Abstract
Farnesol, a quorum-sensing molecule, inhibits C. albicans hyphal formation, affects its biofilm formation and dispersal, and impacts its stress response. Several aspects of farnesol’s mechanism of action remain incompletely uncharacterized. Among these are a thorough accounting of the cellular receptors and transporters for farnesol. This work suggests these themes are linked through the Zn cluster transcription factors Tac1 and Znc1, and their induction of the multi-drug efflux pump Cdr1. Specifically, we have demonstrated that Tac1 and Znc1 are functionally activated by farnesol through a mechanism that mimics other means of hyperactivation of Zn cluster transcription factors. This is consistent with our observation that many genes acutely induced by farnesol are dependent on TAC1, ZNC1, or both. A related molecule, 1-dodecanol, invokes a similar TAC1/ZNC1 response, while several other proposed C. albicans quorum sensing molecules do not. TAC1 and ZNC1 both bind to and up-regulate the CDR1 promoter in response to farnesol. Differences in inducer and DNA binding specificity lead to Tac1 and Znc1 having overlapping, but non-identical, regulons. TAC1 and ZNC1 dependent farnesol induction of their target genes was inversely related to the level of CDR1 present in the cell, suggesting a model in which induction of CDR1 by Tac1 and Znc1 leads to an increase in farnesol efflux. Consistent with this premise, our results show that CDR1 expression, and its regulation by TAC1 and ZNC1, facilitates growth in the presence of high farnesol concentrations in C. albicans, and certain strains of its close relative C. dubliniensis.
Introduction
Candida albicans is a major opportunistic human fungal pathogen that can cause life-threatening systemic infections in immune-compromised individuals (1–3). Multiple important C. albicans virulence-related traits, including the morphological switch between yeast and hyphal growth (4, 5), biofilm formation and dispersal (6), interspecies communication with bacteria (7) and response to oxidative stress (8) can be modulated by its quorum sensing molecule (QSM), farnesol, the first identified QSM for eukaryotes (9–14).
Among multiple Candida species, C. albicans has been found to produce the most significant amounts of farnesol, followed by its close relative C. dubliniensis (15, 16). Dense cultures of C. albicans, in certain media, can accumulate farnesol to concentrations as high as 50 μM (15, 16). The known mechanisms underlying the biological activity of farnesol in C. albicans include modulation of signaling pathways such as the Ras1-Cyr1/cAMP-PKA cascade in part via direct inhibition of Cyr1 (17–19). Farnesol exposure also results in a transcriptional response in C. albicans in both sessile and planktonic cells (12, 20–23).
Among the outstanding questions regarding farnesol activity in C. albicans are the existence of specific farnesol receptors and transporters (13). Adenylyl cyclase Cyr1 is a cytoplasmic target of farnesol as it binds and is inhibited by farnesol (18). Transcription factors that directly respond to farnesol as a nuclear receptor/effector to regulate gene expression, however, have not been identified. Growth of C. albicans, in the ‘white’ cell form, is remarkably resistant to growth inhibition by high concentration of farnesol, compared to other fungal species (14, 24). This property might result from an efficient farnesol efflux by certain transporter(s). The ABC (ATP-binding cassette) transporter Cdr1 was found up-regulated upon 2-24 hour farnesol treatment (21, 22) and has been proposed to play a role in farnesol efflux (22). Expression of CDR1 and another ABC transporter CDR2 in C. albicans is regulated by the Zn(II)Cys6 transcription factor Tac1 (25). Gain of function mutations in TAC1 are often found in clinical isolates of C. albicans that are resistant to treatment with azole drugs, due to high levels of CDR1 expression (25–27). Tac1 binds to a 13 base-pair drug-responsive-element (DRE) at the CDR1 and CDR2 promoters, and activates transcription upon acquisition of gain of function mutations, or treatment with certain xenobiotics such as fluphenazine (25, 26, 28). C. albicans TAC1 gene locates in a ‘zinc cluster region’ on Chromosome 5 (25), where it neighbors two other transcription factors from its family, Hal9 and Znc1. Interestingly, Znc1, when activated artificially, also increases CDR1 expression (30)
In this work we investigated whether Tac1 functions as a farnesol nuclear receptor/effector to activate CDR1 expression, and searched for other transcription factors with a similar function. Our work showed that Tac1 and Znc1 contributed individually in and tandem to the transcriptional activation response to farnesol. We also found that CDR1 expression, and its regulation by TAC1 and ZNC1, facilitates growth in the presence of farnesol in both C. albicans and C. dubliniensis.
