Cyclodextrins (CDs) contain six or more α-D-glucopyranoside units, and have lipophilic interior cavities and hydrophilic exterior surfaces [1]. Due to this unique physical characteristic, the solubility of a large amount of indissoluble drug was increased. This amazing change was usually considered as the including effect of CDs. In recent years, as more as analyzing to CDs, variety CD/drug structure was put forward, such like self-association and non-inclusion [2]. Hydroxypropyl methylcellulose (HPMC) was found to increase the solubility of polyvinylpyrrolidone (PVP) in original [3]. As HPMC is similar as an expanded CD, this result also indicates that CD may combine to drugs without inclusion of which is due to the lipophilic interior cavity. Furthermore, self-association of CDs was proved with phase-solubility diagrams and semi-permeable cellophane membrane method [4,5]. Because all these study are base in liquid phase, we wonder to discuss about the structure of CD complexes in solid phase.

CDs modified with substituents such as 2-hydroxypropyl, methyl, maltosyl, glucuronyl, sulfobutyl ether, and amino groups have been developed to increase the aqueous solubility, bioavailability, physical and biological stability, to reduce cytotoxicity and irritation, and to mask bitterness and odor [6,7]. 2-Hydroxypropyl-β-cyclodextrin (HP-β-CD; hydroxypropylbetadex) requires only simple chemical modifications, although it exhibits excellent complexation with drugs. It shows improved aqueous solubility of drugs and resistance to chemical and photodegradation, and it reduces irritation [8,9]. HP-β- CD has been approved for clinical use in oral and intravenous products and suppositories in the United States, Belgium, and Switzerland [1].

The calcium channel blocker nifedipine (NIF; CAS 21829-25-4, Scheme 1) is a 4-aryl-1,4-dihydropyridine-3,5-dicarboxylate ester derivative (1,4-DHP), which is widely used for moderate hypertension and cerebral disease [10]. Several derivatives, nimodipine (CAS 66085- 59-4), felodipine (CAS 72509-76-3), nilvadipine (CAS 75530-68-6), and lacidipine (CAS 103890-78-4), are in clinical trials [11]. In particular, nicardipine (NIC; CAS 55985-32-5, Scheme 1) and amlodipine (CAS 88150-42-9) are unusual because of their ammonium cation moieties, which are connected by an aliphatic chain [12-14], whereas other derivatives, including NIF, are neutral.

Scheme 1: Structures of NIF and NIC salt.

In this study, we used NIF and NIC as typical examples of a neutral and a cationic drug, respectively. Complexing NIF and NIC with HP- β-CD increases the bioavailability of these drugs and improves their solubility. Chutimaworapan et al. demonstrated an intermolecular interaction between the host HP-β-CD and the guest NIF by showing that complexing NIF with HP-β-CD converted the crystalline NIF to an amorphous state [15]. Wang et al. showed that controlling the viscosity of solutions of various concentrations of HP-β-CD with hypromellose could control the release rate of NIF, and this improved NIF bioavailability through reducing first-pass metabolism [16].

Forming a solid dispersion of HP-β-CD and NIC produces a higher dissolution rate than dispersions of β-CD [17]. NIC/HP-β-CD tablets containing glycerol monostearate ester and PEG 4000 controlled the release of NIC over 8 h [18]. NIF/HP-β-CD or NIC/HP-β-CD are increasingly being investigated for treating cardiovascular disorders; however, the interaction of NIF and NIC amorphous with HP-β-CD in host-guest complexes is not as well understood. Therefore, in this work, we investigate the intermolecular interactions and structure of NIF and NIC with HP-β-CD to improve the physicochemical properties of the complexes by altering the production processes and auxiliary materials.

Materials

HP-β-CD, NIF, NIC hydrochloride (NIC-salt) and all other chemicals and solvents were purchased from Wako Pure Chemicals (Kyoto, Japan). Because NIF and NIC are light sensitive, samples were protected from light by wrapping the containers in aluminum foil.

The degree of substitution (DS) of HP-β-CD was confirmed by electrospray ionization mass spectrometry (910, Varian, Palo Alto, CA). The DS of HP-β-CD was calculated as 6.32 ± 0.74 from the mass spectrum peak heights.

