Kinetic and Structural Characterization of Trypanosoma cruzi Hypoxanthine–Guanine–Xanthine Phosphoribosyltransferases and Repurposing of Transition-State Analogue Inhibitors

Abstract

Over 70 million people are currently at risk of developing Chagas Disease (CD) infection, with more than 8 million people already infected worldwide. Current treatments are limited and innovative therapies are required. Trypanosoma cruzi, the etiological agent of CD, is a purine auxotroph that relies on phosphoribosyltransferases to salvage purine bases from their hosts for the formation of purine nucleoside monophosphates. Hypoxanthine–guanine–xanthine phosphoribosyltransferases (HGXPRTs) catalyze the salvage of 6-oxopurines and are promising targets for the treatment of CD. HGXPRTs catalyze the formation of inosine, guanosine, and xanthosine monophosphates from 5-phospho-d-ribose 1-pyrophosphate and the nucleobases hypoxanthine, guanine, and xanthine, respectively. T. cruzi possesses four HG(X)PRT isoforms. We previously reported the kinetic characterization and inhibition of two isoforms, TcHGPRTs, demonstrating their catalytic equivalence. Here, we characterize the two remaining isoforms, revealing nearly identical HGXPRT activities in vitro and identifying for the first time T. cruzi enzymes with XPRT activity, clarifying their previous annotation. TcHGXPRT follows an ordered kinetic mechanism with a postchemistry event as the rate-limiting step(s) of catalysis. Its crystallographic structures reveal implications for catalysis and substrate specificity. A set of transition-state analogue inhibitors (TSAIs) initially developed to target the malarial orthologue were re-evaluated, with the most potent compound binding to TcHGXPRT with nanomolar affinity, validating the repurposing of TSAIs to expedite the discovery of lead compounds against orthologous enzymes. We identified mechanistic and structural features that can be exploited in the optimization of inhibitors effective against TcHGPRT and TcHGXPRT concomitantly, which is an important feature when targeting essential enzymes with overlapping activities.

Neglected tropical diseases are defined by the World Health Organization as a diverse group of 20 different diseases, including Chagas Disease (CD), that disproportionally affect the world’s poorest populations.¹ Approximately 8 million people are infected with CD worldwide, causing over 10 000 deaths annually.^2,3 It is traditionally considered a Latin American disease, where it is endemic in 21 out of 33 countries, with nearly 85% of cases originating from rural areas. The burden of CD has spread to non-endemic countries and nonrural areas, with nearly 75 million people currently at risk of infection, raising global interest in developing effective therapies.^2,4 Over 300 000 cases are currently estimated in the U.S. according to serological screenings,² likely an underestimation considering the scarcity of surveillance data. California, Texas, and Florida show the highest incidence rates,⁴ with approximately 20% of cases reported in Texas from 2013 to 2019 confirmed to be locally acquired infections.⁵ CD is a parasitic infection caused by the protozoan Trypanosoma cruzi, and its current treatment includes the trypanocidal drugs benznidazole and nifurtimox, both of which are ineffective in treating the advanced, chronic stage of the disease.^6,7 The high frequency of adverse side effects, apparent lack of efficacy in chronic infections, and length of required drug regimens often result in abandonment of treatment.⁸⁻¹² Thus, new drugs with fewer side effects and improved efficacy against both the early and late stages of the disease are urgently needed.

All neglected tropical diseases, including CD, are often overlooked in comparison with the “big three”: malaria, tuberculosis, and HIV/AIDS.¹³ Malaria is also a parasitic infection, caused by protozoa from the Plasmodium genus, with the most widespread and lethal species being Plasmodium falciparum.¹⁴ Malaria affects over 200 million people worldwide, causing more than 400 000 deaths per year, and is endemic in 91 countries of Africa and Asia.³ Common drugs used to treat malaria include chloroquine, doxycycline, mefloquine, and quinine,^3,15−19 and a vaccine has recently been introduced, showing, however, modest efficacy and only partial protection in Phase 3 studies.^20,21 In addition, mutations in the parasite genome have emerged, resulting in the rapid spread of resistance against many commercial drugs.^22,23 Indeed, two of the most used drugs, chloroquine and mefloquine, are no longer effective in many parts of the world.^19,24 Therefore, the development of novel chemotherapies against malaria is paramount to improving disease outcomes.

T. cruzi and P. falciparum are both incapable of de novo purine synthesis and rely exclusively on the salvage of preformed purine bases for the biosynthesis of purine nucleoside monophosphates (NMPs).²⁵ A highly selective nucleobase/proton symporter system^26,27 facilitates the uptake of purine nucleobases from the host to the parasite, and purine phosphoribosyltransferases (PPRTs) convert salvaged purines into NMPs. The dominant nucleobases in human serum and cerebrospinal fluid are hypoxanthine (Hx) and xanthine (Xan), respectively, whereas other purine bases are present at much lower or even undetectable concentrations, indicating purine salvage funnels through inosine monophosphate (IMP) and xanthosine monophosphate (XMP) in Trypanosoma(28) and Plasmodia.²⁹T. cruzi and P. falciparum are incapable of growth in the absence of Hx salvage and metabolism.^28,30,31 Hypoxanthine–guanine–xanthine phosphoribosyltransferases (HGXPRTs, EC 2.4.2.8) catalyze the magnesium-dependent formation of IMP, guanosine monophosphate (GMP), and XMP from 5-phospho-d-ribose 1-pyrophosphate (PRPP) and the respective purine bases Hx, guanine (Gua), and Xan (Figure 1) and are the only means for incorporation of Hx and Xan into the purine pool. The substrate preference of these PPRTs is noted within their nomenclature; some isoforms can use all three 6-oxopurine bases as substrate (HGXPRTs), while others are selective to Hx (HPRTs, hypoxanthine phosphoribosyltransferases) or to Hx and Gua (HGPRTs, hypoxanthine–guanine phosphoribosyltransferases). RNAi studies in the closely related organism Trypanosoma brucei brucei demonstrate that the activity of HG(X)PRTs is essential for the parasite’s survival in cell cultures.³² Consequently, HG(X)PRTs are promising drug targets for the development of novel chemotherapies against both CD and malaria.^28,33

Figure 1.

Reaction catalyzed by HGXPRT (EC 2.4.2.8). The atom numbers of the purine scaffold and ribose ring are shown in gray on product NMP.

P. falciparum expresses a single HGXPRT isoform (PfHGXPRT). Comprehensive characterization of the catalytic mechanism using steady-state kinetics, kinetic isotope effects, and computational quantum chemistry elucidated its transition-state (TS) structure.³⁴⁻³⁷ Immucillins, transition-state analogue inhibitors (TSAIs) initially developed as potent inhibitors of human and P. falciparum purine nucleoside phosphorylase, have been demonstrated to be exceptionally potent and selective inhibitors of PfHGXPRT.^{31,35,38−42} The T. cruzi genome encodes four HG(X)PRT isoforms. We recently characterized two isoforms as HGPRT (TcHGPRTs—isoforms TcA and TcC), demonstrating their activity and catalytic equivalence, substrate preferences for Hx and Gua, kinetic mechanism, and identification of the rate-limiting step(s).⁴³ We also reported on a subset of Immucillins that are inhibitors of TcHGPRT activity in vitro, the most potent of which exhibited low nanomolar affinity.⁴³

Here, we report on the catalytic activity and structure of the remaining T. cruzi isoforms (TcB and TcD) and demonstrate for the first time an enzyme with xanthine phosphoribosyltransferase (XPRT) activity within the T. cruzi purine salvage pathway, therefore identifying these enzymes as true HGXPRTs (TcHGXPRT). The presence of multiple HG(X)PRT isoforms in T. cruzi suggests that blocking one enzyme may not be sufficient to inhibit T. cruzi growth, as studies in T. brucei brucei indicate that the residual activity of any one of the HG(X)PRTs allows for parasite persistence.³² Therefore, we evaluated the inhibitory potential of Immucillins and derivatives against the TcHGXPRT isoforms. We exploited the catalytic and structural similarities of TcHG(X)PRTs and the single PfHGXPRT to evaluate these TSAIs as potential inhibitors of TcHG(X)PRTs. Our results indicated that Immucillins previously shown to be effective against TcHGPRTs are also potent inhibitors of TcHGXPRTs, although the potencies varied according to substrate specificity of the isoforms. Taken together, the data indicate that Immucillins and their derivates can be repurposed as potential lead compounds to treat CD.

Materials and Methods

Unless otherwise indicated, all materials were purchased from Sigma-Aldrich/Sigma Millipore (Burlington, MA). Effective concentrations of PRPP were ascertained and corrected for purity as indicated by the supplier (75% purity), as previously described.⁴³

Expression and Purification of TcHGXPRTs (TcB and TcD), EC 2.4.2.8

The coding sequences of TcB (TcCLB.509693.80, UNIPROT Q4DRC3) and TcD (TcCLB.506457.40, UNIPROT Q4DGA2), were codon-optimized individually for expression in Escherichia coli, synthesized, and cloned into a pET-28a(+) expression vector using NdeI and HindIII restriction sites (GenScript, USA), with a 6×His-tag upstream of the N-terminus. Optimal expression of TcB and TcD was obtained in BL21(DE3) and C43(DE3) cells respectively, using Terrific Broth (TB) media. The plasmids pET-28a(+)::TcB and pET-28a(+)::TcD were individually transformed by heat shock into BL21(DE3) or C43(DE3) cells and incubated overnight at 37 °C on plates containing 50 μg mL^–1 of kanamycin. A single colony of the respective strain was used to inoculate a 50 mL starter culture of TB media in the presence of 50 μg mL^–1 kanamycin. Starter cultures were incubated overnight at 37 °C in an orbital shaker set at 180 rpm. A 13 mL sample of starter culture was used to inoculate 500 mL of TB media in the presence of kanamycin at 50 μg mL^–1. Expression cultures were grown at 37 °C and set to 180 rpm until an OD₆₀₀ of 0.4–0.6 was reached. Protein expression was induced by the addition of IPTG (1 mM final concentration). Cells were grown at 18 and 37 °C, for TcB and TcD expression, respectively, for an additional 24 h. Expression cultures were harvested by centrifugation at 5000 rpm for 45 min, at 4 °C, and cell pellets were stored at −20 °C until purification. Cell pellets were resuspended in Buffer A (50 mM Na₂HPO₄/NaH₂PO₄, pH 7.4, 300 mM NaCl). Lysozyme was added at a final concentration of 0.2 mg mL^–1, and cells were disrupted by sonication (60% amplitude for 15 cycles of 10 s). Cell lysates were clarified by centrifugation (18 000 rpm for 45 min at 4 °C). A 10 mL aliquot of HisPur Ni-NTA resin (Thermo Fisher Scientific) pre-equilibrated with Buffer A containing 10 mM imidazole (EQ Buffer) was used to purify the His-tagged proteins at room temperature (RT). The resin was washed with 5 column volumes (CV) of EQ Buffer, followed by 5 CV of Buffer A containing 20 mM imidazole, followed by 5 CV of Buffer A containing 50 mM imidazole. Protein was eluted with 25 CV of Buffer A containing 300 or 600 mM imidazole, in a stepwise gradient. Analysis with 12% SDS-PAGE stained with Coomassie Blue was used to determine fractions containing the proteins of interest (Figure S1), which were then pooled and dialyzed against 50 mM EPPS (pH 8.3), 300 mM NaCl, 50 mM l-arginine, and 50 mM l-glutamate at 4 °C, using 12–14 kDa MWCO dialysis membranes (Spectrum Spectra/Por 4 regenerated cellulose). Homogeneous recombinant proteins were stored in the presence of 10% (v/v) glycerol at −80 °C. TcB and TcD purification yields were, respectively, 11 and 56 mg of homogeneous protein per gram of cell paste.

