|Home | About | Journals | Submit | Contact Us | Français|
A quantitative real-time RT-PCR system was established to identify which secreted aspartyl proteinase (SAP) genes are most highly expressed and potentially contribute to Candida albicans infection of human epithelium in vitro and in vivo. C. albicans SC5314 SAP1–10 gene expression was monitored in organotypic reconstituted human epithelium (RHE) models, monolayers of oral epithelial cells, and patients with oral (n=17) or vaginal (n=17) candidiasis. SAP gene expression was also analysed in Δsap1–3, Δsap4–6, Δefg1 and Δefg1/cph1 mutants to determine whether compensatory SAP gene regulation occurs in the absence of distinct proteinase gene subfamilies. In monolayers, RHE models and patient samples SAP9 was consistently the most highly expressed gene in wild-type cells. SAP5 was the only gene significantly upregulated as infection progressed in both RHE models and was also highly expressed in patient samples. Interestingly, the SAP4–6 subfamily was generally more highly expressed in oral monolayers than in RHE models. SAP1 and SAP2 expression was largely unchanged in all model systems, and SAP3, SAP7 and SAP8 were expressed at low levels throughout. In Δsap1–3, expression was compensated for by increased expression of SAP5, and in Δsap4–6, expression was compensated for by SAP2: both were observed only in the oral RHE. Both Δsap1–3 and Δsap4–6 mutants caused RHE tissue damage comparable to the wild-type. However, addition of pepstatin A reduced tissue damage, indicating a role for the Sap family as a whole in inducing epithelial damage. With the hypha-deficient mutants, RHE tissue damage was significantly reduced in both Δefg1/cph1 and Δefg1, but SAP5 expression was only dramatically reduced in Δefg1/cph1 despite the absence of hyphal growth in both mutants. This indicates that hypha formation is the predominant cause of tissue damage, and that SAP5 expression can be hypha-independent and is not solely controlled by the Efg1 pathway but also by the Cph1 pathway. This is believed to be the first study to fully quantify SAP gene expression levels during human mucosal infections; the results suggest that SAP5 and SAP9 are the most highly expressed proteinase genes in vivo. However, the overall contribution of the Sap1–3 and Sap4–6 subfamilies individually in inducing epithelial damage in the RHE models appears to be low.
Candida albicans is the most common fungal pathogen of humans. At the most serious level mortality rates from systemic candidiasis are high. However, the majority of patients, notably immunosuppressed individuals, experience some form of superficial mucosal candidiasis, most commonly thrush, and many suffer from recurrent infections.
C. albicans possesses a number of putative virulence attributes and among the most widely studied are the secreted aspartyl proteinases (Saps), encoded by a family of ten genes (SAP1–10) that can be divided into subfamilies based on amino acid sequence homology alignments [SAP1–3, SAP4–6, SAP9–10 (SAP7 and SAP8 are divergent and are not represented as subfamily members)]. Specifically, SAP4–6 are regulated during hypha formation (Hube et al., 1994), which in turn is regulated by transcription factors such as Efg1 and Cph1 (Liu et al., 1994; Stoldt et al., 1997). The discovery that the SAP gene family was differentially expressed under a variety of environmental conditions in vitro (Hube et al., 1994; White & Agabian, 1995) led to the proposition that different members of the Sap family might also be differentially expressed in vivo and contribute to different C. albicans infections. This concept, together with the knowledge that C. albicans inhabits a diverse number of host niches, was the driving force behind subsequent studies that investigated SAP gene expression in several model systems (human samples, animal models and in vitro experimental infections) to ascertain which proteinases or their corresponding genes were expressed during the infective process (reviewed by Naglik et al., 2003a).
Previously, we described a qualitative reverse transcription (RT)-PCR protocol enabling us to analyse SAP1–10 expression in C. albicans isolates obtained directly from the oral cavity and vagina of human patients. We found that SAP2, SAP5, SAP9 and SAP10 were the most common genes expressed during infection and carriage, and that certain SAP genes were correlated with oral and/or vaginal disease (Naglik et al., 1999, 2003a; Albrecht et al., 2006). In addition, using oral reconstituted human epithelium (RHE) and qualitative RT-PCR, we determined a specific order to SAP gene expression: SAP1 and SAP3, followed by SAP6, and finally SAP2 and SAP8 in late phases of infection (Schaller et al., 1998). However, using an RHE model of vaginal candidiasis a different SAP expression profile was observed: SAP2, SAP9 and SAP10, followed by SAP1, SAP4 and SAP5, and finally SAP6 and SAP7 (Schaller et al., 2003). Notably, in both oral and vaginal RHE models using SAP-deficient mutants, it was the Sap1–3 subfamily that contributed to tissue damage (Schaller et al., 1999, 2003), supporting a role for Sap1–3 in establishing C. albicans infections at mucosal surfaces.
