The alkaline ambient pH signal transduction pathway component PalC has no assigned molecular role. Therefore we attempted a gene-specific mutational analysis and obtained 55 new palC loss-of-function alleles including 24 single residue substitutions. Refined similarity searches reveal conserved PalC regions including one with convincing similarity to the BRO1 domain, denoted PCBROH, where clustering of mutational changes, including PCBROH key residue substitutions, supports its structural and/or functional importance. Since the BRO1 domain occurs in the multivesicular body (MVB) pathway protein Bro1/Vps31 and also the pH signal transduction protein PalA (Rim20), both of which interact with MVB component (ESCRT-III protein) Vps32/Snf7, this might reflect a further link between the pH response and endocytosis.
REGULATION by ambient pH has been extensively studied in Aspergillus nidulans where it is mediated by the PacC/Pal regulatory circuit and major contributions have also been made by studies on the equivalent Rim systems in the yeasts Saccharomyces cerevisiae, Candida albicans, and Yarrowia lipolytica.
Response to ambient pH in A. nidulans (reviewed by Peñalva and Arst 2002, 2004; Arst and Peñalva 2003) is mediated by PacC (Caddick et al. 1986; Tilburn et al. 1995), which is activated by a two-step proteolysis of the full-length form, PacC72 (Orejas et al. 1995; Mingot et al. 1999; Díez et al. 2002), in response to the alkaline ambient pH signal transduced by the six-membered Pal signaling pathway (Arst et al. 1994). The functional PacC 250-residue form, PacC27, is an activator of alkaline-expressed genes (Espeso and Peñalva 1996) and repressor of acid-expressed genes (Espeso and Arst 2000).
Signal transduction components PalH and PalI (S. cerevisiae homologs Rim21p and Rim9p, respectively) are predicted seven- and four-pass membrane proteins (Li and Mitchell 1997; Denison et al. 1998; Negrete-Urtasun et al. 1999) and strong candidates as ambient pH sensors. PalB (S. cerevisiae Rim13p), a calpain-like cysteine protease (Denison et al. 1995; Lamb et al. 2001; Sorimachi and Suzuki 2001), is probably responsible for the first, pH-sensitive, signaling proteolysis. PalA (S. cerevisiae Rim20p) (Negrete-Urtasun et al. 1997; Xu and Mitchell 2001) contains the ∼160-residue BRO1 domain (PFAM domain PF03097; http://www.sanger.ac.uk/Software/Pfam/index.shtml), first identified in yeast Bro1p (Nickas and Yaffe 1996). PalA apparently enables the signaling proteolysis by interacting both with PacC, through two YPXL/I motifs flanking the signaling proteolysis site, and with Vsp32/Snf7, the endosomal sorting complex required for transport-III (ESCRT-III) protein, as demonstrated by Vincent et al. (2003). This agrees with the model described for S. cerevisiae (Xu and Mitchell 2001; Xu et al. 2004) where Rim20p (PalA) interacts with both Rim101p (PacC) (Xu and Mitchell 2001) and Vsp32p/Snf7p, which also interacts with Rim13p (PalB) (Ito et al. 2001), to form a scaffold-promoting interaction between the Rim101p cleavage site and the Rim13p protease. A functional link between pH signal transduction and multivesicular body (MVB) pathway sorting complexes has been firmly established in S. cerevisiae (Xu et al. 2004) and shown to be conserved in C. albicans (Kullas et al. 2004; Xu et al. 2004).
Possible molecular roles remain elusive for PalF (S. cerevisiae Rim8p) and PalC, which has no S. cerevisiae homolog (Li and Mitchell 1997; Maccheroni et al. 1997; Negrete-Urtasun et al. 1999). As the PalC primary amino acid sequence revealed no evident sequence signature and PalC appeared absent from the hemiascomycete lineage, we carried out mutational analysis of this protein along with sequence profile similarity searching, exploiting the recent publication of a number of fungal genomes.