Size-Exclusion Chromatography (SEC) of Homogeneous TcB and TcD

TcB and TcD were subjected to SEC to verify the oligomeric state and to ensure purified protein was homogeneous and not aggregated. A 200 μL aliquot of TcB (9.1 mg mL^–1) or TcD (8.2 mg mL^–1) was applied to a GE Superdex 200 Increase 10/300 GL SEC column using an ÄKTA pure (GE Healthcare), equilibrated with SEC Buffer (50 mM HEPES, pH 7.8, 300 mM NaCl, 50 mM l-arginine, 50 mM l-glutamate), at 4 °C. Proteins were eluted off with 1.2 CV of SEC Buffer at a flow rate of 0.25 mL min^–1. Prior to sample application, the column was calibrated with lyophilized protein standards (BioRad, Cat #1511901) containing a mixture of thyroglobulin (670 kDa), bovine γ-globulin (158 kDa), chicken ovalbumin (44 kDa), equine myoglobin (17 kDa), and vitamin B12 (1.3 kDa). The lyophilized standards were resuspended in SEC Buffer. Blue dextran (15 mg mL^–1) resuspended in SEC Buffer was used to measure the void volume of the column. Samples were compared to these standards to ascertain their apparent oligomeric state.

Kinetic Assays for Measuring HGXPRT Activity

Steady-state kinetic assays for the forward and reverse reactions were measured using continuous spectrophotometric assays as previously described.⁴³ Briefly, all assays were carried out in Assay Buffer (50 mM EPPS, pH 8.3, 12 mM MgCl₂) and measured in Greiner Bio-One UV-Star 96-well microplates using a BioTek Synergy HTX plate reader (final volume of 250 μL). The steady-state formation of IMP, GMP, and XMP was monitored spectrophotometrically at, respectively, 245, 257.5, and 253 nm (Δε₂₄₅ = 1227 M^–1 cm^–1, Δε_257.5 = 4044 M^–1 cm^–1, Δε₂₅₃ = 4250 M^–1 cm^–1). Prior to the determination of kinetic rate constants, the concentration of each protein was varied in the presence of saturating concentrations of both substrates (1 mM PRPP and 120 μM 6-oxopurine) to determine the optimal protein concentration to obtain an initial rate for kinetic analysis.

Apparent kinetic parameters for TcB and TcD (app K_m, app k_cat, and app k_cat/K_m) were determined in Assay Buffer from initial rate measurements in triplicate with at least seven variable concentrations of substrate and at saturating concentrations of the second substrate. Apparent kinetic parameters for TcB (100 nM) with the 6-oxopurine substrates were determined with reaction mixtures containing variable concentrations of Hx (5–150 μM), Gua (10–250 μM), or Xan (1.5–100 μM) in the presence of saturating concentrations of PRPP (1 mM). Apparent kinetic parameters for TcD (100 nM) with the 6-oxopurine substrates were determined with reaction mixtures containing variable concentrations of Hx (25–350 μM), Gua (25–350 μM), or Xan (2.5–150 μM) in the presence of a saturating concentration of PRPP (1 mM). Apparent kinetic parameters for TcB (200 nM) and TcD (200 nM) in the presence of variable concentrations of PRPP (4–115 μM) were determined with reaction mixtures containing a saturating concentration of Gua (1 mM). Apparent kinetic parameters for the cofactor Mg²⁺ were determined with reaction mixtures containing variable concentrations of MgCl₂ (0–100 mM) in the presence of saturating concentrations of PRPP (1 mM), Hx (120 μM), and TcHGXPRT (400 nM). All of the apparent kinetic parameters were calculated by fitting of initial rate data to eq 1.

All further kinetic assays were performed using TcHGXPRT isoform B (TcB), except for the crystallographic studies (see below).

Initial rate studies with TcHGXPRT (100 nM) were performed in triplicate with different fixed concentrations of Gua (5, 10, 25, 50, and 100 μM) vs. variable concentrations of PRPP (1.15–80 μM) or fixed concentrations of PRPP (1.5, 2.5, 3.3, 5, and 10 μM) vs. variable concentrations of Gua (10–300 μM). GMP biosynthesis was measured as described above for guanine phosphoribosyltransferase (GPRT) activity. Initial rate data where one substrate was fixed at multiple different and fixed concentrations vs. variable concentrations of the second substrate were fit to eq 2.

Dead-End Inhibition Studies

The inhibition pattern of 9-deazaxanthine vs. Xan for TcHGXPRT (100 nM) was determined at a saturating concentration of PRPP (1 mM), fixed concentrations of 9-deazaxanthine (0, 5, 10, 25, 100, and 250 μM), and variable concentrations of Xan (3–100 μM). The inhibition pattern of 9-deazaxanthine vs. PRPP for TcHGXPRT (300 nM) was ascertained at a saturating concentration of Xan (120 μM), fixed concentrations of 9-deazaxanthine (0, 100, 250, 400, 500, and 600 μM), and varied concentrations of PRPP (23–750 μM). Initial rate data were collected in triplicate and fit to eqs 3–5 to determine if the inhibition pattern was competitive (eq 3), noncompetitive (eq 4), or uncompetitive (eq 5). Initial rate data were subsequently plotted as double-reciprocal plots to confirm the inhibition pattern. XMP formation was measured as previously described for the XPRT activity assays above.

Product Inhibition

The inhibition pattern of PPi vs. Xan for TcHGXPRT (100 nM) was determined with initial rate data at a saturating concentration of PRPP (1 mM), different fixed concentrations of PPi (0, 25, 50, 100, 200 μM), and variable concentrations of Xan (2.5–150 μM). The inhibition pattern of XMP vs. PRPP on TcHGXPRT catalysis was determined with reaction mixtures containing a saturating concentration of Xan (120 μM), variable concentrations of PRPP (11–750 μM), different fixed concentrations of XMP (0, 8, 15, 30, and 60 μM), and TcHGXPRT (300 nM). Initial rate data were collected in triplicate, fit to eqs 3 to 5 to determine the inhibition pattern (competitive, noncompetitive, or uncompetitive), and subsequently plotted in double-reciprocal form. XMP formation was measured as described for the XPRT activity assays above.

Viscosity Effects

Viscosity effects on TcHGXPRT were determined by measuring the initial rate in the presence of varying relative viscosity (η_rel) levels. Reaction mixtures containing either 0, 4, 8 and 12% (v/v) glycerol (microsviscosogen; η_rel = 1–1.45) or 0, 0.5, 1, and 2% (w/v) PEG₁₀₀₀₀ (macroviscosogen; η_rel = 1–1.46) were used to evaluate their effect on initial rates. The kinematic viscosity values were measured using a Cannon-Fenske viscometer (capillary no. 100) and were calculated using eq 6. The dynamic viscosities (reflecting internal resistance to fluid) were calculated by using eq 7. Values of η_rel were calculated using the relationship described in eq 8. Reactions mixtures contained 300 nM TcHGXPRT, 20 μM Hx and variable PRPP (0–300 μM) or 1 mM PRPP and variable Hx (4.7–150 μM), and fixed concentrations of glycerol (0, 4, 8, and 12% v/v) or PEG₁₀₀₀₀ (0, 0.5, 1.0, and 2.0% w/v) in Assay Buffer. Kinetic data were collected in triplicate. IMP formation was measured as previously described for the HGXPRT activity assays. Kinetic parameters were determined by fitting initial rate data to eq 9,⁴⁴ to directly afford the values of k_cat and k_cat/K_m at each of the values of η_rel. Data were plotted as the normalized kinetic parameter (i.e., [k_cat/(k_cat)_η] or [(k_cat/K_m)/(k_cat/K_m)_η]) as a function of η_rel and fit to eq 10.⁴⁵

Screening of Inhibitors

Inhibitors of TcHGXPRT were evaluated in assays containing Hx (120 μM), PRPP (1 mM), PPi (100 μM), varying concentrations of inhibitor, and TcHGXPRT (300 nM), in which IMP formation was measured as described above. The apparent potencies of TcHGXPRT inhibitors (IC₅₀) were determined with initial rate data collected in triplicate and fit to eq 11. Intrinsic potencies (K_i) were determined by using the relationship described in eq 12. The inhibition pattern (competitive, uncompetitive, or noncompetitive) was determined with concentration–response curves performed with 15, 30, 60, 90, 120, and 150 μM Hx in the presence of 1 mM PRPP and 400 nM TcHGXPRT at different fixed concentrations of (S)-2 (0, 0.6, 1.1, 2.3, 4.5, and 9 μM). Initial rate data of (S)-2 vs. Hx were fit to eqs 3 and 5, and the type of inhibition was determined by the model that provided the best fit. The selectivity index (SI) was calculated as the app K_i for human HGPRT (HsHGPRT) divided by the app K_i for TcHGXPRT. The synthesis of the TcHG(X)PRT inhibitors presented in this study, the inhibition assays, and data for TcHGPRT, HsHGPRT, and PfHGXPRT have been previously described.^39,41,43

X-ray Crystal Structure Determination of TcHGXPRT in Complexes with Xanthine, Phosphate, and Inhibitors

Prior to crystallization, TcHGXPRT (TcD isoform) was applied to a GE HiPrep 26/60 Sephacryl S-100 HR SEC column using ÄKTA pure (GE Healthcare). TcHGXPRT was eluted off with 1.2 CV of SEC Buffer (20 mM HEPES, pH 7.4, 150 mM NaCl, and 50 mM (NH₄)₂SO₄) at a flow rate of 0.5 mL min^–1 at 4 °C. TcHGXPRT (8 mg mL^–1) was preincubated with substrate Xan (1 mM) and product PPi (1 mM) for 0.5 to 1 h to allow for the formation of the putative TcHGXPRT·Xan·PPi dead-end ternary complex. Crystallization of TcHGXPRT·Xan·PPi was performed using the hanging drop vapor diffusion method at 18 °C. Each drop (2–3 μL) was prepared by pipetting and mixing the TcHGXPRT·Xan·PPi solution to mother liquor (0.2 M LiSO₄, 24% PEG₃₃₅₀, 0.1 M HEPES, pH 8.0) in ratios of 1:1, 1:2, and 2:1 (protein:mother liquor), all of which yielded crystals with similar size and morphology. Large crystals appeared after 2 days and matured in ∼2 weeks.

Inhibitors (S)-2, (R)-2, 6, 8, 9, and DADME-ImmH^46,47 (0.5 mM solutions in EPPS 50 mM, pH 8.3) were added directly to the TcHGXPRT·Xan·PPi complex crystals using the soaking method and incubated overnight at 18 °C. All crystals were cryoprotected with 20% glycerol and flash frozen in liquid N₂. Crystals soaked with DADME-ImmH did not show the inhibitor in the active site but instead yielded a TcHGXPRT·Xan·Pi structure with improved resolution compared to the crystals not submitted to the soaking processing.

