Secreted aspartic proteinase (Sap) activity contributes to tissue damage in a model of human oral candidosis

Authors


Bernhard Hube. E-mail hube@botanik.uni-hamburg.de; Tel. (+49) 40 42816 411; Fax (+49) 40 42816 513.

Abstract

Secreted aspartic proteinases (Saps) are important virulence factors during Candida albicans mucosal or disseminated infections. A differential expression of individual SAP genes has been shown previously in a model of oral candidosis based on reconstituted human epithelium (RHE), and in the oral cavity of patients. In this study, the ultrastructural localization of distinct groups of Sap isoenzymes expressed during RHE infection was shown by immunoelectron microscopy using specific polyclonal antibodies directed against the gene products of SAP1-3 and SAP4-6. Large amounts of Sap1-3 antigen were found within C. albicans yeast and hyphal cell walls, often predominantly in close contact with epithelial cells, whereas lower quantities of Sap4-6 were detected in hyphal cells. To elucidate the relevance of the expressed Saps during oral infections, we examined the effect of the aspartic proteinase inhibitor, pepstatin A, during infection of the RHE. The extent of lesions caused by the strain SC5314 was found to be strongly reduced by the inhibitor, indicating that proteinase activity contributes to tissue damage in this model. To clarify which of the SAP genes are important for tissue necrosis, the histology of RHE infection with Δsap1, Δsap2, Δsap3, Δsap4-6 and three Δsap1/3 double mutants were examined. Although tissue damage was not blocked completely with these mutants, an attenuated phenotype was observed for each of the single sap null mutants, and was more strongly attenuated in the Δsap1/3 double null mutants. In contrast, the lesions caused by the Δsap4-6 triple mutant were at least as severe as those caused by SC5314. During infection with the mutants, we observed that the SAP gene expression pattern of the Δsap1 and the Δsap1/3 mutants was altered in comparison with the wild-type strain. Expression of SAP5 was observed only during infection with the Δsap1/3 mutant, whereas upregulation of SAP2 and SAP8 transcripts was observed in the Δsap1 and the Δsap1/3 mutants. These results suggest that Sap1-3, but not Sap4-6, contribute to tissue damage in this model. Furthermore, C. albicans may compensate for the deletion of certain SAP genes by upregulation of alternative SAP genes.

Introduction

Hydrolytic enzymes, such as secreted aspartic proteinases (Saps), have often been suggested to be involved in the virulence of the opportunistic pathogen Candida albicans (Cutler, 1991; Odds, 1994). Proteinases of C. albicans consist of a family of isoenzymes encoded by at least 10 SAP genes (Monod et al., 1994; 1998; Hube, 1996; A. Felk, W. Schäfer and B. Hube, unpublished data). These genes have been shown to be regulated differentially in vitro, and a distinct role for different SAP genes has been suggested to occur at different times of the infection process and during different types of infection (Morrow et al., 1992; Hube et al., 1994; Hube, 1996; White and Agabian, 1995; Monod et al., 1998). The importance of Saps in establishing systemic and vaginal C. albicans infections was demonstrated by the protective role of the proteinase inhibitor, pepstatin A, or aspartic proteinase antibodies in experimental animal models (De Bernardis et al., 1997; Fallon et al., 1997). Progress to clarify the relevance of distinct Saps was achieved by the construction of different SAP null mutants (Hube et al., 1997; Sanglard et al., 1997). The importance of Saps for disseminated infections was demonstrated by an increasing survival rate of mice and guinea pigs infected intravenously with Δsap1, Δsap2, Δsap3 (Hube et al., 1997) or Δsap4-6 (Sanglard et al., 1997). The contribution of Sap2 to systemic infection could be explained partially by proposing a role for this enzyme in endothelial cell damage. However, adherence of Δsap1, Δsap2 and Δsap3 mutants to endothelial cells was not affected (Ibrahim et al., 1998). The expression of the closely related genes SAP4-6 was observed by C. albicans in murine macrophages, which exhibited a 53% more effective killing towards the sap4-6 mutant compared with the wild-type strain (Borg-von Zepelin et al., 1998). Using the same mutants, an important role for SAP4-6 but not for SAP1-3 was also shown in a mouse peritonitis model (M. Kretschmar, B. Hube, D. Sangland, M. Schröder et al., submitted). These data, therefore, indicate a role of SAP4-6 in the pathogenesis of invasive candidosis.

In contrast, Δsap2 mutants were almost avirulent during experimental vaginal infections in the rat, and the virulence of Δsap1 and Δsap3, but not Δsap4-6, was attenuated in comparison with the parental strain (De Bernardis et al., 1999).

A temporal progression of SAP gene expression, in the order SAP1 and SAP3 > SAP6 > SAP2 and SAP8, has been previously described in a model of oral candidosis based on reconstituted human epithelium (RHE). The expression of SAP1-3 and SAP6 was also detected in samples from patients with oral candidosis (Schaller et al., 1998; 1999). A possible role of SAP1-3 during oral infection may be the attachment to the epithelial surface, as a reduced adherence of the single mutants (Δsap1,Δsap2,Δsap3) to buccal epithelial cells was observed, whereas Δsap4-6 was significantly increased in adherence (Watts et al., 1998). However, the virulence phenotypes of these proteinase-deficient mutants during experimental oral infections remains to be investigated.

To clarify the relevance of Saps for oral infections, we investigated the consequence of proteinase inhibitors and the virulence phenotype of sap null mutants in the established model of oral candidosis (Schaller et al., 1998). Histological alterations and SAP gene expression of the mutants in comparison with that of the parental strain were investigated by reverse transcription polymerase chain reaction (RT-PCR) during infection. The ultrastructural localization of distinct groups of Sap antigens was examined in yeast and hyphal cells during different stages of interaction with the RHE by immunoelectron microscopy using polyclonal antibodies directed against Sap1-3 or Sap4-6 (Borg-von Zepelin et al., 1998). Here, we show that Sap1-3, but not Sap4-6, contribute to tissue damage in this model of oral candidosis.

