Escherichia coli O157 strains belonging to a distinct lineage and expressing different O-antigen (Oag) lengths were isolated. Although the function of wzz in E. coli has not been adequately investigated, this gene is known to be associated with regulation of Oag length. Using E. coli O157:H7 ATCC43888 (wild-type), several wzz mutants of E. coli O157, including a wzz deletion mutant, were generated and the relationship between the length of Oag modulated by the wzz gene and sensitivities to serum complement investigated. SDS–PAGE, immunoblot analyses and sensitivity tests to human serum complement were performed on these strains. The lengths of the O157-antigen could be modulated by the wzz gene mutations and were classified into long, intermediate and short groups. The short chain mutant was more serum sensitive than the wild-type strain and the other wzz mutants (P < 0.001). In conclusion, Oag chain length modulated by the wzz gene in E. coli O157 influences its sensitivities to serum complement. The present findings suggest that E. coli O157 strains with intermediate or long length Oag chains might show greater resistance to serum complement than those with short chains.
- E. coli
enterohemorrhagic Escherichia coli
normal human serum
- S. enterica serovar Typhimurium
Salmonella enterica serovar Typhimurium
- S. flexneri
Lipopolysaccharide, a major component of the outer membrane of gram-negative bacteria, consists of lipid A, core oligosaccharide and O-specific polysaccharide or Oag that protects the bacteria from antibiotics and host defenses [1-4]. The LPS of most E. coli strains has a basic Oag chain length of 10 to 18 O units [5, 6]. However, among E. coli strains with the same Oag (O2, O7, O16 and O18), Oag chain length is associated not only with chemical differences, but also with defense variations in LPS [6-8]. Chart et al. showed that the LPSs of E. coli O157 is expressed heterogeneously: there are three types (short [7–16], intermediate [10–18] and long [16–25] Oag chains) . In particular, E. coli O157 is known to express 15–20 units to form Oag  and E. coli O157:H7 is the main E. coli serotype responsible for severe, bloody diarrhea (hemorrhagic colitis)  and hemolytic uremic syndrome .
The number of O units per molecule has a characteristic modal distribution which has been shown in several systems to be determined by the chain length determinant Wzz (previously Cld or Rol) protein [13-16]. The wzz gene is located downstream of the Oag synthesis gene cluster . Wzz is a membrane protein with two transmembrane helices, known as the PCPs, which are involved in regulation of chain-lengths of Oags [18, 19]. The PCP family contains proteins not only for Oags but for other polysaccharide biosynthesis such as capsular polysaccharides . Wzz acts as a polymerase and chain length regulator with Wzy and Wzx, which encode Oag polymerase and flippase, respectively [21, 22]. Wzz, a representative protein in the Wzy-dependent pathway belonging to the PCP family, plays a role in determining the distribution of lengths of the synthesized polysaccharide chain . The role of different Wzz proteins and various Oag chain lengths has also been studied in S. flexneri and Salmonella [24-26]. In E. coli O157:H7, the presence of WzzFepE confers a very long Oag chain length (>100 repeat units) , whereas WzzE confers a short length (5–7 repeat units) . In S. enterica serovar Typhimurium, WzzB (ST) confers a long Oag (16–35 repeat units) . PmrA-dependent pbgE2 and pbgE3 gene products control formation of a short Oag in S. enterica serovar Typhimurium . Their structures are very similar to each other at the promoter level . Additionally, in S. flexneri, the presence of WzzSF confers a short Oag chain length (11–17 repeat units), whereas WzzpHS-2 confers a very long Oag chain length (>90 repeat units) [25, 30].
The functions of the wzz gene in E. coli O157 have not been fully investigated. A recent study used the deletion mutants of virulence genes, flagellar genes, and LPS Oag related genes (rfbE and waaL) to investigate the function of E. coli O157:H7 . Based on this background, we used O157:H7 and non-H7 strains carrying different Oag lengths to generate a wzz-deficient mutant and its complements and examined the relationship between the length of the Oag modulated by the wzz gene and sensitivities to serum complement in E. coli O157 strains.
