Understanding the molecular mechanism of substrate channeling and domain communication in protozoal bifunctional TS-DHFR

Abstract

Most species, such as humans, have monofunctional forms of thymidylate synthase (TS) and dihydrofolate reductase (DHFR) that are key folate metabolism enzymes making critical folate components required for DNA synthesis. In contrast, several parasitic protozoa, including Leishmania major (Lm), Plasmodium falciparum (Pf), Toxoplasma gondii (Tg) and Cryptosporidium hominis (Ch), contain a unique bifunctional thymidylate synthase-dihydrofolate reductase (TS-DHFR) having the two sequential catalytic activities contained on a single polypeptide chain. It has been suggested that the bifunctional nature of the two catalytic activities may enable substrate channeling. The 3D structures for each of these enzymes reveals distinct features for each species. While three of the four species (Pf, Tg and Ch) contain a junctional region linking the two domains, this is lacking in Lm. The Lm and Pf contain N-terminal amino acid extensions. A multidisciplinary approach using structural studies and transient kinetic analyses combined with mutational analysis has investigated the roles of these unique structural features for each enzyme. Additionally, the possibility of substrate channeling behavior was explored. These studies have identified unique, functional regions in both the TS and DHFR domains that govern efficient catalysis for each species. Surprisingly, even though there are structural similarities among the species, each is regulated in a distinct manner. This structural and mechanistic information was also used to exploit species-specific inhibitor design.

Introduction

Two enzymes crucial for DNA synthesis and one-carbon transfers are thymidylate synthase (TS) and dihydrofolate reductase (DHFR). TS catalyzes the formation of the nucleotide deoxythymidine monophosphate (dTMP) from deoxyuridine monophosphate (dUMP) utilizing (6R)-L-5,10-methylene-tetrahydrofolate (CH₂H₄F), which is converted to dihydrofolate (H₂F) upon the transfer of one carbon unit (Hardy et al., 1987; Schiffer et al., 1995). DHFR subsequently converts the H₂F to tetrahydrofolate (H₄F) through a hydride transfer step donated by the co-factor NADPH (Fig. 1A) (Fierke et al., 1987). As illustrated in Fig. 1, for most organisms, folate metabolism and nucleotide synthesis are linked via the TS and DHFR catalytic activities (Gangjee et al., 2007; Sharma and Chauhan, 2012; Sharma et al., 2013). While these enzymes are expressed on separate polypeptides in almost all other organisms including humans, in the protozoal parasites such as Leishmania major, Cryptosporidium hominis, Toxoplasma gondii and Plasmodium falciparum, TS and DHFR are tethered together to form a bifunctional enzyme, TS-DHFR (TS-DHFR and DHFR-TS are used interchangeably since genetically the DHFR activity resides at the N-terminus while functionally the TS activity is the first catalytic reaction followed in sequence by DHFR.) (Fig. 1B) (Hardy et al., 1987; Knighton et al., 1994; Yuvaniyama et al., 2003; O'Neil et al., 2003a,b; Senkovich et al., 2009; Begley et al., 2011). It has been suggested that bifunctional forms of TS-DHFR might exhibit ‘substrate channeling’, which is defined as the transfer of a metabolite from one enzyme active site to another without allowing diffusion of the molecule into bulk solution (Meek et al., 1985; Trujillo et al., 1997; Anderson, 1999). In this scenario, dihydrofolate is directly transferred from the TS site to the DHFR site (Fig. 1A). Substrate channeling has been suggested to play an integral role in numerous essential cellular functions, which have implications for an understanding of metabolic regulation, and in this instance, offer a survival advantage for the parasite (Anderson, 1999).

Fig. 1.

(A) The reaction scheme showing catalytic activities for TS (red) and DHFR (blue). (B) Diagram showing organism-specific differences in mono and bifunctional TS-DHFR. The N-terminal tail (yellow) of the DHFR active site (blue) can be of variable length or completely absent. This is also true of the linker/junctional region (green) linking DHFR (blue) and TS (red). Some species, Plasmodium and Leishmania, have an additional N-terminal extension (yellow); however, this is absent in both Cryptosporidium and Toxoplasma. A junctional/linker region (green) is present in Cryptosporidium, Toxoplasma and Plasmodium but absent in Leishmania. Possible substrate channeling of H₂F, from the TS to the DHFR site, is denoted by solid arrow. This figure is available in black and white in print and color in online.

