MSK1 and MSK2 have been shown previously to be activated via both the classical MAPK cascade and the SAPK2/p38 pathway in several cell lines (9
), and we have confirmed that this is also the case in primary murine fibroblasts by the use of specific inhibitors. Although PD-184352 blocks the activation of MKK5 and its substrate ERK5, as well as the classical MAPK cascade, higher concentrations of this compound are required to block the MKK5/ERK5 pathway in fibroblasts (29
). In particular, incubation of cells with 2 μM PD-184352 completely blocks the classical MAPK cascade, at which concentration the MKK5/ERK5 pathway is unaffected. Although most of the experiments presented in this paper were carried out at 5 μM PD-184352, we have also established that the mitogenic activation of MSK1 and MSK2 and the phosphorylation of CREB are completely blocked at 2 μM PD-184352 (36
), excluding an involvement of the MKK5/ERK5 pathway.
In primary murine fibroblasts, TPA activates MSK1 and MSK2 via the classical MAPK cascade while anisomycin activates them largely via the SAPK2/p38 pathway. Both signaling pathways contribute to the activation of MSK1 and MSK2 by EGF and UV-C radiation (Fig. ). We have also shown that, as in other cells (9
), the inhibition by PD-184352 and/or SB-203580 of the activation of MSK1 and MSK2 by TPA, EGF, anisomycin, or UV-C radiation correlates with inhibition of the phosphorylation of CREB and ATF1 in response to the same stimuli. In the present study we used gene targeting to produce knockout mice that lack MSK1, MSK2, or both of these protein kinases in order to evaluate whether these or other protein kinases mediate mitogen- and stress-induced activation of CREB.
Primary embryonic fibroblasts from these mice were used to determine the effects of the knockouts on MAPK signaling cascades. The single knockout of MSK1 or MSK2 and the double knockout of both did not affect other components of the classical MAPK cascade since the activation of ERK1, ERK2, and MAPKAP-K1b/RSK2 in the knockout cells was similar to that in wild-type cells. Similarly, the single and double knockouts did not affect the activation of SAPK2/p38 or its substrate MAPKAP-K2 (Fig. ). The knockout of MSK1 did not lead to a compensatory increase in the level of MSK2 protein or activity, and vice versa.
Our results clearly demonstrate that MSK1 and MSK2 are the major, if not the only, protein kinases that mediate the phosphorylation of CREB at Ser133 and of ATF1 at Ser63 in fibroblasts by agonists that activate SAPK2/p38 in fibroblasts, because the phosphorylation of these transcription factors in response to anisomycin or UV-C radiation was virtually abolished in the MSK1/MSK2 double knockouts. A partial reduction of CREB and ATF1 phosphorylation was observed in the single knockouts, indicating that MSK1 and MSK2 both contribute to the stress-induced phosphorylation of CREB and ATF1 (Fig. ). The absence of CREB phosphorylation in the double-knockout cells was not due to a decrease in the expression of the CREB protein, and, in addition, CREB was phosphorylated normally by PKA in response to cyclic AMP-elevating agents (Fig. ).
Consistent with our earlier report that MSK1 knockout ES cells show a considerable reduction in CREB and ATF1 phosphorylation in response to TPA and EGF, the phosphorylation of CREB and ATF1 was also reduced in MSK1 knockout fibroblasts (Fig. ). However, the reduction was not as great as in ES cells, and this may be due to the higher levels of MSK2 in fibroblasts than in ES cells, where MSK2 activity could not be detected. The double knockout of MSK1 and MSK2 caused a further reduction in CREB and ATF1 phosphorylation compared to that in the MSK1 knockout fibroblasts, indicating that MSK2 contributes to CREB and ATF1 phosphorylation in response to TPA and EGF.
Some residual phosphorylation of CREB and ATF1 was seen in fibroblasts from the double knockout of MSK1 and MSK2 in response to TPA and EGF (Fig. ), suggesting that another protein kinase can phosphorylate CREB in the absence of MSK1 and MSK2. Moreover, this residual phosphorylation was blocked by prior incubation with PD-184352 but was unaffected by SB-203580, indicating that it is likely to be catalyzed by another protein kinase activated by ERK1 or ERK2. No additional MSK isoform appears to be present in the human genome. Moreover, the residual phosphorylation was unaffected by H-89 at concentrations that inhibit MSK1 and MSK2 (9
) and PKA (11
) in cells. The effects of the inhibitors on the residual CREB phosphorylation are consistent with it being catalyzed by an MAPKAP-K1/RSK isoform (26
). However, this appears not to be the case, because a potent cell-permeable inhibitor of MAPKAP-K1b/RSK2 that has recently been developed and that does not inhibit MSK1 and many other protein kinases tested has no effect on the residual TPA-induced phosphorylation of CREB in the MSK1/MSK2 double-knockout cells under conditions where the phosphorylation of other MAPKAP-K1/RSK substrates is blocked completely (G. Sapkota, D. R. Alessi, and G. Wiggin, unpublished experiments).
