ERK1/2 (extracellular-signal-regulated kinase 1/2) MAPKs (mitogen-activated protein kinases) are tightly regulated by the cellular microenvironment in which they operate. Mxi2 is a p38α splice isoform capable of binding to ERK1/2 and ensuring their translocation to the nucleus. Therein Mxi2 sustains ERK1/2 phosphorylation levels and, as a consequence, ERK1/2 nuclear signals are enhanced. However, the molecular mechanisms underlying this process are still unclear. In the present study, we show that Mxi2 prevents nuclear but not cytoplasmic phosphatases from binding to and dephosphorylating ERK1/2, disclosing an unprecedented mechanism for the spatial regulation of ERK1/2 activation. We also demonstrate that the kinetics of ERK1/2 extranuclear signals can be significantly altered by artificially tethering Mxi2 to the cytoplasm. In this case, Mxi2 abolishes ERK1/2 inactivation by cytoplasmic phosphatases and potentiates ERK1/2 functions at this compartment. These results highlight Mxi2 as a key spatial regulator of ERK1/2 functions, playing a pivotal role in the balance between ERK1/2 nuclear and cytoplasmic signals.

INTRODUCTION

ERK (extracellular-signal-regulated kinase) 1/2 MAPKs (mitogen-activated protein kinases) are cytoplasmic serine/threonine kinases that participate in the transduction of signals from the surface to the interior of the cell. ERK1/2 become activated in response to multiple stimuli, including those that regulate cellular proliferation, differentiation and survival. ERK1/2 function within a signalling module containing several tiers of protein kinases, sequentially including: Raf family MAPKKKs (MAPK kinase kinases) and MEK (MAPK/ERK kinase) 1 and 2 dual-specificity MAPKKs (MAPK kinases). This cascade is regulated at its origin by Ras GTPases [1]. Once activated, ERK1/2 disperse throughout the cell and phosphorylate a broad spectrum of substrates localized at different subcellular compartments, including the nucleus and the cytoplasm [2]. The balance between the cytoplasmic and nuclear components of ERK1/2 signals is critical for the biological outcome resulting from ERK1/2 activation [3,4].

ERK1/2 activity is subject to strict spatial and temporal inactivation, primarily mediated by the removal of the phosphates from the regulatory Thr183 and Tyr185 at its ‘activation lip’ [5]. This inactivation process can occur by the action of serine/threonine phosphatases, such as PP2A (protein phosphatase 2A) [6], tyrosine phosphatases, including PTP-SL (striatal-enriched phosphatase-like protein tyrosine phosphatase) [7], or DUSPs (dual-specificity phosphatases), termed MKPs (MAPK phosphatases) [8]. The regulation of ERK1/2 dephosphorylation is complex and occurs at multiple levels [9]. For example, MKPs are encoded by inducible genes whose expression responds to multiple stimuli, in many cases regulated by ERK1/2 themselves [8]. Some MKPs exhibit a short half-life and are readily degraded by the proteasome following ubiquitination, a process that can be prevented by ERK1/2-mediated phosphorylation [10]. Finally, the spatial segregation of different phosphatases in distinct cellular compartments also plays an important role in the specificity of ERK1/2 inactivation [9]. What seems to be clear is that a direct interaction between ERK1/2 and its cognate phosphatases must take place for its dephosphorylation and subsequent inactivation to take place [8,9,11]

Mxi2 is a splice isoform of p38α. It is identical with p38α from amino acids 1 to 280, but it harbours a distinctive 17-amino-acid C-terminus. Its distribution in human tissues resembles that of p38α [12], being most abundant in the kidney [13]. We have reported previously that some of the biochemical properties of Mxi2 make it unique: Mxi2 is insensitive to inhibition by pyridinyl imidazoles and its activity with bona fide p38α substrates is very low [14]. More interestingly, we have identified Mxi2 as a stimulus-independent nuclear shuttle for ERK1/2 [15]. Mxi2 directly binds to ERK1/2 and enhances their accumulation in the nucleus by increasing their affinity for nucleoporins [15]. Furthermore, Mxi2 binding to ERK1/2 prolongs their phosphorylation state, as a consequence of which ERK1/2 nuclear but not cytoplasmic signals are enhanced [16].

To date, the mechanism whereby Mxi2 sustains ERK1/2 phosphorylated status is completely unknown. In the present study, we have investigated this phenomenon in further depth. We report that Mxi2 prevents ERK1/2 dephosphorylation by nuclear but not cytoplasmic phosphatases, by precluding ERK1/2 binding to such proteins. Importantly, we demonstrate that the interaction of ERK1/2 with phosphatases and the duration of its signals can be modulated by the localization of Mxi2. As such, cytoplasmic Mxi2 can prevent ERK1/2 binding to cytoplasmic phosphatases and enhance ERK1/2 signals in that compartment. Overall, our results provide a novel regulatory mechanism for ERK1/2 inactivation and identify Mxi2 as a key modulator in the balance between ERK1/2 nuclear and cytoplasmic signals.

MATERIALS AND METHODS

Plasmids

The expression vectors encoding for Myc-tagged MKP1, MKP3 and DUSP5 were provided by Dr S.M. Keyse (Molecular Pharmacology Unit, Biomedical Research Centre, Dundee, U.K.) and HA (haemagglutinin)-tagged PTP-SL was provided by Dr R. Pulido (Centro de Investigación Príncipe Felipe, Valencia, Spain). GFP (green fluorescent protein)–ERK1c, ERK2-NLS (nuclear localization signal) and HA-tagged Mxi2 and Mxi2 E293A have been described previously [15,17]. To generate NES (nuclear export signal)-Mxi2, the MEK1 NES (ALQKKLEELELDEQQRKRLE) was introduced by PCR immediately after the initiating methionine in Mxi2. Sequences of the oligonucleotides utilized are available upon request.

Cell culture and transfection

COS-7, HEK (human embryonic kidney)-293T and MDCK (Madin–Darby canine kidney) cells were grown in DMEM (Dulbecco's modified Eagle's medium)/10% fetal calf serum. Subconfluent cells were transfected using the calcium phosphate technique [16]. EGF (epidermal growth factor) and PD98059 were from Calbiochem.

