SARS-CoV-2 spike S2-specific neutralizing antibodies

ABSTRACT Since the onset of the coronavirus disease 2019 (COVID-19), numerous neutralizing antibodies (NAbs) against severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) have been developed and authorized for emergency use to control the pandemic. Most COVID-19 therapeutic NAbs prevent the S1 subunit of the SARS-CoV-2 spike (S) protein from binding to the human host receptor. However, the emergence of SARS-CoV-2 immune escape variants, which possess frequent mutations on the S1 subunit, may render current NAbs ineffective. In contrast, the relatively conserved S2 subunit of the S protein can elicit NAbs with broader neutralizing potency against various SARS-CoV-2 variants. In this review, the binding specificity and functional features of SARS-CoV-2 NAbs targeting different domains of the S2 subunit are collectively discussed. The knowledge learned from the investigation of the S2-specific NAbs provides insights and potential strategies for developing antibody cocktail therapy and next-generation coronavirus vaccine.

The core of SARS-CoV-2 consists of the nucleocapsid proteins and a single-stranded genomic RNA [11]. The viral core is encapsulated by phospholipid bilayers, which contain the S proteins, membrane proteins and envelope proteins [11], to form 80-120 nm spherical or pleomorphic particles. S proteins form homotrimers on the viral surface and are composed of two subunits, S1 and S2 ( Figure 1A). At the junction of these two subunits is an S1/S2 furin cleavage site [12]. The S1 subunit (S 1-685 ) contains the N-terminal domain (NTD, S  and the receptor-binding domain (RBD, S 319-541 ), which mediates the binding of virus to angiotensinconverting enzyme 2 (ACE2) receptor of the host cell [11,12]. The S2 subunit (S 686-1273 ), which contains the fusion peptide (FP, S 816-855 ), heptad repeat 1 (HR1, S 918-983 ), and heptad repeat 2 (HR2, S 1163-1203 ), is responsible for triggering the virus-host membrane fusion [11][12][13]. After SARS-CoV-2 S protein binding to the ACE2 receptor, the S1/S2 cleavage site is cleaved by host protease to shed the S1 subunit, resulting in the exposure of S2' cleavage site on the S2 subunit [11,12]. The S2' cleavage site is subsequently cleaved by transmembrane serine protease 2 (TMPRSS2) on the host cell membrane, leading to the exposure of the FP for initiating membrane fusion [11,12]. The FP inserts into the membrane of the host cell and then triggers the rearrangement of HR1 and HR2 into a postfusion six-helix bundle structure to pull together the membranes of the virus and the host cell [11,14].
Since S protein plays the key roles in mediating viral entry, it has become the main target of neutralizing antibodies (NAbs). In addition to antiviral drugs, such as nirmatrelvir/ritonavir (Paxlovid) and molnupiravir (Lagevrio), SARS-CoV-2 NAbs can also mitigate or completely inhibit viral infection, and therefore significantly reduce COVID-19 patient's hospitalizations and disease severity [15][16][17]. SARS-CoV-2 NAbs can either be actively elicited by natural infection or vaccination, or be passively administered into individuals for prophylactic or therapeutic purposes [18]. SARS-CoV-2 NAbs may neutralize virus through inhibiting S1-ACE2 interaction or blocking the S2-mediated virus-host membrane fusion [18]. Besides reducing viral titres through Fabmediated neutralization, NAbs can also inhibit viral infection through the Fc-mediated effector functions such as antibody-dependent cell-mediated cytotoxicity (ADCC), antibody-dependent cellular phagocytosis (ADCP), and complement-dependent cytotoxicity (CDC) [19]. Since the outbreak of COVID-19 pandemic, numerous NAbs against SARS-CoV-2, such as bebtelovimab, tixagevimab, and cilgavimab [20,21], have been developed and authorized for emergency use as COVID-19 therapeutics [18]. Most of SARS-CoV-2 NAbs target the S1 subunit, especially the RBD, to block virus binding to the host cell [3]. The RBD is also an important antigen of COVID-19 vaccines for the induction of NAbs [22]. Clinical studies revealed that several RBDspecific antibody therapies and vaccines lost their neutralizing potency due to the emergence of SARS-CoV-2 VOCs [18]. For example, REGEN-COV, an antibody cocktail therapy consisted of two RBD-specific NAbs, lost the neutralizing ability against the Omicron variant [23]. Another antibody therapy, sotrovimab, which also targets the RBD, lost the neutralizing potency against the Omicron BA.2 variant [24]. In response to the immune escape mutations of SARS-CoV-2, it is very important to develop NAbs with broader neutralizing spectrum. Compared to the S1 . RBD, receptor-binding domain (cyan). S1/S2, S1/S2 furin cleavage site. S2', S2' cleavage site R815 (blue spheres). FP, fusion peptide (bright purple). HR1, heptad repeat 1 (green). HR2, heptad repeat 2 (brown). (B) The trimeric S protein ectodomain. (C) The prefusion (left) and postfusion (right) structures of the trimeric S2 subunits. HR2 was not seen in the crystal structure of the trimeric S protein ectodomain (PDB ID: 6XR8). subunit, the amino acid sequences of the S2 subunit are relatively conserved among different SARS-CoV-2 VOCs [25]. Several studies have found that some antibodies targeting special S2 epitopes of the SARS-CoV-2 S protein tend to have broadly neutralizing activities [17,[26][27][28]. In this review, the specificity and functionality of the S2-specific SARS-CoV-2 NAbs are discussed to provide insights and potential strategies for developing antibody therapy and nextgeneration coronavirus vaccine.
