Metabolic adaptation in Cryptococcus neoformans during early murine pulmonary infection

Summary

The pathogenic fungus Cryptococcus neoformans generally initiates infection in mammalian lung tissue and subsequently disseminates to the brain. We performed serial analysis of gene expression (SAGE) on C. neoformans cells recovered from the lungs of mice and found elevated expression of genes for central carbon metabolism including functions for acetyl-CoA production and utilization. Deletion of the highly expressed ACS1 gene encoding acetyl-CoA synthetase revealed a requirement for growth on acetate and for full virulence. Transcripts for transporters (e.g., for monosaccharides, iron, copper and acetate) and for stress-response proteins were also elevated thus indicating a nutrient-limited and hostile host environment. The pattern of regulation was reminiscent of the control of alternative carbon source utilization and stress response by the Snf1 protein kinase in Saccharomyces cerevisiae. A snf1 mutant of C. neoformans showed defects in alternative carbon source utilization, the response to nitrosative stress, melanin production and virulence. However, loss of Snf1 did not influence the expression of a set of genes for carbon metabolism that were elevated upon lung infection. Taken together, the results reveal specific metabolic adaptations of C. neoformans during pulmonary infection and indicate a role for ACS1 and SNF1 in virulence.

Introduction

Cryptococcus neoformans causes life-threatening meningoencephalitis in patients with immune deficiency (Casadevell and Perfect, 1998). The fungus is found in the environment as desiccated cells and/or basidiospores produced by sexual reproduction or monokaryotic fruiting; inhalation of these cells initiates a pulmonary infection in mammals (Lin and Heitman, 2006). Human exposure to C. neoformans is thought to be common based on the presence of antibodies in the majority of normal individuals, and many cases of cryptococcosis in immunocompromised individuals may result from reactivation of latent asymptomatic infections (Goldman et al., 2001). C. neoformans is a facultative intracellular pathogen during early stages of murine pulmonary infection (Feldmesser et al., 2000; Feldmesser et al., 2001). In fact, Feldmesser et al., (2000) found that the percentage of fungal cells internalized in alveolar macrophages reached a peak during the first 24 hours of infection. Most of the yeast cells were extracellular by 24 hours, a time associated with macrophage cytotoxicity, and both intracellular and extracellular cryptococci were observed at 48 hours and 7 days. Fungal cells eventually disseminate from the lung via the bloodstream and reach the brain to cause meningoencephalitis.

Although the disease process is well characterized, relatively few studies have examined C. neoformans gene expression during growth in vivo (Steen et al., 2003, Rude et al., 2002, Fan et al., 2005). In an initial study, a transcriptional profile was generated by differential display RT-PCR for C. neoformans cells during meningitis in an immunosuppressed rabbit model of meningitis (Rude et al., 2002). This analysis revealed elevated expression for the gene ICL1 encoding isocitrate lyase, a key enzyme in the glyoxylate cycle, suggesting that this cycle might be important for fungal growth in the host. However, disruption of ICL1 did not influence virulence in two animal models nor cause a growth defect in macrophages (Rude et al., 2002). Similarly, deletion of the MLS1 gene encoding malate synthase (another glyoxylate cycle enzyme) did not influence virulence (Idnurm et al., 2007). Subsequent transcriptional profiling of C. neoformans from the cerebrospinal fluid (CSF) of infected rabbits by serial analysis of gene expression (SAGE) revealed relatively high expression of genes involved in energy production, stress response, and small molecule transport, as well as carbohydrate, amino acid, and lipid metabolism (Steen et al., 2003). The transcriptional response of C. neoformans cells upon phagocytosis by murine macrophages has also been examined (Fan et al., 2005). The fungus responds to phagocytosis with elevated expression of genes at the MAT locus and in the cAMP/protein kinase A pathway. Additionally, elevated expression was seen for genes involved in autophagy, peroxisome function, membrane transport, lipid metabolism and the response to oxidative stress (Fan et al., 2005). These studies provide the first insights into C. neoformans gene expression during infection or upon phagocytosis in vitro. Further studies are needed, however, because gene expression profiles could vary substantially for chronic versus acute infections or in different host tissues, as demonstrated for other pathogens such as Candida albicans (Barelle et al., 2006; Fradin et al., 2003; Lorenz and Fink, 2001).

Studies in other pathogenic fungi examined gene expression during infection and explored the roles of glycolytic functions, the glyoxylate cycle, gluconeogenesis and β-oxidation in virulence (Lorenz and Fink, 2001, Lorenz et al., 2004; Idnurm and Howlett, 2002; Solomon et al., 2004; Wang et al., 2003; Piekarska et al., 2006; Barelle et al., 2006; Schöbel et al., 2006; Ramirez and Lorenz, 2007; Thewes et al., 2007; Olivas et al, 2008). For example, the transcriptional response of C. albicans upon phagocytosis by macrophages includes the up-regulation of functions for the glyoxylate cycle, gluconeogenesis and fatty acid degradation (Lorenz et al., 2004; Prigneau et al., 2003). In addition, phagocytosis by neutrophils induced an amino acid deprivation response in both C. albicans and S. cerevisiae (Rubin-Bejerano et al., 2003). Mutants defective in genes encoding glyoxylate pathway functions (e.g. isocitrate lyase), gluconeogenesis functions (phosphoenolpyruvate carboxykinase) and glycolytic functions (phosphofructokinase and pyruvate kinase) are attenuated for virulence (Lorenz and Fink, 2001; Barelle et al. 2006). For β-oxidation, deletion of the FOX2 gene encoding the second enzyme in the pathway also resulted in attenuated virulence (Piekarska et al., 2006; Ramirez and Lorenz, 2007). However, this phenotype may result from an influence on the glyoxylate cycle because deletion of the PEX5 gene for peroxisomal biogenesis did not cause a virulence defect (Piekarska et al., 2006). In human blood, differentially expressed genes in C. albicans encoded functions for a general stress response, an antioxidative response, the glyoxylate cycle, and virulence factors (Fradin et al., 2003). Barelle et al. (2006) used GFP fusions to examine the expression of genes in the glyoxylate cycle, gluconeogenesis and glycolysis in more detail in C. albicans. They found that the genes for the glyoxylate pathway and gluconeogenesis were repressed by the concentration of glucose found in blood and that the genes were induced during phagocytosis (as found by other investigators). Interestingly, these genes were not expressed in fungal cells in infected kidneys. In contrast, glycolytic genes were not induced upon phagocytosis but were expressed in cells in infected kidneys. Barelle et al. (2006) concluded that the glyoxylate cycle and gluconeogenesis may be important early in infection and that glycolysis is important during systemic disease. In light of these results, the finding that components of the glyoxylate cycle were not required for virulence in C. neoformans suggests that this fungus may have different nutritional requirements during infection (Idnurm et al., 2007). Similarly, deletion of the glyoxylate cycle genes for isocitrate lyase and malate synthase did not result in virulence defects in Aspergillus fumigatus (Schöbel et al., 2007; Olivas et al., 2008).

In this study, we examined the early transcriptome changes that occur upon C. neoformans deposition in the murine lung. We performed SAGE analysis on C. neoformans cells recovered from lungs at 8 and 24 hours after infection and compared the data with previously described SAGE libraries from cells grown in culture, and from cells harvested from the central nervous system of infected rabbits. The SAGE data suggested that murine pulmonary infection represents a nutrient-limiting environment for invading C. neoformans cells; specifically, genes in several functional categories showed elevated transcripts including alternative carbon source utilization, central carbon and lipid metabolism, stress response and virulence. To examine carbon source utilization in more detail, we characterized the ACS1 gene encoding acetyl-CoA synthetase and the SNF1 gene encoding a predicted serine/threonine protein kinase in C. neoformans. Loss of ACS1 resulted in poor growth on acetate and a mild attenuation of virulence. The SNF1 gene was examined because the pattern of gene expression resembled the regulatory influence of the well-characterized Snf1 protein kinase in S. cerevisiae; this protein mediates glucose sensing, alternative carbon source utilization and the response to stress. Deletion of the SNF1 gene in C. neoformans gene resulted in poor growth on acetate and ethanol at 37°C, reduced melanin production and a complete loss of virulence in a murine model of cryptococcosis.

