Engineering Lipophilic Aggregation of Adapalene and Adamantane-Based Cocrystals via van der Waals Forces and Hydrogen Bonding

Lipophilic aggregation using adamantanes is a widely exploited molecular property in medicinal and materials chemistry. Adamantanes are traditionally installed to molecular units via covalent bonds. However, the noncovalent installation of adamantanes has been relatively underexplored and presents the potential to bring properties associated with adamantanes to molecules without affecting their intrinsic properties (e.g., pharmacophores). Here, we systematically study a series of adamantanecarboxylic acids with varying substitution levels of methyl groups and their cocrystals with bipyridines. Specifically, single-crystal X-ray diffraction shows that while the directionality of single-component adamantanes is notably sensitive to changes in methyl substitution, hydrogen-bonded cocrystals with bipyridines show consistent and robust packing due to π-stacking predominance. Our observations are supported by Hirshfeld surface and energy framework analyses. The applicability of cocrystal formation of adamantanes bearing carboxylic acids was used to generate the first cocrystals of adapalene, an adamantane-bearing retinoid used for treating acne vulgaris. We envisage our study to inspire noncovalent (i.e., cocrystal) installation of adamantanes to generate lipophilic aggregation in multicomponent systems.

The solution was left for one week of slow crystallization, and colorless plate-like crystals suitable for the synchrotron were collected.

Instruments and methods:
Single crystal X-ray diffraction (SCXRD) data was collected on a Rigaku XtaLAB Mini II diffractometer with a CCD area detector (λMoKα = 0.71073 Å, monochromator: graphite) equipped with an Oxford Cryostream low-temperature device.Experiments were conducted at 100 K with a range of 2θ = 3-62°.The collected data was refined with CrysAlisPro through standard data reduction and background corrections (multiscan for ada, dimet-ada, trimet-ada, 2(ada)•(azop), 2(dimet-ada)•(azop), 2(trimet-ada)•(azop)).Crystals were mounted in Paratone oil on a Mitegen magnetic mount.Structure solution and refinement were performed using SHELXT 1 and SHELXL, 2 respectively, within the Olex2 3 graphical user interface.Non-hydrogen atoms were refined anisotropically, and hydrogen atoms were placed in geometrically calculated positions using a riding model.Synchrotron radiation was used to collect 2(adp)•(azop) and 2(adp)•(bpeta); The intensity data were collected with APEX3, 4 integration and corrections were applied with SAINT v8.40a, 5 absorption and other corrections were made using SADABS 2016/2 6 for 2(adp)•(azop) and TWINABS 2012/1 for 2(adp)•(bpeta).Dispersion corrections appropriate for this wavelength were calculated using the Brennan method in XDISP 7 within refined using SHELXL 2019/2.All non-hydrogen atoms were refined anisotropically.Hydrogen atoms were placed geometrically on the carbon atoms, refined with a riding model.On the H-O groups they were found in the difference map and allowed to refine freely.Crystal structures were generated using Mercury.
Additional refinement notes on 2(adp)•(azop): Displacement parameter restraints were used to model C11' more reasonably.All non-hydrogen atoms were refined anisotropically.Hydrogen atoms were placed geometrically on the carbon atoms, refined with a riding model.On the H-O groups they were found in the difference map and allowed to refine freely.Displacement parameter restraints were used to model C11' more reasonably.PLATON and CHECKCIF found 93% fit to Pbca.The refinement was tried in Pbca but the R1 stuck around 15% cf 7.1% in Pca21 and the refinement output did not flag any correlation in the refinement.Therefore, the structure was refined in Pca21.
Additional refinement notes on 2(adp)•(bpeta): Several crystals were tried, and this was the best crystal.The diffraction pattern showed twinning.Using Cell_now two orientation matrices were determined, the relationship between these components was determined to be 180 degrees about real axis 0 0 1.The data were integrated using the two matrices in SAINT, TWINABS was used to produce a merged HKLF4 file, for structure solution and initial refinement, and HKLF5 file for final structure refinement.The HKLF5 file contained the merged reflections first component and those that overlapped with this component, which were split into 2 reflections.
TWINABS indicated the twin faction to be 64:36 The structure was solved using the HKLF4 file, but the best refinement was given by the HKLF5 file.All non-hydrogen atoms were refined anisotropically.Hydrogen atoms were placed geometrically on the carbon atoms, refined with a riding model.On the H-O groups they were found in the difference map and allowed to refine freely.
Powder X-ray diffraction (PXRD) data was collected on a Scintag XDS-2000 diffractometer using CuKα1 radiation (λ = 1.5418Å).The samples were mounted and collected on glass slides typically in the range of 5−40° two-theta (scan type: step size: 0.02°, rate: 3 deg/min, continuous scan mode).The equipment was operated at 40 kV and 30 mA, and data was collected at room temperature.Fourier-transform infrared spectroscopy (FT-IR) spectra were captured using a ThermoFischer Scientific iS5 IR spectrometer from 600 to 4000 cm-1 using a diamond attenuated total reflectance (ATR) accessory.

S12
Table S8.Selected hydrogen bonds and interactions in single component crystals.

Figure S10 .
Figure S10.Selected projection interaction percentages of the reported single-component structures.

Figure S11 .
Figure S11.Selected projection interaction percentages of the reported cocrystal structures.