Stability and Properties of Ultraviolet Filter Avobenzone under Its Diketo/Enol Tautomerization Induced by Molecular Encapsulation with β-Cyclodextrin

Abstract

Inclusion complexation of the sunscreen ingredient avobenzone (AVB) with β-cyclodextrin (β-CD) was investigated to improve its aqueous solubility and photostability; another ultraviolet (UV) filter, oxybenzone (OXB), and the phytochemical antioxidant curcumin (CUR) served as a comparison. In this study, the 1-octanol/water partition coefficients, acid dissociation constants, phase-solubility diagrams with β-CD, and ultraviolet-visible (UV-vis) spectral changes induced by UVA1 (365 nm) irradiation were evaluated. β-CD at concentrations 50-100 times that of AVB most effectively protected the photostability of AVB. Additionally, an UVA1-insensitive species with a diketo tautomer, which has an UVC-absorbing band and the potential to cause photodegradation, was stored in the inclusion complex. Acetonitrile-water mixtures at various volume ratios were screened to mimic the internal cavity of β-CD for the AVB tautomeric species using nuclear magnetic resonance (NMR) spectral integrals for the components. The results indicated that β-CD provides a hydrophobic environment similar to that of a 40-50% acetonitrile aqueous solution and enhances the photostability of AVB. However, excess β-CD induced a hyperchromic effect on the diketo tautomer. Aggregation of the AVB/β-CD inclusion complexes at β-CD concentrations of ≥2 mM enhances UVC band absorption. To avoid excess β-CD, a molar ratio of 50-100 of β-CD to AVB is recommended as the optimal composition. This study newly exhibited that the cavity of β-CD mitigates the reactivity of UVA1 toward AVB by inducing the diketo tautomer form of AVB within the cavity.

Chart 1. Chemical Structures and Properties of the Dominant Chemical UV Filter Ingredients

Figure 1

pH profiles of distribution coefficients, log D, for (a) AVB, (b) OXB, and (c) CUR. The 1-octanol/water (the modified Britton–Robinson buffers at pH 5–11.5) partition coefficients, log P, the acid dissociation constants, pK_a, were optimized using curve fitting approximated to log D = (log P)/(1 + 10^pH–pK_a), which is Avdeef’s diagram derived from the Henderson–Hasselbalch expression for the equilibrium between the protonated and deprotonated species. Curve-fitting procedures employed the Solver module of Microsoft Excel 2016 with the implemented GRG nonlinear option. The approximated curves corresponding to these optimized parameters were drawn, and the pK_a and log P values for AVB, OXB, and CUR were represented. The inset photograph shows the equilibrium CUR partitioned in 1-octanol/water phases at several pH values.

Figure 2

Phase-solubility diagrams of (a) AVB, (d) OXB, and (c) CUR with β-CD and (d) AVB, (e) OXB, and (f) CUR with HP-β-CD in 25 mM P_i/NaOH buffer at pH 6.8. The regression lines for panels a, b, and e provided the determination coefficients, 0.9223, 0.9994, and 0.9872. The parabolic curves for panels c, d, and f brought the determination coefficients, 0.9946, 0.9995, and 0.9901.

Figure 3

UV–vis spectra of 50 μM (a) AVB, (b) OXB, and (c) CUR irradiated with the UVA1 (365 nm) lamp in 25 mM Tris–HCl (pH 7.4). The spectra of panels d–f correspond to these drugs irradiated with the UVA1 lamp in 8 mM β-CD.

Figure 4

(a) UV–vis spectral titration for 50 μM AVB with 0–10 mM β-CD. In 25 mM P_i buffer (pH 6.5). The spectra of AVB were represented with gradually diluted curves depending upon the concentration of β-CD. They peaked at 388 nm, and spectra gradually morphed and descended with 0, 1/40, 1/20, 1/10, 1, and 2 mM β-CD. Furthermore, twin peaks at 274 and 365 nm expanded with 2, 4, 6, 8, and 10 mM β-CD. The signal at 388 nm would be assigned to the keto-enol species in the aqueous phase, while the signals at 274 and 365 nm would be to the diketo and chelated keto-enol species in β-CD’s hydrophobic internal cavity, respectively. UVA1 irradiation insignificantly degraded AVB with (b) 2 mM and (c) 4 mM β-CD for 0–6 days.

Figure 5

(a) Reversed-phase HPLC chromatogram of intact AVB. Stationary and mobile phases were the ODS column and the Pi buffer (pH 2.5) in H₂O/acetonitrile = 1:1, respectively. Signals were observed with the photodiode array (PDA) at 200–600 nm wavelength. Panels b and c represented the cross sections of spectra at the retention times t_R of 2 and 4 min, respectively. The keto-enol form is planar and absorbs in the UVA1 range (340–400 nm), while the diketo tautomer absorbs in the UVC range (200–280 nm).

Figure 6

Time evolution of the degradation of (a) 25 μM AVB, (b) 50 μM OXB, and (c) 25 μM CUR in the absence (black circles) and presence (equimolar, Bordeaux triangles; 4 mM, indigo rhombuses) of β-CD in isochoric acetonitrile solvent under the UVA1 irradiation (365 nm). AVB, OXB, and CUR signals and their calibration lines were determined wavelengths at 360, 290, and 430 nm, respectively. The calibration lines provided r² = 0.999 and 3σ = 0.022. The obtained diagrams of the logarithm of AVB concentration to the reaction time seemed to have downward convex curves rather than first-order linear functions. Figures S4–S6 represented the corresponding observed spectra.

Figure 7

Absorbance of AVB at 270 and 360 nm in a solvent with various proportions of (a) acetonitrile/H₂O and (b) methanol/H₂O. It represented intact ABZ with circles and the equimolar mixture of AVB and β-CD with triangles. Supposing the solvent molecules are homogeneous volume spheres, the geometrical kissing number in three dimensions is 12. It indicates a hexagonal close-packed lattice where a central sphere contacts 3–4 surrounding atoms. Hence, if the solvent component contains a higher volume ratio than 66–75%, the surroundings of a solute would be almost filled with these solvent molecules. As acetonitrile and methanol molecules are distorted, we considered that this threshold proportion would be attenuated. That could be why the bending points are at about 60% (v/v).

Figure 8

Time evolution of the keto-enol and diketo molarity increase/decrease in the (a) aqueous buffer, (b) 40% acetonitrile, and (c) 70% MeOH upon adding none (circles), equimolar (25 μM, triangles), or excess (4 mM, rhombuses) β-CD. UVA1 (365 nm) irradiation gradually degraded a part of the keto-enol form so that the amount of the keto-enol form (closed signs) descended. Meanwhile, the diketo form (open signs) was elevated. The protecting effect of the equimolar or excess β-CD on the photodegradation seemed equivalent to 40% acetonitrile. The 70% methanol solution experiments were limited, but we obtained similar results within the examinable range.