The CCAAT-binding complex (CBC) is a highly conserved eukaryotic transcription factor, which is composed of three core subunits named NF-YA, NF-YB and NF-YC in mammals and plants, HapB, HapC and HapE in filamentous fungi and Hap2, Hap3 and Hap5 in yeasts (Zhao et al., 2016; Hortschansky et al., 2017; Li et al., 2018; Mao and Chen, 2019). In fungi, the CBC is constitutively bound to its target genes and requires a fourth regulatory subunit to control transcription in a condition-specific manner. For instance, in the model yeast Saccharomyces cerevisiae, CBC interacts with the Hap4 protein to positively control the expression of “respiratory genes” involved in respiration, TCA cycle, gluconeogenesis, heme biosynthesis and mitochondria biogenesis (Bolotin-Fukuhara, 2017). This allows this yeast to induce respiration specifically upon glucose deprivation or during growth in non-fermentable carbon sources such as glycerol. Hence, Hap4 is subjected to glucose catabolic repression and its expression level is under complex transcriptional and post-translational regulations (Forsburg and Guarente, 1989; Bouchez et al., 2020). Hap4 interacts with the Hap5 core CBC subunit through a short 16 amino acids motif, called the Hap4-Like CBC binding domain (Hap4L), which is located at the N-terminus of the protein (McNabb and Pinto, 2005; Bolotin-Fukuhara, 2017). Of note, Hap4 does not have a specific DNA binding domain and its target selection is thought to occur indirectly through its association with the CBC. Yet, the non-specific interaction of Hap4 with the DNA surrounding the CCAAT box was shown to be required for its proper association with the CBC (McNabb and Pinto, 2005).

Clear orthologues of Hap4 have been found only in Hemiascomycetous yeasts (Petryk et al., 2014). However, in many other fungal species, including numerous pathogens, another CBC regulatory subunit, named HapX, has been identified and described as a major regulator of iron homeostasis in these species (Misslinger et al., 2021). HapX only shares with Hap4 the Hap4L domain, allowing its association with the CBC. Upon iron deprivation, the HapX-CBC complex acts as a repressor of genes involved in iron-dependent pathways, i.e. genes encoding iron-sulfur cluster and heme containing proteins or involved in the biogenesis and assembly of these two prosthetic groups, and genes encoding transporters responsible for vacuolar sequestration of iron (Hortschansky et al., 2007; Jung et al., 2010; Singh et al., 2011; Lopez-Berges et al., 2012; Kröber et al., 2016; Wang et al., 2019; Do et al., 2020). In the human and plant pathogens Aspergillus fumigatus, Fusarium oxysporum and Verticillium Dahliae, HapX is also responsible for the induction of some of these “iron consuming genes” in high iron conditions, therefore playing a key role in iron tolerance (Gsaller et al., 2014; Wang et al., 2018). However, this latter role of HapX seems to be less conserved across fungi than its function as a repressor upon iron deprivation (Skrahina et al., 2017).

