Gonadotropin-Releasing Hormone and Protein Kinase C Signaling to ERK: Spatiotemporal Regulation of ERK by Docking Domains and Dual-Specificity Phosphatases

Abstract

Activated ERK translocates to the nucleus to regulate transcription. Spatiotemporal aspects of this response dictate biological consequences and are influenced by dual-specificity phosphatases (DUSPs) that can scaffold and dephosphorylate ERK. In HeLa cells, GnRH causes transient and protein kinase C (PKC)-dependent ERK activation, but termination mechanisms are unknown. We now explore DUSP roles using short inhibitory RNA to knock down endogenous ERK, adenoviruses to express GnRH receptors and add-back ERK2-GFP, and automated microscopy to monitor ERK location and activation. GnRH caused rapid and transient increases in dual phosphorylated ERK2 (ppERK2) and nuclear to cytoplasmic ERK2-green fluorescent protein (GFP) ratio, whereas responses to a PKC-activating phorbol ester were more sustained. In cells expressing D319N ERK2-GFP (D319N mutation impairs docking-domain-dependent binding to DUSPs), GnRH caused more sustained increases in ppERK2 and nuclear to cytoplasmic ERK2-GFP ratio and also had more pronounced effects on Egr-1 luciferase (a transcriptional reporter for ERK activation). Cycloheximide caused more sustained effects of GnRH and phorbol ester on ppERK, suggesting termination by nuclear-inducible DUSPs. GnRH also increased expression of nuclear-inducible DUSP1 and -4, but their knockdown did not alter GnRH-mediated ERK signaling. Screening a short inhibitory RNA library targeting 16 DUSPs (nuclear-inducible DUSPs, cytoplasmic ERK MAPK phosphatases, c-Jun N-terminal kinase/p38 MAPK phosphatases, and atypical DUSPs) revealed GnRH effects to be influenced by DUSPs 5, 9, 10, 16, and 3 (i.e. by each DUSP class). Thus, GnRH-mediated ERK responses (like PKC-mediated ERK responses) are dependent on protein neosynthesis and docking-domain-dependent binding, but for GnRH activation (unlike PKC activation), this does not reflect dependence on nuclear-inducible DUSPs. Termination of these GnRH effects is apparently dependent upon a preexisting rapid turnover protein.

GnRH receptors mediate transient ERK activation in HeLa cells. This response is shaped by protein neosynthesis, D-domain-dependent binding and by phosphatases of each DUSP class.

GnRH acts via G_q-coupled seven-transmembrane (7TM) receptors to stimulate the synthesis and secretion of LH and FSH and thereby mediates central control of reproduction (1, 2, 3). Like many other 7TM receptors, GnRH receptors (GnRHRs) activate the prototypic MAPK, ERK (4). In quiescent cells, ERKs are typically anchored in the cytosol. Upon dual phosphorylation and activation by MAPK/ERK kinase (MEK), ERKs can translocate to the nucleus where they can in turn phosphorylate transcription factors and immediate-early gene products (5, 6, 7, 8). This leads to altered gene expression and, in the case of the GnRHR, ERKs play an important role in the control of gonadotropin secretion (transcription of the common α-subunit as well as the distinct β-subunits of LH and FSH) (4). Typically, G_q-dependent PKC activation plays a major role in GnRH-stimulated ERK1/2 activation. However, mechanisms can vary according to cellular context because, for example, GnRHRs mediate a PKC-dependent transactivation of epidermal growth factor (EGF) receptors with consequent activation of ERK in GT1-7 neurons but cause a largely PKC-dependent and EGF receptor-independent activation of ERK in LβT2 and HeLa cells (9, 10, 11, 12). Sustained activation of 7TM receptors typically causes their desensitization and/or internalization, processes mediated by phosphorylation (most often within the receptor’s C-terminal tail) and consequent binding to arrestins (6, 13). For many 7TM receptors, binding of arrestins to Raf and ERK mediates ERK activation so that arrestins can shape ERK activation by both arrestin-mediated receptor desensitization and arrestin-mediated ERK activation (4, 6, 13). However, type I mammalian GnRHRs are unique in that they lack carboxyl-terminal tails (2), and where investigated, they do not show agonist-induced receptor phosphorylation, do not bind arrestins, do not rapidly desensitize or internalize, and do not cause arrestin-mediated ERK1/2 activation (12, 14, 15, 16, 17). Accordingly, these receptors provide models with which to explore downstream determinants of ERK signaling in the absence of such upstream adaptations.

The control of ERK binding to substrates and scaffolds is governed by a series of docking (D)-domains, the best characterized of which are the D-domains that bind to ERK in a region opposite the catalytic site (6, 18, 19, 20). Association with D-domains can be abrogated by D319N mutation of ERK, which is analogous to the Drosophila sevenmaker gain-of-function mutation (19, 20, 21, 22, 23). The largest group of proteins that directly regulate ERK inactivation belong to the dual-specificity phosphatase (DUSP) family of enzymes (21, 24, 25, 26). These are collectively known as the MAPK phosphatases (MKPs) because of their ability to bind directly to MAPKs and remove both Thr and Tyr phosphate groups, returning them to their inactive state (21, 24, 26). The MKPs typically have motifs directing them to nuclear or cytoplasmic regions and can therefore inactivate MAPKs in multiple compartments (21, 25, 27). Most MKPs also contain D-domains, subtle differences around which can 1) determine whether DUSPs target ERK or the closely related c-Jun N-terminal kinase (JNK) or p38 MAPKs and 2) whether DUSPs remain associated with their substrate after dephosphorylation (28, 29, 30). When considered alongside their tight transcriptional regulation, it is clear that the MKPs act in a complex negative feedback network to control duration, magnitude, and localization of ERK signals and also its cross talk with the JNK and p38 MAPK pathways (31, 32, 33).

