Article Cite This: J. Phys. Chem. B 2018, 122, 7426−7435
pubs.acs.org/JPCB
Association and Internal Morphology of Self-Assembled HPPhOx/ BSA Hybrid Nanoparticles in Aqueous Solutions Aristeidis Papagiannopoulos,*,† Eleni Vlassi,† Stergios Pispas,† and Judith Elizabeth Houston‡ †
Downloaded via KAOHSIUNG MEDICAL UNIV on August 18, 2018 at 07:00:40 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
Theoretical and Physical Chemistry Institute, National Hellenic Research Foundation, 48 Vassileos Constantinou Avenue, 11635 Athens, Greece ‡ Jülich Centre for Neutron Science (JCNS) at Heinz Maier Leibnitz-Zentrum (MLZ), Forschungszentrum Jülich GmbH, Lichtenbergstrasse 1, 85748 Garching, Germany ABSTRACT: We investigate the formation of hybrid polyelectrolyte/protein nanoparticles by associations between aggregates of partially hydrolyzed poly(2-phenyl-2-oxazoline) (HPPhOx) and bovine serum albumin (BSA) in aqueous solutions. Light scattering experiments show that at conditions of low salt, BSA creates interaggregate bridges and increases the size of the HPPhOx nanoparticles. At high salt contents, breaking of aggregates leads to well-defined nanoparticles. The interior of the formed nanoparticles is probed by small-angle neutron scattering. At low salt, diffuse arrangements are observed, whereas at high salt concentration, scattering is dominated by well-defined hydrophobic domains enhanced by the incorporation of BSA. This system shows that the combination of hydrophobic and electrostatic interactions in random-amphiphilic-polyelectrolyte/protein complexes can be used to determine the properties of self-assembled hybrid multifunctional nanoparticles.
■
INTRODUCTION Functional nanoparticles produced by self-assembly of proteins and charged polymers offer unique possibilities in the fields of biotechnology1,2 and nanodelivery of proteins and drugs.3,4 Polysaccharides, which are mostly hydrophilic polyelectrolytes, have been used extensively in protein containing nanoassemblies because they are biocompatible and biodegradable.5 In particular, bovine serum albumin (BSA) has been used in polysaccharide-coated nanoparticles6 and hierarchical polysaccharide-based scaffolds7 for biomedical applications. Synthetic polyelectrolytes have well-defined properties and versatile synthesis routes have been devised, but at the same time they raise questions regarding toxicity and biodegradability. At the same time, they offer the ability to tune their hydrophilic (charged) content in relation to their hydrophobic content. Polyoxazolines and polyoxazoline-containing polyelectrolytes are a promising class of polymers8−10 for biomedical11 and pharmaceutical applications12 with the potential of satisfactorily low biotoxicity.13−15 Polyoxazoline-containing polyelectrolytes16,17 may be synthesized in ways providing both hydrophobic and charged groups within their polymer chain and hence they can interact with proteins, DNA, and surfactants. In thermoresponsive random copolymers of n-propyl-2-oxazoline and ethylenimine (PPrOx−PEI), the interaction of ethylenimine (EI) units with BSA was confirmed by increasing the EI content of the polymers. More importantly, the authors illustrated that above the transition temperature, globular protein-mimicking mesoglobules are formed with their positive © 2018 American Chemical Society
charges exposed in their exterior, leading to more effective interactions with DNA.18 In partially hydrolyzed thermoresponsive polyoxazoline (POx−PEI), polyplexes were formed by mixing with DNA, resulting in well-defined nanoparticles that could also be coated by a cross-linked poly(N-isopropylacrylamide) shell.19 The polyplexes had significantly lower cytotoxicity in comparison to the standard PEI vectors. Complexation between proteins and polyelectrolytes is a subject under intense research in the last two decades,20,21 aiming at basic understanding of interactions and structure formation, as well as in view of their potential applications in drug, protein/peptide, and gene delivery. Different polyelectrolyte systems, e.g., spherical polyelectrolyte brushes,22 selfassembled core−shell micelles,23,24 and linear polyelectrolytes,25,26 lead to a vast range of possibilities for nanostructure formation. These systems are often studied by scattering techniques, like static and dynamic light scattering (DLS),27,28 and by small-angle neutron and X-ray scattering.25,29 Light scattering (LS) methods elucidate the overall size and shape of the formed nanoparticles, and they are very sensitive to aggregation and self-association. On the other hand, smallangle scattering methods provide more details on length scales that are relevant to the internal morphology of aggregates (1− 100 nm). Received: May 8, 2018 Revised: June 19, 2018 Published: June 27, 2018 7426
DOI: 10.1021/acs.jpcb.8b04364 J. Phys. Chem. B 2018, 122, 7426−7435
Article
The Journal of Physical Chemistry B
In dynamic light scattering (DLS), the field autocorrelation functions are obtained at a broad angular range and are converted to the field autocorrelation functions by the Siegert relation β|g1(τ)|2 = g2(τ) − 1 where β is a normalization factor.32 We used CONTIN analysis to Laplace-transform the field autocorrelation functions into time-distribution functions f(τ). The relaxation rate Γ was obtained by the characteristic time τmax of the maximum of the distribution function at every angle by Γ = 1/τmax. The diffusion coefficient D may be calculated by the slope of the Γ versus q2 plots. Eventually, the hydrodynamic radius is taken by the Stokes−Einstein equation Rh = kT/6πηD. The size polydispersity index (PDI) is taken from cumulant analysis at θ = 90°. Small-Angle Neutron Scattering. Experiments were performed on KWS-2, the high-intensity/wide-q small-angle neutron scattering diffractometer, at the Jülich Center for Neutron Science (reactor FRM II). In small-angle neutron scattering (SANS), the scattering vector (q) range was 0.002− 0.1 Å−1 and was covered by three separate detection configurations: 2, 8, and 20 m detection length, with neutron wavelength λ = 4.5 Å. Raw data was treated by standard corrections and reduction, and the collected isotropic twodimensional raw data was transformed into one-dimensional scattered intensity I(q). All numerically calculated intensities mentioned in the Results and Discussion section are convoluted33 with a Gaussian function so that instrumental resolution is taken into account.34,35 Small-Angle Neutron Scattering Data Analysis. A three-level hierarchical Beaucage model36,37 was used to model the SANS data described in eqs 1−3. Each level a and b consists of a Guinier regime that defines its forward scattering Gi and radius of gyration Rg,i and a power-law regime with a characteristic exponent di. We assign level a as the large and level b as the small hierarchical structure, i.e., Rg,a > Rg,b. Alternatively, level b may be denoted as “internal domains” and level a as “aggregates”. The error function in eq 2 constrains the powerlaw at q > R−1 g,i . At higher q, terms a and b are limited at high q by a
In our previous work,30 we have synthesized and characterized partially hydrolyzed poly(2-phenyl-2-oxazoline) (HPPhOx), leading to amphiphilic charged random copolymers. These macromolecules contain hydrophobic (PhOx) and hydrophilic ethylenimine (EI) units in a random sequence. We have previously observed that the HPPhOx copolymers form aggregates ∼100 nm in size with positive surface charge and illustrated that they interact with proteins of fetal bovine serum. In this study, we explore the associations of these aggregates with BSA at neutral pH where the two components are oppositely charged. We perform light scattering and small-angle neutron scattering (SANS) to resolve the overall structure and internal morphology of the HPPhOx/BSA nanoparticles at several solution conditions, leading to intermolecular self-assembly. The system under study provides hybrid multifunctional nanoparticles with the ability for complex hydrophobic and electrostatic interactions at the molecular level and has potential for hydrophobic drug loading, protein separation, and tissue engineering applications.
