*Review* **Development and Structural Analysis of Antibody Therapeutics for Filoviruses**

**Xiaoying Yu <sup>1</sup> and Erica Ollmann Saphire 1,2,\***

<sup>2</sup> Department of Medicine, University of California, San Diego, La Jolla, CA 92093, USA

**\*** Correspondence: erica@lji.org; Tel.: +1-858-752-6791

**Abstract:** The filoviruses, including ebolaviruses and marburgviruses, are among the world's deadliest pathogens. As the only surface-exposed protein on mature virions, their glycoprotein GP is the focus of current therapeutic monoclonal antibody discovery efforts. With recent technological developments, potent antibodies have been identified from immunized animals and human survivors of virus infections and have been characterized functionally and structurally. Structural insight into how the most successful antibodies target GP further guides vaccine development. Here we review the recent developments in the identification and characterization of neutralizing antibodies and cocktail immunotherapies.

**Keywords:** antibody therapeutic; cryo-EM; ebolavirus; filovirus; marburgvirus; structural biology; X-ray crystallography

#### **1. Introduction**

The Filoviruses belong to the family *Filoviridae* and are among the world's deadliest pathogens. Among the six genera are the *Ebolaviruses*, the *Marburgviruses*, the *Cuevavirus*, the recently discovered *Dianlovi* rus [1], *Striavirus* [2], and the *Thamnovirus* [2]. There are 11 species in total, of which the ebolaviruses and marburgviruses are known to cause severe disease in humans. The genus *ebolavirus* includes six known viruses that are each antigenically distinct and named after the location of the disease outbreak where they were first identified. These include Ebola virus (EBOV), Sudan virus (SUDV), Bundibugyo virus (BDBV), Reston virus (RESTV), Taï Forest virus (TAFV), and Bombali virus (BOMV). The *Marburgvirus* genus contains Marburg virus (MARV) and its variant Ravn (RAVV).

The first filovirus to be identified, MARV, was discovered in 1967 when several laboratory workers in Germany developed hemorrhagic fever after handling tissues from non-human primates (NHPs). A total of 31 people were infected, and 7 died [3]. EBOV was first identified in 1976 when two separate outbreaks occurred in northern Zaire (now the Democratic Republic of Congo, DRC) and southern Sudan. Each outbreak resulted in hundreds of cases with 88% and 53% case fatality, respectively [4,5]. *Ebolavirus* has since appeared sporadically in Africa. The largest outbreak to date, which occurred in West Africa from 2014 to 2016, caused more than 28,600 infections and more than 11,300 deaths from Ebola Virus Disease (EVD). In 2021, two additional outbreaks occurred in the Democratic Republic of Congo [6] and Guinea [7]. Other filoviruses have similar outbreak potential and lethality. The most recent significant emergence of MARV in Angola had a case fatality rate of 90% [8]. Meanwhile, Sudan virus (SUDV) and the newly emergent Bundibugyo virus (BDBV) have case fatality rates of ~50% and 25–50% [9], respectively.

Symptoms of EVD include fever, headache, muscle pain, weakness, fatigue, diarrhea, vomiting, stomach pain, and hemorrhage (severe bleeding) [10,11]. The infection prodrome of filoviruses is virtually identical to common, co-circulating diseases like typhoid fever and malaria [10]. As such, early diagnosis, particularly for those cases early in a disease

**Citation:** Yu, X.; Saphire, E.O. Development and Structural Analysis of Antibody Therapeutics for Filoviruses. *Pathogens* **2022**, *11*, 374. https://doi.org/10.3390/ pathogens11030374

Academic Editors: Philipp A. Ilinykh and Kai Huang

Received: 22 February 2022 Accepted: 17 March 2022 Published: 18 March 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

<sup>1</sup> Center for Infectious Disease and Vaccine Discovery, La Jolla Institute for Immunology, 9420 Athena Circle, La Jolla, CA 92037, USA; xiaoyingyu@lji.org

outbreak, is challenging. Given this potential delay in diagnosis, the therapeutic window for potential treatments must be broad so that treatments are effective even if delivered late in a disease course. Traditional approaches involving post-exposure vaccination and small molecule interventions required almost immediate administration to be effective [12]. Currently, monoclonal antibody (mAb) therapy has been shown to be the most effective route of therapy after symptoms appear, and can confer 100% protection for non-human primates (NHPs), even if administered as late as 5 days post-challenge [13,14]. Therefore, studies of antibodies against filoviruses are an important source for potential reliable therapeutics.