Results

Correlation between SAP gene expression and histological evaluations

Experiments to control the reproducibility of the RT-PCR method and to verify the absence of genomic DNA have been described previously (Schaller et al., 1998).

In order to examine the reproducibility of the results, histological evaluations and RT-PCR investigations of the infected epithelia were carried out after 0, 12 and 36 or 48 h in two replicate experiments. Incubation periods of the pepstatin A experiments were 12, 36 and 48 h. The C. albicans strains used were the clinical isolate SC5314 (Gillum et al., 1984), Δsap1, Δsap2, Δsap3 (Hube et al., 1997), Δsap4-6 (Sanglard et al., 1997) and the three isogenic Δsap1/3 double mutants described in this study. Each set of infection experiments included infection with SC5314 as a control. In a previous study, we demonstrated a distinct SAP expression pattern during the course of infection in this in vitro model of mucosal infections (Schaller et al., 1998).

In this study, we confirmed a correlation of the SAP gene expression with epithelial lesions. These histological changes of the RHE were evaluated on the basis of 50 sections from five different sites for each infected epithelium.

Growth ofΔsap mutants and SC5314 after the application of pepstatin A

To examine whether gene disruption or the application of pepstatin A affected growth rate during infection, growth was measured by counting cells 12, 36 and 48 h after the infection of the RHE. After the application of pepstatin A, growth rates for all Δsap mutants and the parental strain were identical to that of the parental strain SC5314.

Morphology of uninfected and infected reconstituted human epithelium (RHE) in the presence or absence of pepstatin A

The RHE consisted of stratified keratinocytes without stratum corneum. The morphology of uninfected epithelial tissue has been described previously (Fig. 1A) (Schaller et al., 1998). The correlation between the expression of distinct SAP genes and the onset of morphological alterations described therein (Schaller et al., 1998) has been confirmed. This, however, is not direct proof that the lesions are caused by Sap activity as other virulence factors might be induced at the same time to result in the aforementioned histological alterations. However, reduced histological damage after the addition of aspartic protease inhibitors, such as pepstatin A, would indicate that Sap activity contributes to the virulence in this in vitro model of oral candidosis.

Figure 1.

. Light micrographs of untreated reconstituted human epithelium (RHE) before infection with C. albicans. A. Stratified keratinocytes without stratum corneum (bar = 20 μm). B. RHE 12 h after inoculation with C. albicans cells (SC5314). Adherence and invasion of Candida cells, subepidermal split formation and vacuolation of the epithelium and severe oedema and detachment (acantholysis) of the cell layers has taken place (bar = 20 μm). C. RHE 12 h after inoculation with C. albicans cells (SC5314) in the presence of 15 μm pepstatin A. Adhesion of a few Candida cells to the superficial keratinocytes is seen, but no marked morphological alterations are visible (bar = 20 μm). Similar results were obtained with 10 μm pepstatin. D. RHE 36 h after infection with SC5314, showing mucosal erosion with severe acantholysis, oedema and enlarged intercellular spaces of all keratinocyte layers. Only a few Candida cells are adhering to the superficial keratinocytes as the pseudomembranous layer has been separated from the underlying RHE (bar = 20 μm). E. RHE 36 h after infection with C. albicansΔsap3 cells shows multiple yeast and hyphal cells. Vacuolation and oedema are only seen within the three uppermost epithelial layers (bar = 20 μm). Similar results were observed with Δsap1 and Δsap2. F. RHE 36 h after infection with C. albicans Δsap4-6 cells. Penetration of the epithelium by hyphal cells. Severe mucosal erosion with vacuolation, oedema and enlarged intercellular spaces within all keratinocyte layers (bar = 20 μm).

Therefore, we analysed the effect of this inhibitor during RHE infection with the parental strain. The infection of untreated RHE with SC5314 resulted in marked epithelial lesions, some with subepidermal splitting (Fig. 1B). In addition, oedema, vacuoles and epithelial detachment were observed. Because C. albicans penetrated the epithelial cell layers, several fungal cells were found below the basal epithelial cells. These morphological alterations were strongly decreased when pepstatin A was added at concentrations of 10 μM and 15 μM (Fig. 1C). Oedema and detachment were observed mainly in the uppermost keratinocyte layer. Vacuolation was strongly reduced and subepidermal split formation was completely blocked by pepstatin A. Differences in the histological alterations with and without pepstatin A were more significant after 12 h compared with 36 h and 48 h. Incubation of uninfected reconstituted epithelium with pepstatin A at concentrations of 10 μM and 15 μM demonstrated no histological alterations.

Morphology of RHE after infection with C. albicans SAP null mutants

The findings of the pepstatin A experiments imply that Saps may be involved in colonization and infection in the in vitro model. In order to elucidate the contribution of single Sap isoenzymes, we tested the SAP null mutants Δsap1, Δsap2, Δsap3 (Hube et al., 1997) and the triple mutant Δsap4-6 (Sanglard et al., 1997) in the in vitro model of oral candidosis. For each time course, a control infection with the parental strain was carried out. Twelve hours after infection with SC5314, slight intracellular oedema of the uppermost keratinocytes was seen. After 36 h, structural alterations of the epithelium increased and keratinocytes were enlarged and showed severe oedema and detachment (acantholysis) (Fig. 1D). Sometimes this layer, which consists of necrotic keratinocytes and C. albicans yeast and hyphal cells, separated from the mucosal equivalent during the fixation and embedding process.

