INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

During mitosis the chromatin fiber is fully condensed and most of the transcriptional activity is inhibited (24, 34). These processes are associated with modifications in the tail of the core histones (11, 18, 23, 62) and with changes in the location of DNA and chromatin binding proteins (31, 41). Components of the transcriptional apparatus, transcription factors (35, 50, 51), and chromatin remodeling complexes (43, 54), are displaced from mitotic chromatin, while proteins that promote condensation seem to be enriched in this chromatin (14, 26, 32, 58).

Mitotic chromosomes are also depleted of high-mobility group B (HMGB) (HMG1 and -2) and HMGN (HMG14 and -17) (7, 21, 27, 29, 53) groups of proteins that serve as architectural elements and affect the structure and function of chromatin (5, 10). (Note that the nomenclature of the high-mobility-group proteins has been recently revised; HMGN1 was HMG-14, and HMGN2 was HMG-17. For a full description see the HMG Chromosomal Proteins Nomenclature Home Page http://www.informatics.jax.org/mgihome/nomen/genefamilies /hmgfamily.shtml.) HMGN proteins are the only nuclear proteins that specifically recognize the generic structure of the 147-bp nucleosome core particle (8). The proteins bind cooperatively to nucleosomes and form complexes containing two molecules of either HMGN1 or HMGN2 (49). Several types of experiments suggest that HMGN1 and HMGN2 proteins decompact chromatin and enhance transcription from chromatin, but not DNA, templates (reviewed in references 5 and 10). The proteins decompact the chromatin fiber by targeting two main elements known to compact chromatin: histone H1 (1, 19) and the amino terminus of histone H3 (60).

The intracellular organization of HMGN1 and -N2 proteins is dynamic (46) and related to both transcriptional activity (28) and cell cycle events (27). In cells that are highly active in transcription, the proteins are finely dispersed throughout the entire nucleus and colocalize with sites of active transcription. Upon inhibition of transcription, the proteins are found in large aggregates. The proteins colocalize with DNA in interphase and prophase but not in metaphase or early anaphase. At the end of mitosis, concomitantly with the appearance of the nuclear envelope, the proteins are actively imported into the nucleus and associate with the DNA (27).

The mechanisms regulating the intracellular localization of HMGN proteins are not known. Both HMGN1 and HMGN2 are acetylated (3, 25, 56, 57) and phosphorylated (2, 13, 39, 55; see also references 20 and 30), modifications which are known to affect their target interactions and cellular localization (3, 25, 39, 40, 44, 55).

We address here the mechanisms whereby HMGN1 and -N2 proteins are prevented from binding to mitotic chromosomes. We demonstrate that the nucleosomal binding domain (NBD) of HMGN1 and -N2 proteins is highly and specifically phosphorylated only during mitosis and that this phosphorylation abolishes the ability of the proteins to bind to chromatin. We use point mutants to demonstrate that both in vivo and in vitro, the negative charge in the NBD of HMGN proteins is directly responsible for their inability to bind to chromatin. We demonstrate that, in living cells, the chromatin residence time of HMGN1 is affected by the charge of the NBD, thereby providing the first experimental evidence that this protein domain is functional in vivo. We suggest that the specific phosphorylation of the NBD of HMGN proteins serves to prevent the interaction of these proteins with their chromatin targets during mitosis.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Generation and characterization of site-specific antibodies. The phosphopeptide PKRKVS(P)SAEGA was used to elicit anti-S6P, an antibody against HMGN1 phosphorylated at serine 6. The phosphopeptide PKRRS(P)ARLS(P)AKP was used to elicit anti-NBD2P, an antibody specific for the phosphorylated nucleosome binding domains of HMGN1 and -N2. The antibodies were affinity purified by sequential purifications on columns containing either nonphosphorylated or phosphorylated peptide (Quality Controlled Biochemicals, Inc.). The affinity-purified antibodies were characterized by enzyme-linked immunosorbent assay and Western blotting analysis using recombinant HMG-14 and -17 phosphorylated in vitro at specific sites.

