Experimental design
Female Wistar rats (Rattus novergicus) with weights in the range of 200–300 g were used for the experimental procedures. The rats were maintained under controlled humidity, temperature, and constant light/dark cycles. All procedures were performed in accordance with the ethical principles set forth by the National Council of Animal Experimentation (CONCEA) and with the approval of the Ethics Committee in Animal Experimentation of São Paulo State University (CEUA/FMB, UNESP, protocol no. 1243–2017). The animals were divided into four groups. In the sham group (n = 5), the sciatic nerve was surgically exposed without any changes. The proximal and distal segments were resected, forming a gap of 12 mm, and sutured with perineural stitches in the animals of the autograft group (n = 5). In the PCL group, a gap of 12 mm was formed with nerve resection, and an NGC empty was fixed (n = 5). In the PCL+MSC group, a gap of 12 mm was formed, and the NGC was fixed and multi-functionalized with AdMSCs embedded in HFB (n = 5). The sciatic functional index (SFI) and tibial functional index (TFI) were evaluated in vivo for 12 weeks after injury. Gait analysis was evaluated using the Catwalk system, and nerve conduction velocity (NCV) was measured at 8 and 12 weeks. Morphometric analysis was performed 8 and 12 weeks post-injury. To evaluate the production of neurotrophic factors at 4 weeks, brain-derived neurotrophic factor (BDNF), glial cell line-derived neurotrophic factor, hepatocyte growth factor (HGF), and the cytokines interleukin – 6 (IL-6) and interleukin – 10 (IL-10) in the spinal cord, real-time PCR (RT-qPCR) was performed nerve in both sham and PCL+MSC groups (n = 3). In addition, immunohistochemical analysis of the sciatic nerve for BDNF, GDNF, p75 neurotrophin receptor (p75NTR), S-100 and neurofilament were performed in both sham and PCL+MSC groups (n = 3).
Isolation, differentiation, and characterization of canine AdMSCs
Subcutaneous canine adipose tissue was obtained from healthy young female dogs undergoing elective surgery in accordance with a previously published protocol [47]. Adipose tissue was digested in 0.04% type 1A collagenase (1 mg/mL, Thermo Fisher Scientific, São Paulo, Brazil) for 1 h at 37°C with gentle shaking. Digested tissue was blocked, centrifuged, and filtered (BD Falcon cell strainer, 70 µm, San Jose, CA, USA). Canine AdMSCs were isolated based on their inherent property of plastic adherence in culture media containing 90% Dulbecco’s modified Eagle’s medium (DMEM), 10% fetal bovine serum (FBS), and 1% penicillin/streptomycin (100 U/mL) (all from Gibco, Grand Island, NY, USA). Cellular expansion was continued until the third passage, and the cells were cryopreserved to induce differentiation, for immunophenotypic analysis and transplantation later on.
Canine AdMSCs were tested for their ability to differentiate into of adipocyte, osteoblast, and chondrocyte lineages. Differentiation was induced in cells that underwent third passage using StemPro adipogenesis, chondrogenesis, and osteogenesis differentiation kits (Gibco, Grand Island, NY, USA) following the manufacturer’s recommendations. The cells were fixed in paraformaldehyde (4%, pH 7.34) 2 weeks after stimulation, and the evaluation of osteogenic and adipogenic differentiation were performed using histological stains, namely, Alizarin red (2%, pH 4.2) and Oil red (0.5% in isopropanol) (Sigma-Aldrich, Saint Louis, MA, USA), respectively. Three weeks after chondrogenic differentiation, the cells cultured as a micromass were fixed in 10% formalin, embedded in paraffin, and stained with hematoxylin-eosin. Samples were analyzed and photographed under an inverted light microscope using LAS 4.0 software (DM IRB; Leica Microsystems, Wetzlar, Germany).