HapX and its orthologues are transcriptionally induced by iron starvation and sense iron through conserved cysteine rich domains (CRD) in their C-terminal part (Hortschansky et al., 2007; Schrettl et al., 2010; Hsu et al., 2011; Singh et al., 2011; Lopez-Berges et al., 2012; Gsaller et al., 2014; Rietzschel et al., 2015; Kröber et al., 2016; Do et al., 2020). Importantly, HapX also contains a basic leucine zipper (bZIP) domain and belongs to the Yeast AP-1 (YAP) family of specific DNA binding transcriptional regulators (Tanaka et al., 2002; Singh et al., 2011; Rodrigues-Pousada et al., 2019). However, in contrast with the other members of this family, HapX does not specifically recognize Yap Response Elements (YRE, canonical sequence TTA(C/G)TAA) in its target promoters. The main consensus sequence found in HapX target promoters is the CCAAT box (CCAAT in most species, C(C/G)AAT in C. neoformans) which is bound by the CBC (Chen et al., 2011; Gsaller et al., 2014; Do et al., 2020; Furukawa et al., 2020). Still, the bZIP domains of HapX in A. fumigatus and its orthologue in the pathogenic yeast Candida albicans (named Hap43) are essential for growth in low iron conditions, optimal DNA binding and optimal gene expression regulation (Singh et al., 2011; Srivastav et al., 2018; Furukawa et al., 2020). Moreover, it was shown in A. fumigatus and A. nidulans that direct DNA binding of HapX to sequences downstream of the CCAAT box, although with a relatively low specificity and a high variability in the recognized DNA motifs, contributes to target selection and binding affinity of the CBC-HapX complex (Gsaller et al., 2014; Hortschansky et al., 2015; Furukawa et al., 2020). Consistent with the key role of iron in microbial proliferation and nutritional immunity of the host, HapX orthologues have been shown to be important virulence factors in several humans, insect and plant fungal pathogens (Jung et al., 2010; Schrettl et al., 2010; Hsu et al., 2011; Lopez-Berges et al., 2012; Peng et al., 2020).

Candida glabrata is an opportunistic human pathogen, which is phylogenetically closely related to the model yeast S. cerevisiae (Gabaldon and Fairhead, 2019). As such, its genome encodes a clear orthologue of Hap4 but earlier studies did not identify obvious C. glabrata equivalents of HapX (Petryk et al., 2014), which was consistent with the fact that the regulatory networks controlling iron homeostasis have significantly diverged between Saccharomycetaceae yeasts (e.g. S. cerevisiae and C. glabrata) and other fungal species (e.g. C. albicans, A. nidulans, C. neoformans, S. pombe, etc.) (Gerwien et al., 2018). However, by combining high-throughput chromatin immunoprecipitation experiments (ChIP-seq) with transcriptomics and co-immunoprecipitation analyses of the Hap5 CBC core subunit, we previously showed that the C. glabrata CBC has a dual role in respiration and iron homeostasis (Thiébaut et al., 2017). The C. glabrata CBC is required for the expression of respiratory genes and it is essential for their strong induction upon growth with non-fermentable carbon sources. As in S. cerevisiae, this function seems to require Hap4, since the deletion of HAP4 severely reduced the expression of the ATP2 gene, encoding the beta subunit of the mitochondrial ATP synthase and taken as a model for respiratory genes in C. glabrata (Thiébaut et al., 2017). In addition to this “S. cerevisiae-like” function, and more unexpectedly, the C. glabrata CBC was found to bind the promoters of genes known in S. cerevisiae to be required for adaptation to toxic iron concentrations (Thiébaut et al., 2017). More specifically, these iron related targets of C. glabrata CBC include the vacuolar iron transporter encoded by CCC1, the glutaredoxin involved in iron signaling and iron sulfur cluster assembly encoded by GRX4, the iron-containing enzymes aconitase (encoded by ACO1), sodium dehydrogenase (encoded by SDH2), glutamate synthase (encoded by GLT1) and wybutosine synthase (encoded by TYW1), the essential iron-sulfur cluster containing termination factor encoded by RLI1, a protein involved in the maturation of iron-sulfur cluster mitochondrial proteins (encoded by ISA1) and the porphobilinogen deaminase (encoded by HEM3) which catalyzes a key step in the biosynthesis of heme. The deletion of HAP5 in C. glabrata had few impact on the expression of these “iron tolerance genes” in optimal growth conditions, but it completely abolished their induction in response to toxic concentrations of iron. However, in contrast with the role of the CBC in other fungal pathogens, Hap5 had a very limited impact on gene expression in iron starvation conditions in C. glabrata (Thiébaut et al., 2017).