Results
Farnesol and 1-Dodecanol rapidly induce CDR1 expression
Since specific xenobiotic inducers evoke an acute activation of the C. albicans CDR1 promoter (28, 31), we tested whether the known physiologic inducer of CDR1, farnesol (FOH), also led to a rapid transcriptional induction of the efflux pump gene. FOH addition to exponentially growing cells led to an increase in CDR1 mRNA expression with an amplitude and temporal pattern comparable to fluphenazine (FNZ), a well-studied inducer of CDR1 (Fig. 1A). FOH induces CDR1 expression in a dose dependent manner, starting at concentrations as low as 4 (Fig. 1B). The 12-carbon backbone and hydroxyl group of FOH are required for its full inhibition of C. albicans hyphal growth (9, 32). Several different terpene alcohols and FOH derivatives were tested for their ability to rapidly induce CDR1. Geraniol and farnesyl acetate are unable to induce levels of CDR1 expression comparable to FOH (Fig. 1C). 1-Dodecanol (1-DD), however, another 12-carbon molecule that inhibits hyphal growth (33) induces CDR1 expression at similar concentrations to FOH (Fig. 1C). Tryptophol (Fig. 1C), an aromatic amino acid derived alcohol with fungal quorum sensing activity (34, 35), and tyrosol (Fig. S1), another C. albicans quorum sensing molecule (34), do not induce CDR1. Tracking the expression of a 3xHA tagged CDR1 allele by immunoblot analysis confirmed that FOH and 1-DD also induced Cdr1 at the protein level (Fig. 1D). Cdr1 protein levels, especially in response to FNZ and FOH, appear to stay at the peak induced level longer than them RNA, indicating that Cdr1 is fairly stable.
Tac1 is required for the induction of some, but not all, FOH and 1-DD target genes
Induction of CDR1, CDR2 and RTA3 by xenobiotics, such as FNZ and estradiol, is dependent on the zinc cluster transcription factor Tac1 (25, 28, 30, 36). The observation that CDR1, CDR2 and RTA3 expression was induced by FOH and 1-DD treatment (Fig. 1A, 2A and 2B) suggested Tac1 hyperactivation as a mechanism for FOH and 1-DD induced transcription. Unlike FNZ induction of CDR1, CDR2 and RTA3, which was entirely Tac1 dependent, only FOH and 1-DD induction of CDR2 was entirely dependent on Tac1 (Fig. 2A to 2C). The residual CDR1 induction, and virtually unaffected RTA3 induction, suggests that additional transcription factors respond to FOH and 1-DD at these promoters.
Znc1 contributes to the induction of multiple FOH and 1-DD target genes
The first candidate transcription factor that we tested for Tac1-independent induction of CDR1 and RTA3 by FOH (and 1-DD) was Znc1. Znc1 is a zinc cluster transcription factor that is encoded adjacent to TAC1, and whose sequence bares the greatest similarity to Tac1 of all other members of the C. albicans zinc cluster transcription factor family (Candida Genome Database; (25)). Znc1 was previously identified as a potential regulator of CDR1 and RTA3 in an experiment in which a potent activation domain was fused to the full-length wild type Znc1 (30). RTA3 induction by FOH and 1-DD was decreased in a znc1∆/∆ strain, while RTA3 induction by FNZ was largely unaffected (Fig. 3A). FNZ, FOH and 1-DD induction of CDR1 was largely unaffected in the znc1∆/∆ strain, however FOH and 1-DD induction of CDR1 and RTA3 was decreased in a tac1∆/∆ znc1∆/∆ strain compared to either single mutant (Fig. 3A and 3B). The pattern of Cdr1 protein expression was consistent with the epistasis analysis of CDR1 mRNA expression in the tac1 and znc1 strains (Fig. 3C). FOH induction of CDR1 in the tac1∆/∆ znc1∆/∆ strain was restored by complementation with either TAC1 or ZNC1 (Fig. 3D), while CDR2 induction was only restored upon TAC1 complementation (Fig. 3E).
In addition to CDR1 and RTA3, the previous Znc1-activation domain fusion analysis (30) suggested that several other genes, including orf19.320 and IFD1 (or19.4476), were direct Znc1 targets. Both orf19.320 and IFD1 are induced by FOH and 1-DD, but not by FNZ (Fig. 3F and 3G). Induction of orf19.320 and IFD1 by FOH and 1-DD was ZNC1 dependent, but not affected by TAC1 (Fig. 3H). Consistent with these results, FOH induction of orf19.320 in a tac1∆/∆ znc1∆/∆ strain was restored by re-introduction of ZNC1, but not TAC1 (Fig. 3H).
Hyperactivation of Tac1 and Znc1 by farnesol and 1-dodecanol
The mechanism by which small molecules regulate the ability of Tac1, as well as other members of the PDR1 family, to activate transcription has not been clearly defined. Direct binding to these small molecules/xenobiotics has been shown, in the case of C. glabrata PDR1 (37), to play an important role in this process. Among the strongest pieces of evidence for direct xenobiotic hyperactivation in C. albicans is the finding that a heterologous DNA binding domain fused to Tac1 (minus the DNA binding domain) activated reporter genes in response to xenobiotic (29, 31). We used a C. albicans one-hybrid assay to test whether Tac1 or Znc1, with its native DBD replaced by the LexA DBD, could activate a LacZ reporter in response to FOH and 1-DD. The LexA-Tac1 fusion protein activated its reporter gene in response to both FOH and 1-DD at levels comparable to, although somewhat lower than, the levels previously observed (31) for FNZ (Fig. 4A). The LexA-Znc1 construct induces the LacZ reporter in response to FOH and 1-DD, but doesn’t respond to treatment with FNZ. The fold activation by FOH and 1-DD are similar for LexA-Znc1 and LexA-Tac1, while the LexA-Znc1 has a basal activation potential that is slightly higher than the Tac1 construct. Hal9, the C. albicans transcription factor with next highest similarity to Tac1 and Znc1, fused to LexA does not activate the reporter in response to FNZ, FOH or 1-DD (Fig. 4).