GnRH is known not only to activate ERK but also to increase expression of the nuclear-inducible MKPs, DUSP1, and DUSP4 in gonadotrope-lineage αT3-1 cells (31, 32, 33). In this model, DUSP induction exerts negative feedback effects on JNK but has less effect on the ERK response (31, 32, 33). Thus, the mechanisms terminating GnRH effects on ERK are unknown. Indeed, the key published information is all negative; the response is not mediated by GnRHR desensitization, is not associated with a switch from G protein-mediated to arrestin-mediated signaling, and is not mediated by DUSP1 or -4. We have recently explored ERK activation in HeLa cells transduced with adenovirus (Ad)-expressing GnRHRs and found that GnRH mediates a protein kinase C (PKC)-dependent and EGF receptor-independent activation of ERK (12). Here, we have used short inhibitory RNA (siRNA) to knock down endogenous ERKs in GnRHR-expressing HeLa cells, and recombinant Ad to add back either wild-type (WT) ERK2-GFP or D319N mutated ERK2-GFP to interfere with D-domain binding. Using immunofluorescent staining and a semiautomated system for image acquisition and analysis, we now use this model to explore how DUSPs shape ERK responses to GnRH and a PKC-activating phorbol ester.

Results and Discussion

Kinetics of ERK activation by GnRH and phorbol 12,13-dibutyrate (PDBu)

In the first experiments, we determined the time course of effects of GnRH and PKC activation (with PDBu) on activation of endogenous ERKs by fluorescence immunohistochemical staining for dual phosphorylated ERK1/2 (ppERK1/2) followed by semiautomated image acquisition and analysis. As shown (Fig. 1), stimulation of Ad mouse (m)GnRHR-transduced HeLa cells with 1 μm GnRH caused a rapid and transient activation of ERK (whole-cell ppERK1/2 was maximal at 5 min and returned to basal by 60 min), whereas 1 μm PDBu caused a slower and more sustained activation (maximal at 5–15 min and above basal for at least 4 h). These data parallel similar data obtained by Western blotting (12, 30, 34) and demonstrate the utility of the imaging assay for monitoring activation of endogenous ERKs.

Fig. 1.

Kinetics of GnRH and PDBu effects on activation of endogenous ERKs. Left panels, HeLa cells were cultured in 96-well plates and transduced with Ad mGnRHR before stimulation with 1 μm PDBu or 1 μm GnRH for the times indicated. They were then fixed, stained, and imaged for calculation of whole-cell ppERK1/2 intensity as described in Materials and Methods. Right panel, LβT2 cells were treated exactly as described for HeLa cells except that they express endogenous mGnRHR and were therefore not transduced with Ad GnRHR. The data shown are pooled from four separate experiments (mean ± sem, n = 3) each with duplicate or triplicate wells and typically with more than 100 cells imaged per well. Effects of both stimuli were statistically significant at all time points and in both cell types (*, P < 0.05) with the exception of the values at 60, 120, and 240 min in GnRH-stimulated HeLa cells, which were not significantly different (P > 0.1) from the control value before stimulation (dotted bar).

Interestingly, the responses to these stimuli were more sustained in the gonadotrope-lineage LβT2 cell line, which has endogenous mGnRHR. In these cells, GnRH caused a more sustained response, which was maximal at 5–60 min with significant elevation throughout the 240-min stimulation period (Fig. 1) (35). Dose-response studies revealed comparable potency [negative log of 50% effective concentration (pEC₅₀) values of −8 to −9] in both cell types, and we have observed similar transient responses to GnRH in HeLa cells stimulated with 1 nm GnRH (not shown), so the difference in response kinetics do not reflect differences in receptor occupancy. The LβT2 cell line was derived by targeted expression of SV40 T antigens in murine gonadotropes (36, 37). T antigens cause cell immortalization by interacting with a wide range of target proteins. These notably include protein phosphatase 2A, which together with protein phosphatase 1, accounts for more than 90% of Ser/Thr phosphatase activity in most cells and tissues and is markedly inhibited by the small T antigen (38, 39, 40). Such inhibition is well known to affect ERK responses (41), and we suspected this could influence the kinetics of ERK dephosphorylation in LβT2 cells. The subsequent studies were therefore performed in HeLa cells only.

