If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Department of Electrical and Computer Engineering, University of Miami, Coral Gables, FL, USABiochemistry and Molecular Biology, University of Miami, Miami, FL, USA
A study through a reduced in vitro model (on E18 rat hippocampal cell cultures) for the first time demonstrates how 30-nm MENPs can be used to wirelessly induce neural activity via application of magnetic fields, with a sub-25-msec temporal response.
•
The property of the nanoparticles such as the magnetic coercivity is used as a wireless switch to activate action potential in selected regions.
•
The validity of MENPs-based neural firing approach is confirmed through different positive and negative control measurements.
•
The importance of having MENPs adequately dispersed to ensure the desired wireless neural firing control operation is being demonstrated.
Abstract
The in vitro study demonstrates wirelessly controlled modulation of neural activity using magnetoelectric nanoparticles (MENPs), synchronized to magnetic field application with a sub-25-msec temporal response. Herein, MENPs are sub-30-nm CoFe2O4@BaTiO3 core-shell nanostructures. MENPs were added to E18 rat hippocampal cell cultures (0.5 μg of MENPs per 100,000 neurons) tagged with fluorescent Ca2+ sensitive indicator cal520. MENPs were shown to wirelessly induce calcium transients which were synchronized with application of 1200-Oe bipolar 25-msec magnetic pulses at a rate of 20 pulses/sec. The observed calcium transients were similar, in shape and magnitude, to those generated through the control electric field stimulation with a 50-μA current, and they were inhibited by the sodium channel blocker tetrodotoxin. The observed MENP-based magnetic excitation of neural activity is in agreement with the non-linear M − H hysteresis loop of the MENPs, wherein the MENPs’ coercivity value sets the threshold for the externally applied magnetic field.
Background
An optimal method for controllably stimulating localized regions of neurons without requiring surgically implanted electrodes or genetic modification remains an open question. Magnetoelectric nanoparticles (MENPs) have a capability to overcome this challenge because of their magnetoelectric (ME) effect, which converts a wirelessly applied magnetic field into dipole electric fields [1]. In theory, if a MENP, serving as a dipole, is located on the cellular membrane, a sufficiently strong electric field from the nanoparticle could affect the membrane potential to generate an action potential. By inducing local action potentials deep in the brain via application of external magnetic fields, MENPs offer a wireless alternative to existing local deep-brain stimulation (DBS) approaches using surgically implanted electrodes [
]. Furthermore, given the spatial resolution of the MENPs-based stimulation is fundamentally limited only by the nanoparticle size and the ability to control spatiotemporal patterns of the remotely applied magnetic field, a single-neuron wireless activation could be theoretically achieved with MENPs.
In the last decade, a number of approaches have been proposed to attempt a ‘wireless’ mechanistic approach; each with its own significant drawbacks. Optogenetics, one of the most viable approaches, uses blue light to control ion channels and influence the membrane potential of neurons. However, it faces limitations in the difficulty of getting the light to penetrate through the skull and brain tissue, as well as the need to genetically modify neurons to express light-sensitive ion channels or pumps [
]. Neural activation by a focused ultrasound (FUS) waves is another wireless stimulation alternative; similarly, the main roadblock is the limited penetration in the skull and tissues at safe energy levels [
]. This approach is based on controlling heat-sensitive TRPV4 receptors in neurons, using magnetic nanoparticles to generate local temperature gradients. In the study, they were able to increase neuronal activity with magnetic fields in the frequency range of up to 1 MHz. One of the main advantages of this approach is the easy penetrability of the magnetic field through the skull and brain tissues and potentially high spatial resolution. However, like the optogenetic approach, the magnetothermal approach requires genetic modification, thus limiting its applications in humans; also, its reliance on temperature to control stimulation makes it difficult to implement.
The MENPs-based approach described here overcomes the mentioned roadblocks. It does not require the use of genetic modification, nor does it use temperature as a control parameter. Furthermore, the ability to spatially localize the MENPs using magnetic fields allows to significantly surpass the spatial resolution limitation of existing transcranial magnetic stimulation (TMS) approaches; the resolution in the centimeter-size range renders TMS inadequate for most medical applications which require localized stimulation [
Recently, through in vitro, ex vivo and in vivo experiments, MENPs have been shown to be capable of modulating neuronal activity in the central nervous system (CNS), though no field-synchronized firing has been shown yet [
]. The MENPs' capability to controllably cross the blood-brain barrier (BBB) when administrated intravenously or intranasally via application of a magnetic field gradient on the order of 1000 Oe/cm has already been demonstrated both in vitro and in vivo [
]. However, despite the promising physics, no nanoscale MENPs-based, controllable, field-synchronized neural modulation has been demonstrated yet. The current limitation can be explained by the fact that to date the MENPs' highly non-linear nanoscale physics has not been properly exploited to provide the wirelessly controlled high-efficacy magnetic-to-electric energy conversion as required for local synchronized stimulation. Hence, the goal of this study is to overcome the stumbling block by building on fundamental understanding of the MENPs' non-linear physics. Our approach uses a reduced system, i.e., neuronal cell cultures, to enable controlled experiments aimed to understand the nanoparticles' and applied field’ conditions.
