Abstract
The human pathogenic yeast Candida albicans can cause an unusually broad range of infections reflecting a remarkable potential to adapt to various microniches within the human host. The exceptional adaptability of C. albicans is mediated by rapid alterations in gene expression in response to various environmental stimuli and this transcriptional flexibility can be monitored with tools such as microarrays. Using such technology it is possible to (1) capture a genome-wide portrait of the transcriptome that mirrors the environmental conditions, (2) identify known genes, signalling pathways and transcription factors involved in pathogenesis, (3) identify new patterns of gene expression and (4) identify previously uncharacterized genes that may be associated with infection. In this review, we describe the molecular dissection of three distinct stages of infections, covering both superficial and invasive disease, using in vitro, ex vivo and in vivo infection models and microarrays.
Introduction
The polymorphic yeast Candida albicans is the most important fungal pathogen of humans. However, as well as being a successful pathogen, C. albicans exists as part of the normal human microbiota and is not found in environmental reservoirs such as soil. In the context of the rest of the fungal kingdom, this lifestyle is remarkable: of the estimated 1.5 million (over 100 000 of which have been confirmed) fungal species, only a very small percentage (150–200 species) is capable of causing infections in humans. Some of these species are specialized for infections of the skin (dermatophytes or Malassezia species), while others are also able to cause more serious, systemic infections. Most of the major pathogens in the latter group (such as Aspergillus fumigatus, Cryptococcus neoformans and Histoplasma capsulatum) are environmental fungi, capable of exogenously infecting susceptible individuals. Candida albicans on the other hand is, under normal circumstances, a benign colonizer of human mucosal surfaces and, therefore, highly specialized for life on or within the human host. However, certain alterations to the host environment can result in the transition from a commensal to a pathogenic phase, permitting infection of virtually every organ of the human body and resulting in severe infections. Even a mildly compromised immune system or a minor imbalance of the microbiota can be sufficient for C. albicans to cause infections of the skin or of mucosal surfaces. These superficial infections are extremely common – for example, c. 75% of all women experience vulvovaginal candidosis during their lifetime and a significant proportion suffer from recurrent infections. In addition, oral and oesophageal candidosis are particularly common in HIV-positive individuals and, without intervention with highly active antiretroviral therapy, occurred in up to 90% of HIV patients (Calderone, 2002). The severity of candidosis, however, increases dramatically in patient populations with predisposing factors such as severely impaired immunity (e.g. neutropenia), cancer (e.g. leukaemia), disruption of natural barriers (e.g. by burn injury or disruption of gut mucosal barriers by abdominal surgery), the presence of indwelling catheters, dialysis or solid organ transplantation (Ruhnke & Maschmeyer, 2002; Perlroth, 2007). These mostly hospital-acquired (nosocomial) bloodstream and invasive infections are life-threatening diseases and candidaemia is now the third most common form of nosocomial bloodstream infection responsible for 9% of such infections (Wisplinghoff, 2004; Perlroth, 2007). Not only are invasive Candida infections common in intensive care units of hospitals, they are also difficult to treat (Perlroth, 2007). Even with first-line antifungal therapy, disseminated candidosis has an attributable mortality of up to 50% in some patient populations, depending on the underlying illness (Perlroth, 2007). The mortality of severe sepsis caused by Candida is >50%, and is therefore higher than the mortality from sepsis due to any bacterium, including Pseudomonas aeruginosa. Invasive candidosis is usually due to fungal entry into the blood resulting in candidaemia or haematogenous dissemination to internal organs. Candida albicans can also cause candiduria, perotinitis, endocarditis, pericarditis, endophthlamitis, meningitis and pneumonia (Calderone, 2002). It is this extraordinary ability of C. albicans to successfully infect virtually every anatomical site of the human host that makes it such an important organism to study – both from a medical and biological perspective.
The exceptional adaptability of C. albicans is mediated by rapid alterations in gene expression in response to environmental stimuli, such as changes in nutrient availability, pH, osmolarity, temperature or attack by cells of the immune system (Fradin, 2003; Prigneau, 2003; Rubin-Bejerano, 2003; Bensen, 2004; Hube, 2004; Brown, 2007; Kumamoto, 2008) and this transcriptional flexibility can be monitored with tools such as microarrays. Indeed, this decade has been witness to numerous genome-wide studies on C. albicans gene expression. These, mainly in vitro studies have described the transcriptional response of C. albicans to different environmental conditions such as the presence of inducers of hyphal development, various stresses or treatment with antifungal agents (Nantel, 2002; Enjalbert, 2006). They have also been essential in the identification of targets of regulatory circuits (e.g. the regulon repressed by Tup1 –Garcia-Sanchez, 2005).
