Heterologous Expression of the Transcriptional Regulator Escargot Inhibits Megakaryocytic Endomitosis*

Certain cell types escape the strict mechanisms imposed on the majority of somatic cells to ensure the faithful inheritance of parental DNA content. This is the case in many embryonic tissues and certain adult cells such as mammalian hepatocytes and megakaryocytes. Megakaryocytic endomitosis is characterized by repeated S phases followed by abortive mitoses, resulting in mononucleated polyploid cells. Several cell cycle regulators have been proposed to play an active role in megakaryocytic polyploidization; however, little is known about upstream factors that could control endomitosis. Here we show that ectopic expression of the transcriptional repressorescargot interferes with the establishment of megakaryocytic endomitosis. Phorbol ester-induced polyploidization was inhibited in stably transfected megakaryoblastic HEL cells constitutively expressing escargot. Analysis of the expression and activity of different cell cycle factors revealed that Escargot affects the G1/S transition by influencing Cdk2 activity and cyclin A transcription. Nuclear proteins that specifically bind the Escargot-binding element were detected in endomitotic and non-endomitotic megakaryoblastic cells, but down-regulation occurred only during differentiation of cells that become polyploid. As Escargot was originally implicated in ploidy maintenance of Drosophila embryonic and larval cells, our results suggest that polyploidization in megakaryocytes might respond to mechanisms conserved from early development to adult cells that need to escape normal control of the diploid state.

In most somatic cells, the ploidy of progeny is maintained by tightly controlling the proper alternation of DNA replication and mitosis. However, polyploidization occurs in different cell types from plants to vertebrates as part of their physiological differentiation programs. For instance, most embryonic cells that do not give rise to adult structures are highly polyploid, probably as a means to potentiate the high metabolic rate needed to sustain the developing embryo (1). In adult tissues, a few highly differentiated and mature cells also appear to re-quire an increase in DNA content, as is the case for mammalian hepatocytes and the platelet precursor cell, the megakaryocyte. In the final stages of maturation, megakaryoblasts stop proliferating and undergo from three to five truncated cell cycles, resulting in polyploid, fully mature megakaryocytes.
To increase their DNA content to levels up to 128N, megakaryocytes repeatedly enter S phase after escaping mitosis without completing anaphase in a process traditionally called endomitosis (2,3). In vitro and in vivo experimental approaches have shown that both G 1 /S and G 2 /M cell cycle transitions have to be regulated for megakaryocytes to achieve polyploidization (4 -12). Specifically, both cyclin D3-Cdk2 and cyclin E-Cdk2 complexes have been proposed as the most likely candidates to drive megakaryocytic polyploidization (4,11,12). However, little is known about the mechanisms upstream of the cell cycle machinery that ultimately control the proper establishment of endomitotic cycles in differentiating megakaryoblasts.
In embryonic systems, the entrance into re-replication cycles seems to be inhibited by proteins belonging to the Snail family of transcriptional repressors (for recent reviews, see Refs. 13 and 14). All Snail proteins contain four to six highly conserved zinc fingers at the C terminus of the protein through which they bind DNA. In vitro binding site selection experiments using pools of random sequences have revealed that the consensus binding element corresponds to an E2 box that is recognized by a variety of basic helix-loop-helix transcriptional activators. It has been proposed that the Snail proteins may compete directly with basic helix-loop-helix proteins for the same binding sequences, although this has been observed only in a limited number of cases (19,20). Genetic studies in Drosophila, Xenopus, Caenorhabditis elegans, and mouse have shown that Snail proteins participate in a variety of processes that include mesoderm determination, neurogenesis, and apoptosis (13,14). More recently, Snail proteins have been implicated in the invasiveness of cancer cells (15,16) and in the development of B lymphomas (17), revealing that they might have important roles in adult tissues.
