OUP user menu

Saccharomyces cerevisiae — a model to uncover molecular mechanisms for yeast biofilm biology

Rasmus K. Bojsen, Kaj Scherz Andersen, Birgitte Regenberg
DOI: http://dx.doi.org/10.1111/j.1574-695X.2012.00943.x 169-182 First published online: 1 July 2012


Microbial biofilms can be defined as multi-cellular aggregates adhering to a surface and embedded in an extracellular matrix (ECM). The nonpathogenic yeast, Saccharomyces cerevisiae, follows the common traits of microbial biofilms with cell–cell and cell–surface adhesion. S. cerevisiae is shown to produce an ECM and respond to quorum sensing, and multi-cellular aggregates have lowered susceptibility to antifungals. Adhesion is mediated by a family of cell surface proteins of which Flo11 has been shown to be essential for biofilm development. FLO11 expression is regulated via a number of regulatory pathways including the protein kinase A and a mitogen-activated protein kinase pathway. Advanced genetic tools and resources have been developed for S. cerevisiae including a deletion mutant-strain collection in a biofilm-forming strain background and GFP-fusion protein collections. Furthermore, S. cerevisiae biofilm is well applied for confocal laser scanning microscopy and fluorophore tagging of proteins, DNA and RNA. These techniques can be used to uncover the molecular mechanisms for biofilm development, drug resistance and for the study of molecular interactions, cell response to environmental cues, cell-to-cell variation and niches in S. cerevisiae biofilm. Being closely related to Candida species, S. cerevisiae is a model to investigate biofilms of pathogenic yeast.

  • fungal biofilm
  • CLSM
  • variability
  • flocculation
  • pseudohyphal
  • invasive


Most human infections are associated with microbial biofilm formation (NIH, 1999). A biofilm is defined by two criteria. The cells must (1) adhere to a surface and (2) produce an extracellular matrix (ECM; Costerton et al., 1999). While bacterial biofilms have been studied intensively (O'Toole et al., 2000; Hall-Stoodley et al., 2004; Høiby et al., 2011), much less is known about the development and architecture of fungal biofilms (Finkel & Mitchell, 2011). However, fungal infections have become a major nosocomial problem because of an increase in the use of immunosuppressive drugs, broad-spectrum antibiotics and invasive devices (Viudes et al., 2002; Sandven et al., 2006; Tortorano et al., 2006; Pfaller & Diekema, 2007; Arendrup et al., 2011). Candida albicans and Candida glabrata are the most frequent causes of fungal infections in humans in the Northern Hemisphere, with an increasing number of human isolates (Pfaller & Diekema, 2007; Arendrup, 2010; Arendrup et al., 2011). However, investigating the pathogenicity of Candida spp. through genetic modifications is difficult because of its diploid nature. Candida glabrata is phylogenetically more closely related to Saccharomyces cerevisiae than to other Candida spp. (Barns et al., 1991; Dujon et al., 2004; Dujon, 2006). Both S. cerevisiae and C. glabrata can produce biofilms as haploids (Whelan et al., 1984; Hawser & Douglas, 1994; Reynolds & Fink, 2001) and form a thin biofilm layer of budding yeasts (Seneviratne et al., 2009; Haagensen et al., 2011). Saccharomyces cerevisiae is genetically tractable and has several properties that make it a favoured model organism (Guthrie & Fink, 1991). Saccharomyces cerevisiae is rarely pathogenic (McCusker et al., 1994), has a high rate of homologous recombination and has a highly versatile DNA transformation system (Rothstein, 1983; Wach et al., 1994). Because of its use in the food industry and as a cell biology model, it has been studied extensively. Saccharomyces cerevisiae was the first eukaryotic genome to be sequenced (Goffeau et al., 1996), making it amenable to global genetic and phenotypic analysis. In addition, both transcriptomic (DeRisi et al., 1997; Velculescu et al., 1997) and proteomic (Zhu et al., 2001) studies were first applied in S. cerevisiae. Consequently, advanced genetic tools have been developed for this fungus. Ten years ago, Reynolds and Fink introduced S. cerevisiae as a model for yeast biofilm studies (Reynolds & Fink, 2001). Biofilm formation of S. cerevisiae and its regulation are conserved in opportunistic pathogenic Candida spp. (Rigden et al., 2004; Desai et al., 2011). Hence, understanding of adherence and its regulation in S. cerevisiae contributes to our understanding of the orthologous mechanisms in Candida spp. Other properties of yeast biofilms may also be conserved, such as quorum sensing (QS) mechanisms (Chen et al., 2004; Chen & Fink, 2006) and the presence of an ECM (Hawser & Douglas, 1994; Kuthan et al., 2003). Taken together, these make S. cerevisiae an attractive model for biofilm studies.

