key: cord-0299404-0yfiaqm7
authors: Erausquin, Elena; López-Sagaseta, Jacinto
title: A single de novo substitution in SARS-CoV-2 spike informs enhanced adherence to human ACE2
date: 2021-07-16
journal: bioRxiv
DOI: 10.1101/2021.07.16.452441
sha: 3f9e17b776a4207ddfbc1f90784eada7d2046701
doc_id: 299404
cord_uid: 0yfiaqm7

SARS-CoV-2 initiates colonization of host cells by binding to cell membrane ACE2 receptor. This binding is mediated by the viral spike receptor binding domain (RBD). The COVID-19 pandemic has brought devastating consequences at a clinical, social and economical levels. Therefore, anticipation of potential novel SARS-causing species or SARS-CoV-2 variants with enhanced binding to ACE2 is key in the prevention of future threats to come. We have characterized a de novo single substitution, Q498Y, in SARS-CoV-2 RBD that confers stronger adherence to ACE2. While the SARS-CoV-2 β variant, which includes three simultaneous amino acid replacements, induces a 4-fold stronger affinity, a single Q498Y substitution results in 2.5-fold tighter binding, compared to the Wuhan-Hu-1 SARS-CoV-2 2019 strain. Additionally, we crystallized RBDQ498Y complexed with ACE2 and provide here the structural basis for this enhanced affinity. These studies inform a rationale for prevention of potential SARS-causing viruses to come.

As the earlier severe acute respiratory syndrome coronavirus (SARS-CoV) (1) and the Middle 27

East Respiratory Syndrome (MERS) coronavirus (2, 3), SARS-CoV-2 (4) belongs to the genus 28 betacoronavirus, and is the causative agent of the Coronavirus disease 2019 (Covid19) 29 pandemic that stroke mankind in late 2019. 30

Covid19 is currently an ongoing threat counteracted by means of massive campaigns of 31 vaccination in humans. While vaccination routines associate with a reduction in severity and 32 clinical outcome, new SARS-CoV-2 variants emerge and pose new challenges for which there 33 is no currently a universal preventive measurement. 34

Since the first Covid19 outbreak in 2019, new SARS-CoV-2 variants are surfacing and have 35

triggered new viral outbreaks (5, 6). Some of these variants are characterized by mutations that 36 affect the RBD binding site on ACE2. A variant bearing the N501Y was detected in the United 37

Kingdom, referred as Variant of Concern 202012/01 or alpha variant, following the World 38

Health Organization convention (7), and has been linked to a higher transmissibility (8). The 39 501.V2 variant, or beta variant, was originally identified in South Africa in December 2020. 40

While the prevalence of this variant was higher among the youths (9), it correlates with a more 41 severe clinical condition. Further, this variant appears to disseminate at a higher rate, compared 42

to previously identified SARS-CoV-2 variants. SARS-CoV-2 b holds three amino acid 43 substitutions in the RBD. These are N501Y, K417N and E484K, which appear to enhance the 44 binding strength of the spike to ACE2 (10). 45 The delta variant was originally identified in India, is already predominant in many countries 46 and includes L452R and T478K substitutions in the RBD. The delta plus variant includes an 47 extra amino acid replacement, K417N (11) . 48 Additionally, increased evidences point to transmissions to humans due to zoonosis. Zoonosis 49 poses further challenges as novel infectious pathogens can be spread from animals to humans. 50

For instance, from 2017 to 2020, five zoonotic avian-related transmissions to humans have 51 been reported (12). Since 2003, three coronaviruses have brought devastating consequences to 52 humans at clinical, economic and social levels. Investigation and anticipation are therefore 53 critical in the prevention of novel pathogens to come with potential to infect humans, in the 54 form of either novel viruses or variants of currently circulating germs. 55 SARS-CoV-2 initiates contacts with the host through its viral envelope-anchored spike (S) 56

protein, which binds with very high affinity to the Angiotensin-converting enzyme 2 (ACE2) 57 on the surface of human cells (13). Many structural studies have evinced with great detail the 58 structural and molecular blueprint of this binding, which involves a nourished array of both 59 polar and non-polar interactions between both the spike RBD and ACE2 (14-16). The binding 60 interface covers a molecular area of 850 Å2. Eighteen residues on the spike RBD establish 61 direct interactions with ACE2 through bonds of different nature, including hydrogen bonds, 62

