Recent studies have suggested a connection between pathogen mediated actin re-organization and serum response factor (SRF) transcriptional programs 
. We screened a panel of gastrointestinal tract-associated pathogens for the ability to induce nuclear accumulation of MAL-GFP. Surprisingly only EPEC caused a significant change in MAL-GFP localization under the infection conditions tested. We suspected S.
Typhimurium would have some effect on MAL-GFP localization. It has been shown that S.
Typhimurium induces actin ruffles during entry and activates host cell Rho-GTPases 
. However, unlike EPEC infection, Salmonellae
rapidly return the host cell cytoskeleton to its resting state following engulfment, via the action of the effector protein SptP 
. Perhaps this down-modulation of actin polymerization by S.
Typhimurium is sufficient to stifle the activation of SRF, whereas the prolonged actin remodelling induced by EPEC infection is not.
We confirmed that MAL translocation correlates with upregulation of SRF target genes during EPEC infection ( and S1
). EPEC infection selectively activates SRF target genes, most significantly EGR2 and IL-6, but also CDC42EP3, ARHGDIB, ACTA2 and VAV3, relative to uninfected controls. None of these genes were activated by EPEC Δtir
infection. CDC42EP3, ARHGDIB, and VAV3 are all involved in Rho, Rac or Cdc42-mediated signalling and are consistent with the Rho dependent pathway of MAL translocation and SRF activation 
. Whether upregulation of these genes is required for pedestal formation or pathogen survival, or is a natural consequence of pedestal formation is unclear at this time, but warrants further study.
EGR2 is an immediate-early, zinc finger transcription factor with two serum response elements in its 5′ flanking sequence 
. EGR2 can be activated by a number of infectious agents including viruses (Human T-cell Leukemia virus type 1), bacteria, and parasites (Toxoplasma gondii
. Interestingly in T. gondii
infection EGR2 induction was dependent on rhoptry secretion, a process analogous to secretion of proteins into a host cell by the bacterial type III secretion system 
. Likewise, we find the secreted protein Tir to be essential for EPEC-induced activation of EGR2. In other infections EGR2 expression is often accompanied by EGR1 and c-FOS. Under our experimental conditions the expression of EGR1 and c-FOS was not induced. This may suggest that this is an EPEC-specific response rather than a general innate pathogen response.
It is clear that host signalling pathways are activated in response to many infectious agents, suggesting they are functioning in innate immunity. Although IL-6 is a well-known SRF target 
its expression can be induced by a number of bacteria 
, it is possible therefore, that IL-6 may function as an innate sentinel in this context. The fact that none of these genes were induced by infection with EPEC Δtir
demonstrates that pedestal formation is fundamental to this signalling cascade.
Tir is an essential effector for the assembly of F-actin pedestals. Following secretion, Tir inserts into the host cell membrane, presenting an extracellular domain that binds the bacterial surface protein intimin 
. The C-terminal region of TirEHEC
is phosphorylated at Tyr474 by host-cell kinases 
in a manner similar to host receptor phosphorylation 
. Phosphorylated Y474 and its flanking residues bind Nck via its SH2 domain 
. Nck subsequently recruits and activates N-WASP stimulating ARP2/3 driven F-actin assembly. In addition, TirEPEC
can promote weak actin polymerization in an Nck-independent manner via phosphorylation of Tir residue Y454 
. In this report we show that Tir is essential for EPEC-induced MAL-GFP nuclear accumulation and subsequent transcriptional activation of selective SRF target genes. Infection of epithelial cells with EPEC Δtir
does not induce MAL-GFP nuclear accumulation, but this phenotype is rescued by the exogenous expression of Tir (). This is consistent with actin rearrangement driven by pedestal formation being key for SRF activation rather than a translocated effector activating SRF directly. In further support of this idea TirEHEC
could also rescue the EPEC Δtir
phenotype (). TirEHEC
is functionally divergent from TirEPEC 
lacks a residue equivalent to Tyr474 
, is not tyrosine phosphorylated in cells 
and does not bind Nck 
. To efficiently form actin pedestals EHEC requires a second translocated effector EspFU
is recruited indirectly to Tir by IRTKS 
, where it can than activate N-WASP which results in actin polymerization. Although the initial signalling methods used to recruit and activate host cell nucleation factors between the related pathogens are different, the net result is the same. Likewise, single mutations of either TirEPEC
Y454 or Y474 to non-phosphorylatable phenylalanines drastically reduced the nuclear accumulation of MAL-GFP to similar levels ( and S2
), suggesting that Nck dependent or independent activation of N-WASP is irrelevant to EPEC-induced MAL-GFP nuclear accumulation. In addition, knockdown of SRF reduced EPEC-induced MAL-GFP accumulation in the nucleus to near uninfected levels (). This is likely the result of altered cytoskeletal gene expression, resulting from the loss of SRF.
