Recovery of motor function of chronic spinal cord injury by extracellular pyruvate kinase isoform M2 and the underlying mechanism

In our previous study, we found that pyruvate kinase isoform M2 (PKM2) was secreted from the skeletal muscle and extended axons in the cultured neuron. Indirect evidence suggested that secreted PKM2 might relate to the recovery of motor function in spinal cord injured (SCI) mice. However, in vivo direct evidence has not been obtained, showing that extracellular PKM2 improved axonal density and motor function in SCI mice. In addition, the signal pathway of extracellular PKM2 underlying the increase in axons remained unknown. Therefore, this study aimed to identify a target molecule of extracellular PKM2 in neurons and investigate the critical involvement of extracellular PKM2 in functional recovery in the chronic phase of SCI. Recombinant PKM2 infusion to the lateral ventricle recovered motor function in the chronic phase of SCI mice. The improvement of motor function was associated with axonal increase, at least of raphespinal tracts connecting to the motor neurons directly or indirectly. Target molecules of extracellular PKM2 in neurons were identified as valosin-containing protein (VCP) by the drug affinity responsive target stability method. ATPase activation of VCP mediated the PKM2-induced axonal increase and recovery of motor function in chronic SCI related to the increase in axonal density. It is a novel finding that axonal increase and motor recovery are mediated by extracellular PKM2-VCP-driven ATPase activity.

The identification of putative PKM2 direct target protein in cultured neurons using drug affinity responsive target stability (DARTS) analysis. DARTS analysis was performed as described previously 11,12 . Cortical neurons were cultured for four days. The cell lysate was prepared using mammalian protein extraction reagent (M-PER) lysis buffer (Thermo Fisher Scientific, Waltham, USA) containing a protease and phosphatase inhibitor cocktail (Thermo Fisher Scientific). Cell lysate (2 μg protein) was incubated with 50 and 500 ng/ml recombinant mouse PKM2 for 30 min at 24 °C and each of these were proteolysed with 0.5 and 5 ng/mL thermolysin (Wako, Osaka, Japan), respectively, in a reaction buffer (50 mM Tris-HCl, pH 8.0; 50 mM NaCl; 10 mM CaCl 2 ) for 30 min at 37 °C. To stop the reaction, 0.5 M ethylenediaminetetraacetic acid (pH 8.0) was added to each sample at a 1:10 ratio on ice. The samples were separated on 10% sodium dodecyl sulphatepolyacrylamide gel (SDS-PAGE). The proteins in the gels were visualized using a SilverQuest Kit (Invitrogen, Carlsbad, CA, United States). Four protein bands (indicated in Supplementary Fig. 1) were thinner in the sample treated with 500 ng/mL PKM2 compared to those of the sample treated with vehicle solution. The bands were excised from the gel, digested with trypsin, and analysed by mass spectrometry using a nano liquid chromatography-tandem mass spectrometry (LC-MS/MS) system (Japan Proteomics, Sendai, Japan).
Binding assay by immunoprecipitation 11 . Recombinant mouse PKM2 (100 pmol) and recombinant human VCP (10 pmol) were coincubated for 1 h at 25 °C. Fifty µl of Dynabeads Protein G (Thermo Fisher Scientific) were treated with 1% bovine serum albumin for blocking in 0.01% Tween phosphate buffered saline for 30 min at 4 °C with rotation. Then, the protein G was incubated with a rabbit anti-PKM2 antibody (2 µL, Cell Signaling Technology, Danvers, MA, US) or normal mouse IgG (2 µL, Santa Cruz Biotechnology, Dallas, TX, US) for 30 min at 4° with rotation. Protein G and antibodies were cross linked by 50 mM dimethyl pimelimidate dihydrochloride (Tokyo Chemical Industry Co., Ltd., Tokyo, Japan) for 1 h at 25 °C with rotation. After washing protein G, incubated recombinant proteins were added, and the mixture was incubated for 2 h at 25 °C with rotation. For the elution of immunoprecipitants, the samples were mixed with LDS sample buffer and 0.1 M glycine-HCl (pH 2.8) for 5 min at 95° and then were loaded onto 10% polyacrylamide gels and electrophoresed. After electrophoresis, PKM2 or VCP in the gel were detected by Western blotting using a rabbit monoclonal anti-PKM2 antibody (1:1000, Cell Signaling) or a mouse monoclonal anti-VCP antibody (1:2000, Abcam) as first antibodies. Secondary antibodies against rabbit IgG (1:2000, Cell Signaling) and mouse IgG (1:2000, Abcam) were used. Western blotting. Western blotting was performed using samples after proteolysis in the DARTS analysis 11 .
Lysates of the plasma membrane fraction of cultured neurons were prepared with the Mem-PER Plus Membrane Protein Extraction Kit (Thermo Fisher Scientific) after four days of in vitro culturing following the manufacturer's protocol. Samples from the DARTS reaction were loaded onto a 10% SDS-PAGE. After electrophoresis, proteins in the gel were transferred to a nitrocellulose membrane (Bio-Rad, Berkeley, CA, United States) and blocked with 0.1% T-TBS containing 5% skim milk (Wako) at 24 °C. Subsequently, the membrane was incubated with a mouse monoclonal anti-VCP antibody (1:5000 for Fig. 1, Abcam)  www.nature.com/scientificreports/  Measurement of axonal length. Evaluation of axonal length was performed as previously described 1 .
For measurement of the density of axons, the cells were treated with or without recombinant mouse PKM2 or the vehicle solution (distilled water) for four days. The neurons were fixed with 4% paraformaldehyde for 90 min and immunostained with a polyclonal antibody against microtubule-associated protein 2 (MAP2, 1:2000, Abcam, Cambridge, UK) as a neuron marker. A monoclonal antibody against phosphorylated neurofilament-H (1:250, SMI-35, Covance, Dedham, MA, USA) was used as an axonal marker. Alexa Fluor 594-conjugated goat antirabbit IgG (1:600) and Alexa Fluor 488-conjugated goat anti-mouse IgG (1:600) were used as secondary antibodies (Molecular Probes, Eugene, OR, USA). Nuclear counterstaining was performed using DAPI (1 μg/mL, Sigma-Aldrich). The fluorescence images were captured with a 20 X objective lens using a fluorescence microscope system (Cell Observer, Carl Zeiss, Tokyo, Japan). Nine to thirty-three images (

