Neglected filoviruses

Abstract

Eight viruses are currently assigned to the family Filoviridae. Marburg virus, Sudan virus and, in particular, Ebola virus have received the most attention both by researchers and the public from 1967 to 2013. During this period, natural human filovirus disease outbreaks occurred sporadically in Equatorial Africa and, despite high case-fatality rates, never included more than several dozen to a few hundred infections per outbreak. Research emphasis shifted almost exclusively to Ebola virus in 2014, when this virus was identified as the cause of an outbreak that has thus far involved more than 28 646 people and caused more than 11 323 deaths in Western Africa. Consequently, major efforts are currently underway to develop licensed medical countermeasures against Ebola virus infection. However, the ecology of and mechanisms behind Ebola virus emergence are as little understood as they are for all other filoviruses. Consequently, the possibility of the future occurrence of a large disease outbreak caused by other less characterized filoviruses (i.e. Bundibugyo virus, Lloviu virus, Ravn virus, Reston virus and Taï Forest virus) is impossible to rule out. Yet, for many of these viruses, not even rudimentary research tools are available, let alone medical countermeasures. This review summarizes the current knowledge on these less well-characterized filoviruses.

While Ebola virus dominates the headlines, other filoviruses remain largely uncharacterized but may pose equal risks to humans.

Graphical Abstract Figure.

While Ebola virus dominates the headlines, other filoviruses remain largely uncharacterized but may pose equal risks to humans.

INTRODUCTION

Viruses with single-stranded negative-sense RNA genomes can be assigned to three unofficial supergroups based on whether these genomes are multisegmented, circular or unsegmented, and on evolutionary relationships of the genome-encoded RNA-dependent RNA polymerases (Li et al. 2015). All of the unsegmented viruses are currently assigned to the rapidly expanding order Mononegavirales. The family Filoviridae is one of the eight mononegaviral families (Afonso et al. 2016). Filoviridae members are differentiated from other mononegaviruses not only by genomic sequence, but also by the formation of characteristically shaped, filamentous virions. In addition, filoviruses are also differentiated from other mononegaviruses by comparatively long (≈19 kb) genomes containing gene overlaps, unique transcriptional initiation and termination signals, and an open reading frame (ORF) encoding a unique structural protein without obvious homologs in other mononegaviruses (i.e. VP24) (Kuhn et al. 2010, 2011). Filoviruses are further assigned to seven species included in the three genera Cuevavirus, Ebolavirus and Marburgvirus (Table 1) based on individual molecular properties (Kuhn et al. 2010, 2011; Bao, Chetvernin and Tatusova 2012; Lauber and Gorbalenya 2012). Six of the currently eight recognized filoviruses are known human pathogens causing two diseases officially recognized by the World Health Organization: Ebola virus disease (EVD) and Marburg virus disease (MVD). The remaining two filoviruses are known or suspected animal pathogens (Table 2). In general, filoviruses can be considered exotic pathogens. Human and/or animal filovirus disease outbreaks are rare. Only 37 outbreaks have been recorded in the almost 50 years since the discovery of filoviruses, and of those, only 11 included more than 100 cases each (Table 3) (Kuhn 2015).

Table 1.

Official taxonomy of the family Filoviridae as of 2016 (Kuhn et al. 2010, 2011; Bukreyev et al. 2014).

Table 2.

Official human filovirus disease classification and nomenclature as of 2016.

Table 3.

Cumulative human filovirus disease cases (updated from Kuhn 2015, as of March 27, 2016).

Based on the publication records (Kuhn 2008), the filoviruses that are currently most characterized and understood are Ebola virus (EBOV), Sudan virus (SUDV) and Marburg virus (MARV). The filovirus best known and feared by the general public is EBOV. This notoriety is in part due to numerous popular science publications, fiction and movies (Semmler 1998; Kuhn 2008; Blakey et al. 2015) and in part because EBOV had caused the most recorded human filovirus infections of all filoviruses, i.e. 1101 until 2012 (Kuhn 2015). At the end of 2013, EBOV caused an unprecedented EVD outbreak in Western Africa, with thus far 28 646 human infections and 11 323 deaths (case-fatality rate [CFR] = 39.51%) (World Health Organization 2016). Unsurprisingly, this public-health emergency resulted in massively increased funding for basic and translational EBOV research, including the development and testing in clinical trials of potential medical countermeasures.

However, despite tremendous progress in understanding of filovirus infections in vitro and in vivo, where and how filoviruses are maintained in nature remain unclear. In addition, the circumstances or mechanisms that lead to the occasional filovirus emergence in human and other mammal populations are also unknown. To emphasize research more or less exclusively on EBOV, SUDV and MARV may quite literally be a fatal mistake (Anthony and Bradfute 2015; Kozak and Kobinger 2016). One cannot exclude the possibility of a future disease outbreak of the scope of the Western African outbreak caused by other filoviruses, including those viruses currently thought to be apathogenic for humans.

This review provides an overview of the current knowledge of these neglected filoviruses: Bundibugyo virus (BDBV), Lloviu virus (LLOV), Ravn virus (RAVV), Reston virus (RESTV) and Taï Forest virus (TAFV).

Bundibugyo virus

Bundibugyo virus (BDBV; pronounced ˌbʊndiː'bʊdʒɔː vɑɪrəs) was discovered in Eastern Africa during a human viral hemorrhagic fever (VHF) outbreak that probably began in or around August 2007 and lasted until January 2008. The outbreak affected the Bundibugyo and Kikyo townships in Bundibugyo District of Western Uganda Administrative Region, Uganda and ultimately amounted to 149 suspected cases and 37 deaths (CFR = 24.8%; alternative published statistics, based on different case definitions and index case identifications, are as follows: 116 cases and 39 deaths [33.6%]; and 131 cases and 42 deaths [32.1%]) (Towner et al. 2008; MacNeil et al. 2010; Wamala et al. 2010; Centers for Disease Control and Prevention 2015). A second EVD outbreak due to BDBV was recorded in August of 2012 around Isiro, Haut-Uele District, Province Orientale, in northeastern Democratic Republic of the Congo, roughly 400 km northwest from Bundibugyo (Fig. 1). When that outbreak was declared over in November 2012, 62 people had been infected and 34 had died (CFR = 54.8%; an alternative statistic, based on a different case definition, is 77 cases and 36 deaths [CFR = 46.8%]) (Albariño et al. 2013; Centers for Disease Control and Prevention 2015; Kratz et al. 2015). Therefore, the average CFR range of human BDBV infections (32.3%–41.1%; Table 3) is comparable to that of EBOV infections (≈41.4%). The often repeated notion that BDBV is ‘less virulent’ than EBOV in humans cannot be upheld based on available data.