Immunoblotting and immunoprecipitation

Immunoblotting and immunoprecipitations were performed as described previously [16]. Briefly, cells were collected and lysed in 20 mM Hepes (pH 7.5), 10 mM EGTA, 40 mM 2-glycerophosphate, 1% NP-40 (Nonidet P40), 2.5 mM MgCl2, 1 mM DTT (dithiothreitol), 2 mM vanadate, 1 mM PMSF, 20 μg/ml aprotinin and 20 μg/ml leupeptin. Proteins were fractionated by SDS/PAGE and transferred on to nitrocellulose filters. Immunocomplexes were visualized by ECL (enhanced chemiluminescence; Amersham Bisociences), using HRP (horseradish peroxidase)-conjugated goat anti-rabbit or anti- mouse secondary antibodies (Santa Cruz Biotechnology). Mouse monoclonal anti-HA (SC-7379), -ERK1/2 (SC-154) and -phospho-ERK (SC-7383) antibodies were from Santa Cruz Biotechnology, and anti-phospho-Myc (9401S) and -phospho-RSK1 (ribosomal S6 kinase 1) (9344S) antibodies were from Cell Signaling Technology. Rabbit polyclonal anti-Myc epitope (SC-789), -MKP1 (SC-370), -MKP3 (SC-28902), -DUSP5 (SC-46926), p38 (N-terminal) (SC-728), -Myc (SC-746), -Elk (SC-355), -RSK1 (SC-231) and lamin A (SC-20680) were from Santa Cruz Biotechnology. Immunoprecipitations were performed in 20 mM Hepes (pH 8), 2 mM MgCl2, 2 mM EGTA, 150 mM NaCl and 1% NP-40. Protein G–Sepharose pellets were washed three times with the same buffer.

Nucleus/cytoplasm fractionation

Nucleus/cytoplasm fractionations were performed in 20 mM Hepes (pH 7.4) buffer, basically as described previously [4]. Briefly, cells were collected in 50 mM 2-glycerophosphate (pH 7.3), 1 mM EDTA, 1 mM EGTA and 1 mM DTT, centrifuged and lysed in 40 mM Hepes (pH 7.5), 5 mM EGTA, 0.1% NP-40, 5 mM MgCl2, 1 mM DTT, 1 mM sodium orthovanadate and 1 mM benzinamide. The lysate was vortex-mixed vigorously and was centrifuged to obtain the cytoplasmic fraction as the supernatant. Nuclei were resuspended in 50 mM 2-glycerophosphate (pH 7.3), 0.2 mM EDTA, 420 mM NaCl, 1.5 mM MgCl2, 1 mM DTT and 25% glycerol, sonicated briefly on ice, vortex-mixed and centrifuged, and the precipitated cell debris was discarded.

RESULTS

Mxi2 sustains ERK1/2 phosphorylation levels at the nucleus

Previous results from our laboratory have demonstrated that Mxi2 binds to ERK1/2 and sustains them in a phosphorylated status [16]. To study this phenomenon further, we evaluated whether, in response to Mxi2 expression, the maintenance of ERK1/2 phosphorylation levels take place throughout the cell or, in contrast, whether it was restricted to a defined sublocalization. To this end, the duration of ERK1/2 phosphorylation, as induced by EGF stimulation, was monitored in the nuclear and cytoplasmic fractions from COS-7 cells transfected with Mxi2. It was found that, in the presence of Mxi2, nuclear ERK1/2 phosphorylation levels remained maximal for 10 h, a dramatic prolongation in comparison with vector-transfected cells, in which ERK phosphorylation levels began to decrease after 10 min (Figure 1A). In contrast, the levels of phosphorylated ERK1/2 in the cytoplasm evolved similarly in Mxi2-tranfected and control cells (Figure 1A). ERK1c is an isoform particularly enriched at the Golgi complex [17]. Thus it was of interest to explore whether Mxi2 influenced the phosphorylation state of this isoform. It was found that, in the presence of Mxi2, ERK1c phosphorylation was sustained for barely 5 min longer than in control cells (Figure 1B). These results demonstrate that the effects of Mxi2 on ERK1/2 phosphorylation exhibit spatial specificity, taking place mainly at the nucleus.

Mxi2 prolongs ERK1/2 phosphorylation at the nucleus

Figure 1
Mxi2 prolongs ERK1/2 phosphorylation at the nucleus

(A) Effects of Mxi2 on ERK1/2 phosphorylation in different compartments. COS-7 cells transfected with (+) or without (−) HA–Mxi2 (1 μg) were stimulated with EGF (100 ng/ml) following starvation. Phosphorylated ERK levels were examined at the indicated time points in the nuclear (nuc) and cytoplasmic (cyt) fractions. Lower panel, expression levels of HA–Mxi2. The purity of the nuclear and cytoplasmic fractions was ascertained by immunoblotting with lamin A and Rho-GDI lysate respectively. tot., total (guanine-nucleotide-dissociation inhibitor). (B) Effects of Mxi2 on ERK1c phosphorylation. HEK-293T cells were transfected with GFP–ERK1c (4 μg) in addition to HA–Mxi2 (1 μg) as indicated. Cells were stimulated with EGF and phosphorylated ERK levels were examined at the indicated time points in total lysates. Lower panel, expression levels of GFP–ERK1c and HA–Mxi2. The results shown are representative of three independent experiments.

Figure 1
Mxi2 prolongs ERK1/2 phosphorylation at the nucleus

(A) Effects of Mxi2 on ERK1/2 phosphorylation in different compartments. COS-7 cells transfected with (+) or without (−) HA–Mxi2 (1 μg) were stimulated with EGF (100 ng/ml) following starvation. Phosphorylated ERK levels were examined at the indicated time points in the nuclear (nuc) and cytoplasmic (cyt) fractions. Lower panel, expression levels of HA–Mxi2. The purity of the nuclear and cytoplasmic fractions was ascertained by immunoblotting with lamin A and Rho-GDI lysate respectively. tot., total (guanine-nucleotide-dissociation inhibitor). (B) Effects of Mxi2 on ERK1c phosphorylation. HEK-293T cells were transfected with GFP–ERK1c (4 μg) in addition to HA–Mxi2 (1 μg) as indicated. Cells were stimulated with EGF and phosphorylated ERK levels were examined at the indicated time points in total lysates. Lower panel, expression levels of GFP–ERK1c and HA–Mxi2. The results shown are representative of three independent experiments.

