Identification of Shigella flexneri IcsA Residues Affecting Interaction with N-WASP, and Evidence for IcsA-IcsA Co-Operative Interaction

The Shigella flexneri IcsA (VirG) protein is a polarly distributed outer membrane protein that is a fundamental virulence factor which interacts with neural Wiskott-Aldrich syndrome protein (N-WASP). The activated N-WASP then activates the Arp2/3 complex which initiates de novo actin nucleation and polymerisation to form F-actin comet tails and allows bacterial cell-to-cell spreading. In a previous study, IcsA was found to have three N-WASP interacting regions (IRs): IR I (aa 185–312), IR II (aa 330–382) and IR III (aa 508–730). The aim of this study was to more clearly define N-WASP interacting regions II and III by site-directed mutagenesis of specific amino acids. Mutant IcsA proteins were expressed in both smooth lipopolysaccharide (S-LPS) and rough LPS (R-LPS) S. flexneri strains and characterised for IcsA production level, N-WASP recruitment and F-actin comet tail formation. We have successfully identified new amino acids involved in N-WASP recruitment within different N-WASP interacting regions, and report for the first time using co-expression of mutant IcsA proteins, that N-WASP activation involves interactions with different regions on different IcsA molecules as shown by Arp3 recruitment. In addition, our findings suggest that autochaperone (AC) mutant protein production was not rescued by another AC region provided in trans, differing to that reported for two other autotransporters, PrtS and BrkA autotransporters.


Introduction
Shigella flexneri is a human pathogen that causes bacillary dysentery by infecting and colonising the colonic epithelium [1]. IcsA (VirG), one of the key virulence factors of S. flexneri, is polarly distributed at the outer membrane (OM). IcsA is required for inter-and intracellular spreading of S. flexneri within the host intestinal epithelium [2,3,4,5,6]. Inside the host cytoplasm, Shigella multiplies and interacts with the host actin regulatory protein neural Wiskott-Aldrich syndrome protein (N-WASP), which in turn activates the Arp2/3 complex that initiates actin polymerisation by recruiting the globular actin monomers to form the filamentous actin (F-actin) comet tails [7,8,9]. The formation of Factin comet tails allows bacterial actin-based motility (ABM) [2,7,8].
IcsA is a member of the autotransporter (AT) family (Type Va secretion system), the largest family of secreted proteins in Gramnegative bacteria [10,11]. Like other family members, IcsA consists of three major domains: an extended N-terminal signal sequence (amino acids [aa] 1-52), a functional passenger adomain (aa 53-758), and a C-terminal translocation b-domain (aa 759-1102) that mediates the translocation of the passenger domain across the OM via the BAM (beta-barrel assembly machine) complex [11,12,13,14,15]. The b-barrel domain is anchored in the OM, while the functional IcsA passenger domain is exposed on the bacterial surface and is responsible for N-WASP recruitment and ABM [3,8].
In spite of the diversity in sequence, function and length, the passenger domains of most ATs possess a b-helical structure [16]. IcsA was recently classified to a subgroup of self associating ATs (SAATs) that mediate bacterial aggregation and biofilm formation [17,18]. The ability of IcsA to self-associate has recently been shown via reciprocal co-precipitation of differentially epitopetagged IcsA proteins [19]. A putative autochaperone (AC) region at the C-terminal of IcsA passenger domain (aa 634-735), which forms part of the SAAT domain, is required for IcsA biogenesis [18,20]. To date, only the crystal structure of the IcsA AC region (aa 591-758) is available, and it possesses two coils of a righthanded parallel b-helix [21].
N-WASP is a key regulator of the actin cytoskeleton and functions as a link between signalling pathways and de novo actin polymerisation, leading to host cell motility and morphological changes [8,22,23,24,25]. IcsA 103-433 which contains glycine-rich repeats (GRRs; aa 117-307) is sufficient for N-WASP binding in vitro [24,26], and in vitro assays showed that IcsA 53-508 is adequate to induce actin polymerisation [8]. Five amino acids (5 aa) linker-insertion mutations (IcsA i ) [20] confirmed the involvement of aa 185-312 (N-WASP interacting region [IR] I), and a second region (aa 330-382 [N-WASP IR II]) in N-WASP recruitment. N-WASP IR II overlapped with the S. flexneri IcsB and host autophagy protein, Atg5, binding region (aa 320-433) [27]. In addition, a new IcsA region (aa 508-730 [N-WASP IR III]) that contains the IcsA AC region was also suggested to be important for N-WASP recruitment and F-actin comet tail formation [20]. However, specific N-WASP binding/interaction site or residues within IcsA have not yet been reported.
From the 5 aa linker-insertion mutagenesis study, IcsA i mutants (IcsA i633 , IcsA i643 , IcsA i677 , and IcsA i716 ) with mutational alterations within the AC region have markedly reduced protein production in S. flexneri strain with native LPS O-antigen (LPS Oag) (smooth LPS, S-LPS) and hence, were unable to form plaques, recruit N-WASP or form F-actin comet tail [20]. When the same IcsA i mutants were expressed in S. flexneri strain without LPS Oag (rough LPS, R-LPS), IcsA i mutant production was restored to a level comparable to wild-type (WT) IcsA [28]. With the restoration of IcsA i production in S. flexneri R-LPS strains, we have identified that IcsA i633 , IcsA i643 , IcsA i677 , but not IcsA i716 , were able to recruit N-WASP and form F-actin comet tails [28].
In this study, we investigated the ability of IcsA mutants with point mutations within N-WASP IR II and III to recruit N-WASP and form F-actin comet tails in both S-LPS and R-LPS S. flexneri backgrounds. Using this approach, we have identified specific residues affecting N-WASP binding/interaction. In addition, we have also co-expressed IcsA mutant proteins with mutations at two N-WASP interaction regions, and showed for the first time that all three N-WASP interacting regions are essential for N-WASP and Arp3 recruitment, supporting a role for oligomeric IcsA in this interaction.

