Induction of apoptosis by the c-Myc helix-loop-helix/leucine zipper domain in mouse 3T3-L1 fibroblasts.

The cellular proto-oncogene c-myc is involved in cell proliferation and transformation but is also implicated in the induction of programmed cell death (apoptosis). The c-Myc protein is a transcriptional activator with a carboxyl-terminal basic region/helix-loop-helix (HLH)/leucine zipper (LZ) domain. It forms heterodimers with the HLH/LZ protein Max and transactivates gene expression after binding DNA E-box elements. We have studied the phenotype of dominant-negative mutants of c-Myc and Max in microinjection experiments. Max mutants with a deleted or mutated basic region inhibited DNA synthesis in serum-stimulated 3T3-L1 mouse fibroblasts. In contrast, mutants of c-Myc expressing only the basic region/HLH/LZ or HLH/LZ domains rapidly induced apoptosis at low and high serum levels. Co-expression of the HLH/LZ domains of c-Myc and Max failed to do so. We suggest that the c-Myc HLH/LZ domain induces apoptosis by specific interaction with cellular factors different to Max.

Programmed cell death is an intrinsic death program operating to eliminate unwanted cells during normal development. It also has been suggested to kill cells after aquirement of growth factor-independent growth properties due to genetic alterations (Ellis et al., 1991;Evan and Littlewood, 1993). Common morphological features of programmed cell death are blebbing of the cytoplasmic membrane, chromatin condensation, and breaking of the dead cell into apoptotic bodies (Wyllie, 1980(Wyllie, , 1987. This kind of cell death, often termed apoptosis, does not elicit an inflammatory response in the tissue and can therefore clearly be distinguished from cell necrosis in which cells die as the result of acute injury (Kerr et al., 1972).
The apoptotic program appears to be installed in all animal cells and can operate in the presence of inhibitors of RNA and protein synthesis (Ellis et al., 1991). In some cellular systems it is even induced by these inhibitors, indicating that short-lived proteins or RNAs may negatively control the apoptotic machinery. Jacobson et al. (1994) recently proposed a model in which the process leading to apoptosis is divided into three phases: (i) an activation phase in which control systems of apoptosis are activated or derepressed (this phase can be sensitive to inhibitors of RNA and protein synthesis), (ii) an effector phase in which the activated control system acts on multiple targets in the cell, and (iii) a degradation phase in which the dying cell is broken down. In the latter two phases, inhibitors of RNA and protein synthesis are not effective to block the apoptotic program.
The cellular proto-oncogene c-myc has been implicated in the control of proliferation and apoptosis. Expression of c-myc is tightly linked to mitogenic stimuli and is a prerequisite for cell growth (for reviews, see Lü scher and Eisenman (1990) and Marcu et al., 1992). Moreover, post-translational activation of a c-Myc estrogen receptor chimera in resting cells is sufficient to induce entry into the cell cycle (Eilers et al., 1989(Eilers et al., , 1991. Expression of an exogenous c-myc gene renders hematopoietic cells and fibroblasts unable to exit from the cell cycle upon withdrawal of growth factors or serum. Instead, these cells continue cycling and concomitantly undergo apoptosis (Askew et al., 1991;Evan et al., 1992;. Expression of c-myc is also required for activation-induced apoptosis of T-cell hybridomas (Shi et al., 1992). These observations suggest a model in which proliferation and cell death are processes that are co-induced by c-Myc and which are subsequently modulated by cytokine action (Evan and Littlewood, 1993;Harrington et al., 1994).
The c-Myc protein has features of a transcription factor with a transcriptional activation domain in the amino-terminal region (Kato et al., 1990). The carboxyl-terminal region contains a basic region (BR), 1 helix-loop-helix (HLH), and leucine zipper (LZ) domain (Murre et al., 1989;Landschulz et al., 1988) in a contiguous array essential for specific DNA binding (BR) and dimerization (HLH/LZ) of c-Myc with the BR/HLH/LZ protein Max (Blackwood and Eisenman, 1991;Blackwood et al., 1992;Kato et al., 1992). c-Myc⅐Max heterodimers and Max⅐Max homodimers bind specifically to the E-box motif CACGTG (Blackwell et al., 1990). Homodimers of c-Myc are not found in vivo. Since Max lacks a transcriptional activation domain, c-Myc⅐Max heterodimers have been suggested to act as transcriptional activator and Max⅐Max homodimers as repressors (Kretzner et al., 1992;Amati et al., 1992). The biological functions of c-Myc reported so far including cell transformation (Stone et al., 1987, Amati et al., 1993a, transcriptional activation (Kretzner et al., 1992;Amati et al., 1993b), and induction of proliferation and apoptosis in quiescent cells Amati et al., 1993b) require dimerization of c-Myc with Max and sequence-specific binding of the heterodimer to DNA. Hopewell and Ziff (1995) recently reported proliferation of the nerve growth factor responsive PC12 cell line in the absence of a functional Max. Whether c-Myc can induce apoptosis in a Max-independent manner in these cells, however, is not yet known.