■
MATERIALS AND METHODS Materials and Sample Preparation. HPPhOx synthesis and solution properties have been described in detail else-
Scheme 1. HPPhOx Random Copolymer (poly(EI-coPhOx))
where.30 HPPhOx with Mn = 2.03 × 103 g mol−1, Mw/Mn = 1.13 (PPhOx precursor), and 67% molar content in PhOx was utilized for the present studies. This random copolymer comes from the partial hydrolysis of PPhOx and contains hydrophobic PhOx and cationic EI units (Scheme 1). It was directly dissolved in D2O at the desired concentration and left overnight to equilibrate at 4 °C. BSA was supplied by Sigma-Aldrich and used without further treatment. BSA was dissolved in D2O and left overnight to equilibrate at 4 °C before mixing with any polymer solution. HPPhOx/BSA mixtures were prepared by mixing volumes of equilibrated (overnight at 4 °C) separate solutions. In salt-containing solutions, appropriate volumes of 1 M NaCl were added. Light Scattering. Light scattering (LS) was performed on an ALV/CGS-3 compact goniometer system (ALVGmbH, Germany) with an ALV-5000/EPP multi tau digital correlator and a He−Ne laser (λ = 632.8 nm). In static light scattering (SLS), the Rayleigh ratio31 was extracted and treated by the
(
I(q) = Ia(q) + Ib(q) + Ic(q)
NAλ 4
(1)
where
i 1 y i 1 2 y z· Ii(q) = Gi expjjj− q2R g,2 i zzz + Bi expjjj− q2R cut, z iz k 3 { k 3 { [erf(qR g, i/ 6 )]3di ·q−di with i = a, b
(2) −d GidiRg,i i[6d2i (2
The prefactor Bi = + di)−1(2 + 2di)−1]di/2Γ(di/2) bridges the two regimes smoothly.38
R(q)
4π 2n02
)
length scales lower than the gyration radius of the next smaller level, the correlations within that smaller level dominate.
Guinier approximation Kc = M exp( −1/3q2R g2) so that the molar mass M and radius of gyration Rg are obtained. The 4π n θ scattering wave vector is q = λ 0 sin 2 , with θ the scattering angle and n0 the solvent’s refractive index, whereas c is the particle mass concentration. K is given by K =
1
2 cutoff function exp − 3 q2R cut, i with Rcut,i = Rg,i+1 because at
Ic(q) = Bc ·[erf(qR g, c / 6 )]3dc ·q−dc
(3)
The third hierarchical level is described by eq 3. It is a power-law function that contributes at q > R−1 g,c , and it is normally used for correlations within aggregates, e.g., scattering by individual chains within micelles.24 This kind of scattering is also predicted by analytical core−shell models where the explicit conformations of macromolecular chains within micellar coronas are taken into account39 or modeled by empirical formulas.29
(∂n/∂c)2 ,
where ∂n/∂c is the refractive index increment of the scattering particles in the solvent ([∂n/∂c]HPPhOx = 0.22 m g−1, [∂n/∂c]BSA = 0.18 m g−1). 7427
DOI: 10.1021/acs.jpcb.8b04364 J. Phys. Chem. B 2018, 122, 7426−7435
Article
The Journal of Physical Chemistry B yz c·Ni·M p ijj 1 ·jjj ·(ρp − ρD O )zzzz Gi = 2 z NA j d p k {
2
(4)
Forward scattering of a structural level, i.e., Ii (q = 0) = Gi is connected to the number of individual chains (Ni) inside the specific structure40 using eq 4. Mp is the molecular weight of the chains, dp (≈1.1 g cm−3) is the dry mass density, and c is the mass solution concentration of the structures assumed here as the nominal solution concentration, i.e., all polymer chains in the solution are incorporated in the hierarchical structures. The neutron scattering length densities of the polymer and D2O we used are ρp = 1.5 × 10−6 and 6.4 × 10−6 Å−2, respectively. Electrophoretic Light Scattering. The experiments were performed on a Zetasizer Nano-ZS (Malvern Instruments Ltd.). ζ potential was calculated by the Henry equation in the Smoluchowski approximation. Averages of 10−20 measurements at 173° angle are reported. All experiments were performed at room temperature. Circular Dichroism (CD). A Jasco J-815 CD spectrophotometer with a Peltier model PTC-423S/15 thermostabilizing system was used for circular dichroism (CD) measurements. The spectrums were accumulated by averaging six successive runs with 0.5 nm wavelength step at 100 nm min−1 rate, response time 1 s, and bandwidth 1 nm. Fourier Transform Infrared (FTIR) Spectroscopy. Fourier Transform Infrared (FTIR) spectra were recorded on a Bruker Equinox 55 Fourier Transform Instrument, equipped with an attenuated total reflectance (ATR) diamond accessory, from SENS-IR, and a press. The dried solutions were placed at the center of the sample holding device, and 64 scans were performed in the range 500−5000 cm−1, at a resolution of 2 cm−1. Scheme 2. (a) Hydrophobic Interactions in HPPhOx SelfAssemblies; (b) Hydrophobic (i), Electrostatic, (ii) and Bridging (iii) Interactions in HPPhOx/BSA Complexesa
Figure 1. Distribution of Rh extracted by CONTIN analysis at θ = 90° for aqueous solutions of HPPhOx (0.75 mg mL−1) mixed with BSA at 0.1 (black), 0.2 (red), and 0.4 (blue) mg mL−1 protein concentration and for 0 M (a), 0.03 M (b), and 0.15 M (c) NaCl concentration.