More recently, we and other groups have used microarray technology in the context of C. albicans infection biology to (1) capture a genome-wide portrait of the transcriptome that mirrors the environmental conditions (thus enabling us to examine the pathogenic processes in detail), (2) identify known genes, signalling pathways and transcription factors involved in pathogenesis, (3) identify new patterns of gene expression and (4) identify previously uncharacterized genes (unknown function genes) that may be associated with infection. As described above, the range of tissue types that C. albicans can infect is extensive; however, generally infections can be grouped into two major types: mucosal or systemic candidosis (also referred to as superficial or invasive infections, respectively). In our laboratory, we use a cyclical approach for dissection of the mechanisms of host–pathogen interactions during distinct types of C. albicans infections and identifying genes involved in the infection process. First, infection models are established and investigated using a combination of microscopic and biochemical techniques to determine the temporal phases of infection. Based on this, appropriate time points are selected and global transcriptional profiling performed. From the expression data, genes of interest are selected, deleted and the resultant knockout mutants tested for attenuation in the infection model in question. If attenuated, further in-depth analysis of gene functions will be performed. In this review, we describe the molecular dissection of three distinct types or stages of infections covering both superficial and invasive disease: oral candidosis (mucosal), bloodstream and liver infection (systemic/invasive).
Mucosal infections: oral candidosis
Recently, we have focused on characterizing oral candidosis and the molecular mechanisms underpinning this type of infection using a combination of different in vitro infection models [e.g. the reconstituted human oral epithelium (RHE) and monolayers of oral epithelial cells] and by comparing these data with results generated from clinical samples (mainly from HIV-positive patients suffering from oral candidosis). Based on our observations made in the experimental infection models, we identified three different substages during the pathogenesis of oral infections: an early/colonization phase, characterized by adhesion of the fungus to the upper layers of the host tissue and fungal proliferation; an invasion phase, associated with hyphal formation and penetration of the upper cell layers of the oral tissue; and a late phase, associated with substantial tissue destruction (Fig. 1a and b). By combining histological analysis, scanning electron microscopy, cellular cytotoxicity measurements and global gene-expression analysis, we were able to study these different stages of oral candidosis at both the cellular and molecular level. Of course, naturally, the time scale of oral candidosis is different. Here, initial colonization can precede the onset of infection by many years and it is only upon compromise of the host that oral candidosis manifests. However, despite the artificiality of the RHE model, it appears to at least partially mimic the clinical setting as the majority of genes constitutively expressed in the RHE were also upregulated by C. albicans infecting the oral cavity of HIV-positive patients (Zakikhany, 2007).
The distinct stages of oral Candida albicans infection. (a) The progression of C. albicans infection of human oral epithelial cells, characterized by attachment, induced endocytosis, active penetration and tissue destruction; (b) representative scanning electron micrographs of the different stages of infection; note the engulfment of fungal cells by epithelial-derived pseudopod-like structures during induced endocytosis; (c) expression profiles of selected genes during infection of the human oral RHE. The colours indicate the degree of (fold) expression: <0.4, green; 0.4–0.7, light blue; 0.7–1.5, beige; 1.5–2, yellow; 2–3, light orange; 3–5, orange; >5, red.
Early-phase oral candidosis
The early phase (0–3 h) in the experimental setting represents the establishment of infection: inoculated C. albicans yeast cells that come into contact with the epithelium adhere to host cells and this contact results in the yeast to hypha transition (Fig. 1a and b). These two events – hyphal formation and adhesion – were also reflected at the molecular level with a number of known genes encoding adhesins or other hyphal-associated genes upregulated during the early phase, including HWP1, FKH1, ATP2, TEF1, ALS3 and SOD5 (Fig. 1c). Following contact with the epithelium, hyphal formation allows tight receptor-mediated contact between fungal and host cells. This interaction in turn results in reorganization of the host cytoskeleton, envelopment of the fungal cell by membrane-derived pseudopod-like structures and subsequent uptake of the fungal cell (Fig. 1a and b). Such a microorganism-triggered epithelial-driven invasion process, known as induced endocytosis, is well described for bacteria such as Salmonella, Shigella or Yersinia (Isberg, 1996; Goosney, 1999; Tran Van Nhieu, 2000) and has, in the case of C. albicans, recently been shown to be mediated by binding of Als3 on the surface of the fungus to oral epithelial cell E-cadherin (Filler & Sheppard, 2006; Phan, 2007). The fact that ALS3 expression is highly induced following C. albicans–epithelial contact corroborates this finding (Fig. 1c). Therefore, during the early attachment phase, C. albicans has already begun epithelial invasion via induced endocytosis.