The diffraction data were collected on a Pilatus3 6 M detector, 23ID of the GM/CA-CAT facilities of the Advanced Photon Source at Argonne National Library. JBlulce-EPIC software was used to collect images with an exposure time of 1.0 s/frame. The data were collected at the Se–K X-ray absorption edge at 12 658 eV. Data collection, unit cell parameters, and refinement statistics are given in the Supporting Information. (Table S1). Data sets were processed using Phenix.⁴⁸TcHGPRT (TcA isoform, 1TC2.pdb⁴⁹) was used as the template for molecular replacement. Solutions were subjected to simulated annealing, multiple refinement cycles of maximum likelihood energy minimization, and B-factor refinement with Phenix.⁴⁸ CCP4 was used to convert sca to mtz files. Coot⁵⁰ was used to fit amino acid side chains into σ-weighted 2F₀–F_c and F₀–F_c electron density maps. Water molecules were added automatically and inspected manually using Coot.⁵⁰ PyMOL 2.4.1⁵¹ and LigPlot+⁵² were used for figure preparation of solved X-ray structures. RMSD calculations were done using Phenix⁴⁸ and Deep View/Swiss-Pdb Viewer 4.1.0.⁵³ Amino acids that have not been fitted either are missing as coordinates or are part of the highly flexible active site loop (loop 2).

The PDB coordinates are 8FWY.pdb (TcHGXPRT·Xan·Pi), 8FX0.pdb (TcHGXPRT·(S)-2), 8FX1.pdb (TcHGXPRT·(R)-2), 8FWZ.pdb (TcHGXPRT·6), 8FX2.pdb (TcHGXPRT·8), and 8FX3.pdb (TcHGXPRT·9).

Data Analysis

Steady-state and inhibition data of TcHGXPRTs were fit using GraphPad Prism 9.4 software.

Initial rate data (v) obtained at variable concentrations of substrate A and at apparent fixed saturating concentrations of second substrate B were fit to eq 1, in which v is the initial rate, E_t is the enzyme concentration, k_cat is the turnover number, A is the concentration of the variable substrate, and K_a is the apparent Michaelis constant.

1

Initial rate data (v) obtained at variable concentrations of a single substrate A at several fixed concentrations of the second substrate B were fit to eq 2, in which E_t is the concentration of enzyme, k_cat is the turnover number, K_a and K_ia are the respective Michaelis constants and dissociation constant of substrate A, and K_b is the Michaelis constant of substrate B.

2

Initial rate data (v) obtained during inhibition studies where double-reciprocal plots of inhibitor (I) vs. variable substrate (A) apparently conformed to competitive, noncompetitive, or uncompetitive inhibition were fit to eqs 3 to 5, respectively. E_t is the concentration of enzyme; k_cat and K_a are, respectively, the turnover number and the Michaelis constant for substrate A, and K_is and K_ii are, respectively, the apparent slope and intercept inhibition constants. All inhibition data were fit to eqs 3 and 5 to evaluate the best fit inhibition model.

3

4

5

The relationship described in eq 6 was used to determine the kinematic viscosity (η_kinematic) values, where K is the viscometer constant and t is the efflux time. The dynamic viscosities (η_dynamic) were calculated with eq 7, where d is the density of the liquid. The η_rel values were obtained according to eq 8, where η₀ is the viscosity when there is 0% viscosogen in the solution and η_x is the viscosity when there is X% viscosogen in the solution.

6

7

8

The initial rate data obtained at fixed concentrations of viscosogen, variable concentrations of the substrate A, and at saturating concentrations of the second substrate B were fit to eq 9,⁴⁴ to directly afford k_cat and k_cat/K_m parameters at each η_rel value, where E_t is the enzyme concentration, A is the concentration of variable substrate, k_cat is the turnover number, and k_SP is the k_cat/K_m for the varied substrate. The kinetic parameters were normalized to the viscosogen-free control (i.e., [k_cat/(k_cat)_η] or [(k_cat/K_m)/(k_cat/K_m)_η]) and plotted as a function of η_rel. Viscosity data were fit to eq 10, where k/k_η is the normalized kinetic parameter, and m is the slope of the line, which represents the dependence of viscosity on the normalized kinetic parameter.⁴⁵

9

10

The TcHGXPRT inhibitors were assessed for inhibitory activity in terms of % activity (% v_i/v₀) data and were fit to eq 11, in which, I is the concentration of inhibitor, IC₅₀ is the half-maximal inhibitory concentration, and h is the Hill slope.⁵⁴ Inhibition constants of TcHGPRT and TcHGXPRT inhibitors (K_i) were determined by use of the Cheng and Prusoff equation for uncompetitive inhibition (eq 12),^54,55 in which IC₅₀ is the half-maximal inhibitory concentration, A is the concentration of Hx, and K_a is the Michaelis constant for Hx, assuming uncompetitive inhibition as the mode of inhibition.

11

12

Results and Discussion

As the complex Mg·PRPP is the true substrate for all PPRTs,⁵⁶ the concentration of Mg²⁺ was kept at a constant, nonlimiting, and noninhibitory concentration in all assays (12 mM, Figure S2). In assays where PRPP is included, excess Mg²⁺ in solution converts all free PRPP into Mg·PRPP, and thus, the Mg·PRPP complex acts as the true substrate. When discussing the sequence and/or structure of the HG(X)PRTs, residues are numbered according to the TcHGXPRT primary sequence (Figure S3). When relevant, the equivalent position numbering of the orthologous enzymes is provided.

Steady-State Kinetics Reveals XPRT Activity and Substrate Preference of TcHGXPRT Isoforms

The reference strain used for T. cruzi genome sequencing (CL Brener) is a hybrid strain whose diploid set of genes is derived from the parental lineages Esmeraldo and non-Esmeraldo.⁵⁷ The T. cruzi genome encodes multiple isoforms of HG(X)PRTs; more specifically, these isoforms are currently annotated as one HGPRT enzyme (TcA - TcCLB.509693.70, UNIPROT Q4DRC4) and three putative HGPRTs: TcC (TcCLB.506457.30, UNIPROT Q4DGA3), TcB (TcCLB.509693.80, UNIPROT Q4GRC3), and TcD (TcCLB.506457.40, UNIPROT Q4DGA2). We have recently characterized the TcHGPRT isoforms (TcA and TcC), corroborating the TcA annotation and clarifying the TcC activity as an HGPRT while demonstrating their catalytic equivalency in vitro.⁴³TcHGPRTs share >98% sequence identity,⁴³ while the remaining isoforms, TcB and TcD, differ by only one amino acid at position 125 (Lys in TcB, and Gln in TcD—Figure S3). TcHGPRTs and the TcB/TcD isoforms only share ∼35% sequence identity (Figure S3); however, low conservation of primary sequence is not uncommon for this family of enzymes, the Type I Phosphoribosyltransferases (PRTases), which includes HG(X)PRTs, orotate phosphoribosyltransferases, and glutamine-PRPP amidotransferases.^56,58

We expressed and purified TcB and TcD individually to characterize their catalytic activity, establish substrate preference, and evaluate their independent activities in vitro. TcB and TcD both displayed strong preference for substrate Xan over Hx and Gua and present highly conserved kinetic parameters for all three 6-oxopurines, except for an approximate 10-fold difference in values of K_m and k_cat/K_m for Hx (13 μM and 92 × 10³ M^–1 s^–1, respectively, for TcB and 150 μM and 7 × 10³ M^–1 s^–1 for TcD; Table 1, Figure S4). Therefore, the TcB and TcD isoforms were characterized as true HGXPRT enzymes. The HGXPRT isoforms are less efficient at catalyzing the formation of IMP and GMP from PRPP and Hx and Gua, respectively (k_cat/K_m ≤ 1 × 10⁴ M^–1 s^–1) when compared to the TcHGPRTs (k_cat/K_m ≥ 1 × 10⁵ M^–1 s^–1), which can also be observed in the large differences in values of k_cat (30–60-fold lower for the TcHGXPRTs). This trend reverses when Xan is the substrate, with TcHGXPRTs showing k_cat/K_m ≥ 1 × 10⁵ M^–1 s^–1, while TcHGPRT activity using Xan as substrate is negligible (k_cat/K_m ≥ 1 × 10¹ M^–1 s^–1) when considering cellular conditions,⁴³ with values of k_cat 160–400-fold higher for the TcHGXPRTs.

Table 1. Kinetic Parameters Obtained from the Initial Rate Data of TcHG(X)PRTs.

Enzyme (Activity)	Reaction	Variable substrate	Fixed substrate	Apparent kinetic parametersa
TcB (HGXPRT)				Km(μM)	kcat(s–1)	kcat/Km(× 103M–1s–1)
	GMP biosynthesis	Gua (10–250 μM)	PRPP (1 mM)	72 ± 3	0.63 ± 0.02	9 ± 1
		PRPP (5–115 μM)	Gua (1 mM)	6 ± 1	0.51 ± 0.03	85 ± 15
	IMP biosynthesis	Hx (5–150 μM)	PRPP (1 mM)	13 ± 2	1.2 ± 0.1	92 ± 16
	XMP biosynthesis	Xan (1.5–100 μM)	PRPP (1 mM)	4.2 ± 0.3	0.97 ± 0.04	230 ± 20
TcD (HGXPRT)				Km(μM)	kcat(s–1)	kcat/Km(× 103M–1s–1)
	GMP biosynthesis	Gua (25–350 μM)	PRPP (1 mM)	45 ± 4	0.30 ± 0.01	7 ± 1
		PRPP (4–115 μM)	Gua (1 mM)	10 ± 4	0.6 ± 0.1	62 ± 20
	IMP biosynthesis	Hx (25–350 μM)	PRPP (1 mM)	150 ± 30	1.1 ± 0.1	7 ± 2
	XMP biosynthesis	Xan (2.5–150 μM)	PRPP (1 mM)	11 ± 0.4	2.5 ± 0.1	230 ± 10
TcA (HGPRT)				Km (μM)	kcat (s–1)	kcat/Km (× 103 M–1 s–1)
	GMP biosynthesis	Gua (5–200 μM)	PRPP (1 mM)	19 ± 1b	32 ± 1b	1700 ± 100b
		PRPP (20–215 μM)	Gua (70 μM)	32 ± 5c	32c	1010c
	IMP biosynthesis	Hx (5–150 μM)	PRPP (1 mM)	13 ± 4b	32 ± 1b	2500 ± 700b
		PRPP (20–215 μM)	Hx (60 μM)	31 ± 5c	23c	750c
	XMP biosynthesis	Xan (50–500 μM)	PRPP (1 mM)	160 ± 40b	0.005 ± 0.0006b	0.03 ± 0.008b
TcC (HGPRT)				Km(μM)	kcat(s–1)	kcat/Km(× 103M–1s–1)
	GMP biosynthesis	Gua (5–200 μM)	PRPP (1 mM)	14 ± 1b	30 ± 1b	2100 ± 200b
		PRPP (6–200 μM)	Gua (100 μM)	33 ± 3b	31 ± 1b	930 ± 90b
	IMP biosynthesis	Hx (5–150 μM)	PRPP (1 mM)	14 ± 2b	35 ± 2b	2500 ± 300b
		PRPP (6–200 μM)	Hx (100 μM)	39 ± 2b	24 ± 1b	620 ± 40b
	XMP biosynthesis	Xan (50–500 μM)	PRPP (1 mM)	150 ± 10b	0.008 ± 0.0002b	0.05 ± 0.005b

^aAll assays were performed at 37 °C (pH 8.3) in the presence of 12 mM MgCl₂. Data were fit to eq 1 to afford apparent kinetic parameters.

