The Molecular Basis of Sulfonamide Resistance in Toxoplasma gondii and Implications for the Clinical Management of Toxoplasmosis

Abstract

Polymerase chain reaction amplification and DNA sequencing of the Toxoplasma gondii dihydropteroate synthase gene (dhps) identified 4 alleles among parasite populations from 32 cases of human toxoplasmosis. Heterologous expression and enzyme assay reveal that 3 of these alleles encode sulfadiazine (Sdz)-sensitive enzymes. The fourth, generating a highly Sdz-resistant enzyme, differs from 1 of the other 3 at only a single residue (407) of dhps. Of interest, a fifth allele, found in a laboratory-induced Sdz-resistant line, also differs from another of these 3 drug-sensitive forms by the same single mutation that affects residue 407 of dhps. Significantly, residues corresponding to dhps-407 are implicated in sulfonamide resistance in other microorganisms. The human-derived allelic form encoding the Sdz-resistant enzyme was found in T. gondii associated with a fatal infection, and its presence within clinical material may have implications for sulfonamide use, particularly in cases of toxoplasmosis in which the initial response to drug treatment is poor.

Toxoplasma gondii, an obligate, intracellular protozoan parasite, is an increasingly important agent of human disease. It has aworldwide distribution and can infect allwarm-blooded animals [1, 2]. Infection of an immunocompetent human host is usually asymptomatic but leaves the subject with a life-long latent infection in the form of quiescent tissue cysts within which the parasite apparently continues to replicate very slowly. These cysts can release viable parasites that can reinitiate an active infection [3]. In an immunosuppressed host, this reactivation of a previously latent infection can have severe consequences and can often be fatal. Toxoplasmosis is a significant problem in congenitally infected and pharmacologically immunosuppressed patients, but its greatest impact is in late AIDS, in which up to 25% of patients will develop toxoplasmic encephalitis (TE) [4].

Prophylaxis and treatment of T. gondii infection is often by combination chemotherapy with sulfadiazine (Sdz) and pyrimethamine (PM), which act to provide a synergistic blockade to the folate biosynthetic pathway by inhibiting dihydropteroate synthase (dhps) and dihydrofolate reductase (dhfr), respectively [2, 5]. Reduced folate cofactors are essential for a number of biochemical processes, including the formation of DNA. T. gondii, like many microorganisms, synthesizes folates de novo because, unlike its mammalian host, it is thought to be unable to use preformed dietary folates. The enzymes of this pathway are therefore attractive targets for antimicrobial chemotherapy.

The response to treatment in TE is generally good, with 90% of patients responding by day 14. However, 10%of patients with verified TE (via biopsy or necropsy) never respond to treatment, and 10%-20% of responders relapse during long-term PM-Sdz maintenance therapy [3, 6, 7]. Although there are currently no published descriptions of naturally acquired antifolate resistance in T. gondii, the above data indicate that populations of this type may exist. Moreover, long-term drug treatment has the potential to contribute to clinical failure by selecting drug-resistant parasite variants.

Notwithstanding the absence of any reported incidence of resistant genotypes within natural populations of T. gondii, parasite lines showing laboratory-induced resistance have been described elsewhere [8, 9]. Growth of one of these, termed R-Sul^R-5, has been shown to be 300 times more resistant to sulfonamides than the wild-type RH strain [9]. Among unicellular organisms, a common mechanism of resistance to antifolate inhibitors is variation of the gene sequences encoding the drug target. This is well exemplified by the case of the closely related apicomplexan parasite Plasmodium falciparum, for which resistance to PM and sulfadoxine is a consequence of alterations in the sequences of the dhfr and dhps genes, respectively. Accumulated single base changes within these genes can increase drug resistance by several orders of magnitude [10–12]. Similarly, evidence from the AIDS-related eukaryotic pathogen Pneumocystis carinii, in which point mutations within the dhps gene are also observed [13, 14], is also consistent with the view that prolonged use of antifolates for prophylaxis and treatment can lead to acquired resistance.

