31 Marburg and Ebola viruses

Infection with Marburg and Ebola viruses causes haemorrhagic fevers that are characterized by organ malfunction, bleeding complications, and high mortality. The viruses are members of the family Filoviridae, a group of membrane-enveloped filamentous RNA viruses. Five distinct species of the genus Ebolavirus have been reported, and the Marburgvirus genus contains only one species. Both Marburg and Ebola virus diseases are zoonotic infections where the primary hosts are thought to be bats. The initial human infection is acquired from wildlife and subsequent person-to-person spread propagates the outbreak until it is brought under control. Ebola and Marburg viruses are classified as hazard or risk group 4 pathogens because of the very high case fatality rates observed for Ebola and Marburg virus diseases, the frequency of person-to-person transmission and community spread, and the lack of an approved vaccine or antiviral therapy. This mandates that infectious materials are handled and studied in maximum containment laboratory facilities.

Epidemics have occurred sporadically since the discovery of Marburg virus in 1967 and Ebola virus in 1976. While some of these outbreaks have been relatively large, infecting a few hundreds of individuals, they have generally occurred in rural settings and have been controlled relatively easily. However, the 2013–16 epidemic of Ebola virus disease in West Africa was different, representing the first emergence of the Zaire species of Ebola in a high-density urban location. Consequently, this has been the largest recorded filovirus outbreak in terms of both the number of people infected and the range of geographical spread. Many of the reported and confirmed cases were among people living in high-density and impoverished urban environments.

The chapter summarizes the most up-to-date taxonomic status of the family Filoviridae. It focuses on Marburg and Ebola viruses in a historical context, culminating in the 2013–16 outbreak of Ebola virus in West Africa. Virus biology of the most well-studied member is described, with details of the viral genome and the protein machinery necessary to propagate viruses at the molecular and cellular level. This information is used to build a wider-scale virus–host perspective with detail on the pathology and pathogenesis of Ebola virus disease. The consequences of cell infection are examined, together with our current understanding of the immune response to Ebola virus, leading to a broader description of the clinical features of disease. The chapter closes by drawing information together in a section on diagnosis, ecology, prevention, and control.

Ebola and Marburg viruses are filoviruses classified within the order Mononegavirales, a large group comprising five families of enveloped viruses possessing linear, non-segmented, negative-sense single-stranded RNA genomes (Pringle 1991). Filoviruses were originally grouped into the Rhabdoviridae family, based on their initial appearance under the electron microscope. A separate family—the Filoviridae—was subsequently conceived, based on unique properties that were defined from detailed studies of Marburg virus and Ebola virus isolates (Kiley et al. 1982). Further characterization demonstrated that they represent divergent lineages, which were significant enough to warrant the formation of the genera Marburgvirus and Ebolavirus. The taxonomy of Marburg viruses and Ebola viruses has changed several times since the discovery of the type Marburgvirus in 1967 (Siegert et al. 1967) and the type Ebolavirus in 1976 (Bowen et al. 1977; Johnson et al. 1977). These periodic changes in nomenclature have often caused confusion and in 2010, an improved nomenclature for the family Filoviridae was devised to reflect the standard terminology that had been adopted by laboratory scientists, while upholding the rules and regulations of the International Committee on Taxonomy of Viruses (ICTV) (Kuhn et al. 2010). This review uses this terminology.

Currently the Marburgvirus genus contains a single species termed Marburg Marburgvirus. Although a number of strains have been identified since 1967, they do not exhibit enough genetic variation to constitute distinct species. The species currently contains two distinct viruses: Marburg virus (MARV) and Ravn virus (RAVV). There is greater divergence within the Ebolavirus genus and currently five species are recognized: Zaire ebolavirus (the type species; a virus in this species is Ebola virus, abbreviated EBOV), Sudan ebolavirus (including Sudan virus (SUDV)), Reston ebolavirus (including Reston virus (RESTV)), Tai Forest ebolavirus (including Tai Forest virus (TAFV)), and Bundibugyo ebolavirus (including Bundibugyo virus (BDBV)). For several years, the family Filoviridae was composed of only the two genera Marburgvirus and Ebolavirus; however, in 2011, a new and discrete filovirus sequence designated Lloviu virus (LLOV) was obtained from bats in Spain (Negredo et al. 2011). While a virus has yet to be isolated in culture or otherwise, the complete genome of LLOV shows that it represents a distinct and single species of Lloviu cuevavirus in the new genus Cuevavirus of the Filoviridae family.

Filoviruses first came to medical attention in summer of 1967 when laboratory and animal technicians suddenly and simultaneously developed acute haemorrhagic fever in the rural town of Marburg and the city of Frankfurt. In Marburg, the epidemic started with three laboratory workers who had processed tissues and organs from African green monkeys (Cercopithecus aethiops). These had been imported from Uganda to establish kidney cell cultures for the production of attenuated polioviruses. An additional 19 patients were hospitalized, including two medical personnel who contracted the disease nosocomially. A few months later, a further case occurred in a woman who is believed to have become infected through sexual intercourse with her husband, 12 weeks after his clinical recovery from the infection (Martini 1969). In Frankfurt, where tissues from the same African green monkeys had been processed, there were six cases, including two secondary infections (Stille et al. 1968). Finally, 1 month after the first outbreak in Marburg, two cases were identified in Belgrade; the first was a veterinarian who was infected while performing a necropsy on a dead monkey, and he went on to transmit the virus to his wife (Martini 1969). In all, 31 cases were identified, which resulted in seven fatalities.

The virus isolated from patients’ blood and tissue was morphologically unique when observed by electron microscopy, and it was antigenically unrelated to any known mammalian pathogen. It was named Marburg virus after the town in which the initial cases occurred. The disease was initially called Marburg haemorrhagic fever, but the term Marburg virus disease (MVD) has more recently become the accepted official name.

MARV remained an obscure curiosity until 1975, when three cases were reported in Johannesburg (Gear et al. 1975), although the source of the infection was not determined (Conrad et al. 1978). Two further episodes of MVD were reported from Kenya in 1980 (Smith et al. 1982) and 1987 (Johnson et al. 1996). Interestingly the index cases associated with these incidents were both linked with travel to the Mt Elgon region of Kenya, which is located close to Lake Kyogo where the African green monkeys that had initiated the 1967 outbreak had originally been trapped. The first large outbreak of MVD occurred in 1998 in Durba/Watsa, located in the north-eastern region of the Democratic Republic of Congo (DRC). Primary cases were mainly gold miners who started multiple, usually short, chains of human-to-human transmission within their families. In all, 154 cases were reported with a case fatality rate (CFR) of 83% (World Health Organization 1999). The largest outbreak of MVD in a community setting to date took place in Uige Province in northern Angola; 252 cases were reported with a CFR of 90%. This outbreak started in October/November 2004, with the main peak of infections occurring in February/March 2005; the last confirmed case died in July 2005. Sequence analysis of virus isolates from this outbreak suggested a single introduction into the community (Jeff et al. 2007). This was followed by a small outbreak in 2007 in the Kamwenge district of Uganda associated with a gold mine, which resulted in four cases and a CFR of 25%. In 2008, two single cases of MVD were reported in the USA and The Netherlands in returning travellers who had visited a cave in Maramagambo in Uganda. The Dutch patient died (Timen et al. 2009), while the US traveller was discharged from hospital after making an uneventful recovery from a 4-day illness (Centers for Disease Control and Prevention 2009). The next outbreak of MVD occurred in 2012 over a 3-week period in the Ugandan districts of Kabale, Ibanda, and Kampala, resulting in 15 cases and a CFR of 27% (Albarino et al. 2013). To date, the most recent episode of MVD occurred in 2014, again in the Kampala district of Uganda. It involved one fatal case, but no confirmed secondary cases (World Health Organization 2014).