In S. cerevisiae, the high iron induction of iron tolerance genes had previously been shown to be dependent on Yap5 (Li et al., 2008; Li et al., 2011; Pimentel et al., 2012). Yap5 is a bZip transcription factor of the YAP family (Fernandes et al., 1997), which constitutively and specifically binds a canonical YRE motif in the promoter sequence of its target genes (Li et al., 2008; Tan et al., 2008; Pimentel et al., 2012). Like HapX, Yap5 senses iron excess by conserved CRDs at its C-terminus (Rietzschel et al., 2015). This interaction changes Yap5 conformation, turning it into a strong transcriptional activator (Li et al., 2008; Rietzschel et al., 2015). Unlike HapX, Yap5 has very few impact on gene expression upon iron deprivation (Pimentel et al., 2012; Merhej et al., 2016).

The role of Yap5 in iron tolerance was shown to be well conserved in C. glabrata and, until the discovery that Hap5 was also required for adaption to high iron in this species, Yap5 was thought to act on his own in this process (Merhej et al., 2016). However, careful examination of Yap5 sequence in C. glabrata revealed that it actually exhibits a truncated Hap4L domain lacking the last four C-terminal amino acids of the canonical motif (Merhej et al., 2015). This domain was shown to be required for Yap5 binding to its target promoters and to mediate interaction of Yap5 with Hap5 (Thiébaut et al., 2017). Moreover, the presence of Hap5 was absolutely required for Yap5 binding to its targets, explaining why the CBC is essential to the high iron response in C. glabrata (Thiébaut et al., 2017). These findings revealed that, although their roles in the regulation of iron homeostasis considerably diverged, Yap5 and HapX are actually homologous proteins harboring many similarities in their functioning, and that they probably derived from a common ancestor.

The two roles of CBC in C. glabrata were found to be independent: Hap4 was shown to be completely dispensable for the high iron induction of GRX4, used as a model for iron tolerance genes, and reciprocally, the deletion of YAP5 had no impact on the expression of ATP2, used as a model for respiratory genes (Thiébaut et al., 2017). Interestingly enough, the sets of iron consuming genes regulated by HapX in A. fumigatus, C. neoformans or C. albicans includes both orthologues of the C. glabrata respiratory genes supposed to be regulated by Hap4 (e.g. CYB2, CYC1, RIP1, SDH1, CYT1) and orthologues of the C. glabrata iron tolerance genes controlled by Yap5 (e.g. CCC1, GLT1, HEM3, ISA1, RLI1, TYW1, ACO1, SDH2) (Chen et al., 2011; Hsu et al., 2011; Singh et al., 2011; Gsaller et al., 2014; Do et al., 2020; Furukawa et al., 2020), suggesting that the C. glabrata CBC regulatory network might have evolved by rewiring the roles of the Hap4 and Yap5 CBC regulatory subunits from an ancestral HapX-like situation.

Still, the co-existence of these two active CBC regulatory subunits in C. glabrata raises an important question. Since Hap4 and Yap5 both have the ability to bind CBC, how do they distinguish between the CBC bound to respiratory genes and the CBC bound to iron tolerance genes? In other terms, what are the underlying mechanisms rendering the two pathways independent from one another and preventing functional interferences between them? In the present work, we addressed this question by combining directed mutagenesis, ChIP and gene expression analyses in various mutants. We showed that the regulation of iron tolerance genes by Yap5 is dependent, not only on its interaction with CBC, but also on its specific binding to YRE, thus restricting its action to promoters in which a CCAAT and a YRE are present close to one another. The presence of Yap5 at these promoters considerably strengthened the binding of CBC on these genes, therefore suggesting a strong synergy between the CCAAT-CBC-Yap5-YRE interactions. Additionally, ChIP-seq experiments identified significant Hap4 binding at respiratory and iron tolerance genes, indicating that Hap4 apparently does not distinguish between these two gene categories. However, in contrast to respiratory genes, Hap4 binding at iron tolerance genes was apparently much lower than Hap5 binding at those same promoters, suggesting that Hap4 is not the preferred partner of CBC on these genes. Moreover, transcriptomic analyses confirmed that Hap4 has no role in the high iron response, while it is required for proper expression of respiratory genes. Analyses of GRX4 expression levels in a HAP4 and YAP5 double deletion mutant indicated that the competition with Yap5 prevents Hap4 to act on the transcription of iron tolerance genes. More precisely, in the absence of Yap5, GRX4 showed a residual expression level which was independent of iron and dependent on Hap4, hence harboring a “respiratory gene-like” pattern of expression.

