Ethics statement. Human brain tissue was obtained from a local biobank operated through the Department for Neurosurgery at the Faculty of Medicine, University of Freiburg (AZ 472/15_160880). All experiments carried out in this study were approved by the local ethics committee (AZ 593/19) and/or were performed according to German animal welfare legislation and local authorities [approved by the animal welfare officers of the Faculty of Medicine at the University of Freiburg (AZ X-17/04C)]. Animals were maintained under a 12-hour light/dark cycle with access to food and water ad libitum. Every effort was made to minimize animal pain and distress.

Preparation of acute human cortical slices. After resection, cortical access tissue was immediately transferred to an oxygenated extracellular solution containing (in mM): 92 NMDG, 2.5 KCl, 1.25 NaH₂PO₄, 30 NaHCO₃, 20 HEPES, 25 glucose, 2 thiourea, 5 Na-ascorbate, 3 Na-pyruvate, 0.5 CaCl₂, and 10 MgSO₄, (pH = 7.3–7.4) at approximately 10°C [NMDG-aCSF; (Gidon et al., 2020; Ting et al., 2018)]. Prior to slicing, cortical tissue was embedded in low melting point agarose (Sigma Aldrich, #A9517; 1.8% (w/v) in phosphate-buffered saline). Tissue sections (400 µm) were cut with a Leica VT1200S vibratome perpendicular to the pial surface in the same solution at 10°C under continuous oxygenation (5% CO₂/95% O₂). Slices were transferred to cell strainers with 40-µm pores and placed in NMDG-aCSF at 34°C. Subsequently, sodium levels were gradually increased as previously described (Ting et al., 2018). After recovery, slices were maintained for further experimental assessment at room temperature in an extracellular solution containing (in mM): 92 NaCl, 2.5 KCl, 1.25 NaH₂PO₄, 30 NaHCO₃, 20 HEPES, 25 glucose, 2 thiourea, 5 Na-ascorbate, 3 Na-pyruvate, 2 CaCl₂, and 2 MgSO₄. Cortical slices from all human samples were macroscopically normal and showed no overt pathology.

Preparation of acute mouse cortical slices. Adult mice [C57BL/6J, B6.129-Synpo^tm1Mndl/Dllr (referred to as Synpo^d/d) and B6.Cg-Synpo^tm1MndlTg(Thy1-Synpo/GFP)1Dllr/Dllr (referred to as Thy1-GFP/Synpo^T/- x Synpo^d/d); 6–11 weeks old] were used in this study. In Synpo^d/d animals, the synaptopodin coding sequence is replaced by the lacZ sequence (encoding β-galactosidase), thereby achieving a null synaptopodin allele (Deller et al., 2003). This mouse strain was backcrossed onto the C57BL/6 genetic background for at least 10 generations. In experiments involving synaptopodin-deficient preparations, age-matched C57BL/6J animals served as controls, and experimental findings were confirmed using wild type littermates from Synpo^d/d mice. For preparing acute slices, animals were anesthetized with isoflurane and rapidly decapitated. Brains were rapidly removed, washed in chilled (approximately 10°C) aCSF, and embedded in low melting point agarose (Sigma-Aldrich #A9517; 1.8% w/v in phosphate-buffered saline). Coronal sections of the mPFC were prepared using a Leica VT1200S vibratome in NMDG-aCSF with the brain tilted dorsally at a 15° angle. Slice recovery and maintenance prior to experimental assessment were performed as described above for acute human cortical slices.

Pharmacology. Acute cortical slices prepared from individual brain samples were randomly assigned to atRA or control (vehicle-only) treatment groups. Treatment with atRA was performed after slice recovery by adding atRA (1 µM, Sigma Aldrich, #R2625) to the extracellular holding solution at a final concentration of 0.05% (v/v in DMSO). The control group from the same set of slices was handled identically but treated with vehicle-only (DMSO). Anisomycin (10 µM, Abcam, #ab120495) and actinomycin D (5 µg/ml, Sigma Aldrich, #A9415) were added to the holding solution 10 minutes before the addition of atRA. Sections were treated for at least 6 hours before experimental assessment.

