Advertisement

Secreted Neutral Metalloproteases of Bacillus anthracis as Candidate Pathogenic Factors*

Open AccessPublished:August 22, 2006DOI:https://doi.org/10.1016/S0021-9258(19)84053-X
      To evaluate the pathogenic potential of Bacillus anthracis-secreted proteases distinct from lethal toxin, two neutral zinc metalloproteases were purified to apparent homogeneity from the culture supernatant of a non-virulent delta Ames strain (pXO1, pXO2). The first (designated Npr599) is a thermolysin-like enzyme highly homologous to bacillolysins from other Bacillus species. The second (designated InhA) is a homolog of the Bacillus thuringiensis immune inhibitor A. These proteases belong to the M4 and M6 families, respectively. Both enzymes digested various substrates, including extracellular matrix proteins, endogenous inhibitors, and coagulation proteins, with some differences in specificity. In addition, InhA accelerated urokinase-mediated plasminogen activation, suggesting that InhA acts as a modulator of plasmin in the host inflammatory system. Relevant to epithelial barrier function, Npr599 and InhA significantly enhanced syndecan-1 shedding from cultured normal murine mammary gland cells without affecting their viability through stimulation of the host cell ectodomain shedding mechanism. In addition, Npr599 and InhA directly cleaved recombinant syndecan-1 fused to glutathione S-transferase. Mass spectrometric analysis suggested that the cleavage sites of Npr599 and InhA are the Asp39–Asp40 and Gly48–Thr49 bonds, respectively. We propose that Npr599 and InhA from B. anthracis are multifunctional pathogenic factors that may contribute to anthrax pathology through direct degradation of host tissues, increases in barrier permeability, and/or modulation of host defenses.
      Bacillus anthracis is a highly pathogenic Gram-positive bacillus that secrets two major virulence factors, lethal toxin and edema toxin, encoded by megaplasmid pXO1. Another plasmid (pXO2) encodes the anti-phagocytic capsule, which substantially contributes to the virulence of the microbe. Lethal toxin is necessary for pathogenicity, as deletion of its gene renders the microbe avirulent, whereas edema toxin-knock-out strains are only partially attenuated (
      • Pezard C.
      • Berche P.
      • Mock M.
      ). Lethal toxin consists of a heptameric protective antigen noncovalently associated with lethal factor. Lethal toxin is a zinc metalloprotease that cleaves and thus inhibits MAPK
      The abbreviations used are: MAPK, mitogen-activated protein kinase; ECM, extracellular matrix; HS, heparan sulfate; MES, 4-morpholineethanesulfonic acid; GST, glutathione S-transferase; pNA, p-nitroanilide; uPA, urokinase-type plasminogen activator; NMuMG, normal murine mammary gland; PMA, phorbol 12-myristate 13-acetate; MCA, 7-methoxycoumarin-4-acetyl; DNP, 2,4-dinitrophenyl; Nva, l-norvaline; DPA, l-diaminopropionyl; ERK, extracellular signal-regulated kinase; JNK, c-Jun N-terminal kinase.
      2The abbreviations used are: MAPK, mitogen-activated protein kinase; ECM, extracellular matrix; HS, heparan sulfate; MES, 4-morpholineethanesulfonic acid; GST, glutathione S-transferase; pNA, p-nitroanilide; uPA, urokinase-type plasminogen activator; NMuMG, normal murine mammary gland; PMA, phorbol 12-myristate 13-acetate; MCA, 7-methoxycoumarin-4-acetyl; DNP, 2,4-dinitrophenyl; Nva, l-norvaline; DPA, l-diaminopropionyl; ERK, extracellular signal-regulated kinase; JNK, c-Jun N-terminal kinase.
      kinase family members in vitro and in vivo, resulting in defective host cell signaling (
      • Guidi-Rontani C.
      • Levy M.
      • Ohayon H.
      • Mock M.
      ,
      • Pellizzari R.
      • Guidi-Rontani C.
      • Vitale G.
      • Mock M.
      • Montecucco C.
      ), with broad implications for the host innate and adaptive immune responses (
      • Agrawal A.
      • Lingappa J.
      • Leppla S.H.
      • Agrawal S.
      • Jabbar A.
      • Quinn C.
      • Pulendran B.
      ,
      • Pickering A.K.
      • Osorio M.
      • Lee G.M.
      • Grippe V.K.
      • Bray M.
      • Merkel T.J.
      ,
      • Popov S.G.
      • Popova T.G.
      • Grene E.
      • Klotz F.
      • Cardwell J.
      • Bradburne C.
      • Jama Y.
      • Maland M.
      • Wells J.
      • Nalca A.
      • Voss T.
      • Bailey C.
      • Alibek K.
      ). Based on its properties, lethal factor is considered to be a major target for new anthrax therapies (
      • Shoop W.L.
      • Xiong Y.
      • Wiltsie J.
      • Woods A.
      • Guo J.
      • Pivnichny J.V.
      • Felcetto T.
      • Michael B.F.
      • Bansal A.
      • Cummings R.T.
      • Cunningham B.R.
      • Friedlander A.M.
      • Douglas C.M.
      • Patel S.B.
      • Wisniewski D.
      • Scapin G.
      • Salowe S.P.
      • Zaller D.M.
      • Chapman K.T.
      • Scolnick E.M.
      • Schmatz D.M.
      • Bartizal K.
      • MacCoss M.
      • Hermes J.D.
      ,
      • Turk B.E.
      • Wong T.Y.
      • Schwarzenbacher R.
      • Jarrell E.T.
      • Leppla S.H.
      • Collier R.J.
      • Liddington R.C.
      • Cantley L.C.
      ). Specific lethal toxin blockers are expected to complement the existing antibiotic treatments, which alone are successful only in 55% of inhalation anthrax patients (
      • Jernigan D.B.
      • Raghunathan P.L.
      • Bell B.P.
      • Brechner R.
      • Bresnitz E.A.
      • Butler J.C.
      • Cetron M.
      • Cohen M.
      • Doyle T.
      • Fischer M.
      • Greene C.
      • Griffith K.S.
      • Guarner J.
      • Hadler J.L.
      • Hayslett J.A.
      • Meyer R.
      • Petersen L.R.
      • Phillips M.
      • Pinner R.
      • Popovic T.
      • Quinn C.P.
      • Reefhuis J.
      • Reissman D.
      • Rosenstein N.
      • Schuchat A.
      • Shieh W.J.
      • Siegal L.
      • Swerdlow D.L.
      • Tenover F.C.
      • Traeger M.
      • Ward J.W.
      • Weisfuse I.
      • Wiersma S.
      • Yeskey K.
      • Zaki S.
      • Ashford D.A.
      • Perkins B.A.
      • Ostroff S.
      • Hughes J.
      • Fleming D.
      • Koplan J.P.
      • Gerberding J.L.
      ). It has been reported that the synthetic inhibitor of lethal toxin proteolytic activity in combination with ciprofloxacin is protective in rabbits (
      • Shoop W.L.
      • Xiong Y.
      • Wiltsie J.
      • Woods A.
      • Guo J.
      • Pivnichny J.V.
      • Felcetto T.
      • Michael B.F.
      • Bansal A.
      • Cummings R.T.
      • Cunningham B.R.
      • Friedlander A.M.
      • Douglas C.M.
      • Patel S.B.
      • Wisniewski D.
      • Scapin G.
      • Salowe S.P.
      • Zaller D.M.
      • Chapman K.T.
      • Scolnick E.M.
      • Schmatz D.M.
      • Bartizal K.
      • MacCoss M.
      • Hermes J.D.
      ). However, a number of anthrax pathological features such as massive hemorrhages and intensive organ and tissue damage cannot be explained by the sole activity of lethal toxin and edema toxin, indicating the involvement of other virulence factors. Our animal experiments using culture supernatants of the non-toxigenic B. anthracis delta Ames strain demonstrated their hemorrhagic effect on skin and high toxicity upon intratracheal administration (
      • Popov S.G.
      • Popova T.G.
      • Hopkins S.
      • Weinstein R.S.
      • MacAfee R.
      • Fryxell K.J.
      • Chandhoke V.
      • Bailey C.
      • Alibek K.
      ). Consistent with this, several broad-spectrum protease inhibitors, as well as immune sera against anthrax M4 and M9 metalloproteases, display high efficacy in the post-exposure treatment of murine systemic anthrax (
      • Popov S.G.
      • Popova T.G.
      • Hopkins S.
      • Weinstein R.S.
      • MacAfee R.
      • Fryxell K.J.
      • Chandhoke V.
      • Bailey C.
      • Alibek K.
      ). These observations prompted us evaluate the pathogenic potential of the secreted proteolytic enzymes, which might serve “accessory” functions to lethal toxin and therefore might be required for the full virulence of B. anthracis. To the best of our knowledge, B. anthracis extracellular proteases have not been previously characterized with respect to their enzymatic properties and activity toward pathologically relevant substrates.
      Bacterial proteases may exert tissue damage directly by cleaving the extracellular matrix (ECM) components, including collagen, laminin, fibronectin, and elastin (
      • Miyoshi S.
      • Shinoda S.
      ,
      • Harrington D.J.
      ). Another general mechanism involves microbial interference with the homeostatic balance between endogenous proteases and their inhibitors, which determines tissue integrity. One of the examples is the effect of Pseudomonas aeruginosa elastase on the balance between neutrophil elastase and the inhibitors α1-protease inhibitor and α2-macroglobulin (
      • Dinwiddie R.
      ). The mammalian plasminogen system can also serve as a target of bacterial proteases, which can accelerate the urokinase-catalyzed activation of plasminogen and degrade endogenous plasmin inhibitors such as α2-antiplasmin and α2-macroglobulin (
      • Lahteenmaki K.
      • Edelman S.
      • Korhonen T.K.
      ). As a result, the activated plasmin can directly digest laminin, a major glycoprotein of basement membranes, and indirectly further damage tissue barriers by activating latent matrix metalloproteases (
      • Lahteenmaki K.
      • Virkola R.
      • Pouttu R.
      • Kuusela P.
      • Kukkonen M.
      • Korhonen T.K.
      ).
      In addition to ECM degradation, bacterial proteases are also involved in the pathogenic cleavage of host cell-surface molecules in a process of ectodomain shedding (
      • Park P.W.
