The zinc cluster proteins Upc2 and Ecm22 promote filamentation in Saccharomyces cerevisiae by sterol biosynthesis‐dependent and ‐independent pathways

The transition between a unicellular yeast form to multicellular filaments is crucial for budding yeast foraging and the pathogenesis of many fungal pathogens such as Candida albicans. Here, we examine the role of the related transcription factors Ecm22 and Upc2 in Saccharomyces cerevisiae filamentation. Overexpression of either ECM22 or UPC2 leads to increased filamentation, whereas cells lacking both ECM22 and UPC2 do not exhibit filamentous growth. Ecm22 and Upc2 positively control the expression of FHN1, NPR1, PRR2 and sterol biosynthesis genes. These genes all play a positive role in filamentous growth, and their expression is upregulated during filamentation in an Ecm22/Upc2‐dependent manner. Furthermore, ergosterol content increases during filamentous growth. UPC2 expression also increases during filamentation and is inhibited by the transcription factors Sut1 and Sut2. The expression of SUT1 and SUT2 in turn is under negative control of the transcription factor Ste12. We suggest that during filamentation Ste12 becomes activated and reduces SUT1/SUT2 expression levels. This would result in increased UPC2 levels and as a consequence to transcriptional activation of FHN1, NPR1, PRR2 and sterol biosynthesis genes. Higher ergosterol levels in combination with the proteins Fhn1, Npr1 and Prr2 would then mediate the transition to filamentous growth.


Introduction
Many fungal species form filaments in response to extracellular stimuli such as nutrient deprivation (Cullen and Sprague, 2012). In the budding yeast Saccharomyces cerevisiae, filamentation can be observed when cells are grown on solid medium with limited nutrients (Cullen and Sprague, 2012). Filamentation in haploid cells is also termed invasive growth and is triggered by the lack of a fermentable carbon source such as glucose (Cullen and Sprague, 2000). In diploids, filamentous growth is also called pseudohyphal growth and can be induced by low nitrogen levels (Gimeno et al., 1992). Under these conditions, round yeast cells become more elongated and do not separate following cytokinesis. Cells also attach to and penetrate the substratum they grow on. Together, these mechanisms allow cells to forage for nutrients. Several signalling cascades are critical for filamentous growth including a mitogen-activated protein kinase (MAPK) pathway, the cAMP-dependent protein kinase A pathway and the target of rapamycin (TOR) pathway (Cullen and Sprague, 2012). These signalling pathways regulate a complex network of transcription factors that includes Flo8, Mga1, Phd1, Sok2, Ste12 and Tec1 (Borneman et al., 2006). These transcription factors alter the gene expression pattern which then drives the transition to filamentous growth.
Sut1 and Sut2 were originally identified as regulators of sterol uptake (Bourot and Karst, 1995;Ness et al., 2001). Under anaerobic conditions, ergosterol, the predominant sterol in yeast, cannot be synthesized because this process requires oxygen, and sterols are therefore imported from the extracellular medium (Jacquier and Schneiter, 2012). In the absence of oxygen, Sut1 upregulates the expression of Aus1 and Dan1, which mediate sterol uptake Alimardani et al., 2004).
Sterol uptake is also regulated by Upc2 and its paralogue Ecm22, which like Sut1 and Sut2, are members of the zinc cluster protein family (Schjerling and Holmberg, 1996;Crowley et al., 1998;Shianna et al., 2001). Like Sut1, Upc2 induces expression of AUS1 and DAN1, and another gene involved in sterol uptake, PDR11, under anaerobic conditions (Abramova et al., 2001;Wilcox et al., 2002). In addition, Upc2 seems to regulate the expression of nearly a third of anaerobically induced genes (Kwast et al., 2002). The role of Ecm22 under anaerobic conditions and sterol import is less clear. However, Ecm22 seems to induce DAN1 expression in the absence of oxygen (Davies and Rine, 2006).
Ecm22 and Upc2 also regulate sterol biosynthesis (Vik and Rine, 2001). Both proteins bind to sterol regulatory elements in the promoter of ergosterol biosynthesis (ERG) genes (Vik and Rine, 2001). Under normal laboratory growth conditions, Ecm22 seems to be the main activator, whereas when sterols are depleted, Ecm22 is replaced by Upc2 (Davies et al., 2005). It was shown that Upc2 acts as a sterol sensor (Marie et al., 2008;Yang et al., 2015). Under sterol-rich conditions, Upc2 is predominantly cytosolic and directly binds to ergosterol. When sterol levels drop, ergosterol dissociates from Upc2, which leads to the translocation of Upc2 to the nucleus where it induces expression of ERG genes.
In this study, we demonstrate that Ecm22 and Upc2 are important regulators of filamentation. In contrast to Sut1 and Sut2, which repress filamentous growth, Ecm22 and Upc2 are activators of filamentation. Ecm22 and Upc2 regulate the expression of PRR2, NPR1, FHN1 and ERG genes, which are all involved in filamentous growth, and upregulated in an Ecm22/Upc2-dependent manner during filamentation. ERG11 expression is also under control of several transcription factors that play a crucial role in filamentation, suggesting that ergosterol biosynthesis is critical for filamentous growth. We further show that UPC2 transcription is regulated by Sut1 and Sut2 and that UPC2 levels increase during filamentation. Thus, zinc cluster proteins not only have overlapping functions in filamentation, they also regulate each other.

