Experimental model and subject details
The adult male Sprague–Dawley rats (aged 7–10 weeks and weighing 200–240 g) used in this study were supplied by the Experimental Animal Center of the Jilin Yisi Company (Chang Chun, China). NDRG2flox/flox and NDRG2−/− astrocyte conditional knockout (CKO) mice with a C57BL/6 background were generated by gene targeting in embryonic stem cells (Cyagen Bioscience Inc.; Guangzhou, China). NDRG2flox/flox mice were crossed with GFAP-Cre mice to generate NDRG2−/− CKO mice. The resulting NDRG2flox/flox mice and NDRG2−/− mice were viable, fertile, and showed no obvious phenotypes. The mice were used in the experiments at 6–8 weeks of age. All animals were housed in a specific pathogen-free, temperature-controlled facility with a regular 12-h light–dark cycle and ad libitum access to food and water.
Focal cerebral ischemia was induced by MCAO via the thread embolism method for 8 h in male rats and mice. MCAO was performed as previously described . Briefly, rats were anesthetized intraperitoneally with 3% pentobarbital sodium (1–2 mL/kg), and mice were anesthetized with 1% pentobarbital sodium (10 mL/kg) until no pinch-paw reflex was observed. Body temperature was maintained at 37 °C using an electric blanket. The experiments were performed in a blinded manner. A vertical midline cervical incision was performed, and the right common carotid artery (CCA) was exposed and ligated at its proximal end. All branches of the external carotid artery (ECA) were isolated, coagulated, and transected. The internal carotid artery (ICA) was isolated, and the pterygopalatine artery (PPA) was ligated close to its origin. The CCA was ligated with a slipknot in the middle of the origin of the PPA and ECA to block blood flow temporarily. A slot was cut in the CCA above the ligation near the proximal end, and a thread embolism (Cinontech, Beijing, China) was inserted along the CCA. A slipknot was made above the cut to prevent leakage. The slipknot was opened between the PPA and ECA, and the thread embolism was advanced through the ICA until resistance was felt, occluding at the origin of the middle cerebral artery. The slipknot was tightened to fix the thread embolism. The ipsilateral side was analyzed, and the contralateral side of the same rat or mouse was used as the control.
Cell isolation and cell culture
Primary astrocyte isolation was performed as described previously . Briefly, the cortices of newborn Sprague–Dawley rats 1–3 days old were dissociated in DMEM/F12 (1:1) medium (Hyclone Laboratories, Inc., Logan, UT, USA; SH30023.01) and digested in 0.05% trypsin/EDTA (ethylenediaminetetraacetic acid; Gibco, Waltham, MA, USA; 25200056) with 1% DNAse (Worthington Industries Medical Center, Inc., Columbus, OH, USA; IDLS006331) at 37 °C for 25 min to yield a single-cell suspension. Cells were collected and resuspended in DMEM/F12 (Dulbecco’s modified Eagle’s medium/nutrient mixture F-12; 1:1) medium containing 10% fetal bovine serum (FBS; Sigma, Australia, F8318) and plated in culture flasks. After 5 days, the other cell types (oligodendrocytes and microglia) growing on top of the astrocytes were removed by shaking the flasks in an orbital shaker for 16 h. Cells were passaged once they reached over 90% confluence using 0.25% trypsin/EDTA. Cells at passage 1 were used for the experiments.
We used an in vitro OGD model to mimic ischemic conditions. Hypoxic (< 0.1% oxygen) culture conditions were achieved using the Anaero Pack system in a 2.5 L jar (MGC, Tokyo, Japan); this system works as an oxygen absorber and CO2 generator. The composition of the gas mixture reached approximately 78% N2 and 22% CO2 within 1 h, as measured by the color change of the oxygen indicator, and the pressure inside the jar was equal to that outside the jar. The ‘lactate + ’ culture medium contained DMEM without glucose (Gibco, 11966025), as well as 30 mM L-lactate sodium (Sigma, St. Louis, MO, USA; 71718), 1 mmol/L sodium pyruvate (Sigma; P5280), 15 mM HEPES (N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid, Sigma, H4034), and FBS (fetal bovine serum; 1%). The pH was set to 6.87 ± 0.02.
