Structural analysis of Pneumocystis carinii and human DHFR complexes with NADPH and a series of five potent 6-[5′-(ω-carboxyalkoxy)benzyl]pyrido[2,3-d]pyrimidine derivatives

These data reveal that the ethyl esters of a series of 5′-(ω-carboxyalkoxy)benzylpyrido[2,3-d]pyrimidines have flexible side-chain conformations that do not optimize interactions with Arg75 in pcDHFR. Also, a novel conformation not reported in other DHFR structures was observed for one conformer of one of the compounds that had a disordered side chain.

Abstract

Structural data are reported for five antifolates, namely 2,4-diamino-6-[5′-(5-carboxypentyloxy)-2′-methoxybenzyl]-5-methylpyrido[2,3-d]pyrimidine, (1), and the 5′-[3-(ethoxycarbonyl)propoxy]-, (2), 5′-[3-(ethoxycarbonyl)butoxy]-, (3), 5′-[3-(ethoxycarbonyl)pentyloxy]-, (4), and 5′-benzyloxy-, (5), derivatives, which are potent and selective for Pneumocystis carinii dihydrofolate reductase (pcDHFR). Crystal structures are reported for their ternary complexes with NADPH and pcDHFR refined to between 1.4 and 2.0 Å resolution and for that of 3 with human DHFR (hDHFR) to 1.8 Å resolution. These data reveal that the carboxylate of the ω-carboxyalkoxy side chain of 1, the most potent inhibitor in this series, forms ionic interactions with the conserved Arg75 in the substrate-binding pocket of pcDHFR, whereas the less potent ethyl esters of 2–4 bind with variable side-chain conformations. The benzyloxy side chain of 5 makes no contact with Arg75 and is the least active inhibitor in this series. These structural results suggest that the weaker binding of this series compared with that of their pyrimidine homologs in part arises from the flexibility observed in their side-chain conformations, which do not optimize intermolecular contact to Arg75. Structural data for the binding of 3 to both hDHFR and pcDHFR reveals that the inhibitor binds in two different conformations, one similar to each of the two conformations observed for the parent pyrido[2,3-d]pyrimidine, piritrexim (PTX), bound to hDHFR. The structure of the pcDHFR complex of 4 reveals disorder in the side-chain orientation; one orientation has the ω-carboxy­alkoxy side chain positioned in the folate-binding pocket similar to the others in this series, while the second orientation occupies a new site near the nicotinamide ring of NADPH. This alternate binding site has not been observed in other DHFR structures. Structural data for the pcDHFR complex of 5 show that its benzyl side chain forms intermolecular van der Waals interactions with Phe69 in the binding pocket that could account for its enhanced binding selectivity compared with the other analogs in this series.

1. Introduction

Opportunistic infections by Pneumocystis pathogens are still a leading cause of mortality and morbidity among patients with AIDS or other immunosuppressed conditions (Thomas & Limper, 2004 ▶). The primary treatment for Pneumocystis pneumonia (PcP) combines a sulfonamide drug with tri­methoprim (TMP; Fig. 1 ▶), which targets the folate pathway. TMP has been widely used as an antibiotic as it is a highly selective competitive inhibitor of bacterial dihydrofolate reductase (DHFR) and is currently used in the treatment of PcP (Rosowsky et al., 2003 ▶), although its use in combination therapy with a sulfonamide that targets the dihydropteroate pathway is not always successful. TMP has limited efficacy in PcP patients and can result in drug resistance (Nahimana et al., 2004 ▶). Therefore, there is still a need to develop more effective treatments.

Figure 1.

Schematic representation of the antifolates under study. 2,4-diamino-6-[5′-(5-carboxypentyloxy)-2′-methoxybenzyl]-5-methylpyrido[2,3-d]pyrimidine, (1); 2,4-diamino-6-{5-[3-(ethoxycarbonyl)propoxy]-2-methoxybenzyl}-5-methylpyrido[2,3-d]pyrimidine, (2); 2,4-diamino-6-{5-[3-(ethoxycarbonyl)butoxy]-2-methoxybenzyl}-5-methylpyrido[2,3-d]pyrimidine, (3); 2,4-diamino-6-{5-[3-(ethoxycarbonyl)pentyloxy]-2-methoxybenzyl}-5-methylpyrido[2,3-d]pyrimidine, (4); and 2,4-diamino-6-(5-benzyloxy-2-methoxybenzyl)-5-methylpyrido[2,3-d]pyrimidine, (5).

