Epac2 Elevation Reverses Inhibition by Chondroitin Sulfate Proteoglycans In Vitro and Transforms Postlesion Inhibitory Environment to Promote Axonal Outgrowth in an Ex Vivo Model of Spinal Cord Injury

Cortical neuron culture.

Cortices of Sprague Dawley rats at postnatal days 0–1 (mixed sexes) were harvested as a source for culturing cortical neurons. The tissue was dissociated enzymatically with 50 U/ml papain (Worthington) in retinal buffer at pH 7.4 composed of 15 mm HEPES (Sigma-Aldrich) buffered Hanks balanced salt solution (Invitrogen) containing 300 μm d-l cysteine (Sigma-Aldrich) and incubated at 37°C for 30 min. The papain action was stopped by using 10% FBS (Thermo Fisher Scientific), and cells were resuspended in Neurobasal medium (Thermo Fisher Scientific) supplemented with 2% B-27 (Thermo Fisher Scientific), 1% Glutamax (Thermo Fisher Scientific), and 100 U/ml penicillin and 100 μg/ml streptomycin (Sigma-Aldrich). Cortical neurons were plated at 40,000 neurons/ml on round 13 mm glass coverslips (BDH) coated overnight with 10 μg/ml poly-d-lysine (PDL; Sigma-Aldrich) and cultured for up to 48 h at 37°C in a 5% CO₂/95% air incubator (NU-581DE; Nuaire).

DRG neuron and explant cultures.

DRGs were dissected from Sprague Dawley rats at postnatal days 0–5 (mixed sexes), collected in Ham's F12 medium (Thermo Fisher Scientific), and trimmed to remove roots. Explants were plated directly when needed. For dissociating DRG neurons, ganglia were transferred to 1 ml retinal buffer containing 50 U/ml papain as described above. The tissue was then transferred to 100 μl Hanks balanced salt solution containing 0.25 mg/ml trypsin inhibitor (Sigma-Aldrich) and 50 μg/ml DNase (Sigma-Aldrich), followed by trituration using a Gilson P200 pipette until a single-cell suspension was achieved. The dissociated neurons were diluted to the required density with Neurobasal medium supplemented as described above plus nerve growth factor (100 ng/ml; Sigma-Aldrich). DRG neurons were plated at 5000 neurons/ml on 13 mm coverslips coated overnight with PDL as described above and 2 μg/ml laminin (Thermo Fisher Scientific) and cultured for up to 48 h at 37°C in a 5% CO₂/95% air incubator.

Microglial and astrocyte cultures.

Primary mixed microglia and astrocytes were cultured as previously described (Georgieva et al., 2018) from the cortices of Sprague Dawley rats at postnatal days 3–6 (mixed sexes) and plated on 10 μg/ml PDL (Sigma-Aldrich) coated 60 mm Petri dishes (Biofil Triple red TCD010060). After 4 d of incubation at 37°C in a 5% CO₂/95% air incubator, the presence of two layers of cells was confirmed: one layer of astrocytes firmly attached to the surface of the dish and a layer of microglia visible as large spherical bright cells on top of the astrocyte layer. For microglia plating, medium was collected after a gentle swirl, centrifuged at 200 × g for 10 min at 4°C and the pellet resuspended in warm medium. The microglia suspension was plated on 18 mm uncoated sterile glass coverslips and used the following day. For astrocyte plating, once microglial cells were collected, cultures were washed repeatedly with warm PBS and incubated with trypsin/EDTA 0.25% (Sigma-Aldrich) for 5 min at 37°C. Trypsin action was stopped by addition of 10% FBS, centrifuged at 200 × g for 10 min at 4°C, and replated in two PDL-coated flasks. Cultures were maintained for 3 weeks to ensure maturation of the astrocytes (Smith et al., 1990). Medium was changed every 2 d, and cells were split when confluent. Microglia and astrocytes were maintained in medium composed of DMEM with 10% FBS and 100 U/ml penicillin and 100 μg/ml streptomycin.

Soluble agonist and antagonist treatments for neurons in vitro.

Soluble agonist Sp-8-BnT-cAMPS (S-220) and antagonist ESI-05 (both Biolog) were used according to the manufacturer's instructions for specific Epac2 activation and inactivation in neurons, respectively. In some experiments, chicken extracellular CSPGs (CC117; Sigma-Aldrich) were added to culture media at 0.5 μg/ml at the time of cell plating.

In vitro treatments for microglia and astrocytes.