SCI surgical operation and continuous administration of PKM2.
Eight-week-old female ddY mice (SLC, Japan) were used for the SCI experiments. All mice were housed with access to food and water ad libitum and kept in a constant environment (22 ± 2 °C, 50 ± 5% humidity, 12 h light cycle starting at 07:00). The mice were anaesthetised with butorphanol tartrate (5 mg/kg, i.p., Meiji Seika Pharma Co., Ltd., Tokyo, Japan), medetomidine hydrochloride (0.75 mg/kg, i.p., Zenyaku Kogyo Co., Ltd., Tokyo, Japan) and midazolam (4 mg/ kg, i.p., Fuji Pharma Co., Ltd., Tokyo, Japan). After laminectomy, contusion injury was given by dropping a 6.5-g weight from a height of 3.5 cm onto the exposed spinal cord at the level of T13 using a stereotaxic instrument (Narishige, Tokyo, Japan), as described previously 1 . During and after surgery, the mice were placed on a heating pad to maintain body temperature. Thirty-two days after SCI surgery, the mice were divided into three groups; vehicle solution (ACSF) (n = 10), 1 ng/mL PKM2 (n = 9), and 1 ng/mL PKM2 + 100 nM CB-5083 (n = 8). Mice were placed in a stereotaxic apparatus, and the head was kept in a fixed position. The scalp was shaved, followed by a sagittal midline incision to expose the skull. A cannula (Brain Infusion Kit 3, DURECT Corporation, CA, USA) was inserted to a lateral ventricle as follows: bregma − 0.22 mm, lateral to the left + 1 mm and − 2.5 mm depth. The free end of the cannula was connected to a micro-osmotic pump (Alzet model 1004) via a 3.5 cm piece of polyvinylchloride (PVC) tubing (Alzet). The cannula was fixed to the skull with Aron Alpha A "Sankyo" (Daiichi Sankyo, Tokyo, Japan). The pump was placed into a subcutaneous pocket in the back of the mouse. The infusion rate of the micro-osmotic pump was 0.11 ml/hr. As the vehicle solution, ACSF (containing 130 mM NaCl, 24 mM NaHCO 3 , 3.5 mM KCl, 1.3 mM NaH 2 PO 4 , 2 mM CaCl 2 , 2 mM MgCl 2 ·6H 2 O, and 10 mM glucose at pH 7.4) was filled into the micro-osmotic pump and connected PVC tube. The micro-osmotic pump and tube were filled with 164 ng/mL of PKM2 and 16.4 mM CB-5083 that was dissolved in ACSF, considering that the pump efflux was 0.11 μL/hr and CSF was produced at a speed of 0.325 μL/min 13 . Thus, the final concentrations of PKM2 and CB-5083 were always approximately 1 ng/ml and 100 nM, respectively, when they were delivered to the CSF of the SCI mice. These doses are enough to induce axonal extension by PKM2, inhibiting the effect of PKM2 by CB-5083 (Fig. 2).

