Spatiotemporal terahertz modulation enhances NMDAR-mediated miniature EPSCs

Abstract

N-Methyl-D-aspartate receptors (NMDAR) are essential for synaptic plasticity and cognitive function, making their modulation a promising strategy for treating disorders like schizophrenia and cognitive impairment. However, methods to selectively modulate NMDAR activity in the lesion’s nucleus of the central nervous system remain limited. In this study, using whole-cell patch-clamp recordings, we demonstrated that frequency-specific (42.5 THz) terahertz irradiation significantly enhanced both the frequency and amplitude of NMDAR-mediated miniature excitatory postsynaptic currents (mEPSCs), a response closely linked to Ca²⁺ currents. The mechanism is elucidated via molecular dynamics (MD) simulations, revealing that 42.5 THz irradiation effectively alters the free energy landscape of Ca²⁺ permeating through the NMDAR channel. Specifically, THz photons resonated with key carboxyl groups at the Ca²⁺ binding site, leading to an increase in Ca²⁺ permeability and consequently enhanced mEPSCs. These findings suggest a novel physical therapy approach for treating cognitive deficits and neurological disorders associated with impaired NMDAR function.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-08207-8.

Introduction

NMDA receptors are multi-subunit, ligand-gated channels that are highly expressed in the central nervous system of mammals and humans¹. They co-mediate the majority of excitatory synaptic transmission in the central nervous system alongside alpha-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptors^2,3. Following activation by glutamate and its co-agonists, the NMDA receptor opens a non-selective cation channel that exhibits high permeability to Ca²⁺, thereby generating excitatory postsynaptic currents (EPSCs)^4,5.

NMDAR plays a key and complex role in synaptic plasticity and cognitive function^6,7. Overactivation or hypofunction of these receptors can lead to the onset or progression of various brain diseases⁸. Specifically, attenuation of NMDAR neurotransmission can result in the loss of neuronal plasticity and cognitive deficits^9–11, while enhanced NMDAR neurotransmission has been shown to improve cognitive and behavioral deficits^12,13. Previous efforts to regulate NMDAR function are mostly focused on drug development^14–16. However, due to the limited ability of drugs to specifically target affected brain regions, serious side effects often arise^17,18. Thus, enhancing NMDAR function at the appropriate target sites, specifically at synapses with NMDAR hypofunction, could mitigate the risk of excessive activation in regions where NMDAR signaling remains intact, thereby reducing the likelihood of eliciting other types of neurological disorders, such as Huntington’s disease and chronic pain^19,20. Despite the superiority of non-invasive techniques such as repetitive transcranial magnetic irradiation (rTMS)²¹ and focused ultrasound (FUS)²² utilized for neurological therapies, there is a growing need for precise, lesion-targeted approaches to modulate or repair NMDAR function^23,24, yet such methods remain underexplored.

THz waves, acting in a non-thermal, reversible, and remotely manner, have emerged as a promising tool in modulating molecule structure for biomedical applications due to absorption characteristics of biomolecules in the generalized THz band^25,26. Previous research indicated that frequency-specific THz photons can be resonantly absorbed by various biological molecules, including nucleic acids, proteins, and neurotransmitters^27–32, leading to structural changes and ultimately modulating macro-biological functions^33–35. In the field of neurobiology, THz waves/photons have also shown potential in the treatment of mental disorders. For example, Yu et al. demonstrated that 34.5 THz irradiation can improve cognitive function in mice with post-traumatic stress disorder (PTSD)³⁶. Additionally, Peng et al. found that 36 THz irradiation can alleviate neuropathic pain in mice by inhibiting the activity of pyramidal neurons in the anterior cingulate cortex³⁷. Furthermore, 53.6 THz photons have been shown to enhance the excitability of cortical neurons, improving the learning efficiency of mice by 50% during sound association learning tasks³⁸. These results highlight the advantage of THz photons in modulating neuronal excitability through direct regulation of the signaling molecules or ions associated. Notably, our theoretical study suggests that 42.5 THz photons could increase the permeability of the second messenger (Ca²⁺) by promoting hydrogen bonding between hydrated Ca²⁺ and key functional groups within voltage-gated calcium channels (VGCCs)³⁹. Given the importance of neuronal calcium signaling, it is pressing to experimentally validate this predicted non-thermal modulation so as to promote the THz-based physical intervene means into medical use. The NMDA receptor channel, known for its high permeability to Ca²⁺, is a suitable candidate for experimental study of the THz modulation of NMDAR-mediated EPSCs in neurons and the underlying mechanism.