As illustrated in Fig. 1B, the overall domain architecture among the bifunctional TS-DHFRs differs substantially for individual protozoal parasites. The L. major and P. falciparum contain an N-terminal extension of the DHFR (shown in yellow, Fig. 1B) whereas C. hominis, T. gondii and P. falciparum contain a junctional region between the DHFR and TS (shown in green, Fig. 1B) (Roos, 1993; O'Neil et al. , 2003a,b). The sequence identity among the four bifunctional TS-DHFR parasitic enzymes at the TS domain is reasonably high, while the sequence identity at the DHFR domain is much lower. Structural studies on monofunctional forms of DHFR from multiple species have shown that DHFR often exhibits high functional homology in spite of possessing low primary sequence identity. Interestingly, flanking the TS and DHFR active sites in parasitic TS-DHFR's is distant, junctional or ‘non-active site’ regions, which have no homology between organisms (Peterson et al., 1988; Ivanetich and Santi, 1990; Knighton et al., 1994) (Fig. 1A). These regions are completely absent in humans. As illustrated in Fig. 1B, sequence alignment reveals some noteworthy and distinct aspects of the Toxoplasma, Cryptosporidium and Plasmodium TS-DHFR enzymes in contrast with the Leishmania TS-DHFR. The 22-amino acid N-terminal tail (shown in yellow) found in Leishmania is missing in the Toxoplasma and Cryptosporidium enzymes, and shortened to 5 amino acids in the Plasmodium enzyme. On the other hand, there is an insertion of an additional 55, 69 or 91 amino acid linker region (shown in green) between the TS and DHFR domains in the Cryptosporidium, Toxoplasma and Plasmodium, respectively, as compared with the absence of an amino acid linker region in Leishmania.

The first crystal structure solved for a bifunctional TS-DHFR was that of L. major (Lm) (Knighton et al., 1994). This structure, complexed with TS and DHFR active site inhibitors, revealed a homodimer in which each monomer of the enzyme contains an N-terminal DHFR domain that rests on the shoulders of the TS domain. In the Lm structure, the TS domain forms the majority of the dimerization interface, while the DHFR domains are spatially separated (Fig. 2A). More recent structures for those bifunctional enzymes containing a junctional region between the TS and the DHFR domains [P. falciparum (Pf), T. gondii (Tg) and C. hominis (Ch)] show distinct structural differences (Fig. 2B–D). In contrast to the Lm structure, the structures of these three bifunctional enzymes with a junctional region show that DHFR domains from each monomer are closer together and a part of this junctional region forms a crossover or ‘donated’ helix that appears to interact with the catalytic B-helix of the DHFR domain adjacent monomer (Knighton et al., 1994, Yuvaniyama et al., 2003; O'Neil et al., 2003a,b; Senkovich et al., 2009; Begley et al., 2011; Sharma et al., 2013). These key differences between the bifunctional TS-DHFR enzymes from various parasitic species have led to their classification of Class I or Class II in which the presence of the junctional region defines Class I, while its absence defines Class II (O'Neil et al., 2003a,b). See Figure 2A–D. Each TS-DHFR is a homodimer and the DHFR domain (dark and light blue) is N-terminal to the TS domain (red and pink). A comparison of the four structures reveals that the Plasmodium, Toxoplasma and Cryptosporidium are more similar as compared with Leishmania. These include: the relative orientations of the TS and DHFR domains and active sites and the presence of structured linker regions of the DHFR that are in direct contact with the opposite DHFR domain of the dimer. As C. hominus, T. gondii and P. falciparum have a long linker or junctional region structurally, their active sites are on different faces of the enzyme. The L. major enzyme has essentially no linker and has TS and DHFR active sites on the same face.

Fig. 2.

