Preparation of nanomaterials
Magnetite nanodiscs (MNDs) were synthesized following a two-step solvothermal synthesis procedure3,4. In the first step, hematite nanodiscs (HNDs) were produced by solvothermal synthesis. In the second step, HNDs were reduced to MNDs while maintaining the morphology. The crystal structures of these nanodiscs were characterized using Transmission Electron Microscopy (TEM) and X-ray diffraction (XRD) (Fig. 1d, e, g). The XRD and VSM results indicate that HNDs are fully reduced to MNDs after a two-step synthesis process (Fig. 1g, S1a). HNDs, which served as our control group, were not magnetized by external magnetic fields. Additionally, BaTiO3 nanoparticles (BTO), sourced from US Research Nanomaterial, Inc., were characterized by TEM and XRD (Fig. 1f, h). All nanomaterials were functionalized for cellular applications; BTOs were coated with PEG to facilitate attachment to cell membranes18 and further functionalized with neutravidin for linkage with nanodiscs. Consistent with previous studies 3,4, both MNDs and HNDs were coated with PMAO. The PMAO-coated nanodiscs were conjugated with biotin for linkage with BTOs.
MagTIES induced neuronal responses in vitro
To validate our hypothesis regarding the efficacy of MagTIES, we initially applied the functionalized BTOs to primary cultured hippocampal neurons. Subsequently, we introduced the functionalized MNDs or HNDs to these neurons. The biotinylated MNDs or HNDs were linked to the neutravidin-conjugated BTOs through biotin-neutravidin binding (Fig. 1i). To visually confirm the binding of BTOs and MNDs in the primary cultured cells, we employed Förster Resonance Energy Transfer (FRET). For this purpose, the functionalized 100 nm BTOs (BTO100) and 250 nm MNDs (MND250) were conjugated with Alexa-488 and Alexa-594, respectively. Following applying BTO100 and MND250 to the cultured neurons, illumination with blue light at 470 nm was used to excite the Alexa-488 on the BTO100. This resulted in the emission of green fluorescence from Alexa-488, which subsequently transferred energy to the nearby Alexa-594 on the MND250 (Fig. 1i). The FRET ratio was significantly increased after the application of MNDs, indicating successful binding (Fig. 1j, S1b). Moreover, the Scanning Electron Microscopy (SEM) image of a neuron in the BTO100/MND250 group shows the BTOs were attached to the cell membrane, and the MNDs were sitting on top of BTOs (Fig. 1k). Similar results can be observed in SEM image of neuron with BTO100 and 250 nm HNDs (HND250) (Fig. 1l)
We then measured the magnetoelectric stimulated Ca2+ responses in the cultured hippocampal neurons with a custom-made air-core coil in the upright fluorescence microscope (Fig. 2a, S2). Upon applying the functionalized BTO100 and MND250, magnetic field-induced Ca2+ responses were observed during the application of slow and weak AMF of 50 mT at 10 Hz (Fig. 2b). Notably, repeated AMF at this intensity and frequency elicited multiple Ca2+ responses (Fig. 2e). Such responses were not observed in the control group with BTO100 and HND250 (Fig. 2c, f). The fluorescence responses were significantly higher in the BTO100/MND250 group (Fig. 2h), indicating the specificity of the MagTIES-induced responses. Interestingly, if we rearranged the application sequence to place the MND250 between BTO100 and the cell membrane, the AMF-induced responses were significantly smaller (Fig. 2c, g-h). This result indicates that the arrangement of BTOs between MNDs and cell membranes is critical for inducing neuronal activity.
To further elucidate the temporal precision of MagTIES, we employed Di-8-ANEPPS, a radiometric voltage-sensitive dye (VSD), to measure action potentials in cultured neurons at a temporal resolution of 1 kHz. Di-8-ANEPPS shifts its emission spectrum upon neuronal depolarization when excited at 470 nm, enabling the quantification of membrane potential changes by comparing green to red emission light (Fig. 2i).21,22 Fluorescence imaging revealed that Di-8-ANEPPS predominantly localizes at the plasma membrane (Fig. 2j), ensuring the accuracy of potential change measurements. By applying short alternating magnetic field (AMF) pulses of 100 ms at 50 mT and 10 Hz, we observed MagTIES-induced neuronal spikes within milliseconds (Fig. 2k). Each AMF cycle alternates the external magnetic field between two directions. This bidirectional alternation in a cycle can generate two distinct torque forces by the magnetite nanodiscs (MNDs). Consequently, we recorded one or two spikes per 10 Hz stimulation cycle over 100 ms periods, demonstrating the millisecond-scale temporal precision enabled by MagTIES.
