Flo11p, drug efflux pumps, and the extracellular matrix cooperate to form biofilm yeast colonies

In natural environments, microorganisms preferentially form organized multicellular communities, such as biofilms and colonies. These structures possess unique attributes that provide resistance against chemicals and other threats and also adaptability to changing conditions, allowing the community to survive in a hostile natural environment (Donlan and Costerton, 2002; Palková, 2004). Features implicated as being essential for the formation of complex fungal biofilms include adhesion to surfaces, the production of an ECM, multidrug resistance (MDR) plasma membrane transporters, and specialized cell subpopulations, such as stationary cells that are more resistant to various stresses (Douglas, 2003; Blankenship and Mitchell, 2006).

Cell–cell and cell–surface adhesion are often mediated by specific cell wall–adhesive glycophosphatidylinositol-anchored proteins (Dranginis et al., 2007). In Saccharomyces cerevisiae, the FLO family of protein adhesins is considered the most important, with pleiotropic Flo11p involved in cell adhesion to inert substrates (Verstrepen et al., 2004), filamentous growth (Lambrechts et al., 1996), and the formation of structured colonies (Vopálenská et al., 2010), flor biofilms (Ishigami et al., 2004), and mats (Reynolds and Fink, 2001). In contrast to Flo1p, Flo5p, and Flo9p, which are crucial for cell–cell adhesion during flocculation (Guo et al., 2000), the role of Flo11p in this process is strain specific (Bayly et al., 2005; Douglas et al., 2007). The ALS family and Hwp1p, which are related to FLO adhesins, mediate the adhesion of Candida albicans and are important for normal biofilm development (Nobile et al., 2008).

An ECM is exclusively present in structured S. cerevisiae colonies but not in their domesticated counterparts (Kuthan et al., 2003), and it is found in air–liquid flor biofilms (Zara et al., 2009). In addition, flocculating S. cerevisiae cells expressing the FLO1 gene secrete a mixture of polysaccharides that blocks the permeation of large molecules (e.g., antibodies; Beauvais et al., 2009). In addition to these findings in S. cerevisiae populations, an ECM is regularly encountered in biofilms of Candida species (Baillie and Douglas, 2000). In both structured S. cerevisiae colonies and yeast biofilms, the ECM facilitates the formation of pores for water and nutrient flow (Douglas, 2003; Kuthan et al., 2003) and protects the communities against dehydration (Flemming and Wingender, 2010).

S. cerevisiae MDR transporters belonging to the ATP-binding cassette family are involved in the ATP-dependent efflux of a wide variety of unrelated compounds, including drugs and other noxious substances. Deletions of these pumps lead to the cells becoming drug hypersensitive (Rogers et al., 2001; Sipos and Kuchler, 2006). The function of the MDR pumps in S. cerevisiae multicellular structures has not yet been examined. Flocculating cells up-regulate some MDR transporter genes (Smukalla et al., 2008), but their function in floc resistance remains undocumented. It has been shown, however, that C. albicans biofilms increase their expression of the MDR genes CDR1, CDR2, and MDR1 (Ramage et al., 2002), and deletions of these genes decrease the resistance of biofilms (Mukherjee et al., 2003).

In contrast to the smooth colonies of laboratory strains, wild S. cerevisiae strains form structured colonies possessing attributes common to fungal biofilms and are hence referred to as biofilm colonies. Their attributes include the production of an ECM rich in polysaccharides and high water retention capacity and the production of the adhesin Flo11p for substrate adhesion and 3D architecture formation (Šťovíček et al., 2010). Using two-photon excitation confocal microscopy (2P-CM; Váchová et al., 2009) in combination with fluorescent protein tagging and staining methods, we show here the dynamics of colony development. Colony structure was composed of an aerial part with an internal cavity and subsurface pseudohyphae, was strengthened by intercellular fibers formed in the presence of Flo11p, and was protected by Pdr5p and Snq2p transporters together with an ECM that was secreted by internal cells.

