|Home | About | Journals | Submit | Contact Us | Français|
The study of fungal regulatory networks is essential to the understanding of how these pathogens respond to host environmental signals with effective virulence-associated traits. In this study, a virulence-associated DEAD-box RNA helicase–encoding gene (VAD1) was isolated from a mutant defective in the virulence factor laccase. A Δvad1 mutant exhibited a profound reduction in virulence in a mouse model that was restored after reconstitution with WT VAD1. Loss of VAD1 resulted in upregulation of NOT1, a gene encoding a global repressor of transcription. NOT1 was found to act as an intermediary transcriptional repressor of laccase. Vad1 was located within macromolecular complexes that formed cytoplasmic granular bodies in mature cells and during infection of mouse brain. In addition, VAD1 was shown by in situ hybridization to be expressed in the brain of an AIDS patient coinfected with C. neoformans. To understand the role of VAD1 in virulence, a functional genomics approach was used to identify 3 additional virulence determinants dependent on VAD1: PCK1, TUF1, and MPF3, involved in gluconeogenesis, mitochondrial protein synthesis, and cell wall integrity, respectively. These data show that fungal virulence-associated genes are coordinately regulated and that an analysis of such transcriptomes allows for the identification of important new genes involved in the normal growth and virulence of fungal pathogens.
The basidiomycetous yeast Cryptococcus neoformans has emerged as one of the major causative agents of meningoencephalitis in immunocompromised hosts, such as persons with AIDS, organ transplant recipients, and patients receiving high doses of corticosteroid treatment, although systemic infections can also occur in immunocompetent individuals (1, 2). However, the spectrum of disease in AIDS has been changing with the advent of highly active antiretroviral therapy with the recognition of an immune reconstitution syndrome associated with cryptococcosis (3). As rates of infection have diminished in developed countries, attention is increasingly being focused on the high rates of cryptococcosis in the developing countries of Africa and Asia where cryptococcosis was found to account for an estimated 17% of AIDS-related deaths in one cohort in Uganda (4). C. neoformans has been divided into 4 serotypes, A–D, of which A and D cause the vast majority of infections in Europe and the United States (5).
Molecular and immunological studies over the past years have provided numerous insights into the basis of virulence of this yeast. The 3 best-known virulence-associated traits of C. neoformans include: (a) its ability to grow at 37°C dependent on calcineurin (6, 7); (b) the presence of a copper-containing laccase that is capable of producing melanin pigments (8, 9); and (c) a high molecular-weight polysaccharide capsule (10–15). In addition, genetic knockout experiments have established the role of a number of other traits required for survival and virulence of C. neoformans, including purine metabolism (16), myristylation (17), signal transduction (18–20), α-mating type locus (21), mannitol synthesis (22, 23), phospholipase (24), urease (25), and manosylation activities (26). Despite this extensive repertoire of identified virulence-associated genes, little is known about how the organism regulates pathogenesis by organizing the expression of these genes.
One approach for the study of virulence regulation has been to model cryptococcal virulence on signal transduction pathways evident in the model yeast, Saccharomyces cerevisiae. For example, the MAPK pathway has been studied in both serotype A and serotype D C. neoformans because of its known role in mating and activation of filamentous growth for food foraging under starvation conditions in S. cerevisiae. However, while many of the MAPK members have profound effects on mating in C. neoformans (GPB1, STE7, STE11, and CPK1), only 2 (STE12, STE20) have been implicated directly in virulence-related gene regulation, and this was found to be strain dependent (27–29). The variability in the penetrance of this pathway in Cryptococcus has led one investigator to propose that STE12 functions in parallel to the MAPK cascade in this fungus by an additional unknown pathway (27). A second signaling pathway implicated in virulence of C. neoformans has been the nutrient-sensing Gα protein cAMP-protein kinase A pathway, although the role of this pathway in virulence also appears to be strain dependent (18, 30, 31).
To identify additional regulatory pathways not evident from the study of model yeasts, we employed random insertional mutagenesis and plasmid rescue in C. neoformans and screened for mutants defective in the virulence factor laccase. Plasmid recovered by excision from genomic DNA of one such mutant showed that the disrupted gene encoded a member of the RCK/p54 subfamily of RNA DExD/H-box proteins, which was subsequently named virulence-associated DEAD-box RNA helicase-encoding protein (Vad1). Members of this family of proteins have been identified as components of the CCR4-NOT complex, a global regulator of transcription and mRNA stability (32). Further studies showed that VAD1 is an important regulator of stress response and plays a role in the expression of several new virulence determinants involved in diverse traits, such as stress-responsive growth, salt tolerance, and virulence, each accounting for a portoin of the Δvad1 phenotype.
