Electrostatic Interactions Enable Nanoparticle Delivery of the Flavonoid Myricetin.

Flavonoids are natural polyphenolic compounds with myriad biological activities and potential as prophylactic and therapeutic agents. However, poor aqueous solubility and low bioavailability have limited the clinical utility of flavonoids, suggesting that drug delivery systems (DDSs) may improve their clinical relevance. Therefore, loading of a representative flavonoid (i.e., myricetin) into a diblock, polymeric nanoparticle carrier (NPC) DDS with a cationic corona and hydrophobic core was investigated. Absorbance and fluorescence spectroscopy results revealed association constants and standard Gibbs free energy values that align with previously reported values (K a = ∼1-3 × 104 M-1; ΔG° = -5.4 to -6.0 kcal mol-1), suggesting that NPCs load myricetin via electrostatic interactions. The zeta potential and gel electrophoresis analysis confirmed this loading mechanism and indicated that NPCs improve myricetin solubility >25-fold compared to myricetin alone. Finally, the dual-drug loading of NPCs was tested using a combination of myricetin and a hydrophobic drug (i.e., farnesol). Electrostatic loading of NPCs with myricetin at concentrations ≤1.2 mM did not affect NPC core loading and release of farnesol, thus demonstrating a novel formulation strategy for the dual-drug-loaded NPC. These findings offer key insights into the NPC DDS design that may enhance the clinical relevance of flavonoid-based therapeutic approaches.

Introduction

Among the largest and most diverse group of biologically active polyphenol compounds, flavonoids exhibit antioxidative, anti-inflammatory, anticarcinogenic, antidiabetic, cardioprotective, neuroprotective, analgesic, antivenom, antiviral, antibacterial, and antibiofilm properties.[1−7] These versatile functions have fostered interest in flavonoids for therapeutic and cosmetic applications.[1,2] Moreover, the established safety profile associated with regular human consumption of flavonoid-rich food and beverages has received recognition recently by the pharmaceutical industry as well as the general public.[2,3,8,9] Over the past 2 decades, more than two dozen clinical trials have been conducted in the United States on polyphenol compounds, which are officially classified as “botanical drugs” by the Food and Drug Administration (FDA). During that time, the FDA approved two botanical drugs, Polyphenon E in 2006 and crofelemer in 2012, and published an industry guidance document on botanical drug development.[8,10−12]

Despite the significant promise of flavonoids, poor aqueous solubility and low bioavailability have limited their development, clinical translation, and approval as botanical drugs.[2,6,13−15] Indeed, flavonoids are rarely used in liquid formulations because of poor aqueous solubility.[14] Furthermore, flavonoids are sensitive to environmental factors, such as temperature, pH, and light,[6] suggesting that use of a drug delivery system (DDS) may increase their clinical utility. Therefore, multiple DDSs, primarily polymeric and lipid-based DDSs, have been considered for use with flavonoids, such as quercetin, apigenin, and daidzein.[6,16] For example, polymeric micelles, such as poly(ethylene glycol)-derivatized phosphatidylethanolamine, Pluronic P123, and Solutol HS 15, have been shown to increase quercetin solubility >100-fold[17] and apigenin solubility ∼150-fold,[18] while lecithin-based micelles increased daidzein solubility >500-fold.[16] Moreover, multiple studies have sought to improve the solubility and bioavailability of a common, well-defined, and powerful antioxidant flavonoid, myricetin, using the following types of DDSs: cyclodextrins,[19] solid lipid nanoparticles,[20] and liposomes.[21] These studies have increased myricetin solubility ∼19-fold,[19] ∼12-fold,[20] and ∼5-19-fold,[21] respectively. However, to the best of our knowledge, no one has investigated the use of cationic, polymeric nanoparticles to enhance flavonoid delivery.

This work sought to improve the clinical relevance of a representative flavonoid, myricetin, by increasing its solubility using a diblock, polymeric nanoparticle carrier (NPC) DDS with a cationic corona and hydrophobic core. Because flavonoid solubility and mechanisms of interaction when combined with micelle carriers are influenced by surface charge,[22,23] a thorough evaluation of myricetin interactions with cationic NPCs was investigated. Additionally, the ability to coload NPCs with flavonoids and core-loaded hydrophobic drugs was studied.

Results and Discussion

Despite significant advances in understanding the therapeutic potential of flavonoids, poor aqueous solubility has limited their clinical translation.[1,2,6,14] Therefore, the use of a nanoparticle DDS to improve flavonoid solubility may enhance the clinical relevance of these natural products. Here, the flavonoid solubility was enhanced >25-fold using a diblock, polymeric NPC DDS (Figure A). In particular, this study evaluated and confirmed the mechanism of NPC loading for a representative flavonoid, myricetin, using spectroscopic and charge-based assays. Additionally, drug loading and release assays yielded key insights pertaining to the impact of flavonoid loading on the ability of NPCs to coload hydrophobic drugs. The effects of myricetin loading on hydrophobic drug loading and release were investigated using the terpenoid farnesol as a model hydrophobic drug. Overall, the findings presented here demonstrate the ability to coload NPCs with both a flavonoid and a hydrophobic drug, thus opening the door for future development of synergistic dual-drug-loaded (i.e., flavonoid and hydrophobic drug-coloaded) NPC therapeutic approaches.

Figure 1

Cationic NPCs hypothesized to coload hydrophobic drugs and flavonoids using different mechanisms. (A) Scheme showing the cationic NPC polymer synthesis process, diblock composition, and micelle self-assembly in aqueous conditions. (B) Cartoon illustrating the known mechanism of hydrophobic drug (e.g., farnesol) loading within the NPC hydrophobic core. (C) Cartoon illustrating the hypothesized mechanism of flavonoid (e.g., myricetin) loading with cationic NPCs via an electrostatic interaction. (D) Cartoon illustrating the proposed mechanism of coloading NPCs with farnesol and myricetin. (A,B) Reproduced with permission from ref (27).

