Patients infected with the human immunodeficiency virus experienced severe hematopoietic toxicity after treatment with the deoxynucleoside analog 3'-fluorothymidine (FLT). Using several methods for the analysis of genome integrity, including histochemical staining of the 3' ends of DNA and both conventional and pulsed-field agarose gel electrophoresis, we demonstrated that FLT caused extensive DNA fragmentation in CEM cells that was not observed when these cells were treated with other, less toxic thymidine analogs. In addition, a distinctive pattern of small DNA fragments that is characteristic of cells in the process of programmed cell death was observed in the genomic DNA of CEM cells treated with FLT. We conclude that FLT induces DNA fragmentation and apoptosis in a human cell line of hematopoietic origin, and we offer this observation as a possible explanation for the severe toxicity of FLT observed in vivo.

AbstractChemInform is a weekly Abstracting Service, delivering concise information at a glance that was extracted from about 100 leading journals. To access a ChemInform Abstract of an article which was published elsewhere, please select a “Full Text” option. The original article is trackable via the “References” option.

5-Chloro-2',3'-dideoxy-3'-fluorouridine (935U83) is a selective anti-human immunodeficiency virus (HIV) agent. When tested in phytohemagglutinin-stimulated normal human peripheral blood lymphocytes against fresh clinical isolates of HIV type 1 (HIV-1) obtained from patients naive to AZT (3'-azido-3'-deoxythymidine [zidovudine]), 935U83 inhibited virus growth with an average 50% inhibitory concentration (IC50) of 1.8 microM; corresponding IC50s were 0.10 microM for FLT (3'-deoxy-3'-fluorothymidine) and 0.23, 0.49, and 0.03 microM for the approved agents AZT, ddI (2',3'-dideoxyinosine), and ddC (2',3'-dideoxycytosine), respectively. Importantly, 935U83 retained activity against HIV strains that were resistant to AZT, ddI, or ddC. Of additional interest, we were unable to generate virus which was resistant to 935U83 by passaging either HXB2 (AZT-sensitive) or RTMC (AZT-resistant) strains in the presence of high concentrations of 935U83. The anabolic profile of 935U83 was similar to that of AZT, and 935U83 triphosphate was a potent inhibitor of HIV-1 reverse transcriptase. Pharmacokinetic evaluation showed good oral bioavailability (86% in mice and 60% in monkeys) and less extensive metabolism to the glucuronide relative to AZT. 935U83 showed low toxicity. In an in vitro assay for toxicity to a human erythrocyte progenitor, erythroid burst-forming unit (BFU-E), the IC50 for 935U83 (> 400 microM) was more than 1,000-fold those of FLT (0.07 microM) and AZT (0.30 microM). Mild reversible reductions in erythrocytes and associated parameters were seen in mice dosed orally with 2,000 mg of 935U83 per kg per day for 1 and 6 months. In monkeys dosed orally with up to 700 mg/kg/day for 1 and 6 months, the only possible treatment-related finding was cataracts in 1 of 12 animals given the intermediate dose of 225 mg/kg/day. At the highest doses in mice and monkeys, maximal concentrations in plasma were more than 100-fold the anti-HIV IC50s against clinical isolates. This safety profile in animals compares very favorably with that of any of the anti-HIV drugs approved to date and has led us to begin evaluation of 935U83 in patients with HIV infection.

Syntheses of several new inhibitors of thymidylate synthase (TS) structurally related to folic acid are described in which the pterin portion of the folate molecule is replaced by a benzo[f]quinazoline moiety, but which retain the natural methyleneamino link to the benzoylglutamate side chain. The effect on enzyme activity and cytotoxicity of various changes in the structure of the (p-aminobenzoyl)glutamate side chain are reported. Replacement of the benzamide portion of the (p-aminobenzoyl)glutamate moiety with 2-fluorobenzamido, 2-isoindolinyl, 1,2-benzisothiazol-2-yl, and 2-thenamido moieties varied in effect from a 9-fold diminution of TS activity to a 5-fold enhancement, while cytotoxic potency on SW-480 and MCF-7 tumor lines showed increases ranging from 3.6- to 450-fold. The detrimental effect on enzyme activity and cytotoxicity of alkyl substitution on the PABA nitrogen is confirmed for these compounds, in contrast with several series of previously reported quinazoline antifolates (2). Substitution of a C3-methyl substituent for 3-amino had little effect on TS activity but was beneficial in terms of solubility and cytotoxicity. The excellent combination of TS inhibitory activity, FPGS substrate activity, and affinity for the reduced folate transport system in the most potent of these derivatives, 3e, resulted in IC50 values of 0.2-0.8 nM against these tumor lines.

