Search

Transfer of drug resistance genes to hematopoietic cells is being studied as a means to protect against the myelosuppression associated with cancer chemotherapy and as a strategy for the in vivo selection and amplification of genetically modified cells. The goal of this study was to test if retroviral-mediated gene transfer of a dihydrofolate reductase (DHFR) variant (L22Y) could be used for in vivo selection of transduced myeloid cells and to determine what proportion of transduced cells was required for protection from myelosuppression. Based on previous work suggesting that selection with antifolates may also require inhibition of nucleoside transport mechanisms, mice transplanted with DHFR-transduced bone marrow cells were treated with trimetrexate and the nucleoside transport inhibitor prodrug nitrobenzylmercaptopurine riboside phosphate. In vivo selection of transduced myeloid progenitors was seen in the bone marrow and in circulating mature peripheral blood cells following drug treatment. These results show that the novel combination of the L22Y-DHFR cDNA, trimetrexate and nitrobenzylmercaptopurine riboside phosphate can be used to select for transduced myeloid cells, and that this approach warrants further study in large animal models. A bicistronic vector containing a human CD24 reporter gene was used to determine the number of modified cells needed for chemoprotection. Partial protection from neutropenia was seen when greater than 10% of myeloid cells expressed the vector, and high levels of protection were obtained when the proportion exceeded 30%. These results suggest that gene transfer may be useful for myeloprotection in certain pediatric cancers, but that more efficient gene transfer will be required to apply this approach to adult cancer patients.

Mammalian cells cultured in the presence of the chemotherapeutic agent, methotrexate, develop resistance to this drug. Sometimes this is due to mutations in the gene for dihydrofolate reductase, the primary target of methotrexate. However, it has not been possible to link such polymorphism to resistance of neoplastic disease to therapy with methotrexate. Nevertheless, interest in this possibility lead to the introduction of many mutations into the cDNA for human DHFR by mutagenesis. Most of the corresponding enzyme variants have been expressed in Escherichia coli and characterized. Many mutations in codons for hydrophobic residues at the active site greatly decrease inhibition by methotrexate, and by the related substrate analogue, trimetrexate, while allowing the retention of considerable catalytic efficiency. Introduction of some of these mutants into mammalian cells by retroviral transfer provides substantial protection from toxic effects of the inhibitors, and has promise for the myeloprotection of patients receiving therapy with methotrexate or trimetrexate. Another potential use is in therapy for inherited disorders of hematopoiesis, where genetic modification of enough cells is a perennial problem. After transplantation of bone marrow that has been transduced with a bicistronic vector encoding both the mutant DHFR and a therapeutic gene, subsequent administration of methotrexate or trimetrexate should permit selection and enrichment of genetically modified hematopoietic cells. Hum Mutat 11:259–263, 1998. © 1998 Wiley-Liss, Inc.

Retroviral transduction of antifolate-resistant variants of human dihydrofolate reductase (hDHFR) into cells can increase their resistance to the cytotoxic effects of these drugs. We evaluated the ability of wild-type hDHFR and 20 mutant enzymes (13 with single-amino acid substitutions, 7 with two substitutions) to prevent growth inhibition in antifolate-treated CCRF-CEM cells. The wild-type enzyme and all of the variants significantly protected transduced cells from trimetrexate (TMTX)-induced growth inhibition. However, only half of the variants conferred more protection than does the wild-type enzyme. For the variants tested, the observed protective effect was higher for TMTX than for methotrexate (or =7.5-fold increased resistance), piritrexim (or =16-fold), and edatrexate (negligible). Transduction of the variants L22Y-F31S and L22Y-F31R led to the greatest protection against TMTX (approximately 200-fold). Protection from loss of cell viability was similar to protection from growth inhibition. The protection associated with a particular mutant hDHFR did not result from the level of expression: Efficient protection resulted from low affinity of the variant for antifolates, reasonable catalytic activity, and good thermal stability. Clones isolated from a polyclonal population of transduced cells varied by as much as 30-fold in their resistance to TMTX, the resistance differences depending on hDHFR expression levels.

Antifolates such as methotrexate (MTX) and trimetrexate (TMTX) are widely used in the treatment of cancer and nonmalignant disorders. Transient, yet sometimes severe myelosuppression is an important limitation to the use of these drugs. It has previously been shown that clonogenic myeloid progenitors and colony-forming units-spleen are resistant to antifolates, suggesting that myelotoxicity occurs late in hematopoietic development. The goal of this study was to define the mechanisms by which primitive hematopoietic cells resist the toxic effects of antifolate drugs. To test the hypothesis that myeloid progenitors may salvage extracellular nucleotide precursors to resist TMTX toxicity, a defined liquid culture system was developed to measure TMTX toxicity in expanding progenitor populations. These in vitro experiments showed that both human and murine progenitors can resist TMTX toxicity by importing thymidine and hypoxanthine from the serum. As predicted from these findings, several drugs that block thymidine transport sensitized progenitors to TMTX in vitro, although to differing degrees. These nucleoside transport inhibitors were used to test whether progenitors and hematopoietic stem cells (HSCs) could be sensitized to TMTX in vivo. Treatment of mice with TMTX and nitrobenzylmercaptopurineriboside phosphate (NBMPR-P), a potent transport inhibitor, caused significant depletions of clonogenic progenitors within the bone marrow (20-fold) and spleen (6-fold). Furthermore, NBMPR-P administration dramatically sensitized HSCs to TMTX, with dual-treated mice showing a greater than 90% reduction in bone marrow repopulating activity. These studies demonstrate that both myeloid progenitor cells and HSCs resist TMTX by using nucleotide salvage mechanisms and that these pathways can be pharmacologically blocked in vivo using nucleoside transport inhibitors. These results have important implications regarding the use of transport inhibitors for cancer therapy and for using variants of dihydrofolate reductase for in vivo selection of genetically modified HSCs.

