Spexin Enhances Bowel Movement through Activating L-type Voltage-dependent Calcium Channel via Galanin Receptor 2 in Mice

A novel neuropeptide spexin was found to be broadly expressed in various endocrine and nervous tissues while little is known about its functions. This study investigated the role of spexin in bowel movement and the underlying mechanisms. In functional constipation (FC) patients, serum spexin levels were significantly decreased. Consistently, in starved mice, the mRNA of spexin was significantly decreased in intestine and colon. Spexin injection increased the velocity of carbon powder propulsion in small intestine and decreased the glass beads expulsion time in distal colon in mice. Further, spexin dose-dependently stimulated the intestinal/colonic smooth muscle contraction. Galanin receptor 2 (GALR2) antagonist M871, but not Galanin receptor 3 (GALR3) antagonist SNAP37899, effectively suppressed the stimulatory effects of spexin on intestinal/colonic smooth muscle contraction, which could be eliminated by extracellular [Ca2+] removal and L-type voltage-dependentCa2+ channel (VDCC) inhibitor nifedipine. Besides, spexin dramatically increased the [Ca2+]i in isolated colonic smooth muscle cells. These data indicate that spexin can act on GALR2 receptor to regulate bowel motility by activating L-type VDCC. Our findings provide evidence for important physiological roles of spexin in GI functions. Selective action on spexin pathway might have therapeutic effects on GI diseases with motility disorders.


Supplementary
. The mRNA levels of galanin and galanin receptors in intestine and colon of mice under the starvation condition. Mice were starved for 24 hours and then the total RNA of proximal colon, distal colon, jejunum and ileum were isolated, cDNA was synthesized and quantitative real-time PCR for galanin (A), GALR1 (B), GALR2 (C) and GALR3 (D) were conducted. The β-actin (E) showed no significant difference between groups as an internal control. Each group contained 8 mice. Statistical differences between individual groups were evaluated using One way ANOVA. *, P＜0.05 compared with paired saline-treated controls.

Homology Modeling
The mouse spexin (mSPX) is a 14AA peptidewhich share the identical sequence (NWTPQAMLYLKGAQ) with human spexin (hSPX) and goldfish spexin (gSPX). The solution structure of gSPX was studied by nuclear magnetic resonance (NMR), which showed that there was an α-helix spanning from Gln5 to Gln14 1 . Although there was no NMR models deposited into the public database, we built the initial structure of mSPX by according to the structural features of gSPX model with Protein Builder module in MOE 2 .The 3D model of mSPX was amidated at the C-terminus. It was further minimized with Amber12:EHT forcefield and R-Field solvation model. The secondary structures (SS) of the first four residues in mSPX initial model are random coil, and the remaining are majorly α-helix.

MD Simulation
The initial structure of mSPX was further relaxed in all-atoms, explicit water,10 ns MD simulation with ff12SB forcefield. Initially, it was solvated into a periodic boundary, cubic, and TIP3P 3 explicit water box with a 12 Å buffer distance by LEaP module in Amber Tools 14 4 .The charge of whole system was neutralized by adding counter ions.
The prepared system was minimized and equilibrated by sanderin three stages: (1) heating from 100 K to 300 K in 20 ps; (2) adjusting solvent density to 1 g/mL in 20 ps; (3) further equilibrating in 200 ps with constant pressure and constant temperature (NPT). Then, 10 ns, and NPT production simulation was carried out by CUDA-accelerated PMEMD 5 . A 2 fs time step was used and bonds involving hydrogen were constrained by SHAKE algorithm 6 for all equilibration and production stages.

Trajectory Analysis
The trajectories during the production stage were analyzed by cpptraj 7 . To calculate the RMSD of protein or spexin, the C α was used for structural superposition. The RMSD value was calculated by comparing with the initial and end structures for all snapshots during the production stage. Snapshots captured from 10 ps interval were clustered on C α RMSD to generate clusters using average-linkage, stopping when either 3 clusters are reached or minimum distance between clusters is 4.0. The representative conformation from the biggest cluster was used for next step.

Template Choosing
The mouse GALR2/3 protein sequences (UniProtID,O88854 and O88853) were retrieved from UniProt and GALR1/2/3 belong to the same GPCR subfamily A; (2) the resolved OPRL1 structure owned reasonable aligned sequence length (254 and 278) and the best sequence identity (31% and 28%) with GALR2/3. Taken all above, the crystal structure of OPRL1 was chosen as template in building homology models for mouse GALR2/3.

Sequence Alignment
The pdb file was downloaded and imported into MOE, only chain A and the bound ligand were kept and others were removed. The mouse GALR2/3 sequences were also imported and aligned with 4EA3 chain A in presence of GPCR constrains.

Homology Modeling
The models of mouse GALR2/3 were built by following standard homology modeling proceduresin MOE. The C-terminal and N-terminal modeling were disabled. Fifty intermediated models were generated in the process of mainchain and sidechain sampling. Medium refinement was chosen for both intermediates and final model refinement with the Amber12:EHT forcefield and R-field solvation model. The hydrogens were added by Protonate3D 9 before refining the final model. The final model was chosen based on the Generalized Born/Volume Integral (GB/VI) electrostatic solvation energy 10 .

