No statistical methods were used to predetermine sample size. Animals used were randomly assigned to groups before experiments and the investigators were not blinded to allocation during experiments and outcome assessment.
All animal care and experimentation were ethically performed according to procedures approved by the Institutional Animal Care and Use Committee at Fudan University. Mice were housed 3–5 per cage (except in metabolic chamber in which mice were singly caged or preparing for core temperature detection) in a 12 h light/12 h dark cycle with ad libitum access to regular chow and water, and maintained at an ambient temperature of 23℃ and a relative humidity of 50%. We used 6 ~ 8-week-old WT male C57BL/6J mice (SLAC company), Wistar rats (SLAC company), Vglut2-ires-Cre (Jackson Lab), Vgat-ires-Cre (Jackson Lab). All mouse lines are in a WT (C57BL/6J) background.
Pdxk flox/flox mice were generated by CRISPR/Cas 9-mediated genome engineering. Exon2 of Pdxk-201 (ENSMUST00000041616.14) transcript, containing 55 bp coding sequence, is recommended as the knockout region based on the structure of pdxk. In brief, sgRNA was transcribed in vitro, donor vector was constructed. Cas9, sgRNA and donor were microinjected into the zygotes of C57BL/6JGpt mice. Zygotes were transplanted to obtain positive F0 mice which were confirmed by PCR and sequencing. A stable F1 generation mouse model was obtained by mating positive F0 generation mice with C57BL/6JGpt mice.
P5756 and its derivatives were kindly provided from Biao Yu’s group by totally synthesized. Pyridoxal (PL, 322481) and pyridoxal 5′-phosphate (PLP, 303712) were obtained from J&K Scientific. Pyridoxamine (PM, P9380), pyridoxine (PN, P9755), 4-pyridoxic acid (4-PA, P9630), pyridoxal-(methyl-d3) (D3-PL, 705187), celastrol (cela, C0869), ethacrynic acid (EA, SML1083) and 2,3,5-triphenyltetrazolium chloride (TTC, 8877) were obtained from Sigma Aldrich. α-angelica lactone (HY-N0548) was obtained from MCE. Glutathione (GSH, 101814) was obtained from MP Biomedicals.
AAV2/9-hSyn-DIO-jGCaMP7s (S0590-9), AAV2/9-hSyn-eGFP-WPRE-pA (S0237-9-H20), and AAV2/9-hSyn-Cre-eGFP-WPRE-pA (S0230-9-H50) were purchased from TaiTool. All virus were diluted with PBS to a final concentration of 5×1012 per mL before stereotaxic delivery into the mouse brain.
Body temperature recording
The rectal temperatures were recorded by animal temperature measuring apparatus (FT3400, Kew Basis) (Fig. 3C,3G and 3H; fig.S3C-E) every 30 minutes and the viscera temperature was recorded by IR digital thermographic camera (FLIR T430sc) every 5 minutes (Fig. 5F, 5G). The other internal temperatures were measured and recorded by Anilogger core temperature monitoring system every 15 minutes. The mice were adapted for 4 days before test. All temperature recording experiments were performed under ambient temperature at 24°C.
Middle cerebral artery occlusion (MCAO) and infarct volume evaluation
Adult male wistar rats were anesthetized with isoflurane. Rats were subjected to a right side MCAO12,57 and heating lamps were utilized to maintain rectal temperature at 36.5 to 37.5°C. Reperfusion was achieved by the withdrawal of the filament after 2 h occlusion. In all ischemia models, solvent or P57 was immediately injected intraperitoneally once priming reperfusion.
Infarct volume was evaluated by TTC staining12. Briefly, rats were sacrificed at 24 h following reperfusion and their brain were quickly removed and chilled. Six coronal brain slices with a 2-mm thickness were cut for the treatment with 1% TTC solution at 37°C for 20 min, then fixed in 10% formalin solution. The images of the coronal slices were taken with a digital camera and analyzed to quantify the infarct area with Adobe Photoshop software. Infarct volumes were determined by the integration of the infarct area of each slice and the distances between them.
