GMP Synthase Is Required for Virulence Factor Production and Infection by Cryptococcus neoformans*

Over the last four decades the HIV pandemic and advances in medical treatments that also cause immunosuppression have produced an ever-growing cohort of individuals susceptible to opportunistic pathogens. Of these, AIDS patients are particularly vulnerable to infection by the encapsulated yeast Cryptococcus neoformans. Most commonly found in the environment in purine-rich bird guano, C. neoformans experiences a drastic change in nutrient availability during host infection, ultimately disseminating to colonize the purine-poor central nervous system. Investigating the consequences of this challenge, we have characterized C. neoformans GMP synthase, the second enzyme in the guanylate branch of de novo purine biosynthesis. We show that in the absence of GMP synthase, C. neoformans becomes a guanine auxotroph, the production of key virulence factors is compromised, and the ability to infect nematodes and mice is abolished. Activity assays performed using recombinant protein unveiled differences in substrate binding between the C. neoformans and human enzymes, with structural insights into these kinetic differences acquired via homology modeling. Collectively, these data highlight the potential of GMP synthase to be exploited in the development of new therapeutic agents for the treatment of disseminated, life-threatening fungal infections.

In the past four decades there has been a dramatic escalation in the number of immunocompromised individuals (1). Although many of these are due to advances in immunosuppressive and chemotherapeutic technologies, the largest cohort is a direct result of the AIDS pandemic. Foreshadowed in 1981 with the occurrence of opportunistic Pneumocystis or human herpesvirus 8 infections in previously healthy individuals in Los Angeles (2,3), AIDS and the lentivirus causing the disease soon gained broad public awareness (4 -6). With the advent of commercial blood testing, national blood bank screening programs were commenced in an effort to slow the spread of this emerging pandemic (7), and in 1987 the first treatment, zidovudine, marked what was thought to be an end to the crisis (8,9). It was not, as was hoped, a miracle drug; the spread of AIDS continued across the globe and is now believed to have infected over 70 million people and killed 55 million to date (10).
Since the discovery of the very first patients identified with Pneumocystis pneumonia, opportunistic fungal pathogens have been tightly linked with the AIDS pandemic. One of the key fungi often encountered in this context is Cryptococcus neoformans, a basidiomycete yeast responsible for cryptococcosis and a major cause of AIDS-related mortality (11). Most commonly found associated with purine-rich bird guano, spores or desiccated yeast cells from this environmental niche are inhaled into the lungs where, in an immunocompromised individual, the fungus can disseminate to the purine-poor central nervous system to cause meningoencephalitis.
The fundamental treatment for cryptococcosis has not changed significantly in over two decades and consists of induction with amphotericin B and flucytosine followed by consolidation and maintenance phases employing fluconazole (12)(13)(14)(15)(16). Each of these therapeutic agents exploit an aspect of fungal physiology that differs from the human host: the presence of ergosterol rather than cholesterol in the cell membrane, the existence of the pyrimidine salvage enzyme cytosine deaminase, or changes in the sterol biosynthetic pathway, respectively. Even so, each of these agents still cause side effects, epitomized by the compromised renal function typical of amphotericin B use.
The similarity in physiology between human host and fungal pathogen means there are few gross differences that can be easily targeted via rational drug design. However, differences such as slight changes in the active site of enzymes in a highly conserved essential pathway may be exploited, as typified by fluconazole. One such potential target is de novo purine biosynthesis, the pathway in which the process of rational drug design was pioneered (17)(18)(19). Proliferating cells require vast quantities of ATP and GTP to meet the demands of replication, transcription, and energy metabolism. Purine biosynthesis is therefore essential in rapidly dividing immune cells, cancers, or infecting microbes. The pathway is already a target for anticancer drugs (such as mercaptopurine and lometrexol) and immunosuppressants (such as mycophenolic acid) (20 -24). However, the investigation of purine metabolism as an antifungal target has been limited (25)(26)(27).
Purine metabolism is of particular interest in C. neoformans because of the gross disparity in purine concentrations in bird guano compared with the central nervous system of the infected human host (28 -30). The importance of de novo biosynthesis rather than scavenging while the fungus infects this purine-poor environment was highlighted by the discovery that IMP dehydrogenase, catalyst of the rate-limiting and first committed step in the de novo GTP biosynthesis pathway, is essential for C. neoformans virulence in a murine model. Importantly, structural studies of this enzyme revealed potentially exploitable characteristics of the fungal enzyme not shared with the human isoforms (25).
IMP dehydrogenase is not the only enzyme in de novo GTP biosynthesis that has potential as an antimycotic target. Following IMP dehydrogenase in the guanylate branch of de novo purine biosynthesis is GMP synthase (E.C. 6.3.5.2). Originally identified in ground rabbit bone marrow (31) and pigeon liver (32), the key work that underpins our current understanding of GMP synthase was performed in Escherichia coli (33,34), and this system continues to be the source for a detailed biochemical understanding of the activity of the enzyme (35)(36)(37).
