| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
,2
* Laboratory for Toxicopathology, Institute of Pathology, Rikshospitalet-Radiumhospitalet Medical Centre; and
Department of Nutrition Research, Institute of Basal Medical Sciences, University of Oslo, Oslo, Norway
1Correspondence: Laboratory for Toxicopathology, Institute of Pathology, Rikshospitalet-Radiumhospitalet Medical Centre, University Hospital, N-0027 Oslo, Norway. E-mail: ellen.skarpen{at}medisin.uio.no
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Key Words: proliferation • survival • hepatocytes • adhesion
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
. When ERK is activated by nonmitogenic signals, ERK remains in the cytoplasm. From the cytoplasm, active ERK is able to enhance survival during oxidative stress (2)
, whereas nuclear accumulation is required for induction of cyclin D1 and entry into the S-phase (3
, 4)
. ERK-mediated cell cycle progression depends on stimulation by growth factors in synergism with anchoring of cells to an appropriate extracellular matrix (ECM), with concurrent organization of the cytoskeleton (5
, 6)
.
Cell adhesion to the ECM is largely mediated by the integrin family of transmembrane receptors. Integrins associate with, or activate, a number of cytosolic kinases, and these kinases initiate many of the downstream integrin signaling events. Adhesion-dependent Rac-signaling from the β1 integrin cytodomain, mediated through the Rac effector protein p21-activated kinase (PAK), is necessary to integrate signals directed to promote ERK nuclear accumulation (7)
. In response to cell attachment, PAK1 phosphorylates Raf on its S338 site (8)
, and MEK1 on S298 (9
, 10)
. These modifications enhance signaling through the Raf-MEK-ERK module (9
10
11)
.
MEK1 and MEK2 bind directly to ERK2 through a region in the N terminus of MEK (1
, 12
, 13)
. In addition, a proline-rich (PR) regulatory sequence in MEK is also involved in MEK-ERK association and signal propagation (10
11
12
, 14
, 15)
. The coupling between MEK1 and ERK2 is enhanced through phosphorylation on S298 in the MEK1 PR region, whereas phosphorylation on MEK1 T292 releases the complex (15)
. MEK1 T292 is a substrate of ERK2, but the site is also phosphorylated at a basal level when ERK2 is inhibited, suggesting several regulators of this site (15)
. Although the S298 site in MEK2 has been conserved, it lacks the T292 phosphorylation site, and it is not a substrate of PAK1 (9)
. Consequently, during cellular adhesion, and in response to Rac-PAK-signaling, the association between MEK1 and ERK2, but not MEK2 and ERK2, is increased (14)
. However, in nonadherent cells, ERK2 preferentially binds to MEK2 compared to MEK1 (14)
. Thus, MEK represents an adhesion-regulated step in growth factor signaling.
Here, we show that MEK1 and MEK2 sort ERK2 to different intracellular compartments following growth factor treatment. When activated through MEK1, ERK2 translocated to the nucleus. In contrast, MEK2-activated ERK2 preferably located to the cytoplasm. MEK1-mediated ERK2 relocation from cytoplasm to nuclei depended on phosphorylation of MEK1 on both S298 and T292. Furthermore, sorting of ERK2 to the nucleus by MEK1 correlated with a proliferative response, whereas MEK2-activated ERK2-mediated survival from the cytoplasm. A shift in the MEK1 biological response from proliferation to survival was obtained when a nuclear export sequence (NES) was attached to ERK2. We conclude that ERK2 intracellular localization determines whether growth factors mediate proliferation or survival and that the sorting occurs in an adhesion-dependent manner at the level of MEK1 and MEK2.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Animals, hepatocyte cultures, and transfections
Young adult male Wistar rats (Møllergaard and Blomhoff, Odense, Denmark) weighing 200–220 g were kept on a 12 h light-dark cycle and fed water ad libitum. Hepatocytes were isolated and seeded as described previously (16)
. Transfection was carried out with DOTAP, according to the manufacturer’s protocol. On average, one of three cells did express the various constructs when examined by microscopy based on identification with tags.
