Candida albicans is the most common oral fungal pathogen of humans, but the mechanisms by which C. albicans invades and persists within mucosal epithelium are not clear. To understand oral pathogenesis, we characterized the cellular and molecular mechanisms of epithelial–fungus interactions using reconstituted human oral epithelium (RHE). We observed that hyphal formation facilitates epithelial invasion via both active (physical penetration) and passive (induced endocytosis) processes. Genome wide transcript profiling of C. albicans experimental RHE infection was compared with that from 11 patient samples with pseudomembranous candidiasis to identify genes associated with disease development in vivo. Expression profiles reflected the morphological switch and an adaptive response to neutral pH, non-glucose carbon sources and nitrosative stress. We identified several novel infection-associated genes with unknown function. One gene, upregulated in both RHE infection and patients, named EED1, was essential for maintenance of hyphal elongation. Mutants lacking EED1 showed transient cell elongation on epithelial tissue, which enabled only superficial invasion of epithelial cells. Once inside an epithelial cell, Δeed1 cells could proliferate as yeasts or pseudohyphae but remained trapped intracellularly. Our results suggest that the adaptive response and morphology of C. albicans play specific roles for host–fungal interactions during mucosal infections.
Candida albicans is normally a harmless commensal fungus of mucosal surfaces in healthy individuals but can cause several types of infections in predisposed patients, ranging from superficial to life-threatening disease. C. albicans frequently causes oral infections in human immunodeficiency virus (HIV+)-infected patients (Challacombe and Naglik, 2006). During the most common type of oral infections, oropharyngeal candidiasis (Schaller et al., 1999a), C. albicans invades the oral mucosa and persists within the epithelium causing superficial lesions (Fidel, 2006). Usually, fungal proliferation, colonization and invasion are hindered by dense epithelial layers with high turnover rate and innate defence mechanisms such as the secretion of antimicrobial compounds. However, C. albicans has developed strategies to overcome these defence mechanisms for both commensal growth and infection (Hube, 2004). One strategy includes the morphological switch from spherical yeasts to filamentous hyphae and the virulence attributes associated with the transition (Kumamoto and Vinces, 2005a). Morphology change is a common feature of dimorphic fungi and can be triggered by fungal sensing genes in response to specific host environments (Nemecek et al., 2006). C. albicans can form yeast cells, pseudohyphae and true hyphae. True hyphal formation can be induced by diverse stimuli in vitro, such as serum, starvation, contact, temperature and pH and is regulated by several proteins including those of the cAMP or MAP kinase pathway (Kumamoto and Vinces, 2005b). In vivo, unlike for bacterial pathogens where mechanisms of epithelial entry are well characterized (Pizarro-Cerda and Cossart, 2006), the specific roles for C. albicans hyphae in epithelial invasion and persistence are unclear.
In this study, we combined a series of histological, microscopical, cellular, transcriptional and molecular approaches to investigate the interaction of C. albicans with oral epithelial tissue in vitro. An experimental oral candidiasis model based on reconstituted human epithelium (RHE) was used to dissect the different stages of fungus–cell interactions from attachment via invasion to tissue destruction. Samples from HIV+ patients who suffered from pseudomembranous oropharyngeal candidiasis were used to directly compare the genome wide transcriptional profile obtained from each of these stages with the in vivo situation and to identify genes associated with oral candidiasis. One of the genes identified as being associated with oral candidiasis in vitro and in vivo was orf19.7561 (IPF946; EED1). This gene was shown to be essential for epithelial tissue dissemination.
Cellular kinetics of fungal attachment, invasion and tissue destruction
To analyse the pathogenesis of oral C. albicans infections, we first analysed the histological, cellular and kinetic processes of an experimental oral infection with C. albicans (SC5314) based on a RHE model (Schaller et al., 1999b) over a period of 24 h. As no ‘commensal’ stage can be observed in the RHE model, we use the term ‘infection’ in this study as a process where the fungus causes an inflammatory-like host cell response and damage (Schaller et al., 2002).
Light and electron microscopy analysis allowed the experimental infection process to be dissected into three phases: an early attachment phase (1–3 h), an invasion phase (3–12 h) and a late phase (12–24 h) which was characterized by strong tissue damage (Fig. 1). In the early phase, the initial physical contact between the epithelium and fungus stimulated a switch to the hyphal form (> 98% within 1 h) and subsequent adherence to the epithelium (40% within 1 h, compared with < 3% for Saccharomyces cerevisiae). As the phenomenon of contact-dependent hypha formation could also be observed with epithelial cell lines (TR146 and HEp2) when other known stimuli of hyphal formation (CO2, cell culture medium) were excluded, we concluded that the physical contact between the fungal cells and the surface of the epithelia cells had the strongest influence on the morphological switch.
