The coronavirus disease 2019 (COVID-19), caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), broke out in Wuhan, China in December 2019. Since then this epidemic has been spread and intensified internationally and defined as a Public Health Emergency of International Concern (PHEIC) by the World Health Organization (WHO) on January 30, 2020. As of September 26, 2021, 231,614,338 cases of COVID-19 have been confirmed, including 4,744,918 deaths caused by SARS-CoV-2-induced inflammatory infections or other complications [https://coronavirus.jhu.edu/map.html].

SARS-CoV-2 is a single-stranded RNA virus, typical symptoms of infected patients are fever, cough, chest discomfort, and respiratory distress syndrome (RDS) often occurs in severe cases (1). The inherent error-prone characteristics of viral RNA-dependent RNA polymerase (RdRp) result in the random introduction of mutations in the viral genome during replication. Although the virus encodes an exonuclease (ExoN, nsp14) with a proofreading function, it cannot eliminate the occurrence of viral mutations (2, 3). With continued spreading and viral replication, chronic infection will increase the possibility of virus adaptive mutation (4, 5). The multiple emerging mutations of SARS-CoV-2 variants confer worrisome epidemiologic, immunologic, or pathogenic characteristics (6). WHO developed a Variant Classification scheme, and the categories of variants of concern (VOC) and variants of interest (VOI) were proposed. VOC refers to the SARS-CoV-2 variants that can increase the transmissibility or cause detrimental change in COVID-19 epidemiology by strongly impairing the effectiveness of vaccines and neutralizing antibodies ( Table 1 ), or can cause more serious disease conditions. VOI means the SARS-CoV-2 variants that harbor mutations which are predictable or known to affect viral characteristics, such as infectivity, disease severity, immune escape, or showing a sudden risk to global public health security [https://www.who.int/en/activities/tracking-SARS-CoV-2-variants/]. At present, there are mainly four kinds of VOC: B.1.1.7 (Alpha, originated in the United Kingdom), B.1.351 (Beta, originated in South Africa), P.1 (Gamma, originated in Brazil), and B.1.617.2 (Delta, originated in India) ( Figures 1 , 2 ). VOI mainly includes C.37 (Lambda, first detected in Peru) and B.1.621 (Mu, first detected in Colombia) ( Figures 3 , 4 ) [https://www.who.int/en/activities/tracking-SARS-CoV-2-variants/].

Variants Under Monitoring (VUM) means a SARS-CoV-2 variant with specific mutations that affect virus characteristics, and may pose a potential threat in the future, but there is no definitive evidence of phenotypic or epidemiological impact at present [https://www.who.int/en/activities/tracking-SARS-CoV-2-variants/]. Currently designated VUMs include B.1.427/B.1.429 (first detected in California, USA), R.1 (first detected in several countries in January 2021), B.1.466.2 (first detected in Indonesia), B.1.1.318 (detected in multiple countries in January 2021), B.1.1.519 (detected in several countries in January 2021), C.36.3 (detected in several countries in January 2021), B.1.214.2 (detected in several countries in November 2020), B.1.1.523 (detected in several countries in May 2020), B.1.619 (detected in several countries in May 2020), B.1.620 (detected in several countries in November 2020), C.1.2 (first detected in South Africa), B.1.525 (Eta, detected in several countries in December 2020), B.1.526 (Iota, first detected in the United States), and B.1.617.1 (Kappa, first detected in India) [https://www.who.int/en/activities/tracking-SARS-CoV-2-variants/]. Some former VOIs but no longer designated as VUMs include P.2 (first detected in Brazil) and P.3 (first detected in the Philippines) ( Supplementary Figures 1 , 2 ) [Variants: distribution of cases data, 20 May 2021-GOV.UK (www.gov.uk)].

In this article, we provide a systematic and comprehensive summary of the key genetic variants of SARS-CoV-2 and elucidate the impacts of pivotal mutations on viral transmissibility, infectivity, and immune escape of vaccines and antibodies.

