Aquaporins in yeasts and filamentous fungi




Recently, genome sequences from different fungi have become available. This information reveals that yeasts and filamentous fungi possess up to five aquaporins. Functional analyses have mainly been performed in budding yeast, Saccharomyces cerevisiae, which has two orthodox aquaporins and two aquaglyceroporins. Whereas Aqy1 is a spore-specific water channel, Aqy2 is only expressed in proliferating cells and controlled by osmotic signals. Fungal aquaglyceroporins often have long, poorly conserved terminal extensions and differ in the otherwise highly conserved NPA motifs, being NPX and NXA respectively. Three subgroups can be distinguished. Fps1-like proteins seem to be restricted to yeasts. Fps1, the osmogated glycerol export channel in S. cerevisiae, plays a central role in osmoregulation and determination of intracellular glycerol levels. Sequences important for gating have been identified within its termini. Another type of aquaglyceroporin, resembling S. cerevisiae Yfl054, has a long N-terminal extension and its physiological role is currently unknown. The third group of aquaglyceroporins, only found in filamentous fungi, have extensions of variable size. Taken together, yeasts and filamentous fungi are a fruitful resource to study the function, evolution, role and regulation of aquaporins, and the possibility to compare orthologous sequences from a large number of different organisms facilitates functional and structural studies.

Abbreviations used:

mitogen-activated protein kinase


transmembrane domain


During the last few years, a number of fungal genomes have been sequenced. This has provided opportunities to study important principles of genome evolution and to employ sequence conservation of orthologous proteins as a suitable tool for functional analysis. Comparative genomics can provide information about structurally and functionally important residues and domains, serving as a basis for designing mutational studies. In the present study, we have investigated the presence and conservation of aquaporins in fungal genomes. Orthodox aquaporins mediate rapid and selective flux of water across biological membranes and hence play important roles in the osmoregulation of cells and organisms. Aquaglyceroporins on the other hand, facilitate transmembrane transport of small uncharged molecules like polyols, urea, arsenite and many more, thereby playing roles in nutrient uptake, osmoregulation and probably other processes (Borgnia et al., 1999; Hohmann et al., 2001). The aquaglyceroporins are divided into Fps1-like (defined by a conserved regulatory region in the N-terminus), Yfl054-like (having a very long N-terminal extension including a conserved stretch) and a third group of proteins, not falling into any of these categories. Detailed studies on the precise physiological roles and functional properties have been reported only for a small number of the fungal water and glycerol channels. The present study tries to motivate future studies on fungal aquaporins as they bear potential to reveal novel information on the function and physiological roles of members of this ancient protein family.

Yeasts and filamentous fungi

Fungi form one of the five kingdoms of life and are a large and diverse group of eukaryotic organisms. Only a very small fraction of all fungal species have been described to date. Fungi, ranging from single-celled organisms (yeasts) to the edible mushrooms, have enormous ecological importance and they encompass pathogens for humans, animals and plants. Fungi are useful, not only for eating, but also in various biotechnological processes, including production of antibiotics, pharmaceutical proteins and, last but not the least, beer, wine, cheese and bread. Because of these reasons, fungi have attracted the interest of scientists. Several fungi are excellent experimental organisms. Recently, fungal genome sequencing has become popular because of the rather small genomes (10–40 Mbp). Some 20 genomes have so far been (partially) sequenced and many more are in the pipeline, generating a valuable resource of sequence information for comparative genomics. Efforts will, however, have to be invested into proper annotation of the genome information.

The yeast species whose genomes have so far been sequenced belong to the hemiascomycete family (Table 1). They include four Saccharomyces species: Saccharomyces cerevisiae, S. paradoxus, S. mikatae and S. bayanus (Kellis et al., 2004), which are closely related to each other. Lower coverage genome information is also available for S. kluyveri, S. kudriavzevii and S. castelli (Cliften et al., 2003). The genome of S. cerevisiae, the most important cellular model organism, was already completely sequenced 10 years ago (Goffeau et al., 1996) and is by far the best annotated. Candida albicans (Jones et al., 2004) is the most common human pathogenic yeast, whereas C. glabrata is the second most important causative agent (Dujon et al., 2004). Although these two species share a common family name, C. glabrata is clearly more closely related to S. cerevisiae. Kluyveromyces lactis (Dujon et al., 2004) is phylogenetically placed in the middle of the hemiascomycetes and used for genetic studies and in industrial applications (the name stands for ‘milk yeast’). The K. waltii genome has recently been sequenced for studies on genome duplications (Kellis et al., 2004). Ashbya gossypii is a filamentous yeast used as a model to study polar growth. It has industrial importance for the production of vitamin B2. The A. gossypii genome is the smallest of all free-living eukaryotes yet sequenced (9.2 Mb, encoding 4718 proteins; Dietrich et al., 2004). Debaromyces hansenii is a halotolerant yeast found on fish and salted dairy products (Dujon et al., 2004). Yarrowia lipolytica is an alkane-using yeast (Dujon et al., 2004). As is also apparent from aquaporin sequences, this yeast shares properties with filamentous fungi. It is used in genetic studies and for heterologous protein production. Finally, the fission yeast Schizosaccharomyces pombe (Wood et al., 2002) is quite distinct from all other yeasts (as distinct from S. cerevisiae as either of the two from human!) and is like budding yeast S. cerevisiae, a widely used model system in molecular cell biology.