No tissue damage was observed in the early phase, as measured by the release of an epithelial marker enzyme [lactate dehydrogenase (LDH)] (Fig. 1A). Contact of hyphae, but not yeast cells, to epithelial cells in the early phase frequently caused host response mechanisms such as formation of epithelial cell protrusions surrounding the hyphae and membrane ruffling characteristic for induced endocytosis (Fig. 2A–C) (Cossart and Sansonetti, 2004; Park et al., 2005). The invasion phase was characterized by invading hyphae and correlated with moderate tissue damage in the mid phase (6 h) and severe tissue destruction in the late phase (12–24 h), with high levels of LDH release (Fig. 1A). Some hyphal cells grew along the surface of the tissue in a thigmotropic manner (Gow et al., 2002) (not shown) or invaded the tissue actively by penetrating directly through the epithelial cells, often embedded in membrane structures (Fig. 2H and I).
Molecular dissection of the experimental infection process by transcript profiling
DNA microarrays are a powerful tool for the investigation of large-scale transcriptional changes. To overcome the restriction of low fungal RNA amounts in the in vivo and in vitro samples due to low numbers of fungal cells we amplified and labelled the fungal RNA with a linear amplification system. Linear amplification systems have been shown to be reliable and highly reproducible by our and other laboratories (Wang et al., 2000; Pabon et al., 2001; Feldman et al., 2002; Hu et al., 2002; Zhao et al., 2002; Schneider et al., 2004; Wang, 2005). To identify which C. albicans genes are associated with the damage of oral tissue, the expression profile of C. albicans (SC5314) cells was examined 1, 3, 6, 12 and 24 h post inoculation of the reconstituted tissue (RHE). To further dissect individual phases of the epithelial–pathogen interaction we modified the experimental set-up by incubating C. albicans with mechanically fractured epithelial tissue for 3 and 6 h (‘disrupted RHE’ see Experimental procedures) to identify genes induced by host cell lysis.
Based on stringent filtering criteria, a reliable expression signal was detectable for ∼4300 open reading frames (75% of total gene numbers) for all time points of the experimental RHE time-course infection (reliably expressed genes). During the infection time-course 15% of the reliably expressed genes were ≥ 2-fold upregulated at one or more time points. Some genes remained either unchanged for their expression at all time points or showed a varying increase of expression during the different time points (75%).
Adaptive response of C. albicans during experimental oral candidiasis
Based on histological, cellular and kinetic observations, we predicted that the expression profile of the early phases (1–3 h) reflected the initiation of infection, and the late phases (12–24 h) the maintenance and progression of the disease process. Therefore, we focused the analysis on these phases. In the early phase (1 h), 164 genes were upregulated, of which 29 were significantly > 2-fold upregulated exclusively at this time point (Table S1A). Twenty-one of these 29 exclusively upregulated genes had no described function in C. albicans; the remaining genes with known function were correlated with cellular functions such as transcription processes.
Stringent filtering processes revealed that 38 genes had increased (≥ 1.5-fold) mRNA levels throughout the entire 24 h experiment, 21% (9 genes) had no described function in C. albicans (Table 1). Seventeen of these 38 genes were ≥ 2-fold upregulated in the early phases (1 h and 3 h). Among these genes we identified several hypha-specific (HWP1) or hypha-associated genes (FKH1, ATP2, TEF1, ALS3, SOD5), corroborating the observation that hyphae are formed immediately after epithelial contact. Other hypha-specific genes (RBT1, ECE1 and HYR1) were upregulated (≥ 2-fold) in the mid and late phases of the RHE infection. The data indicate that hyphal formation via direct contact to epithelial cells is a dominant event in the early phase (Figs 2 and 3).
Table 1. Increased expressed and upregulated genes (> 1.5-fold) at all time points of the experimental time-course infection (RHE) 1, 3, 6, 12 and 24 h.
Agglutinin like protein
Cell elongation protein
F1F0-ATPase complex, F1 beta subunit
Fatty-acyl-CoA synthase, alpha chain
Fork head protein type transcription factor
Formate dehydrogenase, 3-prime end
GAP for RHO1 by homology
High-affinity inorganic phosphate/H+ symporter
Hyphal wall protein
Involved in regulation of nuclear pre-mRNA abundance
Nucleolar shuttling protein with an RNA recognition motif
Peroxisome import protein – peroxin
Plasma membrane H+-transporting ATPase
Protein with putative transcription activation domain
Putative endocytosis and cytoskeleton protein
Putative GPI-anchored protein
Similar to S. cerevisiae Mal32p alpha-glucosidase
Similar to superoxide dismutase
Transcription elongation protein
Translation elongation factor eEF1
Weak similarity to glutenin
In the late phase (12–24 h), 78 genes were significantly upregulated at 12 and/or 24 h with as many as 28% with no described function (Fig. 1B and Table S1B). The genes with described function were mainly associated with general metabolism. Expression profiles represented adaptive response to distinct carbon sources, limited nitrogen sources and nitrosative stress. Similar to C. albicans cells exposed to macrophages (Lorenz et al., 2004) the metabolic balance point for utilizing a carbon source is shifted towards non-glucose molecules. This was indicated by the upregulation of the maltose transporter gene MAL31 (upregulated 11-fold). Furthermore, PCK1, MLS1 and ICL1, all key genes of the gluconeogenesis pathway or the glyoxylate cycle were all strongly upregulated (2- to 20-fold), suggesting that two-carbon compounds are utilized. These are possibly lipid-derived compounds, as genes associated with the β-oxidation pathway were strongly upregulated (FOX3, PXA1 and FOX2). This is consistent with the increased transcript levels of several members of the extracellular lipase gene family (e.g. LIP1), which are upregulated under lipid-rich conditions (Hube et al., 2000). Genes coding for the glyoxylate cycle and some of those involved in β-oxidation and lipolysis were also upregulated in the ‘disrupted RHE’ experiment indicating that fatty acid oxidation and lipolysis might be required for epithelial invasion when C. albicans has access to the cell content of epithelial cells (Fig. 3). To gain other essential nutrients, such as nitrogen and phosphate, C. albicans cells expressed several genes related to amino acid sensing and transport (e.g. GNP1, CAR1 and CAR2) or phosphate transport (PHO84) at increased levels.
