MAP kinase and cAMP filamentation signaling pathways converge on the unusually large promoter of the yeast FLO11 gene

Steffen Rupp, Eric Summers, Hsiu‐Jung Lo, Hiten Madhani, Gerald Fink

Author Affiliations

  1. Steffen Rupp1,
  2. Eric Summers1,
  3. Hsiu‐Jung Lo1,
  4. Hiten Madhani1 and
  5. Gerald Fink*,1
  1. 1 Whitehead Institute for Biomedical Research, Nine Cambridge Center, Cambridge, MA, 02142, USA
  1. *Corresponding author. E-mail: fink{at}
View Full Text


In Saccharomyces cerevisiae, two major signal transduction pathways, the Kss1 MAPK pathway and the cAMP‐regulated pathway, are critical for the differentiation of round yeast form cells to multicellular, invasive pseudohyphae. Here we report that these parallel pathways converge on the promoter of a gene, FLO11, which encodes a cell surface protein required for pseudohyphal formation. The FLO11 promoter is unusually large, containing at least four upstream activation sequences (UASs) and nine repression elements which together span at least 2.8 kb. Several lines of evidence indicate that the MAPK and cAMP signals are received by distinct transcription factors and promoter elements. First, regulation via the MAPK pathway requires the transcription factors Ste12p/Tec1p, whereas cAMP‐mediated activation requires a distinct factor, Flo8p. Secondly, mutations in either pathway block FLO11 transcription. Overexpression of STE12 can suppress the loss of FLO8, and overexpression of FLO8 can suppress the loss of STE12. Finally, multiple distinct promoter regions of the FLO11 promoter are required for its activation by either Flo8p or Ste12p/Tec1p. Thus, like the promoters of the key developmental genes, HO and IME1, the FLO11 promoter is large and complex, endowing it with the ability to integrate multiple inputs.


On solid media containing high glucose and low nitrogen, diploid cells of Saccharomyces cerevisiae form pseudohyphae, which are comprised of chains of elongated cells that form invasive filaments (Gimeno et al., 1992). On rich medium, haploid but not diploid cells manifest invasive growth (Roberts and Fink, 1994). The nutritional signals that result in both pseudohyphal growth and haploid invasive growth involve several pathways. One of these is a MAP kinase pathway required for both mating and filamentation (Liu et al., 1993). This cascade involves three protein kinases, Ste20p, Ste11p and Ste7p, that act in sequence. Their activation results in the conversion of the Kss1p MAPK from an inhibitor of the transcription factor Ste12p/Tec1p into its activator (Liu et al., 1993; Madhani et al., 1997). One of the genes regulated by the Kss1 MAPK cascade is FLO11, a cell wall protein required for invasive and filamentous growth (Lo and Dranginis, 1998).

A second signal transduction pathway involved in filamentous growth is the cAMP/PKA pathway. Increasing the level of cAMP by mutation of the high‐affinity phosphodiesterase Pde2p, by activating mutations of Ras2p GTPase or by the exogenous application of cAMP, enhances filamentous growth (Gimeno et al., 1992; Ward et al., 1995; Lorenz and Heitman, 1997). Both Ras2p, a known regulator of cAMP levels, and the G protein α subunit homolog, Gpa2p, appear to act upstream of adenylate cyclase. Deletion of either RAS2 or GPA2 results in reduced filamentation presumably because of reduced cAMP levels in the cell (Kubler et al., 1997; Lorenz and Heitman, 1997). In several filamentous fungi, the cAMP/PKA pathway plays a role in the regulation of filamentation that has been connected to their virulence (Gold et al., 1994; Xu and Hamer, 1996; Alspaugh et al., 1997; Durrenberger et al., 1998; Madhani and Fink, 1998).

Two genes, the transcription factor FLO8 and the cell surface protein FLO11, have also been reported to be required for invasive and filamentous growth (Lambrechts et al., 1996; Liu et al., 1996; Lo and Dranginis, 1996, 1998). The FLO genes encode proteins required for cell–cell adhesion (Teunissen and Steensma, 1995). FLO8 has been localized to the nucleus (Liu et al., 1996) and was reported to be a putative transcriptional activator of FLO1, a dominant flocculation gene that encodes a cell wall‐associated protein (Kobayashi et al., 1996). FLO11 is localized to the cell surface and appears to be required for cell–cell adhesion and the integrity of pseudohyphal filaments (Lo and Dranginis, 1996).

In this report, we show that FLO11 is a target for both the MAP kinase and cAMP pathways. Our results suggest that FLO8 is required for activating FLO11 transcription via the cAMP/PKA pathway. Ste12p, a second transcription factor important for FLO11 regulation (Lo and Dranginis, 1998), transmits the Kss1 MAPK signal to sites within the promoter of FLO11 that are distinct from the FLO8 target sites. In addition to these, there is a plethora of positive and negative cis‐acting sites spread over at least 2.8 kb that define the FLO11 promoter as one of the largest in the yeast genome, integrating MAPK, cAMP/PKA, mating type and nutritional signals.


A strain dependent on cAMP for growth

Since both the Kss1 MAPK pathway and the cAMP/PKA pathway are activated for filamentation by the same activator, Ras2p (Toda et al., 1987; Mosch et al., 1996), analysis of the distinct role of cAMP on filamentous growth requires the ability to activate the PKA branch independently of RAS. To achieve this goal, we constructed a strain (ras1 ras2 pde2) lacking both RAS genes (RAS1 and RAS2) and PDE2, the gene for the high‐affinity cAMP phosphodiesterase. Since Ras1p and Ras2p are required for the activation of adenylate cyclase, Cyr1p (Toda et al., 1985), and PDE2 encodes the phosphodiesterase required for cAMP hydrolysis, a ras1 ras2 pde2 strain is impaired in the synthesis and breakdown of cAMP. Such a strain should be dependent upon exogenous cAMP for induction of the A kinase and, as no Ras‐induced signal can be transmitted to the MAPK cascade, the effects of cAMP on filamentation should be independent of the RAS/MAPK signal.

The ras1 ras2 pde2 strain (SR957) requires cAMP for growth on YPD (yeast extract, peptone, dextrose), but grows without cAMP on SC (synthetic complete), SLAD (synthetic low ammonia dextrose) and YNB (yeast nitrogen base) media, where it displays hyperaccumulation of glycogen, indicative of low cAMP levels. We presume that the ability of this triple mutant to grow without added cAMP, as has been observed by others (Nikawa et al., 1987), results from basal cyclase activity that is sufficient to provide internal cAMP. The fact that the growth of our ras1 ras2 pde2 strain depends upon a functional cyclase (CYR1) gene supports this explanation (data not shown).

