If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
The human fungal pathogen Candida albicans responds to iron deprivation by a global transcriptome reconfiguration known to be controlled by the transcriptional regulators Hap43 (also known as Cap2), Sef1, and the trimeric Hap2-Hap3-Hap5 complex. However, the relative roles of these regulators are not known. To dissect this system, we focused on the FRP1 and ACO1 genes, which are induced and repressed, respectively, under iron deprivation conditions. Chromatin immunoprecipitation assays showed that the trimeric HAP complex and Sef1 are recruited to both FRP1 and ACO1 promoters. While the HAP complex occupancy at the FRP1 promoter was Sef1-dependent, occupancy of Sef1 was not dependent on the HAP complex. Furthermore, iron deprivation elicited histone H3-Lys9 hyperacetylation and Pol II recruitment mediated by the trimeric HAP complex and Sef1 at the FRP1 promoter. In contrast, at the ACO1 promoter, the HAP trimeric complex and Hap43 promoted histone deacetylation and also limited Pol II recruitment under iron deprivation conditions. Mutational analysis showed that the SAGA subunits Gcn5, Spt7, and Spt20 are required for C. albicans growth in iron-deficient medium and for H3-K9 acetylation and transcription from the FRP1 promoter. Thus, the trimeric HAP complex promotes FRP1 transcription by stimulating H3K9Ac and Pol II recruitment and, along with Hap43, functions as a repressor of ACO1 by maintaining a deacetylated promoter under iron-deficient conditions. Thus, a regulatory network involving iron-responsive transcriptional regulators and the SAGA histone modifying complex functions as a molecular switch to fine-tune tight control of iron homeostasis gene expression in C. albicans.
Living cells actively and efficiently maintain iron homeostasis, a process leading to the maintenance of optimal intracellular iron levels. The ability of pathogens to adapt to host environments is intimately linked to how gene expression is modulated by transcription factors and chromatin regulators. Candida albicans, the most prevalent human fungal pathogen, is an opportunistic pathogen and becomes virulent in immunocompromised host (
). Iron is an essential micronutrient required for various metabolic steps, but has poor bioavailability in the mammalian host with vastly different concentrations in blood stream, gastrointestinal tract, and deep-seated tissues (
C. albicans cells respond to iron levels in the culture media by global transcriptome reconfiguration, involving upregulation of numerous genes including those in the multiple iron uptake pathways, and downregulation of genes encoding iron-requiring enzymes and proteins (
) and is summarized in Figure 1. Iron deprivation elicited the transcriptional induction of SEF1, leading to the induced expression of HAP43, HAP2, and HAP3 genes (Fig. 1). The Hap2, Hap3, and Hap5 proteins form a trimeric HAP complex that binds to the CCAAT motif in vitro (
). C. albicans mutants bearing deletions of HAP2, HAP3, HAP5, or HAP43 genes led to misregulation of the expression of iron homeostasis genes and impaired cellular growth under iron deplete conditions (
Genome-wide analyses showed that Hap43 and Sef1 have a widespread role in transcriptional control of iron homeostasis. Microarray analysis showed that of the Hap43-regulated genes, ∼56% were repressed and ∼44% induced were under Fe-deplete conditions (
), suggesting that both Hap43 and Sef1 have dual regulatory functions. Interestingly, comparison of the Sef1 and Hap43 transcriptomes showed that ∼35% are induced and ∼14% repressed commonly in a Hap43-and Sef1-dependent manner (
Genome-wide ChIP-Chip and microarray analyses showed that Sef1 is recruited to the promoters of ∼20% (∼60 genes) of low iron-induced genes including several iron uptake genes (for example, FRP1 encoding ferric reductase) and to the promoters of HAP2, HAP3, and HAP43 transcriptional regulator genes (
). When iron is surfeit, Sfu1, a GATA-type transcriptional regulator binds to the promoters of several iron uptake genes including FRP1 and that of the SEF1 regulator to down regulate their expression (
). Several key aspects of the transcriptional regulation by the HAP genes and the SEF1 gene are poorly understood. First, it is unclear how the HAP complex and the SEF1 genes coordinate transcriptional activation of iron homeostasis gene expression. Second, the mechanism of how the HAP trimeric complex and the Hap43 mediate transcriptional repression of iron utilizing genes. In this study, we have identified the hierarchy of regulation of target genes by Sef1, HAP complex, and the Hap43 to orchestrate a combinatorial regulatory network resulting in modulation of SAGA-mediated histone acetylation and RNA polymerase (Pol) II recruitment, thereby providing tight transcriptional control of iron homeostasis.
Interdependent expression of HAP complex subunits and Sef1
To investigate the steady-state protein levels of each of the HAP complex and Sef1 transcriptional regulators, we TAP-tagged the respective genes at their endogenous loci and carried out quantitative western blot analysis of cell extracts prepared from cells grown in two different media—either SC-LIM or the YPD media containing BPS (−Fe) or BPS plus FAS (+Fe). Quantification of the western blot data showed that although Hap2, Hap3, Hap5, and Sef1 are all expressed under both −Fe and +Fe conditions, only Hap3 and Sef1 proteins were upregulated (∼3.5-fold) in the BPS-containing YPD (Fig. 2, A and B) and SC-LIM media (Fig. S1, A and B). The elevated Sef1 and Hap3 protein levels in −Fe medium (Fig. 2, A and B) correlated well with the upregulation of the cognate mRNAs (
), our western blot data did not reveal an increase in steady-state level of the Hap2 protein in −Fe compared with +Fe conditions.
To determine if HAP43 expression is controlled at the level of transcription, we constructed a HAP43-lacZ promoter fusion and analyzed expression in WT, hap43Δ, and sef1Δ mutants. The reporter fusion was activated by approximately 6-fold in WT strain in −Fe medium (Fig. S2A), and this activation was not impaired in the hap43Δ mutant indicating that HAP43 expression is not controlled by Hap43. In the sef1Δ mutant, however, the activation of HAP43-lacZ reporter was only partially reduced (approximately 2-fold) compared with the WT level (Fig. S2A). Therefore, we quantified HAP43 mRNA level in −Fe versus +Fe condition and found that the mRNA is induced ∼6-fold in WT, but the induction was lost in the sef1Δ mutant cells (Fig. S2B). Western blot analysis showed that although Hap43 protein level was also substantially reduced (Fig. S2, C and D), the Hap43 protein was still detectable in sef1Δ/Δ mutant cell extracts, indicating Sef1-independent partial expression of HAP43 in iron-starved cells.