B.1.1.7 is the earliest prevalent variant, which was originally identified in the United Kingdom in September 2020 (18), and has a significant transmission superiority with a higher reproduction number (R) than non-VOC lineages (19). On December 18, 2020, it was designated by the Public Health England (PHE) as a VOC lineage [https://researchportal.phe.gov.uk/en/]. This variant with 10 mutations in spike (S) protein (20, 21) can be divided into three subgroups. The major strain contains Δ69-70, and the other two subgroups lack this deletion, which indicated that Δ69-70 obtained selective advantages in the variation process of SARS-CoV-2 (22). As of January 13, 2021, there were 76 confirmed cases of B.1.1.7 in 12 states in the United States (23), and this variant has been circulating in 174 countries so far, indicating that B.1.1.7 is highly contagious [https://cov-lineages.org/]. According to the survey of PHE, B.1.1.7 would lead to a 30%-50% increase in secondary attack rate (18). In some studies, convalescent plasma and vaccine sera were applied in B.1.1.7 neutralization assays, and no widespread immune escape was observed (22). Pseudovirus neutralization assay showed that the neutralization of BNT162b2-immune sera against B.1.1.7 was largely preserved, indicating that the variants have difficulty escaping from vaccine-mediated immune protection (20). Wang et al. evaluated the neutralization effect of two vaccines BBIBP-CorV (Sinopharm) and CoronaVac (Sinovac) developed in China against B.1.1.7. The results showed that compared with wildtype (WT), the BBIBP-CorV-immune serum remained neutralizing potency to B.1.1.7, but the geometric mean titers (GMTs) of CoronaVac-immune serum against B.1.1.7 decreased significantly (24). The above studies suggested that B.1.1.7 did not pose a great threat to the protective efficacy of COVID-19 vaccines. However, B.1.1.7 variant infection was related to higher virus titer in nasopharyngeal swabs, which accounted for the increased mortality (25). Therefore, the necessity of continuous SARS-CoV-2 sequence surveillance should be highlighted.

B.1.351 was discovered in South Africa in May 2020, and then expanded rapidly to become the dominant lineage in South Africa. It was related to the sharp increase of infected cases nationwide in mid-December, which strongly indicated that this variant had a selective advantage (26). Through the analysis of virus sequencing, it was found that B.1.351 also has three subgroups: 501Y.V2-1, 501Y.V2-2, and 501Y.V2-3. Compared with the receptor binding domain (RBD) and N terminal domain (NTD) sequences of the other two subgroups, 501Y.V2-3 contains R246I mutation and lacks L18F mutation, implying the evolution of SARS-CoV-2. The RBD of B.1.351 harbors three notable mutation sites: K417N, E484K and N501Y, and the combined effect of these three mutations could enhance the affinity of viral spike protein to ACE2 (27). A previous study showed that 21 of 44 convalescent plasma samples lost neutralizing activity against B.1.351 (28). Pseudovirus neutralization assay confirmed that 12 of 17 monoclonal antibodies were ineffective against B.1.351 (29). Compared with WT, the 50% plaque reduction neutralization titer (PRNT₅₀) of 14 convalescent serum against live B.1.351 virus decreased by 3.2- to 41.9-fold (30). The impaired efficacy of vaccine-immune sera including mRNA-1273 (Modena) and BNT162b2 (Pfizer) against 501Y.V2 was demonstrated (11), and 20 of 25 BBIBP-CorV (Sinopharm) vaccinated serum samples showed complete or partial neutralization loss against B.1.351, and the CoronaVac vaccinated sera showed a significant decrease of GMTs against B.1.351, accompanied by the complete or partial neutralization loss of most samples (24). The efficacy of the ChAdOx1 nCoV-19 vaccine (AZD1222) against symptomatic infection caused by B.1.351 was only 10.4%, which directly led to the suspension of the ChAdOx1 nCoV-19 vaccine in South Africa (12). The neutralizing activity against B.1.351 of three potent monoclonal antibodies (2-15, LY-CoV555, and REGN10933) approved for emergency use also decreased significantly (11). These studies raised concerns about the efficacy of vaccines and antibodies against B.1.351. Interestingly, a study carried out in South Africa showed that the convalescent serum from B.1.351-infected patients retained the neutralization activity against the virus in the first-wave epidemic with only a 2.3-fold decrease (30). However, the serum obtained from the first wave of the epidemic infection could not effectively neutralize B.1.351 (30). These findings suggested that the sera from individuals infected with B.1.351 possess cross-neutralization activity to other variants, and the antibodies elicited by the variants with stronger immune escape ability may have more extensive neutralization ability ( Figure 5 ). On the other hand, in the presence of cross-neutralization activity, the neutralization efficacy of antibodies induced by original strains against other strains still decreased to some extent. Therefore, the existence of the cross-neutralization provided an idea for vaccine optimization. Universal vaccines using variants in the epidemic as seed strains against multiple variants could quickly and effectively restrain the spread of the epidemic (21).

