The data presented here demonstrate a role for various alcohols in both cellular and colony morphology in S. cerevisiae. Some of these alcohols are products of amino acid metabolism, such as isoamyl alcohol and butanol, which accumulate specifically in conditions of nitrogen starvation. These alcohols stimulate haploid cells to differentiate into a filamentous form similar to diploid-specific, nitrogen starvation–induced, pseudohyphal development. Both of these phenomena involve an elongated cell morphology, alterations in budding pattern, and a dependence on elements of the pheromone-responsive MAPK pathway.
These phenotypes are strongly dependent on the particular yeast strain under study. The structures formed by strains of the Σ1278b background are strikingly similar to pseudohyphal cells, with elongated, cylindrical cells. W303 derivatives, however, adopt a variety of morphologies, including ellipsoidal yeast-form cells, elongated, cylindrical shapes, and rounded yeast cells projecting a thin, hyphal projection reminiscent of germ tube formation in C. albicans. In contrast, strains of the S288c lineage seem completely unaffected by the presence of these alcohols.
This phenotype represents the third process that the pheromone-responsive MAPK pathway regulates in haploid cells. First is mating response, which is activated by pheromone binding to a cell surface receptor/G protein complex and signals through the FUS3 MAPK associated with the STE5 scaffolding protein. Second is haploid invasion, a process stimulated by unknown stimuli, which does not require STE5 and signals primarily through the KSS1 MAPK. Third is butanol-induced filamentation, which also uses the same MAPK pathway. We have not, however, analyzed the roles of STE5, FUS3, and KSS1 in detail. Reporter genes responsive to either pheromone activation (FUS1-lacZ) or filamentation/invasion activity (FRE-lacZ) do not respond to butanol, indicating again that there is a specialization of this signaling pathway. Furthermore, the phenotypes of Δste20 mutations differ in these assays. In the Σ1278b background, Δste20/Δste20 mutant strains have a severe defect in pseudohyphal growth—more severe, in fact, that other ste mutants. However, Δste20 haploid strains have only a modest defect in mating and essentially no defect in butanol-induced filamentation (our unpublished observations). Thus, there must be pathway-specific specialization at the STE20 step, as there is for STE12 and the MAPKs KSS1 and FUS3.
Although these morphological changes are at least partially dependent on the STE MAPK pathway, they are independent of elements of the nutrient sensors, including GPA2, GPR1, and MEP2. This suggests that the alcohols are bypassing the need for nitrogen starvation, an idea supported by the behavior of strains grown in rich (YPD) liquid medium plus butanol (Figure ) and by the filamentous colony morphology on nitrogen-rich, glucose-poor medium (Figure ). This would be similar in concept to haploid invasive growth, in which haploid cells grown on rich solid medium invade the agar substrate. Because this occurs on rich medium, it is unlikely to be a nutrient response, and the mutations in the nutrient-sensing machinery have either no defect in invasive growth (Δmep2
; Lorenz and Heitman, 1998a
) or a severe defect (Δgpa2
; Pan and Heitman, 1999
A screen for additional mutants that affect butanol-induced filamentous growth identified several genes not previously appreciated to have filamentation phenotypes. Genes such as BEM1, BEM4, BUD8, and FIG1 have been implicated in polarized growth; thus, their involvement in this phenomenon is not surprising. The effects of other genes, such as CHD1, which encodes a transcription factor homologue, are less obvious. With the exception of only the putative mitochondrial helicase HMI1 (and perhaps FIG1), mutations that block butanol-induced (haploid) filamentation also block nitrogen-induced (diploid) filamentation.
In addition to the haploid phenotypes, we found that ethanol induces hyperfilamentation of diploid strains on low-nitrogen medium. Ethanol has no effect on the colony or cellular morphology of haploid cells. Again, this phenotype requires TEC1 and the pheromone-response elements that regulate filamentous growth, but it does not require GPA2, GPR1, and MEP2. Thus, as with the haploid phenotype, ethanol stimulates filamentation in a manner that bypasses the nutrient-sensing machinery.
Why do these alcohols have effects on colony morphology? It has been suggested that pseudohyphal differentiation is a means by which yeast cells scavenge for nutrients in scarce conditions (Gimeno et al., 1992
). Because these alcohols bypass the presumptive nutrient-sensing apparatus, it is possible that they represent an alternative means to sense nutrient availability. As yeast cells metabolize the available nutrients and their own proteins and amino acids as nitrogen sources, the concentration of by-products such as isoamyl alcohol and, in particular, ethanol increases. Yeast may have a mechanism to estimate nutrient availability based on the levels of its own by-products. Alternatively, these alcohols may be toxic to the cell, and high concentrations may stimulate filamentous growth to allow the cell to escape the poisoned environment. Indeed, the presence of butanol, isoamyl alcohol, or ethanol (at high concentrations) slows growth rates, supporting this idea.