Here we have studied the effect of dominant-negative mutants of c-Myc and Max on proliferation. Max mutants with a mutated or deleted basic region efficiently blocked serum-induced DNA synthesis. Unexpectedly, c-Myc mutants expressing only the BR/HLH/LZ or HLH/LZ domain rapidly induced apoptosis.
Plasmid DNAs-The plasmid pUHD10 -1 (Deuschle et al., 1989) containing a cytomegalovirus (CMV) enhancer/promoter was used to express different c-Myc and Max mutants. Upstream of the open reading frame, a T7-promoter and a sequence of a translation start site was inserted (pCMV-T7). A hemagglutinin epitope, which is recognized by the monoclonal antibody 12CA5 (Field et al., 1988), was added at the carboxyl terminus of the mutant Max proteins, MycHLH, and MycLZ. A poliovirus epitope, which is recognized by the monoclonal antibody C3 (Taylor et al., 1990) was added at the carboxyl terminus of ctMyc and ctMycBR Ϫ . All mutants were constructed by polymerase chain reactionaided cloning techniques and were subsequently controlled by sequence analysis. Positions in the c-Myc proteins refer to the smaller, 439-amino acid c-Myc protein (64 kDa) (Gazin et al., 1984), in the Max protein to the larger, 159-amino acid-long protein (22 kDa) (Blackwood and Eisenman, 1991). The nuclear localization signal (NLS) in the c-Myc mutants corresponds to the authentic NLS M1 of c-Myc, which maps upstream of the basic region (Dang and Lee, 1988). Synthesis of full-length mutant proteins was confirmed by in vitro transcription/translation experiments. The plasmid expressing the ␤-galactosidase gene under the control of a CMV enhancer/promoter was obtained from Clontech (Palo Alto, CA). Plasmid DNA was purified from bacterial cultures using a commercial purification kit (Qiagen, Diagen, Dü sseldorf, Germany).
Proliferation Assay and Microinjection-Mouse 3T3-L1 fibroblasts were grown on glass coverslips (Cellocate, Eppendorf, Hamburg) with a system of coordinates that facilitates the relocalization of microinjected cells. 1.5 ϫ 10 5 cells were seeded in 35-mm dishes containing a coverslip and allowed to attach for 24 h. For serum starvation, the semiconfluent monolayers were washed with medium without serum and incubated for 48 h in 4.5 ml of Dulbecco's modified Eagle's medium with low FCS (0.5%). Plasmid DNA (150 ng/l) in 10 mM Tris-HCl, pH 7.5, 1 mM EDTA, was loaded into injection needles pulled from 1.2-mm diameter glass capillaries (Clark Electro Medical Instruments, Reading, UK) using a capillary puller (Kopf Instruments, Tujunga, CA). Cells were microinjected using a semiautomatic microinjector 5242 (Eppendorf, Hamburg, Germany) and an inverse Axiovert-10 microscope (Zeiss, Oberkochen, Germany). Approximately 30% of microinjected cells did not survive the procedure of microinjection. Cell death immediately after microinjection was probably due to severe damage of the cell membrane. The damaged cells died within 5-10 min after microinjection. This cell death showed no characteristics of apoptosis. 30 min after microinjection, these dead cells were detached from the surface of the coverslip and did not interfere with the subsequent analysis of the apoptotic cells. The rate of cells damaged by microinjection was relatively constant and not reduced when TE or only vector DNA was injected. Microinjected cells were returned to the incubator within 30 min. 5 h after microinjection, cells were stimulated with 10% FCS, and 5-bromo-2Ј-deoxyuridine (BrdUrd) labeling solution (Cell-Proliferation Kit, Amersham Corp.) was added to each dish at a concentration of 1 l/ml. The incubation was continued for further 20 h before the cells were fixed.
