INTRODUCTION

Top Previous Next References

A member of the transcription factor family of proteins called polyomavirus enhancer binding protein 2 (PEBP2) or core binding factor (CBF) is composed of dimers of  and  subunits. The subunit possesses regions responsible for direct DNA binding and transactivation, while the  subunit facilitates DNA binding by the  subunit (12, 32). Gene cloning studies revealed three  subunit encoding genes, PEBP2A/ CBFA1/AML3, PEBP2B/CBFA2/AML1, and PEBP2C/ CBFA3/AML2, in mammals, and two genes, runt and lozenge, in Drosophila. The  subunit is encoded by a single gene, PEBP2/CBFB, in mammals, and by two genes, brother and big brother, in Drosophila.

Both the  and  subunit-encoding genes are independently rearranged in the chromosomal abnormalities associated with leukemia. Chromosomal translocations involving AML1 include t(8;21) in the French-American-British (FAB) M2 subtype of acute myelogenous leukemia (AML) and t(12;21) in childhood acute lymphoblastic leukemia (ALL). The t(8;21) and t(12;21) translocations produce the chimeric proteins AML1/ETO(MTG8) and TEL-AML1, respectively (5, 9, 23). The  subunit is rearranged in inv(16) of the FAB M4Eo subtype of AML, producing a chimeric protein, CBF/PEBP2-SMMHC (18) (referred to here as -SMMHC). The incidences of these fusion proteins involving PEBP2 subunits in leukemias are as follows: AML1/ETO, 12% of AML; -SMMHC, 12% of AML; and TEL-AML1, 20% of ALL. These values represent a significant contribution to human leukemia in general (19).

Recently, both the  and  subunits were shown to be essential for hematopoiesis in mice. Targeted disruption of either the PEBP2B/CBFA2/AML1 or PEBP2/CBFB gene resulted in nearly identical phenotypes of embryonic lethality with accompanying hemorrhage of the central nervous system and defects in fetal liver definitive hematopoiesis (25, 28, 30, 37, 38). These studies thus proved that both subunits of PEBP2 are indispensable for its in vivo function. Interestingly, heterozygous mice having targeted insertion (knockin) of genes complementary to AML1/ETO and CBF/PEBP2-MYH11 (coding for the -SMMHC protein) displayed phenotypes similar to those of the corresponding targeted disruptions (4, 41). This finding indicated that both chimeric proteins acted as dominant negative effectors.

The  subunit has been shown by electrophoretic mobility shift assays (EMSA) to act as a cofactor for DNA binding. This subunit is derived from the single PEBP2/CBFB gene by alternative splicing to give at least three isoforms, termed 1, 2, and 3 (26, 36). None of the isoforms exhibited DNA binding by themselves. 1 and 2 intensified DNA binding of the  subunit and gave a characteristic band supershift. In contrast, 3 only mildly intensified DNA binding of the  subunit, and the band supershift was not readily apparent. Therefore, it was tentatively concluded that 1 and 2 were functional but 3 was not (13, 26, 38). The only structural difference between 1 and 3 is the absence of the exon 5-encoded region from 3 (26) (Fig. 1A). Accordingly, two of the three independent -gene knockout studies adapted a strategy to target exon 5 of the gene, leaving only the 3 isoform to be expressed (30, 38). These mice exhibited phenotypes identical to those observed when exon 1 was targeted to achieve the authentic null condition (25), a result that may simply imply that 3 is not functional in vivo either. However, well-controlled analysis of transcriptional potential of 1, 2, and 3 will be required before any firm conclusions can be drawn.

View larger version (24K): [in this window] [in a new window]   FIG. 1.   Schematic illustration of the structures of full-length and deletion derivatives of the  (A) and  (B) subunits of PEBP2. Numbers denote the positions of amino acids. (A) 1, 2, and 3 are naturally occurring isoforms of the  subunit. The structure of -SMMHC is also shown. E5 denotes the region encoded by exon 5. Two different splice donor sites are present in exon 5. The black regions in 1 and 3 are encoded by exon 6 but translated in a frame different from that of 2. (B) The Runt domain spans aa 50 to 177, and VWRPY is a stretch of the last 5 aa of B1. Transactivation domain (AD) and inhibitory domain (ID) are also shown.

