Recombinant Murine IP-10/CXCL10 (P6740) was purchased from Beyotime Biotechnology Inc. (Shanghai, China). Recombinant neuregulin-1$\beta$ (Nrg1, CYT-1186) was obtained from Prospec-Tany Technogene Ltd. (Ness-Ziona, Israel).

Female C57BL/6 mice (three months old, from Guangdong Medical Laboratory Animal Center, Guangzhou, China) were housed at the Animal Center of Shantou University Medical College (SUMC) under specific pathogen-free conditions (25 °C, 12 h day/night controlled light cycle) and provided with food and water ad libitum. Animal experimental protocols followed the rules and regulations of the Ethics Committee of SUMC (SUMC2014-004) and conformed to the guidelines of the Chinese Animal Welfare Agency. HT22 and NSC34 mouse neuronal cells were purchased from Guangzhou Jennio Biotech Co., Ltd. (Guangzhou, China) and grown in dulbecco’s modified eagle medium (DMEM) containing 10% fetal bovine serum. Both cell lines were authenticated at Jennio Biotech Co., Ltd. by short tandem repeat (STR) profiling. No contamination of mycoplasma has been identified by the company. HT22 cells and NSC34 cells are both neural cell lines. NSC34 cells are obtained from mouse spinal cord neural cells and HT22 cells are mouse hippocampal neurons. They are both used as excellent in vitro models for research of neuronal damage. Some studies use NSC34 cells as in vitro models for SCI while others use HT22 cells [17, 18, 19, 20]. Therefore, in this study both cell lines were employed as in vivo models, which allows for a more comprehensive study of the effects of CXCL10 and Nrg1 on neurons.

Microarray datasets GSE42828 and GSE93561 were downloaded from Gene Expression Omnibus (GEO) and collected using the following platforms: GPL1261 [Mouse430_2] Affymetrix Mouse Genome 430 2.0 Array. Gene expression matrices from two datasets were merged, inter-batch differences were removed, and the R package “ggplot2” was used for data visualization.

Mice were injected intraperitoneally with a mixture of xylazine (5 mg/kg, Sigma-Aldrich, Taufkirchen, Germany) and ketamine (100 mg/kg, Fujian Gutian Pharmaceutical, Gutian, Fujian, China). A laminectomy was then performed at the T9-T11 levels. In the sham group mice, the spinal segment was exposed without damaging the dura, while in the SCI group mice the spinal cord was compressed for five seconds using a pair of forceps (RWD Life Science Co., Ltd., Shenzhen, China). Bladders were manually expressed once daily until a bladder reflex was established. Eight weeks post injury, three mice in each group were used for morphological assay and perfused with 4% paraformaldehyde solution (BL539A, Biosharp, Hefei, Anhui, China). After treatment with a 30% sucrose solution, spinal cord tissues were cryosectioned at 8 µm thickness prior to further experiment.

Standard immunocytochemistry was performed as described previously [21]. Sections were rehydrated through a graded series of ethanol into phosphate buffer saline (PBS) and then subjected to heat-mediated antigen retrieval in sodium citrate antigen retrieval solution (C1010, Solarbio, Beijing, China) at 99 °C for 40 min. Sections were then blocked in 3% H${}_2$O${}_2$ to clear endogenous peroxidase and incubated overnight at 4 °C with primary antibodies: rabbit anti-GFAP antibody (sc-33673, Santa Cruz Biotechnology, Dallas, TX, USA; 1:200), mouse anti-CD68 antibody (sc-20060, Santa Cruz, Dallas, TX, USA; 1:100), and rabbit anti-CXCL10 antibody (10937-1-AP, ProteinTech, Rosemont, IL, USA; 1:200). After washing three times in PBS for 10 min, sections were incubated with biotinylated secondary antibody and streptavidin-peroxidase conjugate at room temperature for 2 h. The AEC kit (Zhongshan Goldbridge Biotechnology Co., LTD., Zhongshan, Guangdong, China) was used to visualize antigen-antibody complexes. The stained sections were imaged using a Jiangnan light microscope (DN-10B, Jiangnan, Nanjing, Jiangsu, China). Quantification of the staining intensity was performed by densitometry using Image J software (V1.6, National Institutes of Health, Bethesda, MD, USA).