Table 1.  Fungal aquaporinsFunctional studies have been performed on the proteins from S. cerevisiae (cited in the text), S. pombe (Kayingo et al., 2004), Z. rouxii (Wang et al., 2002; Neves et al., 2004) and C. albicans (Carbrey et al., 2001b). Fps1 orthologues from K. lactis and K. marxianus complement the S. cerevisiae fps1Δ mutant (Neves et al., 2004). The others are predicted from genome sequences.
SpeciesOrthodox aquaporins*Fps1-like aquaglyceroporins*Yfl054-like aquaglyceroporins*Other aquaglyceroporins*CommentNames
  • *

    Name of the aquaporin/aquaglyceroporin is given along with the number of amino acids in parentheses.

S. cerevisiaeAqy1 (305)Fps1 (669)Yfl054 (646)  S.cer_Aqy1,2
 Aqy2 (289)    S.cer_Fps1
S. bayanusAqy1 (278)Fps1 (661)Yfl054 (654)  S.bay_Aqy1
S. paradoxusAqy1 (305)Fps1 (673)Yfl054 (648)  S.par_Aqy1
S. mikataeAqy1 (305)Fps1 (218)Yfl054 (648) Fps1 incompleteS.mik_Aqy1
S. kluyveriAqy1 (266)Fps1 (427)  Fps1 incompleteS.klu_Aqy1
S. kudriavzeviiAqy1 (306)Fps1 (259)  Fps1 incompleteS.kud_Aqy1
S. casteliiAqy1 (245)    S.cas_Aqy1
S. pombe  SPAC977.17 (598)  S.pom_SPAC977.17
K. lactisKLLA0B10010g (193)KLLA0E00550g (563)  K.lac_Aqp1 incompleteK.lac_KLLA0B10010g
K. marxianus Q6QHK5 (571)  No genome sequenceK.mar_Fps1
Z. rouxii Q6RW11 (692)  No genome sequenceZ.rou_Fps1
K. waltii K.wal_20572 (600)K.wal_15269 (666)  K.wal_20572
C. glabrataCAGL0A01221g (293)CAGL0C03267g (652)   C.Gla_CAGL0A01221g
 CAGL0D00154g (290)CAGL0E03894g (602)   C.Gla_CAGL0D00154g
C. albicansCaAqy1 (273)    C.alb_Aqy1
D. hanseniiDEHA0F28787g (310)    D.han_DEHA0F28787g
A. gossypiiAGL266C (450)Acl068w (476)   A.gos_AGL266C
Y. lipolyticaYALI0F01210g (284)  YALI0F00462g (392) Y.lip_YALI0F01210g
    YALI0E05665g (385) Y.lip_YALI0F00462g
N. crassaNCU08052.1 (239)    N.cra_NCU08052.1
A. nidulansAN7168.2 (959) AN2822.2 (463)AN3915.2 (386)AN7168.2 inA.nid_AN7168.2
   AN7618.2 (612)AN0830.2 (286)annotation fused to amino acid transporterA.nid_AN2822.2
F. gramineumFG10816.1 (547) FG03780.1 (548)FG03248.1 (343) F.gra_FG10816.1
 FG00811.1 (318)    F.gra_FG00811.1
 FG03680.1 (286)    F.gra_FG03680.1
M. grisea MG03904.4 (536) MG05880.4 (390) M.gri_MG03904.4
  MG04162.4 (270)   M.gri_MG04162.4
  MG10783.4 (278)   M.gri_MG10783.4
U. maydis UM00223.1 (420) UM01508.1 (363) U.may_UM00223.1
  UM02842.1 (503) UM02169.1 (396) U.may_UM02842.1
    UM01930.1 (570) U.may_UM01508.1

