Characterization of oleic acid-coated iron oxide nanoparticles

The oleic acid-coated iron oxide nanoparticles were first observed by transmission electron microscopy (TEM). As shown in Fig. 1a and b, the iron oxide nanoparticles had a quasi-spherical morphology with a particle size ranging from 5 to 11 nm (average diameter: 7.4 nm, polydispersity index: 0.145). The images also evidenced that the nanoparticles were moderately monodisperse and tended to form aggregates. These results are in agreement with the features of oleic acid-coated iron oxide nanoparticles reported in previous works [30,31,32].

Fig. 1

Characterization of oleic acid-coated iron oxide nanoparticles: a) Representative transmission electron microscopy (TEM) images in bright field (BF) mode at different magnifications; b) Size distribution derived from TEM images; c) Chemical mapping using energy dispersive X-ray spectroscopy (EDS) in scanning transmission electron microscopy (STEM); d) Selected area electron diffraction (SAED) pattern according to raw data (blue). The obtained intensity profile is compared with a powder X-ray diffraction pattern calculated for the reference ICSD structure magnetite (ICSD 26410, orange); e) Thermogram obtained by thermogravimetric analysis (TGA), showing the percentage of weight loss (green) and its derivative (blue) at increasing temperatures; f) Comparison of iron oxide weight percentage in oleic acid-coated iron oxide nanoparticles as determined by the 1,10-phenanthroline-based colorimetric assay (blue) and by TGA (red). Results are represented as mean value ± standard error mean (p > 0.05); g) Magnetic properties measured by vibrating sample magnetometry (VSM). The figure on the right represents a zoom of the figure on the left at low specific magnetizations and low magnetic fields to evidence the hysteresis loop and determine remanence and coercivity. Remanence can be inferred from the specific magnetization intercept at zero magnetic field, whereas coercivity can be inferred from the magnetic field intercept at zero specific magnetization

Then, elemental analysis of the oleic acid-coated iron oxide nanoparticles was conducted through energy-dispersive X-ray spectroscopy (EDS) to further evidence the presence of iron and oxygen in the selected STEM image shown on the left part of Fig. 1c. The EDS data confirmed the presence of C, Cu, O and Fe. The C signal stems from the backbone of the oleic acid, the O signal stems from both the end groups of oleic acid and the oxygen atoms in iron oxide nanoparticles, and the Fe signal stems from the iron atoms in iron oxide nanoparticles. Furthermore, the C and Cu signal are attributable to the TEM grid. In this case, no La or Gd signals were observed as the UranyLess solution was only used to stain lipid samples (see Characterization of magnetic nanocapsules). Altogether, results showed that the elemental analysis was consistent with the expected composition of the oleic acid-coated iron oxide nanoparticles.

Selected area electron diffraction (SAED) was conducted to study the crystalline structure of the oleic acid-coated iron oxide nanoparticles [26, 33]. SAED is a crystallographic technique fundamentally analogous to X-ray diffraction but can serve to examine areas as small as just few hundreds of squared nanometers. The SAED pattern of the oleic acid-coated iron oxide nanoparticles is shown in Fig. 1d. The concentric ring-like diffraction pattern obtained is typical for suspensions of crystalline nanoparticles since the diffraction pattern is made of the 360-degree azimuthal superimposition of spots arising from Bragg reflection from individual nanoparticles. Six characteristic peaks at different crystallographic interplanar distances, namely 1.48760, 1.62175, 1.72528, 2.11733, 2.56123 and 2.99579 Ǟ were observed for the iron oxide nanoparticles. These interplanar distances correspond to the reciprocal radii of each concentric ring in the SAED pattern. This experimental SAED profile matches the position and relative intensity of the peaks of the powder X-ray diffraction pattern calculated for the reference ICSD 26410 structure (magnetite) [34]. In fact, the observed interplanar distances correspond, respectively, to the Miller indices (440), (511), (422), (400), (311) and (220), which define the inverse spinel structure of magnetite. Conversely, this profile does not match with the powder X-ray diffraction pattern calculated for the reference ICSD structure 15,840 (hematite), the other major iron oxide crystal type (Fig. S1).