Several other groups have also investigated SAP gene expression or activation during the infection process using a number of different animal models of mucosal C. albicans infection. However, a common finding was that although most SAP genes transcripts were detected, different expression profiles were observed. Using in vivo expression technology (IVET), SAP5 and SAP6 were activated in an oesophageal murine model, whereas SAP5 and SAP2 were activated in a disseminated murine model, with SAP2 being detected at later stages of infection (Staib et al., 2000). In a murine vaginitis model IVET demonstrated only SAP4 and SAP5 activation, with SAP5 induced soon after initial infection and SAP4 at later time points and in fewer cells compared with SAP5 (Taylor et al., 2005). In a murine model of gastrointestinal infection IVET also showed a high percentage of SAP4–6 induction, which increased steadily during the course of infection (Kretschmar et al., 2002). This was supported by qualitative RT-PCR that detected SAP4 and SAP6 mRNA in all mice. SAP2, SAP3 and SAP5 mRNA was only detected in some animals and SAP1 mRNA was not detectable. In another murine model of gastrointestinal infection, qualitative RT-PCR demonstrated expression of all ten SAP genes, with similar expression profiles in different oral tissues (Schofield et al., 2003), Finally, in a murine model of oropharyngeal candidiasis, SAP5 and SAP9 were the most strongly expressed genes throughout the course of infection, but sustained expression of most genes was observed – only SAP7 and SAP8 were transiently expressed (Ripeau et al., 2002).
Whilst these studies have been vital in demonstrating SAP gene expression in C. albicans infections, the diversity of data leads to uncertainty about which SAP genes might be required for different mucosal infections. In addition, there appears to be a discrepancy between the above expression studies and previous functional studies using Δsap null mutants, which indicate that the Sap1–3 subfamily contributes more to mucosal infections than the Sap4–6 subfamily (Naglik et al., 2003a). This is despite the observation that nearly all the expression studies indicate either high/enhanced expression or activation of the SAP4–6 subfamily. Therefore, in this study we wanted to revisit and expand on previous studies using a quantitative real-time RT-PCR system, not previously used to analyse C. albicans SAP gene expression, to determine the relative levels of all members of this virulence gene family in vitro and in vivo and which SAP genes are actively upregulated during colonization and invasion of oral and vaginal tissues. In addition, we wanted to determine whether compensatory upregulatory mechanisms exist after gene disruption of specific SAP subfamilies and hypha-promoting genes and whether any correlation might exist between upregulation of certain SAP genes and epithelial cell damage.
C. albicans strains SC5314 (Gillum et al., 1984), CAF2-1 (Fonzi & Irwin, 1993), Δsap1–3 (Kretschmar et al., 2002), Δsap4–6 (Sanglard et al., 1997), Δefg1 (Stoldt et al., 1997) and Δefg1/cph1 (Lo et al., 1997) were used in this study. Prior to experimentation, the mutant status was reconfirmed on the genomic level. Strains were stored on Microbank cryovial beads (Prolab Diagnostics) at −70 °C. For all experiments, for the first preculture, C. albicans strains (from frozen beads) were grown in YPD [1 % yeast extract, 2 % peptone (Beckton Dickinson: 211677), 2 % glucose] medium for 24 h at 30 °C on an orbital incubator to obtain stationary-phase cells. For the second preculture, C. albicans (start concentration ~1×105 cells ml−1) was regrown for 16 h at 30 °C to obtain early stationary-phase cells (~5×108 cells ml−1). To assess quantitative SAP gene expression in culture, C. albicans SC5314 was further grown in YCB/1 %BSA [yeast carbon base/1 % bovine serum albumin (Sigma: A-8022)] and in 10 % human serum in PBS for up to 24 h.
All patients attended the Oral Medicine clinic at Guy’s Hospital or the Genitourinary Medicine clinic at St Thomas’s Hospital, London, UK, and all samples were collected according to institutional ethical guidelines and with patient consent. Full details of sample collection procedures, detailed criteria for clinical diagnosis, and C. albicans speciation were described previously (Naglik et al., 1999, 2003b); however, the patient cohort used in this study was different to that used in our previous studies. Duplicate whole unstimulated saliva or vaginal swab specimens were collected from each subject. Oral candidiasis patients (n=17) were classified as presenting with clinical signs and symptoms of disease and harbouring >2×103 C. albicans c.f.u. per ml saliva. Vaginal candidiasis patients (n=17) matched the criteria of being culture positive and having symptomatic disease with at least one sign and one symptom of infection.
Oral TR146 and vaginal A431 (both squamous cell carcinomas) RHE was purchased from SkinEthic Laboratories and maintained and utilized as described previously (Schaller et al., 2006). Histologically, these three-dimensional organotypic models closely resemble human oral and vaginal mucosa in vivo. For monolayer experiments, oral epithelial TR146 cells were grown in DMEM (Dulbecco’s modified Eagle’s Medium) supplemented with 10 % FBS (fetal bovine serum) and grown to confluency onto the same inserts (Nunc) as used to create the SkinEthic models. Stationary YPD-grown C. albicans SC5314, Δsap1–3, Δsap4–6, Δefg1 and Δefg1/cph1 cells were inoculated at 2×106 cells (in 50 μl PBS; 4×107 cells ml−1) onto the monolayers or RHEs in duplicate and incubated at 37 °C in 5 % CO2 for 1, 3, 6, 12 or 24 h. Non-infected controls contained 50 μl PBS alone. Other controls included inoculation of C. albicans onto empty inserts (without epithelium) but including culture medium, and a 1 h pre-incubation of the RHE and C. albicans with 15 μM pepstatin prior to the addition of C. albicans cells.
