Many Saccharomyces cerevisiae Cell Wall Protein Encoding Genes Are Coregulated by Mss11, but Cellular Adhesion Phenotypes Appear Only Flo Protein Dependent

The outer cell wall of the yeast Saccharomyces cerevisiae serves as the interface with the surrounding environment and directly affects cell−cell and cell−surface interactions. Many of these interactions are facilitated by specific adhesins that belong to the Flo protein family. Flo mannoproteins have been implicated in phenotypes such as flocculation, substrate adhesion, biofilm formation, and pseudohyphal growth. Genetic data strongly suggest that individual Flo proteins are responsible for many specific cellular adhesion phenotypes. However, it remains unclear whether such phenotypes are determined solely by the nature of the expressed FLO genes or rather as the result of a combination of FLO gene expression and other cell wall properties and cell wall proteins. Mss11 has been shown to be a central element of FLO1 and FLO11 gene regulation and acts together with the cAMP-PKA-dependent transcription factor Flo8. Here we use genome-wide transcription analysis to identify genes that are directly or indirectly regulated by Mss11. Interestingly, many of these genes encode cell wall mannoproteins, in particular, members of the TIR and DAN families. To examine whether these genes play a role in the adhesion properties associated with Mss11 expression, we assessed deletion mutants of these genes in wild-type and flo11Δ genetic backgrounds. This analysis shows that only FLO genes, in particular FLO1/10/11, appear to significantly impact on such phenotypes. Thus adhesion-related phenotypes are primarily dependent on the balance of FLO gene expression.

showed that an increase in repeat length can be directly correlated with the increase in Flo1-dependent phenotypes such as flocculation and plastic adherence. FLO gene expression has also been correlated with changes in the general physical-chemical properties of the cell wall. For instance, the overexpression of individual FLO gene family members strongly and differentially impacts on cell wall hydrophobicity (Govender et al. 2008(Govender et al. , 2010. However, information regarding the regulation of genes responsible for cell-wall dependent phenotypes remains limited, with the exception of FLO11 and to a lesser degree FLO1 (Chen and Thorner 2007). FLO11 is the only nonsubtelomeric FLO family member and thus is not subjected to telomere silencing. However, the gene has been shown to be under epigenetic control (Halme et al. 2004;Octavio et al. 2009). Flo11 is required for and/or contributes to the formation of pseudohyphae (Lambrechts et al. 1996;Dranginis 1996, 1998), "flor" formation (Ishigami et al. 2006), "mat" formation (also referred to as "biofilm formation" or "yeast sliding motility") (Reynolds and Fink 2001), as well as flocculation in S. cerevisiae var. diastaticus (Bayly et al. 2005). Although increased expression of other adhesin encoding genes can compensate for the absence of Flo11, as has been shown for Figure 2 and FLO10, whose overexpression can support pseudohyphal development in yeast carrying a FLO11 deletion (Guo et al. 2000), the biological relevance of such artificially generated phenotypes remains uncertain. FLO1 encodes a dominant flocculation factor and appears to be exclusively required for cell-cell adhesion (Goossens and Willaert 2010;Goossens et al. 2011). Overexpression of the FLO1 homologs FLO5 and FLO9 induces flocculation of a broadly similar nature to that of FLO1 but at different levels of intensity that appears largely strain-dependent (Govender et al. 2008). Although the expression of specific adhesins, or alleles thereof, leads to different phenotypic outcomes, it remains unclear how adhesion-encoding genes are differentially regulated to facilitate specific phenotypic outcomes that would be appropriate in specific environmental conditions. Importantly, it also remains to be clarified whether the expression of other proteins that may be coregulated with the adhesins is contributing to specific cell wall2 related phenotypes.
Mss11 performs a central role in the regulatory mechanisms by controlling FLO11 and FLO1 expression (Bester et al. 2006;Van Dyk et al. 2005), and Flo11-dependent phenotypes are all Mss11-dependent (Barrales et al. 2008;Gagiano et al. 1999b). Here, we identify other genes whose transcription is significantly altered in strains overexpressing or carrying deletions of MSS11. For this purpose, two commonly used and phenotypically diverging laboratory strains, S288c and S1278b, were investigated by means of whole transcriptome analysis. S1278b is generally used to study the formation of pseudohyphae and the ability of yeast to grow invasively into agar containing media. S288c, on the other hand, is the most commonly used laboratory yeast but is unable to form pseudohyphae or grow invasively because of a nonsense point mutation (flo8-1) in another transcriptional activator of FLO1 and FLO11, Flo8 (Liu et al. 1996). Restoration of the genomic copy of FLO8 leads to the reestablishment of both flocculation and invasive growth in this strain (Bester et al. 2006;Liu et al. 1996).
Our analysis shows that most of the genes identified as being strongly affected by changed concentrations of Mss11 in both S288c and S1278b genetic backgrounds encode cell wall mannoproteins, suggesting that Mss11 is primarily involved in the modulation of cell wall properties. However, a genetic analysis suggests that none of these genes appears to contribute to the phenotypes that depend on FLO gene expression. Furthermore, our data show that some of the genes that are up-regulated in response to increased expression of MSS11 in fact respond to the increased expression of FLO11 observed in such strains and are probably not direct targets of Mss11.