Mature astrocytes were plated in 13 mm PDL-coated coverslips at a density of 45,000 cells/ml. Microglia were plated in 18 mm uncoated coverslip at a density of 70,000 cells/ml. Both cell types were allowed to attach and grow overnight. The next morning, media were changed to serum-free media before treatments. Three conditions were studied: control, lipopolysaccharide (LPS), and simultaneous LPS and S-220 treatment. Therefore, 10 μg/ml LPS was added to the media alone or together with S-220 at 2.5 μm, a concentration showing effects in promoting neurite outgrowth. The treatments were left for 4 h.

siRNA transfection.

For siRNA knockdown, DRG neurons were transfected with 3 μg of siRNA duplex (Epac2: ATC CGT GAA TGT AGT CAT TTA) (QIAGEN) using the Amaxa Nucleofector II device (Lonza Walkersville) according to the manufacturer's instructions. To establish transfection efficiency using this method, 3 μg of 3′-fluorescein-tagged Allstars negative control siRNA (QIAGEN) was transfected into neonatal DRG neurons, which were plated on 2 μg/ml laminin-coated glass coverslips. After 16 h, cells were fixed in 4% PFA (Sigma-Aldrich), immunolabeled using anti-GAP43, and analyzed using an AF6000LX microscope (Leica Microsystems). Fluorescein-labeled neurons were compared with the total number of GAP43-positive neurons to obtain the transfection efficiency, which was calculated as 80%.

Lentiviral vector (LV) production.

We had technical limitations for the construction of LVs carrying the specific Epac2 cDNA because the available sequence was not suitable for the subcloning in the lentiviral transfer vector. Therefore, as a proof of concept, we constructed Epac-LV as follows. Human Epac1-YFP construct was kindly provided by Dr. Kees Jalink at The Netherlands Cancer Institute (Amsterdam, The Netherlands). hEpac1-YFP cDNA was subcloned downstream of the CMV promoter in the lentiviral transfer vector pRRL. Self-inactivating LVs were made by cotransfecting HEK293T cells using a three-plasmid system (Dull et al., 1998). pRRL/hEpac1-YFP or pRRL/GFP plasmids were transfected together with pPACK and pENV plasmids using the calcium precipitation method (Dull et al., 1998). Lentiviral particles were harvested 45 h after transfection and concentrated using polyethylene glycol precipitation. Viral particles were resuspended in sterile PBS after precipitation. The titers were determined by transduction of HEK293T cells using serially diluted viral stocks. The titers were 4.6 × 10⁹ TU/ml for LV/hEpac1-YFP (referred as LV/Epac hereafter), and 2.2 × 10⁹ TU/ml for LV/GFP. The viral stocks were aliquoted and stored at −80°C. Cortical neurons to be transduced with the LV were plated in 24 well plates (Greiner) at a density of 60,000 cells/ml; 2.2 × 10⁶ TU of LV/GFP or LV/hEpac1-YFP was added per well to allow transduction. Media were replaced the day after and then changed every 2 d. Cells were maintained for 7 d to ensure full expression of the transgenes. Anti-GFP-labeled neurons were compared with the total number of β-tubulin-(III) -positive neurons to obtain the transduction efficiency.

FRET SE microscopy.

Fluorescent resonance energy transfer sensitized emission (FRET SE) was performed in live DRG neurons as previously described (Tucker, 2014). FRET SE microscopy allowed us to demonstrate the level of Epac2 protein activity in the presence of the agonist, S-220, or the antagonist, ESI-05, respectively, in the growth cones of live DRG neurons. Two fluorophores (CFP and YFP) were fused to the two termini of Epac2 protein in the Epac FRET construct (see below). When the distance between the two is <10 nm, which occurs when the protein is inactive, a FRET signal is produced. On activation of the protein, a conformational change occurs and the distance between the two fluorophores becomes >10 nm, abolishing the FRET signal. Thus, high FRET signals correlate with low Epac2 activation and low FRET signals correlate with high Epac2 activation. Briefly, neurons were transfected with the construct mTurqDel-Epac (dDEPCD) cp173Venus (d)-Venus (d) (H74) (Klarenbeek et al., 2011) (kindly provided by Dr. Kees Jalink, The Netherlands Cancer Institute, Amsterdam, The Netherlands) using the Amaxa Nucleofector II device (Lonza Walkersville). The cells were diluted in medium after transfection and cultured in 50 mm glass-bottomed dishes (MatTek) coated with PDL and 5 μg/ml laminin. One day after transfection, the medium was topped up to 5 ml and buffered with 15 mm HEPES buffer. Using an AF6000LX imaging system (Leica Microsystems) configured correctly for FRET, images were taken of a selected growth cone in the CFP, YFP, and FRET channels every 2 min for a total of 30 min. For each experiment, the first four images were taken without application to act as a baseline as the system stabilized; then the reagent was added after the fourth time point. The Epac2 agonist (S-220) was diluted to a concentration of 5 μm and the antagonist (ESI-05) to 10 μm. Analysis of each image (combining all 3 channels) was performed using Application Suite-Advanced Fluorescence software (Leica Microsystems) by selecting a “region of interest” within the growth cone and recording the FRET value. These data were then expressed as a percentage change from time 0.