Plasmids
The plasmid pCMV-FLAG-7.1-ERK2 was a kind gift from Dr. M. Cobb (University of Texas Southwestern Medical Center, Dallas, TX, USA); the pEGFPC1-ERK2 was a generous gift from Dr. R. Seger (The Weizmann Institute of Science, Rehovot, Israel); the pMCL-HA-MEK1, pMCL-HA-MEK1–8E (dominant negative MEK1), pMCL-HA-MEK1-R4F (constitutive active MEK1), pMCL-HA-MEK2, and pMCL-HA-MEK2-KW71 (constitutive active MEK2) were kindly provided by Dr. N. Ahn (University of Colorado, Boulder, CO, USA); the pRc/CMV-HA-MEK2-K101A (dominant negative MEK2) was generously provided by Dr. S. Meloche (University of Montreal, Montreal, QC, Canada); the pCHA-MEK1-S298A and the pCHA-MEK1-T292A plasmids were kind gifts from Dr. A. D. Catling (University of Virginia Medical School, Charlottesville, VA, USA); the pMT2HA-RSK1 plasmid was kindly provided by Dr. M. Frödin (Biotech Research & Innovation Centre, Copenhagen, Denmark); and the pEYFP-Mem/gap43 was purchased from Clontech Laboratories (Mountain View, CA, USA).
The plasmid pRK5-Myc-Rac1 was a generous gift from Dr. A. Hall (University College London, London, UK). Rac1 was subcloned into pBluescript, and the mutant Rac1-T17N was generated using QuickChange site-directed mutagenesis kit from Stratagene (La Jolla, CA, USA), according to the manufacturer’s recommendations. T17N designed oligo 5'-GCT GTA GGT AAA AAT TGC CTA CTG ATC-3'. The construct was fully sequenced and the cDNAs subcloned as a BamHI–EcoRI fragment into pRK5-Myc.
The plasmid pEGFPC1-ERK2-AAA was created by the exchange of T183, E184, and Y185 in the kinase domain of the activation loop to alanine using QuickChange Site-Directed Mutagenesis Kit from Stratagene and designed oligonucleotide primers from Sigma-Aldrich: raterk2aaaf 5'-CAT ACA GGG TTC TTG GCA GCT GCA GTA GCC ACG CTG TGG-3' and raterk2aaar 5'-CCA ACG CGT GGC TAC TGC AGC TGC CAA GAA CCC TGT ATG-3'. The mutated rat ERK2 was introduced to a pEGFC1 vector from Clontech. The mutant clone was verified by sequencing analysis at Eurofins Medigenomix GmbH (Martinsried, Germany).
The plasmid pEGFPC1-ERK2-NES was created by adding a nuclear export signal (NES) to pEGFPC1 using ExSite PCR-Based Site-Directed Mutagenesis Kit from Stratagene and oligonucleotide primers from Sigma-Aldrich: 5'-GAA GCT TGA GCT CGA GAT CTG-3' and 5'-G CTG CAG AAG AAG CTG GAG GAG CTG GAA CTT TCG AAT TCT GCA GTC GAC GGT AC-3'. The nucleotide sequence of the pEGFPC1-ERK2-NES clone was verified by sequencing analysis at Eurofins Medigenomix GmbH (Martinsried, Germany).
Western blot analysis, measurement of DNA synthesis, and confocal fluorescence microscopy
Both Western blot analysis, DNA synthesis measurements, and confocal fluorescence microscopy were performed essentially as described previously (16)
. In Western blot analysis, equal protein loading was controlled with an antibody to β-tubulin. For calculations of the mean percentage of EGF-induced ERK2-GFP nuclear translocation, confocal microscopy of 100 cells were counted as either 1) strong ERK2-GFP-positive nuclei characterized by an exclusive nuclear localization of ERK2-GFP with negative cytoplasm, 2) positive ERK2-GFP nuclei characterized by both a distinct nuclear localization and the presence of ERK2-GFP in the cytoplasm), or 3) negative ERK2-GFP nuclei characterized by the lack of distinct ERK2-GFP nuclear localization with cytoplasmic staining. Cells from control and EGF-stimulated cells from three independent experiments were counted.
