Marc B. Taraban‡
a,
Li Yu‡b,
Yue Fenga,
Elena V. Jouravlevac,
Mikhail A. Anisimovcd,
Zhong-Xing Jiangb and
Y. Bruce Yu*a
aDepartment of Pharmaceutical Sciences, University of Maryland, Baltimore, MD 21201, USA. E-mail: byu@rx.umaryland.edu; Fax: +1-410-706-5017; Tel: +1-410-706-7514
bSchool of Pharmaceutical Sciences, Wuhan University, Wuhan, Hubei 430071, China
cLight Scattering Center, Institute for Physical Science and Technology, University of Maryland, College Park, MD 20742, USA
dDepartment of Chemical and Biomolecular Engineering, University of Maryland, College Park, MD 20742, USA
First published on 20th October 2014
Amphiphiles comprise a hydrophobic moiety and a hydrophilic moiety. A common property of many amphiphiles is to self-associate in aqueous solutions, driven by the need to shield the hydrophobic moiety from water. This feature has been utilized extensively to create various nano-scale architectures from amphiphiles. However, to effectively control amphiphile behavior, one should have the ability to both promote and prevent self-association. Fluorinated amphiphiles are especially prone to self-association, thus presenting a big challenge in developing non-associative amphiphiles. In this work, we solve this challenge by creating steric hindrance to association. The resulting fluorinated asymmetric amphiphile remains monomeric well above its apparent critical micelle concentration and up to its solubility limit, as demonstrated by small-angle X-ray and neutron scattering, dynamic light scattering and NMR diffusometry techniques. Not being able to associate intermolecularly, the amphiphile undergoes an intramolecular conformational transition, akin to protein folding, to wrap its hydrophilic moiety around its hydrophobic fluorocarbon moiety to shield it from water. This work demonstrates that steric hindrance is an effective tool in creating non-associative amphiphiles.
Generally, in supramolecular chemistry of amphiphilic molecules, one could distinguish intermolecular assembly—the aggregation of a number of such molecules into certain nano-scale construct; or intramolecular assembly—the reorganization of a single molecule into the most thermodynamically stable structure. Intermolecular assembly is more easily experimentally detected and, hence, extensively studied and better understood5 as opposed to intramolecular assembly, which is often an intermediate step in inter-molecular assembly.7 A notable example of standalone intramolecular assembly is the transition of charged polyamidoamine (PAMAM) dendrimers, which change their conformations in response to changes in solution pH and ionic strength.8 Such intramolecular assembly is attributed to the effects of counterions, which prevent intermolecular assembly.9
Both inter- and intramolecular assembly of amphiphiles are driven by a plethora of non-covalent forces, including, but not limited to, electrostatic interactions, hydrophobic interactions, hydrogen bonding, donor–acceptor interactions, and van der Waals forces, etc. For instance, a number of building units have been synthesized capable to self-assemble into three-dimensional supramolecular constructs5 driven, e.g., by hydrogen bonding,10 or by electrostatic interactions triggered in the presence of transition or rare earth metal ions.11 Of course, whether an amphiphile will undergo intermolecular association or intramolecular conformational transition will critically depend on its structure. Understandably, intermolecular assembly is also regulated by the shape of the building unit, since steric hindrances often control the assembly process. Indeed, the overwhelming majority of building units have conic or truncated conic shapes with the assembly triggering function located at the ‘tip’ (‘top’) of the cone. Such shape minimizes the hindrances when the above units are organized first into spheres (or disks), then into cylinders, which in turn can further pack into cubic lattices.5 Conical shape of the single unit also facilitates the preparation of the so-called dendronized linear polymers where a number of dendrimers are attached to numerous binding sites along the polymeric backbone.12 The structural organization of the dendronized polymers affords even wider variety of nano-scaled shapes—toroids, nanoribbons, nanotubes, etc.5
Amphiphilic dendrimers,13 which comprise a hydrophobic core and a hydrophilic dendron, are some of the most versatile building units for nano-scale structures due their strong tendency to self-associate.3 However, such structural organization might also provide an excellent structural motif for non-associate amphiphiles for two reasons. On the one hand, their bi-conical shape may create steric hindrance to intermolecular association in water. On the other hand, amphiphiles with hydrophilic dendrons have the potential to undergo intramolecular conformational transitions. For example, amphiphiles made of hydrophobic aromatic dendrons covalently linked to hydrophilic poly(ethylene glycol) (PEG) chains switch conformation in response to change of solvent polarity. In nonpolar solvents such as THF, the hydrophilic PEG chains are tightly packed and shielded from the apolar solvent by the hydrophobic moiety; but in polar solvents such as MeOH, the hydrophilic PEG chains become extended and wrap around the hydrophobic dendrons to shield them from the polar solvent.14 Although these previously observed conformational changes in organic solvents were in response to changes in solvent polarity, they nonetheless suggest the possibility that one part of an amphiphile could be covered by another part, thereby providing an alternative to intermolecular association.