To examine the requirement of the HAP subunits for HAP complex formation, we conducted CoIP assays using epitope-tagged Hap5, Hap2, and Hap3 in WT, hap3Δ/Δ, or hap2Δ/Δ strain backgrounds. In each case, we used TAP-tagged subunits to CoIP and probe the western blots with anti-FLAG antibody to identify protein–protein interactions. The data showed that Hap5 strongly pulled down Hap3 from WT cell extracts grown in −Fe medium, but not from +Fe cell extracts (lanes 8 and 16; Fig. 2C), likely due to the diminished expression of the Hap3 protein in +Fe compared with −Fe (Fig. 2B). The Hap2 protein was efficiently coimmunoprecipitated by Hap5 from cell extracts grown in +Fe or −Fe media (lanes 8 and 16; Fig. 2D). Moreover, the Hap2-Hap5 interaction was greatly impaired by the hap3Δ mutation in −Fe but not in +Fe condition (lanes 8 and 16; Fig. 2E). The CoIP data also showed that Hap5 interacts with Hap3 in both −Fe and +Fe cell extracts that is largely unaffected by hap2Δ/Δ mutation (Fig. 2F). Together these CoIP data indicated that the induced expression of Hap3 under −Fe condition is essential for Hap5-Hap2 interaction and assembly of Hap3 to form the HAP trimeric complex under iron starved condition. In +Fe condition, however, Hap5 interacted with Hap2 despite the diminished Hap3 protein level, thus indicating Hap3-independent interaction of Hap5 and Hap2.
Selectivity of promoter occupancy by the HAP trimeric complex subunits and Sef1
To address how the functions of these transcriptional regulators impinge on iron-responsive promoters, we carried out ChIP assays to confirm the in vivo occupancy of each subunit of the HAP complex and the Sef1 regulators. Our previous studies showed that under iron deprivation conditions Hap5 was recruited to FRP1 promoter (−84 to −281 with respect to ATG) and ACO1 promoter (−201 to −340) bearing the CCAAT motif (
). Therefore, we chose these promoters to analyze the occupancy of Hap5, Hap2, Hap3, Hap43, and Sef1 under iron deplete and iron-replete conditions using TAP-tagged C. albicans strains. Because HAP43 lost its activity upon epitope tagging (
Our ChIP data showed that the occupancy of each of the three components of the HAP trimeric complex, Hap5, Hap2, and Hap3 is enhanced at the FRP1 promoter (Fig. 3A) and at the ACO1 promoter (Fig. 3B) in −Fe condition compared with +Fe condition. We also extended the analyses to other low-iron-induced promoters (FTR1, SIT1) as well as to another low-iron-repressed promoter (CYC1) using the same ChIP samples. As expected, occupancy of Hap5 was elevated at all three promoters under −Fe compared with +Fe condition (Fig. 3C). While Hap43 occupancy was undetectable at the FRP1 promoter (Fig. 3A), its occupancy was substantially high at the ACO1 promoter (Fig. 3B) and the CYC1 promoter under Fe-deplete condition (Fig. 3C). The ChIP analysis also showed Sef1 occupancy at FRP1, FTR1, SIT1, and ACO1 promoters but not to the CYC1 promoter (Fig. 3C). Given that Sef1 is a transcriptional activator, it was surprising to detect Sef1 occupancy at the ACO1 promoter under −Fe condition. Indeed, previous ChIP-chip data also seemed to identify Sef1 binding to the promoters of ACO1, ISA1, HEM3, and SEF2 genes whose expression is repressed under iron-deficient condition (
), although the significance of this data was not clear.
Together these ChIP data confirmed that the HAP trimeric complex is recruited to low-iron-induced iron uptake gene promoters as well as to the low-iron-repressed iron utilizing gene promoter under iron deprivation conditions. Hap43, however, showed a robust recruitment only to the low-iron-repressed promoters. Furthermore, Sef1 occupancy to low-iron-induced promoters (FRP1, FTR1, and SIT1) was elevated and, surprisingly, to the low-iron-repressed ACO1 promoter but not to the CYC1 promoter. Thus, iron deprivation elicited the induction of HAP trimeric complex occupancy at both iron uptake and iron utilizing gene promoters. Furthermore, iron deprivation also elicited the induction of Sef1 and Hap43 occupancy at the iron uptake and the iron utilizing gene promoters, respectively.
Interdependency of the HAP complex and Sef1 regulators for promoter occupancy
Next, we asked if the HAP trimeric complex, Sef1, and Hap43 transcriptional regulators are mutually dependent for the recruitment of each regulator to promoters in vivo. We constructed Hap5-TAP expressing strains in WT or in hap2Δ/Δ, hap43Δ/Δ, or sef1Δ/Δ mutant backgrounds, and Sef1-TAP strain in the background of hap3Δ/Δ mutation, and carried out ChIP assays with either IgG-Sepharose beads or affinity purified anti-Hap43 antibody. The ChIP data revealed that the recruitment of Hap5 to both FRP1 (Fig. 4A) and ACO1 promoters (Fig. 4B) in −Fe medium was abolished in the hap2Δ/Δ mutant indicating that Hap2 is required for Hap5 promoter occupancy. Our previous study showed that Hap5 recruitment is also dependent on Hap3 (
), thereby showing that Hap5, Hap2, and Hap3 function together in promoter binding of the HAP trimeric complex.
Next, to address the requirement of Sef1 for recruitment of the HAP complex, we conducted ChIP assays in sef1Δ/Δ strain expressing HAP5-TAP. The ChIP-qPCR results demonstrated that the Hap5 recruitment to both FRP1 and ACO1 promoters was lost in the sef1Δ/Δ mutant (Fig. 4, A and B). Furthermore, deletion of hap43Δ/Δ also abolished Hap5 recruitment at ACO1 and FRP1 promoters in Fe-deplete condition (Fig. 4, A and B). Together these data demonstrated that the HAP complex recruitment is highly dependent on both Hap43 and Sef1.
