Requirement of Rsk-2 for Epidermal Growth Factor-Activated Phosphorylation of Histone H3

See allHide authors and affiliations

Science  06 Aug 1999:
Vol. 285, Issue 5429, pp. 886-891
DOI: 10.1126/science.285.5429.886


During the immediate-early response of mammalian cells to mitogens, histone H3 is rapidly and transiently phosphorylated by one or more unidentified kinases. Rsk-2, a member of the pp90rsk family of kinases implicated in growth control, was required for epidermal growth factor (EGF)–stimulated phosphorylation of H3. RSK-2mutations in humans are linked to Coffin-Lowry syndrome (CLS). Fibroblasts derived from a CLS patient failed to exhibit EGF-stimulated phosphorylation of H3, although H3 was phosphorylated during mitosis. Introduction of the wild-type RSK-2 gene restored EGF-stimulated phosphorylation of H3 in CLS cells. In addition, disruption of the RSK-2 gene by homologous recombination in murine embryonic stem cells abolished EGF-stimulated phosphorylation of H3. H3 appears to be a direct or indirect target of Rsk-2, suggesting that chromatin remodeling might contribute to mitogen-activated protein kinase–regulated gene expression.

In mammalian cells, various environmental stimuli induce a Ras-dependent MAP (mitogen-activated protein) kinase cascade that results in the transcriptional activation of immediate-early–responsive genes (1, 2). These transcriptional responses are thought to depend on modulation of the nuclear localization, DNA binding, and activation properties of transcription factors, but the roles of MAPK phosphorylation in this process remain poorly defined (1).

Remodeling of chromatin structure appears to have a primary role in transcriptional regulation (3), and posttranslational modifications of histones are thought to contribute to this remodeling. Widespread phosphorylation of histones, particularly histones H1 and H3, correlates with mitosis in many cells (4). Additionally, rapid and transient phosphorylation of a subset of histone H3 molecules correlates with the activated expression of immediate-early genes such as c-fos and c-jun in mammalian cells after mitogen stimulation (5). These observations suggest that H3 phosphorylation may contribute to chromatin remodeling during mitosis and transcriptional activation.

We have described a polyclonal antiserum (pS10) to a synthetic peptide mimic of histone H3 monophosphorylated at serine-10 that recognizes H3 phosphorylated during mitosis (6). To determine whether the pS10 antiserum detects mitogen-stimulated phosphorylation of H3, we isolated histones from mouse fibroblasts deprived of serum and then treated with a buffer control or EGF. The proteins were resolved by SDS–polyacrylamide gel electrophoresis (PAGE), transferred to a polyvinylidene difluoride (PVDF) membrane, and probed with pS10 antiserum. A rapid and transient increase in phosphorylation of H3 was detected in EGF-treated cells but not in unstimulated cells (Fig. 1A). Similar results were obtained when tetradecanoyl phorbol acetate was used in place of EGF (7, 8).

Figure 1

Correlation of EGF-activated histone H3 phosphorylation with increased activity of a 90-kD H3 kinase in nuclear extracts. (A) NIH 3T3 mouse fibroblasts were serum-starved and then stimulated with EGF for 5, 30, or 180 min (24). (Upper) pS10 immunoblot of acid-soluble nuclear proteins resolved by SDS-PAGE (25). The corresponding region of a gel run in parallel and stained with Coomassie blue is shown in the lower panel. (B) Equivalent amounts of nuclear extract from NIH 3T3 cells after serum-starvation (−EGF) and after EGF treatment for 30 min (+EGF) were resolved by SDS-PAGE in an 8% gel containing histone H3 and processed to detect kinase activity (25, 26). The arrow denotes a band in the autoradiogram with an apparent molecular mass similar to that of pp90rsk, as shown in (C). The identity of the band at 50 kD is not known. (C) Parallel in-gel H3 kinase assays were performed as in (B) except that after electrophoresis, the 90-kD region for each sample was excised and again subjected to electrophoresis on a second SDS gel. The arrow denotes the position of pp90rskdetected when this second gel was analyzed by immunoblotting with pp90rsk antiserum.

