The biofilm life style helps bacteria resist oxidative stress, desiccation, antibiotic treatment, and starvation. Biofilm formation involves a complex regulatory gene network controlled by various environmental signals. It was previously shown that prophage insertions in mlrA and heterogeneous mutations in rpoS constituted major obstacles limiting biofilm formation and the expression of extracellular curli fibers in strains of Escherichia coli serotype O157:H7. The purpose of this study was to test strains from other important serotypes of Shiga toxin-producing E. coli (STEC) (O26, O45, O103, O111, O113, O121, and O145) for similar regulatory restrictions. In a small but diverse collection of biofilm-forming and non-forming strains, mlrA prophage insertions were identified in only 4 of the 19 strains (serotypes O103, O113, and O145). Only the STEC O103 and O113 strains could be complemented by a trans-copy of mlrA to restore curli production and Congo red (CR) dye affinity. RpoS mutations were found in 5 strains (4 serotypes), each with low CR affinity, and the defects were moderately restored by a wild-type copy of rpoS in 2 of the 3 strains attempted. Fourteen strains in this study showed no or weak biofilm formation, of which 9 could be explained by prophage insertions or rpoS mutations. However, each of the remaining five biofilm-deficient strains, as well as the two O145 strains that could not be complemented by mlrA, showed complete or nearly complete lack of motility. This study indicates that mlrA prophage insertions and rpoS mutations do limit biofilm and curli expression in the non-serotype O157:H7 STEC but prophage insertions may not be as common as in serotype O157:H7 strains. The results also suggest that lack of motility provides a third major factor limiting biofilm formation in the non-O157:H7 STEC. Understanding biofilm regulatory mechanisms will prove beneficial in reducing pathogen survival and enhancing food safety.
Citation: Chen C-Y, Hofmann CS, Cottrell BJ, Strobaugh Jr TP, Paoli GC, Nguyen L-H, et al. (2013) Phenotypic and Genotypic Characterization of Biofilm Forming Capabilities in Non-O157 Shiga Toxin-Producing Escherichia coli Strains. PLoS ONE 8(12): e84863. https://doi.org/10.1371/journal.pone.0084863
Editor: Mark Alexander Webber, University of Birmingham, United Kingdom
Received: September 20, 2013; Accepted: November 20, 2013; Published: December 27, 2013
This is an open-access article, free of all copyright, and may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for any lawful purpose. The work is made available under the Creative Commons CC0 public domain dedication.
Funding: This research was funded by the National Program 108 Food Safety Project of the United States Department of Agriculture-Agriculture Research Service (USDA-ARS). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Shiga toxin-producing Escherichia coli (STEC) has emerged in the last three decades to become one of the most important causes of food-associated illnesses in the United States and abroad [1-3]. It has been estimated that serotype O157:H7 is isolated from about one-half of the cases of STEC-induced diarrheal diseases in North America, while non-O157:H7 serotypes comprise the remaining isolates . Of the non-O157:H7 STEC submitted from state public health laboratories for confirmation by the Centers for Disease Control and Prevention (CDC), six O-serogroups (O26, O45, O103, O111, O121, O145) accounted for 71% of all isolates in the period from 1983-2002 . In recognition of the safety concerns and the higher incidence associated with these 6 O-serogroups, the United States Department of Agriculture, Food Safety Inspection Service (FSIS) declared STEC serogroups O26, O45, O103, O111, O121, and O145 to be adulterants in raw non-intact beef products and product components in 2011 and initiated a testing program for these 6 STEC in beef trimmings in June 2012 . Recent studies have identified O113:H21 as an additional important STEC serotype, being the most frequent serotype isolated from ground beef samples  and belonging to the same seropathotype as serotype O121:NM in the classification scheme conceived by Karmali et al. .
The importance of biofilms in human disease pathogenesis and in the food industry has been well established [7,8]. In E. coli, biofilm formation in the majority of strains is dependent on the expression of the curli fimbriae (encoded by csgBA) and synthesis of polysaccharides such as cellulose . The expression of curli and cellulose was found to be highly variable at 28°C and 37°C in a study of commensal E. coli isolated from the human gastrointestinal tract . Expression of both curli and cellulose is dependent on the DNA-binding transcription factor CsgD . The csgD gene is expressed from a relatively weak promoter located in the intervening region between the divergently transcribed csgDEFG and csgBAC operons, and relies on a complex network of regulatory factors, both activators and suppressors, to coordinate transcription under a variety of environmental conditions [11-15]. CsgD expression in Salmonella enterica serovar Typhimurium and E. coli had been shown to be affected by environmental conditions such as temperature, pH, oxygen concentration, osmolarity, and various nutrients . Transcription of csgD predominately utilizes the stationary phase sigma factor RpoS, which activates ydaM and mlrA in a temporal fashion before cooperating with both to activate csgD transcription . Both MlrA and YdaM are essential for csgD expression [15,16].
Although there have been numerous studies focused on the mechanisms and regulation of biofilm formation in both non-pathogenic and pathogenic E. coli, there is a shortage of similar data on STEC strains, especially the clinically important non-O157:H7 serogroups [17,18]. In a previous study using 55 clinical isolates of E. coli serotype O157:H7, we determined that only 4 of the strains showed a weak affinity for the curli-binding dye, Congo red (CR), when cultured on CRI agar at 25°C . None of the strains could generate significant biofilm on polystyrene at 25°C when cultured in LB broth without salt (LB-NS) . Moreover, we determined that these deficiencies were the result of either a lack of mlrA expression due to prophage insertions in the proximal coding region of mlrA (53/55 strains), or attenuating or inactivating lesions in RpoS (40/55 strains). Approximately 70% (38/55) of the O157:H7 strains carried a prophage in mlrA and possessed an RpoS mutation. Apparently, if strains of serotype O157:H7 are to assume a curli and biofilm expressing phenotype, they must make regulatory adjustments or accumulate mutations that overcome these barriers. It is unclear if other serotypes of pathogenic E. coli are subject to the same regulatory constraints. In two surveys of non-O157:H7 STEC strains, prophage insertions in mlrA (yehV) were found at a lower frequency in the non-O157:H7 STEC strains than in serotype O157:H7 strains suggesting that disruption of mlrA may not be an important barrier to curli expression in the non-O157:H7 STEC serotypes [20,21].