GMP synthase is a monofunctional enzyme in all organisms studied so far, with two catalytic modules working in concert to ensure efficient amination of xanthine monophosphate (XMP) 2 to GMP. The magnesium-dependent ATP pyrophosphatase (ATP-PPase) domain reversibly adenylates XMP to form a covalent O 2 -adenyl-XMP intermediate; this activates the C 2 carbon of XMP for attack by the amide side chain of glutamine, liberated by the class I glutamine amidotransferase (GATase) domain (33,34,38).
In contrast, few studies of GMP synthase have been performed in fungi. The GMP synthase gene has been identified and mutated in Saccharomyces cerevisiae, Candida albicans, and Aspergillus fumigatus; as expected, these mutants were all guanine auxotrophs (26, 44 -46). Furthermore, the GMP synthase mutants of C. albicans and A. fumigatus were avirulent in murine infection models (26). No kinetics or structural studies of fungal GMP synthases are available.
Here, we describe the characterization of GMP synthase from C. neoformans, showing the potential of the enzyme as a broad spectrum antifungal target. Using genetic techniques, we demonstrate that GMP synthase is required for C. neoformans virulence factor production and successful infection of the host. With the aid of biochemical analyses and structural modeling, we identify key functional differences between GMP synthase from C. neoformans and humans, thereby delineating its suitability as a potential antifungal drug target.

Results
Identification of the GMP Synthase-encoding Gene in C. neoformans-To begin our characterization of GMP synthase from C. neoformans, the corresponding gene in the type strain H99 genome was identified via a reciprocal best hit BLAST analysis employing the S. cerevisiae ortholog GUA1 (45). A single hit was observed, indicating that the gene is present in single copy. Located on chromosome 11, this locus was designated as CNAG_018770 in the published H99 genome (47); subsequently employing CNAG_018770 as the query sequence in a BLAST search of the S. cerevisiae genome identified GUA1 as the only hit in that species. The gene CNAG_018770 has therefore been named GUA1 after the S. cerevisiae ortholog, whose predicted product is 65% identical at the amino acid level. In comparison, Gua1 is 50% identical to GuaA in E. coli, 35% identical to human GuaA isoform 1, 23% identical to the human GuaA, and 23% identical to human GuaA isoform 2.
Gua1 Is Essential for Guanine Protrophy in C. neoformans-To verify that the identified gene designated as GUA1 encodes GMP synthase, we performed a biolistic targeted gene deletion in C. neoformans type strain H99. Consistent with our bioinformatic predictions, the gua1⌬ strain could not grow on YNB minimal medium, but growth was restored (albeit incompletely) upon the addition of guanine (Fig. 1B). The gua1⌬ strain was not, however, able to utilize the guanine naturally present in rich YPD medium, nor was growth restored by the addition of exogenous guanine to YPD (data not shown); this unusual phenotype was consistent with that previously reported for the IMP dehydrogenase imd1⌬ mutant (25). Reintroducing a wild-type copy of the gene to create strain gua1⌬ϩGUA1 restored growth on both YNB and YPD medium (Fig. 1B). Together, these data are consistent with GUA1 encoding GMP synthase.
To gain insight into potential growth defects that may be exhibited during infection independent of guanine starvation, the effect of deleting GUA1 on the growth rate was investigated in RPMI 1640 medium supplemented with 1 mM guanine. Again, even in the presence of guanine, the gua1⌬ mutant displayed slower growth compared with both wild type and the complemented gua1⌬ϩGUA1 strain because of an extended lag phase (Fig. 1C). Furthermore, the mutant reached stationary phase at an optical density lower than the wild-type or complemented strains. The delayed and reduced growth of the mutant, even in the presence of guanine concentrations far exceeding the limited amounts in mammalian cerebrospinal fluid (ϳ0.5 M (29)), suggested the mutant was likely to exhibit severe attenuation or avirulence in infection models.
Loss of GMP Synthase Affects Production of C. neoformans Virulence Determinants-During the infection process, C. neoformans relies on a number of key virulence factors that play an important role in protecting the pathogen from host defenses, facilitating dissemination and growth. To observe the effects of the gua1⌬ mutation on the production of these virulence factors while limiting the confounding influence of guanine starvation influencing growth, in vitro virulence factor production assays were performed in the presence of exogenous guanine. The complex polysaccharide capsule protects the fungal cells against phagocytosis by immune cells during infection. When grown in guanine-supplemented capsule-inducing RPMI 1640 medium, the gua1⌬ mutant produced significantly less capsule than wild type and the complemented gua1⌬ϩGUA1 strain after 30 h at both 30 and 37°C (p Ͻ 0.0001; Fig. 2A). The virulence factor melanin protects C. neoformans against the effects of oxidants, such as those produced by phagocytic cells. When plated on guanine-supplemented L-3,4-dihydroxyphenylalanine (L-DOPA) medium, production of melanin was delayed; this delay was more pronounced at 37°C. Production of proteases enables C. neoformans to degrade host tissue and facilitate dissemination, as well as obtain nutrients for growth. On guanine-supplemented protease-inducing medium, protease activity was undetectable in the gua1⌬ mutant at either 30 or 37°C (Fig. 2). In summary, loss of GMP synthase was associated with delayed or abolished production of the known virulence factors capsule, melanin, and proteases, and this was more pronounced at human body temperature, providing further evidence that the gua1⌬ mutant was likely to exhibit impaired virulence.
