The three-dimensional structures of two toxins from snake venom throw light on the anticoagulant and neurotoxic sites of phospholipase A2.

The three-dimensional structures of the class II anticoagulant phospholipase A2 (PLA2) toxin RVV-VD from the venom of Russell's viper, Vipera russelli russelli, and the class I neurotoxic PLA2 Notechis II-5 from the, Australian tiger snake, Notechis scutatus scutatus, were determined to 2.2 A and 3.0 A resolution, respectively. Both enzymes are monomeric and consist of 121 and 119 residues, respectively. A comparison of ten class I/II PLA2 structures showed, among other differences, that the beta-sheet of these enzymes (residues 76-83) is about 90 degrees less twisted in class I than in class II PLA2s. This, along with the insertion of some residues in the region 57-59 in class I enzymes (the elapid loop), could be the main reason for the significant difference in the anticoagulant and (presynaptic) neurotoxic properties between the two classes of PLA2. It seems apparent from sequence and structural comparisons that the toxic site of PLA2 responsible for the strong anticoagulancy of these toxins consists of a negatively charged part, Glu53, together with a positively charged ridge of lysine residues free for intermolecular interactions. These lysines differ between the two classes of PLA2.

Changes to biological activity following acetylation of dendrotoxin I from Dendroaspis polylepis (black mamba).

The potassium channel blocker dendrotoxin I was acetylated with acetic anhydride. Mono-acetyl derivatives of all seven lysine residues (N-terminus blocked) and a di-derivative were isolated by chromatography on the cation-exchanger Bio-Rex 70 and reversed-phase high-performance liquid chromatography. The derivative acetyl-Lys 29 and the di-derivative of Tyr 24 and Lys 28 had more than 1000 times lower affinity than the native toxin as determined by inhibition of the 125I-dendrotoxin binding to synaptosomal membranes from rat brain. Lys 29 is part of the triplet Lys-Lys-Lys (28-30) which also occurs in the homologous alpha-dendrotoxin where the triplet is not in the functional site, as shown by site-directed mutagenesis. Acetylation of Lys 29 may have produced large structural perturbations that inactivated the toxin. Acetylation of Lys 28 alone had little effect, but the toxin became almost inactive when both Lys 28 and Tyr 24 were modified. Ten experiments were conducted under similar conditions, but a derivative of Tyr 24 was obtained only three times. In these cases the toxin apparently had a different structure, with Tyr 24 accessible to the reagent. This may depend on freeze-drying, which can alter the structure of proteins. The third derivative with low activity was acetyl-Lys 5, with affinity decreased 20-fold. Lys 5 has a protruding side-chain that does not interact with any other group in the toxin molecule. Therefore, Lys 5 is probably part of the functional site for dendrotoxin's binding to the voltage-dependent K+ channels.

Crystallographic investigation of the dependence of calcium and phosphate ions for notexin.

The crystal structure of the neurotoxic phospholipase A2, notexin, revealed three binding sites for sulphate ions which were suggested to be phosphate binding sites of importance for the activity of the toxin. The present investigation shows that the sulphate ion bound to the major binding site alters the structure of residues 60-75. In the absence of sulphate and phosphate, the structure of this loop has a conformation which partly resembles the non-neurotoxic PLA2s. The affinity of notexin for phosphate is 17 microM, as measured by the increase in fluorescence at 345 nm. Since the concentrations of phosphate and sulphate ions in blood plasma are 3 and 1 mM, respectively, the binding site must be occupied under physiological conditions. This major sulphate/phosphate binding site explains the specific affinity labelling by pyridoxal phosphate. Pyridoxal phosphate binds to this anion binding site which allows the reaction with Lys-88 or Lys-89. The structure of notexin in the presence and absence of Ca2+ shows only small local structural differences.

A Collagen-Binding S-Layer Protein in Lactobacillus crispatus.

Two S-layer-expressing strains, Lactobacillus crispatus JCM 5810 and Lactobacillus acidophilus JCM 1132, were assessed for adherence to proteins of the mammalian extracellular matrix. L. crispatus JCM 5810 adhered efficiently to immobilized type IV and I collagens, laminin, and, with a lower affinity, to type V collagen and fibronectin. Strain JCM 1132 did not exhibit detectable adhesiveness. Within the fibronectin molecule, JCM 5810 recognized the 120-kDa cell-binding fragment of the protein, while no bacterial adhesion to the amino-terminal 30-kDa or the gelatin-binding 40-kDa fragment was detected. JCM 5810 but not JCM 1132 also bound (sup125)I-labelled soluble type IV collagen, and this binding was efficiently inhibited by unlabelled type IV and I collagens and less efficiently by type V collagen, but not by laminin or fibronectin. L. crispatus JCM 5810 but not L. acidophilus JCM 1132 also adhered to Matrigel, a reconstituted basement membrane preparation from mouse sarcoma cells, as well as to the extracellular matrix prepared from human Intestine 407 cells. S-layers from both strains were extracted with 2 M guanidine hydrochloride, separated by electrophoresis, and transferred to nitrocellulose sheets. The S-layer protein from JCM 5810 bound (sup125)I-labelled type IV collagen, whereas no binding was seen with the S-layer protein from JCM 1132. Binding of (sup125)I-collagen IV to the JCM 5810 S-layer protein was effectively inhibited by unlabelled type I and IV collagens but not by type V collagen, laminin, or fibronectin. It was concluded that L. crispatus JCM 5810 has the capacity to adhere to human subintestinal extracellular matrix via a collagen-binding S-layer.