It also has been demonstrated directly that alterations of the T. gondiidhfr gene at residues analogous to those associated with PM resistance in P. falciparum confer comparable levels of resistance when the modified gene is transfected into a sensitive strain of T. gondii [15–17]. T. gondii can thus become highly resistant to antifolates by structural variation of the enzymes that are targeted by such inhibitors. It is therefore a reasonable hypothesis that this situation may also occur naturally in T. gondii. The previous characterization of both the dhfr and dhps genes of T. gondii [18, 19] provides reference data with which sequences of these genes from clinical isolates of Toxoplasma species can be compared. The primary aims of this study were to establish whether allelic variants of dhfr and dhps are found in human cases of toxoplasmosis and whether any such variations affect antifolate sensitivity and thus possibly influence clinical outcome.

Materials and Methods

Clinical specimens. Specimens from patients were obtained and submitted, together with all appropriate clinical information, to the Toxoplasma Reference Unit, Public Health Laboratory Service (Swansea, UK) for routine reference laboratory investigation by polymerase chain reaction (PCR). Detailed molecular analyses of dhfr and dhps genotypes were subsequently carried out on clinical specimens or isolates (grown in mice or tissue culture) within the Department of Biomolecular Sciences, University of Manchester Institute of Science and Technology (Manchester,UK). Five samples were of nonhuman origin, and 32 were from cases of human toxoplasmosis: 8 TE cases in patients with human immunodeficiency virus-AIDS (1 from brain biopsy tissue, 3 from cerebrospinal fluid [CSF], and 4 from peripheral blood-derived tissue cultures); 20 patients with congenital toxoplasmosis (11 from placenta/amniotic fluid and 9 from fetal/neonatal tissues); 3 immunosuppressed patients (1 each fromcardiacmuscle, CSF, and lung tissue); and 1 immunocompetent patient (lymph node derived).

PCR amplification and sequence analysis of clinical samples. We have developed a PCRmethod based on that described by Panaccio et al. [20] that allows for direct amplification of products from a wide range of material (including tissue culture, CSF, amniotic fluids, and solid tissue samples) without the need for template purification. Reactions (100 µL) containing 8µL dNTP stock solution (final concentration each dNTP, 250mM; Roche), 300 ng of each of the appropriate outer PCR primer pairs (table 1), 10 µL of 10 × buffer (final Mg2⁺ concentration, 2.0 mM), 2.5 U Taq polymerase (Roche), and 18% formamide were amplified over 40 cycles in a Perkin-Elmer Gene-Amp PCR system 2400, using the following cycling conditions: 30 s at 84°C for denaturation, 30 s at 40°C for annealing, and 90 s at 60°C for extension. Cycle 1 was preceded by an additional 3-min denaturation step at 94°C. Cycle 40 was followed by a 10-min incubation at 60°C. Five microliters of this primary PCR then was used as template in a second round of amplification, using the corresponding inner primers (table 1) and the reaction conditions described above. Aliquots (25 µL) of each secondary PCR were resolved by agarose gel electrophoresis (2% wt/vol Nusieve GTG agarose; Flowgen Instruments), and the PCR products were purified for sequencing using a PCR purification kit (Qiaquick; Qiagen). DNA sequencing was performed using AmpliTaqFS BigDye terminator chemistry and analyzed on a model 377 sequencer (both from PE Applied Biosystems). In this way, the entire coding sequence of both the dhfr and dhps domains from the parasites was characterized.

Constructs for expression in Escherichia coli. Specific oligonucleotide primers (table 1), incorporating 5′ Nde I and 3′ Xho I restriction sites to facilitate cloning into the expression vector pET-15b (Novagen), were designed to amplify the full-length, type A (RH) hydroxymethyldihydropterin pyrophosphokinase-dihydropteroate synthase (pppk-dhps) cDNA sequence. When cloned in this way, expression from pET-15b inserts a vector-encoded hexahistidine tag at the N-terminus of the mature PPPK-dhps protein, which facilitates protein purification.

The PCR fragment was purified using a PCR purification kit (Qiaquick; Qiagen) and cut with Nde I and Xho I. The cut fragment was purified in the same way and cloned into the pET-15b expression vector, and the construct sequence was verified. The nucleotide polymorphisms seen among parasite samples studied were introduced into the type A sequence, using commercially available kits (Chameleon or QuikChange XL from Stratagene and Gene-editor from Promega), and the constructs were sequence verified.

