Release of Shiga toxin from Shigella dysenteriae 1 by polymyxin B

Three Dimensional Vero Cell-Platform for Rapid and Sensitive Screening of Shiga-Toxin Producing Escherichia coli

Shiga-toxin producing Escherichia coli (STEC) is a serious public health concern. Current Vero cell assay, although sensitive, is lengthy and requires 48-72 h to assess STEC presence in a sample. In this study, we investigated if Vero cells in a three-dimensional (3D) platform would provide improved sensitivity for rapid screening of STEC. Vero cells (epithelial kidney cell line) were grown as a monolayer (2D) or in a collagen-matrix (3D) and exposed to Shiga-toxin (Stx) preparation or STEC cells that were pre-exposed to antibiotics (mitomycin C, ciprofloxacin, or polymyxin B) for toxin induction. Lactate dehydrogenase (LDH) release from Vero cells was used as a biomarker for cytotoxicity. Modified tryptic soy broth (mTSB) as enrichment broth containing mitomycin C (2 μg/ml) or ciprofloxacin (100 ng/ml) significantly induced Stx production, which was further confirmed by the dot-immunoblot assay. The 3D Vero platform detected STEC after 6 h post-infection with cytotoxicity values ranging from 33 to 79%, which is considerably faster than the traditional 2D platform, when tested with STEC. The cytotoxicity for non-Stx producing bacteria, Salmonella, Listeria, Citrobacter, Serratia, and Hafnia was found to be below the cytotoxicity cutoff value of 15%. The detection limit for the 3D Vero cell assay was estimated to be 107 CFU/ml for bacteria and about 32 ng/ml for Stx in 6 h. STEC-inoculated ground beef samples (n = 27) resulted in 38-46% cytotoxicity, and the bacterial isolates (n = 42) from ground beef samples were further confirmed to be stx1 and stx2 positive in a multiplex PCR yielding a very low false-positive result. This 3D cell-based screening assay relies on mammalian cell pathogen interaction that can complement other molecular techniques for the detection of cell-free Stx or STEC cells from food samples for early detection and prevention.

Shigella dysenteriae type 1 toxin induced lipid peroxidation in enterocytes isolated from rabbit ileum

To evaluate the role of reactive oxygen species (ROS) in Shigella dysenteriae 1 toxin (STx) mediated intestinal infection, the ligated rabbit small intestinal loops were injected with STx. The enterocytes isolated from STx treated rabbit ileal loops had a significantly higher level of lipid peroxidation as compared to enterocytes isolated from control rabbit ileum. To study the role of second messengers in STx mediated intestinal damage, the in vivo and in vitro effects of modulators of lipid peroxidation of enterocytes were used. The presence of Ca2+-ionophore A23187 enhanced the extent of lipid peroxidation in enterocytes isolated from the control and STx treated rabbit ileum. However, 1-verapamil only marginally decreased the lipid peroxidation level of enterocytes isolated from STx treated rabbit ileum. The in vitro effect of modulators was in agreement with in vivo studies. Dantrolene significantly decreased the extent of lipid peroxidation of enterocytes isolated from STx treated rabbit ileum. PMA significantly increased the lipid peroxidation level of enterocytes isolated from control ileum. However, PMA could not further enhance the lipid peroxidation level of enterocytes isolated from STx treated rabbit ileum. The presence of H-7 significantly decreased the extent of lipid peroxidation of enterocytes isolated from STx treated rabbit ileum. In vitro effect of PMA and H-7 was in agreement with that of in vivo findings. The role of arachidonic acid metabolites, prostaglandins (PGs), in mediating STx induced lipid peroxidation was also studied. The presence of indomethacin (a PG synthesis inhibitor) significantly decreased the lipid peroxidation induced by STx. These findings suggest that lipid peroxidation induced by STx is mediated through cytosolic calcium. The increase in (Ca2+)i leads to activation of PKC. A significant decrease in the enterocyte levels of antioxidant enzymes superoxide dismutase, catalase and reduced glutathione in STx treated rabbit ileum as compared to control was seen. A significant decrease in vitamin E levels was also observed. This suggests that there is decreased endogenous intestinal protection against ROS in STx mediated intestinal infection which could contribute to enterocyte membrane damage that ultimately leads to changes in membrane permeability and thus to fluid secretion.