In this review, we focus on the traits common to bacterial and pathogenic yeast biofilms that are also found in S. cerevisiae, specifically adhesion, ECM, QS, drug resistance and evolution of cell surface variation. The knowledge of molecular mechanisms for cell–cell and cell–surface adherence in S. cerevisiae is detailed and well reviewed (Brückner & Mösch, 2011). As adhesion is essential for biofilm, environmental cues and pathways regulating adhesion are also expected to affect biofilm development. Because less is known about the molecular mechanisms for matrix formation, QS and drug resistance, the last part of the review contains a discussion of novel microscopic techniques and state-of-the-art molecular genetics that can be applied to identify and investigating factors for S. cerevisiae biofilm development.

Molecular basis for cell surface adhesion

Attachment of S. cerevisiae to foreign surfaces such as polystyrene is dependent on the cell surface protein Muc1/Flo11 (Reynolds & Fink, 2001). In S. cerevisiae, Flo1, Flo5, Flo9 and Flo10 are homologues to Flo11 and they consist of an A, B and C domain (Verstrepen et al., 2004). The N-terminal A domain provides the adhesive properties (Hoyer et al., 1998; Kobayashi et al., 1998). In Flo1, Flo5, Flo9 and Flo10, the A domain is a conserved β-barrel structure denoted the PA14 domain (Rigden et al., 2004; Veelders et al., 2010), which is homologues to the EPA gene products of C. glabrata (Rigden et al., 2004), suggesting similar functions for these gene products. While Flo1, Flo5, Flo9 and Flo10 confer cell–cell adhesion via mannose binding, Flo11 expression in the biofilm-forming S. cerevisiae Σ1278-b strain background confers agar and polystyrene adhesion, but not strong cell–cell adhesion (Guo et al., 2000). In S. cerevisiae var. diastaticus, however, Flo11 expression confers flocculation (cell aggregation) and this Flo11-mediated cell–cell binding is inhibited by mannose (Douglas et al., 2007). The Flo B domain is variable in length and consists of tandem repeats rich in serine and threonine residues. The serine/threonine residues are susceptible to N- or O-linked glycosylation and both Flo1 (Straver et al., 1994; Bony et al., 1997) and Flo11 (Douglas et al., 2007) have been shown to be glycosylated. Finally, the C domain in the C-terminal region contains a site for covalent attachment of a glycosyl phosphatidylinositol anchor (GPI) that can link the Flo adhesins to the plasma membrane (Bony et al., 1997; Caro et al., 1997).

Regulation of adhesion in response to environmental cues

Besides its role in biofilm development, FLO11 is also shown to be essential for pseudohyphae development in diploid cells upon nitrogen starvation (Lo & Dranginis, 1998) and haploid invasive growth on agar (Cullen & Sprague, 2000). Even though these phenotypes are different from biofilm development on polystyrene, many of the factors regulating FLO11 in biofilm can be expected to be the same for invasive and pseudohyphal growth.

FLO11 expression in the Σ1278b background is regulated at the transcriptional level by a number of environmental cues and signalling pathways. A mitogen-activated protein kinase (MAPK) pathway regulates FLO11 via the GTP-binding protein Ras2 (Mösch et al., 1996, 1999; Lo & Dranginis, 1998). Upon MAPK pathway activation, the DNA-binding protein Tec1 induces FLO11 transcription (Roberts & Fink, 1994; Köhler et al., 2002; Heise et al., 2010) either on its own or cooperatively with Ste12 (Madhani & Fink, 1997; Rupp et al., 1999; Heise et al., 2010).

Another master regulator of FLO11 expression is the protein kinase A (PKA) pathway (Rupp et al., 1999), which controls the FLO11 promoter trough transcriptional interference by a noncoding RNA, ICR1 (Bumgarner et al., 2009). ICR1 overlaps the FLO11 promoter and part of the open reading frame and its transcription inhibits FLO11 transcription. Transcription of the interfering ICR1 is dependent on the Sfl1 transcription factor (Bumgarner et al., 2009). Thus, Sfl1 is effectively a negative regulator of FLO11 (Robertson & Fink, 1998; Pan & Heitman, 2002). FLO11 is in a transcriptional permissive state when ICR1 is repressed. This state is dependent on the transcription factor Flo8 and the histone deacetylase Rpd3L (Bumgarner et al., 2009). Flo8 and Sfl1 are regulated by the PKA pathway through the Tpk2 protein kinase (Robertson & Fink, 1998; Pan & Heitman, 2002). Competition between Flo8 and Sfl1 for binding to the FLO11 promoter (Pan & Heitman, 2002) determines whether ICR1 upstream of FLO11 is transcribed and whether FLO11 is in a silenced or a transcriptionally competent state (Bumgarner et al., 2009).

