YopJ of Yersinia pseudotuberculosis is required for the inhibition of macrophage TNF-α production and downregulation of the MAP kinases p38 and JNK

Authors

  • Lance E. Palmer,

    1. Department of Molecular Genetics and Microbiology, School of Medicine, State University of New York at Stony Brook, Stony Brook, NY 11794-5222, USA.
    Search for more papers by this author
  • Silke Hobbie,

    1. Department of Molecular Genetics and Microbiology, School of Medicine, State University of New York at Stony Brook, Stony Brook, NY 11794-5222, USA.
    Search for more papers by this author
  • Jorge E. Galán,

    1. Department of Molecular Genetics and Microbiology, School of Medicine, State University of New York at Stony Brook, Stony Brook, NY 11794-5222, USA.
    Search for more papers by this author
  • James B. Bliska

    1. Department of Molecular Genetics and Microbiology, School of Medicine, State University of New York at Stony Brook, Stony Brook, NY 11794-5222, USA.
    Search for more papers by this author

James B. Bliska E-mail bliska@asterix.bio.sunysb.edu; Tel. (516) 632 8782; Fax (516) 632 9797.

Abstract

Exposure of macrophages to lipopolysaccharide (LPS) leads to production of the pro-inflammatory cytokine, tumour necrosis factor alpha (TNF-α). Previous studies have suggested that pathogenic Yersinia spp. inhibit LPS-mediated production of TNF-α in macrophages, and that one of the Yop proteins secreted by the plasmid-encoded type III pathway is required for this activity. We found that TNF-α production was inhibited when J774A.1 murine macrophages were infected with wild-type Y. pseudotuberculosis but not with an isogenic ysc mutant defective for Yop secretion. We inactivated multiple yop genes to identify which of these factors are required for the inhibition of TNF-α production. A mutant unable to express yopJ was defective for the inhibition of TNF-α production. Production of TNF-α is regulated at the transcriptional and translational levels by several mitogen-activated protein (MAP) kinases. The MAP kinases p38 and JNK underwent sustained activation in macrophages infected with the yopJ mutant. Conversely, p38 and JNK were downregulated in macrophages infected with the wild-type strain. The ability of the yopJ mutant to downregulate p38 and JNK and to inhibit production of TNF-α was restored by the expression of yopJ+in trans. Therefore, YopJ is required for Y. pseudotuberculosis to downregulate MAP kinases and inhibit the production of TNF-α in macrophages.

Introduction

Monocytes and macrophages release a number of pro-inflammatory cytokines, including tumour necrosis factor α (TNF-α) upon exposure to bacterial lipopolysaccharide (LPS) (Adams and Hamilton, 1984; Beutler et al., 1992). TNF-α plays a major role in the inflammatory response by recruiting and activating monocytes and neutrophils (Beutler et al., 1992). In response to TNF-α, phagocytes migrate into infected tissues, adhere to extracellular matrix proteins, release pro-inflammatory and antimicrobial compounds and phagocytose bacteria. This process ultimately leads to the formation of granulomas and the eradication of bacteria.

The mechanism by which LPS stimulates signal transduction and the production of TNF-α in macrophages is not fully understood, but certain key events have been identified recently (Ulevitch and Tobias, 1995). LPS interacts initially with a serum glycoprotein known as LPS-binding protein (LBP). The LBP–LPS complex then binds to CD14, a glycosylphosphatidylinosital-anchored receptor found on the surface of monocytic cells. Although CD14 lacks an intracellular domain, the binding of LPS–LBP to CD14 stimulates tyrosine kinase activation and the phosphorylation of several proteins on tyrosine. Tyrosine kinase inhibitors block LPS-induced TNF-α production, suggesting that tyrosine phosphorylation is required for the LPS-stimulated production of TNF-α (Ulevitch and Tobias, 1995).

Several members of the mitogen-activated protein (MAP) kinase family are activated following LPS stimulation of macrophages (Ulevitch and Tobias, 1995; Hambleton et al., 1996; Sanghera et al., 1996). These include the extracellular signal-regulated kinases 1 and 2 (ERK1/2), the p46 and p54 isoforms of c-Jun N-terminal kinase (JNK) and p38, a mammalian homologue of the yeast HOG1 (Kyriakis and Avruch, 1996). After activation by dual phosphorylation, these kinases translocate to the nucleus and activate transcription factors, such as Elk, c-Jun and ATF2 (Kyriakis and Avruch, 1996; Witmarsh and Davis, 1996). Exposure of macrophages to LPS also activates the transcription factor NF-κB through activation of the IκBα kinase and subsequent phosphorylation and degradation of the NF-κB inhibitor IκBα (Baldwin, 1996). Recent studies have suggested that JNK and the IκBα kinase are controlled by a common upstream regulatory factor (Hirano et al., 1996; Meyer et al., 1996; Lee et al., 1997). The promoter of the TNF-α gene contains several enhancer sequences to which NF-κB and other transcription factors may bind (Beutler et al., 1992). There is also evidence that p38 regulates TNF-α expression at the post-translational level (Lee et al., 1994; Lee and Young, 1996). Therefore, the stimulation of TNF-α synthesis in macrophages by LPS appears to be regulated at both transcriptional and translational levels by multiple components of MAP kinase pathways.

In this study, we examined the ability of the pathogenic bacterium, Yersinia pseudotuberculosis, to inhibit the production of TNF-α by cultured murine macrophages. Y. pseudotuberculosis is an enteric pathogen of humans and animals and is capable of causing acute ileitis, mesenteric lymphadenitis and septicaemia (Brubaker, 1991). Like other pathogenic Yersinia species, Y. pseudotuberculosis harbours a 70–75 kb virulence plasmid that is required for sustained bacterial replication in host lymphoid tissues (Portnoy and Martinez, 1985). This virulence plasmid encodes a type III protein secretion pathway that mediates the translocation of multiple effector proteins, known as Yops, into mammalian cells (Straley et al., 1993; Forsberg et al., 1994; Galán and Bliska, 1996; Cornelis and Wolf-Watz, 1997). The plasmid-encoded type III pathway and the Yops constitute an integrated system that allows extracellular Yersinia to neutralize inflammatory cells. YopH is a protein tyrosine phosphatase that disrupts focal adhesions (Black and Bliska, 1997; Persson et al., 1997). The precise functions of other effector Yops, including YopE, YopJ, YopM and YpkA, remain to be determined (Cornelis and Wolf-Watz, 1997).

