Identification of key mutations responsible for the enhancement of receptor-binding affinity and immune escape of SARS-CoV-2 Omicron variant

Abstract

The Omicron variant of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has raised concerns worldwide due to its enhanced transmissibility and immune escapability. The first dominant Omicron BA.1 subvariant harbors more than 30 mutations in the spike protein from the prototype virus, of which 15 mutations are located at the receptor binding domain (RBD). These mutations in the RBD region attracted significant attention, which potentially enhance the binding of the receptor human angiotensin-converting enzyme 2 (hACE2) and decrease the potency of neutralizing antibodies/nanobodies. This study applied the molecular dynamics simulations combined with the molecular mechanics-generalized Born surface area (MMGBSA) method, to investigate the molecular mechanism behind the impact of the mutations acquired by Omicron on the binding affinity between RBD and hACE2. Our results indicate that five key mutations, i.e., N440K, T478K, E484A, Q493R, and G496S, contributed significantly to the enhancement of the binding affinity by increasing the electrostatic interactions of the RBD-hACE2 complex. Moreover, fourteen neutralizing antibodies/nanobodies complexed with RBD were used to explore the effects of the mutations in Omicron RBD on their binding affinities. The calculation results indicate that the key mutations E484A and Y505H reduce the binding affinities to RBD for most of the studied neutralizing antibodies/nanobodies, mainly attributed to the elimination of the original favorable gas-phase electrostatic and hydrophobic interactions between them, respectively. Our results provide valuable information for developing effective vaccines and antibody/nanobody drugs.

Graphical abstract

1. Introduction

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) continuously evolves to acquire mutations that may affect its transmissibility or immune-evasive ability [1,2]. So far, many variants have appeared, among which the recently emerged Omicron variant and its sub-lineages have led to new waves of infection worldwide [3]. The Omicron variant first arose as BA.1 sub-lineage, which carries 15 mutations in the receptor-binding domain (RBD) of the spike protein compared to the prototype virus [4,5]. The spike RBD directly interacts with the receptor human angiotensin-converting enzyme 2 (hACE2) on the host cells, and the far more mutations in Omicron RBD than earlier variants enable it to be more transmissible and immune-escapable [6]. Statistical analysis of genomic surveillance data showed that the transmissibility of Omicron was about 3.31-fold higher than that of Delta [7,8]. Computational and experimental studies suggested that the higher transmissibility of Omicron was potentially attributed to its increased binding affinity to the receptor hACE2 [9,10]. In addition, numerous experimental tests demonstrated that the Omicron variant substantially evades host immunity induced by vaccination or previous infection [11,12], and escapes the neutralization of existing therapeutic monoclonal antibodies [11,13].

Not all the mutations occurring in RBD contribute equally to the binding of Omicron with the receptor or the antibodies. Identifying key mutations responsible for the changes in RBD-receptor and RBD-antibody binding affinities is important for better understanding the mechanism behind Omicron's increased transmissibility and immune escapability, which can provide valuable information for broad-spectrum vaccine design and therapeutic drug development. Some previous studies have employed computational method to investigate the binding strength of Omicron RBD to the receptor hACE2 as well as the specific therapeutic antibodies [[14], [15], [16], [17], [18], [19], [20], [21], [22], [23]]. However, it is still not completely understood which and how amino acid mutations essentially alter the Omicron RBD-hACE2 interactions. In addition, a comprehensive understanding of the mechanism behind the immune escape of Omicron from a panel of neutralizing antibodies is still lacking. In the present study, all atomic molecular dynamics (MD) simulation combined with molecular mechanics generalized Born surface area (MM/GBSA) was used to evaluate the impacts of the mutations on the binding of Omicron RBD with the receptor hACE2 and a panel of representative antibodies. Then, per-residue energy decomposition analysis was performed to identify the key mutations primarily contributing to the changes in the binding affinity of Omicron RBD with hACE2 and representative antibodies.

2. Materials and methods

2.1. Preparation of the RBD-hACE2 and RBD-antibody complex structures

The coordinate file for the prototype RBD complexed with the receptor hACE2 was obtained from the protein data bank (PDB) with the accession code 6M0J [24]. The complex structures formed by the prototype RBD and 14 neutralizing antibodies/nanobodies were also obtained from PDB with the accession codes shown in Table 1 . The UCSF chimera software was used to introduce the mutations appearing on the RBD of the Omicron variant, with the sidechain conformation of the mutations determined by the Dunbrack 2010 library.

