key: cord-1028631-o990w70t
authors: Ragotte, Robert J.; Pulido, David; Donnellan, Francesca R.; Gorini, Giacomo; Davies, Hannah; Brun, Juliane; King, Lloyd D. W.; Skinner, Katherine; Draper, Simon J.
title: Human basigin (CD147) does not directly interact with SARS-CoV-2 spike glycoprotein
date: 2021-02-23
journal: bioRxiv
DOI: 10.1101/2021.02.22.432402
sha: 370d543f9e224b2d85cb29b308fcebbe267796bd
doc_id: 1028631
cord_uid: o990w70t

Basigin, or CD147, has been reported as a co-receptor used by SARS-CoV-2 to invade host cells. Basigin also has a well-established role in Plasmodium falciparum malaria infection of human erythrocytes where it is bound by one of the parasite’s invasion ligands, reticulocyte binding protein homolog 5 (RH5). Here, we sought to validate the claim that the receptor binding domain (RBD) of SARS-CoV-2 spike glycoprotein can form a complex with basigin, using RH5-basigin as a positive control. Using recombinantly expressed proteins, size exclusion chromatography and surface plasmon resonance, we show that neither RBD nor full-length spike glycoprotein bind to recombinant human basigin (either expressed in E. coli or mammalian cells). Given the immense interest in SARS-CoV-2 therapeutic targets, we would caution the inclusion of basigin in this list on the basis of its reported direct interaction with SARS-CoV-2 spike glycoprotein. Importance Reducing the mortality and morbidity associated with COVID-19 remains a global health priority. Critical to these efforts is the identification of host factors that are essential to viral entry and replication. Basigin, or CD147, was previously identified as a possible therapeutic target based on the observation that it may act as a co-receptor for SARS-COV-2, binding to the receptor binding domain of the spike protein. Here, we show that there is no direct interaction between the RBD and basigin, casting doubt on its role as a co-receptor and plausibility as a therapeutic target.

Since the emergence of SARS-CoV-2 as the cause of the ongoing COVID-19 pandemic, 35 there has been a rush to identify therapeutic targets that could reduce the immense human 36 and economic toll of COVID-19. Receptors required for viral entry are a natural consideration 37 for druggable targets, as receptor-blockade could both prevent infection, if a drug is delivered 38

prophylactically, or treat infection in a therapeutic setting by stopping the spread of the virus 39 to other tissues and organs. Moreover, there may be existing monoclonal antibodies (mAbs) 40 approved for clinical use that target these receptors. 41

After the release of the genome sequence of SARS-CoV-2, the primary entry receptor 42 was rapidly identified as angiotensin converting enzyme 2 (ACE2) (1-5). This is the same entry 43 receptor used by some other coronaviruses, most notably SARS-CoV-1, a highly similar 44 coronavirus that emerged in 2002 (1, 6). Since this initial identification of ACE2, there has 45 been significant discussion in the literature, both peer-reviewed and pre-print, about other 46 co-receptors or co-factors required for entry (5, 7-9). Transmembrane protease, serine 2 resonance (SPR). Meanwhile, recombinant RH5 shows clear binding to both glycosylated and 71 non-glycosylated basigin through the same methods. 72 mutations that stabilise the protein in a pre-fusion conformation (removal of a furin cleavage 78 site and the introduction of two proline residues: K986P, V987P), was expressed as described 79 previously (26). This construct includes the endogenous viral signal peptide at the N-terminus 80 (residues 1-14), while a C-terminal T4-foldon domain is incorporated to promote association 81 of monomers into trimers to reflect the native transmembrane viral protein. The RBD 82 construct utilised the native SARS-CoV-2 spike signal peptide (1-14) fused directly to residues 83 R319-F541 of the spike glycoprotein which encompasses the binding site for the human 84 receptor ACE2 (26). Both constructs include a C-terminal hexa-histidine (His6) tag for nickel-85 based affinity purification. FL-S and RBD were transiently expressed in Expi293™ cells (Thermo 86

Fisher Scientific) and protein purified from culture supernatants by immobilised metal affinity 87 followed by gel filtration in Tris-buffered saline (TBS) pH 7.4 buffer. 88

Full-length SARS-CoV-2 Nucleoprotein (FL-NP, NCBI Reference Sequence: 89 YP_009724397.2, residues M1-A419) was transiently expressed in Expi293™ cells (Thermo 90

