2023 Volume 71 Issue 8 Pages 633-640
Nobiletin (NOB) is a flavonoid with attractive pharmaceutical characteristics, including anti-Alzheimer’s, anti-inflammation, and anti-cancer properties, but it has low solubility in water, resulting in reduced bioavailability. Its solubility must be improved to develop NOB as a drug. Cocrystal engineering can change the physicochemical properties of an active pharmaceutical ingredient and generate remarkable drug candidates that are superior in drug formulation. In this report, extensive co-crystal screening of NOBs with 31 cocrystal formers (coformers) with various functional groups was carried out by the liquid-assisted grinding method. As a result, four cocrystals (NOB with urea (URE), oxalic acid, gallic acid and salicylic acid) and one solvate crystal (NOB with formic acid (FOR)) were found. Powder X-ray diffraction and thermal analysis revealed the unique crystal morphology of all the obtained samples. In addition, the crystal structures of two of them (NOB-URE and NOB-FOR) were determined by single crystal X-ray diffraction. The results revealed that NOB-URE and NOB-FOR are new cocrystals or solvate crystals consisting of molar ratios of 1 : 2 and 1 : 0.73, respectively. In NOB-URE, we could observe a transient increase in solubility due to supersaturation, suggesting that URE is one of the better coformers of NOB.
Nobiletin (NOB; C21H22O8; Fig. 1a) is a flavonoid extracted from citrus peels.1) NOB and its metabolites have been reported to have a wide variety of medicinal properties such as anti-Alzheimer’s, anti-inflammation, anti-cancer, anti-oxidation, anti-apoptosis, and neuroprotection.2–10) Therefore, NOB is expected to be an active pharmaceutical ingredient (API) for these diseases. A recent study demonstrated that the use of choline and geranic acid was effective for improving the transdermal administration of NOB.11) However, NOB is poorly soluble in water. Improving its solubility is necessary to put NOB into practical use as an oral administrative formulation.9)
(a) Chemical structure of NOB. The atom numbers of carbon and oxygen atoms were shown in black and red, respectively. (b) Chemical structures of coformer used in this study. The atom numbers of nitrogen and hydrogen atoms in URE and FOR were shown in blue and black, respectively.
Recently, protein crystal structures have become readily available as a drug target and can be applied to structure-based drug design. We then have been able to design compounds with high selectivity for the target protein and design molecules that increase pharmacological activity. Although these compounds have a high drug activity, they have a complicated chemical structure, high molecular weight, and high hydrophobicity. They are poorly soluble in water and cause a decrease in bioavailability. Actually, about 40% of drugs launched in recent years and 70–90% of drugs in the discovery pipeline are classified as class II or IV drugs in the Biopharmaceutics Classification System.12) To overcome these problems, the solubility of the compounds was improved using several methods. The primary method for improving the solubility is salt formation, which has many examples in medicine because of its convenience and low developmental cost.13) However, this method can only be applied to ionizable compounds because the salt is formed by the ionic bonds between the drugs and additives so that the dissociating groups are necessary for drugs.14) Another method is cocrystallization, which can be applied to ionizable and non-ionizable compounds. The cocrystal is formed via noncovalent interactions such as hydrogen bonding or hydrophobic stacking interactions.15–21) Consequently, cocrystallization has been extensively studied in recent years as a method to improve physical properties without salt formation because the dissociating groups of drugs are unnecessary.22) Moreover, there have been reports on improvements in the solubility, hygroscopicity and physical stability of the compounds.23–27) Examples of drugs using the co-crystallization method include the sodium-glucose transporter 2 (SGLT2) inhibitors for diabetes, Suglat® (Ipragliflozin/L-proline cocrystal) and Steglatro® (Ertugliflozin/L-polyglutamic acid cocrystal), and the neprilysin inhibitor for cardiac insufficiency, Entresto® (Sacubitril/Valsartan cocrystal).
