Growth of cells in contact with an abiotic or biological surface profoundly affects cellular physiology. In the opportunistic human pathogen, Candida albicans, growth on a semi-solid matrix such as agar results in invasive filamentation, a process in which cells change their morphology to highly elongated filamentous hyphae that grow into the matrix. We hypothesized that a plasma membrane receptor-type protein would sense the presence of matrix and activate a signal transduction cascade, thus promoting invasive filamentation. In this communication, we demonstrate that during growth in contact with a semi-solid surface, activation of a MAP kinase, Cek1p, is promoted, in part, by a plasma membrane protein termed Dfi1p and results in invasive filamentation. A C. albicans mutant lacking Dfi1p showed reduced virulence in a murine model of disseminated candidiasis. Dfi1p is a relatively small, integral membrane protein that localizes to the plasma membrane. Some Dfi1p molecules become cross-linked to the carbohydrate polymers of the cell wall. Thus, Dfi1p is capable of linking the cell wall to the plasma membrane and cytoplasm.
As a human commensal, C. albicans resides on mucosal surfaces such as the tongue, gastrointestinal and genitourinary tracts (Odds, 1987; Soll et al., 1991). Immunocompromized hosts are susceptible to superficial C. albicans mucosal infections such as thrush or vaginitis, and to life-threatening disseminated disease. In a compromised host, C. albicans produces characteristic invasive lesions in which filamentous cells, either hyphae or pseudohyphae (Sudbery et al., 2004; Berman, 2006), invade the tissue of the host. This behaviour allows the organism to breach epithelial barriers and reach the blood stream, resulting in disseminated disease.
In the laboratory, C. albicans cells respond to growth in contact with agar medium by producing filaments that invade the agar. Production of invasive hyphae during growth in laboratory medium may occur by the same mechanism that is involved in production of invasive lesions during candidiasis. How fungal cells sense that they are growing on an agar matrix and respond by producing invasive filaments is not well understood. However, signalling events occurring in such cells have been detected. In response to growth on an agar surface, C. albicans cells activate a mitogen-activated protein kinase (MAPK) of the ERK1/2 (extracellularly regulated kinase) superfamily called Mkc1p (Kumamoto, 2005). Among plant pathogens, homologues of Mkc1p and a second MAPK are necessary for tissue invasion and pathogenesis (Doehlemann et al., 2006). Therefore, the goal of the present study was to identify a C. albicans plasma membrane protein important for initiation of invasive filamentation and matrix-dependent MAPK signalling.
Here we describe a gene, Orf19.7084, renamed DFI1 (defective in filamentous invasion 1), which encodes a cell surface glycoprotein that promotes matrix-dependent activation of Cek1p. Dfi1p is also required for full C. albicans virulence in the intravenously inoculated mouse model of disseminated candidiasis. A glycine-rich transmembrane segment containing a GxxxG motif, similar to the dimerization motif found in the mammalian red blood cell protein glycophorin A (Brosig and Langosch, 1998; Smith et al., 2001), is important for Dfi1p function. A subpopulation of Dfi1p is covalently bound to the cell wall, allowing Dfi1p to monitor events affecting the cell wall. Initiation of invasive filamentation may thus involve responses to events that alter the properties of the carbohydrate-rich cell wall.
Signal transduction pathways activated during growth on a semi-solid surface
The MAPK Mkc1p is activated during growth of C. albicans on agar matrix (Fig. 1A, P-M) (Kumamoto, 2005). To identify other MAPKs that are similarly activated, extracts of cells grown in liquid medium (YPS) were compared with extracts of cells grown on the surface of agar medium (YPSA) by Western blotting with antiserum that detects dually phosphorylated forms of ERK1/2 superfamily MAPKs. Activation of a second MAPK was detected in extracts of surface-grown cells at higher levels than in liquid-grown cells (Fig. 1A, P-C). The electrophoretic mobility of this MAPK was consistent with the molecular weight of the MAPK Cek1p (49 kDa) (Fig. 1A, third panel). Extracts of strain CCC55 (cek1 null mutant; Csank et al., 1998) yielded no 49 kDa signal with either anti-phospho-MAPK or anti-Cek1p antiserum (Fig. 1A, lane 1). Furthermore, extracts of surface-grown cells of strain CCC81, which lack a phosphatase thought to act on phospho-Cek1p (Csank et al., 1997), showed increased amounts of phospho-Cek1p (Fig. 1A, lane 2). Therefore, Cek1p is activated during growth on the surface of agar matrix.
Previous studies showed that a mutant lacking Cek1p is defective in filamentation during growth on the surface of several types of agar media such as mannitol-containing Lee's medium, Spider medium or low ammonia medium (Csank et al., 1998). In addition, when grown on the surface of YPSA, CCC55 cells (cek1 null mutant) failed to adhere to the agar and did not produce invasive filaments; when embedded within YPS agar, the mutant was delayed in producing filaments (data not shown). In contrast, strain CCC81, lacking the phosphatase Cpp1p, is hyperinvasive when grown under non-hypha inducing conditions, such as growth on agar medium at 25°C (Csank et al., 1997). These findings argue that activation of Cek1p promotes adhesion to an agar surface and invasion but that filamentation when embedded in agar can occur in the absence of Cek1p.
A gene required for C. albicans invasive filamentation
To understand events that lead to activation of MAPKs and invasion of a semi-solid material, we sought to identify a plasma membrane protein that initiates the signalling for matrix-dependent activation of Mkc1p or Cek1p. Several C. albicans candidate genes encoding putative membrane proteins were deleted (Table 1). Four candidate genes encoded signalling proteins or were homologous to S. cerevisiae proteins involved in activation of MAPKs (orf19.4772, orf19.1490, orf19.5867 and orf19.5537) (Roman et al., 2005; 2009). Four candidate genes lacked clear orthologues in S. cerevisiae and were chosen on the basis of predicted structure or presence of motifs (orf19.7084, orf19.207, orf19.4906 and orf19.1488). Three independent null mutants were generated for each gene tested. Invasive filamentation when embedded or plated on the surface of agar medium, and growth in the presence of the cell wall active agents Congo Red and Calcofluor White were tested.