In this study, by exposing mouse neurons to THz photons at different frequencies and recording NMDAR-mediated mEPSCs using whole-cell patch-clamp techniques, we observed distinct frequency-dependent response patterns of NMDAR-mediated mEPSCs, providing direct experimental evidence supporting our previous theoretical prediction that frequency-specific terahertz photons enhance Ca⁺ currents during neural activity. Additionally, MD simulations further validated these findings and identified critical factors underlying the effects of 42.5 THz photons on these mEPSCs.

Results

To isolate NMDAR-mediated mEPSCs, we added 20 µM CNQX to Mg²⁺ free artificial cerebrospinal fluid (ACSF) to block AMPA receptor. The target neurons were illuminated with a 3 mW THz light source from a distance of 300 μm (Fig. 1A). In the absence of THz irradiation, the average frequency of NMDAR-mediated mEPSCs was 0.08 Hz. In contrast, exposure to 42.5 THz irradiation significantly increased this frequency to 0.2 Hz (Fig. 1B, C). Although the Kolmogorov-Smirnow test showed no statistical difference in the probability distribution of the inter-event intervals of NMDAR-mediated mEPSCs before and during 42.5 THz exposure, the overall leftward shift in the cumulative probability curve indicated an increase in the frequency of NMDAR-mediated mEPSCs. Meanwhile, the amplitude of NMDAR-mediated mEPSCs also showed a significant increase with the THz irradiation, accompanied by a rightward shift in the cumulative probability curve (Fig. 1D).

Fig. 1.

Exposure to 42.5 THz photons significantly enhanced both the frequency and amplitude of NMDAR-mediated mEPSCs in neurons. (A) Schematic of THz irradiation on NMDA receptors. During THz exposure, AMPA receptors on the cell membrane were blocked using the specific antagonist CNQX, and extracellular Mg²⁺ was removed to enhance the NMDA currents. (B) Representative traces of NMDAR-mediated mEPSCs before (black curve) and during (orange curve) exposure to 42.5 THz irradiation (cell = 8). (C) Left panel: Frequency of NMDAR-mediated mEPSCs before and during 42.5 THz irradiation. Paired sample Wilcoxon signed-rank test, * p < 0.05. Right panel: Cumulative probability curve of the inter-event intervals of NMDAR-mediated mEPSCs before and during 42.5 THz irradiation. K–S test, p = 0.2827. (D) Left panel: Amplitude of NMDAR-mediated mEPSCs. Paired sample Wilcoxon signed-rank test, ** p < 0.01. Right panel: cumulative probability curve of the amplitude of NMDAR-mediated mEPSCs. K–S test, p = 0.2827. (E) Representative traces of NMDA-mediated mEPSCs before (black curve) and during (blue curve) 34.5 THz irradiation (cell = 8). (F) Left panel: frequency of NMDA-mediated mEPSCs before and during 34.5 THz irradiation. Paired sample Wilcoxon signed-rank test, p = 0.2500, where ns denotes no significant difference. Right panel: cumulative probability curve of the inter-event intervals of NMDA-mediated mEPSCs before and during 34.5 THz irradiation. K–S test, p = 0.9801. (G) Left panel: Amplitude of NMDA-mediated mEPSCs. Paired sample Wilcoxon signed-rank test, p = 0.1484. Right panel: Cumulative probability curve of the amplitude of NMDA-mediated mEPSCs. K–S test, p = 0.9801. (H) Left panel: The temperature sensor monitored temperature changes caused by THz irradiation at a distance of 300 μm from the optical fiber surface. Right panel: The temperature growth induced by 42.5 and 34.5 THz irradiation (n = 5). Mann-Whitney U-test, p = 0.8413.

