A fluorine-labeled methotrexate as a probe for monitoring tumor antifolate pharmacokinetics: Synthesis, in vitro cytotoxicity, and pilot in vivo ¹⁹F magnetic resonance spectra

Noninvasive in vivo monitoring of the tumor uptake and retention of anti-neoplastic agents in patients via magnetic resonance spectroscopy (MRS) is often of limited utility due to the low plasma concentrations of drug achievable without unacceptable patient toxicity (1). There are exceptions wherein previously published reports have demonstrated the in vivo MR detection of chemotherapeutic agents administered at high doses including ¹³C-labeled temozolomide (2), iproplatin investigated via ¹H MRS with multiple quantum coherence transfer techniques (3), and the ³¹P-containing agents ifosfamide and cyclophosphamide (4). However, achieving acceptable signal-to-noise ratio in monitoring pharmacokinetics is challenging. One case in which more robust sensitivity can be achieved is the in vivo monitoring of the uptake and metabolism of 5-fluorouracil (FU) via ¹⁹F MRS (see, e.g., Ref. 5). In vivo monitoring of pharmacokinetics of FU is possible due to the high plasma concentration of drug that is achieved clinically, the relatively high sensitivity of the ¹⁹F nucleus (83% of that of ¹H) in the MRS experiment, and the absence of background ¹⁹F signal from endogenous metabolites.

Another antineoplastic agent routinely administered at very high dosage, achieving plasma concentrations up to 1 mm, is methotrexate. However, this antifolate is not ¹⁹F-labeled. Herein we report on the synthesis and characterization of a fluorine-labeled methotrexate, 3′-fluoromethotrexate (FMTX). The in vitro cytotoxicity of the labeled compound is compared with that of the parent compound [methotrexate (MTX)] against a methotrexate-sensitive human sarcoma cell line (HT-1080). These results are consistent with molecular modeling studies that indicate similar binding affinities of MTX and FMTX with the enzymes dihydrofolate reductase (DHFR) and thymidine synthetase. Although previous reports in the literature detail the synthesis of other ¹⁹F-labeled antifolates (6, 7), herein we demonstrate the first in vivo observation of tumor uptake and retention of a fluorine-labeled antifolate monitored noninvasively via ¹⁹F MR in a xenograft tumor model in mice.

General. High-resolution ¹H nuclear magnetic resonance (NMR) and ¹³C NMR spectra were recorded on a Bruker AMX-400 spectrometer using tetramethylsilane as the internal chemical shift reference standard. IR spectra were recorded on a Perkin-Elmer 1600 Series Fourier transform spectrometer. Low-resolution electron ionization (EI) mass spectra were obtained on a PE SCIEX API 100 LC/MS system. All reagents were obtained either from Sigma-Aldrich (St. Louis, MO) or from Lancaster (Windham, NH) and vacuum-dried under P₂O₅ overnight before use. All solvents were reagent grade and distilled before use. Silica gel used for chromatography, MN-Kieselgel 60, was purchased from EM Science (Gibbstown, NJ). All reactions were carried out under argon using glassware dried in an oven at 80°C overnight and cooled under vacuum. The reaction mixtures were mechanically stirred using a magnetic stirring bar and stirring plate. Melting points were determined using a Mel-Temp II melting point apparatus fitted with a digital Barnart 115 thermocouple thermometer, and were uncorrected. Optical rotations were recorded on a Fasco DIP-370 digital polarimeter with a sodium lamp at ambient temperature and recorded as [α]_D²⁰ (c = g/100 ml). Molecular modeling and graphics renderings were performed using the SYBYL 6.8 (Tripos Associates Inc., St. Louis, MO) software package on a Silicon Graphics Octane2 R12000 workstation.

Synthesis of FMTX. Synthesis of FMTX was achieved via two alternative synthetic schemes (Schemes A and B) wherein the reaction product 6 is common to both methods as detailed in Figs. 1 and 2.