HMGN1 and -N2 proteins. Recombinant wild-type and mutant HMGN1 and -N2 were generated and purified as previously described (6). HMGN1 and -N2 point mutants were generated from the cDNA constructs by site-directed mutagenesis using the Quickchange Site-Directed Mutagenesis Kit (Stratagene). The mutant constructs were subsequently subcloned into either the pet30A expression vector for bacterial expression or the pCI-neo (Promega) or pEGFP-N (Clontech) vectors for mammalian expression.

In vitro phosphorylation of the recombinant HMGN1 and -N2 proteins. Recombinant HMGN1 and -N2 and recombinant mutated HMGN1S6A were phosphorylated with Rsk2 as recommended by the manufacturer (Upstate Biotechnology), with protein kinase A (PKA) in buffer A (45 mM Tris-Cl, 45 mM boric acid, 10 mM MgCl₂, 100 µM ATP, 2 mM CaCl₂, 0.5 mM EDTA, 1 mM EGTA, 1 mM dithiothreitol, 10% glycerol) and with protein kinase C (PKC; Upstate Biotechnology) in buffer A supplemented with a lipid coactivator solution at a final concentration of 50 µg of phosphatidyl serine and 50 µg of diacylglycerol per ml.

Electrophoresis and Western blotting. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis was done according to the Laemmli protocol, and acid-urea electrophoresis was performed as previously described with small modifications (37). After electrophoresis, gels were stained with Coomassie blue for densitometric analysis or proteins were transferred using a semidry system onto nitrocellulose membrane for Western blot detection performed with the ECL Plus Kit (Amersham).

Core particle mobility shift assays. Mobility shift assays to test for the binding of HMG proteins to core particles were done in 2× Tris-buffered saline buffer exactly as described previously (47).

Preparation of HMGN1 and -N2 proteins from the synchronized HeLa cell cultures. HeLa S3 cells (American Type Culture Collection) were grown in suspension in Eagle essential medium with the Joklik modification (Sigma), supplemented with 10% fetal bovine serum (FBS), 2 mM L-glutamine, and 1% of penicillin-streptomycin. Logarithmically growing cells (5 × 10⁵ to 6 × 10⁵ cells/ml) were synchronized in S phase by growing the cells in medium supplemented with 5 mM thymidine for 24 h and then allowing them to recover in medium without thymidine for 4 h. G₂-phase cells were generated by treatment with thymidine as in the S-phase cells, except that 8 h was allowed for the recovery. HeLa cells were enriched for M phase by an 18.5-h treatment with 0.4 µg of nocodazole per ml (33). The fraction of mitotic cells was determined microscopically using Giemsa (LabChem, Inc.)-stained metaphase spreads. Flow cytometry was used to determine the degree of cell synchronization. Cell cultures enriched for each phase of the cell cycle were harvested by centrifugation, and the HMGN1 and -N2 proteins were extracted and purified as previously described (3).

Transfection into HmgN1^/ fibroblasts and HeLa cells. HmgN1^/ fibroblasts were generated from 13-day embryos of mice bearing a targeted disruption of the HmgN1 gene. Northern, Western, and immunocytochemical analyses demonstrated that the mice and cells derived from these mice do not express HMG-14 mRNA and do not contain HMG-14 protein (Y. Birger, unpublished results). Plasmids expressing either native or mutant human HMGN1 were transfected into the fibroblasts by the Lipofectamine 2000 method (Gibco-BRL). Cells expressing the proteins were detected by confocal immunofluorescence microscopy. Plasmid expressing either HMGN1-green fluorescence protein (GFP) fusion protein or HMGN1S20,24E-GFP were transfected into HeLa cells also by the Lipofectamine 2000 method.

Fluorescence loss in photobleaching (FLIP). HeLa cells expressing either HMGN1-GFP wild-type protein or HMGN1S20,24E-GFP were used for the experiment. A spot ~1 µm in diameter was repeatedly bleached at intervals of 5 s with a 250-ms laser pulse using the 488-nm laser line of an Ar laser of 20-mW nominal output at a 100% intensity. The fluorescence intensity in areas distant from the bleached spot was measured after each bleach pulse. As a control, the fluorescence intensity in a neighboring, unbleached nucleus was measured. Values represent averages from at least seven cells ± the standard deviation.