Human thymidylate synthase is a polymeric protein composed of two subunits with identical primary structures. In this study we determined the binding affinities of 5,10-methylene tetrahydropteroyltetraglutamate (folate substrate) and a group of close structural folate analog inhibitors. Thymidylate synthase bound both mono and polyglutamylated folate substrates and analogs more tightly in the presence of deoxyuridylate. These results and product inhibition studies confirmed that the orders of substrate addition and product release from thymidylate synthase were similar for mono and polyglutamylated substrates. Equilibrium dialysis studies showed that the folate substrate in a ternary complex with deoxyuridylate bound to one of the subunits (site A) with a Kd of 720 nM. The binding of the substrate to the second subunit (site B) was much weaker, and the Kd could not be determined by this method. However, dissociation constants for each subunit could be measured for the folate analog inhibitors, and, depending on the inhibitor, the relative Kd value for each subunit varied substantially. For example, formyl-5,8-dideazafolate and tetraglutamylated 10-propargyl-5,8-dideazafolate bound to both sites with similar Kd values, whereas D1694Glu4 bound to subunit A with a higher affinity (Kd = 1.0 nM) than to subunit B (Kd = 30 nM). In contrast, 1843U89 (mono or diglutamylated form) had a much higher affinity for subunit B (Kd approximately 0.1 nM) compared with subunit A (Kd approximately 400 nM). Enzyme inhibition kinetic analyses showed that the Ki values of 1843U89 were quite low (0.1 nM) and that the inhibition was noncompetitive. In contrast, the other folate analogs inhibited the enzyme via mixed inhibition (i.e. both the Km for the folate substrate and the Vmax were altered). We conclude that the two subunits of thymidylate synthase bind folate substrates and analogs differently and that the asymmetric binding of the ligands is the major factor that determines the inhibition kinetics of each folate analog inhibitor.

Dihydropteroate synthase (DHPS) (EC 2.5.1.15) is the target for the class of compounds known as sulfonamides (1, 4, 11, 16). We recently reported the cloning and sequencing of the DHPS gene, folP, from Escherichia coli (6). We used the cloned gene as a hybridization probe to determine which bacteriophages from the Kohara miniset phage collection carry folP (7). Two phages were found to be probe positive, 18H7 (520) and 14F11 (521), and inspection of the Kohara phage

Several folate-like thymidylate synthase inhibitors are described in which the pteridine nucleus of the folic acid molecule is replaced by a benzoquinazoline moiety, which in turn is attached to the benzoylglutamate side chain by a sulfonamide link. The most potent compounds had K(i) values as low as 2.5 nM against the human enzyme, were good substates for the cellular reduced folate transport system and for folylpolyglutamate synthetase, and had IC50 values for growth inhibition of tumor cell lines as low as 70 nM.