Table 1. Phenotypes of mutant strains lacking the following putative membrane or cell wall proteins.
Mutants lacking orf19.7084 exhibited defects in filamentation during embedded growth (Fig. 2 and Table 1); mutants lacking CaSHO1 (orf19.4772), or CaMSB2 (orf19.1490), were delayed in producing filaments under these conditions (Fig. 2 and Table 1). The other mutants were not defective in invasive filamentation (Fig. 2 and Table 1). The rest of this report will focus on orf19.7084 (termed DFI1 for defective in filamentous invasion). DFI1 is predicted to encode a 36.5 kDa transmembrane protein with a hydrophobic N-terminal region, a large, extracellular, serine and threonine-rich portion, a second hydrophobic region and a short C-terminal tail. Orthologues of Dfi1p are encoded in the genomes of a few other fungi such as Candida tropicalis, Lodderomyces elongisporus and Debaryomyces hansenii but are missing from more distantly related fungi such as S. cerevisiae, plants or animals. Similarities to the Pfam HMM Mid2 protein family (30% identity over the C-terminal 35% of Dfi1p) and to the SKG6 transmembrane α-helix domain (39% identity over the transmembrane and juxtamembrane regions) were detected. Similarities to BLOCKS from carbohydrate binding WSC proteins and flocculins, and motifs for N-linked glycosylation were also found.
dfi1 null mutants are defective in invasion of agar medium
When grown on the surface of YPSA at either 25°C or 37°C, cells of the wild-type (WT) strain adhered to the surface and invaded the agar (Fig. 3A). Under these conditions, the dfi1 null mutant was deficient in adhering to and invading the agar (Fig. 3A). Similarly, when the strains were grown embedded in agar medium at 25°C or 37°C, the WT strain produced invasive filaments but the dfi1 null mutant failed to do so (Fig. 3B and data not shown).
However, dfi1 null mutants were not defective in growth rate (e.g. in YPD liquid medium at 30°C, doubling times for WT and the dfi1 null mutant were 2.0 h). The heterozygous DFI1/dfi1 strain behaved like the WT parent strain (data not shown). Introduction of either a WT, untagged DFI1 allele or tagged DFI1 alleles at the DFI1 locus restored agar invasion and filamentation under embedded conditions, demonstrating that the defect was due to the lack of Dfi1p (Fig. 3A–C).
In YPD liquid medium supplemented with serum (10%; Fig. 3D), Spider liquid medium (data not shown) or RPMI liquid medium (data not shown) at 37°C, dfi1 null mutants germinated and produced elongated hyphae to the same extent as WT cells, demonstrating that dfi1 null mutants retained the ability to produce hyphae. The dfi1 null mutant also adhered to polystyrene and germinated normally under biofilm forming conditions (data not shown). Thus, cells lacking Dfi1p failed to produce filaments in response to growth in contact with agar matrix, but were not inherently unable to form filaments.
Dfi1p function promotes matrix-dependent activation of Cek1p kinase
To determine the effect of Dfi1p on activation of MAPKs, extracts were made from WT or dfi1 null mutants grown in liquid medium or on the surface of agar. As shown in Fig. 1A, extracts obtained from surface-grown dfi1 null cells (Fig. 1A, lane 6) contained lower than WT levels of phospho-Cek1p (Fig. 1A, lane 5) but levels of phospho-Mkc1p were unchanged. In liquid medium neither WT nor dfi1 null mutant strains contained high levels of activated Cek1p or Mkc1p (Fig. 1A, lanes 3 and 4). Low levels of Cek1p activation were observed in liquid cultures in exponential phase, or early or late post-exponential phase (data not shown), consistent with previous observations (Roman et al., 2009). Levels of total Cek1p (Fig. 1A, third panel) and actin (Fig. 1A, bottom panel) were comparable in all samples. The defect in activation of Cek1p was complemented by introduction of a WT copy of DFI1 into the dfi1 null mutant (Fig. 1B). Therefore, the dfi1 null mutant failed to activate Cek1p to WT levels in response to growth on an agar surface.
Increasing levels of phospho-Cek1p in a dfi1 null mutant restores invasive filamentation
During growth on the surface of agar medium, Cek1p is not activated to normal levels in the absence of Dfi1p. If this reduction in phospho-Cek1p level is important, then increasing the amount of activated Cek1p in a dfi1 null mutant might restore agar invasion. Because the phosphatase Cpp1p dephosphorylates phospho-Cek1p (Csank et al., 1997 and Fig. 1A), reducing the gene dosage of CPP1 should increase phospho-Cek1p levels. Therefore, one allele of CPP1 phosphatase was deleted in WT or dfi1 null mutant strains. As shown in Fig. 4A, during growth on agar medium, strain pcz6 (dfi1/dfi1, CPP1/cpp1) accumulated higher levels of phospho-Cek1p than the dfi1 null strain. The amount of phospho-Mkc1p signal was much higher than the phospho-Cek1p signal and did not change substantially in different strains (Fig. 4B).
When grown on the surface of YPSA medium, cells of strain pcz6 (dfi1/dfi1, CPP1/cpp1) adhered to the surface and invaded the agar (Fig. 4C). When embedded in YPSA, strain pcz6 (dfi1/dfi1, CPP1/cpp1) produced filaments at levels resembling the WT strain (Fig. 4D). Therefore, increased invasive filamentation accompanied an increased level of phospho-Cek1p. Furthermore, the CPP1/cpp1 heterozygote strain was hyperinvasive when compared with the WT strain, in agreement with the literature (Csank et al., 1997). These results support the conclusion that activation of Cek1p because of the activity of Dfi1p promotes adhesion and invasive filamentation in response to growth in contact with semi-solid material.