A diploid ras1 ras2 pde2 (SR959) strain grows on SLAD medium without cAMP but does not form pseudohyphae. However, on SLAD medium containing cAMP, the strain is extremely filamentous (Figure 1A). Moreover, the addition of cAMP leads not only to induction of filamentation, but also to invasion of the substrate (Figure 1B). In the presence of cAMP, the ras1 ras2 pde2 strain (SR959) is invasive on all media tested (YPD, SC and SLAD; Figures 1A, B and 4C). Since invasive growth is usually observed with haploid strains on YPD, both the cell type signal and the nutritional signal can be bypassed by high cAMP levels in the cell.

Figure 1.

cAMP is required for induction of invasive and filamentous growth. (A) Pseudohyphal development of strains on cAMP plates. Shown is the growth of strain SR607 (wild‐type) and strain SR959 (ras1/ras1, ras2/ras2, pde2/pde2) on SLAD medium or SLAD medium containing 2 mM cAMP. Plates were incubated for 2 days at 30°C. (B) The upper half of the panel shows strain SR959 (ras1/ras1, ras2/ras2, pde2/pde2) streaked to a SC plate (left) and a SC plate containing 2 mM cAMP (right). Following incubation, the upper plates were washed under a stream of water and re‐photographed to yield the lower plates. (C) Shown are cells of strain SR959 (ras1/ras1, ras2/ras2, pde2/pde2) grown in SC (left) or SC with 2 mM cAMP (right) to OD = 1. Cells grown with 2 mM cAMP show strong flocculation resulting in chains of cells, whereas cells grown without cAMP are mostly either single cells or cells with a single bud. (D) The Northern blots prepared from total RNA of cells shown in (C) were probed with FLO11 probe or FLO1 probe, stripped and subsequently re‐hybridized with ACT1 probe as normalization control.

Figure 2.

Sequence elements involved in regulation of FLO11. (A) The β‐galactosidase activities of 14 individual 200 bp deletions of the FLO11 promoter region were compared with the full‐length 3 kbp promoter in exponentially growing cells (rich medium), post‐diauxic cells (pd) (glucose depletion) and nitrogen starvation conditions (N‐starvation) (SLAD, see Table II). A ratio of β‐galactosidase activity (deletion/3 kbp promoter) of 3 or higher led to that deletion's association with a URS (star, hexagon, square). At least 30% lower expression than the wild‐type reporter under all conditions is represented as a UAS (circle). N refers to cell type specific regulation. (B) The β‐galactosidase activity of 15 isolated 400 bp elements of the FLO11 promoter region, that activate a CYC1::lacZ reporter, was determined in wild‐type cells or in cells deleted for the transcription factors FLO8, STE12, TEC1 or STE12 and TEC1 in exponentially growing cells or post‐diauxic cells. The activity of these constructs was also tested in the presence or absence of cAMP in strain SR957 (see Table III). The symbols (× = rich medium, + = post‐diauxic cells) are placed on a line in a position that indicates which of the 400 bp fragments stimulated β‐galactosidase activity. Each line represents the FLO11 promoter in a different genetic background. The first row [denoted WT (UAS)] denotes sequence elements showing a >2‐fold elevation of the reporter over a plasmid without an insert as measured in wild‐type cells. The next three lines, ste12, tec1 and flo8, represent sequence elements showing a >3‐fold reduction of the β‐galactosidase activity in the mutant (e.g. ste12) as compared with the activity of the same element in wild‐type cells. Elements that give enhanced expression in strain SR957 grown with cAMP as compared with the same strain grown without cAMP are indicated on the line in the last row (cAMP induced). Note that URS elements could mask the contribution of adjacent UAS elements (e.g. FLO11‐11/10 and FLO11‐10/9 provide evidence for a UAS in segment 10; however, the deletion analysis revealed no UAS activity for flo11‐10).

Figure 3.

FLO11 transcription is induced by high intracellular or extracellular cAMP levels and requires Flo8p but not Ste12p for cAMP‐mediated induction. The Northern blots were probed with ACT1 as a normalization probe and a FLO11 probe. (A) Total RNA was prepared from strains SR599 (ira1), 10560‐2B (wild‐type), SR1133 (ira1 ste12), SR1132 (ira1 flo8), L5795 (ste12), L5816 (flo8), L6149 (tec1) and SR1101 (flo11), grown in SC to OD = 1.0–1.2. (B) Total RNA was prepared from strains SR607 (wild‐type), SR959 (ras1/ras1, ras2/ras2, pde2/pde2), SR1081 (flo8/flo8, ras1/ras1, ras2/ras2, pde2/pde2), SR1088 (ste12/ste12, ras1/ras1, ras2/ras2, pde2/pde2) and SR1101 (flo11), grown in SC with or without cAMP to OD = 1.0–1.2.

To determine whether cAMP induces filamentation by activating the MAPK pathway, we measured the expression of the Kss1 MAPK pathway‐specific reporter FG::Ty1‐lacZ (Mosch et al., 1996) in ras1 ras2 pde2 strains grown on SLAD plates containing concentrations of cAMP that induce filamentation. The level of expression of the FG::Ty1‐lacZ reporter in the ras1 ras2 pde2 strain (SR959) is not altered by these cAMP levels (data not shown). Moreover, cAMP induces filamentation in a ste12 ras1 ras2 pde2 deletion strain (SR1088) (Figure 4A). These results agree with those of Lorenz and Heitman (1997) and argue that the cAMP/PKA pathway acts in parallel with the MAPK pathway.

FLO11 is induced by cAMP

The essential role of FLO11 in filamentation (Lo and Dranginis, 1998) suggested that it might be a downstream target of cAMP. It was possible to demonstrate the induction of FLO11 mRNA by cAMP in the ras1 ras2 pde2 strain (Figure 1D), but not in a wild‐type background (Figure 3B). The FLO11 transcript is undetectable when the triple mutant strain is grown without cAMP and is strongly induced in the presence of 2 mM cAMP. FLO1, which encodes another cell surface protein required for flocculation that is 26% identical to Flo11p, is only moderately induced (1.4‐fold) under the conditions used (Figure 1D). This result shows that the strong induction of FLO11 is not a general feature of all flocculation genes. The correspondence between cAMP induction of FLO11 and the morphological changes observed when cells are grown in the presence of cAMP is supported by the phenotype of the FLO11 deletion: in the flo11 ras1 ras2 pde2 strain (SR1121), cAMP fails to induce either invasion or filamentation (Figure 4). These data suggest that FLO11 is a key target of a cAMP‐dependent signaling pathway, one that is required for the induction of invasive and filamentous growth.