Calcium and protein kinase C play a significant role in response to Shigella toxin in rabbit ileum both in vivo and in vitro

The role of second messengers in Shigella toxin (STx) induced fluid secretion in rabbit ileum was evaluated. In vivo and in vitro studies were carried out in presence or absence of following modulators: Ca2+ ionophore A23187 (15 microM), l-verapamil (200 microM), phorbol-12-myristate-13-acetate (PMA, 200 ng), 1-(5-isoquinolinyl-sulphonyl)-2-methyl-piperazine (H-7, 15 microg) and indomethacin (20 microM). In in vivo studies, the fluid accumulation into rabbit ileal loops in response to STx was measured in presence or absence of these modulators. In in vitro studies, unidirectional fluxes of Na+ and Cl- were carried out in presence or absence of these modulators. The addition of Ca2+ ionophore A23187 along with STx further increases the amount of fluid already induced by STx. Whereas the presence of l-verapamil along with STx did not decrease the amount of fluid induced by STx. In vitro findings were in consonance with the in vivo studies. A significant increase in inositol triphosphate (IP3) levels was observed in enterocytes isolated from STx treated rabbit ileum. The addition of PMA into rabbit ileal loops in presence of STx mimicked the effect of STx while the presence of H-7 reversed the secretion caused by STx to absorption. Similar results were obtained while determining unidirectional fluxes of Na+ and Cl- in presence of PMA and also with H-7. A significant increase in PKC levels was observed in the membrane fraction of enterocytes isolated from STx treated rabbit ileum as compared to control. Further a marked decrease in PKC levels was observed in the presence of H-7 in membrane fraction of enterocytes isolated from STx treated rabbit ileum. The addition of indomethacin into rabbit ileal loops reversed the secretion (caused by STx) to absorption. In vitro findings were in consonance with in vivo studies. Besides, there was a significant increase in PG-E levels in enterocytes isolated from STx treated rabbit ileum as compared to control. These findings suggested that STx induced enteritis involves the role of PKC, intracellular calcium stores and prostaglandins. The extracellular calcium pool probably does not play a significant role in this process.

Evidence that the A2 fragment of Shiga-like toxin type I is required for holotoxin integrity

Escherichia coli Shiga-like toxin type I (SLT-I) is a potent cytotoxin consisting of an enzymatically active A subunit and a pentameric B subunit that mediates toxin binding to susceptible eukaryotic cells. Evidence that the carboxy-terminal 38 amino acids of the A subunit are involved in holotoxin 1A:5B association is presented. We compared the ability of purified recombinant SLT-I B subunit (Slt-IB) to combine in vitro with purified recombinant SLT-I A subunit (Slt-IA; full-length subunit A includes amino acids 1 to 293) and its ability to combine with purified recombinant SLT-I A1 subunit (Slt-IA1; truncated subunit A includes amino acids 1 to 255). Each mixture was analyzed for biological and physical evidence of toxin assembly. Although Slt-IA successfully combined with Slt-IB to form a molecular species similar to holotoxin that was detectable by nondenaturing polyacrylamide gel electrophoresis and immunoblotting and yielded a molecule which was cytotoxic to cultured Vero cells, Slt-IA1 did not have this ability. Slt-IA1 was 36-fold more active than Slt-IA in an in vitro protein synthesis inhibition assay. These findings suggest that the Slt-IA2 fragment is crucial for formation of SLT holotoxin and stabilizes the interaction between the A and B subunits.

