The present study was conducted using adult female Sprague-Dawley (SD) rats (n=108; weighing 200–250 g). All rats used in this study were purchased from the Animal Experiment Center of Hubei University of Medicine (Wuhan, China). The study protocol was approved by the Ethics Committee on Animal Experiments of Hubei University of Medicine (Shiyan, China; protocol no. SYXK20 110008). The rats were housed 3 per cage for approximately 1 week prior to the start of the experiments, with free access to food and water and maintained in a suitable environment at 21°C, 60% air humidity and a 12-h light/dark cycle.

Following a baseline behavioral assessment, a model of SCI was established in rats using the modified Allen's test, as previously described (33). The rats were anesthetized with an intraperitoneal injection of 10% chloral hydrate (0.3 ml/100 g) following 12 h of pre-operative fasting from food and water. Under sterile conditions, a dorsal laminectomy was performed, centering on the T9 spinous process, and the spinal cord was then exposed. Subsequently, each rat sustained a contusive SCI from a Spinal Cord Impactor (W.M. Keck Center for Collaborative Neuroscience, Rutgers, Piscataway, NJ, USA), with a 10 g impact rod vertically dropped from a height of 5 cm which then impinged on the spinal cord in a circular zone with a 2 mm diameter. The dwell time for each injury was 10 sec which was sufficient to cause a moderate contusion. The release weight, height of drop, and velocity of each SCI were determined using the Spinal Cord Impactor software (version 7.5). The signs of the successful infliction of SCI were as follows: the rat tails swung spastically, there was a retraction of the lower limbs and a torso-like flutter, and both lower extremities exhibited flaccid paralysis. In the sham-operated group, the spinal cord was exposed in the same manner, but without subjecting the rats to a contusive SCI.

The BMSCs were isolated as previously described (34). Briefly, another 10 SD rats (weighing 120±20 g) were sacrificed following anesthesia with an intraperitoneal injection of 10% chloral hydrate (0.3 ml/100 g). Under aseptic conditions, the BMSCs were separated from the tibias and femurs and all conterminal tissues were carefully removed. The extracted BMSCs were centrifuged at 1,000 × g for 5 min and then aseptically plated in Dulbecco's modified Eagle's medium (DMEM; HyClone, Logan, UT, USA) supplemented with 10% fetal bovine serum (FBS; Invitrogen, Carlsbad, CA, USA), 2 ml glutamine and 100 U/ml penicillin. The cells were incubated at 37°C, in 5% CO₂ for 24 h and the culture medium was then replaced. This procedure was performed every 72 h, through which the other non-adherent cells were eliminated. The remaining BMSCs were examined everyday with an inverted phase contrast microscope (Eclipse Ti-E/U/S; Nikon, Tokyo, Japan) to verify that the cells were further amplified. The BMSCs were subcultured at 1:3 every 4 days when they reached approximately 80–90% confluence. Those used for transplantation were subjected to trypsin preconditioning and washed with phosphate-buffered saline (PBS). To trace the internal migration of the BMSCs toward the lesion site following SCI, transfection was carried out using a recombinant adenovirus encoding green fluorescent protein (Ad-GFP) plasmid with Lipofectamine 2000 (Cyagen Biosciences Inc., Santa Clara, CA, USA), in accordance with the manufacturer's instructions. At 6 h after transfection, the medium was exchanged with all nutrition culture medium and transgene expression was examined 2 days later using a fluorescence microscope (Eclipse Ti-E/U/S; Nikon).

A total of 108 rats were randomly divided into 6 groups (n=18/group) as follows: i) the sham-operated group; ii) the model control group; iii) the EPO group; iv) the BMSC group; v) the BMSC + EPO group; and vi) the BMSC + EPO + AMD3100 group. Each animal in the experimental groups was subjected to SCI as described above. In the EPO group, recombinant human EPO (rhEPO; Beijing Four Rings Bio-Pharmaceutical Co., Ltd., Beijing, China) was administered at a distance of 2 mm cranially and then 2 mm caudally from the site of injury (5×10³ IU/kg) using a prefilled syringe. In the BMSC group, 10 μl of BMSC suspension containing approximately 3×10⁴ cells was administered via a microsyringe (Hamilton, Reno, NV, USA). In the BMSC + EPO group, BMSCs and EPO were administered as described above. In the BMSC + EPO + AMD3100 group, in addition to the injection of BMSCs and EPO, AMD3100 (a chemokine receptor antagonist; Pfizer, Inc., New York, NY, USA) was administered in the same manner (5 mg/kg) through a microsyringe (Hamilton). The AMD3100 dosage was selected in accordance with a previous study (35). In the model control and sham-operated groups, 10 μl of PBS was injected. Following withdrawal of the microsyringe, the muscles and the skin were stitched in respective layers.