It is generally believed that pathogens become stressed when confronted with host cells. Apart from the response to nutrient restriction, there was no evidence for oxidative, osmotic or thermal stress for C. albicans. However, we did detect a clear response to nitrosative (NO) stress, known to be part of the innate response of epithelial cells against microbes. YHB5 (2.3-fold) and SSUI (1.6-fold), marker genes involved in NO detoxification (Hromatka et al., 2005), had increased expression levels in the late phase of infection (12–24 h), but not in the ‘disrupted RHE’ experiment (not shown). Similarly, the NO marker gene, YHB1, was upregulated in all samples at 12–24 h, albeit not calculated as significant (average 2.3-fold) (Fig. 3). Therefore, direct contact with viable epithelial cells is required to cause nitrosative stress in C. albicans. Other stimuli, such as elevated pH, may have contributed to hyphal formation at the later stages (Saporito-Irwin et al., 1995; Bensen et al., 2004). In fact, we observed that pH-responsive genes (PHR1, PRA1) and other alkaline pH-induced genes linked to the Rim101-pathway (Bensen et al., 2004) were upregulated in late-phase RHE infection.
In vivo transcript profiling confirmed trends observed during experimental oral candidiasis
To determine whether our experimental RHE model observations were consistent with in vivo infections, we analysed the transcriptional profiles of 11 smear samples isolated from HIV+ patients with pseudomembranous candidiasis. The in vivo expression profiles demonstrated unexpected high degree of heterogeneity, as a relative low number of genes were upregulated in all patient samples. Nevertheless, 189 genes were statistically significantly increased in expression (> 1.5-fold), of which 52 were upregulated more than twofold (as compared with the common control; see Experimental procedures) (Table 2). Interestingly, 18 genes have no described orthologue in the non-pathogenic yeast S. cerevisiae.
Table 2. Significantly upregulated genes (> 2-fold) in the in vivo patient samples.
Agglutinin like protein
Putative permease for dicarboxylic amino acids
P-type ATPase involved in Na+ efflux
High affinity glutamine permease
Hyphal wall protein
Putative GPI-anchored protein (IHD1)
Hexose transporter, 3-prime end
Crd1p cardiolipin synthase
Alpha-glucosidase that hydrolyses sucrose (MAL2)
Putative transcriptional regulator
Mevalonate diphosphate decarboxylase
GPI-anchored pH-responsive glycosyl transferase
GAP for RHO1 by homology
Secreted aspartyl proteinase
Transcription elongation protein
ATP synthase regulatory factor
Translation elongation factor
Despite the heterogeneity among the patient samples, several genes that were increased in expression at all time points over the entire RHE time-course experiments (38 genes) showed also an increased expression in the patient samples, indicating a basal correlation between the experimental and in vivo infections. Numerous hypha-associated genes (HWP1, SAP5, SOD5, TEF1, ALS3, orf19.988, SPT6, IHD1 or STF2) were upregulated, indicating that hyphae may constitute the predominant cell population in vivo (Fig. 3). Certain alkaline-responsive genes (e.g. PHR1) (Fig. 3) were also upregulated, indicating exposure of C. albicans to neutral or alkaline pH in vivo. Furthermore, the glucose transporter gene HGT12, known to be expressed at low glucose concentrations (Fan et al., 2002), the maltose permease gene MAL31, the alpha-glucosidase gene orf19.7668 and the gene coding for the key enzymes of the glyoxylate cycle, ICL1, were all upregulated, indicating a metabolic shift towards non-glucose molecules as observed in the RHE model. This strongly supports the view that in vivo C. albicans utilizes two-carbon compounds via the glyoxylate cycle rather than six-carbon compounds via the glycolytic pathway.
Finally, cells in the oral cavity were also exposed to nitric oxide as genes involved in NO-detoxification (SSU1, YHB1 and YHB5) were expressed at high levels. However, it should be noted that, as expected, some putatively infection related genes such as the iron acquisition genes CFL2 and FRE4 were upregulated in the patient samples but not in the RHE model.