Stimulation of gastrointestinal antibody to Shiga toxin by orogastric immunization in mice

Shiga toxin (ST) is a protein toxin of Shigella dysenteriae type 1, a causative agent of severe diarrhoea and dysentery. In this report we describe the gastrointestinal secretory antibody response of mice following orogastric immunization with ST. Gastrointestinal secretions were sampled by a gastrointestinal lavage technique weekly for 5 weeks after initial immunization. Assay of lavage samples by ELISA showed that mice vaccinated orogastrically with various doses of ST developed gastrointestinal antibody to ST in a dose-dependent manner. Serum anti-ST activity developed by 5 weeks after initial immunization. The ability of ST to act as a mucosal immune adjuvant was investigated by coadministration of ST and keyhole limpet haemocyanin. In contrast to cholera toxin, a potent adjuvant, ST did not demonstrate adjuvant activity. The mouse gastrointestinal lavage model could be useful for further analysis of the cellular basis of ST immunogenicity.

Pathogenesis of Providencia alcalifaciens-induced diarrhea

Providencia alcalifaciens is a member of the family Enterobacteriaceae. There are reports that P. alcalifaciens can cause diarrhea, but the mechanism(s) by which it causes diarrhea is known. We studied P. alcalifaciens isolated from a child and two adults with diarrhea for enteropathogenicity. The three isolates did not exhibit any characteristic adherence to cultured HEp-2 cell monolayers, and they did not produce enterotoxins, cytotoxins, or keratoconjunctivitis in the Sereny test. Two isolates invaded cultured HEp-2 cell monolayers, producing localized bacterial clusters and actin condensation. The pattern of actin condensation was different from that produced by enteropathogenic Escherichia coli but similar to that produced by Shigella flexneri. Invasion and actin condensation were poor for the third isolate. Histology of adult rabbit small intestinal loops inoculated with all three isolates revealed bacterial attachment to, penetration of, and microulcer formation on the surface epithelium and hyperemia, edema, and polymorphonuclear cell infiltration of lamina propria. All the isolates produced diarrhea in rabbits with removable intestinal ties, and some of these rabbits developed hindlimb paralysis. Intestinal histology of the rabbits with removable intestinal ties which developed diarrhea showed changes similar to that in adult rabbits on which ileal loop assays had been performed. Transmission electron microscopy of intestinal tissues also confirmed tissue penetration by the isolates. Nerve tissue histology of two rabbits that developed hindlimb paralysis showed focal mononuclear cell infiltration around peripheral nerve sheaths. It is concluded that some strains of P. alcalifaciens are enteropathogenic and that they cause diarrhea by invading the intestinal mucosal epithelium. However, the relevance to human disease of the hindlimb paralysis observed in this animal model is not clear.

Hafnia alvei, a probable cause of diarrhea in humans

Hafnia alvei, a member of the family Enterobacteriaceae, was the only species of bacteria cultured from the stool of a 9-month-old child who was admitted with a 3-day history of watery diarrhea. The isolated strain of H. alvei failed to produce heat-labile or heat-stable enterotoxins or Shiga-like toxin I or II and did not invade HeLa cells, nor did it cause keratoconjunctivitis (determined by the Sereny test) in a guinea pig's eye. The strain, however, induced diarrhea in 8 of 12 adult rabbits with removable intestinal ties (removable intestinal tie-adult rabbit diarrhea [RITARD] assay) and in 1 of 2 orally fed animals. No diarrhea could be induced with Escherichia coli K-12 in eight RITARD assay rabbits and three orally fed rabbits, respectively. Microscopic examination of affected animals revealed moderate inflammatory cellular infiltration of the intestinal mucosa, in which bacterial attachment to the surface epithelium and loss of the microvillus border were evident in the ileum and colon. Electron microscopy demonstrated cellular modifications of the apical surface, with cupping or pedestal formation and increased terminal web density at sites of bacterial "attachment-effacement," a well-known characteristic and mechanism of diarrhea of enteropathogenic E. coli. Identical lesions were also induced by H. alvei in rabbit ileal loops, which ruled out naturally occurring rabbit enteropathogenic E. coli strains, which are known to produce similar lesions. It is concluded that at least some strains of H. alvei have the potential to cause diarrhea and that attachment-effacement is a virulence characteristic shared by bacteria other than E. coli.