MATERIALS AND METHODS
Plasmids, strains, media, and culture conditions Plasmids and S. cerevisiae strains used in this study are listed in supporting information, Table S1 and Table S2, respectively. All strains are isogenic to either the S288c or S1278b genetic backgrounds. FLO8 replacement in strains carrying the flo8-1 allele was performed as described previously (Bester et al. 2006). By using genomic DNA isolated from the corresponding European Saccharomyces cerevisiae Archive for Functional Analysis (EUROSCARF) gene deletion library strains as a template, we amplified gene deletion cassettes containing the KanMX4 selection marker via polymerase chain reaction (PCR) with the primers listed in Table S3. These deletion cassettes were subsequently used to generate deletions in the BY4742 flo8-1D::FLO8-LEU and S1278b genetic backgrounds. Yeast transformations were performed according to the lithium acetate method (Ausubel 2004). Yeast cultures were grown at 30°except for the assessment of "mat" formation (see Mat formation). Yeast peptone dextrose (YPD) was used as rich media. Minimal media contained 0.67% yeast nitrogen base with preadded ammonium sulfate but without amino acids supplemented with 2% glucose (w/v) and the required amino acids (SCD media) according to the auxotrophic growth requirements of the relevant strain. Low nitrogen (SLAD) media was prepared similar to SCD except that 0.17% yeast nitrogen base without amino acids or ammonium sulfate was used with the addition of ammonium sulfate to a final concentration of 50 mM. Selection for the KanMX4 marker was performed by supplementing media with 200 mg/L Geneticin (G418; Sigma-Aldrich, South Africa).
Preparation of yeast total RNA Yeast cultures were grown in 5 ml of SCD media from an optical density of 0.1 to between 1 and 2 as determined by spectrophotometric absorbance at a wavelength of 600 nm. Cells were harvested, washed with H 2 O, and resuspended in an ice-cold buffer containing 50 mM sodium acetate, 10 mM EDTA at a pH of 5.0. Total RNA was extracted as described previously (Schmitt et al. 1990). For transcript analysis, total RNA from three independent biological repeats was analyzed.
Microarray analysis, data normalization, and differential expression Probe preparation and hybridization to Genechip microarrays (Affymetrix; Santa Clara, CA) were performed according to Affymetrix instructions, starting with 6 mg of total RNA extracts. Results for each strain were derived from three independent culture replicates. Quality of total RNA, cDNA, cRNA, and fragmented cRNA were analyzed by use of the Agilent Bioanalyzer 2100. Probe hybridization to GeneChip Yeast Genome 2.0 Arrays was performed on the integrated Affymetrix GeneChip 3000 platform. Chip scanning and data collection were performed with the Affymetrix GeneChip Operating Software (GCOS) version 1.4. (http://www.affymetrix.com/support/ technical/manuels.affx). Data sets are available from the Gene Expression Omnibus web site under the series records GSE17716 and GSE29371. The microarray data were background corrected and normalized with RMA (Irizarry et al. 2003) and the resultant log 2 transformed intensity values compared for each of the overexpression or deletion strains with their respective wild-type strains. Determination of differential gene expression between strains was conducted by the creation of an R script by use of the linear fitting and empirical Bayes methods of limma (Smyth 2004(Smyth , 2005. Significant differential gene expression was determined by the use of a threshold for a Benjamini and Hochberg corrected P value , 0.05. Differential expression is reported as log fold changes. Gene ontology enrichment analysis Differentially expressed probe sets were analyzed for Enrichment of Gene Ontology Terms by GOEast with default settings (Zheng and Wang 2008), including a calculated (Benjamini and Yekutieli 2001) FDR threshold of 0.1.
Quantitative real-time PCR (qPCR) analysis DNA contamination in total RNA samples was eliminated by DNase I (Roche Diagnostics, Indianapolis, IN) treatment. One microgram of total RNA was used as template for cDNA synthesis with the ImProm-II reverse transcription system according to the manufacturer instructions (Promega, Madison, WI). cDNA samples were diluted 50 times with H 2 O before qPCR analysis. Primers and hydrolysis probes used for detection and quantification of cDNA were designed with Primer Express ver. 3 (Applied Biosystems, Carlsbad, CA) and are listed in Table S4. Detection reagents were purchased from Applied Biosystems and Kapa Biosystems (Cape Town, South Africa). qPCR runs and collection of spectral data were performed with a 7500 cycler (Applied Biosystems). Except for cDNA corresponding to transcripts of FLO1, FLO5, and FLO9, amplicon formation was monitored with SYBR Green fluorescence with individual primer concentration of 100 nM. Specific labeled hydrolysis probes (Taqman) and primers were designed to differentiate between the cDNA species corresponding to the highly homologous FLO1, FLO5, and FLO9 genes. Hydrolysis probes were modified by the addition of a 39 minor groove binding protein and nonfluorescent quencher, as well as the 59 attachment of fluorescent dyes as described before for the FLO1 and FLO5 specific hydrolysis probes and primer sets (Govender et al. 2008).
The hydrolysis probe (Applied Biosystems) and primer set used for FLO9 cDNA detection are listed in Table S4. Hydrolysis probe and primer concentrations were kept at 250 nM and 900 nM, respectively, for reactions containing probe primer combinations. Cycling conditions during qPCR were as follows: 50°for 2 min, 95°for 10 min, 40 cycles of 95°for 15 sec, followed by 60°for 1 min. When we used SYBR Green for amplicon quantification, a dissociation curve analysis was included after the cycling program to verify amplicon authenticity. Preliminary data analyses were performed with Signal Detection Software, ver 1.3.1. (Applied Biosystems). Individual qPCR reaction runs were performed at least in duplicate. The relative expression value for each sample was defined as 2 -Ct (target) , where Ct (target) represents the cycle number at which a sample reaches a predetermined threshold signal value for the specific target gene. Relative expression data were normalized to the relative expression value of the reference gene PDA1 (Wenzel et al. 1995) in each respective sample, thus giving normalized relative expression for a target gene as 2 -Ct (target) /2 -Ct (PDA1) . Fold change was calculated by log 2 -converting the data followed by subtracting the value for the reference condition/strain.