Functional analysis of stage specific genes
Based on the genes identified during experimental infection and the in vivo patient samples we chose a set of eight genes (mostly unknown function), identified as being associated with specific phases of infection, for functional analysis. Gene disruption experiments provided tentative data about the importance of these genes for oral infections. orf19.7561 (IPF946), orf19.7300 (IPF2830), orf19.7194 (IPF2147), orf19.3117 (IPF12297), orf19.322 (IPF6758), orf19.93 (IPF14895), orf19.3373 (IPF14155), and MAL31 homozygote mutant strains were constructed and tested for tissue damage in the RHE model. Mutants lacking orf19.93, orf19.7194 and orf19.7561 showed a reduced ability to cause tissue damage (13–40 U l−1 extracellular LDH activity, compared with 85 U l−1 and 77 U l−1 for wild-type and parental strains at 24 h) (Fig. 4). Microarray analysis revealed that orf19.7561 was highly expressed in the early phase of RHE infection (twofold) (although also upregulated in late stages), the patient samples (twofold) and the ‘disrupted RHE’ experiment (Fig. 3)). The strain lacking orf19.7561 was the most attenuated in causing tissue damage (13 U l−1 LDH). This indicated that orf19.7561 might not only be important for the early initiation of the infection process but also for the maintenance and persistence within the epithelium, in particular when C. albicans had damaged the epithelial tissue. We renamed this gene EED1 (Epithelial Escape and Dissemination) and analysed the Δeed1 mutant in more detail.
EED1 plays a critical role in filamentation
Reduced growth was excluded as the cause of attenuated tissue damage in the RHE model because growth rate and morphology of Δeed1 yeast cells were similar to the wild-type (Fig. S1). As the ability to switch to the filamentous form is a vital prerequisite for interaction with epithelial tissue, we tested whether the Δeed1 mutant was dysfunctional in hyphal formation. None of the tested solid media (serum, neutral pH, spider medium, milk-tween agar and embedded conditions) induced filamentous growth of the Δeed1 mutant and no agar invasion was observed (Fig. 5). Similarly, no filaments were observed when Δeed1 mutant cells were incubated for > 12 h in liquid hyphal inducing media or using a quorum sensing-based protocol of hyphal induction (dilution of stationary phase cells into fresh pre-warmed YPD medium) (Enjalbert and Whiteway, 2005). Furthermore, dilution of a saturated culture into 10% serum (Kadosh and Johnson, 2005), which is a potent trigger of hyphal formation, did not induce true hyphal formation and cells remained as yeasts or short pseudohyphal chains. Only very powerful stimuli, such as RPMI including 10% serum, combined with a reduced inoculum size (< 107 cells ml−1) induced elongation of some Δeed1 cells (Fig. 6A). However, the elongations were shorter compared with wild-type cells (Fig. 6A and D), classified as elongated pseudohyphal cells [morphology index (MI) 2.5–3.4] (Merson-Davies and Odds, 1989), and were not stable beyond 4 h (Fig. 6A).
As similar phenotypes were observed for heterozygote (EED1/Δeed1) and homozygote (Δeed1/Δeed1) mutant strains (Figs 5 and 8), it must be concluded that the disruption of only one allele of EED1 is sufficient to cause functional abnormality under hyphal inducing conditions. Therefore, we created a revertant strain by re-integrating the native EED1 gene into the native EED1 locus of the heterozygote mutant strain which completely restored the EED1 associated hyphal defective phenotype (Figs 5–8) (see: Experimental procedures).These data suggest that EED1 expression is essential for true hyphal formation in C. albicans.
Δeed1 induces endocytosis, proliferates intracellularly, but is unable to disseminate interepithelially
As the Δeed1 mutant was incapable of causing tissue damage in the RHE model and was unable to produce filaments under the described conditions (Figs 4 and 5), we analysed the interaction of Δeed1 cells with host cells in more detail. Monolayers of epithelial cells (TR146 cell line) were infected with Δeed1 and parental strain cells and samples were analysed after 1, 3, 5, 7, 9 and 24 h of incubation. We observed short filaments of Δeed1 mutant cells at 1 h and pseudohyphal elongated cells at 3–7 h, which permitted induced endocytosis (Fig. 7A–E). In contrast, no elongated cells were discovered after 24 h (Fig. 7F). However, we observed intraepithelial inclusions which comprised yeast shaped Δeed1 cells and short pseudohyphal chains (MI 0–2.5). In contrast, wild-type and revertant strain cells formed dense filamentous layers (MI > 3.4) (Fig. 7G and H). Given that hyphal formation could not be induced in vitro under various conditions, this suggests that contact with epithelium is a highly potent inducer of filamentation in vivo.
We next determined whether RHE tissue also induced the transient induction of elongated cells in the Δeed1 mutant by analysing histological section after 1 and 24 h (Fig. 8). Contact of Δeed1 cells with the epithelium after 1 h also appeared to induce membrane ruffling, induced endocytosis, and invasion into epithelial cells (Fig. 8B and C). After 24 h, scanning electron microscopy (SEM), transmission electron microscopy (TEM) and light microscopy revealed intraepithelial inclusions and bubonic-like structures in the upper part of the RHE, containing only yeast cells and short pseudohyphal chains (Fig. 8D–G). In contrast, wild-type and parental strains formed dense filamentous layers (MI > 3.4) throughout the whole RHE tissue (as shown in Fig. 2). Therefore, it appears that epithelial contact is able to trigger pseudo-filamentation events in Δeed1 that permit epithelial invasion, and once inside an epithelial cell, the Δeed1 cells are able to proliferate as yeasts or short pseudohyphae but remain trapped intracellularly. This suggests that EED1 expression is not essential for initial invasion into epithelial cells, but is required for epithelial escape and interepithelial dissemination, a property not previously assigned to any fungal gene.