Di-Tert-Butyl N-(3-Fluoro-4-Nitrobenzoyl)-l-Glutamate 3. A stirred slurry of 3-fluoro-4-nitrobenzoic acid 2 (1 g, 5.4 mmol, 1 eq) in dry toluene (20 ml) was treated with thionyl chloride (0.6 ml, 8.10 mmol, 1.5 eq) and the mixture was heated under reflux for 2 h. The solution was cooled to room temperature (RT) and the solvent and excess thionyl chloride were evaporated under vacuum. A solution of the resulting acid chloride in 20 ml of CH₂Cl₂ was added dropwise to a stirred mixture of di-t-butyl l-glutamate hydrochloride (1.6 g, 5.4 mmol, 1 eq) and Et₃N (1.5 ml, 10.8 mmol, 2 eq) in 20 ml of CH₂Cl₂ at 0°C (6). The reaction mixture was allowed to warm to RT and then stirred for 2 h. The solution was washed with water, dried over Na₂SO₄, and evaporated under vacuum. The crude product was purified by column chromatography on silica gel (Hex:EtOAc, 5:1 to 4:1) to give 1.6 g (69% yield) of product 3 as a yellow syrup. ¹H NMR (CDCl₃), δ 8.12 (t, 1H, J = 8.3 Hz), 7.80 (d, 1H, J = 11.1 Hz), 7.74 (d, 1H, J = 8.7 Hz), 7.58 (d, 1H, J = 6.6 Hz, NH), 4.62 (m, 1H), 2.44–2.34 (m, 2H), 2.20 (m, 1H), 2.09 (m, 1H), 1.50 (s, 9H), 1.44 (s, 9H); ¹³C NMR (CDCl₃), δ 173.1, 170.6, 163.5 (d, J =1.3 Hz), 156.5, 153.9, 140.6 (d, J = 7.2 Hz), 138.8 (d, J = 7.8 Hz), 126.2 (d, J = 2.5 Hz), 122.9 (d, J = 4.2 Hz), 117.8, 117.5, 82.7, 81.3, 53.5, 31.5, 27.9, 27.9, 26.4; MS: [M + H]⁺ 427.1, [M + Na]⁺ 449.0, Calcd for C₂₀H₂₇N₂O₇F, 426.18.

Di-Tert-Butyl N-(3-Fluoro-4-Aminobenzoyl)-l-Glutamate 4. A solution of the nitro compound 3 (1.54 g, 3.61 mmol, 1 eq) in 25 ml of EtOAc was stirred with 10% Pd-C (134 mg) under an atmosphere of H₂ until complete reaction was indicated by TLC (1 h). The reaction mixture was filtered through Celite and evaporated under vacuum. The crude mixture was purified by column chromatography of silica (Hex:EtOAc, 2:1) to give the amino compound 4 (1.18 g, 85% yield) as a white crystal. Mp 83–84°C; ¹H NMR (d₆-DMSO), δ 8.23 (d, 1H, J = 7.6 Hz), 7.59 (dd, 1H, J = 1.8, 12.8 Hz), 7.52 (d, 1H, J = 1.8, 8.3 Hz), 6.78 (t, 1H, J = 8.7 Hz), 4.34 (m, 1H), 2.33 (t, 2H, J = 7.5 Hz), 2.13 (br s, 2H, NH₂), 2.02 (m, 1H), 1.92 (m, 1H), 1.39 (s, 9H), 1.24 (s, 9H); ¹³C NMR (d₆-DMSO), δ 171.5, 171.3, 150.5, 148.1, 139.8, 139.7, 124.6, 120.8 (d, J = 5.3 Hz), 114.5 (d, J = 4.7 Hz), 114.2, 114.0, 80.2, 79.6, 52.3, 31.3, 27.6, 27.5, 25.9; MS: [M + H]⁺ 397.0, [M + Na]⁺ 419.0, Calcd for C₂₀H₂₉N₂O₅F, 396.21.