Mxi2 counteracts the effect of ERK1/2 nuclear phosphatases

The ability of Mxi2 to prolong ERK1/2 phosphorylation specifically at the nucleus could be due to Mxi2 counteracting the action of nuclear but not cytoplasmic phosphatases. To investigate this hypothesis, COS-7 cells were transfected with two nuclear phosphatases, MKP1 and DUSP5, and two cytoplasmic phosphatases, MKP3 and PTP-SL, all of which are bona fide mediators in ERK1/2 dephosphorylation [8]. As expected, all of these phosphatases were able to markedly decrease EGF-induced ERK1/2 phosphorylation levels, in comparison with vector-transfected cells (Figure 2A, upper two panels). However, in the presence of ectopic Mxi2, ERK1/2 dephosphorylation by nuclear phosphatases was completely abolished, whereas ERK1/2 dephosphorylation by the cytoplasmic phosphatases was unaltered (Figure 2A, lower two panels).

Mxi2 prevents ERK1/2 dephosphorylation by nuclear phosphatases

Figure 2
Mxi2 prevents ERK1/2 dephosphorylation by nuclear phosphatases

(A) COS-7 cells were transfected with vector (control) or HA–Mxi2 (+Mxi2) in addition to the indicated Myc-tagged nuclear (nuc) and cytoplasmic (cyt) phosphatases (1 μg each). Following starvation, cells were stimulated with EGF (100 ng/ml for 5 min) where indicated. Total and phosphorylated ERK1/2 levels were analysed by immunoblotting. (B) Phosphatase protein levels were unaffected by Mxi2. Endogenous protein levels of the indicated phosphatases were analysed by immunoblotting in total lysates from COS-7 cells transfected with (+) or without (−) HA–Mxi2 (1 μg). (C) An Mxi2 defective for binding ERK1/2 cannot prevent their dephosphorylation. Cells were transfected with HA–Mxi2 E293A, in addition to the indicated phosphatases (1 μg each). Total and phosphorylated ERK1/2 levels were analysed by immunoblotting after EGF stimulation where indicated. The results shown are representative of three independent experiments.

Figure 2
Mxi2 prevents ERK1/2 dephosphorylation by nuclear phosphatases

(A) COS-7 cells were transfected with vector (control) or HA–Mxi2 (+Mxi2) in addition to the indicated Myc-tagged nuclear (nuc) and cytoplasmic (cyt) phosphatases (1 μg each). Following starvation, cells were stimulated with EGF (100 ng/ml for 5 min) where indicated. Total and phosphorylated ERK1/2 levels were analysed by immunoblotting. (B) Phosphatase protein levels were unaffected by Mxi2. Endogenous protein levels of the indicated phosphatases were analysed by immunoblotting in total lysates from COS-7 cells transfected with (+) or without (−) HA–Mxi2 (1 μg). (C) An Mxi2 defective for binding ERK1/2 cannot prevent their dephosphorylation. Cells were transfected with HA–Mxi2 E293A, in addition to the indicated phosphatases (1 μg each). Total and phosphorylated ERK1/2 levels were analysed by immunoblotting after EGF stimulation where indicated. The results shown are representative of three independent experiments.

We wanted to understand the mechanism whereby Mxi2 was preventing the effect of nuclear phosphatases on ERK1/2 dephosphorylation. It was conceivable that Mxi2 could affect the expression of these proteins. Thus we monitored the protein levels of MKP1, DUSP5 and MKP3 in COS-7 cells transfected with Mxi2. It was found that the presence of Mxi2 did not alter the expression levels of any of these phosphatases (Figure 2B), ruling out this possibility. Since Mxi2 can bind to ERK1/2 [16], it was possible that its ability to inhibit the action of the phosphatases resided in this effect. To test this, we utilized Mxi2 E293A, a mutant form defective in binding to ERK1/2 [15]. It was observed that, in contrast with wild-type Mxi2, Mxi2 E293A did not impede the dephosphorylation of ERK1/2 by the nuclear phosphatases described above (Figure 2C). These results suggest that Mxi2 must be capable of binding to ERK1/2 in order to neutralize their dephosphorylation by nuclear phosphatases.

Mxi2 prevents nuclear phosphatases binding to ERK1/2

In light of our results, it was possible that Mxi2 could be out-competing phosphatases for binding to ERK1/2, thereby preventing their action. To evaluate this hypothesis, we analysed the ability of phosphatases to associate with ERK1/2 in the presence of Mxi2. In COS-7 cells transfected with Myc-tagged MKP1 and DUSP5, significant levels of endogenous ERK1/2 were detected in the anti-Myc immunoprecipitates. Noticeably, upon co-transfection with Mxi2, the amount of ERK1/2 found in association with the phosphatases diminished dramatically (Figure 3A), demonstrating that Mxi2 blocked the association of ERK1/2 with the nuclear phosphatases. Since ERK1/2 bind to phosphatases and to substrates through the same anchoring domains [18], we tested whether Mxi2 could also affect the capacity of ERK1/2 binding to their substrates. Most interestingly, the presence of Mxi2 did not alter the ability of ERK1/2 to complex with one of their nuclear substrates, the transcription factor Elk-1 (Figure 3B). Overall, these results demonstrate that Mxi2 sustains nuclear ERK1/2 phosphorylation by preventing the association between ERK1/2 and their cognate nuclear phosphatases by some mechanism that does not compromise ERK1/2 binding to their nuclear substrates.

Mxi2 prevents phosphatases binding to ERK1/2

Figure 3
Mxi2 prevents phosphatases binding to ERK1/2

(A) COS-7 cells were transfected with Myc-tagged MKP1 and DUSP5, with (+) or without (−) HA-Mxi2 (1 μg each) as indicated. The presence of endogenous ERK1/2 in anti-Myc (α myc) immunoprecipitates (IP) was analysed by immunoblotting. TL, total lysates. (B) Mxi2 did not prevent ERK1/2 binding to substrates. COS-7 cells were transfected with (+) or without (−) HA–Mxi2 (1 μg). The presence of endogenous Elk1 was determined by immunoblotting in total lysates (TL) and anti-ERK1/2 immunoprecipitates (IP), in cells starved or treated with EGF (100 ng/ml, 5 min) as indicated. The results shown are representative of three independent experiments.