For many human mucosal pathogens the molecular mechanisms involved in colonization, persistence and virulence are poorly understood. In the eukaryote C. albicans, this is compounded by the relative contributions of yeast, hyphal and pseudohyphal growth forms to colonization and infection. To understand this complex host–fungus interaction we used a combination of microscopical, cellular and molecular approaches and an in vitro RHE model of oral candidiasis to dissect the infection process. The early phase (1–3 h) is characterized by epithelial attachment, which is a potent inducer of hyphal formation, and the subsequent upregulation ofhypha-specific and ‘epithelial-contact’ genes. Attachment is probably mediated by fungal adhesins including members of the Als family and Hwp1, which are known to be expressed in vivo (Green et al., 2006; Naglik et al., 2006). Initiation of filamentation triggers both active (physical penetration) and passive (induced endocytosis) mechanisms of entry into epithelial cells. Similar passive processes are induced by invasive bacteria (Cossart and Sansonetti, 2004) and our and previous data (Park et al., 2005) demonstrate that passive uptake is driven by epithelial activity. Phan et al. (2007) recently demonstrated that induced endocytosis of oral epithelial cells is initiated via the first invasin (Als3) of C. albicans in vitro. This is in agreement with our study because we show induced endocytosis and a simultaneous strong expression of the ALS3 gene in both our model of oral candidiasis and patient samples, indicating the importance of our findings and those described by Phan et al. (2007). In addition, preliminary transcriptional data show that ALS3 is upregulated (although not statistically significant) in the Δeed1 mutant (not shown), which may explain why this mutant is still endocytosed but does not produce true hyphae.
Under normal conditions in vivo, the fungus would be kept in check through natural innate defences (antimicrobial peptides, salivary flow and epithelial desquamation). However, under suitable predisposing conditions (e.g. immunosuppression, antibiotic removal of bacterial flora) the fungus can proliferate, disseminate into deeper epithelial layers and cause cell damage, leading to clinical symptoms of disease. We propose this ‘disease’ state is representative of the mid-late phases of infection in our RHE model (6–24 h), which is characterized by cell damage due to fungal hyphae penetrating through epithelial tissue. Invasion and cell damage coincides with a fungal adaptive response to the changing environmental milieu (increased exposure to epithelial cell contents), resulting in the upregulation of various metabolic processes involved in the alkaline pH-response, non-glucose carbon utilization and nitrogen uptake. The switch to metabolizing non-carbon glucose is also detected when C. albicans is exposed to macrophages (Lorenz et al., 2004), possibly indicating that the fungus has access to similar nutrients in both these host cell types. Throughout the infection process epithelial cells are likely to counteract C. albicans invasion through the production of antimicrobial compounds such as nitric oxide (Hromatka et al., 2005), as detected by the transcriptional nitrosative stress response of C. albicans.
Strikingly similar transcriptional adaptive responses were also evident in patients with pseudomembranous candidiasis, albeit with some heterogeneity. This indicates that the RHE infection model mirrors the active disease state in vivo, irrespective of the lack of immune cells (dendritic cells, T and B cells, and polymorphonuclear lymphocytes) that would otherwise restrain widespread C. albicans invasion and cell damage in an immunocompetent or mildly immunosuppressed host. The basal compliance between the in vitro and in vivo systems suggests that, in vivo, C. albicans responds predominantly to the local epithelial environment rather than to specific downstream adaptive immune responses induced by epithelial infection.
The comparison of mid-late stage RHE infections with patient samples led to the discovery of several disease-associated C. albicans genes. One gene, EED1 (IPF946; orf19.7561), which was previously shown to be coregulated with in vitro hyphal production (Nantel et al., 2002; Lorenz et al., 2004; Enjalbert and Whiteway, 2005; Kadosh and Johnson, 2005), was found to be essential for true hyphal formation in vitro. However, Δeed1 could still filament, albeit weakly, upon contact with epithelial cells and induce membrane ruffling, indicating that EED1 is not required for adhesion or induced passive endocytosis. Indeed, preliminary data show that several hypha-associated genes (HWP1, RBT1, SOD5, PCK1, FKH2, RBT4 and BNI4) were highly expressed in the Δeed1 mutant and that the transcriptional profile of Δeed1 and wild-type cells were surprisingly similar in the early phase (1 h) (not shown). The novel virulence function of EED1 emerges after initial epithelial invasion (whether active or passive) in allowing C. albicans to escape from within the epithelial cell in order to disseminate via extended hyphal production, because the Δeed1 mutant remained trapped within the invaded cells and reverted back to yeast form growth. Together with the fact that the Δeed1 mutant was significantly attenuated in causing epithelial damage, we propose that this process of escape and dissemination within the epithelium (whether through EED1 alone or other fungal genes with similar functions) is a major mechanism of fungal pathogenicity at mucosal surfaces.
