(Received for publication, April 24, 1997)
From the MAX is a basic helix-loop-helix-leucine zipper
protein that plays a central role in the transcriptional control of Myc
oncoproteins. MYC-MAX heterodimers stimulate transcription, whereas MAX
homodimers, or heterodimers between MAX and members of the MAD family
of basic helix-loop-helix-leucine zipper proteins, repress
transcription. Max exists in two major isomeric forms, MAX(L) and
MAX(S), which differ from one another only by a 9-amino acid
insertion/deletion. We show here that MAX(L) is much more effective at
homodimeric DNA binding than MAX(S). In NIH3T3 cells, MAX(L) was able
to repress a c-Myc-responsive reporter gene whereas MAX(S) either
stimulated the reporter gene or had little effect on its expression. In
comparison to control cell lines or those stably over-expressing
MAX(S), MAX(L)-over-expressing cell lines showed reduced expression of transiently expressed or endogenous c-Myc responsive genes, grew more
slowly, possessed a higher growth factor requirement, and showed
accelerated apoptosis following growth factor deprivation. Differential
effects on growth and apoptosis represent two previously unrecognized
properties of MAX proteins. These can at least partly be explained by
the differences in their DNA binding abilities and their effects on
target gene expression.
The c-MYC oncoprotein is an important member of the basic
helix-loop-helix-leucine zipper family of transcription factors and is
involved in such cellular processes as proliferation, transformation, differentiation, and apoptosis (1-4). Like most proteins of this class, c-MYC possesses sequence-specific DNA binding properties (5-7)
and can transactivate synthetic or natural promoters containing c-MYC
binding sites (8-12). At extremely high protein concentrations in vitro, c-MYC binds DNA as a homodimer (13). However, at
lower concentrations and within cells, c-MYC homodimerization does not occur to any detectable degree (14, 15). Thus, the active DNA binding
moiety appears to be a heterodimer comprised of c-MYC and another basic
helix-loop-helix-leucine zipper protein, MAX (16, 17). Most of the
biological properties of c-MYC have been attributed to its association
with MAX (18-21). These heterodimers bind DNA strongly and activate
transcription (18). Max also homodimerizes avidly, and in
vitro and in vivo, both c-MYC-MAX heterodimers and MAX
homodimers can be easily detected (14, 22-25). However, because MAX
lacks a transactivation domain (26), MAX over-expression has been
reported to repress the transcription of genes bearing c-MYC binding
sites (8, 11, 27-29). Several model systems have demonstrated that the
expression of such genes is directly dependent upon the levels of
c-MYC-MAX heterodimers and inversely dependent upon the levels of MAX
(8, 10, 11, 28, 29).
Alternate mRNA splicing produces two major MAX isoforms, the
shorter of which (MAX(S)) is a 151-amino acid polypeptide with an
apparent molecular mass of 21 kDa as determined by
SDS-PAGE1 (16, 17, 23). The longer form
MAX(L) is a 160-amino acid, 22-kDa polypeptide, identical to MAX(S)
except for a 9-amino acid insert between residues 12 and 13. Both
proteins, and their respective transcripts, are expressed at
approximately equal levels in most cell types examined and, unlike
c-MYC, are highly stable and unresponsive to the proliferative state of
the cell (23, 30, 31). Both MAX isoforms are highly conserved between
humans and such lower species as chickens, Xenopus, and
zebrafish (32, 33).
Little is known regarding how the two MAX proteins differ from one
another functionally, as most studies have been performed with only one
of the two isoforms and/or have been of restricted biological scope.
These studies are also difficult to compare as they have utilized
different types of expression vectors, growth conditions, and
biochemical and biological read-outs. Recent work has also emphasized
the seemingly contradictory results of c-MYC reporter gene assays that
can occur due to differences in cell density at the time of
transfection (34). We have previously reported that MAX(L) and MAX(S)
differ significantly in their DNA binding abilities (35). Using several
synthetic oligonucleotides whose c-MYC binding sites differed only in
their flanking regions, we showed that, in all cases, MAX(L) bound
>20-fold more strongly to these sites than MAX(S). However, we could
not absolutely rule out that these differences were, at least in part,
the consequence of amino-terminal hexahistidine tags that had been
introduced to facilitate purification of the recombinant, bacterially
expressed proteins.
In the present study, we have compared the abilities of full-length,
unmodified MAX(L) and MAX(S) to bind DNA and to modulate transiently
transfected and endogenous c-MYC-responsive genes. We have also
examined the effects of the two proteins on proliferation, growth
factor requirement, and apoptosis in two different cell lines. We find
that MAX(L) and MAX(S) differ markedly from one another in their
abilities to affect these cellular properties. Because not all of the
observed effects appear to be the result of the antagonism of c-MYC-MAX
heterodimers, our results suggest that, depending upon the identity of
the expressed isoform, as well as on its biological context, MAX
proteins can either enhance or oppose many of the known activities of
c-MYC. Because MAX plays a central role in both the positive and
negative regulation of c-Myc activity, such opposing effects may
provide a more sensitive control over this process.
