The Phosphothreonine Lyase Activity of a Bacterial Type III Effector Family

See allHide authors and affiliations

Science  16 Feb 2007:
Vol. 315, Issue 5814, pp. 1000-1003
DOI: 10.1126/science.1138960

This article has a correction. Please see:


Pathogenic bacteria use the type III secretion system to deliver effector proteins into host cells to modulate the host signaling pathways. In this study, the Shigella type III effector OspF was shown to inactivate mitogen-activated protein kinases (MAPKs) [extracellular signal–regulated kinases 1 and 2 (Erk1/2), c-Jun N-terminal kinase, and p38]. OspF irreversibly removed phosphate groups from the phosphothreonine but not from the phosphotyrosine residue in the activation loop of MAPKs. Mass spectrometry revealed a mass loss of 98 daltons in p-Erk2, due to the abstraction of the α proton concomitant with cleavage of the C-OP bond in the phosphothreonine residue. This unexpected enzymatic activity, termed phosphothreonine lyase, appeared specific for MAPKs and was shared by other OspF family members.

Gram-negative bacterial pathogens often use the type III secretion system (TTSS) to inject into host cells effector proteins (1) that interfere with host signal transduction pathways to promote pathogen infection. MAPK signaling plays an important role in activating host innate immune responses in both plants and animals and is a frequent target of pathogenic effectors (24). Identification of host targets of TTSS effectors and elucidation of their biochemical functions are critical in understanding the mechanism and the evolution of bacterial pathogenesis (3, 5, 6).

The TTSS effector OspF is present in all the four known pathogenic species of Shigella (79), a causative agent of bacillary dysentery (10). OspF, together with SpvC and HopAI1, represents a family of TTSS effectors conserved in both plant and animal bacterial pathogens (11) (fig. S1). None of the OspF family effectors has an established biological or biochemical function. To define the biochemical function of OspF and the OspF family of effectors as well as to identify their potential in vivo targets, we examined effects of OspF on the host immune signaling pathways, including those of nuclear factor κB(NFκB) and MAPKs.

Transient expression of OspF in human 293T cells efficiently blocked extracellular signal– regulated kinases 1 and 2 (Erk1/2) and c-Jun N-terminal kinase (JNK) signaling, whereas it did not alter NFκB activation (Fig. 1A). OspF abrogated phosphorylation of endogenous Erk1/2, JNK, and p38 kinases (Fig. 1, B and C). These results suggest that OspF harbors a specific activity of down-regulating multiple MAPKs but not NFκB signaling when overexpressed in mammalian cells. To define the specific step(s) in the MAPK pathway blocked by OspF, we turned to the canonical Erk pathway that is activated by a phosphorylation cascade of Raf, MAPK kinase (MEK), and Erk kinases (12). OspF blocked Erk activation in 293T cells transfected with either RasV12, v-Raf (constitutive active Raf), or MEK1-ED (constitutive active MEK1) (Fig. 1D). Meanwhile, RasV12-induced MEK1 phosphorylation remained intact despite the disappearance of Erk1/2 phosphorylation in cells expressing OspF (Fig. 1E). Thus, inhibition of Erk phosphorylation by OspF is downstream of MEK1 along the classical Ras-Raf-MEK-Erk cascade.

Fig. 1.

Inhibition of multiple MAPK pathways downstream of MAPK kinase by OspF in 293T cells. (A) Luciferase assays of effects of OspF expression on Erk1/2, JNK, and NFκB pathways. Erk1/2 and JNK activations were achieved by coexpression of constitutive active RasV12 and RacL61, respectively, and NFκB was stimulated by tumor necrosis factor α treatment. V, F, and J denote vector, OspF, and YopJ, respectively. Error bars indicate standard deviation. (B and C) Phosphospecific immunoblotting assays of inhibition of MAPKs activation by OspF. p-Erk, p-JNK, and p-p38 antibodies recognize the dual phosphorylated TXY motif (pT-X-pY) in Erk1/2, JNK1/2, and p38, respectively. HA, hemagglutinin. (D) Luciferase assays of Ras-, Raf-, and MEK1-induced Erk activation in the presence of OspF. v-Raf and MEK1 ED are constitutive active variants of Raf and MEK1. Error bars indicate standard deviation. (E) Phosphospecific immunoblotting of activation of MEK1 and Erk1/2 in the presence of OspF.