In November 2020, a second wave of COVID-19 epidemic broke out in Brazil, causing 76% of the population to be infected (31), which was similar to the epidemic in South Africa. After virus sequence analysis, the variant was named P.1. The RBDs of P.1 and B.1.351 contain mutations in site 417, 484, and 501 residues, except that P.1 harbors K417T, B.1.351 harbors K417N. In addition, the L18F mutation located in the NTD was shown to be related to immune escape potentiality (32). Because of its strong infectivity, P.1 was designated as VOC by WHO on January 11, 2021 (20, 33). The transmissibility of P.1 could be 1.7-fold to 2.4-fold higher than that of non-P.1 lineages, and the convalescent serum from non-P.1 SARS-CoV-2 infected patients could only provide 54%-79% protection against the infection of P.1 (16). According to data from WHO, a variant named P.2 with E484K mutation was originally sequenced in Brazil in April 2020 [https://www.who.int]. P.2 variant showed significant resistance to the vaccinated serum, as the neutralization efficacy of BNT162b2 (Pfizer) against the Brazilian/Japanese P.2 strain decreased 5.8-fold and mRNA-1273 (Moderna) decreased 2.9-fold. Similarly, the vaccine neutralizing activity against the Brazilian/Japanese P.1 strain also decreased significantly (6.7-fold for BNT162b2 and 4.5-fold for mRNA-1273) (17). Few studies have been published on P.2 since it did not cause large-scale outbreaks in Brazil and other countries, but further analysis of the sequence of P.1 and P.2 may reveal the evolution of the virus. Another SARS-CoV-2 variant named P.3 was reported in the Philippines in March 2021. This lineage has several notable mutations in the S protein, including E484K, N501Y, and P681H ( Supplementary Figures 1 , 2 ). P.3 as well as P.1 belong to B.1.1.28 lineage (https://www.who.int/en/activities/tracking-SARS-CoV-2-variants/).

With increased infectivity and virulence, a new variant with an L452R mutation in S protein appeared in California in March 2020, which was named B.1.427/429. Since October 2020, a peak of the COVID-19 epidemic occurred in southern California. According to the sequence analysis of local strains, it was found that most of the variants came from clade 20C (34), which first emerged in Europe and then mutated in Britain and other places, becoming the most widely distributed variant B.1.1.7. From September 2020 to January 2021, 2172 nasal/nasopharyngeal swabs from 44 counties in California were sequenced and it was found that the B.1.427/429 positive sample ratio increased from 0% to more than 50% (34). There are four missense mutations in the S protein of B.1.427/B.1.429, including L452R, S13I, W152C, and D614G, among which the L452R mutation is located in the RBD region ( Supplementary Figures 1 , 2 ). It was found that the virus shedding amount of this variant in vivo was 2-fold higher than that of WT, and the production of pseudovirus containing L452R in cell culture and lung tissue also increased, but was lower than that of B.1.1.7, B.1.351, and P.1, which showed that this variant had higher infectious ability (35). Furthermore, B.1.427/B.1.429 had a certain immune escape ability, which was demonstrated by the significantly reduced neutralization ability of plasma from Pfizer/BioNTech BNT162b2 or Moderna mRNA-1273 vaccinated participants and impaired effects of convalescent serum from patients against B.1.427/B.1.429 (36). Worryingly, more mutations may be accumulated on the basis of B.1.427/B.1.429 in the future, which would further increase the possibility of immune escape (37).