The neat grinding (milling), the solvent-assisted grinding (solvent drop grinding), the slurry, the eutectic, and the nano-spot methods have been reported as cocrystal screening methods.28–32) In addition, it has been suggested that non-toxic and commercially available compounds such as carboxylic acids, amino acids, and sugars should be used as coformers for cocrystallization.33) The neat grinding method has been shown to avoid the use of excessive solvents and tends to be more efficient in cocrystal formation compared to solution-based approaches.28) The solvent-assisted grinding method was introduced as a means of increasing the rate of cocrystal formation in the solid state using a few drops of solvent. It has proven to have advantages such as allowing solvents to be switched easily and is suitable for a significantly wider range of products, including polymorphs and solvates of the cocrystals.34–36) Cocrystallization by the slurry technique is a method in which the API and conformer are added together in a solvent at concentration above their respective solubilities to promote crystallization of the cocrystal by Ostwald’s step rule in suspension. It demonstrates the importance of geometrical fit as well as the hydrogen bonding in cocrystal formation.30) Furthermore, the nano-spot technique has been developed as a method to determine whether microcrystals are cocrystals or not in combination with Raman spectroscopy.32) This method is reported to be useful because the amount of API used can be significantly reduced, which helps to reduce costs. However, the mechanism of cocrystal formation is not yet understood. It is also difficult to predict the physicochemical properties of the cocrystals from the chemical structure and the solubility of the APIs and coformers used. Moreover, computational methods are also not well established, so the studies through the experimental cocrystal screening is necessary, although it is time-consuming.
In this report, we tried to obtain cocrystals of NOB to improve its solubility. For this purpose, we first carried out a cocrystal screening by the solvent-assisted grinding method, where the physicochemical properties of the samples after preparation can be immediately characterized by powder X-ray diffraction and thermal analysis. As a result, we obtained four cocrystals and one solvate crystal of NOB. We also demonstrated for the first time that the solubility of NOB is improved by cocrystallization albeit temporarily. The single crystal structures were also determined for two of them. Furthermore, the solubility of each cocrystal or solvate crystal in water was investigated.
NOB was provided by Ushio ChemiX Corporation (Shizuoka, Japan). The sample was used without further purification. A total of 31 coformers, consisting of acids and amides listed in Supplementary Fig. 1, were used for the screening. These coformers were purchased from Nacalai Tesque, Inc. (Kyoto, Japan), Kanto Chemical Co., Inc. (Tokyo, Japan), Tokyo Chemical Industry Co., Ltd. (Tokyo, Japan), or FUJIFILM Wako Pure Chemical Corporation (Osaka, Japan), respectively. All other reagents used were of the highest grade available from commercial sources.
Cocrystal Screening and Powder X-Ray Diffraction (PXRD)The cocrystal screening of NOB with respective coformers was conducted using the solvent-assisted grinding method.29) Initial cocrystal screening was performed by mixing NOB (50 mg, 0.12 mmol) with equimolar amounts of each coformer, and the formation of cocrystals was confirmed by PXRD performed under the following conditions. Then, we optimized the molar ratio of the co-formers to NOB, which gave new patterns. After optimization, NOB (50 mg, 0.12 mmol) was mixed individually with urea(URE) (14.9 mg, 0.24 mmol), formic acid (FOR) (100 µL, 2.36 mmol), gallic acid (GAL) monohydrate (21.1 mg, 0.12 mmol), oxalic acid (OXA) dihydrate (7.8 mg, 0.06 mmol), and salicylic acid (SAL) (17.2 mg, 0.12 mmol) as a co-former, respectively (Fig. 1b). While grinding, 100 µL of ethanol (EtOH) was added dropwise as an assistant solvent except for the case of liquid FOR. The mixture was ground three times for approximately 20 min with an agate mortar and a pestle at ambient environmental conditions. The solid products after grinding were examined by PXRD.
PXRD patterns in the 2θ range of 5 to 30° of the solid products and starting materials were collected using a Mini Flex 600 (Rigaku Corporation, Tokyo, Japan) with Cu-Kα radiation at room temperature.
Thermal AnalysisThermal gravity (TG)/differential thermal analysis (DTA) was performed using a TG8120 instrument (Rigaku Corporation, Tokyo, Japan). Samples of approximately 3 mg were placed in open aluminum pans and heated under a dry nitrogen gas purge (200 mL/min) with a heating rate of 10 °C/min. The calibration of the TG/DTA instrument was checked using calcium oxalate monohydrate.
Simultaneous PXRD and Differential Scanning Calorimetry (DSC)Simultaneous PXRD and DSC were conducted using the RINT2100 and Thermo Plus XRD/DSC II (Rigaku Corporation, Tokyo, Japan) differential scanning calorimetry instrument at a heating rate of 10 °C/min. PXRD patterns in the 2θ range of 5 to 25° of the products were collected.