Table 1.

The complex formed by RBD and neutralizing nanobodies/antibodies studied in this work.

Neutralizing Antibodies	PDB code	Reference	Neutralizing Antibodies	PDB code	Reference
COVA2-04	7jmo	[25]	TY1	6zxn	[31]
S2E12	7k4n	[26]	C102	7k8m	[32]
C119	7k8w	[27]	S2M11	7k43	[26]
H11-H4	6zhd	[28]	CB6	7c01	[33]
NB6	7kkk	[29]	C002	7k8s	[27]
P17	7cwu	[30]	CR3022	6w41	[34]
H11-D4	6z43	[28]	Fab 2-4	6xey	[35]

2.2. All atomic molecular dynamics (MD) simulation

All atomic MD simulation were performed both for the complex structures formed by the prototype RBD and those by the Omicron RBD using Amber16 software with the ff14SB force field [36,37]. The complex structures were first prepared and hydrogen atoms were added with the tLEaP module of AmberTools16, and then solvated with TIP3P waters in a cubic box with a minimum distance of 1.0 nm from any protein atom to the box edges [38]. The net charges of the protein system were neutralized by counter ions. Subsequently, the system was subjected to a two-step energy minimization. In the first step, the positions of all protein atoms were restrained using a harmonic force constant of 10 kcal mol⁻¹ Å⁻², and the solvent molecules were energy-minimized using 1000 cycles of steepest descent algorithm followed by additional 1000 cycles of conjugated gradient algorithm. In the second energy minimization step, all position restraints were removed, and the whole system was minimized again using the steepest descent and conjugated gradient algorithms as in the first step. After energy minimization, a 200 ps MD simulation in the NTV ensemble was carried out, in which all protein atoms’ positions were restrained with a force constant of 10 kcal mol⁻¹ Å⁻², and the temperature of the system was gradually increased to 300 K. Then, a 1 ns MD simulation in an NTP ensemble was performed to progressively release the position restraints on protein atoms with the force constants of 5, 1, 0.25, 0.05, and 0 kcal mol⁻¹ Å⁻², respectively. After that, a 30 ns NTP MD simulation without any position restraint was carried out, and the last 20 ns simulation trajectory was used for the molecular mechanics generalized Born surface area (MM/GBSA) calculation [39]. During the simulation, a time step of 2 fs was utilized, and all hydrogen bonds were constrained using the SHAKE algorithm [40]. The cutoff value for nonbonded interactions was set to 10 Å, and the PME method was used to calculate long-range electrostatic interactions [41]. The temperature and pressure of the system were kept at 300 K and 1 atm using Langevin coupling and Berendsen algorithms, respectively [42]. Both the complex structures formed by the prototype RBD and those formed by the Omicron RBD were simulated, and the changes in binding free energies between them were calculated to evaluate the effects of mutations.

2.3. Molecular mechanics generalized born surface area (MM/GBSA) method

For each complex structure, 200 snapshots were extracted from the last 20 ns trajectory of the simulation, and the binding free energy between RBD and the receptor hACE2 (or the neutralizing antibodies) was calculated by MM/GBSA method using the Amber16 software. The generalized Born-Neck2 (GBn2) implicit solvation model (igb = 8) was used in the MM/GBSA method [43]. The binding free energy was evaluated by summing the following four items [44,45],

$$\Delta G_{bind}=\Delta E_{MM}+\Delta G_{polar}+\Delta G_{non-polar}-T\Delta S$$

where $\Delta E_{MM}$ is the molecular mechanics contribution in a vacuum, which contains internal, van der Waals, and electrostatic interactions; $\Delta G_{polar}$ represents the contribution of polar solvation free energy calculated by using the generalized Born approach; $\Delta G_{non-polar}$ is the contribution of non-polar solvation free energy estimated with the solvent accessible surface area; $T\Delta S$ denotes the contribution of entropy. The calculation of entropy often brings big errors for extensive protein systems, and thus this term was usually neglected in many studies [46]. In this work, the entropy was also not considered. To investigate the impacts of residue mutations on the binding affinities between RBD and hACE2 (or neutralizing antibodies), the same MM/GBSA calculations were performed both for the complex structure formed by the prototype RBD and the corresponding ones formed by the Omicron RBD, and the differences between them were analyzed. Additionally, per-residue energy decomposition analysis was also performed to identify the key mutations primarily responsible for the changes in binding free energies.

