**2. Neutralising Antibodies Following a Natural Infection**

#### *2.1. Neutralising Antibodies against EBOV*

Mechanisms of antibody activity include neutralisation and Fc receptor-mediated effector functions (e.g., antibody-dependent cellular cytotoxicity, which is the killing of a target cell coated with antibodies by an effector immune cell; antibody-dependent cellular phagocytosis, which is a mechanism of clearance of antibody-coated pathogens or tumour cells by macrophages and natural killer cells and antibody-dependent complement deposition, which is the deposition of a complement component on infected cells mediated by IgG or IgM).

In the context of EBOV infection, it is not fully clear which function is the most associated with protection [46]. However, the characterisation of neutralising antibodies isolated from survivors and their subsequent evaluation in animal challenge models helped to improve our insight into neutralising antibody responses. It is crucial to underline that a lot of studies that analysed antibody responses to EBOV focused on IgG, especially total IgG in serum. However, the neutralisation activity can also involve pentameric IgM and monomeric/dimeric IgA present in serum and particularly abundant in mucosae [61].

Some studies suggested that neutralising antibody levels were modest early postinfection in humans but increased over time. Luczkowiak et al. analysed the neutralising activity of the plasma of three 2013–2016 West Africa epidemic survivors using a lentiviral EBOV-GP pseudotyped infection assay. They found that neutralising antibody titres increased up to 9 months post-infection [62]. Another study in Western patients confirmed this. Williamson et al. analysed B cell responses in four acute *Ebolavirus*-infected patients that had been repatriated to the US during the 2013–2016 West Africa epidemic. They found that between 1 and 3 months post-recovery, there was a low frequency of EBOV-specific B cells encoding for antibodies that displayed low neutralising activity. However, one neutralising antibody isolated in this study led to protection in a mouse EBOV challenge model [63]. A high diversity of neutralising antibodies may be needed for an efficient neutralisation, which could explain the delay in the development of neutralising responses [46]. However, it was clearly observed in several studies that long-term survivors developed robust and sustained neutralising antibody responses mainly targeting GP. Antibodies able to neutralise a pseudotyped EBOV GP were detected in survivors from the 1976 Yambuku outbreak 40 years later [64]. In addition, mAb 114, a potent neutralising mAb against pseudotyped EBOV GP lentivirus particles was first isolated from a survivor of the 1995 Kikwit outbreak, 11 years post-recovery [54]. This mAb is now a component of the recently approved drug named Ebanga®. The presence of persistent neutralising antibodies was confirmed in a larger cross-sectional study. Halfmann et al. measured anti-EBOV-specific humoral responses in 214 survivors and 267 close contacts (including 56 healthcare workers) in Sierra Leone, 15–32 (median 28) months post-recovery [65]. The study revealed 97.7% of survivors had antibodies against at least one of the three antigens (GP, VP40, NP) with 85% harbouring antibodies against all three. Of the survivors with a detectable antibody response against GP, all but one had neutralising titres, typically in the range of 1:128 to 1:152, but some exceeded >1:2048 [65]. A longitudinal study performed by Thom et al. also confirmed the maintenance of total and neutralising antibody responses over the course of several years. Thom et al. performed a longitudinal study looking at 117 survivors and 66 contacts, sampling patients from 3 to 14 months post-discharge, with follow up collections 12 and 24 months later [48]. The study found similar findings to Halfmann et al., where at 3–14 months 96% of survivors developed IgG specific response and this correlated well with total antibody titres (*r* 0.85; *p* < 0.0001). Interestingly, 96% of survivors maintained high titres of neutralising antibody against live Ebola virus, with a mean of 1/174. This

is ten-fold greater than the titres observed in patients one-month post-vaccination with an EBOV vaccine, however, it is not known what it means for the protection as correlates of protection are unclear [48]. A smaller longitudinal study investigating B cell responses carried out by Davis et al. followed the four 2013–2016 West Africa epidemic survivors repatriated to the US (previously mentioned) from discharge to 2 years post-infection. Davis et al. concluded that IgM levels typically declined after a few months, while IgG and IgA levels remained elevated [66]. This was hypothesised to be a result of antigen restimulation. However, opposing evidence by Thom et al. indicated no antigen stimulation of T cells was detected during their longitudinal study, suggesting there might be a different reason for the retention of high titres. This topic is still debated [48]. Davis et al. also characterised EBOV GP-specific monoclonal antibodies and they found that only a subset was capable of recognising cell-surface GP, a subset that contained neutralising antibodies [66]. Some research groups characterised the regions targeted by the neutralising antibodies. Bornholdt et al. analysed 349 monoclonal antibodies specific to GP isolated from a 2014 EBOV Zaire outbreak survivor. They found that 77% of these antibodies could neutralise live EBOV. After analysing the epitopes recognised by the neutralising antibodies, they reported that mAbs which targeted the GP stalk region proximal to the viral membrane were particularly effective at protecting mice against lethal EBOV challenge [59]. Recently, Khurana et al. described specific sites on GP mounting a neutralising antibody response in rabbits, which were protective in a lethal EBOV mouse model [67]. EBOV GP regions targeted by neutralising antibodies will be discussed in the Section 5.2.

