Cytotoxic necrotizing factor type 1 (CNF1) and CNF2 are highly homologous toxins that are produced by certain pathogenic strains of Escherichia coli. These 1,014-amino-acid toxins catalyze the deamidation of a specific glutamine residue in RhoA, Rac1, and Cdc42 and consist of a putative N-terminal binding domain, a transmembrane region, and a C-terminal catalytic domain. To define the regions of CNF1 that are responsible for binding of the toxin to its cellular receptor, the laminin receptor precursor protein (LRP), a series of CNF1 truncated toxins were characterized and assessed for toxin binding. In particular, three truncated toxins, ⌬N63, ⌬N545, and ⌬C469, retained conformational integrity and in vitro enzymatic activity and were immunologically reactive against a panel of anti-CNF1 monoclonal antibodies (MAbs). Based on a comparison of these truncated toxins with wild-type CNF1 and CNF2 in LRP and HEp-2 cell binding assays and in MAb and LRP competitive binding inhibition assays and based on the results of confocal microscopy, we concluded that CNF1 contains two major binding regions: one located within the N terminus, which contained amino acids 135 to 164, and one which resided in the C terminus and included amino acids 683 to 730. The data further indicate that CNF1 can bind to an additional receptor(s) on HEp-2 cells and that LRP can also serve as a cellular receptor for CNF2.Cytotoxic necrotizing factor type 1 (CNF1) is produced by many strains of uropathogenic Escherichia coli (UPEC), which are agents that are responsible for the majority of uncomplicated urinary tract infections (9). CNF1 is a 115-kDa cytoplasmic protein that is a member of a family of toxins that target small GTPases. Specifically, CNF1 deamidates glutamine 63 of RhoA and glutamine 61 of Rac1 and Cdc42, modifications that result in constitutive activation of these small GTPases (1). This activation leads to the formation of stress fibers and focal adhesions (RhoA), lamellipodia (Rac1), and filopodia (Cdc42) in CNF1-intoxicated cells and ultimately results in rearrangement of the cytoskeleton (12,18,28). Phenotypically, CNF1 causes multinucleation of various tissue culture cells (8, 11) but can also be cytotoxic against certain cell lines, including Swiss 3T3 and 5637 bladder cells (19,22). In vivo, CNF1 evokes necrosis when it is injected intradermally into rabbit skin (4). Moreover, members of our laboratory, in collaboration with colleagues, demonstrated that in two animal systems CNF1 expression contributes to the virulence of UPEC strains. In a rat model of acute prostatitis, we found that intraurethral infection with a CNF1-positive strain leads to a significantly enhanced inflammatory response compared to that elicited by an isogenic, CNF1-negative mutant, even when the bacterial counts are equivalent (26). Similarly, in a mouse model of urinary tract infection, the production of CNF1 by UPEC strains results in higher bacterial counts and increased inflammation compared to the results for cnf1 isogenic mutants, in part due to...

c Long-term stability is a desired characteristic of vaccines, especially anthrax vaccines, which must be stockpiled for large-scale use in an emergency situation; however, spontaneous deamidation of purified vaccine antigens has the potential to adversely affect vaccine immunogenicity over time. In order to explore whether spontaneous deamidation of recombinant protective antigen (rPA)-the major component of new-generation anthrax vaccines-affects vaccine immunogenicity, we created a "genetically deamidated" form of rPA using site-directed mutagenesis to replace six deamidation-prone asparagine residues, at positions 408, 466, 537, 601, 713, and 719, with either aspartate, glutamine, or alanine residues. We found that the structure of the six-Asp mutant rPA was not significantly altered relative to that of the wild-type protein as assessed by circular dichroism (CD) spectroscopy and biological activity. In contrast, immunogenicity of aluminum-adjuvanted six-Asp mutant rPA, as measured by induction of toxin-neutralizing antibodies, was significantly lower than that of the corresponding wild-type rPA vaccine formulation. The six-Gln and six-Ala mutants also exhibited lower immunogenicity than the wild type. While the wild-type rPA vaccine formulation exhibited a high level of immunogenicity initially, its immunogenicity declined significantly upon storage at 25°C for 4 weeks. In contrast, the immunogenicity of the six-Asp mutant rPA vaccine formulation was low initially but did not change significantly upon storage. Taken together, results from this study suggest that spontaneous deamidation of asparagine residues predicted to occur during storage of rPA vaccines would adversely affect vaccine immunogenicity and therefore the storage life of vaccines.

