Inhibitory Activity of a Heterochromatin-associated Serpin (MENT) against Papain-like Cysteine Proteinases Affects Chromatin Structure and Blocks Cell Proliferation*

MENT (Myeloid andErythroid Nuclear Termination stage-specific protein) is a developmentally regulated chromosomal serpin that condenses chromatin in terminally differentiated avian blood cells. We show that MENT is an effective inhibitor of the papain-like cysteine proteinases cathepsins L and V. In addition, ectopic expression of MENT in mammalian cells is apparently sufficient to inhibit a nuclear papain-like cysteine proteinase and prevent degradation of the retinoblastoma protein, a major regulator of cell proliferation. MENT also accumulates in the nucleus, causes a strong block in proliferation, and promotes condensation of chromatin. Variants of MENT with mutations or deletions within the M-loop, which contains a nuclear localization signal and an AT-hook motif, reveal that this region mediates nuclear transport and morphological changes associated with chromatin condensation. Non-inhibitory mutants of MENT were constructed to determine whether its inhibitory activity has a role in blocking proliferation. These mutations changed the mode of association with chromatin and relieved the block in proliferation, without preventing transport to the nucleus. We conclude that the repressive effect of MENT on chromatin is mediated by its direct interaction with a nuclear protein that has a papain-like cysteine proteinase active site.


MENT (Myeloid and Erythroid Nuclear Termination stage-specific protein) is a developmentally regulated chromosomal serpin that condenses chromatin in terminally differentiated avian blood cells. We show that MENT is an effective inhibitor of the papain-like cysteine proteinases cathepsins L and V. In addition, ectopic expression of MENT in mammalian cells is apparently sufficient to inhibit a nuclear papain-like cysteine proteinase and prevent degradation of the retinoblastoma protein, a major regulator of cell proliferation.
MENT also accumulates in the nucleus, causes a strong block in proliferation, and promotes condensation of chromatin. Variants of MENT with mutations or deletions within the M-loop, which contains a nuclear localization signal and an AT-hook motif, reveal that this region mediates nuclear transport and morphological changes associated with chromatin condensation. Noninhibitory mutants of MENT were constructed to determine whether its inhibitory activity has a role in blocking proliferation. These mutations changed the mode of association with chromatin and relieved the block in proliferation, without preventing transport to the nucleus. We conclude that the repressive effect of MENT on chromatin is mediated by its direct interaction with a nuclear protein that has a papain-like cysteine proteinase active site.
MENT (Myeloid and Erythroid Nuclear Termination stagespecific protein), a developmentally regulated nuclear protein, is present in three main avian blood cell types (erythrocytes, lymphocytes, and granulocytes) where it is the predominant non-histone protein associated with compact heterochromatin (1). In vitro, MENT brings about a dramatic remodeling and condensation of chromatin higher order structure by forming protein "bridges" connecting separate nucleosomes in nucleosome arrays (for review see Ref. 2). MENT has no homology with other known chromatin proteins but belongs to the intracellular branch of the serpin superfamily (3). Serpins were originally characterized as serine proteinase inhibitors; however, more recently certain members have been shown to be capable of inhibiting other proteinase classes such as caspases (the viral serpin crmA (4)) and papain-like cysteine proteinases (SCCA-1 (5)). The inhibitory members of the serpin family are notable for their ability to undergo a large scale conformational transition (for review see Ref. 6) that is critical for inhibition of target proteinases (7,8) and for self-association or polymerization (9 -11). Multiple sequence alignments and comparison with known serpin structures reveal that MENT contains a large insertion, the "M-loop," between the C-and D-helices. This loop contains two critical functional motifs as follows: a classical nuclear localization signal (NLS) 1 that is required for nuclear import, and an AT-hook motif that is involved in chromatin and DNA binding (3). Like other serpins, MENT possesses a reactive center loop (RCL) 2 through which interaction with a cognate proteinase occurs. The presence of an inhibitory hinge motif (12) in its RCL suggests that MENT may retain the propensity for serpin-like conformational transitions and hence proteinase inhibition.
An important question is whether MENT is indeed a proteinase inhibitor, and if so, whether this activity plays any role in the interaction between MENT and chromatin or fulfils an unrelated second function of the protein. We show that MENT is an effective inhibitor of the papain-like cysteine proteinase cathepsins K, L, and V, as well as a cysteine proteinase from CV-1 cells with characteristics similar to the recently described nuclear proteinase, SPase (13). Thus MENT is the first known chromatin-associated cysteine proteinase inhibitor. We demon-strate that MENT imposes a strong repressive effect on cell proliferation when expressed in CV-1 and NIH/3T3 cells. Furthermore, by using a number of engineered variants (Fig. 1), we have found that the inhibitory activity of MENT, rather than DNA binding per se, is crucial for blocking cell proliferation. We thus show for the first time that a cysteine proteinase-inhibitor interaction can regulate cellular proliferation and modify chromatin structure.

EXPERIMENTAL PROCEDURES
Materials-Unless otherwise noted, all fine chemicals were from Sigma. Sepharose resins and HiTrap TM columns were from Amersham Biosciences AB. Human cathepsins K, L, and V were expressed in Pichia pastoris and purified as described previously (14,15). Papain was purchased from ICN (New South Wales, Australia). Sheep cathepsin L was purified from sheep liver as described previously (16). Active enzyme concentration was determined by titration with the inhibitor E-64.