Figure 1.

Geographical location of primary index cases causing disease outbreaks due to neglected filovirus infections. Countries with index cases are shown in light brown with outbreak locations marked as bold dots (disease outbreaks due to more prominent filoviruses, EBOV, SUDV or MARV, are indicated via faded dots for reference). Former country names are listed in parenthesis under present names. COD: Democratic Republic of the Congo; COG: Republic of the Congo. Adopted from (Kuhn 2015).

A number of BDBV isolates have been obtained (Table 4), and Ugandan Bundibugyo virus/H.sapiens-tc/UGA/2007/Butalya-811250 (BDBV/But-811250) was designated the type BDBV isolate (Kuhn et al. 2014). To date, all published in vitro and in vivo experiments involving BDBV have been performed exclusively with this isolate (Albariño et al. 2013). BDBV isolates from Uganda and Democratic Republic of the Congo share ≈98.6% genome identity. Genomic analyses revealed the BDBV genome organization to be identical to that of other ebolaviruses (Fig. 2) (Towner et al. 2008; Albariño et al. 2013).

Table 4.

Neglected filovirus isolates and sequences.

Figure 2.

Organization of neglected filovirus genomes. Genes and open reading frames (ORFs) are shown as rectangles and horizontal arrows, respectively. Wavy lines indicate incomplete sequencing of 3′ leader or 5′ trailer sequences. Entries for prominent filoviruses, EBOV, SUDV or MARV, are muted.

Genomic analyses also indicated that each of the two outbreaks was in all likelihood caused through singular introductions into the human populations with subsequent human-to-human transmission (Albariño et al. 2013). However, how these introductions occurred and why no further BDBV introductions have occurred since the end of the second outbreak in 2012 remain unclear. The Bundibugyo District borders the Democratic Republic of the Congo. A large portion of the district is part of the Rwenzori Mountains, the Semliki National Park and Game Reserve with its associated wildlife, including non-human primates, and domestic or economical activities, such as cocoa farming, hunting and fishing (Wamala et al. 2010). These potential contacts with wildlife suggest that the BDBV introduction into the human population was a zoonotic event. However, the precise beginning of the 2007 Ugandan EVD outbreak and the identity and history of the human index case is unclear (Towner et al. 2008; MacNeil et al. 2010; Wamala et al. 2010). Likewise, no data have been published on the behavior and recent history of the suspected human index case of the 2012 EVD outbreak.

MVD outbreaks due to MARV and RAVV could be directly linked to infected frugivorous bats from which replicating virus isolates could be obtained (Towner et al. 2009; Amman et al. 2012; Wahl-Jensen et al. 2013). EVD outbreaks due to EBOV have been loosely associated with bats via anti-EBOV antibody or EBOV genome fragment detection in bat sera in the absence of virus isolation (Leroy et al. 2005; Wahl-Jensen et al. 2013). Likewise, massive ape (central chimpanzee and western lowland gorilla) population declines have been temporary and spatially associated with human EVD outbreaks, and while EBOV isolation was unsuccessful, EBOV genome fragments were detected in a low number of apes (Huijbregts et al. 2003; Walsh et al. 2003; Bermejo et al. 2006). However, neither bat nor ape associations could be established for BDBV thus far. Injection of BDBV into Egyptian rousettes (Rousettus aegyptiacus), the presumed natural bat reservoir of MARV and RAVV (Amman et al. 2012), did not result in replication (Jones et al. 2015). Therefore, the ecology of BDBV remains a mystery. Only one ecological survey for BDBV was reported. IgG antibodies against BDBV glycoprotein (GP_1,2) antigen were detected by ELISA and western blot in 9/353 (2.6%) of orangutans (Pongo pygmaeus) sampled on Kalimantan Island, Indonesia, in 2006. BDBV-specific IgM could not be detected (Nidom et al. 2012). These results may suggest that the BDBV distribution is not restricted to Eastern Africa. However, caution is advised as the inter-ebolavirus and non-filovirus GP_1,2 cross-reactivity of naturally occurring antibodies is undefined (antibodies to yet unknown filoviruses or unrelated antibodies may react with BDBV GP_1,2). For instance, IgG and IgM antibodies from survivors of the 2007 BDBV/EVD outbreak strongly and weakly cross-react with other, non-BDBV, ebolavirus antigen preparations, respectively (Macneil, Reed and Rollin 2011). In the absence of BDBV genome detection by deep sequencing or BDBV isolation in cell culture, serological results at best hint toward the whereabouts of BDBV in nature. Thus, at the moment, future EVD outbreaks due to BDBV can neither be geographically nor temporally anticipated, let alone be prevented.

