Structural features of Cryptococcus neoformans bifunctional GAR/AIR synthetase may present novel antifungal drug targets

Abstract

Cryptococcus neoformans is a fungus that causes life-threatening systemic mycoses. During infection of the human host, this pathogen experiences a major change in the availability of purines; the fungus can scavenge the abundant purines in its environmental niche of pigeon excrement, but must employ de novo biosynthesis in the purine-poor human CNS. Eleven sequential enzymatic steps are required to form the first purine base, IMP, an intermediate in the formation of ATP and GTP. Over the course of evolution, several gene fusion events led to the formation of multifunctional purine biosynthetic enzymes in most organisms, particularly the higher eukaryotes. In C. neoformans, phosphoribosyl-glycinamide synthetase (GARs) and phosphoribosyl-aminoimidazole synthetase (AIRs) are fused into a bifunctional enzyme, while the human ortholog is a trifunctional enzyme that also includes GAR transformylase. Here we functionally, biochemically, and structurally characterized C. neoformans GARs and AIRs to identify drug targetable features. GARs/AIRs are essential for de novo purine production and virulence in a murine inhalation infection model. Characterization of GARs enzymatic functional parameters showed that C. neoformans GARs/AIRs have lower affinity for substrates glycine and PRA compared with the trifunctional metazoan enzyme. The crystal structure of C. neoformans GARs revealed differences in the glycine- and ATP-binding sites compared with the Homo sapiens enzyme, while the crystal structure of AIRs shows high structural similarity compared with its H. sapiens ortholog as a monomer but differences as a dimer. The alterations in functional and structural characteristics between fungal and human enzymes could potentially be exploited for antifungal development.

Purines are key biological molecules involved in a plethora of roles including energy metabolism, cell signalling, and encoding the genome for all organisms. To survive, an organism must either salvage purines from the environment via a family of dedicated phosphoribosyltransferases or synthesize purines through the highly conserved enzymes of the de novo biosynthesis pathway. Historically, the purine metabolic pathway has revealed bountiful targets for drug development. Several important drugs that act by inhibiting specific enzymes in the pathway are routinely used to this day; these include acyclovir for treating viral infections, allopurinol for treating gout, and azathioprine for preventing organ transplant rejection (1, 2, 3). Among the drugs that have been developed by targeting purine biosynthesis, antimicrobials that act specifically against bacterial or fungal pathogens have yet to be discovered.

There is a limited number of effective antifungals for treatment of disseminated fungal infections caused by species such as the basidiomycete yeast Cryptococcus neoformans (4). Infections caused by this opportunistic fungus present mainly as meningoencephalitis, a condition commonly encountered particularly in sub-Saharan Africa due to the high prevalence of AIDS in this region (5). The gold standard treatment for meningoencephalitis is amphotericin B, flucytosine, and fluconazole. These treatments have not changed significantly for decades despite high toxicity, excessive cost, limited effectiveness, as well as their association with drug resistance and cross-reactivity (6, 7, 8, 9, 10).

The enzymes of the de novo purine biosynthesis pathway are a potential drug target for antifungals against C. neoformans as the environmental niche of the fungus, purine-rich bird guano, is a stark contrast to the purine-poor human central nervous system it infects (11, 12). Consistent with this, previous research has revealed that successful infection in a murine model requires multiple enzymes that catalyze the conversion of inosine monophosphate (IMP) to adenosine triphosphate (ATP) and guanosine triphosphate (GTP), the final steps of de novo purine biosynthesis (13, 14, 15).

De novo biosynthesis of the purine base is comprised of two major components: the five-step synthesis of an imidazole ring, followed by a six-step process to form the second ring, pyrimidine, to create the first heterocyclic purine base, IMP. De novo biosynthesis of the first part of this structure begins with phosphoribosyl-pyrophosphate (PRPP) amidotransferase (PRPPa, EC 2.4.2.14), which hydrolyzes l-glutamine and transfers the liberated amine group to PRPP, creating phosphoribosyl-amine (PRA; Fig. 1A) (16). The second enzyme, phosphoribosyl-glycinamide (GAR) synthetase (GARs, EC 6.3.4.13), catalyzes the ATP-dependent ligation of l-glycine to PRA, via a phosphorylated intermediate, to yield GAR (17). GAR transformylase (N¹⁰-fTHF-GARt, EC 2.1.2.2) performs the third step that ligates the formyl group from 10-formyltetrahydrofolate (N¹⁰-fTHF) to GAR, producing phosphoribosyl-formylglycinamide (FGAR) (18). The fourth enzyme, phosphoribosyl-formylglycinamidine (FGAM) synthetase (FGAMs, EC 6.3.5.3), activates the FGAR amide oxygen in the ATPase domain to produce an iminophosphate intermediate, which is amidated by ammonia, channelled via a structural domain from the glutaminase domain to create FGAM (19). Finally, phosphoribosyl-aminoimidazole (AIR) synthetase (AIRs, EC 6.3.3.1) catalyzes the ATP-dependent activation of the FGAM formyl oxygen, which reacts with a nearby nitrogen to close the imidazole ring of the unfinished purine base and forms AIR (20).

Figure 1.

Cryptococcus neoformans ade5,7Δ mutants are adenine auxotrophs.A, de novo purine biosynthesis pathway. Bifunctional GARs/AIRs are encoded by ADE5,7 in C. neoformans. Purine biosynthesis enzymes are represented as coloured hexagons. Enzyme fusions are represented by hexagons joined together. B, growth phenotypes of wild-type, ade5,7Δ, and ade5,7Δ+ADE5,7 strains. Strains were serially tenfold diluted and 5 μl of each dilution spotted on YPD, YNB, and YNB supplemented with 1 mM adenine, then grown at 30 and 37 °C. Pictures were taken after 48 h of growth. The ade5,7Δ strain shows no growth on YNB agar without supplementation with adenine.

While these first five enzymes in de novo purine biosynthesis mostly exist as monofunctional proteins in the Archaea and Bacteria, in many Eukaryotes the genes encoding these enzymes have fused to encode multifunctional proteins consisting of up to three components of the pathway in a single open reading frame (21). One of these multifunctional proteins consists of GARs, responsible for the second step of the pathway, fused at its C terminus to AIRs, which performs the fifth step. The bifunctional enzyme is found in the Amoebozoa and Fungi. In the Metazoa, this fusion protein has expanded to include GARt, mediating the third step of de novo purine biosynthesis, forming a trifunctional enzyme (TrifGART). Crystal structures have been determined for unfused bacterial GARs and AIRs from Escherichia coli, Geobacillus kaustophilus, and Thermus thermophiles (17, 22, 23, 24, 25). Homo sapiens is the only eukaryote with crystal structures determined for single unit GARs, AIRs, and GARt. The missing piece to this picture is the structure of the fungal bifunctional GARs/AIRs, which represents an intermediate state between the single unit enzymes of the Bacteria and triple unit fusion of the Metazoa. This major difference may potentially be exploited in the development of new, fungal enzyme-specific purine biosynthesis inhibitors to treat disseminated life-threatening mycoses.

