We show here that by conducting Ca2+ ions across target cell membrane, CyaA decelerates its endocytic uptake and escapes from rapid macropinocytic removal from cell membrane and destruction in endosomes. By redirecting the toxin into a decelerated clathrin-dependent uptake pathway, the calcium-conducting activity of toxin translocation intermediates protracts toxin pore persistence within cytoplasmic membrane, thus extending phagocyte permeabilization and maximizing cytotoxic action of CyaA, as summarized in the model proposed in .
Model of the calcium influx-triggered positive feedback mechanism of self-exacerbating potassium efflux due to protracted cell permeabilization by toxin pores.
This mechanism could be directly demonstrated here for the CyaA-AC− toxoid and CyaA-ΔAC hemolysin variants that lack the AC enzyme activity and could be used at sufficiently high concentrations for live cell imaging. Imposing on CyaA-ΔAC a conformation that enabled it to conduct Ca2+ into cells, indeed, rescued the hemolysin from the macropinocytic pathway, and by decelerating its removal form cell membrane, it particularly enhanced the otherwise very low cell-permeabilizing and cytolytic capacity of CyaA-ΔAC.
Using intact toxin (AC+
) purified form B. pertussis
cells at close to physiologically relevant concentrations (200 ng/ml), we found that inhibition of endocytic uptake of Bp
-CyaA enhanced its capacity to lyze cells. It was, however, not possible to design an experiment that would directly prove the role of calcium-induced deceleration of CyaA uptake in its cytotoxic action. Such demonstration would only be possible if inhibition of Ca2+
influx would leave the other cytotoxic activities of CyaA unaffected. This would then allow to test whether cytolytic activity of CyaA is reduced by acceleration of the endocytic uptake of the intact toxin. However, Ca2+
influx into cells is a prerequisite for CyaA relocation into lipid rafts and subsequent AC domain translocation into cell cytosol 
. Therefore, blocking of Ca2+
influx by whatever means inevitably ablates also the major component of the cytotoxic activity of CyaA, namely the capacity of its AC enzyme to reach cell cytosol and catalyze unregulated dissipation of cytosolic ATP into cAMP to impair cellular signaling.
We have previously shown that upon initial binding of the CD11b/CD18 integrin, the CyaA toxin inserts into the membrane lipid bilayer as a translocation precursor, in which the segments of the AC domain cooperate with segments of the pore-forming domain in forming a novel calcium-conducting path across the phagocyte membrane that mediates influx of extracellular Ca2+
ions into cells 
. Activation of calpain by incoming Ca2+
then yields cleavage of the talin tether of CD11b/CD18 and liberates the toxin-receptor complex for recruitment into membrane lipid rafts 
. This process appears to be entirely independent of the cAMP-generating capacity of CyaA in CD11b+
J774A.1 macrophage cells, as a quite comparable amplitude and even faster initial kinetics of calcium mobilization into cells is observed with the enzymatically inactive dCyaA variant (CyaA-AC−
) than by intact CyaA 
. The results presented in this study then show that it is rather the permeabilization of cells for influx of Ca2+
, than the toxin relocation into lipid rafts itself, which allows the toxoid to escape the rapid macropinocytic removal from cell surface with the cytoplasmic membrane. Moreover, lipid rafts are tiny microdomains in the membrane that would accommodate only limited amounts of the hemolysin, while a quantitative escape from rapid macropinocytosis of dCyaA-KP toxoid and its redirection for decelerated endocytosis was observed upon co-incubation with dCyaA (). This indicates that by permeabilizing cells for calcium influx, the AC-translocating CyaA conformers provoke also deceleration of endocytosis of the bystander CyaA pores that can form outside the rafts, in the bulk phase of cell membrane.
Such conclusion would also be in line with the observed opposing phenotypes of the CyaA-E509K+516K (CyaA-KK) and CyaA-E570Q+K860R (CyaA-QR) variants of CyaA. While the CyaA-KK exhibits a strongly enhanced specific pore-forming activity, its capacity to promote Ca2+
influx into cells is very low 
. On the opposite, the CyaA-QR toxin exhibits a very low capacity to form pores, to permeabilize cells and elicit K+
efflux, while it mediates normal levels of calcium influx into cells 
. This suggests that K+
efflux and Ca2+
influx are two parallel and independent processes that are associated with distinct conformational and/or oligomeric states of the two co-existing conformers of membrane-inserted CyaA.
In this respect, it is puzzling that binding of the 3D1 antibody to CyaA-ΔAC exacerbated at the same time its capacity to conduct calcium ions into cells, as well as its cell permeabilizing capacity. This would, perhaps, be best explained by the bystander effect mentioned above. Due to association-dissociation equilibrium, the antibody would lock only a fraction of membrane-inserted CyaA-ΔAC molecules into a calcium-conducting conformation. The other fraction of hemolysin molecules, not bound with 3D1, would hence be free to form the cell-permeabilizing oligomeric pores promoting K+ efflux, benefiting from the calcium-induced deceleration of the endocytic uptake of the hemolysin pores. It remains, nevertheless, to be conclusively shown that bridging of CyaA-ΔAC dimers by the bivalent 3D1 antibody does not also contribute to the enhanced cell permeabilization capacity and K+ efflux mediated by the CyaA-ΔAC hemolysin complexes with the antibody. 3D1 might, indeed, potentially stabilize the formed oligomeric pores against dissociation within the membrane. At present it can neither be formally excluded that the complex of membrane-inserted CyaA-ΔAC with 3D1 might be conducting at the same time the Ca2+ ions into cells and the K+ ions out of the cell. For intact CyaA the accumulated evidence strongly argues against such possibility (see above). Due to absence of the translocated AC domain, however, the 3D1-locked conformers of CyaA-ΔAC may be capable of oligomerization into pores enabling K+ efflux, or the calcium-conducting path formed by these conformers may be accessible for efflux of cytosolic K+ ions in the absence of the AC domain.