Compared with hydrocarbons, which are hydrophobic but lipophilic, fluorocarbons are hydrophobic and lipophobic at the same time. Consequently, fluorocarbons tend to form their own phase, the so-called fluorous phase.15 As a result, fluorinated amphiphiles typically are more prone to intermolecular association as opposed to their non-fluorinated counterparts. For example, sodium perfluorooctanoate (a linear perfluorinated amphiphile) has a critical micellar concentration (CMC) of ca. 30 mM,16 while its non-fluorinated analog, sodium octanoate, has a CMC of ∼400 mM.17 From this viewpoint, it is particularly challenging to develop a fluorinated amphiphile that does not self-associate in aqueous solutions. In this paper, we describe the synthesis of a non-associative fluorinated amphiphile with 27 equivalent fluorine atoms, FIT-27 (Fig. 1), as well as its structural characterization in aqueous solutions in a wide concentration range. Structurally, this asymmetric fluorinated amphiphile is comprised of a hydrophobic fluorocarbon dendron and a hydrophilic hydrogenous dendron as the solubility enhancer in aqueous media.
Here, we demonstrate that instead of self-association, this fluorinated amphiphile undergoes a conformational transition in which the hydrophilic chains wrap around the hydrophobic dendron to shield it from water. In 19F NMR, such conformational transition manifests itself very similar to micellization with a characteristic critical concentration. Therefore, to reveal such intramolecular changes, we investigated the solution structure of this amphiphilic dendrimer using a combination of small-angle X-ray and neutron scattering (SAXS/SANS), dynamic light scattering (DLS), and NMR diffusometry.
To explore the possibility of micellization by FIT-27, we used a well-established approach based on the dependence of δ(19F) vs. the concentration reciprocal, 1/C.19 It is expected that below the CMC, δ(19F) will remain essentially unchanged since the amphiphiles should exist predominantly as non-associated monomers. Above the CMC, significant upfield shift in δ(19F) is expected as a result of micelle formation, which transfers the fluorocarbon moiety from an aqueous to a fluorous environment. Thus, the plot of δ(19F) vs. 1/C should yield two straight lines intersecting at 1/CMC. Indeed, as seen from Fig. 2(B), the δ(19F) vs. 1/C plot of FIT-27 resulted in the expected pattern with the intersection point of two straight lines corresponding to a CMC of ∼7.5 mM.
The results presented in Fig. 2(B) on FIT-27 are entirely consistent with those from its two close analogs, which also display an intersection around 7–8 mM in the δ(19F) vs. 1/C plot.20 Further, such behaviors are consistent with micelle formation, as observed in other fluorinated amphiphiles, where micellization leads to a higher-field shift of the 19F chemical shift.16 Hence one might conclude that FIT-27 forms micelles with CMC of ∼7.5 mM. However, the absolute changes of δ(19F) in Fig. 2(B) amounts only to ∼0.05 ppm in the concentration range from 0.2 mM to 100 mM, which is much smaller than the 19F chemical shift change observed in the micellization of other fluorinated amphiphiles, which is ∼2 ppm.19 This raises the question as to what kind of micelles had formed by FIT-27, or even, whether micelles had formed at all. To find out, we used scattering and diffusometry techniques, including SAXS, SANS, DLS and PFG NMR to directly observe FIT-27 at three concentrations: 1 mM (well below CMC), 10 mM (slightly above CMC), and 100 mM (well above CMC).
