Catalysis and Sulfa Drug Resistance in Dihydropteroate Synthase

See allHide authors and affiliations

Science  02 Mar 2012:
Vol. 335, Issue 6072, pp. 1110-1114
DOI: 10.1126/science.1214641


The sulfonamide antibiotics inhibit dihydropteroate synthase (DHPS), a key enzyme in the folate pathway of bacteria and primitive eukaryotes. However, resistance mutations have severely compromised the usefulness of these drugs. We report structural, computational, and mutagenesis studies on the catalytic and resistance mechanisms of DHPS. By performing the enzyme-catalyzed reaction in crystalline DHPS, we have structurally characterized key intermediates along the reaction pathway. Results support an SN1 reaction mechanism via formation of a novel cationic pterin intermediate. We also show that two conserved loops generate a substructure during catalysis that creates a specific binding pocket for p-aminobenzoic acid, one of the two DHPS substrates. This substructure, together with the pterin-binding pocket, explains the roles of the conserved active-site residues and reveals how sulfonamide resistance arises.

Drug resistance has led to a decrease in the clinical utility of virtually all marketed antibacterial agents (1), and the sulfonamide class of antibiotics (sulfa drugs) was an early victim of this phenomenon (2, 3). Sulfa drugs interrupt the essential folate pathway in bacteria and primitive eukaryotes; they target the enzyme dihydropteroate synthase (DHPS), which catalyzes the condensation of 6-hydroxymethyl-7,8-dihydropterin pyrophosphate (DHPP) with p-aminobenzoic acid (PABA) in the production of the folate intermediate, 7,8-dihydropteroate (4). Sulfa drugs target both Gram-positive and Gram-negative bacterial infections, and combination therapies such as co-trimoxazole—a mixture of the sulfa drug sulfamethoxazole (SMX) and the dihydrofolate reductase (DHFR) inhibitor trimethoprim—are effective against many pathogenic microorganisms (5). However, DHPS mutations have been frequently characterized in many clinical isolates, relegating sulfonamide-based therapies to second- or third-line options.

Co-trimoxazole has proven to be effective against several emerging threats, including community-acquired multidrug-resistant Staphylococcus aureus (MRSA) (6, 7) and Pneumocystis jiroveci infections in immune-compromised patients (8). DHPS therefore remains an important drug target, and we are developing new inhibitors that target the DHPP-binding pocket of the enzyme (911). Understanding the DHPS catalytic mechanism and the mechanistic basis of sulfa drug resistance is crucial for these drug discovery efforts. DHPS has a TIM barrel α/β structure, and many of the drug resistance point mutations are located within two flexible and conserved loops that appear to make important contributions to the active site (9, 1115). The inability to observe these loops in their catalytic and/or substrate-bound conformations in the available crystal structures has hampered efforts to understand the structural basis of catalysis and sulfa drug resistance.

DHPS catalyzes the formation of a bond between the amino nitrogen of PABA and the C9 carbon of DHPP, with pyrophosphate as the leaving group. It has been suggested that the reaction proceeds by an SN2-like mechanism (14), but the NH2 group of PABA is a poor nucleophile, and crystal structures of DHPS with the substrate analog 6-hydroxymethyl pterin pyrophosphate (PtPP) and the product analog pteroate suggest that the required attack geometry is sterically disfavored (9). To visualize DHPP and PABA in the active site, we soaked both compounds into Bacillus anthracis DHPS (BaDHPS) crystals (table S1) (16). We found that both molecules in the asymmetric unit (molecules A and B) performed catalysis, leaving the product 7,8-dihydropteroate bound at the active site (Fig. 1A). This was unexpected because these crystals are grown at pH 9.0 and in 1.4 M sulfate, conditions that would be expected to hinder catalysis. The costructure closely resembles the pteroate costructure (9), and both have a sulfate ion instead of the eliminated pyrophosphate of DHPP in the anion-binding pocket. A new feature is the partial ordering of loop 2 in molecule A that packs onto the PABA moiety, which is now sandwiched between loop 2 and Lys220. This is the first indication that one role of loop 2 is to stabilize the binding of PABA at the active site.

