The IRFMN adheres to the principles set out in the following laws, regulations, and policies governing the care and use of laboratory animals: Italian Governing Law (D.lgs 26/2014; Authorisation no. 19/2008-A issued March 6, 2008 by Ministry of Health); Mario Negri Institutional Regulations and Policies providing internal authorization for people conducting animal experiments (Quality Management System Certificate–UNI EN ISO 9001:2008–Reg. N° 8576-A); the NIH Guide for the Care and Use of Laboratory Animals (2011 edition) and EU directives and guidelines (EEC Council Directive 2010/63/UE). They were reviewed and approved by the Mario Negri Institute Animal Care and Use Committee which includes ad hoc members for ethical issues and by the Italian Ministry of Health (Decreto no. D/07/2013-B).

C57BL/6J mice were bred (one male and two females per cage) in a specific pathogen free vivarium at a constant temperature (21 ± 1°C) and relative humidity (60 ± 5%), with a 12 h light–dark cycle and ad libitum access to food and water. All efforts were made to minimize animal suffering and to reduce the number of animals used.

Organotypic cortical brain slices were obtained under sterile conditions from the prefrontal cortex of C57BL/6J or Cx3Cr^+/GFP (expressing green fluorescent protein in microglial cells) mouse pups (P1-P3), as previously described (Dossi et al., 2013; Pischiutta et al., 2016) and as represented in Supplementary Figure 1. Briefly, the brain was removed from the skull and embedded into an agarose solution (3% w/V, Promega). Agarose blocks containing mesencephalic and forebrain areas were fixed onto the specimen stage of a vibratome (Leica, VT 1000S), with super glue. The fixed brain was kept in ice-cold (4°C) artificial cerebral spinal fluid (ASCF, NaCl = 87 mM, NaHCO₃ = 25 mM, NaH₂PO₄ = 1.25 mM, MgCl₂ = 7 mM, CaCl₂ = 0.5 mM, KCl = 2.5 mM, D-glucose = 25 mM, sucrose = 75 mM, Penicillin = 50 U/ml, Streptomicin = 50 μg/ml) and oxygenated with 95% of oxygen and 5% CO₂. Prefrontal cortex slices with 200 μm of thickness were cut and transferred onto petri dishes with cold ACSF (4°C). Intact slices were transferred to membranes of tissue culture inserts (Millicell Culture insert, 0.4 μm pore size, Merck-Millipore). The slices were maintained in culture (37°C, 5% CO₂), two by two per insert in a 6-well plate. Each well was filled with 1 ml of culture medium [MEM-Glutamax 25%, basal medium eagle 25% (Invitrogen), horse serum 25% (Euroclone), glucose 0.6%, Penicillin 100 U/ml, Streptomicin 100 μg/mL (Euroclone); pH = 7.2] for 2 days, and then cultured in neurobasal medium supplemented with B27 (NB/B27, B27 1:50, L-glutamine 1:100, penicillin 100 U/mL, streptomycin 100 μg/mL) with medium changes every other day.

After 1 week in culture, cortical slices were subjected to controlled cortical impact (CCI) delivered by an electromagnetic impactor device (Impact One^®, Leica, USA) to model TBI. A stereotactic apparatus, connected to the impactor was used to set the trauma depth. Each insert with slices was taken from the plate and placed under the impactor, with the proper medium supply. The damage was driven by a 1 mm diameter impactor tip, at 1.5 m/s of speed and 0.1 sec dwell time. At first, we tested three different deformation depths: 0 μm (the piston just touched the slice, without penetrating it), 100 μm or 200 μm, the latter corresponding to the thickness of the whole slice (experimental design in Figure 1A). The 100 μm depth induced a significant increase in PI incorporation with no tissue perforation, thus it was selected for the next experiments (experimental design in Figure 1C). Control slices were subjected to the same protocol, without injury.

Controlled cortical impact induced cell death was measured by propidium iodide (PI) incorporation assay at tissue level. At 24 h and 48 h after injury, slices were incubated with PI (2 μM diluted in NB/B27) for 30 min, washed with PBS and photographed using an Olympus IX71 inverted microscope. Fluorescence intensity per slice was measured using ImageJ’s Integrated Density function and the value was normalized over the slice area. In order to evaluate cell death in a region-specific manner, the fluorescence intensity was also evaluated in specific regions of interest (ROIs), drawn in a concentric manner, spaced 0.5 mm, starting on the lesion site to the slice periphery.