In recent years, progress has been made towards vaccine and treatment development. The first vaccine to be approved, Ervebo, is rVSV-based and was tested in an open-label, cluster-randomized ring vaccination trial in Guinea in 2015 [15], deployed in 2018 in the DRC under compassionate use, before gaining approval from the United States Food and Drug Administration (FDA) in late 2019 [16,17]. Another vaccine candidate utilizes a two-dose heterologous vaccination regimen with a replication-deficient adenovirus type 26 vector-based vaccine expressing a Zaire Ebola virus glycoprotein (Ad26.ZEBOV) and a modified vaccinia Ankara (MVA) vector-based vaccine, encoding glycoproteins from the Zaire EBOV, SUDV, and MARV as well as TAFV nucleoprotein (MVA-BN-Filo). This vaccine has been granted marketing authorization by the European Medicines Agency in the European Union [18,19].

The mAb monotherapy mAb114 and antibody cocktail REGN-EB3 were tested in clinical trials and proved to be effective against EBOV; both were granted FDA approval to treat EVD, and showed superior outcomes in reducing mortality compared to ZMapp and remdesivir [20–22]. The longer and multiple-dose regimen required for ZMapp and remdesivir administration could contribute to the slower rate of viral clearance of patients in those groups, and further lead to the difference in mortality between groups [20]. The intrinsic difference between the patient conditions among the four groups may contribute to the variation in treatment protection outcomes [20]. Notably, however, the therapeutic antibodies for humans approved thus far are only effective against EBOV. None show broad reactivity against other pathogenic filoviruses. Efficacious treatments against a range of pathogenic filoviruses are urgently needed.

#### **2. Viral Entry and Glycoprotein Structure**

Filoviruses are enveloped, non-segmented, negative-sense RNA viruses that have a characteristic filamentous shape. The genome has seven genes that encode eight proteins (Figure 1). Six proteins are encoded by the corresponding viral genes, including VP24, NP, VP30, VP35, VP40 (matrix protein), and the RNA-dependent RNA polymerase (L). The ebolavirus *GP* gene expresses two major products: the trimeric glycoprotein, termed GP, which is displayed on the viral surface, and a dimeric, soluble version, termed sGP that represents the majority (80%) of the transcripts and is abundantly shed from infected cells [23–26]. GP and sGP share 295 amino acids and have some similarities in the folding of the monomeric unit [23–26]. The function of sGP remains unclear [23,27], but it has been proposed to act as an immune decoy [28]. Indeed, multiple antibodies cross-react with sGP and GP [29,30]. These antibodies may be absorbed by the much more abundant sGP and thus unavailable to neutralize virtual-surface GP. Many cross-reactive antibodies have a higher affinity for sGP [29,31], and its abundance indicates it may be a major antigen in a natural infection. Interestingly, marburgviruses and dianloviruses do not produce sGP; cuevaviruses express sGP similarly to ebolaviruses [31].

**Figure 1.** Schematic of the Ebola virus genome and virion. The glycoprotein GP (red) is the only viral protein displayed on the virion surface.

