Anthrax edema toxin disrupts distinct steps in Rab11-dependent junctional transport

Abstract

Various bacterial toxins circumvent host defenses through overproduction of cAMP. In a previous study, we showed that edema factor (EF), an adenylate cyclase from Bacillus anthracis, disrupts endocytic recycling mediated by the small GTPase Rab11. As a result, cargo proteins such as cadherins fail to reach inter-cellular junctions. In the present study, we provide further mechanistic dissection of Rab11 inhibition by EF using a combination of Drosophila and mammalian systems. EF blocks Rab11 trafficking after the GTP-loading step, preventing a constitutively active form of Rab11 from delivering cargo vesicles to the plasma membrane. Both of the primary cAMP effector pathways -PKA and Epac/Rap1- contribute to inhibition of Rab11-mediated trafficking, but act at distinct steps of the delivery process. PKA acts early, preventing Rab11 from associating with its effectors Rip11 and Sec15. In contrast, Epac functions subsequently via the small GTPase Rap1 to block fusion of recycling endosomes with the plasma membrane, and appears to be the primary effector of EF toxicity in this process. Similarly, experiments conducted in mammalian systems reveal that Epac, but not PKA, mediates the activity of EF both in cell culture and in vivo. The small GTPase Arf6, which initiates endocytic retrieval of cell adhesion components, also contributes to junctional homeostasis by counteracting Rab11-dependent delivery of cargo proteins at sites of cell-cell contact. These studies have potentially significant practical implications, since chemical inhibition of either Arf6 or Epac blocks the effect of EF in cell culture and in vivo, opening new potential therapeutic avenues for treating symptoms caused by cAMP-inducing toxins or related barrier-disrupting pathologies.

Fig 1. EF inhibits Rab11 downstream of GTP loading.

(A-C) Wing imaginal discs expressing wild-type (wt) and mutant forms of YFP-tagged Rab11, using the strong wing-specific MS1096 GAL4 (abbreviated as 1096GAL4), and stained with a Rabbit anti-GFP antibody, reveal different sub-cellular distributions of Rab11. Insets show corresponding Z-sections. (A) Rab11wt, showing apical restriction. (B) Rab11 activated (or Rab11*), showing apical restriction plus junctional concentration, (C) Rab11 Dominant-Negative (or Rab11DN) showing loss of apical/junctional staining. The bipartite staining of Rab11* is even more pronounced in salivary glands (D-I). (D) 1096GAL4>Rab11wtYFP. (E) 1096GAL4>Rab11*YFP with higher magnification in (H). Arrows point at intercellular junctions where Rab11* accumulates. (F) 1096GAL4>Rab11*YFP +EF. EF blocks Rab11* targeting to the junctions, higher magnification shown in (I). Arrows point to intercellular junctions where Rab11* no longer accumulates when co-expressed with EF. (G) 1096GAL4>Rab11DNYFP, showing that Rab11DN does not concentrate at the junctions, but alters the morphology of secretory granules (thin arrows). (J-M): detection of the endogenous Rab11 reveals that it collects near junctions in a punctate pattern (J), with higher magnification in (L). This preference is abrogated in EF-expressing glands (K), with higher magnification in (M). Selective staining of the activated component of endogenous Rab11 (Rab11*) reveals that this functionally relevant form accumulates next to cell junctions in salivary glands (N), sagittal view in (P), an effect that is reduced in EF-expressing glands (O), sagittal view in (Q). The effect of EF on endogenous Rab11* is also visible in wing imaginal discs: (R) wt, (S) 1096GAL4>EF. Co-labeling of Rab11 and D-Ecad, the Drosophila ortholog of E-Cadherin (T-W) reveals co-localization of Rab11 and D-Ecad at the AJs in wt salivary glands (T), (U) higher magnification. Thick arrow indicates AJs, thin arrow indicates punctate stain near the AJs. Co-localization is lost upon EF expression (V) in salivary glands, (W) higher magnification.

Fig 2. PKA* and Rap1* block Rab11 at distinct steps of junctional trafficking.