To calculate the volume of marketed iron oxide nanoparticles to be added during the formulation of lipid nanocapsules, the content of iron oxide in oleic acid-coated iron oxide nanoparticles was first quantified by TGA analysis (Fig. 1e), which demonstrated that the weight percentage of iron oxide with respect to the total nanoparticles weight was 78.60 ± 3.62 according to the remaining final mass at the end of the heating up to 800ºC. These results were supported by the 1,10-phenanthroline-based colorimetric assay. Notably, since SAED diffraction pattern revealed that the nanoparticles consisted of the crystalline structure of magnetite (chemical formula Fe3O4), to calculate the concentration of iron oxide in the nanoparticles the extrapolated iron concentration from the colorimetric assay was divided by 0.72, which represents the weight percentage of iron in magnetite. The weight percentage of iron oxide in oleic acid-coated nanoparticles was 75.78 ± 8.02. Altogether, averaging colorimetric and TGA results, the oleic acid coating can be assumed to account for ≃ 23% of the total nanoparticle weight (Fig. 1f). Higher [32], analogous [35] and lower [26] weight percentages of oleic acid coating have been reported for other iron oxide nanoparticles. This coating is supposed to favor dispersibility of the iron oxide nanoparticles in organic solvents by preventing particle agglomeration and to enable iron oxide encapsulation within oily phases in nanocapsules.

The magnetic properties of oleic acid-coated magnetite nanoparticles were evaluated by vibrating sample magnetometry (VSM). The left part of Fig. 1g shows the complete magnetization versus magnetic field curve (M-H curve) at room temperature. The saturation magnetization of oleic acid-coated magnetite nanoparticles was 48.63 emu/g. This value is lower than the saturation magnetization of bulk magnetite (85–90 emu/g [36, 37]) but it is in line with the surface effect described for magnetite nanoparticles, wherein a higher proportion of the atoms are near the particle surface where the exchange field is lower, which ultimately accounts for a decrease in magnetization with a decrease in particle size [38,39,40]. Moreover, the presence of a nonmagnetic coating further decreases the saturation magnetization of iron oxide nanoparticles [41]. Altogether, this saturation magnetization is in the range reported for other oleic acid-coated iron oxide nanoparticles (40–60 emu/g) [9, 30, 42]. The right part of Fig. 1g also shows the M-H curve in the range of ± 10 Oe as an inset to observe the hysteresis loop. With a remanence of 0.48 emu/g and a coercivity of 6.10 Oe, the oleic acid-coated iron oxide nanoparticles behaved as soft ferrimagnets with nearly zero remanence and coercivity. Since the superparamagnetic limit for magnetite state has been estimated to be 25 nm [43], this indicates that, based on their size below this threshold, the thermal energy is enough to randomize the magnetic moments and the magnetite nanoparticles should therefore be in their superparamagnetic state [41]. The superparamagnetic state manifests as a magnetic property that arises in the presence of a magnetic field and disappears upon the removal of the magnetic field. This feature is significant for preventing the interaction of magnetite nanoparticles with iron in biological systems. The small remanent magnetization observed when the magnetic field applied is zero may be accounted for by the presence of a population of blocked nanoparticles due to interparticle interactions in the iron oxide aggregates observed by TEM, as hypothesized in [29]. Overall, the coercivity and remanence values were in agreement with those reported for analogously sized oleic acid-coated magnetite nanoparticles [12, 44].

Characterization of magnetic nanocapsules

Magnetic nanocapsules were prepared by the low-energy phase inversion temperature method, as previously described elsewhere [24], by further adding a suspension of the oleic acid-coated magnetite nanoparticles in the initial mixture. For biological assays, fluorescently labeled magnetic lipid nanocapsules encapsulating DiO were prepared for particle tracking purposes as indocarbocyanine dyes have been used in lipid-based nanocarriers due to their lipophilic nature and lack of premature release [29, 45, 46]. The characterization features of the resulting magnetic nanocapsules are summarized and compared with those obtained for the oleic acid-coated iron oxide nanoparticles in Table 1.