Van der Waals forces and salt bridges. 63

Following the aforementioned need to inform potential novel threats, we sought to identify 64 amino acid substitutions that might enhance the adherence for ACE2. We performed an in 65 silico mapping of the RBD binding interface and identified a single substitution, Q498Y, that 66 was predicted to enhance binding to ACE2. We provide evidences for this increased adherence 67 using biolayer interferometry approaches. In addition, we report structural grounds that support 68 the molecular mechanism underlying this increased affinity. Genetic mutations leading to Q>Y 69

replacements are not within the most likely variations to occur in nature. Nevertheless, the 70 clinical records collect several isolates of Q>Y replacements in humans infected with SARS-71

CoV-2. 72

These studies inform a de novo substitution in SARS-CoV-2 that results in a tighter binding to 73 ACE2 and which might lead to a compromised clinical setting. 74 75

In silico prediction of RBD mutations that enhance binding to ACE2 77

Using a computational pipeline for high-throughput in silico mutagenesis, we explored 78 substitutions in the RBD wild type (RBDWT)-human ACE2 (ACE2) interface leading to a 79 predicted increase in affinity by means of a reduced free binding energy score. The mutational blueprint calculated for the RBD-ACE2 interface (Fig. 1) , showed a 90 heterogeneous plot whereby position 475 in RBD, corresponding to Alanine, gathers the most 91 favorable substitutions in terms of affinity. In contrast, Tyr489 is, apparently, the less favorable 92 residue to enhance binding to ACE2 upon substitution. Looking at the chemical nature of the 93 replacement amino acid, substitutions to Tyr and Pro result in the most and less beneficial 94 variations, respectively. Here, we focused on the substitution with the greatest score, Q498Y, 95 which produced a predicted affinity change (DDG) of 2.0 kcal/mol, to experimentally verify its 96 impact on the binding affinity of RBD for ACE2. Gln498 is located within the RBD-ACE2 97 binding interface, particularly in the RBD region near Tyr41 and Gln42 in ACE2 N-terminal 98 a-helix, with which establishes polar and VDW contacts (Fig. 1B) . More specifically, there is 99 a putative H-bond with Gln42 side chain at a distance of 3.4 Å. VDW interactions are more 100 populated and participate in binding both Tyr41 and Gln42. 101

In order to investigate the Q498Y mutation, we produced recombinant human ACE2, RBDWT 103

RBDQ498Y and RBD b proteins in sf9 cells. Purified proteins were used to perform biolayer 104 interferometry assays for an accurate characterization of the binding affinities. We first 105 captured ACE2 on the surface of a nickel sensor through a 12xHis c-terminal tag. Increasing 106 concentrations of RBDwt, RBDQ498Y or RBDb were then loaded and the association and 107 dissociation curves were monitored in real time (Fig. 2) . The RBDb version served as a 108

reference for a naturally-ocurring SARS-CoV-2 variant (Fig. S1 ). 109 2.8x10 -4 s -1 koff value, i.e., ~2 and ~4.5-fold slower dissociation rate than RBDQ498Y and RBDWT, 135

respectively. Nevertheless, and despite the three amino acid substitutions, the overall binding 136 affinity of the RBDb form is only 1.6 times stronger than that of RBDQ498Y ( Fig. 1 , Table I ). 137

This results in a relative change in affinity of 1.4-fold per amino acid replaced, while a single 138 Q498Y replacement confers a 2.5-fold increased in binding affinity. 139 