Whole-cell patch-clamp recordings. Whole-cell patch-clamp recordings of superficial (layer 2/3) cortical pyramidal neurons were carried out at 35°C in a bath solution containing (in mM): 92 NaCl, 2.5 KCl, 1.25 NaH₂PO₄, 30 NaHCO₃, 20 HEPES, 25 glucose, 2 thiourea, 5 Na-ascorbate, 3 Na-pyruvate, 2 CaCl₂, and 2 MgSO₄. For experiments with acute mouse brain slices, superficial (layer 2/3) pyramidal cells in the dorsomedial prefrontal cortex were visually identified using an LN-Scope (Luigs & Neumann, Ratingen, Germany) equipped with infrared dot-contrast and a 40x water-immersion objective [numerical aperture (NA) 0.8; Olympus]. For experiments with human cortical slices, superficial (layer 2/3) pyramidal cells were visually identified on the pia/white matter axis at a distance of 500–1000 µm from the pial surface. Electrophysiological signals were amplified using a Multiclamp 700B amplifier, digitized with a Digidata 1550B digitizer, and visualized with the pClamp 11 software package. For sEPSC and intrinsic cellular property recordings, patch pipettes (tip resistance: 3–5 MΩ) contained (in mM): 126 K-Gluconate, 4 KCl, 10 HEPES 4 MgATP, 0.3 Na₂GTP, 10 PO-Creatine, 0.3% (w/v) biocytin (pH = 7.25 with KOH; 285 mOsm/kg). For sEPSC recordings, pyramidal neurons were held at -70 mV in voltage-clamp mode. To record intrinsic cellular properties in current-clamp mode, a pipette capacitance of 2.0 pF was corrected and series resistance was compensated using the automated bridge balance tool of the Multiclamp commander. Input-output (I-V)-curves were generated by injecting one-second square pulse currents starting at -100 pA and increasing in 10 pA increments (sweep duration: two seconds). Series resistance was monitored and recordings were discarded if the series resistance reached >30 MΩ.

One superficial (layer 2/3) cell in human neocortical slices (Fig. 1; atRA group) with sEPSC amplitude = 38.4 pA and sEPSC frequency = 6.5 Hz showed interneuron characteristics and was therefore excluded from the analysis. The series resistance of one cell from a human cortical slice (Fig. 1; control group) exceeded 30 MΩ during I-V curve recording. The respective I-V-curve was therefore excluded from further analysis. In five human superficial pyramidal neurons (Fig. 1), the number of sweeps in the I-V-curve recordings was lower compared to other recordings (40 sweeps vs. 60 sweeps). Thus, cells were excluded from further analysis of action potential frequency. Furthermore, one I-V-curve recording in the actinomycin-only treated group became unstable during the last sweeps and was consecutively excluded from action potential frequency analysis (Fig. S5). In addition, one cell from a mouse neocortical slice (Fig. 3, S3, and S6; Synpo^WT/WT, atRA group) was excluded from further analysis because the signal displayed a marked electrical interference that caused disturbances in the baseline of the recordings. In the same data set, I-V-curve recording from one cell (Fig. S3, Synpo^WT/WT, control group) is excluded due to I-V duplication in each sweep. Finally, one whole-cell patch-clamp recording of intrinsic cellular properties (Fig. S3; Thy1-GFP/Synpo, control group) lost its integrity during the recording and was therefore excluded from further analysis.