Here we present the characterization of the bifunctional GARs/AIRs from C. neoformans. We have shown that this enzyme is critical for virulence in a murine infection model and have identified differences in enzyme substrate specificity compared with metazoan species. We have also determined C. neoformans X-ray crystal structures of the GARs and AIRs units that revealed potentially exploitable differences in the GARs glycine- and ATP-binding sites when compared with the human ortholog.

Results

Identification of the gene encoding GARs and AIRs in C. neoformans

To identify GARs and AIRs in C. neoformans, a reciprocal best hit BLAST analysis was performed to identify the homolog of S. cerevisiae Ade5,7 in the H99O-type strain genome. Consistent with our previous observations that these enzymes appeared to be fused in all fungi, querying the C. neoformans genome with the S. cerevisiae protein identified a single locus, CNAG_06,314. The predicted protein sequence translated from CNAG_06314 was used in a BLASTp search against the S. cerevisiae genome and returned ADE5,7 as the single hit. As with other C. neoformans genes, we named CNAG_06314 ADE5,7, based on the S. cerevisiae nomenclature.

A protein alignment of S. cerevisiae and C. neoformans Ade5,7 revealed 52.1% identity, indicating a moderate level of conservation. Investigating the conservation of the individual GARs and AIRs functional units of the fusion protein showed conservation was not restricted to a single domain, with 50.1% and 52.6% identity, respectively. By comparison, the C. neoformans bifunctional Ade5,7 protein alignment with the equivalent enzymatic units of H. sapiens TrifGART yields an identity of 42.8% for GARs and 45.4% for AIRs.

C. neoformans ade5,7Δ mutants are adenine auxotrophs

To determine whether the ADE5,7 gene is essential for de novo purine production, an ADE5,7 gene deletion construct was created and introduced via biolistic transformation-mediated homologous recombination into the C. neoformans H99O-type strain genome. Transformants were analyzed by Southern blot to identify and validate deletion mutants. The ade5,7Δ strain could grow when exogenous adenine was supplemented in the yeast nitrogen base (YNB) media, but not without; it is an adenine auxotroph as expected (Fig. 1B). On nutrient-rich YPD media, the ade5,7Δ mutant could also grow, but the colonies were smaller compared with the wild-type strain at both 30 °C and human body temperature. This poor growth could be resolved by the addition of exogenous adenine. Reintroducing the wild-type ADE5,7 gene by inserting it into the Safe Haven locus on chromosome 1 (26) to create the ade5,7Δ+ADE5,7 strain produced wild-type characteristics on all media, verifying that the observed phenotypes were due to the loss of ADE5,7.

Deletion of ADE5,7 in C. neoformans does not affect production of major virulence traits

After confirming the ade5,7Δ adenine auxotrophic phenotype, the deletion and complemented strains were tested for the production of the major virulence traits of C. neoformans. Virulence trait production was measured by spotting the strains on BSA agar to test for protease, l-3,4-dihydroxyphenylalanine (l-DOPA) agar for melanin, egg yolk agar for phospholipase B, and Christensen’s urea agar for urease. Another virulence trait, capsule formation, was visualized under the microscope using India ink following growth in RPMI 1640 medium.

There was no detectable difference in the production of protease, melanin, phospholipase B, urease, and capsule by the ade5,7Δ mutant compared with the wild-type or complemented strains (Fig. S1). The results indicate that unlike a number of other purine enzyme deletion mutants in C. neoformans that have been studied previously, the loss of the bifunctional GARs/AIRs enzyme did not have a visible effect on virulence trait production (13, 14, 15).

ADE5,7 is essential for C. neoformans virulence in a murine inhalation infection model

The GARs/AIRs deletion strain was employed in a murine inhalation infection model to determine whether the loss of the gene encoding this bifunctional enzyme has an impact on the virulence of C. neoformans in vivo. As the availability of purines in the mouse CNS is comparable to the purine-poor human CNS environment (27), it was predicted that the deletion of ADE5,7 would cause a loss of virulence.

Groups of ten mice were infected with either the wild-type, the deletion mutant, or the complemented strain. Mice were monitored daily after infection and were euthanized based on either the symptoms of meningoencephalitis or a 20% loss of body weight; the spleen, kidney, liver, lungs, and brains from all euthanized mice were collected, homogenized, and plated on YNB media supplemented with adenine to quantitate fungal burden.

All mice infected with the wild-type and complemented ade5,7Δ+ADE5,7 strains succumbed to illness within 3 weeks. In stark contrast, all mice infected with the ade5,7Δ mutant survived and appeared healthy at the end of the experiment; the mutant is avirulent (Fig. 2A). Consistent with this result, while the fungal burden load for all organs was equivalent in mice infected with the wild-type and ade5,7Δ+ADE5,7 strains (p-value > 0.05), the mice infected with the ade5,7Δ mutant had no fungal burden detected in their homogenized organs (Fig. 2B). These results demonstrate that ADE5,7 is essential for C. neoformans during the infection process, highlighting its potential as an antifungal target.

Figure 2.

ADE5,7is essential forCryptococcus neoformansvirulence in a murine inhalation infection model.A, deletion of the gene encoding bifunctional GARs/AIRs abolishes the virulence of C. neoformans. Six-week-old mice were infected intranasally with 1 × 10⁵ cells of wild-type, ade5,7Δ mutant, or ade5,7Δ+ADE5,7 complemented strains. Kaplan–Meier survival curves are plotted and significance determined by log-rank tests. All mice infected with the ade5,7Δ mutant survived, and there was no significant difference between the mice survival of wild-type and ade5,7Δ+ADE5,7 complemented strains (p-value > 0.05). B, fungal organ burden of mice infected with wild-type, ade5,7Δ mutant, and ade5,7Δ+ADE5,7 complemented strains. Organs collected from the mice after euthanasia were homogenized and spotted on YNB plates supplemented with adenine. The fungal burden for mice infected with the wild-type and ade5,7Δ+ADE5,7 was significantly higher (∗∗∗∗p-value < 0.05) than the mice infected with the ade5,7Δ mutant, which did not have any fungus present.

Oligomeric states of C. neoformans bifunctional GARs/AIRs and its single enzymatic units

Because GARs/AIRs are essential for C. neoformans virulence, the bifunctional GARs/AIRs and its functional units (GARs and AIRs) were structurally and biochemically characterized to identify differences between the fungal and human proteins that may be employable in the future for antifungal design. Recombinant His-tagged C. neoformans bifunctional GARs/AIRs, and the GARs and AIRs units alone, were expressed in E. coli and purified using immobilized metal affinity chromatography (IMAC) and size-exclusion chromatography (SEC). The predicted sizes of purified bifunctional GARs/AIRs (80 kDa), GARs (50 kDa), and AIRs (40 kDa) were confirmed by SDS-PAGE (Fig. 3, A1–A3, respectively).

