Ambient pH sensing in filamentous fungi: Pitfalls in elucidating regulatory hierarchical signaling networks
Department of Biochemistry and Immunology, Ribeirão Preto Medical School, University of São Paulo, São Paulo, Brazil
Address correspondence to: Antonio Rossi, Department of Biochemistry and Immunology, Ribeirão Preto Medical School, University of São Paulo, 14049-900 Ribeirão Preto, São Paulo, Brazil. Tel.: +55 16 3602 3112. Fax: +55 16 3602 0222. E-mail: firstname.lastname@example.org.
The diverse molecular mechanisms used by filamentous fungi to sense and transduce environmental signals are mediated by a broad range of biochemical networks, in which interwoven metabolic pathways support cellular sensing and the transduction of a myriad of ambient signals . One such signal that is sensed and transduced by fungal cells is ambient pH [2-4]. All microbial organisms must adapt physiologically to ambient pH changes by, for example, appropriately adjusting the secretion of nutritional enzymes with optimal activity accordingly, that is, phosphate (Pi)-repressible acid and alkaline phosphatases have their secretion increased at pH 5.0 and 8.0, respectively [2, 3]. Coordination of environmental and nutrient sensing with the control of genetic responses to molecular signals is crucial for the growth and proliferation of most living organisms, including fungi, because the failure to regulate cellular responses to changing environmental pH would have a negative impact on the competitiveness of an organism. Thus, fungi have evolved elaborate signal transduction networks to allow adaptation to ambient pH changes [2, 3, 5-7]. These metabolic pathways regulate the secretion of hydrolytic enzymes, membrane remodeling, cell adhesion, cellular transport, growth, development, iron uptake, stress, molecular defense, and virulence responses at both acidic and alkaline pH [8-12]. Gene regulation by pH involves the highly conserved PacC/Pal signal transduction pathway that mediates a myriad of metabolic events in filamentous fungi and is widely accepted that it governs the response to neutral to alkaline pH. The pacC gene codes for a Zn-finger transcription factor, and the six pal genes (pal A, B, C, F, H, and I) are members of a signaling cascade that senses alkalinity and promotes the proteolytic activation of PacC [4, 13-18]. The alkaline ambient pH signal is sensed by the transmembrane proteins PalI and PalH  and transmitted from the plasma membrane complex to the endosomal membrane complex on the surface of the endosome, possibly with the involvement of proteins PalC and PalF . The PacC protein undergoes a two-step proteolytic processing in response to neutral to alkaline pH . The PalA protein interacts with the YPXL/I motifs in PacC72 (full-length version of PacC) mediating a protein–protein interaction, which is required for the proteolytic conversion of PacC72 to PacC53. This proteolytic step is possibly mediated by PalB, a calpain-like cysteine protease. In the second, pH-independent step, PacC53 is converted to PacC27. The PacC27 form prevents transcription in neutral to alkaline conditions of those genes expressed preferentially at acidic pH. Thus, whatever the ambient pH, loss-of-function mutations in the pal genes shall lead to a wild-type acidic growth phenotype, that is, the full-length version of PacC would be inactive in the absence of pal signaling or under acidic growth pH . However, it is uncertain whether the highly conserved PacC pathway governs only the expression of alkaline-responsive genes, and even if it is the sole mechanism that governs pH-responsive gene expression in fungi.