Fisher Scientific) intracellularly. The FL-NP construct included a C-tag peptide (EPEA) at the C-91 terminus (27) for affinity chromatography purification using a C-tag affinity resin (Thermo 92 Basigin, either mammalian-or E. coli-expressed, was immobilised on a CM5 chip 136 through amine conjugation using NHS/EDC coupling using a Biacore X100 (GE Healthcare). 137

Samples were run at 30 µL/min with an injection time of 60 s and a dissociation of 200 s. Then 138 5-step two-fold dilution curves of either RH5, RBD or FL-NP were run over starting at 2 µM, 139 with regeneration of the chip via injection of 10 mM glycine pH 2 for 30 s. Between runs, a 140 single injection of RH5 at 2 µM was carried out to confirm there was no loss in binding activity. 141

For CR3022 binding affinity, approximately 400 response units (RU) of antibody were 142 captured on a protein A chip. Steady-state affinity was determined through an 8-step dilution 143 curve beginning at 1 µM, with 10 mM glycine pH 2 used to regenerate the chip between curves. All curves included one duplicate concentration and were evaluated using the Biacore 145 X100 evaluation software. 146 147 PNGase F treatment 148

PNGase F treatment was conducted as per manufacturer's protocol (New England 149 Biolabs). Briefly, 10 µL of basigin at 0.5 µg/µL expressed in either E. coli or Expi293 TM cells 150 underwent denaturation at 95 ºC for 10 min in glycoprotein denaturing buffer (New England 151 Biolabs) followed by immediate cooling on ice for 10 s. Then, the denatured protein was 152 mixed with 2 µL of GlycoBuffer 2, 2 µL 10 % NP-40, 6 µL of water and 1 µL of PNGase F. After 153 incubation for 1 h at 37 ºC, samples were analysed by non-reducing SDS-PAGE and stained 154 with Coomassie Blue. 155

Expi293 TM and E. coli respectively and glycosylation states confirmed by PNGaseF digest (Fig.  162   S1 ). Correct folding of RBD and FL-S was confirmed via dot blot using CR3022, a known SARS-163

CoV-2 RBD and FL-S binding mAb (28), as the primary antibody (Fig. 1C) . These data showed 164 all proteins expressed as expected and demonstrated high levels of purity. The FL-S and RBD 165 also showed stability upon freeze-thawing, and retained binding of the conformation-166 sensitive mAb CR3022 after three freeze-thaw cycles (Fig. 1C) . 167 We next confirmed SARS-CoV-2 RBD and FL-S binding to human ACE2 using SEC (Fig.  168   2A) . When both RBD and ACE2 were incubated together, the complex eluted at an earlier 169 retention volume as compared to ACE2 alone, indicative of the formation of a higher 170 molecular weight complex. Complex formation was then confirmed using SDS-PAGE whereby 171 both RBD and ACE2 eluted within the same peak at approximately 10 mL, whereas RBD alone 172 normally elutes at approximately 16 mL ( Fig. 2A) . 173

This was next confirmed in the same manner with FL-S trimer, which also eluted as a 174 complex with ACE2 when incubated together, as shown by SDS-PAGE (Fig. 2B) . Although there 175

is only a small change in retention volume between FL-S alone and the FL-S-ACE2 complex, 176 this can be attributed to the use of an S200 column, whose resolution limits are less than the 177 expected size of the FL-S-ACE2 complex (approximately 680 kDa). Nonetheless, it is clear that at approximately 11 mL (Fig. 2B) . We next demonstrated there was no interaction between 180 the RH5 malaria antigen and ACE2 (as expected), given both proteins eluted at the same 181 retention volume whether alone or mixed together (Fig. 2C) . These results confirm that our 182 recombinant FL-S, RBD and ACE2 demonstrate the established interactions. 183

Having confirmed the interaction between FL-S/RBD and ACE2, we proceeded to 186 assess FL-S and RBD binding to glycosylated human basigin using the same methodology. RH5, 187 which acted as the positive control, showed clear binding to basigin, forming a stable complex 188 in solution as confirmed by SEC and SDS-PAGE (Fig. 2F) . Binding affinity between RH5 and 189 basigin is weaker than the reported values for RBD and basigin (approximately 1µM for RH5 190 (22, 23) compared to 185 nM for RBD (9)) indicating that this assay should be sufficiently 191 sensitive to detect the RBD-basigin interaction. 192 SARS-CoV-2 FL-NP was used as a negative control and did not form a complex with 193 basigin. Coincidentally, both FL-NP and basigin elute at the same retention volume, but the 194 absence of any shift to a higher order molecular weight complex when incubated together is 195 consistent with no complex formation (Fig. 2G) . Next, we observed that there was no 196 detectable binding between either RBD or FL-S and glycosylated basigin, with both RBD and 197 FL-S eluting separately from basigin (Fig. 2D,E) . Thus, it did not appear that any complex could 198 be formed in solution between these proteins. 199