Canine AdMSCs were characterized by the presence of the surface marker CD90 or absence of surface markers CD45, CD34, and CD71 [48,49]. The concentration of cells in the third passage was counted and adjusted to 1x105 cells. Subsequently, the cells were incubated with primary antibody conjugates CD90-PerCP (BD Pharmigen™, San Diego, CA, USA), CD71-FITC (BD Pharmigen™), CD45-PE (BD Pharmigen™), and CD34-FITC (BD Pharmigen™). Antibodies were incubated for 30 min at room temperature. Cells were then washed with phosphate-buffered saline (PBS) and FACSCalibur® 4-color cytometer (Becton Dickinson Company, San Jose, CA, USA) was used to acquire and analyze the samples, standardizing a total of 2x104 events collected per tube. Cells incubated without primary antibodies were used as controls to distinguish non-specific fluorescence. The gate on canine AdMSCs population was based on the parameters of size (forward scatter) versus cell granularity (side scatter), following the phenotypic characterization. The analyses were performed using CellQuestPro® and FlowJo® software.
Stimulation of Canine Ad-MCSs with interferon-gamma
Canine AdMSCs were activated via direct stimulation to evaluate the properties of neurotrophic and anti-inflammatory molecules using a recombinant inflammatory mediator relevant to nerve injury, following a previously described protocol with minor modifications [50]. Cells were stimulated with canine interferon-gamma (IFN-ã) in the third passage. Triplicates were obtained with 2 × 105 cells /cm2 per well in a 24-well plate (Costar®, TC-treated, Corning, NY, USA). Subsequently, cells were stimulated with 0.75 mL basal medium containing IFN-ã (50 ng/mL, IFN-ã canine recombinant; Kingfisher Biotech, Saint Paul, USA) for 96 h. At this point, the cells were collected using TRIzol reagent (Invitrogen, São Paulo, Brazil) and stored at -80°C for RNA extraction and analysis of gene expression. For the control, cells were cultured in basal culture medium containing DMEM and 10% FBS (all from Gibco).
Gene expression of neurotrophic factors (BDNF, GDNF, and HGF) and anti-inflammatory molecules (IL-10) was quantified. Cells were lysed and homogenized with TRIzol reagent, and RNA extraction was performed using the Mini RNAeasy kit (Qiagen, São Paulo, Brazil). RNA was eluted with RNA-free water and quantified and analyzed by spectrophotometry using a NanoDrop 2000 spectrophotometer (Thermo Fischer Scientific, Wilmington, USA) for the absorbance ratios 260/280 nm and 260/230 nm. Total RNA extracted from the cells was of high quality and purity, indicating that the extraction method was efficient. cDNA was synthesized using a High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems™, Thermo Fischer Scientific, Carlsbad, USA), followed by amplification using a Veriti 96 Well Thermal Cycler (Applied Biosystems™, Thermo Fischer Scientific). The cDNA samples were cryopreserved and used as templates for PCR reactions.
The reactions were performed in triplicate, using the cDNA produced in previous steps as a template, with a PowerUp SYBR Green Master Mix (Applied Biosystems™, Thermo Fischer Scientific Baltics, Vilnius, Lithuania), RNA-free water, and canine primers (Thermo Fisher Scientific, São Paulo, Brazil) (Additional file 1: Table S1). The samples were tested with two reference genes, glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and hypoxanthine phosphoribosyltransferase (HPRT). The qPCR reaction was performed using the QuantStudio™ 12K Flex Real-Time PCR System thermocycler (Applied Biosystems, Life Technologies, Carlsbad, CA, USA) with the following parameters: 50°C for 2 min, 95°C for 2 min, and 45 cycles of 95°C for 1 s and 60°C for 30 min. Relative quantification of expression of the genes of interest was performed using the ΔΔCt method [51].
Fabrication and assembly of NGCs
The NGCs were assembled from 3D-printed PCL membranes. The membranes fabrication was based on a material extrusion process called fused filament fabrication (FFF) using FAB@CTI (Renato Archer Information Technology Center - CTI, São Paulo, Brazil) an experimental 3D printing platform [52]. Previously, the filament extrusion head was adapted to different diameters and melting temperatures, which allowed the molding of a thermoplastic polymer via an orifice (open-ended die) [53]. Previous studies have evaluated the interactions between MSCs and 3D-printed PCL matrices [54]. The printing parameters were defined using FAB@CTI software (Renato Archer Information Technology Center - CTI, São Paulo, Brazil). The following parameters were set: jog speed 2,400 Hz, deposition rate 0.07, path speed 8.8 mm/s, path width 0.3 mm, path height 0.3 mm, and temperature of 80°C. The 3D-printed membranes were sputter-coated with gold (MED 010; Balterz Union) and visualized using a scanning electron microscope (ESEM Quanta 200; Fei Company, Oregon, USA). The geometric parameters were evaluated using image analysis software (ImageJ, National Institute of Health, Bethesda). During the assembly of NGCs, the membranes were wrapped around a 1.5-mm support and sealed with controlled heating. The NGCs were sterilized by washing with a 70% ethanol solution for 10 s, followed by washing with distilled water. After drying at room temperature, the NGCs were subjected to UV irradiation (200–280 nm) for 2 h.