Crystallization and Single Crystal Structure Determination of the NOB CocrystalsCrystallization of single crystals of NOB-URE and NOB-FOR was performed using the solvent evaporation technique. A mixture of 50 mg of NOB with 77.3 mg URE was dissolved in 0.4 mL of EtOH. The single crystals of NOB-URE were obtained after 1 d of evaporation at room temperature. To obtain NOB-FOR solvate crystals, an excess amount of NOB was dissolved in 0.5 mL of FOR. The undissolved NOB powder was filtered off. The individual crystals of NOB-FOR solvate were obtained after 1 d of evaporation at room temperature.
Diffraction datasets were collected on a Rigaku XtaLAB P200 system diffractometer (Rigaku Corporation, Tokyo, Japan) with Cu-Kα or Mo-Kα radiation for NOB-URE or NOB-FOR, respectively. Data collection, indexing, integration, and scaling were carried out using Rigaku CrystalClear software. Crystal structure determination and refinement were performed using SHELXT, SHELEXL and ShelXle.37–39) The data collection and structure refinement results are summarized in Table 1. The atomic coordinates and diffraction data have been deposited in the Cambridge Structural Database (CCDC number 1992852 and 1992853).
| NOB-URE | NOB-FOR | |
|---|---|---|
| Crystal data | ||
| Chemical formula | C21H22O8·2CH4N2O | C21H22O8·0.73CH2O2 |
| Space group | P1̅ | P1̅ |
| Cell parameters (Å, deg.) | ||
| a | 7.1787 (5) | 11.7904 (10) |
| b | 11.7743 (6) | 14.3344 (4) |
| c | 15.5162 (6) | 15.0681 (11) |
| α | 75.146 (4) | 64.138 (8) |
| β | 88.437 (4) | 89.2120 (10) |
| γ | 89.004 (5) | 65.8670 (8) |
| Cell volume (Å3) | 1267.12 | 2048.86 |
| Z/Z′ | 2/1 | 4/2 |
| R-Factor | 0.0773 | 0.0506 |
| Data collection | ||
| Temperature (K) | 293 | 93 |
| Wavelength (Å) | 1.5418 | 0.7108 |
| (sinθ/λ)max (Å−1) | 0.622 | 0.649 |
| No. of reflections | ||
| Measured | 14344 | 33356 |
| Independent | 4828 | 9264 |
| Observed [I > 2σ(I)] | 4460 | 7234 |
| Rint | 0.0486 | 0.0343 |
| Refinement | ||
| No. of reflections | 4828 | 9264 |
| R [F2 > 2σ(F2)] | 0.0773 | 0.0506 |
| wR(F2) | 0.2376 | 0.1742 |
| Δρmax, Δρmin (e Å−3) | 0.626, −0.433 | 0.611, −0.415 |
To confirm whether co-crystallization could improve the NOB solubility, an excess amount of NOB alone (approximately 3 mg), or each cocrystal or solvate crystal (approximately 3 mg converted to the weight of NOB) was placed in 1 mL of distilled water and stirred at 30 rpm using a small-size culture rotator (Rotator RT-50, TAITEC Co., Ltd., Tokyo, Japan) for 120 min at 37 °C. During incubation, samples were taken at each time point as follows; 0.5, 1, 5, 10, 30, 60, and 120 min. The aliquots from each sample were filtered through a 0.45 µm syringe filter (Toyo Roshi Kaisha, Ltd., Tokyo, Japan) that was prewarmed at 37 °C. The filtered aliquot was diluted with an equal volume of EtOH and assayed by UV and visible spectrophotometric analysis using Spectra max 190 (Molecular Devices Corporation, Sunnyvale, CA, U.S.A.) at 340 nm, which is an absorption maximum of NOB and where each conformer has no absorption. Dissolution curves of NOB alone, and cocrystals or solvate crystals were made by plotting the incubation time versus the molar concentration of solubilized NOB. For comparison, same experiments were carried out with the physical mixtures (PM) corresponding to each cocrystal or solvate crystal.