A laccase-deficient mutant was produced by insertional mutagenesis using a pMUT8 plasmid derived from a URA5 transformation marker inserted into pBluescript as described previously (33). A 7-kb plasmid that contained a fragment of the interrupted gene region was rescued from the genomic DNA of the cryptococcal mutant. Through the use of plasmid sequence, the National Center for Biotechnology Information (NCBI) database, and the alignment tool, BLAST, the interrupted gene was found to encode a putative 616 AA polypeptide and to display homology with a conserved subfamily of RNA-binding DExD/H-box proteins referred to as RCK/p54, after the human protein (34). Members within this family include the 485 AA Ste13 of Schizosaccharomyces pombe (74% identity, e–180), the 506 AA Dhh1 from S. cerevisiae (71% identity, e–170), and a 484 AA human p54 (66% identity, e–164). Members of this class of proteins have been implicated in nuclear functions (35) as well as in modulation of the cytoplasmic mRNA decapping and deadenylation complex (36). The cryptococcal protein is composed of a highly conserved DExD/H-box region of 400 AAs (74% identity to Ste13; 72% identity to Dhh1) and a less conserved glutamine-rich (27%) 200-AA C-terminus (see Supplemental Figure 1; supplemental material available online with this article; doi:10.1172/JCI200523048DS1.). The C-terminus of other proteins of this class are also glutamine-rich (Dhh1-26%; Ste13-12%), but are much shorter (100 AA) than the cryptococcal protein. Differences between sequences of the C-terminus of the cryptococcal protein and those of others of the class may suggest functions specific to the cryptococcal protein. Because of these structural differences, we decided not to name the gene after the S. cerevisiae gene, DHH1. Rather, because of the association with the virulence factor laccase, the cryptococcal protein was named Vad1.
In addition to typical yeast traits, C. neoformans expresses several factors that are involved in its role as a human pathogen and were assessed in the VAD1 strains. As shown in Figure Figure1A,1A, laccase activity and transcription were reduced in the Δvad1 mutant and were restored after complementation with WT VAD1. In contrast, capsule and urease activity were unaffected by the mutation. The VAD1 strains were next assessed for virulence in a mouse dissemination model because this model was previously used to establish a role for laccase in the virulence of C. neoformans (9). Despite the presence of residual laccase activity, the Δvad1 strain showed marked attenuation in virulence in this model even after a large inoculum of 106 organisms (P < 0.001; Figure Figure1B).1B). The degree of hypovirulence was surprising, given that previous studies of strains completely lacking laccase showed minor residual virulence (9). This suggests that the Δvad1 mutant exhibits defects other than laccase expression that impair virulence. Reconstitution of the Δvad1 mutant with WT VAD1 completely restored virulence, confirming that the alteration of virulence was not due to a secondary, unintended mutation.
Analysis of the Δvad1 mutant showed that this gene shares traits associated with the RCK/p54 subfamily but also possesses features unique to C. neoformans. Growth rates of the mutants in yeast extract–peptone–dextrose (YPD) broth were virtually identical at 30°C and 37°C (doubling time in hours at 30°C: 2.5, 2.5, 2.5; at 37°C: 2.3, 2.5, 2.3 for WT, Δvad1, and Δvad:VAD1 strains, respectively) and on asparagine salt agar with 2% glucose (Figure (Figure2A),2A), but the Δvad1 mutant showed significantly reduced growth on the same agar media supplemented with 8 mM caffeine or when the carbon source was replaced by the nonfermentable substrates glycerol or lactate, with the mutant forming petite colonies on the latter (Figure (Figure2A).2A). The doubling time of the WT and Δvad1 mutant in medium containing lactic acid as the sole source of carbon was 8.8 hours (± 1.2) and 12.85 hours (± 0.35) respectively, providing verification of the spot-plate data. In addition to exhibiting increased doubling time, the Δvad1 mutant grew to a lower final density (OD600 = 0.19) than the WT (OD600 = 0.23) in medium supplemented with lactic acid. Reduced growth was also observed in media containing 1M NaCl and 1.8 M sorbitol (data not shown). Sensitivity to salt, sorbitol, and caffeine are indications of a defect in cell wall integrity (32). Cell wall defects and reduced ability to utilize nonfermentable substrates are phenotypes shared by Δvad1 mutants and mutants of the S. cerevisiae RCK/p54 member DHH1 (50, 74). However, the Δvad1 mutant was different from that of its closest homolog, S. pombe STE13 (37), in that it retained the ability to mate and produce basidiospores (Figure (Figure2B),2B), although less vigorously than the WT. Reconstitution of the mutant with WT VAD1 restored all defective phenotypes.
