The transcriptional response to environmental alkalization involves both Komagataella phaffii Pho4 transcription factors

Abstract

Background

Pho4 is a fungal transcription factor essential for the response to phosphate (Pi) starvation in many fungi and is usually encoded by a single gene. However, we recently reported that the methylotrophic yeast Komagataella phaffii encodes two functional Pho4 proteins, named Pho4(A) and Pho4(B), and that only Pho4(B) was required for the transcriptional response to Pi limitation. Pho4 is also important in many yeasts for adaptation to an alkaline environment. We evaluate here the participation of Pho4(A) and Pho4(B) in the transcriptional response to alkaline pH.

Results

We show that the expression of nearly one-hundred genes is altered in a pho4(A) pho(B) mutant. In contrast to Pi starvation, Pho4(A) becomes important under high pH stress for the short-time expression of genes required for Pi homeostasis, such as PHO89, VTC1, VTC2, VTC4 and PHO4(B), suggesting that both transcription factors respond to different signals. Integration of the transcriptional data with that obtained from crz1 or rim101 mutants shows that nearly 80 genes are regulated by at least two of these transcription factors. Thus, we show that induction of PHO89 by Pi starvation or high pH is fully abolished in both double pho4(A) pho4(B) and single rim101 mutants, and that mutation of CRZ1 and RIM101 partially eliminates the sensitivity to high pH of the pho4(A) pho4(B) strain, revealing unexpected genetic interactions between these mutations. Finally, we identify an alkali-sensitive region in the PHO89 promoter and demonstrate that its duplication further increases the promoter’s response to alkalinization.

Conclusions

The transcription factor Pho4(A) is not required for the transcriptional response to Pi starvation, but it contributes to the response to alkalinization. Gene expression remodeling upon alkalinization in K. phaffii is controlled by diverse signaling pathways, including Pho4, Crz1 and Rim101 as well as other components still to uncover. Detailed knowledge of the signaling mechanisms triggered upon alkalinization and promoter mapping of the genes responsive to high pH will be useful tools for the generation of alkaline pH-regulated hybrid synthetic promoters in K. phaffii with improved features.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12934-026-03054-2.

Background

Most yeasts grow better at acidic pH, and a sudden shift to slightly above neutrality in environmental pH triggers an adaptive response that comprises changes in the expression of hundreds of genes [1]. The mechanisms involved in these responses have been well characterized in a few fungi, such as Saccharomyces cerevisiae, Candida albicans, and Aspergillus nidulans [2–6], and these studies underscore the participation of several signaling pathways involving key transcription factors such as Rim101/PacC, Crz1, and Nrg1. However, these mechanisms are largely unknown in many other fungi.

Komagataella phaffii (formerly known as Pichia pastoris) is a methylotrophic yeast widely used for heterologous protein production. Its strengths are the capability to grow at high cell density, an efficient secretion of the produced proteins to the medium, and a relatively simple genetic manipulation [7–9]. Growth of K. phaffii on methanol as carbon source depends on the methanol utilization (MUT) regulon, whose first component, the AOX1 gene, encodes most of the alcohol oxidase activity. The AOX1 promoter is widely used to drive heterologous protein expression because of its strength and fine regulation. However, the use of methanol presents several constraints in addition to the toxicity and flammability of this carbon source [10–13]. These limitations have prompted diverse strategies to develop methanol-free expression systems [14–18]. In this regard, the study of the adaptation of K. phaffii to alkalinization revealed a wide and strong transcriptional response [19] and this information has been utilized to select several alkaline pH-inducible promoters that can efficiently drive expression of proteins of industrial interest [20]. A further step for development of improved expression systems in K. phaffii would be the generation of hybrid synthetic alkaline pH-inducible promoters, as was recently demonstrated in S. cerevisiae [21]. To this end, a detailed understanding of the signaling pathways and the transcription factors controlling gene expression under alkaline conditions is needed.

The adaptive responses to alkalinization and to phosphate starvation have many points in common. The mechanism involved in adaptation to inorganic phosphate (Pi) limitation has been characterized in several fungi and, in overall traits, it works as follows. Under conditions of sufficient phosphate, inositol phosphates inactivate the cyclin-dependent kinase inhibitor Pho81 [22, 23], thus allowing the activation of the Pho85-Pho80 cyclin-dependent kinase (CDK) complex [24, 25]. This complex phosphorylates the key basic helix-loop-helix (bHLH) Pho4 transcription factor (TF) at several sites, leading to its inactivation and cytosolic location [26–30]. When Pi becomes scarce, the Pho85-Pho80 kinase is inhibited by Pho81, leading to Pho4 dephosphorylation. Dephosphorylated Pho4 enters the nucleus [31], where it binds to conserved 5’-CACGTG-3’ sequences. In S. cerevisiae, Pho4 binding to consensus sequences requires the participation of another TF, Pho2, and this interaction increases the selectivity of the response to phosphate starvation [32]. This response is restricted to the induction of the expression of ≈ 20–30 genes that are essential to thrive under Pi limitation (the so called PHO regulon), including various secreted phosphatases, such as PHO5 and PHO11, high affinity Pi transporters, such as PHO84 (encoding a H⁺/Pi cotransporter) and PHO89 (a Na⁺/Pi cotransporter), as well as components of the vacuolar polyphosphate synthetic machinery (VTC1, VTC2/3, and VTC4). However, in other fungi the requirement for Pho2 cooperation differs. Thus, C. albicans and C. glabrata contain PHO2 homologs but their mutation does not perturb the ability of these fungi to grow under Pi limitation [33, 34]. In other fungi, such as Neurospora crassa and Cryptococcus neoformans, no Pho2 orthologs have been identified. Interestingly, it seems clear that in those organisms where Pho4 is less dependent on Pho2, the range of genes under the control of Pho4 is expanded and the TF becomes relevant for adaptation to a wide array of stress conditions [35].