In this study we assembled 19 curli-producing and non-producing strains from 7 important, non-O157:H7 O-serogroups to determine whether deficiencies in mlrA and rpoS impose major barriers to curli production and biofilm formation as in clinical isolates of serotype O157:H7.
Materials and Methods
Bacterial strains, growth conditions, plasmids and primers
Two or more strains from each of the seven E. coli O-serogroups O26, O45, O103, O111, O113, O121, and O145 were collected and used in this study (Table 1). The selection criterion was based on the varying degrees of biofilm-forming capabilities assessed by CR dye affinity following 48-h growth on Congo red indicator (CRI) agar at 25°C . Serotype O157:H7 strain ATCC 43895 was used as a negative control, and strain 43894OR, a constitutive curli-producing strain , was used as a positive control for curli isolation and for CR affinity. Overnight cultures were grown in Luria-Bertani (LB) broth at 37°C or 30°C unless otherwise specified. Plasmids used for complementation experiments (pUC19::mlrA and pTOPO::rpoS) were described previously . Complementation plasmids were introduced into all the STEC strains by electroporation using an E. coli Pulser (Bio-Rad, Hercules, CA).
DNA isolation, PCR and sequencing
DNA was isolated from 18-24 h cultures using the DNeasy Blood and Tissue Kit (Qiagen, Valencia, CA). Primers used in this study are shown in Table 2. O-serogroup and the status of the mlrA gene were investigated using previously described primers in a multiplex format [19,20,27]. PCR verification for stx1 and stx2 genes were performed using published primers . All PCR assays (25 µl) were set up using the Qiagen Multiplex PCR Master Mix (Qiagen), 0.5X Q-Solution, 1.5 µl template DNA, and primers at a final concentration of 0.5 µM each. Cycling conditions were as recommended by the manufacturer: 95°C for 15 min, followed by 35 cycles of 30 s at 94°C/ 90 s at 52°C for stx1-comF/R and GK5/GK6, or 57°C for all others/ 90 s at 72°C, and then 10 min at 72°C. The mlrA and rpoS genes, including flanking regions, were amplified using the primer pairs MlrA-F/MlrA-R and RpoS-F/RpoS-R, respectively. The csgBA genes and the intergenic region between csgB and csgD were amplified using CsgB-PF/CsgA-PR and CsgDup2/CsgDdown1, respectively. Amplified products were sequenced using custom primers , the Big Dye Terminator (v3.1) and ABI 3730 DNA Analyzer (Life Technologies) at our core facility. Sequences were assembled using Geneious (v5.6; Biomatters) or Sequencher (v5.1; Gene Code). Sequence annotation and analyses were performed using Geneious (v5.6). Sequences were deposited in NCBI GenBank under accession numbers KF662824-KF662843 for mlrA, KF662844-KF662863 for rpoS, KF662816-KF662823 for csgBA (8 strains), and KF680304-KF680323 for csgD promoter regions.
|Primer or plasmid||Sequence (5’ → 3’) or description||Reference||Purpose|
|A (yehVRJfor)||AAGTGGCGTTGCTTTGTGAT||||Multiplex mlrA PCR|
|B (yehVLJrev)||AACAGATGTGTGGTGAGTGTCTG||||Multiplex mlrA PCR|
|E (yehVRJrev)||GATGCACAATAGGCACTAGGC||||Multiplex mlrA PCR|
|F (yehVLJfor)||CACCGGAAGGACAATTCATC||||Multiplex mlrA PCR|
|O26F||CAATGGGCGGAAATTTTAGA||||O serotype PCR|
|O26R||ATAATTTTCTCTGCCGTCGC||||O serotype PCR|
|O45F||TGCAGTAACCTGCACGGGCG||||O serotype PCR|
|O45R||AGCAGGCACAACAGCCACTACT||||O serotype PCR|
|O103F||TTGGAGCGTTAACTGGACCT||||O serotype PCR|
|O103R||GCTCCCGAGCACGTATAAAG||||O serotype PCR|
|O111F||TGTTTCTTCGATGTTGCGAG||||O serotype PCR|
|O111R||GCAAGGGACATAAGAAGCCA||||O serotype PCR|
|O113F||TGCCATAATTCAGAGGGTGAC||||O serotype PCR|
|O113R||AACAAAGCTAATTGTGGCCG||||O serotype PCR|
|O121F||TCCAACAATTGGTCGTGAAA||||O serotype PCR|
|O121R||AGAAAGTGTGAAATGCCCGT||||O serotype PCR|
|O145F||TTCATTGTTTTGCTTGCTCG||||O serotype PCR|
|O145R||GGCAAGCTTTGGAAATGAAA||||O serotype PCR|
|CsgDup2||ACTACCTTCTTGCGCAACAACC||This study||csgD promoter activity|
|CsgDdown1||GAAGGATCCAGGATAAGCTTTTTATCCGCTTCC||This study||csgD promoter activity|
|CsgB-PF||TCCGCAGACATACTTTCCATCG||This study||csgBA amplification/sequence|
|CsgA-PR||ATTTGAAAGTGCGGCAAGGAGC||This study||csgBA amplification/sequence|
|CsgB-PR||AATTCGGGCCGCTGTTATTACC||This study||csgBA sequence|
|CsgA-PF||TGCAGAGACAGTCGCAAATGG||This study||csgBA sequence|
|pUC19::mlrA||mlrA + 224 bp upstream||||Trans-complementation|
|pTOPO::rpoS||rpoS + 804 bp upstream||||Trans-complementation|
Congo red dye affinity assay
CR binding was assessed by spotting 3 µl of overnight LB cultures on CRI plates and incubating for 48 h at 25°C. Plates were imaged using Epson Perfection 3200 Photo scanner and Epson Scan software (Professional mode) with a blue background at 600 dpi/48-bit color.