Di-Tert-Butyl N-(3-Fluoro-4-Benzylaminobenzoyl)-l-Glutamate 5. Sodium cyanoborohydride (1.67 g, 26.54 mmol, 2.4 eq) was added in small portions to a stirred solution of 4 (4.39 g, 11.06 mmol, 1 eq) benzaldehyde (1.6 ml, 15.49 mmol, 1.4 eq) and bromocresol green (∼1 mg) in 50 ml of ethanol. After complete addition, acetic acid was added dropwise to adjust the solution to slightly acidic pH as determined by the indicator. The reaction solution was stirred for 24 h at RT, and during this time, acetic acid was added dropwise as necessary to maintain a slightly acidic solution. The acidic solution was extracted with EtOAc/Et₂O (1:1, v/v). The organic phase was washed with 10% NaOH, water, and saturated NaCl solution, dried over Na₂SO₄, and evaporated under vacuum to afford the crude product. Purification by column chromatography on silica gel (Hex:EtOAc, 3:1 to 1:1) gave the product 5 (3.54 g, 66%) as a white solid and recovered starting material 4 (909 mg, 21%). Mp 80–82°C; ¹H NMR (CDCl₃), δ 7.53 (dd, 1H, J = 2.0, 12.4 Hz), 7.45 (dd, 1H, J = 1.7, 8.4 Hz), 7.37 (m, 5 H), 6.78 (d, 1H, J = 7.3 Hz), 6.63 (t, 1H, J = 8.4 Hz), 4.63 (m, 1H), 4.42 (s, 2H), 2.39 (m, 1H), 2.32 (m, 1H), 2.20 (m, 1H), 2.02 (m, 1H), 1.59 (br s, 1H), 1.48 (s, 9H), 1.27 (s, 9H); ¹³C NMR (CDCl₃), δ 172.6, 171.4, 165.9, 151.8, 149.4, 139.6, 139.5, 138.0, 128.8, 127.5, 127.3, 124.0, 121.9 (d, J = 5.7 Hz), 113.9, 113.6, 110.9 (d, J =3.5 Hz), 82.3, 80.8, 52.7, 47.3, 31.6, 28.0, 27.5; MS: [M + H]⁺ 487.3, [M + Na]⁺ 509.3, Calcd for C₂₇H₃₅N₂O₅F, 486.25.

Di-Tert-Butyl N-(2-Fluoro-4-Benzylmethylaminobenzoyl)-l-Glutamate 6. Sodium cyanoborohydride (722 mg, 72.7 mmol, 1.5 eq) was added in small portions to a stirred solution of benzylamino compound 5 (3.54 g, 10.91 mmol, 1 eq) and formaldehyde (37% aqueous, 5.5 ml, 72.5 mmol, 10 eq) in 50 ml acetic acid at RT. The reaction solution was stirred for 1 h at RT. The acetic acid was evaporated under vacuum and the resulting residue was dissolved in EtOAc/Et₂O (1:1, v/v), washed with 1 N NaOH, water, and saturated NaCl solution, dried over Na₂SO₄, and evaporated under vacuum to afford the crude product. Purification by column chromatography on silica gel (Hex:EtOAc, 3:1) gave the product 6 (3.57 g, 98%).