The pacC/pal Genes and Mutations
Several conserved genes in which mutations alter the response of fungi to ambient pH were first identified in the fungus Aspergillus nidulans by Dorn in the early 1960s . A. nidulans has optimal growth pH on solid medium at a pH near to 6.5. At least six genes have been identified for which a mutation interferes with pH regulation of the synthesis of Pi-repressible extracellular enzymes and permeases. Mutations in the pal A, B, C, E, and F genes increase colony staining for acid phosphatase at pH 4.8. By contrast, mutations in the pacC gene increase colony staining for alkaline phosphatase at pH 8.2. Dorn suggested that the pacC gene encoded a Pi-repressible acid phosphatase and that the pal genes encoded Pi-repressible alkaline phosphatases. These predictions assume that increased colony staining corresponds to an increased expression of both enzymes. Indeed, the early work of Dorn revealed that the pacC5 mutant of A. nidulans synthesizes temperature-sensitive acid phosphatase, mutant palB7, a temperature-sensitive alkaline phosphatase, and mutant palcA1, temperature-sensitive acid and alkaline phosphatases . In an attempt to interpret these and other data generated by classical genetics studies, Dorn proposed a hypothesis whereby acid and alkaline phosphatases may be composed of two different monomers, with one monomer common to two of these enzymes (PalcA), thus implying a structural function for the pacC, palB, and palcA genes. However, the classic work of Caddick et al.  in the 1980s showed that mutations in any of the pal or pacC genes (e.g., palA1, palB7, palF15, pacC5, and pacC14) increased the levels of Pi-repressible acid (PacA) and alkaline (PalD) phosphatase, respectively, at pH 6.5. This finding implies that at a pH near 6.5, the pal and pacC mutants show a response similar to the wild-type strain at a more acidic and alkaline pH, respectively. The pacC5 and the pacC14 mutations are alleles and code for truncated proteins in which the acidic C-terminal segment is absent; however, the zinc-finger region of PacC is conserved . A null mutant (pacC−) deleted for the entire pacC coding region mimics acidic growth conditions more extremely and, in addition, results in cryosensitive growth, poor growth, and conidiation. This strain is very sick probably because both the positive and negative actions of PacC are prevented. Classical and molecular genetics analyses have supported that the pacC gene is a wide-domain regulatory gene whose products directly mediate pH regulation, whereas the products of the palA, B, C, E, and F genes are involved in a metabolic pathway that leads to the synthesis of an effector molecule that interacts with the pacC product [3, 13]. This cognate effector molecule would be able to prevent the positive and negative action of the pacC gene; in other words, the product of the pacC gene acts positively on the transcription of palD and negatively on the transcription of pacA at alkaline pH . Thus, in the absence of a functional pacC product, the presence or absence of the cognate effector molecule would be of no consequence. In the absence of the cognate effectors, the derepression of pacA transcription at alkaline pH would occur. Interestingly, although the promoters of alkaline-expressed genes such as ipnA, pacC, and the alkaline protease-encoding prtA contain a number of PacC-binding sites (consensus 5′-GCCARG-3′), the pH-regulated acid phosphatase-encoding pacA gene contains none in the 1,311-bp upstream of the initiation codon [4, 22]. Regardless of the underlying reason, it is clear that the structural and regulatory hypotheses are incompatible . As the regulatory nature of the pacC gene is now well documented, is it possible to reconcile these two hypotheses?
Secretion of Pi-Repressible Phosphatases and Other Hydrolytic Enzymes
Most microorganisms modulate their metabolic activities through a complex network that involves several genes that sense the availability of Pi. Under Pi shortage, this metabolic pathway signals for the derepression of phosphatases and transporters to satisfy the cellular demand for Pi. High Pi concentrations repress the function of PalcA, a transcriptional regulator homologous to the Neurospora crassa NUC-1 protein. PalcA is a highly conserved DNA-binding protein that is involved in the sensing of Pi deprivation. The promoter region of genes pacA and palD contains NUC-1-binding sites (consensus 5′-CACGTG-3′), and under conditions of Pi shortage, the NUC-1 repressor is removed. Consequently, the Pi-repressible phosphatases are transcribed [24-26].
