The Core Binding Factor (CBF) α Interaction Domain and the Smooth Muscle Myosin Heavy Chain (SMMHC) Segment of CBFβ-SMMHC Are Both Required to Slow Cell Proliferation*

We have expressed several variants of core binding factor β (CBFβ)-smooth muscle myosin heavy chain (SMMHC) from the metallothionein promoter in Ba/F3 cells. Deletion of amino acids 2–11 from the CBFβ segment, required for interaction with CBFα, prevented CBFβ-SMMHC from inhibiting CBF DNA binding and cell cycle progression. Deletion of 283 carboxyl-terminal residues from the SMMHC domain, required for multimerization, also inactivated CBFβ-SMMHC. Nuclear expression of CBFβ(Δ2–11)-SMMHC was decreased relative to CBFβ-SMMHC. CBFβ(Δ2–11)-SMMHC linked to a nuclear localization signal still did not slow cell growth. The ability of each CBFβ-SMMHC variant to inhibit CBF DNA binding and cell proliferation correlated with its ability to inhibit transactivation by an AML1-VP16 fusion protein. Thus, CBFβ-SMMHC slows cell cycle progression from G1 to S phase by inhibiting CBF DNA binding and transactivation.

CBF activates the expression of several lymphoid and myeloid genes, suggesting that lack of differentiation accounts for the phenotypes of CBF null mice (27)(28)(29)(30). We expressed CBF␤-SMMHC from the zinc-responsive MT promoter in 32D cl3 myeloid and Ba/F3 B-lymphoid cells (14). Induction of CBF␤-SMMHC resulted in decreased CBF DNA binding and slowed proliferation during G 1 phase. The differentiation of 32D cl3 cells in response to granulocyte colony-stimulating factor was unaffected. We proposed that initial genetic alterations occur during leukemogenesis that bypass the growth inhibitory effect of CBF␤-SMMHC, potentiating inhibition of differentiation.
The tail domain of SMMHC is ␣-helical and consists of multiple, related 28-amino acid regions. One face of the helix is hydrophobic, allowing dimerization, and the other face has alternating positively and negatively charged zones and allows multimerization, which occurs with a 98-amino acid (3.5 repeat) stagger. In addition, SMMHC has a non-helical C terminus required for optimal multimerization (35). Human SMMHC has two isoforms (SMMHC 204 and SMMHC 200 ) that differ in the lengths of their non-helical C termini as a result of alternative splicing (36). Herein we refer to CBF␤-SMMHC 200 as INV and CBF␤-SMMHC 204 as INVa. INVa is more abundant than INV in M4Eo AML (37).
We demonstrate that INV and INVa have identical effects on Ba/F3 cells. Variants of INV lacking two-thirds or all of the SMMHC segment, and so incapable of oligomerization, did not affect cell growth. Also, the SMMHC segment, in isolation or linked to CBF␤(⌬2-11), did not inhibit cell proliferation, even when directed to the nucleus. In addition, the ability of each INV mutant to inhibit CBF DNA binding and cell proliferation correlated with its ability to inhibit transactivation.

EXPERIMENTAL PROCEDURES
Cell Culture and Transfection-Ba/F3 cells (38) were maintained in RPMI 1640 medium with 10% heat-inactivated fetal calf serum and 1 ng/ml murine interleukin-3 (R&D Systems). CV-1 cells were maintained in Dulbecco's modified Eagle's medium with 10% calf serum. All cultures contained penicillin/streptomycin. 100 M zinc chloride was added when indicated. Ba/F3 cells were stably transfected by electroporation as described (14). Single-cell clones were isolated by limiting dilution. Viable cell numbers were determined by enumerating cells that excluded trypan blue dye using a hemocytometer. Incorporation of tritiated thymidine to assess proliferation rate was carried out as described (14). Transient transfection of CV-1 cells and luciferase assays were carried out as described (39).
Western Blot, Gel Shift, and Immunofluorescence Analyses-Total cellular extracts were subjected to polyacrylamide gel electrophoresis and Western blotting using CBF␤ antiserum as described (14). Gel shift analysis and indirect immunofluorescence were carried out as described (14).