We have recently developed a model in which endogenous ERKs are knocked down with siRNAs and a GFP fusion protein reporter is added back with recombinant Ad-expressing WT ERK2-GFP, enabling simultaneous quantification of ERK activity (ppERK2) and location (ERK2-GFP). To validate this approach, we used the automated imaging system to test for effects of the knockdown (siRNA sequences targeted to noncoding regions of ERK1 and -2 transcripts) and addback (Ad ERK2-GFP) on ERK phosphorylation. As shown (Fig. 2, left panel), PDBu caused the expected increase in whole-cell ppERK1/2 levels in control cells, whereas no such stimulation was seen in cells receiving ERK1/2 siRNAs, and the response was recovered in cells receiving Ad ERK2-GFP (in the presence of the siRNAs). We also used Western blotting to confirm that GnRH causes a transient increase in ERK phosphorylation in this model, that the ERK1/2 siRNAs are effective at knocking down endogenous ERK expression, and that Ad-mediated addback recovers total effects of ERK expression (Fig. 2, right panel). These results confirm previous findings with Western blotting and imaging that the knockdown prevents phosphorylation responses and that the addback restores ERK expression levels and phosphorylation responses to levels seen in control cells (19, 30, 34).

Fig. 2.

Validation of a knockdown and addback model for studying ERK regulation. Left panel, HeLa cells were transfected with control siRNAs (ctrl) or ERK1/2 siRNAs and were transduced with Ad ERK2-GFP before incubation for 15 min in control medium (open bars) or in medium with 1 μm PDBu (filled bars) as indicated. They were then fixed, stained, and imaged for calculation of whole-cell ppERK1/2 intensity as above. The data shown are pooled from four separate experiments (mean ± sem, n = 4) each with duplicate or triplicate wells. **, P < 0.01, compared with the corresponding control value (without PDBu). Right panel, HeLa cells grown in 6-cm plates were transfected with control siRNAs (ctrl) or ERK1/2 siRNAs and were transduced with Ad mGnRHR with or without Ad ERK2-GFP as indicated, before incubation for 0, 15, or 120 min with 1 μm PDBu. They were then processed for Western blotting with antibodies targeting total ERK1/2, ppERK1/2, or β-actin (loading control) as described in Materials and Methods. The position of bands showing ERK1/2, ERK2-GFP, and the corresponding phosphoproteins were determined by comparison with molecular weight markers, and the data shown are representative of those obtained in three similar experiments.

We next assessed time courses of responses to GnRH and PDBu in HeLa cells using the knockdown/addback protocol. As shown (Fig. 3, upper left panel), GnRH caused a rapid and transient increase in whole-cell ppERK2 levels (maximal at 5 min, returned to basal within 60 min), whereas PDBu caused a slower and more sustained response (maximal at 5–15 min, remaining above basal for at least 4 h). These results are similar to those obtained with endogenous ERK1/2 (Fig. 1) and provide a further demonstration that the knockdown/addback protocol accurately recapitulates behavior of the endogenous ERKs. When the distribution of ERK2-GFP was examined, both stimuli caused pronounced initial increases in nuclear to cytoplasmic (N:C) ERK2-GFP ratio, but the responses were very different after the first 5 min. PDBu induced a sustained but biphasic increase in N:C ERK2-GFP (peaks at 5–15 and 120 min), whereas GnRH caused a transient increase with a peak at 5 min, reducing rapidly to less than 50% of the maximal response within 15 min (Fig. 3, lower left panel). Thus, although GnRH and PDBu cause comparable maximal activation of ERK2, the PDBu-induced ppERK2 response is relatively sustained and is associated with very high and sustained nuclear localization of ERK2-GFP. This distinction is also evident in the representative images shown in Fig. 3. These stimulus-specific response profiles are indicative of stimulus-specific mechanisms controlling termination and compartmentalization of ERK signaling.

Fig. 3.

Spatiotemporal characteristics of GnRH- and PDBu-stimulated ERK regulation using the knockdown and addback model. Cells were transfected in 96-well plates with ERK1/2 siRNAs and transduced with Ad ERK2-GFP and Ad mGnRHR before stimulation with 1 μm GnRH or 1 μm PDBu for the times indicated. They were then fixed and stained before image acquisition and analysis (as above) for the calculation of whole-cell ppERK2 intensity (upper left panel) and the N:C ERK2-GFP ratio (lower left panel). The data shown are pooled from seven separate experiments (mean ± sem, n = 5–7), each with duplicate or triplicate wells. The effects of GnRH were statistically significant (P < 0.05) at 5, 15, and 30 min, and the effects of PDBu were statistically significant at all time points (for both endpoints). Representative regions of cell images are also shown for DAPI, ERK2-GFP, and ppERK2 in cells stimulated with 1 μm GnRH or PDBu as indicated (right panels). Note that despite comparable initial responses, appreciable levels of ppERK2 and nuclear retention of ERK2-GFP are only seen at 120 min in the PDBu-stimulated cells. Scale bars, 30 μm.

Effects of D-domain mutation and protein synthesis inhibition

ERKs bind to many proteins containing modular D-domains, and because these proteins include scaffolds and anchors, perturbation of such binding can influence ERK compartmentalization and signaling (19, 30, 34). To explore this, we introduced a D319N mutation that impairs D-domain binding (19, 20, 22, 23). This mutation does not affect ERK activation or catalytic activity and is therefore useful for identifying classes of proteins involved in shaping stimulus-specific ERK responses (19, 22, 30, 34). Cells transfected with ERK1/2 siRNAs were transduced with Ad-expressing WT or D319N ERK2-GFP. Previous validation of this protocol has revealed that the D319N ERK2-GFP is expressed at comparable levels to WT ERK2-GFP and to endogenous ERK1/2 in control cells, that the mutant does not show stimulus-independent activity, and that it is activated (by EGF and PDBu) with similar potency and efficacy to that seen with the WT ERK2-GFP (30, 34) (Fig. 4). Cell imaging revealed that GnRH caused the expected transient increase in ppERK2 in cells expressing WT ERK2-GFP, but this response was more sustained in cells expressing D319N ERK2-GFP. As a measure of response duration, we estimated the half-time for inactivation by fitting the data (from 5–120 min) to a one-phase exponential decay curve. This revealed slower inactivation in cells expressing D319N ERK2-GFP than in cells with WT ERK2-GFP (half-times 24 and 9 min, respectively). GnRH also caused the expected transient increase in N:C ERK2-GFP ratio, and this effect was also prolonged by the mutation (half-times were 8 and 17 min for cells expressing WT and D319N ERK2-GFP, respectively).

