3.3.1. EBOV Vaccines

The role of antibody responses in protection following vaccination with a replication competent recombinant VSV virus encoding for Zaire EBOV Kikwit 1995 GP (rVSV-ZEBOV) was firstly analysed in preclinical studies. Marzi et al. vaccinated five NHP groups 28, 21, 14, 7 and 3 days before challenge [102]. Surprisingly, an induction of EBOV-GP specific IgG responses was already reported 3–7 days post-vaccination in NHPs. The research group found a partial protection of animals vaccinated 3 days before the challenge and a full protection of animals immunised ≥7 days before the challenge. This indicates that rVSV-ZEBOV may elicit very rapid humoral response and protection [102]. Another preclinical study reported that antibodies were sufficient at protecting mice from infection following immunisation with rVSV-ZEBOV. Meanwhile the depletion of CD8+ T cells did not compromise protection [103]. Similar results were observed by Marzi et al. in NHPs. CD8+-depletion did not impact on survival of rVSV-vaccinated animals following a challenge [104]. Wong et al. also analysed the role of humoral response following rVSV-ZEBOV vaccination in NHPs and found significantly higher total IgG titres in the serum of animals which survived post-challenge compared to the non-survivors [105]. The same team also confirmed long-term protection by challenging vaccinated animals 6–12 months post-immunisation with rVSV-ZEBOV. They observed that the levels of EBOV GP-specific IgG antibody, measured immediately before challenge, correlated with protection, whereas neutralising antibody were not always a reliable measure of protection in their animal model [106]. Another research group determined whether rVSV-ZEBOV vaccine could be used as a post-exposure treatment in Rhesus macaques. They found that four of eight macaques were protected if treated up to 30 min following a lethal infection. While the differences in cellular responses where minimal between the animals that survived and those that succumbed to the virus, there was a significant difference in neutralising responses. Indeed, neutralising antibodies were detected on days 14–36 post-challenge in animals that survived the infection, while the humoral response was not detected in animals that succumbed to the infection, suggesting a critical role for the humoral response [107]. Another study showed that NHPs vaccinated with rVSV-ZEBOV were also protected from EBOV aerosol challenge. Interestingly, upon measurement of circulating rVSV-ZEBOV specific-IgG responses post-vaccination and post-challenge, antibody responses were found to increase post-challenge. While the neutralising capacity of the antibodies was not analysed, they found there was no evidence of IFNγ or TNFα production in CD4<sup>+</sup> or CD8+ before or after the challenge, which further suggests a major

role of humoral response in protection [108]. Qiu et al., compared the protective immune responses in NHPs immunised with rVSV-ZEBOV by intramuscular (IM), intranasal (IN) or oral route (OR). They observed that IgG, IgA and IgM antibody responses were detected in the serum of all vaccinated animals independent of the route of vaccine administration. Post-challenge, IgG and IgA titres increased, while IgM titres did not exceed the levels observed post-vaccination. Globally, IgM titres ranked IN ∼= OR > IM, with an IN titre 2.4 times higher than IM, while IgA and IgG responses ranked IN > OR > IM (IN 6.8-fold > IM) and IN > OR ∼= IM (IN 9.0-fold > IM), respectively. They also analysed the level of neutralising antibodies in the sera 21 days post-vaccination and a few days prior to the challenge. Whilst neutralising antibody titres were relatively low in all animals, the OR (OR > IN >IM) produced the highest neutralising antibody titres. This study clearly highlighted the impact of immunisation with rVSV-ZEBOV on the induction of neutralising antibody responses [109]. Interestingly, NHPs previously infected with simian-human immunodeficiency virus were vaccinated with rVSV-ZEBOV. Following EBOV challenge, 4/6 animals survived. None of the six animals had a detectable antibody response by the day of challenge but three animals which survived developed a modest antibody response post-challenge, which suggests a role of antibodies in survival of these animals [110].