GP mediates attachment and entry into target cells. As the only surface-exposed protein on mature virions, GP is the focus of current filovirus therapeutic mAb discovery efforts (Figure 2) [32]. Filovirus GPs share a common core fold and trimeric organization, but are antigenically distinct due to species-specific sequence differences of up to 70%. The ebolavirus GP monomer comprises GP1 and GP2 subunits that are anchored together by a single GP1-GP2 disulfide bond [33]. The larger GP1 subunit harbors the receptor-binding site (RBS), the glycan cap domain, and the heavily glycosylated mucin-like domain (MLD). GP2 contains the membrane fusion machinery, including the internal fusion loop (IFL), two heptad repeats (HR1 and HR2), the membrane-proximal external region (MPER), and the transmembrane domain (TM) [25]. The GP2 subunit, particularly the IFL and stalk regions, has greater sequence conservation than GP1 among filoviruses. Marburgvirus GPs have a similar arrangement (Figure 2C,D). However, in marburgvirus GP, the furin cleavage site is shifted towards the N-terminus (residue 435 for Marburgvirus vs. 501 for ebolavirus), the region corresponding to the MLD is split into two halves: the major portion of the MLD is attached to the C terminus of GP1. The minor portion (residues 436–501) is attached to the N terminus of the GP2 and is termed the wing domain [34,35].

Filoviruses enter cells via macropinocytosis [36–38]. Once in the endosome, GP is cleaved by host endosomal cathepsins that remove both the MLD and the glycan cap [39–41] to form cleaved GP (GPCL). In GPCL, the RBS for the cellular receptor Niemann-Pick C1 (NPC1) is exposed at the apex of GP1 [42]. Following receptor binding, the fusion subunit in GPCL rearranges into a six-helix bundle that mediates fusion between host and virus membranes [43].

**Figure 2.** Epitopes on the GP surface. (**A**) Surface representation of Ebola Virus GP structure (PDB: 5JQ3) colored by domain. Side view and top view of Ebola virus GP are illustrated. (**B**). Schematic of the EBOV GP sequence. Amino acid numbering is at top, and polypeptide regions that form key domains are numbered in the center of the schematic blocks. 1: portions of the N-terminus of GP1 that form the base, 2: receptor-binding head, 3: glycan cap, 4: mucin-like domain (MLD), 5: GP2 N-terminal peptide; 6: fusion loop, 7: Heptad repeat 1 (HR1), 8 and 9: Heptad repeat 2 (HR2); 9: stalk; and 10: and membrane-proximal external region (MPER), respectively. Other regions include SP: signal peptide, and TM: transmembrane domain. The organization of sGP is illustrated below. The first 295 residues are identical to those in GP1 (labelled sGP-1). Residues 296 through 324 are unique to sGP (labelled sGP-2). The C-terminal sequence, termed delta peptide, is released from sGP by furin cleavage. (**C**) The surface representation of Marburg Virus GP structure (PDB: 6BP2) colored by domain. Side view and top view of Marburg virus GP are illustrated. (**D**) Schematic of the MARV GP sequence. Amino acid numbering is at top. 1–2: GP1, with 2 for RBS; 3: glycan cap, 4: MLD, 5: wing; 6: N-terminal loop: 7: fusion loop, 8: HR1, 9: HR2; 10: MPER; SP: signal peptide, and TM: transmembrane domain.

#### **3. Efforts for Antibody Discovery**

Analyses of antibody responses in human survivors of virus infection can outline key characteristics of antibodies elicited in response to infection and provide an important basis for the development of therapeutic antibodies.

Beginning in the early 1990s, isolation of neutralizing mAbs was mostly enabled by display library approaches, such as phage display [44,45]. Early antibodies like KZ52 and others were discovered from human survivors of the 1995 Kikwit outbreak in DRC [46]. These antibodies facilitated and enhanced understanding of the virus and permitted the determination of the structure of EBOV GP in its pre-fusion conformation [25]. Screening of antibody-secreting hybridomas from immunized mice produced a panel of novel antibodies that could be categorized into multiple different epitope groups [47]. This classification guided the formation of several successful mAb cocktails, including MB-003, ZMAb, and ZMapp, which had non-overlapping binding sites and high efficacy in NHP studies of EBOV infection [13,48,49].

Technological advancements, such as single B cell isolation and next-generation sequencing, accelerated large-scale mAb discovery, direct functional analysis, and exploration of the Ab maturation process. Potent antibodies were subsequently discovered in immunized animals [50–52], human survivors [30,53–55], and vaccinated human volunteers in clinical trials [56,57]. Notably, antibodies in these panels target a spectrum of epitopes on GP, and several neutralize broadly are active against several filovirus species. Several therapeutic antibody cocktails, including REGN-EB3, FVM04/CA45, MBP134AF, rEBOV-520/548, rEBOV-442/515, and 1C3/1C11, were generated and shown to be highly effective [52,58–62].