MAX
cDNAs were amplified by the polymerase chain reaction using the
following primers: forward, 5 Electrophoretic
mobility shift assays were performed essentially as described (35)
using 50 ng of each purified protein or approximately 200 ng of crude
bacterial lysate, equivalent to 20-50 ng of MAX. 5 µl of each
protein was incubated at room temperature for 30 min with 5 µl of
2 × binding buffer containing 1 µg of poly(dI-dC) (Pharmacia
Biotech Inc.) and 10,000 dpm of the 32P-end-labeled,
double-stranded EO(GAC) oligonucleotide containing a single c-MYC
consensus binding site (specific activity approximately 108
dpm/µg). Bound and free labeled complexes were resolved at room temperature on native 4% polyacrylamide gels in 1 × TBE buffer as described (35).
cDNA inserts for
c-Myc (1.5-kilobase XhoI fragment, Ref. 36), MAX(L), and
MAX(S) (16) were excised from their parental vectors and cloned into
the XhoI site of the pSVLneo vector. This was derived from
the pSVL vector (Pharmacia) by inserting into the unique
EcoRI site a 2.5-kilobase EcoRI-linkered DNA
fragment containing the neomycin resistance gene under the control of
the SV40 early promoter (37). The p3xMyc-E1b-luc wild-type and
p0Myc-E1b-luc mutant c-Myc reporter plasmid (11) were kind gifts from
Dr. Roger Davis. The pSV2 NIH3T3 cells were
routinely grown in Dulbecco's modified minimal essential medium
supplemented with 1 mM glutamine, 100 µg/ml streptomycin,
100 units/ml penicillin G, and 10% supplemented calf serum (Hyclone,
Logan, UT). The 32D clone 3 murine myeloid cell line and the WEHI-238
murine B cell line were routinely grown in RPMI 1640 medium
supplemented as described above except that fetal calf serum was
substituted for supplemented calf serum. In addition, the medium used
for 32D cell growth was supplemented with 10% (v/v) conditioned medium
obtained from the IL-3-producing WEHI-238 cell line.
For transient expression assays, NIH3T3 cells were transfected using a
standard calcium-phosphate-based procedure (38). Cultures were
routinely plated at 5 × 105 cells/100-mm dish,
transfected the following day, and harvested 2 days later for the assay
of To obtain stably transfected 32D clones, 2 × 107
cells were collected by low speed centrifugation, washed twice in
tissue culture medium containing IL-3 and fetal calf serum, and
resuspended in 0.3 ml of the same medium. 10 µg of
NdeI-linearized vector DNA in 25 µl of water was added.
Electroporation was carried out in a 0.4-cm disposable cuvette
(Bio-Rad) using a Bio-Rad apparatus set at 300 V and a capacitance
extender set at 960 microfarads. Following electroporation, cells were
incubated on ice for 10 min before being replated in standard IL-3
containing medium. Two days later G-418 was added to a final
concentration of 400 µg/ml and maintained continuously.
G-418-resistant clones were generally observed after about 2 weeks.
Pooled cells representing at least several hundred independent clones
were used for all experiments.
NIH3T3 cell monolayers
(approximately 80% confluency) were washed in methionine + cysteine-free Dulbecco's modified minimal essential medium (ICN
Biomedical, Costa Mesa, CA). The medium was removed and replaced with 1 ml of the same medium containing 100 µCi of
35S-Translabel (ICN, specific activity 1.21 Ci/mmol) for
6 h. The monolayers were then washed exhaustively in ice-cold PBS
and collected in 1 ml of standard RIPA buffer containing the same
protease inhibitors as used for bacterial cultures (see above).
Following centrifugation at 14,000 × g at 4 °C,
2 × 107 dpm of lysate supernatant in a total volume
of 1 ml of RIPA buffer was incubated at room temperature for 2 h
with a rabbit polyclonal anti-human MAX antibody at a final dilution of
1:300. Antigen-antibody complexes were removed by the addition of 50 µl of protein A-Sepharose (Bio-Rad) for an additional 2 h.
Precipitates were washed three times in RIPA buffer, disrupted in
loading buffer, and resolved by 12% SDS-polyacrylamide gel
electrophoresis (SDS-PAGE). 32D cells were processed in an
identical manner except that labeling was performed following the
resuspension of 5 × 106 cells in 1 ml of
methionine/cysteine-free Dulbecco's modified minimal essential medium
containing 100 µCi of Translabel.
Cells were pelleted by centrifugation
at 4 °C, washed in PBS, and fixed in 1 ml of ice-cold 70% ethanol
for at least 1 h. The cells were pelleted twice in ice-cold PBS
and resuspended in PBS containing 100 µg/ml RNase A and 50 µg/ml
propidium iodide (Sigma). Cell cycle analyses were performed on a
Becton-Dickinson FACSTAR fluorescence-activated cell sorter. 20,000 cells were analyzed for each assay with quantitation of cell cycle
parameters being performed using single histogram statistics.
32D cells were cultured at 3 × 104 cells/ml in 30 ml of IL-3-supplemented medium
containing 10% fetal calf serum. 24 h later, the cells were
pelleted by low speed centrifugation and resuspended in the above
medium minus IL-3. Aliquots of cells were removed at timed intervals,
and the fraction of viable cells was determined by trypan blue
exclusion. To evaluate the integrity of DNA, cells were washed twice in
PBS and resuspended in 1 ml of lysis buffer (10 mM
Tris-HCl, pH 8.0; 0.1 M EDTA; 0.5% SDS, 20 µg/ml RNase A, and 100 µg/ml proteinase K). After incubating at 37 °C for 4 h overnight, the lysates were phenol-extracted and
ethanol-precipitated. DNA was redissolved in water and 10 µg was
electrophoresed on a 2% agarose gel.