Table provides a summary of poly(dimethylaminoethyl methacrylate)-b-poly(dimethylaminoethyl methacrylate-co-butyl methacrylate-co-propylacrylic acid) (p(DMAEMA)-b-p(DMAEMA-co-BMA-co-PAA)) NPC characteristics used in this work. The following groups were evaluated (Table ): NPCs with varied block 1 molecular weights (Mn) and similar block 2 Mn, NPCs with similar block 1 Mn and varied block 2 Mn, and multiple NPCs with similar block 1 and block 2 Mn (outlined in black, dashed line). All NPCs used had diameters between ∼12 and 47 nm and zeta potential values between ∼11 and 23 mV, each of which varied depending on the block 1 Mn and block 2 Mn parameters used. These differences enabled the evaluation of how the various molecular weights, sizes, and zeta potential values affected the capability of NPCs to load myricetin herein. Previous work has shown that hydrophobic drugs, such as farnesol, load into the hydrophobic core of polymer NPCs, as illustrated in Figure B.[24−26] Based on previous findings that flavonoid interaction mechanisms with micelle carriers are influenced by surface charge,[22,23] the mechanism of interaction between myricetin and cationic NPCs was investigated. Because a similar flavonoid, quercetin, was found to interact with cationic hexadecyltrimethyl ammonium bromide micelles via electrostatic mechanisms,[22] myricetin was hypothesized to interact electrostatically with cationic NPCs (Figure C). Upon confirmation of this interaction mechanism, a novel dual-drug delivery approach where NPCs were coloaded with myricetin within the NPC corona via electrostatics and the terpenoid farnesol within the core via hydrophobic interactions was assessed. Figure D depicts the proposed mechanism of coloading NPCs with a hydrophobic drug (e.g., farnesol) and a flavonoid (e.g., myricetin) investigated here.

Table 1

Characterization Summary of Polymers and NPCs Used in Myricetin Coloading Studiesa

	polymers
	corona block			core block					diblock copolymers			NPC
polymer	DP	block 1 Mn1 (kDa)	PDI1	DP	block 2 Mn1 (kDa)	% DMAEMA2	% BMA2	% PAA2	overall Mn1 (kDa)	PDI1	CCR	size3 (d nm)	size PDI3	Ζ4 (mV)
NP13/3	90	12.8	1.01	50	3.0	35	18	46	15.8	1.03	4.3	12.5 ± 3.8	0.19	14.3 ± 7.3
NP13/11	90	12.8	1.01	325	10.7	7	68	25	23.5	1.06	1.2	27.1 ± 7.0	0.09	23.2 ± 4.2
NP19/9	160	18.5	1.04	315	9.1	10	74	15	27.6	1.06	2.0	28.2 ± 6.8	0.06	22.3 ± 5.1
NP25/10	265	24.5	1.04	310	9.8	15	57	29	34.3	1.04	2.5	30.6 ± 8.2	0.09	16.8 ± 4.7
NP26/10	300	25.8	1.11	235	10.0	9	67	24	35.8	1.07	2.6	35.2 ± 9.1	0.08	19.8 ± 5.0
NP26/7	300	25.8	1.11	160	7.1	9	54	37	32.9	1.07	3.6	31.8 ± 8.1	0.09	19.9 ± 5.4
NP26/15	300	25.8	1.11	375	15.0	4	60	36	40.8	1.08	1.7	43.2 ± 10.4	0.06	20.5 ± 4.5
NP26/29	300	25.8	1.11	780	29.2	10	57	33	55.0	1.07	0.9	46.7 ± 11.7	0.07	20.9 ± 4.7
NP37/11	490	37.4	1.10	495	10.9	18	56	27	48.3	1.06	3.4	36.6 ± 7.2	0.11	12.3 ± 4.7

As characterized by 1gel permeation chromatography, 2proton nuclear magnetic resonance spectroscopy, 3DLS, and 4electrophoretic light scattering. Abbreviations: CCR, corona-to-core molecular weight ratio; DP, target degree of polymerization; PDI, polydispersity index (Mw/Mn); ζ, zeta potential.

The interaction between myricetin and cationic NPCs resulted in a color shift from clear to dark red because of a phenomenon known as copigmentation (Supporting Information Figure S1). Copigmentation occurs when pigments or copigments, such as anthocyanidins (e.g., the flavonoid myricetin), and other noncolored organic components (e.g., cationic, diblock copolymer NPCs) are combined in a solution and form weak supramolecular associations or noncovalent complexes that exhibit unexpected color changes far greater than the sum of their individual colors.[28,29] Commonly analyzed via absorption spectroscopy,[29] copigmentation generally results in an absorbance value enhancement, termed a hyperchromic shift, and a bathochromic shift (i.e., a red shift to a longer wavelength) in maximum absorbance wavelengths of the pigment.[28] Furthermore, copigmentation is a concentration-dependent process that is sensitive to changes in pH, temperature, and solvent.[29]

Flavonoids are hydroxylated polyphenolic compounds consisting of two phenyl rings (A and B) linked together via a heterocyclic pyran ring (C), as shown in Figure A.[2,6] This C6–C3–C6 structural backbone is known to exhibit a unique absorbance spectrum containing two main absorption bands.[2] Band I (λ ≈ 300–450 nm) represents the cinnamoyl moiety (blue, B and C rings in Figure A) and is associated primarily with the B ring adsorption.[2,30] Band II (λ ≈ 240–290 nm) represents the benzoyl moiety (green, A and C rings) and corresponds mainly to the A ring adsorption.[2,30]Figure B shows these bands in the absorbance spectra for multiple myricetin concentrations in phosphate buffer at pH 7.2. The band I peak is located at 330 nm (Figure B, center of the shaded region) with a shoulder at 380 nm, as shown previously,[30] and the band II peak is located at 230 nm. As shown in Figure C, these wavelengths of maximum absorbance—the band I peak (330 nm) and the band II peak (230 nm)—of the myricetin absorbance spectra underwent bathochromic shifts upon addition of a constant NPC concentration to each of the myricetin concentrations. The band I peak shifted to 380 nm (Figure C; horizontal blue, dashed arrow), and the band II peak shifted to 240 nm (Figure C; horizontal green, solid arrow). Moreover, the intensity of band I increased (Figure C; vertical blue, dashed arrow indicating a hyperchromic shift), while the intensity of band II decreased (Figure C; vertical green, solid arrow indicating a hypochromic shift). Although no entirely predictive and quantitative relationship of such effects currently exists, these spectral changes and an understanding of the spectroscopic features found in myricetin and NPCs offer mechanistic insights into the nature of the myricetin–NPC interaction.

