(Received for publication, May 17, 1996, and in revised form, November 11, 1996)
From the Departments of ¶ Molecular Cancer Biology,
Microbiology,
Medicine, and
§ Biochemistry, Duke University Medical Center,
Durham, North Carolina 27710
In recent years it has become apparent that the
cellular machinery governing cell cycle progression and transcription
control are often homologous in yeast and mammalian cells. We and
others have previously shown that the SP family of mammalian
transcription factors regulates the transcription of a number of genes
whose activities are governed by the product of the retinoblastoma (Rb) susceptibility gene, including c-FOS, c-MYC,
TGF-1, IGF-II, and c-JUN. To
determine whether a similar pathway of transcriptional regulation may
function in yeast, we explored the possibility that transcription
factors with nucleotide-binding specificities akin to those of the SP
family are expressed in Saccharomyces cerevisiae and
Schizosaccharomyces pombe. Here we report the detection of
novel yeast proteins (S. cerevisiae, p180; S. pombe, p200) that specifically bind Rb-regulated promoter
elements in vitro dependent on nucleotides that are also
required for binding and trans-activation by SP family
members in vivo. Our results indicate that the S. cerevisiae retinoblastoma control element-binding activity 1)
requires zinc for association with DNA; 2) does not bind to SCB, MCB,
or E2F sites in vitro; 3) is cell cycle-regulated in a
SWI6-independent fashion; and 4) maximally stimulates
retinoblastoma control element-mediated transcription in early- to
mid-S phase. Taken together, these data suggest that p180 may regulate
the transcription of a subset of yeast genes whose expression is
coincident with the onset and/or progression of DNA replication.
Functional inactivation of the retinoblastoma (Rb)1 protein is associated with the genesis of a number of human tumors, including retinoblastoma, osteosarcoma, and breast, bladder, and small cell lung carcinomas (for reviews see Refs. 1, 2). The Rb protein is believed to control cell proliferation at least in part via the transcriptional regulation of a wide variety of growth-related genes. Although capable of associating with DNA in a sequence-nonspecific fashion, Rb regulates transcription indirectly via its physical or functional interaction with trans-acting factors that specifically bind to DNA (3). To date, nearly a dozen sequence-specific DNA-binding proteins have been shown to be targets of Rb function in vivo. Interestingly, the functional consequence of Rb's interaction with these transcription factors is dependent on the factors themselves and the cell types in which their interaction is analyzed.
The transcriptional response of a subset of Rb-regulated genes,
including c-FOS, c-MYC, TGF-1,
IGF-II, and c-JUN, is dependent on GC-rich
promoter elements termed retinoblastoma control elements (RCEs, Refs.
3-8). At least three ubiquitously expressed nuclear proteins
(retinoblastoma control proteins, RCPs), including SP1 and SP3, bind to
RCEs in vitro, and the interaction of one or more of these
proteins with RCEs in vivo is required for RCE-mediated transcription (7, 9-11). Co-expression of Rb and SP1 or SP3 in
transient transfection assays leads to a marked stimulation of RCE
transcription, a phenomenon we have termed "superactivation" (10,
11). Regions of Rb that are targets of mutation in human tumors are
required for Rb-mediated superactivation, suggesting that the
functional interaction of Rb with SP1/SP3 plays a significant role in
the regulation of cell cycle progression (11). The mechanism(s) by
which Rb stimulates SP-mediated transcription has yet to be clearly
defined. Physical interactions between Rb and members of the SP family
of transcription factors have not as yet been detected in
vitro or in vivo perhaps suggesting that Rb interacts with these transcription factors in the context of a large
macromolecular complex. Consistent with this supposition, Rb has been
proposed to modulate the transcriptional activity of SP family members via their liberation from negative regulators or by indirectly "bridging" their trans-activation domains to components
of the basal transcription machinery (8, 12, 13).
Trans-activation mediated by transcription factors such as
ATF-2, NF-IL6, MYOD, and myogenin is also stimulated and/or facilitated
by Rb in vivo (14-17). Unlike SP family members, Rb forms
physical complexes with these latter factors although the mechanism by
which Rb augments their transcriptional activity has not as yet been
determined. In contrast to these functional effects, interactions of Rb
with transcription factors such as E2F, ELF-1, and UBF lead to the suppression of transcriptional activity (18-20). Rb forms cell cycle-regulated complexes with factors such as E2F and ELF-1 in vivo, sequestering their trans-activation domains from
components of the basal transcription complex (18, 19, 21). Hence, Rb
is believed to control the transcription of at least some cell cycle-regulated genes via periodic interactions with sequence-specific DNA-binding proteins.
The Rb protein is phosphorylated in concert with the progression of the mammalian cell cycle. Quiescent cells, post-mitotic cells, and cells in early G1 carry un- or underphosphorylated Rb protein (22-25). In cycling cells, Rb becomes increasingly phosphorylated on serine and threonine residues beginning in late G1 (the "restriction point") and extending through G2 and then is abruptly de-phosphorylated in anaphase (26). Rb is a substrate of a number of cyclin-dependent kinases (cdks), including cyclin D/CDK4 and cyclin E/CDK2 (27). In addition, a novel cell cycle-regulated Rb and histone H1 kinase has recently been described that associates with the Rb amino terminus in G2/M phases (28, 29). Given that the initiation of DNA synthesis occurs subsequent to Rb phosphorylation, it is widely suspected that phosphorylation of Rb is a necessary step for normal cells to transit through the G1/S boundary. This view is consistent with the observations that 1) transcription factors that control gene expression at the G1/S boundary, such as E2F, are bound exclusively by un- or underphosphorylated Rb; and 2) phosphorylation of Rb by cdks in vitro inactivates Rb as an inhibitor of E2F-mediated transcription (18, 21, 30).