Finally, in order to confirm whether glycosylation may affect binding, we performed 200 the experiment again using basigin ectodomain expressed in E. coli (Fig. S1) , as described by 201

Although it was clear the reported FL-S/RBD-basigin complex was not stable enough 206 to detect via SEC, we next sought to confirm the previously reported SPR data showing the 207 RBD-basigin interaction (9). To begin, we confirmed that the RBD protein interacted with 208 CR3022 with the expected affinity, via an 8-step dilution curve beginning at 1 µM. The steady 209 state affinity was determined to be 190 nM, consistent with published data on this interaction 210 (25) (Fig. 3A,B) . 211

Next basigin, either glycosylated ( Fig. 4A-C) or non-glycosylated ( Fig. 4D-F 4A,D). However, RBD did not show any discernible binding to either glycosylated (Fig. 4B) or 217 non-glycosylated basigin (Fig. 4E) . FL-NP also did not bind to either form of basigin, as 218 expected (Fig. 4C,F) . 219

Discussion SPR-and ELISA-based assays (9). The use of anti-basigin mAb in clinical trial has begun on the 225 basis of the original observation that basigin may be required for host cell entry (24). We 226 believe it is necessary to proceed with caution when interpreting the trial data, as further 227 investigation is warranted to determine what role, if any, basigin has in the SARS-CoV-2 228 invasion process. 229

Our findings here are also supported by another independent investigation (31). Shilts 230 et al. also show evidence that there is no direct interaction between CD147 and full-length 231 spike or its S1 domain using a different set of methods than used here (avidity-based 232 extracellular interaction screening and tetramer-staining of HEK293 cells expressing basigin) 233 (31). Their studies complement the work described here, as they also evaluated this 234 interaction using two different isoforms of basigin and in a cellular invasion assay (31), 235

whereas here we only evaluated the far more abundant basigin-2 isoform (32). 236 Initial evaluation of meplazumab, an anti-basigin antibody, for treatment of SARS-237

CoV-2 pneumonia suggested there could be a benefit; however, we interpret these claims 238 with the utmost caution due to the lack of a peer reviewed publication at present and 239 exceedingly small group sizes (24). If indeed these findings hold, it could be due to non- Figure S1 . PNGase F digest of E. coli-expressed (non-glycosylated) and Expi293 TMexpressed (glycosylated) basigin. The lower molecular weight of glycosylated basigin after PNGase F treatment is consistent with the loss of glycans. The heavier molecular weight of glycosylated basigin treated with PNGase F compared to non-glycosylated basigin can be attributed to the presence of the rat CD4 domains 3+4 (CD4d3+4) solubility tag (33 kDa). 

CoV-2 associated molecules in tissues and immune cells in 343 health and in asthma, COPD, obesity, hypertension, and COVID-19 risk factors

CD147 as a Target for COVID-19 Treatment: Suggested 346 Effects of Azithromycin and Stem Cell Engagement

Potential therapeutic targets 348 and promising drugs for combating SARS-CoV-2

Structural variations and expression profiles of the 350 SARS-CoV-2 host invasion genes in lung cancer

Cardiorenal tissues express SARS-CoV-2 entry genes and basigin (BSG/CD147) 354 increases with age in endothelial cells

Expression of SARS-CoV-2 entry receptors in the 357 respiratory tract of healthy individuals, smokers and asthmatics

COVID-19 and Genetic Variants of Protein Involved in the SARS-CoV-2 Entry into 361 the Host Cells

Accelerating the clinical 393 development of protein-based vaccines for malaria by efficient purification using a 394 four amino acid C-terminal "C-tag

A human monoclonal antibody 397 blocking SARS-CoV-2 infection

Invasion Potentiate Malaria-Neutralizing Antibodies

Production of full-length soluble

Plasmodium falciparum RH5 protein vaccine using a Drosophila melanogaster

Schneider 2 stable cell line system

Basigin is a druggable target for host-oriented antimalarial interventions