The bifunctional enzyme and its single enzymatic units were each characterized using mass photometry to determine the oligomeric state of the proteins in solution. E. coli GARs was previously shown to be a monomer by SEC with standard molecular weights, and E. coli AIRs to be a dimer (17, 23). Mass photometry revealed that the C. neoformans bifunctional GARs/AIRs consisted of ∼65% monomers (∼80 kDa) and ∼25% dimers (∼170 kDa); GARs as a single enzymatic unit were composed of ∼80% monomers (∼50 kDa) and ∼20% dimers (∼100 kDa); AIRs as a single enzymatic unit existed only as dimers (∼80 kDa) (Fig. 3, B1–B3, respectively). By comparison, small-angle X-ray scattering (SAXS) of human TrifGART previously showed the formation of dimers in solution (24). Therefore, the C. neoformans bifunctional GARs/AIRs have similarities to the human enzyme in terms of forming dimers, with both of the component units displaying the ability to dimerize on their own.

Figure 3.

Oligomeric states ofCryptococcus neoformansbifunctional GARs/AIRs and its single enzymatic units.A, size-exclusion chromatography of recombinant His-tagged C. neoformans bifunctional GARs/AIRs and single enzymatic units (GARs and AIRs). Recombinant His-tagged C. neoformans protein was expressed and purified using IMAC and SEC. The predicted sizes of purified bifunctional GARs/AIRs (A1, 80 kDa), single unit GARs (A2, 50 kDa), and single unit AIRs (A3, 40 kDa) were confirmed by SDS-PAGE. B, oligomeric state determination of C. neoformans bifunctional GARs/AIRs and the single enzymatic units GARs and AIRs. Protein oligomerization in solution was determined by mass photometry, revealing monomers and dimers for bifunctional GARs/AIRs (B1), monomers and dimers for single unit GARs (B2), and dimers for single unit AIRs (B3).

C. neoformans GARs differ from characterized metazoan GARs in its substrate specificity and turnover

We performed a biochemical characterization of bifunctional GARs/AIRs and single unit GARs to enable comparison to metazoan TrifGART (GARs/AIRs/GARt) in order to identify functional differences that may be exploitable for antifungal drug design. The recombinant proteins were characterized using steady-state enzyme kinetics of GARs activity. The substrate PRA was chemically synthesized from ribose 5-phosphate and ammonium chloride using the methodology of Schendel and colleagues (28). A coupled enzyme assay with excess molar concentrations of 8 U/ml pyruvate kinase (PK) and 10 U/ml lactate dehydrogenase (LDH) from rabbit muscle was used to measure GARs activity, where the consumption of NADH was detected spectrophotometrically at 340 nm (Fig. 4A).

Figure 4.

Cryptococcus neoformansGARs differ from characterized metazoan GARs in its substrate specificity and turnover.A, GARs activity measured using a coupled enzyme assay with pyruvate kinase (PK) and lactate dehydrogenase (LDH). The continuous assay was recorded spectrophotometrically by measuring the consumption of NADH at 340 nm. The GARs reaction is highlighted with a light blue roundrect. The coupled reaction with PK and LDH is highlighted with a light purple roundrect. Purine biosynthesis enzymes are represented as colored hexagons as in Figure 1A. B, comparisons of the native bifunctional GARs/AIRs and single unit GARs Michaelis–Menten curves for ATP, glycine, and PRA, and enzyme kinetics parameters. Bifunctional GARs/AIRs curve and data points (mean with standard error) are shown in pink and the single unit GARs curve and data points are shown in blue. Specific activity, k_cat, K_m, and k_cat/K_m for the enzymes are recorded in the table.

GARs is responsible for the second enzymatic reaction in purine biosynthesis (Fig. 1A). Studies of the enzymes from E. coli and the thermophilic bacteria Thermus thermophilus, G. kaustophilus, and Aquifex aeolicus have shown that the GARs reaction occurs in three substeps (17, 22). First, the γ-phosphate of ATP is attacked by the oxygen from glycine, resulting in the dephosphorylation of ATP through a nucleophilic substitution. Second, glycine is phosphorylated to form a glycylphosphate intermediate. Third, the glycyl component from the carbonyl carbon of the glycylphosphate is transferred to the amine nitrogen of PRA via a nucleophilic acyl substitution that forms the product GAR. The enzyme specificity for the substrates participating in each of the three sub-steps was measured by varying the amounts of ATP, glycine, and PRA independently, while keeping the other substrates at a saturating concentration equivalent to ten times K_m concentrations for maximal enzyme activity.

ATP dephosphorylation for both bifunctional GARs/AIRs and single unit GARs displayed Michaelis–Menten kinetics, with very similar K_{m app_ATP} values of 48 ± 3 μM and 52 ± 3 μM, respectively (Fig. 4B), and turnover numbers (k_cat) of 27.3 ± 0.4 s⁻¹ for bifunctional GARs/AIRs and 17.3 ± 0.3 s⁻¹ for the single unit GARs. As a result of the approximately twofold lower k_cat for the single unit GARs, k_cat/K_{m_ATP} for bifunctional GARs/AIRs ((5.7 ± 0.4) × 10⁵ M⁻¹ s⁻¹) was almost two times higher compared with single unit GARs, which had a k_cat/K_{m_ATP} of (3.3 ± 0.2) × 10⁵ M⁻¹.

The phosphorylation of glycine also followed Michaelis–Menten kinetics with K_{m app_glycine} values of 496 ± 20 μM for the bifunctional GARs/AIRs and 364 ± 9 μM for single unit GARs and had very similar k_cat/K_{m_glycine} of (5.5 ± 0.2) × 10⁴ M⁻¹ s⁻¹ and (4.8 ± 0.1) × 10⁴ M⁻¹ s⁻¹ (Fig. 4B).

Finally, Michaelis–Menten kinetics for the ligation of the glycyl component to PRA revealed a K_{m app_PRA} of 131 ± 15 μM for bifunctional GARs/AIRs, which was about two times lower than the value determined for the single unit GARs, which had a K_{m app_PRA} of 240 ± 15 μM. This reaction proceeded with increased specificity for the bifunctional GARs/AIRs (k_cat/K_{m_PRA}: (2.1 ± 0.2) × 10⁵ M⁻¹ s⁻¹), which was approximately three times higher than for single unit GARs (k_cat/K_{m_PRA}: (7.2 ± 0.6) × 10⁴ M⁻¹ s⁻¹).

In summary, the bifunctional GARs/AIRs specificity constants for ATP dephosphorylation and for the ligation of the glycyl group to PRA were considerably higher than for the single unit GARs, while the specificity constant for the phosphorylation of glycine was similar for the two proteins. These findings indicate that while the single unit GARs are sufficient to convert PRA to GAR, the dephosphorylation of ATP and ligation of the glycyl group to PRA are more efficient when the AIRs unit is attached.

