HAL Id: hal-00902567https://hal.archives-ouvertes.fr/hal-00902567

Submitted on 1 Jan 1999

HAL is a multi-disciplinary open accessarchive for the deposit and dissemination of sci-entific research documents, whether they are pub-lished or not. The documents may come fromteaching and research institutions in France orabroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, estdestinée au dépôt et à la diffusion de documentsscientifiques de niveau recherche, publiés ou non,émanant des établissements d’enseignement et derecherche français ou étrangers, des laboratoirespublics ou privés.

Necrotoxic Escherichia coli (NTEC): two emergingcategories of human and animal pathogens

Jean De Rycke, Alain Milon., Eric Oswald

To cite this version:Jean De Rycke, Alain Milon., Eric Oswald. Necrotoxic Escherichia coli (NTEC): two emerging cate-gories of human and animal pathogens. Veterinary Research, BioMed Central, 1999, 30 (2-3), pp.221-233. <hal-00902567>

Review article

Necrotoxic Escherichia coli (NTEC): two emergingcategories of human and animal pathogens

Jean De Rycke* Alain Milon. Eric Oswald

Unité de recherche associée Inra/ENVT de microbiologie moléculaire,Institut national de la recherche agronomique et École nationale vétérinaire,

23, chemin des Capelles, 31076 Toulouse cedex, France

(Received 27 October 1998; accepted 9 December 1998)

Abstract - Necrotoxic Escherichia coli (NTEC) were originally defined as strains of E. coli producinga toxin called cytotoxic necrotising factor (CNF). Two types of CNF have been identified, each of thembeing genetically linked to several other specific virulence markers, a situation that allows the defi-nition of two distinct homogeneous categories of NTEC callcd NTEC- and NTEC-2. CNF1 and CNF2are highly homologous holoproteins containing 1 014 amino acids that exert both lethal and necroticactivities in vivo and induce multinucleation and actin stress fibres in cell cultures. The activity of CNFson mammal cells is due to their ability to constitutively activate by deamidation the Rho proteins, afamily of small GTPases that regulate the physiology of the cell cytoskeleton. In NTEC-1, the geneencoding CNF1 belongs to a pathogenicity island which also comprises the genes encoding for a-haemolysin and P-fimbriae. In NTEC-2 strains, CNF2 is encoded by a plasmid that also encodes, in100 % of the isolates, a new member of the cytolethal distending toxin family (CDT-III) and inabout 50 °I° of the isolates, the Fl7b-fimbrial adhesin that confers the ability to adhere to calf intcsti-nal villi. The presence of CDT is also suspected in a large majority of NTEC-1 strains. NTEC-1 Istrains can be found in humans and in all species of domcstic mammals, whereas NTEC-2 strains haveonly been reported in ruminants. The implication of NTEC strains has been clearly established in extra-intestinal infections of humans and animals, for instance in urinary tract infections for NTEC-I Istrains. Their role in severe dysenteric syndromes, both in humans and animals, is substantiated byseveral clinical reports, hut there is little published information on this pathogenicity in animal mod-els of infection. The combined production of several powerful toxins (haemolysin, CNF, CDT) byNTEC strains makes them, however, potentially aggressive pathogens which deserve to be searchedfor on a larger scale. Moreover, NTEC-1 from man and animals appear to be highly related accord-ing to available molecular markers, which indicates that domestic animals could constitute reser-voirs of NTEC strains which are pathogenic for humans. &copy; Inra/Elsevier, Paris.

E.veherichia coli / cytotoxic necrotising factor / cytolethal distending toxin / P-fimbriae /Fl7-fimbriae / pathogenicity island

* Correspondence and reprintsTel.: (33) 5 61 19 38 85; fax: (33) 5 61 19 39 75; e.mail: j.derycke@envt.fr

Résumé - Eschetichia coli nécrotoxiques (NTEC) : deux catégories émergentes de pathogèneshumains et animaux. Les colibacilles nécrotoxiques (NTEC) désignent des souches produisant unfacteur cytotoxique et nécrosant (CNF). Il existe deux types de CNF, liés chacun à des marqueurs devirulence spécifiques, définissant ainsi deux catégories de souches appelées NTEC-1 et NTEC-2. CNFIet CNF2 sont des holoprotéines de forte homologie contenant 1 014 acides aminés. Elles exercent uneactivité létale et nécrosante sur les animaux de laboratoire et un effet de multinucléation et de formationde cables d’actine dans les cellules en culture. L’activité des CNF sur les cellules de mamifères estdue à leur capacité d’activer constitutivement par déamidation les protéines Rho, une famille depetites GTPases régulant la physiologie du cytosquelette. Chez les NTEC-l, le gène codant pourCNFI appartient à un îlot de pathogénicité contenant aussi les gènes de l’hémolysine-a et desfimbriae P. Chez les NTEC-2, CNF2 est codé par un plasmide qui code également, dans 100 % dessouches, pour une toxine cytolétale et distendante (CDT-III), et, dans environ 50 % des souches,pour les fimbriae Fl7b déterminant l’adhésion aux villosités intestinales de veau. La présence d’uneCDT est également suspectée dans une large majorité des NTEC-1. Les souches NTEC-1 ont été détec-tées chez l’homme et toutes les espèces de mammifères domestiques, et les NTEC-2 chez les rumi-nants uniquement. L’implication des NTEC est établie dans les infections extra-intestinales del’homme et des animaux, en particulier du tractus urinaire pour les NTEC-1. Leur rôle dans des syn-dromes sévères de dysenterie a été suspecté sur la base d’observations cliniques, mais il existe peude données d’infection expérimentale. La production simultanée par les NTEC de plusieurs toxinespuissantes (hémolysine, CNF et CDT) fait toutefois des NTEC des pathogènes potentiellement agres-sifs dont la détection mériterait d’être effectuée à plus large échelle. De plus, les NTEC-1 isoléschez l’homme et chez les animaux semblent très proches, selon les marqueurs moléculaires actuel-lement disponibles, ce qui indique que les animaux domestiques pourraient constituer des réservoirsde souches pathogènes pour l’homme. &copy; Inra/Elsevier, Paris.

Escherichia coli / facteur cytotoxique et nécrosant / toxine cytolétale et distendante / fimbriae P/ fimbriae F17 / îlots de pathogénicité

1. DISCOVERY AND PRELIMINARYCHARACTERISATIONOF CYTOTOXIC NECROTISINGFACTORS

Early work on CNF has been reviewedcomprehensively [30, 55]. The existenceof a new cytopathic factor later called cyto-toxic necrotising factor (CNF) was origi-nally reported in extracts of E. coli isolatesfrom child enteritis in Italy [20]. In severalepithelial cell lines (CHO, Vero, HeLa,Hep-2), the original toxin was shown toinduce the progressive formation of giantmultinucleated cells [20, 21]. More impor-tantly, the toxin induced a profound reor-ganisation of the cytoskeleton characterisedmainly by the irreversible formation ofthick bundles of actin stress fibres, a phe-nomenon that could account for the inhi-bition of cell division and the subsequentprocess of multinucleation [45]. In addi-tion to their cytopathic properties, the sameE. coli extracts exerted a necrotic reactionin the rabbit skin, an observation whichjustified the denomination of the cytotoxicnecrotising factor for the putative toxinresponsible for both effects [20]. The toxinwas later purified to homogeneity as a holo-protein of 115 kDa apparent molecular sizeby gel electrophoresis, which accountedfor both the cytopathic and necrotic prop-erties [20, 34]. CNFl was highly lethal formice, with a 50 % lethal dose estimated atabout 20 ng of purified material [31-!. Theclinical and histopathological effects pro-duced by purified CNFI were studied inthe lamb inoculated intravenously. Thetoxin-induced death was preceded byneurological signs and mucous diarrhoea.The most striking lesions were oedemaand haemorrhages of the central nervoussystem [3 1 ].

