The N-terminal domain of ERK1 accounts for the functional differences with ERK2

Abstract

The Extracellular Regulated Kinase 1 and 2 transduce a variety of extracellular stimuli regulating processes as diverse as proliferation, differentiation and synaptic plasticity. Once activated in the cytoplasm, ERK1 and ERK2 translocate into the nucleus and interact with nuclear substrates to induce specific programs of gene expression. ERK1/2 share 85% of aminoacid identity and all known functional domains and thence they have been considered functionally equivalent until recent studies found that the ablation of either ERK1 or ERK2 causes dramatically different phenotypes. To search a molecular justification of this dichotomy we investigated whether the different functions of ERK1 and 2 might depend on the properties of their cytoplasmic-nuclear trafficking. Since in the nucleus ERK1/2 is predominantly inactivated, the maintenance of a constant level of nuclear activity requires continuous shuttling of activated protein from the cytoplasm. For this reason, different nuclear-cytoplasmic trafficking of ERK1 and 2 would cause a differential signalling capability. We have characterised the trafficking of fluorescently tagged ERK1 and ERK2 by means of time-lapse imaging in living cells. Surprisingly, we found that ERK1 shuttles between the nucleus and cytoplasm at a much slower rate than ERK2. This difference is caused by a domain of ERK1 located at its N-terminus since the progressive deletion of these residues converted the shuttling features of ERK1 into those of ERK2. Conversely, the fusion of this ERK1 sequence at the N-terminus of ERK2 slowed down its shuttling to a similar value found for ERK1. Finally, computational, biochemical and cellular studies indicated that the reduced nuclear shuttling of ERK1 causes a strong reduction of its nuclear phosphorylation compared to ERK2, leading to a reduced capability of ERK1 to carry proliferative signals to the nucleus. This mechanism significantly contributes to the differential ability of ERK1 and 2 to generate an overall signalling output.

Figure 1. ERK1-GFP translocates in the nucleus of NIH-3T3 cells after stimulation.

A) Cells transfected with ERK1-GFP in control conditions and after treatment with 80 ng/ml of FGF4 (calibration bar 20 µm). B) Time course of the normalized Concentration Index of cells stimulated with FGF4 (n = 18). The vertical bars are the standard error of the mean at the given point and are representative of the experimental variability of the entire data set. C) ERK-GFP fusion proteins are phosphorylated following serum stimulation, as demonstrated by western blot with a phospho-specific ERK antibody (upper panel). The fusion proteins have the correct molecular weight also when assayed with an anti-GFP antibody.

Figure 2. Comparison of the nucleo-cytoplasmic shuttling of ERK1 and ERK2.

A) After photobleaching of the nucleus of starved NIH 3T3 cells, the ERK1 fluorescence recovered more slowly than ERK2, indicating a slower turnover of ERK1 across the nuclear membrane. Calibration bar 10 µm. B) Time course of the recovery of the cells showed in A. The data point were fitted with a single exponential with time constant τ of 235 sec (ERK2) and 430 sec (ERK1). C) Summary data for the sampled cells. The turnover of ERK1 is slower both in starved cells and after stimulation with FGF. In ERK1 starved cells the turnover is 3.7 time slower (653 s) than ERK2 (178 s) and in the stimulated cells ERK1 turnover (266 s) is 3.1 time slower than ERK2 (84 s). The blue symbol shows the turnover for a GFP dimer. This molecule is smaller than ERK-GFP but it crosses the nuclear membrane more slowly than ERK1, indicating that also for ERK1 is operating a mechanism of facilitated diffusion. D) Scatter diagram showing the recovery and Concentration Index of all paired measurements. Lines join observations relative to the same cell before and after stimulation. The filled symbols are the averages of the two groups.

Figure 3. Alignment of the amino acid sequences of rat ERK1 and ERK2.

The N-terminus is shown with a larger font. The 20 aa present only in ERK1 are displayed in bold.

Figure 4. The N-terminal domain is responsible for the slow nucleo-cytoplasmic shuttling of ERK1.

A) At the N-terminus, ERK1 wild type (wt) contains 20 residues not present in ERK2. We produced fusion with GFP of three different deletions of ERK1, as indicated in the diagram (mouse sequence is exemplified). B) The time constant of the nucleo-cytoplasmic shuttling of ERK1 fusion proteins is strongly affected by the different deletions of the N-terminus. In this and the next figure, the number of cells is indicated over each column.

Figure 5. Fusion of ERK2-GFP with the N-terminus of ERK1 (Δ39 E2/E>E1).

A) Alignment of the N-terminals of ERK1, ERK2 and Δ39 E2. The domain of ERK1 that has been fused to ERK2 is indicated in bold. B) The additional domain added at the N-terminus of ERK2 considerably decreases the speed of nucleo-cytoplasmic turnover both in the starved state and after stimulation with FGF4.

Figure 6. Computational estimate of the functional consequences of the different shuttling rates of ERK1 and 2.