A GxxxG glycophorin A motif in the transmembrane domain of Dfi1p is important for Dfi1p function
The putative transmembrane domain of Dfi1p contains a GxxxG motif (Fig. 5A) that is very similar to the GxxxG dimerization motif first described in glycophorin A (Brosig and Langosch, 1998), and later observed in numerous transmembrane proteins (Smith et al., 2001; Curran and Engelman, 2003). To determine whether this motif was important for Dfi1p function, a mutant allele in which glycine 273 and glycine 277 were replaced with leucines (dfi1G273,277L) was constructed. When grown on the surface of YPSA medium, a strain that expressed mutant Dfi1G273,277Lp-TAP as the sole form of Dfi1p failed to adhere to the agar or to produce filaments that invaded the surface (Fig. 5C). The mutant was also defective in invasive filamentation when grown embedded within agar medium (Fig. 5D).
In addition, the dfi1G273,277L mutant failed to activate Cek1p to WT levels when grown on an agar surface (Fig. 5B). The mutant protein was expressed and localized normally (discussed below). Therefore, the GXXXG motif is important for the ability of Dfi1p to promote Cek1p activation and invasive filamentation during growth on an agar surface, suggesting that multimerization is important for the function of Dfi1p.
Localization and processing of Dfi1p
The DFI1 gene is predicted to encode a 36.5 kDa glycosylated protein (both O- and N-linked glycosylation) with two hydrophobic domains. To determine the localization of the protein, DFI1 was tagged at the C-terminus with GFP or HA and the tagged proteins were analysed. All tagged alleles were under the control of the DFI1 promoter, retained their native 3′ untranslated regions, and yielded functional proteins as determined by the ability to complement the defect of dfi1 null mutants in invasive filamentation (Fig. 3C).
To determine the subcellular localization of Dfi1p, cells containing DFI1-GFP were grown on a YPD agar plate, scraped off the plate and observed. The cells showed a marked ring of fluorescence around the cell perimeter and at the cell septum (Fig. 6, panels 3 and 5). Identical Dfi1p-GFP localization was seen in cells grown in liquid under conditions that promote growth as either yeast or hyphae (data not shown).
Biochemical fractionation of C-terminally HA-tagged alleles showed that Dfi1p-HA fractionated with the membrane pellet. Following centrifugation of a total extract (Fig. 7A, lane 1) at 300 000 g for 1 h, Dfi1p-HA molecules were recovered from the pellet (Fig. 7A, lane 3) but not the supernatant (Fig. 7A, lane 2). When the pellet fraction was treated with nonionic detergent (Fig. 7A, lanes 4, 5, 8 and 9), Dfi1p-HA molecules were extracted. However, treatment with 4 M urea (Fig. 7A, lanes 6 and 7), or 0.1 M NaCO3 (Fig. 7A, lanes 10 and 11) or with buffer only (Fig. 7A, lanes 2 and 3) failed to extract Dfi1p-HA. Therefore, the extractable Dfi1p molecules are integrally associated with a membrane. Based on these combined data, we conclude that Dfi1p is an integral plasma membrane protein.
Dfi1p-HA migrated on SDS-PAGE as a series of isoforms ranging in apparent molecular weight from 170 to 220 kDa (in contrast to the predicted 36 kDa). To determine whether O-linked glycosylation contributed to the slow mobility of Dfi1p, tagged DFI1 alleles were introduced into mutants that were defective in glycosylation [mnt1/mnt1, mnt2/mnt2 and mnt1/mnt1 mnt2/mnt2 double null mutants (Munro et al., 2005)]. Mnt1p and Mnt2p are required to add the second to fifth mannosyl residues at O-glycosylation sites. In addition, digestion of extracts with peptide : N-glycosidase F (PNGase F; New England Biolabs), an amidase that removes N-linked oligosaccharides from glycoproteins, was used to remove N-linked glycosylation. For these experiments, Dfi1p was tagged with a tandem His6x and HA tag (Dfi1p-TAP), which yielded a functional tagged protein (Fig. 3C). Dfi1p-TAP extracts digested with PNGase F showed a reduction in apparent molecular weight from ∼170–220 without digestion (Fig. 7B, lane 5) to ∼110 kDa after digestion (Fig. 7B, lane 1) indicating that Dfi1p is N-glycosylated.
In the mnt null mutants (Fig. 7B, lanes 2–4), the apparent molecular weight of PNGase F-digested Dfi1p-TAP decreased further to approximately 90 kDa in the mnt1/mnt2 double mutant. Increasing enzyme concentration or time of digestion did not increase the mobility of Dfi1p (data not shown). In addition, sequential treatment of Dfi1p-TAP with α-mannosidase and PNGase F resulted in a species that migrated at approximately 90 kDa, similar in size to the species shown in Fig. 7B, lane 4 (data not shown). Therefore, Mnt1p and Mnt2p are involved in the O-glycosylation of Dfi1p-TAP.
The Dfi1p products obtained by PNGase F and α-mannosidase digestion or by using glycosylation mutants still exhibited slower mobility than would be expected based on the predicted molecular weight of the Dfi1p polypeptide. Because the first mannose residue in the oligosaccharide remains attached to the polypeptide following α-mannosidase digestion and mnt1/mnt2 mutations do not prevent addition of the first residue, this residual carbohydrate could be responsible for the altered mobility of the Dfi1p polypeptide. Therefore, Dfi1p-HA was expressed in an S. cerevisiae sec53-1 mutant, a mutant blocked in the first step of both N- and O-glycosylation (Ruohola and Ferro-Novick, 1987). Following incubation at the non-permissive temperature (37°C), a Dfi1p-HA species with an apparent molecular weight of 58 kDa, rather than the predicted molecular weight of 40 kDa for the tagged species, was produced (Fig. 7C). Therefore, we concluded that unglycosylated Dfi1p exhibits an aberrant mobility in SDS-PAGE. In summary, these results show that Dfi1p is a cell surface-associated glycoprotein.