Affinity chromatography experiment and iTRAQ analysis
500 µL brain lysate was pre-incubated 1 hour with DMSO or 20 µM P57 at 4℃ with rotation, and then incubated 1 hour with DMSO or 2 µM P57-Bio at 4°C with rotation. The samples were added to 20 µL pre-washed high-capacity streptavidin agarose beads and incubated 1 hour at 4°C with rotation. The beads were pelleted and washed 4 times with lysis buffer. The wash buffer was aspirated completely from the beads. Then 50 µL 1× sample buffer was added and incubated for 5 min on a 95°C-heat block. The sample buffer containing proteins off the beads was carefully transferred to a new 1.5 mL Eppendorf tube, and was loaded onto 15% SDS-PAGE.
After silver staining, the gel was washed with deionized water and gel lanes were excised using surgical scissors into pieces. Then gel pieces were reduced with 5 mM dithiothreitol (DTT) at 56°C for 1 h. Alkylation was followed by incubation with 15 mM iodoacetamide (IAA) in dark for 45min. After removed the supernatant, the samples were incubated with trypsin covering at 37°C for overnight. Transferred the supernatant into a clean tube. Squeezing the gels used 100% ACN (v/v) to ensure all the peptides could be extracted. Combined all the peptides and dried them by Speed-Vac. Then the peptides were dissolved in 0.1% TFA and further desalted with C18 Zip-tip.
LC–MS/MS analysis was conducted using an EASY-nLC 1000 HPLC system (Thermo Fisher Scientific) and an Orbitrap Fusion (Thermo Fisher Scientific). The sample was reconstituted using buffer A (0.1% formic acid, 2% ACN) and then loaded onto C18 reversed phase capillary analytical column (3 µm particle size, 90 Å pore size) by the autosampler which connected to an EASY-nLC 1000 HPLC system. The peptides were eluted at constant flow rate of 300 nL/min by increasing buffer B (0.1% formic acid, 90% ACN) from 8–32% over 58 min, then 48% in 6 min followed a steep increase to 80% in 2 min.
The mass spectrometric analysis was carried out in a data dependent acquisition (DDA) mode with a cycle time of 3 s. The peptides with a range of m/z 350–1300 were detected by an Orbitrap mass analyzer. And the resolution was set as 120,000 at m/z 200. Automatic gain control (AGC) target and maximum ion injection time (IT) were 5.0 × 105 and 50 ms, respectively. The isolated precursor ions were subjected to fragmentation via higher-energy collisional dissociation (HCD) with a collision energy of 32%, and analyzed by ion trap analyzer. The dynamic exclusion was set as 60 s, and the charge inclusion state was set to 2 ~ 6+.
The raw data were analyzed by Mascot search engine (v2.3.0) against the UniProt Mus musculus database (downloaded 2013). The parameters were set as following: Carbamidomethylation (C) and acetylation (protein N-terminal) were set as fixed modifications and oxidation (M) as variable modification. Trypsin was specific enzyme and allowed up to 2 missed cleavage. Mass tolerances were set to ± 10 ppm for precursor ions, ± 0.5 Da for fragment ions.
Whole brain protein lysate was extracted using buffer containing 20 mM Tris-HCl (PH = 7.4), 100 mM KCl, 1% Triton-X-100 and complete protease inhibitors. The following antibodies were used for immunoblotting: PDXK (1:1000; 15309, Proteintech), GSTP1 (1:1000, 15902, Proteintech) and GAP43 (1:1000, 16971, Proteintech). Membranes were incubated overnight with primary antibody at 4°C, followed by 3X washes in PBST (PBS with 0.1% Triton-X-100, vol/vol) before incubation in secondary antibodies for 1 hour at room temperature. Images were captured using ChemiScope system (Clinx Science Instruments Co. Ltd).
Protein Expression and Purification
The coding sequences of wild-type full-length human PDXK and the site-directed mutant proteins (T47V, Y84E, D235A) were respectively subcloned into pET-28a vector resulting in the addition of an N-terminal His tag and thrombin cleavage site. The plasmids were transformed into Escherichia coli OverExpress C43(DE3) strain (2nd Lab, cat# EC1040) and induced with 0.4 mM IPTG at 18°C overnight. The fusion proteins were purified by HisTrap FF column (GE healthcare life sciences) after lysis of cells by sonication, followed by size-exclusion chromatography using a Superdex 75 column (GE healthcare life sciences). Once purified, the proteins were subsequently frozen in aliquots and stored in buffer (25 mM HEPES-NaOH, pH 7.5, 200 mM NaCl and 5% glycerol).
