OUP user menu

Candida species fail to produce the immunosuppressive secondary metabolite gliotoxin in vitro

Claudio Kupfahl, Thomas Ruppert, Annebärbel Dietz, Gernot Geginat, Herbert Hof
DOI: http://dx.doi.org/10.1111/j.1567-1364.2007.00256.x 986-992 First published online: 1 September 2007

Abstract

Yeasts of the genus Candida are a major cause of morbidity and mortality in immunocompromised patients. Despite new insights in recent years, the pathogenesis of Candida infection is still incompletely understood. Previous studies have suggested that gliotoxin, a secondary fungal metabolite with well-known immunosuppressive effects, is produced by various species of the genus Candida, and a possible role of gliotoxin as a virulence factor of C. albicans has also been discussed. However, until now, no definitive evidence has been provided that members of the genus Candida are able to produce gliotoxin. To clarify this question, we tested a total of 100 clinical isolates of C. albicans,C. glabrata,C. tropicalis,C. krusei and C. parapsilosis for gliotoxin production using a highly sensitive HPLC protocol, and, for selected isolates, confirmed our findings by tandem MS. This approach did not detect intracellular or extracellular gliotoxin production by any of the isolates examined, although various culture conditions were applied. Therefore, in contrast to previous studies, our data strongly suggest that at least the Candida species investigated in this study are not able to produce the secondary metabolite gliotoxin.

Keywords
  • Candida
  • virulence
  • gliotoxin

Introduction

During the past two decades, the incidence of infection with pathogenic fungi has increased dramatically. The most frequently encountered infections are caused by yeast species belonging to the genus Candida, which can cause a wide array of human diseases, ranging from superficial mucosal infection to life-threatening invasive infections (Weig et al., 1998; Sullivan et al., 2004). In particular, Candida albicans represents a major cause of morbidity and mortality in immunocompromised hosts, but other Candida species, such as C. glabrata and C. krusei, have emerged in recent years (Richardson,2005).

The pathogenesis of Candida infections has been studied extensively, and a panel of virulence attributes seems to be involved in the infective process (Calderone & Fonzi,2001; Naglik et al., 2003). Of these putative virulence factors, morphologic changes, adhesins and secreted hydrolytic enzymes, such as secreted aspartyl peptidases (Saps), have been most widely investigated (Ghannoum,2000; Naglik et al., 2003,2004; Yang,2003). However, many of the interactions with the host are still incompletely understood, and almost nothing is known about the potential role of secondary fungal metabolites in the pathogenesis of Candida infection.

In previous studies, Shah et al. suggested that Candida species can produce the epipolythiodioxopiperazine metabolite gliotoxin in vitro and in vivo, and discussed a possible role of gliotoxin in the pathogenesis of Candida infection (Shah & Larsen,1991; Shah et al., 1995). Gliotoxin, which is produced by molds, in particular by Aspergillus fumigatus, is a specific inhibitor of nuclear factor kappa B (Pahl et al., 1996) and is known to mediate a broad spectrum of immunosuppressive effects in vitro, e.g. the inhibition of cytokine production, reduction of cytotoxic T-cell activity, and impaired production of reactive oxygen species by macrophages (Yamada et al., 2000; Nishida et al., 2005; Kupfahl et al., 2006a). Furthermore, administration of gliotoxin to mice results in strong immunosuppression in vivo (Sutton et al., 1994). Moreover, it was shown that gliotoxin can be detected in the lungs of mice with experimentally induced invasive aspergillosis (Lewis et al., 2005). As a result of these observations, gliotoxin has been discussed as a virulence factor for the fungus in invasive aspergillosis (Bauer et al., 1989; Sutton et al., 1994,1996; Wright et al., 2001; Upperman et al., 2003; Gardiner et al., 2005).

Therefore, the observation that Candida is also able to produce this mycotoxin could have important implications for the pathogenesis of Candida infection. However, due to technical limitations,Shah & Larsen (1991) were not able to definitely prove that Candida species can produce gliotoxin.