Studies on a series of benzoquinazoline folate analogues as inhibitors of human thymidylate synthase led to the selection of 1843U89 for further evaluation. This compound had a Ki of 90 pM versus human thymidylate synthase and was noncompetitive with (6R,S)-5,10-methylenetetrahydrofolate. It was a good substrate for the addition of the second glutamate by hog liver folylpolyglutamate synthetase, having a Vmax/Km value 7.8-fold higher than (6R,S)-tetrahydrofolate. The data indicate that 1843U89 was transported into cells via the reduced folate carrier. The Kt for 1843U89 in MOLT-4 cells was 0.33 microM, which was 3-fold lower than that for methotrexate and 16-fold lower than that for (6S-5-formyltetrahydrofolate. V/K values were 20.3 for 1843U89 versus 1.2 and 1.9 for methotrexate and (6S)-5-formyltetrahydrofolate, respectively. It was a potent inhibitor of the growth of human cells, having 50% inhibitory concentrations below 1 nM for all cell lines tested. Growth inhibition was reversed by thymidine alone, indicating that thymidylate synthase was the only site of action of this compound. Growth inhibition was not affected by (6R,S-5-formyltetrahydrofolate at concentrations below 5 microM. However, the 50% inhibitory concentration increased when the concentration in the medium was increased to 100 microM, presumably due to competition for transport. Relative to the human cell lines used, murine cell lines were 80-1300-fold less sensitive to 1843U89 and the other benzoquinazolines tested. This decreased sensitivity appeared to be due, at least in part, to decreased transport or accumulation in murine cells. Ki values for inhibition of methotrexate transport for the benzoquinazolines were 5-17-fold higher in L1210 cells than in MOLT-4 cells. 1843U89, the benzoquinazoline which was transported most efficiently and which was the most potent inhibitor of the growth of human cells, exhibited the largest difference between binding to the MOLT-4 human and L1210 murine transporter. The V/K for L1210 transport was 80-fold less than that for MOLT-4. Initial antitumor studies, using the human thymidine kinase-deficient line GC3TK- to circumvent problems associated with murine transport as well as the high circulating thymidine levels in mice, indicated that 1843U89 had marked in vivo antitumor activity.

The synthesis and thymidylate synthase (TS) inhibitory activity of a series of simple benzo[f]-quinazolin-1(2H)-ones are described. Fully aromatic 3-amino compounds with compact lipophilic substituents in the 9-position were found to have I50 values as low as 20 nM on the isolated enzyme, and represent the first examples of potent, folate-based TS inhibitors that completely lack any structural feature corresponding to the (p-aminobenzoyl)glutamate moiety of the cofactor. A number of the compounds also showed moderate growth inhibitory activity against a human colon adenocarcinoma cell line (SW480), with IC50 values as low as 2 microM.

The Escherichia coli gene coding for dihydropteroate synthase (DHPS) has been cloned and sequenced. The protein has 282 amino acids and a compositional molecular mass of 30,314 daltons. Increased expression of the enzyme was realized by using a T7 expression system. The enzyme was purified and crystallized. A temperature-sensitive mutant was isolated and found to express a DHPS with a lower specific activity and lower affinities for para-aminobenzoic acid and sulfathiazole. The allele had a point mutation that changed a phenylalanine codon to a leucine codon, and the mutation was in a codon that is conserved among published DHPS sequences.

The gene coding for the Escherichia coli enzyme 7,8-dihydro-6-hydroxymethylpterin-pyrophosphokinase has been cloned and sequenced. This gene, designated folK, codes for a protein of 159 amino acids, including an amino-terminal methionine. The protein was overexpressed in E. coli MC4100 by cloning the gene behind the lacUV5 promoter in a high-copy-number plasmid. The enzyme was purified to homogeneity. Amino-terminal analysis of the purified protein showed that the amino-terminal methionine had been removed. The compositional molecular mass (17,945 Da) was identical to the molecular mass determined by mass spectrometry. The enzyme was observed to have a large number of proline residues and migrated anomalously in sodium dodecyl sulfate-polyacrylamide gels, with an apparent molecular mass of 23,000 Da.