As seen from the SAXS scattering profiles of I(Q) vs. Q (Fig. 3), FIT-27 solutions at the three concentrations show no evidence of aggregation at low Q, which is flat. In contrast, aggregation would have caused an uptick in the low Q range (below 0.03 Å−1).21 Moreover, the Guinier plots for globular particles, lnI(Q) vs. Q2,22 are perfectly linear for both 1 mM and 10 mM FIT-27 solutions (inset in Fig. 3), which strongly suggest the absence of aggregation and the monodispersity of scattering particles at 1 mM and 10 mM. No Guinier analysis of 100 mM data could be performed due to the strong structure factor peak centered around 0.1 Å−1 (Fig. 3).
Indirect Fourier transformation of the scattering data for the 1 mM and 10 mM FIT-27 solutions result in the corresponding pair-wise distance distribution functions, P(r), shown in Fig. 4(A) and (B). From P(r), the low resolution structure of the scattering particles can be constructed, as shown in Fig. 4(C) and (D). As seen from Fig. 4, solution structures of FIT-27 are different at 1 mM and 10 mM (see also the differential of their corresponding P(r) functions in the ESI, Fig. S20†).
At 1 mM, FIT-27 has an elongated drop-like shape, with maximum dimension dmax ∼ 51 Å, and radius of gyration Rg = 17.8 Å (Fig. 4(A) and (C)). Based on the FIT-27 chemical structure, one might suggest that the top smaller portion of the drop-like shape (∼25 Å in diameter) belongs to the fluorocarbon moiety, while the bulkier bottom portion (∼35 Å in diameter) is formed by the four more flexible and longer ethylene oxide chains (Fig. 4(C)). When the concentration of FIT-27 increases to 10 mM, the conformation of the molecule changes, which is also consistent with critical point of the chemical shift δ(19F) change in Fig. 2(B). At 10 mM, FIT-27 in solution acquires the shape of an oblate spheroid with maximum dimension dmax ∼ 47 Å, and radius of gyration Rg = 17.2 Å (Fig. 4(B) and (D)). Both dmax and Rg suggest that the structure of FIT-27 at 10 mM is more compact than at 1 mM.
The SAXS scattering profile of FIT-27 at 100 mM also shows no signs of large assemblies—the scattering profile at 100 mM is essentially flat in the range of Q < 0.03 Å−1 (Fig. 3). On the other hand, at 100 mM, an apparent peak, which is absent at 1 and 10 mM, is observed around Q ∼ 0.1 Å−1 (Fig. 3). Similar peaks are commonly attributed to interparticle interaction due to certain ordering of their organization,8 and their appearance at high concentrations is attributed to the intermolecular structure factor.23 In general, SAXS scattering intensity could be expressed as I(Q) ∼ P(Q) × S(Q), where P(Q) is the form factor of the scattering particles, and S(Q) is the structure factor; in dilute solutions, S(Q) = 1.24 Here, the form factor P(Q) characterizes the shape of the individual scattering particle, while in non-dilute systems, the structure factor S(Q) reflects the interference due to the dense packing of the scattering particles when the distance between particles becomes of the same order of magnitude as the size of a particle itself.
