Zn(II) can mediate self-association of the extracellular C-terminal domain of CD147

CD147 (cluster of differentiation 147), also known as basigin or extracellularmatrixmetalloproteinase inducer (EMMPRIN), is a cell-surface type I transmembraneglycoprotein expressed at different levels in various cells and tissues, especially at high levels in tumor cells (Grass and Toole, 2015). CD147 plays important roles in multiple physiological processes, such as spermatogenesis, neural network formation, T-cell activation and in the progression of several diseases including tumor metastasis rheumatoid arthritis (RA) atherosclerosis malaria and HIV infection (Muramatsu, 2016). CD147 was identified to be the cell surface receptor for cytokine cyclophilin A calprotectin S100A9 RH5 of Plasmodium falciparum (Muramatsu, 2016). Meanwhile, it was reported that CD147 can function as receptor for itself through self-association (Yoshida et al., 2000; Ding et al., 2017). Subsequent studies revealed that tumor cell-associated CD147 and soluble CD147 can lead to CD147-mediated intracellular signaling (Belton et al., 2008) and stimulate multiple matrix metalloproteinases (MMPs) production in neighbouring cells. MMPs induction was inhibited by specific antibody against multimerized CD147 (Sun and Hemler, 2001). Therefore, self-association is very important for the MMPs-inducing function of CD147. As a member of the immunoglobulin superfamily (IgSF), CD147 is composed of an extracellular portion (residue 22– 205, CD147) with two Ig domains separated by a 5-residue flexible linker, a single transmembrane domain and a short intracellular domain. There are three conserved N-linked glycosylation sites on CD147, Asn44 on the N-terminal domain, Asn152 and Asn186 on the C-terminal domain (Grass and Toole, 2015). It is suggested that glycosylated CD147 contained a series of high-mannose and complex-type N-linked glycan structures. Glycosylated CD147 purified from lung carcinoma tissue specimen or produced from eukaryotic expression (residue 32–190) can also stimulate MMPs production (Huang et al., 2013), while prokaryotic expressed CD147 (non-glycosylated) or deglycosylated endogenous CD147 were reported to have no MMP-inducing activity on fibroblasts (Sun and Hemler, 2001; Huang et al., 2013). The three-dimensional structure of prokaryotic expressed CD147 was determined by X-ray crystallography (Yu et al., 2008), and it was reported that Lys63 and Ser193 are essential for the dimerization of CD147 (Cui et al., 2012). However, biochemical studies, including native gel analysis, gel filtration, ultracentrifugation and small-angle X-ray scattering (SAXS), suggest that the prokaryotic recombinant CD147 exists as a monomer in solution (Schlegel et al., 2009; Chen et al., 2014; Wright et al., 2014). To clarify the oligomerization properties of CD147, the soluble protein of the extracellular portion of CD147 (CD147, residue 22–205) was overexpressed and purified from the E. coli strain origami B (DE3) (Fig. S1A and S1B). The recombinant CD147 protein was eluted in a single fraction peak as a monomer on size-exclusion chromatography (Fig. 1A, left). We analyzed CD147 using native PAGE and crosslinking experiments, and only protein band corresponding to monomeric CD147 could be detected (Fig. 1A, right). Laser light scattering (LLS) showed that the hydrodynamic radius of CD147 is ∼2.3 nm (Fig. 1A, middle), which is in agreement with the size of monomeric CD147. In addition, no concentration dependent NH signal change in 2D H-N-HSQC spectra was observed, suggesting no protein self-association. Therefore, the recombinant CD147 protein only exists as a monomer in solution, consistent with previous studies (Schlegel et al., 2009; Wright et al., 2014). As glycosylation was suggested to play a role in the selfassociationofCD147,weexamined the interactionofCD147 with different glycans using NMR titration experiments. Three related polysaccharides (N, N’-diacetylchitobiose, 3’-sialyllactose, 3α, 6α-mannopentaose) and five related monosaccharides (sialic acid, D-mannose, D-glucose, glucosamine hydrochloride, D-galactose) were used in NMR titration experiments. Comparison of 2D H-N HSQC spectra of CD147, with or without 10-fold excess of glycans, showed that there was no NH signal chemical shift perturbation caused by all the glycans (Fig. S2A–D), except sialic acid (Fig. S3A).

mM NaCl (pH 7.0). CD147 EC and C-CD147 EC domain were concentrated and quantified for further experiments.