The inv(16) chromosome rearrangement produces fusion proteins collectively called -SMMHC (18). The protein is composed of the amino-terminal portion of the PEBP2  subunit and the carboxy-terminal portion of the smooth muscle myosin heavy chain (SMMHC). Major chromosomal breakpoints occur in intron 5 of the PEBP2 gene, which result in the conservation of the first 165 amino acids (aa) of the  subunit in the -SMMHC protein. The function of -SMMHC has been studied by EMSA using nuclear extracts from -MYH11 gene-transfected cells (2). The study showed a reduction in DNA binding of the normal - dimers and simultaneous appearance of --SMMHC complexes. However, the function of -SMMHC with respect to transactivation activity of PEBP2 was not addressed.

Analysis of the function of the  subunit as well as that of -SMMHC in transactivation requires experimental systems that meet two criteria: little or no transactivation by either subunit alone and strong transactivation when both are present together. It is frequently observed that the  subunit alone apparently induces PEBP2 site-dependent transactivation (1, 6, 17, 22, 27, 33, 42). The nature of -subunit-independent transactivation is unknown, especially in terms of the extent to which this activity depends on the endogenous  subunit. This latter point needs to be emphasized, as it is possible that other transcription factors can also facilitate DNA binding of the  subunit. Also, an earlier study on the effect of the  subunit was undermined by the relatively high activity obtained with transfection of the  subunit alone (42). Recently, we have found that luciferase assays performed in Jurkat T cells by using a reporter containing a regulatory element from the macrophage colony-stimulating factor (M-CSF) receptor promoter meet the above criteria (14). In the present study, we have analyzed the transcriptional properties of the  subunit as well as those of -SMMHC. We show that -SMMHC fails to support PEBP2-dependent transactivation despite the fact that it contains the region essential for -subunit function. Subsequently we show that -SMMHC sequesters the  subunit in the cytoplasm, thereby precluding it from acting in the nucleus.

MATERIALS AND METHODS

Top Previous Next References

Plasmid construction. Mammalian expression vector pEF-BOS (24) was used to make a series of expression plasmids: pEF-B1, pEF-B1(1-243), and pEF-B1(1-183) (1); pEF-B1(1-446), pEF-1, pEF-2, pEF-3, and pEF--SMMHC (previously referred to as pEF-b/MYH11) (20); pEF-AML1(453) (previously referred to as AML1b) (45); and pEF-B1(1-411), pEF-B1(1-371), pEF-B1(1-331), and pEF-B1(1-291) (14). For the -deletion constructs [pEF-165 (human), pEF-135, and pEF-117], corresponding regions of the  subunit were PCR amplified and ligated into the XbaI site of pEF-BOS. The authenticity of the PCR-amplified sequences was confirmed by sequencing. pMSV-C/EBP (3) and the luciferase reporter pM-CSF-R-luc (43) have already been described. pME-B1-GRLBD was constructed as follows. The glucocorticoid receptor ligand binding domain (GRLBD) sequence was cut out from pRShGR_NX (7) by XhoI/BamHI digestion and ligated to the carboxy-terminal sequence of B1 by using the internal BssHII site and a synthetic linker to adjust the reading frame. The entire B1-GRLBD sequence was then cloned into pME18S, an SR promoter-driven expression plasmid (34).

Transfection and luciferase assays. Jurkat human T cells were maintained in RPMI 1640 supplemented with 10% fetal calf serum (FCS) and antibiotics. The cells were transfected with a reporter plus expression plasmids via electroporation, and relative luciferase activities were assayed as described previously (14). The compositions of the transfected plasmids are described in the figure legends. The total amount of transfected DNA was always kept at 10 µg, using the empty vector pEF-BOS.