^bGlockzin et al., 2022.⁴³

^cWenck et al., 2004.³⁰

Our combined results show that the T. cruzi genome encodes two pairs of HGPRT/HGXPRT isoforms (TcA/TcB and TcC/TcD), with different substrate preference and distinct genetic origin: non-Esmeraldo and Esmeraldo, respectively. The main substrate for the TcHGXPRTs is Xan, showing a remarkably lower catalytic efficiency for the other two 6-oxopurine substrates when compared to that of the TcHGPRTs. The role of the substitution of a Lys residue (TcB) for a Gln (TcD) at position 125 (Figure S3) and the concomitant apparent impact on the affinity for substrate Hx is currently under investigation using site-directed mutagenesis studies. As both isoforms show strong conservation of their catalytic activity, they are herein collectively referred to as TcHGXPRT.

Hx is the purine base with the highest available concentration in human serum (8 μM), followed by Xan (1.4 μM), whereas Xan is the most abundant purine in human cerebrospinal fluid (9 μM), followed by Hx (3.4 μM).²⁸ It is interesting to speculate if the expression of the pairs of TcHGPRT/TcHGXPRT is related to the different stages of CD, where TcHGPRT would be preferentially expressed during acute stages when T. cruzi is circulating in the bloodstream, and TcHGXPRT activity would become more relevant as the acute disease progresses into encephalitis⁵⁹ and in chronic stages, when the parasite migrates to and persists in several organs, including the brain.⁶⁰ Studies of the differential expression of HGPRT and HGXPRT isoforms in T. brucei brucei demonstrated that the expression of the single HGXPRT (TbbHGXPRT) and one HGPRT isoform (designated TbbHGPRT-1) in both the bloodstream and procyclic forms of the parasite, while the second HGPRT (TbbHGPRT-2) isoform is only expressed in the procyclic stage.³² The same study also indicates that TbbHGPRT-1 and TbbHGXPRT do not show the stark difference in catalytic efficiency or strong substrate preference between the isoforms³² as observed for the T. cruzi enzymes. Further studies of differential gene expression during varied life stages of the parasite and during the progress of CD could help clarify whether the activity of TcHGPRT or TcHGXPRT is predominant under different metabolic conditions or stages of CD. Another distinction between HGPRTs and HGXPRTs is their subcellular localization. TbbHGXPRT contains the C-terminal peroxisomal targeting signal type 1 (PTS1) tripeptide serine–lysine–leucine (SKL), and direct immunofluorescence experiments showed its presence solely in the glycosome, while TbbHGPRT-1 is cytosolic.³² The primary sequence of the TcHGXPRTs includes the PTS1 tripeptide alanine–histidine–leucine (AHL), a variant of the main motif SKL,³² whereas TcHGPRTs do not (Figure S3). Therefore, it is likely that there is conservation in the subcellular localization of HGPRTs and HGXPRTs among the two trypanosome species.

TcHGXPRT Kinetic Mechanism

Since both TcHGXPRT isoforms (TcB and TcD) presented nearly identical kinetic parameters, the TcB isoform was further used to determine the kinetic mechanism and rate-limiting steps of TcHGXPRT catalysis. The kinetic mechanism of catalysis was determined by initial rate, dead-end inhibition, and product inhibition studies. Initial rate data for Gua vs. PRPP and PRPP vs. Gua are in accordance with the apparent kinetics results (Table 1); the catalytic turnover (k_cat) and specificity constants (k_cat/K_m) for TcHGXPRT are significantly lower than the values determined for TcHGPRT for the GPRT activity (Table 2). TcHGXPRT shows a remarkable 30- to 50-fold reduction in K_m for PRPP (K_PRPP) and a 2-fold increase in K_m for Gua (K_Gua). Initial rate data, when plotted in double-reciprocal form, conformed to an intersecting pattern to the left of the ordinate (Figure S5), suggesting that TcHGXPRT follows a sequential mechanism, as observed for TcHGPRT and other Type I PRTases.^{30,36,61−63}

Table 2. TcHGXPRT and TcHGPRT Initial Rate Data (GMP Biosynthesis) and TcHGXPRT Dead-End and Product Inhibition Data^a,^b.

Enzyme	Variable substrate	Kinetic parametersc
	(μM)	KPRPP(μM)	Ki,PRPP(μM)	kcat/KPRPP(× 103 M–1 s–1)	KGua(μM)	kcat/KGua(× 103 M–1 s–1)	kcat(s–1)
TcHGXPRT	PRPP (1–78)	1.1 ± 0.2	0.6 ± 0.3	40 ± 7	28 ± 3	16 ± 2	0.44 ± 0.02
	Guanine (10–300)	0.6 ± 0.1	0.7 ± 0.2	77 ± 13	30 ± 3	15 ± 2	0.46 ± 0.01
TcHGPRT	PRPP (20–215)	32 ± 5d	NDe	1010d	12 ± 1d	2650d	32d
Enzyme	Variable Substrate	Inhibitor	Fixed substrate	Inhibition pattern	Apparent kinetic parametersc
	μM	μM			Km(μM)	Kis(μM)	Kii(μM)
TcHGXPRT	Xan (3–100)	9-deazaxanthine (0–250)	PRPP (1 mM)	Ce	2.4 ± 0.3	70 ± 10	--
	PRPP (23–750)	9-deazaxanthine (0–600)	Xan (120 μM)	UCe	130 ± 10	--	490 ± 40
	Xan (2.5–150)	PPi (0–200)	PRPP (1 mM)	NCe	9 ± 1	190 ± 40	1400 ± 400
	PRPP (11–750)	XMP (0–60)	Xan (120 μM)	Ce	100 ± 10	33 ± 5	--

^aInitial rate data were fit to eq 2 to afford kinetic parameters.

^bInitial rate data for 9-deazaxanthine vs. Xan and XMP vs. PRPP fit best to eq 3, initial rate data for 9-deazaxanthine vs. PRPP fit best to eq 5, and initial rate data for PPi vs. Xan fit best to eq 4, which afforded the apparent kinetic parameters.

^cAll assays were performed at 37 °C (pH 8.3), 12 mM MgCl₂.

^dWenck et al., 2004.³⁰

^eND = not determined, C = competitive inhibition, UC = uncompetitive inhibition, NC = noncompetitive inhibition

Further kinetic characterization using the Xan substrate mimic 9-deazaxanthine, which is a dead-end inhibitor, was performed to determine whether TcHGXPRT substrate binding is random or ordered. Dead-end inhibitors are substrate analogues that compete for its binding site but do not undergo chemical transformation. Double-reciprocal plots of initial rate data for 9-deazaxanthine vs. Xan show an intersecting pattern on the y-axis, an indication of competitive inhibition (Table 2, Figure S6). These data suggest that Xan and 9-deazaxanthine bind to the same enzyme form during catalysis, which is unsurprising considering the similarities in structure (Figure S6). Due to the structural similarities among Hx, Gua, and Xan, it is expected that all 6-oxopurine substrates follow the same binding order and mechanism in the reaction(s) catalyzed by TcHGXPRT. Double-reciprocal plots of initial rate data for 9-deazaxanthine vs. PRPP provided a parallel pattern, which is indicative of uncompetitive inhibition (Table 2, Figure S6), revealing that 9-deazaxanthine binds to TcHGXPRT after PRPP binding (E·PRPP).

The order of product release was investigated by product inhibition studies. The inhibition pattern of PPi vs. Xan was determined by measuring the initial rate when PPi was fixed at five concentrations (0, 25, 50, 100, and 200 μM), Xan was varied (2.5–150 μM), and PRPP was fixed at a saturating concentration (1 mM). Double-reciprocal plots of the initial rate data provided an intersecting pattern to the left of the ordinate, indicative of noncompetitive inhibition (Figure S7). These data suggested that Xan cannot overcome PPi inhibition, thus ruling out a rapid equilibrium random kinetic mechanism.⁶⁴ The inhibition pattern for XMP vs. PRPP was determined by measuring the initial rate when XMP was fixed at five concentrations (0, 8, 15, 30, and 60 μM), PRPP was varied (11–750 μM), and Xan was fixed at a saturating concentration (120 μM). Initial rate data were for XMP vs. PRPP fit best to eq 3, and the double-reciprocal plot provided an intersecting pattern on the y-axis (Figure S7), which is indicative of competitive inhibition. These results suggest that PRPP and XMP bind to the same enzyme form, and since PRPP binds to free enzyme, XMP must be the last product released.

Initial rate data for Gua vs. PRPP and PRPP vs. Gua fit best to the kinetic model describing a steady-state ordered or a rapid-equilibrium random kinetic mechanism (eq 2). Dead-end inhibition suggested that PRPP binds to free TcHGXPRT, followed by the 6-oxopurine. Product inhibition studies indicated that PRPP and XMP bind to the same enzyme form; therefore, XMP is the last substrate to leave the enzyme–substrate complex. Additionally, the noncompetitive pattern observed for PPi vs. Xan indicated that TcHGXPRT must conform to a steady-state ordered kinetic mechanism,⁶⁴ in which PRPP binds first, followed by the 6-oxopurine, and after catalysis, PPi is released, followed by the respective NMP (Figure 2). TcHGPRT also follows a steady-state ordered kinetic mechanism.^30,43 Indeed, most enzymes in the Type I PRTases enzyme family follow a highly conserved ordered kinetic mechanism,⁵⁶ where the phosphoribosyl donor (PRPP) binds first followed the secondary substrate.

Figure 2.

TcHGXPRT conforms to an ordered kinetic mechanism, where PRPP binds to the free enzyme form (E), followed by binding of the 6-oxopurine substrate (Hx, Xan, or Gua). After catalysis, PPi is the first product to be released, followed by the respective NMP (IMP, XMP, or GMP).