After 36 h infection of RHE with the Δsap1, Δsap2, and Δsap3 mutants, intracellular oedema and detachment was seen only in the two uppermost keratinocyte layers (Fig. 1E). Surprisingly, histological damage 36 h after infection with the Δsap4-6 triple mutant seemed to be increased compared with the parental strain (Fig. 1D and F). A less significantly attenuated virulence of the single mutants and an enhanced virulence of the triple mutant compared with SC5314 was observed after 12 h.

SAP gene expression by C. albicans wild type and mutant strains during the course of infection

During the first 36 h after infection with the parental strain, a temporal pattern of gene expression was observed. SAP1 and SAP3 were expressed after 12 h and SAP2, SAP6 and SAP8 occurred by 36 h (Table 1). This expression pattern was identical to that published previously (Schaller et al., 1998). Using RT-PCR, we then examined gene expression in Δsap1, Δsap2, Δsap3 and Δsap4-6 backgrounds 12 h and 36 h after infection. In the Δsap1 mutant SAP2, SAP3 and SAP8 transcripts were detected after 12 h. Transcripts of SAP6 were expressed after 36 h (Table 1). This SAP expression pattern of the Δsap1 mutant suggested compensatory SAP2 and SAP8 upregulation after deletion of SAP1. This upregulation was confirmed in further infection experiments with Δsap1, which repeatedly demonstrated the temporal expression pattern of SAP2, SAP3 and SAP8 > SAP6 transcripts. Except for the absence of the transcripts of the disrupted proteinase genes, infection experiments with Δsap2, Δsap3 and Δsap4-6 mutants revealed no different SAP expression pattern compared with that of SC5314 (Table 1).

Table 1. . SAP gene regulation of C. albicans wild-type and sap mutant strains during RHE infection. SAP genes expressed 12, 36 and 48 h after infection of RHE, as obtained by RT-PCR, are shown. For Δsap1/3 mutants, we observed early expression of SAP8 or, in one case, early expression of SAP2, SAP5 and SAP8. ND, not determined.Thumbnail image of

Ultrastructural localization of the Sap antigen

As the results of this study demonstrated attenuated virulence for Δsap1, Δsap2 and Δsap3 mutants but not for the Δsap4-6 triple mutant, we concluded that Sap1-3 may be important for this type of experimental candidosis. For clarification of the ultrastructural localization of different Saps in the in vitro model, we used two previously described polyclonal specific antibodies directed against Sap1-3 or Sap4-6 (Borg-von Zepelin et al., 1998). For the detection of Sap immunoreactivity within C. albicans cells or keratinocytes, post-embedding labelling with Sap antibodies conjugated to 10 nm gold particles was carried out 12, 36 and 48 h after infection of RHE with SC5314. Twelve hours after infection, yeast cells were the predominant infectious form. At the same time SAP1 and SAP3 transcripts could be demonstrated by RT-PCR (Table 1).

Labelling with the Sap1-3 antibody demonstrated the intensive density of Sap immunoreactivity in all C. albicans cells, including those found in the pseudomembrane, within epithelial cells and in the intercellular spaces between the epithelial cells. When gold particles were counted in 10 randomly selected cells, we found an average of 416 ± 22 (standard deviation) intracellular and 16 ± 9 extracellular particles. Sap immunoreactivity of adherent yeast cells was localized predominantly at the site of interaction between Candida and epithelial cells (Fig. 2). Labelling with the Sap antibody directed against Sap4-6 revealed no or very few gold particles (14 ± 8 intracellular and 9 ± 6 extracellular particles average in 10 cells), which is in agreement with the lack of SAP4-6 transcripts at this time (Table 1). At 36 h and 48 h after infection of the wild-type strain, expression of SAP1, SAP2, SAP3, SAP6 and SAP8 was detected by RT-PCR (Table 1), and hyphal cells were common. These often penetrated keratinocytes or were found within epithelial cells. All the yeast (Fig. 2A) and hyphal cells (Fig. 2B) of the pseudomembrane, or within the RHE, showed a homogenous distribution of Sap1-3 in the cell wall. At the stage of adhesion or penetration, gold particles were mainly observed in close contact with Candida and epithelial cells (Fig. 2C). Few Sap1-3 proteins were detected within the cytoplasm of the fungus, whereas larger amounts of these proteins were observed at the endoplasmatic reticulum (Fig. 2A). In contrast, only hyphal cells within the RHE were labelled with gold particles after incubation with the Sap4-6 antibody (Fig. 2D), whereas all other C. albicans yeast and hyphal elements of the pseudomembrane or in the stage of adhesion (Fig. 2E) were unlabelled. Control experiments without the addition of the polyclonal antibodies Sap1-3 or Sap4-6 showed no non-specific gold labelling 12 h and 36 h after infection with SC5314.

Figure 2.

. Electron microscopy with post-embedding immunogold-labelling, using 10 nm gold particles in a sample of RHE taken 48 h after infection with SC5314. Large amounts of Sap1-3 antigen are observed within the cell wall of a C. albicans yeast (A) and two hyphal cells (B) found in the pseudomembrane. Note the labelling at the endoplasmatic reticulum (arrows) in the cytoplasm (A). Enhanced deposition of Sap1-3 antigen was discovered within the cell wall of C. albicans yeast cells in direct contact with superficial epithelial cells (C). In contrast, very few gold particles were found within the fungal cell wall of yeast cells at the stage of epithelial adhesion after labelling with the Sap4-6 specific antibody (E). Only hyphal cells within the RHE were labelled with the Sap4-6-specific antibody during infection (D). In each case the scale bar represents 0.5 μm.