A previous VOI, currently designated VUM variant B.1.526 was first identified in the New York region in November 2020, and began to spread at an alarming rate (38). The most significant mutations in the spike of this lineage are L5F, T95I, D253G or S477N, and D614G (39) ( Supplementary Figure 2 ). Preliminary data analyzed by the New York City Department of Health and Mental Hygiene (DOHMH) suggested that the B.1.526 variant was not associated with an increased risk of breakthrough infection or reinfection after vaccination and with more serious disease conditions (38). The antibodies elicited by infection and vaccines (Pfizer BNT162b2 and Moderna mRNA-1273) could retain the complete neutralization titer against B.1.526 harboring S477N, but the neutralization was 3.5-fold lower against B.1.526 harboring E484K than that of D614G strain (40). In addition, the titer of E484K neutralized by REGN10933 monoclonal antibody decreased by 12-fold, but the neutralization activity of its combined cocktail with REGN10987 against B.1.526 was completely retained (40). Other studies also evaluated the resistance of B.1.526 variant to neutralizing antibodies and ACE2 blocking antibodies induced by the mRNA-1273 vaccine within 7 months. These results suggest that current vaccines still retained neutralization ability for B.1.526 and the variant did not show widespread immune escape (41).

B.1.525, also known as 20A/S: 484K, was discovered and expanded rapidly in many countries in December 2020 (https://covariants.org/variants/20A.S.484K). Phylogenetic analysis showed that the lineage originated from Nigeria (42). Although it did not circulate all over the world, it was defined as the lineage of international significance (43) and was classified as VUM for the possibility of increasing infectivity, virulence and reducing the effectiveness of the vaccine by notable mutations carried by the variant (https://www.who.int/en/activities/tracking-SARS-CoV-2-variants/). Eight mutations (D614G, Q677H, E484K, F888L, A67V, Δ69/70, Δ144/145, and Q52R) in the Spike protein of B.1.525 were identified ( Supplementary Figure 2 ). Q677H harbored by B.1.525 can regulate the transmissibility (42). E484K exists in B.1.351, P.1, and P.2 variants and is related to immune escape. Δ69/70 and Δ144/145 were detected in B.1.1.7, Δ69/70 was shown to have a selective advantage (42), and 144/145 site was verified to be a binding epitope of multiple antibodies targeting NTD (36). The average viral load of the upper respiratory tract between this variant and B.1.1.7 infected patients was similar (43). At the same time, B.1.525 was resistant to neutralization of convalescent serum, vaccine-elicited serum, and monoclonal antibodies (44).

As of September 26, 2021, India’s confirmed cases in has reached 33,652,745, becoming the country with the second largest cumulative number of confirmed cases in the world, which was propelled by variant B.1.617 [https://coronavirus.jhu.edu/map.html]. Through genome sequencing of local confirmed cases, it was found that the local epidemic variant B.1.617 harbored E484K, L452R, and P681R mutations, all of which have been present in other epidemic strains. E484K was detected in B.1.351 while the 681 residue was detected as H681 in B.1.1.7. This showed that the B.1.617 may obtain the characteristics of both B.1.1.7 and B.1.351, and it was considered as one of the most concerned epidemic variants (45). With the spread of the local epidemic in India, B.1.617.1, B.1.617.2, and B.1.617.3 strains emerged successively. B.1.617.2, named Delta, was first sequenced in India in October 2020 and then spread globally. It posed a greater threat to global public health security than the other two lineages and was recognized as VOC by WHO on May 11, 2021 [https://www.who.int/publications/]. At present, at least 163 countries were under the shadow of B.1.617.2 [https://outbreak.info/situation-reports]. Under the background of the prevalence of B.1.1.7, B.1.617.2 strain also appeared in the United Kingdom, and gradually increased or even became a dominant strain, indicating that this variant had a significant competitive advantage (46). In May 2021, a local outbreak caused by Delta occurred in Guangzhou, China, where the variant spread four generations in only 10 days, indicating its remarkable infectivity. A research team collected the clinical information of 159 Delta infection cases in Guangzhou, China and analyzed their clinical characteristics and viral dynamics. Compared with the WT strain, the Delta strain harbored a significantly shorter median incubation period (4 days vs. 6 days), and was featured with a higher viral load (median Ct 20.6 vs. 34.0). Moreover, patients with Delta infection were associated with a shorter time of the deterioration to critical illness, a higher risk of critical status and longer period for RNA-negative conversion than WT (47). These clinical data confirmed that Delta carried unique properties that require continuous monitoring and follow-up.

