Crystal structures of chiral 2-[bis(2-chloroethyl)amino]-1,3,2-oxazaphospholidin-2-one derivatives for the absolute configuration at phosphorus

The structures and absolute stereochemistry of two pairs of diastereomeric nitrogen mustards related to the chemotherapeutic cyclophosphamide were determined to test 31P NMR chemical shift trends proposed based on the spatial relationship of the bis(2-chloroethyl)amine moiety and the chiral substituent of the amino alcohol.


Chemical context
Bis(2-chloroethyl)amine moieties, also known as a 'nitrogen mustard', are of interest due their ability to alkylate DNA, which hinders the cellular growth and replication of cancer cells (Einhorn, 1985). 2-[Bis(2-chloroethyl)amino]-1,3 2 ,2oxazaphosphinane 2-oxide, commercially sold as cyclophosphamide, features such a nitrogen mustard moiety and is registered as an FDA-approved chemotherapeutic due to its cytotoxic ability. The bioactivation mechanism of cyclophosphamide is well known. Hydroxylation occurs on the C-4 position through cytochrome P450 type enzymes and the cyclophosphamide -eliminates into acrolein and an enantiomeric mixture of the cytotoxic phosphoramide mustard (Takamizawa et al., 1975;Borch & Millard, 1987;Sladek, 1988). Studies support an enantioselective metabolism via the administration of enantiomerically pure cyclophosphamide, as expected for an enzyme-catalyzed reaction (Cox et al., 1976;Fernandes et al., 2011;Castro et al., 2016). Therefore, it is of pharmaceutical interest to be able to readily identify the absolute configuration at phosphorus of cyclophosphamide and other related nitrogen mustard derivatives.
Diastereomeric 2-[bis(2-chloroethyl)]-1,3,2-oxazaphospholidin-2-ones, a five-membered ring derivative of cyclophosphamide, have been previously synthesized from l-and d-serine, but lacked X-ray diffraction data to determine the absolute configuration at the P atom (Foster, 1978;Jackson et al., 1992). Instead, the spectroscopic trends and X-ray ISSN 2056-9890 diffraction analysis of an l-serine-derived 2-methoxy-1,3,2oxazaphospholidin-2-one was applied and the absolute configuration was determined by analogy (Thompson et al., 1990). It was described that oxazaphospholidinones with a downfield 31 P NMR chemical shift had a syn configuration with respect to the exocyclic methoxy group and the chiral substituent of the amino alcohol, and vice versa for the anti configuration.
Herein we report the synthesis and absolute configuration at phosphorus of chiral 2-[bis(2-chloroethyl)amino]-1,3,2oxazaphospholidin-2-ones in attempts to support these spectroscopic trends for the analysis of future potentially chemotherapeutic analogues. Bis(2-chloroethyl)amine phosphoramidic dichloride was synthesized following the experimental procedure described by Friedman & Seligman (1954). Enantiomerically pure chiral amino alcohols were purchased and used to synthesize pairs of diastereomeric oxazaphospholidinones, which allowed for easy separation via flash column chromatography.

Structural commentary
No single crystals of 3a of X-ray diffraction quality could be obtained, and compound 2a was isolated as an oil. Compounds 2b and 3b, however, have been analyzed by single-crystal diffraction (Figs. 1 and 2). The molecular structures of 2b and 3b are similar. The five-membered rings in both structures feature the expected envelope conformation, with the flap at the C atom connecting to the phenyl and isobutyl groups, respectively. An overlay of the two structures, guided by the position of the phenyl and isobuytl groups (Fig. 3), indicates that the positions of the aza and oxo groups are swapped between 2b and 3b. Another slight difference between the conformations between the two rings is evident, caused by the close to planar configuration of the methylamine N atom of 2b (the sum of angles around N1 is 359.97 ), giving 3b a slightly more 'buckled' appearance than 2b. The chloroethyl moieties in 3b are extended all-trans. In 2b, one is also trans, while the other is gauche with an N2-C11-C12-Cl1 torsion angle of À65.89 (9) .
The conformation of both 2b and 3b appear at first sight to be stabilized by a number of weak intramolecular hydrogenbond-like interactions. In 2b, this involves C12-H12BÁ Á ÁO1 and C11-H11BÁ Á ÁN1, with atoms O1 and N1 being the O and N atoms of the oxazaphospholidin-2-one five-membered ring (see Table 1). In 3b, similar interactions are observed for C8-H8BÁ Á ÁO1 and C7-H7AÁ Á ÁN1. Bond lengths and angles for these interactions are, however, quite unfavorable (see Table 2). In particular, atom N1 in 2b, being essentially planar and sp 2 -hybridized, appears to be an unlikely acceptor for an actual hydrogen bond. The observed close contacts are most likely not significantly contributing to the stability of the molecular geometry realized in the solid state.