Flocculation and hydrophobicity assay
Ca 2+ -dependent flocculation of yeast cultures was determined by a method based on the Helm's sedimentation test (Bester et al. 2006). Yeast hydrophobicity was measured by assaying the partition-ing of yeast cells between an aqueous and hydrophobic hydrocarbon phase after vigorous mixing (Rosenberg 2006). Yeast cultures were deflocculated by the addition of ethylene diamine tetra-acetic acid (EDTA), after which the spectrophotometric absorbance was determined at a wavelength of 600 nm (measurement A) as described in the flocculation protocol. A total of 1 ml of yeast culture was transferred to a microcentrifuge tube, washed, and resuspended in phosphate, urea, magnesium buffer consisting of 127.45 mM K 2 HPO 4 , 53.35 mM KH 2 PO 4 , 30 mM urea, and 0.8 mM MgSO 4 (Hinchcliffe et al. 1985). Finally 100 ml of p-Xylene (1,4-dimethylbenzene) was added. Samples were vortex-mixed vigorously for 30 s and left to stand for 15 min, whereupon the spectrophotometric absorbance of the aqueous phase was determined at a wavelength of 600 nm (measurement B). The hydrophobicity index (HI) was defined as 1 2 (B/A), where greater values reflect a yeast population of an increased hydrophobic nature.

Invasive growth determination
To investigate the ability of yeast cultures to grow invasively into agarcontaining medium, 10 ml of yeast suspensions grown overnight to stationary phase were deposited on 2% agar plates with various media composition as indicated for each specific experiment. Flocs in flocculating cultures were disrupted by repetitive pipetting, and a sample was immediately removed and the OD 600 determined as described previously. Cultures were adjusted so as to contain the same concentration of cells, washed with water, and spotted on plates. Spotted macrocolonies from flocculating cultures have a granular appearance because of cells that reform flocs on the plate after spotting. After allowing for yeast growth at 30°for times depending on specific experiments as indicated, we washed cells off of the agar surface by vigorous rubbing with a gloved finger under running water, revealing only those cells that have grown into the medium.