For each mutant, the first PCR amplification step was conducted between either M49 and one of the reverse primers or between M36 and one of the forward primers. Following this, the PCR products from the two reactions were mixed and the internal primers removed, and the PCR was continued using the flanking primers M49 and M36. PCR amplifications were carried out using Taq DNA polymerase (Invitrogen). Reaction cycles were carried out in a 30-l volume using highmelting wax for "hot start" conditions, with 25 cycles each containing three 1-min steps at 94, 55, and 72°C and finally 10 min at 72°C. The PCR products were inserted between the HindIII and BstBI restriction sites in pSG109, replacing the MENT WT ORF. MENT OV was amplified using the reverse primer M36 and a pair of "nested" oligonucleotide primers: Mov2, GCTGGAGTGGATGCTGCTAGCGTCTCTGAAAAAT-TTAAGGTTGATCACC, and Mov1, AAAGGTCGCGAGGTGGTAGGG-TCAGCAGAGGCTGGAGTGGATGCTGCT. The nested PCR product was used to replace MENT P14R by cutting and ligating at the NruI and BstBI restriction sites. To verify the DNA sequences, all plasmids were sequenced on both strands at the Hershey Medical Center sequencing facility. The resulting mutant ORFs were then excised by digestion with XhoI and EcoRI and ligated into XhoI/BamHI-cut pET-15b vector (Novagen) using EcoRI/BamHI linker DNA (GeneWorks, Australia). The location of the mutations is shown in Fig. 1.
Expression and Purification of MENT and Mutants-The pET-15b constructs were used to express MENT and its variants in the BL21 Escherichia coli strain according to the vendor's protocol (18), with the exception that expression was induced at an A 600 of 1-1.1. Harvested cells were resuspended in lysis buffer (50 mM NaPO 4 , pH 8.3, 2 M NaCl, 1% (v/v) Triton X-100) and disrupted using a final concentration of 1 mg/ml lysozyme and sonication. Following centrifugation, the soluble fraction was applied to nickel-charged chelating Sepharose resin and washed with 10 column volumes of lysis buffer. A buffer containing 50 mM NaPO 4 , pH 8.0, 0.5 M NaCl, 25 mM imidazole was applied to the resin until the A 280 returned to base line, and protein was eluted with 50 mM HEPES, 50 mM NaCl, 250 mM imidazole, pH 7.4. The eluate was diluted 2-fold into buffer A (50 mM HEPES, pH 7.4, 50 mM NaCl) and loaded onto a 5-ml HiTrap TM SP column. After washing in the same buffer, a salt gradient was applied over 12 column volumes using 50 mM HEPES, pH 7.4, 1 M NaCl; intact MENT eluted in the middle of the gradient. These fractions were diluted 1:10 into 25 mM HEPES, pH 7.8, 1.7 M (NH 4 ) 2 SO 4 , loaded onto a 5-ml HiTrap TM phenyl column, and eluted using a gradient over 10 column volumes into 25 mM HEPES, pH 7.8. The (single band) purity of the proteins was verified using SDS-PAGE and their monomeric nature by using acid-PAGE (method appears below). The concentration of the protein was determined using an A 280 measurement with reference to the predicted extinction coefficient (19). After dialysis into buffer A and concentration to ϳ1 mg/ml, proteins were stored at Ϫ80°C until use.
Formation of Peptide-complexed MENT-A 20 M concentration of MENT WT was incubated with a 2 mM solution of a peptide corresponding to the P 14 -P 9 residues of the antithrombin RCL (sequence SEAAAS, Auspep, Parkville, Australia) in buffer A at 37°C for 16 h. A sample of MENT WT incubated in the absence of peptide did not lose activity under these conditions. Native gel electrophoresis confirmed the formation of complex with the peptide and, in conjunction with gel filtration, that peptide-complexed MENT did not polymerize. The gel system consisted of a 10% running gel (no stacking gel) with 260 mM Tris, pH 7.8, running gel, and 1ϫ TBE tank buffer, pH 8.0, with migration of the protein toward the cathode.
Determination of Kinetic Parameters-Enzyme assays were undertaken in pH 5.5 buffer (0.1 M acetate, pH 5.5, 1 mM EDTA, 0.1% (w/v) Brij-35, 10 mM cysteine) or an AMT buffer at pH 7.0 (0.05 M acetate, 0.05 M MES, 0.1 M Tris, pH 7.0, 1 mM EDTA, 0.1% (w/v) Brij-35, 10 mM cysteine). The stability of cathepsin proteinases at neutral pH is often regarded as being low, although more recent studies (20) have indicated that the proteinases are more stable than previously thought in some buffer systems. Under the assay conditions used here, the cathepsins were found to be stable over the entire period of the experiment.
The stoichiometry of inhibition (SI) was determined in pH 5.5 buffer. Briefly, 10 l of 20 nM enzyme was mixed with varied concentrations of inhibitor, incubated for 45-100 min at 30°C, diluted to 200 l in buffer containing a final concentration of 10 M N-Cbz-Phe-Arg-methylcoumarin substrate, and the residual enzyme activity measured using a Fluostar Galaxy plate reader (BMG Labtechnologies, Germany) with excitation/emission wavelengths of 370/460 nm, at 30°C. Data analysis was carried out as described previously (21).