To identify kinases that mediate EGF-dependent phosphorylation of histone H3, we used kinase activity gel assays with purified histone H3 or H3 NH2-terminal peptide substrates to detect H3-specific kinases. We consistently detected H3 phosphorylation by a polypeptide with an apparent molecular size of ∼90 kD in nuclear extracts from EGF-stimulated cells that was not detected in extracts from quiescent cells (Fig. 1B). This activity did not appear to be due to kinase autophosphorylation because phosphorylation was not detected when identical samples were analyzed on gels containing bovine serum albumin instead of H3 (9).

The molecular size and activation by EGF stimulation suggested that the catalytically active species might be a member of the pp90rsk (ribosomal S6 kinase) family of mitogen-activated serine-threonine kinases implicated in cell proliferation and differentiation (10). To test this hypothesis, nuclear extracts were run on an H3 kinase activity gel, and a portion of each gel lane encompassing the 90-kD region was excised immediately after electrophoresis. Proteins in these gel strips were subjected to electrophoresis on a second SDS gel and analyzed by immunoblotting with antiserum to pp90rsk. Consistent with the known nuclear localization of pp90rsk isoforms after mitogenic stimulation (11), an immunoreactive band with electrophoretic mobility similar to that of the kinase apparent in the activity gel was detected in nuclear extracts from EGF-stimulated, but not unstimulated, cells (Fig. 1C).

The pp90rsk proteins phosphorylate several different protein substrates in vitro (10). In mammals, the pp90rsk family comprises three functionally nonredundant isoforms: Rsk-1, Rsk-2, and Rsk-3 (12) [also referred to as MAPKAP-K1a, b, and c, respectively (2)]. We found that biochemically purified rabbit Rsk-2 [containing little or no Rsk-1 and Rsk-3 (13)] phosphorylated only histone H3 in both free histone and nucleosomal substrates (Fig. 2A). H3 phosphorylated by Rsk-2 in vitro was recognized by the pS10 antiserum (Fig. 2A), and microsequence analysis confirmed that Ser-10 is the only site in the NH2-terminus of H3 phosphorylated by Rsk-2 in vitro (Fig. 2B).

Figure 2

Preferential phosphorylation of histone H3 by Rsk-2 in vitro. (A) Free histones (Free) or mononucleosomes (Nuc.) were incubated with rabbit Rsk-2 in vitro (27). These reactions were then resolved by SDS-PAGE (12% gel) and analyzed by Coomassie blue staining, autoradiography and protein immunoblotting with pS10 antiserum. (B) After incubation with [γ-32P]ATP and rabbit Rsk-2, histone H3 was recovered by rpHPLC and subjected to NH2-terminal microsequencing analysis (28). About 50% of the products from each sequencing cycle were collected for liquid scintillation counting and the remainder used for residue identification by HPLC. They-axis values represent total 32P cpm eluted in each sequencing cycle plotted as a function of the residue identified in that cycle.

Because mutations in Rsk-2, but not Rsk-1 or Rsk-3, are associated with CLS (14), we compared pS10 staining of Epstein-Barr virus (EBV)–transformed fibroblasts derived from a CLS patient with that of similarly transformed control fibroblasts from an unaffected sibling. After stimulation with EGF, phosphorylated H3 was distributed in nuclei of normal (control) cells in a speckled pattern (Fig. 3) similar to that reported for mitogen-stimulated mouse fibroblasts (8). In contrast, nuclear staining was not detected in interphase CLS cells even though mitotic chromosomes were stained in CLS cells undergoing mitosis (Fig. 3). These data indicate that H3 is a target of (at least) two distinct signaling pathways: one that is dependent on Rsk-2 and linked to activation by EGF, and another that is Rsk-2-independent and associated with mitosis.

Figure 3

Effects of Rsk-2 deficiency on mitogen-stimulated but not mitotic phosphorylation of histone H3 in CLS cells. Representative fields from indirect immunofluorescence microscopy of transformed normal (upper) and CLS (middle) human fibroblasts, stimulated for 30 min with EGF and costained with DAPI and pS10 antiserum (23, 24), are shown in the left and right panels, respectively. The punctate nuclear localization of phosphorylated H3 in the normal cells (open arrowheads) is indicated. The lower pair of panels shows a second field of CLS fibroblasts in which a cell in mitosis (arrowheads) is visible.

