A novel nonhistone protein (MENT) promotes nuclear collapse at the terminal stage of avian erythropoiesis

EXPERIMENTAL

CELL

RESEARCH

198,268-275

(1992)

A Novel Nonhistone Protein (MENT) Promotes Nuclear Collapse at the Terminal Stage of Avian Erythropoiesis S. A. GRIGORYEV,’ Department

V. 0. SOLOVIEVA, of Molecular

Biology,

K. S. SPIRIN, Moscow

State

INTRODUCTION

In addition to trans-acting factors eukaryotes have an alternative mechanism for gene repression that is the heterochromatization (chromatin condensation) of particular genome domains. Certain heterochromatic re-

0014~4827192 $3.00 Copyright Q 1992 by Academic Press, All rights of reproduction in any form

Moscow,

MATERIALS

U.S.S.R.

AND

METHODS

Cells, nucki, and microscopy. Erythrocytes were obtained from the blood of lo-day-old (EE-10) and 15-day-old (EE-15) embryos and from the blood of adult chickens (AE). An erythroblast cell line (HD3) transformed by a temperature-sensitive virus of avian erythroblastosie was obtained from Dr. T. Graf (Heidelberg, Germany) and 268

Inc. resewed

University,

gions (constitutive) are established early in embryogenesis and are inherited in cell lines in a position-dependent manner. Other chromatin regions become heterochromatic (facultative) in the course of cell differentiation [l, 21. One can distinguish three levels of heterochromatin propagation: the lower level is associated with repression of limited genomic regions extended for several dozen kbp (e.g., locus “white” of Drosophila [3] ); at the second level, heterochromatin is extended over a whole X-chromosome of female mammals [4]; and, finally, at the third level, the bulk of interphase chromatin in a given cell type, e.g., in the nucleated erythrocytes of lower vertebrates, becomes hypercondensed and transcriptionally inert [5-71. The mechanism and the driving force of progressive heterochromatization remain unknown. It is also unclear whether there are any differences in chromatin structure between euchromatin and heterochromatin fractions or whether these two chromatin portions differ only in the amount of free space between the chromatin fibers. We have isolated a nonhistone chromosomal protein (MENT, mature erythrocyte nuclear termination stage-specific protein) which is selectively bound to the repressed chromatin fraction. Hyperexpression of MENT coincides with complete metabolic inactivation and heterochromatization of mature chicken erythrocytes. When reconstituted with immature erythroeyte nuclei in vitro this protein induced the structural transitions (condensation) observed for the chromatin deposited in the nuclei but not for the soluble chromatin. Thus MENT is a probable candidate for a factor inducing the heterochromatization of chromatin at the terminal stage of erythroid differentiation. Its ability to dissociate polynucleosomes from the nuclear framework suggests that MENT may act by releasing the euchromatin domain boundaries, thus allowing chromatin condensation to spread from neighboring heterochromatic regions.

The terminal stage of differentiation of nucleated chicken erythrocytes is associated with an overall gene repression and a condensation of the repressed chromatin portion. Two-dimensional DNP electrophoresis has been used to separate transcriptionally active and repressed chromatin of mature chicken erythrocytes. The repressed chromatin fraction is shown to be enriched with histone HS as well as with a 42-kDa nonhistone chromosomal protein. The 42-kDa protein designated here as MENT (mature erythrocyte nuclear termination stage-specific protein) is hyperexpressed at the terminal stage of chicken erythropoiesis and is accumulated in adult chicken erythrocyte nuclei. This protein was purified by ion-exchange chromatography from 0.4 M NaCl extracts of the erythrocyte nuclei. It appeared to be a basic polypeptide (~19.2) which, however, precipitated at low PH. When reconstituted in uitro with immature erythrocyte nuclei, MENT promoted condensation of intact nuclear chromatin and enhanced the solubilization of nuclease-digested polynucleosomes, thus mimicking the processes occuring in uiuo at the final stage of erythrocyte maturation. The extent of dissociation of specific gene sequences from the nuclear matrix in MENT-treated nuclei is in striking correlation with their transcriptional activity. No other basic proteins (HS, cytochrome c, RNase A) added to the nuclear preparation at the same level as MENT (protein/ DNA = 0.005) caused any effect on nuclear organization. No alterations were observed when MENT was mixed with erythroblasts and nonerythroid nuclei having little or no histone H5. We propose that MENT cooperates with histone HS to complete the nuclear collapse in mature nucleated erythrocytes. Q1992 Academic PZ~SS. I~C.