To address the requirement of the HAP trimeric complex and Sef1 for Hap43 recruitment, we conducted Hap43 ChIP analysis in hap5Δ/Δ and sef1Δ/Δ mutant strain backgrounds. The ChIP data showed that whereas Hap43, as expected, is not recruited to FRP1 (Fig. 4C), its recruitment to ACO1 promoter is completely dependent on the Hap5 subunit of the HAP trimeric complex (Fig. 4D). Next, we investigated the requirement of Sef1 for Hap43 recruitment to the ACO1 promoter. The sef1Δ/Δ mutation did not abolish Hap43 recruitment to the ACO1 promoter but was only reduced by ∼2.5-fold (Fig. 4D) indicating that Sef1 is not essential but required for full potential of Hap43 recruitment to promoter in vivo, consistent with partial dependence of HAP43 mRNA expression on Sef1 (Fig. S2D).
To assess if the HAP complex is required for Sef1 recruitment, we carried out ChIP assays using TAP-tagged SEF1 in WT and hap3Δ/Δ strain backgrounds. This experiment showed that Sef1 occupancy at the FRP1 promoter, as shown above, was stimulated upon iron deprivation, but interestingly the hap3Δ/Δ mutation did not impair the Sef1 occupancy in Fe-depleted medium (Fig. 4E). Furthermore, the Sef1 occupancy at the ACO1 promoter was also independent of Hap3 (Fig. 4E), indicating that Sef1 is recruited independent of the HAP complex at both FRP1 and ACO1 promoters.
HAP complex occupancy is dependent on both HAP- and Sef1-binding sites
To determine how Sef1 and the HAP complex mediate transcriptional activation, we chose the FRP1 promoter since both Sef1 and the HAP complex are recruited to the same location in the FRP1 promoter (Fig. 3A). The FRP1 promoter contains the CCAAT motif at −134 with reference to the FRP1 start codon, and this motif was shown to be critical for Hap5 binding and regulation of FRP1 expression in response to iron deprivation (
), the exact sequence/s has not been experimentally demonstrated. Therefore, we looked for a putative Sef1 site, bearing a 5′CGG motif at the FRP1 promoter and identified an elaborate sequence 5′CGG(N6)GGC(N18)CCAAT(N10)CCG with two apparent 5′CGG motifs in opposite orientations beginning at position −164 that nested the CCAAT motif (Fig. 5A).
To test the functional significance of the 5’CCAAT motif in vivo, we cloned the FRP1 promoter and mutated the 5’CCAAT sequence to 5’TGCGC by site-directed mutagenesis. As the in vivo requirement of the 5’CGG motif was not known, we introduced mutations in the two 5’CGG motifs and constructed FRP1 knock-in strains bearing either WT as control or those bearing point mutations within the CCAAT and CGG motifs (Fig. 5A) and examined the occupancy of Sef1 and HAP complex. The ChIP analysis showed Hap5 occupancy at the WT FRP1 promoter in the knock-in cells in −Fe medium (Fig. 5B). However, the ChIP data also showe′d that the 5′CCAAT– mutation impaired Hap5 occupancy to FRP1 promoter in vivo (Fig. 5B), consistent with the requirement of the 5′CAAT motif for Hap5 binding to FRP1 promoter in vitro reported previously (
Next, we conducted ChIP assays to examine Sef1 recruitment and found robust Sef1 recruitment to the WT FRP1 sequence under iron-deplete condition (Fig. 5B). But the CGG– mutation abrogated Sef1 occupancy (Fig. 5B), indicating the requirement of this promoter sequence for Sef1 binding to chromatin in vivo. Surprisingly, Hap5 occupancy was also impaired by the 5′CGG– mutation (Fig. 5B) suggesting that a WT 5′CGG motif also promotes HAP complex binding. Interestingly, Sef1 occupancy was not impaired by the 5′CCAAT– mutation (Fig. 5B), suggesting that Sef1 occupancy at FRP1 promoter in vivo occurs independent of the 5′CCAAT motif. Together these data suggest that the HAP complex recruitment is dependent on Sef1 occupancy but recruitment of Sef1 seemed to be independent of the HAP complex occupancy at the FRP1 promoter in vivo.
Modulation of H3K9 acetylation by HAP complex and Sef1 at promoters
Next, we assessed the status of nucleosomal histone H3-Lys9 acetylation, a transcriptional activation-linked chromatin mark, at the two iron homeostasis gene promoters. The H3-K9Ac level, normalized to total H3, was determined at the FRP1 promoter by ChIP assay. The data showed that FRP1 promoter was hypoacetylated in Fe-replete medium in the WT strain, but was strongly hyperacetylated in Fe-deplete medium (Fig. 6A). Deletion of SEF1, HAP5, or HAP43 reduced the hyperacetylation in Fe-deplete medium (Fig. 6A). At the ACO1 promoter, the H3-K9Ac level was strongly elevated in Fe-replete medium in the WT strain (Fig. 6B), and this hyperacetylation was dependent on both Sef1 and Hap43, but not on Hap5 (Fig. 6B). In Fe-deplete medium, the ACO1 promoter was not hyperacetylated at the H3-K9, but interestingly, hap43Δ/Δ and hap5Δ/Δ mutations led to hyperacetylation (Fig. 6B). Three conclusions can be drawn based on these ChIP data. First, at the FRP1 promoter, Sef1, Hap5, and Hap43 stimulate H3K9 hyperacetylation under −Fe condition. Second, at the ACO1 promoter, Sef1 and Hap43 promote H3K9 hyperacetylation under +Fe condition, and third, Hap5 and Hap43 repress H3K9 acetylation at the ACO1 promoter under −Fe condition.