The structures of the four bifunctional TS-DHFR enzymes, L. major, C. hominis, T. gondii and P. falciparum, are illustrated as ribbon diagrams in (A–D). Each TS-DHFR is a homodimer and the DHFR domain (dark and light blue) is N-terminal to the TS domain (red and pink). The junctional region for one monomer is highlighted in green. (A) L. major, (B) P. falciparum, (C) T. gondii and (D) C. hominis 3D structures. This figure is available in black and white in print and color in online.

In this paper, we present an overview summarizing our important findings in each system and compare and contrast each protozoal species. We have used a multidisciplinary approach employing transient kinetic methodology, site-directed mutagenesis and structural analysis to elucidate the molecular details involved in TS and DHFR catalysis. In exploring the mechanism and underlying structural determinants influencing function for each bifunctional enzyme, we summarize our findings in this report. We have found that there are key differences in TS and DHFR domain interactions and how these may regulate catalysis as well as whether substrate channeling occurs for each species. We have also discovered that these unique features between parasitic and human enzymes can be exploited for design of new inhibitors.

Materials and methods

A combination of transient kinetic methodologies employing rapid chemical quench and stopped-flow were used to explore each mechanism.

Chemicals

All buffers and other reagents employed were of the highest commercial purity. 7,8-Dihydrofolate (H₂F) was chemically prepared by the reduction of folic acid (Sigma) with sodium hydrosulfite (Blakley, 1960). Tritium-labeled and unlabeled CH₂H₄F were synthesized as previously described using tritiated or unlabeled folic acid, respectively, as starting material, as previously described (Martucci et al., 2009). NADPH was purchased from Sigma and dUMP from MP Biomedicals. The propargyl dideazfolate (PDDF) used in the studies was generously provided by Dr Roy Kisliuk and Dr Ann Jackman.

All buffers and other reagents employed were of the highest commercial purity. 7,8-Dihydrofolate (H₂F) was chemically prepared by the reduction of folic acid (Sigma) with sodium hydrosulfite (Blakley, 1960). Tritium-labeled and unlabeled CH₂H₄F were synthesized as previously described using tritiated or unlabeled folic acid, respectively, as starting material, as previously described (Martucci et al., 2009).

Purification of bifunctional TS-DHFR

Each bifunctional TS-DHFR protein was expressed and purified as previously described (Liang and Anderson, 1998a,b; Johnson et al., 2002; Atreya and Anderson, 2004; Dasgupta, 2008; Dasgupta and Anderson, 2008; Sharma et al., 2013). Briefly, the bifunctional protein was purified using an affinity column of methotrexate bound to agarose (Sigma), as described previously (Matthews et al., 1977; Grumont et al., 1988). A PD-10 column (Amersham Biosciences) was also used to remove residual H₂F after purification. The concentration of purified bifunctional TS-DHFR was determined by spectrophotometrically at 280 nm, using the respective calculated extinction coefficients, i.e. for T. gondii, 71 850 M⁻¹cm⁻¹. The enzyme was flash frozen in storage buffer 25 mM Tris 7.3, 10 mM DTT, and 10% glycerol and stored in −80 °C.

Steady-state kinetic experiments

Steady-state TS and DHFR kinetic experiments were performed by incubating 100 nM of enzyme with a saturating concentration of one TS or DHFR ligand (100 μM, confirmed to be well above saturating) and measuring the reaction rate at varying concentrations of the complementary TS or DHFR ligand (up to 250 μM). Experiments were performed in 1× enzyme buffer: 50 mM Tris pH 7.8, 25 mM MgCl₂, 1 mM EDTA, 2 mM DTT. The DHFR activity was measured by a decrease in absorbance at 340 nm as NADPH and H₂F were converted to NADP⁺ and H₄F (Δε = 10 mM⁻¹cm⁻¹). The TS activity was measured by an increase in absorbance at 340 nm as dUMP and CH₂H₄F were converted to dTMP and H₂F (Δε = 6.4 mM⁻¹cm⁻¹). Rate constants for steady-state experiments were estimated by fitting the data to a Michaelis–Menten hyperbolic curve (v = V_max[S]/(K_m + [S]), where v is the reaction rate, [S] the concentration of substrate and K_m the Michaelis constant) using the curve-fitting program, Kaleidagraph (version 4.03, Synergy Software).