Sut2 regulates the expression of Sut1 target genes
We have previously shown that the zinc cluster protein Sut1 regulates filamentous growth (Foster et al., 2013). SUT1 overexpression using a multicopy plasmid and the strong constitutive PMA1 promoter leads to inhibition of haploid invasive growth and diploid pseudohyphal growth (Foster et al., 2013). We therefore tested whether overexpression of SUT2, a paralogue of SUT1, has the same effect. Increased levels of SUT2 indeed led to the inhibition of haploid invasive growth (Fig. 1A), which is consistent with a previous observation (Rützler et al., 2004). Diploid cells overexpressing SUT2 also failed to undergo the transition to filamentous growth (Fig. 1B), suggesting that Sut2 is equally important for filamentation in both cell types. However, for this study we decided to focus on haploid cells.
The transcription factor Sut1 regulates filamentation through its targets GAT2, HAP4, MGA1, MSN4, NCE102, PRR2, RHO3 and RHO5 (Foster et al., 2013). Under optimal growth conditions, Sut1 represses the expression of these genes, whereas under filamentation-inducing conditions, this repression is lifted. Increased expression of the Sut1 targets then contributes to filamentation. Because of the similarity between Sut1 and Sut2, we tested whether Sut2 also acts as a repressor for Sut1 target genes. We have shown before that Sut2 negatively regulates the expression of NCE102, PRR2 and RHO5 (Blanda and Höfken, 2013). SUT2 overexpression also decreased the levels of GAT2, HAP4, MGA1, MSN4 and RHO3 (Fig. 1C). Increasing SUT2 levels did not affect the expression of other genes such as RHO4 (Fig. 1C), indicating that the observed downregulation is specific.
SUT1 expression is negatively regulated by Ste12 (Foster et al., 2013), a key transcription factor for the switch to filamentous growth (Liu et al., 1993;Roberts and Fink, 1994). As a consequence of Ste12 activation during filamentation, SUT1 levels decrease and expression of Sut1 targets increases. SUT2 is regulated in the same way. Overexpression of STE12 reduces SUT2 expression (Fig. 1D). Taken together, Sut1 and Sut2 seem to play the same role in filamentation. They are both negative regulators, they control expression of the same genes, and their expression is regulated by Ste12.

Ecm22 and Upc2 are positive regulators of filamentation
As overexpression of SUT1 and SUT2 leads to inhibition of filamentous growth, we asked whether Ecm22 and Upc2, which are like Sut1 and Sut2 zinc cluster proteins that regulate sterol import (Bourot and Karst, 1995;Schjerling and Holmberg, 1996;Crowley et al., 1998;Ness et al., 2001;Shianna et al., 2001), also control filamentous growth. Rather unexpectedly, overexpression of either ECM22 or UPC2 resulted in much stronger haploid invasive growth compared with the wild type ( Fig. 2A). Thus, Ecm22 and Upc2 are activators of filamentation, unlike Sut1 and Sut2, which function as inhibitors. Expression levels of the filamentation marker FLO11 were also considerably higher in cells overexpressing ECM22, and even more increased in the UPC2 overexpression strain (Fig. 2B). Higher levels of either ECM22 or UPC2 in diploid cells led to a marked increase in pseudohyphal growth (Fig. 2C), indicating that Ecm22 and Upc2 regulate filamentation in a positive manner in haploids and diploids. Nevertheless, for the further characterization of Ecm22 and Upc2, we focused on haploid cells.
Next, it was tested whether the deletion of ECM22 or UPC2 affects invasive growth. No phenotype was observed for single mutants (Fig. 2D). In contrast, simultaneous deletion of ECM22 and UPC2 resulted in a strong defect in invasive growth (Fig. 2D). In line with this observation, expression of the filamentation marker FLO11 was decreased in ecm22Δ upc2Δ cells but not in the corresponding single deletion strains (Fig. 2E). In summary, these data indicate that Ecm22 and Upc2 have an important and redundant role in filamentation.

Identification of target genes of Ecm22 and Upc2 that play a role in filamentation
Sut1, Ecm22, Upc2 and possibly Sut2 seem to control the expression of a similar set of genes for sterol uptake A. SUT2 overexpression results in decreased haploid invasive growth. Haploid wild-type cells (PPY966) carrying either a SUT2 overexpression plasmid (pMC10) or the corresponding empty vector (pNEV-N) were spotted onto a selective medium plate and were grown for 5 days at 30°C. Pictures were taken before (total growth) and after (invasive growth) rinsing with water. B. Cells overexpressing SUT2 have a defect in diploid pseudohyphal growth. Diploid cells (PC344) carrying either an empty vector (pNEV-N) or a SUT2 overexpression plasmid (pMC10) were grown on low-nitrogen SLAD medium for 6 days at 30°C. C. Sut2 negatively regulates the expression of Sut1 target genes. Cells harbored either a SUT2 overexpression plasmid (pMC10) or the corresponding empty vector (pNEV-N) in combination with the lacZ reporter fused to the indicated promoter regions (pMC6, pSH23, pHU36, pTH391, pTH387, pMC7). Shown is the average β-galactosidase activity with standard deviation of four independent cultures. *, P < 0.01 compared with the wild type carrying an empty plasmid. D. Ste12 downregulates the expression of SUT2. SUT2-lacZ expression (pTH415) was determined for the wild-type strain (PPY966) and cells overexpressing STE12 from the GAL1 promoter (THY762). Bars indicate the average with standard deviation of four independent cultures. *, P < 0.01 compared with the wild type. under anaerobic conditions, including AUS1 and DAN1 Wilcox et al., 2002;Alimardani et al., 2004;Davies and Rine, 2006). It is therefore conceivable that Ecm22 and Upc2 also regulate the expression of Sut1/Sut2 target genes for filamentatous growth.