RNA polymerase II recruitment during iron homeostasis transcriptional control
To examine the recruitment of Pol II to FRP1 and ACO1 promoters, we carried out ChIP analysis in the WT and the hap43Δ, hap5Δ, and sef1Δ mutant strains using an RNA pol II CTD phospho Ser5 antibody. We found that Pol II occupancy at the FRP1 promoter is highly dependent on Sef1 and Hap5 in −Fe medium (Fig. 6C). Interestingly, the hap43Δ mutation led to elevated Pol II levels at the FRP1 promoter compared with that in WT strain in −Fe medium (Fig. 6C). In +Fe medium, however, the sef1Δ, hap5Δ, and hap43Δ mutations did not significantly alter Pol II levels (Fig. 6C) at FRP1 promoter.
At the ACO1 promoter, we observed a high level of Pol II enrichment in +Fe medium in a manner dependent on Sef1 but not Hap5 and Hap43 (Fig. 6D). In contrast, no Pol II enrichment was seen at this promoter in −Fe medium. Interestingly, the hap5Δ and hap43Δ mutations, but not the sef1Δ mutation, led to a significant increase in Pol II occupancy compared with that in WT strain at the ACO1 promoter in −Fe medium (Fig. 6D).
To correlate the H3K9 acetylation and Pol II occupancy to the transcription state of FRP1 and ACO1 genes, we measured mRNA levels by qRT-PCR. FRP1 mRNA was highly induced (∼80-fold) in −Fe medium compared with +Fe medium in WT strain (Fig. 6E). The deletion of SEF1, HAP5, or HAP43 genes substantially reduced FRP1 induction in −Fe medium. The ACO1 mRNA was strongly downregulated in the WT strain in −Fe medium compared with +Fe medium (Fig. 6F). Interestingly, deletion of HAP5 or HAP43, but not SEF1, caused derepression of ACO1 expression as the −Fe/+Fe expression ratio was ∼1.0 in the mutants compared with ∼0.04 in the WT strain indicating the repressive activity of HAP5 and HAP43 at the ACO1 gene (Fig. 6F). Moreover, the derepression of ACO1 expression upon HAP5 or HAP43 deletion is consistent with the elevated H3K9Ac level (Fig. 6B) and Pol II occupancy (Fig. 6D). Although hap43Δ/Δ mutation led to elevated Pol II recruitment at both FRP1 and ACO1 promoters under −Fe conditions (Fig. 6, C and D), only ACO1 mRNA level is derepressed in −Fe medium suggesting a requirement for further activation of FRP1 promoter postrecruitment of Pol II. Together, the H3-K9Ac and Pol II occupancy data along with the mRNA data indicated that while Sef1 functions as an activator of Pol II recruitment and transcription of FRP1 in −Fe medium and that of ACO1 in +Fe medium, the HAP trimeric complex functions as a dual regulator of Pol II recruitment and transcription, i.e., as an activator at FRP1, and as a repressor at ACO1 promoters.
SAGA complex mediates iron homeostasis gene expression
The Gcn5-containing SAGA complex is the major histone acetyl transferase that mediates H3K9 acetylation in yeast (
). Therefore, to test if the SAGA complex is required for iron homeostasis gene regulation, we constructed three C. albicans strains bearing deletion mutations in SAGA subunits, namely GCN5 encoding the CaKAT2 histone acetyl transferase and two key SAGA-specific subunits SPT7 and SPT20. The mutants and the control WT strains were tested for growth in −Fe condition (+BPS plates). Compared with the WT parental strain, the gcn5Δ, spt7Δ, and spt20Δ null mutants were impaired for growth in −Fe plate (Fig. 7A). Of the three, the spt20Δ mutant showed the strongest growth defect as compared with the gcn5Δ and spt7Δ mutants. The three mutants also showed slow-growth phenotype in the control SC-LIM plate, and thus the BPS-sensitive growth defect appears to be less pronounced in BPS-containing plate (−Fe; Fig. 7A). However, iron supplementation to the BPS plate rescued most of the growth defect of the mutants (+Fe; Fig. 7A). Also, reintroduction of a WT copy of GCN5, SPT7, and SPT20 genes in the respective mutants complemented the growth defects in −Fe as well as in the control plate (Fig. 7A) indicating that the three SAGA genes are required for C. albicans response to iron deprivation.
Next, we analyzed if the SAGA subunit genes are required for expression of the iron homeostasis genes FRP1 and ACO1. The expression of FRP1 mRNA was induced ∼100-fold in −Fe (+BPS) medium compared with +Fe medium (+FAS) in the WT strain (Fig. 7B). The FRP1 mRNA induction, however, was significantly reduced in all three SAGA mutants, namely gcn5Δ/Δ, spt7Δ/Δ, and spt20Δ/Δ (Fig. 7B), with the reduction between ∼2.0-fold in gcn5Δ/Δ and ∼10.0-fold in the spt20Δ/Δ mutant, suggesting that mutations in the structural components of SAGA had a major impact possibly due to the spt7Δ/Δ and spt20Δ/Δ mutations substantially affecting the integrity of the SAGA complex as reported previously for the budding yeast SAGA (
). The ACO1 mRNA expression had a fold change (plotted as +Fe/−Fe ratio) of ∼15.0, indicating that ACO1 expression is repressed in −Fe medium compared with +Fe medium in the WT strain (Fig. 7B). In the gcn5Δ, spt7Δ, and spt20Δ mutants, unexpectedly, the ACO1 mRNA expression was further elevated in +Fe medium (Fig. 7B) indicative of a repressive function for SAGA at this promoter under Fe-replete conditions. In any case, these data provided evidence that the SAGA complex is a critical effector of iron homeostasis gene regulation in C. albicans.