Ethics Statement
The anti-IcsA antiserum was produced under the National Health and Medical Research Council (NHMRC) Australian Code of Practice for the Care and Use of Animals for Scientific Purposes and was approved by the University of Adelaide Animal Ethics Committee.

Bacterial Strains and Plasmids
The strains and plasmids used in this study are listed in Table 1.

Growth Media and Growth Conditions
All strains used in this study were routinely grown in Luria Bertani (LB) medium. S. flexneri strains were grown from a Congo Red positive colony as previously described [29]. Bacterial cultures were cultured for 18 h, diluted 1:20 and grown to mid-exponential phase (2 h) with aeration at 37uC. Where appropriate, antibiotics were used at the following concentrations: ampicillin (Ap, 100 mg mL 21 ), chloramphenicol (Cm, 25 mg mL 21 ), kanamycin (Km, 50 mg mL 21 ), or tetracycline (Tet, 10 mg mL 21 ).

DNA Methods
E. coli K-12 DH5a was used for all cloning. DNA manipulation, PCR, transformation and electroporation into S. flexneri were performed as previously described [30,31].

Antibodies and Antisera
Affinity purified rabbit polyclonal antibody to IcsA was produced as previously described [32]. The antiserum was produced under the National Health and Medical Research Council (NHMRC) Australian Code of Practice for the Care and Use of Animals for Scientific Purposes and was approved by the University of Adelaide Animal Ethics Committee. The anti-IcsA antibody was used at 1:100 in immunofluorescence (IF) assay or 1:1000 in Western immunoblotting (WB). The rabbit polyclonal anti-N-WASP [20] and monoclonal anti-Arp3 (BD Biosciences) antibodies were both used at 1:100 in IF.

Preparation of Whole Cell Lysate
The equivalent of 5610 8 bacteria were centrifuged (2,200 xg, 1 min) and resuspended in 100 ml of 1x sample buffer [33]. Samples were heated to 100uC for 5 min prior to electrophoresis.

Western Transfer and Detection
Western immunoblotting was performed as described previously [20] with some modifications. Briefly, proteins were separated on sodium dodecyl sulfate-7.5% or 12%-polyacrylamide gel electrophoresis (SDS-PAGE) gels and transferred to a nitrocellulose membrane. The membrane was blocked with 5% (w/v) skim milk in TTBS (Tris-buffered saline, 0.005% (v/v) Tween-20) for 20 min and incubated with anti-IcsA antibody in TTBS containing 2.5% (w/v) skim milk for 18 h. After three washes in TTBS, the membrane was incubated with horseradish peroxidise-conjugated goat anti-rabbit secondary antibody (Biomediq DPC) for 2 h, followed by three washes in TTBS, then three times in TBS. The membrane was incubated with CPS3 chemiluminescence substrate (Sigma) for 5 min, followed by exposure of the membrane to either X-ray film (Agfa) or imaged with a Kodak Image Station 4000 MM Pro (Carestream Molecular Imaging), to visualise the reactive bands. The film was developed using a Curix 60 automatic X-ray film processor (Agfa).

Site-directed Mutagenesis
Single and multi site-directed mutagenesis were performed according to the QuikChangeH Lightning Site-directed or Quik-ChangeH Lightning Multi Site-directed protocol (Stratagene), respectively. Specific primers with the desired substitutions were designed for single site-directed mutagenesis (Table S1) and degenerate primers were designed for multi site-directed mutagenesis (Table S1). pMG55 (pIcsA::BIO) was used as the DNA template for the construction of the mutants. IcsA::BIO protein possesses a BIO epitope (GLNDIFEAQKIEWH; a substrate for metabolic biotinylation by the BirA biotin-protein ligase [34]) introduced at aa 87 which does not affect the IcsA production levels, polar localisation and function [19]. Briefly, the resultant PCR products were treated with DpnI and transformed into XL 10-Gold ultracompetent cells (Stratagene). Mutations were confirmed by DNA sequencing.