Several observations have indicated that the plasmid-encoded type III pathway is required for Yersinia spp. to inhibit the production of TNF-αin vivo. First, several studies have shown that wild-type strains of Yersinia, but not plasmid-defective mutants, suppress granuloma formation in animal lymphoid tissues (Lian et al., 1987; Straley and Cibull, 1989; Simonet et al., 1990). Secondly, there is evidence based on immunohistochemical analysis that wild-type Y. enterocolitica can inhibit TNF-α production in murine Peyer's patches (Beuscher et al., 1995). Similarly, Nakajima and Brubaker (1993) showed that the production of TNF-α in mouse tissues was inhibited by wild-type Y. pestis but not by a plasmid cured derivative. Finally, priming of mice with exogenous TNF-α in combination with interferon γ (IFNγ) was shown to inhibit the replication of wild-type Y. pestis (Nakajima and Brubaker, 1993), while the injection of anti-TNF-α antibodies into mice was found to enhance the virulence of wild-type Y. enterocolitica (Autenrieth et al., 1993). Taken together, these findings suggest that the plasmid-encoded type III pathway functions to inhibit the production of TNF-α and that this process is important for the replication of Yersinia in host tissues.

Two proteins that are components of the plasmid-encoded type III pathway, LcrV (V antigen) and YopB, are thought to play a role in the suppression of TNF-α production during infection (Nakajima and Brubaker, 1993; Beuscher et al., 1995; Nakajima et al., 1995). A role for LcrV in this process was based on the observation that mice were protected from lethal doses of Y. pestis and produced high levels of TNF-α in their tissues when they were passively immunized with anti-LcrV antibodies or actively immunized with a purified protein A–LcrV fusion protein (Nakajima and Brubaker, 1993; Nakajima et al., 1995). In addition, the injection of mice with the protein A–LcrV fusion protein suppressed the production of TNF-α after challenge with plasmid-cured Y. pestis (Nakajima et al., 1995). The precise function(s) of LcrV is not known, although it clearly has a regulatory role, as non-polar lcrV mutants of Y. pestis are defective for the secretion of Yops (Skrzypek and Straley, 1995). More recently, Beuscher et al. (1995) suggested that YopB functions to inhibit the production of TNF-α during infection, based on the finding that passive immunization of mice with anti-YopB antibodies lead to increased production of TNF-α and decreased replication of wild-type Y. enterocolitica in Peyer's patches. In addition, high concentrations of purified YopB suppressed LPS-induced production of TNF-α by cultured murine macrophages in vitro, indicating a direct role for YopB in this process (Beuscher et al., 1995). YopB contains putative transmembrane domains and is required for the translocation of Yop effector proteins across the eukaryotic cell membrane, but it is not itself translocated (Hakansson et al., 1993; 1996; Boland et al., 1996). Therefore, like LcrV, it is unclear how YopB could function directly to inhibit the production of TNF-α by macrophages, unless it also exhibits effector activity.

Although the results of the studies described above are intriguing, there is no genetic evidence to support the idea that LcrV or YopB are necessary or sufficient to inhibit TNF-α production by pathogenic Yersinia spp. To address this issue, we wanted to determine if wild-type Y. pseudotuberculosis was able to inhibit the production of TNF-α in cultured macrophages and to define genetically which effector Yop was necessary for this activity. We demonstrate that Y. pseudotuberculosis can inhibit the production of TNF-α in cultured macrophages and that YopJ is necessary for this activity. In addition, we show that inhibition of TNF-α production by Y. pseudotuberculosis is associated with downregulation of the MAP kinases p38 and JNK in macrophages. As LcrV and YopB are presumably required for the secretion and translocation of YopJ, these results provide a rational explanation for the involvement of these factors in the suppression of TNF-α production.

Results

Inhibition of macrophage TNF-α production by Y. pseudotuberculosis requires protein secretion by the plasmid-encoded type III pathway

To determine if Y. pseudotuberculosis inhibits the production of TNF-α by macrophages, J774A.1 macrophages were infected with a wild-type strain (YP126) or an isogenic ysc mutant (YP71) that is defective for Yop secretion (Table 1). Before infection, the bacteria were grown at 37°C in media low in free Ca2+ to stimulate maximal expression and secretion of Yops (Experimental procedures). Macrophage culture supernatants were collected 60 min after infection, and the production of TNF-α was quantitated in a bioassay (Experimental procedures). Supernatants from macrophages infected with YP126 contained 10 units ml−1 TNF-α after 60 min (Fig. 1B). In contrast, supernatants from macrophages infected with YP71 contained 150 units ml−1 TNF-α after 60 min (Fig. 1B). Similar results were observed when the infection was extended to 120 min: supernatants from macrophages infected with YP126 contained 40 units ml−1 TNF-α, while supernatants from macrophages infected with YP71 contained 400 units ml−1 TNF-α (data not shown). Proteins secreted into the bacterial growth media were analysed by SDS–PAGE to demonstrate that YP71 was defective for Yop secretion (Fig. 1A, lane 2). These results showed that wild-type Y. pseudotuberculosis inhibited the production of TNF-α by cultured macrophages by at least 10-fold over a 2 h time period and that the secretion of Yops by the type III pathway was required for this inhibitory activity. Addition of purified LPS at 100 ng ml−1 to macrophages stimulated TNF-α production at a level equivalent to that seen in cells infected with the ysc mutant (data not shown). In addition, mock-infected macrophages did not produce TNF-α (data not shown). Therefore, it was likely that LPS shed from Y. pseudotuberculosis was responsible for stimulating TNF-α production during infection.

Table 1.  . Strains.Thumbnail image of
Figure 1.