However, levels of GAT2, HAP4, MGA1, MSN4, NCE102, RHO3 and RHO5 in the ecm22Δ upc2Δ double mutant were indistinguishable from the wild type (data not shown). As an example, NCE102 expression was also tested in cells overexpressing either ECM22 or UPC2. A. Overexpression of ECM22 and UPC2 leads to increased haploid invasive growth. The wild type (PPY966) and the sterol import mutant aus1Δ pdr11Δ (SHY68) carrying the indicated plasmids (pNEV-N, pTH408, pMC8) were spotted onto selective medium plates and were grown for 3 days at 30°C. Pictures were taken before (total growth) and after (invasive growth) rinsing with water. This was done early when filamentation just started in the wild type to demonstrate the stronger invasive growth of strains overexpressing ECM22 and UPC2. B. Overexpression of ECM22 and UPC2 leads to increased FLO11 levels. Wild-type cells (PPY966) harboring a plasmid on which lacZ was fused to the FLO11 promoter (pSH23), and carrying the indicated plasmids (pNEV-N, pTH408, pMC8) were grown in selective medium. Shown is the average β-galactosidase activity with standard deviation of four independent cultures. *, P < 0.01 compared with the wild type carrying an empty plasmid. C. Overexpression of ECM22 and UPC2 results in increased diploid pseudohyphal growth. Wild-type cells (PC344) carrying the indicated plasmids (pNEV-N, pTH408, pMC8) were grown on low-nitrogen SLAD medium for 4 days at 30°C. This was done early when filamentation just started in the wild type to demonstrate the stronger pseudohyphal growth of strains overexpressing ECM22 and UPC2. D. Simultaneous deletion of ECM22 and UPC2 results in a defect in haploid invasive growth. The indicated strains (PPY966, MCY19, MCY21, THY760) were spotted onto YPD plates and were grown for 2 days at 30°C. Pictures were taken before (total growth) and after (invasive growth) rinsing with water. E. Deletion of both ECM22 and UPC2 results in decreased FLO11 expression. β-galactosidase activity was determined for the indicated strains (PPY966, MCY19, MCY21, THY760) all carrying a FLO11-lacZ plasmid (pSH213). Bars indicate the average with standard deviation of four independent cultures. *, P < 0.01 compared with the wild type. Again no effect was observed (data not shown). Thus, the expression of GAT2, HAP4, MGA1, MSN4, NCE102, RHO3 and RHO5 is not under the control of Ecm22 and Upc2, and an altered expression of these genes is not the cause of the filamentation phenotypes of the ecm22Δ upc2Δ mutant and the ECM22 and UPC2 overexpression strains. Interestingly, expression of the Sut1/Sut2 target PRR2 is lowered in ecm22Δ upc2Δ cells but not in the corresponding single mutants (Fig. 3A). Furthermore, PRR2 expression is strongly increased in cells overexpressing UPC2 and to a lesser extent in cells overexpressing ECM22 (Fig. 3B), indicating that PRR2 is a target of Upc2 and Ecm22.
We also analyzed the expression of NPR1, a paralogue of PRR2 (Byrne and Wolfe, 2005). Interestingly, expression patterns of NPR1 and PRR2 are quite similar. Reduced NPR1 levels were observed in the ecm22Δ upc2Δ double mutant but not in cells lacking only one gene (Fig. 3C). Furthermore, NPR1 expression is increased in cells overexpressing either ECM22 or UPC2 and reduced in strains overexpressing SUT1 or SUT2 (Fig. 3D). NPR1 and PRR2 are thus the only genes with a potential role in filamentation that are not only regulated by Sut1 and Sut2 but also by Ecm22 and Upc2.
As mentioned above, NCE102 expression is not affected by deletion or overexpression of ECM22 or UPC2 (data not shown). Nevertheless, we analyzed FHN1, the functional homologue of NCE102 (Byrne and Wolfe, 2005;Loibl et al., 2010). ECM22 deletion had no effect on FHN1 expression, whereas UPC2 deletion led to lower FHN1 levels ( Fig. 4A). This was further reduced in a strain lacking both ECM22 and UPC2. FHN1 expression is strongly increased in cells overexpressing ECM22, and even stronger in cells overexpressing UPC2 (Fig. 4B). In contrast, overexpression of either SUT1 or SUT2 had no effect on the expression of FHN1 (Fig. 4B). Thus, FHN1 expression is regulated by Ecm22 and Upc2 but not by Sut1 or Sut2, whereas its paralogue NCE102 is under control of Sut1 and Sut2 but not of Ecm22 and Upc2.
Next, we examined whether the newly identified targets of Ecm22 and Upc2 (PRR2, NPR1 and FHN1) play a role in filamentation. As shown before, PRR2 expression is strongly upregulated during haploid and diploid filamentation (Foster et al., 2013). Furthermore, PRR2 expression A. Deletion of ECM22 and UPC2 leads to decreased PRR2 expression. β-galactosidase activity was determined for the indicated strains (PPY966, MCY19, MCY21, THY760) carrying a PRR2-lacZ plasmid (pHU37). Shown is the average β-galactosidase activity with standard deviation of four independent cultures. *, P < 0.01 compared with the wild type. B. Overexpression of ECM22 and UPC2 leads to increased PRR2 expression levels. Wild-type cells (PPY966) harboring a PRR2-lacZ plasmid (pHU37) in combination with the indicated plasmids (pNEV-N, pTH408, pMC8) were grown in selective medium, and β-galactosidase activity was determined for four independent cultures. *, P < 0.01 compared with the wild type carrying an empty plasmid. C. Deletion of ECM22 and UPC2 results in decreased NPR1 expression. β-galactosidase activity was determined for the indicated strains (PPY966, MCY19, MCY21, THY760 carrying pTH421) (n = 4). *, P < 0.01 compared with the wild type. D. NPR1 expression is regulated by Ecm22, Upc2, Sut1 and Sut2. Cells harbored a NPR1-lacZ plasmid (pTH421) in combination with the indicated vectors (pNEV-N, pTH408, pMC8, pNF1, pMC10). Shown is the average β-galactosidase activity with standard deviation of four independent cultures. *, P < 0.01 compared with the wild type carrying an empty plasmid.
