Segments crucial for membrane translocation and pore-forming activity of Bordetella adenylate cyclase toxin

Bordetella adenylate cyclase toxin-hemolysin (CyaA, AC-Hly, or ACT) permeabilizes cell membranes by forming small cation-selective (hemolytic) pores and subverts cellular signaling by delivering into host cells an adenylate cyclase (AC) enzyme that converts ATP to cAMP. Both AC delivery and pore formation were previously shown to involve a predicted amphipathic alpha-helix(502-522) containing a pair of negatively charged Glu(509) and Glu(516) residues. Another predicted transmembrane alpha-helix(565-591) comprises a Glu(570) and Glu(581) pair. We examined the roles of these glutamates in the activity of CyaA. Substitutions of Glu(516) increased specific hemolytic activity of CyaA by two different molecular mechanisms. Replacement of Glu(516) by positively charged lysine residue (E516K) increased the propensity of CyaA to form pores, whereas proline (E516P) or glutamine (E516Q) substitutions extended the lifetime of open single pore units. All three substitutions also caused a drop of pore selectivity for cations. Substitutions of Glu(570) and Glu(581) by helix-breaking proline or positively charged lysine residue reduced (E570K, E581P) or ablated (E570P, E581K) AC membrane translocation. Moreover, E570P, E570K, and E581P substitutions down-modulated also the specific hemolytic activity of CyaA. In contrast, the E581K substitution enhanced the hemolytic activity of CyaA 4 times, increasing both the frequency of formation and lifetime of toxin pores. Negative charge at position 570, but not at position 581, was found to be essential for cation selectivity of the pore, suggesting a role of Glu(570) in ion filtering inside or close to pore mouth. The pairs of glutamate residues in the predicted transmembrane segments of CyaA thus appear to play a key functional role in membrane translocation and pore-forming activities of CyaA.     

Almost half of the RTX domain is dispensable for complement receptor 3 binding and cell-invasive activity of the Bordetella adenylate cyclase toxin

The whooping cough agent Bordetella pertussis secretes an adenylate cyclase toxin (CyaA) that through its large carboxy-proximal Repeat-in-ToXin (RTX) domain binds the complement receptor 3 (CR3). The RTX domain consists of five blocks (I–V) of characteristic glycine and aspartate-rich nonapeptides that fold into five Ca²⁺-loaded parallel β-rolls. Previous work indicated that the CR3-binding structure comprises the interface of β-rolls II and III. To test if further portions of the RTX domain contribute to CR3 binding, we generated a construct with the RTX block II/III interface (CyaA residues 1132–1294) linked directly to the C-terminal block V fragment bearing the folding scaffold (CyaA residues 1562–1681). Despite deletion of 267 internal residues of the RTX domain, the Ca²⁺-driven folding of the hybrid block III/V β-roll still supported formation of the CR3-binding structure at the interface of β-rolls II and III. Moreover, upon stabilization by N- and C-terminal flanking segments, the block III/V hybrid-comprising constructs competed with CyaA for CR3 binding and induced formation of CyaA toxin-neutralizing antibodies in mice. Finally, a truncated CyaAΔ_1295-1561 toxin bound and penetrated erythrocytes and CR3-expressing cells, showing that the deleted portions of RTX blocks III, IV, and V (residues 1295–1561) were dispensable for CR3 binding and for toxin translocation across the target cell membrane. This suggests that almost a half of the RTX domain of CyaA is not involved in target cell interaction and rather serves the purpose of toxin secretion.     

Invasive adenylate cyclase toxin of Bordetella pertussis

Bordetella pertussis produces an adenylate cyclase which is a toxin. The enzyme penetrates eukaryotic cells and, upon activation by host calmodulin, generates high levels of intracellular cAMP; as a result bactericidal functions of immune effector cells are considerably impaired. The toxin is composed of a single polypeptide that possesses both the catalytic and the toxic functions. It penetrates the host cell directly from the plasma membrane and is concomitantly inactivated by a proteolytic degradation.     