Cell death was evaluated by a Lactate dehydrogenase (LDH) assay in the culture medium collected at time 0 (before injury), 4, 8, 24, and 48 h after injury and stored at −20°C. LDH release was measured using Cytotox 96 Non-Radioactive Assay (Promega) following manufacturer instructions. Absorbance was read at 490 nm (TECAN, Infinit^®200 PRO).

To assess neuronal damage, we quantified the release of three different neuronal markers in the culture medium: the neurofilament light chain (NfL), the tubulin associated unit (Tau) and the Ubiquitin C-terminal hydrolase L1 (UCH-L1). Injury to astrocytes was assessed by measuring the release of the glial fibrillary acidic protein (GFAP). Culture medium was collected 48 h post-injury, and stored at −20°C. Biomarker releases in the culture media were measured by single molecule array (simoa) immunoassay (Quanterix, Billerica, MA, USA). Analyses were run using commercial simoa assays on a SR-X Analyzer: NfL advantage assay (catalog #103400) and Neurology 4-Plex B Kit (catalog # 103345). Samples were diluted 1:200 in diluent buffer. A single lot of reagents was used for all samples. All samples were examined in duplicate.

Forty-eight hours post-CCI, the slices were fixed 1 h in paraformaldehyde (PFA) 4%, dehydrated two overnights in a 30% sucrose solution, rapidly frozen in N-pentane (−45°C) and stored at −80°C. For immunofluorescence staining, the slices were detached from the transwell and incubated overnight in free-floating with the primary antibodies: anti microtubule-associated protein (MAP-2, 1:500 Abcam, ab11267), glial fibrillary acidic protein (GFAP, 1:2000, MERCK, AB5804), and ionized calcium binding adaptor molecule 1 (Iba-1, 1:1000, Wako, 019-19741), followed by 1 h incubation with secondary antibodies anti-mouse alexa Fluor^®488, or anti-rabbit alexa Fluor ^®594 (1:500, Invitrogen).

Confocal microscopy was done on a Nikon A1 confocal scan unit, managed by NIS elements software. Whole cortical brain slices were imaged at laser excitations of 488 or 594 nm with a sequential scanning mode to avoid bleed-through effects. An overview was obtained with a 10 × 0.5 NA and a concentric grid centered in the lesion site and distanced 500 μm was used to define concentric areas: the lesion, perilesion and periphery (Figure 2A). Within each area, 4 ROIs were acquired with a 40 × 0.75 NA objective. Each image was 1,024 × 1,024, having a pixel size of 0.21 μm and a step size of 0.85 μm for MAP-2 and a pixel size of 0.31 μm and a step size of 1 μm for GFAP and Iba-1. In order to standardize the image acquisition and minimize inter-sample variation, all acquisition parameters were kept constant throughout microscopy sessions.

We quantified the complexity of neuronal arborizations in terms of number, size and density of branches on MAP-2 stained slices, by using an originally developed ImageJ algorithm to segment and skeletonize the signal as previously described (De Paola et al., 2023). Briefly, the background was normalized imposing a minimum gray value of 1,200, and the image was corrected with a median filter with a radius of 5 and binarized over the z stack by the Li method (Li and Tam, 1998). An extended focus image of the binary image was obtained and the skeletonize function was applied. Then the ‘Analyze Skeleton 2D/3D’ ImageJ plugin was applied with no endpoint pruning.

Glial fibrillary acidic protein and IBA1% stained area was measured on 5 focal plans using image J. Microglia morphology was analyzed, as described (Zanier et al., 2015; Moro et al., 2021). A size threshold (40–500 μm²) was used to select cells to be analyzed for the following shape descriptors: area, perimeter, Feret’s diameter and circularity. Mean single-cell values for each parameter were used for statistics.

Results within lesion, perilesion and periphery are shown as the mean values of the 4 ROIs. Since no differences were present between concentric areas in CTRL slices, results are shown as the average of the 12 ROIs.

