CD73 reduces inflammation-related microglia cell death in spinal cord injury through PI3K/AKT/Foxo1 signaling

From January 2019 to June 2019, 20 SCI patients from Huashan Hospital, Fudan University were enrolled. Inclusion criteria included the following: (a) a clear history of trauma, and there were no neurological abnormalities of SCI before injury; (b) existing neurological abnormalities such as limb sensation, motor abnormality, and dysfunction of the bowel and bladder; (c) MRI examination showed spinal cord compression and spinal cord signal changes. Patients with treatment of methylprednisolone before blood taking, with a history of brain disease, with a history of spinal surgery were excluded. Peripheral blood was collected from 20 patients with SCI and 20 healthy donors with similar gender and age distribution, respectively. Clinical characteristics of SCI patient and healthy donors have been showed in Table 1. The Japanese orthopedic association (JOA) score, cervical dysfunction index (neck disability index [NDI]), and American spinal injury association (ASIA) classification were used to assess the severity of SCI. The whole blood RNA was extracted using GeneJET Stabilized and Fresh Whole Blood RNA Kit (ThermoFisher, CA) in accordance with the manufacturer's protocol.

Total RNA was extracted using TRIzol reagent (Invitrogen, San Diego, CA) according to the manufacturer's instructions, and quantification of mRNAs was performed using a 10‐μL final reaction volume using SYBR Green polymerase chain reaction (PCR) Master Mix (ThermoFisher, CA). GAPDH mRNA was used as a housekeeping gene, and relative expression levels of mRNAs were calculated using the comparative ΔΔCT method. Raw data have been supplied as File S1.

C57BL/6J CD73 knock out (KO) male mice used in our study were reproduced from those gifted by Prof. Thompson, Oklahoma Medical Research Foundation, Oklahoma City. Wild‐type (WT) mice used in the experiments were the descendants of the first generation which were bought from Shanghai SLAC Laboratory Animal Co., Ltd. (Shanghai, China). The specific pathogen‐free facility with a 12 hours light/dark cycle and a controlled temperature and humidity were used to house mice. All the experiments were conducted according to the protocols approved by the Institutional Animal Care and Use Committee of Fudan University.

Just as descripted in our previous study,24 mice were anesthetized with pentobarbital by intraperitoneal injection (35 mg/kg). Each mouse was inflicted with spinal crush injury at the T8‐T9 with Dumont‐type forceps with a 0.2 mm spacer. We carried out the vertebrae laminectomies of T8‐T9 with a pair of fine rongeurs, and the dura mater of mice was protected. The mice SCI model was made by lateral compression of the spinal cord with a depth of 0.2 mm for 20 seconds. After operation, the mice were given intramuscular injection of penicillin, 20 000 units once to resist infection. For further studies the effect of AKT and HIF‐1α on SCI, SC79 (40 mg/kg in DMSO) and BAY87‐2243 (0.5 mg/kg and 4 mg/kg in DMSO) were injected intraperitoneally into mice every day after the establishment of mice SCI model. The mice in the control group were injected with the same volume of DMSO at the same time point.

The locomotor behavior after SCI was detected by the Basso Mouse Scale (BMS) locomotor recovery scale. This scale is based on the observation of ankle movement, plantar placement, weight support, stepping, coordination, paw position, and trunk stability in an open field and consists of a 9‐point scale. Sham group, SCI+DMSO, and SCI+SC79 group were assessed on postoperative days 1, 3, 7, 14, 21, and 28. An inspector blinded to the treatments assessed them for 4 minutes in an open field and got the score from 0 (no spontaneous locomotor activity) to 9 (normal coordinated gait with parallel paw placement).