The existence of two distinct types ofCNF was then reported in E. coli from calfenteritis [32, 35]. Both types of cytotoxins,which are highly lethal in mice inoculated bythe intraperitoneal route, share the abilityto induce giant multinucleated cells in the

HeLa epithelial cell line and necrosis in therabbit skin. They are, however, only par-tially related as demonstrated by seroneu-tralisation studies and by responses in bio-logical assays: CNF1, which is similar tothe original type, is more potent than CNF2in inducing multinucleation but less necroticin the rabbit skin test and mouse footpad[35]. Purification of CNF2 led to the iden-tification of a 110-kDa protein whichmigrates faster than CNFI in gel elec-trophoresis but cross-reacts with a CNF1 Iantiserum in western blotting [72]. CNF1 Iand CNF2 are, however, sufficiently dis-tinct on a molecular basis to allow their dif-ferential detection by enzyme-linkedimmunosorbent assay [77, 89] or by DNAhybridisation [75].

2. MOLECULARCHARACTERISATIONOF CNFI AND CNF2 ANDIDENTIFICATION OFTHEIR CELLULAR TARGET:THE SMALL GTPASE RHO

Whereas CNFI is encoded by the chro-mosome [42J, CNF2 is determined by aplasmid, which, in the S5 E. coli prototypestrain, is transmissible by conjugation [73].CNF1 and CNF2 are both encoded by a sin-gle structural gene with a low GC content(35 %) [43, 76 which might indicate a rel-atively recent acquisition by E. coli whoseaverage GC content is 50 %. Sequence anal-ysis showed the close relatedness of the twotoxins which both contain 1 014 predictedamino acids and share 84 % identicalresidues and 90 % conserved residues. The

predicted molecules are characterised bytheir overall hydrophilic properties, the exis-tence of two putative hydrophobic trans-membrane domains partially overlapped bytwo predicted a-helices, and by the absenceof a classical signal peptide in the N-termi-nal end. CNFs share homologous aminoacid sequences with two other der-monecrotic bacterial toxins: Pa,rteurella

multocida toxin (PMT), in an extensiveregion of the N-terminal end, and der-monecrotic toxin of Bordetella pertussis ina short region of the C-terminal amino acids(DNT). The N-terminal region of CNF1contains the cell-binding domain whereasthe catalytic activity is in the C-terminalregion [65].

The specific activity of CNFs on thecytoskeleton is due to their ability to con-stitutively modify the Rho proteins, a fam-ily of small GTPases that regulate the phys-iology of cell cytoskeleton [46, 65, 76]. Themost striking effect of this modification ofRho is to trigger the irreversible formation ofthick bundles of actin stress fibres in thecell. Another consequence of Rho activa-tion by the CNFs is to activate DNA syn-thesis [63] and even to stimulate quiescentdifferentiated cells to enter into the S-phaseof the cell cycle (our unpublished informa-tion). How CNFI (and most probablyCNF2) is able to modify Rho has beenrecently elucidated. CNF catalyses thedeamidation of a glutamine residue at posi-tion 63 (Gln 63) to give a constitutivelyactive Rho protein [48, 821. In addition,CNFI possesses a transglutaminase activ-ity [83]. DNT, whose catalytic domain ishomologous to that of CNF, happens to exertexactly the same activity [59!. These twodermonecrotoxins are the first examples oftoxins acting by deamidation of a specifictarget protein. In spite of their similar bio-logical activity, however, the toxins haveseveral distinct characteristics. Detection of

proteins [32p]ADP-ribosylated by clostridialC3 exoenzyme in DNT-treated cellsrevealed multiple bands with slower andfaster electrophoretic mobilities than that ofthe untreated control [58]. In contrast, pro-tein [3’P]ADP-ribosylated by Stophvlococ-cu.s aureus epidermal differentiationinhibitor (EDIN) or by C3 in CNF-treatedcells revealed a single band in one-dimen-sional slab gel electrophoresis and the bandmigrated more slowly than that of theuntreated control [76].

3. CNF AS A POTENTIAL FACTOROF PATHOGENICITY

The necrotic and highly lethal effectsexerted by CNF1 and CNF2 in experimen-tal animals [31, 33], show that, if released insufficient amounts during an infection, thetoxin is potentially able to exert serious tox-aemic effects. The 50 % lethal dose of CNF1 I

by the intravenous route was estimated tobe about 20 ng for mice, which means thatCNF1 should be ranked among the highlylethal bacterial toxins [31]. At the cellularlevel, the original biological activitiesexerted by CNF, on the cell cytoskeleton inparticular, may also be relevant to NTECpathogenicity (see section 5). First, by induc-ing permanent stress fibres and rufflingactivity, CNF confers the ability to ingestany particles in contact with the cell mem-brane, such as latex beads or bacteria, toexposed cells [44]. Acquisition of thisphagocytic capacity, which reportedlyrequires an exposure of the epithelial cells tothe toxin of at least 48 h, appears to berelated to the progressive inability to enterapoptosis [47]. This correlation can be inter-preted as a coherent strategy of the pathogento induce a long-lasting state of phagocyte-like activity in epithelial cells, allowing bac-terial multiplication and even transcytosis.Another in vitro observation may bear somerelevance to the pathogenicity of NTECstrains toward epithelia. CNFI effacesmicrovilli and decreases transmigration ofpolymorphonuclear leukocytes in intestinalT84-epithelial cell monolayers, without theapparent loss of epithelial polarity [57].

The role of CNF1 in the pathogenicityof NTEC-I strains has also been examinedin cellular models of interaction [36, 40,60 In these models, parental strains werecompared with mutants specifically deletedfor CNF I or for a-haemolysin, another toxinassociated with CNF1 in NTEC- strains.The main difficulty in these models is that a-haemolysin exerts a rapid cytolytic effectthat prevents the observation of the specificeffect caused by CNF1 [40, 60!. The spe-

cific effect of CNF, consisting of the devel-opment of bundles of actin stress fibres,was, however, observed with interactionassays on HeLa cells: either with NTEC-1 Istrains after preliminary seroneutralisation ofhaemolysin [36] or with NTEC-2 strains[79]. The latter observations indicate thatCNF toxins, which do not possess a N-ter-minal signal sequence and whose activityis not detectable in culture supernatants, are

processed through the bacterial membranesand internalised by the target eukaryoticcells. The interesting in vitro observationssummarised above could account for thevarious modalities of NTEC infections inthe field but they deserve to be confirmed invivo, either in the natural disease or in exper-imental models of infection (see section 6).

4. LINKAGE OF EACH TYPEOF CNF WITH SPECIFIC VIRU-LENCE MARKERS

Neither CNF 1 nor CNF2 occurs as a sin-

gle independent characteristic in E. coli. Eachof them is closely associated with specificvirulence markers, an association that cor-

responds to some degree of genetic linkage.