Enzymatic parameters for GARs have been reported for the enzymes from three bacterial species, A. aeolicus, E. coli, and Klebsiella pneumoniae, as well as for two metazoan species, Gallus gallus and H. sapiens (Table 1). The C. neoformans bifunctional GARs/AIRs are most similar in overall specific activity (20.5 ± 0.27 U/mg) to the E. coli GARs for which a specific activity of 20.5 ± 1.5 U/mg has been reported. There is a key difference in enzyme specific activity between the fungal and metazoan enzymes; C. neoformans bifunctional GARs/AIRs have the highest reported specific activity (20.5 ± 0.27 U/mg) among the eukaryotic enzymes, 90 times greater than H. sapiens TrifGART (0.228 U/mg) and five times greater than G. gallus TrifGART (4.0 U/mg). Related to this is the higher turnover number of the fungal enzyme compared with the metazoan enzymes; C. neoformans bifunctional GARs/AIRs turnover (27.3 ± 0.4 s⁻¹) is around 20 times greater than H. sapiens TrifGART (0.4 s⁻¹) and nearly four times greater than G. gallus TrifGART (7.0 s⁻¹). Interestingly, despite having a higher specific activity and faster turnover compared with the metazoan enzymes, a higher concentration of glycine and PRA are required to reach half maximal velocity for C. neoformans bifunctional GARs/AIRs compared with G. gallus TrifGART. The phosphorylation of glycine for C. neoformans bifunctional GARs/AIRs has a K_{m app_glycine} (496 ± 20 μM) that is around ten times higher than the reported K_{m app_glycine} (42 ± 5 μM) for G. gallus TrifGART. The ligation of the glycyl group to PRA for C. neoformans bifunctional GARs/AIRs has a K_{m app_PRA} (131 ± 15 μM) that is approximately three times higher than G. gallus TrifGART K_{m app_PRA} (41 ± 3 μM).

Table 1.

Comparison of kinetic parameters for GARs activity among enzymes from different species

Species	Enzyme form M.W.	Specific activity (U/mg)	kcat (s−1)	Km (ATP) (μM)	kcat/Km (ATP) (M−1 s−1)	Km (Gly) (μM)	kcat/Km (Gly) (M−1 s−1)	Km (PRA) (μM)	kcat/Km (PRA) (M−1 s−1)	Ref.
Cryptococcus neoformans	BiF GARs/AIRs 80 kDa									This work
Homo sapiens	TriF GART 112 kDa									(54)
Gallus gallus	TriF GART 106 kDa									(55)
Klebsiella aerogenes	GARs 45 kDa									(56)
Escherichia coli	GARs 45 kDa									(46)
Aquifex aeolicus	GARs 47.5 kDa									(57)

Abbreviation: ND, no data provided.

Assay conditions were as follows: C. neoformans bifunctional GARs/AIRs and single unit GARs–100 mM Tris HCl, pH 8.0, 37 °C; Homo sapiens TrifGART (trifunctional GARs/AIRs/GARt)–100 mM Tris HCl, pH 7.5, 37 °C; G. gallus TrifGART–10 mM Hepes, pH 7.5, 25 °C; K. pneumoniae GARs–85 mM Tris HCl, pH 8.0, 37 °C; E. coli GARs–100 mM Tris HCl, pH 8.0, 18 °C for Km and 37 °C for specific activity; Aquifex aeolicus GARs–50 mM Tris HCl, pH 7.9, 90 °C.

Reported values from publications are shown in dark gray. Values calculated from publication-reported enzyme functional parameters are shown in light gray.

Overall, these comparisons indicate that there are functional differences between the fungal and metazoan enzymes with variations in the observed values of specific activity, k_cat and K_{m app}. Whether these functional differences may be exploited for antifungal development will depend on the identification of structural differences that can be exploited for the design of specific inhibitors.

Crystal structure of C. neoformans GARs

To compare the C. neoformans enzyme with the human ortholog, we aimed to determine the crystal structure of the C. neoformans bifunctional GARs/AIRs. As with human TrifGART, multiple attempts to crystallize the C. neoformans bifunctional GARs/AIRs as a single protein were unsuccessful (24). Therefore, the enzymatic units of GARs and AIRs were expressed and crystallized separately.

Purified recombinant C. neoformans GARs protein expressed in E. coli was successfully crystallized without ligands. The crystals diffracted to a resolution of 1.8 Å (Table S5). The crystal structure (PDB ID: 7LVO) contains one monomer consisting of 451 residues (excluding the 25 residue recombinant expression 6-His with a TEV protease cleavage site) per asymmetric unit, with four missing residues (position 196–199) in the model due to the absence of electron density, presumably due to flexibility. This enzyme belongs to the ATP grasp superfamily, a group of proteins that have a unique ATP-binding site known as the ATP grasp fold and that includes enzymes involved in cell metabolism, gluconeogenesis, fatty acid, and de novo purine biosynthesis (29). In this latter group, the ATP grasp fold has been identified in CAIR synthetase, as well as the formate-dependent forms of GAR transformylase and AICAR transformylase found exclusively in the Archaea and Bacteria (30, 31, 32).

Similar to GARs from other species, the C. neoformans structure has four domains: N (residues 34–160, five alpha-helices, four beta-strands); B (residues 161–237, three alpha-helices, three beta-strands); A (residues 238–381, six alpha-helices, five beta-strands); and C (residues 382–467, one alpha-helix, three beta-strands), which combine to form the binding sites for PRA, ATP, and glycine (Fig. 5A). A DALI (33) search of the Protein Data Bank (PDB) revealed that the C. neoformans GARs was most structurally similar to the human GARs, which is the only other available eukaryotic GARs structure (24). By superimposing the C. neoformans and human structure, there is a high level of overall conservation with a root mean square deviation (RMSD) of 1.25 Å. The binding sites of PRA, ATP, and glycine for the C. neoformans and human structure are similar and key residues for binding are well conserved.

Figure 5.

Crystal structure ofCryptococcus neoformansGARs.A, Four domains of C. neoformans GARs. The N domain is shown in pink, A in purple, B in yellow, and C in green. B and C, comparison of the glycine and ATP-binding sites from C. neoformans and H. sapiens GARs. The C. neoformans GARs model is shown in orange, the H. sapiens GARs model (PDB ID: 2QK4) is shown in purple. The side chains of the residues that make up the binding site are shown as a stick representation. The amino acid differences identified in these binding sites are highlighted in boxes with dotted lines. The green ATP molecule is from the H. sapiens GARs structure.

Based on the conservation of key binding residues of the H. sapiens and C. neoformans structures, the key binding residues for the C. neoformans enzyme can be predicted based on the structural superimposition. The PRA-binding site in the C. neoformans enzyme includes Arg48 and Glu49 in the N domain, Asp338 in the A domain, and Arg427 in the C domain (24). The Mg²⁺/ATP-binding site includes Lys262 (B domain). The glycine-binding site includes Asp260 (B domain), Arg335 (A domain), Asp338 (A domain), and Glu340 (A domain). These residues are all 100% conserved in the human ortholog.