The physical mixture (PM) method consisted of mechanically mixing NIF or NIC with HP-β-CD at various molar ratios for 5 min with an agate mortar and pestle. The solution mixture (SM) method consisted of preparing solutions of NIF or NIC and HP-β-CD at various molar ratios in acetone and ethanol at 313 K, and then evaporating the solvent at room temperature for 3 days to obtain a powder.

Differential scanning calorimetry

Differential scanning calorimetry (DSC; Thermo Plus 8230, Rigaku, Co., Tokyo, Japan) was performed on 5 mg samples in a closed aluminum pan. The sample was heated to 310-455 K at 10 K min^-1 under a nitrogen gas flow of 30 mL min^-1. The change in the enthalpy of fusion, Δ_fusH, was estimated from the area enclosed by the endothermic curve and baseline. The change in the fusion entropy, Δ_fusS, was obtained as Δ_fusH divided by the melting temperature, T_m, according to Clausius’ definition.

X-ray powder diffraction

X-ray powder diffraction (XRPD; RINT 2000, Rigaku) was performed with monochromatic Cu Kα radiation as an X-ray source. Samples were analyzed by the parallel beam method using cross-beam optics through a 2θ range from 5° to 40° at a scanning velocity of 0.02 steps.

Fourier transform infrared spectrometry

Fourier transform infrared (FTIR) spectroscopy (Frontier, Perkin- Elmer, Inc., Waltham, MA) was performed in the range 4000-600 cm^-1 with sample 5 ± 1 mg in room temperature collected with resolution of 1 cm^-1, in reflection mode with a universal attenuated total reflection sampling accessory. Absorbance, which is proportional to the compound concentration, was calculated as the common logarithm of reciprocal transmittance, T^-1, according to the Beer-Lambert law.

Theoretical study with molecular dynamics

Atomistic molecular dynamics were explored by Materials Studio 6.1 to calculate the stable simulation of the complex of NIF/HP-β-CD and NIC/HP-β-CD. The guest drug, namely NIF or NIC, oriented to the host molecule HP-β-CD were put into an amorphous cell. Geometry optimization (Forcite GeomOpt) and Annealing Dynamics (Forcite Anneal) were used to refine the structures of complexes, NIF/HP-β-CD and NIC/HP-β-CD under the above environment. The intermolecular interactions of NIF/HP-β-CD and NIC/HP-β-CD were simulated by molecular dynamics calculation under the NVT ensemble at 300 K in 500 ps, monitored at the time intervals of 1 fs.

Thermal analyses of NIF and NIF/HP-β-CD mixtures

We recorded DSC thermograms of neutral NIF, HP-β-CD, and NIF/HP-β-CD mixtures. The mixtures were prepared by PM, in which HP-β-CD and NIF powders were mechanically mixed, or by SM, in which solutions of HP-β-CD and NIF were mixed and then dried to a powder. A series of NIF/HP-β-CD mixtures containing 0%-100% (molar fraction, intervals of 10%) NIF were prepared by PM. The DSC thermograms of the PM mixtures in the temperature range of 310-455 K are shown in Figure 1 (a). The thermograms of pure HP- β-CD powder and the NIF/HP-β-CD mixtures (10%-90%) contained low, broad endothermic peaks at about 378 K, corresponding to the dissociation of water from HP-β-CD. Endothermic peaks appeared in the range 444.7-445.5 K in the DSC thermograms of pure crystalline NIF and the NIF/HP-β-CD mixtures (10%-90%), indicating that these peaks corresponded to crystalline NIF fragments fusing (intrinsic melting point 445-447 K) as they dispersed in the mixtures. According to Cabral Marques et al. when guest molecules are incorporated into the interior cavity of CD, the melting, boiling, or sublimation points of complexes usually shift or disappear in the temperature range where CD decomposes. There was no significant shift in the NIF fusion temperature (Figure 1 (a)), indicating that there was little interaction between NIF and HP-β-CD under these conditions.

Figure 1: DSC thermograms of pure NIF, NIF/HP-β-CD mixtures (90%–10% NIF), and pure HP-β-CD prepared by the PM (a) and SM (b) methods. Subscript numbers on the right indicate the molar ratios of NIF:HP-β-CD.