The clinical presentation of EVD due to BDBV infection in Uganda in 2007 is summarized in Table 5 (MacNeil et al. 2010; Roddy et al. 2012). Based on these very limited data and additional statistics published for the 2012 outbreak (Kratz et al. 2015), the overall distribution of clinical signs and symptoms of BDBV infections appears highly similar to that of EBOV, SUDV and MARV infections (not reviewed here, see Siegert et al. 1967; Egbring, Slenczka and Baltzer 1971; Isaäcson et al. 1978; Piot et al. 1978; Smith, Francis and Simpson 1978; Bwaka et al. 1999; Bausch et al. 2006; Barry et al. 2014; Maganga et al. 2014; Schieffelin et al. 2014; Bah et al. 2015; Dallatomasina et al. 2015; Lado et al. 2015; Qin et al. 2015; Yan et al. 2015). Consequently, BDBV infection cannot be diagnosed based on clinical observation alone. During the 2007 EVD/BDBV outbreak, the mean incubation period was 6.3 days (the longest measured incubation period was 25 days) (MacNeil et al. 2010, 2011) and, therefore, maybe somewhat shorter than that of EBOV infection [12.7 ± 4.3 days in a 1995 EVD/EBOV outbreak (Eichner, Dowell and Firese 2011) and 11.4 days in the 2013–2016 EVD/EBOV outbreak in Western Africa (World Health Organization Ebola Response Team 2014)]. Asthenia, diarrhea and headache were the most frequent symptoms/signs. Hemorrhagic signs were observed in approximately half of the cohort (MacNeil et al. 2010; Roddy et al. 2012), which is typical for all ebolavirus infections. As has been reported for EBOV, SUDV and MARV infections (Baltzer et al. 1979; Bwaka et al. 1999; Kibadi et al. 1999; Rowe et al. 1999; Qureshi et al. 2015), survivors of disease caused by BDBV may suffer of long-term sequelae years after convalescence. These sequelae include arthralgia, difficulty swallowing, hearing loss, ocular deficits (blurred vision and retro-orbital pain) and sleeplessness (Clark et al. 2015). Whether BDBV is able to persist in convalescent patients, as has been reported for EBOV and MARV (Martini and Schmidt 1968; Nikiforov et al. 1994; Rodriguez et al. 1999; Mate et al. 2015; Varkey et al. 2015), is unclear.

Table 5.

Comparison of clinical signs or symptoms of human BDBV infection during the 2007 Uganda EVD outbreak.

On the molecular level, EVD due to BDBV appears to differ from other ebolavirus infections. Whereas strong inflammatory responses (‘cytokine storms’) have been reported during EBOV infections and were associated with concomitant multi-organ dysfunction syndrome (Baize et al. 1999, 2002; Villinger et al. 1999), lethal human BDBV infections were associated with low concentrations of proinflammatory cytokines (interleukin [IL]-1α, IL-1β, IL-6, tumor necrosis factor-α) and high concentrations of antiinflammatory IL-10 (Gupta et al. 2012). These results indirectly confirm in vitro experiments with BDBV-infected peripheral blood mononuclear cells (PBMCs). BDBV-infected cells produced 10–100-fold less progeny virions and reacted with 2–10-fold lower expression of tumor necrosis factor-α, monocyte chemoattractant protein-1, IL-1β and macrophage inflammatory protein 1-α compared to EBOV-infected control cells. Interestingly, in contrast to the in vivo data, IL-10 expression was also low (Gupta et al. 2010).

Thus, pathogenetic mechanisms or properties from relatively well-characterized filoviruses (EBOV, SUDV and MARV) cannot be generalized to other, less characterized filoviruses. Consequently, promising medical countermeasures against one filovirus should not be assumed automatically to be promising avenues for related filoviruses. This lack of translatability is all the more concerning in regard to BDBV, as the virus itself is completely uncharacterized. With the exception of genomic sequence determination and functional deductions one can possibly make from alignment with other ebolavirus sequences, only three reports have been published addressing BDBV molecular biology directly. BDBV uses Niemann-Pick C1 (NPC1) as a cell-entry receptor (Ng et al. 2015); cell entry is dependent on cathepsin B in vitro (Misasi et al. 2012); and BDBV Δ-peptide has no effect on MARV replication (Radoshitzky et al. 2011).

The pathology of human BDBV infection remains unknown as neither autopsies nor biopsies were performed during the two disease outbreaks. Until future outbreaks might occur, the delineation of BDBV pathogenesis is solely dependent on animal models. Whereas numerous animal models are available for EBOV, SUDV and MARV infection (Table 6), the only established animal model for BDBV infection is based on intramuscular inoculation of crab-eating macaques (Macaca fascicularis). The conditions to achieve 100% fatality in this model have yet to be determined (typically, 25%–34% of infected animals survive) (Hensley et al. 2010; Falzarano et al. 2011; Mire et al. 2013). No detailed clinical or pathological descriptions of BDBV infection in crab-eating macaques are available. The establishment of a mouse model of BDBV infection was attempted using type 1 interferon α/β receptor knockout (IFN-α/βR^−/−) mice (Table 6). When exposed to BDBV, these mice did not develop signs of clinical disease, weight loss or lethality (Brannan et al. 2015). However, in some of BDBV-infected mice, transient viremia and increased tissue titers (spleen and liver) were noted. Results from serial sampling experiments indicated hepatic disseminated intravascular coagulation (DIC). Splenic lymphoid hyperplasia occurred concurrently with mild lymphoid depletion, and immunohistochemical analysis detected viral antigen in red pulp macrophage-like cells. Platelets were transiently decreased, and white blood cells and absolute lymphocytes were transiently increased.

Table 6.

Overview of animal models for filovirus infections.

The absence of true rodent models for BDBV infection excludes the possibility of high-throughput in vivo drug screening and initial candidate countermeasure evaluation. The only partially lethal non-human primate model requires countermeasure evaluation to be performed with large animal numbers to achieve statistical significance of outcomes, which is both ethically problematic and in most cases prohibited by cost and logistics in limited biosafety level-4 space. Despite these obstacles, the BDBV crab-eating macaque model has been used for limited evaluation of candidate medical countermeasure platforms that are under advanced development for the prevention or treatment of EVD due to EBOV infection. At least two candidate vaccine platforms (i.e. DNA prime/recombinant adenovirus boost, recombinant vesicular stomatitis Indiana virus) appear promising against BDBV (Table 7) (Hensley et al. 2010; Falzarano et al. 2011; Mire et al. 2013). Theoretically, administration of a pan-filovirus vaccine comprised of mosaic ebolavirus proteins could elicit antibody responses to a number of ebolaviruses, including BDBV (Fenimore et al. 2012), but such a vaccine has not been tested in animals with BDBV infection.

