The Arenaviridae Family: Knowledge Gaps, Animal Models, Countermeasures, and Prototype Pathogens

Abstract

Lassa virus (LASV), Junin virus (JUNV), and several other members of the Arenaviridae family are capable of zoonotic transfer to humans and induction of severe viral hemorrhagic fevers. Despite the importance of arenaviruses as potential pandemic pathogens, numerous gaps exist in scientific knowledge pertaining to this diverse family, including gaps in understanding replication, immunosuppression, receptor usage, and elicitation of neutralizing antibody responses, that in turn complicates development of medical countermeasures. A further challenge to the development of medical countermeasures for arenaviruses is the requirement for use of animal models at high levels of biocontainment, where each model has distinct advantages and limitations depending on, availability of space, animals species-specific reagents, and most importantly the ability of the model to faithfully recapitulate human disease. Designation of LASV and JUNV as prototype pathogens can facilitate progress in addressing the public health challenges posed by members of this important virus family.

Several members of the Arenaviridae family (order, Bunyavirales), including Lassa virus (LASV), Lujo virus (LUJV), Junin virus (JUNV), Machupo virus (MACV), Guanarito virus (GTOV), and Chapare virus (CHAPV), are capable of infecting humans [1, 2]. Infection by these viruses can lead to severe viral hemorrhagic fevers (VHFs) in a subset of individuals. Each are classified as biosafety level 4 (BSL-4) agents; however, JUNV may be used at BSL-3 if investigators are vaccinated. LASV is endemic in Western Africa and, after dengue virus, is the most common hemorrhagic fever virus imported into Europe and the Americas. Numerous other arenaviruses are endemic to the United States, although among them only Whitewater Arroyo virus has been isolated from cases of fatal VHF [3]. Here, we review the current state of knowledge about arenaviruses to identify gaps that hinder development of countermeasures.

TAXONOMY

The Arenaviridae family includes viruses that infect rodents, bats, shrews, snakes, and fish as well as humans. The family is comprised of 4 genera, each recognized as distinct, well-supported groups in phylogenies estimated based on the RNA-dependent RNA polymerase (L) protein (Figure 1 and Table 1). Old world (OW) and new world (NW) arenaviruses of mammals are placed taxonomically in the genus Mammarenavirus. These complexes were initially defined based on both serologic classification and geographic origin, but they are also clearly supported at the phylogenetic level [4]. OW mammarenaviruses are found in African rodents from the family Muridae [5–7], and in Asian rodents from various Rattus and Mus species as well as shrews (Suncus murinus) [8–10]. OW arenaviruses do not form clear phylogenetic clades. However, 4 established LASV lineages are broadly distributed across West Africa, with 3 additional proposed lineages recently emerged [11]. Lymphocytic choriomeningitis virus (LCMV), an OW mammarenavirus that is frequently used to model antiviral immune responses, is distributed worldwide in common house mice, Mus domesticus and Mus musculus [1].

Figure 1.

Phylogenetic relationships among species from the Arenaviridae family. Maximum-likelihood tree for L protein sequences representative of the genetic diversity within the Arenaviridae family. The phylogeny was inferred with the IQ-TREE software, using the LG + G4 model. Branch support was assessed using the ultrafast bootstrap approximation with 1000 replicates. Nodes with branch support >95% are indicated with circles. Tree leaves are according to the main groups within the family, namely the genera Hartmanivirus, Antennavirus, and Reptarenavirus, as well the new world complex and old world complex of the Mammarenavirus genus. These main groups are also shown by rectangles on the right part of the plot. For the new world complex, the main clades are indicated by their corresponding letters (A to D). For the Lassa virus species, lineages are indicated as part of the sequence name (*indicates that the lineage assignment is tentative).

Table 1.

