Transcription of the S. pombe fbp1 gene is under the control of PKA and MAPK signaling pathways that are largely responsible for glucose repression and glucose starvation-induced derepression, respectively. In this study, we examined the fbp1 promoter and identified two elements required for transcriptional activation. Both of the signaling pathways regulate protein-DNA interactions at each element, although the mechanisms by which these pathways exert their control differ at the two sites.
UAS1, located 900 bp upstream from the fbp1
transcriptional start site, resembles a CRE that is bound by members of the ATF and CREB family of bZIP transcriptional activators (16
). The discovery of this site as a key element in fbp1
transcription is not surprising, since the atf1-pcr1 bZIP transcriptional activator is required for fbp1
). A base change in the core of this sequence greatly reduces fbp1
transcriptional activation in vivo (170mut2 promoter construct; Fig. B) and eliminates the in vitro binding by a starvation-induced protein complex (Fig. A and B). This base change reduces fbp1
derepression to a greater extent than do deletions of this region (170 and SP deletion constructs; Fig. A), suggesting that additional sequences in this region negatively regulate fbp1
transcription. As expected, the atf1-pcr1 heterodimer is responsible for the starvation-induced UAS1 binding activity (Fig. C and D).
Further characterization of the UAS1 binding activity shows that both the MAPK and PKA pathways regulate this protein-DNA interaction. Deletion of the wis1
MAPKK gene results in a loss of UAS1 binding activity similar to that to that conferred by an atf1
deletion (Fig. ). Furthermore, loss of PKA activity allows for binding in the absence of a starvation signal, while PKA activation by a mutation affecting the cgs1 regulatory subunit of PKA has little effect. These data suggest that PKA activity inhibits atf1-pcr1 binding to UAS1, while the spc1/sty1 MAPK activity overcomes this inhibitory effect. Therefore, activation of the MAPK pathway is not required for atf1-pcr1 binding to UAS1 in a pka1
mutant, while the elevated PKA activity in a cgs1
mutant does not prevent starvation-induced MAPK stimulation of atf1-pcr1 binding to UAS1. Yet, we previously showed that cgs1
mutants are grossly defective in derepression of fbp1
), indicating that activation of the MAPK pathway is not sufficient to overcome the elevated PKA activity. This apparent contradiction may be explained by our discovery here that both the MAPK and PKA pathways exert multiple regulatory effects on fbp1
transcription. These varied interactions might also explain why a mutation that eliminates the only consensus PKA phosphorylation site within atf1
) was seen to have only a moderate effect on mating efficiency (22
While the S. pombe
PKA and MAPK pathways have been previously shown to antagonistically regulate a wide range of biological processes (4
), this is the first demonstration of a direct interaction between these pathways and control of atf1 activity at the level of DNA binding. Previous studies have suggested that the atf1-pcr1 heterodimer is constitutively bound to DNA (6
) and that mutations in the PKA pathway do not affect transcriptional activation by this complex (55
). One possible explanation for the discrepancy between those studies and this one comes from the fact that we are examining binding to an endogenous site rather than a consensus CRE. Signaling through the MAPK and PKA pathways might only confer a detectable change in atf1-pcr1 binding affinity to suboptimal binding sites. The atf1-pcr1 heterodimer may bind a consensus CRE site with such a high affinity that this interaction is insensitive to PKA action. Loss of regulation due to the optimization of a protein binding site has been observed for the E. coli
cAMP receptor protein (CRP), which activates transcription of a large number of operons subject to glucose repression. The CRP binds an optimized palindromic sequence with a 450-fold greater affinity than it binds the endogenous sequence from the lac
operon promoter (10
). In vivo, a galP1
promoter bearing an optimized CRP binding site is no longer subject to glucose repression (13
The one study that has shown a positive role for the spc1/sty1 MAPK on DNA binding by atf1-pcr1 (also known as mts1-mts2) (26
) examined a site involved in meiotic hot spot recombination. While a mutation in the spc1/sty1
gene reduced binding to a similar DNA element at the ade6-M26
recombination hot spot, this element is not involved in transcriptional regulation (27
). There is no evidence for a role of the PKA pathway in this interaction.
UAS2, located 250 bp upstream from the transcriptional start site, contains overlapping consensus sequences for both transcriptional activators and repressors. A base change in the STRE-like sequence CCCCT eliminates the in vivo derepression of fbp1 transcription from this element (compare the Fig. A Pac construct with the Fig. B 170 and 170mut2 constructs) and alters in vitro binding to a probe carrying this sequence by S. pombe extracts (Fig. A). While the CCCCT element is clearly required for transcriptional activation, adjacent sequences may also positively or negatively regulate fbp1 transcription.