The second-order association rate constant (k a ) was determined under pseudo first-order conditions using continuous kinetics. Within a given assay, the enzyme concentration was held constant, at a value between 0.025 and 0.5 nM, and the inhibitor concentration was always at least 20-fold higher, varying between 0.5 and 100 nM. All assays were conducted at 30°C, using a 20 M concentration of N-Cbz-Phe-Argmethylcoumarin substrate in a total volume of 200 l. Upon addition of enzyme to the inhibitor/substrate mixture, cleaved substrate fluorescence was monitored continuously using excitation/emission wavelengths of 370/460 nm. These fluorescence data were satisfactorily fitted, using least squares regression in PRISM (GraphPad Software), to Equation 1 describing slow, tight inhibitor binding (22), where F is the absolute fluorescence value at time t; V 0 is the velocity at time 0; V s is the residual velocity once the reaction has run to completion; and k obs is the apparent second-order rate constant. The uncorrected second-order association rate constant (k unc ) was obtained from the slope of the linear regression through a plot of k obs against inhibitor concentration. The substrate-corrected rate constant was finally obtained as shown in Equation 2, K m values for human cathepsin L and human cathepsin V were determined in this study for the hydrolysis of the fluorogenic substrate and were found to be essentially identical in both buffer systems. The K m values of 7.5 M for human cathepsin K and 9.8 M for sheep cathepsin L were obtained from previous studies (23,24). Preliminary experiments showed that proteinase activity did not drop below 95% over the time course of the assays. Assessment of Complex Stability-Complex stability was assessed by incubating human cathepsin V with a 2-fold molar excess of MENT WT in pH 5.5 buffer at room temperature. Samples were removed at 0, 12, 24, 36, and 48 h, and activity was measured using the same conditions as when determining the SI (see above). An inhibitor-free control was also used and remained Ͼ80% active throughout the course of the assay.
Mass Spectrometry-Molecular weight determinations of purified MENT WT and its products following digestion with cathepsin L were conducted using a linear MALDI-TOF mass spectrometer (Perseptive Biosystems Voyager) at the Pennsylvania State University College of Medicine laboratory core facility. Molecular masses were found to be within 0.15%, using external and internal calibrants. The correspondence between mass determinations and the potential proteolytic products was analyzed using Protein Prospector 3.2.1 software.
Native Protein Gel Electrophoresis-Complexes between MENT and cathepsins L or V were visualized using a modified version of a previously described native acid gel protocol (25), in which the stacking gel, running gel, and tank buffers were 90 mM acetate, pH 6; 260 mM acetate, pH 5; and 40 mM ␤-alanine, pH 5, respectively. Reaction mixtures with a final volume of 20 l containing 4 M MENT and 0 -4 and 8 M cathepsin V or 4 M human cathepsin L in pH 5.5 buffer were incubated at room temperature for 5 min prior to electrophoresis. Protein migration was toward the cathode. Proteins were visualized by staining with Coomassie Brilliant Blue R-250 (ICN, Australia).
SDS-PAGE-4 M MENT (20 l volume) was incubated with 2 M human cathepsin L or cathepsin V in 20 l of pH 5.5 buffer for 5 min at room temperature and electrophoresed on a 12% SDS-PAGE gel (26).
Cells and Transfection-CV-1 cells (ATCC CCL-70, derived from African green monkey kidney) were propagated in Eagle's BME medium (Invitrogen) with 0.1 mM non-essential amino acids, 1.5 g/liter sodium bicarbonate, 1 mM sodium pyruvate, 10% fetal calf serum, and 1ϫ antibiotic/antimycotic solution. NIH/3T3 cells (ATCC CRL-658 derived from NIH Swiss mouse embryo) were propagated in Dulbecco's modified Eagle's medium with 4.5 g/liter glucose (Mediatech Inc.), 1.5 g/liter sodium bicarbonate, 1ϫ antibiotic/antimycotic solution, and 10% (v/v) fetal calf serum. Cells were grown in a 5% CO 2 incubator at 37.5°C. Cells were transfected at about 80% confluency on 30-mm dishes using 12 l of LipofectAMINE-2000 reagent (Invitrogen) for 5 h in serum-free growth media, as described in the vendor's manual. After transfection, cells were allowed to continue growing in their normal media for 48 h.
Immunoprecipitation and Western Blotting-Rabbit polyclonal anti-MENT IgG was obtained and purified as described (27). Chicken polyclonal anti-sheep cathepsin L antibodies were obtained as described (28). For the immunoprecipitation reaction, 1.25 g of MENT was mixed with 1.7 g of sheep cathepsin L in 80 ml of buffer containing 50 mM NaCl, 50 mM HEPES, 4 mM dithiothreitol, pH 6.5, and incubated for 30 min at 37°C. The reaction mixtures were then incubated with 5 l of rabbit anti-MENT IgG at room temperature for 1 h. The reaction mixtures were rocked in a 1:1 suspension with 20 l of protein A-Sepharose, washed 3 times in reaction buffer, and finally resuspended by boiling in 50 l of Laemmli sample buffer. PAGE of proteins, Western blotting, and detection of MENT with anti-MENT antibodies and peroxidase-conjugated anti-rabbit IgG (Amersham Biosciences RPN 2108) were conducted as described previously (1). For detection of sheep cathepsin L, Western blots were first treated with chicken anti-cathepsin L antibodies (1:1000) and then with peroxidase-conjugated antichicken IgG (1:20,000). For detection of Rb protein, Western blots were first treated with mouse anti-human Rb antibodies (BD PharMingen; concentration 0.001 mg/ml in PBS containing 3.5% (w/v) BSA) and then with peroxidase-conjugated anti-mouse IgG (Amersham Biosciences, dilution 1:2500).