The GABA (γ-aminobutyrate) technique is a powerful tool for the selection of pacC (Mingot et al. 1999; Fernández-Martínez et al. 2003) and pal (Arst et al. 1994; Denison et al. 1998; Negrete-Urtasun et al. 1999) loss-of-function mutations. It relies on the ability of these acidity-mimicking mutations to suppress areAr (= areA−, nitrogen metabolite repressed) mutations for utilization of γ-aminobutyrate (GABA) as the nitrogen source, through derepression of acid-expressed gabA specifying the GABA permease (Caddick et al. 1986; Hutchings et al. 1999; Espeso and Arst 2000). In haploid strains, the GABA technique yields mutations in any of the seven pH regulatory genes. To target mutations to palC, we employed an areAr palC+/areAr palC− diploid. This diploid cannot use GABA as the nitrogen source but suppression can be achieved by acidity mimicry, resulting from mutation of the palC+ allele or through mitotic recombination yielding palC− homozygosity. To avoid the latter, we constructed diploid R using inoB2 (inositol auxotrophy) distal to palC40 (Negrete-Urtasun et al. 1999) and in repulsion to areAr18 (Arst et al. 1989), a reciprocal translocation of chromosomes III and IV, including palC and inoB (Figure 1). Consequently, palC40 homozygosity without inositol auxotrophy can occur only through alignment of the translocation-containing chromosome III and the untranslocated chromosome IV with recombination both between the areAr18 breakpoint and palC40 (4 cM) and between palC40 and inoB2 (22 cM) (Figure 1).
The first experiment Table 1 yielded 48 acidity-mimicking—as determined by impaired growth on pH 8 medium (Cove 1976)—palC mutants. These included 24 new truncations, 8 single and one double missense, two with three-base deletions, and one rearrangement mutation (data not shown) plus 12 palC40 mitotic recombinants, despite precautions. In a second attempt and to avoid null phenotype palC truncations, we screened for leaky or temperature-sensitive growth on pH 8.0 medium and obtained another 20 palC missense mutants, thus totaling 55 new palC alleles (Table 1). Alignment (Figure 2) illustrates that most of the missense mutations affect amino acids conserved in the majority of the ascomycete PalCs shown. Of the single residue change mutations, only palC80 (Gly321Asp) and palC162 (ΔArg442) are complete loss-of-function mutations. All truncating mutations result in complete loss of function except the leaky palC131 (1–454 + 2), which removes 53 C-terminal residues including a completely conserved C-terminal di-aromatic motif. This motif (di-tyrosine in A. nidulans PalC) resembles the di-phenylalanine motif, a C-terminal transport motif facilitating ER export (Nufer et al. 2002). As palC131 is partially functional, this motif and other residues C-terminal to residue 454 cannot be completely essential for PalC structure and function. The complete loss-of-function truncating mutations palC159 (1–427 + 29) and palC153 and palC179, truncating the protein cleanly after residue 426, indicate that at least some of residues 427–454 are essential, which agrees with the clustering of mutations in this window. Mutations substituting conserved residues in regions “BRO1 similar,” “LALA,” and “ERRE” (Figure 4B) further support the structural and/or functional importance of these regions. We caution, however, that the phenotypes of any of the palC mutations characterized here might be due to reduced PalC protein levels resulting from protein misfolding with consequent instability or even from messenger instability rather than from PalC dysfunction.
PalC orthologs are present in members of three major fungal phyla, ascomycota, basidiomycota, and zygomycota (Figures 2 and 3). Among ascomycete PalCs the Y. lipolytica ortholog Figure 2 is the first to be detectable in hemiascomycetes (yeasts). This dispels the notion, based on the failure to detect PalC homologs in the previously available hemiascomycete genomes (supplementary Table S1 at http://www.genetics.org/supplemental/) of Ashbya gosypii, C. albicans, Candida glabrata, Debaryomyces hansenii, Kluyveromyces lactis, S. cerevisiae, and Schizosaccharomyces pombe, that PalCs are exclusive to euascomycetes (filamentous fungi). As strong comparative genomic evidence indicates that Y. lipolytica separated early from the main hemiascomycete line and has not been subjected to the significant constraints in genome size characterizing other yeasts (Dujon et al. 2004), we suggest that palC was present in a universal fungal ancestor and that palC orthologs might have been lost from most hemiascomycetes.
Y. lipolytica palC is the only homolog of an A. nidulans pH regulatory gene not identified by mutations affecting pH regulation of extracellular proteases (Tréton et al. 2000; Gonzalez-Lopez et al. 2002). Thus its functional involvement in pH regulation remains to be established.