The recent identification of homologues of E2F and Rb in Drosophila and suggestions of an Rb-like protein in plants serves to support the contention that a conserved pathway of transcriptional regulation may operate in many, if not all, eukaryotic cells (31-35). To date, a structural homologue of Rb has not been identified in yeast, but proteins that are similar to mammalian targets of Rb function have been noted. For example, a 47-kDa factor in Saccharomyces cerevisiae that specifically binds E2F sites in a cell cycle-dependent fashion has been reported, and a similar factor has been detected in Schizosaccharomyces pombe (36, 37). In concert with these findings, when expressed in S. cerevisiae Rb is phosphorylated by yeast cdks at sites that are targets of phosphorylation in mammalian cells (38). Moreover, Rb phosphorylation in yeast appears to be temporally controlled in a manner that is similar to that which occurs in mammalian cells; Rb is phosphorylated prior to the initiation of DNA synthesis in yeast coincident with the cell cycle checkpoint termed "Start" (27). Given these observations, it is tempting to speculate that yeast may harbor proteins functionally analogous to Rb that integrate progression of the cell cycle with transcriptional regulation. Yet, exogenous expression of human Rb in yeast does not appreciably alter cell cycle progression, suggesting that should yeast carry Rb-like proteins their targets of function may not be closely related to their mammalian counterparts (37).
In S. cerevisiae, cell cycle-regulated transcription of a
number of critical genes has been shown to be at least partly dependent on two heterodimeric transcription factors, SBF and MBF, and their interaction with SCB (SWI4-SWI6 cell cycle box) and MCB
(MluI cell cycle box) promoter elements, respectively (39,
40). SCB promoter elements (5-CACGAAA-3
) govern the periodic
transcription of genes such as the HO endonuclease and
cyclins (CLN1 and CLN2). MCB elements
(5
-ACGCGTNA-3
) direct the transcription of a variety of genes
including many required at the G1/S boundary for entry into
S phase, such as thymidylate synthase (TMP1) and B-type
cyclins (CLB5 and CLB6). SBF is composed of the
SWI4 and SWI6 proteins, whereas active MBF complexes result from the
heterodimerization of MBP1 and SWI6 proteins. Transcription factors
that are structurally and functionally similar to SWI4, SWI6, and MBP1
have also been isolated from S. pombe (40). Given that RCEs
share limited sequence homology (5
-GCGCCACC-3
) with yeast SCB and MCB
elements, we hypothesized that RCEs might represent a related family of
cell cycle-regulated yeast promoter elements. Furthermore, we
speculated that yeast RCE-binding proteins might be functionally, and
perhaps structurally, homologous with mammalian RCPs. In this report we characterize the biochemical and functional properties of a novel cell
cycle-regulated RCE-binding protein, p180, that is synthesized in
S. cerevisiae and whose DNA-binding domain is functionally homologous to that of SP1, SP3, and perhaps other members of the SP
family of mammalian transcription factors.
Table I lists the yeast strains used in this study. Cells were grown in YEPD (1% yeast extract, 2% peptone, 2% dextrose) at 30 °C or in selective SDmin media for strains transformed with URA3-based plasmids. For cell cycle studies, temperature-sensitive cdc mutants were grown in YEPD at 23 °C to an A600 nm of 0.4, pelleted, resuspended in prewarmed YEPD, and incubated at 37 °C for 5 h. Arrest was confirmed by documenting terminal arrest phenotypes under light microscopy. Cdc17 cells arrest heterogeneously at S phase (38 °C) or G2 (34 °C) and were treated at 38 °C in this study (41).
|
A high copy, URA3-based
plasmid (pJLB; 42) containing a UAS-less cytochrome c
(CYC1) promoter upstream of the LACZ gene was a
kind gift from Stephen Johnston (University of Texas-Southwestern, Dallas, TX). A synthetic oligonucleotide and its complement carrying three tandem copies of an octameric p180-binding site (pNUT;
5-TCGAGCGCCACCGCGCCACCGCGCCACC-3
) or a mutated derivative and its
complement lacking a p180-binding site (pWEE;
5
-TCGAGCTTCACCGCTTCACCGCTTCACC-3
)
were cloned upstream of the CYC1 promoter at a unique
XhoI site. Plasmids containing one or two copies of pNUT or
one to three copies of pWEE upstream of LACZ were identified
by double-stranded DNA sequencing (43) and named pNUT1, pNUT2, pWEE1,
pWEE2, and pWEE3, respectively. To further differentiate between
independent clones that were examined for
-galactosidase activity, a
lowercase letter was appended to each plasmid name (e.g.
