Yersinia pestis , the causative agent of plague, exports a set of virulence proteins called Yops upon contact with eukaryotic cells. A subset of these Yops is translocated directly into the cytosol of host cells. In this study, a novel protein tag-based reporter system is used to measure the translocation of Yops into cultured eukaryotic cells. The reporter system uses a small bipartite phosphorylatable peptide tag, termed the Elk tag. Translocation of an Elk-tagged protein into eukaryotic cells results in host cell protein kinase-dependent phosphorylation of the tag at a specific serine residue, which can subsequently be detected with phosphospecific antibodies. The YopN, TyeA, SycN, YscB and LcrG proteins function to prevent Yop secretion before host cell contact. The role of these proteins was investigated in the translocation of Elk-tagged YopE (YopE 129 –Elk) and YopN (YopN 293 –Elk) into HeLa cells. Y. pestis yopN , tyeA , sycN and yscB deletion mutants showed reduced levels of YopE 129 –Elk phosphorylation compared with the parent strain, indicating that these mutants translocate reduced amounts of YopE. We also demonstrate that YopN 293 –Elk is translocated into HeLa cells and that this process is more efficient in a Yersinia yop polymutant strain lacking the six translocated effector Yops. Y. pestis sycN and yscB mutants translocated reduced amounts of YopN 293 –Elk; however, tyeA and lcrG mutants translocated higher amounts of YopN 293 –Elk compared with the parent strain. These data suggest that TyeA and LcrG function to suppress the secretion of YopN before host cell contact, whereas SycN and YscB facilitate YopN secretion and subsequent translocation.
Secretion of Yop proteins via the TTS pathway is not a constitutive process; instead, export is triggered by contact between the bacterium and a eukaryotic cell. In vitro, Yop secretion is blocked in the presence of calcium and triggered during growth at 37°C in the absence of calcium (Michiels et al., 1990). The block in Yop secretion in the presence of calcium and before eukaryotic cell contact is dependent upon the secreted YopN protein and the cytosolic TyeA, SycN, YscB and LcrG proteins (Forsberg et al., 1991; Skrzypek and Straley, 1993; Nilles et al., 1997; Iriarte et al., 1998; Day and Plano, 1998; Cheng and Schneewind, 2000). Mutational inactivation of any of these genes results in secretion of Yops in the presence or absence of calcium and before host cell contact.
In this study, we develop a novel reporter system designed to study the translocation of Yops into host cells. We use this technique to investigate the role of Y. pestis YopN, TyeA, SycN, YscB and LcrG in the translocation of Yops into HeLa cells. We show that these proteins are not directly required for translocation of YopE; however, their presence increases the efficiency of this process, presumably by preventing Yop release before contact with the surface of a eukaryotic cell. We also demonstrate that YopN is translocated into HeLa cells and that YopN secretion and subsequent translocation are promoted by the SycN–YscB chaperone complex and inhibited by TyeA and LcrG.
Development of the Elk tag system to detect Yop translocation
The translocation of Yops into eukaryotic cells has been measured using a calmodulin-dependent adenylate cyclase (Cya) reporter system developed by Sory et al. (1994). Alternatively, Lee et al. (1998) have used a subcellular fractionation technique that utilizes digitonin, a cholesterol-specific detergent that solubilizes eukaryotic, but not prokaryotic, cell membranes, to detect and measure Yop translocation. The Cya reporter system requires that the protein of interest be fused with the 400-residue adenylate cyclase catalytic domain which, in our hands, significantly reduced the secretion of the resultant hybrid protein (Fig. 2). In addition, Yop translocation is measured indirectly by determining the increase in intracellular cAMP, a time-consuming and relatively expensive assay. The digitonin fractionation method uses immunoblotting techniques to detect the protein of interest directly; however, the release of extracellular surface-bound proteins into the digitonin-soluble fraction (eukaryotic cell cytosolic fraction) is also possible and difficult to control for, especially when using strains that release large amounts of Yops into the extracellular media (calcium-blind or los phenotype).
In this study, we developed a novel reporter system designed to allow the specific identification of Yop proteins that have been translocated into eukaryotic cells. This reporter system is based on a bipartite phosphorylatable peptide tag, termed the Elk tag (Fig. 1). The Elk tag consists of the SV40 large tumour antigen nuclear localization signal (NLS) and a portion of the eukaryotic transcription factor Elk-1 (amino acids 375–392) that is recognized and phosphorylated at serine 383 by eukaryotic protein kinases, including the extracellular signal-related protein kinase Erk2 (Kortenjann et al., 1994). Phosphorylation of the Elk tag at serine 383 allows specific detection with phosphospecific Elk-1 antipeptide antibodies. In theory, Elk-tagged proteins translocated into a eukaryotic cell are phosphorylated, whereas proteins within the bacterial cell, in the extracellular medium or on the bacterial or host cell surface are not.
In practice, the DNA sequence encoding the Elk tag is fused in frame with the sequence encoding a particular Yop, and the vector encoding the Elk-tagged hybrid protein is expressed in the Y. pestis strain of interest. A suitable mammalian cell monolayer is then infected with the recombinant strain for sufficient time to allow translocation of the hybrid protein into the infected cells and phosphorylation of the Elk tag at Elk-1 serine 383. After the infection is terminated, the culture medium is discarded (or analysed separately), the infected cells are lysed with SDS sample buffer and the lysates are analysed in duplicate by SDS-PAGE and immunoblot analysis with Elk-1 and phosphospecific Elk-1 antibody preparations (Cell Signaling Technology). The Elk-1 antibody preparation detects total Elk-tagged protein expression levels (phosphorylation state independent), whereas the phosphospecific Elk-1 antibody preparation detects only Elk-tagged proteins that are translocated and phosphorylated at Elk-1 serine 383.
YopE–Elk fusions are stably expressed in Y. pestis and secreted into the extracellular environment
YopE has been shown to be translocated into HeLa cells using the Cya fusion approach, by digitonin fractionation and by immunocytochemistry combined with confocal microscopy (Rosqvist et al., 1994; Sory et al., 1994; Lee et al., 1998; Lee and Schneewind, 1999). The N-terminal 129 amino acid residues of YopE are sufficient to target YopE–Cya fusions into HeLa cells. In order to test the Elk tag method, we engineered plasmids pYopE129-Elk (Fig. 1) and pYopE219-Elk. These plasmids carry the sycE gene, encoding the YopE-specific chaperone and the region of yopE encoding amino acid residues 1–129 or 1–219 (full length), respectively, fused to the DNA fragment encoding the Elk tag. In addition, plasmids containing the sequence encoding amino acids 375–392 of Elk-1 (no NLS) fused directly to the sequences encoding YopE129 or YopE219 were constructed to assess the contribution of the NLS to Elk tag function.
