Protein Interactions and Regulation of EscA in Enterohemorrhagic E. coli

Infections caused by enterohemorrhagic Escherichia coli (EHEC) can lead to diarrhea with abdominal cramps and sometimes are complicated by severe hemolytic uremic syndrome. EHEC secretes effector proteins into host cells through a type III secretion system that is composed of proteins encoded by a chromosomal island, locus for the enterocyte effacement (LEE). EspA is the major component of the filamentous structure connecting the bacteria and the host's cells. Synthesis and secretion of EspA must be carefully controlled since the protein is prone to polymerize. CesAB, CesA2, and EscL have been identified as being able to interact with EspA. Furthermore, the intracellular level of EspA declines when cesAB, cesA2, and escL are individually deleted. Here, we report a LEE gene named l0033, which also affects the intracellular level of EspA. We renamed l0033 as escA since its counterpart in enteropathogenic E. coli has been recently described. Similar to CesAB, EscL, and CesA2, EscA interacts with EspA and enhances the protein stability of EspA. However, EscA is also able to interact with inner membrane-associated EscL, CesA2, and EscN, but not with cytoplasmic CesAB. In terms of gene organizations, escA locates in LEE3. Expression of EscA is faithfully regulated via Mpc, the first gene product of LEE3. Since Mpc is tightly regulated to low level, we suggest that EscA is highly synchronized and critical to the process of escorting EspA to its final destination.


Introduction
Enterohemorrhagic Escherichia coli (EHEC) O157:H7 [1,2] forms typical histological lesions termed attaching and effacing lesions on the large intestine tissue of infected individuals. This lesion-generating capability is attributed largely to a set of virulence genes clustering on the bacterial chromosome that is called the locus of enterocyte effacement (LEE) [3,4,5,6]. The LEE contains 41 open reading frames (Orfs) that are largely organized into five poly-cistronic operons. These Orfs have been studied in detail to a variety of different degrees [7] and the better characterized gene products have been categorized into a number of groups, namely components of a type III secretion system (TTSS) (Sep and Esc), secreted translocators and effector proteins (Esp), chaperones (Ces), regulatory factors (Ler, GrlA, GrlR, and Mpc), an adhesin (Eae) and a translocated receptor (Tir). Gene l0033 in EHEC is among the least well characterized genes and the equivalent genes in Citrobacter rodentium and enteropathogenic E. coli (EPEC) have been shown to play critical roles in the TTSS [8,9]. However, information on the mechanism of action of this gene still remains limited.
TTSSs exist in several Gram-negative bacteria that are either pathogenic or symbiotic [10]. The main function of any TTSS is to efficiently and specifically deliver effector proteins from the bacterial cytoplasm into the eukaryote target cell cytoplasm by a direct contact through a syringe-like conduit. In the case of EHEC, one phenotype after delivery of the effector proteins is an induction of rearrangement of the cellular actin by altering the signaling pathways of the infected cells [11]. These alterations result in the formation of pedestal structures. The syringe-like conduit formed is evolutionarily similar to that associated with flagella systems [12]. Ultra-structurally, the whole structure is organized into a needle complex and a basal body. The needle complex of EHEC consists of EspA in the filamentous portion with the tip being EspA in complex with EspB and EspD. On the other hand, the basal body structure consists of an inner membrane ring and an outer membrane ring [13]. In addition to the structural consideration described above, in order to assemble a mature and effective secretion apparatus, timing is highly critical. It is believed that the needle complex is secreted by the TTSS and is then assembled onto the basal body to produce the distal elements of the apparatus [14].
To better understand the mechanistic activity of the TTSS in EHEC, it is highly important to characterize the functions of the less well understood genes in these operons because no gene within the LEE island is dispensable if the TTSS is to be functional and effective [8]. Orf l0033 of EHEC consists of 378 base-pairs and this encodes a protein of 125 amino acids with a calculated molecular weight of 14.66 kDa. By protein sequence comparison, L0033 shares 100% identity with a counterpart in EPEC and 95% identity with that of a counterpart in C. rodentium [9]. Furthermore, there are no other homologues found outside the LEE family. Recent studies in EPEC [9] have confirmed previous findings for C. rodentium [8] whereby the counterparts of l0033 are essential for the secretion of both translocator and effector. Since the gene in these two strains has been renamed escA, we will from now on refer to l0033 as escA to allow easier discussion. Moreover, no functional and evolutionary origin for EscA has been suggested based on previous bioinformatics analysis of the LEE [15]. Using yeast twohybrid system analysis [16], no LEE protein has been found to interact with EscA. However, in contrast to the yeast two-hybrid results, in a recent bacterial pull-down assay EscA had been found to interact with EscC [9]. We have therefore reasoned that there must be additional function(s) carried out by EscA. In this study, our findings suggest that EscA is able to affect the stability of the EspA. In addition to this, EscA was found to interact independently with two EspA-binding proteins (CesA2 and EscL) and EscN, but not a third EspA-binding protein (CesAB). Finally, when regulation in EHEC is examined, it is interesting to note that the expression of EscA has been linked to the presence of Mpc, a protein that is absolutely needed for TTTS functionality; however, this protein is tightly regulated and maintained a low level.