GMP Synthase Is Critical for C. neoformans Virulence in Both Nematode and Murine Models-The compromised growth rate and virulence factor production of the gua1⌬ mutant strongly suggested the strain would be unlikely to successfully cause infection. Evaluation of virulence was first performed in the nematode Caenorhabditis elegans, a natural predator of C. neoformans (64). Nematodes were introduced onto a lawn of wild-type, gua1⌬, or gua1⌬ϩGUA1 cells grown on rich brainheart infusion (BHI) medium, minimal nematode growth medium (NGM), and minimal NGM supplemented with 1 mM guanine. No difference in virulence was observed between the wild-type, gua1⌬, and gua1⌬ϩGUA1 strains when grown on BHI or NGM supplemented with guanine (p Ͼ 0.1; Fig. 3A). Given the abundance of guanine in bird guano, this result was consistent with what could be expected in the environmental niche where C. neoformans encounters this predator. However, on NGM lacking guanine supplementation, the gua1⌬ mutant was avirulent (p Ͻ 0.0001; Fig. 3A).
The nematode data indicated the importance of GMP synthase for virulence was influenced by the abundance of guanine available. Given the low concentrations of available purines in cerebrospinal fluid (28 -30), we investigated the effect of gua1⌬ on virulence using a murine inhalation model of crypotococcosis. Mice infected with the wild-type and gua1⌬ϩGUA1 strains displayed equivalent progression of disease, succumbing to the infection within 20 days (Fig. 3B). In contrast, rather than losing weight and displaying other disease-associated symptoms, all mice infected with the gua1⌬ strain gained weight and were healthy until the 60-day end point of the experiment (Fig. 3B). Furthermore, fungal burden analysis from sacrificed animals showed the infection had been cleared from all mice infected with the gua1⌬ mutant, similar to previous observations (25).
C. neoformans GMP Synthase Enzyme Kinetics Show Differences to the Human Ortholog in Binding Cooperativity-Given the importance of Gua1 during murine infection, we expanded our analyses to investigate the biochemical function of C. neoformans GMP synthase. To identify possible functional differences between the fungal and human enzymes that could be exploited in therapeutic agent development, His-tagged C. neoformans Gua1 was expressed in E. coli and purified, and the histidine tag was removed prior to use of recombinant Gua1 in steady-state kinetic analysis.
The dual catalytic action of GMP synthase is required for the amination of XMP to GMP and involves two main steps: initially adenyl-XMP is produced by the ATP pyrophosphatase (ATP-PPase) domain in the presence of Mg 2ϩ , and this product then reacts with ammonia produced by the hydrolysis of glutamine that is catalyzed by the N-terminal glutamine amidotransferase (GATase) (33,38). The amination of XMP in C. neoformans GMP synthase protein displayed Michaelis-Menten kinetics with a K m of 65.9 Ϯ 13.0 M and a k cat of 0.4 s Ϫ1 for the overall reaction. ATP adenylation also exhibited Michaelis-FIGURE 1. The de novo and salvage purine biosynthesis pathway and GMP synthase role in GTP biosynthesis. A, IMP dehydrogenase and GMP synthase catalyze the first two steps of the guanylate branch of de novo purine biosynthesis. B, 10-fold serial dilutions of indicated strains were spotted onto YNB medium supplemented with specified purines (1 mM) and incubated for 2 days at 30°C. C, strains were grown for 6 days in RPMI 1640 medium supplemented with 10% serum and 1 mM guanine at 30°C, with A 600 readings collected every 12 h. FEBRUARY 17, 2017 • VOLUME 292 • NUMBER 7

Gua1 Is Required for C. neoformans Infection
Menten kinetics and had a K m of 77.5 Ϯ 6.0 M. In contrast, kinetic data for Mg 2ϩ , which is required as a co-factor for the reaction of XMP to adenyl-XMP, revealed a sigmoidal response to increases in substrate concentration, which is indicative of cooperative binding and was best fitted with a Hill coefficient (n) of 2. Compared with previously identified GMP synthases, the kinetic parameters of the C. neoformans enzyme were most similar to those of the Mycobacterium tuberculosis GMP synthase, which was reported to have similar binding affinity. For XMP, the binding affinity was only moderately lower between C. neoformans (65.9 Ϯ 13.0 M) and M. tuberculosis (45.0 Ϯ 1.0 M) and different between human (35.6 Ϯ 1.8 M) and C. neoformans, and unlike M. tuberculosis and human enzymes, the C. neoformans enzyme did not show cooperative binding of XMP (58,59). The K m of ATP is moderately higher than reported in M. tuberculosis (27 Ϯ 2 M) but considerably lower than human GMP synthase (132 Ϯ 7 M). The affinity for glutamine and Mg 2ϩ in C. neoformans GMP synthase was most like M. tuberculosis (Table 1). However, the magnitude of the positive cooperativity observed for XMP binding in the human enzyme was moderate, with a Hill (n) coefficient of 1.48 (59), whereas M. tuberculosis GMP synthase had a Hill (n) coefficient of 2.4 (58). Overall, the human GMP synthase protein had a significantly higher (ϳ12ϫ) turnover number than the C. neoformans protein, and the concentration of glutamine required to reach saturation was almost three times higher for C. neoformans (1130 Ϯ 162 M) compared with human GMP synthase enzyme (406 Ϯ 49 M) (58, 59), confirming potentially important kinetic differences in these enzymes.