. Inhibition of macrophage TNF-α production requires a functional type III protein secretion and translocation system in Y. pseudotuberculosis. The indicated bacteria (Table 1) were grown at 37°C under low Ca2+ conditions to induce maximal expression and secretion of Yops (Experimental procedures). A. Yops secreted into the growth medium by the indicated strains were precipitated with TCA and analysed by SDS–PAGE. B. Bacteria grown as described were used to infect cultures of J774A.1 macrophages. After 1 h, levels of TNF-α (units ml−1) in tissue culture supernatants were determined in a bioassay (Experimental procedures). Values are the average (± standard deviation) of three independent determinations.

YopB and YopJ are necessary for inhibition of macrophage TNF-α production

To determine which of the Yops that are secreted by the plasmid-encoded type III pathway are required for the inhibition of TNF-α production, we constructed a series of Y. pseudotuberculosis strains that are defective for the production of one or more of these proteins (Table 1; Experimental procedures). We first tested a strain that contained an in frame, non-polar deletion in the translocation factor yopB (YP18; Table 1). Analysis by SDS–PAGE confirmed that YP18 was unable to secrete full-length YopB, while the secretion of all other Yops was unaffected (Fig. 1A, lane 3). Macrophages infected with YP18 consistently produced twofold more TNF-α compared with macrophages infected with YP71 (Fig. 1B), indicating that YopB was required for the inhibition of TNF-α production. In this context, YopB could be acting either directly, as proposed by Beuscher et al. (1995), or indirectly by mediating the translocation of another Yop (Boland et al., 1996; Hakansson et al., 1996).

To discriminate between these two possibilities, we next examined strains that were defective for the expression of yopE (YP6), yopH (YP15) or both yopE and yopH (YP17) (Table 1). The inability of these mutants to secrete YopE and/or YopH was confirmed by SDS–PAGE (data not shown and Fig. 2A, lane 3). Both single mutants as well as the double yopH yopE mutant suppressed macrophage production of TNF-α as well as the wild-type strain (data not shown and Fig. 2B). Thus, neither YopH nor YopE alone or in combination were necessary for this inhibitory activity. Next, a series of strains derived from YP17 that contained frameshift mutations in yopK (YP22), yopM (YP23) or ypkA (YP24) were tested (Table 1; Experimental procedures). As shown in 2Fig. 2A, the secretion profiles of these strains showed that YP22 was unable to secrete YopK (lane 4), YP23 was unable to secrete YopM (lane 5) and YP24 was unable to secrete YpkA (lane 6). Of these mutants, only YP24 showed any significant defect in the suppression of macrophage TNF-α production. Approximately one third as much TNF-α was produced in YP24-infected cells compared with YP71-infected cells (Fig. 2B). Careful examination of the profile of Yops secreted by YP24 revealed that the frameshift mutation introduced into ypkA had a partial polar effect on the secretion of YopJ (Fig. 2A, lane 6), which is encoded downstream of ypkA in the same operon (Galyov et al., 1994). To determine if the partial phenotype of YP24 was caused by reduced expression of yopJ, we analysed the parental strain of YP24 (YP25), which contains the suicide plasmid pLP8 integrated into ypkA (Table 1). The integrated suicide plasmid behaves as a polar insertion downstream of ypkA and therefore prevents the expression of yopJ. As shown in 3Fig. 3A, YP25 secreted YpkA but was defective for the secretion of YopJ (lane 4). YP25 was also defective for the inhibition of macrophage TNF-α production (Fig. 3B), suggesting that YopJ was necessary for this inhibitory activity.

Figure 2.

. YopH, YopE, YopK, YopM and YpkA are not required for the inhibition of macrophage TNF-α production. The indicated bacteria (Table 1) were analysed for their ability to secrete Yops (A) or inhibit TNF-α production (B), as described in the legend to Fig. 1. Note that YP24 (lane 6) secretes lower amounts of YopJ compared with its parental strain YP17 (lane 3). In YP17 and its derivatives (lanes 3–6), the insertion of a kanamycin resistance cassette in yopE results in the secretion of a truncated YopE protein, designated YopE1-131.

Figure 3.

. A mutant unable to express yopJ is defective for the inhibition of macrophage TNF-α production. The indicated bacteria (Table 1) were analysed for their ability to secrete Yops (A) or inhibit TNF-α production (B), as described in the legend to Fig. 1. The YpkA protein secreted by YP25 (lane 4) was engineered for the purpose of adding an epitope tag and therefore migrates more slowly because of the fusion of additional residues to its C-terminus.

Complementation analysis

To confirm that YopJ was necessary for the inhibition of TNF-α production, we complemented YP25 by expressing yopJ+in trans. The polymerase chain reaction (PCR) was used to amplify a DNA fragment corresponding to the published sequence of the 264-amino-acid yopJ open reading frame (ORF) (Krause et al., 1991; Galyov et al., 1994) (Experimental procedures). Sequence analysis of this DNA fragment revealed the presence of an error in the published sequence (Krause et al., 1991). This error corresponds to the deletion of a G residue after nucleotide 723. Deletion of this G residue results in a frameshift in the predicted ORF, leading to premature termination at a UAG after codon 264. The wild-type yopJ ORF is predicted to encode a protein of 288 amino acids. Therefore, a larger DNA fragment containing the entire wild-type yopJ ORF was amplified by PCR. This fragment was inserted into an IPTG-inducible expression vector (Experimental procedures), and the resulting plasmid (pLP11) or the empty vector (pMMB67EH) was introduced into YP25. YP25(pLP11) and YP25(pMMB67EH) were grown under low Ca2+ conditions in the presence or absence of IPTG and examined for their ability to secrete YopJ by SDS–PAGE. YP25(pLP11) secreted full-length YopJ when grown in the presence of IPTG but not in the absence of IPTG (Fig. 4A, lanes 3 and 4). YP25(pLP11) and YP25(pMMB67EH) grown in the presence or absence of IPTG were then examined for their ability to inhibit TNF-α production by macrophages. YP25(pLP11) grown in the presence of IPTG inhibited macrophage production of TNF-α (Fig. 4B), confirming that YopJ was required for this activity.

Figure 4.