Indirect immunofluorescence staining. REF52 rat fibroblasts were cultured in Dulbecco modified Eagle medium supplemented with 10% FCS and antibiotics. Cells were trypsinized and electroporated with 15 µg of each expression plasmid. The transfected cells were seeded onto chamber slides (Nalge Nunc, Naperville, Ill.). After 48 h, cells were either untreated or treated with 1 µM dexamethasone (DEX) for 1 h, fixed, and stained with a rabbit antiserum raised against the Escherichia coli-expressed B1 and/or a hamster antiserum raised against the E. coli-expressed 2 as described previously (20, 45). Jurkat T cells were electroporated with 15 µg of each expression plasmid and after 48 h either untreated or treated with 1 µM DEX for 1 h. The cells were then cytospun, fixed, and stained.

Proliferation assay of ME-1 cells. ME-1 cells (40) were grown in RPMI 1640 supplemented with 10% FCS and antibiotics. Proliferation assays were performed with synthesized sense or antisense oligonucleotides as previously described (29). The sequence of the sense oligonucleotide corresponding to the junctional region of -MYH11 (18) is GAAATGGAGGTCCATGAG. The sequence of the corresponding antisense oligonucleotide is CTCATGGACCTCCATTTC.

RESULTS

Top Previous Next References

-SMMHC is required for maintenance of the proliferative state of ME-1 leukemic cells with inv(16). The leukemic cell line ME-1 (40) was established from a leukemic patient with inv(16), and it was from this cell line that -MYH11 (encoding -SMMHC) was originally cloned (18). We examined whether the expression of -SMMHC is essential for the maintenance of the proliferative state of this cell line. We incubated ME-1 cells with an antisense oligonucleotide complementary to the junctional sequence of the fusion transcript. As shown in Fig. 2, the antisense oligonucleotide, but not the corresponding sense oligonucleotide, blocked the proliferation of ME-1 cells. In contrast, neither the antisense nor the sense oligonucleotide blocked proliferation of HL-60 leukemic cells, which do not express -MYH11. After the antisense oligonucleotide treatment, ME-1 cells showed an increase in the population of differentiated CD13-positive cells (data not shown). The results suggest that the fusion product, -SMMHC, may act dominantly over the normal  subunit and prevent myeloid cells from entering into terminal differentiation, thereby supporting their continued proliferation.

View larger version (22K): [in this window] [in a new window]   FIG. 2.   Expression of -SMMHC is necessary for proliferation of ME-1 leukemic cells with inv(16). ME-1 cells (left) or HL-60 cells (right) were grown in the absence or presence of the sense or antisense oligonucleotides complementary to the junctional region of the -MYH11 transcript. The results represent three independent experiments.

The naturally occurring three isoforms of the  subunit function as transactivational cofactors of PEBP2. The result described above motivated us to investigate the function of -SMMHC at the molecular level. We first tried to address the function of -SMMHC and its normal counterpart, the  subunit, in terms of PEBP2-mediated transactivation. Using the M-CSF receptor promoter linked to the luciferase gene, we previously established a reporter assay system in Jurkat T cells, in which the function of the  subunit can be analyzed. This system requires both the  and the  subunits for significant transactivation, as we have shown by using 2 as the  subunit (14). Here we performed similar function assays for the 1 and 3 isoforms. Neither 1 nor 3 alone activated the promoter to any extent (Fig. 3A and B, lanes 5 to 8). Increasing amounts of AML1(453) [herein, we refer to full-length human and mouse proteins encoded by the PEBP2B/CBFA2/AML1 gene as AML1(453) and B1, respectively] only marginally transactivated the reporter activity when expressed alone (Fig. 3A and B, lanes 1 to 4). However, in the presence of 1 or 3, AML1(453) transactivated the promoter in a dose-dependent manner (Fig. 3A and B, lanes 9 to 12). Conversely, in the presence of a fixed amount of AML1(453), a wide range of 1 or 3 concentrations enhanced AML1(453)-dependent transactivation, suggesting that the effect of 1 or 3 was saturating above a certain concentration (Fig. 3A and B, lanes 13 to 17). Therefore, the naturally occurring isoforms of the  subunit, 1, 2, and 3, are all efficient for cooperative transactivation with the  subunit in vivo. Cooperation between the  subunit and the different  subunits was not confined to AML1(453) alone but was a general property of other members of the  subunit family, since similar results were also obtained using PEBP2A or PEBP2C as the  subunit (data not shown).

