Pathogenic bacteria are able to perturb the eukaryotic cell cytoskeleton through signal transduction systems that initiate upon molecular contact at the plasma membrane or through direct interactions of either shed or secreted proteins that gain entry to the cytoplasm. Previously, our laboratory reported that whole bacteria, outer membrane preparations, as well as the native Msp from T. denticola
were able to induce actin reorganization in fibroblasts, including a loss of stress fibers 
. Msp also induced the formation of new actin FBEs 
. The resultant de novo
subcortical actin filament assembly was a crucial determinant for two entirely novel downstream effects: 1) conformational inhibition of store-operated calcium influx and 2) inside-out modulation of the affinity of extracellular ligand engagement 
. In the current investigation, we traced in detail significant pathways and mechanisms that may account for actin filament assembly following Msp induction of FBEs. Our major findings were that 1) Msp stimulates PIP2 production and its local recruitment in fibroblasts, probably leading to the removal of capping proteins from actin filaments to expose barbed ends, and 2) though Msp activates PI3K upstream of small GTPases, neither Rac1 nor PI3K are required for Msp-induced actin rearrangement.
A previous study by our group found that Msp inhibited Rac1 activation selectively in neutrophils, resulting in impaired polarization and chemotaxis in response to a chemoattractant 
. Yet, by an identical method that analyzed bound GTP, we found that Msp clearly activated Rac1 in a time-dependent manner in fibroblasts, as early as 5 min following Msp treatment, suggesting a role for Rac1 in early signalling events in these stromal cells. It is possible that the difference in responses between these cell types may derive from the expression of Rac2 as well as Rac1 by neutrophils, whereas fibroblasts are known to express only Rac1 
RhoA activation is normally associated with the assembly of actin stress fibers and focal adhesions 
. Yet, in our current study, treatment with the native Msp complex resulted in the activation of RhoA along with coincident loss of stress fibers. Under some circumstances, Rho signalling may be uncoupled from stress fiber formation in fibroblasts. Ras-transformed fibroblasts lack actin stress fibers, yet also show high levels of RhoA-GTP 
. It has been determined that prolonged MAPK signalling may also play a major role in stress fiber uncoupling at the level of impairing the response to downstream Rho-kinase function 
. This led us to analyze the levels of Ras-GTP expression in Msp-treated fibroblasts. Using affinity pulldown assays, we observed increased activation of Ras initially at 5 min following Msp exposure. As well, prolonged activation of the MAPKs p38, ERK 
and JNK (M.B. Visser, unpublished), maximal between 15 to 30 min, was observed in Msp-treated fibroblasts, suggesting that these mechanisms may be involved in the coincident Msp-induced RhoA activation and observed loss of stress fibers.
The initial steps in cell migration involve formation of a leading edge that is visible as actin-rich membrane protrusions or ruffles, processes that require Rac1 and PI3K. In the early stages of Msp treatment, Rac1 was recruited to the plasma membrane actin-rich areas, some of which appear to be membrane ruffles. This is similar to the actin filament pattern observed in cells expressing activated Rac1 or by external growth factor stimulation 
. The major mechanism of small GTPase activation involves binding of PIP3 to GEFs 
. Using chemical inhibition studies, we established that Msp activates the PI3K pathway, determined by its requirement for activation of the indirect downstream effector, Rac1. However, the well established PI3K downstream mediator Akt was not activated by Msp. Our studies here suggest that Msp was able to activate PI3K, which in turn would phosphorylate membrane lipids to produce PIP3. PIP3 could then bind GEFs to translocate and activate Rac1 at the plasma membrane as well as bind the PH domain of Akt to mediate its translocation to the plasma membrane. However, activation of Akt by phosphorylation was not observed. Thus, Msp is able to act at the plasma membrane to activate the PI3K pathway; however, individual downstream components of this pathway are evidently affected differently. How Msp acts at the plasma membrane of fibroblasts is not currently known. However, it is unlikely that Msp penetrates the plasma membrane 
; Msp probably acts through activation of cell signalling cascades upon extracellular contact with the plasma membrane. For example, Msp has been demonstrated to form pores in model lipid bilayers and HeLa cell membranes, as well as to depolarize and induce acute calcium ion influxes across plasma membranes 
Although Rac1 and PI3K are well established in the regulation of actin dynamics, it appears that neither of these elements alone are required for Msp-induced actin rearrangement. Neither chemical inhibition of PI3K nor transfection with a Rac1DN construct was able to prevent F-actin reorganization by Msp. Even though studies have shown that expression of either active Rac1 or PI3K alone are able to cause actin rearrangement in fibroblasts 
, our results suggest that Msp impacts either multiple pathways or acts via a further upstream regulator of actin dynamics.