Since no aggregation of FIT-27 is observed at 100 mM, one can assume that the form factor P(Q) at 100 mM could be approximated by normalizing the scattering profile of FIT-27 at 10 mM, i.e.,
P(Q, 100 mM) ≅ 10 × P(Q, 10 mM) = 10 × I(Q, 10 mM) | (1) |
The absence of the aggregates at 10 mM as well as the perfect linearity of the Guinier plot (Fig. 3, inset) suggests that at this concentration of FIT-27 the interparticle interference could be negligible, i.e., S(Q, 10 mM) ≅ 1. Subsequent division of the scattering data for 100 mM FIT-27 by such normalized values for 10 mM solution will give the desired structure factor S(Q) at 100 mM (Fig. 5), i.e.,
![]() | (2) |
This S(Q) vs. Q profile could be instructive to select the proper structure factor when modeling size distribution, volume fraction, and the shape of the scattering particles with certain order of organization in non-dilute solutions. Since the S(Q) vs. Q in Fig. 5 is very similar to the theoretical description of the structure factor for hard spheres in non-dilute solutions (see Fig. S21, ESI†), we modeled the experimental I(Q) vs. Q profile for the 100 mM FIT-27 solution using the corresponding IRENA 2.46 (ref. 25) routine based on the structure factor for hard spheres from the NIST SAS software.26 The results of the modelling show that the low resolution shape of the scattering particle is close to the one suggested at 10 mM within the range of ±3 Å (see Fig. S22 and description, ESI†). In general, one might get an approximate estimate of the average distance d between the scattering particles in 100 mM solution based on the position of structure factor peak Qs (∼0.091 Å−1, Fig. 3) using the standard Bragg's law:22
d = 2π/Qs, d ∼ 70 Å | (3) |
To get more clear insight with respect to the location of the fluorocarbon moiety in the above low-resolution structure of FIT-27, we have opted to perform SANS experiments on 1 mM, 10 mM, and 100 mM FIT-27 solutions in H2O-based PBS (pH 7.4)—same as in SAXS experiments. There are numerous experimental evidences suggesting that due to the big difference between coherent neutron scattering length densities (SLD) of fluorinated and hydrogenous moieties of a molecule,27 one might try to selectively visualize the fluorinated part in H2O without D2O additions. Indeed, e.g., for the linear C8F17-nonioic amphiphiles, neutron SLD for the hydrophobic (fluorinated) part is more than 7 times higher as compared to the hydrophilic (hydrogenous) part (cf. 4.3 × 1010 cm−2 vs. 0.6 × 1010 cm−2).28 Therefore, we attempted to reconstruct the low-resolution 3D shapes of FIT-27 from SANS scattering profiles in expectation that partial contrast matching in H2O will allow us to visualize the fluorinated moiety, i.e., water will at least partially mask the hydrogenous moiety due to their closer SLDs (−0.56 × 1010 cm−2 for water vs. 0.6 × 1010 cm−2 for hydrogenous part) with no such effect on the fluorinated part due to their very different SLD (−0.56 × 1010 cm−2 for water vs. for 4.3 × 1010 cm−2 for fluorocarbons).28
Unlike SAXS data, due to the strong background signal from H2O, SANS profile of FIT-27 at 1 mM is very noisy (Fig. 6(A)). Fortunately, at 10 mM, scattering from FIT-27 is much stronger than the background. Indirect Fourier transform of I(Q) of 10 mM resulted in P(r) function (Fig. 6(B)) with dimensional parameters very close to those observed for the same solution in SAXS experiment—dmax ∼ 48 Å, Rg = 17.3 Å. Reconstructed low-resolution 3D shape of FIT-27 in 10 mM solution from SANS data also appears to be very similar to the oblate spheroid reconstructed from SAXS data (cf. dark green shapes for results from SAXS and light green shapes for results from SANS, Fig. 6(C)). However, in SANS experiments, partial contrast matching between the buffer and the hydrogenous moiety of FIT-27 has allowed us to mask a portion of the hydrogenous moiety—this resulted in a Saturn-like shape with a visualized central fluorinated core (light green shapes in Fig. 6(C)). This Saturn-like shape is also consistent with the observed shoulder peak in P(r) at r ∼ 35 Å (Fig. 6(B)), which corresponds to the size of the fluorinated core (Fig. 6(C)).
Based on the SAXS and SANS results, we conclude that the abrupt change of the 19F chemical shift at the critical concentration is not the result of micellization. Rather, it is the result of a transition between an extended conformation and a compact conformation of FIT-27. Below the critical concentration, FIT-27 adopts an extended drop-shaped conformation with the fluorocarbon moiety exposed to water; above CMC, FIT-27 adopts a compact conformation with the PEG chains wrapping around the fluorocarbon moiety to shield it from water, as illustrated Fig. 7.
Note that PEG chains are much more polar than fluorocarbons. This explains the very small change of the 19F chemical shift displayed by FIT-27 during this process. In essence, intra-molecular conformation transition transfers the fluorocarbons from an aqueous to a PEG environment while inter-molecular micellization would have transferred the fluorocarbons from an aqueous to a fluorous environment. The former causes much smaller change in environment polarity that the latter, and thereby much smaller 19F chemical shift change.