The pattern of protein-DNA complexes present in EMSAs using a UAS2-containing probe is surprisingly complex. At least four distinct complexes appear to be affected by changes in the sequence of the probe (Fig. A), the loss of the scr1 glucose repressor (Fig. B), mutations affecting the PKA pathway (Fig. A), or mutations affecting the MAPK pathway or atf1 (Fig. B). However, unlike the UAS1 binding activity, atf1 does not appear to be a component of the UAS2 binding activity (Fig. C). The most likely candidates for the UAS2-specific activators and repressors are members of a family of proteins containing two zinc fingers that resemble the DNA binding domains of the S. cerevisiae
Msn2 and Msn4 transcriptional activators, as well as the Mig1, Mig2, and Nrg1 repressors (35
). These transcriptional regulators are all associated with either stress-induced transcription or glucose repression and bind to sequences similar to that of UAS2. Analysis of the S. pombe
genome reveals at least seven gene products with similar zinc fingers, including the scr1 repressor (56
), the rsv1 activator (17
), and several uncharacterized proteins. Figure displays an alignment of the zinc fingers from five of these proteins, along with those of the S. cerevisiae
transcriptional regulators mentioned above. To further complicate matters, two pairs of these S. pombe
proteins, one that includes scr1, differ in length by only three residues. Further work will be needed to determine what if any redundant functions are carried out by these proteins.
FIG. 6 Alignment of zinc fingers from related S. cerevisiae and S. pombe proteins. Pairs of zinc fingers from S. cerevisiae proteins Msn2 (6323680), Mig1 (6321403), Mig2 (6321229), and Nrg1 (6320248) and S. pombe proteins C3H1.11 (Q10076), scr1 (CAB10978), rsv1 (more ...)
Since the EMSA studies conducted here utilize whole-cell extracts, it is quite possible that we are detecting binding activities that are cytoplasmic at the time of extract preparation. This complicates our ability to assign roles for any of the UAS2 complexes in either transcriptional activation or repression. It has been shown that phosphorylation of the S. cerevisiae
Mig1 repressor and Msn2 activator determines whether these proteins are nuclear or cytoplasmic (8
). However, even though similar regulation is likely to occur in S. pombe
, this does not detract from our evidence that UAS2 is the target of both transcriptional activators and repressors. Introduction of a base change in the set of four C's results in the loss of transcriptional activation from this site (Fig. B) and dramatically reduces complex A and B formation (Fig. A). Conversely, loss of the Mig1-like scr1 repressor also alters the protein-DNA interactions at UAS2 (Fig. B) and causes an increase in expression from both the full-length promoter and a UAS2-driven promoter (Table ). The Mig1 protein, responsible for glucose repression of many S. cerevisiae
genes, binds to a sequence resembling the UAS2 element and recruits the corepressor Ssn6-Tup1 (58
). The scr1 repressor may function in a similar manner, because we and others have observed that S. pombe
possesses two TUP1
-like genes that encode redundant repressors of fbp1
; R. Janoo and C. S. Hoffman, unpublished observations). However, since a promoter lacking the UAS2 sequence is also partially derepressed by the loss of scr1 (Table ), UAS2 cannot represent a unique site for scr1 action within the fbp1
Our data suggest that the PKA and MAPK pathways each employ at least two independent mechanisms to antagonistically regulate fbp1 transcription. PKA negatively regulates transcriptional activation from both UAS1 and UAS2. At UAS1, PKA inhibits atf1-pcr1 binding; however, PKA must have additional roles, since we have observed that an atf1Δ strain is partially derepressed for fbp1-lacZ expression by the subsequent deletion of pka1 (data not shown). PKA could stimulate the activity of a repressor such as scr1 and/or inhibit the activity of a UAS2-specific activator. The MAPK pathway stimulates atf1-pcr1 binding at UAS1 and promotes the expression or function of other activators at UAS2. These two distinct roles for atf1 are confirmed by the in vivo activation data. UAS1-driven expression is absolutely dependent upon atf1 (Table ), as expected, since the UAS1 site is an atf1-pcr1 binding site (Fig. ). On the other hand, while UAS2-driven expression is largely atf1 dependent, some regulation is still evident in an atf1Δ strain, consistent with an indirect role for atf1.
While it seems unnecessarily complicated that both the PKA and MAPK pathways employ multiple mechanisms to regulate fbp1 transcription, such mechanistic diversity would allow cells to regulate a collection of genes by the same two pathways, but with different degrees of sensitivity. The fbp1-lacZ constructs (Fig. ) are all subject to glucose regulation, yet display regulation over a 5-fold to 200-fold range, with a 3-fold range in absolute repressed levels (excluding the virtually inactive 171 construct). As with these different constructs, the S. pombe genes controlled by the PKA and MAPK pathways in response to nutrient and stress conditions may utilize only subsets of the mechanisms observed here. This is likely to be a general theme of eukaryotic transcriptional regulation, allowing a finite number of signaling pathways to exert qualitatively different outcomes in gene expression over a large number of genes that are subject to the same environmental signals.