The RHE was excised around the circumference with a scalpel and fixed immediately in 10 % buffered formalin. RHE samples were placed in tissue-processing cassettes and run on an overnight program using a Leica ASP 300 tissue processor (Leica Microsystems) and then embedded in paraffin wax. For each RHE, 10–30 3 μm paraffin wax sections were prepared using a Leica RM2055 microtome and silane-coated slides. After dewaxing in xylene, the sections were stained using the periodic acid Schiff (PAS) technique, counterstained with Harris’ haematoxylin and mounted in DPX mountant (VWR, Lutterworth, UK). Histological changes during the time-course of the infection were examined by light microscopy at ×200 magnification.
In all RHE experiments, the release of lactate dehydrogenase (LDH) from epithelial cells into the surrounding medium was used as a measure of epithelial cell damage. Extracellular LDH activity was analysed spectrophotometrically and is given as IU ml−1 at 37 °C.
Total RNA from C. albicans grown in culture, C. albicans-infected RHE samples and patient saliva/swab samples was prepared using the RNeasy RNA isolation system (Qiagen), incorporating glass beads (400–600 μm, Sigma) as previously described (Naglik et al., 1999, 2003b). Total RNA was treated for 30–45 min with Turbo DNA-free DNase (Ambion) and purified RNA from each sample was confirmed DNA-free by the absence of an amplified product after real-time PCR using C. albicans ACT1 (actin) and CEF3 (translation elongation factor 3) genes.
TaqMan primer and probe (5′ FAM, 3′ TAMRA) sets were designed based on unique sequence regions in each individual SAP1–10 gene using PrimerExpress software (PE Applied Biosystems) (Table 1). ACT1 and CEF3 were used as housekeeping control genes for normalization. Calibration and efficiency of all 12 primer/probe sets was assessed in titration experiments using C. albicans SC5314 genomic DNA (500 ng to 5 pg) in serial dilutions. Real-time RT-PCR was used to determine the quantitative levels of SAP1–10 mRNA transcripts in RNA samples using the GeneAmp ABI 5700 and 7500 PCR machines (PE Applied Biosystems) and the QuantiTect TaqMan probe RT-PCR kit (Qiagen) according to the manufacturer’s instructions. Each reaction mixture consisted of 1× RT-PCR buffer, 4 mM MgCl2, 600 nM forward and reverse primer, 200 nM TaqMan probe, HotStarTaq DNA polymerase, omniscript and sensiscript RT enzymes, and template RNA. Following reverse transcription at 50 °C for 30 min and initial denaturation at 95 °C for 15 min, thermal cycling conditions were 94 °C for 15 s followed by 60 °C for 1 min, for 45 cycles. Negative (water) and positive (C. albicans SC5314 DNA) controls were included in each run. SAP1–10 gene expression was normalized to the housekeeping genes ACT1 or CEF3 and analysed using both the two standard curve method and the comparative Ct method (ΔΔCt). The efficiencies of the ACT1 and CEF3 primer sets were comparable and both genes were expressed at similar levels. Data are presented as either mRNA transcripts (arbitrary units) relative to ACT1 or CEF3, or fold difference in expression relative to preculture cells. Each experimental condition was performed in duplicate or triplicate (biological replicates) and reactions (25 μl) were performed in duplicate (technical replicates). In addition, each experiment was repeated at least once on different days for reproducibility and the data verified using independent biological and technical replicates.
For differences in SAP gene expression changes, the Wilcoxon signed-rank matched-pair test was utilized. For differences between SC5314, Δsap1–3 and Δsap4–6, Kruskal–Wallis analysis of variance was utilized. For differences between epithelial models, the Mann–Whitney U test was utilized.
SAP1–10, ACT1 and CEF3 TaqMan primer/probe sets were calibrated and they demonstrated similar efficiency in titration experiments using C. albicans SC5314 genomic DNA (500 ng to 5 pg) in serial log10 dilutions (data not shown). Further quality control experiments were undertaken to specifically assess SAP2 expression in YCB/1 % BSA (inducing medium for SAP2 mRNA expression) and SAP4–6 expression in PBS/10 % serum (inducing solution for the SAP4–6 subfamily expression). As expected, SAP2 mRNA levels were induced by 3 h in YCB/1 % BSA and strongly upregulated by 6 h, with peak mRNA transcript levels at approximately 24 h (~1000-fold increase compared with YPD exponential-phase precultured cells) (Fig. 1a). Likewise, SAP4–6 mRNA production was strongly induced by 3 h in PBS/10 % serum, with peak mRNA transcript levels at approximately 6 h (~10 000–60 000-fold increase compared with preculture cells (Fig. 1b). The data demonstrate the reliable and efficient detection of large-scale expression changes for individual SAP gene members putatively associated with mucosal and systemic infections (reviewed by Naglik et al., 2003a, 2004).