The spinal cord section pretreatment was described above in the section “Immunohistochemical staining”. Following incubation with antibody dilution buffer (Cat no. A1800, Solarbio, Beijing, China), sections were treated with a mixture of the primary rabbit anti-mouse CXCL10 antibody (10937-1-AP, ProteinTech, Rosemont, IL, USA; 1:200) and mouse anti-mouse $\beta$III-Tubulin antibody (sc-80005; Santa Cruz, Dallas, TX, USA; 1:200) [22]. Counterstaining was performed using 4’,6-diamidino-2-phenylindole (DAPI; Beyotime Biotechnology, Shanghai, China). Fluorescence microscopy was performed using an Axio Imager Z2 fluorescence microscope (objective: 40$\times$/0.95; Carl Zeiss LSM880, Zeiss, Jena, Germany).

Cell viability was determined using a CCK-8 assay (C0041, Beyotime Biotechnology, Shanghai, China). HT22 and NSC34 cells were seeded into 96-well plates (1 $\times$ 10${}^3$ cells/well) for 12 h. After treatment with CXCL10 (0–10 nM) or Nrg1 (0–10 nM) as well as a mixture of CXCL10 (5 nM) and Nrg1 (5 nM) for 12 h, HT22 and NSC34 mouse neuronal cells were treated with fresh DMEM containing 10% CCK-8. Absorbance was then measured using a microplate reader (Bio-Tek, Winooski, VT, USA). Cell viability was expressed as the percentage of vehicle control absorbance at 450 nm.

Cells were seeded in a 12-well plate at a concentration of 2 $\times$ 10${}^5$ cells/well. When confluence reached 80%, cultured cells were scratched across the cell surface vertically by a 200 µL pipette tip and unattached cells were washed away by fresh DMEM. 1 mL DMEM with CXCL10 (5 nM), Nrg1 (5 nM), or a mixture of CXCL10 (5 nM) and Nrg1 (5 nM) were then individually added. Cells in the control group were treated with DMEM. The scratch area and wound coverage of the total area were determined at 0, 6, 12, 24 and 48 hours post injury. The percentage of wound closure was calculated using the following formula, in which A${}_t$${}_{\mathrm{=}\mathrm{0}}$ is the initial wound area, and A${}_{t\mathrm{=}\mathrm{\Delta }\mathit{}t}$ is the wound area after $\mathrm{\Delta }$t hours of the initial scratch [23].

(1)$$\%\text{of Wound Closure}=\left(\frac{A_{t=0}-A_{t=\mathrm{\Delta }t}}{A_{t=0}}\right)\times 100\%$$

HT22 and NSC34 cells were lysed in RIPA lysis buffer (P0013B, Beyotime Biotechnology, Shanghai, China) with Phenylmethanesulfonyl fluoride (PMSF, ST506, Beyotime Biotechnology, Shanghai, China) and Phosphatase inhibitor (PPI, P1081, Beyotime Biotechnology, Shanghai, China) to prepare protein samples. Proteins were separated by SDS–polyacrylamide gel electrophoresis and transferred to polyvinylidene difluoride membranes (0.45 µm; Merck Millipore Ltd., Darmstadt, Germany). After blocking for one hour with 5% bovine serum albumin, membranes were incubated with the appropriate primary (Anti-phosphor-ErbB4 antibody (bs-3220R, Bioss, Beijing, China), anti-ErbB4 antibody (sc-71071, Santa Cruz, Dallas, TX, USA), anti-phosphor-ERK1/2 antibody (sc-136521, Santa Cruz, Dallas, TX, USA), anti-ERK1/2 antibody (sc-514302, Santa Cruz, Dallas, TX, USA), anti-caspase 9 antibody (AF1264, Beyotime Biotechnology, Shanghai, China), anti-caspase 3 antibody (P17, Boster, Wuhan, Hubei, China) and anti-GAPDH antibody (sc-365062, Santa Cruz, Dallas, TX, USA)), and secondary antibodies. The protein bands were quantified using a gel imaging analysis system (Tanon-2500B, Tanon, Shanghai, China). The signal intensity was quantified by densitometry using Image J software (V1.6, National Institutes of Health, Bethesda, MD, USA).

Statistical testing was implemented using Statistical Package for the Social Sciences software (SPSS, Version 20.0, Chicago, IL, USA). Differences between two groups were compared by Student’s t-test. The data were given as the mean $\pm$ standard error of the mean (SEM). The p-values are given in the Figure legends.