Scheme B: An Alternate Three-Step Synthesis of Compound 6. Benzylmethylamine (1.5 ml, 11.85 mmol, 5 eq) was added dropwise to a solution of 3,4-difluorobenzonitrile 10 (330 mg, 2.37 mmol, 1 eq) in dry dimethylformamide (DMF) (5 ml) at 0°C. The reaction mixture was allowed to stir at RT overnight. The DMF was evaporated under vacuum and the residue chromatographed on silica gel (Hex:EtOAc, 9:1) to give 319 mg of pure product 11 (56%). This product was hydrolyzed with 50% of sodium hydroxide in methanol to reflux overnight. The reaction mixture was cooled to 0°C and concentrated HCl was added dropwise to pH = 2. The acidic solution was extracted with EtOAc. The organic phase was washed with saturated NaCl, dried over Na₂SO₄, and evaporated under vacuum to give 341 mg of the crude carboxylic acid 12 (99% yield). A stirred slurry of the carboxylic acid (341 mg, 1.31 mmol, 1 eq) in dry toluene (8 ml) was treated with thionyl chloride (144 μl, 1.97 mmol, 1.5 eq) and the mixture was heated under reflux for 2 h. The solution was cooled to RT and the solvent and excess thionyl chloride were evaporated under vacuum. A solution of the resulting acid chloride in 8 ml of CH₂Cl₂ was added dropwise to a stirred mixture of di-t-butyl l-glutamate hydrochloride (388 mg, 1.31 mmol, 1 eq) and Et₃N (0.4 ml, 2.62 mmol, 2 eq) in 8 ml of CH₂Cl₂ at 0°C. The reaction mixture was allowed to warm to RT and then stirred for 2 h. The solution was washed with water, dried over Na₂SO₄, and evaporated under vacuum. The crude product was purified by column chromatography on silica gel (Hex:EtOAc, 3:1) to give 633 mg (96% yield) of product 6 as a white solid. Mp = 71–72°C; IR (NaCl) 3346 (br m), 2978 (s), 2933 (m), 1730 (s), 1641 (s), 1616 (s), 1537 (m), 1509 (s), 1453 (m), 1368 (s), 1255 (s), 1153 cm⁻¹ (vs); ¹H NMR (CDCl₃), δ 7.54 (dd, 1H, J = 2.1, 14.3 Hz), 7.48 (dd, 1H, J = 2.1, 8.4 Hz), 7.34–7.24 (m, 5H), 6.89 (d, 1H, J = 7.4 Hz), 6.81 (t, 1H, J = 8.7 Hz), 4.65 (dt, 1H, Jt = 7.8 Hz, Jd = 4.5 Hz), 4.44 (s, 2H), 2.86 (s, 3H), 2.43 (ddd, 1H, J = 6.9, 8.3, 16.6 Hz), 2.31 (ddd, 1H, J = 6.5, 7.9, 16.6 Hz), 2.26–2.17 (m, 1H), 2.08–1.98 (m, 1H), 1.49 (s, 9H), 1.42 (s, 9H); ¹³C NMR (CDCl₃), δ 172.7, 172.3, 166.2 (d, J =1.3 Hz), 154.8, 152.4, 142.9 (d, J = 7.5 Hz), 138.3, 128.8, 128.1, 127.5, 127.4, 125.3 (d, J = 6.6 Hz), 123.9, 117.6 (d, J = 3.4 Hz), 116.1, 115.8, 82.5, 80.9, 58.7 (d, J = 6.8 Hz), 53.1, 39.2, 32.1, 28.4, 27.6; MS: [M + H]⁺ 501.2, [M + Na]⁺ 523.2, Calcd for C₂₈H₃₇N₂O₅F, 500.27; [α]_D²⁰ +8.1 (c 1.13, CHCl₃).

Di-Tert-Butyl N-(2-Fluoro-4-Methylaminobenzoyl)-l-Glutamate 7. Ammonium formate (1.12 g, 17.81 mmol, 5 eq) was added to a stirred suspension of benzylmethylamino compound 6 (3.57 g, 7.12 mmol, 1 eq) and 10% palladium on carbon (264 mg) in 50 ml of methanol at RT. The mixture was heated to reflux until TLC showed complete reaction. The reaction mixture was filtered through Celite and evaporated under vacuum. The crude mixture was purified by column chromatography of silica (Hex:EtOAc, 2:1) to give the amino compound 7 (2.82 g, 96% yield) as a white solid. Mp 95–96°C; IR (NaCl) 3377 (br), 2979, 2933, 1726, 1658, 1641, 1620, 1580, 1548, 1513, 1502, 1451, 1367, 1154 cm⁻¹; ¹H NMR (CDCl₃) δ 7.53–7.47 (m, 2H), 6.86 (d, 1H, J = 7.5 Hz), 6.62 (t, 1H, J = 8.5 Hz), 4.65 (m, 1H), 4.29 (br s, 1H, NH), 2.92 (d, 3H, J = 5.2 Hz), 2.43 (m, 1H), 2.31 (m, 1H), 2.22 (m, 1H), 2.03 (m, 1H), 1.49 (s, 9H), 1.42 (s, 9H); ¹³C NMR (CDCl₃) δ 172.5, 171.5, 166.0 (d, J = 2.1 Hz), 151.6, 149.3, 140.7, 140.6, 123.9, 121.2 (d, J = 5.8 Hz), 113.3, 113.1, 109.8 (d, J = 3.7 Hz), 82.1, 80.6, 52.6, 31.6, 29.6, 27.9, 27.87, 27.40; MS: [M + H]⁺ 411.1, [M + Na]⁺ 433.1, Calcd for C₂₁H₃₁N₂O₅F, 410.22; [α]_D²⁰ +10.5 (c 1.48, CHCl₃).