As mentioned above, exposing cells to alkalinization invokes a response reminiscent to that of phosphate starvation. For instance, it has been shown that in S. cerevisiae, alkalinization causes nuclear translocation of Pho4 [4] and that the TF is required for pH-triggered induction of PHO84 and PHO12 [36]. In addition, lack of PHO2, PHO4 or PHO81 results in cells strongly sensitive to mild alkalinization [3, 37]. Pho4-deficient strains of C. albicans [38] and C. neoformans [39] also display growth defects at alkaline pH. It is commonly accepted that, when present, Pho4 is encoded by a single gene. However, we have shown very recently that K. phaffii is an exception [40], as it contains two functional Pho4 TFs, encoded by genes PAS_chr1-1_0265 and PAS_chr2-1_0177 (named PHO4(A) and PHO4(B), respectively). We demonstrated that Pho4(B) plays a major role in adaptation to Pi starvation, and we hypothesized that Pho4(A) might be important in mediating tolerance to diverse stresses, including alkaline pH. Cells lacking both TFs grow well under standard conditions, but they are extremely sensitive to a variety of stresses. For instance, whereas pho4(A) or pho4(B) mutants display normal tolerance to high pH, the double mutant grows poorly even at pH 7.0 [40]. Our laboratory has recently initiated the study of the signaling pathways underlying the transcriptional response to alkalinization [41] by generating mutant strains for the Rim101 and Crz1 transcription factors and demonstrating that both contribute to the remodeling of gene expression triggered by alkalinization. However, the mutation of CRZ1 or RIM101 did not alter the response of nearly 40% of the genes whose expression was affected by alkalinization, indicating that additional alkali-responsive pathways should exist. Therefore, we considered it necessary to investigate whether any of the K. phaffii Pho4 TFs could play a significant role in the transcriptional response to high pH. To this end, we took advantage of the recently constructed strains carrying single and double pho4 mutations [40] and performed a comparative transcriptomic profiling by RNA-seq. Our results indicate that Pho4(A) is more relevant than Pho4(B) in orchestrating the alkaline response and that, in combination, these TFs control around 100 genes of which about one third are also influenced by Crz1 and/or Rim101. We have also conducted phenotypic analyses of strains combining mutations in PHO4, CRZ1 and RIM101, which highlight unexpected genetic interactions between these mutations. Finally, we show that the identification of alkali-responsive regulatory elements allows generating improved versions of native promoters regulated by high pH.

Results

Pho4(A) and Pho4(B) differentially contribute to the alkaline pH transcriptional response

Exposure of wild-type K. phaffii cells to alkalinization (pH 8.0) for different times up to 60 min resulted in a wide transcriptional response, yielding 542 differentially expressed genes (DEGs) at least at one tested time point when compared with non-stressed cells. Of these 277 were induced and 265 were repressed. These overall figures were not significantly changed by mutation of PHO4(A), PHO4(B) or in the double mutant (Fig. 1A and Additional Table 1). The response in the wild-type strain was very fast and transient: 75% of the alkaline pH responsive genes were already differentially expressed after 15 min and this percentage decreased to around 50% after 60 min (compare Fig. 1A and B). This temporal profile was not altered by the individual mutations, but the transcriptional changes were more persistent over time in the double mutant, both for induced and repressed genes.

Fig. 1.

Overall transcriptional changes in response to alkalinization in wild-type cells and pho4 mutant strains. Wild-type X-33 (WT) and the indicated mutants were shifted to pH 8.0 and samples taken at T = 0 and after 15, 30 and 60 min were processed for RNA-seq. A The number of differentially expressed genes (DEGs), either induced (Ind) or repressed (Rep) at least at one time-point with respect to the wild-type strain at T = 0 (log2 ≥ 1 or ≤ -1 and p-value ≤ 0.01), was calculated and plotted. B The number of DEGs, induced and repressed genes were calculated for wild-type cells and the indicated mutants for each time upon shifting the cells to alkaline pH (8.0) in comparison with the WT strain at T = 0

Figure 2 presents a Gene Ontology analysis of these changes. Among the induced genes, an antioxidant response was clearly perceived, mostly due to peroxidase activity encoding genes, such as PRX1, AHP1, GTT1, CTA1 or TSA1. This response was transient except in double mutant cells, in which persisted after 60 min of alkalinization. An early induction was also observed for genes involved in ribosome biosynthesis (mostly genes encoding proteins required for rRNA processing) but it was not observed in pho4(A) pho4(B) mutant cells. Genes involved in sulfur metabolism and Cys biosynthesis were induced, and this effect was not affected by lack of the TFs, neither individually nor in combination. Among the repressed genes in wild-type cells, we found an enrichment in those involved in metabolism and energy generation. These combined glycolysis/gluconeogenesis (including three pyruvate decarboxylase isoforms and the ADH2 alcohol dehydrogenase) and tricarboxylic acid cycle related genes. This enrichment was not observed in the double mutant (Fig. 2 and Additional Table 1).

Fig. 2.

Gene Ontology analysis of induced and repressed genes in wild-type and PHO4 single and double mutants upon shifting cells to pH 8.0. Comparisons were made against the expression level in the wild-type (WT) strain at T = 0. KEGG, Kyoto Encyclopedia of Genes and Genomes. pho4(AB) denotes the pho4(A) pho4(B) mutant

An alternative view of the impact of the pho4 mutations in the response to alkalinization can be obtained by comparison of the expression of the genes in each mutant with that of the wild-type strain at a given time after the onset of the high pH stress (Additional Table 1). Fig. 3A shows that the impact of the mutation of PHO4(A) under alkaline pH stress, albeit limited (12 DEGs), was stronger than that of PHO4(B) (only 3 DEGs). The effect of the pho4(A) mutation was mainly observed after 15 min of stress and vanished with time. Mutation of both PHO4 genes multiplied the changes by one order of magnitude (98 DEGs in total) and these accumulated at later times. Of these genes, 22 were negatively affected by the pho4(A) pho4(B) mutation and 56 showed increased expression when compared with the wild-type strain. Analysis of the upstream regions of these genes for diverse TFs consensus binding sequences revealed a nearly three-fold excess of Pho4 signatures in the set of genes whose expression decreased in the mutant when compared with the set of induced genes, or with a set of randomly selected genes (Additional File 1, Figure S1). A less pronounced increase in Rim101 recognition sequences was also found, but no relevant differences were observed for Crz1, Nrg1, Mig1/2 or Aft1. The position of these regulatory elements is compiled in Additional File 1 (Figure S2).