NIH3T3 cells were
plated at 105 cells/ml and maintained in log-phase growth
for 2 days. Cells were then washed twice with PBS, scraped into conical
centrifuge tubes, and pelleted at 1,000 × g for 10 min. ODC assays, using 200 µg of cellular extract, were performed as
described (39) except that an incubation period of 2 h was
used.
We previously demonstrated that
bacterially expressed MAX(L) binds strongly to a c-Myc binding site,
whereas MAX(S) shows little detectable binding (35). In these
experiments, however, each protein was expressed as a hexahistidine
fusion to facilitate its purification, and it was therefore formally
possible that these extraneous amino acids altered the DNA binding
properties. To address this, we expressed both MAX proteins as
non-fusion polypeptides in E. coli. Examination of
IPTG-induced E. coli lysates by SDS-PAGE showed that MAX
accounted for approximately 25% of the total protein. Following
purification by DEAE-cellulose chromatography under non-denaturing
conditions, MAX proteins were tested for their ability to bind the
above c-Myc site. MAX(L) bound the labeled oligonucleotide strongly,
whereas binding by the equivalent amount of MAX(S) was barely
detectable (Fig. 1A). This effect was not due
to an inadvertent loss of DNA binding ability by MAX(S) during its
purification as it retained the ability to bind the probe in
association with purified N-Myc protein (lane 4). The
differences in DNA binding by the two MAX proteins were also apparent
prior to purification (Fig. 1B). In some experiments, and at
higher protein concentrations, we have observed some DNA binding by
MAX(S), but it has always been at least 20-fold less than seen with a comparable amount of MAX(L). These results, with both purified and
crude MAX proteins, confirmed our previous observations (35) and
indicated that the observed differences in DNA binding were not the
artifactual result of changes in polypeptide structure.
The observed
differences in DNA binding by MAX(L) and MAX(S) in vitro
suggested that the two proteins might have dissimilar effects on a
c-Myc-responsive reporter gene in vivo. We tested this idea
using transient transfections in NIH3T3 cells. In the first series of
experiments, a constant amount of a c-Myc responsive reporter construct
(11) was co-transfected with increasing amounts of either a MAX(L) or
MAX(S) expression vector. We assumed that the basal expression of
luciferase activity by the reporter in the absence of any other
co-transfected plasmids at least partly reflected the levels of
endogenous c-Myc and MAX protein activities in these cells. As seen in
Fig. 2A, MAX(L) suppressed luciferase activity in a dose-dependent manner, consistent with the
idea that at high DNA concentrations, MAX(L) homodimers predominate over c-Myc-MAX heterodimers, compete for c-Myc binding sites, and
repress reporter gene transcription (10, 11, 29). The small amount of
residual luciferase activity seen at the highest MAX(L) concentrations
probably represents the c-Myc/MAX-independent activity of the core E1b
promoter. In contrast, co-transfected MAX(S) resulted in an increase in
reporter gene activity which, at the highest concentrations tested, was
3-4-fold over basal levels. At these concentrations the differences
between MAX(L) and MAX(S) were over 7-fold. This observation suggested
that high concentrations MAX(S) allowed its heterodimerization with
endogenous c-Myc, resulting in an increase in the number of
transcriptionally active c-Myc-MAX heterodimers. However, once all
available c-Myc was dimerized, further increases in MAX(S) were unable
to repress transcription. This is consistent with the weak DNA binding
activity of MAX(S) protein purified from E. coli (Fig.
1).
We next tested the effect of expression of each MAX isoform on reporter
gene expression in the presence of co-expressed c-Myc. As shown in Fig.
2B, c-Myc alone stimulated reporter gene activity an average
of 4-fold, consistent with the degree of activation reported by many
other groups (8, 10, 11, 29). With increasing amounts of the
co-transfected MAX(S) expression plasmid, only minimal (1.5-2 fold)
additional stimulation of the reporter was observed. In contrast, the
transfection of the MAX(L) expression plasmid resulted in a
dose-dependent decrease in reporter gene expression of up
to 8-fold. The difference between MAX(L) and MAX(S) at the highest
concentrations examined was over 10-fold. These results were consistent
with those presented in Fig. 2A and indicated that, under
the conditions employed for these experiments, MAX(S) was stimulatory
whereas MAX(L) was generally inhibitory toward c-Myc reporter genes.
Similar results were observed with a CAT reporter construct driven by
an ornithine decarboxylase (ODC) promoter, a known physiologic c-Myc
target (9) (not shown). In control experiments, no regulation by c-Myc
or MAX was seen in transient transfections employing either
E1b-luciferase or ODC-CAT vectors containing absent or mutant c-Myc
binding sites (9, 11) (not shown).
To study
the biological consequences of MAX protein over-expression in more
detail, we established lines of NIH3T3 cells stably transfected with
either MAX(L) or MAX(S) expression plasmids. We selected representative
clones and measured the amount of MAX protein by metabolic labeling.
Fig. 3 shows that both the parental NIH3T3 cell line as
well as single cell clones transfected with the empty vector alone
expressed very low levels of Max. In contrast, MAX-transfected clones
expressed the expected individual isoform at high levels.