The structural analysis and alanine scanning experiment further showed that residues F 1148 , E 1151 , L 1152 , D 1153, Y 1155 , and F 1156 are critical residues for S2P6 recognition [26] ( Figure 2C and Table 1). It was proposed that S2P6 binding to the S2 stem helix might sterically interfere with S2 conformational change ( Figure 3A and B), hence blocking virus-host membrane fusion [26]. Based on the binding specificity and sequence analysis, it is suggested that S2P6 might be arisen in response to previous HCoV-OC43 infection and acquired neutralization efficacy against SARS-CoV-2 through the somatic mutation [26]. 1249A8, an antibody isolated from COVID-19 convalescent patient, could broadly recognize the S protein of various β-CoVs including SARS-CoV-2, SARS-CoV, MERS-CoV, HCoV-OC43, and HCoV-HKU1 (Table 1). 1249A8 also exhibited neutralizing potency against SARS-CoV-2, SARS-CoV, and MERS-CoV, and showed prophylactic and therapeutic activities against SARS-CoV and SARS-CoV-2, including Omicron variant, through the Fab-mediated neutralization and the Fc-mediated effector function [17]. The binding epitope of 1249A8, 1147-SFKEELDKYFKN 1158 , forms an amphipathic α-helix and is embedded in the groove formed by 1249A8 CDRs that interact with residues F 1148 , E 1151 , L 1152 , D 1153 and F 1156 ( Figure 2C and Table 1). The structural analysis further suggested that 1249A8 mimicked the loop structure of the postfusion six-helix bundle to interact with the S2 stem helix and thus disrupted the formation of the postfusion structure [42]. Moreover, 1249A8 possessed high somatic hypermutation (SHM) and exhibited cross-reactivity with the S proteins of HCoV-OC43 and HKU1, suggesting that it might arise from a pre-existing memory B cell against seasonal CoV, and then acquired neutralization efficacy against SARS-CoV-2 infection [17].

S2 stem helix-specific NAbs from immunized mice
The S2 stem helix-specific NAbs could also be produced by immunizing mice with SARS-CoV-2 S protein or S protein-based vaccine. WS6 is an S2 stem helix-specific mAb isolated from mice immunized with mRNA encoding the SARS-CoV-2 S protein [34]. WS6 recognized the S protein of SARS-CoV-2 and exhibited lower affinity against the S proteins of SARS-CoV, MERS-CoV, HCoV-OC43, and HCoV-HKU1. In addition to SARS-CoV-2 Alpha (B.1.1.7), Beta (B.1.351), Gamma (P.1), Delta (B.1.617.2), and Omicron (B.1.1.529) strains, WS6 also neutralized SARS-CoV by inhibiting S protein conformational change [34]. The structural analysis found that WS6 targeted the conserved epitope 1143-PELDSFKEELDKYFKNH 1159 of SARS-CoV-2, which was a part of stem helix and embedded into a groove formed by all CDRs of WS6 heavy chains and light chains [34]. Epitope mapping experiments indicated that F 1148 , L 1152 , Y 1155 , and F 1156 are critical binding sites targeted by WS6, and K 1149 , E 1151 , D 1153 , and K 1157 are also involved in antibody recognition [34] ( Figure 2C and Table 1). It is also worth mentioning that the Fabs of CC40.8 and WS6 have very similar structures while binding with their epitopes ( Figure  3C) although they were derived from different antibody production systems.