There are five filamentous fungi included in the present study, the first four all being ascomycetes (like yeasts). Neurospora crassa was first described in 1843, as the causative agent of a mould infection in French bakeries. It is a multicellular fungus comprising 28 morphologically distinct cell types (Borkovich et al., 2004). It has a long history as an experimental organism: Beadle and Tatum in the 1940s discovered that genes affect enzymes using N. crassa genetics (Horowitz, 1991). Aspergillus nidulans is a widely used experimental organism (the genome sequence has not been published). It is a multicellular fungus in which nuclei migrate from a central spore out into newly formed filaments after division. Fusarium gramineum (the genome sequence has not been published), also known as Gibberella zeae, is a mycotoxin-producing filamentous fungus causing scab of wheat and barley, with big impact on U.S. agriculture in the past decade. Magnaporte grisea is a haploid filamentous ascomycete with a relatively small genome (genome sequence has not been published; 40 Mb). It is an excellent model organism for studying fungal phytopathogenicity and host—parasite interactions. It is the causal agent of rice blast disease and such a big threat to the world food supply that it is considered as a biological weapon. Ustilago maydis (genome sequence has not been published) is a basidiomycete (like common mushrooms) inducing tumours on host plants, e.g. maize. The haploid stage can be propagated in culture as yeast-like cells. It is an important model system for studies on phytopathogenicity and host—parasite interactions as well as for the discovery of antifungal drugs and has been studied by plant pathologists for more than 100 years.

Aquaporin sequences in fungal genomes

Table 1 summarizes the aquaporin protein sequences we found to be encoded in the fully sequenced genomes of yeasts and filamentous fungi. The aquaglyceroporins were divided into three subgroups: proteins similar to S. cerevisiae Fps1, proteins similar to S. cerevisiae Yfl054 and a third group of proteins that do not seem to resemble any of the first two.

Interestingly, the number of aquaporins seems to vary significantly even between closely related species. There are several yeasts and one of the filamentous fungi (N. crassa) that only possess a single protein of the family. Those organisms that appear to lack orthodox aquaporins (note that the genome sequences of K. marxianus and Zygosaccharomyces rouxii are incomplete) usually possess an aquaglyceroporin. Perhaps this is a bifunctional water/solute channel, as was found in the malaria parasite Plasmodium falciparum (Hansen et al., 2002). Even proteins that appear to be orthodox aquaporins based on sequence similarity might have a wider substrate spectrum, as shown for a Toxoplasma protein (Pavlovic-Djuranovic et al., 2003). Overall, the number of aquaporin family members can range from one to five, with a tendency for higher numbers in filamentous fungi. Such proteins may therefore play a role in differentiation processes or be expressed in different cell types. In organisms with five aquaporins, the number for each subtype differs: A. nidulans has four aquaglyceroporins and F. gramineum has three orthodox aquaporins. It certainly would be most interesting to study the localization, expression and function of these proteins.

Orthodox aquaporins in Saccharomyces cerevisiae

Fungal orthodox aquaporins have so far been studied only in S. cerevisiae (Bonhivers et al., 1998; Laizé et al., 1999, 2000a, b; Carbrey et al., 2001a; Meyrial et al., 2001) and in C. albicans (Carbrey et al., 2001b). Remarkably, most of the S. cerevisiae laboratory strains have spontaneous inactivating mutations in both their orthodox aquaporin genes. In Aqy1, there are three point mutations in residues that are conserved among human and plant aquaporins, and one frameshift mutation leading to an 18 residues shorter C-terminus (Bonhivers et al., 1998). Interestingly, truncation of the C-terminus of Aqy1 from the reference strain Σ1278b (functional aquaporins) results in increased water permeability in Xenopus oocytes (Laizé et al., 1999). The AQY2 gene in most laboratory strains is interrupted by an 11 bp deletion, leading to a premature stop codon and hence a non-functional protein (Carbrey et al., 2001a, b). In addition, when comparing S.cer_Aqy2 with Aqy2 of S. chevaleri (probably a strain of S. cerevisiae), which is a functional water channel when expressed in Xenopus oocytes, it differs only at position 141 [proline versus serine residue; Carbrey et al., 2001a). Substitution (P141S) improved localization of S.cer_Aqy2 to the plasma membrane. A serine or threonine residue occurs in 11 out of 22 fungal water channels in this position, whereas all the remaining 11 possess hydrophobic amino acids (alanine, glycine and valine).

Investigation by a PCR approach of orthodox aquaporin alleles in 52 laboratory, wild and industrial yeast strains showed that a functional Aqy1 allele was present in all wild and industrial strains but only in a small number of laboratory strains (Laizé et al., 2000b). On the other hand, the Aqy2 allele was non-functional in most strains, indicating that there seems to be a selective pressure to maintain Aqy1 but not Aqy2 (Laizé et al., 2000b). This appears to fit with the observation that most yeast species related to S. cerevisiae have only one orthodox aquaporin. On the basis of available information, it is known that the aquaporin gene is in a syntenic position (and hence probably functionally equivalent) to AQY1. Aqy1 and Aqy2 are highly similar (88% identical) indicating a recent gene duplication. On the other hand, as discussed below, AQY1 and AQY2 expression is regulated differently, indicating functional specialization. More genome sequences, especially of high sequence quality, will be needed to study aquaporin evolution in yeasts thoroughly.

