Fever raises body temperature (T_b) from ~37 °C to beyond 38.4 °C to combat pathogens. While generally well tolerated below 40 °C, in rare cases fever can abnormally elevate neural activity and induce seizures in neurotypical children aged 2–5 years. This study investigates the mechanisms by which neuronal activity is maintained and stabilized during exposure to fever-range temperatures. Recordings of layer (L)4-evoked spiking in L2/3 pyramidal neurons (PNs) of mouse somatosensory cortex revealed four outcomes as temperature increased from 30°C to 36 °C and 39 °C (fever-range): neurons remained inactive, stayed active, ceased activity, or initiated activity. Roughly equal proportions of neurons ceased or initiated spiking, making the subset of “STAY” PNs, those that remain active across temperatures, crucial for maintaining stable cortical output. STAY PNs were more prevalent at younger postnatal ages. Their firing stability was supported by a distinct ion channel composition, including the thermosensitive channel TRPV3, which enables continued spiking by adjusting depolarization to meet spike threshold. Intracellular blockade of TRPV3, but not TRPV4, significantly reduced the proportion of STAY PNs and suppressed spiking at 39 °C. Moreover, in Trpv3^-/- mice, temperature increases to 39 °C reduced both spiking and postsynaptic potential amplitude, and these mice exhibited a delayed seizure onset. Together, these findings suggest that TRPV3 contributes to the preservation of cortical activity during fever.

Animals

All experimental procedures were performed in accordance with the National Institutes of Health guidelines for care and use of laboratory animals and the EC Council Directive of September 22, 2010 (2010/63/EU). All experiments were approved by the Animal Care and Use Committee of the National Institute on Alcohol Abuse and Alcoholism (NIAAA) (Protocol # LIN-MA-1) and the Animal Care Committee of the HUN-REN Research Centre for Natural Sciences (RCNS) and by the National Food Chain Safety Office of Hungary (license number: PE/EA/1004-5/2021). Mice were housed under standard conditions with ad libitum access to food and water, under a 12-h light/dark cycle. The following mouse strains were obtained from The Jackson Laboratory (JAX):C57BL/6J (JAX 000664), FVB/N mice (JAX 004828), and Trpv3 (JAX 010773) (Moussaieff A et al., 2008; Moqrich et al., 2005). The day of birth was denoted as postnatal day (P)1. Both male and female mice were used for body temperature (T_b) recordings and electrophysiological experiments.

Body Temperature Measurements

All mice used for body temperature (T_b) recordings were implanted with sterile IPTT-300 Implantable Programmable Temperature transponders (Bio Medic Data Systems, LLC) at 2 weeks of age. The injection site was pre-cleaned with a betadine solution, and the sterile transponders were injected subcutaneously using a pre-loaded 12-gauge syringe into animals anesthetized with 4-5% isoflurane. The injection site was on the left side, approximately 11 mm from the base of the hip, with the animal in the prone position. Post-implantation, the mice were able to move, eat, and drink autonomously, with no adverse phenotypes noted. T_b recording commenced at least 5 days after injection. Prior to recording, mice were brought into the procedure room and allowed to habituate for 1 hour. T_b was recorded every 5 minutes for 6 hours using an IPTT-300 Implantable Programmable Temperature reader (DAS-8027) in a 37L x 16W x 13H cm cage. T_b was recorded in the presence (Figure 1D) or absence (Figure 1A-C) of infrared light. Infrared light exposure was provided via a 250-W temperature-controlled infrared lamp (catalog #50320, Stoelting), set to 32.5-33.5 °C. The lamp was positioned ~15 cm above the cage, and the ambient temperature was monitored every 5 minutes (Figure 1D).

