3.1. Self-Assembly by Film Hydration under Orbital Agitation Versus Magnetic Stirring
First assays were performed with PEG
45PLA
69 (2000:5000 KDa) before testing other copolymers. The results indicated that film hydration by agitation in a rotary-evaporator resulted in the presence of large polymeric bulk material (
Figure S1A) and bimodal distribution with high polydispersity index (PDI) values (
Table 2) even after sonication. The orbital process does not exert a direct shear force on the hydration liquid and, therefore, leads to detachment of large slices of the film from the surface of the glass flask. This bulky polymer film may remain in the solution and coexist with vesicles and/or micelles, resulting in large polydispersity and/or bimodal size distribution [
11,
12]. Although literature reports that bulk polymeric film may coexist with PL and other aggregates [
13], these bulk polymeric films sediment out of the solution in dilute conditions in which intermolecular interactions between the vesicles have no significant role in maintaining the bulk film dispersed in solution. Those systems may be interesting only for individual assembly characterization and are not adequate for drug encapsulation due to heterogeneity and potential loss of the material upon sedimentation [
14].
To avoid the presence of bulky polymer after film hydration, magnetic stirring overnight was employed. The macroscopic difference in the system was remarkable and sedimentation was not observed (
Figure S1B). Magnetic stirring improved the copolymers self-assembly since the shear force leads to detachment of the unillamelar polymeric film from the flask surface and favors film breaking into smaller vesicular structures [
15]. Diffusion of the aqueous phase occurs on a larger surface area, which in turn favors the hydrophobic effect and polymeric vesicle formation. In contrast, orbital agitation with lower shear forces allowed only larger bulk polymeric film pieces to detach, which leads to limited aqueous phase penetration into the film. TEM was not performed for these systems since broad size distributions were observed. Nonetheless, either PL or other complex self-assembled nanostructures were formed.
Following, we investigated the influence of polymer MW and hydrophilic fraction (
f) on PL polydispersity.
Table S1 shows the results for size distribution by intensity and number based on dynamic light scattering (DLS) measurements after overnight hydration by magnetic stirring. As already mentioned, wide size distributions are inherent to the self-assembly since this technique is not precise in producing nanostructures with a good control of size [
8]. One of the reasons for this is that the energetic penalty involved in amphiphiles self-assembly into vesicles from a membrane with zero natural net curvature results from the sum of the mean and Gaussian curvature and not from the radius [
5].
Even with lower visual polymeric bulk film sedimentation, the DLS indicated a broad distribution of size with high PDI values. Cumulants fit were not used to analyze the z-average hydrodynamic radius, instead the scattering profile by number distribution was presented in
Table S1 in order to avoid over representation of large objects that we believe are part of smaller populations of remaining non-spherical and irregular aggregates (
Figure 1). The presence of these large structures even in lesser proportions causes significant light scattering and occupy a larger volume in the sample as compared to the smaller structures that might constitute the major population by number distribution in the heterogenous system.
According to the ANOVA, the PDI values for the three copolymers are statistically different (
p > 0.05) and Tukey test confirmed that PDI for PEG
45PLA
69 is different from the other two copolymers. The correlation function (
Figure 2A) confirmed this result, showing difference in the decay times: nearly 180 µs for PEG
45-PLA
69 and 100 µs for PEG
114-PLA
153 and PEG
114-PLA
180. The correlation function was widely discussed by Bhattacharjee (2016) [
16] and is a factor that indicates clear differences among samples. The autocorrelation decays more slowly for bigger particles due to slower relaxation of the particle dynamics.
A tail was identified in the correlation function for all copolymers emphasizing that all systems contained large particles that contributed to high PDI. Usually high PDI values in PL systems might be attributed to the presence of micelles, as reported by Bartenstein [
11], but in our study we attributed these to larger structures as no sign of micelles was observed. In addition, although ANOVA indicated statically different PDI values, these values were higher than ~0.4 for the three copolymers indicating that all the samples still have a very broad size distribution (a monodisperse system would have a PDI < 0.3 [
11].
These results indicated that, from the microscopic point of view, film hydration conditions were still unsatisfactory and even though higher shear of stirring improved the PDI values, other efforts would be necessary to achieve more homogenous PL preparations.