A representative clone from each group was transfected with the
p3xMyc-E1b-luc plasmid to determine the effect of stable
over-expression of each MAX isoform on a c-Myc-responsive gene. As seen
in Fig. 4A, the expression of the plasmid was
only slightly elevated above the control in MAX(S)-over-expressing
cells, whereas in MAX(L)-over-expressing cells, plasmid expression was
reduced by about 7-fold. In control experiments, a reporter plasmid
with mutant c-Myc binding sites (11) was expressed at equivalent levels
in all three cell lines thus indicating that the above results did not
simply reflect a generalized lower rate of transcription in the
MAX(L)-over-expressing cell line.
To determine how each MAX isoform affected the expression of an
endogenous, c-Myc-responsive gene, we also measured ODC enzyme levels
in each of the above clones. As seen in Fig. 4B, the average ODC level was reduced more than 7-fold in the MAX(L)-over-expressing cell line, whereas no significant change from the control was seen in
the MAX(S)-over-expressing cells.
Growth rates for each of the clones shown in Fig. 3 were also
determined. Fig. 5 shows that each of the MAX(S)
over-expressing cell lines grew at rates that were slightly faster than
those of controls. More strikingly, each of the three
MAX(L)-over-expressing clones was growth retarded, showing 8-20-fold
fewer cells compared with controls after 7 days in culture. These
differences were not due to variations in initial plating efficiencies
of MAX(L)-over-expressing cells2 and
persisted throughout the duration of the experiment.
To determine the basis for the differences in growth rates among the
three groups of cell lines, we studied their cell cycle distribution at
various times after plating. Surprisingly, only small and transient
differences were seen in cell cycle patterns among all of the clones
during logarithmic growth (not shown). This suggested that the reduced
growth rate of MAX(L)-over-expressing cells was attributable to a
reduced rate of progression through each phase of the cell cycle
without altering the overall distribution pattern. To test this
directly, control, MAX(L), or MAX(S)-over-expressing cells were
arrested in G0 by serum deprivation for 24 h. The
cells were then stimulated with serum in the presence of nocodazole to
prevent cell cycle progression beyond the first mitotic stage. This
allowed us to clearly measure the rate of progression through a single
cell cycle by flow cytometry. As seen in Fig. 6,
MAX(S)-over-expressing cells traversed the cell cycle slightly faster
than control cells. This was seen best at 16 and 18 h after serum
stimulation. At the latter time, virtually all of the MAX(S)
over-expressing cells had reached G2/M, whereas nearly half
the control cells remained in G1 or S phase. By 24 h,
however, both these lines had completely entered G2/M. This
finding was consistent with the slightly faster growth rate of
MAX(S)-over-expressing cells seen in Fig. 5. In contrast to these
subtle findings, MAX(L)-over-expressing cells showed a significant
delay in cell cycle progression at all stages. For example, at 14 h after serum addition, the MAX(L) population had barely begun to exit
the G0/G1 phase, whereas a majority of cells in
both the control and MAX(S) population had entered S phase and even
progressed to G2/M. Similarly, by 20 h, a significant fraction of the MAX(L) population remained in
G0/G1 and S phases, whereas virtually the
entire control population had traversed the entire cell cycle.
Independent confirmation of this lag was obtained by determining the
mitotic index of each of the three populations of cells 20 h after
serum stimulation. At this time, both control and
MAX(S)-over-expressing cultures showed 90-95% of the populations to
be in mitosis whereas, in MAX(L)-over-expressing cultures, only 46% of
the cells had entered mitosis. Two additional experiments gave
virtually identical results as did similar studies performed with each
of the other clonal cell populations (not shown). Our findings thus
indicate that the slower growth of MAX(L)-over-expressing cells is
largely attributable to a prolonged cell cycle that results from a
lengthening of all phases without any change in the overall distribution pattern.
We also examined each of the MAX(L)-over-expressing clones to determine
if their slower rate of growth might be explained by a higher rate of
apoptotic cell death. No such differences in the rates of cell death
were seen under standard conditions of logarithmic growth (not shown),
and this was confirmed by the absence of any cells with subdiploid
amounts of DNA in our cell cycle analyses (Fig. 6). Nevertheless, in a
number of settings following growth factor withdraw, c-Myc
over-expression accelerates apoptotic cell death through a process that
is dependent upon Max (3, 20, 40-42). Therefore, to determine whether
either of the MAX protein isoforms could influence cell death, we
deprived representative clones of serum and monitored the fraction of
surviving cells. Under these conditions, control and
MAX(S)-over-expressing cells showed identical rates of cell death,
whereas MAX(L)-over-expressing cells died at a 2-3-fold faster rate
(Fig. 7A). This was confirmed by light
microscopic studies of cell monolayers that showed a significantly
greater number of rounded, refractile cells among the MAX(L) population
(Fig. 7B) and by electron microscopic studies demonstrating
the morphological features typical of apoptosis, including nuclear
condensation and cytoplasmic "blebbing" (not shown). Despite
repeated attempts, we were unable to demonstrate the presence of the
nucleosome-sized DNA fragments that are a cardinal feature of
apoptosis. The inability of some murine fibroblasts, including NIH3T3
cells, to display this property following treatment with known
apoptotic stimuli has been previously noted (43-45). Nevertheless, our
results are consistent with the idea that MAX(L) over-expression could,
under appropriate conditions, modulate cell death in NIH3T3 cells and
that this process probably utilized apoptotic pathways.