Plaque Assays
Plaque assays were performed as described previously [20] using the method of Oaks et al. [35]. Briefly, HeLa cells (Human, cervical, epithelial cells ATCC #CCL-70) were grown to confluence overnight in minimal essential medium (MEM)-10% (v/v) foetal calf serum (FCS), washed twice with Dulbecco's PBS (D-PBS) and once in Dulbecco's modified Eagle medium (DMEM) prior to infection. 1610 8 bacteria mL 21 mid-exponential phase bacteria were diluted to 1:1000 in DMEM, and 0.25 mL was added to each well. Trays were incubated in a CO 2 (5%) incubator, and the trays were gently rocked every 15 min. At

Indirect Immunofluorescence of Whole Bacteria
Indirect IF labelling of bacteria was performed as previously described [20]. Briefly, mid-exponential phase bacteria were fixed in formalin (3.7% (v/v) paraformaldehyde in 0.85% (w/v) saline) and centrifuged onto poly-L-lysine-coated coverslips. Bacteria were incubated with anti-IcsA antibody, washed with PBS and labelled with Alexa 488-conjugated donkey anti-rabbit secondary antibody (Molecular Probes) (1:100). Microscopy was performed as previously described [20], using an Olympus IX-70 microscope with phase-contrast optics using a 100x oil immersion objective. Fluorescence and phase-contrast images were false colour merged using the Metamorph software program (Version 7.7.3.0, Molecular Devices).

Infection of Tissue Culture Monolayers with S. Flexneri and IF Labelling
Infection of HeLa cell monolayers and IF staining were performed as previously described [20]. HeLa monolayers were inoculated with mid-exponential phase bacteria and incubated for 1 h at 37uC, CO 2 (5%). The infected monolayers were washed three times with D-PBS and incubated with MEM containing gentamycin for a further 1.5 h. Infected cells were then washed and fixed for 15 min in 3.7% (v/v) formalin, incubated with 50 mM NH 4 Cl in D-PBS for 10 min, and permeabilised with 0.1% (v/v) Triton X-100 for 5 min. The infected cells were blocked with 10% (v/v) FCS in PBS and incubated with the desired primary antibody at 37uC for 30 min. After washing in PBS, coverslips were incubated with Alexa Fluor 594-conjugated donkey anti-rabbit, Alexa Fluor 488-conjugated donkey antirabbit or Alexa Fluor 594-conjugated donkey anti-mouse secondary antibodies (Molecular Probes) (1:100), where appropriate. Factin was visualised by staining with Alexa Fluor 488-conjugated phalloidin (2 U mL 21 ), and 49,69-diamidino-2-phenylindole (DAPI) (10 mg mL 21 ) was used to counterstain bacteria and cellular nuclei.

N-WASP Interacting Region II
May & Morona [20] previously identified a region in IcsA (aa 330-382) that was involved in N-WASP recruitment, based on linker-insertion mutagenesis. Protein production of IcsA i mutants with insertion mutations within this region was unaffected but the mutants were defective in ABM when expressed in S. flexneri DicsA S-LPS [20]. This region was classified as N-WASP IR II. Since the 5 aa linker insertion may disrupt local protein structure conformation (IcsA i330 and IcsA i381 ), it was of interest to determine if altering specific residues would reproduce the same phenotype, thus, enabling the identification of specific amino acids involved in N-WASP recruitment within this region. Multi site-directed mutagenesis was performed on pIcsA::BIO (pMG55) to randomly substitute both amino acids, 330-331 and 381-382, respectively, by using degenerate primers listed in Table S1. Mutants were isolated, DNA sequenced, and listed in Tables 2 and 3. IcsA::BIO T330*G331* and IcsA::BIO T381*V382* mutant production in S. flexneri DicsA was assessed by Western immunoblotting with anti-IcsA antibody (data not shown). All of these IcsA mutant proteins were produced by S. flexneri at WT levels (Tables 2 and 3).
The ability of S. flexneri DicsA expressing IcsA::BIO T330*G331* or IcsA::BIO T381*V382* mutants to recruit N-WASP and form F-actin comet tail was investigated by IF microscopy using anti-N-WASP antibody and Alexa Fluor 488-phalloidin, respectively (data not shown). The results are summarised in Tables 2 and 3. T330 and G331 are both polar amino acids, and IcsA::BIO T330*G331* mutants that have at least one non-polar amino acid (proline) substitution at either position (e.g. encoded by pMYRM539 and pMYRM543) did not recruit N-WASP (Table 2). Notably, mutants encoded by pMYRM536 [IcsA::BIO T330S G331G] and pMYRM539 [IcsA::BIO T330P G331G] both possess single mutations at residue 330 (in bold) but only mutant T330S G331G (both polar amino acids) was capable of recruiting N-WASP. However, it cannot be ruled out that the proline substitution at position 330 or 331 is causing a localised conformation change thereby affecting IcsA function. Furthermore, mutants with charged amino acids (encoded by pMYRM537, pMYRM538 and pMYRM556) recruited less N-WASP and formed less F-actin capping/tails ( Table 2), suggesting that polar residues are required for efficient N-WASP recruitment.
In contrast, T381 and V382 are polar and non-polar amino acids, respectively. Based on the IF microscopy results (Table 3), IcsA::BIO T381*V382* mutants with residue 382 substituted with either charged or polar amino acids (e.g. encoded by pMYRM544, pMYRM545, pMYRM546 and pMYRM549) were unable to recruit N-WASP, whilst mutants that had V382 substituted with another non-polar amino acid (e.g. encoded by pMYRM547 and pMYRM548) recruited N-WASP, regardless of the amino acid polarity at residue 381. The data suggest that a non-polar amino acid is required at residue 382 in order for IcsA to be capable of recruiting N-WASP.
IcsA::BIO T330*G331* mutants that recruited N-WASP were selected to examine their ability to perform intercellular spreading by plaque assay (Fig. S1). As expected, T330*G331* that recruited N-WASP formed plaques but with different plaque sizes, depending on the amino acids substitution. Similar results were obtained for IcsA::BIO T381*V382* mutants wherein mutants that had non-polar amino acids substituted at residue 382, and also recruited N-WASP, formed plaques. In all cases, there was a consistent correlation between N-WASP recruitment, F-actin comet tail/capping formation and plaque formation. Taken together, the data suggest that both residues 330-331 of IcsA need to be polar while residue 382 needs to be non-polar in order for IcsA to be able to recruit N-WASP.

