P. aeruginosa possesses an impressive array of cell-associated and secreted factors to subvert host cell functions and resist host cell defenses. A key protein in this host pathogen interplay is the type III secreted protein ExoT. Upon translocation into the host cell, this toxin contributes to disruption of the epithelial cell barrier in several ways. It causes disruption of the host cell actin cytoskeleton, leading to cell rounding and loss of cell-cell junctions. It prevents epithelial cell migration in the context of wound healing. Finally, it inhibits bacterial internalization into epithelial cells and macrophages. Avoiding uptake by macrophages may further contribute to bacterial evasion of host cell defenses.
In previous studies, we have shown that the GAP activity of ExoT towards Rho, Rac, and Cdc42 contributes to each of these processes but that mutations in the GAP domain, despite their complete loss of GAP activity, only partially eliminated cell rounding and anti-internalization activity (19
). In this report, we definitively demonstrate by several assays that the ADPRT domain clearly contributes to the biological functions of ExoT. After first eliminating the possibility that other portions of the GAP domain contribute to its biological activity, we found that deletions in the ADPRT domain diminished ExoT activity as did mutating the two key catalytic glutamic acid residues in the ADPRT domain (which results in loss of in vivo ADPRT activity (48
). Inactivation of both domains led to a complete loss of activity in the wound healing, cell rounding, and internalization assays. Expression of either an active GAP or ADPRT domain in yeast inhibited growth, suggesting that both domains are catalytically active and that there are yeast targets for these two enzymatic activities. While other bacterial GAPs have been shown to be toxic to yeast (34
), to the best of our knowledge, this is the first demonstration that a bacterial ADPRT inhibits yeast growth.
The role of the ADPRT domain was most striking in the wound healing experiments. Indeed, inactivation of the ADPRT domain, even in the context of an active GAP domain, reduced its ability to inhibit epithelial cell migration and wound closure. The activity of the GAP mutant in this assay was also diminished to a similar extent, suggesting that both domains contribute equally to this physiologically important activity. This loss of function was observed as early as 4 h after infection, though it became more accentuated at later time points. Interestingly, we have previously observed that the subcellular localization of paxillin and focal adhesion kinase was differentially affected by different mutants of ExoT (21
). These two proteins are components of focal adhesions and their activity and/or phosphorylation states are modulated as cells migrate (37
). When scrape-wounded cells were exposed to wild-type ExoT, these proteins, along with actin, redistributed from focal adhesions to a glob-like structure. In contrast, exposure to the GAP mutant [ExoT(R149K)] resulted in a cytoplasmic staining pattern of these two proteins. This observation suggests that the ADPRT activity of ExoT affects focal adhesions.
In contrast, in the cell rounding experiments, the GAP activity was most important for the early changes in cell morphology. This observation is consistent with our previously published findings that ExoT-mediated cell rounding could be blocked by overexpression of constitutively active RhoA (29
). However, video microscopy revealed the effect exerted by the ADPRT domain. Consistent with the static pictures, the GAP activity caused extensive cell rounding, with loss of cell-cell contacts. In contrast, the ADPRT domain appeared primarily to affect cell-substrate adhesion and while leaving cell-cell contacts unaffected. The effect of wild-type ExoT appeared to be a sum of the GAP and ADPRT activities.
As suggested by our previous demonstration that P. aeruginosa
enters cells through a RhoA-dependent pathway (30
), the GAP activity of ExoT accounted for most of the anti-internalization activity. However, in both the cell rounding and anti-internalization assays, the double mutant had a more severe loss of function phenotype than either single mutant, again underscoring the contribution of the ADPRT domain.
While this work was under review, Sun and Barbieri identified three to five targets of ExoT, depending upon the cell type (48
). Besides auto-ADP ribosylation of itself, ExoT ADP ribosylates two isoforms of Crk. This SH2- and SH3-containing adaptor protein links integrin-mediated signaling through focal adhesions via Rac and Rap1 to multiple host cell responses, including cell migration and adhesion (reviewed in reference 9
). Our results, together with theirs, provide an explanation as to why the ADPRT activity of ExoT is prominent in wound healing and cell migration as well as a mechanism for its effects on host cell morphology. We would suggest that the anti-internalization activity of ExoT on epithelial cells reflects the action of its GAP domain on RhoA-modulated processes. The effect on epithelial cell morphology again is primarily through the inactivation of Rho by the GAP activity of ExoT, but there is also a contribution by the ADPRT domain. The loss of cell substrate adhesion seen with the ADPRT domain alone may reflect the disruption of focal adhesions via an effect on Crk. Finally, the inhibition of wound migration involves the inactivation of Rho, Rac, and Cdc42 by the GAP activity as well as the modulation of Crk activity by the ADPRT domain of ExoT. There are likely other targets of the ADPRT activity of ExoT which may affect additional processes. Consistent with this prediction is our observation that expression of the ExoT ADPRT domain in yeast is toxic, despite the fact that S. cerevisiae
lacks a Crk homolog.
ExoT thus joins a growing family of bifunctional proteins that are translocated by the type III secretion system from gram-negative bacteria into eukaryotic cells. There are now several examples in which a GAP domain is fused to an ADPRT domain, including the closely related protein ExoS as well as AexT, a recently identified type III secreted protein in Aeromonas salmonicida
). In the case of SptP of Salmonella enterica
serovar Typhimurium, the GAP domain is fused to a domain with tyrosine phosphatase activity (15
). It is interesting to speculate on the evolutionary advantages of having a protein that encodes two different enzymatic activities. It may simply be a more cost-efficient way to introduce bacterial proteins into host cells (“two for the price of one”). Alternatively, both domains may share a substrate or benefit from being colocalized in the host cell. For example, Rac is apparently a substrate for both the GAP and ADPRT domains of ExoS (14
) and the ExoS ADPRT domain preferentially modifies membrane-bound over cytoplasmic Ras (41
). Finally, the activity of one domain may modify the activity of the second domain. In fact, it has recently been shown that the ADPRT domain of ExoS modifies the catalytic arginine in its own GAP domain and thus has the potential to down-regulate its own GAP activity (42
). Likewise, ExoT exhibits auto-ADP ribosylation activity in vivo (48
). It will be interesting to determine if ExoT exhibits a similar regulatory mechanism. We have observed that the when the GAP domain is expressed in the absence of a functional ADPRT domain, cell rounding occurs faster and the in vivo GAP activity is greater (P. Balachandran, S. Shafikhani, and J. Engel, unpublished data). Together, these observations suggest that ExoT may similarly downregulate its GAP activity by ADP ribosylation of the critical catalytic arginine residue.
In summary, our work suggests that the ADPRT domain of ExoT is functional in vivo and that the modulation of target proteins synergizes with the GAP activity to disrupt the actin cytoskeleton. These activities likely contribute to the role of ExoT in the pathogenesis of P. aeruginosa infections.