Distinct effects of the C-terminal octapeptide of cholecystokinin and of a cholera toxin pretreatment on the kinetics of rat pancreatic adenylate cyclase activity

(1) The kinetic parameters of rat pancreatic adenylate cyclase were evaluated, using GTP, p[NH]ppG or GTPγS as nucleotide activator, cholecystokinin as peptide hormone, and GDPβS and dibutyryl cyclic GMP as inhibitors of guanosine triphosphate and CCK-8, respectively. The time courses of activation and the degree of activation at steady state (E_A/E_TOT) were compatible with a simple two-state model of activation-deactivation based on a pseudo-monomolecular activation process (rate constant k₊₂, and a deactivation process (rate constant k_off) that included, depending on the activating nucleotide, the hydrolysis of GTP (rate constant k₂) and/or the dissociation of the intact nucleotide (rate constant k_?1), so that E_A/E_TOT = k₊₁/(k₊₁ + k₂ + k_?a). (2) The hormone CCK-8 increased the value of k₊₁ with GTP dose-dependently, from 0.2 to 10.9 min^?1. The value of k_?1 increased 0.01 to 0.3 min^?1 but the value of k₂ was unaltered at 7 min^?1, so that E_A/E_TOT increased 15-fold, from 4% to 61%. (3) A cholera toxin pretreatment at 30 μg/ml allowed also a large increase in E_A/E_TOT with GTP (up to 51%) but the underlying mechanism was different. It consisted of a 14-fold decrease in the k_off value of the GTP-activated enzyme (from 7 min^? to 0.5 min^?1) that corresponded to a reduction in GTPase activity. When testing the system with p[NH]ppG, two added effects of the cholera toxin pretreatment were observed: a 4-fold increase in the value of k₊₁ (from 0.2 to 0.8 min^?1) and the occurrence of a significant 0.3 min^?1 value for k_?1.     

Distinct effects of the C-terminal octapeptide of cholecystokinin and of a cholera toxin pretreatment of the kinetics of rat pancreatic adenylate cyclase activity

(1) The kinetic parameters of rat pancreatic adenylate cyclase were evaluated, using GTP, p[NH]ppG or GTP gamma S as nucleotide activator, cholecystokinin as peptide hormone, and GDP beta S and dibutyryl cyclic GMP as inhibitors of guanosine triphosphate and CCK-8, respectively. The time courses of activation and the degree of activation at steady state (EA/ETOT) were compatible with a simple two-state model of activation-deactivation based on a pseudo-monomolecular activation process (rate constant kappa+1), and a deactivation process (rate constant kappa off) that included, depending on the activating nucleotide, the hydrolysis of GTP (rate constant kappa 2) and/or the dissociation of the intact nucleotide (rate constant kappa-1), so that EA/ETOT = kappa+1/(kappa+1 + kappa 2 + kappa-1). (2) The hormone CCK-8 increased the value of kappa+1 with GTP dose-dependently, from 0.2 to 10.9 min-1. The value of kappa-1 increased 0.01 to 0.3 min-1 but the value of kappa 2 was unaltered at 7 min-1, so that EA/ETOT increased 15-fold, from 4% to 61%. (3) A cholera toxin pretreatment at 30 micrograms/ml allowed also a large increase in EA/ETOT with GTP (up to 51%) but the underlying mechanism was different. It consisted of a 14-fold decrease in the kappa off value of the GTP-activated enzyme (from 7 min-1 to 0.5 min-1) that corresponded to a reduction in GTPase activity. When testing the system with p[NH]ppG, two added effects of the cholera toxin pretreatment were observed: a 4-fold increase in the value of kappa+1 (from 0.2 to 0.8 min-1) and the occurrence of a significant 0.3 min-1 value for kappa-1.     

Adenylate cyclase activity in lymphocyte subcellular fractions. Characterization of non-nuclear adenylate cyclase.

Human peripheral lymphocytes were broken in a Dounce homogenizer and subcellular fractions enriched in plasma membranes or microsomal particles and mitochondria were isolated by centrifugation through a discontinuous sucrose gradient. Various agents that promote cyclic AMP accumulation in intact lymphocytes were compared in their ability to stimulate adenylate cyclase activity in the individual fractions. Plasma-membrane-rich fractions that were essentially free of other subcellular particles as judged by electron microscopy and marker enzyme measurements responded to fluoride, but weakly or not at all to prostaglandin E1 and other prostaglandins. Microsomal and mitochondrial-rich fractions responded markedly to both prostaglandin E1 and fluoride. In some, but not all, experiments phytohaemagglutinin produced a modest increase in enzyme activity in plasma-membrane-rich fractions. Catecholamines, histamine, parathyrin, glucagon and corticotropin produced little or no response. In the absence of theophylline, adenosine (1-10 micronM) stimulated basal enzyme activity, although at higher concentrations the responses to prostaglandin E1 and fluoride were inhibited. GTP (1-100 micronM) and GMP(5-1000 micronM) respectively inhibited or stimulated the response to fluoride, whereas the converse was true with prostaglandin E1.     

