Cytophysiologic effects and molecular inhibition of a functional actin specific ADP ribosyltransferase CDT from clostridium difficile 2

1992, reported that most adults secrete anti-toxin A IgA antibody into the colonic lumen that could block toxin binding with intestinal surface receptors.. Since monoclonal antibodies ag

Chapter 1 Review of Literature

1.1 Clostridium difficile

Clostridium difficile is a spore-forming gram-positive bacilli The organism was first referred to as “the difficult Clostridium” in 1935 because of its fastidious and slow growth in culture (Hall and O'Toole 1935) C difficile produces an array of acid fermentation products which can be detected by gas-liquid chromatography (Hill, Osterhaut et al 1988) Its isolation from stool specimen is significantly enhanced in selective agar medium supplemented with cefoxitin, cycloserine, fructose and egg yolk called cycloserine-cefoxitin fructose agar (CCFA)

C difficile produces the volatile fatty acid isocaproic acid, the tyrosine metabolic by-product cresol and a yellow fluorescence on blood containing media (Bongaerts and Lyerly 1997)

p-The organism has been isolated from a variety of sources including humans especially in the hospital setting, farmyard, domestic animals, soil sites, swimming pool and tap water samples (Al Saif and Brazier 1996; Wilcox, Cunniffe et al 1996) Subterminal spore formation which makes this organism persistent and difficult to eradicate can be stimulated by sodium cholate and sodium taurocholate for improved recovery from clinical specimen Resistant spores survive in adverse conditions, hence serving as the vector by which the infection is spread by food handlers

or healthcare personnel (Samore, Venkataraman et al 1996) This explains why hospitalized elderlies and neonates become colonized with this bacterium

The pathogen can cause a spectrum of disease symptoms ranging from mild self-limited diarrhea called C difficile associated diarrhea (CDAD) to the formation of pseudomembrane leading to severe colitis called pseudomembranous colitis (PMC) and toxic megacolon (Kelly, Pothoulakis et al 1994) PMC was first described in 1893 as diphtheritic colitis and it was not until the 1970s that C difficile was implicated as the etiologic agent (Tedesco, Stanley et al 1974a; Larson, Parry et al 1977; Larson, Price et al 1978) PMC was earlier thought to be caused by viral infection or mucosal ischemia until it was realized that stool specimen from

patients contained a toxigenic factor with cytopathic effect in tissue culture cells (Larson, Parry et

al 1977) Afterwhich, C difficile was identified as the source of the cytotoxin (Bartlett, Chang et

al 1978; Larson, Price et al 1978), followed by reports on effective therapy using vancomycin (Keighley, Burdon et al 1978; Tedesco, Markham et al 1978)

1.2 Diagnosis

Although colonoscopy or sigmoidoscopy have been useful as diagnostic tools for colitis, these techniques must be used judiciously due to the invasiveness of procedure (Hurley and Nguyen 2002) The “gold standard” for diagnosis of C difficile infection is the cytotoxicity assay which detects toxin B The test has 94% to 100% sensitivity and 97% specificity (Bond, Payne et

al 1995; Bartlett 1998) However, it requires a tissue culture facility and 1-2 days to complete It should also be noted that toxin B is heat-labile, therefore, the shortest transport time and refrigeration of specimen is imperative so as to minimize proteolytic degradation of the toxin Growth of C difficile in egg yolk-enriched CCFA medium is a good adjunct but less specific than the cytotoxin assay Positive assay results have been confounded by significant proportion of asymptomatic hospitalized patients who are colonized with C difficile (George, Sutter et al 1979)

Rapid toxin testing methods can generate results within few hours, however, sensitivity is lower than the cytotoxicity assay at 85% with specificity at 100% Enzyme-linked immunosorbent assay (ELISA) is the most widely used method in clinical setting (Knoop, Owen

et al 1993; Lyerly and Wilkins 1995; Brazier 1998) Another test which uses latex particle agglunation (LPA) detects glutamate dehydrogenase produced by C difficile (Lyerly, Barroso et

al 1991) Caution should be observed however when interpreting LPA results because glutamate dehydrogenase produced by other anaerobes including Clostridium sporogenes, certain types of Clostridium botulinum and Peptostreptococcus anaerobius can cross react with antibody against the enzyme

Recently, Alfa et al (2002), compared various diagnostic methods and found that cell culture cytotoxin detection is most specific and the Triage C difficile test (TCT) as a more sensitive rapid screening test than ELISA TCT detects toxin A and glutamate dehydrogenase surface antigen directly from stool samples within the day of receipt This is the reason why cytotoxin test is recommended only for stools that required further testing Another rapid diagnostic test that is comparable to ELISA involves the use of polymerase chain reaction (PCR) with 96% sensitivity and 100% specificity Although primarily adopted in research settings, its application in clinical diagnosis has become popular

1.3 Epidemiology

Clostridium difficile is the most frequently implicated cause of antibiotic-associated diarrhea (AAD) and colitis (Frost, Craun et al 1998; Djuretic, Wall et al 1999) In the U.S., C difficile infection is estimated at 3 million cases of diarrhea and colitis annually Most cases are nosocomially acquired with only about 20,000 diagnosed from outpatient setting (Kelly and LaMont 1998) However, public incidence maybe underestimated as community-acquired diarrhea is not routinely investigated for the presence of C difficile or its toxin (Riley, Cooper et

al 1995) As disease incidence has been increasingly reported especially in healthcare facilities for the elders and large medical centers, C difficile is now recognized as a major cause of nosocomial diarrhea in industrialized countries (Lyerly 1993)