‘To whom correspondence and reprint requests should be addressed at present address: Zoology Department, University of Massachusetts at Amherst, Amherst, MA 01993.

AND I. A. KRASHENINNIKOV

MENT,

A NONHISTONE

PROTEIN

FIG. 1. A-particle formation is associated with terminal erythroid differentiation. Ethidium-stained agarose gels after two-dimensional native DNP/DNP electrophoresis of soluble chromatin samples obtained from an erythroblast cell line (HD3), from lo-dayold (EE-10) and 15-day-old (EE-15) chicken embryo erythrocytes, and from erythrocytes of adult chicken (AE). Autoradiographs (two lower photos) were obtained after two-dimensional separation of AE chromatin and blot hybridization with probes Ov and 0.

grown as described [8]. For in uiuo labeling, the transformed cells and primary red blood cells were cultivated in leucine-deficient medium containing 1 &i/ml [“Clleu. Nuclei were obtained from each type of the erythroid cells as described and stored in 50% glycerol at -2O’C [lo]. For light microscopy the nuclear suspensions in RBS (3 mM MgCl,, 10 m&f NaCl, 10 m&f Tris-HCl, pH 7.6) were fixed with 2.5% glutaraldehyde and stained with methyl blue. For preparation of electron microscopy sections the nuclear samples were fixed and embedded in Araldite as described [ 111, but using RSB instead of cacodylate buffer during the nuclear fixation with glutaraldehyde and 0~0,. Chromutin isolafion. For chromatin isolation an aliquot of the nuclear preparation (A, = 20) was resuspended in 5 ml of RSB, containing 0.5 mM PMSF. CaCl, and micrococcal nuclease (Boehringer) were added to give final concentrations of 1 mM and 3 units/pi, respectively, and the reaction was carried out with stirring at +37”C. Aliquota were taken periodically and the DNA was analyzed by agarose gel electrophoresis in order to estimate the time interval needed to achieve the desired extent of digestion. To obtain a “standard” chromatin preparation (the mean number of nucleosomes in a chain, N = 4) the reaction was terminated after 15 min by adding 0.5 ml of 0.1 M Na-EDTA and 5 ml of ice-cold TE (10 mM Tris-HCl, 1 n&f EDTA, pH 7.6). The digested nuclei were pelleted for 10 min at 10,000 rpm (JA-20 rotor, Beckman), resuspended in 5 ml TE, and pelleted once more under the same conditions. The supematant (S2) and the residual nuclear pellet (NR) were collected and used for further experiments. DNA ond hybridization probes. DNA from nuclei and from agarose-derived samples was purified by proteinase K treatment and phenol extraction. DNA concentration in soluble and insoluble chromatin was estimated spectrophotometrically in the presence of 0.5%