Fig. 3.

Impact of the pho4(A) and pho4(B) mutations in the response to environmental alkalinization. A The mRNA levels of each mutant were compared at each specific time-point after shift to high pH with that of the wild-type strain, and the number of DEGs was plotted. The figures denote the total number of unique genes in each category. B GO analysis of the genes represented in panel A. MF, Molecular Function; BP, Biological Process; CC, Cellular Component. C Changes in the expression of genes involved in methanol utilization. FDH1, formate dehydrogenase, AOX1, alcohol oxidase, AOX100, alternative oxidase encoded by PAS_chr3_0408. The asterisk indicates that the associated p-value is > 0.01. The light-yellow box delimits the fold-change threshold. D Heat map showing the change in expression in response to Pi starvation and high pH of the 28 components of the PHO regulon reported in S. cerevisiae [42]. Low Pi vs. high Pi (-/+ Pi) values were taken from [40]. Data was hierarchically clustered (Euclidean distance, average linkage) with Gene Cluster v.3.0 [43] and clusters visualized using Java TreeView [44]

Gene Ontology analysis of the genes affected by the single or double pho4 mutations (Fig. 3B) highlighted several interesting features. FDH1 encodes a NAD⁺-dependent formate dehydrogenase catalyzing the conversion of formate to CO₂. This gene was repressed transiently by alkalinization (Fig. 3C). Such repression was relieved by mutation of PHO4(A), but not by that of PHO4(B), and the gene was potently induced in the double mutant. In contrast, other genes encoding members of the methanol utilization pathway, such as methanol oxidases AOX1 (Fig. 3C) or AOX2 (Additional Table 1) were not affected by alkaline pH or mutation of the TFs. However, PAS_chr3_0408, encoding an alternative oxidase (AOX100), was potently inhibited by alkalinization, albeit this effect was Pho4-independent (Fig. 3C). The effect of the double mutation on genes involved in the degradation of aromatic and branched chain amino acids (Fig. 3B), prompted us to examine how alkaline pH affected the Ehrlich pathway, that is, the 3-step pathway that converts these amino acids into fusel alcohols or acids. As shown in Additional File 1 (Figure S3), alkalinization transiently decreases the expression of genes involved in the transamination (ARO9 and BAT1) and decarboxylative (ARO10, PDC1) steps of the pathway. Notably, the elimination of both Pho4 TFs clearly attenuates this repressive effect. Genes involved in the subsequent reductive (ADH2, GRE2, GRE3) or oxidative (ALD6, ALD100) steps were not affected by high pH (except for ADH2, which was also repressed) but in all cases the absence of the TFs resulted in at least two-fold increase in expression (Figure S3).

The enrichment profile of the repressed gene set (Fig. 3B) reflects the impact of alkaline pH on phosphate metabolism. The specific effects caused by the mutations in response to alkalinization are presented in Fig. 3D, in comparison with recently reported transcriptomic data upon phosphate starvation [40]. It can be observed that numerous genes corresponding to the canonical PHO response, such as PHO89, VTC1, VTC2, or PHO81 were induced under both stresses. However, a few exceptions were detected. For instance, PHO5, VIP1, or GDE1 were only marginally induced or not at all at high pH compared with Pi starvation (Fig. 3D). DOG1, encoding a putative 2-deoxyglucose-6-phosphate phosphatase, was induced by Pi starvation but repressed by high pH, whereas PHO84-2 was repressed by both Pi starvation and alkaline pH (Fig. 3D).

In line with the profiles presented in Fig. 3A, evaluation of specific genes showed that at short times (T = 15 min) the impact of the PHO4(A) mutation in the response to alkalinization was more relevant than that of PHO4(B) (Fig. 4). This can be clearly seen for specific phosphate-related genes responsive to both Pi starvation and high pH, such as PHO89, VTC1, VTC2, VTC4 or even PHO4(B). In any case, most changes induced by high pH were fully lost in the double mutant suggesting that these transcription factors cooperate to play an important regulatory role for a subset of high pH-induced genes. There are, however, some exceptions. For instance, ENA2 (encoding a Na⁺-ATPase), was induced both by Pi starvation and by high pH, but whereas the induction in response to Pi starvation was largely lost in the pho4(A) pho4(B) strain, upon high pH stress this mutant expressed ENA2 nearly at wild-type levels (Fig. 4). A second example was PHO84-2, which was repressed to a similar extent by Pi deficiency and by alkalinization. However, whereas the effect caused by Pi shortage was largely dependent on Pho4(B), repression derived from high pH was fully maintained even in the double mutant. Notably, expression of PHO4(A) was barely affected by lack of Pi or high pH but was enhanced specifically by mutation of both TFs in the absence of Pi. Taken together, these results suggest that the promoters of these genes can integrate different signals elicited by diverse stimuli.

Fig. 4.

Comparative response to Pi starvation and high pH of selected genes in wild-type and pho4 mutants. The changes in expression (log2 Fold Change stress vs. no stress) in the wild-type strain (WT) and the different mutants in response to Pi starvation [40] and alkaline pH (this work) for diverse genes belonging to the PHO regulon and sulfur metabolism are plotted

An overview of the similarities and differences between the transcriptional effects derived from Pi shortage and alkalinization is presented in Fig. 5. Fifty-four genes were repressed by both conditions. This set was enriched in chaperone-related genes, which were repressed in a Pho4-independent fashion after 15 min of stress and gradually returned to basal levels (Additional Table 1). Sixteen genes were induced by high pH but repressed by Pi starvation, and this subset included numerous genes related to sulfur and sulfur-containing amino acid metabolism. These effects were independent on the presence or absence of both PHO4 genes, except for SUL1 and MET10, whose repression by low Pi was attenuated in the pho4(A) pho4(B) mutant (Fig. 4). Many genes encoding ribosomal proteins were repressed in low Pi but remained unaffected in high pH. In contrast, the set of genes induced specifically at alkaline pH but unaffected by low Pi was enriched in genes involved in antioxidant activity, as well as rRNA processing and ribosome assembly (notably, excluding most ribosomal protein encoding genes). None of the genes belonging to these two groups were affected by the pho4(A) pho4(B) mutation. Finally, genes repressed by alkaline pH but not affected by Pi starvation were mostly related to carbon metabolism and oxidative phosphorylation and were largely independent of the Pho4 transcription factors.