Metabolic effects and mode of cytotoxicity of 5-deazaacyclotetrahydrofolate (5-DACTHF, BW543U76), a glycineamide ribonucleotide transformylase inhibitor, were studied in MOLT-4 cells, a human T-cell leukemia line. 5-DACTHF inhibits purine synthesis with 50% inhibitory concentration values of 0.5 microM and 0.08 microM following 6- or 24-h exposure to drug, respectively. At 6 h, adenine nucleotide synthesis is preferentially inhibited over guanine nucleotide synthesis. A similar effect was observed with another glycineamide ribonucleotide transformylase inhibitor, 5,10-dideazatetrahydrofolate. GTP was depleted to 40% of control and ATP to 10% of control by 5 microM 5-DACTHF. After a transitory increase, UTP and CTP were depleted to 30% of control. Deoxynucleotides were also depleted by the drug; dCTP was depleted to the greatest extent, followed by dATP, dTTP, and dGTP, respectively. MOLT-4 cell growth was inhibited by 5-DACTHF with a 50% inhibitory concentration of 0.066 microM. Complete reversal was effected by hypoxanthine, and there was no reversal by thymidine. The drug was cytotoxic to MOLT-4 cells in the range 0.25 to 5.0 microM, but a minimum of 48 h was required for trypan blue-staining dead cells to appear. The rate and extent of kill with the thymidylate synthase inhibitor 2-methyl-10-propargyl-5,8-dideazafolate was greater than with 5-DACTHF, which indicates that kill by inhibition of thymidylate synthase is more effective than that by inhibition of purine synthesis. Electron microscopy of MOLT-4 cells exposed to 5-DACTHF showed electron-dense mitochondria and nuclear changes reminiscent of apoptosis. These morphological changes were accompanied by the appearance of DNA strand breaks at approximately 180-base pair intervals (internucleosomal breaks). Concomitant proteolysis of nuclear proteins poly(ADP-ribose) polymerase and lamin B was observed.

The enzymes 7,8-dihydroxymethylpterin-pyrophosphokinase (HPPK) and 7,8-dihydropteroate synthase (DHPS), which act sequentially in the folate pathway, were purified to homogeneity from crude extracts of Escherichia coli MC4100. The enzymes represent less than 0.01% of the total soluble protein. HPPK was purified greater than 10,000-fold; the native enzyme appears to be a monomer with a molecular mass of 25 kDa and a pI of 5.2. DHPS was purified greater than 7,000-fold; the native enzyme has an apparent molecular mass of 52 to 54 kDa and is composed of two identical 30-kDa subunits. The amino-terminal sequences for both enzymes have been determined.

Highly purified preparations of signal peptidase I (36 kDa) were found to undergo an apparent inter-autocatalytic degradation at 4 degrees C and 37 degrees C. The disappearance of the 36 kDa protein coincided with the stable appearance of a 31 kDa and a 5 kDa species. Amino-terminal sequencing of the 31 kDa product indicated a site specific cleavage following Ala38-Gln-Ala of signal peptidase I. The 31 kDa fragment was purified and shown to have 100-fold less activity than the native enzyme, with pre-maltose binding protein as a substrate.

Prompted by recent disclosures concerning the potent antitumor activities of 5-deaza-5,6,7,8-tetrahydrofolic acid and 5,10-dideaza-5,6,7,8-tetrahydrofolic acid (DDATHF), we have prepared 5-deazaisofolic acid (3a) and 5-deaza-5,6,7,8-tetrahydroisofolic acid (4a). Reductive condensation of 2,6-diamino-3,4-dihydro-4- oxopyrido[2,3-d]pyrimidine with di-tert-butyl N-(4-formylbenzoyl)-L-glutamate and subsequent deprotection with trifluoroacetic acid yielded 5-deazaisofolic acid in good yield. Catalytic hydrogenation of this analogue then gave 4a. The 9-CH3 and 9-CHO modifications of 3a and the 9-CH3 derivative of 4a were also synthesized. Each of the new analogues was evaluated with a variety of folate-requiring enzymes as well as MCF-7 cells in culture. Compound 4a had an IC50 of ca. 1 microM against MCF-7 cells and was nearly 100-fold less potent than DDATHF in this regard. The three oxidized isofolate analogues were all poor inhibitors of tumor cell growth.

Structural modifications at the pyrimidine ring and at the C9,N10-bridge region of the thymidylate synthase (TS) inhibitors N10-propargyl-5,8-dideazafolate (1; PDDF; CB 3717), 2-desamino-N10-propargyl-5,8-dideazafolate (2, DPDDF), and 2-desamino-2-methyl-N10-propargyl-5,8-dideazafolate (3, DMPDDF) have been carried out. Methods for the synthesis of 2-desamino-N10-propargyl-1,5,8-trideazafolate (4), 2-desamino-2-methyl-N10-propargyl-3,5,8-trideazafolate (5a), and 2-desamino-2-methyl-N10-propargyl-5,8-dideaza-1,2-dihydrofolate (6) have been developed. The bridge-extended analogues isohomo-PDDF (7) and isohomo-DMPDDF (8) contain an additional methylene group interposed between N10 and the phenyl ring of 1 and 3, respectively. All new compounds were evaluated as inhibitors of TS and the growth of tumor cells in culture. Selected analogues were tested as substrates of folylpolyglutamate synthetase (FPGS) and striking differences in substrate activity were observed among these compounds, indicating that structural modifications at the pyrimidine ring of classical antifolates profoundly influence their polyglutamylation. Enzyme inhibition data established that both N1 and N3-H of the pyrimidine ring are essential for efficient binding of quinazoline-type antifolates to human TS.