PROMOTING

NUCLEAR

COLLAPSE

269

SDS (1 mg/ml DNA gives an A, = 20). The relative concentration of specific gene DNA was assayed by dot-hybridization [12] on Hybond N membranes (Amersham). The probe for the coding region of the @-globin gene (8) was a 1364-bp MspI fragment of plasmid pCA@Gl [13] obtained from Dr. G. Felsenfeld (Bethesda, MD), the probe for the coding region of the ovalbumin gene (Ov) was a 2.4-kbp EcoRI fragment [14] of plasmid pOv12 obtained from Dr. B. O’Malley (Houston, TX), the probe for the coding region of the histone H5 gene (H5) was a 2.4-kbp EcoRI-EarnHI fragment of plasmid pH5 (151 obtained from Dr. J. Engel (Evanston, IL), and the probe for the myc oncogene was a 1.5-kbp PstI fragment of chicken v-myc [17] inserted in pBR322 obtained from Dr. E. Zaborovsky (Moscow, USSR). Probes were labeled with [‘*P]dNTPs by nick translation. Autoradiographs (several exposures) were scanned with an Ultrascan XL densitometer (LKB). The optical density values were plotted against the known DNA concentration and the slopes of the linear regions of the curves were taken to calculate the relative gene concentrations. Ekctrophoretic techniques. DNP electrophoresis in the first direction was carried out as described [8,9] in 1% agarose type IV (Sigma). Before the second direction was run, the gel was dialyzed for 2 h against three changes of EA buffer (1.5 r&f Na-EDTA, 1 mMsodium acetate, pH 5.6). The electrophoresis in the second direction was carried out at 3 V/cm for 5 h at +4”C with constant buffer recirculation. Southern blotting onto Hybond N nylon filters, nick-translation, and hybridization were as described [15]. For further analysis the gel slices corresponding to “A, ” “B,” and “C” zones found on the DNP 2D gel (see Fig. 1) were cut out of the gel, twice frozen at -7O”C, mixed with 2 vol of 5~ SDS-electrophoresis sample buffer [16], and extensively shaken. Agarose remnants were pelleted using an Eppendorf tabletop centrifuge and the supernatants were taken for either DNA or protein electrophoreses. Protein electrophoresis in 7-21% gradient polyacrylamide gels was carried out as described by Laemmli [16]. Protein isolation and reconstitution with nuclei. For isolation of nonhistone proteins, nuclei of adult chicken erythrocytes were extracted with 0.4 M NaCl for 15 min at +4”C with extensive stirring. Nuclei were pelleted in a JA-20 rotor (Beckman) for 10 min at 20,000 rpm. Supernatant was dialyzed against 0.02 M Hepes-NaOH buffer, pH 7.6, and applied to a Mono S FPLC column (Pharmacia). Proteins were eluted with 0.02 M Hepes buffer, containing a gradient of O-l M NaCl. MENT protein eluted as a single peak at an NaCl concentration of 0.55 M. MENT-containing fractions were dialyzed against 0.02 M Hepes-NaOH, pH 7.6, and the protein concentration was estimated by Bradford’s method [ 181 calibrated by a quantitative amino acid analysis. The concentration of MENT in chromatin fractions was estimated by densitometry of Coomassie R-250-stained gels on an Ultrascan XL laser densitometer (LKB). The densitometry of chromatin proteins was carried out in parallel with that of an aliquot of purified MENT protein of known concentration. For reconstitution experiments, 1 ml of nuclear suspension in RSB (A, = 20) was mixed with 50 pl of a 0.1 mg/ml MENT solution to give a MENT/ DNA ratio of 0.005 (close to that found in AE chromatin) and agitated with a magnetic stirrer at +4”C for 12 h. Soluble and insoluble chromatin fractions were obtained from MENT-reconstituted nuclei as described above.

RESULTS

A-Particle Formation Is Associated with Chicken Erythrocyte Maturation transcriptionally As has been shown by Weintraub, inactive chromatin is retarded during electrophoresis in agarose because of its ability to form aggregates (A-particles) [9]. On the basis of this observation we have developed a native two-dimensional electrophoretic sys-

270

GRIGORYEV

ET

AL.

.MENT

1234 FIG. 2. MENT protein: differentiation stain), Laemmli electrophoresis of proteins soluble chromatin proteins of HD3 (4), EE-10 shows a purified MENT protein sample and obtained after electrophoresis of “C-labeled (14) nuclei.

5

6

789

10

stage specificity and selective association with repressed chromatin fraction. Lanes l-3 (silver obtained from A (l), C (2), and B (3) particles. Lanes 4-10 (Coomassie stain), electrophoresis of (5), AE (6), and of AE nuclei treated with 0.35 MNaCl (pellet-7, and supernatant-8). Lane 9 lane 10 contains a molecular weight marker-protein mixture. Lanes 11-14, autoradiographs were proteins of HD3 (11) and EE-10 (12) nuclei and of 0.35 M NaCl extracts of HD3 (13) and EE-10

tern which, in contrast to that of Weintraub, does not require DNA deproteinization [lo]. Using this procedure, electrophoresis in the first direction, causing the retardation of repressed chromatin, is followed with a low ionic strength electrophoresis in the second direction, during which A-particles depart from the main diagonal. Two-dimensional DNP electrophoresis of AE chromatin and blot-hybridization of the resulting agarose gel show that a transcriptionally active gene (p-globin) comigrates with the B-particles and a repressed gene (ovalbumin) comigrates with the A-particles (Fig. 1, /I and Ov). The C-particle zone contains the nucleosome cores not seen after hybridization. A-particles from adult erythrocyte nuclei are rather stable as compared with chromatin obtained from more metabolically active cells [9] which gives little of this type of aggregation under these experimental conditions. To gain more information about the generation of A-particles in the course of erythroid maturation, we obtained soluble chromatin from cells taken at different sequential stages of differentiation: (1) HD3 cells, erythroblasts transformed by avian erythroleukemia virus at the stage when globin is not yet expressed; (2) EE-10, immature globin-producing erythroid cells (predominantly late polychromatic erythrocytes [ 191) obtained from chicken embryos incubated for 10 days; (3) EE-15, mature definitive embryonic erythrocytes obtained from embryos incubated for 15 days; and (4) AE, metabolically silent erythrocytes obtained from the blood of adult chickens. Figure 1 shows that the ability of chromatin to yield A-particles is dependent on the degree of erythrocyte maturation and thus correlates with transcriptional inactivation. The erythroblast chromatin