Table 31.1 lists the known dates and origins of MARV outbreaks, with CFRs and relevant information pertaining to each episode.

Nine years after the discovery of Marburg, two further epidemics of haemorrhagic fever occurred almost simultaneously in the village of Yambuku in the DRC (formally Zaire) in August 1976 and the town of Nzara in Sudan (now in South Sudan) that September. These were large outbreaks causing 318 cases (CFR of 88%) and 284 cases (CFR of 53%), respectively. High levels of transmission were later understood to have been promoted through the use of contaminated syringes and needles and a lack of barrier nursing practices, although secondary transmission to close contacts were also significant. Virus isolations were eventually made from patient samples from both outbreaks in the UK (Bowen et al. 1977) and USA (World Health Organization 1978a; World Health Organization 1978b), and the virus was named after the river in north-western DRC close to the village of Yambuku. EBOV was morphologically identical to, but antigenically distinct from, MARV (Bowen et al. 1977) and the epidemics were subsequently shown to have been caused by two distinct viruses—EBOV and SUDV, respectively (now classified in the Zaire ebolavirus and Sudan ebolavirus species, respectively) (Kuhn et al. 2010). The following year, a single fatal EBOV case was reported from Tandala in the DRC, approximately 300 km west of Yambuku (Heymann et al. 1980). SUDV also re-emerged in 1979 in Nzara and Yambio in Sudan. Hospitalization of the patient led to four nosocomial infections and further transmission to five families, culminating in 34 cases with a CFR of 65% (World Health Organization 1979).

Following the control of these outbreaks, Ebola haemorrhagic fever—now termed Ebola virus disease (EVD)—was not recognized again until nearly 16 years later when the virus was detected in the Ogooué-Ivindo Province in north-east Gabon (approximately 1500 km west of Yambuku). With a total of 52 cases and a CFR of 60% (Georges et al. 1997); transmission seemed to be associated with hunting and butchering of wildlife, often great apes. Also in 1994, a new species member of Ebolavirus—Tai Forest ebolavirus (TAFV)—was isolated from an ethnologist who had become ill while working in the Tai Forest reserve of Ivory Coast in 1994. This non-fatal infection is thought to have occurred while the ethnologist was performing a necropsy on a dead chimpanzee (whose troop had lost several members to EVD) (LeGuenno et al. 1995; Formenty et al. 1999). Interestingly the investigating team discovered a second non-fatal human seroconversion in nearby Liberia (World Health Organization 1996), which extended the geographic distribution of known members of the Ebolavirus genus to include further regions of West Africa. Thus, it was known that members of the Ebolavirus genus were circulating in West Africa 18 years before the first description of what became the largest and most devastating epidemic of EVD.

The following year, a strain of EBOV re-emerged once again in the DRC, causing a large hospital and community outbreak of EVD in and around Kikwit (Khan et al. 1999) (approximately 1000 km south-west of Yambuku). In total, there were 315 cases with a CFR of 81%. This incident was the first to highlight the risks of virus transmission during preparations of bodies for burial (Muyembe-Tamfum et al. 1999). In 1996, two more outbreaks were reported from Ogooué-Ivindo Province in north-east Gabon. The first of these started in early February and included 37 cases (CFR of 57%); the second episode, which started in summer, resulted in 60 cases (CFR of 75%). The latter epidemic included an imported case in South Africa where an ill Gabonese physician went on to infect a nurse who died of EVD (two cases; CFR of 50%). The first reported epidemic that crossed the border into the Republic of Congo (RC) began in late November of 2001 with the index case again reported from Ogooué-Ivindo Province in north-east Gabon. The epidemic spread to Mekambo and Makokou and from there into the RC by an ill Gabonese who sought medical care from traditional healers. In total, there were 65 cases (CFR of 82%) and 59 cases (CFR of 75%) in Gabon and the RC, respectively (Leroy et al. 2002). The next occurrence of EVD was a large epidemic reported in the districts of Mbomo and Kelle in Cuvette Ouest region in the RC from late 2002 to May 2003 with 143 cases (CFR of 90%), followed in late 2003 by a smaller episode in the district of Mbomo with 35 cases (CFR of 83%) (Formenty et al. 2003). A neighbouring area (Etoumbi) was affected in 2005 by a small outbreak of EVD with 12 cases (CFR of 75%) and so far, this has been the last reported outbreak in the RC region.

Soon after the early reports of EBOV in Gabon and the RC, Uganda witnessed the second largest epidemic of EVD to date, which lasted from 2000 to 2001, culminating in 425 cases and a CFR of 53%. On this occasion, a species of SUDV was shown to be responsible; this was the first appearance of EBOV in Uganda. The epidemic was mainly concentrated in the Gulu district (approximately 800 km south-east of Nzara in South Sudan), a savannah area located in the north of the country. Human-to-human transmission including nosocomial infections were the main protagonists of this outbreak. Significantly the epidemic was the first to be supported by laboratory diagnostics performed in the field, which made useful contributions to outbreak management (World Health Organization 2001). In 2004, southern Sudan was again affected by a small SUDV outbreak with 17 cases and a CFR of 41%. The index case had butchered a monkey and human-to-human transmission was shown to be significant between close contacts. The years 2007 and 2008/09 witnessed the next outbreaks of EVD which occurred in the Kasai occidental province of the DRC (approximately 900 km south of Yambuku). The first larger outbreak included 264 reported cases with a CFR of 71%; the second smaller outbreak had 32 cases with a CFR of 47% (World Health Organization 2009). Both outbreaks affected rural communities in the vicinity of the city of Luebo and were thought to be related to hunting and handling of migratory fruit bats (Leroy et al. 2009).

In 2007, a new species of the Ebolavirus genus, designated Bundibugyo ebolavirus (Bundibugyo virus, BDBV), was identified as the causative agent for an outbreak that occurred in the Bundibugyo district in western Uganda (Towner et al. 2008). In total, there were 149 reported cases, with 37 deaths and a CFR of 25%.

The fifth species of Ebolavirus—Reston ebolavirus—is noteworthy for its lack of pathogenicity in humans. Indeed Reston virus (RESTV) was originally recognized by accident, following a campaign to confirm the aetiology of an incident thought to be simian haemorrhagic fever (SHF) in cynomolgus macaques (Macaca fascicularis) that had been housed in a quarantine facility in Reston in Virginia in 1989. Virus isolation attempts failed to confirm the suspected Arterivirus culprit, SHF virus, but instead showed classic filovirus morphology under the electron microscope, leading to the name Reston (Jahrling et al. 1990). Interestingly and fortuitously, this epizootic virus failed to cause EVD in humans despite clear evidence of infection in several animal handlers at the time (Miranda et al. 1999). The origin of this virus was never determined, but following the resumption of non-human primate (NHP) importation, further incidents of RESTV occurred in similar circumstances in 1990 and 1996 in the USA (Rollin et al. 1990) and in 1992 in Siena in Italy (World Health Organization 1992). Subsequent investigations traced NHP shipments to a supplier in the Philippines, but the source of infection was never ascertained. Interestingly, in 2009, RESTV emerged in pigs in the Philippines, again identified by accident following the confirmation of porcine respiratory and reproductive virus (PRRS) aetiology affecting a pig farm (Barrette et al. 2009). Pigs were determined to be co-infected, however, but the actual disease potential of RESTV in these animals remains unclear. Despite infections/exposures never resulting in clinical disease in humans, RESTV has strong genetic similarity to other pathogenic members of the genus and this discovery raises important issues around food production and the potential for zoonotic transmission.