Mixed glial cell cultures were generated from the cerebral cortex of postnatal (24 hours old) mice and cultured in Dulbecco's modified Eagle's medium/F12 containing 10% fetal bovine serum (FBS; Gibco, Carlsbad, CA) and an antibiotic mixture (1% penicillin/streptomycin) (Invitrogen, Carlsbad, CA) at 37°C and 5% CO₂ for 10 days. Cultures were shaken for 6 hours at 180 rpm at 37°C to collect and purify microglia. BV2 cells were cultured in DMEM (Gibco, Carlsbad, CA) supplemented with 10% FBS (Gibco, Carlsbad, CA), 50 g/mL streptomycin (Invitrogen, Carlsbad, CA), and 50 U/mL penicillin in a humidified atmosphere of 95% air and 5% CO₂. The plasmid of RNAi‐CD73, pcDNA‐CD73, and RNAi‐HIF‐1 were constructed by Genechem Co., Ltd. (Shanghai, China) and then transfected into BV2 cells using Lipofectamine 2000 reagent (Invitrogen, Carlsbad CA) following the protocol stipulated by the manufacturer. Different combinations of Lipopolysaccharide (LPS) (1 μg/mL), adenosine (10μM), MRS1706 (0.3μM), MK2206 (3μM) were added to the media 24 hours after plasmid transfection, and 8 hours later, mRNA and protein were harvested.

Total RNA was isolated from cells using the Trizol (Invitrogen, Carlsbad, CA) according to the manufacturer's protocol. RNA integrity was evaluated using the Agilent 2200 TapeStation (Agilent Technologies, Santa Clara, CA), and each sample had the RIN above 7.0. Subsequently, the purified RNAs were subjected to first strand and second strand cDNA synthesis following by adaptor ligation and enrichment with a low‐cycle according to instructions of NEBNext Ultra RNA Library Prep Kit for Illumina (NEB, Ipswich, MA). The purified l library products were evaluated using the Agilent 2200 TapeStation and Qubit2.0 (Life Technologies, Carlsbad, CA) and then diluted to 10 pM for cluster generation in situ on the pair‐end flow cell followed by sequencing (2 × 150 bp) HiSeq3000. The clean reads were obtained after removal of reads containing adapter, ploy‐N and at low quality from raw data. HISAT2 was used to align the clean reads to the mouse reference genome mm 10 with default parameters. HTSeq was subsequently employed to convert aligned short reads into read counts for each gene model. Differential expression was assessed by DEseq using read counts as input. The Benjamini‐Hochberg multiple test correction method was enabled. Differentially expressed genes were chosen according to the criteria of fold change >2 and adjusted P‐value <.05. All the differentially expressed genes were used for heat map analysis and Kyoto Encyclopedia of Genes and Genomes (KEGG) ontology enrichment analyses. For KEGG enrichment analysis, a P‐value <.05 was used as the threshold to determine significant enrichment of the gene sets.

In cell culture supernatants or mouse spinal cord homogenate, the measurement of protein levels of IL‐1β, IL‐6, and TNF‐α was taken using the commercial ELISA kits from Sigma (Sigma‐Aldrich, St. Louis, MO) according to the manufacturer's instructions.

The release of lactate dehydrogenase (LDH) was detected to determine the cytotoxicity using the LDH Cytotoxicity Assay Kit (Beyotime, Shanghai, China) following the manufacturer's instructions.

Protein of spinal cord tissue and BV2 cells was homogenized in radioimmunoprecipitation assay lysis buffer, and protein concentrations were determined using the bicinchoninic acid assay. Protein samples were fractionated using sodium dodecyl sulfate (SDS)‐polyacrylamide gel electrophoresis, and transferred to nitrocellulose membranes. Following blocking for 1 hour with 5% skimmed milk in Tris Buffered saline Tween (TBST), membranes were incubated with antibodies included NLRP3 (1:1000, Abcam, ab214185), PI3K (1:1000, CST, 4228), AKT (1:1000, CST, 2920), p‐AKT (1:1000, CST, 4070), mTOR (1:1000, CST 2972S), p‐mTOR (1:1000, CST, 5536S), IKK‐β (1:1000, CST, 2678S), p‐IKK‐β (1:1000, CST, 2697S), GSK‐3β (1:1000, CST, 2456S) p‐GSK‐3β (1:1000, CST, 5558S), Bad (1:1000, CST, 9268S), p‐Bad (1:1000, CST, 5284S), Foxo1 (1:1000, CST, 14952), p‐Foxo1 (1:1000, CST, 9641), GSDMD (1:1000, Abcam, ab210070), ASC (1:1000, CST, 67824), and GAPDH (1:2000, Abcam, ab8245) overnight at 4°C. After being washed in TBST, membranes were incubated with HRP‐conjugated secondary antibody at room temperature for 1 hour, and protein was detected using an enhanced chemiluminescence kit. Signal intensities were quantified by gel imaging system (UVP LLC, Upland, CA). The quantification of blots was performed by Image J.