does not give A-particles at all and very faint particles can be seen on the EE-10 DNP/DNP pattern. In the same experiment both EE-15 and AE chromatin samples gave clear A-particle electrophoretic patterns (Fig. 1). MENT, A Nonhistone Repressed Chromatin

Protein that Is Accumulated of Mature Erythrocytes

in

To identify the proteins associated with chromatin fractions shown on Fig. 1, we have extracted them from the agarose slices corresponding to the A-, B-, and Cparticles and subjected the eluates to SDS-polyacrylamide gel electrophoresis. Using this and other approaches, it has been observed previously [lo, 20,26,27] that repressed chicken erythrocyte chromatin is enriched with histone H5 and transcriptionally active chromatin is enriched with high mobility group proteins and with the ubiquitinated form of histone H2A. Here we show (Fig. 2, lane 1) that the A-particle fraction contains a previously unidentified 42-kDa polypeptide. This polypeptide is the only nonhistone protein that is confined to inactive chromatin. An electrophoresis of proteins from unfractionated chromatin (Fig. 2, lanes 4-6) shows that it is one of the most abundant nonhistone proteins in AE chromatin (lane 6). Densitometry of the lanes containing the chromatin proteins and the purified 42-kDa protein has shown that its protein/DNA weight ratio in AE chromatin was about 0.005, approximately 1 polypeptide/55 nucleosomes. However, it occurs at about an order of magnitude less in embryonal chromatin (lane 5) and is not detected in HD3 chromatin (lane 4) or in chromatin samples from other chicken tissues not shown here (liver, brain).

MENT,

A NONHISTONE

PROTEIN

PROMOTING

NUCLEAR

COLLAPSE

FIG. 3. Chromatin condensation in MENT-treated EE-10 nuclei. Left column, light microscopy of intact nuclei reconstituted with MENT. A blend of the two preparations is shown below. Magnification, 1000X. Right (magnification, 10,000X) of nuclear sections of intact EE-10 nuclei, of EE-10 nuclei reconstituted with MENT,

The newly synthesized proteins of the erythroid cells were labeled by incubation of the cells in a culture medium containing ‘*C amino acids. An electrophoretic analysis of labeled proteins shows that the newly synthe-

271

EE-10 nuclei and of EE-10 column, electron microscopy and of intact AE nuclei.

sized 42-kDa protein is not detected in the cytoplasm of any of the cell types studied nor in the nuclei of HD3 cells (Fig. 2, lane ll), but that it is present in the nuclei of EE-10 (lane 12) where it is associated with the low

272

GRIGORYEV 25 Number

r

of nuclei

ET

AL.

MENT Promotes Condensation Erythrocyte Nuclei

EE-10

of Isolated

Embryo

20 15. 10

EE-10

25

+ MENT

BLEND

To gain insight into the role of MENT in the structural organization of chicken erythrocyte chromatin we have reconstituted a purified protein preparation with nuclei isolated from immature erythroid cells. MENT was mixed with nuclear samples at a protein/DNA weight ratio of 0.005, close to that observed in adult erythrocyte chromatin. The resulting nuclear preparation was compared by light and electron microscopy techniques with nuclei incubated under the same conditions either without any protein added or after an addition of other proteins or salts as discussed below. Terminal differentiation of chicken erythrocytes is associated with a decrease in the nuclear diameter from 5 pm (erythroblasts) to 2.2 pm (adult chicken erythrocytes) [7]. Using light microscopy (Fig. 3, left column) we have observed that the isolated EE-10 nuclei were still distinctly larger than the AE nuclei, but, when reconstituted with MENT in vitro, they underwent an ob-

20 15 10 5 0 2.1

2.3

2.5

2.7 2.9 3.1 nuclear diameter

3.3 3.5 (pm)