Recombinant VSV-ZEBOV (Erbevo®, Merck, Kenilworth, NJ, USA) is now the first fully licensed EBOV vaccine. This vaccine is delivered in humans in one dose and its safety and efficacy were also evaluated in clinical trials. Open-label, dose-escalation phase 1 trials were performed in healthy volunteers in Europe and Africa. They measured a persistent EBOV GP-specific antibody response in all vaccinated participants and higher neutralising responses when individuals received a higher dose of rVSV-ZEBOV vaccine [111]. Other phase 1 clinical trials reported very similar results in the US, Canada or Europe. In each study, a dose-dependent neutralising response was observed but occasionally neutralising responses failed to be detected in some individuals who received a lower dose of rVSV-ZEBOV vaccine [112–114]. One study provided clinical data on rVSV-ZEBOV efficacy, a phase 3 trial in Guinea, where a ring vaccination program was employed to include cases, contacts and contacts of contacts. The study reported 100% efficacy after 10 days and found that, 32 days following the detection of the first case in a vaccinated cluster, no new cases were reported, highlighting the ability of this vaccine to prevent transmission [115]. Despite the potential for bias relating to the standard of care between the experimental and control groups in the protocol of this study (as reported by some researchers), this latter highlights the high efficacy of the rVSV-ZEBOV vaccine in humans [116]. Halperin et al. showed that rVSV-ZEBOV produced strong antibody titres with 94% of participants becoming seropositive after 28 days and 91% remaining seropositive after 24 months. Geometric mean titres (GMT) rose from below the assay detection limit of 36.11 to 1262.0 after 28 days before gradually decreasing but retaining a high GMT of 920 after 24 months. The GMTs of neutralising antibodies, measured by PRNT, were high and continued to increase, peaking at 18 months with a plateau at 24 months [117]. Interestingly, Khurana et al. studied the human antibody repertoire in individuals vaccinated with rVSV-ZEBOV. They reported a high initial neutralising IgM immune response before IgG becomes the dominant subtype, which may explain the rapid protection provided by the vaccine. They demonstrated a higher diversity of antibody epitopes in vaccinees who received 20 million plaque-forming units (PFU) compared to those who received 3 or 100 million PFU. Another finding was that higher levels of neutralising GP-specific antibodies were induced after a single vaccination with 20 or 100 million PFU. A boost did not improve neutralising antibody response [118].

Adenoviral vector vaccines have been also developed in the context of EBOV. These vaccines are well known to induce robust T cell responses [119–121]. Wong et al. evaluated the antibody responses in the context of vaccine candidates based on AdHu5 expressing Zaire EBOV GP [105]. They compared the immune responses and survival rates of knockout mice (Rag-1−/−, B cell−/−, CD8<sup>+</sup> T cell−/−, IFN-γ−/−, and CD4<sup>+</sup> T cell−/−), reporting that B cell and CD4+ T cell responses were the most critical in the development of a protective immune response against a mouse adapted strain of EBOV [105]. When repeated in guinea pigs, the average titres of anti-EBOV GP antibodies and neutralising antibody postchallenge were significantly higher in survivors than in non-survivors [105]. Interestingly, they also performed a similar study in NHPs. In this animal model, they found a higher anti-GP IgG response in survivors, but they did not detect any differences in neutralising antibody response between the survivors and non-survivors. However, a difference in T cell responses between the cohorts was detected in this latter model [105]. Chen et al. observed a durable EBOV-neutralising response in mice vaccinated with a prime-boost vaccine regimen based on a chimpanzee serotype 7 adenovirus expressing EBOV GP and a truncated version of EBOV GP1 protein [122].