Fig. 4.

Effects of docking domains and protein synthesis inhibition on ERK activation by GnRH. Cells were transfected in 96-well plates with ERK1/2 siRNAs and transduced with Ad ERK2-GFP and Ad mGnRHR before stimulation with 1 μm GnRH as described under Fig. 3, except that the ERK2 addback was with Ad-expressing WT ERK2-GFP or D319N ERK2-GFP as indicated (left panels), or cells transduced with Ad WT ERK2-GFP were stimulated with GnRH in the presence or absence of 30 μm CHX as indicated (right panels). They were then fixed and stained before image acquisition and analysis (as above) for the calculation of whole-cell ppERK2 intensity (upper panels) and the N:C ERK2-GFP ratio (lower panels). The data shown are pooled from seven separate experiments (mean ± sem, n = 3–7), each with duplicate or triplicate wells. Two-way ANOVA revealed stimulation to be a significant variable (P < 0.001 for all four panels). The D319N mutation was also a statistically significant variable (P < 0.05) for both endpoints (left panels), and CHX was a significant variable for the ppERK2 measurements (P < 0.05, upper right panel) but not for the ERK2-GFP measures (P > 0.1, lower right panel).

We have previously shown that D319N mutants also prolong responses to PDBu (30, 34), so our data are consistent with roles for D-domain-containing proteins in termination of ERK responses to both stimuli. Because both GnRH and PKC mediate synthesis of nuclear MKPs containing D-domains (30, 31, 32, 33, 34), we compared the effects of docking domain mutations and the protein synthesis inhibitor cycloheximide (CHX). As shown (Fig. 4, right panels), CHX had no effect on basal ppERK2 levels in cells expressing WT ERK2-GFP but caused a pronounced prolongation of the ppERK2 response to GnRH (half-time increased from 9–57 min). It also reduced the maximal effect of GnRH on N:C ERK2-GFP ratio and increased the inactivation half-time (from 8–14 min). We have previously described similar experiments using EGF and PDBu, and comparison with the current GnRH data (60-min time point) reveals a remarkable degree of stimulus-specific regulation (30, 34). As shown (Fig. 5), CHX and the D319N mutation had no measurable effect on ppERK2 levels in unstimulated cells. The three stimuli used caused comparable maximal ppERK2 responses (Fig. 4) (30, 34), but sustained responses differed. PDBu caused a characteristic sustained increase in ppERK2 levels, and this effect was increased by CHX and by the D319N mutation. EGF characteristically causes a less sustained activation, and this response was enhanced by the D319N mutation but not by CHX (Fig. 5). GnRH caused no measurable increase in ppERK2 at 60 min, but clear stimulation was seen in cells expressing D319N ERK2-GFP or in cells treated with CHX (Fig. 5). Because ERK activation characteristically stimulates Egr-1 luciferase activity, we performed similar studies with this endpoint to address the functional relevance of the observed changes in ERK activation. As shown (Fig. 6), GnRH, PDBu, and EGF each caused dose-dependent activation of Egr-1 luciferase, and in each case, the effect was markedly inhibited by knockdown of ERK1/2 and recovered by addback of Ad ERK2-GFP, paralleling the data from ERK imaging with the knockdown/addback protocol. These data provide further functional evidence for the validity of the knockdown/addback protocol and also demonstrate that the GnRH effect on Egr-1 luciferase is primarily ERK mediated (residual responses in cells transfected with ERK1/2 siRNAs are likely due to incomplete knockdown). The D319N ERK2-GFP mutation increased basal Egr-1 luciferase activity and further increased responses to each of the stimuli, demonstrating that the increase in ERK activation seen with this mutation (Figs. 4 and 5) leads to amplification of this downstream readout.

Fig. 5.

Comparison of docking domains and protein synthesis inhibitor effects on ERK activation by GnRH, PDBu, and EGF. Cells were transfected in 96-well plates with ERK1/2 siRNAs and transduced with Ad-expressing WT or mutated ERK2-GFP reporters and Ad mGnRHR before stimulation for 60 min with 1 μm GnRH, 1 μm PDBu, or 10 nm EGF in the presence or absence of 30 μm CHX as described under Fig. 3. They were then fixed and stained before image acquisition and analysis for the calculation of whole-cell ppERK2 intensity. For the purpose of this comparison, GnRH data from Fig. 4 have been plotted alongside previously published EGF and PDBu data (30 ). The data shown are pooled from seven separate internally controlled experiments (mean ± sem, n = 3–7), each with duplicate or triplicate wells. *, P < 0.05; **, P < 0.01 compared with the corresponding WT for each stimulus.

Fig. 6.