Heterologous Fibrin Biopolymer (HFB) scaffold
The HFB was kindly supplied in sufficient quantity for this study by the Center for the Study of Venoms and Venomous Animals at São Paulo State University, Brazil. The components and formula of the applied HFB are contained in its patents (registry number: BR1020140114327 and BR1020140114360). The product is distributed in three vials, stored at -20 °C, and must be mixed and applied immediately at the site of interest [39-44].
Experimental injury and repair with NGCs
Sciatic nerve experimental injury was induced in rats under the influence of anesthesia containing isoflurane (Isoforine®; Cristalia, São Paulo, Brazil) using a microsurgical microscope (DF Vasconcelos, São Paulo, Brazil). The experimental lesion consisted of a gap of 12 mm, which was considered to be above the experimental critical level in rats [55]. In the sham group, the nerves were exposed without any modifications. In the autograft group, the proximal and distal segments were resected, inducing a gap of 12 mm and suturing with perineural stitches (9/0 nylon; Shalon, Brazil). In the PCL group, nerve stumps were introduced and fixed 1 mm into the NGC (9/0 nylon; Shalon). In the PCL+MSC group, nerve stumps were introduced and fixed 1 mm into the NGC (9/0 nylon; Shalon).
Thereafter, NGCs were multi-functionalized with 1 × 106 canine AdMSCs embedded in HFB that has been previously tested as a cell scaffold [40, 41, 56]. Fibrin polymerizes rapidly following the mixing of three components, namely, cryoprecipitated from water buffalos (Bubalus bubalis) blood, calcium chloride, and thrombin-like protein purified from South American rattlesnake (Crotalus durissus terrificus) [39, 41]. First, 106 AdMSCs were mixed with 25 µL of cryoprecipitated. The nerve guidance conduit was loaded slowly and homogeneously with a cryoprecipitated + AdMSC solution using a microsyringe (50 µL, 22s-gauge, point style 2; Hamilton, Nevada, USA). Subsequently, a solution of 12.5 µL of calcium chloride and 12.5 µL thrombin-like was administered, resulting in a final suspension with a volume of 50 µL. This process allowed the formation of a homogeneous cell/fibrinogen suspension into the NGC at the first step, which was coagulated after contact with thrombin + CaCl2 within the NGC. Following surgical procedures, the musculature was co-opted in layers. Rats were administered tramadol intraoperatively (20 mg/kg/SC) and in the postoperative periods (2.5 mg/day in water for 5 days).
Sciatic and tibial nerve functional indices
Functional indices were evaluated preoperatively and weekly during the 12-week observation period in the sham, autograft, PCL, and PCL+MSC groups. The plantar surface of the hind limbs was moistened with black ink. The rats walked with a standard walk trace on a sheet of white paper where the footprints were recorded. Subsequently, the distance between the third toe and the hind limb pads (print length, PL), the first and the fifth toes (toe spread, TS), and the second and fourth toes (intermediary toe spread, ITS) were measured. These parameters were evaluated with the right (lesioned) and left (unlesioned) hind limbs, and the values were calculated using the following formulas described by Bain et al., 1989 [57]: sciatic functional index: -38.3 ([EPL-NPL]/NPL) + 109.5 ([ETS-NTS]/NTS) + 13.3 ([EIT-NIT]/NIT) - 8.8 (30, 31); IFP = 174.9 (EPL˗NPL/NPL) + 80.3 (ETS˗NTS/NTS) - 13.4; tibial functional index: -37.2 ([EPL-NPL]/NPL) + 104.4 ([ETS-NTS]/NTS) + 45.6 ([EIT- NIT]/NIT) - 8.8. Sciatic and tibial functional indices equal to -100 indicated total impairment of the sciatic and posterior tibial nerves, whereas values oscillating around 0 reflected a normal function of the three nerves. The mean ± standard deviation was calculated with three gait cycles for each experimental group each week.