Construction ofΔsap1/3 mutants

Development of initial lesions on the in vitro model was accompanied by the expression of SAP1 and SAP3 (Schaller et al., 1998). Attenuated virulence of both the Δsap1 and Δsap3 compared with SC5314 was observed in the in vitro model of oral candidosis. This suggests that Sap1 and Sap3 may play an important role in the early stage of experimental oral candidosis. Further evidence could be expected by testing a Δsap1/3 double mutant in the in vitro model. Consequently, we disrupted the SAP3 gene in the Ura strain BH24-15-1-3 (Hube et al., 1997), using the URA-blaster protocol (Fonzi and Irwin, 1993). A SAP3:hisG::URA3::hisG::SAP3 cassette (Hube et al., 1997) was used to disrupt the two copies of the SAP3 gene by integrative transformation. SAP3 disruption in both alleles of BH24-15-1-3 was confirmed by PCR (Fig. 3) and Southern analysis (Fig. 4). After disruption of the first allele, the ura3 genetic marker was regenerated by treatment with 5-fluoro-orotic acid (5-FOA) (Hube et al., 1997; Sanglard et al., 1997). Three isogenic strains of the Δsap1/3 double mutant were created (Table 2). As demonstrated in Fig. 3, single and double fragments of the expected lengths (a 1.4 kb fragment of the wild-type gene and a 1.9 kb fragment of the hisG-disrupted allele after FOA treatment) were detected by PCR analysis during the disruption procedure. The inserted hisG::URA3::hisG-cassette of the transformants was too large to be amplified under normal PCR conditions. For Southern blot analysis, DNA was digested with EcoRI, separated by agarose gel electrophoresis and transferred to a nylon membrane. Hybridization to a SAP3 probe showed a 1.0 kb band for the wild-type genes, a 1.7 kb band for the FOA-resistant segregants and a 3.0 kb band for the Ura+ transformants (Fig. 4). Each of these isogenic mutants had an identical pattern when tested by PCR and Southern blot analysis, and showed the same growth rates when tested in complex and defined growth media (data not shown). For infection experiments Δsap::hisG/Δsap::hisG::URA3::hisG mutants Ura+ were used (Table 2).

Figure 3.

. Disruption of SAP3 demonstrated by PCR analysis of genomic DNA. SAP3 specific primers amplified a 1.4 kb fragment of the wild-type gene (lane 2). Heterozygote mutant after first round of transformation (lane 3). Ura heterozygote mutant after FOA selection. In addition to the 1.4 kb fragment, a 1.9 kb fragment was amplified (lane 4). Δsap1/3 null mutant after second round of transformation (lane 5). Ura homozygote mutant after second FOA selection (lane 6). Molecular mass marker pBR322 DNA/MvaI (M) (MBI Fermentas) with fragments of 1857, 1058 and 929 bp are shown in lane 1.

Figure 4.

. Disruption of SAP3 demonstrated by Southern analysis. Genomic DNA was restricted with EcoRI and hybridized with a SAP3 probe. The restriction pattern of the wild type (lane 2), a first round transformant (lane 3), a Ura heterozygote mutant after first FOA selection (lane 4), a second-round transformant (lane 5) and a Ura homozygote mutant after second FOA selection (lane 6) is shown. A Dig-labelled Molecular Weight Marker VII (Boehringer Mannheim) was used in lane 1.

Table 2. . C. albicans strains used in this investigation and genotypes of Δsap mutants produced in this study. Three isogenic strains of the Δsap1/3 double mutant were produced (Gillum et al., 1984; Hube et al., 1997).Thumbnail image of

Morphology of RHE and SAP expression after infection with theΔsap1/3 double mutants

For comparison, three isogenic Δsap1/3 double mutants were tested on the RHE as well as the parental strain SC5314. In this set of infection experiments, severe oedema and mucosal erosion of the RHE with enlarged intercellular spaces of the keratinocytes was observed 12 h after infection with SC5314. The structural alterations of RHE and the SAP gene expression pattern (SAP1, SAP2, SAP3, SAP6 and SAP8) were identical to that found 36 h after infection with SC5314 in the former infection experiment (Fig. 1D). Twelve hours after the infection of the RHE with the Δsap1/3 double mutants, only keratinocytes of the uppermost epithelial layer showed slight intracellular oedema and detachment (Fig. 5). This attenuated virulence of the double mutant as compared with the wild-type infection, was also significant after 36 h.

Figure 5.

. Appearance of RHE 12 h after infection with C. albicans cells (Δsap1/3). Multiple yeast or hyphal cells causing vacuolation and oedema are only seen within the uppermost epithelial layer. Infection of RHE with wild-type cells, 12 h after inoculation, is shown in Fig. 1A (bar = 20 μm).

In contrast to infection with SC5314, SAP8 was the first SAP gene expressed by the Δsap1/3 double mutant (Fig. 6A). Thirty-six hours later SAP2, SAP5 and SAP6 expression was detected (Fig. 6B). Further studies of SAP gene expression at the initial stages of infection revealed SAP8 as the first transcript in infections caused by the double mutant. In an additional experiment, expression of SAP2, SAP5 and SAP8 was observed after 12 h. This upregulation of SAP8 and SAP2 and the expression of SAP5 was confirmed in duplicate infection experiments with all three isogenic Δsap1/3 double mutants.

Figure 6.

. Analysis of RT-PCR products of RNA (lanes 2–8) from samples taken 12 h (A) and 48 h (B) after infection with Δsap1/3. A 526 bp fragment obtained by amplification with EFB1 primers demonstrates the cDNA origin of the template (lane 9). Note the shift in the 891 bp fragment after PCR amplification of intron-containing genomic DNA with the same primer pair (lane 10). In lane 1 the molecular mass marker pBR322 DNA/MvaI (M) (MBI Fermentas) was used giving fragments of 1857, 1058, 929 and 383 bp in size.

