The budding yeast Saccharomyces cerevisiae is a eukaryotic microorganism that is able to choose between different unicellular and multicellular lifestyles. The potential of individual yeast cells to switch between different growth modes is advantageous for optimal dissemination, protection and substrate colonization at the population level. A crucial step in lifestyle adaptation is the control of self- and foreign adhesion. For this purpose, S. cerevisiae contains a set of cell wall-associated proteins, which confer adhesion to diverse biotic and abiotic surfaces. Here, we provide an overview of different aspects of S. cerevisiae adhesion, including a detailed description of known lifestyles, recent insights into adhesin structure and function and an outline of the complex regulatory network for adhesin gene regulation. Our review shows that S. cerevisiae is a model system suitable for studying not only the mechanisms and regulation of cell adhesion, but also the role of this process in microbial development, ecology and evolution.
Adhesion of cells to each other and to foreign surfaces is key for multicellular development, colonization and pathogenesis. Adhesive properties are predominantly conferred by specific cell surface proteins, the adhesins. In metazoa, cell adhesion is mediated by several classes of diverse cell adhesion molecules, whose structures and functions during development have been studied in detail (Beckerle, 2002; Abedin & King, 2010). In bacteria, the central role of adhesins during biofilm formation and host–microorganism interactions is becoming more and more evident and is at the focus of many investigations that include diverse bacterial species (Kline et al., 2009; Flemming & Wingender, 2010). In fungi, functional studies in the budding yeast Saccharomyces cerevisiae and in pathogenic Candida species have shown that adhesins allow agglutination of sexual partners before cell fusion, enable the formation of protective and invasive multicellular growth forms, mediate adherence to foreign biotic and abiotic surfaces and confer interactions with solid substrates and host cells (Hoyer, 2001; Kaur et al., 2005; Verstrepen & Klis, 2006; Dranginis et al., 2007). Comparative genomics fueled by the growing number of genome sequences has helped to identify putative adhesin genes and families in other fungal species as well (Verstrepen et al., 2004; Galagan et al., 2005; Linder & Gustafsson, 2008; Ramage et al., 2009). However, many fungi exhibit only a poor experimental tractability. Therefore, detailed insights into fungal adhesin structure and function as well as into the regulatory networks involved must rely on studies with tractable model systems.
Saccharomyces cerevisiae is a formidable organism to investigate almost every aspect of eukaryotic molecular and cell biology. This also includes adhesion of sexual partner cells as well as nonsexual, vegetative adhesion. Sexual adhesion of budding yeast is well understood and is mediated by cell-type-specific adhesins called agglutinins, which are produced by mating partners after exchange of pheromones and confer cell–cell adherence by high-affinity heterotypic protein–protein interactions (Lipke & Kurjan, 1992; Chen et al., 2007; Dranginis et al., 2007). In contrast, detailed molecular insights into vegetative adhesion in S. cerevisiae have been lacking, most likely because many laboratory strains have acquired mutations that suppress this process (Liu et al., 1996). On the other hand, the process of flocculation, the reversible adhesion of vegetative cells, has been widely studied by classical genetics in industrial production strains (Bauer et al., 2010). However, detailed biochemical and structural studies of the responsible adhesins (flocculins) have been performed only recently (Douglas et al., 2007; Veelders et al., 2010; Goossens et al., 2011). The ‘rediscovery’ that S. cerevisiae is able to develop diverse multicellular growth forms such as filaments and biofilms (Gimeno et al., 1992; Roberts & Fink, 1994; Reynolds & Fink, 2001) has led to a significant boost in the research of vegetative adhesion in this organism. These efforts have uncovered a highly complex regulatory network for the control of vegetative adhesion and multicellular development and provided detailed insights into adhesin structure and function. In addition, these studies provided new clues for the ecophysiological significance of adhesion.
In this review, we summarize the current knowledge on the role of vegetative adhesins in the establishment of multicellular lifestyles in S. cerevisiae. We first discuss the diverse growth forms with respect to their importance for basic research and industrial applications. Here, we also focus on the potential ecological significance of different S. cerevisiae lifestyles and their role as models for virulent growth forms of human pathogenic yeasts. In a second part, we present new insights into the structure and function of S. cerevisiae vegetative adhesins and we discuss their impact on our understanding of the molecular mechanisms that underlie cell–cell and cell–substrate adhesion. Finally, we will describe the complex regulatory network that controls adhesin gene expression in S. cerevisiae through numerous conventional and epigenetic mechanisms. Here, we also discuss the mechanisms that enable individual cells or whole populations to adapt specific adhesive properties in response to changing environments. Our review will not cover other aspects of multicellular development in S. cerevisiae such as intercellular communication, cell cycle control or regulation of cell polarity and morphogenesis, as these topics have been reviewed recently (Rua et al., 2001; Palkova & Vachova, 2003; Bähler, 2005; Arkowitz, 2009; Piel & Tran, 2009; Slaughter et al., 2009; Wang, 2009; Enserink & Kolodner, 2010; Moore & Cooper, 2010; Perez & Rincon, 2010).
Lifestyles of S. cerevisiae
The observation that vegetative cells of S. cerevisiae can adopt a multicellular growth form was made more than 100 years ago by Emil Hansen, who described the formation of large yeast aggregates (‘skins’) in industrial strains after fermentation (Hansen, 1883). Later on, additional multicellular forms were described using diverse industrial and laboratory strains and they were named as flocs (Lindquist, 1952), flors (Allan, 1939), biofilms (Cruess, 1938; Reynolds & Fink, 2001) and pseudohyphal filaments (Morris, 1958; Gimeno & Fink, 1992; Gimeno et al., 1992). The principal features of these growth forms are shown in Fig. 1, with a focus on the adhesion properties involved. Basically, development of the different lifestyles requires adhesion of yeast cells to each other (self-adhesion) or adhesion of cells to foreign biotic or abiotic surfaces (foreign adhesion). In this context, foreign surfaces are defined as structures not present on S. cerevisiae cells. Multicellular development is further influenced by the nature of the environment, which in principle can be a solid substrate directly exposed to air or a liquid medium with interfaces to the air and to the solid bottom.
As shown in Fig. 1, S. cerevisiae can choose between a number of different lifestyles. (1) On solid substrates exposed to air, cells that do not produce adhesins are sessile and will develop nonadhesive colonies. This is the typical growth form of many laboratory strains on solid agar media. (2) When expressing genes for self-adhesion, yeast cells can aggregate and form nondissolvable colonies that do not adhere to the substrate. (3) In order to develop nonremovable biofilms, adhesins must be produced that confer foreign adhesion. Typically, cells in biofilms adhere to each other and to the foreign surface. Foreign adhesion is also important for the formation of invasive filaments, which are able to penetrate solid substrates. A simple test for self- and foreign adhesion is a wash assay, in which colonies or larger cell patches are exposed to a stream of water (Roberts & Fink, 1994). This will dissolve nonadhesive colonies and remove cells from the agar surface. (4) In a liquid medium, nonadhesive yeast cells are planktonic and will produce turbid cultures of individual cells. This is the classical growth form of most laboratory strains in liquid media. (5) When producing proteins for self-adhesion, yeast cells can form aggregates that may sediment to the bottom (flocs) or that can float on the liquid surface (flor). These growth forms are observed in diverse industrial strains during the production of alcoholic beverages. (6) In liquid substrates, biofilms and/or filaments may also develop if yeast cells are able to adhere to the surface at the bottom.
Although multicellular growth forms of S. cerevisiae have been observed since the 19th century, the molecular basis for adhesion remained unclear until the isolation of the responsible genes. Initial studies uncovered genes for floc formation, which encoded proteins that were accordingly named flocculins (Miki et al., 1982a, b; Stratford, 1989a, b, 1992). To date, at least eight different adhesins have been identified in diverse industrial and laboratory strains that confer vegetative adhesion (Fig. 2). They include FLO1, FLO5, FLO9, FLO10, FLO11 (or MUC1), FIG2 and AGA1, which all encode cell wall-associated surface proteins that belong to the yeast adhesin family (Dranginis et al., 2007). Although the expression of S. cerevisiae adhesin genes is highly heterogeneous depending on the genetic background and the environmental conditions, they have been found to lead to the formation of flocs, flors, biofilms or filaments when sufficiently expressed. Thus, the encoded proteins are likely to be involved in the binding of specific ligands present on other yeast cells or on foreign surfaces.