In our earlier studies, cell damage and invasion of Δsap mutants was measured solely by histological analysis (Schaller et al., 1999) rather than by a quantitative assay, such as LDH release. Furthermore, cell damage induced by the Δsap1–3 mutant has not been previously measured by LDH release. Therefore, a comparative time-course of LDH release and histological analysis between SC5314, Δsap1–3 and Δsap4–6 was necessary in this study. Histologically, little difference in invasion potential was evident over 24 h between SC5314, Δsap1–3 or Δsap4–6 (Fig. 2). However, invasion all the way to the basal filter was not observed in the Δsap1–3 mutant but was in SC5314 and the Δsap4–6 mutant, which is similar to observations in other studies (Jayatilake et al., 2006). Surprisingly, over a series of experiments, there was also little difference in LDH release between SC5314, Δsap1–3 and Δsap4–6 after 24 h (Table 2). The addition of pepstatin A, however, significantly reduced LDH release by SC5314, but it was still significantly higher than the PBS control (Table 2). In contrast to the Δsap1–3 and Δsap4–6 mutants, the Δefg1 and Δefg1/Δcph1 mutants were not invasive and surface ‘biofilms’ did not develop (Fig. 2). Although Δefg1 cells were clearly present at 12 h and 24 h, indicating epithelial adherence and possibly cell proliferation, the presence of Δefg1/Δcph1 cells was difficult to detect even at 24 h. With both Δefg1 and Δefg1/Δcph1 mutants LDH release was significantly lower than with SC5314 (Table 2). Together, the data indicate that although the Sap family as a whole does appear to contribute to cell damage, other factors, particularly hypha formation, are the major contributors to mucosal invasion and cell damage.
SAP1–10 mRNA levels were first monitored over a 24 h period after oral and vaginal RHE inoculation with C. albicans SC5314 (Tables 3 and and4).4). All SAP genes were detected in YPD preculture cells, albeit at very low levels, except SAP9, which was expressed at relatively high levels compared with the control gene CEF3 (only ~2–3-fold lower). In both oral and vaginal RHE, SAP9 was clearly the most highly and constitutively expressed gene, although it was downregulated within the first 1 h. SAP10 was also constitutively expressed but at lower levels than SAP9. SAP1 and SAP2 were expressed at similar levels to SAP10 but, notably, were not upregulated in either RHE model, although SAP2 expression trended higher in the vaginal RHE. Interestingly, the hypha-associated genes SAP4 and SAP6 were expressed at very low levels throughout the time-course, despite the presence of hyphae, although SAP6 trended higher at 24 h but only in the vaginal RHE. The only large-scale significant changes in gene expression in both RHE models was observed with SAP5, which was upregulated ~10–20-fold within the first 3 h and ~200–700-fold by 24 h compared with preculture cells (Tables 3 and and4)4) (P<0.05). SAP3, SAP7 and SAP8 remained expressed at very low levels throughout the 24 h time-course, being barely detectable, with no change in expression.
Given that the SkinEthic culture medium strongly induces filamentation, we also examined C. albicans SAP1–10 expression during 24 h growth in culture medium alone (on empty inserts) and in the presence of TR146 oral monolayers grown on the same inserts (Table 5). This was to determine whether the presence of epithelial cells or medium alone strongly induces SAP5 but not SAP4 or SAP6 expression. Interestingly, both in oral monolayers and in medium, SAP5 and SAP6 were hugely upregulated within the first hour compared with RHE infection (SAP5: ~100-fold upregulation vs ~5–10-fold in RHE; SAP6: ~15-fold upregulation vs no change in RHE) whereas SAP4 was downregulated. However, this large initial spike in SAP5 and SAP6 expression, which was not observed in the RHE, was transient, as the expression of both genes decreased markedly by 3–6 h. Thereafter, all three members of the SAP4–6 subfamily demonstrated strong upregulation by 24 h. This strong upregulation was only observed for SAP5 in the RHE system. When comparing the three systems (RHE vs monolayer vs medium) at 24 h, one can see a clear trend in expression for the SAP4–6 subfamily. SAP4–6 were upregulated ~8-, ~250-, ~2-fold in RHE; ~80-, ~500-, ~35-fold in monolayers; and ~280- ~800-, ~200-fold in medium, respectively. Although comparative experiments were not performed using the vaginal RHE, the data demonstrate that epithelial cells might have an inhibitory effect on SAP4–6 expression. In support of this, microscopic observation demonstrated a greater hyphal mass on the plain inserts compared with the inserts with TR146 monolayers. However, whether SAP4–6 upregulation is modulated in vivo remains to be confirmed, but what can be concluded is that SAP5 is upregulated during mucosal infections in vivo. The data also indicate that SAP4 and SAP6 can be controlled independently of filamentation given their very low expression even in the presence of extensive hypha formation in the RHE models.