4.1. Characters associated with CNF1

The close association of CIVFl with a-

haemolysin and P-fimbriae (pap) was recog-nised in early studies with E. coli strainscausing septicaemia or urinary tract infec-tions in humans [8, 14, 18, 22] and in cattle[80]. Moreover, isolates which are positivefor the three characters Hly, CNF1 and P-fimbriae belong to a limited number ofserotypes, essentially 02, 04, 06 and 075.The latter isolates also display a specificesterase electrophoretic pattern called B2[53]. The striking similarity observedbetween animal and human strains withinthe 04, 06 and 078 serogroups with regardto esterase and outer membrane protein pat-terns suggests their close relationship [25,

26]. Both in human and animal NTEC-1 Iisolates, the genetic locus encoding P-fim-briae is the class III allele of the pap Gadhesin (called prs), an adhesin whose speci-ficity is thus possibly not restricted to thehuman host [4, 38, 69].

The association between CNF1, Hly andP-fimbriae has now been confirmed on amolecular basis. First, cnf7 and hly wereshown to be closely linked in a 37-kb clonedDNA fragment from a human E. coli uri-nary tract isolate of serotype 04:K12:H5(E-B35). In the same study, this genetic link-age was extended to six additional extra-intestinal isolates belonging to the 02, 04,06, 022, 075 and 085 serogroups [42].Another human uropathogenic E. coli strainprototype of serotype 04:K6:H5, called J96,contains two a-haemolysins (hly I and hlyII), pap fimbriae (pap and prs), F1C-fim-briae (fbc) and type 1-fimbriae !m). Spon-taneous variants of J96 are characterised bythe combined loss of cnfl, hly II and prsexpression, an observation which indicatesthat these three virulence determinants couldbe part of a pathogenicity island (PAI) [16,39, 56]. The detailed analysis of the entireDNA region confirmed that cnfl , hly Il andprs belongs to a PAI of about 110 kb, calledPAI-5, inserted in the chromosome next toPheU (PheR), the gene encoding the tRNAof phenylalanine [88]. The organisation ofthe DNA virulence cluster to which cnflbelongs was recently compared in severalNTEC-1 from various animal and geo-graphical origins. Overall, a large poly-morphism was observed but the genesencoding CNF1, haemolysin and P-relatedadhesins are clustered together on PAIs andfall into two distinct groups: i) PAIs that areinserted near PheU as is the case for thePAI-5 of strain J96; and ii) strains that con-tain PAI-5-type islands that are inserted else-where in the chromosome (H6rault et al.,unpublished results). Interestingly enough,the 900-bp DNA fragment separating thehlyD gene of the hly operon and the cnfl isconserved in all NTEC- tested so far. Theseresults could suggest that cnf7 and hly were

genetically associated before the insertionof the PAI in the chromosome.

4.2. Characters associated with CNF2

In NTEC-2 strains, CNF2 is encoded bya Vir plasmid, a family of large plasmids ofabout go- 106 bp whose transferability to lab-oratory strains was shown with specifiedprototype strains [67]. CNF2 is probablythe incompletely characterised Vir lethaltoxin described as early as 1974 [84]. Theplasmid determinism of CNF2 was demon-strated through the transferability of its invitro and in vivo biological activities by bac-terial conjugation 1721. ] .We know at least two other virulence-

associated characters that are encoded byVir plasmids along with CNF2: a F17-related fimbriae, called F17-b, which conferson the strains the capacity to specificallyadhere to calf intestinal villi [41, 74], and anew type of afimbrial adhesin (AFA) [69]. ] .In E. coli from ruminants, the F17 familyincludes the Fl7a, Fl7b, Fl7c and F111 Itypes [3]. F 17-related fimbriae are present inabout 53 % of CNF2-producing strains,56 % of which belong to the Fl7b type.Moreover, Fl7b fimbriae appear to be exclu-sively associated with CNF2-producingstrains 14, 74 Fl7c and F1 I 1 account forabout 22 % each of F17-positive CNF2-pro-ducing isolates [3] but their association withthe Vir plasmid has not yet been investi-gated (see the review on F17 [64]).

In addition to the F17-related fimbriae,certain NTEC-2 strains may express afim-brial adhesins. The presence of afa-relatedsequences has been reported in severalNTEC-2 strains [69]. Certain of these afa-related sequences were located on the Vir

plasmid [69]. The qfa NTEC-2 strains were,however, negative under high stringencyhybridisation conditions and in the afa-spe-cific PCR derived from the sequence anal-

ysis of the conserved region of the humanafa operons, suggesting that this aja operonwas structurally different from the operons

of human E. coli isolates. Indeed, the afadeterminants located on the Vir were

recently shown to be members of a new typeof afa gene cluster: afa-8 (Lalioui and LeBouguenec, pers. comm.). The afa familyincludes closely related gene clusters thatare expressed by uropathogenic and diar-rhoea-associated E. coli (see the review onAFA [64].

5. CYTOLETHAL DISTENDINGTOXIN (CDT): A NEW TOXINCLOSELY ASSOCIATED WITHCNF IN NTEC STRAINS

Recently, an additional biological activ-ity was detected in NTEC strains followingtheir interaction with the HeLa epithelialcell line. Short exposure of these cells to aculture of NTEC-1 BM2-1 prototype straininduces the formation of giant mononucle-ated cells with two major characteristics:the formation of actin stress fibres (an effectdue to CNF1) and an irreversible cell cycleblock at stage G2/M (an effect notattributable to CNFI) [36, 37]. NTEC-2strains also induce this cell cycle block atstage G2/M, and the bacterial locus thatdetermined this specific activity was iden-tified in one clone of a DNA library con-structed from the Vir plasmid of prototypeNTEC-2 strain 1404 [79]. This locus hap-pened to encode a new cytolethal distendingtoxin (CDT-III), distinct from the two pre-vious types of CDT reported in non-NTECE. coli. These two former types also pro-duced the G2/M block when tested in HeLacells [79J. This original property also prob-ably applies to CDTs produced by other bac-terial species of medical and veterinary rel-evance, such as Shigella [71 ], Haerreophilus[29J and Actinobacillus [87].CDT is the first known bacterial toxin

that specifically blocks the eukaryotic cellcycle in G2. The block is due to the main-tenance of the so-called mitosis promotingfactor (MPF), the complex that determinesentry into mitosis, in an inactivated state.

Cdkl (cdc2), the cyclin-dependent kinasethat constitutes the catalytic part of MPF,remains hyperphosphorylated, thus pre-venting the transition between stages G2and M of the cell cycle [27].

According to our as yet unpublishedobservations, CDT-111 is present in 100 % ofNTEC-2 isolates. We observed that follow-

ing the interaction of NTEC- with HeLacells, about 75 % of isolates produce a G2/Mblock, in addition to the specific cytoskele-tal reorganisation attributable to CNFI [36].This observation suggests that a majority ofNTEC-1 strains also produce a member ofCDT. The putative new CDT of NTEC-1 Iis, however, not sufficiently related toknown CDT members to be detected byhybridisation and PCR methods (our unpub-lished results). CDT would then appear as anew character tightly associated with NTECstrains. This close association between CDTand CNF in both groups of NTEC cannotbe considered fortuitous and could be hypo-thetically explained by the necessity of bio-logical co-operation between the two tox-ins during the interaction with the host cells.We observed that on fully differentiated andconfluent epithelial cells, CDT-111 alone hadno effect, whereas CNF2 stimulated theentry of the quiescent epithelial cells intothe S phase. Interestingly, these CNF-stim-ulated confluent cells became sensitive toCDT activity (Sylvie POr!s, unpublishedresults). This synergy could be critical invivo on fully differentiated epithelium andmight impair the intestinal barrier.