MATERIALS AND METHODS
Bacterial strains and media
Three strains of E. coli, O157:H16 (PV807), O157:H43 (PV00-24) and O157:HNM (H non-motile; PV05-43) were analyzed in a previous study , and E. coli O157:H7 strains of EDL933 , Sakai  and ATCC43888 (no Stx1 or Stx2 production) were used in our analyses (Table 1). For bacterial culture, LB broth (BD, Becton Dickinson, Franklin Lakes, NJ, USA) and plates were used. LB broth and plates each containing one of the following antibiotics–100 µg/mL ampicillin (Wako, Osaka, Japan), 50 µg/mL chloramphenicol (Wako) or 50 µg/mL kanamycin (Wako)–were used as media for selection of strains carrying a drug-resistance gene.
|Strains or plasmids||Genotype and/or phenotype||Source of reference|
|DH5a||E. coli K-12||Promega|
|EDL933||E. coli O157:H7; stx1+, stx2+||(33)§|
|Sakai||E. coli O157:H7; stx1+, stx2+||(34)§|
|ATCC43888||E. coli O157:H7; stx1−, stx2−||ATCC|
|ATCC43888 Δwzz::kan||This study|
|ATCC43888 Δwzz::kan + pAT438||This study|
|ATCC43888 Δwzz::kan + pOP8||This study|
|ATCC43888 Δwzz::kan + pOP18||This study|
|ATCC43888 Δwzz::kan + pOP14||This study|
|pKD46||bla PBAD gam bet exo pSC101 oriTS||(38)§|
|pKD3||bla FRT cat FRT oriRy||(38)|
|pKD4||bla FRT kan FRT oriRy||(38)|
|pGEM-T easy||PCR cloning vector, bla||Promega|
|pAT438||pGEM-T easy with the complete wzz gene of ATCC43888||This study|
|pOP8||pGEM-T easy with the complete wzz gene of PV807||This study|
|pOP18||pGEM-T easy with the complete wzz gene of PV00-24||This study|
|pOP14||pGEM-T easy with the complete wzz gene of PV05-43||This study|
DNA and plasmid extraction
After overnight culture in LB broth at 37°C, DNA extraction was performed using an illustra acteria genomicPrep Mini Spin Kit (GE Healthcare Japan, Tokyo, Japan) and plasmid extraction was performed using an illustra plasmidPrep Mini Spin Kit (GE Healthcare).
Lipopolysaccharide sodium dodecyl sulfate-polyacrylamide gel electrophoresis and immunoblot analysis
Lipopolysaccharide analysis was performed as previously described [35, 36]. Culture adjusted to OD600 nm = 0.8 in LB broth was centrifuged and the pellets suspended in sterile physiological saline. This process was repeated three times, after which it was suspended with sterile water. An equal volume of SDS sample buffer (4% SDS, 8% 2-mercaptoethanol, 20% glycerol, 0.1 M Tris pH 6.8, bromophenol blue) was added and incubated at 100°C for 10 min. Then 10 mL of 10 mg/mL Proteinase K (Wako) was added and incubated at 60°C for 1 hr. These were used as LPS samples in the following analysis [35, 36]. The LPS samples were preserved at −30°C. The LPS samples were separated on 12% polyacrylamide gels (Atto, Tokyo, Japan) using a Tricine–SDS buffer system, after which silver staining was performed [35, 36]. First, the gels were fixed in Solution 1 (25% isopropanol, 7% acetic acid) for 2 hr followed by oxidation in Solution 2 (8.33 mg/mL periodic acid, 0.65% isopropanol, 0.18% acetic acid) for 10 min. After four washes in distilled water for 10 min per wash, silver staining was performed in Solution 3 (0.019 N of NaOH, 0.19% ammonium hydroxide, 0.14% silver nitrate) for 10 min. After two washes in distilled water for 10 min each, development was performed in Solution 4 (0.05 mg/mL citric acid, 0.06% formaldehyde) for 10 min. The reaction was stopped by acetic acid followed by a wash in distilled water for 5 min. A densitometry analysis of silver-stained gels was performed using Image J software . LPS samples were also separated on 12% SDS–PAGE gels and transferred to hydrophobic polyvinylidene difluoride (Hybond-P) membranes (GE Healthcare) at 2 mA/cm. The membranes were blocked with Western Blocking Reagent solution (Roche Diagnostics, Indianapolis, IN, USA). Incubations with anti-O157 (1:7500 dilution) and secondary antibody (1:10,000) were carried out for 1 hr at room temperature. Blots were developed with ECL Plus Western Blotting Detection Reagents (GE Healthcare). The visible ladders, not bands, were considered significant findings of Oag .