      • Pier G.B.
      • Hinkes M.T.
      • Bernfield M.
      ,
      • Park P.W.
      • Pier G.B.
      • Preston M.J.
      • Goldberger O.
      • Fitzgerald M.L.
      • Bernfield M.
      ). Shed ectodomains play pivotal roles in diverse pathophysiological events, including septic shock, host defense, and wound healing (
      • Haynes III, A.
      • Ruda F.
      • Oliver J.
      • Hamood A.N.
      • Griswold J.A.
      • Park P.W.
      • Rumbaugh K.P.
      ,
      • Marshall L.J.
      • Ramdin L.S.P.
      • Brooks T.
      • Charlton P.A.
      • Shute J.K.
      ,
      • Elenius V.
      • Go¨tte M.
      • Reizes O.
      • Elenius K.
      • Bernfield M.
      ). During infection, secreted pathogenic factors enhance host ectodomain shedding, which contributes to epithelial barrier disruption, endothelial damage, and tissue penetration by bacilli (
      • Park P.W.
      • Pier G.B.
      • Hinkes M.T.
      • Bernfield M.
      ,
      • Park P.W.
      • Pier G.B.
      • Preston M.J.
      • Goldberger O.
      • Fitzgerald M.L.
      • Bernfield M.
      ). For instance, LasA, a secreted virulence factor of P. aeruginosa, enhances shedding of syndecan-1, which belongs to a family of cell-surface heparan sulfate (HS) proteoglycans. The resulting soluble syndecan-1 ectodomains enhance bacterial virulence in newborn mice (
      • Park P.W.
      • Pier G.B.
      • Hinkes M.T.
      • Bernfield M.
      ,
      • Park P.W.
      • Pier G.B.
      • Preston M.J.
      • Goldberger O.
      • Fitzgerald M.L.
      • Bernfield M.
      ). Quite notably, inhibition of syndecan-1 shedding or neutralization of the HS of the shed ectodomain prevents P. aeruginosa lung infection (
      • Park P.W.
      • Pier G.B.
      • Hinkes M.T.
      • Bernfield M.
      ). These facts indicate that proteolysis of the ECM and shedding of the cell-surface ectodomain can play roles not only in signaling, but also in establishment of infection by acting as mediators of lethality, perturbing different mechanisms of the host defense response.
      We report here the purification, biochemical properties, and substrate specificity, with regard to ECM molecules, plasma proteins, and the cell-surface protein syndecan-1, of two neutral zinc metalloproteases. The first (designated Npr599 for neutral protease) is a thermolysin-like enzyme highly homologous to bacillolysins from other Bacillus species (
      • Wetmore D.R.
      • Wong S.-L.
      • Roche R.S.
      ,
      • Narasaki R.
      • Kuribayashi H.
      • Shimizu K.
      • Imamura D.
      • Sato T.
      • Hasumi K.
      ). The second (designated InhA for immune inhibitor A metalloprotease) is a homolog of the Bacillus thuringiensis immune inhibitor A. These proteases belong to the M4 and M6 families, respectively. Both of these enzymes can serve as possible pathogenic factors, enhancing tissue destruction, bacterial invasion, and perturbation of host defense responses. Inhibition of Npr599 and InhA activities in vitro correlates with the protective effects of the anti-protease treatments reported previously (
      • Popov S.G.
      • Popova T.G.
      • Hopkins S.
      • Weinstein R.S.
      • MacAfee R.
      • Fryxell K.J.
      • Chandhoke V.
      • Bailey C.
      • Alibek K.
      ), indicating that they can be considered as potential therapeutic targets.

      EXPERIMENTAL PROCEDURES

      Microbial Strain, Cultivation, and Supernatant Preparation—The non-encapsulated and non-toxigenic B. anthracis strain delta Ames (pXO1, pXO2) was kindly provided by Dr. J. Shiloach (National Institutes of Health, Bethesda, MD). To obtain a culture supernatant, 1 liter of LB medium was inoculated with an overnight seed culture (50 ml), incubated at 37 °C with vigorous agitation until the cells had reached stationary phase, and centrifuged at 17,000 × g for 10 min. The supernatant was removed and passed through a 0.22-μm cellulose acetate filter.
      Purification—All operations during enzyme purifications were performed at 4 °C unless indicated otherwise. Solid ammonium sulfate was added to 940 ml of culture supernatant to 75% saturation. The precipitated proteins were then collected by centrifugation at 17,000 × g for 20 min, dissolved, and dialyzed against 50 mm Tris-HCl (pH 7.6) containing 3 mm sodium azide. The resulting proteins were loaded onto a DEAE-cellulose anion-exchange column (bed volume of 60 ml) equilibrated with 50 mm Tris-HCl (pH 7.6) containing 3 mm sodium azide. Elution was achieved stepwise with 10, 50, 100, 200, 500, and 1000 mm NaCl in the same buffer. The substances of two peaks were collected: a flow-through fraction (P1) and a 200 mm NaCl eluate (P2). The P1 and P2 protease fractions were loaded onto a Sephacryl S-200 gel filtration column equilibrated with 20 mm Tris-HCl (pH7.6) and 150 mm NaCl and eluted at flow rate of 1.3 ml/min. Fractions (5 ml) were collected, and protease activity was assayed.
      Protease Assay—Protease activity was assayed during purification using an EnzChek Ultra protease kit for casein hydrolytic activity, an EnzChek gelatinase/collagenase kit for gelatin hydrolytic activity, and an EnzChek elastase kit for elastin hydrolytic activity, respectively, according to the manufacturer’s recommendation (Molecular Probes). Briefly, 5 μl of supernatant or fractions in 45 μl of digestion buffer (100 mm Tris-HCl (pH 8.0), 0.1% Triton X-100, 5 mm EDTA, and 1 mm phenylmethylsulfonyl fluoride) were mixed with 50 μl of fluorescein-labeled substrate, and then fluorescence intensity was measured after a 1-h incubation at 37 °C at 485-nm excitation and 510-nm emission wavelengths. One unit of protease activity was defined as the amount of protease required to liberate 1 mmol of the fluorescent dye from substrate-dye conjugates in 1 min.
      Characterization of the Proteases—To study the effect of pH on protease activity, the proteases were assayed at 37 °C in the following 0.1 m NaCl-containing buffers at various pH values: 50 mm sodium acetate/acetic acid buffer (pH 4–5.5), MES/NaOH buffer (pH 6–7), and 50 mm Tris-HCl (pH 7.5–10). The optimum temperature was determined by measuring the caseinolytic activity of the protease at 21, 37, 50, and 70 °C for 1 h (without the temperature-pH correction). To test the effect of inhibitors on protease activity, the proteins were preincubated with inhibitors, divalent ions, or other reaction components in 10 mm Tris-HCl (pH 7.8) for 30 min at room temperature. An equal volume of 2× casein substrate was then added, followed by further incubation at 37 °C for 1 h.
      SDS-PAGE and Determination of Protein Concentration—Proteins were separated by Tris/glycine/SDS-PAGE using 10 or 14% gels under reduced and denaturing conditions. The gels were stained using Coomassie Brilliant Blue R-250 and then destained. Protein concentration was determined colorimetrically using the Bio-Rad protein assay (Bradford) dye reagent and the bovine serum albumin as standard.
      N-terminal Amino Acid Sequencing—Partial N-terminal amino acid sequencing of the purified proteases was performed on polyvinylidene difluoride-electroblotted proteins at the Midwest Analytical, Inc. (St. Louis, MO), using an automated Edman degradation sequencer (Applied Biosystems, Foster City, CA).
      Substrate Digestion by Proteases—Approximately 0.2 μg of proteases was incubated for 4 h at 37 °C with various proteins in 20 mm Tris-HCl (pH 7.4) containing 1 mm CaCl2 and 1 mm MgSO4. Denaturation of human collagen types I, III, and IV was performed at 95 °C for 2 min. A plasmid expressing recombinant rat syndecan-1 with a glutathione S-transferase (GST) tag at the N terminus was kindly provided by Dr. Eok-Soo Oh (Ewha Womans University, Seoul, Korea). Recombinant syndecan-1 protein was partially prepared from host Escherichia coli BL21(DE3) cells using a glutathione-Sepharose affinity column. Digested substrates were separated by SDS-PAGE.
      Determination of Kinetic Parameters—Synthetic collagenase substrates (0.4–25 μm) were prepared in assay buffer (50 mm Tris-HCl (pH 7.5), 1 mm CaCl2, and 0.01% Tween 20). A collagenase assay was carried out in the substrate solution by incubation with 4 nm enzymes at 37 °C. Fluorescence was measured at λex = 320 nm and λem = 390 nm. Initial velocities were obtained from plots of fluorescence versus time using the data points corresponding to <40% full hydrolysis. The slopes of these plots were divided by the fluorescence change corresponding to complete hydrolysis and then multiplied by the substrate concentrations to obtain initial velocity in units of μm s–1. N-(3-(2-furyl)acryloyl)-Leu-Gly-Pro-Ala-hydrolyzing activity was measured by continuously monitoring the decrease in absorbance of the substrate at 324 nm after the addition of enzyme.
      Assay for Plasmin and Plasminogen Activation—Plasmin activity was assayed by monitoring Val-Leu-Lys-p-nitroanilide (pNA) hydrolysis at a molar enzyme/α2-antiplasmin ratio of 0.3 in 50 mm Tris-HCl (pH 7.5) and 1 mm CaCl2. α2-Antiplasmin was preincubated with proteases at 37 °C for 4 h. Plasminogen activation in the presence of plasma fibrin was assayed by monitoring Val-Leu-Lys-pNA hydrolysis. Human plasminogen (8.3 μg) was incubated at 37 °C with 2 μg of protease or streptokinase (positive control) in 50 μl of 50 mm Tris-HCl (pH 7.5) and 1 mm CaCl2. The resulting reactions were diluted 20-fold and added to 100 μm Val-Leu-Lys-pNA (50 μl) in the presence of fibrin. Urokinase-type plasminogen activator (uPA)-catalyzed plasminogen activation was carried out with 200 units/ml uPA; 0.1 unit/ml plasminogen; 100 μm Val-Leu-Lys-pNA; and 2, 5, or 10 μg/ml purified proteases in a 100-μl reaction volume. The release of pNA from the chromogenic substrate was monitored at 405 nm.