pNUT2b is a second clone with two copies of a wild-type trimer). Cell
transformations and
-galactosidase assays were performed using
previously described protocols (44). To quantify p180-mediated
transcription as a function of cell cycle progression, a mutated
derivative of the CLN2 gene (Cln2x/s; a kind gift of David
Stuart, Scripps Research Institute, La Jolla, CA) encoding a
nonfunctional protein was cloned downstream of a
p180-dependent promoter by linking together a 1.5-kb DNA
fragment containing the pNUT2b promoter, a 1.5-kb DNA fragment of
pUC19CLN2x/s,2 and plasmid pRS306, a single
copy URA3-containing vector (45). A plasmid dependent on a
promoter lacking a p180-binding site, pWEE3a, was prepared in a similar
manner. The resulting plasmids, pNUT2bCLN2x/s and pWEE3aCln2x/s, carry
the defective CLN2 gene in the same transcriptional
orientation as URA3. High copy plasmids carrying these
CLN2 reporter genes were prepared by transferring pNUT- or
pWEE-dependent genes to plasmid pRS202, a
URA3-containing vector,3
producing plasmids P5Cln202 and W1Cln202, respectively.
Yeast extracts were prepared by
bead-beating 1 × 109 cells for 30 min at 4 °C
following their suspension in 1.0 M Tris-HCl (pH 7.5), 0.2 M NaCl, 5 mM EDTA, 20% glycerol, 90 mM -mercaptoethanol, 1 mM
phenylmethylsulfonyl fluoride, 20 µg/ml pepstatin A and leupeptin (Sigma). This suspension was clarified by
centrifugation at 16,000 × g at 4 °C. Supernatants
were transferred to tubes containing an equal volume of 0.5 M HEPES (pH 7.4), 30 mM KCl, 0.4 mM
EDTA, 20% glycerol, 1 mM dithiothreitol, 1 mM
phenylmethylsulfonyl fluoride, 20 µg/ml pepstatin A, and 20 µg/ml
leupeptin; protein concentrations were determined by a colorimetric
assay (Bio-Rad), and aliquots were stored at
80 °C. Radiolabeled
probes were prepared and employed in protein-DNA binding assays as
described previously (9, 46). Oligonucleotides used as radiolabeled
probes or competitor DNAs are listed in Table II.
Radiolabeled probes had a specific activity of 105 cpm/ng
DNA, and 1-2 × 105 cpm of probe was typically
combined with 5 µg of yeast proteins. Competition experiments with
unlabeled oligonucleotides typically employed a 50-200-fold molar
excess of DNA relative to radiolabeled probes. Following resolution on
polyacrylamide gels and transfer to paper, protein-DNA binding assays
were exposed to film (Kodak XAR-5) for 2 days at
80 °C or directly
analyzed in a PhosphorImager (Molecular Dynamics, Inc., Sunnyvale,
CA).
|
A plasmid carrying six copies of the
sequence 5-GCCACCCCTCT-3
(RCE7, Table II) was cleaved with
EcoRI and XbaI liberating a DNA fragment that was
used as a template for primer extension from the EcoRI end
of the probe. Primer extension was carried out using a synthetic primer
(5
-CGAGCTCGCCC-3
), bromodeoxyuridine (BUdR, Sigma),
dTTP, dATP, radiolabeled dGTP and dCTP (3000 Ci/mmol; ICN Biomedicals,
Inc., Irvine, CA), and Klenow enzyme (New England BioLabs, Beverly, MA)
as described elsewhere (9). A BUdR-substituted probe (12 × 106 cpm) with a specific activity of 109 cpm/ng
was employed in a protein-DNA binding assay with 35 µg of yeast
proteins. Following resolution on a polyacrylamide gel, protein-DNA
complexes were irradiated by UV light in situ and visualized
by exposure of the gel to film, and excised complexes were applied to
an SDS-polyacrylamide gel. Following electrophoresis, dried gels were
exposed to film for 2 days at
80 °C. Apparent molecular weights of
resulting protein-DNA complexes were determined by comparison with
molecular weight markers resolved in parallel.
Yeast
cultures were grown in YEPD to approximately
A600 nm = 0.5 at 30 °C, and factor
pheromone (Sigma) was added to a final concentration
of 10 ng/ml. After 2.5 h of incubation, cell cycle arrest was
confirmed by microscopic examination. Cells were then pelleted,
resuspended in 3 ml of sterile deionized water, sonicated, and added to
fresh, prewarmed YEPD media. Subsequently, 50 ml of cells were
collected by centrifugation at 15-min intervals for the preparation of
protein extracts or RNA. Small aliquots of cells were also fixed in 4 volumes of 3.7% formaldehyde, 0.15 M NaCl to establish a
budding index by microscopic inspection. Fresh, prewarmed YEPD was
added every 45 min during synchronous growth to maintain a stable
concentration of cells. Cultures of cells containing high copy reporter
plasmids were grown in YEPD because synchrony was not optimal in
selective minimal media. No more than 7% of cells (as determined by
replica plating onto YEPD and selective minimal media plates) lost
their respective plasmids during synchronous growth in YEPD.