3.7

3.9

FIG. 4. Histograms of nuclear diameter. The measurements of nuclear diameters (in two perpendicular directions) were obtained from the micrographs such as those shown in Fig. 3 (left column) but the samples were mounted on a slide containing a lo-pm-wide calibration grid.

salt-insoluble nuclear matrix. The 42-kDa protein is designated here as MENT. As shown in Fig. 2 (lanes 7, 8, and 14) MENT is readily eluted from erythrocyte nuclei and from chromatin by 0.35 M NaCl. We have purified the protein from the nuclear extract by ion-exchange FPLC chromatography (Fig. 2, lane 9). Among the nuclear nonhistone proteins, MENT was one of the most tightly binding to the cation-exchange resin Mono S. By chromatofocusing we have estimated its isoelectric point, PI, as 9.2, a value close to that for histone Hl (~19.3). Analysis of the amino acid content of MENT revealed that about 20% of the amino acids are basic while the acidic amino acids taken together with their amides (Asx + Glx) constitute not more than 18% of the total. The amino acid composition of MENT resembles that of high mobility group nonhistone proteins, containing both basic and acidic domains. Surprisingly, MENT is readily precipitated with 1% TCA and thus should be formally ascribed to the low mobility group of chromatin proteins.

EE-10

EE-10

l MENT

FIG. 5. A dot-hybridization analysis of soluble and insoluble chromatin of erythroid cells. Autoradiographs were obtained after dot-hybridization of DNA from the soluble (S2) and insoluble (NR) chromatin samples obtained from AR and HD3 nuclei (top) and from intact and MENT-treated EE-10 nuclei (bottom). Three aliquots of each sample, containing 30, 20, and 10 pg of DNA, were applied on nylon membranes and hybridized with p-globin, histone H5, v-myc, and ovalbumin gene probes.

MENT, b) Relative

A NONHISTONE

PROTEIN

DNA wntent

100 =

6.2

EE-IO

[X1NR

60

EE-10 + MENT

60

total

ovalbumin

betaglobin

FIG. 6. MENT promotes the from the low-salt matrix of EE-10 relative gene concentrations in total scanning autoradiographs like those the DNA concentration ratio in the

H5

wc

dissociation of c-myc oncogene nuclei. Histograms showing the S2 and NR fractions obtained by shown in Fig. 5 and corrected for chromatin samples.

vious shrinking. The diameters of the nuclei from each of the samples were measured in two perpendicular directions to show that the treatment with MENT reduces the nuclear diameter by 1 f 0.2 pm (Fig. 4). To rule out possible artifacts of the sample preparation and of microscope focusing, we have mixed both the intact and the MENT-treated nuclei. As seen in Figs. 3 and 4, both types of nuclei are present in the mixed sample, confirming that the size difference is not due to specimen preparation. Electron microscopy of the nuclear sections (Fig. 3, right column) shows that the nuclear compression is associated with an increase in electron density typical of AE nuclei which had undergone a complete heterochromatization [7, 111. In contrast to AE nuclei, the MENT-treated EE-10 nuclear preparation maintained the round shape characteristic of embryonic erythrocytes. It is known that mono- and divalent cations as well as pH values below 7.5 promote condensation of isolated chicken erythrocyte nuclei [30, 311. MENT is a basic protein and therefore one can expect that it acts as a nonspecific polycation. Therefore, we made several control experiments in which we reconstituted the loday erythrocyte nuclei with several other basic proteins

PROMOTING

NUCLEAR

COLLAPSE

273

small enough to migrate into the nuclei: cytochrome c (Sigma), histone H5 isolated as described [35] from the chicken erythrocyte nuclei, and ribonuclease A (Pharmacia). The addition of RNase also served as a control for an influence of nuclear RNA integrity on the nuclear shape. Neither of these proteins when added at the same protein/DNA ratio as MENT (0.005) or in a fivefold excess induced the transitions of the nuclei similar to those observed with MENT. We have also carried out MENT-reconstitution experiments with erythroblast nuclei and nuclei from nonerythroid chicken cells (liver, brain). Neither the nuclease-protection assay nor microscopy showed any signs of chromatin condensation. These results indicate that the chromatin structural transitions reported here are specific both for lo-day erythrocytes and for MENT protein. In addition to nuclear shrinking, the terminal stage of erythrocyte differentiation is accompanied by the appearance of A-particles (Fig. 1) which represent the aggregation-prone repressed chromatin portion [9, lo]. To study the influence of MENT on chromatin structure, MENT-treated nuclei were digested with micrococcal nuclease. The soluble and the insoluble chromatin fractions were obtained and compared with the standard chromatin preparations. Densitometry of chromatin proteins revealed that 45% of the input MENT protein was recovered in the S2 fraction. No changes in nuclease reaction rates were observed, but the MENT-treated chromatin showed a slower rate of linker trimming (data not shown) typical of nuclei from mature chicken erythrocytes [lo]. Since we initially identified MENT as the A-particlespecific protein we expected that it would promote Aparticle formation in lo-day nuclei. However, in several experiments (sedimentation, DNP-electrophoresis, nuclease cutting) we noticed no differences except for a slight retardation of mono- and oligonucleosomes caused by the direct association with MENT. No A-particles were observed in the 2D DNP electrophoretic pattern of chromatin from MENT-treated lo-day erythrocytes. MENT Detaches a Housekeeping Matrix