Confocal microscopy. Cells grown on coverslips in Dulbecco modified Eagle medium (Life Technologies, Inc.) supplemented with 10% FBS (Gibco-BRL) at 37°C in 5% CO₂ incubator were washed with phosphate-buffered saline (PBS), fixed with 4% formaldehyde in PBS for 10 min at room temperature, washed with PBS, and incubated for 20 min in PBS containing 0.1% Triton X-100-1% FBS-0.1% NaN₃ (TNBS buffer). The primary antibody (at about 1 µg/ml in TNBS) treatment was done overnight in room temperature in a moist chamber. The coverslips were washed with PBS and incubated with secondary antibody labeled with either Texas red or fluorescein for 2 h at room temperature in a moist chamber. DNA was stained with TO-PRO3 (Molecular Probes). Following the final wash steps, coverslips were inverted onto glass slides using the ProLong anti-fade reagent (Molecular Probes) as the mounting medium.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

The NBD of HMGN1 and -N2 proteins is highly phosphorylated during mitosis. We previously noted that Hep2 mitotic chromosomes are devoid, or highly depleted of HMGN1 and -N2 (HMG-14 and -17) proteins (27). To better understand the molecular mechanism involved in the displacement of HMGN1 and -N2 from mitotic chromatin, we first studied the mitotic location of the proteins in HeLa cells, which have a high content of HMGN1 and -N2 proteins (9). Confocal immunofluorescent microscopy with these cells indicates that, while during interphase the proteins are dispersed throughout the entire nucleus and colocalize with DNA (not shown), HeLa mitotic chromosomes are devoid of HMGN2 (and HMGN1; not shown). In metaphase the proteins are dispersed throughout the entire cell (Fig. 1A4), and their location is clearly distinct from that of the mitotic DNA (Fig. 1A5). In contrast, the linker histone H1, whose nucleosomal location overlaps with that of HMGN1 and -N2 (1), colocalizes with the DNA throughout the entire cell cycle and is present in mitotic chromosomes (Fig. 1A1 to A3). Taken together with previous results (27), these data indicate that the intracellular organization of HMGN1 and -N2 changes during the cell cycle and that the proteins are not present in mitotic chromosomes.

View larger version (51K): [in this window] [in a new window]   FIG. 1.   (A) HMGN1 and -N2 proteins do not colocalize with DNA during metaphase. Confocal laser scanning immunofluorescence of a HeLa metaphase was done, with the antibodies indicated at the top of the panels. The images were recolorized to facilitate comparison of the mitotic localization of the proteins. (B) Schematic diagram of the phosphorylation pattern of recombinant proteins by various kinases. Phosphorylation was determined by incorporation of 32P into the various proteins. HMGN1N16, deletion mutant lacking the first 16 amino acids; HMGN1S6A, a point mutant in which the serine at position 6 was mutated to alanine; HMGN1S7C, a point mutant in which the serine at position 7 was mutated to cysteine; HMGN1S20,24E, a double mutant in which the serines at position 20 and 24 were mutated to glutamic acid (the result marked by an asterisk was due to the presence of serines in the C-terminal region of HMGN1; this mutant incorporated >20% of radioactivity compared to the wild type); HMGN2C37, deletion mutant lacking the 37 C-terminal amino acids; HMGN2S24,26A, a double mutant in which the serines at positions 24 and 26 were mutated to alanine. The plus and minus signs indicate whether the kinases indicated at the top incorporated radioactivity into the proteins. ND, not determined. (C) Characterization of antibodies to specific phosphorylation sites in HMGN1 and -N2 proteins. Western analysis of either wild-type HMGN1 or the HMGN1 Ser6Ala mutant was done as follows: panels 1 to 3, anti-Ser6P (anti-HMGN1 phosphorylated at serine 6); panels 4 to 6, anti-NBD2P (elicited against HMGN1 and -N2 phosphorylated in the two conserved serines present in the NBD); or 7, the corresponding Coomassie blue-stained gel demonstrates equal loading. The kinases used to phosphorylate the proteins are indicated to the left of each panel. The phosphorylation sites in the HMG proteins are diagrammed above the panels. (D) Western analysis of proteins extracted from either logarithmically growing or mitotic cells with the following: 1, anti-Ser6P; 2, anti-NBD2P; or 3, a mix of anti-HMGN1 and anti-HMGN2.