To verify the suggested production of gliotoxin by Candida species and to further study the species distribution and frequency of gliotoxin production within the genus Candida, we investigated 100 clinical isolates of C. albicans,C. glabrata,C. krusei,C. parapsilosis and C. tropicalis for gliotoxin production using a highly sensitive HPLC protocol, and, for selected isolates, we confirmed our findings by tandem MS.

Materials and methods

Fungal strains

One hundred clinical Candida isolates were collected at the University Hospital of Mannheim. Forty C. albicans strains were isolated from either blood (10), bronchoalveolar lavage flluid (10), pharyngeal swabs (10) or vaginal swabs (10). Fifteen isolates each of C. glabrata,C. krusei,C. parapsilosis and C. tropicalis were isolated from various clinical specimens. Species identification of strains was performed using CHROMagar candida (Becton Dickinson, Heidelberg, Germany). If necessary, presumptive identification was verified with the API20C AUX profile (Bio Mérieux, Nürtingen, Germany). All isolates were preserved using standard techniques. Before testing, all strains were subcultured on Saboraud agar (Oxoid, Wesel, Germany) at 35°C for 5 days. An Aspergillus fumigatus strain known to produce gliotoxin (ATCC 46645) was used as a control (Kupfahl et al., 2006b).

Determination of fungal growth kinetics

The growth rates of all isolates were determined by measurement of fungal biomass using a tetrazolium hydroxide (XTT) reduction assay as previously described (Meletiadis et al., 2001). In brief, yeast cells were adjusted to 1 × 105 cells mL−1 in RPMI-1640 medium buffered to pH 7.0 with HEPES (Sigma-Aldrich, Taufkirchen, Germany) using a haemocytometer. Two hundred microliters was then inoculated in a 96-well flat-bottomed microtiter plate and incubated in a rotary incubator at 37°C and 130 r.p.m. Metabolic activity corresponding to fungal biomass was determined after 48 h as the OD at 450 nm (OD450 nm) 2 h after the addition of 20 μL of sterile filtered XTT solution (EZ4U, Biozol, Eching, Germany). Medium without cells served as a control. All isolates were tested in duplicate.

Culture conditions for gliotoxin screening

For the standard culture condition, a liquid culture was performed as described previously (Kupfahl et al., 2006b). In brief, yeast cells were adjusted to 2 × 104 mL in RPMI-1640 (25 mM HEPES; Sigma-Aldrich, Taufkirchen, Germany) using a haemocytometer. In total, 20 mL of conidial suspension was put into Erlenmeyer flasks and incubated in ambient air in a rotary shaker at 37°C under gentle agitation at 130 r.p.m. for 7 days. Alternatively, Eagle's MEM medium (25 mM HEPES; Sigma-Aldrich, Taufkirchen, Germany) supplemented with 5% fetal bovine serum was used as culture medium for some strains. Cultures were incubated at 37°C for 7 days under a 5% CO2 atmosphere as described by Shah & Larsen(1991). Under these conditions, C. albicans and C. tropicalis grew as yeast cells, pseudohyphae, and hyphae, respectively. Candida krusei and C. parapsilosis grew as yeast cells and pseudohyphae. These Candida species are not known to be able to produce true hyphae (Ahearn & Schlitzer,1984). Candida glabrata, known to be unable to produce hyphae or pseudohyphae (Ahearn & Schlitzer,1984), grew as yeast cells and blastospores. Furthermore, yeast cells of some strains were examined for gliotoxin after 7 days of culture at 37°C in liquid RPMI medium (130 r.p.m.), on Sabouraud agar and on potato agar (Oxoid, Wesel, Germany).