Figure 2

Absorbance spectroscopy revealed that electrostatic interactions occur between myricetin and the NPC corona. (A) Chemical structure of myricetin highlighting the cinnamoyl moiety (blue, B and C rings) responsible for the band I absorbance peak (λ = 300–450 nm) and the benzoyl moiety (green, A and C rings) responsible for the band II absorbance peak (λ = 240–290). (B) Representative absorbance spectra for multiple myricetin concentrations (triangle down open: 0.08 mM, diamond solid: 0.10 mM, triangle up open: 0.12 mM, square solid: 0.14 mM, and circle open: 0.16 mM) in a pH 7.2 phosphate buffer; the band I peak located at 330 nm (center of the shaded region) with a shoulder at 380 nm and the band II peak located at 230 nm. (C) Representative absorbance spectra for multiple myricetin concentrations (triangle down open: 0.08 mM, diamond solid: 0.10 mM, triangle up open: 0.12 mM, square solid: 0.14 mM, and circle open: 0.16 mM) combined with a constant NPC concentration (circle solid: 0.008 mM) in a pH 7.2 phosphate buffer; blue, dashed arrows indicate the band I peak red-shifted to 380 nm with increased intensity, and green, solid arrows indicate the band II peak red-shifted to 240 nm with decreased intensity. (D) Double-reciprocal plot of absorbance data (λ = 380 nm) showing strong correlation (R2 = 0.97) between inverse absorbance and inverse myricetin concentration, resulting in an association constant (y-intercept/slope, Ka) of 1.9 × 103 M–1.

Hydrogen bonding, π–π interactions, and van der Waals forces likely combined to affect myricetin–NPC interactions. Intramolecular hydrogen bonding occurs between the 3-OH and 6′-H groups in myricetin (Figure ), resulting in planarity of the three rings[2,31] (i.e., A, B, and C in Figure ). Planarity is common among flavonoids, such as apigenin, kaempferol, and quercetin,[31−34] and facilitates ring π–π interactions, stabilizes the compound via resonance effects (i.e., charge delocalization), and contributes to planar stacking, which leads to copigmentation.[28,29,35] The π–π interactions among unsaturated rings also increase the extent of conjugation among the aromatic compounds and enable π–π* transitions, resulting in absorption band shifts to longer wavelengths. However, a stimulus is needed to initiate these events. For myricetin–NPC interactions, the electrostatic forces between tertiary amines within the NPC corona and specific hydroxyl groups in myricetin serve as the stimulus. In neutral conditions, such as those used in this study (e.g., pH 7.0–7.2), approximately 50% of NPC tertiary amines are protonated, while the 4′-OH group of myricetin (pKa = 6.33) is fully deprotonated.[36,37] Upon interaction with myricetin’s 4′-OH group, the tertiary amines become closely positioned to unsaturated chromophores (e.g., C=C) in myricetin, thereby enhancing the band I absorbance intensity and causing a bathochromic shift in the band I absorbance to a longer wavelength (e.g., ∼330 to ∼380 nm), as previously described.[38] Moreover, the planar structure of myricetin promotes π-delocalization from C to B,[31,32] resulting in shifts of band I absorbance spectra upon addition of NPCs (Figure C). The observed changes in band II, however, result from deprotonation[36,37] caused by NPC–myricetin interactions. After the 4′-OH group, the next myricetin deprotonation occurs in the 7-OH group (pKa = 7.57) on the A ring.[37] Although the buffer pH remained below the pKa of 7-OH in these studies, localized alkaline microenvironments due to NPCs may enable deprotonation, similar to other cationic micelles,[22,39] as solutions of NPCs in distilled, deionized water exhibit a pH of 8.3–8.6 (data not shown). 7-OH deprotonation results in multiple resonance conformations for myricetin (Supporting Information Figure S2), including two that potentiate hydrogen bonding between 3-OH and 4-C=O on C, ultimately forming a double bond between the B and C rings. This interaction transiently distorts the geometry of the molecule (i.e., decreases planarity), which is known to have a hypochromic effect on absorbance values,[38] resulting in the hypochromic shift observed for myricetin in the band II absorbance spectra in the presence of an NPC (Figure C).

Because of the hypothesis that myricetin and NPCs interact electrostatically, a host–guest binding model associated with an affinity-based DDS was employed to determine an association constant for the NPC–myricetin interaction.[40] By using the absorbance data compiled in Figure B,C, a double-reciprocal plot of the absorbance data was developed. This double-reciprocal plot showed a strong linear correlation (R2 = 0.97) between the inverse of absorbance value differences at λ = 380 nm and the inverse of myricetin concentrations resulting in a calculated association constant (y-intercept/slope, Ka) of 1.9 × 103 M–1 (Figure D). This association constant corresponded to a standard Gibbs free energy change (ΔG° = −RT ln(Ka)) of −4.4 kcal mol–1 of the myricetin–NPC interaction. These estimated values were very similar to the association constants and standard Gibbs free energy results for host–guest electrostatic interactions observed elsewhere.[22,41−43] In sum, the molecular level understanding of the absorbance spectral shifts observed combined with the estimated Ka supports hypothesized electrostatic interactions between myricetin and NPCs and yields key insights for consideration during future development of DDSs.

To substantiate the absorbance spectroscopy results, fluorescence analyses were also used. Fluorescence is capable of yielding more precise insights about host–guest (i.e., NPC–myricetin) interactions, such as association constants (Ka).[44−46] Furthermore, fluorescence enhancement occurs when a drug molecule with fluorescent properties forms a complex with a host molecule,[40] and the number of peaks is correlated with the number of tautomer species in solution. Fluorescence enhancement that occurs when NPCs are combined with myricetin in solution (Figure A) was used to understand the NPC–myricetin interaction. Steady-state fluorescence emission spectra for wavelengths between λEm = 500 and 620 nm using an excitation wavelength of λEx = 280 nm were recorded for a single NPC concentration (0.025 mM) titrated with increasing myricetin concentrations (Figure A). These results showed that fluorescence enhancement occurred at λEm = 560 nm as the myricetin-to-NPC molar ratio increased. Only a single band centered at λEm = 560 nm (λEx = 280 nm) was observed, and fluorescence intensities directly correlated with the myricetin-to-NPC molar ratio.