Viscosity Studies and the Determination of Rate-Limiting Step(s) of TcHGXPRT Catalysis

We evaluated the effects of added viscosogens on the initial rate of TcHGXPRT to ascertain the rate-limiting step(s) of catalysis. Initial rate data were collected at a saturating concentration of Hx (120 μM) and variable concentrations of PRPP (0–300 μM) or at a saturating concentration of PRPP (1 mM) and variable concentrations of Hx (4.7–150 μM), with fixed concentrations of viscosogen (glycerol: 0, 4, 8, and 12% v/v or PEG₁₀₀₀₀: 0, 0.5, 1, and 2% v/v). Microviscosogens, such as glycerol, affect the diffusional steps of catalysis, whereas macroviscosogens such as PEG₁₀₀₀₀ serve as a control to ensure observed effects are due to diffusion and not a crowding effect.⁴⁵ The kinetic parameters k_cat/K_m and k_cat are plotted vs. η_rel. The values of k_cat/K_PRPP, k_cat/K_Hx, and k_cat did not significantly change in the presence of macroviscosogen (PEG₁₀₀₀₀) (Figure 3). The k_cat/K_PRPP values and k_cat/K_Hx values decreased proportionally in the presence of increasing glycerol and provided slopes of 0.9 ± 0.1 and 0.9 ± 0.1, respectively (Figure 3), when plotted to eq 10. These results suggest that product formation is limited by the diffusion of substrate(s) to the active site;⁴⁵ however, the magnitude of k_cat/K_m values (Table 1) might imply otherwise. These data suggest a conformational change necessary for substrate binding is contributing to the observed effect, as previously suggested for the TcHGPRT⁶⁵ and HsHGPRT⁶⁶ enzymes. The k_cat values also decreased proportionally in the presence of increasing glycerol concentrations and provided slopes of 1.0 ± 0.1, regardless of which substrate was varied (Figure 3). A slope of 1 on the k_cat effect suggests that the overall turnover for TcHGXPRT catalysis is also limited by a postchemistry step such as the release of products from the active site or by a conformational change that precedes it and is necessary for product release.

Figure 3.

TcHGXPRT viscosity effects on the values of (A) k_cat/K_PRPP and (B) k_cat when PRPP is variable, and the values of (C) k_cat/K_Hx and (D) k_cat when Hx is variable in the presence of glycerol (blue circles). A reference dashed line (black) with a slope of 1 is included in each plot. The kinetic parameters were not significantly different at increased levels of PEG₁₀₀₀₀ (red triangles).

Type I PRTases have a characteristic flexible loop that closes over the active site upon binding of both substrates.⁵⁸ This flexible loop (loop 2) has been implicated in shielding the reaction TS during chemistry and in preventing nonproductive PRPP hydrolysis. Previously, we determined that the opening of loop 2 (after production of the E·PPi·NMP complex) is rate limiting to TcHGPRT catalysis, which could be concomitant with release of the first product, PPi.⁴³ Conversely, data for TcHGXPRT suggested that catalysis is also limited by binding of substrates and release of products from the active site, potentially due to conformational changes such as the opening and closing of loop 2. One might speculate that the substantial effects on the kinetic parameters k_cat and k_cat/K_m at relatively low η_rel values are due to a significant conformational change, including the movement of large protein loops and not desorption of a small molecule. However, further experimental investigation is necessary to determine if these results truly include a diffusional effect or if conformational changes of TcHGXPRT during catalysis (the opening or closing of the active site loop 2) are the overall rate-limiting step(s), in which binding of substrates and release of products are intrinsically correlated to this isomerization step.

TcHGXPRT Three-Dimensional Structure

TcHGPRT (TcA isoform) has been previously characterized by crystallographic studies to identify key residues involved in catalysis.^49,67 However, no structure of TcHGXPRT, to the best of our knowledge, has been determined. We, therefore, sought to solve the X-ray crystal structure of TcHGXPRT to reveal possible structural implications on catalysis and substrate preference. The structure of the apoenzyme could not be determined, as multiple crystallization attempts failed to produce any crystals, possibly due to the high flexibility of the four loops that compose the Type I PRTases’ active site.^56,58 The structure of TcHGXPRT (TcD isoform) was determined by cocrystallizing it with its preferred substrate Xan and product PPi (E·Xan·PPi), a dead-end complex that has been previously predicted as a possible species in the TcHGPRT kinetic mechanism.³⁰ The crystals soaked with inhibitors (S)-2, (R)-2, 6, 8, and 9 resulted in inhibitor-bound structures as a result of displacement of the ligands in the E·Xan·PPi complex by the inhibitors. All such structures show high structural conservation (RMSD values ranging from 0.15 to 0.24 Å), as expected for isomorphous crystals. In the crystallographic structures determined for the E·Xan·PPi complex and for TcHGXPRT bound to inhibitor 6, phosphate (Pi) is observed in the active site and not the PPi that was added in the preincubation mixture.

Notwithstanding the low sequence identity with TcHGPRT (Figure S3), the secondary and three-dimensional structures of TcHGXPRT and TcHGPRT are highly conserved, with an RMSD of 1.25 Å for 177 α-carbon atoms (Figure S8). The strong structural conservation despite significant variability in amino acids sequence is a common feature in the Type I PRTase family.^56,58,67 All HG(X)PRTs characterized to date comprise monomers containing two domains—a structurally conserved central core domain and a structurally variable hood domain composed by residues from the N- and C-termini, with the active site located in the cleft between them.^56,58,67 The core domain is responsible for binding the phosphoribosyl donor substrate PRPP, and the hood domain binds the variable 6-oxopurine substrate(s).⁴⁸ The TcHGXPRT core domain contains a central β-sheet with five parallel strands and one antiparallel strand flanked by three alpha helices (Figure 4A, cyan). The hood domain (Figure 4A, magenta) is oriented above the core domain and consists of an extended loop (loop 4, Figure 4B, orange) and an antiparallel β-sheet. TcHGXPRT active site is composed of four highly flexible loops⁵⁸ (Figure 4B). Loop 1 (residues 85 to 88, red), loop 2 (residues 116 to 122, blue), and loop 3 (residues 145 to 150, green) are part of the core domain, while loop 4 (residues 181 to 215, orange) is part of the hood domain. Poor electron density for loop 2 was observed only in one subunit of TcHGXPRT in complex with inhibitor 9 (discussed below). In this complex, the active site loop is present in an open conformation (Figure 4B).

Figure 4.

(A) Quaternary structure of TcHGXPRT (depicted as Richardson diagram) bound to Xan (white sticks) and Pi (orange sticks). Subunit A is colored in gray. The hood domain (magenta), core domain (cyan), and anchor domain (purple) are indicated on the secondary structure of subunit B. (B) Active site loops, colored in red (loop 1), blue (loop 2—open conformation as observed in the TcHGXPRT·9 complex, 8FX3.pdb), green (loop 3), and orange (loop 4).

TcHGXPRT crystallized as a homodimer, which corroborates the quaternary structure determined by SEC (Figure S9). PDBePISA calculates a buried surface area for the dimer of 5000 Ų, indicative of a stable dimer,⁶⁸ unlike TcHGPRT,^49,67 which has a dimerization interface with a buried surface area of 4480 Ų. TcHGXPRT contains an additional β-hairpin domain composed of residues 58 to 72 located on the opposite side of the core domain in relation to the hood domain (Figure 4A, purple). This additional β-hairpin, which is also present in TbbHGXPRT, interacts with the C-terminal portion of the adjacent subunit, and is hypothesized to stabilize the dimeric quaternary structure.⁶⁹ We therefore named this β-hairpin the anchor domain. In addition, the carbonyl oxygen of the C-terminal residue of TcHGXPRT (Leu231) projects into the dimer interface, forming hydrogen bonds with the backbone amines of either Lys133 (2.3 Å) or Ala134 (2.3 Å) on the opposing subunit (Figure S10).

There is a single amino acid substitution between the two TcHGXPRT isoforms (residue 125), which is located on strand β9. In solution, this residue would be completely exposed to the solvent in the structures determined by us. This side chain (a Gln in TcD) is oriented away from the dimer interface and active sites, and its main chain forms two hydrogen bonds with the main chain of Ala113 located on β8, stabilizing the secondary structure (Figure S11). In the crystal, the oxygen of the Gln side chain forms apparent hydrogen bonds with symmetry mates and, depending on the monomer, participates in one of two different crystal packing interactions. No crystals were obtained for the TcB isoform, most likely because the Gln (TcD) to Lys (TcB) variation would prevent the formation of the stabilizing hydrogen bonds at the crystallographic interface and insert a highly entropic and charged residue at this key location. This side chain substitution is not expected to affect the structure or catalytic mechanisms of the isoforms; therefore, the difference in affinity observed for substrate Hx (Table 1) cannot be simply explained by this variation in their primary sequence.

TcHGXPRT Active Site Structure—Loops 1 and 3, the Phosphate Binding Loops

Loop 1 (Figure 4B, red) is proposed to play a role in the coordination of the two subunits^70,71 and contains a nonproline cis-peptide bond between Leu84 and Lys85. This unusual bond is conserved in TcHGPRT,^49,67,71PfHGXPRT,³⁶ and HsHGPRT⁷² and is directly implicated in cooperativity between the subunits, as the Lys side chain extends away from the active site and forms an interaction with the adjacent subunit in the dimer interface. In agreement with these observations, we previously reported inhibitors which displayed cooperative binding to TcHGPRT.⁴³ In TcHGXPRT, Lys85 forms a 2.1 Å hydrogen bond with the backbone carbonyl oxygen of Val107 and a 2.3 Å hydrogen bond with oxygen delta 1 of Asp108 located on strand β3 of the adjoining subunit (Figure S12).

The conserved PRPP binding motif (140-VLILEDIVDSGK-151 – Figure S3) includes residues on loop 3 (Figure 4B, green), where the invariant residue Asp148 (indicated in bold on the motif sequence) has been identified as a catalytic general acid/base for HsHGPRT⁶⁶ and PfHGXPRT.⁵⁸ The corresponding residue in TcHGPRT (Asp115) was proposed to also act as a strong hydrogen bond acceptor.⁷³ While an Asp residue in this position is strictly conserved in all HG(X)PRTs,⁷⁴TcHGPRT mutagenesis studies suggested that any amino acid capable of forming a strong hydrogen bond to the purine substrate N7 is sufficient for the stabilization of the reaction TS.⁷³ The PRPP binding motif in TcHGXPRT includes the conserved negatively charged residues Glu144 and Asp145, shown to interact with divalent metal ions in the active site,^67,69 aiding the proper positioning of PRPP for catalysis. The C-terminal residues on the PRPP binding motif (positions 149 to 151) present more variability among the Type I PRTases but, in all cases, form extensive hydrogen bonds to the 5′-phosphate group of substrate PRPP and product PPi.

Both loops 1 and 3 participate in the binding of the Pi and PPi groups of substrate PRPP and are conserved across the Type I PRTases. The Xan-bound structure of TcHGXPRT shows a Pi molecule in the active site at an analogous position to that occupied by the 5′-phosphate group of PRPP and NMPs in homologous HG(X)PRT structures (Figure 5). The Pi is held in place by a network of hydrogen bonds to residues Asp148, Ser149, Gly150, and Thr152, part of the PRPP binding site on loop 3. The structure of TcHGXPRT bound to inhibitor 6 reveals a second Pi binding site (Figure 5). The second Pi interacts with amino acid residues from loop 1 (forming hydrogen bonds to residues Lys85 and Gly86 and additional hydrophobic contacts with Leu84) and loop 4 (a hydrogen bond to Arg211 and hydrophobic contact with Glu205). In the structures of PfHGXPRT bound to TSAIs,^34,40,75 an additional PPi molecule is also bound to the active site through the same interactions with residues from loop 1 and loop 3 as observed in the TcHGXPRT structure, with the addition of interactions with residues from loop 2 and a Mg²⁺ atom (Figure S13). The structure of TcHGXPRT·6·Pi suggests that the presence of a PPi molecule may also play a role in T. cruzi HGXPRT inhibition mediated by these TSAIs (see discussion below).

