Serodiagnosis of infection with Verocytotoxin-producing Escherichia coli


Dr Henrik Chart, Laboratory of Enteric Pathogens, Central Public Health Laboratory, 61 Colindale Avenue, London, NW9 5HT, UK.


Human sera (167) were screened for antibodies to lipopolysaccharide (LPS) prepared from strains of Verocytotoxin-producing Escherichia coli (VTEC) belonging to a range of serogroups, secreted proteins expressed by attaching and effacing VTEC, enterohaemolysin and H = 7 flagellar proteins. Twelve sera (about 7%) contained antibodies to the LPS of E. coli 05 (one), 026 (two), 0115 (two), 0145 (one), 0163 (one) and 0165 (five). Sera containing antibodies to the LPS of E. coli O26 and O145 also contained antibodies to secreted proteins of 100 and 40 kDa. An additional 34 sera, known to contain antibodies to the lipopolysaccharide of E. coli O157, were examined for antibodies to enterohaemolysin, H = 7 flagellar antigens and bacterial cell surface-associated proteins of 5, 6 and 22 kDa. Three sera contained antibodies to enterohaemolysin and one serum contained antibodies to flagellar proteins. Antibodies to membrane-associated proteins were not detected. It was concluded that enterohaemolysin, H = 7 flagellar proteins and the cell surface-associated proteins were unsuitable for use in immunoassays for providing evidence of infection with VTEC.

Verocytotoxin-producing Escherichia coli (VTEC), particularly strains of serotype O157:H7, are a major cause of diarrhoea, bloody diarrhoea and haemolytic uraemic syndrome (HUS) (Karmali 1989). Infection with VTEC results in serum antibodies to the lipopolysaccharide (LPS) of this organism, and detection of these antibodies has formed the basis of serological tests for providing evidence of infection when E. coli belonging to serogroup O157 could not be cultured from patients’ faeces (Chart et al. 1989, 1991; Thomas et al. 1996). Although VTEC belonging to serogroup 0157 have been most frequently associated with HUS, strains of VTEC belonging to serogroups other than 0157 have also been isolated from humans suffering from HUS and bloody diarrhoea (Johnson et al. 1996). Patients infected with these VTEC have been shown to produce antibodies to the LPS of several E. coli serogroups including O5, O115, O145, O153 and O165 (Chart & Rowe 1990; Chart et al. 1993b, 1996), and a simple dot blotting procedure has been devised to detect these antibodies (Chart & Rowe 1997).

Certain strains of enteropathogenic E. coli (EPEC) adhere to human intestinal epithelial cells, causing ‘attaching and effacing’ (AE) lesions (Jerse et al. 1991). The genes encoding the AE mechanism are located on the locus for enterocyte effacement (LEE) region of the E. coli chromosome (Donnenberg & Kaper 1991; Donnenberg et al. 1993; Foubister et al. 1994; Jarvis et al. 1995; Kenny & Finlay 1995; Jarvis & Kaper 1996). Genes in this locus include eae A, which encodes the 97 kDa intimin protein (Donnenberg & Kaper 1991), and eae B, which encodes a secreted protein of 37 kDa involved in triggering host cell signal transduction and the phosphorylation of tyrosine (Foubister et al. 1994). Within the 35 kb LEE region are located also sep A, B, C and D genes which encode the transport proteins necessary for the membrane translocation of proteins involved in the attaching and effacing process. The LEE locus also encodes proteins of 100, 40, 39 and 25 kDa (Kenny & Finlay 1995). The 39 kDa protein resembles the enzyme glyceraldehyde-3-phosphate dehydrogenase, but the function of the remaining proteins is unknown, though they all are actively secreted by a type III secretory system (Jarvis & Kaper 1996). Patients infected with VTEC, carrying the LEE region, may produce serum antibodies to one or more of the secreted proteins, and these antibodies have been used to provide evidence of infection by VTEC (Chart et al. 1998).