The peak area of the NIF fusion peak in the DSC thermogram corresponded to the fusion entropy per mole of NIF in the NIF/HP- β-CD mixture. The peak area of both pure crystalline NIF and the NIF/HP-β-CD mixtures (10%-90%) prepared by the PM method are plotted in Figure 2. The entropy change required to melt 1 mol of crystalline NIF (Δ_fusS) remained constant at 74.83 ± 4.14 J/mol/K, indicating that preparing the HP-β-CD complex by the PM method was thermodynamically ineffective for crystalline NIF. We calculated the entropy change to melt 1 mol of crystalline NIF according to Yalkowsky’s empirical formula, ΔS_m=50-Rlnσ+Rlnφ, where 50 is Walden’s constant, R is the gas constant, and the parameters σ and φ are derived from the indexes allocated to portions of the chemical structure [19-21]. The fusion entropy of crystalline NIF was estimated to be 67.4 J/mol/K, which was slightly less than the experimental value within an acceptable range. The NIF-fusion temperature and the fusion entropy showed that the NIF molecule did not absorb sufficient heat to enter the HP-β-CD interior cavity during PM preparation.

Figure 2: Phase diagrams of the fusion entropies of crystalline NIF for pure NIF, NIF/HP-β-CD mixtures (90%–10% NIF), and pure HP-β-CD prepared by the PM method (squares), and by the SM method (circles). Values are the mean ± standard deviation of three separate measurements.

NIF/HP-β-CD mixtures were prepared by SM with the same ratios as the mixtures prepared by PM, and Figure 1 (b) shows the DSC thermograms. The DSC thermogram of pure NIF prepared by SM showed an endothermic peak at 441.2 to 445.0 K corresponding to recrystallization of NIF. Similar peaks were observed in DSC thermograms of NIF/HP-β-CD mixtures containing 60%-90% NIF, whereas no recrystallization peaks were observed for NIF/HP-β-CD mixtures containing 10%-50% NIF (Figure 1 (b)).

Figure 2 shows the fusion entropies per mole of NIF in the mixture obtained by SM (Δ_fusS). For NIF/HP-β-CD mixtures containing less than 50% NIF, Δ_fusS of NIF was not observed in the DSC thermograms, whereas for mixtures containing more than 50% NIF, Δ_fusS increased with the amount of NIF. The results indicated that the 50% NIF/HP-β- CD mixture formed a complex, and in mixtures containing more than 50% NIF, Δ_fusS of NIF corresponded to the phase transformation of the excess NIF.

The fusion behavior of the mixtures prepared by PM and SM indicated that only SM produced a host-guest complex between NIF and HP-β-CD. In addition, the endothermic peak at about 378 K corresponding to water immobilized in HP-β-CD, was the same as for PM, suggesting that the water in the HP-β-CD cavity was hard to displace.

Thermal analyses of NIC and NIC/HP-β-CD mixtures

Next, we studied the complexation of the cationic drug, NIC, in HP-β-CD. In accordance with the PM and SM methods, NIC/HP-β- CD mixtures were prepared at concentrations from 0%-100% NIC at intervals of 10%. The DSC thermograms of the NIC/HP-β-CD mixtures prepared by PM are shown in Figure 3 (a). An endothermic peak appeared at 428.7-440.0 K in the DSC thermogram of pure NIC-salt, corresponding to the melting point of crystalline NIC-salt. Although similar peaks were observed in the DSC thermograms of NIC/HP-β- CD mixtures containing 20%-90% NIC-salt, the temperatures varied. As the HP-β-CD content increased in the PM samples, the NIC-salt melting point decreased. The heat absorbed to melt the NIC-salt crystal was used to estimate the fusion entropy per mole of NIC-salt (Δ_fusS) (diamonds, Figure 4). The phase diagram of Δ_fusS of NIC-salt showed a linear relationship between the Δ_fusS of NIC-salt and the amount of NICsalt in the mixtures, unlike the NIF/HP-β-CD mixtures. Because Δ_fusS of NIC-salt was estimated as an average of crystalline and amorphous phases, we assumed that the intermolecular interaction between NIC and HP-β-CD was different from that of NIF. The estimated entropy of the melting of crystalline NIC was 106.6 J/mol/K [20,21], which was slightly larger than the experimental value (100.9 J/mol/K), probably because of the flexibility of the tertiary amine chain.

Figure 3: DSC thermograms of pure NIC-salt, NIC/HP-β-CD mixtures (90%–10% NIC), and pure HP-β-CD prepared by the PM (a) and SM (b) methods. Subscript numbers on the right indicate the molar ratios of NIC:HP-β-CD.