Materials and Methods
Bacterial strains and culture conditions EHEC O157:H7 (ATCC strain 43888) [17] was used as the parental strain for creating specific gene deletions and is referred to here as the wild-type (WT) strain. E. coli K-12 strain JM109 (New England Biolabs) was routinely used for DNA cloning and manipulation. Bacteria were regularly grown at 37uC aerobically in Luria-Bertani (LB) broth (Difco). Media were supplemented with antibiotics when necessary: ampicillin, 100 mg/ml; chloramphenicol, 25 mg/ml; and tetracycline, 10 mg/ml. To stimulate EHEC to express the LEE proteins, M9 minimal medium (Difco) in the presence of 5% CO 2 was used for bacterial cultivation [18,19]. When grown in LB under standard conditions, LEE expression is repressed [19].

Primers and expression plasmids
All primers used in this study are listed in Table 1. Plasmids encoding EspA and EspB were constructed by inserting PCR products of the entire Orf (without the stop codon) into pQE60 (Qiagen). The resulting plasmids (pQE-EspA and pQE-EspB) encode the target proteins fused with a hexahistidine (His 66 )-tag at the C-terminus [20]. To express EscA, EHEC chromosomal DNA was amplified by PCR using the paired primers PEscAF and PEscAR. The PCR product was digested with BamHI and BglII, and subsequently ligated into the same enzyme-restricted pQE60 to give pQE-EscA. The plasmids pQE-Mpc, pQE-EspF, and pQE-Map were constructed similarly except for using different primer pairs and a second set of restriction enzymes (NcoI/BglII).
To couple the expression of tag-free EspA with His 66 -tagged EscA, pQE-EscA was used as a backbone, to which a DNA fragment spanning from T5 promoter to the end of espA (without the His 66 -tag-coding codons) and amplified from pQE-EspA was inserted. The resulted pT5-EspA_His-EscA has individual T5 promoters in front of tag-free EspA and His 66 -EscA, respectively. A similar strategy was used to construct pT5-EspB_His-EscA. The resulted plasmids were screened and clones containing the expected restriction patterns were finally confirmed by DNA sequencing. pT5-EscA_His-Mpc was generated by a similar strategy except that the plasmid backbone was pEQ-Mpc and the template used for PCR amplification of the tag-free EscA-coding fragment was pQE-EscA; during the cloning, primers paired for PCR were PRBS-HindIII-F and PEscA-HindIII-R and the restriction enzyme used for digestion and cloning was HindIII.

Constructions of deletion mutant
The method of Gene Gorging was carried out as described previously [21]. This takes advantage of a system that uses the lambda Red recombinase and in vivo linearization of a donor plasmid carrying the desired mutation. In brief, a DNA fragment spanning from sepQ to escN was amplified from the EHEC chromosomal DNA by PCR using the primers of PescNF and PsepQR. The PCR product was subsequently cloned into pTZ57R/T (Fermentas) to result in pTZ31-34. Reverse PCR was then performed using pTZ31-34 as the template and PEscA-R-KO and PEscA-F-KO as paired primers. The PCR product was then self-ligated to produce the escA-deleted plasmid. After sequence confirmation, the resulting plasmid was used for PCR re-amplification with primers PescNF and PI-SecI. The PCR product was then cloned into another pTZ57R/T to give pTZ-33KO that was used as the donor plasmid for the subsequent escA deletion transformation.
Sample preparation (from Bacterial lysate and concentrated Supernatant) and Western blotting analysis were carried out as previously described [18]. The final step of blotted membrane development was carried out using Western Lightning TM Chemiluminescence Reagent Plus (PerkinElmer) and after this the signals were captured by exposing the membranes to X-ray films (Fuji).