In a separate set of experiments conducted under similar conditions, we recorded NMDAR-mediated mEPSCs while providing 34.5 THz irradiation; however, this radiation did not produce significant effects on either the frequency or amplitude of the NMDAR-mediated mEPSCs (Fig. 1E-G). To examine whether the enhancements in NMDAR-mediated mEPSCs by THz photons were due to thermal effects, we monitored the temperature growths during 42.5 THz and 34.5 THz irradiation, finding negligible increases of 0.1090 ℃ and 0.1066 ℃, respectively (Fig. 1H). This implied that frequency-specific THz photons could non-thermally enhance both the frequency and amplitude of NMDAR-mediated mEPSCs.

To confirm the contribution of NMDA receptors to the recorded mEPSCs, we added the NMDA receptor antagonist AP5 (50 μm) to the ACSF (Fig. 2A). In the subsequent recordings from 5 cells, we observed a complete absence of mEPSC events, and 42.5 THz irradiation was unable to induce any mEPSCs (Fig. 2B). In contrast, under conditions with only CNQX present, three out of the eight recorded cells exhibited a complete absence of mEPSCs. However, 42.5 THz irradiation reversed this effect, resulting in an increase in mEPSCs from a baseline of zero to an average frequency of 0.145 Hz, with an average amplitude of 6.11 pA (Figs. 1C, D). These findings indicate that the mEPSCs recorded in the presence of 20 µM CNQX were mediated by NMDA receptors.

Fig. 2.

Exposure to 42.5 THz photons enhanced the amplitude of NMDA-mediated mEPSCs in neurons by increasing Ca²⁺ permeability. (A) NMDA receptors were blocked using the NMDA receptor antagonist AP5, while THz irradiation was simultaneously applied to the cells. (B) AP5 completely inhibited NMDAR-mediated mEPSCs (black curve), even in the presence of 42.5 THz irradiation (orange curve) (cell = 8). (C) Schematic representation indicating the removal of Ca²⁺ from the ACSF. (D) Representative trace of NMDAR-mediated mEPSCs in Ca²⁺ free ACSF before (black curve) and during (orange curve) the exposure to 42.5 THz irradiation (cell = 7). (E) Left panel: Frequency of NMDAR-mediated mEPSCs before and during 34.5 THz irradiation. Paired sample Wilcoxon signed-rank test, p = 0.2500, indicating no significant difference. Right panel: Cumulative probability curve of the inter-event intervals of NMDAR-mediated mEPSCs before and during 34.5 THz irradiation. K–S test, p = 0.5752. (F) Left panel: Amplitude of NMDAR-mediated mEPSCs. Paired sample Wilcoxon signed-rank test, p = 0.6250, denoting no significant difference. Right panel: Cumulative probability curve of the amplitude of NMDAR-mediated mEPSCs. K–S test, p = 0.9627.

To investigate the role of Ca²⁺ in the enhancement of NMDAR-mediated mEPSCs by 42.5 THz irradiation, we subsequently removed Ca²⁺ from the ACSF (Fig. 2C). Under conditions of extracellular Ca²⁺ deficiency, we observed that 42.5 THz irradiation lost its ability to enhance both the frequency and amplitude of NMDAR-mediated mEPSCs (Figs. 2D-F). Specifically, when the brain slices were perfused with Ca²⁺ free ACSF, the frequency of NMDAR-mediated mEPSCs did not show a significant increase upon the application of 42.5 THz irradiation, and there was no evident trend in the probability distribution of the inter-event intervals of NMDAR-mediated mEPSCs. Furthermore, the amplitude of NMDAR-mediated mEPSCs also remained unchanged. These results suggest that Ca²⁺ is indispensable in the process by which THz irradiation enhances NMDAR-mediated mEPSCs.