Gait analysis
Functional locomotor recovery was evaluated using the CatWalk System (Noldus, Wageningen, Netherlands). Catwalk analysis was performed preoperatively and after 8 and 12 weeks in sham, autograft, PCL, and PCL+MSC groups. The CatWalk walkway consisted of a glass roof (100 × 15 × 0.6 cm). Rats were placed on the CatWalk walkway and allowed to walk freely. The LED light emitted from an encased fluorescent lamp was reflected along the glass plate, thereby intensifying the areas on which the front limbs and hind limbs were in contact with the glass plate. The contact areas were captured by a high-speed video camera positioned underneath the glass plate connected to a computer running Catwalk software v10.5 (Noldus). The camera was calibrated, and the signals were digitized, frame-by-frame, using the PCImage-SG video card (Matrix vision GmH, Oppenheimer, Germany) and sent to the matrix for classification. Three runs were performed and classified from each animal and the parameters were obtained for each animal at each time point. The following parameters were recorded: maximum contact area (ipsilateral (left)/contralateral (right) ratio), maximum contact intensity (ipsilateral (left)/contralateral (right) ratio), swing speed (ipsilateral (left)/contralateral (right) ratio), and swing (seconds) (swing exercised by the limbs when they are not in contact with the glass plate) and stand time (seconds).
Nerve conduction velocity
Nerve conduction velocity (NCV) was calculated preoperatively and after 8 and 12 weeks in the sham, autograft, PCL, and PCL+MSC groups, according to a previously published protocol [58,59]. Under anesthesia, the sciatic nerve was stimulated with single electrical pulses (200-µs duration) and supramaximal stimulation that ensured maximal amplitude. Using needle electrodes, the sciatic nerve was percutaneously stimulated proximal to the lesion site at the level of the sciatic notch and distal to the lesion at the level of the ankle. Compound muscle action potentials (cMAP) of the plantar muscles were recorded using monopolar needles inserted into the muscle bellies and displayed with an oscilloscope (Sapphire II 4ME; Teca medelec, USA). Motor NCV was calculated by dividing the distance between stimulation sites by the average latency evoked from two sites (sciatic notch and ankle). The mean ± standard deviation was calculated for each experimental group and at each evaluated time point.
Specimen preparation and morphometric analysis
Nerves were harvested after 8 and 12 weeks from the sham, autograft, PCL, and PCL+MSC groups. Under general anesthesia with isofluorane (Isoforine®; Cristalia, Brazil), rats were euthanized with barbiturate overdose (Thiopentax, Cristalia, São Paulo, Brazil). The vascular system was rinsed by transcardial perfusion with phosphate-buffered saline (PBS; 0.1 M, pH 7.4). Fixation was performed in 2% glutaraldehyde and 1% paraformaldehyde in PBS (0.2 M, pH 7.34), and nerves containing NGC were immersed in the same solution for 24h at 4°C. The sciatic nerve segment into the NGC was dissected and divided into two parts: proximal and distal. Nerves were washed with PBS (0.1 M, pH 7.4) and post-fixed for 3 h in 1% osmium tetroxide solution mixed with PB (pH 7.4). The specimens were dehydrated and embedded in glycol methacrylate resin (Leica Microsystems, Heidelberg, Germany). The blocks were trimmed, and semi-thin sections (1–2 µm) were obtained with an ultramicrotome (Leica RM 2265; Leica Microsystems CMS), which were stained with toluidine blue (0.25%). Morphometric analysis was performed by sampling at least 30% of the cross-section of each nerve using a bright-field microscope (Leica DM 4000 B-M; Leica Microsystems CMS) [60]. The analysis was performed with two sampled fields from each nerve (magnification of 100×) using Adobe Photoshop CC 2019. Morphometric parameters evaluated included myelinated axon diameter, myelinated fiber diameter, myelin thickness (fiber diameter ˗ axon diameter/2), and "g" ratio (axon diameter/fiber diameter). The mean ± standard deviation was calculated for each experimental group and at each evaluated time point.