Identification of a target molecule of extracellular PKM2 in neurons.
Using DARTS analysis, direct binding proteins of PKM2 in neurons were comprehensively explored. The concept of this method is based on change of the structural conformation of the protein when a ligand binds to a target protein. Modified structure of the protein by binding alters the resistance against proteolysis, and it becomes harder or easier to degrade the target protein 16 . Whole-cell neuron lysates were incubated with recombinant PKM2 or vehicle. After the proteolysis reaction using thermolysin, the lysates were electrophoresed and silver stained. Four bands were thinner on the gel in 500 ng/mL PKM2 treatment than the vehicle solution-treated ( Supplementary  Fig. 1). Among those candidates, bands a, b, and c showed no reproducibility; therefore, we focused on molecular weight around 90 kDa of protein. The band was analysed by nanoLC-MS/MS, and the results indicated with a high possibility that the band was valosin-containing protein (VCP) (Score: 368, coverage 26%). To confirm VCP as the target protein of PKM2, we performed DARTS followed by western blotting. Considering that extracellular PKM2 seems to not enter inside cells, we supposed binding proteins of extracellular PKM2 was possibly located on the plasma membrane. Therefore, plasma membrane lysates of the cortical neuron were used. Incubation with recombinant PKM2 facilitated degradation of VCP (Fig. 1a,b, Supplementary Fig. 1). Direct binding of PKM2 and VCP was confirmed by co-immunoprecipitation experiment ( Supplementary Fig. 3). The mixture of recombinant PKM2 and recombinant VCP was immunoprecipitated by anti-PKM2 antibody or control normal mouse IgG. Precipitants were immunoblotted by anti-PKM2 and anti-VCP antibodies. Results indicate VCP is coprecipitated with PKM2, suggesting direct interaction of PKM2 and VCP. VCP has many functions such as ER-associated protein degeneration 17 , cell division, organelle biogenesis, nuclear envelope formation, and protein degradation via the ubiquitin-proteasome system 18 . Since those varieties Figure 3. Effects of extracellular PKM2 i.c.v. infusion on locomotor function of mice with contusive spinal cord injury (SCI). Thirty-two days after SCI, recombinant PKM2 or vehicle solution was i.c.v. injected using a brain infusion cannula and an osmotic pump for 28 days. PKM2 was co-infused with or without CB-5083.
Extracellular PKM2-VCP signalling increases axonal density and improves motor function in chronic spinal cord-injured mice. We investigated the effect of extracellular PKM2-VCP signalling on functional recovery in the chronic phase of SCI mice. Thirty-two days after injury, continuous administration of recombinant PKM2 or vehicle solution, artificial cerebrospinal fluid (ACSF) into the lateral ventricle, was started using a micro-osmotic pump for 28 days. The concentrations of PKM2 and CB-5083 were maintained at approximately 1 ng/mL and 100 nM in the CSF, respectively, during the administration period, which are effective doses in culture cell experiments (Fig. 2d). The hindlimb motor functions were evaluated by Basso Mouse Scale BMS (Fig. 3a), TMS (Fig. 3b), and vertical cage scale (Fig. 3c) (Fig. 3c). However, simultaneous infusion of CB-5083 significantly reduced the score.
Wet weights of tibial anterior and gastrocnemius were significantly reduced in SCI mice compared with shamoperated mice (Fig. 3d,e). PKM2 i.c.v. infusion significantly enhanced the weight of the anterior tibial muscle and PKM2/CB-5083 i.c.v. co-infusion did not increase muscle weight. No significant changes in body weights were observed among the three groups during the experimental period ( Supplementary Fig. 5).
After the behavioural tests, spinal cord and brain tissues were isolated. Slices of the spinal cord that included the lesion area were prepared and immunostained with the raphespinal tract marker 5-HT (Fig. 4a). The raphespinal tract is serotonergic and one of the major descending tracts that modulates the excitability of motor neurons. To decide the lesion area, the GFAP-positive glial scar area was stained. PKM2 i.c.v. infusion increased the density of 5-HT-positive axons at the caudal lesion site in mice with chronic SCI (Fig. 4a,b). CB-5083 infusion completely inhibited PKM2-induced increase in the density of axons. Areas of glial scar were not changed by PKM2 or PKM2/CB-5083 treatment (Fig. 4c).
A retrograde transsynaptic tracer, WGA, was injected in the sciatic nerves of the right and left hindlimbs. We counted WGA-positive neuronal cell bodies in the entire area of the raphe nucleus, including the RMg, Rob and RPa. The number of WGA-labelled neurons was significantly increased in PKM2-treated SCI mice compared to vehicle-treated SCI mice (Fig. 5b,c). While CB-5083 treatment significantly decreased the numbers of WGApositive neurons in the raphe nucleus. Hematoxylin and eosin staining of brain slices indicates no obvious cell losses in the motor area and raphe nucleus area in three groups ( Supplementary Fig. 6).

