5-Fluorocytosine / Isocytosine Monohydrate. The First Example of Isomorphic and Isostructural Co-Crystal of Pyrimidine Nucleobases

: To date, despite the crucial role played by cytosine, uracil, and thymine in the DNA / RNA replication process, no examples showing isomorphic and isostructural behavior among binary co-crystals of natural or modiﬁed pyrimidine nucleobases have been so far reported in the literature. In view of the relevance of biochemical and pharmaceutical compounds such as pyrimidine nucleobases and their 5-ﬂuoroderivatives, co-crystals of the molecular complex formed by 5-ﬂuorocytosine and isocytosine monohydrate, C 4 H 4 FN 3 O · C 4 H 5 N 3 O · H 2 O, have been synthesized by a reaction between 5-ﬂuorocytosine and isocytosine. They represent the ﬁrst example of isomorphic and isostructural binary co-crystals of pyrimidine nucleobases, as X-ray di ﬀ raction analysis shows structural similarities in the solid-state organization of molecules with that of the (1:1) 5-ﬂuorocytosine / 5-ﬂuoroisocytosine monohydrate molecular complex, which di ﬀ ers solely in the H / F substitution at the C5 position of isocytosine. Molecules of 5-ﬂuorocytosine and isocytosine are present in the crystal as 1 H and 3 H -ketoamino tautomers, respectively. They form almost coplanar WC base pairs through nucleobase-to-nucleobase DAA / ADD hydrogen bonding interactions, demonstrating that complementary binding enables the crystallization of speciﬁc tautomers. Additional peripheral hydrogen bonds involving all available H atom donor and acceptor sites of the water molecule give a three-dimensional polymeric structure. In the absence of H · · · F hydrogen-bonding interactions, the robustness of the supramolecular architectures based on three-point recognition synthons is responsible for the existence of isostructurality between the two molecular complexes.


Introduction
In the formation of binary co-crystals, a second substance may be incorporated into the crystal lattice by replacing the molecules of the first component, leaving unit-cell dimensions and space groups unaltered (isomorphism). This can happen particularly if molecules of the two substances are structural homologues. However, isomorphism does not necessarily imply isostructurality, i.e., the situation where two crystal structures of different chemical composition exhibit the same spatial arrangement of molecules, in the formation of crystals or co-crystals [1,2]. Although small changes in the molecular structure can significantly affect molecular disposition in the crystal, specific functional groups or atoms can be interchanged without modifying crystal packing, provided that such groups (or atoms) are not involved in hydrogen bonds [3]. Concerning the exchange of the hydrogen/fluorine atoms, a systematic investigation in the Cambridge Structural Database (CSD) of almost 125,000 pairs of crystal structures yielded 645 pairs containing molecules which differed only in the substitution of H with F atoms, and 21% of them showed structural similarities in the solid-state organization [4].
Considering this, and the fact that the properties of molecular substances in the solid state depend essentially on their chemical nature, their stoichiometry and their reciprocal disposition in space, from the structural chemistry point of view pyrimidine nucleobases and their fluorosubstituted derivatives deserve special attention. On one side, the fundamental importance of cytosine, uracil, and thymine for structural molecular biology and supramolecular chemistry has been well-known for decades. On the other hand, it has been shown that the formation of haloderivatives of cytosine in DNA can be responsible for several modifications in the structure of DNA observed in human tumors [5], and mass spectrometry has been applied to detect halocytosines as free nucleobases to evaluate cytosine halogenation as an endogenous mutagenic pathway in human tissue [6]. Therefore, it is surprising that hardly any example showing isomorphic and isostructural behaviors between co-crystals of natural pyrimidine nucleobases and co-crystals of their fluorosubstituted derivatives can be found in the literature. In addition, although isostructurality is relatively common in crystals of multi-component systems, the author's literature search found only four cases of crystal packing similarity of centro/noncentrosymmetric co-crystals formed by pyrimidine