Escargot was first described in relation to the control of Drosophila imaginal cell development (18). Further analysis revealed that Escargot is involved in the maintenance of diploidy in embryonic and larval cells (19). Thus, ectopic expression of escargot inhibits polyploidization in salivary glands, whereas its ablation prevents the regular progression through mitotic cycles in the imaginal tissues and, as a result, causes abdominal histoblasts to undergo endocycles and become polyploid (18,19). Interestingly, Escargot inhibits polyploidization of murine trophoblast giant cells as efficiently as its murine homolog, Snail (20). In addition, the endogenous expression pattern of both escargot and snail further suggests an inhibitory activity on Drosophila and mouse embryonic endo-cycles, respectively, as both proteins are down-regulated in those cells that become polyploid (18,20).
Here we have asked whether a similar role could be played by proteins related to Escargot in megakaryocytic endomitosis. The megakaryoblastic cell line HEL normally responds to the phorbol ester TPA 1 by undergoing terminal differentiation including polyploidization. We have isolated stable clones of HEL cells constitutively expressing escargot and monitored their response to TPA in terms of differentiation marker expression, extent of polyploidization, and expression and activity of different cell cycle regulators. In addition, we have explored whether the presence of endogenous Escargot-like proteins capable of specifically binding to an Escargot-binding element (EBE) correlates with the ability of megakaryoblastic cells to become polyploid. Our results show that escargot expression interferes with the regulation of the G 1 /S transition of endomitotic cycles and thereby with megakaryocytic polyploidization. They also suggest that down-regulation of proteins that specifically interact with the EBE could be required for megakaryoblastic cells to become polyploid.

MATERIALS AND METHODS
Cell Culture-Cells were cultured in RPMI 1640 medium supplemented with 10% (v/v) fetal calf serum (BioWhittaker), 2 mM L-glutamine, and 60 mg/ml gentamycin. Cells were maintained at 37°C under 5% CO 2 and 95% air in a humidified incubator. In all experiments, exponentially growing cells were subcultured into Nunc 96-, 24-, or 6-well plates with or without appropriate treatment. The number of viable cells was determined by trypan blue exclusion in a hemocytometer chamber. To induce megakaryocytic differentiation, 0.15-0.20 ϫ 10 6 cells/ml were grown in the presence or absence of 10 Ϫ8 M TPA (Sigma) for the indicated times.
Constructs and Oligonucleotides-Full-length escargot DNA was obtained from Dr. Mary Whiteley. Expression vector pcDNA3-ESG was generated by subcloning an XbaI fragment containing the ESG open reading frame into the pcDNA3 polycloning site (Invitrogen). The construction of the cycA(Ϫ875)luc reporter plasmid containing the human cyclin A promoter has been described elsewhere (11).
Transfection and Isolation of escargot-expressing HEL Cells-HEL cells (American Type Culture Collection) were transfected by electroporation at 125 microfarads/300 V in a 0.4-cm cuvette. 5 ϫ 10 6 cells were transfected with 5 g of linearized pcDNA3 or pcDNA3-ESG plasmids. After 24 h, cells were washed with phosphate-buffered saline and resuspended in fresh medium supplemented with 500 g/ml G418 antibiotic (Life Technologies, Inc.). Cells were subcultured in 96-well plates at 1 ϫ 10 5 cells/ml. Single cell clones were isolated by limiting dilution of the G418-resistant cells.
For transient transfection experiments, cell lines were transfected by electroporation. 2 ϫ 10 6 cells were transfected with 2.5 g of plasmid DNA by electroporation. For TPA treatments, electroporated cells were divided into two aliquots and allowed to recover for 12 h. One aliquot was then treated with TPA. 48 h after electroporation, the cells were harvested, and cell extracts for assaying luciferase activity were made by three cycles of freeze-thaw lysis.
Phenotypic Characterization of Cells-Morphological Wright stain was performed on cytospun cells. Cell-surface immunofluorescent staining was performed by incubating cells with antibodies against glycoprotein IIIa/CD61 or glycophorin A conjugated to FITC (Becton Dickinson) as previously described (11). DNA content was determined by staining with 50 g/ml propidium iodide. Cell cycle analysis was performed using a FACScan analyzer using CellQuest software (Becton Dickinson).