      Analysis of Shedding in Cultured Cells—Syndecan-1 shedding from normal murine mammary gland (NMuMG) cells was assayed as described previously (
      • Popova T.G.
      • Millis B.
      • Bradburne C.
      • Nazarenko S.
      • Bailey C.
      • Chandhoke V.
      • Popov S.G.
      ). Briefly, cells were grown in Dulbecco’s modified Eagle’s medium in 96-well plates, cultured to 1-day post-confluence, and stimulated with the indicated proteins using serum-free medium. After stimulation, culture supernatants (100 μl) were collected and acidified with 900 μl of acidification buffer (150 mm NaCl, 50 mm NaOAc, and 0.1% Tween 20 (pH 4.5)). Samples were applied to Immobilon-Ny+ membrane using a Bio-Dot microfiltration apparatus (Bio-Rad). The membrane was washed with acidification buffer; blocked in milk; and incubated with rat anti-mouse syndecan-1 monoclonal antibody (clone 281-2; Pharmingen), followed by horseradish peroxidase-conjugated goat anti-rat secondary antibody. The membranes were developed using an ECL Plus Western blotting detection kit (Amersham Biosciences) and Kodak BioMax light film (Sigma). The results were quantified by scanning the exposed film and evaluating the intensity of exposed dots using the software AlphaEase FC (Alpha Innotech Corp.). Results are expressed as the amount of syndecan-1 shed relative to the control using a calibration curve generated by 2-fold dilutions of culture supernatants from mouse epithelial cells treated with anthrolysin O as a positive control for shedding as described previously (
      • Popova T.G.
      • Millis B.
      • Bradburne C.
      • Nazarenko S.
      • Bailey C.
      • Chandhoke V.
      • Popov S.G.
      ). For each set of measurements, the mean and 95% confidence intervals were calculated using Student’s t test.
      Western Blotting of Syndecan-1 Ectodomains—The conditioned medium from NMuMG cells stimulated for 4 h with purified proteases (250 ng/ml) or phorbol 12-myristate 13-acetate (PMA; 1 μm) was collected, and 1.3% (w/v) potassium acetate and 3 volumes of 95% EtOH were added. After being kept overnight at –20 °C, the samples were dissolved in digestion buffer, and 0.5 volume of each sample were digested overnight with 20 milliunits/ml heparinase II and 20 milliunits/ml chondroitin ABC lyase at 37 °C. These digested and undigested samples were fractionated by SDS-PAGE using 4–20% gradient gels and electrophoretically transferred to Immobilon-Ny+ nylon membrane. The membrane was probed with rat anti-mouse syndecan-1 monoclonal antibody and then horseradish peroxidase-conjugated goat anti-rat secondary antibody and developed by the ECL detection method.
      Mass Spectrometry—Protease-treated proteins were separated by SDS-PAGE. Protein bands were excised from the gel and digested with trypsin (Promega) according to a published procedure (
      • Shevchenko A.
      • Jensen O.N.
      • Podtelejnikov A.V.
      • Sagliocco F.
      • Wilm M.
      • Vorm O.
      • Mortensen P.
      • Shevchenko A.
      • Boucherie H.
      • Mann M.
      ). Tryptic peptides were analyzed by reverse-phase liquid chromatography/nanospray tandem mass spectrometry using an LTQ linear ion trap mass spectrometer (Thermo Electron Corp.) and a fused silica capillary column (100 μm × 10 cm; Polymicro Technologies) with a laser-pulled tip and packed with C18 resin (5 μm, 200-Å pore size; Michrom Bioresources, Inc.). After sample injection, the column was washed for 5 min with mobile phase A (0.4% acetic acid), and peptides were eluted using a linear gradient of 0% mobile phase B (0.4% acetic acid and 80% acetonitrile) to 50% mobile phase B over 30 min at 0.25 μl/min and then to 100% mobile phase B over an additional 5 min. The LTQ mass spectrometer was operated in a data-dependent mode, in which each full mass spectrometric scan was followed by five tandem mass spectrometric scans, where the five most abundant molecular ions were dynamically selected for collision-induced dissociation using a normalized collision energy of 35%. Tandem mass spectra were searched against the rat NCBI Database with SEQUEST (
      • Yates III, J.R.
      • Eng J.K.
      • McCormack A.L.
      • Schieltz D.
      ). For a peptide to be considered legitimately identified, it had to achieve cross-correlation scores of 1.5 for [M + H]+, 2.0 for [M + 2H]2+, and 2.5 for [M + 3H]3+ and a maximum probability of randomized identification of 0.001.

      RESULTS

      Npr599 and InhA Are Abundant in the Culture Supernatant of the B. anthracis Delta Ames Strain—In anthrax pathology, the extracellular secreted proteins distinct from lethal toxin represent virulence factors that cause hemorrhage and other tissue damage (
      • Bonventre P.F.
      • Eckert N.J.
      ,
      • Abramova F.A.
      • Grinberg L.M.
      • Yampolskaya O.V.
      • Walker D.H.
      ). We previously determined the toxic properties of anthrax culture supernatants and the ability of the elastase-like neutral protease (BA3442) belonging to the M4 family to induce hemorrhage in mice (
      • Popov S.G.
      • Popova T.G.
      • Hopkins S.
      • Weinstein R.S.
      • MacAfee R.
      • Fryxell K.J.
      • Chandhoke V.
      • Bailey C.
      • Alibek K.
      ). To understand the molecular mechanisms of anthrax infection and to develop new therapeutic approaches, we undertook the purification and characterization of proteases secreted by B. anthracis. The delta Ames strain (pXO1, pXO2) was cultured in a nutrient-limiting LB medium at 37 °C with vigorous agitation until stationary phase. Proteins from the culture supernatant were precipitated with ammonium sulfate and used for DEAE-cellulose anion-exchange chromatography. Two major peaks with enzymatic activities against casein and elastin were eluted from the column in the flow-through fraction (P1) and in the 200 mm NaCl eluate (P2). The P1 and P2 protease fractions were pulled and further purified to apparent homogeneity on a Sephacryl S-200 gel filtration column. Upon reduced and denaturing SDS-PAGE, the purified enzymes showed a single protein band for P1 with a molecular mass of 36 kDa and two co-purified protein bands for P2 with molecular masses of 46 and 18 kDa (Fig. 1). The proteases are highly abundant (
      • Chitlaru T.
      • Gat O.
      • Gozlan Y.
      • Ariel N.
      • Shafferman A.
      ) and therefore require a purification rate of only 3.2 over the crude culture supernatant.
      Figure thumbnail gr1
      FIGURE 1Purification and identification of Npr599 and InhA. Two proteases were purified from a culture of the B. anthracis delta Ames strain by ammonium sulfate saturation and DEAE-cellulose and Sephacryl S-200 column chromatography. Lane M, prestained molecular mass markers of (from top to bottom) 250, 148, 98, 64, 50, 36, 22, and 16 kDa. C-term, C terminus.
      To identify the proteases and to determine whether the isolated proteins correspond to the particular maturation forms of preproenzymes, we sequenced the N-terminal amino acids by automated Edman degradation. We determined that the P1 protease contains KPVTGTNAVG as a major sequence and VTGTNAVG as a subsequence. These results identify the sequences as the alternatively cleaved N-terminal parts of the catalytic domain of the M4 thermolysin-like neutral protease (NP_843132) with a calculated molecular mass of 34.1 kDa. (The observed molecular mass is 36 kDa.) The full P1 gene (BA0599 in the B. anthracis Ames genome) encodes the protein, which is 99.3% identical to Lactobacillus hydrolase (BAA06144), 99.1% identical to Bacillus cereus neutral protease (AAZ42070), 97.7% identical to bacillolysin (YP034856), and 72.3% identical to bacillolysin MA (BAD60997), all of which belong to the neutral protease family, and which has low homology (33%) to P. aeruginosa LasB (DQ150629) (
      • Read T.D.
      • Peterson S.N.
      • Tourasse N.
      • Baillie L.W.
      • Paulsen I.T.
      • Nelson K.E.
      • Tettelin H.
      • Fouts D.E.
      • Eisen J.A.
      • Gill S.R.
      • Holtzapple E.K.
      • Okstad O.A.
      • Helgason E.
      • Rilstone J.
      • Wu M.
      • Kolonay J.F.
      • Beanan M.J.
      • Dodson R.J.
      • Brinkac L.M.
      • Gwinn M.
      • DeBoy R.T.
      • Madpu R.
      • Daugherty S.C.
      • Durkin A.S.
      • Haft D.H.
      • Nelson W.C.
      • Peterson J.D.
      • Pop M.
      • Khouri H.M.
      • Radune D.
      • Benton J.L.
      • Mahamoud Y.
      • Jiang L.
      • Hance I.R.
      • Weidman J.F.
      • Berry K.J.
      • Plaut R.D.
      • Wolf A.M.
      • Watkins K.L.
      • Nierman W.C.
      • Hazen A.
      • Cline R.
      • Redmond C.
      • Thwaite J.E.
      • White O.
      • Salzberg S.L.
      • Thomason B.
      • Friedlander A.M.
      • Koehler T.M.
      • Hanna P.C.
      • Kolsto A.B.
      • Fraser C.M.
      ). We designated P1 as Npr599.
      The N-terminal sequences of isolated P2 protease were determined as TGPVRGGLNG for the 46-kDa protein and SNGTEKKSHN for the 18-kDa protein. Both of the proteins originate from the M6 family member InhA, which is 98% similar to the homologs from B. cereus and B. thuringiensis (
      • Charlton S.
      • Moir A.J.
      • Baillie L.
      • Moir A.
      ) and which is encoded by the BA1295 gene. Based on this sequence homology, it is likely that the 18-kDa protein (calculated molecular mass of 18.1 kDa) represents the autoprocessed product of InhA, as has been previously shown for B. cereus (
      • Charlton S.
      • Moir A.J.
      • Baillie L.
      • Moir A.
      ).