To understand in detail how S. cerevisiae uses adhesin genes to choose the appropriate lifestyle, at least three key issues must be addressed. (1) The specific requirement of adhesin genes for the development of the diverse growth forms must be determined. This should also involve a more detailed characterization of the known growth forms and even aim at the discovery of as yet unknown lifestyles. (2) The structural and functional properties of the encoded proteins must be characterized in detail. Here, the domains at the cell surface are of special interest as well as the ligand molecules that are specifically bound. (3) The regulation of adhesin genes and proteins by the environment must be studied. Specifically, the signaling pathways and the exact mechanisms that control adhesin gene expression and protein function must be uncovered in detail.
The following paragraphs will summarize the current knowledge on the diverse growth forms, whereas structure–function relationships and regulation of adhesins will be discussed in the subsequent chapters.
Flocculation has been studied extensively in industrial S. cerevisiae strains, because of its relevance for many biotechnological applications. Specifically, flocculation is a fast, cost-effective and environment-friendly way to remove yeast cells at the end of fermentation processes in the production of, for example beer, wine, ethanol or biodiesel (Bauer et al., 2010; Soares, 2010). In addition, flocculent yeast strains that effectively bind Ca2+ ions have been used in bioremediation to remove other divalent ions, for example heavy metals from contaminated sites (Wang & Chen, 2006). Industrial strains have been widely analyzed with respect to flocculin gene variability and expression as well as with regard to other factors that influence flocculation, such as nutrient availability, pH, temperature, oxygen, cell density, agitation, ions, sugars, ethanol and cultural age. Brewing yeast strains for instance have been historically classified according to their ability to form flocs that either rise to the surface or sediment to the bottom (Dengis et al., 1995; Dengis & Rouxhet, 1997). Top-fermenting (ale) yeasts were predominantly classified as S. cerevisiae, whereas bottom-fermenting (lager) yeast strains are natural hybrids between S. cerevisiae and Saccharomyces bayanus and were named Saccharomyces pastorianus (Yamagishi & Ogata, 1999; Dunn & Sherlock, 2008; Ogata et al., 2008). Initial genetic analysis revealed that S. pastorianus strains carry a special type of flocculin gene called Lg-FLO1 (Kobayashi et al., 1998). In contrast to Flo1 present in many laboratory strains, Lg-Flo1 is sensitive not only to mannose, but also to glucose. This mannose–glucose-sensitive adhesion has been named NewFlo phenotype and is of considerable industrial importance, because the relaxed sugar specificity of Lg-Flo1 ensures that flocculation occurs only at the end of fermentation, when all sugars that inhibit floc formation are consumed (Stratford, 1989a, b; Stratford & Assinder, 1991; Verstrepen et al., 2003). In recent years, a remarkable flocculation gene variability in different industrial brewing yeast strains has been revealed (Sato et al., 2001, 2002; Ogata et al., 2008; Van Mulders et al., 2010). This explicit genetic variation has been attributed to the extraordinary mutational frequency observed for flocculin genes and might hamper the stability of flocculation in industrial processes (Verstrepen et al., 2005; Rando & Verstrepen, 2007). Therefore, present and future strategies for optimized control of flocculation include global genetic analysis of diverse yeast strains and their targeted genetic improvement by, for example controlled expression of flocculin genes that confer specific adhesion properties (Govender et al., 2008, 2010; Bauer et al., 2010; Saerens et al., 2010). Comparative genomics of diverse industrial strains might also help to reveal the molecular basis for other factors that determine whether flocs rise to the surface or sediment to the bottom, for example the inclusion of carbon dioxide or the adsorption to rising gas bubbles (Soares, 2010).
The relevance of flocculation in industrial applications has long been recognized, but the ecophysiological significance of this phenomenon has not been investigated at the molecular level until recently. In principle, the formation of multicellular aggregates will protect cells in the center against a harmful environment, but concomitantly, these cells can be deprived from nutritional supply. Thus, the ability to flocculate can provide a yeast population with an evolutionary advantage, but it also comes at the cost of a reduced efficiency to couple nutrient availability to growth. Indeed, it has been shown that the expression of single flocculin genes, for example FLO1, is sufficient to confer the formation of flocs that are protective to environmental stresses such as ethanol or fungicides (Smukalla et al., 2008). This social protection comes at an individual cost, because FLO1-expressing cells grow significantly slower than flo1 cells. In order to prevent nonexpressing individuals from acting as cheaters and taking advantage of the protection by aggregates without paying the cost associated with FLO1 expression, flocs formed by mixed populations are coated with flo1 cells that act as a first line of defense against harmful environments (Smukalla et al., 2008). As such, flocculin genes fulfill the definition of greenbeard genes, which direct cooperation towards other carriers of the same gene (Hamilton, 1964; Dawkins, 1976). Floc formation of S. cerevisiae also represents a social behavior (Brown & Buckling, 2008; Queller, 2008), which can be observed in other microorganisms, for example the bacteria Proteus mirabilis (Gibbs et al., 2008) or the slime mold Dictyostelium discoideum (Queller et al., 2003). The molecular mechanism underlying this behavior is cooperative binding of flocculin gene-expressing cells to each other by binary self-recognition, whereas binding of nonexpressing cells to flocs can occur only by one-way interaction via their surface oligosaccharides. The structural basis for flocculin-mediated social behavior of S. cerevisiae has been elucidated recently through the example of Flo5 (Veelders et al., 2010). This study shows that cooperativity is mainly achieved by heterophilic interactions between the Flo5 protein and specific oligomannoside structures present in the cell wall of adjacent cells. This mode of self-recognition of S. cerevisiae differs from that observed in D. discoideum, where cooperative binding is provided by a direct homophilic interaction between csA adhesin molecules present at the surface of neighboring cells (Benoit et al., 2000). However, the resulting multicellular aggregates appear to fulfill similar ecological functions in both of these microorganisms by providing a mechanism for protection to harmful environments and a strategy for long-time survival (Palkova & Vachova, 2006; Brown & Buckling, 2008; Williams, 2010).
A yeast flor is defined as an air–liquid interfacial layer of floating cells that are attached to each other and form a biofilm, which has also been named as yeast velum or flotation (Cruess, 1938; Hohl & Cruess, 1939; Freiberg & Cruess, 1955; Martinez et al., 1995; Zara et al., 2005). Similar to flocculation, flor formation has long been recognized to be of considerable importance for the production of certain alcoholic beverages, particularly sherry wines (Allan, 1939; Pretorius, 2000). At the end of alcoholic fermentation, yeast strains involved in sherry maturation rise to the surface of the wine and form a flor that is exposed to air (Martínez et al., 1997). This layer of yeast cells has access to higher amounts of oxygen and is therefore able to produce metabolites such as acetaldehyde from ethanol and to consume other nonfermentable carbon sources such as glycerol and ethyl acetate (Cortes et al., 1999; Zara et al., 2010). The resulting compounds from aerobic metabolism are essential for the characteristic aroma of flor sherry wines and include dozens of different metabolites (Berlanga et al., 2001; Moyano et al., 2002; Peinado et al., 2004). The characterization of natural yeast strains from flors of sherry wine revealed that they belong to different races of S. cerevisiae that can grow and survive in the presence of high concentrations of ethanol (over 15% v/v), but that show a huge heterogeneity with regard to the nuclear and mitochondrial genome (Martinez et al., 1995). A common feature of flor-forming yeast strains is that they are able to adapt their hydrophobicity to allow cells to aggregate (Iimura et al., 1980a, b; Martínez et al., 1997). As a result, the aggregates retain gas bubbles that originated in the respiration process and become less dense than the wine (Zara et al., 2005).