Mobility Gel Shift Analysis-Nuclear extracts were obtained by lysing cells in high salt buffer as previously described (30). Binding reactions were carried out for 20 min at room temperature in binding buffer (20% glycerol, 1 mM dithiothreitol, 20 mM HEPES (pH 7.9), 5 mM MgCl 2 , and 0.2 mM EDTA) containing 0.02 pmol of 32 P-labeled oligomer, various amounts of protein extracts as indicated, 40 g/ml poly(dI-dC), and 100 mM KCl. When needed, competitor oligonucleotides were added in excess to the binding reaction. The probe or wild-type competitor (EBE) was GCGGCCTGACAGGTGCTTTG. The mutant competitor (EBE M ) was GCGGCCTGCACACTGCTTTG, and the non-related competitor was GCGGCCTGCGTTACCTTTG. DNA-protein complexes were separated from unbound labeled oligonucleotide on 6% nondenaturing polyacrylamide gels in Tris borate/EDTA buffer. After drying, the shifted complexes were visualized by autoradiography.
Immunoblotting-Western blotting was performed as previously described (11). Briefly, total cellular proteins were extracted in lysis buffer (20 mM Tris-HCl (pH 7.4), 10 mM EDTA, 100 mM NaCl, and 1% Triton X-100) containing protease and phosphatase inhibitors. Protein extracts (30 g) were subjected to SDS-polyacrylamide gel electrophoresis, and gels were transferred to BioTrace TM polyvinylidene difluoride membranes (Pall Corp.) for 1 h at 2 mA/cm 2 on a semidry transfer apparatus (Amersham Pharmacia Biotech). After blocking in 5% skim dry milk in Tris-buffered saline containing 0.1% Tween 20 (TTBS), filters were incubated overnight at 4°C with the appropriate primary antibody diluted in TTBS. After washing and incubation with goat anti-rabbit or goat anti-mouse IgG conjugated to horseradish peroxidase (Dako), signals were detected using the enhanced chemiluminescence system (Pierce).
Immunoprecipitation and Kinase Activity Determination-Extracts were isolated as described for Western blot analyses. Total protein was incubated for 2 h at 4°C in the presence of anti-Cdk2 antibody. Immunocomplexes were then isolated by the addition of 0.1 volume of 50% (v/v) protein A-Sepharose and incubation for 2 h at 4°C on a rotating wheel. Immunoprecipitates were washed four times with 1 ml of icecold lysis buffer and once with 1 ml of kinase buffer (50 mM Tris-HCl (pH 7.4), 10 mM magnesium chloride, and 1 mM dithiothreitol). Pellets were then resuspended in 40 l of kinase buffer with 250 ng of the carboxyl-terminal part of retinoblastoma protein (Santa Cruz Biotechnology). Reactions were initiated by the addition of 5 M ATP and 10 Ci of [␥-32 P]ATP (3000 Ci/mmol), incubated at 30°C for 30 min, stopped by the addition of 4ϫ Laemmli sample buffer, and boiled for 5 min. 25 l of each reaction was analyzed by SDS-polyacrylamide gel electrophoresis on a 10% polyacrylamide gel, and bands were detected by autoradiography.
Immunocytochemical Detection of Cyclins-For the simultaneous analysis of cell cycle and cyclin expression, cells were washed with phosphate-buffered saline and fixed in suspension in cold 75% ethanol. After washing and permeabilization, cells were pelleted and incubated with either FITC-conjugated anti-cyclin A antibody or isotype control antibody or with anti-cyclin E antibody and an appropriate irrelevant IgG1 control (all antibodies were obtained from Pharmingen). Anticyclin E binding was detected after washing by incubation with BODIPY-FL-conjugated goat anti-mouse IgG (Molecular Probes, Inc.). After appropriate washing, cells were labeled with propidium iodide in phosphate-buffered saline. Analysis was performed using a FACScan analyzer using CellQuest software.