Both in the oral monolayers and in culture medium, SAP9 was more strongly downregulated within the first hour compared with the RHE models. SAP9 expression stabilized thereafter and then slowly increased at later time points, similar to the RHE. A similar temporal profile was seen for SAP10 but at lower expression levels. Again, SAP2 expression remained relatively unchanged and the expression levels of SAP1, SAP3, SAP7 and SAP8 remained very low throughout the time-course.
We have previously hypothesized that one possible reason why C. albicans possesses a family of ten SAP genes is that the fungus may utilize alternative Sap proteins when another becomes compromised, is removed, or otherwise lost (Hube & Naglik, 2001). Given that numerous studies have implicated the Sap1–3 and Sap4–6 enzymes in mucosal infections (reviewed by Naglik et al., 2003a) we tested this hypothesis by investigating whether C. albicans compensated for the functional loss of these subfamilies by upregulating alternative SAP genes during RHE infections. It should be noted that SAP1–3 expression was not evaluated in the Δsap1–3 mutant, and SAP4–6 expression was not evaluated in the Δsap4–6 mutant. Although we did not observe significant differences in SAP1–10 expression in either mutant compared with SC5314, in the oral RHE SAP5 trended higher in Δsap1–3 mutant throughout the time-course and SAP4 and SAP6 were upregulated at 24 h (Tables 3 and and4).4). Likewise, SAP2 trended higher in the Δsap4–6 mutant compared with SC5314. No differences were observed in the vaginal RHE.
Hyphal forms are present in vivo in both disease and carrier states (Naglik et al., 2003b, 2006) and contribute to cell damage and the disease process (Zakikhany et al., 2007). Given that SAP gene expression is associated with the different morphological forms of C. albicans, we investigated the degree to which SAP1–10 mRNA levels, and in particular SAP4–6, were affected in two hyphal mutants (Δefg1 and Δefg1/cph1) during oral RHE infection. Relative to the respective preculture and SC5314 RHE-infecting cells, in both mutants there was a moderate upregulation of SAP1, SAP3, SAP7 and SAP8 after 12 h but this diminished by 24 h, after which only SAP1 and SAP8 appeared slightly upregulated (Fig. 3). The most profound change relative to SC5314 RHE-infecting cells was the decrease in SAP5 expression in both mutants, especially in Δefg1/cph1. However, relative to preculture cells, SAP5 was still expressed at a much higher level at 12 h and 24 h in Δefg1 compared with Δefg1/cph1 (Fig. 3), despite the absence of hyphae (Fig. 2). In both hypha-deficient mutants, SAP2, SAP4, SAP6, SAP9 and SAP10 expression remained largely unchanged (Fig. 3).
The RHE model does not possess immune cells (i.e. dendritic cells, T and B cells, and polymorphonuclear lymphocytes) or members of the normal microbial flora that would otherwise restrain widespread C. albicans invasion and damage in an immunocompetent or mildly immunosuppressed host. Therefore, we also analysed the SAP1–10 expression profile in 17 patients with oral disease and 17 patients with vaginal disease to determine whether the RHE infections were reflective of active disease in vivo (Fig. 4). As in the RHE models, SAP9 was the most highly expressed gene in both patient groups, with mRNA levels similar to or exceeding that of CEF3 (although SAP9 was not detected in three oral patients). SAP5 was also relatively highly expressed in vivo, with mRNA levels equivalent to that seen in the RHE models at 12 h and 24 h, probably indicating the presence of hyphae in vivo. Interestingly, SAP4 and SAP6 expression remained relatively low, especially in vaginal infections. This corresponds well with the low-level expression of SAP4 and SAP6 in the RHE models and supports the hypothesis that complex epithelial tissue specifically inhibits or negatively controls the expression of these two SAP genes in vivo. SAP1, SAP2 and SAP10 mRNA transcripts were detected at relatively steady and equal levels, but SAP1 was not commonly expressed in oral disease. Low mRNA levels were detected for SAP3, SAP7 and SAP8. Qualitatively, the most commonly expressed genes were SAP2, SAP5, SAP9 and SAP10, and the least commonly expressed were SAP3, SAP4 and SAP7 (Fig. 4). Quantification of SAP1–10 mRNA from C. albicans carriers (colonizers) was attempted but was inconsistent and non-reproducible, probably because the amount of message was below the necessary threshold for reliable detection by this method (data not shown).
This is to our knowledge the first study to fully quantify SAP gene expression levels during human mucosal infections; it aimed to identify which members of this virulence gene family were actively upregulated during infection of oral and vaginal tissues. Our data indicate that (i) SAP5 is the only gene whose expression is significantly altered during epithelial infection, (ii) the hypha formation process regulated by Efg1 and Cph1 is the predominant mechanism by which C. albicans invades and causes mucosal tissue damage, and (iii) the proteinase family as a whole (Sap1–10) does contribute to epithelial damage, but the Sap1–3 and Sap4–6 subfamilies individually contribute little. We therefore conclude that there is little correlation between the expression of specific SAP genes and epithelial cell damage and that the role of the Sap proteins during mucosal infection probably relates more to other biological properties, and subtle direct effects on the host epithelium.