The reason that MENPs can enable wirelessly controlled stimulation of neurons in the nanoparticles' immediate vicinity is their relatively high ME coefficient (∼10 V/cm/Oe) [
]. Like other nanoparticles in the sub-30-nm size range, MENPs can cross the BBB, and, due to the significantly increased surface-to-volume ratio, can interact with the biological microenvironment at the sub-cellular level [
]. However, unlike any other nanoparticles, due to the significant ME effect, MENPs enable the use of wirelessly applied magnetic fields to induce relatively strong electric fields in the nanoscale proximity of the MENPs (Fig. 1). This property of the MENPs unlocks a new way to electrically affect biological properties at the cellular and even sub-cellular level. Unlike magnetic fields, electric fields can directly interact with neurons in the brain by modulating their membrane potential. However, broad electric fields cannot be used to wirelessly stimulate individual neurons, because they will interfere with the entire neural circuit within the range of the electric field. On the other hand, magnetic fields, if they are relatively weak (<1 T) and slow-varying (≪10 kT/s), do not induce electric fields or currents in the brain, and are thus not capable of activating neurons by themselves. For comparison, in the traditional TMS approach, the magnetic fields are relatively high (>∼1T) and rapidly changing (>10 kT/s), leading to Eddy currents which in turn stimulate relatively large brain regions. Consequently, neither electric nor magnetic fields alone are capable of wirelessly controlled local stimulation with an adequate spatial resolution. In contrast, MENPs, owing to their ME effect, can achieve this objective by converting a broad magnetic field into local electric fields around the MENPs so that only neurons in the vicinity of the MENPs are activated. Therefore, the spatial resolution of the resulting stimulation would be enhanced by the ability to control the localization and properties of the MENPs. The nanoparticles' electric fields act on the neuronal membrane in the immediate vicinity of the nanoparticles which are approximately 30 nm in diameter. A MENP's dipole electric field dies off rapidly with the distance away from the nanoparticles (∼1/r3). Therefore, for the electric field of the nanoparticles to remain significant for the purpose of controlling the neural firing, the nanoparticles must be less than approximately 30 nm from the membrane. Furthermore, given different conductivity values of the intra-/extra-cellular space and the membrane regions, the field is screened off differently depending on the specific MENPs' location. As a result, the highest electric field would be reached when the nanoparticles are in/across the membrane region. If the MENPs generate a local electric field that depolarizes neurons beyond the action potential threshold, by having a sufficient number of these magnetically driven electric-field-generating nanoparticles in a selected local brain region (e.g., guided through magnetic field spatiotemporal patterns), one could selectively induce local action potentials, without affecting the surrounding neuronal circuitry. There are two alternative ways to localize neural modulation with MENPs, i.e., through localization of MENPs or magnetic fields, respectively. In the MENPs' localization case, nanoparticles could be localized using time-varying magnetic fields to induce a metastable diamagnetic response of the nanoparticles. Using antibodies could further provide cell-specific or sub-cellular-region-specific targeting [
]. In the field localization case, the nanoparticles would be administrated globally or sub-globally into specific organs or regions; then the modulation would be localized by using 3D magnetic field gradients matched to specific magnetic characteristics of the nanoparticles, e.g., the magnetic coercivity.
Fig. 1Underlying mechanism. A high-level illustration to show how, due to the magnetoelectric effect of MENPs, a wirelessly induced magnetic field, H, can be used to induce a local dipole electric field, E, in the vicinity of a nanoparticle (shown as a red circle). We hypothesize that the local electric fields of the MENPs activate voltage-gated ion channels in proximity to the nanoparticle. H and E are the magnetic and electric fields, respectively; M and P are the nanoparticles' magnetization and polarization, respectively. It can be noted that the actual electric field induced at the dipole pole can be substantially higher than the field predicted by the linear phenomenological Landau equation. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)
where P and α are the MENPs' polarization (electric dipole moment per unit volume) and ME coefficient, respectively, and H is the applied magnetic field. Given α of 10 V/cm/Oe, the local electric field in the sub-diameter vicinity of the nanoparticle in response to a 1000-Oe magnetic field would be approximately 10000 V/cm (10 mV/10 nm). This field is on the order of the field required to overcome the action potential threshold by activating local ion channels through voltage gating [
]. Here, it is noteworthy that the actual electric field induced at either pole of the dipole can be substantially higher than the value predicted by this linear phenomenological equation. Therefore, given the membrane thickness and the nanoparticle size are on the order of 10 and 30 nm, respectively, the field effect would be substantially increased if the nanoparticle could be placed on the membrane with one of the dipolar poles touching the membrane. Furthermore, the assumed linear dependence cannot be justified, given the highly non-linear physics of the core-shell MENPs used in current studies, particularly because of the highly non-linear field dependence of the magnetostrictive core. The ME effect in these nanostructures exists because of the lattice-matched crystallographic interface between the magnetostrictive core and the piezoelectric shell [
]. To maximize the ME effect, it is critical to maximize both the core's magnetostrictive coefficient and the shell's piezoelectric coefficient. For biological applications, the most popular shell material is barium titanate [
]. As for the core material, cobalt ferrite is one of the most commonly used compounds because of its relatively high saturation magnetostriction on the order of 200 ppm [
]. The high magnetostriction of this compound comes at the cost of a relatively high magnetic anisotropy, which in turn leads to a highly hysteretic M − H dependence. If this hysteresis is not considered during application of the control magnetic field, the magnetically induced electric field of MENPs could be substantially smaller compared to the field required for the synchronized stimulation. This basic non-linear physics is described in more detail below in Section Discussion and also shown through basic concept illustrations in Supplementary Materials Fig. S1. This study aims to demonstrate wirelessly controlled stimulation of neurons by considering this non-linear physics and ME properties of MENPs, and compare the stimulation signals to those achieved by the traditional invasive electrode-based DBS.