Like associative amphiphiles prone to intermolecular association, this fluorinated amphiphile has its hydrophobic group exposed to water below its critical concentration. Above the critical concentration, this non-associative fluorinated amphiphile covers its hydrophobic group through intramolecular conformational transition rather than through intermolecular association, as in associative amphiphiles. Such concentration-induced conformational transition bears certain similarity to crowding-induced compaction and folding of polypeptide chains/proteins.29 One possible driving force of such transformation could be the solubilization energy of the hydrophobic fluorocarbon moiety of FIT-27 by water molecules. Below the critical concentration (∼7.5 mM), the free energy of solubilization of the fluorocarbon moiety of FIT-27 is probably low enough, and the molecule favors drop-like conformation of high conformational entropy with extended hydrophilic tails. As the concentration of FIT-27 reaches the critical point and beyond, the crowding effect drives up the free energy of solubilizing the fluorocarbon moiety by water. This leads to conformational transition from extended to compact shape where the hydrophobic moiety is shielded from water molecules. Micellization/association was not observed presumably due to the unfavorable sterical hindrances created by the bulky shape of the fluorocarbon moiety.
The absence of large FIT-27 assemblies was also confirmed in DLS experiments. As seen from Fig. 8(A), at all three concentrations (1, 10, and 100 mM), three very close values of the decay times of the autocorrelation functions in FIT-27 solutions were observed: τ (1 mM) = 32 μs; τ (10 mM) = 37 μs; τ (100 mM) = 37 μs. From these correlation times the collective diffusion coefficient Dc of FIT-27 can be obtained, which is 8.3 × 10−11, 7.2 × 10−11 and 7.1 × 10−11 m2 s−1 at 1, 10 and 100 mM, respectively. From Dc, the average hydrodynamic radius Rh of hydrated FIT-27 molecules can be calculated using the Stokes–Einstein equation, which is 31, 33 and 34 Å, respectively at 1, 10 and 100 mM.
Fig. 8(B) shows the distribution of Rh at these three concentrations. The width of the distribution in 10 and 100 mM solutions is very close to each other (cf. green and blue traces, Fig. 8(B)), consistent with the similar oblate spheroid shapes of the molecule reconstructed from SAXS data (Fig. 4(D)). The noticeably broader Rh distribution of FIT-27 in 1 mM solution (red trace, Fig. 8(B)) could be due to more asymmetrical drop-like shape of FIT-27 with extended ethylene oxide chains (suggested based on SAXS data). The observed peaks of the hydrodynamic radii (Fig. 8(B)) could be attributed to non-aggregated FIT-27 monomers, and are in a good agreement with the dimensional parameters of FIT-27 molecule concluded on the basis of SAXS and SANS data. Since the radius of gyration Rg from SAXS/SANS is the averaged distance of all scattering elements from the center of gravity of the scattering particle weighted by the scattering contrasts, while the hydrodynamic radius Rh from DLS reflects the radius of the hydrated particle and is derived from diffusion measurements, Rg is usually smaller than Rh. For example, for solid spheres, theory gives Rg/Rh ∼ 0.7–0.8.30 It has been shown that, for compact dendrimers, Rg/Rh ∼ 0.5–0.7,31 which is consistent with our observations in the present work (e.g., for 1 mM FIT-27 solution, Rg/Rh ∼ 0.57).
The distribution of the decay times of the autocorrelation function at 100 mM also shows a peak with very slow relaxation times τ ∼ 40 ms (Fig. 8(A)). Such along relaxation times could not be ascribed to the Brownian diffusion motions of particles. Rather, this peak reflects the average times of collective motion of organized densely packed FIT-27 particles with overlapping hydration layers at 100 mM in the laser spot. Indeed, such slow motions are consistent with the very slow dynamics of FIT-27 at 100 mM—in a full agreement with the SAXS data (Fig. S22, ESI†), which favor the very dense packing of the FIT-27 particles restricting the faster motion of the particles.
Effects of dense packing and the absence of large assemblies in FIT-27 solutions were also confirmed by the measurements of the self-diffusion coefficient Ds using the PFG NMR technique through the 19F signal from FIT-27. To account for any concentration-driven viscosity effects, the FIT-27 self-diffusion coefficient Ds(FIT-27) is normalized by the water self-diffusion coefficient Ds(H2O) measured in the same solution through the 1H2O signal. Fig. 9(A) shows the concentration dependence of normalized Ds(FIT-27)/Ds(H2O) values as well as the values of individual diffusion coefficients (Table inset, Fig. 9(A)).