Stopped-flow kinetic experiments

Stopped-flow experiments were performed using a Kintek SF-2001 apparatus (Kintek Instruments, Austin, TX). To determine the rate of the DHFR reaction, coenzyme fluorescence resonance energy transfer experiments were carried out with 290 nm excitation and emission using an interference filter at 450 nm. Concentrations of enzyme and substrates described are those after mixing. To find the single turnover rate (k_chem) of the DHFR reaction, enzyme (50 µM in 2x enzyme buffer: 100 mM Tris pH 7.8, 50 mM MgCl₂, 2 mM EDTA, 2 mM DTT) was incubated with NADPH (250 µM) and then mixed with H₂F (25 µM). Changes in fluorescence upon mixing were monitored and the resultant data were fit in Kaleidagraph to a single exponential curve, Fluorescence =Ae^{(−kchem × time)}, where A is the amplitude of fluorescence and k_chem is the exponential rate constant. Stopped-flow traces were smoothed using the Kintek software smoothing program, which averages together adjacent points to reduce noise.

For the TS burst reaction, enzyme (25 µM) was preincubated with excess dUMP (1 mM) and mixed with excess CH₂H₄F (500 µM). Changes in absorbance at 340 nm were monitored, and the trace was fit to a burst curve Fluorescence = Ae^{(−kburst × time)} + k_ss × time, where where A is the amplitude of absorbance, k_burst is the exponential phase rate constant and k_ss is the linear phase rate constant.

Rapid Chemical Quench

The bifunctional TS-DHFR reaction was measured via using a Kintek RFQ-3 Rapid Chemical Quench Apparatus (Kintek Instruments, Austin, TX). The reactions were initiated by mixing 15 µL enzyme solution (enzyme + 2X enzyme buffer) with 15 µL radiolabeled CH₂-H₄F. The reactions were terminated by quenching with 89 µl of a solution of 0.78 N KOH, 10% sodium-ascorbate and 200 mM 2-mercaptoethanol. The TS-DHFR single turnover bifunctional reaction was monitored by addition of [³H]-CH₂H₄F (10 µM) to enzyme (50 µM), dUMP (500 µM) and NADPH (500 µM). For the pulse-chase experiment, 50 µM of enzyme was incubated with 500 µM dUMP and 500 µM NADPH. The reaction was initiated by mixing the enzyme solution with 25 µM of [³H]-CH₂H₄F and 150 µM unlabeled H₂F. The data from the single enzyme turnover experiments were fit to a single exponential equation using Kaleidagraph.

High performance liquid chromatography analysis

Tritiated products of the rapid chemical quench experiments were quantified by high performance liquid chromatography (HPLC) in combination with a radioactivity flow detector as described previously (Atreya and Anderson, 2004; Martucci et al., 2009). The HPLC separation was performed using a BDS-Hypersil C18 reverse phase column (Thermo) with a flow rate of 1 ml min⁻¹. An isocratic separation using a solvent system of 10% methanol in 200 mM triethylammonium bicarbonate at pH 8.0 was used. The elution times were as follows: H₄F, 9 min; H₂F, 18 min; CH₂H₄F, 20 min.

Results and Discussion

Species-specific kinetic differences for domain–domain interactions in protozoal TS-DHFR enzymes

Investigating substrate channeling and domain–domain communication for L. major TS-DHFR

Our earlier studies used rapid chemical quench and stopped-flow fluorescence studies to investigate substrate channeling and domain–domain interactions for the Lm TS-DHFR (Liang and Anderson, 1998a,b). One of the key experiments to assess substrate channeling is termed a pulse-chase experiment as illustrated in Scheme 1. The TS-DHFR bifunctional single turnover reaction involving the conversion of radiolabeled (denoted by *) *CH₂H₄F to *H₂F at the TS active site and *H₂F to *H₄F at the DHFR site is performed. As illustrated in the experimental design in Scheme 1, if the bifunctional TS-DHFR exhibits substrate channeling, the presence of cold H₂F in solution would be expected to have a minimal effect on the bifunctional reaction (Scheme 1A). However, if the enzyme did not channel, radiolabeled H₂F would be able to freely diffuse into solution after being produced by the TS reaction (Scheme 1B). The excess cold H₂F in solution would therefore be able to outcompete the H₂F produced by the TS for the H₂F binding sites in the DHFR domain. A substantial buildup of radioactive H₂F and a decrease in the apparent rate of radiolabeled H₄F production would therefore be expected.