To facilitate immunolocalization of the Vad1 protein, a 10-AA c-myc tag was inserted within the WT VAD1 gene at AA position 58 and used to complement the Δvad1 mutant. Proper functioning of the VAD1–c-myc construct was verified by restoration of laccase expression in the reconstituted Δvad1:VAD1–c-myc strain (Figure (Figure3A,3A, bottom panel). Western blot analysis using a c-myc monoclonal antibody showed a single band on SDS-PAGE corresponding to the expected molecular mass of the Vad1 protein (Figure (Figure3A,3A, top panel, lane 2), which was not evident on cell extract from the Δvad1 mutant (Figure (Figure3A,3A, top panel, lane 1). Cell extract of cryptococcal cells expressing the VAD1–c-myc construct was subjected to gel filtration on a TSK-GEL G6000 (Supelco Chromatography) high-performance liquid chromatography (HPLC) column and analyzed by dot blot hybridization of column fractions. As shown in Figure Figure3B,3B, comparison of elution profiles of the c-myc reactive material to molecular mass standards showed the presence of a large Vad1 multiprotein complex migrating at a size predominately at 0.5–2.0 mDa. The finding of large multiprotein complexes in vitro suggests that the Vad1 protein is part of a large regulatory complex, which is consistent with the functions of other members of the RCK/p54 subfamily. DHH1, the S. cerevisiae homolog of VAD1, has been shown to interact with members of the CCR4-NOT regulatory complex (32).
To determine the cellular localization of Vad1, the Δvad1 strain was complemented with a WT VAD1 fragment containing a C-terminal GFP tag. To mimic WT expression, the construct was expressed under the control of the native VAD1 promoter and was inserted into the genome as determined by Southern blots of uncut DNA, which localized the construct to genomic DNA (data not shown). Complementation of the Δvad1 mutant with the VAD1-GFP construct restored laccase expression, confirming the functionality of the fusion protein (data not shown). During active budding in the presence of glucose, immature daughter cells that did not yet contain nuclei observable by DAPI staining showed accumulation of Vad1 in the cytoplasm, seen as a diffuse fluorescence specific to cells expressing GFP-Vad1 (Figure (Figure3C).3C). Once daughter cells matured and nuclei were observed, Vad1 was consistently found in cytoplasmic locations in structures similar to the recently described P-bodies containing the RCK/p54 submember of S. cerevisiae and other gene products involved in transcript stability (38). The cytoplasmic granular appearance of Vad1 was then maintained during glucose starvation, although the granules appeared slightly larger than those of nonstarved mature cells (Figure (Figure3C).3C). Untransformed WT cells showed no specific cellular localization and weak autofluorescence that was evident at multiple wavelengths, whereas Vad1-GFP fluorescence was only visible at GFP-specific wavelengths. Vad1-GFP fluorescence never colocalized with DAPI, which suggests that it does not exhibit nuclear localization, making a role for VAD1 in nuclear DNA-binding unlikely. This is in contrast to what has been reported for the human homolog p54 (39). While the RNA helicase proteins are numerous and typically have diverse housekeeping functions, the recently described RCK/p54 subfamily appears to have a more restricted, possibly regulatory function (32). It has been proposed that human and mouse p54 RCK/p54 homologs are protooncogenes (39). The S. cerevisiae homolog, Dhh1, has been proposed as having a role in transcription due to its genetic and physical interactions with the regulatory CCR4-NOT complex (40, 41). In addition, Dhh1 has also been found to have a role in transcript stability by physically interacting with and modulating the function of the mRNA decapping complex (36, 38). These data would suggest a role for VAD1 in cytoplasmic processes such as RNA stability as proposed for yeast Dhh1 (36) rather than a direct role in transcription initiation.
To assess the expression and cellular localization during neuropathogenesis, cells expressing the Vad1-GFP fusion protein were inoculated into mice and subsequently recovered from mouse brains (colony counts from brain at time of sacrifice: 70,600 ± 4,900 cfu/g; n = 3). As shown in Figure Figure3D,3D, cells recovered from mouse brains during infection (n = 200) showed expression of Vad1 protein, and all cells were observed to have the granular appearance of mature cells, indicating a predominance of the mature cell type during infection of the mouse brain (right panels). WT cells exhibited only nonspecific autofluorescence (left panels). The observance of predominantly mature cells during brain infection suggests that survival of C. neoformans in brains is primarily due to the fungus’s resistance to killing by the host rather than by rapid fungal growth and effective host killing.