To further establish a direct role for acetylation by Gcn5, we assessed nucleosomal histone acetylation at the two iron homeostasis gene promoters in WT and gcn5Δ mutant strains. ChIP assays were carried out from chromatin extracts prepared from WT and gcn5Δ mutant strains cultured under iron-deplete or iron-replete conditions, using either histone H3-K9 acetyl antibody or the total H3 antibody, and FRP1 and ACO1 promoters were probed using qPCR. The results depict that histone H3-K9 is hyperacetylated at the FRP1 promoter in −Fe medium compared with +Fe in WT cells (Fig. 7C). However, in the gcn5Δ mutant strain, the hyperacetylation was completely lost indicating that Gcn5 is the critical histone acetyl transferase under iron-deficient conditions (Fig. 7C). At the ACO1 promoter, however, we found hyperacetylation in +Fe consistent with a high level of ACO1 mRNA expression in +Fe compared with −Fe medium (Fig. 7C). The gcn5Δ abrogated H3K9 acetylation at ACO1 promoter in +Fe medium (Fig. 7C), demonstrating that SAGA-mediated acetylation is also critical for ACO1 expression in +Fe medium.
Fungal pathogens encounter vastly differing iron levels in human host environments. Accordingly, C. albicans and other fungal pathogens maintain iron homeostasis by modulating their intracellular iron pool through multiple iron acquisition pathways and by conserving iron pools by downregulating iron-requiring proteins. The iron homeostasis gene regulation is controlled by multiple transcriptional regulators in C. albicans and other fungi (
). There were also indications that this control could involve interlinked expression of the transcriptional regulators, but the extent of connectivity and the mechanism of their control were not understood. Therefore, in this work we rigorously examined the mutual transcriptional control of the HAP complex, Sef1, and the Hap43 under iron-deplete and iron-replete conditions. Moreover, how these iron-responsive regulators exert transcriptional control of iron homeostasis genes also was not understood.
We first probed the regulatory interactions within the HAP transcriptional regulators and found that the Hap2 and Hap5 formed a heterodimeric complex in cell extracts independent of the iron status in cells. However, the Hap2-Hap5 recruitment to FRP1 and ACO1 promoters was intimately linked to forming a heterotrimeric complex with the Hap3 subunit under iron starvation conditions. The expression of Hap3 in turn is controlled by Hap43 and Sef1 transcriptional regulators under iron starvation conditions, although the HAP3 homolog NFY-B is not known to be transcriptionally controlled in metazoans. However, C. albicans genome encodes an additional HAP3-like gene HAP31 (
). Therefore, to determine the interdependency, if any, of the HAP trimeric complex and Hap43 for promoter recruitment, we first carried out ChIP-qPCR validations for iron-deprivation-induced FRP1, FTR1, SIT1, and iron-deprivation-repressed ACO1 and CYC1 promoters (Fig. 3). Whereas Hap2, Hap3, and Hap5 were recruited to both classes of promoters, Hap43 was recruited to only the repressed gene promoters. By conducting the ChIP assays in the different HAP mutant strains, we found that Hap5 recruitment to FRP1 and ACO1 is dependent on Hap2 as well as Hap43 (Fig. 4, A and B), indicating that although Hap43 is not recruited to the FRP1 promoter, the HAP trimeric complex recruitment is Hap43-dependent. Given that Hap43 is bound to multiple iron uptake genes that are induced under −Fe conditions, we would not be able to rule out if the Hap43 cross-linking efficiency at the FRP1 promoter is weaker than that at other promoters identified in the Chip-ChIP study. Besides, Hap3 is essential for Hap2-Hap5 interaction in cell extracts (Fig. 2, C–F) indicating that the efficient assembly of the trimeric complex is also dependent on intersubunit interactions within the HAP complex. Thus, it is conceivable that the HAP complex assembly in C. albicans is a two-step process similar to that in Aspergillus spp. and mammals (
). Together these results highlight a mutual dependence between the subunits of the HAP complex and between Hap43 and the HAP complex for promoter recruitment.
It was previously reported that ∼7% of the iron-responsive gene promoters are bound by both Hap43 and Sef1 indicating a synergy between the two regulators. Therefore, to test the interdependency of Sef1 and the HAP complex for their recruitment to FRP1 promoter, we mutated the binding site for the HAP complex (CCAAT box) and the presumptive Sef1-binding site (CGG motif) on the FRP1 promoter and found that while the HAP complex recruitment was dependent on Sef1, the recruitment of Sef1 itself was independent of the HAP complex indicating that Sef1 acts upstream to the HAP complex at the FRP1 promoter. As Sef1 has been shown to be recruited to the HAP3 promoter and induce transcription (
), Sef1 likely promotes the assembly and recruitment of the HAP complex to the FRP1 promoter (Fig. 8B). Furthermore, the Sef1 bound at the ACO1 promoter could also potentially promote the recruitment of the HAP trimeric complex and Hap43 to the ACO1 promoter (Fig. 8B).
A previous nucleosome mapping study identified a reduced nucleosome occupancy at the FRP1 promoter under Fe-deplete conditions (
). These differential nucleosome occupancy data are consistent with the transcriptional states of the two promoters under Fe-deplete and Fe-replete conditions (Fig. 8, A and B) indicating a chromatin-mediated control of iron homeostasis. Our results established that the SAGA complex subunits encoded by GCN5, SPT7, and SPT20 are required for cellular growth under Fe-depleted conditions and for optimal regulation of FRP1 and ACO1 promoters under Fe-deplete and Fe-replete conditions respectively (Fig. 7, A and B). Although FRP1 expression was impaired in all three SAGA mutants tested, the spt7Δ and spt20Δ mutations had a greater impact on gene expression as compared with gcn5Δ (Fig. 7B), with spt20Δ being most debilitating. There are compelling previous reports indicating such differential SAGA subunit effects. Proteomics studies of ySAGA (
) on the different ySAGA subunit deletions showed that spt7Δ and spt20Δ impacted a larger number of genes at global level as compared with gcn5Δ. Additionally, the single subunits deletions (gcn5Δ or spt7Δ or spt20Δ) have less effect on mRNA levels than the double deletions (gcn5Δ spt3Δ). Moreover, deletion of subunits of a particular SAGA module showed similar expression profiles, whereas deletion of subunits of different SAGA modules impacted different subset of genes (
). So again, based on the different effects of spt7Δ, spt20Δ, and gcn5Δ on FRP1 and ACO1 expression, there seems to be different requirements for SAGA in activation and repression, also indicating that multiple SAGA functions could be invoked in gene expression control of iron homeostasis.