This early phase of germ tube formation, attachment and induced endocytosis is followed by the invasion phase (3–12 h). The invasion phase is characterized by extensive epithelial penetration via prolific hyphal growth and the expression of several hyphal-associated genes. Although this phase is associated with considerable invasion of the epithelium, substantial tissue damage is not observed until the late phase (12–24 h), suggesting that initial invasion alone is not sufficient to cause damage and not the only factor contributing to tissue destruction. Invasion during the mid and late phases is mediated predominantly by active penetration (Fig. 1a and b). Active penetration is an invasion mechanism distinct from induced endocytosis as it does not rely on the host's cellular machinery, but exclusively on fungal attributes possibly including physical pressure applied by the advancing hyphal tip and the secretion of extracellular hydrolases, which have been presumed to assist in the invasion process via the degradation of host cell components (Schaller, 2005).
Late-phase oral candidosis
Finally, the late phase, characterized by substantial tissue destruction, was – at the transcriptional level – associated with numerous adaptive responses (Fig. 1c). By 24 h, expression of the two alkaline-responsive genes, PHR1 and PRA1, was induced, as was that of the alkaline-responsive transcription factor, RIM101, suggesting that the fungal cells were in an environment of neutral-alkaline pH. Although it is possible that the induction of these genes was influenced by the pH of the RHE maintenance medium, it is probable that these genes are truly induced during oral epithelial tissue destruction, as their transcript levels were >1.5-fold upregulated in patient samples. Candida albicans appears to face a glucose-poor environment in the oral tissue as indicated by the upregulation of the glucose and maltose transporter genes HGT12 and MAL31 and key components of the gluconeogenesis pathway and glyoxylate cycle (PCK1, MLS1 and ICL1). A number of genes involved in amino acid sensing and transport (GNP1, CAR1 and CAR2) were also upregulated suggesting that C. albicans cells sense a nitrogen-poor environment on or within the oral tissue. In addition, strong induction of the high-affinity phosphate transporter PHO84 suggested that ready access to phosphate sources was also limited in this tissue. Apart from the limitation of certain nutrients, the only other clear stress condition encountered by the fungus appeared to be nitrosative stress as indicated by the upregulation of the marker genes YHB5, SSU1 and YHB1 involved in detoxification of nitric oxide (Hromatka, 2005). It is known that epithelial cells produce nitrogen monoxide radicals as part of their innate immune response against microorganisms and it would therefore appear that the experimental RHE model is capable of mounting at least this innate defence mechanism against C. albicans infection. Surprisingly, many well-known genes involved in iron acquisition – a major virulence determinant of virtually all pathogenic microorganisms (Andrews, 2003; Howard, 2004; Sutak, 2008) – were not upregulated in oral RHE tissue. It is possible that the experimental set-up of the RHE allowed exposure of C. albicans to unnaturally high levels of iron from the surrounding medium. Alternatively, C. albicans may be utilizing a novel iron source during invasion of this tissue. Our group has recently demonstrated that C. albicans can utilize iron from host ferritin during oral infections and have shown that this event is mediated by the multifunctional cell surface protein Als3 (Almeida, 2008). Given the high expression of ALS3 during our models of oral infection, this novel iron-acquisition strategy is a likely possibility.
Molecular analysis of an oral candidosis-associated gene
Among the genes that were upregulated during oral infection were a substantial number with no known function or with no homologue in the brewer's yeast Saccharomyces cerevisiae. We reasoned that these unknown function infection-associated genes constitute good candidates for novel virulence factors in C. albicans. One such gene, orf19.7561 (renamed EED1) has no obvious homologues in any other sequenced organism and was upregulated in patient samples and during both the early and the late phases of RHE infection. Given the expression profile of this gene we predicted that it might play a role (1) during the onset of and (2) in the maintenance/persistence of infections of the oral cavity. For functional characterization of the role of EED1 during oral infections, both copies of the gene were deleted. The resultant eed1Δ mutant had severe hyphal-formation defects, growing as yeast or short chains of pseudohyphae under standard laboratory hyphal-induction conditions. Growth in the presence of very strong stimuli (e.g. RPMI with 10% serum) resulted in elongated pseudohyphal germ tube formation of eed1Δ; however, following extended incubation time (5 h), the eed1Δ cells were unable to maintain filamentous growth and switched back to the yeast morphology. These results suggested that EED1 is required for both the initiation and maintenance of filamentous growth. As hyphal development is a prerequisite for invasion of oral epithelial cells, it was predicted that the eed1Δ mutant would be highly attenuated in our oral infection models. Surprisingly, despite the observed in vitro filamentation defects, upon infection of TR146 oral cells the eed1Δ mutant cells switched to filamentous growth, reinforcing the view that fungal contact with epithelial cells itself is a potent inducer of filamentation and bypasses the requirement for EED1. Moreover, these filaments were able to invade the epithelial cells via induced endocytosis. Despite initial filamentation and invasion, by 24 h, eed1Δ had reverted to the yeast morphology and existed within intraepithelial inclusion bodies while wild-type cells disseminated throughout the epithelial tissue via extensive hyphal formation. The gene was therefore named EED1 for epithelial escape and dissemination. Although other genes (EFG1 and CPH1) have been shown to be required for escape from professional phagocytic cells, such as macrophages, EED1 is the first example of a fungal gene required for escape from and dissemination within oral epithelial tissue.