In order to identify the host signaling cascades that are co-opted by bacterial virulence factors to regulate the cytoskeleton, we sought a scheme to identify genes generally employed in mammalian cytoskeleton control. We picked known and putative actin-associated or regulatory genes and tested their ability to induce nuclear accumulation of MAL-GFP. Novel genes inducing MAL-GFP nuclear accumulation with probable involvement in actin-cytoskeletal rearrangement were then evaluated for involvement in host-pathogen interactions.
We identified FLRT3, TESK1 and C22orf28 as novel inducers of MAL nuclear accumulation and confirmed the involvement of FLRT3 in EPEC induced MAL translocation by siRNA (). Overexpression of ABRA induced nuclear accumulation of MAL-GFP consistent with published data for the Murine homologue STARS 
. Knockdown of ABRA significantly decreased EPEC induced accumulation of MAL-GFP in the nucleus, suggesting that ABRA is a necessary component in the signaling pathway. In addition we found ABRA was enriched in EPEC pedestals and that ABRA knockdown adversely affected pedestal morphology. STARS has been shown to activate SRF and stabilize the F-actin cytoskeleton in a RhoA dependent manner, with the carboxy terminal being sufficient and necessary to activate SRF and bind actin 
. The pedestal phenotype observed in ABRA knockdown cells is consistent with ABRA stabilizing the F-actin cytoskeleton in this context (). Loss of this stabilization function in microcolonies leads to the dissolution of discreet pedestals and results in a structure more similar to a ruffle.
Under our experimental conditions overexpression of SRF also resulted in the nuclear accumulation of MAL-GFP. The specific reason for this is currently unclear. Currently the prevailing hypothesis states that MAL continually shuttles between the cytoplasm and the nucleus. Perhaps MAL has a higher binding affinity for SRF than G-actin, and upon entering the nucleus, preferentially complexes with SRF and is retained in the nucleus.
Transcription of SRF is controlled by SRF its self 
, the overexpression of SRF may be interpreted by the cell as activation of the pathway, leading to an upregulation of SRF target genes and subsequent decrease in G-actin. These are just two potential hypotheses that may not be mutually exclusive, but warrant further study.
Of the proteins identified in this study ABRA and FLRT3 localized to the EPEC pedestal (), and were both necessary for EPEC-induced translocation of MAL-GFP (). Epistasis analysis showed that knockdown of FLRT3 could significantly reduce ABRA-induced nuclear accumulation of MAL-GFP, but ABRA knockdown had no effect on FLRT3-induced nuclear accumulation of MAL-GFP ( and S4
). This places FLRT3 downstream of ABRA and identifies it as an intermediary protein from pedestal to nucleus. Based on this data we hypothesize a new model (), where EPEC-induced remodeling of the actin cytoskeleton, via Tir, activates SRF in an ABRA and FLRT3 dependent manner. Our findings therefore reveal a novel mechanism for pathogen-induced activation of a host transcription factor. They shed light on the relationship between ABRA and SRF and identify FLRT3 as a new component of this signalling pathway.
A model for EPEC-induced activation of host-cell transcription factor SRF.