Mat formation
The ability of yeast strains to form spreading growth mats (also referred to as "biofilm" formation or "sliding motility") on plates was determined as described previously (Reynolds and Fink 2001). In brief, 10 ml of a yeast suspension grown overnight in liquid media as described previously was deposited in the center of an YPD plate containing 0.3% w/v agar and incubated at room temperature (20225°). "Mat" formation was monitored by measuring the diameter of growth of at least three independent biological repeats. Measurements were always taken by use of the same reference point on the plate.

Polystyrene adherence assay
To measure the ability of yeast cells to adhere to polystyrene plastic surfaces, liquid cultures (100 ml) were incubated at room temperature in flat bottom polystyrene 96-well plates (Sterilin). After incubation (~2 hr) an equal volume of a solution of 1% (w/v) crystal violet was added to the cells followed by further incubation for 15 min at room temperature. The wells were repeatedly washed with H 2 O, leaving only stained cells remaining attached inside the wells. Sodium dodecyl sulfate was added to the wells to desorb the crystal violet from the cells and increase the visibility of attachment (Reynolds and Fink 2001).

RESULTS
Altered MSS11 expression affects transcription in S288c and S1278b To discover novel targets of Mss11, genome-wide expression levels of strains with modified MSS11 expression were monitored through DNA microarray analysis. These strains included the S1278b and S288C wild-type strains transformed with the multicopy shuttle vector YEpLac195 containing MSS11 (2m-MSS11) as well as the S1278b MSS11 deletion strain (mss11). The same strains transformed with the YEpLac195 (2m) without insert were used as controls. Total RNA for transcriptome analysis was isolated from three biological replications of transformants grown to the mid-exponential growth phase.
As expected no MSS11 transcript was detected in S1278b mss11D:: LEU2 (data not shown), whereas strains with 2m-MSS11 displayed a 3.2-and 4.3-fold up-regulation in S288c and S1278b respectively. This finding is in agreement with previous findings showing that this multicopy expression system results in the up-regulation of Mss11specific targets such as FLO1 and FLO11 (Bester et al. 2006;Gagiano et al. 2003;Van Dyk et al. 2005).
In S1278b ECM34, FLO11, HMS1, and TIR3 were found to be induced upon MSS11 overexpression and repressed in an MSS11 deletion strain ( Figure 1B). HMS1 was previously identified as a regulator of pseudohyphae formation (Lorenz and Heitman 1998), and ECM34 is an uncharacterized gene with suspected cell wall function (Lussier et al. 1997).

Common gene targets of both FLO11 and MSS11 overexpression
To determine to what extent high levels of Flo11 may account for the effects observed in the MSS11 overexpressing strains, we performed a microarray analysis of strain S1278b overexpressing FLO11 by using the constitutive strong promoter from PGK1 and with wild-type S1278b as reference. The experimental conditions were similar as for the MSS11 overexpression analysis, and in this manner 139 genes were found to be significantly regulated in response to FLO11 overexpression. GO enrichment analysis (Table S7) indicates that these genes are predominantly involved in metabolic functions. Five of these genes were also and similarly affected by upon MSS11 overexpression ( Figure  1, C and D) and are involved in metabolic or mitochondrial functions. These genes are NCA3, HMS1, ISF1, NCE103, and as expected FLO11. NCE103 encodes the only carbonic anhydrase for yeast (Amoroso et al. 2005). The MSS11 and FLO11 data sets display unique levels of variation across their respective biological repeats. As a consequence, more genes appear to be affected in a statistically significant way in the FLO11 overexpressing strain than in the MSS11 overexpressing strain, although FLO11 levels are similarly up-regulated in both strains.