Gene from the Nuclear

The disappearance of the insoluble chromatin portion (nuclear matrix) is known to be the third major alteration in the chromatin organization of differentiating avian erythrocytes [21, 221. The nuclear matrix has been reported to be enriched with transcriptionally active genes [22,36,37]. In the course of our experiments with MENT-treated nuclei we have noticed that pretreatment with MENT increases the proportion of the soluble chromatin coming out of the nuclease-digested nuclei. Therefore, we have carried out hybridization experiments aimed at monitoring the association of spe-

274

GRIGORYEV

cific gene sequences with insoluble chromatin of intact and MENT-treated nuclei. The procedure for chromatin solubilization used here is similar to the low-salt nuclear matrix assay which has been used previously to show the preferential association of the coding sequences of the actively transcribed genes with the nuclear matrix [36,37]. To look for possible alterations in the matrix-binding properties of particular gene sequences, we have applied aliquots containing similar amounts of DNA from soluble and insoluble chromatin fractions of HD3 cells, EE-10, and AE to nylon membranes. The gene probes used for hybridization were the coding regions of a gene repressed in all cell types studied (ovalbumin [14] ), a housekeeping gene that is expressed in all proliferating cells including erythroblasts but switched off in the process of terminal erythroid differentiation (c-myc, [33] ), a gene expressed in erythrocytes but not in erythroblasts (fi-globin [13]), and a gene transcribed in all erythroid cells (histone H5 [15]). The most soluble chromatin fraction obtained during the chromatin isolation (Sl) comprises mononucleosoma1 and subnucleosomal DNA fragments and gives practically no hybridization signal. The results obtained after the hybridization of S2 and NR DNA are shown here. As can be seen from Fig. 5, the actively transcribed histone H5 and c-myc genes, but not the nonexpressed P-globin and ovalbumin genes, are preferentially associated with the insoluble nuclear material of HD3 nuclei and lose their matrix association in the nuclei of mature erythrocytes where the whole genome is repressed. In EE-10 nuclei, the P-globin gene which is actively transcribed is preferentially associated with the matrix as are the H5 and c-myc genes (Fig. 5, bottom). When we analyzed the DNA of MENT-treated EE-10 nuclei we observed that the /3-globin and the H5 genes still maintained the preferential matrix association of active genes with the matrix. On the contrary, the c-myc gene has been completely detached from the matrix (Fig. 5, bottom, and Fig. 6). No dissociation of c-myc was observed for MENT-treated HD3 nuclei (data not shown). It thus appears that MENT interfers only with those matrix-binding sites that are destined to be lost at the terminal stage of erythrocyte differentiation. We have also checked out the possibility that MENT, being a basic protein, dissociates the matrix nonspecifically by neutralizing the chromatin DNA. When we analyzed the soluble and insoluble chromatin from nuclei treated with 100 mM sodium chloride and with histone H5 (exogeneous histone/DNA ratio = 0.05), we observed no depletion of the low-salt nuclear matrix in the active and inactive genes. In fact, the nuclei treated with histone H5 showed a general increase in the insoluble chromatin portion. These results confirm our previous suggestion that MENT acts in a cell- and gene-specific manner, and its action is not mediated through DNA charge neutralization.