To study the ME-caused stimulation effects, we used the above core-shell MENPs made of the cobalt ferrite (CoFe2O4) magnetostrictive core and the barium titanite (BaTiO3) piezoelectric shell. The nanoparticles were coated by a thin layer of polyethylene glycol (PEG) (∼1 nm) with the two-fold goal to ensure their (i) biocompatibility and (ii) adequate dispersion. The spinel ferrimagnetic cobalt ferrite core and the perovskite piezoelectric barium titanite shell were lattice matched at the interface to maximize the magnetic-to-electric energy transfer, thus maximizing the ME coefficient. Both the synthesis and nanoprobe-based characterization of these nanostructures were described in detail in recent publications [
]. A transmission electron microscopy (TEM) image of a typical naked MENP is shown in Fig. 2a. Atomic force microscopy (AFM) and magnetic force microscopy (MFM) images of naked MENPs are shown in Fig. 2b left and right, respectively. The core-shell crystallographic configuration is visible in the TEM image, with the magnetic core showing a single domain crystallographic orientation. The difference between the magnetic properties of the core and the shell is also clear in the MFM image; the dominant signal stems from the core, with no significant signal coming from the shell.
Fig. 2MENPs. (A) TEM image of a naked CoFe2O4–BaTiO3 core-shell MENP. (B) AFM (left) and MFM images of naked core-shell MENPs. The scale bar is 30 nm. (C) Major M − H loops measured via AGM of the core-shell MENPs and the control purely magnetic nanoparticles made of the same material as the core component, CoFe2O4, of the MENPs (right). M − H loops in perpendicular and parallel field orientations are shown by solid and broken lines, respectively. The averaged (over 5 measurements) perpendicular coercivity fields of the core nanoparticles and MENPs are 328 and 283 Oe, respectively. The averaged parallel coercivity fields for the two nanoparticle types are 276 and 271 Oe, respectively.
M − H loops of the core-shell MENPs as well as of the control magnetic nanoparticles, with no ME effect, made of the same material as the MENPs' magnetic core, i.e., cobalt ferrite (CoFe2O4), measured via alternating gradient magnetometry (AGM), are shown in Fig. 2c. To detect any possible anisotropic effects, the nanoparticles were deposited on a glass substrate under application of a 500-Oe magnetic field in the out-of-plane orientation, thus having their “easy” axes oriented perpendicular to the substrate. Perpendicular and parallel M − H loops show the measurements with a magnetic field applied perpendicular and parallel to the substrate, respectively. For the MENPs and the magnetic core nanoparticles (just cores without shells), the saturation magnetization values are approximately 0.5 and 37.1 emu/g, respectively, the perpendicular coercivity fields are approximately 328 and 283 Oe, respectively, the parallel coercivity fields are 276 and 271 Oe, respectively. The coercivity field is an extrinsic measure of the nanostructure, it indicates the field required to reverse the magnetization. Another important observation from these M − H loops is the fact the fields required to saturate the MENPs’ magnetization in the perpendicular and longitudinal orientations are on the order of 1000 and above 5000 Oe, respectively. The saturation field in the longitudinal orientation is indicative of the magneto-crystalline anisotropy field, an intrinsic measure to characterize the magnetic energy density contained in the magnetic core. In turn, this value indicates the applied magnetic field required to induce the highest magnetostrictive effect, thus leading to the highest local electric field due to the ME coupling between the core and the shell. The ME effect of the nanoparticles was also qualitatively measured using a voltage sensitive fluorescent dye placed in the proximity of the nanoparticles, as shown in Supplementary Materials Fig. S2. This experiment directly shows that when exposed to a magnetic field MENPs generate relatively strong electric fields according to the magnetoelectric physics.
For the discussed nanoscale physics to work, it is critical to ensure the nanoparticles are adequately dispersed (Supplementary Materials Fig. S3). Otherwise, their key properties due to the nanoscale physics would be compromised because of the substantially reduced effective surface to-volume ratio of the agglomerated and thus clustered nanoparticles. To achieve the adequate level of dispersion, as confirmed through dynamic light scattering (DLS) and Zeta potential measurements, we used a method of gravity filtration along with PEGylation of the nanoparticles, as described in detail in Section Methods.
2.2 Ca2+ ion imaging correlated with magnetic field application
Hippocampus cell cultures grown from Sprague Dawley embryonic day-18 rat neurons were used in all the experiments reported within. To detect neuronal activity, Ca2+ imaging was performed on cells loaded with Cal 520 fluorescent dye from AAT Bioquest. A typical fluorescence image of a cell culture is shown in Fig. 3A, which normally contains 40–200 neurons. Regions of interest (ROIs) were extracted using automated local thresholding in ImageJ – an image processing program, which was manually corrected to separate out clusters of neurons. An effort was made to tag every neuron in the view, excluding those at the edges that tend to produce artifacts. The default dose of the administrated nanoparticles was approximately 5 μg per dish, added to the cell media in a 5-μl DI water solution. A representative fluorescence image of a cell culture with MENPs is shown in Fig. 3B. Typical scanning electron microscopy (SEM) images of fixed cells, with MENPs, are shown in Fig. 4. These images indicate that the nanoparticles tend to concentrate on the dendrites.