In humans, the two-phase vaccine candidate Ad26.ZEBOV (Zabdeno®) and MVA-BN-Filo (Mvabea®), produced by Johnson&Johnson (New Brunswick, NJ, USA), is a vaccine candidate that has progressed well through clinical trials and has received marketing authorisation by the European Commission in July 2020. This is a viral vector vaccine, that uses two heterologous doses. The first dose is a non-replicating Adenovirus type-26 encoding EBOV GP and the second dose is a multivalent recombinant Modified Vaccinia Ankara vector-based vaccine encoding the GPs from EBOV, SUDV, MARV and the nucleoprotein of TAFV [123]. Anywaine et al. compared the immunogenicity of heterologous two-dose Ad26.ZEBOV and MVA-BN-Filo vaccination regimens in a phase 1 trial in Uganda and Tanzania. They demonstrated a robust immunogenicity when Ad26.ZEBOV was administered as the first dose followed by MVA-BN-Filo. This regimen promoted higher neutralising and total antibody titres with 93% of participants achieving seropositivity at the time of the second dose, reaching 100% 21 days after the second dose. Neutralising antibody titres were initially low but peaked at 21 days following the second dose 100% of participants in the 56-day interval group, before decreasing and stabilising 180 days post-initial vaccination. Titres plateaued until the final time-point at day 365 [124]. A similar phase 1 clinical trial was run in Kenya, where high levels of neutralising GP-specific antibodies were detected and sustained up to 360 days after the first dose [125].

Globally, preclinical and clinical studies demonstrated a key role of antibody responses in protection post-vaccination, particularly in the context of rVSV-ZEBOV. However, the correlation between neutralising responses and protection following vaccination is still debated.

Finally, there is uncertainty regarding the efficacy of postexposure mAb treatments following a recent vaccination with an EBOV vaccine. This scenario of exposure may happen early post-injection when vaccination is not yet fully effective. Cross et al. demonstrated in rhesus macaques that vaccination with rVSV-ZEBOV 1 day prior to EBOV challenge followed by anti-EBOV GP mAb MIL77 treatment 3 days later increased the rate of survival compared to animals vaccinated or treated with MIL77 only [126]. However, additional data is needed to draw a robust conclusion about a potential synergy between vaccination and antibody-based immunotherapy.

## 3.3.2. MARV Vaccine Candidates

To date no MARV vaccine has been approved yet, but several vaccine platforms have been attempted, including a recombinant VSV vector expressing MARV GP which showed efficacy in NHPs both post-exposure and prophylactically [127]. Jones et al. demonstrated that a single intramuscular injection led to 100% protection of vaccinated NHPs from a lethal MARV challenge [128]. Daddario-DiCaprio et al. confirmed these results with the same vaccine by proving the 100% protection following a challenge with heterologous MARV strains and RAVV [129]. Both studies reported high IgG titres in vaccinated NHPs but an absent or very low neutralising response. More recently, Mire et al. conducted a study involving six NHPs immunised with a rVSV-MARV-GP platform prior to challenge 13 months later. Immunisation elicited strong total antibody titres (between 1600 and 12,800) which remained elevated throughout the study. Regarding the neutralising antibody responses determined by PRNT, all NHPs had a neutralising response at day 28 post-vaccination but the titres were relatively low and by the day of challenge, two had

no neutralising antibodies [130]. Together, these results suggest a less important role of neutralising antibodies in protection from MARV challenge.

Interestingly, a follow-up study by Daddario-DiCaprio et al. showed that immunisation with rVSV-MARV vaccine up to 20–30 min post-challenge protected all NHPs. All treated animals showed low to moderate amounts of IgM by day 6 post-challenge. Four of the five vaccinated animals developed a moderate IgG response by day 10 post-challenge. However, PRNT demonstrated low amounts of neutralising antibodies between days 6 and 37 in the plasma of all immunised animals [131]. Similar results were confirmed by Geisbert et al. In addition, a partial protection was reached when the vaccine was administered 24 h and 48 h post-challenge, when five out of six and two out of six macaques were protected, respectively. All animals which survived to challenge showed moderate to high levels of MARV-specific IgG in sera, whereas animals that died did not have any detectable IgG responses [132]. The sum of these results suggests the importance of humoral response following rVSV-MARV vaccination even though the neutralising activity may be less important with rVSV-MARV vaccine compared to rVSV-EBOV vaccine.