Immunofluorescent Microscopy and Image Analysis-For fluorescent microscopy, CV-1 and NIH/3T3 cells expressing MENT were grown in 30-mm culture dishes on microscopy coverslips (thickness number 1, Fisher). Cells attached to the cover glasses were fixed with methanol/ acetone (1:1 for 25 min at Ϫ20°C), washed twice with PBS for 5 min, and blocked in PBS containing 3.5% (w/v) BSA for 5 min at room temperature. The specimens were overlaid with PBS containing 3.5% (w/v) BSA and anti-MENT IgG (dilution 1:400) and incubated for 45 min at room temperature. The washing and blocking was repeated, and the specimens were overlaid with PBS containing 3.5% (w/v) BSA and fluorescent secondary antibodies, Alexa Fluor 594 goat anti-rabbit IgG (Molecular Probes), and incubated for 45 min at room temperature. The cells were then washed with PBS for 5 min, stained with 0.1 g/ml Hoechst 33258 in PBS for 20 min at room temperature, and mounted in PBS containing 50% (v/v) glycerol and 0.5% (v/v) n-propyl gallate as an anti-bleach reagent.
To detect the BrdUrd-incorporating nuclei, the coverslip-attached cells were stained with anti-MENT IgG as described above, washed for 5 min in PBS, and fixed by immersing in absolute methanol for 5 min at room temperature. Cells were then air-dried overnight in a lightprotected space and rehydrated by immersing in PBS for 3 min. DNA was denatured by incubating coverslips in 4 M HCl for 30 min at room temperature. To neutralize the acid, the slips were immersed in 0.1 M sodium borate buffer, pH 8.5, and the buffer was changed twice over a 10-min period. They were then washed 3 times for 3 min with PBS and overlaid with a 0.1-ml solution containing anti-bromodeoxyuridine-fluorescein (Roche Molecular Biochemicals). Specimens were incubated for 60 min at room temperature in a dark, humidified chamber, washed twice for 5 min with PBS, and mounted for fluorescent microscopy as above.
Fluorescence microscopy was conducted using an Eclipse E1000 automated microscope (Nikon Corp.) equipped with an Orca model C4742-95 digital camera (Hamamatsu Corp.). Image capturing and analysis was performed using Image-Pro Plus 4.1 software (Media Cybernetics), and image deconvolution was performed using the threedimensional blind deconvolution algorithm by AutoDeblur 7.5 software (AutoQuant Imaging).

MENT Is an Efficient Inhibitor of the Cysteine Proteinases
Cathepsin L and Cathepsin V-Previous studies (1, 3) demonstrated that MENT was unable to inhibit a variety of serine proteinases. We predicted that the bulky hydrophobic phenylalanine residue in the putative P 2 position would allow MENT to inhibit members of the papain-like cysteine proteinase family. This is not unprecedented, because previous work by Schick et al. (29) showed that the serpin SCCA-1 was capable of inhibiting cathepsins K, L, and S. It was found that recombinant MENT is an effective inhibitor of the human cysteine proteinases cathepsin L and V and sheep cathepsin L at pH 5.5, with k a values of 1.4 ϫ 10 6 M Ϫ1 ⅐s Ϫ1 , 4.0 ϫ 10 5 M Ϫ1 ⅐s Ϫ1 , and 2.2 ϫ 10 5 M Ϫ1 ⅐s Ϫ1 , respectively (Tables I and II (Table I; Fig. 2a, left panel). MENT was also capable of inhibiting human cathepsins L and V at neutral pH (Table I). Intact MENT purified from chicken granulocyte nuclei (MENT NA ) demonstrated similar inhibitory behavior (Table I) and was as efficient in immunoprecipitating sheep cathepsin L protein using anti-MENT antibodies as recombinant material, visualized by Western blot with antibodies against sheep cathepsin L (Fig. 2b, top left panel). A mutant lacking the M-loop (MENT MLOOP-, Fig. 1) gave almost identical SI and k a values to wild-type material, confirming that this region is not important for inhibitory activity (Table I).
Serpins inhibit serine proteinases using a unique "inhibition by distortion" mechanism (8), in which the RCL is cleaved in the final complex. To confirm the P 1 -P 1 Ј residues of MENT with respect to cathepsin L, a MALDI-TOF analysis of cathepsin L-treated MENT was performed. A single predominant peptide, not seen with either cathepsin L or MENT alone, was observed after the reaction of cathepsin L with MENT (Fig. 2c). This peptide had a molecular mass of 4694.5 kDa (average of 18 measurements, S.D. ϭ 5.8). This closely matches the predicted molecular mass (4691 kDa) of a product resulting from cleavage between the Thr 371 and Thr 372 residues, which corresponds with the consensus P 1 -P 1 Ј bond in serpins. As analyzed by acid native PAGE, MENT WT was found to be capable of forming a complex with both human cathepsin V and cathepsin L (Fig.  2b, bottom left and right panels). This complex was essentially irreversible; incubation of a 2-fold molar excess of MENT WT to cathepsin V showed no regain of proteolytic activity after 48 h at room temperature, whereas cathepsin V alone retained 70% activity. Consistent with previous studies investigating the complexes between crmA-ICE (30) and SCCA-1-cathepsin L (31), SDS-PAGE revealed that the complex between MENT and cathepsin L or cathepsin V was not SDS-stable (Fig. 2d). This is in contrast to a serpin-serine proteinase interaction, in which an SDS-stable complex is readily observed. Presumably this difference arises as a result of the greater susceptibility of a thioester bond to nucleophilic attack, when compared with the acyl bond formed with serine proteinases (32). The 1:1 stoichiometry and the high k a values strongly support the hypothesis that the physiological target of MENT is a papain-like cysteine proteinase. MENT WT was also able to inhibit other members of the cysteine proteinase family. It was found to be an effective inhibitor of human cathepsin K and a relatively less effective inhibitor of papain but not an inhibitor of human cathepsin B (Table II).