nucleobases/5-fluorosubstituted derivatives with polymorphic centro/noncentro symmetric crystals of the 5-fluorosubstituted coformers. These four examples (Scheme 1), in which there was no indication of C-H· · · F intermolecular interactions, refer to the crystal structures of: i) the monoclinic P2 1 /c forms of monohydrate (1:1) 5-fluorocytosine/cytosine, SAVKIM [7], and monohydrate 5-fluorocytosine, BIRMEU02 [8]; ii) the monoclinic P2 1 /n forms of (1:1) 5-fluorocytosine/cytosine, ADAJAT [9], and 5-fluorocytosine, MEBQEQ [8]; iii) the tetragonal P4 1 2 1 2 forms of (1:1) 5-fluorocytosine/cytosine, ADAJAT01 [9], and 5-fluorocytosine, MEBQEQ01 [8]; iv) the orthorhombic Pnma forms of (1:1) N,N-dimethyluracil/N,N-dimethyl-5-fluorouracil, N,N-dimethyl-5-fluorouracil, OKATEB and KAMSAS01, respectively [10], and N,N-dimethyluracil, DMURAC01 [11]. Notably, in all the crystal structures of the isostructural pairs SAVKIM/BIRMEU02, ADAJAT/MEBQEQ and ADAJAT01/MEBQEQ01 5-fluorocytosine and cytosine are present as 1H-ketoamino tautomers. Consequently, the DAA/ADD triple hydrogen bond interaction between nucleobases is precluded, and dimerization is obtained through two independent DA/AD interactions based on the same constituent hydrogen bonds: (ring)N-H· · · N(imino) and (amino)N-H· · · O (carbonyl) (Figure 1). In light of previous observations, this paper reports the single-crystal X-ray structure of co-crystals of (1:1) 5-fluorocytosine/isocytosine monohydrate, (I). The solid-state arrangement of molecules of (I) has then been compared with the structurally related fluorocytosine/5-fluoroisocytosine monohydrate molecular complex, PANLAS [12]. The latter molecular complex differs from (I) in changing F/H atoms at the 5 position of isocytosine (Scheme 2). The choice of the components of this binary co-crystal was made from consideration of the following points.
5-Fluorocytosine (Flucytosine) is a synthetic monofluorinated analog of cytosine. This compound can exist in different tautomeric forms, due to the presence of solvent-exchangeable protons. In the crystal, it exhibits polymorphism and is known only in the 1H-ketoamino form, the most stable tautomer in aqueous solution [13]. It is structurally very similar to its parent nucleobase and satisfies steric requirements at enzyme receptor sites, therefore it is commonly used as antimetabolite. Flucytosine can act as an antifungal agent, as it is deaminated by cytosine deaminase of the fungal cells to give 5-fluorouracil, resulting in RNA miscoding. 5-Fluorocytosine also plays an important role in the treatment of different types of cancer, although its use is limited by severe side effects [14].
Isocytosine (2-aminouracil), an isomer of cytosine, deserved special attention in experiments with an artificial pyrimidine base pair to expand the genetic alphabet of DNA, and to understand how the complementarity originally appeared in the form of the G-C [15]. Isocytosine is subject to prototropic tautomerism between the 1H and 3H-ketoamino forms, and it can form in solid-state intermolecular complexes with one or the other tautomer. In the 3H-ketoamino form, the most stable tautomer in polar solvents [16,17], isocytosine undergoes base pairing in a WC manner with unnatural isoguanine, and in a reversed WC manner with natural guanine, and has been used for structural studies of nucleic acids [18] as well as for hydrogen-bonding in nucleobases [19]. It has been shown that isocytosine could provide a base for a putative new prodrug 5-fluoroisocytosine, when used together with a putative isocytosine deaminase [20].
In this work, analysis of the degree of similarity of (I) with PANLAS demonstrated that (1:1) 5-fluorocytosine/5-fluoroisocytosine monohydrate was the first example of isomorphic and isostructural co-crystals of pyrimidine nucleobases. The experimental results were complemented with Hirshfeld surface analysis to evaluate the contribution of different intermolecular interactions in the crystal packing.
Interestingly, in PANLAS, 5-fluoroisocytosine is present as a 3H-ketoamino tautomer, a tautomeric form never found in crystals of the isomer 5-fluorocytosine but observed for pure isocytosine. This tautomer favors pairing in the WC manner with 5-fluorocytosine through an intermolecular triple hydrogen bond. Any attempts to grow crystals to determine the solid-state structure of pure 5-fluoroisocytosine, which has not been reported so far, failed.