Characteristics of Selected Arenaviruses

Taxonomic Position	Virus Common Name	Reservoir	Distribution	Disease	Vaccines	Therapeutics
Old world mammarenaviruses	Lassa virus	Multimammate mouse (Mastomys natalensis) and others	West Africa	In humans: Lassa fever	Multiple vaccine candidates in development	Small molecule drugs and a monoclonal antibody combination in development
	Lujo virus	Unknown		In humans: Lujo HF	None	None
	Lymphocytic choriomeningitis virus	House mice (Mus musculus and Mus domesticus)	Worldwide	In humans: influenza-like illness; aseptic meningitis; meningoencephalitis; congenital microencephaly	None	None
New world mammarenaviruses	Junin virus	House (Mus musculus), Vesper (Calomys callosus) and grass mice (Akodon azarae)	Argentina	In humans: Argentine HF	Candid-1 live attenuated vaccine	Passive survivor plasma; monoclonal antibodies in development
	Machupo virus	Vesper mouse	Bolivia	In humans: Bolivian HF	None	None
	Guanarito virus	Alston's cotton rat (Sigmodon alstoni)	Venezuela	In humans: Venezuelan HF	None	None
	Sabiá virus	Unknown	Brazil	In humans: São Paulo HF	None	None
	Chapare virus	Pygmy rice rats (Oligoryzomys microtis)	Bolivia	In humans: Chapare HF	None	None
	Whitewater Arroyo virus	Wood rats (Neotoma spp.)	Southwest United States	In humans: California HF	None	None
	Wōnlnzhōnlu virus	Rats (Rattus spp.), mice (Mus spp.), and shrews (Suncus spp.)	China, Cambodia, Thailand	In humans: fever and respiratory symptoms	None	None
Hartmaniviruses	Haartman Institute snake virus 1	Captive boas (Booidea) and pythons (Pythonidae)	Worldwide	None observed	None	None
Reptarenaviruses	California Academy of Sciences virus	Captive boas and pythons	Worldwide	In snakes: Boid inclusion body disease	None	None
Antennaviruses	Wēnlng frogfish arenaviruses	Frogfish (Antennarius striatus)	China (likely more widespread)	Unknown	None	None
	Salmon pescarenavirus	Wild and farmed Chinook (Oncorhynchus tshawytscha) and sockeye (Oncorhynchus nerka) salmon	Northeast Pacific Ocean	In salmon: spleen and liver inflammation; tubule necrosis and hyperplasia in the kidney	None	None

Abbreviation: HF, hemorrhagic fever.

NW mammarenaviruses infect rodents of the family Cricetidae [3, 12], with the exception of Tacaribe virus that infects artibeus bats [13] (Figure 1 and Table 1). Species from the NW complex form 4 well-defined clades [14]. Clades A–C are distributed across Latin America and the Caribbean, while all current representatives of clade D have been found in North America. NW mammarenaviruses in North America are maintained in reservoir rodents from the Cricetidae subfamily Neotominae, whereas Latin American and Caribbean mammarenaviruses are maintained in rodents from the subfamily Sigmodontinae [15, 16]. Arenaviruses of snakes [1, 17, 18] are classified in 2 genera, Reptarenavirus and Hartmanivirus. Arenaviruses first detected in frogfish (Antennarius striatus) [19], and later in salmon [20], represent a new genus, Antennavirus [21].

Knowledge Gap

1. The extent of arenavirus diversity is unknown. The Arenaviridae family likely extends to other yet uncharacterized vertebrate hosts, including additional species of mammals, reptiles, and fish. Because of the potential for spillover to humans it would be prudent to further characterize NW mammarenavirus diversity, which is much less well understood than for OW viruses. Screening of rodents, shrews, bats, and other potential arenavirus reservoirs by metagenomic sequencing should be utilized to elucidate the as yet undersampled diversity of NW mammarenaviruses.