Gliotoxin extraction from culture supernatant and yeast cells

Gliotoxin extraction was performed as described before (Kupfahl et al., 2006b). In brief, for detection of gliotoxin in culture supernatant, fungal biomass was harvested by filtration through a 0.22-μm filter (GHP Acrodisc; PALL, Dreieich, Germany). Culture filtrates were extracted three times with chloroform, and the extracts were pooled, dried in a rotary vacuum concentrator at 42°C, reconstituted in 250 μL of acetonitrile, and stored at 4°C until use. Medium inoculated with a gliotoxin standard (Sigma-Aldrich, Taufkirchen, Germany) was used as a control. For detection of gliotoxin in yeasts, cells were harvested and ground to fine powder under liquid nitrogen in a prechilled mortar. Ground cells were resuspended in a Triton-X 0.5%/chloroform mixture (1:1, v/v) and stirred for 2 h. The chloroform layer was collected and evaporated as described above. Dried material was reconstituted in 250 μL of acetonitrile and stored at 4°C until use.

HPLC and tandem MS detection of gliotoxin

HPLC (System Gold, Beckman, Munich) was used for the quantification of gliotoxin in culture extracts as described previously (Kupfahl et al., 2006b). In brief, for HPLC, a reversed-phase C-18 column (300 × 3.9 mm internal diameter, 15 μm spherical packing; Waters, Eschborn, Germany) was used. HPLC analysis of samples (50 μL) was performed as gradient elution, and the absorbance of the effluent was monitored at 280 nm. A standard curve was obtained with a gliotoxin standard (Sigma-Aldrich, Taufkirchen, Germany) ranging from 62.5 to 8000 ng. Mean interassay and intra-assay coefficients of variation over the range of the standard curve were <10%. The overall recovery of gliotoxin standard from culture supernatants was 72.4%, determined as the ratio of defined amounts of gliotoxin standard extracted from RPMI medium with subsequent HPLC quantification and the same amounts of gliotoxin standard directly injected into the HPLC system. The detection limit of the HPLC system was less than 10 ng mL−1 gliotoxin as assessed by serial dilution of gliotoxin standard in RPMI culture medium with subsequent gliotoxin extraction and quantification by HPLC. The results of the HPLC method were confirmed for at least two samples of each species by tandem MS. With this method, even 2 ng mL−1 gliotoxin were detected with a signal-to-noise ratio higher than 8. For detection of gliotoxin with MS, the HPLC effluent of appropriate time for gliotoxin elution was collected, concentrated in a rotary evaporator, and stored at 4°C until further analysis with MS. For tandem MS, 2 μL of the collected fraction was pipetted into a precoated borosilicate nanoelectrospray needle (MDS Protana, Odense, Denmark). MS analysis was performed on a quantitative time-of-flight (Q-TOF) mass spectrometer (Qstar Pulsar; Applied Biosystems, Darmstadt, Germany) equipped with a nano-electrospray ionization ion source (MDS Protana, Odense, Denmark). The potential was set to 900 V, and the declustering potential and focusing potential were set to 40 and 180 V, respectively. Fragmentation of selected compounds was usually performed with a collision energy of 17 V.

Results

Growth rates of fungal strains

To ensure that all strains showed sufficient growth under the standard culture conditions used, the growth kinetics of all isolates were monitored using an XTT biomass assay. Preliminary studies with well-characterized isolates of each species showed that the fungal biomass peaked around day 2–3 after inoculation of the liquid cultures (data not shown). Thus, in order to estimate the growth kinetics of individual isolates, the biomass was measured after 2 days. As shown in Fig.1, the growth rates of individual isolates differed somewhat among the Candida species and, to a lesser extent, within a species. However, after 2 days of culture, all strains grew substantially.

1

Growth rates of Candida strains tested for the production of gliotoxin. The individual growth rate of all isolates (n=100) was measured as the fungal biomass after 48 h of culture in RPMI-1640 medium using a colorimetric XTT reduction assay. The bar represents the median fungal biomass of each group.