Motility was measured using the soft agar swimming assay modified from previously described . Briefly, bacterial strains were grown in LB-NS for 48 h at 25°C without shaking. Soft agar plates (1% tryptone, 0.5% NaCl and 0.3% agar) were inoculated by stabbing the center of the plate with a toothpick wetted with the broth cultures. The diameters of the halos grown at 25°C were measured after 24 h.
Strains were screened for catalase activity by placing a small loop-full (~1 µl) of bacteria collected from an 18-24-h colony onto a glass slide and exposing the bacteria to a drop of 30% H2O2. Results were graded by the robustness of bubbling, from “- (very little or no bubble formation)” to “+++ (immediate, strong bubble formation)”.
Biofilm (crystal violet-binding) assays
Starter cultures were grown in LB-NS at 37°C for 18 h. Cultures were diluted 1:100 in fresh LB-NS, and 100-µl samples of each strain were dispensed into 6 wells of a 96-well plate (TPP, Trasadingen, Switzerland). The plates with samples were incubated at 25°C for 48 h in a sealed humidity chamber to prevent loss of sample volume. Cultures were removed and the presence of biofilm/pellicles was visually inspected. The wells were then washed three times with water using a Nunc-Immuno Wash 8 plate washer (Nalge Nunc International). Biofilm was detected by addition of 200 µl of a 0.1% crystal violet (CV) solution per well for 30 min, followed by three water washes and de-staining with 200 µl of 95% ethanol for 30 min. The absorbance was read at 590 nm using Safire2 (TECAN).
Curli enrichment and isolation
Curli fibers, which are resistant to SDS solubilization, were isolated and solubilized in formic acid using the procedures described previously [10,19]. Samples were loaded on an Any kD TGX gel (Bio-Rad) and resolved by electrophoresis in Tris-Glycine-SDS buffer (25 mM Tris, 192 mM Glycine, 0.1% w/v SDS, pH 8.3) (Bio-Rad). CsgA subunits were visualized with Bio-Safe Coomassie Stain G250 (Bio-Rad).
csgD promoter activity
Six representative csgD promoter types (Table 1) were amplified from strains ATCC 43895, SJ13, 05-6545, 03-4064, SJ24, and E59 using primers CsgDup2 and CsgDdown1 and cloned into SmaI-BamHI digested pMLB1034 in E. coli EC100 strain to generate lacZ translational fusions. Cultures were grown in LB at 37°C and samples were collected at various time points and assayed for β-galactosidase activity . Experiments were repeated three times, and one representative time course was shown.
Our previous study with 55 E. coli O157:H7 isolates showed that MlrA and RpoS functionality played a key role in determining the biofilm-forming capabilities of strains in that serotype; the majority of the strains carried a stx1 prophage insertion in mlrA and/or an rpoS mutation that disrupted the curli regulatory pathway, thus diminishing their ability to form biofilms as evidenced by CR and CV binding. In this study, nineteen isolates of the 7 important non-O157 STEC O-serogroups O26, O45, O103, O111, O113, O121, and O145 were selected based on their varied affinities for CR dye (Figure 1). The O serogroup and the status of the stx1, stx2 and mlrA genes are shown in Tables 1 and 3, and Figure S1. Results of the motility assay and catalase tests are shown in Table 3. Strain SJ15, previously thought to be O111:NM, tested negative in the latex agglutination assays developed for the “Top 6” non-O157 serotypes . Molecular serotyping by PCR in this study showed that it belongs to serogroup O113, which was not included in the targeted O groups in the agglutination assays . The serogroup result was confirmed by conventional serotyping at the E. coli Reference Center (University Park, PA; Chobi DebRoy, personal communication). SJ15 was found to be motile, however, the H-type was not determined.
Strains carrying mutant mlrA or rpoS genes were marked with black boxes. AmpR/KanR: antibiotic resistance preventing the complementation experiments. WT: strain carries wild-type RpoS protein thus not complemented. NA: stable transformant not achievable.
|FCL1||-||WT (340bp)||+++||WT||- (1+/6)||1.12±0.67||R+++d||++||2.66±0.03||R+++d||++|
|DA-33||+||824bp, weak 702bp/340bp||+++||WT||-||0.13±0.02||W||-||1.61±0.09||R++(+)||++|
|SJ15||+||824bp, weak 340bp||+++||WT||- (1+/6)||0.16±0.07||W||-||0.53±0.10||R+++d||(+)|
|SJ29||+||WT||-||FS87T98 (Ins)||- (1+/6)||0.11±0.02||W||-||0.08±0.001||W||-|
|SJ16||+||WT||+||FS78T98 (Ins)||- (1+/6)||0.11±0.01||W||-||0.09±0.003||B/R+||-|
|SJ24||-||multiple bands||++||WT||- (1+/6)||0.11±0.02||W||-||0.89±0.09||W/P||-|
|43895||+||824bp/702bp, weak 340bp||+++||WT||-||0.10±0.004||W||-||1.76±0.12||R++||+|
|43894OR||+||824bp/702bp, weak 340bp||+++||WT||+++||2.11±0.14||R+++d||++||0.89±0.09||R+++d||++|
The mlrA gene is intact in most isolates and without stx1 prophage insertion
Most STEC strains (15 out of 19) used in this study carried an intact mlrA gene, although 13 strains were stx1+ by PCR, indicating that stx1-encoded bacteriophage may prefer a different chromosomal insertion site in certain non-O157:H7 STEC strains (Tables 1 and 3; Figure S1). DA-33 and SJ15 strains, both stx1-positive, produced a strong 824-bp F/B band with weak 340-bp A/B band by mlrA multiplex PCR, indicating that they likely carried prophage insertions in the proximal region of mlrA as described for strains of E. coli serotype O157:H7  (Table 3; Figure S1). Both O145 strains SJ24 and E59 were PCR-negative for stx1 but produced weak A/B (340 bp), A/E (702 bp), and F/B (824 bp) bands indicative of the prophage insertion in mlrA, with additional larger molecular-weight bands suggestive of primer binding at other prophage sites in the genome. Sequences of the intact mlrA genes from the 15 STEC strains and MG1655 were compared. All the predicted MlrA proteins are extremely conserved, showing only 1 amino acid difference at residue 239; Five strains (2 O103 strains FCL1 and SJ10 and all 3 O45 strains) had a leucine (L239) and the rest of 11 strains had a phenylalanine (F239) (data not shown).
