All procedures were approved by the Institute of Animal Care and Use Committee of Tangdu Hospital, Fourth Military Medical University. All methods were performed in accordance with “the Regulations on the Care and Usage of Laboratory Animals” issued by the Ministry of Science and Technology of the People's Republic of China. Total 100 adult female (8-week old) Sprague-Dawley (SD) rats were purchased from Laboratory Animal Center of Fourth Military Medical University (Xi'an, China). They were housed in facility maintained at 22–24°C, 45–55% humidity, 12:12 h light: dark cycle with free access to food and water. All rats were allowed to acclimate for 1 week before experiments.

A rat SCI model was established with a modified Allen's method (Chen et al., 2013). Briefly, rats were anesthetized by intraperitoneal injection of pentobarbital sodium at 50 mg/kg and fixed on a stereotaxic apparatus (DL Naturegene, Beijing, China). After shearing hair and disinfecting with 75% ethanol, an about 3 cm incision was made at the 12th rib, skin was opened and muscles around the spinous process were separated. The spinal cord was exposed by removing the T11-T12 vertebral plate and opening the canalis spinalis to expose dura mater spinalis. The T10 and T13 spinous processes were clamped to ensure the location of injury. The contusion was introduced with a customized IH-0400 Spinal Cord Impactor (PSI, Lexington, KY, USA) at the spinal cord corresponding to the T12 spinous process. The rod was 20 g and 2.5 mm in diameter. The striking force was 20 × 2.5 g•cm and contact time was 1 s. Then the muscles were realigned and wound was closed. The rats in sham group were undergone the same procedure except spinal cord contusion. The body temperature was maintained at 36.0–37.0°C with incandescent lamp and rectal thermometer monitoring throughout the operation. After operation, rats were individually housed and given tetracycline (10 mg/kg, SQ) and buprenorphine (0.05 mg/kg, SQ) twice daily for 3 days. The bladder was emptied manually three times daily until rats were capable for reflex bladder emptying. Nine rats were excluded from further experiments due to death (2 rats from time-course study, 1 from bumetanide group), insufficient injury (1 rat from SCI group, locomotor function recovered in <24 h), and no or too little eating/drinking (4 rats from time-course study, 1 from TGN-020 + bumetanide group). Rats were euthanized at specified time from 0 to 7 days (n = 7 at each time point) by intraperitoneally administering 150 mg/kg pentobarbital with 25 mg/kg phenytoin. Spinal fluid was collected by direct cistern magna puncture (Mahat et al., 2012). Spinal cord around the striking site (2 cm in length) was collected and divided into 4 portions across the epicenter for assessing water content (rostral ventral section), for RNA work (rostral dorsal section), for protein work (caudal ventral section), and fixed for histological works (caudal dorsal section).

For drug treatment, 200 mg/kg of TGN-020 (Sigma-Aldrich, St. Louis, MO) (Igarashi et al., 2011) and/or 0.3 mg/kg of Bumetanide (Sigma-Aldrich) (Cleary et al., 2013), dissolved in 0.1 ml normal saline was administered intra-peritoneally 15 min before induction of SCI (n = 7 in each treatment group). Those animals were sacrificed 48 h after SCI.

The 21-point (0–21) Basso, Beattie, and Bresnahan (BBB) locomotor rating scale was used to assess the behaviors of rats before and 0.5, 1, 3, and 7 days after injury (n = 7), which was based on the observation of hindlimb movements of a rat freely moving in an open field (Basso et al., 1995, 1996). During the evaluation, rats were allowed to freely walk on the open field for 4 min.

The level of spinal cord edema was expressed by water content in the spinal cord tissue. A 1.5-cm spinal cord tissue band centered around the injury site was weighed for the wet weight and weighed again for the dry weight after it was dried for 24 h in an 80°C oven. The water content in the spinal cord tissue was calculated as (wet weight–dry weight)/wet weight × 100% (Li et al., 2016; Cabrera-Aldana et al., 2017).