The Sef1, HAP complex, and Hap43 promote H3-K9 hyperacetylation of the FRP1 promoter under Fe-deplete conditions (Fig. 6A). Furthermore, Sef1 and the HAP complex are also required for Pol II recruitment at the FRP1 promoter under Fe-deplete conditions (Fig. 6C). At the ACO1 promoter, however, Hap5 and Hap43, but not Sef1, are required to maintain this promoter in a hypoacetylated state under Fe-deplete conditions (Fig. 6B), consistent with the repressive function of the Hap43-HAP complex at the ACO1 promoter. Consistent with the transcriptional repressive function of the Hap43-HAP complex at ACO1, Pol II recruitment is greatly stimulated upon hap5Δ/Δ and hap43Δ/Δ deletion under Fe-deplete conditions (Fig. 6D). Furthermore, we are surprised that the hap43Δ/Δ mutation led to a greater than WT level of Pol II recruitment at the FRP1 promoter (Fig. 6C); but the reduced FRP1 mRNA level in the hap43Δ/Δ mutant perhaps suggests that Pol II recruitment is not sufficient to activate the FRP1 promoter under Fe-deplete conditions indicating a postrecruitment control of Pol II activation for FRP1 transcription. We posit that the HAP complex could be minimally sufficient to directly stimulate H3-K9 acetylation upon FRP1 promoter binding, and Sef1 could just mediate HAP complex recruitment. In this context, other transcriptional activators stimulate H3 acetylation by SAGA recruitment through activation domain interactions (
), although an activation domain is yet to be identified in either Sef1 or the HAP complex.
Based on the data presented in this study together with previous studies, a model emerges wherein iron deprivation elicited the transcriptional induction of SEF1 and HAP43 mRNAs as well as the repression of SFU1 mRNA expression. Sef1 and Hap43 in turn upregulated the transcription of HAP3 and HAP2 under iron-deplete conditions leading to a functional HAP trimeric complex. Under +Fe conditions, however, SEF1 and HAP43 transcription is repressed. The GATA factor Sfu1 is a negative regulator of iron uptake genes (
). Moreover, as the HAP complex and Sef1 are absent under +Fe, the FRP1 promoter remains deacetylated (Fig. 8A). Paradoxically, the loss of HAP complex in +Fe led to a hyperacetylated state of ACO1 promoter (Fig. 8B). It is conceivable that the ACO1 hyperacetylation in +Fe is likely due to SAGA activity recruited by Hap43, as the deletion of HAP43 (Fig. 6B) or GCN5 (Fig. 7C) impaired H3K9Ac at ACO1 under +Fe condition.
Under −Fe conditions, a functional Hap2-Hap3-Hap5 trimeric complex and Sef1 are recruited to FRP1, while the HAP trimeric complex, Hap43, and Sef1 are recruited to the promoter of ACO1 (Fig. 8B). The recruitment of the trimeric HAP complex and Sef1 stimulates SAGA activity and Pol II recruitment at FRP1 leading to transcriptional activation of FRP1 (Fig. 8B). Although Hap43 is not occupied at FRP1, Hap43 is required for HAP3 expression and thereby stimulates HAP recruitment to FRP1. The Sef1 regulator induces the expression of HAP43 as well as HAP3 under −Fe conditions and thus also stimulates HAP complex assembly at the FRP1 and ACO1 promoters (Fig. 8, A and B). In contrast, recruitment of trimeric HAP complex and Hap43 to the ACO1 promoter leads to inhibition of SAGA activity and the consequent loss of ACO1 expression in −Fe medium (Fig. 8B). Thus, the Hap2-Hap3-Hap5 trimeric complex and Hap43 have dual roles in regulation of iron homeostasis by transcriptional activation at FRP1 and repression of ACO1 (
Together the data presented here indicated that C. albicans has evolved a fail-safe mechanism for iron homeostasis transcriptional control by incorporating the HAP complex, Sef1, and Hap43, together with SAGA-mediated chromatin modification, as subswitches in turning on the transcriptional control in this highly pervasive human fungal pathogen. The SAGA-mediated control of iron homeostasis gene expression also provides a highly tunable system for a rapid response of C. albicans to changing iron milieu in the host. It would be fascinating to explore what elements of this intricate control are also operative in the vastly differing iron levels in mammalian tissues that are infected by C. albicans. Given that the genes encoding the HAP complex and Sef1 and the SAGA subunit genes ADA2, GCN5, SPT20 and SPT3 are required for C. albicans virulence (
). Iron-depleted media used was either SC-LIM or YPD with 100 μM BPS or 200 μM BPS respectively. For iron-replete media, 100 μM FAS (SC-LIM) or 200 μM FAS (YPD) was supplemented to deplete media. C. albicans cells were cultured in either SC-LIM or YPD media and grown for 14–16 h at 30 °C and diluted to an initial OD of 0.25 (WT) or 0.5 (mutants) and grown for 5 h and harvested, and cell pellets were used immediately or snap frozen in liquid nitrogen and stored at −80 °C till use.
Construction of strains and plasmids
The C. albicans strains and plasmids used in this study are listed in the Table S1 and S2. All strains in this study are derived from C. albicans parental strain SN152 and SN95 (
). Oligonucleotides used in this study are listed under Table S3.
Details regarding construction of plasmids and C. albicans strains are available on request. Briefly, genes encoding the HAP complex subunits Hap2, Hap3, and Sef1 were genomically TAP-tagged in C. albicans strain SN152 using the plasmids Ip21 and Ip22 (
) as DNA templates and ONC645/646, ONC649/650, ONC637/638 respectively as primers and the tagging cassettes were amplified by PCR using Phusion HF DNA polymerase. Correct integration of the tagging cassettes in C. albicans strains KNC398, KNC410, and KNC379 was confirmed by PCR, and protein expression was confirmed by western blotting using rabbit polyclonal anti-TAP antibody (Thermo Fisher). The Hap5-TAP strain RPY453 was described previously (
). The mutant strains KNC430, KNC453, and RPY471 were constructed by deleting sef1Δ, hap2Δ, and hap43Δ in strain RPY453. The mutant strain KNC402 was constructed by TAP tagging the SEF1 gene in strain RPY431. The various deletions were confirmed by PCR and the expression of TAP-tagged proteins by western blotting.
