Magnetically Induced Brownian Motion of Iron Oxide Nanocages in Alternating Magnetic Fields and Their Application for Efficient siRNA Delivery

Abstract

Hyperthermia of superparamagnetic nanoparticles driven by Néel relaxation in an alternating magnetic field (AMF) has been studied in biomedical areas; however, Brownian motion, induced by another magnetic relaxation mechanism, has not been explored extensively despite its potential in intracellular mechanoresponsive applications. We investigated whether superparamagnetic cage-shaped iron oxide nanoparticles (IO-nanocages), previously demonstrated to carry payloads inside their cavities for drug delivery, can generate Brownian motion by tuning the nanoparticle size at 335 kHz AMF frequency. The motivation of this work is to examine the magnetically driven Brownian motion for the delivery of nanoparticles allowing escape from endosomes before digestion in lysosomes and efficient delivery of siRNA cargoes to the cytoplasm. superconducting quantum interference device (SQUID) measurements reveal the nanocage size dependence of Brownian relaxation, and a magnetic Brownian motion of 20 nm IO-nanocages improved the efficiency of siRNA delivery while endosomal membranes were observed to be compromised to release IO-nanocages in AMFs during the delivery process.

Graphical Abstract

While nanoparticle carriers have been used widely to deliver therapeutic RNAs for gene therapy, overcoming endosomal escape to enhance transfection efficiency remains a major challenge.^1–6 Most RNA carriers are trapped in endosomes, fused with lysosomes, and degraded before they can be released into the cytoplasm, which is the major reason for low transfection efficiency.^7,8 While various strategies such as rupture, pore formation, and fusion of endosomal membranes^9,10 via inflow of protons, chloride ions, and water in endosomes were developed for the endosomal escape of nanoparticle-based drug carriers,^9,11–13 balancing the tradeoff between endosomal escape efficiency and cytotoxicity is still difficult to accomplish.^14–16 Lipids have been the gold standard for transfection reagents, and it has been reported that lipid nanoparticles could alter intracellular transport along the cytoskeleton and avoid lysosomal degradation due to their random Brownian movement.¹⁷ Thus, we developed the concept that an efficient endosomal escape of nanoparticle carriers and an improvement in the efficiency of therapeutic reagent delivery could be achieved if their Brownian motion and diffusion could be externally optimized to release them to the cytoplasm.

Superparamagnetic iron oxide nanoparticles (SPIONs) can undergo two types of magnetic relaxations in alternating magnetic fields (AMFs). At a certain range of frequencies of AMFs along with specific sizes of SPIONs, the phase difference of the oscillating magnetic moment with respect to the field direction of the AMF can be dissipated by either Néel relaxation or Brownian relaxation.^18,19 In the Néel mechanism, the magnetic moment of the nanoparticle oscillates rapidly in attempting to match the frequency of the AMF, resulting in an internal dissipation of heat through the crystal lattice of the particle. In the Brownian mechanism, the entire particle physically rotates to align with the magnetic moment with the field direction as SPIONs catch up with the frequency of the AMF, dependent on the friction and viscosity of the surrounding medium (i.e., magnetically induced Brownian motion).¹⁸ Whether the Brownian or Néel mechanism is more dominant is dependent on various factors, including the nanoparticle size and the frequency of the AMF. In this report, we investigated whether superparamagnetic cage-shaped iron oxide nanoparticles (IO-nanocages), previously demonstrated to carry payloads inside the cavity for therapeutic molecular delivery, can be controlled to undergo magnetically induced Brownian motion, dependent on size, at the conventional AMF frequency of 335 kHz. The combination of SQUID (superconducting quantum interference device) measurements and computational simulations of the magnetic relaxation time for superparamagnetic IO-nanocages reveals their size dependence on these two relaxation modes. For example, Brownian relaxation (i.e., induction of rotational motion) is dominant for IO-nanocages in the size range of >17 nm, while Néel relaxation (i.e., heat generation) becomes the dominant mechanism for IO-nanocages smaller than 17 nm.