Most strains of E. coli O157:H7 produce an enterohaemolysin of approximately 110 kDa (Beutin et al. 1989; Schmidt et al. 1995), which appears to be expressed during pathogenesis as patients have been reported to produce antibodies to this protein (Schmidt et al. 1995). Also, strains of E. coli belonging to serotype O157:H7 have been reported to express low molecular weight outer membrane proteins (OMPs) of 5 and 6 kDa (Levine et al. 1987; Padhye & Doyle 1991; Clark et al. 1995).

In the present study, 167 sera from patients with suspected VTEC infection were screened for antibodies to the LPS of a range of VTEC to evaluate more fully the dot immunoassay and to establish the prevalence of antibodies to the secreted antigens. Sera (34) from patients with serological evidence of infection with E. coli O157 were also used to determine the serum antibody response of patients to enterohaemolysin, the low molecular weight proteins (LMWPs) and also H = 7 flagellar proteins, to assess their antigenic properties and establish their potential as a basis for immunoassays suitable for diagnosing infections with VTEC.



Twelve strains of E. coli belonging to serogroups O5 (E41787), O26 (E36039), O55 (E40230), O104 (E32627), O105 (E43549), O111 (E52849), O115 (E47747), O128 (E41509), O145 (E38938), O153 (E31695), O163 (E31708) and O165 (E40235) were used for the preparation of LPS. Enteropathogenic E. coli strain E20513 (O111:H2) carried the eae A gene and was used to produce secreted proteins (Chart et al. 1998). Escherichia coli strain TPE 1302 is a strain of K12 (C600), carrying plasmid pEO40, encoding enterohaemolysin. Strain K12 C600 was used as an enterohaemolysin-negative control. Dr Lother Beutin (Robert Koch Institute, Berlin, Germany) provided strain TPE 1302. To ensure that plasmid carriage was maintained by strain TPE 1302, all media were supplemented with ampicillin (100 μg ml−1), and carriage was monitored by gel electrophoresis and DNA hybridization with a probe for part of the enterohaemolysin gene, CVD 419 (Levine et al. 1987). Escherichia coli strain E119317 (O = untypable: H = 7) was used to prepare flagellae. Twenty-one strains of O157:H7 VTEC were used for studies involving LMWPs. Six strains of non-VT-producing E. coli O157, with flagellar antigens H19 (two), H45 (two) and H8 (two), and expressing LPS profiles A, B and C, respectively (Chart et al. 1993a), were also used. Strains of E. coli were grown in 10 ml nutrient broth (37 °C, 8 h) prior to subculture onto nutrient agar and culture overnight (37 °C). For expression of flagellae, bacteria were subcultured from Craigie agar tubes onto nutrient agar (37 °C, 16 h). All bacteria were from the culture collection held by the Laboratory of Enteric Pathogens. Bacteria were maintained on Dorset's egg agar slopes.

Human sera

A total of 213 sera were used in this study; 167 were from patients and had been referred to the Laboratory of Enteric Pathogens (LEP) for screening for antibodies to the LPS of E. coli O157 (Chart et al. 1991). These sera had been referred for analysis because of a high clinical suspicion of VTEC infection, but they had been shown not to contain antibodies to the O157 LPS somatic antigen and were used to test for antibodies to the LPS of VTEC other than O157. These sera had originated from patients from England and Wales, and were received during 1996. From the available data, 57 patients were male and 30 were female.

Of the remaining 46 sera, 34 had been submitted to the LEP for routine E. coli O157 serology, and had been shown to contain antibodies to the LPS of E. coli O157. Of the 34 patients submitting sera, 12 were female and 10 were male; the sex of two patients was not known. Thirteen patients had symptoms of HUS and four had diarrhoea. Twelve sera were from apparently healthy adults and originated from the national blood transfusion laboratory, London, UK. All sera were stored at −10 °C.

Rabbit sera

Serum antibodies were produced to H = 7 flagellar antigens in a New Zealand white rabbit (Chart & Rowe 1992).