■
RESULTS AND DISCUSSION The size and shape of the HPPhOx/BSA complexes was addressed by DLS and SLS. Size-distribution functions at the tested solution conditions are shown in Figure 1. At no added and intermediate amounts of salt, the distributions are broad. In the case of no added NaCl at the highest amount of BSA, there is a clear contribution of a small size population at 4.2 nm, which signifies the presence of free BSA in solution. The hydrodynamic radius of BSA reported in the literature41−43 is 3.5−3.7 nm. The increased size observed in our case can be related to some adsorption of free polymer on BSA monomers. At 0.03 and 0.15 M NaCl, the signature of BSA globules at the maximum BSA content is not detected. This is because the molar mass of the
a
Gray circles denote the region of interaction. EI units, PhOx units, and BSA are represented by green circles, black circles, and red ellipsoids, respectively. 7428
DOI: 10.1021/acs.jpcb.8b04364 J. Phys. Chem. B 2018, 122, 7426−7435
Article
The Journal of Physical Chemistry B
Figure 3. SLS data from aqueous solutions of HPPhOx (0.75 mg mL−1) mixed with BSA at 0.1 (black), 0.2 (red), and 0.4 (blue) mg mL−1 protein concentrations and for 0 M (a), 0.03 M (b), and 0.15 M (c) NaCl concentration. Lines are fits with the Guinier approximation.
Figure 2. Relaxation rate from aqueous solutions of HPPhOx (0.75 mg mL−1) mixed with BSA at 0.1 (black), 0.2 (red), and 0.4 (blue) mg mL−1 for 0 M (a), 0.03 M (b), and 0.15 M (c) NaCl. Lines are polynomial fits, as explained in the main text.
Table 1. Results from SLS and DLS from HPPhOx/BSA Complexes with HPPhOx at 0.75 mg mL−1 [NaCl] (M) 0
0.03
0.15
cBSA (mg mL−1)
M (×105 g mol−1)
Rg (nm)
Rh (nm)
ρ = Rg/Rh
PDI (θ = 90°)
ζ (mV)
0.1 0.2 0.4 0.1 0.2 0.4 0.1 0.2 0.4
2.49 2.08 1.50 1.57 2.15 39.1 2.49 452 47.4
143 146 75.5 45.9 53.1 220 75.9 78.4 64.1
111 107 51.7 33.2 36.4 60.3 76.3 88.8 55.6
1.29 1.36 1.46 1.38 1.46 3.65 0.99 0.88 1.15
0.54 0.44 0.54 0.48 0.50 0.52 0.41 0.022 0.27
18.8 ± 4 21.5 ± 2 29.1 ± 2 20.1 ± 1 27.8 ± 2 20.8 ± 2 17.9 ± 2 13.4 ± 3 16.9 ± 2
7429
DOI: 10.1021/acs.jpcb.8b04364 J. Phys. Chem. B 2018, 122, 7426−7435
Article
The Journal of Physical Chemistry B
polymer.26 In Scheme 2, a representation of the main interactions considered in this work is presented. The relaxation rate from HPPhOx/BSA complexes is shown in Figure 2. The diffusion coefficient was obtained by fitting a quadratic formula Γ(q) = D · q2 + c · (q2)2 because the trends were not linear. The cause of the curved Γ(q) is most probably the polydispersity of the complexes. Indeed, at high salt content where the distribution functions appear narrow (Figure 1), the corresponding Γ(q) profiles are nearly linear (Figure 2c). The SLS data are fitted by the Guinier model (Figure 3). When Rg is relatively small, they appear linear. When Rg is high, a low-q upturn appears because the internal morphology of complexes is probed and because of polydispersity. In those cases, the low-q linear part of the plots is fitted (Figure 3a,b). In Table 1, the results from SLS, DLS, and ELS are presented. The apparent molar mass, radius of gyration, and hydrodynamic radius of the HPPhOx aggregates of this work in similar conditions (in H2O) have been reported previously.30 The reported values were M = 0.32 × 105 g mol−1, Rg = 41 nm, and Rh = 24 nm at 0 M (at HPPhOX 1 mg mL−1), M = 35 × 105 g mol−1, Rg = 229 nm, and Rh = 219 nm at 0.03 M, and M = 4.5 × 105 g mol−1, Rg = 84 nm, and Rh = 89 nm at 0.15 M. In the absence of salt, the complexes have significantly higher mass and size than those of the BSA-free copolymer aggregates. Protein globules cause secondary aggregation to the initial bare copolymer aggregates because they create interaggregate bridges.23 At the highest amount of BSA, the smallest size and molecular mass of complexes is obtained with the maximum ζ potential. This shows that bridging is not so effective at high protein content. The increased number of protein globules in solution leads to small-sized complexes with high apparent surface charge density. The enhanced presence of BSA introduces an increased number of hydrophobic sites that couple to the hydrophobic PhOx monomers. This apparently causes conformational rearrangements in the amphiphilic copolymer chains. A more pronounced exposure of charged EI units occurs in the outer surface of the nanoparticles, increasing their surface charge. These hybrid nanoparticles are smaller than 100 nm in radius and can be ideal for drug delivery applications. At intermediate salt concentration, nanoparticles are obtained with size in the range of 40−50 nm and molecular weights similar to those of 0 M NaCl. The exception is that, at 0.4 mg mL−1, BSA Rg and M appear significantly larger and similar to those of the BSA-free aggregates. This originates from the strong low-q upturn and the related formation of large complexes. The smaller complexes at 0.1 and 0.2 mg mL−1 BSA are formed by a
Figure 4. (a) SANS profiles from HPPhOx 3.75 mg mL−1 at 0 M (black), 0.03 M (red), and 0.15 M (blue) NaCl in D2O. Lines are fit to the experimental data by eqs 1−3. (b) The separate contributions Ia (q), Ib (q), and Ic (q) of eq 1 are illustrated (dashed lines) for the case of HPPhOx 3.75 mg mL−1 at 0 M.