Examining the population of a wild S. cerevisiae BR-F strain, we show (Fig. 1) that a 25-h-old population of mostly rounded cells had already formed a small colony with short pseudohyphae (i.e., elongated cells joined into filaments; Gimeno et al., 1992) invading the agar at its base. Several hours later, an internal cavity appeared within the mound colony (Fig. 1 A, 34–42 h). The colony then expanded primarily horizontally with an elevated ridge at its margin (Fig. 1 A, the earlike structure in the cross section) and an internal cavity (Fig. 1 A, 48–60 h). The colony also formed abundant rootlike pseudohyphae that grew radially into the agar from its central bottom region (Fig. 1, A and B). Later, the central colony area further expanded horizontally, pushing the ridge apart (Fig. 1 A, 3 d), and the colony became cup shaped (Fig. S1 A). During this period, secondary roots began to grow from the base of the ridge, anchoring it to the agar. The continued horizontal growth of the central cell layer thereby led to its undulation and the formation of wrinkles (Fig. 1 A, 7 d; and Video 1), as also observed in some types of biofilms (Uppuluri et al., 2009). The architecture of structured colonies formed by various wild S. cerevisiae strains is comparable, thus indicating universal underlying principles.

The cells within colonies are interconnected by long, thin fibers that extend from cell walls and form a velcrolike structure (Figs. 1 E and S1 B). The fibers interconnect oval cells in the upper layer and pseudohyphae in the roots (Fig. S1 B). They can hold cells close to one another but can also be stretched up to 460 nm. These fibrous interconnections appear to be composed of two halves, each belonging to one of the interconnected cells, as they are more electron dense in the adhesion area. Adhesin Flo11p participates in the formation of the fibers, which are absent in colonies from the strain lacking Flo11p (BR-F–flo11; Fig. S1 B). Two Flo11p molecules are long enough to form a fiber interconnection 460 nm long, even if they are partially coiled, and they may aggregate into bundles via amyloid formation (Ramsook et al., 2010). Alternatively, the effect of Flo11p could be indirect, e.g., eliciting a change that leads to the production of other surface adhesins. The finding of velcrolike interconnections is of particular interest in C. albicans, in which brushlike fibrillar structures composed of long fibers (100–160 nm) were observed on the surface of cells growing in liquid cultures (Tokunaga et al., 1986), and an involvement of adhesins has been proposed (Klis et al., 2009). Much shorter fibers related to Flo1p were observed on the surface of flocculating S. cerevisiae cells (Beauvais et al., 2009).

To distinguish between the areas of dividing and stationary cells within the developing colony, we used two detection systems: (1) a BR-F–Hmg1p-GFP strain with a GFP gene fused to the HMG1 gene encoding the hydroxymethylglutaryl-CoA reductase of the nuclear envelope (Koning et al., 1996), which exhibits typical distributions in dividing and stationary cells (Fig. 2 B), and (2) the BR-F–cdc3^ts strain with the ts mutation in the CDC3 gene for septin (Fig. 2 A; Hartwell, 1971). Dividing cells were spread uniformly throughout the colonies that were 24–36 h old (Fig. 2 C). Beginning at ∼40 h, the colony became stratified with the upper cells becoming mostly stationary while the rest of the colony contained dividing cells (Fig. 2 D). In colonies that were 3 d old, the surface layers of the ridge and central plateau mostly consisted of stationary cells. The internal layers consisted of sporadically or slowly dividing, but still relatively young, cells, whereas only the ridge interior and the root bases and tips still contained actively dividing cells (Fig. 1, C and D). These observations demonstrate that clearly demarcated zones of dividing and nondividing cells are discernible within the structured colonies (Fig. 1 F).