To examine the hypothesis that the Brownian motion of superparamagnetic IO-nanocages triggered by AMFs can influence siRNA delivery with increased endosomal escape, we studied the internalization profile of IO-nanocages, the delivery efficiency of firefly luciferase siRNA, and the endosomal escape of IO-nanocages in AMFs using luciferase-expressing reporter B16-F10 cells. When cells were treated with the firefly luciferase siRNA-loaded IO-nanocages and the AMF was applied at a frequency of 335 kHz, the siRNA delivery efficiency by IO-nanocages with dominant Brownian motion was improved to 51% while minimal siRNA delivery was detected by the same delivery protocol without the AMF. The comparison of siRNA delivery efficiencies with IO-nanocages in the Brownian relaxation mode and the Néel relaxation mode showed that IO-nanocages in the Brownian mode can deliver 5-fold higher siRNAs. It should also be noted that this AMF treatment did not affect cell viability. Consistently, under the same experimental conditions to trigger the Brownian mode, endosomal membranes of cells were observed by transmission electron microscopy (TEM) and fluorescence (FL) images of a calcein leakage assay to be compromised to release IO-nanocages to the cytoplasm.

Thus, here we introduce a new strategy to improve the efficiency of siRNA delivery with the efficient magnetic endosomal escape of nanocarriers. Although previously various enhanced gene delivery approaches with SPIONs have been reported using their characteristic properties such as hyperthermia via enhanced photothermal conversion and magnetically guided cellular uptake,^20,21 the novelty of this approach is to apply the magnetic Brownian motion of IO-nanocage carriers in AMFs for diffusing siRNAs to the cytoplasm, which to our knowledge has not previously been explored. Advantages of this approach include minimal heat generation by IO-nanocages in AMFs, reducing the possibility of degradation of therapeutic siRNA cargoes in addition to enhanced endosomal escape.

MAGNETIC PROPERTY OF IO-NANOCAGES: CONTRIBUTION OF BROWNIAN AND NÉEL MECHANISM IN THE ALTERNATING MAGNETIC FIELD

Magnetic nanoparticles (MNPs) below 100 nm can be treated as a single magnetic dipole whose response to AMFs occurs primarily through two mechanisms referred to as Néel and Brownian relaxations.^22,23 The relative strength of each mechanism can be evaluated by a comparison of the relaxation time τ, the time taken for the direction of magnetic moment of SPIONs to reach equilibrium with the external magnetic field after each reversal of the field polarity. Néel relaxation is affected by the movement of magnetic moments of single-domain MNPs in AMFs,²⁴ and the relaxation time of the Néel mechanism depends on the magnetic anisotropy and magnetic volume of SPIONs.²⁵ In contrast, Brownian relaxation involves the movement of MNPs relative to the surrounding medium²² and the relaxation time of Brownian mechanism depends on the hydrodynamic volume and the viscosity of the surrounding fluid of SPIONs.²⁵ Since the dominance of each mechanism depends on how fast the magnetic moment is realigned, the magnitude of dominance of each mode can be evaluated by the breakdown of τ to Brownian and Néel contributions (i.e., the shorter the relaxation time, the more dominant the relaxation mode is; see the section II in the Supporting Information for more information).