The largest and most widespread epidemic of EVD to date emerged in December 2013 as a single entry into the human population when a 2-year old became infected in Guéckédou in Guinea, a small village close to the border with Sierra Leone and Liberia (Baize et al. 2014). The virus aetiology was not determined to be Zaire EBOV until March 2014 (Baize et al. 2014). Consequently, the virus had spread unchecked through local communities for several months; many people in this population were highly mobile and this exacerbated its sustained transmission. Official notification was received by the WHO of a rapidly developing outbreak of EVD on 23 March 2014. Soon afterwards, EVD was reported in four districts of Guinea. The neighbouring countries of Liberia and Sierra Leone also reported suspected EVD.

In Guinea, the outbreak progressed steadily; a total of 86 suspected cases, including 59 deaths, were reported by 24 March, and on 28 May, the total number of cases reported had reached 281, with 186 deaths. In late May, the outbreak was reported in Conakry, the capital city, which had a population of approximately 2 million.

In Liberia, EVD was reported from four counties by the middle of April 2014 and in Monrovia, the capital, in June. In Sierra Leone, the outbreak was first recognized at the end of May, in Nigeria on 20 July, in Senegal at the end of August, and in Mali in October. In August 2014, the WHO declared that the epidemic was a ‘public health emergency of international concern’ (Fauci 2014). Soon after this international recognition, imported EVD cases were reported from three countries outside of West Africa—Spain, the UK, and the USA. A broader and more sustained international response then established treatment centres and additional mobile laboratories in key areas of the affected countries. The epidemic was gradually brought under control, although sporadic cases, almost certainly arising from sexual transmission associated with persistence of the virus in the semen of male survivors, were still occurring at the beginning of 2016. Further sporadic cases are to be expected.

By 20 January, 2016, the cumulative reported numbers of Ebola cases and fatalities for each of the main West African countries affected were: 3804 cases with 2526 deaths in Guinea, 10 675 cases with 4809 deaths in Liberia, and 14 123 cases with 3956 deaths in Sierra Leone, bringing a total of 28 638 cases and a CFR of 40% (World Health Organization 2016). In terms of reported morbidity and mortality, this EVD epidemic has been much greater than all previous outbreaks combined; moreover, the real number of those who have been infected and died is likely much higher (WHO Ebola Response Team 2014).

While most media attention was focused on the West African outbreak, a separate and unrelated outbreak of EVD was declared in August 2014 in Jeera County in Equateur Province of the DRC. Its early identification and the subsequent rapid international response involving early engagement with community leaders and the organization of safe burials was key to the control of this epidemic which was declared over in November 2014 after 66 cases and 49 deaths.

Table 31.2 lists the Ebola outbreak dates, the Ebolavirus species responsible, and country origins with CFRs and relevant information pertaining to each episode.

The majority of studies have focused on EBOV of the type species Zaire Ebolavirus. In the genus Marburgvirus, the most commonly studied virus has been MARV and the following text is based on these two viruses. They are believed to be representative of the other Filoviridae.

Filovirus particles have a distinctive filamentous morphology, which inspired their name (filum—thread). When viewed directly with negative staining techniques under transmission electron microscopy, virions are seen to be highly pleomorphic, appearing as either U-shaped, 6-shaped, or as torus configurations, particularly for MARV (EBOV is generally more filamentous). Filovirus particles also occur as elongated filaments of varying length, occasionally forming branched structures. The unit length associated with peak infectivity for MARV and EBOV particles has been measured to be 860 and 1200 nm, respectively (Geisbert and Jahrling 1995). They have a uniform diameter of 80 nm, within which is arranged a helical nucleocapsid of 50 nm in diameter. Although all members of order Mononegavirales have filamentous helical nucleocapsids, the filoviruses are unique in being the only animal viruses in which the whole virion, not only the nucleocapsid, is highly filamentous. The surface virion is decorated with membrane-anchored glycoprotein spikes, which project approximately 10 nm from the surface, an appearance often described as ‘moth-eaten’ (Fig. 31.1).

As for other members of the order Mononegavirales, the filovirus genome consists of a non-segmented, single-stranded, negative-sense RNA molecule. The gene organization generally conforms to those of paramyxoviruses and rhabdoviruses, but their complexity is more akin to those of paramyxoviruses. Filovirus genomes are approximately 19 000 bases in length, making them the largest in the order; they contain seven sequentially arranged genes: nucleoprotein (NP) – virion protein (VP) 35 – VP40 – glycoprotein (GP) – VP30 – VP24 – polymerase (L). Genes are delineated by conserved transcriptional signals and begin close to the 3′ end of the genomic sequence with a start site, and end with a polyadenylation stop site. The genes are separated by short intergenic regions of one or more nucleotides. An unusual feature of all filovirus genomes is the presence of gene overlaps. The extragenic sequences at the 3′ end of all filovirus genomes (often called the leader sequences) are short, ranging in length from 50 to 70 nucleotides, while the length of the 5′ end (trailer) sequences are variable. The extreme 3′ and 5′ ends are conserved and show a high degree of complementarity (Fig. 31.2).

The structural proteins of filoviruses can be subdivided into two groups: (i) those that form the nuclear capsid (NC) RNA–protein complex and (ii) those that are associated with the membrane envelope. The NC-associated proteins are involved in transcription and replication of the genome, whereas the majority of the envelope-associated proteins have a role in either the assembly of the virion or virus entry into cells.

The expression of a non-structural soluble glycoprotein (sGP) as the primary product of the GP gene of EBOV is unusual and is an important distinction from MARV (Sanchez et al. 1996). Approximately 300 N-terminal amino acids of sGP are identical to those of the structural GP, but the C-terminus is unique in sequence. sGP is produced from a precursor molecule that is also cleaved by furin near the C-terminus, to release a short peptide that contains exclusively O-linked glycans; this has been named ‘delta peptide’ (Volchkov 1999) and to date no biological activity has been attributed to it. Biochemical and antigenic analyses of the EBOV sGP have shown that it is structurally distinct from GP and is secreted from infected cells as a homodimer in a parallel orientation held together by disulfide bonds. It is possible that sGP contributes to disease and there is evidence that it has multiple roles in pathogenesis (de La Vega et al. 2015); it is evident that large amounts circulate in the blood of acutely infected humans (Sanchez et al. 1999).

An additional non-structural glycoprotein, termed small soluble glycoprotein (ssGP), has also been identified, partially characterized (Mehedi et al. 2011), and appears to be a truncated version of sGP. As with GP, ssGP is expressed through transcriptional editing and is expressed at a low level (approximately 5% of sGP + GP); it has structural properties similar to sGP with many N-linked glycans, although it does not contain O-linked glycans. ssGP also exists as a parallel homodimer; its true function has yet to be defined, although it appears to lack the anti-inflammatory property reported for sGP (Mehedi et al. 2011).

The major matrix protein of the filoviruses is VP40, with the VP24 protein playing a contributory function. VP40 is the most abundant protein in the virion, while only small amounts of VP24 are incorporated into virus particles. Both proteins have an affinity for membranes and are associated with the virion envelope; however, they do not contain membrane-spanning regions. VP40 plays a critical role in viral budding, as it initiates and drives the envelopment of the NC by the plasma membrane (Jasenosky and Kawaoka 2004). In addition, it has been reported that both VP40 and VP24 of EBOV contribute to the regulation of genome replication and transcription (Watanabe et al. 2007; Hoenen et al. 2010). VP24 is hydrophobic and has an affinity for the plasma membrane and perinuclear region of infected cells. It is also capable of forming homotetramers, which are induced by pH and divalent cation changes. In addition to its role in NC formation, it is involved in the modulation of the host immune response to infection through evasion mechanisms (Reid et al. 2007). It has an important role in pathogenesis, and mutations in VP24 have been linked to the adaptation of EBOV in mice and guinea pigs to produce lethal disease (Ebihara et al. 2006; Mateo et al. 2011).