LDH activity in spinal fluid was analyzed using an LDH activity assay kit (NJJCBio, Nanjing, China) according to manufacturer's instruction. Briefly, 20 μl of spinal fluid was mixed with 250 μl matrix buffer, 50 μl coenzyme I working solution and incubated at 37°C for 15 min. Then 250 μl 2,4-dinitrophenylhydrazine was mixed in and the mixture was incubated another 15 min at 37°C. The reaction was stopped by 250 μl of 0.4 mol/L NaOH and left at room temperature for 3 min before read at 440 nm. Spinal fluid LDH activity (U/ml) = (OD_Sample−OD_Control)/(OD_Standard−OD_Blank) ^* Standard Concentration ^* dilution factor/sample volume (ml).

Total RNA from rat spinal cord tissues was extracted with RNeasy Mini Kit (Qiagen, Shanghai, China) according to manufacturer's manual. The first strand cDNA was synthesized with a reverse transcription kit from Tiangen Biotech (Beijing, China) according to manufacturer's protocol. Quantitative real-time PCR was performed with TransStart Top Green qPCR SuperMix from TransGen (AQ131-01, Beijing, China) on a ABI 7300 (Applied Biosystems, Foster City, CA) with primers CTCAACGCACCTAACAGGGA and GACGGAAGGCGGTTTTCAAG for NKCC1; CTGGGGGCAGGCAATGAGAG and GGGAGGTCCACACTTACCCC for AQP4; and GATGTGGATCAGCAAGCAGGA and AAAACGCAGCTCAGTAACAGTCC for Actb. The reaction program was consisted of 95°C for 3 min followed by 40 cycles of 95°C 30 s, 55°C 20 s, and 72°C 20 s. The relative mRNA levels were calculated by 2^−ΔΔCt method with Actb as the internal control.

The total protein of the spinal cord tissues was extracted using RIPA lysis buffer. Total proteins of each group (40 μg) were resolved in 8% sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and then transferred onto polyvinylidenefluoride (PVDF) membranes (Millipore, Bedford, MA). After the membranes were blocked in5% non-fat milk at room temperature for 30 min, they were incubated with anti-AQP4 (ab46182, Abcam, Cambridge, MA), anti-NKCC1 (ab59791, Abcam), or anti-β-actin (A2066, Sigma, St Louis, MO) antibody at 4°C overnight. After 3 washes with PBST, the membranes were incubated with proper horseradish peroxidase-conjugated goat anti-rabbit IgG antibody at room temperature for 60 min, and visualized with the enhanced chemiluminescence (ECL) substrate (ThermoFisher, Shanghai, China). The images were scanned and analyzed with ImageJ (NIH, Bethesda, MD).

The histology of spinal cord was evaluated by hematoxylin and eosin (HE) staining and Pischinger methylene blue staining (Khedkar et al., 2012). The left ventral section of spinal cord was fixed in formalin and embedded into paraffin. The 6 μm coronal sections were cut from the tissue block and stained with HE stain or Pischinger's methylene blue before evaluated microscopically with an Olympus IX71 Inverted Fluorescence Phase Contrast Microscope (Olympus, Shanghai, China). The areas of cavities were assessed using Digimizer software (MedCalc Software bvba, Ostend, Belgian) and number of dendrites was quantified with ImageJ software (NIH, Bethesda, MD).

Paraffin-embedded spinal cord tissue at injury site was cut at a thickness of 6 micron. Slides were deparaffinized and rehydrated by washing in xylene and passing through ethanol gradient. Endogenous peroxidase activity was quenched with freshly made 0.3% hydrogen peroxide and antigens were heat retrieved in citrate buffer. Slides were blocked with normal goat serum at 37°C for 30 min, incubated with antibody against AQP4 (ab46182, Abcam) or NKCC1 (ab59791, Abcam) overnight at 4°C, rinsed with 0.01 M PBS 3 min for 3 times, incubated with FITC- (for AQP4) or Cy5- (for NKCC1) conjugated goat anti-rabbit IgG antibodies at 37°C for 30 min, conterstained with DAPI. The slides were dehydrated with alcohol gradient, xylene cleared, and mounted with neutral gum. The images were obtained with an Olympus IX71 Microscope.