To test for association of Dfi1p with the cell wall, cell wall preparations were isolated from strain pcz11 carrying 2 tagged alleles of DFI1. Cell wall fractions were treated with or without Quantazyme, a recombinant β-1,3-glucanase, to digest the cell wall and release cell wall-bound proteins. The β-glucanase-treated sample released Dfi1p-TAP molecules (Fig. 7D, lane 2) while the untreated sample did not (Fig. 7D, lane 1). No protein was released upon treatment of isolated cell walls with chitinase or 4 M urea (data not shown). Because cell wall-bound Dfi1p was detected via its C-terminal tag, at least some of the wall-bound Dfi1p must retain its transmembrane domain, thus linking the cell wall to the plasma membrane.
In addition, treatment of cell wall preparations from dfi1/DFI1-TAP, and DFI1-TAP/DFI1-TAP strains with PNGase F, the amidase that removes N-linked oligosaccharides from glycoproteins, released Dfi1p-TAP molecules from the cell wall (Fig. 7D). No Dfi1p-TAP signal was released from dfi1 null cell walls or cell walls incubated without enzyme. PNGase F may cleave a peptide-carbohydrate bond that links Dfi1p to the cell wall or remove N-linked oligosaccharides that are cross-linked to the wall. We conclude that there is a subpopulation of Dfi1p that is released from cells with detergent and another subpopulation that is released only by enzymatic digestion.
To determine whether the material released by enzymatic digestion constituted a major or minor fraction of the total Dfi1p-TAP, cells were extracted with either detergent treatment alone or with detergent plus PNGase treatment. Following extraction in the presence of detergent, neither Dfi1p-TAP nor the control protein actin was extracted from the insoluble pellet fraction by further treatment with detergent alone (Fig. 7E). However, some Dfip-TAP molecules were extracted from this insoluble fraction by PNGase F treatment (Fig. 7E). Of the total full-length Dfi1-TAP molecules that could be extracted from cells (Fig. 7E, lanes labelled T), the majority (about 80–90%) were extracted with detergent alone (Fig. 7E, lanes labelled S) and the remainder (10–20%) required PNGase digestion in order to be released from the insoluble fraction (Fig. 7E, lanes labelled I).
The Dfi1G273,277Lp-TAP mutant protein was produced at normal levels and acquired N-linked glycosylation (Fig. 7F). The mutant protein fractionated similarly to WT protein, became cross-linked to the cell wall and localized to the cell perimeter (Fig. 6, panel 7). Therefore, the mutant protein was properly localized but failed to promote Cek1p activation during growth in contact with semi-solid material and failed to adhere to or invade agar medium.
Dfi1p is needed for growth in the presence of cell wall targeting agents
Because some Dfi1p molecules are covalently bound to the cell wall, Dfi1p may play a role in monitoring the status of the cell wall. The dfi1 null mutant was found to be hypersensitive to the glucan synthase inhibitor Caspofungin (Fig. 8) and the cell wall disturbing agents Congo Red (Fig. 8) and Calcofluor White (data not shown). This hypersensitivity was not remediated by sorbitol. Furthermore, dfi1 null mutants grew like WT strains in the presence of other stresses (1.5 M NaCl, 1 M sorbitol, 0.3 M CaCl2, 0.1 M LiCl, 10 mM MnSO4, 0.05% SDS, pH 8.8, 18°C or 42°C; data not shown). These results show that Dfi1p is involved in the maintenance of the cell wall.
In contrast, the dfi1G273,277L mutant strain grew like WT in the presence of Congo Red and Caspofungin (Fig. 8A). Because the mutant protein was properly localized, cross-linked to the wall and allowed WT resistance to drugs, the cell wall of the dfi1G273,277L mutant strain did not appear to be compromised. Therefore, the failure of this mutant protein to promote matrix-dependent activation of Cek1p demonstrates that one of the functions of Dfi1p is to participate in activation of signalling in response to growth on a semi-solid surface.
Strain pcz6 (dfi1/dfi1, CPP1/cpp1) was as sensitive to Congo Red and Caspofungin as the dfi1 null mutant (Fig. 8B) despite its ability to invade agar and accumulate matrix-activated Cek1p. Therefore, unlike the defect in invasive filamentation, a reduction in the amount of Cpp1p phosphatase did not bypass the sensitivity of the dfi1 null mutant to cell wall active drugs. Furthermore, under the conditions of our experiment, strain CCC55 (cek1 null mutant) grew like WT in the presence of Caspofungin (Fig. 8C). These results show that in addition to its function in promoting Cek1p activation and invasive filamentation, Dfi1p has other functions in cellular physiology.
C. albicans dfi1 null mutant was attenuated in virulence
To determine whether Dfi1p was important for virulence, the behaviour of the mutant lacking Dfi1p (pcz25) was analysed in the hematogenously disseminated murine model, in comparison with a WT strain (pcz24) and the reconstituted null mutant (pcz26). As described in Experimental procedures, 3 × 105 cells from each strain were injected into 16 mice per strain via the lateral tail vein and survival of the mice was monitored as a function of time (Fig. 9). The results showed that both the WT strain and the DFI1+ reconstituted strain killed the majority of the mice while killing by the dfi1 null mutant was reduced. The survival curves for mice inoculated with WT or dfi1 null mutant were significantly different (P < 0.001 by log rank test with Sidak correction for multiple comparisons). Therefore, the strain lacking Dfi1p was attenuated in its ability to cause lethal systemic infection.
Numerous species of fungi sense and react to the presence of a surface. For example, the differentiation of Uromyces appendiculatus in response to contact with ridges (Hoch et al., 1987; Zhou et al., 1991) and the response of C. albicans hyphae to contact with obstructions (Brand et al., 2007; 2009) depend on the rapid activation of mechanosensitive ion channels. In contrast, responses to sustained contact with a matrix, as seen in matrix-induced invasive filamentation of C. albicans (Brown et al., 1999), are less well understood. Here we demonstrate that the cell surface protein Dfi1p promotes matrix-dependent Cek1p activation and invasive filamentation of C. albicans.