Figure 3

Fluorescence spectroscopy-substantiated electrostatic interactions occur between myricetin and NPC corona. (A) Representative fluorescence spectra showing fluorescence enhancement at λEm = 560 nm (using λEx = 280 nm) as the myricetin concentration is increased for a single NPC concentration (0.025 mM). The molar ratios of myricetin to NPC used ranged from 0 to 14. (B) Equilibrium binding of myricetin to NPC measured by the increase in intrinsic fluorescence at λEx = 280 nm, λEm = 560 nm as myricetin was titrated into NPC solutions. The plot shows F/Fmax (y-axis) vs the ratio of myricetin concentration to NPC concentration (x-axis) for NPCs containing varied block 1 Mn (12.8, 18.5, 24.5, and 37.4 kDa) and similar block 2 Mn (∼10 kDa). Data shown as mean ± standard deviation from n = 3–5 independent experiments. **p < 0.01 for differences between NPCs with block 1 Mn of 12.8 and 37.4 kDa using two-way ANOVA with Tukey’s multiple comparisons test. (C) Representative Scatchard plot for NPC loaded with myricetin in a pH 7.2 phosphate buffer, indicating an association constant (negative slope, Ka) of ∼1.7 × 104 M–1. Data shown as mean ± standard deviation from n = 3 independent experiments. (D) Scatter plot of Ka vs number of DMAEMA monomers in diblock copolymers using NPCs with different block 1 Mn values and similar block 2 Mn values, showing a statistically significant increase in Ka values as more DMAEMA is present in NPCs. Data shown as mean ± standard deviation from n = 3 independent experiments. Abbreviations: DMAEMA, dimethylaminoethyl methacrylate; F, fluorescence; Fmax, maximum fluorescence; Mn, number average molecular weight; Myr, myricetin; and NPC, nanoparticle carrier.

A singular peak suggests that only one tautomeric form of myricetin exists at equilibrium in solution and is produced by a ground-state complex of NPC–myricetin, where myricetin is in its keto tautomeric form (Supporting Information Figure S2, structure B).[47,48] The initial deprotonation that leads to the 4′-phenoxide anion produces this keto tautomer via one of two previously described conditions.[30,36] First, in neutral pH conditions, deprotonation of the 4′-OH group occurs and prompts the formation of the 4′-phenoxide anionic state and excitation of the fluorophore to a higher vibrational level (e.g., S1 or S2) associated with the enolic tautomer resonance state (Supporting Information Figure S2, structure B1), where the 3-OH proton transfers to the 4-C=O group through intramolecular hydrogen bonding. However, an internal conversion process, known as excited-state intramolecular proton transfer (ESIPT), rapidly relaxes the molecule to the lowest vibrational level within S1 associated with the keto tautomer prior to fluorescence emission.[49] Therefore, the keto tautomer provides the thermally equilibrated excited state that is viewed as the steady-state fluorescence spectra in Figure A. The keto form is likely favored to maintain the planarity of the myricetin molecule as the enolic form results in a transient double bond connecting the B and C rings, as described above for the band II absorbance change. Although seemingly contradictory, this scenario elegantly explains the hypochromic shift in band II absorbance (Figure C) and the single fluorescence band (Figure A), given that the absorption process occurs first, followed by the internal conversion process and finally fluorescence emission (see the Jablonski diagram in Supporting Information Figure S3).

The keto form coupled with exposure to either protonated tertiary amines or intercalated water molecules within the hydrophilic NPC corona may also inhibit hydrogen bonding between the 4-C=O and 3-OH groups and therefore disrupt ESIPT tautomer formation. Tertiary amine effects are more probable, given that the oxygen molecules within water serve as fluorescence quenchers,[49,50] yet enhancements were observed. Additionally, detergent use has been shown to increase the flavonoid fluorescence signal[22,51] likely because of isolation of drug molecules from aqueous conditions.[40] Thus, it seems plausible that the self-assembly of NPC micelles may physically isolate myricetin from the aqueous conditions, thereby eliminating quenching by oxygen and enhancing the fluorescence signal. The Coulombic effects of electric double layers inherent to colloidal entities, such as NPCs, in aqueous conditions may contribute to this behavior by segregating coronally located myricetin from the solute–solvent interface.[47] Because the latter does not explain the hypochromic shift in band II absorbance (Figure C), the first explanation appears more likely. However, the two are not necessarily mutually exclusive, and some combination of these behaviors may ultimately be responsible for the results observed.

Based on the fluorescence results, equilibrium binding plots (Figures B and Supporting Information S4) and Scatchard plots (Figure C) of solutions containing NPCs with various block 1 and block 2 molecular weights were assessed. An equilibrium binding plot of solutions containing NPCs with varied block 1 Mn (12.8, 18.5, 24.5, and 37.4 kDa) and similar block 2 Mn (∼10 kDa) titrated with myricetin demonstrated fluorescence enhancement at λEm = 560 nm until saturation was achieved at the myricetin-to-NPC molar ratios of 10 and 14 (Figure B). Similar results were also obtained when NPCs with similar block 1 Mn (25.8 kDa) and varied block 2 Mn (7.1, 15.0, and 29.2 kDa) were evaluated (see Supporting Information Figure S4).

Scatchard plots, which can be used to assess reversible binding interactions and determine binding constants, were used to evaluate the fluorescence data, as shown in Figure C (R2 = 0.89, p < 0.05). The NPC–myricetin association constant (Ka = 1.7 × 104 M–1) was calculated from the slope of this plot and corresponded to a standard Gibbs free energy change of −5.6 kcal mol–1. By repeating this process for all of the NPCs tested in Figure B, a scatter plot of Ka versus DMAEMA repeats in each diblock copolymer was obtained (Figure D). This analysis also revealed a statistically significant correlation (p < 0.01) between NPC DMAEMA repeats and NPC–myricetin association constants despite a weak correlation coefficient (R2 = 0.61). This finding is not surprising considering that the DMAEMA tertiary amines provide electrostatic interactions with hydroxyl groups present in myricetin. Scatchard plot analyses performed on multiple NPCs (Table ) yielded Ka and ΔG° results of ∼1–3 × 104 M–1 and −5.4 to −6.0 kcal mol–1, respectively. Again, these values are similar to the electrostatic interactions determined using absorbance spectroscopy mentioned above (Figure ) as well as those found in previous studies.[22,41−43] In particular, others have determined the average Ka and ΔG° values for host–guest interactions between organic hosts and organic molecule guests to be 1 × 103.4 M–1 and −4.6 kcal mol–1, respectively,[43] which align well with the spectroscopy findings presented here. Therefore, the fluorescence analysis corroborated the absorbance-based conclusion that the electrostatic interactions occur between myricetin and NPCs.