RpoS gene and catalase activity
DNA fragments containing the rpoS ORF and flanking regions were amplified and sequenced (GenBank Accession # KF662844-KF662863). Five different mutations in the rpoS coding region that resulted in amino acid alterations, compared to the RpoS of strain ATCC 43895, were identified in the 19 strains in this study, including 2 strains with an amino acid substitution, and 3 strains with a frame-shift due to single nucleotide insertion or deletion (Table 3 and Figure 2). None of these mutations were the same as those found in the O157:H7 strains PA1-PA52 examined previously . All strains that produced mutant RpoS proteins showed decreased catalase activity using the hydrogen peroxide test (Table 3).
Congo red affinity and curli production
Results on CRI agar are shown in Figure 1 and listed in Table 3. Serotype O157:H7 strains ATCC 43895 (negative control) and 43894OR (positive control) were included for comparison. In contrast to O157 strains, non-O157 STEC strains showed a wider range of pink, brown, and red colors on CRI plates, likely due to the differences in their genetic backgrounds. All four strains that carried an interrupted mlrA appeared white on CRI. Four of the five rpoS mutant strains also appeared white, and one rpoS mutant strain (SJ7) showed a light pink color on CRI.
Production of curli/CsgA in the strains agreed with the CR affinity reasonably well (Figure 3 and Table 3). All 9 strains that were white on CRI did not produce curli. Most CR-binding isolates produced curli as evidenced by the CsgA bands on SDS-PAGE. However, four strains with weak to intermediate CR affinities showed little or no curli production: strains SJ7, 03-4064 and 05-6545 (light pink, pink/red+ and brown/red+, respectively) did not produce detectable levels of curli/CsgA, while strain 98-8338 (brown) produced only a trace amount. This is likely due to CR binding with other cell surface components (such as cellulose) in those strains (data not shown).
Biofilm formation (crystal violet-binding)
Biofilm formation was assayed in 96-well polystyrene plates using crystal violet (CV) dye binding (Table 3). Although most strains that showed intermediate to strong CR binding also bound CV strongly and formed visible biofilm, there were a few exceptions. Strains SJ13 and FCL1 bound CR strongly and had a dry appearance (Figure 1), but visual biofilm was observed in only 1 out of 6 replicate wells (1+/6) (Table 3). This variability may result from deficiencies in initial attachment due to their lack of motility, a stage of biofilm development prior to and thus independent of curli expression (Table 3). Lack of motility may also explain why strains SJ14 and 03-4064 showed little to no biofilm formation in spite of the fact that these strains bound CR to a modest degree, scoring pink/red+ (P/R+) on CRI. Strain 05-6545 was unique in that it was the only strain that produced abundant biofilm (CV OD590 = 2.05) and bound moderate amounts of CR but failed to express curli. No other curli-deficient strain generated enough biofilm to register CV OD590 values above 0.20. Western hybridization of the curli prep from strain 05-6545 with an E. coli CsgA specific antibody revealed a faint band after prolonged exposure, indicating that a low level of CsgA expression was present (data not shown). It is unclear whether the production of curli in 05-6545 was poor, or the curli fibers were structurally different from the norm thus eluding the purification procedure.
Complementation by pUC19::mlrA or pTOPO::rpoS plasmids
Fifteen of the 19 strains were susceptible to ampicillin and were transformed with pUC19::mlrA. The pUC19::mlrA plasmid restored CR binding and curli production in two of the four strains carrying a disrupted mlrA gene (strains DA-33 and SJ15) (Figures 1 and 3, and Table 3). However, CR binding was not restored in the two O145 mlrA mutant strains (SJ24 and E59) or in strains SJ29 and DEC10 (the latter two strains carried rpoS mutations). Biofilm formation (evaluated by CV binding) was enhanced in all 4 pUC19::mlrA complemented mlrA mutant strains, and in only 1 of 5 rpoS mutant strains (SJ7). Since both SJ24 and E59 strains showed strong catalase activity, encoded wild-type RpoS proteins, and were not trans-complemented by mlrA, it must be assumed that their deficiencies in curli production and CR affinity resulted from defects in structural or regulatory genes other than mlrA or rpoS.
Four of the 5 strains that carried a defective rpoS gene were susceptible to kanamycin (DEC10, SJ7, SJ16, and SJ29) and were transformed with pTOPO::rpoS. Strain 96-3285 was resistant to ampicillin, kanamycin and tetracycline (Table 1), thus it was excluded from the complementation study. Strain SJ29, although not resistant to kanamycin, was also excluded due to failure to stably maintain the pTOPO::rpoS plasmid. The catalase activity was successfully restored to near wild-type level by pTOPO::rpoS in strains DEC10, SJ7, and SJ16 (data not shown). SJ7 and SJ16 rpoS-complemented strains also showed slightly increased binding to CR, although not as strong as those complemented by pUC19::mlrA (Figure 1). CR-binding was unaffected by pTOPO::rpoS in strain DEC10, suggesting that this strain may have other defects in curli and/or CsgD expression in addition to the rpoS mutation.