ECC1385 Is an Inhibitor of C. neoformans GMP Synthase-Given the biochemical differences uncovered above and the importance of GMP synthase in virulence in C. neoformans, we next investigated the ability to phenocopy the gua1⌬ phenotype with a chemical inhibitor. The best known inhibitor of GMP synthase is 6-diazo-5-oxo-L-norleucine (DON), a well characterized glutamine agonist that targets several glutamine binding proteins, in particular phosphoribosylformylgly-FIGURE 2. Loss of GUA1 compromises virulence factor production. A, C. neoformans strains were incubated in RPMI 1640 medium, 10% fetal bovine serum, and 1 mM guanine at 30 and 37°C. At 30 h, the cells were stained with India ink. The relative capsule diameter value shows the mean, and the error bars represent S.E. ****, p Ͻ 0.0001. B, melanin production was determined on L-DOPA medium with strains incubated at 30 and 37°C for 48 h. C, protease production was determined on YNB medium with amino acids and 0.1% bovine serum albumin plates with strains at 30 and 37°C for 48 h. . Virulence of gua1⌬ mutant in nematode and murine host systems. A, Bristol N2 C. elegans were transferred onto lawns of the indicated C. neoformans strains grown on BHI medium, minimal medium, and minimal medium supplemented with 1 mM guanine and incubated at 28°C, and survival was monitored. For minimal medium, the gua1⌬ mutant strain was avirulent, differing from the wild-type or gua1⌬ϩGUA1 strains (p Ͻ 0.0001). No significant difference in virulence was found between strains when grown on rich medium or minimal medium supplemented with guanine (p Ͼ 0.1). B, virulence of the gua1⌬ mutant in the murine inhalation model of infection. 6-week-old female BALB/c mice were infected with H99, gua1⌬, or gua1⌬ϩGUA1 C. neoformans strains (n ϭ 10), and survival was monitored over 60 days. Survival of mice with wild-type or gua1⌬ϩGUA1 strains showed no significant difference (p Ͼ 0.05), and survival was significantly lower than in those infected with the avirulent gua1⌬ mutant strain (p Ͻ 0.0001). C and D, posthumous organ burden was calculated in colony-forming units (CFU)/g of organ as described.

Gua1 Is Required for C. neoformans Infection
cinamidine synthetase and CTP synthetase (65)(66)(67). In the context of antimycotic development, however, a nonspecific inhibitor is of little interest because a number of off target effects in the host would be anticipated, and so it was deemed unsuitable for this study. Ones potentially important lead compound was ECC1385, a synthetic compound that showed potent activity against a number of other fungal pathogens (26). A genetics-based investigation revealed that ECC1385 had potency against the GMP synthases from C. albicans and A. fumigatus and that the compound had a different mode of action to a glutamine agonist (26).
We tested ECC1385 as a direct inhibitor of purified C. neoformans GMP synthase using concentrations between 0 and 200 M, which revealed an IC 50 of 4.4 M, providing the first biochemical evidence that ECC1385 acts directly on the activity of purified GMP synthase. However, subsequent broth dilution assays to determine the inhibitory action of ECC1385 against C. neoformans failed to inhibit growth, despite employing a concentration range reaching 100 M, far exceeding the potency determined in the enzyme assay. Importantly, although this compound cannot serve as an antifungal itself, it provides evidence that the activity of GMP synthase can be inhibited without depending on a nonspecific agent.
Predicted Differences between Human and C. neoformans GMP Synthase-The process of rational drug design can be greatly facilitated by the knowledge of the three-dimensional structure of the target. To this end, we endeavored to solve the crystal structure of C. neoformans GMP synthase; however, despite extensive efforts, we were unable to obtain a protein structure because of poor crystal quality.
To gain insight into GMP synthase structure in the absence of a crystal structure, we used size exclusion chromatography (SEC) coupled with multiangle laser light scattering (MALLS) to determine the oligomeric state of GMP synthase in solution. Human GMP synthase is a monomer in solution with a small dimer population; however, in the active form, when all substrates are bound, there is evidence that it forms a dimer (43). In contrast, E. coli GMP synthase is a homodimer is solution (39,68). The elution profile for C. neoformans GMP synthase demonstrates that this protein predominantly forms a dimer (120 kDa) in solution, with a small tetramer population (Fig. 4). The quaternary structure of C. neoformans GMP synthase is therefore more similar to E. coli than the human protein.