Detection of caspase-3 activity
Cells cultured in 6-well plates were washed once with PBS and harvested in 300 µl modified RIPA buffer (50 mM Tris–HCl, pH 7.4, 1% Igepal, 0.25% sodium deoxycholate, 150 mM NaCl, 1 mM EGTA) for 5 min at RT. The cell lysate was transferred to a 96-well plate and supplemented with 20 µM caspase-3 fluorometric substrate Ac-DEVD-amc. Modified RIPA buffer with 20 µM substrate was used as a negative control. The cell lysates were incubated for 2 h at 37°C before fluorescence was measured at 460 nm (Perkin-Elmer HTS 7000 Plus, Bio Assay Reader). Protein was measured as described above.
Statistics
Qualitative data are presented as mean and SEs of the mean (SEM). Statistically significant differences between treatments were analyzed with paired sample Wilcoxon Ranks Test. All calculations were made with SPSS® version 14.0 for Windows (Statistical Packages for the Social Sciences, Chicago, IL, USA). A value of P < 0.05 was considered significant.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
. Nuclear accumulation of ERK2 is an adhesion-dependent step (5
, 6)
. Also, cellular adhesion is known to enhance S298 phosphorylation of MEK1 through the activation of Rac and PAK1 (7
, 9
, 11)
. Although modulation on S298 strengthens the interaction between MEK1 and ERK2, phosphorylation of the ERK-regulated site T292 releases the MEK1-ERK2 complex (14
, 15)
. The possibility that these opposing phosphorylation events could represent the adhesion-dependent mechanism that regulate growth factor-induced ERK2 nuclear translocation was investigated in cultured hepatocytes with the mutants MEK1-S298A and MEK1-T292A.
Nuclear accumulation of ERK was investigated when activated by MEK1wt compared to when activated by MEK1 mutated at S298 and T292. Cultured hepatocytes were transiently transfected with HA-tagged MEK1 constructs, and two-color immunofluorescence staining was performed on the cells with an antibody to ERK (red color) combined with an antibody to HA (green color). This allowed us to examine the localization of endogenous ERK in the transfected hepatocytes only. As shown in Fig. 1
A, ERK was localized in the cytoplasm of control cells in MEK1wt- (Fig. 1Aa
), MEK1-S298A- (Fig. 1Ad
), and MEK1-T292A-expressed cells (Fig. 1Ag
), and in the surrounding hepatocytes. In response to EGF, ERK accumulated strongly in the nucleus of both surrounding and MEK1wt-transfected cells (Fig. 1Ab, c
). Comparably, hepatocytes that expressed the MEK1 mutants S298A (Fig. 1Ae, f
) and T292A (Fig. 1Ah, i
), showed less ERK nuclear accumulation.
|
To further study ERK and MEK interactions, confocal microscopy was performed on hepatocytes cotransfected with ERK2wt-GFP together with HA-tagged MEK1wt, MEK1-S298A, and MEK1-T292A, which allowed the preferable activation of ERK2-GFP through MEK1wt, and the MEK1 mutants, respectively (Fig. 1B
). In control cells (Fig. 1Ba, e, i
), ERK2wt-GFP (green color) was evenly distributed throughout the cells and colocalized (yellow color) with HA-stained MEK1 (red color). In response to EGF, ERK2wt-GFP translocated strongly from the cytoplasm to the nuclei in hepatocytes coexpressed with MEK1wt (Fig. 1Bc, d
), which resulted in less yellow staining compared to control cells (compare Fig. 1Bc
with Fig. 1Ba
). However, when activated by the MEK1 mutants S298A (Fig. 1Bg, h
) and T292A (Fig. 1Bk, l
), ERK2wt-GFP nuclear translocation was impaired, and no reduction in yellow color compared to control cells was found (compare Fig. 1Bg
with Fig. 1Be
and Fig. 1Bk
with Fig. 1Bi
). In Fig. 1C
, the percentage of ERK2wt-GFP that was retained in the cytoplasm when activated through MEK1wt, MEK1-S298A, or MEK1-T292A is illustrated. The columns represent the percentage of cells from three independent experiments that showed either cytoplasmic localization alone, or cytoplasmic and nuclear staining of ERK2wt-GFP. In control cells transfected with MEK1wt, ERK2wt cytoplasmic staining was found in 90% of the cells. In response to EGF, the cytoplasmic staining was reduced to 50%, compared to only 96% of the MEK1-S298A-transfected cells and 91% of the MEK1-T292A-transfected hepatocytes. Thus, mutations at the MEK1 S298 and T292 sites caused cytoplasmic retention of ERK2wt-GFP in response to EGF.