Two additional assays were completed to examine how surface charge affects the NPC–myricetin electrostatic interactions. For these studies, NPCs with the highest Ka values were used. As shown in Figure D, NPCs containing between 165 and 250 DMAEMA repeats within the diblock copolymer (e.g., NP25/10 and NP26/15) demonstrated Ka values greater than 2 × 104 M–1. Because no statistical difference in Ka existed between these two NPCs, additional factors, such as corona-to-core molecular weight ratio (i.e., CCR) and size (i.e., diameter), were also considered. Previous work has shown that CCRs do not affect either drug-loading capacity (DLC) or NPC binding to hydroxyapatite, but higher CCRs exhibit greater environmentally responsive drug release.[26] In addition, among NPCs with similar corona molecular weights, NPCs with smaller diameters demonstrated significantly better electrostatic binding to negative charges in hydroxyapatite surfaces,[26] which could be advantageous for some applications, such as bone targeting for fracture healing as well as antibiofilm therapies. Therefore, NP25/10 was selected because its CCR was higher (2.5) and its diameter was smaller (30.6 nm) than that of NP26/15 (1.7 and 43.2 nm, respectively).

The zeta potential values of NP25/10 alone or NP25/10 combined with increasing myricetin concentrations (i.e., 0.7, 1.4, and 2.8 mM) in phosphate-buffered saline (PBS) at pH values above and below the myricetin’s published pKa value range (∼6.3–6.6[20,36,52]) were measured. Figure A shows the results from the testing in PBS at pH 6.2. Altogether, these data demonstrate that the addition of increasing myricetin concentrations did not affect the NPC zeta potential values (e.g., ∼20–24 mV) at pH 6.2. Similarly, neither increasing myricetin concentrations nor pH differences affected NPC size (i.e., diameter), as shown in Supporting Information Figure S5. However, the pH 7.2 test results revealed multiple statistically significant decreases in NPC zeta potential values as higher concentrations of myricetin were used (Figure B). Of particular note, the use of 2.8 mM myricetin effectively neutralized the NPC surface charge. These results highlight that myricetin loading is pH-dependent and largely influenced by the pKa of the myricetin 4′-OH group (pKa = 6.33).[36,37] In sum, pH values below this pKa did not affect the protonation state of myricetin, so no electrostatics-mediated loading could occur; however, pH values above this pKa caused myricetin to become anionic because of the disassociated proton from the 4′-OH group and thus facilitated electrostatic interactions to occur between myricetin and the cationic NPC corona.

Figure 4

Zeta potential and gel electrophoresis analysis confirmed that electrostatic interactions occur between myricetin and NPC corona. (A) Zeta potential values (mean ± standard deviation) for NPC alone and NPC plus increasing concentrations of myricetin (i.e., 0.7, 1.4, and 2.8 mM) using PBS at pH 6.2 show no significant difference. (B) Zeta potential values (mean ± standard deviation) for NPC alone and NPC plus increasing concentrations of myricetin (i.e., 0.7, 1.4, and 2.8 mM) using PBS at pH 7.2 show significant decreases in zeta potential as higher drug concentrations are used. Data shown as mean ± standard deviation from n = 5–9 independent measurements. *p < 0.0001 compared to 0 mM myricetin group, $p < 0.05 compared to 0.7 mM myricetin group, #p < 0.05 compared to 1.4 mM myricetin group, &p < 0.0001 compared to 2.8 mM myricetin group using one-way ANOVA with Tukey’s correction for multiple comparisons. (C) Representative gel electrophoresis image showing increasing myricetin concentrations (≥2.8 mM) hinder cationic NPC migration to the anode (shown using black arrows), while free and excess myricetin migrates to the cathode (shown using white arrows).

To further substantiate that electrostatic interactions occur between myricetin and the NPC corona, a qualitative gel electrophoresis assay (Figure C) was completed and confirmed that ∼2.8 mM myricetin neutralizes the NPC surface charge. Anodic migration of myricetin-loaded cationic NPC at ≥2.8 mM myricetin was hindered, while free and excess myricetin migrated toward the cathode (Figure C). No impact on NPC migration was observed at ≤1.4 mM myricetin. Therefore, the use of a lower concentration (≤1.4 mM) was deemed necessary to ensure that myricetin is loaded within the NPC corona for release characterization or therapeutic use. Altogether, these results validated the spectroscopy conclusions and provided some initial quantifiable data, confirming the extent of the electrostatic interactions between myricetin and NPCs.

Multiple synergistic effects have been observed from combinations of flavonoids with other therapeutic compounds, such as antibacterial and antiviral small-molecule drugs[3,4,7] as well as antitumor, cardioprotective, and antitoxin agents.[53,54] Therefore, after establishing the NPC loading mechanism of myricetin, coloading NPCs with myricetin and a second, potentially synergistic antibiofilm drug with established NPC core loading characteristics (e.g., farnesol)[24−26,55] were evaluated. In particular, the effect of NPC-loaded myricetin on farnesol loading and release was evaluated using NP26/10 (Figure ). Serendipitously, an existing chromatography method for measuring farnesol concentration also yielded a clear elution peak for myricetin (Figure A). Therefore, DLC, the NPC mass percentage due to loaded drug, and drug-loading efficiency (DLE), the percentage of the drug feed successfully loaded into NPCs, for both drugs were measured simultaneously. As shown in Figure B, NPCs combined with 1.2 mM myricetin demonstrated a DLE of ∼27% and a DLC of ∼4%, while the use of 2.4 mM myricetin optimized these parameters (∼22% DLE and ∼6.2% DLC). Although seemingly low, these levels of DLE and DLC results are not uncommon for crystalline small-molecule drugs.[56−58] These results improved myricetin solubility >25-fold compared to myricetin alone and at least 44% greater than known myricetin solubility increases because of using a DDS.[20,21] Moreover, despite the presence of the polar aprotic solvent dimethyl sulfoxide (DMSO), which is used to dissolve myricetin for NPC drug loading and could lead to poor farnesol loading if present in excessive amounts, Figure C shows no impact of myricetin on NPC farnesol loading (i.e., DLC = ∼24%, which is comparable to previous data[25−27]) for myricetin concentrations ≤1.2 mM. However, starting at 2.4 mM myricetin, a statistically significant decrease (p ≤ 0.001) in farnesol DLC, albeit only to ∼22.5% for 2.4 mM myricetin, was observed (Figure C). Based on the DLC and DLE data acquired for both myricetin (Figure B) and farnesol (Figure C), myricetin concentrations ≤1.2 mM did not impact farnesol loading.