Fig. 5.

Overlap between genes regulated by Pi starvation and by alkalinization. The Venn diagram shows the number of genes induced or repressed by Pi starvation [40] and/or alkalinization (this work). Significantly enriched functional families (p-value < 0.05) were extracted with g: Profiler [45]. Green color indicates that expression of the specific genes was not altered by the pho4(A) pho4(B) mutation, whereas red color specifies that they were sensitive to it. The functional relationship between the proteins of several families was illustrated with STRING

Overall, our data indicates that, in combination, Pho4(A) and Pho4(B) control the expression of nearly 100 genes whose expression is affected by alkalinization, although the participation of Pho4(A) seems predominant. In addition, the expression of nearly 200 genes is affected by both Pi starvation and alkaline pH, in most cases with the same effect (induction or repression). These results reinforce the notion that both stress conditions share similar traits.

The integrative contribution of Pho4, Crz1 and Rim101 pathways to the transcriptional response to alkalinization

Figure 6A shows the integration of the transcriptomic data in response to alkalinization presented in this work with that of the recently reported crz1 and rim101 mutants [41]. The number of genes affected by the mutation of both Pho4 TF was similar to that found for the crz1 strain and, therefore, relatively small when compared with the extensive Rim101-dependent changes. About two-thirds of the Pho4-dependent genes (64%) were not dependent on Crz1 or Rim101. As shown in Fig. 6A and B, only one gene was dependent on Pho4 and Crz1, but not on Rim101 (PAS_chr2_1_0242, encoding a protein similar to the S. cerevisiae Cld1 phospholipase). Thirty-one genes were dependent on Pho4 and Rim101, but not on Crz1, and this set included seven genes whose expression was decreased by alkalinization in Pho4-deficient cells but resulted increased in the rim101 mutant. Interestingly, all of them were related to phosphate (PHO5, PHO84-1, PHO84-2) or iron (FTR1, FET3, FRE2-1, PAS_chr2-1_0103) homeostasis. Finally, four genes were affected by all four mutations. All of them were induced in the pho4(A) pho4(B) strain and the other two mutations gave all four possible induced/repressed combinations. The integration of both datasets yielded 863 genes induced or repressed by alkalinization, of which 535 (62%), are affected by the absence of at least one of these four TFs. It should be noted that in some cases the change in mRNA levels is observed even in the absence of stress (Additional Table 1). Altogether, the integration of the three pathways accounts for a substantial fraction of the global transcriptional response triggered by alkalinization.

Fig. 6.

Integrative overview of the PHO4, RIM101 and CRZ1 transcription factors on the transcriptional response to alkalinization. A Data from this work (Additional Table 1) and from reference [41] was integrated and the genes grouped according to the dependence of the response to alkaline pH to the indicated TFs. B The genes identified in this work as dependent on both Pho4(A) and Pho4(B) (denoted as pho4(AB)) were assigned according to their dependence (or absence of) from the Rim101 and Crz1 TFs. ↑, denotes induced by the mutation and ↓ indicates repressed by the mutation

The PHO89 promoter integrates multiple regulatory inputs under Pi starvation and high pH stresses

As shown in Fig. 4, RNA-seq data indicated that induction of PHO89 in response to both Pi scarcity and alkaline pH was dependent on the presence of Pho4(A) and Pho4(B). However, our recent work showed that the increased expression of PHO89 under high pH was also fully blocked in a rim101 mutant [41]. Thus, we sought to examine this behavior in more detail with the aid of a reporter strain in which the PHO89 promoter was fused to GFP. As shown in Fig. 7A, when cells were cultured in low Pi conditions, lack of Crz1 resulted in partial loss of GFP expression, and mutation of either RIM101 or PHO4(B) fully blocked GFP expression from the PHO89 promoter. When cells were exposed to high pH (8.0), lack of CRZ1 caused a modest but significant increase in expression after 90 min. As expected, the absence of either Rim101 or both Pho4(A) and Pho4(B) totally abolished GFP production (Fig. 7B). Therefore, these results further prove that both Rim101 and Pho4 pathways are essential for the regulation of PHO89 expression under Pi starvation and high pH stress conditions, while Crz1 has a less relevant contribution.

Fig. 7.

SEM

Alkaline response elements can be used to construct novel pH-regulatable promoters in K. phaffii

Functional mapping of the PHO89 promoter (to be discussed elsewhere) showed that removal of the region between nt -772 and -476 potentiated about 2-fold GFP production in response to alkaline pH (compare strain RW-04 in Fig. 7C with pPHO89-GFP in panel B). However, a further deletion up to nt 390 (strain RW-04) fully abolished the response (not shown). This implied that the region between nt 476 and 390 should contain activator element(s). To test if these elements could potentiate expression, we placed a duplicate of the -476/387 region upstream of the construct in strain RW-04, giving rise to strain RW-21 and tested their response at pH 8.0. As shown in Fig. 7C, strain RW-21 produced after alkalinization up to 1.6-fold higher GFP levels than RW-04, while maintaining an undetectable response at acidic pH (5.5). Therefore, this artificial promoter is about 4-fold stronger than the native version described originally (Fig. 7B and reference [19]). This result shows that specific pH-responsive elements can be useful components for building novel promoter versions, thus opening new opportunities for the rational design of alternative, pH-responsive promoters in K. phaffii.