Choice assay.

Mature astrocytes obtained as described above were plated at 5000 cells/ml on 5 μg/ml laminin custom-made dishes. These dishes consisted of 60 mm Petri dishes (CellStar, Greiner Bio-One) that had three 13-mm-diameter holes in the base of the dish using a hole puncher; 18 mm coverslips were mounted onto the base of the dish using RTV 3140 coating glue (Dow Corning) that is nontoxic to cells. Astrocytes were left to grow for 5 h before plating of DRG neurons among the astrocytes. Cocultures were further incubated at 37°C overnight in a CO₂ incubator. S-220 was added just before time lapse imaging. Time lapse recordings were taken over 30–60 min using NIS-element software and an Okolab chamber-mounted eclipse Ti inverted microscope (Nikon). The response of each DRG neurite was categorized into two different categories: crossover and reflect/retract. Each response was calculated as a percentage of the total number of interactions in each experimental condition, control or S-220 treatment. Data were collated for at least 10 growth cones per condition and pooled from at least 3 separate experiments.

Growth cone turning assays.

Assays were performed as previously described (Murray and Shewan, 2008). Briefly, DRG neurons were cultured in low-walled glass-bottomed 50 mm Petri dishes (MatTek) coated with 5 μg/ml laminin. The culture medium, supplemented with 15 mm HEPES buffer, was overlaid with a thin monolayer of warmed vegetable oil and transferred to a Nikon Diaphot inverted-stage microscope equipped with a stage-mounted incubator heated to 37°C. Growth cones were visualized using a Nikon Diaphot inverted microscope connected to a PC running QWin version 2.1 software (Leica Microsystems). Borosilicate glass micropipettes (Intracel) were pulled in a Sutter pipette puller (Intracel) so that they had a bore width of 1 μm. Growth cones were oriented at 45° and 100 μm from a glass micropipette containing 3–5 μl of either F12 as a control or S-220 treatment. Each reagent was ejected by an air pulse of 3 psi at 2 Hz and 10 ms duration, applied to the pipette by a Picospritzer III (Intracell). Each growth cone was assayed for 30 min, and only growth cones that had advanced at least 10 μm from the original position at time 0 were included. The angle between the position of the growth cone at 30 min, its original position at time 0, and its original direction of growth was calculated using ImageJ software. Data were collated for at least 10 growth cones per condition, pooled from at least 3 separate experiments.

Hydrogel characterization.

Fmoc hydrogel is commercially available (BiogelX) and can also be made using published methods (Alakpa et al., 2016). Hydrogel was prepared at 2.2 mg/ml using ThinCert cell inserts (Greiner Bio-One) and following the manufacturer's instructions. RGD (Arg-Gly-Asp) peptide (ab142698; Abcam) at different concentrations was incorporated into the gel according to the manufacturer's instructions. Rheology experiments (Discovery HR-2, TA Instruments) were performed at different time points during degradation to assess changes in gel stiffness, using Frequency sweep test 0–100 Hz, 1% strain at 37°C. Gels were also weighed at different time points during degradation in PBS. Buffer was completely aspirated, and the insert weight was subtracted from the total weight. The Griess assay was used to quantify nitrite release by primary microglia when cultured with the hydrogel for 6 h. LPS (1 μg/ml; Sigma-Aldrich) was used as a positive activated control. Fluorescein sodium salt (Sigma-Aldrich) was used to assess the release profile since it has similar molecular weight compared with S-220 (MW = 376.27 Da) and can be measured by a fluorescence plate reader (Omega microplate reader; BMG Labtech) at 440–520 nm with 700 ms gain (Perale et al., 2012; Wilems and Sakiyama-Elbert, 2015).

Immunocytochemistry.