Immunocytochemistry-Cells were washed in phosphate-buffered saline solution (PBS) and fixed in 4% (w/v) paraformaldehyde in PBS for 30 min at 4°C. The cells were then permeabilized by treatment with 0.2% Triton X-100 in PBS for 20 min at room temperature, washed, and blocked in a solution of PBS containing 10% FCS overnight at 4°C to reduce unspecific staining. Primary and secondary antibodies were appropriately diluted in PBS containing 10% FCS. Cells were incubated 45 min with primary and 30 min with the secondary antibodies at room temperature. Between and after these incubations, the cells were washed 3 times with PBS, each time for 5 min. As primary antibodies, the biotinylated monoclonal mouse antibody 12CA5 (hemagglutinin epitope) (Field et al., 1988), C3 (polio epitope) (Taylor et al., 1990), 9E10 (c-Myc) (Evan et al., 1985), and a monoclonal antibody directed against bacterial ␤-galactosidase (Boehringer, Mannheim) were used. Anti-IgG goat antibodies conjugated to rhodamine and streptavidine conjugated with aminomethylcoumarin (Dianova, Hamburg, Germany) were used for indirect staining. For detection of incorporated BrdUrd, the cells were treated with 2 M hydrochloric acid for 15 min at 37°C. After washing in PBS, the cells were incubated with a fluoresceine-coupled mouse monoclonal antibody directed against BrdUrd (Boehringer Mannheim) for 30 min at 37°C. After washing, the cells were analyzed using an Axioscop fluorescence microscope (Zeiss). By switching the respective filters, the rate of BrdUrd incorporation in cells expressing the different proteins was determined. Photographs were made with a 456070 camera and MC100-regulatory unit (Zeiss) and Kodachrome 400 films (Eastman Kodak Co.).
Electrophoretic Mobility Shift Assay (EMSA)-The CM1 probe (Blackwell et al., 1990) containing a c-Myc⅐Max binding site (boldface) was obtained by annealing two complementary synthetic oligonucleotides (5Ј-GATCCCCCCACCACGTGGTGCCTGA-3 and 3Ј-GGGGGTG-GTGCAC-CACGGACTCTAG-5Ј) and was labeled by a fill reaction using Klenow polymerase and [␣-32 P]dCTP. The binding reaction (20 l) with reticulocyte lysate, binding mix (10 mM Hepes, pH 7.9, 100 mM KCl, 1 mM EDTA, 1 mM dithiothreitol, 1 g of poly(dA-dT), 4% Ficoll), and 50,000 disintegrations/min of labeled CM1 probe was incubated at room temperature for 30 min. Samples were subjected to electrophoretic separation at room temperature on a 4.5% nondenaturing polyacrylamide gel in 1 ϫ TBE buffer (1 ϫ TBE ϭ 89 mM Tris borate; 2 mM EDTA, pH 8.0) at 250 V for 3 h. The gels were dried and exposed to Kodak x-ray films between intensifying screens. For competition experiments a CM1mut oligonucleotide was used where the central nucleotides CG were exchanged by GC.

Construction of Dominant-Negative Mutants of c-Myc and
Max-We have used the approach of dominant-negative mutants to interfere with the function of c-Myc in vivo. Two types of mutants were constructed to inhibit the formation of functional c-Myc⅐Max heterodimers in vivo (Fig. 1). Dominant-negative mutants of Max lacking the BR (MaxBR Ϫ ) or carrying point mutations in the BR (MaxBRmut) can still dimerize with c-Myc (Reddy et al., 1992;Billaud et al., 1993). Overexpression of the Max mutants under the control of the CMV enhancer/ promoter should disrupt c-Myc⅐Max and other c-Myc-or Maxcontaining complexes and shift the equilibrium from DNA binding complexes to non-DNA binding complexes. As a consequence, the c-Myc⅐Max-specific E-box motifs remain unoccupied.