To validate the study of VAD1 as it relates to virulence, in situ hybridization was performed on sections of human brain infected with C. neoformans. The patient was a 42-year-old male who presented with mental status changes and was found to have a CSF cryptococcal antigen of 1:16,348 and a computed tomography (CT) scan suggestive of cerebral edema. The patient was subsequently found to be HIV positive with a CD4+ T cell count of 51 cells/mm3, and in spite of 1 dose of Amphotericin B, became unresponsive and died. On autopsy, evidence of effacement of the uncus was found consistent with brain herniation. CSF cultures grew C. neoformans, and brain sections showed numerous mucicarmine-stained yeasts. Figure Figure44 shows an autopsy specimen from the cerebral cortex meninges. In situ hybridization was performed with an antisense RNA strand or a sense strand (negative control) of a 579 bp fragment encoding a nonconserved region of the C-terminus and a 3′ untranslated region of VAD1. Positive hybridization signal was detected only in sections probed with an antisense probe (Figure (Figure4A).4A). Sections probed with the corresponding sense probe showed no hybridization to yeast forms (Figure (Figure4B).4B). Sections were also stained with mucicarmine specific for C. neoformans capsule (Figure (Figure4C)4C) and H&E (Figure (Figure4D).4D). These data demonstrate the expression of VAD1 during infection of the human brain by C. neoformans and provide rationale for further elucidation of the mechanisms by which loss of VAD1 results in severe virulence attenuation.
Due to the pleiotrophic phenotype of the Δvad1 mutant as well as the large degree of virulence attenuation with the mutation, a search was made for other genes, in addition to laccase, that exhibit altered expression in the Δvad1 mutant. WT and Δvad1 cells were subjected to glucose starvation to derepress laccase expression, and mRNA was compared in duplicate by 24-primer differential display. Four genes, including laccase, showed reduced expression under glucose-starvation conditions, and 1 gene showed increased expression in the Δvad1 mutant. The contribution of each to the Δvad1 phenotype was assessed by construction of genetically altered strains.
A differential display tag found to be highly enriched in the Δvad1 strain relative to the WT exhibited identity to annotated fragment 179.m00475 of The Institute of Genomic Research (TIGR) Cryptococcus neoformans database. A BLAST search revealed homology to the S. cerevisiae NOT1 gene, encoding a global negative regulator of transcription, and the core component of the CCR4-NOT complex, of which the VAD1 homolog, DHH1, is also a component. In the WT, the NOT1 transcript exhibited a moderate level of glucose repression under the conditions tested but was verified to be upregulated in the Δvad1 mutant by Northern blot (Figure (Figure5A).5A). To determine the contribution of NOT1 mRNA accumulation to the Δvad1 laccase-deficient phenotype, an episome expressing interfering RNA to the NOT1 transcript was introduced into the Δvad1 mutant. RNA interference (RNAi) was the method of choice since the S. cerevisiae NOT1 is an essential gene and deletion of any 2 components of the CCR4-NOT complex results in a synthetic lethal phenotype. Expression of NOT1-interfering RNA restored laccase activity to the Δvad1 mutant as measured on asparagine medium lacking glucose and supplemented with norepinephrine whereas a control plasmid had no effect (Figure (Figure5B,5B, top panel). A Northern blot of total RNA isolated from the WT and the Δvad1 mutant containing either the iNOT1 or iControl plasmid was probed with NOT1 and ACT1, which demonstrated that NOT1 mRNA is reduced in the Δvad1 mutant expressing NOT1-interfering RNA (Figure (Figure5B,5B, bottom panel). These data suggest that accumulation of NOT1 mRNA contributes to the laccase deficiency of the Δvad1 mutant and is compatible with roles of other members of the RCK/p54 family in RNA degradation (36). To determine the effect of NOT1 mRNA accumulation on laccase expression independent of VAD1 mutation, full-length NOT1 coding sequence under the control of a GPD1 promoter was introduced as a transgene into WT C. neoformans, creating strain NO20. Overexpression of NOT1 in the WT resulted in decreased expression of laccase as demonstrated by Northern blot (Figure (Figure5C,5C, top panel) as well as reduction in laccase activity (Figure (Figure5C,5C, bottom panel), which suggest that NOT1 is a negative regulator of laccase expression in C. neoformans.