At the third day after SCI, different groups' mice were deeply anesthetized with 10 % chloralic hydras (3.5 mL/kg, i.p.) and perfused with 0.9 % NaCl, followed by 4 % paraformaldehyde in 0.01 M phosphate‐buffered saline (PBS, pH = 7.4). Spinal cord segments near the lesion epicenter were collected. Three transverse paraffin sections (25μm thick) were mounted on poly‐L‐lysinecoated slides. For immunohistochemical analysis, deparaffinized sections were incubated with H₂O₂ and methanol for 10 minutes to block endogenous peroxidase, and then incubated for 30 minutes with serum‐blocking solution. The sections were then incubated with primary antibodies of CD73 (1:100, Abcam, ab175396), GSDMD (1:100, Abcam, ab210070), and CASP‐1 (1:100, Abcam, ab1872) for 1 hour, followed by incubation with HRP‐conjugated anti‐rabbit secondary antibodies for 30 minutes. Bound antibodies were visualized by incubation with 3,3N‐Diaminobenzidine Tertrahydrochlori for 10 minutes. All images were captured using a Nikon ECLIPSE Ti microscope (Nikon, Japan).

Spinal cord tissue samples were harvested as described above. BV2 cell samples were fixed with 4% paraformaldehyde in 0.1 M phosphate buffer (pH 7.4) for 15 minutes after the treatment described in cell cultures. Then all samples were blocked for 1 hour with 1 % bovine serum albumin containing 0.3 % Triton X‐100, followed by incubation (overnight at 4°C) with GSDMD (1:100, Abcam, ab210070) and CASP‐1 (1:100, Abcam, ab1872) the primary antibodies. Then all the samples were incubated (for 2 hours at room temperature) with their respective secondary antibody: Dylight (Dy)488‐ and Dy594‐conjugated secondary antibodies (all 1:1000; Jackson ImmunoResearch, West Grove, PA). All images were acquired using Nikon ECLIPSE Ti microscope (Nikon, Japan).

A −2000/+200 bp mouse GSDMD promoter of C57BL/6J mouse genomic DNA was cloned and inserted in the pGL3 basic vector (Promega, Madison, WI). Further deletion generated −1450/+200 bp, −900/+200 bp, −300/+200 bp, +50/+200 bp reporter plasmids. Mutant GSDMD reporter plasmids were generated based on −300/+200 bp plasmid. BV2 cells were cotransfected with luciferase reporter plasmids, pRL‐TK reporter plasmid (control reporter), and foxo1 plasmid (pc‐foxo1). After transfection for 24 hours, cells were harvested and measured using the toolVeritas 9100‐002 (Turner BioSystems, Sunnyvale,CA), and luciferase activity was divided by the Renilla luciferase activity to normalize for transfection efficiency.

ChIP assay was performed using a ChIP assay kit (Abcam, Cambridge, UK) according to the manufacturer's protocol. Per condition, 107 cells were treated with 1% formaldehyde for 10 minutes prior to three washes with ice‐cold PBS. Cells were then resuspended in 5 mL SDS lysis buffer containing 1 × protease inhibitor cocktail II and sonicated 15 times with 20 seconds pulses at 4°C. Insoluble material was removed by centrifugation for 10 minutes at 12 000 xg, 4°C. For each immunoprecipitation (IP), 100 μL of prepared chromatin was added to 900 μL of dilution buffer. The lysates were then pre‐cleared with Protein G agarose before being incubated with appropriate antibodies overnight at 4°C. The following day 60 μL of Protein G agarose was added to each IP and rotated for 1 hour at 4°C. The Protein G agarose‐antibody/chromatin complexes were then washed sequentially in 1 mL of the provided buffers. The DNA/protein complexes were eluted in 200 μL elution buffer before reversal of crosslinks with 5 M NaCl for 4 hours at 65°C. Samples were then treated with RNase A at 37°C for 30 minutes followed by incubation with Proteinase K for 2 hours at 45°C. The resulting DNA was purified over the provided spin filter columns, with elution in 50 μL of elution buffer C. The purified DNA was then subjected to quantitative Polymerase Chain Reaction (qPCR). Each qPCR reaction performed in triplicate containing 5 μL purified DNA and was quantified as described above.