The novel furopyrimidine N-(4-{N-[(2,4-diaminofuro[2,3-d]pyrimidin-5-yl)methyl]methylamino}benzoyl)-L- glutamate (MTXO), a classical antifolate with antitumor activity comparable to that of methotrexate (MTX), has been studied as inhibitor-cofactor ternary crystal complexes with wild-type Pneumocystis carinii (pc) and recombinant human wild-type dihydrofolate reductase (hDHFR). These structural data provide the first direct comparison of the binding interactions of the same antifolate inhibitor in the active site for pc and human DHFR. The human ternary DHFR complex crystallizes in the rhombohedral space group R3 and is isomorphous to the ternary complex reported for a gamma-tetrazole methotrexate analogue, MTXT. The pcDHFR complex crystallizes in the monoclinic space group P2(1) and is isomorphous to that reported for a trimethoprim (TMP) complex. Interpretation of difference Fourier electron-density maps for these ternary complexes revealed that MTXO binds with its 2,4-diaminofuropyrimidine ring interacting with Glu32 in pc and Glu30 in human DHFR, as observed for MTXT. The presence of the 6-5 furopyrimidine ring instead of the 6-6 pteridine ring results in a different bridge conformation compared with that of MTXT. The bridge torsion angles for MTXO, i.e. C(4a)-C(5)-C(8)-N(9) and C(5)-C(8)-N(9)-C(1'), are -156.5/51.9 degrees and -162.6/51.8 degrees, respectively for h and pc, compared with -146.8/57.4 degrees for MTXT. In each case, the p-aminobenzoylglutamate conformation is similar to that observed for MTXT. In the pcDHFR complex, the active-site region is conserved and the additional 20 residues in the sequence compared with the human enzyme are located in external loop regions. There is a significant change in the nicotinamide ribose conformation of the cofactor which places the nicotinamide O atom close to the 4NH(2) group of MTXO (2.7 A), a shift not observed in hDHFR structures. As a consequence of this, there is a loss of a hydrogen bond between the nicotinamide carbonyl group and the backbone of Ala12 in pcDHFR. In the human ternary complexes, the cofactor NADPH is bound with a more extended conformation, and the nicotinamide O atom makes a 3.5 A contact with the 4NH(2) group of MTXO. Although the novel classical antifolate MTXO is not highly active against pcDHFR, there are correlations between its binding interactions consistent with its lower potency as an inhibitor of h and pcDHFR compared with MTX.

Class 1 aldehyde dehydrogenases (ALDH-1) function as drug resistance gene products by catalyzing the irreversible conversion of aldophosphamide, an active metabolite of cyclophosphamide, to an inert compound. Because the dose-limiting toxicity of cyclophosphamide is myelosuppression, retrovirus-mediated transfer of ALDH-1 to bone marrow cells has been proposed as a protective strategy. Here we show that expression of ALDH-1 vectors was problematic due to low levels of ALDH-1 mRNA accumulation. A number of vectors containing several different ALDH-1 cDNAs were introduced into a variety of different cell lines either by transfection or transduction. Detectable ALDH-1 protein and enzyme activity was only seen in one transfected cell clone. Cells transduced with ALDH-1 retroviral vectors had no detectable protein expression and very low levels of ALDH-1 mRNA. Analogous vectors containing other drug resistance cDNAs led to much higher levels of steady-state mRNA. The mRNA half-life from ALDH-1 vectors was less than 2 hr suggesting that vector-derived mRNAs were destabilized by ALDH-1 coding sequences. These results suggest that methods which increase the stability of ALDH-1 mRNAs will be important for increased drug resistance in retrovirally transduced hematopoietic cells.

Although substitution of tyrosine, phenylalanine, tryptophan, or arginine for leucine 22 in human dihydrofolate reductase greatly slows hydride transfer, there is little loss in overall activity (kcat) at pH 7.65 (except for the arginine 22 variant), but Km for dihydrofolate and NADPH are increased significantly. The greatest effect, decreased binding of methotrexate to the enzyme-NADPH complex by 740- to 28,000-fold due to a large increase in the rate of methotrexate dissociation, makes these variants suitable to act as selectable markers. Affinities for four other inhibitors are also greatly decreased. Binding of methotrexate to apoenzyme is decreased much less (decreases as much as 120-fold), binding of tetrahydrofolate is decreased as much as 23-fold, and binding of dihydrofolate is decreased little or increased. Crystal structures of ternary complexes of three of the variants show that the mutations cause little perturbation of the protein backbone, of side chains of other active site residues, or of bound inhibitor. The largest structural deviations occur in the ternary complex of the arginine variant at residues 21-27 and in the orientation of the methotrexate. Tyrosine 22 and arginine 22 relieve short contacts to methotrexate and NADPH by occupying low probability conformations, but this is unnecessary for phenylalanine 22 in the piritrexim complex.

Substitution of glycine or alanine for phenylalanine 31 in human dihydrofolate reductase produces variants that are inhibited less by methotrexate (MTX) than the previously reported serine variant. The 100 times decrease in MTX affinity for the glycine variant is due to slower binding, and to inability of the initial complex to isomerize to a nondissociating conformer. A polar group at position 31 is unnecessary for resistance, but residues larger than serine confer no resistance. The glycine variant best fulfills criteria for gene therapy: low Km for H2folate, high kcat, and good stability. Although kcat is unaltered by these mutations, the rate of hydride transfer is greatly decreased. Presteady-state measurements have enabled a complete catalytic scheme to be constructed for the glycine variant that predicts observed steady-state behavior. The crystal structures of inhibitor complexes of the serine, alanine, and glycine mutants and of the wild-type enzyme show that the mutations cause little perturbation of the protein backbone, of side chains of residues at the active site, or of the bound inhibitor. A molecule of bound water occupies the space vacated by the phenyl group.

Nuclear magnetic resonance (NMR) spectra for [2-amino,3-15N2]folate and [2-13C]folate complexed with human dihydrofolate reductase, and for complexes of similarly labeled dihydrofolate, show that the N-3 proton of bound folate or dihydrofolate exchanges slowly with solvent and that the bound substrates are in the imino-keto tautomeric form. Previously proposed schemes for substrate protonation that require bound substrate to be in the enolic tautomer are therefore unlikely. The NMR spectra for bound folate are unchanged by raising the pH from 7 to 9.5, whereas those for free folate show marked changes due to ionization for the N-3 proton. The fraction of bound folate with the N-3 proton ionized at pH 9.5 is therefore very small, and the rate constant for the dissociation of the ionized species must be at least 320 times faster than for the protonated species. Comparison of NMR spectra over the pH range 5 to 7 gives no indication of a change in ionization state of the Glu30 carboxyl group over this pH range. This raises doubts about whether the apparent pKa of approximately 6 that describes pH dependence of hydride transfer is due to ionization of this carboxyl group.