Results

Serial analysis of gene expression during early pulmonary infection

To investigate pathogen gene expression during pulmonary cryptococcosis, we generated SAGE libraries from fungal cells harvested by bronchoalveolar lavage (BAL) from mouse lungs at 8 (21,510 tags) and 24 hours (h) (20,129 tags) after infection. The details of the SAGE analysis are presented in Experimental Procedures and in Tables S1, S2 and S3 of the Supplementary material. A comparison of the two libraries identified 382 tags with differential levels at the two stages of infection suggesting dynamic changes during adaptation to the host environment. The SAGE libraries were also compared with two previously described libraries constructed from cells grown in low iron medium (LIM) (77,829 tags; Hu et al., 2007) or in YNB broth at 37°C (84,039 tags; Steen et al., 2002). These are referred to as in vitro libraries. In addition, the SAGE data for fungal cells from rabbit cerebral spinal fluid (CSF) (66,217 tags; Steen et al., 2003) were compared with the pulmonary infection libraries (in vivo libraries). Overall, these libraries provided snapshots of RNA abundance in cells from quite different conditions of temperature (i.e., 37°C for both in vitro and the in vivo lung libraries, and 39.5°C for rabbit), nutrients (i.e, YNB broth, low iron medium (LIM) and host environment) and stress (i.e., presence or absence of host defense responses). The SAGE data were normalized to 20,000 tags per library to facilitate comparisons between all five conditions. For this analysis, we focused mainly on comparisons between in vivo (mouse lung and rabbit CSF) and in vitro (LIM and YNB broth) conditions. Following annotation to match tags to genes, we identified several functional categories containing differentially expressed genes as shown in Tables 1 and 2, and Tables S4 and S5 of the Supplementary material. A notable general observation was that the patterns of expression were most similar for the libraries prepared with cells from LIM and rabbit CSF (Table S2). For the bulk of the analysis described below, we employed the data for the 8 h lung library and the LIM library to define differential expression with the primary goal of identifying genes that showed elevated expression during lung infection.

Table 1.

SAGE tags for genes encoding predicted functions in carbon and lipid metabolism, and transport.

Tag	Low Iron Medium	Yeast Nitrogen Base	Lung, 8 hours	Lung, 24 hours	Rabbit, Cerebral Spinal Fluid	TIGR Gene number	GenBank Accession number	Predicted function and gene name (if known)
Glyoxylate cycle and acetate/ethanol/pyruvate metabolism
CAGCAGTGTA	3	0	144	46	7	CNF03900	XM_571348	aldehyde dehydrogenase
CGCGAGGGCA	1	9	80	40	0	CNA07740	XM_567122	acetyl-CoA synthase, ACS1
GCCCAGCAGG	0	1	62	25	0	CNH02910	XM_572462	malate synthase, MLS1
CATCACTCTT	13	2	39	48	13	CNJ00950	XM_567475	pyruvate decarboxylase
ACCTTTATAT	0	0	7	3	0	CNC06260	XM_569885	alcohol dehydrogenase
Tricarboxylic acid cycle
CCAATGATTA	1	0	19	6	8	CND02620	XM_570245	aconitase
CATTTTCTCA	0	0	15	6	1	CNG03480	XM_572036	succinate dehydrogenase
GGTTACGCCG	1	55	6	3	1	CNG03490	XM_572038	malate dehydrogenase
AACTTTGTTC	13	0	4	2	15	CNA04610	XM_566837	isocitrate dehydrogenase
CATTTTGATT	8	0	0	0	2	CNF03780	XM_571672	malic enzyme
TTGGCTCCCA	5	1	0	0	2	CNC01700	XM_569517	fumarase
Glycolysis
ACTCAGGTTG	1	65	19	17	3	CNB00300	XM_568771	fructose 1,6-biphosphate aldolase
TGGTTTTTGT	0	0	11	1	0	CNJ01080	XM_567487	phosphofructokinase
TGTGAATGTG	1	11	9	6	4	CNB02660	XM_568853	hexokinase, HXK2
GTCGTAGAGT	81	3	9	3	57	CNC00160	XM_569379	enolase
GGTCGTTTAT	12	0	0	0	7	CNB04050	XM_569228	glucose-6-phosphate isomerase
TATGCAACAC	6	0	0	0	4	CNF00810	XM_571493	phosphoglycerate mutase
Gluconeogenesis
TTTCAGAAGG	0	1	68	27	2	CNI03590	XM_572603	phosphoenolpyruvate carboxykinase, PCK1
Pentose phosphate pathway
GTATTGACC	0	72	56	13	7	CNK00070	XM_567776	phosphoketolase
CATTACTGCA	5	0	17	7	25	CND02280	XM_570275	oxidoreductase
TAGTGTCCCG	70	6	41	9	32	CNK01050	XM_567793	6-phosphogluconate dehydrogenase
AAGTGTGGTA	12	2	0	0	18	CNK03170	XM_567910	transaldolase
Other functions related to carbon metabolism
CATTATGACA	4	0	19	5	5	CNA01100	XM_567089	glycerol 3-phosphate dehydrogenase
ATTCACGCGC	1	0	17	5	3	CND02900	XM_570206	glutamine:fructose-6- phosphate amidotransferase
CCATATGTTT	3	14	15	2	1	CNF03530	XM_571431	glycogen phosphorylase- like protein
GACATTGTTG	2	22	18	8	1	CNC05340	XM_569729	D-arabinitol dehydrogenase
CAACGGGGGG	35	0	7	1	11	CNH00170	XM_572212	phosphomannomutase
CTTCCTGTTC	23	1	9	6	20	CNN00430	XM_568570	phosphoglucomutase
CATCCCAAAG	15	0	0	0	20	CNC03700	XM_569599	UDP-glucose pyrophophorylase
TATAAGACCG	6	0	0	0	4	CNF03930	XM_571680	glycogen metabolism- related protein
Lipid metabolism
TAACCATATA	19	0	96	17	7	CNJ02970	XM_567576	butyrylcholinesterase, triacylglycerol lipase, CGL1
AACCACACTA	1	34	39	15	14	CNN01770	XM_568678	similar to Benzodiazepin receptor
TCATTTCACT	2	0	33	10	3	CNI01560	XM_572814	sterol-binding protein
GCGAAGTACT	0	0	27	3	0	CNN01550	XM_568632	acyl-CoA oxidase
GCTGCCTTTG	2	13	22	6	1	CNI00240	XM_572730	enoyl-CoA hydratase/isomerase
CAATGGGAAT	2	0	12	3	0	CNI03690	XM_572896	2,4-dienoyl-CoA reductase (NADPH)
TATGTGAATC	3	0	9	3	4	CNM00830	XM_568422	oxidoreductase, short chain dehydrogenase/reductase
CTACAAATGA	1	0	9	1	1	CNH02950	XM_572465	oxidoreductase, short chain dehydrogenase/reductase
GATAAGGTGT	4	0	8	6	8	CNI02880	XM_572653	oxysterol-binding protein
TATTATTGTT	0	0	8	1	2	CNA03320	XM_566658	enoyl-CoA hydratase
AAGCAGAGGA	2	9	7	3	5	CNI00710	XM_572753	long-chain-fatty-acid CoA ligase
TAGTTTATAT	2	0	7	3	2	CNG02680	XM_571931	hydroxymethylglutaryl-CoA synthase
ATTGAATGTA	0	0	7	2	25	CND01120	XM_570351	fatty acid beta-oxidation- related protein
Transporters
TATACGATTT	26	0	612	143	2	CNB02680	XM_568855	monosaccharide transporter, HXT1
CCCATCGTAT	82	0	136	79	24	CNM02430	XM_568258	plasma membrane iron permease, CFT1
CTTGACAGCG	0	5	42	46	0	CNF02960	XM_571371	acetate transporter, ADY2
TATCTGATTT	19	1	34	8	10	CNG04240	XM_572092	hexose transporter
TATTGGTACA	16	0	28	13	0	CNC03960	XM_569629	phosphate transporter
GTCATTTATG	2	0	27	3	2	CNF02890	XM_571700	neutral amino acid transporter
CGGGAATACG	0	0	24	12	1	CNH03640	XM_572526	GPR/FUN34 family protein/acetate transporter, ATO2
CTGTTCGGCA	9	3	21	13	11	CND01080	XM_570353	high affinity copper transporter, CTR4
CACTGTATGC	1	0	21	18	0	CNE02570	XM_570958	succinate:fumarate antiporter
GCCGGGCTGT	1	15	20	14	2	CNC02540	XM_569547	organic acid transporter
CCGAGAAACG	2	0	19	7	2	CND04380	XM_570596	monovalent inorganic cation transporter
AGCACTTATG	1	0	16	2	0	CNI03490	XM_572882	maltose permease, trehalose transporter
CAGTATCACT	2	0	15	2	2	CNL05010	XM_568040	gaba permease
GTGGTTTTCT	1	0	11	9	2	CNF04760	XM_571440	neutral amino acid permease
CATATATAAC	2	0	7	0	1	CNH00950	XM_572289	spermine transporter
GGAAGAGGAA	1	0	6	3	2	CNE00440	XM_571108	Inner membrane solute transporter MRS4
ATCGGTACCC	4	47	0	0	4	CNN01030	XM_568544	inorganic phosphate transporter

Tags are listed by abundance based on the 8 hour library; these tags have higher levels in this library relative to the LIM library. The remaining tags in the shaded rows are listed by their abundance in the LIM library. The tag counts were normalized to 20,000 tags for each library.

Table 2.

Virulence-associated and stress-response genes.