To study the effect
of MAX protein over-expression on apoptosis in a better system, as well
as to generalize our findings in NIH3T3 cells, we utilized the 32D
murine myeloid cell line (46). Proliferation and survival of 32D cells
are dependent on the hematopoietic cytokine IL-3 whose withdrawal
results in rapid and easily quantifiable apoptotic cell death (40).
Previous work has shown that c-Myc over-expression accelerates
apoptosis in these cells following IL-3 withdrawal, thus suggesting
that MAX proteins might be involved in this process (40, 47). Stably transfected populations of 32D cells were therefore generated, and the
over-expression of MAX proteins was confirmed in pooled clones by
immunoprecipitation of [35S]methionine-labeled cell
extracts (Fig. 8). We then compared the growth rates of
control and MAX-overexpressing 32D cells at several IL-3
concentrations. At the highest IL-3 concentration tested, control and
MAX(S)-overexpressing cells were indistinguishable and showed robust
growth, whereas MAX(L)-overexpressing cells grew at a rate one-half to
two-thirds that of control cells (Fig. 9A).
More marked differences were seen as the IL-3 concentration became
limiting. For example, when the amount of cytokine was reduced by
one-half, MAX(L)-over-expressing cells showed significant growth
impairment, whereas MAX(S)-over-expressing and control cells remained
indistinguishable (Fig. 9B). Even more profound differences
among the three cell lines became apparent at the lowest IL-3
concentration tested (Fig. 9C). Although control cultures showed a significant impairment in growth, they still proliferated significantly faster than MAX(L)-over-expressing cells. In marked contrast, MAX(S)-over-expressing cells proliferated 4-fold more rapidly
than control cultures and 10-fold more rapidly than
MAX(L)-over-expressing cells. These results indicate that, compared
with control 32D cells, MAX(L)-over-expressing cells have decreased
IL-3 sensitivity, whereas MAX(S)-over-expressing cells have an
increased sensitivity.
The increased requirement for IL-3 by MAX(L)-over-expressing 32D cells
suggested that they might be more sensitive to apoptotic cell death
following complete removal of the cytokine. This was confirmed when all
three cell lines were studied in parallel (Fig. 10A). Over the 24-h course of the
experiment, MAX(L)-over-expressing cultures showed a 5-8-fold greater
accumulation of dead cells than either control or
MAX(S)-over-expressing cells. That death was due to apoptosis was
confirmed by examination of nuclear DNA which showed the expected
increase in nucleosome-sized fragments (Fig. 10B).
The ability of MAX to homodimerize or heterodimerize with c-MYC
suggests that MAX can participate in both the transcriptional activation and repression of c-MYC-responsive genes (29). Additional modes of MAX regulation, however, have made it apparent that the model
is subject to many other positive and negative influences and is thus
more complicated than originally proposed. For example, either CKII
phosphorylation or dimerization with a MAX isoform lacking a basic
domain can negatively regulate DNA binding (30, 48-50). At least four
additional members of the basic helix-loop-helix-leucine zipper family,
known as MAD proteins, can also heterodimerize with MAX and repress
transcription (51-53). Their contribution to MYC-regulated
transcription may depend upon their abundance as well as certain cell
type-specific factors. Qualitative differences among the various
MAD-MAX complexes may also exist but have not been explored.
Although MAX(L) and MAX(S) represent the major isoforms of the protein
and are highly evolutionarily conserved (33), how they differ
functionally from one another has not been determined. Most published
studies on the effects of c-MYC and MAX have examined only one MAX
isoform and/or have assessed a limited number of properties (8, 10, 28,
29, 31). To our knowledge, none of the work has compared the properties
of cell lines stably expressing either MAX isoform.
We have re-examined several of the known properties of MAX proteins and
have shown that MAX(L) and MAX(S) regulate DNA binding, cell cycle
progression, and apoptosis in distinct ways. One of the most striking
differences between MAX(L) and MAX(S) is in their intrinsic DNA binding
properties (Fig. 1). These differences, previously noted with
hexahistidine-tagged polypeptides (35), have now been confirmed with
full-length, unmodified proteins. Although we have occasionally
detected DNA binding by MAX(S), it is observed at a level similar to
that seen with "forced" c-MYC homodimers (5) and, in our hands, is
always at least 20-fold less than obtained with comparable amounts of
MAX(L).
It is possible to understand our in vivo results in light of
the above DNA binding studies. Thus, the findings in transient transfection assays that MAX(L) is a more potent repressor of a c-MYC
reporter construct than MAX(S) (Fig. 2) can be interpreted as
indicating that MAX(L) homodimers compete with c-Myc-MAX heterodimers for the same DNA binding sites. In contrast, DNA binding by
c-Myc-MAX(S) remains largely unopposed by MAX(S) homodimers so that
only reporter gene stimulation is seen. However, the extent of this
effect may depend upon the amount of endogenous c-Myc protein, and at
sufficiently high MAX(S):c-Myc ratios, repression may also occur.
Similar findings were obtained in cells stably over-expressing
individual MAX isoforms. Thus, fibroblasts over-expressing MAX(L)
showed significantly less expression of either a transiently expressed
or endogenous c-Myc-responsive gene (Fig. 4). In a more biological
setting, those lines expressing MAX(L) were growth retarded, whereas
those expressing MAX(S) either behaved no differently from control
lines or were growth-stimulated (Figs. 5 and 9). However, the nature and extent of these opposing effects was dependent upon the cellular context (for example, see Fig. 9).