Numerous studies of serpins that inhibit serine proteinases have shown that mutation of the P 14 position to a large polar residue results in substrate-like rather than inhibitory behavior (33), presumably because the mutation interferes with RCL insertion. To determine whether such a structural transition was required in the inhibition of cathepsin proteinases by MENT, a P 14 Thr 3 Arg mutant (MENT P14R , Fig. 1) was constructed. Although some residual inhibitory activity was observed against cathepsin V (k a of 3.4 ϫ 10 3 M Ϫ1 ⅐s Ϫ1 ), MENT P14R acted predominantly as a substrate, with an SI of 50. The limited residual inhibitory activity observed is consistent with previous studies (33) on the interaction between P 14 Arg antitrypsin and serine proteinases. Correction of the k a value for the high SI (by multiplying the two) yielded a value close to that seen with MENT WT (1.7 ϫ 10 5 M Ϫ1 ⅐s Ϫ1 ). This indicates that the P 14 Arg mutation does not interfere significantly with the initial docking between MENT and cathepsin V but does perturb the transition to an irreversible complex. Consequently, the MENT P14R mutant interacted with but was unable to sustain an interaction with cathepsin proteinases; it did not co-precipitate sheep cathepsin L protein in the immunoprecipitation assay (Fig. 2b, top left panel, lanes 5 and 6).
Incubation of wild-type MENT with a 100-fold molar excess of the peptide SEAAAS (corresponding to the P 14 -P 8 residues of the RCL of antithrombin) resulted in a MENT-peptide complex (Fig. 2b, top right panel) that demonstrated no inhibitory activity (Table I). It is suggested that the 6-mer peptide inserts into the A-sheet in a similar fashion to that seen in the structure of PAI-1 in complex with two pentapeptides, preventing RCL insertion (34).
Finally, it was found that a construct of MENT in which the P 15 -P 4 Ј residues of the RCL were replaced with those of the non-inhibitory serpin ovalbumin (MENT OV ) was also unable to inhibit cathepsin L and was an extremely poor cathepsin V inhibitor (Table I).
Ectopic Expression of MENT in CV-1 Cells Is Sufficient to Protect Rb1 Protein from Degradation by SPase, a Nuclear Papain-type Cysteine Proteinase-Although cathepsins L and V are lysosomal proteins (35), cysteine proteinases with properties very similar to cathepsin L have been reported to be localized to the nucleus and involved in the degradation of a number of important nuclear regulators (35)(36)(37)(38). In particular, work by Nishinaka et al. (13) demonstrated the presence in CV-1 cells  and several other cell lines of a cathepsin L-like proteinase, SPase, whose activity correlated with degradation of underphosphorylated Rb protein (39), a factor that plays a pivotal role in regulating the cell cycle (40,41).
To examine the influence of MENT on SPase activity in vitro, we prepared nuclear extracts from wild-type (non-transfected) CV-1 and NIH/3T3 cells. Fig. 3a shows that the nuclear extract from wild-type CV-1 cells contained a strong cathepsin-like protease (SPase) activity. A nuclear extract prepared from NIH/3T3 cells, which do not express SPase (13), did not demonstrate such enzymatic activity. Addition of either E-64d or bacterially expressed MENT WT to the CV-1 extract completely abolished the SPase activity, whereas addition of bacterially expressed MENT P14R had no effect (Fig. 3a).
To monitor the stability of Rb protein, we then analyzed the CV-1 nuclear proteins using SDS-PAGE followed by Western blotting with an anti-Rb antibody. In the nuclear extract of CV-1 cells, the bands at around 105 kDa correspond to intact and multiphosphorylated forms of Rb and the 50-kDa band to the main product of Rb proteolysis by SPase (13). This latter degradation product was observed in the nuclei of control CV-1 cells but was dramatically reduced in cells isolated in the presence of 50 g/ml E-64d (Fig. 3b).
To characterize the potential influence of MENT on the stability of Rb protein, CV-1 cells were transfected with expression vectors containing MENT and mutant MENT ORFs, transcribed from a strong constitutive cytomegalovirus promoter. The stability of Rb protein was monitored in nuclear extracts derived from CV-1 cells expressing MENT WT , MENT P14R , and MENT OV (the latter two of which are essentially incapable of cathepsin L or V inhibition, see above) or transfected with expression vector without an insertion. A strong protection of the 105-kDa Rb bands and the absence of the 50-kDa product was observed immediately following nuclear extract preparation (Fig. 3c, lane 1) similar to that achieved by treatment with E-64d (Fig. 3b). No additional degradation was observed throughout a prolonged incubation period in cells expressing MENT WT (Fig. 3c, lanes 1-5). This is consistent with inhibition of the SPase-mediated intracellular degradation of Rb protein (13). In contrast, no protection of Rb was observed in cells expressing similar levels of MENT P14R (lanes 6 -10) and MENT OV (lanes [11][12][13][14][15] or transfected with control vector (lanes 16 -20). It was thus concluded that ectopic expression of MENT is sufficient to inhibit the nuclear cysteine proteinase activity that potentially plays a role in regulating the metabolism of the key proliferation control factor, Rb protein, in vivo.