A total of 31 hydrogen-donating compounds with carboxylic acids or amides as functional groups that can form hydrogen bonds with NOB were used as a coformer for cocrystal screening. Cocrystal screening of NOB with co-formers using the solvent-drop grinding method provided new crystalline solids with PXRD patterns that differed from those of the starting materials, as shown in Fig. 2. As described below, the crystal structures of the NOB cocrystals with URE or solvate crystals with FOR were determined by single-crystal X-ray diffraction.
DTA analysis of NOB-URE showed that the endothermic peak corresponding to the melting of the cocrystals was observed at 140 °C, which was higher than those of the starting materials, NOB (136 °C) and URE (133 °C) (Fig. 3a). In the TG and PXRD-DSC analysis, no weight loss and change of PXRD patterns of the cocrystals were observed in the heating process below the melting point. Thus, we presumed that the cocrystals were non-solvates (Figs. 3a, 4a). In contrast, DTA analysis of NOB-FOR showed the two endothermic peaks at 50 and 134 °C (Fig. 3b). The first peak was unique to the NOB-FOR and the second was nearly identical to the melting point of NOB alone. In the TG thermogram, the decreased mass corresponded to 0.76 molecules of formic acid. The PXRD-DSC analysis also showed the two endothermic peaks at 53 and 137 °C, which corresponded to the values observed in the DTA analyses (Fig. 4b). During the heating process, the PXRD patterns of the sample were unique to the starting materials before the first endothermic peak. However, the pattern changed when it was heated from 53 to 137 °C, and was identical to that of the NOB crystals. Therefore, it can be concluded that the NOB-FOR was solvated with the partial-occupancy formic acid, and the temperature to remove the formic acid was approximately 50 °C. In addition to the new PXRD patterns obtained, NOB-OXA, NOB-GAL, and NOB-SAL cocrystals also had a unique melting points: 143, 213, and 111 °C, respectively, which differed from the melting points or the dehydration temperature of each compound, OXA dihydrate (59 and 161 °C), GAL monohydrate (70 and 255 °C) and SAL (158 °C), respectively (Figs. 3c–e). Moreover, no weight loss of these cocrystals was observed in the heating process before the melting points, indicating that the cocrystals were non-solvates.
(a) NOB-URE, (b) NOB-FOR, (c) NOB-OXA, (d) NOB-GAL, (e) NOB-SAL. The TG and DTA curves are shown in dotted line and solid line, respectively.
Single-crystal X-ray diffraction of the NOB-URE cocrystal revealed that the asymmetric unit of space group P1̅ consisted of one NOB molecule and two URE molecules (URE1 and URE2) (Fig. 5a). The hydrogen atoms, H1A and H2A, of each amide group from the URE1 form hydrogen bonds with the oxygen atoms, O8 and O7, of two methoxy groups of NOB with distances of 2.07 and 2.49 Å respectively (Table 2). Conversely, the hydrogen atoms H1A and H2A, of each amide group from the URE2 form hydrogen bonds with the oxygen atoms O5 from methoxy groups and O4 from the carbonyl group of NOB with distances of 2.11 and 2.19 Å, respectively (Table 2). In the crystal, the O6 methoxy group of the NOB formed a CH…O hydrogen bond with the O6 methoxy group of the horizontally adjacent NOB to the flavone skeleton (Fig. 5b). Moreover, in the vertical direction of the flavone skeleton, the NOB molecules were stacked due to the π–π stacking interaction (Fig. 5c). The channel-like structures existed between the stacked NOB molecules, and URE molecules were located within the channels and formed the crystals by linking the stacked NOB molecules (Fig. 5d).
(a) Crystal structure of NOB-URE cocrystal in the asymmetric unit. The hydrogen bonds between NOB and URE are shown in the black dotted line with the distance. (b and c) Crystal packing analysis of NOB-URE cocrystal. The right panels show magnified views of the frame in the left figure. The hydrogen bonds between two NOB molecules are shown in the blue dotted line. In (c), the π–π stacking interactions are indicated by black arrow at both ends. (d) Layered structure and channel-like spaces in the NOB-URE cocrystal. (e) Crystal structure of NOB-FOR solvate crystals in the asymmetric unit. The hydrogen bonds between NOB and FOR are shown in the black dotted line with the distance. (f and g) Crystal packing analysis of NOB-FOR solvate crystals. The right panels show magnified views of the frame in the left figure. The hydrogen bonds between two NOB molecules are shown in the blue dotted line. In (g), the π–π stacking interactions are indicated by black arrow at both ends. (h) Layered structure and channel-like spaces in the NOB-FOR solvate crystals.