Recently, Gunn et al. described the profile of humoral responses in EVD survivors from Sierra Leone. Interestingly, this study highlighted the development of both neutralising and polyfunctional IgG1 and IgA in survivors [68]. It is probable that there is a synergy between the neutralising activity and the innate immune effector functions via Fc receptor of antibodies. To our knowledge, it has not been shown in the context of EBOV, but it has been recently described in the context of SARS-CoV-2 [69].

The delay in neutralising antibody response early post-infection but long-term persistence of neutralising antibodies in survivors may suggest a role of neutralising antibodies in the clearance of the Ebola virus rather than in the early stages of infection resolution. The characterisation of GP epitopes inducing protective neutralising antibodies, as well as a better understanding of antibody polyfunctionality at systemic and mucosal levels could be very valuable and may help to improve the efficacy of current antibody-based therapeutics and vaccines.

#### *2.2. Neutralising Antibodies against Non-EBOV Filoviruses*

Aside from EBOV epidemics, SUDV, BDBV and MARV outbreaks make up the vast majority of other recorded human filovirus infections and so it is important to consider the analysis of immune responses to these viruses. This is especially true because historically their small outbreaks have often occurred in remote regions of Africa with limited health infrastructure which restricts the collection of patient samples and data. Consequently, our understanding of these diseases is less detailed and, despite being equally deadly, there are currently no licensed therapeutic options for non-EBOV filoviruses.

Sobarzo et al. analysed the neutralising antibody responses of survivors of the 2000–2001 Gulu [70] and 2012 Kibaale [71] outbreaks caused by SUDV. Serum samples were collected 12 years or 3 years post-recovery, respectively. Similarly to EBOV, a robust and persistent SUDV-specific antibody response, mainly targeting GP and NP, was observed in both cohorts. Five of five Kibaale survivors and five of six Gulu survivors displayed antibodies capable of neutralising the whole SUDV by plaque reduction neutralisation assay (PRNT) [70,71].

Marburg has been responsible for two outbreaks exceeding case numbers of 100. Stonier et al. performed a longitudinal study following PCR positive survivors of the 2012 MARV outbreak in Uganda sampling patients 9, 15, 21 and 27 months after the outbreak [72]. All survivors had an antibody response to MARV GP and NP and most had

a response to VP30 and VP40. Whilst total antibody titres remained high in all patients throughout the study, at 9 months, only two of six survivors demonstrated, via PRNT, a neutralising antibody response between 1:20 and 1:40 which diminished by 21 months and disappeared by month 27 [72]. These results suggest a more rapid waning of neutralising antibodies following MARV infection compared to EBOV infection. Could this difference be linked to the presence of extra GP products with EBOV, which is not observed with MARV? The mechanisms behind these results are unclear.

Other researchers isolated monoclonal antibodies from peripheral blood B cells from survivors of the 2007 BDBV outbreak in Uganda. They found a large proportion of BDBV GP-specific monoclonal antibodies including some antibodies which strongly neutralised BDBV but also SUDV and EBOV using chimeric filoviruses [57].

Together these results show that robust neutralising antibody responses are also induced following an infection with non-EBOV filoviruses.

#### **3. Antibody-Based Therapeutics and Vaccines**

#### *3.1. Monoclonal Antibodies*

Therapeutic mAbs have been a field of great interest. Two EBOV specific mAb therapeutics, REGN-EB3 (Inmazeb™, Regeneron Pharmaceuticals, Tarrytown, NY, USA) and mAb114 (Ebanga™, Ridgeback Biotherapeutics, Miami, FL, USA) having recently been approved by the Food and Drug Administration (FDA, Silver Spring, MD, USA).

Monoclonal antibodies can be created in a laboratory or isolated from a specific B cell clone from survivors or vaccinated animals. Monoclonal antibody therapy is the passive transfer of immunity in the form of one, or a small number of, antibodies targeting a single epitope of a protein, usually GP [73]. This differs from polyclonal antibodies elicited from vaccination or infection where antibodies target a higher number of epitopes on the viral protein [73]. One of the advantages is that mAbs have an immediate and potent effect and can be used as a post-exposure therapeutic and can also be administered prophylactically with a relatively long half-life, e.g., mAb114 24.2 days [56,74].