Next, we proceeded to characterize the 3 transcripts reduced in the Δvad1 mutant in parallel to the virulence factor laccase. As shown in Figure Figure6,6, 1 fragment was subcloned and sequenced and found to be nearly identical to annotation fragment 162.m02888 of the TIGR Cryptococcus neoformans database. A conceptual translation of the annotation fragment resulted in a 607 AA fragment having homology to PCK1 of S. cerevisiae (65% identity to NP013023). PCK1 encodes the enzyme phosphoenolpyruvate carboxykinase, which catalyzes the only irreversible step in gluconeogenesis and is a major regulatory checkpoint for the control of gluconeogenesis (42, 43). Targeted knockout strains of PCK1 were constructed and confirmed by Southern blots digested with BamHI/KpnI and hybridized with a PCK1 fragment (Figure (Figure6B;6B; WT: 8.0, 6.0, 1.1 kb; Δpck1: 8.0, 6.0 kb). The Δpck1 mutant was found to show normal growth on glucose-containing media (Figure (Figure6C,6C, upper panel), but showed reduced growth on the 3-carbon substrate lactate (Figure (Figure6C,6C, lower panel), similar to that shown by the Δvad1 mutant. Growth in YPD and asparagine liquid with 2% glucose at 37°C showed identical rates (doubling times — YPD: 2.5, 2.5; 2% glucose/asparagine: 4.0 and 3.9 hours for WT and Δpck1 strains, respectively). Laccase activity was unaffected by the Δpkc1 mutation (data not shown). As shown in Figure Figure6D,6D, the Δpck1 mutant was found to be markedly less virulent than WT cells (P < 0.001) even at a large inoculum (1 × 106 cells), with restoration of virulence with PCK1 complementation, showing the importance of this enzyme during infection.
The finding of a role for an RCK/p54 protein in the regulation of PCK1 has not, to our knowledge, been reported previously and suggests an additional regulatory pathway for this enzyme in addition to the cAMP and Mig-1–dependent pathways previously proposed from studies in S. cerevisiae (43). Furthermore, establishment of a role for PCK1 in cryptococcal virulence provides insights into the pathogenesis of cryptococcal infections. Gluconeogenesis is the pathway for biomass production during periods of glucose deprivation. This anabolic pathway utilizes either 3-carbon fragments such as lactate or 2-carbon fragments such as acetate produced by the glyoxylate shunt pathway and would be expected to be active in environments having low glucose, such as is typical in the brain during cryptococcal meningoencephalitis (44). Recently, the glyoxylate shunt pathway, which utilizes 2-carbon fragments, has been shown to be dispensable for the virulence of C. neoformans (45). In combination with this study, the finding that PCK1 is required for virulence of C. neoformans suggests that 3-carbon substrates such as lactate rather than 2-carbon fragments may be preferred for biomass production during infection. Indeed, while lactic acid is produced during fungal meningitis (46), spectroscopy studies have shown a preponderance of the 2-carbon substrate, acetate, during cryptococcal brain infections (47). The accumulation of acetate and consumption of lactate may be due to an inability of the pathogen to optimally utilize acetate by the glyoxylate pathway that is induced only late in infection (45), whereas lactate is successfully consumed and converted into biomass during infection through the gluconeogenic pathway.
The next transcript reduced in the Δvad1 mutant was found to be nearly identical to annotation fragment 177.m03152 from the TIGR Cryptococcus neoformans database, which is predicted to encode a 480-AA fragment exhibiting 65% identity to TUF1 from S. cerevisiae (NP014830.1). TUF1 is a nuclear-encoded elongation factor required for translation of mitochondrial proteins; it typically shows reduced growth on all media, especially nonfermentable substrates such as glycerol (48). TUF1 mRNA abundance was reduced in the Δvad1 mutant under both glucose repressed and derepressed conditions (Figure (Figure7A).7A). The TUF1 transcript, unlike those of PCK1 and laccase, is not glucose repressible, which suggests that modulation of transcription by VAD1 is not limited to glucose repressible genes. Attempts to produce a knockout of the cryptococcal TUF1 were unsuccessful in spite of analysis of approximately 1000 transformants, so RNAi was used to assess the growth phenotypes after TUF1 transcription inhibition. As shown in Figure Figure7B,7B, suppression of TUF1 by RNAi resulted in a reduction of growth under all conditions, including in the presence of glucose, and caused a severe growth reduction on glycerol, consistent with the growth reduction on this nonfermentable substrate demonstrated by the Δvad1 mutant. Reduction of TUF1 expression by RNAi had no effect on laccase expression, which suggests that the reduction of TUF1 mRNA levels in the Δvad1 mutant does not contribute to the laccase-deficient phenotype. The severe impairment of normal growth exhibited by the strains expressing TUF1-interfering RNA would most likely result in decreased virulence and was not tested further.