MOLECULAR VIROLOGY

Arenaviruses have pleomorphic enveloped virions ranging in size from 80 to 150 nm (Figure 2). Electron-dense granules, considered to be host-derived ribosomes, give virions their “sandy” granular appearance in electron micrographs that led to the family name (Latin arenosus: sandy) (Figure 2A). Genetic differences are readily apparent when comparing mammalian, reptilian, and fish arenaviruses (Figure 2B). Mammarenaviruses, reptarenaviruses, and hartmaniviruses have bisegmented genomes [22]. The small segment is ambisense and encodes the glycoprotein complex (GPC) and the nucleocapsid protein (NP). The large segment encodes the RNA-dependent RNA polymerase (L) and, in mammarenaviruses and reptarenaviruses in which the large segment is ambisense, a zinc-binding or matrix protein (Z). Hartmaniviruses and antennaviruses do not encode a Z protein [23]. Antennaviruses contain 3 genome segments [19], the largest of which encodes L; GPC and another small protein of unknown function are encoded by another segment with the NP homolog encoded separately by the third genome segment.

Figure 2.

Arenavirus proteins and genomes. A, The arenavirus virion is enveloped and contains 2 or 3 RNA subunits, plus cellular ribosomes. B, Genome structures of the 4 genuses in the Arenaviridae family. X is a protein of unknown function encoded by antennaviruses. Plus symbol indicates that the gene is positive polarity. Negative symbol indicates the gene is negative polarity. Lines indicate proteins with significant sequence similarity.

GPC of mammarenaviruses, hartmaniviruses, and antennaviruses include an unusual long stable signal peptide (SSP) that is retained in the virion as part of the GPC. The receptor-binding glycoprotein 1 subunit (GP1) and a transmembrane fusion subunit GP2 are arranged on the virion surface as a trimer of dimers [24]. The glycoprotein trimer bears a thick glycan shield [25], and recognition by neutralizing antibodies is rare [24, 26]. Reptarenavirus glycoproteins are distinct from other arenavirus glycoproteins, but are related to glycoproteins of filoviruses and retroviruses [27]. Reptarenaviruses do not encode a SSP.

Due to a transcriptionally regulated requirement for a threshold amount of NP, this protein is the most abundantly produced protein during arenaviral infections. NP is divided into N- and C-terminal domains (Figure 3). The N-terminal half of NP adopts a clam-shell shape in which N- and C-terminal subdomains encase 8 bases of single-stranded RNA (ssRNA) [28] (Figure 3A). The C-terminal domain functions as a 3′–5′ exoribonuclease that specifically and rapidly digests double-stranded RNA (dsRNA) [29–31], a key inducer of the interferon system and innate immunity (Figure 3B). The exoribonuclease appears to play a key role in LASV-induced immunosuppression. In the absence of RNA, NP forms a trimer with N- and C-terminal domains oriented in a head-to-tail fashion in an RNA-free trimeric ring of NP monomers [32]. This oligomeric arrangement of NP is incompatible with RNA binding. This assembly could represent a conformation-driven chaperoning strategy of the unbound NP. Indeed, binding of RNA requires the opening of an α-helical gate (Figure 3A) to expose the ssRNA pocket [28]. Relocation of this gate would prevent trimerization.

Figure 3.

Structure and RNA binding of Lassa virus nucleoprotein (LASV NP). A, Cartoon representation of the N-terminal domain of LASV NP showing the deep groove through which the single-stranded RNA (ssRNA) channels. Conformational changes open and close the RNA-binding pocket. α-helix6 (α6) is positioned on top of the pocket, preventing RNA from binding. When α6 shifts away from the pocket as shown, the RNA gate opens to accommodate ssRNA. B, Cartoon representation of LASV NPΔ340, the C-terminal, immunosuppressive domain of LASV NP. Double-strand RNA (dsRNA), a potent interferon inducer, is degraded by the exonuclease activity of NP. Residues with similarity to the DEDD superfamily of exonucleases show coordination of a magnesium (Mg) ion. At the entrance to the active site of the exonuclease are several positively charged amino acids. Coordination of a zinc (Zn) ion is indicated.

