| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
J. Biol. Chem., Vol. 278, Issue 44, 43384-43393, October 31, 2003
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
**
From the
Department of Biochemistry and Molecular Biology, Milton S. Hershey Medical Center, Pennsylvania State University College of Medicine, Hershey, Pennsylvania 17033, the Department of Molecular Biology, M. V. Lomonosov Moscow State University, Moscow 119899, Russia, the ¶Department of Biochemistry and Molecular Biology, Monash University, Victoria 3800, Australia, and the ||Biology Department, University of Massachusetts, Amherst, Massachusetts 01003
Received for publication, July 15, 2003 , and in revised form, August 19, 2003.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
Heterochromatin spreading in terminally differentiated cells is associated with profound remodeling of higher order chromatin structure. Linker DNA at the entry/exit site of the nucleosome becomes more condensed, and extensive lateral interactions form between chromatin fibers (reviewed in Ref. 9). Such changes are driven by developmentally regulated factors. Surprisingly, the expression of major non-histone proteins that promote heterochromatin-specific repression in proliferating cells, such as Ikaros and heterochromatin protein 1, declines in terminally differentiated cells (10, 11) and thus cannot account for heterochromatin spreading.
As developing avian blood cells enter the final maturation stage and RNA and protein synthesis declines, few proteins capable of carrying out chromatin condensation are up-regulated. Nucleated erythrocytes in vertebrates (other than mammals) accumulate histone H5, an erythrocyte-specific member of the histone H1 family that is involved in modifying higher order chromatin structure (12, 13). In other blood cells such as lymphocytes and polymorphonuclear granulocytes, the histone types and levels are not changed, suggesting that other factor(s) must participate in chromatin condensation in these types of cells.
The first identified and best characterized non-histone chromatin-condensing protein is MENT (myeloid and erythroid nuclear termination stage-specific protein). This abundant nuclear protein accumulates in terminally differentiating chicken blood cells, binds to heterochromatin, and promotes its condensation during terminal cell differentiation (14, 15). MENT belongs to a structurally diverse but functionally related group of proteins, including Sir3p, Tup1p, and MeCP2, that assemble tight higher order chromatin structures in repressed chromatin (3, 9, 16). The mechanism(s) of chromatin compaction by these proteins and the structural features that mediate chromatin remodeling remain unknown.
MENT is a member of the serpin (serine protease inhibitor) family and contains a conserved inhibitory serpin reactive center loop (RCL)1 (17). Previous studies of MENT expressed in NIH/3T3 cells have shown that the RCL affects MENT interactions with chromatin in vivo (18). MENT sequencing also reveals a putative chromatin-binding domain (the M-loop) that contains an AT-hook DNA-binding motif that is absent in other serpins (17).
The goal of this work was to understand how chromatin structure is modified by MENT and specifically to determine the individual roles of the M-loop and RCL of MENT in chromatin folding and self-association. Here, we show that these two domains fulfill distinct functions in chromatin condensation. The M-loop mediates side-by-side juxtaposition of naked DNA and is necessary for tight folding of chromatin via its interaction with linker DNA. The RCL is not required for DNA folding, but is necessary for chromatin self-association and MENT oligomerization. Our findings dissect, for the first time, chromatin condensation into two independent structural transitions caused by a single factor (MENT): compaction of individual chromatin fibers and formation of protein bridges that interconnect distinct nucleosome arrays.
|
EXPERIMENTAL PROCEDURES |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
-D-thiogalactopyranoside at a cell density of A600 = 0.6, and cells were harvested 2.5 h postinduction by centrifugation. Frozen cell pellets were lysed using BPER according to the manufacturer's instructions, with the exception of the addition of 10 mM sodium phosphate (pH 7.4), 0.5 M NaCl, 10 mM imidazole, and 25 µg/ml DNase I (Worthington) to the BPER reagent before cell lysis. Lysate was purified on HisTrap affinity columns (Amersham Biosciences) according to the manufacturer's instructions. MENT-containing fractions were further purified on HiTrap SP cation-exchange columns using an AKTA FPLC unit (Amersham Biosciences). MENT-containing fractions were dialyzed into 20 mM HEPES (pH 7.5) and 50 mM NaCl and were stored at +4 °C for use over a period of 30 days. Nucleosome Reconstitution—Nucleosome 12-mer arrays were reconstituted from DNA of the sea urchin 5 S rRNA gene containing 207x12 nucleosome positioning sequences (19) and core histones as described (20). Core histones and DNA were mixed on ice to a final concentration of 0.05 mg/ml histones, 1.9 A260 units of DNA, and 2 M NaCl and dialyzed in Slide-A-Lyzer units (Mr 3500 cutoff, Pierce) stepwise into reconstitution buffer (10 mM HEPES (pH 7.5), 0.2 mM EDTA, 0.1% Nonidet P-40) containing 2, 1.5, or 0.75 M NaCl for 3 h at 4 °C, followed by dialysis into reconstitution buffer containing 0.5 M NaCl for 12 h and finally into buffer containing 5 mM NaCl for 6 h.