Table 1 Comparison of physicochemical characterization of oleic acid-coated iron oxide nanoparticles and magnetic nanocapsules

To study the hydrodynamic size distribution of the resulting nanocarriers, dynamic light scattering (DLS) measurements were conducted in water. According to the predictive univariate mathematical model defined in [24] for blank lipid nanocapsules prepared by this method, a hydrodynamic diameter of around 50 nm was to be expected for the theoretical initial 1.2 oil/surfactant (i.e., Labrafac WL1349/Kolliphor HS15) weight ratio used for the preparation of the formulation of magnetic lipid nanocapsules, as widely reported experimentally [21, 47,48,49]. However, as shown in Fig. 2a, after purification by centrifugation, magnetic nanocapsules showed an average hydrodynamic diameter of 256.7 ± 8.5 nm. Hence, in this case, and conversely to what had been previously reported [50, 51], the addition of magnetite nanoparticles greatly increased by approximately fivefold the particle size of their blank lipid nanocapsules counterparts (Fig. S2a). A plausible explanation could be the one given by Cui et al., who directly correlated the growing size of PLGA-based nanoparticles with increasing content of magnetic nanoparticles [9]. Regarding the polydispersity index of magnetic nanocapsules though, a highly monodisperse size distribution was obtained with a polydispersity index of 0.089 ± 0.0034, in agreement with results reported for non-magnetic lipid nanocapsules [48, 52,53,54]. Previous studies have reported both magnetic lipid-based [55, 56] and polymer-based [8, 17, 57] nanoparticles of 200–250 nm for magnetic targeting purposes to the BBB. Hence, the size distribution of magnetic nanocapsules was deemed to be adequate to benefit from magnetic targeting to the BBB.

Fig. 2

Characterization of magnetic nanocapsules: a) Representative intensity distribution profile (%) as a function of the hydrodynamic diameter (nm) measured by dynamic light scattering (DLS); b) Representative ζ-Potential (mV) distribution measured by electrophoretic light scattering (ELS); c) Representative transmission electron microscopy (TEM) images in bright field (BF) mode at different magnifications; d) Chemical mapping of the squared area in the left image of c) using energy dispersive X-ray spectroscopy (EDS) in bright field transmission electron microscopy (TEM): iron (green), oxygen (red), carbon (blue); e) Thermogram obtained by thermogravimetric analysis (TGA), showing the percentage of weight loss (green) and its derivative (blue) at increasing temperatures; f) Comparison of iron oxide weight percentage in oleic acid-coated iron oxide nanoparticles as determined by the 1,10-phenanthroline-based colorimetric assay (blue) and by TGA (red). Results are represented as mean value ± standard error mean (p > 0.05); g) Magnetic properties measured by vibrating sample magnetometry (VSM). The figure on the right represents a zoom of the figure on the left at low specific magnetizations and low magnetic fields to evidence the hysteresis loop and determine remanence and coercivity. Remanence can be inferred from the specific magnetization intercept at zero magnetic field, whereas coercivity can be inferred from the magnetic field intercept at zero specific magnetization

The ζ-potential of the dispersion of magnetic nanocapsules in water was measured by electrophoretic light scattering (ELS). As shown in Fig. 2b, magnetic nanocapsules had an average ζ-potential of—30.4 ± 0.3 mV and a ζ-deviation of 4.99 ± 0.30 mV. Slightly negative ζ-potentials had been reported previously for lipid nanocapsules, although the lower ζ-potentials in absolute value may be related to the ionic strength of the dispersion medium, which tends to decrease the ζ-potential [51]. However, the ζ-potential value reported herein correlates with the trend observed in [58], where it was observed that the higher the particle size of lipid nanocapsules, the higher the ζ-potential in absolute value: in particular, 200 nm-sized lipid nanocapsules had a ζ-potential of ≃—20 mV. The observed negative ζ-potential is expected to prevent non-specific adsorption of the lipid nanocapsules to the negatively charged cell membranes [59]. This can ultimately prevent nonspecific systemic uptake from occurring and increase the availability of nanocapsules for magnetic targeting to the BBB.