Microscale Thermophoresis (MST) Assay
A Monolith NT. Automated from NanoTemper Technologies was used for MST binding assay. Wild-type or mutant PDXK proteins were fluorescently labeled with the RED-tris-NTA (NanoTemper, cat# MO-L008) according to the manufacturer’s instructions. All affinity measurements were performed in PBS buffer mixed with 0.05% Pluronic F-127 (Invitrogen, cat# P6866). Indicated compounds, arrayed at different concentrations, were incubated with proteins for 30 min before applied to Monolith NT standard treated capillaries. Thermophoresis was then determined at 25°C with 15–20% excitation power and middle MST power. The data analysis was performed using the analysis software (MO. Affinity Analysis) after the measurement and plotted by GraphPad Prism 8.
Surface Plasmon Resonance (SPR) Assay
SPR experiments were performed on Biacore T200 instrument (GE healthcare). PDXK protein was covalently immobilized onto CM5 chip as standard procedure. The chip was equilibrated overnight first with HBS buffer (10 mM HEPES (PH = 7.4), 150 mM NaCl, 0.05% (v/v) surfactant P20 and 0.2% dimethyl sulfoxide). P57 was gradiently diluted with HBS buffer and then injected at 30 µL/min for 120 s contact time followed by dissociation for 300 s. Sensorgrams were fitted to the Langmuir binding equation for a 1:1 interaction model using Biacore T200 Evaluation Software v3.0 to determine kinetic parameters and equilibrium dissociation constants. Each experiment was performed at least three times.
Determination of Kinetic Constants
All PDXK kinetic measurements were carried out at 37℃ with 125 nM protein in 384-well plates (Corning, cat# 3702). Initial velocity studies for the conversion of PL to PLP were performed at 380 nm in an Agilant 8454 spectrophotometer (Thermo Fisher) in 100 mM sodium HEPES buffer, pH 7.4. The concentration of PL (J&K, cat# 322481) and Mg-ATP (Sigma, cat# A2383) were fixed at 2 mM and 1 mM respectively when titrating another substrate. Values for Km and Kcat were determined from Michaelis-Menten by GraphPad Prism 8.
Docking studies were performed with methods as previously published (https://doi.org/10.1016/j.ejmech.2019.111767) using Schrödinger suite (version 2009), which includes all of the programs described below. P57 was prepared through LigPrep panel integrated to generate all low-energy stereoisomers and possible ionization states in the pH range of 7.0 ± 2.0 with Epik. The crystal structure of PDXK in complex with PLP and ATP (PDB code 3KEU) was prepared using the Protein Preparation Wizard with default parameters. The ligand and protein were energy- minimized using OPLS-2005 force field. A PLP-centered receptor grid was generated with the program Glide. Then docking study was performed using Glide in extra-precision (XP) mode with enhanced planarity of conjugated pi groups, and strain correction terms applied.
For the rotarod test, the mice were trained 2 days before test. On day 3, mice were placed on an accelerating rotarod cylinder, and the latency time of the animals was measured. The speed was increased from 5 to 40 rpm within 240 seconds. A trial ended if the animal fell off the rungs or gripped the device and spun around for 2 consecutive revolutions without attempting to walk on the rungs. Motor test data are presented as mean of latency time on the rotarod.
In the open-field test, each mouse was gently placed in the center of the box after intraperitoneal injection of P57, PL or solvent for 40 min respectively. Total distance travelled was recorded using the Ethovision XT video tracking software system (Noldus Information Technologies, Leesburg, VA, USA).
Metabolic Studies (CLAMS)
6 ~ 8-week-old C5BL/6J male mice were maintained individually in a metabolism chamber (Comprehensive Lab Animal Monitoring System, CLAMS) with free access to food and water for 72 hours. Mice were housed for 24 hours for adaption. On day 2, mice were intraperitoneally injected with solvent or P57 (12.5 mg/kg or 25 mg/kg). Metabolic parameters including O2 consumption, CO2 production, respiratory exchange ratio (RER), energy expenditure and total locomotor activity were recorded at 24 min intervals in a standard light-dark cycle (light 7:00–19:00 and dark 19:00–7:00) at 25°C. Respiratory quotient is the ratio of carbon dioxide production to oxygen consumption (VO2). RER = VCO2/VO2. Energy expenditure was calculated as the product of the calorific value of oxygen (3.815*VO2 + 1.232*VCO2)58.