RNA extracts were obtained using a modification of a previously described procedure (47). Briefly, cells were resuspended in a sodium acetate buffer and treated with 1% SDS. Phenol equilibrated in sodium acetate buffer was added to disrupted cells and incubated at 70 °C for 4 min with periodic agitation. The mixture was cooled rapidly for 10-15 s in an ethanol/dry ice bath and subsequently clarified by centrifugation. The aqueous phase was re-extracted with phenol and finally with chloroform, and RNA was precipitated and resuspended in distilled H2O. Following quantification by UV spectroscopy, 10 µg of RNA was resolved in a 1.8% agarose-formaldehyde gel as described previously (48) and then transferred to BioTrans+ filters (ICN Biomedicals, Inc. Irvine, CA). Blots were baked, prehybridized, and hybridized with 2 × 106 cpm/ml (specific activity of 0.8-1.0 × 109 cpm/µg DNA) of radiolabeled probe. Radiolabeled probes used for these studies were prepared from a 1.5-kb BamHI fragment of pUC19Cln2x/s, a BamHI-BglII fragment of pSKACT1 (ACT1 probe), and a HindIII fragment of pMS202 (copy II of the yeast histone H4 gene; Ref. 49). Probes were prepared using a random primer kit following instructions of the manufacturer (Promega, Inc., Madison, WI). Following washing, blots were quantified by PhosphorImaging (Molecular Dynamics) and subjected to autoradiography (Hyperfilm-MP, Amersham Corp.). Blots were stripped prior to the addition of supplementary probes by agitating in several exchanges of boiling 0.1% SDS.
Analysis of S. cerevisiae Genome Data BaseThe S. cerevisiae genome data base maintained by the Stanford University Department of Genetics was examined for the occurrence of RCE sites by employing the FASTA algorithm (50) available at their website (http://genome-www.stanford.edu/Saccharomyces/).
To determine
whether S. cerevisiae and/or S. pombe express
RCE-binding proteins (RCE-BPs), a 31-base pair oligonucleotide corresponding to the human c-FOS RCE was radiolabeled and
employed with yeast extracts in protein-DNA binding ("gel-shift")
assays. As shown in Fig. 1, several protein-DNA
complexes resulted when this oligonucleotide was incubated with either
yeast extract. To determine if one or more of these protein-DNA
complexes results from the specific association of yeast proteins with
the c-FOS RCE, excess unlabeled homologous or heterologous
oligonucleotides were added as competitor DNAs. For both S. cerevisiae and S. pombe extracts, a single slowly
migrating protein-DNA complex is eliminated by excess unlabeled
homologous oligonucleotides (Fig. 1, +Fos) but not by an
irrelevant oligonucleotide carrying binding sites for the transcription
factor AP-1 (Fig. 1, +AP-1). The abundance of additional
faster migrating protein-DNA complexes was not appreciably affected by
the inclusion of either competitor oligonucleotide, and we have found
that their abundance in yeast extracts is quite variable, often being
undetectable (for example see Fig. 2). To determine if
yeast RCE-BPs require nucleotides for complex formation that are shared
with mammalian RCPs, we employed two mutated RCE oligonucleotides that
have previously been shown to strongly interact (Fig. 1,
5Fos-4) or not interact (Fig. 1,
5
Fos-5) with SP1 and SP3 (9). As shown in Table
II, oligonucleotides 5
Fos-4 and 5
Fos-5 include 18 nucleotides derived
from the 5
-half of the c-FOS RCE and carry dinucleotide
substitutions that respectively increase or abolish mammalian
RCP-binding activity relative to wild-type sequences. Akin to our
previous result with mammalian RCPs, the abundance of the slowest
migrating yeast protein-DNA complexes was abolished by inclusion of an
excess of 5
Fos-4 and not appreciably diminished by 5
Fos-5 (Fig. 1).
Wild-type oligonucleotides corresponding to RCEs within the
TGF
-1 and c-MYC promoters also eliminated the
slowest migrating protein-DNA complex when employed as competitors in
parallel protein-DNA binding assays (data not shown). Thus, S. cerevisiae and S. pombe express at least one protein
that specifically binds RCEs in vitro. As is more clearly illustrated in the rightmost panel of Fig. 1, we note that
the specific protein-DNA complex detected in S. cerevisiae
extracts migrates slightly faster than its counterpart in S. pombe extracts.
Yeast and Mammalian RCE-binding Proteins Require Identical Nucleotides and Zinc for DNA Binding
To extend our comparison of
mammalian and yeast RCE-binding proteins and map nucleotides that are
important for protein-DNA complex formation, protein-DNA binding assays
were performed in the presence of a large panel of synthetic
oligonucleotides whose RCP-binding activities have been previously
established with mammalian cell extracts (9). As shown in Table II,
these additional oligonucleotides carry hexanucleotide or dinucleotide
substitutions within discrete portions of the c-FOS RCE.
Consistent with our previous results for SP1, the slowly migrating
protein-DNA complex detected in S. cerevisiae extracts was
abolished in competition experiments with the 5 (5
Fos-WT) and the 3
(3
Fos-WT) octameric repeats of the c-FOS RCE (Fig.
2A; Ref. 10). These results indicate that as for SP1, both
octameric sites within the c-FOS RCE are bound by yeast
RCE-BPs. This conclusion is supported by evidence that mutational
disruption of both octameric sites (dbl RCP
) generates an
oligonucleotide with little or no protein-binding activity, whereas
mutations within either the 5
(5
RCP
) or 3
(3
RCP
) octameric
repeat result in oligonucleotides that function as competitors (Fig.