Systemic infections: survival in the blood and liver invasion
Systemic infections are characterized by three major events: dissemination via the bloodstream followed by escape from the bloodstream and the infection of deep-seated organs. In order to study these events, we have established bloodstream and liver infection models. A number of mechanisms as to how C. albicans enters the bloodstream have been proposed; these include so-called ‘natural’ routes, such as via the penetration of epithelial cells at mucosal surfaces, or iatrogenic (artificial) routes, such as implantation of medical devices, surgery, trauma or depletion of the natural microbial flora by antibiotic treatment (Mavor, 2005). Once inside the bloodstream, C. albicans is able to disseminate and can potentially infect almost every organ of the host; however, for this to occur, the fungus must first survive within the bloodstream and then escape via traversal of the endothelium.
Survival in blood
The bloodstream is a harsh environment for any pathogenic microorganism due to the presence of numerous immunoactive cells and molecules; however, in order to cause systemic infections, pathogenic microorganisms must possess mechanisms to resist attack by the immune system. In order to investigate the fungal response to the hostile environment of the blood, we incubated C. albicans with human blood and measured the transcriptional response of the fungus over a time course experiment (10, 20, 30 and 60 min) using microarrays compromising 2002 genes (Fradin, 2003). In this study, we showed that C. albicans rapidly adapts to the blood environment. Such rapid adaptation relies on the expression of a distinct subset of genes and translation into the corresponding proteins necessary to meet the requirements of the new environment. This was reflected by the strong upregulation of genes related to protein synthesis at 10 min. This early adaptation event was followed by the upregulation of genes involved in the glyoxylate cycle, fermentation, glycolysis and response to oxidative stress. In addition, several known hyphal-associated genes were upregulated upon exposure to blood, in agreement with microscopy observations that 42% of cells had undergone the yeast to hypha transition during the time course of the experiment (Fig. 2a). The fact that genes of the glyoxylate cycle were upregulated at the same time as genes involved in glycolysis is surprising: in general, the glycolytic pathway and the glyoxylate cycle are utilized in the absence or presence of carbohydrates, respectively, and not at the same time by the same cell. One explanation for this is that the fungal cells existed in at least two distinct subpopulations, one with and one without ready access to sugars. This split-population hypothesis may also be supported by the morphogenic heterogeneity of fungal cells exposed to blood and may be explained by the fact that blood consists of a heterogeneous mixture of different cell types, which may act differentially on distinct fungal cells.
Interaction of Candida albicans with blood. (a) The effect of blood or blood fractions on C. albicans morphology, note the heterogenous morphology of fungal cells in whole blood, repression of hyphal formation in the PMN (neutrophil) fraction and hyphal development in the MNC (monocyte/lymphocyte) and red blood cells (RBC) (erythrocyte) fractions; (b) representative micrographs of human blood cells; (c) expression profiles of selected genes following 30-min incubation with indicated blood component. The colours indicate the degree of (fold) expression: <0.4, green; 0.4–0.7, light blue; 0.7–1.5, beige; 1.5–2, yellow; 2–3, light orange; 3–5, orange.
In order to dissect which blood factors are involved in combating C. albicans, and which fungal factors may resist this assault, we further analysed the cellular and transcriptional response of C. albicans to whole blood and to various blood fractions enriched in particular host cell types (Fradin, 2005). One major virulence trait – the yeast to hypha transition – is also presumed to aid in C. albicans escape from the bloodstream by assisting in traversal of the endothelial lining of blood vessels (Filler & Sheppard, 2006) and has been shown to mediate escape following phagocytosis by macrophages (Lorenz & Fink, 2001). The morphology of C. albicans exposed to the various blood fractions for 30 min was therefore determined. In blood fractions lacking neutrophils (erythrocyte, monocyte/lymphocyte and plasma fractions) most cells (80–85%) formed germ tubes. However, exposure of C. albicans to the polymorphonuclear (PMN) neutrophil fraction almost completely blocked hyphal development with 96.5% of cells remaining in the yeast morphology (Fig. 2a). This dominant effect of PMNs on the morphology of C. albicans was also reflected at the transcriptional level: cluster analysis showed that cells incubated in whole blood or in the PMN fraction shared a similar profile that was distinct from cells incubated in the presence of plasma, erythrocytes, monocytes or in whole blood depleted of PMNs.