Unlike the smooth laboratory strain colony (Váchová et al., 2009), the biofilm colony was not covered by a protective cell monolayer (unpublished data). However, we did identify a distinct property of the surface cell layers that was important for protecting the population as a whole. Surface cells (visualized by ConA-AF [ConA conjugated with Alexa Fluor 488] cell wall staining in vertical transverse colony cross sections) could not be stained with Nile red (NR), which targets lipid granules and membranes (Greenspan and Fowler, 1985). This NR-free cell layer was present throughout the BR-F colony (Fig. 3, A and B), including the pseudohyphae tips (Fig. 3 C). This layer was thick (17.8 ± 2.5 µm) in colonies 24–36 h old, persisted over the next 2 d, became thinner in the aerial part (11.5 ± 1.2 µm in a colony 3 d old), and, with the exception of the root tips, disappeared in older colonies (7 d old). After formaldehyde fixation or treatment with NaN₃ causing rapid cellular ATP depletion, entire colonies could be stained with NR (Fig. 3 D). This demonstrated that lipid particles are present in untreated colonies, but NR is probably removed from the surface cells. We considered MDR pumps to be prime candidates for this process. Indeed, the disruption of the PDR1 gene, encoding a transcription activator of some MDR transporters (Balzi et al., 1987; Fardeau et al., 2007), significantly diminished the removal of NR (Fig. 3 D). Deleting various combinations of genes coding for individual MDR pumps (Snq2p [Servos et al., 1993], Pdr5p [Balzi et al., 1994], and the latter’s close relatives Pdr10p and Pdr15p [Wolfger et al., 1997]; Fig. 3 H and Table I) demonstrated that Pdr5p and Snq2p are essential for the extrusion of NR from the surface cells of colonies, as dye export was completely blocked in the BR-F–pdr5snq2 strain (Fig. 3 D). Moreover, we demonstrated that the Pdr5p-GFP transporter was mainly present in the plasma membrane of cells located at the surface of BR-F–Pdr5p-GFP colonies that were 1.5 d old. Meanwhile, internal cells exhibited mostly vacuolar GFP fluorescence, which is indicative of Pdr5p degradation (Fig. 3 F). Deletions of other potential MDR regulator genes, PDR3, STB5, and YAP1 (Jungwirth and Kuchler, 2006), in the BR-F–pdr1 strain had no effect on Pdr5p/Snq2p function (Fig. 3 H). This indicates that the expression of PDR5 and SNQ2 in the surface layer of BR-F colonies is regulated by the Pdr1p transcription factor, probably jointly with an additional, yet unidentified, MDR gene regulator (Fig. 3 E). Together, these findings imply an important protective role of the surface cells equipped with Pdr5p, Snq2p, and possibly other MDR exporters in the removal of external toxic compounds (Fig. 3 G).

In contrast to the tightly packed cells within smooth colonies, pores are present within structured colonies (Kuthan et al., 2003) that are similar to those allowing nutrient flow in biofilms (Douglas, 2003). To study nutrient flow within the biofilm colony, we set up detection systems based on the expression of GFP under the control of a regulatable promoter, P_GAL1 or P_CUP1, in the BR-F strain (Table I). 45 min after feeding the colonies of such strains with the inductor (galactose or copper ions) from the agar side, GFP fluorescence was detected not only in the tips of the pseudohyphae (i.e., closest to the inductor) but also in all of the cells in the surface layer, implying a very efficient propagation of the inductor within the colony. Surprisingly, a majority of the root cells and internal parts of the ridge remained uninduced even after 5 h of induction (Fig. 4 A). However, all colony cells were induced when the exposed vertical transverse cross section of the colony was placed flat on agar soaked with the inductor (Fig. 4 C). Therefore, we hypothesize that the uninduced part of the colony produces an ECM (different from the velcrolike fibers that are present throughout the entire colony), completely blocking the penetration of certain chemical species, even small ones, to the inner parts of the intact colony. The low-permeable ECM begins to be produced in the central area of young (33–35 h old) colonies, and the area expands with the colony growth (Fig. 4 B).