When 2-chloro-2'-deoxyadenosine 5'-triphosphate (CldATP) is incorporated into DNA by human polymerases alpha and beta (Hpol alpha, Hpol beta) the rate of chain extension decreases. In the present study primer extension has been quantitated by estimating the concentration of each successive oligonucleotide product at a series of time points. This has permitted calculation of pseudo-first-order rate constants for successive nucleotide additions to primer. By this method it has been shown that rate constants for CldATP addition are 79-100% of those for dATP in the case of Hpol alpha, and 26-153% with Hpol beta. The concentrations of CldATP for half maximum velocity is 0.6 microM for Hpol alpha, and 6 microM for Hpol beta, each about twice the value for dATP. Thus, CldATP is a good substrate for both enzymes but is more efficiently used by Hpol alpha. Addition of a single analogue residue by Hpol beta to any of seven primers decreases the rate constant for addition of the next nucleotide to 2-7% of that after dAMP addition and further extension is negligible. Consecutive additions of analogue residues by Hpol alpha progressively decrease the rate of subsequent extension, and after five consecutive additions extension virtually terminates. These effects probably make a major contribution to the cytotoxicity of chlorodeoxyadenosine and its therapeutic usefulness as an antileukemic agent.

Four spin-labeled inhibitors of dihydrofolate reductase (DHFR) have been synthesized, each of which has the 2,2,6,6-tetramethyl-1-piperidinyloxy (TEMPO) reporting group at a different distance from the 2,4-diaminopyrimidine moiety by which the inhibitors are anchored and oriented in the active site. Inhibitors in which the TEMPO group is attached by a short side chain are weakly bound to DHFR from bacteria (Streptococcus faecium and Lactobacillus casei), to the bovine enzyme and to recombinant human DHFR. However, binding is sufficiently tight, especially in the ternary complexes with NADPH, for recording of the EPR spectra of the bound ligands. The spectra indicate that when these inhibitors are bound to the enzyme the TEMPO group is highly immobilized with correlation time, tau c, 4-20ns. Inhibitors that have the reporter group attached to the glutamate moiety of methotrexate bind to all four DHFRs more tightly than the inhibitors with shorter side chains by factors of up to 10(6). However, in most complexes formed by the inhibitors with longer side chains immobilization of the TEMPO group is slight (tau c 0.2-4 ns). These results are in general agreement with predictions from X-ray crystallographic results including thermal factors but there are some unanticipated differences between some results for bacterial and eukaryotic enzymes. Three of the splin-labeled inhibitors would provide good probes for distance measurements in and around the active site of mammalian DHFR.

2-Chloro-2'-deoxyadenosine 5'-triphosphate (CldATP) was compared with dATP as a substrate for DNA synthesis by bacterial and viral DNA polymerases in vitro. Lengths of chain extension and DNA synthesis pause sites were determined by comparison with products generated by dideoxynucleotide sequencing methods on the same end-labeled primer/template duplex after high-resolution polyacrylamide gel electrophoresis. Reverse transcriptase (RT) from human immunodeficiency virus (HIV-1) and avian myeloblastosis virus (AMV) incorporated CldATP efficiently. DNA strand elongation continued past most chloroadenine (ClA) insertion sites but resulted in shorter chains than when dATP was inserted. Phage T4 DNA polymerase incorporated CldATP least efficiently; Klenow fragment of Escherichia coli DNA polymerase I and modified T7 DNA polymerase (Sequenase) showed intermediate ability to utilize the analogue. Incorporation of several consecutive ClA residues into the replicating strand dramatically reduced the ability of Sequenase, Klenow fragment, and T4 DNA polymerases to continue strand elongation. In the absence of the corresponding normal deoxyribonucleoside triphosphate during DNA synthesis, ClA was frequently misincorporated as thymine, cytosine, or guanine by both AMV RT and HIV-1 RT but rarely, if at all, by Klenow fragment, Sequenase, and T4 DNA polymerase. Except T4, for most DNA polymerases, CldATP at 10-20-fold molar excess over dATP was not a strong competitive inhibitor of dATP, as judged by the amount of strand extension and polymerase pause sites during DNA synthetic reactions. Our results indicate that the degree of strand extension in the presence of CldATP, the number and location of polymerase pause sites, and the amount of misincorporation of the analogue are both polymerase- and sequence-dependent.

The dissociation constants (pKa) for the pteridine ring system of dihydrofolate (H2folate) have been redetermined, and those for dihydrobiopterin (H2biopterin) have been determined. Determination of the pKa for N5 of H2folate is complicated by the low solubility and instability of H2folate at pH 2-4, and other complicating factors. The initial rate of absorbance change due to degradation is a maximum at pH 2.5, and the products depend on the oxygen concentration: under aerobic conditions, (p-aminobenzoyl)glutamic acid and 7,8-dihydropterin-6-carboxaldehyde are major products. H2Biopterin is much more soluble and more stable at low pH. For protonation of N5, the pKa is 2.56 +/- 0.01 for H2biopterin and 2.59 +/- 0.03 for H2folic acid. Spectrophotometric determination of the pKa for the N3-O4 amide group of H2folate is subject to serious errors when a wavelength between 220 and 235 nm is used. These errors arise from the pH-dependent absorbance of mercaptoethanol often present in the preparation. The amide group has a pKa of 10.41 +/- 0.04 in H2biopterin and 10.85 +/- 0.04 in H2folate. The redetermined value for the pKa of N5 of H2folate has implications for mechanistic models for dihydrofolate reductase, and revised kinetic constants have been calculated for one model.

The apoenzyme of wild-type (WT) dihydrofolate reductase (DHRF) from Escherichia coli exists in two conformational states, Et and Ew, which differ in affinity for NADPH and in kinetic competence. Dissociation constants for the binary complex of NADPH with the two conformers differ by over 100-fold (KDt = 0.17 microM, KDw = 22 microM). Rate constants governing the interconversion of conformers are small (t1/2 for Ew----Et = 71 s), and since Ew is not catalytically competent, this conversion is accompanied by an increase in catalytic velocity. The equilibrium proportion of Et in the absence of ligands is 63%, but binding of NADPH greatly increases this proportion, and t1/2 for conversion of Ew.NADPH to Et.NADPH is 30 s. This conformational equilibrium has also been examined in mutant enzyme in which aspartate 27 is replaced by asparagine (D27N E. coli DHFR). Although ASp27 is an active site residue, it does not interact directly with bound NADPH, and in the mutant the rate constant for NADPH binding to Et is unchanged as are the dissociation constants for NADPH complexes with Et or Ew. However, for mutant apoenzyme, the proportion of Et is decreased to 18% in the absence of ligands so that the overall KD for NADPH is increased (0.15 microM for WT E. coli DHFR, 0.68 microM for D27N E. coli DHFR). The lower proportion of Et is due to a decreased rate for Ew----Et (t1/2 = 221 s) and an increased rate for Et----Ew (t1/2 = 50 s versus 120 s for WT E. coli DHFR).