To compare potential structural differences between the human and C. neoformans enzymes, a homology model (63) of C. neoformans GMP synthase was generated using the E. coli enzyme as a template; there are no reported fungal GTP synthase structures, and of those species for which a structure has been determined, the E. coli enzyme exhibited the highest protein identity (50%) with C. neoformans (34).
Overall, the structural model of C. neoformans, based on the E. coli structure, has three distinct regions: dimerization, ATP-PPase, and GATase domains (Fig. 5). The human GMP synthase has been reported to have two distinct groups of its dimerization domain (D1 and D2) as well as the ATP-PPAse and GATase domains, and in this respect C. neoformans GMP synthase is more closely related to E. coli (Fig. 5). Like E. coli, C. neoformans GMP synthase is a dimer in solution (Fig. 4), whereas human GMP synthase is a monomer in solution and believed to dimerize upon substrate binding.
Although lacking two subsections, the dimerization domain of E. coli and C. neoformans includes most amino acids required for XMP binding, and the highly conserved dimerization subdomain with the consensus sequence Val-Gly-Val-Xaa-Gly-Asp-Xaa-Arg-Xaa-Tyr (supplemental Fig. S3). This subdomain is present in the D1 domain of human GMP synthase rather than D2, which has a high level of identity with C. neoformans and E. coli GMP synthase.
In the human GATase domain, three key residues are involved in the binding of glutamine: Cys 104 , His 190 , and    FEBRUARY 17, 2017 • VOLUME 292 • NUMBER 7

Gua1 Is Required for C. neoformans Infection
Glu 192 . These residues form a catalytic triad that is highly conserved among the amidotransferases, and it is also conserved in C. neoformans GMP synthase (supplemental Fig. S3), as well as E. coli (data not shown) (43). The XMP binding site would be a more viable drug target of GMP synthase given that glutamate analogs, such as DON, target other enzymes as well. Of the residues that interact with XMP in human GMP synthase, one residue, Gln 610 , was particularly interesting because it is part of the D1 domain that is absent from C. neoformans GMP synthase (43). Two other residues were not identified in the model but can be seen in the sequence alignment as being conserved. Given the large conformation change believed to occur during XMP amination and the homology model based on AMP-bound E. coli GMP synthase, the model may not represent the XMP-bound state. The D2 domain contains other amino acids, which contribute to XMP binding, and for the most part, these amino acids are conserved; however, at position Glu 694 in the humans enzyme, C. neoformans GMP synthase has a leucine.

Discussion
C. neoformans GMP synthase was identified based on similarity to the S. cerevisiae ortholog GUA1. The mutant phenotype following deletion by biolistic transformation showed that the enzyme is essential for fungal growth on minimal medium and required for multiple phenotypes associated with viru-lence. These virulence factors are key to C. neoformans survival in the human host; defects in production of melanin, capsule, and proteases have been demonstrated to cause attenuated virulence. Furthermore, both capsule and melanin production are regulated via a heterotrimeric G protein that requires GTP to function and, by extension, are dependent on the synthesis of GTP themselves (69).
Given the auxotrophy and lag observed in the production of virulence factors, we investigated the importance of GUA1 during infection. Both the nematode and murine models of infection showed that GUA1 is essential for virulence, mirroring the reports for C. albicans and A. fumigatus (26,44). Furthermore, in the murine inhalation model, the gua1⌬ strain was completely cleared by the host immune system.
Together, the in vitro phenotypic assays and the in vivo virulence models show that GMP synthase could be an useful drug target. Beyond the obvious requirements for GTP in DNA synthesis, transcription, and energy metabolism, many cellular processes are dependent on regulation via GTP-binding proteins. Furthermore, the production of the characteristic capsule of C. neoformans, the pathogen's most well known virulence factor, requires GDP-mannose for its biosynthesis (70). Inhibition of GMP synthase would prevent these key biological processes and thus be even more detrimental to C. neoformans survival in the hostile host environment. The biochemistry and the structure of GMP synthase is key information that could enable the exploitation of this enzyme as an antifungal drug target, helping address the urgent need for novel chemotherapeutic agents to combat disseminated mycoses. GMP synthase consists of two active sites: one for the binding of glutamine and the other for XMP and ATP. The homology model and sequence alignment of C. neoformans GMP synthase to the human enzyme shows that the catalytic triad required for glutamine binding is conserved, although kinetic data show C. neoformans GMP synthase protein to have a lower affinity for this substrate, maybe because of other residues hindering binding. The best known inhibitor of GMP synthase in other species is DON, a well characterized glutamine agonist that targets several glutamine binding proteins. However, given the lower affinity of the C. neoformans GMP synthase protein for glutamine as a substrate, it is likely an unsuitable target for DON-based inhibition and additionally be a nonspecific target (65)(66)(67). In contrast, XMP amination requires interactions with eight residues in human GMP synthase, of which only five are homologous to the C. neoformans enzyme. The fact that the human GMP synthase residue Gln 610 is not present in its fungal homolog, along with residues around the XMP binding site, may make targeting this site in antifungal development worthwhile.