The matrix protein Z of mammarenaviruses occurs in monomeric and oligomeric forms [33]. In mammarenaviruses, the glycine at position 2 is myristoylated, which is important for association with the plasma membrane, and the really interesting new gene (RING) domain of Z contains 2 zinc-binding domains [33, 34]. In reptarenaviruses, Z also has a RING domain, but unlike mammarenaviruses, the reptarenavirus Z contains a transmembrane domain [35]. For the mammarenaviruses, several recent studies have described the binding of monomeric Z to L where the RNA product exits [36], and locking of the different domains of L together [37], suggesting a role of Z in regulating the polymerase [36–39].

Knowledge Gaps

1. Z is known to negatively regulate L, but the concentration dependence is not sufficiently understood. The conformational and functional switch between the function of Z in control of L in the cytosol, and function of Z as the viral matrix protein at the plasma membrane also remain unclear. The dodecameric ring form of Z has no known function. Furthermore, the role of the Z myristol group in the soluble Z–L is unknown including whether Z traffics to the membrane separately from L, the role of the oligomer, and what conformational changes permit and then destabilize the Z–L interaction allowing L to begin RNA synthesis. Further elucidation of the roles of Z and L in regulating viral RNA synthesis could potentially facilitate development of arenavirus small molecule drugs targeting this process.

2. Mechanisms of immunosuppression by Z and other arenaviruses proteins are incompletely described. Filling this gap may allow development of medical countermeasures that block this process.

3. Assembly mechanisms of hartmaniviruses and antennaviruses that do not encode a Z protein are unknown.

ENTRY AND REPLICATION

Cell entry of LASV and other arenaviruses is a multistep procedure. Several host membrane proteins can serve as cell surface receptors/factors for the prefusion (virion) form of GPC. The major and most efficient cell surface receptor for LASV entry is matriglycan on α-dystroglycan (α-DG) [40–45]. Clade C NW mammarenavirses also utilize α-DG for entry. In contrast, the major cell surface receptor of clade B NW mammarenaviruses is the transferrin receptor. No receptors have yet been identified for clade A and D mammarenaviruses.

After binding to surface receptors, LASV virions are internalized via endocytosis. Prefusion GPC responds to the acidic endosomal pH inducing a conformational shift in prefusion GPC termed “GPC priming,” and dissociation from α-DG. Primed GPC binds to the endosomal receptor, lysosomal-associated membrane protein 1 (LAMP-1) [46, 47]. Following LAMP-1 binding, GPC undergoes additional conformational changes that mediate virus-endosomal membrane fusion and LASV genome release into the cytosol. LAMP-1 is not absolutely required for GPC-mediated membrane fusion in very late endosomes, but if it is not present, entry is inefficient [48, 49]. Like LASV, LCMV and its close relatives were recently found to switch receptors from α-DG to CD164 in the endosome [50]. Similarly, a genetic screen implicates the endosomal resident CD63 in LUJV entry [51].

After the viral RNA is unpacked, RNA replication and transcription take place in the cytoplasm. In the case of the bisegmented arenaviruses, primary transcription of the S and L genomic RNAs produces mRNAs that encode NP and L, respectively. S and L RNA genomes serve as templates for antigenomic RNAs, which are the templates for GPC and M mRNAs as well as for the full-length progeny genomes. A stem-loop structure within the intergenomic region terminates transcription from genomes and antigenomes. Z functions as a negative regulator of L. Arenaviruses use a cap-snatching strategy involving the endonuclease activity of L, which is required for effective translation from cellular mRNAs [52]. The GPC precursor is tracked from the endoplasmic reticulum to the Golgi, where it is heavily N-glycosidated and processed by cellular proteases (SPase, SKI1/SP1) into its mature form comprising noncovalently linked GP1, GP2, and SSP. NP encases the viral genome via its N-terminal domain. Cleaved glycoproteins are incorporated into the virion envelope when N, M, and L and the genomic RNA are assembled. The virus then buds and is released at the cell surface membrane.