Tags	Low Iron Medium	Yeast Nitrogen Base	Lungs, 8 hours	Lungs, 24 hours	Rabbit, Cerebral Spinal Fluid	TIGR Gene number	GenBank	Predicted Function and gene name (if known)
TATATGTGTA	35	0	247	60	262	CNG04220	XM_572088	heat shock protein 12
CACGTCCACG	27	61	95	30	5	CND01490	XM_570285	Cu/Zn superoxide dismutase, SOD1
ATAAGCTTTC	12	0	74	14	25	CNG00600	XM_571772	mannitol-1-phosphate dehydrogenase
CATTTTATGT	14	0	35	16	24	CNM01520	XM_568451	heat shock protein 90
TAACATAATG	16	0	30	3	17	CNA01260	XM_567249	glutathione S-transferase
TGTTTCTACA	2	1	24	3	3	CNG04220	XM_572088	heat shock protein 12
CTCTTCATTT	8	4	23	14	10	CNA03160	XM_566757	heat-shock protein, SKS2
GCGGATAAAA	6	5	23	6	8	CNB00910	XM_569098	stress response RCI peptide, cation transport-related protein
GCTGCCTCTG	0	4	21	32	1	CNA01500	XM_566622	alternative oxidase
ATGTTTTATT	9	0	20	7	3	CNG02560	XM_572003	UDP-glucuronic acid decarboxylase, UXS1
AGATGGACGA	4	19	19	8	4	CNG01060	XM_571865	cyclophilin-like protein
AATTGATGAG	0	0	18	10	2	CNC07160	XM_569844	flavohemoprotein, FHB1
TTGATTTTTT	0	0	16	3	0	CNM01860	XM_568294	macrolide-binding protein FKBP12
GCGACAGCCA	0	1	15	2	1	CNC01490	XM_569482	peptidyl prolyl cis-trans isomerase, ESS1
ACGGAGTGTA	1	0	14	9	2	CND00260	XM_570333	FK506-resistant calcineurin B regulatory subunit
TCGATGGCGA	2	0	11	5	14	CNE05040	XM_571033	glyoxal oxidase precursor
CAACGATGAT	5	0	10	3	66	CNF02910	XM_571367	LEA domain protein
GCTTAATTAT	0	1	8	2	1	CNN00360	XM_568558	protein kinase, SCH9
GTCTTCATCA	2	0	8	2	1	CND05600	XM_570476	heat shock protein 12
TTTCTGAAAA	2	0	7	1	1	CND05940	XM_570487	capsule associated protein, CAP1
TGTGTAAAAG	0	0	6	1	0	CNE00710	XM_570793	mannitol-1-phosphate dehydrogenase
TCAGAAGTTG	78	8	11	1	51	CNC04200	XM_569667	thioredoxin
TGTTATCGGT	60	4	22	6	54	CNM01520	XM_568451	heat shock protein 90
CATAATTGGC	35	1	13	1	14	CNM02070	XM_568283	heat shock protein

Tags are listed by abundance based on the 8 hour library and show higher levels in this library relative to the LIM library. The remaining tags in the shaded rows are listed by their abundance in the LIM library. The tag counts were normalized to 20,000 tags for each library.

Adaptation during infection involves changes in central carbon metabolism

Initially, we identified a number of tags for genes involved in carbon metabolism that were expressed at high levels during lung infection. In particular, tags for functions in the glyoxylate pathway, the TCA cycle, gluconeogenesis, lipid catabolism and two-carbon metabolism were elevated relative to the two in vitro SAGE libraries (Table 1). Notably, these results suggested a metabolic emphasis on the generation and utilization of acetyl-CoA upon infection. The position of acetyl-CoA in central carbon metabolism is summarized in Fig. 1. For the glyoxylate cycle, the tag for the gene encoding malate synthase was notably higher in the lung libraries (60 times higher at 8 h and 25 times at 24 h) compared with the in vitro libraries. In contrast, the tag for this gene was not present in the normalized data for the CSF library. Disruption of the MLS1 gene for malate synthase in C. neoformans eliminates the ability of the fungus to grow on acetate but does not affect virulence (Idnurm et al., 2007). Tags for genes encoding the TCA cycle enzymes aconitase and succinate dehydrogenase were also elevated in the lung libraries. However, tags representing three other TCA cycle enzymes were lower in the lung libraries than in one or both of the in vitro libraries (Table 1). The SAGE analysis also revealed that a tag for the gene encoding phosphoenolpyruvate carboxykinase (PCK1), which controls the only irreversible step in gluconeogenesis, was higher in both in vivo libraries (68 and 27 times at 8 h and 24 h after infection, respectively, versus the in vitro libraries). This result suggests that glucose may be limiting during early infection resulting in the activation of gluconeogenesis in at least a portion of the cells in the population. Panepinto et al., (2005) found that disruption of PCK1 caused poor growth on lactate and markedly reduced virulence, thus indicating the importance of gluconeogenesis during infection.

Figure 1. The position of acetyl-CoA in central carbon metabolism.

Metabolic processes and metabolites are indicated in upper and lower case letters, respectively. Note that neither the cellular compartments for the various reactions nor the import processes for metabolites such as ethanol and acetate are indicated.

The possibility of glucose limitation is supported in part by observations that the tags for genes encoding glycolytic functions including glucose-6-phosphate isomerase and phosphoglycerate mutase were relatively low in the lung libraries. However, this pattern was not consistent for all glycolytic functions because the tags for fructose 1,6-biphosphate aldolase, hexokinase and phosphofructokinase were present at similar levels between the libraries or elevated in one of the lung libraries. Additionally, the tag for enolase was present at low levels in the lung libraries compared to elevated levels in the libraries from cells grown in low iron medium or isolated from cerebral spinal fluid. Similarly, tags for different genes involved in the pentose phosphate pathway (PPP) showed a mixed pattern of elevated (e.g., 6-phosphogluconate dehydrogenase), similar (e.g., phosphoketolase) or reduced (e.g., transaldolase) expression in the in vivo versus one or both of the in vitro libraries (Table 1).

The SAGE data also indicated that functions for lipid degradation and fatty acid catabolism were elevated during infection. Specifically, a tag for a predicted butyrylcholinesterase (triacylglycerol lipase, CGL1) was one of most abundant in the 8 h lung library, but dropped in abundance at 24 h. The level of this tag was four times higher in the 8 h library versus the LIM library and >90 times higher than in the YNB broth library; tag abundance was similar in the 24h library and the LIM library (Table 1). Lavage fluid from mammalian lungs is rich in phospholipids (Rooney et al., 1994; Rose et al., 1994) and fungal lipases such as the putative Cgl1 protein may contribute to carbon acquisition during lung colonization. Interestingly, the expression of this gene was induced during phagocytosis but attempts at disruption were not successful suggesting that the gene is essential (Fan et al, 2005). Fatty acids generated by phospholipases and triacylglycerol lipases would enter the β-oxidation pathway and eventually yield acetyl-CoA. We did identify tags for components in the β-oxidation pathway in the lung libraries (e.g., enoyl-CoA hydratase/isomerase, and 2,4-dienoyl-CoA reductase (NADPH)), suggesting a role for β-oxidation upon pulmonary infection (Table 1). We also identified tags for functions in sterol biosynthesis and regulation, including sterol-binding proteins and a hydroxymethylglutaryl-CoA synthase (Parks and Casey, 1995; Olkkonen et al., 2006).

The ability to convert acetate to acetyl-CoA contributes to virulence

The expression of genes encoding enzymes for the production of acetyl-CoA from pyruvate and acetate were elevated during pulmonary infection. In particular, elevated expression was found for the genes for acetyl-coenzyme A synthetase, (ACS1), pyruvate decarboxylase, and aldehyde dehydrogenase, as well as two acetate transporters (Table 1). Acetate utilization or production is potentially relevant to the pathogenesis of C. neoformans because Himmelreich et al. (2003) showed that it was one of the major metabolites present in infected tissue. Acetyl-CoA is generated by fatty acid catabolism, by decarboxylation of pyruvate via the pyruvate dehydrogenase complex or by direct activation of acetate by the Acs1 enzyme (Fig. 1). We also found tags for functions involved in the conversion of pyruvate, acetaldehyde, and ethanol to acetate (e.g., pyruvate decarboxylase and alcohol dehydrogenase) that were elevated in the lung libraries (Table 1). A tag for a related function, succinate:fumarate antiporter, was also elevated and this protein is required for growth on ethanol or acetate in yeast (Table 1; Palmieri et al., 1997). Notably, a tag for aldehyde dehydrogenase was one of the most abundant in both in vivo libraries (144 copies at 8 h and 46 at 24 h). A similar gene in Ustilago maydis is required for growth on ethanol as a sole carbon source (Basse et al. 1996). Taken together, these tags suggest that acetate and acetyl-CoA play important metabolic roles during pulmonary infection.