The structural differences between GMP synthase in humans and C. neoformans identified via homology modeling are consistent with this difference potentially arising from marked changes in the dimerization domain that contributes key residues to the human XMP binding site as seen in the solved E. coli structure. These differences in the binding site of XMP may enable the development of selective compounds that do not interact with the human enzyme.
Limited insights into potential starting points for GMP synthase inhibitor development were provided in our study of ECC1385, which demonstrated potent activity against the purified C. neoformans GMP synthase but was unable to show activity in a whole cell assay. Rodriguez-Suarez et al. (26) demonstrated whole cell activity against C. albicans in vitro but not in vivo, instead finding that increased concentrations lead to toxicity in the murine model. ECC1385 has specificity toward GMP synthase; however, it is not adequately active toward fungi. This could simply be due to an inability to enter the cell.
The mechanism of action of GMP synthase requires all bound substrates to undergo an intermediate reaction before coming together to form GMP (34). How this secondary reaction (whereby the adenyl-XMP and GATase domain product are brought together) occurs is largely unknown, because the catalytic domains are ϳ30 Å apart in the available crystal structures. To enable GMP synthesis to occur, the enzyme must either channel the intermediates or undergo a large conformational change whereby both domains are brought together (34). It has been suggested that this large conformational change could be facilitated by the presence of XMP, ATP and Mg 2ϩ . When the substrates are bound, the structure likely transitions into a compact state, creating an ammonia channel that enables the product of the GATase domain to react with adenyl-XMP (34,36). The ability to undergo such a large conformational change may have contributed to our inability to acquire high quality diffraction data from our crystallization efforts.
Overall, we have demonstrated that GMP synthase is crucial for virulence factor production and pathogenesis of C. neoformans in both murine and nematode models. Differences between human and C. neoformans in enzymatic activity and amino acids of the XMP binding site make GMP synthase an attractive antifungal drug target. Historically, the de novo purine biosynthesis pathway has been a target for many drugs, in particular against cancer, because they are able to selectively target the demands of rapidly proliferating cells by starving them of essential nucleotides. Similarly, disseminating microbes rapidly proliferate during infection, making this an attractive target for antibiotics. Given the lack of current drugs on the market against fungi, novel targets like GMP synthase are urgently needed.

Experimental Procedures
Bioinformatic Analyses-The C. neoformans type strain H99 genome sequence was reported by Janbon et al. (47). The GMP synthase-encoding gene was identified in the C. neoformans genome via reciprocal best hit BLAST analysis querying with the S. cerevisiae Gua1 protein.
Molecular Techniques-The sequences of oligonucleotides used are listed in supplemental Table S1. The deletion construct for the GUA1 gene was generated using overlap PCR, employing primers UQ1736 and UQ1739 to join the GUA1 5Ј region (primers UQ1736 and UQ1737), the G418 resistance marker NEO (primers UQ234 and UQ235), and the GUA1 3Ј region (primers UQ1738 and UQ1739). H99 genomic DNA was used as the template for GUA1, and the plasmid pJAF1 was used for NEO (49). The deletion construct was transformed into type strain H99 via biolistic transformation with a Bio-Rad He-1000 Biolistic device (Bio-Rad) with selection on medium containing 100 g/ml G418 and 1 mM guanine. For complementation, the GUA1 gene was PCR-amplified (primers UQ1736 and UQ1739) from H99 genomic DNA, digested with SpeI and XhoI and cloned into the nourseothricin resistance vector pCH233 cut with SpeI/XhoI to generate pAK05. The pAK05 GUA1 NAT fragment was subsequently purified as a 6167-bp SpeI/FspI fragment and transformed into the gua1⌬ mutant, FEBRUARY 17, 2017 • VOLUME 292 • NUMBER 7 selecting for nourseothricin resistance (100 g/ml). For strain validation, genomic DNA was prepared using the CTAB protocol of Pitkin et al. (50), digested, electrophoresed on TAE-agarose gels, and Southern-blotted onto Hybond-XL membrane (GE Healthcare, UK) using standard procedures (51). Probes (primers UQ1736 and UQ1739) were generated using the Rediprime II kit and [␣-32 P]dCTP (PerkinElmer Life Sciences). The blots were hybridized at 65°C, and the membranes were exposed onto Fuji Super RX medical X-ray film (Fujifilm).

Gua1 Is Required for C. neoformans Infection
Phenotypic Assays-Production of melanin was assayed on L-DOPA medium (52) supplemented with 1 mM guanine. Urease assays were performed on Christensen's agar (53), and protease assays were performed on YNB with amino acids and ammonium sulfate supplemented with 2% glucose, 0.1% BSA, and 1 mM guanine. Images were collected after 24 -92 h at 30 or 37°C. All growth tests were performed in triplicate.