Bacterial proteins binding to the mammalian extracellular matrix.

Pathogenic bacteria frequently express surface proteins with affinity for components of the mammalian extracellular matrix, i.e. collagens, laminin, fibronectin or proteoglycans. This review summarizes our current knowledge on the mechanisms of bacterial adherence to extracellular matrices and on the biological significance of these interactions. The best-characterized bacterial proteins active in these interactions are the mycobacterial fibronectin-binding proteins, the fibronectin- and the collagen-binding proteins of staphylococci and streptococci, specific enterobacterial fimbrial types, as well as the polymeric surface proteins YadA of yersinias and the A-protein of Aeromonas. Some of these bacterial proteins are highly specific for an extracellular matrix protein, some are multifunctional and express binding activities towards a number of target proteins. The interactions can be based on a protein-protein or on a protein-carbohydrate interaction, or on a bridging mechanism mediated by a bivalent soluble target protein. Many of the interactions have also been demonstrated on tissue sections or in vivo, and adherence to the extracellular matrix has been shown to promote bacterial colonization of damaged tissues.

P fimbriae of uropathogenic Escherichia coli as multifunctional adherence organelles.

P fimbriae are the major single virulence factor of uropathogenic Escherichia coli strains. Recent analyses have shown that P fimbriae possess two distinct binding specificities mediated by different fimbrial subunits. P fimbriae bind to Gal alpha (1-4)Gal-containing globoseries of glycolipids of epithelial cells; this binding is mediated by the lectin-like minor protein G of the filament. In vitro mapping of the human urinary tract for binding sites of P fimbriae has revealed that they bind in a Gal alpha (1-4)Gal-inhibitable manner to epithelia of kidney and bladder. On the other hand, P fimbriae bind to immobilized fibronectin and its amino- and carboxyterminal fragments; this binding is dependent on the E and the F minor proteins of the P-fimbrial filament and seems to be based on a protein-protein interaction. The P fimbriae-fibronectin interaction has been demonstrated also on frozen sections of kidney. P fimbriae thus possess two tissue-adherence properties: one specific for epithelial glycoconjugates and the other for fibronectin of subepithelial extracellular matrices. P-fimbrial binding to epithelial glycoconjugates seems to be important in determining the host tropism and enabling the ascent of E. coli urinary tract infections. Binding to fibronectin may be important in secondary phases of the infection, e.g. after epithelial injury.

Binding of bacterial adhesins to rat glomerular mesangium in vivo.

Two well characterized bacterial adhesins, the O75X fimbriae of Escherichia coli and the type-3 fimbriae of Klebsiellae, with in vitro affinities to type IV and V collagens, respectively, were used to test whether bacterial components with affinity for glomerular matrix could bind to glomeruli in vivo. The purified fimbrial proteins were injected into rats, and kidney samples were studied by immunofluorescence at two hours to nine months postinjection. The O75X, but not the type-3 fimbriae, formed mesangial deposits that persisted for months. Preincubation of the O75X fimbriae with type IV collagen significantly reduced the glomerular binding. The fimbrial deposits were extracellular, as anti-O75X IgG injected into rats bound to glomeruli. Proteinuria or histological damage could not be detected even after passive or active immunizations of the rats. The results demonstrate that bacterial adhesins may bind in vivo to and persist in glomeruli by their specific affinities. The results also indicate that additional factors provided by the bacteria or the host are needed for glomerular damage to take place.

Immobilization of plasminogen on Escherichia coli flagella.

The interaction of plasminogen with flagella of Escherichia coli was investigated. Plasminogen bound to flagella purified from E. coli LE392, a commonly used cloning host, and E. coli IH3069, an O25H1 strain isolated from a case of newborn bacteremia. The binding was inhibited by the lysine analog epsilon-aminocaproic acid, suggesting involvement of the lysine-binding Kringle domains of plasminogen in the binding. Purified flagella enhanced the formation of plasmin activity in the presence of tissue-type plasminogen activator; a similar enhancement was observed with flagella-expressing LE392 cells.