The strategies of epithelial invasion, escape and dissemination of C. albicans are similar to bacteria, but also show unique features. For example, bacterial access to epithelial cells predominantly depends on host cell activity or components, and bacteria such as Listeria or Salmonella either proliferate within the cell prior to dissemination or lay dormant by hijacking the cell machinery and evading the host immune response (Cossart and Sansonetti, 2004). In contrast, C. albicans appears to gain access via either active or passive invasion and then immediately escapes through hyphal formation in order to disseminate, without evidence of prior proliferation or dormancy. We conclude that this major difference in C. albicans and bacterial virulence mechanisms is due to the fungal ability to modify morphology and strongly supports the belief that filamentation is required for optimal pathogenicity.
We should note that other hyphal mutants lacking the key regulators Efg1 and Cph1 have also been shown to remain intracellular, in contrast to wild-type cells, after phagocytosis by macrophages (Lorenz et al., 2004). However, these data were obtained with professional phagocytic cells and not with epithelial cells. It remains to be investigated whether these mutants are able to invade epithelial cells. Although the behaviour of Δeed1 is distinct from Δefg1, it seems that EED1 is linked to the Efg1 pathway because EED1 is downregulated in Δefg1 mutants (Nantel et al., 2002). Thus, Efg1 may be a possible initiator of EED1 expression, which in turn seems to be essential for the maintenance of hyphal elongation. It is not clear how EED1 regulates these processes but structurally EED1 has striking similarities to Def1, a regulator of RNA polymerase II (RNAPII) in S. cerevisiae with multiple functions (Woudstra et al., 2002; Somesh et al., 2005). Both proteins are of comparable length and unusually rich in glutamine residues (30% Eed1; 32% Def1) over a 200–300 amino acid central region. As Def1 is involved in ubiquitination and proteolysis of RNAPII, it is possible that we have identified a C. albicans homologue that regulates fungal morphology via a similar mechanism.
Media and growth conditions
All C. albicans strains were routinely grown in YPD (1% yeast extract, 1% peptone, 2% dextrose) or SD [0.67% yeast nitrogen base without amino acids, 2% glucose, supplemented with arginine or histidine (20 μg ml−1) or uridine (40 μg ml−1) as required] at 30°C or 37°C. For solid media 2% agar was added. Yeast-to-hypha transition was induced using fetal calf serum (FCS) 10%, RPMI 1640, M199, Spider or N-acetylglucosamine medium, neutral to alkaline pH values, and a ‘quorum sensing’ protocol (Enjalbert and Whiteway, 2005). For agar invasion and embedded experiments ‘milk-tween agar’ (Jitsurong et al., 1993) and YPS (1% yeast extract, 1% peptone, 1% saccharose, 1% agar) was used.
All C. albicans strains were derivatives of strain SC5314 (Gillum et al., 1984). Strain BWP17 (Wilson et al., 1999) was used to create all mutant strains. The genotypes of all strains constructed in this study are listed in Table 3.
Table 3. Wild-type and mutant strains used in this study.
Null mutants lacking either orf19.7561 (IPF946, EED1), orf19.7300 (IPF2830), orf19.7194 (IPF2147), orf19.3117 (IPF12297), orf19.322 (IPF6758), orf19.93 (IPF14895), orf19.3373 (IPF14155) or MAL31 were constructed as described using pFA-His, pFA-Arg and pFA-Ura plasmids (Gola et al., 2003). Briefly, cassettes for transformation were amplified by primers which included >100 bp of gene specific sequences at their 5′-end and the corresponding annealing regions to the pFa vector (Fig. 3, annealing region in bold; Table S2). The transcription protocol was according to Walther and Wendlend (2003).
Ura– mutant strains were recovered by integrating the plasmid CIp10 (URA3) into the RPS1 locus (Murad et al., 2000; Brand et al., 2004). A parental strain was created by integrating CIp30 (URA3; HIS1; ARG4) (Dennison et al., 2005) into the RPS1 locus of BWP17. CIp10 and CIp30 were kindly provided by A. Brown, Aberdeen. All experimental strains were compared with the wild-type (SC5314) and parental strain (BWP17 + pCIp30). As the heterozygote (eed1/EED1) and the homozygote (eed1/eed1) showed comparable phenotypes (Figs 5 and 8), a Δeed1 revertant strain (EED1/EED1) was created by reintegrating the EED1 gene into the heterozygote mutant strain. To produce a revertant strain, we first disrupted one allele of EED1 with pFA-URA (H2; eed1/EED1) (Table 3) and confirmed that this heterozygote mutant had phenotypes similar to the homozygote mutant. In a second step, wild-type homozygosity was restored by a transformation of a PCR-amplified version of the native EED1 gene into strain H2 (eed1/EED1). Strains which had a restored wild-type locus (EED1/EED1) via recombination of EED1 into the disrupted locus were selected on fluorootic acid plates (Fonzi and Irwin, 1993). All gene manipulations in this study were confirmed by PCR and Southern blot analysis.