In the absence of neutrophils, numerous genes associated with protein synthesis, glycolysis and hyphal formation were expressed, suggesting that cells were metabolically active and in concordance with observation that the majority of cells had undergone the yeast to hypha transition. In stark contrast, in the presence of neutrophils, fungal cells underwent growth arrest, and hyphal morphogenesis was almost completely blocked. Nutrient starvation certainly appeared to contribute to the observed growth arrest. The environmental nitrogen level of cells exposed to neutrophils was low as indicated by the upregulation of the ammonium permeases MEP2 and MEP3. Furthermore, the amino acid starvation-responsive transcriptional regulator GCN4, as well as several genes associated with arginine, leucine, lysine and methionine biosynthetic pathways were induced as described previously by Rubin-Bejerano (2003). Vacuolar proteases (Prb1, Prb2 and Apr1) and carboxypeptidases (Prc1 and Prc2) – known to be involved in the utilization of endogenous nitrogen sources – were also upregulated in response to neutrophils. In addition to this clear nitrogen-starvation response, C. albicans also appeared to face a carbohydrate-poor environment, as genes associated with glycolysis were downregulated while components of the glyoxylate cycle were strongly induced (Fig. 2c).
One of the proposed antimicrobial mechanisms of neutrophils is the production of reactive oxygen species, which contribute to the killing of pathogenic microorganisms by attacking multiple cellular components such as DNA, proteins and lipids. It appears that neutrophils exert substantial oxidative stress on C. albicans as a number of genes involved in detoxification of reactive oxygen species – such as superoxide dismutases, catalase, glutathione peroxidase, glutathione reductase, glutathione S-transferase and thioredoxin – were strongly induced in response to incubation with neutrophils (Fig. 2c). To further investigate the role of oxidative stress on C. albicans during interactions with neutrophils we fused a green fluorescent protein (GFP) reporter to the oxidative stress-responsive SOD5 promoter (Martchenko, 2004). The SOD5 reporter was found to be induced in yeast cells that were either attached to or phagocytosed by neutrophils. This finding suggests that neutrophils elicit an oxidative stress response, which does not rely on phagocytosis of the fungal cell. Finally, it was shown that sod5Δ mutant cells had reduced survival following incubation with neutrophils, reinforcing the view that an appropriate oxidative stress response is critical for fungal survival in the hostile environment of the blood and subsequent dissemination throughout the body.
Liver invasion
Having survived the bloodstream for a certain time period, C. albicans cells must escape from the circulation via traversal across the endothelial lining of blood vessels in order to avoid the killing activities of blood cells. Following escape from the blood stream, the fungus has the potential to infect virtually every internal organ. As a model of infection of a deep-seated organ we analysed C. albicans invasion of the liver. In order to identify genes associated with tissue invasion of organs, we designed two comparative experiments (Thewes, 2007a, b): (1) comparison of C. albicans gene expression following in vivo intraperitoneal infection of mouse liver and ex vivo infection of perfused pig liver to identify genes commonly associated with infection of this organ; (2) analysis of genes expressed in the liver model by the invasive strain (SC5314) but not by the noninvasive strain (ATCC10231) to more stringently define those genes actually involved in the invasion process, as opposed to those that are simply more highly expressed in the host than in broth culture. Preliminary histological analysis revealed that in both infection models, the infection process of the invasive strain was characterized by initial attachment and hyphal formation followed by penetration of the liver capsule and invasion into the tissue by hyphal cells (Thewes, 2007b; Fig. 3a and b).
Candida albicans invasion of liver tissue. (a) Candida albicans infection of in vivo mouse or ex vivo pig liver, characterized by initial attachment and hyphal formation, superficial invasion of the liver capsule followed by extensive invasion into deeper liver tissue; (b) representative histological micrographs of C. albicans liver infection; (c) expression profiles of selected genes during infection of the mouse liver at the time of attachment (0.5 h), superficial (3 h) and deeper (5 h) invasion. The colours indicate the degree of (fold) expression: 0.4–0.7, light blue; 0.7–1.5, beige; 1.5–2, yellow; 2–3, light orange; 3–5, orange; >5, red.