3. Results and discussion

3.1. Key mutations contributing to the binding affinity enhancement between Omicron RBD and the receptor hACE2

Fig. 1(a) and (b) display the locations of the 15 mutations carried by Omicron RBD. Most of the mutations are located in the receptor binding motif (RBM), which are involved in the direct interaction with the receptor hACE2, as shown in Fig. 1(b).

Fig. 1.

Illustration of the Omicron RBD mutant residues and changes of the RBD mutation-induced binding free energy. (a) The mutant residues occur in Omicron RBD. (b) The spatial structure of the prototype RBD and hACE2 complex. The mutation sites in Omicron are labeled in the structure. (c) The comparison of binding free energies between the prototype and Omicron RBD-hACE2 complexes.

MD simulations were carried out both for the prototype and Omicron RBDs complexed with the receptor hACE2. The root mean square deviation (RMSD) profiles of the simulated complex systems converged to a relative plateau after the MD runs, indicating that the complex systems reached equilibrium, as shown in Fig. S1(a). Then, based on the MD simulation trajectories, the binding free energy of the prototype RBD and the Omicron RBD with the receptor hACE2 were, respectively, computed by the MM/GBSA approach. The changes in the binding free energy between these two complex systems were obtained to evaluate the impacts of the mutations emerging in Omicron RBD on its binding affinity to hACE2. The calculation results are shown in Fig. 1(c). The binding affinity of the Omicron RBD-hACE2 complex was higher than that of the wild-type complex. The binding free energies for the wild-type and Omicron RBDs with the receptor hACE2 were −60.36 kcal/mol and −79.88 kcal/mol, respectively, as shown in Table 2 . The result indicates that the mutations emerging in Omicron RBD significantly increase the binding affinity between RBD and hACE2, which may be responsible for the significantly improved transmissibility of Omicron variant compared to the ancestral virus.

Table 2.

The binding free energies calculated by MMGBSA method for the wild-type and Omicron RBDs, respectively, in complex with the receptor hACE2^a.

RBD	$\Delta E_{vdw}$	$\Delta E_{ele}$	$\Delta G_{polar}$	$\Delta G_{non-polar}$	$\Delta E_{vdw}+\Delta G_{non-polar}$	$\Delta E_{ele}+\Delta G_{polar}$	$\Delta G_{bind}$
WT-RBDb	−92.76	−704.11	749.81	−13.30	−106.06	45.70	−60.36
Omicron-RBD	−102.17	−1493.51	1529.98	−14.18	−116.35	36.47	−79.88

^aThe unit of the values is kcal/mol.

^bWT-RBD represents wild-type receptor binding domain.

To further explore the mechanism behind the changes in the binding free energy between prototype and Omicron RBD-hACE2 complexes, the MM/GBSA free energy was analyzed separately in different contributing terms. The sum of the van der Waals and the non-polar solvation energies, i.e., $\Delta E_{vdw}+\Delta G_{non-polar}$, represents the hydrophobic interactions between the protein and its receptor. The sum of the gas-phase electrostatic interactions and the polar solvation energy, i.e., $\Delta E_{ele}+\Delta G_{polar}$, accounts for the burial of the charged or polar groups upon binding. The calculation results of different energy terms in MM/GBSA are summarized in Table 2. The results indicate the hydrophobic interactions are favorable for both wild-type and the Omicron RBD-hACE2 complexes, whereas the burial of the charged and polar groups is unfavorable. However, the mutations in Omicron RBD enhance the hydrophobic interactions between RBD and hACE2 mainly by increasing the van der Waals interactions. The energy value of $\Delta E_{vdw}+\Delta G_{non-polar}$ decreases from −106.06 kcal/mol to −116.35 kcal/mol, as shown in Table 2. Besides that, the mutations in Omicron RBD also distinctly improve the gas-phase electrostatic interactions to alleviate the unfavorable burial of the charged and polar groups upon the binding of RBD to hACE2. The energy value of $\Delta E_{ele}$ significantly increases from −704.11 kcal/mol to −1493.51 kcal/mol, and the $\Delta E_{ele}+\Delta G_{polar}$ energies decrease from 45.70 kcal/mol to 36.47 kcal/mol.

