Animals
All investigations within this project have been approved by the institutional ethics commissions of TU Dresden and the Central Institute of Mental Health, Mannheim and the regional authorities of the federated states of Saxony (Landesdirektion Sachsen) and Baden-Württemberg (Regierungspräsidium Karlsruhe). Experiments were performed in accordance with the guidelines of the Directive 2010/63/EU on the protection of animals used for scientific purposes of the European Commission with great attention to avoid suffering and to reduce number of animals used [44].
We used male Wistar rats from the breeding colony at the CIMH. Rats were housed in single cages (Makrolon®, Type III, Tecniplast Deutschland GmbH) on sawdust bedding (Ssniff - Bedding 3/4 S, Altrogge) with Bed-r'Nest material (Datesand Ltd.). Pelleted food (V1534-300, ssniff Spezialdiäten GmbH) and water were available ad libitum. Housing rooms were temperature (20–22°C) and humidity (40–55%) controlled with a 12 h automatic light-dark cycle (lights on at 6.00 am).
Long-term alcohol consumption with repeated deprivation periods
Following two weeks of habituation to the animal room, rats were given ad libitum access to ethanol (VWR International GmbH) solutions of 5%, 10%, and 20% (v/v) besides tap water. Concurrent access to several alcohol concentrations has been shown to increase the magnitude and duration of the ADE [31, 32]. The positions of bottles were changed weekly. After eight weeks of continuous alcohol availability, bottles with alcohol solution were removed from the cages and reintroduced after a deprivation period of two weeks. Phases of free access to alcohol and deprivation subsequently alternated randomly with variable durations of drinking between 4–6 weeks and deprivation lasting 2–3 weeks, aiming to prevent habituation and behavioural adjustment.
The long-term alcohol exposure procedure, including all drinking and deprivation phases, continued over 13 months. During the final two-week period of alcohol deprivation, animals (n = 10) were habituated to the recording set-up and underwent surgery to implant the neuroprosthetic device (Fig. 1a). Ten alcohol-naïve control animals underwent the same procedure [30].
Manufacturing and implantation of neuroprosthetic interfaces
The procedures to fabricate, characterise and implant the used neuroprosthetics have been described in detail before [30, 45, 46]. Briefly, micro-EcoG interfaces were produced by an additive manufacturing approach using the 3D bioprinter 3DDiscovery™ Evolution (regenHU Ltd.). The devices consisted of soft silicone for the base (DOWSIL™ SE 1700, Dow Inc., Midland, USA) and isolating (SE 734, Dow Inc., Midland, USA) layers embedding a 3 × 3 electrode array of conductive platinum ink (chemPUR). The electrode interconnects were attached to stainless steel microwires (∅: 0.23 mm, 7SS-2T, Science Products GmbH) soldiered to a plug-in connector (BKL 10120653, BKL-Electronic Kreimendahl GmbH). An additional microwire was fixed to a microscrew drilled into the skull during surgery and served as a reference electrode. Implantation was performed under subcutaneous (s.c.) anaesthesia (fentanyl (0.005 mg/kg, Hameln Pharma), midazolam (2 mg/kg, Ratiopharm), medetomidine hydrochloride (0.135 mg/kg, Orion Pharma) in a stereotactic surgery involving trepanation of the skull (∅ 6.0 mm, 330205486001060, Meisinger) to position the devices epidurally on the prefrontal cortex with the frontal electrode row located at 3.2 mm anterior to bregma. External parts of the implant were fixed to the skull using dental cement (Paladur, Kulzer GmbH), and the wound was sutured. After completion of the surgery, anaesthesia was antagonised (s.c., naloxone hydrochloride (0.12 mg/kg, Inresa Arzneimittel GmbH), flumazenil (0.2 mg/kg), atipamezolhydrochloride (0.75 mg/kg, Orion Pharma). Animals received meloxicam (1 mg/kg, s.c., Boehringer Ingelheim Vetmedica GmbH) as an analgesic right after surgery and the following day.
Electrocorticography recording and acute pharmacological modulation of neural activity
ECoG recordings without any further interventions were implemented three days after the animals had undergone surgery. Recordings were performed at a sampling rate of 3 kHz using the Intan RHD2000 USB interface system cable connected to the implant plug-in module. The initial recording measured a state of abstinence in alcohol dependent rats compared to matched alcohol-naïve rats.
Recordings were performed within an electrically shielded and sound-insulated audiometry booth where one animal at a time was placed in a rodent sling (Lomir Biomedical Inc.) to reduce movement artefacts. To record auditory event-related activity, sound stimuli were presented through a stereo loudspeaker located at a distance of 40 cm and an angle of 45° centrally above the animal's head. Auditory stimuli have been generated using the Psychophysics Toolbox (Version 3) for Matlab (Version R2019b, The Mathworks Inc.) and were composed of frequent (standards: 50 ms, 1 kHz, 70 dB sound pressure level (SPL), 87% of trials) and rare (deviants: 50 ms, 2 kHz, 80 dB SPL, 13% of trials) sinusoidal sounds with 5-ms onset/offset ramps, presented in 6 blocks of 5 min with 1 s interstimulus interval. Deviant sounds have been interspersed with at least one standard tone to avoid successive occurrences.