Cross-linking Analysis-cDNAs in pGEM/CMV or pBS were transcribed and translated using the TNT reticulocyte lysate kit (Promega) in the presence of [ 35 S]methionine (Amersham Pharmacia Biotech) following the manufacturer's instructions. 3 l of each extract was then diluted to 15 l with phosphate-buffered saline and exposed to 0.0025% glutaraldehyde at room temperature for 1 h. The reactions were stopped by adding glycine to 192 mM and Tris (pH 6.8) to 25 mM. Samples were then subjected to SDS-polyacrylamide gel electrophoresis and autoradiography. Fig. 1. INV(⌬C283) has a deletion of 283 residues from the 440-amino acid SMMHC segment. INV(␤⌬2-11) and INVa(␤⌬2-11) lack amino acids 2-11 of the CBF␤ segment. Deletion of residues 2-11 from INV prevents heterodimerization with AML1B in vitro (16). CBF␤(165) contains the 165 CBF␤ residues found in INV. Additional variants are described below.

CBF␤ and SMMHC Domains Are Required for INV Activities-The variants of INV investigated herein are diagrammed in
To verify that the intact SMMHC domain can mediate homodimerization and to determine whether INV(⌬C283) has lost this activity, we expressed several INV variants and AML1B in reticulocyte lysates and subjected them to glutaraldehyde cross-linking (Fig. 2). INVa, INV, and INVa(␤⌬2-11) contain intact SMMHC domains and dimerized efficiently. CBF␤, INV(⌬C283), and AML1B did not form dimers.
Clonal Ba/F3 cell lines expressing INV(⌬C283), INV(␤⌬2-11), or CBF␤(165) from the MT promoter were obtained by electroporation and G418 selection. Ba/F3 cells were chosen in lieu of 32D cl3 cells for these experiments as it had been much easier to obtain MTINV lines with Ba/F3 cells (14), perhaps because the MT promoter is leakier in 32D cl3 cells. In addition, the effects of INV on CBF DNA binding and cell cycle progression were identical in Ba/F3 and 32D cl3 cells (14). The MT promoter offers the advantage over other inducible systems of higher level expression, and some leakiness is tolerable when expressing a dominant-interfering protein. Expression of these three INV variants Ϯ zinc was assessed by Western blotting (Fig. 3A) The effect of each of these INV variants on DNA binding by endogenous CBF was assessed by gel shift analysis using a strong CBF-binding site present in the myeloperoxidase gene (Fig. 3B). Zinc did not affect CBF DNA binding in vectortransfected CB6 cells, but induction of INV in the INV-3 cells reduced CBF DNA binding severalfold as described (14). The specificity of the observed gel shift complex was verified by competition with unlabeled wild-type oligonucleotide or a mutant oligonucleotide that does not bind CBF (29). Induction of INV(⌬C283), INV(␤⌬2-11), or CBF␤(165) did not affect DNA binding by CBF. Equivalence between paired extracts was verified by gel shift assay with a USF-binding site.
We also expressed INV(␤141) in two Ba/F3 lines (Fig. 4, left  panel). This protein contains the minimal CBF␤ domain required for strongly binding CBF␣ subunits and the entire SMMHC segment present in INV. Induction of INV(␤141) reduced CBF DNA binding (Fig. 4, right panel).
The effect of these variants on the proliferation of Ba/F3 cells was assessed by viable cell counting and by tritiated thymidine incorporation (Fig. 5). INV slowed cell proliferation as described (14), whereas INV(⌬C283), INV(␤⌬2-11), or CBF␤(165) did not. INV(␤141) slowed proliferation as effectively as INV. Results are shown for days 2 and 3, as differences were less evident during day 1. These results indicate that both the CBF␤ and SMMHC domains of INV are required to inhibit CBF DNA binding and cell proliferation and suggest that these activities require heterodimerization via the CBF␤ domain and homodimerization via the SMMHC domain.