Dfi1p is an integral plasma membrane protein, a fraction of which is covalently bound to the cell wall. Full-length Dfi1p can be released from a cell wall fraction by enzymatic digestion of the β-glucan polymer or by digestion with PNGase F but it is not released by SDS or chaotropic agents. In addition, the Dfi1p polypeptide is required for normal resistance of C. albicans to drugs that perturb the cell wall, such as Caspofungin and Congo Red.
Sensing the presence of matrix
The cell wall is a cross-linked mesh of carbohydrate polymers containing proteins that are covalently and non-covalently associated (Chaffin, 2008; Sosinska et al., 2008). Other carbohydrate polymer-rich materials such as mammalian mucus are affected by numerous parameters of the environment, which alter their viscoelastic properties. For example, several factors including the degree of hydration, ion homeostasis, pH and the presence of oxidizing agents affect lung mucus (Singh and Hollingsworth, 2006). Analogously, contact of a fungal cell wall with a semi-solid surface such as agar may alter the viscoelastic properties of the wall. Seen in this light, the fungal cell wall is not merely a protective structure designed to prevent damage to the cell. Rather, the carbohydrate layers of the cell wall may have an additional important function as sensors of environmental stimuli.
The ability to sense cell wall perturbation underlies the function of other fungal proteins. In S. cerevisiae, the plasma membrane proteins Wsc1p and Mid2p have been proposed to be mechanosensors that interact with the cell wall and sense perturbations of the wall (Levin, 2005). A Wsc1p derivative that was engineered to extend through the yeast cell wall was shown to have spring-like properties (Dupres et al., 2009). Importantly, the spring constant of the protein was altered by treatments that perturbed the cell wall, showing that changes in the wall impacted the mechanical properties of the protein (Dupres et al., 2009). Perhaps perturbation of the cell wall affects the behaviour of Wsc1p because, like Dfi1p, Wsc1p interacts tightly, sometimes covalently, with the cell wall. Because insoluble fractions are not routinely treated with PNGase F to detect linked molecules, it may not be unusual for cell surface glycoproteins to be cross-linked to the cell wall.
Sensing of environmental parameters through their effects on the viscoelastic properties of carbohydrates occurs in other cell types. For example, mammalian mucins, highly glycosylated proteins found on the apical surfaces of most simple secretory epithelia (Cullen, 2007), in conjunction with about one hundred other proteins, form mucus, an interconnected mesh of carbohydrate and protein. A membrane tethered mucin, Muc1p, has been proposed to relay changes in pH and ionic composition through their effects on Muc1p conformation. In endothelial cells, the glycocalyx, an extracellular layer composed of polysaccharide-rich molecules, participates in mechanosensing and the response to shear stress (Tarbell and Ebong, 2008). Thus, sensing the effects of the environment on the properties of carbohydrate polymers may be a general mechanism for detecting critical environmental parameters.
MAPK pathways and invasive filamentation
In accord with the model that Dfi1p is needed for the cellular response to changes in the cell wall, the dfi1 null mutant is hypersensitive to cell wall disturbing agents such as Caspofungin and Congo Red. This hypersensitivity probably reflects a second function for Dfi1p that is not mediated through activation of Cek1p kinase. Although liquid-grown cells challenged with high levels of Congo Red transiently accumulate activated Cek1p (Eisman et al., 2006), reduction in the level of phospho-Cek1p phosphatase Cpp1 did not reduce the hypersensitivity of the dfi1 mutant strain to Congo Red. Conversely, the DFI1G273,277L mutant was resistant to Congo Red despite the fact that this mutant failed to accumulate normal levels of activated Cek1p in response to surface contact. In addition, strain CCC55 (cek1 null mutant) was not hypersensitive to Caspofungin under the conditions of our experiment, while the dfi1 null mutant was very sensitive to this agent. These results show that in addition to its ability to signal, Dfi1p has at least one function that is not mediated through activation of the Cek1p pathway.
Our data suggest that a certain threshold of activated Cek1p kinase is needed to promote invasive filamentation because, in the dfi1 null mutant, low levels of activated Cek1p were detected in surface-grown cells even though these cells failed to produce invasive filaments. In contrast, the DFI1 heterozygous strain, which produced slightly reduced levels of contact-activated Cek1p when grown on agar, filamented like the WT strain. Increasing levels of phospho-Cek1p in a dfi1 null mutant to levels equivalent to the levels in the WT strain by deleting one allele of CPP1 resulted in invasive filamentation. Deleting one allele of CPP1 in a WT strain increased both invasive filamentation and the levels of phospho-Cek1p but did not result in constitutive filamentation, consistent with the observation that increased levels of phospho-Cek1p were not observed in liquid-grown cells lacking Cpp1p (Eisman et al., 2006). All together, these data substantiate the conclusion that there is a critical ratio of phospho-Cek1p to Cek1p that must be achieved in order for a colony to undergo invasive filamentation.
Strain CCC55 (the cek1 null mutant; Csank et al., 1998) was defective in adhering to an agar surface and did not invade the agar. Strain CCC55 also exhibited a delayed embedded filamentation phenotype; however, the phenotype of CCC55 was weaker than the strong filamentation defect observed for pcz5 (the dfi1 null mutant). The difference in the phenotypes of CCC55 and pcz5 may be related to the fact that CCC55 lacks both phospho-Cek1p and unphosphorylated Cek1p. As seen in S. cerevisiae (Cook et al., 1997), unphosphorylated Cek1p may have an inhibitory effect on invasive filamentation.
In addition, residual matrix-dependent Cek1p activation occurs in the dfi1 null mutant showing that other pathways for matrix-dependent activation of Cek1p exist. Two other plasma membrane proteins that are known to signal to the Cek1p pathway, Sho1p (Roman et al., 2005) and Msb2p (Roman et al., 2009) play roles in the response of cells to growth in contact with matrix. Mutants lacking either of these proteins exhibited delays in filamentation when grown embedded within agar (Table 2). Therefore, the response of cells to matrix involves multiple sensors of cell surface stimuli.