Filoviruses have broad tropism for many different cell types, which relates in large part to the binding properties of the GP spikes that cover the surface of the virion. The molecular mechanism leading to EBOV GP-mediated infection is currently unclear. However, a body of evidence indicates that EBOV uses macropinocytosis and clathrin-mediated endocytosis (Hunt et al. 2011) to enter host cells. EBOV GP binds to attachment factors, such as DC-SIGN (Alvarez et al. 2002) and TIM-1 (Kondratowicz et al. 2011). Once viral particles have been taken into late endosomes and lysosomes, GP₁ is cleaved by cathepsins B and L, which are cellular cysteine proteases. This sets off a conformational change in the GP₂ trimer that triggers the deployment of the fusion machinery, which produces a fusion-competent intermediate (Dube et al. 2009). Recent work has shown that this intermediate binds to the Niemann–Pick type C1 (NPC1) (Miller et al. 2012) intracellular receptor present in the endosome. Several studies have demonstrated that EBOV GP-mediated fusion (Bale et al. 2011; Brecher et al. 2012; Miller et al. 2012) is effected by NPC1, low pH, and possibly mild reduction of GP. Ultimately GP₂ fusion peptides are inserted into the endosomal membrane, causing membrane fusion—an event which links and brings together the viral and host membranes—releasing the NC into the cytoplasm and establishing the infection process.

Following filovirus entry, negative-strand RNA biology dictates that transcription of the viral genome occurs before anything else. Thus, polyadenylated monocistronic mRNAs are synthesized from virus genes in a 3′ to 5′ direction from the encapsidated genomic RNA template. Transcription involves a process of starting and stopping as the polymerase complex encounters conserved initiation and termination/polyadenylation sites along the genome. Polyadenylation is believed to occur by slippage or stuttering of the polymerase at a section of five to six uridines ending the termination site. A characteristic that is unique to the transcriptional signals of filoviruses is a common pentanucleotide sequence 3′-UAAUU, present at the 5′ end of start sites and at the 3′ end of stop sites (Feldmann et al. 1992). Transcripts are also capped at the 5′ end (7MeG5′-ppp5′-R) by the L protein (Ferron et al. 2002).

The organization and transcription of the GP genes of EBOV are unusual and contrast with those of MARV. In MARV, the GP gene encodes a single product GP in a conventional open reading frame, whereas all Ebolavirus species encode their GP in two open reading frames (−0 and −1 frames). Expression of the full-length EBOV GP requires a transcriptional editing event (Mehedi et al. 2011) (comparable to the editing of the P gene in paramyxoviruses). Translation of the unedited transcript of the EBOV GP gene results in the production of sGP, the smaller non-structural secreted glycoprotein. The transcriptional editing event that leads to GP expression occurs at a series of seven uridines on the genomic RNA template and results in the insertion of an additional adenosine, which connects the GP coding frames; approximately 20% to 25% of the transcripts are edited. This mechanism of insertion is thought to have evolved out of the polymerase’s ability to polyadenylate by stuttering on a poly (U) template. Insertion of a single nucleotide at the editing site appears to occur with a high degree of fidelity; however, insertion of two adenosines can also occur (in approximately 5% of GP gene transcripts) and this leads to the synthesis of low levels of ssGP (Mehedi et al. 2011). The editing of EBOV GP gene transcripts is the only example of a virus glycoprotein that is expressed through this type of mechanism. Sequence analysis of the GP genes of MARV isolates indicates that a nucleotide sequence that corresponds to the editing region of EBOV GP genes is totally absent (Feldmann et al. 1992).

In addition to transcription, the promoter at the 3′ end of the genomic RNA also drives the synthesis of full-length complementary/antigenomic RNA from the encapsidated template. As with other non-segmented, single-stranded, negative-sense RNA viruses, the ends of the genome have a high degree of sequence complementarity and stem-loop structures are predicted to form at the 3′ and 5′ ends of genomic and antigenomic RNAs. These structures are believed to be essential to the replication of filoviruses (Crary et al. 2003). Initial expression of virus genes leads to a build-up of viral proteins (especially NP), which, at a certain threshold, is thought to trigger a switch from transcription to replication. This switch results in the synthesis and encapsidation of antigenomic RNA molecules, which in turn serve as templates for genomic RNA which is also rapidly encapsidated. Depletion of capsid proteins is believed to cause a return to transcription, and eventually an equilibrium is established wherein transcription and replication are concurrent processes. As replication progresses in the infected cell, NC assemblies containing genomic RNA accumulate and are directed to the plasma membrane to meet with the structural membrane proteins for virion assembly. Significant insight into filovirus replication has been achieved from studies that have adopted developments in reverse genetics, and since 2005 reconstituted genomes and mini-reporter genome transcription/replication systems based on plasmid cDNA have enabled useful advances in filovirus biology (Boehmann et al. 2005; Enterlein et al. 2006). For MARV, the NP, VP35, and L proteins are all that is required to transcribe and replicate minigenomes (Enterlein et al. 2006; Nanbo et al. 2010) in a rescue system. Interestingly similar systems developed for EBOV also require VP30 (Neumann et al. 2002) (including the NP, VP35, and L) as the minimal complement of proteins for replicating a mini-reporter genome RNP.

When sufficient levels of NCs and envelope-associated proteins are reached, these components assemble into virus particles at the plasma membrane (Jasenosky and Kawaoka 2004). Filovirus-infected cells develop prominent inclusion bodies, visible even under the light microscope. These inclusions are predominantly induced by NP but also contain other proteins of the NC and are a source sometimes referred to as virus factories of NCs which can be seen in association with inclusions. Mature NCs interact with VP40 molecules in the budding process. Structural and functional studies of VP40 have provided important insights into the assembly of filovirus virions (Dessen et al. 2000; Harty et al. 2000). Post-translational processing and intracellular trafficking of VP40 result in the deposition of VP40 at the plasma membrane via the vacuolar protein sorting pathway. VP40 can mediate its own release from mammalian cells, forming enveloped virus-like particles (VLPs); these, however, are more efficiently produced when GP and NP are present (Licata et al. 2004). The N-terminus of filovirus VP40 molecules contain late (L) domain motifs which are important in post-translational processing and trafficking events that facilitate virus budding (Harty et al. 2000). Membrane/lipid rafts have been identified as platforms for the assembly of filovirus virions (Panchal et al. 2003). Raft-associated VP40 is believed to associate with NCs, drawing them tightly to the membrane where they are enveloped and extruded from the host cell as infectious virions. Membrane rafts are rigid microdomains (containing sphingolipids and cholesterol) present in biological membranes which are isolated from the fluid phospholipids surrounding them. GP trimers conveyed to the plasma membrane through vesicular transport have an affinity for these lipid rafts and are associated with palmitoylation of their membrane-spanning anchor sequences here. Electron tomography studies of MARV budding indicate that the entire length of nucleocapsids associate laterally with the plasma membrane (much like a rising submarine), before its protrusion and the release of the mature virion particle by being pinched off at the trailing end (Urata et al. 2007).

Clinical investigations from episodes and outbreaks of human EBOV and MARV infections have contributed significant information about the pathology and pathogenesis of these infections. While information derived from such incidents prior to 2014 has often been sparse and fragmentary, the West African outbreak of Ebola provided an important opportunity to learn more about EBOV in the human host. Many of these studies are ongoing and new information will become available in due course.

Controlled studies carried out in the laboratory with a range of susceptible laboratory animals have provided the most comprehensive information about disease pathogenesis (Warfield et al. 2009). Generally, small rodents such as guinea pigs and genetically engineered mice have provided useful insight into host–filovirus interactions, virus adaptation studies (Dowall et al. 2014), and rapid assessment of antiviral therapy options. However, disease pathogenesis in rodent models is not as relevant to the human viral haemorrhagic fever (VHF) disease as that observed in NHPs (Bray et al. 2001; Geisbert et al. 2002). Consequently, detailed disease pathogenesis data have been obtained from experimental infections of NHPs and when possible human clinical studies.