ET AL. DISCUSSION

During avian erythrocyte differentiation the apparently uniform chromatin of erythroblasts is divided into two structurally distinct chromatin portions, a smaller active chromatin fraction (B-particles), retaining the organization of total chromatin in the precursor cells, and a repressed chromatin portion that undergoes radical structural transitions. These transitions include accumulation of H5 and MENT, loss of other nonhistone proteins and of matrix association, gain of linker protection from the nuclease, and the ability to form A-particles [9, 10, 20-22, 26, 271. The structural divergence is accompanied by the spatial partition of the active and repressed genes [34]. The data reported here suggest that MENT is a probable regulator of chromatin inactivation. The fact that erythroid cells become competent for MENT-induced chromatin condensation only at the final stage of differentiation suggests that there must be at least one other factor stimulating chromatin condensation in mature erythrocytes. The most probable one is histone H5, which accumulates in erythrocyte nuclei throughout erythroid differentiation [25] and which binds preferentially to the repressed chromatin portion [lo, 26, 271. It has also been shown that the nucleation centers for cooperative binding of H5 to DNA are within the matrix-associating DNA sequences [28]. Taken together with our results, these findings suggest that MENT may cooperate with histone H5 to induce chromatin condensation by detaching the matrix-binding DNA sites from the matrix itself. The boundaries of previously decondensed euchromatin domains could thus be opened for spreading condensation to the chromatin regions that had accumulated a sufficient amount of linker histones. It has been reported previously that the condensation of avian erythrocyte nuclei proceeds through a spatial rearrangement of chromatin (more tight packing) rather than through transitions of higher-order chromatin folding [7]. In experiments not shown here we have observed that MENT did not induce any conformational transitions of either total soluble chromatin or chromatin associated with specific genes (myc, H5, and fi-globin). This suggests that protein acts at the nuclear level, presumably by destroying specific sites in the nuclear skeleton and allowing a closer intrafiber packing, rather than directly modifying the structure (higherorder folding) of the chromatin fibers themselves. The proposed function of MENT in facilitating the transition of euchromatin into facultative heterochromatin has some analogies with that assigned to a class of heterochromatin-specific proteins, including those encoded by Su(var) genes and the satellite DNA-binding protein Dl of Drosophila [2, 31. Dl resembles MENT not only in its amino acid composition but also in the association with a specific chromatin fraction

MENT,

A NONHISTONE

PROTEIN

protected from terminal nuclease trimming [ 291. Therefore, it can be proposed that MENT is related to this class of proteins. In contrast to the latter, which are limited in their action, MENT is hyperexpressed at the terminal stage of red blood cell differentiation, and thus may cause the general repression of mature erythrocyte chromatin. The gene-specific effect of MENT on chromatin solubility is perhaps the most intriguing result of the present work. The actual matrix-attachment sites are believed to represent the association of transcriptional complexes with the nuclear matrix [37]. Whether the removal of a gene from the matrix is a prerequisite for or a consequence of its repression is yet unknown, but the possibility that MENT is a general repressor of the genes that are committed to inactivation in erythrocytes should be considered. Very recently, an unknown protein specific for mature erythrocytes was reported to act as a nonspecific transcription repressor [32]. As we have shown here, MENT can be purified in preparative amounts from mature erythrocytes. This makes the protein a convenient tool in studies of the 3D organization of interphase nuclei and of DNA-protein interactions by conventional biochemical and cytological methods. Other approaches, especially gene cloning and expression experiments, are needed to clarify the role of MENT protein as a putative nuclear repressor and its possible application as an antagonist of the proliferation of oncogenically transformed erythroid cells. We thank Dr. K. Frese (Moscow, USSR) who helped us with cell cultures, Dr. E. Kiseleva (Novosobirsk, USSR) for help with electron microscopy, and Dr. C. L. Woodcock (Amherst, USA) for critical reading of the manuscript. We are also grateful to Dr. T. Graf (Heidelberg, Germany) for HD3 tissue culture and Dr. G. Felsenfeld (Bethesda, USA), Dr. B. O’Malley (Houston, USA), Dr. J. D. Engel (Evanston, USA), and Dr. E. Zabarovsky (Moscow, USSR) for plasmid samples.

REFERENCES

PROMOTING

NUCLEAR

COLLAPSE

275

8. Beug, H., PaImieri, S., Freudenstein, C., Zentgraf, H., and Graf, T. (1982) CeU 28,997-919. H. (1984) Cell 38,17-27. Grigoryev, S. A., Spirin, K. S., and Krasheninnikov, I. A. (1999) Nucleic Acids Res. 18.7397-7406. Zentgraf, H., Deumhng, B., and Franke, W. W. (1969) Exp. Cell Res. S&333-337. Meinkoth, J., and Wahl, L. (1984) Anal. Bbchem. 138,267-284. Wood, W. I., Nickel, J., and Felsenfeld, G. (1981) J. Bill. Chem.

