In this paper, we have reported the molecular cloning and characterization of a novel DSP termed MKP-M from a mouse macrophage cDNA library. MKP-M is localized in the cytoplasm and preferentially deactivates JNK among the three subsets of MAPKs. In macrophages, LPS stimulation induced MKP-M mRNA expression at least partly through p38 MAPK activation. The expression of a dominant-negative mutant of MKP-M in macrophages enhanced LPS-stimulated JNK activation in both transient and stable expression systems, suggesting that MKP-M is involved in the downregulation of JNK activity in macrophages after LPS stimulation.
DSPs form a family of proteins characterized by a highly homologous catalytic domain and selectively dephosphorylate MAPKs (27
). They play important roles in the maintenance of cellular functions, because dysregulation of MAPKs can be detrimental to cells. Comparison of the primary sequence of MKP-M with those of related DSPs revealed an extended active-site sequence motif, (V/L)X(V/I)HCXAG(I/V)SRSXT(I/V)XXAY(L/I)M (where X is any amino acid) that appeared to be a hallmark of this family of proteins. Additional areas of homology included a rhodanase homology domain in the N-terminal region, which may play a role in defining substrate specificity or serve as a recruiting site for other proteins (29
). In addition, the presence of a proline-rich carboxyl extension in MKP-M is interesting. Although it is not conserved among most DSP family proteins, this extension is homologous with those of a mammalian DSP, hVH-5 (or M3/6 for its mouse version) (35
), and a Caenorhabditis elegans
tyrosine phosphatase, VHP-1. Although the biological significance of this extension in MKP-M remains obscure, it contains a putative PEST sequence, KLCQFSPVQEVSEQSPETSPD (Fig. A). An abundance of proline (P), glutamate (E), serine (S), and threonine (T) residues is commonly found in the most rapidly degraded eukaryotic proteins, and the half-lives of proteins containing PEST sequences are frequently less than 1 h (48
). Thus, the putative PEST sequence may be utilized as signals for rapid degradation of MKP-M. As the MKP-M mRNA level in macrophages is rapidly increased by LPS stimulation but remains elevated for at least 24 h (data not shown), the temporal abundance of MKP-M may be regulated at both the mRNA and protein levels. Consistently, a C-terminally truncated version of MKP-M without the PEST sequence was easily expressed at much higher levels than the full-length version in stable transfection of macrophages. Interestingly, hVH-5 also contains PEST sequences in the homologous proline-rich extension (35
Other than the predominantly expressed mRNA isoform (A1), we identified three additional mRNA variants (A2, B1, and B2), which are probably produced by alternative splicing. The mRNA levels for these variants were significantly lower than that for the major mRNA isoform (A1), as Northern blot analysis revealed a dominant mRNA band that was confirmed to be the A1 isoform. In transient-transfection assays, only one of the three variants (B1) produced a stable protein of 45 kDa. As this putative short form of MKP-M protein does not contain the portion of the protein used to make the polyclonal antibody, we could not confirm the endogenous expression level of this 45-kDa protein. Although the MKP-M B1 protein lacks the C-terminal portion of the phosphatase domain and may not encode an active phosphatase, it contains an extended active-site sequence motif. The significance of these alternative spliced forms of MKP-M remains unknown.
The major form of MKP-M mRNA (A1) encodes an active MKP of 80 kDa. In cells, MKP-M displayed substrate selectivity for JNK2 compared with ERK2 and p38α. This substrate selectivity of MKP-M (JNK
ERK = p38) is unique and is different from those of MKP-1 (also called hVH1 or CL100) (p38 > JNK > ERK) (13
), MKP-3 (also called PYST1) (14
) and MKP-4 (ERK > JNK = p38) (42
), MKP-2 (also called hVH-2 or TYP-1) (ERK = JNK > p38) (9
), MKP-5 (p38 = JNK > ERK) (52
), or PAC-1 (ERK = p38 > JNK) (9
). Interestingly, it is somewhat similar to that of hVH-5 and M3/6 (p38 = JNK
), to which MKP-M is most homologous. It has recently been reported that the RILPHLYL sequence in the N-terminal domains of hVH-5 and M3/6, which shares significant homology with the JNK-binding site of c-Jun protein (called the delta domain), is important for the ability of these phosphatases to dephosphorylate JNK (25
). Although the ILP and L(Y/F)LG elements of this motif occur in all known mammalian DSPs, the presence of the N-terminal arginine, which is unique to hVH-5 and M3/6, makes this sequence conform to the delta domain consensus known to be critical for binding to JNK. Like hVH5 and M3/6, MKP-M contains arginine in the N-terminal position of this motif. As substrate binding to some DSPs is accompanied by catalytic activation through the stabilization of the active phosphatase conformation (5
), this delta domain-like motif may contribute to the selective JNK dephosphorylation by MKP-M.
It seems unreasonable that both hVH-5 (or M3/6) and MKP-M are localized mainly in the cell cytoplasm, considering that JNK efficiently phosphorylates several transcription factors and was reported to translocate to the nucleus in response to UV or hypoxia (26
). However, others have found that JNK constitutively exists in both cytoplasm and nuclei and can be activated without evident nuclear translocation at least by gamma irradiation (8
). More recently it has been reported that activated JNK translocates to mitochondria and interacts with an antiapoptotic protein, Bcl-xL
). Therefore, it seems reasonable to speculate that controlling JNK activity in the cytoplasm is essential for some of the physiological functions of JNK.