The functional activity of antibodies isolated during discovery efforts can be evaluated by in vitro neutralization assays to determine whether they block infection by one or more ebolaviruses together with structural biology to reveal the molecular basis for protective activity. Neutralization can be analyzed using authentic virus under BSL-4 containment [63] or at lower biosafety levels using model systems. The biologically contained Ebola virus ΔVP30 system can be performed at BSL-3 [64]. In this system, the entire open reading frame of VP30, which is an essential transcription factor for EBOV replication, is deleted to generate a replication-deficient particle. The VP30 needed for replication is supplied in *trans* through Vero cells that stably express VP30 such that EBOVΔVP30 can undergo multiple replication cycles only in VeroVP30 cells, but not the parental cells that lack VP30. In this system, all viral antigens and proteins, including sGP, are stably expressed. A second neutralization system involves recombinant vesicular stomatitis virus (VSV) bearing EBOV GP [65] and can be performed at BSL-2. This method uses a recombinant VSV (rVSV) in which the VSV-G gene is replaced with filovirus GP that is then displayed on the rhabdoviral surface [65–68]. These pseudovirus constructs typically carry a fluorescence reporter such as GFP to monitor infection, and are frequently modified so that only GP, and not secreted sGP, is expressed.

A systematic analysis of 171 mAbs by The Viral Hemorrhagic Fever Immunotherapeutics Consortium (VIC) compared the readouts of 3 different neutralization assays by epitope and level of in vivo protection [31]. There were several differences in the performance of the 171 mAbs across the 3 assays. For example, the authentic EBOV assay was more forgiving: a group of glycan-cap-directed antibodies only neutralized authentic EBOV and no model system. The ΔVP30 system was more stringent: fewer antibodies demonstrated neutralization overall. The results that correlated best with in vivo protection, however, were those assays that contained sGP (i.e., authentic EBOV and ΔVP30), as well as the fraction left un-neutralized, which was measured in rVSV assay. Antibodies that failed to completely neutralize rVSV at the highest concentration similarly failed to protect in the mouse model.

A small fraction of antibodies had no neutralization activity in any assay, yet nevertheless protected in a mouse model of EBOV infection. In high-throughput systems-serology assays, which examined the contribution of immune effector functions like phagocytosis

and activation of natural killer (NK) cells to in vivo antibody efficacy, these antibodies had high immune effector function activity, indicating the potential contribution of Fcmediated activity to protection [31,69]. By comparison of the in vivo activity of the Fc variants of mAbs from different epitope groups, a previous study revealed the differential requirements for FcγR to achieve protection [70]. MAbs targeting the membrane-proximal regions (HR2 and MPER) or the MLD do not inspire Fc–FcγR interactions. However, the mAbs that contact to chalice bowl region or the fusion loop are linked to FcγR engagement. Moreover, by applying an Fc engineering platform, a library of Fc variants of a specific mAb could be generated to compare how Fc effector functions correlate with mAb protection performance. Based on the results of functional characterization of the variants and the in vivo protection in mice, Fc variants with high complement activity, yet moderate and balanced ADCC activity are more protective against viral infection in vivo [71]. Together, the previous results suggested that effective antibody treatments should comprise mAbs that achieve both neutralization and immune effector functions. The VIC study and other work [31,51,53,54] also revealed the epitopes at which broader ebolavirus cross-reactivity can be achieved.

Structural biology has served a vital role in increasing our understanding of the molecular mechanisms underlying antibody-mediated neutralization of many viral infections. In particular, atomic resolution structures obtained using cryogenic electron microscopy (cryo-EM) or X-ray crystallography allow exploration of the fine details of binding between antibodies and virus proteins like GP to reveal crucial information about molecular interactions and the basis of cross-reactivity. Combining structural studies with virology and in vitro biochemistry allows a thorough evaluation of therapeutic candidates and their target interactions.