The Inhibitory Reactive Center but Not the M-loop Domain of MENT Is Essential for a Proliferation Block-It was noted that during ectopic expression experiments all MENT WT -expressing cells ceased to proliferate and did not give rise to stable colonies producing even a low level of MENT. Normally, MENT is expressed in terminally differentiated blood cells, which may reflect a role in the repression of cell division during terminal differentiation. We therefore sought to determine which structural elements of MENT could be responsible for the cellular repression. Two unique, putatively important structural regions in this regard are MENT's chromatin-binding domain (the M-loop (3)) and the RCL, shown above to be essential for cysteine protease inhibition and capable of influencing the level of cell cycle regulators such as Rb. We used the murine fibroblast NIH/3T3 cells, which lack SPase activity (Fig. 3a), for comparison with CV-1 cells. Cells transiently expressing wildtype and MENT variants (Fig. 1) were identified by cell staining with anti-MENT antibodies and subsequent immunofluorescent microscopy (Fig. 4). The proliferation rates were monitored by co-immunofluorescence with antibodies against MENT and BrdUrd, a marker of DNA replication (43), and quantitation of the double positive cells with Image-Pro Plus software (Fig. 5).
In transfected CV-1 cells, MENT WT showed a predominantly nuclear localization as revealed by immunofluorescence with anti-MENT antibodies and counterstaining with a DNA-specific dye, Hoechst 33258 (Fig. 4, panels 1 and 2). This is consistent with the presence of a putative NLS in the MENT M-loop (3). Nuclei of MENT WT -expressing CV-1 cells (arrows on panels 1 and 2) showed an altered chromatin organization with more prominent Hoechst-positive foci and diminished nuclear diameters indicating partial chromatin condensation. Similar results were obtained with NIH/3T3 cells (Fig. 4, arrows on  panels 3 and 4). In contrast to cells transfected with vector alone, there was a very low rate of proliferation in either CV-1 or NIH/3T3 cells expressing MENT WT on the second day after transfection (Fig. 5A). Less than 3% of the MENT WT -positive CV-1 or NIH/3T3 cells survived a 30-day incubation in the culture media, and none gave rise to stable MENT WT -expressing colonies under neomycin selection.
To determine the role of the AT-hook (42), which lies within the M-loop, a variant of MENT lacking this motif was constructed by replacing the surrounding residues RPSRGRP with a similarly charged sequence, KGAKAKG (MENT ATHOOK-). FIG. 3. Inhibition of a nuclear cysteine proteinase (SPase) by MENT in CV-1 cells. a, cysteine proteinase activity in nuclear extracts (protein concentration 10 g/ml) isolated from NIH/3T3 cells (᭜) and CV-1 cells complemented with 25 g/ml E-64d (), 0.5 g/ml MENT WT (q), 0.5 g/ml MENT P14R (OE), and 0.5 g/ml control BSA (⅜). b, SDS-PAGE and Western blotting of the protein from the CV-1 nuclei (protein concentration 150 g/ml) without (Control) and with 50 g/ml E-64d (ϩ E-64d). c, SDS-PAGE and Western blotting of the protein from the nuclear extracts isolated from CV-1 cells transfected with cDNA encoding MENT WT (lanes 1-5), MENT P14R (lanes 6 -10), MENT OV (lanes [11][12][13][14][15], and control vector (lanes 16 -20) and incubated at 37°C for 0 min (lanes 1, 6, 11, and 16), 20 min (lanes 2, 7, 12, and 17), 1 h (lanes  3, 8, 13, and 18), 3 h (lanes 4, 9, 14, and 19) and 20 h (lanes 5, 10, 15, and 20). Western blots (b and c) were detected using antibodies against Rb. The positions of the major bands corresponding with intact Rb and its main proteolytic fragment at 50 kDa are indicated. This mutation significantly affected the association of MENT with nuclear chromatin, causing the localization of a substantial portion of the protein to the cytoplasm (Fig. 4, panel 5) but failed to rescue the block in proliferation (Fig. 5A). The mutant in which the M-loop was completely deleted, MENT MLOOP-, was more abundant in the cytoplasm than the nucleus in both CV-1 (Fig. 4, panel 6) and NIH/3T3 cells (Fig. 4, panel 8). This is consistent with the deletion of the NLS motif found in the MENT M-loop (3). Although the nuclear morphology of CV-1 cells expressing MENT MLOOP-(panel 7 on Fig. 4) did not show the signs of chromatin condensation observed with MENT WT (panel 2), the deletion of the M-loop did not rescue the proliferation block in either CV-1 or in NIH/3T3 cells (Fig. 5A).
It was thus concluded that the M-loop was not essential for the repressive effect imposed by MENT. Because the MENT MLOOPmutant retains full cathepsin inhibitory activity (Table I), we set out to determine whether this inhibitory activity could be linked to the effect on chromatin structure and cell proliferation.