Deoxynucleoprotein (DNP) Electrophoresis—Native agarose gel electrophoresis of protein-DNA complexes was performed as described (15). Briefly, MENT was mixed with native chicken nucleosome arrays, naked DNA (207x12 5 S nucleosome positioning sequence), or nucleosome array reconstitutes and incubated for 30 min on ice in 10-µl reactions containing 0.6 A260 units of chromatin, 13 mM HEPES (pH 7.5), 37 mM NaCl, 0.4 mM EDTA, 5% glycerol, and 0.4–2.5 µM MENT. Samples were centrifuged at 16,000 x g for 10 min at 4 °C. Supernatants were loaded directly onto 1% type IV agarose gels (Sigma) in 20 mM HEPES (pH 8.0) and 0.2 mM EDTA and resolved at 80 V for 90 min. Gels were stained in ethidium bromide following electrophoresis. Reaction precipitates were dissolved in 1% SDS and loaded onto 1% type I agarose gel containing TAE (0.04 M Tris-acetate, pH = 7.8; 0.02 M Na-acetate; 2 mM EDTA) and ethidium bromide. All gels were visualized and digitized using EagleEye software. Precipitates were quantified using OptiQuant software to determine the percentage of total chromatin or DNA precipitated in the reaction.
Electrophoretic Mobility Shift Assays (EMSAs)—The DNA probe used in EMSAs was an 88-bp PCR product of the second intron of the chicken A-globin gene, and the pCAG1 plasmid (21) was used as a template. The primers used were 5'-GTG AGA AAG GAA TGA AGG GGA TGG and 5'-GCC ACC TCC TCA CTG CCT TCC. The product was purified using a QIAquick PCR purification kit. DNA fragments were 5'-end-labeled with T4 kinase (Invitrogen) according to the manufacturer's instructions. Labeled DNA was purified using a Bio-Spin P-30 column (Bio-Rad), followed by ethanol precipitation.
For EMSA (22), 50-µl reactions containing 10 mM HEPES (pH 7.5), 20 mM NaCl, 0.4 mM EDTA, 5% glycerol, 200 µg/ml bovine serum albumin, <1 nM labeled DNA, and 0–50 nM MENT were incubated on ice for 30 min and then centrifuged 16,000 x g for 10 min at 4 °C. Supernatants were loaded directly onto 6% acrylamide gels (35:1 acrylamide/bisacrylamide) in 25 M HEPES (pH 7.5) and 0.125 mM EDTA that had been prerun for 30 min at 20 mA. Electrophoresis was performed at 20 mA/gel for 30 min. Gels were fixed in 10% acetic acid and 10% methanol for 30 min, dried, and exposed to a PhosphorImager (Amersham Biosciences). Densitometry was performed using OptiQuant software and by measuring the disappearance of unbound DNA probe.
Electron Microscopy—For standard transmission electron microscopy of naked DNA and protein-DNA complexes (A260 = 1.0), samples identical to those taken for DNP electrophoresis (see above) were fixed by the addition of 0.1% glutaraldehyde for at least 16 h at +2 °C. Fixed samples were applied to glow discharged thin carbon films and stained with ethanolic phosphotungstic acid (positive staining) or aqueous uranyl acetate (negative staining) as described (23). For negative staining, fixed samples on carbon films were washed with double-distilled water and then negatively stained with 2% uranyl acetate and air-dried. For positive staining, samples were rinsed in 0.3% PhotoFlo 100 in distilled water that had been brought to pH 9.0 with sodium borate buffer and air-dried. Grids were then stained by immersion for 30 s in 1% sodium phosphotungstate in 75% ethanol, rinsed briefly in 95% ethanol, dehydrated in 100% ethanol, and air-dried (23).