Figure 3
Mxi2 prevents phosphatases binding to ERK1/2

(A) COS-7 cells were transfected with Myc-tagged MKP1 and DUSP5, with (+) or without (−) HA-Mxi2 (1 μg each) as indicated. The presence of endogenous ERK1/2 in anti-Myc (α myc) immunoprecipitates (IP) was analysed by immunoblotting. TL, total lysates. (B) Mxi2 did not prevent ERK1/2 binding to substrates. COS-7 cells were transfected with (+) or without (−) HA–Mxi2 (1 μg). The presence of endogenous Elk1 was determined by immunoblotting in total lysates (TL) and anti-ERK1/2 immunoprecipitates (IP), in cells starved or treated with EGF (100 ng/ml, 5 min) as indicated. The results shown are representative of three independent experiments.

Mxi2 regulates the balance between ERK1/2 nuclear and cytoplasmic signals

Previous results from our laboratory have made clear that Mxi2 can dramatically alter ERK1/2 nucleo-cytoplasmic distribution [15]. As such, it would be conceivable that the resulting re-allotment could have an impact on ERK1/2 signal output in these compartments. Following this rationale, we tested this possibility in the physiological environment provided by MDCK cells, which express Mxi2 endogenously (Figure 4A) [15]. In this cellular setting, we monitored ERK1/2 nuclear activity by assaying the phosphorylation of the transcription factor c-Myc, phosphorylated by ERK1/2 at Thr58 and Ser62 [19]. In the cytoplasm, we examined ERK1/2 activity by analysing the phosphorylation of its cytoplasmic substrate the kinase RSK1 [20]. Treatment with EGF induced the phosphorylation of c-Myc with biphasic kinetics: a sustained early peak that spanned from 5 to 30 min after stimulation, and a later peak 4 h after addition of the agonist. In the cytoplasm, RSK1 phosphorylation also occurred in a biphasic fashion, with a peak 30 min after the addition of EGF and a second sustained peak after 2–4 h (Figure 4B). We compared these effects with those in MDMx− cells, an MDCK-derived cell line devoid of Mxi2 by the stable expression of an shRNA (small hairpin RNA) [15] (Figure 4A). In these cells, we observed that the pattern of c-Myc phosphorylation induced by EGF was completely different: basal c-Myc phosphorylation levels were much lower and its phosphorylation rate was remarkably slower, taking 30 min before any phosphorylated Myc became apparent and increasing thereafter to reach a maximum after 4 h. On the other hand, the pattern of phosphorylation exhibited by RSK1 was largely unaltered in MDMx− cells compared with parental MDCK cells (Figure 4B).

Endogenous and ectopic Mxi2 expression affects ERK1/2 nuclear but not cytoplasmic signalling

Figure 4
Endogenous and ectopic Mxi2 expression affects ERK1/2 nuclear but not cytoplasmic signalling

(A) Expression levels of Mxi2 in MDCK and MDMx− cell lines and in COS-7 cells, control and expressing ectopic Mxi2 (COS/Mx), as revealed by immunoblotting with an anti-p38 antibody specific for its N-terminus [α p38 (Nt)]. (B) Effects of endogenous Mxi2 down-regulation on ERK1/2 nuclear and cytoplasmic signalling. MDCK and MDMx− cells were serum-starved for 18 h and stimulated with EGF (100 ng/ml) for the indicated periods. The levels of c-Myc, RSK1 and ERK1/2, total and phosphorylated, were analysed in total lysates by immunoblotting. (C) Effects of ectopic Mxi2 overexpression on ERK1/2 nuclear and cytoplasmic signalling. The levels of c-Myc, RSK1 and ERK1/2 were analysed in total lysates from COS-7 cells transfected with vector (COS) or with Mxi2 (COS/Mxi2) (1 μg) after stimulation with EGF for the indicated periods. The values given indicate the fold phosphorylation levels relative to untreated cells. The results shown are representative of five independent experiments. 5′ (etc.), 5 min (etc.).

Figure 4
Endogenous and ectopic Mxi2 expression affects ERK1/2 nuclear but not cytoplasmic signalling

(A) Expression levels of Mxi2 in MDCK and MDMx− cell lines and in COS-7 cells, control and expressing ectopic Mxi2 (COS/Mx), as revealed by immunoblotting with an anti-p38 antibody specific for its N-terminus [α p38 (Nt)]. (B) Effects of endogenous Mxi2 down-regulation on ERK1/2 nuclear and cytoplasmic signalling. MDCK and MDMx− cells were serum-starved for 18 h and stimulated with EGF (100 ng/ml) for the indicated periods. The levels of c-Myc, RSK1 and ERK1/2, total and phosphorylated, were analysed in total lysates by immunoblotting. (C) Effects of ectopic Mxi2 overexpression on ERK1/2 nuclear and cytoplasmic signalling. The levels of c-Myc, RSK1 and ERK1/2 were analysed in total lysates from COS-7 cells transfected with vector (COS) or with Mxi2 (COS/Mxi2) (1 μg) after stimulation with EGF for the indicated periods. The values given indicate the fold phosphorylation levels relative to untreated cells. The results shown are representative of five independent experiments. 5′ (etc.), 5 min (etc.).

We then investigated whether similar events could be reproduced in an ectopic expression system by manipulating Mxi2 levels. Unlike MDCK cells, COS-7 cells do not express Mxi2 endogenously. Thus the transfection of Mxi2 into these cells would allow us to study the effects of both suppressing physiological Mxi2 expression, as in the MDCK model, and of inducing its ectopic expression, as in COS-7 cells. Noticeably, in COS-7 cells transfected with Mxi2 the pattern of c-Myc phosphorylation induced by EGF was almost identical with that encountered in MDCK cells. Likewise, in parental COS-7 cells, lacking Mxi2, the changes in c-Myc phosphorylation were similar to those exhibited by MDMx− cells (Figure 4C). On the other hand, the phosphorylation of RSK1 followed similar oscillations regardless of the presence of Mxi2, with peaks at 30 min and 2 h after EGF stimulation. In order to ascertain whether c-Myc phosphorylation was a consequence of ERK1/2 activity, we analysed it in MDCK cells that had been treated previously with the MEK inhibitor PD98059. Inhibition of ERK1/2 activation completely prevented the early peak of EGF-induced c-Myc phosphorylation. Surprisingly, the second peak of c-Myc phosphorylation, occurring after 4 h, was unaffected by the blockade of ERK1/2 signals (Figure 5A), suggesting that some other EGF-induced kinase, maybe GSK3 (glycogen synthase kinase 3), known to phosphorylate c-Myc at the same residues as ERK [19], could be responsible for the late c-Myc phosphorylation. In summary, these results indicate that, both under physiological conditions and in a heterologous expression model, the presence of Mxi2 could profoundly influence ERK1/2 nuclear but not cytoplasmic signals.

