Genomewide identification of pheromone-targeted transcription in fission yeast

The fission yeast Schizosaccharomyces pombe provides an attractive model system for studying cellular differentiation processes in response to environmental cues. It is a unicellular, haploid organism with a conventionally organized cell cycle [1], and it is easily subjected to genetic analysis [2]. Furthermore, since its entire genome sequence is known [3], genome-wide analysis tools can be readily applied.

In response to nitrogen starvation fission yeast cells undergo sexual differentiation. In this process haploid P and M cells first mate to form dipoid zygotes [4], which then undergo meiosis and sporulation [5]. Nitrogen starvation leads to the establishment of a pheromone signalling system by which M and P cells communicate with each other (see Figure 1A), causing growth arrest in the G₁ phase of the cell cycle. Pairs of G₁ arrested M- and P- cells subsequently agglutinate, mate, and undergo karyogamy, after which the resulting diploid zygotes enter meiosis and finally sporulate.

The HMG-box transcription factor Ste11 is a master regulator of the sexual differentiation programme [6]. Upon starvation, Ste11 activates numerous genes required for mating and meiosis, including the mat genes specifying cellular mating type, many genes encoding components of the pheromone communication system, and mei2 that encodes the ultimate activator of meiosis. Ste11 binds to one or several Ste11-sites (AACAAAGAAA) present in the promoter regions of the genes it directly controls.

The two pheromones are small, secreted peptides that bind to and activate cell-type specific 7TM receptors on the cell surface (Mam2 in M cells and Map3 in P cells), thereby activating the receptor-coupled G-protein, Gpa1. This stimulates a MAP kinase cascade, consisting of Byr2 (a MAP3K), Byr1 (a MAP2K) and Spk1 (a MAPK). The pheromone response, which is shared by the two cell types, activates genes required at various steps throughout the differentiation process, with entry into meiosis being the most downstream event reported to rely on the signal [7].

Several lines of evidence suggest that Ste11 is also the transcription factor activated by pheromone signalling. Pheromone-controlled genes often have Ste11-boxes in their promoter regions, and hyper-activation of the MAP2K Byr2 causes constitutive activation of Ste11 [8]. Spk1 can in fact directly phosphorylate Ste11 on threonines 305 and 317 in vitro, and mutational analysis of these residues in vivo is consistent with a direct activation mechanism [8]. However, since Ste11 is absolutely required for the expression of many components in the response pathway, it is difficult to evaluate whether pheromone signalling can stimulate transcription in the absence of Ste11.

The analysis of pheromone-controlled transcription is further complicated by the nutritional control over the differentiation pathway. In order to make them respond to mating pheromones, it is necessary to starve cells for nitrogen, which in itself causes dramatic changes in the expression of many genes [9, 28]. Furthermore, the transcriptional responses to nitrogen starvation and pheromone stimulation are partially overlapping, and Ste11-controlled genes appear to be induced to varying degrees by both these environmental cues (Figure 1B). Some genes, such as ste6 [10] are already strongly induced by nitrogen starvation and become further enhanced by pheromone (Figure 1B, panel 1). Expression of other genes like fus1 [11] has an absolute requirement for both nitrogen starvation and pheromone signalling (Figure 1B, panel 2). Genetic analysis has shown that byr2, byr1 and ste11 are required for induction of pheromone controlled genes by nitrogen starvation [12], suggesting that this signal is also conveyed by the pheromone-responding MAPK pathway.

An additional layer of regulation is conferred by cellular mating type. Whilst most pheromone-controlled genes appear to be equally expressed in the two cell types, a small subset is either M-specific (Figure 1B, panel 3) or P-specific (Figure 1B, panel 4). These genes are involved in synthesis and processing of the cell-type specific extracellular components of the pheromone-communication system, i.e. the pheromones and receptors. Their expression is controlled by the allele present at the mat1 locus of the cell [13]. In M cells the mat1-Mc gene causes activation of M-specific genes, while the mat1-Pc gene activates P-specific genes in P cells. Mat1-Mc is an HMG-box protein that appears to recruit Ste11 to a weaker version of its binding site (AACAAAGA) present in the promoter region of M-specific genes [14]. The mechanism by which Mat1-Pc activates the P-specific genes is less clear. Genetic analysis suggests that it cooperates with the Map1 protein, a putative transcription factor of the MADS-box family [15, 16].