Figure 5.

Structure of TcHGXPRT (gray) bound to Xan (white sticks) and Pi (orange sticks). The Pi molecule forms hydrogen bonds (yellow dashed lines) with residues from loop 3 (green): Asp148, Ser149, Gly150, Lys151, and Thr152 (5′ phosphate binding site). Inhibitor 6 (green sticks) and the extra Pi (orange sticks) are also shown, highlighting the conservation of binding its purine and 5′-phosphate moieties and presenting the PPi binding site. The second Pi molecule binding site comprises amino acid residues from loop 1 (Lys85 and Gly86, red) and loop 4 (Arg211, orange). The electron density for 6 and the extra Pi molecule are Polder maps contoured at 3σ. See also Figure S13A,B.

TcHGXPRT Active Site Structure—Loop 2 and Rate-Limiting Step(s)

The active site loop 2 is commonly referred to as the catalytic or flexible loop (Figure 6).^67,76 Indeed, the flexibility of loop 2 results in broken or missing electron density in many Type I PRTases crystal structures.⁵⁸ Previously determined structures of TcHGPRT showing the complete active site (1TC2.pdb⁴⁹ and 1P18.pdb⁷³) revealed that loop 2 undergoes a significant conformational change (∼17.5 Å movement) upon binding of both substrates, changing from an open, solvent-exposed conformation, into a closed solvent-protected conformation (Figure S14). Structural studies of TcHGPRT showed that upon closure of loop 2, both substrates became properly positioned for catalysis through the formation of novel interactions between loop 2 and PRPP and the induced repositioning of the purine base. This allows the formation of a hydrogen bond with Asp148 located in loop 3 and the N7 atom on the purine, which is proposed to mediate the protonation of the base at the N7 position, lowering the energy barrier and allowing the chemical step of the reaction to take place.^49,58,77 The significance of this specific interaction and importance for formation of the reaction’s TS is well-described by the kinetic studies and structure of PfHGXPRT bound to its transition-state inhibitor ImmGP (compound 9 – see Figure 10 for structure).³⁹ Despite its role in catalysis, the loop 2 primary sequence is not well-conserved between orthologous enzymes, apart from the invariant Ser114–Tyr115 dipeptide located on its N-terminal portion.⁷⁷ Once loop 2 closes, these two residues form apparent hydrogen bonds with PRPP, positioning it for nucleophilic attack of the purine base (N9) on the C1′ atom of PRPP.⁷⁷ These interactions highlight the importance of this loop on positioning of substrates, catalysis, and the possible implications on PPi release.⁷⁷ Opening of loop 2 after product formation moves the dipeptide Ser114–Tyr115, now bound to PPi, away from the active site, a motion that may be intrinsic to the release of the first product PPi, as it is pulled from the active site during the enzyme conformational change.^49,65,67,77

Figure 6.

Structure of TcHGXPRT (gray, 8FX3.pdb) bound to 9 (white sticks). Active site loops are colored as in Figure 4B. Loop 2 (blue) is shown in its open conformation. The structure of TcHGPRT (light cyan, 1TC2.pdb⁴⁹) in complex with PRPP, allopurinol (yellow sticks), and Mg²⁺ (green sphere) is overlaid, showing loop 2 in its closed conformation. See also Figure S14.

Figure 10.

Binding of 6 to TcHGXPRT, in the presence of a Pi molecule that interacts with loop 1 and 4 residues. This complex is hypothesized to mimic the E·I·PPi dead-end complexes described for PfHGXPRT. Interactions are represented as in Figure 9.

All structures of TcHGXPRT reported here show an incomplete loop 2, except for one subunit of TcHGXPRT in complex with inhibitor 9, in which poor density is observed for an open conformation (Figures 4B and 6). In this structure, the catalytic Asp148 residue is already positioned within a hydrogen bond distance (3.0 Å, see below) to the N7 position of the purine base, suggesting that closing of active site loop 2 may not be necessary to orient the catalytic Asp residue. This is consistent with the aforementioned finding that Asp148 may not serve as catalytic acid/base but may only be required for orientation of the substrate.⁷³ We previously demonstrated that opening of loop 2 after the chemical step, and releasing of first product PPi, are rate limiting for TcHGPRT.⁴³ Viscosity studies of TcHGXPRT (Figure 3) indicated, however, that there is a diffusional component of substrate binding and product release from the active site to the rate-limiting steps(s). Considering the insights provided by structural studies, and the role of loop 2 on catalysis, these conformational changes (active site opening and closing) may play an important part, especially postchemistry. Further experiments are necessary to independently measure the rates of substrate binding and product release as well as loop 2 conformational changes to completely elucidate the TcHGXPRT catalytic mechanism.

TcHGXPRT Active Site Structure—Loop 4 and Implications for Purine Substrate Selectivity

Residues 181 to 215 from the hood domain form the long active site loop 4 (Figure 4B, orange). This loop is responsible for binding the 6-oxopurine bases,⁵⁸ and comparison of the amino acid composition of loop 4 from TcHGXPRT and TcHGPRT⁶⁷ provides clues on the mechanism of substrate selectivity of these enzymes. Three active site amino acid substitutions are striking: TcHGXPRT Tyr198 is replaced by a Phe, Glu205 is replaced by an Asp, and Phe204 is replaced by a Leu residue (Figure 7, magenta). The structure of TcHGXPRT bound to Xan and Pi shows the 6-oxopurine base is locked in place by a network of 10 hydrogen bonds to loop 4 residues Glu205, Tyr198, Val199, and Lys177, a conserved water molecule, and loop 3 residue Asp148. Residue Tyr198 also forms a conserved π–π stacking with the aromatic ring of the bound purine substrate (or NMP product, after catalysis).^56,58 The C2-oxo group of Xan, the preferred substrate of TcHGXPRT, distinguishes it from the other two 6-oxopurine substrates, Hx and Gua (Figure 1). The C2-oxo group is accommodated in the active site by five hydrogen bonds to Tyr198, Glu205, the conserved water molecule, and Val199. The hydroxyl group on Tyr198 also makes a hydrogen bond to the main chain carbonyl of Glu205, avoiding repulsion between the dipole of this carbonyl group and the C2-oxo group of Xan. At this position, the N7 of Xan is already within hydrogen bonding distance from the proposed catalytic residue Asp148 (3.0 Å – Figure 7), even when the active site is in its open conformation. This observation agrees with the viscosity data indicating diffusion of substrate(s) into the active site may contribute to the rate-limiting step(s), since in this case the rearrangement of the purine substrate and active site residue Asp148 upon closure of loop 2 is not necessary to position the 6-oxopurine for catalysis. Closing of loop 2 is still necessary, however, for proper positioning of substrate PRPP through the formation of hydrogen bonds to Ser114–Tyr115 (see above), an observation that also agrees with the viscosity data, which includes product release and a postchemistry conformational change as components of the reaction rate-limiting step(s).

Figure 7.

TcHGXPRT (represented as in Figure 4B, 8FWY.pdb) bound to Xan (white sticks) and Pi (orange sticks). The hydrogen bonds formed with 6-oxopurine are shown as yellow dashed lines. The hydrogen bond between Tyr198 and Glu205 is represented as a pink dashed line. Residues from active site loop 4 are represented with their carbon atoms in orange, and Asp148, the proposed reaction catalytic residue, is represented with its carbon atoms in green. The amino acid substitutions observed in TcHGPRT (Tyr198Phe, Glu205Asp, and Phe204Leu) are shown in magenta. WAT indicates the conserved active site water molecule (blue sphere).

The presence of a Phe residue at position 198 in TcHGPRT does not allow for the formation of hydrogen bonds with the equivalent residue at position 205 (an Asp). The structure of the closely related TbbHGPRT-1 bound to XMP (6MXG.pdb⁶⁹) shows that the 6-oxopurine portion of XMP adopts a different position in the HGPRT active site, displaced 2 Å toward the Asp at position 205, with rotation of Phe at position 198, and an increased distance to the catalytic residue Asp148 (Asp115 in TcHGPRT, 4.3 Å distance). These differences explain the less preferable binding of Xan to TcHGPRT and lower catalytic efficiency when compared with TcHGXPRT (Table 1).

Comparison of TcHGXPRT bound to Xan and the structure of TbbHGXPRT bound to XMP (6MXB.pdb⁶⁹) shows the conservation of positioning of the 6-oxopurine within the active site as well as of the residues that interact with the reaction product and overall three-dimensional structure (0.7 Å RMSD) (Figure S15A,B). The structure from T. brucei brucei also lacks the active site loop 2 and is equivalent to the open active site, postchemistry, and post PPi release enzymatic species in the reaction mechanism shown in Figure 2. TbbHGXPRT structures have also been determined in complex to GMP (6MXC.pdb⁶⁹) and IMP (6MXD.pdb⁶⁹), revealing the interaction of the C2-amino group of GMP with the carbonyl oxygen of Glu205 and the conserved water, forming an apparent hydrogen bond with the C2-amino group. This is in contrast with its role in hydrogen bonding with the C2-oxo atoms of Xan and XMP (Figure S15). The authors proposed the role of the conserved water molecule as an adapter to facilitate binding of both XMP and GMP to HGXPRT.⁶⁹ This water plays no role in the binding of Hx, as its C2-carbon cannot form hydrogen bonds but instead makes hydrophobic interactions with Phe204 (a Leu in TcHGPRT, the last of the three substitutions observed between the T. cruzi isoforms). Binding of XMP, GMP, and IMP to the active sites of TbbHGXPRT is highly conserved. Considering the conservation between the TbbHGXPRT and TcHGXPRT active sites (Figure S15), the implications of substrate selectivity derived from T. brucei brucei structures can be, most certainly, extrapolated to the T. cruzi isozymes. Moreover, the structure of TcHGXPRT bound to Xan corroborates our hypothesis of a distinct catalytic mechanism, where the diffusion of substrates into the active site also contributes to the reaction rate-limiting step(s) in catalysis.