Knowledge Gaps

1. While the molecular structures of proteins of LASV and certain NW arenaviruses, including MACV and JUNV, have been determined, there is limited structural information for proteins of other arenaviruses, particularly those outside the Mammarenavirus genus. Structural information for GPC coupled with identification of key epitopes can inform development of vaccines and immunotherapeutics. Structural information for other arenaviruses proteins, such as L, could facilitate development of small-molecule drugs. Further clarity on functional significance of each protein during replication and in the context of infection are also needed.

2. Z appears sufficient for virus budding in those arenaviruses that encode a Z. The mechanisms by which Z mediates virion formation are unknown. How arenaviruses lacking Z assembe virion particles and egress from the cell is unclear.

3. The exonuclease domain of NP rapidly mediates dsRNA digestion, and this plays a role in blocking innate immune sensing in cell culture. Furthermore, exonuclease knockout mutations are still impaired in interferon-knockout cells, suggesting that the exonuclease function may play an additional fundamental role in viral replication and propagation or in proofreading. Improved understanding of the role of exonuclease domain of NP in vivo could potentially be exploited for developing medical countermeasures (MCM).

4. The function of the RNA-free NP trimer is unknown. It may be an artifact of recombinant overexpression. Another possibility is that it represents an “NP0” in which trimerization of NP performs the equivalent function of the P protein of other negative-strand viruses, which arenaviruses do not explicitly encode. Identification of the specific role of the trimer may guide design of inhibitors targeting arenavirus replication.

5. There is an incomplete understanding of entry for NW mammarenaviruses. It is unknown whether the receptor switch utilized by LASV, LCMV, and LCMV-like viruses is common to other OW mammarenaviruses.

EPIDEMIOLOGY

Rural areas of the West African nations of Nigeria, Sierra Leone, Liberia, and Guinea are endemic for Lassa fever [53]. Mounting evidence points to expansion of the Lassa fever zone to include Burkina Faso, Mali, Togo, Benin, Senegal, and Côte d’Ivoire. The true incidence of Lassa fever has not been determined, in part because of the lack of available diagnostic assays. Most LASV-infected persons in West Africa are likely never seen by health professionals due to limited access to medical care. The initial symptoms of Lassa fever are like other febrile illnesses, leading to misdiagnoses of other common endemic illnesses such as malaria. Extrapolations from prior limited serological studies suggested that there may be as many as 500 000 LASV infections and 5000 deaths per year in West Africa [54, 55]. The development of recombinant antigen enzyme-linked immunosorbent assay (ELISA) has recently enabled large-scale serological surveys that are underway at the time of this writing in several West African countries. Ongoing seroprevalence studies of LASV in West Africa are essential for conducting human clinical trials of MCM such as vaccines or therapeutics.

Argentine hemorrhagic fever was first described in 1955 [56]. It manifests as a number of catastrophic vascular, renal, and neurological symptoms and has a mortality rate of 15%–30% in untreated individuals. The etiologic agent of Argentine hemorrhagic fever is an arenavirus, JUNV. Before the development of a vaccine in 1992, there were hundreds of cases of Argentine hemorrhagic fever each year [57].

Knowledge Gaps

1. Arenavirus reverse genetics systems exist, but need development and support. These may be critical in times of an outbreak of a novel arenavirus to quickly begin working safely and expeditiously on MCM, such as diagnostics, vaccine, and drug development, as soon as genetic information is available.

2. Arenaviruses tend to be geographically restricted, reflecting the range of their reservoir hosts. JUNV, MACV, GTOV, Sabia virus (SABV), and CHAPV, cause Argentine, Bolivian, Venezuelan, Brazilian, and Chapare hemorrhagic fevers, respectively. The prevalence of Argentine hemorrhagic fever has been described; however, the incidence and prevalence of other South American arenaviruses is largely unknown.

3. Importation of samples for characterization of arenaviruses under high containment has challenges, including export/import hurdles and regulations involving select agents. Fortification of clinical and basic research infrastructure, particularly in areas known to be endemic for arenaviruses, is needed to circumvent these challenges. Another measure can be establishment and support of symmetrical international research collaborations with prenegotiation of import and export licenses.