In both mucosal RHE models and in patient samples SAP9 was consistently the most highly, constitutively expressed gene. We have previously shown that SAP9 is frequently expressed both in the commensal stage and during infection in vivo and that (together with its homologous counterpart Sap10) Sap9 is linked to regulatory processes on the fungal cell surface that contribute to cell surface integrity and adhesion to oral epithelial cells (Albrecht et al., 2006). Although SAP9 was also highly expressed in preculture growth (Tables 3 and and4),4), the presence of high levels of SAP9 transcripts throughout the 24 h time-course supports the notion that Sap9 is probably a key proteinase that promotes cell wall integrity, general cell growth and fitness of C. albicans at mucosal surfaces, and is independent of morphology. Although SAP10 mRNA levels were ~10-fold lower than SAP9 levels, SAP10 transcripts were consistently detected both in vitro and in patient samples at steady levels, which may also support a contributory role for SAP10 in C. albicans cell growth and fitness (Albrecht et al., 2006).
In wild-type SC5314 RHE infections, only SAP5 (and not the other hypha-associated genes SAP4 and SAP6) was significantly upregulated during the disease process. Since SAP4 and SAP6 were strongly upregulated within 1–3 h in vitro when C. albicans filamentation was stimulated in the presence of serum (Fig. 1) or culture medium (Table 5 –for SAP6), this suggests that only the Sap5 proteinase from within this subfamily is required to facilitate mucosal colonization, penetration and infection by C. albicans. In support of this, two recent studies have demonstrated that C. albicans (Frank & Hostetter, 2007) and Sap5 in particular (Villar et al., 2007) is able to target and degrade E-cadherin, a major protein in epithelial cell junctions, to facilitate invasion. However, Sap5 does not appear to be directly inducing epithelial cell damage, since the Δsap4–6 mutant was not attenuated in its ability to induce LDH release (Table 2) and because the hypha-deficient Δefg1 mutant did not cause damage even though SAP5 was expressed (Table 2, Fig. 3). Furthermore, histological analysis does not indicate a reduced ability of Δsap4–6 null strain in penetrating oral or vaginal RHE models (Fig. 2). Although the histological and LDH data indicate a less convincing role for Sap5 in mucosal infections, the strong upregulation of SAP5 in both RHE models (Tables 2 and and3),3), its relatively high expression in nearly all patient samples (n=34) (Fig. 4), Sap5’s ability to target E-cadherin (Villar et al., 2007), and other studies demonstrating an association of SAP5 or the SAP4–6 subfamily with infection (Staib et al., 2000; Taylor et al., 2005; Ripeau et al., 2002; Malic et al., 2007), all suggest that Sap5 may facilitate C. albicans infection/penetration of mucosal surfaces in vivo.
In this study we found that compared with monolayers the RHE model appears to specifically reduce the strong upregulation of SAP5 and SAP6 at early time points (1 h) and of SAP4 and SAP6 at later time points (24 h) (Table 5), despite the presence of culture medium and extensive hyphal growth (Fig. 2). This (and other immunological data: J. R. Naglik and others, unpublished data) most likely indicates a natural and innate ability of complex, organotypic epithelium to actively control and modulate C. albicans gene expression as a direct result of its structural organization. However, we previously showed that the initial physical contact between C. albicans and epithelial cells is a strong inducer of the yeast-to-hypha switch (Zakikhany et al., 2007). Therefore, it appears that complex epithelial tissues specifically induce hypha formation in C. albicans but can simultaneously affect or influence hypha-specific gene expression. Another possibility is that C. albicans senses that the RHE structure is more complex (e.g. different surface extracellular matrix proteins and differentiation markers compared with monolayers) and actively self-suppresses the expression of certain SAP genes during hypha formation.
Some of our data differ from other mucosal-based qualitative RT-PCR studies in mice. For example, in a gastrointestinal model SAP4 and SAP6 were constitutively detected and SAP5 was only occasionally expressed (Kretschmar et al., 2002). Similarly, IVET showed that together with SAP5 activation, SAP6 was activated in an oesophageal model (Staib et al., 2000) and SAP4 in a vaginal model (Taylor et al., 2005). Possible reasons for these apparent differences could be the different infection models, or the different molecular techniques employed, which assess different aspects of gene expression/activation. Alternatively, the data might simply highlight the difficulties in comparing animal with human studies, in that C. albicans is a natural colonizer/pathogen of humans but not of mice (Naglik et al., 2008). Clearly there are differences between the interactions of C. albicans with these two mammalian species, which may account for some of the discrepancies between our human data and previous mouse data.