Immunoblot analyses confirmed that H3 phosphorylation was not readily detected in serum-starved or EGF-treated CLS fibroblasts (Fig. 4A), nor was it detected after stimulation of these cells with serum or ultraviolet (UV) irradiation (9). A weak signal, equivalent in serum-starved and EGF-stimulated CLS cells, which we attribute to mitotic H3 phosphorylation in these slowly growing cultures, was detected only after lengthy exposures (9). Because the amounts of Rsk-1 and Rsk-3 expressed in these fibroblasts and several CLS lymphoblast lines are equivalent to those in normal cells (14, 15), these results suggest that the Rsk-1 and Rsk-3 isoforms are unable to compensate for the loss of Rsk-2 function.

Figure 4

Deficiency in EGF-stimulated phosphorylation of histone H3 and Rsk-2–associated H3 kinase activity in CLS cells. (A) Transformed normal and CLS human fibroblasts were dissolved in sample buffer containing SDS, resolved by SDS-PAGE (12% gel), and analyzed by immunoblotting with the pS10 antiserum and anti-tubulin (22). Phosphorylated H3 was detected in normal cells after serum starvation (−) and after the addition of EGF. The high level of phosphorylated H3 detected in these cells after serum deprivation appears to be associated with transformation-induced proliferation [see (B) and text]. Phosphorylated H3 was not detected in Rsk-2–deficient CLS cells regardless of the presence or absence of EGF. Tubulin levels demonstrate equivalent loading. (B) Transformed normal human fibroblasts [as in (A)] were synchronized by using coupled aphidicolin and nocodazole treatments (29). Metaphase-arrested cells were released into serum-free media and harvested 6 or 24 hours later with or without exposure to EGF for 30 min. Cells were dissolved in sample buffer containing SDS, resolved by SDS-PAGE (10% gel), and analyzed by immunoblotting with the pS10 antiserum and anti-CREB. The strong induction of H3 phosphorylation by EGF treatment 24 hours after release of the cells from mitotic arrest is apparent. CREB levels demonstrate equivalent loading. (C) Anti–Rsk-2 immunoprecipitates were prepared from whole-cell extracts of randomly growing transformed normal human and CLS fibroblasts as described (14). Immunoprecipitates were solubilized by boiling in SDS-PAGE sample buffer, and a portion (90%) of the material released was analyzed on a kinase activity gel containing purified H3 (left) (26). The remaining 10% of the released material was resolved on an SDS-PAGE gel and analyzed by protein immunoblotting with Rsk-2–specific antiserum (right). Rabbit skeletal muscle Rsk-2 (0.05 U) was used as a positive control (std). H3 kinase activity is markedly diminished in the CLS Rsk-2 immunoprecipitate.

Phosphorylation of histone H3 was observed in both serum-starved and EGF-stimulated cultures of the EBV-transformed control fibroblasts (Fig. 4A). This might reflect increased proliferation of these cells upon transformation. To test this hypothesis, we treated cultures with aphidicolin and nocodazole to arrest these cells in mitosis and monitored EGF-stimulated phosphorylation of H3 after releasing cultures from this cell cycle block. Augmented phosphorylation of H3 was observed in nocodazole-arrested cells (Fig. 4B). Six hours after releasing cells into serum-free medium, phosphorylation of H3 was not observed, regardless of whether cells were stimulated with EGF or not. However, by 24 hours, EGF treatment induced phosphorylation of H3 (Fig. 4B). Stimulation with serum or UV irradiation induced similar amounts of H3 phosphorylation in parallel cultures at this point (9). The low amount of phosphorylated H3 detected in unstimulated cells at 24 hours (Fig. 4B) may represent mitotic phosphorylation of H3 as the synchrony of cultures declined over time. Mitogenic induction of c-fos expression and MAPK signaling are conserved in these cells (15); thus, the H3 phosphorylation observed in asynchronous cultures of these cells may mask the mitogenic H3 phosphorylation response but appears not to be associated with alterations in mitogen response pathways per se.