### **4. Methods to Measure Neutralising Antibody Responses**

Researchers use different assays to evaluate neutralising antibody titres in serum or other biological fluids. To measure EBOV-specific neutralising antibody responses, the gold standard is the PRNT, which uses authentic live virus to measure the number of plaques formed upon infection of a cell line, such that if the sample is capable of neutralisation, fewer plaques would be observed [133]. Briefly the biological fluid (e.g., serum) containing the antibodies is mixed with the authentic live virus, usually for 1–2 h, and used to infect a permissive cell line. An agarose overlay is added for 5–7 days to avoid a 'too rapid' spreading of the virus. Finally, the cells can be fixed and plaques are counted, often manually. The PRNT takes several days and its development at a large-scale can be limited. It is the reason why microneutralisation assays (MNA) were also developed in the context of EBOV [97]. The concept of MNA is very similar to PRNT but 1 h post-incubation of cells with the mix EBOV/serum, the cells are washed and fixed for 1–2 days. Finally, infected cells can be detected using anti-EBOV and secondary antibodies. The number of spots can be counted using an imaging system [58]. Both methods, PRNT and MNA, are robust. However, EBOV is classified as a biosafety safety level (BSL) 4 organism as it is a very dangerous pathogen. BSL-4 work requires Class III safety cabinets, very specific pressure conditions in the lab, very strict decontamination procedures of the materials and trained staff. This increases the time and cost of such experiments, as well as limiting the number of sites with such capabilities [134].

Pseudotyped virus neutralisation assays can be a more flexible and manageable approach. Pseudotyped viruses (pseudoviruses) are recombinant viruses with core and envelope proteins derived from different viruses. They carry full or partial/modified sequence genomes to give rise to replication proficient or deficient virus. A reporter gene (e.g., luciferase, green fluorescent protein) is incorporated into the genome of the vector to detect a reduction of luminescence or fluorescence by neutralising antibodies, following infection of a permissive cell line. In the context of EBOV, recombinant VSV or lentiviral viruses are often used as a backbone to express EBOV GP. The advantage is that the recombinant pseudotyped viruses can usually be used at lower containment levels than EBOV. However, data generated from different pseudovirus systems can vary. Indeed, parameters such as the type of backbone, the reporter system and the expression of GP on the backbone may impact the sensitivity and the specificity of the assays, as well as their ability to accurately detect neutralising antibodies. Therefore, some studies have investigated conditions to improve the correlation with live EBOV neutralisation assays. We compared side by side two systems of EBOV GP pseudoviruses. Steeds et al. showed that the VSV luciferase pseudovirus system had a greater correlation (r = 0.85 + *p* < 0.0001) to the live EBOV assay than the HIV-1 system (r = 0.54 + *p* = 0.0004) [135,136]. Similarly, Wilkinson et al. ran a study across several laboratories and reported that labs using the VSV

system reported higher correlation with a wildtype Ebola virus neutralisation assay (r = 0.84 and r = 0.96). Konduru et al. also confirmed a high correlation between the VSV platform and live EBOV neutralisation assay [137,138]. A limitation of the pseudovirus system is the absence of proteins aside from GP. While neutralisation antibodies mainly target EBOV GP, we cannot exclude the possibility that neutralising antibodies may be directed against other key proteins (e.g., VP40, other forms of GP) and cannot be detected using a pseudovirus system. For instance, VSV systems do not produce sGP, which could have an impact on neutralisation outcomes. Interestingly, Saphire et al. evaluated the impact of sGP on the results generated by three different neutralisation assays: VSV platform, authentic live EBOV, EBOVΔVP30-RenLUc virus. The VSV system did not express sGP whereas the other systems expressed wildtype sGP. Globally, they observed that the presence of sGP did not prevent neutralisation and in some cases, neutralisation was higher in the sGP-expressing assays [81]. When pseudoviruses are replication-deficient, this system can only detect the antibodies which prevent the entry of the virus into the cell. For example, the system cannot detect the impact of antibodies on the production of new viral progeny.