correlates with filamentation phenotypes. PRR2 levels are reduced in strains that have a filamentation defect such as the ecm22Δ upc2Δ double mutant and strains that overexpress either SUT1 or SUT2 (Fig. 3A) (Blanda and Höfken, 2013;Foster et al., 2013). In strains that are hyperfilamentous due to overexpression of either UPC2 or ECM22, PRR2 expression levels are increased (Fig. 3B). Together these data strongly suggest that Prr2 plays an important role in filamentation. However, a PRR2 deletion strain does not display a filamentation defect (Fig. 5A) (Foster et al., 2013). Because PRR2 has a paralogue, NPR1 (Byrne and Wolfe, 2005), it is conceivable that no defect was observed for the prr2Δ strain because both genes have overlapping functions in filamentation. We therefore examined filamentous growth of the npr1Δ prr2Δ double mutant and the npr1Δ mutant. Both strains had an equally strong defect in invasive growth (Fig. 5A), establishing a role for NPR1 in filamentation but not for PRR2. However, cells overexpressing PRR2 exhibited increased invasive growth (Fig. 5B), suggesting that PRR2 like its paralogue NPR1 are involved in filamentous growth.
Deletion of either FHN1 or NCE102 or both genes did not affect filamentous growth (data not shown) (Foster et al., 2013). However, as for PRR2, overexpression of either FHN1 or NCE102 resulted in increased invasive growth (Fig. 5B), suggesting that the corresponding proteins play a positive role in filamentous growth.
Ecm22 and Upc2 control the expression of genes that are involved in ergosterol biosynthesis in the presence of oxygen, and sterol import from the extracellular medium under anaerobic conditions (Crowley et al., 1998;Shianna et al., 2001;Vik and Rine, 2001). It is therefore conceivable that sterol biosynthesis and/or uptake contribute to filamentation. However, as no sterol was added to the medium the cells grow on and penetrate, it is unlikely that invasive growth requires sterol import. Furthermore, an aus1Δ pdr11Δ double mutant, which is unable to import sterols (Wilcox et al., 2002), displays normal invasive growth ( Fig. 2A) (Foster et al., 2013). Finally, the hyperfilamentation phenotype of strains overexpressing either ECM22 or UPC2 is not affected in the sterol uptake mutant aus1Δ pdr11Δ ( Fig. 2A). Together, these data suggest that under the conditions examined here, invasive growth does not require sterol import.
ERG genes are also important targets of Ecm22 and Upc2 (Vik and Rine, 2001;Wilcox et al., 2002). We chose ERG3, ERG11 and NCP1 to analyze the role of ERG genes in Ecm22/Upc2-mediated filamentation. Erg3 and Erg11 directly catalyse steps in the biosynthetic pathway (Kalb et al., 1987;Arthington et al., 1991), whereas Ncp1 transfers electrons to several Erg enzymes (Yoshida,Fig. 4. FHN1 expression is regulated by Ecm22 and Upc2. A. Cells lacking ECM22 and UPC2 have reduced FHN1 levels. Shown is the average β-galactosidase activity with standard deviation of four independent cultures of the indicated strains (PPY966, MCY19, MCY21, THY760 carrying pTH407). *, P < 0.01 compared with the wild type. B. Overexpression of ECM22 and UPC2 leads to increased FHN1 expression. β-galactosidase activity was determined from four independent cultures of the wild-type strain (PPY966) harboring an FHN1-lacZ plasmid (pTH407) and the indicated vectors (pNEV-N, pTH408, pMC8, pNF1, pMC10). *, P < 0.01 compared with the wild type carrying an empty plasmid. A. NPR1 deletion causes a defect in filamentation. The indicated strains (PPY966, SHY4, THY808, THY809) were spotted onto YPD medium and grown at 30°C. After 2 days, pictures were taken before (total growth) and after (invasive growth) rinsing with water. B. Overexpression of PRR2, NCE102 and FHN1 results in stronger invasive growth. Wild-type cells (PPY966) carrying the indicated vectors (pRS426, pTH402, pTH422, pTH401) were spotted onto selective medium plates and incubated for 3 days at 30°C. Pictures were taken before (total growth) and after (invasive growth) rinsing with water. This was done early when filamentation just started in the wild type to demonstrate the stronger invasive growth of strains overexpressing FHN1, NCE102 and PRR2.
1988; Aoyama et al., 1989;Kelly et al., 1995). The expression of ERG3, ERG11 and NCP1 is downregulated in ecm22Δ upc2Δ cells but not in the corresponding single mutants (Fig. 6A), which is consistent with published data (Vik and Rine, 2001;Wilcox et al., 2002). Notably, overexpression of either SUT1 or SUT2 does not affect levels of ERG3, ERG11 or NCP1 (data not shown), suggesting that the expression of these genes is specifically regulated by Ecm22 and Upc2, and not by Sut1 and Sut2. Importantly, ERG3, ERG11 and NCP1 are all required for invasive growth (Fig. 6B), suggesting that sterol biosynthesis plays an important role in filamentation.