. YopJ is required for the inhibition of macrophage TNF-α production. The indicated bacteria (Table 1) were analysed for their ability to secrete Yops (A) or inhibit TNF-α production (B). Bacteria were grown as described in the legend to Fig. 1 except that IPTG was added where indicated (Experimental procedures).

Yop secretion is required for downregulation of p38

As MAP kinases play an important role in regulating TNF-α biosynthesis in response to LPS stimulation (Lee and Young, 1996), we wanted to determine if wild-type Y. pseudotuberculosis inhibited MAP kinase activity. We focused initially on p38, as this kinase appears to regulate TNF-α biosynthesis at the translational level (Lee and Young, 1996). Macrophages were infected with YP126 or YP71 for 15 or 45 min and then lysed. p38 was immunoprecipitated from the lysates and incubated in an in vitro kinase reaction with [γ-32P]-ATP and a purified transcription factor substrate (GST-ATF2). The kinase reactions were then analysed by SDS–PAGE and autoradiography to quantitate levels of GST-ATF2 phosphorylation (Experimental procedures). As expected, p38 activity was not detected in mock-infected cells (Fig. 5, lane 1). There was an initial stimulation of p38 activity at 15 min and a subsequent decline in activity by 45 min when macrophages were infected with YP126 (Fig. 5, lanes 2 and 3). Quantitation by densitometry indicated that p38 activity decreased severalfold between 15 and 45 min (Fig. 5, lanes 2 and 3). In contrast, p38 activity was sustained in macrophages infected with YP71 (Fig. 5, lanes 4 and 5). Thus, we observed transient activation of p38 when macrophages were infected with wild-type bacteria and sustained activation of p38 when macrophages were infected with bacteria unable to secrete Yops. As shown in 6Fig. 6A lanes 1–5, similar results were obtained when the activation of p38 was measured by immunoblot analysis of macrophage lysates using a phosphospecific antibody that specifically recognizes the dual-phosphorylated form of p38 (Experimental procedures). A duplicate filter probed with a standard anti-p38 antibody demonstrated that equivalent amounts of p38 were loaded in each lane (Fig. 6B, lanes 1–5). These results indicated that p38 was activated by bacterial infection, most probably as a result of LPS stimulation. Subsequently, wild-type bacteria were able to downregulate p38 through a process that required Yop secretion.

Figure 5.

. Downregulation of p38 activity by Y. pseudotuberculosis requires Yop secretion. Macrophages were left uninfected or infected with YP126 or YP71 pregrown under low Ca2+ conditions. After 15 or 45 min, the macrophages were lysed and p38 activity was assayed in an in vitro kinase reaction using GST–ATF2 as a substrate (Experimental procedures). An autoradiographic exposure of the SDS–polyacrylamide gel is shown. The relative levels of p38 activity (fold activation) calculated from densitometric scans of the autoradiograph are indicated below each lane. Results are representative of three independent experiments.

Figure 6.

. YopJ is required for downregulation of p38 and JNK. The indicated bacteria (Table 1) were analysed for their ability to downregulate p38 or JNK by immunoblotting. Bacteria were grown as described in the legend to Fig. 4. Macrophages were left uninfected or infected with bacteria for the indicated times and then lysed. Lysates were analysed by immunoblotting with phosphospecific anti-p38 (A) or anti-JNK (C) antibodies, as described in Experimental procedures. Duplicate filters were probed with standard anti-p38 (B) or anti-JNK antibodies (D) to control for loading. In (D), the standard anti-JNK antibody specifically recognizes the p46 and p54 isoforms of JNK, but also cross-reacts with a protein doublet that migrates between the p46 and p54 isoforms.

YopJ is required for downregulation of p38

As immunoblotting with phosphospecific antibodies provides a convenient method for measuring p38 activation in a large number of samples, we used this approach to identify which Yop was required for the downregulation of p38. p38 was downregulated when macrophages were infected with YP17 for 45 min, indicating that neither YopH nor YopE was required for this activity (Fig. 6A, lane 7). However, p38 was not downregulated when macrophages were infected with YP25 for 45 min, indicating that YopJ was necessary for this activity. Macrophages were infected with YP25(pMMB67EH) or YP25(pLP11) grown in the presence or absence of IPTG. Downregulation of p38 was only observed when macrophages were infected with YP25(pLP11) grown in the presence of IPTG (Fig. 6A, lane 17). Therefore, YopJ was necessary for the downregulation of p38.

YopJ is required for downregulation of JNK

To determine if JNK was similarly downregulated in a YopJ-dependent manner by Y. pseudotuberculosis, the activation level of this kinase was measured by immunoblot analysis of cell lysates using a phosphospecific anti-JNK antibody. The results obtained for JNK were essentially identical to those obtained for p38: JNK was transiently activated and then downregulated in macrophages with YP126 or YP17, while JNK showed sustained activation in macrophages infected with YP71 or YP25 (Fig. 6C, lanes 1–9). In addition, we were able to complement the defect in YP25 by growing YP25(pLP11) in the presence of IPTG (Fig. 6C, lane 17). A control anti-JNK immunoblot demonstrated that equivalent amounts of JNK were loaded in each lane (Fig. 6D, lanes 10–17). Therefore, YopJ was required for Y. pseudotuberculosis to downregulate JNK.

YopJ is not redundant to YopE or YopH

To demonstrate that YopJ was required for the suppression of TNF production even in the presence of YopH and YopE, an additional mutant was constructed by precise deletion of the yopJ gene in an otherwise wild-type strain. The yopJ + expression vector, pLP11, or the empty vector, pMMB67EH, was introduced into this mutant (YP26; Table 1), and the resulting strains were tested for the ability to secrete YopJ when grown in the presence of IPTG. As shown in 7Fig. 7A, YP26(pMMB67EH) was defective for the secretion of YopJ (lane 3), while YP26(pLP11) secreted YopJ at native levels when grown in the presence of IPTG (lane 4). In addition, YP26(pLP11) inhibited TNF production when grown in the presence of IPTG (Fig. 7B). These results demonstrate that the ability of YopJ to inhibit TNF production is not redundant to YopE or YopH.

Figure 7.