Effects of ERK1/2 inhibition on EGF-induced c-Myc phosphorylation

Figure 5
Effects of ERK1/2 inhibition on EGF-induced c-Myc phosphorylation

(A) MDCK cells were serum-starved and pre-treated with 10 μM PD98059 1 h before stimulation with EGF for the indicated periods. The levels of phosphorylated c-Myc were analysed by immunoblotting. In the PD98059 time course, a lysate corresponding to 5 min of EGF treatment (C+) was included as a positive control for ERK1/2 phosphorylation. (B) A nuclear ERK2 does not fully the mimic Mxi2-mediated effects. Left-hand panel, lysates from vector and ERK2-NLS-transfected cells were separated into nuclear (N) and cytoplasmic (C) fractions, and ERK2 protein levels were analysed by immunoblotting. Right-hand panel, cells transfected with ERK2-NLS (1 μg) were starved and stimulated with EGF for the indicated periods. The levels of phosphorylated c-Myc were analysed by immunoblotting. The values given indicate the fold phosphorylation levels relative to untreated cells. The results shown are representative of five independent experiments. 5′ (etc.), 5 min (etc.).

Figure 5
Effects of ERK1/2 inhibition on EGF-induced c-Myc phosphorylation

(A) MDCK cells were serum-starved and pre-treated with 10 μM PD98059 1 h before stimulation with EGF for the indicated periods. The levels of phosphorylated c-Myc were analysed by immunoblotting. In the PD98059 time course, a lysate corresponding to 5 min of EGF treatment (C+) was included as a positive control for ERK1/2 phosphorylation. (B) A nuclear ERK2 does not fully the mimic Mxi2-mediated effects. Left-hand panel, lysates from vector and ERK2-NLS-transfected cells were separated into nuclear (N) and cytoplasmic (C) fractions, and ERK2 protein levels were analysed by immunoblotting. Right-hand panel, cells transfected with ERK2-NLS (1 μg) were starved and stimulated with EGF for the indicated periods. The levels of phosphorylated c-Myc were analysed by immunoblotting. The values given indicate the fold phosphorylation levels relative to untreated cells. The results shown are representative of five independent experiments. 5′ (etc.), 5 min (etc.).

Considering that the expression of Mxi2 causes an increase in ERK1/2 nuclear levels [15], it was of interest to determine whether increasing ERK1/2 protein levels at the nucleus would be sufficient to bring about the changes in c-Myc phosphorylation observed in our experiments described above. To this end, we added the SV40 (simian virus 40) NLS to the C-terminus of ERK2 in order to bolster its constitutive accumulation in the nucleus. Indeed, ERK1/2 nuclear levels were markedly enriched when the ERK2-NLS was transfected into COS-7 cells (Figure 5B, left-hand panel). In these cells, it was found that the ERK2-NLS altered the EGF-induced c-Myc phosphorylation pattern to resemble that encountered following Mxi2 transfection. However, the early phosphorylation peak was short-lived, being evident just at the 5 min time point (Figure 5B, right-hand panel), whereas Mxi2 was capable of sustaining c-Myc phosphorylation up to 30 min (Figure 4C). As such, it can be concluded that an increase in ERK1/2 nuclear levels is not sufficient to fully recapitulate the effects of Mxi2 on ERK1/2 functions at the nucleus.

A cytoplasmic Mxi2 bolsters ERK1/2 cytoplasmic signals

Finally, we investigated whether it was possible to alter ERK1/2 compartment-specific signal output by modifying the subcellular localization of Mxi2. For this purpose, we generated an Mxi2 tethered to the cytoplasm by the addition of a MEK1 NES to its N-terminus. In contrast with Mxi2, NES-Mxi2 localized almost exclusively to the cytoplasm when transfected into COS-7 cells, which subsequently led to an enrichment in ERK1/2 in that compartment (Figure 6A). Interestingly, EGF-induced c-Myc phosphorylation was remarkably different in cells transfected with Mxi2 (Figure 4C) compared with those transfected with NES-Mxi2 (Figure 6B). Noticeably, in the latter case, the c-Myc phosphorylation curve resembled that detected in control COS-7 cells (Figure 4C), indicating that NES-Mxi2 did not have an effect on ERK1/2 nuclear signals. Conversely, when ERK1/2 cytoplasmic functions were evaluated by analysing RSK1 activation, it was found that NES-Mxi2 brought about a dramatic increase in RSK1 phosphorylation, both in intensity and in duration, sustaining RSK1 phosphorylation from 30 min up to 4 h after EGF stimulation (Figure 6B). Once again, in contrast with Mxi2, this did not alter the cytoplasmic events (Figure 4C). These results demonstrate that Mxi2 could influence ERK1/2 signals, either at the nucleus or at the cytoplasm, depending on its localization.

A cytoplasm-tethered Mxi2 affects ERK1/2 cytoplasmic but not nuclear signalling

Figure 6
A cytoplasm-tethered Mxi2 affects ERK1/2 cytoplasmic but not nuclear signalling

(A) Mxi2-NES localizes to the cytoplasm. Lysates from COS-7 cells transfected with Mxi2 (wt) and Mxi2-NES (NES) (1 μg each) were separated into nuclear (N) and cytoplasmic (C) fractions, and Mxi2 and ERK2 protein levels were analysed by immunoblotting. (B) Mxi2-NES has an impact on ERK cytoplasmic functions. Cells transfected with Mxi2-NES were stimulated with EGF (100 ng/ml) for the indicated times. The levels of c-Myc and RSK1, total and phosphorylated, were analysed in total lysates by immunoblotting. (C) A cytoplasmic Mxi2 prevents ERK dephosphorylation by cytoplasmic (cyt) but not nuclear (nuc) phosphatases. COS-7 cells were transfected with Mxi2-NES in addition to the indicated Myc-tagged nuclear and cytoplasmic phosphatases (1 μg each). After starvation, cells were stimulated with EGF (100 ng/ml for 5 min) as indicated. Total and phosphorylated ERK1/2 levels were analysed by immunoblotting. The values given indicate the fold phosphorylation levels relative to untreated cells. The results shown are representative of five independent experiments. 5′ (etc.), 5 min (etc.).