Extension of synthetic primers by purified human polymerase alpha (Hpol alpha) with the (+)-strand of M13mp18 DNA as template encounters numerous specific pause sites on the M13 template. Some of these are regions of template secondary structure, at others the template codes for incorporation of the same base in multiple consecutive positions, but at some the responsible feature in the sequence is not obvious. 2-Chloro-dATP (CldATP) substitutes efficiently for dATP in such chain extension, with 2-chloroadenine (ClA) incorporation into many positions coding for A. However, there are more sites where extension is interrupted than with all four normal nucleotide substrates, particularly (but not exclusively) at template secondary structure and sites of multiple consecutive ClA insertion. DNA synthesis from normal substrates by Hpol beta in this system shows less frequent and less marked pauses, but with CldATP substituted for dATP chain extension is limited because of marked slowing of extension at sites of multiple consecutive ClA insertion. With either polymerase, the rate of extension is decreased even more at such regions when bromo-dATP is used as substrate. Some misincorporation of ClA instead of G or T can occur at certain sites in absence of the corresponding normal substrate, but misincorporation as C is rare. CldATP is a very weak inhibitor of chain extension by Hpol alpha, but a somewhat better inhibitor of Hpol beta. These results may account in part for the inhibition of DNA synthesis in cells exposed to 2-chlorodeoxyadenosine or 2-bromodeoxyadenosine.

The state of protonation of substrates bound to mammalian dihydrofolate reductase (DHFR) has significance for the mechanism of catalysis. To investigate this, dihydrofolate and dihydropteroylpentaglutamate have been synthesized with {sup 15}N enrichment at N-5. {sup 15}N NMR studies have been performed on the binary complexes formed by bovine DHFR with these compounds and with (5-{sup 15}N)dihydrobiopterin. The results indicate that there is no protonation at N-5 in the binary complexes, and this was confirmed by {sup 13}C NMR studies with folate and dihydrofolate synthesized with {sup 13}C enrichment at C-6. The chemical shift displacements produced by complex formation are in the same direction as those which result from deprotonation of the N-3/C-4-O amide group and are consistent with at least partial loss of the proton from N-3. This would be possible if, as crystallographic data indicate, there is interaction of N-3 and the 2-amino group of the bound ligands with the carboxylate of the active site glutamate residue (Glu{sup 30}).

Association and dissociation rate constants obtained by stopped-flow spectroscopy have permitted definition of a kinetic scheme for recombinant human dihydrofolate reductase that correctly predicts full time course kinetics of the enzymatic reaction over a wide range of substrate and product concentrations. The scheme is complex compared with that for the bacterial enzyme and involves branched pathways. It successfully accounts for observed rapid hysteresis preceding steady state and for the nonhyperbolic dependence of steady-state rate on substrate and product concentrations. The major branch point in the catalytic cycle occurs at E.NADP.H4folate because either NADP or H4folate can dissociate from the ternary product complex (koff = 84 s-1 and 46 s-1, respectively). The rate of conversion of enzyme-bound substrates to products is very fast (k = 1360 s-1) and nearly unidirectional (Kequ = 37) so that other steps limit the catalytic rate. At saturating substrate concentrations these steps include release of NADP and H4folate from E.NADP.H4folate and release of products from the two abortive complexes E.NADPH.H4folate (koff = 225 s-1) and E.NADP.H4folate (koff = 4.6 s-1). Since NADP dissociates slowly from E.NADP.H2folate nearly 90% of the enzyme accumulates as this complex at steady state. Nonetheless, the catalytic rate is maintained at 12 s-1 by rapid flux of a small portion of the enzyme through an alternate branch. At physiological concentrations of substrates and products the steady-state rate is limited primarily by the rate of H2folate binding to E.NADPH so that the enzyme is extremely efficient.

Tumors resistant to chemotherapeutic oxazaphosphorines such as cyclophosphamide often overexpress aldehyde dehydrogenase (ALDH), some isozymes of which catalyze the oxidization of aldophosphamide, an intermediate of cyclophosphamide activation, with formation of inert carboxyphosphamide. Since resistance to oxazaphosphorines can be produced in mammalian cells by transfecting them with the gene for human ALDH isozyme 3 (hALDH3), it seems possible that patients receiving therapy for solid tumors with cyclophosphamide might be protected from myelosuppression by their prior transplantation with autologous bone marrow that has been transduced with a retroviral vector causing overexpression of hALDH3. We investigated whether retroviral introduction of hALDH3 into a human leukemia cell line confers resistance to oxazaphosphorines. This was examined in the polyclonal transduced population, that is, without selecting out high expression clones. hALDH3 activity was 0.016 IU/mg protein in the transduced cells (compared with 2x10(-5) IU/mg in untransduced cells), but there was no detectable resistance to aldophosphamide-generating compounds (mafosfamide or 4-hydroperoxycyclophosphamide). The lack of protection was due, in part, to low catalytic activity of hALDH3 towards aldophosphamide, since, with NAD as cofactor, the catalytic efficiency of homogeneous, recombinant hALDH3 for aldophosphamide oxidation was shown to be about seven times lower than that of recombinant hALDH1. The two polymorphic forms of hALDH3 had identical kinetics with either benzaldehyde or aldophosphamide as substrate. Results of initial velocity measurements were consistent with an ordered sequential mechanism for ALDH1 but not for hALDH3; a kinetic mechanism for the latter is proposed, and the corresponding rate equation is presented.

One of the main impediments to effective gene therapy of blood disorders is the resistance of human hematopoietic stem cells to stable genetic modification. We show here that a small minority of retrovirally transduced stem cells can be selectively enriched in vivo, which might be a way to circumvent this obstacle. We constructed two retroviral vectors containing an antifolate-resistant dihydrofolate reductase cDNA transcriptionally linked to a reporter gene. Mice were transplanted with transduced bone marrow cells and then treated with an antifolate-based regimen that kills unmodified stem cells. Drug treatment significantly increased the percentage of vector-expressing peripheral blood erythrocytes, platelets, granulocytes, and T and B lymphocytes. Secondary transplant experiments demonstrated that selection occurred at the level of hematopoietic stem cells. This system for in vivo stem-cell selection provides a means to increase the number of genetically modified cells after transplant, and may circumvent an substantial obstacle to successful gene therapy for human blood diseases.