. The ability of YopJ to inhibit TNF production is not redundant to YopE or YopH. The indicated bacteria (Table 1) were analysed for their ability to secrete Yops (A) or inhibit TNF-α production (B), as described in the legend to Fig. 4.

Discussion

TNF-α is a potent cytokine that plays a critical role in the inflammatory response against a variety of pathogens (Beutler et al., 1992). In this paper, we have demonstrated that wild-type Y. pseudotuberculosis is capable of inhibiting the production of TNF-α by murine macrophage-like J774A.1 cells. Our results also provide direct genetic evidence that YopJ, a protein secreted by the plasmid-encoded type III pathway in Yersinia, is required for the inhibition of TNF-α production.

Using a bioassay to measure the release of TNF-α by macrophages, we showed that macrophages infected with wild-type Y. pseudotuberculosis produced 10-fold less TNF-α compared with macrophages infected with a mutant defective in the secretion of Yops. We believe that LPS shed from the surface of Y. pseudotuberculosis is largely responsible for stimulating TNF-α production during these infection experiments, although we cannot rule out the possibility that other bacterial factors contribute to this effect. By inactivating several different yop genes that encode potential effector molecules of the plasmid-encoded type III pathway, we obtained evidence that YopJ is necessary for the inhibition of TNF-α production. Our results also rule out the possibility that either LcrV or YopB are sufficient to inhibit the production of TNF-α in cultured macrophages, as a mutant defective for the expression of yopJ secreted native levels of both LcrV and YopB but was completely defective for the inhibition of TNF-α production. YopB and LcrV most probably play an indirect role in this process by mediating the secretion and or translocation of another Yop (e.g. YopJ). This idea is consistent with data showing that neutralizing antibodies directed against LcrV or YopB prevent the inhibition of TNF-α production by Yersinia spp. in host tissues (Nakajima and Brubaker, 1993; Beuscher et al., 1995). However, we are unable to explain the following observations: (i) that the injection of mice with a protein A–LcrV fusion peptide suppressed the production of TNF-α after challenge with plasmid-cured Y. pestis (Nakajima et al., 1995); and (ii) that purified YopB inhibited the LPS-stimulated production of TNF-α by macrophages in vitro (Beuscher et al., 1995). In the latter case, it is conceivable that, at the high concentrations of purified protein used (10–40 μg ml−1) (Beuscher et al., 1995), YopB may inhibit TNF-α production in a non-specific fashion, possibly by disrupting the plasma membrane of cultured macrophages. The recent demonstration that purified YopB has membrane-disrupting activity in vitro (Hakansson et al., 1996) would be consistent with this hypothesis.

To examine the mechanism by which Y. pseudotuberculosis inhibits the production of TNF-α in cultured macrophages, we analysed the effect of bacterial infection on two of the MAP kinases, p38 and JNK. Initially, we focused on p38, because this kinase appears to regulate TNF-α expression at the translational level (Lee et al., 1994; Lee and Young, 1996). Using in vitro kinase reactions to assay p38 activity directly, we detected transient p38 activity in macrophages infected with wild-type Y. pseudotuberculosis. Conversely, sustained p38 activity was detected in macrophages infected with a Yop secretion mutant. Thus, p38 activity was stimulated in response to infection with either strain. Subsequently, wild-type bacteria were able to inhibit p38 activity through a mechanism that required Yop secretion. The apparent time lag in the inhibitory effect may reflect the time required for the secretion and translocation of Yops, which takes several minutes to occur (Bliska and Black, 1995). Immunoblotting with phosphospecific antibodies to p38 demonstrated that the decrease in p38 activity in wild-type infected cells was caused by downregulation of p38 by dephosphorylation. Lysates of macrophages infected with various yop mutants were analysed by phosphospecific immunoblotting to identify the Yop required for downregulation of p38. These experiments demonstrated that YopJ was required for downregulation of p38. Analysis of JNK activation by the same technique yielded essentially identical results. Therefore, Y. pseudotuberculosis is able to downregulate both p38 and JNK using a mechanism that requires YopJ. Our evidence is consistent with the idea that YopJ performs its essential function after it is delivered into macrophages via the type III pathway (Fig. 8). However, as it has not been demonstrated that YopJ is translocated, it is possible that YopJ is required for the function of another translocated protein. Transfection or microinjection experiments will be required to demonstrate that YopJ is sufficient to downregulate p38 and JNK and inhibit TNF production.

Figure 8.

. A model for the inhibition of macrophage TNF-α production by Y. pseudotuberculosis. Bacterial LPS stimulates activation of p38 and JNK MAP kinase signalling cascades, leading to transcriptional and translational upregulation of TNF-α production. Y. pseudotuberculosis inhibits TNF-α production by downregulating p38 and JNK. Downregulation of p38 and JNK requires a functional type III pathway and YopJ. In this model, YopJ is translocated into macrophages via the type III pathway and directly inhibits a step of the pathway upstream of both p38 and JNK.

Regulation of TNF-α production in response to LPS stimulation in macrophages is complex and appears to involve multiple components of MAP kinase pathways acting at several regulatory levels. There is evidence that p38 regulates the production of TNF-α at the translational level (Lee and Young, 1996). LPS also stimulates TNF-α expression at the transcriptional level (Beutler et al., 1992), and the promoter region of the TNF-α gene contains several enhancer regions to which transcription factors, such as NF-κB and AP-1, may bind (Shakhov et al., 1990; Beutler et al., 1992). JNK has been implicated in the regulation of AP-1 (Karin, 1995; Witmarsh and Davis, 1996) and, more recently, a direct link between MEKK1 (an upstream regulator of JNK) and the activation of NF-κB has been established (Hirano et al., 1996; Meyer et al., 1996; Lee et al., 1997). From a pathogenic standpoint, it would make sense to attack this complex cascade at an early step that is controlled by one key regulatory molecule. Thus, we favour the idea that YopJ is required for Y. pseudotuberculosis to inhibit an early step in the MAP kinase pathway (Fig. 8). There are several types of regulatory molecules that could be targeted by Y. pseudotuberculosis in order to inhibit p38 and JNK activation. These kinases are controlled by a signalling pathway that is distinct from the classical ERK1/2 MAP kinase regulatory cascade. Common upstream regulators of p38 and JNK may include the PAK-related kinases and the Rho family GTPases, Cdc42 and Rac1 (Vojtek and Cooper, 1995). Thus, it is conceivable that a block placed at the level of Cdc42, Rac or Pak could account for the downregulation activity observed here.