Cells and explants were fixed with 4% PFA in PBS for 30 min, followed by incubation with 10% normal goat serum for 1 h, all at room temperature. Cells/explants were then incubated with primary antibodies overnight at 4°C. We used the following primary antibodies: anti-β-tubulin-III (1:1000, mouse; RRID:AB_1844090; Sigma-Aldrich), anti-growth associated protein-43 (GAP-43) (1:500, rabbit; RRID:AB_10622060; Sigma-Aldrich), anti-glial fibrillary acidic protein (GFAP) (1:1000, mouse; RRID:AB_94844; Merck Millipore), anti-ionized calcium binding adaptor molecule-1 (Iba-1) (1:1000; rabbit; RRID:AB_2314666; Wako), anti-inducible nitric oxide synthase (iNOS) (1:100; mouse; RRID:AB_397719; BD Biosciences), and anti-GFP (1:100, chicken; RRID:AB_300798; Abcam). Following three washes with PBS, cells/explants were incubated for 2 h at room temperature with the appropriate secondary antibodies, including goat anti-mouse, anti-rabbit, or anti-chicken antibodies (1:400, RRID:AB_221544, AB_221544, AB_143160, AB_141672, and AB_141359, respectively; Invitrogen). Coverslips were mounted with PBS/glycerol (1:8 ratio) after counterstaining with Hoechst 33342 (2 μg/ml in PBS; Sigma-Aldrich). All primary/secondary antibodies and goat serum were prepared with PBS containing 0.2% Triton X-100 (Sigma-Aldrich) and 0.1% sodium azide (Sigma-Aldrich).

Imaging and quantification for neurite outgrowth.

An Eclipse Ti-E microscope (Nikon) was used to acquire all images. HCA-vision (CSiRO) was used to quantify total neurite length. The plugin Neuron J for ImageJ was used to manually quantify Maximal Neurite Length of dissociated neurons and maximal distance (D_max) of neurite growth from DRG explants (Torres-Espín et al., 2014). Analyses were performed for at least 100 neurons and 10 DRG explants per condition from three different experiments.

Analysis of astrocytes and microglia activation in vitro.

For each coverslip, 10 representative micrographs were taken with a 20× objective. A minimum of 100 cells per condition were analyzed. The activation of the astrocytes was determined by the quantification of GFAP intensity (Sofroniew, 2009). All images were converted into 8-bit files, and the function threshold in ImageJ was applied to select only area with GFAP immunoreactivity. Intensity of staining was determined by the mean optical density value. The mean background fluorescence proximate to each cell was subtracted from the measured fluorescence intensity of the cell area to give a corrected fluorescence intensity. The activation of microglia was quantified by calculating the percentage of iNOS/Iba-1 double-immunoreactive cells in the total Iba-1-immunoreactive cells. Data for same conditions from three different replicates were averaged and compared.

Production of spinal cord slices: ex vivo modeling of SCI.

The ex vivo model using rat spinal cord tissue was adapted from a study using mouse spinal cord tissue (Weightman et al., 2014). The SD pup's (postnatal days 0–3; mixed sexes) dorsal trunk skin surface was sprayed with 95% ethanol; then a dorsal midline incision was made and skin flaps were retracted to expose the spinal column. A midline incision was made along the length of the spine using fine micro-dissecting Vannas spring scissors (Fine Science Tools). The spinal cord was rapidly dissected from the thoracolumbar region and placed in ice-cold slicing media. The meninges were gently removed, and a 0.7–0.9 cm length of cord was cut from the thoracic region. The cord was transferred onto an ice-cold chopping plate of a preset McIlwain tissue chopper and sliced lengthways in the parasagittal plane (350 μm thickness). Three or four slices could be derived from each cord. Slices were incubated in ice for 90 min and then selected under a dissecting microscope and transferred to precut Omnipore membrane “confetti” (Merck Millipore), resting on the Mill cell culture insert membrane (Merck Millipore), using a plastic Pasteur pipette. The process of slice plating was performed under a dissecting microscope to ensure the correct position of the slice. To avoid variability, the selected slices from each pup were plated together in one dish fitting up to four slices. Slices were cultured at the air–medium interface for up to 10 d in a humidified incubator with 5% CO₂ at 37°C with 80% medium changes every other day. The experimental unit, n, refers to slice numbers obtained from different pups.

Lesioning organotypic spinal cord slice cultures.