All results were expressed as mean ± standard deviation. Student's unpaired t‐tests and two‐way analysis of variance (ANOVA) followed by Dunnett's test were used to analyze data. A P value of less than .05 was considered to be statistically significant. All statistical analyses were done with the SPSS 14.0 software.

To investigate whether NLRP3/GSDMD plays a role in neuroinflammation after SCI, 40 blood samples (20 normal vs 20 SCI) were subjected to RT‐PCR assay. Then 20 SCI patients were divided into high‐expression and low‐expression groups based on whether NLRP3/GSDMD gene expression reached an average level. As shown in Table 2, various clinical parameters in SCI patients were compared between two groups. Specifically, NLRP3/GSDMD high‐expression group had a lower JOA score and a higher NDI compared to NLRP3/GSDMD low‐expression group. Strikingly, quantitative analysis revealed that the expression of NLRP3/GSDMD was significantly higher in the SCI patient samples in comparison to normal samples (Figures 1A and 1B). Furthermore, linear regression analysis confirmed a positive correlation between NLRP3/GSDMD expression and JOA score, while a negative correlation between CD73/GSDMD expression and NDI index (Figures 1C and 2D). The receiver operating characteristic curve (ROC) curve further demonstrated they may be good diagnostic markers for SCI clinically (Figure 1E). Collectively, all these above results indicate the expression level of CD73 and GSDMD in SCI patients correlate with the severity of injury.

To examine the inflammasome expression in SCI systematically, the inflammasome‐related genes were assessed in spinal cord tissue samples from mice with SCI or sham surgery. All inflammasome‐associated genes were detectable in mice spinal cord tissue 3 days after surgery, with increased IL‐1β, IL‐18, CASP‐1 and GSDMD transcript levels in SCI group (Figure S1A). Besides, all inflammasome sensors examined were also detectable, and NLRP3 and AIM2 transcript levels were upregulated (Figure S1B) after SCI. To investigate the influence of CD73 on inflammasome activation, the mRNA detection of the above genes was performed once again in CD73 KO mice with SCI. As shown in Figures 2A and 2B, compared with WT mice, various inflammasome‐associated genes (IL‐1β, IL‐18, CASP‐1, ASC, and GSDMD) as well as three inflammasome sensors (NLRP3, NLRP1, and AIM2) showed higher levels in CD73 KO mice 3 days post‐injury. In addition, the release activity of proinflammatory factor (IL‐1β, IL‐6, and TNF‐α) and LDH were also effectively increased in CD73 KO mice after SCI (Figures 2C and 2D). Consistently, western bolt and immunochemistry revealed similar protein expression patterns of NLRP3, GSDMD, ASC, and CASP‐1 as their mRNA expression (Figures 2E, 2F, and 2H). In particular, levels of full length GSDMD, N‐terminal GSDMD (GSDMD‐N, an active form of GSDMD), p20 (an active form of CASP‐1), were increased (Figures 2E and 2F). More importantly, Immunofluorescence results (Figure 2G) showed that the increased expression of GSDMD coincides with CD68 (a marker of microglia) suggesting the cell pyroptosis was occurred mainly in microglia. In addition, the inflammasome‐related genes were assessed in BV2 cells and primary microglia. As shown in Figures 2J, 2L, and 2M, the treatment with LPS increased NLRP3 and CD73 gene levels, facilitated the activation of GSDMD and CASP‐1. Consistently, the treatment of LPS had similar effect on primary microglia (Figures 2K, 2L, and 2N). Particularly, NLRP3, GSDMD, CASP‐1, and ASC showed higher levels in primary microglia from CD73 KO mice (Figures 2K, 2L, and 2N). Together, these results demonstrate that CD73 deficiency exacerbates microglia pyroptosis, and this effect may relate to NLRP3 inflammasome.