As part of a larger program to design lipophilic DHFR inhibitors that would contain structural features of both piritrexim (PTX) and TMP (Fig. 1 ▶), used in the treatment of Pneumocystis pneumonia (PcP), Rosowsky and coworkers have synthesized a number of compounds that incorporate the 2′,5′-dimethoxybenzyl substitution pattern of PTX with the diaminopyrimidine-5-(3′,4′,5′-trimethoxy­ben­zyl) pattern of TMP (Chan et al., 2005 ▶; Rosowsky et al., 2004 ▶). To further define inhibitor classes that combine the high potency of PTX with the high antiparasitic versus mammalian selectivity of TMP, a similar series of ω-carboxyalkoxypyrido[2,3-d]pyrimidines, both as carboxylates and their ethyl esters, were designed and tested (Chan et al., 2005 ▶). These data revealed that the ester derivatives were less potent than the carboxy­late parent compounds as measured by their IC₅₀ values using recombinant protein (Table 1 ▶).

Table 1. Enzyme inhibition against DHFR for selected inhibitors (Fig. 1 ▶; Chan et al., 2005 ▶; Rosowsky et al., 2004 ▶).

Inhibitor	IC50, pcDHFR (nM)	IC50, rat DHFR (nM)	Rat DHFR/pcDHFR selectivity ratio
TMP	13000	180000	13.8
PTX	13	3.3	0.26
PY957	0.049	3.9	80
PY1011	1.0	5000	5000
1	1.3	2.0	1.5
2	6.5	4.2	0.65
3	28	14	0.50
4	35	21	0.60
5	96	170	1.8

Activity data for the series of diaminopyrimidines that have the TMP template revealed that the optimal length of the ω-carboxy­alkoxy side chain was for the carboxybutyloxybenzyl side chain of PY957 and the carboxy-1-pentyloxybenzyl side chain of PY1011 (Fig. 1 ▶), which had selectivity ratios of 80 and 5000, respectively, for rat DHFR/pcDHFR IC₅₀ (Rosowsky et al., 2004 ▶; Table 1 ▶). The sequence of rat DHFR has 89% identity and 96% similarity to that of human DHFR. None of the differences involve active-site residues. Structural data for PY1011 and PY957 revealed that the inhibitor carboxylate forms a strong salt bridge to the con­served Arg75 in the pcDHFR active site, similar to that ob­served for the folate sub­strate or methotrexate inhibitor complexes (Cody et al., 1999 ▶, 2006 ▶). In this series of pyrido[2,3-d]pyrimidines selectivity was uniformly reduced, with the most selective of these five inhibitors being 1, which contains a 5′-carboxy­butyloxybenzyl side chain (Fig. 1 ▶). Compound 3 also has a butyloxybenzyl side chain but is an ethyl ester and has a selectivity ratio about half that of compound 1 (Table 1 ▶).

In order to validate the binding mode of these pyrido[2,3-d]pyrimidines, we report structural data for ternary complexes of pcDHFR with NADPH and a series of five ω-carboxy­alkoxy inhibitors (Fig. 1 ▶) that explore the effects of side-chain length and ester formation on pcDHFR binding. Additionally, the crystal structure of 3 is reported as a ternary complex with human DHFR (hDHFR).

2. Methods

2.1. Crystallization and X-ray data collection

Recombinant pcDHFR was cloned, expressed and purified as described previously (Delves et al., 1993 ▶). The protein was washed in a Centricon-10 three times with 10 mM MES buffer pH 6.0, 100 mM KCl and concentrated to 15.5–21.7 mg ml⁻¹ pcDHFR for the five complexes. The protein was incubated with the antifolates (1–5; Fig. 1 ▶) and NADPH prior to crystallization using the hanging-drop vapor-diffusion method at 277 K. The protein droplets for the pcDHFR complexes contained 33–36% PEG 2K, 46–52 mM MES pH 6 with 100 mM KCl. Cryogenic solutions for the pcDHFR crystals were prepared using mother liquor with a 16–24% gradient of ethylene glycol.

Recombinant hDHFR was expressed in pDS5 vector in Escherichia coli BL21 (DE3) cells and purified as described previously (Cody et al., 2009 ▶). The protein was concentrated to 6.9 mg ml⁻¹ and incubated with 3 and NADPH prior to crystallization using the hanging-drop vapor-diffusion method at 287 K. The protein droplets contained 100 mM potassium phosphate pH 6.9 and 30% ammonium sulfate. The reservoir consisted of 100 mM potassium phosphate pH 6.9, 65% ammonium sulfate and 3%(v/v) ethanol. The hDHFR complex crystal was cryoprotected with Paratone-N oil (Hampton Research, Aliso Viejo, California, USA).