Discussion

Proteolytic activity due to Saps seems to be an important virulence factor for C. albicans infections (Rüchel, 1992; Goldman et al., 1995; Hoegl et al., 1996; Hube, 1996; 1998) and has been implicated in both systemic and local infections (MacDonald and Odds, 1983; Kwon-Chung et al., 1985; Ghannoum and Abu Elteen, 1986; Borg-von Zepelin and Rüchel, 1988; Ray and Payne, 1988; Homma et al., 1991; Ollert et al., 1993; De Bernardis et al., 1995; Fallon et al., 1997). This role was confirmed by studies that demonstrated a protective effect by the use of aspartic proteinase inhibitors, such as pepstatin A, or proteinase-specific antibodies during infection (Borg-von Zepelin and Rüchel, 1988; Ray and Payne, 1988; Ollert et al., 1993; De Bernardis et al., 1997; Fallon et al., 1997). Molecular tools, such as targeted gene disruption and the use of expression vectors, have enabled the construction of distinct SAP null mutants or strains constitutively overexpressing SAP genes (Hube et al., 1997; Sanglard et al., 1997; Dubois et al., 1998). Testing the virulence phenotype of these SAP mutants in specific in vitro or animal models has led to a better understanding of the relevance of Saps in pathogenicity during different types of candidosis (Hube et al., 1997; Sanglard et al., 1997; Borg-von Zepelin et al., 1998; Dubois et al., 1998; Ibrahim et al., 1998; De Bernardis et al., 1999; M. Kretschmar, B. Hube, D. Sangland, M. Schröder et al., submitted). These virulence studies were mainly focused on animal models for systemic or vaginal candidosis. To clarify the relevance of distinct Saps for oral infections, we tested the above mentioned mutants (Δsap1, Δsap2, Δsap3, Δsap4-6) in an established in vitro model for oral candidosis based on RHE (Schaller et al., 1998).

The usefulness of the in vitro model for mimicking oral candidosis in man and for studying the expression of secreted aspartic proteinases using RT-PCR and immunoelectron microscopy (Schaller et al., 1998) was confirmed by the results of the present study. However, first lesions were not always observed at the same time points. There appeared to be certain factors that influence the transition from growth without causing damage and the aggressive stages of growth. For example, these factors may be secreted components of the epithelial cells, such as antimicrobial peptides (Harder et al., 1997). In fact, the β-defensin hBD-2 secreted by human skin is known to act effectively against C. albicans. Using RT-PCR, HBD-2 transcripts were found in our in vitro model during infection (S. Malz, W. Schäfer and B. Hube, unpublished).

Infection of RHE with the wild-type strain SC5314 showed a strong correlation between the onset of epithelial lesions and the pattern of SAP gene expression, and served as a control. Invasion of the RHE occurred mainly by penetration through the intercellular spaces, thus implying that the penetration procedure is associated with the detachment of intercellular structures such as desmosomes. In contrast to our previous investigations, direct penetration of keratinocytes was observed in the present study. This kind of epithelial invasion was also seen in vivo as demonstrated by electron microscopy studies of samples from patients with oral candidosis (Farrell et al., 1983; Reichart et al., 1995). These results suggest that intercellular and intracellular epithelial proteins are potentially targets for the proteolytic activity of Saps.

Immunoelectron microscopy studies using two specific polyclonal antibodies directed against Sap1-3 or Sap4-6 confirmed the results of our gene expression studies. Corresponding to the expression of SAP1 and SAP3 12 h after infection with SC5314, C. albicans cells showed an intensive Sap1-3 but no Sap4-6 labelling within the cell wall. The transport of all Sap proteinases via the secretory pathway from the endoplasmatic reticulum to the plasma membrane and secretion through the cell wall was reflected in the pattern of intracellular localization. The ultrastructural localization of Sap immunoreactivity within the cell wall was also demonstrated previously during experimental rat vaginitis (Stringaro et al., 1997). In our study, hyphal and yeast cells demonstrated different patterns of Sap1-3 and Sap4-6 localization after 36 h and 48 h. For example, Sap1-3 isoenzymes were present in all C. albicans yeast and hyphal cells of the pseudomembrane and in those in direct contact with the RHE, which may indicate a role in adhesion and/or invasion. In contrast, Sap4-6 labelling was only evident in the cell wall of yeast and hyphal cells within the RHE, but not on the epithelial surface or in the pseudomembrane. Evidence for the contribution of Saps to tissue damage in this in vitro model of oral candidosis was provided by the strong attenuation of the histological pathology after treatment with the specific aspartic proteinase inhibitor, pepstatin A. However, epithelial damage was not completely blocked by pepstatin A. This suggests that Sap activity was not eliminated, or that other factors could also cause tissue damage. Incomplete in vitro inhibition of Saps by pepstatin A has been demonstrated previously (Ollert et al., 1993; Colina et al., 1996; Korting et al., 1999), which may explain the epithelial damage observed at later incubation periods (36 h and 48 h) in the presence of pepstatin A. Evidence for the relevance of Sap1-3 for the development of the mucosal erosion was provided by the attenuated virulence phenotype of the single mutant strains in comparison with the parental strain infection. Surprisingly, the histological damage of the triple mutant, Δsap4-6, seemed to be increased. This may be explained by an enhanced attachment to the epithelial cells, because the adherence of these SAP null mutants on buccal cells as an initial stage for pathogenesis of oral candidosis was significantly higher compared with SC5314 (Watts et al., 1998; M. Borg-von Zepelin, I. Meyer, D. Sanglard and M. Monod, unpublished results). For each of the single Δsap1, Δsap2 and Δsap3 mutants, a moderate attenuation of adherence was described (Watts et al., 1998), indicating that SAP1-3 may contribute in a cumulative way to adherence to buccal epithelial cells. One possible reason for the enhanced adherence of the triple mutant Δsap4-6 in the in vitro assay may be a compensatory upregulation of SAP1-3 caused by the deletion of SAP4-6 (M. Borg-von Zepelin, I. Meyer, D. Sanglard and M. Monod, unpublished results).