Erk1/2 phosphorylation can be reconstituted in the cell-free extracts system by adding upstream signaling molecules (3). Addition of bacterially expressed and highly purified OspF (fig. S2A) in 293T cell extracts abolished both RasV12 and active B-Raf–induced Erk1/2 phosphorylation, whereas MEK1 phosphorylation was not affected (Fig. 2A). This excludes the transcriptional regulation of MAPK pathway by OspF and also provides a system that recapitulates the inhibition of MAPK signaling by OspF in cells. To further test whether OspF directly targets the Raf-MEK1-Erk1/2 cascade biochemically, we reconstituted the phosphorylation cascade of B-Raf–MEK1–Erk2 by using purified kinases. Addition of OspF to this reaction largely abolished Erk2 phosphorylation (Fig. 2B). Furthermore, recombinant OspF could inhibit phosphorylation of glutathione S-transferase (GST)–Erk2 by MEK1-ED (Fig. 2C), suggesting that OspF directly targets Erks or MEKs. Recombinant Erk2 fused to maltose binding protein (MBP-Erk2), but not MBP alone, effectively precipitated OspF (fig. S2A). Flag-OspF also coprecipitated with endogenous Erk1/2 but not MEK1 from 293T cells (fig. S2B). Thus, OspF targets Erk and attenuates its phosphorylation status.

Fig. 2.

Immunoblotting analyses of removal of phosphate groups from Erk, JNK, and p38 kinases by OspF in vitro. (A) Effects of OspF on Erk1/2 activation reconstituted in the cell-free extract system. Active B-Raf protein was added to activate Erk pathway in the presence or absence of OspF. (B) Effects of OspF on in vitro reconstituted Erk2 phosphorylation downstream of Raf. (C) Effects of OspF on Erk2 phosphorylation by MEK1 in vitro. (D) Selective removal of phosphate groups from the phosphothreonine residue in the pT-X-pY motif in Erk2 in vitro. GST-p-Erk2 was treated with OspF or MAPK phosphatase 3 (MKP3), followed by p-Erk and a phosphotyrosine-specific antibody (p-Tyr) immunoblotting. (E) Removal of the phosphate groups from p-JNK and p-p38 by OspF in vitro. (F) Phosphotyrosine immunoblotting of Myc-Erk2 modified by OspF in 293T cells.

Given the direct association between OspF and Erk2 and that the basal amount of phospho-Erk1/2 (p-Erk1/2) was diminished by OspF addition in the cell-free reconstitution assay (Fig. 2A), we reasoned that OspF could harbor a phosphatase activity to reverse Erk phosphorylation. When in vitro phosphorylated GST-Erk2, JNK, and p38 were incubated with recombinant OspF, dephosphorylation indeed occurred (Fig. 2, D and E). Furthermore, OspF selectively removed phosphate groups from the threonine but not from the tyrosine residue in the pT-X-pY (13) motif in p-Erk2 (Fig. 2, D and F). Thus, OspF harbors an in vitro enzymatic activity of specifically removing phosphate groups from the phosphothreonine in the pT-X-pY motif in MAPKs, including Erk1/2, JNK, and p38.