Next, we examined whether the activation of ERK2wt was reduced when activated by the MEK1 mutants compared to MEK1wt, since this could lead to reduced nuclear translocation. Also, previous studies have suggested that modulation on MEK1 S298 enhances growth factor-mediated ERK-activation (9
10
11)
. Thus, Western immunoblotting with an antibody recognizing ERK phosphorylated at its MEK consensus site, T-E-Y, was used to compare ERK2wt activated by MEK1wt, MEK1-S298A, or MEK-T292. As shown in Fig. 1D
, a similar level of ERK2wt phosphorylation was found when activated by MEK1wt, MEK1-S298A, or MEK1-T292A. Thus, neither the S298 nor the T292 site influenced the level of EGF-induced ERK2 phosphorylation, only the ERK2 nuclear translocation.
MEK1 translocates ERK2 to the nucleus more efficiently than MEK2
Having established that the S298 and T292 sites in MEK1 are essential for growth factor-induced ERK2 nuclear accumulation, we hypothesized that MEK2, which lacks the T292 site and is a poor substrate of PAK1, to a lesser extent than MEK1 was able to relocate ERK2 from the cytoplasm to the nucleus in response to growth factor. First, we wanted to establish that phosphorylation of S298 was restricted to MEK1 also in primary cultures of hepatocytes. Western immunoblotting with an antibody recognizing MEK phosphorylated at S298 was performed on hepatocyte lysates from cells transfected with MEK1wt or MEK2wt in the absence or presence of Racwt. As shown in Fig. 2
A, MEK1, but not MEK2, was phosphorylated at S298, and the level was enhanced in the presence of Racwt. The S298 phosphorylation of MEK1 was not elevated in EGF-stimulated cells, which indicated that S298 was not a site for direct growth factor regulation, as described previously (15
, 17)
.
|
Nuclear accumulation of ERK when activated by MEK2 was studied in hepatocytes transiently transfected with HA-tagged MEK2. As shown in Fig. 2B
, endogenous ERK (red color) was localized in the cytoplasm of control cells (Fig. 2Ba
) of both surrounding and MEK2wt-expressed cells. In response to EGF (Fig. 2Bb, c
), the ERK nuclear accumulation was impaired in the MEK2wt-expressed cells (green color).
The individual interaction of MEK2 with ERK2 was further examined by confocal fluorescence microscopy of hepatocytes cotransfected with ERK2wt-GFP in combination with HA-tagged MEK2wt (Fig. 2C
). In the absence of EGF (Fig. 2Ca, b
), ERK2wt-GFP (green color) was distributed throughout the whole cell and colocalized (yellow color) with MEK2wt (red color). When stimulated with EGF (Fig. 2Cc, d
), ERK2wt-GFP was mostly retained in the cytoplasm colocalized with MEK2wt (yellow color).
In Fig. 2D
, the percentage of ERK2wt-GFP with cytoplasmic localization when activated through MEK2wt compared to MEK1wt is illustrated. In control cells transfected with MEK2wt, cytoplasmic staining was found in 94% of the cells, compared to 90% in MEK1wt-transfected cells. When stimulated by EGF, an average of 88% of MEK2wt-expressing cells showed cytoplasmic retention compared to only 50% of the MEK1wt-transfected cells. Thus, when activated by EGF through MEK2, ERK2 was mostly retained in the cytoplasm.