Similarly, De Bernardis et al. (1999) showed that the virulence of all the single sap1-3 mutants, but not of the triple mutant Δsap4-6, was significantly attenuated in experimental vaginitis. However, in contrast to our results, which demonstrated a similar reduction of virulence for all three mutants, Δsap2 was almost avirulent in the rat vaginitis model (De Bernardis et al., 1999). The important role of Sap2 during experimental vaginitis at acidic pH values might be explained by the large amounts of this proteinase and its optimal proteolytic activity at low pH (Hube et al., 1994). The pH optimum for Sap2 and Sap3 was 3.5, whereas Sap1 was optimally active at pH 4.5, and Sap4-6 at pH 5.0 (Borg-von Zepelin et al., 1998).

In the neutral milieu of oral candidosis, the progressive transcription of SAP1, SAP3 and finally SAP2 correlated with the progression of diseases. As SAP1-3 expression was shown to be expressed in several laboratory and clinical strains (Hube et al., 1994), we suggest that the observed expression pattern is not strain dependent. Evidence for the relevance of Sap1-3 was demonstrated by the attenuated virulence phenotype of Δsap1-3 mutants. Expression of SAP1 and SAP3 was also observed in samples from patients suffering from oral candidosis, suggesting a similar function in the in vitro model and in vivo (Schaller et al., 1998). A correlation of SAP1 and SAP3 with oral candidosis was also suggested by Naglik et al. (1999), who demonstrated the expression of these two SAP genes in all samples from patients suffering from clinical signs of oral candidosis, but not in samples from healthy C. albicans carriers. We therefore decided to delete both these SAP genes, which seem to be the most relevant SAP genes, for the initial steps of epithelial tissue damage. Consequently, we constructed Δsap1/3 double mutants. Three independent isogenic double mutants were tested to rule out any unforeseen genetic changes that affect the strain phenotype. Infection of the RHE with each of the double mutants resulted in a stronger reduction of the histological damage compared with the infections with each of the single mutants. Thus SAP1 and SAP3 apparently both contribute to the initial virulence phenotype of the parental strain infection in mucosal infections. The remaining epithelial lesions after infection with the single and the double mutants may be due to the activity of other potential hydrolytic enzymes, for example the lipases and phospholipases secreted by C. albicans or other Saps. Hence, we investigated the expression pattern of the remaining SAP genes. Interestingly, we found an initial upregulation of SAP8, followed by earlier expression of SAP2 relative to the normal expression pattern of the wild type. In addition, SAP5 expression occurred, which had never been observed in previous studies. This is the first time that upregulation of C. albicans genes due to targeted gene disruption has been demonstrated, indicating that knock-out experiments, especially when gene families are involved, should be carefully interpreted. Furthermore, it suggests that C. albicans may be able to compensate for the loss of important genes by upregulation of alternative genes.

In summary, these results illustrate that several SAP genes contribute to epithelial lesions in the in vitro model for oral candidosis. The expression hierarchy during a course of infection seems to be established. Also the relevance of these genes to epithelial tissue damage was shown to differ. We found that Sap1-3, but not Sap4-6, are important in this model. It remains to be tested whether this is also the case during in vivo infections. Investigations to prove the relevance of Sap1-3 during in vivo oral infections are currently in progress.

Experimental procedures

Candida strains

The clinical Candida albicans wild-type strain SC5314 (Gillum et al., 1984) and the SAP null mutant strains Δsap1, Δsap2 and Δsap3 (Hube et al., 1997) and Δsap4-6 (Sanglard et al., 1997) were used in the study. For gene disruptions, the UraΔsap1 strain of C. albicans (BH24-15-1-3) and a disruption cassette of SAP3 (pAS3-URA3) was used (Hube et al., 1997). Disruption of the SAP3 gene followed the conventional URA-blaster protocol (Fonzi and Irwin, 1993).

Culture media and growth conditions

YPG (1% yeast extract, 2% peptone and 2% glucose) complex medium was used for disruption experiments. Transformants generated using the URA-blaster protocol, were grown in synthetic dextrose (SD) medium, which contained 0.17% yeast nitrogen base (Difco) and 2% glucose.

For the infection of the reconstituted epithelium, inocula were prepared by culturing yeast cells for 24 h at 37°C on Sabouraud–dextrose agar (Difco). A sample of the culture was washed three times in 0.9% NaCl and cells were counted. A sample of ≈ 2 × 105 cells were then suspended in 10 ml of YPG medium. The suspension was cultured for 16 h at 25°C through orbital shaking and cells were again counted. A suspension of 4 × 106 cells was again incubated through shaking in fresh medium for 24 h at 37°C. After washing three times with phosphate-buffered saline (PBS), the final inoculum of these semisynchronized cells was then adjusted to the desired density with PBS solution.