Structural and functional characterization of an essential RTX subdomain of Bordetella pertussis adenylate cyclase toxin

The adenylate cyclase toxin (CyaA) is one of the major virulence factors of Bordetella pertussis, the causative agent of whooping cough. CyaA is able to invade eukaryotic cells by a unique mechanism that consists in a calcium-dependent, direct translocation of the CyaA catalytic domain across the plasma membrane of the target cells. CyaA possesses a series of a glycine- and aspartate-rich nonapeptide repeats (residues 1006-1613) of the prototype GGXG(N/D)DX(L/I/F)X (where X represents any amino acid) that are characteristic of the RTX (repeat in toxin) family of bacterial cytolysins. These repeats are arranged in a tandem fashion and may fold into a characteristic parallel beta-helix or beta-roll motif that constitutes a novel type of calcium binding structure, as revealed by the three-dimensional structure of the Pseudomonas aeruginosa alkaline protease. Here we have characterized the structure-function relationships of various fragments from the CyaA RTX subdomain. Our results indicate that the RTX functional unit includes both the tandem repeated nonapeptide motifs and the adjacent polypeptide segments, which are essential for the folding and calcium responsiveness of the RTX module. Upon calcium binding to the RTX repeats, a conformational rearrangement of the adjacent non-RTX sequences may act as a critical molecular switch to trigger the CyaA entry into target cells.     

Characterization of adenylate cyclase fromEscherichia coli

Adenylate cyclase activity was detected and characterized in cell-free preparations of different strains ofEscherichia coli; it was localized not only in the membrane fraction but also in the cytoplasm, the localization differing from strain to strain. The adenylate cyclase activity is highly dependent on the method used for disintegration of cells. The best results were obtained when using vortexing of the cell suspension with ballotini beads. The pH optimum of adenylate cyclase in cell-free preparations was found to be 9.0 –9.5. The enzyme has an absolute requirement for Mg²⁺ and is inhibited by sodium fluoride and inorganic diphosphate. Release of adenylate cyclase from the membrane leads to an immediate loss of the activity; it was found that adenylate cyclase is quite labile and hence it could not yet been purified. The method used to determine adenylate cyclase activity and cyclic AMP is described.     

Effects of cholera toxin and 5'-guanylylimidodiphosphate on human spermatozoal adenylate cyclase activity

The effects of cholera toxin and 5′-guanylylimidodiphosphate (Gpp(NH)p) on human spermatozoal adenylate cyclase activity were tested. Cholera toxin had no demonstrable effect on adenylate cyclase activity in human spermatozoa at concentrations between 5 and 20 μg/ml, whether the toxin was preincubated with intact spermatozoa between 5 min and 5 h prior the adenylate cyclase assay, or was added to lysed spermatozoa, where the adenylate cyclase would be accessible to the toxin. In contrast, Gpp(NH)p at concentrations between 10 and 100 μM was effective in activating human spermatozoal adenylate cyclase activity.     

Adenylate cyclase activity in lymphocyte subcellular fractions. Characterization of a nuclear adenylate cyclase.

Nuclei from purified human peripheral lymphocytes were prepared by incubations with Triton X-100 to disrupt the cells, followed by sucrose-density gradient centrifugation. The nuclei were pure as judged by phase-contrast microscopy and had low contents of non-nuclear marker enzymes. In addition, nuclei prepared from lymphocytes surface-labelled with 125I had only 2-7% of the radioactivity bound to intact lymphocytes. At 3.3 mM-Ca2+ and 100 micronM-ATP a fluoride-sensitive adenylate cyclase was demonstrated in nuclei prepared in 0.2% Triton X-100 or 0.33% Triton X-100. There was linear accumulation of cyclic AMP for 10 min in both preparations. The apparent Km for ATP was 90 micronM. Adenylate cyclase activity was augmented by 1.0 mM-Mn2+ and inhibited at higher concentrations. Ca2+ showed two peaks of stimulation, at 1.0-2.5 mM- and above 10 mM-Ca2+. Mg2+ was inhibitory at all concentrations. EDTA OR EGTA only slightly decreased adenylate cyclase activity, suggesting that another metal ion may be necessary for activity. Adenylate cyclase activity was stimulated by 10mM-isoproterenol and 10 micronM-adrenaline in the presence of a phosphodiesterase inhibitor. Phytohaemagglutinin and prostaglandin E1 alone or in combination with isoproterenol had no effect on nuclear adenylate cyclase activity in either nuclei preparation. These results indicate that human lymphocyte nuclei contain one or several adenylate cyclases which differ from adenylate cyclases found in other subcellular fractions of these cells with regard to their bivalentcation requirements and responsiveness to pharmacological agents.     