Preparation of LPS

For the preparation of whole-cell LPS, bacteria were digested with Proteinase K (Sigma; Chart et al. 1989). Bacteria were placed into pre-weighed, screw-capped Eppendorf tubes and suspended in SDS-PAGE solubilization buffer to give a cell concentration of 1 mg cells 30 μl−1 of solubilization buffer. The suspensions were incubated at 100 °C for 10 min to denature proteins and facilitate subsequent proteolysis. Heat-denatured bacteria (50 μl) were mixed with 50 μl SDS-PAGE solubilization buffer containing 100 μg Proteinase K prior to incubation at 60 °C (60 min). The resultant preparations were used for both SDS-PAGE and the dot immunoassay.

Preparation of secreted proteins

Bacteria were sedimented by centrifugation (5000 g, 4 °C, 30 min), supernatant fluids were millipore-filtered (0·45 μm) and bacterial proteins were concentrated by acetone precipitation (Chart et al. 1998).

Dot immunoassay

For the dot immunoassay, 3 μl Proteinase-K-digested preparation was diluted with 1·5 ml PBS and 25 μl volumes were applied to nitrocellulose sheets using a 96-well microfiltration apparatus (Bio-Rad, Hemel Hempstead, UK; Chart & Rowe 1997). The nitrocellulose sheets were dried (37 °C, 2 h) and cut into eight strips containing the LPS prepared from all 12 VTEC strains. Cutting the left-hand-side of the nitrocellulose paper with pinking shears facilitated the subsequent orientation of the strips. Each strip was ‘blocked’ with 5 ml skimmed milk solution (3% (w/v) in phosphate-buffered saline (PBS)) and mixed with 30 μl serum in 5 ml skimmed milk solution (1 h). For the routine screening of sera, antigen-antibody complexes were detected with an alkaline phosphatase-conjugated antihuman polyvalent antiserum (Sigma) and an enzyme substrate buffer comprising 20 ml 0·1 mol l−1 Tris, 0·09 mol l−1 NaCl and 0·15 mol l−1 MgCl2 containing 90 μl nitroblue tetrazolium (Sigma) solution (75 mg ml−1 in 70% (v/v) aqueous dimethyl formamide) and 70 μl 5-bromo-4-chloro-3-indolylphosphate (Sigma) solution (50 mg ml−1 deionized water).


SDS-PAGE was carried out based on the method of Laemmli (1970) as described by Chart et al. (1989). Lipopolysaccharide, prepared from whole cells by Proteinase K digestion (see above), was applied to gels comprising a 4·5% (w/v) acrylamide stacking gel and a 12·5% (w/v) separation gel. Electrophoresis was performed using an Atto mini-gel system (Genetic Research Instrumentation Ltd, Braintree, UK) with a constant current of 50 mAmp for 0·5 h. Gels were either stained with silver for LPS, or used for immunoblotting.

Gel staining

SDS-PAGE profiles of LPS prepared by Proteinase K digestion of whole bacteria were stained with a silver stain specific for carbohydrate (Tsai & Frasch 1982; Hitchcock & Brown 1983).


SDS-PAGE LPS profiles were transferred onto nitrocellulose membranes using the method of Towbin & Gordon (1984). Individual LPS profiles were prepared by cutting the nitrocellulose sheets into strips with pinking shears to enable realignment. Each nitrocellulose strip was blocked in 5 ml 3% (w/v) skimmed milk in PBS and reacted with a patient's serum (30 μl lane−1) for 1 h at room temperature. Following washing (×3) with PBS-Tween (PBS containing 0·05% (v/v) Tween-20), antibody-antigen complexes were detected using a goat anti-human polyvalent antibody conjugated with alkaline phosphatase (Sigma) 5 μl in 5 ml skimmed milk-PBS for 1 h. Bound alkaline phosphatase was detected using an enzyme substrate buffer as described (above) for the dot immunoassay. Antigen-antibody complexes were detected with a peroxidase-conjugated antihuman polyvalent antiserum (Sigma) in association with an ECL luminescence kit (Amersham, UK) and autoradiography.