6. ROLE OF NTEC STRAINS INHUMAN AND ANIMAL DISEASES

6.1. In humans

There is repeated convincing evidence,established on both epidemiological andclinical grounds, that NTEC-1 are respon-sible for various extra-intestinal infections inhumans. This conclusion is based on the

comparison of the frequency of virulencefactors in samples of E. coli isolates from

clinically defined extra-intestinal infectionsas compared to faecal isolates. Whereas iso-lates producing both haemolysin and CNF1 Iare seldom found in faeces from healthyindividuals, their frequency reaches an aver-age of 40-50 % in strains from urinary tractinfections [ 14, 17, 49, 61, 94], prostatis [1,90] and other extra-intestinal infections [9,18, 22, 25, 61]. The frequency of NTEC-1 isreported as being similar in cystitis,pyelonephritis and asymptomatic bacteriura[14]. It is worth noting that a majority ofNTEC-I strains incriminated in extra-intesti-nal infections also produce adhesins suchas P-fimbriae (see also section 4.1) and thatthey belong to a restricted number ofserotypes, mainly 02, 04 , 06, 08 and 075[8, 9, 13, 23, 53]. Among these serotypes, E.coli 04:H5, associated with about 2 % ofextra-intestinal infections in the USA and

typified by the prototype strain J96, is con-sidered as clonal [62 1.

Although discovered for the first time inneonatal enteritis [20], the participation ofNTEC-1 in diarrhoeal diseases in humanshas not been as clearly established as inextra-intestinal infections [6]. However, sev-eral clinical and epidemiological observa-tions suggest that NTEC-1 do exert a

pathogenic effect on the intestine, at leastin young children. First, two separate out-breaks of neonatal diarrhoea occurring intwo different Spanish hospitals werereported to be associated with NTEC-1 ofthe 06: K 13 serotype [ 10], which provides agood suspicion about the potential role ofNTEC-1 in enteritis in infants. Second,pyelonephritic E. coli strains, which werelater characterised in our laboratory as typ-ical NTEC-1 strains expressing haemolysin,CNF and P-fimbriae were incriminated intwo cases of ecchymotic colitis in new-borninfants [50]. Finally, it should be mentionedthat cases of diarrhoea in children due to

haemolytic E. coli were reported in Aus-tralia 1541 and in Somalia [70]. Isolates ofthe Australian study were later described asalso producing CNF I , and therefore can bequalified as typical NTEC-1 strains [40].

6.2. In pigs

NTEC-1 but not NTEC-2 strains havebeen reported in pigs. According to a surveyof 65 Spanish piggeries in different areasof Spain, about 1.5 % of the animals, eitherdiarrhoeic or healthy, were shedding NTEC-1 strains in their faeces [51 ]. In England andWales, among 3 595 E. coli porcine isolatesexamined by the Central Veterinary Labo-ratory between 1986 and 1991, 6.8 % wereCNF-positive. These figures indicate thatNTEC-1 is far from uncommon in pigs. Theparticipation of NTEC-1 strains with diar-rhoeal disease in piglets was suggested inepidemiological studies on the basis of fae-cal shedding by diarrhoeic animals [68].More convincing evidence on the actualpathogenic role of NTEC-1 in neonatal pigswas then provided experimentally throughexperimental infection by the oral route [93]. ] .The two tested NTEC-1 strains, belonging tothe 088 and 032 serogoups, caused an earlyenteritis, progressing to enterocolitis andcharacterised clinically by blood-staineddiarrhoea, and a bacteriaemic spread to thelungs. The histopathological changes wereconsistent with a toxaemic process andshowed that, aside from their pathogenic-ity for the intestine, NTEC-1 strains has aparticular predilection for the lungs. Theparticipation of NTEC-1 strains in a lethaloutbreak of necrotic enteritis accompaniedby interstitial pneumonia in weaner pigs wasalso reported in South Africa [78!. It is worthmentioning that NTEC- strains from pigsbelong in large majority to the sameserotypes as NTEC-1 from humans, mainly02, 04, 08, 054, 078 and 083 52 1.

6.3. In ruminants

6.3.1. NTEC-I I

NTEC-1 strains, which are not commonlyfound in the faeces of young and adult cat-tle as compared to NTEC-2 [15], have pre-viously been associated with calf diarrhoea!7!, specifically with mucous enteritis inseverely depressed calves [32]. An outbreak

of septicaemic diseases in young beef cattlecharacterised by haemorrhagic purpura dueto NTEC-1 strain has been reported [28].Three clustered cases of lethal haemorrhagiccolitis in 14-month-old heifers due to aNTEC-1 strain of the 02 serogroup and

expressing the fimbrial antigen F6 (alsocalled 987P) have also been described. Theiractual role in disease is corroborated by theirdemonstration in the ulcerative inflamma-

tory lesions of the colon and rectum of dis-eased animals [91 ].

6.3.2. NTFC-2

The septicaemic ability of the ovine S5prototype NTEC-2 strain has been demon-strated on different animal models, includ-ing calves [85], and the Vir plasmid of E.coli S5 was shown to confer bacteriaemic

properties to laboratory strains of E. coli[66, 85, 86[. This indicates that upon cer-tain circumstances, NTEC-2 are able toovercome the host defence to exert their

pathogenicity, either in the digestive tractor after invasion of the organism.

NTEC-2 strains are commonly found inthe faeces of healthy calves and adult cattle,reaching a prevalence of about 50 % for theformer and 20 % for the latter [ 13, 15, 19].Their shedding in sheep faeces is also prob-ably high, as suggested by the frequency ofCNF-positive isolates in E. coli isolatesexamined in England and Wales at the Cen-tral Veterinary Laboratory, estimated to be9.1 °Io of all isolates [92 This spontaneousadaptation of NTEC-2 to the digestive tractof ruminants without apparent harmfuleffects, could be interpreted as a lack ofpathogenicity. The possible role of NTEC-2 strains in severe haemorrhagic enterocol-itis (Schelcher and De Rycke, unpublishedresults) should, however, be further inves-tigated.

6.4. In cats and dogs

A few reports incriminate NTEC- Istrains in gastro-enteritis of dogs [81 ] and inurinary tract and other extra-intestinal infec-

tions of both cats and dogs [95]. It is worthmentioning that up to about 50 % of healthycats shed NTEC-1 in their faeces accordingto a Spanish report [11]. Moreover, NTEC-1 strains from dogs and cats belong in themajority to the same serogroups as NTEC-1 from humans, i.e. 02, 04, 06 and 023[11, 81] (see review on E. coli as a pathogenin dogs and cats [5].