CsgBA structural proteins
To rule out the possibility that the failure to produce curli was due to mutation or deletion in the curli structural genes, the csgBA coding regions were amplified and sequenced from the seven strains that failed to produce curli (SJ16, SJ29, DEC10, 96-3285, 03-4064, SJ24, E59) and 43895. There were over 40 nucleotide differences in the 1.16-kb region (pairwise identity 98.6%), but few resulted in amino acid changes (GenBank Accession # KF662816-KF662823). All seven non-O157 strains encoded full-length CsgB and CsgA. The deduced amino acid sequences of CsgA and CsgB were highly conserved (Figure 4). The CsgA proteins from the 7 non-O157 strains varied only by 1 amino acid at residue 134 (S or T). In addition to several amino acid substitutions, the most striking feature of the CsgA from O157:H7 strain 43895 was that it is one residue (glycine) longer in the glycine-rich region than those of the non-O157 strains due to a 3-bp insertion. However, this variation in the number of glycine residues was found in CsgA from other strains of E. coli in GenBank, thus it is not a rare occurrence (data not shown). The pI of CsgA protein from 43895 is 5.43, compared to 4.89 of the rest. All strains encoded identical CsgB proteins except for the two O145 strains (SJ24 and E59), which contained a phenylalanine rather than a valine at residue 79 and an asparagine rather than a serine at residue 107. These substitutions did not result in significant change in hydrophobicity, pI (9.58), or the predicted secondary structure of the CsgB protein. BLASTP results also showed that both amino acid changes were present in the CsgB from other strains of E. coli either singularly or in combination (data not shown). It is not clear whether these protein sequence variations could affect CR binding or biofilm formation, but it is certain that the lack of curli formation in these 7 strains was not due to truncations of the curli structural proteins.
Sequence and activity of the csgD promoters
As the extensive csgD promoter in E. coli contains numerous sequence-dependent binding sites for regulatory factors that modulate its expression, variations in the promoter sequence could affect promoter activity. The csgD promoter regions, including partial csgB and csgD coding sequences determined to be required for optimal expression of csgD in S. Typhimurium , was sequenced in each of the 19 non-O157 STEC strains for comparison with the sequence from E. coli O157:H7 strain ATCC 43895 (Figure 5). Five distinct promoter patterns (types B-F) were identified among the 19 strains, with type B being the predominant promoter type (Table 1). Promoter type B was found in each of the O26, O111, and O113 strains plus one O103 strain, DA-33, which carries 3 additional synonymous nucleotide changes in the csgD coding region (designated pB/G). Promoter type C was present in the three O45 strains and the 2 remaining O103 strains. Promoter type D was exclusively found in the three O121 strains. Types E and F, the promoters from the two O145 isolates, formed a separate cluster and differed from each other by only 1 bp. Thus there seemed to be some O-serogroup specificity for the csgD promoter sequences.
(A) Alignment of csgD promoters and (B) β-galactosidase activities (Miller Units) over time. A representative time course was presented.
A csgD::lacZ fusion plasmid was constructed from each non-O157 promoter type B-F and from E. coli O157:H7 ATCC 43895, and transformed into the curli-expressing commercial K12 strain EC100. The cloned csgD coding regions were identical except for strains SJ24 and E59, which included three synonymous mutations. Comparisons of the β-galactosidase assays were performed at various time points of growth to assess the promoter strengths. There was a 30% difference at the late log and stationary phases between the two promoters with the highest activities (type E from SJ24 and the 43895 promoter) and the rest of the promoters (Figure 5). However, it is difficult to derive relationships between csgD promoter strengths and biofilm-forming-capabilities when many of the strains in the study carried mutations that limited biofilm formation. Both SJ24 and 43895 strains carried mlrA insertions and failed to form biofilms. Promoter types B-D were each represented by strains capable of biofilm formation and biofilm-deficient strains with defects in mlrA, rpoS, or motility. It is also important to consider that each promoter type may be optimized for, or function differently in, its own respective background. With that limitation recognized, these results indicate that csgD promoter sequence variations play a relatively small role in determining promoter strength.
In the previous study of E. coli serotype O157:H7 strains, we determined that prophage insertions in mlrA and heterogeneous mutations in rpoS constituted major obstacles limiting biofilm formation and the expression of extracellular curli fibers under a defined set of conditions. In this study we used similar conditions to investigate the regulation of biofilm formation in a small but diverse collection of non-serotype O157:H7 STEC isolates. The relative ease with which we identified both biofilm-forming and non-biofilm-forming strains in most serotypes suggested that the non-serotype O157:H7 STEC may not encounter the same regulatory restrictions as the O157:H7 strains. Consistent with that observation, we found only 4/19 strains (21%) with prophage insertions in mlrA, in contrast to serotype O157:H7 where 53/55 (96.4%) strains carried the prophage . There were two other reports that surveyed the status of mlrA (yehV) in non-O157 STEC strains, both with a limited number of strains for each serotype [20,21]. One study found no mlrA prophage insertion in 15 strains of 11 different non-O157:H7 STEC serotypes , all of those serotypes were different from this study. In a different report, the incidence of prophage insertion in mlrA was 28% among approx. 70 strains from 42 non-O157 STEC serotypes, compared to 98% for the serotype O157:H7 strains . Although only 1 of the 7 serogroups (O113) reported here was represented in that study , their results also suggest that prophage insertions in mlrA occur at a much higher frequency in serotype O157:H7 than in many other STEC serogroups. More studies with larger sample sizes are needed to confirm strain, serotype, and serogroup insertion frequencies.
RpoS mutations were also found in 26% (5/19) of the strains in this study and likely contributed to the biofilm deficiencies. However, assumptions concerning the frequency of rpoS mutations and their consequences regarding biofilm formation in comparison with serotype O157:H7 would not be appropriate here since the sampling is limited and might be biased. In a study of RpoS mutations in strains from different sources (soil, food, bovine, or clinical), mutation frequencies in RpoS varied greatly depending on the environment from which strains were collected . Such trends may also apply to non-O157:H7 strains.