Based on the expression profile of so-called ‘marker genes’, a number of inferences could be drawn about the nutritional state of the environment encountered by C. albicans during liver invasion (Thewes, 2007a; Fig. 3c). For example, as opposed to the situation in the oral cavity, there appeared to be sufficient access to sugar for utilization as a carbon source as indicated by upregulation of PFK1 (encoding a key enzyme of glycolysis), PDA1 and PDX1 (involved in acetyl-CoA biosynthesis) and KGD1 and KGD2 (encoding key enzymes of the tricarboxylic acid cycle). Similarly, there appeared to be sufficient levels of nitrogen in the liver as genes associated with amino acid starvation were not upregulated. In spite of this apparent abundance of carbon and nitrogen sources, C. albicans appeared to face severe iron limitation upon liver invasion as indicated by the upregulation of FET5, FTR1, ZRT1, CFL1, RBT5, FRE5 and CTR1– all genes associated with an iron-poor environment or with iron acquisition. There did not appear to be any specific response to oxidative, nitrosative or osmotic stress in the liver, although genes associated with general/thermal stress such as HSP78, HSP90, HSP104, HSP12 and HSP70 were upregulated. Whether this is due to actual thermal stress or cross-protection against an as-yet unknown stress in the liver remains to be determined. Upregulation of well-known hyphae-associated genes such as SAP5, ALS3 and HWP1 was in accordance with the histological observation that cells were growing predominantly in the hyphal morphology. Finally, upregulation of the alkaline-responsive PHR1 indicated that the majority of cells encountered a neutral-alkaline pH.
Molecular analysis of a gene involved in invasive candidosis
The next step of this study involved dissection of the transcriptional data to identify genes intimately associated with the process of liver tissue invasion. To focus our analysis in this direction, we looked for genes that were upregulated by the invasive SC5314 during both mouse (3 and 5 h postinfection) and pig (12 h postinfection) liver invasion but not by the noninvasive strain, ATCC10231. This detailed analysis yielded five genes, one of which (DFG16) was selected for further analysis. DFG16 encodes a member of the PalH/RIM21 super-family, which constitutes putative pH sensors and has been shown to function in the Rim101 pathway (Barwell, 2005). To characterize the function of this gene, both alleles were deleted and the resultant dfg16Δ mutant tested. In accordance with the protein's predicted role as a pH sensor, dfg16Δ behaved normally at acidic pH but was unable to grow under iron- or phosphate-limited conditions at alkaline pH. Moreover, dfg16Δ had reduced osmotic stress tolerance at alkaline, but not acidic, pH and failed to form filaments at pH 8. Finally, the virulence of dfg16Δ was attenuated in a mouse model of haematogenously disseminated candidosis (Thewes, 2007a) and dfg16Δ cells had reduced potential to invade liver tissue following intraperitoneal infection (our unpublished data).
The fact that transcript levels of DFG16, encoding a putative pH sensor, are increased during liver invasion underscores the importance of environmental pH sensing and adaptation during the progression of systemic candidosis. Unlike environmental human fungal pathogens, which receive clear host-associated signals to initiate infection (e.g. a shift to 37 °C), C. albicans must be able to dramatically reprogramme its behaviour based on more subtle environmental cues. Probably the most extensively studied trait of C. albicans, the yeast to hypha transition, is under the control of an extensive network of signalling pathways, which integrate the receipt of a wide range of environmental signals (temperature, pH, oxygen, CO2, nutrients and physical contact to name only a few) to control cellular morphology. Why must C. albicans sense such a vast array of environmental signals to determine its morphology? Firstly, unlike many environmental pathogenic fungi, which form either nonpathogenic/saprophytic or host-associated/pathogenic morphologies, the pathogenicity of C. albicans relies on the reversible yeast to hypha transition (mutants unable to form yeast also have reduced virulence). Secondly, because C. albicans is continuously in contact with the host (even in the nonpathogenic stage as commensal) and, therefore, in an environment of physiological temperature, the formation of a single morphological state in response to temperature alone would be inappropriate. Thirdly, although the gross morphology of C. albicans is dependent on certain, sometimes quite different, combinations of environmental signals, any given subset of signals may specifically result in the expression of a different subset of genes not strictly coexpressed with a given morphology; these discrete transcriptomes reflecting the given requirements encountered at particular anatomical niches: for example, RBT5, encoding the cell surface-localized haemoglobin-binding protein (Weissman & Kornitzer, 2004), is highly expressed by hyphal cells invading liver tissue but not by hyphal cells invading oral tissue.
Given the highly dynamic response of C. albicans to its environment, combined with the diverse niches it occupies, it is not surprising that the genome of this fungus contains a large number of genes encoding described and putative sensors. The correct sensing of – and response to – environmental signals is crucial and relies on the expression of relevant sensor-encoding genes in a given microenvironment. The reduced virulence of dfg16Δ demonstrates that perturbations in the sensing of a single environmental factor (pH) can block the pathogenicity of C. albicans and illustrates the delicacy of the cross-talk between the host and the pathogen.
Discussion and outlook: the future of C. albicans infection models and identification of infection-associated genes
In summary, we have presented three examples of how, using carefully designed infection models combined with global gene expression analysis, we can confirm presumed features of pathogenicity, identify genes that are expressed during infection and discover novel aspects of host–pathogen interactions during representative C. albicans infections.