All the rats were administered penicillin (3×10⁴ U/kg) and an analgesic, sufentanil (0.05 μg/kg), for the first 3 days. The urinary bladders were manually emptied by squeezing twice a day until the auto-urination function was restored. Food and water were supplied in slender drinking tubes by placing them at the bottom of the cage, until the rats were able to lift their heads to reach the food placed at a normal height/position in the cage. Rats diagnosed with bacterial infections during the experiment were treated immediately with cefazolin (10 mg/kg/day; Sandoz International GmbH, Holzkirchen, Germany).

Dual hind limb motor function was evaluated at 1, 3, 7, 14, 21 and 28 days post-SCI using the Basso-Beattie-Bresnahan (BBB) Locomotor Rating Scale developed by Basso et al (36). A 120×30 cm foam box with a smooth bottom was manufactured for the assessment. The rats were placed in the box 2 days pre-operatively, twice a day in order to familiarize them with the environment. The dual hind limb motor function was quantified using a scale ranging from 0 to 21, where 0 corresponds to no locomotor activity and 21 corresponds to normal performance. A sensorimotor grid walk test was also conducted to estimate the capacity of the rats to precisely control hind paw placement, as previously described (37). The rats crawled freely on a 120×120 cm grid with 1.0 cm regularly spaced horizontal holes. Slips were recorded when the animal misplaced the paw down through the hole in the grid, until a maximum value of 20 missteps were reached. The rats that could not perform with at least proficient gait in the BBB test were unable to finish the walkway and were classified according to the maximum value. To decrease the average error and improve the accuracy, all evaluations were performed using a double blind technique: two individuals observed from different sides and recorded the scores independently, and the mean values were then acquired.

Thoracic spinal cord segments (approximately 1-cm-thick) centered on the site of injury were serially collected and then immediately homogenized at 1, 7 and 28 days post-operation. The samples were centrifuged at 1,800 × g at 4°C for 10 min, and subsequently, the detached spinal cord samples were stored at −80°C for further research. The levels of cytokines, including tumor necrosis factor-α (TNF-α) and SDF-1 levels were measured using DuoSet ELISA kits (Elabscience Biotechnology Co., Ltd., Shanghai, China) in accordance with the manufacturer's instructions. All analyses were carried out in duplicate, using the proposed substrates, buffers and diluents. Each experiment was repeated 4 times. The final outcomes were pooled as the average concentration of cytokines.

To identify the distribution of the BMSCs in the injured spinal cord, an immunofluorescence assay was performed 2 weeks following surgery. On days 1, 7 and 28 following treatment, 6 animals from each group were randomly selected and were sacrificed by anesthesia. The injured spinal cord tissues were carefully extracted. At each time point, 4 specimens from each group were fixed with 4% paraformaldehyde for 24 h and then stored at −80°C in a constant temperature refrigerator for biological detection. Two specimens were frozen and embedded in OCT (Leica Biosystems, Shanghai, China), and then cut into 7-μm-thick longitudinal cryostat sections. To reduce non-specific staining, the sections were treated with diluted normal goat serum (Boster Bio-Engineering, Wuhan, China) at room temperature for 30 min. To stain the nuclei, 4′,6-diamidino-2-phenylindole (DAPI) was added and followed by incubation in the dark for 5 min. The excess DAPI was washed 4 times for 5 min with Tween-20 in PBS (PBST; Beyotime Institue of Biotechnology, Beijing, China) following incubation. The slices were then dried with absorbent paper and mounted with anti-fluorescent quencher (SouthernBiotech, Wuhan, China). Immunofluorescence was observed under a fluorescence microscope (Olympus BX51; Olympus, Tokyo, Japan).