To examine the role of ACS1 in more detail, we first confirmed the expression of the gene during infection by real-time quantitative PCR (Fig. 2A). In S. cerevisiae, ACS1 is expressed during growth on non-fermentable carbon sources and under aerobic conditions, and Acs1 functions in the acetate utilization pathway (Schüller, 2003). ACS1 expression is also regulated by the Snf1 protein kinase in yeast (Young et al. 2003) and upregulated during the growth of C. albicans in blood (Fradin et al., 2003). We therefore reasoned that loss of Acs1 function would impair the growth of C. neoformans on acetate. Indeed, deletion of ACS1 resulted in a growth defect on acetate and ethanol, and also caused poor growth on glycerol (Fig. 2B). No growth defect was observed on sucrose, arabinose, galactose or lactic acid (data not shown). In contrast, the S. cerevisiae acs1 mutant can grow on ethanol but there are conflicting reports about growth on acetate (De Virgilio et al., 1992; Kratzer and Schüller, 1995). Complementation of the C. neoformans acs1 mutant with ACS1 gene restored growth to WT (wild type) levels (Fig. 2B). The acs1 mutant did not show differences in capsule or melanin formation, or growth at different temperatures, compared with the WT strain H99 (data not shown). However, the mutant did show a delay in disease progression in a mouse inhalation model of cryptococcosis (Fig. 2C). That is, mice infected with the mutant survived 7–10 days longer than those infected with the WT or complemented strains (P < 0.001). These results indicate that the ability to produce acetyl-CoA directly from acetate makes a modest contribution to growth in the host. It is possible that the synthesis of acetyl-coA from pyruvate, from β-oxidation or by other acetyl-CoA synthetases may limit the impact of loss of Acs1.

Figure 2. Analysis of ACS1 expression during pulmonary infection and the role of the gene for growth on alternative carbon sources and virulence.

(A). Quantitative real time PCR to confirm the elevated expression of the ACS1 gene during pulmonary infection relative to growth in vitro. (B) Growth of the WT strain H99, the acs1 mutant and the complemented strain on the carbon sources indicated above the panels. (C) Virulence of the acs1 mutant compared with the WT strain H99 and the complemented strain. The strains were each inoculated into 10 mice by inhalation and the mice were monitored for signs of illness. The virulence of the acs1 mutant was statistically different from that of the WT (P = 0.0001) or the complemented strain (P = 0.0008) by the Krustal-Wallis test. A representative graph is shown from a total of three experiments, and the acs1 mutant displayed a similar attenuation of virulence in each experiment.

We have considered the possibility that other genes encode acetyl-CoA synthetase in C. neoformans. Our BLASTp searches with Acs1 and Acs2 from S. cerevisiae identified the C. neoformans gene product that we designated Acs1 (expect value = 0.0, 56% identity). However, two genes encoding proteins of lower similarity were also found and preliminarily named ACS2 and ACS3. Acs2 had an expect value of E-32 (26% identity) and Acs3 had an expect value of E-13 (21% identity). For context, a BLASTp comparison of Acs1 and Acs2 from S. cerevisiae yielded an expect value of 0.0 with 57% identity. Thus, for C. neoformans, there was one clear ortholog for ACS1/ACS2 and two genes with weaker similarity. Alignments of the yeast and C. neoformans polypeptides are shown in Fig. S1 (Supplementary material). Preliminary experiments indicated that disruption of the C. neoformans ACS2 gene yielded mutants that lacked notable phenotypes in terms of growth on alternative carbon sources or virulence (data not shown). The construction of double and triple mutants will be needed to determine whether loss of these genes exacerbates the phenotypes of the acs1 mutant.

Ttransport functions show elevated expression during lung infection

The SAGE data indicated that C. neoformans cells contained elevated transcripts for several putative transporters upon growth in the lung environmental (Table 1). Notably, a tag for a gene encoding a putative glucose transporter (designated HXT1) was the most abundant transcript in both lung libraries, a finding suggestive of glucose limitation (Table S3, Supplementary material). This gene shared high similarity to the yeast RGT2 and SNF3 genes, and was also regulated by protein kinase A (PKA) as determined by SAGE analysis (Hu et al., 2007). We confirmed the elevation of the HXT1 transcript for the lung libraries by quantitative real-time PCR and found that deletion of HXT1 resulted in early melanin production (Fig. S2, Supplementary material). The influence on melanin formation is interesting because elevated glucose is known to suppress this process (Zhu and Williamson, 2004). The hxt1 mutant did not show a virulence defect (data not shown).

The other transporters had predicted functions in the movement of molecules such as amino acids, sugars (e.g., trehalose), metals (iron and copper), organic acids (e.g., acetate) and phosphate (Table 1). Trehalose transport is of interest because the trehalose pathway in C. neoformans is involved in survival in the host, the response to high-temperature stress and glycolysis (Petzold et al., 2006). The tag for a putative phosphate transporter (Pho84) was up-regulated at both 8 and 24 h during lung infection, and this gene was also induced in C. neoformans during phagocytosis (Fan et al., 2005). In S. cerevisiae, Pho84 is involved in sensing nutrient signals and activates the cAMP-PKA pathway (Giots et al., 2003; Fan et al., 2005). The candidate metal transporters included the high-affinity iron permease (CFT1, Jung et al., 2006; 2008) and a copper transporter (CTR4, Waterman et al., 2007). CFT1 was previously identified as an iron-regulated gene in C. neoformans cells and the gene is also regulated by PKA in C. neoformans (Lian et al., 2005; Hu et al, 2007). CFT1 encodes an iron permease and is required for the use of iron from transferrin and for virulence in C. neoformans (Jung et al., 2008). Overall, the elevated tags for transport functions suggest that specific assimilation activities are required to support fungal proliferation in host tissue.

Elevated expression of virulence-associated and stress response genes indicates that the lung is a stressful environment for C. neoformans

The SAGE analysis also identified a group of elevated tags representing functions related to virulence and stress (Table 2). For example, two tags for the flavohemoprotein gene FHB1 (flavohemoglobin denitrosylase) were abundant at 8 h and 24 h after infection. FHB1 was induced in fungal cells during murine macrophage infection and deletion of the gene resulted in hypersensitivity to nitrosative stress and attenuation of virulence in a mouse model (de Jesus-Berrios et al., 2003; Fan et al., 2005). A tag representing the gene encoding Cu, Zn superoxide dismutase (SOD1) was also elevated during lung infection, especially in the 8 h library. Superoxide dismutase has been shown to contribute to the virulence of C. neoformans and C. gattii (Cox et al., 2003; Narasipura et al., 2003; Narasipura et al., 2005). Interestingly, a tag representing the gene for the Sch9 protein kinase was elevated at 8 h. In C. neoformans, Sch9 modulates capsule formation and thermal tolerance, and contributes to virulence both independently of and in conjunction with the cAMP-PKA pathway (Wang et al., 2004). We also found that a tag for a gene for peptidyl prolyl cis-trans isomerase (Ess1), an enzyme that mediates the folding of target proteins, was elevated at 8 h. Ess1 is dispensable for growth, haploid fruiting and capsule formation, but is required for virulence in C. neoformans (Ren et al., 2005). A tag matching the gene for the FK506-resistant calcineurin B regulatory subunit was also identified in the early infection library. Calcineurin, a serine-threonine-specific calcium-activated phosphatase, is the target of the immunosuppressive drugs cyclosporine A and FK506, and this protein influences the mating, growth at 37°C and virulence in C. neoformans (Fox et al., 2001; Fox and Heitman, 2005). Finally, the in vivo SAGE libraries revealed that the expression of two genes encoding mannitol-1-phosphate dehydrogenase (Mpd1), an enzyme involved in mannitol synthesis, were elevated during infection (Table 2). Mannitol synthesis is thought to be important in cryptococcosis (Chaturvedi et al., 1996), and the MPD1 gene is induced under nitric oxide stress (Chow et al., 2007). Overall, the elevated expression of genes involved in stress response and functions known to contribute to virulence likely reflects C. neoformans adaptation to a hostile environment in the host.

A homolog of SNF1 influences melanin production and virulence

In evaluating the SAGE results for pulmonary infection, we noticed that the differential expression of genes for carbon metabolism, transport and stress showed similarities to the regulatory pattern defined for Snf1 in S. cerevisiae (Young et al., 2003; Hong and Carlson, 2007). Snf1 is a serine/threonine protein kinase that plays a major role in nutrient response and cellular metabolism, especially in gluconeogenesis and growth on alternative carbon sources (Celenza and Carlson, 1986). We therefore identified and characterized a candidate SNF1 homolog in C. neoformans to test the hypothesis that the Snf1 regulatory pathway influences the patterns of expression that we observed during infection (Fig. S3, Supplemental material). Initially, we confirmed that a cDNA of the SNF1 gene from C. neoformans was able to restore the growth of a S. cerevisiae snf1 mutant on sucrose and raffinose (data not shown). Subsequently, a SNF1 deletion mutant was constructed in C. neoformans to examine phenotypes related to carbon source utilization and virulence. The snf1 mutant as well as the WT and complemented strains were able to grow on either glucose or sucrose at 30°C and therefore did not show the growth defect on sucrose observed in S. cerevisiae (Fig. 3, Celenza and Carlson, 1984). A similar result was obtained for the plant pathogenic fungus Cochliobolus carbonum in which a SNF1 mutant was able to grow on both glucose and sucrose at 30°C (Tonukari et al., 2000). At 30°C, the C. neoformans snf1 strain also grew as well as the WT and complemented strains on arabinose, fructose, raffinose, galactose, mannose, lactate, acetate, ethanol, and glycerol (Fig. 3, and data not shown). Interestingly, different phenotypes were observed at 37°C in that the C. neoformans snf1 mutant displayed a noticeable reduction in growth on sucrose and ethanol, markedly reduced growth on acetate, and similar growth to the WT and complemented strains on other carbon sources (glycerol, lactate, galactose, raffinose and arabinose; Fig. 3 and data not shown).