Plasmids pYopE129-Elk, pYopE219-Elk and pYopE129-Cya were introduced into Y. pestis KIM8-3002.P39 (ΔyopE) [ΔyopE], and the expression and secretion of the YopE–Elk and YopE–Cya hybrid proteins were compared with that of full-length YopE [parent] after 5 h of growth in TMH medium at 37°C in the presence and absence of calcium (Fig. 2). The amount of YopE or YopE-hybrid protein present in the cell pellets and culture supernatants was determined by SDS-PAGE and immunoblotting with antiserum specific for YopE. Fusion of the Elk tag to YopE amino acid residue 129 or 219 had a minimal effect on the expression and secretion of the YopE-hybrid proteins; in contrast, fusion of the adenylate cyclase domain to residue 129 of YopE dramatically reduced the secretion of the YopE129–Cya hybrid protein. Specifically, 48% of YopE, 45% of YopE129–Elk and 30% of YopE219–Elk were secreted in the absence of calcium, but only 9% of YopE129–Cya was secreted under identical conditions. These results demonstrate that the addition of the 35-amino-acid residue Elk tag to a TTS substrate does not significantly deter export of the tagged protein. Owing to the cytotoxicity associated with the effector domain of full-length YopE, the YopE129–Elk expressing construct was used in further experiments.
Cell pellet and culture supernatant fractions from Y. pestis KIM8-3002.P39 (ΔyopE) carrying plasmid pYopE129-(no NLS)Elk or pYop129-Elk were analysed further with the Elk-1 and phosphospecific Elk-1 antibody preparations (Fig. 3A). The Elk-1 antibody detected YopE129–Elk and YopE129–(no NLS)Elk in both cell pellet and culture supernatant fractions, indicating that each of the tagged proteins was stably expressed and secreted under inductive conditions (37°C, no calcium). Neither of the Elk-tagged proteins was detected using the phosphospecific Elk-1 antibody, indicating that the Elk tag (±NLS) was not phosphorylated in the bacterial cell or in the culture supernatant.
Translocation and phosphorylation of Elk-tagged YopE in HeLa cells
To determine whether the Elk-tagged YopE hybrids could be delivered into HeLa cells and phosphorylated at Elk-1 serine 383, HeLa cell monolayers were infected with YopE129–Elk- and YopE129–(no NLS)Elk-expressing strains. Y. pestis KIM5-3001.P39 (ΔyopE) carrying pYopE129-Elk or pYopE129-(no NLS)Elk was used at a multiplicity of infection (MOI) of 30 bacteria per HeLa cell. After 3.5 h of infection at 37°C, the cell culture medium was discarded, and the infected HeLa cell monolayer was lysed by the addition of SDS-PAGE sample buffer containing phosphatase inhibitors and protease inhibitors (see Experimental procedures). Protein lysates from the infected HeLa cells were analysed by SDS-PAGE and immunoblotting with the Elk-1 and phosphospecific Elk-1 antibody preparations (Fig. 3B). The Elk-1 antibody detected both YopE129–Elk and YopE129–(no NLS)Elk proteins in the HeLa cell lysates, indicating that the Elk-tagged proteins were expressed during the infection. The YopE129–Elk protein was also detected in the HeLa cell lysates using the phosphospecific Elk-1 antibody, suggesting that YopE129–Elk was translocated and phosphorylated in the HeLa cells. The YopE129–(no NLS)Elk protein was not detected using the phosphospecific antibodies, suggesting that the Elk tag NLS plays a critical role in targeting YopE to the nucleus for subsequent eukaryotic protein kinase-dependent phosphorylation.
Phosphorylation of YopE129–Elk by HeLa cells and RAW 264.7 cells is not influenced by co-expression of YopJ
The translocated effector protein YopJ has been shown to be a cysteine protease that inhibits MAP kinase and NF-κB signalling pathways in eukaryotic cells (Palmer et al., 1998; Schesser et al., 1998; Orth et al., 2000). Phosphorylation of the Elk-1 transcription factor at serine 383 is partially dependent upon the MAP kinase Erk2; therefore, YopE129–Elk phosphorylation was examined in both YopJ-expressing (Y. pestis KIM5-3001.P39 [ΔyopE]) and YopJ non-expressing (KIM5-3001.P40 [ΔyopE yopJ]) strains carrying plasmid pYopE129-Elk. Analysis of HeLa cell lysates and RAW 264.7 murine macrophage cell lysates infected with each strain revealed no YopJ-dependent differences in the level of YopE129–Elk phosphorylation (Fig. 4A), suggesting that the level of activated Erk2 in infected cells (± YopJ) is sufficient to phosphorylate YopE129–Elk or that other eukaryotic protein kinases are involved in Elk tag serine 383 phosphorylation. In order to exclude YopJ-dependent effects on Elk tag phosphorylation in other studies, further experiments were carried out with strains deleted for yopJ. These results also demonstrate that Elk tag phosphorylation can be used to measure Yop translocation into both human epithelial and murine macrophage-like cells.
Phosphorylation of YopE129–Elk is dependent upon functional type III secretion and translocation machineries
To confirm that Elk tag phosphorylation is representative of Yop translocation, we measured YopE129–Elk expression and phosphorylation in Y. pestis strains defective in Yop secretion (ΔyscV) or translocation (ΔyopB). HeLa cells were infected with Y. pestis KIM5-3001.P40 (ΔyopE yopJ), KIM5-3001.P41 (ΔyopE yopJ yopB) and KIM5-3001.P42 (ΔyopE yopJ yscV) carrying pYopE129-Elk at an MOI of 30 for 3.5 h. Analysis of the infected HeLa cell lysates with the Elk-1 antibody revealed that both yopB and yscV deletion mutants expressed YopE129–Elk at levels comparable to that of the parent strain; however, no phosphorylation of YopE129–Elk was detected in either of the infected cultures (Fig. 4B). Complementation of the yopB and yscV deletion mutants with plasmids pYopB2 and pYscV2, respectively, restored YopE129–Elk phosphorylation to levels comparable to that of the parent strain. These data indicate that YopE129–Elk is phosphorylated only upon delivery into a eukaryotic cell, a process that requires both a secretion- and a translocation-competent Y. pestis strain. The complete absence of YopE129–Elk phosphorylation in the Y. pestis yopB mutant and subsequent restoration of YopE129–Elk phosphorylation with plasmid pYopB2 confirms that YopB is essential for Yop translocation.