Rsk-2 immunoprecipitates from serum-supplemented control cells contained more H3 kinase activity than immunoprecipitates prepared in parallel from CLS cells (9). When immunoprecipitates from control cells were analyzed, bands migrating just ahead of a rabbit Rsk-2 standard were detected on an H3 kinase activity gel and on a parallel Rsk-2 immunoblot (Fig. 4C). The heterogeneity resolved on the activity gel may reflect MAPK phosphorylation of Rsk-2. In contrast, Rsk-2 was not detected in the immunoprecipitate from CLS cells by immunoblotting, and a weak H3 kinase activity was barely detected in activity gels (Fig. 4C). Although the absence of Rsk-2 in immunoblots is consistent with the expectation that Rsk-2 protein expressed in these cells lacks the COOH-terminal epitopes recognized by the antiserum to Rsk-2 (14, 15), attempts to detect Rsk-2 expression in these cells with antiserum to an NH2-terminal epitope of pp90rsk were also unsuccessful (15). Thus, the weak signal in the activity gel associated with the Rsk-2 immunoprecipitate from CLS cells may be attributable to contaminating kinases, possibly Rsk-1 or Rsk-3. Other bands were not apparent in any of the samples, and the activity observed was not due to kinase autophosphorylation because activity was not detected when immunoprecipitates were assayed in activity gels prepared without protein substrate (7). We conclude that the H3 kinase activity in immunoprecipitates from normal cells was intrinsic to Rsk-2, and that H3 kinase activity in immunoprecipitates from CLS cells was diminished due to the lesion in the RSK-2 locus in this patient.

We assayed whether ectopic expression of Rsk-2 could restore mitogen-stimulated phosphorylation of histone H3 in CLS cells. Phosphorylation of H3 in response to EGF was detected in CLS fibroblasts transfected with RSK-2 DNA but not in cells transfected with vector DNA alone, even though MAPK was activated by phosphorylation upon EGF stimulation in both cell cultures (Fig. 5A). Changes in the amounts of Rsk-1 and Rsk-3 were not observed upon Rsk-2 expression (Fig. 5A). Furthermore, CLS fibroblasts transfected withRSK-2 DNA, but not those transfected with vector DNA alone, displayed punctate nuclear pS10 staining after EGF treatment (Fig. 5B). These results argue strongly that Rsk-2 can directly or indirectly mediate EGF-stimulated phosphorylation of H3 in vivo.

Figure 5

Restoration of mitogen-stimulated phosphorylation of histone H3 in CLS cells after ectopic expression of Rsk-2. (A) After transfection with a human Rsk2 expression vector (Rsk-2) or the empty vector (vector only), CLS fibroblasts were serum-starved and mock-treated (−EGF) or treated with EGF for 30 min (+EGF) and dissolved in boiling SDS buffer (30). Cellular proteins were resolved on 12% SDS-PAGE gels, and the expression of the indicated proteins was analyzed by immunoblotting with the respective antisera (13). Although MAPK phosphorylation is induced by EGF in both types of transfected cells, H3 phosphorylation was induced by EGF only in cells transfected with Rsk-2 DNA. The levels of Rsk-1, Rsk-3, and MAPK demonstrate equivalent loading. (B) Representative fields from indirect immunofluorescence microscopy of CLS fibroblasts that were serum-starved and stimulated with EGF for 30 min after transfection with a human Rsk-2 expression vector (lower) or the empty vector (upper) as in (A). Cells costained with DAPI and pS10 antiserum are shown on the left and right, respectively. The punctate nuclear localization of phosphorylated H3 in the Rsk-2-transfected cells (open arrowheads) is indicated.