Initially, the molecular basis for cell surface hydrophobicity during flor formation has been attributed to an elevated unsaturation level and mean chain length of fatty acid residues in membranes (Iimura et al., 1980a, b; Farris et al., 1993). However, the addition of ergosterol and/or oleic acid has no effect on flor formation, whereas incubation with proteases was found to break up the velum, suggesting that the process might depend on the synthesis of hydrophobic proteins rather than lipids (Martínez et al., 1997). A classic genetic study indicated that flor formation in S. cerevisiae is under the control of a single Mendelian gene (Santa Maria & Vidal, 1973). Later on, molecular genetic analysis has shown that in many industrial strains, flor formation depends on the efficient expression of the FLO11 gene (Ishigami et al., 2004, 2006; Zara et al., 2005, 2009; Fidalgo et al., 2008; Govender et al., 2008). These studies also indicate that FLO11 confers flor formation by inducing increased cell surface hydrophobicity, which is in agreement with the finding that Flo11 is a hydrophobic cell wall protein (Reynolds & Fink, 2001; Mortensen et al., 2007; Purevdorj-Gage et al., 2007). In contrast to FLO11, the dominant expression of FLO1 or FLO5 in a laboratory strain was not sufficient to induce flor formation, although these proteins are able to significantly increase cell surface hydrophobicity (Govender et al., 2008; Van Mulders et al., 2009). This indicates that efficient flor formation might depend on factors other than hydrophobicity, which might be provided by Flo11, but not by other flocculins. Other nonflocculin proteins have also been described that confer cell surface hydrophobicity and are involved in velum formation in French wine yeast (Alexandre et al., 2000) and foam formation in sake-producing yeast (Shimoi et al., 2002). In addition, the formation of flors might be further stabilized by an extracellular matrix (ECM) of unknown composition that has been observed in certain yeast strains (Kuthan et al., 2003; Zara et al., 2009).
On solid or semi-solid agar surfaces exposed to air, S. cerevisiae can form biofilms in the form of adhesive colonies or mats (Fig. 1). In addition, adhesive colonies can adopt different morphologies (Kuthan et al., 2003; Casalone et al., 2005; Vopalenska et al., 2005; St'ovicek et al., 2010) and even build stalk-like structures (Engelberg et al., 1998; Scherz et al., 2001). A number of genetic studies have shown that agar adhesion can be mediated by FLO11, FLO10 and FIG2, but not by FLO1, FLO5 or FLO9 (Lambrechts et al., 1996a, b; Lo & Dranginis, 1998; Guo et al., 2000; Govender et al., 2008; Van Mulders et al., 2009). How exactly the Flo11, Flo10 and Fig2 proteins confer agar binding is not known. With respect to biofilm development, the clonal formation of mats from single cells on semi-solid agar is an interesting model. A growing mat will form two visually distinct populations of cells at the edge of the mat (rim) and in the interior (hub), whose development depends on the creation of concentric glucose and pH gradients in the medium (Reynolds & Fink, 2001). Based on their agar adhesion properties, these structures can be physically separated from one another. Interestingly, FLO11 is expressed in both portions of mats, indicating that the difference in adhesion between the rim and the hub might stem from reduced Flo11 activity that is caused by an elevated pH at the rim (Reynolds et al., 2008). In addition, processing of Flo11 protein might provide a fluid layer that surrounds yeast mats to enable spreading growth by sliding motility similar to biofilm-forming bacteria (Karunanithi et al., 2010). In natural environments, mat formation might be a lifestyle that is adopted by S. cerevisiae on semi-solid substrates and help the organism to more rapidly colonize preferred habitats. It should be pointed out, though, that ecologically important surface structures that are recognized by the organism to initiate biofilm formation are not yet known. It is also not clear whether S. cerevisiae biofilm formation includes the production of a protective ECM, as has been observed for bacterial biofilms. Transmission electron microscopy revealed that flocculin gene-expressing yeast strains secrete a mixture of glucose and mannose polysaccharides that surrounds the cells (Beauvais et al., 2009). Although this matrix does not seem to play a role in the resistance of flocs against drugs and ethanol, it indicates that S. cerevisiae is in principle able to produce ECM-like structures that may be important in natural habitats.
Saccharomyces cerevisiae does not belong to the classical filamentous fungi that grow in the form of true hyphae, which are defined as long and branching filamentous structures that arise by continuous tips growth of hyphal cells and subsequent fission of cells through the formation of septa (Gow, 1994; Carlile, 1995; Borkovich & Ebbole, 2010). However, diploid strains of S. cerevisiae are able to produce filaments in the form of pseudohyphae, which are defined as chains of attached, elongated cells that are formed from one another by budding (Lodder et al., 1958; Gimeno & Fink, 1992; Gimeno et al., 1992; Kron & Gow, 1995; Gow, 1997; Mösch, 2000, 2002; Gancedo, 2001). A major difference between hyphal and pseudohyphal filaments is the mode of origin and not the end-product, which, in both cases, is a mycelium (Scherr & Weaver, 1953; Shepherd, 1988). In addition, numerous studies using other ascomycetes, for example Candida albicans, have shown that further significant differences exist between hyphae and pseudohyphae including the organization of the polarisome or the regulation of cell cycle events (Sudbery et al., 2004). Therefore, S. cerevisiae ranks among the dimorphic yeasts that are able to interconvert between unicellular and filamentous growth phases (Kron, 1997; Ernst & Schmidt, 2000). In order to produce a filament, pseudohyphal cells of S. cerevisiae adhere to each other and to the substrate. In addition, linear chain formation depends on oriented cell division, which is not necessary for the formation of, for example flocs or flors. For this purpose, pseudohyphal cells maintain a unipolar budding pattern, where buds continuously emerge from the site opposite to the birth end, the distal cell pole (Chant & Herskowitz, 1991; Gimeno et al., 1992; Zahner et al., 1996; Taheri et al., 2000; Krappmann et al., 2007). Pseudohyphal cells also adapt the timing of their cell cycle to divide symmetrically and to acquire an elongated morphology by targeting the actin cytoskeleton (Kron et al., 1994; Kron, 1997). Genetically, the processes of substrate adhesion, bud site selection and cell morphogenesis, which together are required for filament formation in S. cerevisiae, can be dissected from each other (Mösch & Fink, 1997).
Adhesion to foreign surfaces is one of the key factors that enable fungal filaments to penetrate solid substrates and to grow invasively (Carlile, 1995; Borkovich & Ebbole, 2010). In S. cerevisiae, adhesion to agar enables pseudohyphal cells to act as an anchor for the cell at the apex (Gimeno et al., 1992). Anchored filaments in combination with the force that is produced by unipolar cell divisions might be the key to propel pseudohyphae or hyphae through a solid substrate. It has been shown that nonadhesive diploid yeast strains are able to produce filaments by unipolar cell division, but they remain at the surface of the substrate (Mösch & Fink, 1997). In the laboratory, agar invasion is mediated by FLO11, FLO10 and FIG2 that confer adhesion to this substrate (Lambrechts et al., 1996a, b; Lo & Dranginis, 1998; Guo et al., 2000). In the case of FLO11, filament formation well correlates with expression of the gene. Numerous conditions have been found that induce both FLO11 expression and agar invasion by filaments, for example starvation for nitrogen, glucose or amino acids (Gimeno et al., 1992; Ljungdahl et al., 1992; Cullen & Sprague, 2000). In natural environments, S. cerevisiae might therefore use substrate adhesion to penetrate solid surfaces and to explore substrates in the third dimension (Palkova & Vachova, 2006; Vopalenska et al., 2010). However, neither the molecular structure of natural substrate surfaces nor their interaction with different members of the S. cerevisiae adhesin family is known so far.
Finally, the study of S. cerevisiae filament formation has medical relevance. Similar to human pathogenic yeasts, pseudohyphal cells of S. cerevisiae not only become adhesive, they also secrete enzymes such as hydrolases and proteinases that support invasion of the substrate (Odds, 1994; Madhani et al., 1999; Murphy & Kavanagh, 1999; Kaur et al., 2005; Zhu & Filler, 2010). The precise molecular mechanisms for substrate adhesion and invasion are likely to differ between S. cerevisiae and human pathogenic yeasts. However, central signaling pathways that control dimorphism in S. cerevisiae have also turned out to be relevant for filament formation in pathogenic yeasts, for example in C. albicans (Lo et al., 1997). In fact, investigation of dimorphism in S. cerevisiae has been instrumental to uncover many of the signaling routes that control hyphal growth and virulence in a growing number of human pathogenic fungi (Lengeler et al., 2000; Bahn et al., 2007).