In the present study we use genomic microarrays to identify pheromone-induced genes in fission yeast. In order to discriminate pheromone-induction from the effect of nitrogen starvation, we utilize the fact that cyr1 strains defective in cAMP production are derepressed for mating activities, and hence respond to pheromone without being starved [17]. We have identified classes of genes induced by M-factor in P cells and genes induced by P-factor in M-cells. As expected, these two groups of genes were highly overlapping. Moreover, a direct comparison of M and P cells allowed us to identify several new genes that are either M- or P-specific. Our results support the notion that Ste11 is the major transcription factor bringing about pheromone-stimulated transcription.

In fission yeast, nitrogen starvation causes induction of gene expression by at least three different signalling pathways. First, it leads to a modest reduction in cellular cAMP, which in turn down regulates the glucose sensing Pka1 protein kinase that directly represses transcription of genes such as fbp1 and ste11 [25]. Secondly, nitrogen starvation activates the Spc1/Sty1 stress response pathway [26], which causes induction of the atf1 transcription factor that also controls ste11 [27]. Finally, genetic analysis suggests that the pheromone-responding MAPK pathway conveys the signal of nitrogen starvation to pheromone controlled promoters [12].

In the present study we have used a cyr1 genetic background to separate pheromone-induced transcription from the general effects of nitrogen starvation. Consequently, we found that pheromone stimulation of cyr1 cells caused a much less dramatic change in expression profiles than nitrogen starvation of wild-type cells [9, 28]. By this approach we have identified a total of 163 genes that were more than two-fold induced by pheromone stimulation, including several genes previously reported to be pheromone controlled.

Among the pheromone-induced genes we found the cAMP degrading phosphodiesterase Cgs2, suggesting that pheromone signalling enforces the reduction in cAMP caused by nitrogen starvation [29]. This gene was also shown to be induced by activation of the Spc1/Sty1 stress response pathway [30], suggesting that cgs2 constitutes a junction for cross talk between different starvation sensing pathways. When cAMP levels drop, the APC/C E3 ubiquitin ligase becomes activated [31], and this causes G1 arrest via degradation of the S-phase activating cyclin Cig2 [32]. Consistent with an increased APC/C activity during differentiation, we found apc2 encoding a subunit of APC/C among the pheromone-induced genes.

Our identification of components in the pheromone signal transduction pathway (spk1,ste4 and ste11) suggests that initial pheromone stimulation triggers a positive feedback loop enforcing the response. Furthermore, the whi5 gene, whose counterpart in S. cerevisiae encodes an inhibitor of transcriptional activities required for vegetative S-phase entry [33, 34], was among the induced genes, suggesting that pheromone signalling inhibits vegetative growth. Taken together these observations underscore the autocatalytic and irreversible nature of the developmental switch.

On the other hand, functions that dampen the pheromone response were also identified. Thus, the msa1 gene encoding an RNA-binding protein that negatively regulates differentiation specific transcripts [35], and rgs1 coding for a GTPase that down-regulates the receptor-coupled G-protein Gpa1 [36] were both pheromone induced, stressing the importance of a tight modulation of the response during differentiation.

The results presented here are consistent with Ste11 being the transcription factor activated by pheromone signalling. The pheromone-regulated genes in Figure 5 contain perfect matches to a matrix constructed from known sites and they together account for about 30% of the pheromone-induced transcripts. The remaining pheromone-regulated genes almost invariably contain more diverged derivatives of the Ste11-binding site, often in multiple copies. It is thus likely that Ste11 regulates the majority of pheromone-regulated genes, perhaps in some cases via co-operative binding to a number of relatively weak binding sites as previously reported [14]. It is, however, likely that some pheromone-regulated genes are indirectly induced by other transcription factors, either alone or in combination with Ste11. Indeed, several transcription factors such as rep1 and mei4 are found within the group of pheromone-induced genes.

The isp6 gene, which we found to be pheromone induced, was actually reported to be induced by nitrogen starvation independently of ste11 [37], and it does not have an obvious Ste11-box in its promoter region [see Additional file 1]. The cAMP pathway negatively regulates isp6 transcription [37], and presumably the pheromone effect on this gene is caused indirectly by cross talk with other signalling pathways as discussed above.