2A). To further map the nucleotides required for yeast
RCE-binding activity, synthetic oligonucleotides containing dinucleotide mutations within the 5
18 bases of the c-FOS
RCE were employed as competitor DNAs. An excess of mutants 5
Fos-1, -2, -3, and -5, which contain dinucleotide mutations within the octameric
repeat, were unable to compete for the formation of the single yeast
protein-DNA complex. In contrast, inclusion of synthetic
oligonucleotides carrying dinucleotide substitutions outside the
octameric site (mutants 5
Fos-4, -6, -7, and -8) did not diminish
RCE-binding activity. Taken together, these data indicate that the
minimum RCE nucleotides required for binding of yeast RCE-BPs is the
octameric 5
-GCGCCACC-3
sequence. A similar series of experiments
performed with S. pombe extracts gave identical results
(data not shown). To confirm that mutated oligonucleotides carrying
dinucleotide substitutions retained or lost their ability to bind yeast
RCE-BPs, a panel of wild-type and mutated c-FOS oligonucleotides were examined directly for yeast RCE-BP-binding activity. Each oligonucleotide was radiolabeled and incubated with
yeast extracts, and protein-DNA complexes were resolved by electrophoresis. Consistent with earlier protein-DNA binding assays, those oligonucleotides that do not function as competitors (mutants 5
Fos-1, -2, -3, and -5) also did not form protein-DNA complexes when
examined as radiolabeled probes (Fig. 2B). Additionally, mutated oligonucleotides that function as competitors had wild-type protein-binding activity (Fig. 2B, mutants 5
Fos-6, -7, and
-8). Interestingly and consistent with our previous results using
mammalian cell extracts, mutant 5
Fos-4 has greater protein-binding
activity than wild-type RCE oligonucleotides (Fig. 2B). In
summary, we conclude from these results that extracts prepared from
S. cerevisiae and S. pombe carry at least one
protein that forms specific protein-DNA complexes with the
c-FOS RCE. Moreover, nucleotides required for the formation
of these complexes are identical to those required for the binding of
mammalian RCPs, such as SP1 and SP3. These data suggest that the
DNA-binding domains of yeast and mammalian RCE-binding proteins are
functionally, and perhaps structurally, conserved. Consistent with the
notion that yeast RCE-BPs may be structurally related to SP family
members, addition of potent zinc chelating agents, such as EDTA or
1,10-phenanthroline, to yeast extracts inhibits the formation of
protein-RCE complexes in a zinc-dependent manner (Fig.
2C and data not shown).
Since
RCE nucleotides required for the binding of yeast proteins in
vitro are similar to those found in yeast SCB and MCB elements, we
wished to determine whether such elements could compete for the
formation of yeast RCE-BP·RCE complexes. Additionally, since the
RCE-BP-binding site of a mutated RCE oligonucleotide with increased
affinity for yeast proteins (5Fos-4; 5
-TTGCGCCACC-3
) resembles the
consensus site for the mammalian transcription factor E2F
(5
-TTTCGCGC-3
), we wished to determine whether yeast RCE-BPs might be
related to a previously described yeast E2F-like activity (36). Thus,
oligonucleotides carrying E2F sites from the mouse DHFR and
adenovirus E2 promoters as well as oligonucleotides carrying consensus SCB and MCB elements were analyzed as competitors in protein-DNA binding assays (Table II and Ref. 52). As shown in Fig.
3A, inclusion of a 50- and 100-fold molar
excess of oligonucleotides (HIP and E/J; Ref. 51) carrying wild-type
and mutated E2F sites from the DHFR promoter did not result
in a reduction in RCE protein-DNA complexes. In contrast, an E2F site
containing oligonucleotide (E2-WT) derived from the E2
promoter diminished protein-RCE complexes; however, a mutated
derivative (E2-Mut) in which the E2F sites were destroyed could also
effectively compete for DNA-binding activity (52). Taken together,
these data indicate that the DNA-binding domains of yeast RCE-binding
proteins are not likely to be functionally analogous to that of E2F. To
determine if yeast RCE-BPs bound SCB and MCB elements in
vitro, oligonucleotides carrying consensus elements were examined
in protein-DNA binding assays. As shown in Fig. 3A, a
100-fold molar excess of these unlabeled oligonucleotides did not
diminish the recovery of RCE protein-DNA complexes. To establish if
factors important for the binding and trans-activation of
SCB and MCB elements may be necessary for yeast RCE-binding activity,
extracts were prepared from yeast strains carrying disruptions of
SWI4 or SWI6, and these extracts were employed in
protein-DNA binding assays. As shown in Fig. 3B, disruption
of these genes had no discernible effect on the abundance or mobility
of protein-DNA complexes. Thus, we conclude that 1) under the
conditions we have employed, yeast RCE-BPs do not bind to E2F, SCB, and
MCB elements in vitro and 2) yeast RCE-BPs do not require
SWI4 or SWI6 for RCE-binding activity.