Cords were lesioned after 3 d in culture. A slice lesioning tool was custom-made using a preassembled, double-bladed scalpel with a spacer creating a gap of ∼700 μm. The tool was aseptically assembled before lesioning by taping together two surgical blades (size 15) with a spacer in the middle secured into a scalpel holder. The same lesioning tool was used in each individual experiment. The lesioning procedure was implemented inside a horizontal laminar flow hood under a dissecting microscope. To stabilize the slice in the process of lesioning, the confetti was cut in a T shape, allowing it to be secured with forceps while producing the lesion. A small lateral movement was used to ensure complete transection of the cord without damaging the confetti. An aspirator fitted with a pipette tip was used to remove all the debris resulting from the lesioning.

Assessment of the viability of spinal cord slices.

To assess viability, three intact slices were randomly chosen before and after the lesion and incubated with the live/dead assay kit (Invitrogen) containing calcein (1 μl/ml) for live cells and ethidium bromide (3 μl/ml) for dead cells, for 15 min at 37°C. Following the incubation period, slices were mounted with antifade mounting media Vectashield (Vector Laboratories). Immediately afterward, fluorescence micrographs were taken (664 μm × 834 μm) with a 10× objective with consistent exposure settings and converted to grayscale. The integrated density (mean gray value per unit area) was measured from both live- and dead-stained micrographs. Thus, the viability for each slice was calculated by expressing the corrected integrated density of the live-stained micrograph as a percentage of the sum total from both stained groups (Cho et al., 2009). The quantification was conducted using ImageJ software.

Treatment of spinal cord slice.

Treatment was administered immediately after the lesion either in the medium or incorporated into the gels. Treatment gels were prepared 2 h before lesioning and precut into 700 μm × 700 μm pieces using the McIlwain tissue chopper. Gels were incorporated into the injury site under the microscope using needles. Treatment was applied for 7 d.

Immunohistochemistry.

All slices were fixed with 4% PFA 7 d after the lesion for 30 min, followed by incubation with 10% normal goat serum for 1 h at room temperature. Slices were then incubated with primary antibodies for 24 h at 4°C. Primary antibodies used were as follows: anti-β-tubulin-III (1:1000, mouse), anti-GFAP (1:500, rabbit; RRID:AB_2109645; Merck Millipore), anti-GFAP (1:1000, mouse), anti-Nestin (1:400, mouse; RRID:AB_2151130; Merck Millipore), and anti-Iba-1 (1:1000; rabbit). Following three washes with PBS, cells were incubated for 4 h at room temperature with appropriate secondary antibodies, including goat anti-mouse or anti-rabbit secondary antibodies. Slices were mounted with Vectashield. All primary/secondary antibodies and goat serum were prepared with PBS containing 0.2% Triton X-100 and 0.1% sodium azide.

Image and axonal outgrowth analysis on spinal cord slices.

All fluorescence images were captured with an Eclipse Ti-E microscope (Nikon). Single panoramic images were generated when appropriate using the Large Image tool in Nis Elements for advanced research software. Maximal Intensity Projection images using multiple Z planes were also used when appropriate. Where applicable, fluorescence images of immunostained slices were merged using Photoshop. Quantitative analysis of β-tubulin-III stained profiles was conducted across the lesion gap from one lesion margin to the other according to previously published methods (Weightman et al., 2014), with 20 division lines at an equal distance of 35 μm between each line using a multiselection plugin from ImageJ. Peak analysis plugin from ImageJ was then used to convert the staining into optical density at each division line covering the whole width of the lesion gap, which is equal to the width of the slice. Finally, the peaks (number of staining profiles) per square millimeter per slice (Weightman et al., 2014) were calculated by averaging the optical densities and then divided by the area of the lesion gap.

Analysis of astrocyte activation in spinal cord slices.

Quantitative analysis of astrocyte activation was conducted by semiquantification of GFAP staining intensity within the lesion gap. Thus, three fluorescence images (478 μm × 598 μm) were taken with a 20× objective from the front areas of the GFAP-immunoreactive glial scar from one-half of the lesion first: the central one aligned with the midline perpendicular to the lesion margin and then one right and one left with equal distance of 200 μm from the central one. We then repeated with another three fluorescence images for the other half of the lesion as described above. Consistent exposures were applied for all captured images, which were then converted to grayscale followed by applying a threshold using the ROI manager plugin from ImageJ. Finally, the intensity of the GFAP staining from the six images per slice was averaged.