S2 stem helix-specific Abs can bind SARS-CoV-2 S protein but lacking neutralizing activity B6 was a pan-β-CoV antibody isolated from mice receiving the prime immunization of the prefusionstabilized MERS-CoV S ectodomain trimer and a subsequent booster of the prefusion-stabilized SARS-CoV S ectodomain trimer [36]. B6 can tightly bind to the epitope peptides of MERS-CoV S 1230-1244 , HCoV-OC43 S 1232-1246 , HCoV-HKU4 S 1231-1245 , SARS-CoV S 1129-1143 , and SARS-CoV-2 S 1147-1161 . Notably, B6 failed to neutralize SARS-CoV-2 and SARS-CoV [36]. The crystal structure of the complex of B6 Fab and its epitope peptide 1230 DFQDELDEFFKNVST 1244 of MERS-CoV S protein showed that it folded as an amphipathic ɑ-helix and docked its hydrophobic face, composed of F 1231 , L 1235 , F 1238 and F 1239 (corresponding to F 1148 , L 1152 , F 1155 , and F 1156 of SARS-CoV-2 S protein) ( Figure 2C, into a hydrophobic groove formed by B6 heavy and light chains [36], Figure 3A and B). The superimposition of the Fabs of S2P6 and B6 showed that they bound to the similar epitope with very different orientations ( Figure 3D), suggesting that the epitope accessibility and the correct binding orientation should be the key factors for the determination of the neutralizing activity of the S2 stem helix-specific antibody.
Another S2 stem helix-specific murine antibody, IgG22, was isolated from mice immunized with the stabilized S2 subunit of MERS-CoV S protein. IgG22 can bind to the S proteins of MERS-CoV, SARS-CoV, SARS-CoV-2, but only showed neutralizing potency against MERS-CoV. Similar with B6, IgG22 binds to MERS-CoV S protein through recognizing the key residues of L 1235 , F 1238 , and F 1239 (corresponding to L 1152 , Y 1155 , and F 1156 of SARS-CoV-2 S protein) [37] ( Figure 2C). Interestingly, superimposition of the Fabs of B6 and IgG22 showed that they have very similar structures upon binding with their epitopes ( Figure 3E). This evidence also supports the criteria that the epitope accessibility and the binding orientation of the S2 stem helix-specific antibody may affect the efficacy of the NAbs although they recognize the same conserved regions found in various CoVs.

NAbs targeting the S2'cleavage site or FP
Proteolytic cleavage at the S2' site of S protein and outstretch of FP are necessary for the S protein-mediated membrane fusion. The sequence of FP is 100% conserved among SARS-CoV-2 VOCs ( Figure 4A) and remains conserved in α-CoVs and β-CoVs ( Figure  4B). Antibodies targeting FP exhibited broad binding ability to S proteins of α-CoVs and β-CoVs, including SARS-CoV-2 VOCs [47]. In addition, FP was identified as an immunodominant epitope [32,[48][49][50]. Previous studies showed that the depletion of antibodies targeting FP reduced the SARS-CoV-2 neutralizing ability of the COVID-19 positive sera [51]. Therefore, it is expected that the FP-specific antibody may have neutralizing activity against CoVs.