It is unclear what constitutes a typical infectious dose for the different routes of exposure leading to human infection. Infection follows direct contact most commonly with an infected patient or recently deceased cadavers (Khan et al. 1995). Abrasions in the skin, parenteral inoculation, and exposure of mucosal surfaces to infected body fluid are all potential routes of entry. Infectious filoviruses or more commonly viral RNA have been isolated from both semen and genital secretions (Rodriguez et al. 1999; Rowe et al. 1999) and virus detected by immunohistochemistry (IHC) in skin samples of human cases (Zaki et al. 1999). Virus has also been found in nasal secretions and the skin of NHPs. Viraemia associated with late-stage illness in fatal cases is often intense and can reach levels of 10⁷ to 10⁸ plaque-forming units (pfu)/ml in humans (World Health Organization 1978). These viral loads are readily observed by direct electron microscopic inspection of post-mortem fluids and tissues.

Infection through butchering of bush meat was identified as a likely route of human infection in outbreaks of EBOV in Gabon (Georges-Courbot et al. 1997). While adequate heating of food during cooking will inactivate infectious filoviruses, consumption of contaminated foods has not been completely eliminated as a possible route of infection. Infectivity titres in organs of filovirus-infected NHPs are frequently in the 10⁷ to 10⁹ pfu/g range (Geisbert et al. 2008). Ingestion has not been identified as a route of infection in human cases, but oral administration of EBOV to rhesus macaques leading to a fatal infection has been documented (Jaax et al. 1996).

Pathologic changes seen in patients dying with filovirus infections seem consistent; all exhibit extensive necrosis of the parenchymal cells in most organs, with little inflammation within infected tissues (Dietrich et al. 1978; Zaki and Goldsmith 1999). A range of organs are affected: liver, spleen, kidney, and the gonads. Of the affected organs, the liver shows the most characteristic histopathologic features—hepatocellular necrosis is widespread, with intact, hyalinized, ghost-like cells often remaining in place amid large amounts of karyorrhectic debris. Exceptional numbers of virions are often present in these debris and intracytoplasmic inclusion bodies are present within intact hepatocytes. Studies using a range of approaches, including light microscopy, electron microscopy, IHC, and in situ hybridization, show consistent cellular damage and the widespread presence of viral antigens and nucleic acid, suggesting that direct viral damage is a major element in the pathogenesis of the disease.

Filoviruses have broad cell tropism; monocytes, macrophages, dendritic cells, endothelial cells, fibroblasts, hepatocytes, adrenal cortical cells, and several types of epithelial cells all support replication of these viruses (Baskerville et al. 1978; Zaki et al. 1999). The exact sequence of events during infection, however, is not fully resolved. Studies in NHPs involving sequential infection with EBOV and MARV suggest that the initial sites of replication are monocytes, macrophages, and dendritic cells (Hensley et al. 2011). These cells, which are mobile, play a central role in the dissemination of virus from the initial sites of infection to regional lymph nodes. This most likely occurs through the lymphatics and through the blood to the liver and spleen. At these locations, filoviruses infect resident macrophages and dendritic cells.

Several lines of evidence suggest that filovirus-infected monocytes/macrophages release various soluble factors that recruit additional monocytes/macrophages to the site of infection; this recruitment makes more target cells available, which further amplifies the infection (Geisbert et al. 2003; Bray and Geisbert 2005). The liver and adrenal glands, as well as the spleen, also appear to be important target organs for both EBOV and MARV, and this tropism plays a key role in disease pathogenesis. Raised liver enzyme levels are striking observations (Fisher et al. 1992) in most filovirus infections, as are various degrees of hepatocellular degeneration and necrosis. The hepatocellular lesions are generally not severe enough to be the cause of death. However, impairment of the liver could contribute to the overall pathogenesis. Decreased synthesis of coagulation factors and other plasma proteins as a result of severe hepatocellular necrosis may contribute to the bleeding diathesis. Adrenocortical infection and necrosis are also observed in filovirus infections of humans and NHPs (Geisbert et al. 2007). The adrenal cortex has a key role in homeostasis and particularly the maintenance of blood pressure. Reduced secretion of steroid-synthesizing enzymes following adrenal damage leads to hypotension and sodium loss with hypovolaemia, which have been noted in nearly all cases of filovirus disease. This suggests that reduced adrenocortical function may contribute to the development of shock that characterizes late-stage MVD and EVD.

In both EBOV and MARV infection in experimentally infected NHPs (Alves et al. 2010), depletion and necrosis of lymphoid cells are commonly seen in the spleen, thymus, and lymph nodes of fatal cases in the absence of an inflammatory response. Lymphopenia is a consistent feature of filovirus infections of humans and NHPs. Lymphocytes have not been shown to support infection with filoviruses, despite the massive loss of lymphocytes during filovirus infection. Apoptosis of large numbers of lymphocytes occurs during both EBOV and MARV infections in humans and experimentally infected NHPs (Alves et al. 2010), which may partly explain the progressive lymphoid depletion seen in fatal infections. The mechanism(s) leading to apoptosis and the loss of ‘bystander’ lymphocytes during the course of filovirus infection are unknown but are thought to be initiated through a range of agonists or pathways. These are likely to include the tumour necrosis factor (TNF)-related apoptosis-inducing ligand (TRAIL), the Fas death receptor pathways (Geisbert et al. 2003), impairment of dendritic cell function, and abnormal production of soluble mediators such as nitric oxide (NO). Direct interactions between lymphocytes and filovirus proteins may play a role, and the recognition of an immunosuppressive motif in the C-terminal region of the EBOV and MARV glycoproteins (Yaddanapudi et al. 2006) supports this idea.

Filovirus infection of humans and NHPs induces the secretion of inflammatory mediators including the interferons (IFNs), interleukin (IL)-6, IL-8, IL-10, IL-12, interferon-inducible protein (IP)-10, monocyte chemoattractant protein-1 (MCP-1), the chemokine RANTES, TNF-α, reactive oxygen, and nitrogen species (Wauquier et al. 2010). In vitro filovirus infection of a range of primary human cells also leads to the secretion of many of the same inflammatory mediators. Ultimately, it appears that virus-induced secretion of these molecules leads to an immunologic imbalance (an immunopathology) that contributes to disease progression. Despite these clues, details of the inflammatory response after filovirus infection are still not fully understood. Conflicting results, such as the high levels of circulating IFN, were noted in acute-phase sera of patients infected with EBOV in one study (Villinger et al. 2010), but not detected in a subsequent similar study (Baize et al. 2002), and are not yet resolved. Investigations of EBOV patients who have survived, including cases of very mild disease or asymptomatic infection (Leroy et al. 2001), have suggested that an initial increase in cytokines, including IL-1β, IL-6, and TNF-α, is important for controlling infection; this is followed by a return to baseline levels. This is in contrast to the general picture of pathogenesis and immunopathology that may also be induced by an inappropriate pro-inflammatory response in severe cases. This is a key line of research being undertaken by the EVIDENT programme using samples obtained from the West African EBOV outbreak in 2015 (EVIDENT).

Inhibition of the type I IFN response appears to be a feature of filovirus pathogenesis, which was initially indicated by studies of EBOV-infected endothelial cells (Harcourt et al. 1998). Subsequently, VP35 was shown to block IFN regulatory factor (IRF-3) activation, which prevents transcription of IFN-β, so that it functions as a type I IFN antagonist (Basler et al. 2000). VP24 expression also interferes with type I IFN signalling, and mutations in VP24 have also been linked to the adaptation of EBOV leading to fatal disease in guinea pigs (Dowall et al. 2014). Interestingly, in MARV, a different mechanism for evading the host IFN response has been described; MARV VP40 has been shown to block Janus kinases phosphorylation and their target STAT proteins in response to type I and type II IFN and IL-6 (Valmas et al. 2010). Similar to VP24 in EBOV, mutations in MARV VP40 have been linked to adaptation leading to fatal disease in guinea pigs.