The ζ-deviation, although not frequently reported in literature, is a measure of dispersion of the ζ-potential distribution. The ζ-deviation value below 5 mV observed for the magnetic nanocapsules is consistent with a highly monodisperse ζ-potential distribution. Overall, the ζ-potential provides information on the electrostatic repulsive forces between the dispersed magnetic nanocapsules. The addition of magnetite nanoparticles did not modify the ζ-potential of their blank lipid nanocapsules counterparts (Fig. S2b), which aligns well with encapsulation of the oleic acid-coated iron oxide nanoparticles within the lipid core of the nanocapsules. A ζ-potential above |30| mV as in this case is often correlated with high colloid stability in the literature as per the electrostatic component of the DLVO theory [60]. Altogether, in terms of surface properties, the magnetic nanocapsules were foreseen suitable for magnetic targeting to the BBB.

Hydrodynamic diameter, polydispersity index and zeta potential of fluorescently labeled magnetic nanocapsules were analogous to the unlabeled magnetic nanocapsules (data not shown).

TEM imaging was performed to observe the morphology and structure of the magnetic nanocapsules. As shown in Fig. 2c, TEM images confirmed the spherical morphology of the lipid nanocapsules, which associated with the brighter regions. The particle size of the magnetic nanocapsules as observed by TEM ranged from 210 to 250 nm, slightly below the hydrodynamic diameter obtained by DLS. This trend was to be expected given that hydrodynamic diameter stems from solvated samples, whereas TEM imaging occurs on dry samples under vacuum conditions, which ultimately often leads to size obtained by DLS being bigger than that observed in TEM images [60]. Within the lipid core of the nanocapsules, electron-dense spots were observed forming grape-like aggregates. The particle size of these electron-dense spots ranged from 8 to 15 nm, which matches with the particle size observed for the oleic acid-coated magnetite nanoparticles under TEM imaging (Fig. 1a, b). Moreover, a significant increase in size was observed for nanocapsules that contained the dark spots in comparison with those devoid of them (as showed on the left image of Fig. 2c). However, artifacts can occur during TEM imaging upon staining with uranyl acetate. Hence, to prevent misinterpretation, elemental analysis was further conducted to verify the efficient encapsulation of the iron oxide nanoparticles within the core of the lipid nanocapsules. Then, elemental analysis through energy-dispersive X-ray spectroscopy (EDS) was conducted to evidence the location of the elements on the lipid nanocapsules in the selected area highlighted in green in Fig. 2c. As shown in Fig. 2d, the EDS analysis revealed the presence of C, O, Fe, Cu, La and Gd. The C signal stems from both the backbone of the polyethylene glycol (15)-hydroxystearate and the medium-chain triglycerides of caprylic and capric acids, the O signal stems from the ethylene oxide monomers of the polymer, from the glycerol moieties of triglycerides and from the oxygen atoms in magnetite, and the Fe signal stems from the iron atoms in magnetite. Moreover, the C and Cu signals are attributable to the TEM grid, whereas La and Gd signals are introduced by the staining UranyLess solution. The Fe peak on the dark particles in the TEM image confirmed that these electron-dense particles were iron-containing particles. Interestingly, the analysis revealed a homogeneous distribution of the C, O and Fe elements on a single nanocapsule. Altogether, the co-localization of Fe signal with the electron-dense particles in the core of the nanocapsules in the corresponding TEM image and with O and C signals in the elemental analysis confirmed the efficient encapsulation of iron oxide nanoparticles within the oily core of the lipid nanocapsules. Overall, this analysis demonstrates more strongly the encapsulation of iron oxide nanoparticles within lipid nanocapsules than the sole two previous studies that attempted to perform likewise [50, 51]. On the one hand, Moura et al. claimed to have iron oxide nanoparticles co-localized within the oily core of the lipid nanocapsules based solely on single and scarce electron-dense spots in TEM images [51]. Instead of using EDS analysis to demonstrate colocalization of Fe signal within the oily core of lipid nanocapsules, this analysis was only used to evidence a functionalization process through the presence of the S signal. On the other hand, Bohley et al. also relied on TEM images where the lipid nanocapsules themselves were entirely electron-dense instead of showing the grape-like morphology usually reported for iron oxide nanoparticles under TEM imaging [50]. Overall, this lack of efficient encapsulation may also account for the fact that neither of the authors did report a significant increase in particle size upon the addition of iron oxide nanoparticles as observed in this study. The encapsulation of oleic acid-coated magnetite nanoparticles within the oily core of lipid nanocapsules is likely a result of, on the one hand, hydrophobic interactions between the oleic acid coating stabilizing the magnetite nanoparticles and the medium chain triglycerides forming the core of the nanocapsules and, on the other hand, of the purification steps by centrifugation that enriches the samples in nanocapsules loaded with magnetite nanoparticles.