The secretion of both acid and alkaline Pi-repressible phosphatases (among other proteins and enzymes) responds to ambient pH under the regulation of the Pal/PacC signal transduction pathway. Thus, if the pal genes represent metabolic steps involved in the activation of a single regulatory protein (transcription factor PacC), the molecular properties of the secreted Pi-repressible acid phosphatase, for example, should be identical in all pal mutants and wild-type strain. However, this is not the case, as indicated by the molecular mass, electrophoretic mobility, chromatographic behavior, thermolability, kinetic properties, and glycosylation of the acid phosphatases that are secreted by the palA1, palB7, palC4, and palF15 mutant strains [27-29]. We hypothesized that each of the pal genes perform specific functions in the activation of PacC including, in addition to proteolysis, other post-translational modifications, such as glycosylation. It is worth noting that a large number of enzymatic steps are necessary to accomplish the full glycosylation needed to secrete Pi-repressible phosphatases in the native form. Incomplete post-translational modifications may change the affinity of enzymes involved in tagging PacC, which probably occurred with the truncated PacC14 protein, that is, PacC14 affinity for the PacC-binding sites has probably been changed (Fig. 1). It is likely that the alterations in steady-state kinetic properties, temperature sensitivity, and electrophoretic mobility reflect structural changes in the enzyme molecule synthesized by the pal and pacC mutant strains. For example, the acid phosphatase secreted by the palB+ strain at pH 6.5 showed hydrolytic activity over p-nitrophenyl phosphate (PNP-P) with negative cooperativity, whereas the enzyme secreted by the palB7 mutant showed Michaelian behavior . This difference could explain the approximately twofold increase in the enzyme activity secreted by the palB7 strain, as shown by colony staining or enzymatic hydrolysis of PNP-P. This occurs because enzymes showing Michaelian behavior hydrolyze substrate faster than isoenzymes showing negative cooperativity, assuming that substrate concentrations are equivalent . The pacCc14 mutant should poorly secrete the Pi-repressible acid phosphatase, as irrespective of the ambient pH, the secretion of alkaline phosphatase would be predominant; however, this is also not the case. Indeed, the level of pacA-encoded acid phosphatase secreted by the pacCc14 mutant at pH 5.0 was comparable with that secreted by the pacC+ strain when the temperature of the fungal culture was decreased from 37 °C to 22 °C . Moreover, the pacC14 mutation drastically reduced the mannose and N-acetylglycosamine content of the pacA-encoded acid phosphatase that was secreted at pH 5.0. In addition, the pacA-encoded acid phosphatase secreted by the pacC14 mutant strain was not observed by staining for this enzyme in cultures grown at the restrictive temperature of 37 °C. Thus, it seems that one of the PacC-dependent metabolic responses to pH signaling in both N. crassa and A. nidulans is not transcriptional regulation of phosphatases, but rather the pH-dependent multistep glycosylation of secreted Pi-repressible phosphatases.
Is Proteolysis of the Transcription Regulator PacC Imperative for Its Function?
The molecular characterization of the pal/pacC genes of A. nidulans helped to improve the understanding of ambient pH signaling in fungi. The pal genes (palA, B, C, F, H, and I) are putative members of a signaling cascade involved in both acidic and alkaline ambient pH sensing. Their function is thought to be promotion of the proteolytic activation of PacC, as expression of C-terminal-truncated PacC derivatives suppresses pal mutant effects, as shown in the pacCc5, pacCc14, and other supposedly gain-of-function mutations. An interesting finding is that unlike pacC, transcription of the six pal genes is apparently not pH-regulated [19, 33]. The pal genes would be necessary for the proteolytic activation of PacC, that is, the full-length version of PacC (PacC72) would be converted to PacC27, the active form of PacC. However, this active metabolic phenotype was not confirmed in the pacCc14 strain carrying a truncated pacC gene that transcribes a truncated PacC52 protein (Figs. 1 and 2). A full-length version of the pacC-1 gene cloned from N. crassa complemented the pacCc14 mutation of A. nidulans. The acid and alkaline phosphatases secreted at pH 5.0 by the pacCc14 mutant strain were both remediated in the pacC+ transformant, confirming a loss-of-function (rather than a gain-of-function) effect of the pacCc14 mutation in A. nidulans . This finding implies that for the proteolytic activation of PacC it requires, in addition to the interaction of PacC72 with two PalA molecules , pal-dependent synthesis of another effector molecule that tags the pacC product. Interestingly, sequence prediction analyses suggest that both PacC53 and PacC52 truncated proteins do not have an adequate conformation to show in silico DNA-binding properties (Fig. 