Figure 4.

FLO8 is required for cAMP‐mediated filamentous and invasive growth. (A) Filamentous growth of strains on SLAD plates containing 2 mM cAMP. All strains are diploid. The plate shows strains SR607 (wild‐type) SR959 (ras1/ras1, ras2/ras2, pde2/pde2), SR1081 (flo8/flo8, ras1/ras1, ras2/ras2, pde2/pde2), SR1088 (ste12/ste12, ras1/ras1, ras2/ras2, pde2/pde2) and SR1121 (flo11/flo11, ras1/ras1, ras2/ras2, pde2/pde2) that were streaked on a SLAD plate containing 2 mM cAMP. Following 5 days of incubation at 30°C, photographs of the colonies were taken. (B and C) Invasive growth of strains patched on YPD + 2 mM cAMP. Strains were grown on YPD + 2 mM cAMP for 2 days at 30°C. Pictures were taken before (B) and after wash (C). All strains are diploid. The upper row of patches shows the same strains as in (A). As a control, the invasive growth phenotype of SR607 (MATa/α, wild‐type), SR1097 (flo8/flo8), SR1022 (ste12/ste12) and SR1098 (flo11/flo11) is shown in the lower row of patches.

The enhanced FLO11 transcription in the presence of cAMP is correlated with a change in cellular morphology. Cells grown in liquid SC with 2 mM cAMP show pseudohyphal‐like chains of cells, whereas the majority of cells grown in liquid SC without cAMP are either single cells or cells with a single bud (Figure 1C). The effect of cAMP on cell–cell attachment is much more pronounced than the effect of the cyclic nucleotide on cell elongation.

Analysis of the FLO11 promoter

Analysis of the intergenic regions of yeast (Table I) suggests that FLO11 has the longest 5′ non‐coding region in the yeast genome, a stretch of ∼3.6 kbp upstream of the ATG that initiates the putative coding region. To determine how much of this 5′ region of FLO11 is required for the regulated expression of FLO11, we examined the behavior of plasmid‐based FLO11::lacZ reporter constructs containing deletions in this non‐coding region. Fourteen serial 200 bp deletions were constructed that span the region between the 2800 bp upstream of the FLO11 initiation codon. Expression of the deletions was assayed in haploid and diploid strains. Since FLO11 expression varies with the growth phase of the cells (Lo and Dranginis, 1998), we analyzed exponentially growing cells on SC, cells grown on SC until the glucose has been depleted (post‐diauxic) and cells on SLAD, a medium that is high in glucose and low in nitrogen.

View this table:
Table 1. The 10 largest putative intergenic regions in the S.cerevisiae genome

Enzymatic assays of the individual flo11lacZ promoter deletions reveal an unusually long promoter with many sites. In subsequent sections, a site is tentatively assigned as a URS (upstream repression site) if its deletion leads to at least 3‐fold enhanced expression of FLO11, and as a UAS (upstream activation site) if its deletion leads to at least 30% reduced expression.

Deletion analysis of promoter elements

Analysis of cis‐acting elements by enzymatic assays of the individual flo11lacZ promoter deletions reveals that the intact FLO11 promoter is highly repressed. The deletions define at least nine URS elements (Table II; Figure 2A), whose activity depends on the state of growth, nutrient conditions and cell type. One of the URS elements is defined by flo11‐14 (−2600 to −2800 bp), showing that cis‐acting elements are present at least 2.8 kb from the putative FLO11 coding region (Table II; Figure 2A). In general, haploid strains show stronger repression than diploid strains.

View this table:
Table 2. Relative β‐galactosidase activity of 14 FLO11::LacZ reporter constructs, each containing a different 200 bp deletion in the 3 kb segment in the 5′ region upstream of the putative ATG that initiates translation

A clear way to visualize the activity of these sites of repression is by comparison of the lacZ activity for each of the conditions with that of a haploid grown on SLAD (Table II). This comparison reveals URS elements within flo11‐4, ‐5, ‐7, ‐8, ‐12, ‐13 and flo11‐14. A subset of these elements is key to repression on all media, but the strongest effect is in haploids grown on SLAD. Haploid‐specific effects are found for deletions flo11‐4, flo11‐10, flo11‐11, flo11‐12 and flo11‐13. Clearly, there are sites in flo11‐4, flo11‐12 and flo11‐13 that function in the haploid‐specific nitrogen regulation of FLO11.

There are other differences between haploid and diploid strains, the most notable of which are: (i) in diploids on SC (exp), flo11‐4 has a 2‐fold reduced expression level, whereas in haploids it has a 33‐fold elevated expression; and (ii) in diploids after the diauxic shift, flo11‐11 has a 3‐fold reduced level whereas in haploids it has 12‐fold elevated expression. In diploid cells, flo11‐4, flo11‐10 and flo11‐11 act as UAS elements. flo11‐5 is a strongly nitrogen‐regulated site in both haploids and diploids.

Deletion analysis also revealed sequence elements required for expression of FLO11. Under all conditions tested, flo11‐6 had a dramatic reduction in expression, suggesting that the sequence deleted in flo11‐6 (bp −1000 to −1200) contains a strong UAS. Furthermore, flo11‐1, flo11‐2 and flo11‐3 show consistently lower activity, as compared with the wild‐type FLO11–lacZ reporter construct, identifying these as UAS elements in the FLO11 promoter. Recent studies of the TATA element in the STA2 gene of S.cerevisiae var. diastaticus, a close homolog of FLO11, reported transcriptional initiation of STA2 at −100 of the putative ATG (Vivier and Pretorius, 1998). The authors concluded by analogy that the same transcriptional initiation site is used in FLO11. The reduced expression of flo11‐1 is likely to be a consequence of the deletion of the transcription initiation site in this construct.