Fig. 3(A). W/O MENPs: A 10X fluorescence image of a high-density primary neuronal cell culture used in the experiment: hippocampus dissociated cells from E18 Sprague Dawley Rat (BrainBits). The culture was tagged with a fluorescent calcium ion indicator Cal-520. (B). With MENPs: A merged brightfield and fluorescence image of a high-density primary neuronal cell culture (hippocampus dissociated cells from E18 Sprague Dawley Rat), tagged with a fluorescent calcium ion indicator, Cal-520, and with administrated MENPs (5 μl of DI water with 5 μg of the nanoparticles). The nanoparticles are localized around the dendrites. Some of them are shown through red broken-line enclosures. Only agglomerates of nanoparticles with a net size of more than 100 nm can be detected. Image taken by a 100X oil immersion lens. (C). Neuron stimulation with MENPs: A sample zone with magnetic stimulation correlated fluorescent change highlighted in red. Brighter neurons are less strongly highlighted due to their smaller overall change in fluorescence relative to their base value. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)
Fig. 4SEM images of neuron culture with PEG-coated MENPs. (A). Zoomed-out (top) and (B) zoomed-in SEM images of cell cultures with 100 μl of nanoparticles solution administrated into the petri dish under study. Nanoparticles' amount was chosen to be substantially higher (x100) than the usual amount used in the experiments amount in order to obtain sufficient signal under SEM imaging. The whitish dots are the regions with nanoparticles.
The following magnetic field sequence was used to wirelessly trigger neuronal activity with the MENPs. After recording a 60 s baseline, five 2-s-long trains of 20 bipolar square pulses with 50% duty cycle were driven through an audio amplifier, which produced a 1200-Oe bipolar magnetic field (Supplementary Materials Fig. S4). The actual field profile measured at the location of the cells is shown in Supplementary Materials Fig. S4c. Both the magnetic and electric signal circuits are shown in Supplementary Materials Fig. S4d. The electromagnet used is shown in Supplementary Materials Fig. S4e. These magnetic field pulses were applied approximately every 10 s in the orientation perpendicular to the cell culture dish. 30 s after application of magnetic field, a control sequence of three to five electric pulses (20 mA current) were applied every 10 or 20 s respectively using two, non-magnetic, high-purity, silver microelectrodes, placed in diametrically opposite locations in the 35-mm dish. The purpose of this electric pulse sequence, applied with the same temporal pattern of 2-s long 20 pulse/sec trains of bipolar pulses with a duty cycle of 50%, was to compare the wirelessly controlled MENPs-based magnetic stimulation to the standard electric stimulation achieved with physical microelectrodes.
The typical waveform response to such sequential application of contactless magnetic and electrode-based electric stimulation pulses is shown in Fig. 5a. The neuronal temporal response to the application of both magnetic and electric fields was faster than the measurement temporal step of 25 ms. We found a high level of similarity between the Ca2+ transients evoked by the MENP-based wireless magnetic stimulation and the wired, electrode-based stimulation. Neuron pulse-triggered averages from 850+ cells across 5 dishes after MENP-magnetic and electrode stimulation without and with MENPs and after adding TTX are shown in Fig. 4b. This showcases that there were no statistically detectable neuronal firing events in the cases of magnetic field alone without MENPs, or after TTX was added for both MENP-based wireless magnetic stimulation and the wired-electrode-based stimulation. All cells with TTX had shown firing events from magnetic and electric stimulation before the TTX was added (see statistical comparisons in Table 1).
Fig. 5Ca2+transients in hippocampus cell control and MENP cultures, tagged with Cal520 fluorescent dye, in response to contactless magnetic and electrode-based electric stimulation pulses. After a 60 s baseline recording, five 2-s-long trains of 20 bipolar pulses (50% duty cycle) of a ±1200-Oe magnetic field were applied every 11 or 12 s (blue vertical lines) in the orientation perpendicular to the cell dish. Then, after a 30-s break, a control sequence of three electric field pulses (red vertical lines), with a current magnitude of 20 mA, was applied every 20 s using two microelectrodes placed in diametrically opposite sites in the 35-mm dish. This experiment was conducted on four conditions. Control dishes with no particles, control dishes with the MENPs and a sub-coercivity field of 300-Oe instead of the default 1200-Oe field, dishes with MENPs and the above coercivity field, and those same dishes after adding 1 mM of a sodium ion channel blocker. (A) Fluorescence normalized time traces from each of the experimental conditions (B) Neuron pulse-triggered averages of the magnetic and electric pulses from responsive neurons for each of the experimental conditions. The shaded region corresponds to the duration of the applied signal. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)
Table 1Experimental summary table for all experimental conditions for in vitro studies. P-values were calculated using the Kolmogorov-Smirnov test comparing each condition's baseline, magnetic, and electric inter-spike interval distributions. The null hypothesis was that the stimulation would not change the distribution of firing events. P-values < 0.05 are bolded. All electrical stimulation except in dishes with TTX were statistically significant, whereas magnetic stimulation was only significant with MENPs and above threshold fields.