Effect of the G331W Mutation on N-WASP Recruitment and Intercellular Spreading
We hypothesised that residues 330-331 require the polar-polar combination to recruit N-WASP. Thus, site-directed mutagenesis was performed on pIcsA::BIO using primer pairs listed in Table  S1 to mutate G331 to tryptophan (W), which is a non-polar, aromatic amino acid. IcsA::BIO G331W (encoded by pMYRM588) was expressed in S. flexneri DicsA, and protein production was examined by Western immunoblotting (Fig. 1A, lane 2) and IF microscopy (Fig. S2). IcsA::BIO G331W was expressed on the bacterial surface and was produced at a WT comparable levels. As seen by IF microscopy, IcsA::BIO G331W mutant was capable of recruiting N-WASP and formed F-actin comet tails but seemed less frequent than the functionally equivalent IcsA::BIO ( Fig. 1B(b)). Its ability to promote intercellular spreading was also tested by plaque assay. Plaques formed by IcsA::BIO G331W were half the size (***P,0.001) of those formed by S. flexneri expressing IcsA::BIO (Fig. 1C). Collectively, the data suggest that the G331W mutation affected intercellular spreading of S. flexneri, presumably by affecting N-WASP recruitment efficiency. In addition, the data also indicate that a polar-polar combination for residues 330-331 is not essential for N-WASP recruitment.

Effect of the V382R Mutation on N-WASP Recruitment and Intercellular Spreading
We hypothesised that it is essential for IcsA residue 382 to be a non-polar amino acid for N-WASP recruitment. Site-directed mutagenesis was performed on pIcsA::BIO to mutate V382 to arginine (R) that is positively charged. IcsA::BIO V382R (encoded by pMYRM590) mutant production by S. flexneri DicsA was assessed by Western immunoblotting (Fig. 1A, lane 3) and IF microscopy (Fig. S2). IcsA::BIO V382R was expressed on the bacterial surface and produced at a WT comparable levels. Its ability to recruit N-WASP and form F-actin comet tails within HeLa cells was examined. IF microscopy showed that the V382R mutant was defective in N-WASP recruitment and F-actin comet tail formation (Fig. 1B(c)). Consequently, S. flexneri expressing the IcsA::BIO V382R mutant was unable to form plaques (Fig. 1C).
These data suggest that a non-polar amino acid (aa 382) is a prerequisite for N-WASP recruitment and residue 382 is likely to be an N-WASP binding/interaction site.

N-WASP Interacting Region III
We have previously characterised several IcsA i mutants which had 5 aa insertion mutations within the IcsA AC region for their ability to recruit N-WASP, and identified that IcsA i716 was defective in N-WASP recruitment [28]. These IcsA i mutants had extremely low protein production in the S-LPS background but WT comparable expression levels in the R-LPS background [28]. The IcsA i716 mutant has a 5 aa insertion mutation after residue 716 which affects the Aro-X-Aro (aromatic residue-any-aromatic residue) motif, Y 716 -D 717 -Y 718 . This motif is commonly found in the b-domain of ATs and is thought to be a preferential binding site of SurA [36,37]. Hence, it was of interest to determine if altering residues 716 and 717 would reproduce the effect of the insertion mutation that may have an effect on local protein conformation, in terms of IcsA biogenesis and N-WASP recruitment ability. The recently identified IcsA AC crystal structure (Fig.  S3) shows that the predicted locations for residues 716-717 appeared to be surface exposed, while residue 718 is predicted to be buried. Hence, multi site-directed mutagenesis was performed on pIcsA::BIO to substitute both Y716 and D717 with random amino acids at these positions. Seven pIcsA::BIO Y716*D717* mutants were isolated, DNA sequenced, and are listed in Table 4. The production of IcsA::BIO Y716*D717* mutants in both S. flexneri DicsA S-LPS and S. flexneri DicsA DrmlD R-LPS strains was determined by Western immunoblotting. The IcsA::BIO Y716*D717* mutant production was markedly reduced in the S-LPS strain, regardless of the amino acids substitutions ( Fig. 2A). As expected, similar to the IcsA i mutants [28], IcsA::BIO Y716*D717* mutant production in the R-LPS strain was restored to a level that was comparable to IcsA::BIO ( Fig. 2A).
We then investigated the effect of individual residues (Y716 and D717) by performing site-directed mutagenesis on pIcsA::BIO to mutate Y716 to phenylalanine (F) or glycine (G), and D717 to glycine (G) using primer pairs shown in Table S1. Plasmid constructs pIcsA::BIO Y716F (pMYRM592), pIcsA::BIO Y716G (pMYRM595) and pIcsA::BIO D717G (pMYRM608) were transformed into both S. flexneri DicsA S-LPS and S. flexneri DicsA DrmlD R-LPS strains and protein production was examined by Western immunoblotting (Fig. 2B) and IF microscopy (Fig. S4). The Y716F mutation did not affect IcsA production (Fig. 2B, lanes 2 and 6) but the Y716G mutation (Fig. 2B, lanes 3 and 7) resulted in reduced IcsA production, in both S-LPS and R-LPS backgrounds. Similar to other IcsA mutants with mutations within the AC region [28], the D717G mutation greatly reduced IcsA production in S. flexneri DicsA S-LPS (Fig. 2C, lane 2) but the production was restored in S. flexneri DicsA DrmlD R-LPS, albeit slightly lower than IcsA::BIO (Fig. 2C, lane 5). Hence, whereas aa 717 is essential for IcsA biogenesis, the aromatic aa 716 is also important but less dependent on LPS structure.