The transcriptome of C. albicans as a tool to explore the physiological environment during infection
One possible application of genome-wide profiling of a pathogenic microorganism during infection is the concept of the transcriptome as a biosensor, which may enable us to monitor the physiological conditions encountered by the microorganism. For example, we verified that when C. albicans was exposed to blood, neutrophils elicit a strong oxidative stress response. This was not the case in the liver or oral cavity, where general or nitrosative stress responses, respectively, dominated, thus demonstrating that C. albicans faces diverse challenges depending on the anatomical niche in which it finds itself. On the other hand, some features of the host environment appeared to be common and there was significant transcriptional overlap between different infection types. For example, key components of the glyoxylate cycle were upregulated during both oral and bloodstream (but not liver) infections. Similarly numerous hyphal-associated genes were upregulated in both oral and liver infections, but not during interaction with neutrophils. Interestingly, very few genes were strongly induced under all infection conditions, suggesting that, in general, there is no general response to infection-associated growth and that the transcriptome of C. albicans within the human host is overall niche specific. This may be explained by the commensal lifestyle of C. albicans. Environmental fungi receive clear signals upon the onset of infection. For example, synthesis of one of the major virulence factors of C. neoformans, the polysaccharide capsule, is induced upon infection via the fungus sensing a shift to an environment of iron limitation and physiological temperature (Jung & Kronstad, 2008). Candida albicans on the other hand, outside of the laboratory, is in constant contact with the human host and while colonizing mucosal surfaces is constitutively exposed to an environment of physiological temperature and limited iron. Only disturbance of the normal bacterial microbiota and/or a weakened immune system triggers activity associated with infection and the colonization of new niches within the human body. Given the niche specificity of C. albicans, the principle of using the transcriptome as a biosensor may be applicable for studying the conditions faced by the fungus at diverse body sites. For example, based on detailed in vitro studies, the expression of certain marker genes such as PHR1, PRA1, PHR2, SOD5, CAT1, YHB1, ICL1 and MEP2 may indicate alkaline or acidic pH, oxidative stress, nitrosative stress, low glucose or nitrogen conditions, respectively (De Bernardis, 1998; Sentandreu, 1998; Lorenz & Fink, 2001; Martchenko, 2004; Biswas & Morschhauser, 2005; Corvey, 2005; Hromatka, 2005). However, further in vitro gene expression experiments are needed to identify not only single marker genes, but also sets of genes (regulons) and signatures associated with a certain physiological situation. For example, the identification of particular transcriptional signatures associated with growth on defined nutrient sources could be used to overlay the transcriptome during infection and may actually pinpoint what nutrients are being utilized in vivo. Bignell and colleagues (McDonagh, 2008) have recently used such an approach to study the initiation of infection during pulmonary aspergillosis. By comparing transcriptional signatures obtained from specific in vitro conditions to the expression pattern observed upon infection of the murine lung, they demonstrate that A. fumigatus adaptation to the host is associated with iron limitation, alkaline stress and nutrient deprivation.
Unknown function genes
One major obstacle in interpreting such global studies is the overall lack of annotation of a large set of genes in the C. albicans genome. Although d'Enfert (2005) assigned ‘tentative functional assignments’ for 92% of the C. albicans genome, the true number of genes with known function is probably much lower. Braun (2005) reported gene ontology (GO) terms (excluding unknown function) for only 56% of the genome. Moreover, 19% of genes do not share significant sequence homology with other organisms (Braun, 2005). Because C. albicans is highly specialized to exist in association with warm-blooded mammals, it is reasonable to hypothesize that C. albicans possesses unique factors involved in interactions with its host environment. Although these host-interaction factors likely evolved to maintain a commensal life-style, when the host environment undergoes certain changes (antibiotic treatment, immune compromise, etc.), they may become virulence factors involved in the onset and progression of invasive infections. Therefore, one of our central premises is that unknown function genes transcriptionally upregulated during infection constitute promising host-interaction/virulence factors, required for the pathogenesis of C. albicans. For example, we have shown that EED1, a previously undescribed gene with no obvious sequence homology in any other sequenced organism is induced during in vivo and in vitro oral infections and was essential for dissemination within oral tissue (Zakikhany, 2007). This example therefore illustrates that there exist genes of unknown function, induced during infection, which are essential for the pathogenesis of C. albicans. We are therefore continuing to explore the role of unknown function genes in the pathobiology of C. albicans.
Experimental design
A major issue to consider during the interpretation of current transcriptional profiling studies is the choice of control. In our group we cohybridize experimental RNA against a ‘common control’ of RNA from cells grown in YPD at 37 °C to mid-logarithmic phase. The rationale behind this is that virtually all genes are expressed under this condition, and many infection-associated genes should be more highly expressed during infection than in batch culture. However, care should be taken when using this approach as ‘upregulation’ is dependent on the relative expression in the control and thus genes expressed at the same or even lower level in vivo compared with YPD may still be important for the infection process.