(i) Characterization of the antibodies to phosphorylated HMGN1 and -N2 proteins. To test the specificity of the antibodies, we treated recombinant native and mutant proteins with specific kinases. It was previously noted that PKA preferentially phosphorylates serine 6 in human HMGN1 (61), that RSK2 specifically phosphorylates serine 6 of mouse HMGN1 (59), and that PKC phosphorylate both HMGN1 and HMGN2 with similar efficiency (45). As schematically outlined in Fig. 1B, we found that PKA and RSK2 also efficiently phosphorylate recombinant human HMGN1, as measured by ³²P incorporation after labeling with [-³²P]ATP. RSK2 did not incorporate ³²P into an HMGN1 point mutant in which serine 6 was mutated to alanine, into a deletion mutant of HMGN1 lacking the first 16 amino acids, or into HMGN2 protein. PKA efficiently phosphorylated a point mutant of HMGN1 in which serine 7 was mutated to cysteine. Thus, RSK2 and PKA phosphorylate only serine 6 of HMGN1. PKC incorporated ³²P into wild-type HMGN1, wild-type HMGN2, a deletion mutant of HMGN1 lacking the first 16 amino acids, and an HMGN2 mutant lacking the C-terminal 37-amino-acid residues. In contrast, PKC did not phosphorylate an HMGN2S24,26A, a double mutant in which the only serines present in the protein, located in the NBD, were mutated to alanine. Likewise, the amount of ³²P incorporated into the HMGN1S20,24E mutant, in which the two serines located in the NBD were mutated to glutamic acid, was less than 20% of that incorporated into the wild-type protein. We assume that the small amount of phosphate was incorporated into the seven serines present in the C-terminal region of the protein. Thus, PKC specifically phosphorylates the serine residues in the conserved NBDs of HMGN1 and -N2 (Fig. 1B).

View larger version (33K): [in this window] [in a new window]   FIG. 2.   Cell cycle-dependent phosphorylation of HMGN1 and -N2. (A) Flow cytometry analyses. Dark black tracing, logarithmically growing cells; gray tracing, synchronized cells. (B) Western analyses with proteins extracted from the stages indicated at the top with: 1, anti-Ser6P; 2 and 3, anti-NBD2P; 4, anti-HMGN1 (demonstrates equal loading).

(ii) Mitotic phosphorylation of HMGN1 and -N2 proteins. To test the phosphorylation levels of HMGN1 and -N2 proteins during mitosis, we performed immunoblot analysis with proteins obtained from either nocodazole-arrested or logarithmically growing HeLa cells. Anti-NBD2P produced a strong signal with both HMGN1 and HMGN2 proteins extracted from mitotic cells but not with the HMGN proteins prepared from either phosphatase-treated mitotic extracts or logarithmically growing cells (Fig. 1D). Thus, the NBDs of both HMGN1 and HMGN2 are highly phosphorylated during mitosis. Likewise, the phosphorylation level of serine 6 in HMGN1 is significantly elevated during mitosis (Fig. 1D1).