The strains KNC692 (WT) and KNC693 (hap2Δ) bearing Hap5-TAP and Hap3-FLAG were constructed by integration of the His6-FLAG3 cassette from plasmid pSH26-5 (
) into the HAP3 locus in strains RPY453 and KNC453 respectively. The strain KNC553 bearing Hap2TAP and Hap5-FLAG was constructed by integration of the His6-FLAG3 cassette from plasmid pSH26-5 into the HAP5 locus in strain KNC398. The strain KNC558 was constructed in multiple steps as follows. First, the HAP5 locus was FLAG-tagged in strain RPY431, and subsequently, the HAP2 locus was TAP-tagged to obtain strain KNC558.
To construct gcn5Δ/Δ, spt7Δ/Δ, and spt20Δ/Δ deletion strains, we carried out sequential deletion of both alleles using the HIS1-ARG4-HIS1 (HAH) deletion cassette from plasmid pHAH1 (
), using primers for GCN5 (ONC601/ONC115 and ONC601/ONC114), SPT7 (ONC602/ONC115 and ONC603/ONC114), or SPT20 (ONC605/ONC115 and ONC606/ONC114) and amplified the respective up-split and down-split fragments, combined and transformed into C. albicans strain SN95 and selected for Arg+ colonies. The correct integrations were confirmed using PCR, and the resultant heterozygous strains KNC641, KNC651, and KNC653 were grown nonselectively in YPD and Arg+ His+ colonies representing homozygous deletion strains KNC642, KNC652, and KNC654 were selected and confirmed by PCR.
For reintegration of GCN5, SPT7, and SPT20 into the homozygous deletion strains KNC642, KNC652, and KNC654, plasmid pNIM1 was sequentially digested with BglII/MluI to obtain nourseothricin resistance marker CaSAT1 while plasmid CIp10 was digested with NotI/XbaI and end filled with Klenow. Then linearized plasmid pCIp10 and insert CaSAT1 was ligated and transformed to electrocompetent bacterial cells DH10B to obtain plasmid pCIp10-SAT1(pPI1). Then, genes were amplified along with 1000 bp upstream and 500 bp downstream of ORF sequences using primer pairs ONC1156-ONC1157 (GCN5), ONC1154-ONC1155 (SPT7), and ONC1152-ONC1153 (SPT20) respectively. The amplified product was digested with BamHI (for SPT7 and SPT20) and SacI (for GCN5) and cloned into plasmid pPI1 digested with the respective enzymes and plasmids pPI2 (GCN5), pPI3 (SPT7), and pPI4 (SPT20) were obtained.
The plasmids pPI2, pPI3, and pPI4 were linearized with PmeI, BstEII, and Afl II respectively and transformed into KNC642, KNC652, and KNC654 strains, and positive transformants were confirmed by PCR using ORF specific primer pairs ONC661/ONC662, ONC663/ONC664, ONC708/ONC709 and obtained PRI1 (gcn5Δ/GCN5), PRI2 (spt7Δ/SPT7), and PRI3 (spt20Δ/SPT20) respectively.
Strains for β-Gal assays were constructed as follows. The PACT1::lacZ-ACT1t insert from plasmid pAU36 (
) to construct pNA36. A 1kb HAP43 promoter region was amplified from SC5314 strain genomic DNA and cloned as a KpnI-SalI fragment into the KpnI-XhoI digested pNA36 to replace PACT1 and obtained plasmid pNA40. The plasmid pNA40 was linearized with StuI and transformed into SN152, RPC75, and TF015 (sef1Δ) by electroporation to integrate the reporter plasmid at the RPS1 locus and obtained KNC332, KNC340, and KNC348. The correct integration was confirmed by PCR.
The strains bearing WT and mutant FRP1 promoters were constructed in several steps. First, the WT FRP1 coding sequence along with 500 bp upstream region was deleted in strains RPY453 and KNC379 to obtain strains KNC467 and KNC481 respectively. The homozygous frp1Δ deletion was confirmed by PCR. Next, a DNA fragment containing the FRP1 coding sequence flanked by 1kb upstream region and 300 bp downstream region was amplified as a KpnI-SacI fragment using ONC630-ONC631 and cloned into pLitmus28 to obtain pNA51. The WT FRP1 fragment was then subcloned into pSN69 (
) and obtained plasmid pNA61, which was integrated at the native FRP1 locus in strain KNC467 and KNC481 and obtained strains KNC479 and KNC486 respectively. The correct integration was confirmed by PCR.
To construct point mutations in the HAP complex binding site (CCAAT motif) in the FRP1 promoter, the mutagenic oligonucleotide pair ONC901-ONC902 was designed using the Quick Change Primer Design tool (Agilent Technologies) and the CCAAT motif (−134 with reference to ATG) was mutated to TGCGC as reported previously (
). The plasmid pNA61 was used as a template along with the mutagenic primers and amplified using the Quick-Change II XL Site-Directed Mutagenesis Kit and obtained plasmid pNA68. Similarly, the putative Sef1-binding site (
) was mutated using the mutagenic oligonucleotides ONC929-ONC930 and ONC931-ONC932 and obtained plasmid pNA76. The WT and the two mutant insert sequences in the plasmid clones were verified by Sanger sequencing. Next, the plasmids pNA68 and pNA76 were linearized with Bpu101, gel purified, and integrated at the chromosomal frp1 locus in strains KNC467 and KNC481 and obtained strains KNC494, KNC498, KNC505, and KNC509. The correct integrations were confirmed by PCR.