Despite the high level of conservation of key binding residues with the human enzyme, analysis of the binding sites of GARs identified two amino acid differences, one in the glycine-binding site and the other in the ATP-binding site, at the position where the adenine ring would fit (Fig. 5, B and C). The first amino acid difference was identified at the glycine-binding site, where Phe317 of C. neoformans GARs corresponds to Tyr274 in human GARs. This residue interacts with the ammonium group of glycine (Fig. 5B). The side chains of both amino acids have a similar orientation, indicating that the position of the amino acid could be crucial for glycine attachment. Phenylalanine and tyrosine both contain an aromatic ring, the only difference being the presence of the hydroxyl group in tyrosine, causing it to be more polar. The difference in polarity might be used to influence specific inhibitor design to the C. neoformans enzyme. The two other major fungal pathogens, Candida albicans and Aspergillus fumigatus, also have the same corresponding phenylalanine residue (Fig. S2), thus exploiting this feature for antifungal drug design might enable broad spectrum efficacy.

The second amino acid difference corresponds to Tyr235 at the ATP-binding site of C. neoformans GARs, where the corresponding residue is Leu192 in human GARs (Fig. 5C). Although the side chains of Tyr235 and Leu192 are oriented in a similar direction, their side chains have different properties. Tyrosine has an aromatic amphipathic side chain, while leucine has an aliphatic nonpolar side group. The tyrosine also extends into the adenine-binding pocket by approximately 5 Å more than leucine; a specifically designed inhibitor could utilize the differences in the properties of the side chain, as well as the size of the binding pocket, to favor the C. neoformans binding site over the H. sapiens enzyme binding site. This amino acid is not conserved across the three major fungal pathogens; A. fumigatus has the same tyrosine, while C. albicans has a phenylalanine (Fig. S2). However, tyrosine and phenylalanine are much more similar than a leucine, thus designing an inhibitor to fit this binding pocket for fungal pathogens might still be possible.

Crystal structure of C. neoformans AIRs

The purified recombinant C. neoformans AIRs expressed in E. coli was successfully crystallized without ligands. The crystals diffracted to a resolution of 2.2 Å and the crystal structure (PDB ID: 7LVP) contained one monomer per asymmetric unit (Table S6). Excluding the His-tag (25 residues), each monomer consists of 307 residues, forming 11 alpha-helices and 9 beta-strands (Fig. 6). Residues 91–98 and residues 122–137 were not modeled (Fig. S3) as there was no electron density observed for these regions; they are presumably disordered. Based on mass photometry data, this protein exists as a dimer, consistent with structures of AIRs in other species (23, 24, 25). The pair of monomers forming the dimer in the crystals is related by a crystallographic twofold axis.

Figure 6.

Comparison of the dimeric structures of C. neoformans and H. sapiensAIRs. The C. neoformans AIRs model is shown in pale cyan (subunit A) and cyan (subunit B). The H. sapiens AIRs model (PDB ID: 2QK4) is shown in light pink (subunit A) and deep purple (subunit B). Binding sites of human AIRs are indicated in the model. Residue modification from Tyr87 to Phe74 is highlighted in the model with a black box and is in stick representation. The superposition of the dimers yields an RMSD value of 2.55 Å.

Structural analysis of the C. neoformans AIRs indicated that it belongs to the PurM superfamily, which includes AIRs from other species as well as the ATPase domain of FGAM synthetase. The dimer has one binding site for ATP and one for the substrate FGAM, each of which lies in the homodimer interface of the four beta-strands from monomer A and four beta-strands of monomer B. Based on the superimposition of ligand-bound T. thermophilus AIRs and C. neoformans AIRs, and the mechanism proposed previously for the T. thermophilus enzyme, the key binding residues for ATP and FGAM are fully conserved. In the C. neoformans AIRs structure, ATP would be predicted to bind to conserved residues Asp77 (monomer A), Asp48 (monomer B), Glu124 (B), Gly123 (B), Asp90 (B), and a Mg²⁺ ion would be bound by Asp77 (A) and His231 (A) at the dimer interface of the C. neoformans AIRs. FGAM is predicted to be bound by the conserved residues Gly49 (both monomers A and B), Val50 (B), Tyr91 (B), Ser174 (A and B), and Asn175 (B), causing a conformational change in the N-terminal region of the enzyme. Based on previous mechanism proposed for T. thermophilus AIRs, the correct positioning of the substrates in the C. neoformans AIRs enables the hydrolysis of ATP to activate the FGAM formyl oxygen, which is coordinated by the conserved Gly46 (A) and Glu124 (B); the nitrogen then triggers a nucleophilic substitution on the activated formyl oxygen and transfers the oxygen to the phosphate formed during ATP hydrolysis. This reaction closes the five-membered imidazole ring of the purine base and forms AIR (20).

Based on a DALI search of the PDB, the top hit is human AIRs as expected as this is the only other eukaryotic structure available for this enzyme. A structural alignment of the monomeric units of C. neoformans and human AIRs shows high conservation, with an RMSD of 0.99 Å. However, when the dimeric unit of the C. neoformans AIRs is structurally superimposed with the human AIRs dimer, the RMSD is increased to 2.55 Å, suggesting a different orientation of the monomers in each dimeric unit (Fig. 6). By aligning one monomer of the dimer unit from C. neoformans and human AIRs, and using the third alpha helix as an anchor, the orientation of second monomers differs by an angle of 22.2° (Fig. 6). Additional investigation of the dimer interface (using PISA (34)) reveals that there are more residues that form hydrogen bonds in the dimer interface for the H. sapiens AIRs compared with the C. neoformans AIRs, which could explain the differences in dimer arrangement. Furthermore, there is a residue modification in the dimer-interacting beta-strands of the C. neoformans AIRs from a Tyr87 to a Phe74 in H. sapiens AIRs that might have caused the change in dimer conformation (Fig. 6). There could be potential for drugs to target the differences in the interface, disrupting dimer formation and by extension creation of the active site.

Discussion

Unlike a number of de novo purine biosynthesis enzymes that have alternative enzymes that can perform equivalent biochemical roles, there are no alternative enzymes that perform the role of GARs or AIRs (21). In the Archaea and the Bacteria, these conserved GARs or AIRs proteins are almost exclusively encoded by individual genes. Rare exceptions do exist, such as the intracellular pathogens of the Francisella genus in the Bacteria where GARs and a different enzyme in the pathway, SAICAR synthetase, are fused together. In fungi, the genes encoding GARs and AIRs are fused, and this fusion is later expanded with the addition of GARt in the Metazoa and some protists. This addition of GARt, an entire enzymatic unit almost 30 kDa in size, is a major structural difference that may potentially be exploited alongside smaller key amino acid changes in the development of new, fungal-specific purine biosynthesis inhibitors, to treat disseminated life-threatening mycoses without interfering with the host purine metabolism.