Several flo11 promoter mutations were inserted in the chromosome at the FLO11 locus by integrative transformation. One, flo11‐16, deleted virtually the entire putative promoter region (from −150 to −2947), leaving the proposed TATA site intact (Vivier and Pretorius, 1998). Strains carrying either flo11‐16 or flo11‐6 are completely defective in haploid invasive growth and, as diploids (e.g. flo11‐6/flo11‐6), show severely reduced filamentation. The haploid invasion defect of these strains is as severe as that of a deletion of the FLO11 coding region. FLO11 mRNA levels in flo11‐16 and flo11‐6 are only 10% of the wild‐type level. The results with flo11‐6 support our conclusions based on the data from the lacZ plasmid constructs, which identified the segment deleted in the flo11‐6 construct as a critical UAS for FLO11 expression.

Strains containing integrated flo11‐3, flo11‐4 and flo11‐14 also have phenotypes consistent with the data obtained from the corresponding flo11–lacZ reporter constructs: flo11‐3 has a strongly reduced haploid invasion defect, whereas flo11‐4 has a hyperinvasive phenotype. When the level of FLO11 mRNA in these strains was compared with that in wild‐type, we found that flo11‐3 was reduced by 80% and flo11‐4 was elevated 3‐fold. Haploid flo11‐14 strains in exponential growth phase showed a 1.5‐fold increase in FLO11 mRNA, supporting the regulatory role of this distant promoter element.

Analysis of individual promoter fragments

In a second approach to identify UAS elements of the FLO11 promoter, we designed 14 individual 400 bp sequence elements, overlapping by 200 bp, to test activation of a lacZ reporter (Guarente and Ptashne, 1981). This reporter construct series (Table III; Figure 2B) identified UAS elements in the segments FLO11‐2/1 and FLO11‐3/2 that overlap between bp −200 and −400, FLO11‐6/5 and FLO11‐7/6 that overlap between bp −1200 and −1000, and FLO11‐10/9 and FLO11‐11/10 that overlap between bp −2000 and −1800. These sequence elements show a >2‐fold increase in β‐galactosidase activity as compared with the reporter plasmid without any insert. The activity of FLO11‐2/1, FLO11‐3/2 and FLO11‐11/10 is induced in post‐diauxic cells, leading to an induction of up to 200‐fold for FLO11‐11/10. This result suggests that elements FLO11‐2, FLO11‐3 and FLO11‐11 are required for enhanced FLO11 expression in later stages of growth. Taken together with the deletion analysis, these data suggest that there are at least four UAS elements in the FLO11 promoter. The FLO11 promoter has extensive homologies to the promoter regions of the STA1, STA2 and STA3 genes from S.cerevisiae var. diastaticus, where a complex regulation pattern, similar to that of FLO11, has been observed (Lambrechts et al., 1994).

View this table:
Table 3. β‐Galactosidase assays of individual 400 bp FLO11 promoter elements cloned into the UAS of a CYC1::lacZ reporter construct

Trans‐acting elements

Northern analysis revealed that yeast strains deleted for FLO8 (L5816), STE12 (L5795) and TEC1 (L6149), transcription factors known to be required for pseudohyphal and invasive growth, show strongly reduced levels of FLO11 transcription (Figure 3A). Previous work (Lo and Dranginis, 1998) suggested that ste12 mutants reduce FLO11 transcription. The effects of Ste12p on FLO11 are a direct consequence of activation by the Kss1 MAPK pathway because strains containing STE11‐4, a dominant‐active allele of the MAPKKK, have a 30‐fold greater amount of FLO11 mRNA than wild‐type strains (STE11). Moreover, deletion of STE12 in a STE11‐4‐containing strain (L5577) leads to a reduction of FLO11 transcript levels to only 30% of the wild‐type level.

To identify the regions of the FLO11 promoter that are targeted by these trans‐acting elements, we transformed the 14 individual 400 bp sequence elements into strains deleted for FLO8, STE12, TEC1 and for both STE12 and TEC1 (Table III; Figure 2B). Deletion of FLO8 leads to a severe reduction in the expression of FLO11‐6/5 and FLO11‐7/6 in both exponential (8‐ and 4‐fold reduction, respectively) and post‐diauxic growth phase (5‐fold for both elements). These elements are also induced by high cAMP levels. FLO8 function is also required for maximum expression of FLO11‐3/2 and FLO11‐8/7 in the post‐diauxic growth phase, but not in exponential growth, a result suggesting that Flo8p may function differently depending upon nutrient conditions.

Deletion of STE12 has a stronger effect in exponentially growing cells than in post‐diauxic cells. In exponentially growing cells lacking Ste12p, the FLO11‐6/5, FLO11‐9/8, FLO11‐10/9 and FLO11‐11/10 segments show at least a 3‐fold reduction in expression. In post‐diauxic cells, only FLO11‐10/9 shows reduced expression in a STE12 deletion strain.

Deletion of TEC1 has significant effects on expression of the FLO11 insertion reporter series, but the reduction can be observed only in exponentially growing cells and is less severe than that observed for strains deleted for STE12 or FLO8. This observation agrees with the Northern analysis (Figure 3A) of the intact FLO11 promoter, where FLO11 mRNA levels show less of a reduction in a strain deleted for TEC1 (L6149) than in strains deleted for STE12 (L5795) or FLO8 (L5816). FLO11‐6/5, FLO11‐10/9 and FLO11‐12/11 are dependent upon TEC1 in exponentially growing cells. Deletion of TEC1 in a Δste12 strain shows a modest reduction in the expression of FLO11‐10/9 as compared with the single mutant strains. However, since deletion of STE12 leads to a >10‐fold stronger effect than deletion of TEC1, this result may indicate a Tec1p‐independent role for Ste12p at this site.

Our results suggest that Flo8p and Ste12p/Tec1p act via multiple sites in the FLO11 promoter that are largely non‐overlapping. This becomes most evident in post‐diauxic cells. The strongest effect of Flo8p and Ste12p in post‐diauxic cells is observed with two distinct sequence elements of the FLO11 promoter. Flo8p acts on the sequence elements FLO11‐3/2, FLO11‐5/6, FLO11‐6/7 and FLO11‐8/9, whereas Ste12p acts on FLO11‐10/9. The effect of Ste12p on element FLO11‐5/6 can be observed only in exponentially growing cells. FLO11‐12/11 and FLO11‐10/9 are targeted both by Tec1p and Ste12p. The existence of spatially distinct sites for different transcription factors provides strong evidence in support of the dual control over FLO11 transcription.