To further confirm this trend, we assembled an additional but indirect BLI approach whereby 144

RBDwt was covalently immobilized on the surface of an AR2G sensor. ACE2 at a concentration 145 of 400 nM was then pulsed and the binding curves were compared with similar concentrations 146 of ACE2 but previously incubated with stoichiometric amounts of RBDWT, RBDQ498Y or RBDb 147 (Fig. 3) , i.e., we set out to investigate the binding strenght to ACE2 in solution. This alternative 148 strategy allowed us to confirm a stronger binding between ACE2 and RBDQ498Y, when 149 compared with RBDWT, as we could notice a stronger competition in the association of ACE2 150

to surface RBDWT. As in the direct assay, the RBDb variant showed the tightest binding to 151 ACE2, as mirrored in the strongest inhibition of ACE2 binding to surface-RBDWT (Fig. 3) . 152

To explore the structural basis for this enhanced affinity, we prepared a complex between 154 RBDQ498Y and ACE2 (Fig. 4) . Incubation of RBDQ498Y with ACE2 showed a sharp shift in the 155 retention time of the protein through size exclusion column chromatography. 156 

We then screened over 750 crystallization conditions to obtain crystals of RBDQ498Y-159 complexed with ACE2. Crystals of the RBDQ498Y-ACE2 complex appeared in with 0.1 M 160 sodium phosphate pH 6.5 and 12% PEG 8000. A full dataset to 3.3 Å was collected and 161 processed in P212121 space group. Initial phases were calculated with the molecular 162 replacement method, and yield two RBDQ498Y-ACE2 complexes per assymetric unit. The 163 overall docking mode is almost identical to that of the complex with RBDWT (Fig. 5A) . The 164

binding footprint is preserved and shows an elongated interface where RBDQ498Y interacts 165 primarily with ACE2 N-terminal a-helix ( Fig. 5 and Table S2 ). The buried surface area (BSA) 166 of the interaction covers 820 and 837 Å 2 in each complex structure in the assymmetric unit. As 167

for the complex with RBDWT, a strong electron density signal is also found for a Zn 2+ ion, 168 which is coordinated by His374, His378, Glu375 and Glu402 in ACE2. The electron density 169 in position 498 allowed us identify a preserved backbone structure, with Tyr498 side chain 170 projected towards the RBD. More specifically, Tyr498 side chain occupies a pocket in the 171 RBD-ACE2 in a manner similar to that of Gln98 in the complex structure with the wild type 172 counterpart ( Fig. 5B-C) . However, this structural arrangement and the bulkier side chain of 173 tyrosine in RBDQ498Y promotes differences with respect to glutamine in RBDWT. 174 

In the complex structure with the RBDWT structure, Gln498 contacts Gln42 in ACE2 through 180 a single 3.4 Å H-bond. On the contrary, the presence of the bulky Tyr498 side chain promotes 181 a minor alteration of Gln42 from its position in the complex with RBDWT to accommodate the 182 long tyrosine side chain (Fig. 6A) . This arrangement favors, on one side, a shorter (2.6 Å) H-183 bond between Tyr498 side chain oxygen and Gln42 side chain oxygen in ACE2, and an 184 additional putative 3.6 Å H-bond with Gln side chain nitrogen ( Fig. 5 and Table S2 ). Moreover, 185

Tyr498 locks Tyr449 through a 2.4 Å-side chain-side chain H-bond, which consequently brings 186 it closer to Asp38 in ACE2. Following this arrangement, the Tyr449-Asp38 H-bond distances 187 are 2.9 Å and 2.5 Å with Asp33 side chain oxygens, vs 2.7 Å and 3.2 Å in the complex with 188

RBDWT. In addition, the number of non-polar interactions is higher in the case of the RBDQ498Y 189 complex (Fig. 5C) . These additional contacts are observed with Asp38, Gln 42 and Lys353. 190 Altogether, the presence of a tyrosine instead of glutamine supports a more populated network 191 of interactions that helps stabilize the RBD-ACE2 complex. 192