Recent work by Walker (2009) analysing gene expression of C. albicans infecting the rabbit kidney further highlighted certain issues associated with the choice of control. The authors report a strong downregulation in the expression of well characterized and reportedly ‘hyphae-associated’ genes such as ALS3 and ECE1, despite histological observations suggesting hyphal growth in the kidney lesions. However, in this study the control chosen was RNA from cells grown in RPMI 1640 at 37 °C and, as the authors discuss, the downregulation of these genes does not necessarily mean that they are not being transcribed in vivo, but simply that the expression is significantly higher in the control cells.
Two strategies can be used to circumvent these issues. The first is to design a temporal experiment where the transcriptome is captured at different time points during the infection process including the preculture. The dynamics of expression can then be analysed over time, effectively independent of the control signal. The second approach is to perform parallel experiments where only one variable is changed between the two, for example, the comparison of C. albicans infecting reconstituted human epithelium either with or without the presence of neutrophils (Schaller, 2004). It should be noted that, for some experimental designs, one colour labelling, may be more appropriate. Readers are directed to two excellent reviews describing the design principles of microarray experiments (Yang & Speed, 2002; Bryant, 2004).
Although microarrays represent the current ‘gold standard’ for unbiased global analysis of gene expression in C. albicans, other technologies may be more appropriate for certain applications. For example, a number of reporter systems have been described that can monitor expression of single genes during models of infection. For example, in vivo expression can be measured very sensitively by placing the site-specific recombinase FLP under control of the promoter of interest. Using this method, transient expression of the gene of interest can be detected during animal models of infection at the single cell level (Staib, 1999). A second single cell profiling approach involves fusion of the promoter of interest to a GFP reporter. This method is particularly effective for monitoring gene expression in a mixed population of cells as demonstrated by Barelle (2006).
Finally, although the C. albicans genome now stands at its 21st assembly (van het Hoog, 2007), it is likely that numerous small transcripts, potentially involved in pathogenesis, have been overlooked and, therefore, not included on current microarrays. This and the above-discussed problems may be overcome by serial analysis of gene expression of C. albicans during infection as has recently been described for C. neoformans infecting the murine lung (Hu, 2008). Using this approach, it will be possible to determine which genes are actually expressed at high levels during infection and to uncover truly novel transcripts involved in the infection process.
The future of C. albicans infection models
The past half decade has witnessed the establishment of numerous postgenomic technologies and refined infection models for the study of C. albicans, and this review has described the use of such technologies to characterize the behaviour of this fungus in different niches of the human host. However, this is only part of the story, as all forms of candidosis (whether superficial or invasive infections) are dependent on the host status and the host response must always be considered as an element of pathogenesis (Richardson & Rautemaa, 2009). Therefore, we need to also investigate the specific host responses for the different types of infection and the different stages of disease development to understand the underlying mechanisms that contribute to C. albicans infections.
Based on infection models such as those presented in this review, it should be possible to extend current studies on C. albicans host–pathogen interactions to more comprehensively cover infection-associated parameters: both fungal pathogenicity mechanisms and host immune factors. For example, we have shown that in blood, it is neutrophils that have the greatest impact on C. albicans morphology, viability and gene expression (Fradin, 2003, 2005) lending direct experimental evidence to clinical observations that neutrophils are the primary defence mechanism against systemic candidosis (Chauhan, 2006). As an example of extension of such a study, Fradin (2007) went on to characterize the transcriptional response of neutrophils exposed to C. albicans, identifying an enrichment in genes involved in proinflammatory cell–cell signalling. And while the innate immune system is vital for the first line of protection against systemic candidosis, T-cell-mediated cellular immunity is generally considered particularly important at oral mucosa (reviewed in Saunus, 2008). Recently, Schaller and colleagues (Schaller, 2004; Weindl, 2007) have utilized the RHE model to show that neutrophils induced a protective T-helper type 1 immune response in human oral epithelial cells and that this protection is directly mediated by TLR4 receptors.
The future of C. albicans infection biology will surely extend such approaches as those reviewed here, simultaneously integrating analysis of pathogen and host factors to assemble both a detailed and global picture of host–pathogen interactions. The inherent complexity of these interactions means that systems biology must also play a role in shaping our understanding of this delicate cross-talk. Attaining such an inclusive portrait of candidosis will unquestionably drive forward the discovery of novel diagnostic tools and the development of effective antifungal therapies.
Acknowledgements
Our own research was supported by the Robert Koch-Institute (RKI), the Hans-Knoell-Institute (HKI), the Deutsche Forschungsgemeinschaft (DFG), the European Union (EU), and the Federal Ministry of Education and Research (BMBF). We would also like to thank the reviewers of this manuscript for their helpful suggestions and Brice Enjalbert for interesting discussions on data interpretation.
References
Author notes
Editor: Teun Boekhout