Forty-eight hours after CCI, slices were collected, and total RNA was extracted by using the Ribospin II extraction kit (GeneAll) following the manufacturer’s protocol. The RNA concentration was determined with a Nanodrop spectrophotometer and 200 ng/μl RNA from each sample was reverse transcribed to cDNA using TaqMan^® qPCR RT Master Mix (Applied Biosystems by life technologies) following the manufacturer’s protocol. Real-time RT-PCR was performed, and the relative gene expression to the reference gene RPL27 was determined by ΔΔCt method. Data are expressed as log₂ of the fold difference from the CTRL group. The primers used are listed in Table 1. Primers were designed using the Primer-BLAST tool (NCBI, United States),¹ and Prime3 open-source software.²

Mesenchymal stromal cells (MSCs) were isolated from human umbilical cord perivascular cells (Fraga et al., 2013) and the secretome was collected as a conditioned medium in passage 8 (P8) as previously described (Mendes-Pinheiro et al., 2019). Briefly, cells were seeded at a density of 4,000 cells/cm² and incubated with α-MEM medium (Gibco, USA), supplemented with 5% human platelet lysate (HPL) 1% penicillin-streptomycin, 0,04% heparin. After 3 days in culture, cells were washed 4 times with PBS without Ca^2+/Mg²⁺ (Invitrogen) and incubated for 24 h with Neurobasal-A medium (Gibco, USA) supplemented with 1% of penicillin-streptomycin (Fraga et al., 2013). On the next day, the conditioned medium containing the MSC-secretome was collected and centrifuged at 1,200 rpm (Thermo Electron Corporation, EUA) for 5 min to remove any cell debris. Then, secretome was snap frozen in liquid nitrogen and stored at −80°C. Before treating the slices, the collected secretome was supplemented with 2% B27 (Euroclone, IT) and 1% L-glutamine (Euroclone, IT), and applied to the organotypic culture. Treatment was applied 1 h post-CCI and no medium changes were performed during the following 48 h.

For the interactome pathway analysis, we used Elsevier’s Pathway Studio version 10.001.³ To establish the various relationships among the different altered validated proteins, ResNet Pathways Studio Propriety database was utilized to infer all the interactions. Interactome network was generated using a “direct interaction” algorithm for cellular processes, biological process mapping as well as the proposed pathway interactions.

Cortical slices were allocated to injury and treatments by a list randomizer.⁴ All evaluations were done blinded to injury/treatment status. Data are presented as mean ± SEM. The choice between parametric or non-parametric tests was based on passing the Shapiro–Wilk normality test, and data distribution was inspected by QQ plot. For each experiment, the figure legend reports the statistical analysis of data. For statistical analyses we used GraphPad Prism (GraphPad Software Inc., USA, version 9.2.0). For the pathway analysis, we used Fisher’s statistical test to determine if there are non-random associations between two categorical variables organized by specific relationship (protein interaction and biological process). The algorithm compares the sub-network distribution to the background distribution using a one-sided Mann–Whitney U-Test, and calculates a p-value indicating the statistical significance of the difference between two distributions.

Traumatic brain injury severity was tailored by impacting cortical slices with CCI deformation depth of 0, 100, or 200 μm (Figure 1A). PI incorporation was highest after CCI at 100 μm deformation depth (p < 0.01, Figure 1B). CCI at 200 μm depth resulted in the complete perforation of the brain tissue and loss of the lesion core (Figure 1B) leading to lower overall PI integrated density values. Thus, CCI at 100 μm depth was selected for the subsequent experiments (Figure 1C).

We quantified regional PI incorporation by drawing concentric ROIs centered in the lesion core (Figure 1D). PI incorporation in TBI slices showed the maximum value in the lesion core (p < 0.001) and a gradual decrease toward peripheral ROIs at both 24 and 48 h post-injury (Figure 1D). AUC in TBI slices was higher at 48 compared to 24 h (p < 0.05) indicating a spatial and temporal increase in cell death (Figure 1D right insert).

After TBI, LDH release progressively increased up to 48 h with a significant difference already observed at 4 h (p < 0.05) compared to controls (Figure 1E) thus confirming the increase of cell death over time. CTRL slices showed a low PI incorporation and LDH release uniform over space and time.

