The data in the present study were collected from 9 adult male Long-Evans rats (307–574 g). The rats were procured from Charles River Laboratory (Wilmington, MA) and were acclimated in their home cages for a period of 7–10 days before undergoing the stereotaxic surgical procedure. The rats were housed in Plexiglas cages and had access to standard lab chow and water ad lib., under a 12-h light and 12-h dark cycle, with the temperature set to 22–24 degrees Celsius. All experiments conducted in this study received approval from the Pennsylvania State University Institutional Animal Care and Use Committee.

After the initial acclimation, aseptic stereotaxic surgery was performed to introduce the viruses into the brain. Rats were initially anesthetized with isoflurane, followed by intramuscular injections of ketamine (Midwest Veterinary Supply, Lakeville, MN; 40 mg/kg) and xylazine (12 mg/kg). Subcutaneous (s.c.) injections of dexamethasone (0.5 mg/kg) and enrofloxacin (2.5 mg/kg) were given to prevent tissue inflammation and bacterial infections. Buprenorphine (1.0 mg/kg) was injected subcutaneously to induce post-surgery analgesia. After exposing the skull, small holes were drilled towards the area of interest and the following injections were administered into the ACC:

Viral expression was under the control of the hSyn promoter, effectively limiting the DREADD receptor expression solely to neurons. This method is also referred to as ‘pan-neuronal’ targeting, since all neuronal subtypes (e.g., inhibitory interneurons and pyramidal cells) are targeted with the hSyn promoter (Wang et al., 2023). The viral volumes and titers and the stereotaxic coordinates for viral injections were as follows: 3 × 1 μL virus or control solution was injected towards ACC: +1.5 mm A/P, +0.5 mm M/L and − 3.8, −2.8 and − 1.8 mm D/V (Paxinos and Watson, 2014). Viral solutions were injected undiluted at ≥2 × 10¹² vg/mL [hM3D(Gq)] and ≥ 7 × 10¹² vg/mL [hM4D(Gi)] viral titers, using an automated Stoelting™ Quintessential Stereotaxic Injector (Stoelting Co., Wood Dale, IL) with the utilization of a 5 μL capacity Hamilton syringe (Neuros Syringe, Model 75 RN, 33 gauge, Point Style 4, Hamilton Co. Reno, NV). Viral vectors of pAAV-pAAV-hSyn-hM3D(Gq)-mCherry and pAAV-hSyn-hM4D(Gi)-mCherry were purchased from Addgene (plasmids #50474 and #50475, respectively; Addgene Co., Watertown, MA). The craniotomy was filled with a surgical silicone adhesive (Kwik-Cast, World Precision Instruments, Sarasota, FL) and the skin was closed above the wound. Rats remained in homecages for at least 7 weeks to recover and to allow for proper viral transduction prior to the subsequent electrophysiological recording session.

To induce an initial anesthesia, 3–4% isoflurane was applied followed by a single low dose injection of xylazine (5 mg/kg, i.p., AnaSed Injection, Akorn, Lake Forest, IL). During this deeper anesthetic state, the animal’s head was fixed into the stereotaxic frame, the previous head incision was re-opened, and the surgical silicone adhesive was removed from the craniotomy that was drilled during the viral injection. To record neuronal action potentials and local LFPs, a 16-channel silicon electrode array probe (type: A1x16-5 mm-150-177, NeuroNexus Co., Ann Arbor, MI) was slowly dropped into the ACC (AP: +1.5, ML: +0.5, DV: −3.8). We are confident that the electrode placement overlapped considerably with the virus expression, as the craniotomy from the viral injection surgery was still present and aligned with the stereotaxic coordinates at the time of electrophysiological recordings. The 16 recording channels were located 150 μm distance from each other on the probe’s surface, covering a 2.25 μm-long dorsoventral recording area (DV: approx. -3,8–1.55 mm), fully spanning the previous 3 viral injection sites (Supplementary Figure S1). The reference wire of the probe was connected to the probe ground and together they were inserted carefully into a small craniotomy drilled at the contralateral hemisphere. The reference wires were gently touching the brain surface/dura, without causing any damage to the brain or bleeding, and the instrumental ground wire was connected to the stereotaxic frame. After the initial surgical preparations had been completed, a much lower, sedation-level dose of isoflurane (0.3–0.5%) was set and a light anesthesia was maintained during the subsequent ~1 h recording session.

After recording a stable 20+ min baseline spontaneous firing activity from the ACC, we administered a s.c. injection of 0.3 mg/kg DCZ. We used a water soluble DCZ dihydrochloride salt (Hello Bio Co., #HB9126), which was easily and perfectly dissolved in sterile saline. We used a freshly made DCZ solution for each stimulation.

Each recorded electrode channel was separately analyzed (single neuronal firing activity and LFP analyses) from every animal, due to our observation that the 150 μm distance from each neighboring recording channels were sufficiently large to record from completely separate single neuronal spiking networks (for a demonstration recording sample please see Supplementary Figure S2).