Expression and purification of induced protein. Expression conditions were optimized with regard to time of induction, temperature, growth medium, and oxygen availability (data not shown). The expression constructs were routinely transformed into E. coli BL21 (DE3) pLysS cells and grown on Luria broth agar plates with appropriate antibiotic selection. A single colony was grown overnight at 37°C in Luria broth, again with antibiotic selection. The overnight culture was diluted 1:20 and grown at 30°C to an optical density of 0.4 at 600 nm. A 1-µL uninduced control sample was removed, and the culture was then induced with 0.1 mM isopropyl b- D-galactopyranoside. The culture was grown for a further 4 h before the cells were pelleted by centrifugation at 1664 g at 4°C for 5 min. The soluble fraction, which contained 20%–25%of the induced protein, was then purified using Bugbuster (Novagen), according to the manufacturer's recommended protocol. The resulting supernatant was further purified by metal affinity chromatography, using Ni- NTAagarose (Qiagen), according to themanufacturer's recommended protocol. The purified protein then was stored in 50 mM sodium phosphate (pH 8.0) and 300 mM NaCl at 4°C for immediate use. For long-term storage, 10 mM dithiothreitol and 10% glycerol were added to the protein sample, which was then rapidly frozen in liquid nitrogen and stored indefinitely at 280°C. Protein concentration was determined by use of the Bio-Rad protein assay, and protein purity was determined visually by SDS-polyacrylamide gel electrophoresis analysis of the samples in a Bio-Rad Mini Protean II apparatus, using a discontinuous buffer system [21].

Preparation of DHPS substrate. 6-Hydroxymethylpterin diphosphate lithium salt (Schircks Laboratories) was first fully oxidized by solution in 0.4 M potassium ascorbate/KOH (pH 5.0) and then reduced to 6-hydroxymethyl-7,8-dihydropterin diphosphate by treatment with excess sodium dithionite [22]. Conversion from oxidized to reduced form was monitored by the shift of the absorbance peak from340 nm(oxidized sample) to 315 nm (reduced sample). The reduced substrate was stored in aliquots under nitrogen at280°C in the dark. A fresh aliquot was used for each experiment.

Enzyme susceptibility to sulfa drugs. An enzyme assay measuring the synthesis of [¹⁴C]dihydropteroate from [¹⁴C]p-aminobenzoic acid (PABA) and 6-hydroxymethyl-7,8-dihydropterin diphosphate was developed on the basis of the assay described by Walter and Konigk [23]. A standard assay reaction (100 µL) consisted of 1 mM [14C]PABA (specific activity, 58 mCi/mmol; concentration, 0.1 mCi/ µL; Moravek), 15 mM 6-hydroxymethyl-7,8-dihydropterin diphosphate, 50 mMTris-HCl (pH 8.0), 5 mMMgCl2, 5 mMdithiothreitol, 250 ng (36 nM) of purified PPPK-dhps, and various concentrations of Sdz (0–10mM). The reaction mixture was incubated at 37°C for 15 min, after which the reaction was terminated by the addition of 20 mM EDTA (pH 8.0) and heating at 95°C for4min. After cooling on ice for 10 min, the reactions were mixed thoroughly, and triplicate 20-µL aliquotswere spotted onto polyethyleneimine cellulose-coated Polygram sheets (Machery-Nagel) and developed with ascending chromatography in 0.2 M LiCl. Under these conditions, neither dihydropteroate nor its oxidized product, pteroate, moved from the origin, whereas PABA migrated with an Rf value of 0.3. Radioactivity on the chromatograms was quantitated by use of a Typhoon scanner (AP Biotech) and ImageQuant software (Molecular Dynamics). Comparative studies among different sulfa drugs were carried out as described above except that, for reasons of limited solubility, stocks were made up in dimethyl sulfoxide (DMSO) and assays contained a final concentration of 10% DMSO.