To locate the sites within the promoter where the cAMP signal activates FLO11‐transcription, we transformed the 400 bp reporter series into SR957, a strain where internal cAMP levels can be manipulated by adding cAMP to the media. Addition of 2 mM cAMP to SR957 leads to a 3‐fold induction of FLO11‐6/5 and to a 2‐fold induction of FLO11‐7/6, FLO11‐8/7 and FLO11‐10/9 as compared with SR957 grown without cAMP. These results suggest that trans‐acting elements up‐regulate FLO11 expression through a cAMP‐mediated signal via more than one cis‐acting element. The segment most strongly induced by cAMP is defined by the overlapping elements FLO11‐6/5 and FLO11‐7/6 (within bp −1000 to −1200), the same element targeted by Flo8p. As shown earlier, this element is required for induction of invasive and filamentous growth.

FLO8 is required for cAMP‐mediated FLO11 transcription

FLO8, a gene essential for filamentous growth (Liu et al., 1996), is also required for the expression of FLO11. FLO11 transcripts are not detectable in a strain (L5816, flo8‐2) that contains a deletion of the FLO8 gene (Figure 3A). Furthermore, FLO11 induction by cAMP is blocked in a ras1 ras2 pde2 strain (SR1081) carrying a deletion of FLO8 (Figure 3B).

The ste12 ras1 ras2 pde2 strain (SR1088), like the flo8 ras1 ras2 pde2 (SR1081) strain, has dramatically reduced expression of FLO11. However, FLO11 transcription can be induced by cAMP in the ste12 ras1 ras2 pde2 strain (Figure 3B). The induction of FLO11 by cAMP in the ste12 mutant and not in the flo8 mutant is consistent with the phenotypes of the corresponding strains: a flo8 ras1 ras2 pde2 strain is unable to form filaments on SLAD plates or to invade the agar on YPD plates even in the presence of cAMP, whereas ste12 ras1 ras2 pde2 is both invasive and able to form filaments on YPD or SLAD plates containing 2 mM cAMP (Figure 4). Thus, high cAMP levels can bypass the requirement for the MAPK cascade transcription factor Ste12p, but not the requirement for Flo8p in the activation of FLO11 transcription.

The suppression patterns of ste12 and flo8 mutants by overexpression of FLO8 and STE12, respectively, in an otherwise wild‐type background support a model of their joint control over FLO11. Overexpression of FLO8 in a ste12 strain (SR1021) and overexpression of STE12 in a flo8 strain (SR1134) suppress the pseudohyphal and invasion defects of the mutants (Figure 5A). The morphologies of the pseudohyphae in these ‘suppressed’ strains is not identical to that of wild‐type. In flo8 strains overexpressing STE12, the cells of each pseudohyphal strand seem more elongated than the cells of wild‐type pseudohyphae. In ste12 strains overexpressing FLO8, the cells are not longer than wild‐type; however, they have a denser network of filaments than wild‐type strains. This colony morphology is similar to that of strains that are induced to form filaments by cAMP (SR959; Figure 1C).

Figure 5.

Cross‐suppression of ste12 and flo8 defects by overexpression of Flo8p and Ste12p. (A) Strains SR607 (wild‐type), SR1097 (flo8/flo8), SR1022 (ste12/ste12), SR1134 (flo8/flo8 pGAL1::STE12) and SR1021 (ste12/ste12, pRS202FLO8) were streaked on SLAD plates. Pictures of the colonies were taken after 16 h of incubation at 30°C. (B) Northern blots probed subsequently with ACT1 as a normalization probe and a FLO11 probe. Total RNA was prepared from strains SR1101 (flo11), SR607 (wild‐type), SR1097 (flo8/flo8), SR1022 (ste12/ste12), SR1021 (ste12/ste12, pRS202FLO8) and SR1134 (flo8/flo8 pGAL1::STE12) grown in SC to OD = 1.0–1.2. The two lanes shown for flo8/flo8 pGAL1::STE12 represent cells grown in SC glucose or SC galactose (SC‐Gal).

These patterns of suppression by overexpression are reflected in the FLO11 expression pattern. Overexpression of FLO8 enhances the expression of FLO11 10‐fold in a ste12 mutant (SR1021; Figure 5B). The reciprocal experiment in which STE12 is overexpressed is more difficult because high levels of Ste12p are toxic. To control the levels of STE12, we used a GAL::STE12 construct that could be regulated by galactose. In a flo8 strain (SR1134) containing this GAL::STE12 construct on a plasmid, the FLO11 transcript levels are increased 3‐fold on SC glucose medium as compared with the strain transformed with the control plasmid (SR1097). This increase probably represents incomplete repression of the GAL promoter. If STE12 is induced by incubation for 4 h in SC galactose medium, FLO11 expression is increased 10‐fold (when normalized to the level of ACT1 message; Figure 5B, last lane). Overexpression of STE12 in the ras1 ras2 pde2 strain leads to induction of FLO11 even in the absence of cAMP (data not shown).

To test whether cAMP activates transcription directly via Flo8p, we used a LexA–FLO8 fusion construct on a plasmid and a strain containing a β‐galactosidase reporter activatable by a proficient LexA (Golemis and Brent, 1992). The reporter is strongly activated by LexA–FLO8 (20‐fold) but not by LexA alone, demonstrating that Flo8p can act directly as a potent transcription factor. However, in this LexA system, we were unable to demonstrate an effect of cAMP on the ability of FLO8 to activate the reporter.

FLO11 transcription is also stimulated by high internal levels of cAMP

To determine whether a high level of internal cAMP stimulates FLO11 transcription, we deleted IRA1 in an otherwise wild‐type strain. Ira1p is a Ras‐GAP that inactivates Ras‐GTP by converting it to Ras‐GDP. Loss of Ira1p function leads to a higher proportion of activated Ras and thus to elevated cAMP levels in the cell (Tanaka et al., 1989). In the ira1 mutant strain (SR599), FLO11 transcripts are strongly induced (Figure 3A). This FLO11 induction is reflected in the hyperinvasive phenotype of strains devoid of IRA1 function (Figure 6). Both a ste12 ira1 and a ste11 ira1 mutant are hyperinvasive, illustrating that at least some of the cAMP signal is independent of the MAP kinase pathway. However, flo11 ira1 (SR1079) or flo8 ira1 (SR1132) strains fail to invade (Figure 6). These results are consistent with the interpretation that Flo8p is required for induction of FLO11 by high internal and external cAMP levels. A strain (SR1032) lacking both FLO8 and IRA1 has dramatically reduced levels of FLO11 transcripts, whereas deletion of STE12 in an ira1 background (SR1133) still shows FLO11 transcript levels comparable with wild‐type (Figure 3A). These results are consistent with the hyperinvasive phenotype of the ira1 ste12 strain and the non‐invasive phenotype of the ira1 flo8 strain.