Reactivity of astrocytes was also estimated by the overlapping of GFAP and Nestin as previously described (O'Neill et al., 2017). Thus, three fluorescence images (478 μm × 598 μm) were taken with a 20× objective from the front areas of the GFAP-Nestin-immunoreactive glial scar from both halves of the lesion as described above. Briefly, captured images were converted to grayscale followed by the application of the “auto-threshold” function using the “Default” setting in ImageJ. This generates a binary image with a black signal in a white background. “Image Calculator” function in ImageJ using the “AND” operation was used to generate a third image representing the regions of overlap between binary green and red. The number of overlapping pixels was calculated by using the “Histogram” function in ImageJ to obtain the total number of pixels with value 255 (black). A minimum of three images was analyzed per spinal cord slice and averaged to get a final value per slice.

Analysis of microglial activation in spinal cord slices.

Quantitative analysis of microglial activation was conducted by semiquantification of the cell body perimeter of Iba-1-immunoreactive cells within the lesion gap. Thus, three fluorescence images (316 μm × 395 μm) were taken with a 30× objective as described above for astrocyte activation analysis. The perimeter (Kozlowski and Weimer, 2012) of at least 20 Iba-1-immunoreactive cells per micrograph was then measured using ImageJ and averaged for each slide. For nonlesioned microglial activation analysis, three images were taken from at least 1 mm away from the lesion site, where Iba-1-immunoreactive cells displayed nonreactive morphologies.

Treatment of contusion injured spinal cord with S-220-loaded hydrogel.

For all in vivo experiments, young male adult Wistar rats weighing between 180 and 200 g were purchased from Envigo. The skin and muscle overlying the spinal column were incised, and a laminectomy was performed between the T9 and T10 levels. Following fixation of the adjacent T9 and T11 vertebral body to suspend the target region, a standardized moderate (12.5 mm) thoracic spinal contusion was inflicted at T10 thoracic vertebral level by the use of a NYU impactor device (Mascis rat model III; Rutgers University) as previously described (Young, 2002; Kjell and Olson, 2016). Following the injury, muscle layers were sutured and the skin was closed. Postinterventional care included manual voiding of the bladder twice a day until voiding reflex was regained. Three weeks after the induction of the injury, rats were randomly allocated to treatment groups. For rats receiving treatment, a 10 μl Hamilton syringe was loaded with the pregel format of the hydrogel using RN 26S gauge needle (26S/102/pst2/tapY) (Hamilton; Supelco). The Hamilton syringe was mounted on a stereotaxic frame and armed into 45° angle. The dura was opened, and the needle was introduced into the dura with the fine controllers. A maximum of 3 μl was injected at1 μl/min. The needle was left in place for 10 min after injection to prevent the leakage of the pregel solution. The dura was covered with a piece of saline-soaked gel foam, muscle layers were sutured and the skin was closed. Buprenorphine (0.3 mg/kg; Vetergesic; Alstoe Animal Health) was given subcutaneously for the management of acute pain after surgery and saline (0.9% NaCl, 5 ml) was administered subcutaneously to compensate for any blood loss. Locomotor function was assessed by two observers independently using the 21-point Basso, Beattie & Bresnahan (BBB) open field locomotor scale on postoperative days 1, 3, 14, 21, 28, 35, 42, and 49.

Statistical analysis.

Sigma-Aldrich Plot 13.0 and Origin Pro 2018b software was used for all statistical analyses. The normality test Shapiro–Wilk was performed to ensure normal distribution of the data and suitability of parametric tests. Equal variance was tested using the Brown–Forsythe test. Statistical analyses were performed using unpaired Student's t test and one-way ANOVA, using Bonferroni's multiple-comparison test for post hoc analysis. If normality was not met, nonparametric tests were used: the Mann–Whitney rank sum test and the Kruskal–Wallis one-way ANOVA on ranks with Dunn's method or Tukey test post hoc correction. For in vitro studies, all data represent mean ± SEM.

For ex vivo studies, between 4 and 6 replicates were done per condition and experiment and are represented by a circle in the quantification graph. All values quoted are expressed as the mean ± SEM (box limits) ± 5%–95% (error bars). Two-way repeated-measure ANOVA was performed for locomotor recovery (BBB scale) analysis, followed by Bonferroni's post hoc multiple-comparison adjustment to calculate the significant levels. “Treatment” is a between-subjects variable, whereas “time” is a within-subjects variable. The in vivo data were presented as mean ± SEM. All statistical analyses were performed using Origin Pro 2018b and SigmaPlot 14.0 software.