Slice Preparation

Using standard methods, acute primary somatosensory cortex (S1) slices (350 µm thick) from P7-8, P12-14, or P20-23 mice were cut in the "across-row" plane, oriented 35° toward coronal from midsagittal (Antoine et al., 2019). The cutting solution contained (in mM): 85 NaCl, 75 sucrose, 25 D- (+)-glucose, 4 MgSO₄, 2.5 KCI, 1.25 NaH₂PO₄, 0.5 ascorbic acid, 25 NaHCO₃, 0.5 CaCl₂. Once cut, slices were transferred to a submerged-style holding chamber containing standard Ringer's solution (in mM: 119 NaCl, 2.5 KCI, 1.3 MgSO₄, 1 NaH₂PO₄, 26.2 NaHCO₃, 11 D-(+)-glucose and 2.5 CaCl₂) and then incubated at 32°C for 30 minutes. Both solutions were at neural pH (i.e., 7.3), 300 mOsm, and saturated with 95% O₂ and 5% CO₂. Slices were kept at room temperature for at least 30 minutes before being transferred to a submerged recording chamber.

In Vitro Physiology

Whole-cell current-clamp recordings were made from pyramidal neurons (PNs) visually identified via infrared DIC optics. Physiological verification for regular spiking was done in current clamp. Recordings were made using 3-6 MΩ micropipettes containing (in mM):116 K gluconate, 20 HEPES, 6 KCI, 2 NaCl, 0.5 EGTA, 4MgATP, 0.3 NaGTP, 10 Na phosphocreatine, with a Multiclamp 700B amplifier (Molecular Devices, Sunnyvale, CA). Where applicable, forsythoside B (50 µM, TRPV3 blocker, Millipore Sigma Cat #: PHL83313) or RN1734 (10 µM, TRPV4 blocker, Tocris Bioscience Cat #: 3746/25) was added to the internal solution. Camphor (MedChem, Cat. No: HY-N0808) was added to the bath solution for experiments in Figure 8.

Signals were filtered (2-6 kHz) and digitized (10-20 kHz). Perfusate temperature for all in vitro experiments was regulated by a PC-connected Peltier heater (SM-4600, Scientifica, UK) and temperature controller (Linlab2, Scientifica, UK), with a high-accuracy, low-noise temperature control system, containing both a built-in temperature sensor and bath sensor for accurate feedback. ACSF flow rates of ~3 ml/min were used to facilitate efficient heating/cooling.

To determine the effect of temperature on an activated cortical network, layer 4-evoked spiking was quantified during perfusate temperature increases from 30°C to 36°C, and then to fever range (~ 39°C) while recording membrane potential. The slicing plane facilitated clear identification of whisker barrel columns, and neuronal activity was evoked by stimulating the barrel center using a bipolar electrode (0.2 ms pulses) at a stimulation of 1.4 Eθ. Eθ is defined as the minimal intensity that evoked a consistent excitatory postsynaptic current (EPSC) during more than 3 of 5 consecutive sweeps with 10 s ISI (Antoine et al., 2019). Eθ was determined in voltage clamp for each recorded cell prior to recording PSPs and spiking in current clamp. A new brain slice was used for each recording.  L4- evoked feedforward post synaptic potentials (PSPs) and spiking were recorded in single L2/3 PNs from a pre-stimulus baseline membrane potential (V_m) of -50 mV. This V_m, just below spike threshold, was selected to mimic in vivo conditions, as PNs in vivo can reside within this V_m range during whisker exploration of objects or surfaces (Yamashita et al., 2013). Evoked spikes were analyzed over a 150-ms interval, starting 2-3 ms post-stimulus, for 11 sweeps at 10 s inter-sweep interval (ISI). Spike threshold was defined as the membrane potential (V_m) at which the second derivative of V_m was >5 SDs above the pre-stimulus period. Intrinsic spiking excitability was measured in the presence of glutamate and GABA-A receptor blockers (in μM: 100 APV, 10 NBQX, 3 gabazine). F-I curves were obtained from PNs held at –80 mV by applying incremental current injections relative to this V_m.