Secretion and translocation of YopE129–Elk expressed from plasmids pYopE129-Elk and pCD1::yopE129-Elk
Plasmid pYopE129-Elk expresses YopE129–Elk from a pBluescript cloning plasmid that is stably maintained at a relatively high copy number in Y. pestis. In contrast, plasmid pCD1 is estimated to be maintained at only three to five copies per cell. To evaluate the effect of gene dosage on YopE129–Elk expression and delivery, the yopE129–Elk gene fusion was moved into plasmid pCD1 of Y. pestis KIM8-3002 (Pla–), generating Y. pestis KIM8-3002.P54 (Pla–; yopE129–Elk). Expression and secretion of YopE and YopE129–Elk by Y. pestis KIM8-3002 [parent] and KIM8-3002.P54 (Pla–; yopE129-Elk) [yopE129-Elk], respectively, were determined after growth in TMH medium at 37°C in the presence and absence of calcium (Fig. 5A). As expected, the expression and secretion of both YopE and YopE129–Elk were calcium regulated; however, the amount of YopE129–Elk secreted (32%) was decreased compared with YopE (61%) or YopE129–Elk expressed from pYopE129-Elk (see Fig. 2).
To examine the effect of gene dosage on the translocation and subsequent phosphorylation of YopE129–Elk, HeLa cell monolayers were infected with KIM8-3002.P40 (ΔyopE yopJ) carrying pYopE129-Elk and KIM8-3002.P54 (Pla–; yopE129–Elk). The level of YopE129–Elk expression and translocation (phosphorylation) was significantly greater when the protein was expressed from the high-copy-number pYopE129-Elk plasmid (Fig. 5B). Based upon these results, the pYopE129-Elk expression plasmid was chosen for use in further studies.
Yersinia pestis mutants defective in the production of YopN, TyeA, SycN or YscB translocate reduced amounts of YopE129–Elk
YopN, TyeA, SycN, YscB and LcrG are necessary to prevent Yop secretion in the presence of calcium and before contact with a eukaryotic cell. Previous studies using a Y. pestis lcrG deletion mutant have shown that LcrG is not required for Yop translocation; however, expression of LcrG promotes efficient Yop translocation (Sarker et al., 1998; Fields et al., 1999; DeBord et al., 2001). More recently, Cheng et al. (2001) have used digitonin fractionation to show that Y. enterocolitica yopN, tyeA, sycN and yscB deletion mutants translocate reduced amounts of YopE into HeLa cells. To elucidate further the role of YopN, TyeA, SycN and YscB in Yop translocation, we measured the translocation of YopE129–Elk by Y. pestis yopN, tyeA, sycN and yscB deletion mutants. To accomplish this, we moved the yopN, tyeA, sycN and yscB deletions into Y. pestis KIM5-3001.P40 (ΔyopE yopJ ), generating Y. pestis strains KIM5-3001.P44 (ΔyopE yopJ yopN), KIM5-3001.P45 (ΔyopE yopJ tyeA), KIM5-3001.P46 (ΔyopE yopJ sycN) and KIM5-3001.P47 (ΔyopE yopJ yscB). We then compared YopE129–Elk translocation by the Y. pestis parent strain with that of the yopB, yopN, tyeA, sycN and yscB deletion mutants with and without complementing plasmids pYopB2, pYopN2, pTyeA2, pSycN2 and pYscB2 respectively (Fig. 6). Analysis of lysates from HeLa cells infected with the Y. pestis parent strain or with the yopB, yopN, tyeA, sycN or yscB mutants showed similar levels of YopE129–Elk expression (Elk-1 antibody). In contrast, the phosphospecific Elk-1 antibody detected only low amounts of phosphorylated YopE129–Elk in the lysates of HeLa cells infected with the yopN, tyeA, sycN and yscB deletion mutants, confirming that YopE129–Elk translocation was significantly reduced in these strains. Complementation of the yopB, yopN, tyeA, sycN and yscB deletion mutants with plasmids pYopB2, pYopN2, pTyeA2, pSycN2 and pYscB2, respectively, restored normal levels of YopE129–Elk translocation, confirming that the translocation defects associated with these strains result from the lack of YopB, YopN, TyeA, SycN or YscB and not from polar effects or other mutations in plasmid pCD1.
A yopN deletion strain secretes YopE129–Elk into the extracellular medium during HeLa cell infection experiments
After contact between the bacterium and a eukaryotic cell, effector Yops are targeted directly into the eukaryotic cell and are not found in substantial amounts in the extracellular milieu, indicating that Yop translocation is polarized (Rosqvist et al., 1994; Persson et al., 1995). Mutational inactivation of yopN, tyeA, sycN or yscB in the enteropathogenic yersiniae results in uncontrolled secretion before host cell contact and release of effector Yops into the extracellular medium (Rosqvist et al., 1994; Iriarte et al., 1998; Cheng et al., 2001). Owing to expression of the plasmid pPCP1-encoded outer membrane protease (Pla), quantification of Yop release into the extracellular medium by Y. pestis strains is difficult to measure (Sodeinde et al., 1988). To facilitate the localization of YopE129–Elk during tissue culture infections, plasmid pPCP1-cured (Pla–) Y. pestis strains [KIM8-3002.P40 (ΔyopE yopJ) and KIM8-3002.P44 (ΔyopE yopJ yopN)] were used to infect HeLa cell monolayers. After a 3.5 h infection, the extracellular media were removed from the infected monolayers and centrifuged to separate the non-adherent bacteria from the tissue culture media containing secreted proteins. The non-adherent bacteria, bacteria-free media and the infected HeLa cell monolayer from each cell culture infection were analysed separately with the Elk-1 or phosphospecific Elk-1 antibodies (Fig. 7). As expected, almost no YopE129–Elk protein was present in the media supernatant fraction obtained from the HeLa cell monolayer infected with the pPCP1-cured parent strain (ΔyopEJx), demonstrating that YopE129–Elk was predominantly directed into the HeLa cells and not released into the extracellular medium. In contrast, the pPCP1-cured yopN deletion mutant (ΔyopEJNx) released large amounts of YopE129–Elk into the extracellular medium, confirming that YopN was required for the directional translocation of YopE129–Elk into the eukaryotic cells. As expected, the YopE129–Elk protein was difficult to detect in the media supernatant fraction obtained from monolayers infected with the Pla protease-expressing strains (ΔyopEJ and ΔyopEJN); however, several Elk-1 antibody-reactive proteolytic digestion products were detectable in the media supernatant fraction isolated from the yopEJN-infected monolayer, suggesting that YopE129–Elk was released and subsequently degraded by Pla. Similar results were seen for the Pla protease-expressing tyeA, sycN and yscB mutants (data not shown). These data suggest that the Y. pestis yopN, tyeA, sycN and yscB mutants translocate reduced amounts of YopE129–Elk, not because of a defect in secretion or translocation, but because of an inability to block Yop secretion before cell contact.