To confirm that MEK1 and MEK2 activated ERK2 to a similar level, Western immunoblotting was performed with an antibody recognizing ERK phosphorylated at the MEK T-E-Y consensus site (Fig. 2E
). A similar level of ERK2wt phosphorylation was observed when activated by EGF through MEK1wt compared to MEK2wt.
DNA synthesis is lower in cells expressing MEK2wt, MEK1-S298A, or T292A-mutants compared to MEK1wt
When activated by potent growth factors, ERK translocates from the cytoplasm to the nucleus, where it phosphorylates transcription factors like c-Myc and Elk1. This event leads to growth progression, allowing cells to proceed through the S-phase (3
, 4)
. We wanted to examine whether cytoplasmic retention of ERK2 correlated with reduced ability to initiate cell cycle progression. Thus, activation of c-Myc was examined by confocal immunofluorescence staining with an antibody recognizing activated c-Myc in hepatocytes transfected with ERK2wt-GFP combined with MEK1wt, MEK1-S298A, MEK1-T292A, or MEK2wt. As shown in Fig. 3
A, control cells from hepatocytes transfected with all four combinations (Fig. 3Aa, d, g, j
) showed low levels of pMyc staining. When stimulated with EGF, increased pMyc staining was found in hepatocytes transfected with MEK1wt (Fig. 3Ab, c
). Comparably less pMyc staining was found in hepatocytes transfected with MEK1-S298A (Fig. 3Ae, f
), MEK1-T292A (Fig. 3Ah, i
), and MEK2wt (Fig. 3Ak, l
). Thus, nuclear accumulation of ERK2 correlated with pMyc activation.
|
Next, DNA synthesis in response to EGF was measured in hepatocytes transiently transfected with MEK1wt compared to MEK2wt, or MEK1-S298A, MEK1-T292A, or dominant negative MEK1–8E (Fig. 3B
). Compared to hepatocytes transfected with MEK1wt, both MEK2wt, MEK1-S298A, MEK1-T292A, and MEK1–8E-expressed cells showed a significant reduction in DNA synthesis. Thus, reduced EGF-induced DNA synthesis correlated with reduced ERK2 nuclear translocation. The amount of thymidine incorporated into DNA was measured in the total pool of hepatocytes, both transfected and untransfected, since primary hepatocytes are unable to reattach after sorting. On average, 
1/3 of the hepatocytes did express the construct of interest, which must be taken into consideration during interpretation of the data. Comparative protein expression of the various MEK constructs is shown in Fig. 3C
.
MEK2, but not MEK1, provides survival from TGFβ-induced apoptosis
In response to oxidative stress in hepatocytes, ERK is activated but does not accumulate in the nucleus to enhance proliferation. However, in the cytoplasm, ERK is able to mediate survival signals by activation of RSK, which relocates to the nucleus to enhance survival (2)
. Since MEK2-activated ERK2 to a large extent retained in the cytoplasm, we investigated whether MEK2 was a stronger activator of RSK1 than MEK1. Confocal fluorescence immunostaining of RSK1 was studied in hepatocytes transiently expressed with constitutive active MEK1 and MEK1. As shown in Fig. 4
A, nuclear localization of RSK1 (red color) was more pronounced in cells transfected with MEK2-CA (Fig. 4Ac, d
) than in MEK1-CA-expressed cells (Fig. 4Aa, b
).
|
Next, we wanted to examine whether MEK2 was a stronger provider of survival than MEK1. Apoptosis was measured by Caspase-3-activity in TGFβ-exposed hepatocytes transfected with constitutive active (CA) mutants of MEK1 or MEK2 (Fig. 4B
). The total cell population was measured, consisting of 1/3 of transfected cells overexpressing MEK1-CA or MEK2-CA, and 2/3 of surrounding hepatocytes. When hepatocytes were transfected with MEK1-CA, enhanced survival from TGFβ-induced apoptosis was not found compared to nontransfected cells. However, when MEK2-CA was overexpressed, a reduction in the TGFβ-induced caspase-3-level was found. Thus, MEK2, but not MEK1, provided survival from TGFβ-induced apoptosis. In C, comparable protein expression of MEK1-CA and MEK2-CA constructs is shown.