6.5. In other species

Outbreaks of lethal enteritis due toNTEC-1 of serogroup 02 were reported inrabbits and in horses [2]. In both cases thestrains were recovered in pure cultures fromthe intestinal contents of several diarrhoeic

animals, an observation that suggests theirrole in the symptoms observed. In the rabbitoutbreak reported above, about one third ofthe animals developed non-bloody diarrhoea(14 000 out of 50 000) from which morethan 10 % died. In the horse outbreak

reported above, three 2-month-old animalsdied after a short stage of dysentery withscanty haemorrhagic faeces, which are clin-ical symptoms reminiscent of those observedin heifers with NTEC-1 strains of the 02

serogroup [91 It is worth noting thatNTEC-1 strains have also been found inferal animals such as wild ruminants andmarmots [24!. j.

7. CONCLUSION

There is presently ample well-foundedevidence that NTEC-1 strains are harmful

pathogens in both humans and domestic ani-mals. Previously designated as

uropathogenic E. coli, their pathogenicityis in fact not restricted to the urinary tractand encompasses both the digestive tractand other organs. As described in section

4.1, isolates of this group of pathogenic E.coli have fairly constant characteristicsgrouped in a pathogenicity island (includ-ing, at least, hly, CNF1 and the class IIIallele of the pap G adhesin), and they belongto a limited number of serotypes. Most

importantly, animal strains of this group donot appear, according to current knowledge,to be different from human strains, whichposes the problem of a potential threat forhuman health through contact with animalreservoirs. In this context, the frequent shed-ding of NTEC-1 strains by cats in particularshould be of a particular concern in veteri-nary public health.

The role of NTEC-2 in pathology,restricted in any case to ruminants, remainsmore controversial. Early observationsshowed that some prototype NTEC-2 strainsdo have a septicaemic potential associatedwith the Vir plasmid, but to our knowledgethere is no published report allowing theepidemiological assessment of the occur-rence of NTEC-2 strains in extra-intestinaldiseases of ruminants. The participation ofNTEC-2 strains in calf enteritis is even moredifficult to demonstrate because these strainsare commonly harboured by young rumi-nants without apparent pathological conse-quences, in a much similar way as verotox-

inogenic E. coli or Salmonella. To be of anyvalue, the demonstration of the role ofNTEC-2 strains in diseases of young rumi-nants should refer to carefully defined clin-ical cases (for instance to haemorrhagic col-itis in beef calves) and should include thedirect demonstration of the NTEC-2 strainsin the intestinal lesions and/or in other

organs.

ACKNOWLEDGEMENT

The NTEC programme conducted in our lab-

oratory is supported by a grant from the Euro-pean Community Program FAIR (# 1335).

REFERENCES

III Andreu A., Stapleton A.E., Fennel C., LockmanH.A., Xercavins M., Fernandez F., Stamm W.E.,Urovirulence determinants in Escherichia colistrains causing prostatitis, J. Infect. Dis. 176(1997)464-469.

!2! Ansuini A., Candotti P., Vecchi G., Falbo V.,Minelli F.. Caprioli A., Necrotoxigenic

E.scherichia coli in rabbits and horses, Vet. Rec.134 ( I 994) 608.

[3] Bertin Y., Martin C., Oswald E., Girardeau J.P.,Rapid and specific detection of F17-related pilinand adhesin genes in diarrheic and septicemicEscherichia coli strains by multiplex PCR, J. Clin.Microbiol. 34 (1996) 2921-2928.

[4] Bertin Y., Martin C., Girardeau J.P., Pohl P., Con-trepois M., Association of genes encoding P fim-briae, CS31A antigen and EAST I toxin amongCNFI-producing Escherichia coli strains fromcattle with septicemia and diarrhea, FEMS Micro-biol. Lett. 162 (1998) 235-239.

[5] Beutin L., Escherichia coli as a pathogen in dogsand cats, Vet. Res. 30 ( 1999) 285-298.

[6] Bisicchia R., Ciammarughi R., Caprioli A., FalboV., Ruggeri F.M., Toxin production and haemag-glutination in strains of Escherichia coli fromdiarrhoea in Brescia, Italy, J. Hyg. 95 ( 1985)353-361. 1 .

[7] Blanco J., Gonzalez E.A., Garcia S., Blanco M.,Regueiro B., Bemardez I., Production of toxins byEscherichia coli strains isolated from calves withdiarrhoea in Galicia (north-western Spain), Vet.Microbiol. 18 ( 1988) 297-311. 1 .

[8J Blanco J., Alonso M.P., Gonzalez E.A., BlancoM., Garabal J.I., Virulence factors of bacteraemicEscherichia coli with particular reference to pro-duction of cytotoxic necrotising factor (CNF) byP-fimbriated strains, J. Med. Microbiol. 31 (1990)175-183.

191 Blanco J., Blanco M., Alonso M.P., Blanco J.E.,Gonzalez E.A., Garabal J.1., Characteristics ofhaemolytic Escherichia coli with particular ref-erence to production of cytotoxic necrotizing fac-tor type I (CNFI), Res. Microbiol. 143 (1992)869-878.

[ 10! Blanco J., Gonzalez E.A., Espinosa P., BlancoM., Garabal J.I., Alonso M.P., Enterotoxigenicand necrotising Escherichia coli in human diar-rhoea in Spain, Eur. J. Epidemiol. 8 ( 1992)548-552.

[11] I J Blanco J., Blanco M., Wong I., Blanco J.E.,Haemolytic Escherichia coli strains isolated fromstools of healthy cats produce cytotoxic necro-tizing factor type 1 (CNFI), Vet. Microbiol. 38(1993)157-165.

!!2! Blanco J.E., Blanco J., Blanco M.. Alonso M.P.,Janscn W.H., Serotypes of CNF1-producingE.scherichia coli strains that cause extraintestinalinfections in humans, Eur. J. Epidemiol. 10 (1994)707-71 1 .

[ 13J Blanco M., Blanco J., Blanco J.E., Ramos J.,Enterotoxigenic, verotoxigenic, and necrotoxi-genic Escherichia coli isolated from cattle inSpain, Am. J. Vet. Res. 54 (1993) 1446-1451. 1.

! 14! Blanco M., Blanco J.E., Alonso M.P., BlancoJ.B., Virulence factors and 0 groups ofEscherichia coli isolates from patients with acutepyelonephritis, cystitis and asymptomatic bac-teriuria, Eur. J. Epidemiol. 12 ( 1996) 191-198.

[ 15J Blanco M., Blanco JE., Mora A., Blanco J.B.,Distribution and characterization of faecal necro-

toxigenic Escherichia coli CNFI+ and CNF2+isolated from healthy cows and calves, Vet.Microbiol. 59 (1998) 183-192.

[16] Blum G., Falbo V., Caprioli A., Hacker J., Geneclusters encoding the cytotoxic necrotizing factortype 1, Prs-fimbriae and alpha-hemolysin formthe pathogenicity island II of the uropathogenicEscherichia coli strain J96, FEMS Microbiol.Lett. 126 ( 1995) 189-195.

[ 17] Brauner A., Katouli M., Tullus K., Jacobson S.H.,Production of cytotoxic necrotizing factor, vero-cytotoxin and haemolysin by pyelonephritogenicEscherichia coli, Eur. J. Clin. Microbiol. Infect.Dis. (1990) 762-767.

[18] Brauner A., Katouli M., Ostenson C.G., P-fim-briation and haemolysin production are the mostimportant virulence factors in diabetic patientswith Escherichia coli bacteraemia: a multivari-ate statistical analysis of seven bacterial virulencefactors, J. Infect. 31 (1995) 27-31. 1.