Fig. 5a shows temporal profiles of calcium transients with and without MENPs present. Due to the constraints of our calcium imaging methodology, we defined spikes as fluorescence changes exceeding a threshold of 20% the average fluorescence change of neurons during electrical stimulation. This was chosen to directly compare the ability of MENPs to trigger neural activity against that of traditional electrodes. The five blue and three red dashed vertical lines correspond to the wireless magnetic and wired-electrode-based electric stimulation, respectively. To verify reproducibility, such alternating sequences of magnetic and electric field trains were applied multiple times in the same zone. At the end of the measurements of each dish, 1 mM of TTX was added to block sodium ion channels and then, after waiting for approximately 30 min, the above two stimulation sequences were measured again.
To show the similarity between the MENP-based wireless magnetic stimulation and the wired-electrode-based stimulation, the data are visualized in the inter-event interval histograms for cell cultures with MENPs and control cell cultures with no MENPs, including their baseline (no stimulation) as well as MENP-magnetic and electrode stimulation intervals in Fig. 6b . The histograms clearly show that the intervals between calcium transients synchronize to the stimulation sequence interval in both the MENP-based wireless magnetic and wired-electrode-based stimulation. Without MENPs, this synchronization is not present when the magnetic field pulse sequences are applied but continue to be present in the electrode-based stimulation. The above conclusions were drawn from statistical comparisons laid out in Table 1. The same experiment was conducted using magnetic pulse sequences with a strength of 100 Oe (Supplementary Materials Fig. S5). As expected, given the field is significantly less than the coercivity field (of 300 Oe), no rapid response (within a second) was observed for the low magnetic field application. The statistical analysis of the results is summarized in Table 1 and also presented in Fig. 7.
Fig. 6Calcium event trains and inter-event intervals for control and MENP cultures in response to contactless magnetic and electrode-based electric stimulation pulses. (A) Temporal alignment of calcium transients with magnetic and electrical stimulation (with μA) with and without MENPs. Events defined as calcium transients of similar magnitude to the average electric stimulation transient. The first five blue dashed vertical lines correspond to magnetic field stimulation instances while the latter three red dashed vertical lines correspond to the electric field stimulation instances. The color bar refers to the calcium transient magnitude normalized to the average magnitude of an electrically evoked Ca2+ transient for that experiment. (B) Inter-event interval histogram for all experimental conditions, comparing their baseline (no stimulation), magnetic and electrode stimulation intervals. The highlighted histogram bins correspond to the stimulation intervals used to stimulate the neurons. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)
Fig. 7Neuron response rate to stimulation. The percentage of neurons that respond to either magnetic or electric stimulation in a 4-s time window following stimulations. Responses were measured as the peak of a fluorescence change spike exceeding a threshold of 20% the average calcium fluorescence change during electric stimulation. Though fluorescence change happens immediately with the onset of stimulation, peaks typically occur 1–2 s following stimulation, as shown.
For further comparison, the rate of evoked calcium transients was measured for three different MENPs, type A, B, and C, with three different parallel coercivity fields, 272 Oe, 990 Oe, 70 Oe, respectively (Fig. 8A), by application of the magnetic field pulses with two different strengths, 100 and 1200 Oe, respectively (Fig. 8B). These experiments further confirmed that the wireless response was observed only when the magnetic field exceeds the coercivity field.
Fig. 8Three types of MENPs Based on Coercivity Field: (A) M − H loops in perpendicular (left columns) and parallel (to the substrate) orientations for three types of MENPs under study, A, B, and C, respectively. The three MENPs types were created by controlling their relative core-to-shell volume ratios (A: 1:3; B: 1:5; C: 1:7). To detect the magnetocrystalline anisotropy during M − H measurements using an alternating gradient magnetometer (AGM) (LakeShore MicroMag2900), the nanoparticles were deposited on an AGM substrate under application of a d.c. magnetic field on the order of 100 Oe in the perpendicular to the plane orientation. The coercivity values of the three MENPs' types measured in the parallel orientation were approximately 272, 990 and 70 Oe, respectively, while their saturation magnetizations were 0.9, 0.8 and 0.6 emu/g, respectively. (B). Neuronal firing frequency (frequency of spikes) in arbitrary units (au) for three types of MENPs for two values of the magnetic field strength applied, ∼100 and ∼1200 Oe, respectively. Measurements conducted from 24+ dishes.
The presented in vitro experiments showed that MENPs could wirelessly modify neural activity, with a response time in the sub-25-ms range. The coercivity field of the nanoparticles served as a wireless magnetic switch required to stimulate neurons. More precisely, to ensure complete nanoparticles’ magnetization reversal, a magnetic field strength above 2HC was used to activate neurons. The experiments showed that the magnetically stimulated and optically detected calcium transient waveforms were of the same form and magnitude as those stimulated electrically with a 50-μA current using traditional, high purity silver microelectrodes.