A) Cells transfected with ERK1-GFP and ERK2-GFP were treated with FGF4 for 15 min to allow complete nuclear translocation and then with the ERK blocker U0126. The inactivation of the ERK pathway caused the immediate loss of nuclear accumulation of ERK-GFP unmasking the action of nuclear dephosphorylation. Calibration bar 20 µm. B) By fitting the time course of the decay of nuclear ERK2-GFP (empty red diamonds) with the model output (continuous red line), we could estimate the dephosphorylation rate. This estimate was included in the model together with the shuttling rates measured for ERK1 to predict the outcome of this experiment practiced on the cells transfected with ERK1-GFP. The model prediction (green line) describes with great accuracy the experimental points (filled green symbols). C) Reaction scheme of the model. It has been considered an equilibrium among 4 different states regulated by first order kinetics: pERK and ERK indicates the concentration of the phosphorylated and not-phosphorylated pools in the cytoplasm (Cyt) or in the nucleus (Nuc); the rate constants α and β are the time constant of recovery of FRAP experiments; γ is the phosphorylation rate; γ' and δ' are the dephosphorylation rates, respectively, in the cytoplasm and in the nucleus. D) The model was used to compute the phosphorylation in the nucleus as a function of the shuttling speed. The empty symbols represent the result of a single simulation run and the filled symbols are the averages of each group. The phosphorylation level has been normalized to ERK2, therefore the computation shows that the total level of phosphorylation of ERK1 is only about half of ERK2. The numbers under each set of data points are the time constants (in seconds) of nuclear shuttling in the starved (above) and stimulated conditions (below).

Figure 7. Comparison between the activation of ERK1 and 2.

A) NIH 3T3 cells have been starved for 24 hr before treatment for 15 min with increasing concentrations of serum as indicated. The densitometric analysis of the gel blotted with the phospho-specific antibody has been performed with a linear imager to quantify the intensity of the phospho-ERK1 and 2 bands. From each experiment we computed the ratio pERK2/pERK1 which is a measure of the relative activation of the two kinases. The graphs shows that ERK1 activation lags behind ERK2. B) Effects of phosphatase inhibition on the relative activations of ERK1 and 2. Cells have been starved for 24 hr before a 30 min treatment with serum 10% and/or a cocktail of phosphatase inhibitors. Inhibition of phosphatases in presence of serum caused a further increase of phosphorylation compared to serum only. This increase was larger for ERK1, indicating a stronger dependence of ERK1 on de-phosphorylation. Notice that there is no vertical correspondence between the western blot and the quantification.

Figure 8. Functional consequences of N-terminus mutations of ERK1 and ERK2.

A) In vitro assay testing the capacity of the indicated ERK to be activated by MEK and to phosphorylate the downstream target MBP. The upper lanes show the purified protein revealed by the anti-GFP antibody. The fusion proteins showed correct molecular weights (ERK1, E2>E1 71 kD; ERK2, E1>E2 69 kD). The lower panel indicates that each protein is able to phosphorylate MBP. The experiment was performed in triplicate with similar results. B) Representative examples of the cell colonies transfected with the indicated constructs. C) Immunoblot anti GFP shows that all the colonies expressed the GFP tagged proteins; the anti-HA and anti-Myc staining show that the mutated forms of Ras (respectively: H-Ras G12V and H-Ras Q61L), were expressed in the colonies with constitutive activation of the pathway, but they were absent in the wild type background. D) Quantification of the effect on proliferation of the various expressed fusion proteins. The symbols indicates the number of colonies counted after transfection with the specified vector, normalized to the colonies counted after transfection with the empty vector. The empty symbols represent the result of a single experiment and the filled symbols are the averages of each group. Expression of constitutively active Ras (H-Ras G12V and H-Ras Q61L) causes a large increase in proliferation that is inhibited by co-expressing ERK1 but not by co-expressing ERK2. The mutant of ERK1 characterized by fast shuttling (E1Δ39 indicated as E1>E2) did not prevent H-Ras G12V and Q61L-induced proliferation behaving similarly to ERK2. In contrast the slow mutant of ERK2 (Δ39E2 indicated as E2>E1) inhibited proliferation. Statistical significativity was essayed by t-test.

Figure 9. Effects of the overexpression of N-terminus mutations of ERK1/2 on the nuclear target MSK.

A) Representative fields of fibroblast treated for 24 hr with 10% serum. In this example, cells have been transfected with ERK2-GFP and were stained with the pMSK specific antibody (red). The left panel show that the treatment caused nuclear translocation of ERK2-GFP and activation of MSK, which is well visible both in the transfected cell (the nucleus is yellow) and in nearby non-transfected cells. The inhibitor U0126 (25 µM) was administrated 6 hr before fixation. In these conditions ERK did not accumulate in the nucleus and there was not detectable phosphorylation of MSK, indicating that, MSK requires ERK activity to be phosphorylated after serum treatment. Bar 20 µm. B) Intensity of the pMSK signal measured in cells transfected with ERK1, ERK2, E1Δ39 (indicated as E1>E2) and Δ39E2 (indicated as E2>E1). The experiment was repeated in triplicate and the fluorescence of each cell has been normalised to the mean fluorescence of the ERK1 group to allow the comparison of the different experiments. Number of cells indicated nearby each symbol. U0126 suppressed almost completely the pMSK signal (mean = 0.48±0.04, n = 76; not shown). Since cell fluorescence was not normally distributed, significativity was assessed with the Kolmogorov-Smirnoff test. ERK1≠ERK2 (p≤0.0005); E1>E2≠E2>E1 (p≤0.0005). E1 and E2>E1 are not significantly different (p≤0.2). E1>E2 is larger that ERK1 (p≤0.0005) but it is also slightly but significantly smaller than ERK2 (p≤0.01).