Although the fungal water channels possess all the known sequence characteristics of orthodox aquaporins, they show a minor divergence in the NPA motifs, which are part of the pore constriction formed by loops B and E. The water channel in Y. lipolytica has an NPC in the B loop and F. gramineum and N. crassa have NPV in the E loop (see discussion on NPA motifs below). It appears that several fungal water channels have N- and C-terminal extensions of unknown function and poor sequence conservation. In the case of the N. crassa aquaporin, the C-terminal extension is almost certainly due to a sequence or annotation mistake, as this extension by itself has similarity to a complete amino acid transporter with 12 TMDs (transmembrane domains), probably derived from a different gene.

The fungal orthodox aquaporins are an amazingly diverse group of proteins. Overall, only 20 residues are completely conserved. Still, BLAST searches with these fungal aquaporins detect the other fungal water channels with highest similarity, even if the sequence identity between yeast and fungal aquaporins decreases to approx. 35%. This suggests that fungal orthodox aquaporins have evolved from a common ancestor since the eukarya split into three kingdoms. Orthodox aquaporins from yeasts appear to be more similar to plant plasma membrane intrinsic proteins compared with animal water channels, whereas the situation is just the opposite for those of filamentous fungi.

S. cerevisiae Aqy1

The AQY1 gene becomes abundantly expressed when diploid S. cerevisiae are shifted to sporulation conditions (nutrient starvation; Chu et al., 1998). The gene is only poorly expressed in haploid and diploid vegetative cells. Sporulation of diploid yeast cells is tightly coupled with meiosis and results in the formation of four haploid spores, surrounded by an ascus wall. Spores differ from vegetative cells in their differently composed cell wall, their decreased water content and diminished metabolic activity. This makes them much more resistant to harsh environmental conditions. Overall, the yeast sporulation process resembles in many respects gametogenesis in animals (Engebrecht, 2003).

When a heterozygous AQY1/AQY1-GFP strain undergoes sporulation, the green fluorescent protein signal is detected only in two of the four spores (Sidoux-Walter et al., 2004). Therefore Aqy1 is produced during the later stages of sporulation, i.e. once spores have formed and separated from each other. Mutants lacking Aqy1 show a 30% decreased spore viability and results suggest that this is due to events occurring during spore formation rather than during spore maintenance or germination (Sidoux-Walter et al., 2005). One possible explanation for these observations is that Aqy1 plays a role in decreasing the spore water content, which is approx. half of that of vegetative cells. This is partly due to spores accumulating large amounts of trehalose. Although this potentially generates a driving force for water into the developing spore, this force is counteracted by the rigid spore wall and hence turgor pressure. Therefore the combination of production of material inside the cell combined with a rigid wall and an active water channel could be a means to decrease cell/spore water content (Sidoux-Walter et al., 2004).

S. cerevisiae Aqy2

The expression of AQY2 also appears to be tightly regulated. The gene is expressed in exponentially growing cells, but not at all in resting cells (F. Sidoux-Walter and S. Hohmann, unpublished data). It has been reported that AQY2 expression is stimulated by the protein kinase A isoform Tpk2, which could potentially explain growth phase regulation (Robertson et al., 2000). In addition, expression of AQY2 is diminished when cells are shifted to high osmolarity and it reappears when cells are again shifted to lower osmolarity. Down-regulation by osmoshock at least partly depends on the osmosensing high osmolarity glycerol MAPK (mitogen-activated protein kinase) pathways (F. Sidoux-Walter and S. Hohmann, unpublished data; Hohmann, 2002). The Aqy2 protein is located mainly in the plasma membrane, although localization in the endoplasmic reticulum occurs in laboratory strains that carry the Pro141 allele (see above; Carbrey et al., 2001a).

A phenotype for the deletion of AQY2, which fits with the observed expression pattern has not been reported. Agre and co-workers observed that deletion of either AQY1 or AQY2 renders yeast cells more tolerant to repeated osmotic shifts (Bonhivers et al., 1998; Carbrey et al., 2001a). Such conditions are, according to our data, associated with poor expression of both water channels. Thevelein and co-workers reported diminished survival of rapid freezing regimes for deletion of either orthodox aquaporins, whereas overexpression enhanced survival. It was not reported whether these two aquaporins are normally expressed under the conditions employed in these experiments (Tanghe et al., 2002, 2004).