Animal Surgery and In vivo Electrophysiological Recordings

In vivo experiments were done under general anesthesia where five mice (age, P24-26; body weight, 7-10 g; both genders) received an intraperitoneal injection of ketamine (100 mg/kg) and xylazine (10 mg/kg). Regular doses of the ketamine/xylazine cocktail were given intramuscularly to maintain the depth of anesthesia during the experimental sessions. Up until the thermal fever protocol started, the T_b of the animals was kept at 36°C using a homeothermic heating pad connected to a temperature controller (Supertech, Pécs, Hungary). To measure the internal body temperature, a Type T thermocouple microprobe (MT-29/5; Physitemp Instruments, Clifton, NJ, USA) was placed in the rectum of the animal. The microprobe has a shaft diameter of 330 µm, a time constant of 0.025 second and 0.1°C accuracy.

To perform high-density extracellular electrophysiological recordings, mice were placed in a stereotaxic frame (David Kopf Instruments, Tujunga, CA, USA), then two circular craniotomies (~1.5 mm in diameter) were made with a dental drill above the left and right barrel cortices. The craniotomies were centered at the following stereotaxic coordinates: anterior-posterior (AP): −1.0 mm; medial-lateral (ML): 3.5 mm (with respect to the bregma; Paxinos and Franklin 2001). Two commercially available Neuropixels 1.0 silicon probes (imec, Leuven, Belgium; Jun et al., 2017) mounted on two motorized stereotaxic micromanipulators (Neurostar, Tubingen, Germany) were implanted into the brain, one into the left and the other into the right barrel cortex, to a depth of 1.5 mm. An insertion speed of 2 µm/s was used to decrease the mechanical trauma caused by the probe insertion (Fiáth et al. 2019).

In order for the probe tracks to be perpendicular to cortical layers, the probes were inserted at an angle of 20 degrees from vertical. The dura mater in the craniotomy was left intact, except when it was too thick and thus the silicon probe could not pierce through this layer (which was indicated by significant probe buckling; n = 3 insertions). In these cases a 36 gauge, slightly bent needle was used to carefully cut the dura above the targeted cortical area. After the probe reached its final insertion depth, to allow the brain tissue to settle, we waited at least 10 minutes before electrophysiological recording was started. Spiking activity of cortical neurons (action potential band, 300–10.000 Hz) was recorded on 384 channels (768 channels in total for the two probes), with a sampling rate of 30 kHz/channel and with a gain of 500 (yielding a resolution of 2.34 μV per bit). Data were acquired using the SpikeGLX open-source software (http://billkarsh.github.io/SpikeGLX/). A common stainless steel wire inserted into the neck muscle of the animal served as the external reference and ground electrode. To avoid the dehydration of the cortex, the skull and the craniotomy was kept moist during the whole experiment using body temperature, sterile physiological saline solution and Gelaspon. Before starting the experimental protocol, manual whisker stimulation (by repetitively touching the whiskers of the animal with a cotton swab) was used to verify the recording position. In all cases (n = 10 probe insertions), we detected strong whisker-evoked neuronal activity.

Thermal Fever Protocol during Electrophysiological Recordings

First, we recorded cortical activity for 45 minutes at _{36°C body temperature ("baseline " period), then the body temperature of the animal was elevated with the aid of the heating pad and a power bank with warming capability which was placed next to the mouse. The body temperature was increased from 36°C to 39°C in about 5 minutes (0.01°C/s). The elevated body temperature was maintained for 45 minutes ("thermal fever " period). Next, we turned off the heating until the Tb} reached physiological temperatures (36°C; ~5 min; ~0.01°C/s). Cortical activity was recorded for another 45 minutes at 36°C body temperature ("recovery " period). Continuous recordings with a total duration of 145 minutes were obtained for each mouse.