Summary To achieve widespread dissemination in the host, Bacillus anthracis cells regulate their attachment to host endothelium during infection. Previous studies identified BslA (Bacillus anthracisS‐layer Protein A), a virulence factor of B. anthracis, as necessary and sufficient for adhesion of vegetative cells to human endothelial cells. While some factors have been identified, bacteria‐specific contributions to BslA mediated adhesion remain unclear. Using the attenuated vaccine Sterne 7702 strain of B. anthracis, we tested the hypothesis that InhA (immune inhibitor A), a B. anthracis protease, regulates BslA levels affecting the bacteria's ability to bind to endothelium. To test this, a combination of inhA mutant and complementation analysis in adhesion and invasion assays, Western blot and InhA inhibitor assays were employed. Results show InhA downregulates BslA activity reducing B. anthracis adhesion and invasion in human brain endothelial cells. BslA protein levels in ΔinhA bacteria were significantly higher than wild‐type and complemented strains showing InhA levels and BslA expression are inversely related. BslA was sensitive to purified InhA degradation in a concentration‐ and time‐dependent manner. Taken together these data support the role of InhA regulation of BslA‐mediated vegetative cell adhesion and invasion.

SummaryCytotoxic necrotizing factor type 1 (CNF1) and dermonecrotic toxin (DNT) share homology within their catalytic domains and possess deamidase and transglutaminase activities. Although each toxin has a preferred enzymatic activity (i.e. deamidation for CNF1 and transglutamination for DNT) as well as target substrates, both modify a specific glutamine residue in RhoA, Rac1 and Cdc42, which renders these GTPases constitutively active. Here we show that despite their similar mechanisms of action CNF1 and DNT induced unique phenotypes on HEp-2 and Swiss 3T3 cells. CNF1 induced multinucleation of HEp-2 cells and was cytotoxic for Swiss 3T3 cells (with binucleation of the few surviving cells) while DNT showed no morphological effects on HEp-2 cells but did induce binucleation of Swiss 3T3 cells. To determine if the enzymatic domain of each toxin dictated the induced phenotype, we constructed enzymatically active chimeric toxins and mutant toxins that contained single amino acid substitutions within the catalytic site and tested these molecules in tissue culture and enzymatic assays. Moreover, both site-directed mutant toxins showed reduced time to maximum transglutamination of RhoA compared with the parent toxins. Nevertheless, the substitution of threonine for Lys 1310 in the DNTbased mutant, while affecting transglutamination efficiency of the toxin, did not abrogate that enzymatic activity.

Shiga toxin (Stx)-producing Escherichia coli (STEC) strain B2F1 produces Stx type 2d, a toxin that becomes more toxic towards Vero cells in the presence of intestinal mucus. STEC that make Stx2d are more pathogenic to streptomycin (Str)-treated mice than most STEC that produce Stx2a or Stx2c. However, purified Stx2d is only 2- or 7-fold more toxic by the intraperitoneal route than Stx2a or Stx2c, respectively. We hypothesized, therefore, that the toxicity differences among Stx2a, Stx2c, and Stx2d occur at the level of delivery from the intestine. To evaluate that hypothesis, we altered the toxin type produced by stx2d+ mouse virulent O91:H21 clinical isolate B2F1 to Stx2a or Stx2c. Because B2F1 encodes two copies of stx2d, we did these studies in a derivative of B2F1 in which stx2d1 was deleted. Although the strains were equivalently virulent to the Str-treated mice at the 1010 dose, the B2F1 strain that produced Stx2a was attenuated relative to the ones that produced Stx2d or Stx2c when administered at 103 CFU/mouse. We next compared the oral toxicities of purified Stx2a, Stx2c, and Stx2d. We found that purified Stx2d is more toxic than Stx2a or Stx2c upon oral administration at 4 µg/mouse. Taken together, these studies suggest that Stx2 toxins are most potent when delivered directly from the bacterium. Furthermore, because Stx2d and Stx2c have the identical amino acid composition in the toxin B subunit, our results indicate that the virulence difference between Stx2a and Stx2d and Stx2c resides in the B or binding subunit of the toxins.