Approximately 15%-25% of hospitalized adults and debilitated patients carry the organism particularly those receiving chemotheraphy prior to surgery (Thomas, Bennett et al 1990; Privitera, Scarpellini et al 1991; Zimmerman 1991; Yassin, Young-Fadok et al 2001) CDAD incidence in ambulatory adults has been estimated at 7 to 12 cases per 100,000 person-years and up to 50% among infants and children (Bacon, Fekety et al 1988; Hirschhorn, Trnka et

al 1994; Levy, Stergachis et al 2000) The resistance of infants to C difficile infection may provide some answers to the discovery of new treatment and information on pathogenesis of the

organism (Lyerly and Wilkins 1995) Toxigenic strain is carried by 50% of asymptomatic infants with the protective mechanism involved still unclear (Lyerly and Wilkins 1995) Some argue that infants usually carry the weakly toxigenic strains belonging to serogroup F These strains have toxin A-/B+ phenotype, produce lower levels of toxin and not typically associated with adult disease Another proposed explanation is the immaturity of toxin receptors on infant enterocyte membrane (Eglow, Pothoulakis et al 1992)

Some research findings suggest that immunity against C difficile toxins may prove to be

an effective means of preventing infection Kelly et al (1992), reported that most adults secrete anti-toxin A IgA antibody into the colonic lumen that could block toxin binding with intestinal surface receptors Furthermore, IgG antibody has been detected in about 60% of children and adults in the U.S., implicating possible humoral immunity (Viscidi, Laughon et al 1983) This is supported by experimental results showing children with recurrent CDAD have lower serum levels of anti-toxin A IgG than age-matched controls Moreso, those with inadequate humoral immune response were more predisposed to relapse (Leung, Kelly et al 1991)

1.4 Pathogenesis

The development of CDAD is dependent on several factors (Fig 1.1) The first factor is acquisition of bacteria (Kelly and LaMont 1998) Since most patients become colonized during hospital or nursing home stay and develop asymptomatic sequelae, colonization of the colon is not sufficient for the development of disease Instead, previous or concurrent antibiotic therapy is the aggravating factor The organism competes with normal intestinal flora when the latter is disturbed by antibiotics which leads to overgrowth of C difficile and elaboration of toxins There

is a strong association between clindamycin use and development of CDAD (Tedesco, Barton et

al 1974b) while broad-spectrum penicillins and cephalosporins are most commonly implicated because of their widespread use (Nolan, Kelly et al 1987) Studies involving molecular typing reported that virulent C difficile strains produce asymptomatic colonization more often than

Figure 1.1 Development and possible outcomes of C difficile infection Modified from Abigail A Salyers and Dixie D Whitt, “Bacterial pathogenesis a molecular approach, 2ndedition (2002)”

High number of non-spore forming anaerobe over Normal Gut Flora

Microflora alteration

C difficile proliferates Symptoms

Cessation of

treatment

Relapse 10-20%

Death

Antibiotic Therapy

Asymptomatic

CDAD (Johnson, Clabots et al 1990; Shim, Johnson et al 1998), suggesting that extrinsic bacterial factors like host immunity and timing and dosage of antimicrobial exposure must be involved (Johnson and Gerding 1998)

As C difficile grows, the toxins are released upon autolysis then enter the host cell via receptor-mediated endocytosis and generalized pinocytosis In rabbit, glycoprotein receptors for toxin A on enterocyte membrane were found to be linked to nucleotide regulatory protein (Pothoulakis, La Mont et al 1991) Toxin A is an enterotoxin that causes excretion of fluid from bowel whereas toxin B is primarily cytotoxic causing disruption in the signal transduction pathway and disassembly of filamentous actin that leads to the collapse of cytoskeleton and cell rounding (Hecht, Potoulakis et al 1988) Bowel fluid released from damaged epithelial cells containing polymorphonuclear neutrophil (PMN), lymphocyte, serum protein, erythrocyte and mucus is inflammatory

The toxins can also invoke inflammation through their ability to act as chemoattractant for PMNs and stimulate release of mediators such as tumor necrosis factor alpha (TNF∝) and interleukin 1 and 6 (Pothoulakis, Castagliuolo et al 1993; Henderson, Wilson et al 1999) The infiltration and damage to the colonic mucosa result in the accumulation of fibrin, mucin, dead cells and leukocytes forming yellowish patches of separate lesions on the mucosal surface These eventually coalesce into a sheetlike layer called pseudomembrane that distinguishes PMC from other types of colonic infection (Price and Davies 1977; Kelly and LaMont 1998) (Fig 1.2A,B) PMC is a potentially lethal gastrointestinal disease characterized by exudative plaques with necrosis of the intestinal mucosal surface

1.5 Disease management

Aside from diarrhea, abdominal pain, tenesmus and fever are other common symptoms of

C difficile-mediated colitis (McClane and Mietzner 1999) The disease can be fatal as it can lead

to colonic perforation or systemic toxicity if left untreated The treatment of choice involves

discontinuance of use of offending antibiotic and commencement of efficacious drugs against C difficile such as oral vancomycin or oral metronidazole (Briceland, Quintiliani et al 1988; Peterson and Gerding 1990) Clindamycin, lincomycin, ampicillin or the cephalosphorins were involved in many cases of PMC and CDAD whereas aminoglycosides, trimethoprim-sulfamethoxazole, erythromycin and the fluoroquinolones were less likely causes (Silva, Fekety

et al 1984; Bingley and Harding 1987; McFarland 1998; Apisarnthanarak, Razavi et al 2002; Hurley and Nguyen 2002; Safdar and Maki 2002)

Once therapy is discontinued, relapses occur in 10 to 20% of cases due to failure to clear the organism and restore the normal microbiota In this case, various management approaches have been recommended like improvement on handwashing and use of barrier precautions such

as isolation of symptomatic patients (Samore 1999), fluid and electrolyte replacement and administration of agents that slows intestinal motility (i.e., Lomotil), slow and tapering vancomycin therapy (Tedesco, Gordon et al 1985), use of rifampin or cholestyramine (Tedesco 1982; Buggy, Fekety et al 1987), bacteriotherapy with fecal enemas (Tvede and Rask-Madsen 1989), oral administration of nontoxigenic C difficile (Seal, Borriello et al 1987), and treatment with the yeast Saccharomyces boulardii (Surawicz, Mc Farland et al 1989)

