Cross‐Neutralizing and Broadly Neutralizing Human Antibodies Against Nonenveloped Viruses

3.1. Parvoviridae

Adeno‐associated viruses (AAV) (genus Dependovirus, Parvoviridae) are small single‐stranded DNA viruses which are classed as nine distinct serotypes, of which AAV2 and AAV9 have been developed for use as gene therapy vectors. Their T = 1 icosahedral capsids are composed of 60 viral capsid proteins (CPs) which form characteristic features at the symmetry axes: fivefold channels, threefold protrusions, and twofold depressions. The CPs of different AAV serotypes share a conserved jelly‐roll core with variable surface regions that are responsible for the structural differences between serotypes. Infectious entry of AAV begins with attachment to host glycans, the nature of which varies between AAV types (heparan sulfate for AAV2, and galactose for AAV9). Attachment is followed by internalization mediated by the transmembrane AAVR protein, which is the common entry receptor for all AAV serotypes [26]. The presence of both pre‐existing and treatment‐induced neutralizing antibodies represents a serious obstacle to AAV‐vectored gene therapies, and in particular there is a degree of cross‐neutralization between wild‐type AAVs circulating in human populations worldwide and AAV9 capsids [27]. The structural basis of this cross‐neutralization was revealed by the characterization of antibodies secreted by cloned B‐cells from three infants treated with Zolgensma, an AAV9 gene therapy product for spinal muscular atrophy [12]. These three children all developed high titers of AAV9 NAb, with significant serological cross‐reactivity to AAV1, 2, 5, and 8. Seven antibodies were cloned and expressed from each child, and the cryo‐EM structures of the 21 capsid‐Fragment antigen‐binding (Fab) complexes were determined. Almost all (6 or 7 out of 7 antibodies from each child) of these mAbs were neutralizing, and while most (17 out of 21) were type‐specific, four mAbs showed some cross‐specificity, of which two (mAb1‐6 and 2–7) were able to neutralize multiple AAV serotypes. Cryo‐EM structures showed that the twofold axis of symmetry of the virion was immunodominant, being recognized by 16 of 21 mAbs, fifteen of which were strictly type‐specific. These immunodominant epitopes overlapped with the binding site of the AAVR, indicating that type‐specific antibodies neutralized the virus by blocking binding to the entry receptor. In contrast, mAb1‐6 and 2–7 recognized an epitope at the top of the fivefold axis of the particle that did not overlap with either the galactose or AAVR binding sites (Figure 2A). Although the neutralizing mechanism was not determined, the authors surmised that bNAb binding might interfere with externalization of the VP1u domain, which is required for endosomal escape and nuclear targeting of AAV virions during infectious entry, since mAb1‐6 binding induced a conformational change in the capsid, reducing the diameter of the pore at the fivefold axis [13].

3.2. Hepeviridae

HEV is a small, nonenveloped (+) strand RNA virus. Its capsid protein, encoded by ORF2, self‐assembles into an icosahedral T = 3 structure composed of 90 homodimers. Each monomer consists of three domains, shell (S), middle (M), and protruding (P), with the P domain forming prominent spikes that mediate interactions with host factors. HEV encompasses eight genotypes (HEV‐1 to HEV‐8), with HEV‐1 to HEV‐4 responsible for most human infections. Transmission occurs primarily via contaminated water (HEV‐1, HEV‐2) or zoonotic foodborne routes (HEV‐3, HEV‐4). Despite its genetic diversity, HEV is classified as a single serotype, with conserved neutralizing epitopes on the capsid. HEV is associated with an estimated 20 million infections and 30,000–40,000 deaths annually, with severe disease particularly affecting pregnant women and immunocompromised patients. Chronic infections can occur in transplant recipients and other vulnerable groups. Although an effective vaccine (Hecolin) is licensed in China, it is not widely available, and the off‐label therapeutic use of ribavirin faces efficacy limitations and emerging resistance. The need for therapeutics is reinforced by the emergence of Rocahepevirus ratti (ratHEV) as a zoonotic pathogen responsible for confirmed cases of acute and chronic hepatitis in humans across Asia and Europe [28]. Despite sharing only 55%–60% amino acid identity with human HEV, ratHEV is able to cross species barriers and infect both immunocompetent and immunocompromised individuals. In this context, bNAbs represent a promising approach due to their ability to target conserved viral sites and provide pan‐genotypic protection, in some cases extending across distinct hepevirus species.