76E1 is an FP-specific antibody isolated from a COVID-19 convalescent patient [52]. 76E1 can cross-react to S proteins of some α-CoVs (e.g. HCoV-229E and HCoV-NL63) and β-CoVs  (Table 2). Furthermore, 76E1 provided in vivo protection against SARS-CoV-2 and HCoV-OC43 challenges in both prophylactic and therapeutic murine models [52]. 76E1 can bind to S 809-833 which contains the highly conserved S2' cleavage site R 815 and the Nterminal region of FP. Epitope mapping experiments identified that R 815 , E 819 , and F 823 were important binding residues of 76E1, and D 820 , L 822 , and K 825 were also involved in antibody recognition ( Figure  4C). Since FP is partially buried in the prefusion state of S trimer ( Figure 1B) and R 815 is shielded by the N-terminal loop (residues 804-806) of FP, 76E1 is expected to be incapable of binding to the prefusion S trimer. Structural analysis showed that binding of 76E1 to the S2' cleavage site R 815 may inhibit S2' cleavage and the subsequent cascade of S2 conformational rearrangement, implying that FP is a promising target for neutralizing antibody or vaccine candidate [52]. However, it is controversial that two FP-specific antibodies 555D6 and 125C1 isolated from COVID-19 convalescent patients did not exhibit neutralizing activity against SARS-CoV-2 although they can also bind to the FP region (S 809-823 ) [52]. The different neutralizing potency between 555D6, 125C1, and 76E1 revealed that only certain FP-specific antibody exhibited SARS-CoV-2 neutralizing activity [52]. Hence, the immunization strategy for specifically eliciting 76E1-like NAbs is a challenge for developing pan-coronavirus vaccines. As mentioned previously, FP will become exposed after S protein binding to ACE2 and the cleavage of the S2' site is the key step to facilitate S2 subunit undergoing the prefusion-to-postfusion conformational transition. Based on these observations, it is predicted that ACE2 binding to S protein is the pivotal step to trigger the exposure of FP for 76E1 binding. Therefore, the cocktail therapy combining 76E1 with ACE2 was applied to examine the enhancement of the neutralizing potency against SARS-CoV-2. The results showed that the combination of 76E1 with ACE2 exhibited synergistic effect in preventing SARS-CoV-2 infection [52]. Interestingly, an RBD-specific antibody CB6 also exhibited synergistic effect with 76E1 in neutralizing SARS-CoV-2, implying that CB6 is an ACE2-mimicking antibody [52]. In addition, P2C-1F11, a previously reported antibody with ACE2 mimicry feature [53], also synergistically enhances the neutralizing potency of 76E1 to combat SARS-CoV-2. These data suggest a potential combination therapy by using 76E1 and ACE2 or ACE2-mimicking antibody to synergistically neutralize SARS-CoV-2.
COV44-62 and COV44-79 are also FP-specific SARS-CoV-2 NAbs isolated from COVID-19 convalescent patients. COV44-62 and COV44-79 can inhibit virus-host membrane fusion [54] and neutralize SARS-CoV, HCoV-OC43, HCoV-NL63, HCoV-229E, and SARS-CoV-2 VOCs including Omicron BA.2 and BA.5 subvariants ( Table 2). COV44-62 could further neutralize MERS-CoV. In addition, COV44-62 and COV44-79 possessed in vivo protection ability against SARS-CoV-2 challenge in the Syrian hamster model [54]. COV44-62 and COV44-79 target 815 RSFIEDLLF 823 within the N-terminal region of FP, which adopts a helical structure and is partially solvent-exposed in the prefusion S protein. Binding analysis showed that COV44-62 and COV44-79 have higher affinity to S2 subunit than to prefusion S protein, suggesting that its binding epitope might be shielded by the S1 subunit. Interestingly, epitope mapping experiments indicated that COV44-62 and COV44-79 bind to the same face of the FP helix with different angles. K 814 , R 815 , E 819 , D 820 , L 822 , F 823 , K 825 , and D 830 are key residues for COV44-62 binding, but the critical residues for COV-44-79 binding are partially different, including R 815 , I 818 , E 819 , D 820 , L 822 , F 823 , and D 830 ( Figure 4C and Table 2). Although FP was predicted as an immunodominant epitope and was important for serum-neutralizing potency, the serological analysis found that FP-specific antibodies were only elicited in some COVID-19 patients and were rarely found in vaccinated donors [55]. This realistic situation might be due to the less exposure of COV44-62 and COV44-79 binding epitopes which are shielded by the S1 subunit and therefore rarely presented to B cells during natural infection or vaccination.