Conformational changes induced by Q498Y substitution. 193 Interestingly, we observed conformational changes induced by the Q498Y substitution. First, 194

Gln42 shows a moderate displacement away from the interface (Fig. 6A ). This displacement 195 avoids steric hindrance and enables a suitable pocket for acommodation of Tyr498. 196

However, we could notice an additional conformational change in a loop delimited between 197

Ser443 and Asn448 (Fig 6B) . Again, the bulkier volume of Tyr498 suggests a rearrangement 198 of this loop, which shifts its backbone conformation upwards. Two consecutive glycine 199 residues in the loop, Gly446-Gly447, likely favor plasticity in this region. Thus, the SARS-200

CoV-2 RBD features a flexible RBD structure with ability for adaptation to mutations and 201 preserve or even enhance binding to the human ACE2. 

Aware of the low likelihood of a Q>Y substitution to occur, we looked into the prevalence of 209 Q>Y substitutions reported in isolates from individuals infected with SARS-COV-2 as of April 210 2021 and relative to the Wuhan-Hu-1 strain genome. Analysis of putative substitutions 211 identified in the GISAID database yield a total of 185 mutation events leading to Q>Y 212 substitutions (Fig. 7) . Here, we harnessed an in silico computational pipeline to map the SARS-CoV-2 RBD-human-238 ACE2 binding interface and searched for putative substitution in the RBD that could lead to an 239 enhanced affinity for ACE2. Of all possible substitutions, Q498Y resulted in the highest score, 240 with a predicted affinity change (ΔΔG) of 2.004 kcal/mol over its wild type counterpart, i.e., 241

Q498Y was predicted to further stabilize the SARS-CoV-2 RBD-human-ACE2 interaction. 242

In order to validate this prediction, we expressed recombinantly human ACE2, RBDWT, and 243 RBDQ498Y, and compared the binding kinetic profiles using biolayer interferometry. Our 244 binding studies showed an enhanced binding with RBDQ498Y, thus confirming the predicted in 245 silico values. To gain understanding on the structural bases for this enhanced affinity, we grew 246 crystals of the RBDQ498Y-ACE2 complex and solved the structure at a resolution of 3.3 Å. The 247 electron density maps were of sufficient quality to depict the RBDQ498Y-ACE2 complex, which 248

had an almost identical docking mode when compared with the complex with RBDWT (14, 15) . 249

Importantly, we could observe relevant changes in the region surronding position 498 in the 250 RBD. While the overall docking interface is preserved, we noticed conformational 251 rearrangements, apparently, due to the presence of the bulkier tyrosine residue, as opposed to 252 glutamine. In particular, Gln42 in ACE2 shifts away from its position in the crystal structure 253 complex with RBDWT, while a 6-residue loop near position 498 moderately alters its 254 conformation thanks to two consecutive glycine residues. Both conformational changes avoid 255 steric clashes and hindrance to accommodate Tyr498 side chain. Despite these structural 256

rearrangements, Tyr498 provides a more populated network of polar and hydrophobic 257

interactions with ACE2 that supports the findings through biolayer interferometry, and 258 ultimately points to a tighter stabilization of the RBD-ACE2 complex. 259

Interestingly, the RBD of SARS-CoV, which originated a SARS epidemic in 2002, contains a 260 tyrosine in the position equivalent to RBD498 in SARS-CoV-2 (18 

In silico mutational screening 279 We performed a high-throughput in silico mutagenesis using structural information from the 280 interface between RBD and ACE2. This information was collected from the atomic coordinates 281 deposited in the Protein Data Bank under the accession code 6M0J and used to feed the mCSM-282 PPI2 pipeline (17). The method relies on the structure of a protein-protein template, the 283 reference residue environment, the physicochemical properties of the replacement residue, the 284 degree of evolutionary conservation and the thermodynamics of the protein-protein complex 285

to predict the impact of such mutations. An amino acid substitution matrix focused on the wild 286 type RBD binding site was generated and used to plot a heatmap to evaluate variations in 287 protein-protein binding energy, determined as changes in Gibbs free energy (DDG) upon single-288 residue mutagenesis. 289 290