Structure and function of S. cerevisiae adhesins
The vegetative adhesins of S. cerevisiae belong to the large family of fungal glycosylphosphatidylinositol-linked cell-wall glycoproteins (GPI-CWPs) (Verstrepen et al., 2004; Verstrepen & Klis, 2006; Dranginis et al., 2007; Linder & Gustafsson, 2008). These secreted proteins confer unique adhesion properties and their primary amino acid sequence often shares only a low degree of similarity. However, GPI-CWPs have a common overall architecture and consist of at least three different domains (Fig. 3). The N-terminal or A domain that follows the secretion signal is exposed at the cell surface and confers the recognition and binding of ligand molecules presented in trans. This domain is followed by a segment of variable length (middle or B domain) that is extremely rich in serine and threonine residues and that is highly glycosylated. At the carboxy-terminal region (or C domain), a GPI anchor is added for localizing adhesins to the cell wall. Of the proteins known to confer vegetative adhesion in S. cerevisiae, the flocculins Flo1, Flo5, Flo9, Flo10 and Flo11 fulfill these structural criteria (Dranginis et al., 2007; Linder & Gustafsson, 2008). In contrast, Fig2 and Aga1 do not fully match the general architecture of GPI-CWPs, with the most obvious difference of lacking a clearly defined A domain (Verstrepen et al., 2004).
With respect to understand the precise function of Flo proteins during self-recognition and cell–cell adhesion, long-standing issues have been the identification of domains and residues that confer carbohydrate binding and discrimination between closely related hexoses. Carbohydrate competition studies with strains expressing FLO1, FLO5, FLO9 or FLO10 and different hexoses have suggested that the encoded Flo proteins bind to mannose-containing carbohydrate structures present at the cell surface of neighboring cells (Stratford, 1989a, b; Stratford & Assinder, 1991; Van Mulders et al., 2009; Veelders et al., 2010). Specific variants of FLO1 such as Lg-FLO1 were found to confer not only mannose-, but also glucose-inhibitable flocculation (Kobayashi et al., 1998; Ogata et al., 2008). These studies uncovered a conserved VSWGT motif encompassing residues 226–230 of Flo1 to be involved in carbohydrate binding and pointed towards the tryptophan residue at position 228 to be participating in discriminating between mannose and glucose. Notably, the VSWGT motif can be structurally and functionally related to the EYDGA motif, which is found in the N-terminal PA14 domain of the epithelial adhesin Epa1 of the human pathogenic yeast Candida glabrata and is involved in sugar recognition (Zupancic et al., 2008; Goossens & Willaert, 2010). Furthermore, the molecular role of Ca2+ has remained unclear. It has been proposed that Ca2+ is either involved in the formation of semi-rigid rod-like superstructures of the heavily O-glycosylated B region of Flo proteins (Jentoft, 1990) or plays a crucial role in carbohydrate recognition as exemplified by classical C-type lectins (Miki et al., 1982a, b; Stratford, 1989a, b; Kuriyama et al., 1991).
The first high-resolution insight into Flo proteins was recently provided by solving the three-dimensional structure of the Flo5A domain complexed to its cognate ligands derived from yeast high-mannose oligosaccharides at resolutions of up to 0.95 Å (Veelders et al., 2010). This study has provided answers to many of the long-standing questions regarding domain organization, the role of Ca2+ and ligand binding/discrimination. The overall structure of Flo5A shows a bipartite organization comprising a large β-sandwich domain that is topologically related to the PA14 domain and an additional insertion, the Flo5 subdomain (Fig. 4). This Flo-specific subdomain (N84-A120) consists of five short β-strands that are stabilized by two disulfide bridges. A second region unique for Flo5A is formed by its N- and C-termini, which extend as an L-shaped stretch from the core domain and are fixed to it by disulfide bridges to two consecutive cysteines. This loop (G168-T189) is considerably longer than in the PA14 domain and, together with a second loop (Y46-T68), seals the surface of the underlying β-sheet from solvent access.
The structural and biochemical analysis of the Flo5A domain revealed that Ca2+ is directly involved in carbohydrate binding (Veelders et al., 2010). The crystal structure of the mannose/Ca2+ complex of Flo5A shows a C-lectin-type mode of carbohydrate binding via Ca2+-mediated recognition of the 2′- and 3′-hydroxyl groups, where the Ca2+ ion is bound between the sugar and the protein (Fig. 4). The Ca2+-binding site of Flo5A is a unique and characteristic hallmark of the PA14/Flo5-like protein family. The Ca2+ ligands belong to the carbohydrate-binding loops, CBL1 and CBL2. CBL1 is unusual by bearing a rare cis-peptide between the nonproline residues D160 and D161. So far, a DcisD motif has only been found in the nucleotide-binding site of the ATP synthase (Abrahams et al., 1994) as well as in a Zn2+-dependent aminoprotease (Chevrier et al., 1994). This motif is present throughout the PA14/Flo5-like family, but is missing in the PA14 domain itself. The crucial role of the unique DcisD motif in the PA14/Flo5-like family of Ca2+- and carbohydrate-dependent adhesins is reflected by a complete lack of flocculation, when either of the two D residues is mutated (Veelders et al., 2010). Multiple sequence alignment shows that 45% of the known PA14 domains, including human proteins like galactosyltransferases and fibrocystin, contain a double D at the equivalent position (Sonnhammer et al., 1997).
The study with Flo5A also suggests that terminal arms of yeast high-mannose oligosaccharides act as cognate ligands in vivo. The yeast cell wall has a complex architecture (Klis et al., 2006; Lesage & Bussey, 2006) and contains a large proportion of mannoproteins with either N- or O-linked oligosaccharides (Lehle et al., 2006; Lommel & Strahl, 2009). These carbohydrates have long been discussed to be involved in flocculation (Stratford, 1992), but the exact structures of the ligands that are recognized by flocculins were not known. Fluorescence-titration and crystal soaking experiments revealed that only α1,2-linked mannobiosides, but neither α1,3- nor α1,6-linked mannoses, are bound. Moreover, α1,2-linkage to a second mannose increased the affinity ninefold (KD=3.5 mM) when compared with mannose. The recognition of the second mannose moiety is achieved by hydrogen bonds between its 3′-hydroxyl group and S227 from CBL2 and, in addition, Q117 residing in the Flo5 subdomain (Fig. 4). Interestingly, mutation of S227 to alanine was found to increase the affinity towards mannose without affecting specificity. α1,2-α1,2-Mannotriose shows an affinity and binding mode that is comparable to α1,2-mannobiose. The same is true for a synthetic mannopentaose, which mimics a yeast N-linked core oligosaccharide (Fig. 4). Apparently, the Flo5 subdomain supports discrimination between the branches of the core oligosaccharide by steric hindrance or the contribution of a hydrogen bond between the fourth mannoside and residues N104.
The high-resolution view of Flo5A allows to explain the behavior of NewFlo-type flocculins with broadened sensitivity against different sugars like glucose (Stratford & Assinder, 1991). Initial work suggested that NewFlo-type variants like D202T and W228L are directly affected in carbohydrate binding (Kobayashi et al., 1998). However, the Flo5A structure reveals that these residues are not directly involved in carbohydrate recognition and indirectly affect the structure of the primary binding site. Lg-Flo1, which also promotes NewFlo-type flocculation, lacks the complete Flo5A subdomain and therefore exhibits relaxed carbohydrate recognition and less efficient sugar binding. Here, the Flo5A structure allows to predict and engineer novel NewFlo-type flocculins with unaltered flocculation efficiency. A prominent example is the Q98A variant, for which a loss of interaction with the axial 2-hydroxyl group of mannose and therefore a reduced ligand specificity can be predicted. Indeed, the Q98A variant confers unaltered flocculation efficiency, but completely lacks discrimination between mannose and glucose.