We also feel that some clarity is required concerning some of our previous publications, specifically regarding the lack of SAP5 expression and differences in expression patterns. The SAP gene primers used in some of the initial papers (Schaller et al., 1998, 1999) were not optimal for a qualitative analysis of SAP gene expression. Some primer sets were designed to hybridize upstream of the ATG start site, rather than within the open reading frames. At the time the authors were unclear about the length of the transcripts and so not all the mRNAs may have been detected. Also, the primer sequences cited for SAP5 were listed incorrectly (both primers were actually from the same strand). This was an error that occurred whilst preparing the manuscript, but the actual primers used were correct and gave signals with genomic DNA. Thus, the primer sets used in early publications were not appropriate for comparative qualitative expression analysis. In later publications these errors were corrected (Schaller et al., 2003; Korting et al., 2003), as better, more specific primers were used, and for example strong signals for SAP5 in the oral RHE were clearly evident, albeit at a qualitative level.
In this study, we find no evidence that SAP1–3 are specifically upregulated during the course of infection (Tables 3 and and4)4) or that the Δsap1–3 mutant is attenuated in invasion (Fig. 2) or induces less tissue damage (Table 2). This current dataset may be regarded as contradictory to some of our previous data, which indicated that individual members of the Sap1–3 subfamily contribute to tissue damage in the RHE models (Schaller et al., 1999, 2003). However, in these initial studies only histological analysis was utilized as a measure of cell damage (Schaller et al., 1999), which is not a sufficient marker when used alone. Furthermore, only individual Δsap1, Δsap2, Δsap3 or Δsap1/3 double mutants were tested rather than a Δsap1–3 triple mutant, which has not previously been tested in its ability to induce LDH release from the RHE models. Nevertheless, it is indeed surprising that the individual null mutants were attenuated in inducing damage whereas in this study the Δsap1–3 mutant had no detectable phenotype. We have no logical explanation for this, except possible URA3 effects. URA3 was reintegrated into the SAP1 locus in the Δsap1 mutant, the SAP2 locus in the Δsap2 mutant, and the SAP3 locus in the Δsap3 and the Δsap1–3 mutants. Although unlikely, this difference in the locus site for URA3 reintegration may have somehow affected the ability of the Δsap1–3 mutant to induce epithelial damage (LDH release) compared with the individual sap mutants, particularly Δsap2. All we can say is that after extensive reinvestigations (histology and LDH release), we find little evidence for a role for Sap1–3 in directly causing epithelial cell damage, and suggest that previous data may have been overinterpreted.
Although LDH levels were found to be similar, we observed that the Δsap1–3 mutant does not invade quite as efficiently as SC5314 or the Δsap4–6 mutant (Fig. 2). A recent comparable observation is that C. albicans strain ATCC 10231 was found to be less invasive than SC5314 in a liver invasion model but the induced LDH values were similar (S. Thewes & B. Hube, unpublished data). We found that although ATCC 10231 possesses similar proteinase as well as lipase/phospholipase activity to SC5314, ATCC 10231 produces fewer or shorter hyphae. Therefore, it is possible that in the RHE system SC5314 and Δsap1–3 induce similar LDH values but that the overall invasiveness of the Δsap1–3 mutant is slightly reduced (by an unknown mechanism). This would support our current and past histological observations and also concur with other recent studies (Jayatilake et al., 2006). This hypothesis would also support our current thoughts that hypha formation is the primary factor that assists C. albicans mucosal invasion.
Previously, in humans, we demonstrated that SAP1, SAP3 and SAP8 were more commonly expressed during vaginal rather than oral disease (Naglik et al., 2003b). The present study, using a different set of patient samples, supports these findings, as we found a similar higher frequency of SAP1, SAP3 and SAP8 expression (SAP1: oral 8/17, vaginal 14/17; SAP3: oral 2/17, vaginal 10/17; SAP8: oral 10/17, vaginal 14/17). We also previously suggested that the frequency of SAP1, SAP3, SAP7 and SAP8 expression was correlated with mucosal disease rather than carriage (Naglik et al., 2003b). Such a comparison could not be performed using our real-time system due to inconsistencies when working with low levels of Candida message from carriers. However, we found little evidence for strong upregulation of these four genes in RHE wild-type (SC5314) infections. Also, the expression levels of SAP1, SAP3, SAP7 and SAP8 were very low in the RHE models and generally low in patient samples when detected. This may explain our previous qualitative findings, in that the low levels of message for these three genes might have been detected in patient samples but not carriers. Based on these observations, the direct individual contribution of SAP1, SAP3, SAP7 and SAP8 to the disease process in humans would appear to be minimal. The lack of evidence for a virulence role for SAP7 is supported by previous work using a murine vaginal infection model demonstrating no alterations in virulence using a Δsap7 null strain (Taylor et al., 2005). However, it should be noted that not all Sap enzymes may have equal activity, and that if certain Sap proteins are highly active in vivo they could be very effective when in low abundance. Furthermore, the protein levels of proteinases and their respective half-life may also be different. Therefore, it is possible that although some of the SAP genes are expressed at very low levels they may still have biologically relevant functions, especially those that are localized in the cell wall. However, even if this is the case, the data indicate that these functions do not appear to be correlated with epithelial invasion and cell damage.