Di-Tert-Butyl N-(2-Fluoro-4-Amino-4-Deoxy-N-Methyl-Pteroyl)-l-Glutamate 8. The amine 7 (887 mg, 2.16 mmol, 1 eq) was stirred with bromomethyldiaminopteridine [which was prepared according to the method of Piper and Montgomery (8) as modified by Boyle and Pfleiderer (9); 80% purity, 1.02 g, 2.46 mmol, 1.1 eq] in 6 ml of dry DMAC at 90°C for 1 h. “Proton sponge” (Aldrich Chemical Co., Milwaukee, WI; 480 mg, 2.24 mmol, 1 eq) was then added and the reaction was continued for an additional hour. The solvent was removed under vacuum. The dark residue was treated with 50 ml of CHCl₃ containing 1 ml of Et₃N and adsorbed in silica gel. Column chromatography on silica gel (CHCl₃:EtOH, 4:1) provided almost pure product. Another column chromatography was performed to obtain 477 mg of pure product 8 (38% yield) as a yellow solid. Mp 160–164°C; ¹H NMR (d₆-DMSO), δ 9.09 (br s), 8.65 (s, 1H), 8.40 (d, 1H, J = 7.5 Hz), 7.87 (br s), 7.67 (m, 2H), 7.56 (br s), 7.06 (t, 1H, J = 8.7 Hz), 6.82 (br s), 4.62 (s, 2H), 4.29 (m, 1H), 3.00 (s, 3H), 2.31 (t, 2H, J = 7.3 Hz), 2.03–1.87 (m, 2H), 1.39 (s, 9H), 1.37 (s, 9H); ¹³C NMR (d₆-DMSO), δ 171.4, 171.1, 165.1 (d, J = 1.5 Hz), 162.7, 162.8, 155.0, 153.6, 151.2, 149.3, 145.5, 141.4, 141.3, 125.0 (d, J = 6.4 Hz), 124.3, 121.2, 117.6 (d, J = 3.6 Hz), 115.3, 115.1, 80.5, 76.7, 56.9 (d, J = 6.6 Hz), 52.4, 40.3, 31.3, 27.6, 27.6, 25.1; MS: [M + H]⁺ 585.4, [M + Na]⁺ 607.4, Calcd for C₂₈H₃₇N₈O₅F, 584.28.

N-(2-Fluoro-4-Amino-4-Deoxy-N-Methyl-Pteroyl)-l-Glutamic Acid 9. The protected fluoromethotrexate 8 (750 mg, 1.28 mmol) was dissolved in 12 ml of CH₂Cl₂ and 6 ml of trifluoroacetic acid (TFA) was added dropwise. After stirring for 2 h at RT, the reaction mixture was thoroughly evaporated under vacuum. The residue was dissolved in 0.01 N NaOH. The pH was the adjusted to 4.2–4.4 by addition of 1 N HCl giving a bright yellow precipitate. Filtration and washing with cold water provided the final product 9 (554 mg, 92% yield) as a bright yellow solid. Mp: 220–223°C (decomp.); ¹H NMR (d₆-DMSO), δ 8.65 (s, 1H), 8.44 (d, 1H, J = 7.6 Hz), 7.76 (br s, 1H), 7.69 (m, 2H), 7.42 (br s, 1H), 7.06 (t, 1H, J = 8.8 Hz), 6.76 (br s, 2H), 4.62 (s, 2H), 4.36 (m, 1H), 3.00 (s, 3H), 2.32 (t, 2H, J = 7.5 Hz), 2.07 (m, 1H), 1.92 (m, 1H); ¹³C NMR (d₆-DMSO), δ 173.9, 173.5, 165.1, 162.7, 162.5, 154.6, 153.7, 151.3, 145.9, 141.4, 141.4, 125.2, 125.2, 121.3, 117.7, 57.0, 51.9, 40.0, 30.5, 26.0; MS: [M + H]⁺ 472.9, Calcd for C₂₀H₂₁N₈O₅F, 472.16; [α]_D²⁰ +16.5 (c 0.87, 1 N NaOH).