1.6 Virulence Factors

1.6.1 Large clostridial cytotoxins (LCT)

The most studied diseases caused by C difficile are those with symptoms caused by the largest known single-molecule bacterial toxins, toxin A and toxin B (Dove, Wang et al 1990) These are well-studied amongst the clostridial exotoxins and cytotoxins and are encoded within a 19.6 kb pathogenicity locus (PaLoc) of the C difficile chromosome (Fig 1.3) PaLoc contains a putative positive regulator gene tcdD, LCT genes tcdA and tcdB, a putative holin gene tcdE and a negative regulator gene tcdC The toxins have no recognizable signal sequence and do not appear

Figure 1.3 The pathogenicity locus (PaLoc) of C difficile VPI 10463 showing conserved regions (GenBank accession nos X51797, X53138, X92982, U25131, U25132) Several gene portions encode for conserved structural features including the glucosyltransferase or catalytic domains (striped block), nucleotide binding sites (solid block), hydrophobic transmembrane domains (checkered block), repeating units (open block) and binding domain for attachment to host cell receptor (speckled arrow) Solid circles represent the DXD motif which is part of the catalytic domain responsible for binding of Mn2+ whereas –SH symbolize conserved cysteines The length of genes are numerically indicated below the arrows (not drawn to scale) while arrowheads show the transcriptional direction and line segments represent the size of monocistronic and polycistronic transcripts The figure is not drawn to scale for simplicity.

to be proteolytically activated, as both toxins are released upon bacterial autolysis (Dove, Wang

et al 1990)

They share around 49% of amino acid sequence identity with extensive structural similarity in the C-terminal third consisting of small repeating subunits within larger units (Fig 1.3) This portion is involved in receptor-binding specifically to galactose-rich residues and has similarity to glucosyltransferases of Streptococcus mutans and Streptococcus sobrinus (GtfB, GtfC and GtfI) which can bind to carbohydrates (von Eichel-Streiber, Laufenberg-Feldmann et al 1992) Since monoclonal antibodies against the repeating subunits (amino acid residues starting

at 2097 and 2355) neutralize toxin A enterotoxic activity and inhibit its binding to carbohydrate receptors, the repeating subunit portion appears to be immunodominant Galα1-3Galβ1-4GlcNAc on human intestinal cells has been identified as the toxin A receptor and it was suggested to be the same receptor for toxin B (Lyerly and Wilkins 1995) Differences in receptor composition and distribution may contribute to the level of toxicity among intestinal cells

With about 50% homology, the N-terminal regions of LCTs are composed of a central hydrophobic domain that represents a membrane spanning region involved in receptor-binding, translocation of enzyme portion into the cell cytoplasm and intracellular processing It is a conserved domain with 4 cysteine residues and a putative nucleotide-binding domain that is involved in the glucosylation of G-proteins (Fig 1.3) Site-directed modification of toxin B histidine residue of the nucleotide-binding site to glutamine resulted in 90% loss of toxic activity (Aktories and Just 1995; Hofmann, Busch et al 1997) The enzymatic activity of toxins A and B was traced to a 63 kDa recombinant fragment located at around 516 to 542 residues of the N-terminal region These findings highlight the critical role of the N-terminal region in cytotoxicity (Barroso, Moncrief et al 1994)

Consistent to their considerable sequence homology, recent studies on both toxins suggest similar molecular action (Jander, Rahme et al 2000) Upon binding to membrane receptors and internalization, the toxins act as monoglucosyltransferases capable of modifying

and consequently inhibiting members of the Rho family proteins including Rac, Cdc42 and Rho subtypes RhoA, RhoB and RhoC The toxins can cleave UDP-glucose and attach a glucose moiety to residue Thr37 of Rho protein This modifies Rho from an active (UDP-glucose + Rho)

to an inhibited conformation (Glucose-Rho + UDP)(Fig 1.4)

Rho proteins are members of a subfamily of small GTP-binding proteins (G-protein) that act as “molecular switches” capable of regulating a number of essential functions in mammalian cell including cell adhesion, microfilament organization, nuclear signaling, pseudopod formation and re-shaping (intravasation) of phagocytes and many signal transduction pathways (Aktories and Just 1995) The toxins preferentially glucosylate the GDP-bound form of G-protein because its configuration exposes the threonine residue (Fig 1.4) In contrast, the threonine gets buried in the GTP-bound form that becomes inaccessible to toxin activity As a consequence of Rho glucosylation, GTPase activity of G-protein is reduced through intensified release of GTP which disrupts signaling control and many cellular processes This mechanism has been exploited by molecular biologists as a useful tool the studies of cytoskeletal dynamics

GTP-bound G-protein

effect Signal

Weak binding of GTP, Low GTPase activity

Glc

GTP

Human analogs of this core carbohydrate have been identified and designated as X antigen (Galβ1-4[Fucα1-3]GlcNAcβ1-3Galβ1-4α[Glc]), Y antigen with the highest affinity to toxin has

an extra α1-2-linked fucose to X antigen and I antigen with 2 core structures (Galβ1-4GlcNAc) per oligosaccharide (Tucker and Wilkins 1991) Carbohydrate moieties similar to X antigen are expressed on secretory component of immunoglobulin and PMNs during inflammation PMNs initially attach to these receptor molecules on the surface of endothelial lining prior to extravasation Thus, a similar mechanism may explain the chemotactic and binding properties of the toxin to PMN

Other toxin A receptors on human epithelial cell may be involved since treatment with galactosidase which does not affect the X,Y,I antigens (Lewis group) was found to reduce binding (Smith, Cooke et al 1997) The receptor was demonstrated to be a glycoprotein when protease treatment likewise resulted in partial binding and later identified as a sucrase-isomaltase glycoprotein on rabbit ileal brush border (Pothoulakis, La Mont et al 1991; Pothoulakis, Gilbert

α-et al 1996) Eventually other molecules such as substance P receptors were identified and shown

to be essential in causing enteritis (Pothoulakis, Castagliuolo et al 1998) Through electron microscopy, internalization of toxin A was observed via its localization on clathrin-coated pits Moreso, an acidic environment was required for its release from the endosome (Henriques, Florin

et al 1987; Fiorentini and Thelestam 1991) Endosomal acidification leads to increased hydrophobicity, revelation of toxin transmembrane structure, membrane insertion and translocation (Qa'Dan, Spyres et al 2000)