Intracellular enterohaemolysin was extracted from strain TPE 1302 by sonic disruption of whole bacteria (Heat Systems W-225, Ultrasonics Inc.; Life Sciences Laboratories, Luton, UK). Bacterial debris was sedimented by centrifugation (5000 g, 15 min, 4 °C) and the supernatant fluid sterilized by millipore filtration (0·45 μm; Chart et al. 1998).

Bacterial outer membranes

Outer membranes (OMs) were prepared from sonicated bacteria by the selective solubilization of inner membrane proteins with Sarkosyl (Merck; Chart & Griffiths 1985).

Heated extracts

Bacteria suspended in 0·5 ml PBS were incubated at 37, 45, 55 or 60 °C for 30 min prior to the sedimentation of bacteria (12 500 g, 10 min) and the removal of supernatant fluid. Protein concentration was determined using the Lowry method.

Electron microscopy

Strains of E. coli were examined for surface structures by transmission electron microscopy (Chart et al. 1995) following staining with 2% (w/v) aqueous ammonium molybdate.

Plasmid profiling

Plasmid DNA was prepared by the alkaline extraction method of Birnboim & Doly (1979), and separated on 0·7% (w/v) agarose gels with a borate buffer (100 V, 4 h). The DNA was stained with ethidium bromide and viewed under ultra-violet light (260 nm).

DNA hybridization

Bacteria were examined for the ability to hybridize with enterohaemolysin probe CVD 419 probe, a 3·4 kb HindIII fragment prepared from a plasmid carried by E. coli O157:H7 (Takeda et al. 1979; Levine et al. 1987).


Sera from 167 patients were reacted with LPS purified from 12 strains of VTEC using a dot immunoassay. A total of 12 reacted with LPS from E. coli 05 (one), 026 (two), 0115 (two), 0145 (one), 0163 (one) or 0165 (five) (Table 1). The antibody binding reactions observed with SDS-PAGE and immunoblotting confirmed these sera. The results in Fig. 1 show patients’ antibodies binding to the LPS of E. coli O5 (lane 1), O26 (lane 2), O115 (lane 3), O145 (lane 4), O163 (lane 5) and O165 (lane 6). For comparison, profiles of these LPSs are shown in Fig. 2,(E). coli O5 (lane 1), O26 (lane 2), O115 (lane 3), O145 (lane 4), O163 (lane 5) and O165 (lane 6). For 22 patients with HUS, serum antibodies were not detected to any of the 12 LPS preparations and the cause of disease in these patients remains unknown.

Table 1.  Patients’ details for sera containing antibodies to the LPS of VTEC
LPS serogroupPatients
Age (years)
  1. na, Not available.

O1651renal failure
Figure 1.

Twelve of 167 sera were found to react with the LPS of Escherichia coli 05 (lane 1), 026 (lane 2), 0115 (lane 3), 0145 (lane 4), 0163 (lane 5) or 0165 (lane 6). Each profile contained LPS prepared from 500 μg cell mass and each SDS-PAGE profile was reacted with 30 μl of patient's serum

Figure 2.

For comparison with the results in Fig. 1, silver-stained LPS profiles are provided for Escherichia coli O5 (lane 1), O26 (lane 2), O115 (lane 3), O145 (lane 4), O163 (lane 5) and O165 (lane 6). Each profile contained LPS prepared from 500 μg bacterial cell mass

The sera with antibodies to the LPS of E. coli O26 and O145 also contained antibodies to the 100 kDa and 40 kDa secreted proteins; sera with antibodies to the other LPS types did not contain antibodies to the secreted proteins.