Chemical cross-linking experiments
Protein (75 μM) with or without Zn(II) (150 μM, 300 μM or 450 μM) was cross-linked with 150 μM ethylene glycolbis (succinimidyl succinate) (EGS) (Pierce) in the reaction buffer (20 mM MOPS, 50 mM NaCl, pH 7.0). Before cross-linking by EGS, EDTA was added with the final concentration of 8 mM when it was used. The reaction mixture was incubated at room temperature for 30 min, and then the reaction was quenched by adding Tris-HCl (1 M, pH 7.5) to a final concentration of 100 mM for 30 min.

Laser light scattering (LLS)
Samples contained 50 μM CD147 EC in 20 mM MOPS buffer (pH 7.0), along with 50 mM NaCl and were filtered (0.22 μm) to remove dust prior to measurements.
Dynamic light scattering (DLS) was conducted using A commercialized spectrometer equipped with a BI-200SM Goniometer and a BI-Turbo-Corr Digital Correlator. A solid-state laser (100 mW, 532 nm, Changchun, China) polarized at the vertical direction was used as the light source.

NMR titration experiments of Zn(II)
The NMR samples all contained 0.4 mM uniformly 15 N labeled protein in 20 mM MOPS, 50 mM NaCl (pH 7.0) with 90% H 2 O/10% D 2 O. A series of 2D 1 H-15 N HSQC spectra with gradually increased Zn(II) concentration (0.2 mM, 0.4 mM, 0.8 mM) were collected at 298 K on a Bruker Avance 700 MHz spectrometer (Bruker, Germany) with cryoprobes. An excess of EDTA (4 mM) was added to the NMR sample for the final 2D 1 H-15 N HSQC spectrum.

Surface plasmon resonance (SPR) assay
Surface plasmon resonance assays were performed using the Biacore T200 with streptavidin (SA) sensor chip. The biotinylated C-terminal domain (C-CD147 EC ) was immobilized on a SA sensor chip via biotin-streptavidin interaction. A concentration series of CD147 EC as analyte (0.045 μM, 0.225μM, 0.45 μM, 0.9 μM, 1.8 μM, 3.6 μM, 4.5 μM, 9 μM and 18 μM) with or without 200 μM Zn(II) was injected over the C-CD147 EC -coated chip for 180 s at rate of 30 μL/min, followed by a 720 s dissociation time. The sample without Zn(II) was also add 1 mM EDTA for SPR experiments. The chip surface was then regenerated with a pulse of glycine-HCl (pH 2.0) at the end of each cycle. All experiments were performed at 25 o C in 10 mM HEPES (pH 7.4), 150 mM NaCl, 0.05% P 20 . The side-chain of histidines should adopt one of the three tautomeric states (N  -H, N  -H, and charged), which can be distinguished based on cross-peak patterns and relative peak intensity in 2D 1 H-15 N HSQC spectrum of the imidazole group ( was also found in charged state. For His102 and His170, the chemical shift differences between 15 N 1 and 15 N 2 are much larger, and the 1 H  -15 N 1 signal is missing, consistent with the N  -H tautomer. Interestingly, His115 showed broad multiple peaks, which indicates that there are multiple conformations in slow-to-medium timescale exchange for its imidazole ring. With increased temperatures, the 1 H-15 N signals of His115 became less broadened, suggesting that multi-conformational exchange rate is getting faster. The chemical shifts and correlation pattern of His115 sidechain at 308K shows that the inmidazole should mainly exist as a N 2 -H neutral tautomer in solution.

Structure calculation
The distance restraints were obtained by analyzing the NOSEY spectra and dihedral angle restraints were obtained by TALOS (Cornilescu et al., 1999). selected for the further AMBER12 refinement. Finally, SANE-AMBER calculation was carried out until no angle violation was bigger than 5° and no distance violation bigger than 0.2 Å. Twenty structures with the lowest AMBER energies were selected and a mean structure was generated by SUPPOSE. PROCHECK_NMR (Laskowski et al., 1996) and MOLMOL (or PyMOL) were used to analyze the structure (Koradi et al., 1996).