A number of environmental cues are detected by the MAPK and PKA pathways for regulation of FLO11 and might as such affect biofilm development. Glucose acts on the protein kinase, Tpk2, via the transmembrane G-protein receptor Gpr1, the G-protein alpha subunit Gpa2 and cAMP (Colombo et al., 1998; Kraakman et al., 1999). Another protein kinase, Tpk1, is responsible for derepression of FLO11 in response to low levels of glucose. Tpk1 phosphorylates Yak1 at high glucose levels (Zhu et al., 2000; Malcher et al., 2011), which targets Sok2 for binding and repression of the FLO11 promoter (Borneman et al., 2006). At low glucose levels, this Tpk1 repression is relieved and FLO11 activated. Glucose starvation also acts on FLO11 expression through the derepressing Snf1 protein kinase pathway (Carlson et al., 1981; Kuchin et al., 2002; Van de Velde & Thevelein, 2008).

Low levels of ammonium regulate cAMP/PKA and MAPK pathways in diploid cells via the ammonium permease Mep2 (Lorenz & Heitman, 1998a, b). Lorenz and Heitman observed that pseudohyphal growth is lost in a diploid mep2/mep2 mutant (Lorenz & Heitman, 1998a, b). This phenotype was repressed with cAMP and dominant RAS2 and GPA1 alleles, suggesting that both Ras2 and Gpa1 are activated by Mep2 (Lorenz & Heitman, 1998a, b). Ras2 signals to the PKA pathway (Toda et al., 1985) as well as to the MAPK pathway (Mösch et al., 1996). Thus, the ammonium signal via Mep2 appears to induce FLO11 via both pathways. The degradation products of tryptophan, tyrosine, tryptophol and tyrosol also induce FLO11 transcription via Tpk2, but the upstream components are unknown (Chen & Fink, 2006).

Several lines of evidence indicate that amino acids also regulate FLO11 gene expression. Low levels of proline and glutamine induce pseudohyphal growth in diploid cells (Gimeno et al., 1992; Lorenz & Heitman, 1998a, b). Lorenz and Heitman suggest that amino acid transporters might transduce this signal. This hypothesis is indirectly supported by the findings of Ljungdahl and co-workers that loss of the Ptr3 regulatory component of the amino acid-sensing pathway leads to increased adhesive growth in haploid cells (Klasson et al., 1999). The ptr3 mutant has increased activity of the general amino acid permease, Gap1 (Klasson et al., 1999), which could mediate FLO11 expression, according to the presence of amino acids in the environment via an unknown pathway.

In addition to transcription factors regulated by the PKA and MAPK pathways, a number of DNA-binding factors are known to regulate FLO11 expression (Lambrechts et al., 1996; Lorenz & Heitman, 1998a, b; Gagiano et al., 1999; Van Dyk et al., 2003, 2005; Kim et al., 2004; Prusty et al., 2004; Bester et al., 2006; Borneman et al., 2006). The exact role of these factors in FLO11 transcription and most environmental cues regulating their activity has not been clarified, but because of their impact on FLO11, they are expected to be involved in S. cerevisiae biofilm development.

Variability and bistability of cell surface adherence

The adhesive properties of S. cerevisiae vary more than most other traits in this species (Hahn et al., 2005; Van Mulders et al., 2010). This variability arises through: (1) epigenetically inherited changes in expression patterns of the FLO genes, (2) mutations affecting regulatory genes and elements of FLO genes, (3) deletions and insertions affecting the number of repeats in the B domain of Flo proteins and (4) point mutations affecting substrate affinity of the A domain as discussed earlier. Phenotype switching might therefore be a mechanism by which a biofilm population can disperse via nonadhesive planktonic cells.

Regulation of FLO11 by the histone deacetylase, Hda1, allows for epigenetic inheritance of the FLO11 transcriptional state (Halme et al., 2004). In a population of clonal diploid cells, subpopulations of cells might repress FLO11 in an Hda1-dependent manner while others express FLO11, leading to morphological variation in the population. This epigenetic switch is likely to play a similar role for FLO11 expression in biofilm-forming haploid cells so that only a subpopulation of cells form a biofilm, while the remaining exist in a planktonic form.

The presence of several FLO genes in the S. cerevisiae genome allows for a variety of cell surface properties and biofilm morphotypes depending on their expression (Van Mulders et al., 2010). FLO11 is located on chromosome IX in the middle of the right arm (Lo & Dranginis, 1996), where it is conditionally expressed in the Σ1278b background. FLO1, FLO5, FLO9 and FLO10 are in subtelomeric regions (Teunissen et al., 1993, 1995; Carro et al., 2003; Verstrepen et al., 2004), where they are repressed and restricted in their influence on morphotype (Guo et al., 2000; Halme et al., 2004; Van Mulders et al., 2010). Expression of FLO1, FLO5, FLO9 and FLO10 from plasmids or in brewer strains shows that all four genes infer adhesive properties (Guo et al., 2000; Van Mulders et al., 2010) making the genes reservoirs for cell surface variability in biofilms.

Subtelomeric localization and the repetitive motifs of the FLO genes may also be important in the ability of S. cerevisiae biofilms to evolve. Subtelomeric regions and repetitive motifs increase evolution rates (Louis & Haber, 1990), and the repetitive motifs within FLO1 have been shown to trigger frequent recombination events causing expansions and contractions of the gene (Verstrepen et al., 2005). Mutation in the repetitive region of FLO11 has been observed in sherry yeast (Fidalgo et al., 2006), which results in cells that rise to the surface of the sherry during fermentation. Hence, minor mutations enabled by the location and gene structure of the FLO might be important for cell surface variability in S. cerevisiae biofilms.