Preparation of competent cells
Competent cells were prepared using a previously described method . E. coli K-12 DH5α (Promega, Madison, WI, USA) was diluted at 1:200 in 40 mL fresh LB broth containing 100 µg/mL ampicillin and cultured at 37°C for 2–3 hr. E. coli O157:H7 ATCC43888 carrying pKD46 was diluted and cultured at 37°C for 4–5 hr in 40 mL LB broth containing 50 µg/mL kanamycin. Following centrifugation at 4400 g at 4°C for 5 min and two washes, the pellets were resuspended in an equal volume of 10% glycerol. These cells were stored at −80°C and were used in electroporation as competent cells.
Preparation of wzz mutants
To investigate the relationship between the length of the Oag modulated by the wzz gene and sensitivities to serum complement in the E. coli O157 strains, the wzz-deficient mutant and its complements of Stx-nonproducing E. coli O157:H7 ATCC43888 were prepared using the one step method of Datsenko and Wanner et al. . The region containing the kanamycin resistance gene was amplified from pKD4 by PCR using dwzz1 and dwzz2 (Table 2). The purified PCR products were transferred to E. coli ATCC43888 carrying pKD46 by electroporation. Following incubation in super optimal broth with catabolite repression medium (2.0% tryptone, 0.5% yeast extract, 10 mM NaCl, 2.5 mM KCl, 10 mM MgCl2, 20 mM glucose) at 37°C for 1.5 hr, the mixture was cultured on LB plates containing 50 µg/mL kanamycin at 37°C overnight . Deficiency of the wzz gene was confirmed by PCR and sequence analysis. PCR amplifications of wzz genes from E. coli ATCC43888 (wild-type), PV807 (long Oag), PV00-24 (intermediate Oag) and PV05-43 (short Oag) were performed using the sequence of the wzz gene determined in the present study (Table 2). The purified PCR products were cloned into pGEM-T Easy Vector (Promega), followed by wzz-deficient mutant transformation with each plasmid (Table 1). SDS–PAGE and immunoblotting were then performed on the wzz-inserted strains. All genomic sequences were compared among the EDL933, Sakai and ATCC43888 strains, which are identified as the same clone for evolutionary and systematic purposes except that, as mentioned above, the ATCC43888 strain lacks stx1 and stx2 phages, which provide verotoxins . Therefore we used the E. coli ATCC43888 strain as the host strain for the preparation of wzz mutants (Fig. 1a).
|Primer name||Sequence (5′–3′)||Primer size (bp)||Target gene||Refs.|
|dwzz1||TCACCACCCTGCCCTTTTTCTTTAAAACCG||56||kan on pKD4||This study|
|wzz_Sakai_F||ATGCGGACTTGGAAATTTCCG||21||ATCC43888 or PV05-43 or PV00-24 wzz||This study|
|wzz_Sakai_R||TTACTTCGCGTTGTAATTACGC||22||ATCC43888 wzz||This study|
|wzz_PV05-43_R||TTACTTCGCGTTATAATTACGC||22||PV05-43 or PV00-24 wzz||This study|
|wzz_EC95-42_F||ATGAGAGTAGAAAATAATAATG||22||PV807 wzz||This study|
Sensitivity tests to human serum
To examine the function of Wzz in resistance to the exogenous stress of serum-mediated killing, the concentration of serum complement in the present assays was based on those used for checking stress responses in previous studies [4, 26, 30, 31]. Sensitivity to human serum (Lonza, Basel, Switzerland) was tested as described by Hölzer et al.  with slight modifications. Overnight bacterial cultures were diluted at 1:200 in fresh LB broth, grown at 37°C until OD660 nm = 0.6, and then washed twice with 0.9% NaCl. HIS was prepared by incubation for 30 min at 56°C. The bacterial suspensions were diluted 10-fold in 0.9% NaCl with NHS or HIS. The final concentrations of bacterial suspensions were diluted to 30% or 50% as volumes of NHS and HIS, respectively, and incubated for 60 min at 37°C. Serial dilutions of the samples with 0.9% NaCl were prepared and plated onto LB plates to determine viable bacteria. The survival rates of bacteria exposed to NHS were compared to controls with HIS in order to normalize all samples and the percent survival of wild-type bacteria calculated. Statistical analysis was performed by ANOVA and the distribution of percent survival between wild-type strain and mutant strains compared by Tukey's post-hoc test using PASW Statistics 17.0 software package (for Windows; SPSS, Chicago, IL, USA). Data are presented as the mean ± SD and a value of P < 0.05 was considered statistically significant.