EspA and EscA stability assay
The stability assay was carried out as previously described [23]. In brief, bacteria harboring pQE-EspA or pQE-EscA were grown at 37uC aerobically in LB broth overnight. The cultures were then 1:50 inoculated into M9 medium and incubated at 37uC in the presence of 5% CO 2 for 5.5 h. Then, IPTG was added to 1 mM and the cultures were further incubated for 0.5 hr. Bacterial protein synthesis was then stopped by adding chloramphenicol to a final concentration at 200 mg/ml. At different incubation times, the cells were sampled, spun down, disrupted by boiling in SDScontaining sample buffer and analyzed for the presence of EspA or EscA by Western blotting using anti-His 66 antibodies.

Affinity chromatography
Affinity binding of His 66 -tagged proteins to Ni 2+ -NTA agarose beads (Bio-Rad) was carried out as previously described [20] except that the buffer used for equilibration and lysis was 100 mM Tris-HCl, pH 7.4, containing 1 mM PMSF. Wash buffer was consisted of 100 mM Tris-HCl, pH 7.4, containing 50 mM imidazole whereas elution buffer was identical to the wash buffer except that imidazole was added up to a concentration of 250 mM.

RNA isolation and real-time quantitative RT-PCR (qRT-PCR)
RNA isolation method was carried out as previously described [23]. Briefly, total RNAs were extracted from 6 h-cultured EHEC in M9 by using TRIzol reagent (Invitrogen). After extraction, the RNAs were treated with DNase I (Macherey-Nagel) to remove contaminated DNA and reversed to cDNA with the RevertAid TM First Strand cDNA sysnthesis Kit (Fermentas). Quantitative PCR reaction was carried out in triplicates with the Power SYBR Green Master Mix (Protech) and performed by the Stratagene Mx3000P TM Real-Time PCR System. The primers used for qPCR are listed in the Table 1. Comparative CT was used to determine the relative escA mRNA expression fold when normalized to an expression level of internal control gene (ompC).

Bacterial two-hybrid analysis
The system (Stratagene) used has been described previously [18]. In brief, the bait protein generated from pBT plasmid was fused to the C-terminus of a?lcI the target protein, produced from the pTRG plasmid, was fused with an N-terminal RNA polymerase a subunit. The degree of interaction between the bait and target proteins in the system was reflected by measuring the relative b-galactosidase activity [18].

EscA of EHEC is essential for TTSS
The LEE3 operon encodes seven proteins and, among them, EscV, EscN, and SepQ are located on bacterial membrane and are essential for EHEC to deliver proteins from the bacterium into an infected cell [15,24,25]. EscA is the fourth gene of the LEE3 operon. To investigate the role of EscA, escA was deleted [21] from the chromosome of EHEC to give a deletion strain called DescA ( Figure 1A) hereafter. We examined the expression of a number of representative LEE proteins in this mutant strain. Figure 1B (right panel) shows that EspA, EspB, Tir were absent in the supernatant of the DescA mutant, a result that is completely consistent with the previous observations in C. rodentium [8]. The defective secretion could be restored by complementation with expressed His 66 -tagged EscA from the plasmid pQE-EscA. When intracellular proteins were examined ( Figure 1B, left panel), the EspA level of the DescA mutant was decreased as compared to that of the WT strain. However, the levels of EspB and Tir were not affected. This decreased level of intracellular EspA could be restored by ectopic expression of EscA when the DescA mutant was transformed with pQE-EscA. However, this was not the case when transformation was carried out using the vector control, pQE60.
To test whether the escA deletion also affects the secretion of effectors Map and EspF, a different strategy of using anti-His 66 to detect the plasmid-encoded proteins in the bacterial samples was used. Figure 1C (left panel) shows that both Map and EspF, which were tagged with His 66 , were well expressed and detected similarly in both bacterial lysates of the parental strain (WT) and the DescA mutant. However, these tagged Map and EspF were absent in the culture supernatants of the DescA strain but present unequivocally in that of the WT strain (Fig. 1C, right panel). Therefore, deletion of escA disturbs the type-III secretion severely and affects many secreted proteins.