complexes is much higher in these two cases (as will be shown in the following), hence dominating the scattered intensity. At 0.15 M NaCl, the distributions are significantly narrower than in the other two salinities. This is a sign that at conditions where electrostatic interactions are highly screened, better-defined nanoparticles are formed. In complexes between BSA and quaternized poly(chloromethyl styrene) copolymers, addition of the protein (at no added salt) resulted in an increase in molar mass and not in size, indicating the predominance of intraaggregate hydrophobic interactions between protein and
Table 2. Extracted Parameters from SANS on HPPhOx Solutions in D2O [NaCl] (M)
0
0.03
cHPPhOx (mg mL−1)
1.875
3.75
7.5
Ga (cm−1) Rg,a (nm) da Gb (cm−1) Rg,b (nm) db Bc (×10−4 cm−1 Å−dc) Rg,c (nm) dc Na Nb
11.0 ± 0.7 62 ± 5 3.04 ± 0.08 3.3 ± 0.2 29 ± 2 2.15 ± 0.05 0.76 ± 0.04 3.3 ± 0.2 1.69 ± 0.04 890 ± 60 270 ± 20
93 ± 5 69 ± 6 3.36 ± 0.08 9.6 ± 0.5 23 ± 2 2.71 ± 0.05 2.1 ± 0.1 2.6 ± 0.2 1.54 ± 0.03 3800 ± 200 390 ± 20
99 ± 5 50. ± 4 2.38 ± 0.07 18.9 ± 0.8 14 ± 2 3.44 ± 0.07 5.6 ± 0.3 3.3 ± 0.2 1.65 ± 0.03 2000 ± 100 380 ± 20 7430
0.15 3.75
10.7 ± 0.6 51 ± 4 3.09 ± 0.08 3.8 ± 0.2 9±1 4.2 ± 0.07 2.0 ± 0.2 4.4 ± 0.3 1.77 ± 0.02 430 ± 30 150 ± 10
235 ± 10 72 ± 6 3.21 ± 0.08 6.8 ± 0.5 10 ± 2 3.44 ± 0.06 1.5 ± 0.1 4.1 ± 0.3 1.76 ± 0.03 9500 ± 400 280 ± 20 DOI: 10.1021/acs.jpcb.8b04364 J. Phys. Chem. B 2018, 122, 7426−7435
Article
The Journal of Physical Chemistry B
Figure 5. SANS profiles from HPPhOx 1.875 (black), 3.75 (red), and 7.5 (blue) mg mL−1 in D2O with no added salt. Lines are fit to the experimental data by eqs 1−3.
mechanism where the presence of BSA most probably causes breaking of the initial large copolymer aggregates. At 0.15 M NaCl, high molar masses are found while the sizes are below 100 nm. The shape factors ρ only at high salt and moderate protein content approach the ones for homogeneous spherical particles44 (∼0.775) contrary to those at low and no added salt where more open/loose microgel-type morphologies are indicated.45 In particular, at 0.15 M and 0.4 mg mL−1, BSA high molar mass hybrid nanoparticles are formed with spherical shape. In general, ζ potential measurements show that BSA’s negative charges do not decrease the positive surface charge of the complexes. HPPhOx contributes one charge per EI monomer, whereas BSA (at pH 7) contributes about nine net negative charges per globule.46 At the concentrations studied, the BSA/HPPhOx charge ratios are not higher than 1%. Hence, the surface charge of the complexes is dominated by the presence of HPPhOx chains. Similarly, in polyplexes between DNA and random copolymers of n-propyl-2-oxazoline and ethylenimine (PPrOx−PEI), positive surface charge is observed at high amine-to-phosphate groups (N/P) ratios.18 It has to be mentioned that in PPrOx−PEI thermoresponsive copolymers, the formation of globular aggregates with significant positive ζ potential occurred above the transition temperature, i.e., at 37 °C.18 In HPPhOx, formation of aggregates with positive surface charge occurs at room temperature.30 This is possibly caused by the comparably higher content of EI units in HPPhOx and the hydrophobicity of PhOx, which in our case, although it does not induce thermoresponsive behavior,30 it can effectively drive selfassociation at room temperature in contrast with PrOx.
Figure 6. SANS profiles from HPPhOx (3.75 mg mL−1) mixed with BSA at 0 (black), 0.469 (red), 0.938 (blue), and 1.875 (green) mg mL−1 for 0 M (a), 0.03 M (b), and 0.15 M (c) NaCl in D2O. Lines are fits to the experimental data.
Table 3. Extracted Parameters from SANS on HPPhOx (3.75 mg mL−1)/BSA Complexes [NaCl] (M) −1
0
0.03
0.15
cBSA (mg mL )
0.469
0.938
1.875
0.469
0.938
1.875
0.469
0.938
1.875
Ga (cm−1) Rg,a (nm) da Gb (cm−1) Rg,b (nm) db Bc (×10−4 cm−1 Å−dc) Rg,c (nm) dc
68 ± 4 51 ± 4 1.93 ± 0.05 6.9 ± 0.4 22 ± 2 2.49 ± 0.05 2.0 ± 0.1 2.5 ± 0.3 1.57 ± 0.03
55 ± 3 47 ± 4 2.55 ± 0.05 7.3 ± 0.4 21 ± 2 2.56 ± 0.05 2.5 ± 0.1 3.2 ± 0.3 1.50 ± 0.03
58 ± 3 46 ± 4 2.83 ± 0.04 7.2 ± 0.4 20.0 ± 2 2.54 ± 0.05 2.2 ± 0.1 3.6 ± 0.2 1.66 ± 0.04
25 ± 2 62 ± 5 1.70 ± 0.05 4.3 ± 0.2 12 ± 1 2.83 ± 0.05 2.9 ± 0.1 3.3 ± 0.3 1.68 ± 0.03
50 ± 83 ± 7 2.03 ± 0.06 6.7 ± 0.3 16 ± 2 2.81 ± 0.06 2.8 ± 0.2 2.0 ± 0.2 1.79 ± 0.04
60 ± 3 75 ± 7 1.89 ± 0.04 6.6 ± 0.3 17 ± 2 2.48 ± 0.05 2.9 ± 0.2 1.9 ± 0.2 1.61 ± 0.04
10 ± 1 53 ± 6 3.13 ± 0.08 3.9 ± 0.2 8±1 4.20 ± 0.08 1.6 ± 0.2 5.5 ± 0.4 1.76 ± 0.05
10 ± 1 51 ± 5 3.47 ± 0.08 5.0 ± 0.4 9±1 3.67 ± 0.06 1.8 ± 0.2 6.4 ± 0.5 1.77 ± 0.04
10 ± 1 51 ± 5 3.46 ± 0.09 6.3 ± 0.4 11 ± 1 2.87 ± 0.05 1.7 ± 0.1 5.9 ± 0.5 1.79 ± 0.04
7431
DOI: 10.1021/acs.jpcb.8b04364 J. Phys. Chem. B 2018, 122, 7426−7435
Article
The Journal of Physical Chemistry B
Figure 7. Circular dichroism from HPPhOx/BSA complexes. Concentration is at 0.2 mg mL−1 and that of HPPhOx is 0 (black), 0.5 (red), and 1.0 (blue) mg mL−1 with no added salt.