Expression of genes encoded by mitochondrial DNA is essential for oxidative phosphorylation and hence for a large part of ATP production in a cell. The importance of mitochondrial functions has been demonstrated in other cryptococcal studies, in which deletion of a mitochondrial gene encoding the alternative oxidase of mitochondria resulted in defects in tolerance to the stress environment in macrophages and attenuation in cryptococcal virulence (49). TUF1 regulation by VAD1 thus suggests a novel pathway for the regulation of mitochondrial functions of eukaryotes in general and fungal pathogens in particular.
The final transcript reduced in the Δvad1 mutant is nearly identical to annotation fragment 184.m04573, which is predicted to encode a 416 AA polypeptide. The predicted protein contained a putative hydrophobic leader sequence and glycosyl phosphatidylinositol anchor site and was found to have 40% serines with a ser/thr-rich carboxyterminus (AA 321–396), properties also possessed by a class of cryptococcal mannoproteins recently described as having T cell stimulatory properties (50, 51). Because of these structural similarities to the 2 previous characterized proteins and a lack of a homolog in other organisms as determined by a BLAST analysis of the NCBI database, the gene was named after this class of proteins: mannoprotein of Filobasidiella neoformans number 3 (MPF3). As shown in Figure Figure8A,8A, MPF3 exhibited glucose repression and showed reduced expression in the Δvad1 mutant (50% ± 5%), but this reduction was not as marked as that observed for laccase, PCK1, or TUF1. Deletion of MPF3 was confirmed by Southern blot (Figure (Figure8B)8B) and resulted in a subtle yellow discoloration of the strains but no observable difference in growth in glucose or glycerol, and the growth rate in 10% human serum was nearly identical to that of the WT and complemented strain (doubling times — WT: 2.7 ± 0.05 hours; Δmpf3: 2.7 ± 0.06 hours; Δmpf3:MPF: 2.6 ± 0.16 hours). No effect on laccase activity was seen in the Δmpf3 mutant (data not shown). However, a marked growth defect was observed when the Δmpf3 mutant was grown on media containing 1M NaCl or 1.8 M sorbitol (Figure (Figure8C),8C), suggesting a defect in cell wall integrity (32) that was also exhibited by the Δvad1 mutant. Injection of mice with a large inoculum (106) did not result in a significant alteration in survival of mice (data not shown), but injection of a smaller inoculum (103) did result in attenuation of virulence in a mouse model (Figure (Figure8D;8D; P < 0.01). These data show a role for MPF3 in tolerance to osmotic and salt stress and suggest that MPF3 contributes to the virulence of C. neoformans.
The categories of target genes dependent on VAD1 expression appear remarkable in their wide spectrum of activity in cellular functions. However, all appear to have a role during exposure to stress, implicating VAD1-dependent regulation in the cellular stress response. In the mammalian host during pathogenesis, the organism is subjected to a wide variety of cellular stresses (52). As an opportunistic pathogen, C. neoformans most likely evolved its stress-regulatory pathways to maximize its chances for survival within its environmental niche, which includes the hollows of trees and occasional encounters with pathogenic amoeba (53, 54). However, the response to environmental stress, mediated by evolved regulatory networks, has also proven effective protection against the mammalian host defense. For example, the VAD1-dependent protein laccase is a cell wall enzyme that acts during pathogenesis in the brain to convert dopamine to immunomodulatory products and to neutralize the active iron form within macrophages (55, 56). However, in the hollows of trees, laccase may be considered a foraging enzyme, having a role in lignin degradation for the breakdown of wood (53, 57). Similarly, the induction of PCK1 that is required for biomass production during glucose starvation in trees can allow utilization of lactic acid produced by host cell macrophages during infection. In addition, TUF1, as a required gene involved in the production of mitochondrial proteins, plays a role in efficient ATP generation from oxidative versus fermentative growth (58) that is essential for normal growth in host cells as well as in wooded environments. Finally, MP3, required for protection against osmotic and ionic stress, perhaps during dehydration of the environment, can also serve to protect the pathogen during infection. The combined effects of reduced transcription of these important genes in the Δvad1 mutant undoubtedly contribute to the marked attenuation in virulence of the regulatory mutant. These studies are, to our knowledge, the first functional genomics studies of a C. neoformans transcriptome and demonstrate the utility of investigating virulence-associated transcriptomes for the identification of new virulence factors in fungal pathogens. Furthermore, identification of virulence-associated master regulators such as Vad1 may allow the pharmacological targeting of multiple virulence phenotypes simultaneously, thus providing more effective methods for the control and prevention of cryptococcosis.