Ecm22 and Upc2 are the main regulators of ERG gene expression (Vik and Rine, 2001), and little is known about other transcriptional regulators. However, a global screen for binding sites of the key transcription factors for filamentation Flo8, Mga1, Phd1, Sok2, Ste12 and Tec1 revealed that promoter regions of many ERG genes contain binding sites for these factors (Borneman et al., 2006). To our knowledge, it has not been examined whether these transcription factors actually regulate the expression of ERG genes. As all six transcription factors examined by Borneman et al. (2006) associate with the ERG11 promoter, we further analyzed this link. Using chromatin immunoprecipitation (ChIP), we found that Flo8-3HA expressed from its own promoter binds to the ERG11 promoter (Fig. 6C). Flo8-3HA overexpressed from the GAL1 promoter associated more strongly with the promoter region of ERG11. The ERG11 expression level was increased by 2.1 ± 0.15 in cells overexpressing Flo8-3HA as determined by quantitative real-time polymerase chain reaction (PCR). Thus, there is a clear correlation between Flo8-3HA levels, association of Flo8-3HA with the ERG11 promoter and ERG11 expression. As for FLO8, we also found that overexpression of MGA1, PHD1 and STE12 resulted in increased ERG11 expression A. Deletion of ECM22 and UPC2 results in decreased expression of ERG genes. β-galactosidase activity was determined for the indicated strains (PPY966, MCY19, MCY21, THY760 carrying pTH376, pTH379 or pSH24). Given is the average β-galactosidase activity with standard deviation (n = 4). *, P < 0.01 compared with the wild type. B. ERG genes are required for invasive growth. The indicated strains (PPY966, THY784, THY827, MBY16) were spotted onto YPD plates and grown for 2 days. Pictures were taken before (total growth) and after (invasive growth) rinsing with water. C. Flo8 binds to the ERG11 promoter. Cells expressing FLO8-3HA from the endogenous promoter (THY839), cells expressing 3HA-tagged FLO8 from the GAL1 promoter (THY841), and cells expressing untagged FLO8 from their own promoter (PPY966) were grown in galactose medium and subjected to ChIP. The immunoprecipitates (IP) were tested for the presence of the ERG11 promoter region. As a control for the PCR, cell lysates were tested without any anti-HA precipitation. D. Regulation of ERG11 expression by transcription factors that promote filamentous growth. ERG11-lacZ (pTH379) expression was determined for the wild-type strain (PPY966) and cells overexpressing the indicated transcriptional regulators from the GAL1 promoter (THY768, THY769, THY765, THY771, THY762, THY767). Bars indicate the average with standard deviation of four independent cultures. *, P < 0.01 compared with the wild type. (Fig. 6D). Notably, these strains have been shown to display strongly increased filamentous growth (Foster et al., 2013). Thus, there is a clear correlation between ERG11 expression and filamentation.
As increased levels of SOK2 and TEC1 did not affect ERG11 expression (Fig. 6D), we also analyzed SOK2 and TEC1 deletion strains. ERG11 levels in sok2Δ and tec1Δ mutants were comparable with the wild type (data not shown). Thus, there is no evidence that Sok2 and Tec1 control ERG11 expression, but Flo8, Mga1, Phd1 and Ste12 regulate ERG11 expression in a positive manner. The fact that ERG11 expression is regulated by so many transcription factors that promote filamentation is a further indication that ERG11 and probably other ERG genes play a crucial role in filamentous growth.

Targets of Ecm22 and Upc2 are upregulated during filamentation
The Ecm22/Upc2 target genes examined here are either essential for filamentation (NPR1, ERG3, ERG11 and NCP1) (Figs 5A and 6B) or at least play a positive role in this process (PRR2 and FHN1) (Fig. 5B). It therefore seems likely that their expression increases during filamentous growth. We have previously shown a strong increase of PRR2 expression under filamentationinducing conditions (Foster et al., 2013). The other Ecm22/Upc2 targets ERG3, ERG11, NCP1, FHN1 and NPR1 were also all upregulated during filamentation, in contrast to the control RHO4 (Fig. 7A). This induction is not affected by the deletion of either ECM22 or UPC2 but

Fig. 7. Expression of Ecm22/Upc2 targets increases during filamentation.
A. Expression of Ecm22/Upc2 target genes during filamentous growth. β-galactosidase activity was determined for the indicated genes (pTH376, pTH379, pSH24, pTH407, pTH421, pMC7) in cells (PPY966, MCY19, MCY21, THY760) grown for 14 h at 30°C on minimal medium plates lacking glucose. Cells grown in liquid minimal medium containing glucose served as reference. Shown is the average increase of four independent replicates with standard deviation. *, P < 0.01 compared with the wild type. B. ERG11 expression increases only under conditions that induce filamentation. Wild-type cells (PPY966) carrying an ERG11-lacZ plasmid (pTH379) were either grown in liquid minimal medium with or without glucose, or alternatively cells were grown for 14 h on minimal medium plates with or without glucose. Shown is the average β-galactosidase activity with standard deviation (n = 4). *, P < 0.01 compared with cells grown in high-glucose liquid medium. C. Erg11 and Prr2 protein levels increase during filamentation. Cells expressing either Erg11-9Myc or Prr2-9Myc (THY837, SHY6) were grown in liquid high-glucose minimal medium or on plates lacking glucose. Cells were lyzed and equal amounts were analyzed by immunoblotting using antibodies against the Myc epitope and Cdc11 (loading control). D. Ergosterol levels increase during filamentation. Sterols were extracted from wild type cells (PPY966) grown in liquid minimal medium with 2% glucose or from plates lacking glucose. Ergosterol levels were determined from three independent cultures. *, P < 0.01 compared with cells grown in high-glucose liquid medium. reduced in strains lacking both genes (Fig. 7A). These data suggest that Ecm22 and Upc2 are partly responsible for the upregulation but that other transcription factors are involved as well. The expression of the Ecm22/Upc2 target genes only increased when cells were grown on plates with limited nutrients, as shown here for ERG11 (Fig. 7B). In liquid medium without glucose, and on glucose-rich plates ERG11 was expressed at levels comparable with liquid medium containing glucose (Fig. 7B). Thus, gene expression correlates with filamentous growth that only occurs when cells are grown on solid medium with limited nutrients (Gimeno et al., 1992;Cullen and Sprague, 2000). We next examined whether altered transcription observed here results in changes at protein level. Erg11 levels were significantly higher in cells grown under filamentation-inducing conditions (Fig. 7C). This effect was even more pronounced for Prr2, which was barely or not detectable in liquid cultures with glucose but strongly expressed in cells grown on plates without glucose (Fig. 7C). This correlates well with the five-to sixfold increase of ERG11 expression during filamentation determined by β-galactosidase assays ( Fig. 7A and B), and a 90-fold increase for PRR2 that we observed previously using quantitative real-time PCR (Foster et al., 2013). As a consequence of higher ERG gene expression during filamentous growth, the ergosterol content could also increase. In fact, we observed significantly higher ergosterol levels in cells grown on plates with limited nutrients (Fig. 7D). In summary, targets of Ecm22 and Upc2 are upregulated at transcriptional and protein level during filamentation. This probably results in physiological changes such as higher ergosterol levels.