Yeast RCE-binding Activity Consists of at Least One High Molecular Weight Protein
To determine the number and apparent molecular weight(s) of the yeast RCE-binding proteins, we employed a photoaffinity labeling technique (UV cross-linking) with a multimerized RCE probe (RCE7) that was radiolabeled and substituted with bromodeoxyuridine (BUdR). This probe contains six contiguous copies of an oligonucleotide that spans the middle six bases of the octameric sequence previously shown to be critical for the binding of yeast RCE-BPs (Table II). We have previously used this multimerized probe for the analysis of mammalian RCPs, and prior to UV cross-linking experiments we confirmed that RCE7 would efficiently compete for the binding of yeast RCE-BPs (data not shown). Following radiolabeling and substitution with BUdR, RCE7 was incubated with extracts prepared from S. cerevisiae and S. pombe; protein-DNA complexes were resolved on nondenaturing gels, and bound proteins were covalently cross-linked to the probe in situ via irradiation with UV light. Protein-DNA complexes were excised and subsequently resolved on an SDS-polyacrylamide gel. As shown in Fig. 3C, a single protein-DNA complex with an apparent molecular mass of 180 kDa (denoted p180) was recovered from S. cerevisiae extracts. Consistent with the lessened relative mobility of protein-DNA complexes produced by extracts prepared from S. pombe, similar cross-linking assays performed with S. pombe extracts resulted in the detection of a single protein-DNA complex of approximately 200 kDa (Fig. 3C). The simplest interpretation of these data is that S. cerevisiae and S. pombe express a single large RCE-binding protein that may be cross-linked to DNA. However, we are mindful of the possibility that additional yeast proteins may participate in RCE protein-DNA complexes that were not detected by UV cross-linking. The remaining studies in this report were performed with S. cerevisiae cells and extracts.
RCE-mediated Transcription Is Dependent on an Intact p180-binding SiteTo determine if an intact p180-binding site is required for
RCE-mediated transcription, a multimer carrying three tandem copies of
the octameric site shown to be a target for p180 binding (pNUT; Fig.
4A and Table II) or a multimer carrying three
copies of a mutated octamer that lacks a binding site for p180 (pWEE;
Fig. 4A and Table II) were cloned upstream of a
LACZ reporter gene whose expression is directed by a basal
CYC1 promoter (42). In control experiments p180 bound pNUT
but not pWEE when examined in protein-DNA binding assays both as probes
and unlabeled competitors (data not shown). Asynchronously growing
populations of yeast transformed with either construction or the parent
plasmid were permeabilized with SDS and chloroform and assayed for
-galactosidase activity. As shown in Fig. 4B, plasmids
carrying one or two copies of the pNUT trimer increased
-galactosidase activity 31- and 370-fold, respectively, relative to
the parent plasmid lacking a p180-binding site. Similar levels of pNUT
trans-activation were also obtained in cells lacking
SWI6 function (BY600; data not shown). pNUT-mediated
trans-activation was also independent of the orientation of
the octameric repeat cloned upstream of
-galactosidase (Fig.
4B). In contrast to results with pNUT, one to three copies of the pWEE trimer resulted in little or no increase in
-galactosidase activity (Fig. 4B). We conclude from these
results that an intact p180-binding site is required for RCE-mediated
yeast transcription. Given that a dinucleotide substitution that
ablates p180-binding activity in vitro also inactivates
transcriptional activity in vivo, these data also strongly
suggest that p180 can function as a stimulatory transcription
factor.
p180 Abundance and/or DNA-binding Activity and Resulting Transcription Is Cell Cycle Regulated
Should RCEs function as
transcriptional targets of a Rb-like pathway in yeast, we speculated
that p180 DNA-binding and/or transcriptional activity might vary in
concert with cell cycle progression. This speculation was buoyed by the
observation that extracts prepared from cells arrested with nocodazole
showed significantly lower p180 DNA-binding activity relative to
extracts prepared from asynchronously growing cultures (data not
shown). To further examine the abundance of p180 DNA-binding activity
during the cell cycle, extracts were prepared from a panel of S. cerevisiae temperature-sensitive cdc mutants grown at the
permissive and nonpermissive temperatures. All cdc mutant strains grown
asynchronously at the permissive temperature (30 °C) contained
significant amounts of p180 DNA-binding activity (Fig.
5, top row of left panel). At the
nonpermissive temperature (38 °C), mutants whose functions are
required during G1, G2, or M phases showed
little or no p180 DNA-binding activity (Fig. 5, center row
of left panel), yet these cell extracts are clearly
functional since they retained GCN4 DNA binding activity (Fig. 5,
bottom row of left panel). In contrast, mutants
whose functions are required in early to mid-S phase (cdc6, cdc7, and cdc17) contained wild-type levels of p180
DNA-binding activity. Interestingly, a distinct S phase mutant,
cdc2, did not have detectable p180 DNA-binding activity at
the nonpermissive temperature for function (Fig. 5, center
row of left panel). Importantly, the abundance of
p180·DNA complexes were not appreciably altered in extracts prepared
from a wild-type strain grown at either temperature. Although these
results suggested that p180 abundance and/or DNA-binding activity is
maximal during early to mid-S phase, we were concerned that these
results might be compromised by artifacts induced by cell cycle arrest
and wished to perform similar analyses with synchronously growing cell
populations. Thus, cells were synchronized by incubation with factor, extensively washed and incubated in growth medium, and protein
extracts were prepared in 15-min intervals. Equivalent amounts of total
cell proteins were subsequently examined in protein-DNA binding assays
(Fig. 6A). Microscopic inspection of
factor-treated cells showed that greater than 95% of cells entered S
phase synchronously during the course of two population doublings (Fig.
6B). Consistent with evidence from growth-arrested cdc
mutants, maximal p180 DNA-binding activity was detected at time points
that immediately precede the peak of accumulated budded cells (Fig. 6).
The abundance of p180 DNA-binding activity throughout the cell cycle
was directly quantified by PhosphorImaging and determined to vary at
least 2-fold during the course of two cell cycles.