Recent studies by Molinos‐Albert et al. and Ssebyatika et al. have characterized two major classes of human bNAbs targeting the apex of the protruding (P) domain of the HEV capsid, which has emerged as a critical site of viral vulnerability [14, 15]. On the one hand, Group VI antibodies (including Es1.327, Es4.431, and Es5.127) recognize conserved loops located at the apex of P‐domain monomers (Figure 2B). On the other hand, a distinct class of glycan‐sensitive antibodies (p60.1 and p60.12) recognizes a conserved quaternary epitope located at the interface between P‐domain dimers (Figure 2B), directly involving residue N562 (the N3 glycosylation site). A unique feature of these glycan‐sensitive antibodies is their selective recognition of the nonglycosylated form of the capsid protein, which is found exclusively on infectious virions. This specificity allows them to evade the decoy effect of the heavily glycosylated soluble ORF2 proteins, which are abundantly secreted into the bloodstream during infection.

Both classes of bnAbs exhibit picomolar neutralizing potency and broad activity against all major human HEV genotypes (HEV‐1 to HEV‐4) and display pan‐hepevirus activity by cross‐neutralizing ratHEV, despite its substantial genetic divergence from human HEV strains. On the other hand, the neutralization mechanism of these bNAbs was not specifically investigated, and since the attachment receptor for HEV is not known, it is not possible to infer whether bNAbs block receptor binding from structural data. Interestingly, p60.1 and p60.12 were also able to neutralize quasi‐enveloped HEV particles, suggesting postattachment neutralization [29].

3.3. Picornaviridae

The Picornaviridae family is a diverse group of (+) strand RNA viruses, including the Enterovirus genus, which comprises many human pathogens such as rhinoviruses (RV), coxsackieviruses (CV), and poliovirus. Enterovirus capsids are composed of three external (VP1, VP2, and VP3) and one internal (VP4) structural proteins. Pentamers of VP1 define the vertices of the icosahedral capsid, and trimers of VP2–VP3 form a hexagonal structure at the threefold axis. The interface between the VP1 pentamer and the adjoining VP2‐VP3 hexagon often forms a trench, which is referred to as the viral canyon which generally constitutes the receptor‐binding pocket. Enteroviruses use a variety of host entry receptors, and even within a single enterovirus species, there is extensive serological diversity—for example, 75 distinct serotypes of RV‐A have been described. Despite this diversity, the general mechanism of infectious entry is conserved within the Enterovirus genus. First, binding to host cells is mediated by attachment receptors; then a second interaction with an uncoating receptor, combined with exposure to low pH, leads to expansion of the capsid into an uncoating intermediate known as the “A‐particle”. Finally, the viral genome is delivered into the cytoplasm through a transmembrane channel composed of hydrophobic domains of the VP1 and VP4 proteins.

Multiple distinct serotypes of an enterovirus species can use the same attachment or uncoating receptor—for example, most RV‐A serotypes use ICAM‐1 as a receptor—which suggests that monoclonal antibodies with the capacity to neutralize enteroviruses across serotypes may exist. Some support for this view has come from work in mice documenting cross‐neutralization of different B‐species Coxsackie viruses [30, 31], and the characterization of a monoclonal antibody with the ability to neutralize both EV71 and CVA16 [32]. In humans, a broadly neutralizing antibody, EV68–228, isolated from pediatric patients with documented EV‐D68 infection, showed high neutralizing potency across viral clades within the EV‐D68 species, conferred strong protection in animal models [16], and has completed a phase I clinical trial (NCT06444048). EV68–228 binds around the fivefold axes of the EV‐D68 capsid, targeting a structurally conserved site at the canyon rim that includes the VP1 DE and the VP2 EF loops. The authors suggested that by occupying this site, EV68–228 could block the interaction of the virus with sialic acid‐containing receptors, thus preventing viral attachment and entry. However, the recent identification of the EV‐D68 entry receptor MFSD6 [33] revealed that the EV68‐228 binding footprint also includes capsid residues that directly interact with MFSD6 (Figure 2C), suggesting that its neutralizing potency may be related to the inhibition of infectious entry, rather than simply blocking attachment to susceptible cells.

Similarly, several cross‐neutralizing poliovirus‐specific human monoclonal antibodies have been described [34, 35], including one bNAb, 9H2, with the ability to neutralize all three poliovirus serotypes. Cryo‐EM structures showed that 9H2 binds across and into the viral canyon, overlapping significantly with the poliovirus receptor (PVR/CD155) binding footprint (Figure 2D), thus directly preventing attachment to the cellular receptor [17].