4. It is unclear if all isolates of OW and NW arenaviruses, including those that have been demonstrated to cause VHFs, cause serious disease. Nonlethal strains of mammarenaviruses may exist that could inform studies of disease pathogenesis that could be modelled in experimental animals.

ECOLOGY

With no approved vaccines or therapeutics available, Lassa fever is among the most consequential rodent-borne virus infection in terms of morbidity, mortality, and societal costs. Individuals living in endemic hotspots of Lassa fever have recurrent exposure to LASV via spillover from the primary host reservoir Mastomys natalensis especially in or near human habitations. Additional rodents, including Rattus sp., the African wood mouse Hylomyscus pamfi, and the Guinean multimammate mouse Mastomys erythroleucus, also appear to be involved in LASV transmission to humans. Individuals contract LASV through inhalation of aerosolized urine in droplets or dust particles, as well as via ingestion of food or water or contact with fomites contaminated with rodent urine or droppings. Knowledge regarding rodent habitats, reproduction and fecundity, movement patterns, and spatial preferences are essential for implementing effective preventative measures against Lassa fever. The efficacy of rodent abatement practices, including the efficacy of feline and chemical rodenticides, is of equivalent importance to drive stringent rodent control.

Like LASV, the detected range of JUNV incidence is not in perfect alignment within the geographic range of its primary host, Calomys musculinus [58, 59]. The first outbreaks of Argentine hemorrhagic fever were primarily in the humid pampas of central Argentina, 200–300 km west of Buenos Aires. Since then, the endemic area has expanded northwards and westwards from its historical foci to its present extent of approximately 150 000 km² [59, 60]. JUNV is also present in animals outside the historic Argentine hemorrhagic fever endemic zone [57]; however, the geographic range of JUNV has not expanded to the full range in which the host species are found. Since the discovery of JUNV, several other South American mammarenaviruses causing severe hemorrhagic fevers, including MACV (Bolivia), SABV (Brazil), GTOV (Venezuela), and most recently CHAPV (Bolivia), were discovered associated with sporadic outbreaks of severe human disease [61]. The incidence of exposure to the human population by these viruses is largely unknown as only severe cases are the most likely to be reported.

Knowledge Gaps

1. Despite M. natalensis being broadly distributed across sub-Saharan Africa, Lassa fever is only found in West Africa. Factors that contribute to the geographical distribution of LASV are poorly described. The features of M. natalensis populations that carry LASV and those that do not should be determined.

2. Known arenaviruses, including LASV, may have a broader host range than currently documented. Other rodents could be involved as transient or intermediate vectors. Likewise, the ecology and geographic distribution of NW mammarennaviruses has not been well elucidated. A better understanding of arenavirus ecology, including several rodent species that serve as reservoirs or intermediates of LASV transmission to humans are needed, can inform public health interventions.

PATHOGENESIS

Infection with LASV can result in a range of outcomes from a mild or asymptomatic infection to a severe life-threatening illness. Signs and symptoms of Lassa fever occur 1–3 weeks after exposure and are highly variable. After 4–7 days of mild illness, an estimated 20% of infected individuals develop symptoms of a severe VHF. Sore throat, vomiting, and coughing are common, but each case presentation is distinct [54, 62, 63]. Around 40% of Lassa fever patients experience bleeding from the nose, mouth, other orifices, and mucosal surface, which confers a poor prognosis. Death from Lassa fever generally occurs between 10 and 14 days of symptom onset and is attributed to diminished effective circulating volume, shock, and multiorgan system failure [63]. Infection is approximately 90% lethal to women in the third trimester of pregnancy. Neurological problems have also been described following LASV infection, including tremors, and encephalitis. Temporary or permanent unilateral or bilateral deafness has been reported to occur in approximately 30% of patients with Lassa fever [64, 65].