View larger version (27K): [in this window] [in a new window]   FIG. 3.   Cooperative activation of the M-CSF receptor promoter by the  and  subunits of PEBP2. Jurkat T cells were transfected with 4 µg of pM-CSF-R-luc and indicated amounts (micrograms) of pEF-AML1(453) and pEF-1 (A) or pEF-3 (B). Relative luciferase activities obtained with pM-CSF-R-luc and an empty plasmid were set at 100. For presentation, lanes 9, 13, and 16 are duplicated from lanes 7, 3, and 11, respectively.

-SMMHC does not support cooperative transactivation with full-length B1 or its deletion derivatives. When -SMMHC was used instead of the naturally occurring  subunit isoforms, -SMMHC did not cooperate with the  subunit (Fig. 4). To compare the effects of 2 and -SMMHC, we examined a series of deletion constructs of B1 (Fig. 1B) in the presence of 2 or -SMMHC. With 2, peak activities were obtained with B1(1-371) and B1(1-331) (Fig. 4, lanes 5 and 6), confirming that the transactivation domain lies between aa 291 and 371 (14). In contrast, -SMMHC did not lead to strong transactivation with any of the B1 deletion constructs, demonstrating a clear difference from 2. All of the data suggest that -SMMHC interferes with the  subunit function.

View larger version (23K): [in this window] [in a new window]   FIG. 4.   Effects of 2 and -SMMHC on transactivation with full-length B1 and its deletion derivatives. Jurkat cells were transfected with 4 µg of pM-CSF-R-luc, 1.5 µg of pEF-2 (lanes 1 to 9) or pEF--SMMHC (lanes 10 to 18), and 4.5 µg of the pEF-BOS-based expression plasmid for B1 or its deletion derivative as indicated. For lanes 1 and 10, 4.5 µg of an empty vector, pEF-BOS, was used. Relative luciferase activity obtained with cells transfected with 4 µg pM-CSF-R-luc and 6 µg of pEF-BOS was set at 100.

Minimal region of the  subunit required for - cooperative transactivation. -SMMHC consists of the amino-terminal 165 aa of the  subunit and the coiled-coil tail structure of SMMHC (18, 21) (Fig. 1A). To examine whether the inability of -SMMHC to cooperate with the  subunit is due to the lack of the carboxy-terminal sequences present in the normal isoforms of the  subunit, we constructed a series of carboxy-terminal deletion constructs of the  subunit as shown in Fig. 1A in expression plasmids and tested their activities. As shown in Fig. 5A and B, 135 enhanced AML1(453)-dependent transactivation, in contrast to 117, which did not exhibit any significant effect. Figure 5C shows a direct comparison of the transactivation abilities of 2, 165, 3, and 135. The results show that these constructs are equally functional, indicating that the amino-terminal 135-aa region in the  subunit is the minimal region required for - cooperation. Therefore, the inability of -SMMHC to cooperate with the  subunit cannot be due to the absence of the cooperation domain with the  subunit. Rather, the results seem to suggest that the presence of the SMMHC region prevents the -subunit region from properly cooperating with the  subunit.

View larger version (16K): [in this window] [in a new window]   FIG. 5.   Comparison between the natural isoforms and deletion derivatives of the  subunit with respect to their capacity to cooperate with the  subunit in transactivation. (A and B) Jurkat cells were transfected with 4 µg of pM-CSF-R-luc and indicated amounts (micrograms) of pEF-AML1(453) and pEF-135 (A) or pEF-117 (B). As in Fig. 3, lanes 9, 13, and 16 are duplicated from lanes 7, 3, and 11, respectively, for presentation. (C) Jurkat cells were transfected with 4 µg of pM-CSF-R-luc, 4 µg of pEF-AML1(453), and 2 µg of pEF-2 (lane 3), pEF-165 (lane 4), pEF-3 (lane 5), or pEF-135 (lane 6) as indicated. Relative luciferase activities obtained with pM-CSF-R-luc and an empty plasmid (pEF-BOS) were set at 100, and the results represent three independent experiments.