Actin filament assembly, including FBE formation, is directly regulated by both cytoskeleton binding effectors and local changes in polyphosphoinositides such as PIP2 
. It is dependent upon initial formation of FBEs and subsequent elongation of filaments 
. Previously, we reported that Msp-induced actin FBE formation in both fibroblasts and neutrophils 
. Here, we demonstrate for the first time that cell exposure to Msp induces the removal of the capping proteins gelsolin and CapZ from actin barbed ends in fibroblasts. We propose that uncapping is the primary mechanism of FBE formation caused by cell exposure to Msp. PIP2 is known to induce dissociation of both gelsolin and CapZ from actin termini in multiple cell types, including platelets and fibroblasts 
. Our experiments using the PLCδ1-PH-GFP probe to monitor PIP2 dynamics 
together with an ELISA based assay indicated that Msp increases cellular PIP2 levels and induces recruitment of PIP2 to local areas of actin rearrangement at the plasma membrane. Using a cell-permeant peptide PBP-10, which mimics the PIP2 binding site on gelsolin 
, we showed that PIP2 is required for Msp-mediated actin uncapping. Notably, inhibition of PIP2 binding using the gelsolin-derived peptide, PBP-10, is also able to prevent removal of CapZ from actin filaments. We propose that removal of capping proteins by PIP2 is a key mechanism involved in Msp-mediated actin FBE formation.
Changes in PIP2 levels can result from synthesis by phosphoinositol phosphate kinases (PIPK) 
or degradation of PIP3 by phosphoinostide phosphatases (PP) 
. Our studies here, using soluble PIP3 as a substrate, showed that the amount of free phosphate increased following Msp treatment, confirming the involvement of a PP in Msp-mediated PIP2 production. As PLC recognizes the (4,5) PIP2 form rather than the (3,4) PIP2 form 
, PIP3 is probably dephosphorylated at the 3-OH following cell exposure to Msp. The 3- phosphoinositide phosphatase PTEN (phosphatase and tensin homologue deleted on chromosome 10) is able to dephosphorylate PIP3 to PIP2 
and is a compelling candidate that may be involved in PIP2 production by Msp. Similarily, our data using a synthetic bisperoxovanadate compound which is able to inhibit multiple phosphatases, including PTEN 
, indicate that PIP2 is formed by Msp through PTEN.
The actin binding protein cofilin has two functions in the cell, to depolymerize actin filaments, resulting in monomer turnover, and to sever filaments to create FBEs. Cofilin is normally inactive in resting cells, inhibited either by binding to PIP2 or by phosphorylation at serine 3 
. Our data indicate that cofilin remains inactive due to phosphorylation, following Msp exposure. Small GTPases, including RhoA are also able to activate kinases responsible for cofilin phosphorylation 
, which may be at play following Msp exposure, supported by our observed Msp-induced activation of RhoA. Furthermore, Msp-mediated PIP2 production may also be involved in concurrent cofilin inactivation. Therefore, normal actin dynamics and subsequent cell movement are probably impaired, as activation of cofilin is known to be among the key regulators for protrusion of the cell's leading edge 
The pathways by which T. denticola
Msp induces actin filament reorganization appear to be unique among bacterial factors that have been reported to perturb a target host cell's cytoskeleton. Little evidence exists that T. denticola
invades host cells as a significant feature of its pathogenicity 
. Similarly, its Msp does not appear to pass through the plasma membrane into the cytosol (
, M. A. Magalhaes, unpublished). Further we demonstrate here that Msp in its native state in whole bacteria does play a role in host cell cytoskeleton alteration, as a Msp mutant had significantly diminished stress fiber perturbing activity. The analysis of the T. denticola
35405 genome has not found any deduced secretion systems which allow penetration of host cell membranes and injection of bacterial proteins into the cytosol (type III, IV or VI), although other general secretion systems to the extracellular environment are present 
. We suggest that Msp perturbs the fibroblast actin cytoskeleton through a signalling cascade that is novel among those so far reported for bacterial pathogens (). Rather than invading or secreting effectors inside the cell to subvert the host machinery, extracellular contact with Msp induces a fibroblast phosphatase that enhances PIP2 production and recruitment, leading to actin filament uncapping and subsequent de novo
subcortical actin assembly.
A model for the action of Msp in subcortical actin filament assembly in fibroblasts.