Screening for gliotoxin production

With the gradient elution HPLC protocol described above, gliotoxin eluted as a single peak after 17.6 min. Figure2 shows a representative HPLC chromatogram of gliotoxin standard extracted from RPMI medium (a), the standard curve of gliotoxin standard extracted from medium (b), and a representative chromatogram of extracted culture supernatant of the A. fumigatus strain known to produce gliotoxin (c). On the basis of these results, the extracted culture supernatants of all isolates were tested after 7 days of cultivation in RPMI medium. Remarkably, despite sufficient growth of all Candida isolates, gliotoxin was not detected in any of the 100 isolates, independent of the species tested. Figure3 shows a representative HPLC chromatogram for each species (a, c–f). As a further control, culture supernatants of at least two strains of each species were spiked with 20 ng mL−1 gliotoxin standard. As shown in Fig.3b, the culture supernatant of C. albicans did not interfere with the extraction of gliotoxin, so a technical problem at this step could be excluded. To detect very small amounts of gliotoxin that could escape detection by HPLC, the results of the HPLC were confirmed by tandem MS for at least two samples of each Candida species. Gliotoxin was definitely detected in the HPLC effluent of the sample of A. fumigatus by tandem MS, thus verifying the result of the HPLC (Fig.4a and b). However, gliotoxin was not detected in the HPLC effluent of any of the Candida species tested (Fig.4c and d).

2

HPLC detection of gliotoxin. (a) Sample of gliotoxin standard extracted from RPMI medium. Shown is the HPLC chromatogram of gradient elution with UV detection at 280 nm over a range of 22 min. Gliotoxin eluted after 17.6 min. (b) Linearity and reproducibility of the HPLC method. Defined gliotoxin standard concentrations were extracted from RPMI medium, and a standard curve was created using the peak values of HPLC detection at 280 nm. Values are given as the mean of three replicates±SD. (c) HPLC chromatogram of extracted culture supernatant of an Aspergillus fumigatus control strain known to produce gliotoxin. The strain was grown in liquid RPMI medium for 7 days. The HPLC effluent at the indicated retention time of gliotoxin (17.6 min) was collected for further analysis by tandem MS.

3

Representative HPLC chromatograms of each Candida species. For better orientation, the retention time of gliotoxin is marked in each chromatogram. (a) HPLC chromatogram of an extracted culture supernatant of a representative Candida albicans isolate from 40 isolates tested. (b) The chromatogram of the culture supernatant of the same isolate of Candida albicans, but with the addition of 20 ng mL−1 gliotoxin standard before extraction. (c–f) Representative HPLC chromatograms of Candida glabrata (c), Candida krusei (d), Candida tropicalis (e) and Candida parapsilosis (f). In total, 15 isolates of each species were tested.

4

Tandem MS verification of results. (a) Mass spectrum of the collected HPLC fraction of the Aspergillus fumigatus control strain. The small insert shows the mass spectrum of the gliotoxin standard. The compound with 327.06 m/ z, which is the known mass of gliotoxin, was selected for further collision-induced dissociation. (b) Four major fragments with 263, 245, 227 and 111 m/ z resulted after collision-induced fragmentation of the compound with 327.06 m/ z. An identical fragmentation pattern was found for the compound with 327.05 m/ z of the gliotoxin standard (small insert). (c, d) Mass spectra of collected HPLC fractions of Candida albicans (c) and Candida glabrata (d). No compound with 327.05 m/ z was detectable in the mass spectrum, and high-resolution tandem MS showed none of the gliotoxin fragments (small insert). At least two isolates of each species were tested.

As it is known that culture conditions can influence the production of gliotoxin (Belkacemi et al., 1999; Watanabe et al., 2004b), we next cultured at least four strains of each Candida species in Eagle's MEM medium using the culture conditions described by Shah & Larsen(1991). Under these altered culture conditions, gliotoxin was also not detected in culture supernatants. To further prove the absence of gliotoxin in Candida species, we next analyzed yeast cells for intracellular gliotoxin. Therefore, after 7 days of culture, yeast cells were harvested from liquid RPMI medium or from Sabouraud agar and potato agar cultures, and subjected to the gliotoxin extraction procedure; extracts were then analyzed by HPLC. As in the culture supernatants, gliotoxin was not detectable in any of the strains investigated.