Gene duplication is a major source of functional diversification (Innan and Kondrashov, 2010; Voordeckers et al., 2015). However, the co-existence of two paralogues with similar biochemical properties but diverging functions can lead to potentially detrimental competition between the duplicates. This phenomenon has been named “paralogue interference” and is particularly relevant in the case of transcriptional regulators, which work very often as cooperative complexes (Baker et al., 2013). In previously described cases, this paralogue interference constraint was overcome by mutations in one or both paralogues, which prevented or biased the competition between them to allow sub-functionalization without losing important functions under strong selection pressure (Bridgham et al., 2008; Baker et al., 2013; Pougach et al., 2014). In many cases, this led to a complexification of the derived regulatory networks and, potentially, to the appearance of new regulatory modes and new functions (Voordeckers et al., 2015).

In the present work, we addressed this question by studying the interplay between Yap5 and Hap4, two regulatory subunits of the CBC in C. glabrata. Yap5 and Hap4 are not considered as paralogues per se, because their sequence similarity is very low. However, they both have a Hap4L domain which confers them the ability to interact with the CBC, making each other potential competitors for the regulation of CBC targets. Still, it was shown previously that the Yap5 and Hap4 pathways seemed to work independently (Thiébaut et al., 2017). In other terms, mechanisms evolved to avoid detrimental interference between these two regulatory networks. Based on a work on HapX in A. fumigatus, which showed that the C-terminal part of the Hap4L domain is involved in low specificity DNA binding and target selection of the HapX-CBC complex, it was proposed that Hap4 could similarly distinguish between iron tolerance promoters and respiratory genes in C. glabrata (Furukawa et al., 2020). Actually, our observations do not exclude a potential contribution of Hap4 to DNA target recognition, but they do not go into that direction either. Especially, our ChIP-seq experiments showed that Hap4 can be found at the promoters of both respiratory and iron tolerance genes, suggesting that it is not able to clearly discriminate between the two types of genes. Our results rather suggest that Yap5 is the main actor of the selectivity that is observed in this system. Yap5 seems to be dependent, not only on its interaction with the CBC bound to the CCAAT box, but also on the presence of a YRE located at 10 to 14 bp of the CCAAT box. This functioning of Yap5 is very probably the same in most Saccharomycetaceae species, as the CCAAT-YRE bipartite motifs in GRX4, CCC1, ISA1 and TYW1 promoters were remarkably conserved in this clade. This strict DNA binding specificity physically restricts Yap5 regulation to a handful of genes containing a bipartite CCAAT-YRE DNA motif and probably explains why Yap5 is not able to efficiently compete with Hap4 for the regulation of respiratory genes, which do not contain this bipartite motif. Conversely, this strict DNA binding specificity seems to provide Yap5 with an advantage in its competition with Hap4 for the regulation of iron tolerance genes. Hence, the presence of Yap5 seems to strengthen the binding of Hap5 to the promoter of GRX4. Moreover, according to our results, this is the presence of Yap5 which prevents Hap4 to act on the expression of GRX4, although this effect was not observed for other Yap5 targets, which suggests that other parameters besides the presence of Yap5 interfere with a potential Hap4 regulation of those genes. The only exceptions to this model are SDH2 and ACO1, which only exhibit a half-YRE motif close to the CCAAT box (TTAGT and TTAG, respectively) in their promoters but yet are bound and induced by Yap5 in response to iron excess. Interestingly, SDH2 and ACO1 have a hybrid regulation, being both respiratory genes controlled by Hap4 and part of the high iron response induced by Yap5. Hence, an equilibrium seems to have been selected for those two genes in the competition between Yap5 and Hap4, which would have been compromised by the presence of a perfect CCAAT-YRE in their promoters. In summary, the Yap5/Hap4 competition appears to be a quantitative and dynamic balance which is controlled by their respective capacity to interact with CBC on the one hand, and by the more or less strict specificity of Yap5, and possibly Hap4, for the DNA sequences surrounding the CBC CCAAT binding sites on the other hand ( Figure 6 ). Quantitative biochemical analyses of the Yap5-CBC-DNA and of the Hap4-CBC-DNA interactions using techniques such as surface plasmon resonance (Hortschansky et al., 2007; Hortschansky et al., 2015; Furukawa et al., 2020) will be required to better understand the molecular basis of this balance.