Toxin A acts as an enterotoxin when injected into ligated ileal loops inducing release of bloody, viscous secretion upon mucosal damage (Triadafilopoulos, Pothoulakis et al 1987) Similar to toxin B, pertussis and diphtheria toxin however, toxin A is an intracellular-acting cytotoxin which is internalized into target cell (Kushnaryov and Sedmak 1989) Various mammalian cell lines were susceptible to toxin A including Chinese Hamster Ovary (CHO), human cervix (HeLa), mouse adrenocortical (Y1), human larynx (Hep-2), human lung fibroblast

(MRC-5) and other cell lines (Donta, Sullivan et al 1982; Henriques, Florin et al 1987; Lima, Lyerly et al 1988; Kushnaryov and Sedmak 1989; Fiorentini, Malorni et al 1990)

Many cellular effects of the toxin were observed For example, cytoskeletal collapse caused cell rounding and marginalization of nucleus in HeLa and CHO cells After 3 h of toxin A treatment, parallel filament bundles of 11 nm diameter were transiently visible in the nuclei with extensive development of golgi apparatus, smooth endoplasmic reticulum and lysosomes indicative of disturbance in synthetic and secretory functions of the cells (Kushnaryov and Sedmak 1989; Fiorentini, Malorni et al 1990) In IEC-6 cells, surface blebbing and nuclear fragmentation ensued after early cytoskeletal disaggregation (Fiorentini, Donelli et al 1993) Intoxication of T8 cells showed increased tight junction permeability while induction of interleukin 8 (IL-8), apoptosis and cell detachment were evident in human intestinal cells (Hecht, Potoulakis et al 1988; Mahida, Makh et al 1996) Other observed biological effects were induction of chloride secretion, actin condensation leading to membrane retraction and decline in membrane potential, morphological changes of mitochondria and disaggregation, abrupt increase

in cellular ATP levels (Moore, Pothoulakis et al 1990; He, Hagen et al 2000) Overall, these essential physiologic functions of cells were hindered directly through toxin enzymatic action or indirectly as a result of Rho monoglucosylation or cytoskeletal disassembly

The UDP-glucose hydrolase and glucosyltransferase activities of toxin A have been attributed to the first 695 (Faust, Ye et al 1998) or 562 amino acid residues of the N-terminus (Ciesla and Bobak 1998) Similarly, glucosyltransferase activity of toxin B was detected within the first 900 residues (Hofmann, Busch et al 1997) owing to high sequence homology of LCTs The catalytic portion contains a conserved DXD motif flanked by hydrophobic regions (Wiggins and Munro 1998) Amino acid residues composing this region are likely essential to function since mutations in Asp286 and Asp288 of C sordelii lethal toxin which is 90% homologous to toxin B (Just, Selzer et al 1996), inhibited both glycohydrolase and glucosyltransferase activities and attachment of azido-UDP-glucose to the N-terminus of toxin (Busch, Hofmann et al 1998)

The DXD motif was found to be important in Mn binding to allow the correct positioning of the cosubstrate UDP-glucose which is subsequently cleaved and transferred and attached to Thr-37 of Rho

1.6.1.2 Toxin B

Toxin B is similarly large with its molecular weight at 270 kDa and pl at 4.1 (Sullivan, Pellett et al 1982) It is non-enterotoxic, non-cytotonic but a more potent cytotoxin than toxin A that causes cell rounding but not fluid hypersecretion (Pothoulakis, Barone et al 1986) Like toxin A, toxin B is also composed of 3 major structural domains, has repetitive oligopeptides suggestive of membrane binding via glycoprotein receptors and the presence of a flanked hydrophobic region of 172 residues which was proposed to function for intracellular translocation (von Eichel-Streiber, Laufenberg-Feldmann et al 1992) A drop in intra-endosomal pH was also observed, such that toxin B may assume a hydrophobic structural fold necessary for membrane insertion and translocation, a process requiring the presence of Ca2+ and calmodulin (Caspar, Florin et al 1987; Gilbert, Pothoulakis et al 1995; Qa'Dan, Spyres et al 2000) Therefore, similarity in physical properties between toxins A and B is not surprising due to their conserved nature In addition, toxin B likewise functions in the UDP-glucosylation of small GTP-binding proteins Rho, Rac and Cdc42 (Just, Selzer et al 1995a) but not Rap which is a toxin A substrate (Chaves-Olarte, Weidmann et al 1997) The enzymatic domain lies within the N-terminus (located at the first 546 or 467 amino acid residues) possessing a conserved residue Trp102 which

is crucial for UDP-glc binding (Busch, Hofmann et al 1998)

Some of the more prominent effects of toxin B are directed towards immune cells Macrophages were demonstrated to release lipooxygenases including LTB4 and tumor necrosis factor (TNFα) upon intoxication (Siffert, Baldacini et al 1993; Souza, Melo-Filho et al 1997) Moreso, human monocytes released inflammatory mediators IL-1, IL-6 and TNFα which are potent PMN chemoattractants Therefore, it is evident that toxin B works through primary

induction of the release of proinflammatory cytokine by myeloid cells prior to biomolecular disruption Expectedly however, similar cellular effects of toxin A was observable on toxin B treatment For example, toxin B also induced calcium influx for actin disassembly (Gilbert, Pothoulakis et al 1995); nuclear fragmentation and chromatin condensation which are typical features of apoptosis (Fiorentini, Fabbri et al 1998); loss of anchorage due to actin, vinculin, talin and vimentin reorganization (Ottlinger and Lin 1988; Ciesielski-Treska, Ulrich et al 1989); inhibition of protein synthesis (Pothoulakis, Barone et al 1986); potocytotic surface blebbing characterized by the presence of bleb matrix with ribosomes but devoid of organelles and morphological alterations such as retraction of cytoplasmic projections and rounding (Wedel, Toselli et al 1983) However, unique effects of toxin B (100X more cytotoxic than toxin A) were also reported (Riegler, Sedivy et al 1995; Chaves-Olarte, Weidmann et al 1997) Toxin B was also observed to prevent proper signaling in human embryonic kidney cells via muscarinic acetylcholine receptor and in rat basophilic leukemia cells via phospholipase D receptor (Ojio, Banno et al 1996; Schmidt, Rumenapp et al 1996)