Oral epithelial cells
The epithelial cell line TR146 cell line was derived from a squamous cell carcinoma of buccal mucosa (Rupniak et al., 1985). TR146 cells were routinely grown in DMEM medium supplemented with 10% FCS. TR146 cells were used after four passages. Prior to inoculation, 90% confluent cells on poly-l-lysine (Biochrom AG, Berlin, Germany) coated glass coverslips (Roth, Karlsruhe, Germany) were washed with phosphate-buffered saline (PBS) and incubated with serum-free DMEM medium in 6 well tissue culture plates. Each coverslip was infected with ∼105C. albicans cells and incubated for the time periods indicated. The epithelial cells were incubated in a humidified incubator at 37°C in 5% CO2.
Reconstituted human oral epithelium
The RHE for the in vitro model of oral candidiasis was based on cultured TR146 cells and supplied by SkinEthic Laboratories (Nice, France). The RHE was maintained in serum-free Maintenance medium (SkinEthic), on a 0.5 cm2 microporous polycarbonate filter (insert) (Schaller et al., 2004). Infection experiments were performed with different C. albicans strains. RHE was infected with 2 × 106Candida cells in 50 μl PBS for different incubation periods (1–24 h). Non-infected controls contained 50 μl PBS alone. To analyse the transcriptional profile of C. albicans cells exposed to intracellular material and cellular debris (as expected for the later stages of infection) we mechanically degraded the RHE prior to inoculation with C. albicans (‘disrupted RHE’). For experiments with the disrupted RHE, the tissue was removed from the polycarbonate filter with a scalpel and destroyed mechanically by resuspending the tissue with a pipette for several times. For inoculation with fungal cells the destroyed tissue was placed back on a polycarbonate filter and inoculated and incubated as described for the non-disrupted tissue samples.
All RHE experiments were performed in a humidified incubator at 37°C in 5% CO2. For microarray experiments all samples were immediately shock frozen and stored at −70°C.
Tissue damage assay
Epithelial cell damage in the RHE infection, caused by the different strains of C. albicans, was determined by the release of LDH into the surrounding medium. Extracellular LDH activity was analysed spectrophotometrically by measuring the rate of NADH disappearance at 340 nm during the LDH-catalysed conversion of pyruvate to lactate (Wroblewski and Ladue, 1955). The LDH activity is given as U l−1 at 37°C and was determined from at least six independent experiments. For statistical analysis, P-values < 0.05 were considered significant.
The endocytosis assay was performed according to Park et al. (2005). Briefly, wells with infected TR146 cells were washed and fixed with Histofix 4% (Roth) for 1 h. Adherent C. albicans cells were stained with an anti-C. albicans monoclonal antibody CA1 (targeting α-1,2 linked mannosides) (kindly provided by D. Poulain, Lille) for 1 h and counterstained with secondary anti-rat IgM conjugated with Alexa Fluor 488 (Molecular Probes). After rinsing, epithelial cells were permeabilized with 1% Triton X-100 and the staining procedure was repeated to detect invading cells using Alexa Fluor 568 (Molecular Probes). Each experiment was performed in duplicates and repeated three times.
The mean MI was used to evaluate C. albicans morphology (Merson-Davies and Odds, 1989). Briefly, MI represented the following morphologies: MI 1 (0–1.5), yeast cells, single or small groups; MI 2 (1.5–2.5), elongated, ovoid cells (‘short pseudohyphae’); MI 3 (2.5–3.4), elongated cells, pseudohyphal, obvious constrictions, sides almost parallel; MI 4 (> 3.4), true filaments, parallel sided, minimal constrictions at sites of septum formation. A minimum of 100 cells were observed for each morphology evaluation.
Light microscopy studies of the RHE were performed as described (Schaller et al., 2004). The morphology of C. albicans was directly assigned in standard microscopy analysis and a minimum of 100 cells were considered for evaluation.
For electron microscopy tissue samples were immersed overnight in 2.5% glutaraldehyde (in 0.05 M Hepes buffer, pH 7.2) and gently washed with distilled water prior to post-fixation. For SEM, post-fixed samples (1% OsO4, 1 h) were rinsed with distilled water, dehydrated with alcohol (stepwise 30–96%), dried [critical point dried in CO2 (CPD 030, BAL TEC, Vaduz, Liechtenstein)] and sputter-coated with 7 nm Au (Polaron Sputter Coating Unit E 5100, GaLa Instrumente, Bad Schwalbach). The samples were examined using a LEO 1530 Scanning Electron Microscope (Carl Zeiss SMT AG, Oberkochen) at 3 kV.
For TEM, post-fixed samples (1% OsO4, 1 h) were rinsed with distilled water, block-stained with uranyl acetate (2% in distilled water), dehydrated in alcohol (stepwise 30–96%), immersed in propylenoxide and embedded in Epon (polymerized 48 h at 60°C, Serva, Heidelberg). Ultra-thin sections were stained with uranyl acetate (2% in distilled water) and lead citrate, stabilized by carbon evaporation (BAE 250, BAL TEC; Vaduz, Liechtenstein) and examined with a TEM 902 (Carl Zeiss SMT AG, Oberkochen) at 80 kV. Images were digitized using a 1K slow scan CCD – camera (Proscan, Scheuring).