Discussion
This study is the first report to indicate that extracellular PKM2 infusion to the lateral ventricle recovers motor function in the chronic phase of SCI mice. The improvement of motor function was associated with axonal increase, at least of raphespinal tracts connecting to motor neurons directly or indirectly. Since target molecules of extracellular PKM2 in neurons have not been determined, a comprehensive analysis was performed. VCP was identified as a direct target protein of PKM2, particularly its ATPase activity related to induction of axonal extension. The VCP-mediated axonal extension is a novel finding, although the downstream pathway of the upregulation of ATPase remains unknown. Additionally, the binding modes of PKM2 and VCP have not been identified yet. Various cofactors directly interacting with VCP have been identified by proteome analysis [19][20][21] . Although cofactors for VCP usually interact with VCP interaction motif 22 , ubiquitin regulatory X domain 23 , and VCP binding motif 24 , such domains are contained in the PKM2 sequence. Further study for identifying the binding modes of PKM2 and VCP in detail are expected.
Generally known VCP function covers a variety of phenomena, such as protein degradation of ER-related 25-27 , mitochondria-related 28,29 , and ribosome-related 30 , controlling autophagy 18 , molecular chaperone 31,32 and so on. VCP function relating to axonal formation has hardly been investigated. Only one study indicated that VCP bound to the slow Wallerian degeneration protein (Wld s ). However, VCP is not involved in the Wld s -mediated axon protection 33 .
VCP protein expresses in the entire region of the brain and spinal cord (the human protein atlas: https ://www. prote inatl as.org/). VCP expression and PKM2 expression in the spinal cord of chronic (eight weeks post-injury) SCI mice are not decreased compared to that in an uninjured spinal cord 34 . As we previously found, extracellular PKM2 is secreted from skeletal muscle and transferred to the central nervous system by systemic circulation 1 .

Scientific Reports
| (2020) 10:19475 | https://doi.org/10.1038/s41598-020-76629-7 www.nature.com/scientificreports/ Although we detected an increase in the density of raphespinal tracts in this study, other descending tracts and interneurons might also be extended by extracellular PKM2 stimulation. Skeletal muscle atrophy is observed to be severe in SCI mice (Fig. 3d,e), and it partially recovered with PKM2 i.c.v. infusion. It is unlikely that PKM2 infused into lateral ventricle affected the hindlimb muscles. If anything, improved hindlimb movement might increase muscle mass. In summary, extracellular PKM2 increases the density of axons and motor function when applied to the brain in the chronic phase of SCI mice. The increase in axons and motor recovery are mediated by extracellular PKM2-VCP-driven ATPase activity.

Data availability
All data needed to evaluate the conclusions in the paper are present in the paper or the Supplementary Materials. www.nature.com/scientificreports/