Effects of docking-domain mutants on GnRH-stimulated Egr-1 luciferase activity. Cells were transfected in 96-well plates and transduced with Ad mGnRHR and Ad Egr-1 luciferase. ERK knockdown (KD) was achieved by transfection with ERK1/2 siRNAs, and addback was with Ad WT ERK2-GFP or D319N ERK2-GFP as indicated (ctrl, control cells that received neither knockdown nor addback). After culture, these cells were stimulated for 4 h with the indicated concentrations of GnRHR, PDBu, or EGF. They were then processed for measurement of luciferase (luc) activity as described in Materials and Methods. The data shown are pooled from four separate experiments (mean ± sem, n = 2–4) each with duplicate or triplicate wells, and the figure shows Egr-1 luciferase activity in relative light units (RLU).

The relevance of nuclear-inducible MKPs

The data above reveal that protein neosynthesis and D-domain-dependent binding are involved in termination of ERK responses to GnRH (and PDBu) in the nucleus and that such termination is functionally relevant in terms of downstream signaling. They are therefore suggestive of a role for the nuclear-inducible MKPs, including DUSP1, -2, and -4, which are rapidly synthesized in response to external stimuli, localize to the nucleus, and bind to ERK through D-domain-dependent interactions (30, 42). Consistent with this, and with previous data in αT3-1 cells (31, 32, 33), we found that GnRH and PDBu both increased levels of DUSP1 mRNA and DUSP4 mRNA. Interestingly, these effects on DUSP1 mRNA were prevented by knockdown of endogenous ERKs, whereas the DUSP4 mRNA response was not (Fig. 7). Moreover, PDBu caused a pronounced increase in DUSP2 mRNA that was also blocked by ERK1/2 knockdown but was not mimicked by GnRH (Fig. 7). To test for potential roles of these phosphatases in shaping ERK responses to GnRH, we initially used siRNAs to knock down DUSPs 1, 2, and 4 individually but observed no measurable effect on GnRH-stimulated ppERK2 responses (not shown). However, we have previously found that knockdown of one DUSP can cause compensatory increases in effects of PDBu on other DUSPs (30) and therefore tested to see whether this also occurs in GnRH-stimulated cells. We found that the effect of GnRH on DUSP1 mRNA was greatly enhanced by knockdown of DUSP2 mRNA (Fig. 8, left panel) but not by knockdown of DUSP4 mRNA (not shown) and that the effect of GnRH on DUSP4 mRNA was greatly enhanced by knockdown of DUSP1 (Fig. 8, right panel) but not by knockdown of DUSP2 (not shown). We suspected that such compensatory changes may have undermined our attempt to define roles for these DUSPs by individual knockdown (above) and therefore used a triple-knockdown protocol. As shown (Fig. 9), the effects of GnRH on ppERK2 levels, N:C ERK2-GFP ratio and Egr-1 luciferase activity were unaltered by the combined knockdown of DUSPs 1, 2, and 4. We have previously found that effects of EGF are also insensitive to the triple knockdown of DUSPs 1, 2, and 4 (30) (Fig. 9, C and D), and such insensitivity may be characteristic of stimuli causing only transient activation, although the triple knockdown did cause a modest increase in EGF-stimulated Egr-1 luciferase activity (Fig. 9E). Importantly, the GnRH data differ markedly from those obtained with PDBu, in which the sustained ppERK2 response and Egr-1 luciferase activity are enhanced by the triple knockdown and the effect on N:C ERK2-GFP is reduced by the triple knockdown (30) (Fig. 9, C–E). Indeed, the effect of CHX on responses to PDBu was mimicked by the combined knockdown of DUSPs 1, 2, and 4 (Figs. 4 and 9), suggesting that PDBu causes neosynthesis of these phosphatases, which act in concert to dephosphorylate ERKs and scaffold them in the nucleus. In sharp contrast to this, the effect of CHX on GnRH responses is not mimicked by knockdown of these phosphatases, suggesting protein synthesis-dependent but DUSP1-, 2-, and 4-independent termination mechanisms.

Fig. 7.

Effects of GnRH, PDBu, and ERK knockdown on nuclear-inducible DUSP mRNA. Cells were transfected in six-well plates with control siRNAs (ctrl) or ERK1/2 siRNAs and transduced with Ad mGnRHR before incubation for 120 min in control medium or with 1 μm PDBu or 1 μm GnRH, as indicated. Total RNA isolates were analyzed for relative levels of DUSP1, -2, or -4 mRNA by qPCR as described in Materials and Methods. The data shown are normalized values obtained from three separate experiments, each with duplicate readings (mean ± sem, n = 3). *, P < 0.05; **, P < 0.01, comparing control siRNA-transfected cells to ERK1/2 siRNA-transfected cells, using Student’s t test.

Fig. 8.

Effects of DUSP knockdown on DUSP induction by GnRH. Cells were transfected in six-well plates with control siRNAs (ctrl), DUSP2 siRNAs, or DUSP1 siRNA, transduced with Ad mGnRHR, and incubated for 120 min in control medium (−) or with 1 μm GnRH (+), as indicated. Total RNA isolates were analyzed for relative levels of DUSP1 or -4 mRNA by qPCR as described in Materials and Methods. The data shown are normalized values obtained from three separate experiments, each with duplicate readings (mean ± sem, n = 3). *, P < 0.05; **, P < 0.01, comparing control and GnRH-treated cells, using Student’s t test. In parallel experiments, effects of GnRH on DUSP2 mRNA levels were also assessed, and these were not measurably altered by knockdown of DUSP1 or -2 (not shown).