The Pho4, Rim101 and Crz1 pathways functionally interact in K. phaffii

To explore further the possible interaction among these pathways we mutated PHO4(A) or PHO4(B) in crz1 and rim101 backgrounds (Additional Table 2). In addition, we created various triple mutants by mutating CRZ1 or RIM101 in the PJA-026 background (pho4(A) pho4(B)) and these strains were used to carry out specific phenotypic analyses. We first tested the tolerance to calcium ions, because previous work had shown that mutation of CRZ1 as well as that of both PHO4 TFs decreased tolerance to this divalent cation [40, 41]. As shown in Fig. 8A, the effect of the mutation of CRZ1 on CaCl₂ sensitivity was stronger than that of the double pho4(A) pho4(B) mutation. The combination of the mutations of CRZ1 and either PHO4(A) or PHO4(B) did not further increase sensitivity. In contrast, calcium sensitivity was exacerbated in two independent strains lacking all three TFs, indicating an additive effect of the mutations. These mutants carried in CRZ1 the same mutation as the single crz1 strain, i.e., a stop at codon 201 (denoted as 201* in Fig. 8A and B). Interestingly, a 6-nucleotides deletion in CRZ1 that eliminates Ile185 and Lys186 (denoted as ΔI¹⁸⁵K¹⁸⁶ in Fig. 8A and B) but does not interrupt the protein yielded cells whose tolerance was similar to that of the double pho4(A) pho4(B) mutant, indicating that deletion of these two residues does not affect the function of Crz1. We found that mutation of RIM101 or CRZ1 did not alter the growth defect on Pi limitation of the double pho4(A) pho4(B) strain. Similarly, the sensitivity to Mn²⁺ ions of the Pho4-deficient strain remained unaffected (Additional File 1, Figure S4).

Fig. 8.

Genetic interactions of the lack of both Pho4 K. phaffii transcription factors with the mutations of CRZ1 and RIM101. The indicated strains were grown as described in Materials and Methods in the presence of increasing concentrations of CaCl₂ or different pHs. The effects of the different CRZ1 (panels A and B) and RIM101 (panel C) mutations in the respective proteins are indicated. For example, 201* denotes that the mutation creates a stop in CRZ1 at codon 201, and ΔI¹⁸⁵K¹⁸⁶ indicates that the mutation eliminates two residues in CRZ1, but the polypeptide is not interrupted. Similarly, 40*, 46* and 47* indicate stops at the respective codons in RIM101 (see Additional Table 2 for details). Data are mean ± SEM from at least seven independent cultures

We also monitored the alkaline pH tolerance in strains combining the mutations of the PHO4 TFs with that of CRZ1 or RIM101. Previous work showed that the mutation of CRZ1 results in a mild alkaline pH sensitive phenotype detectable only above pH 7.8 [41], whereas the lack of both Pho4 TFs yields cells highly sensitive even to pH 6.8-7.0 [40]. These phenotypes are confirmed in Fig. 8B. Notably, the triple mutant strains containing the CRZ1 201* mutation were more tolerant to alkaline pH than the double pho4(A) pho4(B) mutant, suggesting that lack of CRZ1 has a beneficial effect on alkaline pH tolerance in the absence of the Pho4 TFs. The combination with the CRZ1 ΔI¹⁸⁵K¹⁸⁶ mutation, which was neutral concerning calcium tolerance, was also irrelevant in rescuing the high pH sensitivity caused by the absence of both Pho4(A) and Pho4(B) (Fig. 8B). Mutation of RIM101 causes a stronger decrease in tolerance to high pH than that of CRZ1 [41], and we found that the additional mutation of either PHO4(A) or PHO4(B) did not alter the tolerance level of the RIM101 mutant (Fig. 8C). We then tested strains carrying three different truncations at the N-terminus of RIM101-encoded protein (stops at codon 40, 46 and 47) in the pho4(A) pho4(B) background (the stop at codon 46 is the same mutation that is present in the single rim101 strain). Remarkably, all strains showed increased tolerance compared to the double pho4 mutant, mimicking the profile of the single rim101 strain up to pH 7.4. At higher pHs, the triple mutants became more sensitive than the rim101 strain, to eventually become indistinguishable from pho4(A) pho4(B) cells at pH 7.8 and above (Fig. 8C). Therefore, our results show that, in K. phaffii, the mutation of either CRZ1 or RIM101 shows clear genetic interactions with the double PHO4 mutation regarding alkaline pH tolerance. In the case of RIM101, the effects are dependent on the intensity of the pH stress.

Discussion

We recently showed that, in contrast to all previously examined fungal genomes, the yeast K. phaffii encodes two functional Pho4 transcription factors, named Pho4(A) and Pho4(B) [40]. Pho4(B) is crucial for the transcriptional response to Pi starvation, whereas Pho4(A) has virtually no role under this condition. However, their expression in a S. cerevisiae pho4 mutant proved that both proteins were functional TFs. In addition, while individual mutation of each TF gave rise to a few specific phenotypes, the simultaneous mutation of both TFs rendered cells highly sensitive to multiple stresses. This suggested that Pho4(A) and Pho4(B) might have specific functions and that, in combination, they were major components of the cellular stress response mechanisms.

Because exposure of yeast cells to alkaline pH mimics the response to phosphate deficiency, we considered it necessary to investigate the individual role of each K. phaffii Pho4 protein in the adaptation to an alkaline environment. Interestingly, the relevance of Pho4(A) and Pho4(B) in mediating the transcriptional response differs depending on the specific stress condition. Thus, Pho4(B) is essential for the response to Pi starvation, but Pho4(A) is not [40]. In contrast, we show here that Pho4(A) plays a more prominent role under high-pH conditions, while loss of Pho4(B) has very little effect. It is worth noting that four out of the six genes whose induction by alkalinization specifically decreased in pho4(A) cells (Fig. 3A) are phosphate-related (PHO89, VTC2, VTC1, and PHO84). A fifth gene, PHO86, although was not induced by high pH, was also repressed in the pho4(A) mutant (Fig. 3D and Additional Table 1). This suggests that specific signals elicited by alkalinization, but not by Pi starvation, might activate Pho4(A). Among the 22 genes showing a decrease in induction at high pH in the pho4(A) pho4(B) double mutant, eleven contained one or more putative Pho4 consensus sequence, suggesting direct control by Pho4. Notably, the three iron-related genes (FTR1, FET3 and FRE2-1) did not contain Pho4 or Aft1 binding sites (Aft1 is a central TF in adaptation to iron scarcity), suggesting alternative Pho4-dependent regulatory processes. It is worth noting, however, that promoter regulation is not the only mechanism controlling mRNA levels. In this regard, work in S. cerevisiae has shown that, in contrast to heat-shock, osmotic and oxidative stress, alkaline stress affects substantially mRNA stability [46].