Measurement of PL level in hypothalamus
Pyridoxine (PN), pyridoxamine (PM), pyridoxal (PL), 4-pyridoxic acid (4-PA), and pyridoxal 5′-phosphate (PLP) were analyzed on an Agilent 1260 Infinity liquid chromatography (Agilent, CA, USA) coupled with an AB SCIEX QTrap 6500 Mass Spectrometer (AB SCIEX, MA, USA) according to a method reported by Midttun et al. previously59. Pyridoxal-(methyl-d3) (PL-d3) was used as the internal standard to tract the extraction efficacy. The accuracy of each analyte was validated to be of more than 90% with the RSD of less than 10%.
Acute slice preparation and electrophysiology
Mouse brain slices were prepared according to previously published procedures 60. Mouse was deeply anesthetized by isoflurane, and followed by transcardial perfusion with ice-cold, oxygenated (95% O2 / 5% CO2) artificial cerebrospinal fluid (ACSF) containing (in mM): 127 NaCl, 2.5 KCl, 25 NaHCO3, 1.25 NaH2PO4, 2.0 CaCl2, 1.0 MgCl2, and 25 Glucose (osmolarity ~ 310 mOsm/L). After perfusion and decapitation, mouse brain was removed and immersed in ice-cold and oxygenated ACSF. Tissue was blocked and supported by a block of 4% agar, and transferred to a slicing chamber containing ice-cold and oxygenated ACSF. Coronal slices of 250-µm-thick were prepared on a vibratome (Series1000, Tissue Sectioning System, Natural Genetic Ltd., USA), in rostral/caudal direction. Slices were transferred to a chamber containing constantly oxygenated ACSF, and incubated at around 34 ˚C for 20 min before recording.
Brain slices were transferred to a recording chamber perfused with oxygenated ACSF at a flow rate of 1.5-2 ml/min, and temperature was maintained at ~ 30°C during recording by feedback in-line heater (TC-324C; Warner Instruments, Hamden, CT). MPA GABA and non-GABA neurons were visualized in slices using IR/DIC microscopy, and identified based on tdTomato signal in Vgat-ires-Cre; Ai14 mice. Cell-attached recordings were established with glass pipettes (3–5 MΩ) containing the following (in mM): 135 K-gluconate, 4 KCl, 10 HEPES, 10 Na-phosphocreatine, 4 MgATP, 0.4 Na2GTP, and 1 EGTA (with pH 7.2–7.3, and osmolarity ~ 295 mOsm/L). P57 (2 µM) or PL (10 µM) were applied by bath perfusion. Recordings were made using 700B amplifier and Digidata 1440A interface (Axon Instruments, Union City, CA). Signals were filtered at 4 kHz, sampled at 10 kHz, and analyzed using Clampex 10.7 (Molecular Devices).
Tissue processing, immunohistochemistry and imaging
Mice were deeply anesthetized with isoflurane, and perfused transcardially with 4% paraformaldehyde (PFA) in 0.1 M phosphate buffered saline (PBS). Brains were removed and post-fixed for 24 hours in 4% PFA at 4°C, followed by dehydrating with 20% and 30% sucrose for 24h in sequence. Dehydrated brains were embedded in OCT, and cut at 40 µm using a cryostat (CM1950, Leica). Tissues were chosen and pretreated in 0.2% Triton-X100 for an hour at room temperature (RT), then blocked with 0.05% Triton-X100, 10% bovine serum albumin (BSA) in PBS for one hour at RT and rinsed in PBS. Tissues were transferred into primary antibody solution (Rabbit anti-cFos, 1:1000, Cell Signaling) in PBS with 0.2% Triton-X100 and incubated for 24 hours at 4°C. Tissues were rinsed in PBS for three times, and incubated with secondary antibody solution (Goat anti-Rabbit 647, 1:800, Life Technologies) in PBS for 2 hours at RT, then rinsed with PBS for three times and mounted onto slides, dried and covered under glycerol: TBS (3:1) with Hoechst 33342 (1:1000, ThermoFisher Scientific). Sections were imaged with an Olympus VS120 slide scanning microscope. Confocal images were acquired with a Nikon A1 confocal laser scanning microscope with an X25 objective. Images were analyzed in ImageJ (FIJI).