[ 19] Burns A.L., Ball H.J., Finlay D.A., CNF produc-ing Escherichia coli isolated from cattle in North-ern Ireland, Vet. Microbiol. 49 ( 1996) 235-241. 1.

[20) Caprioli A., Falbo V., Roda L.G., Ruggeri F.M.,Zona C., Partial purification and characterizationof an Escherichia coli toxic factor that induces

morphological cell alterations, Infect. Immun. 39(1983) 1300-1306.

[21] Caprioli A., Donelli G., Falbo V., Possenti R.,Roda L.G., Roscetti G., Ruggeri F.M., A cell divi-sion-active protein from Escherichia coli,Biochem. Biophys. Res. Commun. 118 (1984)587-593.

[22] Caprioli A., Falbo V., Ruggeri F.M., BaldassarriL., Bisicchia R., Ippolito G., Romoli E., DonelliG., Cytotoxic necrotizing factor production byhemolytic strains of Escherichia coli causingextraintestinal infections, J. Clin. Microbiol. 25(1987) 146-149.

[23! Caprioli A., Falbo V., Ruggeri F.M., Minelli F.,Orskov 1., Donelli G., Relationship between cyto-toxic necrotizing factor production and serotypein hemolytic Escherichia coli, J. Clin. Microbiol.27(1989)758-761. 1 .

[241 Caprioli A., Donelli G., Falbo V., Passi C.,Pagano A., Mantovani A., Caprioli A., Antimi-crobial resistance and production of toxins inE.scherichia coli strains from wild ruminants andthe alpine marmot, J. Wildt. Dis. 27 ( 1991 ) 324.

[25 Cherifi A., Contrepois M., Picard B., Goullet P.,De Rycke J., Fairbrother J., Barnouin J., Factorsand markers of virulence in E.scherichia coli fromhuman septicemia. FEMS Microbiol. Lctt. 58( I 990) 279-283.

[26! Cherifi A., Contrepois M., Picard B., Goullet P.,Orskov L, Orskov F., De Rycke J., Clonal rela-tionships among Escherichin coli serogroup 06isolates from human and animal infections, FEMSMicrobiol. Lett. 64 (1991) 225-30.

[27! Comayras C., Tasca C., Peres S.Y., DucommunB., Oswald E., De Rycke J., Escherichia cnlicytolethal distending toxin blocks the HeLa cellcycle at the G2/M transition by preventing cdc2protein kinase dephosphorylation and activation,Infect. Immun. 65 ( 1997) 5088-5095.

[28] Contrepois M., Ribot Y., Etude des colibacillesisol6s lors de septic6mies bovines : I. avec symp-tomes meninges ; 2. avec syndrome immuno-d6pressif et purpura hémorragique, Bull. Acad.vét. Fr. 59 (1987) 465-473.

129] Cope L.D., Lumbley S., Latimer J.L., Klesney-Tait J., Stevens M.K., Johnson L.S., Purven M.,Munson R.S. Jr, A diffusible cytotoxin ofHaell10phillls dllcreri, Proc. Natl. Acad. Sci. USA94 ( 1997) 4056!06 1 .

1301 De Rycke J., Les colibacilles producteurs de cyto-toxines : importance en m6decine v6t6rinaire ct ensaiit6 publique, Ann. Rech.Vct. 22 ( 1991 )105-126.

[3 De Rycke J., Plzissiart G., Toxic effects for lambsof cytotoxic nccrotising factor from Escherichincoli, Res. Vet. Sci. 49 (1990) 349-354.

[32! Dc Rycke J., Guillot J.F., Boivin R., Cytotoxinsin non-enterotoxigenic strains of* ruliisolated from fcces of diarrheic calves, Vet.Microbiol. 15(1987) 137-150.

[33! Dc Rycke J.. Oswald E., Boivin R., An in vivoassay for the detection of cytotoxic strains ofEscherichia coli. Ann. Rcch. Vet. 20 ( 1989)39--46.

1341 Dc Rycke J., Phun-Thanh L., Bernard S.,Immunochcmical identification and biologicalcharacterization of cytotoxic necrotizing factorfrom Escherichia cnli, J. Clin. Microbiol. 27( 1989) 983-988.

1351 Dc Rycke J., Gonzalez E.A.. Blunco J., OswaldE., Blanco M., Boivin R., Evidence for two typesof cytotoxic necrotizing factor in human and ani-mai clinical isolates of Escherichia coli. J. Clin.Microbiol. 28 ( 1990) 694-699.

[ 36 j De Rycke J., Mazars P., NougayredeJ.P., TascaC.. Boury M., Hcrault F., Valette A., Oswald F.,Mitotic block and dclaycd Icthality in HeLacpithclial cells cxposed to Elcherichia coli BM2-I producing cytotoxic necrotizing factor type I,infect, 1111111LIll. 64 (1996) !694-t&dquo;705.

! 37 Dc Rycke J., Nougayrede J.P.. Oswald E.. MazarsP.. interaction of L:cr herichiu coli producing cyto-toxic nccrotizing factor with HeLa epithelial cells.Adv. Fxp. Med. Biol. 412 ( 1997) 363-366.

1381 Dozois C.M., Clcment S., Dcsautels C., OswaldE., Fairbrother J.M., Expression ol’P, S, and F I Cadhesins by cytotoxic nccrotizing factor I-pm-ducing Esclrerichio coli from scpticemic and diar-rhcic pigs, FEMS Microbiol. Lett. 152 (1997)307-312.

1391 Dozois C.M., Curtiss IIl R., Pathogcnic diversityof /’.Br/7!’/r/7!/ coli and the emergence of exotic icislands in the gcne stream. Vct. Res. 30 (1999)I .57-179.

[40] Elliott S.J., Srinivas S., Albert M.J., Alam K.,Robins-Browne R.M., Gunzburg S.T., Mee B.J.,Chang B.J., Characterization of the roles ofhemolysin and other toxins in enteropathy causedby alpha-hemolytic E.scherichia coli linked tohuman diarrhea, Infect. Immun. 66 (1998)2040-2051. 1 .

[41 ! El Mazouari K., Oswald E., Hernalsteens J.P.,Lintermans P., De Greve H., F17-like fimbriaefrom an invasive !’.tcAfnc/!f; coli strain produc-ing cytotoxic necrotizing factor type 2 toxin,Infect. Immun. 62 (1994) 2633-2638.

[42 ! Falbo V., Famiglictti M., Caprioli A., Gene blockencoding production of cytotoxic necrotizing fac-tor 1 and hemolysin in E.tc/x’nr/t/a coli isolatesfrom extraintestinal infections, Infect. Immun.60(1992)2182-2187.

[43 ! Falbo V., Pace T., Picci L., Pizzi E., Caprioli A.,Isolation and nucleotide sequence of the geneencoding cytotoxic necrotizing factor I ofEscherichia roli, Infect. Imtnun. 61 (1993)4909!1914.

[441 Filzano L., Fiorentini C., Donelli G., Michel E..Kocks C.. Cossart P., Cabanie L., Oswald E.,Boquct P., Induction of phagocytic behaviour in nhuman epithelial cells by E.scherichia coli cyto-toxic necrotizing factor type I , Mol. Microbiol. 9(1993)1247-1254.