      Both Npr599 and InhA Are Neutral Zinc Metalloproteases—The caseinolytic activities of Npr599 and InhA were assayed in buffers at a pH range of 4–10. The highest activity at 37 °C was found in the Tris-HCl buffer at pH 7–8, indicating that the isolated enzymes belong to the class of neutral proteases. To estimate the optimum temperature, the proteases were assayed for caseinolytic activity at 21, 37, 50, and 70 °C in the Tris-HCl buffer at pH 7.8. Both enzymes displayed high activity at 37 °C and remained fully active at 50 °C.
      The effect of various inhibitors on protease activity is presented in Table 1. Both Npr599 and InhA were rapidly inhibited by metal-chelating agents such as EDTA and 1,10-phenanthroline, whereas InhA was less sensitive to phosphoramidon and galardin compared with Npr599. High concentrations of the disulfide bond-reducing agent dithiothreitol (10 mm) inhibited both proteases, but milder thiol-reducing compounds like β-mercaptoethanol and l-cysteine (at 1 mm) showed no substantial activity, which was expected because of the absence of cysteine residues in both protein sequences. Of note, the addition of 3.5 μm SDS activated Npr599 by ∼2.4-fold, similar to the effect of Brij 35 on leukocyte elastase activity reported by Cook and Ternai (
      • Cook L.
      • Ternai B.
      ), suggesting that these detergents could mimic a biologically relevant activation mechanism. The divalent metal ions Cu2+, Fe2+, and Zn2+ inhibited the caseinolytic activities of Npr599 and InhA, whereas Ca2+, Mg2+, and Mn2+ did not (Table 2). Both enzymes appear to require zinc for hydrolytic activity because depletion of the metal ion from the active center with 1 mm 1,10-phenanthroline completely abolished the activity against casein, which could not be restored by the addition of excess CaCl2 (1 mm). Both Npr599 and InhA contain a HEXXH motif, which is defined as a zinc-binding domain of metalloprotease (
      • Hooper N.M.
      ). Collectively, the above data agree with the primary structure-based identification of Npr599 and InhA proteases as M4 and M6 zinc metalloenzymes (
      • Barrett A.J.
      ), respectively (see above).
      TABLE 1Effect of protease inhibitors on Npr599 and InhA activities
      InhibitorRemaining activityTypical specificity against the type of proteases
      Npr599InhA
      %
      Control100100
      EDTA
      10 mm6 ± 0.328 ± 5.0Metallo
      1 mm9 ± 0.829 ± 1.4Metallo
      0.1 mm10 ± 1.133 ± 8.0Metallo
      1,10-Phenanthroline
      10 mm0 ± 0.00 ± 0.0Metallo
      1 mm3 ± 0.00 ± 0.0Metallo
      0.1 mm55 ± 4.834 ± 13.8Metallo
      Phosphoramidon
      5 mm1 ± 0.245 ± 1.5Metallo
      0.5 mm2 ± 0.580 ± 0.8Metallo
      0.05 mm12 ± 0.687 ± 1.6Metallo
      Galardin
      1.28 mm0 ± 0.070 ± 1.5Metallo
      0.128 mm12 ± 0.8104 ± 1.3Metallo
      0.0128 mm37 ± 0.8103 ± 3.2Metallo
      PMSF
      10 mm88 ± 10.2111 ± 10.7Serine
      1 mm100 ± 5.4110 ± 8.5Serine
      0.1 mm97 ± 2.395 ± 2.2Serine
      Leupeptin
      10 mm95 ± 5.175 ± 5.9Serine
      1 mm106 ± 3.2105 ± 2.8Serine
      0.1 mm106 ± 2.6111 ± 3.5Serine
      Pepstatin A
      5 mm68 ± 2.232 ± 0.9Acid (carboxylic)
      0.5 mm75 ± 2.986 ± 3.1Acid (carboxylic)
      0.05 mm78 ± 0.588 ± 4.7Acid (carboxylic)
      E-64
      5 mm74 ± 0.559 ± 1.8Thiol (cysteine)
      0.5 mm76 ± 0.178 ± 7.5Thiol (cysteine)
      0.05 mm80 ± 3.984 ± 0.4Thiol (cysteine)
      SDS
      0.35 mm21 ± 0.615 ± 1.8Surfactant
      0.035 mm87 ± 17.278 ± 7.5Surfactant
      0.0035 mm243 ± 5.884 ± 0.4Surfactant
      l-Cysteine
      10 mm0 ± 0.00 ± 0.0Cysteine
      1 mm108 ± 6.8100 ± 2.7Cysteine
      0.1 mm119 ± 4.1116 ± 1.1Cysteine
      β-Mercaptoethanol
      10 mm85 ± 11.886 ± 13.3Sulfhydryl
      1 mm133 ± 7.6106 ± 8.0Sulfhydryl
      0.1 mm121 ± 6.6113 ± 9.4Sulfhydryl
      Dithiothreitol
      10 mm35 ± 2.715 ± 3.5Sulfhydryl
      1 mm84 ± 2.865 ± 8.7Sulfhydryl
      0.1 mm123 ± 1.2115 ± 11.3Sulfhydryl
      TABLE 2Effect of divalent ions on Npr599 and InhA activities
      Divalent ionRemaining activity
      Npr599InhA
      %
      Control100100
      Ca2+
      1 mm86 ± 1.591 ± 5.2
      0.1 mm97 ± 1.494 ± 3.5
      Cu2+
      1 mm0 ± 0.00 ± 0.0
      0.1 mm1 ± 1.00 ± 0.0
      Fe2+
      1 mm0 ± 0.30 ± 0.0
      0.1 mm63 ± 1.848 ± 5.2
      Mg2+
      1 mm83 ± 2.876 ± 6.1
      0.1 mm97 ± 1.497 ± 1.7
      Mn2+
      1 mm79 ± 0.2103 ± 2.5
      0.1 mm96 ± 4.391 ± 5.1
      Ni2+
      1 mm46 ± 1.445 ± 3.4
      0.1 mm77 ± 1.469 ± 1.3
      Zn2+
      1 mm21 ± 0.70 ± 0.0
      0.1 mm78 ± 2.351 ± 6.2
      Potential Substrates for Npr599 and InhA Include ECM- and Hemorrhage-related Proteins—To evaluate the isolated proteases as pathogenic factors, we next surveyed their target molecules that are related to inflammation and innate immune responses. When the internally quenched fluorescent substrates of casein, gelatin, and elastin were used, Npr599 had strong activity for casein (14.1 units/mg) and elastin (17.5 units/mg) and weaker activity for gelatin (6.5 units/mg), whereas InhA had strong activity for casein (14.3 units/mg) and gelatin (16.3 units/mg) and weaker activity for elastin (4.3 units/mg). Individual kinetic parameters for synthetic collagenase substrate hydrolysis were evaluated by Lineweaver-Burk plot as summarized in Table 3. Among the synthetic substrates tested, MCA-Pro-Leu-Gly-Leu-(DNP)-Ala-Ala-Arg-NH2 (kcat = 1.8 s–1) and MCA-Pro-Leu-Ala-Nva-(DNP)DPA-Ala-Arg-NH2 (kcat = 54.5 s–1) were the best turnover substrates for Npr599 and InhA, respectively. InhA was ∼15-fold more catalytically efficient than Npr599 in cleaving MCA-Pro-Leu-Gly-Leu-(DNP)-Ala-Ala-Arg-NH2 in terms of kcat/Km.
      TABLE 3Kinetic parameters for hydrolysis of synthetic collagenase substrates by Npr599 and InhA
      SubstrateKmkcatkcat/Km
      μms-1μm-1 s-1
      Npr599
      MCA-Pro-Leu-Ala-Nva-(DNP)DPA-Ala-Arg-NH21.1 ± 0.260.2 ± 0.090.1 ± 0.04
      MCA-Glu-Val-Lys-Met-Asp-Ala-Glu-Phe-(DNP)-Lys-OH0.3 ± 0.020.1 ± 0.040.4 ± 0.16
      MCA-Pro-Leu-Gly-Leu-(DNP)-Ala-Ala-Arg-NH27.0 ± 0.721.8 ± 0.660.3 ± 0.12
      FALGPAaNDND
      DQ gelatin0.1 ± 0.024.5 ± 0.3246.6 ± 9.41
      InhA
      MCA-Pro-Leu-Ala-Nva-(DNP)Dpa-Ala-Arg-NH26.7 ± 1.0554.5 ± 4.418.3 ± 1.10
      MCA-Glu-Val-Lys-Met-Asp-Ala-Glu-Phe-(DNP)-Lys-OH4.0 ± 0.2517.5 ± 1.934.3 ± 0.38
      MCA-Pro-Leu-Gly-Leu-(DNP)-Ala-Ala-Arg-NH211.3 ± 1.2852.5 ± 6.494.6 ± 0.09
      FALGPANDND
      DQ gelatin0.3 ± 0.0251.2 ± 6.33190.9 ± 27.91
      Because bacterial protease may cause tissue damage by directly degrading host tissues, significant host proteins were tested as substrates of the purified proteases. For example, ECM proteins such as fibronectin, laminin, and collagen types I and IV, which can be degraded during inflammation and bacterial infections, are candidate targets of B. anthracis proteases (
      • Miyoshi S.
      • Shinoda S.
      ,
      • Travis J.
      • Potempa J.
      • Maeda H.
      ). Fig. 2A shows that both Npr599 and InhA indeed effectively cleaved fibronectin and collagen type I, whereas Npr599 was more active with laminin and less active with collagen type IV compared with InhA. In addition to the extracellular structural proteins, α2-macroglobulin, α2-antiplasmin, and α1-protease inhibitor are the most important serum protease inhibitors regulating the activity of plasmin and blood elastase (
      • Borth W.
      ). Fig. 2B shows that both of these proteins were partially degraded by the proteases, which could potentially have high pathological relevance. On the other hand, the purified proteases did not prominently digest IgG, IgM, and interferon-γ, which are important components of humoral and cell immunity (Fig. 2C). Under the same conditions, mucosal IgA was partially cleaved. With regard to the blood coagulation cascade, fibrinogen Aα- and Bβ-chains were completely cleaved by Npr599 within 4 h, unlike the γ-chains, which remained visible in the gel. On the other hand, the fibrinogen Aα-, Bβ-, and γ-chains were completely cleaved by InhA (Fig. 2D).