Translocation of YopN293–Elk into HeLa cells by a Y. pestis yop polymutant strain
Results presented by Lee and Schneewind (1999) and Cheng et al. (2001), obtained using the digitonin fractionation technique, strongly suggest that Y. enterocolitica YopN is translocated into HeLa cells. In contrast, Boland et al. (1996) failed to detect the translocation of several truncated YopN–Cya hybrid proteins. To examine targeting of YopN in Y. pestis, we constructed plasmid pYopN293-Elk, which encodes YopN amino acid residues 1–293 (full length) fused to the Elk tag. Plasmid pYopN293-Elk was initially transformed into Y. pestis KIM5-3001.6 (ΔyopN) (Jackson et al., 1998), and Yop secretion was measured after growth at 37°C in the presence and absence of calcium. The yopN deletion mutant carrying pYopN293-Elk secreted YopE and YopN293–Elk in the presence and absence of calcium (data not shown), indicating that the YopN293–Elk protein was defective in its ability to regulate Yop secretion. Similarly, Cheng et al. (2001) reported that a YopN–neomycin–phosphotransferase (NPT) hybrid protein is defective in the regulation of Yop secretion, indicating that YopN's regulatory function can be disrupted by the addition of either a small peptide or a protein to the C-terminus of YopN. Thus, to examine YopN293–Elk translocation, pYopN293-Elk was moved into the calcium-regulated YopN-expressing Y. pestis KIM5-3001.P40 (ΔyopE yopJ) strain, which was then used to infect HeLa cells as described previously. HeLa cell lysates were analysed with both the Elk-1 and the phosphospecific Elk-1 antibody preparations (Fig. 8A). A low, but reproducible, level of YopN293–Elk phosphorylation was detected, indicating that YopN was translocated into HeLa cells in the presence of wild-type YopN and other secreted Yops.
It is conceivable that YopN293–Elk was translocated inefficiently because of competition for the secretion and/or translocation machinery from wild-type YopN and/or the other Yops. Therefore, we moved plasmid pYopN293-Elk into a Y. pestis yop polymutant strain KIM5-3001.P48 (ΔyopEJTMHA lcrQ) that is defective for expression of the six translocated effector Yops (YopE, YopJ, YopT, YopM, YopH and YpkA) and LcrQ, a secreted protein that functions to repress pCD1 gene transcription before the triggering of secretion (Rimpiläinen et al., 1992; Pettersson et al., 1996; Stainier et al., 1997). Immunoblot analysis of HeLa cell lysates infected with the derepressed yop polymutant strain carrying pYopN293-Elk revealed a significant increase in the level of YopN293–Elk translocation and phosphorylation compared with the yopEJ strain (Fig. 8A).
Secretion and translocation of YopN293–Elk is promoted by the SycN–YscB complex and inhibited by TyeA and LcrG
The YopN, TyeA, SycN, YscB and LcrG proteins are necessary to prevent Yop secretion in the presence of calcium and before contact with a eukaryotic cell. YopN interacts directly with the SycN–YscB chaperone complex and TyeA (Day and Plano, 1998; Iriarte et al., 1998; Cheng et al., 2001). LcrG, on the other hand, binds to LcrV, a secreted protein that is required for the translocation of effector Yops into eukaryotic cells (Nilles et al., 1997; Matson and Nilles, 2001). To investigate the role of SycN, YscB, TyeA and LcrG in the translocation of YopN, we measured YopN293–Elk translocation into HeLa cells by Y. pestis sycN, yscB, tyeA and lcrG deletion mutants. The yopB, tyeA, sycN, yscB and lcrG deletion mutations were moved individually into KIM5-3001.P48 (ΔyopEJTMHA lcrQ) generating strains KIM5-3001.P49 (ΔyopEJTMHA lcrQ yopB), KIM5-3001.P50 (ΔyopEJTMHA lcrQ tyeA), KIM5-3001.P51 (ΔyopEJTMHA lcrQ sycN), KIM5-3001.P52 (ΔyopEJTMHA lcrQ yscB) and KIM5-3001.P53 (ΔyopEJTMHA lcrQ lcrG) respectively. These strains (with and without complementing plasmids) were transformed with plasmid pYopN293-Elk, and the resultant transformants were used to infect HeLa cell monolayers. Expression and translocation (phosphorylation) of the YopN293–Elk protein was examined by SDS-PAGE and immunoblot analysis with the Elk-1 and phosphospecific Elk-1 antibody preparations respectively (Fig. 8B). The Elk-1 antibody detected approximately equal amounts of YopN293–Elk in each of the infected HeLa cell lysates. As expected, no phosphorylated YopN293–Elk was detected in lysates from HeLa cells infected with the translocation-defective yopB mutant. The sycN and yscB deletion mutants exhibited reduced phosphorylation of YopN293–Elk compared with the parent strain, suggesting that the SycN–YscB chaperone complex functions to facilitate YopN secretion and subsequent translocation. In contrast, lysates from HeLa cells infected with the tyeA or lcrG deletion mutants showed significantly increased amounts of phosphorylated YopN293–Elk protein compared with the parent strain, suggesting that TyeA and LcrG function, in part, to prevent YopN secretion and subsequent translocation. Complementation of the sycN and lcrG deletion mutants with plasmids pSycN2 and pLcrG2, respectively, restored parental levels of YopN translocation. Complementation of the tyeA and yscB strains with plasmids pTyeA2 and pYscB2 only partially restored parental YopN translocation levels. These data agree with the results presented by Cheng et al. (2001), suggesting that TyeA functions to inhibit entry of YopN into the Yop secretion/translocation pathway. The identification of a role for LcrG in suppressing YopN secretion is novel and represents the first experimental evidence suggesting that TyeA, SycN, YscB and LcrG function together to regulate YopN secretion and subsequent translocation.