As reported for the oleic acid-coated iron oxide nanoparticles, the iron oxide weight percentage in the magnetic nanocapsules was quantified using both TGA and the 1,10-phenanthroline-based colorimetric assay. The weight percentage of iron oxide in magnetic nanocapsules determined by TGA analysis (Fig. 2e) was 12.77 ± 3.21 according to the remaining final mass at the end of the heating up to 800ºC. These results were supported by the colorimetric assay, which demonstrated that the weight percentage of iron oxide with respect to the total nanocapsules weight was 11.18 ± 6.45. Altogether, averaging colorimetric and TGA results, the iron oxide can be assumed to account for ≃ 12% of the total nanocapsules weight (Fig. 2f), which represents a 79.83% encapsulation yield with regards to the theoretical 15% initial iron oxide weight percentage. The increase of the organic component in magnetic nanocapsules in comparison with oleic acid-coated iron oxide nanoparticles from ≃ 23% to ≃ 88% is consistent with the increase in weight percentage due to the polymer shell and oily core included in the formulation of lipid nanocapsules. Indeed, in Fig. 2e, the weight loss curve (in green) and the corresponding derivative weight loss curve (in blue) show that the total weight of the nanocapsules decreases in two steps as the temperature increases due to the thermal decomposition of the different organic excipients (i.e., lipid and polymers), which have distinct decomposition temperatures. The first peak observed in the derivative weight loss curve in Fig. 2e at ≃ 240ºC can be ascribed to medium-chain triglycerides of caprylic and capric acids (Labrafac lipophile WL1349) according to the information provided by the supplier in its safety data sheet. The second peak in the derivative weight loss curve at ≃ 360ºC matches the thermal decomposition temperature reported for polyethylene glycol (15)-hydroxystearate (Kolliphor HS15) by the supplier. Above 360ºC and up to 800ºC there is no further weight loss since the polymeric and lipid excipients have already been thermally decomposed and remaining iron oxide does not degrade at these temperatures. No significant weight loss was observed at temperatures below 100ºC that could be due to water evaporation since samples had been freeze-dried prior to TGA analysis. Notably, the first thermal decomposition event was associated with a weight loss of about 75.8% of the total nanocapsules weight, whereas the second thermal decomposition event was associated with a weight loss of about 12.6%. This may explain why magnetic nanocapsules showed a bigger particle size than expected according to previous studies. In fact, TGA analysis seems to outline that after the purification steps by centrifugation only those nanocapsules loaded with iron oxide nanoparticles and with a bigger size sedimented. This accounts for the fact that despite a theoretical initial 1.2 Labrafac WL1349/Kolliphor HS15 weight ratio was used for the preparation of lipid nanocapsules, the final formulation was formed by a 6.0 mass ratio between both excipients. Applying the linear univariate mathematical model to predict nanocapsules sizes as a function of the Labrafac WL1349/Kolliphor HS15 weight ratio described elsewhere [24], volume diameter of nanocapsules is expected to be approximately 180 nm, which more closely matches the sizes experimentally reported herein. Altogether, the iron oxide weight percentage in the final magnetic nanocapsules is in line with or slightly above than that reported in other studies for nanocarriers prepared for magnetic targeting purposes [35, 55].