Astrocytes were seeded into six-well plates at a density of 5 × 105 cells per well 24 h prior to transfection. For siRNA-mediated gene knockdown, three different siRNAs targeting NDRG2 (RiboBio) were delivered using Lipofectamine RNAiMAX (Thermo Fisher, Waltham, MA, USA; 13778075) following the manufacturer’s protocol. Since the silencing efficiencies of all three siRNAs were found to be similar, one of the siRNAs was randomly selected for subsequent experiments.
NDRG2 was overexpressed by the transfection of cells with pcDNA3.1-NDRG2 plasmid DNA, which was accompanied by pCDNA3.1 as the control group. pcDNA3.1 and pcDNA3.1-NDRG2 were synthesized by PPL (Public Protein/Plasmid Library, China). The cell transfection was delivered using the X-tremeGENE HP DNA Transfection Reagent (Roche, Basel, Switzerland; 06366244001) following the manufacturer’s protocol. Plasmids with site-directed mutations were purchased from PPL (Public Protein/Plasmid Library, China). Different forms of NDRG2-overexpressing plasmids were constructed using the pcDNA3.1 vector containing Rattus norvegicus NDRG2 with a C-terminal 3 × Flag-tag (DYKDDDDK-tag). A Lys176 single-mutation plasmid and an Arg240 single-mutation plasmid were constructed by changing the relevant codon to a GCA codon encoding alanine. For the NDRG2 overexpression vector, three forms of NDRG2-overexpressing plasmids (wild-type [WT], site-directed mutagenesis of Lysine 176 to alanine [K176A], double site-directed mutagenesis of Lysine 176 to alanine and Arginine 240 to alanine [K176A + R240A]) were delivered using the X-tremeGENE HP DNA Transfection Reagent.
Total RNA was isolated from astrocytes or tissues using TRIzol reagent (Invitrogen, Waltham, MA, USA). For tissues, the animals were killed at pre-specified time points, and the brains were rapidly removed and coronally sliced into 2-mm sections in cold phosphate-buffered saline (PBS). The ipsilateral side was analyzed, and the contralateral side of the same rat or mouse was used as the control. Both sides were quickly frozen with liquid nitrogen until they cracked, after which they were ground in TRIzol. Total RNA was extracted using the miRNeasy Mini Kit (Qiagen, Hilden, Germany), and cDNA was synthesized using TransScript One-Step gDNA Removal and cDNA Synthesis SuperMix (Transgene, Strasbourg, France) according to the manufacturer’s instructions. Briefly, the reaction was performed in a final volume of 20 μL with 500 ng of total RNA, 10 μL of 2 × TS reaction mix, 1 μL of oligo(dT)15, 1 μL of gDNA remover, and 1 μL of RT/RI enzyme mix, followed by incubation at 42 °C for 30 min and at 85 °C for 5 s. RT-qPCR was conducted using TransStart Green qPCR SuperMix (Transgen, Beijing, China) in an ABI7300 qRT–PCR machine. The primers used are listed in Additional file 4: Table S3. Gene expression was normalized to that of the internal control (β-actin) using the 2−ΔΔCt method.
Western blot analysis
Total protein content was extracted from the cultured cells or tissues, and protein concentrations were determined using the Pierce BCA assay kit (Beyotime, Haimen, China; P0012S). Antibodies against GFAP (1:10,000; Abcam, Cambridge, UK; ab7260), NDRG2 (1:5000; Sigma-Aldrich; HPA002896), p-STAT3 (1:2000; CST, Naples, FL, USA; 9145P), STAT3 (1:2000; CST, 4904 T), p-Akt (1:2000; CST, 4060S), Akt (1:2000; CST, 9272S), c-Jun (phospho S63) (1:1000, CST, 91952 T), c-Jun (1:1000, CST, 9165 T), TNFα (1:500, Abcam, ab6671), HIF1α (1:1000, Sigma, SAB5200017), GAPDH (1:2000; CMC-TAG, AT0002) and β-actin (1:2000; CMC-TAG, AT0001) were used for western blot analysis. The band intensities of the target proteins were quantified by densitometry using ImageJ software (National Institutes of Health, Bethesda, MD, USA). GAPDH and β-Actin were used as the loading control. The results were derived from three independent experiments.