qPCR gene expression analysis
To confirm the data, DAN1, FLO1/5/9/11, and TIR1/2/3/4 were analyzed by means of quantitative real-time PCR (qPCR). In addition DAN4, FIG1, FIG2, and FLO10 were also included in this analysis to explore potential functions of other members of these gene families.
qPCR analysis shows that FLO5 and FLO9 are not regulated as the microarray analysis suggests (Figure 2). This difference is likely because of the nature of the Affymetrix Genechip probe sets, which cannot efficiently differentiate between FLO1, FLO5, and FLO9 signals because of very high sequence homologies. FLO1, FLO11, and NCA3 up-regulation in the MSS11 overexpression S288c and S1278b strains are confirmed, as well as FLO11 repression in response to MSS11 deletion. FLO10 and all TIR members follow the same expression pattern as FLO11, albeit with lower magnitude. As for DAN1, a signal with substantial variation between biological repeats, was detected in overexpression strains but not in the reference or deletion strain (Figure S1), suggesting that MSS11 overexpression does impact on this gene. The data furthermore show that DAN4 is induced in the S1278b MSS11 overexpressing strain.
In the S288c genetic background, FIG1 and FIG2 are the only genes down-regulated in response to MSS11 overexpression. Both of these genes are important for mating (Aguilar et al. 2007;Erdman et al. 1998;Muller et al. 2003;Zhang et al. 2002), suggesting a possible function of Mss11 in reducing mating while increasing other adhesionrelated phenotypes.
We further extended the qPCR analysis to compare the MSS11 overexpressing strain with a S1278b strain constitutively overexpressing FLO11 (Figure 3). FLO11 is induced to a very similar degree (4.7and 4.1-fold) in the FLO11 and the MSS11 overexpressing strains, but none of the cell wall2associated genes included in this analysis was significantly affected in the strain overexpressing FLO11.
NCA3, the only non-cell wall protein assessed here, is strongly induced in response to high levels of FLO11 expression, similar to what is observed in the Mss11 overexpressing strain. We also included AQY2 in the analysis because it has been shown to be regulated in a manner similar to FLO11 (Furukawa et al. 2009), but our data suggest a 1.4-fold reduction in the expression of this gene in response to FLO11 overexpression.

