*1.3. Cellular Entry of Ebola Virus*

Cell entry is a critical stage in the lifecycle of any virus. GP exposed on the surface of the virion has been demonstrated to be essential for cell entry. In which case GP is the most important target for neutralising antibodies.

In EBOV, co- or post- translational editing (e.g., transcriptional slippage) of the GP transcript produce various GP products (reviewed in detail by Lee et al.) [12]. Similar posttranslational modifications are not observed in MARV (Figure 1A). In EBOV, the pre-GP gene is transcribed as one protein but is cleaved by the furin protease into two subunits, GP1 and GP2, joined by a disulphide bridge to form a heterodimer [16]. Surface GP exists as a trimer of the GP heterodimers expressed on the surface of the virion and interacts with NPC1 for cell entry [17]. However, surface GP is not the main gene product. The primary product, which is expressed from unedited RNA transcripts, is a non-structural protein called secreted GP (sGP). Secreted GP is a dimer of GP1 and a truncated GP2 bound by disulphide bridges [18] which is hypothesised to act as a decoy antigen to sequester antibodies or cause antigenic subversion. Surface GP is only produced upon the addition of an adenosine residue to the GP transcript in 20% of cases. A deletion of one adenosine or addition of two adenosines in the transcript leads to the production of another nonstructural protein, small secreted GP (ssGP) [12,19]. The role of ssGP is unclear in EBOV pathogenesis. Lastly, during EBOV infection, GP can be shed from infected cells in a soluble form due to cleavage by TACE metalloprotease. This soluble GP is named shed GP and was shown to sequester EBOV-specific neutralising antibodies directed towards GP [20].

The entire GP is required for cell entry. GP1 is associated with cell attachment and receptor binding, whilst GP2 has functions regarding fusion with the host cell membrane. Both subunits are therefore targets of neutralising antibodies. GP1 contains four subdomains: the head, base, mucin-like domain (MLD) and glycan cap (GC). The GP1 head, when arranged in its trimeric conformation forms a three lobed chalice containing the receptor binding domain (RBD). The GP1 base clamps the internal fusion loop (IFL) and heptad repeat 1 (HR1) and heptad repeat 2 (HR2) from GP2 for arrangement into its pre-fusion conformation (Figure 1B) [12]. There are also two heavily glycosylated regions on the GP1, GC and the MLD [12]. The GC is a large chain of sugars, some of which are thought to act

as attachment factors to host cells e.g., binding to C-type lectins. Both the MLD and GC are highly variable regions and can block neutralising antibodies binding to the GP, e.g., the MLD overhangs the GP to protect critical epitopes [21].

GP2 is docked into the viral membrane by its transmembrane domain. GP2 is also responsible for fusing the virus membrane and the host endosome membrane. It contains several domains that are key to this process such as HR1 and HR2 that change conformation to aid the IFL insertion into the host endosome membrane ultimately leading to fusion (Figure 1B) [12].

EBOV uses a broad range of binding factors to attach to a variety of host cell types before being internalised by the host cell into an endocytic compartment by receptor mediated endocytosis or macropinocytosis [12,22]. In the late endosome, cathepsins cleave the GP removing the MLD and GC and exposing the RBD in a stage called priming, allowing better access to the RBD for receptor binding [17,23]. The RBD then binds to the NPC1 cholesterol transporter [24]. The base of GP1 is a clamp, that when released triggers the conformational changes required to expose the hydrophobic IFL which inserts into the endosomal membrane [12,25]. HR1 and HR2 pull on the IFL causing it to fuse with endosomal membrane, creating a pore for the release of the ribonucleoprotein into the cell cytoplasm [12,26].

#### *1.4. EVD and Immunity*

EVD is characterised as a systemic inflammatory response ("septic-shock-like-syndrome") with coagulation abnormalities and multi-organ failure, immunosuppression and lymphopenia [4]. This pathology is a result of GP-mediated activation of innate immune cells.

EBOV preferentially targets dendritic cells (DC), monocytes and macrophages but can also productively infect epithelial and endothelial cells, adrenal fibroblasts and hepatocytes [27]. GP-mediated activation of the TLR4 pathway in DCs and macrophages [28,29] results in the expression and secretion of inflammatory cytokines (e.g., IL-6, IL-1β, TNF) [30–33]. In macrophages, EBOV elicits a strong upregulation of IFN-I signalling [34], a crucial feature of the immune response to viral infection that provides the link between innate and adaptive immunity [35]. However, the role of Type I IFN in EBOV clearance and protection is not fully understood. Several studies demonstrated that IFN-I impacted EBOV replication in vitro and survival in animal models [36–40], while higher levels of IFN-α were associated with EVD fatal cases [41]. With that said, the picture is likely more complicated as EBOV has also developed some strategies to dampen innate immune responses [42].

Following the activation of the innate immune system, adaptive immune responses are induced. Early studies in individuals who succumbed to EVD described robust T cell activation, followed by a collapse in the T cell population [43,44]. More recent studies demonstrated the impact of T cell dynamics, kinetics and phenotype on EVD outcome [44–46]. It is known that T cells, particularly CD8<sup>+</sup> T cells, are important in viral infection. However, the role of CD4<sup>+</sup> and CD8<sup>+</sup> T cells in protection is still under discussion. For example, it was demonstrated that CD8<sup>+</sup> T cell deficient mice, but not CD4<sup>+</sup> T cell or B cell deficient mice, succumbed to a subcutaneous infection with a mouse-adapted EBOV strain suggesting a crucial role of CD8+ T cells in protection [47]. Following the 2013–2016 West Africa epidemic, more studies analysed the activation of T cells in EVD survivors. We reported that the dominant CD8<sup>+</sup> polyfunctional T cell phenotype was IFNγ+, TNF+, IL-2<sup>−</sup> in EVD survivors in Guinea [48]. We also found that both CD4+ and CD8<sup>+</sup> T cells contributed to specific T cell memory but with a differing cytokine profile. CD4<sup>+</sup> T cells produced IFNγ, TNFα and IL-2, whereas CD8<sup>+</sup> T cells only produced IFNγ and TNFα [49]. Sakabe et al. found that CD8+ responses to the NP were immunodominant in survivors in Sierra Leone [50]. More details about T cell responses following a natural EBOV infection can be found in several reviews [46,51]. Regarding the humoral response, some studies have suggested that early development of IgM and IgG was associated with a positive outcome [52] while antibody deficiencies were reported in fatal cases [53]. Furthermore, monoclonal antibodies (mAbs) were isolated from EVD survivors, some of which were

shown to be neutralising and protective in animal models [54–59]. However, it is still debated whether the antibody titres or the neutralisation activity of antibodies is the stronger forecaster of protection [60]. This review will discuss the role of neutralising antibody responses following a natural infection and in the context of therapeutics and vaccination.