Penetration of fimbriate enteric bacteria through basement membranes: a hypothesis.

A mechanism for penetration of basement membranes by Escherichia coli is presented. The mechanism is based on the ability of the S fimbriae of meningitis-associated E. coli to bind to vascular endothelium and choroid plexuses in brain and to basement membranes. On the other hand, the S and the type 1 fimbriae of E. coli immobilize plasminogen and tissue-type plasminogen activator; this process generates proteolytic plasmin activity on the surface of fimbriate cells. Our hypothesis is that bacterium-bound plasma activity, directed to basement membranes through fimbrial binding, promotes bacterial penetration through basement membranes.

The three-dimensional structure of notexin, a presynaptic neurotoxic phospholipase A2 at 2.0 A resolution.

The three-dimensional structure of notexin has been solved by molecular replacement methods. The structure has been refined at 2.0 A resolution to a crystallographic R-value of 16.5% with good stereo-chemistry. The core of the protein is very similar to other phospholipase A2s (PLA2 s) but several parts of the molecule are distinctly different. The most significant differences from PLA2 s from bovine pancreas and rattlesnake occur in the stretches 56-80 and 85-89. Residue 69, which has been shown to be important for phospholipase binding, has a different conformation and different interactions than in other known PLA2s. The C alpha positions for residues 86-88 differ by about 6 A from both the bovine and the rattlesnake enzyme. The crystals contain no Ca2+ ions. Instead, a water molecule occupies the calcium site.

Multifunctional nature of P fimbriae of uropathogenic Escherichia coli: mutations in fsoE and fsoF influence fimbrial binding to renal tubuli and immobilized fibronectin.

P fimbriae of the F7(1) serotype of Escherichia coli are composed of a major subunit, FsoA, and of three minor proteins named FsoG, FsoE, and FsoF. FsoG is the Gal alpha(1-4)Gal-specific lectin. We assessed mutated recombinant strains each deficient in one fimbrial component for adhesion to frozen sections of rat cortical kidney and to fibronectin immobilized on glass. Rat kidney lacks the Gal alpha(1-4)Gal-containing glycolipids. The fsoG mutant strain was as adhesive to sections of rat kidney and to fibronectin-coated glass as was the recombinant strain expressing the complete fso gene cluster. The fsoA mutant strain was highly adhesive to fibronectin and to kidney sections. In the rat kidney, the adhesion of these strains was predominantly localized to sites of basolateral membranes of tubuli. The fsoE and the fsoF mutant strains were slightly less adhesive to kidney structures and failed to adhere to fibronectin. The fsoE, fsoF double mutant strain adhered neither to fibronectin nor to kidney sections. None of the fso recombinant strains reacted with soluble fibronectin, suggesting that the interaction is dependent on the conformation of the fibronectin molecules. Recombinant strains expressing the F7(2), F8, F11, F13, and F14 serovariants of the P fimbria also showed adherence to immobilized fibronectin. The results show that in addition to binding to globoseries of glycolipids via the G protein, the P fimbriae of uropathogenic E. coli exhibit a tissue-binding property influenced by fsoE and fsoF gene products and with affinity for basolateral membranes and fibronectin.

Type V collagen as the target for type-3 fimbriae, enterobacterial adherence organelles.

Tissue-binding specificity of the type-3 fimbriae of pathogenic enteric bacteria was determined using frozen sections of human kidney. A wild-type Klebsiella sp. strain and the recombinant strain Escherichia coli HB101(pFK12), both expressing type-3 fimbriae, as well as the purified type-3 fimbriae effectively bound to sites at or adjacent to tubular basement membranes, Bowman's capsule, arterial walls, and the interstitial connective tissue. Bacterial adherence to kidney was decreased after collagenase treatment of the tissue sections. Recombinant strains expressing type-3 fimbriae specifically adhered to type V collagen immobilized on glass slides, whereas other collagens, fibronectin or laminin did not support bacterial adherence. In accordance with these findings, specific binding of purified type-3 fimbriae to immobilized type V collagen was demonstrated. Specific adhesion to type V collagen was also seen with the recombinant strain HB101(pFK52/pDC17), which expresses the mrkD gene of the type-3 fimbrial gene cluster in association with the pap-encoded fimbrial filament of E. coli, showing that the observed binding was mediated by the minor lectin (MrkD) protein of the type-3 fimbrial filament. The interaction is highly dependent on the conformation of type V collagen molecules since type V collagen in solution did not react with the fimbriae. Specific binding to type V collagen was also exhibited by type-3 fimbriate strains of Yersinia and Salmonella, showing that the ability to use type V collagen as tissue target is widespread among enteric bacteria.

Binding of Escherichia coli S fimbriae to cultured human endothelial cells.