Decreased stability of EspA in the absence of escA
To examine the possibility that the decreased level of intracellular EspA is controlled transcriptionally at the promoter, we introduced T5 promoter-driven expression of EspA-His 66 into the DescA mutant and the parental strain (WT) separately. In these systems, the transcription of the tagged EspA was straightforwardly governed by the T5 cis-element of the plasmid. As a control in parallel, EspB-His 66 was similarly introduced. After 6 h-cultivation of the bacteria in M9 medium, the expression levels of EspA-His 66 and EspB-His 66 were determined by Western blotting. Figure 2A shows that the expression levels of EspB-His 66 were similar in both the WT and DescA strains. However, EspA-His 66 could hardly be detected in the DescA strain but was abundantly found in the parental WT strain. These results strongly suggest that the intracellular level of EspA is down-regulated in a novel manner in the DescA mutant and that transcriptional regulation is unable to account for the differences in expression levels. To confirm that the expression of EspA in strain DescA is not regulated at the mRNA level, a transcriptional fusion reporter [23] was used. Consistent with the above results, it was found that the b-galactosidase activity derived from the espA-lacZ fused transcripts of the WT and DescA transformants showed no significant difference (data not shown).  To account for the decreased intracellular EspA level in the DescA strain, we next hypothesized that, in the absence of EscA, EspA might be unstable. To validate this notion, the stability of EspA in the parental WT and that in the DescA strain were compared after de novo protein synthesis in the bacteria was suppressed by treatment with chloramphenicol. Figure 2B shows the Western blotting results for EspA in the bacterial lysates that were harvested at various time points. Apparently, EspA is rather unstable in the DescA mutant when compared to the parental WT strain. The protein level of EspA in the DescA mutant was lower than that in the parental strain at time zero and the protein had decayed greatly after 1 h, as revealed by quantitative measurement of the relative protein abundance using densitometry ( Figure 2C). In contrast, EspA in the WT strain was maintained at a relatively high level during the sampling period over 16 h.
Whether the unstable EspA in the DescA mutant could result from a decrease of CesA2 and EscL, proteins known to interact with EspA [23,26] and EscA (see below), was further examined. By an approach similar to that used in Fig. 2A, we examined CesA2 and EscL that were encoded by pQE-CesA2 and pQE-EscL, respectively, and expressed in the plasmid-transformed DescA mutant. The results were then side-by-side compared with that seen in the parental WT strain. Figure 2D shows that CesA2 (left panel) and EscL (right panel) were detected indistinguishably in both strains, a result suggesting that CesA2 and EscL are as stable as in the WT strain when the bacteria lack EscA.

Interaction of EscA with EspA
To address whether the instability of EspA in the DescA mutant is due to a lack of protein interaction contributed by EscA, we generated pT5-EspA_His-EscA ( Figure 3A) to express tag-free EspA and His 66 -tagged EscA simultaneously. We examined whether the two proteins could be co-purified because of a possible interaction. After washing and affinity purification of His 66 -tagged EscA from the Ni 2+ -NTA column, EspA was readily detected in the fractions where EscA was eluted ( Figure 3B). For a comparison, a similar construct of pT5-EspB_His-EscA ( Figure 3A) was used to express tag-free EspB and His 66 -tagged EscA simultaneously. Figure 3C shows that although EspB was found unequivocally in the flow-through fraction of the Ni 2+ -NTA column, it was not detected in the fractions containing the eluted His 66 -tagged EscA. Therefore, these findings confirm that EscA interacts specifically with EspA.

