(Received for publication, November 7, 1996, and in revised form, January 14, 1997)
From the Fred Hutchinson Cancer Research Center,
Basic Sciences Division, Seattle, Washington 98104
Two promoter elements have been defined that activate G1/S-specific transcription in Saccharomyces cerevisiae. SCB elements (CACGAAA) are activated by the Swi4-Swi6 complex, and MCB elements (ACGCGTNA) are activated by the Mbp1-Swi6 complex. CLN1 encodes a cyclin which is expressed during this interval, and requires Swi4 and Swi6 for peak transcription, but it has no consensus SCB elements in its promoter. Two SCB-like sequences had been previously noted and suggested to be the functional promoter elements. Our studies indicate that these sequences are unable to activate transcription of a lacZ reporter construct, or to bind Swi4-Swi6 complexes in vitro. However, a cluster of three sequences resembling MCB sequences are active promoter elements, sufficient to confer G1/S-specific transcription to a reporter. These sites are the predominant activation elements in the CLN1 promoter, and despite their resemblance to MCB elements, they bind Swi4-Swi6 complexes in vitro and require Swi4 and Swi6 for their activity in vivo. This indicates that the sequences that promote Swi4/Swi6 binding have not been fully defined, or that there are multiple Swi4- and Swi6-containing complexes with distinct DNA binding specificities. In addition to these novel Swi4/Swi6-binding sites, these studies also show that there must be at least one novel promoter element that can confer G1/S-specific transcription to CLN1, because when all the potential SCB- and MCB-like sequences are eliminated the transcript is still cell cycle regulated.
Passage of cells through the cell cycle of Saccharomyces cerevisiae requires the activation of the Cdc28 kinase. Cdc28 is activated by association with different regulatory subunits, called cyclins. As the name implies, cyclins are unstable proteins whose levels vary within the cell cycle, primarily by the cell cycle regulation of transcription (1, 2). During late G1, there are two predominant cyclins (Cln1 and Cln2) which can form active complexes with Cdc28. The promoters of both of these genes show strong cell cycle regulation which is reflected both at the level of the protein and associated Cdc28 kinase activity (3, 4).
The timing of expression of CLN1 and CLN2 is similar to that of the HO endonuclease and many of the DNA synthesis genes of S. cerevisiae. Transcription of the HO endonuclease is absolutely dependent on the activity of two transcription factors, Swi4 and Swi6, which form a complex and bind to repeated sequence elements within the HO promoter known as Swi4/Swi6-regulated cell cycle boxes (SCBs)1 (5-7). Swi4 is the DNA-binding component of the complex (6, 8) and brings Swi6 to the DNA via association between their COOH termini (8-10). Many of the DNA synthesis genes are also transiently expressed in late G1 and their cell cycle regulated transcription is also dependent on Swi6 (11, 12). However, transcription of these genes seems to be mediated by an alternative DNA sequence element, the MluI cell cycle box (MCB) (13, 14), which is bound by a complex of Swi6 (11, 12, 15) with a different DNA binding partner, Mbp1. MBP1 was identified through its homology to SWI4, and like Swi4, Mbp1 binds DNA through the amino-terminal region of the protein, and associates with Swi6 through its COOH terminus (16).
Deletion of SWI4 and SWI6 is lethal (17) and deletion of SWI4 alone is lethal in some strains (18). To identify the essential transcriptional targets of the Swi4-Swi6 complex, high copy suppressors of this lethality were sought. CLN1 and CLN2 were identified in these screens (18, 19) and their transcription was shown to be largely dependent upon Swi4 and Swi6. Furthermore, complexes formed in vitro on the SCB sequences within the HO and CLN2 promoters include the Swi4 and Swi6 proteins (6, 18, 19).
Although CLN1 has been considered to be coordinately regulated with CLN2, no direct analyses of the CLN1 promoter have been performed. Within the CLN1 promoter, there are two SCB-like sequences, CTCGAAA, which differ from the consensus SCB sequence at the second position, with T replacing A. Because of the similarity of the SCB-like sequence to the consensus, and because of the transcriptional induction of CLN1 by the Swi4 and Swi6 factors, it has been widely thought that the CLN1 SCB-like sequences are functional in binding Swi4/6 and mediating cell cycle-regulated expression. However, mutational analysis has indicated that all the bases within the SCB element are critical for its activity (20). To clarify this issue, we have characterized the CLN1 promoter and found that the SCB-like sequences play little or no role in transcriptional induction of CLN1. Instead there are three MCB-like sequences that are critical for activation and cell cycle regulation of this promoter. Furthermore, we find that Swi4 plays a more important role in binding and activating these MCB-like sites than does the "MCB-binding protein" Mbp1. In addition to the novel Swi4-Swi6 complexes that form on the CLN1 promoter, our studies show that there is at least one other cell cycle regulatory element which also confers G1/S-specific transcription to the CLN1 promoter.