Pertussis toxin blocks melatonin-induced inhibition of forskolin-stimulated adenylate cyclase activity in the chick brain.

The high-affinity guanine nucleotide-sensitive receptor sites for melatonin in the mammalian hypothalamus and pars tuberalis mediate inhibition of adenylate cyclase (AC) activity. Therefore, we have examined whether similar sites in the chick brain and retina also modulate AC activity. Melatonin did not alter basal or forskolin-stimulated AC activity in whole forebrain or retinal homogenates. In contrast, melatonin significantly inhibited forskolin-stimulated AC activity in forebrain synaptosomal membranes and partially purified retinal membranes in a concentration-dependent manner. Maximal inhibition (approximately 25-30%) of stimulated AC activity was observed at 10-100nM melatonin, while the concentrations (EC50's) which caused half-maximal effects were 22 +/- 6 pM and 30 +/- 5 pM in the brain and retina respectively. Pretreatment of forebrain slices with pertussis toxin abolished the inhibitory effect of melatonin on stimulated AC activity. These data provide the first evidence that melatonin suppresses AC activity in the chick CNS via a pertussis toxin-sensitive G-protein.     

Characterization of Neurospora crassa mutant stratins deficient in adenylate cyclase activity.

Mycelial extracts from Neurospora crassa strains having any one of five different mutations of the cr-1 allele (“crisp”) exhibited an adenylate cyclase specific activity 2 to 3% of that found in wild-type strains. The enzyme deficiency seemed to be specific for the cr-1 mutation but not for the “crisp” morphology and recessive in heterocaryons carrying mutated and wild-type cr-1 alleles. The reduced adenylate cyclase activity detected in extracts from cr-1 mutants was not due to an impairment in the extraction of membranes, to a preferential inactivation of the enzyme after extraction, or to the presence of inhibitors.     

A proper transmembrane Ca2+ gradient is essential for the higher enzymatic activity of adenylate cyclase

Adenylate cyclase from bovine brain cortex was reconstituted into liposomes with (1000 fold) or without transmembrane Ca2+ gradient. The highest enzyme activity (the active center of enzyme exposing outside) was observed in the vesicles with lower Ca2+ concentration outside (approximately 10(-6) M, similar to physiological condition). If the transmembrane Ca2+ gradient was in the inverse direction (i.e. higher Ca2+ concentration outside, 1 mM), a lowest enzyme activity would appear. Such a difference could be diminished following addition of A23187. Obtained results showed that a proper transmembrane Ca2+ gradient is essential for the optimal fluidity of phospholipid bilayer, favouring the formation of suitable conformation of adenylate cyclase with higher enzyme activity.     

Calmodulin regulation of adenylate cyclase activity

Calmodulin-dependent stimulation of adenylate cyclase was initially thought to be a unique feature of neural tissues. In recent years evidence to the contrary has accumulated, calmodulin-dependent stimulation of adenylate cyclase now being demonstrated in a wide range of structurally unrelated tissues and species. Demonstration of the existence of calmodulin-dependent adenylate cyclase has in nearly all instances required the removal of endogenous calmodulin. It is not yet clear whether calmodulin-dependent and calmodulin-independent forms of the enzyme exist and whether some tissues (such as heart) lack a calmodulin-dependent adenylate cyclase. The presence of calmodulin appears largely responsible for the ability of the adenylate cyclase enzyme to be stimulated by submicromolar concentrations of calcium; it may not be relevant to the inhibition of the enzyme which occurs at higher concentrations of calcium. The physical relationship of calmodulin to the plasma membrane bound enzyme (or to the soluble forms of the enzyme) is not known nor is the mechanism of adenylate cyclase activation by calmodulin clear; current data suggest some involvement with both the N and C units of the enzyme. Finally, it is possible that in vivo calcium contributes to the duration of the hormone stimulated cyclic AMP signal. Thus current in vitro data suggest that optimal hormonal activation of calmodulin-dependent adenylate cyclase occurs at very low intracellular calcium concentrations, comparable to those found in the resting cell; conversely the enzyme is inhibited as intracellular calcium increases, following for example agonist stimulation of the cell. These higher calcium concentrations would then activate calmodulin-dependent phosphodiesterase. Such differential effects of calcium on adenylate cyclase and phosphodiesterase would ultimately restrict the duration of the hormone-induced cyclic AMP signal.     

