The mitogen-activated protein kinase (MAPK) signaling pathways are crucial in cell function and homeostasis. MAPKs regulate pathophysiological processes by controlling signal translation and cellular response such as survival, proliferation, differentiation and migration [1,2,3][1][2][3]. They are activated by a double phosphorylation process on tyrosine and threonine residues in a conserved Thr-X-Tyr motif (X being any amino acid) [4,5][4][5]. Activation of MAPK pathways triggers multiple intracellular signaling cascades. Each cascade is initiated by a specific signal and leads to the activation of a particular MAPK [6]. Once activated, MAPK can phosphorylate various cytoplasmic and/or nuclear substrates and induce changes in the function of target proteins and gene expression [6]. Spatial localization of MAPKs also determines the target substrates and subsequent cellular effects [7].
Phosphatases reverse phosphorylation and return MAPKs to an inactive state. The dual specificity phosphatase (DUSP) family belongs to the 199 phosphatases encoded in the human genome. This family is composed of 61 phosphatases capable of downregulating MAPKs by dephosphorylating both tyrosine and serine/threonine residues in a single substrate [12,15][8][9]. The phosphorylation of proteins is a reversible process. This prevents the abnormal activation of the signal and fine-tunes its activity and downstream effects [3,12][3][8]. The balance between phosphorylation and dephosphorylation controls the expression, function, activity and localization of many proteins [12,16][8][10]. Dephosphorylation by DUSPs regulates the duration, intensity and spatiotemporal profile of the MAPK signaling cascade [17][11]. This dephosphorylation takes place thanks to the highly conserved phosphatase site which contains arginine, cysteine and aspartic acid [3,12,18][3][8][12]. In addition to the active site common to all DUSPs, some DUSPs contain a MAP kinase-binding motif (MKB), also called a kinase-interacting motif (KIM), which interacts with the common docking domain of MAPKs to allow the interaction between the enzyme and the substrate [3,12,18,19][3][8][12][13]. Ten DUSPs containing the KIM domain are classified as typical DUSPs or MAP kinase phosphatases (MKPs) (Table 1), while those which do not have this domain (16 phosphatases in total) are called atypical DUSPs [3,5,20][3][5][14]. However, there are a few exceptions. DUSP2, DUSP5 and DUSP8 are typical DUSPs and contain the KIM domain but they are not called MKPs. On the other hand, DUSP14 and DUSP26, which are atypical DUSPs and do not contain a KIM domain, are called MKP6 and MKP8, respectively (Table 1) [3,20][3][14]. Typical DUSPs are the best characterized within the DUSP family [20][14] and this comprises the typical DUSP named DUSP9 or MKP4, which was first described in 1997 by Muda and collaborators [18][12]. This 42-kDa protein dephosphorylates several substrates including JNK, p38, the MAPKKK apoptosis signal-regulating kinase 1 (ASK1) and ERK1/2 with a high specificity for ERK kinases [3,18][3][12].
The MKB/KIM motif is composed of two CDC25 homology domains and an intermediate group of basic amino acids mediating the interaction with the common domain of MAPKs
[16,18,19,44][10][12][13][39].
DUSP9 is able to undergo a conformational rearrangement allowing it to alternate between a partially active structure and a fully active structure
[43][38]. A crystallographic representation of the catalytic site of
DUSP9 showed a unique structure with significant differences between the catalytic core and several surrounding loops compared to other MKPs. The catalytic site of
DUSP9 deviates considerably from the canonical conformation of DUSPs, which may explain the low catalytic activity of this protein in the absence of specific substrates
[19,43][13][38]. The
DUSP9 protein alone has very low catalytic activity but binding to MAPK through the MKB/KIM domain significantly increases its phosphatase activity
[18,19][12][13]. The binding of the substrate likely triggers a conformational change and thus increases its catalytic efficiency
[19,43][13][38].