      Figure thumbnail gr2
      FIGURE 2Potential substrates for Npr599 and InhA. Biologically important substrates were digested with 0.2 μg of Npr599 (P1 lanes) or InhA (P2 lanes) or without protease (No lanes) in each reaction for 4 h at 37°C. Boiled samples were separated by SDS-PAGE and stained with Coomassie Blue. A, digestion of ECM proteins was analyzed on 14% gel for fibronectin (FN; lanes 2–4) and laminin (LM; lanes 5–7) and on 10% gel for collagen type I (Col-I; lanes 9–11) and collagen type IV (Col-IV; lanes 12–14). B, digestion of endogenous serum protease inhibitors was analyzed on 10% gel for α2-macroglobulin (α2-MG; lanes 15–17) and α1-proteinase inhibitor (α1-PI; lanes 18–20) and on 4–20% gel for α2-antiplasmin (α2-AP; lanes 21–23). C, Digestion of immune response proteins was analyzed on 4–20% gel for IgG (lanes 25–27) and IgM (lanes 28–30) and on 10% gel for IgA (lanes 31–33) and interferon-γ (IFN-γ; lanes 34–36). D, digestion of blood coagulation- or tissue damage-related response proteins was analyzed on 10% gel for fibrinogen (Fbg; lanes 38–40) and plasminogen (Plg; lanes 41–43). Lanes 1, 8, 24, and 37 contained molecular mass markers (M).
      In addition, we further examined whether denaturation of collagens leads to enhanced degradation and if the proteases exhibit antiplasmin-inactivating activity. Fig. 3 (A and B) shows that collagen types I, III, and IV became more susceptible to proteolysis after denaturation, albeit native collagens were also degraded effectively. In addition, proteolysis of α2-antiplasmin by Npr599, but not by InhA, led to complete loss of plasmin inhibitory activity (Fig. 3C).
      Figure thumbnail gr3
      FIGURE 3Collagen proteolysis (A and B) and α2-antiplasmin inactivation (C) by proteases. Native (A) and denatured (B; at 95 °C for 2 min) human collagen types I (Col-I; lanes 1–3), III (Col-IIII; lanes 4–6), and IV (Col-IV; lanes 7–9)(2 μg each) were digested with 0.2 μg of either Npr599 (P1 lanes) or InhA (P2 lanes) and without protease (No lanes) for 4 h at 37 °C. Digested fragments were analyzed on 10% gel and visualized by Coomassie Brilliant Blue staining. Lane M contained molecular mass markers. α2-Antiplasmin (alpha2-AP) preincubated with proteases at 37 °C for 4 h was subjected to the plasmin assay as described under “Experimental Procedures” (C). Residual plasmin activity is expressed relative to the untreated control without inhibitor after subtraction of background activity. Error bars represent arithmetic means ± S.D. (n = 3).
      InhA May Modulate the Blood Coagulation Cascade through Regulation of Plasmin Activity—As mentioned above, bacterial proteases can activate the mammalian plasminogen system to induce fibrinolysis and ECM degradation. We next investigated whether protease-mediated cleavage of plasminogen generates plasmin activity. As shown in Fig. 2D, InhA was more active than Npr599 in cleaving human plasminogen and produced a cleavage pattern of five major bands similar to that of bacillolysin MA (
      • Narasaki R.
      • Kuribayashi H.
      • Shimizu K.
      • Imamura D.
      • Sato T.
      • Hasumi K.
      ). We analyzed next protease-catalyzed plasmin activity using the chromogenic synthetic substrate Val-Leu-Lys-p-NA. The degradation of plasminogen did not generate plasmin activity, in contrast to the streptokinase of Staphylococcus aureus used as a positive control (Fig. 4A). This demonstrates that neither Npr599 nor InhA itself is a bacterial plasminogen activator. On the other hand, in the incubation of plasminogen with uPA, the addition of InhA elevated the initial rate of uPA-mediated plasminogen activation (Fig. 4B). This result suggests that InhA, but not Npr599, is a modulator of uPA-catalyzed plasminogen activation. Taken together, the direct cleavage of the endogenous plasmin inhibitors α2-macroglobulin and α2-antiplasmin (Fig. 2B), the inactivation of α2-antiplasmin by Npr599 (Fig. 3C), and the modulating effects of InhA on plasmin activity during anthrax infection (Fig. 4) suggest that the direct proteolytic effects of the secreted proteases during the infectious process are likely to prevent initiation of both blood coagulation and clot fibrinolysis through modulation of the host plasmin-mediated system.
      Figure thumbnail gr4
      FIGURE 4Acceleration of urokinase-dependent plasminogen activation by InhA. A, Npr599 (P1) and InhA (P2) are not bacterial plasminogen activators. Human plasminogen (Plg; 8.3 μg) was incubated at 37 °C with 2 μg of either the protease or streptokinase (SK). The 20-fold diluted resulting reactions were added to 100 μm Val-Leu-Lys-pNA in the presence of fibrin, and the release of pNA was monitored during the incubation. B, uPA-catalyzed plasminogen activation is accelerated by InhA. The reaction (100 μl) contained 200 units/ml uPA; 0.1 units/ml plasminogen; 100 μm Val-Leu-Lys-pNA; and 2, 5, or 10 μg/ml purified proteases. The release of pNA from the chromogenic substrate was monitored at 405 nm for 10 min. Error bars represent S.D. values determined from triplicate measurements.
      Npr599 and InhA Activate Syndecan-1 Shedding through Stimulation of the Host Cell Shedding Mechanism—The proteolytic activity of Npr599 and InhA against components of the ECM prompted us evaluate the effect of these proteases on intercellular interactions in epithelial monolayers. We were specifically interested in the fate of syndecan-1 ectodomains, which are involved in the maintenance of barrier permeability, cytoskeleton organization, and intercellular signaling and which have been recently implicated as mediators of lethality, perturbing different mechanisms of the host defense response (
      • Park P.W.
      • Pier G.B.
      • Hinkes M.T.
      • Bernfield M.
      ,
      • Bernfield M.
      • Go¨tte M.
      • Park P.W.
      • Reizes O.
      • Fitzgerald M.L.
      • Lincecum J.
      • Zako M.
      ). We tested whether anthrax extracellular proteases can modulate syndecan-1 shedding from host cells using a culture of NMuMG epithelial cells. Fig. 5 shows that both Npr599 and InhA functioned as sheddases, releasing soluble syndecan-1 molecules into the culture medium in a time- and dose-dependent manner. Maximum stimulation was reached at a concentration of 250 ng/ml for both Npr599 (∼7-fold increase) and InhA (∼22-fold increase) (Fig. 5A). Furthermore, shedding activation by Npr599 was rapid and saturable by 8 h, whereas InhA was not saturable by this time point (Fig. 5B). At high concentrations (>250 ng/ml), Npr599 became ineffective in the acceleration of syndecan-1 shedding (Fig. 5A). Both Npr599 and InhA were shown to have minimum toxic effects on host cells when tested using the lactate dehydrogenase release assay (data not shown).
      Figure thumbnail gr5
      FIGURE 5Enhancement of syndecan-1 shedding by Npr599 and InhA. Confluent NMuMG cells in 96-well plates were incubated at 37 °C with Npr599 (62.5, 250, and 500 ng/ml) and InhA for 4 h (A) or with protease (250 ng/ml) for 1,4, and 8 h (B). Shed syndecan-1 ectodomain levels were measured by dot-blot analysis as described under “Experimental Procedures.” Error bars represent S.D. values determined from triplicate measurements. AU, arbitrary units.
      Because ectodomain shedding from host cells is inhibited by a variety of substances active in a number of receptor- and stress-activated signaling pathways involving protein-tyrosine kinases protein kinase C, and MAPKs (
      • Park P.W.
      • Foster T.J.
      • Nishi E.
      • Duncan S.J.
      • Klagsbrun M.
      • Chen Y.
      ,
      • Arribas J.
      • Borroto A.
      ), we next analyzed the effects of the inhibitors of several of these pathways on shedding activity. Shedding by both Npr599 and InhA was strongly inhibited by piceatannol, a specific inhibitor of the cytoplasmic Syk family protein-tyrosine kinases (Fig. 6), indicating that cell protein-tyrosine kinase activity is involved in bacterial protease-induced shedding. Suramin is a multipotent inhibitor (
      • Ralevic V.
      • Burnstock G.
      ) that, among other activities, modulates protein-tyrosine phosphatases involved in cell adhesion, integrin signaling, and cell cycle progression (
      • Zhang Y.L.
      • Keng Y.F.
      • Zhao Y.
      • Wu L.
      • Zhang Z.Y.
      ,
      • Stoker A.W.
      ). Specifically, suramin inhibits several protein-tyrosine phosphatases in the low μm range and activates them at higher concentrations. Because of low bioavailability, >50 μm suramin has to be used for the activation effect (
      • McCain D.F.
      • Wu L.
      • Nickel P.
      • Kassack M.U.
      • Kreimeyer A.
      • Gagliardi A.
      • Collins D.C.
      • Zhang Z.Y.
      ). Fig. 6 shows that, similar to piceatannol, suramin stimulated syndecan shedding at 20 μm. At higher concentrations, suramin effectively inhibited syndecan-1 shedding induced by proteases, supporting the notion that the effects of Npr599 and InhA on shedding are mediated by the host cell sheddase(s) through signaling pathways that involve proteintyrosine kinases and protein-tyrosine phosphatases.
      Figure thumbnail gr6
      FIGURE 6Effect of inhibitors on syndecan-1 ectodomain shedding from NMuMG cells enhanced by Npr599 and InhA. NMuMG cells in 1% fetal calf serum were preincubated with the indicated concentrations of inhibitors for 1 h and then exposed to shedding inducers (250 ng/ml Npr599 or InhA) for 24 h. Data are expressed relative to shedding observed without inhibitors in cells either left untreated or treated with Npr599 or InhA. The dashed line represents control syndecan-1 ectodomain shedding by Npr599 and InhA in the absence of inhibitors. SB, SB202190; PD, PD98059; JNK inh. II, JNK inhibitor II; PTK, protein-tyrosine kinase; PTP, protein-tyrosine phosphatase. Error bars represent S.D. values determined from triplicate measurements. *, p < 0.05 compared with the protease-treated control in the absence of inhibitors (paired Student’s t test).