To further reveal the key mutant residues responsible for the enhancement of binding affinity between Omicron RBD and hACE2, the binding free energy calculated by MM/GBSA method was decomposed into the contributions of each residue, as shown in Fig. 2 (a). The results indicate that five mutations including N440K, T478K, E484K, Q493R, and G496S, predominantly contributed to the decrease of binding free energy. In the wild-type RBD-hACE2 complex, the binding free energy contributed by Glu484 is a positive value, which implies the residue is unfavorable for the binding to hACE2. However, upon the substitution of Glu484 with Ala, the unfavorable energy becomes neutral, as shown in Fig. 2(a). The binding free energy contributed by the Gln493 is a negative value, which indicates the original residue is favorable for the binding of RBD to hACE2. The substitution of Gln493 with Arg results in a remarkably lower value of binding free energy, implying the mutation improves the beneficial interactions between RBD and hACE2. The binding free energy contributed by the Asn440, Thr478, and Gly496 are close to 0, which implies the residues have negligible contributions to the binding of RBD to hACE2. However, the substitution of Asn440 with Lys, Thr478 with Lys, or Gln493 with Arg obviously decreases the value of binding free energy, indicating the acquirement of these mutations by Omicron favors its binding to the receptor. On the other sides of the binding interface, the residues Glu35, Glu37, Asp38, and Tyr83 in hACE2 also exhibit improved contributions to the binding free energy after introducing the mutations of Omicron, as shown in Fig. 2(b). Most of these residues are negatively charged ones, indicating electrostatic interactions may play important roles in the attachment of Omicron RBD with the receptor hACE2. The residues Glu35 and Tyr83 exhibit stronger interactions to Omicron RBD than the wild-type counterpart, and the binding free energies contributed by the residues Glu37 and Asp38 in hACE2 are converted from unfavorable to favorable ones upon introducing the Omicron-carrying mutations.

Fig. 2.

The binding free energy contributed by each of the mutant residues on RBD and key residues on hACE2, are responsible for improving the RBD-hACE2 binding affinity. (a) The binding free energy contributed by mutant residues in wild-type RBD and Omicron RBD. (b) The binding free is energy contributed by the key residues in hACE2.