NMDA receptors are essential heteromers, composed of tetramers formed by the coassembly of various combinations of GluN1, GluN2A-GluN2D, GluN3A, and GluN3B subunits, which together create Ca²⁺-permeable ion channels⁴⁰. To elucidate the mechanism underlying 42.5 THz photons enhanced NMDAR-mediated mEPSCs, we conducted the MD simulations to explore the effect of THz irradiation on Ca²⁺ permeation through the NMDA receptor channel. The NMDAR employed in our MD simulations comprises two GluN1b and two GluN2B subunits, which are activated by the co-agonist glycine and the neurotransmitter agonist glutamate⁴¹. Agonists and competitive antagonists interact with the receptor at ligand-binding domains (LBDs), while allosteric modulators exert their regulatory effects on ion channel gating by binding to either the amino terminal domains (ATDs) or the LBDs, thereby influencing the transmembrane domain (TMD)⁴². Analysis of crystal structures highlighted the significance of the Asp-Arg-Pro-Glu-Glu-Arg motif (residues 679–684) in GluN1b subunits, which played a crucial role in calcium binding and high flux between the LBD-TMD linkers⁴³ (Fig. 3A). By analyzing the vibrational spectra of these key residues especially the charged ones in whole or in part, we identified that, the peak at 42.5 THz was primarily derived from the carboxyl groups of Asp and Glu, while the 34.5 THz peak originated from the guanidinium group of Arg (Fig. 3B). We then calculated the free energy landscapes of single Ca²⁺ permeating the channel via metadynamics simulations conducted by GROMACA-patched PLUMED^44,45. The energy maps were drawn as a function of the radial (r) and axial (z) distances from the origin, which was the lower gate for the calcium flux (Fig. 3A, right panel).

Fig. 3.

Mechanism underlying 42.5 THz irradiation modulating NMDA-mediated mEPSCs. (A) Left panel: schematic of Ca²⁺ passing the NMDA receptor under THz irradiation (orange light). Right panel: magnified view of the boxed region in left panel, highlighting Ca²⁺ (blue ball) passing through the putative calcium binding site, composed of residues 679–684 (pink chain). (B) Vibration spectra of bulk water (gray curve) and side chains of binding residues (orange curve for the carboxyl groups of Asp and Glu; green curve for the guanidinium group of Arg). (C-E) Heat maps of the free energy profiles under conditions without external irradiation (C) and with irradiations at 42.5 THz (D), 34.5 THz (E). (F-H) The number of hydrogen bonds between hydrated Ca²⁺ and Asp679 (F), Glu682 (G) and Glu683 (H) at the binding site with irradiation at 42.5 THz.

As shown in Fig. 3C, in the absence of THz irradiation, there were basically three energy minima, corresponding to the binding sites for Ca²⁺. More specifically, markers 1 and 2 denote binding sites at Asp-679 from two GluN1b subunits in distinct Ca²⁺-coordinated ways, that is, the calcium bound -COO⁻ of Asp-679 in GluN1b(α) was oriented inward the channel lumen, while the GluN1b(β) counterpart was directed outward (Figs. S1A-B). Notably, the former positioned the calcium ion closer to the channel lower gate, directly facilitating its passage through the pore, while the latter supported ion translocation synergistically. On the contrary, since the two Glu-682 bound Ca²⁺ in the same way (Fig. S1C and Fig. S2), they formed a single binding site at the marker 3. Overall, the energy map indicated a preference for Ca²⁺ passage along the channel lumen rather than its center due to strong electrostatic attractions from the negatively charged -COO⁻ groups of Asp and Glu.