On the other hand, no cross‐neutralizing mAbs against human RV have been characterized, despite the description of HRV‐A16 and HRV‐A39 cross‐reactive memory B‐cells during acute infection in humans [36]. These memory B‐cells preferentially secrete IgG and rapidly migrate to the upper respiratory tract following experimental RV infection; however, the structural and functional characteristics of the antibodies they secrete, as well as the breadth of their cross‐specificity, have not been reported to date.

3.4. Caliciviridae

Noroviruses (NVs) are small (+) strand RNA viruses of the Caliciviridae family, with a T = 3 icosahedral capsid composed of the VP1 protein comprising an interior shell (S) domain and an exposed protruding (P) domain, with a P‐domain dimer making up each protrusion. Human noroviruses are the leading cause of viral gastroenteritis, with genogroups I and II responsible for the majority of human infections. Genogroups are themselves made up of multiple genotypes, with GII.4 and GII.17 currently predominant worldwide [37]. The NV GII.4 capsid is subject to antigenic drift, leading to the emergence of successive waves of new GII.4 variants over the past four decades. Seven vaccine candidates are under clinical development [38], and the extent to which both vaccination and infection are able to induce cNAbs continues to be the focus of much research. These efforts have been hampered by the difficulty in propagating NV in cell culture systems, which has prevented the development of scalable neutralizing assays. Indeed, culture of human NVs has only been possible since 2016, when human intestinal enteroids, rather than cell lines, were first successfully used as an in vitro host‐cell system [39]. Therefore, it has only recently become possible to study neutralizing mechanisms other than the direct blocking of receptor binding.

NVs bind to histoblood group antigen (HBGA) glycans present at the surface of enterocytes, with the P2 subdomain of VP1 responsible for HBGA binding. Symptomatic infection leads to the development of VP1‐specific antibodies that block binding to HBGA, which are correlated with protection against subsequent infection [40]. Although most HBGA‐blocking antibodies bind directly to the HBGA binding site at the tip of P2 and have narrow specificity, several cross‐specific HBGA‐blocking mAbs, isolated from healthy blood donors or recipients of experimental NV vaccines, have now been characterized [18, 41, 42, 43]. The breadth of these different mAbs varies, with specific antibodies able to block HBGA binding of multiple GII.4 variants (A1431 [41]), multiple GII genotypes (NORO‐320 [18], VX20, VX22 [43]), or multiple GI genotypes (16E10 [42]). These antibodies bind far from the HBGA site (Figure 2E), either at the junction of the P2 and P1 subdomains (A1431), or at the side of the P1 subdomain, with few or no contacts on P2 (NORO‐320, VX22, 16E10). The A1431 epitope is accessible on the surface of the virion and appears to inhibit HBGA binding indirectly by steric hindrance. A different mechanism has been proposed for NORO‐320, which, because of its divalent nature as an IgA, can block HBGA binding by aggregating VLPs together. However, when virus neutralization was tested in a human intestinal enteroid assay, the Fab form, which was incapable of blocking HBGA binding, was at least as potent as the IgA form, showing that NORO‐320 can still neutralize NVs independently of virus aggregation and HBGA blocking [18].

On the other hand, it is not clear how NORO‐320, VX22, and 16E10 can bind to intact NV particles, as their lateral binding to epitopes on the P1 subdomain would imply a significant structural clash with the adjacent P‐domain dimer. Indeed, binding of both VX22 and 16E10, either as Fab or full IgG, to NV VLPs was shown to result in profound and irreversible changes in capsid structure [42, 43]. This type of antibody‐induced disassembly appears to be related to the dynamic flexibility of GII.4 NV virions [44], such that the VX22, 16E10, and NORO‐320 epitopes may be transiently accessible as the P‐domains alternate between “resting” and “raised” states. After initial binding, these antibodies may then act as a molecular crowbar to prise off VP1 dimers from the particle, with catastrophic consequences for virion infectivity. Interestingly, although VX22 was cross‐neutralizing, its potency was approximately 1000‐fold lower against GII.2 and GII.3 viruses compared with GII.4, even though its HBGA‐blocking IC50 was equivalent against these three NV genotypes [43]. This observation suggests that the vulnerability of infectious NV particles to antibody‐induced disassembly may differ between genotypes, and consequently, that the antiviral efficacy of antibodies such as NORO‐320, VX22, and 16E10 may be less broad than would be expected based on their inhibition of VLP binding to HBGA.