The control measurements further supported this hypothesis. The sodium channel blocker, TTX, ended the high rate of random spikes as well as both magnetic and electric stimulation evoked spikes, indicating that the observed calcium signals were due to neural activity, not mechanical or other artifacts. Furthermore, all reported post-TTX measurements were conducted only on the dishes which showed calcium response to the magnetic field application prior to the TTX administration, to avoid false negatives. Also, application of a field with a strength of 100 Oe, i.e., substantially smaller than the coercivity field (∼300 Oe), did not lead to any correlated change in neuronal activity. As in the TTX case, only the dishes which positively responded to the application of 1200 Oe were counted. The control value of 100 Oe is essential to understand the results of the studies in the past in which they used MENPs to increase neuronal activity [
]. Although they did show an increased rate of activation, they did not demonstrate the same degree of forced time alignment of neural activity and field application. To wirelessly activate neurons on demand, it is critical to apply a field to the neurons significantly higher than the MENPs’ coercivity value, given the nanoparticles are on the membrane. Only in this case, the MENPs-induced local electric field could achieve its maximum value, thus in turn effectively depolarizing the membrane potential to the threshold value required for firing.
The SEM and optical fluorescent measurements indicated that the nanoparticles were mostly concentrated in the dendrite regions. This suggests that the observed somatic calcium transients result from the integration of local depolarization of dendritic membranes driven by the MENPs' magnetic-to-electric field coupling. Each MENP acts as a field-controlled switch to control local voltage-gated ion channels next to the nanoparticle. When the switch is on, via application of the magnetic field, local ions and thus electric-field energy could be transferred across the membrane. Action potential firing occurs when the collectively transferred energy from the dendrites and soma exceeds a required threshold. For this switch to function properly, the local electrical field due to the nanoparticle needs to generate an electric field on the order of 10–15 mV across the membrane as required for activation of the local ion channels. According to Equation (1), given the ME coefficient of 10 V/cm/Oe, application of 1000 Oe should induce a local electric field on the order of 10 mV/10 nm, which is comparably high. With the nanoparticles attached to the cell membrane, considering the M − H hysteresis (Fig. 2c), the complete reversal of the magnetization would be possible only if the applied a.c. magnetic field strength significantly exceeds the coercivity field, HC (>∼ 2HC). Otherwise, the complete magnetization reversal would be prohibitive, thus leading to a relatively small magnetization change. The difference between the two cases of the applied a.c. magnetic field can be explained by the difference between the magnetization changes occurring in the major and minor M − H hysteresis loops, respectively. For example, given the hysteresis loop in Fig. 2c, with a coercivity of 300 Oe, an applied a.c. magnetic field with a strength of 1200 Oe would saturate the loop, which in turn would lead to the magnetization change on the order of 2Ms, which is the maximum possible change of the magnetization. In contrast, application of a 100-Oe a.c. magnetic field would lead to a magnetization change of <0.1MS, following a horizontal plateau of the minor loop. Consequently, the transferred electric field energy would differ by an order of magnitude between the major and minor loop cases, respectively. The observation is important because in several recent in vitro and in vivo studies that used MENPs to stimulate neurons, trigger behavioral changes in animals, induce neurogenesis, or enhance bone cell proliferation, they applied an a.c. magnetic field with a strength significantly smaller than the coercivity of the nanoparticles [
]. The fact that these experiments showed an increased neuronal activity, though did not demonstrate synchronization of the activation events with the magnetic field application, can be explained by the relatively small electric field (<1mV/10 nm) induced by the small magnetic field (100 Oe). In this case, although the MENPs’ induced electric field did reduce the membrane potential by approximately 1 mV, thus leading to an increased rate of neural activation, this reduction was still significantly lower than the threshold value (∼10 mV) required for field-synchronized firing. In contrast, when the applied magnetic field was larger than 2Hc, the membrane potential was depolarized by approximately 10 mV. Such a high induced electric field is close to the threshold required to neuronal firing, thus leading to the neuronal activation events being synchronized with the magnetic field application. Another important factor which needs to be kept in mind when using MENPs to induce wirelessly controlled neural firing is the location of the nanoparticles with respect to the neural network. The nanoparticles need to be adequately dispersed to ensure the nanoscale physics can be effective. To further significantly increase the stimulation efficacy, the nanoparticles need to be placed on the membrane, preferably with their poles touching the membrane surface. Both effects, i.e., the adequate dispersion and the location on the membrane surface, can be achieved by a special surface treatment of MENPs, e.g., coating them with biocompatible and possibly biodegradable polymers/copolymers.
In summary, this in vitro study demonstrates a field controlled wireless activation of neural firing owing to the ME effect. The MENPs’ magnetic field coercivity, HC, serves as a wireless field-controlled On/Off switch of each firing event. This conclusion was additionally supported through the experiment in which three types of MENPs, with coercivity fields of 272, 990, and 70 Oe, respectively, were used to induce neural firing by application of magnetic field pulses with two different strengths, 100 and 1200 Oe, respectively (Fig. 8). As a last remark, it should be understood that these nanoparticles represent the first generation of MENPs. They are sufficient to show the feasibility of the novel wireless brain stimulation technology. Future research could lead to development of biodegradable MENPs with a significantly higher ME coefficient, thus enabling a robust technology to wirelessly stimulate neural circuits with a single-neuron spatial resolution.