When a 42.5 THz THz irradiation was applied along the z-axis, the initial minimum at marker 3 was eliminated, and most importantly the free energy at marker 1’ decreased significantly from − 209 kJ/mol to −220 kJ/mol, facilitating Ca²⁺ passage through the channel (Fig. 3D). The underlying mechanism lied in the resonant enhancement of the -COO⁻ vibrations by the 42.5 THz photons, leading to a rearrangement of calcium binding coordination (Fig. S1D). Since Ca²⁺ interacted with the -COO⁻ groups through its hydration waters, the hydrogen bonds formed in between played a crucial role in Ca²⁺ permeation. We therefore calculated the number of hydrogen bonds between Ca²⁺ shell waters and -COO⁻ groups of key residues at three binding sites. Figures 3F-H all demonstrated an increase of the hydrogen bond number under the 42.5 THz irradiation, with a notable growth from 2.48 to 3.74 for hydrogen bonds formed with Asp-679. This exactly aligned with the free energy drop (enhanced affinity) at the marker 1’ in Fig. 3D. It follows that 42.5 THz irradiation significantly enhanced NMDA receptor channel permeability through carboxyl resonance. In contrast, under the 34.5 THz irradiation in resonance with vibrations of Arg side chains instead of the -COO⁻, no energy minima near the lower gate were observed (Fig. 3E and Fig. S3) and hence no enhanced Ca²⁺ permeability. These findings underscored the importance of frequency-specific and core residues-targeted regulation in modulating the NMDAR function.

Discussion

Proper development and refinement of neural circuit require sufficient functionality and activity of NMDARs⁴⁶. In this study, we demonstrated that 42.5 THz photons could enhance NMDA receptor-mediated excitatory synaptic transmission and the enhancement was Ca²⁺-dependent. Additionally, this effect exhibited frequency selectivity and dependence on the channel structure. In fact, numerous studies have investigated the effects of THz radiation on biological functions⁴⁷. The results indicate that cell types, exposure parameters (such as frequency, power, duration), and thermal conditions vary significantly across the studies. Our data further contribute to the understanding of the biological effects of THz radiation under specific parameter conditions.

More importantly, there is an urgent clinical need for disease intervention methods with high spatiotemporal precision⁴⁸, while terahertz therapy can effectively complement this underdeveloped field. The development of neurological diseases is usually caused by dysfunction in specific nuclei or neural circuits^49–51. Correct manipulation of local neural circuits can reverse related disease phenotypes^51–54. In our research, we employed THz photons delivered via optical fibers to specifically target brain regions, enhancing NMDAR-mediated mEPSCs in the lateral septum (LS). LS neurons integrate excitatory inputs from the hippocampus and other regions to provide context-dependent (top-down or higher-order) regulation of emotion and motivational behavior⁵⁵. This innovative approach has promising implications for improving learning capacities^56,57, suppressing obesity⁵⁸, and alleviating anxiety and depression⁵⁹. Research studies have demonstrated that deficiency of NMDARs in neurons within the LS was closely associated with impaired social behaviors in animals^60,61. And the loss of NMDAR function in LS neurons may serve as a critical factor contributing to cognitive dysfunction following stroke⁶². Additionally, in schizophrenia research, accumulating evidence suggests that hypofunction of NMDA receptors on γ-aminobutyric acid GABAergic neurons may lead to underdeveloped GABAergic circuitry and reduced inhibitory tone, thereby inducing various schizophrenia-like symptoms⁵⁰. Given that LS neurons are predominantly GABAergic⁶³, our findings may establish a novel therapeutic strategy for schizophrenia and other central nervous system disorders associated with NMDAR hypofunction. To explore this potential, future research will be crucial in utilizing animal models of psychiatric disorders to determine whether 42.5 THz photons can effectively reverse impaired NMDAR-mediated excitatory synaptic transmission and alleviate the associated pathological symptoms.