Repurposing Transition-State Analogue Inhibitors (TSAIs) of PfHGXPRT against TcHG(X)PRTs

Despite the low degree of sequence similarity between TcHGXPRTs and the orthologous PfHGXPRT, the overall structures of their subunits are extensively conserved (with an overall RMSD of 1.55 Å), including the structure of their active sites (Figure S16A,B). PfHGXPRT has a homotetrameric quaternary structure, whereas TcHGXPRT is a homodimer. The plasmodial enzyme also presents the core and hood domain architecture characteristic of the Type I PRTases;⁵⁶ however the anchor domain is absent, suggesting that this may be unique to the trypanosomal HGXPRTs. A distinction worth noting is the structure of PfHGXPRT loop 2—the highly disordered residues in T. cruzi assume an antiparallel β-hairpin conformation in P. falciparum (for example, see Figure S16B), and the densities of these residues are well-defined in the available high resolution crystallographic structures (≤2 Å),^34,40,75 with B-factors consonant with a structurally more rigid segment of the protein. The more defined organization of the secondary structure of these residues and PfHGXPRT-determined intrinsic kinetic isotope effects³⁵ suggest that opening and closing of its active site does not play a relevant role in its kinetic mechanism as it does in T. cruzi. The structures of PfHGXPRT bound to an acyclic immucillin inhibitor⁴⁰ reveal the conservation in TcHGXPRT of all amino acid residues that interact with the ligand (Figure S16C,D), except for the residues implicated in substrate selectivity (TcHGXPRT Tyr198, Phe204, and Glu205) that in P. falciparum are, respectively, Phe197, Leu203, and Asp204—the same residues observed in TcHGPRT. It is worth noting that although the plasmodial enzyme is regularly referred to as a HGXPRT, its kinetic characterization has shown that Xan is a very poor substrate, with a specificity constant of k_cat/K_m = 0.005 μM^–1 s^–1 which is 150-fold and 300-fold lower when compared to Hx (k_cat/K_m = 0.75 μM^–1 s^–1) and Gua respectively (k_cat/K_m = 1.5 μM^–1 s^–1) and apparent K_m values in the 189–420 μM range³⁷ under assay conditions almost identical to the methods used in this study. These results indicate that PfHGXPRT should be more correctly classified as an HGPRT, and its kinetic parameters agree with the proposed role of the triad of residues discussed above on the ability of using Hx, Gua, and Xan as substrate or Hx and Gua, respectively (where trypanosomal HGXPRTs contain Tyr/Phe/Glu and trypanosomal HGPRTs and PfHGXPRT contain a Phe/Leu/Asp).

PfHGXPRT follows an S_N1-like catalytic mechanism.³⁵ We have previously demonstrated the conservation of this reaction mechanism in TcHGPRT, along with the feasibility of repurposing high affinity TSAIs developed against the plasmodial enzyme as potent inhibitors of TcHGPRT.⁴³ These inhibitors contain a secondary amine that upon protonation structurally mimics the putative ribooxacarbenium ion predicted in the TS of the reaction catalyzed by PfHGXPRT (Figure 8, center) as well as the P. falciparum and human purine nucleoside phosphorylases.^39−42,78 Considering the active site conservation between TcHGXPRT and PfHGXPRT as well as the experimental evidence of the importance of inhibiting, concomitantly, both the HGPRT and HGXPRT isozymes in Trypanosoma,³² we reevaluated this set of compounds (Figure 8) as potential inhibitors of TcHGXPRT.

Figure 8.

PfHGXPRT S_N1-like TS structure³⁵ (center) and TSAIs known as Immucillins (8, 9, and 10) and their derivatives, tested as potential inhibitors of TcHGXPRT.

The apparent potencies (IC₅₀) of each compound were determined with concentration–response curves (Figure S17) at saturating concentration of substrates (PRPP 1 mM, Hx 120 μM), and data were fit to eq 11. Inhibition constants (K_i) were determined using eq 12 (Tables 3, S2). Notably, the addition of PPi to the reaction mixture (100 μM) potentiates inhibition of these TSAIs against TcHGXPRT (Figure S18), providing evidence for the formation of E·I·PPi dead-end complexes, as described for PfHGXPRT^40,75 and as suggested by the TcHGXPRT·6 complex structure (Figure 5). All of the tested inhibitors displayed positive cooperativity against TcHGXPRT (Figure S17). The importance of interactions formed between residues of loop 1 and of the adjacent subunit as the structural basis for cooperativity have been previously established and demonstrated.^{36,49,67,71,72} However, the observed positive cooperativity may be due to the effect of the bound PPi in enhancing the potency of the TSAIs by the formation of dead-end complexes.

Table 3. Immucillin Inhibitors of TcHGXPRT, TcHGPRT, and PfHGXPRT^a.

Inhibitor	TcHGXPRT (app Ki μM)	TcHGPRTb(app Ki μM)	PfHGXPRTb(app Ki μM)
(S)-1	0.6 ± 0.1	0.006 ± 0.002	0.0005 ± 0.00005c
(S)-2	1.6 ± 0.1	0.004 ± 0.002	0.0006 ± 0.00004
(R)-2	38 ± 4	0.5 ± 0.03	0.023 ± 0.002
(S)-3	5.4 ± 0.3	0.017 ± 0.002	0.0012 ± 0.0001
(S)-4	0.7 ± 0.1	1.4 ± 0.03	0.11 ± 0.03
(S)-5	0.9 ± 0.1	13.4 ± 0.9	8 ± 0.5
(±)-6	11 ± 1	0.17 ± 0.02	0.009 ± 0.0001
(±)-7	35 ± 3	1.1 ± 0.2	0.015 ± 0.004c
ImmHP (8)	16 ± 1	0.17 ± 0.03	0.001 ± 0.0003
ImmGP (9)	36 ± 5	1.1 ± 0.4	0.014 ± 0.002
ImmXP (10)	0.20 ± 0.02	36.7 ± 0.7	>100
11	8.4 ± 0.3	0.7 ± 0.2	0.06 ± 0.003
12	8 ± 1	3.6 ± 0.3	0.11 ± 0.006

^aTcHGXPRT assays were performed at 37 °C (pH 8.3) in the presence of 12 mM MgCl₂ and 100 μM PPi. Concentration–response data was fit to eq 11, and K_i values were calculated using eq 12.

^bGlockzin et al., 2022.⁴³

^cGai et al., 2022.⁷⁹

The mode of inhibition against TcHGXPRT was determined using (S)-2 as a representative inhibitor for this set of compounds. Different fixed levels of (S)-2 (0.6–9 μM) at variable concentrations of Hx (15–150 μM) and a fixed saturating concentration of PRPP (1 mM) revealed a curvilinear decrease of IC₅₀ values at increasing concentrations of Hx, as expected for uncompetitive inhibition⁵⁴ (Figure S19). The initial rate data fit best to eq 5, and double-reciprocal plots provided an apparent parallel pattern (Figure S20), also indicative of uncompetitive inhibition. The intrinsic potencies (K_i) were determined by using the Cheng and Prusoff equation for uncompetitive inhibitors (eq 12) and are shown in Table 3.

The striking conservation among the most potent inhibitors of PfHGXPRT and TcHGPRT (although not in the same magnitude of potency) has been previously shown and discussed.^43,79 Both enzymes show high preference for Hx-based TSAIs (Hx being their preferred substrate), except for compound 1 (Gua-based), while Xan-based compounds showed lower inhibitory activity, as expected due to their active site composition at positions corresponding to residues 198, 204, and 205. In contrast, TcHGXPRT shows a remarkable preference for Xan as a substrate (Table 1), and such a property is reflected in the potent inhibition observed for compounds 5 (K_i = 0.9 μM) and 10 (K_i = 0.2 μM). Compound 10 (Immucillin-XP, or ImmXP) is a 9-deaza-xanthine analogue of a putative S_N1-like TS in XPRT catalysis, where the ribooxacarbenium ion is mimicked by a protonated amine (Figure 8). The two-atom replacement on XMP (Figure 1) to form 10 provided a 165-fold increase in binding affinity (Table 2K_is,XMP = 33 ± 5 μM; Table 3K_i = 0.20 ± 0.02 μM). Compound 5 is the 3′-OH acyclic version of 10, and this substitution resulted in a 4.5-fold decrease in potency (K_i = 0.9 ± 0.1 μM). This same substitution on the Gua-based and Hx-based immucillins resulted, conversely, in a significant improvement in potency (compounds (S)-1 and (S)-2, see below). Compounds 8 (ImmHP) and 9 (ImmGP) are TS mimics for HPRT and GPRT catalysis, respectively. Acyclic versions of compounds 8 and 9 (11 and 12, respectively) provided a 2- to 4-fold increase in potency (Table 3). The introduction of a hydroxyl at C5′ on acyclic 11, resulting in compound 6, does not significantly alter the potency. The addition of the 3′-OH in compounds (S)-2 and (S)-1 resulted, on the other hand, in 10-fold and 60-fold increases in potency, respectively, when compared to parent compounds 8 and 9. These data suggest that the 3′-OH group is important for inhibitor binding, as observed in the inhibition of TcHGPRT and PfHGXPRT (Table 3). Notably, phosphonate (S)-2 binds 22-fold more tightly than phosphate 7. All protozoan HG(X)PRTs presented an unconventional pattern where the phosphonate binds tighter than the native phosphate ester. Similarly, all protozoan HG(X)PRTs bind more tightly to the (S)-2 enantiomer over the (R)-2 enantiomer, with TcHGXPRT binding 24-fold tighter to (S)-2.

Structural Implications for TcHGXPRT Inhibition by Immucillins

TcHGXPRT is the orthologue most refractory to inhibition by the panel of TSAIs tested, with differences in K_i ranging from 10-fold to 100-fold when the activity of the same compound is compared with PfHGXPRT and TcHGPRT (Table 3), suggesting that, unlike TcHGPRT and PfHGXPRT, TcHGXPRT may adopt a different TS structure, as also implicated by viscosity data. However, it should be noted that the relative affinities for TcHGXPRT are within the order of the reduced values of k_cat/K_m for substrates Hx and Gua, which could reflect a change in the TS stability rather than the TS structure, suggesting that further kinetic analysis may be necessary to evaluate this possibility. Additionally, the underlying structural conformation and amino acid identities implicated in the strong preference for Xan (Tyr/Phe/Glu) may also play a more important role in ligand binding affinity, as demonstrated by the improved K_i values of the xanthine-based compounds (S)-5 and (S)-10 against TcHGXPRT, which show poor inhibition against TcHGPRT and PfHGXPRT.

The structures of TcHGXPRT bound to ImmHP (8) and ImmGP (9) revealed a well-conserved binding mode to the active site (Figure S21A), and all contacts with the purine and 5′-phosphate moieties are conserved for both inhibitors, but 8 forms additional hydrogen bonds between its 2′-OH and amino acid residues from loop 1 (Gly86) and loop 4 (Arg211), both mediated by water molecules (Figure 9). The structure of PfHGXPRT bound to 8(75) shows the conservation in the interactions of the purine and 5′-phosphate moieties, in addition to contacts formed between residues from loop 2 and the 5′-phosphate portion of 8 (Figure S22), as expected to also occur in TcHGXPRT upon closure of loop 2.

Figure 9.

Interactions formed in the active site of TcHGXPRT with (A) ImmGP (9) and (B) ImmHP (8). Hydrogen bonds are represented as green dashed lines, and hydrophobic contacts are represented as red half circles. Water molecules are represented as cyan spheres. The interactions that are unique to each inhibitor are highlighted in red on each interaction map.