Strains BY 602 (Mata, ade2-1, trp1-1, can 1-100, leu2-3, 112, his3-11,15, ura3, met, HO::lacZ46) and the related
swi4 and swi6 deletion strains (BY606 and BY600)
have been described previously (21), as have the mbp1
deletion strain K3294 and its isogenic wild type strain K1107 (16). All
these strains were derived from W303-1a, but they are not isogenic with
it. The BY1231 wild type is isogenic with W303-1a
(Mata, ade2-1, trp1-1, can 1-100, leu2-3,112,
his3-11,15, ura3). Yeast cultures were grown in YEP medium or YC
minimal media supplemented with essential nutrients as required and
containing 2% glucose (22). For cell cycle synchronization, cells were
grown overnight in media lacking either uracil or tryptophan,
transferred to rich media and grown to A600 of 0.2, before addition of 4-5 µg/ml -factor.
-Factor arrests were
monitored by phase microscopy. For measurement of transcript levels in
logarithmically growing cells containing plasmids, the cells were grown
under conditions that selected for the plasmid to an
A600 of about 0.15. Then the cells were transferred
to rich media and allowed to double once before RNA was isolated.
Region of the CLN1
promoter encompassing nucleotides 618 to
479 was amplified by
PCR1 from CLN1 in YEp13 (23) with
oligonucleotide primers BD1497 (GCCAGCTCGAGGCCTGCAAGAGACGCGTTCAAGG) and
BD1507 (GCCAGCTAGCGGCCGCGGCACCATTCAGCTCAATTCC). The
480 to
367
region of the CLN1 promoter was amplified from the same plasmid
with BD1495 (GCCAGCCTCGAGCCAGGTCGAGGCTGGGAGGG) and BD1496
(GCCAGCTAGCGGCCGCGGGGACCCACGCTG). Both sets of primers incorporate
XhoI and NotI restriction sites so that they
could be subcloned into pBD1442, which is a derivative of pSH144 (24) with a NotI restriction site inserted downstream of the
XhoI restriction site by introduction of the annealed
linker oligonucleotides BD1392 (TCGAGAAAGTATGCGGCCGCTAA)
and BD1393 (TCGATTAGCGGCCGCATACTTTC), generating plasmids
pBD1512 (CLN1 MCB:lacZ) and pBD1513
(pCLN1 SCB:lacZ), respectively.
The promoter fusion construct pCLN1:HIS3 was
generated by subcloning the BamHI/ClaI fragment
(674 to +177) of CLN1 from pJHB1a (23) into pRS314 to
generate pBD1582. The HIS3 gene (
22 to +813) was amplified
by PCR from pRS303 with oligonucleotides BL29 (TTCCATCGATGCAAGATAAACGAAGGC) and BL30 (GCCCAGCTCGAGGCGCGCCTCGTTC) and
subcloned into ClaI/XhoI digested pBD1582 to
generate pBD1584 (pCLN1:HIS3). The mcb
mutated version of this plasmid (pBD1586) was generated by
site-directed mutagenesis (25) of pBD1584 using the mutagenic
oligonucleotides BL20
(CTGCAAGAGACCTTCAAGGAAGAATTCCATTTTAC) and BL22 (ATCTCGCACCCTTAGTTAGTTTCC). The mcb,
scb mutated version of this construct, pBD1632, was generated by
mutagenesis of pBD1586 with the mutagenic oligonucleotide BL47
(GACTAATCAAGTGACGAAGATCAAATTAA). Changes from the wild type sequence are underlined.
The
MATa1, LacZ, and CLN1
probes have been described previously (26-28). The probe used to
detect the CLN1:HIS3 transcripts was generated by
subcloning the KpnI/HindIII fragment of pBD1584 (334 of CLN1 to +305 of HIS3) into pBS KSII+
(Stratagene) to generate pBD1587, or by subcloning the same fragment
into M13mp18 to produce pBD2022. The analysis of transcript levels by
S1 nuclease protection was performed essentially as
described (29). S1 protection data was quantified by
PhosphorImaging (PhosphorImager 400A, Molecular Dynamics Corp.,
Sunnyvale, CA).