Effect of Y716 and D717 Mutagenesis on N-WASP Recruitment and F-actin Comet Tail Formation
We have previously shown that IcsA i716 was defective in N-WASP recruitment and proposed that residues 716-717 could be an N-WASP binding/interaction site [28]. To clarify this hypothesis, the IcsA::BIO Y716*D717* mutants, IcsA::BIO Y716F, IcsA::BIO Y716G and IcsA::BIO D717G mutants were assessed for their ability to recruit N-WASP and form F-actin comet tail by IF microscopy (in both S-LPS and R-LPS backgrounds). The results are summarised in Table 4. For the    IcsA::BIO Y716*D717* mutants, regardless of the amino acids substitutions, none of the mutants recruited N-WASP in HeLa cells (in both S-LPS and R-LPS backgrounds), despite the higher protein expression levels in the R-LPS background ( Fig. 2A), which suggest that residues 716-717 are an N-WASP binding/ interaction site. Investigation of individual residues, Y716 and D717, on N-WASP recruitment showed that S. flexneri DicsA S-LPS strain expressing IcsA::BIO Y716F recruited N-WASP and formed Factin comet tails (Fig. 3A(c)), while the S. flexneri DicsA S-LPS strain expressing IcsA::BIO Y716G recruited N-WASP (slightly less than IcsA::BIO) but had only F-actin capping formation and F-actin comet tails were not observed (Fig. 3A(e)). However, both Y716F and Y716G mutants expressed by the S. flexneri DicsA DrmlD R-LPS strains recruited N-WASP and formed F-actin comet tails ( Fig. 3A(d,f)), suggesting that the aromatic property of residue 716 is not essential for N-WASP binding/interaction. As expected, S. flexneri DicsA S-LPS expressing IcsA::BIO D717G did not recruit any N-WASP or form F-actin comet tail, likely due to the very low protein level (Fig. 3A(g)). In contrast, despite the higher level of protein production in the R-LPS background, N-WASP staining was barely detectable and F-actin capping was detected on a minor population of S. flexneri DicsA DrmlD expressing IcsA::BIO D717G (,15%) ( Fig. 3A(h)). Collectively, the data suggest that residue D717 has a greater impact on N-WASP recruitment in comparison to Y716.