An interesting feature of many DSPs is their tight and rapid transcriptional induction by growth factors and/or cellular stresses. Indeed, MKP-1 was first identified as an immediate-early gene rapidly induced by mitogens, heat shock, or oxidative stress (7
). Other DSP genes also undergo transcriptional upregulation in response to various stimuli (4
). In many instances, induction of each DSP gene seems to be specific to given stimuli (4
). In mouse macrophage cell lines, a small amount of MKP-M mRNA was constitutively expressed. The amount of MKP-M mRNA significantly increased in response to LPS stimulation but not in response to IL-1β, TNF-α, IFN-γ, IL-15, or IL-2 stimulation. In contrast, MKP-1 mRNA responded to all of these cytokines, and PAC-1 mRNA responded to IL-1β and IFN-γ stimulation in addition to LPS stimulation. Interestingly, the mRNA level of M3/6, the closest homologue of MKP-M, also increased in response to LPS and lipid A stimulation but not in response to cytokine stimulation, indicating that M3/6 and MKP-M may share similar transcriptional control elements. The mRNA expression level of M3/6, however, seemed rather low in macrophages.
LPS-stimulated MKP-M mRNA induction seemed to be independent of ERK activation and to be mediated at least partly by the activation of the p38 MAPK pathway, as treatment with SB208530, a specific inhibitor of p38 activation, successfully inhibited the MKP-M mRNA increase after LPS stimulation, suggesting that ERK activation may have an inhibitory effect on MKP-M gene expression. In RAW264.7 cells, p38 MAPK is strongly activated by LPS treatment but only weakly activated by IL-1β, TNF-α, IFN-γ, IL-2, or IL-15 treatment (data not shown). Thus, we speculate that this LPS-specific response of MKP-M transcription may be due to its dependence on p38 MAPK activation. We presume that significant involvement of JNK in MKP-M transcriptional induction is unlikely, as dicoumarol, which partially inhibited LPS-mediated JNK activation, did not significantly affect MKP-M mRNA induction (Fig. B). However, as dicoumarol could not inhibit the JNK activation completely and no other powerful and specific JNK inhibitor was available, we could not rule out the possibility that a low-level activation of JNK is involved in the MKP-M mRNA increase mediated by LPS. On the other hand, the expression of the MKP-1 gene is dependent on the activation of ERK in CCL39 cells (3
) or of JNK in NIH 3T3 cells (2
). PAC-1 gene expression was reported to be dependent on ERK activation in T cells (15
). These reports are consistent with our finding that both MKP-1 and PAC-1 are responsive to different sets of stimuli than is MKP-M.
LPS, a cell wall component of gram-negative bacteria, is a complex glycolipid composed of a hydrophilic polysaccharide region and a hydrophobic domain known as lipid A that is responsible for most of the biological functions of LPS (49
). LPS stimulates host cells to produce endogenous mediators, including bioactive lipids (e.g., platelet-activating factor and thromboxane A2), reduced oxygen species (e.g., NO), and, in particular, cytokines such as IL-1, IL-6, IFN-γ, and TNF-α (49
). It activates a transcriptional factor, NF-κB (33
), and MAPKs, including ERKs (17
), JNKs (18
), and p38 MAPKs (19
). On the other hand, these responses have to be carefully regulated, as excessive and long-lasting inflammation often causes tissue damages. Thus, it is important to understand the mechanisms by which the JNK activity is regulated in macrophages in response to bacterial components such as LPS.
Proximally, JNKs are activated by the dual-specificity MKKs 4 and 7, which in turn are activated by the MEK kinases. MEK kinase 1 has recently been reported to be essential for JNK activation in embryonic stem cells (58
). In spite of these upstream kinases, JNK activation in macrophages after stimulation with LPS is a temporary process even in the continued presence of activating stimuli, indicating that the duration of JNK activity also provides important control mechanisms for LPS-induced inflammatory responses. However, the exact mechanisms of JNK deactivation after LPS stimulation have remained unknown.
Our results showed that the expression of a dominant-negative form of MKP-M increased both the magnitude and duration of JNK activity induced by LPS stimulation in a mouse macrophage cell line. It also decreased TNF-α secretion induced by LPS (Fig. D). These findings indicate that MKP-M, the transcription of which is induced by LPS, plays an important role in the JNK deactivation process in LPS-treated macrophages. It was rather surprising, as other MAPKs (at least MKP-1 and PAC-1) are also induced by LPS in these types of cells. However, in contrast to MKP-M, most of the other MKP members dephosphorylate ERKs or p38 MAPKs more preferentially, and M3/6, which is structurally related to MKP-M and preferentially dephosphorylates JNKs, is specifically expressed in brain, lung, and heart in adult mice (54
). In fact, the expression of the catalytically inactive mutants of MKP-1 or M3/6 in RAW264.7 cells did not have significant effects on LPS-mediated JNK activation (Fig. A).
In summary, these observations provide evidence that MKP-M, a newly identified MKP protein, specifically dephosphorylates JNK and plays an important role in regulating the LPS-mediated JNK activity and cytokine production in macrophages. MKP-M may become a new therapeutic target for the control of inflammatory responses caused by bacterial infection.