Fig. 1

Products generated by crystalline BaDHPS. (A) The product 7,8-dihydropteroate after soaking crystals in DHPP, PABA, and Mg2+. (B) DHP+ and PHBA after soaking crystals in DHPP, PHBA, and Mg2+. (C) DHP+ after soaking crystals in DHPP and Mg2+. In each figure, loop 1 is shown in magenta, loop 2 in green, and the N terminus of helix αloop7 in teal; a sulfate ion (yellow/red) occupies the anion-binding pocket. The surrounding protein is shown in light gray; carbon, oxygen, and nitrogen atoms are shown as black, red, and blue balls, respectively; covalent bonds are shown in yellow, and hydrogen bonds are shown with red dashes. The FobsFcalc electron densities were generated from refined structures in which the indicated ligands were omitted and are contoured at 3σ. Below each panel is the structure of the key molecule bound in the complex. DHP+ is bound in both (B) and (C), but the structure is only shown in (C) with the pterin ring atoms numbered. W indicates an ordered water molecule in the pterin-binding pocket. Abbreviations for amino acids: D, Asp; H, His; I, Ile; K, Lys; N, Asn; R, Arg; S, Ser.

To investigate whether PABA is locked into place before product formation, we replaced PABA with p-hydroxybenzoic acid (PHBA), a less reactive PABA analog (16). The structure (table S1) showed that PHBA indeed binds in the same location as PABA, with a partially ordered loop 2 clamping it in place (Fig. 1B). The structure also revealed that DHPP had lost its pyrophosphate group in both molecules A and B, leaving the dihydropterin core in the pterin-binding pocket. There is no evidence of an OH group at C9 that would result from hydrolysis of an unstable carbocation. This structure confirms that the pyrophosphate is not removed by an SN2 nucleophilic attack but is eliminated in a manner consistent with an SN1 reaction.

We explored alternatives to an SN2 mechanism by performing quantum chemical modeling of the initial step of a “pure” SN1 reaction: the cleavage of the C9–O bond of DHPP in the absence of PABA (figs. S1 and S2) (16). Three key results resulted from these computational analyses: (i) The barrier to bond breaking is only ~24 kcal mol−1; (ii) the essential Mg2+ ion (17) adds the leaving pyrophosphate α-oxygen to its coordination shell, thereby acting as a Lewis acid and assisting pyrophosphate elimination; and (iii) the carbocation formed at the C9 position is stabilized by charge delocalization into the pterin ring. This predicted scenario is analogous to the SN1 mechanism of the prenyltransferases (18, 19). Natural bond order analyses (20) before and after bond breaking indicate resonance stabilization of the carbocation that includes a partial iminium character of N8. We propose that this cationic intermediate, which we term DHP+, is the dihydropterin core species that we observe in the crystal structure.

The calculations suggest that DHPS can slowly release pyrophosphate from DHPP independent of PABA binding at the active site. To test this, we soaked BaDHPS crystals in DHPP without PABA (16), and the structure (table S1) revealed that pyrophosphate had indeed been released from DHPP (Fig. 1C). Loop 2 was completely disordered, which supports its role in helping to lock PABA onto the surface of Lys220. The calculations also support the essential role of the Mg2+ ion (17), and we confirmed this experimentally for BaDHPS (fig. S3A) (16). To visualize the effect of removing Mg2+, we presoaked crystals in EDTA to remove Mg2+, and then added EDTA and DHPP for a further 3 hours of soaking (16). The resulting structure (table S1) showed that pyrophosphate is still cleaved from DHPP but remains trapped in the anion-binding pocket, where it appears to stabilize the conformations of loop 1 and loop 2 (fig. S4). Therefore, Mg2+ is not absolutely required for the cleavage of pyrophosphate from DHPP but may play a key role in its release from the enzyme.

We also completed two crystal structures of Yersinia pestis DHPS (YpDHPS) (16): the apo structure (fig. S5A and table S2) and the complex with pteroate (fig. S5B and table S2). Although both structures closely resemble those of BaDHPS (9), two features particularly recommend them for mechanistic studies: Loop 1 and loop 2 are both unconstrained by crystal contacts in the apo structure and are free to adopt functional conformations, and the crystals are grown in more physiological conditions (pH 6 to 7 and 12% PEG 20,000).

Soaking DHPP and PABA into YpDHPS crystals gave a structure (table S2) that apparently shows the enzyme near the transition state (Fig. 2A, fig. S6A, and movie S1). In both molecules in the asymmetric unit, PABA, DHP+, pyrophosphate, and a Mg2+ ion are all present within a highly organized loop 1 and loop 2 substructure prior to product formation. The released pyrophosphate occupies a pocket comprising residues Ser32, Ser34, and Asp35 from loop 1 (the latter two residues via an ordered water molecule), Ser66 and Thr67 from loop 2, and Arg254 and His256 from the anion-binding site (we use BaDHPS numbering; the sequence alignment is shown in fig. S7). The Mg2+ ion is octahedrally coordinated with the two distal oxygen atoms of the pyrophosphate, the Oδ1 oxygen of Asn27, and three water molecules. The entire active site is covered by the distal end of loop 1 encompassing Pro30 to Gly37, which forms a β-ribbon structure. The component residues make a number of stabilizing interactions; Asp31 is clamped between Arg68 and Arg82, Ser32 interacts with the pyrophosphate as noted above, Ser34 interacts with Arg219, and Asp35 interacts with Asn27.

