Organotypic culture of post-mortem adult human brain explants exhibits synaptic plasticity

Background: Synaptic plasticity is an essential process encoding fine-tuned brain functions, but models to study this process in adult human systems are lacking. Objective: We aim to test whether ex vivo organotypic culture of post-mortem adult brain explants (OPABs) retain synaptic plasticity. Methods: OPABs were seeded on 3D microelectrode arrays to measure local field potential (LFP). Paired stimulation of distant electrodes was performed over three days to investigate our capacity to modulate specific neuronal connections. Results: Long-term potentiation (LTP) or depression (LTD) did not occur within a single day. In contrast, after two and three days of training, OPABs showed a significant modulation of the paired electrodes ’ response compared to the non-paired electrodes from the same array. This response was alleviated upon treatment with dopamine. Conclusion: Our work highlights that adult human brain explants retain synaptic plasticity, offering novel approaches to neural circuitry in animal-free models.


Introduction
The human brain is a highly organized tissue composed of an interwoven network of neural cells, in which information is transmitted through intercellular synaptic structures.The neuronal network conveys electric activity over long distances, and the strength of the pre/postsynaptic interactions is a critical parameter encoding neurological functions.Both murine and more recently cerebral organoid models have been instrumental in uncovering seminal paradigms, which however, could not be always transferable to the human brain, highlighting the need for complementary approaches [1].
Hebbian plasticity postulates that paired stimulations of the pre-and post-synaptic compartments modulates the strength of synaptic connections, best exemplified in protocols of Spike-Timing Dependent Plasticity (STDP) [2,3].Such approaches induce either timing-dependent-long-term-potentiation (LTP) or -depression (LTD), according to the specific stimulations exerted.Numerous STDP-based studies have been performed in mouse models, but do not address the extensive differences between human and mouse brains [4].Accordingly, STDP protocols induce very different outcomes depending on whether the sample is of human or mouse origin [1].Human induced pluripotent stem cells derived neurons represent a versatile model for assessing human neural plasticity and LTP in vitro [5], but such approaches neither recapitulate the diverse architecture of the human brain nor its complete cellular composition.STDP has been shown in human brain resections in patch-clamp settings [6], but induction of LTP or LTD in local field potential (LFP) and multi-electrode arrays (MEAs) remains challenging.
We recently characterized the organotypic culture of human postmortem adult brain explants (OPAB), and recorded spontaneous electrical activity ex vivo using 3D microelectrode arrays (MEA) [7].We showed that spontaneous LFP generated by OPABs can be exploited to characterize external perturbations, such as viral infections, using dedicated machine learning frameworks.However, intrinsic heterogeneity of the brain explants and absence of synchronicity lead to relatively high variability of the spontaneous LFP signal.
Hence, we investigated whether the ex vivo culture of human brain explants can exhibit Hebbian plasticity by applying a STDP protocol, and specifically paired stimulation electrodes.Although synaptic plasticity did not change significantly a couple of hours post-stimulation, we found that OPABs exhibited increased response the following days.Interestingly, treatment of the OPAB with Dopamine prevented such inter-day plasticity, which illustrated the fine-tuned regulation of this process.

Ethics
The research presented in the study complies with all relevant ethical regulations.The protocol for the use of post mortem brain explants for research purposes included in this study was approved by the institutional review board (IRB) of the CHU of Montpellier (approval ID: 202,000,643) and the French Biomedicine Agency (agreement number PSF20-025).It permits human brain resection for research purposes during autopsy performed at the forensics institute of the CHU de Montpellier, upon obtaining local prosecutor authorization and family consent.

Donor enrolment
Excluding criteria were causes of death involving the head and neck (e.g. car crashes, stroke), and previously known severe neuropathology (e.g.epilepsy, neurodegenerative diseases).Altering substances (e.g.alcohol, opiates) were not included in selection criteria but are reported for each donor (Supplementary Table S1).Post-mortem interval for all selected donors was between 1 and 22 h.The data presented in this study originate from 9 donors aged 27-76 years old.