Finally, the crystal structure of Flo5A provides a possible answer to a problem that is posed by the fact that high concentrations of mannose ligands are not only present on the surface of neighboring cells (trans position), but also on the site of the flocculin (cis position). An interesting question is how flocculins preferentially interact with ligands presented in trans without being compromised by cis ligand interactions, which would prevent cell–cell adhesion. Interestingly, a secondary carbohydrate-binding site is observed at the back of the Flo5A domain, although molar concentrations of mannose or glucose are required and binding is Ca2+ independent (Veelders et al., 2010). It is therefore possible that the secondary carbohydrate-binding site may help to fix the A domain at the cell surface by interacting with cis ligands and thereby allow the primary binding site to preferentially bind to trans ligands (Fig. 5). A similar mechanism is observed in the case of certain cell-adhesion molecules of higher eukaryotes, for example the human neural cell adhesion molecule (NCAM), where secondary binding with adhesion molecules in cis ensures efficient primary binding with the adhesin molecules present in trans (Kiselyov et al., 2005).
Flo1, Flo9 and Flo10
The Flo5A structure allows to model the structure of the A domains of the closely related Flo1 (94% identical), Flo9 (89% identical) and Flo10 (66% identical) proteins and to predict functional characteristics. In the case of Flo1A, all residues that are crucial for Ca2+ and ligand binding in Flo5A are conserved, suggesting identical functions (Watari et al., 1994; Groes et al., 2002; Veelders et al., 2010). An interesting difference is observed in Flo9A, which carries an alanine at position 227 instead of a serine. This suggests that Flo9A might confer enhanced floc formation as has been observed for the S227A variant of Flo5A. The most striking differences can be found in Flo10A, which has an extra insert in the Flo-specific subdomain and carries aspartate residues at positions 98 and 104. However, the Ca2+-binding site is conserved in Flo10A, and computer modeling predicts a similar overall structure when compared with Flo5A. This suggests that Flo10A has an altered carbohydrate-binding spectrum when compared with Flo1A and Flo5A, which might explain its special adhesion properties in vivo. Strikingly, FLO10 not only confers flocculation when overexpressed, but also agar and plastic adhesion (Guo et al., 2000; Van Mulders et al., 2009). It will be interesting to determine as to which of these functions are mediated by the Flo10A domain.
In contrast to Flo5 and its close relatives, the exact structure and molecular function of the Flo11 N-terminal domain is not known. The Flo11 A domain is conserved and can be found in related species within the Saccharomycotina as well as in Schizosaccharomyces pombe (Linder & Gustafsson, 2008). Because Flo11 confers hydrophobic properties to yeast cells, it has been suggested that this flocculin functions similar to fungal hydrophobins (Reynolds & Fink, 2001; Ishigami et al., 2006). However, whether hydrophobicity is conferred by the A domain of Flo11 has not been investigated. Interestingly, FLO11 has been found to mediate strain-specific phenotypes. In strains of S. cerevisiae var. diastaticus, FLO11 expression confers Ca2+-dependent and mannose-sensitive flocculation, but no agar adhesion or filament formation (Bayly et al., 2005; Douglas et al., 2007). In contrast, strains of the Σ1278b genetic background are nonflocculent, although they express FLO11 at high levels (Guo et al., 2000). Instead, these strains require FLO11 for agar adhesion and filamentation (Lo & Dranginis, 1998; Rupp et al., 1999), and genetic analysis suggests that adhesive functions are conferred by the A domain (Veelders et al., 2010). Biochemical studies with secreted Flo11 proteins from the two strains attached to microscopic beads have shown that it confers the ability to bind to S. cerevisiae var. diastaticus cells, but not to Σ1278b cells (Douglas et al., 2007). Moreover, this function of Flo11 is mannose sensitive and depends on the expression of FLO11 in the S. cerevisiae var. diastaticus target cells. These findings suggest homotypic adhesive mechanisms that might involve the binding of Flo11 to mannose residues presented by other Flo11 molecules in trans. It has been suggested that these differences in Flo11 functions might arise from strain-specific mannosylation of the adhesin (Bayly et al., 2005; Douglas et al., 2007). However, functional variance might also be attributed to the structural differences observed in the two A domains. Interestingly, Flo11A from Σ1278b strains carries an insert of an extra 15 amino acids when compared with S. cerevisiae var. diastaticus and lacks two of the highly conserved residues of the Flo11 family. Thus, Flo11A might have diverged in Σ1278b strains to recognize new ligands and to confer alternate adhesion properties. In the future, high-resolution structure–function analysis of different Flo11A variants will be required to answer these questions.
In contrast to the N-terminal domains, much less is known about the structure and function of the middle parts of S. cerevisiae adhesins. Bioinformatic analysis reveals that they have a length of several hundred amino acids and are highly enriched in threonine and serine residues predicting extensive N- and O-linked glycosylation (Lehle et al., 2006; Lommel & Strahl, 2009). In addition, middle domains are often composed of conserved tandem repeats of variable length (Verstrepen et al., 2004; Dranginis et al., 2007; Linder & Gustafsson, 2008) and they contain sequences enriched for β-branched aliphatic amino acids, indicating a high potential for β-aggregation and amyloid formation (Ramsook et al., 2010). Although the functions of these diverse structural features are not yet understood in full detail, a number of studies provide initial insights.
Middle domains appear to play an important role in properly presenting the N-terminal A domains at the cell surface. In the case of Flo1 and Flo11, it has been shown that the length of the middle domain directly correlates with the functionality of the proteins during flocculation and biofilm formation (Verstrepen et al., 2005; Zara et al., 2009). These findings are in agreement with studies on Candida adhesins (Frieman et al., 2002; Frieman & Cormack, 2004) and they show that longer middle domains usually enable more efficient adhesion, while smaller domains confer less avid adherence. This suggests that longer middle parts are required to expose A domains to neighboring cells during flocculation or to foreign surfaces during biofilm formation, while shorter parts cause A domains to remain buried in the cell wall. A further interesting twist comes from the observation that tandem repeats allow for rapid changes in repeat numbers by triggering frequent recombination events, causing a variation in the size and composition of middle domains (Verstrepen et al., 2005; Fidalgo et al., 2006, 2008). This mechanism might enable fungal populations to rapidly adapt adhesion properties to the environment (Levdansky et al., 2007; Verstrepen & Fink, 2009).
Vegetative adhesins of S. cerevisiae contain a large number of residues susceptible to N- and O-linked glycosylation. It has been shown that Flo1 and Flo11 are heavily glycosylated, although their exact glycan profile has not yet been determined (Straver et al., 1994; Bony et al., 1997; Douglas et al., 2007; Karunanithi et al., 2010). The functional role of this glycosylation for adhesion is not clear. It has been suggested that O-linked oligosaccharide side-chains enable the middle domains to obtain a rod-like structure that is further stabilized by Ca2+ ions (Jentoft, 1990). The resulting semi-rigid stalks may help to project the protein through the exterior of the cell wall and enable the A domains to interact efficiently with ligands (Frieman et al., 2002; Verstrepen & Klis, 2006). In addition, the glycans provided by the middle domains of adhesin might affect the hydrophobicity. In C. albicans, for example, this correlation between the cell wall mannosylation and cell surface hydrophobicity has been demonstrated (Masuoka & Hazen, 1997, 2004). In S. cerevisiae, hydrophobicity has initially been linked to the expression of FLO11 (Reynolds & Fink, 2001). However, a more recent study shows that strains expressing FLO1, FLO5, FLO9 or FLO10 also become highly hydrophobic (Van Mulders et al., 2009). This implicates that hydrophobicity is conferred by glycan structures, because the amino acid composition of adhesin middle domains can be highly divergent. Similarly, surface adhesion observed for strains that produce adhesins without an obvious N-terminal A domain, for example Fig2, might be conferred by oligosaccharides (Guo et al., 2000).
Bioinformatic screening has uncovered that a number of S. cerevisiae adhesins contain one or more sequences with a high β-aggregation potential including Flo1, Flo11, Aga1, Fig2 and adhesins from C. albicans (Otoo et al., 2008; Frank et al., 2010; Ramsook et al., 2010). Functional studies show that Flo1 and Flo11 proteins form amyloids in vivo and in vitro and that cell–cell aggregation mediated by these adhesins is sensitive to amyloid-binding dyes. This suggests that amyloid formation of β-aggregation-prone sequences in the middle domains leads to the generation of adhesin multimers. These structures might affect the functionality of the N-terminal A domains by increasing the avidity or the potential for ligand binding in trans (Fig. 5).