In addition to the FLO genes, a number of genes encode homologues of one or several of the A, B or C domains. Because these genes do not encode all three domains, they may not function in cell surface adhesion. They might, however, serve as a genetic pool for a rapid evolution of novel cell surface properties through recombination with the FLO genes (Verstrepen et al., 2004).

The genetic and epigenetic mechanisms for variability in S. cerevisiae adhesive properties could reflect a selective pressure for high evolvability of adhesion in the natural environment of this species. Organisms adapt to ever-changing environments by stochastic genetic and epigenetic switches that ensure subpopulations with traits that, while not necessarily advantageous for the given environment, might be in another (Acar et al., 2008; Veening et al., 2008). Genetic switches are known to affect the cell surface properties of biofilm-forming microorganisms and might enable migration and establishment of novel populations, and in the case of pathogens, immune system evasion (Justice et al., 2008).

Extracellular matrix

An ECM has been identified in biofilms of organisms as diverse as bacteria, algae, archaea and fungi (Flemming & Wingender, 2010). ECM-like substances have also been shown in S. cerevisiae using electron microscopy (Kuthan et al., 2003; Beauvais et al., 2009; Zara et al., 2009; St'ovicek et al., 2010). So far, matrix has been identified in S. cerevisiae colonies on agar and in multicellular consortia such as flor or flocs, and we expect that S. cerevisiae biofilms also contain matrix and thus follow the classical definition of a biofilm.

The S. cerevisiae ECM-like structure observed with electron microscopy has been extracted with EDTA and is found to contain mono- and polysaccharides (Beauvais et al., 2009). In addition, a protein unrelated to flocculins has been extracted with Tween and SDS detergents from fluffy colonies (Kuthan et al., 2003).

Matrix in flocculating cells has been shown to contribute to exclusion of high molecular weight molecules such as dyes, but the matrix does not contribute to stress resistance to small molecules such as ethanol (Beauvais et al., 2009). A function of the matrix could be protection of cells within the biofilm by lowering the permeability of antifungal compounds (Beauvais et al., 2009; Vachova et al., 2011). In addition to an excluding function, the space within a matrix might serve as reservoirs for nutrients and waste products (Kuthan et al., 2003) as in bacterial biofilms (Sutherland, 2001).

Quorum sensing

QS is the process in which cells sense each others' presence through self-produced QS molecules (autoinducers). Increasing the concentration of a QS molecule to a threshold level induces a population-wide phenotypic change (Nealson et al., 1970; Bassler et al., 1993). This form of social behaviour has been shown to be important for the formation of bacterial biofilms (Vuong et al., 2000) and pathogenic yeast (Ramage et al., 2002; Chen et al., 2004).

QS has been shown to regulate FLO11 and thus might have an impact on the development of S. cerevisiae biofilms. S. cerevisiae uses ethanol and the aromatic alcohol tryptophol and phenylethanol as autoinducers in a cell density-dependent manner (Chen & Fink, 2006; Smukalla et al., 2008). When the cell density is sufficiently high, the production of ethanol and aromatic alcohols reaches a threshold, activating FLO11 expression via the PKA pathway (Chen & Fink, 2006). Hence, tryptophol and phenylethanol likely influence S. cerevisiae biofilm development through the regulation of FLO genes. Candida albicans uses the structurally related aromatic alcohol tyrosol as a QS molecule (Chen et al., 2004), while tryptophol and phenylethanol do not induce phenotypic changes in C. albicans (Chen & Fink, 2006).

Cell-to-cell communication has been described in S. cerevisiae with ammonia as an airborne signalling molecule, produced by one cell and sensed by another to induce oriented growth (Palkova et al., 1997). Although ammonia is not a quorum molecule in the strict sense, it is an example of communication between individual S. cerevisiae cells in two subpopulations.

Biofilm resistance

Biofilms are known for their resistance to antimicrobial agents (Kuhn et al., 2002; Olson et al., 2002). Reduced accessibility of the antibiotics to cells in a biofilm and phenotypic variability within the biofilm population are suggested as mechanisms responsible for the reduced susceptibility (Hoyle et al., 1990; Costerton et al., 1999; Høiby et al., 2010).

In S. cerevisiae, the majority of cells in a flocculating population can survive concentrations of amphotericin B that are 100-times higher than the minimum inhibitory concentration for planktonic cells (Smukalla et al., 2008). Fink et al. found that only the outer layer of cells in a floc are affected by amphotericin B and the flocculating lifestyle is a physiological state that indicates reduced growth. Reduced growth rate and dormancy are believed to be involved in antibiotic persistence of bacterial biofilms and could be caused by a nutrient-limiting gradient across the biofilm (Brown et al., 1988; Gilbert et al., 1997; Lewis, 2007).