Molecular Modeling. The Protein Data Bank (PDB) coordinate file for the thymidylate synthetase (TS)-methotrexate cocrystal structure was obtained from PDB (1AXW). The atom types for the inhibitor and cofactor dUMP were corrected, hydrogen atoms were added, and the protein C and N endgroups were fixed using the SYBYL/BIOPOLYMER module. Protein atomic charges were assigned with the Kollman all-atom charge set and inhibitor/cofactor charges were calculated using the Gasteiger-Hückel method. The complex was minimized using the Powell method, the Tripos Force Field, and 0.05 kcal/mol • Å rms gradient as the convergence criterion. All protein and cofactor heavy atoms (inherent to the crystal structure) were constrained in an aggregate during minimizations. Surfaces and cartoon diagrams were created within SYBYL using the MOLCAD surface dialog. Because we cannot directly compare the position of the ligand atoms in the X-ray structure with a minimized model, we required a minimized MTX:TS template. To create this model, the MTX:TS complex was minimized using the methods above. The conformation of MTX in the binding pocket changed only very slightly after minimization. MTX was replaced with FMTX and was again minimized as above.

An analogous study of the DHFR:MTX and FMTX complexes with NADPH was performed beginning with the X-ray structure of the MTX:DHFR complex of the Mycobacterium tuberculosis DHFR enzyme (10).

The cytotoxicity of FMTX was compared with that of the parent compound MTX against the MTX-sensitive human sarcoma cell line HT-1080 (ATCC, Rockville, MD). Cultured cells were maintained as monolayer cultures in RPMI 1640 culture medium supplemented with 10% FCS at 37°C under a humidified 5% CO₂ atmosphere. For cytotoxicity assays, monolayer cells were trypsinized and plated in six-well culture plates (10 cm² per well) at a density of 1000 cells/well. After a period of 48 h to allow for cell attachment and the establishment of cell proliferation, the medium was aspirated and replaced with fresh medium. During drug exposure, culture medium was either normal medium or thymidine-free medium to determine the effect of thymidine salvage (11, 12). Thymidine-free medium was prepared from normal medium via treatment with thymidine phosphorylase (Sigma, St. Louis, MO) for 60 min at 37°C followed by heat inactivation of the thymidine phosphorylase at 55°C and filtration (0.22 μm filter). MTX and FMTX stock solutions (10 mm) were prepared in isotonic saline with pH adjusted to 7.4. After a period of 24 h drug exposure, culture medium was aspirated and replaced with fresh normal medium. Cells were incubated for a further 72–96 h and cell viability was determined via the sodium 3′-[1-(phenylaminocarbonyl)-3,4-tetrazolium]-bis(4-methoxy-6-nitro) benzene sulfonic acid hydrate (XTT)/phenazine methosulfate (PMS) assay (13, 14). This spectrophotometric assay indicates the level of cellular biochemical redox activity in each culture well relative to control, untreated cells as a measure of cell viability. Cytotoxicity (repeated in triplicate) was evaluated from drug concentration response curves and is reported in terms of IC₅₀, the concentration of drug that reduces HT-1080 cell viability by 50%. Results for MTX and FMTX were compared via Student's t test.

A pilot in vivo ¹⁹F MR study was performed according to institutionally approved protocols for the safe and humane treatment of animals. A LNCaP (human prostate cancer) tumor xenograft in a nude mouse was used to determine in vivo MR visibility of FMTX. Tumor growth was initiated by the injection of 0.2 ml of a slurry of ∼10⁵ cells. The tumor cell slurry was inoculated s.c. into the left flank of 6-week-old male nude (athymic) mice.