All 4 strains carrying prophage in mlrA possessed wild-type rpoS genes, two (DA-33 and SJ15) were successfully complemented for CR binding and curli production by pUC19::mlrA but the two O145 strains were not. As the curli structural genes were intact, the O145 strains may carry mutations elsewhere in the regulatory cascade. On the other hand, attempts to complement the CR-binding phenotype in three rpoS mutant strains were mostly unsuccessful, even though the catalase activity was restored in these strains. As expected, the two rpoS mutant strains SJ29 and DEC10, which showed no catalase activity, were unable to be complemented by pUC19::mlrA, indicating that defects in these rpoS mutants cannot be overcome by over-producing MlrA on a multi-copy plasmid, and that the expression of mlrA on pUC19::mlrA was still regulated by RpoS in the same manner as that of the chromosomal copy. On the other hand, CR binding was moderately complemented by pUC19::mlrA in strains SJ7 and SJ16; both strains showed low levels of catalase activity and probably produced an attenuated RpoS that was made more effective by increasing the level of MlrA. It seems unorthodox that CR binding in those strains was moderately enhanced by extra copies of the mlrA gene, but not so much by adding functional copies of the rpoS gene. Since genetic background and the regulation of biofilm-related genes in the non-O157 STEC serotypes may be different from those of the O157:H7 strains, it would be important to conduct genomic comparisons of these strains to gain a more comprehensive understanding of the molecular mechanism involved. We are currently planning to pursue transcriptome analyses of these strains.
Some strains with an intact mlrA within each serotype failed to express curli or form biofilm under the conditions used in this study. In the previous study of O157:H7 strains, none of the 55 strains formed biofilm (CV OD590 <0.13) and that deficiency could be explained by either prophage insertions or RpoS mutations in all but 1 strain (PA6) . In this study of non-serotype O157:H7 strains, 14 strains showed weak or no biofilm formation (CV OD590 <0.29). Nine of those 14 strains (64.3% of the weak biofilm-formers, 47.4% of total) had either prophage insertions in mlrA or deficiencies in rpoS. Thus 5 of the 19 non-serotype O157:H7 strains (26.3%) had mutations in the biofilm regulatory pathway other than rpoS or mlrA. In all of those 5 strains, there was complete or nearly complete loss of motility. Flagella have been shown to be essential in the early stages of E. coli biofilm development and strains that lacked or produced paralyzed flagella formed only small, dense clusters of attached cells . If lack of flagella-driven motility did compromise biofilm formation in those 5 strains, it would be expected that curli expression, which occurs at a later stage, would remain unaffected as both mlrA and rpoS were intact. However, while curli was strongly expressed in three of the strains with weak or no motility (FCL1, SJ14, and SJ13), it was present at low levels or absent in strains 98-8338 and 03-4064. Plasmid pUC19::mlrA-transformed strains SJ24 and E59 represent additional strains with functional copies of mlrA and rpoS that are unable to form curli; both showed no motility. While the failure to produce curli could be due to mutations in one of the many other csgD regulators, it raises the possibility that certain defects in motility could also have negative effects on curli expression and thus affecting biofilm formation by both inhibiting early motility and then blocking later curli expression. Recent studies have identified factors that regulate both the flagella and csgD operons to orchestrate the switch from a motile to a non-motile state . Additional studies are needed to determine if the lack of curli production in these non-motile strains is caused by deficiencies in a common CsgD/flagella regulator.
Prophage insertions in mlrA, RpoS mutations, and loss of motility provided reasonable explanations for the biofilm and curli deficiencies of the strains in this study; however, the failure to complement the DEC10 rpoS mutation (DEC10 is motile and has an intact mlrA) suggests an additional regulatory mutation in that strain. Mutations in any of the many csgD regulators would be expected to have detrimental effects on curli expression and biofilm formation but they would likely remain phenotypically silent in the face of the widespread nature of the 3 conditions identified in this study. As such, large-scale comparative genomic sequencing of STEC strains will be needed to identify additional mutations of individual csgD regulators. It is conceivable that variations in the csgD promoter region where many of these factors bind could cause strain as well as serotype differences in csgD/curli expression. Sequencing of the csgD promoters in this study indicated that there were conserved differences among and between serogroups. We also tested promoter strengths from each promoter type in a common K12 strain background. Our results indicated that there were serotype differences in the strength of the csgD promoters but these differences may not play an important role in CR-affinity or curli production. Additional studies in each of the STEC backgrounds and with integrated single-copy-fusions will be needed to fully test the consequences of these promoter differences.
All biofilm-deficient strains in this study could be characterized as having either: 1) prophage insertion in mlrA; 2) rpoS mutation; or 3) impaired motility. There was little overlap in these three groups. Of the 14 strains in this study that possessed at least one of the three impairments, none carried both mlrA and rpoS defects, and only two O145 strains (SJ24 and E59) were non-motile with mlrA prophage insertions. In contrast, 38 of the 53 serotype O157:H7 strains (71.7%) had altered RpoS in addition to prophage insertions in mlrA, suggesting that the accumulation of these two genotypes were not mutually exclusive. The only strains in the present study that had prophage insertions and one other condition (lack of motility) were the two O145 strains, which may indicate that there was selection pressure against accumulating multiple defects in the biofilm pathways in most non-O157:H7 STEC strains. Interestingly, O145 strains had repeatedly grouped closely with serotype O157:H7, the most successful pathogenic serogroup among the STEC, either by sequence similarity, promoter strength, or other phenomena examined in this study. These results agreed with the genome comparisons showing that O145 and O157 were closely related (X. Yan, personal communication of unpublished results).
In this study we tested a small collection of 19 non-serotype O157:H7 STEC strains from the 7 important O-serogroups (O26, O45, O103, O111, O113, O121, and O145) to evaluate whether prophage insertions in mlrA and mutations in rpoS present major barriers to curli expression and biofilm formation as they do in strains of serotype O157:H7. We discovered that, while both genotypes are encountered, mlrA prophage insertions are unlikely to be as common as they are in serotype O157:H7. We also discovered that deficiencies in motility form a third general condition that, along with prophage insertions and rpoS mutations, provide a plausible explanation for the lack of biofilm formation noted in all biofilm-deficient strains in this study. While mutations in other csgD regulators, as observed for strain DEC10, await further studies, we have described conserved serotype variations in the csgD promoter that add to the regulatory complexity of curli and biofilm regulation in the STEC strains.