Brain slice preparation and culture
During autopsy, after incising and reclining the scalp, the skull was opened with an electric saw and the whole brain was extracted.Macroscopic examination attested to the absence of brain lesions, and a 2-3 cm thick coronal slice of the primary motor cortex (M1, frontal lobe) was dissected, and immediately transferred for transportation into cold medium containing 1X Neurobasal media (Thermofisher), supplemented with N2 (Gibco), 1X Glutamax (Thermofisher), and 0.5 % Penicillin/Streptomycin (Thermofisher), hereafter referred to as N2 media.Samples were kept at 4 • C during the transfer to the lab (<30 min) to be processed.Under sterile conditions, the dura mater was removed, individual folds were isolated and embedded into 3 % low melting-point agarose (Thermofisher).Slices of 300-μm thick were sectioned using a vibratome (PELCO Easyslicer, Ted Pella).Suitable slices were selected based on cytoarchitectural integrity of the cortical structure, i.e. presence of white matter and all layers of the grey matter.For culture, brain slices no younger than 5 days in vitro were transferred to an insert lifting a PET membrane with 0.4 μm pores (Sabeu), hanging over 2.5 ml of medium (changed every 2-3 days) in a 6-well plate.Slices were maintained at 37 • C, 5 % CO 2 and 95 % humidity atmosphere.

Electrophysiology
For sensitive LFP measurement and stimulation of the tissue, OPABs were plated on 3D MEA (MultiChannel System), containing an array of 60 TiN-coated conical electrodes of 12 μm diameter active site and 80 μm height, spaced by 200 μm each.The OPABs were transferred to MEAs the day prior to experiment using a brush and spatula, with the membrane side down, and the basal side of the cortex oriented towards the ground electrode.The OPABs were maintained in 150 μl BrainPhys aCSF (StemCell Technologies), supplemented with N2 and Glutamax.The medium was changed daily for up to 3 days post seeding, and individual MEAs were kept in a humidified Petri dish to limit evaporation.
For recordings, MEAs were dried of any moisture on the connective parts and quickly transferred under a sterile hood onto a MEA2100-Mini headstage (MCS) connected to a digital amplifier (MCS).The headstage was shielded using tin foil and placed in an incubator at 37 • C, 5 % CO 2 and 95 % humidity atmosphere for recording.LFP data was acquired at a 10 kHz sampling rate.
For selecting the electrodes for the experiment, first the most active electrodes were visually identified during a 2 min baseline recording prior to experiment.A 35 mV monophasic pulse was applied for 200 μs to one of the active electrode, and the response of the stimulation was assessed for electrodes separated by at least two electrodes.Once an electrode eliciting a response of more than 5 mV was isolated, it was next stimulated and we verified if this stimulation elicited a response (>5 mV amplitude) in the first stimulated electrode.This process was repeated until 2 suitable electrodes eliciting responses (>5 mV amplitude) on either direction was isolated.These were chosen for the paired stimulation experiment.

Paired stimulation protocol
The experiment extended over a period of three days.Correspondingly, we present the data labeled as being from 0 h (zero hour), 24 h (24 h) and 48 h (48 h), relative to the start of the recording.The MEA mounted OPABs were exposed to 9 stimulation blocks each day, comprising of 5 alternating 'test' blocks and 4 'training' blocks (starting with a test block).
Each training block consisted of 300 pairs of 1 Hz stimulations of the stimulus-electrode (orange electrode in Fig. 1D) followed by the stimulation of the paired-electrode (cyan electrode in Fig. 1D).We tested a few stimulation parameters (not shown) to converge on the currently used constant monophasic stimulation of 35 mV applied for 200μs, applied to both the stimulus-electrode and paired-electrode.In the training blocks, the two stimulations were applied with a time gap of 12 ms between them.In the test block, the same stimulation was applied only to the stimulus-electrode, 10 times, again at 1 Hz.We recorded the resulting response (defined later) in the paired-electrode from 10 ms prior to the stimulus-electrode stimulation until 90 ms after the stimulus-electrode stimulation.A period of 60 s separated each training and test block.Note that all the data in Figs. 1 and 2 are of the response of the pairedelectrode area in the test blocks (block represented by numbers) over the three days (hours represented as subscript).

Treatment
In the dopamine-treated slices (10 slices from 5 donors), 10 μM Dopamine (Sigma-Aldrich) was added 2 min before the first test block.The dopamine treatment was performed while the MEA was on the headstage, followed by gentle shaking.Slices were washed 3 times with aCSF 1 h after the end of the last test block of every experiment day, in order to avoid overnight exposure to dopamine.