A similar mode of action is expected for a dominant-negative c-Myc mutant lacking the transcriptional activation domain and BR (ctMycBR Ϫ ). Overexpression of ctMycBR Ϫ should sequester Max protein, which is then no longer available for dimerization with wild-type c-Myc. A second c-Myc mutant, ctMyc, differs from the previous mutant by having an intact BR. Heterodimers of ctMyc⅐Max have lost their transactivation potential but should be able to bind specifically to DNA.
To ensure nuclear transport of all mutant c-Myc proteins, the NLS M1 of c-Myc (Dang and Lee, 1988), which maps outside of the BR/HLH/LZ region, was cloned 3Ј of the LZ. The Max mutants contain their NLS at the authentic position in the carboxyl terminus. All mutant proteins were tagged with epitopes of viral proteins of influenza virus (MaxBR Ϫ , Max-BRmut, MycHLH, MycLZ) or polio virus (ctMycBR Ϫ , ctMyc) in order to detect them by indirect immunofluorescence staining. For unknown reasons, the polio epitope did not work satisfactorily. Instead, we used the antibody 9E10 (Evan et al., 1985), which recognizes the LZ domain of c-Myc. A survey of all mutants used in this work is shown in Fig. 1.
Functional Analysis of Dominant-Negative Mutants of c-Myc and Max in Gel-shift Experiments-The mutants were expressed after in vitro transcription/translation, and correctly sized proteins were demonstrated (data not shown). The prop-erties of the mutant proteins were examined in gelshift experiments. Two functions were tested: (i) formation of ctMyc⅐Max dimers and specific binding to an oligonucleotide that contains the E-box recognition motif and (ii) the ability of MaxBR Ϫ , MaxBRmut, ctMycBR Ϫ , MycHLH, and MycLZ mutants to compete for the formation of ctMyc⅐Max complexes.
Unprogrammed reticulocyte lysate extract (translation reaction without addition of exogenous RNA) already revealed a binding activity (Fig. 2A, lane 2, marked with an asterisk). This shift was competed by oligonucleotides CM1 containing the CACGTG motif (Blackwell et al., 1990) and CM1mut with an inversion of the middle CG of the E-box motif ( Fig. 2A, lanes 4  and 5), indicating that this activity is not specific for the CACGTG-binding motif. Additionally, this complex could not be disrupted or supershifted by the addition of antibodies directed against Max (Fig. 2B, lane 2) and c-Myc (data not shown) and therefore contains no Max⅐Max homodimers or c-Myc⅐Max heterodimers. The amount of this endogenous shift activity varied considerably depending on the batch of the reticulocyte lysate.
When ctMyc was added to the binding reaction, an additional shift appeared that most likely is produced by a heterodimer of ctMyc and endogenous Max, which is already present in the extract ( Fig. 2A, lane 3). This shift can be competed with the oligonucleotide CM1 but not with CM1mut ( Fig. 2A, lanes 4  and 5). We observed that the Max protein alone did not produce a specific shift ( Fig. 2A, lane 6). This was not unexpected since phosphorylation of Max at an amino-terminal casein kinase II (CKII) site in the reticulocyte lysate inhibits DNA-binding of Max⅐Max homodimers but not c-Myc⅐Max heterodimers (Berberich and Cole, 1992). We could confirm this observation by using the mutant MaxCKII Ϫ , which has replaced Ser-11 by Ala ( Fig. 2A, lanes 7-9). Mixing of ctMyc and Max generated a strong shift that can be explained by synergistic action of both proteins ( Fig. 2A, lanes 10 -12).
The specificity of the observed shifts was also analyzed by using antibodies. The ␣-Max antibodies reduced and/or supershifted ctMyc⅐Max complexes (Fig. 2B, lanes 1-4). Antibodies specific for the c-Myc LZ domain or the viral tag of Max only inhibited formation of ctMyc⅐Max specific shifts (lanes 6 and 7).
We next tested whether the presence of the mutant Max and c-Myc proteins in the binding reaction can compete for the formation of the ctMyc⅐Max shift. Addition of a 2-fold excess of MaxBR Ϫ , MaxBRmut, or ctMycBR Ϫ protein to the binding reaction resulted in a clear reduction of the ctMyc⅐Max-specific shift (Fig. 2C, lanes 2, 3, and 6), whereas addition of unprogrammed reticulocyte lysate (data not shown) or MycHLH and MycLZ had no effect (Fig. 2C, lanes 4 and 5). In summary, ctMycBR Ϫ , MaxBR Ϫ , and MaxBRmut were able to disrupt a ctMyc⅐Max complex in vitro and therefore act in a dominantnegative manner as expected.