MD Simulation
To further refine the model and remove unreasonable contacts, we carried out 10 ns MD simulation for mouse GALR2/3 in the cell membrane environment, built by Membrane Builder in CHARMM-GUI 11 .
Firstly, the homology model of mouse GALR2/3 in PDB format were uploaded to the CHARMM-GUI website. Secondly, the whole protein was aligned by the first principal axis along Z. The heterogeneous lipids, POPC and POPE, were added into upper leaflet and lower leaflet. The 0.15 M KCl was also included to neutralizing the charge of whole system with Monte-Carlo placing method. Finally, the prepared membrane system was transformed for MD simulation in Amber 14. The prepared system was generated by LEaP in AMBER 14. The ff12SB and Lipid14 12 were chosen as forcefield for protein and lipids, respectively.
To equilibrate and simulate the mouse GALR2/3 membrane protein systems, there were 4 stages: (1) heating to 100 K in 5 ps; (3) heating system from 100 K to 303 K in 100 ps; (4) equilibrating system by repeating 10 times hold protocol, 500 ps for each time; (5) 10 ns, NPT production simulation. The 2fs time step and SHAKE algothrim 6 were used in all equilibration and production stages.

Trajectory Analysis
The trajectory analysis method was same as the one used in building mSPX model. The representative conformations of mouse GALR2/3 from the biggest cluster was used for next step.

Building the Mouse GLAR/2/3-Spexin Complex Models
Defining the Binding Site complexed with OPRL1 in the template, was used as reference ligand to define the binding site for molecular docking. However, compared with mSPX, it was too small to represent a proper binding site for peptides. After searching the PDB database, we found a structure that the CXCR4 chemokine GPCR complexed with 16 AA long peptide CVX15 (PDB ID, 3OE0) 13 , where the ligand was of a close size with mSPX. By superposing this receptor with mouse GALR2/3 models, CVX15 was used as reference ligand to define the possible binding site of mSPX.

Flexible Docking
The flexible docking were carried out in MOE with Dock module. The mSPX was defined as ligand and the mGALR2/3 was defined as receptor. The mGPX was kept rigid, proxy triangle was used as placement method and London dG was used for scoring at the first stage. The docking poses were further refined with molecular mechanics for both mSPX and receptor residue 8 Å around the mSPX.
The final poses were scored by GBVI/WSA dG scoring function. In both the placement and refinement stages, at most 30 poses were retained.

Analyzing the Docking Results
The docking results were analyzed based on the binding free energy, ligand conformational energy, as well as the important contacts from prior knowledge. The best pose was used for analyze the important interactions between mSPX and mGALR2/3.

The Mouse Spexin Model
In a 10 ns MD simulation, the overall conformation of mSPX in the water solution was stably evolved.
Compared with the initial structure, the C α RMSD of mSPX residues, except the flexible N-and Cterminus residues, were between 1 and 3 Å (Fig. S2A).The representative conformation of mSPX showed distinct secondary structure features: (1) the N-terminal residues, Asn1, Trp2, and Thr3 were coiled randomly; (2) the residues at the middle, Pro4, Gln5, Ala6, Met7, Leu8, Tyr9, and Leu10 were in alpha helix; (3) the remaining residues at the C-terminal, Lys11, Gly12, Ala13 and Gln14 were in random coil (Fig. S2B). It was worth noting that, although in the same secondary structure, the N-terminal residues are more rigid than the C-terminal residues, with the backbone C α root mean structural fluctuation (RMSF) less than 1.5 Å as compared to 2 to 4 Å (Fig. S2C).

The Mouse GALR2/3 Model
The mGALR2/3 is belong to the G protein-coupled receptors (GPCR) superfamily and class A subfamily. To align the sequence of mGALR2/3 with the template 4EA3 chain A, the seven transmembrane (TM) domains of both query and template sequences were annotated firstly by comparing with a database of GPCR in MOE. The whole sequences were aligned with emphasizing the importance of TM residues (Fig. S3). From the alignment results, the disulfide bonds were predicted: C98 and C174 in mGALR2, C95 and C172 in mGALR3. Those disulfide bonds were kept in homology modeling to ensure the intact structures. Due to lack of structural information in template, the N-terminal outgaps (M1-G19 in mGALR2, M1-P12 in mGALR3) and the C-terminal outgaps (R315-C371 in mGALR2, L305-Q370) were omitted in homology modeling.
The conformation of mGALR2/3 within membrane environment was stable in 10 ns MD simulation, with C α RMSD around 2 Å (Fig. S4A). The overall structure of representative conformation of mGALR2/3 show distinct GPCR structural features: the 7 TM helixes crossed the cell membrane and were linked with several loops in the exterior cell (EC) and inner cell (IC) (Fig. S4B). The binding cavity was also clear shown. The EC loop linking TM4 and TM5, the IC loop linking TM5 and TM6, and the EC loop linking TM6 and TM7, were of the most flexibility (Fig. S4C).Those results suggested that the binding cavity for ligands could be flexible and the induce-fit binding may happen.