The chemistry of the bifunctional alkylating agent phosphoramide mustard and model phosphoramides was probed by multinuclear NMR spectroscopy as a function of pH. Between pH 1 and 11, both the 31P and 15N resonances for phosphoramide mustard displayed a single monobasic titration curve with a pKa of 4.9. The protonation below pH 4.9 correlates with the loss in reactivity of the mustard. The 17O NMR spectrum of 17O-enriched phosphoramide mustard shows little change with pH. The data on the mustard was compared to 15N and 31P NMR data on 15N-enriched phosphoramidic acid, phosphorodiamidic acid, and phosphoric triamide. Contrary to the conclusions of previous studies, our combined 31P, 15N, and 17O NMR results are more consistent with N-protonation of phosphoramide mustard rather than an O-protonation. Theoretical calculations on the phosphoramidic acid, phosphorodiamidic acid, and phosphoric triamide show O-protonation to be more stable in the gas phase. For the latter two compounds, the calculations suggest that N-protonation may be the most stable protonated form in the aqueous phase. These findings influence our understanding of the structure-activity relationships of phosphoramide mustards.

The alkylating agent isophosphoramide mustard (IPM) spontaneously forms a relatively stable aziridine derivative which can be directly observed using NMR spectroscopy. The protonations of IMP and its aziridine were probed using 1H, 31P, 15N, and 17O NMR spectroscopy. The positions of the 31P, 15N, and 17O resonances of IPM between pH 2 and 10 each exhibit a single monobasic titration curve with the same pKa of 4.31 +/- 0.02. On the basis of a comparison with other compounds and our earlier work with phosphoramide mustard, the NMR results for IPM indicate that protonation occurs at nitrogen and not oxygen. Over this same pH range, each of the 1H, 31P, and 15N resonances of IPM-aziridine also show a single monobasic titration with a pKa of 5.30 +/- 0.09. The magnitude of the change in chemical shifts suggests that the protonation of the IPM-aziridine occurs at the ring nitrogen. Theoretical gas-phase calculations of PM, IPM, and IPM-aziridine suggest O-protonation to be more likely; however, aqueous phase calculations predict the N-protonated forms to be most stable. Furthermore, for PM and IPM-aziridine, which contain nonequivalent nitrogens, the theoretical calculations and experimental data both agree as to which nitrogen undergoes protonation. These results suggest that the IMP-aziridine remains unprotonated under physiological conditions and may, in part, explain the lower alkylating activity of IPM as compared to PM.

To investigate the mechanism(s) of bisalkylation by isophosphoramide mustard (IPM), IPM-beta,beta,beta',beta'-d(4) was synthesized and the products of its reaction with thiosulfate (at pD 7.0) were analyzed by NMR. By both (1)H and (13)C NMR, the distribution of deuterium in the products was consistent with bisalkylation through sequential aziridinyl intermediates [(NCH(2)CD(2)S):(NCD(2)CH(2)S) = 53:47]. Under the given reaction conditions, label scrambling as a result of thiosulfate acting as a leaving group was ruled out through control experiments. The data gave a calculated kinetic isotope effect of 0.97 per deuterium. For the initial aziridine species formed from IPM, ab initio quantum chemical calculations gave a hybridization value of sp(2.4)(-)(2.5) for each of the C-H bonds of the reaction centers, and this correlated with the observed inverse isotope effect. Other structure and bond order data were also determined for this aziridine intermediate and related compounds.

Hydrogenolysis of 3-(benzyloxy)cyclophosphamide (10) using Pd/C catalyst and ethyl acetate as solvent leads to the formation of 3-hydroxycyclophosphamide (3, approximately 20%) and cyclophosphamide (1, approximately 10%), accompanied by regioselective hydrogen-exchange reactions at the C-4 and C-5 positions in 3 and 1. A variety of oxidizing reagents and liver microsomal incubation failed to provide evidence (31P NMR) for conversion of 1 into 3, whereas identical incubation of 3 led to its reduction to 1. Compound 3 is stable at pH 6.5-8.2, 37 degrees C, and exhibits anticancer activity comparable to 1 when tested against L1210 leukemia in mice. Data are discussed with regard to a previously reported suggestion that metabolism of 1 may involved oxidation to give 3 followed by rearrangement of 3 to 2.