Constitutive Expression of escargot Does Not Affect the Growth or Morphology of Megakaryoblastic HEL Cells-En-
domitotic megakaryoblastic HEL cells were transfected with full-length Esg cDNA under the control of a cytomegalovirus promoter. 10 clones were selected in the presence of G418 and analyzed for the presence of functional ESG protein. Nuclear extracts of transfectants were analyzed by electrophoretic mobility shift assays using a labeled double-stranded oligonucleotide containing the consensus binding site for ESG (ACAG-GTG, hereinafter referred to as the EBE). As shown in Fig. 1a, extracts from four of the clones transfected with the ESGexpressing vector gave a pattern of shifted proteins that was not present in a G418-resistant clone of HEL cells transfected with an empty vector. The latter was subsequently used as control HEL (referred to as HELc) cells. A similar pattern of protein-EBE complexes was observed in COS cells transiently transfected with the same ESG-expressing construct (Fig. 1c). The specificity of binding to the EBE of proteins derived from the HEL/ESG clones was assessed by competing with a 100-fold excess of the EBE oligonucleotide, a mutant version of the EBE that does not bind to ESG (EBE M ), or an oligonucleotide containing an unrelated sequence. This revealed that the two predominant complexes of higher mobility represented specific 1 The abbreviations used are: complexes between the ectopically expressed ESG and EBE, whereas the lower mobility complexes seen in both transfected and control clones were nonspecific and unrelated to the presence of ESG (Fig. 1b). Based on these preliminary observations, the HEL/ESG clones HA1 and HD12 were selected for further analysis.
To determine whether constitutive expression of ESG was affecting the intrinsic phenotypic properties of HEL cells, clones HA1 and HD12 were examined with respect to their proliferation under normal growth conditions. In addition, the clones were tested for differentiation in response to treatment with the phorbol ester TPA, which triggers the differentiation of HEL cells, leading to decreased cell division and acquisition of mature megakaryocytic features. Constitutive expression of ESG did not affect the exponential growth of HEL cells in standard growth medium or the growth arrest response to TPA (Fig. 2a). The morphology of exponentially growing HELc and ESG-expressing clones was very similar, with an immature morphology typical of erythroleukemic cells (Fig. 2c). Likewise, the TPA-driven up-regulation of megakaryocytic markers such as glycoprotein IIIa (CD61) and down-regulation of the erythrocytic marker glycophorin A remained unchanged in the ESGexpressing cells compared with HELc cells (Fig. 2b). The morphology of exponentially growing HELc and ESG-expressing clones was very similar, with a blast morphology typical of erythroleukemic cells (Fig. 2c). However, profound differences were observed when HELc, HA1, and HD12 cells were treated with TPA. Hence, a significant proportion of differentiated HELc cells appeared as large cells with a polylobulated nucleus, whereas ESG-expressing HEL cells showed no cytoplasmic or nuclear maturation and the presence of large vacuoles in a high proportion of cells. Therefore, it appeared that ESG was specifically interfering with the completion of the TPA-driven megakaryocytic differentiation program.
Escargot Interferes with Phorbol Ester-induced Polyploidization-Next we analyzed the nuclear DNA content of TPAtreated cells to determine whether the key feature of megakaryocytic maturation, i.e. the degree of polyploidization, was also affected in the presence of ESG. Control and ESGexpressing HEL clones were seeded at 1.5 ϫ 10 5 cells/ml in the presence or absence of TPA and stained with propidium iodide 48 and 96 h after treatment. More than 50% of the HELc cells showed a DNA content of 4N or higher (10 and 1% of cells being 8N and 16N, respectively) 48 h after treatment, whereas HA1 and HD12 cells failed to become polyploid even after 96 h in the presence of TPA (Fig. 3). Instead, the propidium iodide staining pattern indicated that ESG-expressing HEL cells remained arrested in G 1 and G 2 . This behavior appeared to be reminiscent of megakaryoblastic K562 cells, which respond to TPA in terms of platelet antigen up-regulation, but do not become polyploid (Fig. 3). As Escargot DNA-binding properties were not affected by the TPA treatment (data not shown), we conclude that Escargot inhibits the establishment of endomitotic cycles in megakaryoblastic cells.