Pseudovirus platforms are surrogate systems to identify neutralising antibodies. However, the production of recombinant filoviruses expressing a reporter protein, which can help to monitor and quantify the infection, may be useful to improve the characterisation of antibodies able to neutralise EBOV or other filoviruses. Thus, an alternative approach to measuring filovirus-specific neutralising antibody responses is to use chimeric filoviruses. Chimeric filoviruses are man-made viruses composed of components from two filoviruses. Usually, one live filovirus is used as a backbone and its GP is replaced by a heterologous filovirus GP (e.g., EBOV GP). The potential difference in neutralisation-sensitivity between a chimeric filovirus and a pseudovirus has to be considered in order to avoid producing misleading results in the detection of neutralising antibodies [139]. Llinykh et al. compared the neutralisation-sensitivity of VSV-expressing filovirus GP with EBOVexpressing heterologous filovirus GP from BDBV, SUDV, MARV and LLOV. They reported that chimeric filoviruses were as sensitive as authentic filoviruses expressing the same GP. However, VSV chimeric viruses were more sensitive to antibody neutralisation than authentic filoviruses [139]. The analysis of neutralising antibody responses can also be performed by surrogate virus neutralisation tests. Whilst these tests have been broadly developed for other viruses such as SARS-CoV-2, to our knowledge they have yet to be developed for EBOV or other filoviruses.

To conclude, the choice of platform used to measure neutralising antibody responses is crucial in order to yield the most accurate results. However, the neutralising antibody titres generated with different platforms may be difficult to compare. It is the reason why the use of a WHO research standard in each assay can help to compare the titres between the studies and consequently, the neutralisation efficacy determined in each study.

#### **5. Viral Neutralisation**

#### *5.1. Main Mechanisms of Neutralisation*

Neutralising antibodies bind epitopes which block critical amino acids in the virus's lifecycle, typically epitopes involved in cell entry or the production of new viral progeny. The different mechanisms of neutralisation are shown in Figure 2 (pink boxes).

The vast majority of EBOV neutralising antibodies are targeted towards the GP as it contains multiple crucial epitopes exposed on the surface, particularly in the RBD [140]. Neutralising antibodies which bind to the EBOV GP block the interactions between the EBOV GP and NPC1 and prevent receptor binding via a direct binding to the RBD or in a way to block access to the RBD, thus preventing any subsequent binding [90]. The presence of neutralising antibodies can also prevent EBOV fusion with the endosomal membrane, usually necessary for the virus to escape into the cytoplasm. Indeed, neutralising antibodies can impede this by binding to the GP1/GP2 interface, clamping the GP and preventing the conformational changes in the GP required for this process (Figure 2) [17]. Finally, cathepsin cleavage usually removes the GC and MLD to expose the RBD mediating access

for receptor binding. Neutralising antibodies can bind to GP in such a way to partially or totally prevent cleavage. For example, mAb CA45 disrupts cleavage to produce partially cleaved GPs, while mAb100 almost completely inhibits cleavage [54,140,141]. Though not clearly described in the context of EBOV, some antibodies can neutralise by altering the structure or conformation of the viral GP to the point that it is unable to perform a critical function, e.g., receptor binding (see Figure 2) [142]. VP40 is critical for viral budding and the production of new progeny and therefore its potential as a therapeutic target has been investigated. For example, anti-VP40 mAbs have been shown to neutralise EBOV by preventing the formation of new progeny (Figure 2). This is bolstered when administrated in cocktails with mAbs targeting different sites on the VP40 protein or with GP-specific antibodies [143,144]. Antibodies, particularly polymeric IgM and IgA, can also agglutinate virus particles, forming aggregates that reduce the contact between the virus and the cell and affect subsequent entry, in addition to signalling, for antibody-dependent phagocytosis (Figure 2) [145–147].

**Figure 2.** The mechanisms of action of neutralising antibodies (pink boxes) and non-neutralising antibodies (yellow boxes). Diagram was created with BioRender.com (©BioRender 2021, accessed in June 2021) and 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.