Five genes repressed by alkalinization increased their expression in the pho4(A) strain (Fig. 3A). Among them was FDH1 (Fig. 3C), which encodes a formate dehydrogenase that belongs to the MUT pathway [47]. However, no other members of the pathway, such as AOX1, AOX2, FGH1 (PAS_chr3_0867) or FLD1 (PAS_chr3_1028) were affected by alkalinization or mutation of both Pho4 TFs (Fig. 3C and Additional Table 1), indicating that the effect on FDH1 is specific. The K. phaffii FDH1 promoter has attracted some attention as an alternative to the AOX1 promoter because it can be induced by methanol even in the presence of glycerol [48]. Recent work reported that both pAOX1 and pFDH1 are downregulated by lactate and upregulated upon overexpression of the genes encoding the TFs MXR1 and MIT1 [49]. However, little is known about the specific regulation of FDH1. Our data indicates that the FDH1 promoter contains regulatory elements different from those found in other genes involved in the MUT pathway and that the Pho4 transcription factors might be involved in these differential regulatory properties. This notion is reinforced by our recent observation that FDH1 (but not AOX1) is induced by Pi starvation in wild-type cells but not in the double pho4(A) pho4(B) mutant [40].

The Ehrlich pathway is important for assimilation of specific amino acids, mainly branched chain and Phe [50] (see Figure S3 in Additional File 1). We observe that alkalinization represses the expression of enzymes required for the first two steps of the pathway and that genes involved in all three steps are derepressed in the absence of both Pho4(A) and Pho4(B) (Figure S3). This is interesting because the Ehrlich pathway is not regulated by Pho4-mediated mechanisms [51] and, among the promoters of the K. phaffii genes shown in Figure S3, only GRE3 shows a putative Pho4-binding sequence. However, the positive effect of the pho4(A) pho4(B) mutation on the expression of these genes could be explained if it is considered that the last step of the Ehrlich pathway involves a redox reaction and that the redox status affects metabolism through the Ehrlich pathway [52]. On the other hand, a cross-regulation between low-NAD⁺ and low-Pi signaling has been described, showing activation of the PHO pathway in low-NAD⁺ mutants [53]. Because fusel metabolites are profusely used in the production of flavors, chemicals, and pharmaceuticals [51], the effect of K. phaffii Pho4 TFs in the expression of the genes of the pathway might find future applications.

Our recent work showed that K. phaffii cells lacking Pho4(A) and Pho4(B) are strongly sensitive to different kinds of stress, including alkaline pH [40]. The Ena Na⁺-ATPase is a major component of the mechanism for tolerance to saline and alkaline stress [54–57]. While ENA2 was induced by Pi starvation in a Pho4-dependent way [40], its induction by alkaline pH is not affected by the pho4(A) pho4(B) double mutation (Fig. 4). Therefore, the alkaline pH sensitive phenotype of pho4(A) pho4(B) cells cannot be attributed to a defect in the expression of the ATPase gene. In addition, genes encoding other important components of pH homeostasis, such as the Pma1 ATPase, or the Nha1 and Kha1 cation/H⁺ antiporters were not induced by alkalinization and did not change in the pho4(A) pho4(B) double mutant. In contrast, we show that the induction by high pH of several genes required for adaptation to Pi starvation does not occur in the pho4(A) pho4(B) strain (Figs. 3D and 4). Thus, because alkalinization of the medium mimics a situation of Pi starvation, the inability to cope with Pi shortage could impair growth at high pH. It is also known that defects in copper and iron metabolism decrease tolerance to alkalinization in S. cerevisiae [37] and Schizosaccharomyces pombe [58]. Furthermore, alkalinization triggers the induction of genes required for uptake and utilization of these transition metals in many fungi [1] and references therein]. Therefore, a further contribution to the growth defect of the K. phaffii Pho4-deficient strains at high pH could derive from the impaired expression of several genes (FTR1, FET3 and FRE2-1) related to iron homeostasis, and that of PAS_chr3_0517 (ZPS1). This latter gene encodes a protein highly similar to C. albicans PRA1, which has been found induced by alkalinization and linked to zinc binding in this organism [59]. K. phaffii contains a gene encoding a putative Pho2 TF (PAS_chr2_2_0043). Its protein sequence is distant from that of S. cerevisiae and other fungi (20–30% of identity), although it must be noted that this transcription factor is poorly conserved in sequence through evolution. The function of Pho2 in K. phaffii, as far as we know, is yet to be determined. However, the relatively large number of genes affected by lack of Pho4(A) and Pho4(B) upon Pi starvation [40] or alkaline pH (this work), as well as the wide array of phenotypes observed in the K. phaffii double pho4 mutant [40], are in sharp contrast with that observed in S. cerevisiae, a yeast requiring Pho2 for Pho4 function (see Introduction section). This suggests that the dependence of K. phaffii Pho4 TF on Pho2 could be relatively weak or even null.

Recent work [41] has highlighted the relevance of the Rim101 and the Crz1 signaling pathways in response to alkalinization. We show here that 36 genes are regulated by Pho4(A) and Pho4(B) in collaboration with Rim101 or Crz1. Among them, seven are less expressed upon alkalinization in the pho4(A) pho4(B) strain and more expressed (often derepressed) in the rim101 mutant. Curiously, they are related to phosphate and iron metabolisms. In contrast, PHO81 and PHO89 are repressed in both kinds of mutants, whereas the expression of VTC1, VTC2, VTC3 and PHO4(B) is attenuated only in the pho4(A) pho4(B) strain. This indicates that, as it occurs in S. cerevisiae [4], alkalinization triggers different signaling pathways that can be integrated into specific alkali-responsive promoters. The strong regulation of PHO89 expression by Rim101 and the Pho4 TFs was confirmed by our experiments using a pPHO89-GFP reporter (Fig. 7A and B) and suggest that cooperative interaction between both kinds of TFs is necessary for efficient expression from this promoter. In summary, our results demonstrate that, collectively, both Pho4 TFs play a relevant role in tolerance to alkalinization in K. phaffii, being involved in the regulation of nearly one hundred genes. The combination of our current data with the recently reported effect of the crz1 and rim101 mutations [41] can explain, at least in part, about 65% of the transcriptional response to alkaline pH. This implies that nearly 35% of alkali pH responsive genes are not significantly affected by these mutations and, therefore, other still unexplored pathways must exist.