Compared to optogenetics or chemogenetics that require exogenous gene delivery, our THz irradiation method achieves comparable spatial precision in neural modulation without genetic manipulation, thereby eliminating risks associated with exogenous gene delivery. Additionally, current evidence suggests that THz radiation within certain parameters may have limited detrimental effects on cellular systems. Studies have demonstrated that THz exposure did not induce genotoxicity, morphological changes, or heat shock protein expression in HCE-T cells⁶⁴, nor did it cause chromosomal damage or alter cell cycle dynamics⁶⁵. Additionally, THz exposure did not result in DNA damage or affect skin cell proliferation⁶⁶. At the in vivo level, research by Zhang et al. revealed that THz irradiation could modulate cortical neuron activity in mice while maintaining stable intracellular Ca²⁺ levels and neuronal viability post-irradiation, as observed through two-photon microscopy³⁸. However, while these findings indicate minimal toxicity, the potential long-term effects of THz exposure on neuronal physiology remain unclear. Therefore, comprehensive investigation of wavelength-specific THz effects on multidimensional neuronal activity parameters is warranted, with particular attention to identifying potential adverse physiological impacts that could inform safety thresholds for biomedical applications.

Through molecular dynamics simulations, we revealed that 42.5 THz electromagnetic fields facilitate Ca²⁺ permeation through NMDAR channels, which may contribute to an increased amplitude of NMDAR-mediated mEPSCs. Since the amplitude of mEPSCs is predominantly determined by the quantity or functional state of postsynaptic glutamate receptors⁶⁷, and considering that our brief THz irradiation protocol theoretically would not induce significant increases in the quantity of postsynaptic NMDARs⁶⁸, the observed enhancement in amplitude can primarily be attributed to improved NMDAR activity. Furthermore, THz irradiation significantly increased the frequency of NMDAR-mediated mEPSCs, potentially resulting from enhanced presynaptic glutamate release^69,70. Our previous study revealed that 42.5 THz fields can promote Ca²⁺ permeation through VGCCs³⁹, which are the primary drivers of presynaptic neurotransmitter release⁷¹. Thus, 42.5 THz irradiation may augment presynaptic calcium influx via VGCCs, thereby potentiating glutamate release and ultimately increasing the frequency of NMDAR-mediated mEPSCs.

Previous studies have also demonstrated that 30–45 THz photons can resonate with glutamate or GABA, enabling their absorption of photon energy to transition into higher energy states, thereby enhancing both EPSCs and inhibitory postsynaptic currents (IPSCs)⁷². This finding strongly supports the active role of THz photons in modulating neuronal signal transduction. However, the relatively high THz power (averaging 30 mW) used in this previous work raises the possibility that accompanying thermal effects could also be an unavoidable factor in enhancing of both EPSCs and IPSCs. In our study, we employed THz photons at an average power of 3 mW to enhance NMDAR-mediated mEPSCs through non-thermal mechanisms, further elucidating the target action of THz photons. In fact, there are also some limitations. We only assessed the impact of 42.5 THz radiation on NMDAR-mediated mEPSCs over a duration of 3–5 min. Whether 42.5 THz photons can sustain enhanced EPSCs in neurons over a longer time span remains to be validated at the individual level.

Conclusion

In conclusion, we presented, for the first time, enhanced mEPSCs via non-thermal, and spatiotemporally-controlled modulation of NMDAR by 42.5 THz photons, which was attributed to increased calcium flux. The mechanism was uncovered resonant absorption of THz photons by -COO⁻ groups along the Ca²⁺ permeation pathway in NMDAR channels, altering the free energy landscapes and promoting Ca²⁺ permeability in a frequency-dependent way. This work not only provides direct experimental evidence for THz field promoted Ca²⁺ channel conductance, but suggests a potential therapeutic strategy for neurodegenerative diseases associated with reduced NMDA receptor function.

Materials and methods

Animals

All experiments were conducted in accordance with relevant guidelines and regulations, and approved by the IACUC (Institutional Animal Care and Use Committee) of SIAT, Chinese Academy of Sciences (CAS, SIAT-IACUC-230927). In this study, we utilized adult male C57BL/6J mice (2–3 months old) from the Charles River Laboratories (Beijing, China). They were housed under a 12-h light and 12-h dark cycle at a temperature of 22–25 °C. All efforts were made to minimize animal suffering and the number of animals used. All methods used in this study conformed to the ARRIVE guidelines (version 2.0).