Data for the pcDHFR and hDHFR inhibitor complexes were collected to 1.4–2.0 Å resolution on beamlines 11-1 or 9-­2 at the Stanford Synchrotron Resource Laboratory (SSRL) using the remote-access protocol (McPhillips et al., 2002 ▶; Cohen et al., 2002 ▶; González et al., 2008 ▶). All data were processed in HKL-2000 using DENZO and were scaled with SCALA (Otwinowski & Minor, 1997 ▶). The unit-cell parameters and diffraction statistics ▶ for these ternary complexes are listed in Table 2.

Table 2. Crystal properties and refinement statistics for inhibitors 1–5 bound to pcDHFR and hDHFR as NADPH ternary complexes.

Values in parentheses are for the highest resolution shells.

	pcDHFR–1	pcDHFR–2	pcDHFR–3	pcDHFR–4	pcDHFR–5	hDHFR–3
PDB code	3nz6	3nz9	3nza	3nzb	3nzc	3nzd
Space group	P21	P21	P21	P21	P21	H3
Unit-cell parameters
a (Å)	37.1	36.9	36.9	36.9	36.9	84.6
b (Å)	42.8	42.7	42.7	42.7	42.7	84.6
c (Å)	60.8	60.1	60.6	60.6	59.9	77.6
β (°)	94.4	94.8	94.6	94.6	94.8
Beamline	SSRL 11-1	SSRL 11-1	SSRL 11-1	SSRL 11-1	SSRL 11-1	SSRL 9-2
Resolution	1.80 (1.86–1.80)	1.60 (1.66–1.60)	1.70 (1.76–1.70)	1.30 (1.37–1.30)	1.80 (1.86–1.80)	1.80 (1.90–1.80)
Wavelength (Å)	0.975	0.975	0.975	0.975	0.975	0.975
Rmerge	0.070 (0.58)	0.028 (0.33)	0.070 (0.38)	0.046 (0.46)	0.060 (0.34)	0.062 (0.34)
Completeness (%)	96.3 (98.8)	96.6 (90.3)	97.8 (98.0)	87.5 (47.3)	95.8 (83.8)	95.3 (74.2)
Observed reflections	17172	46806	21070	154924	17462	67454
Unique reflections	12344	24011	19449	44341	16720	18315
〈I/σ(I)〉	14.4 (0.58)	13.1 (5.0)	14.4 (4.4)	12.8 (1.7)	4.9 (0.5)	15.0 (3.3)
Multiplicity	3.7 (1.2)	3.6 (3.2)	2.5 (2.0)	3.4 (2.1)	3.5 (3.0)	3.7 (3.2)
Resolution range	32.7–2.0 (2.1–2.0)	24.5–1.6 (1.7–1.6)	27.8–1.8 (1.9–1.8)	24.7–1.4 (1.5–1.4)	32.5–1.9 (2.0–1.9)	53.3–1.8 (1.9–1.8)
Reflections used	11724	16107	13884	30117	11845	17366
R factor	0.21	0.20	0.21	0.24	0.21	0.16
Rfree	0.28	0.27	0.28	0.29	0.27	0.23
Total protein/ligand atoms	1931	1879	1893	1892	1782	1878
Total waters	166	114	116	73	48	296
Average B factor (Å2)	32.0	30.0	29.3	25.4	41.3	19.6
Error in Luzzati plot	0.29	0.21	0.23	0.21	0.25	0.17
R.m.s. deviation from ideal
Bond lengths (Å)	0.019	0.023	0.022	0.022	0.022	0.025
Bond angles (°)	2.11	2.34	2.00	2.02	1.98	2.23
Ramachandran plot (%)
Most favored	91.2	94.6	92.1	93.6	89.2	96.7
Additional	5.4	3.4	4.0	3.9	7.8	2.7
Disallowed	3.4	1.5	4.0	2.5	2.9	0.5
Missing density†	1–4, 83–89	1, 4, 83–89	1–4, 83–89	1–4, 83–89	1–4, 83–89

^†Residues without interpretable density and not refined.