The data were expressed as mean ± standard error (SE). The sample size was 7 rats in each group. Statistical analyses were performed using the SPSS16.0 (IBM, Chicago, IL). The differences between groups were analyzed using student t-tests or one-way analysis of variance (ANOVA) followed by post-hoc bonferroni test. P-values less than 0.05 were considered statistically significant.

The locomotor activity of rats was lost after immediately SCI shown by the plunge of BBB score to near 0 and remaining there for about a day before visible joint movements. BBB score showed modest increase from day 1 to day 7 after SCI (Figure 1A).

Traumatic SCI caused rapid occurrence of spinal edema which peaked at 24 h and lasted at least 7 days after SCI. The spinal cord water content was seen significantly increased 1 h (72.92 ± 0.59%), peaked at 24 h after SCI (76.41 ± 1.09%), and reduced thereafter but kept at significantly higher levels than that of uninjured spinal cord (70.54 ± 0.71%; Figure 1B). Accompanied by spinal cord edema, the histology of spinal cord tissue was gradually distorted and damaged as cavitations appeared 6 h after injury, then the destruction of spinal cord structure became progressively profound during the duration monitored (7 days; Figures 1C,D).

Meanwhile, the LDH activity in cerebrospinal fluid was significantly increased from 1614.67 ± 26.08 to 2892.33 ± 25.97 U/g protein 6 h after SCI and peaked at 24 h after injury (3953.46 ± 58.50 U/g protein), which then reduced but maintained at high levels until at least 7 days after SCI (Figure 2A). Pischinger's methylene blue staining showed reduced number of dendrites and increase of tissue cavities from 6 h after SCI and deteriorated until at least 7 days after SCI (Figures 2B,C).

The mRNA levels of AQP4 (Figure 3A) and NKCC1 (Figure 3B) of spinal cord were significantly elevated 6 h, peaked 24 h, and decreased from the peak but maintained at high levels until 7 days after injury. Similarly, AQP4 and NKCC1 protein rapidly accumulated in spinal cord after SCI, peaked around 24 h after SCI, and reduced thereafter but maintained at significantly higher levels compared to uninjured spinal cords (0 h) through at least 7 days after injury (Figures 3C,D).

Rats receiving AQP4 inhibitor TGN-020 or NKCC1 blocker bumetanide before SCI had significantly lower spinal cord water content compared to untreated SCI rats, and concomitant administration of TGN-020 and bumetanide resulted in further reduction of spinal cord edema (Figure 4A). Meantime, TGN-20 and or bumetanide significantly inhibited SCI-caused increase of cerebrospinal fluid LDH activity from about 30% to more than 50% (Figure 4B). Consequently, SCI caused spinal cord tissue destruction was substantially reduced by TGN-020, or bumetanide, or both (Figures 4C–E). TGN-020 and bumetanide reduced cavity area by about 30% respectively, and the combination of TGN-020 and bumetanide reduced the cavity area by more than 50% (Figures 4C,D). Meanwhile, TGN-020 and bumetanide significantly reduced SCI-caused loss of dentrites in monotherapy or combinatorial treatment (Figures 4C,E).

Blockade of AQP4 with TGN-020 not only inhibited SCI induced upregulation of AQP4 mRNA (Figure 5A) and protein (Figures 5B–D) levels of spinal cord, but reduced NKCC1 expression as well (Figure 5). Similarly, bumetanide inhibited both SCI-induced NKCC1 and AQP4 overexpression (Figure 5). Blocking both AQP4 and NKCC1 resulted in markedly decrease of their mRNA and protein levels after SCI compared to treatment with either bumetanide or TGN-020 alone (Figures 5A–D).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. The reviewer JG and handling Editor declared their shared affiliation.