Purification and characterization of an Escherichia coli Shiga-like toxin II variant

A Shiga-like toxin II variant was purified to homogeneity from Escherichia coli TB1(pCG6), which contained the toxin genes cloned in multicopy plasmid pUC18. The purification scheme involved polymyxin B extraction of the toxin from bacterial cells, followed by differential (NH4)2SO4 precipitation, anion- and cation-exchange fast-protein liquid chromatography, and immunoaffinity exclusion chromatography. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis of purified toxin revealed three protein bands that migrated with calculated molecular weights of 33,000, 27,500, and 7,500. These bands correspond to values for the A, A1, and B subunits, respectively, that would be expected on the basis of the nucleotide sequence and comparison with data for Shiga toxin and other Shiga-like toxins. Electrophoresis under nonreducing conditions resulted in disappearance of the 27,000-molecular-weight band. Western blot (immunoblot) analysis revealed three protein bands with molecular weights of 33,000, 27,500, and 7,500. The purified toxin induced typical signs of edema disease in pigs injected intravenously with doses as small as 3 ng/kg of body weight. The 50% cytotoxic doses for Vero, PK15, and Madin-Darby bovine and canine kidney cells were 0.5, 2.0, 8.0, and 8.0 pg, respectively. The 50% lethal dose of purified toxin for mice was 0.9 pg by the intraperitoneal route. Approximately 75 micrograms of purified toxin was required to induce a 1-ml/cm fluid response in rabbit ileal loops. Antiserum to the Shiga-like toxin II variant neutralized homologous toxin, Shiga-like toxin II, and Verotoxin 2 but not Shiga-like toxin I.

Evidence that glutamic acid 167 is an active-site residue of Shiga-like toxin I

Escherichia coli Shiga-like toxin I, a close relative of Shiga toxin and a distant relative of the ricin family of plant toxins, inhibits eukaryotic protein synthesis by catalyzing the depurination of adenosine 4324 in 28S rRNA. By comparing the crystallographic structure of ricin with amino acids conserved between the Shiga and ricin toxin families, we identified seven potential active-site residues of Shiga-like toxin I. The structural gene encoding Shiga-like toxin I A chain (Slt-IA), the enzymatically active subunit, was engineered for high expression in E. coli. Oligonucleotide-directed mutagenesis of the gene for Slt-IA was used to change glutamic acid 167 to aspartic acid. As measured by an in vitro assay for inhibition of protein synthesis, the specific activity of mutant Slt-IA was decreased by a factor of 1000 compared to wild-type Slt-IA. Immunoblots showed that mutant and wild-type Slt-IA were synthesized as full-length proteins and were processed correctly by signal peptidase. Both proteins were equally susceptible to trypsin digestion, suggesting that the amino acid substitution did not produce a major alteration in Slt-IA conformation. We conclude that glutamic acid 167 is critical for activity of the Shiga-like toxin I A chain and may be located at the active site.