      To determine whether p38, ERK, and JNK signaling pathways are involved in protease-mediated acceleration of syndecan shedding, we tested SB202190 (an inhibitor of p38), PD98059 (an inhibitor of MEK1/2 (MAPK/ERK kinase-1/2; ERK pathway)), and JNK inhibitor II. As shown in Fig. 6, low concentrations of either PD98059 or JNK inhibitor II (5 μm) had some stimulatory effect on syndecan-1 shedding, but all three inhibited the InhA-induced syndecan-1 release in the concentration range of 5–50 μm typical for their activity, although Npr599 seemed to be susceptible only to JNK inhibitor II. The metalloprotease (sheddase) inhibitors galardin, phenanthroline, and phosphoramidon inhibited Npr599-activated shedding only partially and did not inhibit InhA-activated shedding (Fig. 6). On the other hand, these inhibitors were active in the direct caseinolytic assay (Table 1). For example, phenanthroline at >1 μm completely abrogated digestion of casein, but had no activity even at 20 μm with regard to shedding by both proteases. Taken together, these experiments highlight differences between Npr599 and InhA. Although the activities of both converge on the activation of protein-tyrosine kinases and protein-tyrosine phosphatases, Npr599, in contrast to InhA, seems to function rather independently of the cellular stress pathways.
      Npr599 and InhA Can Further Accelerate Syndecan-1 Shedding through Direct Proteolytic Cleavage of the Ectodomain—To investigate whether Npr599 and InhA directly cleave the ectodomain of syndecan-1, we prepared recombinant rat syndecan-1 tagged with GST at the N terminus and expressed in E. coli BL21 host cells. GST-syndecan-1 was purified using glutathione-Sepharose 4B beads. When incubated with Npr599 and InhA, GST-syndecan-1 protein was completely degraded within 1 h (Fig. 7A). However, lethal factor, a metalloprotease component of lethal toxin, had no significant activity on syndecan-1 proteolysis (data not shown). To identify the degraded fragments, Western blot analysis was performed using anti-GST antibody and anti-syndecan-1 antibody (N-18, Santa Cruz Biotechnology, Inc.), raised against a 15–20-amino acid peptide that maps within the first 50 amino acids of syndecan-1 of mouse origin. As shown in Fig. 7B, the major digestion products generated by Npr599 and InhA were ∼32 kDa in size, although the mobilities of the fragments were slightly different. This suggests that recombinant GST-syndecan-1 protein is cleaved at the site adjacent to the N terminus, right after the HS attachment sites.
      Figure thumbnail gr7
      FIGURE 7Direct cleavage of the N terminus of recombinant syndecan-1 by Npr599 and InhA. A, partially purified recombinant syndecan-1 core protein tagged with GST (800 ng) was incubated without (lane 1) or with 100 ng of Npr599 (lane 2) or InhA (lane 3) for 4 h at 37 °C and analyzed by 4–20% SDS-PAGE. After electrophoresis, the gel was immunoblotted with anti-GST antibody. B, the immunoblot was incubated with antibody against the N terminus of the syndecan-1 epitope. Lanes 1–3 are the same as described for A. GST-SDC1, GST-syndecan-1; GST-SDC1 (N-term), the N-terminal fragment of GST-syndecan-1.
      To determine how the digestion pattern of the recombinant protein is relevant to that of syndecan-1 shed from the cell surface, we analyzed the syndecan-1 ectodomains after treatment of NMuMG cells with the purified proteases and B. anthracis culture supernatant. Fig. 8 shows that the sizes of the intact ectodomains (with HS chains attached to the core proteins) shed by both the purified proteases and culture supernatant were different from those of the unstimulated or PMA-stimulated cells (used as an endogenous sheddase activation control) (
      • Subramanian S.V.
      • Fitzgerald M.L.
      • Bernfield M.
      ). The ectodomains after additional heparinase II and chondroitin ABC lyase digestions (with the core proteins stripped from the HS chains) became similar in size in all cases and revealed an additional small fragment generated by the proteases as well as by the culture supernatant. Together with the data on the N-terminal cleavage of recombinant syndecan-1, these findings suggest that Npr599 and InhA are capable of syndecan-1 shedding through the direct proteolysis of the ectodomain at a site apart from that of the cellular sheddase.
      Figure thumbnail gr8
      FIGURE 8Western immunoblotting of syndecan-1 ectodomains shed by the B. anthracis culture supernatant and purified proteases into the culture medium of NMuMG cells. The cells were left unstimulated (lane 1) or were stimulated with 250 ng/ml Npr599 (lane 2) or InhA (lane 3), 10% (v/v) B. anthracis delta Ames culture supernatant in LB medium (LB supt; lane 4), or 1 μm PMA (lane 5). Shed syndecan-1 was separated by 4–20% SDS-PAGE, and the proteins were transferred onto a cationic nylon membrane and probed with anti-syndecan-1 ectodomain antibody (clone 281-2). A, intact syndecan-1 ectodomains migrated as smears because of the heterogeneous length and different number of attached HS chains. B, samples as in A before analysis were additionally digested with 20 milliunits/ml heparinase II and 20 milliunits/ml chondroitin ABC lyase. The positions of PMA-shed syndecan-1 before (*) and after (**) removal of HS chains in the region of 80 kDa are indicated. The fragments generated by direct proteolysis of syndecan-1 ectodomains by exogenous proteases migrated in the region of 60 kDa (***).
      Npr599 and InhA Cleave the Syndecan-1 Fusion Protein at the Asp39–Asp40 and Gly48–Thr49 Bonds, Respectively—Finally, to determine the cleavage sites for each protease, recombinant GST-syndecan-1 fusion protein was incubated with Npr599 and InhA and separated by SDS-PAGE. The major digestion products with the GST tag were trypsinized, and the resulting peptides were subjected to reverse-phase liquid chromatography/nanospray tandem mass spectrometry. In addition to peptides with typical trypsin-cleaving amino acids (Lys and Arg), (R)L17QPALPQIVTANVPPEDQDGSGD39(D) (MH+, 2361.16 Da) was found among the Npr599-digested peptides, whereas peptide (R)L17QPALPQIVTANVPPEDQDGSGDDSDNFSGSG48(T) (MH+, 3227.46 Da) was found with high fidelity among the InhA-digested peptides. These data allow the assignment of Npr599 and InhA cleavage sites in the syndecan-1 core protein to the Asp39–Asp40 and Gly48–Thr49 peptide bonds, respectively. The HS chains in syndecan-1 are known to be attached at Ser37, Ser45, and Ser47 (
      • Zhang L.
      • David G.
      • Esko J.D.
      ). Therefore, Npr599-induced shedding is expected to generate the N-terminal ectodomain fragment 1–39 with one HS chain, whereas InhA-induced shedding would result in the single ectodomain fragment 1–49 containing all three HS chains. The differences in the lengths of the core fragments and the number of attached HS chains may influence the affinity of the detection antibody and would contribute to the low intensity of the Western blot band in the case of Npr599 digestion compared with InhA digestion (Fig. 8).

      DISCUSSION

      In this study, InhA and Npr599 (bacillolysin) were isolated as major proteases in the culture supernatants of the avirulent delta Ames strain of B. anthracis, in agreement with previous proteome analysis (
      • Lamonica J.M.
      • Wagner M.
      • Eschenbrenner M.
      • Williams L.E.
      • Miller T.L.
      • Patra G.
      • DelVecchio V.G.
      ,
      • Antelmann H.
      • Williams R.C.
      • Miethke M.
      • Wipat A.
      • Albrecht D.
      • Harwood C.R.
      • Hecker M.
      ). It is likely that these proteases are controlled by the tetratricopeptide protein Cot43 (
      • Aronson A.I.
      • Bell C.
      • Fulroth B.
      ) and contribute to the activity of lethal toxin and other virulence factors. The direct tissue-degrading activities of bacterial proteases are important factors of microbial pathology at all stages of the infectious process. With regard to inhalation anthrax, hemorrhagic mediastinitis and hemorrhagic thoracic lymphadenitis are typical signs of such an activity (
      • Guarner J.
      • Jernigan J.A.
      • Shieh W.J.
      • Tatti K.
      • Flannagan L.M.
      • Stephens D.S.
      • Popovic T.
      • Ashford D.A.
      • Perkins B.A.
      • Zaki S.R.
      ). In this study, both Npr599 and InhA were found to cleave tissue components such as fibronectin, laminin, and collagen types I and IV with some differences in specificity. Recent analyses confirmed that proteolytic degradation of ECM proteins in anthrax may play an important role in the development of hemorrhage and damage to endothelial and epithelial barriers (
      • Popov S.G.
      • Popova T.G.
      • Hopkins S.
      • Weinstein R.S.
      • MacAfee R.
      • Fryxell K.J.
      • Chandhoke V.
      • Bailey C.
      • Alibek K.
      ,
      • Warfel J.M.
      • Steele A.D.
      • D'Agnillo F.
      ). In addition to structural damage relating to tissues, we have shown that Npr599 and InhA specifically cleaved α1-protease inhibitor (a neutrophil elastase inhibitor), α2-macroglobulin (a universal protease inhibitor), and α2-antiplasmin (a plasmin inhibitor). Proteolysis of α2-anti-plasmin by Npr599 led to loss of its plasmin inhibitory activity (Fig. 3C). As a potential consequence, the host protease inhibitor depletion by bacterial proteases may cause extensive tissue damage by disruption of normal host proteolytic balance and abnormal activation of the host neutrophil elastase (
      • Supuran C.T.
      • Casini A.
      • Scozzafava A.
      ). This mechanism may be relevant to anthrax because it has been shown that the intensity of neutrophil infiltration into tissues is inversely correlated with the outcome (
      • Grinberg L.M.
      • Abramova F.A.
      • Yampolskaya O.V.
      • Walker D.H.
      • Smith J.H.
      ). In addition to this, fibrinogen and plasminogen were degraded by Npr599 and InhA. The in vitro cleavage by the proteases did not generate plasmin activity even in the presence of fibrin as an anti-autolytic agent; however, the cleavage pattern of plasminogen produced by the purified proteases in our study is similar to that produced by bacillolysin MA (
      • Narasaki R.