We thank Dr. Chandi Wijey for PCR verification of the O serotypes, Ms. Susan Lawlor and Dr. David Needleman at the Core Facility for their assistance in sequencing. We are grateful for Drs. John Luchansky, Cristobal Chaidez, and Pina Fratamico for supplying the strains.
Disclaimer: USDA is an equal opportunity provider and employer. Mention of trade names or commercial products in this article is solely for the purpose of providing specific information and does not imply recommendation or endorsement by the U.S. Department of Agriculture.
Conceived and designed the experiments: CYC GAU. Performed the experiments: CYC CSH BJC TPS GCP LHN GAU. Analyzed the data: CYC CSH BJC TPS GCP LHN XY GAU. Contributed reagents/materials/analysis tools: GCP XY. Wrote the manuscript: CYC GAU. Drafted and edited the paper: CYC CSH BJC TPS GCP LHN XY GAU.
- 1. Brooks JT, Sowers EG, Wells JG, Greene KD, Griffin PM et al. (2005) Non-O157 Shiga toxin-producing Escherichia coli infections in the United States, 1983-2002. J Infect Dis 192: 1422-1429. doi:https://doi.org/10.1086/466536. PubMed: 16170761.
- 2. Rangel JM, Sparling PH, Crowe C, Griffin PM, Swerdlow DL (2005) Epidemiology of Escherichia coli O157:H7 outbreaks, United States, 1982-2002. Emerg Infect Dis 11: 603-609. doi:https://doi.org/10.3201/eid1104.040739. PubMed: 15829201.
- 3. Takkinen J, Struelens M, Niskanen T, Makela P, Rizzi V, et al. (2011) European Centre for Disease Prevention and Control and European Food Safety Authority. Shiga toxin/verotoxin-producing Escherichia coli in humans, food and animals in the EU\EEA, with special reference to the German outbreak strain STEC O104. . Stockholm: ECDC. Available: . http://www.ecdc.europa.eu/en/publications/Publications/1106_TER_EColi_joint_EFSA.pdf. Accessed 2013 Sept 10.
- 4. Notice FSIS 47-13. FSIS verification testing for non-O157 Shiga toxin-producing Escherichia coli (non-O157 STEC) under MT60, MT52, and MT53 sampling programs. (Re-issued and updated FSIS Notice 40-12.). Available: . http://www.fsis.usda.gov/wps/wcm/connect/33a38418-61e7-4b48-822f-6eecb8c582a1/47-13.pdf?MOD=AJPERES. Accessed 2013 December 9.
- 5. Bosilevac JM, Koohmaraie M (2011) Prevalence and characterization of non-O157 Shiga toxin-producing Escherichia coli isolates from commercial ground beef in the United States. Appl Environ Microbiol 77: 2103-2112. doi:https://doi.org/10.1128/AEM.02833-10. PubMed: 21257806.
- 6. Karmali MA, Mascarenhas M, Shen S, Ziebell K, Johnson S et al. (2003) Association of genomic O island 122 of Escherichia coli EDL 933 with Verocytotoxin-producing Escherichia coli seropathotypes that are linked to epidemic and/or serious disease. J Clin Microbiol 41: 4930-4940. doi:https://doi.org/10.1128/JCM.41.11.4930-4940.2003. PubMed: 14605120.
- 7. Parsek MR, Singh PK (2003) Bacterial Biofilms: an emerging link to disease pathogenesis. Annu Rev Microbiol 57: 677-701. doi:https://doi.org/10.1146/annurev.micro.57.030502.090720. PubMed: 14527295.
- 8. Van Houdt R, Michiels CW (2010) Biofilm formation and the food industry, a focus on the bacterial outer surface. J Appl Microbiol 109: 1117-1131. doi:https://doi.org/10.1111/j.1365-2672.2010.04756.x. PubMed: 20522145.
- 9. Zogaj X, Nimtz M, Rohde M, Bokranz W, Römling U (2001) The multicellular morphotypes of Salmonella typhimurium and Escherichia coli produce cellulose as the second component of the extra cellular matrix. Mol Microbiol 39: 1452-1463. doi:https://doi.org/10.1046/j.1365-2958.2001.02337.x. PubMed: 11260463.
- 10. Bokranz W, Wang X, Tschäpe H, Römling U (2005) Expression of cellulose and curli fimbriae by Escherichia coli isolated from the gastrointestinal tract. J Med Microbiol 54: 1171-1182. doi:https://doi.org/10.1099/jmm.0.46064-0. PubMed: 16278431.
- 11. Gerstel U, Römling U (2003) The CsgD promoter, a control unit for biofilm formation in Samonella typhimurium. Res Microbiol 154: 659-667. doi:https://doi.org/10.1016/j.resmic.2003.08.005. PubMed: 14643403.
- 12. Gerstel U, Park C, Römling U (2003) Complex regulation of csgD promoter activity by global regulatory proteins. Mol Microbiol 49: 639-654. PubMed: 12864849.
- 13. Ogasawara H, Yamamoto K, Ishihama A (2010) Regularory role of MlrA in transcription activation of csgD, the master regulator of biofilm formation in Escherichia coli. FEMS Microbiol Lett 312: 160-168. doi:https://doi.org/10.1111/j.1574-6968.2010.02112.x. PubMed: 20874755.
- 14. Ogasawara H, Yamada K, Kori A, Yamamoto K, Ishihama A (2010) Regulation of the Escherichia coli csgD promoter: interplay between five transcription factors. Microbiology 156: 2470-2483. doi:https://doi.org/10.1099/mic.0.039131-0. PubMed: 20466769.
- 15. Brown PK, Dozois CM, Nickerson CA, Zuppardo A, Terlonge J et al. (2001) MlrA, a novel regulator of curli (AgF) and extracellular matrix synthesis by Escherichia coli and Salmonella enterica serovar Typhimurium. Mol Microbiol 41: 349-363. doi:https://doi.org/10.1046/j.1365-2958.2001.02529.x. PubMed: 11489123.