Briefly, three independent transformants of each strain were grown overnight in SC-LIM and then diluted to 30 ml SC-LIM plus 100 μM BPS (−Fe) at a starting density of 0.3 OD600 and 0.5 OD600 for WT and the mutant strains (hap43Δ/Δ and sef1Δ/Δ), respectively. For iron-replete conditions (+Fe; SC-LIM plus 100 μM BPS and 100 μM FAS), cells were diluted to a starting density of 0.25 OD600 for all the strains and grown for 5 h till ∼1.0 OD. Cells were harvested and cell pellets washed with Z-buffer (
), resuspended in 200 μl Z-buffer with 2 mM PMSF, and vortexed for 2 min in cold room using a Braun homogenizer. Supernatants were collected and protein extracts were cleared by centrifugation at 13,000 rpm at 4 °C for 15 min. The total protein concentration in the cell lysates was determined by Bradford assay using Softmax Pro 2.6 program in SpectraMax 340 microplate reader (Molecular Devices). The β-galactosidase kinetic assay was performed at 37 °C in 96-well microplates in SpectraMax 340 microplate reader with ONPG as substrate (0.75 mg/ml), absorbance monitored, and the β-galactosidase specific activity was calculated as described previously (
) and plotted using GraphPad Prism 6. Statistical analysis of Hap43 protein level and the β-galactosidase activity was performed by comparing expression in −Fe compared with +Fe and statistical significance determined by Student's t-test.
Whole-cell extract preparation, coimmunoprecipitation, and Western blotting
The cell pellets were resuspended in chilled Winston Buffer (40 mM HEPES-NaOH pH 7.5, 350 mM NaCl, 10% glycerol, 0.1% Tween 20 (
) with added protease inhibitors (2.5 μg/ml Aprotinin, 2 mM Benzamidine, 1 mM Dithiothreitol, 2 μg/ml Leupeptin, 2 μg/ml Pepstatin, 100 μM PMSF, 10 μg/ml TPCK, 10 μg/ml TLCK), and vortexed in presence of glass beads. The lysates were cleared by centrifugation at 13,000 rpm for 15 min at 4 °C.
Five milligrams of whole-cell extracts was used for immunoprecipitation with IgG-Sepharose 6 Fast Flow beads (GE healthcare) for 4 h at 4 °C, washed with Winston buffer, followed by TEV buffer. The beads were then incubated with AcTEV protease (Invitrogen) overnight at 4 °C. The eluate was collected and 20% of total eluate (IP) and 100 μg whole-cell extracts (Input) were separated on 10% SDS-PAGE gel, transferred onto a Protran nitrocellulose membrane (GE healthcare) and probed with anti-FLAG, anti-TAP oranti-Glucose-6-P-dehydrogenase antibodies detected with ECL-Prime Western blot detection reagent (GE Healthcare) and exposed to X-ray film.
Total RNA was isolated from C. albicans cells and cDNA was synthesized essentially as described before (
). Briefly, all strains were precultured overnight in SC-LIM medium, grown in fresh SC-LIM medium with 100 μM BPS alone or with BPS and 100 μM FAS for 5 h at 30 °C to an OD600 of 0.8–1.0, cross-linked with 1% (v/v) formaldehyde, quenched with 125 mM glycine, cells harvested, and chromatin extracts prepared by shearing in Bioruptor (model UCD 300, Diagenode). For chromatin immunoprecipitation of TAP-tagged proteins, the sheared chromatin extracts equivalent to ∼25OD600 cells from control and experimental strains were immunoprecipitated with about 25 μl of IgG-Sepharose 6 Fast Flow (pre-blocked with 0.2 mg/ml sheared Calf thymus DNA and 0.1% (w/v) BSA) for 3 h at 4 °C. For Hap43 ChIP, 2 μl affinity-purified antibody (
) was bound to 30 μl beads and immunoprecipitated for 4 h at 4 °C. The chromatin immunoprecipitations were conducted using chromatin extract from ∼35 OD600 equivalent cells and 2 μg of each antibody prebound to Dynabeads Protein G. The following antibodies were used in ChIP: anti-total H3, anti-H3-K9Ac, and RNA pol II CTD phospho Ser5 antibody (4H8, Abcam).
Immunoprecipitation was carried out for 9 h at 4 °C, washed, and eluted in 25 μl 0.1 × TE. The input DNA and immunoprecipitated DNA were purified, diluted 1:10,000 (Input DNA) and 1:5 (IP DNA), and probed by quantitative real-time PCR for specific regions of interest as well as for the control nonspecific region (ca21chr1_1573500–1574000). For all chromatin immunoprecipitations, enrichment was calculated for target regions and for the control nonspecific region (ca21chr1_1573500–1574000) and specific enrichment calculated with respect to input total chromatin. For the TAP chromatin immunoprecipitations, the fold enrichment was determined by further normalizing the background amount of immunoprecipitation obtained in mock immunoprecipitation conducted using chromatin extracts from untagged strain. Similarly, for Hap43 chromatin immunoprecipitations, the specific enrichment was calculated by further normalizing the background amount of immunoprecipitation obtained in the hap43Δ/Δ mutant chromatin immunoprecipitation.
The statistical significance of quantitative data was determined by Student's t-test in Graph Pad Prism 6, and a p-value ≤ 0.05 was considered statistically significant. The p-values are indicated as ∗ (p ≤ 0.05), ∗∗ (p ≤ 0.01), ∗∗∗ (p ≤ 0.001), ∗∗∗∗ (p ≤ 0.0001), and ns (p > 0.05).
The authors declare that they have no conflicts of interest with the contents of this article.
We thank the members of the Natarajan lab for discussions and Alan Hinnebusch, NIH, and Kaustuv Sanyal, JNCASR, for their insightful comments on an earlier version of the article. This study was supported by a grant from DST-SERB (EMR/2017/000161) and partial financial support under UGC-MRP (MRP-MAJOR-MICR-2013-35991) and through departmental funding from DST-PURSE, UGC-CAS (DRS-I), and DST-FIST II program grants. Article publication charge was funded by the SERB project grant and Jawaharlal Nehru University.
M. K. S. and N. A. methodology; M. K. S. and N. A. investigation; M. K. S. and N. A. writing–original draft; M. K. S., N. A., P. P., and K. N. formal analysis; M. K. S., P. P., and K. N. writing–review and editing; K. N. conceptualization; K. N. funding acquisition; M. K. S., N. A., P. P., and K. N. validation.
Funding and additional information
N. A. was supported by research fellowship from UGC, and M. K. S. and P. P. were supported by Junior and Senior Research Fellowships from CSIR.