Escargot-mediated Inhibition of Megakaryocytic Polyploidization Occurs through Alteration of the Endomitotic G 1 /S Transition-Establishment of endomitotic cycles in HEL cells requires active cyclin E-Cdk2 complexes, which in turn maintain cyclin A expression (11). We therefore analyzed the G 1 /S machinery in HELc cells and ESG-expressing clones before and after treatment with TPA. As shown in Fig. 4a, TPA treatment of HA1 and HD12 cells resulted in a decrease in the levels of cyclins E and A and Cdk2 activity, whereas p27 kip1 inhibitor levels remained unchanged or even increased. This was in sharp contrast with the pattern of expression of these proteins in HELc cells, which showed the expected maintenance of cyclin and p27 kip1 levels and Cdk2 activity. Interestingly, the pattern of TPA-induced changes in HA1 and HD12 cells resembled that in non-endomitotic K562 cells with the exception that the ESG-expressing cells demonstrated a more dramatic down-regulation of retinoblastoma protein. These data suggest that a profound inhibition of the G 1 /S transition in TPA-treated cells occurs in the presence of Escargot.
To investigate in greater detail whether Escargot was affecting the maintenance of expression of G 1 /S regulators during TPA-driven differentiation, a time course bivariant flow cytometric analysis was performed. Prior to TPA treatment, cyclins E and A were present in G 1 /S and S/G 2 cells, respectively, as expected for exponentially growing cells (Fig. 4b, t ϭ 0). Subsequent to the addition of TPA, it appeared that the level of both cyclins progressively declined for the first 24 h and that no cells expressed the proteins. In fact, 48 h after differentiation, HA1 and HD12 lacked nuclear cyclins E and A, as was also observed in K562 cells. In contrast, polyploid HELc cells maintained waves of expression of the G 1 /S cyclins (Fig. 4b). These results suggest that the cell populations that are entering S and M phases at the time of TPA treatment respectively progress through replication and mitosis during the first hours of differentiation, but that no further G 1 /S transition takes place in escargot-expressing HEL cells.
Escargot Affects the Transcriptional Regulation of Cyclin A-We have previously observed that TPA inhibits cyclin A expression in non-endomitotic megakaryoblastic cells (e.g. K562) and that ectopic expression of cyclin E enables these cells to reestablish cyclin A expression and, as a consequence, to undergo endomitotic replication (11). Therefore, a possible target for the action of Escargot could be the transcriptional regulation of cyclin A gene expression. To test this, HELc, HA1, and HD12 cells were transiently transfected with the cycA(Ϫ875)luc construct, which contains the luciferase reporter gene under the control of sequences from Ϫ875 to ϩ37 base pairs of the human cyclin A promoter. Promoter activity was then assessed in the presence and absence of TPA. As we had previously shown (11), the activity of the cyclin A promoter in HEL cells was maintained or slightly increased in TPAtreated cells. In contrast, TPA resulted in a Ͼ3-fold decrease in luciferase expression in escargot-expressing cells (Fig. 5). This apparent inhibition of cyclin A promoter activity in ESG-expressing HEL cells following TPA treatment was similar to the degree of inhibition that was observed in TPA-treated K562 cells (40% of activity compared with untreated cells). These results are consistent with the idea that the mechanisms by which Escargot blocks the establishment of endomitotic cycles in HEL cells could be similar to those operating in non-endomi-

totic K562 cells and may involve effects at the level of the regulation of cyclin A gene expression.