Immunostaining, post-hoc labeling and confocal microscopy. Cortical slices were fixed in 4% PFA (prepared from 16% PFA stocks in phosphate-buffered saline according to the manufacturer’s instructions; Thermo Scientific, #28908) at room temperature and stored at 4°C overnight in the same solution. After fixation, slices were washed in phosphate-buffered saline and incubated for 1 hour with 10% (v/v) normal goat serum [NGS; diluted in 0.5% (v/v) Triton X-100/PBS] to reduce non-specific staining and increase antibody penetration. Subsequently, slices were incubated overnight at 4°C with rabbit anti-synaptopodin (Synaptic Systems, #163002; 1:1000) or rabbit anti-NeuN (Abcam, #ab104225, 1:500) antibodies; both antibodies were diluted in 10% (v/v) NGS in 0.1% (v/v) Triton X-100/PBS. Sections were washed with PBS and incubated with goat anti-rabbit Alexa Fluor 488 or goat anti-rabbit Alexa Fluor plus 555-labeled secondary antibodies (Invitrogen, #A-11034 and #A-32732, respectively) overnight at 4°C; both secondary antibodies were diluted 1:1000 in 10% (v/v) NGS in 0.1% (v/v) Triton X-100/PBS. For visualizing patched pyramidal cells, streptavidin-Alexa Fluor 488 (Invitrogen, #S32354; 1:1000) was added during the secondary antibody incubation. Sections were washed again and incubated for 10 minutes in Sudan Black B (0.1% (w/v) in 70% ethanol) to reduce autofluorescence. Sections were then incubated with DAPI for 10 minutes (Thermo Scientific, #62248; 1:5000 in PBS) to facilitate visualization of cytoarchitecture. After the final washing step, sections were transferred onto glass slides and mounted with a fluorescence anti-fade mounting medium (DAKO Fluoromount).

Confocal images were acquired using a Leica SP8 laser-scanning microscope equipped with a 20x multi-immersion (NA 0.75; Leica), a 40x oil-immersion (NA 1.30; Leica), and a 63x oil-immersion objective (NA 1.40; Leica). Image stacks for dendritic spine and synaptopodin cluster analyses were acquired with a 63x objective at 6x optical zoom (resolution: 1024 x 1024; z-step size: 0.2 µm; ideal Nyquist rate). Laser intensity and detector gain were set to achieve comparable overall fluorescence intensities throughout stacks between all groups. Confocal image stacks and single plane pictures were stored as TIF files.

Immunogold labeling of synaptopodin. Acute human cortical slices (400 µm) were fixed with microwave irradiation (Privileg 8020 E, 640 Watt, 2.45 GHz; (Jensen & Harris, 1989)) for 8 seconds on a petri dish filled with ice in 0.1% glutaraldehyde and 4% PFA [dissolved in 0.1 M phosphate buffer (PB) and 0.05 M sucrose]. The slices were kept in the same solution for 1 hour at room temperature and then transferred to 0.1 M PB. After 3 hours, 50-µm sections were prepared using a Leica VT1000S vibratome, washed for 30 minutes in 50 mM Tris-buffered saline (TBS), and incubated for 1 hour in 20% NGS (v/v) in 50 mM TBS. Subsequently, sections were incubated with rabbit anti-synaptopodin [Synaptic Systems, #163002; 1:100 in 2% NGS/50 mM TBS (v/v)] at 4°C overnight. Sections were washed for 1 hour in 50 mM TBS and incubated with a suitable secondary goat anti-rabbit antibody [Nanoprobes, #2004; 1.4 nM gold-coupled, 1:100 in 2% NGS/50 mM TBS (v/v)] at 4°C overnight. After washing for 30 minutes in 50 mM TBS, sections were post-fixed in 1% glutaraldehyde/25 mM PBS (w/v) for 10 minutes. Sections were washed again in PBS, and silver intensification (HQ Silver Enhancement Kit, Nanoprobes, #2012) was performed according to the manufacturer’s instructions. Subsequently, slices were incubated with 0.5% osmium tetroxide for 40 minutes, washed in graded ethanol (up to 50% (v/v)) for 10 minutes each, and incubated with uranyl acetate [1% (w/v) in 70% (v/v) ethanol] for 35 minutes. Slices were then dehydrated in graded ethanol (80%, 90%, 95%, 2 x 100% for 10 minutes each). Two 15-minute washing steps in propylene oxide were performed prior to incubation with durcupan/propylene oxide (1:3 for 45 minutes followed by 3:1 for 45 minutes) and durcupan (overnight at room temperature). After slices were embedded in durcupan, ultra-thin sectioning (55 nm) was performed using a Leica UC6 Ultracut. Sections were mounted onto copper grids (Plano), at which point an additional Pb-citrate contrasting step was performed (3 minutes). Electron micrographs were captured using a Philips CM100 microscope equipped with a Gatan Kamera Orius SC600 (magnification 5200x). Acquired images were stored as TIF files.