Effects of iron and temperature on Shiga-like toxin I production by Escherichia coli

Iron is known to depress Shiga toxin production by Shigella dysenteriae 1, and temperature has been shown to regulate several genes required for Shigella invasiveness. In this study, the influence of iron and temperature on regulation of a highly related toxin, Shiga-like toxin I (SLT-I) of enterohemorrhagic Escherichia coli, was examined in strains lysogenic for the toxin-converting coliphage 933J and in strains carrying the cloned slt-I genes on a high-copy-number plasmid vector. For comparison, S. dysenteriae 1 was included in these studies. As expected, iron suppressed Shiga toxin synthesis, and reduced growth temperature was also found to decrease Shiga toxin production. Iron also suppressed SLT-I synthesis in E. coli lysogenized with phage 933J but did not demonstrably repress toxin synthesis in E. coli strains carrying the cloned slt-I genes. Temperature had no effect on SLT-I synthesis. Mini-Mu lac operon fusions were then isolated in the cloned slt-I genes and used to test for regulation of beta-galactosidase by iron. Iron did not decrease beta-galactosidase production in strains that harbored these operon fusion plasmids. Taken together, these results indicate that iron but not temperature represses SLT-I synthesis when the slt-I genes are phage associated but this suppression is not easily demonstrated when the slt-I genes are cloned on a high-copy-number plasmid.

Cloning of genes for production of Escherichia coli Shiga-like toxin type II

Genes controlling production of Shiga-like toxin type II (SLT-II) in Escherichia coli were cloned from the SLT-II-converting bacteriophage 933W and compared with the Shiga-like toxin type I (SLT-I) genes previously isolated and described from phage 933J. Subcloning analysis identified a region within the 4.9-kilobase EcoRI fragment of phage 933W that was associated with SLT-II production. Experiments with E. coli minicells containing these subclones demonstrated that the 4.9-kilobase EcoRI fragment encodes the structural genes for SLT-II. These experiments additionally showed the genetic organization of the SLT-II genes to be the same as that of the SLT-I genes, with the coding sequence for the large A subunit adjacent to that for the smaller B subunit. The mobilities of the SLT-II subunits in sodium dodecyl sulfate-polyacrylamide gels were slightly greater than those determined for the SLT-I subunits. Although apparent processing of the SLT-I subunits was observed with polymyxin B treatment of the labeled minicells, no processing of the SLT-II subunits was detected. Southern blot hybridization studies suggested that the DNA fragment carrying the SLT-II structural genes shares approximately 50 to 60% homology with the DNA of the SLT-I structural genes.

Detection of Shiga-like (SL) toxins of enteropathogenic Escherichia coli (EPEC) of human, porcine, calf, and lamb origin on Vero and HeLa S3 cells: a comparative study

One hundred and forty eight strains of human, porcine, calf and lamb origin belonging to different enteropathogenic O:H serotypes isolated in 13 countries in four continents were tested for production of Shiga, Shiga-like (SL) and other cytotoxins on Vero cells and HeLa (S3 subline) cells in tissue cultures. Altogether, 45% human strains and 89% porcine strains were defined as strong toxin producers (toxin titre greater than or equal to 1:100) on Vero or HeLa S3 cells while 31% of human and 9% porcine strains were regarded as moderate to weak toxin producers (toxin titre less than 1:100). Twenty three percent of human and 1.5% of porcine strains were negative for Shiga or SL-toxin. Polymyxin release of Shiga or SL toxins from bacterial colonies of blood agar grown cultures is recommended as it is simple and effective method facilitating the detection of even low levels of toxins in EPEC or non-EPEC strains. Of the ten strains from calves and lambs, only four were strong toxin producers when cell-free culture supernatants were tested while a polymyxin release method showed that 8 strains were strong toxin producers. One strain was negative by both methods. The high proportion of Shiga/SL toxin negative strains in all O:H serotypes of human origin (but not of porcine origin, especially O 139 serogroup) suggests that systematic studies should continue to look for new toxins in freshly isolated strains grown under in vivo like conditions, e.g. in iron depleted culture media.