Figure 3. Growth of the C. neoformans snf1 mutant on alternative carbon sources.

Cells of the WT strain (H99), the snf1 mutant and the complemented strain were spotted at decreasing concentrations (from left to right) on the media indicated on the left. YPD and YNB serve as controls for growth on glucose, and YNB with sucrose, sodium acetate or ethanol were used to test alternative carbon sources. The plates were incubated for two days at 30°C or 37°C as indicated. Reintroduction of the SNF1 gene into the snf1 mutant restored growth to the level of the WT strain.

Next, we examined the response of the snf1 mutant of C. neoformans to oxidative, osmotic, salt, and nitrosative stress. We did not observe significant differences between the mutant and the WT strain when cells were grown on YNB medium supplemented with NaCl, KCl, sorbitol, H₂O₂ or menadione, suggesting that SNF1 was not required for protection against these osmotic and oxidative stresses (Fig. 4 and data not shown). However, the mutant exhibited increased sensitivity to sodium nitrite at 37°C (Fig. 4). The complemented strain restored the WT level of growth on sodium nitrite at 37°C, indicating that a functional SNF1 gene is required to withstand nitrosative stress. We also tested the snf1 mutant for altered sensitivity to several drugs known to inhibit fungal growth. We found that the mutant displayed increased sensitivity to the antifungal drug amphotericin B that is commonly used to treat cryptococcosis, but a change in sensitivity to fluconazole was not observed (Fig. 4). We also found that the snf1 mutant was more sensitive to rapamycin, perhaps indicating an interaction with the TOR pathway. The mutant showed equal sensitivity to FK506 and cyclosporine A compared to the WT strain (data not shown). We next examined the three main virulence factors of the fungus: the ability to grow at 37°C, capsule formation and melanin production. We did not observe differences in growth or capsule formation between the wt, snf1 mutant and complemented strains at either 30°C or 37°C (data not shown). It was notable, however, that the snf1 mutant produced melanin at the same level as the WT strain at 30°C, but was unable to produce visible melanin in colonies at 37°C (Fig. 4). Complementation of the snf1 mutation with the SNF1 gene again restored the ability of cells to produce melanin at 37°C (Fig. 4).

Figure 4. Growth and melanin production by the C. neoformans snf1 mutant.

Cells of the WT strain H99, the snf1 mutant and the complemented strain were spot inoculated in dilutions from left to right on YNB medium to test the response to salt (1.5M NaCl), nitrosative stress (8 mM NaNO₂), amphotericin B (0.5 μg/ml), and rapamycin (10 μg/ml). The plates were incubated at 30°C or 37°C for two days. Melanin production was tested on L-DOPA medium after three days of incubation.

The motivation to examine the role of SNF1 was based in part on the pattern of gene expression observed in the SAGE data and also on evidence that alternative carbon sources are important for growth in the mammalian host (Lorenz and Fink, 2001). We therefore used quantitative real-time PCR to determine whether loss of Snf1 influenced the expression of a set of genes selected from the SAGE analysis. In parallel, we tested the influence of growth on low glucose or acetate on expression of the same genes. As shown in Fig. 5A, loss of Snf1 did not have an appreciable effect on the expression of the metabolic genes (e.g., ACS1, PCK1, and MLS1), or the acetate transporters ADY2 and ATO2, when cells were grown on 2% glucose, 0.2% glucose or 2% acetate. In contrast, growth of the WT strain at the lower glucose level resulted in a substantial elevation in the transcripts for all of the genes. The CFT1 gene showed a different pattern in that growth on the low glucose medium resulted in elevated expression and Snf1 was required for part of this response. The transcripts for a subset of the genes responded to acetate with increased levels; these included ACS1 (~2.6 fold), CGL1 (~2.5 fold), CFT1 (~3.7 fold), HXT1 (~3.7 fold) and MLS1 (~2.0 fold). Of these genes, loss of Snf1 had a positive influence on the level of the HXT1 transcript during growth on acetate. Overall, we concluded that glucose levels play a major role in regulating the genes that we observed by SAGE to have elevated transcripts during lung infection, and that Snf1 plays little role for these genes. These results require cautious interpretation, however, because we have not tested whether Snf1 influences gene expression during infection. We did find that SNF1 transcript levels were increased in the cells grown on low glucose or acetate (Fig. 5B), but they were not substantially increased in cells recovered from infected lungs (Fig. S3, Supplementary material).

Figure 5. Influence of glucose, acetate and deletion of SNF1 on the expression of genes identified by SAGE analysis.

(A) Expression of selected genes as measured by quantitative real-time PCR following growth under the conditions indicated in the legend on the right. (B) Expression of the SNF1 gene in cells grown in YNB medium with the carbon sources indicated on the X axis. (C) Expression of the genes encoding laccase or genes involved in the response to stress following growth under the conditions indicated on the right. The transcript of the ACT1 gene was used for normalization.

Given the influence of the snf1 mutation on melanin formation, the response to nitrosative stress and growth on acetate, we also examined whether Snf1 controlled the expression of genes related to these phenotypes. Loss of Snf1 did have an interesting influence on the expression of the LAC1 and LAC2 genes encoding the laccase enzymes for melanin production. The transcript for LAC2 was slightly elevated in WT cells grown on medium with 0.2% glucose or with acetate, and the transcript was more markedly elevated in the snf1 mutant grown on acetate. The LAC1 transcript was ~8.0 fold higher in WT cells grown on 0.2% glucose compared with cells grown on acetate or 2% glucose, and Snf1 was required for this enhanced transcript level. Interestingly, the LAC1 transcript levels were not influenced by growth on acetate in either strain. Complementation of the snf1 mutation with SNF1 restored the WT pattern of regulation (data not shown). These results suggest that part of the melanin defect in the snf1 mutant results from an influence on LAC1 expression. Snf1 did not appear to appreciably influence levels for stress-related genes (e.g., FHB1, SOD1, etc.) under any of the conditions (Fig. 5C).

Finally, we hypothesized that SNF1 would also be required for virulence because of the mutant phenotypes for growth on alternative carbon sources, melanin production and the response to nitrosative stress at 37°C. To test this idea, we infected A/Jcr mice with the WT strain, the snf1 mutant and the complemented strain by intranasal inoculation. All mice infected with the WT and complemented strains succumbed to infection by day 22–24, while those infected with the snf1 mutant survived to the end of the experiment at day 60 (Fig. 6). Thus the snf1 mutant was avirulent compared with the other strains (P < 0.001). At day 60, three mice infected with snf1 mutant were sacrificed and the lung and brain tissue were analyzed for fungal burden. Mutant cells were found at an average burden of 2.95 × 10⁵ CFU/g (SD = 1.73 × 10⁵) in lung tissue, but were not detected in the brain samples for any of the mice. Histopathology of brain tissue at day 60 also failed to detect cells of the snf1 mutant thus supporting the conclusion that the mutant was unable to disseminate to or persist in the brain (Fig. S4, Supplementary material). This observation is interesting, given the melanization defect of the snf1 mutant, because previous work showed that a laccase-defective, non-melanized strain was unable to escape from the lung (Noverr et al. 2004). To specifically test whether the snf1 mutant could disseminate beyond lung tissue, we inoculated mice by inhalation with the mutant and WT strains, and we monitored the numbers of fungal cells in the brains and lungs at day 17. This day was selected because it immediately precedes the time that the mice succumb to infection with the WT strain. This experiment revealed that the snf1 mutant was able to reach the brain, although the fungal burden for the mutant was quite low (average of 48.50 CFUs (SD = 14.64)) compared with the burden of WT cells (average of 1.87 × 10⁵ (SD = 1.02 × 10⁴)). For comparison, the lung tissue at day 17 contained an average of 1.96 × 10⁴ CFUs (SD = 3.90 × 10³) for the mutant and an average of 1.10 × 10⁸ CFUs (SD = 2.48 × 10⁷ CFUs) for the WT. Combined with our previous analysis, these results indicated that the snf1 mutant reached the brain in low numbers but failed to persist. Histopathology also confirmed the presence of the snf1 mutant in the lung (Fig. S4, Supplementary material).

Figure 6.

Analysis of virulence in a mouse inhalation model of cryptococcosis. (A) Ten A/Jcr mice were inoculated with the WT strain H99, the snf1 mutant or the complemented strain and monitored for signs of illness. The virulence of the snf1 mutant was significantly different from that of the WT strain (P < 0.0001) and the complemented strain (P < 0.0001) by the Krustal-Wallis test.