Deletion analyses of the ADE5,7 gene encoding C. neoformans GARs/AIRs revealed several growth phenotypes. Adenine supplementation was required for growth of the ade5,7Δ mutant but guanine supplementation alone was unable to restore growth. Adenine/adenosine deaminase in C. neoformans enables the conversion of adenine/adenosine into hypoxanthine/inosine, which in turn can be converted into guanine. In contrast, there are no enzymes that can convert guanine into hypoxanthine, thus the inability of guanine to restore growth to the mutant. Unlike the imd1Δ, gua1Δ, ade12Δ, and ade13Δ purine biosynthesis mutants previously studied in C. neoformans that could not grow on unsupplemented YPD, the ade5,7Δ mutant was able to grow under this condition, demonstrating its ability to utilize purines in the YPD media (13, 14, 15). However, mechanisms influencing utilization of purines in YPD are still unknown.

Major virulence trait production of the ade5,7Δ mutant was similar to the wild-type strain, unlike the imd1Δ, gua1Δ, ade12Δ, and ade13Δ purine biosynthesis mutants, which were impaired in melanin, capsule, and protease production (13, 14, 15). However, consistent with the other purine enzyme deletion mutants, mice infected with the ade5,7Δ strain remained healthy at the end of the experiment and the infection was cleared from all major organs. This result supports the model that the reduced virulence of purine biosynthetic mutants is primarily due to the inability to synthesize purines de novo during infection of the purine-poor central nervous system and not due to diminished virulence trait production (11).

A comparative analysis of C. neoformans bifunctional GARs/AIRs and its GARs unit alone revealed differences in enzyme kinetic parameters. The single unit GARs have a 3-fold lower preference for PRA compared with the bifunctional enzyme. This decreased preference for PRA in the absence of the AIRs unit suggests that sections of the protein that promote PRA binding may have been removed. The only other eukaryote in which the K_{m app_PRA} has been determined, G. gallus TrifGART, has similar specificity for PRA compared with the C. neoformans bifunctional GARs/AIRs, supporting the model that the inclusion of the second enzymatic unit may have an effect on PRA binding. Comparison between the C. neoformans, G. gallus, and H. sapiens enzymes revealed differences in substrate specificities, which could suggest that the substrates might interact differently with the binding sites. The lower affinity for substrate binding observed for the C. neoformans enzyme suggests that inhibitors that are substrate analogs may be less suitable, as there is a higher likelihood for the H. sapiens enzyme to bind these substrate analogs more strongly due to its higher affinity. Instead, specifically designed inhibitors could target the unique binding properties of the C. neoformans enzyme, which were revealed through structure determination of the C. neoformans GARs unit. In consideration of differing assay conditions in various organisms, direct comparison of these kinetic parameters may be limited and studies to investigate each specific condition could be instigated.

Although the biochemical characterization of AIRs, the second enzyme unit of bifunctional GARs/AIRs, has the potential to reveal functional differences as well, the unavailability of the substrate FGAM from commercial sources and the complexity of FGAM chemical synthesis required to make it ourselves precluded these functional assays for AIRs being performed. However, in the future, there may be opportunities to investigate this activity. Despite this lack of AIRs enzyme kinetic data, we are still well informed about the key differences in the protein through the determination of the crystal structure. While recombinant GARs/AIRs did not crystallize likely due to the presence of a flexible linker between the two enzymatic units, expression of the two enzymatic units as separate constructs enabled successful crystallization. The same strategy had also been employed in the determination of these structures from the H. sapiens enzyme (24). In addition, crystallization of these proteins with ligands was attempted; however, diffracted crystals did not contain the ligands of interest.

The GARs unit crystals diffracted to a resolution of 1.8 Å and the structure was refined to R_free of 0.28. After data quality analysis, it seems that the higher R_free and R_work values for this model are likely due to a large disordered region between residues 194 and 230 in the B-domain, which is known to be flexible based on previous structures. As the core of the protein could be modeled reliably, accurate comparisons of the C. neoformans GARs model can still be made with the H. sapiens GARs crystal structure. Despite the overall C. neoformans structure being highly similar to H. sapiens, a closer analysis of the binding sites for glycine and ATP identified two important amino acid residue differences: H. sapiens Tyr274 corresponds to C. neoformans Phe317 and H. sapiens Leu192 corresponds to C. neoformans Tyr235. The Leu192 to Tyr235 residue mutation results in a change in the size of the binding pocket and has differing side chain properties that have more potential to be exploited in drug design, by creating an inhibitor that can bind specifically to the fungal-binding site. Functional catalytic data could not be directly correlated with structural information as there is limited kinetic data available for the human ortholog that only reported specific activity for GARs.

The C. neoformans AIRs unit crystals diffracted to a resolution of 2.2 Å and the structure was refined to an R_free of 0.26. The monomer of the C. neoformans enzyme is very similar to its ortholog from H. sapiens, with the ATP and FGAM-binding sites highly conserved; however, there is a distinct difference in the dimeric arrangement. The variation in the dimer arrangement of the C. neoformans AIRs is likely due to part of the interface observed in the human enzyme being disordered (residues 91–98 and residues 122–137 based on truncated AIRS domain construct) (Fig. S3). As a consequence, the monomers are held closer together in the human AIRs (Fig. 6). Another reason for this difference could be due to the amino acid residue modification from a Tyr87 to a Phe74 in the human enzyme, which could push monomers within the dimer further apart in the fungal enzyme (Fig. 6). As the dimer interface is crucial for the enzymatic function, a molecule could potentially be designed to specifically target the C. neoformans AIRs dimer interface.

In short, this study reports a multifaceted characterization of C. neoformans bifunctional GARs/AIRs through gene deletion studies of ADE5,7, the first enzyme kinetic analysis of eukaryotic GARs outside the Metazoa and the first structure determination of both GARs and AIRs from a nonhuman eukaryote. Not only could we utilize the subtle differences identified through functional and structural comparison for future drug design, we could also use the similarities as a tool to possibly gain better insights of the eukaryotic GARs and AIRs biochemical and structural mechanisms, which could be relevant even for noneukaryotic organisms as these enzymes are present across all domains of life.

Experimental procedures

Bioinformatic analysis

The ADE5,7 gene and protein sequences were obtained from the C. neoformans H99O-type strain genome published by Janbon et al. (35). The gene encoding GARs and AIRs was identified in the C. neoformans genome via tBLASTn search using the S. cerevisiae homolog. Reciprocal analysis via BLASTp was performed using the translated product of CNAG_06314 to query the S. cerevisiae genome. Sequence alignments of GARs and AIRs from C. neoformans, S. cerevisiae, and H. sapiens were performed using ClustalW v1.4 (MacVector Inc).