VN01H1 and C77G12 are also FP-specific antibodies isolated from COVID-19 convalescent patients [56]. VN01H1 can neutralize β-CoVs (e.g. SARS-CoV-2, SARS-CoV, MERS-CoV) as well as α-CoVs (e.g. HCoV-NL63 and HCoV-229E), whereas C77G12 can only neutralize β-CoVs [56]. Both antibodies could further neutralize SARS-CoV-2 Omicron BA.1 and BA.2 variants and provide in vivo protection against SARS-CoV-2 infection in hamsters [56].  Structural analysis showed that key residues recognized by VN01H1 and C77G12 are R 815 , S 816 , I 818, E 819, D 820 , L 821 , L 822 , F 823 , N 824 , and K 825 ( Figure 4C and Table 2) [56]. VN01H1 and C77G12 likely bind to R 815 through electrostatic interaction for blocking S2' cleavage by TMPRSS2 and therefore preventing the activation of virus-host membrane fusion. The binding epitopes of VN01H1 and C77G12 are buried inside the core region of prefusion S trimer and become exposed after ACE2 binding. In addition, the neutralizing ability of VN01H1 and C77G12 could be improved by engineering the full-length antibodies into the single chain variable fragment (scFv) [56], suggesting that the smaller size of antibodies can increase its accessibility for binding to the epitopes in FP. Furthermore, the ACE2-mimicking antibody S2E12 could enhance VN01H1 and C77G12 binding to S protein and synergistically increase neutralizing effect [56]. Based on these findings, it is hypothesized that ACE2 or ACE2-mimicking antibodies promote exposure of the S2'cleavage site which is then more accessible for the FP-specific antibody binding.
In brief, the FP-specific NAbs mentioned above (76E1, COV44-62, COV44-79, VN01H1, and C77G12) that target the S2' cleavage site and the Nterminal region of FP (Figures 4-5C) were mainly  isolated from COVID-19 patients. These antibodies inhibit viral infection possibly through blocking S2' cleavage by TMPRSS2 or endosomal cathepsins, hence preventing S2 from structural rearrangement for virus-host membrane fusion. However, only a small fraction of FP-specific antibodies possessed broadly neutralizing ability [56]. Thus the strategy for eliciting NAbs using the S2' cleavage site or FP as the immunogen should be carefully designed to efficiently induce neutralizing antibodies.

NAb targeting HR2
The postfusion state of the trimeric S2 is formed by a central HR1 triple-helix bundle and additionally surrounded by HR2 [57] (Figure 1C). Ek1 is a peptide derived from HR2 which can bind to HR1 for blocking HR1-HR2 interaction and neutralizing SARS-CoV-2 through inhibiting virus-host membrane fusion [58]. It is reported that an in silico designed peptide inhibitor based on the HR2 sequences can specifically bind to the HR1 domain of SARS-CoV-2 S protein and prevent the S2 subunit from forming the pre-hairpin conformation, thus potentially blocking SARS-CoV-2 infection [59]. A previous study also found that antibodies targeting HR1 and HR2 of SARS-CoV S protein can neutralize SARS-CoV [60]. However, the NAb targeting HR1 or HR2 of SARS-CoV-2 S protein was poorly investigated.
hMab5.17 is a humanized antibody derived from a murine antibody Mab5, which was elicited by immunization of BALB/c mice with SARS-CoV S protein [57,61]. hMab5.17 can bind to the HR2 domains of SARS-CoV-2 and SARS-CoV S proteins but not the HR2 domains of other β-CoVs, probably due to the different levels of HR2 sequence similarity ( Figure  6A). In addition, hMab5.17 can neutralize SARS-CoV and SARS-CoV-2 VOCs including Alpha (B.1.1.7), Beta (B.1.351), Gamma (P.1), and Delta (B.1.617.2) variants [57]. Notably, though the S protein of Gamma strain has the V1176F mutation in the highly conserved HR2 domain ( Figure 6B), the neutralizing activity of Mab5.17 against Gamma strain remains effective [57]. hMab5.17 also provided in vivo protection against the challenges of SARS-CoV-2 WT and Delta strains in Syrian hamsters and reduced weight loss, viral titre, and pathological changes. hMab5.17 targets the highly conserved epitope 1165 DLGDISGIN 1173 [57] (Figure 6C), which is located at the N-terminal end of the HR2 domain and is well-exposed in both prefusion and postfusion conformational states of S protein. The binding epitope of hMab5.17 folds into an extended N-terminal loop and a C-terminal one-turn helix, and wraps around the central HR1 triple-helix bundle [57]. Sequence alignment further reveals that this epitope is only conserved in SARS-CoV-like viruses but not in other CoVs ( Figure 6A), explaining the binding and neutralizing specificity of Mab5.17 against SARS-CoV-2 and SARS-CoV [57]. Since the HR2 regions are almost 100% conserved among SARS-CoV-2 VOCs (except the Gamma strain containing the V1176F mutation) ( Figure 6B), HR2 has the potential for elicitation of SARS-CoV-2 bNAbs.