View larger version (31K): [in this window] [in a new window]   FIG. 3.   High level of phosphorylation of the HMGN1 and -N2 proteins during mitosis. (A) Acid urea gel of proteins extracted from either logarithmically growing cells (left panel), HeLa cultures in which 55% of the cells were mitotic (center panel), and phosphatase-treated proteins isolated from 55% mitotic cells (right panel). The migration areas of the HMGN1 and HMGN2 bands are indicated on the right side of the gel. (B) Quantitative analysis of mitotic phosphorylation. High-resolution acid urea gels and corresponding densitometric scans of proteins were isolated either from logarithmically growing cells (panel 2), from 75% mitotic HeLa cells (panel 3), or phosphatase-treated proteins isolated from these mitotic cells (panel 4). The elec- trophoretic profile of PKC phosphorylated recombinant proteins, which served as markers, are shown in panel 1. Dotted lines show the positions of nonmodified mono- and diphosphorylated proteins.

Phosphorylation of the NBD abolishes the binding of HMGN1 and -N2 to nucleosome core particles. The high level of mitotic phosphorylation of the NBD of HMGN1 and -N2 raises the possibility that this modification prevents the binding of the proteins to mitotic chromatin. The primary binding target of HMGN1 and -N2 proteins in chromatin is the nucleosome core particle. At physiological ionic strengths, each nucleosome core binds two molecules of either HMGN1 or HMGN2 to form a specific complex that can be detected by mobility shift assays (49). The interaction of HMGN with nucleosomes involves specific contact not only with the nucleosomal DNA but also with the amino termini of the core histones (17, 60). We modified the proteins with the site-specific kinases using trace amounts of ³²P and used mobility shift assays to assess the effect of phosphorylation on nucleosome binding.

View larger version (42K): [in this window] [in a new window]   FIG. 4.   Phosphorylation in the NBD abolishes the binding of HMGN1 and -N2 to core particles. Equal amounts of nonphosphorylated or either HMGN1 or HMGN2 protein trace labeled with [32P]ATP were incubated with 1 µg of the core particles (Cp) at the molar ratios indicated at the top of each lane. (A) Wild-type HMGN1. (B) HMGN1 phosphorylated at serine 6. (C) HMGN1 phosphorylated in the NBD. (D) Wild-type HMGN2. (E) HMGN2 phosphorylated in the NBD. The mobilities of the HMG-core particle complex and of the core particle are indicated at the left of each panel. The asterisk indicates the molar concentration at which 50% of the cores were shifted. The location of the phosphorylated proteins was visualized by autoradiography (right side of panels B, C, and E). The double-headed arrows indicate the mobility of the Cp and the HMG-Cp complex. Panels B', C', and E' are scans and images of the boxed areas (HMG/Cp = 4.3) in the respective panels. Note that the phosphorylation at serine 6 of HMGN1 reduces, but does not abolish, the binding of the protein to nucleosomes. Part of the radioactivity comigrates with the shifted core particles. The HMGN1 and HMGN2 phosphorylated in the NBD failed to generate an HMG-Cp complex. The line with the asterisk indicates the position of the nonspecific aggregates.

In vivo, a negatively charged NBD prevents the binding HMGN1 to chromatin. Although it is generally accepted that the HMGN proteins interact with chromatin mainly through their NBDs, this point has never been examined in intact, living cells. To critically examine the hypothesis that the introduction of two negative charges into the NBD is sufficient to abolish the binding of the protein to chromatin in living cells, we generated a double point mutant of HMGN1, S20,24E, in which the two serine residues in the NBD were mutated to glutamic acid. Mobility shift assays with the bacterially expressed proteins revealed that, just like the phosphorylated wild-type proteins, the S20,24E double mutant does not bind to nucleosomes (Fig. 5A [compare with Fig. 4]). Thus, the negative charge in the NBD abolishes the interaction of the HMGN1 and -N2 proteins with nucleosome cores, in spite of the fact that the binding of HMGN protein to nucleosome core requires the presence of the positively charged histone tails (17, 60).