Interactions of EscA with CesA2, EscL, and EscN
The above results indicate that, in the absence of EscA, the stability of EspA was decreased in EHEC and that EscA possessed the abilities to interact with EspA biochemically. These two properties of EscA are similar to those of three other proteins, namely the two chaperons reported for EspA, CesAB and CesA2 [26,27] and the EspA-binding protein EscL [23]. Thus, we proposed that EscA might conjoin with these proteins in assisting the movement of EspA and perhaps a close interaction between these proteins might exist. To test this notion, pull-down assays were performed to examine a possible interaction between EscA and any one of these three previously known EspA-binding proteins. First, tag-free EscA and His 66 -tagged CesA2-expressing JM109 were separately cultured and total proteins were subsequently harvested. After binding His 66 -tagged CesA2 onto Ni 2+ -NTA column and followed by washing, tag-free EscA was subsequently applied to the column. After additional washing and elution, EscA was found to appear with His 66 -tagged CesA2 ( Figure 4A) in the eluent fractions. A similar experiment was carried out to investigate the interaction between EscA and EscL, and Figure 4B shows a positive result that is similar to that of EscA and CesA2. An irrelevant His 66 -tagged protein (YgfZ) was used as negative control to test for co-elution with EscA using the Ni 2+ -NTA column. Using this control protein, no EscA was detected in the eluent fractions ( Figure 4C). On the other hand, a similar attempt to pinpoint whether there is an interaction between EscA and CesAB using the same strategy gave a clearly negative result, which is shown in Figure 4D.
It has been previously reported that CesAB could not interact with EscN. However, the interaction occurs after CesAB interacts with EspA [28]. Therefore, whether EscA could form complex with EscN prior to the interaction of EscA with EspA is then addressed. First, amino acid sequences of CesAB and that of EscA were compared and no primary sequence similarity could be concluded. Second, an interaction possibility between EscA and EscN was tested by using affinity Ni 2+ -NTA column to retain His 66 -tagged EscN and followed by applying untagged EscA; whether EscA could be co-eluted with EscN was then examined. Figure 4E shows positively that EscA alone could interact with EscN as evidenced by the results of co-elution. To unveil the second layer of possibility, we address whether stable interaction between EspA with CesA2 could be disrupted by the presence of EscA or a ternary complex could be formed after the addition of EscA. His 66 -tagged CesA2 and tag-free EspA interaction was first performed in the Ni 2+ -NTA column. It was then followed by applying tag-free EscA in a clarified bacterial lysate into the column. After extensive washing, proteins coretained with His 66 -tagged CesA2 in the Ni 2+ -NTA column was examined by Western blotting analysis. Figure 5A shows clearly that EspA and CesA2 were detected in the eluent fraction but not EscA, a result suggesting that the complex of CesA2 and EspA remains stable and neither a ternary complex consisting of CesA2-EspA-EscA is stably formed after the addition of EscA. A similar experiment was performed by retaining His 66 -tagged EscL and tag-free EspA first in the column followed by supplementing with EscA. Neither disturbing effect nor ternary multipart formation was seen with the addition of EscA ( Figure 5B). A negative observation was also found when EscA was added into the column that first retained His 66 -tagged CesAB and EspA ( Figure 5C).

Regulation of EscA expression
To explore the possibility that there is an EscA regulatory mechanism in EHEC, we tested whether the expression of EscA was responding to the known gene deletions in LEE that grossly affect TTS of EHEC. We transformed pQE-EscA into individual EHEC variants that contain deletions affecting each of the known regulators in the LEE, namely ler, glrA, glrR, and mpc. We then examined the expression levels of EscA. Figure 6 shows that deleting grlA or grlR had no adverse effect on the expression level of EscA when compared to the parental WT strain. On the other hand, EscA could hardly be detected in the ler deletion strain (Dler). EscA was also absent in strain AC36, in which initiation of Mpc translation has been abolished, but transcription of the rest of . Affinity interaction between EscA with CesA2, between EscA and EscL, and between EscA and EscN as demonstrated by co-elution from a Ni 2+ -NTA column. Experiments were carried out in a manner similar to that in Figure 3 with a slight modification. In essence, a His 66 tag was placed at the C-terminal end of CesA2 (A), EscL (B), YgfZ (C), EscA (D), and EscN (E). Bacterial lysate containing the His 66 -tagged protein was individually applied to the Ni 2+ -NTA column. Then bacterial lysate containing tag-free EscA (A-C and E) or CesAB (D) was applied to the column. After washing and eluting, fractions were analyzed by Western blotting with specific antibodies except that all His 66 -tagged proteins were all detected by anti-His 66 . I: input; F: flow through. doi:10.1371/journal.pone.0085354.g004 Figure 5. Neither disruption nor ternary multipart formation to the preformed complex of CesA2-EspA, EscL-EspA, or EscL-EspA by adding EscA. Experimental procedure was similar to Figure 4 except that tagged-free EscA was individually applied after the columns that were first bound with His 66 -tagged proteins followed by binding of tag-free EspA. I: input; F: flow through; W: wash; E: eluent. doi:10.1371/journal.pone.0085354.g005 LEE3 remains untouched [18]. Since expression of EscA from pQE-EscA is governed by the exogenous promoter T5, it was reasoned that the above expression differences were unlikely to be due to transcriptional variation at the escA mRNA level between strain AC36 (or Dler) and strain DgrlA or DgrlR. Therefore, we proposed that EscA expression is correlated with the presence/ absence of Mpc and/or Ler.