3.5. Polyomaviridae

The BK and JC polyomaviruses (PyV) both provoke disease only in immunosuppressed patients: polyomavirus nephropathy and hemorrhagic cystitis in transplant recipients (BKPyV), and progressive multifocal leukoencephalopathy (JCPyV) in a variety of clinical contexts. The four genotypes of BKPyV represent distinct serotypes [45, 46], while JCPyV has only one serotype. Polyomavirus capsids are composed of 72 VP1 pentamers, each associated with one copy of the VP2/3 protein, which is buried inside the capsid. Attachment receptors are sialic‐acid‐bearing glycans, such as the b‐series gangliosides GD1b and GT1b, for BKPyV. Internalized virions traffic to the endoplasmic reticulum, where capsid reorganization, resulting in exposure of the VP2/3 N‐terminal domain, is required for the transport of the particle to the cytosol. A nuclear localization signal peptide in VP2/3 then allows import of partially disassembled particles into the nucleus.

Screening of memory B cells from healthy donors led to the identification of the bNAb P8D11/MAU868 [5] and 41F17 [21], while 319C07/Potravitug [6] and the 41F17‐like mAb BK120 [22] were isolated from kidney transplant recipients with BKPyV DNemia. P8D11/MAU868 neutralizes all four BKPyV genotypes and has been tested in a phase II trial for the treatment of BKPyV DNemia in kidney transplant recipients (NCT04294472). It binds a conformational epitope on the lateral surface of the VP1 pentamer, distinct from the sialic‐acid binding pocket (Figure 2F). Structural and mutagenesis data suggest that, rather than blocking attachment, P8D11/MAU868 interferes with a postattachment step such as capsid disassembly or genome release [5]. Similarly, 41F17 binds to a quaternary epitope at the interface of two VP1 pentamers on the capsid surface, which, like the P8D11/MAU868 epitope, does not overlap the receptor‐binding site on the capsid surface (Figure 2F). On the other hand, the heavy‐chain CDR3 of 319C07/Potravitug (also now in phase II: NCT05769582) inserts deep into this site, making contacts with the conserved amino acids at the bottom of the pocket, rather than the variable residues around its rim [6]. Interestingly, 41F17 and P8D11/MAU868 also neutralize JCPyV, a property which is shared by the 29B1 antibody that was isolated from a patient recovering from progressive multifocal leukoencephalopathy [47]. The crystal structure of the BKPyV VP1 pentamer in complex with 29B1 ScFv (Figure 2F) shows that the 29B1 epitope has little overlap with the sialic‐acid‐binding pocket [23], but shares some contact residues with 41F17. Overall, at least three distinct epitopes on BKPyV and JCPyV capsids can be recognized by bNAb, two of which do not overlap the receptor‐binding site.

3.6. Papillomaviridae

More than 200 distinct HPV types are known, of which the high‐risk α‐HPVs are responsible for both cervical cancer and a significant proportion of oropharyngeal cancers. HPV capsids are composed of 72 pentamers of the L1 structural protein, associated with one copy of the L2 protein. Despite their diversity, all α‐HPVs appear to share the same two‐step mechanism of infectious entry. The first step, mediated exclusively by the L1 protein, is binding to heparan sulfate proteoglycans, which leads to reorganization of the capsid and exposure of the N‐terminus of the L2 protein. This is required for subsequent internalization of virus particles and their trafficking to the nucleus.

The virus‐like particles (VLPs) used in HPV vaccines only contain VP1, but since the L2 protein is buried inside the capsid, the outer surface of these VLPs is identical to that of the infectious virion. The 9‐valent HPV vaccine, comprising HPV6,11,16,18,31,33,45,52,58 VLPs, protects against the great majority of HPV‐induced cancers, whereas older bi‐ (HPV16,18) and tetravalent (HPV6,11,16,18) vaccines only cover the highest risk genotypes.

Vaccination induces mainly type‐specific neutralization [20, 25], and these antibodies predominantly target exposed structures like the FG loop, which is the main neutralizing epitope [48]. However, studies of the serological and clinical responses to bivalent and 4‐valent vaccines have indicated that there is some cross‐reactivity [25, 49] and cross‐protection [50] between HPV16 and the closely related HPV31, which was not represented in the bivalent and 4‐valent vaccines. The basis of this cross‐protection was determined in a recent study of HPV‐specific mAbs generated after vaccination with the 4‐valent vaccine [24]. In this paper, 6 of 68 human HPV16‐specific mAbs were shown to cross‐neutralize HPV31 to some extent, and four of the five antibodies that were studied further targeted the surface‐exposed FGb loop, which is conserved between HPV16 and HPV31. Although the mechanism of neutralization was not determined for the HPV16‐HPV31 cross‐neutralizing antibodies, previous work has shown that mouse monoclonals binding to an epitope defined by the FG and DE loops neutralize HPV16 at a postattachment step, presumably by preventing the exposure of the minor capsid protein, L2 [51].