Aquaglyceroporins in yeasts and filamentous fungi

Aquaglyceroporins efficiently facilitate specific passive permeation of small uncharged solutes across biological membranes such as glycerol, other polyols, urea, arsenite, antimonite and probably more, commonly in preference to water (Heller et al., 1980; Maurel et al., 1994; Gerbeau et al., 1999; Fu et al., 2000; Wysocki et al., 2001; Liu et al., 2002). GlpF, the glycerol facilitator in Escherichia coli, is best studied and a crystal structure of 2.2 Å resolution has been reported (Fu et al., 2000). Yeasts and filamentous fungi possess a range of genes that encode aquaglyceroporins. Again, those of S. cerevisiae are best studied; Fps1 was, in fact, one of the first five aquaporins discovered approx. 15 years ago (Van Aelst et al., 1991). As discussed below, the role of Fps1 in yeast osmoregulation is very well characterized. Genome sequencing revealed the existence of a second yeast aquaglyceroporin with the systematic name Yfl054 (Hohmann et al., 2000).

The fungal aquaglyceroporins seem to belong to three groups (Figure 1, Table 1): the Fps1-like proteins, the Yfl054-like proteins and those that do not resemble either of these groups. Whereas Fps1-like proteins are found only in yeasts and the third group only in filamentous fungi, Yfl054-like proteins are found in both.

Figure 1.

Cladogram of fungal aquaporins

Aquaglyceroporins can be divided into three subgroups; Fps1-like proteins (defined by a conserved regulatory region in the N-terminus), Yfl054-like proteins (defined by a long N-terminal extension including a conserved stretch) and a third group containing proteins not matching any of these criteria. As shown in the cladogram proteins divided according to these criteria indeed cluster together. The cladogram was made in CLUSTAL W using default parameters and then reconstructed in ‘Tree view’. Abbreviations are explained in Table 1.

Fps1-like proteins

Fps1 plays a central role in the osmoregulation of S. cerevisiae (Hohmann, 2002). The protein controls the intracellular level of glycerol, the compatible osmolyte of proliferating yeast cells. Under hyperosmotic conditions, yeast cells stimulate the activity of the MAPK Hog1 through an elaborate sensing-signalling system (Hohmann, 2002). Activation of Hog1 in turn stimulates, among many others, genes encoding enzymes in glycerol biosynthesis (Albertyn et al., 1994; Rep et al., 2000). Together with the activation of glycolysis (Dihazi et al., 2004), this results in enhanced glycerol production. In order for glycerol to be accumulated in the cell, the activity of Fps1 diminishes under hyperosmotic conditions within seconds (Tamás et al., 1999). Once the cell has accumulated sufficient glycerol or is shifted to hypo-osmotic conditions, Fps1 opens again to release glycerol and hence turgor pressure (Tamás et al., 1999). This picture is consistent with phenotypes associated with FPS1 mutations. Deletion of FPS1 renders yeast cells sensitive to hypo-osmotic shock because excessive turgor pressure cannot be released and hyperactive, unregulated Fps1 causes sensitivity to hyperosmotic stress because the lost turgor pressure cannot be restored.

There is no evidence that Fps1 requires additional proteins for its regulation, although this needs to be demonstrated with reconstituted protein in vitro. So far, however, expression and purification of Fps1 has not been successful. In any case, it appears that Fps1 is an aquaglyceroporin whose activity is regulated by osmotic changes. This regulation requires an approx. 20 amino acid long regulatory domain right in front of the first TMD as well as a short stretch located right downstream of the sixth TMD (Figures 2 and 3) (Tamás et al., 1999; Hedfalk et al., 2004; Karlgren et al., 2004).

Figure 2.

Conserved domains in Fps1-like proteins

Location of the domains with reference to the topology of Fps1 (Figure 3) are as follows: (A) residues 205–280 (S. cerevisiae sequence) correspond to the first TMD 1 and the sequences proximal to it; (B) residues 501–548 correspond to the sixth TMD and sequences distal to it. Alignments were carried out by CLUSTAL W, using default parameters. Abbreviations are explained in Table 1.

Figure 3.

Topology map of Fps1 highlighting important residues

Conserved residues in the MIP family (•) and the C-terminal myc-tag (◯) used for detection on Western blot are highlighted. Enlarged and bold residues correspond to the regulatory domains important for proper channel closure. Squared residues (□) were identified in a genetic screen for hyperactive mutants. Model revised by courtesy of Kristina Hedfalk, Chalmers University of Technology, Göteborg, Sweden.