The ¹⁹F chemical shift of FMTX was first determined at high field (B₀ = 9.4 T). FMTX dissolved in blood plasma was placed in the outer compartment of a coaxial 5-mm NMR tube with the inner compartment (3 mm outer diameter) containing TFA in D₂O. The ¹⁹FMTX resonance was observed at 46.4 ppm with respect to the TFA chemical shift reference.

FMTX was administered via i.v. bolus tail-vein injection at a dosage of 400 mg/kg, a dosage comparable to that used clinically in humans (15). The mouse was unanesthetized for ¹⁹F MR experiments. The mice crawl into a 60-ml syringe barrel (with air holes) which is used as an animal holder with the tumor protruding through a hole into a home-built 2 turn ¹⁹F MR surface coil. The inner diameter of the surface coil was 0.8 cm with tumor volume that offered maximum filling factor (200–350 mm³). The in vivo ¹⁹F MR study was performed at 188 MHz (B₀ = 4.7 T) in a wide-bore (33 cm diameter) small animal imaging system (GE Omega, Fremont, CA) with the tumor in a temperature control/susceptibility matching water bath (16). For the in vivo ¹⁹F MR experiment (pulse and acquire), acquisition parameters included a 60° pulse-width with a pulse repetition time (T_R) of 2 s, 10,000 Hz sweep-width, 1024 complex data points per free induction decay (FID), and a summation of 256 FIDs per spectrum (9 min/spectrum). An external chemical shift reference standard of 100 mm TFA in D₂O was held in an 18-μl glass microsphere.

To optimize the acquisition parameters for future studies, the spin-lattice relaxation time T₁ was determined (17). An estimate of the in vivo FMTX T₁ value was made for a phantom sample of 7 mm FMTX dissolved in blood plasma at 37°C and 4.7 T with the inversion-recovery pulse sequence. The resonance intensity data of the 180°-τ-90°-acquire pulse sequence was fit, as shown in Fig. 3, to the equation:

wherein the constant C allows for any error in the calibration of the 90° or 180° pulses. Resonance intensities were determined via modeling of the time-domain NMR signal data using the jMRUI software package run on a PC (18). From the fit to Eq. A, a T₁ value of 0.60 s was determined.

The synthetic route for FMTX was essentially the same as that of 3′-fluoroaminopterin (6). The synthesis was readily accomplished in eight steps starting from 3-fluoro-4-nitro-toluene 1 as outlined in Scheme A (Fig. 1). Thus, standard sodium dichromate oxidation of 1 afforded 2, which was subsequently converted into the acid chloride by treatment with thionyl chloride in refluxing toluene. Addition of the isolated acid chloride to di-tert-butyl-l-glutamate hydrochloride in dichloromethane containing two equivalents of triethylamine at 0°C gave amide 3 in 69% yield. At this point, the hydrogenolysis of the nitro group to the corresponding amine was done at atmospheric pressure and RT to give p-aminobenzoyl moiety 4 in 85% yield. The resulting amine was transformed into the corresponding benzyl amine 5 by reductive amination with benzaldehyde using the standard sodium cyanoborohydride methodology. With the benzyl group in place, monomethylation of the benzyl amine was easily achieved via a subsequent second reductive amination with formaldehyde to give methylbenzyl amine 6 in 98% yield. Hydrogenolysis using 10% palladium on carbon with ammonium formate in refluxing methanol gave methyl amine 7 in 90% yield. Condensation with 6-(bromomethyl)-2,4-pteridinediamine hydrobromide (8, 9) in dimethylacetamide at 90°C with proton sponge gave the fully assembled di-tert-butyl-3′-fluoromethotrexate 8 in 38% yield after chromatography. Subsequent deprotection with TFA in dichloromethane gave FMTX 9 in 92% yield.

An alternate synthesis of compound 6 (Fig. 2) was accomplished in four steps starting with 3,4-difluoro-benzonitrile (Scheme B). Initial displacement of the fluorine at C-4 with methyl benzylamine in DMF at RT afforded compound 11 in 98% yield. Hydrolysis of the nitrile with 50% sodium hydroxide gave the resulting acid 12 in 98% isolated yield. This acid was then converted to the final compound 9 as previously described in 99% yield.