A determination of the anisotropic constant K is necessary to break down the relaxation time τ in the Néel mechanism for IO-nanocages, as shown in the section II in the Supporting Information. After the superparamagnetism of IO-nanocages was confirmed by measuring the DC magnetic saturation M_s and coercivity H_c (Figure S1), the AC susceptibility was measured by varying the temperature in SQUID (Figure 1a). T_f is the temperature at which the imaginary component of the AC susceptibility is at a maximum. T_f values from three different AMF frequencies of 10, 100, and 300 Hz shown in Figure 1a were used to extrapolate an anisotropic constant K^26,27 from the linear fitting of an Arrhenius plot of reciprocal values of AMF frequencies and T_f for 23 ± 1.0 nm IO-nanocages (Figure 1b), which is a mean DLS size value of IO-nanocages we typically synthesized (equations for the fitting are given in section II in the Supporting Information). The anisotropic constant K obtained from the fitting of an Arrhenius plot shown in Figure 1b is 1.36 × 10⁴ J/m³, which is within the characteristic range of anisotropic constants of magnetic nanoparticles.²⁸ The magnetic volume V_M is assumed to be equal to hydrodynamic volume of V_H. We estimated V_H as $\frac{\pi }{6}{d}^3$, where d is obtained by hydrodynamic diameter of IO-nanocage from dynamic light scattering (DLS) measurements. With these values and K = 1.36 × 10⁴ J/m³ from Figure 1b, the total relaxation time with respect to Brownian and Néel mechanisms was simulated as shown in Figure 1c as the K value is extrapolated at the AMF frequency of 335 kHz. In this figure, the total relaxation time is dominated by Brownian relaxation for a size of IO-nanocages larger than 17 nm (i.e., where the dotted black line overlays the green line in Figure 1c), while the Néel relaxation is predominant for IO-nanocages smaller than 17 nm (i.e., where the dotted black line overlays the red line in Figure 1c). As shown in Figure 1d, experimental specific absorption rates (SARs) from the measurement of heat absorption are directly compared with theoretical SARs with the size of IO-nanocages. In this figure, the experimental SARs and theoretical SARs derived from Figure 1c agree well throughout all particle sizes, which also supports the high reliability of relaxation and energy simulations based on SQUID measurements. Details of the magnetic and SAR simulations with equations based on experimental measurements related to the breakdown of Brownian versus Néel relaxation time are described in section II in the Supporting Information. From SAR values, the magnitudes of Brownian motions of IO-nanocages of 15 nm (in the Néel regime) and 20 nm (in the Brownian regime) can also be calculated and compared using the linear response theory (LRT; see section VI in the Supporting Information for details). Under the assumption that these magnitudes are proportional to the power dissipation of IO-nanocages under AMFs, when we calculate SARs from magnetic Brownian motion using the LRT with eqs 3, 4, and 7 in section II in the Supporting Information, the power dissipation from Brownian motion for a 20 nm IO-nanocage is estimated to be 72.1 W/g while the power dissipation for a 15 nm IO-nanocage corresponds to 0.04 W/g. Based on these values, the Brownian torque of 20 nm IO-nanocages is 4 orders of magnitude larger than that for 15 nm, consistent with the 15 nm IO-nanocage being in the Néel-dominant region while the 20 nm IO-nanocage falls into the Brownian-dominant region as shown in Figure 1c.

Figure 1.

(a) Measurements of the imaginary component of AC susceptibility for nanocages at a size of 23 nm with three AMF frequencies, f. From these plots, T_f values for 10, 100, and 300 Hz at the maximum AC susceptibility are obtained as 190, 208, and 210 K, respectively. (b) Arrhenius plot of f⁻¹ versus $T_f^{-1}$ to extrapolate a value of the anisotropic constant K for the nanocages from a linear fitting.^26,27 The obtained K value is used to simulate the contribution of Neé;el versus Brownian relaxations in total magnetic relaxation time in (c). (c) Simulated relaxation times of Brownian mechanism τ_B (green line) and Neéel mechanism τ_N (red line) with respect to the total relaxation time τ (black-dashed line) with the size of the IO-nanocage. The dotted gray horizontal line for f = 335 kHz corresponds to the maximum energy absorption where the magnetic moment and the AMF are completely out of phase. (d) Simulated specific absorption rates (SARs) based on the linear response theory (LRT) (blue line) and experimental SARs (orange cross-bar)s with the size of IO-nanocages.