For electron microscopy of reconstituted nucleosome arrays contrasted by platinum shadowing, samples (A260 = 1–1.5) in 15 mM HEPES (pH 7.5), 25 mM NaCl, and 1 mM EDTA were fixed by the addition of 0.1% glutaraldehyde for 4 h at +2 °C, and the unreacted glutaraldehyde was removed by dialysis against 15 mM HEPES (pH 7.5), 25 mM NaCl, and 1 mM EDTA. Samples were brought to 75% glycerol and then sprayed onto freshly cleaved mica as described (24). The mica was dried under high vacuum in a turbo-pumped evaporator (BA080, Baltec Inc., Hudson, NH) and, while under vacuum, shadowed with platinum at an angle of 6° and then coated with a thin layer of carbon. To provide both an accurate value for the size of the particles and also a view of the complete shape, the sample was kept stationary for the first one-fourth of the platinum evaporation procedure and then rotated for the remaining three-fourths.
Samples were examined in a Tecnai 12 transmission electron microscope (FEI Co., Hillsboro, OR) operated at 100 kV, and images were recorded on a slow-scan CCD camera (TVIPS GmbH, Gauting, Germany). Most shadowed preparations were examined using tilted beam dark-field optics.
1-Ethyl-3-(3-dimethylaminopropyl)carbodiimide HCl (EDAC) Cross-linking—Native soluble chicken chromatin and linker histone-depleted chromatin samples were isolated as described (15). MENT was reassociated with the isolated chromatin in 10 mM HEPES (pH 7.5), 40 mM NaCl, 0.5 mM EDTA, 0.6 A260 units of chromatin, and 0.9 µM MENT for 30 min at room temperature. EDAC (Pierce) was added to a final concentration of 30 mM, and the samples were incubated 2 h. -Mercaptoethanol was added to a final concentration of 50 mM, and the samples were incubated for an additional 15 min. SDS-PAGE sample buffer was added, and the samples were boiled for 2 min. Samples were then separated on gradient SDS-10–18% acrylamide gels (25) and transferred to polyvinylidene difluoride membrane in 192 mM glycine and 25 mM Tris for 1 h at 100 V. Western blotting was performed with anti-MENT antibodies as described (15).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
R). The two RCL mutations, MENTov and MENTTR, result in the inability of MENT to inhibit target protease. All three mutations have been shown to affect MENT interactions with nuclear chromatin after transfection into cultured cells (18). We expressed the recombinant, N-terminally histidine-tagged wild-type and mutant MENT proteins in E. coli and isolated the proteins by histidine trap affinity and ion-exchange chromatography (18). Expressed proteins were shown by matrix-assisted laser desorption ionization time-of-flight to have the predicted molecular masses. Recombinant wild-type MENT behaved similarly to native MENT in DNP-agarose gel shift assays (15) and protease inhibition assays (18), indicating that neither the histidine tag nor the lack of eukaryotic post-translational modifications interfere with normal MENT function.
|
The affinity of MENT for short naked DNA (under conditions of a vast excess of the protein over DNA) was assessed by EMSA (22). We typically used an 88-bp DNA fragment isolated from a portion of the chicken A-globin gene, which has an intermediate level of bound MENT in situ as determined by chromatin immunoprecipitation with anti-MENT antibodies (11). Other sequences from the chicken genome with either lower or higher levels of MENT in situ did not differ in their affinity for MENT in vitro as determined by EMSA (data not shown), suggesting that MENT does not have a DNA sequence preference in vitro. 32P-End-labeled DNA was titrated with MENTWT, MENTMLoop–, or MENTTR. MENT-DNA complexes were resolved on 6% polyacrylamide gels (data not shown), and the fraction of DNA bound was determined by the disappearance of unbound DNA (Fig. 1B). The unbound DNA fraction (V) versus MENT concentration was plotted in Hill coordinates (Fig. 1, C–E) to reveal the degree of cooperativity in protein-DNA interactions (26, 27). Although MENTWT and both mutants have similar Kd values of 
2 nM (Fig. 1B), there was a striking difference in the binding cooperativity seen in M-loop-deleted MENT. Both MENTWT and MENTTR exhibited strong positive cooperativity (Hill coefficients of 1.32 and 1.79, respectively). MENTMLoop– exhibited strong negative cooperativity, with a Hill coefficient of 0.60. This indicates that the M-loop of MENT, but not its RCL, is necessary for cooperative binding of MENT to DNA.