2.2. Structure determination and refinement

All structures were solved by molecular-replacement methods with MOLREP (Vagin & Teplyakov, 2010 ▶; Collaborative Computational Project, Number 4, 1994 ▶) using the coordinates for pcDHFR (PDB entry 3cd2; Cody et al., 1999 ▶) and hDHFR (PDB entry 1u72; Cody et al., 2005 ▶). Inspection of the resulting difference electron-density maps was made using the program Coot (Emsley & Cowtan, 2004 ▶) running on a Mac G5 workstation and revealed density for ternary complexes in all cases. To monitor the refinement, a random subset of all reflections was set aside for calculation of R _free (5%). The models for the antifolates were generated from the crystal structure of PY957 (Cody et al., 2006 ▶) and optimized with SYBYL (Tripos). The parameter files for the cofactor and inhibitors were prepared using the Dundee PRODRG2 server website (http://davapc1.bioch.dundee.ac.uk/programs/prodrg; Schüttelkopf & van Aalten, 2004 ▶). Refinement was carried out using the program REFMAC5 from the CCP4 suite of programs (Collaborative Computational Project, Number 4, 1994 ▶). Between least-squares minimizations, the structures were manually adjusted to fit difference electron density. The Ramachandran conformational parameters generated by RAMPAGE (Lovell et al., 2002 ▶) for the final models from the last cycle of refinement are listed in Table 2 ▶. Coordinates for these structures have been deposited in the Protein Data Bank and their PDB codes are listed in Table 2 ▶. Superimpositions were computed using the same structure as reference molecule with the SSM function in Coot. Figures were prepared using the modeling program PyMOL (DeLano, 2006 ▶).

3. Results

Inspection of the difference electron-density maps for pcDHFR with the five pyrido[2,3-d]pyrimidine inhibitors (1–5; Fig. 1 ▶) revealed ternary complexes with NADPH in all cases. The lack of clearly interpretable electron density for loop 83–89 in each of these pcDHFR complexes indicates that this region is highly mobile and represents alternate conformational states compared with the structures previously reported in other pcDHFR complexes (Cody et al., 1999 ▶; Cody & Schwalbe, 2006 ▶). Similarly, there was poor electron density for the first four N-terminal residues. These residues were not included in the refinement.

As previously described (Cody et al., 2005 ▶, 2006 ▶; Cody & Schwalbe, 2006 ▶), the side chain of the conserved Arg75 in pcDHFR is held in place by a network of hydrogen bonds to the conserved Thr38 and Thr39 and the backbone functional groups of Lys68 and, through water, to the backbone carbonyl of Phe69. There are also cis-peptide linkages between Arg67 and Pro68 and Gly124 and Gly125, as observed in other pcDHFR complexes. The interactions of the 2,4-diaminopyrido[2,3-d]pyrimidine ring of these antifolates preserve the overall pattern of contacts with invariant residues in the active site. The hydrogen-bond network involving structural water, the conserved residues Thr144, Trp27 and Glu32 and the N1 nitrogen and 2-amino group of the pyrido[2,3-d]pyrimidine ring is also maintained. Similarly, the inhibitor 4-­amino group maintains its contacts with the conserved residues Ile9 and Tyr129 and NADPH. In all these structures, the cofactor NADPH is bound in an extended conformation similar to other pcDHFR or hDHFR cofactor complexes (Cody & Schwalbe, 2006 ▶).

The pyrido[2,3-d]pyrimidine inhibitor 3 is the ethyl ester homologue of the potent TMP analogue PY957 (Fig. 1 ▶), as both inhibitors have a 5′-carboxybutyl side chain. In this series of pyrido[2,3-d]pyrimidines, only inhibitor 1, which has a carboxy­pentyloxybenzyl side chain, is a carboxylate analogue similar to the potent pyrimidine analogs reported previously (Rosowsky et al., 2004 ▶). Structural data for the binding of inhibitor 1 to pcDHFR reveal that despite the difference in size of the pyrimidine ring of the TMP template versus that of the pyrido[2,3-d]pyrimidine ring, the 5′-carboxypentyloxy­benzyl side chain of 1 adjusts its conformation to form a salt bridge with Arg75 similar to that observed for the potent inhibitors PY957 and PY1011 (Cody et al., 2006 ▶) (Fig. 2 ▶). Torsion angles that describe the bridging geometry between ring systems, defined in 1 (Fig. 1 ▶), reveal that the conformation of these inhibitors is similar (Table 3 ▶). In the case of inhibitor 3, the ethyl ester homologue of PY957, the side chain moves away from Arg75, similar to that of inhibitor 2 (Fig. 3 ▶).