Structural data for two independent crystal forms (monoclinic, C2, and orthorhombic, P2(1)2(1)2(1)) of the ternary complex of the potent antitumor agent PT523 [N alpha-(4-amino-4-deoxypteroyl)-N delta-hemiphthaloyl-L-ornithine], reduced nicotinamide adenine dinucleotide phosphate (NADPH), and recombinant human dihydrofolate reductase (hDHFR) reveals multiple binding orientations for the hemiphthaloyl group of the inhibitor. Analysis of these data shows that PT523 binds with its pteridine ring in the same orientation observed for methotrexate (MTX) analogues. However, in each structure, the hemiphthaloyl ring occupies three alternate conformations. In the C2 lattice, the phthaloyl moiety binds in two extended conformations, A and C, with each conformer having a 180 degrees flip of the o-carboxylate group, and a third, lower occupancy conformer B, with the phthaloyl group folded within contact of the active-site pocket. In the orthorhombic lattice, PT523 also has three conformers for the phthaloyl group; however, these differ from those observed in the monoclinic lattice. Two major conformers, A and C, are displaced on either side of the extended position observed in the C2 lattice, one near the folded B conformer of the C2 lattice and the other extended. These conformers form tighter intermolecular contacts than those in the C2 lattice. Conformer B is folded back away from the active site in a unique position. There are also significant differences in the conformation of the adenine-ribose moiety of NADPH in both complexes that differ from that observed for other inhibitor-NADPH-hDHFR ternary complexes. These data suggest that the added intermolecular contacts made by the hemiphaloyl group of PT523 contribute to its tighter binding to hDHFR than MTX, which does not extend as far from the active site and cannot make these contacts. These crystallographic observations of multiple conformations for the hemiphthaloyl group are in general agreement with solution NMR data for the binding of PT523 to hDHFR [Johnson et al. (1997) Biochemistry 36, 4399-4411], which show that the hemiphthaloyl group may adopt more than one conformation. However, the crystallographic data reveal more discretely occupied positions than can be interpreted from the solution data. These results suggest that crystal packing interactions may influence their stability.

Chemoresistance gene transfer is an experimental method to protect hematopoietic cells from the toxicity of anticancer drugs. Because multiple drugs are usually given together in cancer therapy, this strategy will ultimately require vectors expressing multiple chemoresistance genes. For this reason, we designed a bicistronic retroviral vector (HaMID) containing a modified human multidrug resistance-1 cDNA and a mutant human dihydrofolate reductase cDNA bearing a leucine to tyrosine substitution at codon 22 (L22Y). To determine if this vector would confer dual drug resistance to hematopoietic cells, recombinant retrovirus was used to transduce the human CEM T lymphoblastic cell line as well as primary murine myeloid progenitors. Growth suppression assays, using polyclonal transduced CEM cells, demonstrated increased resistance to taxol (13-fold), trimetrexate (8.9-fold), vinblastine (5.6-fold), methotrexate (2.5-fold), and etoposide (1.5-fold) when used as single agents. HaMID-transduced cells also grew at a logarithmic rate in the simultaneous presence of 25 nM taxol and 100 nM trimetrexate while control cells were entirely growth inhibited by this drug combination. Similarly, HaMID-transduced murine myeloid progenitors acquired increased resistance to taxol (2.9-fold) and trimetrexate (140-fold), and were able to form colonies in the simultaneous presence of both drugs. Our results suggest that retroviral transfer of HaMID into primary hematopoietic cells should reduce the myelosuppression associated with the combined use of antifolates and P-glycoprotein-effluxed drugs.

Dihydrofolate reductase (DHFR) catalyzes the reduction of dihydrofolate (H2folate) to tetrahydrofolate by NADPH, and this requires that the pteridine ring be protonated at N5. A long-standing puzzle has been how, at physiological pH, the enzyme can protonate N5 in view of its solution pKa of 2.6 and the fact that the only proton-donating group in the pterdine binding site, Asp-27, hydrogen bonds not to N5 but to the 2-amino group and N3 of the pterin ring. We have determined the pKa of N5 of dihydrofolate in the Escherichia coli DHFR/NADP+/H2folate ternary complex by Raman difference spectroscopy and found that the value is 6.5. In contrast, the pKa of N5 is less than 4.0 in either the binary complex, the ternary complex with an analogue of NADPH (H2NADPH), or the Asp27 to serine mutant DHFR (D27S) ternary complex with NADP+. Thus, one need not invoke proton donation from Asp-27 to N5 via a series of bound water molecules and/or pteridine-ring substituents. We propose instead that the N5 protonated form of H2folate is stabilized directly at the active site in the DHFR/NADPH/H2folate complex by specific interactions that form only in the ternary complex, involving perhaps a bound water molecule, the carboxamide moiety of the coenzyme, and/or the local electrostatic field of the enzyme molecule, to which an important contribution may be made by Asp-27.

R67 dihydrofolate reductase (DHFR) is an R-plasmid encoded protein that confers clinical resistance to the antibacterial drug trimethoprim. To determine whether an acidic titration in kinetic pH profiles is related to titration of histidines 62, 162, 262, and 362, the stability of tetrameric R67 DHFR has been monitored as a function of pH. For the pH range 5-8, tetrameric R67 DHFR reversibly dissociates into dimers, as monitored by ultracentrifugation and molecular sieving techniques. From the crystal structures of dimeric and tetrameric R67 DHFR [Matthews et al. (1986) Biochemistry 25, 4194-4204] (Narayana, Matthews, and Xuong, personal communication), symmetry-related histidines 62, 162, 262, and 362 occur at the two dimer-dimer interfaces and protonation of these residues could destabilize tetrameric R67 DHFR. Ionization of these histidines was confirmed by monitoring the chemical shifts of the C2 proton in NMR experiments, and best fits of an incomplete titration curve yield a pKa of 6.77. Since tryptophans 38, 138, 238, and 338 also occur at the dimer-dimer interfaces, fluorescence additionally monitors the tetramer-two dimers equilibrium. When fluorescence was monitored over the pH range 5-8, a protein concentration dependence of fluorescence was observed and global fitting of three titration curves yielded Kd = 9.72 nM and pKa = 6.84 for the linked reactions: [formula: see text] Modification of H62, H162, H262, and H362 by diethyl pyrocarbonate stabilizes dimeric R67 DHFR and causes a 200-600-fold decrease in catalytic efficiency. Decreased catalytic activity in dimeric R67 DHFR is presumably due to loss of the putative single active site pore found in tetrameric R67 DHFR.

Dihydrofolate reductase (DHFR) catalyzes the NADPH-dependent reduction of dihydrofolate (H2folate) to tetrahydrofolate (H4folate). This enzyme is necessary for maintaining adequate levels of H4folate and its derivatives, which are essential for the biosynthesis of purines, thymidylate and several amino acids. This enzyme is therefore a target for anticancer drugs such as methotrexate (MTX). Cancer patients receiving high dose MTX as part of their therapy frequently develop clinical resistance and molecular mechanisms of such resistance are of interest.