Figure 1
Displacement ellipsoid representation of a molecule of 2b (50% probability level), with the atom-numbering scheme. meters = 0.000 (8) and 0.07 (4), respectively] to test whether their determination from 31 P NMR chemical shift data based on the spatial relationship of the bis(2-chloroethyl)amine moiety and the chiral substituent of the amino alcohol does hold true (Thompson et al., 1990). The single-crystal X-ray structures of 2b and 3b tentatively support the literature trends based on their 31 P NMR chemical shifts. The chiral center(s) of the amino alcohol are syn to the nitrogen mustard moiety and the absolute configurations at phosphorus were found to both be S for 2b and 3b [see Favre & Powell (2014) for assignment of absolute structure for hypervalent atoms such as P or S in tetrahedral geometry]. The 31 P NMR data are shifted slightly downfield compared to their anti diastereomers 2a and 3a, thus confirming the trend proposed by Thompson et al. (1990). The absolute shift values are, however, rather small: 1.40 ppm for the pair of 3a and 3b, and nearly no shift is observed for the pair of 2a and 2b (0.33 ppm) (see Experimental section for all NMR data). Whether the assignment of absolute structure is reliable enough to be used for other related molecules in the absence of structural data from X-ray diffraction is not clear based on the data at hand. For a more reliable estimate, data from a larger library of compounds are needed.

Supramolecular features
Molecule 2b does not feature any acidic H atoms and, as such, does not have any strong hydrogen bonds. The O atom of the phospholidinone unit does, however, act as an acceptor for several C-HÁ Á ÁO hydrogen-bond-like interactions, originating from two methylene and one aromatic C-H unit of neighboring molecules (see Table 1 for metrical details and symmetry operators). The three C-HÁ Á ÁO interactions surrounding O2 are about equally spread, thus giving the O atom of the P O unit a pseudo-tetrahedral environment made up of the P atom on one side, and the three C-H units on the other three. A C-HÁ Á Á interaction, involving C10-H10A towards the density of the benzene ring at (x À 1 2 , Ày + 1 2 , Àz + 1), is also observed, but no significant C-HÁ Á ÁCl interactions and nostacking are found. The combined C-HÁ Á ÁO and C-HÁ Á Á interactions connect molecules into a three-dimensional lattice (Fig. 4).
Compound 3b does, in contrast to 2b, have an acidic functional group, the amide N-H moiety, that is capable of forming a medium-to-strong hydrogen bond. Intermolecular interactions in the structure of 3b are indeed dominated by an N-HÁ Á ÁO hydrogen bond between the amide H atom and the phospholidinone O atom. The graph-set motif for a single interaction is C(4), connecting individual molecules into infinite chains that wrap around a twofold screw axis parallel to the b-axis direction (Fig. 5). The spirals of molecules thus formed are further stabilized by a C-HÁ Á ÁO interaction between C2 and phospholidinone atom O1, and by a weak C-HÁ Á ÁN interaction between atoms C1 and N1 down the chain direction (Fig. 5). Neighboring spiral chains are connected through C-HÁ Á ÁCl interactions involving H8A of one of the methylene groups and Cl1.  Table 2 Hydrogen-bond geometry (Å , ) for 3b. Symmetry codes: (i) Àx þ 2; y À 1 2 ; Àz þ 1; (ii) Àx þ 2; y þ 1 2 ; Àz þ 1; (iii) x; y þ 1; z; (iv) Àx þ 1; y þ 1 2 ; Àz.

Figure 3
Overlay of molecules 2b and 3b (50% displacement ellipsoid probability level). R.m.s. value based on atoms of the five-membered ring and oxygen is 0.111 Å . Color coding: P orange, O red, N blue, Cl green, and C light purple for 2b and light blue for 3b.

Figure 4
Packing arrangement and intermolecular interactions of 2b (50% probability level). Intermolecular contacts are shown as dashed lines (light blue for C-HÁ Á ÁO and purple for C-HÁ Á Á).