3.2. Self-Assembly by Film Hydration under Magnetic Stirring and Sonication
In order to control size distribution, sonication was considered as a next step and the DLS results are presented in
Table S2. Literature shows that sonication reduces aggregates size due to cavitation, which consists in the oscillation of small gas bubbles by expansion and contraction in a liquid exposed to acoustic pressure waves. The gas bubbles eventually collapse resulting in high pressures and this stress breaks up large vesicle aggregates into small vesicles [
17].
The correlation coefficient (
Figure 2B–D), however, was slightly different after sonication, with smoother curve fits for all sonicated preparations, except PEG
114-PLA
180. This could correspond to the narrower size distribution for the smoother curves. Decaying time remained approximately 180 µs for PEG
45PLA
69 and 100 µs for PEG
114-PLA
153 and PEG
114-PLA
180 after sonication steps, as well as tails in the correlograms. The difference in the intercepts was not considered in this study since the concentration of the samples submitted to DLS was not controlled.
For PEG-PLA PL, no significant differences in PDI values were observed after sonication (
Figure 3) neither between the different times of sonication (20 min and 50 min).
The TEM images of the systems obtained under overnight magnetic stirring and sonication by 50 min are shown in
Figure 4. Typical morphology of vesicular structures was confirmed and the measurements done with the Image J
® program indicated mean diameters about 226 nm, 94 nm and 133 nm with membranes thickness (t) of 8, 10.5 and 14 nm for PEG
45-PLA
69, PEG
114-PLA
153 and PEG
114-PLA
180, respectively.
Polymersomes were found to be smaller from TEM analysis as compared to DLS, which is expected since DLS estimates hydrodynamic diameter and, consequently, the hydrophilic corona of hydrated PEG or brush thickness (d) is taken into account which leads to overestimation of the size (there are on average three water molecules per ethylene oxide unit) [
18].
The TEM imaging allowed visualizing the core of the vesicles and its membrane (
Figure 4, (t) -Bottom right). The preparation of TEM grids with phosphotungstic acid stain leads to negative contrast images, however the darker color of the PL membrane can be explained by the fact that phosphotungstic acid could react more strongly with the ester groups of PLA, which in this case creates positive contrast-like images for the hydrophilic corona of the PL [
18].
To overcome broad distribution (high PDI values), different times of magnetic stirring at 400 rpm were investigated, namely 24, 48 and 72 h of magnetic stirring (400 rpm) at room temperature or at 40 °C. Results are presented in
Figure 5 and show that stirring for 24 h results in PDI values smaller than overnight (15 h) agitation. For PEG
45-PLA
69 and PEG
114-PLA
153, a decrease in PDI was observed altogether with a visual enhance in homogeneity. Although the broad size distributions and the presence of more than one size peak (
Table S3–S5) was still the case, the longer stirring time facilitated hydration of the copolymer initially broken from the network of bulk film into smaller, more uniform structures. The hydrophobic block of PEG
114-PLA
180 has the highest molecular weight and glass-transition temperature among the three copolymers, consequently hydration of this glassy copolymer in aqueous solution even at longer times might result in the dispersion of bulk polymer film rather than in the formation of defined spherical unilamellar PL.
According to our results, stirring times higher than 24 h, did not further improve the broad size distributions and the presence of more than one size population for these systems remained. The ANOVA pointed no statistically significant difference for PDI values of PEG
45-PLA
69 and PEG
114-PLA
153 for the stirring times of 24, 48 and 72 h. For the PEG
114-PLA
180, a significantly lower PDI value (
p < 0.05) was observed after 24h of stirring at 40
°C, perhaps because of higher copolymer glass transition temperature.
Figure 6 shows the correlation coefficients and one can see slight differences of decay time as well as improved curve smoothness for PEG
114-PLA
180 indicating more homogenous samples.
Tables S3, S4 and S5 show the presence of peaks corresponding to higher sizes.