Regulation of UPC2 expression
Upc2 has been shown to positively regulate its own expression (Abramova et al., 2001;Wilcox et al., 2002). We therefore tested the possibility that Ecm22, Sut1 and Sut2 are also involved in the regulation of UPC2 expression. Overexpression of UPC2 led to increased UPC2 levels (Fig. 8A), confirming UPC2 autoregulation that has been reported before (Abramova et al., 2001;Wilcox et al., 2002). Higher ECM22 levels had no effect on UPC2 expression, whereas overexpression of either SUT1 or SUT2 decreased UPC2 expression (Fig. 8A). Thus, UPC2 expression is positively regulated by Upc2, and in a negative way by Sut1 and Sut2. In contrast, ECM22 expression was not affected in cells overexpressing either ECM22, UPC2, SUT1 or SUT2 (data not shown).
As the expression of SUT1 and SUT2 is regulated by the transcription factor Ste12 (Foster et al., 2013) (Fig. 1D), it is tempting to speculate that Upc2 is indirectly regulated by Ste12. To test this hypothesis, we examined genetic interactions between STE12 and UPC2. Overex-pression of UPC2 rescues the filamentation defect of the STE12 deletion strain (Fig. 8B). This is a highly specific interaction as increased ECM22 levels have no effect (Fig. 8B). This is consistent with the observation that the Ste12 targets Sut1 and Sut2 regulate the expression of UPC2 but not of ECM22 (Fig. 8A). We also found that STE12 overexpression suppresses the filamentation defect of the ecm22Δ upc2Δ double mutant (Fig. 8C), which further strengthens the link between STE12 and UPC2.
Finally, we tested whether levels of ECM22 and UPC2 change during filamentation. The expression of ECM22 did not change under conditions that induce filamentous growth, whereas UPC2 levels increased during filamentation (Fig. 8D). Taken together, these data suggest that regulation of gene expression is an important control mechanism for Upc2 during filamentous growth. In contrast, Ecm22 seems to be regulated by a different unknown mechanism.
The Ecm22/Upc2 targets examined here are all either essential for filamentous growth or play at least an important role in this process. Furthermore, they are upregu-lated during filamentation in an Ecm22/Upc2-dependent manner. Therefore, activation of Ecm22 and/or Upc2 during filamentous growth probably leads to increased expression of their targets FHN1, NPR1, PRR2 and the ERG genes, which in turn promotes filamentation (Fig. 9). Other studies have shown that Upc2 is primarily activated through reduced sterol levels (Davies and Rine, 2006), which can be achieved through inhibition of sterol biosynthesis enzymes. As sterol synthesis requires oxygen, anaerobic conditions also lead to a reduction of sterol and therefore Upc2 activation. It was proposed that in sterolrich conditions sterol directly binds to Upc2 that keeps it inactive in the cytoplasm (Marie et al., 2008;Yang et al., 2015). Dissociation of sterol from Upc2 leads to nuclear translocation of Upc2 and transcriptional activation. Starving conditions that trigger filamentation might also lead to reduced sterol levels. However, here we show that UPC2 overexpression alone is sufficient to upregulate genes involved in filamentation. The observed increase of UPC2 expression during filamentation might therefore also be sufficient for its role in filamentous growth. UPC2 transcription is repressed by Sut1 and Sut2 and positively regulated by its own gene product. Furthermore, expression of SUT1 and SUT2 is inhibited by the transcription factor Ste12 (Foster et al., 2013), which is activated during filamentation (Liu et al., 1993;Roberts and Fink, Fig. 8. Regulation of UPC2 expression. A. UPC2 expression is under control of Sut1 and Sut2. The average β-galactosidase activity of wild-type cells (PPY966) carrying the indicated plasmids (pTH414 in combination with pNEV-N, pTH408, pMC8, pNF1, pMC10) is given with standard deviation (n = 4). *, P < 0.01 compared with the wild type carrying an empty plasmid. B. Overexpression of UPC2 rescues the filamentation defect of the STE12 deletion strain. The wild type (PPY966) and the ste12Δ mutant (THY842) carrying the indicated vectors (pNEV-N, pTH408, pMC8) were spotted onto selective medium plates and incubated for 4 days at 30°C. Pictures were taken before (total growth) and after (invasive growth) rinsing with water. C. STE12 overexpression suppresses the filamentation defect of the ecm22Δ upc2Δ mutant. The indicated strains (PPY966, THY762, THY760, THY826) were spotted on minimal medium supplemented with galactose and raffinose for STE12 overexpression and grown for 3 days at 30°C. D. UPC2 expression increases during filamentation. β-galactosidase activity was determined for the indicated genes (pTH412, pTH414) in wild-type cells (PPY966) grown either for 14 h on minimal medium plates lacking glucose or grown in liquid minimal medium containing glucose. Shown is the average activity with standard deviation (n = 4). *, P < 0.01 compared with the cells grown in liquid high-glucose medium. 