Scheme 1.

Diagram of the pulse-chase experiment for measuring H₂F channeling. (A) For substrate channeling, conversion of radiolabeled CH₂H₄F to radiolabeled H₂F (yellow) would involve conversion to directly transfer from the TS to the DHFR domain without releasing it into solution, therefore only a small amount of the H₂F intermediate product would be observed. However, an enzyme unable to channel (B) would release radiolabeled H₂F (yellow) into solution, where it would be outcompeted by an excess of unlabeled H₂F (gray) for substrate binding sites in the DHFR domain. In this scenario, a substantial buildup of radiolabeled H₂F in solution would be observed. This figure is available in black and white in print and color in online.

For the Lm TS-DHFR, there was very little buildup (<10%) of the radiolabeled H₂F whereas a large buildup (>50%) was noted in comparison with an experiment which examined monofunctional TS and DHFR for Escherichia coli enzymes combined together in a single enzyme turnover experiment (Liang and Anderson, 1998a,b).

Additional single enzyme turnover experiments examined the DHFR reaction using rapid quench and stopped-flow fluorescence energy transfer to monitor DHFR catalysis. A combination of these two experiments revealed a large activation (almost 10-fold) in the rate of DHFR catalysis when TS active site ligands (fdUMP and CH₂H₄F) were present at the TS site as illustrated in Fig. 3.

Fig. 3.

Diagram illustrating the activation of Lm DHFR catalysis in the presence of the TS active site ligands, fdUMP and CH₂H₄F and the 22 aa N-terminal tail that may play a role in the activation.

One of the interesting questions that arose from the observation of TS ligand-mediated activation of DHFR catalysis was whether there was a structural component involved in this modulation. As shown in Fig. 3, a close look at one Lm TS-DHFR monomer shows that the N-terminal 22-amino acid tail extending from the DHFR domain wraps down around the TS domain near the active site making extensive contacts with the TS domain. Mutational studies were conducted to delete this 22-amino terminal extension to see whether there was an effect on DHFR catalysis. The removal of these residues and analysis of the reaction kinetics revealed a substantial 3-fold activation in DHFR catalysis (45 s⁻¹), indicating that the N-terminal extension in this case plays an autoinhibitory role (Dasgupta and Anderson, 2008).

Probing substrate channeling and domain interactions for P. falciparum TS-DHFR

As shown in Fig. 2B, the structure of Pf TS-DHFR, a member of the Class I bifunctional enzymes, provides the structural determinants for understanding earlier biochemical studies by the Rathod lab who examined the effect of separately expressing the TS and DHFR domains (Shallom et al., 1999). These experiments showed that truncated versions of the Pf TS-DHFR containing only the DHFR domain maintained catalytic activity while a truncated version of the TS domain, having the DHFR domain and the junctional region removed, lacked TS activity. Interestingly, TS activity was restored upon addition of the DHFR/junctional region in trans restored catalytic activity (Shallom et al., 1999).

Our kinetic studies including pulse-chase experiments examining substrate channeling in the Pf TS-DHFR showed that channeling of the H₂F from the TS site to the DHFR site was maintained. An investigation of Pf TS-DHFR and possible interdomain interactions between TS and DHFR and DHFR–DHFR domain interactions were also examined. These studies showed that the mechanism and regulation of TS and DHFR catalytic activities were distinct compared with Lm TS-DHFR (Dasgupta and Anderson, 2008; Dasgupta et al., 2009). The DHFR catalytic rate (60 s⁻¹) for Pf TS-DHFR was almost 5-fold faster than the Lm TS-DHFR; however, in the presence of TS ligands the DHFR rate was increased to 120 s⁻¹ showing a 2-fold activation.