A particularly intriguing observation is that the Ca2+
influx and K+
efflux-promoting activities of the toxin synergize in manipulating the pathway and kinetics of CyaA endocytosis. Indeed, the decrease of intracellular potassium level was repeatedly shown to cause inhibition of clathrin-mediated endocytosis 
. Hence, the more the cell becomes permeabilized for efflux of K+
, upon inhibition of macropinocytic uptake of toxin pores by incoming Ca2+
, the more the potassium leakage from cells through toxin pores further decelerates the clathrin-dependent removal of CyaA pores from cell membrane. Such cooperation of Ca2+
influx and K+
efflux-mediating activities of CyaA would thus generate a positive feedback loop, exacerbating potassium depletion due to steadily increasing extent of cell membrane permeabilization by persisting and/or newly forming CyaA pores.
The above outlined positive feedback loop of potassium efflux was apparently operating under the used experimental conditions, since vesicles containing dCyaA were observed to accumulate as tightly attached to, or located just beneath the cell membrane, for over 20 minutes from toxoid addition ( and Video S1
). This shows that excision of clathrin-coated vesicles from cell membrane was inhibited and protraction of cell membrane permeabilization by the persisting pores then fed back into deceleration of endocytic uptake of dCyaA.
The enzymatically active CyaA could not be used in this study for analysis of endocytic trafficking of CyaA by live cell microscopy imaging, as at the high concentrations of labeled proteins (>1 µg/ml), required for this type of analyzes, the enzymatically active CyaA uncontrollably impairs intracellular trafficking by rapid depletion of ATP and induction of cell death 
. Indeed, the cytotoxic action of enzymatically active CyaA on CD11b+
phagocytes was documented repeatedly at doses lower than 10 ng/ml, where less than 1 ng/ml of active CyaA toxin quantitatively inhibits oxidative burst in neutrophils 
. As low CyaA doses as 5 to 10 ng/ml elicit monocyte ruffling and a near-instant inhibition of CR3-mediated opsonophagocytosis or arrest of the fluid-phase uptake, respectively. This appears to be due to cAMP signaling-mediated inhibition of the small GTPase RhoA and possibly of the PI3 kinase 
. The results obtained here with the toxoids appear, nevertheless, to be relevant also to the understanding of endocytosis and of cytotoxic action of the enzymatically active CyaA. By using a newly developed FACS assay for cell surface accessibility of bound CyaA, we could approach here the kinetics of endocytic removal of intact CyaA from cell surface at close to physiologically low toxin concentrations (200 ng/ml). Under such conditions, the receptor-bound CyaA was found to be progressively removed from cell surface over 15 minutes of incubation with about one quarter of toxin molecules escaping the uptake into endosomes and remaining exposed on the surface of cell membrane. This endocytic uptake of intact CyaA from cell membrane was noticeably slowed down in the presence of a five-fold excess of enzymatically inactive CyaA-AC−
molecules that promoted Ca2+
influx into cells and K+
efflux in trans
. Moreover, inhibition of the endocytic uptake through inhibition of ATP re-synthesis strongly enhanced the capacity of native Bp
-CyaA toxin to permeabilize and lyze cells already at as low toxin concentrations as 200 ng/ml. This points towards a more important contribution of the pore-forming activity to the overall cytotoxic action of CyaA than previously recognized 
Enzymatically inactive but pore-forming CyaA-AC−
toxoids have been extensively used over the past 18 years for delivery of AC-inserted CD8+
T cell epitopes from viruses, mycobacteria or tumors into the MHC class I and II-restricted antigen presentation pathways of CD11b-expressing dendritic cells 
. As dCyaA-based vaccines for cancer immunotherapy are currently in phase I of clinical studies (www.genticel.com
), the here reported insight into dCyaA endocytosis and trafficking paves the way towards deciphering of the efficacy of T cell vaccine delivery by CyaA toxoids. Moreover, an endotoxin-free CyaA-AC−
(dCyaA) toxoid was recently observed to alter the expression levels of a dozen of genes involved in innate immune response signaling in bone marrow-derived macrophages 
. This is likely due to the capacity of the toxoid to promote Ca2+
influx into cells and permeabilize cells for K+
efflux. A long-sought evidence for a role of the pore-forming capacity of CyaA in Bordetella pertussis
infection has, indeed, been recently reported by Dunne and coworkers 
. This study showed that by eliciting K+
efflux from dendritic cells, and perhaps some other CD11b-expressing phagocytes, the pore-forming activity of CyaA contributes to activation of the NALP3 inflammasome and thereby to induction of innate IL-1β response, which supports the clearance of Bordetella
bacteria at later stages of infection. The results reported herein show that the capacity of CyaA to permeabilize cells for K+
efflux depends on the capacity of the toxin to promote Ca2+
influx into cells and escape the rapid macropinocytic removal from target cell membrane. This reveals a further layer of sophistication of CyaA action on host cells, thus underpinning the key role of this toxin with multiple ‘talents’ in the virulence of Bordetellae