Imaging
Following the three days experiment, OPABs were washed in phosphate buffered saline (PBS) and fixed in 4 % paraformaldehyde (PFA) for 1 h at room temperature before staining.Samples were permeabilized with permeabilization buffer (0.5 % BSA, 1 % Triton in PBS) for 24 h on a rocker and all subsequent steps were performed in permeabilization buffer with overnight incubations.OPABs were labeled with primary rabbit polyclonal anti-MAP2 (GeneTex, #GTX133109), and chicken anti-MBP (Thermo Fisher Scientific, #PA110008) antibodies, washed and labeled with appropriate secondary antibodies; donkey anti-rabbit IgG Alexa Fluor Plus 488 (Thermo Fisher Scientific, #A-21206), donkey anti-chicken Alexa Fluor 647 (Jackson ImmunoResearch, #703-605-155) and Dapi nuclear counterstain (Pierce) at a final concentration of 1 μg/ml.For viability assessment, OPAB healthiness was assessed using the LIVE/DEAD Viability/Cytotoxicity kit (Thermofisher Scientific) for 30 min, washed in PBS and fixed in 4 % paraformaldehyde (PFA) for 1 h at RT. Upon fixation, OPABs were fully immerged in RapiClear 1.52 reagent (Sunjin Lab) overnight at RT to clarify the samples before imaging.The images were taken on a spinning disk confocal microscope (Dragonfly, Oxford Instruments) equipped with an ultrasensitive 1024 x 1024 EMCCD camera (iXon Life 888, Andor) using the 405, 488 and 637 nm laser lines with a 40X NA 1.3 oil-immersion objective.Images were processed using Bitplane Imaris x64 (Oxford Instruments) version 9.7.

Data analysis
Data was exported as ASCII files and analyzed using a custom Mat-Lab/Octave (v8.4.0) script.The stimulated area (a 3x3 electrode group neighboring the stimulus-electrode), paired area (a 3x3 electrode group neighboring the paired-electrode) and ground electrode were excluded from background calculation.The response of the paired-electrodes was calculated by averaging the peak activation (after every stimulation of the stimulus-electrode) in the 9 electrodes of the paired area.The response in the nonpaired electrodes was calculated by averaging the peak activations across all the electrodes except those in the paired-area, the stimulated area and the ground electrode.
The response across the 5 test blocks (averaged over the 10 stimulations) is shown for each OPAB at 0 h (Figs.2C), 24 h (Fig. 2D) and 48 h (Fig. 2E) of the experiment.Fig. 2F shows the absolute difference in the response in the first test at 24 h from that of the first test of 0 h (T 24 0 -T 0 0) in cyan from each OPAB.This is compared with the same difference in response (T 24 0 -T 0 0) in the corresponding non-paired electrodes for each OPAB.
Fig. 2G shows a similar difference in first test responses between 48 h and 0 h (T 48 0 -T 0 0) in the paired-electrode (cyan data) and non-paired electrodes (grey data).Fig. 2H-I shows the same measures from the Dopamine treated slices.
Grubb's outlier test was used on each test and consistently removed two slices which had very high basal activation levels, one from the non- dopamine experiment, and one from the dopamine addition experiment.These two slices were excluded from the analysis and the data presented here altogether.
In addition, T 0 1 of slice 10 in the non-dopamine experiment presented a technical issue, and the slice was removed from 1-way paired ANOVA analysis in Fig. 2C, but kept in all subsequent statistical analysis.
Statistical significance of within day trends was analyzed using a 1way ANOVAs while the significance of response differences between the paired-electrode and nonpaired electrodes was assessed using Wilcoxon paired signed-rank test.Schemes were created with Biorender.com.