9. Weintraub, 10. 11. 12. 13.

267,7730-7736. 14. Lawson, G. M., Tsai, M. J., and O’Malley, B. W. (1989) Biochemistry 19,4403-4411. 15. Trainor, C. D., Stamler, S. J., and Engel, J. D. (1987) Nature

328,827-830. 16. Laemmli, U. K. (1970) Nature 227,680-685. 17. Venstrom, B., Moscovici, C., Goodman, H. M., and Bishop, J. M. (1981) J. Viral. 39, 625-631. 18. Bradford, M. M. (1976) Anal. B&&em. 72,248-254. 19. Lukas, A. M., and Jamroz, C. (1961) Atlas of Avian Hematology, Agriculture Monograph 25, U.S. Department of Agriculture, Washington, D.C. 20. Ridsdale, J. A., Rattner, J. B., and Davie, J. R. (1988) Nucleic Acids Res. 16.5915-5926. 21. LaFond, R. E., and Woodcock, C. L. F. (1983) Ezp. Cell Res.

147,31-39. 22. Razin, S. V., Yarovaya, 0. V., and Georgiev, G. P. (1983) Nucleic Acids Res. 13.7427-7444.

23. Farache, G., Razin, S. V., Rzeszowskaya-Wolny,

J., Moreau, J., Recillas Targa, F., and Scherrer, K. (1990) Mol. Cell. Biol. 10,

5349-5358. 24. Zenk, D. W., Ginder, D. W., and Brotherton,

T. W. (1990) Biochemistry 29,5221-5226. 25. Weintraub, H. (1978) Nucleic Acids Res. 5,1179-1188. 26. Kamakaka, R. T., and Thomas, J. 0. (1990) EMBOJ. 9,3997-

4006. 27. Postnikov, J. V., Shick, V. V., Belyavsky, A. V., Khrapko, K. R., Brodohn, K. L., Nikolskaya, T. A., and Mirzabekov, A. D. (1991) Nucleic Acids Res. 19,717-725. 28. Izaurralde, E., Kas, E., and Laemmli, U. K. (1989) J. Mol. Biol.

210,573-585. 29. Levinger, L. (1985) J. Biol. Chem. 260, 14,311-14,318. 30. Leake, R. E., Trench, M. E., and Barry, J. M. (1972) Exp. Cell Res. 71.17-26. 31. Coutelle, C., Belkner, J., Prehn, S., Rosenthal, S., David, H., and Dreher, R. (1974) Ezcp. Cell Res. 88,15-23.

1. John, B. (1988) In Heterochromatin: Molecular and Structural Aspects (Verma, R. S., Ed.), p. l-47, Cambridge, Univ. Press, New York. 2. Alberta, B., and Sternglanz, R. (1990) Nature 344,193-194. 3. Reuter, G. (1987) Mol. Gen. &ret. 210,429-437. 4. Gartler, S. M., and Riggs, A. D. (1983) Annu. Rev. &wet. 17, 155-190. 5. Lemez, L. (1964) Adv. Morphag. 3.197-245. 6. Ringertz, N. R., and Bolund, L. (1974) in The Cell Nucleus (Bush, H., Ed.) Vol. 3, p. 417-446, Academic Press, New York. 7. Dardick, I., and Setterfield, G. (1976) Enp. CeU Res. 100, 159171. Received April 7,199l Revised version received September 23, 1991

32. Walmsley, M. E., Buckle, R. S., Allan, J., and Patient, R. K. (1991) J. Cell Biol. 114,9-19. 33. Kirsch, I. R., Bertness, V., Silver, J., and Hollis, G. F. (1986) J. Cell. Biochem. 32,11-21.

34. Hutchison, N., and Weintraub, H. (1985) Cell 43,471-482. 35. von Holt, C., Brandt, W. F., Greyling, H. J., Lindsey, G. G., Retief, J. D., Rodrigues, J., de A. Schwager, S., and Sewell, B. T. (1989) in Methods in Enzymology (Wassarman, P. M., and Kornberg, R. D., Eds.), Vol. 170, pp. 431-523, Academic Press, San Diego. 36. Stratling, W. H. (1987) Biochemistry 26,495-502. 37. Delcuve, G. P., and Davie, J. R. (1989) Biochem. J. 263, 179186.