1.6.1.3 The Pathogenicity Locus (PaLoc)

1.6.1.3.1 Genetic profile of the PaLoc

In the late 1980s, a small fragment of toxin A gene (0.3 kb) was expressed from lambda gt11 library with the unstable product found to be capable of causing CHO cell elongation (Muldrow, Ibeanu et al 1987) Thereafter, the use of a 4.7 kb PstI restriction fragment that encodes a portion of tcdA as probe for chromosomal walking of VPI 10463 genome, led to the discovery of both tcdA and tcdB sequences and their proximity (Barroso, Wang et al 1990; Dove, Wang et al 1990) The tcdA gene is 8130 bp long of 26.9% GC content which encodes for 2710 amino acid residues (Sauerborn and von Eichel-Streiber 1990) Its fusion protein product showed lethal cytotoxic and enterotoxic activities (Phelps, Lyerly et al 1991) On the other hand, tcdB of

7098 nucleotides and 27.4% GC composition, encodes for 2366 amino acid residues whose

expressed protein exhibiting cytotoxicity and reactivity to toxin B antisera (Barroso, Wang et al 1990) Likewise, von Eichel-Streiber et al (1987; 1989; 1992) cloned the LCT genes which were proposed to have arisen as a result of gene duplication due to their extensive sequence identity

Recently, a toxinotyping scheme was proposed for the standardization of C difficile strain serogrouping (Rupnik, Avesani et al 1998) Classification was further streamlined at the first international C.difficile symposium where unified nomenclature for toxin genes and their products were drafted, with relevant suggestions such as the use of tcd series in the designation of PaLoc genes and renaming of tcdD as tcdR referring to the role of the protein product in regulation (Rupnik, Dupuy et al 2005)

Accordingly, toxin A and toxin B share common structural motifs (Fig 1.3) like the terminal third active site cleft with DXD motif and possessing enzymatic properties, a central domain with 4 cysteines suggested to be critical for enzymatic function, a membrane-spanning hydrophobic region which is necessary for toxin transport (von Eichel-Streiber, Meyer zu Heringdorf et al 1995) and the C-terminal domain which is comprised of repeating units of aromatic amino acids tyrosine and phenylalanine that function for carbohydrate receptor binding (von Eichel-Streiber and Sauerborn 1990) The C-terminal portion was also observed to be immunodominant portion of the toxin (von Eichel-Streiber, Harperath et al 1987) While the importance of the central hydrophobic region in toxin B in cytotoxicity and its function for protein translocation in adenylate cyclase of Bordetella pertussis (Hanski 1989), alpha hemolysin

N-of E.coli and leukotoxin N-of Pasteurella haemolytica has been demonstrated (Strathdee and Lo 1987), the role of this membrane-spanning region in intracellular transport of LCTs is yet to be determined

Recent evidences have shown that bacterial virulence factors could be acquired through horizontal transfer of genes often clustered into genetic blocks or modules called pathogenicity island (Hammond and Johnson 1995; Groisman and Ochman 1996; Hacker, Blum-Oehler et al 1997) Hammond et al (1995), first reported the cluster of toxigenic elements in VPI 10463

chromosome composed of LCT genes and three smaller accessory orfs In 1997, Hundsberger et

al (1997), designated the block as the pathogenicity locus (PaLoc) and the genes as tcdA to tcdE (Fig 1.3) with tcdD, tcdB, tcdE and tcdA transcribed in the same direction in that order and tcdC (adjacent to tcdA) in the opposite direction The 3’ end of tcdB is 1350 bp upstream of the tcdA translation start site (Fig 1.3) tcdE which is composed of 501 bp and flanked by tcdB and tcdA,

is 122 bp downstream of tcdB stop codon Located immediately upstream of tcdB is the 555 bp tcdD gene that codes for TcdD protein with high lysine content of its C-terminus In nontoxigenic C difficile strains, the PaLoc is replaced with either a 115 bp nucleotide segment having transcriptional terminator features (Braun, Hundsberger et al 1996) or a 127 bp segment flanked by AT-rich inverted repeats characteristic of an insertion element (Hammond and Johnson 1995) which are indicative of potential sites for genetic exchange

1.6.1.3.2 Regulation of PaLoc genes

The divergent capabilities of C difficile strains in their level of toxin production despite identity of the PaLoc region, implicate heterogeneity in the regulatory process One PaLoc gene proposed to be involved in regulation is tcdD (txeR) TcdD displays a helix-turn-helix configuration and shares significant identity with UviA, a promoter P1 DNA-binding bacterial response regulator (Moncrief, Barroso et al 1997) and positive regulators of C botulinum neurotoxin BotR (Hauser, Eklund et al 1994) and C tetani neurotoxin TetR (Marvaud, Eisel et

al 1998) Indeed, it was observed that the expression of toxin A and toxin B promoter repeating units (ARU) increased by 500 and 800-fold respectively, when tcdD was supplied in trans Furthermore, there are long intergenic distances between the transcriptional start sites (TSS) and start codons of both toxin A (169 bases) and toxin B (239 bases) promoter regions typical among clostridial sequences These are strong indication of its role in the positive regulation of toxin expression However, when grown under conditions of limited nutrient, the level of toxin A and toxin B was observed to increase (Yamakawa, Karasawa et al 1996) Instead of acting as

transcriptional regulator therefore, TcdD may function as a sigma factor that promotes toxin synthesis in response to stress The PaLoc gene suggested to be involved in negative regulation is tcdC (Hundsberger, Braun et al 1997) It contains 695 bp located downstream of tcdA which encodes for an acidic protein with various lengths of repetitive residues