Crystal Preparation
The 5-fluorocytosine and isocytosine were purchased from Aldrich (99% purity) and subjected to further purification by successive sublimation under reduced pressure. For the mechanochemical synthesis of co-crystals, equimolecular amounts (0.1 mmol in 1:1 stoichiometric ratio) of pure products were taken in an agate mortar and pestle, and then mixed and ground manually without the addition of any solvent. Crystallization of the ground powder was adjusted in methanol. The resulting solution (1 mL) was heated at 60 • C with continuous stirring for 6 h under reflux, and then slowly cooled to room temperature and filtered. Single crystals were obtained via slow evaporation of methanol solution at room temperature after ca three days.

Single Crystal Structure Analysis
The details of crystal data, data collection and structure refinement are summarized in Table 1. The data set was collected on an Xcalibur S CCD diffractometer (graphite-monochromated Mo Kα radiation, λ = 0.710689 Å) at room temperature using the CrysAlisPro software package (Rigaku Oxford Diffraction, Yarnton, United Kingdom, 2018) [37]. The crystal structure was solved by direct methods using SIR2004 [38]. All the non-hydrogen atoms were refined anisotropically by the full-matrix least-squares method based on F 2 using SHELXL-2014/7 [39], within the WinGX system [40]. All H atoms were found through difference Fourier, but for refinement all C-bound H atoms were placed in calculated positions, with C-H = 0.93 Å and U iso (H) = 1.2U eq of the parent C atom, and refined as riding on the adjacent atoms. Positional and isotropic thermal parameters of H atoms of the heteroatoms were freely refined, giving N-H and O-H (water molecule) distances in the range 0.89 (3)-0.94 (3) and 0.87 (4)-0.88 (4) Å, respectively. The molecular and packing diagrams were prepared using the Mercury 3.9 program package [41]. The Hirshfeld surface analysis was carried out using Crystal Explorer17 (University of Western Australia, Crawley, AU, Australia [42]

Structural Analysis
The title compound, (I), crystallized in the triclinic centrosymmetric space group P-1, and the asymmetric unit comprised of two molecules of 5-fluorocytosine and isocytosine and one water molecule of crystallization ( Figure 2). In (I), as found for pure coformers, only the aminooxo tautomers were observed, with the N1a and N3 positions carrying the acidic proton. As already mentioned, in the solid state 5-fluoroisocytosine was present only in the 1H-ketoamino form, whereas isocytosine had two stable 1H and 3H-ketoamino tautomers. In the title compound, the presence of isocytosine as 3H-ketoamino tautomer favored the formation of almost-coplanar WC base pairs with 5-fluorocytosine (the two molecules being inclined by 5.4 (1) • to one another) through a DAA/ADD hydrogen-bonding pattern (three-point interaction, TPI) of R 2 2 (12) graph-set motif [43][44][45]. The values of bond lengths and angles agreed with those of the corresponding geometrical parameters found in the crystal structures showing 5-fluorocytosine monohydrate [8,23,24] and isocytosine [26,27] as 1H and 3H-ketoamino tautomers, respectively. Analysis of crystal structure similarity (CSS) was undertaken for the pair PANLAS/(I), using the graphical user interface of the Mercury program (CSD version 2020.2.0, Materials module). With this methodology, based on the COMPACK algorithm [46], from each CIF file a cluster of molecules (default size equal to 15) was obtained by picking a molecule plus a number of its closest neighboring molecules, and analyzing the overlapping molecules in the two clusters was then attempted. Two crystal structures are isostructural to within a specified tolerance (the default distance and angle tolerances being 20% and 20 • , respectively) if, by comparing a cluster of 15 molecules, the algorithm returns 15 out of 15 molecules in common within the tolerance parameters set. The root mean square deviations in atomic positions (RMSD) were then calculated for the overlapping molecules in the two clusters. Analysis of the PXRD pattern was also performed using the already-cited CSD Materials module. With this method, the comparison of simulated PXRD patterns from each CIF file returned a similarity index (PXS) which ranged from 0 (completely dissimilar patterns) to 1 (identical patterns). These similarity methods were applied to the pair PANLAS/(I) for the comparison of a cluster extended to 20 molecules, as more than one molecule was present in the asymmetric unit [47]. The calculations of CSS, PXS and RMSD showed that the central molecule plus 19 symmetry-related molecules were in a common packing shell (Table 2). 551.52 (15) Comparison of the simulated PXRD patterns shows some differences (Figure 3), as PXRD is sensitive to chemical composition and even to small variations in cell metrics and atomic displacement parameters originated by data collection performed under different temperatures (PANLAS, 150 K; (I), 298 K) [47], but results returned by similarity methods suggests that these co-crystals have the same crystal packing. The isomorphism of the crystal structures was determined by using the unit-cell parameters of the two crystal structures to calculate the unit-cell similarity index (Π) [48]. If Π = 0, then the matched crystal structures are isomorphous. In the present case, the Π index was close to zero (zero up to the first decimal place), suggesting the isomorphism of the pair PANLAS/(I) ( Table 2).