OspF appears to function as a threonine-specific MAPK phosphatase. However, our analysis rules out the possibility of a classical protein phosphatase of OspF in nature [Supporting Online Material (SOM) text]. To reveal the chemical nature of OspF activity, we subjected MEK1-phosphorylated MBP-Erk2 (p-Erk2–control), as well as MBP–p-Erk2 further treated with OspF (p-Erk2–OspF), to tandem mass spectrometry analysis. More than 80% of the MBP-Erk2 sequence was covered by observed tandem peptides. All of the peptides from p-Erk2–control and p-Erk2–OspF were indistinguishable except for the peptide containing the TXYphosphomotif [VADPDHDHTGFL-pT-E-pY-VATR (13)]. The mass of the peptide from p-Erk2–OspF was 2204.9 dalton (Fig. 3A), 98 dalton less than that of the corresponding peptide from p-Erk2–control. The 98-dalton mass decrease occurred on the phosphothreonine residue in the pT-X-pY motif (Fig. 3B). Consistently, when a synthetic phosphopeptide containing the TXY phosphomotif in Erk2 (DHTGFL-pT-E-pY-VATR) was modified by OspF and the resulting product analyzed by mass spectrometry, a mass decrease of 98 dalton on the phosphothreonine residue compared with the control peptide was also observed (Fig. 3C). Thus, upon OspF treatment, the phosphothreonine residue in the TXY motif in both p-Erk and the phosphopeptide substrate underwent a dehydration reaction (18 dalton of mass loss) in addition to the loss of a phosphate group (80 dalton). The reaction of p-Erk2 with OspF probably resulted in a C=C bond between Cβ and Cα in the side chain of the threonine residue (Fig. 3D), suggesting that OspF functions as a phosphothreonine lyase that cleaves the C-OP bond with the concomitant abstraction of the α proton in the phosphothreonine residue in p-Erk2. Furthermore, this chemical modification of p-Erk2 resulting from OspF phosphothreonine lyase activity led to a completely inactive kinase both in vitro and in cells (fig. S3). In addition, OspF-“dephosphorylated” Erk2 could no longer be rephosphorylated by MEK1-ED (fig. S4). Thus, OspF functions as a phosphothreonine lyase, rather than a phosphatase, to irreversibly inactivate MAPKs.

Fig. 3.

The phosphothreonine lyase activity of OspF. (A) The mass spectrometry analysis of the molecular mass of the tryptic peptide (VADPDHDHTGFL-pT-E-pY-VATR) from OspF-treated p-Erk2. Different peaks correspond to the isotopic peaks of the peptide. (B) Electrospray ionization (ESI) tandem mass spectrometry (MS/MS) spectrum of the tryptic peptide shown in (A). b and y ions are marked in the spectrum. The amino acid sequence of the peptide is shown at top left, and TΔ marks the OspF-modified phosphothreonine residue. The fragmentation patterns that generate the observed b and y ions are illustrated along the peptide sequence. (C) MS/MS spectrum of the OspF-modified synthetic Erk2 phosphopeptide (DHTGFL-pT-E-pY-VATR). Notations are the same as described in (B). Inlet shows the molecular mass of the Erk2 phosphopeptide before (control) and after (OspF) treatment. a.u., arbitrary units. (D) Diagram of the chemical alteration of the phosphothreonine residue in Erk2 by OspF. OspF cleaves the C-OP bond with the concomitant abstraction of the α proton, which results in a C=C bond between the Cα and the Cβ in the phosphothreonine residue in Erk2.

To gain further understanding of the OspF phosphothreonine lyase activity, we mutated 10 residues conserved in all the OspF family members (fig. S5A). These mutants had similar expression and purification profiles to those of the wild-type protein, suggesting that they were folded correctly. Most of the mutants retained the Erk2 phosphothreonine lyase activity, whereas mutants of K102→R102 (K102R), H104→A104 (H104A), and K134→A134 (K134A) were largely inactive (fig. S5, A and B). These mutants could still bind to Erk (fig. S2, A and C), suggesting a functional importance for K102, H104, and K134. A nonradioactive phosphatase assay was used to quantitatively measure the enzymatic activity of OspF. OspF efficiently removed phosphate groups from the Erk2 phosphopeptide with a Michaelis-Menton constant (Km) of 95.8 ± 2.4 μM and a catalytic rate constant (kcat) of 0.70 ± 0.04 s–1 (table S1). The K102R, H104A, and K134A mutants were almost completely inactive in the peptide assay (fig. S5C). Furthermore, these mutants were also deficient in down-regulating MAPK signaling when expressed in 293T cells (fig. S6). Thus, K102, H104, and K134 are probably essential for OspF phosphothreonine lyase activity toward MAPKs.