In RHE models, we previously found a distinct order of SAP gene expression during the course of oral (SAP1/3, SAP6, SAP2, SAP8) (Schaller et al., 1998) and vaginal (SAP2, SAP9/10, SAP1, SAP4/5, SAP6/7) (Schaller et al., 2003) infections. However, these data were qualitative and provided no precise information regarding gene expression levels or upregulation, and some of the data (Schaller et al., 1998) were also observed at much later time points (36–96 h) than those used in this study (up to 24 h). Therefore, we would again like to clarify some issues regarding the time-course used in this and previous publications. The early work (Schaller et al., 1998) was somewhat exploratory and used extended time points (up to 96 h) to make damage-orientated observations. Three repeat experiments were performed and in one experiment the histological alterations and the corresponding detection of SAP gene expression started 30 h earlier than the other two experiments. In addition, experiments were often performed with RHE comprising epithelium of different stages of differentiation. At the time we were unaware that this could have an impact on the histological and SAP expression data, as we now know that as the RHEs differentiate they become more resistant and less immunologically reactive to C. albicans infection (unpublished observations in the Naglik and Hube laboratories). In the following years we used 12–36 h (Schaller et al., 1999; Korting et al., 2003; Schaller et al., 2003) and thereafter up to 24 h incubation periods with C. albicans (Schaller et al., 2005; Weindl et al., 2007). In our more recent study using oral RHE (Korting et al., 2003), we observed strong expression of SAP2, SAP5, SAP9 and SAP10 at 36 h; these are also the four SAP genes we find more highly expressed at 24 h in this study, demonstrating conformity between the studies. Currently, no more than 24–36 h is required to study all aspects of C. albicans–host interaction using the SkinEthic models (Schaller et al., 2006).
Although we did not observe strong compensatory changes in SAP gene expression in the Δsap1–3 or Δsap4–6 mutants, alterations were clearly evident with the hypha-deficient mutants Δefg1 and Δefg1/cph1. SAP1, SAP3, SAP7 and SAP8 were upregulated compared with the wild-type after 12 h, and SAP1 and SAP8 after 24 h in both mutants. Although the precise reasons for these differences in SAP expression in the hypha-deficient mutants are unclear, the results potentially demonstrate the versatility of C. albicans in adapting to different environments and situations that the fungus may encounter in vivo. It should be noted that SAP5 was still expressed at relatively high levels in Δefg1 compared with Δefg1/cph1, supporting the notion that SAP5 expression can be hyphal independent and suggesting that both hyphal signal-transduction pathways (Efg1 and Cph1) are important for SAP5 expression in vivo as previously suggested (Felk et al., 2002; Korting et al., 2003; Staib et al., 2002).
In summary, we conclude that the quantitative C. albicans SAP1–10 expression profiles presented in this study are indicative of the transcriptional responses that occur during human mucosal infections in vitro and in vivo. However, we would also like to take this opportunity to state our current beliefs in the role and contribution of the SAP genes in mucosal infections: (i) all SAP genes can be expressed in vivo, (ii) SAP gene expression remains largely unaltered during oral and vaginal (RHE) infection, with the exception of SAP5, which is significantly upregulated; (iii) Sap1–3 are not as critical for mucosal tissue digestion in the RHE models as previously concluded; (iv) individual SAP genes contribute little to the overall invasiveness and damage-inducing ability of C. albicans in vitro, but the SAP family as a whole are involved in causing damage, as pepstatin addition partially inhibits invasion and LDH release; (v) complex, organotypic mucosal tissues appear to be able to regulate and manipulate expression of C. albicans genes, particularly SAP4–6; (vi) hypha formation is probably the most essential factor for inducing epithelial invasion and damage, and in light of recent data demonstrating degradation of E-cadherin by Sap5 we suggest that Sap5 may play an indirect role in facilitating hyphal invasion, but not in inducing epithelial damage directly.
Also in this issue of Microbiology, using a recombinase-based genetic reporter system, Lermann & Morschhäuser (2008) report that only SAP5 is significantly activated during vaginal RHE infection and that SAP1–6 members do not appear to be required for invasion. This is in general agreement with our current findings. However, contrary to our data, the authors demonstrate no effect on RHE invasion and damage using the aspartic protease inhibitor pepstatin A.
We are very grateful to Drs D. Sanglard, J. Ernst and G. Fink for kindly supplying the C. albicans Δsap4–6, Δefg1 and Δefg1/cph1 null mutants, respectively. Many thanks also to Sonia Tait for performing LDH assays, Lynette Fernandes-Naglik for identification of Candida species, and Ron Wilson for assistance in statistical analysis. J. R. N. was supported by a personal Wellcome Trust Value in People (VIP) award and by NIDCR (DE015528 and DE017514). M. S. and G. W. were supported by the Deutsche Forschungsgemeinschaft (Sch 897/1; Sch 897/3). B. H. was supported by the Deutsche Forschungsgemeinschaft (Hu528/8-3 and 4).