Hyperactivation of Tac1 tightly correlates with its phosphorylation by the Mediator complex, and can be detected by a decrease in gel mobility (5). FOH and 1-DD treatment both result in an N-terminal HisFlag tagged Tac1 mobility shift that is slightly lower than the shift caused by FNZ (Fig. S2A). The Tac1 band shift by FOH is unaffected in znc1 deletion mutant (Fig. S2B), ruling out a potential competitive effect by hyperactivated Znc1. The variability of Tac1 phosphorylation pattern suggests inducer specific conformations of hyperactive Tac1 that lead to differential phosphorylation by Mediator. To test if hyperactivated Znc1 is also subject to phosphorylation, we generated strains expressing C-terminally 3xHA tagged Znc1. The tagging does not compromise Znc1 activation competence at the CDR1 and orf19.320 promoters (Fig. S2C and S2D). As opposed to Tac1, FOH or 1-DD treatment does not induce detectable changes in Znc1-3HA mobility in either wild type or tac1 deletion background (Fig. S2A and S2E).
Tac1 and Znc1 promoter occupancy correlates with their impact on target gene induction by FOH
ChIP analysis was performed to test whether Tac1 and Znc1 promoter occupancy determined their FOH induced target gene specificity. Tac1 and Znc1 occupancy are enriched at the CDR1 DRE in the absence of inducer, and this occupancy is enhanced by treatment with FOH (Fig. 5A). There is also a weak, but reproducible, enrichment of Tac1 and Znc1 occupancy at the RTA3 DRE under non-inducing conditions. Similar to our observation that Znc1 was the primary regulator of RTA3 expression in response to FOH, only Znc1 occupancy at the RTA3 DRE was increased by treatment with FOH (Fig. 3B). Tac1 and Znc1 occupancy were specific to the CDR2 and orf19.320 promoters, respectively, under non-inducing conditions and were enhanced by induction with FOH (Fig. 5C and 5D). Another Znc1 dependent promoter, IFD1, is exclusively bound by Znc1, but only after FOH treatment (Fig. 5E). The ChIP assay results allowed identification of potential cis elements for Znc1 at the tested promoters. High Znc1 occupancy at the CDR1 and RTA3 DRE suggests DNA binding preference of Znc1 similar to that of Tac1. Thirteen base pair (bp) DRE-like CGG triplet sequences were found in the orf19.320 and IFD1 promoter regions, whose location correlated to the region of highest local enrichment for Znc1 ChIP signal. Thus, we refer to these 13 bp elements as potential Znc1 binding motifs (PZMs). DREs and PZMs in the tested genes share a core consensus of CGGNNNNCGGAN (Fig. 5F). Multiple bases found in PZMs (labeled red in Fig. 5F) have been reported to impair CDR2 DRE function (26), and may specifically reduce Tac1 binding. The ChIP (Fig. 5B) and expression analysis (Fig. 3A) suggest that one such nucleotide in the RTA3 DRE, P12 A, may be better tolerated by the Znc1 DBD than the Tac1 DBD. One model for the partially redundant function of Tac1 and Znc1 at the CDR1 promoter is that both transcription factors competently bind to the CDR1 promoter in the absence of the other. To test this hypothesis we performed Znc1 ChIP in a tac1∆/∆ strain. Znc1 occupancy is observed at the CDR1, RTA3 or orf19.320 promoters in a tac1∆/∆ strain, and even increases at the CDR1 DRE (Fig. 5G) compared to a wild type TAC1 strain. This last finding indicates Tac1 and Znc1 may compete for DRE binding at promoters where we detected co-occupancy. Additionally we have found in ChIP assays that FNZ treatment has only a minor effect on Znc1 occupancy compared to its effect on Tac1 at the CDR1 promoter (Fig. S3A), or the effect of FOH/1-DD on Znc1 occupancy at the RTA3 promoter (Fig. S3B). Previous studies have shown that Tac1 GOF mutants can confer fluconazole resistance through a mechanism that relies on the ability of Tac1 to bind and activate the CDR1 promoter (25, 26, 28, 31). We have found that the fluconazole MIC in TAC1GOF mutant strains does not decrease in znc1∆/∆ strain (Table S1). Collectively, this evidence supports the hypothesis that Tac1 and Znc1 bind promoters independently of each other.