The N-terminal regulatory domain was initially identified by truncation analysis. Remarkably, large portions of the 250 residues long N-terminus could be removed without affecting regulation (Tamás et al., 1999, 2003). However, deletion of a short element between amino acids 225 and 236, just 20 residues in front of the first TMD, rendered Fps1 constitutively open. This effect can be scored easily as yeast cells expressing hyperactive Fps1 fail to grow on high osmolarity medium. Mutational analysis narrowed the important region to a 12 amino acid long domain (LYQNPQTPTVLP), which is extremely well conserved among yeast species and is a part of the approx. 25 amino acid stretch constituting the most conserved part within the N-termini among Fps1-like sequences (Figures 2 and 3). The distance between this domain and the first TMD, which is also conserved, is important for proper channel regulation (Tamás et al., 2003).

The C-terminal regulatory domain was identified by employing a similar truncation analysis, where parts of the approx. 150 residue long C-terminus were eliminated (Hedfalk et al., 2004). This identified 12 amino acids, residues 535–546, as important for controlling the Fps1 function. Expression of Fps1 lacking this domain also resulted in delayed intracellular glycerol accumulation and sensitivity to hyperosmotic conditions. The first half of this 12 amino acid long domain (HESPVN) is very well conserved among yeast species (Figure 2). In fact, the regulatory elements in the N- and C-terminal extensions are more highly conserved compared with loops B and E.

A genetic screen for hyperactive Fps1 has contributed further information about residues involved in the regulation (Karlgren et al., 2004). The screen was based on the use of a gpd1Δ gpd2Δ mutant that cannot produce glycerol and therefore does not grow in the presence of, for example, 1 M xylitol. However, in cells expressing a constitutively open Fps1, xylitol can be taken up into the cell resulting in equal concentrations on both sides of the plasma membrane, relieving osmotic stress and allowing growth. This was used to screen for random mutations rendering Fps1 constitutively open. So far, mutations in 14 distinct residues have been identified (Figure 3; Karlgren et al., 2004). Five of those are located within the N-terminal regulatory domain, whereas three cause truncation of the C-terminus, thereby confirming previous studies on the importance of the termini. Two conserved residues in the B loop also appear to be critical for channel control. Interestingly, all mutations identified hit residues that are completely conserved among the ten aligned yeast Fps1 proteins. Random mutational analysis and generation of double point mutations will provide a framework for further genetic and structural analysis to understand better the mechanisms that control Fps1 regulation.

The present view on how Fps1 might be regulated is also based on a structural prediction of the conserved N-terminal regulatory domain. The conserved NPQ motif in the centre of this domain reminds of the NPA motif of loops B and E discussed above. Indeed, it seems probable that the regulatory domain can form a loop structurally similar to the B loop (Karlgren et al., 2004). This has stimulated the idea that the regulatory domain could fold back into the membrane and block the B loop. The observation that residues in the B loop itself also contribute to regulation could be accommodated in such a model. The C-terminal regulatory domain may have a more general role on properly positioning the TMDs within the membrane and facilitate an interaction of the N-terminal regulatory domain and the B loop. Eventually, structural information and studies in a reconstituted system will be needed to fully explain the mechanism that controls Fps1.

By searching whole or partial genome sequences, we identified nine proteins that resemble Fps1. In addition, the Fps1 gene from Z. rouxii, an osmotolerant yeast, has been sequenced. It appears that Fps1-like proteins are restricted to yeasts since they have not been found so far in filamentous fungi. This is remarkable, given the important role played by Fps1 in yeast osmoregulation and the apparently sophisticated mechanism of its regulation. As a criteria for Fps1-like proteins, we used the highly conserved, unique regulatory sequence located in front of the first TMD (no other hits in the databases), the long, variable hydrophilic extensions of different lengths and an unusually long (approx. 35 amino acids) but poorly conserved A loop. The predicted proteins differ in length by as much as 216 amino acids from the shortest, A. gossypii Fps1 (476 residues), to the longest, Z. rouxii Fps1 (692 residues). C. glabrata is unique in having two Fps1 homologues. Interestingly, both C. glabrata proteins seem to be somewhat more similar to S.cer_Fps1 than to each other, which is quite unusual for paralogous genes. One possible explanation could be that both C. glabrata genes originate from independent horizontal gene transfer events from the S. cerevisiae gene.