Interactions of wild-type MENT and its mutants with longer DNA (2.5 kilobase pairs (kbp)) were examined by agarose gel shift assays (15) under conditions in which the range of MENT concentrations was close to that of DNA (1–7 molecules/100 bp). Under these stoichiometric binding conditions, the number of MENT molecules bound per DNA and the ability of MENT to fold DNA into ordered structures could be monitored. As with short DNA, both the wild-type and mutant MENT proteins bound DNA with similar affinity as determined by the disappearance of unbound DNA (Fig. 1G) and could accommodate up to 2 MENT molecules/100 bp of DNA without precipitation (Fig. 1F). Association of MENTWT and MENTov with long DNA resulted in the formation of distinct complexes with significantly lower mobility compared with free DNA upon DNP-agarose electrophoresis. These results are similar to those obtained with MeCP2, where the protein was shown to form slowly migrating complexes of condensed DNA in a similar assay (16). In our experiments, MENTov was significantly more efficient in the formation of the retarded complexes compared with MENTWT (Fig. 1G, compare lanes 4–6 and 11–14), apparently due to a stronger self-association of DNA complexes with MENTWT, which become insoluble. This is consistent with the significant difference in DNA self-association (as measured by DNA precipitation) between MENTWT and MENTov shown in Fig. 1F, where MENTWT was about twice as effective as MENTov in causing DNA self-association at >2 MENT molecules/100 bp of DNA.
MENTMLoop– induced only a small reduction in electrophoretic mobility and a smearing of the DNP bands resulting from a heterogeneous population of nucleoprotein complexes (Fig. 1G, lanes 15–19). This confirms our previous conclusion that loss of the M-loop greatly decreases the cooperativity of MENT interaction with DNA, resulting in binding without a large concomitant change in DNA conformation.
The M-loop Mediates DNA Condensation into "Tramline" Structures—The structural features of MENT-DNA complexes formed in our DNP experiments were visualized by transmission electron microscopy. Complexes of MENTWT and MENTov with 1.1 kbp of DNA (3 molecules/100 bp) appeared as thick structures (Fig. 2, E–G and K) that were never seen in samples lacking MENT (Fig. 2A). We attribute the thick complexes to MENT molecules bound to DNA.
|
At lower MENT/DNA ratios (1 molecule/100 bp), the thick complexes occupied only a segment of the DNA strand, whereas the rest of the strand maintained its initial configuration (Fig. 2, B–D). The coexistence of fully covered and fully free patches of DNA is in excellent agreement with the positive cooperativity of MENT-DNA interactions identified in EMSAs. When we compared the length of the thick complexes with the length of the starting DNA (1.1 kbp = 375 nm), we observed that the DNA molecules containing thick patches were shorter than the starting protein-free DNA. Remarkably, the double length of a thick segment plus the length of an open segment within one particle was close to the length of a single open DNA molecule. At the higher MENT/DNA ratios (3 molecules/100 bp) at which the MENTWT-DNA complexes produce prominent supershifted bands on agarose gels (Fig. 1G, lane 4), we observed fully saturated thick MENT-DNA structures (Fig. 2, E and F). The lengths of such structures measured approximately one-half of the free DNA length, consistent with the folding of DNA in half by MENT. We also saw even thicker complexes, obviously containing more than two strands of DNA and originating from extensive self-association of DNA strands (Fig. 2G). These complexes show that DNA folding by MENT is not limited to two closely juxtaposed duplexes.
We suggest that, within the thick segments, the DNA is folded back on itself by MENT, forming parallel side-by-side structures. Similar structures have been previously observed with complexes between DNA and the globular region of histone H5 or MeCP2 and are known as DNA tramlines (16, 28). Indeed, when we performed negative staining of MENT-DNA complexes with uranyl acetate, we observed closely juxtaposed DNA duplexes, confirming the formation of tramlines by MENT (Fig. 2, H–J).
We also used electron microscopy to examine the protein-DNA complexes formed by two mutant MENT variants. MENTov formed DNA tramlines that were nearly identical to the ones obtained with MENTWT (Fig. 2K); however, MENTov did not bridge multiple tramlines together, as seen with MENTWT. This is consistent with the lower tendency of MENTov to precipitate chromatin at high protein/DNA ratios (Fig. 1F). With MENTMLoop–, we observed only irregular protein-DNA complexes, strikingly different from those seen with MENTWT and the inactive RCL mutant (Fig. 2L). Thus, we conclude that the M-loop, but not the RCL domain, is necessary for cooperative binding of MENT to DNA and for folding DNA molecules together in an ordered fashion, resulting in the parallel tramline structures.