- 16. Pesavento C, Becker G, Sommerfeldt N, Possling A, Tschowri N et al. (2008) Inverse regulatory coordination of motility and curli-mediated adhesion in Escherichia coli. Genes Dev 22: 2434-2446. doi:https://doi.org/10.1101/gad.475808. PubMed: 18765794.
- 17. Biscola FT, Abe CM, Guth BEC (2011) Determination of adhesin gene sequences in, and biofilm formation by, O157 and non-O157 Shiga toxin-producing Escherichia coli strains isolated from different sources. Appl Environ Microbiol 77: 2201-2208. doi:https://doi.org/10.1128/AEM.01920-10. PubMed: 21317257.
- 18. Wang R, Bono JL, Kalchayanand N, Shackelford S, Harhay DM (2012) Biofilm formation by Shiga toxin-producing Escherichia coli O157:H7 and non-O157 strains and their tolerance to sanitizers commonly used in the food processing environment. J Food Prot 75: 1418-1428. doi:https://doi.org/10.4315/0362-028X.JFP-11-427. PubMed: 22856565.
- 19. Uhlich GA, Chen C-Y, Cottrell BJ, Hofmann CS, Dudley EG et al. (2013) Phage insertion and variations in rpoS limit curli formation and biofilm formation in serotype O157:H7. Microbiology 159: 1586-1596. doi:https://doi.org/10.1099/mic.0.066118-0. PubMed: 23744902.
- 20. Shaikh N, Tarr PI (2003) Escherichia coli O157:H7 Shiga toxin-encoding bacteriophages: integrations, excisions, truncations, and evolutionary implications. J Bacteriol 185: 3596-3605. doi:https://doi.org/10.1128/JB.185.12.3596-3605.2003. PubMed: 12775697.
- 21. Serra-Moreno R, Jofre J, Muniesa M (2007) Insertion site occupancy by stx2 bacteriophage depends on the locus availability of the host strain chromosome. J Bacteriol 189: 6645-6654. doi:https://doi.org/10.1128/JB.00466-07. PubMed: 17644594.
- 22. Medina MB, Shelver WL, Fratamico PM, Fortis L, Tillman G et al. (2012) Latex Agglutination Assays for Detection of Non-O157 Shiga Toxin Producing Escherichia coli Serogroups O26, O45, O103, O111, O121, and O145. J Food Prot 75: 819-826. doi:https://doi.org/10.4315/0362-028X.JFP-11-430. PubMed: 22564929.
- 23. Fratamico P, Bagi LK (2012) Detection of Shiga Toxin-Producing Escherichia coli in Ground Beef Using the GeneDisc Real-Time PCR System. Frontiers in Cellular and Infection Microbiology 2: 152. PubMed: 23267438.
- 24. Wasilenko JL, Fratamico PM, Narang N, Tillman GE, Ladely S et al. (2012) Influence of Primer Sequences and DNA Extraction Method on Detection of Non-O157 Shiga Toxin Producing Escherichia coli in Ground Beef by Real-Time PCR Targeting the eae, stx, and Serogroup-Specific Genes. J Food Prot 75: 1939-1950. doi:https://doi.org/10.4315/0362-028X.JFP-12-087. PubMed: 23127702.
- 25. Uhlich GA, Keen JE, Elder RO (2001) Mutations in the csgD promoter associated with variations in curli expression in certain strains of Escherichia coli O157:H7. Appl Environ Microbiol 67: 2367-2370. doi:https://doi.org/10.1128/AEM.67.5.2367-2370.2001. PubMed: 11319125.
- 26. Beutin L, Miko A, Krause G, Pries K, Haby S et al. (2007) Identification of Human-Pathogenic Strains of Shiga Toxin-Producing Escherichia coli from Food by a Combination of Serotyping and Molecular Typing of Shiga Toxin Genes. Appl Environ Microbiol 73: 4769-4775. doi:https://doi.org/10.1128/AEM.00873-07. PubMed: 17557838.
- 27. DebRoy C, Roberts E, Valadez AM, Dudley EG, Cutter CN (2011) Detection of Shiga Toxin–Producing Escherichia coli O26, O45, O103, O111, O113, O121, O145, and O157 Serogroups by Multiplex Polymerase Chain Reaction of the wzx Gene of the O-Antigen Gene Cluster. Foodborne Pathog Dis 8: 651-652.
- 28. Burkart M, Toguchi A, Harshey RM (1998) The chemotaxis system,but not chemotaxis, is essential for swarming motility in Escherichia coli. Proc Natl Acad Sci U S A 95: 2568–2573. doi:https://doi.org/10.1073/pnas.95.5.2568. PubMed: 9482927.
- 29. Miller JH (1972) Experiments in molecular genetics. Cold Spring Harbor Laboratory, Cold Spring Harbor, NY..
- 30. Gerstel U, Kolb A, Römling U (2006) Regulatory components at the csgD promoter-additional roles for OmpR and integration host factor and role of the 5’ untranslated region. FEMS Microbiol Lett 261: 109-117. doi:https://doi.org/10.1111/j.1574-6968.2006.00332.x. PubMed: 16842367.
- 31. van Hoek AHAM, Aarts HJM, Bouw E, van Overbeek WM, Franz E (2013) The role of rpoS in Escherichia coli O157 manure-amended soil survival and distribution of allelic variations among bovine, food and clinical isolates. FEMS Microbiol Lett 338: 18-23. doi:https://doi.org/10.1111/1574-6968.12024. PubMed: 23066907.
- 32. Pratt LA, Kolter R (1998) Genetic analysis of Escherichia coli biofilm formation: roles of flagella, motility, chemotaxis and type I pili. Mol Microbiol 30: 285-293. doi:https://doi.org/10.1046/j.1365-2958.1998.01061.x. PubMed: 9791174.
- 33. Pesavento C, Hengge R (2012) The global repressor FliZ antagonizes gene expression by σS-containing RNA polymerase due to overlapping DNA binding specificity. Nucleic Acids Res 40: 4783-4793. doi:https://doi.org/10.1093/nar/gks055. PubMed: 22323519.