Finally, it is important to highlight that based on some studies performed by the Viral Haemorragic Fever Immunotherapeutic Consortium, 5% of non- or weakly neutralising EBOV-specific antibodies tested in the panel were found to be protective when assessed in vivo proving a crucial role for non-neutralising antibodies in protection [81]. Perhaps more protective antibodies have not yet been discovered, as the assays of choice for the analysis of antibody responses are largely based on neutralisation only. Indeed, nonneutralising antibodies can stimulate an array of processes that aid in viral clearance including initiating complement deposition, contributing to viral inactivation, phagocytosis and complement-dependent cytotoxicity (Figure 2) [148]. Monoclonal antibodies are also capable of interacting with Fc receptors activating antibody-dependent cell-mediated

cytotoxicity and antibody-dependent cellular phagocytosis [68,146,148,149]. Furthermore, it has been reported that non-neutralising antibodies have been shown to have synergistic neutralising ability, whereby they bind to the viral GP, altering the structure of an epitope making it more accessible for neutralising antibodies to bind. This has been documented for both EBOV and SUDV [150]. The three main mechanisms of non-neutralising antibodies are mentioned in Figure 2 (yellow boxes). Some monoclonal antibodies isolated from EBOV survivors have been shown, at sub-neutralising concentrations, to cause antibodydependent enhancement via antibody-dependent phagocytosis, especially in macrophages which are thought to be the preferred target cell of EBOV. This could play a potential role in antibody-dependent enhancement of EBOV infection leading to a more severe disease and worse clinical outcomes [151].

#### *5.2. Regions Targeted by Neutralising Antibodies*

Antibodies binding to the GP head are frequently neutralising because they block the function of the RBD, hence inhibiting cell entry and GP binding to NPC1 which is required for escaping the endosome.

The Viral Haemorrhagic Fever Immunotherapeutic Consortium investigated 168 EBOV specific mAbs for their ability to neutralise live virus and pseudotyped VSV and measured their ability to protect BALB/c mice following a challenge with mouse-adapted EBOV. The study found that antibodies specific for the GP1 head, base, the HR2 and IFL provided statistically significant protection. All of the same domains, except the GP1 base were targeted by more potently neutralising antibodies. Consequently, the neutralising ability of the antibodies showed a statistically significant correlation with protection, with a Spearman's rank value of 0.65 (the closer rank value is to 1, the stronger the association between the ranks), based on the percentage of infected cells using an authentic EBOV [81]. HR1, HR2 and IFL are also important neutralisation targets given that HR1 and HR2 both facilitate IFL insertion into the host membrane [12]. Some studies clearly described that the neutralising activity of CA45 or ADI-15878 and ADI-15742 were linked to IFL binding [58,93]. Finally, the base of GP1 is also a target for neutralisation, when an antibody binds, it prevents the release of the base which clamps HR1 and HR2 to instigate the conformational changes required for viral escape from the endosome, thereby preventing these conformational changes from taking place [12].

The GC and MLD are designed to shield GP from humoral responses and are cleaved by cathepsin in the endosome. They play no part in the receptor binding or fusion of the EBOV GP with the cell membrane, hence are not always targets for neutralisation [81]. This is demonstrated by the potently neutralising mAb114 and the non-neutralising 13C6 which have overlapping epitopes binding to the GP1 core and at least partially to the GC. MAb114's binding to the GC isn't critical and remains bound and neutralising when the GC is removed following cathepsin cleavage. Meanwhile, residues on GC are critical for 13C6 binding and when removed, 13C6 no longer remains bound to the EBOV GP, rendering it non-neutralising and allowing EBOV to freely interact with NPC1 [75].

#### **6. Strategies to Prevent Immune Escape from Neutralising Antibodies**

The efficacy of antibody-based therapeutics and vaccine-induced antibody responses may be reduced over time due to viral immune evasion. Immune evasion may impair antibody binding and consequently neutralisation reducing protection provided by the mAbs-based therapeutics, vaccines or a previous natural response.