Fig. 2

The DHPS catalytic mechanism. (A) Stereo view of the Y. pestis (Yp) DHPS ordered active site generated from YpDHPS crystals soaked in DHPP, PABA, and Mg2+. Loop 1 (magenta), loop 2 (green), and an octahedrally coordinated Mg2+ ion (green ball) are all organized by the pyrophosphate (PPi; orange/red), and a PABA molecule is bound within a specific binding pocket. Note that loop 1 caps the active site via a distal β-ribbon substructure, and the distal Pro69 of loop 2 engages the bound PABA. The residue numbering corresponds to BaDHPS. (B) The proposed SN1 chemical reaction catalyzed by DHPS. Pyrophosphate is first removed from DHPP. The resulting cationic intermediate species can adopt the DHP+ resonance forms shown in the square brackets in which the positive charge is delocalized into the pterin rings and stabilized by the pterin-binding pocket. The amine group of PABA finally attacks DHP+ at the C9 carbon atom to generate the product 7,8-dihydropteroate. Abbreviations for amino acids: C, Cys; D, Asp; F, Phe; G, Gly; H, His; I, Ile; K, Lys; N, Asn; P, Pro; R, Arg; S, Ser; T, Thr.

The YpDHPS complex structure (Fig. 2A) explains three key features of the catalytic mechanism and the active site. First, it explains that an essential role of Mg2+ is to order the loop 1–loop 2 substructure, as well as to stabilize the leaving pyrophosphate. The conformations and locations of the active-site residues, the bound substrates, and the Mg2+ ion closely match those of the computed state in which the C–O bond has been broken (fig. S2D, RMSD = 0.77 Å) (16). Second, it explains why the residues within the two loops are so highly conserved. Finally, it explains why the PABA-binding site has been so difficult to visualize: It is only fully formed in this complex. Phe33 from loop 1, Pro69 from loop 2, and Lys220 and Phe189 are all highly conserved, and they combine to form the PABA-binding pocket. Also, loop 1 forms a protective lid over the active site with a restricted entrance that matches the shape and chemistry of PABA. Consistent with our previous results (9), the carboxylate moiety of PABA is accommodated by Ser221 and the helix dipole of helix αloop7.

Our data show that the role of the pterin-binding pocket is to first bind DHPP and then promote the release of pyrophosphate by stabilizing a carbocation on the C9 carbon. The likely roles of the conserved Asp184 and Asp101 are to stabilize resonance forms that move the positive charge away from this primary carbocation and toward the N3/2-amino and N8 nitrogen atoms, respectively, either by ionic interactions or by proton abstraction (16). The electrophilic DHP+ intermediate can then react with the incoming PABA nitrogen via nucleophilic conjugate addition. We showed experimentally (16) that the release of pyrophosphate, and presumably the dissolution of the intermediate-state substructure, is promoted by PABA (fig. S3B). The SN1 mechanism that we propose is shown in Fig. 2B.

We prepared eight active-site BaDHPS mutants to test this proposal (16). The kinetic parameters were measured using an assay that monitors pyrophosphate release (16), and the results are summarized in Table 1. Parallel assays were performed in the presence of high (200 μM) PABA and high (50 μM) DHPP to allow independent measurements of the binding affinities of the two substrates. In general, the mutations support the proposed mechanism and the intermediate-state substructure shown in Fig. 2A. Loop 1 mutations Asn27 → Ala (N27A), Phe33 → Ala (F33A), Phe33 → Leu (F33L), and Asp35 → Ala (D35A) and loop 2 mutation Ser66 → Ala (S66A) all primarily affected the binding of PABA and had little effect on the binding of DHPP. Each of these residues, directly or indirectly, contributes to the PABA-binding site. The pterin-pocket mutants, Asp101 → Asn (D101N) and Asp184 → Asn (D184N), were designed to investigate the effect of removing these negative charges. D184N was unable to bind DHPP because the asparagine side chain adopts a rotamer that prevents its interaction with the pterin ring (table S1 and fig. S8). However, D101N showed efficient binding to both DHPP and PABA but had a much reduced kobs. Finally, Lys220 → Gln (K220Q) showed reduced binding to PABA and DHPP, consistent with the role of Lys220 in binding both substrates.