To understand how individual yeast cells and whole populations adapt their surface properties in response to the environment, it is essential to understand the mechanisms underlying the regulation of adhesin genes and proteins. More specifically, the signaling pathways and transcription factors that target the different promoter regions must be uncovered as well as the post-transcriptional and post-translational mechanisms that control adhesin production and function. In addition, the integration of regulatory responses must be studied under diverse environmental conditions to understand how individual cells control adhesin production and how phenotypic heterogeneity is achieved in yeast populations. A large number of studies have addressed the control of FLO11 expression and revealed a highly complex promoter structure and regulatory pattern for this gene (Rupp et al., 1999; Mösch, 2000, 2002; Gancedo, 2001; Gagiano et al., 2002; Verstrepen & Klis, 2006; Zaman et al., 2008). In contrast, the regulation of FLO1, FLO5, FLO9 and FLO10 is less well understood, especially at the molecular level (Teunissen & Steensma, 1995; Verstrepen et al., 2003). This situation might stem from the fact that these FLO genes are not active in many laboratory strains, for example of the S288c or the Σ1278b genetic background (Liu et al., 1996; Halme et al., 2004). Finally, it is important to understand how S. cerevisiae can adapt adhesin gene structure on a longer evolutionary time scale.
The FLO11 promoter spans more than 3 kb and contains many upstream activation sequences (UASs) and elements for repression (Lo & Dranginis, 1996; Rupp et al., 1999). Numerous genetic studies in combination with genome-wide transcription factor binding analysis have revealed detailed insights into the topology of regulatory pathways at the promoter and its responsiveness to many external and internal signals (summarized in Fig. 6). In addition, single-cell analysis and mathematical modeling have provided a first picture of population-level heterogeneity of FLO11 expression through the integration of conventional and epigenetic regulatory mechanisms (Halme et al., 2004; Vinod et al., 2008; Octavio et al., 2009).
Fus3/Kss1 mitogen-activated protein kinase (MAPK) cascade
Under conditions of ample nutritional supply, cell-surface adhesion and FLO11 expression depend on specific elements of the Cdc42-regulated Fus3/Kss1 MAPK cascade. This signaling module is highly conserved in all eukaryotes and has initially been identified to control sexual mating of haploid yeast cells (Bardwell, 2005; Chen & Thorner, 2007). Execution of vegetative adhesion and mating is under the control of shared and program-specific components. The shared components include the Rho-type GTPase Cdc42, the scaffold protein Ste50, the p21-activated protein kinase (PAK) Ste20, the MAPK kinase kinase (MAPKKK) Ste11, the MAPK kinase (MAPKK) Ste7, the MAPKs Fus3 and Kss1 as well as the transcription factor Ste12 (Liu et al., 1993; Roberts & Fink, 1994; Mösch et al., 1996, 2001; Cook et al., 1997; Madhani et al., 1997; Mösch & Fink, 1997; Roberts et al., 1997; Ramezani Rad et al., 1998; Rupp et al., 1999). These elements control the expression of FLO11 during vegetative growth and they are required for the activation of mating genes in response to peptide pheromones. The mating-specific components include the pheromone receptors, the three subunits for the heterotrimeric G protein and the scaffold protein Ste5, all of which are not required for FLO11-dependent adhesion during vegetative growth (Liu et al., 1993; Roberts & Fink, 1994).
Glucose is generally accepted to activate PKA and FLO11-mediated adhesion via Ras2 and Gpa2, but the exact means by which this environmental signal stimulates the G proteins are still elusive (Gancedo, 2008; Zaman et al., 2008). In addition, the secreted aromatic alcohols tryptophol and phenylethanol stimulate FLO11 expression through a Tpk2-dependent mechanism, suggesting that PKA might be part of a quorum-signaling pathway that links environmental sensing to adhesion (Chen & Fink, 2006; Wuster & Babu, 2010). A further signal that controls PKA is ammonium, which acts via Mep2, a low-affinity ammonium permease required for filamentation (Lorenz & Heitman, 1998a, b; Van Nuland et al., 2006). This indicates that PKA might affect the stimulation of FLO11 in response to nitrogen starvation. Finally, it has been shown that the PKA pathway is connected to the Fus3/Kss1 MAPK cascade, which enables cross-pathway control towards FLO11 (Mösch et al., 1996, 1999; Vinod & Venkatesh, 2007; Chavel et al., 2010).
Amino acid starvation has been demonstrated to activate adhesive growth and FLO11 in the presence of glucose and ammonium (Braus et al., 2003; Kleinschmidt et al., 2005). This effect depends on the sensor kinase Gcn2 and the transcription factor Gcn4, which are central elements of the general amino acid control (GAAC) system (Hinnebusch, 2005). However, Gcn4 seems to regulate FLO11 indirectly by controlling an as yet unknown signaling pathway that confers derepression of FLO11 in response to amino acid starvation (Braus et al., 2003; Valerius et al., 2007). A further connection between amino acid availability and FLO11 comes from the observation that strains mutated for SSY1 or PTR3, which encode elements of the SPS amino acid-sensing pathway, are hyperadhesive (Klasson et al., 1999; Ljungdahl, 2009). However, no connection has been made between downstream regulators of the SPS system, for example the transcription factors Stp1 and Stp2 (Andreasson & Ljungdahl, 2004), and FLO11. Finally, exposure of yeast cells to the plant hormone indole acetic acid (IAA) stimulates the expression of FLO11 by involving members of the Avt family of IAA and amino acid transporters and the transcription factor Yap1 (Prusty et al., 2004). This suggests that S. cerevisiae seems to be able to adapt its adhesion properties in response to plant hormone and nutritional signals, but the exact signaling routes leading to FLO11 remain to be elucidated.
Transcription factors Mss11, Msn1, Mga1, Rme1 and Haa1
FLO11 expression is under further control of several transcription factors including Mss11, Msn1, Mga1, Rme1 and Haa1, which operate through not well-understood mechanisms. MSS11 and MSN1 were initially identified as activators of FLO11 and filamentation, but also as multicopy suppressors of mep2 and ras2 mutations (Gimeno & Fink, 1994; Lambrechts et al., 1996a, b; Vivier et al., 1997; Gagiano et al., 1999). Later studies showed that Mss11 is required for the activation of FLO11 by many other regulators including Tec1, Flo8, Phd1, Nrg1, Nrg2, Sok2 and Sfl1 (Gagiano et al., 2003; van Dyk et al., 2005). How exactly Mss11 operates at the FLO11 promoter is not clear, but its function depends on a conserved glutamine-rich activation domain and may involve physical association with Flo8 (Kim et al., 2004a, b). MSN1 (also known as PHD2 or MSS10) was originally isolated as a multicopy suppressor of an snf1 mutation, indicating that this regulator might function downstream of the Snf1 kinase (Estruch & Carlson, 1990). It was also suggested that Msn1 may act at longer distances to destabilize chromatin (Sidorova & Breeden, 1999). A further regulator of FLO11 is Mga1, a protein with similarity to heat shock transcription factors. The MGA1 gene was originally isolated as a multicopy suppressor of pseudohyphal growth defects of mep2 mutants (Lorenz & Heitman, 1998a, b). Genome-wide location analysis demonstrated that Mga1 – along with Ste12, Tec1, Sok2, Phd1 and Flo8 – binds to the FLO11 promoter in vivo (Borneman et al., 2006). This study also revealed that the promoters of MGA1 and PHD1 were bound by all of these transcription factors, identifying them as master regulators of FLO11 and as key target hubs in the complex regulatory network for adhesion and filamentation. The transcriptional regulator Rme1, a DNA-binding protein that represses entry into meiosis, was also identified to positively regulate FLO11 and adhesion (van Dyk et al., 2003). The expression of RME1 is under direct control of the a1-α2 repressor and is therefore efficiently expressed in haploids, but not in diploids (Covitz et al., 1991). This might at least in part explain why FLO11 expression is significantly higher in haploid than in diploid strains (Rupp et al., 1999). However, Rme1 does not seem to be involved in the ploidy-specific expression of FLO11, which is conferred by as yet unknown mechanisms (Galitski et al., 1999). Finally, FLO11 has recently been found to be under control of the transcription factor Haa1, which is also required for adaptation of yeast to acidic stress (Keller et al., 2001; Aranda & del Olmo, 2004; Fernandes et al., 2005). HAA1 is required for FLO11 expression, but only under acid stress conditions, and it acts as a multicopy suppressor of a yak1 mutation (Malcher et al., 2011). Whether Haa1 is a direct target of the Yak1 kinase and how Haa1 controls the FLO11 promoter is not known.