Expression of the SMMHC Domain in the Nucleus Does Not Slow Cell Proliferation-The subcellular localization of INV(␤⌬2-11) and of several other INV variants was assessed by indirect immunofluorescence (Fig. 6). INV was detected prominently in the nucleus of Ba/F3 cells in a speckled, rod-like pattern (upper right panel) as described (14). INV(␤⌬2-11) was detected most prominently in the cytoplasm (second row), as were INV(⌬C283) and CBF␤(165) (third row).
To verify that nuclear expression of the SMMHC domain does not slow cell growth, we linked the SV40 T-antigen nuclear localization signal to CBF␤(⌬2-11) in the context of both SMMHC isoforms. These proteins, NLS-INV(␤⌬2-11) and NLS-INVa(␤⌬2-11), were expressed from the MT promoter in Ba/F3 cells. At the same time, we prepared lines expressing

FIG. 4. Deletion of amino acids 142-165 from the CBF␤ segment of INV does not affect inhibition of CBF DNA binding. Two
Ba/F3 subclones expressing INV(␤141) from the MT promoter were cultured with (ϩ) or without (Ϫ) zinc for 6 h. Total cellular proteins were then prepared and subjected to Western blot analysis using CBF␤ antiserum (left panel). These extracts were then subjected to gel shift analysis with a CBF-binding site and with a USF-binding site (right panels).
INVa and HA-SMMHC. The expression of these proteins in Ba/F3 cells treated with zinc was confirmed by Western blotting with CBF␤ antiserum (Fig. 7). HA-SMMHC retains 24 CBF␤ residues, and so could be detected with this antiserum raised against glutathione S-transferase-CBF␤ (14).
NLS-INV(␤⌬2-11) was detected most abundantly in the nucleus (Fig. 6, lower left panel), although despite numerous attempts, we never achieved a speckled pattern similar to that detected with INV. Western blot analysis of nuclear and cytoplasmic fractions confirmed that the NLS directed greater than two-thirds of INV(␤⌬2-11) to the nucleus. 3 Strikingly, indirect immunofluorescence detected HA-SMMHC as large bright spots both in the nucleus and cytoplasm, suggesting that the CBF␤ domain of INV limits the extent of its multimerization (Fig. 6, lower right panel).
The ability of these INV isoforms to inhibit CBF DNA binding and to slow cell proliferation was then assessed (Fig. 7B and Fig. 8). INVa inhibited CBF DNA binding and slowed cell proliferation as effectively as INV, whereas the two NLS-INV(␤⌬2-11) variants were ineffective. HA-SMMHC did not reduce CBF DNA binding upon induction with zinc, relative to USF DNA binding, which was reduced in the zinc-containing extract, and which did not slow cell growth. These results support the conclusion that interaction with CBF␣ subunits is the mechanism whereby INV slows cell proliferation.

Inhibition of Transactivation by INV Variants
Correlates with Their Inhibition of Proliferation-To assess the affect of INV variants on transactivation, we employed a reporter containing four CBF-binding sites, derived from the myeloperoxidase gene (29). An internal control was not utilized, as we observed specific inhibition of several viral promoters by INV. 3 AML1B activated p(CBF) 4 TKLUC 6-fold, and coexpression of INV reduced this activation 3-fold, on average. However, these modest effects did not allow us to identify differences between the INV variants. 3 To develop a more reliable assay, we employed a protein containing the AML1B DNA-binding domain and the VP16 transactivating domain. AML1-VP16 did not activate pTK-LUC, but activated p(CBF) 4 TKLUC 15-fold (Fig. 9) VP16 activation by INV and these six variants approximately paralleled their ability to inhibit cell proliferation, suggesting that inhibition of CBF transactivation is the mechanism whereby INV slows cell cycle progression. DISCUSSION By expressing INV from the MT promoter in Ba/F3 and 32D cl3 cells, we provide evidence suggesting that lack of hematopoiesis in INV-expressing mice is due to inhibition of cell proliferation during G 1 phase (14). We have now extended these observations by identifying the domains of INV required for inhibition of CBF DNA binding and transactivation and for inhibition of cell proliferation. Deletion of just 10 amino acids from the CBF␤ segment, required for interaction with CBF␣ subunits, prevented each of these activities. Although the resulting protein, INV(␤⌬2-11), might still interact with other cellular proteins, either via the CBF␤ or SMMHC segment, these interactions were not sufficient to strongly interfere with CBF activities or to slow proliferation. Consistent with this finding, neither the CBF␤ segment nor the SMMHC segment alone was active in these assays. On the other hand, deletion of CBF␤ amino acids 142-165 from INV, which leaves its CBF␣ interaction domain intact, did not prevent its inhibition of CBF DNA binding, transactivation, or proliferation.