B.1.617.2 with Δ156-157, G158R, L452R, T478K and other mutations is considered to pose great challenges to the effectiveness of vaccines and neutralizing antibodies (48). Compared with the WT, the neutralizing antibody titers (NAbTs) of sera from BNT162b2 recipients against B.1.617.2 reduced by 5.8-fold (49), and the neutralization titer of serum from Pfizer Comirnaty vaccine had a 3-fold decrease against B.1.617.2 compared with B.1.1.7 (14). Serum from participants injected with a single dose of AstraZeneca vaccine almost completely lost neutralizing activity against B.1.617.2. The effective dose 50% (ED50) analysis showed that compared with B.1.1.7, the neutralization titer of patients sera in 6 and 12 months after infection against B.1.617.2 decreased by 4- to 6-fold, respectively (14). In addition, a recent study held by PHE showed that even after two doses of vaccine, recipients could still be infected by B.1.617.2. Among the four clinically approved antibodies Bamlanivimab, Etesevimab, Casirivimab, and Imdevimab, Bamlanivimab lost its neutralization activity against B.1.617.2, which is considered to be caused by the mutation at site L452, while the other three antibodies remained neutralizing activity (14). Therefore, it is still necessary to monitor the efficacy of the vaccines and antibodies against circulating variants.

Recently, a new SARS-CoV-2 variant C.37 had infected more than 80% of the population in Peru (14). C.37 has a similar mutation as B.1.617.2 in 452 site (L452Q) as well as a mutation F490S in the antibody-binding epitopes of RBD, which may reduce the neutralization of partial RBD antibodies. WHO named the variant as Lambda and designated it as a variant of interest in June 2021 [https://www.who.int/en/activities/tracking-SARS-CoV-2-variants/]. Given C.37 may have higher contagious and immune escape ability, it is necessary to take continuous surveillance for the Lambda variant.

The SARS-CoV-2 Mu variant (B.1.621), a new variant of interest classified by WHO on August 30, 2021, has been detected in at least 50 countries, predominantly in Colombia [https://outbreak.info/situation-reports]. In the context of Gamma’s dominance, Mu transcended Gamma within only five months, propelling the epidemic in Colombia, implying the conspicuous infectivity. The Mu variant was first detected on August 1, 2021 with nine mutations (D614G, P681H, R346K, N501Y, E484K, T95I, D950N, Y145N, Y144S) in the Spike protein ( Figure 4 ). The distinctive substitutions distributed in NTD (T95I, Y145N, Y144S), RBD (R346K, N501Y, E484K) and furin cleavage site (P681H) were associated with increased viral infectivity and immune escape ability. Pseudovirus neutralization assay showed that the neutralization titers of BNT162b2-vaccinated sera against Mu strain were reduced by 7.6-fold compared with WT, showing significantly more resistant than other VOCs (2.6-fold against Alpha; 8.2-fold against Beta; 4.1-fold against Gamma; 4.0-fold against Delta) and VOIs (3.4-fold against Lambda), and a similar situation also existed in convalescent sera (50).

SARS-CoV-2 variants appeared frequently all over the world, which aroused great global attention. According to previous studies, climatic factors will change the speed of virus transmission to a certain extent and affect the spread of the epidemic (51). However, with highly contagious characteristics of SARS-CoV-2, it seems that the regulatory role of climate in virus spread is attenuated. If effective control measures are not taken in time, large-scale epidemic may still break out in hot and humid climates (52), which is proved by the emergence of B.1.351 and B.1.617 strains.

Spike protein of SARS-CoV-2 contains N-terminal domain (NTD), receptor binding domain (RBD), and other regions (such as fusion peptide region) (1). The RBD and NTD regions are mainly located on the S1 subunit, while the fusion peptides and other regions are located on the S2 subunit. Based on the sequence analysis of the main epidemic strains (B.1.1.7, B.1.351, P.1, B.1.427/429, and B.1.617), it was found that the main mutations located in the RBD region are K417N/T, N439K, L452R, E484K/Q, and N501Y. The mutations located in the NTD region are mainly as follows: L18F, T20N, P26S, Δ69-70, D80A, D111D, D138Y, G142D, Δ144, W152G, R190S, D215G, and Δ242-244; other regions are A570D, D614G, H655Y, P681H/R, A701V, T716I, S982A, and T1027I (22, 26, 29, 45, 53, 54) ( Figures 2 , 4 ).