In addition to the many conventional regulatory mechanisms that target the FLO11 promoter, the expression of FLO11 is also under epigenetic control. In response to nitrogen starvation, diploid yeast strains express FLO11 and initiate filamentation in a highly heterogeneous manner (Halme et al., 2004; Vinod et al., 2008). The high degree of cell-to-cell variation observed is caused by metastable and inheritable silencing of the FLO11 promoter and is based on both promoter and genomic positional information. At least two distinct mechanisms contribute to epigenetic control of FLO11. One mechanism depends on the regulator Sfl1 and the histone deacetylase (HDAC) Hda1, which is part of a complex implicated in transcriptional control (Rundlett et al., 1996; Vogelauer et al., 2000; Lee et al., 2009). In the case of FLO11, Sfl1 has been suggested to provide promoter specificity for silencing by direct binding (Halme et al., 2004). Hda1 is likely to integrate both genome positional information and promoter-specific signals that stem from Sfl1, which could recruit Hda1 via the Tup1/Ssn6 corepressor (Conlan & Tzamarias, 2001). The nature of the positional determinants, however, remains unclear. A second mechanism involves a pair of cis-interfering noncoding RNAs (ncRNAs) and Rpd3L, another HDAC complex involved in transcriptional regulation (Yang & Seto, 2008; Bumgarner et al., 2009). Interestingly, Rpd3L is an activator of FLO11, which is unexpected, given the fact that the complex is present at the FLO11 promoter (Barrales et al., 2008; Bumgarner et al., 2009). However, Rpd3L seems to act via chromatin condensation at the FLO11 promoter at an upstream site that includes the binding sites for Sfl1 and Flo8. This event blocks the access of Sfl1, but promotes Flo8 binding and enables the expression of the ncRNA PWR1, whose transcription is initiated upstream of and in a direction opposite to FLO11. As a consequence, FLO11 expression is promoted because the PWR1 ncRNA interferes with the transcription of ICR1, a second ncRNA. The expression of ICR1 is promoted by Sfl1 binding and represses FLO11 in cis by a ‘promoter occlusion’ mechanism (Martens et al., 2004, 2005; Bumgarner et al., 2009). Because transcription of ICR1 occurs in the same direction as FLO11 and covers most of the FLO11 promoter region, it probably blocks access of the FLO11 core promoter to general transcription factors and to chromatin remodelers required for nucleosome rejection (Martens & Winston, 2003). Additional chromatin remodeling at the FLO11 promoter might be provided by the SWI/SNF complex, a multisubunit DNA-dependent ATPase that regulates transcription by altering chromatin structure and that is involved in FLO11 activation (Peterson & Workman, 2000; Barrales et al., 2008).
Epigenetic mechanisms provide a circuitry for toggling between a silenced and a transcriptional competent state (Bumgarner et al., 2009). In the case of FLO11, they contribute to the observed variegated and bistable expression and the resulting variation of cell surface properties within a yeast population. The observed cis-acting mechanisms are likely to contribute to the finding that two copies of the FLO11 locus present within a single cell switch between a silenced and a competent promoter state in a random and independent manner (Octavio et al., 2009). Epigenetic mechanisms help to classify conventional trans-acting activators of FLO11 expression, based on their ability to stabilize the competent state. It has been shown that Tec1, Ste12 and Phd1 belong to a class of activators, which only weakly stabilize the competent state and cannot effectively activate transcription at a silenced promoter (Octavio et al., 2009). These class I activators regulate fast promoter fluctuations and destabilize the competent state to increase the burst frequency (Bar-Even et al., 2006; Newman et al., 2006). In contrast, Flo8 belongs to the class II of activators that primarily regulate slow promoter fluctuations and stabilize the competent state. At medium levels, Flo8 ‘opens’ the silenced promoter state and enables activation by class I activators. At high levels, Flo8 fully disrupts silencing to induce a homogenous high-level expression of FLO11 within yeast populations.
Transcriptional elongation and post-transcriptional regulation
In addition to the plethora of control mechanisms that act on the promoter of FLO11, the expression of the gene is regulated during transcription elongation and at the post-transcriptional level. It has been found that FLO11 transcription elongation is hindered in the region of the tandem repeats in tho mutants (Voynov et al., 2006). In S. cerevisiae, the multisubunit THO complex has been identified as a possible elongation complex, which is recruited to actively transcribed genes and is involved in the cotranscriptional formation of messenger ribonucleoparticles (mRNP) that are competent to be exported from the nucleus (Jimeno et al., 2002; Strasser et al., 2002; Jimeno & Aguilera, 2010). In the case of FLO11 (and also FLO1), regulation by THO depends on the internal repeats and is partially uncoupled when TOP1 encoding topoisomerase is overexpressed (Voynov et al., 2006). Interestingly, mutations that reduce THO or topoisomerase activity induce the formation of DNA : RNA hybrids (R loops) (Masse et al., 1997; Huertas & Aguilera, 2003; Drolet, 2006). This indicates that the THO complex might help to overcome such inhibitory hybrid structures and thereby allow efficient transcription elongation at internal repeats.
Several studies have reported post-transcriptional control mechanisms for FLO11 expression. Comparison of FLO11 transcript levels with the activity of a coexpressed FLO11-lacZ reporter gene suggests that translation of the FLO11 mRNA is upregulated in response to amino acid starvation and that efficient translation and adhesion requires the ribosomal protein Rps26 at levels that are sufficient for viability (Strittmatter et al., 2006; Fischer et al., 2008). Thus, yeast cells seem to be able to adapt adhesion properties by adjusting FLO11 translation efficiency. A further post-transcriptional mechanism for FLO11 expression involves the RNA-binding protein Khd1, which regulates the asymmetric expression of FLO11 indirectly via the transcriptional regulator Ash1, which is required for the expression of FLO11 and filamentation (Chandarlapaty & Errede, 1998; Pan & Heitman, 2000; Wolf et al., 2010). In addition, Kdh1 directly binds to the FLO11 mRNA and inhibits its translation (Wolf et al., 2010). This regulation allows changes in FLO11 expression between mother and daughter cells to establish the asymmetry that is required for the transition between yeast form and filamentous growth. Finally, Flo11 also seems to be regulated at the post-translational level. It has been shown that Flo11 is shed from cells and involves the protease Kex2, which is required for the cleavage and maturation of the Flo11 protein (Karunanithi et al., 2010). This study also found that Flo11 shedding contributes to the overall balance in adherence properties that is optimal for filamentation and mat formation. Moreover, shed Flo11 is a component of a fluid layer that surrounds mats, which may have functions analogous to the mucus secretions of higher eukaryotes. Whether and how shedding is regulated in response to environmental changes remains to be investigated.
FLO1, FLO5, FLO9 and FLO10
When compared with FLO11, less information is available on the regulation of other FLO genes. In industrial strains, FLO gene expression has been found to be regulated by a number of environmental stimuli (Bauer et al., 2010; Soares, 2010). In brewing strains for instance, FLO1 is regulated by carbon and/or nitrogen starvation (Sampermans et al., 2005), pH (Soares & Seynaeve, 2000) and ionic strength (Jin & Speers, 2000). However, detailed analysis of the regulatory pathways and control mechanisms in production strains has been hampered by their limited experimental tractability.