Dominant-Negative Mutants of Max Block DNA Synthesis in Serum-stimulated Mouse 3T3-L1 Fibroblasts-To test whether dominant-negative mutants of c-Myc and Max could inhibit c-Myc function in vivo, we used a combined microinjection/ proliferation assay (Fig. 3a). Mouse 3T3-L1 fibroblasts were serum-starved for 48 h and microinjected with expression plasmids (Graessmann and Graessmann, 1983) coding for MaxBR Ϫ and MaxBRmut. Subsequently, cells were stimulated with 10% fetal calf serum, and BrdUrd was added to visualize DNA synthesis. The cells were fixed 20 h after serum stimulation and stained with antibodies directed against the viral epitope and BrdUrd. Stimulation of quiescent fibroblasts with serum consistently induced DNA synthesis in about 70% of the uninjected cells. This induction was almost completely blocked in cells expressing MaxBR Ϫ (Fig. 3b). A similar but less strong block was observed in cells expressing MaxBRmut.
Expression of MaxBR Ϫ and MaxBRmut differed in regard to their cellular distribution. MaxBRmut was almost exclusively demonstrable in the nucleus, whereas MaxBR Ϫ was also observed in the cytoplasm in a considerable portion of cells. As a control, an expression plasmid coding for ␤-galactosidase was injected. Expression of ␤-galactosidase had no significant effect on induction of DNA synthesis (Fig. 3b). A quantification of the proliferation assay is shown in Fig. 3c.
Dominant-Negative Mutants of c-myc Induce Apoptosis in Mouse 3T3-L1 Fibroblasts-Examination of the dominant-negative c-Myc mutants in the proliferation assay turned out to be impossible. 6 h after injection of the expression plasmids, many of the cells expressing ctMyc or ctMycBR Ϫ showed morphological changes with characteristic features of apoptosis. These changes included cytoplasmic blebbing, chromatin condensation, and nuclear fragmentation (Fig. 4). Nuclei of uninjected nonapoptotic cells exhibit a regular blue chromatin staining. Chromatin staining of nuclei of apoptotic cells is irregular with bright areas (Fig. 4, arrow). 20 h after microinjection, the microinjected fields on the coverslip were almost free of cells, and the few remaining cells staining positive for the c-Myc  (Blackwell et al., 1990) with the consensus c-Myc-binding site, and binding mix. A, Lanes 2-9, 1.5 l of the indicated lysate was incubated at 37°C for 10 min; lanes 10 -12, 0.75 l of ctMyc lysate was mixed with 0.75 l of Max lysate and incubated at 37°C for 10 min. ctMyc and Max generate a specific gel shift (arrow). Labeled CM1 oligonucleotide alone (lane 1) and unprogrammed reticulocyte lysate (UL) (lane 2) served as controls. The endogenous unspecific complex in the unprogrammed lysate is marked with an asterisk. Competition experiments were performed by addition of a 200-fold molar excess of unlabeled oligonucleotide CM1 (200xCM1) or CM1mut (200xCM1mut). B, lanes 1, 3, and 5, 1.5 l of ctMyc lysate (lane 1) or 0.75 l of ctMyc lysate mixed with 0.75 l of Max lysate mutants were detached and stuck loosely at the surface.
We performed kinetic experiments to determine the onset of apoptosis in cells expressing ctMyc and ctMycBR Ϫ . For this purpose, the injected cells were inspected and photographed in intervals of 2 h after microinjection (Fig. 5). Expression of mutant c-Myc proteins could be detected in the nucleus and in the cytoplasm 2 h after microinjection (data not shown). The rate of expressing cells was Ͼ70% of injected cells and thus in the range as observed for other proteins (MaxBR Ϫ , MaxBRmut, ␤-galactosidase). At this time, cells expressing ctMyc and ct-MycBR Ϫ still showed a normal shape. However, after 4 h, the cells changed their morphology, and bright spots became apparent in the field of injected cells. The number of affected cells consistently increased between 6 to 10 h, while at the same time many of the affected cells disintegrated. After 12 h, cells with a normal shape expressing ctMyc or ctMycBR Ϫ were no longer detectable. Apoptosis was not observed in cells expressing ␤-galactosidase, MaxBRmut (see Fig. 3), and MaxBR Ϫ (Fig. 5).