Figure 5

NPC–myricetin interaction did not affect farnesol loading and release. (A) Representative HPLC spectra showing elution peaks for myricetin (∼1.5–2.5 min) and farnesol (∼15.9–17.4 min). (B) Myricetin DLC (% DLC) and DLE (% DLE) curves for NP26/10 using multiple myricetin concentrations. Data shown as mean ± standard deviation from n = 3 independent measurements. (C) Farnesol and myricetin % DLC curves for NP26/10 using 4.5 mM farnesol in combination with multiple myricetin concentrations. Data shown as mean ± standard deviation from n = 3 independent measurements. ***p < 0.001 and ****p < 0.0001 vs farnesol % DLC result using 0 mM myricetin from one-way ANOVA with Dunnett’s multiple comparison test. (D) Farnesol and myricetin release from farnesol- and myricetin-coloaded NPCs at pH 7.2 over 48 h. Data shown as mean ± standard deviation from n = 3 independent experiments. (E) Half-time of release (t1/2) and release rate constants (k) for NP25/10 from data shown in (D).

Based on these findings, 0.08 mM NP26/10 was combined with 4.5 mM farnesol and 1 mM myricetin to evaluate farnesol and myricetin release under neutral pH (i.e., pH 7.2) conditions. As shown in Figure D and quantified in Figure E, myricetin released quickly with ∼90% release within the first hour (t1/2 ≤ 1 h) and a release rate constant (k) of 0.69 h–1, while farnesol released more slowly (t1/2 = 6.3 h, k = 0.11 h–1) over 24 h, consistent with previous farnesol release studies.[25−27] This type of differential temporal release may be beneficial for applications where sequential disruption of different therapeutic targets is necessary to provide a desired therapeutic outcome. For example, early myricetin release could inhibit key antimicrobial resistance factors within a biofilm, and delayed farnesol release could enable improved drug access to susceptible bacteria to provide an antibacterial effect. Based on the results of the drug loading and release studies shown in Figure , this type of dual-drug-loading approach was deemed feasible and may significantly enhance clinically relevant treatment regimens across a range of flavonoid-based therapeutic applications.

Conclusions

This study showed that a NPC DDS loads myricetin via electrostatic interactions with the NPC corona and improves myricetin solubility >25-fold compared to myricetin alone. The results obtained confirmed that electrostatic interactions occur between myricetin and NPCs with Ka values of ∼1–3 × 104 M–1 (ΔG° = −5.4 to −6.0 kcal mol–1). These data match with the previously reported flavonoid association constants and standard Gibbs free energy values[22,41−43] and indicate spontaneous supramolecular complex formation between NPCs and myricetin. The electrostatic-mediated corona loading of myricetin at concentrations ≤1.2 mM did not affect farnesol loading or release from hydrophobic NPC cores. This approach may enable development and delivery of novel dual-synergistic drug-loaded NPCs to treat a variety of conditions. Overall, these findings offer key insights into the NPC DDS design that may enhance clinically relevant treatment regimens across a range of flavonoid-based therapeutic applications.

Experimental Section

Materials

All materials were supplied by Sigma-Aldrich unless otherwise specified. DMAEMA and BMA were purified by distillation prior to use. The reversible addition–fragmentation chain-transfer polymerization (RAFT) chain-transfer agent (CTA), 4-cyano-4-[(ethylsulfanylthiocarbonyl)sulfanyl]pentanoic acid (ECT), and PAA were synthesized as described previously.[59−61] The RAFT radical initiator, 2,2-azobisisobutyronitrile (AIBN), was recrystallized from methanol. All PBS used in this study was 1× DPBS unless otherwise specified. All water used was deionized and distilled with a resistivity of at least 18 MΩ unless otherwise specified.

Polymer Synthesis and Characterization

Cationic Corona Block (Block 1) Synthesis

Poly(dimethylaminoethyl methacrylate), or p(DMAEMA), was synthesized via RAFT polymerization using the CTA ECT and the radical initiator AIBN. Distilled DMAEMA was mixed with specific molar ratios of ECT and AIBN, so the [monomer]/[CTA]/[initiator] = 450/5/1, 800/5/1, 1325/5/1, 1500/5/1, or 2450/5/1 in dimethylformamide (DMF) at 40 wt % to obtain the different block 1 molecular weights shown in Table . Nitrogen was used with a Schlenk line to purge the reaction vessel for 45 min prior to submerging the vessel in an oil bath at 60 °C for 6 h. The polymerization reaction was terminated by exposing the vessel contents to atmospheric oxygen. The product was precipitated and washed four times in 80:20 pentane/diethyl ether with centrifugation and dried under vacuum overnight.

Hydrophobic Core Block (Block 2) Synthesis

Poly(dimethylaminoethyl methacrylate)-b-poly(dimethylaminoethyl methacrylate-co-butyl methacrylate-co-propylacrylic acid), or p(DMAEMA)-b-p(DMAEMA-co-BMA-co-PAA), was synthesized using RAFT polymerization with p(DMAEMA) as the macro-CTA, AIBN as the radical initiator, and 15–25 wt % DMAEMA, 50–55 wt % BMA, and 25–30 wt % PAA dissolved in DMF (40 wt % monomers, initiator, and macro-CTA to solvent volume). The target degree of polymerization (DP), or [monomer]/[CTA], was varied according to Table to control block Mn, and [ECT]/[AIBN] = 5 for all polymerizations. After 45 min of purging the reaction vessel with nitrogen, the reaction proceeded at 60 °C for 24 h. The polymerization reaction was terminated by removing the vessel from heat and exposing the vessel contents to atmospheric oxygen. The product was precipitated four times using 80:20 pentane/diethyl ether and centrifugation before being dried overnight under vacuum.