Cloning and generation of recombinant baculovirus 291

Codon-optimized gene sequences for human extracellular ACE2 containing a C-terminal 292 12xHis tag, and for SARS-CoV-2 wild type spike receptor binding domain (RBD) with a N-293

terminal TwinStrep tag were synthesized by BioBasic Inc. RBD beta variant RBD (RBDβ) was 294 synthesized by GeneUniversal, being position 334 the first N-terminal native amino acid, and 295

which we referred as short RBD. Sequences were digested from delivery generic vectors using 296

BamHI and NotI restriction enzymes (FastDigest) and cloned in the pAcGP67A transfer vector 297

using Optizyme™ T4 DNA Ligase (Thermo Fisher Scientific). E. coli DH5α cells (Invitrogen)  298 were transformed with the modified transfer vectors and before plasmid DNA extraction using 299

the GeneJET Plasmid Miniprep Kit (Thermo Fisher Scientific) following manufacturer's 300

instructions. An N-terminal TwinStrep tag and 3C site containing version of ACE2 with no His 301 tag was generated by PCR using ACE2 plasmid DNA as template. N-terminal TwinStrep-3C 302 site short wild type RBD (RBDWT) and a similar construct including the Q498Y substitution 303 (RBDQ498Y) were generated by PCR using specific primers. All new PCR products were further 304 digested and cloned into pAcGP67A vectors as described before. Figure S1 . SARS-CoV-2 variants of concern and associated amino acid substitutions in the RBD. Cartoon representation of the RBDWT-ACE2 complexes (cyan and pale yellow colors, 500 respectively) using the atomic coordinates deposited in the PDB under the accession number 6M0J. The amino acid substitution for each variant are highlighted with dots and colors. 501 502

Identification of a novel coronavirus in patients 382 with severe acute respiratory syndrome

Emergence of the Middle East respiratory syndrome 384 coronavirus

A pneumonia outbreak 392 associated with a new coronavirus of probable bat origin

Immune Evasion of SARS-CoV-2 395 Emerging Variants: What Have We Learnt So Far? Viruses

CoV-2 variants, spike mutations and immune escape

Estimated transmissibility and impact of SARS-CoV-2 lineage B.1.1.7 in England

Covid-19. Health Department

Antibody evasion by the P.1 strain of SARS-CoV-2

New 'Delta Plus' variant of SARS-CoV-2 identified; here's what we 422 know so far

Function, and Antigenicity of the SARS-CoV-2 Spike Glycoprotein

428 Structural basis of receptor recognition by SARS-CoV-2

Structure of the SARS-CoV-2 spike receptor-binding domain bound to the 432 ACE2 receptor

Cryo-EM 437 Structures of SARS-CoV-2 Spike without and with ACE2 Reveal a pH-Dependent 438 Switch to Mediate Endosomal Positioning of Receptor-Binding Domains

mCSM-PPI2: predicting 441 the effects of mutations on protein-protein interactions

Structure of SARS coronavirus spike receptor-444 binding domain complexed with receptor

Atomic coordinates and structure factors for the RBDQ498Y-ACE2 complex have been 449 deposited in the Protein Data Bank under the accession code 7P19

We thank the staff of XALOC 453 beamline at ALBA Synchrotron for their assistance with X-ray diffraction data collection. We 454 thank Maria Gilda Dichiara Rodriguez, Ane Ochoa Echeverria and Adela Rodriguez Fernandez 455 for their excellent technical support

Ministry of Science and Innovation

Government of Navarre

Author contributions

Competing interests: The authors declare that they have no conflict of interest