Discussion

Gene expression during pulmonary cryptococcosis

In this study, we employed SAGE to compare the C. neoformans transcriptomes at 8 and 24 h of murine pulmonary infection with expression patterns from in vitro culture or from CSF in a rabbit model of experimental meningitis. These comparisons provided two main insights into the pathogenesis of C. neoformans. First, lung infection resulted in elevated expression of fungal genes encoding functions for the production and utilization of acetyl-CoA. These included enzymes in the glyoxylate pathway, gluconeogenesis, β-oxidation and the conversion of pyruvate, ethanol and acetate to acetyl-CoA. Two candidate acetate transporters also showed elevated expression. This pattern was set against a background of elevated or reduced expression of specific components of the glycolytic pathway and the TCA cycle. As proposed for C. albicans (Fradin et al., 2003), this pattern may reflect variations in gene expression profiles in subpopulations of cells exposed to different host environments (e.g., extracellular and phagocytized cells). More interestingly, it could also reflect adjustments to the expression levels of pathway components to enhance the production of specific metabolic intermediates necessary for growth in host tissue. Given the early times of infection that we analyzed, it is also possible that some of the changes in gene expression for metabolic functions reflect the mobilization of glycogen and/or triglyceride stores to generate glucose and acetyl-CoA. A role for acetyl-CoA as a key metabolic intermediate was tested by deleting the ACS1 gene encoding a predicted acetyl-CoA synthase; the resulting mutant was unable to grow on acetate and showed a delayed ability to cause disease. It is likely that production of acetyl-CoA by other reactions (e.g., β-oxidation, other acetyl-CoA synthetases) partially compensated for loss of ACS1 (Fig. 1).

The second insight from this study was that the transcript profiles observed during lung infection suggested a role for the C. neoformans homolog of the yeast Snf1 kinase, a well-characterized regulator of alternative carbon source utilization and the stress response in S. cerevisiae. Deletion of SNF1 in C. neoformans revealed an involvement in carbon source utilization, virulence and the regulation of melanin production. Interestingly, the snf1 mutant was able to disseminate in quite low numbers to the brain following infection but was unable to persist, perhaps due to problems with carbon source utilization and/or the response to host defense mechanisms. SNF1 was specifically required for the elevated transcript levels of the LAC1 gene during growth on low glucose thus revealing a new regulatory function for melanin formation in C. neoformans. However, loss of SNF1 did not influence the expression of the genes for carbon source utilization that we identified in the SAGE analysis of lung infection. Instead, we found that the transcript levels of these genes were elevated in cells grown under low glucose conditions, a result consistent with the hypothesis that glucose is limiting during pulmonary infection.

Glucose limitation and alternative carbon sources during infection

The availability of glucose and local microenvironments in the mammalian host can have a marked influence on the expression of functions to exploit carbon sources other than glucose (e.g., acetate, lactic acid, fatty acids). For example, Barelle et al. (2006) examined gene expression in C. albicans during infection in light of the observation that glucose is present at sufficiently high levels in blood (0.06–0.1%) to limit expression of functions for alternative carbon source utilization. These studies indicated that C. albicans differentially regulates carbon assimilation pathways depending on the stage of infection (e.g., during interactions with phagocytic cells versus systemic infection) and the host tissue. Similarly, our comparison of the SAGE data for infections of the lung versus the CSF revealed that the patterns of gene expression were quite different, in support of the idea that there may be tissue-specific patterns of gene expression and adaptation for C. neoformans. The patterns of gene expression found by Thewes et al. (2007) for C. albicans during intraperitoneal infection and ex vivo liver infection also indicated that at least some of the cells experience a low glucose environment. These investigators found upregulation of the PCK1 gene encoding phosphoenolpyruvate carboxykinase for gluconeogenesis as well as upregulation of genes for the synthesis of acetyl-CoA from pyruvate. However, specific glycolytic enzymes and TCA cycle enzymes were also upregulated indicating that some of the cells were utilizing six-carbon sugars and respiration for energy production. We found similar sets of genes to be upregulated during cryptococcal pulmonary infection, and many of these were also regulated by glucose in vitro. The secretions from respiratory airways generally have a very low concentration of glucose in healthy individuals (<0.05 mM), and acute illness, inflammation or diabetes can elevate glucose levels (Philips et al, 2003). For example, bronchial aspirates from intubated patients were found to have a mean glucose concentration of 3.5 mM and, interestingly, glucose levels correlated with risk of infection for some bacterial pathogens (Philips et al., 2005). We hypothesize that many of the C. neoformans cells entering the lung experience glucose starvation. This idea is supported by the expression patterns that we observed and, in particular, by the identification of a candidate hexose transporter (HXT1) that was the most highly expressed gene (612 tags) in the 8 h lung library. This tag was present at much lower levels in the LIM (26 tags) and the CSF (2 tags) libraries. Our in vitro analysis indicated that the expression of the gene was dramatically higher at 0.2% glucose compared to 2% glucose. Thus the expression of this gene may be a useful sensor/readout of glucose levels (although other unknown factors could also influence the expression of the gene). The hxt1 mutant also showed precocious melanin formation and this phenotype is consistent with a role for Hxt1 in glucose sensing.

During cryptococcal meningoencephalitis, a low glucose environment is thought to occur in brain tissue and a three to five fold drop in glucose concentrations was found in the cerebrospinal fluid of infected rabbits (Perfect et al., 1980; Kwon-Chung et al., 2000; Rude et al., 2002). However, our analysis of the library from experimental meningitis in the rabbit model (Steen et al., 2003) did not reveal elevated expression of the glucose responsive genes that were observed in the pulmonary infection libraries. These results indicate that a more focused comparison of C. neoformans gene expression in different host tissues is needed.

The role of gluconeogenesis in the virulence of C. neoformans has been explored as a result of the identification of the PCK1 gene for phosphoenolpyruvate carboxykinase as a downstream target of Vad1, a DEAD-box RNA helicase (Panepinto et al., 2005). The tag for PCK1 was elevated in our SAGE analysis of pulmonary infection, but not in the macrophage internalized cryptococcal cells (Fan et al., 2005), suggesting a difference between the pulmonary infection and the intracellular environment. Interestingly, a pck1 mutant of C. neoformans was unable to grow on lactate or to cause disease (Panepinto et al 2005). This observation led to the proposal that 3-carbon sources, rather than 2-carbon molecules such as acetate, are preferred during infection. Our finding that a acs1 mutant did not grow on acetate or ethanol and had reduced virulence is consistent with this idea. Similarly, icl1 and mls1 mutants also can’t use acetate as a sole carbon source but retain virulence (Rude et al., 2002; Idnurm et al., 2007). Further characterization of the icl1 mutant revealed that the mutant also does not grow on ethanol and shows poor growth on lactate and glycerol (J. Perfect, unpublished results).

Roles for acetyl-CoA and acetate during adaptation to the lung environment

Acetyl-CoA is a central metabolite in the balance between carbohydrate metabolism and fatty acid catabolism (Fig. 1). Regulation of the balance may be a specific adaptation to the host environment by C. neoformans, as suggested by the abundance of transcripts for enzymes mediating the production and utilization of acetyl-CoA during pulmonary infection. It is possible that the expression pattern that we observed reflects the utilization of ethanol and acetate as carbon sources during infection. However, as mentioned above, mutants that cannot utilize acetate in culture retain virulence in C. neoformans. One can speculate that the production of acetyl-CoA may be specifically important during infection by C. neoformans because acetyl-CoA is an important precursor for the synthesis of chitin in the cell wall and for O-acetylation of the capsule via acetyltransferase activity (in addition to its key metabolic role). β-1,4 N-acetyl glucosamine has recently been shown to play a role in attachment of the capsule to the cell wall of C. neoformans and treatment with chitinase can release capsule polysaccharide (Rodrigues et al., 2008). Acetylation of capsule polysaccharide plays a role in the ability of capsule polysaccharide to inhibit neutrophil migration and a mutant lacking capsular acetyl groups is hypervirulent (Ellerbroek et al., 2004; Janbon et al., 2001). Thus, enhanced acetyl-CoA production may meet the demands of capsule-related biosynthetic functions during infection.

An additional intriguing aspect of pathogen metabolism during infection is the production of exported metabolites that may condition the host environment and contribute to virulence. The SAGE data for pulmonary infection are particularly interesting in light of the characterization of the metabolites produced by C. neoformans in culture and in cryptococcomas from rat lung and brain (Bubb et al., 1999; Wright et al., 2002; Himmelreich et al., 2003). These studies revealed that trehalose, mannitol, glycerol, acetate, ethanol and glycerophosphorylcholine were particularly abundant among the >30 metabolites that were detected. Wright et al., (2002) discussed the implications of these metabolites for virulence including the contribution of mannitol to defense against oxidative killing and the possibility that acetate acidifies the extracellular environment. In fact, direct measurement of cerebral cryptococcomas revealed a relatively low pH of 5.4 to 5.6. Wright et al. (2002) went on to demonstrate that C. neoformans supernatants at pH 5.5 induced necrosis in neutrophils, reduced superoxide production and influenced chemotaxis by these cells. Our observed elevated expression of transcripts for enzymes involved in the conversion of pyruvate to metabolites such as acetaldehyde, acetate and ethanol in C. neoformans during pulmonary infection is consistent with the appearance of these metabolites in cryptococcomas. In particular, we found tags for two candidate acetate transporters that were elevated during infection and we hypothesize that one or both of these proteins may contribute to acetate export. In addition to an influence on the host immune response, acidification of the local environment could also contribute to the availability of iron by triggering its release from transferrin. Indeed, Friedman et al. (2006) have shown that Staphylococcus aureus remodels its metabolism in response to iron limitation (a key feature of the host environment) to produce acidic products including lactic acid. For fungal pathogens, Thewes et al. (2007) also found a pattern of expression for pH-responsive genes in C. albicans during infection, as well as upregulated expression of transporters for iron, zinc and phosphate. These authors also noted the possible connection between pH modulation of the host environment and iron acquisition. As mentioned earlier, we found that the transcript for the iron permease Cft1 was elevated in pulmonary infection suggesting that C. neoformans is experiencing iron deprivation. We recently showed that iron acquisition during infection is partially dependent on Cft1 and that this permease is required for transferrin utilization (Jung et al., 2008).