Dominant-negative mutants of c-Myc have been successfully used in earlier studies to inhibit transformation (Dang et al., 1989;Sawyers et al., 1992) and induction of DNA-synthesis . These c-Myc mutants carried deletions in the transactivation domain between amino acids 40 and 178 and therefore differ markedly from the mutants employed in this study, which have deleted the amino terminus up to amino acids 354 and 367. Notably, the mutant with the deletion in the transactivation domain did not induce apoptosis in quiescent 3T3-L1 fibroblasts in microinjection experiments data not shown).
Induction of apoptosis by wild-type c-Myc in serum-starved mouse fibroblasts can be inhibited by refeeding the cells with medium containing high serum levels . Therefore, similar kinetic experiments as carried out with serum-starved 3T3-L1 fibroblasts were performed with proliferating 3T3-L1 cells in the presence of 10% FCS. Since proliferating fibroblasts showed a high mobility on the surface of the coverslip with a permanent change of positions, a similar documentation of the results as shown for quiescent fibroblasts in Fig. 5 was impossible. Despite this problem, plasmids coding for ctMycBR Ϫ and ctMyc were microinjected in proliferating cells. Both mutants showed a normal expression rate of 70% after 2 h. Similar to quiescent cells, many of the ctMycBR Ϫ and ctMyc expressing cells showed an apoptotic morphology after 6 -8 h. 12 h after microinjection, cells with a normal morphology expressing the c-Myc mutants were no longer detectable (data not shown).
Max Mutants Inhibit Apoptosis Induced by ctMyc and ctMy-cBR Ϫ -The mutant ctMycBR Ϫ induces apoptosis without binding to DNA. However, this mutant can still dimerize with Max. Therefore, we tested whether co-expression of MaxBR Ϫ could modulate ctMycBR Ϫ -induced apoptosis (Fig. 6). Plasmids encoding ctMycBR Ϫ or ctMyc were mixed with a plasmid coding for MaxBR Ϫ and injected in quiescent 3T3-L1 cells. 12 and 24 h after microinjection, the cells showed a normal morphology, and the rate of ctMycBR Ϫ -expressing cells was in the range of 50 -60% of injected cells. Thus, co-expression of MaxBR Ϫ suppressed ctMycBR Ϫ -induced apoptosis. Apoptosis was also suppressed by MaxBRmut, but less efficiently (Table I).
Separate Expression of MycHLH and MycLZ Does Not Induce Apoptosis-We next studied whether expression of either the HLH (MycHLH) or the LZ domain (MycLZ) of c-Myc is sufficient to induce apoptosis. Therefore, expression plasmids were constructed encoding either the HLH or LZ domain of c-Myc (Fig. 1). Both mutants separately or in combination could not compete for the formation of the ctMyc⅐Max complex in gel-shift experiments (Fig. 2c, lanes 4 and 5; data not shown). Despite high expression, each mutant failed to induce apoptosis in quiescent mouse 3T3-L1 fibroblasts after 12 h (Fig. 7). Cells expressing MycHLH showed an increased rate of vacuolization after 12 h, which might be an indicator of lack of cell health. However, the rate of apoptotic cells was not increased when these cells were examined after 24 h (data not shown). Interestingly, co-expression of MycHLH and MycLZ restored induction of apoptosis with similar kinetics as observed for ctMy-cBR Ϫ (Table I). A survey of the apoptotic activity of all mutants used in this work is shown in Table I.

DISCUSSION
Expression of the c-myc gene in serum-starved mouse fibroblasts induces DNA synthesis and at the same time triggers apoptosis. Both events have been shown to require heterodimerization of c-Myc and Max. Here we have studied dominant-negative mutants inhibiting specifically DNA binding of c-Myc⅐Max heterodimers. The phenotypes of c-Myc and Max mutants after expression in 3T3-L1 mouse fibroblasts differed markedly. While Max mutants inhibited serum-induced DNA synthesis, c-Myc mutants rapidly induced apoptosis.