Methods used previously to detect the translocation of Yop proteins into eukaryotic cells include immunofluorescence microscopy (Rosqvist et al., 1991), the Bordetella pertussis adenylate cyclase (Cya) reporter system (Sory and Cornelis, 1994) and the digitonin fractionation technique (Lee et al., 1998). Each of these methods has been used to measure the translocation of individual Yops into eukaryotic cells. The technique presented here uses a small (35 amino acid) phosphorylatable peptide tag (the Elk tag) and phosphospecific antibodies to detect Elk-tagged proteins that have been translocated into a eukaryotic cell and phosphorylated by resident protein kinases. No phosphorylation of Elk-tagged YopE occurred within bacterial cells or when secretion (ΔyscV)- or translocation (ΔyopB)-defective Y. pestis mutants were used to infect HeLa cells, confirming that phosphorylation of the tag was dependent upon delivery of the tagged protein into a eukaryotic cell.
The Elk tag and Cya reporter constructs tag the C-terminus of their target protein with a short peptide or a reporter enzyme respectively. One disadvantage of these two protein fusion techniques is that attachment of the tag can disrupt the secretion or function of the tagged protein. For example, in our hands, the YopN293–Elk hybrid protein was defective in its ability to regulate Yop secretion; however, fusion of the Elk tag to YopN (data not shown) or YopE had little effect on the secretion of the tagged proteins (Fig. 2). In contrast, the ≈ 400-amino-acid residue Cya reporter dramatically reduced the secretion of the YopE129–Cya hybrid protein. The Elk tag and Cya fusion techniques also require the participation of a specific eukaryotic protein, Erk2 and calmodulin, respectively, to activate or modify (phosphorylate) the reporter. These requirements provide an intrinsic specificity to these reporter systems that allows the specific detection of Yop translocation in the absence of subcellular fractionation. On the other hand, the digitonin fractionation technique measures the translocation of wild-type proteins, but relies upon the quality of the subcellular fractionation procedure and protease control experiments to identify and quantify the translocated proteins reliably (Lee et al., 1998; Lee and Schneewind, 1999; Nordfelth and Wolf-Watz, 2001). Analysis of infected cell cultures using digitonin fractionation or Cya fusions requires extensive processing and time-consuming assays respectively. In contrast, the Elk tag technique offers a simple method for measuring both Yop expression and translocation that does not require subcellular fractionation or rigorous post-infection assays.
The YopN, TyeA, SycN, YscB and LcrG proteins are necessary to block Yop secretion in the presence of calcium and before contact with a eukaryotic cell. Previous studies have shown that YopE and other effector Yops are translocated, although less efficiently, by Yersinia yopN, tyeA and lcrG deletion mutants (Rosqvist et al., 1994; Fields et al., 1999; Cheng and Schneewind, 2000; DeBord et al., 2001). Most recently, Cheng et al. (2001) demonstrated that Y. enterocolitica yopN, tyeA, sycN and yscB deletion mutants are capable of translocating effector Yops, including YopE, into HeLa cells. Our analysis of YopE129–Elk translocation confirms that Y. pestis yopN, tyeA, sycN and yscB deletion mutants are capable of translocating low levels of YopE into HeLa cells. Apparently, the inability of these mutants to prevent secretion before contact with the surface of host cells and their inability to limit secretion to the point of contact between the two cells significantly reduce the efficiency of the delivery process.
Recent studies by Lee and Schneewind (1999) and Cheng et al. (2001) using digitonin fractionation procedures suggest that Y. enterocolitica translocates YopN into eukaryotic cells. These results contradict data presented by Boland et al. (1996), who failed to demonstrate translocation of a truncated YopN118–Cya hybrid protein. We constructed plasmid pYopN293-Elk to measure translocation of YopN into HeLa cells. YopN293–Elk expressed from yopEJ Y. pestis was translocated and phosphorylated (Fig. 8A). YopN293–Elk phosphorylation was enhanced when a derepressed Y. pestis yop polymutant strain carrying pYopN293-Elk was used to infect HeLa cells, suggesting that YopN293–Elk competes with the effector Yops for the secretion and/or translocation machineries. The high level of YopN293–Elk phosphorylation detected in the lysates from HeLa cells infected with the Y. pestis yopEJTMHA lcrQ strain provided a convenient system to analyse the effect of mutations in sycN, yscB, tyeA or lcrG on YopN translocation. The failure of Boland et al. (1996) to detect translocation of a YopN118–Cya hybrid protein in Y. enterocolitica may also be related to problems associated with the expression of a truncated non-functional YopN protein in the presence of wild-type YopN and the other effector Yops.
Yersinia enterocolitica sycN and yscB mutants exhibit a dramatic decrease in the translocation of YopN into HeLa cells ( Cheng et al., 2001 ). In contrast, translocation of YopN 293 –Elk by the derepressed Y. pestis yop polymutant strain carrying sycN or yscB deletion mutations was only slightly reduced compared with the parent strain. The differences in YopN translocation levels observed between the Y. enterocolitica and Y. pestis sycN and yscB mutants are likely to be directly related to the differences in the strain backgrounds. In fact, the yop polymutant strain used to measure YopN 293 –Elk translocation may partially negate the requirement for the SycN–YscB chaperone complex. Boyd et al. (2000 ) have demonstrated that TTS chaperones promote preferential secretion of their cognate Yop, an effect that is reduced or eliminated in a yop polymutant strain. Thus, the role of the SycN and YscB chaperones is minimized in the yop polymutant background, where the majority of the competition has been eliminated.
In contrast to the results obtained with the sycN and yscB mutants, Y. pestis tyeA and lcrG deletion mutants showed a dramatic increase in translocation of YopN293–Elk compared with the parent strain, indicating that TyeA and LcrG play a role in preventing YopN secretion and subsequent translocation. Cheng et al. (2001) have observed a similar increase in YopN translocation in a Y. enterocolitica tyeA mutant. The increase in YopN293–Elk translocation in the tyeA and lcrG deletion strains was not caused by an increase in Yop expression, including YopN, normally associated with calcium-blind mutants, as these experiments were carried out in an lcrQ background that is constitutively induced for Yop expression.
TyeA and LcrG are both small (<100 amino acids) proteins that specifically bind to a C-terminal region of YopN and LcrV respectively (Nilles et al., 1997; Iriarte et al., 1998). The finding that LcrG, like TyeA, functions to limit YopN translocation is novel and suggests that TyeA and LcrG may regulate Yop secretion indirectly through effects on YopN secretion and subsequent translocation. To date, no direct interaction between LcrG and YopN or TyeA has been described. TyeA and LcrG could function directly or indirectly to stabilize the interaction between YopN and the TTS apparatus that prevents Yop secretion in the presence of calcium and before contact with a eukaryotic cell. Conversely, TyeA and/or LcrG may influence the interaction of the SycN–YscB chaperone with YopN. Interestingly, the inhibitory effects of TyeA and LcrG appear to be specific to YopN, as YopE translocation is uniformly reduced in each of the calcium-blind mutants.