The evidence presented in this paper indicates that MAPKAP-K1b/RSK2 is not rate limiting for the phosphorylation of CREB in fibroblasts. Instead, our data indicate that MSK1 and MSK2 are the major mediators of CREB and ATF1 phosphorylation after mitogenic stimulation. This is consistent with CREB phosphorylation in fibroblasts being unaffected by the knockout of the MAPKAP-K1b/RSK2 gene (8
). However it conflicts with the report that immortalized fibroblasts from human patients with Coffin-Lowry syndrome, which possess an inactivating mutation in the MAPKAP-K1b/RSK2 gene (40
), do not phosphorylate CREB in response to EGF (15
). These findings led to the conclusion that MAPKAP-K1b/RSK2 is the major, if not the only, protein kinase that mediates the mitogen-induced phosphorylation of CREB in fibroblasts. We have therefore repeated the experiments with the same immortalized fibroblasts from Coffin-Lowry syndrome patients (a generous gift from A. Hanauer). In our hands, CREB phosphorylation was readily detectable in response to either EGF or TPA in the Coffin-Lowry cells (A. Kieloch, S. Arthur, and D. R. Alessi, unpublished work). While it remains possible that MAPKAP-K1b/RSK2 mediates the phosphorylation of CREB in cells that we have not yet studied, the finding that a dominant-negative MAPKAP-K1b/RSK2 mutant inhibits the mitogen-induced phosphorylation of CREB could be explained by the binding of this mutant to ERK1 or ERK2, which may prevent the mitogenic activation of MSK1 and MSK2.
Mice with knockouts of MSK1, MSK2, or both MSK1 and MSK2 were viable and fertile and showed no obvious health defects. This is in contrast to knockout or transgenic CREB mutants (7
), which have severe phenotypes. This difference may reflect the wider role of CREB in mediating the effects of many agonists that act via different signal transduction cascades. In particular, agonists that elevate the level of cyclic AMP and that activate PKA also induce the phosphorylation of CREB, and this is unimpaired in the MSK1/MSK2 knockout cells (Fig. ).
The phosphorylation of CREB has been implicated in the induction of certain immediate-early genes, such as c-fos
, as CREB phosphorylation appears to be sufficient for induction of these genes via PKA. In contrast, CREB phosphorylation induced by activation of different MAPK cascades is not sufficient for gene induction, and phosphorylation of additional transcription factors, such as TCFs, is also required (1
). It has been shown previously that approximately 50% of c-fos
induction by UV-C radiation in a variety of cell types, including fibroblasts, can be blocked by a dominant-negative form of CREB that cannot be phosphorylated. Mutation of the CRE site in the c-fos
promoter also caused a 50% reduction in c-fos
, and the remaining induction was unaffected by dominant-negative CREB, indicating that the c-fos
promoter can be transcribed independently of the CRE site and CREB (1
). Similar results were also obtained from PC12 cells with a dominant-negative form of CREB that blocked CREB binding to DNA (21
). This was found to reduce c-fos
induction by NGF but completely blocked c-fos
induction by PKA. Thus the lack of CREB phosphorylation in the MSK1/MSK2 double knockout would not be expected to completely block c-fos
induction. Consistent with these previous observations, a 50% reduction in anisomycin-stimulated c-fos
induction in the MSK1/MSK2 double-knockout cells (Fig. ), where there is almost no CREB phosphorylation (Fig. ), was observed. The induction of c-fos
in the MSK1/MSK2 double-knockout cells is likely to result from the direct SAPK2a/p38- and ERK-catalyzed activation of the TCF that binds to the SRE in the c-fos
promoter. The much smaller reduction in c-fos
gene transcription observed in the double-knockout cells after stimulation with TPA or EGF (Fig. ) may be explained by the residual phosphorylation of CREB after stimulation with TPA or EGF (Fig. ). If this is the case, then relatively small increases in CREB phosphorylation may be sufficient to produce maximal effects on the transcription of immediate-early genes.
The induction of junB
in the double-knockout fibroblasts was also reduced. The transcription of junB
is less well studied; however it has been reported to be controlled by SRE- and CRE-like sequences, which are located both 3′ and 5′ to the gene (3
). The reduction of anisomycin-stimulated induction of junB
in the double knockouts may be due to the reduction of CREB phosphorylation in these cells.
gene promoter contains a CRE and multiple SRE sequences. It has been reported that the mutation of the CRE reduces the induction from the egr1
promoter in response to stress and that inhibition of the egr1
promoter by SB-203580 is lost when the CRE site is mutated in 293 cells (34
). However, there was no significant reduction in the transcription of egr1
in response to anisomycin in the fibroblasts from the MSK1/MSK2 double-knockout mice, even though the phosphorylation of CREB was abolished (Fig. and ). The induction of egr1
in response to anisomycin was also inhibited by SB-203580 to the same extent in wild-type and double-knockout fibroblasts (Fig. ).
The TPA- and EGF-stimulated induction of egr1
in fibroblasts from the double knockout animals was also unaffected. It therefore appears that the phosphorylation of CREB is not required for the transcription of egr1
in fibroblasts. It is possible that, in the absence of CREB phosphorylation, another SB-203580-sensitive transcription factor that occupies the CRE is able to stimulate transcription of egr1
in fibroblasts. Alternatively, the results may reflect differences in the relative importance of the CRE and SRE in the egr1
promoter between the primary fibroblasts used in the present study and the 293 cells used previously (34
In summary, we have identified MSK1 and MSK2 as the major protein kinases that phosphorylate CREB and ATF1 in fibroblasts in response to mitogens and cellular stress and have shown that CREB phosphorylation is required for the full induction of c-fos
in response to anisomycin and UV-C radiation. We are currently studying whether CREB phosphorylation is deficient in other cells and tissues in the MSK1/MSK2 double-knockout mice. Since CREB has been implicated in learning and memory (7
), it will clearly also be of great interest to examine whether this process is impaired in the MSK1 and MSK2 knockout mice. These animals should facilitate the identification of additional physiological substrates and roles of MSK1 and MSK2.