The HO promoter probe
used in these assays has been described previously (9). The
TMP1 promoter probe was amplified by PCR from pEM88 (13) and
spans 216 to
112 of the promoter. The CLN1 SCB-like and
MCB core regions were amplified by PCR from pBD1513 and pBD1512, using
oligonucleotides BD1495/BD1496 and BD1497/BD1507, respectively. Probes
were prepared by PCR, using 0.125 mM dNTP mixture, but with
dCTP at 0.06 mM and 30 µCi of [
-32P]dCTP
(3000 Ci/mmol, DuPont NEN). Probes were isolated by electrophoresis through 7% polyacrylamide gels and electroelution. Protein extracts were prepared from exponentially growing cells as described (9), except
that the breaking buffer was supplemented with 5 mM
MgCl2.
Binding reactions were performed for 10 min at 30 °C, with 10 µg
of protein extract, 1 µg of poly(dI-dC) as nonspecific competitor, and other competitor DNAs in assay buffer (9). Up to 1 ng of probe was
added and after a further 10-min incubation, binding mixtures were
chilled prior to loading on a 4% polyacrylamide gel. Gels were pre-run
for 1 h at 200 V at 4 °C before reactions were loaded and
electrophoresed for 1.5-2.5 h. Gels were dried, and autoradiographed
at 70° C with X-AR film (Kodak). In the antibody addition
experiments, sera was added in the amounts indicated for a second
10-min incubation prior to addition of probe.
Competitor oligonucleotides used in the gel retardation experiments
were derived from the CLN1 promoter (616 to
578) using annealed oligonucleotides of plus strand sequence
CTGCAAGAGACGCGTTCAAGGAAGAATTCGCGATTTTAC (BL16 and BL17). This DNA is
referred to as CLN1 MCB(2). The mutant version of this DNA,
called CLN1 mcb(2), was made with oligonucleotides BL20 and
BL21, which covered the same sequence, but with the CGCG sequences at
604 and
586 mutated to CC. The
REB1 duplex DNA is the
610 to
596 region of the
SIN3 promoter (CGAGAGTCCGGGTAATGAT). The HO SCB
sequence was the annealed oligonucleotides BD260/BD261 encompassing
666 to
636 of the HO promoter and contained a single perfect SCB (CACGAAAA) (9) or the HO scb mutant
(ACTAAAA).
The CLN1 promoter has a cluster of three CGCG
sequences and two closely spaced CTCGAAA sequences located within 400 base pairs of the transcription start site (Fig. 1).
None of these sequences match the full consensus MCB (ACGCGTNA) or SCB
(CACGAAA) sequences, but clusters of MCB core sequences (CGCG) have
been shown to activate transcription in other contexts (13, 24). In
contrast, the mutagenic analysis of SCB elements that has been
carried out (20) predicts that the two SCB-like sequences would be poor
activation sequences. To see if either of these regions are upstream
activation sequences (UAS) for CLN1 transcription, we cloned
these regions into a UAS-deficient promoter, which was fused to
lacZ (pBD1442), and measured lacZ transcript
levels. The level of MATa1 transcript was also
measured and served as an internal control for the amount of total RNA
present in each sample. These measurements (see Table I)
show that the SCB-like region of the promoter does not activate
lacZ expression above background. This suggests that there
are no upstream activation sequences in this region. Since SCB elements
are known to activate transcription in this reporter context (5), it is
likely that the T in the second position prevents recognition of these
sites by the Swi4/Swi6 trans-activation complex. In
contrast, the CLN1 MCB-like region is an efficient activator
of transcription, in that it induces a 70-fold increase in
lacZ transcription above background transcription from the UAS-less vector.
|
To see if the transcription activated by the MCB core region of the
CLN1 promoter is cell cycle regulated in a manner similar to
the full-length promoter, we synchronized cells carrying the CLN1 MCB reporter plasmid in G1 with -factor,
and then took samples of cells at 10-min intervals following release
from the arrest. lacZ transcript levels for CLN1
MCB directed transcription were measured by S1 nuclease
protection, and normalized to MATa1 levels as
before. The quantitation of this experiment (Fig. 1, lower
panel) shows that the CLN1 MCB region is sufficient to
promote cell cycle regulated expression of the lacZ
reporter. Furthermore, comparison of the lacZ transcription
profile with the endogenous CLN1 transcript in the same
samples shows that the timing of expression driven by the
CLN1 MCB-like region is very similar to that of the
endogenous CLN1.