Ex vivo electrophysiology

Mouse brain slices were prepared according to previously established methods⁷³. Mice were anesthetized using isoflurane and subsequently decapitated. The brains were quickly extracted and transferred to a slicing solution maintained at 4 °C. Prior to this, the slicing solution was saturated with 95% O₂ and 5% CO₂ for at least 15 min. The composition of the slicing solution included (in mM): 110 choline chloride, 0.5 CaCl₂, 2.5 KCl, 25 NaHCO₃, 1.3 NaH₂PO₄, 7 MgCl₂, 1.3 Na-ascorbate, 0.6 Na-pyruvate, and 25 glucose. During the slicing procedure, a continuous supply of 95% O₂ and 5% CO₂ was maintained. Coronal slices of 300 μm thickness were prepared using a vibratome (Leica VT1200). These slices were stored in ACSF consisting of (in mM): 124 NaCl, 3.0 KCl, 1.5 CaCl₂, 1.0 NaH₂PO₄, 26 NaHCO₃, and 20 glucose, which was also saturated with 95% O₂ and 5% CO₂ at 32 °C. After thirty minutes in the 32 °C environment, the slices were transferred to room temperature (23–25 °C).

Whole-cell patch-clamp recordings were conducted using a Multiclamp 700B amplifier and a Digidata 1550B system (Molecular Devices). Data acquisition was performed at a sampling rate of 10 kHz and subsequently analyzed using either Clampfit (Molecular Devices) or MATLAB (MathWorks).

For whole-cell voltage-clamp recordings, patch pipettes (3–6 MΩ) were filled with a Cs-based low Cl⁻ internal solution containing (in mM): 135 CsMeSO₃, 10 HEPES, 1 EGTA, 3.3 QX-314, 8 Na₂-phosphocreatine, 4 Mg-ATP, 0.3 Na-GTP, and adjusted to 290 mOsm kg⁻¹ (pH adjusted to 7.3 with CsOH). To record NMDA-mediated miniature excitatory postsynaptic currents (mEPSCs), artificial cerebrospinal fluid (ACSF) without Mg²⁺ (Mg²⁺-free ACSF) was utilized^5,74,75. To eliminate the interference of action potentials and inhibitory postsynaptic currents, 1 µM tetrodotoxin (TTX) and 100 µM picrotoxin (PTX) were added to the bath solution. Additionally, an AMPA receptor antagonist, CNQX (20 µM), was applied to the bath. After seal formation and membrane rupture, the cells were allowed to stabilize for 3–5 min before commencing pulse protocols. NMDA-mediated mEPSCs were recorded for 3–5 min. Subsequently, terahertz irradiation was applied to the region of the recorded neurons using a dedicated optical fiber, and this process continued for at least 3.5 min. We selected the mEPSC data from the 30 s following the 42.5 THz irradiation until the end of the irradiation as the mEPSCs of the LS neurons under THz irradiation, which can be considered the THz irradiation group. Given that the mEPSCs before and during the THz irradiation were recorded continuously from the same cell, we used the paired sample Wilcoxon signed-rank test to compare the statistical differences between the two conditions.

The data for this study were obtained from a total of 12 mice. The experimental data on the effect of 42.5 THz on mEPSCs in neurons were obtained from 10 mice. Of course, these 10 mice provided data on three sections: the effect of 42.5 THz irradiation on mEPSCs (Fig. 1B and D), the confirmation of NMDA receptor-mediated mEPSCs (Fig. 2B and C), and the effect of 42.5 THz irradiation on mEPSCs in the absence of Ca²⁺ (Fig. 2D and F). The experimental data on the effect of 34.5 THz irradiation on mEPSCs in neurons were obtained from 5 mice, among which 3 mice also provided data for the 42.5 THz irradiation group.