Results & discussion
We have recently characterized the use of OPAB for antiviral assessment [7], and here we further confirm that OPAB contains neural cells including neurons, myelin and astrocytes, while exhibiting synaptic boutons (Fig. 1A-C).Furthermore, the OPAB were found to produce spontaneous LFP, indicating that a subset of neural circuits has been preserved, despite the post-mortem collection of the tissue and the slicing procedure.Hence, we wondered whether the remaining synaptic connections showed plastic properties and endeavored to test the well-established protocol of coincidental stimulation, known to induce STDP.To this end, we performed training blocks of alternated training and test blocks (Fig. 1D), in order to induce plasticity between two distant areas of the slices.In the test block, the stimulation was applied only to the stimulus-electrode, while we recorded the response in the paired-electrode (Fig. 1D and Material & Methods).
The activation in the paired-electrode showed an increase, followed by a decrease, after which it slowly converged back to the baseline (Fig. 2A).We use the peak voltage of the increase (grey arrow) to quantify the response in each electrode.Heatmaps representing the response change over time as a function of electrode position across tests and days showed that the paired electrode tends to exhibit higher response amplitude compared to non-paired electrodes (Fig. 2B and Supplementary Figs.S1-S2).Spatiotemporal quantitative analysis revealed that single-day training over the 4 training blocks was unable to trigger a significant change in the response, neither in the pairedelectrode (Fig. 2C; F = 1.033, p = 0.354, 1-way repeated measures ANOVA), nor in the non-paired electrodes (F = 2.331, p = 0.1215).Some OPABs showed increased or decreased amplitude of their response over the tests (T 0 0 to T 0 4), but statistical analysis of the absolute difference between the 4th (T 0 4) and 0 th (T 0 0, pre-stimulation) tests of the paired-area versus the non-paired also showed no significant changes (p value = 0.24).
Next, to test whether the response could have longer kinetics, we repeated the train/test blocks 24 and 48 h after the initial test as schematized in Fig. 1D.Paired-electrode response across the test sessions after one day (Fig. 2D, T 24 0 to T 24 4) or after 2 days (Fig. 2E, T 48 0 to T 48 4) showed again no change in the responses within each day, both for 24 h (F = 1.217, p = 0.303, 1-way repeated measures ANOVA) as well as 48 h (F = 1.639, p = 0.2219, 1-way repeated measures ANOVA).As for T 0 , statistical analysis of the absolute difference between the 4th (T 24 4 or T 48 4) and 0 th (T 24 0 or T 48 0, pre-stimulation of the experimental day) tests of the paired-area versus the non-paired also showed no significant changes at both 24 h (p value = 0.29) and 48 h (p value = 0.33).
Quantification of the changes of response across days was assessed by analyzing the difference between the responses of paired-electrode in the first test block on each day (T 0 0, T 24 0 or T 48 0, Fig. 2F-G, cyan dots) and compared it to the change in response in the non-paired electrodes (Fig. 2F-I, grey dots).Although we did observe important inter-slice variability (each slice being color-coded), there was a clear change in the response amplitude across the days.Of the 14 slices we trained, 10 exhibited an increase in the response amplitude across the training blocks, while 4 exhibited a decrease in the response amplitude across the training blocks (Fig. 2F-G).Of note, the absolute response change is shown here.We observed a significant increase in response in the paired-electrodes compared to the non-paired electrodes, both after 24 h (Fig. 2F, p = 0.0353, Wilcoxon paired signed Rank test) and after 48 h (Fig. 2G, p = 0.0031, Wilcoxon paired signed Rank test).Analysis of all the slices revealed that this increase occurred in 71 % (24 h) and 79 % (48 h) of slices (Fig. 2F-G).These findings indicate that our paired stimulation protocol can induce synaptic plasticity in OPAB, that become apparent over a multiday timescale.
Interestingly however, dopamine treatment during training reverted the observed changes in response level (Fig. 2H-I, p = 0.0039, Wilcoxon paired signed Rank test), suggesting that in our model, dopamine prevents synaptic plasticity.This could be due to the preferential activation of inhibitory dopamine receptor classes (D2-4), which would hamper learning upon coincidental stimulation.Further research using specific dopamine receptor agonists/antagonists is needed to determine their role in OPAB synaptic plasticity.
In conclusion, our study highlights that brain explants obtained postmortem and cultured ex vivo, can show plasticity upon paired stimulation training.The design of this work did not aim at correlating how the timing of stimulation impacts the level of response induction (which could then be referred to as STDP), but rather to illustrate that a classical STDP-inspired experimental design of coincidental stimulation readily induces synaptic plasticity in the complex OPAB model.Seminal studies in human samples have previously compared various time-lags using patch clamp [1,8,9].While the rules governing STDP in human cortex remain incompletely understood and widely variable, our approach is likely to provide additional complexity to this issue, and future work shall specifically investigate how the time lag affect synaptic plasticity using MEA-based LFP.
Responses to the test/train protocol was variable, with some slices poorly responding to the stimulation.This could be due to the high interdonor and inter-slice variability, as previously shown [7].Moreover, the intrinsic neural circuitry of each slice is relatively unique and it is likely that some of our test/train electrodes were not adequately connected to induce a STDP response.We chose to perform the more integrative approach of MEA-based LFP recording and stimulation, as this approach can represent a surrogate to current in vivo development of brain-machine interfaces (BMI) implanted in the brain of mice, monkeys and even humans (Neuralink).
Finally, we found that treatment with low concentration of dopamine prevents synaptic plasticity induction in OPAB.Although unexpected, our ability to modulate STDP in OPAB indicates that this model can readily be used to investigate modulators of synaptic plasticity.Previous work suggested that dopamine gates the induction of LTP in mice prefrontal cortex [10].This data contrasts with our findings, but may be explained by the difference of species, model, maturity, and specificity of the brain region.One could hypothesize that the developing cortex may favor the learning of new motor skills through overall dopamine-mediated system activation, whereas a mature motor cortex may be wired for learning fine motor skills, which may refine neuronal circuitry through inhibitory signals and fine-tuned functional reorganization.Such a switch in mouse STDP response has been shown to occur during development, hinting at dramatic changes in STDP outcome depending on the cellular context [11,12].Our OPAB model originates from adults, but the organotypic culture of post-mortem fetal human cortical brain explants would represent an attractive model to compare STDP and evaluate the pleiotropic functions of neuromodulators during development.