In vivo fiber photometry calcium signal recording
For in vivo fiber photometry recording, AAV-hSyn-DIO-jGCaMP7s viruses were injected into MPA of Vgat-ires-Cre and Vglut2-ires-Cre mice with a microsyringe pump controller (NanoJect III, Drummond Scientific Company). Three weeks after virus injection, an optical fiber (400 µm, 0.57 NA, MFC_400/430, Doric Lens) was implanted into MPA. During optical fiber implantation, we monitored the change of fluorescence intensity in real-time with photometry recording system (Doric Lens). GCamp7s was excited with sinusoidally modulated light from laser diode modules at 465 nm (211 Hz) and 405 nm (531 Hz). Signals emitted from GCamp7s and its isosbestic control channel were acquired using a photoreceiver (H10722-20; Hamamatsu Photonics), digitized at 500 Hz, and then recorded by Doric Neuroscience Studio Software. The final depth of implantation was determined by the cessation of further increases of fluorescence intensity, and optical fiber was secured with dental cement. Mice were housed for at least 1 week for recovery before recording. During recording, mouse was placed in its home cage with an optical fiber patch cord connected to the implanted optical fiber. Changes in fluorescence with time were calculated by smoothing signals from the isosbestic control channel. Relative fluorescence intensity was defined as the signal intensity change relative to the first 5-min baseline recording before i.p. injection. Mouse core body temperature was monitored every 5 min with a thermal camera (FLIR T430sc) during recording.
Raw data quality (Fastq format) was assessed and poor-quality reads were removed using fastp (version 0.19.5) program61 with the following parameters: “-n 15 –q 20 –u 50 –w 16”. Cleaned data were then mapped to reference genome (mm10) using the default parameters of Cell Ranger v6.0.2 (10X Genomics). Since both un-spliced pre-mRNA and mature mRNA were captured in snRNA-seq strategy, we set the parameter “--include-introns” in Cell Ranger pipeline, which enables us to counts these unspliced reads.
Seurat (version 4.0)62 was used to process and cluster cells. For this analysis, three S4 objects, one for each group (P57, PL, and control) were created using the Read10X and CreateSeuratObject functions with the parameter min.cell = 3 to filter genes expressed in fewer than 3 cells. The percent of ribosomal protein transcripts (percent.RP), the percent of mitochondrial transcripts (percent.MT), the number of genes detected per cell (nFeature_RNA) and the percent of erythrocyte transcripts (percent.HB) were used to assess cell quality. We removed the low-quality cells by setting cut-off: 200 < nFeature_RNA < 6000, percent.MT < 10, percent.HB < 5, percent.RP < 3. To eliminate the effects of different sequencing depth, we normalized the data using NormalizeData function with the following parameters: normalization.method = "LogNormalize", scale.factor = 10000. Then, using the “vst” method to find 2000 high variable genes (HVGs) by FindVariableFeatures function, these HVGs would be the features in the principal component analysis (PCA). Before PCA analysis (RunPCA function), the data was scaled using the ScaleData function. Clustering was done by using the FindNeighbors function, using the top 15 principal components (PCs) and the FindClusters function.
Cell type annotation was performed with SingleR (version 1.6.1)63. If a cluster co-expresses markers of two cell types, then the cluster likely contains double-droplets. Based on this, we excluded total 965 cells using the default parameters of the R package DoubletFinder (version 2.0) from three groups (P57, PL, and control)64. For visualization and analysis and to eliminate batch effects, the three groups were integrated as a Seurat object using Harmony (version 0.1.0)65. Then, the integrated object was clustered using FindClusters function with the resolution = 0.4. In addition, Glutamatergic neurons and GABAergic neurons were extracted and clustered to 39 subpopulations by setting the resolution = 1.5. We performed FindAllMarkers function to identify the marker gene of each cluster by setting cut-off: min.pct = 0.1, logfc.threshold = 0.25, test.use = "wilcox". The DEGs between different groups was calculated by using FindMarker function, such as P57_vs_Control: FindMarkers (object, ident.1 = “P57”, ident.2 = "control").
To explore the potential function of marker genes in different cell populations or DEGs in different groups, we used R package clusterProfiler (version 4.0.5) to perform KEGG Function enrichment analysis. The org.Mm.eg.db package was used to set the background gene list as all genes detected in sequencing. The Benjamin-Hochberg FDR method was used to adjust the right tail P-value (two-sided binomial statistical test).