FOH induced Znc1 works through a Mediator dependent co-activator mechanism
Our previous work showed recruitment of Mediator complex is critical to FNZ induced Tac1 dependent CDR1 activation (31). Here, we found the Mediator tail module is also important for FOH induced CDR1 expression (Fig. 6A) and that either Tac1 or Znc1 is competent for Mediator recruitment at the CDR1 DRE under these conditions (Fig. 6B). Therefore, Tac1 and Znc1 both show DRE binding and Mediator recruitment at the CDR1 promoter in the presence of FOH that is independent of the other.
Additional transcription factors regulate FOH and 1-DD induced transcription in a promoter specific manner
Despite the FOH induction of CDR1 being severely compromised in the tac1∆/∆ znc1∆/∆ strain, a small residual induction was observed in this background (Fig. 7A). This finding suggested that another transcription factor was involved in CDR1 activation by certain inducers. A genetic screen of zinc cluster transcription factors identified Mrr2, Stb5 and Cta4 as potential regulators of CDR1 (30). The tac1 znc1 cta4 and tac1 znc1 stb5 triple deletion mutants showed unaffected FOH and 1DD induction of all genes tested, compared to a tac1 znc1 double null strain (Fig. 7A). Deletion of mrr2, however, largely eliminates the residual induction of CDR1 mRNA by FOH (and 1-DD) in the tac1∆/∆ znc1∆/∆ strain (Fig. 7A), while deletion of mrr2 also compromises CDR1 activation by FNZ, FOH and 1-DD when Tac1 and Znc1 are both present (Fig. 7B). Deletion of mrr2 in the tac1∆/∆ znc1∆/∆ background also decreases induced Cdr1 protein levels (Fig. 7C). Interestingly, mrr2 deletion appears to cause a greater decrease in the induced Cdr1 protein levels compared to mRNA levels (Fig. 7B and 7C), suggesting Mrr2 may also regulate CDR1 post-transcriptionally. In addition to CDR1, the genes CDR2 (Fig. 2A), PDR16 (Fig. S4A), orf19.7042 (Fig. S4B) and orf19.344 (Fig. S4C) are dependent on TAC1, ZNC1 or a combination of the two under different induction conditions (30). Deletion of mrr2, stb5 or cta4 did not impact the expression of these additional Tac1/Znc1 target genes when we assayed their induction in a tac1∆/∆ tnc1∆/∆ background (Fig S4D-F). Likewise deletion of mrr2 in an otherwise wild type background did not compromise induction of the tested Tac1/Znc1 target genes, other than CDR1, by FNZ, FOH or 1-DD (Fig S4G-K). Among the genes tested, Mrr2 functions as a specific modulator of CDR1 expression, rather than a broad regulator of FOH and 1-DD induction. The transcription factor(s) responsible for the residual FOH or 1-DD induction of PDR16, orf19.7042 and orf19.344 in the tac1∆/∆ znc1∆/∆ strains remain unidentified. Fig. 7D provides a heat map summary of the impact of Tac1, Znc1 and Mrr2 on expression of the fluphenazine, farnesol and 1-dodecanol induced genes analyzed in this work.
Cdr1 mediated feedback regulation of the transcriptional FOH response
It has been proposed that C. albicans Cdr1 can decrease intracellular FOH concentration via its efflux pump activity (22). This led us to hypothesize that FOH induced CDR1 expression may be part of a negative feedback mechanism to down-regulate the cellular response to FOH. To test whether increased CDR1 expression down-regulated the transcriptional response to FOH, we compared FOH induced gene expression in a wild type and cdr1 deletion background. In a cdr1∆/∆ strain, CDR2, orf19.320 and orf19.344 are all expressed at higher levels by the same concentration of exogenous FOH (Fig. 8A). cdr2 deletion does not affect FOH induction, and had a minimal effect when combined with cdr1 deletion, suggesting CDR1 plays a specific role in modulating this response. Based on this finding we sought to determine whether the efficiency of Cdr1-mediated xenobiotic transport governed the transcriptional response to other small molecules. Farnesyl acetate, an FOH like molecule, causes little to no induction of FOH-inducible promoters in a wild-type strain (Fig. 8B). In the absence of CDR1, however, farnesyl acetate and FOH result in comparable induction of several FOH target genes (Fig. 8B). This finding suggests that the Cdr1 dependent transport of farnesyl acetate, rather than an inability to activate the relevant transcription factors, is the limiting factor a response to farnesyl acetate. Compared to FOH induction curve that peaks and then decreases, the induction curve for farnesyl acetate in the cdr1 null shows a plateau or gradual increase (Fig. 8A-B). Similar to FOH, 1-DD treatment induces higher levels of target gene expression in the absence of cdr1 (Fig. S5A and S5B), while geraniol and tryptophol do not (Fig. S5C). Deletion of cdr1 only enhances the FOH/1-DD response to specific inducers rather than a pool of broadly related molecules. The CDR1 facilitated negative feedback model predicts that the increase in FOH target gene expression in the cdr1 null strain will also be dependent on Tac1 and Znc1. Indeed, FOH induction in wild type and cdr1∆/∆ strains exhibits a similar dependence on TAC1 and ZNC1 (Fig. 8C and Fig. S4D-F). The negative feedback model also predicts that over-expression of Cdr1 would increase farnesol efflux and desensitize C. albicans to FOH induction. A strain carrying a TAC1 GOF allele, which over-expresses CDR1, showed dramatically lower degree of Znc1 target gene induction than a wild type strain at identical concentrations of FOH (Fig. 8D). These results further support a model where FOH induced Cdr1 expression facilitates FOH clearance.