The contrasting effects of MAX(L) and MAX(S) on cell cycle progression
and apoptosis represent heretofore unrecognized properties of these
proteins. In retrospect, these attributes are not unexpected, given the
known role for c-MYC in G0/G1 progression and
apoptosis (40, 41, 50, 54, 55). Our observations argue that cellular decisions pertaining to these processes may hinge upon the proper levels of both c-MYC and MAX, as well as other factors. In
the case where mitogenic stimulation results in high levels of
endogenous c-MYC, MAX proteins, particularly MAX(S), may exert either
little negative effect on proliferation and apoptosis or may even
stimulate proliferation. However, as c-MYC becomes limiting, as in the
case of serum-deprived fibroblasts (Fig. 7) or IL-3-deprived 32D cells (Fig. 9), MAX(L) may act to limit the proliferative response. With more
profound c-MYC depletion, severe growth impairment and/or apoptosis may
ensue as a result of the suppression of c-MYC-responsive genes by a
relative excess of MAX(L) (Fig. 10). These results suggest that the
control of cellular proliferation and apoptosis by the c-MYC-MAX
network can tolerate some variation in the relative levels of these
proteins but ultimately results in cell death when these limits are
violated in either direction. Apoptosis may thus be viewed as a
decision based upon the cell's assessment and integration of absolute
and relative c-MYC and MAX levels together with its ability to
proliferate or enter a quiescent state in response to these levels. In
turn, these latter activities will be determined by the availability of
growth factors which by themselves may affect proliferative and
apoptotic pathways.
The observation that MAX(L)-over-expressing cells contain lower levels
of Odc (Fig. 4B) provides a partial, although necessarily incomplete, biochemical explanation for their biological behavior. Odc,
the rate-limiting enzyme in polyamine biosynthesis, is cell cycle-regulated and is required for S phase entry (39, 56). Thus, the
lower Odc levels in MAX(L)-over-expressing cells might be sufficient to
account for their slower rate of cell cycle progression, particularly
through S phase. On the other hand, at least some of the pro-apoptotic
activity of c-Myc has been linked to its direct up-regulation of Odc
(39) although other c-Myc target genes are likely to be important for
this process (42). Although the specific relationship between Odc
levels, proliferation, and apoptosis remains to be established with
regard to each of the Max isoforms, the cell lines reported here should
be useful in establishing these associations.
Because c-Myc over-expression has been associated with accelerated
apoptosis in IL-3-deprived 32D cells (40), we examined c-Myc transcript
levels in all three cell lines following IL-3 removal. In all three
cases, c-Myc was expressed at equivalent levels and became undetectable
within 1-3 h following cytokine withdrawal.2 These
observations argue that the apoptosis mediated by MAX(L) over-expression not only does not require concomitant c-Myc
over-expression but may well be c-Myc-independent. However, we cannot
rule out the possibility that small amounts of c-Myc-MAX(L) heterodimer may be sufficient to promote apoptosis. At the very least, our results
suggest that the apoptotic pathways utilized by MAX(L) are distinct
from those utilized by c-Myc, which generally, but not invariably,
require high level and inappropriate expression of the protein (40, 41,
57).
Our results are consistent with a model in which three types of DNA
binding complexes exist in subconfluent cultures of mammalian cells:
c-Myc-MAX(S) and c-Myc-MAX(L) heterodimers, both of which will be
stimulatory, and MAX(L) homodimers which will be repressive. MAX(S)
homodimers will also be present but will be relatively ineffective at
DNA binding. The overall transcriptional state of a c-Myc-responsive
gene will thus reflect the relative abundance of the DNA binding dimers
(34). Over-expressed MAX(L) will be repressive due to the greater
abundance of MAX(L) homodimers than c-Myc-MAX(L) heterodimers. When
co-expressed with c-Myc, low concentrations of MAX(L) might initially
be stimulatory due to its predilection to heterodimerize. However,
repression will predominate at higher MAX(L) levels as homodimers form
and compete with c-Myc-MAX heterodimers. MAX(S) excess will have little
inhibitory effect on target gene expression due to the poor intrinsic
DNA binding activity of MAX(S) homodimers. Co-expression of c-Myc and
MAX(S) will stimulate reporter gene expression due both to increased
c-Myc-MAX(S) heterodimer formation and the weak repressive activity of
MAX(S) homodimers. High levels of MAX(S) expression, perhaps coupled
with limiting c-Myc expression, might eventually produce some
repression and could account for the few reports of MAX(S)-mediated
suppression that have been described.
It is important to bear in mind that the above model is likely to be
oversimplified as it considers DNA binding only by unmodified MAX
homodimers. CKII-mediated phosphorylation and associations with MAD
proteins may affect each MAX isoform differently (27, 35, 48, 49,
51-53). Differences in the affinities of various MAX-MAD heterodimers
for the mSin3 repressor may also exist but have not been explored (32,
58). Such differences, if they exist, may well be highly cell
type-specific and could also account for much of the variability in MAX
function that has been observed.
Despite these questions, our results indicate distinct roles for MAX(L)
and MAX(S) in c-Myc target gene expression, cellular proliferation, and
apoptosis.
We thank Seth Corey for providing 32D and
WEHI-238 cells and Xiaoying Yin for assistance in Northern
blotting.