To define the activity of the MCB core sequences in the context of the
normal promoter, plasmids were generated containing CLN1
sequence (674 to +177) fused to HIS3 on centromeric
vectors. Then, the analogous plasmid was made with the three MCB core
sequences CGCG mutated to CTCT (cln1:HIS3 mcb). This mutated
sequence cannot compete for SCB and MCB bound factors in
vitro (data not shown). The presence of these mutations reduced
transcript levels about 4-fold (Fig. 2), indicating that
the MCB core sequences are the elements responsible for most of the
transcriptional activation of CLN1 in logarithmically
growing cells. To see if the SCB-like sequences contribute any
activation function in the absence of the MCB cores, additional
mutations were introduced in the SCB-like sequences (CTCGAAA to
GTCTACA) to produce a cln1:HIS3 mcb scb plasmid. A further
2-fold reduction in transcript levels was observed from the mcb scb
mutant promoter. This modest effect, together with the inability of the
CLN1 SCB-like region to significantly activate
lacZ expression in the reporter construct, suggests that these SCB-like sequences are minor contributors to the transcriptional activation of CLN1.
The CLN1 Promoter Contains Novel Cell Cycle Regulatory Element(s)
We have shown that the MCB-like region of
CLN1 is sufficient to activate cell cycle-regulated
expression, but specific mutation of the MCB core sequences did not
completely eliminate transcription, nor did elimination of both MCB-
and SCB-like sequences. This suggests that there are other active
transcription elements in the CLN1 promoter. To see if these
unidentified elements contribute to the cell cycle regulated
transcription of the CLN1 promoter, we compared the profiles
of transcription from the wild type and the mcb scb mutant versions of
the CLN1:HIS3 plasmid through the cell cycle.
Fig. 3 shows that the CLN1:HIS3
transcript driven by the mcb scb mutant promoter (lower
panel) is still clearly cell cycle regulated, and peaks with
similar timing to the endogenous CLN1 transcript and to the
wild type construct (upper panel). Since cell cycle
regulated transcription persists in the absence of all the recognizable
MCB- and SCB-like sequences, there must be a novel sequence element
residing in the promoter that confers G1/S-specific
transcription.
Swi4 and Swi6 Are the Principle Activators of Transcription from the MCB Region of the CLN1 Promoter
Our in vivo results show that the MCB core sequences serve as a major source of transcriptional activation of CLN1. This result was unexpected because peak levels of CLN1 transcription are dependent upon both Swi4 and Swi6 (18, 19), and these proteins have been characterized as activators of expression through SCB sequences (5). To see if Swi4 and Swi6 activate CLN1 transcription through the MCB core sequences, we transformed swi4 and swi6 deletion strains with the CLN1 MCB:lacZ reporter plasmid, and assayed lacZ transcription in asynchronous cell populations by S1 nuclease protection as before (Table II). Transcription from the CLN1 MCB:lacZ reporter dropped about 5-fold in the absence of either Swi4 or Swi6 activity, suggesting that these factors could be responsible for trans-activation via the MCB core sequences. In contrast, deletion of MBP1, the transcription factor thought to have MCB-binding specificity (16), caused less than a 2-fold reduction in lacZ transcription. These results indicate that Swi4 and Swi6 are the principle activators of transcription from the MCB region of CLN1 in vivo. Mbp1 participates to a lesser extent in this activation.
|
We then asked if Swi4 and Swi6 could bind to the CLN1 MCB
elements directly by performing gel-retardation assays on whole cell
extracts prepared from wild type and mutant yeast. To ensure that the
extracts had the potential to form Swi4-Swi6 and Mbp1-Swi6 complexes,
we first assayed complex formation on fragments from the HO
and TMP1 promoters. HO was the first promoter in
which SCB sequences were identified (30), and binding of Swi4/Swi6 to
these sequences is well established (6, 7, 9). Consistent with previous
studies, a low abundance high molecular weight complex (denoted by the
arrow in Fig. 4A) was formed with
wild type cell extracts on the HO promoter. This complex was
absent from the swi4 or swi6 extracts, but was
unaffected by mutation of MBP1. This confirms that our wild
type extracts are competent to form SCB binding complexes. The lower
complexes observed were comparable to those previously observed and
were not specific for the SCB elements.