Microarray dataset profiles GSE42828 and GSE93561 were downloaded from GEO DataSets. After normalization, probes were converted to gene symbols for series matrix files of each dataset and the gene expression data of these two datasets were merged (Fig. 1A). The differential expressions of Cxcl10 and Nrg1 genes in both datasets were tested and it was found that the expression level of Cxcl10 is significantly up-regulated in SCI samples, whereas that of Nrg1 is significantly down-regulated in the three-day and seven-day SCI groups, the young SCI group, and the old SCI group when compared with control samples (Fig. 1B).

Fig. 1.

Expression of Cxcl10 and Nrg1 genes in microarray datasets GSE42828 and GSE93561. (A) Gene expression matrices were merged with no inter-batch differences. (B) Expression score of Cxcl10 and Nrg1 genes in microarray datasets GSE42828 and GSE93561 (**, vs. 0 d control, p 0.01; ***, vs. 0 d control, p 0.001; ${}^{\mathrm{\#}\mathrm{\#}}$, vs. young control, p 0.01; ${}^{\mathrm{\#}}$, vs. young control, p 0.05; , vs. old control, p 0.05; ns, no significance). (C) Log${}_2$ values in both Cxcl10 and Nrg1 expressions. Log${}_2$ value indicates the quotient of expression quantity in the SCI group and that in the sham group taking the logarithm base 2. Nrg1, Neuregulin 1; CXCL10, C–X–C motif chemokine ligand 10; SCI, spinal cord injury.

After log${}_2$ transformation, it was also found that in SCI mice of different ages or at different time points post SCI, the Cxcl10 and Nrg1 genes showed different expression quantities (Fig. 1C). the log${}_2$ value gives the quotient of expression quantity in the SCI group and the sham group after taking the base 2 logarithm. Log${}_2$ values in both the Cxcl10 and Nrg1 gene expressions of the young group were higher than those of the old group. Log${}_2$ value analysis demonstrated that compared to the vehicle control, Nrg1 expression is reduced at three and seven days post injury. It was also noted there was a minimum log${}_2$ value of Cxcl10 at three days post injury. That means that Cxcl10 expression is increased post injury, but moderated at three days post injury.

It was next aimed at determining whether CXCL10 is involved in neuronal injury. To address this, mouse spinal cord slices were harvested for immunohistochemical and immunofluorescence staining at eight weeks following SCI or sham operation. By using immunohistochemical staining, it was demonstrated that CD68, GFAP, and CXCL10 levels were significantly increased eight weeks post injury when compared with sham controls. These data suggest that SCI causes glial cell activation, accompanied by enhanced expression of CXCL10 (Fig. 2A–F).

Fig. 2.

SCI of mice causes inflammation in the spinal cord. (A–F) In SCI mice, high expression levels of GFAP (A,D), CD68 (B,E), and CXCL10 (C,F) were observed in the injured spinal cord, Scar bar: 200 µm. S.E.M. (**p 0.01). (G) Representative immunofluorescence image of the spinal cord of mice eight weeks after SCI staining with $\beta$III-Tubulin and CXCL10 is shown. Scar bar: 50 µm. GFAP, Glial Fibrillary Acidic Protein; CXCL10, C–X–C motif chemokine ligand 10; DAPI, 4’,6-diamidino-2-phenylindole.

Additionally, immunofluorescence staining showed partial localization of CXCL10 adjacent to $\beta$III-Tubulin-positive neuronal fibers, suggesting a direct impact of CXCL10 on neurons under the pathological conditions of SCI (Fig. 2G).

To determine whether CXCL10 intensifies neuronal cell damage, the viability of HT22 and NSC34 cells was initially assessed. A similar phenomenon occurred when HT22 and NSC34 cells were treated for 12 h with a serial concentration of CXCL10 (0–10 nM). It appears that 2, 5 and 10 nM of CXCL10 reduced cell viability in both cell types (Fig. 3A,D). Both cell types were then treated with Nrg1 (0–10 nM) for 12 h, to explore the basic effect of Nrg1. Nrg1 increased cell viability in a dose-dependent manner at doses ranging from 2 to 5 nM (Fig. 3B,E). Nrg1 and CXCL10 were both then assessed, either individually or in combination, at 5 nM to evaluate how the two cooperated in HT22 and NSC34 cells (Fig. 3C,F). The CCK8 assay results showed that the CXCL10-induced decrease in cell viability was significantly rescued by Nrg1.

Fig. 3.

Effect of CXCL10 and Nrg1 on HT22 cells or NSC34 cells viability. (A–C) Effect of CXCL10 (A) and Nrg1 (B), as well as combined effects of the two (C) on HT22 cell viability. n = 10 per group. S.E.M. (***, p 0.001; **, p 0.01; ns, no significance). (D–F) Effect of CXCL10 (D) and Nrg1 (E), as well as combined effects of the two (F) on NSC34 cell viability. n = 10 per group. S.E.M. (***, p 0.001; ns, no significance). Nrg1, Neuregulin 1.