The GP1 head is home to the RBD and therefore an obvious target for mAbs including mAb 114. Mab114 is the only effective monotherapy and is licensed under the name Ebanga™ following impressive results from the Pamoja Tulinde Maisha (PALM) trial where it demonstrated a protective effect by reducing mortality to 35.1% [75,76]. Odesivimab is also part of the licensed Inmazeb™ cocktail but it is non-neutralising because it binds to the GP1 head but a little further away from the RBD than mAb114 and consequently does not neutralise (Table 1) [77]. FVM04 is a broadly neutralising antibody that binds to the GP1 head of the GP and can neutralise EBOV and SUDV in addition to providing good efficacy in mouse models (Table 1) [78].

The GP1 base is potentially the most common target for neutralising antibodies. It is the target of KZ52, which is a research standard for neutralisation, even though it was shown not to be protective in NHP trials. KZ52 binds to an epitope on the GP1 base that at least partially overlaps with several other potential therapeutics e.g., 2G4 and 4G7 [79]. 2G4 and 4G7 are both neutralising antibodies from the ZMapp cocktail which share overlapping epitopes on the GP1 base (Table 1) [80]. Maftivimab neutralising antibody of the Inmazeb™ cocktail also binds to the GP1 base. While it does not share an epitope with the aforementioned mAbs of the Inmazeb™ cocktail, it is thought to have the same mechanism of neutralisation (Table 1) [77].

The GC is a highly variable region that is usually thought of as non-neutralising, but it was found to be a target for several therapeutic mAbs. Saphire et al. via the Viral Haemorrhagic Fever Consortium demonstrated that antibodies targeting GC were mostly non-neutralising but a small subset of neutralising antibodies were detected [81]. The GC is also a target for Atoltivimab which is a neutralising antibody in the licensed Inmazeb™ cocktail and able to activate effector functions (Table 1) [77,81]. Furthermore, the GC contains the epitope for the broadly neutralising EBOV-548. This mAb binds to a conserved GC epitope in EBOV and BDBV. It can neutralise both viruses and effectively

activates effector functions [82]. EBOV-520, which binds to the GP base region, strongly demonstrates co-operative binding. It binds to its epitope with 5× greater affinity once EBOV-548 is bound to the GC. Together, EBOV-520 and EBOV-548 can fully protect NHPs from mortality, though EBOV-520 alone has also demonstrated partial protection in animal models [82–84]. Finally, 13C6 is a non-neutralising antibody from the ZMapp cocktail that binds to the GC but is removed during cathepsin cleavage. It also effectively activates Fc functions [80].

ADI-15878 is a neutralising pan-ebolavirus mAb which binds to conserved regions on the HR1 and IFL domains of EBOV, SUDV and BDBV [85]. When administered alone, survival rates post-challenge in guinea pigs of 33–50% were observed. However, when paired with ADI-23774 in a cocktail called MBP134 at a high dose, it provided 100% protection [86]. CA45 is another pan-ebolavirus mAb with neutralising activity against EBOV, SUDV, BDBV and RESTV that binds to the IFL and GP1. This mAb was shown to be protective in animal models, both individually and in combination with FVM04 in NHP trials (Table 1) [87]. CA45 and FVM04 can also be given in a cocktail with MR191 an anti-MARV and RAVV mAb, that provides 100% protection against NHPs challenged with MARV when tested alone and within the cocktail [87,88]. MR191 binds to residues in the RBD that are essential for receptor binding and are hence highly conserved among related filoviruses. MR191 however, does not neutralise EBOV because the GC obscures the epitope whereas MARV has a disordered and flexible GC that gives the mAb access to this epitope [89,90].

The PALM trial is the only large clinical trial to date evaluating anti-EBOV experimental therapeutics in a field setting during the 2018 outbreak in the Congo, with patients presenting during various stages of disease following positive RT-PCR. It compared ZMapp, mAb114, REGN-EB3 and Remdesivir in 681 participants in a 1:1:1:1 ratio. One striking result was that ZMapp and Remdesivir were found to have no statistically significant effect on reducing EBOV mortality which stood around 50% [76]. The study did find however, that REGN-EB3 (Inmazeb™) and mAb114 (Ebanga™) had a protective therapeutic effect, reducing mortality to 33.5% and 35.1%, respectively. The effectiveness of the treatments was especially high among patients who received treatment early, with an estimated 11% increase in the risk of mortality with each day the treatment was delayed. Furthermore, patients who were classified to have a high viral titre also had a much higher mortality of 67% even with mAb114 or REGN-EB3 [76].

**Table 1.** A table describing the characteristics and epitopes of several key mAbs, with a figure integrated highlighting the residues on the EBOV GP that make up the epitopes of each of the mAbs in the table. The critical/known residues are in red, and the putative residues highlighted in yellow. The protein structure was generated in The PyMOL Molecular Graphics System, Version 2.0 Schrödinger, Delano Scientific LLC, Berkeley, CA, USA using the PDB accession: 5JQ3.