Human Proteins That Bind to the Escargot DNA-binding Site Can Be Detected in Non-endomitotic Cells-The phenotypic similarities between Escargot-expressing HEL cells and nonendomitotic K562 cells suggested that endogenous Escargotlike proteins could be present in these latter cells. To test this possibility, electrophoretic mobility shift assays were performed on nuclear extracts isolated from untreated or TPAtreated HEL and K562 cells. Under the conditions used to assess ectopic escargot expression in transfected cells, no specific bands could be detected in HELc cells (Fig. 1a) or K562 cells (data not shown). However, when 2-3 times higher amounts of protein extract were used, a major complex could be detected in both HEL and K562 cells (Fig. 6, upper left panel). The specificity of the binding was estimated by competition with an excess of unlabeled EBE or of EBE M (Fig. 6, lower panels). A similar binding pattern was detected in a variety of human cell types, including HeLa (epithelial), and U2OS (osteosarcoma) cells (Fig. 6, upper right panel). Perhaps most significantly, the amount of this specific complex was much higher in electrophoretic mobility shift assays performed with extracts from K562 and the other cells analyzed compared with those using HEL extracts. Interestingly, the amount of protein able to form this specific complex with EBE remained constant, or even increased, in TPA-treated K562 cells, whereas it almost disappeared in differentiated endomitotic HEL cells. These data indicate that human cells contain one or more nuclear EBE-binding proteins. They also suggest that the presence of endogenous EBE-binding proteins, like ectopically expressed escargot in HEL, is not compatible with establishment of megakaryocytic endomitotic cycles. DISCUSSION In this study, we have shown that constitutive expression of the transcriptional repressor Escargot inhibits megakaryocytic polyploidization. Previous genetic and biochemical evidence had already demonstrated an inhibitory role for Drosophila Escargot and murine Snail factors in the establishment of endocycles in early development (19 -21). The data presented here extend this evidence and indicate that the Drosophila protein retains its specific function in re-replication control when expressed in adult mammalian cells. The inhibition of endomitosis is also observed when mouse Snail is ectopically expressed in HEL cells, 2 pointing to a general role of certain members of the Snail family in the control of ploidy maintenance. The cloning of a human homolog, SnaH, has recently been reported (29). It could then be speculated that Escargot is affecting target promoters physiologically controlled by the endogenous protein, given the degree of homology between human and mouse proteins (overall 83%, reaching Ͼ95% identity in the DNA-binding region). Although the binding elements that SnaH appears to preferentially recognize are not coincident with the ones shown for its mouse homolog or Escargot (29), it is important to bear in mind that these factors could bind to variants of the sequences tested by in vitro random selections. A putative target of Escargot relevant to megakaryocytic endomitosis could be cyclin A: we have shown here that its expression is down-regulated in escargot-expressing cells and that the transcriptional activity of its proximal promoter region is significantly reduced in these cells. Three E boxes that could account for a repressive effect of Escargot/SnaH on the transcription of the cyclin A gene are present in the proximal promoter region that was present in our cyclin A promoterreporter construct (CACTTG, CATATG, and CAAGTG at Ϫ538, Ϫ469, and Ϫ387). Although none of these sites corresponds to the consensus Escargot-binding element (19), it could be possible that the overexpressed heterologous protein binds in vivo to one or more of them. Work is in progress to investigate whether cyclin A is one of the putative Escargot target genes that mediates its effect on endoreplication and to determine whether SnaH is physiologically relevant to megakaryocytic polyploidization. HELc, HA1, and HD12 cells were transfected by electroporation with 2 g of cycA(Ϫ875)luc reporter constructs. TPA was added to half of the cells; and 48 h later, cells were collected and assayed for luciferase activity. The white and black bars represent untreated and TPA-treated cells, respectively. The activity of the cyclin A promoter has been arbitrarily set to a value of 1 for the exponentially growing untreated aliquot of the three cell lines transfected; the values for the activity obtained in the TPA-treated cells are then given relative to this activity. The -fold stimulation of the promoter induced by TPA is shown at the top.