DUSP9 is capable of binding and being activated by different MAPKs. Measurements in the presence of para-nitrophenylphosphate (pNPP) showed
DUSP9 binding preference and activation by ERK2, JNK and p38 MAPK
[19,34,44][13][27][39].
At a functional level,
DUSP9 phosphatase is unique. It plays an important role in the dephosphorylation and inactivation of specific kinases such as ASK1, ERK1/2, p38 and JNK (
Figure 2). This results in a negative control of MAPK signal transduction and a fine tuning of their duration and intensity
[15,17,32][9][11][25]. The endogenous catalytic activity of
DUSP9 was studied for the first time by Muda and collaborators by measuring the hydrolysis of pNPP in the presence of increasing doses of a purified human recombinant protein.
DUSP9 displayed a dose-dependent catalytic activity which was directly proportional to the amount of protein added
[18][12]. The
DUSP9 protein has a broad specificity for MAPK substrates. It can dephosphorylate ERK-family MAPKs, stress-activated JNK and p38 MAPKs (
Figure 2), but its effect is significantly higher and more specific for ERK kinases
[18,43][12][38]. In hepatic tumor cells,
DUSP9 negatively regulates the RAS/RAF/MEK/ERK signal by dephosphorylating ERK1/ERK2 and a low level of
DUSP9 is correlated with an elevated level of phospho-ERK1/2 in hepatocellular carcinoma (HCC) samples
[15][9]. Co-incubation of ERK2 with increasing concentrations of
DUSP9 results in the dose-dependent blockade of ERK2 target phosphorylation such as stathmin
[18][12]. This catalytic activity is effectively inhibited by sodium vanadate, which is an inhibitor of protein tyrosine phosphatases
[18][12].
Figure 2. Regulation of DUSP9 expression and connection with MAPK pathways. Following the activation of tyrosine kinase receptors (TKR), MAPK pathways are activated through the successive phosphorylations of MAPKKKs (among which ASK1), MAPKKs MKK3/6, MKK4/7 and MEK1/2, and MAPKs p38, JNK and ERK. Phosphorylated MAPKs translocate to the nucleus and induce expression of the downstream targets c-FOS, ERG1, NANOG and HIF1, among others. In murine embryonic stem cells, the binding of BMP4 on its receptor induces the phosphorylation of Smad1/5, which then associates with Smad4. The Smad1/5–Smad4 complex translocates to the nucleus, binds the DUSP9 promoter and induces its transcriptional expression. Transcription factors HIF1-α and ETS can also induce DUSP9 expression. Besides the BMP signal, mTOR, RAR-α and ERβ pathways can also potentiate the transcription of the DUSP9 gene. Following NANOG binding on LINCU promoter, the long non-coding RNA LincU is transcribed and exported in the cytoplasm where it associates with DUSP9 and stabilizes it. Stable DUSP9 can dephosphorylate its substrates, including ERK1/2, p38, JNK and ASK1. In the absence of LincU RNA, DUSP9 protein is unstable and is polyubiquitinylated before its degradation by the proteasome. DUSP9 can also be post-transcriptionally regulated by miR-212 and miR-1246, which target the 3′-untranslated region of its mRNA. In summary, DUSP9 expression is tightly regulated by transcriptional, post-transcriptional and post-translational mechanisms.
In normal mature tissues,
DUSP9 is mainly expressed in kidney, adipose tissue and placenta, while it is only minimally present in brain, ovary, testis and urinary bladder (National Center for Biotechnology Information: available online:
https://www.ncbi.nlm.nih.gov/gene/1852, accessed on 15 July 2021)
[32,46,47][25][41][42]. In order to maintain cell homeostasis, the level and activity of
DUSP9 has to be tightly regulated at transcriptional, post-transcriptional and post-translational levels (
Figure 2)
[3]. At the transcriptional level, the expression of
DUSP9 can be regulated by various transcription factors such as EFS family members and hypoxia-inducible factor 1 alpha (HIF1α). For instance, genetic depletion by gene knock-down or digoxin-induced pharmacological inhibition of
HIF1 blocks the expression of
DUSP9 and induces the loss of its inhibitory effect on the ERK signaling pathway in MDA-MB-231 breast cancer cells
[17][11]. Thus,
HIF1 positively regulates
DUSP9 expression.