The six TMDs show high similarity in agreement with conserved residues in the aquaporin family (Park and Saier, 1996), as well as the AEF motif in TMD1 and its counterpart D in TMD6 (Zardoya and Villalba, 2001). The TMDs are flanked by conserved and often charged or polar amino acids. Although the overall sequence conservation of the termini is poor, there are some not previously recognized conserved sequence elements. In the N-terminus, there is an acidic region around position 75 (S. cerevisiae) followed by a conserved motif around position 110 (FPIQEVIPS). In the C-terminus, there is a conserved element (FKSV) around 75 amino acids distal from the sixth TMD and clustering of acidic and basic amino acids seems to be conserved as well. The sequence of these elements does not reveal any functional role. Since deletion of these sections does not affect channel regulation (Tamás et al., 2003; Hedfalk et al., 2004), the conserved elements might indicate additional, possible regulatory roles of Fps1.

Alterations in NPA motifs

Among aquaporins, the NPA motifs in loops B and E respectively are highly conserved. These loops fold back into the membrane and form part of the central pore constriction (Fu et al., 2000; Nollert et al., 2001; de Groot et al., 2003). Together, the NH donor groups from the asparagine side chains form hydrogen bonds with the hydroxy groups of glycerol (Fu et al., 2000).

Remarkably, in fungal aquaglyceroporins, there are several alterations in the NPA motifs (Figure 4A). In the B loop, only NP is perfectly conserved (one exception, which shows an HPA), and in the E loop, only the asparagine and alanine residues are conserved (one exception with an NPS). In the B loop, the third position of the motif can be taken by alanine, serine, threonine or isoleucine, the central position of the motif in the E loop can be taken by proline, leucine, methionine, phenylalanine, glycine or alanine residue. Since the asparagine residues crucial for glycerol transport are perfectly conserved in fungal aquaglyceroporins, a similar glycerol pathway as for GlpF can be anticipated. It has been shown that Fps1 (S. cerevisiae) tolerates ‘restored’ NPA in both loops (Bill et al., 2001), which is in line with the fact that only the asparagine residues are in contact with glycerol. At the same time, GlpF does not tolerate the NPS and NLA sequences found in Fps1, indicating that altered NPA motifs are compensated by other changes (Bill et al., 2001). Several other residues important for glycerol transport, i.e. Arg206 in the constriction region of GlpF (Sui et al., 2001) and residues known to interact with glycerol (Fu et al., 2000) are well conserved in fungal aquaglyceroporins.

Figure 4.

(A) Sequences of loops B and E of fungal aquaglyceroporins, showing alterations in the NPA motifs

Alignments were carried out by CLUSTAL W, using default parameters. Abbreviations are explained in Table 1. (B) Structural details of the region near the Asn-Pro-Ala motifs in the crystal structure of E. coli GlpF (PDB code 1fx8; Fu et al., 2000). The view is approximately along the quasi-2-fold symmetry axis, along the centre of the membrane plane, looking towards the core of the protein. The periplasmic side is up. The thin black line separates the well-conserved Pro/Ala pair (right of the line; Pro69/Ala205) from the less-conserved pair (left; Pro204/Ala70). Selected residues are shown as stick models, with those of the NPA motifs having green carbons. Two hydrogen bonds from Glu14 to the B loop main-chain amides of residues His66 and Leu67 (grey carbons) and one hydrogen bond to Gln93 are indicated with grey lines. Also shown is the selectivity filter residue Arg206 (grey carbons). Glycerol molecules in the channel are shown as space-filling models.

The strictly conserved proline and alanine (Pro/Ala) residues are Pro69 (Pro353) and Ala205 (Ala482) in the B and E loop NPA motifs respectively (numbered according to GlpF with Fps1 numbers in parentheses). The less conserved Pro/Ala pair is composed of Ala70 (Ser354) and Pro204 (Leu481) residues. In contrast with the asparagine residues of the NPA motifs, the proline and alanine residues do not contribute to the channel surface, but are located roughly 7 Å away from the pore.

A possible explanation for the difference in sequence conservation lies in the asymmetry of the local environment near the NPA motifs. The two Pro/Ala pairs are located close in space near the quasi-2-fold axis, illustrated in Figure 4(B). Within van der Waals contact distance from the side chains of the well-conserved Pro/Ala pair are Val94 (Leu378) and the strictly conserved Gln93 (Gln377) in TMD3, as well as the remainder of the NPA residues. Gln93 (Gln377) is buried in the protein core, and hydrogen bonds to Glu14 (Glu259), which is also well buried and involved in an extensive hydrogen bond network. Hydrogen bonds are also present between the side-chain carboxy group of Glu14 (Glu259) and the backbone amides of His66 (His350) and Leu67 (Leu351), whose carbonyl groups form hydrogen bonds with glycerol molecules that traverse the channel. A mutation of Pro69 (Pro353) or Ala205 (Ala482) may therefore result in structural changes that propagate through the hydrogen bond network and alter the function of the channel. In addition, changes in backbone geometry at Ala205 (Ala482) would also affect the succeeding residue Arg206 (Arg483), which contributes directly to the selectivity filter and is strictly conserved.