4. Methods
MENPs’ Synthesis: Chemicals: Cobalt(II) Nitrate HexahydrateSA (Co(NO3)2·6H2O), Iron(III) Nitrate NonahydrateSA (Fe(NO3)2·9H2O), Sodium HydroxideSA (NaOH), Barium CarbonateFS (BaCO3), Titanium(IV) IsoproproxideSA (Ti[OCH(CH3)2]4), Citric AcidSA (HOC(COOH)(CH2COOH)2), Polyethylene GlycolSA (H(OCH2CH2)nOH, MW: 3000), and EthanolMS (>99.7%), were purchased from Sigma Aldrich, Fisher Scientific, and Millipore Sigma. All reagents were used without further purification.
The default CoFe2O4@BaTiO3 core-shell nanoparticles were made in a two-step process. The cobalt ferrite cores were produced using the following co-precipitation route. In a typical synthesis, 300 mg of cobalt nitrate and 833 mg of Iron nitrate were added to DI water under constant 450 rpm stirring. After the metallic salts had fully dissolved, 3 M NaOH aqueous solution was added dropwise until the mixture reached a pH of 11. The solution was then heated to 80 °C and allowed to react for 30 min. The solution was then cooled, and the Cobalt Ferrite particles separated from solution by centrifuge. The particles were then washed twice with DI water and twice with Ethanol before left to dry overnight on a 60 °C hotplate.
The barium titanate shells were formed using a solgel process. The precursors were prepared in three beakers. The first beaker typically contained 40 mg of dried cobalt ferrite cores with 500 mg of citric acid in 20 ml of DI water, which was bath sonicated for 2 h. The second beaker contained 252 μl of titanium isopropoxide with 500 mg of citric acid in 30 ml of ethanol, and the third beaker had 176 mg of barium carbonate dissolved with 500 mg of citric acid in 30 ml of DI water. The second and third solution s were stirred at 500 rpm separately for half an hour to ensure chelation before combined. After the first solution had been sonicated, it was added to the combined solution and brought to 90 °C under continuous 500 rpm stirring to form the gel. This gel was transferred to an alumina crucible and calcined according to the following temperature profile. 25 °C–380 °C ramp in 180 min, 380°C–780 °C ramp in 280 min, held at 780 °C for 5 h, and then cooled from 780 °C to 25 °C in 220 min. The finished MENPs were then washed twice in DI water and twice in ethanol before left to dry overnight on a 60 °C hotplate.
PEGylation of MENPs: To PEGylate the MENPs, the particles were dispersed via bath sonicator in a 1:1 particle to DI water ratio. Polyethylene glycol (PEG) was added in a 1:3 ratio relative to the particles, and the solution was sonicated for 8 h while maintaining a bath temperature below 38 °C. The conjugated particles were removed and washed in DI water once to remove excess PEG.
Gravity Filtration: To ensure thus synthetized nanoparticles remain adequately dispersed, we employ the following process, called gravity filtration, to control the thermodynamic agglomeration of particulates. Any non-soluble nanoparticles tend to agglomerate to minimize the surface area; for MENPs, this is compounded by attractive forces between the magnetic cores. It is possible to create a semi-stable solution, where the frequency of nanoparticle interactions is low enough to let them remain dispersed for a long time (hours or even days). The agglomerated particles drift down in solution faster than the non-agglomerated counterparts. This approach is functionally similar to the standard centrifuge filtration, with the benefit of avoiding compression of particulates and thus allowing for a fine control over small size and density ranges.
Hippocampus Cell Cultures Preparation: Primary neuronal cultures used in this study were made from hippocampus dissociated cells from E18 Sprague Dawley Rat (Brainbits). Initially untreated 35 mm petri dishes with 14 mm glass bottom culture area (Matsunami Glass D35-14-1-U) were used to improve imaging quality. The dishes were first coated with 50 μg/ml poly-d-lysine (PDL) (Gibco A3890401) overnight at room temperature and rinsed three times with ddH2O and air dried. Then, 100 μg/ml laminin (BioLamina LN511) was coated on top of PDL in 37 °C incubator for 1–3 h. Dissociated cells were then seeded at a density of 60,000 cells/cm2 and ½ -media change was made every 3–4 days with fresh, 37 °C, CO2 equilibrated NbActiv4 (Brainbits). Experiments were conducted on the cultures between day 14 and day 21. A typical image of a high-density neuronal cell culture is shown in Supplementary Materials Fig. S4.
Tagging Hippocampus Cell Cultures With Calcium Dye Cal520: Calcium dye, Cal-520 (AAT Bioquest, Cat.# 21130), was used to record neural activity. 4.5 mM stock solution was prepared in anhydrous DMSO. Dye working solution was prepared in culture media with final concentration of 5-μM Cal-520, and 1.5 mM probenecid (AAT Bioquest, Cat. # 20062). Probenecid was to reduce the intracellular dye indicators. Cell culture media was changed with the dye working solution and incubated in 37 °C, CO2 incubator for 1 h. Then, 5 μg MENPs were added in the culture, gently mixed and incubated in a dark box for 30 min at room temperature.
Ca2+Imaging: After cells were loaded with the calcium dye, Cal520, and placed in the incubator at 37 °C for 1 h, MENPs solution were added to the dish and were incubated at room temperature for 30 min to allow them to settle to the bottom of the dish. Cells with particles were then moved onto the Nikon Eclipse E400 microscope mounted with the in vitro coil system. Cal520 was excited using a white LED with a dichroic filter including a 420-nm excitation filter and 520-nm emission filter. Movies were taken at 10× magnification, and the data was recorded using a PCO Panda camera at 40 FPS.