Then, the binding interface around the key mutant residues discussed above was also analyzed. The average conformations from the MD simulation trajectories for the prototype and Omicron RBD-hACE2 complexes were compared, and the conformational rearrangements around the key mutation sites were analyzed, as shown in Fig. 3 . In the binding interface of the prototype RBD-hACE2 complex, the key residue Glu484 is surrounded by two negatively charged residues from hACE2, i.e., Glu35 and Glu75, which form repulsive electrostatic interactions between them, as shown in Fig. 3(a). The electrostatic repulsion involving Glu484 is unfavorable for the binding of RBD to hACE2. The replacement of Glu484 by Ala in Omicron RBD eliminates the unfavorable repulsive interactions and facilitates the attachment of Omicron with the receptor, as shown in Fig. 3(d). Compared with the prototype RBD-hACE2 complex, the binding free energy contributed by the residue 484 in the Omicron complex decreases from 1.43 kcal/mol to −0.089 kcal/mol, and that of Glu35 decreases from −0.0007 kcal/mol to −1.65 kcal/mol, as displayed in Fig. 2. The key residue Gln493 was neutral in the prototype RBD, surrounded by the negatively charged residues Glu35 and Asp38 from hACE2, as shown in Fig. 3(b). The substitution of Gln493 with the positively charged residue Arg in Omicron variant brings attractive electrostatic interactions between the residue Arg493 with the residues Glu35 and Asp38, and a new salt bridge is formed between Arg493 and Glu35, as shown in Fig. 3(e), which improves the binding of Omicron RBD with the receptor hACE2. After the mutation of Gln493 to Arg, the binding free energy contributed by this residue decreases from −3.54 kcal/mol to −6.48 kcal/mol, and the that of Glu35 decreases from −0.00070 kcal/mol to −1.65 kcal/mol. The residue Gly496 without any sidechain has negligible contributions to the binding affinity between prototype RBD and hACE2, as shown in Fig. 3(c). The replacement of Gly496 by Ser extends the sidechain, which forms new hydrogen interactions across the binding interface with the residue Asp38 from hACE2, as shown in Fig. 3(f). After introducing the G496S mutation, the binding free energy related to this residue decreases from −0.011 kcal/mol to −2.93 kcal/mol, and the that of Glu38 decreases from 0.72 kcal/mol to −0.53 kcal/mol. The residues T478K and N440K, which are located at the edges of RBD structure as shown in Fig. 3(g), are both neutral in the prototype RBD. Whereas in Omicron RBD, they are mutated to positively charged residues. As revealed by our previous study, the receptor hACE2 is overall negatively charged [47]. Therefore, the substitution of N440 and T478 with Lys in Omicron variant improves the attractive electrostatic interactions, which favour the binding of Omicron RBD to the receptor hACE2, as shown in Fig. 3(h). After introducing the N440K and T478K mutations, the binding free energy contributed by these two residues decrease from −0.026 kcal/mol to −1.40 kcal/mol, and 0.098 kcal/mol to −1.035 kcal/mol, respectively. Taken together, the key residue mutations, including N440K, T478K, E484A, Q493R, and G496S, significantly decrease the electrostatic binding energy, which is consistent with the calculation results shown in Table 2. The electrostatic energy is reduced from −704.11 kcal/mol to −1493.51 kcal/mol.

Fig. 3.

The local structural rearrangement of the RBD-hACE2 key mutant residues. The wild-type RBD-hACE2 complex was colored by cyan, and the Omicron RBD-hACE2 complex was colored by orange. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

3.2. Key mutations contributing to the reduction of the binding affinity between the Omicron RBD and most of the studied neutralizing antibodies and nanobodies

In this study, fourteen neutralizing antibodies and nanobodies complexed with RBD were used to investigate the impacts of the Omicron carrying mutations on their binding affinity. All atomic MD simulations were performed for these antibodies/nanobodies in complex with the wild-type and Omicron RBDs, respectively. The time-dependent RMSD values of the simulated complex systems reached a relative plateau after the MD runs as displayed in Figs. S1(b–o), indicating the equilibrium of the systems. Then, by using the MMGBSA method, the binding free energy between RBD and antibodies/nanobodies both for the prototype and Omicron complex systems were calculated, and the results were compared to explore the influence of the mutations on the binding affinity. The calculation results are shown in Fig. 4 . The results demonstrate that the binding free energy decreases significantly for most of the studied neutralizing antibodies and nanobodies in complex with Omicron RBD compared to that of the prototype RBD, indicating that Omicron may be immunologically resistant to most of these antibodies/nanobodies.

Fig. 4.

The comparison of binding free energies between the wild type and Omicron RBD-nanobodies/antibodies complexes.

For twelve out of the fourteen studied neutralizing antibodies/nanobodies, including COVA2-04, S2E12, C119, H11-H4, NB6, P17, H11-D4, TY1, C102, S2M11, CB6, and C002, exhibit significantly reduced binding affinities with the Omicron RBD compared to those with the prototype RBD. The values of binding free energy increase from −76.32 kcal/mol, −88.72 kcal/mol, −65.77 kcal/mol, −46.28 kcal/mol, −51.68 kcal/mol, −49.70 kcal/mol, −44.93 kcal/mol, −46.14 kcal/mol, −57.93 kcal/mol, −63.31 kcal/mol, −92.18 kcal/mol and −63.57 kcal/mol to −41.45 kcal/mol, −29.60 kcal/mol, −38.39 kcal/mol, −35.99 kcal/mol, −21.59 kcal/mol, −25.25 kcal/mol, −33.11 kcal/mol, −36.17 kcal/mol, −51.80 kcal/mol, −41.72 kcal/mol, −76.19 kcal/mol and −50.92 kcal/mol, respectively, after introducing the Omicron carrying mutations, as shown in Table 3 . It should be noted that for antibody CR3022, the mutations in Omicron RBD have modest effects on the binding affinity between them, where the binding free energy slightly increases from −71.13 kcal/mol to −69.80 kcal/mol. For antibody Fab 2–4, the mutations enhance the binding affinity a little between RBD and this antibody, where the binding free energy decreases from −40.26 kcal/mol to −46.51 kcal/mol. In the following discussion, the twelve RBD-antibody/nanobody systems with reduced binding affinities caused by the mutations were investigated extensively, and the key mutations responsible for the binding affinity reduction were identified.