Our previous studies have shown that CD73 regulates microglia polarization through the A_2B adenosine receptor (A_2BAR).24 Therefore, to determine whether A_2BAR also mediates the CD73's ability to alleviate microglia pyroptosis, we constructed BV2 cells that either had downregulated or upregulated CD73 expression, exposed them with adenosine or A_2BAR antagonist MRS1706 after LPS pretreatment. Immunoblot and RT‐PCR results showed that CD73 interference and overexpression had acceptable efficiency (Figures 3A and 3B). Results of RT‐PCR and ELISA showed that CD73 downregulation caused a profound induction of pyroptosis genes in transcript levels including NLRP3, ASC, CASP‐1, and GSDMD, increased the release of IL‐1β, IL‐18, TNF‐α, and LDH, while CD73 upregulation was counterproductive (Figures 3C and 3E). Immunoblot and Immunofluorescence revealed similar changes in protein levels of NLRP3, GSDMD, and ASC replying to CD73 change (Figures 3F and 3I). To further determine the role of A_2BAR in the CD73 mediated BV2 cell pyroptosis, the cells were exposed with adenosine or MRS1706 (A_2BAR antagonist) after LPS pretreatment. Consequently, the presence of adenosine further alleviates the pyroptosis induced by CD73 upregulation, while the addition of MRS1706 obviously abolishes the effects of CD73 upregulation in all above experiments, respectively (Figures 3C and 3I). All these results suggest that A_2BAR may necessary for CD73 to regulate microglia pyroptosis.

To further investigate the mechanism by which CD73 regulates microglia pyroptosis, mRNA sequencing was performed on CD73‐overexpressing BV2 cells after administrated with LPS. A total of 1649 mRNAs were differentially expressed (1081 upregrated and 568 downregulated) between CD73 overexpressed BV2 cells and normal subjects (Figure 4A). Notably, 55 differentially expressed mRNAs were enriched in PI3K/AKT pathway (Figure 4B). To explore whether CD73 regulates pyroptosis through PI3K/AKT signal pathway, the expression of PI3K, AKT, and the common downstream signaling molecules of it (including mTOR, IKK‐β, GS3K‐β, Bad, and Foxo1) were detected in CD73 overexpressed BV2 cells (pretreated with LPS or not). As showed in Figures 4D and 2E, LPS treatment promoted the phosphorylation of mTOR and IKK‐β, while CD73 overexpression only increased the activation of Foxo1. Then the RT‐PCR and immunoblot were used to confirm the relationship between CD73 and PI3K/AKT/Foxo1 axis. As indicated in Figures 4C, 4F, and 4G, the mRNA and protein expressions of PI3K were all increased in CD73 overexpressed BV2 cells, and the elevated level of phosphorylated AKT and Foxo1 were also observed. In addition, we found the mRNA level of PI3K, the phosphorylated AKT and Foxo1were reversed after exposed to MRS1706 in CD73 upregulated cells, and their expression could also be suppressed by an AKT inhibitor MK2206 (Figures 4C, 4F, and 4G). Moreover, MK2206 treatment also caused a remarkably increase in the pyroptosis‐associated genes (including NLRP3, ASC, CASP‐1, and GSDMD) in transcript levels, as well as increased the release of IL‐1β, IL‐6, TNF‐α, and LDH (Figures 4H and 4J). Similar changes in protein levels of NLRP3, GSDMD, and ASC were also observed by immunoblot replying to MK2206 (Figures 4K and 4L). In vivo study, the SC79 (an AKT activator) was injected intraperitoneally (40 mg/kg) once a day after SCI in CD73 KO mice, and the results of immunohistochemistry proved that SC79 could decrease the expression of GSDMD and CASP‐1 at the lesion site (Figure 4M). Furthermore, the BMS scoring in SC79‐treated group was significantly superior to DMSO‐treated group at 28th day after injury (Figure 4N). Taken together, we conclude that CD73 could alleviate microglial pyroptosis in vitro and in vivo, and the effect may mediate via the PI3K/AKT/Foxo1 pathway.