Sensitive method for detecting low numbers of verotoxin-producing Escherichia coli in mixed cultures by use of colony sweeps and polymyxin extraction of verotoxin

High titers of Verotoxin (VT) were released from cell pellets of VT-producing Escherichia coli (VTEC; corresponding to E. coli strains producing "high" levels of Shiga-like toxin) after incubation in polymyxin B (0.1 mg/ml) for 30 min at 37 degrees C. Maximal titers of polymyxin-releasable VT occurred in cells obtained from 5-h Penassay broth cultures and were up to eightfold higher than the peak culture supernatant VT titers which occurred in 8-h cultures. Polymyxin-releasable cell extracts of 5-h broth cultures inoculated with mixtures of VT-positive (VT+) and VT-negative strains had easily detectable VT titers when the proportion of VT+ cells in the mixture was about 1.0%, but culture supernatants were negative for VT even when this proportion was 20%. The results were the same whether the initial inoculum consisted of broth culture mixtures of VT+ and VT-negative strains or colony sweeps (loopfuls of confluent bacterial growth) taken from solid plate media previously inoculated with the broth mixtures. In a clinical study, 80 stool cultures from patients with hemolytic uremic syndrome and family contacts with diarrhea were tested for free fecal VT, VT in polymyxin extracts of colony sweeps (VT/PECS), and VTEC (examination of 20 separate E. coli colonies from primary media for VT production). Of the 80 samples, 40 were positive for at least one of these three tests; all 40 were positive for free fecal VT, and 20 of these were positive for VT/PECS. VTEC (as few as 1 colony out of 20) were only isolated from 14 of the 20 cultures that were positive for VT/PECS. In six cases, the VT/PECS was positive even when none of 20 colonies tested were VT+, suggesting that the procedure was able to detect a proportion of VTEC that was less than one in 20(5%). We conclude that the VT/PECS method is highly sensitive for detecting low concentrations of VTEC in stools and provides a rapid method for screening out stools that are negative for VTEC. The technique should also be of value in epidemiological studies for detecting low numbers of VTEC in animal feces, foods, and environmental samples.

Monoclonal antibodies to leukotoxin of Actinobacillus actinomycetemcomitans

Hybridoma cell lines which produce monoclonal antibodies to a leukotoxin from Actinobacillus actinomycetemcomitans were prepared. The monoclonal antibodies were selected for their ability to neutralize the cytotoxic activity of the leukotoxin and recognize the toxin on nitrocellulose blots. The antibodies belonged to either the immunoglobulin G1 (IgG1) or IgG2 subclass and differed in their ability to bind to the leukotoxin on nitrocellulose blots. However, only slight differences in neutralization titers were observed. Use of the monoclonal antibodies revealed that polymyxin B-extracted or osmotic shock-released leukotoxin could be separated into several high-molecular-weight polypeptides by sodium dodecyl sulfate-polyacrylamide gel electrophoresis. Immunoblot analysis with the monoclonal antibodies also demonstrated that the leukotoxin was present in eight oral strains of A. actinomycetemcomitans that had been previously classified by a biological assay as leukotoxic. The availability of these monoclonal antibodies should facilitate and expand studies concerning the role of the leukotoxin in the pathogenicity of A. actinomycetemcomitans.

Extraction and isolation of a leukotoxin from Actinobacillus actinomycetemcomitans with polymyxin B

A leukotoxin from Actinobacillus actinomycetemcomitans was isolated by a procedure that includes polymyxin B extraction, ion-exchange chromatography, and gel filtration chromatography. The procedure resulted in the recovery of 48% of the toxin with a 99-fold increase in specific activity. The isolated toxin has a molecular mass of 180,000 daltons by gel filtration and 115,000 daltons by sodium dodecyl sulfate-polyacrylamide gel electrophoresis. It retains all the major biological characteristics previously documented for crude leukotoxin preparations, including susceptibility to heat and proteolytic enzymes and neutralization by sera from patients with juvenile periodontitis. The isolated leukotoxin destroys human but not rat or guinea pig polymorphonuclear leukocytes and has no apparent effect on human erythrocytes. The availability of the A. actinomycetemcomitans leukotoxin should facilitate studies on its chemistry and mode of action as well as its role in the pathogenesis of human periodontal disease.