Fig. 1 .
Fig. 1.OPAB and the paired stimulation protocol.(A-B) Confocal microscopy of frontal OPAB, showing neurons (MAP2) and oligodendrocytes (MBP).Nuclei are visible through Dapi staining in blue.Images are representative of a 10 μm Z stack.Scale bar = 30 μm. (C) Confocal microscopy of a frontal OPAB laying on a 3D MEA and labeled with a Live/Dead marker, staining living cells in green (left), and dead cells in red (right).Scale bar = 30 μm.The micrographs highlight that many cells are alive.The lower right inset shows a saturated image of the same area, revealing the pattern of four electrodes (white).(D) top left panel shows the microelectrode array (MEA) and the cartoon of a stimulated electrode (orange), paired-electrode (dark cyan), the paired-area (light cyan) and the non-paired electrodes (grey).The area around the stimulus electrode and the ground (both in white).The right and bottom panels show the schematic representation of the 5 test blocks and 4 training blocks in the paired stimulation protocol.The protocol was repeated for three consecutive days.In the training blocks the stimulus electrode and paired electrodes were stimulated repeatedly with a time difference of 12 ms.In the test blocks, only the stimulus electrode was stimulated and we recorded the 'response' in the paired electrode.(For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Fig. 2 .
Fig. 2. OPAB exhibit inter-day STDP.(A) Representative traces of the average recorded LFP in the paired area and non-paired area of one OPAB over the three days.(B) Representative heatmap of the average absolute response difference in the non-paired electrodes and paired area in one OPAB at T 48 0. Grey: excluded/ absent electrodes.Similar heat maps for 24 and 48 h from all the OPABs are provided as Supplementary Figs.S1-S2, respectively.(C-E) Response (calculated as the averaged peak activity across the paired-area) recorded across the test sessions on day 1 (0 h, C), day 2 (24 h, D) and day 3 (48 h, E).Each data point corresponds to an average of the n = 10 recordings in a test block.No statistical significance was found: C (p = 0.354, 1-way repeated measures ANOVA; paired T-Test between paired and non-paired electrodes of the T 0 4-T 0 0 difference p value = 0.24), D (p = 0.303, 1-way repeated measures ANOVA; paired Wilcoxon signed-rank test between paired and non-paired electrodes of the T 0 4-T 0 0 difference p value = 0.29), E (p = 0.222, 1-way repeated measures ANOVA; paired T-Test between paired and non-paired electrodes of the T 0 4-T 0 0 difference p value = 0.33).(F-G) Average absolute response change in the paired and non-paired electrodes at 24 h (abs (T 24 0 -T 0 0), F) and 48 h (abs (T 48 0 -T 0 0), G).A significantly higher response change was observed in paired electrodes compared to non-paired electrodes both at 24 h (p = 0.035, Wilcoxon signed-rank test) and 48 h (p = 0.0031, Wilcoxon signed-rank test).(F-I) Pie inserts represent the proportion of OPABs with Paired > Non-Paired (P > NP, cyan) and Paired < Non-Paired (grey).(C-G) n = 14 slices from 9 donors.(H-I) To examine the effect of dopamine, OPABs were treated with 10 μM dopamine during the first training block at 0 h.Dopamine treatment prevented a response change in the paired-electrode, with in fact the non-paired electrodes showing larger response change than the paired-electrode at 24 h (++: p < 0.05, Wilcoxon signed-rank test, H).No difference of response was observed between the paired-electrode and non-paired electrodes at 48 h (I).(H-I) n = 10 OPABs from 6 donors.(For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)