Candida transformation and gene disruption

Transformation of the Ura auxotroph Δsap1 C. albicans strain was performed with pAS3-URA3 as described by Hube et al. (1997). For the preparation of protoplasts late log phase, C. albicans cells were pelleted and resuspended in 5 ml of 1 M sorbitol (Sigma) containing 50 mM dithiothreitol (Sigma) and 25 mM EDTA (Sigma) (pH 8.0) for 5 min at 30°C. Pelleted cells were resuspended in 5 ml of 0.1 M sodium citrate (Sigma) (pH 5.8), 1 M sorbitol and 25 mM EDTA, to which 0.05 ml of β-glucuronidase (Sigma) was added. After incubation for 5 min at 30°C, spheroplasts were resuspended in 0.5 ml of: 1 M sorbitol,10 mM CaCl2, 10 mM Tris-Cl (pH 7.5), and used for transformation. Plasmid DNA (10 μg) containing the SAP3 disruption cassette was linearized. Ten microlitres of carrier DNA (3 mg ml−1) (herring sperm DNA; Clontech) was added and incubated with 100 μl of the protoplasts preparation for 15 min at room temperature. Ten volumes of: polyethylene glycol 4000 (20% w/v), 10 mM CaCl2, 10 mM Tris-Cl, pH 7.5, were added and the cells were incubated for 15 min at room temperature. After resuspensation in 100 μl of SD medium the protoplasts were plated on SD agar supplemented with 1 M sorbitol. Colonies having the desired disruption were then grown in the presence of 5-fluoro-orotic acid (FOA) to select for cells where the his repeats recombined, deleting the URA3 gene in the process. These strains were then transformed with the same disruption cassette for the deletion of the second allele. Three isogenic Δsap1/3 mutants were constructed. Ura+ transformants and FOA-resistant segregants of these transformants were analysed by PCR and Southern analysis. For PCR screening, primers were designed to amplify those parts of the gene that were removed during gene disruption. Thus, amplified DNA fragments changed in size after disruption of the wild-type copy.

Southern analyses

C. albicans genomic DNA isolations were carried out as has been described previously (Magee, 1994). After restriction digestion, DNA was size-fractionated by agarose gel electrophoresis and vacuum-blotted onto a membrane. PCR was used to prepare a hybridization probe for Southern blotting. A 1.0 kb fragment containing 0.8 kb of the open reading frame (ORF) of SAP3 was amplified, using 5′-TGGATTGG AACATTTCTAATTC-3′ and 5′-GTTATAATCACTAGTTCC -3′ primers and Dig-labelled-dUTP (Boehringer Mannheim). Prehybridization and hybridization of the membrane with labelled DNA probe were performed at 42°C in a solution containing 40% formamide, 0.1% N-laurylsarcosin, 5× SSC, 0.2% SDS, 100 μg ml−1 of denaturated, degraded herring sperm DNA and 2% (w/v) blocking agent (Boehringer Mannheim). Southern blots were hybridized under high-stringency conditions as described by Hube et al. (1997). Detection of hybridized nucleic acids with Dig-antibodies and chemiluminescent reagent was performed according to the manufacturer's instructions (Boehringer Mannheim).

Reconstituted human epithelium (RHE)

The human epithelium for the in vitro model of oral candidosis was supplied by SkinethicTM Laboratory. It was obtained by culturing transformed human keratinocytes derived from squamous cell carcinoma of the buccal mucosa, of the cell line TR146 on an inert supporting membrane (Rupniak et al., 1985). Cultures were incubated in serum-free conditions in a defined medium based on the MCDB-153 medium (Clonetics), containing 5 μg ml−1 of insulin, on a 0.5-cm2 microporous polycarbonate filter for 7 days. The in vitro model and all culture media were prepared without antibiotics and antimycotics.

For the inhibition of Saps, pepstatin A (Sigma) was dissolved in absolute methanol and administered to 50 μl of PBS, containing 2 × 106C. albicans yeast cells of the SC5314 strain at final concentrations of 10 and 15 μM. Controls contained 50 μl of PBS with 10 and 15 μM pepstatin A alone.

Two replicate infection experiments were performed for each C. albicans strain. RHE was infected with 2 × 106C. albicans yeast cells of the SC5314, Δsap1, Δsap2, Δsap3, Δsap4-6 strains and three isogenic strains of the described Δsap1/3 double mutant in 50 μl of PBS. Controls contained 50 μl of PBS alone. All tissue cultures were incubated at 37°C with 5% CO2 at 100% humidity for 12 h, 36 h and/or 48 h. For investigation of the SAP expression pattern, two additional infection experiments for Δsap1 were performed with incubation periods of 6 h, 12 h, 24 h and 48 h. The medium was changed every 24 h.

The growth rates of the different sap mutants and SC5314 with and without application of pepstatin A were tested by cell counting 12 h and 36 h after infection of RHE.

Light microscopy

A part of each specimen was fixed with 2.5% glutaraldehyde and 2.0% formaldehyde in a 0.05 M phosphate-buffered solution at pH 7.3 (Karnovsky, 1965). Specimens were post-fixed in Dalton's chrome–osmium fixative at room temperature (Dalton, 1955). After several washing and dehydration procedures for 1 h at room temperature, the specimens were embedded in glycide ether. The small blocks of tissue were cut, using an ultramicrotome (Ultracut). Semi-thin sections (1 μm) were studied with a light microscope after staining with 1% toluidine blue and 1% pyronine G (Merck). The sections were viewed at a magnification of 400×.