Table 1

Kinetic parameters of wild-type and mutant BaDHPS. kobs, observed turnover number; Km, Michaelis constant; ND, activity too low to detect.

View this table:

To investigate the mechanism of the sulfa drugs, we soaked BaDHPS crystals in DHPP and sulfathiazole (STZ) and determined the structure (table S1) (16). Molecule A revealed a clear DHP-STZ product (Fig. 3A) similar to that of the normal product (Fig. 1A), whereas molecule B revealed the bound drug before complex formation, similar to the PHBA complex (Fig. 1B). This finding is consistent with previous studies showing that sulfa drugs can replace PABA as DHPS substrates (21, 22). We then soaked DHPP and SMX into YpDHPS crystals and obtained a structure (Fig. 3B, fig. S6B, table S2, and movie S2) similar to that observed with DHPP and PABA (Fig. 2A) (16). In molecules A and B, DHPP is bound and loop 1 and loop 2 are ordered by pyrophosphate and an octahedrally coordinated Mg2+ ion, but only molecule B contains SMX in the PABA-binding pocket. A difference from the YpDHPS PABA/DHPP structure (Fig. 2A) is that pyrophosphate remains attached to DHPP, but this has little effect on the loop conformations because both the location and Mg2+ coordination of the pyrophosphate are unaffected.

Fig. 3

Sulfa drug mechanism and resistance. (A) B. anthracis (Ba) DHPS crystals soaked in DHPP and STZ reveal a covalent DHP-STZ adduct bound at the active site and stabilized by an ordered loop 2. The boxed labeled residues are sites of sulfa drug resistance, and two major sites are labeled in boxed orange. The molecular envelope (light gray) encompasses the binding pockets for pterin, PABA, and the anion (PPi or sulfate). Note that the thiazole ring of STZ extends outside this pocket directly adjacent to Pro69. The structure of DHP-STZ is shown below. (B) Stereo view of the Y. pestis (Yp) DHPS active site occupied by DHPP, sulfamethoxazole (SMX), and an octahedrally coordinated Mg2+ ion. The structure is very similar to that shown in Fig. 2A with PABA in place of SMX. One difference from the PABA structure is that DHPP is intact with the pyrophosphate covalently attached. The residue numbering corresponds to BaDHPS. The structures of DHPP and SMX are shown below. In both figures, loop 1 is shown in magenta, loop 2 in green, and the N terminus of helix αloop7 in teal. Abbreviations for amino acids: A, Ala; D, Asp; F, Phe; G, Gly; I, Ile; K, Lys; N, Asn; P, Pro; R, Arg; S, Ser; T, Thr.

The YpDHPS sulfa drug complex reveals the drug binding site. SMX perfectly fits the PABA-binding pocket, with the negatively charged oxygen atoms of the sulfonyl group matching the PABA carboxyl group and their common phenyl groups engaging the same hydrophobic pocket in the loop 1–loop 2 substructure. Figure 3B shows that the common sites of resistance are all clustered around this substructure. Phe33, Thr67, and Pro69 are frequently observed sites of resistance mutations (14) and form key elements of the PABA-binding site. It has been observed that resistance is typically associated with regions of a drug that extend beyond the substrate envelope (23). Figure 3A reveals that the thiazole and methoxazole rings of STZ and SMX, which have no counterparts in PABA, are positioned outside the DHPS substrate envelope and are located such that mutations at Phe33 and Pro69 can impede sulfa drug binding.

Supporting Online Material

Materials and Methods

Figs. S1 to S8

Tables S1 and S2

References (2440)

Movies S1 and S2

References and Notes

  1. See supporting material on Science Online.
  2. Acknowledgments: We thank R. DuBois, C. Pemble, S. Gajewski, and D. Miller for assistance with the YpDHPS structure determination; J. Bollinger and E. Enemark for technical assistance; D. Hammoudeh and C. Rock for helpful discussions; and staff at SERCAT for assistance with synchrotron data collection. Supported by NIH grant AI070721 (S.W.W. and R.E.L.), NIH Cancer Center (CORE) support grant CA21765, and ALSAC (the American Lebanese Syrian Associated Charities). The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH. Data were collected at Southeast Regional Collaborative Access Team (SER-CAT) 22-ID and 22-BM beamlines at the Advanced Photon Source, Argonne National Laboratory. Supporting institutions may be found at Use of the Advanced Photon Source was supported by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences, under contract W‐31‐109‐Eng‐38. Coordinates and structure factors for all the structures described have been deposited in the Protein Data Bank, and the PDB accession codes are listed in tables S1 and S2.
View Abstract

Stay Connected to Science

Navigate This Article