Figure 6.

Deletion of IRA1 stimulates haploid invasive growth in ste12 but not in flo8 or flo11 strains. (A) The strains are patched onto YPD plates and incubated for 2 days at 30°C. All strains are haploid. The single mutant strains SR599 (ira1), 10560‐2B (wild‐type), L5795 (ste12), L5816 (flo8) and SR1101 (flo11) are patched as a control next to the double mutant strains SR1133 (ira1 ste12), SR1132 (ira1 flo8) and SR1079 (ira1 flo11). (B) Following incubation for 2 days at 30°C, the upper plates were washed under a stream of water and re‐photographed to yield the lower image.


The FLO11 promoter contains cis‐acting sites for diverse external and internal signals. We have shown that in addition to mating type control, nitrogen starvation and phase of growth, the FLO11 promoter is controlled by cAMP. Mutational analysis of the FLO11 promoter identified cis‐acting segments spread over 2.8 kb that mediate the trans‐acting signals. Our experiments support a model in which at least two signal transduction pathways, the cAMP/PKA pathway and the Kss1 MAPK pathway, regulate FLO11 transcription.

FLO11 is a target for the MAPK and cAMP/PKA pathways

FLO11 expression depends on signals from both the cAMP/PKA pathway and the Kss1 MAPK pathway. In agreement with a previous report (Lo and Dranginis, 1998), we find that the Kss1 MAPK pathway via STE12/TEC1 controls the expression of FLO11. In this work, we show that cAMP levels are also critical for FLO11 transcription. In a ras1 ras2 pde2 strain where internal cAMP levels can be altered by the addition of cAMP to the media, FLO11 transcript levels increase with increasing amounts of cAMP. Deletion of IRA1, a Ras‐GAP that leads to increased cAMP levels in the cell, also results in dramatically induced FLO11 transcript levels.

FLO8 is required for the cAMP induction of FLO11

We have found that FLO8, a known transcription factor required for invasive and filamentous growth (Liu et al., 1996), is required for FLO11 transcription. Strains that lack FLO8 fail to express FLO11 and cannot be induced to express FLO11 or to filament in the presence of high concentrations of cAMP. Unlike Flo8p, Ste12p is not required for cAMP induction of FLO11, suggesting that the cAMP pathway is distinct from the STE12/TEC1 branch. In agreement with this result, the Ste12p/Tec1p‐dependent FG::Ty1‐lacZ reporter is not regulated by cAMP.

Recent work has shown that TPK2, one of three catalytic subunits of the PKA in S.cerevisiae, is required for activation of FLO11 (Robertson and Fink, 1998). Although the simplest model is that FLO8 directly mediates the signal from Tpk2p, we were unable to show an effect of cAMP on the ability of FLO8 to activate transcription in a LexA reporter system. It is, of course, possible that the LexA constructs we used do not reflect the FLO8 activity accurately. In view of these results, we conclude that Flo8p is either activated directly by a cAMP/PKA signal or required in addition to the cAMP/PKA signal to induce FLO11 transcripts.

Flo8p and Ste12p control FLO11 expression independently

Increased expression of one of the two pathways that control FLO11 can bypass a block in the other. High cAMP levels or overexpression of FLO8 bypasses the requirement for STE12 for both invasion and filamentation and for induction of FLO11. This bypass occurs regardless of how the increase in cAMP levels occurs (exogenously by addition of the compound to a ste12 ras1 ras2 pde2 strain, SR1088, or endogenously by altering regulation of cAMP production in the ira1 ste12 strain SR1133). In contrast, both a functional FLO8 and FLO11 are required for induction of invasion and filamentation by cAMP. Overexpression of STE12 can also bypass the requirement for FLO8 for invasive and filamentous growth and for induction of FLO11, suggesting that each pathway can act independently of the other.

FLO11 contains multiple cis‐acting elements that respond independently to each signaling pathway

The trans‐acting factors STE12, TEC1 and FLO8 as well as cAMP each target multiple cis‐acting segments in the FLO11 promoter (Table III; Figure 2B). Flo8p acts on a 200 bp element in the FLO11 promoter (−1000 to −1200 bp) defined by the two overlapping segments FLO11‐6/5 and FLO11‐7/6 under all conditions tested. This promoter element, the UAS flo11‐6, is required for FLO11 expression as well as for invasive and filamentous growth. The same two overlapping segments, FLO11‐6/5 and FLO11‐7/6, are also most strongly induced by cAMP. These results support the conclusion that Flo8p is required to transmit the filamentation signal from the cAMP/PKA pathway. In post‐diauxic cells, FLO8 has an additional function in activating segments FLO11‐3/2 and FLO11‐8/7. The overlapping segments FLO11‐2/1 and FLO11‐3/2 define in wild‐type cells a UAS induced in post‐diauxic cells only. Like Flo8p and cAMP, Ste12p acts on multiple UASs, numbering four in total. Ste12p shows also some contribution to segment FLO11‐5/6 in exponentially growing cells. This contribution is not observed in post‐diauxic cells.

Two UASs (FLO11‐12/11 and FLO11‐10/9) also program transcription that is dependent on Tec1p, a known DNA‐binding partner of Ste12p. Interestingly, the dependencies on Ste12p and Tec1p are not perfectly concordant. An extreme example is the FLO11‐10/9 UAS which exhibits a 57‐fold dependency on Ste12p, but only a 3.5‐fold dependency on Tec1p. These data suggest Tec1p‐independent functions for Ste12p in FLO11 expression. Given that Ste12p uses multiple DNA‐binding partners in the mating signaling pathway, the current data may be a hint that Ste12p interacts with heretofore unidentified partners in addition to Tec1p. No effect of α‐factor on FLO11–lacZ expression could be detected.