A number of studies point to an important role for reactive oxygen and nitrogen species in filovirus disease pathogenesis (Geisbert and Jahrling 2004). Abnormal nitric oxide production has been linked to a number of pathological conditions including tissue damage and vascular leakage, which is a key contributor to shock seen in severe disease. In general, NO is induced in monocytes and macrophages to control infection. However, in the case of filoviruses, since monocytes and macrophages are key cells for early viral replication, this may make disease worse due to the production of large amounts of NO. This NO production also results in reduced lymphocyte proliferation, cell damage, and hypotension, which is a consistent feature in severe EVD.

The current data point to a compromised immune response which subsequently leads to high levels of systemic virus. Filoviruses spread via mobile cells, which also secrete pro-inflammatory mediators which are important in late stages of disease. These subsequently lead to the development of haemorrhage and shock. Indeed, the current favoured hypothesis to explain filovirus pathogenesis is that the infection and activation of monocytes/macrophages is central to the pathogenesis of severe disease. The release of pro-inflammatory mediators causes damage to the vasculature and the coagulation system, which leads to multiple organ failure and a syndrome clinically similar to septic shock (Bray and Geisbert 2005).

Clearly the vascular system is damaged by filovirus infections, but whether this is the result of direct viral protein–host interactions or a consequence of an unregulated release of inflammatory mediators following infection is uncertain and has been under investigation for some time. In 2000, a study speculated that EBOV GP was the key viral protein linked to vascular cell injury and that EBOV infection of endothelial cells induces structural change (Yang et al. 2000), contributing to haemorrhage. However, while human and NHP endothelial cells are susceptible to EBOV and MARV infection, other studies have indicated that overt cytopathology is not induced and that endothelial cells only become infected at very late stages of infection (Geisbert et al. 2003a). Additionally, an immunohistochemical survey of a fatal case of MARV infection showed infrequent infection of endothelial cells (Geisbert et al. 1998). Clearly, disturbance of the blood tissue barrier is an important component of filovirus disease, but early histologic observations of autopsy tissues do not show the presence of vascular lesions (Murphy 1978) and there have been no reports of vascular lesions in any subsequent studies to date. There is also no evidence of significant vascular lesions in filovirus-infected NHPs (Ryabchikova et al. 1999; Hensley et al. 2011). Other types of studies have supported a hypothesis of indirect damage to the endothelium as a result of filovirus-induced cytokine release. For example, increased endothelial permeability was associated temporally with the release of TNF-α from MARV-infected human monocytes/macrophages (Feldmann et al. 1996). Subsequent studies showed that EBOV-induced cytokine release led to activation of the endothelium, as demonstrated by a breakdown of barrier function, associated with reorganization of the actin cytoskeleton and integrins (Wahl-Jensen et al. 2005). Indeed most studies indicate that changes in the integrity of the endothelium are influenced primarily by local or systemic increases in levels of cytokines and other host cell factors triggered by infection.

Perturbation of blood coagulation and fibrinolysis during EVD and MVD present clinically as petechiae, ecchymoses, mucosal/nasal haemorrhages and haematomas, or uncontrolled bleeding at venepuncture sites. However, severe blood loss is rare and, when present, is most commonly from the gastrointestinal tract. The scale of blood loss, even in severe cases, is not significant enough to account for death. The coagulopathy seen in filovirus infections is characterized by thrombocytopenia, consumption of clotting factors, and increased levels of fibrin degradation products. Clinical laboratory data suggest that disseminated intravascular coagulation (DIC) is an important feature of human EVD (Rollin et al. 2007), and D-dimer levels were substantially increased in all patients with SUDV infections reaching levels four times higher in patients with fatal disease than in patients who survived (Rollin et al. 2007). However, documentation of DIC in human filovirus infections is sparse, which may be in part due to difficulties in performing studies in inaccessible geographic settings. Perhaps unsurprisingly, the coagulopathy is more clearly defined for NHPs. Studies have documented the histologic and biochemical picture of DIC during EBOV infection in a range of NHP species (Ryabchikova et al. 1999; Jeffs 2006; Ebihara et al. 2011). For MARV, the histologic or biochemical picture of DIC has been reported in a handful of cases studied (Geisbert and Jaax 1998) and in experimentally infected NHPs (Hensley et al. 2011). While it is likely that several factors contribute to the development of filovirus coagulopathy, evidence implicate the expression of tissue factor and release from EBOV-infected monocytes/macrophages as an important factor in the development of the coagulopathy (Geisbert et al. 2003b).

Filovirus infections are commonly thought to be the most severe of the known VHFs; however, there is limited information and close observations of acute human cases have not always been possible. The West African EBOV outbreak of 2013–16 has enabled detailed investigations to take place and new information in relation to EVD is expected to become available soon. Following an incubation period averaging 4 to 10 days (range 2 to 21 days), there is an abrupt onset of flu-like symptoms, including fever, chills, malaise, and myalgia. New symptoms continue to develop, which indicate multisystem involvement. These are systemic, such as prostration, and involve a range of signs and symptoms involving the gastrointestinal system (anorexia, nausea, vomiting, abdominal pain, and diarrhoea), the respiratory system (chest pain, shortness of breath, and cough), the vascular system (conjunctival injection, postural hypotension, and oedema), and the neurological system (headache, confusion, and coma manifestations). Coagulopathy and overt haemorrhage develop in a minority of cases. These include petechiae, bruising, uncontrolled oozing from venepuncture sites, and mucosal haemorrhages. Often, 5 to 7 days after the onset of illness, a maculopapular rash associated with varying degrees of erythema appears, which is a useful feature to aid the differential diagnosis (although it is often difficult to discern) and is usually followed by desquamation in patients who survive. Late-stage disease is characterized by shock and severe metabolic disturbances. Convulsions are common as is a diffuse coagulopathy.

Laboratory parameters are less characteristic than the clinical profile, but the following observations are associated with filovirus disease: an early leucopenia (as low as 1000/ml) with lymphopenia and subsequent neutrophilia, atypical lymphocytes with the majority in immature stages of development; a thrombocytopenia with levels of 50 000–100 000/ml common; prothrombin and partial thromboplastin times are frequently prolonged, and fibrin split products are detectable; elevated serum transaminase levels with the aspartate aminotransferase (AST) typically exceeding alanine aminotransferase (ALT) are seen in many VHFs; hyperproteinaemia and proteinuria are seen.

Later in the course of disease, secondary bacterial infection is seen and may lead to elevated white blood counts. There is an increased risk of abortion for pregnant women. Fatal cases develop clinical signs early during infection, and death typically occurs between days 6 and 16, often due to hypovolaemic shock. Fatal cases are generally associated with viral loads equivalent to 10⁷ copies of EBOV per ml or greater, when measured by polymerase chain reaction (PCR) (Towner et al. 2004), and this has become one the most influential predictors of mortality. Non-fatal cases generally show fever for approximately 5 to 9 days when there is an improvement, which typically occurs at around days 7 to 11. It is noteworthy that this is simultaneous with the development a humoral antibody response (Ksiazek et al. 1999). Recovery can be slow and myelitis, recurrent hepatitis, and psychosis sometimes occur (Rowe et al. 1999; Feldmann and Geisbert 2011; Mehedi et al. 2011). Recent findings in patients linked to the 2013–16 West African outbreak have detailed a common range of pathologic sequelae such as polyarthritis (Howlett et al. 2016), chronic headache, and eye symptoms. In one case, highlighted by uveitis in a convalescent patient, further investigation showed the presence of EBOV in ocular fluid (Varkey et al. 2015). Such viral recrudescence after earlier clearance is extremely rare and while this does not indicate that all eye symptoms are the result of virus persistence, it is clear that there are certain sanctuary sites where EBOV can persist in immune-privileged tissues such as the eyes and also the testes (Christie et al. 2015). These complicate the risk of transmission events following supposed recovery and clearance (Bausch et al. 2007) and are of ongoing concern.