DUSP9 is also transcriptionally regulated by BMP signaling in mouse embryonic stem cells (mESCs). The bone morphogenetic protein 4 (Bmp4) induces the recruitment of Smad1/5 and Smad4 on the promoter region of
DUSP9 and induces
DUSP9 expression at the mRNA level, and thus at the protein level. The positive regulation of
DUSP9 by Bmp4 is accompanied by inhibition of ERK pathway activity and downregulation of its targets Egr1 and Fos (
Figure 2). This
DUSP9-mediated dephosphorylation was not observed for p38 and JNK, showing its specificity for ERK1/2 signaling which is crucial for the renewal and differentiation of mESCs
[31][24]. In the same cells,
DUSP9 can also undergo post-translational regulation by a long non-coding RNA (lncRNA) called
LincU [48][43]. Following induced expression of
LincU by the transcription factor Nanog, this lncRNA directly binds
DUSP9 and maintains the phosphatase in an active and stable conformation. As a result,
DUSP9, bound to
LincU, constantly dephosphorylates ERK1/2 and thus, totally blocks RAS/RAF/MEK/ERK signaling. Interestingly, the phosphorylation level of ERK1/2 was inversely correlated with
LincU expression and the amount of
DUSP9 protein, while the level of
DUSP9 mRNA remained unchanged
[48][43]. Thus,
LincU interacts with
DUSP9 protein and stabilizes it in an active state, thereby triggering a constitutive dephosphorylation state of
DUSP9 kinase substrates (
Figure 2)
[48][43].
DUSP9 can also be transcriptionally regulated by the retinoic acid receptor (RAR). Microarray analysis of differential gene expression in Caco-2 cells treated with RAR agonists, Ch55 and Am580 demonstrated dose-dependent induction of
DUSP9 expression by RAR signaling.
DUSP9 was one of the most induced genes after RAR activation. Induction of
DUSP9 mRNA and protein by RAR signaling was confirmed by RT-qPCR and immunolabelling
[49][44]. Similar results were obtained with HT29 and HeLa cells. On the other hand,
DUSP9 induction was inhibited in Caco-2 cells expressing the dominant negative form of RARα or treated by the specific RAR antagonist LE540. Chromatin immunoprecipitation analysis showed that RAR induces
DUSP9 expression by binding directly to the
DUSP9 promoter through an inverted direct repeat separated by 1 (DR1 element). By inducing
DUSP9, RAR signaling inactivates ERK during the differentiation of colorectal cancer (CRC) cells
[49][44]. In a more recent paper, an upregulation of
DUSP9 was also reported in rat ovaries following Estrogen Receptor-beta signaling induction, supporting the hypothesis that
DUSP9 is one of the key genes involved in gonadotrophin-mediated ovarian follicle development
[50][45].
In the cytoplasm,
DUSP9 can be degraded by the proteasome following its ubiquitination
[3,48][3][43]. Pretreatment with the potent proteasome inhibitor MG132 stabilizes
DUSP9 protein and increases its level in
LincU-deficient mESCs
[48][43]. These data indicate that the ubiquitination-proteasome pathway is involved in the degradation of
DUSP9 induced by
LincU deficiency. Immunoprecipitation performed after pretreatment with MG132 and immunolabelling with an anti-ubiquitin antibody showed polyubiquitinated
DUSP9 bands, suggesting that the knockdown of
LincU causes ubiquitination of
DUSP9 [48][43]. This indicates that
LincU binds to
DUSP9, increases its stability and protects it from ubiquitin-mediated degradation
[3,48][3][43].