Cdr1 facilitates C. albicans and C. dubliniensis resistance to FOH exposure
C. albicans, is able to tolerate significantly higher levels of exogenous FOH compared to other ascomycetes, through a mechanism that is not entirely understood (13, 14, 24). To determine whether the CDR1 transcriptional response pathway described above could detoxify FOH, we investigated the role of Cdr1 in C. albicans FOH survival. Under the conditions tested, our C. albicans wild type strain did not exhibit a major decrease in colony size when grown on YPD agar containing 200 μM FOH (Table 1). The growth of a cdr1 deletion strain, however, was compromised or completely inhibited at 50 or 100 μM FOH (Table 1). Colony growth, within the FOH concentration range tested, was not affected in a cdr2 null strain. FOH tolerance, however, was further decreased in a cdr1∆/∆ cdr2∆/∆ strain versus a cdr1∆/∆ strain. Consistent with this finding, deletion of tac1 further sensitizes a cdr1 null mutant to FOH, while TAC1 GOF mutants, which overexpress Cdr2, increase FOH tolerance in a cdr1∆/∆ strain in a manner that is dependent on CDR2 (Table 1). A killing assay was performed to determine if FOH inhibits growth of cdr1 deletion mutants through a fungistatic or fungicidal effect. The cdr1∆/∆ cdr2∆/∆ strain showed decreased viability upon treatment with 200 μM FOH as only ~1% mutant cells survived after 6 hours treatment (Fig. 9A). Identical treatment with vehicle had no effect on viability (Fig. S6A) The viability of the cdr1∆/∆ cdr2∆/∆ and cdr1∆/∆ strains decrease at a similar rate during the first 6 hours of FOH exposure. Deletion of the transcription factors that drive FOH induction of CDR1 and CDR2 had a lesser impact on viability after FOH exposure than complete deletion of CDR1 and CDR2 (Fig. 9A). Deletion of tac1, znc1 and mrr2 individually, or tac1 znc1 simultaneously, does not affect FOH tolerance (Table 1). The tac1∆/∆ znc1∆/∆ mrr2∆/∆ strain, however, showed mildly compromised colony formation and decreased cell growth upon FOH exposure (Table 1, Fig. 9B). There was no difference in growth sensitivity of the wild type and the cdr1∆/∆ cdr2∆/∆ strains to SDS, indicating that the mutants did not impact the membrane integrity of the cells (Fig. S6B).
The C. dubliniensis genome sequencing strain CD36, a close relative of C. albicans, has been reported to have much higher sensitivity to FOH induced cell death (24). It is also known that multiple C. dubliniensis strains within genotype group I, including CD36, do not express functional full length Cdr1 protein due to a homozygous variant that creates a stop codon at CDR1 amino acid 756 (38). All cdr1756stop/756stop C. dubliniensis strains (CD36, CD38 and Wü284) we tested showed complete or partial growth inhibition by 50 μM FOH (Table 2; Fig. 9C). There are multiple CDR1+/+ C. dubliniensis strains (such as CD57), and we have observed that they exhibit comparable resistance to C. albicans at FOH levels as high as 200 μM (Table 2; Fig. 9C). Additionally we have found that deletion of cdr1 sensitizes CD57 to farnesol, while deletion of cdr1 in the cdr1756stop/756stop strain Wü284 does not further sensitize it to farnesol (Table 2).
Tac1 and Znc1 mediate the CDR1 the transcriptional FOH response in C. dubliniensis
Our standard protocol showed that FOH induces CDR1 expression in CD57 with similar kinetics and amplitude to C. albicans (Fig. 10A). Although CD57 showed an extremely low expression level of CdCDR2 compared to its C. albicans ortholog, FOH treatment results in a comparable fold induction of CDR2 in the two species (Fig. 10A). The ortholog of orf19.320, a C. albicans Znc1 dependent promoter, in C. dubliniensis (Cd36.83180) had an expression pattern in which induction was only observed after one hour of FOH treatment (Fig. 10A). A LexA-CdTac1 fusion protein, constructed in a similar fashion to the earlier LexA-CaTac1 fusion (Fig. 4), activated LacZ expression upon FOH and FNZ treatment when tested in one hybrid reporter assay in C. albicans (Fig. 10B). Representatives of the three different classes of FNZ/FOH induced promoters defined in C. albicans, Tac1/Znc1 dependent (CDR1), Tac1 dependent (CDR2) and Znc1 dependent (Ca orf19.320/Cd36_83180), have a very similar dependence on these same transcription factors in C. dubliniensis (Fig. 10C). An exception was the observation that deletion of znc1 in the tac1∆/∆ CD57 strain did not decease FOH CDR1 induction as strongly as it does in C. albicans, suggesting the existence of other regulator(s) governing FOH induction of the CdCDR1 promoter.