Mechanisms of pathogenesis by LASV and other arenaviruses are incompletely understood, but are thought to include direct cytopathic effects by viral proteins, including increase in vascular leakage. Immunopathogenesis, including induction of a cytokine storm, also appears to be involved. Long-term sequelae, including neurological, immune, and renal defects, have been described in survivors and may involve the induction of autoimmunity. The risk of recrudescence has yet to be ruled out.

Knowledge Gaps

1. The morbidity burden of LASV in endemic regions is unclear. Case fatality rates are based on patients presenting to health care facilities which overrepresents severe LASV infections, and do not include asymptomatic or mild infections that likely predominate in the wider population.

2. Factors that determine virulence, severity of disease, and outcome of arenavirus diseases are unknown. There are likely a multitude of factors, including human and viral genetic differences. Different lineages of LASV have different case fatality rates, which could reflect these genetic factors, as well as differences in clinical management. It is possible that OW and NW mammaraenaviruses employ different pathogenic mechanisms.

3. Field-caught rodents that are carrying arenaviruses do not display obvious signs of illness. However, the pathogenesis of arenaviruses in reservoirs or intermediate species is not known. Infection studies of M. natalensis raised and bred in captivity could be informative in this regard.

4. While sometimes present in healthy snakes, reptarenaviruses are the causative agent of inclusion body disease, which spreads among boid snakes (boas and pythons) in captivity [27, 66–70]. Hartmaniviruses have not been shown to cause inclusion body disease [71].

5. Antennaviruses of salmon, which have been isolated from dead or moribund aquaculture salmon as well as live-sampled wild salmon, appear to be pathogenic. Mechanisms of pathogenesis of arenaviruses other than the mammarenaviruses have not been determined.

NATURAL IMMUNITY AND IMMUNOEVASION

Protective immunity has been correlated with cellular immune responses to LASV rather than humoral responses. Responses to both GPC and NP have been detected. Some T cell epitopes are lineage-restricted, and others are pan-lineage. There is significant cross-reactivity of cellular responses in humans across LASV lineages. Putative conserved T cell epitopes have been identified. While infected humans and experimental animals such as nonhuman primates (NHP) develop strong antibody responses to LASV, those antibodies are often directed to epitopes not associated with viral clearance such as nonneutralizing epitopes or linear epitopes exposed only in the postfusion conformation. The LASV GPC trimer bears 33 N-linked glycans that form a denser shield relative to other human viruses; only the human immunodeficiency virus (HIV-1) has a heavier glycan shield [72]. In addition, LASV infection and overexpression of GPC in immunogen production leave a large proportion of uncleaved GPC molecules that cannot mediate membrane fusion [73]. Consequently, this glycan shield together with GPC in nonprefusion structures is thought to pose a substantial barrier to elicitation of neutralizing antibodies. LASV neutralizing antibodies, if detected at all, are usually only present after recovery. Neutralizing epitopes require prefusion GP and are usually quaternary. Significant cross-reactivity of humoral responses in humans neutralizing antibody titers continue to rise even several months after convalescence has been established, which may indicate constant stimulation of B cells due to low levels of virus persistence.

Knowledge Gaps

1. While convalescent plasma is the standard for treatment of Argentine hemorrhagic fever caused by JUNV, convalescent plasma has not been shown to be an effective treatment for LASV infection. The differences in natural immunity that account for this difference have not been elucidated, but likely may relate to the types of neutralizing antibodies elicited (see below).

2. It remains to be determined whether it will be possible to design vaccines that effectively elicit neutralizing antibodies such as those that manifest during the late stages of Lassa fever convalescence.

3. Natural immunity to LUJV and to NW mammarenaviruses has not been extensively characterized.

4. It is unknown whether T cell immunity induced by natural infection with mammarenaviruses induces cross-protective immunity across lineages. NP and potentially other arenavirus proteins may induce protective T cell–mediated immunity.

5. Some arenaviruses proteins are known to have immunosuppressive properties, but the mechanisms are incompletely understood.