Antibody cocktail therapy
High mutation rates of SARS-CoV-2 led to resistance against antibody treatment. Therefore, cocktail therapy might be used to minimize immune escape of SARS-CoV-2, as the virus required simultaneous mutations at multiple epitopes to avoid neutralization by NAbs [62]. Several cocktail therapies against SARS-CoV-2 had been developed and approved by FDA for clinical use. REGN-COV2 was composed of two RBDspecific antibodies imdevimab and casirivimab, and could effectively neutralize SARS-CoV-2 WT, Alpha, Gamma, and Delta variants. However, REGN-COV2 lost neutralizing ability against the Omicron variant, indicating that the antibody cocktail therapy needs to be further explored once an NAb evasion strain has emerged [63]. Evusheld TM , is a combination of two human monoclonal antibodies, tixagevimab and cilgavimab, targeting the surface S protein of SARS-CoV-2. Tixagevimab + cilgavimab combination demonstrated to neutralize SARS-CoV-2 pre-omicron strains. However, recently emerged Omicron sublineages have exhibited the ability to escape this cocktail's activity. Although Evusheld TM could have maintained some benefit for COVID-19 prophylaxis and treatment in immunosuppressed and vulnerable populations that cannot be vaccinated (for safety reasons or immune diseases), the utility of this cocktail could have compromised [64].
It has been reported that a combination of S2 stem helix-specific NAb 1249A8 and S1-specific NAb 1213H7 through direct respiratory administration could synergistically protect mice from SARS-CoV-2 WT, SARS-CoV-2 Delta variant, and SARS-CoV infections [17]. Another research also found that a combination of S2 stem helix-specific NAb CV3-25 and RBD-specific antibody CV3-1 could effectively protect mice from SARS-CoV-2 infection [44]. Besides S2 stem helix-specific antibodies, FP-specific antibodies were also applied in cocktail therapy. As described previously, the combination of the FPspecific antibody 76E1 and ACE2 or ACE2-mimicking antibodies (CB6 or P2C-1F11) exhibited synergistic effect in preventing SARS-CoV-2 infection [52]. It's also found that the combination of the FP-specific antibody VN01H1 or C77G12 and ACE2-mimicking antibody S2E12 possessed synergistic neutralizing effect [56]. Therefore, it is suggested that the combination of S1-specific and S2-specific antibodies for COVID-19 treatment is a promising strategy to prevent the emergence of the immune escape strains.

Vaccine development
Since the outbreak of COVID-19, numerous vaccines have been developed and several have been authorized for emergency use [65][66][67]. The emerged SARS-CoV-2 variants, especially those possessing RBD mutations, exhibited resistance to vaccine-stimulated NAbs [68]. Furthermore, the SARS-CoV-2 BA.1 and BA2 neutralizing titres of the sera derived from BNT1162b2-vaccinated donors were 20-fold lower than the titres against the D614G strain [69]. Mutations in RBD or NTD of the S protein often caused antibody evasion [70]. Previous studies have identified or predicted S 812-829 , S 1148-1159 , and S 1256-1273 as antigenic hot areas in the S2 subunit [32,49,[71][72][73][74]. S 812-829 -specific IgGs were strongly elicited in both COVID-19 patients and mRNA-vaccinated donor [75], demonstrating its high immunogenicity. S 818-835 -specific antibodies were significantly correlated with the neutralizing IC 50 values and depletion of these antibodies reduced the serum-neutralizing potency for inhibiting SARS-CoV-2 pseudovirus infection [51]. S 811-830 -specific antibodies elicited by previous HCoV infection could mitigate the severity of COVID-19 [32]. Furthermore, it was found that antibodies against S 1148-1159 possessed neutralizing activity [74]. S 1148-1159 -specific IgG responses in the COVID-19 non-survivor group were significantly lower than that in the survivor group, suggesting the protective roles of the S2 stem helix-specific antibodies [71]. Another study also identified that S 1146-1165 was an immunodominant epitope and antibodies against this peptide partially contributed to the SARS-CoV-2 neutralizing potency of the COVID-19-convalescent patient sera [76]. These findings suggested that FP and S2 stem helix-specific antibodies really played important roles in preventing SARS-CoV-2 infection. Additionally, the S2 crossreactive antibodies isolated from COVID-19 patients exhibited broader ranges of binding ability against various SARS-CoV-2 VOCs than antibodies against the S1 subunit [41]. To avoid immune escape and specifically elicit NAbs for combating SARS-CoV-2 WT and various VOCs, the relatively conserved FP and S2 stem helix seem to be better vaccine candidates [77].