Adhesion phenotypes of transformants
Phenotypes dependent on cell wall adhesins were assessed for S1278b, S288c, and S288c carrying a reconstituted copy of FLO8 (see Figure  S2), as well as for the transformants used for the transcriptome analysis (Figure 4). Phenotypes monitored included "mat" formation (Reynolds and Fink 2001) (Figure 4, A Figure S2E). Although S288c wildtype is unable to undergo "mat" formation or flocculate, these phenotypes are restored in the strain with a functional copy of FLO8. "Mat" formation by S288c (FLO8) appears, however, uniquely different from that formed by S1278b, and the strain only forms fully developed growth "mats" after an extended incubation of 3 weeks with extensive morphological variation between biological repeats. In addition, FLO8 replacement increases cell wall hydrophobicity and ability to adhere to polystyrene surfaces. Irrespective of genetic background, an abolishment of all the aforementioned phenotypes is observed upon MSS11 deletion.
In accordance with previous reports (Barrales et al. 2008;Bester et al. 2006;Gagiano et al. 1999b), Mss11 was shown to be absolutely required for flocculation and agar invasion and affects cell wall hydrophobicity. We further show that Mss11 is required for "mat" formation as strain S1278b mss11D is unable to form this specific growth form as displayed by wild type (Figure 4, A and B). The same deletion strain cannot invade agar plates ( Figure 4C) and shows a decrease in cell wall hydrophobicity ( Figure 4D). Remarkably, S1278b displays a low level of flocculation (~5%) under these growth conditions (Figure 4E). MSS11 overexpression restores invasive capability and flocculation in S288c and leads to increased invasion and floc formation (~12%) in S1278b. Both MSS11 overexpressing strains display increased cell hydrophobicity. Interestingly, MSS11 overexpression could not suppress flo8-1 in S288c with regard to the lack of "mat" formation, even after extending the incubation period (data not shown). Furthermore MSS11 overexpression led to "mats" of a smaller diameter in comparison to wild type S1278b. The overexpression of either FLO11 or MSS11 in S1278b results in an increase of invasion, cell wall hydrophobicity, and low levels of flocculation ( Figure S3). Thus both strategies result in similar adhesion phenotypes.
Adhesion phenotype screen of single-and their corresponding flo11D double-deletion strains To assess possible roles of the other cell-wall protein encoding genes that our data show are coregulated with FLO genes, double-deletion strains were constructed in which the deletion of FLO11 was combined with deletions of each of the investigated genes, DAN1, FIG1, FIG2, FLO1, FLO10, NCA3, TIR1, TIR2, TIR3, or TIR4 in both the S1278b and S288c (FLO8) genetic backgrounds. These doubledeletion strains (and the FLO11 single-deletion strain as control) were furthermore transformed with either an empty vector or 2m-MSS11 and spotted on low nitrogen media (SLAD) to investigate invasion.
We previously reported that overexpression of MSS11 in a S1278b flo11D strain led to the reestablishment of some invasive growth. This phenotype was not suppressed in any of the double mutants ( Figure 5). However in S288C (FLO8) flo11D, the deletion of FLO10 was found to abolish agar invasion. Invasion of strain fig1D flo11D is representative of all the other double-deletion mutants. This identifies Flo10 as the only other adhesin required for invasion in the absence of Flo11 in this analysis. Note that the granular nature of the macrocolonies is caused Figure 1 Genes significantly regulated in both of the following strain comparisons. (A) The MSS11 overexpression strains S288c (S288c MSS11; y-axis) and S1278b (Sigma MSS11; xaxis), (B) MSS11 overexpression and deletion in S1278b (Sigma MSS11 and Sigma mss11 on the y-and x-axis respectively), (C) Venn diagram depicting the amount of regulated genes either shared or unique across the overexpression strains, and (D) S1278b either overexpressing FLO11 (Sigma FLO11; y-axis) or MSS11 (Sigma MSS11; x-axis).
by flocculation in the cell suspensions after being dropped on the plate as S288C (FLO8) cells display strong flocculation.
We further tested if single deletions of DAN1, FIG1, FLO1, FLO10, FLO11, MSS11, TIR1, TIR2, TIR3, and TIR4 affect flocculation and cell hydrophobicity in S288c (FLO8) (Figure 6, A and B). Flocculation ability of strains grown to stationary phase was tested in YPD. Strains carrying deletions in FLO1 and MSS11 show no flocculation, thus confirming previous findings (Bester et al. 2006). We further identify the significant requirement of Flo10 as a flocculation factor as FLO10 deletion leads to near total abolishment of flocculation. This requirement was not observed for flo10D strains grown in minimal media (data not shown); thus, this effect is very likely dependent on media composition. FLO11 and TIR4 deletions lead to a small but significant increase in flocculation. The cell wall hydrophobicity of the same strains is shown in Figure 6B. Deletion of FLO11 or MSS11 resulted in more hydrophilic cells supporting previous findings (Barrales et al. 2008) with no significant changes observed for the other deletion strains.
FLO proteins contain internal tandem repeats that display length variation between different strains and directly affect adhesin phenotypes (Verstrepen et al. 2005). Furthermore, yeast progeny from a common parental strain show great variation in flocculation (Smukalla et al. 2008) likely because of variation in the FLO1 internal repeat region. To rule out the possibility that FLO1 repeat variability is responsible for the discrepancies in flocculation observed in Figure 6, we amplified the repeat lengths from genomic DNA as described previously (Verstrepen et al. 2005) from the same strains used for the phenotype assay and found that all strains contained the same length repeat regions as compared with wild-type S288c (FLO8) ( Figure S4).

Differential regulation of FLO10 and FLO11
We further investigated the possibility of FLO1 and FLO10 being regulated similarly. S1278b strains carrying single and double deletions in genes encoding for FLO11 transcriptional control components were analyzed for FLO transcripts by means of qPCR analysis ( Figure  7). Van Dyk et al. (2005) demonstrated that the absence of the Sfl1 repressor leads to the induction of FLO11 transcription. This is blocked by a deletion of FLO8, acting down-stream of the cAMP-PKA pathway but only partially in yeast deleted for STE12 or TEC1, which function downstream of the MAPK pathway. Our analysis confirms these findings for FLO11 regulation and provides novel data for FLO10 using cDNA from the same set of yeast strains (Figure 7). No significant levels of transcript of FLO1 could be detected in wildtype or the deletion strains, supporting previous findings that this gene is silenced in S1278b (Guo et al. 2000). In the wild-type strain, FLO10 transcript is lower compared with FLO11 and the gene appears also repressed by Sfl1. It is partially dependent on the cAMP-PKA pathway, as can be seen from the signal still present in the sfl1D flo8D is indicated in the scale with red representing up-regulation and green down-regulation, respectively. Fold changes falling outside the range of the indicated scale is represented as a numerical value displayed on a saturated color background. As indicated, the left and right panels represent data from the microarray and qPCR analyses respectively. Only microarray data with significant fold changes are shown (see Materials and Methods). No signal was detected for DAN1 in the reference strain as analyzed with qPCR. Color map generated by JColorGrid ver 1.860 (Joachimiak et al. 2006).