The TMP1 promoter contains MCB sequences which are critical
for cell cycle-dependent transcription (12, 13), and
activation of these MCB elements is Swi6- (11, 12) and
Mbp1-dependent (16). Again, consistent with previous
findings (16), a low abundance high apparent molecular weight complex
was formed on the TMP1 MCB elements upon incubation with
wild type extract that was not formed with the swi6 and
mbp1 mutant cell extracts (Fig. 4B). This
confirmed that the wild type extracts were competent to form standard
MCB binding complexes. The MCB specificity of these binding complexes
was verified by competition studies with wild type and mutant MCB
sequences (data not shown). It is worth noting that complex formation
is reduced in the swi4 strain. This could indicate that Swi4
also binds to the TMP1 MCB elements, however, Swi4
antibodies have no apparent effect on these binding complexes (see Fig.
5C).
Knowing that our extracts and conditions permitted formation of both SCB- and MCB binding complexes, we repeated the assay with the SCB-like region of CLN1. With the CLN1 SCB probe, no complexes were formed that were Swi4-, Swi6-, or Mbp1-dependent (Fig. 4C). In contrast, the CLN1 MCB probe formed high molecular weight complexes with proteins from wild type extracts (Fig. 4D) that resembled the HO and TMP1 specific complexes in mobility. These complexes did not form with swi6 and swi4 mutant extracts, but they were as abundant with the mbp1 mutant extract as they were with wild type extracts. This indicates that the CLN1 MCB binding complex is Swi4- and Swi6-dependent, but that Mbp1 is not required.
To address whether Swi4 and Swi6 are present in the CLN1 MCB binding complex, we used affinity purified antibodies raised against recombinant Swi4 and Swi6 proteins (9) in the gel retardation assays. Addition of the Swi4 antibody to the CLN1 MCB binding reaction (Fig. 5A) caused retardation of all of the upper complexes. In fact, the pattern of supershifting on CLN1 MCB was very similar to that produced on incubation of Swi4 antisera with the HO DNA binding reaction using the same wild type extracts (Fig. 5B) (9). This suggests a direct involvement of Swi4 protein in the CLN1 MCB binding complex. The Swi6 antibody also affected the CLN1 MCB complex in that it either shifted the complexes to a higher position in the gel or prevented binding altogether. These results also support a direct involvement of Swi6 in the CLN1 MCB binding complex. In fact, all of the specific CLN1 MCB binding complex is affected by both the Swi4 and Swi6 antibodies. This would not be expected if a significant fraction of the complexes included Mbp1 instead of Swi4 or if they lacked Swi6. Thus, both the in vivo and in vitro data indicate that Swi4-Swi6 complexes are the predominant trans-activators of the CLN1 MCB region.
The Swi4 and Mbp1 proteins are related (16), therefore it was possible that the Swi4 antisera could cross-react with Mbp1. To address this possibility, we incubated Swi4 antibody in a TMP1 binding reaction in which Mbp1-Swi6 is the predominant complex. In this case the Swi4 antibodies had no impact on the mobility or the abundance of the TMP1 binding complex (Fig. 5C). The Swi4 antibody does not cross-react with Mbp1 under the conditions of the gel retardation assay, so it must be detecting only Swi4 in the CLN1 MCB binding complex.
Finally, we wanted to confirm that the MCB core sequences were
responsible for the Swi4-Swi6 complex formation observed with the
CLN1 MCB probe. To do this, we performed competition
experiments using either the CLN1 MCB(2) probe (see Fig. 1)
which includes the first two CGCG core sequences, or the analogous but
mutated competitor (CLN1 mcb(2)), in which both CGCG cores
of the elements are mutated to CTCT. Titration of the wild type
competitor into the binding reaction caused a loss of complex
formation, whereas even high levels of the mutant competitor failed to
compete for binding to CLN1 MCB (Fig. 6).