(A-F) Rab11*YFP distribution (detected by a Rabbit anti-GFP antibody) is differentially affected by PKA* versus Rap1* in Drosophila salivary glands. (A) Rab11* (in 1096GAL4>Rab11*YFP glands) shows a strong preference for the intercellular junctions (see D for higher magnification). (B) PKA* induces ubiquitous redistribution of Rab11* (in 1096GAL4>Rab11*YFP+PKA* glands, see E for higher magnification). (C) Rap1* expression does not alter Rab11* targeting to the junctions, but blocks the final membrane fusion event (in 1096GAL4>Rab11*YFP+Rap1* glands, see F for higher magnification). (G-I) Endogenous Rab11 (detected by a mouse anti-Rab11 antibody) in salivary glands of the indicated genotypes (G) Wild-type (+/+). (H) 1096GAL4>PKA*. Rab11 shows higher levels and loss of junctional preference. (I) 1096GAL4>Rap1*, Rab11 still accumulates near the junctions. (J-L) D-Ecad staining of salivary glands. The salivary gland-specific SglGAL4 was used for these stains. (J) Linear AJs form in wild-type glands. (K) PKA* expression results in weakened junctions and cytoplasmic retention of D-ECad in punctate vesicles. (L) Rap1* expression causes D-Ecad to accumulate around the junctions in small vesicles. (M-R) Wing phenotypes implicating both cAMP effector pathways in Rab11 inhibition, from flies of the following genotypes: (M) wild-type (N) 1096GAL4>PKA*. (O) 1096GAL4>Rap1* (more similar to the EF phenotype, see panel Q). (P) 1096GAL4>Rap1*+PKA* (showing synergism between PKA* and Rap1*). (Q) 1096GAL4>EF. (R) 1096GAL4>EF+Epac^RNAi. Knock-down of Epac expression significantly suppresses the EF phenotype. See S6 Fig showing quantifications of these phenotypes. In contrast, loss-of-function alleles of PKA-C1 do not reduce the EF phenotype significantly (S6 Fig).

Fig 3. EF prevents association between Rab11* and its effector Rip11.

(A) Drosophila salivary glands stained with a Rabbit anti-GFP antibody (Rb α-GFP), showing that dRip11GFP accumulates at cell-cell junctions in 1096GAL4>dRip11-GFP larvae. (B) 1096GAL4>dRip11-GFP+EF glands. Junctional staining is moderately weakened by EF. (C-F) Salivary glands co-stained with a Rat anti-GFP antibody (Rt α-GFP) and a Rabbit anti-dRip11 antibody (Rb α-Rip11), showing that the co-localization between Rab11* and dRip11 detected with this antibody combination is lost or reduced upon co-expression of Rab11*YFP with EF, PKA*, or Rap1*. Lower panels show higher magnifications. Insets show representative examples of vesicles with or without Rab11*/dRip11 co-localization. (C) 1096GAL4>Rab11*YFP, arrows indicate Rab11*/dRip11 co-localization. (D) 1096GAL4>Rab11*YFP+EF. (E) 1096GAL4>Rab11*YFP+PKA*. (F) 1096GAL4>Rab11*YFP+Rap1*. Arrows point to adjacent but non-overlapping punctae. (G) Quantification of Rab11*YFP/dRip11 co-localization in panels C-F measured by the Pearson’s coefficient. (H-I) MDCK cells transfected with constructs expressing human Rip11-GFP and Rab11wt-DsRed. (H) Untreated cells showing Rab11-Rip11 co-localization throughout the cell with higher levels of Rip11 and Rab11 at cell borders. (I) Rip11 and Rab11 no longer co-localize in cells treated with ET (Edema Toxin = EF+PA). While a minor component of Rip11 is still evident at cell borders, Rab11 fails to reach the cell borders.

Fig 4. Activation of Arf6 partially mimics the effect of EF.