As seen from Fig. 9(A), no significant changes in the self-diffusion of FIT-27 molecules were detected when its concentration increased from 1 mM to 10 mM. This observation corroborates the absence of large aggregates of FIT-27 as well as the negligible intermolecular interference at 10 mM, even though this is already above the critical concentration of 7.5 mM. At 100 mM, noticeable retardation of FIT-27 diffusion is observed, and this is in full agreement with the slow motion of non-associated but densely packed FIT-27 molecules with overlapping hydration layers concluded on the basis of SAXS and DLS data for this concentration.
Of note, the observed collective diffusion coefficient Dc of FIT-27 from DLS measurements is generally larger than the self-diffusion diffusion coefficient Ds from PFG NMR experiments (Fig. 9(B)). Such differences can be explained by the analysis of collective vs. self-diffusion coefficients in the framework of frictional formalism for binary solutions.32 It has been shown that collective (Dc) and self-diffusion (Ds) are described by different relationships depending on the friction coefficients between the components of the binary system:
Dc ∼ (f12c1)−1 and Ds ∼ (f12c1 + f22c2)−1 | (4) |
In summary, the abrupt change of the NMR chemical shift at certain critical concentration is conventionally used as an indication of micellization of amphiphiles. However, as have been shown by the combination of SAXS, SANS, DLS, and NMR diffusometry measurements, the observed critical point in the concentration dependence of the NMR chemical shift could also reflect the conformational transition of the amphiphile molecule in the absence of any aggregation. One possible driving force of such transformation could be the solubilization energy of the hydrophobic fluorocarbon moiety of FIT-27 by water molecules. At low concentration, like other surfactants, FIT-27 exists in the monomeric state with its hydrophobic fluorocarbon moiety exposed with extended hydrophilic tails. As the concentration increases, again like other surfactants, FIT-27 needs to have its hydrophobic moiety shield from water, presumably due to the unfavorable free energy of solvating a large number of hydrophobic groups. However, unlike other surfactants, whose shape permit close packing into, e.g., spherical micelles, the drop-like shape of FIT-27 molecule makes such close packing of the bulky fluorine moiety impossible. On the other hand, the hydrophilic chains are long and wide enough to cover the fluorocarbon part. As a result, the hydrophilic chains wrap around the hydrophobic moiety to shield it from water. In the case of dendrimeric amphiphiles, such as FIT-27, micellization in general might be incompatible with the shape of the molecule due to aforementioned reasons. Shielding of the hydrophobic part is achieved through intra-molecular conformational transition, resulting in the hydrophilic part wrapping around the hydrophobic part, a process that bears some resemblance to protein folding in a crowded environment.
Solution scattering profiles I(Q) vs. Q used to study the structures of FIT-27 at 1 mM, 10 mM, and 100 mM were processed using the ATSAS software (European Molecular Biology Laboratory, Hamburg).39,40 This software was also used to restore low resolution 3D structures of FIT-27 solutions based on the simulated annealing algorithm. The analysis of pair-wise distance distribution functions for globular particles P(r) was performed using the linear regularization method of indirect Fourier-transformation using the program GNOM. P(r) is proportional to the probability of finding different vector lengths connecting two unit-volume elements within the scattering particle, and P(r) = 0 happens at the maximum linear dimension of the scattering particle, dmax (i.e., for r ≥ dmax, P(r) = 0). The radius of gyration of the scattering globular particle, Rg, is derived from the second moment of P(r) using the following equation.
![]() | (5) |
Footnotes |
† Electronic supplementary information (ESI) available: Experimental procedures, 1H, 13C and 19F NMR spectra of the intermediate compounds of the FIT-27 synthesis, and their MS and elemental analysis data. HPLC and MS data of final product, SAXS structure factor information, and comparison and differential of pair-wise distance distribution functions of FIT-27 at different concentrations. See DOI: 10.1039/c4ra09752a |
‡ Marc B. Taraban and Li Yu contributed equally to this work. |
This journal is © The Royal Society of Chemistry 2014 |