In theory, the Elk tag technique, or variations of this technique, could be used to measure the translocation of proteins into eukaryotic cells by other bacterial pathogens that use type III or type IV delivery systems. In fact, recent communications confirm that translocation of SipA and SptP by Salmonella enterica can be measured using the Elk tag reporter system (J. B. Day and C. A. Lee, personal communication). Furthermore, the successful use of the Elk tag to follow Yop translocation demonstrates that phosphorylation by compartmentalized kinases may represent a practical method of following the movement of proteins within or between cells. It is conceivable that additional tags can be developed that are specifically recognized by protein kinases limited to specific intracellular compartments. Thus, phosphorylation-specific tags may also be used to follow the trafficking of bacterial proteins (or eukaryotic proteins) after their introduction into eukaryotic cells. As our knowledge of eukaryotic protein kinase function and localization increases, and with the dramatic increase in the availability of commercially produced phosphospecific antibodies, the development and utilization of phosphorylation-specific protein-tracking methodologies should also expand.
Bacterial strains and growth conditions
Escherichia coli and Y. pestis strains used in this study are listed in Table 1 . All Y. pestis strains used are Pgm – and thus avirulent from peripheral routes of infection ( Une and Brubaker, 1984 ). Y. pestis KIM5-3001 ( Lindler et al., 1990 ) and KIM8-3002 (pPCP1 – ) ( Williams and Straley, 1998 ) were used as parent strains for the construction of the Y. pestis mutants used in this study. Plasmid pPCP1 encodes the outer membrane plasminogen activator protease (Pla), which has been shown to degrade exported Yops ( Sodeinde et al., 1988 ). These strains and derivatives of these strains were grown routinely in heart infusion broth (HIB) or on tryptose blood agar base plates (TBA; Difco Laboratories) at a temperature of 30°C. During growth experiments, Y. pestis strains were grown in the presence and absence of 2.5 mM calcium chloride in the defined medium TMH ( Goguen et al., 1984 ) as described previously ( Plano et al., 1991 ). E. coli SY327 λpir ( Miller and Mekalanos, 1988 ) was used to propagate derivatives of the suicide vectors pUK4134 and pRE112 ( Edwards et al., 1998 ), whereas E. coli DH5α was used for routine cloning experiments.
Plasmid DNA was isolated according to the method of Kado and Liu (1981), by the Perfect Prep method (5 Prime→3 Prime) or with Qiagen columns. DNA fragments were isolated and purified from agarose gels using the Qiaex DNA purification kit (Qiagen). Electroporation of E. coli and Y. pestis was done as described previously (Perry et al., 1990). The PCR technique was performed using Taq (Sigma) or Pfu turbo (Stratagene) DNA polymerases and 21–30 nucleotide primers according to the manufacturer's instructions. Double-stranded DNA was sequenced by the University of Miami School of Medicine DNA Sequencing Core Facility using a DyeDeoxy Terminator cycle sequencing kit and an ABI model 373A DNA sequencer (Applied Biosystems). Nucleotide sequences were analysed with Intelligenetics computer software
SDS-PAGE and immunoblot procedures
Whole-cell and secreted proteins prepared as described previously (Day and Plano, 1998) were loaded onto 12–15% SDS-PAGE gels and run overnight at 20 mA per gel. The proteins were then transferred to Immobilon-P membranes (Millipore) and blocked with 5% non-fat milk (NFM) in TBS (20 mM Tris, 150 mM NaCl, pH 7.4) for 30 min at room temperature. Primary antibodies were diluted in TBS containing 1% bovine serum albumin (BSA) and 0.05% Tween 20 and incubated with the blots overnight at 4°C. The Elk-1 (#9182) and phosphospecific Elk-1 (#9181) antibodies were purchased from Cell Signaling Technology. Blots were washed three times with TBS containing 0.1% BSA and 0.05% Tween 20 for 10 min. Secondary antibodies (horseradish peroxidase- or alkaline phosphatase-conjugated anti-rabbit IgG or anti-mouse IgG) were diluted in TBS containing 1% BSA and 0.05% Tween 20 and incubated with the blots for 2 h. Blots were then washed as described above and developed with BCIP/NBT solution (Pierce) or the Phototope-HRP detection system (Cell Signaling Technology). Densitometric analyses of chemiluminescent images captured with the ChemiImager 5500 digital imaging system were quantified using alphaeasefc densitometry software (Alpha Innotech).
Construction of Y. pestis deletion mutants
Deletion of a 1.07 kb fragment encoding the 5′ regions of the divergently transcribed yopE and sycE genes was constructed using the PCR–ligation–PCR technique (Ali and Steinkasserer, 1995) and primers 5′-TTTGAGCTCCCTCT CTGCTCAGCATGTTGATGT-3′, 5′-GAATTCATACTCAGCTT GAGATGCTGG-3′, 5′-ACGTCGCCAGTGGCGTTGATCTAA-3′ and 5′-TTTGGTACCCTGCTGACGGTAAAGCAGGAT-3′. The resultant PCR product was digested with SacI and KpnI (underlined) and inserted into SacI- and KpnI-digested pBluescript SK– (Stratagene). An EcoRI-digested kanamycin resistance cassette (pUC-4K; Amersham Biosciences) was inserted into the EcoRI site (bold) flanking the deletion. The resultant plasmid was digested with SacI and KpnI, and the DNA fragment containing the kanamycin resistance cassette and flanking DNA was inserted into suicide vector pRE112 (Edwards et al., 1998), generating plasmid pRE112-yopE/sycE::Km. Plasmid pRE112-yopE/sycE::Km was used to move the resultant mutation into plasmid pCD1 of Y. pestis KIM5-3001 (Pla+) and KIM8-3002 (Pla–), generating Y. pestis strains KIM5-3001.P39 (Pla+ΔyopE) and KIM8-3002.P39 (Pla–; ΔyopE) respectively.
An in frame deletion in yopJ eliminating the coding regions for amino acid residues 43–237 was constructed using the PCR–ligation–PCR technique essentially as described above using primers 5′-TTTGAATTCGAGAGAAAAGTTGCTAGAG CT-3′, 5′-GGATCCATCCGATATATCAGTTTC-3′, 5′-AACCCG CAGGGAGTTGGTACTGTT-3′ and 5′-TTTGGTACCATCTCT GATTTAGCCTGGTAC-3′. The resultant 1.93 kb amplified product was digested with EcoRI and KpnI (underlined) and inserted into pRE112, generating plasmid pRE112-ΔyopJ. Plasmid pRE112-ΔyopJ was used to move the yopJ deletion into Y. pestis KIM5-3001.P39 (Pla+ΔyopE) and KIM8-3002.P39 (Pla–; ΔyopE), generating Y. pestis strains KIM5-3001.P40 (Pla+ΔyopE yopJ) and KIM8-3002.P40 (Pla–; ΔyopE yopJ) respectively.