Table 3.

The binding free energies calculated by the MMGBSA method for wild-type and the Omicron RBD, respectively, in complex with neutralizing antibodies or nanobodies^a.

Neutralizing antibodies	RBD	$\Delta E_{vdw}$	$\Delta E_{ele}$	$\Delta G_{polar}$	$\Delta G_{non-polar}$	$\Delta E_{vdw}+\Delta G_{non-polar}$	$\Delta E_{ele}+\Delta G_{polar}$	$\Delta G_{bind}$
COVA2-04	WTb	−120.90	−182.76	245.37	−18.03	−138.93	62.61	−76.32
	Omicron	−70.58	25.07	14.54	−10.47	−81.05	39.61	−41.45
S2E12	WT	−121.70	−118.87	167.72	−15.87	−137.57	48.85	−88.72
	Omicron	−74.35	34.50	19.58	−9.32	−83.68	54.08	−29.60
C119	WT	−97.55	−91.84	136.66	−13.04	−110.59	44.82	−65.77
	Omicron	−82.65	110.36	−55.91	−10.18	−92.83	54.45	−38.39
H11-H4	WT	−68.46	2.95	27.51	−8.29	−76.75	30.46	−46.28
	Omicron	−66.75	259.94	−221.99	−7.197	−73.94	37.96	−35.99
NB6	WT	−84.64	−80.48	125.07	−11.62	−96.27	44.59	−51.68
	Omicron	−56.94	254.53	−211.81	−7.37	−64.31	42.72	−21.59
P17	WT	−78.21	−56.49	95.75	−10.75	−88.96	39.25	−49.70
	Omicron	−72.87	398.90	−341.92	−9.36	−82.23	56.98	−25.25
H11-D4	WT	−58.45	−17.75	38.92	−7.65	−66.10	21.17	−44.93
	Omicron	−62.62	281.01	−244.23	−7.26	−69.88	36.78	−33.11
TY1	WT	−84.21	53.05	−4.44	−10.54	−94.75	48.61	−46.14
	Omicron	−79.41	246.60	−193.30	−10.06	−89.47	53.30	−36.17
C102	WT	−99.49	−141.54	197.96	−14.85	−114.35	56.42	−57.93
	Omicron	−93.15	12.48	42.46	−13.59	−106.74	54.94	−51.80
S2M11	WT	−70.44	−89.12	105.55	−9.29	−79.73	16.42	−63.31
	Omicron	−69.87	49.90	−13.03	−8.72	−78.59	36.87	−41.72
CB6	WT	−124.28	−200.05	249.96	−17.81	−142.09	49.91	−92.18
	Omicron	−109.58	−26.06	74.91	−15.46	−125.04	48.85	−76.19
C002	WT	−118.36	−48.56	120.03	−16.68	−135.04	71.47	−63.57
	Omicron	−93.90	224.98	−169.46	−12.54	−106.44	55.52	−50.92
CR3022	WT	−104.84	−292.85	340.33	−13.77	−118.61	47.48	−71.13
	Omicron	−102.56	−317.84	364.94	−14.34	−116.90	47.10	−69.80
Fab 2-4	WT	−80.17	82.30	−31.92	−10.48	−90.65	50.39	−40.26
	Omicron	−83.16	223.60	−175.28	−11.66	−94.83	48.32	−46.51

^aThe unit of the values is kcal/mol.

^bWT-RBD represents wild-type RBD.