Co-expression of Mutated IcsA Proteins in S. flexneri
Our results and those reported by May & Morona [20] showed that N-WASP recruitment was partly affected or completely abolished when IcsA protein with mutation within any of the N-WASP IRs (I, II or III) was expressed by S. flexneri (summary in Table 6. N-WASP recruitment by S. flexneri R-LPS strains co-expressing various IcsA proteins.   Table S2). We have also recently found that IcsA is able to selfassociate [19]. Therefore, we hypothesised that all three N-WASP interacting regions are required for N-WASP binding/activation and co-expression of two IcsA proteins with mutations at different N-WASP interacting regions would be able to complement the N-WASP binding ability and ultimately, restore N-WASP recruitment and activation. To co-express two mutated IcsA proteins, relevant icsA genes were cloned into vector pSU23 which has a similar copy number (intermediate) to pBR322. IcsA mutants with mutation at each N-WASP IR and defective in N-WASP recruitment when expressed individually in S. flexneri were chosen for co-expression/complementation analysis. pSU23 constructs that encode for IcsA WT , IcsA::BIO, IcsA i248 (N-WASP IR I) and IcsA::BIO Y716G D717G (N-WASP IR III) were constructed as detailed in Materials and Methods. As expected, all IcsA proteins were expressed at WT levels that were comparable to those expressed by the pBR322 constructs in both S-LPS and R-LPS backgrounds (Fig. S5A), except IcsA::BIO Y716G D717G (N-WASP IR III) that had extremely low protein production in the S-LPS background (Fig. S5A, lane 7), which was similar to that observed for the pBR322 based clone ( Fig. 2A). Similar results were obtained for IF microscopy where surface IcsA expression was detected at WT levels, except for IcsA::BIO Y716G D717G in the S-LPS background, as expected (Fig. S5B).
To co-express IcsA proteins, various pBR322-IcsA constructs that encode for IcsA WT , IcsA::BIO, IcsA i248 (N-WASP IR I), IcsA::BIO V382R (N-WASP IR II) or IcsA::BIO Y716G D717G (N-WASP IR III) were subsequently electroporated into different S. flexneri strains (S-LPS and R-LPS) carrying the pSU23-IcsA constructs as shown in Tables 5 and 6. S. flexneri S-LPS strains that co-expressed IcsA proteins were assayed for N-WASP recruitment following infection of HeLa cells by IF microscopy, and the results are summarised in Table 5. S. flexneri S-LPS strains that expressed either IcsA WT or the functionally equivalent IcsA::BIO were able to recruit N-WASP, regardless of the co-expressed IcsA protein. However, co-expression of two mutated IcsA proteins (N-WASP IR I+II, II+III or I+III) did not result in N-WASP recruitment. As expected, S. flexneri strains that co-expressed a mutated IcsA protein with an empty vector did not recruit N-WASP. As the production of IcsA::BIO Y716G D717G was extremely low in the S-LPS background but restored to a WT comparable levels in the R-LPS background, various IcsA proteins were also co-expressed in S. flexneri R-LPS. Their ability to recruit N-WASP was examined and summarised in Table 6. The same outcomes were obtained for the S. flexneri R-LPS strains which have the same IcsA co-expression combination in the S-LPS background, with the remarkable exception of the co-expressed combination IcsA::BIO V382R and IcsA::BIO Y716G D717G (N-WASP IR II and III, respectively) that was able to recruit N-WASP (Table 6). To investigate if the N-WASP recruited by S. flexneri R-LPS coexpressing IcsA::BIO V382R and IcsA::BIO Y716G D717G (IR II+III) (MYRM697) was activated, Arp2/3 complex recruitment and F-actin comet tail formation were examined by IF microscopy with anti-Arp3 monoclonal antibody and Alexa Fluor 488phalloidin staining, respectively. Arp3 recruitment was significantly detected at one pole of the bacterium which co-localised with N-WASP ( Fig. 4A(b)) and F-actin comet tail/capping was observed ( Fig. 4B(b)). The data indicate that the complementation of N-WASP interacting regions II and III allowed S. flexneri MYRM697 to activate N-WASP, recruit Arp2/3 complex and form F-actin comet tails. However, MYRM697 displayed reduced N-WASP recruitment frequency and F-actin comet tail/capping formation compared to MYRM704 (expressing the functionally equivalent IcsA::BIO). We have quantitated the recruitment of N-WASP, Arp3 and F-actin comet tail/capping formation by scoring 20 infected HeLa cells (,250-350 bacteria) when both N-WASP/ Arp3 and N-WASP/F-actin tail/capping were observed respectively, in two independent experiments. MYRM697 which coexpressed IcsA::BIO V382R and IcsA::BIO Y716G D717G had an approximately 40-45% reduction in N-WASP (P***,0.001) and Arp3 recruitment (0.001,P**,0.01) (Fig. 4C), as well as Factin comet tail/capping formation (P***,0.001) (Fig. 4D), compared to MYRM704, the positive control strain. Nevertheless, the data shows significant co-operative interaction between N-WASP IRs on different IcsA molecules in N-WASP activation. However, we do not rule out that some degree of co-operativity between N-WASP IRs within the same IcsA molecule may also be occurring in this strain.