Cellular migration was detected using Transwell chambers, which were 6.5 mm in diameter with 8 μm nitrocellulose pore filters (Amersham Biosciences, Piscataway, NJ, USA). In the chemotaxis group, 200 μl of serum-free culture medium containing 1×10⁵ BMSCs were added to the upper chambers, and 800 μl of DMEM containing 10 U/ml of rhEPO (Beijing Four Rings Bio-Pharmaceutical Co., Ltd.) and 10% FBS were added to the lower chambers. In the chemotaxis inhibition group, 200 μl of serum-free culture medium containing 1×10⁵ BMSCs, which had been co-cultured with 10 μg/ml AMD3100 (Pfizer Compounds) at 37°C for 2 h, were added to the upper chambers, and 800 μl of DMEM containing 10 U/ml of rhEPO and 10% FBS were added to the lower chambers. In the control group, 200 μl of serum-free culture medium containing 1×10⁵ BMSCs were added to the upper chambers, and 800 μl of DMEM were added to the lower chambers. Following incubation for 18 h, the non-migrated cells in the upper chamber were cleared and the membranes were fixed with 4% paraformaldehyde for 30 min. The migrated cells were stained with 5% crystal violet dye solution (Saichuang Technology, Wuhan, China) for 20 min, washed with PBS, and then photographed under a microscope (Eclipse Ti-E/U/S; Nikon).

The apoptotic index (AI) of the lesion area was detected by TUNEL assay using an In Situ Cell Death Detection kit (Roche Applied Science, Basel, Switzerland) according to the manufacturer's instructions. The sections were dewaxed with graded alcohols and then digested with protease K (20 μg/ml, pH 7.4) at 25°C for 30 min. The sections were then incubated with TUNEL reaction mixture, which contained 5 μl of enzyme solution and 45 μl of fluorochrome-labeled solution at 37°C without light for 60 min. After being washed with PBS 3 times, the sections were incubated with DAPI (Beyotime Institue of Biotechnology) for 30 min. Subsequently, the sections were dried with absorbent paper and mounted with anti-fluorescent quencher liquid (Southern Biotech, Birmingham, AL, USA) and then observed under a fluorescence microscope (Eclipse Ti-E/U/S; Nikon). Apoptotic nuclei with green fluorescence and DAPI-positive nuclei with blue fluorescence were observed at ×400 magnification. The AI was calculated as the percentage of TUNEL-positive cells among the total number of nucleated cells.

The spinal cord samples were homogenized using radioimmunoprecipitation lysis buffer (RIPA; Beyotime Institue of Biotechnology) and then centrifuged at 18,000 × g for 15 min. The supernatant containing 50 μg total protein was extracted for electrophoresis on a 10% sodium dodecyl sulfate gel (SDS; Beyotime Institute of Biotechnology) and then transferred onto polyvinylidene fluoride membranes (PVDF; Millipore, Billerica, MA, USA). After being blocked with Tris-buffered saline with Tween-20 (TBST; Biosciences, Shanghai, China), the PVDF membranes were incubated at 4°C overnight with anti-CXCR4 (P61073) or anti-EPO antibody (sc-80995; 1:200; Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA) or rabbit anti-GAPDH antibody (A300-639A; 1:200; Xianzhi Biological Co. Ltd., Hangzhou, China) as primary antibodies, followed by incubation with horseradish peroxidase-labeled goat anti-rabbit secondary antibody (A00098; 1:50,000; Santa Cruz Biotechnology, Inc.) at 26°C for 2 h. The immunoreactive complexes were visualized using an ECL enhanced detection kit (Thermo Fisher Scientific, Waltham, MA, USA) and exposed to X-ray films. The densitometry of the bands was analyzed using Image-Pro Plus 6.0 software (Media Cybernetics, Rockville, MD, USA). The expression of CXCR4 and EPO was then normalized to GAPDH.