In frame deletions eliminating the sequences encoding amino acid residues 33–351 of YopH, 1–299 of YopM, 1–202 of YopT, 1–732 of YpkA and 4–80 of LcrQ were constructed using the PCR–ligation–PCR technique and specific oligonucleotide primers (primer sequences available upon request) essentially as described previously. The resultant PCR products were digested with SacI and KpnI and inserted into pRE112, generating plasmids pRE112-ΔyopH, pRE112-ΔyopM, pRE112-ΔyopT, pRE112-ΔypkA and pRE112-ΔlcrQ. The resultant suicide vectors were used sequentially to move the yopH, yopM, yopT, ypkA and lcrQ deletions into the Y. pestis KIM5-3001.P40 (ΔyopE yopJ), generating the derepressed Y. pestis yop polymutant strain KIM5-3001.P48 (ΔyopEJTMHA lcrQ).
An in frame deletion in yopB eliminating the coding sequence for amino acid residues 28–97 of YopB was constructed using the PCR–ligation–PCR technique and primers 5′-TTTGAGCTCATGGATAAAAATTTATATGGT-3′, 5′-GGCGG GCGCTGGTGTCTCGACGTA-3′, 5′-TTTACGCTTGCTTCAC CTGATACA-3′ and 5′-TTTGGTACCGGTCGTGATAATTGGG CTGTC-3. The resultant PCR product was digested with SacI and KpnI (underlined) and inserted into pRE112, generating plasmid pRE112-ΔyopB. The pRE112-ΔyopB suicide vector was used to move the yopB deletion into Y. pestis KIM5-3001.P40 (Pla+ΔyopE yopJ), KIM8-3002.P40 (Pla–; ΔyopE yopJ) and KIM5-3001.P48 (ΔyopEJTMHA lcrQ), generating strains KIM5-3001.P41 (Pla+ΔyopE yopJ yopB), KIM8-3002.P41 (Pla–; ΔyopE yopJ yopB) and KIM5-3001.P49 (ΔyopEJTMHA lcrQ yopB) respectively. Suicide vector pUK4134.3 (Plano and Straley, 1995), which carries an in frame deletion in yscV that eliminates the coding region for amino acid residues 618–644 of YscV, was used to move the yscV deletion into Y. pestis KIM5-3001.P40 (Pla+ΔyopE yopJ ), generating strain KIM5-3001.P42 (Pla+ΔyopE yopJ yscV).
Previously constructed suicide vectors pUK4134.6 (ΔyopN) (Day and Plano, 1998), pUK4134.P2 (ΔtyeA) (Day and Plano, 1998), pUK4134.P3 (ΔsycN) (Day and Plano, 1998) and pUK4134.P1 (ΔyscB) (Jackson et al., 1998) containing in frame deletions eliminating the coding sequence for amino acids 48–197 of YopN, 10–55 of TyeA, 34–65 of SycN and 61–125 of YscB were used to move the yopN, tyeA, sycN and yscB deletions into Y. pestis KIM5-3001.P40 (Pla+ΔyopE yopJ), generating strains KIM5-3001.P44 (Pla+ΔyopE yopJ yopN), KIM5-3001.P45 (Pla+ΔyopE yopJ tyeA), KIM5-3001.P46 (Pla+ΔyopE yopJ sycN) and KIM5-3001.P47 (Pla+ΔyopE yopJ yscB) respectively. Suicide vectors pUK4134.P2 (ΔtyeA), pUK4134.P3 (ΔsycN), pUK4134.P1 (ΔyscB) and pUK4134-ΔlcrG2 (Nilles et al., 1998) were used to move the tyeA, sycN, yscB and lcrG deletions into the Y. pestis yop polymutant strain KIM5-3001.P48 (ΔyopEJTMHA lcrQ), generating strains KIM5-3001.P50 (ΔyopEJTMHA lcrQ tyeA), KIM5-3001.P51 (ΔyopEJTMHA lcrQ sycN), KIM5-3001.P52 (ΔyopEJTMHA lcrQ yscB) and KIM5-3001.P53 (ΔyopEJTMHA lcrQ lcrG) respectively.
Construction of the pYopE129-Elk, pYopE129-(no NLS)Elk and pYopN293-Elk expression plasmids
Plasmid pYopE129-Elk, which expresses a truncated YopE–Elk tag fusion protein consisting of residues 1–129 of YopE fused to the Elk tag (Fig. 1), was constructed by the PCR–ligation–PCR technique. The Elk-1-encoding fragment (amino acids 375–392) was amplified from plasmid pFA2-Elk-1 (Stratagene) using primers 5′-AGCATTCACTTCTGGAGCACC-3′ and 5′-TTTCTCGAGTCACTTGGCCGGGCTACGGGG-3′. The DNA fragment encoding the SV40 large tumour antigen NLS was amplified from plasmid pDELA (Zhang and Lautar, 1996) using primers 5′-TTTATGCATGCGGAATTAATTC CCGAGCCT-3′ and 5′-ACCCAATTCGACCTTTCTCTT-3′. The Elk-1- and NLS-encoding fragments were ligated and reamplified using primers 5′-TTTCTCGAGTCACTTGGCC GGGCTACGGGG-3′ and 5′-TTTATGCATGCGGAATTAATTC CCGAGCCT-3′ carrying NsiI and XhoI sites (underlined) respectively. The resulting NLS-Elk-1 fragment was digested with NsiI and XhoI and inserted in frame into PstI- and XhoI-digested plasmid pYopE (Fig. 1), encoding amino acids 1–129 of YopE and its chaperone SycE, generating plasmid pYopE129-Elk. In addition, the Elk-1-encoding fragment (no NLS) was amplified with primers 5′-TTTCTCGAGAGCAT TCACTTCTGGAGCACC-3′ and 5′-TTTCTCGAGTCACTTG GCCGGGCTACGGGG-3′, digested with PstI and XhoI (underlined) and inserted into PstI- and XhoI-digested plasmid pYopE, generating plasmid pYopE129-(no NLS)Elk.