The minimum substrate sequence recognized by signal peptidase I (SPase I or leader peptidase) was defined by measuring the kinetic parameters for a set of chemically synthesized peptides corresponding to the cleavage site of the precursor maltose binding protein (pro-MBP). The minimum sequence of a substrate hydrolyzed by SPase I at a measurable rate was the pentapeptide Ala-Leu-Ala decreases Lys-Ile. The rates of hydrolysis of this substrate, however, were several hundred-fold lower than those observed for the maturation of MBP in Escherichia coli, suggesting that in addition to these minimal sites involved in recognition, other features of pro-MBP are also needed for the optimal rate of signal peptide cleavage by SPase I. One parameter may be the length of the polypeptide chain. Studies of the synthetic peptides showed that decreasing the length of the polypeptide chain of substrates decreased the substrate efficiency measured as kcat/Km. However, in one case a decrease in the length of a peptide corresponding to -7 to +3 positions of pro-MBP to a nonapeptide (-7 to +2) increased the substrate efficiency by about 900-fold. The nonapeptide is the most efficient substrate for the enzyme in vitro so far reported. It is speculated that better peptide substrates are the ones which are able to adopt folded structures.

Signal peptidases, the endoproteases that remove the amino-terminal signal sequence from many secretory proteins, have been isolated from various sources. Seven signal peptidases have been purified, two fromE. coli, two from mammalian sources, and three from mitochondrial matrix. The mitochondrial enzymes are soluble and function as a heterogeneous dimer. The mammalian enzymes are isolated as a complex and share a common glycosylated subunit. The bacterial enzymes are isolated as monomers and show no sequence homology with each other or the mammalian enzymes. The membrane-bound enzymes seem to require a substrate containing a consensus sequence following the −3, −1 rule of von Heijne at the cleavage site; however, processing of the substrate is strongly influenced by the hydrophobic region of the signal peptide. The enzymes appear to recognize an unknown three-dimensional motif rather than a specific amino acid sequence around the cleavage site. The matrix mitochondrial enzymes are metallo-endopeptidases; however, the other signal peptidases may belong to a unique class of proteases as they are resistant to chelators and most protease inhibitors. There are no data concerning the substrate binding site of these enzymes.

Nine mutant thymidylate synthases were isolated that only differed in sequence at position 147. The wild-type enzyme (which had a histidine residue at 147) and mutant enzymes were purified to near homogeneity and their kinetic properties were compared. Although the kcat values for the mutant enzymes were 10-10,000-fold lower than for the wild-type enzyme, the Km values for both 2'-deoxyuridylate and 5,10-methylenetetrahydrofolate were nearly identical for all the enzymes indicating that His-147 is not significantly involved in initial substrate binding. By comparing the wild-type (His-147) to the glycine (Gly-147) enzyme, the side chain of His-147 was estimated to lower the activation energy of the catalytic step by 1.6-2.9 kcal mol-1. In contrast to the wild-type enzyme, the activity of the Gly-147 enzyme decreased when the pH was raised above 7.5. The activity loss coincided with the deprotonation of a residue that had a pKa of 9.46 +/- 0.2 and an enthalpy of ionization (delta Hion) of 12.1 +/- 0.9. These values are consistent with the involvement of a lysine or an arginine residue in the catalytic process. An inspection of the rates of ternary complex formation among enzyme, 5-fluoro-2'-deoxyuridylate, and 5,10-methylenetetrahydrofolate for the mutant enzymes indicated that His-147 is not needed for the proton removal from C-5 of 2'-deoxyuridylate but rather participates in an initial catalytic step and alters the pKa value of a catalytically important lysine or arginine residue.