All data are expressed as the means ± standard deviation (SD) and compared by one way analysis of variance (ANOVA). All statistical analyses were conducted using Statistical Product and Service Solutions (SPSS) version 19.0 software (IBM Corp., Armonk, NY, USA), with a value of P<0.05 considered to indicate a statistically significant difference.

The levels of SDF-1 in the spinal cord increased significantly following SCI (Fig. 1A). The SDF-1 levels in the BMSC + EPO group were significantly higher compared with those in the sham-operated, model control, EPO, BMSC and BMSC + EPO + AMD3100 groups (P<0.05). However, no significant difference was observed between the model control and BMSC + EPO + AMD3100 groups. On days 1, 7 and 28 post-SCI, the TNF-α levels (Fig. 1B) in the BMSC + EPO group were significantly lower than those in the sham-operated, model control, EPO, BMSC and the BMSC + EPO + AMD3100 groups (P<0.05). However, the difference was not significant between the model control and BMSC + EPO + AMD3100 groups (P>0.05).

Dual hind limb loco-motor function was evaluated using two complementary behavioural tests. The BBB motor rating scale was adopted to evaluate the locomotor capacity of the animals in each group up to 28 days post-operation (Fig. 2). Due to the effect of the anesthesia, no significant difference was observed between the rats subjected to SCI on day 1 post-operation. The mean BBB scores in the sham-operated group declined slightly at the beginning, but returned to normal values by day 7. Over time, the rats in the BMSC + EPO group obtained significantly higher BBB scores than those of the rats in the model control and BMSC + EPO + AMD3100 group (P<0.001) and those of the rats in the EPO and BMSC group (P<0.05). However, no significant differences were observed between the model control and BMSC + EPO + AMD3100 groups (P>0.05). The grid walk test of the hind limbs was applied to estimate locomotor coordination and motor accuracy (Fig. 3). The animals administered the BMSCs in conjunction with EPO exhibited progressive improvement in hind limb control and paw placement accuracy in the grid walk test. Moreover, the BMSC + EPO-treated rats experienced significantly fewer hind limb slips from day 7 than the rats in the model control and BMSC + EPO + AMD3100 groups (P<0.001) and the rats in the EPO and BMSC group (P<0.05); however, no significant differences were observed in these scores between the model control and the BMSC + EPO + AMD3100 groups (P>0.05).

The transplanted cells were detected in the injured spinal cords 2 weeks following SCI (Fig. 4). GFP-labeled BMSCs were observed and they were located at the lesion site in the BMSC and BMSC + EPO groups. However, homologous cells were almost undetectable in the sham-operated, model control, EPO and BMSC + EPO + AMD3100 groups.

Our results (Fig. 5) revealed that EPO significantly promoted the migration of BMSCs when compared with the control group (no EPO; P<0.001); however, the EPO-induced cell migration was markedly inhibited when the BMSCs were preconditioned with AMD3100 (P<0.001). No significant differences were observed between the AMD3100 group and the control group (P>0.05).

The AIs at different time points following SCI in each group are presented in Figs. 6 and 7. The quantity of TUNEL-positive cells on day 1 following SCI showed no significant difference between the model control, EPO, BMSC, BMSC + EPO and BMSC + EPO + AMD3100 groups. On days 7 and 28 post-SCI, the AI in the BMSC group was significantly lower than that in the model control group (P<0.05) and the AI in the BMSC + EPO group was significantly lower than that in the model control, EPO, BMSC and BMSC + EPO + AMD3100 groups (P<0.05); however, the difference between the model control and BMSC + EPO + AMD3100 groups was not statistically significant (P>0.05). These data indicated that EPO enhanced the anti-apoptotic effect of the BMSCs, which was attenuated by AMD3100.

The protein expression levels of EPOR and CXCR4 were measured by western blot analysis (Fig. 8). The protein expression of EPOR increased with the administration of EPO and then gradually decreased from day 7 to 28 (Fig. 8A). No significant differences were observed in EPOR protein expression between the sham-operated, model control and BMSC groups. The CXCR4 expression level in the EPO and BMSC + EPO groups was consistent with the expression data obtaeind for EPOR; however, this effect was antagonized with the administration of AMD3100 (Fig. 8B).