RSK-2 mutations are closely associated, if not causally linked, to CLS (16). However, because mutations at additional loci may contribute to defects in mitogen-responsiveness in CLS cells, we specifically disrupted the RSK-2 gene in murine embryonic stem (ES) cells by homologous recombination (17) and assayed these cells for EGF-stimulated phosphorylation of histone H3. The absence of Rsk-2 expression in these cells was confirmed by protein immunoblotting (Fig. 6A), and disruption of the endogenousRSK-2 gene was confirmed by Southern (DNA) blot analyses (17). Augmented phosphorylation of H3 was detected by immunoblotting after EGF stimulation of serum-deprived wild-type (wt) ES cells (Fig. 6B). In contrast, no stimulation of H3 phosphorylation in response to EGF was detected in Rsk-2 ES cells even though MAPK was activated in these cells after EGF treatment (Fig. 6B). The amount of phosphorylated H3 in serum-deprived and EGF-stimulated Rsk-2 ES cells was similar to that in serum-deprived wt ES cells. Thus, EGF-stimulated phosphorylation of H3 requires Rsk-2. These observations were confirmed by pS10 staining in randomly growing wt and Rsk-2 ES cells. Punctate nuclear staining resembling that described above for normal human fibroblasts was observed in cultures of wt ES cells, but not in cultures of Rsk-2 ES cells (Fig. 6C). Mitotic phosphorylation of H3 in murine cells appears to be independent of Rsk-2 because it was readily detected in both wt and Rsk-2 ES cells (Fig. 6C), in agreement with our data for human fibroblasts (Fig. 3).

Figure 6

Rsk-2 is required for EGF-stimulated phosphorylation of histone H3 in mouse cells. (A) Confluent wild-type (wt ES) and Rsk-2–deficient (Rsk-2) cells (17) were starved for 48 hours in 0.5% FBS and then stimulated with EGF (30 ng/ml) for 30 min. Cells were dissolved in boiling SDS buffer and resolved on a 10% SDS-PAGE gel. Expression of Rsk-1, Rsk-2, and Rsk-3 was then analyzed by immunoblotting with isoform-specific antisera (13). (B) Samples prepared as in (A) from serum-deprived (−EGF) and EGF-stimulated (+EGF) wild-type and Rsk-2 ES cells were resolved on a 10% SDS-PAGE gel, and the levels of MAPK (ERK1/ERK2), phosphorylated MAPK, H3, and phosphorylated H3 were assessed by immunoblotting with the respective antisera (13). EGF stimulated MAPK phosphorylation in both wt and Rsk-2 cells, but H3 phosphorylation was deficient in Rsk-2cells. The levels of (nonphosphorylated) H3 and MAPK confirm equivalent loading. (C) Randomly growing wild-type (upper) and Rsk-2 (lower) ES cells were costained with DAPI and pS10 antiserum (left and right panels, respectively). Solid arrowheads indicate cells in mitosis. Phosphorylated H3 was localized in mitotic chromosomes of both wild-type and Rsk-2 cells, but punctate nuclear localization was apparent only in interphase wild-type ES cells (open arrowheads).

Our results indicate that Rsk-2 is required for EGF-stimulated phosphorylation of histone H3 in vivo. However, the role of H3 phosphorylation in cellular responses to mitogens remains to be defined. Although activation of cAMP response element–binding protein (CREB) phosphorylation and c-fos transcription (whose promoter contains a CREB recognition site) by EGF are defective in CLS cells (15), these same cells resemble normal cells in other responses to mitogenic and other stimuli (15). Thus, it appears that Rsk-2 ablation does not lead to global changes in the activity of other signaling pathways. Furthermore, our finding that EGF-activated phosphorylation of H3 is Rsk-2–dependent whereas EGF-induced phosphorylation of Elk-1 and serum response factor in CLS cells is not (15), indicates that not all MAPK-regulated nuclear responses require Rsk-2.

Evidence for the direct involvement of H3 phosphorylation in gene activation is limited (8). However, a growing body of evidence indicates that conserved regulatory pathways have evolved in eukaryotes wherein chromatin-modifying activities, notably histone acetylases and deacetylases, are recruited to specific promoters through selective interactions with activator and coactivator proteins (18). The growth factor–dependent interaction of pp90rsk with the transcriptional coactivator CBP (19), a histone acetyltransferase (20), raises the possibility that histone acetylation and phosphorylation may act together to facilitate gene expression at mitogen- responsive promoters such as c-fos and c-jun (5, 21).

  • * To whom correspondence should be addressed. E-mail: paolosc{at}; allis{at}

  • These authors contributed equally to this work. The contributions of the Sassone-Corsi and Allis labs were equal.


View Abstract

Navigate This Article