In Σ1278b laboratory strains, FLO gene expression patterns differ from the S288c genetic background. Under standard growth conditions, Σ1278b FLO1, FLO5, FLO9 and FLO10 are epigenetically silenced (Guo et al., 2000; Halme et al., 2004; Reynolds et al., 2008). FLO1 expression is further impaired due to the absence of an efficient Flo8-binding site in the promoter (Fichtner et al., 2007). In Σ1278b, FLO1 has also been found to be regulated during transcription elongation by the THO complex (Voynov et al., 2006). FLO10 expression in Σ1278b was found to be derepressed by the deletion of SFL1 or in the absence of either Ira1 or Ira2, two Ras2 GTPase-activating proteins that negatively control the cAMP-PKA pathway (Halme et al., 2004). The same study also uncovered that FLO10 is regulated by the same transcription factors that control FLO11, Sfl1 and Flo8, but is silenced by a distinct set of histone deacetylases, Hst1 and Hst2. Further epigenetic control is inferred on FLO10 by a mechanism that involves a pair of cis-interfering ncRNAs (Bumgarner et al., 2009). Similar to FLO11, these transcripts are encoded upstream of FLO10 and are likely to control gene expression by a promoter occlusion mechanism. Finally, none of the mutations that derepress FLO10 in Σ1278b are sufficient to induce significant expression of FLO5 or FLO9, indicating that they are regulated by distinct mechanisms. Genome-wide transcription factor-binding studies in Σ1278b strains have revealed that, for example the FLO9 promoter is bound by diverse regulators including Phd1, Sok2 or Nrg1 (Harbison et al., 2004). However, whether these regulators indeed control FLO9 remains to be investigated.
AGA1 and FIG2
AGA1 and FIG2 play well-established roles in sexual agglutination and the expression of both genes is highly induced during mating (Roy et al., 1991; Cappellaro et al., 1994; Erdman et al., 1998; Zhao et al., 2001; Jue & Lipke, 2002; Zhang et al., 2002). Accordingly, AGA1 and FIG2 are under control of the Fus3/Kss1 MAPK cascade and their promoters are bound by the transcription factor Ste12 in vivo (Oehlen et al., 1996; Zeitlinger et al., 2003; Harbison et al., 2004). In addition, both of these genes also confer vegetative adhesion. In Σ1278b flo11 mutant strains for instance, the expression of FIG2 from the strong GAL1 promoter is sufficient to induce agar adhesion and filamentation (Guo et al., 2000). It has therefore been suggested that successful mating not only depends on sexual adhesion of partner cells, but might also require a more complex response before agglutination (Erdman & Snyder, 2001). Such an early response could be induced by low levels of pheromone and enhance the ability of cells to search for mating partners by directed surface growth, which requires vegetative adhesion by Fig2 and Aga1. Further stimuli and regulatory pathways, which in principle could exist to control AGA1 and FIG2, have not yet been discovered.
Saccharomyces cerevisiae as a model for fungal adhesion
The formation of multicellular aggregates by S. cerevisiae has been observed already in the 19th century, but a precise picture of the molecular mechanisms that underlie yeast adhesion has only recently become available. In this review, we have summarized a substantial number of publications that provide first detailed insights into the structure and function of adhesins and reveal a highly complex regulatory network for adhesion in S. cerevisiae. These studies have further allowed to uncover new concepts in gene evolution, to investigate the interplay of conventional and epigenetic mechanisms in gene regulation and to analyze the importance of self-recognition for cooperative cell interaction and social behavior of microorganisms. Thus, cell adhesion of S. cerevisiae is a valuable system to study not only fundamental cellular processes, but also the general principles of microbial development.
In pathogenic fungi, analysis of adhesin function and regulation is often hampered by their elaborate or limited experimental tractability. Here, S. cerevisiae has proven to be a valuable model for the characterization of adhesins from other fungi. A prominent example is the functional analysis of Epa adhesins from C. glabrata in S. cerevisiae cells to study domain structures and ligand-binding specificity (Cormack et al., 1999; Frieman et al., 2002; Zupancic et al., 2008). Similar strategies might be used for studying adhesin molecules from other pathogens. Furthermore, several lessons learned from studying the regulatory network for adhesion in S. cerevisiae have turned out to be true for other fungi (Lengeler et al., 2000). For instance, the finding that the cAMP-PKA and MAPK pathways control S. cerevisiae adhesion has allowed to directly test whether the same is the case in human pathogenic fungi. Indeed, both pathways also act in parallel in C. albicans to control adhesion and virulence (Liu et al., 1994; Lo et al., 1997; Feng et al., 1999). In addition, many of the transcription factors that control FLO11 in S. cerevisiae are related to regulators of adhesion and virulence in fungal pathogens including Tec1, Ste12, Phd1, Sok2 and Mss1 (Liu et al., 1994; Stoldt et al., 1997; Schweizer et al., 2000; Su et al., 2009). Finally, high-resolution structural analysis of S. cerevisiae adhesins allows to produce more precise models for related proteins from pathogenic fungi, for example PA14 domain-related adhesins (de Groot & Klis, 2008; Veelders et al., 2010). These findings are promising indicators that future knowledge gained by studying the S. cerevisiae adhesion model will have an important impact on the research directed in other fungi.
Owing to the fact that we have only limited information on the true ecological significance of S. cerevisiae adhesion, future research in this field must include a detailed analysis of a large number of adhesin genes and proteins and it must also aim at the establishment of more natural test systems. More specifically, such studies must include high-resolution structure and function analysis of many more yeast adhesins, for example members of the Flo10 and Flo11 subfamily, with a clear focus on uncovering the precise mechanisms for ligand recognition and discrimination. A helpful step towards this goal will be the comparative analysis of adhesin genes in a large number of diverse laboratory and industrial S. cerevisiae strains. In addition, such studies should include a wide variety of Saccharomyces sensu stricto strains isolated from diverse natural environments or clinical origin (Fay et al., 2004; Carreto et al., 2008; Klingberg et al., 2008). Uncovering the biodiversity of yeast adhesin structure and function may also provide new insights into the evolutionary adaptation of fungal cell adhesion in general and further develop S. cerevisiae into an attractive model for ecology and evolution (Replansky et al., 2008). A second future effort should be the elucidation of the physiologically relevant conditions and the corresponding regulatory pathways that control adhesin gene expression in natural environments. Clearly, the mechanisms that control adhesin genes in the subtelomeric regions are far from understood. Here, new lessons learned from S. cerevisiae may also contribute to a better understanding of adhesin gene regulation in human pathogenic yeasts, for example C. glabrata, which often carry important epithelial adhesin genes close to telomeres (Castano et al., 2006). A further and very challenging future goal is the identification of ecophysiologically relevant surfaces structures and ligand molecules that are bound by S. cerevisiae adhesins in natural environments and habitats. There is only limited information about the distribution and population structure of wild strains in natural environments (Fay & Benavides, 2005; Aa et al., 2006). Saccharomyces cerevisiae has been isolated from cultivated plants, for example damaged or rotting fruits, but also from uncultivated habitats, for example the bark and fluxes of oaks in natural woodlands (Mortimer et al., 1994; Polsinelli et al., 1996; Naumov et al., 1998; Mortimer & Polsinelli, 1999; Sniegowski, 1999; Sniegowski et al., 2002; Fay et al., 2004). In this context, it is interesting to note that adhesin genes and genes for hydrolytic enzymes, for example pectinases and glucoamylases, are coregulated in S. cerevisiae, indicating a competence for plant substrate adhesion and invasion (Vivier et al., 1997; Madhani et al., 1999). Saccharomyces cerevisiae and other Saccharomyces sensu stricto species have also been found to be associated with insects, for example wasps, honey bees and fruit flies (Phaff & Knapp, 1956; Phaff et al., 1956; Stevic, 1962; Lachance et al., 1995; Naumov et al., 1995, 1996). It has therefore been discussed that S. cerevisiae, which is not an airborne microorganism, may use insects as a vector for dissemination (Stevic, 1962; Mortimer & Polsinelli, 1999). Although these observations point towards possible foreign surfaces that might be specifically recognized by S. cerevisiae, it remains to be determined whether plants or insects have specific surface structures that are recognized by S. cerevisiae adhesin. The future identification of naturally relevant adhesin ligands will also crucially depend on the development of new in vivo and in vitro test systems for S. cerevisiae cell adhesion. These approaches will ultimately allow to understand the ecophysiological roles of adhesins not only for the development of protective multicellular structures, but also for the colonization of substrates and the distribution of S. cerevisiae in natural habitats.
Work in our laboratory was supported by grants from the Deutsche Forschungsgemeinschaft (DFG), MO 825/3-1 and GRK 1216, by the International Max-Planck Research School (IMPRS) for Environmental, Cellular and Molecular Microbiology and by the Marburg Center for Synthetic Microbiology (SYNMIKRO).