Prediction and experimental validation of solid solutions and isopolymorphs of cytosine/5-flucytosine

A computational search for polymorphs of cytosine, 5-flucytosine and a 1 : 1 mixture of the two substances not only rationalised the preferred packing arrangements but also enabled the finding and characterisation of cytosine/5-flucytosine solid solutions. The structures of the new solid forms were determined by combining laboratory powder X-ray diffraction data and computational modelling.


. Cytosine and 5-Flucytosine Tautomer Selection
The two investigated compounds are known to exist (at least in solution) in different tautomeric forms. A survey of the two compounds' structures present in the Cambridge Structural Database 1 revealed that so far only the keto (amino-keto) tautomer has been identified in solid state. Thus, only the keto tautomer was considered in our computational searches for anhydrate polymorphs. The structures were relaxed to a local minimum in intermolecular lattice energy, calculated from the FIT 5 exp-6 repulsion-dispersion potential and atomic charges, fitted to electrostatic potential around the PBE0/aug-cc-pVTz charge density using the CHELPG scheme. 6

Reminimisation: DMACRYS
For each of the searches the lowest energy structures (Table S1) were refined using DMACRYS 7 with a more realistic, distributed multipole model 8 for the electrostatic forces which had been derived using GDMA2 9 to analyse the PBE0/aug-cc-pVTz charge density.

Reminimisation: CrystalOptimizer
The orientation of the amino group (planar vs. pyramidal orientation) of the most stable structures (Table S1) of the two compounds was optimised with the program CrystalOptimizer. 10 Conformational energy penalties and isolated molecule charge densities were computed at the PBE0/aug-cc-pVTz level of theory.  The DFT-D calculations were carried out with the CASTEP plane wave code 11 using the Perdew-Burke-Ernzerhof (PBE) generalised gradient approximation (GGA) exchangecorrelation density functional 12 and ultrasoft pseudopotentials, 13 with the addition of a semiempirical dispersion correction, either the Tkatchenko and Scheffler (TS) model, 14 or Grimme06 (D2). 15 In a first step, the structures were geometry optimised using the TS dispersion correction. Brillouin zone integrations were performed on a symmetrised Monkhorst-Pack k-point grid with the number of k-points chosen to provide a maximum spacing of 0.07 Å −1 and a basis set cut-off of 780 eV. The self-consistent field convergence on total energy was set to 1x10 −5 eV. Energy minimisations were performed using the Broyden-Fletcher-Goldfarb-Shanno optimisation scheme within the space group constraints. The optimisations were considered complete when energies were converged to better than 2x10 −5 eV per atom, atomic displacements converged to 1x10 −3 Å, maximum forces to 5x10 −2 eV Å −1 , and maximum stresses were converged to 1x10 −1 GPa. The energies for the structures were recalculated, without optimisation, with the number of k-points chosen to provide a maximum spacing of 0.07 Å −1 and a basis set cut-off of 780 eV, using the D2 dispersion correction. Isolated molecule minimisations to compute the isolated cytosine (keto tautomer) and 5-flucytosine (keto tautomer, Ugas) were performed by placing a single molecule in a fixed cubic 35x35x35 Å 3 unit cell, then optimised with the same settings as used for the crystal calculations.
H  F exchange was systematically applied to the experimental structures C-I, C-II, F-I and F-II to produce isostructural cytosine, 5-flucyosine and mixed crystal structures thereof. The structures were minimised as described above.

Computationally Generated Low-Energy Structures
All calculated structures are available in .res format from the authors on request.  a Structure ID: f -5-flucytosine and rank CrystalPredictor. The CASTEP minimised structures were checked for higher symmetry using PLATON. 16 fC-I and fC-II -isostructural with C-I and C-II. b Packing Index (%) calculated using PLATON.

Representation of the Experimental Structures
The computational models were successful in reproducing the experimental anhydrate and hydrate structures of cytosine (Table S5 taken    The errors on the stated temperatures (extrapolated onset temperatures) and enthalpy values were calculated at the 95% confidence intervals (CI) and are based on three measurements.
Thermogravimetric Analysis (TGA) was carried out with a TGA7 system (Perkin-Elmer, Norwalk, CT, USA) using the Pyris 2.0 Software. Approximately 4 -6 mg of sample was weighed into a platinum pan. Two-point calibration of the temperature was performed with ferromagnetic materials (Alumel and Ni, Curie-point standards, Perkin-Elmer). A heating rate of 5 °C min -1 was applied and dry nitrogen was used as a purge gas (sample purge: 20 mL min -

Infrared Spectroscopy
Infrared spectra were recorded with a diamond ATR (PIKE GaldiATR) crystal on a Bruker Vertex 70 spectrometer (Bruker Analytische Messtechnik GmbH, D). The spectra were recorded in the range of 4000 to 30 cm -1 with an instrument resolution of 2 cm -1 (256 scans per spectrum).

Powder X-ray Diffraction
Powder X-ray diffraction (PXRD) patterns were obtained using an X'Pert PRO diffractometer (PANalytical, Almelo, NL) equipped with a / coupled goniometer in transmission geometry, programmable XYZ stage with well plate holder, Cu-K1,2 radiation source with a focussing, a 0.5° divergence slit and a 0.02° Soller slit collimator on the incident beam side, a 2 mm antiscattering slit and a 0.02° Soller slit collimator on the diffracted beam side mirror and a solid state PIXcel detector. The patterns were recorded at a tube voltage of 40 kV and tube current of 40 mA, applying a step size of 2 = 0.013° with 400 s per step in the 2 range between 2° and 70°.