Discussion
Our characterization of Tac1, Znc1 and Mrr2 as essential signal targets (direct or indirect) for farnesol provides a new framework for thinking about how the C. albicans cell coordinates its transcriptional response to the quorum-sensing molecule. The demonstration that multiple Zn Cluster transcription factors can be activated by overlapping, yet non-redundant, small molecules reveals a previously underappreciated combinatorial complexity that allows these factors to regulate complex patterns of gene regulation. Additionally, the dependence of CDR1 expression on this transcriptional response combined with the regulation of this response by the action of this important efflux pump supports the mounting evidence for CDR1 serving as a farnesol transporter and suggests the presence of a negative feedback loop modulating its action.
Tac1 and Znc1 act as targets in Candida for farnesol
Earlier work had indicated that CDR1 expression was up regulated upon treatment with farnesol (21, 22), and that Tac1 was an important regulator of CDR1 (25, 27). This work demonstrates that farnesol can activate an acute transcriptional response via the hyperactivation of Tac1 and Znc1. The kinetics and amplitude of this response is very similar to the well-characterized response of Tac1 to xenobiotics, such as fluphenazine and estradiol (28), and expands the range of C. albicans Tac1 hyper-activators into the realm of physiological small molecules. Our discovery that farnesol can lead to the up regulation of CDR1 through the hyperactivation of Znc1 reveals that the control of the efflux pump expression is more complex than previously appreciated. The finding that Tac1 and Znc1 share overlapping, but non-identical, targeting to sequences in the promoter elements in farnesol induced genes extends our knowledge of how a diverse set of genes can be upregulated via farnesol. It is uncertain, at this point, whether farnesol activates Tac1 and Znc1 through a direct binding mechanism, similar to the activation of Pdr1 in C. glabrata by azoles (37), or through an indirect mechanism such as post-translational modification. The observed sub-cellular localization of exogenously-added farnesol includes nuclear enrichment (39), which would allow farnesol to hyperactivate Tac1 and Znc1 through direct binding. Parallels have previously be drawn between the mechanism of action of the C. glabrata Pdr1 zinc cluster transcription factor and mammalian nuclear receptors (37, 40), and it is interesting to note that the mammalian bile acid receptor FXR (Farnesoid X-activated Receptor) is also activated by farnesol (41). However, strong physiological FXR agonists, such as chenodeoxycholic acid (CDCA) and deoxycholic acid (DA) (42, 43), do not activate CDR1 expression in C. albicans (Fig. S1), suggesting the lack of a boarder overlap between Tac1/Znc1 and mammalian FXR inducers. If direct binding of inducers to the Zn cluster transcription factors is the mechanism of hyperactivation, the structural differences between Tac1 inducers (farnesol, dodecanol, fluphenazine, estradiol) suggest a low affinity/specificity interaction similar to that observed for the FXR receptor. The observation that Znc1 can respond to farnesol (and dodecanol), but not fluphenazine, indicates that it is possible for the Zn cluster transcription factors to make structural distinctions between these molecules. Although we cannot rule out post-translational modification as a mechanism, the absence of a mobility shift in Znc1 upon farnesol treatment compared to the hyperactive Tac1 phosphorylation mobility shift (31) suggests that phosphorylation is not a requirement for farnesol activation of Tac1 and Znc1.
Mode of Tac1- and Znc1-DNA interaction
In fungi, paralogous zinc cluster transcription factors have been reported to form both homo- and hetero-dimers, for example Pdr1/Pdr3 (44) and Oaf1/Pip2 (45, 46) in S. cerevisiae. Our observation of distinct binding of Tac1 and Znc1 at promoters where they show non-redundant activation, as well as the binding of Znc1 to promoters in a tac1 null strain, does not support the idea that Tac1 and Znc1 bind promoters as a stable heterodimer. The previous observation that Tac1 requires the presence of both CGG triplet elements in a DRE for gene activation (26) largely rules out the possibility that monomeric Tac1 and Znc1 each bind to one CGG triplet in a single DRE, a mode of DNA binding infrequently observed for zinc cluster transcription factors (47). The evidence presented here supports a model where Tac1 homodimers and Znc1 homodimers competitively bind to co-regulated promoters at a single paired CGG triplet element. Given the variation in DRE and PZM sequences at Tac1 dependent promoters (i.e. CDR2) and Znc1 dependent promoters (i.e. orf19.320), it is likely that Tac1 and Znc1 have overlapping, but non-identical sequence specificity. The DNA binding specificity of zinc cluster transcription factors, namely the sequence between and surrounding the CGG triplet(s), is thought to be determined by the amino acid sequence in the linker region between the zinc clusters and the dimerization domain (48–50). Divergence in Tac1 and Znc1 in their linker regions may contribute to their different occupancy at the DREs and PZMs. Since it has previously been noted that Tac1 binds to DREs regardless of the promoter status (36) it was reasonable to think that hyperactivation of Tac1 did not work by a mechanism that increased DNA binding of the transcription factor. The ChIP analysis shows, however, that DRE/PZM occupancy is clearly increased in an inducer specific manner. It is unclear whether this reflects a direct increase in binding affinity of the transcription factor or the formation of a more stable transcription factor complex involving interactions with co-activators and enhanced chromatin remodeling.