The N-terminal extension for Pf TS-DHFR is much shorter than the Lm TS-DHFR (see Fig. 1, yellow). A deletion of these residues resulted in a 2-fold decrease in DHFR catalytic activity (30 s⁻¹). This is in contrast to Lm TS-DHFR where the N-terminal extension appears to play an autoinhibitory role and removal results in an increase in catalytic activity.

While the junctional region is missing in Lm TS-DHFR, the other three TS-DHFR enzymes contain this subdomain, often referred to as a crossover helix. The role of the crossover helix in Pf TS-DHFR (Fig. 2B, highlighted in green) was examined by mutational analysis of the residues on the face of the crossover helix. The crossover helix in P. falciparum TS-DHFR (residues 283–295) has numerous acidic residues (residues 283–289 are either Asp or Glu, followed by F290, V291, Y292, F293, N294 and F295), which form electrostatic interactions with the many positively charged residues (primarily Lys) on the backside of the DHFR active site. These residues face the B-helix of the adjacent DHFR domain, which were mutated to alanine. Mutational disruption of the crossover helix in the Pf TS-DHFR in Ala-FACE mutant had no effect on either the overall rate or reaction kinetics. Due to difficulties with protein expression and stability, a polyalanine helical, all-Ala mutant and a Δ-helix deletion mutant of Pf TS-DHFR were unable to be examined. This detailed kinetic analysis revealed no changes in the kinetic parameters, indicating that the crossover helix is not important for Pf TS-DHFR (Dasgupta and Anderson, 2008). Since our mutational analysis with Pf TS-DHFR revealed the ala-FACE mutant is not kinetically different from WT enzyme, it is possible that the electrostatic interactions are involved in the stability of P. falciparum enzyme rather than affecting catalysis.

Evaluating substrate channeling and domain interactions for T. gondii TS-DHFR

We recently solved the Tg TS-DHFR structure (Fig. 2C) (Sharma et al., 2013). As shown in Fig. 2C, the overall architecture is similar to that for Pf TS-DHFR. A series of kinetic studies, comparable to those described for Lm and Pf TS-DHFR, including pulse-chase experiments with Tg TS-DHFR, established that the substrate channeling of H₂F from the TS to the DHFR site was preserved in this bifunctional enzyme. Single turnover experiments revealed a surprisingly fast rate (180 s⁻¹) for DHFR catalysis, even in the absence of TS ligands (Johnson et al., 2002). Moreover, the addition of TS ligands did not further enhance the rate, in contrast to the TS-ligand activation observed in both Lm and Pf TS-DHFR. This bifunctional enzyme lacks an N-terminal extension; however, this protein has a junctional region that is shorter than the Pf TS-DHFR and a portion of this region forms a crossover helix as found in Pf TS-DHFR (Fig. 1B). There are several unique features of the crossover helix found in Tg TS-DHFR. The crossover helix is distinct from Pf TS-DHFR in that it is shorter as there is a kink in the helix due to the presence of a proline residue (P292). Notably, the mutation of the proline to an alanine results in a 2-fold rate enhancement for DHFR catalysis. On the other hand, the Δ-helix crossover deletion mutant and all-Alanine helix mutant show a decrease in DHFR catalytic activity (Sharma et al., 2013).

A paradigm shift in substrate channeling and domain interactions for C. hominis TS-DHFR

As shown in Fig. 2D, the overall crystal structure of Ch TS-DHFR is similar to that of the two other Class I TS-DHFR family. While the Ch TS-DHFR has the shortest junctional regional linker region of the three Class I TS-DHFR bifunctional enzymes, the crossover helix is maintained. A series of single enzyme turnover and pulse-chase experiments lead to surprising results in investigating substrate channeling and domain interactions for the Ch TS-DHFR bifunctional enzyme. The pulse-chase experiment which examined radiolabeled CH₂H₄F conversion to H₄F in the presence of an excess of cold H₂F showed that there was a large buildup (>70%) of hot H₂F indicating a complete lack of substrate channeling in this enzyme.

Experiments to evaluate DHFR catalysis and the effect of TS ligands showed that the DHFR rate was fast (130 s⁻¹) and unaffected by the presence of TS active site ligands, fdUMP and CH₂H₄F similar to that observed for the Tg TS-DHFR (Atreya and Anderson, 2004). A comparison of all four species is illustrated in Scheme 2.