As no effective drugs are available for SARS-CoV-2, vaccination becomes an important strategy in preventing and controlling the epidemic. The purpose of vaccination is to induce an immune response similar to natural infection and to produce associated immune cells and antibodies, and the antibody response caused by the vaccine is stronger than that caused by natural infection (36). Neutralization assay confirmed that the neutralization effect of three RBD mutations N439K, Y453F, and N501Y on convalescent serum was greater than that of vaccinated serum, which indicated that mRNA vaccine was more resistant to single RBD mutation than natural infection (9).

Since the outbreak of the SARS-CoV-2 epidemic, the world has accelerated the development process of vaccines (87). At present, there are seven kinds of potential SARS-CoV-2 vaccines: inactivated vaccines, live attenuated vaccines, DNA vaccines, mRNA vaccines, viral vectored vaccines, protein subunit vaccines, and virus-like particle vaccines. As of 26 September 2021, a total of 6,078,264,761 doses of vaccine [https://coronavirus.jhu.edu/map.html] have been administered worldwide. However, the emerging SARS-CoV-2 variants aroused great concern as a variety of vaccine sera showed weakened neutralization effect against variants ( Table 2 ).

Antibodies are important tools for the treatment of infectious diseases (116). Most of the antibodies used in the treatment of COVID-19 target RBD or NTD, as these two regions have higher immunogenicity (117). The neutralizing antibodies targeting SARS-CoV-2 RBD can be classified into four categories ( Figure 6 ) (54). Among these categories, the immune superiority of RBM epitopes directly contacting with ACE2 are more obvious than other epitopes (56). The mechanism of monoclonal antibodies targeting NTD is still not clear (54, 118).

Both antibody therapy and convalescent serum therapy can lead to higher immune pressure (118). Constantly subjected to immune pressure caused by the same class of neutralizing epitopes leads to the emergence of novel mutations in these neutralizing epitopes, providing the possibility of antibody response escape ( Table 3 ). However, the simultaneous use of two antibodies targeting different epitopes will help to alleviate this situation, because viruses with simultaneous double-site mutations are not easy to survive under such conditions (73, 119). In addition, due to the key role of S2 subunit in membrane fusion (120), and the high conservation of S2 sequence, antibodies or vaccines targeting fusion peptide (FP) can be designed to reduce mutations caused by immune pressure and enhance the persistence and effectiveness of antibodies and vaccines. Antibodies targeting FP can play a role by preventing protease-mediated cleavage of S2 site. In view of the fact that RBD has a high mutation entropy, which increases the possibility of vaccine-induced immune escape, these conservative targets in S2 may be considered for future vaccine design (121, 122).

At present, most of the monoclonal antibodies (mAbs) used in clinical (e.g., REGN10933, LY-CoV-555) still maintain high neutralization efficacy against B.1.1.7 (11), but most of the mAbs showed a remarkable decrease against B.1.351 (11, 29). mAbs therapy can produce selective pressure, which would increase the possibility of virus escaping from targeted antigen mutations. The combination of two or more mAbs that target non-overlapping epitopes could reduce the escape possibility (123). Moreover, due to the key roles of L452R, LY-CoV-555 completely lost the neutralization activity to B.1.427/B.1.429 (36). Other studies showed that the cocktail therapy was less affected by B.1.1.7 (22). Therefore, cocktail therapy may become an effective method for the treatment of SARS-CoV-2 variants infection (73, 119). The non-overlapping epitopes and lack of binding competition are the critical factors in mAbs cocktail selection (73). According to this principle, some studies have designed the combination of CoV2-06 and CoV2-14 as cocktail antibodies. The targeted epitopes of CoV2-06 and CoV2-14 were mutated at the same time, which was not conducive to virus survival. Therefore, the epitopes of CoV2-06 and CoV2-14 can be considered in future vaccine design so as to reduce the immune escape of variants (73). Similarly, COVID-19 IgGs can be obtained from the isolated convalescent serum. These IgGs can target different epitopes of S protein to play effective neutralizing roles (124).

In addition, some mAbs have cross neutralization activity such as COVA1-16, which can maintain neutralization activity against a variety of variants (e.g., B.1.1.7, B.1.351) by binding to highly conserved epitopes of S protein (125). The infection of B.1.1.7 or B.1.351 can also cause low levels of cross neutralization antibodies (30, 36).

Notably, Sinopharm announced that they have isolated a potent monoclonal antibody 2B11 with higher neutralization activity from the convalescent sera. The IC50 of antibody 2B11 against Delta variant is 5 ng/mL, which is a promising alternative treatment for COVID-19 (126).