Figure 6
A cytoplasm-tethered Mxi2 affects ERK1/2 cytoplasmic but not nuclear signalling

(A) Mxi2-NES localizes to the cytoplasm. Lysates from COS-7 cells transfected with Mxi2 (wt) and Mxi2-NES (NES) (1 μg each) were separated into nuclear (N) and cytoplasmic (C) fractions, and Mxi2 and ERK2 protein levels were analysed by immunoblotting. (B) Mxi2-NES has an impact on ERK cytoplasmic functions. Cells transfected with Mxi2-NES were stimulated with EGF (100 ng/ml) for the indicated times. The levels of c-Myc and RSK1, total and phosphorylated, were analysed in total lysates by immunoblotting. (C) A cytoplasmic Mxi2 prevents ERK dephosphorylation by cytoplasmic (cyt) but not nuclear (nuc) phosphatases. COS-7 cells were transfected with Mxi2-NES in addition to the indicated Myc-tagged nuclear and cytoplasmic phosphatases (1 μg each). After starvation, cells were stimulated with EGF (100 ng/ml for 5 min) as indicated. Total and phosphorylated ERK1/2 levels were analysed by immunoblotting. The values given indicate the fold phosphorylation levels relative to untreated cells. The results shown are representative of five independent experiments. 5′ (etc.), 5 min (etc.).

Finally, as NES-Mxi2 could bolster the duration and intensity of ERK1/2 cytoplasmic but not nuclear signals, we investigated whether this was due to NES-Mxi2 interfering with cytoplasmic phosphatases, while not affecting those at the nucleus. COS-7 cells were transfected with the nuclear phosphatases MKP1 and DUSP5, and the cytoplasmic phosphatases MKP3 and PTP-SL, in addition to NES-Mxi2. In this case, and in contrast with the results obtained with wild-type Mxi2 (Figure 2A), NES-Mxi2 markedly inhibited ERK1/2 dephosphorylation exerted by the cytoplasmic phosphatases, but was unable to prevent ERK1/2 dephosphorylation by nuclear phosphatases (Figure 6C).

Taken together, our results demonstrate that Mxi2 maintains the ERK1/2 phosphorylated state by its direct interaction with these kinases, thereby preventing their association with phosphatases. We also show that Mxi2 localization determines which ERK subcellular pool will be subject to this regulation.

DISCUSSION

In the present study we have taken a step forward in trying to understand the regulation of ERK1/2 phosphorylation by Mxi2. Our results demonstrate that the ability of Mxi2 to prolong the ERK1/2 phosphorylated state is spatially defined, taking place primarily at the nucleus. In contrast, the dephosphorylation rate of ERK1/2 at the cytoplasm and of ERK1c at the Golgi complex are not significantly altered in cells expressing Mxi2. These observations clearly coincide with our previous results demonstrating that Mxi2, both endogenous and ectopically expressed, is localized primarily at the nucleus, with less than 20% localized in at the cytoplasm [15]. Apparently, this Mxi2 fraction appears too scarce to effectively have an impact on ERK phosphorylation at that subcellular localization.

Unravelling the molecular mechanism whereby Mxi2 sustained ERK1/2 phosphorylation was essential. Our previous results have ruled out Mxi2 exerting a potentiating effect on MEK-mediated phosphorylation, while hinting that Mxi2 could somehow hinder the functions of ERK1/2-deactivating phosphatases [14,16]. In the present study, we have demonstrated that Mxi2 can indeed prevent the dephosphorylation of ERK1/2 undertaken by nuclear phosphatases, such as MKP1 and DUSP5, without affecting their expression levels. Furthermore, we show that Mxi2 is competent for counteracting cytoplasmic phosphatases such as MKP3 and PTP-SL, provided that it can be present at sufficiently high levels in such compartment. These results suggest that the maintenance of ERK1/2 phosphorylation at the nucleus by Mxi2 does not reflect any specific effect on nuclear phosphatases, but rather points to a simple phenomenon of co-localization at the same cellular compartment, a consequence of the preferential localization of Mxi2 at the nucleus.

Importantly, we have shown that the ability of Mxi2 to preclude ERK1/2 dephosphorylation resides in its capacity to directly interact with ERK1/2, since Mxi2 E293A, a mutant form defective in binding ERK1/2, cannot counteract the action of the phosphatases. In agreement with these observations, we have shown that Mxi2 inhibits ERK1/2 binding to phosphatases. This inhibitory effect probably entails a competition for binding sites. We have shown previously that Mxi2 binds to the insert region of ERK1/2 [15], and reports indicate that phosphatases, such as MKP3, also require this region for the formation of a productive complex whereby ERK1/2 can be dephosphorylated [21]. Since the affinity of Mxi2 for ERK1/2 (Kd ~20 nM [16]) is greater that that of phosphatases such as MKP3 (Kd ~180 nM [21]), it can be envisioned that Mxi2 could be capable of dislodging phosphatases from ERK1/2 complexes. Thus our results unveil a novel mechanism whereby ERK1/2 dephosphorylation can be regulated.

Most interestingly, we have demonstrate that, although capable of displacing phosphatases from ERK1/2 complexes, Mxi2 does not prevent ERK1/2 binding to substrates, something that results in the potentiation of their phosphorylation/activation, as demonstrated previously for the transcription factor Elk-1 [16] and as we show in the present study for c-Myc. This lack of interference cannot be explained on the basis of Mxi2 and substrates requiring different sites for binding to ERK1/2, since, as demonstrated previously [22], Elk-1 also requires the ERK1/2 insert region for efficient binding. Once again, it is likely that differences in affinities can account for this phenomenon. As reported previously, multiple binding sites participate in the ERK1/2 interaction with substrates [18,2325], possibly acting in synergy that probably results in a higher binding affinity compared with Mxi2.