Biofilm formation of S. cerevisiae have been found to decrease susceptibility to biocides (Tristezza et al., 2010) and antifungals (Chandra et al., 2001) suggesting that S. cerevisiae biofilms have the common traits of resistance that are observed in other organisms.

Methods for biofilm analysis

Until recently, S. cerevisiae biofilms have been mainly investigated macroscopically using agar plate assays or crystal violet staining of biofilms on polystyrene (Reynolds & Fink, 2001). These methods are suitable for high-throughput screens but do not allow investigation of: (1) individual cells in a biofilm; (2) macromolecular interactions in a biofilm; (3) physiological responses of individual cells in a biofilm to physical changes in the environment or (4) environmental niches of biofilm cells. Recently, two tools have been developed that can be used to address these issues. High-resolution imaging of live biofilm allows characterization of fluorophore-labelled biofilm and macromolecules such as RNA and protein (Fig. 1), and a mutant collection in the biofilm-forming S. cerevisiae Σ1278b strain background permits screening for gene products involved in biofilm development. Combination of the two methods finally gives the opportunity to screen for mutants with altered physiological response to factors in the biofilm or the environment (methods listed in Table 1).

Figure 1

CLSM of Saccharomyces cerevisiae (CEN.PK113.7D sfl1) biofilm after (a, d) 6 h, (b, e) 18 h, (c, f) 30 h growth in synthetic complete medium with amino acids and 2% glucose (SC-ura). Cells were grown in a Lab-Tek™ Chamber Slide™ System; Permanox® (NUNC, Denmark) in 1 mL medium and stained 30 min with SYTO9. a, b and c are 3D reconstructions of biofilm made from 2-µm thick images in stacks of up to 75 individual images. d, e and f are sections in the X–Y, X–Z and Y–Z dimensions of the biofilm. Both reconstruction images and sections through the biofilm were made with imaris software (Bitplane) from raw CLSM images. CLSM was performed with a Zeiss LSM510 microscope using a 63×/0.95NA water immersion lens. Bar 30 µm.

View this table:
Table 1

Methods developed for Saccharomyces cerevisiae that can be applied to study biofilm

Identification of genes involved in biofilm developmentScreening of Σ1278b mutant collection byDowell et al. (2010)
Crystal violet screening of individual mutantsReynolds & Fink (2001)
Quantification of barcode tags in mixed populationsWinzeler et al. (1999), Giaever et al. (2002), Gresham et al. (2011)
Characterization of extra cellular matrixScanning electron microscopyKuthan et al. (2003), Zara et al. (2009), St'ovicek et al. (2010)
Raman microscopySmith & Berger (2009), Wagner et al. (2009)
High-resolution imaging of life cells and cellular responses in biofilmCLSM combined with
Fluorescent dyesChandra et al. (2001), Seneviratne et al. (2009), Haagensen et al. (2011)
Fluorescent-protein visualization of
ProteinHuh et al. (2003)
RNABertrand et al. (1998)
DNAThrower & Bloom (2001)
Protein–protein interactions (FRET)Dye et al. (2005)
Intracellular pH and growth rateOrij et al. (2009)
comstat quantification of features measured with CLSMHeydorn et al. (2000a, b), Seneviratne et al. (2009)
Controlled environmental conditions in flow cellsWeiss Nielsen et al. (2011)
Introduction of molecular markersAutomated selection of double mutants by SGA analysisTong et al. (2001)
  • Only published for Candida spp.

Microscopy of S. cerevisiae biofilms

Scanning electron microscopy offers nanometre-scale resolution (Paddock, 2000) and can be used to obtain information about the architecture and matrix of a biofilm (Kuthan et al., 2003; Zara et al., 2009; St'ovicek et al., 2010). While electron microscopy is suited for visualization of biofilm structures at high resolution, this method cannot be used to follow live biofilm over time.

High-resolution imaging of live cells in developing biofilms can be obtained by confocal laser scanning microscopy (CLSM). Three-dimensional CLSM images of a biofilm are obtained by stacking and reconstructing images from scans through the depth of the biofilm. Because CLSM records a fluorescent signal, any molecule that can be labelled fluorescently can potentially be visualized in a yeast biofilm at micron-scale resolution (Paddock, 2000). CLSM has been used extensively to study bacterial biofilms over the last decade (Klausen et al., 2003; Haagensen et al., 2007; Folkesson et al., 2008; Pamp et al., 2009). Recently, the method has been applied to visualize yeast biofilms of C. albicans, C. glabrata and S. cerevisiae (Chandra et al., 2001; Seneviratne et al., 2009; Haagensen et al., 2011; Weiss Nielsen et al., 2011). CLSM yield valuable three-dimensional information about yeast biofilm architecture and can be used to study, for example, biofilm development over time (Fig. 1). So far, CLSM has not been used to differentiate S. cerevisiae cells within a biofilm. However, the variety of labelling methods and fluorescently labelled libraries developed for this organism offer promising tools for the study of cell–cell variability in S. cerevisiae biofilm by CLSM.