Characterization of activation and nuclear import/export of ERK2-NES
To determine whether ERK2 localization was the point of divergence between MEK1 and MEK2 signaling, we constructed an ERK2 with a nuclear export signal, ERK2-NES. By activation of ERK2-NES through MEK1, the MEK1 signals will be forced to the cytoplasm. We applied this model to study whether this will switch the MEK1-function from proliferation to survival. To characterize the mutant, confocal microscopy was performed on hepatocytes cotransfected with ERK2-NES-GFP and MEK1wt. In control cells, we found ERK2-NES-GFP primarily in the cytoplasm. In response to EGF, only low levels were found in the nuclei when compared to ERK2wt-GFP (compare Fig. 5
Ab with Fig. 1Bd
). Further, when an inactivable ERK2-AAA-GFP mutant was combined with MEK1, the mutant was located throughout the whole cell in both control (Fig. 5Ac
) and EGF-exposed (Fig. 5Ad
) hepatocytes. The proportion of cells with cytoplasmic ERK2-GFP staining in control and EGF-stimulated cells is presented in Fig. 5B
. An average of 97% of the cells transfected with ERK2-NES-GFP showed cytoplasmic retention in response to EGF, compared to 50% with ERK2wt-GFP, and 99% with ERK2-AAA-GFP. Thus, ERK2-NES-GFP showed a cytoplasmic staining pattern when activated through MEK1. To characterize the intracellular movement of ERK2-NES-GFP immediately after EGF stimulation, confocal fluorescence microscopy was performed on ERK2-NES-GFP-transfected hepatocytes treated with leptomycin B, an inhibitor of NES-regulated nuclear export (18)
. With leptomycin B alone, ERK2-NES-GFP was evenly distributed in the cytoplasm (Fig. 5Ca
). In leptomycin B-treated cells stimulated with EGF (Fig. 5Cb
), ERK2-NES-GFP showed strong staining in the nuclei. Thus, ERK2-NES-GFP entered the nucleus in response to EGF but rapidly shuttled back to the nucleus due to its NES site.
|
To confirm that the mutant could be activated to the same extent as the ERK2wt, Western immunoblotting was performed on lysates from hepatocytes transfected with ERK2wt-GFP or ERK2-NES-GFP, combined with MEK1wt or MEK2wt. As shown (Fig. 5D
), ERK2-NES-GFP was phosphorylated at the same level as ERK2wt-GFP, both when activated through MEK1 and MEK2.
MEK1 signaling through ERK2-NES provides survival, but not proliferation
Survival from TGFβ-induced apoptosis was examined in hepatocytes transfected with MEK1-CA or MEK2-CA alone or in combination with ERK2-NES-GFP (Fig. 6
A). A significant increase in survival was found when MEK1-CA was coexpressed with ERK2-NES-GFP compared to when MEK1-CA was expressed alone. However, in cells expressing MEK2-CA, a similar ability to mediate survival was observed with or without ERK2-NES-GFP. Thus, MEK1 gained the ability to enhance survival when ERK2 was routed to the cytoplasm. Comparable protein expression of ERK2-NES-GFP when coexpressed with MEK1-CA or MEK2-CA is shown in 6B.