Discussion

It was suggested previously that Candia species are able to produce the immunosuppressive secondary metabolite gliotoxin (Shah & Larsen,1991,1992; Shah et al., 1995,1998), and these investigations achieved a high profile in the secondary literature. Therefore, we wanted to investigate the frequency and species distribution of gliotoxin-producing strains among clinical Candida species. However, among a total of 100 clinical Candida isolates cultured under various conditions, we did not detect any intracellular or extracellular gliotoxin production. Our results strongly suggest that at least isolates of clinically important Candia species collected in a large German hospital are not able to produce gliotoxin.

The HPLC and the tandem MS method used here are highly sensitive and specific for the detection of gliotoxin (Kupfahl et al., 2006b). However, to exclude the possibility that technical insufficiencies were responsible for the lack of gliotoxin detection in this study, we included several controls, namely validation of all steps of the extraction procedures, and validation of the HPLC and the tandem MS method using a commercially available gliotoxin standard. With regard to the above-mentioned studies describing gliotoxin production by C. albicans, the authors used an isocratic HPLC method for detection of gliotoxin, and as further confirmation they compared the UV absorbance spectrum of the chromatographic peak with that of a gliotoxin standard. However, it is known that identification of substances on the basis of HPLC retention times suffers from several limitations, of which the most important is the coelution of compounds other than the target compound, in particular when an isocratic approach is used, as performed by Shah & Larsen(1991). Therefore, the analysis of the UV absorbance spectrum of selected HPLC peaks over a wide range could be helpful for identification, but also often suffers from lack of specificity (Pragst et al., 2004). Moreover, as the UV spectra observed by Shah & Larsen(1991) were not identical, the authors themselves stated that more rigorous testing is required to verify the production of gliotoxin by C. albicans. However, in subsequent work on this topic, the authors stated that C. albicans is known to produce gliotoxin (Shah et al., 1995,1998), but provided no further evidence. In the last few years, technical limitations in the identification of selected HPLC compounds have been increasingly overcome by sophisticated methods such as tandem MS. Our approach, using a highly sensitive combination of HPLC and tandem MS of fractions of appropriate time for gliotoxin elution, showed the absence of gliotoxin within the detection limit of the method among the Candida isolates tested in the current study.

As it is known that the culture conditions influence the production of gliotoxin (Belkacemi et al., 1999; Watanabe et al., 2004b), we cannot exclude the possibility that Candida species are able to produce gliotoxin under culture conditions other than those chosen for the current study. However, the culture conditions used here included the culture conditions described by Shah & Larsen(1991). Furthermore, the RPMI liquid culture assay was found to be very potent in stimulating gliotoxin production in molds (Watanabe et al., 2004a, b). When the A. fumigatus control strain was cultured in liquid RPMI, gliotoxin was already detectable in the culture supernatant after 24 h of culture (data not shown). However, the incubation time was prolonged to 7 days in order to also detect possibly later and/or weaker gliotoxin production by Candida species.

Candida is known to secrete proteases, phospholipases and lipases (Naglik et al., 2003,2004; Yang,2003). However, few data exist for the production of secondary metabolites by Candida species. In molds, the genes that encode the proteins involved in the biosynthesis of secondary metabolites are often clustered in the genome (Gardiner et al., 2005); well-known examples are penicillin and aflatoxin. Recently, a putative gliotoxin gene cluster was identified in A. fumigatus, and the first step in the putative biosynthetic pathway of gliotoxin in A. fumigatus was confirmed experimentally (Gardiner & Howlett,2005; Kupfahl et al., 2006b). However, such a homologous gene cluster encoding for gliotoxin biosynthesis is not present in the genome of C. albicans, thus corroborating our results at the genetic level [(Gardiner.(2004) and database search].

As we screened for gliotoxin only, this does not exclude the possibility that the strains investigated in our study were able to produce other bioactive secondary metabolites. The identification and the investigation of the toxicity of some HPLC peaks observed in this study, namely a peak at 13.9 min that was repeatedly detectable in C. albicans,C. glabrata and C. krusei, is ongoing in our laboratory and might be helpful in clarifying the possible role of secondary metabolites in the pathogenesis of Candida infection.

Footnotes

  • Editor: Richard Calderone

References

View Abstract