Table 7.

Overview of candidate vaccine development against neglected filoviruses.

Neither post-exposure prophylactics nor antivirals have yet been identified, let alone been tested, against BDBV infections in vivo. Consequently, treatment of human BDBV infections is limited to general supportive measures applied to other ebolavirus infections, such as nutritional supplementation, oral or intravenous fluid rehydration and medication against anxiety (e.g. diazepam), secondary infections (antibiotics, antimalarials, antimycotics), dyspepsia (e.g. cimetidine, ranitidine, omeprazole), nausea/vomiting (e.g., metaclopramide) and pain (e.g. acetominophen, morphine) (Roddy et al. 2012).

Diagnostic tests for BDBV infection are currently limited to various pan-filovirus reverse transcriptase-polymerase chain reaction (RT-PCR) or quantitative reverse transcriptase-PCR (qRT-PCR) protocols (Lu et al. 2015) and BDBV-specific sequence capture probes for next-generation sequencing (Koehler et al. 2014). A handful of BDBV-specific monoclonal or polyclonal murine, rabbit or crab-eating macaque antibodies (Ou et al. 2011; Holtsberg et al. 2015; Keck et al. 2015; Wang et al. 2015), BDBV cross-reactive murine antibodies (Fusco et al. 2015; Wang et al. 2015; Furuyama et al. 2016), BDBV cross-reactive murine single-chain variable domain fragments (scFv) and thermostable single-domain nurse shark antibodies (IgNAR V) raised against inactivated EBOV particles (Goodchild et al. 2011) are available, but have not yet been tested in diagnostic serological assays.

Lloviu virus

Lloviu virus (LLOV; pronounced j'ɔːvju vɑɪrəs) was discovered in 2002 in Cueva del Lloviu in Spain. Next-generation sequencing of tissues collected from several of hundreds of deceased (insectivorous) Schreibers's long-fingered bats (Miniopterus schreibersii) within the cave revealed infection with this novel filovirus (Fig. 1) (Negredo et al. 2011). Therefore, LLOV is only the third (LLOV, MARV and RAVV) of the eight known filoviruses to unambiguously infect bats. However, analysis of available data does not permit drawing conclusions on whether LLOV caused the bats’ deaths. Similar to MARV- and RAVV-infected bats, LLOV-infected bats could have been subclinically and persistently infected with LLOV and died of other causes. As LLOV has not yet been rediscovered, Schreiber's long-fingered bats may not be the natural host reservoir of LLOV.

A single coding-complete genomic sequence of LLOV, Lloviu virus/M.schreibersii-wt/ESP/2003/Asturias-Bat86 (LLOV/Ast-Bat86; Table 4) was assembled from one of the Spanish samples (Negredo et al. 2011) and was designated the type LLOV ‘isolate’ (Kuhn et al. 2014). The LLOV genome is highly reminiscent in organization of other filovirus genomes containing the same overall linear ORF arrangement (Fig. 2). However, LLOV appears to express seven structural proteins (nucleoprotein [NP], VP35, VP40, GP, VP30, VP24, L) from six genes rather than the filovirus-typical seven genes (Negredo et al. 2011). All attempts failed to isolate this virus in culture (Negredo et al. 2011), and all sample material was depleted. Because the 5′ genomic terminus of the virus could not be sequenced, it is unclear whether a functional LLOV genome could be synthesized for in vitro rescue. Consequently, only four studies are published targeting LLOV specifically. These studies relied on recombinantly expressed LLOV proteins (Maruyama et al. 2014; Feagins and Basler 2015) or on virus surrogate systems such as vesiculoviral pseudotypes (Maruyama et al. 2014), retroviral pseudotypes (Ng et al. 2014) or recombinant vesiculoviruses (Ng et al. 2014, 2015) to study parts of the presumed LLOV replication cycle. Results indicate that the LLOV surface GP_1,2 mediates LLOV cell entry by binding to the universal endosomal filovirus receptor NPC1 (Ng et al. 2014, 2015) in a pH-, cathepsin L-dependent (not cathepsin B-dependent) manner reminiscent of EBOV (Maruyama et al. 2014; Ng et al. 2014). As shown for other filoviruses, co-expression of LLOV GP_1,2 and matrix protein VP40 results in the formation of filamentous filovirion-like particles (Maruyama et al. 2014). In terms of interferon response inhibitory functions, LLOV proteins VP35, VP40 and VP24 are functional analogs of EBOV rather than MARV (Feagins and Basler 2015). Finally, LLOV Δ-peptide inhibits cell transduction of retroviral particles pseudotyped with MARV GP_1,2 just like EBOV and MARV Δ-peptides (Ng et al. 2014).

The potential of LLOV to infect humans is unknown. Because of the absence of a replicating LLOV isolate, no animal model of LLOV infection is available. Consequently, possible persistence of LLOV in animals, pathogenicity/virulence, pathogenesis or potential countermeasures against infection are not known. Specific diagnostic tests for LLOV infection have not yet been reported. However, a recently established, novel system using recombinant EBOV expressing LLOV GP_1,2 instead of EBOV GP_1,2 aided in the identification of LLOV GP_1,2-specific antibodies that could possibly be used in diagnostic assays such as immunofluorescent assay (IFA), enzyme-linked immunosorbent assay (ELISA) or western blot, assuming that LLOV GP_1,2 is presented correctly on the recombinant EBOV/LLOV chimeric virion (Ilinykh et al. 2016).