In contrast to the zones of stationary cells that appear in the air-facing aerial part of the colony, the ECM prevents the penetration of chemicals from both the aerial and subsurface colony parts. The existence of the ECM is supported by the earlier scanning EM observation of an abundant extracellular material containing a sugar component (Kuthan et al., 2003). A microscopically similar ECM of C. albicans species biofilm (Al-Fattani and Douglas, 2006) and the extracellular glucan of C. albicans biofilm sequestering antifungals (Nett et al., 2010) both contribute to the biofilm drug resistance.

These developmental principles, as revealed in this study, provide new insights into the differentiation of a biofilm colony and the function of specialized cell subpopulations, and they suggest the presence of unique mechanisms of population protection (Video 2). What are the colony strategies? A colony arising from a single cell grows very quickly, as most cells efficiently divide. In contrast to a smooth laboratory strain colony, biofilm colony also expands substantially in the vertical direction. This expansion may be enabled by the velcrolike interconnection of its cells. Initially, the colony population protects itself from external chemical threats by inducing MDR exporters (Fig. S2) capable of removing toxic compounds (Sipos and Kuchler, 2006). Later, the activity of the exporters persists exclusively in the surface cell layers over the entire colony, and, in parallel, additional protection strategies are initiated. The upper cell layers of the aerial part of the colony become stationary and thus more resistant to chemical and other threats. As the nutrients from agar are efficiently transported, the formation of such nondividing cells appears to be not simply the result of nutrient exhaustion but, more likely, a regulated process that helps to protect the colony surface that is directly exposed to the open air. In parallel, the internal cells near the agar begin to produce the ECM.

The preservation of the velcrolike joints then contributes to the mechanical stability of the expanding colony and may lend flexibility to the layer, forming aerial wrinkles with internal cavities. Subsequent cell generations formed by the dividing inner cells of that layer are thus well protected. The cells in the inner bottom part of the ridge and the pseudohyphae in the subsurface colony regions do not enter a stationary phase. Rather, they continue to produce the ECM that is impermeable to some small compounds such as galactose and to harmful chemicals such as copper ions. Only the pseudohyphae tips protrude from the ECM, but these are still protected by the MDR exporters. The tips may function as the sensors of nutrients and other environmental stimuli important to the colony. The questions remain as to what the chemical nature of the ECM is and how the embedded cells access the nutrients that are essential for their growth. It was previously shown that the ECMs of various microorganisms function as sorptive sponges that sequester organic molecules close to the cells (Decho, 2000) and that they also bind and sequester drugs (Nett et al., 2010). Yeast ECMs vary by their content of different polymeric and monomeric carbohydrates, proteins, and phosphorus (Al-Fattani and Douglas, 2006) and are even preferentially grazed by ciliates (Joubert et al., 2006), suggesting that ECMs have nutritional value. Thus, we hypothesize that in the biofilm colony, the ECM itself may function both as a sequestration barrier and a nutrient pool essential for new cell progeny within the cavities.

In conclusion, the specific architecture of the biofilm colony enables multiple protection strategies (Video 2), yielding a high level of resistance in the wild. Importantly, some of the colony features that we have shown here (e.g., a specific growth pattern, the production of the ECM, and drug efflux pumps) are the traits that are also implicated in the formation of complex fungal biofilms (Blankenship and Mitchell, 2006). The structured yeast colony thus represents a well-defined and powerful in vivo model system that may help to uncover the underlying general principles of microbial biofilm formation.

The wild yeast strain BR-F was supplied from a collection of the Institute of Chemistry (Slovak Academy of Sciences). The temperature-sensitive mutant strain ts104 was obtained from a collection of Charles University in Prague. All other strains were derivatives of the BR-F strain and were prepared in this study (Table I). Colonies were grown on GMA (3% glycerol, 1% yeast extract, and 2% agar) at 28°C unless otherwise indicated.

Strains with gene deletions, C-terminal GFP fusions, and artificial-promoter (P_CUP1 and P_GAL1) constructs replacing the HIS3 gene were prepared according to Gueldener et al. (2002) and Sheff and Thorn (2004) by transforming the cells with DNA cassettes generated by PCR that used the primers and plasmids listed in Tables S1 and S2. The cdc3-1 allele from the ts104 strain was first cloned into the pFA6a-natNT2 vector using BamHI and AscI restriction enzymes. The DNA cassette was then amplified and used to replace the wild-type CDC3 allele in the BR-F–cdc3CDC3 strain. Yeast cells were transformed as described by Gietz and Woods (2002).