| Donor (D) | D–H…A | Acceptor (A) | Distance (Å) | Angle (°) | ||
|---|---|---|---|---|---|---|
| D–H | H…A | D…A | D–H…A | |||
| NOB-URE | ||||||
| URE1 | N1–H1A…O8 | NOB1 | 0.860 | 2.069 | 2.883 | 157.57 |
| URE1 | N2–H2A…O7 | NOB1 | 0.860 | 2.483 | 3.339 | 173.70 |
| URE2 | N1–H1A…O5 | NOB1 | 0.860 | 2.112 | 2.928 | 161.64 |
| URE2 | N2–H2A…O4 | NOB1 | 0.860 | 2.188 | 3.016 | 158.35 |
| NOB-FOR | ||||||
| FOR1 | O2–H2…O4 | NOB1 | 0.840 | 1.814 | 2.652 | 174.47 |
| FOR2 | O2–H2…O4 | NOB2 | 0.840 | 2.141 | 2.804 | 135.66 |
| FOR3 | O2–H2…O4 | NOB2 | 0.841 | 1.896 | 2.653 | 149.07 |
| FOR4 | O2–H2…O4 | NOB1 | 0.840 | 2.532 | 2.785 | 98.64 |
The NOB-FOR solvate crystal also belonged to the space group P1̅ (Fig. 5e). The solvate crystal contained two NOB molecules and several FOR molecules in the asymmetric unit. The electron densities corresponding to the FOR molecules were unclear. Hence, we inserted four molecules of FOR (FOR1–4) under the occupancy settings of 0.45 (FOR1), 0.39 (FOR2), 0.34 (FOR3), and 0.29 (FOR4), respectively. We then finalized the structure refinement. In the final structure, the hydrogen and oxygen atoms connected to the carbonyl carbon atom of FOR2 were not visible in the electron density map, and hence they are thought to be disordered in the crystal. As a result, the asymmetric unit in the crystals contained two NOB molecules and 1.46 FOR molecules, which were almost consistent with the result of the thermal analysis with a value of 0.76 molecules of FOR per one NOB molecule. The H2 atoms of FOR1 and FOR4 form a hydrogen bond with O4 from the carbonyl group of NOB1 with distances 1.81 and 2.53 Å, respectively (Table 2). The H2 atoms of FOR2 and FOR3 form hydrogen bonds with the O4 of NOB2 with distances 2.14 and 1.90 Å, respectively (Table 2). In the crystal, the O3′ methoxy group of the NOB1 formed a CH…O hydrogen bond with the O6 methoxy group of the horizontally adjacent NOB2 to the flavone skeleton (Fig. 5f). The O3′ and O8 methoxy group of the NOB2 also formed a CH…O hydrogen bond with the O7 and O8 methoxy groups of the horizontally adjacent NOB1. Moreover, in the vertical direction of the flavone skeleton, the NOB molecules were stacked due to the π–π stacking interaction and the channel-like structures were also observed as they were in the NOB-URE cocrystals (Figs. 5g, h). The FOR molecules were also located within the channels.
Dissolution Study of the NOB CocrystalDrug absorption depends on solubility and rate of dissolution. These properties are controlled by the dissolution of the cocrystal and the supersaturation of the API with regard to the dissociation and dissolution of the cocrystal.40) Thus, we investigated the dissolution behavior of NOB alone, and the four cocrystals and one solvate crystal against distilled water at 37 °C (Fig. 6a). The concentration in water of NOB was determined to be 26.6 µg/mL (66 µM) with a pH value of 4.33 after 120 min of incubation. The chemical structure of NOB is highly hydrophobic and there are no void spaces in its crystal structure, so the low solubility of NOB is probably low (Supplementary Fig. S2).
The data are expressed as the mean ± standard deviation (n = 3).