Filovirus disease can be suspected in acute febrile patients with a history of travel to an endemic area associated with possible epidemiological exposure. Clinical diagnosis is often difficult, due to a wide variety of more common infectious diseases causing similar clinical symptoms such as malaria and typhoid fever in West Africa. Other infections to consider are shigellosis, salmonellosis, meningococcal disease, plague, leptospirosis, anthrax, relapsing fever, rickettsial infection including typhus, yellow fever, Rift Valley fever, chikungunya, and dengue fever. Risk factors to consider include cave (bat) exposure, treatment in local hospitals, and direct contact with sick or dead persons or wild animals, particularly monkeys and apes. These are useful historical features, especially in travellers returning from endemic areas of Africa. For patients with filovirus disease, prostration, lethargy, and diarrhoea seem to be more severe than with other VHF infections; the appearance of a characteristic rash helps the differential diagnosis, but this is most evident in Caucasian patients. Diagnosis of single cases is difficult, but the occurrence of clusters of cases with prodromal fever followed by severe diarrhoeal illness with or without haemorrhagic diatheses, a high CFR, and person-to-person transmission are suggestive of VHF and require the implementation of containment procedures. During outbreaks, health-care workers, who have direct contact with patients, are at high risk for infection; adequate barrier nursing precautions should be implemented when collecting samples (Peters et al. 1996; Adamson et al. 2015).

Irrespective of the capabilities of the laboratory, the initial presumptive diagnosis of filovirus infection is often based on the described clinical and risk factor assessment. Clinical microbiology and public health laboratories in almost all countries are not equipped to diagnose VHF infections, particularly those caused by filoviruses. Many countries have recommended investigation algorithms to send specimens to national and/or international reference laboratories capable of performing the required assays safely. Laboratory diagnosis of filovirus infections is now generally based on reverse transcriptase (RT)-PCR and antigen detected by enzyme-linked immunosorbent assay (ELISA) (Niikura et al. 2001; Strong et al. 2006) which are the primary test systems to diagnose an acute infection. Laboratory confirmation can also be achieved by detection of specific immune responses in infected individuals. For antibody detection, the most commonly used assays are direct IgG and IgM ELISAs and IgM capture ELISA (Ksiazek et al. 1999; Strong et al. 2006). RT-PCR, antigen detection, and serology can be performed on materials that have been rendered non-infectious by radiation or chemical treatment.

Gamma or equivalent ionizing radiation is one of the most efficient means of inactivating specimens prior to antigen detection and serology; however, this is restricted to specialist laboratories and inactivation with heat and chaotropic agents has become common (Smither et al. 2015). Samples for RT-PCR (requiring nucleic acid extraction) can be treated with guanidinium isothiocyanate, a chaotropic agent that denatures proteins and destroys the functionality of viral enzymes, making samples non-infectious. These methods of inactivation allow the safe manipulation of material outside of the containment laboratory where work can be carried out more expediently.

Viral nucleic acid detected by RT-PCR and antigen detected by ELISA or point of care test can be detected in blood as early as the onset of symptoms and up to 16 days post-onset of symptoms (Feldmann and Geisbert 2011). PCR is significantly more sensitive, but antigen detection is simpler to perform. Serum or whole blood samples are generally used for initial diagnosis, but virus is detectible by PCR in urine after the viraemia has cleared and for a prolonged period in semen samples. IgM antibodies can be detected as early as 2 days post-onset of symptoms and persist for between 30 and 168 days after infection. Virus-specific IgG antibodies are detectable between days 6 and 18 after onset and persist for many years (Feldmann and Geisbert 2011; Mehedi et al. 2011). Other serologic tests that have been used in filovirus diagnosis are the western blot assay and the indirect immunofluorescence assay on gamma (or otherwise) inactivated fixed cells infected with reference laboratory filoviruses or containing expressed recombinant filovirus proteins (Johnson et al. 1983). The relatively high viraemia levels in humans bring diagnostic advantage from electron microscopy techniques for filovirus infections (Jahrling et al. 1990). Significantly, electron microscopy proved key to the initial laboratory diagnosis of EBOV in the 2013–16 West Africa outbreak in 2014 (Baize et al. 2014).

Isolation of infectious virus from serum or other clinical material is a relatively simple and sensitive procedure and should always be attempted if high containment laboratory infrastructure is available. Filoviruses grow in a wide variety of cell lines, although Vero cells (or the E6 clone) have been most often used. During primary isolation work in vitro, the development of cytopathic effect is often subtle and commonly absent. In consequence, guinea pigs have often been used for primary isolation of those filoviruses that do not seem to grow well in tissue culture. In guinea pigs, however, repeated passage is usually required to produce severe/fatal disease; adaptation in this way also results in the development of several mutations in the filovirus genome (Dowall et al. 2014).

Standardization and evaluation of diagnostic procedures for filoviruses are routinely carried out in key reference laboratories, but generally this is difficult to disseminate because of the restricted availability of virological and clinical material. Nevertheless, the European Network for Imported Viral Diseases (ENIVD) provides external quality assurance for filovirus RT-PCR diagnostic procedures (Niedrig et al. 2004) after virus inactivation. Filovirus outbreaks usually occur in remote areas where sophisticated medical support systems are limited and timely diagnostic services are extremely difficult to provide. The provision of a field laboratory offering basic diagnostics for filoviruses and other agents can greatly aid in the management of patients specifically, and the outbreak in general. The development of portable real-time thermocyclers, immunologic assays, and simple portable class III biosafety cabinets—flexible film isolators—for use in the field has made the rapid deployment of a mobile diagnostic laboratory a reasonable undertaking. Indeed several laboratories, both mobile and more permanent, were rapidly established over the course of the 2013–16 West Africa EBOV outbreak (Diers et al. 2015), using such isolators. In particular, the European Mobile Laboratory was the first deployed unit in March 2014 and went on to support the diagnosis of over 50% of the cases in Guinea where it was initially established (European Mobile Laboratory Project).

The natural reservoir(s) of filoviruses remains unclear despite increased numbers of outbreaks and opportunities to investigate their origins (Feldmann et al. 2004; Peterson et al. 2004; Heeney 2015; Leendertz et al. 2016). As classic zoonotic agents, filoviruses are likely to persist in an animal (or several animals) or arthropods without causing disease, which can transmit the virus directly to humans, great apes, NHPs, or an interim amplifying host (Formenty et al. 1999; Monath 1999). However, extensive arthropod field surveys have failed to detect the presence of EBOV and they do not replicate well in arthropod cells nor inoculated arthropods (Turell et al. 1996). It is therefore unlikely that arthropods are an intermediary or a reservoir for filoviruses. There is, however, growing evidence of the involvement of bats in filovirus outbreaks and there has been circumstantial as well as direct evidence that bats may be involved, e.g. in the early SUDV outbreaks in Sudan (Arata and Johnson 1978), the EBOV outbreak in the DRC in 2007 (Leroy et al. 2009), the MARV infections in Kenya (Johnson et al. 1996) and the DRC (Bausch et al. 2003), as well as the EBOV outbreak in West Africa from 2013 to 2016 (Marí Saéz et al. 2014). Interestingly, experimental infection of wild African fruit and insectivorous bats has shown that these animals are capable of supporting the replication of EBOV without becoming ill, despite supporting high levels of circulating virus (Swanepoel et al. 1996). Findings of asymptomatic EBOV and MARV infections in fruit bats are additional evidence that such animals are capable of harbouring filoviruses and may serve as reservoir species. However, while MARV has been isolated from bat species such as R. aegyptiacus (Leroy et al. 2005; Towner et al. 2009), EBOV have so far only been identified from bats by PCR and serological signatures—viruses have not been isolated (Heymann et al. 1980). Nevertheless, since persistently infected hosts are a key requirement for zoonotic diseases such as those caused by filoviruses, chronic infection in bats or other small animal species are likely to be involved in their ecology (Heeney 2015).