Immunohistochemical study of sciatic nerve and RT-qPCR analyses of spinal cord samples
Immunohistochemical analysis (S-100, neurofilament, BDNF, GDNF, and p75NTR) of sciatic nerve samples and qPCR of the spinal cord samples (BDNF, GDNF, HGF, IL-6, and IL-10) were performed for the sham and PCL+MSC (n = 3) groups after 4 weeks. Rats were euthanized with barbiturate overdose (Thiopentax; Cristália). Fresh spinal cord tissue ipsilateral to the lesion was harvested, frozen in liquid nitrogen, and stored at -80°C. Fixation was performed in 4% paraformaldehyde in PB (0.1 M, pH 7.34), and the regenerated nerve was dissected and immersed in the same solution for 12 h at 4°C. Specimens were immersed in ascending order 10%, 20%, and 30% of the sucrose solutions (0.1 M PB, pH 7.4) for 12 h, mixed with Tissue-Tek OCT (Sakura Finetek, Torrance, USA), and frozen at -80°C.
Longitudinal cryostat sections (12 µm) of the sciatic nerves were acclimatized, washed, and incubated in 3% bovine serum albumin solution or 3% donkey serum in PB (0.1 M, pH 7.4) for 1 h, followed by incubation in a moist chamber with primary antibodies against S100, neurofilament H (NF), BDNF, GDNF, and p75NTR for 4 h (Additional file 2: Table S2). After rinsing with PB, the sections were incubated with Alexa Fluor 488, Alexa Fluor 546, or CY2-conjugated secondary antiserum for 45 min at room temperature. The sections were then mounted in a mixture of glycerol/PB (3:1) for quantitative measurements or glycerol/DAPI for qualitative analysis. Representative images were obtained using a fluorescence microscope (BX51; Olympus Corporation, Tokyo, Japan) equipped with a camera (DP 72; Olympus Corporation). Four images of each sample were imported for the determination of the integrated pixel density that represented the intensity of labeling using ImageJ software (version 1.33u, National Institutes of Health, USA), according to a previously published protocol [61,62]. The mean intensity ± standard deviation was calculated for each group.
For RT-qPCR, the spinal cord was finely pricked and homogenized in TRIzol reagent (TRIzol™, Invitrogen™) and chloroform. The samples were vigorously shaken for 30 s using a Precellys Lysing Kit® (Uniscience, São Paulo, Brazil) with a Precellys 24 tissue homogenizer (Bertin Technologies SAS, Montigny-le-Bretonneuz, France). Total RNA was extracted, quantified, and reverse-transcribed to cDNA, which was amplified as described previously in the RT-qPCR assay procedure performed with cells. Assays analyzing the levels of BDNF, GDNF, HGF, IL-10, and IL-6 were performed (all from Thermo Fisher Scientific, São Paulo, Brazil) (Additional file 3: Table S3). Samples were tested with two reference genes, β2-microglobulin and HPRT. The qPCR reaction was performed using the QuantStudio ™ 12K Flex Real-Time PCR System thermocycler (Applied Biosystems™, Thermo Fischer Scientific) with the following parameters: 50°C for 2 min, 95°C for 2 min, and 45 cycles of 95°C for 1 s and 60°C for 30 min. Relative quantification of expression of the genes of interest was performed using the ΔΔCt method [51].
Statistical analysis
Variables, namely, sciatic and tibial functional indices, Catwalk analysis, and NCV were assessed for normality with statistical tests (Shapiro–Wilk or Kolmogorov–Smirnov), descriptive statistics, and graphic analyses (QQ plot). An analysis of variance test (two-way ANOVA, multiple comparisons) was performed followed by Tukey’s test to verify the differences in the means of the variables between each group and the time of the experiment. Other variables (integrated pixel density and relative quantification) were assessed for normality using statistical tests (Shapiro-Wilk), descriptive statistics, and graphic analysis. For parametric data, the t-test was performed with unpaired samples. For non-parametric data, the Mann-Whitney test was performed for unpaired samples. The level of significance between the groups was set at p < 0.05. The differences were denoted by a single asterisk (p < 0.033), two asterisks (p < 0.002), or three asterisks (p < 0.001) (GraphPad Prism version 8 for Mac, San Diego, CA, USA).