Results

Sequence analysis. We previously reported the isolation and characterization of the dhps gene from T. gondii. We showed that, as in the case of P. falciparum, the dhps sequence forms part of a bifunctional gene in which the previous activity in the Sequence analysis. We previously reported the isolation and characterization of the dhps gene fromT. gondii.We showed that, as in the case of P. falciparum, the dhps sequence forms part of a bifunctional gene in which the previous activity in the

Direct automated sequencing of the PCR-amplified fragments, covering the 6 exons of dhps and the 3 exons of dhfr, confirmed that all fragments were of T. gondii origin and allowed rapid construction of a database of complete coding sequences for each of these 2 domains from 37 parasite samples. Analysis of these data identified several sequence polymorphisms within dhps and a single nucleotide polymorphism within the dhfr sequences (table 2). This latter change does not affect the amino acid sequence encoded by the gene and thus cannot alter the drug sensitivity of the dhfr protein it encodes. In contrast, among the 5 polymorphic nucleotide positions (affecting codons 474, 560, 580, 597, and 627) initially identified within the dhps gene, 3 alter the encoded amino acid. Moreover, all the sequences determined fell into 1 of 3 groups, each carrying a particular combination of nucleotides at these positions. Each of these 3 groups effectively represents a different allelic variant of the parasite. One of these putative alleles (type A) is typical of the RH or wildtype parasites. The second (type B) shows specific variation at 4 of the 5 polymorphic loci, and the third (type C) carries these same alterations together with a specific change at the fifth variable locus. Each of these putative alleles encodes a slightly different protein.

Analysis of the laboratory-induced, sulfamethoxazole-resistant strain R-Sul^R-5 revealed the presence of an additional single nucleotide polymorphism superimposed on the RH (type A) sequence seen in the clinical isolates (table 2). This polymorphism alters amino acid residue 407, which is equivalent to one of those functionally linked to sulfa resistance in the related organism P. falciparum [24] and strongly implicated in sulfa resistance in P. carinii [25]. These data thus provide the first demonstration that sulfa resistance in T. gondii can be associated with nucleotide polymorphism in dhps.

We have also identified this same resistance-associated polymorphism in 1 of our clinical samples (Swa-20), in which it is seen in addition to the 5 changes that characterize the type C allelic sequence (table 2). Because the laboratory-induced strain and the clinical isolate that showthe 407 alteration have different allelic backgrounds for both dhps and dhfr (table 2), this result cannot be explained by cross-contamination. We therefore believe that this represents the first description of a putative sulfaresistant T. gondiidhps domain from a clinical isolate and have sought to test whether it does indeed encode a protein with significant sulfa resistance.

Heterologous expression in E. coli and purification of the expressed protein. Using specific PCR primers, we were able to amplify the full-length RHpppk-dhps cDNA sequence and clone it into the pET-15b expression vector. After sequence verification, the gene was expressed in E. coli BL21 (DE3) pLysS cells by induction of T7 polymerase. The PPPK-dhps fusion protein thus produced is 664 aa in length, with a predicted Mr of 75,137. Figure 1 shows the results of typical expression experiments and identifies a protein of the expected size, which is not present in the uninduced, control sample.All 5 variant protein sequences were similarly obtained, and significant purification was achieved via binding of the His-tag incorporated at their N-termini, giving a routine recovery of ~25 mg/L purified protein in all cases.

Sensitivity of expressed proteins to sulfa drugs. We have developed an assay for dhps activity and characterized the heterologously expressed enzyme preparations with respect to their sulfonamide susceptibilities. After expression and metal affinity chromatography purification, 250-ng aliquots of each protein were assayed in the presence of various concentrations of Sdz (0–10 mM). Sdz was used in these experiments because it is the principal sulfonamide used in the treatment and prophylaxis of toxoplasmosis. In the presence of various concentrations of Sdz, the activity of each protein was determined as a percentage of the control activity measured in the absence of drug. The results of these assays are shown in figure 2.

The proteins expressed from the different alleles (A, B, and C) have almost identical Sdz sensitivities (figure 2A), although there are significant differences in their specific activities under these experimental conditions (data not shown). Each of these drug-sensitive proteins has an IC50 value for Sdz of ~160 µM. However, the single mutation found in the A allele background of the dhps gene of the laboratory-induced sulfa-resistant strain (R-Sul^R-5) does indeed confer significant drug resistance on the protein it encodes, shifting its Sdz IC50 value from ~160 µM to ~5mM(figure 2B). These values are in good agreementwith those previously determined for the equivalent enzymes in crude T. gondii extracts [9] and presumably account for the much increased in vivo resistance of this parasite line, as noted by the same authors. Of importance, our current data also showthat this same mutation confers a similar level of drug resistance to the protein expressed from the bifunctional pppk-dhps gene carrying the dhps domain identified in the Swa-20 clinical isolate, where it is combined with the 5 additional mutations that define the C allele. In this case, the IC50 value for Sdz is also ~5 mM (figure 2B).