The phenotypes observed upon combination of the pho4(AB), crz1 and rim101 mutations are consistent with a crosstalk between the corresponding signaling pathways. The additivity of the crz1 and pho4(AB) mutations regarding sensitivity to Ca²⁺ ions points out to the existence of independent mechanisms. The sensitivity of the crz1 mutant could be attributed, at least in part, to a defect in the induction of the gene encoding the vacuolar Ca²⁺-ATPase Pmc1 [41]. However, lack of both Pho4(A) and Pho4(B) does not decrease the expression of key genes involved in calcium transport and homeostasis ([40] and this work). A plausible explanation for the calcium sensitivity of this mutant would be a likely defect in the accumulation of polyphosphates, which are a major form for calcium storage in vacuoles [60]. In fact, genetic interactions between the calcineurin/Crz1 and phosphate metabolism in fungi have been reported in Aspergillus fumigatus [61] and in C. neoformans [62].

In contrast to that observed for calcium tolerance, strains carrying a loss of function mutation in CRZ1 improved tolerance to alkaline pH caused by loss of Pho4(A) and Pho4(B). This effect cannot be attributed to a reversion of the Pho4-deficient phenotype, since the same strains showed the expected phenotypes when grown under Pi deficiency or in the presence of MnCl₂ (Figure S4). Instead, it could be explained, at least in part, by the increase in expression of PHO89 observed in the crz1 mutant in response to alkalinization (Fig. 7B), as Pho89 has been reported to be crucial for alkaline pH tolerance in S. cerevisiae [63]. We also show (Fig. 8C) that introducing loss-of-function RIM101 mutations in a Pho4-deficient background eliminate the high pH sensitivity. Notably, the beneficial effect of the loss of RIM101 is progressively lost at pH 7.4 and above. This genetic interaction could be explained by the opposite effect of the mutations on the expression of key genes involved on phosphate and iron homeostasis (see Results). Given the relevance of phosphate- and iron-related transcriptional responses for adaptation to alkaline pH, it is conceivable that the combination of both mutations could alleviate high pH sensitivity up to some extent. At higher pHs this beneficial effect would be insufficient, and the hypersensitive phenotype conferred by the double pho4(A) pho4(B) mutation would prevail.

We demonstrated recently that alkali pH-regulatable promoters could be useful for heterologous expression of industrially relevant proteins in K. phaffii [20], and showed for S. cerevisiae that multimerization of alkali responsive genetic elements allows to generate improved expression vectors regulated by alkaline pH [21]. We identify here a short alkali-responsive region within the PHO89 promoter whose duplication markedly increases GFP expression, thus providing the first example of this strategy in K. phaffii. The combination of bioinformatic analysis and functional mapping of alkali-regulatable promoters will become a valuable tool to identify specific regulatory elements that could be harnessed to generate hybrid promoters with enhanced capabilities.

Materials and methods

Growth conditions and recombinant DNA techniques

The DH5α Escherichia coli strain was grown at 37 °C in LB medium, supplemented with hygromycin B (100 µg/ml) when used as plasmid host. E. coli transformation and standard recombinant DNA techniques were performed as described [64]. Unless otherwise stated, K. phaffii cells (X-33) were grown at 28 °C on YP medium (Yeast extract 1%, Peptone 2%) supplemented with 2% glycerol (YPGly) or 2% glucose (YPD) as carbon source. For phosphate limitation experiments, cells were grown on synthetic medium prepared with Formedium premade mix (ref. CYN0803, Yeast Nitrogen Base without Amino Acids and Phosphate) supplemented with 15 mM KCl, 2% glycerol and, when needed, 7.35 mM KH₂PO₄, and adjusted to pH 5.8. K. phaffii cells were transformed by electroporation with a Gene Pulser electroporator (Bio-Rad) as described [65]. Unless otherwise stated, 2% of bacteriological agar (Condalab) was employed for growth on plates.

Generation of K. phaffii strains

Methods for constructing the strains PJA-001 (crz1) and PJA-007 (rim101) were reported in [41], whereas pho4(A), pho4(B) and pho4(A) pho4(B) mutants were described in [40]. Strains carrying crz1 pho4(A) and crz1 pho4(B) mutations were constructed by CRISPR/Cas9-mediated gene disruption [66] by transforming strain PJA-001 with vectors p907_sgRNA-Pho4(A).1 or p907_sgRNA-Pho4(B).1, respectively. Strains rim101 pho4(A) and rim101 pho4(B) were made by transforming strain PJA-007 with the same vectors. To construct the triple mutants, strain PJA-026 (pho4(A) pho4(B)) was transformed with vectors p907_sgRNA-Crz1(1) or p907_sgRNA-Rim101(26), respectively [41].

The indicated K. phaffii strains were transformed by electroporation with the different constructs and plated on YPD agar containing hygromycin (200 µg/ml). Colonies were streaked out twice on hygromycin YPD agar plates, and genomic DNA from at least three isolated colonies was amplified and the amplification fragments were sequenced to identify possible mutations. Positive clones were cured for the vector by repeated growth in liquid YPD. The specific genotype of these strains and the effect of the mutations on the produced protein is shown in Additional Table 2.

The effect of the different mutations on the response of the PHO89 promoter was evaluated with the aid of the pPHO89-GFP translational reporter described in [19]. The vector was linearized with PmeI (which cuts within the AOX1 promoter) and the appropriate strains (Additional Table 2) were electroporated with one-hundred ng of DNA and plated on YPD agar plates containing hygromycin (200 µg/ml). The correct insertion of the construct into the chromosomal AOX1 locus was confirmed by PCR. The wild-type strain carrying the pPHO89-GFP insertion (PJA-037) was described previously [19].