Additionally, the presence of a cross-linked polymeric shell is needed to stabilize the complexes at room temperature while it is used in PPrOx−PEI18 and POx−PEI19 polyplexes. PDI values are smaller than those in the other salt contents, in line with the CONTIN analysis of Figure 1. The morphology of HPPhOx/BSA complexes at length scales between 1 and 100 nm in aqueous media was measured by SANS. These length scales correspond to the interior of the investigated nanoparticles, especially the ones with Rh ∼ 100 nm or higher. HPPhOx was at a concentration that provides adequately strong scattering in SANS. In complexes, BSA concentration was chosen so that the protein/polymer mass ratios were in the same range as the ones in LS experiments; i.e., mBSA/mHPPhOx is between 0.13 and 0.5. In Figure 4a, SANS profiles at different salt contents in D2O are presented. In the case of 0 M, the separate contributions of eq 1 are also shown (Figure 4b). The SANS profiles in the absence of salt are apparently dominated by associations with diffuse internal structure as they do not diverge significantly from power-law behavior. At 0.03 M, a well-defined structure appears at intermediate length scales in the form of a Guinier regime between 0.01 and 0.04 Å−1. At 0.15 M, the low-q scattering increases again but the Guinier regime is still evident. The extracted forward scattering from the hierarchical levels a and b follow these trends (Table 2). In the absence of salt, Rg,a is higher than the values obtained by LS of our previous work30 (Rg ≈ 40 nm). This possibly comes from the different concentration regime we test in SANS in comparison with LS. The higher hierarchical level indicates presence of clustered aggregates. As will be discussed, the association in HPPhOx is evidently very sensitive to concentration. At 0.03 and 0.15 M, the largest structural level has a characteristic size Rg,a clearly smaller than the one observed previously by SLS and DLS mentioned above and previously.30 This means that internal structure is probed. Indeed, the corresponding SANS profiles (Figure 4a) show an upward trend at very low q, signifying clustering at even higher length scales. The internal domains’ contribution is dominant with Rg,b ≈ 9 nm (Figure 4a and Table 2). It is related to well-defined domains made of associative hydrophobic polymeric segments of high local concentration. Their morphology does not change significantly upon further salt addition. Their fractal exponent db approaches the one of rough surface fractal in the presence of salt. This points out the compact chain associations under
Figure 8. (a) ATR-FTIR from HPPhOx (0.75 mg mL−1)/BSA (0.4 mg mL−1) complexes with no added salt (black), pure HPPhOx (0.75 mg mL−1) (gray), difference of the two spectra (blue), and pure BSA (0.2 mg mL−1) (red); fitting of pure BSA (b) and BSA inside the HPPhOx complex (c), as described in the main text. Experimental data (blue) and fitting profiles are shown with continuous lines. The separate components representing different wavenumber regions (secondary structure assignments) are presented with black, red, blue, and green dotted lines.
conditions of screened electrostatic interactions. Similarly, the high q power-law exponent, defined by dc, slightly increases from 1.54 to 1.76−1.77, revealing denser interchain associations and less extended chain conformations at the nanometer scale. The relative contribution of levels a and b changes nonsystematically mainly because of the abrupt increase in Ga upon addition of 0.15 M salt (Table 2). Their aggregation numbers obtain their minimum value at 0.03 M NaCl. At high salt, clustering between b-level structures is greatly pronounced. 7432
DOI: 10.1021/acs.jpcb.8b04364 J. Phys. Chem. B 2018, 122, 7426−7435
Article
The Journal of Physical Chemistry B Table 4. Secondary Structure of BSA Determined by Fitted ATR-FTIR Data assignment
β-turn
α-helix
random coil
short-segment chains connecting α-helical segments
intermolecular β-sheet
wavenumbers (cm−1) BSA HPPhOx/BSA1a HPPhOx/BSA2a HPPhOx/BSA3a HPPhOx/BSA4a
1675−1679 14.2% 18.5% 20.1% 23.2% 23.1%
1655−1658 63.9% 33.5% 38.9% 35.2% 33.7%
1640−1642
1632−1639 17.9% 21.3% 17.4% 20.5% 20.9%
1618−1619 4.0% 1.0%
25.6% 23.6% 21.1% 22.3%
From HPPhOx/BSA complexes at 0.75 mg mL−1 HPPhOx. For 1, 3, and 4 BSA concentration at 0.2 mg mL−1 and for 2 at 0.4 mg mL−1. For 1 and 2 NaCl concentrations at 0 M, for 3 at 0.03 M, and for 4 at 0.15 M.
a
that no significant change in BSA conformation is detected by CD after interaction with HPPhOx aggregates. ATR-FTIR was used to elucidate any possible secondary structure changes of BSA upon interaction with HPPhOx. In quaternized chloromethyl styrene cationic polyelectrolytes, we have shown that the hydrophobic interaction led to significant denaturation of the protein’s native structure.26 Hence, ATRFTIR data of this work was analyzed in the amide I region (1700−1600 cm−1). In Figure 8a, the data from pure BSA, pure HPPhOx, and HPPhOx/BSA complex are presented. The difference between the polyelectrolyte/protein complex and the pure protein, which is the signal from the complexed protein, shows that there is a clear structural change in BSA. In denaturation studies of BSA, this enhancement of signal at lower wavenumbers is a sign of denaturation of α-helix, which dominates its native conformation.50,51 Data from uncomplexed and complexed BSA were quantified by fitting the experimental curves with a superposition of Gaussian profiles52,53 with the band assignments of Table 4. The native structure is rich in α-helix conformation (1655−1658 cm−1), whereas upon interaction with HPPhOx, the β-turn (1675−1679 cm−1) and short short-segment chains connecting α-helical segment (1632−1639) contributions increase. At the same time, the (1640−1642 cm−1) random-coil conformation becomes a significant part of the spectrum, whereas it was not detected in pure BSA. We conclude that ATR-FTIR measurements can prove a significant conformational change in BSA from mostly α-helix to mostly random-coil, β-turn, and short short-segment chains connecting α-helical segments.