The energy-minimized structures of the complexes of MTX and FMTX with the enzymes TS and DHFR demonstrate that the introduction of the fluorine label at the 3′ position of MTX results in only very small differences in the binding of the inhibitor in the antifolate-enzyme complexes. Fig. 4 shows the very small differences in the geometry of the inhibitors MTX and FMTX in these energy-minimized computer models. The modeling results also indicate only very slight differences in the binding energy between the MTX-enzyme and FMTX-enzyme complexes. The FMTX complexes appear to be slightly more favorable energetically, but not to a significant extent. Binding energies of the antifolate complexes with TS and DHFR differ by 1.4 and 3.7 kcal/mol, respectively. This suggests that FMTX may bind slightly better than MTX in both proteins.

Table 1 and Fig. 5 display the cytotoxicity data for the two antifolates, MTX and FMTX, against the MTX-sensitive human sarcoma cancer HT-1080. The cytotoxicity of FMTX is slightly greater than that of the unlabeled MTX although these differences are not statistically significant (P > 0.05). These biological results are to be expected based on the similar binding energies of MTX and FMTX for the target enzymes DHFR and TS predicted by molecular modeling. It is also apparent that the presence of exogenous thymidine in the culture medium reduces cytotoxicity.

Fig. 6 demonstrates that tumor uptake of FMTX is readily detectable in vivo with good temporal resolution (9 min per spectrum). The surface coil MR experiment localizes ¹⁹F signal to the sensitive volume of the coil (i.e., the volume filled with tumor), thus it is assured that the signal intensity represents concentration of FMTX in the tumor. In vivo, the chemical shift of FMTX is in the range from 46.5 to 46.9 ppm with respect to the external reference solution of TFA. In this pilot study, we did not follow the full time course of FMTX tumor wash-in and washout as the goal was simply to demonstrate MR visibility of the compound in vivo. Also, this set of spectra was acquired under conditions wherein longitudinal magnetization was essentially fully relaxed. Knowledge of the T₁ of FMTX in plasma (0.60 s, as determined here) provides a good first estimate of the in vivo T₁ value at 4.7 T.

The primary focus of this report is to detail the chemical synthesis of FMTX, demonstrate that it has biological activity (as determined by in vitro cytotoxicity studies) similar to that of MTX, and that the compound can be detected in vivo with ¹⁹F MRS in a pilot study. Both molecular modeling and in vitro cytotoxicity indicate that the presence of the fluorine label in FMTX causes minimal modification of the activity of the parent compound, MTX.

As noted, there have been other reports in the literature of synthesis of fluorine-containing analogues of MTX (6, 7). However, in studies that compared cytotoxic efficacy of MTX with one of these fluorine-containing analogues (in which the fluorine substitution was on the γ-glutamyl moiety of MTX, see Ref. 7), the cytotoxicity of the fluorine-containing species was found to be some three orders of magnitude lower. It was determined that this was the result of impaired poly-glutamylation of the fluorine-substituted species.

In a cohort of HT-1080 tumor xenografts that were subsequently sacrificed (n = 4, not shown), FMTX accumulated in the tumor for time points up to 234 min postinjection and remained at elevated levels until the conclusion of the ¹⁹F MR observation (270 min postinjection). Quantitation of the FMTX resonance in this cohort indicated that the maximum mean intratumor FMTX tumor concentration was 0.73 ± 0.32 mm (mean ± SE) at 234 min postinjection.

Many of the clinically observed modes of resistance to MTX therapy involve reduced uptake and/or reduced intracellular retention (19, 20). We are currently investigating differences in pharmacokinetic parameters as observed by ¹⁹F MR (maximum tumor concentration, rates of uptake, and washout) in a panel of MTX-sensitive and -resistant human sarcoma xenografts to determine the relationship between these pharmacokinetic parameters and therapeutic efficacy. The preliminary data (not shown herein) indicate that ¹⁹F MR observation of tumor uptake/washout may be useful in assessing the likelihood of tumor response to therapy. Sensitive tumors appear to achieve significantly greater intratumoral concentration of FMTX than do MTX-resistant tumor xenografts (up to 3 times higher).