Cells in the cultivation medium were examined at RT under a microscope (DMR; Leica) equipped with a 100×/1.3 oil objective and a GFP filter or Nomarski contrast and then photographed with a charge-coupled device camera (ProgRes MFcool; Jenoptik; Figs. 2 [A and B] and S2). Colony images were captured in incident light with a Navitar objective and a complementary metal-oxide semiconductor camera (ProgRes CT3; Jenoptik; Fig. S1 A). Time-lapse experiments were performed using a camera (DS-5M; Nikon) with a Navitar 12× objective and light-emitting diode illumination (Video 1). NIS-Elements software (Laboratory Imaging) was used throughout.

2P-CM (Figs. 1 [A–D], 2 [C and D], 3 [A–D and F], and 4 [A–C]) was performed according to Váchová et al. (2009). In brief, colonies were embedded in low-gelling agarose directly on the plates at RT and cut vertically down the middle. They were placed on the coverslip (the cutting edge to the glass), and the colony side views were obtained by 2P-CM. When required, the cross sections were stained with the following fluorescent dyes: 2.5 µg/ml NR, 30 µg/ml ConA-AF, and 1 µg/ml calcofluor white. Alternatively, GFP fluorescence was monitored. Images were acquired at RT with a true confocal scanner microscope (SP2 AOBS MP; Leica) fitted with a mode-locked laser (Ti:Sapphire Chameleon Ultra; Coherent Inc.) for two-photon excitation and using 20×/0.70 and 63×/1.20 water immersion plan apochromat objectives. Excitation wavelengths of 920 nm were used for ConA-AF, NR, and GFP, and wavelengths of 790 nm were used for calcofluor white. Emission bandwidths were set to 470–540 nm for ConA, 580–750 nm for NR, 480–595 nm for GFP, and 400–550 nm for calcofluor white. An overview of the morphology of colonies and individual cells was obtained simultaneously with green fluorescence as autofluorescence in the 600–740-nm wavelength range. Images of colonies older than 2 d were composed of two or three stitched fields of view. Details in Fig. 1 (B2 and C1) and Fig. 2 (D4) were obtained by composing two images from neighboring fields of view.

Small blocks (∼1 mm³) of colonies embedded in a 2% agarose gel were fixed by glutaraldehyde/potassium permanganate according to Wright (2000) and embedded in PolyBed 812 (Polysciences, Inc.). Using a transmission electron microscope (1011; JEOL Ltd.) at 80 kV, 70-nm ultrathin sections (prepared on a Reichert-Jung ultramicrotome) stained by uranyl acetate and lead citrate were examined. Images (Figs. 1 E and S1 B) were acquired by a digital camera (MegaView III) and AnalySIS software (Olympus).

Fig. S1 shows the differences between parental BR-F and knockout BR-F–flo11 strains. Fig. S2 shows the early production of Pdr5p-GFP. Tables S1 and S2 list primers and plasmids used in this study, respectively. Video 1 shows BR-F colony growth. Video 2 shows flash animations of biofilm colony development and defense strategies.

We thank Dr. R. Pelc for helpful comments on the manuscript and H. Žďárská for technical assistance.

The study was supported by the Grant Agency of the Czech Republic (grant 204/08/0718 to Z. Palková and L. Váchová) and the Ministry of Education (grant LC531 to Z. Palková and L. Váchová; grant LC06063 to Z. Palková and L. Kubínová; grant MSM0021620858 to Z. Palková and V. Šťovíček; grant AV0Z50200510 to L. Váchová and O. Hlaváček; and grant AV0Z50110509 to L. Kubínová). Z. Palková is a Howard Hughes Medical Institute International Research Scholar.