In Figure 5, the SANS data at different concentrations show that concentration dependence is not trivial. Upon increasing polymer content, both the scattering intensity and the shape of the curves change. A well-defined structure at intermediate q is evident at 7.5 mg mL−1, which corresponds to the internal hydrophobic domains. On the other hand, the forward scattering of the intermediate level Gb increases systematically with concentration at 0 M NaCl resulting to aggregation numbers Nb between 270 and 390. At 1.875 and 7.5 mg mL−1, the b-structure is pronounced (Figure 5). The HPPhOx aggregates show some structural similarities with casein micelles and other disordered protein aggregates. The amphiphilic nature of casein proteins allows their associations to interact and encapsulate hydrophobic substances.47 Casein aggregates48 consist of submicelles (R ∼ 7 nm) into spherical particles (R ∼ 120 nm) and are covered by a “hairy layer” that stabilizes the particles.49 In any case, the nontrivial concentration dependence reveals the soft and fluctuating interior of aggregates that should be susceptible to significant rearrangements under the influence of other objects, e.g., protein globules and hydrophobic drugs. In Figure 6, the fitted SANS profiles from HPPhOx/BSA complexes are depicted. The presence of protein does not seem to drastically change the morphology at 0 M NaCl. The resulting parameters (Table 3) do not show any systematic trend either. From LS at 0 M, interactions with BSA were described as an effect on interaggregate association and dissociation. Consequently, this takes place at length scales of the order of 100 nm or higher, i.e., in the outer regions of the polymeric aggregates. This effect does not influence the internal morphology of the aggregates and hence the SANS picture. At 0.03 M NaCl (Figure 6b), the intermediate length-scale scattering is gradually taken over by the scattering at low q. This means that BSA accumulates preferably at the exterior of the aggregates causing their interassociations. The relative contributions from Ga and Gb (Table 3) are compatible with this conclusion. The extracted Rg,a can be identified with the Rg obtained by SLS at 0.1 and 0.2 mg mL−1 BSA where breaking of the aggregates was concluded. At high salt content (0.15 M NaCl) (Figure 6c), it is clear that protein reduces aggregation at large length scales in favor of the intermediate length-scale domains. This is a qualitatively different situation than in the absence of salt. BSA globules enter the aggregates and interact with the internal hydrophobic domains. Indeed, Gb systematically increases, whereas Rg,b also does so but marginally (Table 3). This is compatible with the increased molar masses observed by SLS at high salt. In Figure 7, the circular dichroism results show the expected signal for solutions of pure BSA. By increasing polymer concentration, the noise level is enhanced and the wavelength range is restricted, probably due to spectral contributions from the phenyl groups of the copolymer. However, we can assume
■
CONCLUSIONS Complexation of the protein BSA with aggregates from the amphiphilic polyelectrolyte HPPhOx was used for the formation of nanoparticles with controllable size and internal morphology. As both electrostatic and hydrophobic interactions drive associations in this system, it was found that screening of electrostatic interactions by addition of salt weakens BSAinduced interaggregate bridging, leads to better-defined nanoparticles, and enhances hydrophobic domains in the interior of the complexes. The native protein conformation was significantly denatured upon interaction with the amphiphilic polyelectrolyte as it was found by ATR-FTIR. This copolymer/protein combination enriches our understanding of protein/hydrophobically modified polyelectrolyte interactions and structure manipulation of resulting complexes and offers candidate stimuli-responsive, potentially biocompatible nanoformulations. It could find applications in protein delivery, as a hydrophobic drug container and as a more general guide for producing multifunctional hybrid macromolecular nanoparticles. 7433
DOI: 10.1021/acs.jpcb.8b04364 J. Phys. Chem. B 2018, 122, 7426−7435
Article
The Journal of Physical Chemistry B
■
(13) He, Z.; Miao, L.; Jordan, R.; S-Manickam, D.; Luxenhofer, R.; Kabanov, A. V. A Low Protein Binding Cationic Poly(2-oxazoline) as Non-Viral Vector. Macromol. Biosci. 2015, 15, 1004−1020. (14) He, Z.; Schulz, A.; Wan, X.; Seitz, J.; Bludau, H.; Alakhova, D. Y.; Darr, D. B.; Perou, C. M.; Jordan, R.; Ojima, I.; et al. Poly(2-oxazoline) based micelles with high capacity for 3rd generation taxoids: Preparation, in vitro and in vivo evaluation. J. Controlled Release 2015, 208, 67−75. (15) Hsiue, G.-H.; Wang, C.-H.; Lo, C.-L.; Wang, C.-H.; Li, J.-P.; Yang, J.-L. Environmental-sensitive micelles based on poly(2-ethyl-2oxazoline)-b-poly(L-lactide) diblock copolymer for application in drug delivery. Int. J. Pharm. 2006, 317, 69−75. (16) Uchman, M.; Hajduová, J.; Vlassi, E.; Pispas, S.; Appavou, M.