Since dihydrofolate reductase (DHFR) is responsible for maintaining intracellular folate pools in their biochemically active reduced state, its inhibition has been the target of continuing antifolate development to improve selectivity, penetration into target cells, and effectiveness against tumors with either intrinsic or acquired antifolate resistance1. For example, investigation of patients with methotrexate (MTX)-resistance has shown that in some tumors the resistance is due to defects in cellular transport or polyglutamation and that new, second generation, antifolate drugs such as 10-ethyl-10-deazaminopterin (10-EDAM) or the lipophilic antifolates piritrexim (PTX) and trimetrexate (TMQ), are more effective than methotrexate (MTX)2 . In other tumors it was found that DHFR encoded mutations which were responsible for the MTX resistance. Studies of a highly MTX-resistant subline of the HCT-8 human colon carcinoma cell line showed that DHFR possessed a phenylalanine to serine mutation at residue 313–5 . Kinetic studies further showed that the Ki for MTX was 100-fold higher than wild type DHFR while a mutation of phenylalanine 34 to serine resulted in over 20,000-fold increase in the Ki for MTX. Similar changes in Ki were noted for trimetrexate as well.

Oligonucleotide-directed mutagenesis has been used by many laboratories to obtain information regarding the significance of the structural features of various enzymes. When an amino acid residue, or group of residues, is deleted from the enzyme structure, or is replaced by other residues by directed mutagenesis, inferences can be drawn regarding the relation of that portion of the structure to the catalytic activity of the enzyme. This approach has been applied extensively to the study of dihydrofolate reductase (DHFR) by Benkovic and his collaborators.

Dihydrofolate reductase (DHFR) catalyzes NADPH-dependent reduction of 7,8-dihydrofolate (H2folate) into 5,6,7,8-tetrahydrofolate (H4folate), which is essential for the synthesis of thymidylate and hence the synthesis of DNA. Since rapidly growing cells can be killed by inhibition of hDHFR by methotrexate (MTX), this and other inhibitory drugs are used in the chemotherapy of cancer. A serious problem in such chemotherapy with MTX is development of MTX-resistance in tumor cells. Spontaneous mutation in the DHFR gene is suspected to be one of the causes for such resistance.

There is marked pH dependence of the rate constant (koff) for tetrahydrofolate (H4folate) dissociation from its ternary complex with human dihydrofolate reductase (hDHFR) and NADPH. Similar pH dependence of H4folate dissociation from the ternary complex of a variant of hDHFR with the substitution Phe31----Leu (F31L hDHFR) causes this dissociation to become rate limiting in the enzyme mechanism at pH approximately 5, and this accounts for the marked decrease in kcat for this variant as the pH is decreased from 7 to 5. This decreased kcat at low pH is not seen for most DHFRs. koff for dissociation of folate, dihydrofolate (H2folate), and H4folate from their binary complexes with hDHFR is similarly pH dependent. For all the complexes examined, the pH dependence of koff in the range pH 5-7 is well described by a pKa of about 6.2 and must be due to ionization of a group on the enzyme. In the higher pH range (7-10), koff increases further as the pH is raised, and this relation is governed by a second pKa which is close to the pKa for ionization of the amide group (HN3-C4O) of the respective ligands. Thus, ionization of the ligand amide group also increases koff. Evidence is presented that the dependence of pH on koff for hDHFR accounts for the shape of the kcat versus pH curve for both hDHFR as well as its F31L variant and contributes to the higher efficiency of hDHFR compared with bacterial DHFR.

The active sites of all bacterial and vertebrate dihydrofolate reductases that have been examined have a tryptophan residue near the binding sites for NADPH and dihydrofolate. In cases where the three-dimensional structure has been determined by X-ray crystallography, this conserved tryptophan residue makes hydrophobic and van der Waals interactions with the nicotinamide moiety of bound NADPH, and its indole nitrogen interacts with the C4 oxygen of bound folate through a bridge provided by a bound water molecule. We have addressed the question of why even the very conservative replacement of this tryptophan by phenylalanine does not seem to occur naturally. Human dihydrofolate reductase with this replacement of tryptophan by phenylalanine has increased rate constants for dissociation of substrates and products and a considerably decreased rate of hydride transfer. These cause some changes in steady-state kinetic behavior, including substantial increases in Michaelis constants for NADPH and dihydrofolate, but there is also a 3-fold increase in kcat. The branched mechanistic pathway for this enzyme has been completely defined and is sufficiently different from that of wild-type enzyme to cause changes in some transient-state kinetics. The most critical changes resulting from the amino acid substitution appear to be a 50% decrease in stability and a decrease in efficiency from 69% to 21% under intracellular conditions.

Lysine-54 of human dihydrofolate reductase (hDHFR) appears to be involved in the interaction with the 2{prime}-phosphate of NADPH and is conserved as a basic residue in other species. Studies have suggested that in Lactobacillus casei dihydrofolate reductase Arg-43, the homologous residue at this position, plays an important role in the binding of NADPH and in the differentiation of K{sub m} values for NADPH and NADH. A Lys-54 to Gln-54 mutant (K54Q) of hDHFR has been constructed by oligodeoxynucleotide-directed mutagenesis in order to study the role of Lys-54 in differentiating K{sub m} and k{sub cat} values for NADPH and NADH as well as in other functions of hDHFR. The purpose of this paper is to delineate in quantitative terms the magnitude of the effect of the Lys-54 to Gln-54 replacement on the various kinetic parameters of hDHFR. Such quantitative effects cannot be predicted solely on the basis of X-ray structures. The ratio of K{sub m}(NADH)/K{sub m}(NADPH) decreases from 69 in the wild-type enzyme to 4.7 in the K54Q enzyme, suggesting that Lys-54, among other interactions between protein side-chain residues and the 2{prime}-phosphate, makes a major contribution in terms of binding energy and differentiation of K{sub m} values for NADPH and NADH.more » Agents at concentrations that show activating effects on the wild-type enzyme such as potassium chloride and urea all inactivate the K54Q enzyme. There appear to be no gross conformational differences between wild-type and K54Q enzyme molecules as judged by competitive ELISA using peptide-specific antibodies against human dihydrofolate reductase and from protease susceptibility studies on both wild-type and K54Q mutant enzymes. The pH-rate profiles using NADPH for K54Q and wild-type enzymes show divergences at certain pH values, suggesting the possibility of alteration(s) in the steps of the catalytic pathway for the K54Q enzyme.« less

The role of the active site residue phenylalanine-31 (Phe31) for recombinant human dihydrofolate reductase (rHDHFR) has been probed by comparing the kinetic behavior of wild-type enzyme (wt) with mutant in which Phe31 is replaced by leucine (F31L rHDHFR). At pH 7.65 the steady-state kcat is almost doubled, but the rate constant for hydride transfer is decreased to less than half that for wt enzyme, as is the rate of the obligatory isomerization of the substrate complex that precedes hydride transfer. Although steady-state measurements indicated that the mutation causes large increases in Km for both substrates, dissociation constants for many complexes are decreased. These apparent paradoxes are due to major mutation-induced decreases in rate constants (koff) for dissociation of folate, dihydrofolate, and tetrahydrofolate from all of their complexes. This results in a mechanism proceeding almost entirely by only one of the two pathways used by wt enzyme. Other consequences of these changes are a much altered dependence of steady-state kcat on pH, inhibition rather than activation by tetrahydrofolate, absence of hysteresis in transient-state kinetics, and a decrease in enzyme efficiency under physiological conditions. The results indicate that there is no quantitative correlation between dihydrofolate binding and the rate of hydride transfer for this enzyme.