Fig. 9.

Effects of DUSP knockdown on ERK activation by GnRH, EGF, and PDBu. A and B, Cells were transfected in 96-well plates with ERK1/2 siRNAs and with either control siRNAs or with siRNAs targeting DUSPs 1, 2, and 4 (triple knockdown). They were also transduced with Ad ERK2-GFP and Ad mGnRHR before stimulation for the indicated periods with 1 μm GnRH. C and D, For the purpose of this comparison, the 60-min GnRH data from A and B have been plotted alongside previously published data from cells stimulated for 30 min with 10 nm EGF and or 1 μm PDBu (30 ). The data shown are pooled from six separate internal experiments (mean ± sem, n = 3–6), each with duplicate or triplicate wells. E, Cells were transfected in 96-well plates with ERK1/2 siRNAs and with either control siRNAs or with siRNAs targeting DUSPs 1, 2, and 4 (triple knockdown). They were also transduced with Ad ERK2-GFP, Ad mGnRHR, and Ad Egr-1 luciferase. After culture, they were stimulated for 4 h with GnRH (1 μm), PDBu (1 μm), or EGF (10 nm) and then processed for measurement of luciferase activity as described in Materials and Methods. The data shown are pooled from four separate experiments (mean ± sem, n = 2–4), each with duplicate or triplicate wells, and the figure shows Egr-1 luciferase activity in relative fluorescence units (RLU). **, P < 0.01; *, P < 0.05 compared with corresponding control values.

siRNA screening for DUSP effects on GnRH-mediated ERK signaling

In a final series of experiments, we screened a library of siRNAs targeting individual DUSPs for effects on GnRH-mediated ERK responses (Fig. 10). The MKP DUSPs can be divided into four major subgroups according to subcellular localization and substrate specificity. DUSPs 1, 2, 4, and 5 constitute the nuclear-inducible MKPs; DUSPs 6, 7, and 9 are the cytoplasmic ERK MKPs; DUSPs 8, 10, and 16 are known as the JNK/p38 MKPs; and the rest are termed the atypical DUSPs (21, 24, 26). Of these, DUSPs 1–6 and DUSP9 are known to directly dephosphorylate ERK in vitro (21, 24, 26), whereas the others can influence ERK activity by virtue of their regulation of JNK/p38 MAPK pathway cross talk with ERK. The data herein extend a previous screen for effects on responses to PDBu and EGF. In validating this approach, we used quantitative PCR (qPCR) to screen for expression and knockdown of 22 separate DUSPs and have included only the 16 siRNAs for which we were able to demonstrate clear expression and knockdown (mRNA reduced >75%) in HeLa cells (34). As expected, these data revealed no significant effect of knockdown of the nuclear-inducible MKPs DUSP1, -2, and -4 on ppERK2 levels or N:C ERK2-GFP ratios after either acute (5 min) or sustained (120 min) stimulation with GnRH. However, siRNAs targeting DUSPs 5, 10, and 3 all reduced ppERK2 levels after acute stimulation with GnRH, whereas siRNAs targeting DUSPs 9, 16, and 3 all reduced N:C ERK2-GFP ratios at 120 min. Thus, we have found 1) that a surprisingly large proportion of these phosphatases (five of 16 tested) can shape ERK responses to GnRH, 2) that GnRHR responses are influenced by phosphatases in each of the major MKP DUSP groups (the nuclear-inducible DUSP5, the cytoplasmic ERK MKP DUSP9, the JNK/p38 MKP DUSP16, and the atypical DUSP, DUSP3), 3) that the two endpoints measured yielded different hits, indicating distinct effects on activation and compartmentalization, and 4) that each of the siRNA hits caused inhibition of the GnRH effect, indicating involvement of the phosphatases in negative feedback control rather than direct dephosphorylation of ERKs. An important caveat to these data is that we have validated the knockdowns by qPCR rather than at the protein level. Validated antibodies are not available to most of the target proteins, and little is known about their relative turnover rates or the cellular levels of these proteins needed for function. Accordingly, for those siRNAs that failed to alter GnRH responses, we cannot exclude the possibility that protein expression was simply insufficiently reduced to reveal their regulatory role.

Fig. 10.

A screen for DUSP effects on GnRH signaling to ERK. Cells were transfected in 96-well plates with ERK1/2 siRNAs and with either control siRNAs or with siRNAs targeting each of the DUSPs indicated. They were also transduced with Ad ERK2-GFP and Ad mGnRHR before stimulation (0, 5, or 120 min) with 1 μm GnRH and then processed for imaging to determine whole-cell ppERK2 levels as well as the N:C ERK2-GFP ratio. Each of the indicated DUSPs are detectable by qPCR in these cells, and the siRNAs reduce expression of the appropriate transcripts by more than 70% (30 34 ). mRNA for DUSPs 13, 15, and 21 were not detectable in these cells, and we were unable to reduce DUSP8 expression by more than 30%, so these were excluded from the screen. None of the siRNAs that affected GnRH responses had any measurable effect on cell number or whole-cell ERK2-GFP levels, demonstrating that they do not have general toxic effects on cell survival, proliferation, transcription, or translation (supplemental table, published as supplemental data on The Endocrine Society’s Journals Online web site at http://mend.endojournals.org). One-way ANOVA revealed the siRNAs as significant variables (for each time point and endpoint), and statistically significant effects of the individual siRNA Smartpools are indicated (*, P < 0.05 by post hoc Bonferroni tests).