Figure 4: Phase diagrams of the fusion entropies of crystalline NIC for pure NIC, NIC/HP-β-CD mixtures (90%–10% NIC), and pure HP-β-CD prepared by the PM method. Endothermic peaks were not obtained for the samples prepared by the SM method in Figure 3(b). Values are the mean ± standard deviation of three separate measurements.

In contrast, the DSC thermograms of the SM NIC/HP-β-CD mixtures did not contain endothermic signals for NIC-salt. Pure NICsalt was converted to the amorphous phase by the SM method, and it was stable for 2 months. We measured the XRPD of the SM-prepared NIC/HP-β-CD mixtures by DSC-XRPD at a heating rate of 5 K min^-1 from 300 to 450 K under a nitrogen gas flow of 30 mL/min. The DSC thermograms contained no NIC-salt peaks, and showed halo patterns (Figure 6). The DSC and DSC-XRPD results confirmed that NIC prepared by SM formed a stable amorphous phase.

Figure 5: XRPD of pure NIF, NIF/HP-β-CD mixtures (90%–10% NIF), and pure HP-β-CD prepared by the PM (a) and SM (b) methods. XRPD of NIF/HP-β-CD mixtures (95%, 90%, 85% NIF) (c). Subscript numbers on the right indicate the molar ratios of NIF:HP-β-CD.

Figure 6: XRPD of pure NIC-salt, NIC/HP-β-CD mixtures (90%–10% NIC), and pure HP-β-CD prepared by PM (a) and SM (b) methods. Subscript numbers on the right indicate the molar ratios of NIC:HP-β-CD.

Glass transition signals at 342, 340, and 339 K were observed in the DSC thermograms of the SM NIC/HP-β-CD mixtures containing 80%, 90%, and 100% NIC, respectively (Figure 3 (b)). Because these mixtures contained a large amount of NIC-salt, polar interactions were dominant among the amorphous protonated NIC cations and the NIC-salt chloride anions. Thus, the mixtures were probably stabilized by these strong interactions at lower temperatures, whereas the the mixtures became fluid above the glass transition point. The molecular weight of HP-β-CD is 1541.54, and is 3.22 times that of the protonated NIC cation; therefore, the molar ratio of NIC:HP-β-CD of 8:2 was similar to the 1:0.8 mass ratio. Consequently, in mixtures containing up to 80% NIC, the majority of HP-β-CD was occupied by NIC. The dominant intermolecular interaction in the mixture was non-polar and was weaker than the electrostatic interaction. Therefore, mixtures containing 0%-70% NIC could not maintain glass phases at low temperatures. This would explain why a glass transition was observed in only those mixtures containing 80%-100% NIC.

X-ray powder diffraction of NIF and NIF/HP-β-CD mixtures

We performed XRPD for NIF to determine whether the mixtures contained single or multiple polymorphs of crystalline NIF. Pure HP- β-CD showed a halo pattern, whereas pure NIF exhibited a crystal pattern containing sharp diffraction peaks at 2θ of 8.3°, 10.6°, 12.0°, 16.4°, and 19.8° (Figure 5 (a)). The NIF diffractogram corresponded to crystalline polymorph A [22]. The diffractograms of the NIF/HP-β- CD mixtures prepared by PM consisted of a combination of the halo pattern of pure HP-β-CD and the sharp peaks of pure NIF (Figure 5 (a)), and the peak intensity was proportional to the molar ratio of NIF and HP-β-CD. NIF remained in the crystalline phase, indicating that the intermolecular interaction with HP-β-CD was weak.

In contrast, the crystalline NIF peaks were not present in the XRPD diffractogram of NIF/HP-β-CD mixtures prepared by SM containing up to 50% NIF, and they were similar to that of pure HP-β-CD, although we obtained the sharp peaks corresponding to crystalline NIF in mixtures containing 60%-90% NIF (Figure 5 (b)). The absence of the crystalline NIF peaks indicated that, in contrast to the PM method, the SM method converted NIF to the amorphous phase through the intermolecular interaction between NIF and HP-β-CD. When crystalline NIF was dissolved in acetone/ethanol, the NIF molecules could interact with HP-β-CD as the solvent evaporated, forming an amorphous phase.