Discussion
IcsA (VirG) is one of the key virulence proteins of S. flexneri, and is able to initiate F-actin comet tail formation through activation of the host N-WASP, leading to bacterial ABM and spreading within the host colonic epithelium. Although N-WASP is known to interact with IcsA and several IcsA regions have been reported to be involved in N-WASP recruitment [20], specific N-WASP interacting residues have yet to be identified. The aim of this study was to determine the N-WASP binding/interaction site(s) on IcsA (N-WASP interacting regions II and III) based on previously identified linker insertion mutations [20]. However, 5 aa linker insertion mutations may have caused local disruption to the IcsA secondary structure, complicating the identification of important amino acids residues involved in N-WASP interaction. Hence, site-directed mutagenesis of specific residues at these sites was undertaken in this study. Interaction between IcsA and N-WASP Specific amino acids within the N-WASP IR II were randomly mutated (aa 330-331 and aa 381-382, respectively). Screening of the resultant IcsA mutants suggested that for efficient N-WASP recruitment, a polar-polar combination is required for residues 330-331, while a non-polar amino acid is required for residue 382 (Tables 2 and 3). Hence, site-directed mutagenesis was subsequently performed on pIcsA::BIO to mutate residues G331 and V382 into non-polar tryptophan (W) and polar arginine (R), respectively. S. flexneri DicsA expressing IcsA::BIO G331W was capable of recruiting N-WASP and forming F-actin comet tail/ capping but not as frequently as the positive control strain (IcsA::BIO). As a result, a reduction in plaque size (,50%) was observed for IcsA::BIO G331W. In comparison, a single amino acid change at position 330 into a proline which is a non-polar amino acid (pMYRM539, T330P), abolished N-WASP recruit-ment. Notably, pMYRM543 (T330G G331P) which possesses proline at position 331 was also defective in N-WASP recruitment. Although both pMYRM543 (IcsA::BIO T330G G331P) and IcsA::BIO G331W mutants have the polar and non-polar amino acids combination, a contradictory effect on function was obtained. Taken together, the results obtained suggest that the proline substitution at either position disrupts local secondary structure of IcsA and prevents N-WASP binding. Nevertheless, the results suggest that N-WASP interaction at this site does not rely on amino acid polarity, and the polar-polar combination is not essential. Indeed, the data suggest that the presence of specific amino acid residues at positions 330-331 is required for efficient N-WASP interaction and recruitment.
Consistent with the IcsA::BIO T381*V382* mutants that also have residue 382 substituted with polar arginine (R) [pMYRM544   (Table 3), which possess a nonpolar amino acid at residue 382 (in bold), were able to recruit N-WASP and form plaques when expressed in S. flexneri, regardless of the amino acid substitution at residue 381. The data suggest that a non-polar residue is essential at position 382 and plays an important role in N-WASP recruitment. Collectively, site-directed mutagenesis within the N-WASP IR II revealed new N-WASP binding/interaction sites, and we have assigned them as N-WASP IR 2a (aa 330-331) and N-WASP IR 2b (aa 382), respectively.
We have previously shown that IcsA i716 which had a 5 aa insertion mutation within the AC region (N-WASP IR III) was defective in N-WASP recruitment [28]. The predicted location for residues Y716-D717 (forming part of the Aro-X-Aro motif) is at the beginning of the first anti-parallel b-strand that appears to be exposed to the external medium (Fig. S3), suggesting that this motif could be involved in protein binding. Hence, IcsA::BIO Y716*D717*, IcsA::BIO Y716F, IcsA::BIO Y716G and IcsA::BIO D717G mutants were created, expressed in both S. flexneri S-LPS and R-LPS backgrounds, and investigated for N-WASP recruitment. The IcsA::BIO Y716*D717* mutant protein production was markedly reduced in the S-LPS background but restored to a WT comparable levels in the R-LPS background ( Fig. 2A), which is consistent with the IcsA i716 mutant phenotype [28]. In spite of the WT comparable levels of IcsA::BIO Y716*D717* mutant expression on the R-LPS bacterial surface, N-WASP recruitment was not observed ( Table 4). The data suggested that this Aro-X-Aro motif is an N-WASP binding/interaction site, in this region of the protein.
Our results showed that IcsA::BIO Y716F (conserved mutation) and IcsA::BIO Y716G (non-conserved mutation) were functional in both S-LPS and R-LPS backgrounds, except that only F-actin capping was observed for Y716G in the S-LPS background. This may be due to the lower levels of IcsA::BIO Y716G expression in the S-LPS background which affected N-WASP recruitment efficiency and/or activation, hence, resulting in F-actin capping formation only. Consistent with these results, S. flexneri S-LPS expressing IcsA::BIO Y716F formed plaques that were comparable with WT while IcsA::BIO Y716G formed smaller plaques (Fig. 3B), reflecting the slightly reduced N-WASP recruitment efficiency and F-actin capping formation by IcsA::BIO Y716G. Together, the data suggest that the aromatic property of residue 716 is not essential for N-WASP recruitment.
IcsA::BIO D717G did not recruit N-WASP in the S-LPS background ( Fig. 3A(g)), which is due to the extremely low IcsA production (Fig. 2C). Consequently, plaques were not observed. In contrast, the restoration of IcsA::BIO D717G expression in the R-LPS strain (Fig. 2C) did not lead to WT comparable N-WASP recruitment efficiency (Fig. 3A(h); Table 4), implying that D717 is involved in N-WASP binding/interaction. Collectively, the data obtained suggest that D717 plays a more important role than Y716 in N-WASP recruitment. Nevertheless, both residues Y716-D717 are important for N-WASP recruitment as mutation of both residues (IcsA::BIO Y716*D717* mutants) abolished N-WASP recruitment completely (Table 4).
In terms of IcsA biogenesis, we showed that the 716 aromatic residue is important for IcsA production but this was largely independent of LPS structure because the non-conservative mutation resulted in reduced IcsA::BIO Y716G protein production in both S-LPS and R-LPS backgrounds (Fig. 2B). This implies that the aromatic property of residue 716 is important for protein folding/stability. In contrast, D717 is a critical residue within the AC region for IcsA biogenesis but in a LPS structure dependent manner, as the IcsA::BIO D717G mutant had markedly reduced IcsA mutant production in the S-LPS background but was mostly restored in the R-LPS background (Fig. 2C, lanes 2 and 5). These results are consistent with our findings regarding the IcsA i AC mutant production [28].