Refined similarity searching reveals the PalC-related protein family:
As sequence similarity searches gave no indication of a possible molecular role for PalC, we also used the relatively sensitive procedures of HMMer (Eddy 1998) and PSI-BLAST (Altschul et al. 1997) to seek PalC-related protein families. Searching the nrdbembl (January 2005) protein database with the 262 N-terminal residues of PalC detected, as well as PalC orthologs, we detected two members of the PFAM BRO1 domain family after the third round of iteration with E-values markedly below the cutoff (E = 0.005) and all 63 PFAM PF03097 BRO1-like domain-containing proteins after the eighth iteration and retrieved no new unrelated sequences subsequently. This relationship to BRO1 domain proteins was highly suggestive in view of the PalA BRO1 domain and was substantiated using a Hidden Markov model derived from the region of similarity corresponding to A. nidulans PalC 38–235 in the 14 PalCs available to query the nrdbembl, which detected three BRO1 domain proteins with reasonable scores (0.0018–0.046). In a third approach, a PalC Hidden Markov model in an HHpred (Söding 2005) search of the PFAM database gave only one significant hit (E-value 3e-05), the BRO1-like domain. These data provide statistical evidence that a region of PalC comprising residues 28–235 is significantly, albeit remotely, related to the PFAM PF03097 BRO1-like domain and a region of similarity extending downstream from it (Figure 4A). We have denoted this region PCBROH (PalC-BRO1 homology). The frequency of mutational changes here and their occurrence in key PCBROH residues such as Tyr68 and Trp111 indicate a major functional and/or structural role in PalC (Figure 4B).
BRO1 domain proteins include, as well as PalA (Negrete-Urtasun et al. 1997), PalA orthologs Rim20p (S. cerevisiae and C. albicans) (Xu and Mitchell 2001), the human PalA homolog AIP1/Alix (Missotten et al. 1999; Vito et al. 1999), and the yeast Bro1p (Nickas and Yaffe 1996), an MVB pathway protein recruiting the deubiquitinase Doa4p to endosomes (Odorrizi et al. 2003; Luhtala and Odorrizi 2004). All these proteins interact with Vps32p/Snf7p (a key component of ESCRT-III) or its homologs (Xu and Mitchell 2001; Katoh et al. 2003; Strack et al. 2003; Vincent et al. 2003; von Schwedler et al. 2003; Katoh et al. 2004; Peck et al. 2004; Xu et al. 2004).
It has been suggested that the BRO1 domain is the region of interaction between Rim20p-Bro1p family members and Snf7p family members (Xu et al. 2004). However, the finding that the BRO1 domain-containing protein human rhophilin-2 did not interact with either of two human hSnf7 proteins tested (Peck et al. 2004) argues against this generalization. Thus a precise role for the BRO1-like domain in PalC cannot be assigned. However, it is tempting to speculate that it reflects a further link between the pH response system and the MVB pathway multiprotein complexes at the endosome and/or cell membrane.
The PacC-mediated pH regulatory system is an important virulence determinant in plant (reviewed by Peñalva and Arst 2004) and animal pathogens, including C. albicans (reviewed by Fonzi 2002; Peñalva and Arst 2002; Davis 2003) and A. nidulans (Bignell et al. 2005). In view of the growing frequency of invasive mold, and in particular of Aspergillus infections (reviewed by Clark and Hajjeh 2002), the uniquely fungal PalC might possibly be a suitable target for therapeutic intervention.
We are grateful to Lily Stanton for technical assistance; Elaine Bignell, Eduardo Espeso, and Olivier Vincent for comments on the manuscript; and Joanna Rudnicka for interesting discussion. We thank the Wellcome Trust, the Dirección General de Investigación Científica y Técnica, and the Consejo Superior de Investigaciones Científicas (CSIC) for their support through grants 067878, BIO2003-0077, and for a CSIC Bioinformatics I3P studentship for J.C.S.-F., respectively.
Note added in proof: The structure of the Bro1 domain in yeast Bro1p has been determined (J. Kim, S. Sitaraman, A. Hierro, B. M. Beach, G. Odorrizi and J. H. Hurley, 2005, Structural basis for endosomal targeting by the Bro1 domain. Dev. Cell 8: 937–947) and establishes that the Bro1 domain contains 367 residues, thereby extending considerably beyond the 160-residue Bro1p PFAM domain. Sequence alignment suggests that the region of A. nidulans PalC corresponding to the structurally based Bro1 domain of Bro1p includes the PalC N-terminal 446 residues, within which the majority of our single residue substitutions fall. Conserved PalC residues Arg47, Tyr68, Trp111, and Glu136 (Figures 2 and 4A) are likely to correspond, respectively, to Bro1p residues Arg51, Tyr70, Trp94, and Glu116, which are central in one of two structure-stabilizing, buried, charged, and polar clusters. The loss-of-function phenotypes of palC Tyr68His or -Asn and Trp111Arg mutations are consistent with the predicted structural importance of these residues.
Communicating editor: A. Mitchell
- Received April 21, 2005.
- Accepted May 6, 2005.
- Copyright © 2005 by the Genetics Society of America