Scheme 2.

Comparison of TS-ligand activation of DHFR catalysis.

The role of the crossover helix in the DHFR catalysis was probed by mutational analysis. The residues on one face of the helix appear to make extensive contacts with the B-helix abutting the DHFR active site of the adjacent DHFR monomer (see Fig. 4A). Mutational analysis of the crossover helix to evaluate the importance of these helical contacts was conducted using: (i) an ala-FACE mutant in which residues were mutated to alanine, (ii) the entire crossover helix was mutated to alanine, and (iii) a series of mutations of individual residues near the adjacent beta sheet face. As illustrated in Fig. 4B, there were large effects on DHFR chemical catalysis, indicating that this crossover helix plays a critical role in modulating catalysis. Additionally, a series of thermal denaturation experiments showed that the crossover helix also plays a role in stabilizing the Ch TS-DHFR.

Fig. 4.

Role of crossover helix for Ch TS-DHFR catalysis. (A) Crossover helix from one DHFR domain monomer interacts with the Helix B of adjacent DHFR domain. (upper) Expanded view of potential helical interactions and residues selected for mutational analysis. (B) Effect of mutations on Ch DHFR catalysis (left) and stability to thermal denaturation (right).

The dramatic difference in role of the crossover helix for Pf TS-DHFR as compared with Ch TS-DHFR can be understood in structural terms. The crossover helices in C. hominis and P. falciparum TS-DHFR vary in their amino acid compositions and interactions. In C. hominis, the face of the crossover is composed of hydrophilic and hydrophobic residues (S195, D198, L202, I206, R210), which forms helix–helix interactions with amino acids of the backbone of the DHFR active site (Y132, N42, F35, F172 and E31). In contrast, the crossover helix in P. falciparum TS-DHFR (residues 283–295) has numerous acidic residues (residues 283–289 are either Asp or Glu, followed by F290, V291, Y292, F293, N294 and F295), which form electrostatic interactions with the many positively charged residues (primarily Lys) on the backside of the DHFR active site.

Structural basis for substrate channeling in bifunctional TS-DHFRs

Since the only bifunctional TS-DHFR enzyme that does not exhibit channeling behavior is the Ch TS-DHFR, one might consider what are the underlying structural features that might govern whether each species exhibits substrate channeling? It is notable, for instance, that while Tg TS-DHFR does show substrate channeling, the species closest in structure, Ch TS-DHFR, does not. To examine whether there might be species differences in surface charge distribution, the structures were evaluated. As illustrated in Fig. 5A–D, a comparison of the Coulombic surface charge distribution of L. major, P. falciparum and T. gondii to C. hominis TS-DHFR revealed that the C. hominis was the only species to contain a strongly negatively charged DHFR domain. The negative charge could result in the dissociation of dihydrofolate into bulk solution explaining the lack of channeling behavior in C. hominis TS-DHFR. As illustrated in Fig. 5, a consideration of the electrostatic surface indicates that the negatively charged dihydrofolate would be directed to the more positively charged DHFR domain for L. major, T. gondii and P. falciparum, while the negatively charged DHFR domain for C. hominis would likely facilitate dissociation of the H₂F into bulk solution.

Fig. 5.

Overall surface charge distributions of channeling and non-channeling TS-DHFRs. Figures of Coulombic charges were created by using Chimera using default settings. Negative charges are colored red and positive charges are colored blue. Pictured are the TS-DHFRs from (A) L. major, (B) P. falciparum (PDB ID: 1J3I), (C) T. gondii (4EIL) and (D) C. hominis (PDB ID: 1QZF). The DHFR domain of each enzyme is circled. An asterisk (*) indicates a version of TS-DHFR capable of channeling (Sharma et al., 2013). This figure is available in black and white in print and color in online.