In vitro magnetic stimulation: A magnetic stimulation setup was designed to fit into a Nikon Eclipse E400 microscope and to hold a single cell culture dish. An electromagnet was used to provide an a.c. magnetic field from underneath the dish. AC signals were generated by a Digilent Discovery 2 board, controlled using a National Instruments DAQ device, and amplified by a Class H 4000-Watt stereo power amplifier. For experiments with a.c. magnetic stimulation, the a.c. magnetic field was a 1.2-kOe bipolar square wave at 20 Hz. The a.c. field magnitudes were verified with a gaussmeter. The sequence of stimulation was as follows: (i) a 1-min baseline was recorded; (ii) magnetic stimulation was applied for a total of 5 pulse trains, every 10 or 20 s, for a duration of 2 s; (iii) a second baseline for 30 s; (iv) a 20-mA current applied through two microelectrodes placed in the cell dish and separated by about 1 cm for a total of 3 pulse trains, every 20 s, for a 250 ms duration, (v) a third baseline for another 30 s. After recording multiple regions in each dish, 5 μL of 1-mM TTX was added to the cell dish containing 2 μL of a media as a control, and the same sequence of stimulation was repeated.
Alternating Gradient Magnetometry: The B–H loops were measured using an alternating gradient magnetometer (AGM) MicroMag 2900 from Princeton Measurements Corporation. The MicroMag 2900 system is capable of measuring up to 200 mg samples with its special robust “P1” probe. It can measure from 1 μemu to 5 emu with a resolution of 0.005% of full scale with 60% overrange capability. Its sensitivity varies with a 10-nemu standard deviation at room temperature with an averaging time of less than 1 s.
Transmission Electron Microscopy: The transmission electron microscopy (TEM) images were obtained using Philips CM 200 TEM microscope.
Atomic Force Microscopy and Magnetic Force Microscopy: The atomic force microscopy (AFM) and magnetic force microscopy (MFM) images were obtained using a standard Bruker Multimode microscope. The images were analyzed using the NanoScope software by Bruker.
Cell Fixation and Dehydration for SEM and AFM Measurements: First, the culture media was replaced with 4% glutaraldehyde in 4 °C PBS and incubated for 1 h at room temperature. Then, the culture was washed with PBS, and kept in cold ethanol with a gradually increased concentration (30%, 50%, 75%, 90%, and 95%) for 1 min at each value. Then, the culture was put in 100% ethanol for 15 min. Lastly, ethanol was aspirated, and the sample was air dried. Before imaging a thin layer of gold was coated on the samples to increase the surface conductivity, thus improving the SEM image quality.
Signal Processing for In Vitro Analysis: This section contains a brief explanation of the data analysis methods used. A more detailed description is provided in the Supplementary Materials Item S7. Image stacks were rigid motion corrected in FIJI and then cropped to avoid the black edges on shifted frames [
]. Fluorescence intensity traces were extracted from the recorded image stacks using manually annotated ROIs. The extracted time traces were then fed into our image processing pipeline, which performed denoising and detrending to remove the inherent intensity decay of Cal520. The pipeline then generated the signal triggered averages, calcium transient timing, and inter-event interval calculations.
CRediT authorship contribution statement
E. Zhang: Writing – original draft, Conceptualization, Methodology, Software, Validation, Formal analysis, Visualization, Investigation. M. Abdel-Mottaleb: Writing – original draft, Investigation, Data curation, Methodology. P. Liang: Conceptualization, Methodology, Supervision, Project administration, Funding acquisition. B. Navarrete: Investigation, Data curation, Methodology. Y. Akin Yildirim: Investigation. M. Alberteris Campos: Investigation. I.T. Smith: Investigation. P. Wang: Investigation. B. Yildirim: Investigation. L. Yang: Conceptualization, Data curation. S. Chen: Investigation. I. Smith: Conceptualization, Resources, Data curation. G. Lur: Conceptualization, Resources. T. Nguyen: Conceptualization. X. Jin: Conceptualization, Resources. B.R. Noga: Conceptualization, Resources, Investigation. P. Ganzer: Conceptualization. S. Khizroev: Conceptualization, Methodology, Formal analysis, Resources, Writing – original draft, Supervision, Project administration, Funding acquisition.
Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgments
This research was supported by the Defense Advanced Research Projects Agency (DARPA) and Naval Information Warfare Center, Pacific (NIWC Pacific) under Contract N66001-19-C-4019. Any opinions, findings and conclusions or N6600119C4019 recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the DARPA or NIWC Pacific. The study was also partially supported by National Science Foundation (NSF) under awards # ECCS-1935841. Furthermore, the authors would like to acknowledge the invaluable help and insightful advice throughout this entire study from Dr. Ibolya Edit Andras of the Department of Biochemistry and Molecular Biology of the University of Miami.
Appendix A. Supplementary data
The following is the Supplementary data to this article:
00227-3&id=gr1.jpg){kind=link}
00227-3&id=gr2.jpg){kind=link}
00227-3&id=gr3.jpg){kind=link}
00227-3&id=gr4.jpg){kind=link}
00227-3&id=gr5.jpg){kind=link}
00227-3&id=gr6.jpg){kind=link}
00227-3&id=gr7.jpg){kind=link}
00227-3&id=gr8.jpg){kind=link}