Phenotypes of Max and c-Myc Mutants-The mutants MaxBR Ϫ and MaxBRmut specifically inhibited binding of ctMyc⅐Max heterodimers to DNA in gel-shift experiments. Both mutants also inhibited DNA synthesis in serum-stimulated 3T3-L1 mouse fibroblasts, indicating that they could act as dominant-negative mutants in vivo. This confirms earlier results showing that induction of cell cycle progression depends on binding of the c-Myc⅐Max heterodimers to DNA (Cogliati et al. 1993;Amati et al., 1993b). Several c-Myc target genes have been identified so far (Eilers et al., 1991;Benvenisty et al., 1992;Bello-Fernandez et al., 1993;Jansen-Dü rr et al., 1993;Reisman et al. 1993;Yang et al., 1993;Gaubatz et al., 1994). Modulation of the activity of these and yet unidentified genes by the Max mutants may be responsible for the inhibition of cell cycle progression. Since Max mutants also affect homodimerization of wild-type Max and heterodimerization of Max with Mad and Mxi1 (Ayer et al., 1993;Zervos et al., 1993), the inhibitory effect of Max mutants could be even more complex and rely not only on the inhibition of DNA binding of c-Myc⅐Max heterodimers.
The mutant ctMycBR Ϫ inhibited binding of the ctMyc⅐Max heterodimer to the cognate E-box to a similar extent as the (lanes 3 and 5) were incubated at 37°C for 10 min. Lane 2, 1 l of ␣-Max antibody and 1.5 l of ctMyc lysate were mixed and incubated at 37°C for 10 min. Subsequently, 0.75 l of ctMyc lysate was added, and incubation at 37°C was continued for 5 min. MaxBR Ϫ , MaxBRmut, and ctMycBR Ϫ but not MycHLH and MycLZ compete for the formation of ctMyc⅐Max-specific gel shifts. For further details, see "Materials and Methods." mutants MaxBRmut and MaxBR Ϫ did. From this observation, we expected an inhibitory effect on DNA synthesis also for the mutant ctMycBR Ϫ . However, this mutant induced apoptosis in serum-stimulated cells before the onset of DNA synthesis could be measured. Apoptosis was induced by this mutant at low and high serum levels and also observed after expression of ctMyc.
Induction of apoptosis by wild-type c-Myc requires dimerization of the HLH/LZ domains of c-Myc and Max (Amati et al. 1993b). The following reasons argue against dimerization of Max and the c-Myc mutants as requirement for induction of FIG. 3. Dominant-negative mutants of Max inhibit DNA synthesis in serum-stimulated cells. a, time scheme of the proliferation assay. b, quiescent 3T3-L1 fibroblasts were microinjected with expression plasmids encoding MaxBR Ϫ , MaxBRmut, or ␤-galactosidase. 20 h after serum stimulation, the cells were fixed and analyzed by immunocytochemistry. DNA synthesis of cells expressing the Max mutants or ␤-galactosidase (these cells are marked by white arrows) was measured by incorporation of BrdUrd. c, quantitative evaluation of at least five experiments. Each experiment was done with more than 200 cells expressing either MaxBR Ϫ , MaxBRmut, or ␤-galactosidase, and the rate of BrdUrd incorporation was determined. BrdUrd incorporation in serum-starved cells was consistently less than 3%, in serum-stimulated cells more than 70%.
apoptosis. (i) The presence of Max does not confer DNA binding activity to the ctMycBR Ϫ ⅐Max heterodimer. Alternatively, ct-MycBR Ϫ may act by sequestering Max from other DNA binding complexes. However, Max is also sequestered by MaxBR Ϫ and MaxBRmut, which do not induce apoptosis. (ii) Instead, the mutants MaxBR Ϫ and MaxBRmut suppress ctMycBR Ϫ -induced apoptosis, indicating that complexes of ctMycBR Ϫ ⅐ MaxBR Ϫ and ctMycBR Ϫ ⅐MaxBRmut are unable to induce apoptosis. We therefore conclude that ctMycBR Ϫ and ctMyc interact probably with other factor(s) than Max to induce apoptosis in 3T3-L1 cells.