1). Admitting that both structures bind DNA, they would have at least a lowered affinity for DNA binding, which would be enough to show differences in the transcription profiling of different genes. Thus, irrespective of ambient pH, if full-length PacC is inactive in the absence of pal signaling, loss-of-function mutations in any pal gene should lead to a defective metabolic phenotype. Indeed, loss-of-function mutations in the palA, palB, or pacC genes lowered the mannose content in the phosphate (Pi)-repressible acid phosphatase secreted by A. nidulans in an acidic milieu. In addition, by using an mRNA differential display, we isolated one cDNA, not detected in the palB7 strain, that encodes a manosyltransferase [8, 28, 29]. In Trichophyton rubrum, disruption of the pacC gene resulted in decreased growth on human nails and decreased secretion of keratinolytic proteases possibly through transcriptional modulation of the O- and N-linked mannosyltransferase genes [6, 35]. Transcription profiling of these two mannosyltransferases revealed a high level of complexity because their transcription were also affected by nutrients, culture pH, and the functioning of the pacC gene . We hypothesized that putative pal-dependent PacC post-translational modifications are determinants for the conversion of PacC to its functional form, irrespective of ambient pH, in which the active PacC protein should carry the post-translational modification promoted by the pal-signaling cascade. Putative candidate post-translational modifications predicted by sequence analysis of PacC72 include phosphorylation, glycosylation, sumoylation, palmitoylation, and/or acetylation of PacC72. Interestingly, phosphorylation is important for the nuclear localization of RIM101 transcription factor (an ortholog of PacC) in Saccharomyces cerevisiae. When the cyclin-dependent kinase Pho85 (an ortholog of PGOV in N. crassa) is functional, RIM101 localizes in the cytoplasm, a physiological condition in which Pi is abundant. When functional Pho85 is absent, RIM101 accumulates in the nucleus [25, 36]. It has been reported that the full-length form of RIM101 could localize in the nucleus as well and bind DNA . Also interesting are the observations that a mammalian PalB homolog shows preferential nuclear localization when expressed in COS cells [28, 38, 39], suggesting that PalB has specific functions in the processing of transcription factors, as does its homolog Rim13p in S cerevisae [40, 41]. PacC is also phosphorylated in A. nidulans; however, in this case, it was speculated that phosphorylation leads to recruitment of the proteasome machinery . Moreover, even in the presence of the pal signal, a fraction of the full-length form of PacC is localized in the nucleus at both acidic and alkaline pH .
Transcription of the pacC gene is also modulated in response to nutrient changes, phosphate and carbon availability, and alternative splicing of the palB gene, which showed variant mRNA splicing patterns in response to changes in growth conditions [43-45]. The palB pre-RNA was spliced in strains cultured in minimal medium, regardless of extracellular pH and Pi changes, whereas it was apparently not spliced in yeast extract (YE) cultures, except when grown in low Pi at pH 5.0. Interestingly, the intronic regions I and III of palB pre-RNA were spliced concomitantly, resulting in a predicted inactive PalB protein when introns I and/or III were not spliced. Thus, we presumed that PalB is inactive in YE cultures at pH 8.0, an experimental condition in which PacC is active, suggesting that PacC protein is potentially activated through an alternative metabolic pathway to PalB proteolysis (Fig. 2). Thus, it remains possible that a full-length version of PacC is functional [8, 43]. Accumulation of higher levels of pacC transcripts in low-Pi YE cultures at pH 8.0 may be correlated with the absence of spliced forms of the palB+ allele. Furthermore, transcription of pacC is downregulated in palB+ and palB− strains grown in high- or low-Pi cultures, respectively. Downregulation of pacC in a palB− background probably occurs because the protease PalB is inactive. Moreover, transcription of Pi-repressible phosphatases is modulated by Pi and pH and, consequently, depends on both PalcA and PacC transcription factors. This finding suggests the occurrence of crosstalk between the pH- and Pi-regulatory circuits, implying synergy between the regulators PalcA and PacC. Thus, the transcription regulator PacC may be functional in both acidic and alkaline ambient pH, in the full-length or proteolyzed form, if it carries in either case the pal-dependent molecular tag. If it is demonstrated that the full-length version is the only active molecular form of PacC, the implication will be that the proteolysis event regulates PacC turnover.
This work was supported by research grants from the Brazilian funding agencies FAPESP (Grant No. 2008/58634-7 and Postdoctoral position No. 2012/03689-7), CNPq, CAPES, and FAEPA. The authors thank C. A. Vieira and S. H. Castrechini for technical assistance.