The M-loop Is Necessary for Folding of Linker DNA in Chromatin—MENT functions in a chromatin context in vivo. To investigate the role of the RCL and M-loop domains in chromatin remodeling by MENT, we reconstituted chicken core histones with 207x12 sea urchin 5 S DNA repeats to obtain "core arrays," defined nucleosome arrays with positioned histone cores (19, 20). Similar arrays have been widely used before and have became a standard object for studying chromatin conformational dynamics in vitro (see, for example, Refs. 3, 16, and 29). Our reconstitutes were tested by micrococcal nuclease/restriction mapping to show that the majority of the nucleosomes occupied the predicted positions on the DNA (data not shown). The resulting core arrays were mixed with a range of wild-type or mutant MENT concentrations, and the oligonucleosomes were analyzed by DNP electrophoresis (15). As with free DNA, we observed that association of MENTWT, MENTov, and MENTMLoop– with the core arrays caused formation of retarded complexes. All proteins were equally efficient in nucleosome array binding, but neither MENTWT nor its mutants gave rise to nucleosome array self-association up to a ratio of 4 MENT molecules/nucleosome (2 molecules/100 bp) (Fig. 3A).
|
The structural features of MENT-reconstituted complexes were analyzed by transmission electron microscopy after contrasting with platinum rotary shadowing. Fig. 3B (first row) shows that, under these experimental conditions, the reconstituted oligonucleosomes had a typical open "beads-on-a-string" structure with even spacing of nucleosomes on DNA. In contrast, at a ratio of 1 MENTWT molecule/100 bp of DNA, the reconstituted oligonucleosomes formed compact structures in which free linker DNA was rarely seen (Fig. 3B, second row). This indicates that MENT brings the nucleosome cores into close proximity, presumably as a result of its interactions with linker DNA. Reconstitutes with MENTov also produced compact structures similar to those with MENTWT (Fig. 3B, third row), which were slightly more condensed than the MENTWT complexes. This supports our model that a functional RCL is not necessary for proper linker DNA folding. In contrast, MENTMLoop– showed complexes with predominantly open linkers (Fig. 3B, fourth row). However, unlike the control oligonucleosomes (Fig. 3B, first row), the core arrays containing MENTMLoop– did not have a plain beads-on-a-string organization, but rather showed many nucleosomes in close proximity to each other (Fig. 3B, fourth row, arrows), connected by "open" linkers to other clusters. This suggests that, although losing the ability to interact with linkers, MENTMLoop– retains the potential for internucleosome bridging. Significantly, because the M-loop deletion retains a full enzymatic inhibitory activity in vitro (18), its altered DNA interaction is unlikely to be due to a general protein structural alteration.
Both the RCL of MENT and Linker Histones Are Necessary for Bridging Oligonucleosome Arrays—Native chromatin compacted by MENT contains 1 molecule of linker histone (H1 and/or H5)/nucleosome (17). Linker histones promote chromatin fiber folding (30) and self-association (12), and the action of MENT is likely to enhance chromatin condensation by linker histones. Upon association of MENTWT with 207x12 reconstitutes containing 1 molecule of linker histone H5/nucleosome ("linker arrays"), we observed a significantly higher rate of chromatin self-association than was seen with core arrays based on the decreased amount of visible material with increasing MENT concentrations (compare Fig. 4A, lanes 1–4; and Fig. 3A, lanes 1–4). Fig. 4B confirms that increasing MENT concentrations brought about progressive self-association and precipitation of linker arrays. In contrast, neither MENTov nor MENTMLoop– could cause any distinct self-association at the same protein/DNA ratio (Fig. 4B), and their complexes with linker arrays resulted in more prominent bands of freely migrating and electrophoretically retarded particles upon native DNP electrophoresis (Fig. 4A, lanes 5–12). Thus, the two MENT mutants were able to bind linker arrays and to cause their shift upon DNP electrophoresis, but were unable to precipitate them. These results show that binding alone is not sufficient to cause the nucleosome array self-association and that the RCL and M-loop of MENT are both necessary for maximum chromatin condensation.
|
Electron microscopy of linker arrays with MENTWT revealed abundant self-associated structures (Fig. 4C, middle row) that were much larger in size than the starting oligonucleosomes (upper row). These particles strongly correlated with the extensive self-association seen upon DNP electrophoresis (Fig. 4A). In contrast, similar arrays that re-associated with MENTov behaved as separate particles, showing no self-association (Fig. 4C, lower row). Thus, electron microscopy confirmed that interarray bridging (self-association) requires linker histone and an intact RCL domain of MENT. It should be noted that, at a similar MENT/chromatin ratio, electron microscopy showed a greater extent of linker array self-association by MENTWT (Fig. 4C) than predicted from the gel data (Fig. 4A). This result probably indicates partial dissociation of this material during electrophoresis.