The proposed regulatory process starts with tcdD as being transcribed at high levels during the early exponential growth phase which was immediately followed by increased expression of tcdB, tcdE and tcdA at the stationary phase (Hundsberger, Braun et al 1997) Therefore, it was deduced that TcdD acts as a positive regulator of PaLoc gene expression As the bacteria enter the late stationary phase, TcdC expression becomes elevated to counter TcdD action and antisense effect of transcripts from other genes Toxin synthesis was most pronounced during the early and mid-exponential growth phase, thus, the high toxin level at the stationary phase is due to toxin accumulation Finally, several studies have demonstrated the combined monocistronic and polycistronic mode of PaLoc transcription through detection of individual and readthrough mRNAs (Hammond, Lyerly et al 1997; Hundsberger, Braun et al 1997; Dupuy and Sonenshein 1998) These connote inherent multiplicity in promoter sequences and discrepancy in the expression level of different genes comprising the PaLoc

6-difficile, C perfringens, and C spiroforme can all cause gastrointestinal diseases in humans as well as animals (Borriello and Carman 1983; Braun, Herholz et al 2000; Stoddart and Wilcox 2002), thus implying common evolutionary lineage

1.6 2.1 ADP-ribosyltransferase

Aside from LCTs, a number of C difficile strains simultaneously produce an ribosyltransferase (ADPRT) designated as CDT toxin (Perelle, Gibert et al 1997a) The enzyme can mediate catalysis of nicotinamide adenine dinucleotide (NAD) and attachment of ADP-ribosyl group to various protein substrates Since several bacteria produce ADPRT (Table 1.1),

ADP-Figure 1.5 The cdt locus of C difficile CD196 showing conserved regions (GenBank accession no L76081) Several gene portions encode for conserved structural features including the catalytic domain (diagonally-striped block), nucleotide binding site (solid block), hydrophobic transmembrane domains (checkered block), docking site (vertically-striped block), oligomerization domain (horizontally-striped block) and binding domains (speckled arrow) Arrows represent relative gene sizes (not drawn to scale) and point to the direction of transcription Relative gene sizes are indicated numerically below the arrows

they have been classified into four groups based on the type of substrate they act on The first group includes the diphtheria toxin (DT) of Corynebacterium diphtheriae (van Ness, Howard et

2631 bp

1392 bp

al 1980) and exotoxin A (ETA) of Pseudomonas aeruginosa (Wretlind, Bjorklind et al 1987) These toxins are involved in attaching the ADP-ribose to eukaryotic elongation factor 2 (EF-2) which is active in protein synthesis Members of the second group ADP ribosylates membrane associated G-proteins This group is comprised of cholera toxin (CT) produced by Vibrio cholerae, pertussis toxin (PT) of Bordetella pertussis, Escherichia coli heat-labile enterotoxins (LT1 and LT2) and Pseudomonas exoenzyme S (ExoS) (Moss and Vaughan 1988) Toxins that act on low molecular weight GTP-binding protein Rho comprise the third group These include the C3 enzyme of Clostridium botulinum (Aktories, Rosener et al 1988; Just, Mohr et al 1992), Clostridium limosum Sa and Sb exoenzymes (Just, Mohr et al 1992) and epidermal cell differentiation inhibitor (EDIN) of Staphylococcus aureus (Sugai, Hashimoto et al 1992) The fourth group are actin-specific ADPRTs which include the C2 toxin of Clostridium botulinum types C and D (Aktories, Barmann et al 1986b; Aktories, Weller et al 1987), iota toxin of Clostridium perfringens type E (Simpson, Stiles et al 1987), Clostridium spiroforme toxin (Stiles and Wilkins 1986a; Simpson, Stiles et al 1989), Clostridium difficile CDT toxin (Popoff, Rubin

et al 1988b) and the Bacillus cereus vegetative insecticidal proteins, VIP1-VIP2 (Seungil, Craig

et al 1999) These binary toxins which act independently as catalytic and membrane-binding proteins are distinct from the classical A-B toxins like cholera toxin that needs to assemble to form a functional complex with two different subunits (Madshus and Stenmark 1992)

1.6 2.2 Actin as substrate

Being one of the most abundant proteins in eukaryotic cells, actin plays a myriad of functions that maintains homeostatic condition and normal physiology (Pollard 1986) Its propensity towards reversible shift from the monomeric (G-actin) to filamentous (F-actin) form and vice versa makes the accomplishment of its versatile roles possible A number of its functions include organization

of organelles, cell division, protein translocation, support, mobility, secretion, phagocytosis and more

Table 1.1 Bacterial toxins produced as binary or preformed A-B structures

Bacteria Toxin (typea) Activity (substrate) Reference

adenylate cyclase and metalloprotease

a (B) represents the binary type while (P) represents preformed type A-B toxin

b

ADPRT is ADP-ribosylating toxin

Actin is a single polypeptide comprised of 375 amino acid residues having a molecular mass of 42 kDa Crystallographic data showed that the molecule is made up of four subdomains with one nucleotide binding cleft for ATP or ADP which attracts a divalent cation like Mg2+(Kabsch, Mannherz et al 1990) Owing to its highly conserved nature, variability in amino acid composition was reported at only 6% (Korn 1982) In mammalian cells, actin was classified into

6 major groups namely: the skeletal muscle α-actin, cardiac muscle α-actin, smooth muscle α- and γ-actin, and cytoplasmic β- and γ-actin, which comprise the 3 isoforms α, β and γ, distinguished by their isoelectric properties (Vandekerckhove and Weber 1979) Alpha-actin is the most acidic while β-actin is partially acidic and the γ isoform is most basic Iota toxin which

is a close homologue of CDT was shown to to act on all isoforms (Schering, Barmann et al 1988) whereas C2 only ADP-ribosylates the β and γ isoforms (Aktories, Ankenbauer et al 1986a; Ohishi and Tsuyama 1986)

The actin filament is a polar molecule having a fast-growing (barbed), positive end a slow growing (pointed), negative end with the ATP-bound actin being more stable than ADP-bound form in F-actin (Carlier 1991) The structural shift is regulated by several factors including its interaction with actin-binding anchor proteins like α-actinin, profilin, gelsolin and β-thymosin (Pollard and Cooper 1986) Therefore, the protein assumes a dynamic configuration as simultaneous polymerization and depolymerization process takes place