Light source and in vitro THz photon irradiation

The 42.5 THz or 34.5 THz irradiation was provided by a quantum cascade laser (QCL). The laser pulse lasted for 2 microseconds, featuring a repetition rate of 200 kHz and a duty cycle of 90%. The THz photons generated by the quantum cascade laser (QCL) were focused through a THz lens before being directed into a THz optical fiber (Artphotonics, PIR240/300). This optical fiber is a standard polycrystalline composite made from infrared AgCl: AgBr, with a core diameter of 240 ± 15 μm, a clad diameter of 300 + 0/−15 μm, and a numerical aperture of 0.3 ± 0.03. At a distance of approximately 300 μm from the recorded neurons, the optical fiber emitted photons at a power output of 3 mW. During the terahertz irradiation, temperature variations in the target brain region were monitored using a temperature control instrument (Warner TC-344 C).

Molecular dynamics simulations

The simulation system is constructed with the membrane builder of the Charmm-GUI⁷⁶. It consists of an NMDA receptor embedded in dipalmitoyl phosphatidylcholine bilayer lipid membranes (oriented along the z-axis). The system is immersed in KCl solution at an ionic concentration of 0.15 mol/L, and the water model employed is the TIP3P model. The dimensions of the simulation system are X = Y = 130 nm and Z = 180 nm. It contains 430 1-Palmitoyl-2-oleoyl-sn-glycero-3-phosphorylcholine molecules, 66,514 water molecules, 180 K⁺, 182 Cl⁻, and 1 Ca²⁺.

The initial crystal structure of the protein was obtained from the RCSB data bank (PDB entry: 6WHT), and the GluN1b-GluN2B NMDA receptor in active conformation was resolved at 4.4 angstrom resolution⁴². The direction of the electric field is along the z-axis with a strength of 0.6 V/nm. We performed metadynamics simulations with Gromacs⁴⁴ and PLUMED code⁴⁵ using the Charmm-36 m force fields to explore efficiently the entire free energy surface and calculated the free energy as a function of the metadynamics collective variables. We selected the midpoint between residues 678 as the origin for the metadynamic simulations, establishing a grid that extends from 0.0 to 1.6 nm in the radial direction and from 0.0 to 2.0 nm in the axial direction. The widths of the Gaussian hills were set to 0.05 and the biasfactor was set to 15. The entire system was simulated for 200 ns, eventually reaching equilibrium.

To obtain infrared vibrational spectra, we recorded the time-dependent evolution of the total dipole moment during MD simulations, computed its autocorrelation function, and performed Fourier transformation. The infrared vibrational intensity at frequency ω can be expressed as:

where ω denotes frequency, t represents time, and µ(t) is the dipole moment autocorrelation function.

We multiply the velocity of each particle in the trajectory file by its charge, replacing the original velocity v in the trajectory file with the current vector vq. We then compute the autocorrelation function of the current vector vq and obtain the vibrational spectrum through the Fourier transform of this autocorrelation function. The Fourier-transformed vibrational spectrum is generated using GROMACS built-in tool, gmx velacc. Following system equilibration, the spectral sampling interval is set to 1 fs, with a total sampling time of 100 ps for the calculation.

Statistical methods

For the frequency and amplitude of mEPSCs before and during THz irradiation, we employed the Paired sample Wilcoxon signed-rank test using GraphPad 9.0. For comparing the thermal effects caused by different THz irradiations, we used the Mann-Whitney U-test. For comparing the probability density distributions of the time intervals and amplitudes of mEPSCs before and during THz irradiation, we used the Kolmogorov-Smirnov (K-S) test. All statistical data can be found in the figure legends. Statistical significance was set at p < 0.05. Data are presented as means ± SEM.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Author contributions

Conceptualization: Y. L., and Z. Z. Methodology: Y. L. and Z. Z. Investigation: S. J., Y. Z., P. C. and A. W. Data curation: S. J., Y. Z., P. C., J. Z. and A. W. Funding acquisition: Z. Z., and Y. L. Writing-original draft: S. J. and Y. Z. Writing-review & editing: Y. L. and Z. Z. All authors reviewed the manuscript.

Data availability

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Declarations

Competing interests

The authors declare no competing interests.