Figure 3
Fold change of expression of selected genes in response to the overexpression of either MSS11 or FLO11 in strain S1278b. double mutant. No transcription could be detected in the sfl1D ste12D double mutant, suggesting that FLO10 transcription also requires MAPK signaling. The transcript is still detected in the sfl1D tec1D strain, suggesting very specific roles of these MAPK pathway components in the regulation of FLO10. Transcription levels in sfl1D mss11D were the same as for the sfl1D flo8D strain.

Mss11 affects multiple cell wall genes
In this study we show that the manipulation of MSS11 expression levels has a significant impact on a number of genes encoding for cell wall related proteins. This assessment holds true for two strains that are geno-and phenotypically divergent, suggesting that Mss11 function is indeed more specifically related to cell wall remodeling. The data also clearly indicate that only FLO1, FLO10, and FLO11 appear to be directly involved in the Mss11-controlled adhesion phenotypes that were assessed here, and that the other co-regulated genes have no specific role in such processes. This leaves the question of the specific roles of these proteins unanswered.
Interestingly in strain S288c, although significantly up-regulating many cell wall protein encoding genes, MSS11 overexpression also had a significant repressive impact on the mating-related genes FIG1 and FIG2. It is noteworthy that Fig1may also play a role in mating-unrelated polarized growth because it was shown that a FIG1 transposon insertion mutant displayed decreased filamentation in response to 1-butanol (Lorenz et al. 2000). The data therefore suggest that Mss11 Figure 4 Adhesion phenotype analysis of the S1278b (labeled Sigma) and S288c transformants used in the transcriptome analysis. Wild-type, MSS11 deletion (mss11), and overexpression (MSS11) strains were analyzed as indicated. (A) "Mat" formation on 0.3% YPD agar after 7 days of growth. (B) Measurement of "Mat" growth at day 5, 7, and 12, respectively. (C) Invasive growth of transformants. Transformants were grown in selective media and spotted on YPD plates: Total growth after 6 days incubation (left) and invaded cells revealed following subsequent plate washing (right). Transformants grown to stationary phase in liquid minimal media (SCD) assayed for (D) the degree of culture hydrophobicity and (E) their ability to flocculate. may play a role in directing cellular differentiation toward nonsexual adhesive phenotypes while repressing mating. It is possible that such a function is of some relevance in an evolutionary framework, similar to what has been observed in the case of RME1 (Magwene et al. 2011;Van Dyk et al. 2003). In a nutrient-poor environment, mating may be undesirable even in the presence of mating partners because it certainly represents an energetically demanding and potentially risky exercise in such unfavorable conditions. Such an interpretation is reinforced by the fact that MSS11 overexpression activates the high-affinity hexose transporter genes, HXT2 (3.3 log 2 fc), HXT4 (3.0 log 2 fc), and HXT6/7 (2.5 log 2 fc) in S288c (data not shown). Furthermore Mss11 controls starch use by the induction of glucoamylase-encoding STA genes that are found in some strains of S. cerevisiae (Gagiano et al. 1999a;Webber et al. 1997). Thus, Mss11 may be part of the switch between the mating and the adhesive, invasive or nutrient scavenging growth forms of S. cerevisiae.
Besides the FLO gene family, the TIR genes appear to be the second most MSS11-affected gene family. Indeed, several members of the group clearly and strongly respond to MSS11 expression levels. However, our data do not add significantly to a better understanding of these genes. Indeed, none of the phenotypes investigated here were affected by deletions of these genes in both wild-type and flo11D genetic backgrounds. It is likely that other conditions will need to be investigated to find Mss11-dependent observable phenotypes associated with these genes.
Adhesion phenotypes are dependent on multiple FLO genes Our results show that the magnitude of specific phenotypes depends on more than one adhesin. In the strain S288c (FLO8), Flo1 is the dominant flocculation protein. However, Flo10 clearly plays a role in the process, and the absence of Flo11 leads to enhanced floc formation. Previously, it was reported that a ssn6 strain with elevated expression levels of both FLO1 (Fleming and Pennings 2001) and FLO11 (Conlan and Tzamarias 2001) displays flocculent behavior, suggesting that Flo1 might be dominant over Flo11 (Conlan and Tzamarias 2001). High Mss11 levels similarly result in greater expression of FLO1 and FLO11, but lead to both increased flocculation and invasion (Bester et al. 2006). It would be interesting to conduct controlled FLO1 and FLO11 co-expression and observe the phenotypic outcome, as that might shed some light on specific adhesin dominance and competitive or cooperative interactions between different adhesins. This study shows that agar invasion is dependent on both FLO10 and FLO11 and that Flo11 clearly is the dominant factor required for this behavior. Previous work has highlighted the level of functional overlap between Flo proteins by means of controlled or over-expression studies (Govender et al. 2008;Guo et al. 2000;Van Mulders et al. 2009). Results from this study strongly suggest that only FLO family members control cellular adhesion properties.
However, we did not investigate the effect of potential redundancy within gene families. For example, TIR gene family members display varying degrees of sequence homology. Different Tir proteins therefore are likely to show some functional overlap and may be able to complement phenotypes of single gene deletions. Thus, future work should focus on whole gene family deletions to rule out gene complementation. However, FLO11 overexpression did not lead to any changes in TIR gene expression (Figure 3), yet resulted in the full range of adhesion-associated phenotypes associated with this protein.
A specific role for Tir proteins in adhesion phenotypes therefore appears unlikely.