INV was expressed most abundantly in the nucleus, whereas INV(␤⌬2-11) was localized more extensively to the cytoplasm. INV has previously been shown to localize predominantly to the cytoplasm of fibroblastic cells, co-localizing with the cytoskeleton (16,41). Perhaps increased nuclear expression of INV in hematopoietic cells results from their having less cytoskeletal structure and higher levels of CBF␣ subunits. Using Western blotting, we also detected INV predominantly in the nuclei of M4Eo patient samples (37) and in ME-1 cells, 3 which derive from an M4Eo AML (42). We sought to determine whether expressing INV(␤⌬2-11) more abundantly in the nucleus allows either its CBF␤(⌬2-11) or SMMHC segment to interact with nuclear proteins and so interfere with cell proliferation. We found that addition of a nuclear localization signal to its N terminus directed INV(␤⌬2-11) to the nucleus, but did not result in reduced CBF DNA binding or cell growth rate.
We expressed a construct, INV(⌬C283), lacking two-thirds of the SMMHC segment. This protein did not dimerize in vitro or interfere with CBF activities or cell proliferation in vivo. In the future, we intend to more precisely identify the regions of the SMMHC domain required for the activities of INV. Such investigations might eventually lead to the identification of relevant SMMHC-interacting proteins. Herein we demonstrate that both INV and INVa inhibit CBF activities and cell proliferation equally, indicating that their different non-helical C-terminal tails (8 or 34 amino acids) are interchangeable. Interestingly, HA-SMMHC appeared to form much larger intracellular aggregates than INV, suggesting that the CBF␤ segment interferes with, but does not prevent, multimerization via the SMMHC domain.
INV/CBF␣ heterodimers are capable of interacting with CBF-binding sites. INV inhibits activation of the myeloid NP-3 promoter by AML1B (43), indicating that these heterodimers are not transactivating, and we reached the same conclusion using the p(CBF) 4 TKLUC reporter. Although this assay is artificial, our results suggest that interaction with CBF␣ subunits and integrity of the SMMHC segment are both required for INV to inhibit CBF-mediated transactivation. INV might inhibit CBF transactivation by sequestering AML1B into multimeric structures away from chromatin. Alternatively, INV/ CBF␣ dimers bound to relevant genes might directly interfere with transcription. Which of these mechanisms predominates is unknown, but our recent finding that INV can redirect AML1B to the cytoplasm in NIH 3T3 cells supports the importance of the sequestration model (16).
Phenotypic differences between M2 AML associated with t(8;21) and M4Eo AML associated with inv(16) might in part result from AML1-ETO being a more potent inhibitor of CBF transactivation than INV. AML1-ETO slowed 32D cl3 cell myeloid differentiation, whereas INV did not (14,43,44), suggesting that cell proliferation is more sensitive than differentiation to inhibition of CBF activities. In the context of additional genetic hits, inhibition of differentiation might become more evident. Also, we recently found that expression of INV or AML1-ETO in Ba/F3 cells reduces induction of p53 when the cells are treated with DNA-damaging agents (45). This effect was not observed with cells expressing INV(␤⌬2-11). Thus, inhibition of CBF activities might prevent apoptosis and encourage leukemic progression in premalignant cells. Finally, although we have demonstrated that inhibition of CBF activities accounts for inhibition of cell proliferation by CBF␤-SMMHC, interactions of the CBF␤ or SMMHC segment with other proteins might contribute to transformation and to the phenotype of M4Eo AML.