A method of insertional mutagenesis was performed using a linearized plasmid pMUT8 as described previously (33). One mutant with a single genomic insertion of pMUT8 (confirmed by Southern blot) and reduced laccase activity was chosen for further investigation. To rescue the inserted pMUT8, genomic DNA was prepared and digested with NdeI to excise the insertional plasmid from genomic DNA as previously described (33). A 7-kb insertional plasmid containing the flanking region of a disrupted gene was obtained. Automated sequencing of the plasmid was performed by standard methods (33). Sequences from both sides of the insertional plasmid were used to design oligonucleotide primers (C53T, C53U5) to amplify a 500-bp fragment from H99 genomic DNA using standard PCR methods.
The 500-bp PCR fragment amplified from H99 genomic DNA was used to screen a previously constructed H99 cDNA library (33) by standard techniques. Clone pcVAD1 contained the largest insert and was sequenced from both strands. Genomic and cDNA sequences were aligned and have been submitted to the GenBank database under accession number AY654620. The 2.5-kb fragment of the cDNA clone was radiolabeled to isolate a 4.5-kb genomic DNA allele (clone pgVAD1) from an H99 genomic library generated in a Lambda ZAP Express vector (Stratagene) according to the manufacturer’s instructions. To complement the Δvad1 mutant, a 2-kb hygromycin B resistant gene was fused with ClaI-restricted plasmid pgVAD1 to generate pVAD1/HgR, which was transformed into strain Δvad1 by electroporation as described (59).
To construct the c-myc–VAD1 fusion, 2 oligonucleotides containing a StuI site encoding the sequence EQKLISEEDL were ligated to the complementary site within plasmid pVAD1/HgR to generate pVAD1/HgR-myc. Correct insertion of the c-myc tag was confirmed by sequencing. The construct was digested with NheI and PvuII to remove the plasmid vector and was transformed into a Δvad1 mutant by electroporation. Transformants were selected on hygromycin B–containing medium, confirmed as having the construct inserted by Western blot of whole-cell extracts, and tested for restoration of laccase expression on norepinephrine agar as previously described (33). The c-myc–VAD1 complemented strain was then grown on YPD media, washed twice with sterile distilled water, and incubated for 3 hours at 30°C in asparagine salts either with 2% glucose or without; cell extracts were subjected to gel filtration on a TSK-GEL G6000 (Supelco Chromatography) HPLC column as described in Supplemental Methods.
A PCR-amplified fragment of VAD1 containing VAD1 promoter elements was inserted upstream of a pBluescript cassette (Stratagene), consisting of a GC-rich GFP (60), an EF1α cryptococcal terminator sequence, and a 1.3-kb fragment of URA5 previously described (33), and was used to complement the Δvad1 mutant to produce strain VGFP-1. VGFP was grown on YPD for 2 days, then inoculated into asparagine liquid containing 2% glucose or identical media without glucose, incubated at 30°C for 3 hours, and prepared for immunofluorescence microscopy as previously described (61). To enable identification of the expression and cellular location of the Vad1 protein during neuropathogenesis, mice were inoculated with 105 cells of either WT C. neoformans or the GFP-Vad1–expressing strain VGFP-1. Mice were monitored for 2 weeks, after which they were sacrificed by CO2 narcosis. Brains were immediately removed and sectioned longitudinally. One half of the brain was homogenized and cultured for colony counts and the other half homogenized in ice-cold phosphate-buffered saline and subjected to sucrose-gradient centrifugation for 20 minutes at 4°C to separate C. neoformans cells from brain tissue as previously described (62), then subjected to fluorescence microscopy as above.
The method of Ausubel et al. (63) was used. Briefly, formaldehyde-fixed tissue was sectioned (4 micron sections), dewaxed with xylene, and rehydrated. A 579-bp Sac I-EcoR I fragment of the VAD1 cDNA was subcloned into pBluescript SK (Stratagene), then transcribed in vitro using T7 polymerase and NTPs and labeled with U-digoxigenin according to the manufacturer’s instructions (Roche Diagnostics Corp.). The tissue section was subjected to RNA denaturation with sequential washes with 0.2 N HCl, then washed twice with SSC at 70°C, followed by blocking with buffer containing iodoacetamide and N-ethylmaleimide and dehydration. Serial sections were then hybridized with either sense or antisense probes prepared as described above at 50°C overnight, then washed at the same temperature in RNAse-free buffer. Slides were then incubated with a second alkaline phosphatase–conjugated secondary antibody overnight at 4°C, washed, and visualized with 5-bromo,4-chloro,3-indolylphosphate/nitroblue tetrazolium substrate as described in ref. 64. This was followed by staining with Fast Red (Sigma-Aldrich).