Lipopolysaccharide analysis of O157 and wzz-mutant strains
In an attempt to understand the genetic basis of Oag chain length heterogeneity, we constructed strains with the wzz deletion mutant (Δwzz) of the wild-type strain (ATCC43888) and with the wzz genes from E. coli PV807 (long Oag), PV00-24 (intermediate Oag) and PV05-43 (short Oag) strains inserted into the Δwzz strain (Table 1). To determine if the effects of the mutations included Oag expression, we isolated LPS from EDL933, Sakai or ATCC43888 (E. coli O157:H7 parental strains) and from its isogenic wzz mutant strains. We analyzed the strains by SDS–PAGE, silver staining (Fig. 1a,b), densitometry analysis (Fig. 1c) and immunoblotting with anti-O157 antibody (Fig. 1d) and found that variation of Oag chain length could be subdivided into lipid A-core, short (S: <15 units), intermediate (I: 15–20 units) and long (L: >20 units) chain length. EDL933, Sakai and ATCC43888 strains showed expression of the intermediate Oag chain (Fig. 1a). We considered the bands that are visible in the long region to be non-specific : our preliminary SDS–PAGE results showed non-specific bands in the long region in all 16 types of parental E. coli strains (data not shown). In the wzz deletion mutant (Δwzz), expression of the intermediate Oag chain length in the wild-type was reduced (Fig. 1b, lane 1) compared with the vector containing the wzz gene (Δwzz + pAT438; Fig. 1b, lane 2 for the intermediate chain). Each mutant showed a different pattern of Oag expression (Fig. 1b). Long (Δwzz + pOP8), intermediate (Δwzz + pOP18) and short (Δwzz + pOP14) Oag mutant strains expressed the long, intermediate and short chains of the donor strains, respectively (Fig. 1b, lanes 3 and 4 for the long, lanes 5 and 6 for the intermediate, lanes 7 and 8 for the short chains), suggesting Wzz function regarding which length of Oag is encoded may depend on the type of strain.
According to densitometry analysis, we classified the chains as short (<15 units), intermediate (15–20 units) and long (>20 units) in E. coli O157 strains and wzz mutants (Fig. 1c). Furthermore, each mutant showed the same pattern of Oag expression according to immunoblotting with anti-O157 antibody (Fig. 1d).
Serum sensitivities of wzz mutants
To determine whether changes in Oag chain length preference resulted in altered sensitivity to stresses, we incubated wild-type and wzz mutant strains in diluted normal human serum. At a 30% serum concentration, the wzz mutant (Δwzz) and this mutant harboring pAT438, pOP8 or pOP18 plasmids showed no significant differences in survival compared with the wild-type ATCC43888 strain, an exception being the short chain mutant (Δwzz + pOP14) (survival rate 33.1 ± 2.7%; P < 0.001). At a 50% serum concentration, the wzz deletion mutant (Δwzz) exhibited reduced survival compared with the wild-type strain (survival rate: 66.8 ± 8.9%; P < 0.05; Fig. 2). At a 50% serum concentration, the Δwzz + pAT438, long chain (Δwzz + pOP8) and intermediate chain mutants (Δwzz + pOP18) showed no significant difference in survival compared with the wild-type strain. The short chain mutant (Δwzz + pOP14; survival rate: 18.8 ± 10.3%) was more serum sensitive than the wild-type strain or the other wzz mutants (P < 0.001).