(A-G) Wings showing EF and Arf6-dependent phenotypes. (A) 1096GAL4>EF. (B) 1096GAL4>Arf6*. Activated Arf6 (Arf6*) causes a phenotype very similar to that caused by EF. (C) 1096GAL4>EF+Arf6*, showing an additive phenotype. (D) 1096GAL4>Arf6wt wing, displaying a wild-type phenotype. (E) 1096GAL4>EF+Arf6wt wing revealing synergy between EF and Arf6wt. (H-I) Wing imaginal discs stained with an anti Rab11 antibody. (H) In wild-type discs, Rab11 displays a dotted apical distribution. (I) In discs expressing Arf6* (1096GAL4>Arf6*), Rab11 levels are reduced, and apical restriction is lost. (J-K) Wing imaginal discs expressing Sec15-GFP. (H) Sec15-GFP expressed at high levels in 1096GAL4>Sec15-GFP discs forms large fluorescent punctae. Like EF, Arf6* expression in (K) 1096GAL4>Sec15-GFP+Arf6* discs blocks formation of Sec15-GFP punctae and reduces Sec15-GFP levels. (L-O) D-Ecad staining in wing imaginal discs. (L) D-Ecad stain in wild-type discs, revealing the apical network of AJs. (M) In 1096GAL4>Arf6* discs, apical D-Ecad levels are severely reduced, an effect similar to that of EF (N). Arf6-RNAi suppresses the EF phenotype and partially restores normal D-Ecad at AJs (O). (P) A 1096GAL4>Rab11*YFP salivary gland showing Rab11* targeting to the AJs, revealed by an anti-GFP stain. (Q) Arf6* blocks Rab11* targeting to cell junctions in 1096GAL4>Rab11*+Arf6* glands. (R) Endogenous Rab11 stain of wild-type salivary glands. (S) Arf6* causes endogenous Rab11 to accumulate in the cytoplasm and diminishes its localization at junctions. (T) D-Ecad stain of wild-type salivary glands. (U) Arf6* results in accumulation of large intracellular inclusions of D-Ecad in 1096GAL4>Arf6* salivary glands.

Fig 5. Arf6 and Epac/Rap1 play key roles in mediating EF toxicity in mammalian systems.

(A-E) HBMECs (Human Brain Microvascular Endothelial Cells) stained with an anti pan-Cadherin antibody. (A) The pan-Cadherin (p-Cad) stain clearly delineates cell borders in untreated cells (UT). (B) ET (EF+PA) treatment results in strong reduction in Cadherin staining and loss of AJs. (C) Co-treatment of cells with ET and the Slit2 peptide, which prevents Arf6 activation via induction of ArfGAP activity, fully restores AJs compromised by ET treatment. (D) Co-treatment with the Epac-specific inhibitor ESI-09 partially rescues p-Cad expression at AJs. (E) Co-treatment with the PKA-inhibitor H89 does not appreciably rescue junctional expression of pCad at cell borders but does restore intracellular p-Cad staining (F-I) ET-induced footpad swelling (edema) can be suppressed by pharmacological intervention. (F) ET-injected footpads develop a striking edema. Pre-treatment with ESI-09 consistently blocks this symptom. Pre-treatment with (G) SecinH3, a compound that indirectly suppresses the activation of Arf6 (****p<0.0001, ***p<0.001), and (H) the Epac pathway inhibitor ESI-09, but not the PKA inhibitor H89, abrogate EF-induced edema (****p<0.0001, ns non significant). (I) AG1024, a compound that may indirectly block Rap1 activation, acts as a potent inhibitor of ET-induced edema at early times, but offered more modest protection at a later time point (****p<0.0001, *p<0.05).

Fig 6. Summary diagram.

EF-induced cAMP overload can activate either or both of the PKA and Epac/Rap1 effector pathways depending on biological context. In this model, uncontrolled cAMP production by EF leads to activation of PKA and/or Epac. PKA stimulation promotes dissociation of Rab11 from its effectors Rip11 and Sec15. This premature dissociation may prevent the activated form of Rab11 from reaching the AJs, and thereby block delivery of cargo proteins (e.g., Cadherins and Notch ligands). In some cellular contexts, dissociation of Rab11-GTP from Rip11 may also lead to Rab11 degradation (e.g., Drosophila wing discs and human endothelial cells), while in others, only to the loss of junctional accumulation (e.g., Drosophila salivary glands). Over-activation of Rap1 by Epac leads to inhibition of exocyst-mediated vesicle fusion at the cell surface. Epac/Rap1 may act indirectly via sequential activation of RalA and Arf6, which cross inhibits Rab11-mediated cargo delivery to junctions, or directly through an unknown mechanism. The relative contribution of each cAMP-responsive pathway may depend on cell type and organism, although, the Epac/Rap1 branch (blocked by ESI09, and possibly by AG1024) appears to be the primary mediator of EF in fly wings, human endothelial cells, and mouse footpads. The Arf6 GTPase inhibits Rab11* targeting to the AJs, an effect that may be mediated by its interaction with exocyst components. Inhibition of Arf6 by Slit2 or Secin H3 provides protection against EF in cultured humans cells and in vivo in mice, respectively.