Ravn virus

Ravn virus (RAVV; pronounced rævn vɑɪrəs) was discovered in 1987 and has been reencountered in 1999, 2007 and 2009 (Fig. 1). Thus far, only three human infections, two of them lethal, have been recorded. These cases result in a mean CFR of 66.7% (higher than that of any ebolavirus infection), but the low case number results in extremely high confidence intervals (Table 3). Of note, several human RAVV infections may have been overlooked during the relatively extensive MVD outbreak in the Democratic Republic of Congo in 1998–2000 (Bausch et al. 2006).

In addition to the scientific literature (Tukei 1988; Johnson et al. 1996), the initial 1987 infection of a 15-year-old Danish boy with RAVV has been vividly described in a popular science book using the synonym ‘Peter Cardinal’ (Preston 1994). The boy had lived with his expatriate sister and parents in Kisumu District of the now dissolved Nyanza Province in Kenya. The boy fell sick on 10 August 1987, with symptoms and clinical signs indistinguishable from MVD. Despite extensive treatment and life-sustaining measures started at Aga Khan Hospital in Mombasa on 13 August and continued at Nairobi Hospital on 18 August, the boy died on 20 August (Tukei 1988; Preston 1994; Johnson et al. 1996). Where and under which circumstances the boy became infected with RAVV are unclear. Epizootiological studies focused primarily on Kitum Cave in Kenya's Mt. Elgon National Park, which the family had visited on 2 August, because this cave was also loosely associated with a fatal MARV infection of a French engineer in 1980 (Smith et al. 1982; Preston 1994). Being large enough to give shelter to animals including bats, large felids and elephants, Kitum Cave was thought to be a possible hotspot for zoonotic viral spillover. However, even after examination of several thousand collected samples and sentinel animals from the cave, no filovirus was found (Tukei 1988; Preston 1994).

The second case of human RAVV infection was recorded in 1999 in Durba, Haut-Uele District, Province Orientale, Democratic Republic of the Congo, during a much larger outbreak of MVD due to MARV infection (1998–2000) including 153 infections and 128 deaths. All these human MARV infections and the single RAVV infection were traced back to an illegal underground gold mine. The original source of the RAVV and MARV infections is unclear. However, the outbreak came to an abrupt end when the gold mine was flooded, indicating a marburgvirus (MARV and RAVV) host reservoir inside of the cave and, therefore, zoonotic transmission (Bausch et al. 2006).

Between June and July 2007, three people developed MVD due to MARV infection in Ibanda District, Western Region, Uganda, after entering and working in the local Kitaka Cave (Adjemian et al. 2011). This lead ore mine was closed and secured by a guard who developed MVD in mid-September due to RAVV infection (Towner et al. 2009; Adjemian et al. 2011).

A total of three human RAVV isolates have been sequenced, one from each outbreak (Table 4). The coding-complete sequence of the isolate obtained from the 1987 case, Ravn virus/H.sapiens-tc/KEN/1987/Kitum Cave-810040 (RAVV/KiC-810040), was designated the type RAVV isolate (Kuhn et al. 2014). RAVV/KiC-810040 is also the isolate that has been used for virtually all laboratory studies with RAVV. The RAVV genome is identical in organization to the MARV genome (Fig. 2) (Johnson et al. 1996). Molecular-biological characterization studies focusing on RAVV have been extremely rare. In one study, RAVV VP40 was found to be a functional analog to MARV VP40, inhibiting the host cell's interferon type I and II responses (Valmas and Basler 2011).

Among the neglected filoviruses, RAVV's ecology is the least shrouded in mystery. RAVV (and MARV) genome-related nucleic acids fragments were detected in Egyptian rousettes (R. aegyptiacus) sampled in Kitaka Cave in 2007–2008 and 2012. Several replicating isolates of RAVV (and MARV) could be isolated from Kitaka Cave Egyptian rousettes sampled in 2007–2008 and from Egyptian rousettes sampled from 2008 and 2009 in Python Cave, a tourist site in Uganda's Queen Elizabeth National Park less than 30 miles away from Kitaka Cave (Table 4) (Towner et al. 2009; Amman et al. 2012, 2014). These findings indicate that Egyptian rousettes are reservoir hosts for marburgviruses. The circumstances under which extremely rare marburgvirus transmission occurs to humans has not been defined, and explanations are not available for the lack of marburgviruses in Egyptian rousette populations elsewhere in Africa (Wahl-Jensen et al. 2013).

The clinical presentation of human RAVV infection has only been described for the Danish boy who died in 1987 (Preston 1994; Johnson et al. 1996). As the source of infection remains unknown, so is the incubation time. The boy first developed headache and fever, accompanied by malaise, anorexia and vomiting. The disease then worsened, resulting in bloody diarrhea and hypotension, ecchymoses, leukocytosis (20 000/mm³) and thrombocytopenia (26 000/mm³). Deterioration was characterized by high fever (40°C), disseminated Pseudomonas aeruginosa superinfection, continuously falling blood pressure reaching unrecordable values, delirium, cyanosis, blood coagulation abnormalities resembling DIC, and increases in serum potassium and urea levels. The boy then died because of cardiac arrest due to shock despite all efforts, including administration of antibiotics, steroids, heparin, fresh plasma and blood, and dialysis. Autopsy revealed massive petechial and purpuric hemorrhages in the skin, conjunctivae and gastrointestinal mucosa; hemorrhages in the lungs, tracheobronchal tree, epicardium, renal cortices and bladder (but not in spleen, pancreas, adrenals or in the acutely congested liver); and retroperitoneal edema; and pleural, pericardial and peritoneal effusions (Preston 1994; Johnson et al. 1996).

RAVV is the best researched neglected filoviruses in regard to medical countermeasures. Immunocompetent and immunodeficient mouse models and guinea pig models of RAVV infection have been established for high-throughput medical countermeasure evaluation, and immunopathological studies and uniformly lethal crab-eating macaque and rhesus monkey models are available for pathogenesis studies and specific countermeasure evaluation (Table 6). RAVV was long seen as an outlier ‘subtype’ or ‘strain’ of MARV. Consequently, many experiments aiming at identification of medical countermeasures against MARV infection included RAVV. This inclusion explains why several promising candidate vaccines for prevention (adenovirus, DNA, virion-like particles [VLP], vesicular stomatitis Indiana virus [VSIV]-based, inactivated and subunit; Table 7) and candidate therapeutic treatments (antisense/interfering RNAs and small molecules; Table 8) of RAVV infection have been studied. However, candidate therapeutic treatments have not been evaluated in non-human primates.