-S.; Š těpán ek, M. Self- and co-assembly of amphiphilic gradient polyelectrolyte in aqueous solution: Interaction with oppositely charged ionic surfactant. Eur. Polym. J. 2015, 73, 212−221. (17) Vlassi, E.; Pispas, S. Solution Behavior of Hydrolyzed Gradient Methyl/Phenyl Oxazoline Copolymers and Complexation with DNA. Macromol. Chem. Phys. 2015, 216, 873−883. (18) Mees, M.; Haladjova, E.; Momekova, D.; Momekov, G.; Shestakova, P. S.; Tsvetanov, C. B.; Hoogenboom, R.; Rangelov, S. Partially Hydrolyzed Poly(n-propyl-2-oxazoline): Synthesis, Aqueous Solution Properties, and Preparation of Gene Delivery Systems. Biomacromolecules 2016, 17, 3580−3590. (19) Ivanova, T.; Haladjova, E.; Mees, M.; Momekova, D.; Rangelov, S.; Momekov, G.; Hoogenboom, R. Characterization of polymer vector systems based on partially hydrolyzed polyoxazoline for gene transfection. Pharmacia 2016, 63, 3−8. (20) Becker, A. L.; Henzler, K.; Welsch, N.; Ballauff, M.; Borisov, O. Proteins and polyelectrolytes: A charged relationship. Curr. Opin. Colloid Interface Sci. 2012, 17, 90−96. (21) Kayitmazer, A. B.; Seeman, D.; Minsky, B. B.; Dubin, P. L.; Xu, Y. Protein-polyelectrolyte interactions. Soft Matter 2013, 9, 2553−2583. (22) Ballauff, M. Spherical polyelectrolyte brushes. Prog. Polym. Sci. 2007, 32, 1135−1151. (23) Papagiannopoulos, A.; Meristoudi, A.; Pispas, S.; Radulescu, A. Micelles from HOOC-PnBA-b-PAA-C12H15 Diblock Amphiphilic Polyelectrolytes as Protein Nanocarriers. Biomacromolecules 2016, 17, 3816−3827. (24) Papagiannopoulos, A.; Meristoudi, A.; Pispas, S.; Keiderling, U. Thermoresponsive behavior of micellar aggregates from end-functionalized PnBA-b-PNIPAM-COOH block copolymers and their complexes with lysozyme. Soft Matter 2016, 12, 6547−6556. (25) Cousin, F.; Gummel, J.; Clemens, D.; Grillo, I.; Boué, F. Multiple Scale Reorganization of Electrostatic Complexes of Poly(styrenesulfonate) and Lysozyme. Langmuir 2010, 26, 7078−7085. (26) Papagiannopoulos, A.; Vlassi, E.; Pispas, S.; Jafta, C. J. Tuning the solution organization of cationic polymers through interactions with bovine serum albumin. Phys. Chem. Chem. Phys. 2017, 19, 18471− 18480. (27) Kayitmazer, A. B.; Bohidar, H. B.; Mattison, K. W.; Bose, A.; Sarkar, J.; Hashidzume, A.; Russo, P. S.; Jaeger, W.; Dubin, P. L. Mesophase separation and probe dynamics in protein-polyelectrolyte coacervates. Soft Matter 2007, 3, 1064−1076. (28) Tang, Y.; Duan, J.; Wu, J. A laser light scattering study of complex formation between soybean peroxidase and poly(N-isopropylacrylamide-co-sodium styrene sulfonate). Colloids Surf., A 2012, 395, 82−87. (29) Rosenfeldt, S.; Wittemann, A.; Ballauff, M.; Breininger, E.; Bolze, J.; Dingenouts, N. Interaction of proteins with spherical polyelectrolyte brushes in solution as studied by small-angle X-ray scattering. Phys. Rev. E 2004, 70, No. 061403. (30) Vlassi, E.; Papagiannopoulos, A.; Pispas, S. Hydrolyzed poly(2plenyl-2-oxazoline)s in aqueous media and biological fluids. Macromol. Chem. Phys. 2018, No. 1800047. (31) Chu, B. Laser Light Scattering, 2nd ed.; Academic Press: New York, 1991. (32) Berne, B. J.; Pecora, R. Dynamic Light Scattering, with Applications to Chemistry, Biology, and Physics; Dover: Toronto, 2000.
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. ORCID
Aristeidis Papagiannopoulos: 0000-0002-5662-9866 Judith Elizabeth Houston: 0000-0001-5205-3620 Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS We acknowledge support of this work by the project “Advanced Materials and Devices” (MIS 5002409), which is implemented under the “Action for the Strategic Development on the Research and Technological Sector”, funded by the operational program “Competitiveness, Entrepreneurship and Innovation” (NSRF 2014-2020), and co-financed by Greece and the European Union (European Regional Development Fund). This work is based upon experiments performed at the KWS-2 instrument operated by JCNS at the Heinz Maier-Leibnitz Zentrum (MLZ), Garching, Germany.
■
REFERENCES
(1) vander Straeten, A.; Bratek-Skicki, A.; Germain, L.; D’Haese, C.; Eloy, P.; Fustin, C.-A.; Dupont-Gillain, C. Protein-polyelectrolyte complexes to improve the biological activity of proteins in layer-bylayer assemblies. Nanoscale 2017, 9, 17186−17192. (2) Valetti, N. W.; Brassesco, M. E.; Picó, G. A. Polyelectrolytes− protein complexes: a viable platform in the downstream processes of industrial enzymes at scaling up level. J. Chem. Technol. Biotechnol. 2016, 91, 2921−2928. (3) Lankalapalli, S.; Kolapalli, V. R. M. Polyelectrolyte Complexes: A Review of their Applicability in Drug Delivery Technology. Indian J. Pharm. Sci. 2009, 71, 481−487. (4) Shu, S.; Sun, L.; Zhang, X.; Wu, Z.; Wang, Z.; Li, C. Polysaccharides-based polyelectrolyte nanoparticles as protein drugs delivery system. J. Nanopart. Res. 2011, 13, 3657−3670. (5) Ulery, B. D.; Nair, L. S.; Laurencin, C. T. Biomedical Applications of Biodegradable Polymers. J. Polym. Sci., Part B: Polym. Phys. 2011, 49, 832−864. (6) Wang, Z.; Wang, K.; Lu, X.; Li, C.; Han, L.; Xie, C.; Liu, Y.; Qu, S.; Zhen, G. Nanostructured Architectures by Assembling PolysaccharideCoated BSA Nanoparticles for Biomedical Application. Adv. Healthcare Mater. 2015, 4, 927−937. (7) Xie, C.; Lu, X.; Han, L.; Xu, J.; Wang, Z.; Jiang, L.; Wang, K.; Zhang, H.; Ren, F.; Tang, Y. Biomimetic Mineralized Hierarchical Graphene Oxide/Chitosan Scaffolds with Adsorbability for Immobilization of Nanoparticles for Biomedical Applications. ACS Appl. Mater. Interfaces 2016, 8, 1707−1717. (8) Hrubý, M.; Filippov, S. K.; Š těpánek, P. Smart polymers in drug delivery systems on crossroads: Which way deserves following? Eur. Polym. J. 2015, 65, 82−97. (9) Luxenhofer, R.; Han, Y.; Schulz, A.; Tong, J.; He, Z.; Kabanov, A. V.; Jordan, R. Poly(2-oxazoline)s as Polymer Therapeutics. Macromol. Rapid Commun. 2012, 33, 1613−1631. (10) Zahoranová, A.; Kronek, J. Hydrogels Based on Poly(2oxazoline)s for Pharmaceutical Applications. Handbook of Polymers for Pharmaceutical Technologies; John Wiley & Sons, Inc., 2015; pp 231−258. (11) Sedlacek, O.; Monnery, B. D.; Filippov, S. K.; Hoogenboom, R.; Hruby, M. Poly(2-Oxazoline)s − Are They More Advantageous for Biomedical Applications Than Other Polymers? Macromol. Rapid Commun. 2012, 33, 1648−1662. (12) Vlassi, E.; Papagiannopoulos, A.; Pispas, S. Amphiphilic poly(2oxazoline) copolymers as self-assembled carriers for drug delivery applications. Eur. Polym. J. 2017, 88, 516−523. 7434
DOI: 10.1021/acs.jpcb.8b04364 J. Phys. Chem. B 2018, 122, 7426−7435
Article
The Journal of Physical Chemistry B
circumneutral and acidic pH: A tale of two nano-bio surface interactions. J. Colloid Interface Sci. 2017, 493, 334−341.