Summary and conclusions

Although many extracellular stimuli activate ERK, it is now well established that differences in activation and inactivation mechanisms generate stimulus-specific responses and that the spatiotemporal aspects of these signals can dictate the biological consequences of ERK activation. Thus, stimuli that cause proliferation characteristically elicit sustained activation of ERK and sustained nuclear accumulation of ERK by mechanisms involving neosynthesis of nuclear-inducible MKPs that can both inactivate and scaffold (retain) ERKs in the nucleus (43, 44). In contrast, nonproliferative stimuli often cause only transient activation and nuclear accumulation of ERKs that occurs independently of protein synthesis, implying dephosphorylation by preexisting protein phosphatases (44). In HeLa cells, PDBu shows the characteristics of a sustained stimulus (30, 34), namely sustained activation and nuclear accumulation of ERK that is associated with increased expression of nuclear-inducible MKPs and is sensitive to (combined) knockdown of nuclear-inducible MKPs and to protein synthesis inhibition with CHX. In contrast, EGF shows the characteristics of a transient ERK activator, namely transient increases in ERK activity and nuclear localization, less pronounced effects on nuclear-inducible MKP expression, and insensitivity to knockdown of nuclear-inducible MKPs or to protein synthesis inhibition (30, 34). In this regard, the GnRHR is particularly intriguing because in HeLa cells, it mediates a transient activation of ERK (even more transient than the response to EGF) despite the fact that its effect in these cells is PKC dependent and EGF receptor independent (12). Moreover, the rapid termination of the GnRH effect cannot be attributed to receptor desensitization because mammalian GnRHRs do not rapidly desensitize (14, 16, 45, 46). Indeed, ppERK levels increase to a maximum at 5 min and then reduce with a half-time of approximately 10 min, despite the fact that [³H] inositol phosphate accumulation experiments reveal no desensitization during the first 60 min of stimulation with GnRH (47), so the ERK response is transient despite maintained upstream activation.

In GnRH-stimulated cells the rapid inactivation of ERK and the lack of sustained nuclear ERK accumulation suggest ERK dephosphorylation by preexisting phosphatases rather than by nuclear phosphatases induced by stimulation with GnRH, and this is confirmed by the fact that the combined knockdown of nuclear-inducible DUSPs 1, 2, and 4 failed to influence GnRH effects on ppERK2 levels of ERK2-GFP compartmentalization. Nevertheless, CHX and the D319N mutation of ERK did inhibit inactivation of the GnRH effect on ppERK2. Thus, GnRH-mediated ERK activation shares characteristics seen with PKC-mediated ERK activation (dependence on protein neosynthesis and D-domain-dependent binding), but in the case of PKC activation, this can be attributed to coordinated effects of nuclear-inducible DUSPs 1, 2, and 4, but with GnRH activation, this is clearly not the case. We know of no other system in which a transient ERK response is inactivated in a CHX-dependent manner and suggest that the GnRH effect may be terminated by a preexisting rapid turnover protein (or proteins).

A puzzling feature of our data is that GnRH causes a brief and PKC-dependent activation of ERK, whereas PDBu activates PKC to cause a characteristically sustained ERK activation. The reasons for this distinction remain unknown, although it clearly cannot be attributed to GnRHR desensitization. We suspect that this may relate to differences in PKC isozymes activated or to differences in compartmentalization of PKC activation. An alternative possibility is that it reflects concomitant activation of other signaling pathways by the GnRHR.

Finally, although our data reveal no evidence for termination of GnRHR signaling by the nuclear-inducible MKPs, DUSP1, -2, or -4, the siRNA screen revealed that a high proportion of DUSPs (five of the 16 tested) can influence either the amplitude of the GnRHR-mediated ppERK2 response or the localization of ERK2-GFP in GnRH-stimulated cells. Indeed, these data provide the first evidence that GnRH effects can be influenced by members of each of the major DUSP subtypes. However, it is important to recognize that in each case, DUSP knockdown actually reduced the GnRH effect (on ppERK2 or on ERK2-GFP localization), arguing against a direct role for these proteins in dephosphorylation of ERK in GnRH responses. This is in accord with our previous screen in which 12 of the 16 siRNAs tested influenced responses to EGF and/or PDBu, and in all but one case, the siRNA effects were inhibitory. This presumably reflects the complexity of signaling networks in which DUSPs may influence the interaction between multiple MAPK pathways as well as the positive- and negative-feedback pathways.

Materials and Methods

Engineering of plasmids and viruses

Ad-expressing WT and D319N ERK2-GFP as well as mGnRHRs and Egr-1 promoter luciferase reporter were prepared, grown to high titer, and purified as described (12, 30, 34). The Ad CMV β-galactosidase vector was a gift from Prof. James Uney (University of Bristol, UK).