The expression of Mxi2 has direct consequences on ERK1/2 signal output at the nucleus. We have shown that Mxi2, both endogenous and ectopically expressed, causes a faster, stronger and prolonged EGF-induced ERK1/2 nuclear signal, as evaluated by the phosphorylation of c-Myc. On the other hand, Mxi2 expression has little impact on ERK1/2-mediated cytoplasmic events, such as the phosphorylation of RSK1. As we have reported previously, Mxi2 has dramatic effects on the nucleo-cytoplasmic distribution of ERK1/2 [15]; however, the sole alteration of ERK1/2 levels in these compartments is not sufficient to account for the observed effects. For example, in MDMx− cells, down-regulation of Mxi2 expression results in a remarkable increase in ERK1/2 cytoplasmic levels [15], something that does not alter the pattern of phosphorylation of RSK1. Furthermore, we have shown that artificially augmenting the presence of ERK1/2 at the nucleus by the addition of an NLS does not entirely mimic the effects of Mxi2: elevated ERK1/2 nuclear levels resulting from ERK2-NLS expression could, to some extent, saturate the available nuclear phosphatases, thereby delaying the inactivation of the ERK1/2 nuclear pool, but a sustained early peak of c-Myc phosphorylation cannot be achieved, as accomplished by Mxi2. This suggests that a direct ERK–Mxi2 interaction, thereby permanently preventing ERK1/2 binding to its deactivating phosphatases, is required for ERK1/2 signals to be sustained. In support of this notion, artificially directing Mxi2 to the cytoplasm by means of an NES, markedly augments and prolongs RSK1 phosphorylation.

It is well known that, depending on the cell type, there are broad differences in the relative distribution and in the activity levels that ERK1/2 display at the nucleus and at the cytoplasm. Even though we do not fully understand the reasons underlying this variability, it is conceivable that it could represent a mechanism to promote/quench ERK1/2 signals at these compartments. Thereby the repertoire of biochemical/biological responses given to a specific stimulus could be increased. The promotion of ERK1/2 signalling at the cytoplasm, by proteins such as PEA15 (phosphoprotein-enriched in astrocytes 15) [26] and most scaffold proteins [27,28], has been described. On the other hand, Mxi2 is a clear representative of proteins that enhance ERK1/2 signalling at the nucleus. However, we demonstrate that, if present in the cytoplasm, Mxi2 can markedly influence ERK1/2 functions therein. As such, it is conceivable that biological processes affecting Mxi2 localization could exert a profound impact on the balance between ERK1/2 nuclear and cytoplasmic signals and, subsequently, remarkably altering key biological outputs.

Abbreviations

     
  • DMEM

    Dulbecco's modified Eagle's medium

  •  
  • DTT

    dithiothreitol

  •  
  • DUSP

    dual-specificity phosphatase

  •  
  • EGF

    epidermal growth factor

  •  
  • ERK

    extracellular-signal-regulated kinase

  •  
  • GFP

    green fluorescent protein

  •  
  • HA

    haemagglutinin

  •  
  • HEK

    human embryonic kidney

  •  
  • MAPK

    mitogen-activated protein kinase

  •  
  • MDCK

    Madin–Darby canine kidney

  •  
  • MEK

    MAPK/ERK kinase

  •  
  • MKP

    MAPK phosphatase

  •  
  • NES

    nuclear export signal

  •  
  • NLS

    nuclear localization signal

  •  
  • NP-40

    Nonidet P40

  •  
  • PTP-SL

    striatal-enriched phosphatase-like protein tyrosine phosphatase

  •  
  • RSK1

    ribosomal S6 kinase 1

AUTHOR CONTRIBUTION

Berta Casar performed all of the experiments, with the exception of those shown in Figure 1(B), which were performed by Gilad Gibor, and in Figure 5(A), which were performed by Javier Rodríguez. Berta Casar and Javier Rodríguez also prepared the Figures and performed the statistical analyses. Piero Crespo conceived the study and wrote the paper. Rony Seger contributed to the conception of the study and revision of the paper.

We are indebted to Dr S.M. Keyse, Dr R. Perona and Dr R. Pulido for providing reagents.

FUNDING

This work was supported by the Spanish Ministry of Education [grant number BFU2008-01728], the European Union Sixth Framework Programme under the GROWTHSTOP project [grant number LSHC CT-2006-037731] and the Red Temática de Investigación Cooperativa en Cáncer (RTICC) [grant number RD06/0020/0105], the Fondo de Investigaciones Sanitarias (FIS), Carlos III Institute, Spanish Ministry of Health. R.S. is an incumbent of the Yale S. Lewine and Ella Miller Lewine professorial chair for cancer research.