For capsule assays, strains were incubated in RPMI 1640 medium (Life Technologies) supplemented with 2% glucose, 10% fetal bovine serum (Life Technologies), and 1 mM guanine with shaking at 30 or 37°C. At 30 h, cells were collected and stained with India ink (Becton Dickinson) and imaged with a Leica DM2500 microscope and DFC425C camera. At least 10 independent images were taken, and the relative capsule diameter of 50 cells from each culture was determined as described by Zaragoza et al. (54). Experiments were performed in biological triplicate, and analysis of variance tests were performed in GraphPad Prism version 7.0 (GraphPad Software) to compare variation between replicates.
Growth curves were conducted in RPMI 1640 medium supplemented with 2% glucose, 10% fetal bovine serum, and 1 mM guanine at 30°C. Starter cultures were grown overnight, diluted to A 600 0.05, and then monitored spectrophotometrically at A 600 every 12 h for 6 days. Growth curves were performed in triplicate.
Nematode Virulence Assays-H99, gua1⌬, and gua1⌬ϩGUA1 strains were grown at 30°C overnight on BHI, minimal medium, and minimal medium supplemented with 1 mM guanine prior to the introduction of 30 -50 synchronized young adult C. elegans worms and incubation at 28°C. Worms were counted every 24 h for 8 days; individuals that did not respond to touch with a platinum wire pick were considered dead and removed. Experimental conditions were tested in triplicate, with Kaplan-Meier survival curves and Mantel-Cox tests performed in GraphPad Prism version 7.0 (GraphPad Software) to determine significance.
Murine Inhalation Model of Cryptococcosis-For murine infection assays, 6-week-old female BALB/c mice (Animal Resources Centre, Murdoch, Australia) were infected by nasal inhalation (55). For each strain, 10 mice were inoculated with a 50-l drop containing 5 ϫ 10 5 C. neoformans cells. A maximum of five mice were housed per individually ventilated cage (Tecniplast) with Bed-o'Cobs 1/8-inch bedding (Andersons), Crink-l'Nest nesting material (Andersons), and cardboard as environmental enrichment. The mice were provided rat and mouse cubes (Specialty Feeds, Glen Forrest, Australia) and water ad libitum. Each mouse was examined and weighed twice daily for the duration of the experiment, with affected mice euthanized via CO 2 inhalation once body weight had decreased to 80% of preinfection weight or they exhibited symptoms consistent with infection. Death after CO 2 inhalation was confirmed by pedal reflex prior to dissection. Brain, lungs, liver, spleen, and kidneys were collected, homogenized in 1 ml of PBS using a TissueLyser II (Qiagen), serially diluted, and plated on YPD supplemented with 100 g/ml ampicillin, 50 g/ml kanamycin, and 25 g/ml chloramphenicol (H99 and gua1⌬ϩ GUA1 infected mice) or YNB supplemented with 1 mM guanine and antibiotics as before (gua1⌬ infected mice). The plates were incubated at 30°C, and after 2 days colonies were counted and used to calculate colony-forming units/g of organ. Kaplan-Meier survival curves were plotted using GraphPad Prism version 7.0 (GraphPad Software). Significance was analyzed using the log-rank test, whereas organ burden significance was determined using a one-way analysis of variance with Tukey's multiple comparisons test. p values of Ͻ 0.05 were considered significant.
Ethics Statement-This study was carried out in strict accordance with the recommendations in the Australian Code of Practice for the Care and Use of Animals for Scientific Purposes by the National Health and Medical Research Council. The protocol was approved by the Molecular Biosciences Animal Ethics Committee of the University of Queensland (approval number SCMB/008/11/UQ/NHMRC). Infection was performed under methoxyflurane anesthesia, and all efforts were made to minimize suffering through adherence to the Guidelines to Promote the Wellbeing of Animals Used for Scientific Purposes as put forward by the National Health and Medical Research Council.
Broth Microdilution Assays-Minimum inhibitory concentration susceptibility of C. neoformans to ECC1385 was determined in accordance with the CLSI M27-A2 guidelines, with the following modifications: YNB medium supplemented with ammonium sulfate and 2% glucose, final inoculum concentration of 1.5-2.0 ϫ 10 3 cells/ml and incubation at 35°C for 72 h (25,56). Drug concentrations ranged from 100 M to 49 nM; the MIC was defined as the concentration that prevented any discernible growth after 72 h. Fluconazole was employed as a control.