[45 ! Fioientini C.. Arancia G., Caprioli A., Falbo V.,Ruggeri F.M., Donelli G., Cytoskeletal changesinduced in HEp-2 cells by the cytotoxic necro-tizing factor of /!st77!!’/!/ coli, Toxicon 26(1988)1047-1056.

1461 Fiorentini C., Fahhri A., Flatau G.. Donelli G..Matarrese P., Lemichez E.. Falzano L.. BoquetP., I’schcrichin coli cytotoxic necrotizing factor I

(CNF 1), ), toxin that activates thc Rho GTPasc, J.Biol. Chcm. 272 (1997) 19532-19537.

[47! Fiorcntini C.. Fabbri A.. Matarrcse P.. Falzano1,.. Boquet P.. Malorni W.. Hinderance of apop-tosis and phagocytic behaviour induced byli.scherichia coli cytotoxic necrotizing factor I:two related activities in epithelial cells, Biochem.Biophys. Rcs. Commun. 241 ( 1997) 341-346.

1481 Flatau G.. Lemichcr E., Gauthier M., Chardin P..Paris S., Fiorentini C., Boquet P., Toxin-inducedactivation of the G protein p21 Rho by deamida-tion of glutuminc, Nature 387 ( 1997) 729-733.

!49! Foxman B.. Zhang L.. Palin K., Tallntan P.. MarrsC.F.. Bacterial virulence characteristics of/’,B<7!7r/?M coli isolates from first-time urinarytract infection, J. Infect. Dis. 171 (1995)1514-1521. 1 .

1501 Gaill;trd J.L.. Chcron G., Mougcnof J.F.. DcslysJ.1’.. Nczclof C.. Vcron M.. Schmitz .I.,1’ycloncphritic Elc/¡erichia c oli strains as intesti-nal pathogens in two newborn inflnts, Lancct I(1989)327-328.

J5 Garahal J.1., Gonzaiez E.A.. V!tzquez F.. BlancoJ., Blanco M.. Toxigcnic E.sclrerichia mli in Span-

ish piggeries from 1986 to 1991, Vet. Microbiol.47 ( 1995) 17-25.

f521 Garabal J.I., Gonzalez E.A., Vazquez F., BlancoJ., Blanco M., Blanco J.E., Serogroups ofEscherichia coli isolated from piglets in Spain,Vet. Microbiol. 48 ( 1996) 1 13-123.

[53 J Goullet P., Picard B., Contrepois M., De Rycke J.,Barnouin J., Correlation between esterase elec-trophoretic polymorphism and virulence-associ-ated traits in extra-intestinal invasive strains ofEscherichia coli, Epidemiol. Infect. 1 12 ( 1994)51-62.

[54J Gunzburg S.T., Chang B.J., Elliot S.J., Burke V.,Gracey M.,. Diffuse and enteroaggregative pat-terns of adherence of enteric Esclrerichia coli iso-lated from aboriginal chidren from the Kimberleyregion of Western Australia, J. Infect. Dis. 167(1993)755-758.

[551 Gyles C.L., E.scherichia coli cytotoxins andenterotoxins, Can. J. Microbiol. 38 ( 1992)734-746.

[561 Hacker J., Blum-Oehler G., MUhldorfer I.,Tschape H., Pathogenicity islands of virulent bac-teria: structure, function and impact on microbialevolution, Mol. Microbiol. 23 ( 1997) 1089-1097.

[57J Hofman P., Flatau G., Selva E., Gauthier M., LeNegrate G., Fiorentini C., Rossi B., Boquet P.,Escherichia coli cytotoxic necrotizing factor Ieffaces microvilli and decreases transmigrationof polymorphonuclea leukocytes in intestinal T84epithelial cell monolayers, Infect. Immun. 66( 1998) 2494-2500.

158] Horiguchi Y., Senda T., Sugimoto N., Katahira J.,Matsuda M., Bordetella brollchiseptica der-monecrotizing toxin stimulates assembly of actinstress fibers and focal adhesions by modifyingthe small GTP-binding protein Rho, J. Cell Sci.108 (1995) 3243-3251. 1 .

[59! Horiguchi Y., Inoue N., Masuda M., KashimotoT.. Katahira J.. Sugimoto N., Matsuda M., Bnr-eletell(i bt-oii(-hisepti(-ti dermonecrotizing toxininduces reorganization of actin stress fibersthrough dcamidation of Gln-63 of the GTP-hind-ing protein Rho, Proc. Natl. Acud. Sci. USA 94( 1 1)1)7 ) i)623-H626.

[6(l! Island M.D., Cui X., Foxman B., Marrs C.F.,Stamm W.E.. Stapleton A.E., Warren J.W., Cyto-toxicity of hcmolytic, cytotoxic nccrotizing fac-tor I-positive and -negative E.schcriclaia coli tohuman T24 bladder cells, Infcct. Immun. 66

(1998)3384-3389.

!61 ! Jacobson S.H., Katouli M., Tullus K.. BraunerA., Phenotypic differcnccs and characteristics ofpyelonephritogenic strains of Escherichia coliisolated from children and adults, J. Inlcct. 21 1( 1990) 279-286.

!62! Johnson J.R.. Stapleton A.E., Russo T.A., ScheutzF., Brown J.J., Maslow J.N.. Characteristics andprevalence within serogroup 04 of a .196-likcclonal group of uropathogcnic I:&dquo;scherichia coli

04:H5 containing the class I and class III allelesof papG, Infect. Immun. 65 (1997) 2153-2159.

[63! Lacerda H.M., Pullinger G.D., Lax A.J., Rozen-gurt E., Cytotoxic necrotizing factor I fromE.scherichia coli and dermonecrotic toxin fromBordetella bronchiseptica induce p21 (rho)-depen-dent tyrosine phosphorylation of focal adhesionkinase and paxillin in Swiss 3T3 cells, J. Biol.Chem. 272 (1997) 9587-9596.

[64] Le Bouguenec C., Bertin Y., AFA and F17 7adhesins produced by pathogenic Escherichiacoli strains in domestic animals, Vet. Res. 30(1999) 317-342.

!65] Lemichez E., Flatau G., Bruzzone M., Boquet P.,Gauthier M., Molecular localization of theEscherichia coli cytotoxic necrotizing factorCNFI cell-binding and catalytic domains, Mol.Microbiol. 24 (1997) 1061-1070.

!66] Lopez-Alvarez J., Gyles C.L., Occurrence of thevir plasmid among animal and human strains ofinvasive Escherichia coli, Am. J. Vet. Res. 41 1(1980)769-777.

167] Lopez-Alvarez J., Gyles C.L., Shipley P.L.,Falkow S., Genetic and molecular characteristicsof Vir plasmids of bovine septicemic E.scherichiacoli. J. Bacteriol. 141 ( 1980) 758-769.

1681 McLaren I., Wray C., Another animal Esc/!fnc/7Mco/i cytopathic factor, Vet. Rec. 119 ( 1986)576-577.

1691 Mainil J.G., Jacquemin E.. H6raL)lt F., OswaldE., Presence of pap-, sfa-, and afa-relatedsequences in necrotoxigenic Escherichia cnli iso-lates from cattle: evidence for new variants ofthe AFA family, Can. J. Vet. Res. 61 ( 1997)193-199.