Media and strains

All strains are listed in Table S1. Wild-type (C. neoformans H99O) and ade5,7Δ+ADE5,7 complemented strains were grown in yeast peptone dextrose (YPD) media (2% bacto-peptone, 1% yeast extract, 2% glucose, 2% agar) at 30 °C, unless otherwise stated. The ade5,7Δ mutant was grown in yeast nitrogen base (YNB) media (0.45% yeast nitrogen base without amino acids and ammonium sulfate, 10 mM (NH₄)₂SO₄, 2% glucose, 2% agar) supplemented with 1 mM adenine at 30 °C, unless otherwise stated. Mach1 E. coli (Thermo Fisher Scientific) strains were used for cloning the expression vectors and BL21 E. coli (Agilent Technologies) strains were used for protein expression. Mach1 E. coli was grown on lysogeny broth (LB) agar (1% tryptone, 0.5% yeast extract, 1% sodium chloride, 2% agar) supplemented with 100 μg/ml ampicillin at 37 °C. BL21 E. coli was grown in autoinduction growth media (0.5% yeast extract, 1% tryptone, 2%, 0.5% glycerol, 0.05% glucose, 0.2% lactose, 0.5 M (NH₄)₂SO₄, 0.5 M KH₂PO₄, 1 M Na₂HPO₄, 1 mM MgSO₄) for protein expression.

Creating ADE5,7 gene deletion and complementation strains

All primers (IDT) used in this study are listed in Table S2 and plasmids in Table S3. Phusion High Fidelity polymerase (New England Biolabs) was used for generation of all PCR amplicons. Genomic DNA extracted from C. neoformans wild-type was used as a template to amplify 5′ upstream and 3′ downstream regions of ADE5,7 and pJAF1 was used to amplify NEO (36). 1 kb upstream of ADE5,7 (UQ4315, UQ3880), the NEO marker (UQ3881, UQ3882), and 1 kb downstream of ADE5,7 (UQ3883, UQ4316) were combined using NEBuilder HiFi DNA Assembly (New England Biolabs) to create a deletion construct that was cloned into pBlueScriptII SK(-) to create pSMHC8. CLC Genomics Workbench 7.5 (CLC bio) was used to analyze sequencing data generated from the Australian Genome Research Facility (AGRF).

Biolistic transformation of the deletion construct was performed using a BioRad He-1000 Biolistic Device (37) into C. neoformans H99O strain on YNB media supplemented with 10 mM adenine, 1 M sorbitol, and 5 μg/ml W7 hydrochloride (38). Transformants were subsequently selected on YNB media supplemented with 10 mM adenine and 100 μg/ml G418. G418-resistant colonies were characterized via diagnostic PCR (UQ3132 and UQ1036) and tested for adenine auxotrophy on YNB media. To verify that ADE5,7 was successfully deleted from C. neoformans, CTAB extracted DNA from the ade5,7Δ mutants (39) was digested with EcoRV and NcoI (New England Biolabs), then run on an agarose gel at 20 V overnight. DNA from the agarose gel was Southern blotted to a Hybond-XL membrane (GE Healthcare) (40). The ADE5,7 probe (UQ3884 and UQ3885) was radioactively labeled using DECAprime II kit (Thermo Fisher Scientific) and dCTP [α-³²P] (PerkinElmer). The blots were hybridized with the radioactively labeled probe overnight at 65 °C, then washed with 0.1% SDS, 2× SSC buffer, exposed on Fuji Super RX medical X-ray film (Fujifilm), and was visualized after development.

Complementation of the ade5,7Δ mutant was performed through reintroduction of ADE5,7 at the Safe Haven locus in chromosome 1 (26). The region consisting of 5′ region of ADE5,7, the ADE5,7 CDS, and 3′ region of ADE5,7 (UQ4315, UQ4316) was cloned into Safe Haven, targeting pSDMA25, to create pSMHC1. pSMHC1 was linearized with PacI and transformed into the Safe Haven location of ade5,7Δ mutant. Transformants were selected on YPD media supplemented with 100 μg/ml nourseothricin (NAT). Transformant colonies were screened by diagnostic PCR (UQ1768, UQ2962, UQ2963, and UQ3348) and for adenine prototrophy (26). Precise integration of ADE5,7 at the Safe Haven location was verified by Southern blot with the same probe used for verifying the ade5,7Δ mutant.

Phenotype and virulence assays

C. neoformans strains were suspended in 1 ml of dH₂O and diluted to OD₆₀₀ = 1.0. For observation of adenine auxotrophy, the cell suspension was diluted by 10-fold serially to 10⁻⁴ and spotted on YPD, YNB, and YNB supplemented with 1 mM adenine. Plates were incubated at 30 °C and 37 °C. Images were taken 48 h postincubation. All growth spotting assays were done in triplicate.

For observation of virulence trait production, 3 μl of 10⁻¹ dilution suspension was spotted on media as follows: l-DOPA agar supplemented with 10 mM asparagine for melanin production (41); Christensen's urea agar for urease production (42); Sabouraud’s dextrose agar supplemented with 8% egg yolk for phospholipase B production (43); and YNB complete agar supplemented with 1% BSA (BSA, Sigma-Aldrich) for protease production. All media used for virulence assays were supplemented with 1 mM adenine to compensate for the adenine auxotrophy of the ade5,7Δ mutant. Plates were incubated at 30 °C and 37 °C. Images were taken 48 h postincubation for l-DOPA agar, Christensen’s urea agar, and BSA agar while images were taken 72 h postincubation for egg yolk agar. All virulence trait production spotting assays were done in triplicate.

For capsule assays, strains were grown in RPMI 1640 (Life Technologies) supplemented with 2% glucose, 10% foetal bovine serum (Life Technologies), and 1 mM adenine in a shaking incubator at 30 °C or 37 °C. After 24 h growth, cells were stained with India ink (BD Diagnostics), viewed, and imaged using a Leica DM2500 microscope and DFC425C camera (Leica). Five images were photographed for each strain. The whole cell and cell body diameter of 5 cells from each image were measured using the ruler tool in Adobe Photoshop CC (Adobe Systems), and capsule measurement was presented as a relative percentage of cell size. The capsule assays were performed in triplicate. A one-way ANOVA test with Sidak’s posttest was used to determine significance of differences in capsule sizes between strains.

Mouse virulence experiments

Ten 6-week-old mice were intranasally infected with 5 × 10⁵ cells of C. neoformans H99O wild-type strain, the ade5,7Δ mutant, and ade5,7Δ+ADE5,7 complemented strain. Five mice were housed in an IVC cage, which was lined with Bed-o’Cobs 1/8” bedding (The Andersons), Crink-L’Nest nesting material (The Andersons), and cardboard. Mice were provided with constant supply of Rat and Mouse Cubes (Specialty Feeds) and drinking water. After infection, the mice were monitored for signs of C. neoformans infection, scored, and weighed daily. When the mice lost 20% of their starting body weight or showed severe symptoms, they were euthanized by carbon dioxide inhalation. The spleen, liver, kidney, lungs, and brains were collected and homogenized using TissueLyser II (QIAGEN). The organ suspensions were serially diluted and spotted on YPD agar supplemented with 100 μg/ml ampicillin, 50 μg/ml kanamycin, and 25 μg/ml chloramphenicol, to prevent bacterial contamination. After 48 h, the colonies were counted to calculate the organ burden. The survival curves and organ burden plots were plotted using Graphpad Prism 7.0 (GraphPad Software Inc). The log rank test and a one-way ANOVA with Tukey’s multiple comparisons test were used to determine the significance of survival and the organ burden, respectively.