We then assessed the role of the phosphate on the tyrosine in pT-X-pY in the MAPK phosphothreonine lyase activity of OspF. When Erk2 (pT-X-pY) was tyrosine-dephosphorylated by protein tyrosine phosphatase (HePTP), subsequent dephosphorylation by OspF was less efficient (fig. S7). Consistently, kinetic analysis using the synthetic threonine-monophosphorylated Erk2 peptide gave a higher Km value compared with that of the dual phosphorylated Erk2 peptide (table S1). Thus, the phosphate on the tyrosine in the substrate probably contributes to substrate recognition by OspF. Further kinetics analysis using different peptide substrates showed that OspF was specific for the phosphothreonine in the classical MAPKs (with TXY motif), with the highest enzymatic activity toward p38 (table S1).

SpvC, a member of the OspF family of TTSS effectors from nontyphoid Salmonella strains, is required for full virulence and systemic infection (14, 15). Recombinant SpvC efficiently removed phosphate groups from GST–p-Erk2 (Fig. 4A). SpvC catalyzed the same enzymatic conversions of the Erk2 phosphopeptide with kinetics comparable to that observed for OspF (kcat = 1.94 ± 0.05 s–1 and Km = 28.5 ± 0.5 μM). Similarly to OspF, overexpression of SpvC in mammalian cells abrogated phosphorylation and activation of Erk1/2, JNK, and p38 kinases (Fig. 4, B to D). In addition, HopAI1, a type III effector from plant pathogen Pseudomonas syringae (11), possessed the same in vitro catalytic activity as described for OspF and SpvC, using either p-Erk2 (Fig. 4A) or the Erk2 phosphopeptide as the artificial substrate (kcat = 0.70 ± 0.02 s–1 and Km = 132.1 ± 8.1 μM). Thus, the entire OspF family of TTSS effectors has an in vitro phosphothreonine lyase activity.

Fig. 4.

The phosphothreonine lyase activity of other OspF family members. (A) Immunoblotting assays of removing phosphate groups from p-Erk2 by indicated OspF family members in vitro. t, time. (B and C) Phospho-Erk immunoblotting assays of down-regulation of Erk(B), JNK, and p38(C) pathways by SpvC transfection in 293T cells. The assay was performed and the results are presented similarly as shown in Fig. 1, B and C, respectively. (D) Luciferase assays of effects of SpvC on Erk signaling activated by RasV12 in 293T cells. Error bars indicate standard deviation.

To examine whether the type III-secreted OspF indeed targeted the host MAPKs, we generated an OspF-deletion strain of Shigella flexneri 2a and performed an infection assay (16). In contrast to a recent report that analyzed p-Erk1/2 at 3 hours postinvasion (17), we analyzed p-p38 at 30 min postinvasion, because OspF was 8 times more active toward the p38 phosphopeptide compared with the Erk2 phosphopeptide (table S1). In addition, several studies have suggested that OspF is only transcribed for 2 hours after bacteria invasion (18, 19). A higher amount of p-p38 was observed in HeLa cells infected with the OspF-deletion strain of Shigella compared with the amount in cells infected with the wild-type strain (fig. S8A). Furthermore, reintroducing wild-type OspF, but not the H104A mutant, into the OspF-deletion strain lowered the amount of p-p38 in infected HeLa cells (fig. S8B). Thus, MAPK is a potential target of the phosphothreonine lyase activity of OspF during Shigella infection.

Phosphothreonine lyase activity has not been described previously. The closest enzyme might be the threonine synthase (EC in bacteria and plants, which uses a pyridoxal phosphate cofactor. This enzyme converts O-phospho-l-homoserine into threonine through an intermediate that has a C=C bond between the Cβ and the Cγ. However, OspF did not show any sequence identity with any phospholyase, including threonine synthase, and required no cofactors. This suggests that the OspF family phosphothreonine lyase has designed an unusual mechanism for enzyme catalysis. Together, OspF, SpvC, and HopAI1 define a family of conserved pathogenic effectors with a unique phosphothreonine lyases activity that probably targets the host MAPKs in pathogenesis.

Supporting Online Material

Materials and Methods

SOM Text

Figs. S1 to S9

Table S1


References and Notes

View Abstract

Stay Connected to Science

Navigate This Article