The RCL mutations in MENT P14R and MENT OV were demonstrated above to virtually abolish the ability of MENT to inhibit cathepsin proteinases in vitro (Table I). Upon expression of MENT P14R or MENT OV in CV-1 cells, the mutants demonstrated a prominent nuclear localization (Fig. 4, panels [11][12][13]. However, in contrast with cells expressing either MENT WT or MENT MLOOP-, there was a dramatic increase in the rate of BrdUrd incorporation among CV-1 and NIH/3T3 cells expressing MENT P14R or MENT OV (Fig. 5A). A slightly (but significantly) lower BrdUrd incorporation in the presence of MENT P14R than of MENT OV was most probably due to residual inhibitory activity of the former. More than 10% of NIH/3T3 cells expressing MENT P14R or MENT OV survived a 30-day incubation in cell culture, and using neomycin selection, stable colonies expressing the mutant MENT could be obtained. 3 These experiments demonstrated that mutations in the RCL were able to rescue the MENT-imposed proliferation block without preventing intranuclear transport or inhibiting the binding of MENT to chromatin or DNA. Because the very same mutations have been shown to compromise the ability of MENT to inhibit cysteine proteinases in vitro and SPase in the CV-1 nuclear extracts (see above), these results for the first time link a proliferation block imposed by a nuclear serpin to its cysteine proteinase inhibitory activity.
To test if papain-like cysteine proteinase inhibition could induce the proliferative block independently of MENT, MENTtransfected cells were incubated with 50 g/ml cell-penetrating inhibitor, E-64d, by adding the inhibitor to the media. At this concentration, E-64d has been shown to block the intranuclear degradation of Rb by SPase (13). E-64d did not significantly inhibit BrdUrd incorporation in control CV-1 and NIH/3T3 cells (Fig. 5A), thus showing that inhibitory activity per se is not sufficient to block proliferation.
RCL Mutations and E-64d Affect Intranuclear Localization of MENT, Evidence for Its Interaction with a Papain Proteaselike Reactive Center in Vivo-NIH/3T3 have a well defined focal distribution of nuclear heterochromatin that can be easily visualized using Hoechst 33258, a fluorescent dye that preferentially stains AT-rich DNA in murine pericentromeric heterochromatin (44). By staining with anti-MENT antibodies, we observed that in NIH/3T3 cells expressing a low level of MENT, 3 3, 5, 6, 8, 9, 11, and 13-16) and 0.1 g/ml Hoechst 33258 dye (panels 2, 4, 7, 10, and 12). Scale bar represents 10 m.
the protein had a focal distribution in the nucleus co-localizing with heterochromatin (arrowheads on Fig. 4, panels 3 and 4).
In moderate-to-high MENT-expressing cells the protein was distributed throughout total nuclear chromatin (arrows on Fig.  4, panels 3 and 4; see also Fig. 5B).
A dramatically different pattern of immunostaining was observed in cells expressing MENT P14R and MENT OV . In these cells, the protein accumulated in a number of foci confined exclusively to the Hoechst-positive pericentromeric heterochromatin (Fig. 4, panels 9, 10, and 14), even at high levels of expression (Fig. 5B). This result shows that a single amino acid mutation abolishing cysteine proteinase inhibition leads to a large scale rearrangement of the distribution of MENT inside nuclear chromatin. We thus concluded that the organization of nuclear chromatin, the distribution of MENT, and repression of proliferation can all be affected by amino acid substitutions in the RCL of MENT that affect cysteine proteinase inhibition.
To investigate whether an exogenous cysteine proteinase inhibitor could substitute for the loss of the inhibitory activity, cells expressing MENT P14R were incubated in the presence of E-64d. This treatment, however, restored neither the anti-proliferative effect nor the chromatin localization of MENT P14R (Fig. 4, panel 15).
Cells expressing MENT WT were incubated in the presence of E-64d to test whether a cysteine proteinase is involved in the MENT-chromatin interaction. Remarkably, after a 24-h treatment with E-64d, the majority of MENT WT was relocalized to the heterochromatic regions, especially noticeable in cells expressing high levels of MENT (Fig. 4, panel 16; Fig. 5B), making the protein distribution similar to that of MENT P14R and MENT OV (Fig. 4, panels 9, 10 and 14; Fig. 5B). The E-64dtreated nuclei also had a normal morphology and shape clearly distinct from untreated, MENT WT -expressing cells (Fig. 4,  panel 3). The E64d-treated cells, however, showed a stronger background staining of euchromatic regions than with MENT P14R and MENT OV (panels 9 and 14). This residual staining probably reflects a less efficient disruption of the interaction between MENT WT and its euchromatic target by E-64d, which relies on competition between the molecules for a binding site.

DISCUSSION
The only heterochromatin protein identified to date that can bring about large scale chromatin condensation both in vitro and in vivo is an abundant, developmentally regulated nuclear protein, MENT, which is also a member of the serpin superfamily.
In this paper we demonstrate that MENT is an efficient inhibitor of cathepsins K, L, and V. Along with SCCA-1 (29), it therefore represents only the second serpin shown to be capable of effective inhibitory activity against cathepsin proteinases. It is worthwhile noting that both proteins arise from the same branch of the superfamily and that both proteins are predominantly intracellular, suggesting that there may be a growing subclass of serpins that, along with cystatins, perform a significant role as physiological intracellular modulators of cathepsin activity. The results presented here also have implications for the nature of the serpin-cysteine protease interaction. The MENT 6mer data (as well as that from MENT P14R and a corresponding SCCA-1 mutant (5)) support the hypothesis that the conformational changes required to inhibit both serine and cysteine proteinases are similar (45).
Given the specific nature of the interaction of MENT with certain papain-like cysteine proteinases, and given its predominantly nuclear localization, it was reasoned that MENT may be involved in regulating the activity of an intracellular, nuclear cysteine proteinase in vivo, in particular one similar to the recently identified enzyme SPase. SPase has been shown previously (39) to regulate levels of the cell cycle control protein Rb, which would therefore be expected to be modulated by the presence of a cysteine proteinase inhibitor. It was found that MENT, expressed in CV-1 cells or added exogenously to a nuclear extract, was capable of inhibiting SPase and preventing Rb degradation. Thus, in addition to inhibitory activity seen in vitro, MENT is capable of acting as a proteinase inhibitor in vivo and is able to inhibit an intracellular proteinase with the characteristics of the cysteine proteinase SPase.