For paraffin wax embedding of the tissue samples, post-fixed samples (1% OsO4, 1 h) were rinsed with distilled water, dehydrated with alcohol (stepwise 30–96%), immersed in xylene and embedded in paraffin wax. Selected sections were placed on microscope slides (12 mm Ø). Paraffin wax was removed with xylene, samples were immersed in alcohol (< 96%), dried and examined according to the SEM protocol.
Samples (smears) were collected from European HIV+ patients (< 200 CD4-cells μl−1) suffering from oral pseudomembranous candidiasis. In accordance with local ethical committee approval from the medical faculty Charité, Berlin, signed consent forms were obtained from each patient. After collection, all patient samples were immediately shock frozen and stored at −70°C. Of 31 patient samples collected, eleven provided sufficient C. albicans RNA for microarrays. A ratio equal or less than 3:1 host to C. albicans RNA was shown to be optimal for successful analysis. The ratio was determined by quantitative RT-PCR (QuantiTect, Qiagen) using the ABI PRISM 7700 (Applied Biosystems).
RNA extraction and labelling
Frozen RHE and patient samples were lysed and homogenized (FastPrep®) using PeqGold RNApure reagent (Peqlab) with glass beads (0.5 mm, Roth). Total RNA was extracted as described (Fradin et al., 2005). Total RNA was linearly amplified and labelled using the ‘Low RNA Input Fluorescent Linear amplification Kit’ (Agilent Technologies, Santa Clara, CA, USA).
Microarray hybridization and analysis
For transcriptional profiling, we used C. albicans microarrays (Eurogentec, Seraing, Belgium) as described (Fradin et al., 2005). Sample RNA (from RHE and patients samples, Cy5-labelled) was cohybridized with a ‘common reference’ (RNA from SC5314 grown in YPD, mid-log phase, 37°C, Cy3-labelled), making dye swap controls unnecessary. Slides were hybridized, washed and scanned as described (Fradin et al., 2005). Data normalization (LOWESS) and analysis were performed in GeneSpring 7.2 software (Agilent Technologies). Reliable expression of genes was defined as normalized expression of present genes that did not vary more than 1.5 standard deviations within replicate arrays. Using Student's t-test, a P-value < 0.05 was considered significant. Genes which showed 1.5- to 2-fold changes compared with the common reference were considered as ‘increased’ and >2-fold as ‘upregulated’. Downregulated genes are not discussed in this study. For RHE infections, basal filtering was performed with reliably expressed genes. Reliable gene expression was determined individually for the different data sets of the experimental RHE infection (‘time-course’ and ‘disrupted RHE’ experiment). Only significant data were considered for further analysis. All RHE time-course experiments were performed with two to five biological replicates and the supporting experiment (‘disrupted RHE’) in duplicates.
For patient samples, arrays were hybridized with 11 individual mRNA samples. Data were pooled after normalization (Allison et al., 2006) and only significant data (P-value < 0.05) were considered for analysis of differentially expressed genes. To test whether host RNA contaminations would influence hybridization, labelled human samples alone were hybridized with the C. albicans microarrays. No significant cross-hybridization was detected. The normalized data in MIAME-compliant standard format are available at http://www.galarfungail.org/data.htm.
The validity of the microarray results was assessed for 11 selected key genes by real-time RT-PCR using the QuantiTect Probe RNA-PCR Kit (Qiagen) and a 7500 Real-Time PCR System (Applied Biosystems). Genes investigated were HWP1, ECE1, PHR1, PHR2, SAP5, YHB1, CAR1, FOX2, ICL1, orf19.93, EED1. Relative transcript abundance was determined with the 2–ΔΔCt method (Livak and Schmittgen, 2001). As a calibrator the common reference (see Microarray hybridization and analysis) was used and the ACT1- abundance was set as an endogenous control. For the time points 1 h and 24 h, 11 genes were analysed in two independent RHE samples from two different experiments, which were the same samples used for the microarray experiments. For the experiments with the disrupted RHE we analysed seven genes at the late time point (6 h). To verify the in vivo data, five genes were analysed in three independent patient samples. Overall, the change in expression of all genes tested by qRT-PCR was in agreement with the direction of fold change as determined by microarray analysis.
We thank the Cancer Research Technology, London, D. Poulain, Lille, A. Brown, Aberdeen, and J. Wendland, Jena, for kindly providing cell line, antibodies and plasmids. We are grateful to M. Özel and S. Goyard for very helpful support; J. Laude and B. Fehrenbacher for excellent technical assistance. This work was supported by the Robert Koch-Institute, the Deutsche Forschungsgemeinschaft (Hu528/10), the European Commission (QLK2- 2000–00795; ‘Galar Fungail consortium’) and the National Institutes of Health (R01 DE017514-01 to J.R.N.).