Since the abundance of p180 DNA-binding activity varies during cell
cycle progression, becoming maximal during S phase, we wished to
determine whether p180-mediated transcription was similarly regulated.
The promoter regions from pNUT2b, containing six copies of a wild-type
p180-binding site upstream of a basal CYC1 promoter, or from
pWEE3a, containing nine copies of a mutated p180-binding site, were
cloned upstream of a disrupted CLN2 gene (clnx/s) and inserted into a high copy URA3-containing vector (pRS202).
The resulting constructions (P5Cln202 and W1Cln202, respectively) were
subsequently used to transform a yeast strain (DLY204) carrying a
disruption of CLN2, and Northern blots carrying total RNAs
from cells containing either construction were examined for exogenous CLN2 mRNA with a CLN2-specific probe. DLY204
has previously been shown to express an unstable CLN2
mRNA of 0.6 kb.4 Each transformed yeast
strain expressed a minor transcript of 2.0 kb (Fig.
7A, top panel; asterisk) whose synthesis was
independent of the integrity of plasmid-borne p180-binding sites. In
contrast, a single prominent transcript of 3.0 kb was expressed
exclusively in cells carrying plasmids with intact p180-binding sites
(Fig. 7A, top panel; CLN). To determine whether the
abundance of the 3.0-kb pair mRNA was cell
cycle-dependent, cells containing P5Cln202 were
synchronized with factor and then incubated in growth medium. Total
RNA was isolated from synchronized cells at 15-min intervals; CLN2 expression was examined with a CLN2-specific
probe, and the abundance of the 3.0-kb CLN2 mRNA was
directly quantified in a PhosphorImager and normalized to the abundance
of actin mRNA (Fig. 7A, middle panel). In successive
experiments, the abundance of p180-dependent
CLN2 mRNA varied by an average of 3-4-fold during the
cell cycle with peaks of transcription occurring coincident with
maximal p180 DNA-binding activity (compare Figs. 6B and
7B). In contrast to these results, the abundance of the
2.0-kb CLN2-related mRNA was largely unchanged during
the course of these experiments (Fig. 7A, top panel). As an
additional measure of cell cycle synchrony and to mark the position of
S phase, Northern blots probed for CLN2 mRNA were
re-examined for the abundance of histone H4 message. With
the exception of RNA harvested from
factor-arrested cells, maximal
amounts of p180-dependent CLN2 mRNA were
detected coincident with the accumulation of histone H4
mRNA (Fig. 7A, bottom panel; Ref. 53). We do not as yet
understand why residual CLN2 mRNA is apparent in
factor-arrested cells. Nonetheless, for populations of synchronously
growing cells, cell cycle-regulated fluctuations in p180 DNA-binding
activity in vitro are temporally correlated with periodicity
in the abundance of RCE-mediated transcription in vivo.
Moreover, RCE-mediated transcription is maximal during early- to mid-S
phase.
The promoters of a number of Rb-responsive genes (e.g.
c-FOS, c-MYC, TGF-1,
IGF2, IL6, c-JUN, and
c-NEU) carry GC-rich sequences, termed retinoblastoma
control elements (RCEs) that are necessary and sufficient for
Rb-mediated transcriptional regulation (3). Previous analyses have
determined that RCEs are bound by several mammalian proteins (RCPs)
in vitro, and these proteins have been revealed to be
members of the SP family of transcription factors. To determine if an
RCE-like pathway of transcriptional control exists in lower eukaryotes,
we sought to identify RCE-binding proteins (RCE-BPs) in yeast cells. We
reasoned that if RCEs are evolutionarily conserved promoter elements
then yeast RCE-BPs may be functional homologues of Rb-targeted
transcription factors (e.g. SP1 or SP3), and the activities
of such proteins might be regulated by a yeast Rb-like molecule. This
report characterizes a novel transcription factor expressed in S. cerevisiae, termed p180, that specifically binds to RCEs during
early- to mid-S phase utilizing nucleotides that are required for
binding and trans-activation by SP1/SP3. Directly correlated
with this binding activity is a cell cycle-regulated stimulation of
RCE-mediated transcription in vivo. We also report that a
200-kDa RCE-binding protein (termed p200) with similar RCE-binding
activity is expressed in S. pombe.
Based on photoaffinity labeling experiments, a single high molecular weight RCE-binding protein expressed in S. cerevisiae and S. pombe may be cross-linked to DNA in vitro. We note that these data do not preclude the possibility that additional yeast proteins participate in p180·DNA and p200·DNA complexes or that RCEs are bound and regulated by as yet undetected proteins in vivo. However, given the tight correlation between the nucleotide binding specificity of p180 in vitro and the transcriptional activity of wild-type and mutated RCEs in vivo, this latter possibility is unlikely. Consistent with our cross-linking results, preliminary affinity chromatography experiments indicate that RCE-binding activity in S. cerevisiae extracts co-fractionates with a polypeptide of approximately 180 kDa as well as several smaller proteins.5 Determining whether one or more of these proteins account for p180 DNA-binding activity will require additional rounds of protein purification. We also note that the large apparent molecular weight of yeast RCE-binding proteins is not without precedent as a number of S. cerevisiae transcription factors ranging between 170 and 220 kDa have previously been noted, including MOT1, RIF1, SIN3, and SNF2 (54, 55).