Coexpression of -SMMHC inhibits normal - cooperative transactivation. There exist at least two possible mechanisms to explain the observed negative effect of the SMMHC region. One is that -SMMHC may not be able to interact with the  subunit because of steric inhibition by the SMMHC region. Another possibility is that complexes of the  subunit and -SMMHC, if formed, may not be able to transactivate, which seems more likely based on already published observations (2, 4). In the former case, -SMMHC should not interfere with the normal - interaction, predicting that coexpression of the -SMMHC should not affect - cooperative transactivation. In contrast, in the latter case, -SMMHC would compete with the normal  subunit for heterodimerization with the  subunit, thus inhibiting transactivation. To test these possibilities, we examined the effect of -SMMHC on cooperative transactivation by B1(1-331) and 2. B1(1-331) was used as the  subunit because it exhibited higher activity than full-length B1 in the presence of 2 (Fig. 4), but similar results were obtained with full-length B1. As shown in Fig. 6A, -SMMHC inhibited transactivation by B1(1-331) and 2, suggesting that -SMMHC competed with 2 for interaction with B1(1-331), a result that favors the latter possibility. In contrast, 165, which contains the exact -subunit region of -SMMHC and is capable of dimerizing with the  subunit, did not inhibit transactivation (lane 6). This is because the  subunit can cooperate with the functional  subunit regardless of whether it is 2 or 165 (Fig. 5C). Furthermore, we examined the effect of -SMMHC on another mode of transactivation. The M-CSF receptor promoter contains a binding site for CCAAT/enhancer binding protein  (C/EBP) in close vicinity to the PEBP2 binding site, and C/EBP and PEBP2 synergistically activate the promoter (44). We have found that the transactivation domain of B1 (aa 291 to 371) is dispensable for synergistic transactivation and that the region of B1 from aa 1 to 291 suffices for synergy with C/EBP (15). B1(1-291) plus 2, which showed only moderate transactivation (Fig. 4, lane 7), exhibited a remarkable increase in transactivation activity in the presence of C/EBP (Fig. 6B, lane 3). In this transactivation mediated by synergistic cooperation between B1(1-291) and C/EBP, the presence of the  subunit is absolutely required. -SMMHC not only failed to substitute for the  subunit for the function (lane 4) but also inhibited the effect of 2 when coexpressed (lane 5). In contrast, 165 exhibited cooperativity similar to that of 2 and did not inhibit the effect of 2 (lanes 8 and 9). Again, this is probably because of cooperation of the  subunit with the functional  subunit regardless of whether it is 2 or 165. On the other hand, 117 neither exhibited the same cooperativity as 2 nor inhibited the effect of 2 (lanes 6 and 7), consistent with the fact that 117 does not dimerize with the  subunit. This makes a clear distinction with -SMMHC, which dimerizes with the  subunit and inhibits transcription.

View larger version (25K): [in this window] [in a new window]   FIG. 6.   Comparison of effects of -SMMHC and 165 on two modes of transactivation achieved by B1 deletion derivatives and 2. (A) Jurkat cells were transfected with 3 µg of pM-CSF-R-luc and different combinations of expression plasmids as indicated [pEF-B1(1-331), 4.2 µg; pEF-2, 1.4 µg; pEF--SMMHC, 1.4 µg; and pEF-165, 1.4 µg]. The results represent two independent experiments, in each of which the relative luciferase activity obtained with 3 µg of pM-CSF-R-luc and 7 µg of pEF-BOS was set at 100. (B) Jurkat cells were transfected with 3 µg of pM-CSF-R-luc and different combinations of expression plasmids as indicated [pMSV-C/EBP, 1 µg; pEF-B1(1-291), 3.6 µg; pEF-2, 1.2 µg; pEF--SMMHC, 1.2 µg; pEF-117, 1.2 µg; and pEF-165, 1.2 µg]. The results represent two independent experiments.