The dual control of FLO11 explains several previous observations

Our model (Figure 7) accounts for a number of complexities that previously were difficult to explain. First, the MAPK pathway mutants are leaky, showing both residual filamentation and invasion (Lo et al., 1997). This phenomenon can be explained if the residual activity would be a consequence of some basal redundant activation of FLO11 by the cAMP branch. If both the cAMP/PKA pathway and the MAPK pathway are blocked (in a flo8 ste12 mutant for example), there is no residual activity. Secondly, our model accounts for earlier observations that the activated alleles of RAS are only partially blocked by mutations in the MAP kinase pathway (Mosch and Fink, 1997). Presumably these activated alleles can stimulate adenylate cyclase with a consequent increase in the levels of cAMP that can account for activation of FLO11 transcription. This explanation can be extended to interpret the enhanced filamentation and invasion of strains deleted for PDE2, the high affinity phosphodiesterase (Ward et al., 1995; Lorenz and Heitman, 1997). Thirdly, the FLO8 requirement for FLO11 activation explains why many laboratory strains that contain a mutation in FLO8 are unable to switch to pseudohyphal growth. The failure to express FLO11 also explains why these strains are not flocculent. Fourthly, the fact that FLO11 has upstream sites that mediate MAPK, cAMP/PKA, mating type and nutrition signals endows this single region with the ability to integrate the known factors that influence diploid pseudohyphal growth and haploid invasive growth.

Figure 7.

Model of two signal transduction pathways, the Ras‐regulated cAMP/PKA and Kss1 MAP kinase pathways, which induce invasive and filamentous growth by activating FLO11 transcription. The Kss1 MAP kinase pathway acts on Ste12p/Tec1p to induce FLO11 transcription. Flo8p is required for activation of FLO11 by the cAMP/PKA pathway. cAMP‐mediated Flo8p activation can be either direct (A) or indirect (B).

The FLO11 promoter is complex

Our analysis of the FLO11 5′ intergenic region underscores the complexity of the FLO11 promoter. First, the promoter is extremely large; direct mutational analysis shows that elements required for expression extend at least 2.8 kbp from the putative ATG. It is interesting to note that among the 10 yeast genes with the longest intergenic regions are two developmental genes known to contain large complex promoter regions, IME1 and HO (Table I). The promoters of each of these genes, like that of FLO11, respond to a variety of environmental and internal signals.

Our mutational analysis could detect four UAS and nine URS sites in the FLO11 promoter that are responsive to environmental signals such as rich medium, glucose depletion (post‐diauxic cells) or nitrogen starvation, but there may be many more. The individual cis‐acting segments can be key regulators under all conditions (flo11‐4, flo11‐7, flo11‐8, flo11‐14) or selectively required for a specific environmental condition (flo11‐10, flo11‐11, flo11‐12, flo11‐13). Most notably, flo11‐4, flo11‐10, flo11‐11 and flo11‐13 show haploid‐specific induction. In addition, flo11‐11 is strongly regulated by the growth phase of the cells, revealing a major induction in post‐diauxic cells (glucose depletion), whereas flo11‐12 and flo11‐13 are regulated by nitrogen availability. flo11‐5 is strongly regulated by nitrogen availability both in haploids and in diploids.

The mere presence of a large number of URS elements, however, increases the complexity of the analysis. Moreover, these elements could mask the contribution of any adjacent UAS element. For example, Lo and Dranginis (1998) pointed out that there is a degenerate FRE (filamentation and invasion response element) at bp −725 to −699. FLO11‐4/3 or FLO11‐5/4 both contain this element; however, neither shows significant stimulation of FLO11 expression nor a requirement for function of STE12 or TEC1. This site contains a URS (flo11‐4 in Table II) so we cannot rule out that the putative FRE site is functional but is masked by URS elements.

The mechanism by which the FLO11 promoter is controlled by trans‐acting proteins may be so complex that it cannot be reconstructed adequately by analysis of mutations in the cis‐ or trans‐acting elements. Although this type of analysis can identify the elements, the number of elements and the multiplicity of interactions may obscure the key interactions. A model for highly complex promoters has been proposed for the β‐interferon promoter (Thanos and Maniatis, 1995). The plethora of interactions at this promoter results in the formation of an enhanceosome, a multiprotein structure that involves interactions between DNA, DNA‐binding proteins and a host of ancillary proteins. A similar detailed in vitro analysis of FLO11 may be necessary to unravel the mechanism by which its transcription is controlled.

Materials and methods

Media and yeast strains

Strains used in this study are listed in Table IV. Standard methods for genetic crosses and transformation were used (Guthrie and Fink, 1991). All strains are congenic to the Σ1278b genetic background. The derivation of ras1, ras2, pde2, ste12, flo8 and flo11 strains has been described (Liu et al., 1993, 1996; Lo and Dranginis, 1996; Mosch et al., 1996; Kubler et al., 1997). Invasive and pseudohyphal growth assays were described (Gimeno et al., 1992; Roberts and Fink, 1994). Invasive growth was tested after 2 days incubation at 30°C. cAMP was added to liquid or solid media directly before use.

View this table:
Table 4. Saccharomyces cerevisiae strains used in this study


Plasmids used in this study are listed in Table V. Yeast–E.coli shuttle vectors have been described previously (Sikorski and Hieter, 1989; Christianson et al., 1992). The FG(TyA)::lacZ reporter plasmids are described in Mosch et al. (1996). Plasmids pira1::LEU2 and pira1::HIS3 were created by amplifying both flanking 400 bp regions of the IRA1 ORF, introducing EcoRI and BglII sites at the ends. The two fragments are cloned into a Bluescript KS+ EcoRI site, and the LEU2 or HIS3 marker is cloned into the BglII site, creating a complete deletion. The same method was used to create plasmid pcyr1::LEU2. The flo11::URA3 plasmid was obtained from Lo and Dranginis (1996).

View this table:
Table 5. Plasmids

Plasmid constructions

The LexA–FLO8 plasmid was constructed by amplifying the entire FLO8 ORF from B3265 using PCR primers that create BamHI sites at the ends. The ORF was subcloned into a modified pMAL bacterial expression vector (Fink Lab collection B3672), and full‐length expression was verified by a C‐terminal FLAG tag. The FLO8 ORF was sequenced and subcloned (without the FLAG tag) into pEG202. Correct fusion to the LexA was confirmed by sequencing.