Table 8.

Overview of candidate peri-exposure treatment against infections with neglected filoviruses.

Reston virus

Reston virus (RESTV; pronounced ‘rɛstən vɑɪrəs) was discovered during a highly lethal VHF epizootic that occurred almost simultaneously in Virgina, Pennsylvania and Texas in the USA in 1989–1990 among captive crab-eating macaques (Geisbert and Jahrling 1990; Jahrling et al. 1990; Miranda and Miranda 2011). The first affected US location was Hazleton Research Products’ Primate Quarantine Unit in Reston, Virginia, for which RESTV was named, and this virus became a household name through the vivid descriptions of the epizootic in Richard Preston's popular science book The Hot Zone (Preston 1994). The affected macaques of all US locations, several of them also co-infected with a yet-to-be-identified simian arterivirus (suspected to be simian hemorrhagic fever virus), had been imported from a Ferlite Scientific Research non-human primate export facility in Calamba, Luzon and Philippines (Fig. 1). Follow-up studies indeed found RESTV circulating in that facility (Hayes et al. 1992; Miranda et al. 1999), suggesting that RESTV may be an Asian filovirus or that the virus had been imported to the Philippines from another location. Since then, RESTV epizootics were recorded three more times. In 1992 and 1996, RESTV once again killed numerous crab-eating macaques in Hazleton non-human primate facilities in Siena, Italy and Alice, Texas, respectively, after being imported from the same Philippine facility in Calamba that was implicated in the 1989 epizootic (Centers for Disease Control and Prevention 1996; Ciorba et al. 1997; Miranda et al. 1999, 2002; Rollin et al. 1999; Miranda and Miranda 2011). Depopulation of the facility terminated RESTV circulation (Miranda et al. 2000). From 2008 to 2009, RESTV was repeatedly isolated in the Philippines from several of hundreds of captive domestic pigs (Sus scrofa) dying of a respiratory and abortion disease and co-infected with a porcine arterivirus (porcine reproductive and respiratory disease syndrome virus) and/or circoviruses (Barrette et al. 2009). Whether RESTV caused or contributed to the observed clinical signs in the affected pigs remains unclear.

Only a few RESTV isolates have been obtained (Table 4). Reston virus/M.fascicularis-tc/USA/1989/Philippines89-Pennsylvania (RESTV/Phi89-Pen) was designated the type RESTV isolate (Kuhn et al. 2014).

Infection of domestic piglets with RESTV in the laboratory only resulted in viremia in the absence of disease signs, indicating that pigs or other suids may be able to maintain subclinical infections and, therefore, serve as RESTV reservoir hosts (Marsh et al. 2011). However, no evidence has been obtained thus far supporting RESTV infection in wild suids. RESTV is widely assumed to be apathogenic for humans. This assumption is based on the absence of recorded clinically overt human infections despite numerous possibilities for non-human primate-to-human or pig-to-human transmission during the various epizootics and on the detection of anti-RESTV antibodies in a few clinically healthy individuals that were exposed to RESTV (Center for Disease Control 1990a,b; Miranda et al. 1991; World Health Organization 2009; Miranda and Miranda 2011). Consequently, ecological studies focusing on RESTV have been sparse as the virus is frequently not considered as an immediate threat to humans as are most other filoviruses. Anti-RESTV antibodies were detected in pteropodid bats sampled in Africa (Ogawa et al. 2015), Bangladesh (Olival et al. 2013), China (Yuan et al. 2012) and the Philippines (Taniguchi et al. 2011; Jayme et al. 2015), and in orangutans in Indonesia (Niikura et al. 2001). Short RESTV NP gene-like fragments (519 bp) were amplified from samples taken from bats trapped in the Philippines (Jayme et al. 2015), and short RESTV L gene-like fragments could be detected by PCR in samples taken from domestic pigs in China (Pan et al. 2014). However, neither replicating RESTV isolates could be isolated from any sample nor could complete or coding-complete genomes be assembled to prove actual infection of these animals. Injection of RESTV into Egyptian rousettes did not result in virus replication (Jones et al. 2015). Consequently, Asian endemicity of RESTV remains a hypothesis at this time, and epizootics due to RESTV infection can neither be geographically nor temporally anticipated.

Clinical descriptions of RESTV infections are limited. Due to the co-infection with simian arteriviruses (non-human primate epizootics) or porcine arteriviruses and circoviruses (pig epizootic), clinical descriptions of the disease outbreaks (Dalgard et al. 1992; Geisbert et al. 1992; Hutchinson et al. 2001; Ikegami et al. 2002a) have to be viewed with extreme caution. Which virus caused which clinical sign and how the various co-infecting viruses interfered with each other remain unclear. Therefore, our understanding of the clinical presentation of RESTV infection relies on results of very few published experimental laboratory infections (Table 6). Initial experiments suggested that RESTV infection in crab-eating macaques presents clinically and pathologically similar to EBOV and SUDV infections, but progresses more slowly (death within 8–14 days after infection compared to 8 days, respectively), is less lethal (<100%), and results in lower viremia. Interestingly, whereas EBOV is known to also cause lethal infection in grivets (Chlorocebus aethiops), RESTV and SUDV do not (Fisher-Hoch et al. 1992; Jahrling et al. 1996b).