The diffractogram of the SM NIF/HP-β-CD mixture containing 90% NIF contained characteristic polymorph B peaks at 2θ of 7.14° and 23.93°, although the corresponding peaks were not observed for pure NIF or for the other NIF mixtures (Figure 5 (b) and (c)) [22]. Burger and Koller reported a XRPD diffractogram of NIF polymorph B crystallized from 1,4-dioxane, which was similar to the 90% NIF mixture [23]. Furthermore, Caira et al. reported that polymorph B crystallized from 1,4-dioxane was converted to polymorph A at a high temperature [24]. Although NIF/HP-β-CD mixtures prepared by PM did not contain polymorph B, Hirayama and Wang reported that polymorph B was partially transformed in mixtures that were heated above the melting point of NIF and were immediately cooled to room temperature [22]. The metastable polymorph was obtainedfrom our NIF mixtures following hot and cold processing. Polymorph A was converted to metastable polymorph B because of the presence of 10% HP-β-CD. The metastable state was probably maintained by the hydroxypropyl substituents attached to the guest polymers and the modified cyclodextrin.

In summary, the DSC thermograms and XRPD diffractograms of PM mixtures of NIF were similar and the size of the crystalline NIF peaks was proportional to the molar ratio of NIF in the mixture. The thermograms and diffractograms of SM mixtures of NIF were also similar, except for the 90% NIF mixture. However, there were no clear endothermic fusion peaks or crystal diffraction peaks for the mixtures containing up to 50% NIF, indicating that crystalline NIF was converted into amorphous NIF. In the mixture containing 50% NIF, NIF formed a host-guest complex with HP-β-CD. This result also complied with XRPD diffractograms which were no obvious peaks present under 50% NIF.

X-ray powder diffraction of NIC and NIC/HP-β-CD mixtures

We obtained XRPD diffractograms of NIC mixtures prepared by the PM method. Crystalline NIC-salt showed signals at 2θ of 7.8°, 14.0°, 20.5°, and 22.8° (Figure 6 (a)). The intensity peaks for the NIC/ HP-β-CD mixtures containing 10%-90% NIC depended on the NIC content (Figure 6 (a)). Although the phase diagram of the Δ_fusS of NIC was different from that of NIF, the XRPD diffractograms of the PMs of NIF/HP-β-CD and of NIC/HP-β-CD were similar. In contrast, no diffraction peaks were obtained for the SMs and pure NIC-salt (Figure 6 (b)). This result indicated that the NIC-salt crystal was converted to the amorphous phase, consistent with the DSC experiments.

The diffraction peaks of crystalline NIC-salt were obtained in the PMs of NIC (Figure 6 (a)). Δ_fusS of NIC-salt decreased linearly with the NIC content (Figure 4). Because the fusion temperature of NICsalt is lower than that of NIF, it should be easier to recrystallize NICsalt than NIF. However, there are many possible conformations of the benzylmethylaminoethyl side chain in NIC and the asymmetric substituents at the dihydropyridine moiety. Furthermore, the hydrochloride salt consists of a chloride anion and the protonated NIC cation. These properties would make the recrystallization of NIC- salt or other intermolecular interactions entropically unfavorable. If the structure of NIC is less suitable for intermolecular interactions than NIF, then the crystalline phase would not be maintained at low NIC contents in the PMs with HP-β-CD. Hence, Δ_fusS was less than equivalent to the mixtures containing crystalline NIC-salt. In addition, recrystallization from the SMs should be impossible, explaining the halo patterns of the SMs and pure NIC-salt.

FTIR Spectra of NIF/HP-β-CD and NIC/HP-β-CD mixtures

We investigated the substituents of the guest molecules that interacted with the host molecule by analyzing the FTIR spectra of the host-guest complexes. The FTIR spectra of pure NIF and the NIF/ HP-β-CD mixtures prepared by PM and SM are shown in Figure 7. In the spectrum of pure NIF, NIF absorbances were observed at 3331, 1680, 1527, 1493, 1432, 1348, 1310, 1225, 1120, 1100, 858, and 828 cm^-1 (underlined wavenumbers indicate absorbances that had intensities that were not proportional to the content of the mixture). The absorbance intensities decreased linearly with the NIF content in the PM NIF/HP-β-CD mixtures (Figure 7 (a)). The signals at 1527 (NH bending), 1432 (aromatic NO₂ stretching), 1310 (C-N symmetric bending), 1225 (ester C-O stretching), and 858-828 cm^-1 (aryl NO₂ stretching), and several lower wavenumbers, corresponded to NIF in the SM NIF/HP-β-CD mixtures. The signals at 3331 (amine N-H stretching), 1680 (ester group C=O stretching), 1493 (CH bending), 1348 (ether and N-H stretching), 1120 (ester C-O stretching), and 1100 cm^-1 (C-O stretching) were not present in the FTIR spectra of the SM mixtures containing up to 50% NIF (Figure 7 (b)).