FIG. 6. Nuclear extracts from human cells contain proteins that bind to the Escargot-binding element. Upper panels, nuclear extracts from TPA-treated (ϩ) or untreated (Ϫ) megakaryoblastic HEL and K562 cells and non-megakaryoblastic HeLa and U2OS cells were incubated with labeled EBE, and the protein complexes were analyzed by electrophoretic mobility shift assays as described under "Materials and Methods." Reactions were carried out with 12 g of protein, except for HA1 (5 g). Lower panels, the binding reaction was carried out in the presence of an excess of unlabeled EBE or of EBE M (25, and 100 times as indicated) in HA1, K562, and HeLa cells. The black arrows show the ectopic Escargot shifted complexes, and the white arrow shows the specific endogenous EBE-bound complex.
Another interesting point to take into account is that the megakaryocytic endomitotic cycle presents substantial differences compared with embryonic endocycles. Although the latter consist of successions of DNA duplication and short "gap" phases in which G 2 /M cyclins are absent (1,22), megakaryocytes complete the G 2 phase and enter an abortive mitosis lacking karyo-and cytokinesis (23). However, despite these significant differences, our results suggest that mechanisms similar to those inhibited by Escargot in embryonic cells may account for polyploidy in these highly differentiated adult cells.
In all the systems analyzed to date, polyploidization is accompanied by maintenance of an active G 1 /S transition and down-regulation of the mitotic machinery (21, 24 -27). In megakaryocytic endomitosis, the G 2 /M transition takes place in the presence of cyclin A and of active, albeit reduced, cyclin B-Cdc2 complexes (3,5,10,11,28). Here we show that the G 1 /S transition machinery, previously implicated in the establishment of endomitosis in TPA-treated megakaryoblastic cells, is inhibited in escargot-expressing cells. Thus, Cdk2 activity is not detected in the differentiating escargot-expressing HEL cells, and this could result in the down-regulation of cyclin A. In this respect, our data slightly differ from previous interpretations that Escargot/Snail specifically interfere with G 2 /M machinery (20,21). Thus, the inhibitory effect of Escargot and Snail was described to affect the obligate down-regulation of mitotic Cdc2 and cyclins A and B that occurs in embryonic endocycles (20,21,26,27). We have also observed lower levels of cyclin B and also of Cdc2 in escargot-expressing cells, 3 but our interpretation is that the poor expression of these G 2 /M factors is an indirect consequence of the majority of cells not proceeding beyond G 1 . We believe that the key point at which Escargot prevents HEL cells from proceeding into endomitotic cycles lies in the regulation of the G 1 /S machinery, as a profound down-regulation of cyclin E, Cdk2, and also of retinoblastoma protein occurs in the overexpressing cells.
An additional observation is that the expression of escargot in HEL cells results in a phenotype that is strikingly reminiscent of non-endomitotic K562 cells. Hence, escargot-expressing cells respond to TPA by up-regulating megakaryocytic and concomitantly down-regulating erythrocytic markers, but do not undergo polyploidization (11). This is also consistent with the idea that Escargot-like proteins could be playing a determinant role in megakaryocytic differentiation. It can then be hypothesized that K562 cells do not enter endomitosis because of the presence of a protein functionally equivalent to Escargot, which would then be down-regulated in differentiating HEL cells. Ectopic expression of escargot in HEL cells might mimic an endogenous factor by interfering with the establishment of endomitotic cycles in these cells. The fact that EBE-binding proteins are present in TPA-treated K562 cells but absent in HEL cells suggests that this could be the case. Although further work is needed to determine whether the EBE-binding activities that we have detected correspond to Snail factors or to another class of DNA-binding proteins, the evidence presented in this study suggests that megakaryocytic endomitotic differentiation could require the down-regulation or inhibition of such E2 box-binding factors.