On the molecular-biological level, RESTV is arguably the best characterized of the neglected filoviruses, but very few studies have been published overall. The genome structure of RESTV is highly similar to other ebolaviruses (Ikegami et al. 2001; Groseth et al. 2002) (Fig. 2). Crystal structures have been determined for parts or the entire polypeptide chain of RESTV NP (Baker et al. 2016), VP35 (Leung et al. 2010), VP30 (Preston 1994) and VP24 (Reid et al. 2007; Zhang et al. 2012). In comparison to EBOV and MARV, RESTV was found to be less able to inhibit cellular type I IFN responses (Kash et al. 2006; Zhang et al. 2012). Just like EBOV, RESTV suppresses the type I interferon response through VP35 and VP24, and the type II interferon response through VP24 (Kash et al. 2006; Zhang et al. 2012). RESTV uses NPC1 for cell entry (Ng et al. 2015) and is dependent on cathepsin B (but not cathepsin L) for cell entry in vitro (Misasi et al. 2012). Akin to LLOV, RESTV Δ-peptide has little effect on MARV replication (Radoshitzky et al. 2011).

A RESTV minigenome system and RESTV reverse genetics have been used to study the RESTV lifecycle (Boehmann et al. 2005; Groseth et al. 2005). No reports on medical countermeasures against RESTV infection are available, most likely due to the perception that RESTV is not an imminent threat to humans. Diagnosis of RESTV infection is possible via pan-filovirus RT-PCR or qRT-PCR protocols (Lu et al. 2015) and RESTV-specific sequence capture probes for next-generation sequencing (Koehler et al. 2014). Several serological or nucleic-acid based RESTV-specific diagnostic systems have been described (Kalter et al. 1995; Ksiazek et al. 1999; Niikura et al. 2001; Ikegami et al. 2002b, 2003a,b; Ou et al. 2011). RESTV-specific monoclonal or polyclonal murine, rabbit or crab-eating macaque antibodies (Ou et al. 2011; Holtsberg et al. 2015; Keck et al. 2015; Wang et al. 2015), RESTV cross-reactive murine antibodies (Fusco et al. 2015; Wang et al. 2015; Furuyama et al. 2016), RESTV cross-reactive murine scFvs and IgNAR Vs raised against inactivated EBOV particles (Goodchild et al. 2011) are available.

Taï Forest virus

Like LLOV, Taï Forest virus (TAFV; pronounced tɑː'iː 'fɔːrɨst vɑɪrəs) is a filovirus that thus far has only been encountered once (Table 3). TAFV was discovered in 1994 through the infection of a Swiss ethologist during a necropsy she performed with two colleagues on a western chimpanzee (Pan troglodytes verus) in Taï National Park, western Côte d'Ivoire (Fig. 1) (Le Guenno et al. 1995; le Guenno, Formenty and Boesch 1999). The ape belonged to a troop under observation since 1979 whose numbers had been reduced by at least two episodes of epizootic hemorrhagic fever in November 1992 (8 deaths) and November 1994 (12 deaths). The 34-year-old ethologist, who performed the necropsy on 16 November 1994 to shed light on the etiological cause of these episodes, developed a febrile disease 8 days later. On 26 November, she was hospitalized in Abidjan, and on 1 December she was transported to a hospital in Basel, Switzerland, where she recovered. Filovirus infection was confirmed by electron-microscopic and serological methods (ELISA and IFA) and by virus isolation in tissue culture (Le Guenno et al. 1995). Molecular characterization identified TAFV as a distinct filovirus most closely related to BDBV with a genome organization similar to all ebolaviruses (Fig. 2), an ebolavirus that only was encountered in Eastern Africa ≈4200 km away (Towner et al. 2008). Taï Forest virus/H.sapiens-tc/CIV/1994/Pauléoula-CI (TAFV/Pau-CI) was designated the type (and only) isolate of TAFV (Table 4) (Kuhn et al. 2014).

Histopathological examination of tissues from the western chimpanzee necropsied by the ethologist strongly indicated that the ethologist acquired TAFV from this animal. Lesions resembled those found in macaques experimentally infected with EBOV: multifocal necroses infiltrated with inflammatory cells, Kupffer cell hyperplasia in the liver, diffuse fibrinoid and hemorrhagic necrosis in the splenic red pulp and lymphoid depletion in all lymphatic tissues. More importantly, macrophages in various affected tissues reacted with TAFV-specific and TAFV-cross-reactive antibodies. However, typical hemorrhagic, thrombotic or vascular lesions of EBOV infection were not present in the ape (Wyers et al. 1999), and ultimate confirmation of TAFV infection in chimpanzees by either virus isolation or next-generation sequencing is lacking. Likewise, how the chimpanzees may have become infected is unclear. Whereas MARV and RAVV subclinically infect bats in nature, and data for EBOV–bat associations are suggestive (Wahl-Jensen et al. 2013), TAFV does not replicate in Egyptian rousette bats (Jones et al. 2015). No clues have been found in regard to TAFV ecology other than possibly TAFV-specific antibodies in Indonesian orangutans (Nidom et al. 2012).

Because of the recording of only a single case, the clinical presentation of EVD due to TAFV infection is unclear. The TAFV-infected ethologist developed fever, headaches, myalgia, chills, cough, abdominal pain and nausea accompanied with acute non-bloody diarrhea and vomiting, a generalized maculopapular rash and hematuria. Clinical tests revealed proteinuria, marked liver enzymes (aspartate aminotransferase, alanine aminotransferase) and lactate dehydrogenase elevations, thrombocytopenia, lymphopenia and neutrophilia, and suggested DIC. The woman was released on day 15 after disease onset and recovered without sequelae other than temporary hair loss (Le Guenno et al. 1995; le Guenno, Formenty and Boesch 1999; Formenty et al. 1999). Thus, the disease resembled non-lethal EVD caused by BDBV (Table 5) and other ebolaviruses and MVD. The CFR of TAFV infection cannot be determined from a single survivor. The molecular and immunological responses to human TAFV infection have not been studied in detail. Consequently, neither cytokine response nor other biomarker data are available. Since the patient fortunately survived and no biopsies had been performed, there are no human pathological data.