Constructs for the deletion of PCK1, TUF1, and MPF3 consisted of the 1.3-kb URA5 selectable marker flanked by approximately 500 bp of genomic sequence homologous to the regions upstream and downstream of the coding sequence for each gene. Each construct was transformed into H99FOA as described previously (9). Transformants were screened for homologous recombination by PCR followed by Southern blot verification. A detailed description of the construct design and screening for each gene is included in Supplemental Methods.
RNAi suppression of TUF1 was performed after the method of Liu et al. (65) using a plasmid derived from pPM8 (66). Briefly, TUF1 was PCR amplified from a H99 cDNA mass-excised library as previously described (33), using primers TUF1-3114SR and TUF1-3657AX and ligated downstream of the ACT1 promoter with a linker fragment consisting of a 500–bp PCR-amplified fragment of intron I of CNLAC1 generated from a CNLAC1 genomic clone (plasmid p5.1) (8) using primer INTRON-XhoS and INTRON-XhoA to produce plasmid EF-TU7A. EF-TU7A was linearized and transformed into H99FOA and selected on minimal media. Control strains were transformed with the identical plasmid EF-TU7B, which contained only one copy of TUF1 without a second antisense fragment. A similar plasmid was constructed for suppression of NOT1 using primers iNOT-F-RI and iNOT-R-Xh. For overexpression of NOT1, the full-length NOT1 coding sequence with a C-terminal c-myc tag was PCR amplified with primers FL-NOT-F-Xb and mycNOT-R-Nt, then cloned downstream of the GPD1 promoter and upstream of the TRP1 terminator in pSL1180 (Amersham Biosciences) containing the hygromycin resistance cassette. The construct was linearized with SpeI and VspI and transformed into H99 by electroporation. Transformants were selected on YPD with hygromycin, and genomic insertion of the construct was verified by Southern blotting of uncut genomic DNA.
Total RNA was isolated from WT (H99) and Δvad1 strains (8) under the following conditions: Cells were grown in YPD broth at 30°C to mid-log phase with absorbance of 0.5–0.6 at 600nm. Cells were subsequently derepressed in asparagine medium without glucose (1g/l asparagine, 0.1g/l MgSO4, 10mM sodium phosphate, pH 6.5) at 30°C for 3 hours. Differential display analysis was performed on both strains using 24 primer sets per the manufacturer’s instructions (GenHunter). A BLAST search of the sequence (http://www.ncbi.nlm.nih.gov/blast/) from recovered fragments was performed against the TIGR Cryptococcus neoformans genome database (http://www.tigr.org/tdb/e2k1/cna1/) to identify the putative open reading frames of the subcloned cryptococcal genomic fragments.
Capsule formation was assessed using methods previously described (33). Urease production was measured by incubation of cells on Christensen’s agar (25), and the method of Liu et al. (55) was used to measure laccase activity. Growth rate in liquid media was measured according to Salas et al. (9). Mating was conducted on sucrose proline media as described (67). Virulence studies were conducted according to a protocol using a previously described mouse meningoencephalitis model (9). Animal studies were approved by the University of Illinois at Chicago Animal Care Committee. Studies involving human tissue were approved by the University of Illinois at Chicago Office for the Protection of Research Subjects and the Institutional Review Board (protocol #2004-0600).
Statistical significance of mouse survival times was assessed by Kruskall-Wallis analysis (ANOVA on Ranks). Pairwise analyses were performed post hoc by using Dunn’s procedure. Intensity of signals from Northern blot analyses was determined by densitometry using the STORM 860 phosphorimager (Amersham Biosciences).
This work was supported, in part, by United States Public Health Service grants NIH AI49371 and AI45995 (to P.R. Williamson), National Research Service Award NIH 1F32AI062124-01 (to J. Panepinto), and an American Heart Association fellowship (to X. Zhu). We would like to acknowledge GenHunter for technical assistance with differential display and Lofstrand Labs Ltd. for technical assistance with in situ hybridization. We also thank TIGR (supported by NIH grant 1 UO1 AI48594-01) and the Fungal Genomic Initiative for preliminary sequence information for the C. neoformans genome.
See the related Commentary beginning on page 593.
Nonstandard abbreviations used: RNAi, RNA interference; TIGR, The Institute of Genomic Research; VAD1, virulence-associated DEAD-box RNA helicase–encoding gene; YPD, yeast extract–peptone–dextrose.Conflict of interest: The authors have declared that no conflict of interest exists.