Department of Molecular Genetics and
Biochemistry,
ABSTRACT
INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
FOOTNOTES
ACKNOWLEDGEMENTS
REFERENCES
Expression and Purification of Recombinant Proteins
-CGC TCA TGA GCG ATA ACG ATG
ACA TC-3
; reverse, 5
-GCG CAA GCT TGC CTG CCC CGA GTG GCT TAG-3
. In
the former case, the italics denote "GC clamps" and NcoI-compatible BspHI sites that were added to
facilitate cloning into the NcoI site of the QE60 vector
(Qiagen, Chatsworth, CA). Following this, cloned cDNAs were
sequenced to confirm their identity. The cloning step permitted the
expression of IPTG-inducible, full-length, non-hexahistidine-tagged MAX
proteins in the pREP4 Escherichia coli strain (Qiagen).
Cultures were induced for 4-6 h with IPTG (final concentration 1 mM) at which time the bacteria were concentrated by
centrifugation, washed in PBS, and lysed by three rounds of freeze-thawing and sonication in 2 × binding buffer (35) that, for MAX(L), consisted of 100 mM KCl; 40 mM
Hepes, pH 6.5; 2 mM EDTA; 2 mM dithiothreitol;
10% glycerol plus 1 mM phenylmethylsulfonyl fluoride
(Sigma); and 1 µg/ml each of aprotinin, pepstatin, and leupeptin
(Boehringer Mannheim). For MAX(S), the identical buffer was used except
it's pH was adjusted to 7.6. Following clarification of each lysate at
10,000 × g for 10 min, a portion of the supernatant ("crude lysate") was used directly in DNA binding assays after adjusting for differences in MAX protein content using the pH 7.6 binding buffer. The remainder of the supernatant was applied to a
DEAE-cellulose column equilibrated with the above pH 6.5 or pH 7.6 2 × binding buffer. MAX(L) and MAX(S) proteins were eluted with a
stepwise KCl gradient, dialyzed against pH 7.6 2 × binding
buffer, and stored in small aliquots at
80 °C.
Hexahistidine-tagged N-Myc protein (35) was purified by nickel-agarose
affinity chromatography essentially to homogeneity and dialyzed against
2 × binding buffer. Preliminary control experiments indicated
that the different pH values used to bind MAX(L) and MAX(S) to
DEAE-cellulose did not influence their DNA binding properties following
their adjustment to pH 7.6 (not shown).
gal plasmid was routinely
included in all transfections to control for transfection efficiency.
Plasmid DNAs were purified using a standard alkaline lysis method
followed by two cycles of purification by CsCl/ethidium bromide
equilibrium centrifugation.
-galactosidase and luciferase (Analytical Laboratories, Ann
Arbor, MI). Stable NIH3T3 transfectants were obtained using the above
procedure except that 2 days after transfection, the monolayers were
split 1:5 and cultured in the presence of 400 µg/ml G-418 (Life
Technologies, Inc.). Individual G-418-resistant clones were isolated
using glass cloning cylinders.
DNA Binding by MAX Proteins
Fig. 1.
DNA binding by purified and crude MAX
proteins. MAX proteins were produced as unmodified, full-length
polypeptides in E. coli using the QE-60 expression vector
(Qiagen). MAX proteins were purified from IPTG-induced cultures by
non-denaturing DEAE-cellulose chromatography to yield individual MAX
proteins of >90% purity (A) or were used directly from
crude bacterial lysates of approximately 25% purity (B).
N-MYC was produced as a hexahistidine fusion protein (50) and purified
using nickel-agarose chromatography. Each protein (approximately 20 ng)
was tested for DNA binding with the 32P-labeled
double-stranded, palindromic EO(GAC) oligonucleotide (50) using
non-denaturing polyacrylamide gel electrophoresis. A, MAX(L)
showed strong binding to the 32-P-labeled probe (lane
2), whereas MAX(S) bound poorly (lane 3). The ability
of MAX(S) to dimerize with purified N-MYC and bind the probe
(lane 4) indicated that MAX(S) was in an active state. Similar results were obtained with crude extracts of the proteins (B), indicating that the differences in DNA binding were not
related to protein purification. Lane 2 shows the absence of
MAX(L) DNA binding activity in crude extracts of uninduced E. coli.
[View Larger Version of this Image (41K GIF file)]
Fig. 2.
Regulation of a c-Myc-regulated reporter by
MAX(L) and MAX(S). A, NIH3T3 cells were transiently
transfected with 5 µg of the 3xMyc-E1b-luc reporter plasmid and the
indicated amounts of MAX(L) (dark boxes) or MAX(S)
(open boxes) expression vectors. Luciferase activity was
measured 2 days later. The values shown are the averages of three
separate experiments ± 1 S.E. B, 5 µg of
3xMyc-E1b-luc was expressed either in the absence () or presence (+)
of an equivalent amount of pSVL-Myc expression vector (hatched boxes) along with the indicated amounts of pSVL-MAX(L) or MAX(S) expression vectors. Luciferase activities were determined as described in A.
[View Larger Version of this Image (19K GIF file)]
Fig. 3.
Immunoprecipitation of MAX proteins from
NIH3T3 cells. NIH3T3 cells were stably transfected with MAX(L) or
MAX(S) expression vectors and selected for G-418 resistance. Single
cell clones were expanded and metabolically labeled with
[35S]methionine/cysteine. MAX proteins were
immunoprecipitated from equivalent amounts of total cell lysate and
resolved by SDS-PAGE. Note that parental NIH3T3 cells and
vector-transfected control clones contain barely detectable amounts of
Max.