As mentioned in the introduction, although their similarity over their whole sequences is relatively weak, Yap5 and HapX have the same functional domains, i.e. a HAP4L domain, a bZip motif and CRDs (Merhej et al., 2015; Rietzschel et al., 2015; Thiébaut et al., 2017). Hence, although their roles in the regulation of iron homeostasis have significantly diverged, they very likely derive from a common ancestor and are, therefore, probably orthologues. Still, their DNA binding properties look significantly different. HapX has been shown to contribute to the target selectivity of the HapX-CBC complex with a relatively low specificity. More precisely, HapX-CBC recognizes a tripartite motif composed of a CCAAT box and of two motifs, RWT and TKAN, positioned at 12 bp and 11 to 23 bp from the CCAAT sequence, respectively (Furukawa et al., 2020). Some versions of the TKAN motif resemble a half YRE site, but clearly HapX does not require the presence of a canonical YRE to bind DNA and regulate gene expression (Hortschansky et al., 2015; Furukawa et al., 2020). So, since their bZip sequences are very similar, why are Yap5 and HapX DNA binding properties different? Based on the available data, at least two complementary hypotheses can be proposed. First, the last four amino acids at the C-terminus of Yap5 Hap4L domain have diverged from the canonical sequence found in HapX (Merhej et al., 2015). This truncated Hap4L may weaken the interaction of Yap5 with CBC and/or with the DNA sequences surrounding the CCAAT (Furukawa et al., 2020). Hence, the specific interaction with a YRE close to the CCAAT may act as a compensatory mechanism to increase the CBC-Yap5-DNA binding energy. In support of this hypothesis, we observed in a previously published study that, while the deletion of HAP5 or of the whole Hap4L in YAP5 completely abolished Yap5 binding to GRX4, single mutations in important positions of the Hap4L domain in Yap5 severely diminished its interaction with GRX4, but a significant ChIP enrichment was still detectable [ Figure 4B in Thiébaut et al. (2017)]. This result suggests that, in the presence of a CCAAT-YRE DNA motif, non-optimal Yap5-CBC interactions can be sufficient to ensure a minimal binding efficiency of Yap5 to its targets. Second, HapX is, to the best of our knowledge, the only Hap4L protein encoded by the A. fumigatus genome, while Yap5 had to cope with the potential competition of Hap4 for CBC binding and gene target regulation. Hence, as discussed above, the accumulation of mutations in Yap5 Hap4L domain on one hand and the presence of a bipartite CCAAT-YRE motif in the promoters of the Yap5-CBC targets on the other hand, may have been selected to resolve this potential conflict between the two CBC regulatory subunit, by providing an advantage to Yap5 over Hap4 at the promoters of the iron tolerance genes, while disadvantaging Yap5 at the promoters of respiratory genes. In this matter, it would be very interesting to study the interplay between Hap4L containing proteins in non-Saccharomycetaceae yeast species such as C. albicans. In this species, there are two Hap4 homologues (Hap41 and Hap42) and one Hap4L-bZip protein (Hap43) which is clearly more similar to HapX than to Yap5, both for its sequence and for its functioning (Hsu et al., 2011; Singh et al., 2011; Merhej et al., 2015). So, other solutions must have been selected to avoid competition between these three Hap4L proteins. Unfortunately, the roles of Hap41 and Hap42 are largely unknown and Hap43 DNA interaction properties have not been investigated in detail. Hence, we have little information to understand the mechanisms at play in this case, besides the fact that Hap43 can act as a repressor of the expression of HAP41 (Chen et al., 2011).