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 (AEC approval number: SCMB/010/17). 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.

Protein expression and purification

The ADE5,7 ORF was amplified from three custom gBlock gene fragments (uqg19, uqg20, uqg21; Table S4) (IDT), using Phusion High Fidelity polymerase in a two-step amplification. The ORF was amplified in two fragments (UQ3890 and UQ3869, UQ3868 and UQ3891) and then a second round of overlap PCR (UQ3890 and UQ3891) to form the entire ADE5,7 ORF. The amplified ADE5,7 ORF was digested with BamHI (New England Biolabs) then cloned into BamHI/SspI-cut pMCSG7 to introduce the His-tag into the ORF and create the protein expression plasmid pSMHC2, which was verified by Sanger sequencing. The ORF for the GARs region of ADE5,7 was amplified from pSMHC2 using primers UQ4282 and UQ4283, then cloned into the SspI/BamHI cut pMCSG7 using NEBuilder HiFi DNA Assembly, to create the plasmid pSMHC10, which was verified by sequencing. The ORF for the AIRs region of ADE5,7 was amplified from pSMHC2 using primers UQ4284 and UQ4285, then cloned into the SspI/BamHI cut pMCSG7 using NEBuilder HiFi DNA Assembly to create the plasmid pSMHC11, which was verified by Sanger sequencing.

Expression constructs were transformed into BL21(DE3) E. coli and grown in a shaking incubator at 37 °C for 3 h in autoinduction growth media (44) supplemented with 100 μg/ml ampicillin until OD₆₀₀ reached a value of 1.0. A 4 L culture was then grown in an incubator shaking at 220 rpm at 18 °C overnight. The cells pelleted from the culture was resuspended in lysis buffer (50 mM HEPES, 300 mM NaCl, 1 mM DTT, 100 mM phenylmethylsulfonyl fluoride). The cells were lysed by sonication and the lysate was collected by centrifugation at 16,000 rpm. A HisTrap Fast Flow column (GE Healthcare) was used for purification and the protein was eluted using an elution buffer containing 150 mM imidazole (50 mM HEPES, 300 mM NaCl, 150 mM imidazole). The eluted protein was then concentrated using an Amicon Ultra 15 ml centrifugal filter device (Merck Millipore). The protein was further purified using a Superdex 200 Increase column at a rate of 2.5 ml/min (GE Healthcare) that was equilibrated with gel filtration buffer (10 mM HEPES pH 7.5, 150 mM NaCl, 1 mM DTT) using an ÄKTA pure FPLC system (GE Healthcare) to separate the protein by size. The fractions of the eluted protein corresponding to a single peak were pooled and concentrated to 10 mg/ml. The purified protein was stored at –80 °C for future use.

Mass photometry

A Refeyn OneMP instrument (Refeyn Ltd) was used to measure the oligomeric states of the proteins in solution (45). Ten microliter of gel filtration buffer followed by 1 μl of the protein solution was applied to the drop to a final concentration of 100 nM and 6000 frames were recorded. The calibration curve was obtained by using three protein standards (66, 146, and 480 kDa) (Thermo Fisher Scientific).

GARs enzyme kinetic assays

For the synthesis of PRA, 5 mg of ribose-5-phosphate, 250 μl of NH₄Cl, and 50 μl 5 M KOH were incubated at 37 °C for 30 min and the product was quantified according to Schendel and others (28). The GARs activity was monitored spectrophotometrically using a Cary60 UV-Vis spectrophotometer (Agilent). All enzyme activity assays were performed in biological triplicates. The standard assay was performed based on the approach used previously for the E. coli GARs, with the modification running the assay at 37 °C using 0.0175 mg GARs and 2 mM PRA (46). As it is unstable at higher temperatures, the reaction was initiated with the addition of PRA rather than the addition of the enzyme. GARs activity was determined by following the oxidation of NADH at 340 nm (ε340 = 6220 M⁻¹ cm⁻¹). The concentration ranges for measuring the K_m of substrates ATP, PRA, and glycine ranged from 4 μM to 2000 μM, 16 μM to 2000 μM, and 8 μM to 4000 μM. Enzyme kinetic data were processed and fit to the Michaelis–Menten equation using Graphpad Prism 7.0 (GraphPad Software Inc).

Crystallization and structure determination of C. neoformans GARs and AIRs

All crystallization attempts were performed via hanging drop vapor diffusion method. Eight commercial screening plates, Grid Screen Sodium Malonate, Grid Screen Ammonium Sulfate, Grid Screen Sodium chloride, Grid Screen PEG/LiCl, PEGION, and PEGRX (Hampton Research), JSCG, Morpheus, PACT, PROPLEX, and ShotGun (Molecular Dimensions) were used to screen crystallization conditions with protein concentrations of 10 mg/ml. A Mosquito crystallization robot (TTP Labtech) was used to set up screening plates. Each drop contained 100 nl protein solution and 100 nl reservoir solution inverted over 100 μl reservoir solution. The crystallization screens were incubated at 20 °C and monitored by the Rock Imager system (Formulatrix). The GARs protein crystallized in 0.1 M sodium citrate, pH 5.0, 30% v/v MPEG550, and the AIRs protein crystallized in 0.1 M SPG (succinic acid, sodium dihydrogen phosphate, and glycine in ratio of 2:7:7), 25% w/v PEG1500.

Diffraction data were collected on the MX2 beamline of the Australian Synchrotron. Data were collected using Blu-Ice (47), and processed and scaled using XDS (48) and Aimless (49). The C. neoformans GARs structure was solved by molecular replacement using Phaser (50) in the PHENIX suite version 1.14 (51), with H. sapiens GARs (PDB ID: 2QK4) as the template model. The C. neoformans GARs model was refined using data between 38 Å and 2 Å resolution, and in-between rounds of refinement model building were conducted with Coot version 0.8.9.2 (52). The data collected were analyzed using Phenix xtriage, verifying the absence of twinning symmetry, outliers, and abnormalities. The C. neoformans AIRs were solved by molecular replacement using Phaser (50) in the PHENIX suite version 1.14 (51) with the Neisseria gonorrhoeae AIRs (PDB ID: 5VK4) as the template model. The AIRs model was refined using data between 42 Å and 2.23 Å resolution, and in-between rounds of refinement model building were conducted with Coot version 0.8.9.2 (52). Structures in the PDB similar to C. neoformans GARs and AIRs were identified using DALI (33). C. neoformans and H. sapiens enzyme structures were superimposed using the Pymol align command. Interacting residues were analyzed using PISA (53).

Data availability

The atomic coordinates and structure factors (PDB ID: 7LVO, 7LVP) have been deposited in the Protein Data Bank (https://www.rcsb.org/).

Supporting information

This article contains supporting information (26, 58).

Conflict of interest

The authors declare that they have no conflicts of interest with the contents of this article.

Edited by Joseph Jez

Supporting information

Figures S1–S3 and Table S1–S6