[View Larger Version of this Image (35K GIF file)]
Fig. 4.
Expression of c-Myc-responsive genes in
representative NIH3T3 cell clones over-expressing MAX(L) (clone L2) or
MAX(S) (clone S18) plus a vector-transfected clone (clone Neo7).
A, cells were transiently transfected with the
c-Myc-responsive 3xMyc-E1b-luc reporter plasmid (32) (filled
boxes) or the mutant p0Myc-E1b-luc mutant plasmid that lacks
functional c-Myc binding sites (32) (hatched boxes). Each
plate received 5 µg of each plasmid as well as 5 µg of
pSV2-gal to control for transfection efficiency. Two days later, cells were harvested and assayed for luciferase activity following correction for differences in
-galactosidase activity. The
results shown represent the average of three independent experiments, each performed in duplicate, ± 1 S.E. B, each of the
indicated clones was plated at a density of 3 × 104
cells/ml and allowed to grow for 3 days before harvesting. Endogenous Odc activity was then measured as described previously (47). The
results shown represent the average of three experiments ± 1 S.E.
[View Larger Version of this Image (24K GIF file)]
Fig. 5.
Growth curves of MAX-over-expressing NIH3T3
clones. 105 cells from the clones analyzed in Fig. 3
were seeded onto 100-mm tissue culture dishes and maintained in 10%
serum. Cells from triplicate plates were counted at the indicated
times.
[View Larger Version of this Image (15K GIF file)]
Fig. 6.
MAX(L)-over-expressing cells show a slower
rate of transit through all phases of the cell cycle.
Representative NIH3T3 clones were serum-starved for 24 h. At time
0, the cells were stimulated with medium containing 10% serum and 0.4 µg/ml nocodazole to prevent passage past the first G2/M
phase. At the indicated times after serum re-stimulation, cultures were
harvested to assess nuclear DNA content by fluorescence-activated cell
sorting.
[View Larger Version of this Image (29K GIF file)]
Fig. 7.
Accelerated death of
MAX(L)-over-expressing NIH3T3 cells following serum removal.
A, vector control clone Neo7 (), MAX(S) clone S3 (
),
and MAX(L) clone L2 (
) were seeded at an initial density of 2 × 105 cells/100-mm plate and allowed to grow for
approximately 24 h. The plates were washed and replaced with
serum-free medium at time 0. At daily intervals thereafter, cells from
both the monolayer and supernatant were harvested and stained with
trypan blue to determine the viable fraction. B,
phase-contrast photomicrographs of monolayers from each clone 4 days
after removal of serum. Note the relative paucity of viable,
fusiform-shaped cells in the MAX(L)-over-expressing cell line.
[View Larger Version of this Image (37K GIF file)]
Fig. 8.
Immunoprecipitation of MAX proteins from 32D
cells. Pooled G-418-resistant clones transfected with the
indicated plasmids were metabolically labeled with
[35S]methionine/cysteine. MAX proteins were
immunoprecipitated and analyzed by SDS-PAGE as described in the legend
to Fig. 3.
[View Larger Version of this Image (42K GIF file)]
Fig. 9.
Growth curves for 32D cells in different IL-3
concentrations. 2 × 104 pooled G-418-resistant
clones of 32D cells transfected with the empty vector alone (), with
pSVL-MAX(S) (
), or with pSVL-MAX(L) (
) were plated in medium
supplemented with 10% (A), 5% (B), or 2%
(C) WEHI-238-conditioned, IL-3-containing medium. Viable
cell counts were determined in triplicate cultures at the indicated intervals and are represented as the average number of cells/plate ± 1 S.E.
[View Larger Version of this Image (11K GIF file)]
Fig. 10.
Accelerated apoptosis in
MAX(L)-over-expressing 32D cells. A, logarithmically growing
control (), MAX(S) (
), or MAX(L) (
) over-expressing 32D cells
(>95% viability) were washed free of IL-3-containing medium and
cultured in IL-3-depleted medium for the indicated times. The
percentage of viable cells was determined by trypan blue staining.
B, nuclear DNA fragmentation was assessed in each cell line
either at the time of IL-3 withdrawal (0 h) or 24 h later. Note
that, at the latter time, MAX(L)-over-expressing cells demonstrate a
greater amount of DNA fragmentation, consistent with the results in
A.
[View Larger Version of this Image (19K GIF file)]
*
This work was supported by National Institutes of Health
Predoctoral Training Grant 538572 (to H. Z.) and by National
Institutes of Health Grant HL33741 (to E. V. P.).The costs of publication of this
article were defrayed in part by the
payment of page charges. The article
must therefore be hereby marked
"advertisement" in accordance with 18 U.S.C. Section
1734 solely to indicate this fact.
To whom correspondence should be addressed: Section of
Hematology/Oncology, Children's Hospital of Pittsburgh, 3705 Fifth Ave., Pittsburgh, PA 15213. Tel.: 412-692-6797; Fax:
412-692-5723.
1
The abbreviations used are: PAGE, polyacrylamide
gel electrophoresis; IPTG,
isopropyl-1-thio--D-galactopyranoside; PBS,
phosphate-buffered saline; ODC, ornithine decarboxylase; IL,
interleukin.
2
H. Zhang and E. V. Prochownik, unpublished
observations.
©1997 by The American Society for Biochemistry and Molecular Biology, Inc.