The electrophysiological signal was sampled at 20 kHz and amplified using a NeuroNexus recording system, with a SmartBox data acquisition device (NeuroNexus, Ann Arbor, MI). The acquired 16-channel raw electrophysiology data was converted to Spike2 software format (*.smr files) and analyzed with the help of the Spike2 program (Spike2, version 7.00, Cambridge Electronic Design Co., Cambridge, UK). A notch filter was applied at 60 Hz for each channel. The following 7 Infinite Impulse Response (IIR) second order band-pass Butterworth digital filtering passing limits (Zhang et al., 2020) were applied to each of the 16 recorded channels to separate 6 LFP band-specific channels and 1 Spiking channel from each raw data channel, for further analysis:

Single neuronal spikes were separated from the continuous digitalized spiking channel, using a ± 25 μV spike amplitude discriminator and further classified using the built-in spike separator function in the Spike2 program, based on action potential amplitude and waveform shape differences. The separated single neuronal data channels were then individually analyzed. First, 1 s-based frequency histograms were built and then changes in firing rates (in Hz, 1/s) after DCZ activation were evaluated. We used 3-min single neuronal firing rate average values as the basis of the statistical analyses, except for determining the type of neuronal response (inhibition-, excitation- etc.) where we used the 1 s-based firing rates. To determine the type of neuronal responses to stimulus, we applied two-sample unequal variance t-test to each post-stimulus 3-min window and compared them to the 1–4 min pre-stimulus baseline firing rates. The modality of single neuronal effects was determined based on significant firing rate changes compared to the pre-stimulus baseline activity. The following neuronal response modalities were specified: excitation (E), inhibition (I), excitation-inhibition (EI), excitation-inhibition-excitation (EIE), inhibition-excitation (IE), inhibition-excitation-inhibition (IEI), excluded or no effect (N). The main exclusion criterion was defined as a significant difference between the pre-stimulus 1–4 vs. 4–7 min baseline activity (unstable baseline, which served as a recording quality check), but the 1–4 vs. 7–10, and the 1–4 vs. 10–13 min differences were also taken into consideration in more complicated cases.

Single neuronal effects were analyzed statistically in a collided (all neurons added together) and a separated (neurons with similar effects added together) manner, applying one-way ANOVA, Tukey HSD post-hoc tests and dependent samples t-tests (Statistica version 6.1, StatSoft Inc. Tulsa, OK).

LFP band-specific power data (in μVolt²) was extracted from the separated band-specific channels from Spike2 (see details above), using pre- and post-stimulus 12-min windows (Pre 1–13, Post 1–13, 13–25, 25–37 min) and was further analyzed using Microsoft 365 Excel (Microsoft Co., Redmond WA) and Statistica programs, using mainly factorial ANOVA and LSD post-hoc tests. Additionally, a normalized LFP band specific power data was calculated from dividing the post-stimulus band specific power data by the pre-stimulus power. ANOVA and LSD post-hoc tests were also carried out on the normalized data set.

Spiking-correlated power data (in μVolt²) was similarly extracted from the separated band-specific channels with Spike2, using the same method as described above (LFP band-specific power data extraction), with the difference that only the ±200 ms peri-spike intervals were extracted, not the whole recording session. This approach led to the creation of a distinct dataset exclusively focused on LFP activity changes synchronized with the neural spiking activity. Data analysis methods were the same as in the case of the LFP band-specific power data (see above).

Following electrophysiological recordings, animals received an overdose of 2:1 ketamine/xylazine. They were then transcardially perfused with 60–120 mL saline [0.9% NaCl (Sigma-Aldrich) in ddH2O] and ~ 240 mL 4% paraformaldehyde [Electron Microscopy Sciences; in 0.1 m phosphate buffer (PB)]. Brains were removed and postfixed for 1 h in the same fixative at room temperature prior to being stored in a sucrose solution [Thermo Fisher Scientific; 20% in 0.1 m PB containing 0.01% sodium azide (Sigma-Aldrich)] at 4°C for at least 24 h.

Brains were then sectioned coronally (35 μm) on a freezing stage microtome (Leica Biosystems SM2010 R) into four parallel series. Three of the series were stored in cryoprotectant solution [30% sucrose in 0.1 m PB containing 30% ethylene glycol (Thermo Fisher Scientific) and 0.01% sodium azide] at 4°C for future processing, and one series was immediately mounted onto Superfrost plus glass slides (Fisher Laboratories) and allowed to dry prior to Fluoroshield with DAPI (Sigma-Aldrich) application and coverslipping.

In a second series containing the ACC, free-floating sections were incubated and washed at room temperature under gentle agitation, in 0.1 M PBS, for at least 4 h prior to additional processing and between all incubations. Sections were then exposed to 1% H₂O₂ (10 min; in PBS) and incubated in blocking solution [2.5 h; PBS containing 0.1% bovine serum albumin (Thermo Fisher Scientific) and 0.4% Triton X-100 (Sigma-Aldrich)]. Sections were then incubated in mouse anti-GAD67 17 h; (EMD Millipore; 1:2,000) and goat anti-mouse conjugated to Dyelight 488 (2.5 h; Invitrogen; 1:100), followed by rabbit anti-RFP (17 h; Rockland; 1:20,000) and goat anti-rabbit conjugated to Alexa 555 (30 min; Invitrogen; 1:100). Sections were mounted on Superfrost plus glass slides (Fisher Laboratories) and coverslipped with Fluoroshield with DAPI (Sigma-Aldrich).