View larger version (34K): [in this window] [in a new window]   FIG. 5.   A negative charge in the NBD inhibits binding of HMGN1 to chromatin either in vitro or in vivo. (A) Mobility shift assays with wild-type HMGN1 and with the HMGNS20,24E double mutant. The HMG/Cp molar ratios are indicated on the top of the lanes. Note that the double-mutant HMGN1 S20,24E does not form a complex with core particles. (B) FLIP analysis in HeLa cells expressing either HMGN1-GFP or the double-mutant HMGN1S20,24E-GFP. Note that the fluorescence intensity of the cells expressing the mutant proteins was lost faster than in the cells expressing the native proteins, indicating that the mutant is more mobile. As a control, the fluorescence intensity in neighboring, unbleached nucleus was also measured. Values represent averages of at least seven cells with the standard deviations. (C) Confocal laser scanning fluorescence microscopy visualizing the mitotic location of either the human HMGN1 or the human double-mutant HMGN1S20,24E in transfected HmgN1/ mouse fibroblasts. The proteins are labeled green, and the DNA is labeled red. Both the wild-type protein and the mutated protein do not colocalize with the metaphase chromosomes.

/

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

We report here that the NBD of the HMGN1 and -N2 protein family is highly and specifically phosphorylated during mitosis and that this phosphorylation has a major functional consequence: it abolishes the interaction of the proteins with its chromatin targets. Moreover, we provide insights into the molecular mechanisms involved in this effect and demonstrate that the abolishment of chromatin binding is accomplished by introducing negative charges in the NBD region of the protein. This finding is relevant to understanding the functional consequences of the mitotic phosphorylation of chromatin binding proteins.

Timing and specificity of HMGN phosphorylation. Several potential phosphorylation sites in HMGN1 and -N2 have been previously identified (reviewed in references 20 and 30); however, recent studies indicate that serine 6 in HMGN1 and the two conserved serines in the NBD region of the HMGN1 and -N2 protein family are the only in vivo phosphorylation sites (2, 39, 55). Even a detailed analysis of okadaic acid-treated cells failed to reveal additional phosphorylation sites in these proteins (39). Since serine 6 of HMGN1 is associated with the expression of immediate-early genes (2, 13, 59), it can be expected that this modification would be present in logarithmically growing, cycling cells. Indeed, mass spectroscopic analysis also detected a low level of monophosphorylated HMGN1 in cycling K562 (39). In full agreement, our immunological analysis also indicates a low level of phosphorylation in logarithmically growing cells, specifically at serine 6 of HMGN1 (Fig. 2B).

Phosphorylation prevents the interaction of HMGN proteins with chromatin. Footprinting (1), cross-linking (4, 16, 60), and site-directed mutagenesis (48) experiments suggest that the interaction of HMGN proteins with nucleosome cores involved multiple precise, but weak, interactions. The interaction of the proteins with nucleosomes is also affected by posttranslational modifications (3, 25, 55). Indeed, we find that phosphorylation of serine 6 in the N-terminal region of HMGN1 weakens the binding of the protein to chromatin, but to a lesser degree than that previously noted by Spaulding et al. (55). In fact, our studies with ³²P-labeled proteins clearly demonstrate that even under cooperative binding conditions, the phosphorylated protein binds to nucleosomes (Fig. 3).

Biological significance of HMGN mitotic phosphorylation. Our data, as well as the information previously available, indicate that the phosphorylation of HMGN proteins is associated with specific metabolic events, suggesting that this process serves a role in modulating the activity of HMGN proteins. HMGN1 and -N2 proteins are architectural elements that decondense chromatin and enhance transcription from chromatin templates (5, 10). The proteins colocalize with active sites of transcription and displacement of HMGN from nuclei inhibits transcription (28). In transcriptionally active, logarithmically growing cells the proteins are highly mobile (46), and conceivably their NBD could be readily targeted by kinases. It is therefore possible that in the nucleus the phosphorylation state of these proteins is in a constant state of turnover and that the equilibrium is favoring greatly the dephosphorylated state. However, phosphorylation of the HMGN NBD is clearly incompatible with a nuclear location and, upon okadaic acid treatment, phosphorylated proteins are found only in the cytoplasm (39). We find a very high level of phosphorylation in mitotic cells when the nuclear membrane is disaggregated but not in logarithmically growing, transcriptionally active cells when the nuclear membrane is intact. Thus, all the evidence is consistent with the notion that phosphorylation of the NBD is exclusively a mitotic event.