A major drawback of mAbs is that they are specific to one epitope. Consequently, they are more vulnerable to immune escape from rapidly mutating viruses, and this has been observed with many other viruses such as influenza, HIV or SARS-CoV-2 [77,152–155]. In an experimental setting, Steeds et al. showed that GP variants G74R, P330S and H407Y were capable of readily escaping the WHO research neutralisation standard KZ52. Worryingly, KZ52 has overlapping epitopes with several other promising mAb therapeutics, e.g., CA45, 4G7 and more (Table 1) [135]. A solution to this concern is the use of mAb cocktails, that

target several distinct epitopes, hence reducing the chance of immune evasion. It is the reason why numerous mAbs discussed previously are part of a cocktail. Furthermore, this strategy also allows the addition of non-neutralising antibodies which may be more effective at initiating effector functions to increase the efficacy of the drug e.g., 13C6 or Odesivimab [77,80]. Another tactic that is employed involves targeting highly conserved epitopes that are often critical to viral fitness.

Currently, there are only two licensed EBOV-specific mAbs, with no licensed therapeutics on the market for the treatment of any other filovirus. Thus, pan-ebolavirus or pan-filovirus mAbs would significantly improve filovirus treatment options. An example of which includes CA45 and FVM04 that when used in combination as a cocktail provide 100% protection against both EBOV and SUDV in NHP and mouse models [87,93]. Furthermore, when MR191, is added to the cocktail, it also provides 100% protection against MARV [87]. Interestingly, some pan-neutralising mAbs were isolated from a 2013–2016 West Africa EBOV epidemic survivor and showed protection in mouse and ferret models [58]. Gilchuk et al. also isolated mAbs from the 2013–2016 West African epidemic and 2018 DRC outbreak survivors targeting epitopes at the base region of the GP which displayed pan-neutralising and protective abilities in mice, guinea pigs and ferrets [83]. Some survivors from the 2014 Boende EVD outbreak also mounted pan-filovirus serum neutralising responses [156].

There has also recently been a push for cross-reactive pan-filovirus vaccine candidates. Experimental DNA- and VSV-based vaccines encoding multiple GPs of various filoviruses have been tested in NHP models. For example, Keck et al. immunised two cynomolgus macaques three times with a trivalent GP cocktail consisting of EBOV/SUDV/MARV. Sera from the macaques had strong IgG measured by ELISA. While neutralising titres against the GP of all three filoviruses, were detected, although significantly lower neutralisation was observed with MARV. As MARV is more distantly related, it is not surprising that less cross-reactivity is shared [157].

The development of pan-neutralising therapeutics and vaccines is likely the future of filovirus research. These therapies may prove to be a highly beneficial resource in the poor regions where filoviruses are endemic.

#### **7. Conclusions**

This review has summarised that antibodies, and in particular neutralising antibodies, are an important line of defence against filovirus infection but current research falls short of defining them as a correlate of protection. Preclinical and clinical studies were not able to determine the level of neutralising antibodies necessary to protect an individual.

Monoclonal antibodies are the first licensed post-exposure therapeutics for EVD, which, coupled with expanded use of vaccines, have the potential to save a large number of lives. However, vaccine access in the field is still limited and mAb therapies are not perfect. In addition, the emergence of EBOV variants which could reduce the efficacy of mAbs and vaccine-induced humoral responses has to be considered. Consequently, newer and more effective antibody-based treatments containing more broadly neutralising antibodies, as well as the development of vaccines based on epitopes eliciting cross-reactive neutralising antibodies, would be very beneficial.

**Author Contributions:** A.H. and S.L. wrote the manuscript and designed the figures. C.B., J.M., T.T. and M.W.C. revised the manuscript. All authors contributed to the article and approved the submitted version. All authors have read and agreed to the published version of the manuscript.

**Funding:** The APC was funded by the US Food and Drug Administration (grant number: 75F40120C 00085).

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Conflicts of Interest:** The authors declare no conflict of interest.
