature (TA), three pairs of primers with the same TA but different melting temperatures (Tm) of the amplicons were combined in the multiplex PCRs. The melting temperatures of PCR products are presented in Table 2. The PCR was optimized with the following conditions: initial denaturation at 95°C for 5 min, followed by 45 cycles of denaturation at 95°C for 10 s, annealing at the TA for 30 s, and extension at 72°C for 30 s. During the HRM analysis stage, the temperature was increased from 65°C to 95°C at the speed of 0.1°C per step and was held for 2 s at each step. Results are presented as means ± standard deviations (SD) for three biological replicates.
Statistical analysis.Data were obtained in 10 biological replicates (for the oxidation of membrane lipids) or 3 biological replicates (for all other assays) and are expressed as means ± standard deviations. Data were analyzed by one-way analysis of variance (ANOVA) using SPSS software, version 21.0 (SPSS Inc., Chicago, IL, USA). The least significant difference (LSD) was used to test the difference among means using a P value of <0.05.
ACKNOWLEDGMENTS
We are grateful to Lars Leichert for providing plasmids with the roGFP2-based probes.
We acknowledge Canada Research Chairs, Alberta Agriculture and Forestry, the Natural Sciences and Engineering Research Council of Canada, Alberta Innovates, and the China Scholarship Council for funding.
FOOTNOTES
- Received 16 September 2019.
- Accepted 28 November 2019.
- Accepted manuscript posted online 6 December 2019.
Supplemental material is available online only.
- Copyright © 2020 American Society for Microbiology.