!70J Nicoletti M., Superti F., Conti C., Calconi A.,Zagaglia C., Virulence factors of lactose-nega-tive Gscherichia coli strains isolated from chil-dren with diarrhea in Somalia, J. Clin. Microhiol.26 ( 1988) 524-529.

1711 Okuda J.. Fukumoto M., Takeda Y., NishihuchiM., Examination of diarrheagenicity of cytolethaldistending toxin: suckling mouse response to theproducts of the (-(ItABC genes of Shigella dnell-teriae, Infect. Immun. 65 ( 1997). 428-433

1721 Oswnld E., De Rycke J., A single protein of I 10 ()kDa is associated with the multinuclcating andnecrotizing activity coded by the Vir plasmid ofEscherichia coli. FEMS Microbiol. Lett. 56(1990)279-284.

1731 Oswald E.. Dc Rycke J.. Guillot J.F.. Boivin R..Cytotoxic effect of multinuclcation in HeLa cellcultures associated with the presence of Vir plas-mid in Escherichia coli strains, FEMS Micro-biol. Lctt. 49 ( 1989) 95-99.

1741 Oswald E., De Rycke J.. Lintermans P.. vanMuylcm K., Mainil J., Daube G., Pohl P., Viru-tence factors associated with cytotoxic necnriiz-ing factor type two in hovine diarrheic and sep-ticemic strains of Escherichia coli, J. Clin.Microbiol. 29 ( 1991 ) 2522-2527.

[751 Oswald E., Pohl P., Jacquemin E., LintermansP., Van Muylem K., O’Brien A.D., Mainil J.,Specific DNA probes to detect E.scherichia cnlistrains producing cytotoxic necrotising factortype I or type 2, J. Med. Microbiol. 40 ( 1994)428-434.

1761 Oswald E., Sugai M., Labigne A., Wu H.C.,Fiorentini C., Boquet P., O’Brien A.D., Cyto-toxic necrotizing factor type 2 produced by viru-lent Escherichia coli modifies the small GTP-

binding proteins Rho involved in assembly ofactin stress fibers, Proc. Natl. Acad. Sci. USA 91 I( 1994) 3814-3818.

!771 Oswald E., Tabouret M., Boivin R., De RyckeJ., Detection of Eschkrichia coli strains producingcytotoxic necrotizing factor type two (CNF2) byenzyme-linked immunosorbent assay, Vet. Micro-hiol. 40 (1994) 209-218.

1781 Penrith M.L., Henton M.M., Clay C.G., CNFI Itoxin-producing strains of Escherichia coli iso-lated from weaner pigs with necrotic enteritis inSouth Africa, Vet. Rec. 136 ( 1995) 493-494.

!79j Peres S.Y., Marches 0., Daigle F., NougayredeJ.P., Herault F., Tasca C., De Rycke J.. Oswald E.,A new cytolethal distending toxin (CDT) fromEscherichia coli producing CNF2 blocks HeLacell division in G2/M phase, Mol. Microbiol. 24(1997)1095-1107.

1801 Pohl P., Daube G., Mainil J., Lintermans P.,Kaeckenbeeck A., Oswald E., Virulence factorsand phenotypes of sixty-one strains of Escherichiacoli of bovine origin, producing cytotoxic necro-tising toxin type I, Ann. Rech. Vet. 23 (1992)83-91. 1 .

1811 Prada J., Baljer G., Dc Rycke J., Steinruck H.,Zimmcrmann S., Stephan R., Beutin L., Charac-teristics of alpha-hemolytic strains of E.scherichiacoli isolated from dogs with gastroenteritis. VetMicrobiol. 29 (1991) 59-73.

1821 Schmidt G., Sehr P., Wilm M., Selzer J., MannM.. Aktories K., Gin 63 of Rho is deamidated byE.scherichia coli cytotoxic necrotizing factor-I, I ,Nature 387 (1997) 725-729.

[83 ! Schmidt G., Selzer J., Lerin M., Aktories K., TheRho-deamidating cytotoxic necrotizing factor Ifrom Escherichia coli possesses transglutaminaseactivity. Cysteinc 866 and histidine 881 are essen-tial for cnzymc activity, J. Biol. Chem. 273 ( 1998)13669-13674.

1841 Smith H.W., A search for transmissible

pathogenic characters in invasive strains ofEscherichin coli: the discovery of a plasmid-con-trolled toxin and aplasmid-controlled lethal char-

acter closely associated, or identical, with col-icine V, J. Gen. Microbiol. 83 ( 1974) 95-1 I 1.

1851 Smith H.W., Observations on Escherichia coliin calves, in: Rutter J.M. (Ed.), Proceedings ofthe first seminar on ’Pathology’ in the EEC pro-gram of coordination of research in beef produc-tion. Perinatal ill-health in calves, EEC, Brussels,1975, pp. 47-59.

[86] Smith H.W., Transmissible pathogenic charac-teristics of invasive strains of Escherichia coli,J. Am. Vet. Med. Assoc. 173 ( 1978) 601-607.

1871 Sugai M., Kawamoto T., Peres S.Y., Ueno Y.,Komatsurawa H., Fujiwara T., Kurihara H.. Sug-inaka H., Oswald E., The cell cycle-specificgrowth-inhibitory factor produced by Acti-nohacillus actinomycetemcomitans is a cytolethaldistending toxin, Infect. Immun. 66 (1998)5008-5019.

1881 Swenson D.L., Bukanov N.O., Berg D.E., WelchR.A., Two pathogenicity islands in uropathogenicEscherichia coli J96: cosmid cloning and sam-ple sequencing, Infect. Immun. 64 (1996)3736-3743.

[ 89 ! Tabouret M., De Rycke J.. Detection of cytotoxicnecrotising factor (CNF) in extracts of Escherichiucoli strains by enzyme-linked immunosorbentassay, J. Med. Microbiol. 32 ( 1990) 73-81. 1.

1901 Terai A., Yamamoto S., Mitsumori K., OkadaY., Kurazono H., Takeda Y., Yoshida 0.,Escherichia coli virulence factors and serotypesin acute bacterial prostatitis, lnt. J. Urol. ( 1997)289-294.

!91 ! Wray C., Piercy D.W., Carroll P.J., Johnson C.T.,Higgins R., Bovine haemorrhagic colitis associ-ated with CNF+ and F6+ (987P) Escherichin coli,Vet. Rec. 131 (1992) 220.

1921 Wray C.. McLaren I.M., Carroll P.J., Escherichiacoli isolated from farm animals in England andWales between 1986 and 1991. Vet. Rec. 133

(1993)439-442.!93j Wray C., Piercy D.W., Carroll P.J., Cooley W.A.,

Experimental infection of neonatal pigs with CNFtoxin-producing strains of F,scherichia cnli, Res.Vet. Sci. 54 (1993) 290-298.

1941 Yamamoto S., Tsukamoto T., Terai A., KurazonoH., Takeda Y., Yoshida 0., Distribution of viru-lence factors in E.scherichia cnli isolated fromurine of cystitis patients, Microbiol. Immunol.39 (1995) 401-404.

!95! Yuri K.. Nakata K., Katae H., Yamamoto S..Hascgawa A., Distribution of uropathogcnic vir-uleiice factors among E,scherichia coli strains iso-latcd from dogs and cats. J. Vet. Mcd. Sci. 60(1998)287-290.