Polymer Purification and Storage

The dried diblock copolymer was removed from vacuum and dissolved in ∼5 mL of 100% ethanol in a 50 mL conical tube. After the raw polymer was completely dissolved, ∼25 mL of PBS was added to the tube. The combined solution was dialyzed using a 6–8 kDa dialysis membrane tubing (Spectrum Laboratories) for at least 4 days with ∼2–3 water changes each day. The dialyzed solution was frozen at −80 °C and lyophilized for at least 4 days using a Labconco FreeZone 2.5 freeze dryer. The lyophilized polymer was stored in closed containers at room temperature prior to use.

Polymer Characterization

First block and diblock copolymer molecular weights and polydispersities (PDI, Mw/Mn) were determined using gel permeation chromatography (Shimadzu Technologies) with a miniDAWN TREOS multiangle light scattering detector (Wyatt Technology) in line with an Optilab T-rEX refractive index detector (Wyatt Technology). High-performance liquid chromatography (HPLC) grade DMF + 0.05 mM LiCl (0.2 μm filtered) was used as the mobile phase at a flow rate of 0.35 mL min–1 through a TSKgel SuperH-H guard column and TSKgel SuperHM-N column (Tosoh Biosciences) at 60 °C. ASTRA 6.1 light scattering software (Wyatt Technology) and a previously reported dn/dc value of 0.06[62,63] were used to calculate the molecular weight. The diblock copolymer percent composition was characterized using proton nuclear magnetic resonance (1H NMR) spectroscopy (Bruker AVANCE 400), as described previously.[59] The representative 1H NMR spectra for p(DMAEMA) block 1 and p(DMAEMA)-b-p(DMAEMA-co-BMA-co-PAA) diblock copolymers are shown in Supporting Information Figure S6.

1H NMR of p(DMAEMA) and block 2 DMAEMA: δ (ppm, from CHCl3 signal at 7.26 ppm) 0.8–1.1 (3H, br m, CH3–C), 1.8–1.9 (2H, br s, CH2–C), 2.33 (3H, br s, N–CH3), 2.6 (2H, br s, N–CH2), 4.07 (2H, br s, CH2–O–C=O).

1H NMR of block 2 BMA: δ (ppm, from CHCl3 signal at 7.26 ppm) 0.9–1.02 (3H, br s, CH3–C), 1.03 (2H, br s, CH–CH3), 1.5 (2H, br s, CH2–CH–CH2), 1.8–1.9 (2H, br s, CH2–C), 3.93 (2H, br s, CH2–O–CHPLCO).

1H NMR of block 2 PAA: δ (ppm, from CHCl3 signal at 7.26 ppm) 0.85 (3H, br s, CH–CH2), 1.4 (2H, br s, CH2–CH–C), 1.5 (2H, br s, CH2–C), 1.6 (2H, br s, CH–CH3).

NPC Self-Assembly Characterization

NPC size and zeta potential were measured using a Zetasizer Nano ZS (Malvern Panalytical). NPC size measurements were performed via dynamic light scattering (DLS) analysis using lyophilized polymer concentrations of ∼0.2–0.3 mg mL–1 fully dispersed in PBS and passed through a 0.45 μm poly(vinylidene difluoride) (PVDF) aqueous syringe filter into disposable cuvettes. The zeta potential was determined using polymer concentrations of ∼0.2–0.5 mg mL–1 in 90:10 water/PBS solutions and filtered using 0.45 μm PVDF aqueous syringe filters into disposable p1070 capillary cells to ensure that sample conductivity values permitted analysis via General Purpose analysis in the Malvern Zetasizer software.

Myricetin Sample Solution Preparation

Myricetin stock was obtained from TCI America (CAS number: 529-44-2 and product number: M2131). Myricetin stock was prepared in bulk at 25 mM in DMSO. The stock was first diluted using DMSO and then diluted further using PBS or phosphate buffer (pH 6.2 or 7.2) to achieve the final sample concentration for each study. Every sample contained some variation of nanoparticle concentration, drug concentration, and/or buffer for a total volume of either 0.5 or 1 mL. All samples were sonicated using a VWR Ultrasonic Cleaner (model 50HT Bath Sonicator) for 10–15 min.

Absorbance Spectroscopy

For absorbance testing, 100 μL of each sample solution was added to a black, quartz cuvette. Absorbance data were collected using an Infinite M200 PRO microplate reader (Tecan, Switzerland) to measure the absorbance of each solution in a cuvette from a wavelength of 240–500 nm in increments of 5 nm. After testing each sample, the cuvette was washed with methanol and water and then dried with a Kim wipe. Once the absorbance data were obtained for myricetin alone samples (Ao) and myricetin plus NPC samples (A), a double-reciprocal plot of the absorbance data was developed using 1/(A – Ao) on the y-axis and 1/(molar myricetin concentration) on the x-axis. An estimated association constant (Ka) was calculated for the plot by dividing the y-intercept value by the slope value as described elsewhere.[40]

Fluorescence Spectroscopy

For fluorescence testing, 500 μL of each sample solution was filtered using Amicon Ultra-0.5 Centrifugal Filter Devices (Amicon Ultra 3K device) in a centrifuge for 15 min at 14,000g. The filtrate (∼400 μL) was collected and set aside. The concentrate (∼100 μL) was obtained by inverting the filter into a separate tube and centrifuging for 2 min at 1000g. The two sample groups were then diluted back to their original volume by adding ∼100 and ∼400 μL of phosphate buffer at pH 7.2, respectively. Then, 100 μL of each sample solution was transferred onto a 96 black-well plate for fluorescence testing. Fluorescence data were collected using an Infinite M200 PRO microplate reader (Tecan, Switzerland) on a 96 black-well plate at an excitation of 280 nm from an emission wavelength of 500–620 nm with a manual gain of 175. Equilibrium binding plots (F/Fmax vs [Myr]/[NPC]) were prepared using fluorescence values (λEx = 280 nm, λEm = 560 nm) from NPC solutions (0.025 mM) titrated with increasing myricetin concentrations, and Scatchard plots were created by plotting (F/Fmax)/([Myr]) versus (F/Fmax). For these plots, F represents the fluorescence for each sample, Fmax represents the saturated fluorescence value, [Myr] represents the molar concentration of myricetin used, and [NPC] represents the molar NPC concentration used. The association constants (Ka) obtained as the negative slope of the Scatchard plots were plotted against the number of DMAEMAs in the diblock copolymer (calculated from Table ).