The hypothesis of a co-evolution of the Hap4 and Yap5 regulons shed a new light on the evolution of the respiratory and iron homeostasis regulatory networks in yeast. Interestingly enough, the yeast species which have a Hap4L-bZip protein of the Yap5 type (i.e. Saccharomycetaceae), are also the ones strongly enriched for crabtree positive species (Dashko et al., 2014; Hagman and Piskur, 2015). Moreover, the Saccharomycetaceae species in which the CCAAT-YRE motif was less conserved, i.e. Kluyveromyces sp and Eremothecium sp ( Figure 2A ), are the ones which are less prone to use crabtree metabolism according to Hagman et al. (Hagman et al., 2013). The crabtree effect is defined as the capacity to favor fermentation over respiration in the presence of sugars, even in aerobic growth conditions. The induction of the respiratory genes by Hap4 is an important aspect of the crabtree metabolism (Perez-Samper et al., 2018). Hence, by allowing the divergence of the regulation of the respiratory genes and the iron tolerance genes, the resolution of the potential interference between Hap4 and Yap5 possibly played a key role in the fixation of this new metabolic feature in Saccharomycetaceae. Reciprocally, it is tempting to speculate that the positive selection of this new regulation of respiration has been a strong driving evolutionary force for the specialization of Yap5 in the regulation of the high iron response in these species. A collateral consequence of this process was that Yap5 lost its role in the repression of iron consuming genes upon iron starvation, which is the common feature of all other HapX homologues (Devaux and Thiébaut, 2019). However, this regulation, which is crucial for the adaptation to iron scarce conditions, is ensured at the post-transcriptional level by the Cth2 RNA binding protein in extent Saccharomycetaceae species, which compensate for the absence of a “HapX-like” regulation of those genes (Puig et al., 2005; Gerwien et al., 2016).

In conclusion, we showed that Hap4 has the potential to interfere with the regulation of Yap5, since Hap4 is found at the promoters of both respiratory genes and iron tolerance genes. This potential interference between the two regulatory pathways is minimized by the strict DNA binding specificity of Yap5 for CCAAT-YRE motifs, which, on one hand prevent Yap5 to bind to most respiratory genes and, on the other hand, strengthen the Yap5-CBC-DNA interaction at the promoters of iron tolerance genes. Consequently, the Hap4 and Yap5 pathways seem to work independently in C. glabrata. Hence, the resolution of the Hap4/Yap5 interference led to the diversification of the regulatory network in C. glabrata, and probably in most Saccharomycetaceae species, by allowing the acquisition of a Hap4-dependent regulation of respiration, while considerably restricting the set of targets of Yap5 as compared to its HapX homologues. We propose that the positive selection of a new metabolic capacity, i.e. the crabtree effect in the Saccharomycetaceae, has shaped the evolution of the iron homeostasis regulatory network in budding yeasts.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: https://www.ebi.ac.uk/arrayexpress/, E-MTAB-https://www.ebi.ac.uk/arrayexpress/, E-MTAB-10675, E-MTAB-10544.

TD performed the ChIP-Q-PCR and RT-Q-PCR analyses, constructed the strains and performed the directed mutagenesis of the GRX4 promoter. MB and AT performed the ChIP-seq experiments. AT performed the ChIP-seq raw data treatment and FD did the peak calling step. AT and FD performed the GRX4 multispecies sequence analyses. FD performed the microarray experiments. AT did the RT-Q-PCR analyses on the LacZ reporter. JM constructed the pSG plasmid and wore a PSG tee-shirt for most of his post-doc. FD interpreted the results, submitted the sequencing and microarray data to public repositories, made the figures and wrote the manuscript. All authors contributed to the article and approved the submitted version.

This work was funded by the Agence Nationale pour la Recherche (CANDIHUB project, grant number ANR-14-CE14-0018-02).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.