To examine whether the RCL mutation has a comparable effect on native chromatin, we reconstituted the wild-type protein and the MENTov mutant with native chicken oligonucleosomes (trimers that contain linker histones H1 and H5). As in the experiments with reconstituted linker arrays (Fig. 4), when we associated MENT with native trimers, we observed a significantly higher rate of chromatin bridging and precipitation by MENTWT than by MENTov (Fig. 5). At a MENT/DNA ratio of 2 molecules/100 bp, the complex of trinucleosomes and MENTWT was almost completely precipitated, whereas MENTov showed significantly weaker precipitation (Fig. 5B) and a prominent band of freely migrating particles upon native DNP electrophoresis (Fig. 5A, lane 10). Thus, with both native and reconstituted nucleosome arrays, the MENTov mutant has a significantly impaired ability to cause large-scale chromatin self-association, presumably because the native RCL domain is essential for this function.
|
The RCL Is Required for DNA- and Chromatin-induced Oligomerization of MENT—For several other heterochromatin proteins such as the Sir-Rap complex (32) and HP1 (33), it has been suggested that protein oligomerization may be linked to heterochromatin assembly. We reasoned that, during the mediation of chromatin bridging or folding, MENT molecules may interact with one another and that, in the presence of ligand, it should be possible to "trap" such multimers. Therefore, to determine whether MENT does in fact oligomerize in the presence of DNA and chromatin, we used EDAC, a chemical crosslinking reagent that forms zero-length bridges between interacting proteins. This reagent has been successfully used in previous studies of linker histone interactions with core histones (34, 35) and other protein-protein interactions. When we treated MENT-DNA complexes similar to those shown in Figs. 2 and 3 with 30 mM EDAC, MENTWT dimers, tetramers, and larger oligomers were readily detected in anti-MENT Western blots (Fig. 6A). These oligomers were also quite prominent in the presence of native soluble chromatin (Fig. 6A, compare lanes 1 and 3), although the pattern was not as strong as with naked DNA (lane 2). Since patterns of oligomeric bands were obtained with both pure DNA and chromatin, the cross-linked species must be MENT oligomers rather than MENT-histone complexes. Removal of linker histones from the soluble chromatin before MENT reconstitution returned MENT oligomerization to the level achieved with naked DNA (Fig. 6A, lane 4). In contrast to MENTWT, MENTMLoop– self-oligomerized efficiently both in the presence and absence of naked DNA, but when bound to chromatin, was weaker in oligomerization than MENTWT (Fig. 6A, lanes 11 and 12 versus lanes 3 and 4). Importantly, we were unable to detect oligomerization of MENTov in the presence of any DNA or chromatin samples tested (Fig. 6A, lanes 5–8). We repeated the cross-linking with intact chicken white blood cell nuclei and found that MENT dimers could be trapped in situ (Fig. 6B), indicating that our observations are unlikely to be an artifact of in vitro reaction conditions.
|
Thus, it is possible to trap both MENTWT and MENTMLoop– oligomers using a zero-length cross-linker. However, MENTov cannot be trapped in a multimeric form in the presence of any of the ligands we tested. MENTov is not able to interact with target proteases (18), and the conformation of its RCL is expected to be quite different from that of an inhibitory serpin. The x-ray crystal structure of ovalbumin reveals that the RCL adopts an 
-helical conformation (47), whereas in other inhibitory serpins, the RCL normally adopts an extended or -like conformation (see, for example, Ref. 48). Taken together, our data reveal that the conformation and/or sequence of the RCL is important for promoting bridging between chromatin arrays, whereas intra-array folding does not involve the RCL, but does require the M-loop. The absence of the RCL-mediated oligomerization with MENTov suggests that it either mediates intrafiber compaction as a monomer or has a different conformation compared with MENTWT that is not readily trapped by the cross-linker.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
The primary structure of MENT and, in particular, the presence of an AT-hook motif in the M-loop (17) correlate well with the new results showing that the M-loop is a primary site mediating DNA binding and chromatin remodeling. AT-hooks are found in a number of architectural chromatin proteins such as high mobility group I(Y), ISWI, and SWI2/SNF2, where they mediate binding to AT-rich DNA and cooperate with other chromatin-binding proteins or protein complexes (36). However, our results also show that the M-loop in MENT is not the sole determinant of DNA binding. The Kd for MENT binding to naked DNA was not significantly affected by its mutation, although the M-loop deletion caused a strong negative cooperativity (Fig. 1) and compromised the close packing of DNA duplexes. This result indicates that there are other DNA-binding site(s) in MENT, but that the M-loop of MENT is responsible for ordered (although apparently not sequence-specific) binding of MENT to DNA. In its absence, the remaining positive charge of MENTMLoop– may cause a repulsion between the protein molecules bound to DNA and thus result in negative cooperativity.