References

References
1
Raman
M.
Chen
W.
Cobb
M. H.
Differential regulation and properties of MAPKs
Oncogene
2007
, vol. 
26
 (pg. 
3100
-
3112
)
2
Yoon
S.
Seger
R.
The extracellular signal-regulated kinase: multiple substrates regulate diverse cellular functions
Growth Factors
2006
, vol. 
24
 (pg. 
21
-
44
)
3
Robinson
M. J.
Stippec
S. A.
Goldsmith
E.
White
M. A.
Cobb
M. H.
A constitutively active and nuclear form of MAP kinase ERK2 is sufficient for neurite outgrowth and cell transformation
Curr. Biol.
1998
, vol. 
8
 (pg. 
1141
-
1150
)
4
Ajenjo
N.
Canon
E.
Sanchez-Perez
I.
Matallanas
D.
Leon
J.
Perona
R.
Crespo
P.
Subcellular localization determines the protective effects of activated ERK2 against distinct apoptogenic stimuli in myeloid leukemia cells
J. Biol. Chem.
2004
, vol. 
279
 (pg. 
32813
-
32823
)
5
Keyse
S. M.
Protein phosphatases and the regulation of mitogen-activated protein kinase signalling
Curr. Opin. Cell Biol.
2000
, vol. 
12
 (pg. 
186
-
192
)
6
Alessi
D. R.
Gomez
N.
Moorhead
G.
Lewis
T.
Keyse
S. M.
Cohen
P.
Inactivation of p42 MAP kinase by protein phosphatase 2A and a protein tyrosine phosphatase, but not CL100, in various cell lines
Curr. Biol.
1995
, vol. 
5
 (pg. 
283
-
295
)
7
Pulido
R.
Zuniga
A.
Ullrich
A.
PTP-SL and STEP protein tyrosine phosphatases regulate the activation of the extracellular signal-regulated kinases ERK1 and ERK2 by association through a kinase interaction motif
EMBO J.
1998
, vol. 
17
 (pg. 
7337
-
7350
)
8
Owens
D. M.
Keyse
S. M.
Differential regulation of MAP kinase signalling by dual-specificity protein phosphatases
Oncogene
2007
, vol. 
26
 (pg. 
3203
-
3213
)
9
Yao
Z.
Seger
R.
The molecular mechanisms of MAPK/ERK inacativation
Curr. Genomics
2004
, vol. 
5
 (pg. 
385
-
393
)
10
Brondello
J. M.
Pouyssegur
J.
McKenzie
F. R.
Reduced MAP kinase phosphatase-1 degradation after p42/p44MAPK-dependent phosphorylation
Science
1999
, vol. 
286
 (pg. 
2514
-
2517
)
11
Camps
M.
Nichols
A.
Gillieron
C.
Antonsson
B.
Muda
M.
Chabert
C.
Boschert
U.
Arkinstall
S.
Catalytic activation of the phosphatase MKP-3 by ERK2 mitogen-activated protein kinase
Science
1998
, vol. 
280
 (pg. 
1262
-
1265
)
12
Zervos
A. S.
Faccio
L.
Gatto
J. P.
Kyriakis
J. M.
Brent
R.
Mxi2, a mitogen-activated protein kinase that recognizes and phosphorylates max protein
Proc. Natl. Acad. Sci. U.S.A.
1995
, vol. 
92
 (pg. 
10531
-
10534
)
13
Faccio
L.
Chen
A.
Fusco
C.
Martinotti
S.
Bonventre
J. V.
Zervos
A. S.
Mxi2 a splice variant of p38 stress-activated kinase is a distal nephron protein regulated with kidney ischemia
Am. J. Physiol. Cell Physiol.
2000
, vol. 
278
 (pg. 
C781
-
C791
)
14
Sanz
V.
Arozarena
I.
Crespo
P.
Distinct carboxi-termini confer divergent characteristics to the mitogen-activated protein kinase p38α and its splice isoform Mxi2
FEBS Letts.
2000
, vol. 
474
 (pg. 
169
-
174
)
15
Casar
B.
Sanz-Moreno
V.
Yazicioglu
M. N.
Rodriguez
J.
Berciano
M. T.
Lafarga
M.
Cobb
M. H.
Crespo
P.
Mxi2 promotes stimulus-independent ERK nuclear translocation
EMBO J.
2007
, vol. 
26
 (pg. 
635
-
646
)
16
Sanz-Moreno
V.
Casar
B.
Crespo
P.
p38α isoform Mxi2 binds to extracellular signal-regulated kinase 1 and 2 mitogen-activated protein kinase and regulates its nuclear activity by sustaining its phosphorylation levels
Mol. Cell. Biol.
2003
, vol. 
23
 (pg. 
3079
-
3090
)
17
Aebersold
D. M.
Shaul
Y. D.
Yung
Y.
Yarom
N.
Yao
Z.
Hanoch
T.
Seger
R.
Extracellular signal-regulated kinase 1c (ERK1c), a novel 42-kilodalton ERK, demonstrates unique modes of regulation, localization, and function
Mol. Cell. Biol.
2004
, vol. 
24
 (pg. 
10000
-
10015
)
18
Jacobs
D.
Glossip
D.
Xing
H.
Muslin
A. J.
Kornfeld
K.
Multiple docking sites on substrate proteins form a modular system that mediates recognition by ERK MAP kinase
Genes Dev.
1999
, vol. 
13
 (pg. 
163
-
175
)
19
Pulverer
B. J.
Fisher
C.
Vousden
K.
Littlewood
T.
Evan
G.
Woodgett
J. R.
Site-specific modulation of c-Myc cotransformation by residues phosphorylated in vivo
Oncogene
1994
, vol. 
9
 (pg. 
59
-
70
)
20
Frodin
M.
Gammeltoft
S.
Role and regulation of 90 kDa ribosomal S6 kinase (RSK) in signal transduction
Mol. Cell. Endocrinol.
1999
, vol. 
25
 (pg. 
65
-
77
)
21
Zhou
B.
Zhang
J.
Liu
S.
Reddy
S.
Wang
F.
Zhang
Z. Y.
Mapping ERK2-MKP3 binding interfaces by hydrogen/deuterium exchange mass spectrometry
J. Biol. Chem.
2006
, vol. 
281
 (pg. 
38834
-
38844
)
22
Lee
T.
Hoofnagle
A. N.
Kabuyama
Y.
Stroud
J.
Min
X.
Goldsmith
E. J.
Chen
L.
Resing
K. A.
Ahn
N. G.
Docking motif interactions in MAP kinases revealed by hydrogen exchange mass spectrometry
Mol. Cell
2004
, vol. 
14
 (pg. 
43
-
55
)
23
Tanoue
T.
Adachi
M.
Moriguchi
T.
Nishida
E.
A conserved docking motif in MAP kinases common to substrates, activators and regulators
Nat. Cell Biol.
2000
, vol. 
2
 (pg. 
110
-
116
)
24
Fantz
D. A.
Jacobs
D.
Glossip
D.
Kornfeld
K.
Docking sites on substrate proteins direct extracellular signal-regulated kinase to phosphorylate specific residues
J. Biol. Chem.
2001
, vol. 
276
 (pg. 
27256
-
27265
)
25
Barsyte-Lovejoy
D.
Galanis
A.
Sharrocks
A. D.
Specificity determinants in MAPK signaling to transcription factors
J. Biol. Chem.
2002
, vol. 
277
 (pg. 
9896
-
9903
)
26
Formstecher
E.
Ramos
J. W.
Fauquet
M.
Calderwood
D. A.
Hsieh
J. C.
Canton
B.
Nguyen
X. T.
Barnier
J. V.
Camonis
J.
Ginsberg
M. H.
Chneiweiss
H.
PEA-15 mediates cytoplasmic sequestration of ERK MAP kinase
Dev. Cell
2001
, vol. 
1
 (pg. 
239
-
250
)
27
Casar
B.
Pinto
A.
Crespo
P.
Essential role of ERK dimers in the activation of cytoplasmic but not nuclear substrates by ERK-scaffold complexes
Mol. Cell
2008
, vol. 
31
 (pg. 
708
-
721
)
28
Casar
B.
Pinto
A.
Crespo
P.
ERK dimers and scaffold proteins: unexpected partners for a forgotten (cytoplasmic) task
Cell Cycle
2009
, vol. 
8
 (pg. 
1007
-
1013
)