      • Kuribayashi H.
      • Shimizu K.
      • Imamura D.
      • Sato T.
      • Hasumi K.
      ). It is therefore plausible that, like bacillolysin, Npr599 and InhA may use the host plasmin for both fibrinolysis and invasion. In support of this suggestion, InhA elevated the initial rate of uPA-mediated plasminogen activation (Fig. 4B), and it remains to be determined whether these or other B. anthracis proteases can activate the fibrinolytic system during infection, similar to the streptokinase of S. aureus (
      • Loy J.A.
      • Lin X.
      • Schenone M.
      • Castellino F.J.
      • Zhang X.C.
      • Tang J.
      ). In a transgene experiment with human plasminogen in mice, streptokinase increased mortality during streptococcal infection (
      • Sun H.
      • Ringdahl U.
      • Homeister J.W.
      • Fay W.P.
      • Engleberg N.C.
      • Yang A.Y.
      • Rozek L.S.
      • Wang X.
      • Sjobring U.
      • Ginsburg D.
      ). Based on the loss of endogenous plasmin inhibitor activity upon cleavage with InhA, it seems likely that the cleavage would cause a shift in the hemostatic balance toward the activation of fibrinolysis.
      The syndecan-1-shedding activity of B. anthracis proteases suggests additional pathogenic mechanisms. Syndecan ectodomains are constitutively shed from culture cells as part of normal cell-surface HS proteoglycan turnover. Shedding is also activated as one of the host responses to stress, tissue injury, and other external stimuli (
      • Park P.W.
      • Pier G.B.
      • Hinkes M.T.
      • Bernfield M.
      ,
      • Li Q.
      • Park P.W.
      • Wilson C.L.
      • Parks W.C.
      ) via endogenous metalloproteases (
      • Park P.W.
      • Pier G.B.
      • Preston M.J.
      • Goldberger O.
      • Fitzgerald M.L.
      • Bernfield M.
      ,
      • Popova T.G.
      • Millis B.
      • Bradburne C.
      • Nazarenko S.
      • Bailey C.
      • Chandhoke V.
      • Popov S.G.
      ,
      • Park P.W.
      • Foster T.J.
      • Nishi E.
      • Duncan S.J.
      • Klagsbrun M.
      • Chen Y.
      ,
      • Arribas J.
      • Borroto A.
      ,
      • Andrian E.
      • Grenier D.
      • Rouabhia M.
      ) such as membrane type 1 matrix metalloprotease, MMP-7, and MMP-9 (
      • Andrian E.
      • Grenier D.
      • Rouabhia M.
      ,
      • Wang Z.
      • Go¨tte M.
      • Bernfield M.
      • Reizes O.
      ,
      • Schmidt A.
      • Echtermeyer F.
      • Alozie A.
      • Brands K.
      • Buddecke E.
      ). During disease, the abnormally increased levels of shed syndecans in the bloodstream could modulate susceptibility to infection or even become toxic (
      • Park P.W.
      • Pier G.B.
      • Hinkes M.T.
      • Bernfield M.
      ,
      • Park P.W.
      • Pier G.B.
      • Preston M.J.
      • Goldberger O.
      • Fitzgerald M.L.
      • Bernfield M.
      ). For instance, shed syndecan-1 can tightly bind to the host enzymes cathepsin G and neutrophil elastase and consequently decrease the affinity of these proteases for their physiological inhibitors (
      • Kainulainen V.
      • Wang H.
      • Schick C.
      • Bernfield M.
      ). This effect would further increase the neutrophil damage discussed above.
      Our previous experiments have shown that secreted factors of B. anthracis such as lethal toxin, the pore-forming toxin anthrolysin O, and the cytolytic lipases ClnA and AnlB accelerate the normal process of host cell syndecan-1 ectodomain shedding (
      • Popova T.G.
      • Millis B.
      • Bradburne C.
      • Nazarenko S.
      • Bailey C.
      • Chandhoke V.
      • Popov S.G.
      ). Secreted α- and β-toxins of S. aureus also induce syndecan-1 shedding (
      • Park P.W.
      • Foster T.J.
      • Nishi E.
      • Duncan S.J.
      • Klagsbrun M.
      • Chen Y.
      ). Our data increase the repertoire of B. anthracis sheddases and further support the notion that microbial shedding could exploit the host cell response to a variety of bacterial stress factors, which are likely to cause a cumulative effect of a pathogenic proportion. The experiments with inhibitors in this and previous reports (
      • Park P.W.
      • Pier G.B.
      • Preston M.J.
      • Goldberger O.
      • Fitzgerald M.L.
      • Bernfield M.
      ,
      • Popova T.G.
      • Millis B.
      • Bradburne C.
      • Nazarenko S.
      • Bailey C.
      • Chandhoke V.
      • Popov S.G.
      ,
      • Park P.W.
      • Foster T.J.
      • Nishi E.
      • Duncan S.J.
      • Klagsbrun M.
      • Chen Y.
      ) demonstrate that the accelerated shedding can involve different signaling pathways, thus reflecting the specific nature of the particular inducer, but seems to converge on the stimulation of cytoplasmic tyrosine kinases, ultimately leading to the activation of cellular sheddase. On the other hand, our study presents evidence that shedding of syndecan-1 can also occur through direct ectodomain cleavage by B. anthracis proteases. Recombinant syndecan-1 is susceptible to digestion, and the cleavage sites are located at the Asp39–Asp40 and Gly48–Thr49 bonds in the case of Npr599 and InhA, respectively. In contrast, both constitutive and PMA-induced host-mediated syndecan-1 shedding are expected to result in cleavage at a juxtamembrane site located in the region adjacent to the putative transmembrane domain of syndecan-1 (
      • Wang Z.
      • Go¨tte M.
      • Bernfield M.
      • Reizes O.
      ). The results of the immunoblot experiments using syndecan ectodomains shed from NMuMG cells are in agreement with direct and host-mediated shedding taking place simultaneously. The effect of chemical inhibitors of proteolysis such as phenanthroline on shedding (Fig. 6) indicates that the contribution of direct cleavage is relatively minor for InhA, although it seems to be more important for Npr599. In support of this, piceatannol as the inhibitor of the endogenous mechanism at the highest tested concentration of 50 μm demonstrated some residual shedding (27 and 7% for Npr599 and InhA, respectively), which might be attributed to the direct effect of bacterial proteases. Previously, it has been shown that the P. aeruginosa shedding enhancer LasA, a known metalloprotease virulence factor, stimulates shedding independently of its strong elastolytic activity (
      • Park P.W.
      • Pier G.B.
      • Preston M.J.
      • Goldberger O.
      • Fitzgerald M.L.
      • Bernfield M.
      ). Recently, thrombin and plasmin have been reported to accelerate syndecan-1 and syndecan-4 shedding by a combination of direct and receptor-mediated mechanisms (
      • Schmidt A.
      • Echtermeyer F.
      • Alozie A.
      • Brands K.
      • Buddecke E.
      ). However, our observation is the first example of bacterial proteases mimicking the direct shedding effect of host proteases. Future therapeutic interventions targeting anthrax proteases should take into account the plural nature of their biological activity.

      Acknowledgments

      We thank Dr. Eok-Soo Oh for providing the rat GST-syndecan constructs and Dr. Barney Bishop and Thomas Huff (George Mason University) for assistance with purification experiments.

      References

        • Pezard C.
        • Berche P.
        • Mock M.
        Infect. Immun. 1991; 59: 3472-3477
        • Guidi-Rontani C.
        • Levy M.
        • Ohayon H.
        • Mock M.
        Mol. Microbiol. 2001; 42: 931-938
        • Pellizzari R.
        • Guidi-Rontani C.
        • Vitale G.
        • Mock M.
        • Montecucco C.
        FEBS Lett. 1999; 462: 199-204
        • Agrawal A.
        • Lingappa J.
        • Leppla S.H.
        • Agrawal S.
        • Jabbar A.
        • Quinn C.
        • Pulendran B.
        Nature. 2003; 424: 329-334
        • Pickering A.K.
        • Osorio M.
        • Lee G.M.
        • Grippe V.K.
        • Bray M.
        • Merkel T.J.
        Infect. Immun. 2004; 72: 6382-6389
        • Popov S.G.
        • Popova T.G.
        • Grene E.
        • Klotz F.
        • Cardwell J.
        • Bradburne C.
        • Jama Y.
        • Maland M.
        • Wells J.
        • Nalca A.
        • Voss T.
        • Bailey C.
        • Alibek K.
        Cell. Microbiol. 2004; 6: 225-233
        • Shoop W.L.
        • Xiong Y.
        • Wiltsie J.
        • Woods A.
        • Guo J.
        • Pivnichny J.V.
        • Felcetto T.
        • Michael B.F.
        • Bansal A.
        • Cummings R.T.
        • Cunningham B.R.
        • Friedlander A.M.
        • Douglas C.M.
        • Patel S.B.
        • Wisniewski D.
        • Scapin G.
        • Salowe S.P.
        • Zaller D.M.
        • Chapman K.T.
        • Scolnick E.M.
        • Schmatz D.M.
        • Bartizal K.
        • MacCoss M.
        • Hermes J.D.
        Proc. Natl. Acad. Sci. U. S. A. 2005; 102: 7958-7963
        • Turk B.E.
        • Wong T.Y.
        • Schwarzenbacher R.
        • Jarrell E.T.
        • Leppla S.H.
        • Collier R.J.
        • Liddington R.C.
        • Cantley L.C.
        Nat. Struct. Mol. Biol. 2004; 11: 60-66
        • Jernigan D.B.
        • Raghunathan P.L.
        • Bell B.P.
        • Brechner R.
        • Bresnitz E.A.
        • Butler J.C.
        • Cetron M.
        • Cohen M.
        • Doyle T.
        • Fischer M.
        • Greene C.
        • Griffith K.S.
        • Guarner J.
        • Hadler J.L.
        • Hayslett J.A.
        • Meyer R.
        • Petersen L.R.
        • Phillips M.
        • Pinner R.
        • Popovic T.