Immunoelectron microscopy

Post-embedding immunogold labelling was carried out for the intracellular detection of Sap antigen in SC5314 infected samples taken after 12, 36 and 48 h. A part of each specimen was fixed in Karnovsky solution for 1 h at room temperature and embedded in LR-White or glycide ether. Sections, 80–100 nm thick, were mounted on nickel grids. The grids were rinsed on drops of dH2O for 10 min, and floated on drops of PBS containing 5% (v/v) normal goat serum for 2 × 10 min. Grids were then incubated with the anti-Sap polyclonal rabbit antibodies (Borg-von Zepelin et al., 1998), directed against Sap1-3 or against Sap4-6 (Borg-von Zepelin et al., 1998), diluted 1:100 in PBS supplemented with 1% ovoalbumin, 0.1% Tween 20 and 0.015 M sodium azide (PBS-OT) for 3 h at room temperature. After washing overnight with PBS, grids were incubated with 10 nm of gold-conjugated goat anti-rabbit IgG (Auroprobe EM Immunogold reagents, Amersham) diluted 1:50 in TBS-OT (0.02 M Tris hydrochloride acid buffer, 0.15 M NaCl, 0.015 M sodium azide, 1% ovoalbumin, 0.1% Tween 20, adjusted to pH 8.2) for 1.5 h at room temperature. In control samples the anti-Sap polyclonal antibodies were omitted. After several washing steps with TBS-OT, grids were fixed with 2% glutaraldehyde in PBS and washed again in aqua dest. Grids were stained with 0.5% uranyl acetate for 10 min, and 2.7% lead citrate for 5 min (Ultrastainer) at 20°C. Grids were examined using a Zeiss EM 902 transmission electron microscope (Zeiss) operating at 80 kV, at magnifications of between 3000× and 85 000×. To back up the data statistically, we counted gold particles for each treatment in 10 cells. Gold particles within the cell or the cell wall were labelled ‘intracellular’, peripheral gold particles within a distance of 1 μm of the cell wall were labelled ‘extracellular’.

RNA isolation

Total RNA from the infected epithelia was prepared as described by Hube et al. (1994). Frozen cells were resuspended in 0.5 ml of extraction buffer (0.1 M Tris-HCl pH 7.5, 0.1 M LiCl, 0.01 M dithiothreitol), containing 0.5 g of glass beads (425–600 microns), 17 μl of 20% SDS, 0.5 ml of phenol–chloroform (50:50, saturated in 1 M Tris-Cl, pH 7.5) and vortexed for 10 min to break the cells. After being spun for 3 min at 3000 × g, the aqueous phase was removed and extracted twice in phenol–chloroform. The RNA was precipitated with 2 volumes of 96% ethanol and stored at −20°C. Precipitated RNA was pelleted and resuspended in 4 μl of 10 mM Tris-HCl pH 7.5, 1 mM EDTA and 1 μl of RNase inhibitor (Gibco BRL, Life Technologies).

cDNA-synthesis (RT)

After DNase I (Gibco BRL, Life Technologies) treatment according to the manufacturer's instructions, annealing was done with Oligo (dT) primer (Gibco BRL, Life Technologies) at 70°C for 10 min, and the product was chilled on ice. The reaction mixture was made up to a final volume of 20 μl, containing RT buffer [50 mM Tris (pH 8.3), 75 mM KCl, 3 mM MgCl2], 10 mM dithithreitol, 10 μM desoxynucleoside triphosphate (dNTP) and 200 U of Superscript II reverse transcriptase (Gibco BRL, Life Technologies). cDNA synthesis was carried out at 42°C for 1 h. The reaction was then stopped by heating for 15 min at 70°C. Finally, RNase H (Gibco BRL, Life Technologies) treatment was carried out for 20 min at 37°C.

Primers

For detection of gene expression, the following pairs of primers were used:

5′-AGGGAAAGGTATTTACACT-3′ and 5′-CAGTTTCAATT CAGCTTGG-3′ or 5′-GATTTGCTTACATAGTAAGTAC-3′ for SAP1; 5′-TGATTGTCAAGTCACTTATAGT-3′ or 5′-CC TAAAGCATTCCCAGGTTAC-3′ and 5′-CTTAGGTCAAGG CAGAAATAC TG-3′ for SAP2; 5′-TGGATTGGAACATTTC TAA TTC-3′ and 5′-CAATCTCCA GAGGAGT ACTTCC-3′ for SAP3; 5′-TTTTCATTAACAACCAACCATTC-3′ and 5′-GTCCTGGTGGCTTCGTTGC-3′ for SAP4; 5′-ATAATTAA TCTAAAGTCAAAGTTC-3′ and 5′-CAATCTCCAGAGGAGT ACTTCC-3′ for SAP5; 5′-TTCTTCAAACGTTTTAATTC TCT-3′ and 5′-CATAAATGACTTCAAAATATAAAT-3′ for SAP6; and 5′-CTCTATAAAGTAGAAATACTTGA-3′ and 5′-GTTGACACAGGTTCTTCTG-3′ for SAP8. For an internal mRNA control and for detection of genomic DNA, we designed a primer pair to identify and amplify a 891 bp fragment of the EFB1 gene (elongation factor), which encodes EF-1β and contains a 365 bp intron (Maneu et al., 1996). The presence of the intron allowed the distinction of amplified signals arising from genomic and cDNA origins. Primers used for EFB1 expression were: 5′-ATTGAACGAATTCTTGGCT GAC-3′ and 5′-CATCTTCTTCA ACAGCAGCTTG-3′.

PCR

A single PCR protocol was used with all nine primer sets. The samples were subjected to 40 cycles of denaturation for 3 min at 94°C, annealing for 3 min at 50°C, and extension for 10 min at 72°C.

Acknowledgements

The authors thank E. Januschke (Ludwig-Maximilians-University, Munich, Germany) for excellent technical assistance and M. Monod (Universitaire Vaudoise, Lausanne, Switzerland) for providing the polyclonal antibodies; N. A. R. Gow (University of Aberdeen, UK) for critical reading of the manuscript. The work was supported by grants from the Theodor-Nasemann-Stipendium and the Manfred-Plempel-Stipendium to M.S. and the Deutsche Forschungsgemeinschaft (Hu 528/8–2).

Ancillary