Electron microscopy. After 6 hours of treatment with atRA or vehicle-only control, slices were fixed in 4% paraformaldehyde (w/v) and 2.5% glutaraldehyde (w/v; phosphate-buffered saline) overnight. After fixation, slices were washed for 4 hours in 0.1 M phosphate buffer (PB). Subsequently, slices were incubated with 1% osmium tetroxide for 45 minutes, washed in graded ethanol (up to 50% (v/v)) for 5 minutes each, and incubated overnight with uranyl acetate (1% (w/v) in 70% (v/v) ethanol) overnight. Slices were then dehydrated in graded ethanol (80%, 90%, 98% for 5 minutes each, 2 x 100% for 10 minutes each). Subsequently, two washing steps in propylene oxide for 10 minutes each were performed prior to incubation with durcupan/propylene oxide (1:1 for 1 hour) and transfer to durcupan (overnight at room temperature). Slices were embedded in durcupan, and ultra-thin sectioning (55 nm) was performed using a Leica UC6 Ultracut. Sections were mounted onto copper grids (Plano), at which point an additional Pb-citrate contrasting step was performed (3 minutes). Electron microscopy was performed using a Philips CM100 microscope equipped with a Gatan Orius SC600 camera at 3900x magnification. Acquired images were saved as TIF files and analyzed by an investigator blinded to experimental conditions.

Quantification and statistics. Electrophysiological data were analyzed using Clampfit 11 from the pClamp11 software package (Molecular Devices). sEPSC properties were analyzed using the automated template search tool for event detection. Specifically, AP detection was performed using the input/output curve threshold search event detection approach, whereas AP frequency was determined based upon the number of APs detected during a given injection. Artifacts in electrophysiological recordings were excluded from further analysis. Immuno-labeled synaptopodin clusters in superficial (layer 2/3) pyramidal cells of the human cortex were analyzed in image stacks of second- and third-order dendritic branches. Synaptopodin clusters that colocalized with dendritic spines (either spine neck or head; Fig. 2D) were explored in further analyses. Single-plane images of both synaptopodin-positive and -negative clusters were extracted from image stacks at the point of maximum spine head cross-sectional area and stored as TIF files. Blinded analyses of spine head cross-sectional area and synaptopodin cluster size were performed manually using the ImageJ software package (available at http://imagej.nih.gov/ij/). Here, the outer borders of synaptopodin clusters and spine heads were marked independently of their overall fluorescence intensity. Data were transferred and stored in Excel files. Ultrastructural analysis of spine apparatus organelles was performed using single-plane images of human cortical excitatory synapses where pre- and post-synaptic structures could be readily identified. The cross-sectional areas of spine apparatus organelles were determined manually using the ImageJ software package, independent of their shape and internal structural organization.

Statistical analyses were performed using the GraphPad Prism 7 software package. Two-group comparisons were performed using a Mann-Whitney-U test; U-values for statistically significant differences are reported in the figure legends. A Kruskal-Wallis test with Dunn’s multiple comparisons was used to compare more than two groups. Correlation of individual data points was visualized by a linear regression fit and analyzed by computing Spearman r. For statistical evaluation of XY-plots from, we used a RM two-way ANOVA test (repeated measurements/analysis) with Sidak’s multiple comparisons. For the comparison of more than two groups in XY-plots, Tukey’s multiple comparisons were applied. For the in-sample analysis of human cortical slices (paired experimental design), we used a Wilcoxon matched-pairs signed-rank test. P values <0.05 were considered statistically significant (*p < 0.05, **p < 0.01, ***p < 0.001); results that did not yield significant differences are designated “ns”. Statistical differences in XY-plots were indicated in the legend of the figure panels (*) when detected through multiple comparisons, irrespective of their localization and the level of significance. In the text and figures, values represent the mean ± standard error of the mean (s.e.m.).

Graphical illustrations. Figures were prepared using Photoshop graphics software (Adobe, San Jose, CA, USA). Image brightness and contrast were adjusted.