Several other studies have implicated type III secretion pathways of pathogenic bacteria in the modulation of cytokine production by host cells. While this manuscript was in preparation, Ruckdeschel et al. (1997) reported that Y. enterocolitica inhibits the production of TNF-α by J774A.1 cells through a process that requires Yop secretion. They also reported that ERK1/2, p38 and JNK activities were inhibited by wild-type Y. enterocolitica. However, their experiments measuring the activity of ERK1/2 or p38 were inconclusive, as crude cell extracts were used in kinase reactions with substrates that are not specific to these kinases (Ruckdeschel et al., 1997). Schulte et al. (1996) reported that the ability of Y. enterocolitica to inhibit the production of interleukin 8 (IL-8) by cultured epithelial cells required a functional plasmid-encoded type III pathway. Interestingly, although YopB was required for this activity, a strain apparently defective for the expression of yopJ inhibited IL-8 production as well as wild type (Schulte et al., 1996). Therefore, inhibition of IL-8 production in epithelial cells may occur by a mechanism distinct from that reported here. Further clarification of this issue awaits the identification of the translocated effector protein that is required for the inhibition of IL-8 production. More recently, Hobbie et al. (1997) have shown that the ability of Salmonella typhimurium to stimulate the production of IL-8 in epithelial cells requires the invasion-associated type III pathway. Stimulation of IL-8 production by S. typhimurium involves the activation of the MAP kinases, ERK1/2, p38 and JNK, and the transcription factors, NF-κB and AP-1 (Hobbie et al., 1997). Thus, MAP kinase pathways appear to represent a common target of bacterial type III secretion pathways, although the biological outcome of the interaction may differ depending upon the organism's pathogenic strategy, the type of host cell encountered and the precise nature of the effector proteins that are translocated.

The role of YopJ as a pathogenicity factor for Yersinia has been examined by several groups. Straley and Bowmer (1986) measured the LD50 of a Y. pestis yopJ mutant injected intravenously into Swiss mice and found that the LD50 values for the yopJ mutant and the wild type were similar. Therefore, YopJ clearly has no role in virulence under these specific conditions. More recently, Galyov et al. (1994) challenged BALB/c mice orally with large, undefined doses of wild-type or yopJ mutant Y. pseudotuberculosis strains and reported that all mice died within 6 days of infection. The ability of the yopJ mutant to colonize Peyer's patches and spleens on different days after oral infection was also measured, and no significant defect was reported. However, given that the bacterial dose was unknown and only two mice were used per time point, it is difficult to make a definitive conclusion from these studies (Galyov et al., 1994).

While this manuscript was in preparation, Monack et al. (1997) reported that YopJ is required for Y. pseudotuberculosis to induce programmed cell death in macrophages. However, significant killing (> 50%) required an 8 h infection (Monack et al., 1997). As only 3% of macrophages were killed after a 2 h infection (Monack et al., 1997), we do not believe that cell death is responsible for the inhibition of TNF-α production that we observe at 1 h. Hardt and Galán (1997) also reported that YopJ shares significant sequence similarity with two other bacterial proteins, AvrA of S. typhimurium and AvrRxv of the plant pathogen Xanthomonas campestris pv. vesicatoria. AvrA is translocated into host cells by the invasion-associated type III pathway in S. typhimurium, although its function is not yet known (Hardt and Galán, 1997). AvrRxv is responsible for inducing a hypersensitive response in certain plant species (Whalen et al., 1993). The hypersensitive response is characterized by rapid, localized tissue necrosis that drastically limits the growth of the pathogen (Staskawicz et al., 1995; Leach and White, 1996). It is possible that YopJ, AvrA and AvrRxv execute similar effector functions during the interaction of pathogenic bacteria and host cells. In view of the findings presented here, YopJ may induce apoptosis in host cells by interfering with signals that inhibit cell death, rather than by activating a programmed cell death pathway.

Experimental procedures

Materials

The following oligonucleotides were obtained from the Stony Brook Oligonucleotide Synthesis Facility or Life Technologies: A1 5′-GGGCGGCCGCATATGAAAAGCGTGAAAATCATGGGAACT-3′; A2 5′-CCGCGGCCGCCCGGGACATCCATTCCCGCTCCAACCGGT-3′; B1 5′-GGGATCCCATATGAGT-GCGTTGATAACCCATGAC-3′; B2 5′-GTTAGCACCGAGT-TTCTTTGATGCGATGCCGGATTTCTT-3′; B3 5′-GGCAT-CGCATCAAAGAAACTCGGTGCTAACACCGCAAG-3′; B4 5′-CGGATTCGAATTCTTAAACAGTATGGGGTCTGCCGG-3′; K1 5′-GGGATCCCAATTGCATATGTTTATTAAAGATACTTATAACATGCGT-3′; K2 5′-CGGATCCGAATTCTCAT-CCCATAATACATTCTTGATCG-3′; J1 5′-CCGGATCCATT-TGGCAATTGCTTAACAATAATTATTT-3′; J3 5′-GGGAA-TTCATATGATCGGACCAATATCACAAATAAAT-3′; J4 5′-GGGAATTCCTACAATGGATTTATTGTTATCAAATCT-3′; J6 5′-CCGGATCCGATATTCAGCGATTGATCAGATCGC-3′; J10 5′-CCAAAATACAAGATCTTTATTTATCCTTATTCAGGGAATTAACA-3′; J11 5′-AAGGATAAATAAAGATC-TTGTATTTTGGAAATCTTGCTCCAGTA-3′; J12 5′-CCG-GATCCGAGCTGTATTTCGCTGAATACTACC-3′; M1 5′-CGGATCCCAATTGCATATGTTCATAAATCCAAGAAATGTATCTAATA-3′; M2 5′-CGGATCCGAATTCCTACTCAAATACATCATCTTCAAGTT-3′.