The dissolution behavior of NOB was slightly improved by cocrystallization with coformers (URE, FOR, OXA, GAL, and SAL). URE was the most effective coformer, improving the concentration of NOB in water 9.1-fold (242.3 µg/mL (602 µM)) after incubation at 37 °C for 2 min (Fig. 6b). However, the concentration in water of NOB gradually decreased, and the equilibrium concentration (incubation at 37 °C for 120 min) stabilized at 35.9 µg/mL (89 µM, 1.3-fold) with a final pH value of 7.28. After 120 min incubation, the precipitates appeared. The PXRD pattern of the precipitate collected by filtration was identical to that of NOB alone (Fig. 6c), indicating that the cocrystal had dissolved and dissociated into the NOB and the coformer during incubation. Subsequently, the NOB crystallized in solitary form. Unlike other coformers, URE is a neutral compound,41) and the NOB-URE cocrystals have a channel-like structure, which is easily penetrated by the solvent. It is believed that the NOB-URE cocrystals continued to dissolve because the solution remained neutral pH even after the cocrystals dissolved in the initial stages of this study. This may have resulted in a transient increase in solubility. The dissolution study of PM was also performed to confirm the effect of URE on the solubility of NOB and the effect of cocrystallization. The PM study showed no solubility improvement, the concentration of NOB was 33.2 µg/mL (83 µM, 1.2-fold) with a final pH value of 6.88 (Supplementary Fig. S3a and Supplementary Table 1). This result indicates that cocrystallization with URE is important for the transient improvement in solubility of NOB, while it is less effective as a hydrotrope for URE.
We found that the NOB-FOR solvate crystals have a maximum solubility of 35.7 µg/mL (89 µM, 1.3-fold) in 30 min. Their equilibrium concentration stabilized at 32.8 µg/mL (81 µM, 1.2-fold) with a final pH value of 3.06. The PXRD pattern of the precipitate collected by filtration was identical to that of initial solvate crystals, indicating that the solvate crystals were hardly dissolved (Fig. 7b). Although NOB-FOR solvate crystals had a channel-like structure like the NOB-URE cocrystal, the solubility of NOB improved only a little probably because the FOR effect made the pH acidic. Because of their channel-like structure, the solvate crystals dissolved very slightly in the initial stage of this study. However, they made the solution acidic, and as a result, the solvate crystals became almost insoluble. Correspondingly, the equilibrium concentration of the PM stabilized at 32.6 µg/mL (81 µM, 1.2-fold) which was almost identical to that of the solvate crystals with a final pH value of 2.03 (Supplementary Fig. S3b and Supplementary Table 1). FOR would not be effective as a hydrotrope.
(a) NOB-URE, (b) NOB-FOR, (c) NOB-OXA, (d) NOB-GAL, (e) NOB-SAL. Black circles and blue stars indicate the diffraction peaks coinciding with NOB alone or cocrystal peaks, respectively.
The NOB-OXA cocrystals had a maximum solubility of 90.7 µg/mL (226 µM, 3.4-fold) in 30 s and the equilibrium concentration stabilized at 37.2 µg/mL (93 µM, 1.4-fold) with a final pH value of 2.19. The PXRD pattern of the precipitate collected by filtration was identical to that of NOB alone (Fig. 7c), indicating that the cocrystal had dissolved and dissociated into the NOB and the coformer during incubation as in the case of NOB-URE cocrystals. Unlike FOR, the solubility of NOB was transiently improved despite acidic conditions, suggesting that NOB-OXA possibly may also have a structure that is easily penetrable by solvents. In addition, it is believed that the NOB-OXA cocrystals continued to dissolve even when the solution became acidic after the cocrystal dissolved in the early stage of this study. However, the equilibrium concentration of the PM stabilized at 33.8 µg/mL (84 µM, 1.3-fold) which was almost identical to that of the cocrystals with a final pH value of 2.48 (Supplementary Fig. S3c and Supplementary Table 1), indicating that OXA is also not effective as a hydrotrope.