The complex regulation of the CDR1 promoter
The new regulatory mechanisms revealed by our study further demonstrate the complexity of the CDR1 promoter. Previous studies of activation of the CDR1 promoter have focused on gain-of-function mutants in Tac1, or xenobiotic hyperactivation of Tac1(25, 26, 28, 51). Our study now adds farnesol and 1-dodecanol treatment to the limited number of conditions where CDR1 expression can be induced chemically in the absence of Tac1 (29). Tac1-indepndent CDR1 induction by farnesol can be largely attributed to Znc1 (and Mrr2) function. Interestingly, unlike Tac1 (27) and Mrr2 (52), no gain-of-function mutants of Znc1 have been reported to drive azole resistance in C. albicans. Of the well-characterized CDR1 inducers, farnesol is the only one considered to be a C. albicans metabolite.
Identification of Tac1 and Znc1 as farnesol induced transcription activators of the CDR1 promoter allowed us to test whether the proposed CDR1 mediated farnesol efflux (22) provided feedback to the transcriptional response. The amplification of the Tac1 and Znc1 driven transcriptional response to farnesol in cells lacking Cdr1 function, as well as the dampening of the Znc1 driven transcriptional response to farnesol in TAC1 gain of function cells that overexpress CDR1 (Fig. 8), both support the idea of Cdr1 serving to regulate intracellular levels of farnesol via an efflux mechanism. However, the observation that the transcriptional response to farnesol still exhibits attenuation after rapid induction in a cdr1 null strain (Fig. 8) indicates that there may be additional farnesol transporters in C. albicans. The contrast of this response compared to the plateau observed for the transcriptional response to farnesyl acetate in cdr1 null strain indicates that Cdr1 may be the sole efflux pump for farnesyl acetate (Fig. 8B).
The role of CDR1 in modulating farnesol sensitivity
Although the phenotypic regulation of C. albicans by farnesol is not the central focus of the work presented here, the suggested circuit involving farnesol, Zn cluster transcription factors and Cdr1 prompted us to begin to investigate how these factors might interact to influence Candida biology. Under our experimental conditions (cells grown in YPD), as well as synthetic media conditions tested in several other studies (53–55), farnesol concentrations as high as 300 showed minor effects on the growth or viability of wild type C. albicans. These concentrations of farnesol are typically toxic to most other fungi (14, 24). Our analysis of farnesol toxicity to C. albicans and C. dubliniensis strains with and without functional Cdr1 consistently suggests Cdr1 mediated farnesol efflux plays a protective role under these growth conditions (Fig. 9, Table 1 and 2). It is somewhat surprising that transcription factor mutants in C. albicans and C. dubliniensis do not show a more dramatic change in farnesol sensitivity given the decrease in CDR1 induction. It is possible that expression of other genes that influence farnesol sensitivity were affected by tac1 or (and) znc1 or (and) mrr2 deletion in a way that compensated for the compromised CDR1 induction in these mutants. Elsewhere it has been reported that farnesol treatment in a non-media condition (PBS) induced apoptosis in C. albicans through a Cdr1-dependent mechanism (22), suggesting Cdr1 may regulate C. albicans farnesol sensitivity in either direction depending on the treatment condition. Since Cdr1, an ATP-dependent transporter, activity is strongly affected by cellular energy status (56, 57), the availability of may nutrients impact the role of Cdr1 during farnesol exposure.
This results of this study lead to the ability to ask new questions about both the general role and mechanism of zinc cluster transcription factors in the response to physiological fungal metabolites, as well as the action of farnesol as a quorum sensing molecule. For instance, how do Tac1 and Znc1 achieve specificity in their response to farnesol? are there other metabolite ligands? and do the other genes in the Tac1/Znc1 farnesol regulon play an important role in quorum sensing. It has been observed that an increase in CDR1 and CDR2 expression in sessile C. albicans cells, compared to planktonic cells (58, 59) contributes to the drug resistance in early C. albicans biofilm, in cooperation with the up regulation of major facilitator transporter, MDR1 (60) . The mechanism underlying the induction of these pumps in biofilm has not been clarified. Farnesol induction of CDR1 and CDR2 suggests accumulation of quorum sensing molecule favored by static growth as a possible answer.
Materials and Methods
Strains and plasmids
Strains and plasmids used in this study are respectively listed in Table S2 and Table S3. Construction of the strains and plasmids are described, in detail, in the ‘Supplemental Methods’ session. Primers used for generating the strains and plasmids are listed in Table S4. C. albicans transformation was performed by electroporation and selected by Clonat resistance (1% Yeast extract, 2% Peptone, 2% Glucose, 2% Agar, 0.1 mM Uridine, 100 μg/mL Clonat) or complementation of auxotrophy (6.7 g/L Difco YNB without amino acids (BD), appropriately supplemented 1.5 g/L Drop-out Mix Synthetic without uracil, histidine, arginine, leucine (US Biological), 2% Glucose, 2% Agar). Expression of flippase was induced by growth in YPMal (1% Yeast extract, 2% Peptone, 2% Maltose, 0.1 mM Uridine) liquid media for 24 hours. Successful eviction of the SAT1 marker was selected by sensitivity to 100 μg/mL Clonat.