|
Next, we studied whether relocating ERK2 to the cytoplasm correlated with reduced ability of MEK1 to mediate proliferation. DNA synthesis was measured in hepatocytes transfected with MEK1wt combined with ERK2wt, ERK2-NES, or the ERK2-AAA mutant. As shown in Fig. 7
A, reduced EGF-induced proliferation was found in cells expressing ERK2-NES and ERK2-AAA compared to cells transfected with ERK2-wt. Thus, the role of MEK1 in mediating a proliferative response correlated with ERK2 nuclear translocation. In Fig. 7B
, comparable protein expression of the various ERK2-GFP constructs is shown.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
. Also, MEK contains regulatory sites where other signals can be integrated. Two regions are quite divergent in MEK1 and MEK2: the N-terminal ERK-binding site and the PR region that allows interaction with signal proteins. MEK1, but not MEK2, is phosphorylated by PAK1 at S298 in the PR region, a modulation associated with enhanced signaling through Raf-MEK-ERK (9
10
11)
. In addition, this region is involved in the MEK-Raf-1 interaction (17
, 20)
. The low homologies in these regulatory sequences suggest that MEK1 and MEK2 may not have fully overlapping functions. In support of this, knockout studies indicate functional differences between the two isoforms. MEK1-deficient mice die due to placental vascularization defects, whereas MEK2 seems to be dispensable for growth and development (21
, 22)
. Further, knockdown of MEK1 or MEK2 with RNAi suggests that MEK1 signaling enhances cyclin D1 levels, thereby promoting cellular growth, whereas activation through MEK2 increases expression of the cell cycle inhibitory protein p21/CIP1 (23)
. In hepatocytes, unique activation patterns of MEK1 and MEK2 occur when EGF is added at different times after plating. Immediately after plating, MEK2 is activated, whereas MEK1 is activated after 24 h in culture (24)
. Thus, it is likely that MEK1 and MEK2 support specific functions in hepatocytes. In our study, we found that MEK2, but not MEK1, provided survival and cytoplasmic retention of ERK2. However, when a NES site was attached to ERK2, the MEK1 function switched from proliferation to survival. This indicated that survival signals from ERK2 are based on cytoplasmic localization, which is in line with our previous study, in which ERK mediated survival through phosphorylation of RSK in the cytoplasm (2)
. Correspondingly, MEK2 mediated activation of RSK1 more than MEK1. Activation and nuclear accumulation of ERK are prerequisites for cell cycle progression (3
, 4
, 25)
. When ERK2 was activated by MEK2, low levels of nuclear translocation were found. This correlated with a reduced ability to mediate proliferative signals when compared to cells expressing MEK1, in which strong nuclear accumulation of ERK2 was observed. Thus, our data indicated that MEK2 mediated survival, whereas MEK1 was involved in proliferation. Further, MEK1-mediated ERK2 nuclear translocation, and proliferation was dependent on and enhanced by modulation of the S298 and T292 sites in the PR region.
ERK2 nuclear translocation is regulated by the MEK1 S298 and T292 sites
When ERK2 is activated by MEK1, the MEK1-ERK2 complex dissociates in the cytoplasm, and ERK2 enters the nucleus independently from MEK1 (1)
. Once dephosphorylated in the nucleus, ERK2 is rapidly exported to the cytoplasm for reassociation with MEK1 (26
, 27)
. MEK1 shuttles between the cytoplasm and the nucleus by passive diffusion and NES-dependent export (27)
, and it has been proposed that inactivated ERK2 is exported to the cytoplasm in complex with MEK1. However, FRET technology combining ERK2-YFP and MEK1-CFP suggests that ERK2-MEK1 complexes are only found in the cytoplasm, prior to agonist exposure, and after stimulation, when the signal has been deactivated (28)
. Further, it was suggested that ERK2 nuclear import and export occurred through passive diffusion. Thus, ERK2 intracellular localization may rely on facilitated diffusion of free phosphorylated or dephosphorylated ERK2 across the nuclear membrane. Consequently, when activated, phosphorylated ERK2 dissociates from MEK1, and the level of free active ERK2 in the cytoplasm increases. This allows it to enter the nucleus along its concentration gradient. Movement of dephosphorylated ERK2 back to the cytoplasm may occur by facilitated diffusion enhanced by sequestration of inactive ERK2 in the cytoplasm, due to its reacquisition of affinity for MEK1. To obtain persistent nuclear localization, ERK2 may bind to newly synthesized nuclear anchoring proteins, which are regulated by proteasomal proteolysis (29)
. These proteins may possess selective affinity toward phosphorylated, and not dephosphorylated ERK2.