Construction of the modified PHO89-GFP reporters

A 502 bp fragment encompassing nt -476/+11 of the PHO89 promoter was amplified from genomic DNA with oligonucleotides PHO89_frag2_HindIII and Pp_Pho89_rev using Q5 DNA polymerase and cloned into the HindIII-KpnI sites of plasmid pAHYB-GFP [67], to yield pRW02. In parallel, a 107 bp fragment covering nt -476/-387 of this promoter was amplified with oligonucleotides PHO89_frag2_HindIII and multRui_rv_HindIII. This fragment, which contains a predicted CDRE (Calcineurin Dependent Response Element) and three putative Pho4 binding sites, was cloned into HindIII-digested pRW02. The orientation of the insert and the absence of unwanted mutations were assessed by DNA sequencing. The vector with the inserted fragment in the same orientation than that the native promoter was named pRW02_A. The sequence of the oligonucleotides used can be found in Additional Table 3. Vectors were linearized with PmeI and one-hundred ng of DNA used to electroporate wild type cells as described above.

RNA preparation and transcriptomic analysis

For evaluation of the transcriptional response of K. phaffii to alkalinization of the medium cells were grown overnight on YPGly medium. Cultures were then diluted in 100 ml of fresh medium at OD₆₀₀=0.2 (∼ 10⁷ cells/ml) and grown until an OD₆₀₀ around 0.6. At this point, two 50 ml aliquots were centrifuged (5 min at 1000xg). One aliquot was resuspended in the same volume of fresh YPGly medium plus 50 mM TAPS pH 5.5 (control cells) or in YPGly supplemented with 50 mM TAPS adjusted to pH 8.0 (stressed cells). Samples (15 ml) were taken immediately after resuspension in medium at pH 5.5 (T = 0) and, for each stressed culture, after 15, 30 and 60 min, placed on ice and immediately centrifuged (5 min at 1000 xg) at 4 °C. Then, one ml of cold H₂O was added, and cells were transferred to pre-cooled 1.5 ml screw cap tubes and centrifuged (2 min at 1000 xg) at 4 °C. Finally, the supernatant was removed, and samples were snap-frozen with liquid N₂ and stored at -80 °C. For assessment of the effects of phosphate starvation, cells were grown as described in [40].

Total RNA extraction was carried out as described in [19] from three independent cultures of wild-type, pho4(A) (strain PJA-004), pho4(B) (strain PJA-021) and pho4(A) pho4(B) (PJA-026) cells. RNA characterization and library preparation was as in [40]. Pooled libraries were subjected to sequencing in a NovaSeq X Plus equipment (Illumina) by Novogene Co., Ltd. Clean reads were generated by fastp [68]. An average ± SD of 2.4E7 ± 3.7E6 clean reads were obtained, yielding 3.56 ± 0.55 clean Gbases (97.1% ± 0.10 of them reaching Q30 or above). Paired-end clean reads were aligned with the K. phaffii GS115 reference genome (assembly GCA_000027005.1) by using the HISAT2 (v2.2.1) software [69], resulting in an average of 96.11% ± 0.92 of mapped reads. Differential gene expression (DGE) analysis was performed with DESeq2 [70] by setting a |log2(Fold Change)| ≥ 1 and a p-value ≤ 0.01 as thresholds. DNA sequence data was deposited at NCBI’s Gene Expression Omnibus (GEO) and can be retrieved under study GSE310523.

Gene Ontology enrichment analysis was carried out with g: Profiler [45] using the K. phaffii GS115 data and default settings (except domain scope, that was set to “All known genes”). The STRING database v12.0 [71] was used to generate interaction networks between proteins. Minimum Interaction Score was set to 0.5 (medium-high confidence). K-means clustering was employed, and the number of clusters ranged from 3 to 5 depending on the protein set.

Phenotypic assays

K. phaffii strains carrying the different mutations were assessed for diverse phenotypes in both liquid and solid media. For evaluation in liquid medium, three hundred µl of the appropriate medium were inoculated with exponential cultures (OD₆₀₀ = 0.004) in Bioscreen honeycomb microplate plates (#9502550) and incubated at 28 °C for 18–24 h in an orbital shaker (Ika-Schuttler MTS-2) at 900 strokes/min. To test growth under phosphate limitation, the Formedium premade mix (ref. CYN0803) was used as described in [40]. Plates were prepared with 1.5% of purified agar (Condalab, Ref. 1806) to ensure a Pi concentration < 15 µM, cultures were diluted to OD₆₀₀ = 0.05, and 5-fold dilutions were prepared. Three µl of the suspensions were spotted on the plates and grown for the indicated time prior documentation.

Expression from the PHO89 promoter with a GFP reporter

To evaluate the response to phosphate starvation and to alkalinization, the appropriate strains carrying the pPHO89-GFP described in [19] or its variants were treated as in [40] and processed for flow cytometry as in [19]. The fluorescence of the samples was analyzed in a CytoFLEX (Beckman Coulter) flow cytometer using the FIT-C filter (525/40).

Other techniques

Search for putative TFs binding sites (including Pho4) was made with the RSAT platform [72] by mining with the matrix-scan algorithm the upstream -800/-1 sequences (relative to the initiating ATG and allowing overlapping with adjacent CDS) of selected genes from the K. phaffii GS115 genome (GCF_000027005.1_ASM2700v1). The upstream (non-overlapping ORFs) sequences of the K. phaffii GS115 genome were used as background and a p-value threshold of 2 × 10^-4 was applied. The relevant matrices were extracted from the JASPAR non redundant fungi database [73]. Sequence logos were created with WebLogo 3 [74].

Authors’ contributions

MA: Investigation, Formal analysis, Data curation, Writing - review & editing. AC: Investigation, Funding acquisition, Writing - review & editing. RW: Investigation. JA: Investigation, Conceptualization, Funding acquisition, Supervision, Validation, Formal analysis, Visualization, Writing - original draft, Writing - review & editing. AG: Investigation, Conceptualization, Validation, Visualization, Writing - review & editing.

Funding

Work supported by grants PID2020-113319RB-I00 and PID2023-150535OB-I00 (AEI, Ministerio de Ciencia, Innovación y Universidades) to JA and AC, and 2023 PROD 00006 (AGAUR, Generalitat de Catalunya) to JA. RW is a recipient of a China Scholarship Council fellowship.

Data availability

The datasets supporting the conclusions of this article are included within the article and its supplementary materials (Additional File 1 and Additional Tables 1, 2 and 3). Raw and processed transcriptome data has been deposited in the NCBI Gene Expression Omnibus (GEO) database under accession number GSE310523.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.