Snf1 and virulence-related gene expression

In yeast, the Snf1 protein kinase is required for the metabolic shift that occurs upon glucose depletion, for growth on alternative carbon sources, and for the response to a variety of environmental stresses such as sodium and lithium salts, alkaline pH, heat shock, and hyperosmolarity (Hardie et al., 1998; Kemp et al., 1999; Kemp et al., 2003; Vyas et al., 2001), (Alepuz et al., 1997; Hong and Carlson, 2007; Portillo et al., 2005; Ye et al., 2006). We found that Snf1 in C. neoformans plays a role in the growth response to acetate, ethanol and sucrose, and that sensitivity of the snf1 mutant was noted for nitrosative stress and for the drugs rapamycin and amphotericin B. We found that all of the snf1 phenotypes in C. neoformans were manifested at the host temperature of 37°C, but not at 30°C. In S. cerevisiae, a snf1 mutant has reduced thermotolerance, and carbon source utilization mutants in C. albicans have more serve defects at 37°C (Thompson-Jaeger et al., 1991; Ramirez and Lorenz, 2007). In C. neoformans, snf1 mutant strains grew as well as the WT strain on media (either YPD or YNB) with glucose (or glycerol, galactose, raffinose, arabinose) as the sole carbon source at 37°C, indicating that loss of SNF1 did not generally reduce thermotolerance, but rather influenced specific phenotypes.

SNF1 has been implicated in the virulence of plant pathogenic fungi. For example, the SNF1 gene in Cochliobolus carbonum is required for the expression of cell-wall-degrading enzymes and for disease symptom formation on maize (Tonukari et al., 2000). The SNF1 gene in Fusarium oxysporum, a pathogen that causes vascular wilt disease in over 100 cultivated plant species, regulates the transcription of genes encoding cell wall degrading enzymes and virulence on both Arabidopsis thaliana and Brassica oleracea (Ospina-Giraldo et al., 2003). So far there has been no description of a role for SNF1 in the virulence of fungal pathogens of animals. A SNF1 homologue in C. albicans is essential for viability, and disruption of one allele resulted in morphological changes and decreased growth, but did not influence virulence (Petter et al., 1997). The discovery of a role for SNF1 in the virulence of C. neoformans reveals a new regulatory connection between carbon source utilization and growth in mammalian hosts. Part of the virulence defect may be due to reduced melanin production at 37°C. Laccase expression in C. neoformans is repressed by elevated temperature and enhanced by glucose starvation as well as copper, iron and calcium (Zhu and Williamson, 2004). Laccase expression also occurs during early pulmonary infection (Garcia-Rivera et al., 2005). Both LAC1 and LAC2 contribute to melanin production and both are regulated by the cAMP pathway; differences have been noted for the two genes, however (Missall et al., 2005; Pukkila-Worley et al., 2005). Notably, LAC2 but not LAC1 is regulated in response to oxidative and nitrosative stresses in manner that is influenced by the TSA1 gene encoding a thiol peroxidase (Missall et al., 2004, 2005). Missall et al., (2005) presented a model for the regulation of LAC2 expression in response to nitric oxide stress and proposed that a stress-activated protein kinase might function downstream of Tsa1.

In terms of contributions to virulence, the laccases produced by the LAC1 and LAC2 have different but overlapping substrate specificities that may allow the utilization of different diphenolic substrates in brain tissue. The LAC1 gene appears to make the larger contribution to virulence in a mouse model of cryptococcosis, although mice still succumb to infection with a lac1 mutant (Pukkila-Worley et al., 2005). Given these results, it is likely that the Snf1 protein influences other aspects of virulence because a snf1 mutant fails to cause disease. It is possible that the melanin defect contributes to the observed poor dissemination of the snf1 mutant from the lung to the brain because melanin formation is known to influence this process (Noverr et al., 2004). However, the snf1 mutant also failed to persist in brain tissue, perhaps indicating that Snf1 influences the expression of additional virulence factors. For example, Snf1 might regulate processes that are important for virulence in C. neoformans such as gluconeogenesis (Panepinto et al. 2005), as well as control the expression of cell surface factors that mediate interactions with the host and/or influence the response to host defenses. Overall, the identification of a role for Snf1 presents opportunities to further characterize the role of the protein in the regulation of central metabolism, the utilization of alternative carbon sources and the response of the fungal cells to the stressful conditions of the host environment.

Experimental procedures

Strains, plasmids and media

The serotype A strain H99 (C. neoformans var. grubii) and the S. cerevisiae strains W303-1A (MATa ade2-1 trp1-1 his3–11, 15 can1-100 ura3-1 leu2–3, 112) and MCY4908 (W303-1A SNF1Δ10) were used in the study. The strains were maintained on YPD medium (1% yeast extract, 2% peptone, 2% dextrose, and 2% agar). Selectable markers for the transformation of C. neoformans were from plasmid pCH233 (nourseothricin resistance) and pJAF1 (neomycin resistance). Plasmid BPH618 was used for yeast transformation. YPD plates containing neomycin (200 μg/ml) were used to select C. neoformans transformants and YPD plates containing noursethricin (100 μg/ml) were used to select the SNF1-complemented transformants. S. cerevisiae transformants were selected on YNB agar (yeast nitrogen base without amino acids and uracil) supplemented with 2% glucose and the other required nutrients. YNB agar (without amino acids and uracil) was supplemented with 2% glucose, 2% sucrose, 2% raffinose, or 2% maltose, and used to examine the growth of the yeast snf1 mutant. YPD and/or YNB plates (YNB with amino acids) supplemented with different inhibitors or chemicals were used for phenotypic experiments. Escherichia coli was grown on LB broth or agar supplemented with 100 μg/ml of ampicillin at 37°C.

Isolation of C. neoformans cells from mouse lungs for SAGE analysis

The C. neoformans cells for mouse inoculation were grown overnight in YPD in a 30°C shaker, washed with phosphate-buffered saline (PBS) and resuspended in PBS at a concentration of 8.2 × 10⁷ cells/50 μl. A total of 20 female A/Jcr mice were anesthetized with ketamine and xylazine, inoculated by nasal inhalation and subjected to bronchoalveolar lavage at 8 and 24 h. At each time point, the treated mice were euthanized, a small incision was made in the trachea, and a capillary tube was inserted toward the lungs. The tube was secured by silk thread and a series of 0.5 ml aliquots of ice-cold water were flushed into the lungs. A total of 10 ml of water was used per mouse and the lavage fluids from each inoculation group were pooled. The cells were washed twice with ice-cold water, frozen at −80°C and lyophilized for RNA isolation. In total, 3.0 × 10⁸ cells were obtained at 8 h and 2.6 × 10⁸ cells were collected at 24 h.

SAGE library construction, sequencing and analysis

SAGE library construction, sequencing and analysis were as previously described (Steen et al., 2002, 2003; Hu et al., 2007). Briefly, RNA was isolated from lyophilized cells by vortexing with glass beads (3.0 mm, acid-washed and RNase-free) for 15 min. in 15 ml of TRIZOL extraction buffer (Invitrogen, Carlsbad, California). The mixture was incubated for 15 min at room temperature, total RNA was isolated according to the manufacturer’s instructions and RNA quality was assessed by agarose gel electrophoresis. Total RNA was used directly for SAGE library construction as described by Velculescu et al., (1995) using the I-Long SAGE kit (Invitrogen). The tagging enzyme for cDNA digestion was NlaIII and 29 PCR cycles were performed to amplify the ditags during library construction. Colonies were screened by PCR (M13F and M13R primers) to assess the average clone insert size and the percentage of recombinants. Clones from the libraries were sequenced by BigDye primer cycle sequencing on an ABI PRISM 3700 DNA analyzer. Sequence chromatograms were processed using PHRED (Ewing and Green, 1998), and vector sequence was detected using Cross_match (Gordon et al., 1998). Fourteen-base-pair tags were extracted from the vector-clipped sequence, and an overall quality score for each tag was derived based on the cumulative PHRED score. Duplicate ditags and linker sequences were removed as described previously (Steen et al., 2002). Only tags with a predicted accuracy of ≥99% were used, and statistical differences between tag abundance in different libraries were determined using the methods of Audic and Claverie (1997).