Zeta Potential Evaluation

For zeta potential testing, nanoparticle (NPC) samples were prepared in phosphate buffer (pH 6.2 or pH 7.2) at 2.7 mg mL–1. Varying concentrations of myricetin were added to create each sample solution. The samples were diluted 5-fold or 10-fold from 2.7 to 0.54 or 0.27 mg mL–1, respectively, using water to prevent damaging the zetasizer cells during measurement because of excess salt concentration. The zeta potential was measured using a Zetasizer Nano ZS (Malvern Panalytical).

Gel Electrophoresis Evaluation

A 1% agarose gel was prepared in 1× tris-acetate-ethylenediaminetetraacetic acid buffer adjusted to pH 7.4. The NPC samples were prepared by massing lyophilized powder for NP25/10 in 20 mL scintillation vials and adding PBS to make a concentration of 2.7 mg mL–1 (0.08 mM). Appropriate volumes of myricetin dissolved in DMSO at 25 mM were added to the respective myricetin-containing samples, and all the scintillation vial solutions were sonicated for 15 min using a VWR Ultrasonic Cleaner (model 50HT Bath Sonicator). Multiple 50 μL samples from these solutions were transferred to Eppendorf tubes, and 4 μL of BlueJuice Gel Loading Buffer was added. After briefly vortexing each solution, 25 μL of each sample was loaded into the 1% agarose gel. Electrophoresis occurred for 50 min. The gel was then fixed and stained with the Bio-Rad Silver Stain Plus kit (Catalog #161-0449) according to the manufacturer’s instructions.

Characterization of NPC Drug Loading

NPCs were loaded with farnesol as described previously.[25−27] Briefly, farnesol/PBS emulsions at 1.0 mg mL–1 were prepared using ∼30 s of tip sonication (Fisher Scientific Sonic Dismembrator model 100 at 4 W power setting) and were then immediately added to preweighed lyophilized diblock copolymers to achieve desired polymer NPC concentrations (e.g., 2.7 mg mL–1) in 20 mL glass scintillation vials. In addition, calculated volumes of myricetin dissolved in DMSO (from either bulk or experimental stock solutions) were added to NPC and/or farnesol solutions where necessary. All of these solutions were bath-sonicated using a VWR Ultrasonic Cleaner (model 50HT Bath Sonicator) for 15 min to enable drug loading. NPCs loaded with farnesol and/or myricetin were concentrated using 3 kDa centrifugal filters (Amicon Ultra 0.5 mL, Millipore, USA). The concentrate was recovered, and PBS was added to return the sample volume to 0.5 mL. This solution was diluted with an equal amount of ethanol and passed through a 0.45 μm PVDF aqueous syringe filter before the amounts of farnesol and/or myricetin loaded were measured via HPLC using a 20 μL injection volume. HPLC analysis was conducted using a gradient mobile phase consisting of HPLC-grade methanol and water (10–90% MeOH) and a Kromasil C18 column (50 mm × 4.6 mm, 5 μm particle size, 100 Å pore size from Supelco, Bellefonte, PA, USA) with a flow rate of 0.5 mL min–1 over 20 min. The column effluent was monitored with a variable wavelength UV–vis detector at 210 nm (Shimadzu Technologies). The relative area-under-the-curve values for the myricetin peaks that occur between ∼1.5 and 2.5 min and the farnesol peaks that occur between ∼15 and 17 min were compared to a standard curve of known drug concentrations to determine the concentration of each drug loaded. The amount of drug loaded was used to calculate the DLE (100% × (wtloaded)/wt0) and DLC (100% × (wtloaded/wtNPC + wtloaded)), where wtloaded is the amount of the loaded drug, wt0 is the initial amount of the drug used, and wtNPC is the amount of the polymer used.

Characterization of NPC Drug Release

NPC drug release was characterized using dialysis, as described previously.[25,26,64] After drug loading, NPC/PBS solution was transferred into a prewetted 6–8 kDa dialysis membrane tubing (Spectrum Laboratories), which was securely clipped at both ends. The filled dialysis membranes were placed in phosphate buffer at pH 4.5 or 7.2 and dialyzed at 37 °C with buffer changes occurring after each sample collection time point to ensure sink conditions. For farnesol and myricetin release studies, samples were collected at 0, 2, 4, 6, 12, 24, and 48 h. The samples were immediately frozen and stored at −80 °C until drug concentration of each sample was assessed via HPLC as described above for drug-loading characterization. First-order release fits of the data were performed using GraphPad Prism software (v.6.07), and release rate constants (kobs) and release half-times (t1/2) were calculated according to the first-order release equation (% release = 100 × (1 – e–)), where % release is the percent of drug release at time t and k is the observed kinetic constant of drug release (kobs). The following relationship between kobs and t1/2 was used for value conversions: t1/2 = ln(2)/kobs. The first derivatives of the release fit equations were calculated by GraphPad Prism to yield graphical depictions of the change in release rates over time.

Drug Solubility Calculations

Fold changes in drug solubility were reported either exactly as stated in referenced publications or calculated by comparing the drug concentrations solubilized by a drug delivery carrier to the highest known soluble concentration of free drug alone. In addition, the differences between myricetin concentrations loaded by NPCs and other DDSs were calculated and directly compared to other systems’ reported concentrations to yield percent differences.

Statistical Analysis

GraphPad Prism 6 software (v.6.07) was used to perform statistical analyses, such as one-way ANOVA with Tukey’s or Dunnett’s correction for multiple comparisons or two-way ANOVA with Sidak’s multiple comparisons at p < 0.05 as indicated in the figure captions.