Significance of non-FLO targets of MSS11 overexpression
The specific roles of other cell wall encoding genes that are regulated by Mss11 remain unknown. Our data show that none of the tested genes directly interferes or impacts on FLO11-dependent phenotypes. Flo protein expression, on the other hand, does not appear to directly  impact on the regulation of these genes, further suggesting that these genes have other, as yet-unknown roles that are unrelated to adhesion. Further phenotypic screening of mutant strains as well as strains carrying deletions of whole gene families may lead to some indication of such function.
Our data clearly show that overexpression of FLO11 has little direct impact on other cell wall protein encoding genes. FLO1, TIR1, TIR2, TIR3, and TIR4 are induced specifically in response to high Mss11 levels. In contrast to this NCA3, involved with cellular energy metabolism, is induced in response to both high Flo11 and Mss11 levels. This result suggests that high Flo11 levels may have metabolic impacts, either through sensing pathways responding to cell wall status or by indirectly changing the environment of individual cells from freefloating to being attached to substrates or other cells. However, considering that S1278b displays low levels of flocculation it is less likely that indirect gene activation in response to MSS11 overexpression is the result of floc formation as reported by Smukalla et al. (2008).
We also show that high Flo11 levels clearly repress AQY2 expression. This appears to be in contrast with findings by Furukawa et al. (2009), showing that Aqy2 affects adhesion phenotypes and cell wall characteristics in a similar fashion to Flo11p, and that AQY2 regulation is similar to that of FLO11. Taken together, these findings suggest that there is a biological link between these two factors which results in differential regulation dependent on the specific experimental conditions.
The regulation of FLO10 FLO10, similar to FLO1 and FLO11, displays strong dependency on Mss11, although not to the same degree. Furthermore, FLO10 responds to the same signaling pathways as FLO11. Our data show that FLO10 transcription appears absolutely dependent on MAPK signaling since Ste12 is essential for its expression. Previous studies have shown that Ste12 and Tec1 regulate the FLO11 promoter (Lo and Dranginis 1998;Madhani and Fink 1997) with Ste12 acting as the general MAPK signaling component and Tec1 as specific filamentous growth transcription factor (Bardwell et al. 1998;Madhani and Fink 1997). We show that these factors have different roles and requirements in the regulation of the FLO10 promoter because the sfl1D tec1D mutant displays low levels of FLO10 transcription. Thus FLO10 transcriptional activation requires MAPK signaling but does not depend on the filamentous growth specific MAPK component Tec1.
It is surprising that the co-regulation of FLO genes with many other cell wall encoding genes appears to have no detectable impact on any of the relevant cell wall-dependent phenotypes investigated here. Indeed, no significant influence of any of these genes on the intensity of phenotypes such as flocculation (with the possible exception of TIR4), agar invasion or cell wall hydrophobicity was observed. These data strongly suggest that differential FLO gene regulation, controlled by overlapping pathways, is responsible for the balance of Flo proteins in the cell wall and that this balance is primarily responsible for governing the adhesion properties of the cell.