1994). We propose a model in which Sut1 and Sut2 partially repress the expression of their targets GAT2, HAP4, MGA1, MSN4, NCE102, NPR1, PRR2, RHO3, RHO5 and UPC2 under optimal growth conditions (Fig. 9A). When cells are grown on a solid medium with limited nutrients, Ste12 is activated, which results in reduced Sut1 and Sut2 levels, and as a consequence, in increased levels of Sut1/ Sut2 targets. Together these targets mediate the transition to filamentous growth. UPC2 levels increase due to autoregulation and the reduced repression by Sut1 and Sut2. This then leads to transcriptional activation of Upc2 target genes. This model is supported by genetic interactions reported here. The filamentation defect of a STE12 deletion strain is rescued by UPC2 overexpression. Increased levels of Upc2 targets, which are downstream of Ste12, are presumably sufficient for this effect. Interestingly, ECM22 overexpression had no effect on the filamentation defect of the ste12Δ mutant that is consistent with other observations. In contrast to UPC2, ECM22 expression does not change during filamentation and is not regulated by the Ste12 targets Sut1 and Sut2. We also observed that STE12 overexpression suppresses the fila- Fig. 9. Model for the regulation of filamentation by zinc cluster proteins. All factors shown represent proteins. Genes are not shown for the sake of simplicity. Activating and inhibitory arrows indicate regulation of expression of the corresponding genes. A. When cells are grown in nutrient-rich liquid medium, Sut1 and Sut2 partially repress the expression of their targets. These include GAT2, HAP4, MGA1, MSN4, NCE102, RHO3 and RHO5, which are only regulated by Sut1 (Foster et al., 2013) and Sut2 (Blanda and Höfken, 2013; this study) but not by Ecm22 and Upc2 (this study). NPR1 and PRR2 are under control of all four transcription factors (this study). Importantly, UPC2 expression is also repressed by Sut1 and Sut2 (this study). B. When cells are grown on solid medium with limited nutrients, Ste12 becomes activated (Liu et al., 1993;Roberts and Fink, 1994) and lowers SUT1 and SUT2 levels (Foster et al., 2013;this study). As a consequence of the loss of repression, expression of the Sut1/Sut2 targets increases and the corresponding gene products contribute to filamentous growth (Foster et al., 2013;this study). UPC2 expression might also increase due to autoregulation (Abramova et al., 2001;Wilcox et al., 2002;this study). Higher Upc2 levels result in increased expression of its targets which include FHN1, NPR1, PRR2 (this study) and ERG genes (Vik and Rine, 2001; this study). The corresponding proteins mediate the transition to filamentous growth (this study). Many targets of Ecm22, Upc2, Sut1 and Sut2 are probably also under control of other transcription factors that promote filamentation. Promoters of many ERG genes have binding sites for Flo8, Mga1, Phd1, Ste12, Sok2 and Tec1 (Borneman et al., 2006;this study). All six transcription factors bind to the promoters of GAT2, HAP4, MGA1, RHO3 and RHO5, and at least one of these transcription factors bind to the promoter regions of PRR2, NCE102 and MSN4 (Borneman et al., 2006;Foster et al., 2013). This suggests that Ecm22, Upc2, Sut1 and Sut1, and their targets are part of an important complex transcriptional network for the induction of filamentation. mentation defect of the ecm22Δ upc2Δ strain. This could be explained by the action of other Ste12 targets that function in parallel to the Upc2 pathway.
GAT2, HAP4, MGA1, MSN4, NCE102, PRR2, RHO3 and RHO5 are not only regulated by Sut1 and Sut2. Their promoter regions also contain binding sites for the transcription factors Flo8, Mga1, Phd1, Sok2, Ste12 and Tec1, which promote filamentation (Borneman et al., 2006;Foster et al., 2013). Likewise, many ERG promoters have a binding site for at least one of these factors (Borneman et al., 2006). All six transcription factors bind to the ERG11 promoter, and we show here that Flo8, Mga1, Phd1 and Ste12 actually control ERG11 expression. It seems very likely that other ERG genes and therefore as a consequence ergosterol biosynthesis are regulated by these transcription factors. This would be a novel and interesting regulatory mechanism for this important metabolic pathway.
What is the function of the Ecm22/Upc2 targets in filamentation? Fhn1, like Nce102, is involved in the formation of a specialized plasma membrane domain termed eisosome (Loibl et al., 2010). This membrane domain could be important for polarized growth during filamentation. Prr2 functions as a mating inhibitor (Burchett et al., 2001). It is not clear how this is relevant for filamentation. The kinase Npr1 stabilizes and activates plasma membranebound nitrogen source transporters when nitrogen is limited (Schmidt et al., 1998;De Craene et al., 2001;Boeckstaens et al., 2014). This includes the ammonium permease Mep2, which also functions as a nitrogen sensor for the transition to filamentous growth (Lorenz and Heitman, 1998;Van Nuland et al., 2006). Npr1 activity is regulated by the TOR pathway. The increase of NPR1 expression that we observed seems to be another regulatory mechanism to allow optimal ammonium transport and sensing during nitrogen limitation.