-SMMHC sequesters the  subunit in the cytoplasm. The next question was why dimers composed of the  subunit and -SMMHC cannot transactivate. We previously observed that the  subunit and -SMMHC localized both in the cytoplasm and in the nucleus when expressed together (20). When expressed separately, the  subunit was exclusively nuclear whereas -SMMHC was cytoplasmic. Thus, it seems that coexpression of the -SMMHC affected subcellular localization of the  subunit, but no direct experimental evidence for this hypothesis was provided (20). To examine this hypothesis more clearly, we established a system in which translocation of one factor can be controlled by an addition of a ligand. We constructed an expression plasmid coding for the fusion protein B1-GRLBD. The intracellular localization of B1-GRLBD can be controlled by DEX treatment (Fig. 7A). While parental B1 by itself is localized exclusively to the nucleus (Fig. 7A, ii), B1-GRLBD was localized to the cytoplasm in the absence of DEX treatment, because it was tethered there by GRLBD (Fig. 7A, iii). With DEX treatment, B1-GRLBD translocated into the nucleus (Fig. 7A, iv). When B1-GRLBD and 2 were coexpressed, both were localized to the cytoplasm in the absence of DEX treatment (Fig. 7B, left panels). With DEX treatment, B1-GRLBD was mostly translocated into the nucleus along with a small fraction of 2 (Fig. 7B, right panels). It is of note that most of the  subunit remained in the cytoplasm. When B1-GRLBD and -SMMHC were coexpressed, both were localized to the cytoplasm without DEX treatment, as expected (Fig. 7C, left panels). Interestingly, with DEX treatment, the majority of B1-GRLBD still remained in the cytoplasm together with -SMMHC (Fig. 7C, right panels; see the cell in which both were expressed). In this experiment, DEX treatment was sufficient to induce ligand-dependent nuclear translocation, because in other cells where only B1-GRLBD was expressed, B1-GRLBD completely translocated it into the nucleus upon DEX treatment. These results show that nuclear translocation of B1-GRLBD upon DEX treatment was blocked by -SMMHC but not by wild-type 2. The same was found for Jurkat T cells, where DEX-dependent nuclear translocation of B1-GRLBD was blocked by -SMMHC (Fig. 7D).

View larger version (57K): [in this window] [in a new window]   View larger version (71K): [in this window] [in a new window]   View larger version (54K): [in this window] [in a new window]   FIG. 7.   DEX-dependent nuclear translocation of B1 is blocked by -SMMHC. (A) Schematic illustration of the B1-GRLBD construct and effect of DEX on the subcellular localization of B1-GRLBD. REF52 cells were transfected with pEF-B1 (ii) or pME-B1-GRLBD (iii and iv) and after 48 h either untreated (ii and iii) or treated with 1 µM DEX for 1 h (iv). Cells were stained with a rabbit anti-B1 antiserum and a fluorescein isothiocyanate (FITC)-conjugated goat anti-rabbit antiserum. (B) REF52 cells were transfected with expression plasmids for B1-GRLBD (pME-B1-GRLBD) and 2 (pEF-2) and either untreated (left panels) or DEX treated (right panels). Cells were double stained with a rabbit anti-B1 antiserum followed by an FITC-conjugated goat anti-rabbit antiserum and with a hamster anti-2 antiserum followed by a rhodamine-conjugated goat anti-hamster antiserum. Signals from FITC were detected by using a 495-nm-wavelength filter, and those from rhodamine were detected by using a 555-nm-wavelength filter; upper panels represent FITC signals (B1-GRLBD), and lower panels represent rhodamine signals (2). (C) REF52 cells were transfected with expression plasmids for B1-GRLBD (pME-B1-GRLBD) and -SMMHC (pEF--SMMHC) and either untreated (left panels) or DEX treated (right panels). For double-stained cells, upper panels show FITC signals (B1-GRLBD) and lower panels show rhodamine signals (-SMMHC). Note that some cells expressed B1-GRLBD but not -SMMHC. (D) Jurkat T cells were transfected with pME-B1-GRLBD alone (i and ii) or pME-B1-GRLBD plus pEF--SMMHC (iii to vi) and either untreated (left panels) or DEX-treated (right panels). Panels iii and iv and panels v and vi show double staining of the same cells. i, ii, iii, and v, FITC signals (B1-GRLBD); iv and vi, rhodamine signals (-SMMHC).