MagTIES with nanoparticles in different sizes
Unlike bulk BaTiO3, the piezoelectric coefficients of single BTOs have been shown to correlate with their size inversely20. Smaller BTOs, under 120 nm, possess higher piezoelectric coefficients, which have a piezoelectric coefficient (d33) more significant than 1500 pC/N20. BaTiO3 nanoparticles larger than 300 nm had d33 smaller than 300 pC/N20. In this context, we further investigated neuronal responses by combining MND250 with BTOs of different sizes, including 100 nm (BTO100), 300 nm (BTO300), and 500 nm (BTO500) (Fig. 3a-c). We found that BTO100/MND250 groups elicited significantly stronger neuronal responses than BTO300/MND250 and BTO500/MND250 groups (Fig. 3d-f). These results indicate that stimulated response is inversely correlated to the size of BTOs.
Conversely, the size of MNDs might also affect the efficacy of MagTIES. The torque force generated by MNDs is positively correlated to the size of nanodiscs. We hypothesized that smaller MNDs might induce less piezoelectric responses from BaTiO3 nanoparticles (BTOs) and thereby affect the neuronal activity induced by MagTIES. To test this hypothesis, we compared the stimulated neuronal responses using MagTIES with BTO100 combined with MND250, 220 nm MNDs (MND220), or 135 nm MNDs (MND135) (Fig. 3g-i). The cultured neurons activated in response to all combinations (Fig. 3g-i). We found that the BTO100/MND250 group had a significantly more robust response than the BTO100/MND135 group (Fig. 3i). The BTO100/MND250 group had a slightly more robust response than the BTO100/MND220 group. However, these two groups have no significant difference (Fig. 3i). These results show that MagTIES with larger MNDs can induce more effective responses. In addition, the cell viability test shows that the cell death rate was meager in both BTO100/MND250 and BTO100/HND250 groups after MagTIES (Fig. S3). These results collectively suggest the effectiveness and biosafety of MagTIES when employing BTO100 combined with MND250; this combination was used in the following stages of this study.
A previous study demonstrated that applying MND250 alone during AMF can activate neuronal activity by triggering the intrinsic mechanosensitive ion channel, TRPC, in cultured neurons4. In MagTIES, a layer of BTOs is positioned between the MNDs and cell membranes, which reduces the direct mechanical force transduced from MNDs to membranes. We found that applying the TRPC-specific antagonist, SKF96365, does not reduce MagTIES-induced neuronal activity (Fig. 4a, d). We further confirmed that MagTIES-induced Ca2+ responses depended on voltage-gated channels (Fig. 4b-e) by using antagonist of voltage-gated Na+ channels, tetrodotoxin (TTX), and the antagonist of voltage-gated Ca2+ channels, mibefradil. These results indicate that TRPC is not critical for the MagTIES-induced activity of voltage-gated ion channels.
MagTIES Induced Neuronal Activity In Vivo
To evaluate the in vivo efficacy of MagTIES, particularly its ability to target deep brain regions, we selected the amygdala, a crucial area for emotion processing located deep within the brain23,24. We utilized stereotactic injection to introduce a combination of BTO100 and MND250 into the amygdala of mice (Fig. 4a). Following a day of recovery, mice were exposed to MagTIES within an 11 cm coil, designed to generate an AMF tailored explicitly for our experiment. The mice were acclimated to the chamber, measuring 10 cm in diameter and 9 cm in height, for 30 minutes before stimulation (Fig. 5a-d). The stimulation protocol consisted of 10 times 30 s stimulation periods interspersed with 30 s rest intervals, with the AMF set at 50 mT and 10 Hz.
To systematically evaluate the effect of MagTIES on neuronal activity, we varied the ratios and amounts of BTOs and MNDs injected. The experimental groups included varying ratios such as 40 µg BTOs with 4 µg MNDs (40B/4M), 20 µg BTOs with 4 µg MNDs (20B/4M), and 4 µg BTOs with 4 µg MNDs (4B/4M). We also examined different total amounts at the same ratio, including combinations like 20 µg BTOs with 20 µg MNDs (20B/20M) and 40 µg BTOs with 40 µg MNDs (40B/40M). Before craniotomy with stereotactic injection, functionalized BTOs and functionalized MNDs were characterized by FRET in vitro to confirm the function of biotin-avidin linkage in each batch of materials (Fig. 1i-j). Intriguingly, we observed that the expression of c-Fos, an immediate early gene marker for neuronal activation, was significantly higher in the injected hemisphere compared to the contralateral side in the 40B/4M and 20B/4M groups (Fig. 4b, d). However, this trend was not as pronounced in the 4B/4M group. No significant differences were observed between the contralateral and ipsilateral sides (Fig. 4d). Interestingly, increasing the total amount of BTO and MND injections in the 20B/20M and 40B/40M groups did not significantly enhance c-Fos expression in the ipsilateral amygdala (Fig. 4d). These findings indicate that MagTIES with a specific amount of 20 µg BTOs and 4 µg MNDs injection can effectively induce neuronal activity in vivo. In contrast, the control group with 20 µg BTOs and 4 µg HNDs injections (20B/4H) showed no significant changes in c-Fos expression (Fig. 4c, e), underscoring the specificity of the MagTIES-induced response.