Different lengths of Oag can be used to determine the effects of different wzz genes on chain length distribution in strains with a common genetic background. In a previous study, we found genetic variations in Oag chain length in E. coli O157 strains (data not shown) . Although most strains containing typical E. coli H7 have intermediate or long chains, some have short chains . In the present study, we subdivided E. coli O157 strains into three groups with short (<15 units), intermediate (15–20 units) and long (>20 units) chains. Comparison of the sequences of the wzz genes of the strains in this study (ATCC43888: wild-type, PV807 [long Oag], PV00-24 [intermediate Oag] and PV05-43 [short Oag]) showed significant divergence among O157 strains, as shown in our previous study . Some strains had short or intermediate length Oags similar to the parental strains; however, one had a long Oag that differed from the parental one (data not shown) .
The length of their Oag chains affects how gram-negative bacteria interact with the complement proteins of the innate immune system. Long chain lengths prevent complement from reaching the bacterial surface to cause lysis . We found that the short chain mutant was more serum sensitive than the wild-type strain or other wzz mutants. These data were partly supported by other studies of wzz mutant S. enterica serovar Typhimurium and S. flexneri 2a bacteria that reported that those with very long Oag lengths are highly resistant to serum [4, 25, 26, 40, 41] and that E. coli O86:B7 with the very short type of LPS is serum sensitive . In addition, PmrA regulator induces expression of the wzzfepE gene and increases the amount of very long chain Oag, which is required for the resistance of Salmonella to human serum complement . Moreover, E. coli O157:H7 deletion mutants (rfbE and waaL), which are LPS Oag-related genes, significantly decreased each survival rate compared to the parental strain in 20% and 50% serum complement . Our study also showed decreased survival of the wzz deletion mutant (Δwzz) as the serum concentration increased from 30% to 50% (P < 0.05), whereas short, intermediate and long chain mutants did not exhibit a decrease in survival at our studied serum concentrations, which we selected on the basis of the previous published reports mentioned above.
Regarding the sequence of the wzz gene, in S. enterica serovar Typhimurium the region of the wzz deletion mutant (Δwzz) has reportedly been identified with that of wzzB in the EDL933 complete genome sequence (Gene Bank accession no. AE005174) and WzzB (ST) conferred long Oag (16–35 repeat units) . Therefore, we consider that our data confirms that Δwzz in E. coli O157 strains retains short chain lengths (Fig. 1b,d) and that intermediate and/or long chain length Oag are expected to play a role that is possibly equal to that of WzzB (ST) but not of WzzFepE, which confers a very long O-antigen chain length . Additionally, SDS–PAGE and immunoblotting did not reveal a very long chain length, as shown in Figure 1. Thus, our interpretation is that E. coli O157 strains might have variations in chain length Oag in each strain. In a recent study, Kintz and Goldberg reported a relationship between Wzz and oligomerization for chain length regulating activity and found differences in complex stability . This report supports our findings that intermediate or long Oag chains may play a facilitating role in the survival of E. coli O157 strains during serum-mediated stimulation in contrast to the short chains of other E. coli O157 strains. We interpret the significance of our study as follows: (i) we demonstrated a relationship between different responses to serum stress and length of Oag by using two different serum concentrations; and (ii) we showed the function of wzz in coding the length of Oag in E. coli O157, which suggests that E. coli O157 significantly increases its resistance to serum stress by changing the length of Oag.
In conclusion, we determined that the wzz genes of E. coli O157 strains affect Oag chain length and found that modulation of short Oag chain length by the wzz gene influences sensitivities to serum complement of these strains. Consistent with what previous studies have shown, intermediate and/or long lengths of Oag of E. coli O157 have been optimized for pathogenesis during evolution and Wzz function regarding which length of Oag was encoded may depend on the type of strain. These results suggest that E. coli wzz genes with varyings lengths of Oag show concomitant variations in sensitivity to serum complement.
The authors declare no conflict of interest or financial support.