The $\Delta E_{vdw}+\Delta G_{non-polar}$ and $\Delta E_{ele}+\Delta G_{polar}$ terms of the MMGBSA free energy account for the contributions of hydrophobic interactions and burial of charged or polar groups upon binding, respectively. From the calculation results in Table 3, the reduction of binding affinity in seven RBD-antibodies/nanobodies systems, including COVA2-04, S2E12, C119, NB6, CB6, C102 and C002, mainly caused by the changes of the $\Delta E_{vdw}+\Delta G_{non-polar}$ term, whose values increase from −138.93 kcal/mol, −137.57 kcal/mol, −110.59 kcal/mol, −96.27 kcal/mol, −114.35 kcal/mol, −142.09 kcal/mol and −135.04 kcal/mol to −81.05 kcal/mol, −83.68 kcal/mol, −92.83 kcal/mol, −64.31 kcal/mol, −106.74 kcal/mol, −125.04 kcal/mol and −106.44 kcal/mol, respectively. Whereas, for the other systems, including H11-H4, P17, H11-D4, TY1, and S2M11, the reduction of binding affinity is mainly attributed to the changes in the $\Delta E_{ele}+\Delta G_{polar}$ term, whose values increase from 30.46 kcal/mol, 39.25 kcal/mol, 21.17 kcal/mol, 48.61 kcal/mol, and 16.42 kcal/mol to 37.96 kcal/mol, 56.98 kcal/mol, 36.78 kcal/mol, 53.30 kcal/mol and 36.87 kcal/mol, respectively.

To identify the key mutation residues responsible for the decrease of the binding affinity between RBD and antibodies/nanobodies, the MMGBSA binding free energy was decomposed into the contributions of each residue. The residues, whose energy changes induced by the mutations are larger than 1.0 kcal/mol, are listed in Table 4 . The calculation results show that the Y505H mutation significantly contributes to the reduction of the binding affinity for most of the RBD-antibody/nanobody complexes with large $\Delta E_{vdw}+\Delta G_{non-polar}$ energy changes (Table 4 and Table S1 in the Supplementary materials). Whereas for the complexes dominated by $\Delta E_{ele}+\Delta G_{polar}$ changes, the mostly contributed residues are located at the antibody/nanobody side, as listed in Table 4 and Table S2 in the Supplementary materials. But, as discussed in the following, most of these key residues directly interact with the residue Glu484 on RBD, and the E484A mutation is responsible for the electrostatic energy changes associated with these residues.

Table 4.

The key residues with energy changes induced by the mutations larger than 1.0 kcal/mol for the studied RBD-antibody/nanobody complexes.

Neutralizing antibodies	Key mutant residues on RBD	Key residues on antibodies/nanobodies
COVA2-04	Y505H	Y33, Y52, S53, L96, R98, A99, S27, V28, S29, Y32
S2E12	S477 N, T478K, G496S, Q498R, N501Y, Y505H	V53, G54, R72, M74, S75, T76, C106, Y50
C119	Y505H	Y33, T100, Y102, Y105, K30, Y90
H11-H4	/	R52, S57
NB6	Q498R, Y505H	S25, G26, I27, F29, G30, R31, N32, T52, R54, D99
P17	/	H35, D54, R98, H99, R96
H11-D4	/	R52, S57, R103, S104
TY1	/	Y35, I53
C102	Y505H	G97, D98, Y99
S2M11	/	F29, T30, Y33, N52, I54, S55, T77, Y103
CB6	K417 N, Y505H	Y52, M101, R31, Y32, T94
C002	Y505H	Y32, E95, F98, S99, I100, G57, R96

To further explore the mechanism for the decrease of binding affinity between RBD and antibodies/nanobodies caused by the Y505H mutation, the average structures obtained from MD simulations were compared between the wild-type and mutated systems to analyze the conformation changes around the mutation site. Four systems, including COVA2-04, S2E12, H11-H4, and C002, were analyzed as case studies, as displayed in Fig. 5 . It is found that the wild-type RBD binds to the neutralizing antibodies/nanobodies tightly. However, in the complexes formed by Omicron RBD, distinct gaps appear at the binding interface around the Y505H mutation site, which significantly reduce the van der Waals interactions and decrease the binding affinity between RBD and the antibodies/nanobodies as shown in Table S1 in the Supplementary materials.

Fig. 5.