A FLO11–lacZ reporter plasmid was constructed by amplifying the 3 kbp region 5′ of the ATG by PCR using primers containing a BglII siteat the end and cloning into YEp355 (Myers et al., 1986) (primers used: CGCACACTATGCAAAGACCGAGATCTTCC and GAAGATCTTCTCCACATACCAATCACTCG). The 14 individual deletions in this reporter were constructed by a primer overlap method (Higuchi, 1989) using a Bluescript KS+ plasmid containing the 3 kbp FLO11 promoter region (subcloned from YEp355‐FLO11::lacZ as an EcoRI–HindIII fragment). After mutagenesis, the partially deleted flo11‐nn promoter sequence was recloned into YEp355. The primers used for this purpose were:






























To generate the individual flo11‐n deletions in the chromosomal FLO11 promoter region, the URA3 gene was replaced by transformation with the respective EcoRI–HindIII‐digested KS+flo11‐Δn. Correct replacements were selected on 5‐fluoro‐orotic acid and verified by PCR.

To determine UAS sequence elements, 14 individual 400 bp elements, overlapping by 200 bp, were amplified by PCR and cloned into pLG669Z (Guarente and Ptashne, 1981) using a restriction site (XhoI) introduced at the 5′ end of the PCR primers. The primers used for this purpose were:






























The resulting 14 plasmids for the deletion series were transformed into strain 10560‐2B. At least three independent clones were tested using filter assays for equivalent expression of β‐galactosidase. Diploid strains were created by mating of the respective 10560‐2B flo11‐nn transformant with strain 10560‐5B by selection on YNB containing only leucine as a supplement.

The 14 plasmids containing the 400 bp sequence elements were transformed into 10560‐2B, L5795, L5816, L6149, L6150 and SR957. At least three independent clones were tested using filter assays for equivalent expression of β‐galactosidase. Filter assays were used to test the effect of α‐factor (5 μM).

The large number of FLO11 promoter elements and multiple conditions that affect those elements required some economy in the experimental design. The measurement of β‐galactosidase from our plasmid‐based reporter containing each of these elements permitted the rapid construction of strains and reproducible measurement of reporter activity. We confined the analysis to selective media so that the strains retained the plasmids. Our β‐galactosidase assays show, in general, a higher activity for haploid strains than for diploid strains. For rich medium, this result is in agreement with that found previously when FLO11 expression was monitored by measurement of steady‐state mRNA levels (Lo and Dranginis, 1998). When we measure FLO11::lacZ after 24–26 h (post‐diauxic growth) or on SLAD medium, haploid cells show an induction of >10‐fold, and diploid cells 5‐fold induction. This large induction of FLO11::lacZ for haploid cells on SLAD was not observed in the experiments where FLO11 mRNA was measured (Lo and Dranginis, 1998). This discrepancy could be due either to a difference in the stability of mRNA versus protein or the differences in the times at which the cultures were sampled. The expression of FLO11 as measured by either method is extremely sensitive to growth conditions and growth phase.

Preparation of RNA and Northern analysis

Total RNA was prepared using hot acid phenol, and Northern blots were performed as described in Ausubel et al. (1987). Strains deleted for ras1 ras2 pde2 were pre‐grown in SC + 1 mM cAMP to OD600 = 1.0. The cells were washed twice using SC and diluted to an OD600 = 0.3 into fresh SC medium or SC medium + 2 mM cAMP. The cells were grown to OD600 = 1.0 and harvested. Then 10–15 μg of RNA were separated on a formaldehyde‐containing agarose gel. For hybridization to FLO11 mRNA, a probe corresponding to bp 3502–4093 of the FLO11 ORF was used. The actin probe used corresponds to the 3′ exon of ACT1. The FLO1 probe corresponds to bp 1–500 of the FLO1 ORF.

Extract preparation and enzyme assays

Cells for β‐galactosidase assays of FG::TyA‐lacZ were incubated on SLAD plates for 3 days; cells for flo11::lacZ expression studies were grown in the respective liquid media and quantitated according to Mosch et al. (1996).

Cells for quantitation of flo11::lacZ expression in exponential growth phase were inoculated from confluent 20 h grown cultures 1:20 in fresh medium and grown for 4–6 h. Cells for quantitation of flo11::lacZ expression after the post‐diauxic shift were grown for 24–26 h. The same cultures were used to inoculate the exponentially growing cultures. Cells for quantitation of flo11::lacZ expression in SLAD medium were pre‐grown for 20 h in SC medium, washed twice with 2% glucose, diluted 1:5 into SLAD medium and grown for 4–6 h. For detection of cAMP‐induced FLO11 promoter segments, strain SR957 harboring the individual plasmids was grown overnight in selective medium containing 1 mM cAMP, transferred to media containing no cAMP for 10 h and split into two cultures containing 2 mM cAMP or no cAMP and grown for 4 h before harvesting.

Cells for the LexA::FLO8 assay were grown for 20 h in liquid SC medium, harvested and β‐galactosidase activity was determined.

Analysis of S.cerevisiae intergenic regions

A list of S.cerevisiae genomic regions that lack annotated features (i.e. that do not correspond to known genes, predicted ORFs >100 amino acids or transposons) was obtained from the Stanford Yeast Genome Database ( and were sorted based on their size. Because gene density decreases dramatically near chromosome ends, entries within 35 kbp of telomeres were eliminated. Regions >2.5 kbp were reexamined systematically for features and, if they were found to contain genes, Ty elements or Ty fragments, these and several of the remaining regions which have been found to encode RNAs (Olivas et al., 1997) were eliminated. The 10 largest regions are shown in Table I. It is important to note that uncharacterized expressed genes with ORFs <100 amino acids are not annotated and, consequently, may be found to exist in regions currently labeled as intergenic. Conversely, predicted ORFs >100 amino acids may occur by chance, which would result in the underestimation of the sizes of the intergenic regions which contain them.


We thank Anne Dranginis for kindly providing plasmids, and Fran Lewitter for help with the analysis of the size of intergenic regions in the yeast genome. We thank Tim Galitski for helpful comments on the manuscript, and the members of the Fink Laboratory for many fruitful discussions. S.R. was supported by a Research Fellowship from the Deutsche Forschungsgemeinschaft (DFG). This work was supported by grants from the National Institutes of Health (GM40266 and GM35010 to G.R.F.). G.R.F. is an American Cancer Society Professor of Genetics.


View Abstract