Although they are not targeted by current HPV vaccines, the subdominant epitopes in the N‐terminal domain of the L2 protein (RG‐1 at residues 17–36 and a second epitope at residues 65–85 of the HPV16 L2 protein) are conserved between HPV types, and mouse antibodies specific for these epitopes are broadly neutralizing. Vaccine strategies based on L2 are under development [52, 53, 54], and their testing in clinical trials (NCT05985681) may lead to the identification of human HPV L2‐specific bNAbs.

3.7. Reoviridae

Rotaviruses (RtVs) belong to the Reoviridae family, which consists of viruses with a segmented double‐stranded RNA genome enclosed in a complex triple‐layered particle (TLP). RtVs are classified into seven antigenically distinct groups (A–G) based on the internal capsid protein VP6, which constitutes the middle layer of the TLP. Among these, only groups A, B, and C infect humans, with group A RtVs being the main cause of acute epidemic gastroenteritis, particularly severe in infants and young children. Within each group, strains are further subdivided into genotypes according to the genes encoding the two outer capsid proteins, VP7 (designated G, for glycoprotein) and VP4 (designated P, for protease‐sensitive protein). Proteolytic cleavage of VP4 generates VP5* and VP8*, which are the functional proteins displayed on the surface of mature virions. Infectious entry begins with the binding of VP8* to glycan receptors (such as HBGAs) on host enterocytes, the nature of which varies with RtV genotype. Virus particles are then internalized by endocytosis, following which partial decapsidation allows double‐layered particles (DLP) to translocate to the cytosol. Although some observations suggest the existence of one or more entry co‐receptors, none have so far been identified. Although six major genotypes (G1P[8], G2P[4], G3P[8], G4P[8], G9P[8], and G12P[8]) have historically caused most human infections, emerging nonvaccine genotypes and geographically variable strains continue to be reported. Licensed vaccines, Rotarix and RotaTeq, provide strong protection against severe disease but only partially cover this genotype spectrum.

In this context, recent studies have identified human monoclonal antibodies that target conserved epitopes on the VP4 protein. Nair et al. used fluorescence‐labelled particles to isolate RtV‐specific B‐cells from human intestinal epithelium, then characterized the antibodies they expressed [55]. Several cNAbs, able to neutralize more than one RtV genotype, were characterized, including VP7‐specific and VP4‐specific antibodies that cross‐neutralized different G or P genotypes, respectively. Of these, antibody #41 proved to be a high‐affinity, VP4‐specific bNAb, able to recognize P[1], P[3], P[4], P[6], P[7], and P[8] genotypes. The structural basis of its broad specificity was elucidated by high‐resolution cryo‐EM, and the Fab41 epitope was mapped to the junction of VP5* and VP8* proteins. The structure revealed that Fab41 neutralizes the virus by locking the VP4 spike in its upright conformation and preventing a structural transition that is critical for endosomal escape, rather than by blocking initial receptor binding [56]. Although the mechanism of the VP7‐specific cNAb described by Nair et al. was not investigated, it is well established that VP7‐specific mouse mAbs neutralize RtV by cross‐linking VP7 subunits on the outer capsid, and thereby inhibiting the uncoating of TLP [57, 58].

Finally, a role for VP6‐specific IgA in protection against RtV infection has been known for 30 years [59], and more recently it has been shown that VP6‐specific IgG also confers protection in a mouse model [60]. Since VP6 makes up the middle layer of TLP, it is only accessible on the DLP that reaches the cytoplasm of infected cells, implying that neutralization by VP6‐specific antibodies must be intracellular. Indeed, the antiviral effect of VP6‐specific IgG was dependent on the cytosolic receptor, TRIM21 [60]. The Rotateq vaccine does not induce antibodies to human RtV VP6, since the recombinant viruses that make up the vaccine are based on a bovine RtV scaffold, with the incorporation of human VP7 (G1, G2, G3, or G4) or VP4 (P[8]) proteins. Future vaccination platforms capable of inducing VP6‐specific antibodies could potentially result in broader protection against multiple RtV genotypes, since all group A RtVs carry the same VP6 serotype.