DISCUSSION

Top Previous Next References

A system for in vivo functional analysis of the  subunit has revealed that 1, 2, and 3 isoforms are all transcriptionally functional. Previously the function of the  subunit was studied mostly by in vitro EMSA (13, 26, 38). In the present study, the M-CSF receptor promoter-luciferase reporter was used to assay -subunit function in Jurkat T cells, and the results showed that the three naturally occurring isoforms of the  subunit, 1, 2, and 3, are all effective in supporting - cooperative transactivation. Previously we and others observed prominent band supershifts of -subunit-DNA complexes with 1 and 2 but not with 3 in EMSA (13, 26, 38). However, 3 did have an effect of intensifying DNA binding of the  subunit. Thus, - dimers may be stable enough for enhanced transactivation in vivo but not stable enough to produce a supershift in EMSA.

Properties of -SMMHC. We have found that -SMMHC does not cooperate with the  subunit for transactivation despite the presence of the minimal region for -subunit function. Rather, -SMMHC seems to inhibit normal - cooperative transactivation. The results suggest that -SMMHC can efficiently dimerize with the  subunit, but that SMMHC region actively participates in undermining the proper functioning of these complexes.

How then does the carboxyl-terminal SMMHC region interfere with the potential harbored in the -subunit region (identical to 165)? One possible mechanism is the sequestration of the otherwise nucleus-located  subunit in the cytoplasm. We showed that -SMMHC blocked nuclear translocation of B1-GRLBD upon DEX treatment, whereas 2 did not. With this GRLBD fusion system, the majority of the --SMMHC complexes were retained in the cytoplasm. As such, the --SMMHC complexes can no longer participate in DNA binding and transactivation. This constitutes a form of dominant negative inhibition. It is important to note that colocalization of -SMMHC and the  subunit was observed both in this and the previous (

Effect of -SMMHC on proliferation. The present study suggests that -SMMHC plays a key role in the maintenance of the proliferative state of ME-1 cells which have inv(16). In contrast, Cao et al. recently reported that -SMMHC blocks proliferation of 32D myeloid cells at the G₁/S transition (2). Further studies will be needed to address this apparent discrepancy between the two different systems. Nonetheless, it is interesting that proliferation of Kasumi-1 cells, in which AML1/ETO is produced as a result of the t(8;21) translocation, was specifically blocked by an antisense oligonucleotide complementary to the junction region of the AML1/ETO mRNA (29). Thus, -SMMHC and AML1/ETO are analogous in some respects: both are dominant negative inhibitors in developing embryos (4, 41), and both are essential for the proliferative properties of leukemic cells that have the corresponding chromosomal abnormalities.

Alternative interpretation of the  knockout and the -SMMHC knockin studies. Previously, the 3 isoform was thought to be inactive as discussed above. Accordingly, the  knockout mice were produced via targeting of exon 5, leaving the 3 isoform to be expressed normally or increased in a compensatory fashion (30, 38). These mice exhibited nearly the same phenotypes as the mice that were targeted in exon 1, which represents the authentic knockout (25). Together, these observations suggest that 3 is inactive in vivo. In fact, however, alternative splicing with the skipping of exon 5 in the former group of mice did not occur so as to quantitatively compensate for the loss of 1 and 2 (reference 38; our unpublished observation [15] on mice generated by Sasaki et al. [30]). Thus the observed phenotypes may be attributed to a dosage effect but not to the disabled function of 3 per se.