Designing species-specific inhibitors

With a detailed knowledge of both structure and unique regions of each bifunctional TS-DHFR enzyme that are important in catalytic function, this information can be used to design species-specific inhibitors. This can be accomplished by focus on a known active site and exploit structural differences between species. Alternatively, a non-active site such as an allosteric site, shown through mutational analysis to be important for catalytic function, can be targeted. We will mention illustrative examples where the structural and mechanistic information has been utilized for inhibitor design. As shown in Fig. 6, the Cryptosporidium TS-DHFR enzyme has several variant residues in the TS active site in the folate binding region that are unique to the parasite and served as a basis for designing parasite-specific inhibitors (Fig. 6A) (Kumar et al., 2013, 2014; Mukerjee et al., 2015). The crossover helical region of the Ch TS-DHFR, shown to be important for DHFR catalysis, was successfully targeted with ɑ-helical mimetics (Vargo et al., 2009; Martucci et al., 2013) (Fig. 6B). For the Toxoplasma TS-DHFR, a unique TS–TS interface in the TS dimer was effectively inhibited using a peptide-based strategy (Landau et al., 2013) (Fig. 6C).

Fig. 6.

Exploiting unique structural features for inhibitor design. (A) Targeting variant TS active site residues for Ch TS-DHFR, (B) crossover helix in DHFR domain of Ch TS-DHFR and (C) TS–TS dimer interface for Tg TS-DHFR.

As illustrated in Fig. 7, novel binding sites on the bifunctional TS-DHFR enzymes identified through mutational analysis and structural studies also have been successfully targeted using a combination of virtual library screening and molecular docking as well as a peptidomimetic strategy. These include molecular and virtual screening to identify inhibitors of Lm TS-DHFR at the TS-DHFR interface in Lm TS-DHFR (Fig. 7A) (Atreya et al., 2003), a pocket near the crossover helix of Ch TS-DHFR (Martucci et al., 2009) (Fig. 7B), a pocket near the TS-dimer interface for Tg TS-DHFR (Zaware et al., 2013) (Fig. 7C) and a pocket near the DHFR active site of Pf TS-DHFR (Fig. 7D).

Fig. 7.

Targeting novel binding or allosteric sites. (A) Targeting the TS-DHFR interface for Lm TS-DHFR, (B) pocket near the crossover helix in DHFR domain of Ch TS-DHFR, (C) region near the TS–TS dimer interface for Tg TS-DHFR and (D) the DHFR active site of Pf TS-DHFR.

Physiological significance for substrate channeling and domain interactions in bifunctional TS-DHFR enzymes

Several putative physiological advantages have been suggested for substrate channeling, including control of metabolic flux, protection of reactive and/or toxic intermediates, increased catalytic efficiency, decreased flux and decreased diffusion of intermediates away from the catalytic sites (Mathews and Sinha, 1982, Ovadi et al., 1989; Ovadi, 1991; Rudolph et al., 1997). It has been suggested that the presence of bifunctional TS-DHFR enzymes in parasites offers a survival advantage.(Ivanetich and Santi, 1990) For example, a knockout experiment in the bifunctional Leishmania TS-DHFR has shown that the DHFR-TS gene is essential for survival (Cruz et al., 1993; Gueiros-Filho and Beverley, 1996). It is a critical metabolic enzyme and validated drug target for P. falciparum (Ivanetich and Santi, 1990) and studies with Pf TS-DHFR have shown that the junctional linker connecting the TS and DHFR domains is essential for catalytic activity (Shallom et al., 1999). Similar experiments in Cryptosporidium are technically difficult due to challenges in genetic manipulations of the organism and inability to continuously maintain parasites in cell culture. The genetic knockout experiments have not been carried out in T. gondii.

The presence of substrate channeling in L. major, P. falciparum and T. gondii and absence in C. hominis may be due to the distinct host cell environments where each parasite has evolved to survive. The de novo folate biosynthesis pathway is present in both T. gondii and P. falciparum; however, the genes for folate biosynthesis are absent in the genome of Cryptosporidium. While Toxoplasma and Plasmodium can both carry out de novo synthesis and salvage folate, Cryptosporidium only has the ability to salvage folate from the host (Hyde, 2008; Hyde et al., 2008).

Funding

This work was supported by National Institute of Allergy and Infectious Diseases at National Institute of Health, AI 083146.