The TA cloning vector pCRII was purchased from Invitrogen. L929 mouse fibroblast cells were obtained from the Stony Brook Tissue Culture Facility. Recombinant mouse TNF-α was obtained from the National Institute for Biological Standards and Control. Actinomycin D was purchased from Sigma. Rabbit antibody to p38 (Derijard et al., 1995) and the c-Jun(1–79) (Hibi et al., 1993) and ATF2(1–109) (Gupta et al., 1995) glutathione S-transferase (GST) fusion proteins were a gift from R. Davis (University of Massachussetts, USA). GST–c-Jun and GST–ATF2 were purified as described previously (Guan and Dixon, 1991). Rabbit polyclonal antibody to JNK was purchased from Santa Cruz Biotechnology. Gluthathione agarose and protein A Sepharose were purchased from Pharmacia. Phosphospecific anti-p38, anti-JNK and anti-ERK1/2 antibodies and standard anti-p38, anti-JNK or anti-ERK1/2 antibodies were purchased from New England Biolabs. Chemiluminescence (ECL) reagents for immunoblotting were purchased from New England Nuclear.

Bacterial strains and growth conditions

The Y. pseudotuberculosis strains used in this study are shown in Table 1. All strains are derived from YP126, a wild-type serogroup III Y. pseudotuberculosis strain, or its isogenic virulence plasmid-cured derivative, YP137. Bacteria were cultivated routinely at 26°C in Luria broth (LB) or on LB agar plates containing appropriate antibiotics. For infection assays, overnight cultures of bacteria were diluted into LB containing 20 mM sodium oxalate and 20 mM MgCl2 to an OD600 of 0.1. Bacteria were grown with shaking at 26°C for 1 h and then shifted to 37°C for 2–3 h to induce maximal Yop expression. In some cases, IPTG was added to a final concentration of 0.1 mM at the time of temperature shift. Bacteria were pelleted and resuspended to an OD600 of 1.0 in Hanks' balanced salt solution (HBSS) and used to infect macrophages. Supernatants of the bacterial cultures (1 ml) were processed by the addition of trichloroacetic acid to a final concentration of 10% to precipitate Yops. After an overnight incubation at 4°C, precipitated proteins were collected by centrifugation, washed with cold acetone and dried under vacuum. Proteins were solubilized by boiling in 1 × Laemmli sample buffer containing 10 mM dithiothreitol (DTT) and separated on 10% SDS–polyacrylamide gels. Gels were stained with Coomassie brilliant blue.

Construction of yop mutants

To construct YP17 (yopE yopH mutant), the virulence plasmid from IP17 (Black and Bliska, 1997) was purified and introduced into YP137 by electroporation, followed by selection for resistance to kanamycin and chloramphenicol. To inactivate yopK, yopM and ypkA, the corresponding ORFs were amplified by PCR using the following pairs of tailed primers: K1 and K2 for yopK, M1 and M2 for yopM and A1 and A2 for ypkA. The amplified fragments were inserted into the BamHI site (for yopK and yopM ) or the NotI site (for ypkA) of pBluescript-II (SK−) (Stratagene). The resulting plasmids, pJB1(yopK+), pJB2(yopM+) and pLP1 (ypkA+), were linearized by partial digestion with the following restriction enzymes: Afl III for pJB1, BsaHI for pJB2 and BstEII for pLP1. This was followed by treatment with Klenow fragment. The blunt-ended DNAs were recircularized by ligation to create frameshift mutations in the respective yop ORFs. The resulting plasmids were designated pLP2 (ypkAΔBstEII), pLP3 (yopKΔAflII) and pLP4 (yopMΔBsaHI). DNA fragments containing the mutated ORFs were excised from the pBluescript plasmids and inserted into the BamHI site (for yopKΔAflII and yopMΔBsaHI) or the NotI site (for ypkAΔBstEII) of the tetracycline-resistant suicide vector pSB890 (W.-D. Hardt and J. Galán, unpublished data), resulting in pLP6 (yopKΔAflII), pLP7 (yopMΔBsaHI) and pLP8 (ypkAΔBstEII) respectively. A suicide plasmid containing an in frame deletion in yopB (nucleotides 496–774) was constructed by recombinant PCR (Higuchi, 1990), using two pairs of tailed primers and two rounds of PCR. In the first round, primers B1 and B2 were used to amplify nucleotides 1–495 of yopB (Hakansson et al., 1993), and primers B3 and B4 were used to amplify nucleotides 775–1206 of yopB. The PCR products from the first round were gel purified and subjected to a second round of PCR using B1 and B4 (Higuchi, 1990). The resulting PCR product was inserted into the TA cloning vector pCRII to create pJB3 (yopBΔ496–774). The DNA insert was then excised by BamHI digestion and introduced into the BamHI site of pSB890, yielding pJB4 (yopBΔ496–774). The suicide plasmid pLP13 containing a precise deletion of yopJ (nucleotides 1–867) was constructed by recombinant PCR in a similar fashion, except that the primer pairs used in the first round were J1 with J10 and J11 with J12, while primers J1 and J12 were used in the second round. All pSB890-derived plasmids were mated from Escherichia coli SM10λpir into YP17 or YP126, and merodiploids were selected on LB plates containing tetracycline and chloramphenicol (for YP17) or minimal media plates (M6) containing tetracycline (for YP126). Loss of the suicide vector was achieved by growing merodiploids in LB in the absence of selection for 24 h, followed by plating on LB plates lacking NaCl and containing 5% sucrose to select against the sacB gene in pSB890. Isolated colonies were tested for loss of tetracycline resistance. To confirm the presence of the mutant allele, PCR was used to amplify the relevant sequences from tetracycline-sensitive colonies, and the resulting products were subjected to restriction analysis.