On day 6 and 9 after initial recordings, alcohol-dependent rats randomly received intraperitoneal (i.p.) injections of psilocybin (2.5 mg/kg, 3-[2-(dimethylamino) ethyl]-1H-indol-4-yl] dihydrogen phosphate (purity 99.7%) obtained from the University of Chemistry and Technology Prague, Czech Republic, dissolved in Ampuwa (Braun Melsungen AG) or LY379268 (1 mg/kg, 1R,4R,5S,6R)-4-Amino-2-oxabicyclo[3.1.0]hexane-4,6-dicarboxylic acid, Tocris Bioscience) each 30 min before recording. Respective dosing was chosen based on previously published data that demonstrated efficacy in reducing alcohol relapse [37, 41].
To capture drug-induced changes in resting-state neural activity, we performed continuous 5-minute recordings right before administering psilocybin or LY379268, and directly after the auditory oddball paradigm described above. Pre- and post-application recordings were approximately one hour apart.
Event-related potentials and oscillatory activity
Data processing was performed using the EEGLAB toolbox [47] (Version 2019.1) for Matlab. Following offline filtering using a 0.1–45 Hz bandpass finite impulse response filter (Kaiser windowed, Kaiser β = 5.65, filter length 54330 points), data were segmented in epochs between − 100 and 700 ms relative to stimulus onset separately for standard and deviant sounds and baseline-corrected using the pre-stimulus interval between − 100 ms to 0 ms. Artefacts and noisy channels were identified and excluded based on a delta criterion of 500 µV and visual confirmation before averaging epochs for single subjects and over all animals (grand average). As neural responses to the frequent standard sounds rarely displayed pronounced amplitudes, indicating habituation effects due to the high repetition rate [48], subsequent analysis was performed on the difference curves (deviant-minus-standard responses). ERP peak latencies were detected within the following time intervals confirmed by visual inspection: P1: 20–70 ms, N1: 35–120 ms, P2: 60–260 ms, N2: 100–320 ms, P3: 130–600 ms. The amplitudes of the ERP components were calculated as peak-to-peak amplitudes (P1N1, N1P2, P2N2, N2P3).
Oscillatory power in the delta (1–4 Hz), theta (4–8 Hz), alpha (8–12 Hz), beta (12–30 Hz) and gamma (30–45 Hz) bands was determined by applying the function pop_newtimef.m in EEGLAB based on a Fast Fourier transform using 400 datapoints and a pad ratio of 64. The resulting event-related spectral perturbation (ERSP) was calculated as decibels (dB) ( ≙ 10*log10 (µV2/Hz)).
Resting-state neural oscillations
Data were initially bandpass filtered, as described above, and segmented into 2 s-epochs with 50% overlap. Following the exclusion of bad epochs (δ = 500 µV, confirmed by visual inspection), we applied the EEGLAB function pop_spectopo.m to determine channel spectra based on the Welch method. The resulting power spectral density (PSD) in the 1–45 Hz frequency range was calculated as dB.
Statistics
Statistical analyses were conducted using SPSS® (Version 28, IBM Corp.) and R (Version 4.2.3, R Foundation for Statistical Computing). Differences in ERPs (peak latencies, amplitudes) and EROs (bandpower (ERSP), latency and frequency of max. power within each frequency band and over the whole frequency range) induced by psilocybin and LY379268 were examined by applying within-subjects repeated measures analysis of variance (rmANOVA) with factors treatment and channel. Multiple comparisons were adjusted using the Sidak correction. Likewise, rmANOVA was applied to analyse resting-state oscillatory activity (PSD) before vs. after administration of psilocybin and LY379268.
Neural parameters of alcohol- and drug-naïve controls were compared with those of long-term alcohol consumers and each pharmacological intervention by applying a between-subjects ANOVA with factors treatment and channel and Sidak adjustment for multiple comparisons.
To explore the potential impact of individual alcohol consumption patterns and susceptibility to the ADE on designated neural parameters, we performed Spearman correlation analyses using the averaged amounts of alcohol (g/kg body weight) consumed by each rat per day during the last week of a drinking phase (= baseline consumption, BL) and on the first day following periods of abstinence (= Alcohol Deprivation Effect (ADE)) as well as relapse intensities (difference of ADE and previous BL). Partial Spearman correlation was performed to relate alcohol consumption patterns with the neural activity following pharmacotreatments, controlling for neural activity without drug administration.