Figure 2.

Stereo superposition of 1 (pink) with PY957 (grey) and PY1011 (cyan) in the structures of their ternary complexes with pcDHFR and NADPH (Cody et al., 2006 ▶). The active-site residue Arg75 is shown.

Table 3. Conformational parameters for methylene bridge of pyrimidine and pyrido[2,3-d]pyrimidine inhibitors (Fig. 1 ▶).

The torsion angles used are defined in 1 (Fig. 1 ▶).

	C5—C6—C7—C1′ (°)	C6—C7—C1′—C2′ (°)	C4—C4a—C5 (°)
pcDHFR–1	153.4	−93.9	123.8
pcDHFR–2	151.7	−95.7	125.6
pcDHFR–3	169.2	−96.0	124.8
pcDHFR–4	−148.2	−114.3	126.3
pcDHFR–4	−174.2	85.4	126.3
pcDHFR–5	174.8	−94.1	129.1
pcDHFR–PY957†	179.0	91.0	—
pcDHFR–PY1011†	177.2	74.1	—
hDHFR–3	115.6	105.7	125.3
hDHFR-L22R–PTX‡	145.1	−89.5	127.0
hDHFR–PTX‡	123.5	108.8	129.9

^†Cody et al. (2006 ▶).

^‡Lewis et al. (1995 ▶).

Figure 3.

Stereo superposition of inhibitors 1 (green), 2 (cyan) and 3 (violet) in their ternary complexes with pcDHFR and NADPH. The active-site residues Arg75 and Phe69 are shown.

Structural data for inhibitor 2, which has the shortest side chain (5′-carbethoxypropyloxybenzyl), shows that its conformation is similar to that of 1; however, there is limited interaction with Arg75 as the ethyl ester forces the side chain to move away from Arg75. Similar interactions were observed for the ternary complex of pcDHFR with 3, which has a 5′-­carbethoxy­butyloxybenzyl side chain (Fig. 3 ▶). Analysis of the structure of the hDHFR ternary complex with 3 reveals that the conformation of the inhibitor differs from that observed in the pcDHFR complex (Fig. 4 ▶), as is shown by the bridging torsion angles (Table 3 ▶). In the hDHFR complex the ethyl ester carbonyl forms a strong hydrogen bond contact (2.8 Å) to the amine of the Asn64 side chain. In the pcDHFR complex the corresponding residue is Phe69, which forms close van der Waals contacts to the methylene C atoms of the inhibitor side chain.

Figure 4.

Comparison of the binding of 3 in the pcDHFR–NADPH ternary complex (violet) and the hDHFR–NADPH ternary complex (blue). Also shown are residues Arg75 and Phe69 in pcDHFR and Arg70 and Asn64 in hDHFR.

Of the five structures determined for this series, 4 is the most interesting in that analysis of the difference electron-density map shows that the 5-carbethoxypentyloxy­benzyl side chain occupies two positions: one conformer is bound in the same way as the other inhibitors in this series, while the second conformer occupies a position observed for the first time binding in a pocket near NADPH (Fig. 5 ▶). Based on the electron-density profile, each side chain was refined with 50% occupancy.

Figure 5.

View of the 2F _o − F _c difference electron-density map (0.5σ, blue) showing two conformations for 4, each refined at half occupancy, in the ternary complex with pcDHFR and NADPH. The green density, contoured at 3σ, is from an F _o − F _c OMIT map calculated without contributions from 4. Key residues in the active site are labeled for orientation.

Inhibitor 5 has a 2′-O-benzyl ether side chain and therefore is not able to interact with Arg75; however, there are good hydrophobic contacts between the benzyl side chain of 5 and Phe69 that enhance the binding to pcDHFR (Fig. 6 ▶).

Figure 6.

Stereo superposition of the binding of 5 (grey) with that of 3 (violet), highlighting the hydrophobic interactions of the benzyl ether with Phe69, in the ternary complexes with pcDHFR and NADPH.