Induction of Apoptosis by the HLH/LZ Domain of c-Myc-Additional factors interacting with c-Myc have recently been identified including the transcription factors TFII-I , YY1 (Shrivastava et al., 1993), and AP2 (Gaubatz et al., 1995). The interaction with all three factors has been shown to involve the HLH/LZ domain of c-Myc. Ternary complexes of c-Myc⅐Max⅐TFII-I or c-Myc⅐Max⅐YY1 could not be demonstrated, indicating that binding of TFII-I or YY1 to c-Myc excludes Max binding. In an attempt to define the carboxylterminal domain of c-Myc responsible for inducing of apoptosis more precisely, the HLH and LZ domains were expressed separately. Both domains could be expressed at high levels in 3T3-L1 cells without any sign of apoptosis. However, co-expression of the HLH and LZ domains restored induction of apoptosis, indicating that both domains are important for induction of apoptosis in our system but that the contiguous arrangement of the HLH and LZ domains is not necessary. Additionally, the results indicate that apoptosis is induced by the specific expression of the c-Myc HLH/LZ domain and not by the experimental approach and expression of high levels of proteins. The data suggest that the c-Myc HLH/LZ domain induces apoptosis by its specific interaction with other cellular factors containing HLH and/or LZ motifs.
Induction of apoptosis by wild-type c-Myc depends on the presence of the amino-terminal transcriptional activation domain  and dimerization of c-Myc with Max (Amati et al., 1993b). This suggests specific gene-regulatory activity of the c-Myc⅐Max complex during induction of apoptosis. In this respect, induction of apoptosis by the c-Myc mutants differs markedly from wild-type c-Myc-induced apoptosis. It is very unlikely that the c-Myc mutants can still regulate target FIG. 4. Apoptotic cell expressing ctMycBR ؊ , which shows the characteristic cytoplasmic blebbing and chromatin condensation. Cells were fixed 6 h after microinjection, ctMycBR Ϫ was detected by immunofluorescence with the ␣-Myc antibody (9E10), and chromatin was stained with 4,6-diamidino-2-phenylindole (DAPI). The apoptotic cell is indicated by an arrow.
genes of wild-type c-Myc. However, since the mutants ctMy-cBR Ϫ and ctMyc probably can still interact with other transcription factors, both mutants may also induce apoptosis by altering gene expression. The changes in gene expression are probably severe and induce apoptosis even in the presence of high serum. Although we are far from understanding how the mutants ctMyc and ctMycBR Ϫ act, the presented data are of importance for our understanding of c-Myc protein function.
A c-Myc mutant with a deleted transactivation domain (amino acids 40 -178) is unable to induce apoptosis in quiescent 3T3-L1 fibroblasts but has been shown to inhibit serum-induced DNA synthesis . This implicates that the central part of the c-Myc protein (amino acids 179 -354) either by itself or by binding of other proteins, controls the interaction of cellular proteins with the carboxyl-terminal HLH/LZ domain of c-Myc. Deletion of the central part of the c-Myc protein renders the HLH/LZ domain highly promiscuous for otherwise tightly controlled interactions with cellular factors. For this reason, c-Myc mutants with a deleted transactivation-domain are more useful as dominant-negative mutants than mutants consisting only of the HLH/LZ domain (Mukherjee et al., 1992;. FIG. 6. Co-expression of MaxBR ؊ inhibits apoptosis induced by ctMyc and ctMycBR ؊ . Expression plasmids were mixed in a ratio of 2:1 (300 ng/l MaxBR Ϫ , 150 ng/l ctMyc or ctMycBR Ϫ ) and microinjected in quiescent 3T3-L1 fibroblasts. Cells were fixed after 12 h, and ctMyc and ctMycBR Ϫ were detected by immunofluorescence staining with the ␣-Myc antibody (9E10). Chromatin was stained with 4,6-diamidino-2-phenylindole.

TABLE I Ability of the c-Myc and Max mutants to induce apoptosis
The rate of apoptotic cells is given in percent and was calculated as the rate of apoptotic cells between 2 and 12 h after microinjection of quiescent 3T3-L1 mouse fibroblasts. All experiments were carried out at least three times, each with a minimal number of at least 200 cells.
FIG. 7. Separate expression of MycHLH and MycLZ does not induce apoptosis. Quiescent 3T3-L1 fibroblasts were microinjected with expression plasmids coding for MycHLH or MycLZ, fixed after 12 h, and the mutants were detected by immunofluorescence staining with the antibody 12CA5. Chromatin was stained with 4,6-diamidino-2-phenylindole.