STRUCTURAL ANALYSIS OF SIRNA-LOADED IO-NANOCAGES

IO-nanocages were coated with 3-(3,4-dihydroxyphenyl)-propionic acid (DHCA) to make them water-soluble before loading with siRNAs.²⁹ The structure of the siRNA-loaded IO-nanocages was analyzed by TEM, dynamic light scattering (DLS), UV/vis absorption, and ζ potential. The TEM image of IO-nanocages shown in Figure 2a indicates the high yield in monodisperse size distribution with the well-defined uniform hollow structure in the cage shape. The size of IO-nanocages can be adjusted by changing the concentration of iron ions used for the galvanic replacement reaction, replacing manganese atoms in the template manganese oxide nanocubes by iron atoms.²⁹ Based on the UV/vis absorbance at 260 nm of RNAs in the supernatant of IO-nanocage solutions after 12 h of siRNA incubation, the average number of siRNAs per IO-nanocage (20 ± 0.8 nm) is estimated to be 45 molecules. The average value of the hydrodynamic diameter of the 20 nm IO-nanocages is increased to 23.9 nm with a polydispersity index (PDI) of 0.143 after the siRNA incorporation and DHCA capping. This low PDI value indicates that IO-nanocages are well dispersed in aqueous media, and it is consistent with the negative ζ potential of these IO-nanocages (−15.3 ± 3.2 mV), which could contribute to reducing the agglomerate via electrostatic repulsion. The diffusion of siRNAs into the cavity of IO-nanocages is considered to be driven by the charge gradient of the nanoparticle surface.³⁰ Due to the steric hindrance of the cage shape, negatively charged carboxyl capping groups of DHCA are expected to be distributed in higher density on the outside of nanocages with respect to the inside of the cavity, which could create a charge gradient on the surfaces of IO-nanocages (i.e., more neutral toward the core). Thus, this charge gradient with the aid of capillary force is highly likely to be a driving force for the loading of negatively charged siRNA inside DHCA-coated IO-nanocages.

Figure 2.

(a) Transmission electron microscope (TEM) image of IO-nanocages. Inset: high-resolution TEM image of an IO-nanocage. (b) Confocal image of luciferase-expressing B16-F10 melanoma cells after IO-nanocages were incubated for 24 h. (red, Cy5-labeled IO-nanocages; green, GFP; blue, DAPI). These cells were fixed and permeabilized before imaging. Scale bar: 100 μm. (c) Quantitative analysis of IO-nanocage uptake by B16-F10 cells, based on confocal images taken at different time points. N = 3. Confocal images for 3, 6, 12, 18 h time points are also available in section IV in the Supporting Information (Figure S3).

EFFECT OF BROWNIAN MOTION OF IO-NANOCAGES FOR SIRNA DELIVERY IN ALTERNATING MAGNETIC FIELDS (AMFS)

As a proof of principle, we loaded firefly luciferase siRNAs in the magnetic IO-nanocage carrier to examine the influence of AMFs on siRNA delivery efficiency in luciferase-expressing B16-F10 cells. Twenty nanometer IO-nanocages were chosen as a siRNA carrier for this study, since this size range in an AMF at a frequency of 335 kHz is in the Brownian relaxation regime close to the highest energy absorption point, as shown in Figure 1c. First, an in vitro uptake profile of IO-nanocages was investigated using IO-nanocages labeled with Cy5. The uptake of Cy5-labeled IO-nanocages in B16-F10 cells was quantified by confocal microscopy at 3–24 h after the IO-nanocage injection (e.g., Figure 2b). As shown in Figure 2c, the fluorescence intensity of IO-nanocage uptake in confocal images was averaged for 50 cells with N = 3 and the change in intensity with time plotted. In this figure, the internalization by IO-nanocages in the cells was saturated at 18 h of incubation, consistent with other iron oxide based nanoparticles whose internalization and saturation times were in the range of 4–24 h,^31–33 and thus in the next step of the bioluminescence experiment the AMF was applied after 18 h of siRNA-loaded IO-nanocage treatment with the luciferase-expressing B16-F10 cells.