The binding of Escherichia coli adhesins to human umbilical vein endothelial cells was studied by a cell monolayer enzyme-linked immunosorbent assay. S fimbriae displayed a concentration-dependent and saturable binding to the endothelial cells which was mediated by their sialylgalactoside-specific lectin activity. P fimbriae exhibited only low binding, and type 1 fimbriae exhibited no binding to these cells.

The O75X adhesin of uropathogenic Escherichia coli is a type IV collagen-binding protein.

Interaction of the basement-membrane binding O75X adhesin of uropathogenic Escherichia coli with various extracellular matrix proteins was studied. The adhesin showed strong binding to type IV collagen immobilized on microtitre plates, whereas other collagens, laminin and fibronectin, were only weakly recognized. Similarly, specific binding of [125I]-labelled type IV collagen to O75X-positive bacteria was shown. Interaction of the two proteins was also demonstrated by affinity chromatography of the O75X adhesin on immobilized type IV collagen. The adhesin bound strongly to the immobilized N-terminal 7S domain of type IV collagen, and the binding of [125I]-labelled type IV collagen to O75X-positive bacteria was inhibited by the soluble 7S domain. Binding of O75X to type IV collagen and to its 7S domain was specifically inhibited by chloramphenicol but was not affected by periodate or endoglycosidase-H treatment of the glycoproteins. Our results show that the 7S domain of type IV collagen is the basement membrane receptor for the O75X adhesin and suggest an interaction based on protein-protein recognition. Inhibition of the interaction by chloramphenicol favours the supposition that a modified tyrosine is involved in the binding site.

A novel lectin-independent interaction of P fimbriae of Escherichia coli with immobilized fibronectin.

Binding of P fimbriae of uropathogenic Escherichia coli to purified human fibronectin and human placental type IV collagen was studied. In an enzyme immunoassay, purified P fimbriae bound strongly to immobilized intact fibronectin and to the aminoterminal 30-kDa fragment and the 120-140-kDa carboxyterminal fragments of fibronectin. Binding to the gelatin-binding 40-kDa fragment of fibronectin was considerably weaker. No binding to immobilized type IV collagen was seen. The interaction between P fimbriae and immobilized fibronectin was not inhibited by alpha-D-Gal-(1-4)-beta-D-Gal-1-O-Me, a receptor analog of P fimbriae. Moreover, a mutated P fimbria lacking the lectin activity behaved similarly in the adherence assays. Recombinant strains expressing the corresponding cloned fimbriae genes bound to immobilized fibronectin, but no binding to soluble 125I-labelled fibronectin was found. The results suggest that P fimbriae interact with immobilized fibronectin and that the binding mechanism does not involve the lectin activity of the fimbriae.

Properties of Escherichia coli isolates from urinary tract infections in boys.

Fifty-five isolates of Escherichia coli from urine of boys younger than three years of age with urinary tract infection (UTI) were compared with strains from girls of the same age range who had UTI. The frequency of P fimbriae, hemolysin, and type 1C fimbriae, previously described as associated with pyelonephritis (PN) in girls, was also high (76%, 60%, and 31%, respectively) in UTI-associated strains from boys. However, in contrast to isolates from girls, strains from lower UTI in boys did not differ from PN-associated strains regarding these three characteristics. In contrast, aerobactin production was significantly associated with PN compared with lower UTI in both sexes. Serotypes O4 and O6 were overrepresented among all UTI-associated strains from boys and PN-associated strains from girls. This overrepresentation was largely accounted for by three clones, one of which was a new clone identified among the UTI-associated strains from boys.

Binding characteristics of Escherichia coli adhesins in human urinary bladder.

We studied domains in the human bladder that acted as receptors for Escherichia coli P, S, type 1, type 1C, and O75X fimbriae or adhesin and domains in the human kidneys that were receptors for E. coli type 1C fimbriae. Binding sites in frozen tissue sections were localized by direct staining with fluorochrome-labeled recombinant strains and by indirect immunofluorescence with the purified adhesins. In the bladder, the P and S fimbriae showed closely similar binding to the epithelial and muscular layers, and the S fimbriae also bound to the connective tissue elements. Type 1 fimbriae bound to vascular walls and to muscle cells, whereas the O75X adhesin bound avidly to connective tissue elements and to some extent to epithelial and muscle cells of the bladder. The type 1C fimbriae bound to distal tubules and collecting ducts of the kidney and to vascular endothelial cells in both the kidney and bladder. The binding of all adhesin types was inhibited by specific receptor analogs or Fab fragments. The results reveal a possible mechanism by which the type 1C fimbriae may help invasion of E. coli in the kidneys but do not support a pathogenetic role for type 1 fimbriae. Similar tissue specificity of P and S fimbriae in the human urinary tract indicates that the presence of binding sites on uroepithelia does not fully explain the virulence properties of P fimbriae in human urinary tract infections.