Cooperative binding of chromatin-repressing proteins such as HP1 and polycomb group with nucleosome arrays has long been thought to promote heterochromatin spreading (reviewed in Ref. 37). With MENT, we observed for the first time that the DNA-binding cooperativity of a protein is directly linked (via the M-loop domain) to the compaction of DNA and chromatin. With chromatin, this phenomenon is best observed with the MENTov mutant and nucleosome core arrays (Fig. 3), where it is not obscured by chromatin self-association. Although the details of linker DNA organization in the compacted arrays were not seen with these imaging techniques, based on our observation of tramlines formed by MENT on naked DNA (Fig. 2), we suggest that the tight folding of a nucleosome array is caused by a similar close juxtaposition of linker DNA within a chromatin fiber (Fig. 7). The juxtaposition of linker DNA within the MENT-compacted chromatin also agrees with our previous cryo-electron microscopy studies (17).
|
MENT belongs to the serpin protein family and, as we reported previously, is a specific inhibitor of a lysosomal protease, cathepsin L; also, as in other serpins, its conserved RCL mediates protease inhibition (18). We also showed that, in MENT-expressing cells, both the M-loop and RCL are essential for the normal interaction of MENT with nuclear chromatin in situ (18). We have shown in this study that, in vitro, in a biochemically defined system, an active RCL domain is essential for the type of chromatin self-association that could lead to large-scale compaction. This result clearly rules out the possibility that an interaction between MENT and a protease is required for chromatin folding, although a cathepsin-like activity may still be involved in regulating chromatin condensation in vivo (18). The presence of a cathepsin-like activity in the nucleus is still not confirmed, and its potential interaction with MENT remains an open question.
Many serpins have been demonstrated to form inactive loop-sheet polymers via the formation of RCL-A-sheet linkages (31, 38–41). One explanation for our observations is that such linkages are responsible for the higher order multimers that we observed during bridging reactions in vitro and in vivo. However, this explanation, while attractive, does not explain all of our experimental data. The formation of loop-A-sheet polymers is irreversible and is associated with the loss of inhibitory activity against target proteases (40). However, MENT isolated from terminally differentiated avian erythrocytes is in the native, inhibitory conformation, suggesting that, in the case of MENT, the oligomerization process is reversible (18). Furthermore, our data reveal that MENTMLoop–, which has inhibitory activity (18), multimerized in vitro in the absence of any ligand (Fig. 6), whereas significant loop-A-sheet polymerization would be expected to result in the loss of inhibitory activity, which we did not see. Taken together, the current data suggest that the RCL mediates reversible higher order multimers of MENT that are distinct from typical loop-A-sheet polymers. We note that the crystal structure of the active serpin plasminogen activator inhibitor-1 reveals edge strand linkages between the RCL of one molecule and the A-sheet of the next, forming an infinite chain within the crystalline lattice (49). It was noted that, in this case, the loop-sheet interaction is reversible since active material was obtained from dissolved crystals. It is possible that reversible MENT oligomerization involves a similar mechanism. However, the precise molecular mechanism of MENT-MENT interactions during bridging and/or folding remains to be determined.
Based on our molecular modeling of MENT (17), we suggest that, when it binds to nucleosome linkers via its M-loop, the RCL, which resides on the opposite side of the molecule, remains exposed and available to form oligomers with other MENT molecules. Because MENT oligomerization is promoted by DNA and chromatin binding (Fig. 6), we suggest that MENT may simultaneously bind nucleosomal DNA and oligomerize through its RCL, thus connecting separate nucleosome arrays by protein bridges and causing chromatin condensation (Fig. 7).