Expressing regulation of EscA is mediated by interaction with Mpc
To test the above possibilities, we examined the strains AC36 and Dler after supplementing with the critical element to determine whether correct EscA expression was recovered. Since Ler is a global LEE regulator that activates multiple lee operons when upregulation is signaled, it would not be surprising that extra Ler was able to trigger an increase in expression of various LEE proteins including EscA. Therefore, we chose rather to ectopically express Mpc from a compatible plasmid pAC-Mpc. Mpc and EscA, both His 66 tagged, were co-expressed in strains AC36 and Dler ( Figure 7A). As a reference, we included strain DescA in the experiments. As seen in Figure 7A, strain DescA that simultaneously harbored pQE-EscA and pAC-Mpc gave an increasing expression of EscA when compared to the same strain harboring pQE-EscA alone (compare lanes 1 and 2). Using strain AC36 ( Figure 7A, lane 3) and as seen in Figure 6, no EscA was detected when this strain was transformed with pQE-EscA alone. However, when pAC-Mpc striking results were seen, however, when the Dler strain was transformed with pQE-EscA; this gave no detectable EscA as seen above ( Figure 6). Co-transformation with both pQE-EscA and pAC-Mpc resulted in significant expression of EscA ( Figure 7A, comparing lanes 1, 5, and 6). Since mpc is the first gene in the LEE3 operon and escA is was co-transformed with pQE-EscA into this strain, this gave identifiable EscA expression ( Figure 7A, lane 4). More located downstream of mpc, it seems likely that Mpc has a direct effect on the expression of EscA.
To examine whether the amounts of mRNA coding for EscA differs between strain AC36 and the WT strain, RNAs from both strains were extracted and analyzed by qRT-PCR. Results in Figure 7B indicated that the levels of escA mRNA had no significant difference between the WT and AC36 strains. The notion that RNA level difference could not account for the adverse dissimilarity between the expression levels of the two strains could be further supported by the experiments in Figure 7A, in which mRNA for EscA was driven all by the T5 promoter. Therefore, regulations that may affect the protein per se were further examined. A careful examination of EscA detected in the pQE-EscA-transformed WT and AC36 strains was then carried out. Figure 7C shows the sequential detection results of EscA after that bacteria received chloramphenicol and previously induced with IPTG for 0.5 h. EscA level in the Mpc-deficient strain (AC36) was obviously lower than that in the WT strain at time zero, and the EscA signal disappeared quickly and was barely detected after 1-h bacterial culture. On the other hand, EscA protein in WT was relative stable and it remained abundant during the assay period  (8 h). The above results suggest that the protein stability of EscA may account for the major difficulty to express EscA in the absence of Mpc. A simple hypothesis is that Mpc might interact with EscA and enhance EscA expression.
To test the above notion, we examined the possible interaction between EscA and Mpc using a bacterial two hybrid system. In this system, interaction between target protein and bait protein is revealed by way of an increase in reporter b-galactosidase activity. As references, three negative controls and one positive control [18] were included in the experiments. The results in Figure 8A showed that EscA does interact with Mpc, albeit not strongly. A reciprocal hybrid, where Mpc was incorporated as the target protein with EscA acting as the bait protein, also gave a positive result (data not shown).
To confirm the above notion that Mpc interacts with EscA, a co-elution assay was also performed. Using a strategy similar to that seen in Figure 3, we generated expression plasmids expressing tag-free EscA alone and with His 66 -tagged Mpc simultaneously. After eluting His 66 -tagged Mpc from the Ni 2+ -NTA column, the fractions were examined for the presence of EscA. Figure 8B confirms that EscA is readily co-eluted with Mpc but the protein was absent from the eluent when no His 66 -tagged Mpc was present in the system.