Figure 3a,b summarizes the bioluminescence assays of siRNA delivery to the reporter B16-F10 cells with and without AMFs via the 20 nm IO-nanocage carrier, which is in the size range that exhibits a dominant Brownian relaxation mechanism at a frequency of 335 kHz of the AMF (Figure 1c). In this experiment, cells were incubated with firefly luciferase siRNA loaded IO-nanocages for 18 h and an AMF (f = 335 kHz) was applied for 5 min inside the magnetic coil (diameter 8 cm). These samples were further incubated for 6 h to allow the post-transcription of siRNA to interfere after the release (i.e., total incubation time 24 h). According to Figure 3a, without the AMF, all of the neat cells (control), neat IO-nanocages, nonspecific siRNA-loaded IO-nanocages (DLS size 26.35 nm (PDI 0.146), ζ potential −16.5 ± 1.7 mV), and firefly luciferase siRNA loaded IO-nanocages had no significant reduction of luciferase expression in B16-F10 cells. This observation indicates that IO-nanocages do not release siRNAs in the cytoplasm even after 24 h in the absence of an AMF. The effect of an AMF on these groups in B16-F10 cells is shown in Figure 3b. As shown in Figure 3b, the luciferase expression showed no suppression for all samples except the experimental group treated with firefly luciferase siRNA loaded IO-nanocages. The luciferase expression level of B16-F10 cells in this group was significantly reduced to 48.9 ± 2.7%. Based on the ANOVA single factor t test, the P value of firefly luciferase siRNA loaded IO-nanocages is elucidated to be P < 0.005 with respect to the control (neat cells). Thus, this in vitro experiment showed that the magnetic Brownian motion of IO-nanocages improved siRNA delivery efficiency to 51% with respect to the same protocol without applying AMF (Figure 3a,b), based on the reduction of bioluminescence in the B16-F10 cells. It should be noted that, when various iron oxide based nanoparticles were previously used to deliver firefly luciferase siRNAs, the typical silencing efficiency of luciferase expression has been reported to be in the range of 40–60%.^34–37 Because no quenching of luciferase expression was observed when firefly luciferase siRNAs were delivered by IO-nanocages in the absence of an AMF (Figure 3a), 51% luciferase expression quenching after AMF application is considered to be a significant reduction.

Figure 3.

In vitro bioluminescence assay of siRNA delivery to luciferase-expressing B16-F10 cells in the (a) absence and (b) presence of the AMF. In (b), the reduction of green emission following the treatment of siRNA-loaded IO-nanocages and application of an AMF at 335 kHz indicates that siRNAs are delivered to silence luciferase expression significantly in B16-F10 cells when siRNA is delivered by IO-nanocages in the AMF. **P < 0.005, N = 4. (c) Comparison of cell proliferation of all groups in (b) under the AMF, measured by an MTT assay. N = 4. A positive control in (c) was performed by inducing cell death with etoposide.

To further examine our hypothesis that the Brownian motion of an IO-nanocage at a 335 kHz AMF frequency plays a role in the transfection efficiency, we performed a siRNA delivery experiment with B16-F10 cells using 15 nm IO-nanocages in which the Néel relaxation mode dominates at 335 kHz (Figure 4). As expected, 15 nm IO-nanocages (15 ± 0.4 nm diameter) with minimal Brownian contribution (Figure 1c) in the AMF delivered only trace amounts of siRNAs throughout all samples, as shown in Figure 4 (black bars in Figure 4), while 20 nm IO-nanocages in the Brownian relaxation mode could reduce bioluminescence about 5 times better than that of 15 nm IO-nanocages (e.g., 11% bioluminescence reduction by using 15 nm IO-nanocages vs 51% bioluminescence reduction by using 20 nm IO-nanocages in the silencing siRNA + IO-nanocage experimental group as shown in Figure 4) due to a higher concentration of siRNAs delivered to the cytoplasm of B16-F10 cells in the presence of an AMF. Thus, this comparison suggests that Brownian motion of IO-nanocages plays a significant role in effective siRNA delivery.

Figure 4.

In vitro bioluminescence assay of siRNA delivery to luciferase-expressing B16-F10 cells by a 15 nm IO-nanocage versus a 20 nm IO-nanocage. Brownian relaxation is more dominant for the 20 nm IO-nanocage, while Neé;el relaxation is more dominant for the 15 nm IO-nanocage in the AMF (f = 335 kHz), based on Figure 1c. **P < 0.005, N = 4.