Because MENT binding to naked DNA (Figs. 1 and 2) and nucleosome folding within a single core array (Fig. 3) do not require an active RCL, we conclude that MENT oligomerization is not essential for intra-array chromatin folding. Furthermore, electron microscopy revealed a stronger folding of nucleosome core arrays by a mutant with inactive RCL (MENTov) than by MENTWT (Fig. 3). This suggests that the folding and bridging activities of MENT not only are separable, but may compete with each other, such that MENT folds more efficiently when its bridging activity is inactivated.
Although MENT is expressed in certain cells at a level unmatched by any other non-histone protein (up to 2 molecules/nucleosome in avian granulocytes (15)), our work suggests that it does not act alone, but may be influenced by other proteins, especially histones and their modifications. Thus, when linker histone H5 is present in the linker arrays, the addition of MENT causes a significantly stronger self-association then when linker histones are absent (Fig. 5). Apparently, without linker histone, a greater number of MENT molecules bind and fold linker DNA. When the histone H1/H5 level is higher, fewer molecules of MENT contribute to intrafiber chromatin compaction, and more are available for chromatin bridging. Indeed, in granulocyte chromatin, where the MENT level is especially high, the linker histone level is not increased compared with proliferating cells with decondensed chromatin. In contrast, in erythrocyte chromatin, where the developmentally regulated histone H5 is highly expressed, the level of MENT is much lower (15), perhaps balancing the stronger linker-folding function performed by histone H5.
An important and universal histone modification regulating heterochromatin spreading is histone H3 methylation at lysine 9 (42). In chicken erythrocyte nuclei, this modification (45) as well as histone H5 (46) are preferentially associated with inactive heterochromatin. Recently, we found that dimethylation of histone H3 at lysine 9 co-localizes with MENT in heterochromatin and directly promotes chromatin condensation by MENT in an RCL-dependent manner (11). The N-terminal tail of histone H3 exits from the nucleosome core at the linker entry/exit site and, together with linker histone, is involved in chromatin compaction (43, 44). The functional parallels between the synergistic effects on MENT of the dimethylated histone H3 N-terminal domain (11) and linker histone (Fig. 4) suggest that both may promote chromatin condensation by contributing to intrafiber compaction and thus favor the interfiber bridging function of MENT.
Several other chromatin-bridging heterochromatin proteins such as Sir3p, Tup1p, and MeCP2 can both bridge nucleosome arrays and promote transcriptional silencing (3, 9, 16). In chicken erythrocytes, MENT is selectively associated with gene-poor heterochromatin, and its direct effect on transcription is thus unclear. However, ectopic expression of MENT in cultured cells causes large-scale chromatin remodeling and has a strong repressive effect on cell proliferation. Importantly, the RCL mutations that (as shown here) inhibit nucleosome array bridging (Fig. 4) also abrogate the proliferation repression and chromatin-remodeling properties of MENT (11, 18). Thus, interarray bridging rather than intraarray compaction is likely to be the key structural transition that is of primary importance in large-scale chromatin compaction leading to inhibition of chromatin transcription and replication. Regulation of specific genes targeted by MENT is currently under investigation.
|
FOOTNOTES |
|---|
** To whom correspondence should be addressed: Dept. of Biochemistry and Molecular Biology, Milton S. Hershey Medical Center, Pennsylvania State University College of Medicine, 500 University Dr., P. O. Box 850, Hershey, PA 17033. Tel.: 717-531-8588; Fax: 717-531-7072; E-mail: sag17{at}psu.edu.
1 The abbreviations used are: RCL, reactive center loop; FPLC, fast protein liquid chromatography; DNP, deoxynucleoprotein; EMSA, electrophoretic mobility shift assay; EDAC, 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide HCl; kbp, kilobase pairs.
|
ACKNOWLEDGMENTS |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  S. J. McBryant, C. Krause, C. L. Woodcock, and J. C. Hansen The Silent Information Regulator 3 Protein, SIR3p, Binds to Chromatin Fibers and Assembles a Hypercondensed Chromatin Architecture in the Presence of Sal |
) (Ov-Swap), and MENTMLoop–. (
). The MENT/DNA ratios (molecules/bp) were as indicated. G, DNP electrophoresis of DNA (2.5 kbp long) reconstituted with MENTWT (lanes 1–7), MENTov (lanes 8–14), and MENTMLoop– (lanes 15–19).