Nuclear and cytoplasmic protein extraction
Nuclear and cytoplasmic proteins were extracted using the Nuclear and Cytoplasmic Protein Extraction Kit (Yeasen, Shanghai, China) according to the manufacturer’s instructions. In brief, cells were washed and harvested by scraping in cold PBS. Cell pellets were obtained after centrifugation. 200 µL of reagent A containing PMSF was added per 20 µL cell precipitation. The sample was vortexed vigorously (highest setting on vortex) for 5 s, then incubated for 15 min in an ice bath. Subsequently, 10 µL of cytosolic protein extraction reagent B was added to the samples and the mixture was vortexed vigorously for 5 s, then incubated for 1 min in an ice bath. Samples were centrifuged at 12,000–16,000g for 5 min at 4 °C. The supernatant was the extracted cytoplasmic protein. 50µL of reagent C containing PMSF was added to the precipitate, then the mixture was incubated for 30 min in an ice bath. Samples were centrifuged at 12,000–16,000g for 10 min at 4 °C. The supernatant was the extracted nuclear protein, which was immediately transferred into a pre-cooled tube.
Histological and immunofluorescence staining
Rats and mice were transcardially perfused with 10% neutral formalin buffer, and the brains were dissected and post-fixed in 10% neutral formalin buffer for an additional 48 h. The fixed samples were embedded in paraffin and sectioned into 5-μm-thick sections. For immunohistochemistry, the sections were deparaffinized, rehydrated, washed with PBS several times, and treated with an UltraSensitive™ SP (mouse/rabbit) immunohistochemistry (IHC) kit (Maixin Biotechnology, Fuzhou, China; KIT-9710). The sections were incubated with rabbit anti-GFAP antibody (1:1000; Abcam, ab7260), rabbit anti-NDRG2 antibody (1:5000; Sigma-Aldrich, HPA002896), rabbit anti-TNFα antibody (1:200, Novus, St. Louis, MO, USA; NBP1-19532), and rabbit anti-c-Jun (phospho S63) antibody (1:100; Abcam, ab32385), and the nuclei were counterstained with hematoxylin. The stained sections were examined using the Aperio Digital Pathology Slide Scanner (Leica, Wetzlar, Germany) and an ImageScope imaging system (Leica). The integrated optical densities were semi-quantified using Image-pro Plus software (version 6.0; Media Cybernetics, Rockville, MD, USA). The following antibodies were used for fixed cells: rabbit anti-c-Jun (phospho S63) antibody (1:100; Abcam, ab32385), chicken anti-GFAP antibody (1:1000; Abcam, ab4674), chicken anti-MAP2 antibody (1:1000; Abcam, ab5392), Alexa Fluor 594 goat anti-rabbit immunoglobulin G (IgG) (1:200; Jackson Healthcare, Alpharetta, GA, USA; 140418), and Alexa Fluor 488 goat anti-chicken IgG (1:500; Jackson, 103545155). Nuclei were counterstained with Hoechst 33342 (1:1000; Beyotime, C1025). Sections were examined using a fluorescent microscope (Olympus, Tokyo, Japan) and the CellSens Entry imaging system (Olympus).
Molecular structure modeling and molecular docking simulation
A 3D model of NDRG2 was constructed using I-TASSER software (Iterative Threading ASSEmbly Refinement) , resulting in a high-value model (C‐score = 1.37, estimated TM‐score [template modeling score] = 0.90 ± 0.06, estimated RMSD [root-mean-square deviation] = 3.3 ± 2.3 Å). The lactate structure was downloaded from the RCSB Protein Data Bank (http://www.rcsb.org/). Based on binding-change mutations at Gly146, Asp172, Lys176, Asp180, and Arg240, lactate was docked into a 15-Å binding pocket formed by the five amino acids of NDRG2 (Additional file 1: Fig. S5). A comparative molecular docking analysis between lactate and NDRG2 was carried out using AutoDock Vina open-source software (Trott et al. 2010).