CLSM can also be used in combination with Raman microscopy (RM) to obtain information about the chemical composition of the ECM (Wagner et al., 2009). RM uses specific Raman scattering signals to detect chemical components with high sensitivity to chemical composition changes (Smith & Berger, 2009; Wagner et al., 2009). As RM does not require staining, it is not limited by the need for specific dyes to identify matrix macromolecules (e.g. polysaccharides, proteins) and does not interrupt biofilm architecture (Wagner et al., 2009).

Fluorescence markers for CLSM

Yeast biofilms have been visualized by CLSM using fluorescent dyes such as the nucleic acid stains SYTO9 and propidium iodide, the cytoplasm stain FUN1 and the glucose- and mannose-binding concanavalin A-Alexa Fluor (Fig. 1; Chandra et al., 2001; Kuhn et al., 2002; Seneviratne et al., 2009). Combinations of fluorescent signals can be used to simultaneously investigate subpopulations in a mixed population. LIVE/DEAD assays with dye combinations of SYTO9 and propidium iodide have been used successfully in bacterial biofilm studies and can be used to differentiate S. cerevisiae cells (Zhang & Fang, 2004; Seneviratne et al., 2009). Propidium iodide penetrates only damaged cell membranes and therefore stains only dead cells. However, the staining procedure results in disturbance of the biofilm by either mechanical stress or growth inhibition. A noninvasive solution for this problem is labelling biofilm-forming cells with a fluorescent protein. The fluorescent proteins GFP (green, excitation (ex): 488 nm; emission (em): 507 nm), YFP (yellow, ex: 514; em: 527), CFP (cyan, ex: 433; em: 475), RFP (red, ex: 584; em: 607) and mCherry (red, ex: 587; em: 610) (Shaner et al., 2004, 2005; Müller-Taubenberger & Anderson, 2007) have been optimized for S. cerevisiae (Sheff & Thorn, 2004). Combinations such as mCherry/GFP or mCherry/YFP/CFP can be used, so that two or three labelled components can be followed simultaneously. Fluorescent labelling has been used successfully to monitor the interaction and dynamics of bacterial biofilm subpopulations (Klausen et al., 2003; Haagensen et al., 2007; Pamp & Tolker-Nielsen, 2007; Macia et al., 2011) and is likely to be a powerful tool for analysis of S. cerevisiae biofilm.

Detection of DNA, RNA and protein by CLSM

Molecules that have been successfully tagged with a fluorescent protein in S. cerevisiae include DNA (Thrower & Bloom, 2001), RNA (Bertrand et al., 1998) and proteins (Huh et al., 2003). Labelling of these molecules with fluorescent proteins such as GFP offers great opportunities to investigate differentiation of S. cerevisiae biofilm and locations of protein, RNA and DNA in yeast biofilm. Besides its application as a method to study differentiation of cells in yeast biofilm, fluorescent labelling of proteins can also be a valuable tool to study experimental evolution in live biofilm. Mutants that explore certain niches of the biofilm can thus be followed by CLSM of labelled proteins that are specifically expressed in the mutant.

CLSM might also be used to determine gene expression levels of individual cells in a biofilm. GFP expression levels correlate with fluorescence intensity (Li et al., 2000). Therefore, relative expression levels of a gene can be monitored if a GFP cassette is placed under control of a promoter controlling the transcription of a particular gene.

Protein–protein interactions in S. cerevisiae biofilms

Another application of CLSM is detection of protein interactions in living cells by fluorescence microscopy of fluorescence resonance energy transfer (FRET). FRET is well suited for studying cell-specific protein–protein interactions in a highly diverse cell population such as a biofilm. The principle of FRET is that emitted light energy of an excited donor fluorophore is transferred to and excites an acceptor fluorophore. This phenomenon occurs only when the two fluorophores are in close proximity. For example, a CFP fusion protein excites an YFP fusion protein only when they are separated by 2 nm or less (Dye et al., 2005).

Another method to visualize protein–protein interactions in living yeast cells is bimolecular fluorescence complementation (BiFC). Interaction between two proteins is tested by fusion of the proteins to different nonfluorescent fragments of a fluorescent protein. Interaction of the proteins forms a fluorescent complex that can be detected microscopically (Kerppola, 2008).

Determination of growth rate with CLSM

Individual cells in a biofilm population are predicted to have diverse growth rates and this might affect both stress resistance and antifungal tolerance (Brown & Donnelly, 1988; Gilbert et al., 1997). Because the growth rate correlates to transcript levels of a large number of genes (Regenberg et al., 2006; Brauer et al., 2008), expression of GFP from growth rate-regulated promoters could be used to monitor the growth of individual biofilm cells.