Expression and Purification of C. neoformans GMP Synthase-Total RNA was isolated from strain H99 using TRIzol reagent (Invitrogen), with intron-free cDNA and then synthesized using a Bioline cDNA synthesis kit (Bioline). The GUA1 ORF was subsequently PCR-amplified (primers UQ2081 and UQ2082), and the product was inserted via ligation-independent cloning into the SspI site of His tag vector pMCSG7 (57) to yield pJLC1 and then co-transformed with pLysS into E. coli strain BL21(DE3) (Merck). Transformed cells were grown at 37°C in LB supplemented with 100 mg/ml ampicillin and 12.5 mg/ml chloramphenicol to an A 600 of ϳ1 then induced with 1 mM isopropyl ␤-D-1-thiogalactopyranoside and grown for a further 17 h at 22°C. The cells were harvested and resuspended in lysis buffer (50 mM HEPES, pH 8.0, 300 mM NaCl, 30 mM imidazole, 1 mM DTT, and 1 mM PMSF) before disruption with a Sonifier W-450 Digital Ultrasonic Cell Disruptor sonicator (Branson). Following centrifugation, supernatant was loaded onto a 5-ml HisTrap Fast Flow column (GE Healthcare) to purify the histidine tagged protein by immobilized nickel affin-Gua1 Is Required for C. neoformans Infection ity chromatography. The protein was eluted in a linear gradient of 30 -500 mM imidazole, with a single elution peak. Peak fractions were pooled, concentrated, and incubated overnight with 500 g of tobacco etch virus protease at 4°C to cleave the N-terminal histidine tag. Following a second round of nickel affinity chromatography, GMP synthase was collected from the flowthrough, concentrated, and further purified using a HiLoad 26/600 Superdex 200 SEC column (GE Healthcare). Protein was eluted at a rate of 2.5 ml/min with SEC buffer (10 mM HEPES, pH 7.5, 150 mM NaCl, and 1 mM DTT) using an Ä KTApurifier FPLC system (GE Healthcare). Peak fractions were combined and concentrated to ϳ32 mg/ml and flash frozen in liquid nitrogen for storage at Ϫ80°C.
Steady-state and Inhibitor Kinetics-GMP synthase activity was monitored spectrophotometrically using a Cary60 UV-visible spectrophotometer (Agilent). Assays were carried out as described for M. tuberculosis by Franco et al. (58) with the exception of 50 mM HEPES, pH 7.5, being used instead of 50 mM Tris, pH 7.5; optimized reaction conditions were determined as 0.15 mM XMP, 1 mM ATP, 5 mM glutamine, and 20 mM MgCl 2 . The temperature used was 40°C, consistent with previous assays conducted for the human, E. coli, and M. tuberculosis enzymes; this temperature also gave the highest activity in C. neoformans GMP synthase (35,58,59). Purified C. neoformans GMP synthase was used at 0.025 mg/ml final concentration in the assay. Assays were performed in triplicate by measuring the decrease of absorbance at 290 nm to follow conversion of XMP (⑀ 290 ϭ 4.080 mM Ϫ1 cm Ϫ1 ) into GMP (⑀ 290 ϭ 3.066 mM Ϫ1 cm Ϫ1 ). ⌬⑀ ϭ 1.014 mM Ϫ1 cm Ϫ1 was used to calculate the amount of GMP formed. The data were fitted to the Hill or Michaelis-Menten equations using Graph-Pad Prism version 7.0 (GraphPad Software) as appropriate ( Figure S1). The Hill equation was used to calculate the Hill coefficient (n) (Equation 1) (60). IC 50 values were determined using standard assay conditions with varying concentrations of inhibitor ECC1385 (0 -100 M) ( Figure S2). Multiangle Laser Light Scattering-SEC coupled with MALLS was performed using a Dawn Heleos II 18 angle lightscattering detector coupled with an Optilab rEX refractive index detector (Wyatt Technology). 500 g of GMP synthase was applied to the HiLoad 26/600 Superdex 200 SEC column at a flow rate of 0.5 ml/min in 10 mM HEPES, pH 7.5, 150 mM NaCl, and 1 mM DTT. Molecular mass calculations were performed using Astra 5.3 software (Wyatt Technology). Input of the refractive increment (dn/dc values) was set at 0.186 in molecular mass calculations (61).
Crystallization Screens and Modeling of GMP Synthase-Crystallization experiments of recombinant C. neoformans GMP synthase were performed by hanging drop vapor diffusion at 20°C with the commercial sparse matrix screens JCSG, Morpheus, PACT, ProPlex (Molecular Dimensions), INDEX, PEG/ Ion and PEGRx (Hampton Research). Two initial lead conditions (PACT F8, 20% PEG 3350, 0.2 M sodium sulfate decahydrate, and 0.1 M Bis-Tris propane, pH 6.5, and PACT G8, 20% PEG 3350, 0.2 M sodium sulfate decahydrate, and 0.1 M Bis-Tris propane, pH 7.5) were chosen for factorial grid screen optimization. Despite significant efforts-co-crystallization with substrates (XMP, AMP-PNP, and MgCl 2 ), streak and microseeding, enhanced nucleation (62), Silver Bullets (Hampton Research), and Additive Screen (Molecular Dimension)the best resolution diffraction results that could be achieved using the MX2 beamline at the Australian Synchrotron was 6 Å. Subsequently we utilized homology modeling of C. neoformans GMP synthase using the automated webserver Modeler (63) with the E. coli GMP synthase (50% protein identity) structure as a template (PDB code 1GPM) (34).