        • Quinn C.P.
        • Reefhuis J.
        • Reissman D.
        • Rosenstein N.
        • Schuchat A.
        • Shieh W.J.
        • Siegal L.
        • Swerdlow D.L.
        • Tenover F.C.
        • Traeger M.
        • Ward J.W.
        • Weisfuse I.
        • Wiersma S.
        • Yeskey K.
        • Zaki S.
        • Ashford D.A.
        • Perkins B.A.
        • Ostroff S.
        • Hughes J.
        • Fleming D.
        • Koplan J.P.
        • Gerberding J.L.
        Emerg. Infect. Dis. 2002; 8: 1019-1028
        • Popov S.G.
        • Popova T.G.
        • Hopkins S.
        • Weinstein R.S.
        • MacAfee R.
        • Fryxell K.J.
        • Chandhoke V.
        • Bailey C.
        • Alibek K.
        BMC Infect. Dis. 2005; 5: 25
        • Miyoshi S.
        • Shinoda S.
        Microbes Infect. 2000; 2: 91-98
        • Harrington D.J.
        Infect. Immun. 1996; 64: 1885-1891
        • Dinwiddie R.
        Respiration. 2000; 67: 3-8
        • Lahteenmaki K.
        • Edelman S.
        • Korhonen T.K.
        Trends Microbiol. 2005; 13: 79-85
        • Lahteenmaki K.
        • Virkola R.
        • Pouttu R.
        • Kuusela P.
        • Kukkonen M.
        • Korhonen T.K.
        Infect. Immun. 1995; 63: 3659-3664
        • Park P.W.
        • Pier G.B.
        • Hinkes M.T.
        • Bernfield M.
        Nature. 2001; 411: 98-102
        • Park P.W.
        • Pier G.B.
        • Preston M.J.
        • Goldberger O.
        • Fitzgerald M.L.
        • Bernfield M.
        J. Biol. Chem. 2000; 275: 3057-3064
        • Haynes III, A.
        • Ruda F.
        • Oliver J.
        • Hamood A.N.
        • Griswold J.A.
        • Park P.W.
        • Rumbaugh K.P.
        Infect. Immun. 2005; 73: 7914-7921
        • Marshall L.J.
        • Ramdin L.S.P.
        • Brooks T.
        • Charlton P.A.
        • Shute J.K.
        J. Immunol. 2003; 171: 2057-2065
        • Elenius V.
        • Go¨tte M.
        • Reizes O.
        • Elenius K.
        • Bernfield M.
        J. Biol. Chem. 2004; 279: 41928-41935
        • Wetmore D.R.
        • Wong S.-L.
        • Roche R.S.
        Mol. Microbiol. 1992; 6: 1593-1604
        • Narasaki R.
        • Kuribayashi H.
        • Shimizu K.
        • Imamura D.
        • Sato T.
        • Hasumi K.
        J. Biol. Chem. 2005; 280: 14278-14287
        • Popova T.G.
        • Millis B.
        • Bradburne C.
        • Nazarenko S.
        • Bailey C.
        • Chandhoke V.
        • Popov S.G.
        BMC Microbiol. 2006; 6: 8
        • Shevchenko A.
        • Jensen O.N.
        • Podtelejnikov A.V.
        • Sagliocco F.
        • Wilm M.
        • Vorm O.
        • Mortensen P.
        • Shevchenko A.
        • Boucherie H.
        • Mann M.
        Proc. Natl. Acad. Sci. U. S. A. 1996; 93: 14440-14445
        • Yates III, J.R.
        • Eng J.K.
        • McCormack A.L.
        • Schieltz D.
        Anal. Chem. 1995; 67: 1426-1436
        • Bonventre P.F.
        • Eckert N.J.
        J. Bacteriol. 1963; 85: 490-501
        • Abramova F.A.
        • Grinberg L.M.
        • Yampolskaya O.V.
        • Walker D.H.
        Proc. Natl. Acad. Sci. U. S. A. 1993; 90: 2291-2294
        • Chitlaru T.
        • Gat O.
        • Gozlan Y.
        • Ariel N.
        • Shafferman A.
        J. Bacteriol. 2006; 188: 3551-3571
        • Read T.D.
        • Peterson S.N.
        • Tourasse N.
        • Baillie L.W.
        • Paulsen I.T.
        • Nelson K.E.
        • Tettelin H.
        • Fouts D.E.
        • Eisen J.A.
        • Gill S.R.
        • Holtzapple E.K.
        • Okstad O.A.
        • Helgason E.
        • Rilstone J.
        • Wu M.
        • Kolonay J.F.
        • Beanan M.J.
        • Dodson R.J.
        • Brinkac L.M.
        • Gwinn M.
        • DeBoy R.T.
        • Madpu R.
        • Daugherty S.C.
        • Durkin A.S.
        • Haft D.H.
        • Nelson W.C.
        • Peterson J.D.
        • Pop M.
        • Khouri H.M.
        • Radune D.
        • Benton J.L.
        • Mahamoud Y.
        • Jiang L.
        • Hance I.R.
        • Weidman J.F.
        • Berry K.J.
        • Plaut R.D.
        • Wolf A.M.
        • Watkins K.L.
        • Nierman W.C.
        • Hazen A.
        • Cline R.
        • Redmond C.
        • Thwaite J.E.
        • White O.
        • Salzberg S.L.
        • Thomason B.
        • Friedlander A.M.
        • Koehler T.M.
        • Hanna P.C.
        • Kolsto A.B.
        • Fraser C.M.
        Nature. 2003; 423: 81-86
        • Charlton S.
        • Moir A.J.
        • Baillie L.
        • Moir A.
        J. Appl. Microbiol. 1999; 87: 241-245
        • Cook L.
        • Ternai B.
        Biochem. Int. 1988; 17: 637-646
        • Hooper N.M.
        FEBS Lett. 1994; 354: 1-6
        • Barrett A.J.
        Curr. Opin. Drug Discovery Dev. 2004; 7: 334-341
        • Travis J.
        • Potempa J.
        • Maeda H.
        Trends Microbiol. 1995; 3: 405-407
        • Borth W.
        FASEB J. 1992; 6: 3345-3353
        • Bernfield M.
        • Go¨tte M.
        • Park P.W.
        • Reizes O.
        • Fitzgerald M.L.
        • Lincecum J.
        • Zako M.
        Annu. Rev. Biochem. 1999; 68: 729-777
        • Park P.W.
        • Foster T.J.
        • Nishi E.
        • Duncan S.J.
        • Klagsbrun M.
        • Chen Y.
        J. Biol. Chem. 2004; 279: 251-258
        • Arribas J.
        • Borroto A.
        Chem. Rev. 2002; 102: 4627-4638
        • Ralevic V.
        • Burnstock G.
        Pharmacol. Rev. 1998; 50: 413-492
        • Zhang Y.L.
        • Keng Y.F.
        • Zhao Y.
        • Wu L.
        • Zhang Z.Y.
        J. Biol. Chem. 1998; 273: 12281-12287
        • Stoker A.W.
        J. Endocrinol. 2005; 185: 19-33
        • McCain D.F.
        • Wu L.
        • Nickel P.
        • Kassack M.U.
        • Kreimeyer A.
        • Gagliardi A.
        • Collins D.C.
        • Zhang Z.Y.
        J. Biol. Chem. 2004; 279: 14713-14725
        • Subramanian S.V.
        • Fitzgerald M.L.
        • Bernfield M.
        J. Biol. Chem. 1997; 272: 14713-14720
        • Zhang L.
        • David G.
        • Esko J.D.
        J. Biol. Chem. 1995; 270: 27127-27135
        • Lamonica J.M.
        • Wagner M.
        • Eschenbrenner M.
        • Williams L.E.
        • Miller T.L.
        • Patra G.
        • DelVecchio V.G.
        Infect. Immun. 2005; 73: 3646-3658
        • Antelmann H.
        • Williams R.C.
        • Miethke M.
        • Wipat A.
        • Albrecht D.
        • Harwood C.R.
        • Hecker M.
        Proteomics. 2005; 5: 3684-3695
        • Aronson A.I.
        • Bell C.
        • Fulroth B.
        J. Bacteriol. 2005; 187: 3133-3138
        • Guarner J.
        • Jernigan J.A.
        • Shieh W.J.
        • Tatti K.
        • Flannagan L.M.
        • Stephens D.S.
        • Popovic T.
        • Ashford D.A.
        • Perkins B.A.
        • Zaki S.R.
        Am. J. Pathol. 2003; 163: 701-709
        • Warfel J.M.
        • Steele A.D.
        • D'Agnillo F.
        Am. J. Pathol. 2005; 166: 1871-1881
        • Supuran C.T.
        • Casini A.
        • Scozzafava A.
        Med. Res. Rev. 2003; 23: 535-558
        • Grinberg L.M.
        • Abramova F.A.
        • Yampolskaya O.V.
        • Walker D.H.
        • Smith J.H.
        Mod. Pathol. 2001; 14: 482-495
        • Loy J.A.
        • Lin X.
        • Schenone M.
        • Castellino F.J.
        • Zhang X.C.
        • Tang J.
        Biochemistry. 2001; 40: 14686-14695
        • Sun H.
        • Ringdahl U.
        • Homeister J.W.
        • Fay W.P.
        • Engleberg N.C.
        • Yang A.Y.
        • Rozek L.S.
        • Wang X.
        • Sjobring U.
        • Ginsburg D.
        Science. 2004; 305: 1283-1286
        • Li Q.
        • Park P.W.
        • Wilson C.L.
        • Parks W.C.
        Cell. 2002; 111: 635-646
        • Andrian E.
        • Grenier D.
        • Rouabhia M.
        Oral Microbiol. Immunol. 2006; 21: 123-128
        • Wang Z.
        • Go¨tte M.
        • Bernfield M.
        • Reizes O.
        Biochemistry. 2005; 44: 12355-12361
        • Schmidt A.
        • Echtermeyer F.
        • Alozie A.
        • Brands K.
        • Buddecke E.
        J. Biol. Chem. 2005; 280: 34441-34446
        • Kainulainen V.
        • Wang H.
        • Schick C.
        • Bernfield M.
        J. Biol. Chem. 1998; 273: 11563-11569