The residues within van der Waals contact distance from the side chains of the less well-conserved Pro/Ala pair [Ala70 (Ser354)/Pro204 (Leu481)] are Pro240 (Pro516), Ile241 (Phe517) and Ala244 (Ala519) in TMD6. Side chains of none of these residues line the channel or contain polar groups and are therefore expected to interact less rigidly with the channel residues and to be more tolerant to structural variation without loss of channel function. Ala70 (Ser354) is succeeded in sequence by Val71 (Ile355), which contributes to the channel surface but in an unspecific manner, reflected in its replacement by other hydrophobic residues across species. The substitution of Ala70 (Ser354) or Pro204 (Leu481) can therefore lead to a functional channel and may be a way to fine-tune channel properties.

Yfl054-like aquaglyceroporins

Yfl054 was identified following sequencing of the genome of S. cerevisiae. Proteins similar to Yfl054 have been found by database searches in other yeasts and a Yfl054-like protein seems to be the only aquaporin in the fission yeast S. pombe.

Yfl054-like proteins are characterized by an approx. 350 amino acids long N-terminal extension and an approx. 50 amino acid C-terminal extension. The core transmembrane part is strikingly well conserved among the different yeast proteins, even when including the homologue from the only distantly related fission yeast. The long N-terminal extension is less well conserved but shows one stretch of significant sequence similarity. This sequence (PVWSLNQPLPV) is perfectly conserved among yeasts, whereas it is partly conserved for the filamentous fungi (PVWSLXXPLPV for A. nidulans and PXXSLXXPLPX for F. gramineum). However, the non-conserved residues generally have the same chemical properties as the yeast sequence.

There have been attempts to associate a physiological role to S. cerevisiae Yfl054 and S. pombe SPAC977.17, so far without success (M.J. Tamàs and S. Hohmann, unpublished data; Kayingo et al., 2004), although one report indicated that it could have a function redundant to Fps1 in ethanol-affected glycerol transport (Oliveira et al., 2003). Attempts to identify the cellular membrane in which Yfl054 is localized have also been unsuccessful. It seems, however, that Yfl054 has a function that is distinct from that of Fps1. Lucas and co-workers reported that deletion of either FPS1 or YFL054 causes enhanced passive diffusion of ethanol (Oliveira et al., 2003), which could be due to altered membrane composition. Such an effect has been reported for the deletion of FPS1 (Toh et al., 2001).

Given the conservation of domain structure and sequence, it is clear that Yfl054 confers a specific role associated with transmembrane solute fluxes and it may be involved in regulatory processes through its long N-terminus. More work is needed to decipher such roles.

Other aquaglyceroporins

The third group of aquaglyceroporins is found only in filamentous fungi. This group appears to consist of members that share limited sequence similarity and differ in size. Whereas BLAST searches using such aquaglyceroporin sequences identify members of this group among the top scores, other aquaglyceroporins, such as those from Trypanosoma or Yfl054 score equally high. Several members of this group also have long extensions, especially at the N-terminus, but those appear to be unrelated to that of Yfl054. Although the Trypanosoma proteins have recently been characterized (Uzcategui et al., 2004), to our knowledge no research has been performed on these fungal proteins and hence nothing is known about expression patterns, localization, cell-type specificity or physiological function.


Fungal aquaporins seem to be quite a diverse group of proteins with unique functional, regulatory and physiological properties. When genome sequences of sufficient coverage and quality become available, it will again be possible to draw more detailed pictures of the evolution and occurrence of aquaporins in fungi. What is clearly needed are further studies on the function, expression, localization and physiological roles of fungal aquaporins, which seems perfectly feasible given the fact that several of the fungi listed in the present study are experimental organisms in which genetic manipulation can be performed fast and with high precision. On the other hand, given the difficulty in associating physiological functions to several of the yeast aquaporins, it apparently requires some scientific creativity to elucidate the precise cellular role of these proteins. In the laboratory, aquaporins are not usually needed for survival or fitness.

This should, however, rather encourage further studies on fungal aquaporins. Given the diversity of these proteins and the knowledge collected so far, for instance on yeast Fps1, it is clear that studies on yeast and filamentous fungi bear potential to discover new principles of aquaporin regulation and physiological function, as well as osmoregulation and cell physiology in general.

Sequence files

The sequences used in the present study as well as several multiple alignments can be viewed and downloaded at


This work was supported by grants from the European Commission (QLK3-CT2000-00778, QLK3-CT2000-52116 and QLK3-CT2001-00987), the Human Frontier Science Organisation (RG-0021-2000-M) and the Swedish Research Council (research position to S.H.).