Association between MEK1 and ERK2 is regulated at two sites in MEK1: an intrinsic N-terminal binding domain, and one PR region containing the S298 and T292 sites (13
, 17)
. Phosphorylation of MEK1 on its S298-site by PAK enhances complex binding between MEK1 and ERK2, whereas ERK2 phosphorylation of MEK1 at T292 releases ERK2 from MEK1 (14
, 15)
. When hepatocytes were transfected with a MEK1-S298A mutant, we found the ERK2 nuclear accumulation in response to EGF to be impaired. A requirement for integrin signaling for ERK nuclear translocation has been shown (6
, 7)
. Further, integrin-mediated adhesion activates Rac and PAK, which phosphorylates MEK1 on S298 (9
10
11
, 30)
. This suggests that MEK1-S298 phosphorylation represents an adhesion-dependent step for ERK2 nuclear accumulation. Also, the T292-site was necessary for ERK2 nuclear accumulation, which indicated that both the S298 and the T292 sites are involved in adhesion-dependent ERK2 nuclear transport. Since phosphorylation of T292 by ERK2 is known to release the MEK1-ERK2 complex (15)
, our results are in line with the assumption that ERK2 enters the nucleus as a free protein, and not in complex with MEK1 (28)
. Although the S298 and T292 MEK1 mutants abrogated the ERK2 nuclear accumulation, neither affected the initial ERK2 T-E-Y phosphorylation. This suggested that immediately following stimulation with EGF, the MEK1 S298 and T292 sites are involved in ERK2 nuclear import/export mechanisms only, and do not interfere with the ERK2 phosphorylation. Also, when activated through MEK2, which lacks T292 and is a poor substrate of PAK, ERK2 accumulated to a low extent in the nucleus, although similar levels of phosphorylation were found. This indicated that when activated by MEK2, ERK2 remained in complex with MEK2, localized in the cytoplasm due to the NES-site on MEK2.
Our studies, combined with others, allow us to propose the following mechanism of ERK2 nuclear accumulation in response to growth factors in adhesion-dependent cells. Cellular adhesion initiates phosphorylation of the S298-site on MEK1, enhancing the complex association between MEK1 and ERK2. In response to EGF, MEK1 becomes phosphorylated on its S218/S222 Raf site. The combined actions of PAK and Raf on MEK1 allow ERK2 feedback phosphorylation at the MEK1 T292 site, which leads to dissociation between MEK1 and active ERK2. At the same time, T292 phosphorylation acts as a negative feedback on S298 phosphorylation due to conformational changes, so it cuts PAK off from further MEK1 phosphorylation (15)
. Thus, ERK2 nuclear translocation is allowed by the concerted action of growth factors and adhesion signals, and the signals are integrated at MEK1.
The sorting of ERK2 to mediate proliferation may be determined by the PR region of MEK1 at more than one level. First, as discussed above, by allowing nuclear accumulation of ERK2 through the subsequent association and release with MEK1, and second, by providing a preferred ability of MEK1 compared to MEK2 of sequestering ERK2 in adherent cells prior to growth factor addition. Adhesion of cells to ECM increases phosphorylation of MEK1 at S298 and a tighter binding between MEK1 and ERK2 (14)
. Contrary, in the absence of adhesion, no PAK1-induced phosphorylation of S298 occurs, which results in a less tight association between MEK1 and ERK2 (14)
. At the same time, the MEK2-ERK2 complex remains constant in adherent and suspended cells (14)
. In this manner, MEK2 will be the preferred partner of ERK2 in nonadherent cells compared to MEK1. Consequently, growth factor stimulation of nonadherent cells will activate ERK2 preferably through MEK2, which enhances survival, while cells attached to ECM allow growth factor-induced activation of ERK2 through MEK1, proving proliferation.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Received for publication March 28, 2007. Accepted for publication August 16, 2007.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