We not only observed transcriptional activation of ERG genes but also increased ergosterol levels during filamentation. It can only be speculated on the role of ergosterol in filamentation. However, eisosomes are rich in sterol (Grossmann et al., 2007). Fhn1, Nce102 and Erg enzymes might therefore act together to mediate filamentation. Interestingly, an Ecm22/Upc2-mediated change of sterol biosynthesis in response to an external signal has been reported before. ECM22 is downregulated upon hyperosmotic stress (Montañés et al., 2011). This results in reduced ERG gene expression and lower sterol biosynthesis, which seems to be an important adaptation mechanism for hyperosmotic stress.
In Candida albicans, the most common fungal pathogen in humans, filamentation plays important roles in host cell adherence, tissue invasion and virulence (Sudbery, 2011;Gow et al., 2012;Höfken, 2013). It would therefore be interesting to study the role of the C. albicans homo-logues of ECM22, UPC2, SUT1, SUT2 and their targets in filamentation and virulence. UPC2, the sole C. albicans orthologue of budding yeast ECM22 and UPC2, is well studied because of its role in antifungal drug resistance (Silver et al., 2004;MacPherson et al., 2005). Many clinically important antifungals target ergosterol. Azoles inhibit Erg11, which results in ergosterol depletion and the accumulation of toxic sterols (Lupetti et al., 2002). Several gain-of-function mutants of UPC2 have been identified from azole-resistant clinical isolates (Dunkel et al., 2008;Heilmann et al., 2010;Hoot et al., 2011;Flowers et al., 2012). Upc2 hyperactivation leads to ERG11 overexpression, which contributes to azole resistance. Upc2 therefore represents a potential new target for antifungal drugs (Gallo-Ebert et al., 2014). It is not clear whether filamentation in C. albicans is regulated by Upc2 and its targets in a similar way as in budding yeast. Upc2 hyperactivation results in reduced filamentation and virulence (Lohberger et al., 2014), which is not consistent with our model. In contrast, deletion of NCE102, the only C. albicans orthologue of budding yeast FHN1 and NCE102, leads to a defect in filamentation and reduced virulence (Douglas et al., 2013), which is in line with our observations in budding yeast. The role of zinc cluster proteins and their targets in C. albicans filamentation and virulence therefore certainly needs to be further examined.

Yeast strains, plasmids and growth conditions
All yeast strains used in this study are listed in Table 1. The strains are in the Σ1278b background. Yeast strains were constructed using PCR-amplified cassettes (Wach et al., 1997;Longtine et al., 1998;Janke et al., 2004). Yeast strains were grown in 1% yeast extract, 2% peptone, 2% dextrose (YPD) or synthetic complete (SC) medium. Synthetic low ammonium dextrose (SLAD) medium for induction of pseudohyphal growth contains 0.67% yeast nitrogen base without amino acids and without ammonium, 2% glucose and 50 μM (NH 4)2SO4. For induction of the GAL1 promoter, yeast cells were grown in medium with 2% galactose and 3% raffinose instead of glucose. All constructs used in this work are listed in Table 2.

Filamentation assays
For agar invasion assays, 10 5 cells of an overnight culture were spotted on YPD or selective medium, and grown at 30°C. Plates were photographed before and after being rinsed under a stream of deionized water.
For pseudohyphal growth assays, cells were grown overnight, and 100 cells were spread on solid SLAD medium. Plates were incubated at 30°C. Colonies were examined with a Zeiss Axioskop 2 microscope equipped with a 5 × objective and images were captured using a ProgRes C12 camera (Jenoptik).
For protein analysis, β-galactosidase assays and determination of ergosterol, cells were grown to exponential phase in SC medium. Cells were washed with water, and 10 5 cells were plated on SC medium lacking glucose and incubated for 14 h at 30°C. For protein analysis and β-galactosidase assays cells were scraped from one plate. Five plates were required for each measurement of the ergosterol content.

Immunoblotting
One milliliter of cells was harvested by centrifugation and resuspended in 1 ml water. One hundred fifty microliters 1.85 M NaOH was added and incubated for 10 min on ice. After adding 150 μl 55% trichloroacetic acid, the samples were incubated for 10 min on ice. Following 20 min centrifugation 13 000 r.p.m. at 4°C, the supernatant was discarded. The pellet was resuspended in SDS sample buffer (150 mM Tris [pH 8.8], 2% SDS, 10% glycerol, 5% β-mercaptoethanol) and heated for 15 min at 65°C. The samples were then clarified by centrifugation at 13 000 r.p.m. for 1 min. Equal amounts were   Myers et al. (1986) separated by SDS-PAGE, transferred to nitrocellulose and incubated with mouse monoclonal anti-Myc (9E10) from Santa Cruz Biotechnology. To test whether equal amounts of protein were loaded, membranes were stripped after development by incubating membranes in stripping buffer (65 mM Tris [pH 6.8], 2% SDS, 20 mM β-mercaptoethanol) for 40 min at 50°C. After thorough washing with PBS, membranes were incubated with rabbit polyclonal anti-Cdc11 (Santa Cruz Biotechnology) as loading control. Secondary antibodies were from Jackson Research Laboratories.

Ergosterol quantification
Ergosterol levels were determined as described by Arthington-Skaggs et al. (1999), with minor modifications. Briefly, cells were harvested, washed with water and the wet weight was determined. Cells were resuspended in 1.5 ml 25% alcoholic potassium hydroxide solution (25 g KOH and 35 ml water were brought to 100 ml with ethanol) and vortexed for 1 min. Cells suspensions were transferred to borosilicate glass screw-cap tubes and incubated in an 85°C water bath for 1 h. The samples were then allowed to cool down to room temperature, and sterols were extracted with a mixture of 500 μl of water and 1.5 ml of n-heptane followed by vortexing for 3 min. Ergosterol content was determined using a Hitachi U-1900 spectrophotometer and calculated as percentage of the wet weight as described by Arthington-Skaggs et al. (1999).