To assess the human serum antibody response to enterohaemolysin, E. coli strain TPE 1302 was used. SDS-PAGE analysis of sonicated extracts prepared from E. coli strains TPE 1302 demonstrated a protein of 110 kDa, not observed in extracts made from E. coli K12 C600. This protein was also present in OMs prepared from E. coli strains TPE 1302 (Fig. 3, lane 2) but absent from OMs prepared from E. coli strain K12 C600 (Fig. 3, lane 1). Outer membrane protein (OMP) profiles prepared from E. coli TPE 1302 were used to screen patients’ sera for antibodies to the enterohaemolysin. Only three patients’ sera contained antibodies to the 110 kDa OMP (Fig. 3, lane 3). These patients comprised two females and one male; one of the females had a diarrhoeal illness but none of the three had symptoms of HUS. The 12 control sera did not contain antibodies to this protein (Fig. 3, lane 4).

Figure 3.

Sonicated extracts prepared from Escherichia coli strains TPE 1302 (lane 2) contained a protein of 110 kDa (arrowed), not observed in extracts made from E. coli K12 C600 (lane 1). Three patients had serum antibodies to 110 kDa enterohaemolysin (lane 3); the 12 control sera did not (lane 4)

Strains of E. coli O157:H7 were examined for OMPs of 5 and 6 kDa (Padhye & Doyle 1991) as a source of antigens to determine the human serum antibody response to these LMWPs. Outer membranes prepared from strains of E. coli O157:H7 were found not to contain OMPs of 5 or 6 kDa. However, heated extracts prepared from the 21 strains were found to contain proteins of 5 and 6 kDa (Fig. 4). Strains of E. coli O157:H19 (LPS profile A; Chart et al. 1993a), O157:H45 (LPS profile B; Chart et al. 1993a) and O157:H8 (LPS profile C; Chart et al. 1993a) did not express the 5 and 6 kDa proteins.

Figure 4.

LMWPs were not extracted from bacterial cells by incubation at 37 °C (lane 1) or 45 °C (lane 2), and only trace amounts of these proteins were detected following incubation at 55 °C (lane 3). Bacteria incubated at 60 °C for 30 min released observably more of LMWs of 22, 6 and 5 kDa (lane 4). Flagellar protein subunits of 66 kDa were released at all four incubation temperatures. Molecular weight standards (lane 5) represent proteins of 94·0, 66·2, 45·0, 31·0, 21·5 and 14·4 kDa

Further studies with the LMWPs showed that incubation at 37 and 45 °C (Fig. 4, lanes 1 and 2) did not extract these proteins from the bacterial cell surface, and that only trace amounts of LMWPs were detected following incubation at 55 °C (Fig. 4, lane 3). However, bacteria incubated at 60 °C for 30 min released these proteins from the bacterial cell surface (Fig. 2, lane 4). Incubation at 60 °C also released a 22 kDa protein from the cell surface, and a 66 kDa protein was detected in supernatant fluids from bacteria incubated at temperatures ranging from 37 to 60 °C (Fig. 4). Bacteria expressing the LMWPs were examined by electron microscopy, but the only surface structures detected were flagella. Immunoblotting with the rabbit serum prepared to H = 7 flagella showed that the 66 kDa protein was flagella protein subunits. SDS-PAGE profiles containing flagella protein and the 5, 6 and 22 kDa proteins were reacted with patients’ sera; only one patient's serum was found to contain antibodies to H = 7 flagella antigens. This patient was female and did not have HUS or a diarrhoeal illness. Antibodies to the LMWPs were not detected in patient or control sera.


In the present study, we evaluated a simple dot immunoassay designed for the rapid screening of patients’ sera for antibodies to the LPS of VTEC associated with human infections. SDS-PAGE and immunoblotting confirmed the results of the dot immunoassay. The study was facilitated by the fact that the LPS preparations could be used for both the dot immunoassay and SDS-PAGE/immunoblotting, and avoided the need to prepare purified LPS by more complicated methods such as the hot-phenol extraction procedure. Of the 167 sera tested, antibodies specific for the LPS of serogroups 05, 026, 0115, 0145, 0163 and 0165 were detected in 12.

Previous studies have identified antibodies to the LPS of E. coli 05, 0115, 0145, 0153 and 0165 in 19% of sera examined, with antibodies to E. coli 05 being most frequently detected (Chart & Rowe 1990). A later study (Chart et al. 1996) identified only three patients with serum antibodies to the LPS of E. coli O5. In the present study, antibodies specific for a range of VTEC LPS types were detected in about 7% of sera, which correlated with previous studies from this laboratory (Chart & Rowe 1990).