Implementing a strategy for the prevention of primary filovirus infections of humans is problematic, as the natural reservoirs and factors that affect filovirus movement in the wild are still largely unknown. Assuming that bats serve as a reservoir, proper education in the context of zoonotic spread would be a feasible way of prevention.

It has generally been difficult to identify a human index case, let alone the type of contact that initiated the infection, and rarely has the epidemic origin (the primary infectious event) been identified. Indeed, among the 21 most documented outbreaks of EVD in Africa, an index case was only identified four times and hypothesized in two other instances (World Health Organization 1978b; Heymann et al. 1980; George et al. 1997; Khan et al. 1999; Leroy et al. 2009; Baize et al. 2014). Once it enters a human or an NHP population, it is clear that the virus is spread through close contact with acutely infected members. Isolation of patients and use of strict barrier nursing procedures, including the use of basic protective equipment (gloves, visor, face mask), have been sufficient to rapidly interrupt transmission in the hospital (Jeffs et al. 2007). Cadavers from fatal cases, however, represent a residual risk for community members, and unprotected handling of corpses should be avoided (Roddy et al. 2007). The scale of the 2013–16 EBOV epidemic has been linked to transmission events from fatal cases and traditional funeral practices (Victory et al. 2015); reducing the opportunities for transmission through the introduction of safe burial practices, together with education about other route exposure, such as sexual transmission, through dialogue with community leaders, was key to its eventual control (Nielsen et al. 2015).

Before the 2013–16 EBOV epidemic in West Africa, filovirus infections were managed solely with supportive therapy, directed towards maintenance of effective blood volume and electrolyte balance. Detailed knowledge of filovirus replication, pathogenesis, and host responses to infections has steadily identified new targets for therapeutic intervention, and several experimental options became available (and were used) during the West Africa outbreak.

The passive administration of antibodies targeting the EBOV surface GP (Saphire 2013) has shown promise for treating severe EVD. The principle was first established in NHPs when immunoglobulin prepared from a macaque, which survived EBOV, conferred protection in rhesus macaques, when it was administered 2 days after infection with EBOV (Dye et al. 2012). Multiple studies have now shown that treatment with EBV-specific monoclonal antibodies (MAbs) confers protection if given post-exposure and that this effectiveness is enhanced by the use of several MAbs given together as a ‘cocktail’ (Takada et al. 2003; Olinger et al. 2012; Qiu et al. 2012). Two of the most widely studied anti-EBOV MAb combinations are ZMAb (MAbs: 2G4, 4G7, and 1H3) (Takada et al. 2003) and MB-003 (MAbs: 13C6, 6D8, and 13F6) (Qiu et al. 2014). All six MAbs in these cocktails were isolated following immunization of mice. ZMAb and MB-003 showed partial protection against EBOV in a range of disease models (mice, guinea pigs, and NHPs). A more potent combination of MAbs (ZMapp) was subsequently produced, which was derived from ZMAb and MB-003 that was effective in reversing clinical signs in six out of six rhesus macaques, even when given as late as 5 days post-EBOV exposure (Lyon et al. 2014). ZMapp and ZMAb have now been used to treat human EBOV infections under emergency compassionate protocols developed for the 2013–16 EVD outbreak in West Africa (Roos 2014). At least seven EBOV-infected patients have been treated with ZMapp and five survived (Bishop 2015; Meyers et al. 2015). Six patients have been treated with ZMAb and all patients survived. The treatment protocols have been reported to be well tolerated. In the absence of a clinical trial, the survival of these patients cannot be directly attributed to treatment with the MAb combinations, but due to their promise for the treatment of human infection, the production and clinical testing of anti-EBOV MAb cocktails are now being rapidly accelerated (McCarthy 2014). A similar approach but one based on a very much cheaper strategy of polyclonal antibody production in sheep has recently been shown to be effective at an in vitro (Dowall et al. 2016) and in vivo level (Dowall 2016, personal communication). While this is currently less well tested than monoclonal approaches, it has the advantage of being more affordable in the developing world. Importantly, this polyclonal approach restricts the opportunity of antiviral resistance through the development of escape mutants, which have been reported for monoclonal treatments (Enterlein et al. 2006).

In addition to the envelope GP, other important antiviral targets have been the viral transcription and replication machinery. Unfortunately, ribavirin, a broad-spectrum synthetic guanosine analogue with virustatic activity against a number of other RNA viruses including arenaviruses and bunyaviruses, has no in vitro or in vivo effect on filoviruses (Huggins 1989). However, other strategies to interfere with transcription and replication include antisense oligonucleotides, phosphorodiamidate morpholino oligomers, and RNA interference (RNAi) (Enterlein et al. 2006; Geisbert et al. 2006; Swenson et al. 2009; Warren et al. 2010). While many of these approaches are promising, they may be limited by sequence specificity (genetic variation of species), production costs, and the route of administration (mainly intravenous). Two potentially useful antiviral drugs for EVD treatment are: favipiravir (van Herp et al. 2015), a pyrazinecarboxamide derivative, developed and licensed in Japan for use against influenza virus, which has been shown to be effective against EBOV (Oestereich et al. 2014); and GS-5734, a pro-drug of adenine nucleotide analogue showing similarly encouraging results in NHPs (Warren et al. 2015). A clinical trial of favipiravir in West Africa has recently reported a potential benefit for all but the most severe cases (Sissoko et al. 2016) and recently reported in preliminary format (Mohammadi 2014). Similar trials of GS-5734 are under way.

A range of vaccine approaches to protect against filovirus infections have been developed and evaluated in established animal models. These have ranged from naked DNA (Yang et al. 2003) and virus-like particles (Ye et al. 2006); replication-deficient vectors such as adenovirus (Wang et al. 2006), Venezuelan equine encephalitis virus replicons (Pushko et al. 2001); and replication-competent live attenuated vectors such as vaccinia virus (Chepurnov et al. 1997), vesicular stomatitis virus (Geisbert et al. 2008), parainfluenza virus type 3 (HPIV3) (Bukreyev et al. 2006), and Newcastle disease virus (DiNapoli et al. 2010). Through advances in reverse genetic technology, an additional concept based on a replication-deficient EBOV which lacks VP35 has also been developed (Halfmann et al. 2009). Most of the vaccine approaches showed protective efficacy in rodent models, but several failed to protect NHPs. Currently, the following vaccine candidates are being further evaluated: ChAd3-EBOV (Phase I), rVSV-EBOV (Phase I), Ad26-EBOV (Phase I), ChAd3-EBOV (Phase II), and VSV-EBOV (Phase III). Interestingly, those vaccines based on recombinant attenuated VSV expressing EBOV GP, instead of their own G protein (VSV-EBOV), seem to be the most effective (Arnemo et al. 2015; Huttner et al. 2015; Matassov et al. 2015). A ring vaccine trial using this candidate was set up in July 2015 comprising approximately 4000 volunteers in Guinea and approximately 3000 frontline workers in Sierra Leone (Henao-Restrepo et al. 2015; Ohimain 2016) has shown promise and it is expected that effective candidates will be identified from the recent advances in the science of Ebola in the near future.