To conduct the in vivo ubiquitination assay, ‘lactate-’ or ‘lactate + ’ culture medium was generated by excluding or adding (respectively) the proteasome inhibitor MG132 (10 μM, Selleck, Berlin, Germany; S2619) from the culture during 8 h of OGD conditions prior to harvesting. Cell lysates were precleared by adding 50 μL of protein A/G immune magnetic beads (Bimake, Houston, TX, USA; B23201) and were immunoprecipitated with anti-NDRG2 antibody. Polyubiquitinated forms of NDRG2 were detected by western blot with an anti-ubiquitin antibody (1:1000, Cell Signaling Technology, A100). Astrocytes were transfected with expression vectors (i.e., different forms of Flag-tagged NDRG2 overexpressing vector [WT, K176A, K176A + R240A]); 72 h later, the medium was changed to a ‘lactate + ’ culture medium by administering MG132 for 8 h under OGD conditions prior to harvesting. Cell lysates were precleared by adding 20 μL of anti-Flag-conjugated magnetic beads (Bimake, B26101) and were subsequently immunoprecipitated with anti-NDRG2 antibody. Proteins immunoprecipitated with anti-Flag magnetic beads were detected by western blot using an anti-ubiquitin antibody.
Total RNA isolation was performed using the NEBNext Poly (A) mRNA Magnetic Isolation Module (New England Biolabs, Ipswich, MA, USA). The libraries were prepared using a KAPA Stranded RNA-Seq Library Prep Kit (Illumina, San Diego, CA, USA), and were sequenced to a mean depth of (19.7 ± 1.7) million reads using an Illumina HiSeq 4000 sequencer (Illumina) for 150 cycles. RNA-seq data were analyzed using R software (version 3.5.3; The R Project for Statistical Computing, Vienna, Austria) ballgown package, and were expressed as fragments per kilobase of transcript per million mapped fragments (FPKM). Bioinformatics analyses were performed using R and Python (version 3.4.10; Fredericksburg, VA, USA). DEGs in the two groups were identified using FPKMs and P-values (fold change > 1.5, P-value < 0.05). Hierarchical clustering was performed to extract distinguishable mRNA expression profiles from the samples. Gene ontology (GO) analysis was performed to investigate three functional domains, namely biological process (BP), cellular component (CC), and molecular function (MF). Pathway analysis was performed to functionally analyze and map the DEGs to KEGG pathways. P-values < 0.05 identified statistically significant GO terms and KEGG pathways correlated with various conditions.
ClueGO clustering analysis
The Cytoscape plugin ClueGO (https://www.cytoscape.org) was used for the functional enrichment analysis. Specifically, ClueGO was implemented to decipher functionally grouped gene ontology and pathway annotation networks. In this calculation process, a kappa coefficient was calculated to reflect the functional correlation between paths or terms based on gene overlap between pathways. Network specificity was scaled to the average.
In reporting the results, functionally similar entries are displayed in the same color. The threshold for enrichment significance was set to P < 0.05. Functionally grouped networks of enriched categories were generated for the candidate target genes. GO terms are represented as nodes, the lines between nodes reflect correlations between signaling pathways, and node size represents enrichment significance. The colors of the nodes reflect the enrichment classification of each node (i.e., which functional group it belongs to), and node pie charts represent the biological processes for these targets.
To detect concentrations of secreted TNF, cell supernatants were assayed for TNFα proteins using ELISA after removing particulates via centrifugation. Eight hours after lactate treatment under OGD conditions, the culture supernatants were harvested and stored at − 70 °C. The concentrations of TNFα were measured using ELISA kits according to the manufacturer’s instructions (R&D Systems, Minneapolis, MN, USA; RTA00). The sample results were scaled to those of the siNC group.
Quantification and statistical analysis
Results are expressed as means ± standard deviations. The sample sizes are shown in each figure. Statistical analyses were performed using GraphPad Prism software (version 7.0; San Diego, CA, USA). The parametric paired t-test was used on pairs of related samples and parametric unpaired t-test was used on pairs of independently sampled groups. Comparisons of more than two groups were conducted by one-way ANOVA with Tukey’s multiple comparison test. P values < 0.05 were considered statistically significant and are labeled with single asterisks (*P < 0.05), while P values > 0.05 are labeled ns (ns = no significance). P values < 0.01 is labeled with double asterisks (**P < 0.01), respectively. The mean of experimental groups beyond the confidence limits was considered statistically significant, but was labeled “ns” when the difference was within the confidence interval.