4. Discussion

This structural study of a series of O-(ω-carboxyalkoxy) analogs of PTX (1–5; Fig. 1 ▶) that have variable length side chains with and without ethyl esters revealed that the carboxybutyl carboxylate side chain of 1 was optimal for interaction with the conserved Arg75 in the active site of pcDHFR. The presence of the ethyl ester causes the side chain to move away from Arg75 to make hydrophobic interactions with Phe69 in the pcDHFR active site. The replacement of Phe69 by Asn64 in hDHFR results in a conformational change in the binding of 3 compared with that to pcDHFR as the carbonyl of the side chain of 3 forms a hydrogen bond to Asn64 (Fig. 4 ▶). In addition, there is a conformational change in the bridging torsion angle of 3 between the two structures (Table 3 ▶). These changes are consistent with the greater potency of 3 for mammalian DHFR compared with pcDHFR (Table 1 ▶).

The most unexpected results were observed for inhibitor 4, which showed that the carbethoxypentyloxybenzyl side chain binds in two alternate conformations, one of which is near the nicotinamide ring of NADPH, a region not previously occupied by a ligand. The position of the second conformer of 4 also differs significantly from the other examples of alternative binding conformers observed in the crystal structure of PY957 bound to the Q35S/N64F double mutant of hDHFR (Cody et al., 2009 ▶; Fig. 7 ▶). Examination of this region of pcDHFR reveals the presence of a pocket below the nicotinamide ring that is open to the surface (Fig. 8 ▶). However, this pocket is not present in the hDHFR structure. These differences suggest that changes in the surface profile of the DHFR can influence the binding environment and could be of importance in understanding structure–activity profiles for ligand-binding efficacy.

Figure 7.

Stereoview of the binding modes of the two alternate conformations of 4 (yellow and violet) in the structure of the pcDHFR–NADPH ternary complex (violet) compared with that of PY957 in the Q35S/N64F double mutant of hDHFR (green). The position of the conserved Arg75 that normally binds the carboxylate of the inhibitor is shown.

Figure 8.

van der Waals surface of the ternary complex of pcDHFR with 4 and NADPH showing the positions of the two alternate conformations of 4 (yellow and pink). The surface profile for pcDHFR shows a pocket that extends toward the nicotinamide ribose ring of NADPH that is not present in hDHFR (surface not shown).

Analysis of the DHFR binding of the parent pyrido[2,3-d]pyrimidine PTX also revealed that alternate conformers were observed for its binding to hDHFR (Lewis et al., 1995 ▶; Cody & Schwalbe, 2006 ▶). Comparison of these PTX structures with the carboxyalkyoxy pyrido[2,3-d]pyrimidine analogs reported here shows a clustering of the conformers observed for these structures. The conformation of 3 in hDHFR is similar to that observed for PTX in wild-type hDHFR, while that of 1 in pcDHFR is similar to that of PTX in the L22R mutant of hDHFR (Table 3 ▶; Fig. 9 ▶). The crystal structure of PTX in complex with pcDHFR has been reported previously (Champness et al., 1994 ▶); however, no coordinates are available and thus no comparisons could be made with the structures of the hDHFR–PTX complexes. Taken together, these structural results suggest that the clustering observed for the conformational parameters of PTX and the pyrido[2,3-d]pyrimidine analogs 1–5, although small, could be the result of subtle changes in the interactions of specific residues in the active site with the ligand. Therefore, these patterns could be of importance when computing preferential binding geometries in homology models.

Figure 9.

Comparison of the binding of PTX in hDHFR–NADPH (gold) (a) with 3 (blue) in pcDHFR and (b) with 1 (green) and PTX (grey) in the L22R mutant of hDHFR (Lewis et al., 1995 ▶). Conformational angles about the methylene bridge for these inhibitors are shown in Table 3 ▶.

In summary, these data show that the pyrido[2,3-d]pyrimidine O-(ω-carboxyalkoxybenzyl) scaffold of this series of DHFR inhibitors results in weaker potency and selectivity compared with the homologous analogs in the pyrimidine series (Chan et al., 2005 ▶; Rosowsky et al., 2004 ▶; Table 1 ▶). Structural data for the variable-length acids and esters of the carbalkyoxybenzyl analogs shows that the ethyl esters are not able to optimize the binding interactions within the p-aminobenzyoylglutamate binding pocket and that their side-chain orientations were highly variable among the analogs studied. This variability in binding that does not utilize strong interactions with the conserved Arg75 observed for folate analogs is consistent with the weaker binding observed for this series.

Supplementary Material

PDB reference: pcDHFR–1, 3nz6

PDB reference: pcDHFR–2, 3nz9

PDB reference: pcDHFR–3, 3nza

PDB reference: pcDHFR–4, 3nzb

PDB reference: pcDHFR–5, 3nzc

PDB reference: hDHFR–3, 3nzd