Furthermore, the cytotoxicity of the AMF treatment was evaluated by performing an MTT assay. As shown in Figure 3c, the percentages of cell proliferation in the three experimental groups neat IO-nanocages, nonspecific siRNA-loaded IO-nanocages, and firefly luciferase siRNA loaded IO-nanocages were all nearly 100% when the AMF was applied. An MTT assay on a positive control (cells killed by etoposide (100 μg/mL)) shown in Figure 3c detected only a trace amount of proliferation, supporting the validity of the assay. Therefore, neither IO-nanocages nor firefly luciferase siRNAs influence the cell proliferation by applying the AMF, and thus the AMF treatment to cells is not cytotoxic. It should also be pointed out that since no reduction of proliferation of B16-F10 cells was observed when all IO-nanocage groups were delivered in the AMF (Figure 3b), the reduction of expression by the firefly luciferase siRNA loaded 20 nm IO-nanocages shown in Figure 3b was not due to the cytotoxicity of Brownian movement of IO-nanocages in AMFs.

ANALYSIS OF ENDOSOME-CONTAINING IO-NANOCAGES WITH AND WITHOUT ALTERNATING MAGNETIC FIELDS

In the previous section, an enhanced siRNA delivery from IO-nanocages was observed when the experimental condition was tuned for Brownian motion of IO-nanocages. To examine the hypothesis that the Brownian motion of superparamagnetic IO-nanocages triggered by an AMF accelerates endosomal escape for an efficient delivery, endo-/lysosomal compartments in cells before and after the AMF application were studied by TEM and a calcein leakage assay. When the AMF was not applied (Figure 5a), we observed in the TEM image that IO-nanocages were entrapped in endosomes and the endosomal membranes were intact; however, endosomal membranes were ruptured and IO-nanocages were released into the cytoplasm when an AMF was applied (Figure 5b). To further support that magnetic Brownian motion is a main contributor to the endosomal escape, we performed the same in vitro experiment using 15 nm IO-nanocages, whose absorbed energy from the AMF has a minimal contribution from the Brownian motion (Figure 1c). In the TEM images shown in Figure S2, 15 nm IO-nanocages were retained in endosomes and endosomal membranes were intact without rupturing even after applying AMFs. Thus, this result is consistent with our finding that the Brownian motion of 20 nm IO-nanocages is responsible for the effective endosomal escape and the efficient RNA delivery in AMFs while the Brownian motion of 15 nm IO-nanocages is not strong enough to give the same effect. To support this observation in the wider range of the intracellular environment, we also performed a calcein leakage assay. The fluorescence (FL) intensity of a membrane-impermeable dye of calcein (150 μg/mL) should be intensified when endo-/lysomal membranes are compromised and calcein leaks out from endosomes to the cytoplasm. When 15 nm IO-nanocages were delivered in the absence and presence of an AMF (Figure 5c), we observed punctuated patterns with dim fluorescence intensity, suggesting that calcein is still entrapped in endo-/lysosomes regardless of the AMF application. Similarly to this result, when 20 nm IO-nanocages were delivered in the absence of an AMF (Figure 5d, left), the fluorescence intensity was dim and punctuated patterns were observed. However, following the application of AMF after cells were incubated with 20 nm IO-nanocages (Figure 5d, right), the fluorescence intensity of calcein (in green) increased in cells, which indicates that the endosomal membrane was ruptured and calcein was leaking out to the cytoplasm. As shown in Figure 5e, the fluorescence intensities of calcein in B16-F10 cells from Figure 5c,d were quantified, and the intensity was increased by 13-fold due to the leakage via the endosomal rupture when an AMF was applied. To further confirm the endosomal escape of IO-nanocages, we imaged by confocal microscopy the colocalization of endosomes and IO-nanocages in B16-F10 cells before and after applying AMFs (section V in the Supporting Information). IO-nanocages were labeled with Cy-5 (red), whereas endosomes were stained with Rab7 primary antibody and Alexa488 secondary antibody (green). As shown in Figure S4a, in the absence of AMFs, the fluorescence of IO-nanocages and endosomes (Rab7) overlapped, as shown by yellow dots, indicating that IO-nanocages were retained inside the endosomes. However, the degree of colocalization of endosomes and IO-nanocages was significantly reduced in the presence of AMFs, suggesting that IO-nanocages were released from these endosomes with the application of AMFs (Figure S4b). Thus, all studies of TEM imaging, a calcein leakage assay, and intracellular confocal imaging showed that endosomal escape of IO-nanocages accelerated under the influence of magnetic fields.