A third possibility is that yeast uses the concentration of these alcohols as a mechanism to sense population density and coordinate development appropriately. Quorum sensing of this nature is common in bacteria. Population density is one signal that regulates competence development in Bacillus subtilis
. A secreted pheromone (ComX) activates a two-component signaling pathway once it passes a threshold concentration (reviewed by Grossman, 1995
). Other bacteria, such as Vibrio fischeri
, control autofluorescence based on population density with the use of secreted autoinducers, mostly related to N
-acyl homoserine lactones (reviewed by Hellingwerf et al., 1998
). Quorum-sensing systems have also been described in both plant and human pathogens (Agrobacterium tumefaciens
and Pseudomonas aeruginosa
, respectively) to regulate expression of virulence factor genes.
Whatever the reason for this phenomenon in yeast, there must be a cellular mechanism to sense the presence of these alcohols. Although the STE MAPK pathway is required for the full expression of butanol-induced phenotypes, this pathway is not likely to represent the only butanol-responsive signaling system in yeast. As shown in Figure and Table , haploid Δste12 mutants change their budding pattern in response to butanol; thus, this element of the phenotype is necessarily independent of the MAPK pathway. For this reason, we expect there to be a system (or systems) to recognize the presence of these alcohols and transduce a signal to multiple downstream pathways, including the MAPK pathway.
It is possible that, rather than sensing the alcohols themselves, cells sense intermediates in the conversion of amino acids to alcohols. By this model, addition of alcohols to the medium would be expected to increase the concentration of these intermediates in the cell. Several distinct pathways have been identified that convert leucine to isoamyl alcohol or valine to isobutyl alcohol (Dickinson et al., 1997
); the pathways are overlapping and interconnected, supporting the idea that an intermediate could be the relevant signaling molecule. We tend not to support this idea, though, because we also see these affects with 1-butanol and tert
-amyl alcohol, which, although related to isoamyl and isobutyl alcohols, are not derived from any common amino acids. Moreover, neither α-ketoisocaproic acid (derived from leucine) nor α-ketoisovaleric acid (derived from valine) affected colony morphology when added to solid SLAD medium (Lorenz, Cardenas, and Heitman, unpublished observations).
There are a few precedents for the idea of branched chain amino acids or their derivatives in signaling roles. One study has proposed a role for branched chain amino acids (such as leucine and valine) in the control of translation (Xu et al., 1998
). Addition of these amino acids to pancreatic β-cells stimulated phosphorylation of the PHAS-I and p70S6k
proteins. Phosphorylated forms of both PHAS-I and p70S6k
stimulate translation, PHAS-I via interactions with the mRNA cap-binding protein eIF-4E and p70S6k
via phosphorylation of ribosomal protein S6. Although this study correlated the effects with essential versus nonessential amino acids, it also showed that α-ketoisocaproic acid had this effect as well, suggesting that the relevant signaling molecule may be a by-product rather than the amino acids themselves. In microorganisms, branched chain amino acid metabolism has been linked to growth and development in the Gram-negative bacterium Myxococcus xanthus
(Toal et al., 1995
). Mutations at the esg
locus confer growth defects in minimal medium and defects in aggregation and differentiation during development of multicellular fruiting bodies. The esg
locus encodes a branched chain keto acid dehydrogenase, an enzyme that converts α-ketoacids (the transamination products of branched chain amino acids) to short, branched chain fatty acids. Indeed, several fatty acids can rescue the developmental defects of esg
mutants. In this case, the signaling molecule is not an alcohol by-product but a fatty acid; nevertheless, the involvement of branched chain amino acid metabolism is shared between development in M. xanthus
and the phenotypes described here for S. cerevisiae
Butanol-induced filamentous growth offers an additional advantage that has thus far been impossible in the analysis of filamentation. The recent advent of DNA array or chip experiments presents the possibility of understanding the transcriptional program of filamentous growth. This is undoubtedly complex, because a large number of transcriptional regulators (perhaps 18 to date) have been linked with filamentous growth in many laboratories. Standard pseudohyphal growth is confined to solid medium, and not all cells become elongated or invasive, thus making array experiments extremely difficult. Butanol allows “filamentous” growth in liquid medium, and virtually all cells show some aspects of this behavior, making the butanol-induced phenomenon amenable to array analysis. These experiments are under way and hopefully will shed light on the regulation of filamentous growth.