Synthesis and crystallization
5.1. Bis(2-chloroethyl)phosphoramidic dichloride, 1 Bis(2-chloroethyl)amine hydrochloride (3.00 g, 16.77 mmol) was suspended in freshly distilled phosphoryl chloride (10 ml, 107 mmol) in a 50 ml round-bottomed flask and heated under reflux overnight. Once all the solids were completely dissolved, excess phosphoryl chloride was distilled off to leave a darkbrown oily residue. The residue was dissolved in an excess of a mixture of petroleum ether-acetone (1:1 v/v), while in a 323 K hot water bath. The hot solution was then filtered to remove any solids and the solvent was removed via rotary evaporation to yield an off-white solid. The solid was recrystallized using a 1:1 (v/v) solution of petroleum ether-acetone to afford phosphoramide mustard 1 (4.04 g, 79.4%) as an off-white crystalline solid (m.p. 327-328 K). 31   Phosphoramide mustard 1 (0.647 g, 2.50 mmol), (1R,2S)-(À)-ephedrine (0.375 g, 2.51 mmol), toluene (20 ml) and triethylamine (0.75 ml, 5.38 mmol) were added to a 50 ml roundbottomed flask at 275 K under an argon atmosphere. The solution was then allowed to stir and warm to room temperature overnight. The reaction mixture was vacuum filtered through 2.0 cm of Celite packed onto a fritted glass funnel and was washed with an additional 60-80 ml of dichloromethane. The solvent was removed via rotary evaporation, which yielded a viscous yellow oil. The oil was purified by flash column chromatography (110 g silica, 100% ethyl acetate, R F = 0.50 and 0.33 in 100% ethyl acetate) and afforded oxazaphospholidinones 2a and 2b (combined yield 0.54 g, 64.6%), based on their order of elution. Approximately 25 mg of oxazaphospholidinone 2b was dissolved in 2 ml of ethyl acetate and allowed to slowly evaporate over several days at room temperature. This yielded colorless crystals for single-crystal X-ray diffraction.
Fast diastereomer (2a)  Packing arrangement and intermolecular interactions of 3b (50% probability level). Hydrogen bonds are shown as dashed lines (blue for N-HÁ Á ÁO, light blue for C-HÁ Á ÁO, and red for C-HÁ Á ÁCl). Molecules 'wrap' around the twofold axis at [0, y, 1 fritted glass funnel and was washed with an additional 60-80 ml of ethyl acetate. The solvent was removed via rotary evaporation, which yielded a viscous yellow oil. The oil was purified by flash column chromatography (60 g silica treated with 1% triethylamine, 100% ethyl acetate, R F = 0.29 and 0.17 in 100% ethyl acetate) to afford oxazaphospholidinones 3a and 3b (combined yield 0.22 g, 72.8%), based on their order of elution. Approximately 25 mg of oxazaphospholidinone 3b was dissolved in 2 ml of ethyl acetate and allowed to slowly evaporate over several days at room temperature. This yielded colorless crystals for single-crystal X-ray diffraction.

Refinement
H atoms attached to C and N atoms were positioned geometrically and constrained to ride on their parent atoms. C-H bond lengths were constrained to 0.95 Å for aromatic C-H groups. Aliphatic CH, CH 2 , and CH 3 groups were constrained to C-H bond lengths of 1.00, 0.99, and 0.98 Å , respectively. The position of the amino H atom was refined and the N-H distance restrained to 0.88 (2) Å . Methyl H atoms were allowed to rotate, but not to tip, to best fit the experimental electron density. U iso (H) values were set to a multiple of U eq (C), with 1.5 for CH 3 and 1.2 for N-H, C-H, and CH 2 units. Crystal data, data collection and structure refinement details are summarized in Table 3

Special details
Geometry. All esds (except the esd in the dihedral angle between two l.s. planes) are estimated using the full covariance matrix. The cell esds are taken into account individually in the estimation of esds in distances, angles and torsion angles; correlations between esds in cell parameters are only used when they are defined by crystal symmetry. An approximate (isotropic) treatment of cell esds is used for estimating esds involving l.s. planes.

Special details
Geometry. All esds (except the esd in the dihedral angle between two l.s. planes) are estimated using the full covariance matrix. The cell esds are taken into account individually in the estimation of esds in distances, angles and torsion angles; correlations between esds in cell parameters are only used when they are defined by crystal symmetry. An approximate (isotropic) treatment of cell esds is used for estimating esds involving l.s. planes. Refinement. The position of the amine H atoms was refined and the N-H bond distance was restrained to 0.88 (2) Angstrom.