The comparison of the average conformation obtained from MD simulations between the wild-type RBD-antibodies/nanobodies and Omicron RBD-antibodies/nanobodies complexes. (a)–(d) present the wild-type RBD complexed with the COVA2-04, S2E12, NB6, and C002 antibodies, respectively. (e)–(h) display the corresponding mutated complexes. The complex structures are shown by surface model. Cyan and orange colors represent RBD and antibodies/nanobodies, respectively. Red region in the figure highlight the mutation residue Y505H. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

To reveal the physical mechanism behind the decrease of RBD-antibody/nanobody binding affinity induced by the E484A mutation, three antibodies/nanobodies, i.e., H11-H4, the P17, and H11-D4, complexed with the Omicron RBD were compared to those with the wildtype RBD, as shown in Fig. 6 . It is found that there are one or more positively changed residues on the antibodies/nanobodies directly interact with the negatively charged residue Glu484 in the wild-type RBD, however, the substitution of Glu484 with Ala in the Omicron RBD removes these favorable electrostatic interactions and significantly increases the $\Delta E_{ele}$ energy, as shown in Tables S2 and S3 in the Supplementary materials. The elimination of the electrostatic interactions caused by the E484A mutation is mainly responsible for the reduction of the binding affinity between Omicron RBD and the neutralizing antibodies or nanobodies, which contributes to the possible immune escape of the Omicron variant from these antibodies/nanobodies.

Fig. 6.

The comparison of the local structure around the E484A mutation site between the wild-type (cyan color) and mutated (orange color) RBD-antibody/nanobody complexes. (a)–(c) display the complex structures formed by RBD with the H11-H4, P17, and H11-D4 antibodies/nanobodies, respectively. In these figures, the residue 484 in RBD as well as the positively charged residues directly interacting with the Glu484 in the antibodies/nanobodies are highlighted in stick model. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

4. Conclusion

SARS-CoV-2 continues to evolve, and numerous variants have emerged. The mutations acquired by the variants may influence the transmissibility and immune escape ability of the virus. The epidemiological surveillance and experimental studies have demonstrated that the newly emerged Omicron variant is significantly more transmissible and also distinctly evades the existing neutralizing antibodies. Therefore, serious attention has been paid to the mutations emerged in the Omicron variant. It is of great significance to identify the key mutations and understand the corresponding molecular mechanism responsible for the changes in the binding affinities of RBD with the receptor hACE2 and the neutralizing antibodies/nanobodies. These key mutations may serve as potential targets for developing effective vaccines and antibody/nanobody drugs against the Omicron variant.

In this study, all atomic MD simulations combined with the MMGBSA method were applied to calculate the receptor binding free energy of the Omicron RBD compared with that of the wild-type RBD, and then the key mutations that mostly contribute to the increase of the receptor binding affinity were identified by using per-residue energy decomposition analysis method. Our results indicate that five mutations, e.g., N440K, T478K, E484A, Q493R, and G496S, are mainly responsible for the increase of the binding affinity by forming new favorable electrostatic interactions with the receptor hACE2.

Furthermore, fourteen neutralizing antibodies/nanobodies targeting the prototype RBD were selected, and the impacts of the mutant residues carried by Omicron RBD on the binding affinity of these neutralizing antibodies/nanobodies to RBD were investigated. MMGBSA and per-residue energy decomposition analyses showed that two mutations, i.e., Y505H and E484A, play critical roles for the potential immune escape of the Omicron variant from the studied neutralizing antibodies/nanobodies by different mechanisms. The Y505H mutation destroys the hydrophobic interactions between RBD and the antibodies/nanobodies, and induces split of the binding interface, which significantly reduces the binding affinity between them. The E484A mutation reduces the RBD-antibody/nanobody binding affinities through breaking the favorable electrostatic interactions formed by Glu484 with the corresponding positively charged residues on the antibodies/nanobodies. These two mutations significantly decrease the binding affinities between RBD and the studied antibodies/nanobodies, which may weaken the effectiveness or even evade the neutralization of these antibodies/nanobodies. Our calculation results provide valuable information for the development of effective vaccines and therapeutic antibodies/nanobodies against the Omicron variant.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following is the Supplementary data to this article:

Data availability

The data that has been used is confidential.