Cell growth and drug treatment
Cells were grown in liquid YPD media (1% Yeast extract, 2% Peptone, 2% Glucose, 0.1 mM Uridine) at 30°C if not specified. Drug treatment was performed by adding fluphenazine (Alfa Aesar, 6 mg/mL aqueous stock), farnesol (MP bio (mixed isoforms), biweekly 50 mM or fresh 200 mM methanolic dilution), 1-Dodecanol (Sigma, 50 mM methanolic dilution), farnesyl acetate (Fluka (mixed isoforms), 50 mM methanolic dilution), geraniol (Sigma, 50 mM methanolic dilution), tryptophol (Sigma, 50 mM methanolic stock), tyrosol (Alfa Aesar, 50 mM aqueous stock), chenodeoxycholic acid (Cayman, 50 mM DMSO sock) or deoxycholic acid (Fisher, 50 mM DMSO stock) into mid-log phase C. albicans or C. dubliniensis culture to the final concentration specified for each experiment in the figure legends. To test cell growth in the presence of farnesol, an overnight culture, after appropriate dilution, was spread on YPD plates supplemented with each concentration of FOH or same volume of methanol. Colony number and size were analyzed by OpenCFU (62) after 40 hours incubation at 30°C. In farnesol killing assays, an overnight culture of each strain to be tested was diluted to OD 0.05 in fresh YPD media for treatment with 200 μM farnesol or same volume of methanol. Aliquots taken at each of the indicated time points, after an appropriate dilution (if needed) were plated on YPD agar (no farnesol). Colony number was counted after 40 hours incubation at 30°C.
RT-qPCR
RNA samples were prepared from collected frozen cell pellets and reverse-transcribed as described previously (61). qPCR was performed using ‘Relative Standard Curve’ method (StepOne, Life Technologies). ACT1 abundance measured by ZL712/ZL713 was used as an internal reference to compare CDR1 (ZL540/ZL541), CDR2 (ZL542/ZL543), RTA3 (ZL544/ZL545), orf19.320 (ZL951/ZL958), IFD1(ZL823/ZL824), PDR16 (M2PT-1/M2PT-2), orf19.7042 (M2PT-23/M2PT-24), orf19.344 (M2PT-15/M2PT-16) or expression across strains and conditions. ZL540/ZL541, ZL1008/ZL1009 and ZL951/ZL958 were used as respective pan-primers to compare expression and induction of CDR1, CDR2 and orf19.320 homologs in C. albicans and C. dubliniensis.
Immunoblotting
Immunoblot analysis was used to compare Cdr1-3HA expression, or 6His3Flag-Tac1 or Znc1-3HA gel mobility. Cell lysates were prepared, resolved by SDS-PAGE and probed by an α-HA (Roche, 3F10) or α-Flag (Sigma, F7425) antibody as described (31). A lower molecular weight region of a gel, which typically did not contain the immunoblotting signals of interest in this study, was stained by Coomassie Blue as the loading reference.
Chromatin Immuno-Precipitation (ChIP)
ChIP experiments were performed as described previously (31) to analyze 6His3Flag-Tac1, Znc1-3HA and Mediator (Med17-3HA) occupancy at target promoters. Results of ChIP experiments are presented in a ‘Relative Recovery of Input’ form. The absolute recovery at the CDR1 promoter ‘1-up’ region in a non-farnesol treated untagged strain (as specified in figure legends) was set to ‘1’ to normalize recoveries at other promoter regions across conditions. Primers used in the ChIP assay are listed in Table S4 with their probing region denoted.
Liquid β-galactosidase activity assays
C. albicans one-hybrid strains was diluted from overnight culture, grown for 3 to 4 hours in fresh YPD media and treated with fluphenazine (~25 μM), farnesol (50 μM), 1-Dodecanol (50 μM) or methanol for 1 hour or 3 hours before collection for β-galactosidase activity measurement by an SDS/Chloroform method (31, 63). β-galactosidase activity in Miller units were calculated by the following simplified formula: 1,000 × A420/ (T X C), where A420 is the absorbance of the reaction product at 420 nm, T is the reaction time in minutes, C is the total amounts of cells in total OD600 used in the reaction. A420 and OD600 values were measured with a Beckman Coulter DU-7300 spectrophotometer. Activity of each LexA fusion protein was tested in at least three confirmed transformants.
Acknowledgments
This study is supported by NIH 5R21AI113390 and 5R21AI115253 to LCM. We thank Dr. Deborah Hogan and the Hogan lab for sharing their expertise and reagents relating to farnesol biology. We also thank Dr. Gary Moran for providing strains.