All sections including the ACC were visually examined for mCherry fluorescence or mCherry and GAD67 using an inverted fluorescence phase-contrast microscope (Keyence BZ-9000 E). In the first series, locations of viral injections were mapped and images were taken using a Texas-R/mCherry filter for viral expression visualization and a DAPI filter for cell nuclei visualization using the BZ-II Viewer software at 4x, 10x and 20x magnification. In the second series, sections including the ACC were imaged using a Texas-R/mCherry filter for visualization of viral expression, a DAPI filter for cell nuclei visualization, and a GFP filter for GAD67 visualization using the BZ-II Viewer software at 20x magnification. BZ-II Analyzer software was used to adjust image brightness, contrast and fluorescent blur, overlay channels and add scales. Three sections from each animal in hM3D(Gq) and hM4D(Gi) groups were chosen, and a 200 μm x 200 μm area within the ACC was used to manually quantify number of DAPI+ cells, number of mCherry+ cells, number of GAD67+ cells, number of mCherry+/GAD67+ cells, percentage of GAD67+ cells that co-express mCherry and percentage of mCherry+ cells that co-express GAD67.

In a separate analysis, we modified the approach used for oscillatory LFP power analysis, as previously discussed and illustrated in Figure 3. However, in this case, we narrowed our focus to the ±200 ms peri-injection LFP signal in close proximity to individual neuronal spikes (action potentials). The purpose of this tailored analysis is to specifically examine changes in oscillatory LFP power that are synchronized with the spontaneous firing activity of local neurons. This method provides an emphasized view on the network oscillations that are locally affected by chemogenetic manipulation.

To demonstrate that low-dose isoflurane sedation did not significantly confound our results, we conducted pilot experiments by replacing isoflurane with ketamine as the primary anesthetic. Ketamine operates through a distinct mechanism of action, interacting with N-methyl-D-aspartate (NMDA), opioid, monoaminergic, and muscarinic receptors, in contrast to isoflurane, which primarily engages with GABA receptors (Chung et al., 2013). Under ketamine anesthesia, we observed the typical increase in firing rates, accompanied by the characteristic enhancement of oscillatory LFP activity in response to hM4D(Gi) stimulation (Supplementary Figure S3). This pattern closely mirrored what was observed under isoflurane sedation.

Viral injections into the ACC region were confirmed through the visualization and mapping of viral vector-tagged mCherry fluorescence. Representative images from an hM3D(Gq)- (Figures 5A,C) and an hM4D(Gi)- (Figures 5B,D) transduced animal are shown in Figure 5, demonstrating expression of cell nuclei (DAPI), viral transduction (mCherry), and GABAergic cells (GAD67) in the ACC. Quantification of a subregion within the transduced area of the ACC (indicated by white box in Figures 5A,B and shown in Figures 5C,D) revealed that 17.98 ± 12.09% of all GAD67+ cells were transduced with hM3D(Gq) and 33.52 ± 10.82% of all GAD67+ cells were transduced with hM4D(Gi) (Figures 5E,F). Additionally, 67.04 ± 4.42% hM3D(Gq)-transduced cells co-expressed GAD67 and 67.15 ± 16.02% hM4D(Gi)-transduced cells co-expressed GAD67 (Figures 5E,F). Overall, these results indicate that the majority (~67%) of transduced cells within the ACC are inhibitory, GABAergic cells.

The use of anesthetics in our study has the potential to introduce unknown and uncontrolled confounding effects. However, the fact that we replicated our results using a different anesthetic with a different neuronal mechanism of action (Supplementary Figure S3) somewhat rules out the pivotal role of the anesthetic mechanism significantly affecting our observations. Nevertheless, follow-up experiments on awake and freely behaving animals would be valuable to confirm our current observations.

Our study observed decreased baseline firing activity in viral-infected animals compared to controls without chemogenetic receptors. While virus titers were within safe limits (10¹³ vg/mL) (Wirtshafter and Stratford, 2016; Goossens et al., 2021), we cannot rule out potential neurotoxic effects. Limited data exist on the long-term consequences of chemogenetic applications, emphasizing the need for further research to optimize safety and efficacy.

Due to the technical nature of our recording setup, we were generally not able to record the firing activity of fast-spiking (>50 Hz) PV interneurons (Nahar et al., 2021). This lack of PV interneuronal data in our sample could be related to the employed extracellular multichannel recording probes (A1x16-5 mm-150-177, NeuroNexus Co., Ann Arbor, MI), which typically have somewhat low impedance (approx. 0.5–3 MOhm). This impedance range is suggested to be strongly biased toward recording form pyramidal cells (Moore and Wehr, 2014).