Interestingly, a single SCB element from the HO promoter
could also compete for complex formation, whereas the homologous
sequence with a mutated scb element failed to compete even at high
concentrations. These experiments show that the CGCG core sequences are
important for complex formation with the CLN1 MCB probe, but
that binding of Swi4-Swi6 to the CLN1 MCB probe can also be
competed by an SCB consensus binding sequence (CACGAAA).
We have shown that the MCB core sequences are the predominant transcription activation elements in the CLN1 promoter during logarithmic growth. The SCB-like sequences in this promoter contribute much less to CLN1 activation. They are unable to activate reporter gene constructs, and they do not serve as Swi4-Swi6-binding sites in vitro. These SCB-like sequences differ from the known SCB sites at the second position, with CTCGAAA instead of the consensus CACGAAA. An A to C transversion at this position causes a 5-fold drop in UAS activity (20), so it is not surprising that a T is also deleterious at the second position.
One of the surprising results of these studies is that transcriptional activation of CLN1 is, in large part, derived from MCB-like sequences which are activated by Swi4-Swi6 complexes. Antibodies to Swi4 react with and supershift all the specific protein complexes formed on the CLN1 MCB core sequences, indicating that Swi4 is present in all of these complexes. This is not due to cross-reactivity of the Swi4 antibodies with Mbp1 (Fig. 6), nor is it due to the absence of Mbp1 from the whole cell extracts because the Mbp1-dependent band shift on TMP1 readily forms under these conditions (Fig. 5B). Mbp1 simply is not a detectable component of the CLN1 MCB complexes that form in vitro. Why, then, is the UAS activity of the CLN1 MCB region reduced at all in an mbp1 mutant strain? Two explanations seem plausible. Mbp1 may participate directly in a subset of the DNA binding complexes instead of Swi4, but our in vitro conditions may not favor its detection. Alternatively, the 2-fold drop in CLN1 MCB transcription that we observe in mbp1 strains may be an indirect effect. Swi4 has three MCB-like elements in its promoter that are critical for full SWI4 transcription (24). Thus, the loss of Mbp1 activity could reduce SWI4 transcription or affect its timing and that could indirectly cause the drop in CLN1 MCB transcription. Whether Mbp1 participates directly or indirectly, it is clear that it plays a minor role compared with Swi4 in these complexes. Swi4-Swi6 complexes form on the MCB core sequences and are the predominant trans-activators of the CLN1 gene.
This is the first clear demonstration that Swi4-Swi6 complexes can activate transcription through MCB-like elements in vivo, but there is other evidence that supports this idea. Cross-competition between SCB and MCB binding complexes has been demonstrated in vitro, but this requires 50-fold molar excess of competitor (12). In addition, in vitro translated fragments of Swi4 can bind to MCB sequences, and fragments of Mbp1 can bind to SCB sequences (16). Interpretation of these experiments is complicated by the fact that only fragments of the Swi4 and Mbp1 proteins were used, and they were present at far higher concentrations than the wild type proteins. Thus, their binding could be an artifact of protein concentration or conformational differences. These fragments are not competent to bind Swi6, so they certainly do not reflect the binding properties of the native complex. Swi4 overproduction in vivo promotes Swi6-independent transcription at SCB (17) and MCB sites (31), but this activity depends upon overproduction of Swi4, and the vast majority of the Swi4 present in such cells is also fragmented (9). With wild type levels of Swi4, both binding and activation of SCB elements absolutely depends upon Swi6 activity (5, 32). So, while these studies are useful for defining what the DNA binding fragment of Swi4 is capable of interacting with, the physiological significance of this interaction could not be assessed until such an activity was demonstrated in vivo with wild type cells. This study shows that the Swi4/Swi6-binding sites in the CLN1 promoter are different and more MCB-like than the ones that have been identified in the CLN2 and HO promoters. This raises the possibility that the complexes themselves may be fundamentally different, and that the Cln1 and Cln2 cyclins may be differentially regulated.