The cig2 gene, encoding an S-phase activating cyclin, is specifically expressed in the G₁ phase of the cell cycle [38], and hence its induction by pheromone could be an indirect consequence of pheromone-mediated G₁ arrest However, the cig2 promoter contains a perfect Ste11-box [see Additional file 1], suggesting that Ste11 and the pheromone pathway may also directly control this gene. The regulation of cig2 represents a paradox: Transcription of the gene is on the one hand induced by pheromone, yet the same signal also mediates proteolysis of the Cig2 protein [32]. Presumably this regulation reflects the fact that Cig2 needs to be down-regulated in order to activate the differentiation pathway, but then Cig2 is needed again soon after for activation of pre-meiotic S-phase [39].

By comparing the effects of M- and P-factor stimulation we were able to identify P-and M-specific genes. Given the extensive hunt over many years for cell-type specific mutants, it was no surprise that we only found very few new genes, and an important conclusion of this work is that the list in Table 2 can be regarded, as close to being complete. The larger number of M-specific genes largely reflects the more elaborate synthesis of M-factor relative to P-factor.

Our results confirm the proposed mechanism of M-specific gene activation [14], where the Mat1-Mc protein both recruits Ste11 to a weak binding site and also binds to an M-box sequence situated approximately 1.5 or 2.5 helical turns upstream of the Ste11-box (Table 3). The control of P-specific transcription is still an enigma that awaits functional promoter dissection studies to be resolved.

Our observation that the P-specific transcript map3 extends into the neighbouring gene SPAC3F10.09 demonstrates the importance of mapping the genome-wide location of transcripts and the use of this information in the design of future DNA microarrays. Using a different array system, SPAC3F10.09 was reported to be induced in meiosis [9], presumably also reflecting an erroneous detection of map3 transcripts transcribed from the opposite strand. Convergent overlapping reading frames are relatively rare in the fission yeast genome, but our finding that the map3 transcript extends more than 500 bases downstream of the reading frame suggests that convergent transcripts may overlap even when the reading frames do not.

We also found 92 genes that were more than twofold down regulated by both pheromones (data nor shown). GO categories representing nucleotide and amino acid metabolism were significantly over-represented among these genes, suggesting that specific mechanism exist either to repress these genes or degrade their transcripts during differentiation.

Pheromone communication in fungi was originally considered to be a mechanism by which cells of opposite mating type identified each other, but subsequent studies have demonstrated that it controls many other functions than mere cell recognition. In fission yeast pheromone signalling brings about G₁ cell-cycle arrest prior to mating, cell elongation towards the partner cell, and is also involved in cell fusion and karyogamy. Finally, it has been known for a decade that entry into meiosis also requires activation of the pheromone pathway. This is because the two meiotic subfunctions mat1-Pm and mat1-Mm are pheromone controlled; if they are expressed from a heterologous promoter, meiosis is readily activated in the absence of pheromone signalling [7]. Our results in the present study show that – although dispensable for completion of meiosis – pheromone control extends beyond the point of meiotic entry. Thus, several genes required for specific meiotic sub-functions in particular for meiotic recombination turned out being pheromone induced. Consistent with this, expression of the dmc1 gene encoding a RecA like helicase was previously shown to be dependent on Ste11 [40]. However, we note that many of these genes do not have optimal Ste11 binding sites in their promoters [see Additional file 1], suggesting that their induction by pheromone may be indirect. Nevertheless, it is evident from our array results that pheromone stimulation in itself can bring about a robust induction of these genes.

Previously, the bqt1 gene mediating telomere clustering prior to meiotic horsetail movement was shown to be pheromone induced [41]. These nuclear movements are already initiated in haploid cells challenged with pheromone [42]. We found that dhc1 and dic1, encoding dynein heavy and intermediate chains respectively were both pheromone induced. Dynein is required for the horsetail movements, and a null mutant in dhc1 significantly reduces meiotic recombination [43]. Furthermore, the dynein-associated Ssm4 protein is also pheromone regulated [44] and this study). Taken together, these observations suggest that pheromone signalling orchestrates the entire process of pre-meiotic chromosome movement and associated recombination. It will be interesting to learn whether oscillations in pheromone signalling play a dynamic role in bringing about these processes.

We have identified 163 pheromone induced genes in fission yeast. Most genes are induced by pheromone in both cell types, and analysis of GO annotations associated with pheromone induction showed that the genes were involved in many aspects of sexual differentiation, including conjugation, and morphological processes associated with mating. Surprisingly, many meiotic gene functions were also pheromone controlled. Our results confirmed that the Ste11 transcription factor controls many pheromone-induced genes. Finally, most cell-type specific genes now appear to be identified in fission yeast.