An alternative method for determining growth rates uses ratiometric pHluorin, which is a pH-sensitive GFP protein that responds to intracellular pH in living S. cerevisiae cells (Miesenböck et al., 1998; Orij et al., 2009). Intracellular pH changes with growth rate (Orij et al., 2009). Therefore, pHluorin can be used to measure the growth rate of individual cells in a biofilm. Recently, pHluorin2 with enhanced fluorescence has been developed (Mahon, 2011).

Finally, fluorescent in situ hybridization (FISH) of rRNA with fluorophore-labelled probes can be used to determine growth rate of individual biofilm cells by CLSM. In several microorganisms, the number of ribosomes is correlated with the growth rate in exponential phase (Kjeldgaard & Kurland, 1963; Waldron & Lacroute, 1975; Poulsen et al., 1993; Møller et al., 1995). A standard correlation between growth rate and ribosomal content as measured by quantitative FISH has been applied to the exponential and stationary phases of bacteria (Yang et al., 2008). Specific probes for S. cerevisiae rRNA have been developed (Inacio et al., 2003) and might be used to determine growth rate of individual cells in S. cerevisiae biofilms.

S. cerevisiae in mixed biofilms

Fungi can co-exist in the same biofilm with bacteria (Adam et al., 2002; Hogan & Kolter, 2002). FISH-rRNA can thus be used to detect and localize different species in a mixed species biofilm (Thurnheer et al., 2004). The results can be visualized by CLSM and could provide valuable information about the architecture of mixed biofilms and possible interspecies interactions.

Controlling biofilm growth conditions for CLSM

Saccharomyces cerevisiae biofilms can be cultured under either static condition in growth chambers or at controlled environmental conditions in flow cells (Fig. 2). Both techniques are compatible with CLSM (Haagensen et al., 2011; Weiss Nielsen et al., 2011). Static growth conditions are obtained by culturing cells in a growth chamber that is attached to a microscope slide (Fig. 2a). The static growth system has the advantage that it is easy to set up and the disadvantage that growth conditions are not easily controlled. Flow cells are composed of a chamber through which medium flows and a cover slip on which biofilm forms (Fig. 2b). The flow-cell system has a continuous supply of nutrients that is easily changed, for example, for administration of antifungals with minimal biofilm disturbance (Weiss Nielsen et al., 2011). CLSM of biofilm formed in flow cells is a powerful tool to study gene regulation upon changing environmental conditions and can be used to study regulation of, for example, FLO genes by the use of FLO promoter-GFP fusions in the biofilm-forming cells. The CLSM flow-cell method can also be used to visualize phenotypic variabilities and bistabilities in the biofilm such as variation in repression of FLO5, 9, 10 and bistabilities in FLO11 expression generated by Hda1. While many bacterial biofilms are formed on glass surfaces, S. cerevisiae biofilms are observed on polystyrene surfaces (Reynolds & Fink, 2001). However, some polystyrenes are autofluorescent and interfere with CLSM recording. Polyvinyl coverslips are an optimal choice as a surface for yeast biofilm development and CLSM imaging, as this plastic supports biofilms and is not autofluorescent in the range of the common fluorophores (430–610 nm) (Haagensen et al., 2011; Weiss Nielsen et al., 2011).

Figure 2

Illustration of culture conditions for CLSM of Saccharomyces cerevisiae biofilm. (a) Biofilm cultured under static condition in growth chamber. Cells attach to a microscope slide and are covered by medium. For imaging, the objective is lowered into the medium. (b) Biofilm cultured under environmental controlled conditions in a flow cell. Cells attach to a coverslip that is glued to a flow cell with silicone forming a watertight seal. The medium flow can be controlled and environmental condition can gradually be changed. The objective does not disturb the biofilm.

Quantification of features in a biofilm

Three-dimensional biofilm structures can be quantified using comstat software, based on the stack of images acquired by CLSM (http://www.comstat.dk). Features calculated by COMSTAT include biovolume, area occupied by cells in each layer, thickness, substratum coverage, fractal dimension, roughness, surface-to-volume ratio, number of microcolonies and microcolony size (Heydorn et al., 2000a, b). Although this software is mainly used for quantification of bacterial biofilms, it will be a valuable tool for objective quantitative analysis of yeast biofilms (Seneviratne et al., 2009).

Genetic tools for identification of molecular factors involved in biofilm development

Fluorescent markers for CLSM are relatively easily integrated in the S. cerevisiae genome. The high frequency of homologous recombination allows for one-step gene replacement between a DNA cassette and a corresponding genomic sequence with as little as 35 bp of genomic homology (Rothstein, 1983; Wach et al., 1994). This unique feature and others have led to the synthesis of two complete deletion strain collections of S. cerevisiae (Giaever et al., 2002; Dowell et al., 2010), and GFP fusions to most S. cerevisiae gene products (Huh et al., 2003).

A powerful resource for identification of genes involved in biofilm development is an almost complete collection of deletion mutants in the biofilm-forming S. cerevisiae background Σ1278b (Dowell et al., 2010). We have recently used this collection for a crystal violet screen for mutants that have reduced biofilm-forming ability (unpublished). The screen revealed 56 genes not previously associated with biofilm development. We foresee that many of these genes are involved in the regulation of FLO1, FLO5, FLO9, FLO10 and FLO11 and understanding of their involvement in biofilm development will aid the understanding of FLO regulation.