Essential role of GTP in the expression of adenylate cyclase activity after cholera toxin treatment.

Expression of activation of rat liver adenylate cyclase by the A1 peptide of cholera toxin and NAD is dependent on GTP. The nucleotide is effective either when added to the assay medium or during toxin (and NAD) treatment. Toxin treatment increases the Vmax for activation by GTP and the effect of GTP persists in toxin-treated membranes, a property seen in control membranes only with non-hydrolyzable analogs of GTP such as Gpp(NH)p. These observations could be explained by a recent report that cholera toxin acts to inhibit a GTPase associated with denylate cyclase. However, we have observed that one of the major effects of the toxin is to decrease the affinity of guanine nucleotides for the processes involved in the activation of adenylate cyclase and in the regulation of the binding of glucagon to its receptor. Moreover, the absence of lag time in the activation of adenylate cyclase by GTP, in contrast to by Gpp(NH)p, and the markedly reduced fluoride action after toxin treatment suggest that GTPase inhibition may not be the only action of cholera toxin on the adenylate cyclase system. We believe that the multiple effects of toxin action is a reflection of the recently revealed complexity of the regulation of adenylate cyclase by guanine nucleotides.     

Identification by in vitro complementation of regions required for cell-invasive activity of Bordetella pertussis adenylate cyclase toxin

The adenylate cyclase toxin (CyaA) of Bordetella pertussis is a 1706-residue protein composed of an amino-terminal adenylate cyclase (AC) domain linked to a 1300-residue channel-forming RTX ( r epeats in t o x in) haemolysin. The toxin delivers its AC domain into a variety of eukaryotic cells and impairs cellular functions by catalysing unregulated synthesis of cAMP from intracellular ATP. We have examined toxin activities of a set of deletion derivatives of CyaA. The results indicate that CyaA does not have a dedicated target cell-binding domain and that structural integrity and co-operation of all domains, as well as the post-translational fatty acylation mediated by an accessory protein CyaC, are all essential for target cell association and toxin activity of CyaA. When tested individually, all toxin derivatives were inactive and impaired in the tight association with the target cell surface. However, pairs of constructs with non-overlapping deletions complemented each other in vitro and exhibited a partially restored cytotoxic activity. This suggests that at least a part of the active toxin may act in the form of dimers or higher oligomers. The complementation analysis revealed that the last 217 residues of CyaA, containing the unprocessed secretion signal, form an autonomous domain essential for toxin activity, and that the region from residue 624 to 780 may be directly involved in delivery of the AC toxin into cells.     

Requirement for guanosine triphosphate in the activation of adenylate cyclase by cholera toxin

The activation of adenylate cyclase in lysed pigeon erythrocytes requires, among several cofactors, a nucleotide which may be ATP, GTP, or many other triphosphates. However, after removal of endogenous nucleotides by gel filtration or by adsorption onto charcoal the requirement can be met only by GTP, or an analog of GTP. The GTP is required during the activation of the cyclase by toxin even if GTP is also included during the subsequent adenylate cyclase assay, conducted without toxin. In the presence of GTP it is possible to assay for the cytosolic protein that is also required for the action of cholera toxin. By gel filtration, its apparent molecular weight is 15,000–20,000.     

Oligomeric behavior of Bordetella pertussis adenylate cyclase toxin in solution

Adenylate cyclase (AC) toxin from Bordetella pertussis inserts into eukaryotic cells, producing intracellular cAMP, as well as hemolysis and cytotoxicity. Concentration dependence of hemolysis suggests oligomers as the functional unit and inactive deletion mutants permit partial restoration of intoxication and/or hemolysis, when added in pairs [M. Iwaki, A. Ullmann, P. Sebo, Mol. Microbiol. 17 (1995) 1015-1024], suggesting dimerization/oligomerization. Using affinity co-precipitation and fluorescence resonance energy transfer (FRET), we demonstrate specific self-association of AC toxin molecules in solution. Flag-tagged AC toxin mixed with biotinylated-AC toxin, followed by streptavidin beads, yields both forms of the toxin. FRET measurements of toxin, labeled with different fluorophores, demonstrate association in solution, requiring post-translational acylation, but not calcium. AC toxin mixed with DeltaR, an inactive mutant, results in enhancement of hemolysis over that with wild type alone, suggesting that oligomers are functional. Dimers and perhaps higher molecular mass forms of AC toxin occur in solution in a manner that is relevant to toxin action.