Cell culture and transfection

HeLa cells were cultured in 10% fetal calf serum (FCS)-supplemented DMEM. LβT2 cells were kindly provided by Prof. P Mellon (University of California, San Diego, CA) and were cultured as described (48). For ERK knockdown experiments, HeLa cells cultured in 96-well plates were transfected with 1 nm nontargeting control siRNAs or siRNAs targeted to noncoding regions of ERK1/2 as described (19, 30, 34), using RNAiMAX reagent (Invitrogen, Paisley, UK). For DUSP siRNA transfection, 10 nm SMARTpool or nontargeting control siRNA mixtures (Dharmacon, Cramlington, UK) were included in transfections. For Ad-mediated protein expression, cells were transduced (16 h after siRNA transfection) with 1.5 × 10⁶ plaque-forming units (pfu)/ml Ad WT or D319N ERK2-GFP and, where specified, with 2 × 10⁶ pfu/ml Ad mGnRHR vector in DMEM with 10% FCS. For luciferase assays, Ad Egr-1-luciferase and Ad CMV β-galactosidase vectors were included at 1 × 10⁶ pfu/ml. The Ad-containing medium was removed after 4–6 h and replaced with fresh DMEM with 0.1% FCS. The cells were then maintained for 16–24 h in culture before stimulation with GnRH (Sigma, Poole, UK), EGF (Calbiochem, San Diego, CA), or PDBu (Sigma).

Western blotting

HeLa cells were simultaneously plated and transfected in six-well plates (2.5 × 10⁵ cells per well) with 1 nm control or ERK1/2 siRNAs. Some cells were then transduced with Ad ERK2-GFP, as above. After treatment noted in figure legends, cells were lysed as described (12, 47) before Western blotting. Total and ppERK1/2 were detected using polyclonal rabbit anti-total ERK1/2 and rabbit anti-ppERK1/2 (1:1000; Cell Signaling Technology, Hitchin, UK), respectively. Loading controls were assayed by staining parallel blots with mouse anti-β-actin (AC-15, 1:5000; Sigma).

qPCR

To quantify expression and regulation of DUSP mRNAs, HeLa cells were simultaneously plated and transfected in six-well plates (2.5 × 10⁵ cells per well) with 1 nm ERK1/2 siRNAs and 10 nm control or SMARTpool siRNAs before Ad transduction as described above. Cells were kept in reduced-serum media before stimulation with 1 μm GnRH or 1 μm PDBu. Extraction of total RNA was performed using an RNeasy kit according to the manufacturer’s instructions (QIAGEN, Crawley, UK). Contaminating genomic DNA was removed from columns using an additional deoxyribonuclease (QIAGEN) digestion step. cDNA was then prepared for 1 μg of each total RNA sample using a cloned AMV first-strand synthesis kit according to the manufacturer’s instructions (Invitrogen, Paisley, UK). cDNAs were then quantified relative to expression of human GTPase-activating protein (hGAP) using previously described primers (30). The PCR primers were mixed with 50 ng RT-PCR template and SYBR green PCR master mix (Applied Biosystems, Warrington, UK), and the comparative cycle threshold method was used to detect relative expression curves on an ABI PRISM 7500 detection system (Applied Biosystems).

Semiautomated image acquisition and analysis

Cells were transfected with siRNA, transduced with Ad vectors, and plated as described above on Costar plain black-wall 96-well plates (Corning, Arlington, UK). After treatment with GnRH or PDBu, cells were washed in ice-cold PBS before fixation and staining for ppERK1/2 and imaging as described (30, 34). In some experiments, cells were treated with 30 μm CHX (Sigma) for 30 min before stimulation. Image acquisition was performed on an IN Cell Analyzer 1000 microscope, using a ×10 objective (GE Healthcare, Amersham, UK). Analysis of ppERK1/2 staining and localization was performed using the Dual Area Object Analysis algorithm in the IN Cell Analyzer Workstation (IN Cell Investigator, GE Healthcare) using 4′,6-diamidino-2-phenylindole (DAPI) and ppERK1/2 images. ERK2-GFP localization and ppERK2 staining was simultaneously analyzed using the Multitarget Analysis algorithm (IN Cell Investigator, GE Healthcare) using ERK2-GFP, ppERK2, and DAPI images (ERK2-GFP and DAPI images were used to define whole-cell and nuclear regions, respectively). Single cells expressing supraphysiological levels of ERK2-GFP were excluded from analysis (∼20% of cells) using appropriate gating parameters to prevent misleading localization data (30, 34). A total of 300–500 cells per field were typically analyzed, and up to four fields per well were captured in experiments performed in duplicate or quadruplicate, meaning that in each experiment, data were normally derived from at least 1000 individual cells per time point. Imaging data are reported as ppERK2 intensity (mean fluorescence intensity per cell) or as a ratio of nuclear to cytoplasmic intensity (N:C ratio) of ERK2-GFP.

Luciferase assays

Cells were transfected with siRNA, transduced with Ad vectors, and plated as described above on Costar plain black-wall 96-well plates (Corning) but including Ad Egr-1-luciferase and Ad CMV β-galactosidase reporter vectors. After treatment with GnRH or PDBu, cells were washed in ice-cold PBS, lysed, and assessed for luciferase activity by chemical luminescence after the addition of luciferin substrate (Promega, Southampton, UK). β-Galactosidase activity was used to correct luciferase activity for transfection efficiency, as measured after the addition of chlorophenol red-β-d-galactopyranoside substrate (Roche, East Sussex, UK).