Wild-type MENT was capable of exerting a strong proliferation block in CV-1 and NIH/3T3 cells. We do not believe that this effect is due to the inhibition of SPase; however, MENT was found to block the proliferation of CV-1 cells, which contain a relatively high level of SPase, as well as NIH/3T3 cells, which lack nuclear SPase (13). Furthermore, it was found that treating cells with E-64d, an inhibitor of papain-like cysteine proteinases, in the absence of MENT, was incapable of inducing a proliferation block. E-64d has been shown previously (13) to prevent the degradation of signaling molecules and transcription factors, including the degradation of Rb by SPase, and this Our previous studies have shown that MENT contains a heterochromatin-binding AT-hook motif within the M-loop region, and it was reasoned that this domain may be important for the repression of proliferation. To our surprise, neither the mutant which lacks the AT-hook nor the one in which the entire M-loop was deleted rescued the proliferation block. The observation that MENT MLOOP-was still capable of inhibiting proliferation may be explained by the passive diffusion of MENT MLOOP-into the nucleus (Fig. 4, panels 6 and 8). In contrast, either a single amino acid exchange (P14R) in the hinge region or the MENTovalbumin RCL loop swap rescued the proliferation block almost completely. Although these data suggest that the inhibitory activity of MENT appears crucial for blocking cell proliferation, the parallel experiments with E-64d show that the cysteine proteinase inhibitory activity per se is not sufficient and some other aspect of MENT is also required.
We were intrigued by the possibility that a nuclear cysteine proteinase may be involved in targeting MENT to specific regions of the nucleus. As described previously, in avian eryth-rocytes expressing low levels of MENT, the protein has a focal distribution in the nucleus (27), whereas at a high expression level, MENT is distributed throughout total nuclear chromatin (1). In NIH/3T3 cells we have also observed an expression level-dependent variation in nuclear localization of ectopically expressed MENT WT (Fig. 4, panel 3; Fig. 5B) but a dramatically different pattern of immunostaining when we expressed the non-inhibitory variant MENT P14R (Fig. 4, panels 9 and 10). In particular, this variant localized strongly to pericentromeric heterochromatin foci and failed to localize to euchromatin, even in cells expressing a high level of MENT (Fig. 5B). Similar results were obtained for the MENT OV variant. Furthermore, it was found that treatment with E-64d caused the distribution of MENT WT to resemble more closely that of MENT P14R (Fig. 4,  panel 16; Fig. 5B). E-64d is known to inhibit specifically papain-like cysteine proteinases by forming a covalent complex with the reactive center thiol (46) and blocking the S 1 -S 3 subsites (47). It is therefore likely that the treatment with E-64d counteracts the repressive effect of MENT on nuclear chromatin by blocking the active site of a MENT-interacting factor. The specificity of E-64d as well as the requirement for an inhibitory reactive center loop in MENT strongly suggest that one or more cysteine proteinases ubiquitously expressed in different cell types are directly involved in the association of MENT with chromatin in the nucleus, and may thus mediate or regulate the repression of cell proliferation caused by MENT.
These data suggest a scenario in which an active turnover of MENT in complex with a proteinase partner is competed for by the EϪ64d inhibitor. We therefore propose that the ability of MENT to associate with and repress euchromatin is mediated by its ability to interact and form a complex with a nuclear cysteine proteinase. We suggest that the association of MENT with AT-rich DNA in heterochromatin, in contrast, is mediated by the M-loop and AT-hook and thus does not require an active RCL. This would explain why mutations that impair MENTproteinase interactions simultaneously push the protein to heterochromatin and prevent the repression of proliferation. A scheme for the proposed targeting of MENT by cysteine proteinase that accounts for the effects of E-64d and the RCL mutations is shown in Fig. 6. We suggest that the MENTproteinase complex may in some way act as a "seed" for the cooperative binding of additional MENT molecules. The recruited MENT molecules would, in turn, form internucleosome bridges connecting chromatin fibers (2) in a conformation-dependent or concentration-dependent fashion. Such a mechanism would lead to the progressive condensation of neighboring euchromatin (Fig. 6). The nature of MENT conformational transitions triggered by the putative protease and the mechanism of its propagation remain to be characterized. FIG. 6. Proposed model for the recruitment of MENT to euchromatin by a cysteine proteinase-like factor. a, MENT expressed at a low level is associated with heterochromatin (solid line) through its DNA binding domain and also reacts with a putative cysteine proteinase associated with euchromatin (slashed line). A low concentration of MENT is insufficient for forming large complexes on euchromatin and causing condensation. b, when expressed at a higher level, excess of MENT WT binds to the previously seeded MENT cooperatively and accumulates in euchromatin at higher level than in heterochromatin. This cooperativity may be induced by the MENT-proteinase complex, although its mechanism is presently unknown. Accumulation of MENT in euchromatin apparently causes its condensation. c, in the presence of E-64d, the cysteine proteinase is inhibited; excess MENT WT binds to and accumulates in heterochromatin at a higher level than in euchromatin. d, the MENT P14R variant fails to form stable complex with the cysteine proteinase and also accumulates in heterochromatin, rather than in euchromatin.