Taken together, all these results indicated (1:1) 5-fluorocytosine/isocytosine monohydrate to be isomorphic and isostructural with (1:1) 5-fluorocytosine/5-fluoroisocytosine monohydrate molecular (Figure 4), which differed solely in the H/F substitution at the C5 position of isocytosine. Neither of the two atoms was involved in the hydrogen bonding as donor or acceptor functional groups, although the F· · · F close contact [2.9003 (15) Å] observed in PANLAS between adjacent TPI dimers within a ribbon was replaced in (I) by the F· · · H contact (3.175 Å).
Crystal packing is mainly controlled by interactions between nucleobases, and the hydrogen-bonding scheme (Table 3) shows that TPI pairs and water molecules have the dual role of hydrogen bond donors and acceptors. Specifically, ribbons running approximately along the c axis were formed by two different R 2 4 (8) hydrogen bond rings which involved strong N-H· · · O (carbonyl) hydrogen bonds of adjacent antiparallel TPI heterodimers (Figure 4a). Nonetheless, water molecules play a fundamental role in the hydrogen bonding interactions, which include four intermolecular contacts. Acting as both hydrogen bond donors and acceptors, water molecules join together two TPI heterodimers by forming R 3 3 (10) hydrogen bond rings within a ribbon. Moreover, water molecules interact as hydrogen bond acceptors with a third TPI heterodimer from an adjacent ribbon through weak C-H· · · O intermolecular interactions, favoring the formation of a two-dimensional network. Neighboring networks are held together by water-mediated Ow-O hydrogen bonds (Figure 4b).

Conclusions
In summary, in this work the crystal structure of (1:1) 5-fluorocytosine/isocytosine monohydrate molecular complex has been analyzed in terms of spatial disposition and compared with that of the structurally-related fluorocytosine/5-fluoroisocytosine monohydrate molecular complex. From the evaluation of results obtained with the similarity methods, it has been shown that this molecular complex exhibits isomorphism and isostructurality with (1:1) 5-fluorocytosine/5-fluoroisocytosine monohydrate molecular complex. The crystal structure was sustained by the base-pairing interaction between the 1H-ketoamino form of 5-fluorocytosine and the 3H-ketoamino form of isocytosine, demonstrating that complementary binding enables the crystallization of specific tautomers. The robustness of the supramolecular architectures based on three-point recognition synthon, which benefits from the presence of isocytosine in the 3H-ketoamino form, was responsible of the existence of isostructurality between (I) and PANLAS. Indeed, particularly in the presence of strong interactions such as TPI, no change in the supramolecular arrangement due to the substitution of the H atom with the fluorine atom is expected, if neither of the two atoms are involved in hydrogen bonding as donor or acceptor functional groups [50]. The solidity of triple hydrogen-bonding interaction has been invoked to tolerate the H/F exchange in isostructural co-crystals of (1:1) barbiturate melamine complexes [51].
The understanding of changes in noncovalent interactions induced by chemical modifications of nucleobases is essential for investigating molecular recognition processes involved in DNA/RNA base pairing. Therefore, crystal structure determination of molecular complexes of nucleobases with their fluorinated derivatives, where covalently bound fluorine atoms cannot compete with stronger heteroatoms as acceptors in the formation of hydrogen bonds [52,53], can provide important experimental insights into the nature of DNA/RNA pairing interactions between natural bases and their close unnatural analogs [54].