Given the significantly increased IC50 value for Sdz of the Swa-20 enzyme, it was of interest to investigate the response of this enzyme to related, clinically useful inhibitors. The results (table 3) show that there is significant cross-resistance to all of the inhibitors tested. Although the IC50 value for dapsone against the mutant enzyme is some 10-fold lower than the values found for the other inhibitors, it should be noted that this drug showed a similarly increased efficacy against the wild-type enzyme. Thus, none of these sulfa drugs is very effective against the Swa-20 enzyme and, of importance, the measured IC50 value exceeds the maximum attainable serum level of the inhibitor in all cases.

Discussion

Sequence analysis of 37 parasite samples, including isolates from patients with AIDS, congenitally infected patients, and transplant recipients, has identified 5 nucleotide polymorphisms that can associate to give 3 different Sdz-sensitive alleles of the pppk-dhps gene in natural populations of T. gondii. Clinical samples can either have no (allele A, such as wild-type RH), 4 (allele B), or 5 (allele C) polymorphisms within the dhps domain. The IC50 values for all 3 of the proteins encoded by these alleles are ~160 µM.

We have also fully sequenced the dhps gene from R-Sul^R-5, a previously characterized, sulfamethoxazole-resistant, laboratory induced isolate of T. gondii [9]. Here, we identified an additional polymorphism within dhps affecting residue 407. This polymorphismis at an analogous position to those observed in related organisms, including P. falciparum [11, 12] and P. carinii [13, 25], which previously have been causally associated with sulfonamide drug resistance. Drug-inhibition assays of the altered protein confirm previous observations that growth of R-Sul^R-5 is sulfonamide resistant [9]; we determined the IC50 value to be ~5 mM. This single dhps mutation, which presumably accounts for the altered properties of the enzyme it encodes [9], is thus strongly implicated as being the causal agent of the sulfonamide drug resistance of this isolate.

Critical to the present study is our observation of this same resistance-associated polymorphism in one of our clinical samples (Swa-20). In addition to the drug-sensitive, type C allelic sequence, Swa-20 also carries the single polymorphism affecting codon 407 found in the sulfamethoxazole-resistant strain (R-Sul^R-5). The expressed protein (pSwa-20) showed almost identical levels of resistance to Sdz as that from R-Sul^R-5, with an IC50 value determined to be ~5 mM, and significant crossresistance to other sulfa drugs. These results substantiate our view that the polymorphisms affecting codons 474, 560, and 597 have no association with sulfa resistance, but that, by contrast, the alteration in codon 407 confers substantial resistance to a range of related inhibitors and that the resistance observed when all 4 amino acid changes are present in dhps is due entirely to the Asn!Asp change at position 407.

One key question raised by our findings is whether, in the case of the Sdz-resistant clinical isolate (Swa-20), resistance was acquired after infection occurred or was already present prior to this event. The latter case would suggest the existence of resistant strains of T. gondii in likely sources of human infection, such as domestic cats or livestock [26]. Review of the medical records for the Swa-20 case confirmed this to be a congenital infection, with documented seroconversion of the mother during the first trimester. No sulfonamides were used in the treatment of the mother during pregnancy, thus suggesting infection by a strain of the parasite already carrying sulfonamide resistance.

This finding, of the potential for infection of humanswith sulfonamide-resistant T. gondii, may necessitate a review of standard treatment protocols, particularly in cases in which sulfonamides are used and the clinical response to anti-T. gondii therapy is poor. In such cases, analysis of the dhps sequence may provide useful insight into the cause of such poor response. To better understand the risk and potential clinical impact of infection with sulfonamide-resistant strains of T. gondii, we are currently investigating whether such resistant strains can be identified in likely sources of human infection.

Acknowledgments

We gratefully acknowledge Janet Francis and Katy Roberts (Toxoplasma Reference Unit, Swansea) and Peter Winstanley and Simon Szwandt (University of Liverpool), for collecting parasite samples; David Roos (University of Pennsylvania), for supplying strain R-Sul^R-5; Jeremy Derrick (University of Manchester Institute of Science and Technology [UMIST]), for useful discussions; and David Gardner (UMIST), for help with DNA sequencing.

References