Figure 7: FTIR spectra of pure NIF, NIF/HP-β-CD mixtures (90%–10% NIF), and pure HP-β-CD prepared by the PM (a) and SM (b) methods. Subscript numbers on the left of the spectra are the same as the sequences in Figures 1 and 3.

If the intensity of an IR absorbance is proportional to the amount of the guest molecule, then the substituents of the guest molecule and the binding site of the host molecule may interact. Shifts in IR absorptions would confirm that the corresponding substituents of the guest molecules interact with the host molecule and that the vibration modes of the substituents couple with the binding site. Therefore, our results suggest that the 1-4-dihydropyridine-3-5-dicarboxylate dimethyl ester acted as binding sites in the interior cavity of HP-β- CD. Chutimaworapan et al. have reported similar findings for the dissolution of water-soluble carriers in NIF solid dispersions.

The FTIR spectra of the pure NIC-salt and NIC/HP-β-CD mixtures are shown in Figure 8. Peaks for pure crystalline NIC-salt were observed at 3253 (N-H stretching), 3182 (C-H stretching), 3079, 2974, 2957, 2494, 1703 (ester C=O stretching), 1647, 1623 (phenyl C-C stretching), 1536 (asymmetric NO₂ stretching), 1494 (phenyl stretching), 1354 (symmetric NO₂ stretching), 1273, 1203, 1102, 1016 cm^-1, and several lower wavenumbers (744, 710, 699 cm^-1), as shown in Figure 8 (a) (underlined wavenumbers were present in crystalline NIC-salt and amorphous NIC spectra). The spectra of crystalline NICsalt and amorphous NIC were different, which was confirmed by the halo pattern in Figure 6 (b). Amorphous NIC gave peaks at 3192, 3079, 2957, 1683, 1647, 1623, 1524, 1484, 1434, 1382, 1347, 1309, 1272, 1208, 1116, 1092, 1018 cm^-1, and several lower wavenumbers (789, 744, 700 cm^-1), as shown in Figure 8 (b). The signal at 1703 cm^-1 for crystalline NIC-salt shifted to 1683 cm^-1 in amorphous NIC in the presence and absence of the host molecule. The signals at 1536 and 1354 cm^-1 in crystalline NIC-salt were shifted to 1524 and 1347 cm^-1 in amorphous NIC in the presence and absence of the host molecules. Therefore, these shifts arose from the crystal/amorphous transition of NIC, rather than the intermolecular interaction between the ester C=O group and the aromatic NO₂ group of NIC with HP-β-CD. The signals at 710 and 699 cm^-1 in crystalline NIC-salt corresponded to the broad signal at 700 cm^-1 in amorphous NIC. Only the intensity of these signals depended on the content NIC of the mixtures prepared by PM and SM in Figure 8.

Figure 8: FTIR spectra of pure NIC-salt, NIC/HP-β-CD mixtures (90%–10% NIC), and of pure HP-β-CD. They were prepared by the PM (a) and SM (b) methods. Subscript numbers on the left of the spectra are the same as the sequences in Figures 1 and 3.

Structure of Host-Guest Complexes Simulated by Molecular Dynamics

NIF/HP-β-CD and NIC/HP-β-CD complexes were prepared in amorphous cells in vacuum as the density with 1:1 (host: guest) in 1 cell. After treating with anneal between 300-500 K and NVT dynamic for 500 ps in 300 K, complexes were considered as finally stable state. The difference value of energy was 206.7 kcal/mol in NIF/HP-β-CD complex forming, and 256.9 kcal/mol in NIC/HP-β-CD forming. More energy releasing of NIC system indicated NIC/HP-β-CD complex were assembled easily than that of NIF, and the reason maybe the muti-site interaction in NIC system on HP-β-CD outside surface.

Obviously in Figure 9, after simulation, the molecular of NIF inserted the HP-β-CD cavity deeply with benzene ring. While, the molecular of NIC closed to the HP-β-CD edge and was caught between 2 HP-β-CDs. This was another evidence to prove the different structure between NIF/HP-β-CD and NIC/HP-β-CD.