Plasmid pYopN293-Elk, which expresses a carboxyl-terminal NLS and the phosphorylatable portion of Elk-1 fused to full-length YopN, was constructed by the PCR–ligation–PCR technique. The DNA fragment encoding full-length yopN was amplified using primers 5′-TTTGGTACCGAAAAATAGC CAAGCAGCACT-3′ and 5′-GAAAGGTCGTACGCCATTAGTT TT-3′. The NLS-Elk-1 fragment was amplified as described above with primers 5′-GCGGAATTAATTCCCGAGCCT-3′ and 5′-TTTCTCGAGTCACTTGGCCGGGCTACGGGG-3′. The yopN and NLS-Elk-1 amplification products were ligated together and used as a template for amplification using primers 5′-TTTGGTACCGAAAAATAGCCAAGCAGCACT-3′ and TTTCTCGAGTCACTTGGCCGGGCTACGGGG-3′. The amplified product was digested with KpnI and XhoI (underlined) and inserted into KpnI- and SalI-digested pTRC99a (Amersham), generating pYopN293-Elk.
Complementing plasmids pYopB2, pYscV2, pYopN2, pSycN2, pYscB2, pTyeA2 and pLcrG2 were constructed from PCR products (primer sequences available upon request) and cloned into plasmid pABC184, a derivative of pACYC184 that carries a PvuII fragment of pBluescript SK– inserted into the HincII site of pACYC184 (New England Biolabs). Plasmid pABC184 allows the expression of inserted genes cloned into the pBluescript SK– multiple cloning region via the pBluescript SK–lac promoter. The chloramphenicol-resistant pACYC184 derivative allows complementation experiments to be carried out in Y. pestis strains carrying ampicillin-resistant colE1 plasmids, such as pYopE129-Elk and pYopN293-Elk.
Insertion of yopE129–Elk into pCD1 of Y. pestis KIM8-3002
The pYopE129-Elk DNA sequence encoding YopE129–Elk was moved into plasmid pCD1 of Y. pestis KIM8-3002 by allelic exchange. Briefly, plasmid pCD1 sequences encoding YopE amino acid residues 130–219 and ≈ 0.8 kb of additional flanking DNA were amplified using primers 5′-TTTGTCGAC AAAAATCATGATCAGTTCGCT-3′ and 5′-TTTGGTACCCTG CTGACGGTAAAGCAGGAT-3′. The amplified product was digested with SalI and KpnI (underlined) and inserted into XhoI- and KpnI-digested pYopE129-Elk, generating plasmid pYopE129-Elk2. An ≈ 2.6 kb PvuII fragment of pYopE129-Elk2 carrying yopE129–Elk and flanking DNA sequences was inserted into SmaI-digested pRE112, generating plasmid pRE112-yopE129-Elk. Plasmid pRE112-yopE129-Elk was used to move the yopE129–Elk gene fusion into plasmid pCD1 of Y. pestis KIM8-3002 (Pla–), generating Y. pestis KIM8-3002.P54 (Pla–; pCD1::yopE129-Elk).
Construction of plasmid pYopE129-Cya
Plasmid pYopE129-Cya encodes a YopE–Cya fusion protein consisting of residues 2–400 of the B. pertussis calmodulin-dependent adenylate cyclase (Cya) fused to residues 1–129 of YopE. The DNA sequence encoding residues 2–400 of the B. pertussis cya gene was amplified from B. pertussis (ATCC 8467) chromosomal DNA using primers 5′-TTTCTGCAG CAATCGCATCAGGCTGGTTAC-3′ and 5′-TTTCTCGAGCTA GCGTTCCACTGCGCCCAG-3′. The resulting cya gene fragment was digested with PstI and XhoI (underlined) and inserted in frame into PstI- and XhoI-digested plasmid pYopE (Fig. 1), encoding amino acids 1–129 of YopE and its chaperone SycE, generating plasmid pYopE129-Cya.
Tissue culture infections
HeLa cells were grown routinely at 37°C in 5% CO2 in Dulbecco's modified Eagle medium (DMEM) (Invitrogen) supplemented with l-glutamine, 10% (v/v) fetal calf serum (FCS) and 100 µg ml−1 penicillin and streptomycin. RAW 264.7 macrophages were grown routinely at 37°C in 5% CO2 in RPMI (Invitrogen) supplemented with 10% (v/v) FCS and 100 µg ml−1 penicillin and streptomycin. HeLa cells and RAW 264.7 macrophages were seeded into six-well tissue culture plates containing 2.5 ml of medium per well at a density of 2.5 × 105 cells per well. Cells were allowed to adhere for 24 h. Before infection with Y. pestis, cultured cells were washed twice with DMEM or RPMI without FCS and without antibiotics and incubated with 1 ml of the same medium for 30 min at 37°C in 5% CO2. Cytochalasin D was added to RAW 264.7 macrophages 30 min before infection at a concentration of 5 µg ml−1 in dimethyl sulphoxide (DMSO). Y. pestis strains were grown overnight at 30°C in HIB and diluted the next day to an OD620 of 0.15 in 2 ml of HIB. After 3 h of growth at 30°C in a shaker bath, bacterial cells were washed once with DMEM or RPMI and resuspended in the same media. Cell monolayers were infected with Y. pestis strains at an MOI of 30 for 3.5 h at 37°C in 5% CO2. Y. pestis strains carrying plasmid pYopN293-Elk were induced with IPTG at a final concentration of 0.1 mM at the time of infection. After 3.5 h, the culture supernatants were removed, and the infected adherent cells were lysed by the addition of 100 µl of 1.5× SDS-PAGE lysis buffer containing mammalian cell protease (P-8340) and phosphatase inhibitor (P-2850) cocktails (Sigma). Samples were boiled for 3 min and loaded onto 14% SDS-PAGE gels, transferred to Immobilon-P membranes and probed with Elk-1 (#9182) or phosphospecific Elk-1 (#9181) antibody preparations (Cell Signaling Technology). Release of YopE129–Elk into the media was analysed by separating the non-adherent bacteria (pellet) from the media supernatant by centrifugation at 12 000 g for 10 min at 4°C. Proteins released into the media supernatant fraction were precipitated with 10% TCA. The non-adherent bacterial pellet fraction and media supernatant TCA pellet were resuspended in 100 µl of 1.5× SDS-PAGE lysis buffer and analysed by SDS-PAGE and immunoblot analysis.
We thank Michael W. Jackson for technical assistance. We also thank Yi Tan at New England Biolabs for helpful discussions concerning phosphospecific antibodies and protein kinase assays. This study was supported by Public Health Service grant AI39575.