Continuous surveillance of viral genome sequence data and the effectiveness of vaccines would help to better understand the drivers of SARS-CoV-2 transmission and evolution, providing a basis for vaccine development and update (123). The same point mutation (e.g., D614G, N501Y) sequenced in different virus strains began to spread globally, suggesting that these mutations have certain evolutionary advantages (118). Two important determinants of variant spread are occurrence frequency in individuals and transmission possibility. Immune escape mutations in a single host may be relatively rare, at least in early infections. However, potential host adaptive mutations can be observed even in the absence of vaccine or antibody therapy selection pressure. It is suggested that transmission-enhancing and/or immune-escape SARS-CoV-2 variants are unlikely to arise frequently, but could spread rapidly if they are successfully transmitted (65). We speculate that early mutations of the virus such as B.1.1.7 mainly enhance its transmission ability, while in the later stage of the epidemic, immune escape strains start to emerge with the increased immune pressure.

The number of people with acquired immunity against COVID-19 continues to increase after natural infection or vaccination, but due to unequal interventions and access to vaccines, the virus would subject to greater immune pressure and require repeated immunization rounds to deal with the continuous arising of virus variants (127). Postponement of the second dose could help more people get vaccination when the vaccine production dose is not enough, but the proposal to change the vaccine scheme to a single dose may accelerate the evolution of virus strains. The antibody titer produced by only one dose of vaccination would not be sufficient for virus infection prevention and virus clearance, which is likely to contribute to the production of vaccine-resistant strains (118).

The emergence of new variants emphasizes the need for continued vigilance. Since vaccine-induced herd immunity increases the probability of immune escape, it is difficult to determine which variants or sequences should be selected to update the vaccine sequence. B.1.351 is the variant of greatest concern, with the strong resistance to vaccine serum and antibodies. Therefore, it is believed that the development of vaccine constructs using B.1.351 is the top priority (5). As COVID-19 continues to spread, more virus variants with the ability to escape the neutralization of antibodies would appear. The protection of these variants can be ensured by the combination of two or more potent neutralizing antibodies against different epitopes (128). In addition, using antibody cocktails to resist virus mutation seems to be a sensible strategy. However, it must be recognized that the use of mAbs for long-term treatment or prevention, especially in chronic infected individuals who are immunocompromised, may lead to the emergence of neutralization resistance mutations. In order to avoid selective pressure and immune escape, it is suggested that antibody therapy might consist of a combination of antibodies targeting non-overlapping or highly conserved epitopes (129). Considering the important role of site 484 for antibody binding and neutralization, it is a good strategy to identify monoclonal antibodies from E484K infected individuals (59).

The global circulation of multiple SARS-CoV-2 variants undermines confidence in whether the current vaccines will provide long-term protection. The vaccines elicited antibody responses against RBD in a manner similar to natural infection (63), suggesting that vaccines could reduce the severity of disease caused by natural infection more or less. In addition, the response of T cells against spikes cannot be disturbed by mutations and could still prevent severe diseases caused by variants (59). Moreover, under the continuous emergence of SARS-CoV-2 mutations, the body’s immune system is also constantly improving its ability to deal with the evolution. The memory B cells did not decrease after 6.2 months of effective vaccination, but continued to evolve and were involved in preventing reinfection. These results strongly suggest that vaccinated individuals can respond quickly and effectively to the virus upon exposure (130).

New variants will continue to emerge (123), and intensive surveillance systems are needed to monitor the arising of new variants. Moreover, breakthrough infections among vaccinees urgently need to be elucidated. The second or even third generation vaccines targeting virus variants, as well as the more extensive development of immunogens targeting ACE2-RBD-independent surfaces, are worthy of further study (59).

HF, JF, and ML designed the research. FL, ML, ZP, LJ, LG, JF, and HF read and analyzed the papers. LT and JH participated in discussion. FL, ML, JF, and HF wrote and revised the manuscript. All authors contributed to the article and approved the submitted version.

This manuscript was funded by grants from National Key Research and Development Program of China (grant No. 2020YFA0712102), Fundamental Research Funds for Central Universities (grant No. BUCTZY2022), and H&H Global Research and Technology Center (grant No. H2021028).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.