Foxo1 is a transcriptional factor and widely participates in the transctipiton process of many genes. In order to continue analyze the molecular mechanisms involved in macroglia pyroptosis, we used the JASPAR database (http://jaspar.genereg.net) to identify potential binding sites of GSDMD promoter region for Foxo1. To measure the activity of potential 3 cis‐acting elements and determine the minimum sequence required for activity, a series of five reporters constructs with progressively larger deletions from the 5′ end of the promoter were generated. Luciferase assay revealed an increased promoter activity of the pGL3–2000/+200 bp as compared to the control group in BV2 cells, indicating the existence of binding site for Foxo1 in the promoter region of GSDMD gene (Figure 5A). Consistently, the ChIP assay confirmed the direct binding of Foxo1 to GSDMD promoter region (Figure 5B). However, when the promoter sequence was deleted to position +50 bp, the promoter activity with pGL3+50/+200 bp decreased significantly compared with the pGL3−300/+200 (Figure 5A), this result demonstrated that positive regulatory elements were located in the −300/+50 promoter region of GSDMD in BV2 cells. As expected, further mutation of this binding site effectively blocked the transcription of GSDMD (Figure 5C). Overexpression of Foxo1 in BV2 cells also increased the expression of GSDMD, while cotreatment of Foxo1 plasmid and CD73 plasmid inhibited the elevation of GSDMD mRNA, as well as reduced the release of IL‐1β, IL‐6, TNF‐α, and LDH (Figures 5D and 5F). Thus, we conclude that Foxo1 mediates the pyroptosis inhibition effect of CD73 by regulating the expression of GSDMD at transcriptional level.

Several reports have shown that HIF‐1α can regulate CD73 expression.27 So, in this study, we exposed LPS‐activated BV2 cells with a HIF‐1α small interference RNA to determine its regulatory role. Immunoblot and RT‐PCR results showed that HIF‐1α interference had acceptable efficiency (Figures 6A and 6C). LPS‐induced pyroptosis could simultaneously promote the expression of HIF‐1α and CD73 in BV2 cells, and the overexpression of CD73 was inhibited after interference of HIF‐1α (Figures 6D and 6F). In vivo, the results also indicated that SCI could increase the expression of HIF‐1α along with the elevation of CD73, while BAY87‐2243 treatment, a HIF‐1α inhibitor, reversed the expression pattern of CD73 (Figures 6G and 6J). Surprisingly, we found that the BAY87‐2243, which was supposed to only affect the HIF‐1α activity, also reduced HIF‐1α expression both in mRNA and protein levels (Figures 6G and 6J). We suspect that CD73 reduction should be responsible for this phenomenon. As we expect, the increase of HIF‐1α induced by LPS or SCI was blocked in CD73 downregulated BV2 cells and in CD73 knockout mice, respectively (Figures 6K and 6P). Moreover, Merighi et al reported a stimulatory effect of adenosine on HIF‐1α by activation of p38 mitogen‐activated protein kinases phosphorylation via A_2B adenosine receptor.28 Our previous study also demonstrated that CD73 can regulate the polarization of macrophages/microglia through this pathway.24 This prompted us to verify whether CD73 regulates HIF‐1α expression is associated with this signal. As suggested in Figures 6Q and 6S, the results of RT‐PCR and immunoblots showed that both a A_2B adenosine receptor inhibitor MRS1706 and a p38 inhibitor SB203580 can reduce the increase in HIF‐1α caused by overexpression of CD73 in BV2 cells. In general, the above studies confirmed that HIF‐1α and CD73 can promote each other and subsequently jointly regulate microglia pyroptosis.

The data that support the findings of this study are available from the corresponding author upon reasonable request.