To investigate TBI-induced modifications of neuronal cytoarchitecture we quantified MAP-2 immunofluorescent signal on whole-mounted slices at 48 h post injury, selecting ROIs at lesion, perilesion, and periphery (Figure 2A). Compared to CTRL (Figure 2B), TBI slices showed a strong decrease of MAP-2 staining in the lesion and perilesional areas (Figure 2C). MAP-2 morphometric parameters showed that TBI decreased the complexity of neuronal arborizations, with significant reduction in the number of branches (p < 0.001, Figure 2D), in the size of the longest branch (p < 0.001, Figure 2E) and in the network density (p < 0.001, Figure 2F) at the lesion core, while showing values comparable to controls at the periphery. In accordance, gene expression analysis at 48 h indicated a TBI-induced downregulation of MAP-2 and the neuronal associated growth factor BDNF (p < 0.01, Figure 2G).

When analyzing culture media, we found a significant increase of NfL (p < 0.05), Tau (p < 0.01) and UCH-L1 (p < 0.01) levels after TBI, confirming neuronal death (Figure 2H).

Histological analysis of GFAP⁺ astrocytic marker, showed a uniform labeling in CTRL slices, cells displayed the typical morphology of cortical astrocytes with defined cell somata and long processes protruding outward (Figure 3A). At 48 h post injury, GFAP stained area significantly decreased in all ROIs (p < 0.001), with the highest effect in the lesion core (Figures 3B, C). TBI also induced a significant release of GFAP in culture media 48 h post-injury (p < 0.01, Figure 3D), concurrent with an upregulation of GFAP gene expression (p < 0.001, Figure 3E).

We found a reduction of IBA1⁺ stained area in the lesion area of TBI slices compared to CTRL (p < 0.01, Figures 4A–C) that progressively recovered toward the periphery. Analysis of shape parameters (Figures 4D–G) showed a decreased cell area (p < 0.05, Figure 4D), perimeter (p < 0.01, Figure 4E) and Feret’s diameter (p < 0.05, Figure 4F) in the lesion core of TBI slices, indicating the polarization of microglial cells toward an amoeboid shape with a less ramified morphology. Accordingly, an increase in circularity value was observed in the lesion core (p < 0.01, Figure 4G). In the periphery, morphometric parameters were comparable to controls. Gene expression analysis revealed that TBI-induced an upregulation of CD11b microglial pan marker (p < 0.01, Figure 4H).

Using the differentially expressed mRNA and proteins released in the culture media, system biology analysis was conducted to identify cellular processes represented in our in vitro TBI model. Altered proteins (NfL, Tau, UCH-L1, MAP2, GFAP) were related with other two TBI well established biomarkers (S100-B and neuron specific enolase, NSE) (Supplementary Figure 2). Pathway analysis revealed an association with neuronal death, nerve injury and degeneration, as well as glial reaction and astrocytosis (Supplementary Figure 2).

We assessed the effects of MSC-derived secretome administration, delivered to TBI slices 1 h after injury in the culture media (experimental design in Figure 5A). Cell death analysis at 48 h by PI incorporation assay revealed a significant protection induced by MSC-secretome in the lesion core and the adjacent regions (p < 0.001, Figure 5B), resulting in a global reduction of cell death (AUC TBI vs TBI + MSC-sec: p < 0.01, Figure 5B insert). Analysis of MSC-induced effects on neurons showed a rescue effect on TBI-induced downregulation of MAP-2 (p < 0.05) and BDNF (p < 0.01, Figure 5C), a decrease of NfL release in the culture media (Figure 5D), and an increase of neurite branches in the periphery of the slice (Figures 5E, H), while no changes were detected for the longest branch size and network density (Supplementary Figure 3). Analysis on glial cells revealed no differences on GFAP and an up-regulation of microglial CD11b expression (p < 0.05) induced by MSC-secretome compared to TBI untreated slices (Figure 5F). Histological analysis of microglia activation revealed an increased stained area in MSC-secretome treated vs untreated TBI slices (Figures 5G, H, Two-way ANOVA p of treatment <0.01), but no changes in shape descriptors (Supplementary Figure 4).