Figure 9: Images of structure of NIF/HP-β-CD and NIC/HP-β-CD complexes drawing by molecular dynamics.

We performed DSC experiments on NIF/PEG 4000 mixtures. The fusion temperature was gradually decreased to 392.35 K at a molar ratio of 90.4% NIF in the NIF/PEG 4000 mixture. NIF and PEG 4000 co-melted samples were prepared by heating to 453.15 K and cooling to room temperature. The results agreed with data published by Zajc and Srčič [25]. Because the fusion entropy per mole of NIF in the mixture decreased from 100% to 90.4% NIF, we constructed a fusion entropy phase diagram for NIF/PEG 4000 mixtures (Figures 9 and 10) to show that NIF/HP-β-CD mixtures and NIF/PEG 4000 mixtures are comparable. Consequently, because the x-intercept of the fusion entropy trendline was about 8/9 (approximately 89%) it was estimated that 1 mol of PEG 4000 enclosed complexed about 8 mol of NIF; there are 8 vertexes in a cuboid or parallelepiped. The threshold in the phase diagram was interpreted as the maximum number of guests per host molecule, namely HP-β-CD or PEG 4000. The phase diagram of NIC/ HP-β-CD mixtures prepared by PM suggested that the HP-β-CD molecules infinitely interact with a protonated NIC cation. Therefore, HP-β-CD may act as the medium for the NIC-salts, and the NIC-salts may act as an adhesive, aggregating the HP-β-CD molecules. A similar conclusion has been proposed, where the drug-CD system may have a non-inclusion complex or micelle-like structure [2].

Figure 10: Fusion entropy phase diagrams of pure NIF, NIF/PEG 4000 mixtures (90%–10% NIF), and pure PEG 4000. The samples were prepared by co-melting. No endothermic peaks were observed for 40%–0% NIF. The molecular weights of NIF and PEG were 346.335 and 4000.000 g/mol, respectively.

Hydrophobic interactions usually drive a guest drug to enter the hydrophobic interior cavity of native or modified CDs. In contrast, all the hydrophilic groups, such as 2-OH, 3-OH, and 6-OH groups on the glucopyranoside units and the other polar substituents of the modified regions (for example, 2-OH of hydroxypropyl groups), are on the exterior interface of native or modified CDs. Therefore, hydrophobic NIF and the hydrophilic NIC cation are oriented to the hydrophobic interior and to the hydrophilic exterior of HP-β-CD, respectively (Scheme 2). Different charges on the guest molecule would result in opposite interactions with the host, even though they contained the same 1,4-DHP backbone.

Scheme 2: Structure of HP-β-CD and schematics of NIF/HP-β-CD and NIC/ HP-β-CD systems.

Hydrophobic NIF may enter the interior cavity of HP-β-CD despite the thermodynamic barrier. In contrast, the polar NIC-salt would interact with the exterior interface of HP-β-CD. Hence, the PM NIC/HP-β-CD could easily melt stiffness in the crystalline NIC-salt. In contrast, NIC/HP-β-CD prepared by SM mixed the CD derivatives and the protonated NIC cations/chloride anions independently of the amount of NIC-salts in the mixture. In an exterior interaction, the CD derivatives acted as the medium in a solid dispersion system, such as PEG or hydroxypropyl cellulose.

SM allows to NIF bind to the interior cavity of HP-β-CD to form an equimolar complex, whereas no complex was formed by PM. In contrast, crystalline NIC-salt was readily converted to the amorphous phase even by PM of NIC/HP-β-CD, and did not form an inclusion complex by PM and or SM. When the amount of NIC-salt was equal to or less than HP-β-CD, the glass transition of the ionic crystal disappeared, indicating that NIC-salt was probably dispersed in the HP-β-CD matrix in the mixtures. This result indicated that the NIC cations and chloride anions were adsorbed on the exterior surface of the HP-β-CD molecule or occupied the gaps between the HP-β-CD molecules (Scheme 2). Therefore, the stoichiometry of the complex formation between NIC and HP-β-CD was not determined. We constructed fusion entropy phase diagrams that showed the different intermolecular interactions in NIF/HP-β-CD and NIC/HP-β-CD.

We thank the Hirose International Scholarship Foundation for supporting the study and research of author Y-P. L.