Establishment of experimental disease caused by TAFV is currently only possible using a partially lethal crab-eating macaque model (Table 6), thereby development of medical countermeasures is challenging. Consequently, results from only a single study on a candidate vaccine for the prevention of TAFV are ambiguous (Table 7), and no TAFV-specific antivirals have been brought forward. Molecular-biological studies on TAFV are close to absent with the exception of the determination of the crystal structure of the TAFV NP C-terminal domain (Baker et al. 2016) and the characterization of TAFV GP_1,2-mediated cell-entry (NPC1 and cathepsin B) requirements (Misasi et al. 2012; Ng et al. 2015). Minigenomes and reverse genetics systems have yet to be established for TAFV. TAFV detection is possible via pan-filovirus RT-PCR (Lu et al. 2015) and TAFV-specific sequence capture probes for next-generation sequencing (Koehler et al. 2014). Very few TAFV-specific antibodies have been described (Ou et al. 2011; Furuyama et al. 2016), including ebolavirus cross-reactive llama single-domain antibodies (Sherwood and Hayhurst 2013). TAFV-specific serological diagnostic assays have not yet been described.

Other filoviruses

The geographically broad distribution of filoviruses (western to eastern Equatorial Africa, Philippines and Spain) suggests that many filoviruses remain to be discovered. Supporting this hypothesis, He et al. (2015) recently described the amplification of filovirus NP-, VP35- and L-like nucleic acids from (frugivorous) Leschenault's rousette bats (Rousettus leschenaultii) captured in Yúnnán Province, China. Whereas most fragments were too short (129–354 bp) to unambiguously assign them to the family Filoviridae, two fragments from the same bat (‘Bt-DH04‘) could be extended to 2750 and 2682-bp length, respectively. The first fragment aligned with the 3′ end of a filoviral NP and almost an entire filoviral VP35-like gene; the second covered a large portion of a filoviral L-like gene. Phylogenetically, these fragments represent a novel clade of filoviruses basal to ebolaviruses and in between ebolaviruses and marburgviruses/cuevaviruses.

CONCLUSIONS

The ecology of filoviruses needs to be defined, and the natural host reservoirs of cuevaviruses and ebolaviruses are still unclear (Wahl-Jensen et al. 2013). The discovery of Egyptian rousettes as natural host reservoirs of marburgviruses in Ugandan caves (Towner et al. 2009) was a major step forward in filovirology, but questions remain as to why MARV and RAVV cannot be found in other Egyptian rousette populations in and outside of Africa. Because of these uncertainties, how and under which circumstances filovirus host-human transmission occurs are unclear. The geographic distribution of filoviruses (i.e. their endemicity) remains undefined despite several ecological filovirus niche modeling studies (Peterson, Bauer and Mills 2004; Peterson et al. 2006; Pigott et al. 2014, 2015). Finally, the discovery of LLOV in Spain (Negredo et al. 2011) and the detection of filovirus-like entities in Chinese bat populations (He et al. 2015) indicate that the family Filoviridae is undersampled. Its members are probably much more diverse and distributed than previously thought.

Because of this lack of ecological knowledge, prediction of when and where human and/or animal filovirus disease outbreaks may occur is impossible. Since knowledge on neglected filoviruses (BDBV, LLOV, RAVV, RESTV and TAFV) is extremely limited, one cannot exclude the possibility of future large EVD or MVD outbreaks caused by these viruses. The recent EVD/EBOV outbreak in Western Africa, involving 28 646 cases and 11 323 deaths, demonstrates for the first time that EVD outbreaks do not necessarily remain geographically confined or involve only dozens to several hundreds of cases (Table 3). Yet, a disease outbreak of the magnitude of that EVD outbreak caused by a neglected filovirus may prove even more disastrous to Africa or the world. Among the already very low global number of research institutes that are permitted to perform biosafety level 4 research on filoviruses, few even have access to neglected filoviruses. Reagents and assays for neglected filoviruses are not available or extremely limited, especially for commercial products, which primarily fulfill the need for research on EBOV, SUDV and MARV research. The only commerical reagents for neglected filoviruses include: irradiated BDBV, RAVV, RESTV and TAFV or isolated genomic RNA; recombinant complete, partial or tagged GP_1,2 (BDBV, RESTV, TAFV), VP40 (TAFV), VP24 (RESTV); polyclonal rabbit anti-GP_1,2 antibodies (BDBV, RESTV, TAFV); and monoclonal antibodies against VP40 (BDBV); commercial ELISA systems for the detection of circulating guinea pig, human, non-human primate and pig antibodies (RESTV) and circulating GP_1,2 antigen (BDBV, RESTV, TAFV); and qPCR systems for RESTV and TAFV (all based on catalog searches of Alpha Diagnostics, BEI Resources, Integrated BioTherapeutics, Genesig and Sino Biological; Anthony and Bradfute 2015). Animal models for neglected filovirus infections are either absent, not 100% lethal, or not established in non-human primates (Table 6). Consequently, there are few specific candidate vaccines and almost no specific therapeutics in the pipeline to prevent or treat neglected filovirus infections (Tables 7 and 8).

Therefore, we appeal to the international filovirus research community, and even more so to the funders of filovirus (currently almost exclusively EBOV) research and development activities, to create and maintain a global, coordinated and highly collaborative program to prospectively create basic reagents, assays, methodologies, databases, animal models and medical countermeasure platforms that include neglected filoviruses on a routine basis.

FUNDING

This work was supported in part through Battelle Memorial Institute's prime contract with the US National Institute of Allergy and Infectious Diseases (NIAID) under Contract No. HHSN272200700016I. LB, JCJ and JW performed this work as employees of Battelle Memorial Institute. A subcontractor to Battelle Memorial Institute who performed this work is: JHK, an employee of Tunnell Government Services, Inc. RB is supported by the Hartmut Hoffmann-Berling International Graduate School of Molecular and Cellular Biology (HBIGS). The content of this publication does not necessarily reflect the views or policies of the US Department of Health and Human Services, or the institutions and companies affiliated with the authors.

Conflict of interest. None declared.