Figure 5.

TEM images of endosomes containing siRNA-loaded IO-nanocages in B16-F10 cells (a) before and (b) after the application of Aan MF (f = 335 kHz). Scale bar: 200 nm. Endosomes and IO-nanocages in cells indicated by red arrows in TEM images are illustrated to the right of each image. (c) Confocal image of a calcein leakage assay when B16-F10 cells were incubated with siRNA-loaded IO-nanocages at a size of 15 nm in the absence (left) and the presence (right) of AMFs. Calcein (green) was used as a membrane-impermeable dye to probe the porosity of the endo-/lysosomes, and nuclei (blue) were stained with DAPI-mounting media. Scale bar: 20 μm. (d) Confocal image of a calcein leakage assay when B16-F10 cells were incubated with siRNA-loaded IO-nanocages at a size of 20 nm in the absence (left) and the presence (right) of AMF. The color distribution is the same as in (c). Scale bar: 20 μm. (e) Quantitative analysis of calcein leakage by B16-F10 cells from the fluorescence intensity distributions of calcein in (c) and (d). N = 4. In (c) and (d), IO-nanocages are not fluorescently labeled.

In conclusion, the outcome demonstrated that superparamagnetic IO-nanocages can be loaded with siRNAs using a simple incorporation protocol, and siRNAs released from IO-nanocages silence the luciferase expression in reporter B16-F10 cells when an AMF is applied at f = 335 kHz under an experimental condition where IO-nanocages undergo Brownian motion. The endosomal membranes were observed to be ruptured under this condition, suggesting that this Brownian motion of IO-nanocages is responsible for compromising endosomes and the escape of IO-nanocage carriers could release a higher concentration of siRNAs in the cytoplasm. The siRNA delivery efficiency by the Brownian-dominant IO-nanocage (20 nm) is 5 times higher than that of the Néel-dominant IO-nanocage (15 nm), supporting the hypothesis that magnetically induced Brownian motion plays a role in escaping endosomes and enhancing the transfection efficiency. It should be noted that AMF frequencies lower than 335 kHz could generate more Brownian torque.³⁸ While we had instrumental limitations to lowering AMF frequencies to less than 335 kHz, the reason we settled with 335 kHz is that efficient endosomal escape with effective RNA delivery was already accomplished at this AMF frequency and a higher torque of Brownian motion by further lowering the frequency could reduce the cell viability. Furthermore, the frequency range and power of our experimental AMF setting are clinically relevant,³⁹ which is important for our future interest in biomedical applications. Therefore, we used a single AMF frequency of 335 kHz to report this interesting intracellular phenomenon. The outcome from this work could further accelerate the clinical translation of gene therapy and other gene technologies such as gene editing, affecting the broad fields of biological and biomedical engineering when efficient magnetic deliveries with Brownian motion can be accomplished by IO-nanocages.

We expect that IO-nanocages could play pivotal roles in other biomedical fields besides gene delivery. For example, IO-nanocages could also act as biosensors, since the oscillating frequency of magnetically driven Brownian motion would vary with the concentrations of nucleic acids/proteins attached on these nanoparticles,^40,41 and the readout would be the optical absorbance response via magnetically induced linear dichroism⁴² or harmonics of distorted magnetization created by the Brownian motion of SPIONs.⁴³ A magnetically driven Brownian movement of nanoparticles could also guide cellular development and carry out mechanotransduction to alter the fate of phenotypes of cells. Furthermore, if the capping agents of IO-nanocages are modified to bind on actin to create cytoskeletal nanofibers, superparamagnetic IO-nanocages could also generate sufficient mechanical force in an AMF to affect the phenotypic evolution of cells.⁴⁴ Thus, biomedical applications driven by the magnetic Brownian motion of IO-nanocages have great potential, and it is critical to examine this concept in animal models and investigate the efficacy of therapeutic reagent delivery of IO-nanocages by magnetically driven Brownian motion.