Additionally, the Mkc1p pathway plays a role in regulating invasive filamentation (Kumamoto, 2005). The involvement of multiple MAPKs in regulating invasion is also seen in other fungi such as the phytopathogenic fungus Magnaporthe grisea, where the Cek1p orthologue is needed for infection structure formation and the Mkc1p orthologue for penetration (Xu and Hamer, 1996; Xu et al., 1998; Zhao et al., 2007). Thus, like filamentation in liquid medium, invasion during growth on a surface is regulated by multiple pathways.
Matrix sensing and host – Candida interactions
In humans, C. albicans is found on the skin, in lungs, in the oral cavity, in the stomach, in the intestines and in the vagina. In healthy individuals, C. albicans colonization of these sites is generally benign. If during colonization, the host's immune status was to decline and C. albicans organisms were able to adhere and grow on a tissue surface, the fungal responses to the presence of semi-solid material described here would activate invasive filamentation. These events would contribute to the development of invasive lesions and potentially life-threatening candidiasis. The observation that a mutant lacking Dfi1p was attenuated in the ability to produce fatal disseminated infection following intravenous inoculation of mice is consistent with this model. Therefore, the Dfi1p protein, a cell wall component that promotes virulence and invasive filamentation, represents an attractive target for future antifungal drug development.
As a colonizer of human mucosal surfaces, C. albicans cells are exposed to many of the same environmental stresses that affect the mucus layers in the host. In both the host and the microorganism, the ability to respond to such stresses with a change in cellular physiology may be mediated through signalling proteins that interact with carbohydrate layers, i.e. Dfi1p for the fungus and mucins for the host. Thus, in this respect, C. albicans and its host share the ability to monitor the same environmental parameters. Such fine tuned environmental sensing may be a key element in the success of C. albicans as a colonizer and opportunistic pathogen.
Candida albicans strains are listed in Table 2. The cek1 and cpp1 null mutants were kindly provided by M. Whiteway and have been described previously (Csank et al., 1997; Csank et al., 1998). The mnt1 single null, mnt2 single null and mnt1/mnt2 double null mutants were the kind gift of Dr N. Gow (Munro et al., 2005). The S. cerevisiae sec53 mutant (Ruohola and Ferro-Novick, 1987) was kindly provided by Dr S. Ferro-Novick. Escherichia coli strains DH5α, XL1Blue or KO1067 (a dam/dcm derivative of E. coli K12) grown in L broth plus required antibiotic were used to propagate plasmids.
Candida albicans strains were derived from strain BWP17 (Wilson et al., 1999). DNA constructs were introduced into C. albicans following the transformation protocol described in Reuss et al. (2004). Null mutant strains were confirmed by Southern blotting (Ausubel et al., 1989). When needed, strains that carried the SAT1 (Reuss et al., 2004) flipper cassette were induced to express the site-specific recombinase by growing the cells on liquid YPS for 24 h. Under these conditions, multiple integrated flipper constructs (SAT1 and URA3 (Morschhauser et al., 1999), or SAT1 and miniHIS1, see below) could be resolved at once. Following growth on YPS, colonies that had lost the markers of interest were identified by replica plating onto appropriate media. Deletion of markers from the genome was confirmed by PCR.
After excision of the URA3 flipper, to restore URA3 prototrophy, primers pz203 and pz204 were used to PCR amplify the URA3 locus from the genome of C. albicans SC5314. This fragment was cloned into pCR2.1 (Invitrogen), generating pCURA, and sequenced. ura3/ura3 strains were transformed with a PstI/NcoI URA3+ fragment purified from pCURA. To construct pcz24 and pcz25, strains were transformed with HIS1+ and ARG4+ PCR products, and integrations were verified by PCR.
Generation of mutant strains lacking one of eight candidate genes
For every null mutant shown in Table 1, the following strategy was used: using SC5314 genomic DNA template, a first round of PCR generated the following two fragments: a 5′ fragment, generated with ‘5′ Fragment, Forward Primer’ (5′ FFP) and 5′ Fragment, Reverse Primer (5′ FRP), and a 3′ fragment, generated with 3′ FFP and 3′ FRP. Primers are described in Table S1. PCR fragments were gel-purified and used as template in a second PCR reaction that included only the 5′ FFP and 3′ FRP primers to yield a fused fragment that was TOPO cloned into pENTR (Invitrogen). The fused fragments had KpnI and SacI restriction sites at the junction between them. Into this junction, the KpnI/SacI fragments containing the SAT and URA flippers described previously (Morschhauser et al., 1999; Reuss et al., 2004) were cloned. The resulting plasmids were digested with PmeI and BssHII to release the flippers flanked by approximately 500 bp of homology on either side of the relevant open reading frame. Gel-purified fragments were used for transformations and deletions were confirmed by Southern blot. Three independent null mutants were generated for each gene tested.
Construction and use of the URA or SAT placers
These plasmids were engineered from the URA or SAT flippers (Morschhauser et al., 1999; Reuss et al., 2004) (Fig. S1) and allow one step cloning to generate disruption cassettes for gene deletion (described in Supporting information).
Construction of tagged alleles of DFI1
To complement the dfi1 null mutation, primers pz03/pz04 were used to generate a PCR fragment encompassing the WT DFI1 locus. The DFI1 locus was cloned into vector pNEB193 (New England Biolabs) with SalI and BamHI (pNEBDFI1). After sequencing, the fragment was cloned into the URAplacer (pUPL) generating pUPLDFI1. To cross the gene back into the chromosome, pUPLDFI1 was cut with BglII, which cuts at a unique site within the 5′ UTR of DFI1 and the linearized DNA was transformed into a dfi1 null strain.
C-terminally tagged DFI1 alleles were constructed as follows: using primers pz175/176, BspEI and BclI sites were introduced after the last codon in DFI1 to generate pNEBDFI1-Ctags. The HA and GFP tags were amplified using primers pz179/180 (HA3), and pz195/196 (GFP) and plasmids pFA6AHA3 (Longtine et al., 1998) and pYEGFP (Cormack et al., 1997) as templates and introduced into the BspEI/BclI sites of pNEBDFI1-Ctags. All tags were linked to DFI1 via a Ser-Gly4 amino acid linker and all constructs retained the native 3′ UTR of the DFI1 locus.