Previous investigations on IcsA regions involved in N-WASP recruitment have been studied on individual N-WASP interacting regions (based on deletion and linker insertion mutations) [20,24,26] and the identification of multiple N-WASP interacting regions on IcsA led us to investigate the mode of action of IcsA in recruiting and activating N-WASP. IcsA is part of the SAATs and has recently been shown to self-associate and oligomerise at the OM [18,19]. On the other hand, allosterically activated N-WASP is able to dimerise/oligomerise and recruit the Arp2/3 complex [38,39]. May et al. [19] also suggested that self-association of IcsA may play a role in N-WASP clustering, as S. flexneri S-LPS coexpressing IcsA WT and IcsA i563 or IcsA i677 displayed reduced efficiency in ABM. We hypothesised that IcsA functions as a complex on the bacterial surface and different regions of IcsA (N-WASP IR I, II and III) are involved in interacting with N-WASP. To investigate our hypothesis, we co-expressed two IcsA mutant proteins in both S-LPS and R-LPS backgrounds in different combinations, to complement one mutated N-WASP IR with another copy of IcsA that had mutation at another N-WASP IR (Tables 5 and 6). Similar IcsA mutant production was obtained when the IcsA mutants were individually expressed by either pBR322 or pSU23 in S-LPS and R-LPS S. flexneri. Hence, the expression levels of co-expressed IcsA proteins (either with functional IcsA or IcsA mutant) were expected to be similar as the compatible intermediate copy plasmids (pBR322 and pSU23) were used. In addition, May et al. [19] showed that the coexpression of functional IcsA with IcsA i mutant did not affect the polar distribution and the amount of functional protein at the bacterial surface.
Our findings showed that none of the IcsA mutants when coexpressed recruited N-WASP in the S-LPS background, except when co-expressed with IcsA WT or the functionally equivalent IcsA::BIO (Table 5). Interestingly, when IcsA with mutations within N-WASP IR II and III (IcsA::BIO V382R and IcsA::BIO Y716G D717G) were co-expressed in the R-LPS background, N-WASP was activated, resulting in Arp2/3 complex recruitment and formation of F-actin comet tail/capping, albeit at a frequency less than the control (,40-45% reduction) (Fig. 4). Co-expression of these IcsA mutant proteins in the S-LPS background did not show N-WASP recruitment which was likely due to the very low IcsA::BIO Y716G D717G production. IcsA::BIO V382R alone was not able to recruit N-WASP. Interestingly, this suggests that the AC region possessed by IcsA::BIO V382R does not seem to rescue the related expression defect of IcsA::BIO Y716G D717G. This is in contrast with the ability of the co-expressed AC region to rescue an AC region mutant in other systems (e.g. PrtS and BrkA ATs) [40,41].
Studies on other autotransporter AC regions (e.g. PrtS and BrkA) were performed on E. coli which is a R-LPS strain, and the additional AC region was provided in trans and located at the OM [40,41]. Our system is different from those previously reported whereby two full length mutated IcsA proteins were co-expressed in both S-LPS and R-LPS S. flexneri, and the AC IcsA mutant (e.g. IcsA::BIO Y716G D717G) production was severely affected in the S-LPS background. We have previously proposed that AC IcsA mutant could be degraded in the periplasm prior to translocation across the OM [28]. Hence, the failure to complement N-WASP recruitment by co-expressing IcsA::BIO V382R and IcsA::BIO Y716G D717G in the S-LPS background, could be due to the absence of IcsA::BIO Y716G D717G at the OM, as a result of degradation in the periplasm, supporting our previous hypothesis. Alternatively, LPS O-antigen could be masking the function of coexpressed IcsA::BIO V382R and IcsA::BIO Y716G D717G. However, it is unclear if IcsA::BIO Y716G D717G was coexpressed at high levels on the S-LPS S. flexneri surface. IcsA proteins with different tags could be co-expressed in S-LPS S. flexneri in future to determine the IcsA::BIO Y716G D717G expression level. In addition, we have previously shown that the expression of individual AC IcsA mutants in both R-LPS S. flexneri and R-LPS E. coli was restored to WT comparable levels, which does not require complementation of an additional AC region to restore IcsA biogenesis [28]. Therefore, it is unclear whether the AC region provided by IcsA::BIO V382R (at the OM) is able to restore IcsA biogenesis/folding of an AC IcsA mutant, as reported in PrtS and BrkA [40,41].
The N-WASP interacting region complementation data ( Table 6; Fig. 4) suggest that IcsA functions as a complex (at least as a dimer), with all three N-WASP interacting regions required for efficient N-WASP recruitment. Furthermore, N-WASP IR II and III on different IcsA molecules interact cooperatively with N-WASP to promote its activation. However, we do not rule out that N-WASP binds to all three N-WASP IRs of one IcsA WT molecule. The function of N-WASP IR I (contains the GRRs) cannot be complemented by another copy of IcsA with mutation at another N-WASP IR (Fig. 5). However, S. flexneri coexpressing both IcsA i248 (N-WASP IR I) and IcsA WT (or the functionally equivalent IcsA::BIO) was capable of recruiting N-WASP, suggesting that despite mutation in N-WASP IR I, the presence of another copy of functional IcsA that provides sufficient functional N-WASP IRs is able to complement the defect, and a negative dominance effect was not detected.
Taken together, we have identified new amino acids within N-WASP IR II and III that are involved in N-WASP recruitment. In particular, the conserved Aro-X-Aro motif (aa 716-718, N-WASP IR III) that is located within the AC region could be a potential drug target site, by utilising the recently crystallised IcsA AC region structure. Furthermore, our N-WASP IR complementation data strongly suggest that IcsA functions as a multimer. We have shown for the first time that all three N-WASP IRs are involved in N-WASP recruitment and activation. Our findings have shed light on N-WASP activation by S. flexneri IcsA protein where the nature of their interactions is poorly understood. However, it is still unclear which region of N-WASP (WHI domain, CRIB domain or both) [26,42] is involved in binding to the IcsA protein and whether N-WASP binds to three different sites of IcsA as a single molecule or as a complex. It is also unclear whether the binding of N-WASP to multiple sites of IcsA is a pre-requisite or if it merely enhances the allosteric activation of N-WASP and/or N-WASP oligomerisation which in turn activates the Arp2/3 complex. These issues warrant further investigations that would reveal the mystery of the complex interaction between the S. flexneri virulence IcsA protein and the host actin cytoskeleton regulatory N-WASP protein.