Modulating the neural oscillation by MagTIES
Brain oscillations are crucial for various functions across different brain regions25–28. Neuromodulation technologies with high temporal precision, such as electrical DBS and optogenetics, can target and modulate these oscillations to specific frequencies27–29. This capability is essential for manipulating neuronal circuitry by controlling neuronal activity with precise timing. However, tuning the frequency of brain oscillation via magnetic nanoparticle-based neuromodulation technologies has not been previously reported. Here, we utilized fiber photometry for recording real-time neuronal activity in vivo30, circumventing interference from magnetic stimulation to the measurements (Fig. 4f). AAV-hSyn-GCaMP7s-WPRE were unilaterally injected into the basolateral amygdala (BLA). After 3 to 7 weeks, the 20 µg BTOs and 4 µg MNDs (20B/4M) or 20 µg BTOs and 4 µg HNDs (20B/4H) were stereotaxic injected into the same location. The optical fiber with 400 µm diameter was implanted into the BLA. Post-surgical recovery, the Ca2+ responses in vivo were measured by fiber photometry in a magnetic apparatus with a diameter of 10 cm and a height of 19 cm (Fig. S5e-g).
First, we perform the MagTIES using 50 mT AMF at 10 Hz for 30 s. In the 20B/4M group, we observed a significant increase in fluorescence change (dF/F) compared to pre-stimulation levels (Fig. 4g-h). In contrast, there was no notable fluorescence change in the control group with 20B/4H (Fig. 4g-h). In 20B/4M group, reducing the magnetic field intensity to 40 mT, 30 mT, and 20 mT still resulted in an observable trend of increased fluorescence, but without statistical significance (Fig. 4i). Interestingly, further analysis using Fast Fourier Transform (FFT) revealed increased brain oscillations at 20 Hz during 10 Hz AMF application in the 20B/4M group (Fig. 4j), a phenomenon not seen in the 20B/4H group. The power spectrum intensity at 20 Hz was significantly increased in 20B/4M injected mice when applying 50 mT AMF (Fig. 4k), which cannot be observed in 20B/4H injected mice. (Fig. 4k). The increased power spectrum intensity trend was also notable when reducing the magnetic field intensity to 40 mT, 30 mT, and 20 mT, but without statistical significance (Fig. 4l).
Different AMF frequencies were used for MagTIES to determine whether the oscillation frequency corresponded to the MagTIES frequency. While we increased the frequency of AMF from 10 Hz to 11 Hz and 12 Hz, we can observe the apparent increment of power intensity at 20 Hz, 22 Hz, and 24 Hz, respectively (Fig. 5a, S6a). Align with our hypothesis, the oscillation of Ca2+ responses was precisely two times the applied frequency of AMF. When using 10 Hz AMF to the 20B/4M group, the change of power intensity at 20 Hz was ~ 7-fold more prominent than the baseline intensity (Fig. 5b, e), which is significantly higher than the change of power intensity in 20B/4H group (Fig. 5b, e). But the shift in power intensity at nearby frequencies doesn’t show the difference between the 20B/4M group and 20B/4H group (Fig. 5e). Similarly, by using MagTIES with 11 Hz, 50 mT AMF, The increase of power spectrum intensity at the 22 Hz was significantly larger in the 20B/4M group than 20B/4H group (Fig. 5c, f). The power intensity was not changed at 20 H and 24 Hz (Fig. 5f). We can observe similar results using MagTIES with 12 Hz and 50 mT AMF. Only a change of power intensity at 24 Hz was increased in the 20B/4M group compared to the 20B/4H group (Fig. 5d), which cannot be observed in other frequencies (Fig. 5g).
Finally, we observed that MagTIES induced responses even when the AMF stimulation period was reduced to 5 seconds or 1 second. Specifically, applying AMF for a 5-second stimulation period, interspersed with 5-second intervals for three cycles, led to a marked increase in power spectrum intensity, notably at 20 Hz during the stimulation periods (Fig. 5h-i). The change in power intensity at 20 Hz, observed 5 seconds before and after stimulation, was significantly greater in the 20B/4M group than in the 20B/4H group (Fig. 5j, S6b). A similar pattern emerged when AMF was applied for 1-second periods, followed by 1-second resting periods for 15 cycles. A significant increase in power spectrum intensity at 20 Hz was noted during these 1-second stimulation periods (Fig. 5k-l). The rise in power intensity at 20 Hz was also significantly more significant in the 20B/4M group (Fig. 5m, S6c). These fiber photometry results indicate the unique capability of MagTIES to manipulate neural oscillations in the amygdala with a high degree of temporal and frequency specificity, marking a significant advancement in the field of neuromodulation.