Discussion
EscA is a small protein rich in coiled-coil structures. Previous studies have shown that EscA is indispensable for the secretion of both translocator and effector in EPEC and C. rodentium [8,9]. It has also been shown that EscA interacts with EscC [9], one of the TTSS components on the outer membrane that protrudes toward the periplasmic space. And EscA has been found mainly in the periplasmic fraction, followed by inner membrane-associated fraction and then in the cytoplasm. The property whereby EscA is found in is consistent with the notion that EscA makes a ''contribution to the structural integrity of the TTS complex'' [9] and backs the physical necessity that EscA interacts with periplasm-protruding EscC. As seen from our findings, EscA also interacts with a number of other LEE proteins in addition to EscC. First, EscA is able to interact independently with EscL and CesA2, two of the three currently known EspA-binding LEE proteins. It is worth noting that, EscL and CesA2 have been localized to the inner membrane-associated fraction [23,26], a place where EscA is also detected. In contrast, CesAB is found in the bacterial cytoplasm and functions as an EspA chaperone in that compartment. However, this EspA-binding protein EscA is unable to interact with CesAB. The fact that EscA by itself is able to interact with EspA means that EscA is the fourth EspA-binding protein.
Furthermore, it shares in common with the other EspA-binding proteins the fact that deletion of the gene coding for the cognate EspA-binding protein decreases the stability of EspA and this reduces the intracellular level of EspA. Therefore, EscA readily fits into this category.
Before being exported out, one rationale is that secretory effector proteins should be quenched with chaperons to keep them EscA and Mpc demonstrated by the bacterial two-hybrid system. Mpc was cloned as a bait fusion protein whereas EscA was cloned as a target [18]. Interaction of the bait and the target is reflected as an increase in b-galactosidase activity when compared to the negative (N) control. Interaction between Mpc and Ler was used as a positive (P) control. *: p,0.001 (B) Co-elution of EscA with Mpc. The experiments were carried out in a manner similar to that of Figure 3. doi:10.1371/journal.pone.0085354.g008 Figure 9. A summary of networked interactions of currently known EspA-binding proteins. Three of the four known EspAbinding proteins are associated with the inner membrane, as shown near a basal body of model TTSS [15]; they are EscL, CesA2, and EscA. EscA alone is able to interact with EscN ATPase, an event that characteristically differs from the interaction between CesAB and EscN, which requires a pre-exposure of CesAB to EspA [28]. EscA also interacts with EscC that is localized to the bacterial outer membrane, perhaps due to that EscA is also present in the periplasm [9]. Solid line indicates an interaction previously suggested, with reference provided; interaction suggested here is asterisked. Broken arrow depicts that a conformational switch of CesAB is needed before CesAB interacts with EscN and the switch is triggered by EspA (linked by an arrow line). Note: EscA does not disrupt the binary interactions of EspA with the other EspA-binding proteins nor it stimulates a ternary complex formation from the existing binary interaction complexes. doi:10.1371/journal.pone.0085354.g009 in the unfolded status. Once the TTSS is activated, these proteins should be escorted by chaperons from the cytosol to cytoplasmic base of an injectisome, a membrane-associated basal apparatus. EscN, an essential ATPase for TTSS locating to the entrance of the injectisome, provides the energy to disassemble the chaperon/ effector protein complex in an ATP-dependent manner [29]. Exemplified is that homodimeric CesAB does not interact with EscN. However, upon interaction with EspA, CesAB homodimer transforms into CesAB/EspA heterodimer that also induces a conformational switch of CesAB [28,30] so that CesAB is able to interact with EscN by exposing the EscN-recognition domain. Here, we found that EscA alone is able to interact with EscN, an observation that is rather similar to the interaction between CesT and EscN [31]. To sum it up, currently known interactions among EspA and its binding proteins are depicted in Figure 9.
Amazingly seen is that EscA is able to interact with Mpc, a protein encoded by the first gene of the LEE3 operon, in which escA is the fourth open reading frame ( Figure 1A). Mpc has been previously shown to be indispensable but is expressed at only a low level since strong expression results in a severe suppression of the TTSS via counteraction with Ler [18]. Here, we have demonstrated that Mpc is essential for the expression of EscA (Figure 7). In terms of LEE3 organization, escA comes after mpc, escV, and escN in the operon and before l0032, sepQ, and espH. The indispensable role of mpc in EscA expression and the interaction of Mpc with EscA suggest to us that EscA also needs to be expressed at a low level and EscA degrades rapidly without the presence of Mpc. It is then conceivable that such expression needs to occur in a timely manner for the purpose of successfully assembling the basal body of the TTSS. To test whether EscA is able to suppress the synthesis of TTSS proteins in a similar manner to Mpc, EscA was overexpressed in the WT strain of EHEC. The results (data not shown) indicate that an increased expression of EscA seems to have no apparent effect on the TTSS. Thus, it is likely that EscA does not play as profound a regulatory role in the LEE as Mpc.