Mutant alleles of DFI1 were generated by overlap PCR. Briefly, using pNEBDFI1-HA (or TAP or GFP) as template, two mutagenic primers were used to generate two overlapping fragments that were fused together in a final round of PCR. The first round primer pairs used to generate the G273,277L mutated allele were pz247/pz63 and pz134/pz248. The second round PCR used primer pair pz134/pz63. This fragment was digested with XhoI and MluI and moved into pSPLDFI1-TAP (DFI1-TAP in the SAT placer) or pSPLDFI1-GFP (DFI1-GFP in the SAT placer) to generate pSPLDFI1G273,277L-TAP or pSPLDFI1G273,277L-GFP plasmids. To generate pSPLDFI1-TAP (DFI1-TAP in the SAT placer) an oligomer encoding GSGGGHHHHHH was cloned in frame at the BspEI site immediately upstream of the HA tag present in pSPLDFI1-HA.
For introduction of TAP-tagged DFI1 alleles into mnt1 null, mnt2 null and mnt1/mnt2 double null backgrounds, a portion of the 3′ UTR from the DFI1 locus (flanked by BssHII and BglII sites) was amplified by PCR and cloned upstream of the 5′ UTR and the rest of the DFI1 locus in pSPLDFI1-TAP (DFI1-TAP in the SAT placer) so that a unique BglII site was present between the 3′ and 5′ UTR fragments. The resulting plasmid, pSPL2XO-DFI-TAP, was linearized with BglII prior to transformation.
For expression of DFI1-HA in S. cerevisiae, the tagged DFI1 locus was amplified with primers pz22 and pz201 and cloned onto pYES2.1/V5-His-TOPO (Invitrogen).
Molecular biology methods
Unless otherwise stated, all PCR reactions used Hi-Fi polymerase (Invitrogen), Pfu Turbo or Ultra (Stratagene) following the manufacturers' recommendations. Primers are listed in Tables S1 and S2. All restriction enzymes and DNA ligase were purchased from New England Biolabs. All constructs generated by PCR were confirmed by sequencing. Protein sequence analysis was performed with the Biology WorkBench, http://workbench.sdsc.edu, or with blast.
Growth of cells
Candida albicans was routinely grown in YPD (1% yeast extract, 2% peptone, 2% glucose), YPS (1% yeast extract, 2% peptone, 2% sucrose) or CM [complete medium minus amino acid or uridine (Ausubel et al., 1989)] at 25, 30 or 37°C. Nourseothricin (Werner BioAgents, Germany) was used at 200 µg ml−1.
Embedded filamentation assays
The growth of colonies under embedded conditions was performed as described previously (Brown et al., 1999) with the following modifications: in order to minimize plate to plate variations in water activities, following 4 h of growth of a diluted overnight culture, 25 ml of lukewarm 1% YPS agar was pipetted on top of a drop of medium containing approximately 150 cells. Solidified plates were placed inside a humidified chamber for the duration of the experiment in order to minimize uneven drying of the plates. The plates were incubated at either 25 or 37°C. Filamentous colonies were defined as colonies with at least 20 visible filaments. Each assay described was repeated at least three times.
Agar invasion assays
Cells were grown on the surface of YPSA medium for 4 days at 25°C. Colonies were washed off the surface of the agar with a stream of water and gentle rubbing. Adherent cells remaining on the surface of the agar were photographed at 4× power. A sliver of agar was cut through the centre of a colony, flipped on its side and photographed in cross-section at 4× power.
Cultures were grown in liquid YPD supplemented with 10% bovine serum, RPMI medium or Spider medium at 37°C. At various times, cells were collected, stained with 10 µg ml−1 Calcofluor White (Sigma), washed and photographed. These assays were performed in triplicate.
Caspofungin or Congo Red treatment
To test for sensitivity to cell wall disturbing agents, cultures grown at 30°C were diluted into fresh YPD and allowed to grow at 30°C to OD600 ∼ 1.5. Serial dilutions were spotted on YPD plates or YPD plates supplemented with either 200 µg ml−1 Congo Red or 90 ng ml−1 Caspofungin. For Fig. 8C, plates were supplemented with 150 ng ml−1 Capsofungin or 250 µg ml−1 Congo Red. Plates were incubated at 30°C for times indicated. Sensitivity was tested at least three times.
Protein extraction and immunoblotting
To characterize the levels of activated MAPK, cells were grown and extracted as described (Kumamoto, 2005). Cells were grown in liquid medium for 72 h or on the surface of agar medium for 72 or 96 h. Extracts were fractionated by SDS-PAGE and analysed by immunoblotting. For Western blotting, 80–100 µg of total protein (for phospho-MAPK or Cek1p) or 5–20 µg (for actin) was loaded per sample and separated on an 8% SDS-PAGE (MAPK or Cek1p) or 12% (actin). After transfer onto polyvinylidene difluoride membranes (Millipore, 0.2 µm) by standard protocols, blots were initially probed with anti-dually phosphorylated p42/p44 MAPK rabbit polyclonal antiserum (Cell Signalling; CS-9101) at 1:1000 (Figs 1 and 4), overnight at 4°C, or with anti-dually phosphorylated p42/p44 MAPK rabbit monoclonal antibody (Cell Signalling; CS-4370) (Fig. 5). Goat-anti-rabbit IgG-HRP (Zymed, 62–6120) was used at 1:5000 dilution as secondary antibody. Following detection using Amersham ECL system, the blot was stripped and reprobed with anti-Cek1 rabbit polyclonal (1:3000; kindly provided by Dr J. Plá) overnight at 4°C. Anti-Cek1 signal was detected using NEN luminol. Actin loading controls were probed at 1:15 000 with rabbit anti-actin (Sigma, A5060) and detected as described above. ImageQuant software was used to quantify results. Comparisons of MAPK activation in WT and dfi1 null mutants were performed more than 10 times, and results were consistent.