The only other known promoter at which functional overlap between Swi4 and Mbp1 may exist is the CLB5 promoter. CLB5 is also a cyclin, whose transcription peaks at the same time as SCB- and MCB-regulated genes. Interestingly, the CLB5 promoter contains five MCB core sequences (CGCG), but it has no perfect matches to either the MCB or SCB consensus (33). It has been suggested that these sites may be activated by both Swi4- and Mbp1-containing complexes because CLB5 transcription is cell cycle-regulated in both swi4 and mbp1 mutants (16, 33). However, the critical cell cycle regulatory elements in the CLB5 promoter have not been identified, so it is also possible that these CGCG sequences play no role in CLB5 transcription at all. If they do, they may be conventional Mbp1-Swi6-binding sites and the residual cell cycle regulation of CLB5 that is observed in mbp1 mutants may be due to a novel cell cycle regulatory element that has not yet been identified. The existence of novel cell cycle-specific regulatory elements has been demonstrated in all three of the G1-specific promoters that have been carefully analyzed (SWI4 (24), CLN2 (34, 35), and CLN1 (Fig. 3)). However, since activation at CGCG sequences occurs principally via Swi4-Swi6 complexes at the CLN1 promoter, it will be interesting to determine if CLB5 is regulated in a similar fashion, or if the CGCG sites in its promoter represent a fourth type of site at which binding by both Swi4-Swi6 and Mbp1-Swi6 complexes is equally likely.
The SCB binding complex, which includes Swi4 and Swi6, has been referred to as SBF, and the MCB binding complex, including Mbp1 and Swi6, is commonly called MBF (12, 36). Our data indicates the existence of a third type of complex, where Swi4 and Swi6 bind to CGCG sequences. This finding makes it clear that the binding site specificity of Swi4-Swi6 complexes is not restricted to SCB sequences as they are currently perceived, so it may be premature to apply one name to all of them. The lack of a clear consensus sequence for Swi4- and Mbp1-mediated transcription also makes it impossible to deduce by inspection the elements or factors required for the expression of any given promoter.
There are several possible hypotheses that could account for the
altered binding specificity of Swi4 in the CLN1 MCB
complexes. One possibility is that sequences adjacent to the CGCG
sequences directly specify the binding of Swi4- rather than
Mbp1-containing complexes. Alternatively, an accessory factor may bind
to an adjacent site on the DNA that promotes this binding. A third
possibility is that another protein binds directly to or modifies a
subset of the Swi4-Swi6 complexes and changes their DNA binding
specificity. It has been shown that the binding of the S. cerevisiae Mcm1 protein to weak consensus sequences can be
strengthened by the presence of an adjacent binding site for the 1
protein (37). Also, the DNA binding specificity of the Oct1
transcription factor is altered by the presence of the Herpes viral
tegument protein, VP16, from ATGCAAAT to TAATGARAAT (38). We cannot
rule out any of these possibilities at the moment. Determining the
binding site specificity of Swi4-containing complexes by site selection
(39) or by mutational analysis (20) is difficult because a single SCB
element does not form a stable band shift
complex2 or activate sufficient
transcription to be reliably measured (5). The involvement of other
proteins also cannot be addressed easily because full-length Swi4
cannot be efficiently produced in the expression systems that have been
tested or by in vitro translation (8, 10). However,
resolving these issues will be important for understanding the large
number of G1/S-specific events that are regulated by the
Swi4-Swi6 or Mbp1-Swi6 transcription complexes.
The other surprising result of this study is that the CGCG sequences are not the sole source of cell cycle regulated transcription within the CLN1 promoter. Elimination of these sequences as well as the SCB-like sequences did not eliminate the cell cycle regulation of the CLN1 promoter. In fact, the timing of the residual cell cycle regulated transcription from the mutant promoter is similar to the timing of the expression from the wild type promoter. There are no other obvious cell cycle regulatory elements within this promoter, so there must be another unrecognized cell cycle regulatory element within the CLN1 promoter which activates G1/S-specific transcription.
Redundancy is a commonly employed strategy for regulating cell cycle-specific events. Cyclin-CDK complexes are typically regulated at the level of transcription, post-translational phosphorylation, and ubiquination, and by the binding of inhibitors which are in turn highly regulated (40). Here is another example, where even the transcriptional control exerted on a cyclin promoter is redundant. It will be interesting to determine how many different promoter elements are employed to ensure the G1/S specificity of transcription and if their mechanisms of activation are equivalent, or whether they respond to different regulatory signals. The latter would afford more protection from deregulation, and more capacity to respond to different environmental cues.
We thank J. Sidorova for reagents and valuable technical advice. We also thank J. Sidorova, C. McInerny, B. Mai, S. Ewaskow, and C. Gordon for critical reading of the manuscript. We are grateful to C. Wittenberg and B. Miller for DNAs and strains. We also thank T. Knight and P. Goodwin for PhosphorImage analysis and P. Huff for help with the manuscript.