It has been reported that the pre-existing immunity against endemic HCoVs may also contribute to modulate immune responses in the SARS-CoV-2-unexposed subjects for mitigating COVID-19 severity [78,79]. It was proposed that some S2 stem helixspecific NAbs, such as CC40.8 and S2P6, might be primed by previously endemic CoV and then be reactivated by SASR-CoV-2 infection for developing more broadly reactive antibodies [28]. The high SHM levels of most FP-specific antibodies also suggested the same hypothesis [56]. It was also found that the B cells of COVID-19 patients have increased SHM and higher affinity against the S proteins of HCoVs than that of the pre-pandemic individuals, indicating that SARS-CoV-2 exposure might train the HCoV-induced B cells to produce more effective SARS-CoV-2 NAbs [41]. Immunogenetic analysis of the S2 stem helix-specific antibody sequences showed strong enrichment of IGHV1-46 (63%) and IGHV3-23 (22%) germline gene families [43]. For light chain gene usage, a strong enrichment of IGKV3-20 (47%) and IGLV1-51 (16%) germline gene families was found [43]. Of note, S2P6 and 1249A8 are encoded by IGHV1-46 and IGKV3-20. CC40.8 is encoded by IGHV3-23 and IGKV3-20. Some unique IgG germline gene repertoires served as the foundation for consisting of the S2 stem helixcross reactive NAbs.
Despite the broadly neutralizing activity, it has been observed that the neutralizing titres of the S2 stem helix-specific antibodies were relatively lower than those of the S1-or RBD-targeting antibodies. CV3-25 neutralized various SARS-CoV-2 strains with IC50 values ranging from 0.05 to 0.2 μg/mL [46]. In contrast, the RBD-specific antibody CV3-1 neutralized various SARS-CoV-2 strains with IC50 values ranging from 0.004 to 0.014 μg/mL [46]. S2P6 neutralized the authentic SARS-CoV-2 with the IC50 value of 1.67 μg/mL, but the RBD-specific antibody S309 exhibited the IC50 value of 0.04 μg/mL [26]. 1249A8 neutralized the SARS-CoV-2 Omicron virus with the IC50 value of 2.4 μg/mL, but the S1-specific neutralizing antibody 1213H7 exhibited the IC50 value of 0.06 μg/mL [17]. These findings revealed that the potency metrics of the S2 stem helix-specific antibodies are 10-40 folds lower than those of S1-or RBD-targeting antibodies. Therefore, the improvement of the potency of the S2 stem helix-specific antibodies remains a challenge to be conquered in the future.
The FP within residues 816-855 of S protein is conserved among α-CoVs and β-CoVs. Antibodies targeting this region showed broad cross-activation against S proteins of α-CoVs and β-CoVs, including various SARS-CoV-2 VOCs. NAbs targeting S2' cleavage site/FP inhibit viral infection through blocking the S2' cleavage process and the subsequent S2 rearrangement for membrane fusion [52,4,56]. Structural analysis showed that S2' cleavage site/FP is buried in the interior core of the trimeric S protein [52,54] and becomes exposed after binding of S protein to the ACE2 receptor. A combination of S2' cleavage site/FP-specific NAbs and ACE2 protein or ACE2mimicking antibodies as a cocktail therapy exhibited synergistic effect in preventing SARS-CoV-2 infection [52,56].
Several studies have developed HR1-specific peptides that can inhibit SARS-CoV-2 infection by blocking the formation of the six-helix bundle [57][58][59]. The HR2-specific mAb hMAb5.17 can neutralize SARS-CoV-2 and SARS-CoV [57]. However, the NAbs against HR1 or HR2 were rarely found, probably due to the poor immunogenicity or the inaccessibility of the region in the prefusion state of the trimeric S protein.

Disclosure statement
No potential conflict of interest was reported by the author (s).

Funding
The work was supported by the National Science and Technology Council, Taiwan (grant number MOST111-2320-B-002-047) and National Taiwan University (grant number 111L880802).