The NOB-GAL cocrystals had a maximum solubility of 55.9 µg/mL (139 µM, 2.1-fold) in 5 min. The solubility of NOB was transiently improved in NOB-GAL cocrystals, although not as much as in the NOB-URE and NOB-OXA cocrystals. Cocrystallization seemed to slightly improve the solubility of NOB, and GAL as a hydrotrope also improved the solubility as discussed below. The equilibrium concentration stabilized at 35.5 µg/mL (88 µM, 1.3-fold) with a final pH value of 3.79. The PXRD pattern of the precipitate collected by filtration showed mostly initial cocrystals, but some diffraction peaks derived from solitary NOB crystals were also observed (Fig. 7d). These results suggest that the NOB-GAL cocrystals may have a higher dissolution rate but their solubility is low. A short-term supersaturation is inferred by the dissolution of a small fraction of the cocrystals, which precipitated as solely as NOB crystals. However, because the solution became acidic due to the dissolution of the small fraction in the early stage of this study, and the remaining cocrystals did not dissolve, resulting in most of them remaining intact. The equilibrium concentration of the PM stabilized at 58.9 µg/mL (147 µM, 2.2-fold) which was higher than that of the cocrystals with a final pH value of 2.29 (Supplementary Fig. S3d and Supplementary Table 1). This result suggests that GAL, unlike URE, FOR, and OXA, is an aromatic compound and that the aromaticity may slightly improve the solubility of NOB. Indeed, it has been reported that the solubility of poorly soluble APIs is significantly improved when the aromatic compounds were used as a hydrotrope.42,43) In particular, nicotinamide and its derivatives have been found to have potential as a hydrotrope for poorly soluble APIs.44,45) In general, it has been suggested that when nicotinamide is used as a hydrortrope for API, nicotinamide and API form a complex through π–π interaction using their respective π-electrons. Therefore, it is possible that GAL may increase the solubility of NOB through the π–π interaction.
The NOB-SAL cocrystals had a maximum solubility of 40.7 µg/mL (101 µM, 1.5-fold) in 10 min and the equilibrium concentration stabilized at 37.3 µg/mL (93 µM, 1.4-fold) with a final pH value of 2.64. The PXRD pattern of the precipitate collected by filtration showed mostly solitary NOB crystals, but some diffraction peaks derived from NOB-SAL cocrystals were also observed (Fig. 7e). These results suggest that the NOB-SAL cocrystals may have a lower dissolution rate than NOB-GAL. As a result, no supersaturation was observed in the NOB-SAL cocrystals, and they required a longer time to give maximum solubility. As with NOB-OXA, the NOB-SAL cocrystals continued to dissolve even when the solution became acidic after the cocrystal dissolved in the early stage of this study, but it is assumed that during the dissolution of the NOB-SAL cocrystals, the NOB crystals were solely precipitated coupled with the acidic conditions. The equilibrium concentration of the PM stabilized at 55.0 µg/mL (137 µM, 2.1-fold) which was higher than that of the cocrystals with a final pH value of 2.48 (Supplementary Fig. S3e and Supplementary Table 1). As with GAL, the aromaticity of SAL may have slightly improved the solubility of NOB through the π–π interactions.
Cocrystal screening of NOB was carried out using the solvent drop grinding method, resulting in the detection of four cocrystals and one solvate crystal. The formation of cocrystals was confirmed by PXRD and thermal analysis. Furthermore, the crystal structures of NOB-URE cocrystals and NOB-FOR solvate crystals were determined by the single crystal structure analysis and their stoichiometric ratios were confirmed to be 1 : 2 and 1 : 0.73, respectively. The cocrystals and the solvate crystals slightly improved the solubility of NOB. Among these, the NOB-URE crystals showed the highest improvement in solubility due to the supersaturation in the early stage of dissolution. This may be because only URE was a neutral compound among the cocrystal coformers obtained, although the reason is unclear. The improvement in the solubility of NOB was limited, at least when the acidic compounds were used as a coformer of cocrystals. In addition, NOB-URE cocrystals had a channel-like structure, which allows solvents to penetrate easily into the cocrystals. These results suggest that the NOB-URE can be effectively utilized in combination with hydrotropes or detergents to maintain higher NOB solubility. They also suggest that the neutral or basic compounds, at least not acidic compounds, with aromaticity as hydrotropes or detergents may further improve the solubility of NOB.
The authors would like to thank Dr. Kazuo Okamoto of Ushio ChemiX Corporation for kindly providing the synthetic nobiletin. The authors thank Mr. Josef Messerklinger for editing a draft of this manuscript. This work was partly supported by the Japan Society for the Promotion of Science KAKENHI Grant Number 20K07001 (to SN) and 18K14641 (to HS), and 20K16058 (to MI), and Grants from the Takeda Science Foundation, from the Japan Prize Foundation, and from the Uehara Memorial Foundation (to HS).
The authors declare no conflict of interest.
This article contains supplementary materials.