ADPRTs of the 4th group labels the monomeric actin form specifically at a conserved acceptor amino acid Arg177 which was revealed through site-directed mutagenesis of the enzymatic component (Aktories, Barmann et al 1986b; Schering, Barmann et al 1988) Since the ADP-ribosylation site is located at the axis of the F-actin helix, the bulky ADP-ribose group attached to Arg177 poses considerable steric hindrance to the polymerization process The ADP-ribose moiety behaves like a capping protein that blocks the positive end the filament Furthermore, ADP-ribosylation inhibits the actin ATPase activity These disturbances may have critical consequences such as breakdown of cytoskeletal network and deterioration of cellular protein movement and function

1.6.2.3 Biology of actin-specific ADPRT

CDT and other ADPRTs employ a binary mode for intoxication of eukaryotic cells The binary nature was first illustrated through the use of cross-reacting and neutralizing antibodies against C spiroforme toxin (Stiles and Wilkins 1986a; Stiles and Wilkins 1986b) The properties

of the binary components were later characterized based on electrophoretic mobility and found to

be nontoxic individually but potently cytotoxic in combination (Stiles and Wilkins 1986a) Thereafter, iota Ia was discovered to be an ADP-ribosyltransferase whose substrate is the actin monomer (Simpson, Stiles et al 1987; Schering, Barmann et al 1988) while Ib mediates translocation of Ia into cells (Stiles, Hale et al 2000; Blocker, Behlke et al 2001; Richard, Mainguy et al 2002)

The functional toxin is composed of a precursor enzymatic or A protein component and a translocation or B protein component forming an A-B complex (Table 1.1) This differs form other bacterial binary toxins that engage target as a preformed holotoxin with an A-B structure (Table 1.1) Intoxication by CDT and C2 initially occur through the binding of the B component

on target cell membrane, then oligomerization into heptamer on the cell surface However, initial processing of the B monomer by proteases (Ohishi and Miyake 1985; Simpson 1989) is required before attachment to membrane receptors and formation of complexes Thereafter, the B-receptor complex functions as a docking site for the enzymatic A molecule The aggregate is subsequently translocated into the cytosol via an acidified endosomal vesicle (Simpson 1989) that is optimal for toxin to assume a conformation capable of membrane insertion and translocation Cytoplasmic toxin can then inhibit various normal cell physiologic functions primarily those involving cytoskeletal structures

al 2003) is still not clear

PaLoc toxin genes tcdA and tcdB were shown to be transcribed both from specific promoters and from promoters of upstream genes with the gene-specific transcripts represented the majority of toxin gene mRNAs and that their expression is subject to catabolite repression by glucose (Dupuy and Sonenshein 1998) This glucose effect was general to many toxinogenic strains having varying levels of toxin production Most contributory to PaLoc toxin synthesis is tcdR (formerly tcdD) which was found to act as RNA polymerase sigma factor In addition, external factors as temperature (Mani, Lyras et al 2002) and endogenous factors like cortocosteroids (Castagtliuolo, Karalis et

gene-al 2001), cysteine (Karlsson, Lindberg et gene-al 2000), intracellular calcium and NF-kappa

B (Jefferson, Smith et al 1999) were reported to have considerable influence on toxin expression

Unlike the regulation on the expression of PaLoc genes, information on cdt regulation is limited However, ample reports have been published on the action of a plasmid pX01-encoded

B anthracis atxA gene producing a 56 kDa protein (Dai, Sirard et al 1995) as a positive regulator for lethal toxins, capsule and other genes encoded on plasmids pX01 and pX02 (Guignot, Mock

et al 1997; Hoffmaster and Koehler 1997) External factors such as pH and temperature also affect the control of protein synthesis (Koehler 2002) As less virulent C difficile strains were observed to be devoid of atxA (Koehler 2002), it is plausible that C difficile produce a similar regulatory protein to enhance CDT secretion and pathogenicity

1.6.2.5 Conserved ADPRT structure and function

Unlike the binding component of ADPRTs, the enzymatic component has been reported

to have substantial homology particularly at important catalytic regions of the toxin Accordingly, several protein structures necessary for NAD binding and hydrolysis were found conserved across different families of ADPRT (Domenighini, Magagnoli et al 1994) Based on

studies involving structural modeling and comparative sequence analysis, ADPRTs were classified into the cholera toxin (CT) group which includes CDT and most bacterial mono-ADP-ribosyltransferases and the diphtheria toxin (DT) group comprised of DT, ETA and eukaroyotic poly-ADP-ribosyltransferases (Domenighini and Rappuoli 1996)

X-ray crystallographic data on A-B type ADPRTs identified a conserved catalytic cavity comprising around 100 residues with consensus sequences essential for NAD catalysis The site forms an α-helix bent over a β-strand structural conformation containing 3 predominantly conserved regions (Domenighini, Magagnoli et al 1994; Takada, Lida et al 1995) The first region possesses a nucleophilic residues arginine or histidine which is preceded by an aromatic amino acid The CT group contains the conserved Arg while the DT group has the conserved His The Arg/His side chain is dictated by the aromatic side chain to assume a distinct arm that extends into the cavity The region was reported to be participate in electron transfer (Okazaki and Moss 1994) and hydrogen bonding with NAD substrate (Takada, Lida et al 1995) The aromatic amino acid-rich region 2 is 50 to 75 residues downstream of region 1and was suggested

to facilitate NAD binding It is part of the β-strand scaffold of the catalytic cavity containing the arom-ph-Ser-Thr-Ser-ph consensus, where arom and ph stands for partially variable aromatic and hydrophobic residues, respectively The third region consists of essential tandem Glu residues of the Glu/Gln-X-Glu motif in the CT group or the Tyr-X-Tyr motif in the DT group (Okazaki and Moss 1994) This region was proposed to mediate hydrogen binding or salt bridge formation with Arg residue with the Glu residues forming the critical structure for NAD binding and activation In addition, adjoining acidic residues play adjunct roles in hydrogen binding with the 2’ hydroxyl ribose in NAD (Domenighini, Montecucco et al 1991) Overall, the Glu and His/Arg residues comprise the NAD binding region before catalysis occurs A diagrammatic illustration of the conserved motifs of the nucleotide cleft was illustrated by Nagahama et al 2000 (Fig 1.6)