To authenticate the feasibility and sensitivity of the developed magnetic nanocapsules as stimuli-responsive nanocarriers, it is important to retain the magnetic properties of the oleic acid iron oxide nanoparticles after encapsulation. Accordingly, the magnetic properties of magnetic nanocapsules were also evaluated at room temperature by VSM. The left part of Fig. 2g shows the complete M-H curve. The saturation magnetization of magnetic nanocapsules was 5.84 emu/g. This value is lower than the saturation magnetization of oleic acid-coated iron oxide nanoparticles. Specifically, the specific saturation magnetization of the magnetic nanocapsules was 12.0% of that of the oleic acid-coated iron oxide nanoparticles, in full agreement with the iron oxide weight percentage in the final formulation as shown by both the colorimetric iron assay and TGA analysis. This reduction in magnetization is therefore a consequence of the weight percentage of nonmagnetic excipients included in the formulation, as also observed in [61]. For a nanocarrier to be suitable for magnetic targeting, high saturation magnetization is needed. Overall, the saturation magnetization of the magnetic nanocapsules is above the saturation magnetization values reported for other nanocarriers prepared for magnetic targeting purposes [6, 55, 56, 62, 63]. As a result, this saturation magnetization value was considered sufficient to provide magnetic targeting responsiveness. Furthermore, lipid nanocapsules become near fully saturated at relatively low magnetic fields (5,000 Oe, equivalent to 0.5 T). The right part of Fig. 2g also shows the M-H curve in the range of ± 10 Oe as an inset to observe the hysteresis loop. Magnetic nanocapsules had a coercivity of 6.60 Oe and a remanence of 0.12 emu/g. In comparison with the values observed for oleic acid-coated iron oxide nanoparticles, the low coercivity value was maintained whereas specific saturation remanence was decreased by fourfold for magnetic nanocapsules, which may be due to the reduced occurrence of iron oxide aggregates upon encapsulation within lipid nanocapsules, as also observed in [29]. Altogether, both near zero coercivity and remanence values contribute to the superparamagnetic-like state of the magnetic nanocapsules. Indeed, Azarmi et al., with higher remanence values for their iron oxide nanoparticles than those reported herein, claimed their particles to be superparamagnetic [12]. Hence, even if the drop in saturation magnetization upon encapsulation of iron oxide nanocapsules in lipid nanocapsules might have been regarded as a caveat to respond to an external magnetic field for magnetic targeting, this encapsulation may likewise help prevent the iron oxide nanoparticles from aggregating contributing thereby to a more superparamagnetic-like behavior.

Interaction of magnetic nanocapsules with the human cerebral microvascular endothelial hCMEC/D3 cell line

First, the biocompatibility of the magnetic nanocapsules was tested on the human cerebral endothelial cell line hCMEC/D3 at different concentrations following 24 h (Fig. 3a) and 72 h (Fig. S3a) of treatment. The WST-1 assay was chosen to infer the cytotoxicity profile from changes in cellular metabolic activity. Notably, magnetic nanocapsules did not reduce the metabolic activity of hCMEC/D3 cells at any of the concentrations tested ranging from 10 to 500 µg/mL after 24 h (Fig. 3a). These data may seem controversial in comparison with a recent study on the effect of blank lipid nanocapsules on metabolic activity of this cell line [51]. In that previous study, a significant reduction in cell viability for the highest concentrations of blank lipid nanocapsules was evidenced after 24 h and even after only 4 h treatment. However, the concentration range tested there, when expressed in the same units as those utilized herein, correspond to ≃ 450 µg/mL-≃ 1750 µg/mL. Hence, the highest concentration evaluated in this study is in the same order than the lowest tested in the previous study, which may well account for the higher cytotoxicity observed there. The viability data obtained for the magnetic nanocapsules also aligns well with the results observed for smaller sized blank lipid nanocapsules following 24 h incubation [