We further considered the effect of temperature on self-assembly by comparing RT (20 °C) and 40 °C. Our results show that increasing the temperature did not significantly influence the vesicles self-assembly, since similar PDI values were obtained (
Figure 5). We did not investigate temperatures higher than 40 °C since this might lead to protein denaturation. Moreover, PEG interacts with water by hydrogen bond and the interactions are weaker at higher temperatures due to the increased kinetic energy of water molecules. As a result, PEG moieties are dehydrated by temperature increase, favoring van der Waals interactions among PEG chains with the copolymer precipitation (hydrophobic effect).
Challenges for self-assembly of copolymers into PL by film hydration method can be understood through the studies of Battaglia and Ryan (2006) [
19,
20] regarding polymeric vesicle formation. Accordingly, hydration condition may lead to slow self-assembly by triggering incomplete formation of vesicles with initial finger instabilities (myelin) that are formed after water addition and copolymer swelling. These finger instabilities can grow and form vesicles owing to the copolymer diffusion in water and the water diffusion in the copolymer [
19].
3.3. Effect of Centrifugation and Extrusion Post-Film Hydration
Based on the results above, 24 h stirring time at room temperature was chose as the preferred condition for the following experiments. Further efforts to reduce PDI and, consequently, obtain a narrower size distribution focused on fractionation of the populations of nanostructures by centrifugation. This technique can separate PLs and bulk polymer film structures.
Figure 7 shows that centrifugation resulted in PDI values of 0.363 and 0.280 for PEG
114-PLA
153 and PEG
114-PLA
180, respectively and decay times around 48 µs. Centrifugation was fast, simple and resulted in successful separation of the polydisperse aggregates/material. Unfortunately, DLS analysis could not be performed for PEG
45-PLA
69 because centrifugation resulted in very diluted samples due to the loss of significant material in the bulk polymeric film fraction.
Similarly, extrusion as a post-hydration method resulted in the loss of material and dilution of PL samples, confirmed by TEM images (
Figure S2). To evaluate the extrusion effect, we increased the initial PEG-PLA concentration to 0.1%. Moreover, we used nanoparticle tracking analysis (NTA) to complement DLS studies (
Figure S3). For PEG
114-PLA
153, the mean diameter values based on NTA are in agreement with the values obtained by DLS. However, significant populations of different sizes are observed by NTA for the other two copolymers. One should keep in mind that DLS is more adequate than NTA for the determination of nanostructures hydrodynamic diameter. Nonetheless, NTA allows an estimation of the concentration of PL [
21] and our results clearly show that the PEG
45-PLA
69 system after the extrusion resulted in more diluted systems in terms of number of polymersomes due to the initial loss of polymeric material in the bulk fraction after the film hydration, what was already visually observed. The PLs prepared with higher concentration of copolymer (0.1% (
m/
v)) (
Figure 8) resulted in narrower size distribution and PDI values of 0.345, 0.144 and 0.081 for PEG
45-PLA
69, PEG
114-PLA
153 and PEG
114-PLA
180, respectively. However, for PEG
45-PLA
69, a tail in the distribution profile and correlogram was still observed and attributed to dust traces, which could be removed by filtration before the measurement. Here we decide not to filter the samples since it usually dilutes PL systems and may affect the vesicle shape, size or concentration (
Figure S4).
Both centrifugation and extrusion lowered the PDI values, however the mechanisms are different. The first one is a purification step with the separation of fractions by size and density [
22]. In extrusion, on the other hand, larger aggregates either deform or break up, and reassemble [
10]. We demonstrate that, although centrifugation and extrusion were reported as steps that generally reduce the polydispersity of PL systems, the presence of excess of bulky polymeric film leads to low concentrations of vesicles and is a limiting factor for taking this method any further. It is important to highlight that usually, for copolymers having a glassy amorphous component, extrusion should be performed above the glass transition temperature, as it was done for PEG
45-PLA
69 (Tg = 23 °C) and PEG
114-PLA
153 (Tg = 39 °C). Considering PL for protein-based drug delivery, preparation methods at temperatures higher than 40 °C were considered unsuitable due to very likely denaturation effect on the protein, therefore, extrusion was performed at 40 °C for PEG
114PLA
180 (Tg = 40 °C). Nonetheless our DLS results confirmed that this temperature was sufficient to avoid PL deformation in the extrusion procedure.
To sum up all the conditions investigated,
Scheme 1 presents the main challenges for PLs preparation from PEG-PLA copolymers self-aggregation.