Figure 1.6 Active-site structure of E coli heat-labile toxin (LT) The Arg-7 (R7),

Ser-61 (SSer-61), Thr-62 (T62), Ser-63 (S63), Glu-110 (E110), and Glu-112 (E112) side chains are shown in bold Sequence was taken from porcine LT (SWISSProt accession no.,

P06717) The residues in bold are those that are conserved in ADP-ribosylating enzymes

The structure and critical residues were also found conserved among binary type ADPRT

to which CDT belongs Recently, the enzyme structure of B cereus VIP, C botulinum C3 exoenzyme and C perfringens iota toxin were reported (Han, Craig et al 1999; Han, Arvai et al 2001; Tsuge, Nagahama et al 2003) Common structural motifs were described for the N- and C-terminal domains of the enzymatic protein which are structurally similar but with limited sequence identity The comparative domain structure was illustrated by Tsuge et al (2003) for iota Ia toxin (Fig 1.7) Both domains consist of a β–sandwich core formed by a 5-stranded mixed β-sheet and a 3-stranded anti-parallel β–sheet Arranged around the β–sandwich core are consecutive α–helices The C-domain cleft where NAD binds also exists in the N-domain, the difference between Ia, C3 and VIP2 lies within the C-domain occluding loop region of the ARTT motif which includes the EXE motif (Glu378 and Glu380) While conformation of Glu380 is the same, Glu378 orientation on the flexible loop upon NAD binding is different It was suggested then that the ARTT motif is important for acceptor recognition (Menetrey et al 2001)

Fig 1.7 A, Stereo view of the Ia complex with NADH (in ball and stick) N-terminus in purple and C-terminus in red while NAD cleft was shown by black arrow The diagram was generated using MOLSCRIPT and Raster3D B, N-domain and C-domain structures α–helices are colored yellow and –sheets are colored purple The secondary structures are labelled α3‘ 61-66 is an additional helix at the N-domain not found in VIP2 Reprinted from Tsuge et al 2003

Furthermore, the C-domain of VIP2 showed significant homology with that of iota Ia (39.9%) and Rho-ADP-ribosylating exoenzyme C3 (30%) from C botulinum but not the N-domain (Han, Craig et al 1999; Han, Arvai et al 2001; Tsuge, Nagahama et al 2003) Indeed, the N-domain of Ia has an additional α–helix (α3’:61-66) compared with VIP2 (Fig 1.7B) The structural similarity of the C-domain (Fig 1.8) and the disparity in the N-domain appear reasonable since ADP-ribosylation of actin molecule at the C-domain must be conserved while N-domain has evolved to function for the recognition of target cell membrane receptors

A

B

Figure 1.8 Stereo view of superposed NAD binding site of ADPRT C-domains A Binary iota Ia C-domain (atoms are represented in standard colors for residues) against binary VIP2 C-domain (green atoms) B Binary iota Ia C-domain against A-B type diphtheria toxin (light blue atoms) The conserved residues are indicated The stereo view of the density map was produced using MOLSCRIPT and Raster 3D and modified from Tsuge et al (2003)

A couple of modular enzymatic processes for bacterial mono-ADPRTs has been proposed The first one follows an SN2-like mechanism which involves a nucleophilic attack on NAD, displacement of nicotinamide and structural modification that is optimal for subsequent interaction with actin (Wilson, Reich et al 1990; Takada, Lida et al 1995) The process is a single-step concerted reaction with no intermediate and a single transition state involving simultaneous displacement of the leaving group and attack of the nucleophile group A representative stereochemical reaction is shown in figure 1.9A

A second pathway had been proposed describing an SN1-type mechanism for the dissociation of the nicotinamide group creating a positively charged oxocarbenium intermediate which is stabilized by the tandem Glu consensus (Bell and Eisenberg 1996) The ADP-ribose remains docked in the cavity until transfer onto actin SN1 reaction is multi-step with several

intermediates and transition states It involves a slow rate-limiting step which results in loss of the leaving group generating an intermediate carbocation that readily undergoes a rapid reaction with the nucleophile A representative stereochemical reaction is shown in figure 1.9B

1.6.3 Other virulence factors

Interestingly, virulence of various C difficile strains differ despite similarity in the amount of toxin produced This suggests the presence of other factors that contribute to disease causation such as proteins for bacterial reproduction, growth or attachment to host receptors to name a few Indeed, some strains produce thin capsule whose synthesis is regulated by the presence of certain metabolites like glucose (Borriello, Davies et al 1990; Bongaerts and Lyerly 1997) Others synthesize antiphagocytic factors like capsular polysaccharide which protects the offending pathogen from opsonizing antibodies which are important for effective innate host defense (Dailey, Kaiser et al 1987) A highly virulent C difficile strain was found to adhere tightly to hamster gut mucosa compared to an avirulent strain by using fimbrial structures at 6 µm length and 9 nm in diameter for efficient host cell adhesion (Borriello, Davies et al 1988) Such surface proteins act as adhesins whose interactive properties with mammalian cells were revealed

by different methods such as electron microscopy and immunohistochemistry (Drudy, O'Donoghue et al 2001; Calabi, Calabi et al 2002) Adherence increases the potency of most virulence proteins by virtue of promoting proximity and hence immediate effect to target cells However, environmental and host factors such as variation in the expression of intestinal receptors influence infection success rate resulting in different clinical outcome and prognosis

Adherence proteins used for target cell attachment include two paracrystalline surface layers or S-layers P36 and P47 (Cerquetti, Molinari et al 2000) The S-layers are superimposed with the outer lattice having symmetrical square configurations while the inner portion assumes hexagonal lattices Their precursor protein called surface layer protein A is encoded by slpA and showed interstrain variation (Waligora, Hennequin et al 2001) Moreso, glycoprotein subunits