A purification procedure is reported for obtaining bovine liver dihydrofolate reductase in high yield and amounts of 100-200 mg. A key step in the procedure is the use of an affinity gel prepared by coupling pteroyl-L-lysine to Sepharose. The purified reductase has a specific activity of about 100 units/mg and is homogeneous as judged by analytical ultracentrifugation, polyacrylamide gel electrophoresis, and titration with methotrexate. The products of the first step of Edman degradation indicated a minimum purity of 79%. The reductase has a molecular weight of about 21500 on the basis of amino acid composition and 22100 +/- 300 from equilibrium sedimentation. It is not inhibited by antiserum to the Streptococcus faecium reductase (isoenzyme 2). Unlike the reductase of many other vertebrate tissues, the bovine enzyme is inhibited by mercurials rather than activated and it has a single pH optimum at both low and high ionic strength. However, the position of the pH optimum is shifted and the activity increased by increasing ionic strength. Automatic Edman degradation has been used to determine 34 of the amino-terminal 37 amino acid residues. Considerable homology exists between this region and the corresponding regions of the reductase from S. faecium and from Escherichia coli. This strengthens the idea that this region contributes to the structure of the binding site for dihydrofolate.

When 15 nM methotrexate was added to the medium in which human peripheral lymphocytes stimulated with phytohemagglutinin were incubated, it caused a 50 per cent decrease in the maximum number of blasts produced, in the number of cells in mitosis and in the incorporation of [ 3 H]deoxyuridine into DNA. However, [ 3 H]thymidine incorporation into DNA was increased by methotrexate concentrations up to 0.5 mM. When 50 nM methotrexate was present continuously, blast formation, mitosis and deoxyuridine incorporation were virtually abolished, but if this concentration was present only during the induction phase (the first 24 hr), the subsequent effect on blast proliferation was slight. In contrast, 24-hr exposure during the proliferative phase (days 3–5) severely affected blast proliferation. The effects of methotrexate were largely reversible by thymidine, but hypoxanthine or purine nucleosides had no significant effect so that the metabolic block appears to be entirely at thymidylate synthetase under the experimental conditions. The inhibitory effects of ten other antifolates on transforming lymphocytes were determined and, with one exception, their relative effectiveness was found to be as predicted from inhibitory effects on highly purified bovine dihydrofolate reductase. The growth of four established lines of human lymphoblastic cells was inhibited to essentially the same extent by six of the antifolates, and these cells were only slightly less sensitive to the antifolates than were the transforming normal lymphocytes.

Dihydrofolate reductase from Streptococcus faecium has been labeled with (..gamma..-/sup 13/C)tryptophan. We have determined changes occurring in the chemical shifts and line widths of the four resonances of the /sup 13/C NMR spectrum of the labeled enzyme, due to its interaction with various ligands. These include the coenzyme, NPDPH and related nucleotides, folate and its polyglutamate derivatives, and many inhibitors including methotrexate and trimethoprim. In addition, paramagnetic relaxation effects produced by a bound spin-labeled analogue of 2'-phosphoadenosine-5'-diphosphoribose on the tryptophan C/sup ..gamma../ carbons have been measured. Distances calculated from the relaxation data have been compared with corresponding distances in the crystallographic model of the NADPH-methotrexate ternary complex of Lactobacillus casei reductase. The paramagnetic relaxation data indicate that the two downfield resonances (1 and 2) correspond to tryptophans (W/sub A/ and W/sub B/) that are more remote from the catalytic site, and from the crystallographic model these are seen to be Trp-115 and Trp-160. The upfield resonances (3 and 4) that show broadening due to chemical exchange correspond to closer residues (W/sub C/ and W/sub D/), and these are identified with Trp-6 and Trp-22. However, the relaxation data do not permit specific assignments within the nearer and farther pairs. Although resonancemore » 3, which is split due to chemical exchange, was formerly assigned to Trp-6, data obtained for the enzyme in the presence of various ligands are better interpreted if resonance 3 is assigned to Trp-22, which is located on a loop that joins elements of secondary structure and forms one side of the ligand-binding cavity.« less

Transient and steady-state kinetics have been examined for dihydrofolate reductase (DHFR) from a number of sources. Rates of hydride transfer at pH 7.65 cover a wide range, from 7 s−1 for DHFR from a strain of Lactobacillus casei (LCDHFR1) to 3000 s−1 for recombinant human DHFR (rHDHFR). In all cases as the pH is increased from 7 to 10, Vmax for the steady-state reaction decreases, and DVmax, the primary isotope effect, increases. This indicates a decrease in the rate of hydride transfer with increasing pH. The cross-over points, at which rates of product release and hydride transfer become equal, were calculated to occur at DVmax = 2.34. The higher the rate of hydride transfer at pH 7.65, the higher the pH of the cross-over point. For LCDHFR1 the low rate of hydride transfer results in this process being partially rate-limiting for the steady-state reaction even at pH 5, with a cross-over point at about pH 7. At pH 7.65 the burst phase associated with the initial conversion of enzyme-bound substrates to enzyme-bound products has an isotope effect of 3 or higher for LCDHFR and for DHFR from Escherichia coli (ECDHFR). In contrast, the vertebrate DHFRs (bovine, BDHFR; chicken, CDHFR; and rHDHFR) exhibit a burst of product formation which is only partially limited by hydride transfer at this pH (Dkb: 2.3, 2.2, and 2.1, respectively). An obligatory isomerization of the ternary substrate complex or of the ternary product complex is postulated to be partially rate-limiting for the vertebrate enzymes. At pH 5 LCDHFR1 and ECDHFR also exhibit evidence of such a rate-limiting obligatory conformational transition of the substrate or product ternary complex.