Solvent Screen
The experimental screen for mixed cytosine/5-flucytosine solid forms encompassed Crystal 16® cycling experiments, slurry experiments in selected organic solvents, dehydration and sublimation experiments.

Crystal16® Cycling Experiments
CF-II (10 -15 mg) and solvents (1.0 mL) were dispensed into 1.8 mL vials. The vials were transferred to a Crystal16™ parallel crystalliser, equipped with programmable heating/cooling, magnetic stirring and turbidity sensors. The suspensions were stirred at 900 rpm at 5 °C for 30 minutes, heated to X °C at 0.1 °C min 1 , equilibrated for another 10 minutes, then cooled to 5 °C at 0.1 °C min 1 under stirring. The cycle was repeated a second time with stirring only upon heating. The solid products were isolated by filtration and analysed with PXRD (Table S7).

Slurry Experiments
Suspensions of cytosine (C-I) and 5-flucytosine (F-I and F-II mixture) were prepared in methanol, ethanol, dimethyl formamide, dimethyl sulfoxide and 1-butanol and then stirred in the temperature range from 10 to 30 °C (40 °C -1BuOH) for at least 96 hours. The wetcakes were analysed by PXRD (measured between two mylar foils to prevent solvent loss).
The solvate stoichiometry was determined with TGA.

Dehydration Experiments
Dehydration studies of the monohydrate solid solution were performed and the resulting product was analysed with PXRD and TGA (Table S13).

Sublimation Experiments
Sublimation experiments of the solid solutions lead to phase separation ("purification") and 5-flucytosine single crystals were obtained ( Figure S1). Cell parameters for CF-I and CF-II, details of the data collection and a list of atomic parameters can be found in Tables S10-S12. Observed and calculated PXRD patterns are shown in Figure   S2.    Figure S2. Observed (black points), calculated (red line) and difference (diff.) profiles for the Rietveld refinements of (a) CF-I and (b) CF-II. Green tick marks denote the peak positions.

Methanol Solvate Powder X-Ray Diffraction
According to the PXRD patterns ( Figure S3) the methanol solvate is not phase pure, but contains traces of the monohydrate. The methanol solvate is isostructural with the 5flucytosine hemimethanol solvate (MEBQOA 22 ). Figure S3. Comparison of anhydrous (red), monohydrate (blue), methanol solvate (green) and desolvated methanol solvate (violet) PXRD patterns. Note that the methanol solvate pattern is not phase pure.

Thermogravimetric Analysis
The methanol solvate loses its solvent molecules immediately when exposed to dry conditions (N2). The TGA curve shows a two-step mass loss, corresponding to the loss of methanol and water (sample not phase pure). The calculated mass loss for phase pure methanol hemisolvate solid solutions is listen in Table S13. Figure S4. TGA curve of methanol solvate/monohydrate mixture. 14.420 a Calculated weight loss relative to wet substance (substance and solvent). b Calculated weight loss relative to dry substance (substance without solvent).

Ethanol Solvate Powder X-Ray Diffraction
The slurry method produced an ethanol solvate with anhydrate II impurities ( Figure S5). The

Thermogravimetric Analysis
The ethanol hemisolvate is, compared to the methanol solvate, stable. Desolvation occurs at temperatures > 80 °C. The measured mass loss in the TGA experiments is lower than expected for a hemisolvate (Table S14). This can be related to the fact that the solvate was contaminated with CF-II.

Dimethyl Formamide Solvate
Powder X-Ray Diffraction The PXRD characteristics of the DMF solvate ( Figure S7) of the solid solution suggests that it is isostructural with the 5-flucytosine DMF monosolvate (see ref. 17). Figure S7. Comparison of anhydrous (red), monohydrate (blue), DMF solvate (green) and PXRD patterns of storage experiments of the DMF solvate (violet).

Thermogravimetric Analysis
The mass loss derived from TGA experiments of the DMF solvate confirms a monosolvate stoichiometry ( Figure S8 & Table S15).  a Calculated weight loss relative to wet substance (substance and solvent). b Calculated weight loss relative to dry substance (substance without solvent).

Dimethyl Sulfoxide Solvate
The slurry method produced a DMSO solvate with CF-II impurities ( Figure S9). The PXRD characteristics of the DMSO solvate of the solid solution suggests that it is isostructural with the 5-flucytosine DMSO solvate (DUKWAI 23 ).
Powder X-Ray Diffraction Figure S9. Comparison of anhydrous (red), monohydrate (blue), DMSO solvate (green) and storage experiments of DMSO solvate (violet) PXRD patterns. Note that the DMSO solvate patters is not phase pure.

Thermogravimetric Analysis
The measured mass loss in the TGA experiments is lower than expected for a monosolvate stoichiometry (Table S16), which can be related to the fact that the solvate was not phase pure but contaminated with CF-II. Figure S10. TGA curve of DMSO monosolvate/CF-II.