Cloning and expression of yopJ

A nucleotide sequence corresponding to the published sequence of the 264 codon yopJ ORF (Krause et al., 1991; Galyov et al., 1994) was amplified by PCR using the tailed primers J3 and J4. The resulting PCR product was introduced into pCRII, yielding pLP9. The DNA inserted into this plasmid was subjected to sequencing using an ABI model 373 automated sequencer and commercially available primers. Analysis of the resulting sequence revealed the presence of an additional G residue after nucleotide 723 that is absent in the published yopJ sequence (Krause et al., 1991). This frameshift error in the published sequence is predicted to result in premature protein termination after codon 264, while the wild-type yopJ ORF is predicted to encode a 288-amino-acid protein. A second DNA fragment, containing the wild-type yopJ ORF and 429 nucleotides of 3′ flanking sequence, was generated using PCR and primers J3 and J6. This fragment was inserted into pCRII, yielding pLP10, and sequenced to confirm its structure. An NdeI to EcoRI fragment containing the yopJ + ORF was isolated from pLP10 and inserted into a derivative of pMMB67EH (Furste et al., 1986), containing a ptac promoter followed by a strong phage T7 ribosome binding site (Bliska and Black, 1995). The resulting plasmids, pLP11 (yopJ +), and the vector, pMMB67EH, were mated from SM10λpir into YP25 or YP26 by selecting for resistance to ampicillin and chloramphenicol.

Cell culture and TNF-α bioassays

J774A.1 murine macrophage-like cells were maintained as continuous cultures as described previously (Bliska et al., 1991). L929 cells were grown in DMEM containing 5% FBS. L929 cells were detached by treatment with trypsin for 5 min at 37°C, resuspended to 5 × 106 cells ml−1 in 90% FBS, 10% dimethyl sulphoxide (DMSO) and frozen at −70°C. Macrophages or L929 cells were seeded into multiwell tissue culture plates on the day before the assay. Macrophages were detached from tissue culture flasks as described previously (Bliska and Black, 1995), resuspended in DMEM containing 10% FBS, and 2 × 105 cells were seeded in 1 ml into the wells of a 24-well plate. L929 cells were thawed, resuspended in DMEM containing 10% FBS, and 5 × 104 cells were seeded in 0.1 ml into the wells of a 96-well plate. Immediately before infection, the media overlaying the macrophages was replaced with 1 ml of DMEM containing 10% FBS. In some cases, IPTG was added to a final concentration of 0.1 mM. Aliquots of 20 μl of bacterial suspension (2 × 107 bacteria) were added to each well, and the plates were placed at 37°C in a CO2 incubator. Samples of the tissue culture media (macrophage supernatant) were removed at various time points, sterilized by passage through 0.2 μm filters and serially diluted twofold in DMEM containing 5% FBS. The media overlaying the L929 cells was aspirated and replaced with 0.1 ml of DMEM containing 5% FBS and 2 μg ml−1 actinomycin D and 0.1 ml of diluted macrophage supernatant or 0.1 ml of diluted purified recombinant mouse TNF-α. The plates were incubated 20 h at 37°C in a CO2 incubator. After aspiration of the media, the wells were rinsed once with PBS, and 50 μl of 0.5% crystal violet in 20% methanol was added to each well. After 5 min at room temperature, the wells were rinsed thoroughly with dH2O. Aliquots of 0.1 ml of 1% SDS were added to each well to solubilize the crystal violet, and the plates were incubated for 30 min at 37°C. The optical density (OD550) of the samples was measured in a microplate reader. Units of TNF-α produced during infections were calculated from the linear range of the standard curve obtained with purified TNF-α.

P38 kinase assay

p38 kinase activity was measured as described previously (Hibi et al., 1993; Derijard et al., 1995; Hobbie et al., 1997) with the following modifications. Macrophages were prepared and infected as for the TNF-α bioassays except that 2 × 106 cells in 6 cm dishes were infected with 1 × 108 bacteria. After 15 or 45 min of infection, the cells were washed once with ice-cold PBS containing 1 mM Na3VO4. Washed cells were gently lysed in 0.5 ml of lysis buffer [10 mM HEPES, pH 7.5, 150 mM NaCl, 10% glycerol, 1 mM Na3VO4, 1% Triton X-100, 1% aprotinin, 1 μM phenylmethylsulphonyl fluoride (PMSF)] at 4°C for 3 min. Cytoplasmic fractions were collected by tilting the dishes, leaving the cytoskeletal and nuclear components attached to the plastic. Lysates were clarified by centrifugation for 10 min at 4°C at 12 000 r.p.m. in a microfuge. p38 or JNK were immunoprecipitated from precleared cytoplasmic fractions as described previously (Hobbie et al., 1997). The kinase activity of the immunoprecipitated p38 was determined using GST–ATF2 as described previously (Hibi et al., 1993; Derijard et al., 1995; Hobbie et al., 1997). Autoradiographs were scanned on an LKB densitometer.

Immunoblotting

Macrophages were prepared and infected as for the kinase assays. Before lysis, monolayers were washed with ice-cold PBS containing 10 mM NaF and 1 mM Na3VO4. Washed cells were lysed for 15 min on ice in 0.5 ml of modified RIPA (50 mM Tris, pH. 8.0, 150 mM NaCl, 1% NP40, 0.5% deoxycholate, 0.1% SDS, 1 mM Na3VO4, 10 mM NaF, 200 μM AEBSF, 20 μM leupeptin and 1 μM pepstatin). Lysates were clarified by centrifugation for 15 min at 4°C at 12 000 r.p.m. in a microfuge. Protein concentrations were determined using the Bio-Rad protein assay. Samples of the lysates containing approximately 20 μg of protein were separated on 10% SDS–polyacrylamide gels and transferred to nitrocellulose membranes. Membranes were developed with phosphospecific anti-p38 or anti-JNK antibodies and ECL reagents according to the manufacturer's instructions. Duplicate membranes were developed with standard anti-p38 or anti-JNK antibodies to control for loading.

Acknowledgements

We thank C. Roy and members of our laboratories for helpful discussions and comments on the manuscript, G. Habicht for assistance with the TNF-α bioassay and R. Rowehl for providing cultured cells. This work was supported by grants from the National Institutes of Health (AI35175) and the Sinsheimer Foundation. J.B.B. is a PEW Scholar in the Biomedical Sciences.

Ancillary