P3-(6-(N-Trifluoracetyl)aminohex-1-yl) deoxyguanosine triphosphate has been prepared by the reaction of N-trifluoroacetyl-6-aminohexanol 1-pyrophosphate with the imidazolide of dGMP and has been characterized. This compound and the corresponding free amine, obtained by removal of the protective trigluoroacetyl group, are activators of ribonucleotide reductase of Lactobacillus leichmannii. An affinity adsorbent for the reductase, prepared by reaction of the amine derivative with CNBr-activated Sepharose, contains dGTP covalently attached through the gamma-phosphate via a six-carbon chain to the matrix. The method of synthesis of the dGTP derivative is generally applicable to the synthesis of P3-(omega-aminoalk-1-yl)nucleoside triphosphate esters for the preparation of analogous affinity adsorbents. Ribonucleotide reductase can be rapidly purified to homogeneity, on a large scale, by use of dGTP-Sepharose and conditions for optimum recovery of the enzyme have been determined. The affinity of ribonucleotide reductase and other proteins for dGTP-Sepharose is increased by either raising the ionic strength or lowering the temperature of the eluent. Elution of the enzyme from the adsorbent can be achieved between pH 5.8 and 7.3, whereas at pH 5.3 the reductase is bound extremely tightly and cannot be recovered. Ribonucleotide reductase can be eluted from the adsorbent with dGTP or urea. Elution with urea is carried out at pH 6.3, where the enzyme is stable and maximum recovery is obtained. Affinity chromatography consistently produces ribonucleotide reductase of high specific activity (170-180 units/mg). In the presence of 0.1 to 1.2 M urea or hydroxyurea, the enzyme is inhibited, but allosteric activation is unchanged. No alteration in the structure or function of the reductase was detected when the enzyme was exposed to 2.0 M urea during elution from the affinity adsorbent, but exposure for longer periods causes some inactivation.

Coalpha-[alpha-(Aden-9-yl)]-Cobeta-adenosylcobamide (pseudocoenzyme B12) purified from Clostridium tetanomorphum has been reacted with ribonucleotide reductase purified from Lactobacillus leichmannii under various conditions, and the properties of the products obtained have been compared by electron paramagnetic resonance (EPR) with those previously reported for products formed from the normal coenzyme (adenosylcobalamin). The rapidly formed intermediate and the slowly formed "doublet" species from the pseudocoenzyme have EPR spectra identical with those formed from the normal coenzyme. This and other considerations make it less likely that the unusual magnetic properties of the rapidly formed intermediate are due to strongly distorted octahedral symmetry about Co(II) as previously postulated. Instead it is probable that the EPR spectrum is due to interaction of the radical pair by both exchange coupling and magnetic dipole--dipole coupling. Although Coalpha-[alpha-(aden-9-YL)]cob(II)amide in solution does not show superhyperfine splitting in the EPR spectrum because of its base-off configuration, the cob(II)amide formed by degradation of the pseudocoenzyme within the catalytic site of the enzyme did show triplets due to a nitrogen axially coordinated to cobalt. This suggests that binding of the cob(II)amide to the reductase catalytic site causes a shift to the base-on form.

Nucleoside deoxyribosyltransferase has been extensively purified from extracts of Lactobacillus leichmannii . The enzyme was initially co-purified with ribonucleotide reductase and then separated from the latter by affinity chromatography on dGTP-Sepharose. A new procedure is described for the synthesis of 6-( p -aminobenzylamino)purine. The latter was diazotized and reacted with Sepharose substituted with m -phenylenediamine to produce an affinity material for chromatography of the transferase. Some properties of the purified transferase are described, including remarkable stability to heat and to guanidine hydrochloride. With thymidine as donor, the transferase had a very broad specificity for the purine acceptor, the only modifications not tolerated being the loss of the tautomeric N in the imidazole ring and, possibly, substitution at the 8-position. The enzyme was used for the convenient synthesis of twelve analogs of 2'-deoxyadenosine on the 100–400 mg scale with an average yield of 64 per cent. The method was readily applicable to larger scale preparations and the enzyme was recoverable from the final reaction mixtures in good yield. Of the nucleosides synthesized, the 2-chloro and 2-bromo analogs of 2'-deoxyadenosine were especially good inhibitors of the growth of KB and CCRF-CEM cells in culture.

Dihydrofolate reductase from wild-type Escherichia coli (WT-ECDHFR) and from a mutant enzyme in which aspartate 27 is replaced by asparagine have been compared with respect to the binding of the inhibitor methotrexate (MTX). Although the Asp27----Asn substitution causes only small changes in the association rate constants (kon) for the formation of binary and ternary (with NADPH) complexes, the dissociation rate constants for these complexes (koff) are increased for the mutant enzyme by factors of about 5- and 100-fold, respectively, at pH 7.65. In binding experiments, the initial MTX binary and ternary complexes of the mutant enzyme were found to undergo relatively rapid isomerization (kobs approximately 17 and 145 s-1, respectively). Although such rapid isomerization of complexes of WT-ECDHFR could not be detected in binding experiments, evidence of a slow isomerization (k = 4 x 10(-3) s-1) of the ternary WT-ECDHFR.MTX.NADPH complex was obtained from progress of inhibition experiments. This slow isomerization increases binding of MTX to WT-ECDHFR only 2.4-fold (much less than previously estimated). From presently available data, we could not determine the contribution of the rapid isomerization of complexes to the binding of MTX to the mutant enzyme. The Asp27----Asn substitution increases the overall dissociation constant (KD) 9-fold for the binary complex and 85-fold for the ternary complex. When it is also taken into account that a proton ultimately derived from the solvent must be added to MTX bound to the WT enzyme, but not to MTX bound to the mutant enzyme, these increases in KD for the mutant enzyme correspond to decreases in binding energy for MTX of 3.9 and 5.2 kcal/mol at pH 7.65 for the binary and ternary complexes, respectively.

Summary [2- 13 C]Methotrexate bound to dihydrofolate reductase from Lactobacillus casei has been studied by nuclear magnetic resonance. The chemical shift for enzyme-bound methotrexate over the range pH 6.0 to 9.0 is constant and close to that for free, protonated methotrexate. Furthermore, the ultraviolet spectrum of bound methotrexate is the same from pH 6.4 to pH 10.0 and considerably different from that of free, unprotonated methotrexate. These data indicate that N-1 of bound methotrexate has pK > 10. From this and from the measured apparent association constants of methotrexate and folate it can be concluded that the charge interaction between the protonated N-1 of bound methotrexate and the active site carboxyl accounts for the particularly tight binding of the inhibitor.