p180 and p200 bind RCEs via nucleotides that are identical to those
bound by SP1 and SP3. Dinucleotide substitutions introduced within a
10-base pair sequence, 5-GCGCGCCACC-3
, within the c-FOS RCE dramatically alter the recovery of yeast protein-RCE complexes in vitro and perturb RCE-mediated transcription in
vivo. For example, several dinucleotide substitutions within the
RCE (5
Fos-1, -2, -3, -5) ablate the binding of p180/p200 and SP family
members to RCEs, and a single dinucleotide substitution (5
Fos-4)
increases their DNA-binding activity concordantly. That identical
mutations similarly affect the DNA-binding activities of yeast and
mammalian RCE-binding proteins strongly suggests that their DNA-binding domains are functionally homologous. It is less certain, however, that
their respective DNA-binding domains are closely related in structure.
Consistent with the notion that yeast and mammalian RCE-binding
proteins may be structurally related, we have shown that each requires
zinc for DNA-binding activity. The SP family of transcription factors
are well characterized "zinc-finger" proteins carrying three tandem
zinc-binding motifs of the Cys2-His2 class
(56). Although likely to be metalloproteins, whether p180/p200 will be
"zinc-finger" proteins of the same or a similar class will require
their eventual cloning and sequencing. Despite their common nucleotide
specificities and zinc requirement, three additional observations
suggest that the primary sequence of the SP family members are not
likely to be closely related to that of p180 or other yeast
transcription factors. First, using polyclonal antisera prepared
against the entirety of the SP1 trans-activation domain, we
have been unable to deplete yeast extracts of p180.5
Second, the expression of SP1 in S. cerevisiae does not
result in the trans-activation of reporter genes regulated
by several SP1-binding sites (57). Moreover, co-expression of a
component of the human basal transcription complex, TATA-box binding
protein, did not facilitate SP1-mediated yeast transcription. This
latter result may indicate that yeast lack one or more general
transcription factors necessary for SP1-mediated transcription, perhaps
co-activators that bridge glutamine-rich trans-activation
domains to the basal transcription complex. Finally, expression of
human Rb in S. cerevisiae does not appreciably alter yeast
cell growth or progression of the cell cycle (38). Thus, should
p180/p200 be structurally related to SP1/SP3 we anticipate that their
homology may not extend further than their respective DNA-binding
domains.
In accord with the notion that p180 is a cell cycle-regulated
transcription factor, we have consistently noted that maximal levels of
RCE-mediated transcription are coincident with the peak of histone
H4 mRNA abundance. Although we have observed on average a 3-4-fold difference in cell cycle-regulated RCE-mediated
transcription, it is possible that this may not accurately reflect the
magnitude of periodic p180 activity for the following reasons. First,
our trans-activation assays were performed with a promoter
construct carrying a wild-type c-FOS RCE cloned upstream of
a basal promoter. It is not as yet clear that the c-FOS RCE
is an optimal binding site for p180, and it is possible that small
perturbations in sequence can greatly affect p180-mediated
transcription. For example, dinucleotide substitutions at two distal
positions within the RCE (5Fos-4) result in a substantial increase in
p180 DNA-binding activity and resulting trans-activation
in vivo.5 Second, it is also likely that the
promoter context of p180-binding sites and their juxtaposition to other
trans-acting factors and the site of transcriptional
initiation may play important roles in the regulation of periodic
p180-mediated transcription. Thus, a more definitive cell cycle
analysis of p180-mediated transcription will require the identification
of genes regulated by p180 and a careful examination of their
promoters. Given that p180 DNA-binding activity and RCE-mediated
transcription are confined to a temporal "window" of early- to
mid-S phase, it is tempting to speculate that p180 may play a role in
the trans-activation of 1) genes associated with the
biogenesis of DNA and/or 2) other S phase genes, such as histones and
the B-type cyclins CLB3 and CLB4 (58, 59).
Although sequences closely related to the c-FOS RCE are not
contained within the promoters of these S phase genes, evaluating whether p180 plays a role in their regulation will require the cloning
and functional characterization of p180. Interestingly, a
computer-assisted search of the S. cerevisiae genome data
base with the c-FOS RCE revealed predicted sites of p180
binding upstream of the KIP1 and UBC13 genes.
KIP1 is a kinesin-related protein required for spindle assembly, and
its abundance is maximal in M phase (60).6
UBC13 is thought to encode a ubiquitin-conjugating enzyme;
however, it is not known whether its expression is cell
cycle-regulated. Whether p180 plays a direct role in the synthesis of
these genes remains to be determined. Since p180 does not bind SCB and
MCB elements in vitro or require SWI4/SWI6 proteins for
DNA-binding or transcriptional activity, we hypothesize that p180 may
function as a regulator of a novel subset of periodically transcribed
RCE-dependent yeast genes. However, since it is possible
that our in vitro DNA binding assays may be insensitive to
weak interactions of p180 with SCB/MCB elements, p180-mediated
transcription may also partially account for cell viability in the
absence of cell cycle regulators such as MBF (39).
We are indebted to Daniel Lew for helpful discussions and for providing yeast strains as well as K. Helen Kranbuhl and Philip Borden for technical assistance. We also thank Daniel Burke and Linda Breeden for providing mutant yeast strains, Peggy Farnham for oligonucleotides, and David Stuart and Joseph R. Nevins for plasmids used in this study.