Scratch-injury assay was used to clarify the role of CXCL10 and Nrg1 in recovery from neuronal injury. Scratch injury activated the primary damage of neurons and then affected the entire neuronal migration, which to a certain degree mimics neuronal injury under the pathological condition of SCI. Therefore, it is appropriate to monitor the neuronal response after SCI in vitro with the use of the scratch assay [19]. In both HT22 and NSC34 cells, microscopic imaging revealed that wound closure failed to be inhibited in the CXCL10 group. Intriguingly, CXCL10 promoted scratch healing in HT22 cells at 24 h, and in NSC34 cells at 24 and 48 h (Fig. 4A–D). Wound closure was significantly increased by Nrg1 at 24 h after scratch in both cell types (Fig. 4A–D). Nrg1 and CXCL10 co-treated HT22 and NSC34 cells showed significant wound closure 24 and 48 h following scratch injury (Fig. 4A–D).

Fig. 4.

Effect of CXCL10 and Nrg1 on HT22 cells and NSC34 cells after scratch injury. (A,C) Effect of CXCL10 and Nrg1 on HT22 cells after scratch injury. n = 4 per group. Scar bar: 200 µm. S.E.M. (***, Nrg1 + CXCL10 group vs. control group, p 0.001; ${}^{\mathrm{\#}}$, CXCL10 group vs. control group, p 0.05; , NRG1 group vs. control group, p 0.05). (B,D) Effect of CXCL10 and Nrg1 on NSC34 cells after scratch injury. n = 4 per group. Scar bar: 200 µm. S.E.M. (**, Nrg1 + CXCL10 group vs. control group, p 0.001; ${}^{\mathrm{\#}\mathrm{\#}}$, CXCL10 group vs. control group, p 0.01; ${}^{\mathrm{\#}}$, CXCL10 group vs. control group, p 0.05; , NRG1 group vs. control group, p 0.05).

Next, we wanted to investigate the effects of CXCL10 and NRG1 on related molecules in HT22 and NSC34 cells. The protein levels of pErbB4, pERK1/2, cleaved caspase 9, and cleaved caspase 3 were detected in HT22 and NSC34 cells (Fig. 5A and Fig. 6A). Nrg1 binds to ErbB4 and promotes its phosphorylation [24]. Cleaved caspase 9 and cleaved caspase 3 play an apoptosis-promoting role [11]. pERK1/2 is involved in cell migration, cell survival and cell proliferation [25, 26]. These molecules may serve as downstream molecules of CXCL10 or Nrg1 to perform critical functions.

Fig. 5.

Effect of CXCL10 and Nrg1 on the molecular expression of HT22 cells. (A) Representative images of pErbB4, pERK1/2, cleaved caspase 9, and cleaved caspase 3. (B–E) Quantitative results of pErbB4, pERK1/2, cleaved caspase 9, and cleaved caspase 3. n = 3 per group. S.E.M. (**p 0.01; *p 0.05; ns, no significance). GAPDH, glyceraldehyde-3-phosphate dehydrogenase.

Fig. 6.

Effect of CXCL10 and Nrg1 on the molecular expression of NSC34 cells. (A) Representative images of pErbB4, pERK1/2, cleaved caspase 9, and cleaved caspase 3. (B–E) Quantitative results of pErbB4, pERK1/2, cleaved caspase 9, and cleaved caspase 3. n = 3 per group. S.E.M. (**p 0.01; *p 0.05; ns, no significance).

In HT22 cells results showed that, compared with the sham group, CXCL10 induced an increase in the expressions of cleaved caspase 9 and cleaved caspase 3, with no distinct difference in the level of pERK1/2 (Fig. 5D,E). High levels of cleaved caspase 9, cleaved caspase 3, and pERK1/2 were observed in CXCL10-treated NSC34 cells (Fig. 6D,E), although Nrg1 treatment induced an increase in the level of pErbB4 and pERK1/2 in both cell lines (Fig. 5B–D and Fig. 6B–D). Additionally, Nrg1 mitigated the expressions of cleaved caspase 9 and cleaved caspase 3 induced by CXCL10, accompanied by enhanced phosphorylation levels of ERK1/2 in both cell lines (Fig. 5D,E and Fig. 6D,E).