Compound 6, the acyclic version of 8 with an additional hydroxyl group, has similar potency to 8 (K_i = 11 and 16 μM, respectively). The hydroxyl group was intended to occupy the binding site equivalent to the 3′-OH of 8; the crystal structure revealed that while the purine base and 5′-phosphate moieties of 6 do occupy the same portions of the active site and form the same interactions, the N4′ of the inhibitor is rotated to a position equivalent to the C2′ of 8 in the active site, and the hydroxyl is positioned where the C4′ of 8 is located (Figures 10 and S21B). Due to the rotation of C1′, N4′, C4′, C5′, and C6′, these atoms do not form the hydrogen network with the active site that would mimic the 3′-OH interactions formed by 8. However, in this complex, a Pi molecule is bound in loop 1 (Figure 5). This Pi forms interactions with loops 1 and 4 that mimic the contacts formed by the 3′-OH group of 8 (residues 84 to 86, Glu205 and Arg211 – Figure 10). In this structure, the density between the purine base moiety and the 5′-phosphate is not well-defined, indicating the possibility of multiple conformations or disorder, while the opposing monomer has a more well-defined density of 6 but poor density in the area of the additional Pi molecule. One interpretation could be that inhibitor 6 is binding to TcHGXPRT in two conformations, one with the 5′-phosphate bound as indicated in Figure 10 and one with its 5′-phosphate bound to loop 1. In fact, design strategies of acyclic nucleoside phosphonate as PfHGXPRT and TbbHG(X)PRTs inhibitors containing a second phosphate group intended to bind to loop 1 have been proposed.^80,81 However, it is well-described that the family of TSAIs tested in this study (Table 3) binds to PfHGXPRT in the presence of a PPi molecule bound to loops 1 and 4,^34,40,75 and our inhibition data revealed the potentiation of the inhibitory activity in the presence of PPi in the in vitro assays, with a specially pronounced effect on K_i values of 6 (Figure S18). These results combined with the structure of 6 bound to TcHGXPRT in the presence of the second Pi (a PPi analogue) suggest a similar mode of binding is expected to occur in T. cruzi.

The structure of TcHGXPRT bound to (S)-2 showed that its additional hydroxyl group overlays the 3′-OH of 8 when the bound active sites are compared (Figure S21C). Interestingly, the interactions formed between the 3′-OH of 8 with amino acids residues from loop 1 and loop 4 (hypothesized to be responsible for the gain in potency from 9 to 8) are not present in the binding of (S)-2. Its 2′-OH, instead, forms extensive contacts with water molecules and residues from loop 3 (Figure 11A). (S)-2 is approximately 10-fold more potent than its parent compound 8 (Table 3), and this gain in potency may be explained by the formation of hydrogen bonds between the purine ring and residues Tyr198 and Asp145 (Figure 11A)—these residues form only hydrophobic contacts with the other ligands here discussed. (R)-2 on the other hand, is less potent than its isomer (24-fold – Table 3). When compared to (S)-2, (R)-2 loses a hydrogen bond with loop 3 but forms an extra one with Val83 located on loop 1 (Figure 11A,B) due to positioning of its 3′-OH group 1.2 Å closer to the active site loop 1 (Figure S21D). The extensive network of hydrogen bonds to water molecules present in the active site of bound (S)-2 is lost in the (R)-2 complex as well as the hydrogen bonds to Tyr198, while most of the remaining interactions are still present. This loss is most likely responsible for the decreased potency and is likely due to a solvation effect, as water-mediated contacts can be highly specific and participate in the recognition of ligands.⁸²

Figure 11.

Binding of (S)-2 (A) and (R)-2 (B) to TcHGXPRT. The (S)-2 purine ring forms hydrogen bonds to Tyr198 and Asp145. The 5′-phosphate portion and 3′-OH substituents are held in place by an extensive network of hydrogen bonds to water molecules. Combined, these features are believed to be responsible for its potency against TcHGXPRT. The contacts with water molecules are mostly absent in (R)-2 binding as well as the hydrogen bond between Tyr198 and the purine ring N3. All interactions are represented as in Figure 9. The interactions that are unique to each inhibitor are highlighted in red on each interaction map.

(S)-1, the Gua-based version of (S)-2, is the most potent inhibitor of TcHGXPRT in the set evaluated, with an approximately 2.5-fold improvement when compared to (S)-2. The structure of PfHGXPRT bound to (S)-1 has been recently determined (7TUX.pdb³⁴). All PfHGXPRT residues that interact with (S)-1 are conserved in TcHGXPRT and adopt the same conformation within the active site (Figure S23A). In the PfHGXPRT·(S)-1 complex, loop 2 is ordered and in a conformation equivalent to the closed conformation in TcHGPRT. When closed, loop 2 pushes the inhibitor deeper into the active site,⁵⁸ forming interactions observed with residues from loop 1 and loop 2 (Figure S23). (S)-1 is 100-fold more potent vs. TcHGPRT and 1200-fold more potent against PfHGXPRT when compared to TcHGXPRT (Tables 3 and S2). (S)-1 is Gua-based, with Gua being the preferred substrate for PfHGXPRT³⁷ and second-preferred for TcHGPRT, with however, near identical values of K_m, k_cat, and k_cat/K_m when compared to Hx as a substrate.⁴³ As the amino acid composition and three-dimensional arrangement are extensively conserved in the active sites of all three enzymes, the differences in potency of (S)-1 and all TSAIs evaluated (Tables 3 and S2) can be explained by differences in the kinetic mechanism, including the rate-limiting conformational changes and variations of the TS structure and in the three amino acid residues responsible for substrate preference. This also explains why the TSAIs are better inhibitors of TcHGPRTs (Table 3), since its kinetic mechanism and the composition of amino acids are identical to PfHGXPRT.⁴³

It is most interesting that there is a conservation of potency for compounds (S)-1 (Gua-based) and (S)-2 (Hx-based) against all protozoan HG(X)PRTs as top inhibitors against TcHGPRT (K_i ≈ 0.01 μM, Tables 3 and S2) and PfHGXPRT (K_i = 0.0005–0.0006 μM, Tables 3 and S2) while showing promising activity against TcHGXPRT, in the low micromolar range (K_i 0.6–1.6 μM, Tables 3 and S2). However, as the concomitant inhibition of TcHGPRT and TcHGXPRT is an important characteristic in the optimization of lead compounds, another inhibitor that shows promising features is TSAI (S)-4. Unlike the other compounds presented in Table 3, whose K_i values show several fold differences against both isoforms, the (S)-4 potency is in the same range for TcHGPRT (K_i = 1.4 μM) and TcHGXPRT (K_i = 0.7 μM). When compared to (S)-2, (S)-4 contains a methyl group as the R₃ substituent (Figure 8), a portion of the ligand’s structure that we hypothesize to be more amenable to chemical modifications, allowing the insertion of moieties able to improve potency. Substitutions in the sugar portion would also avoid the triad of residues putatively involved in substrate preference, located around the purine ring (Figure 7), and the results presented here suggest that a Gua-based inhibitor can be optimized to target both TcHGPRT and TcHGXPRT. Further efforts are currently underway to determine the structure of TcHGXPRT bound to Immucillin inhibitors by cocrystallization in hopes of obtaining a complex (E·I or E·I·PPi) in its closed conformation, allowing the identification of all contacts formed between ligands and amino acid residues from loop 2.

Conclusion

A key step in the development of effective CD therapies that target purine metabolism is the simultaneous optimization of lead compounds that can inhibit both TcHGPRT and TcHGXPRT. We have shown the difference in substrate preference between both types of isoforms, pinpointing for the first time the XPRT activity in T. cruzi and rectifying their previous classification as putative HGPRTs. An important part of developing a dual-target inhibitor includes elucidation of the kinetic mechanisms and structures of both enzymes. Our results reveal that the steady-state ordered kinetic mechanism is conserved for all TcHG(X)PRT isoforms (Table 2) as well as a postchemistry event as the rate-limiting step(s) during catalysis (Figure 3). Further studies are necessary to clarify the role of product release and the postchemistry opening of loop 2 as separate or intrinsic parts of TcHGXPRT rate-limiting step(s) of catalysis. The quaternary structure of TcHGPRT was extensively studied in the late 1990s and early 2000s,^49,71,73 and here we describe, for the first time, the structure of TcHGXPRT bound to its preferred substrate Xan and a subset of TSAIs that are potent inhibitors of PfHGXPRT. Our results indicate that these TSAIs can be repurposed as inhibitors of the orthologous trypanosomal enzymes due to the conservation in their active-site structures. The structures of PfHGXPRT and TcHG(X)PRTs, as well as of other Type I PRTases,^56,58,69 all show the extensive conservation in binding of both the purine and 5′-phosphate moieties of substrates, products, and NMP-based inhibitors. The location of the 5′-phosphate binding site, at the hinge of loop 2, may present challenges for the optimization of more soluble phosphate-mimic substituents, as we recently demonstrated;⁷⁹ therefore, modifications in the purine ring and sugar moiety may hold the key to increased potency and selectivity against all T. cruzi isoforms.

Glossary

Abbreviations

C

competitive

CD

Chagas Disease

CV

column volume

Gua

guanine

GMP

guanosine monophosphate

GPRT

guanine phosphoribosyltransferase

HPRT

hypoxanthine phosphoribosyltransferase

HGPRT

hypoxanthine–guanine phosphoribosyltransferase

Hx

hypoxanthine

HGXPRT

hypoxanthine–guanine–xanthine phosphoribosyltransferase

IMP

inosine monophosphate

NC

noncompetitive

NMP(s)

nucleoside monophosphate(s)

Pi

phosphate

PPi

pyrophosphate

PRPP

5-phospho-d-ribose 1-pyrophosphate

PPRT(s)

purine phosphoribosyltransferase(s)

PTS1

peroxisomal targeting signal type 1

RT

room temperature

SEC

size-exclusion chromatography

TB

Terrific Broth

TcA

TcHGPRT isoform A

TcB

TcHGXPRT isoform B

TcC

TcHGPRT isoform C

TcD

TcHGXPRT isoform D

TS

transition state

TSAI(s)

transition-state analogue inhibitor(s)

Type I PRTases

Type I phosphoribosyltransferases

UC

uncompetitive

Xan

xanthine

XMP

xanthosine monophosphate

XPRT

xanthine phosphoribosyltransferase

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.biochem.3c00116.

• Primary sequence alignments, SDS-PAGE analysis, initial rate data, SEC results, structural overlays, and interaction maps (PDF)

Accession Codes

TcB: TcCLB.509693.80 (TriTrypDB), Q4GRC3 (UNIPROT); TcD: TcCLB.506457.40 (TriTrypDB), Q4DGA2 (UNIPROT); TcA: TcCLB.509693.70 (TriTrypDB), Q4DRC4 (UNIPROT); TcC: TcCLB.506457.30 (TriTrypDB), Q4DGA3 (UNIPROT); PfHGXPRT: P20035 (UNIPROT); HsHGPRT: P00492 (UNIPROT)

Author Present Address

Nuvelus, 371 Sherborne Cv, Cordova, Tennessee 38018, United States. (R.H.)

Author Contributions

K.G. planned and performed experiments, established protocols, analyzed the data, and wrote the manuscript. K.M.M. refined TcHGXPRT structures. R.H. obtained TcHGXPRT crystals and diffraction data. S.W.M. and G.E.P. performed experiments. K.S., K.C., and J.N.B. performed the synthesis of TSAIs. A.L.L. supervised the crystallography refinement and reviewed the manuscript. P.C.T. provided all transition-state analogue inhibitors and provided funding. T.D.M. edited and reviewed the manuscript and provided funding. A.K. planned and executed experiments, established protocols, wrote and reviewed the manuscript, and provided funding. All authors have given approval to the final version of the manuscript.

This research is funded by The National Institute of Health, National Institute of Allergy and Infectious Diseases – NIH/NIAID, under the grant 1R01AI127807, and by the National Science Foundation – NSF, under the grants CHE-2041047 and NSF CHE-2204080.

The authors declare no competing financial interest.