The libraries yielded 21,510 (8 h) and 20,129 (24 h) tags. An overview of the abundance classes for both SAGE libraries is presented in Table S1, Supplementary material, with both the number of different tag sequences and the total number of tags present in each class for the cells from lung tissue at 8h or 24 h after infection. An overview of pair wise analyses of differential expression for all of the libraries is presented in Table S2, Supplementary material, and the 100 most abundant tags in each library are listed in Table 3, Supplementary material. All libraries were normalized to 20,000 to allow direct comparisons, and the tags that appeared less than once in any given library were removed. The EST database available for strain H99 at the University of Okalahoma’s Advanced Center for Genome Technology (http://www.genome.ou.edu/cneo.html) was used for the preliminary assignment of tags to genes. When an EST sequence could not be identified for a particular tag, the genomic sequence for H99 at the Duke University Center for Genome Technology (http://cgt.genetics.duke.edu/data/index.html) and the Broad Institute (http://www.broad.mit.edu/cgi-bin/annotation/fungi/cryptococcus_neoformans/) was used to identify contigs with unambiguous tag assignments. Note that a limitation of the SAGE approach is that some transcripts are not detected because of low abundance and/or the absence of an NlaIII site for transcript processing.

Complementation of a S. cerevisiae snf1 mutation with the SNF1 gene of C. neoformans

To obtain a cDNA for SNF1, total RNA was isolated from frozen C. neoformans cells. The cDNA was synthesized using random hexamer priming and Superscript transcriptase II (Invitrogen Canada). A 2,270 bp PCR product was obtained using primers SNF1-cDNA-5 and SNF1-cDNA-6, cloned into TOPO-TA vector (pSNF1c-topo) and then subcloned into pBPH618 to obtain pCnSNF1c. The same pair of primers was used to amplify the SNF1 gene from genomic DNA of strain H99 and a PCR product of 2950 bp was obtained. This product was cloned into a TOPO-TA vector (pSNF1g-TOPO) and subcloned into pBPH618 to create pCnSNF1g. Both SNF1 inserts in pCnSNF1c and pCnSNF1g were sequenced to confirm the presence of the C. neoformans gene. To complement yeast snf1 mutant with C. neoformans SNF1 homologue, pCnSNF1c and pBPH618 (empty vector) were transformed into strain MCY4908 by PEG and lithium acetate treatment. Transformants were selected on SCD (dexose)-URA and 5-FOA plates for uracil prototrophy, and on SCS (sucrose) for sucrose utilization. The sequences of the primers for this study are given in Table S6, Supplementary material.

Deletion of the SNF1 gene in C. neoformans

A snf1::NEO deletion allele was constructed using a modified overlap PCR procedure (Davidson et al., 2002; Yu et al., 2004). Briefly, the primers SNF1-1/SNF1–3 and SNF1–4/SNF1–6 (Table S6, Supplementary material) were used with genomic DNA to obtain the left and right arms for the deletion construct. The NEO selectable marker was amplified using primers SNF1–2/SNF1–5 and the plasmid pJAF1. The SNF1:NEO allele results in the deletion of the complete open read frame of SNF1 (2926 bp). The resulting PCR product (3854 bp) was used to transform strain H99 by biolistic transformation (Davidson et al., 2000). Transformants were grown on YPD plates containing neomycin and screened by colony PCR with Extaq polymerase (Takara) using primer pairs SNF1–7/SNF1–8 and SNF1–9/hug-Neo. Primer SNF1–9 was designed from the region upstream of SNF1 and hug-Neo was designed for the NEO gene. Transformants in which WT allele was replaced were confirmed by genomic hybridization as described (Hu and Kronstad, 2006). One mutant designated SNF1–22 contained the deletion allele and was studied further. For complementation of the deletion mutation, the SNF1 gene was amplified by PCR using primers SNF1-rec-for and SNF1-rec-rev, and genomic DNA from strain H99. The 4627 bp product was digested with XbaI and cloned into the XbaI site of pCH233, creating the plasmid pSNF1rec. The strain SNF1–22 was transformed with pSNF1rec by biolistic transformation, and transformants were selected on YPD containing nourseothricin (100 ug/ml). Reintroduction of SNF1 was confirmed by colony PCR and genomic hybridization.

Quantitation of gene expression during pulmonary infection or during growth in culture

Total RNA from frozen cells collected from the lungs of infected mice was obtained as described above (from an independent experiment), and DNA was removed by treatment with DNase I for 30 min at 25°C. Subseqently, cDNA was synthesized using random hexamers and Superscript transcriptase II (Invitrogen Canada). The resulting cDNA was used for real-time PCR with primers targeted to the 3′ regions of transcripts. Primers were designed using PrimerExpress v3 (Applied Biosystems). The Power SYBR Green PCR mix (Applied Biosystems) was used according to the manufacture’s recommendations. An Applied Biosystems 7500 Fast Real-time PCR system was used to detect and quantify the PCR products with the following conditions: incubation at 95°C for 10 min followed by 40 cycles of 95°C for 15 sec, and 60°C for 1 min. The cDNA of the ACT1 gene was used to normalize the data. Dissociation analysis on all PCR reactions confirmed the amplification of a single product for each primer pair and the absence of primer dimerization. Relative gene expression was quantified using SDS software 1.3.1 (Applied Biosystems).

Growth conditions for examining the influence of reduced glucose levels, acetate and loss of SNF1 were as follows. Cells (WT, snf1 and snf1::SNF1) were grown in YNB+2% glucose at 37 °C to mid log phase, and washed with the 37 °C pre-warmed medium. Equal numbers of cells (1.5 × 10⁸ cells) were transferred to either YNB+2% glucose or YNB + 0.2% glucose or YNB + 2% acetate, and cultured for 5 hours at 37 °C prior to RNA isolation. The sequences of the primers for the PCR analysis are given in Table S7, Supplementary material.

Stress and drug response assays

To examine the response of C. neoformans WT, snf1 and snf1::SNF1 strains to various stress conditions, exponentially growing cultures were washed, resuspended in H₂O, and adjusted to a concentration of 0.2 × 10⁵ cells/μl. The cell suspensions were diluted 10-fold serially, and 5 μl of each dilution was spotted onto YPD and/or YNB plates supplemented with different chemicals. Plates were incubated for two to five days (depending on the conditions) at 30°C or 37°C, and photographed. The responses of strains to oxidative, nitrosative, osmotic stress and to agents that challenged cell wall integrity were examined. The specific assays were performed on YPD and/or YNB plates supplemented with or without 1.2 M KCl, 1.0 M or 1.5 M NaCl, 75 mM LiCl, 0.1% SDS, 0.5 mg/ml Congo Red, 3 μg/ml menadione, 0.5 mM H₂O₂ and 2 mM, 4 mM or 8 mM sodium nitrite (NaNO₂). For carbon source utilization experiments, YNB agar (yeast nitrogen base, 6.7% g/liter) was supplemented with one of the following carbon sources: 0.5% or 2% glucose, 0.5 or 2% sucrose, 2% raffinose, 2% maltose, 2% galactose, 0.5% or 2% sodium acetate, 0.5% or 2% ethanol, 0.5% or 2% glycerol, 0.5% or 2% lactic acid. Sensitivity to inhibitors of calcineurin and TOR (target of rapamycin) signalling were examined by spotting the cell dilutions on YNB plates containing 2 μg/ml or 4 μg/ml FK506, 125 μg/ml or 250 μg/ml Cyclosporin (CsA), or 1 μg/ml or 10 μg/ml rapamycin. The antifungal drugs amphotericin B (0.5 μg/ml) and fluconazole (0.5 μg/ml or 1 μg/ml) were also tested. The figures show only the conditions where the mutant gave a different response compared to the WT strain.

Capsule formation and melanin production

LIM (low-iron medium) was used to examine capsule formation. A single colony from a YPD plate for each strain was cultured overnight at 30°C in liquid YPD medium. Cells were harvested and diluted in low iron water, and 10⁶ cells were added into 3 ml of LIM for further incubation at 30°C for 48 h. After incubation, the capsule was stained by India ink and examined by differential interference microscopy (DIC). To examine melanin production, a single colony of each strain was incubated overnight at 30°C in liquid YPD medium, washed and diluted to 2×10⁴ cells per ml. Five μl of serial dilutions from this stock were spotted onto L-DOPA plates containing 0.1% glucose. The plates were incubated for three days at 30°C or 37°C, and melanin production was monitored and photographed daily.

Virulence assays

For virulence assays, female A/Jcr mice (4 to 6 weeks old) were obtained from Jackson Laboratories (Bar Harbor, Maine). The C. neoformans cells for inoculation were grown in YPD medium overnight at 30°C, washed in PBS and resuspended at 1.0×10⁶ cells/ml in PBS. Inoculation was by intranasal instillation with 50 ul of cell suspension (5.0×10⁴). The status of the mice was monitored twice per day post-inoculation. Differences in virulence were statistically assessed with the Krustal-Wallis test. For histopathology, infected mice were euthanized by CO₂ inhalation, and organs were excised and placed in 10% buffered formalin. Fixed organs were sent to Wax-It Histology Services (Vancouver, B.C. Canada) for sectioning and staining with Mayer’s Mucicarmine. For determination of the fungal load in organs, infected mice were euthanized by CO₂ inhalation and organs were excised, weighed and homogenized in 1mL PBS using a MixerMill (Retsch). Serial dilutions of the homogenates were plated on Sabouraud dextrose agar plates containing 35 μg/mL chloramphenicol and colony-forming units were counted after an incubation for 48 h at 30°C.