(33) Barker, J. G.; Pedersen, J. S. Instrumental Smearing Effects in Radially Symmetric Small-Angle Neutron Scattering by Numerical and Analytical Methods. J. Appl. Crystallogr. 1995, 28, 105−114. (34) Radulescu, A.; Szekely, N. K.; Polachowski, S.; Leyendecker, M.; Amann, M.; Buitenhuis, J.; Drochner, M.; Engels, R.; Hanslik, R.; Kemmerling, G.; et al. Tuning the instrument resolution using chopper and time of flight at the small-angle neutron scattering diffractometer KWS-2. J. Appl. Crystallogr. 2015, 48, 1849−1859. (35) Vad, T.; Sager, W. F. C.; Zhang, J.; Buitenhuis, J.; Radulescu, A. Experimental determination of resolution function parameters from small-angle neutron scattering data of a colloidal SiO2 dispersion. J. Appl. Crystallogr. 2010, 43, 686−692. (36) Beaucage, G. Small-Angle Scattering from Polymeric Mass Fractals of Arbitrary Mass-Fractal Dimension. J. Appl. Crystallogr. 1996, 29, 134−146. (37) Papagiannopoulos, A.; Zhao, J.; Zhang, G.; Pispas, S.; Radulescu, A. Thermoresponsive transition of a PEO-b-PNIPAM copolymer: From hierarchical aggregates to well defined ellipsoidal vesicles. Polymer 2013, 54, 6373−6380. (38) Hammouda, B. Analysis of the Beaucage model. J. Appl. Crystallogr. 2010, 43, 1474−1478. (39) Pedersen, J. S. Small-Angle Scattering from Surfactants and Block Copolymer Micelles. In Soft Matter Characterization; Borsali, R., Pecora, R., Eds.; Springer: Dordrecht, 2008; pp 191−233. (40) Papagiannopoulos, A.; Karayianni, M.; Mountrichas, G.; Pispas, S.; Radulescu, A. Micellar and Fractal Aggregates formed by two Triblock Terpolymers with different arrangements of one Charged, one Neutral Hydrophilic and one Hydrophobic Block. Polymer 2015, 63, 134−143. (41) Phillies, G. D. J. Diffusion of bovine serum albumin in a neutral polymer solution. Biopolymers 1985, 24, 379−386. (42) Yohannes, G.; Wiedmer, S. K.; Elomaa, M.; Jussila, M.; Aseyev, V.; Riekkola, M.-L. Thermal aggregation of bovine serum albumin studied by asymmetrical flow field-flow fractionation. Anal. Chim. Acta 2010, 675, 191−198. (43) Nyström, B.; Roots, J. Dynamic light scattering studies of protein solutions under high pressure. J. Chem. Phys. 1983, 78, 2833−2837. (44) Müller, A.; Burchard, W. Structure formation of surfactants in concentrated sulphuric acid: a light scattering study. Colloid Polym. Sci. 1995, 273, 866−875. (45) Schmitz, K. S.; Wang, B.; Kokufuta, E. Mechanism of Microgel Formation via Cross-Linking of Polymers in Their Dilute Solutions: Mathematical Explanation with Computer Simulations. Macromolecules 2001, 34, 8370−8377. (46) Mattison, K. W.; Dubin, P. L.; Brittain, I. J. Complex Formation between Bovine Serum Albumin and Strong Polyelectrolytes: Effect of Polymer Charge Density. J. Phys. Chem. B 1998, 102, 3830−3836. (47) Głąb, T. K.; Boratyński, J. Potential of Casein as a Carrier for Biologically Active Agents. Top. Curr. Chem. 2017, 375, 71. (48) Hansen, S.; Bauer, R.; Lomholt, S. B.; Quist, K. B.; Pedersen, J. S.; Mortensen, K. Structure of casein micelles studied by small-angle neutron scattering. Eur. Biophys. J. 1996, 24, 143−147. (49) Dalgleish, D. G. On the structural models of bovine casein micelles-review and possible improvements. Soft Matter 2011, 7, 2265− 2272. (50) Murayama, K.; Tomida, M. Heat-Induced Secondary Structure and Conformation Change of Bovine Serum Albumin Investigated by Fourier Transform Infrared Spectroscopy. Biochemistry 2004, 43, 11526−11532. (51) Lu, R.; Li, W.-W.; Katzir, A.; Raichlin, Y.; Yu, H.-Q.; Mizaikoff, B. Probing the secondary structure of bovine serum albumin during heatinduced denaturation using mid-infrared fiberoptic sensors. Analyst 2015, 140, 765−770. (52) Roach, P.; Farrar, D.; Perry, C. C. Interpretation of Protein Adsorption: Surface-Induced Conformational Changes. J. Am. Chem. Soc. 2005, 127, 8168−8173. (53) Givens, B. E.; Xu, Z.; Fiegel, J.; Grassian, V. H. Bovine serum albumin adsorption on SiO2 and TiO2 nanoparticle surfaces at 7435
DOI: 10.1021/acs.jpcb.8b04364 J. Phys. Chem. B 2018, 122, 7426−7435