Each mutant in the Σ1278b deletion collection carries a gene deletion made by a kanamycin-resistance cassette flanked by unique 20-nucleotide sequences. The 20-nucleotide barcode tags enable identification of each mutant in a mixed population (Fig. 3a). A pool of mutants can thus be grown under selective conditions and the abundance of the individual mutant in a biofilm assessed by the frequency of the individual barcode tags (Winzeler et al., 1999). Barcode frequencies are measured either by array analysis (Winzeler et al., 1999; Giaever et al., 2002) or sequencing (Gresham et al., 2011).

Figure 3

Principles for screening of the Σ1278b barcode-tagged deletion mutant collection (a) and cloning free introduction of GFP-genes fusions in a mutant collection (b). (a) A pool of mutants is screened for nonbiofilm formers. vvvΔ, xxxΔ and yyyΔ: refer to genotypes of three different mutants deleted in gene VVV, gene XXX and gene YYY. Barcode v, barcode x and barcode y: the barcode sequences that are specific for deletion vvvΔ, xxxΔ and yyyΔ. Mutants are grown on a solid surface. The surface is washed, and cells not in the biofilm are removed. The barcode frequencies are determined before and after wash, to identify the mutants that do not form biofilm. (b) Introduction of a GFP gene fusion in mutant xxxΔ barcode x. xxxΔ barcode x is a deletion of wt allele XXX made with a kanamycin-resistance cassette (KanMX) in a strain with the genotype MATa xxxΔ::KanMX barcode x ZZZ can1::STE2p-SpHIS5 lyp1::STE3p-LEU2 his3::HisG leu2 ura3. ZZZ-GFP encodes a protein–GFP fusion linked to the nourseothricin-resistantce gene (natMX). ZZZ is the wild-type allele of the protein of interest. A MATα XXX ZZZ-GFP-natMX his3 is mated with the MATa xxxΔ::KanMX barcode x ZZZ can1::STE2p-SpHIS5 lyp1::STE3p-LEU2 his3::HisG leu2 ura3 haploid from a mutant library. Selection for the heterozygote diploid MATa/α is carried out with nourseothricin and kanamycin. Diploids are subsequently sporulated and haploid xxxΔ::KanMX barcode-x ZZZ-GFP-natMX double mutants selected on nourseothricin and kanamycin. Mating type of haploid MATa can be selected for through expression of the mating-type-specific STE2p-Sphis5 cassette that causes a His+ phenotype. MATα haploids express lyp1::STE3p-LEU2 and are thus Leu+. (b) is based on the principles for SGA analysis described by Boone et al. (Tong et al., 2001).

In 2001, Boone et al. published a procedure called synthetic genetic array (SGA) analysis for selection of double mutants through automated crossing (Tong et al., 2001). Besides its use for analysis of synthetic genetic interactions, this unique method can also be used to cross mutant alleles such as fluorescent proteins into each of the mutants in the Σ1278b collection (Fig. 3b) (Tong et al., 2001; Huh et al., 2003; Dowell et al., 2010; Song et al., 2010). This offers the opportunity to follow gene expression and cell localization in homogenous or mixed biofilm populations over time using CLSM.

Future perspectives for S. cerevisiae biofilm research

In summary, several features of S. cerevisiae make it an ideal model for studies of fungal biofilms. Although nonpathogenic, some S. cerevisiae strains have the ability to form biofilms, and this is controlled by genes homologous to the genes responsible for biofilm formation in pathogenic Candida spp. The varied genetic and cell biology techniques that have been developed for S. cerevisiae will permit studies on the molecular mechanisms underlying yeast biofilm development, cell–cell interactions in yeast biofilms and drug resistance mechanisms.

In addition to the role of S. cerevisiae as a model for biofilms of opportunistic pathogenic yeasts, S. cerevisiae biofilms could be used as models to study other phenomena in biology. Bacterial biofilms have been described as models for social evolution (Diggle et al., 2007). A population of Pseudomonas aeruginosa cells in a biofilm can communicate via QS (Passador et al., 1993). Cells in the population that produce the quorum molecules are designated cooperative, while individuals that do not produce quorum molecules have a fitness advantage and are designated cheaters (Diggle et al., 2007). The ability of S. cerevisiae to produce cell surface adhesins allows closely related cells to interact and benefit from the physical advantages of being part of the biofilm. Cheaters are predicted to exist in this system as nonadhesin-producing cells that still adhere to adhesin-producing neighbours through mannose interactions (Smukalla et al., 2008; Veelders et al., 2010). A number of other social phenomena such as cross-feeding, resistance and QS might also be involved in the biofilm dynamics of S. cerevisiae.


Danish Agency for Science Technology and Innovation is acknowledged for financial support (FTP 10-084027)


  • Editor: Christine Imbert


View Abstract