Spike Sorting and Data Analysis

To extract cortical single-unit activity, spike sorting was performed with Kilosort2 (https://github.com/MouseLand/Kilosort; Pachitariu et al., 2016a, 2016b) using the default parameter set (available in the StandardConfig.m file). Channels containing activity acquired by recording sites located outside the cortex were removed before spike sorting (usually ~130 channels/probe recorded from the barrel cortex). The list of single unit clusters generated by Kilosort2 was visually inspected to remove units considered as noise (e.g., units with abnormal spike waveform shapes) or multi-unit activity (e.g., clusters with a contaminated refractory period). Manual curation of the Kilosort2 results was done with the Phy Python library, which provides a graphical user interface for interactive visualization of high-density data and supply operations for merging, splitting and marking of clusters (https://github.com/cortex-lab/phy; Rossant et al., 2016). In this dataset, we aimed to keep only those single units which had at least 900 spikes (>~0.1 Hz firing rate), a clear refractory period, a consistent waveform shape and whose spikes were present throughout the 145-minute-long recording.

Following manual curation, based on their spike waveform duration, the selected single units (n = 633) were separated into putative inhibitory interneurons and excitatory principal cells (Barthó et al., 2004). The spike duration was calculated as the time difference between the trough and the subsequent waveform peak of the mean filtered (300 – 6000 Hz bandpassed) spike waveform. Durations of extracellularly recorded spikes showed a bimodal distribution (Hartigan’s dip test; p < 0.001) characteristic of the neocortex with shorter durations corresponding to putative interneurons (narrow spikes) and longer durations to putative principal cells (wide spikes). Next, k-means clustering was used to separate the single units into these two groups, which resulted in 140 interneurons (spike duration < 0.6 ms) and 493 principal cells (spike duration > 0.6 ms), corresponding to a typical 22% - 78% (interneuron – principal) cell ratio. Finally, the firing rates of neurons were computed separately during the three distinguished periods (baseline, thermal fever and recovery). To decrease the effect of transient changes (e.g., tissue recovery during the baseline period or the short-term effect of heating during the thermal fever period), for each 45-minute-long period, we used only the last 25-minutes to calculate the mean firing rates of neurons.

Immunofluorescence staining

Mice were anesthetized and transcardially perfused using standard methods. Brains were post-fixed overnight in 4% PFA at 4 °C with shaking, rinsed in 1X PBS, and cryoprotected in 30% sucrose. They were embedded in optimal cutting temperature (OCT) compound, frozen on dry ice, and stored at –80 °C until sectioning. Tissue was cut at 30 µm on a cryostat, mounted onto slides. Sections were permeabilized in 0.3% Triton X-100 (in 1X PBS) for 10 min at room temperature. Endogenous peroxidase activity was quenched by incubation in 3% H₂O₂ (in 1X PBS) at room temperature. Slides were blocked in 10% normal goat serum for 1 h at room temperature, then incubated overnight at 4 °C with anti-TRPV3-Biotin antibody (#ACC-033-B), TRPV3 blocking peptide (#BLP-CC033), and/or anti-TRPV4 antibody(#ACC-034) (Alomone Labs, Israel). For antibody–peptide controls, the anti-TRPV3-Biotin antibody was pre-incubated with its blocking peptide at a 1:2.5 ratio, using 1 mg/ml of each. The following day, sections were incubated for 1 h at room temperature with fluorescently labeled secondary antibodies (streptavidin-conjugated or unconjugated). Nuclei were counterstained with DAPI Fluoromount-G (Electron Microscopy Sciences, USA). Finally, sections were mounted onto slides, cover-slipped, and stored at 4 °C until imaging. Images were acquired using a Keyence BZ-X810 microscope (BZ-X800 Viewer, Version 1.1.1.8) with BZ-X filter sets: DAPI (Ex 360/40 nm; OP-87762), GFP (Ex 470/40 nm; OP-87763), and TRITC (Ex 545/25 nm; OP-87764).