To detect Dfi1p-HA, anti-HA (1:1000 overnight incubation at 4°C) obtained from the Tufts University Monoclonal Grasp Center was used. Goat-anti-mouse-HRP (Cell Signaling, 7076) was used as secondary and the signal was detected with NEN luminol.
Cells were grown in liquid or on plates and collected over ice as described above. All subsequent steps were performed at 4°C. Following a PBS wash, cell pellets were resuspended in 1× EB buffer (0.3 M Sorbitol, 10 mM Tris pH 7.5, 100 mM NaCl, 1 mM MgSO4, 1 mM EDTA) plus protease inhibitors [1 mM PMSF and Fungal Specific protease inhibitor (Sigma P8215) at 1 ml per 20 g cell paste] and broken using a French Press (18 000 psi) or a bead beater (following standard protocols). The extracts were spun twice at 500 g to remove unbroken cells and the supernatant was centrifuged at 300 000 g for 1 h in a Beckman ultracentrifuge (TLA100.3 rotor). The 300 K supernatant was removed and the remaining membrane pellet was homogenized by hand with a Teflon-glass homogenizer in the same buffer (EB), or EB supplemented with 1% Triton X-100, EB + 4 M urea, EB + 4 M urea + 1% Triton X-100 or EB + 0.1 M Na2CO3. Samples were incubated on ice for 30 min, and then centrifuged at 13 000 g for 30 min in a tabletop centrifuge at 4°C. The supernatant was removed, and the pellet was resuspended in a volume equal to the original volume of EB + 1% SDS. Equal volumes of samples were loaded onto 6% SDS-PAGE gels. Fractionation experiments were repeated numerous times.
Cell wall extraction, glucanase, chitinase and PNGase F digestion
Cells were grown in YPD at 30°C to an OD600 of 1.0, and the cell wall fraction was isolated, digested and analysed as described (Pitarch et al., 2008). For digestion, 170 mg wet weight of the cell wall fraction was incubated with 1500 U gm−1 wet weight of Quantazyme (MP) at 37°C for 4–16 h. Alternatively, 200 mg wet weight of cell wall was incubated with 20 000 units of PNGase F (New England Biolabs P0704) at 37°C for 4 h. Samples incubated with or without enzyme were centrifuged at 500 g for 10 min at 4°C, and the supernatants were precipitated with TCA and analysed by Western blotting with anti-HA antibody. Experiments were repeated three times.
For PNGase F digestion of whole cells, cells were grown in YPD at 30°C to an OD600 of approximately 1.5. Cells were harvested by centrifugation and cell pellets were washed three times with ice cold water. The cell pellets (∼50 mg) were resuspended in 600 µl of 1× RIPA buffer (50 mM Tris-HCl pH 8.0, 150 mM NaCl, 0.1% SDS, 1% NP40, 0.5% sodium deoxycholate) plus protease inhibitors (2 mM PMSF, Roche Complete Protease Inhibitor cocktail and Sigma Inhibitor cocktail). After addition of 1/10 volume Denaturation buffer (5% SDS, 400 mM DTT), samples were boiled for 10 min and then frozen in a dry ice/ethanol bath, followed by thawing in an ice/water bath. This process was repeated two more times and then samples were boiled for a final time. Following this fourth boiling, 1/10 volume each of Buffer G7 (0.5 M NaPO4 pH 7.5) and NP40 (10% stock) and 5 µl of Sigma Inhibitor Cocktail mix were added. Half of this sample was transferred to a new tube (‘Total’), and the other half was centrifuged at 15 000 g in a refrigerated table top microcentrifuge at 4°C for 15 min. The supernatant was transferred to a new tube (‘Soluble’). The remaining pellet was washed three times with ice cold water, and collected by centrifugation at 200 g (4°C) for 1 min. The washed pellets were resuspended in the same volume as the Soluble samples, in the same buffer, and labelled ‘Insoluble’. An aliquot from each fraction (Total, Soluble and Insoluble) was transferred to a new tube to serve as an undigested control. The remainder of each sample received PNGase F (500 U µl−1, NEB). Samples with and without PNGase F were incubated at 37°C for 3.5 h. Following digestion, the tubes were centrifuged at 200 g for 15 s, and the supernatants were transferred to new tubes and processed for SDS-PAGE on a 7.5% gel, followed by immunoblotting with anti-HA or anti-actin antibody. A second gel, carrying the same samples, was run, transferred to polyvinylidene difluoride and stained with Coomassie blue; discreet bands were detected in all lanes, and no evidence of protein degradation was observed in any of the lanes.
Intravenously inoculated mouse model of candidiasis
Candida albicans strains were grown overnight in CM-Ura medium and harvested by centrifugation at 3250 g. Cells were washed with PBS twice, resuspended, counted and adjusted to a density of 3 × 106 cells ml−1 in PBS. The cell suspension (0.1 ml containing 3 × 105 cells) was injected into the lateral tail vein of female CF1 mice (18–20 g; Charles River Laboratories, Wilmington, MA). Each C. albicans strain was tested in two experiments in a total of 16 mice and combined results of the two experiments are shown. Survival of mice was monitored daily after infection with C. albicans for 21 days. Statistical analysis of survival curves using the log rank test was performed with SAS version 9.2.
The authors would like to thank the following people for materials, strains and helpful discussions: Dr C. Csank, Dr M. Whiteway, Dr J. Plá, Dr N. Gow, Dr J. Morschhauser, Dr A. Johnson, Dr S. Ferro-Novick, Dr J. Kohler and Dr P. Riggle. We also thank Robin Ruthazer for performing the statistical analysis of survival data. We are grateful to Dr M. Malamy and Dr L. Sonenshein for careful review of the manuscript. This research was supported in part by grants AI38591 and AI081794 from the National Institutes of Health (to C. A. K.). T. R. D. was supported by a predoctoral fellowship from the National Science Foundation.