Only three sera with antibodies to the LPS of O26 and O145 contained antibodies to the 100 kDa and 40 kDa secreted proteins. Patients with serum antibodies to the LPS of O145 have previously been shown to produce antibodies to these secreted proteins (Chart et al. 1998). Twenty-two patients with HUS were found not to have serum antibodies to any of the 12 LPS preparations, and the role of VTEC in these cases remains unknown.

From this part of the study, we concluded that a simple dot immunoassay provided a rapid means of screening sera for antibodies to a limited range of VTEC LPSs, and that results of dot blotting could be confirmed by SDS-PAGE and immunoblotting. Furthermore, the screening of patients’ sera for antibodies to the LPS from a panel of VTEC provides valuable information for the surveillance of infections caused by VTEC other than those belonging to E. coli O157. The value of screening sera for antibodies to the secreted proteins needs to be evaluated further.

The second part of the study was to determine whether patients suspected of having been infected with E. coli O157 expressed antibodies to enterohaemolysin, LMWPs or H = 7 flagella proteins. In a study by Schmidt et al. (1995), 19 of 20 patients with HUS were reported to produce antibodies to enterohaemolysin (Schmidt et al. 1995). In the present study, only three patients’ sera out of a total of 70 contained antibodies to enterohaemolysin. Whether this was due to the enterohaemolysin being poorly antigenic, or whether it was because this toxin was not expressed in vivo, remains unknown. However, it was concluded that enterohaemolysin was not a suitable antigen for providing serological evidence of infection with E. coli O157.

Proteins of 5 and 6 kDa were identified in heated extracts but not in OM preparations as described by Padhye & Doyle (1991). This was thought to be due to the fact that Padhye and Doyle sedimented OMs using 131 000 g for 36 h, which would sediment the 5 and 6 kDa proteins along with OMs. In the present study, OMs were sedimented with 40 000 g for 1 h. The LMWPs were not expressed by strains of E. coli O157 other than those with H = 7 flagella antigens, suggesting that proteins of 5 and 6 kDa were only expressed by VTEC belonging to serotype O157:H7. The LMWPs were extracted maximally by incubating bacteria at 60 °C for 30 min, a regime employed for the extraction of flagella and fimbriae (Chart et al. 1995). The fact that incubation at 45 °C did not cause these proteins to be released from the cell surface suggested that the 5, 6 and 22 kDa proteins had a strong association with the bacterial outer membrane. It should be noted that flagella were dislodged from the cell wall by incubation at temperatures as low as 37 °C, indicating that the LMWPs were attached to the cell surface more avidly than flagella.

Serum antibodies to the LMWPs were not detected in sera with antibodies to the LPS of E. coli O157. This may be due to these proteins not being expressed during pathogenesis, or to the proteins being only poorly antigenic. These proteins have a comparatively low molecular weight, and it is not unusual for E. coli proteins of small molecular size to be poorly antigenic, as is the case for E. coli heat-stable toxin (Takeda et al. 1979). The LMWPs remain to be elucidated fully. The low molecular proteins described by Padhye & Doyle (1991) had initially been detected using a monoclonal antibody (4E8C12). However, Clark et al. (1995) subsequently showed that monoclonal antibody 4E8C12 recognized epitopes on LPS molecules. The LMWPs described here cannot be LPS as they stained with Coomassie brilliant blue. The present studies describe LMWPs which are located on the bacterial cell surface but are not fimbrial structures. It was concluded that these proteins were not suitable as a basis in immunoassay for detecting antibodies in patients infected with VTEC. Similarly, as only one patient expressed antibodies to flagellar proteins, O157 LPS remains the antigen of choice for immunoassays designed to detect serological evidence of infection with E. coli O157.


The authors acknowledge the Department of Health for support for part of this study.