6. For GPC, we now have structures of prefusion GPC in complex with a few neutralizing antibodies, and we understand why neutralizing antibodies may be difficult to elicit. Additional work is needed to turn these predictions into vaccines to determine how a neutralizing antibody response could be elicited by vaccination.

ANIMAL MODELS

Animal models are required to investigate arenavirus-induced disease pathogenesis and for developing therapeutics and vaccines [74, 75]. Inbred Strain 13 guinea pigs infected with wild-type LASV (strain Josiah) show a fatality rate close to 100%. These animals also show a high rate of mortality when infected with LUJV and several NW mammarenaviruses. A limitation to the use of inbred strain 13 guinea pigs is the lack of commercial vendors—researchers that wish to use them require private colony support. Outbred (Hartley) guinea pigs are commercially available. These Hartley guinea pigs, however, show a LASV fatality rate of only approximately 30% [76–78], although a guinea pig adapted LASV has been developed with near 100% lethality in outbred guinea pigs [79]. Guinea pig-adapted strains of LCMV and certain NW mammarenaviruses that can be used at BSL-3 have also been developed, although the outcome of infection is strain dependent. While these models have demonstrated fantastic predictive efficacy for success of vaccines and therapeutics in NHP, they are of limited value for studying host responses due to limited reagents available for immunological characterization in guinea pigs.

Wild-type laboratory mice are not highly permissive to infection by LASV and most NW mammarenaviruses, and mice fail to develop quantifiable pathologies. Conversely, certain strains of immunocompromised or knock-out mice are susceptible. Interferon-α receptor knock-out (IFNAR^−/−), human/mouse-chimeric HLA-A2.1 (humanized HHD), chimeric IFNA^−/−B6, CBA, and STAT deficient mice (STAT1^−/− mice) show varying manifestations of LASV disease, ranging from semi- to fully lethal [74, 75]. These mice have been used to study sensorineural hearing loss after recovery from LASV [80]. Immunocompetent 3-week-old transgenic mice expressing hTfR1 are highly susceptible to JUNV infection and develop lethal disease [81]. The development of this novel small-animal model provides a valuable tool for the study of JUNV pathogenesis and a platform for the evaluation of therapeutic interventions that interfere with virus binding to hTfR1 in vivo. Murine models of arenavirus infection have utility because many immunological reagents are available; however, the impact of each model’s immunological deficiency should be appropriately accounted for.

While rhesus macaques, common marmosets, and squirrel monkeys have been described as LASV models, the most frequently used NHP model is the cynomolgus macaque due to a more consistent susceptibility. Disease manifestations in all NHP models closely mimic that of severe Lassa fever and are likely the most predictive for MCM efficacy in humans. Disease severity in NHP models is understood to be dependent on LASV strain and dose [74, 82, 83]. NHP models benefit from the availability of many useful immunological reagents due to cross-reactivity of monkey antigens with human antigens and heavy investment from the biomedical field. Survival models tied to hearing loss observed in humans have been described [84], but highlight the challenges of performing survival studies with NHP in high containment where studies often cannot go over 45 days postchallenge. Nonetheless, the requirement for BSL-4 facilities, where space and study prioritization factor heavily into study implementation, have been an ongoing limitation. Perhaps most concerning, due to shifts in global animal trade caused by the COVID-19 pandemic, NHP resources are becoming increasingly scarce, which has further compounded difficulties in carrying out pathogenesis studies and critical evaluations of MCMs [85].

Knowledge Gaps

1. Route of challenge may impact disease course and thus MCM efficacy. Most infection models employ an intraperitoneal (rodent) or intramuscular/subcutaneous (NHP) challenge route; yet natural infection (exposure to rodent excreta) likely is the result of mucosal exposure in the host. While limited small-particle aerosol models do exist in the context of modeling deliberate release, there are few if any data on more traditional mucosal challenge models.

2. Some of the first detections of LCMV infections in human were in transplant patients and areas of Africa where coinfections with HIV and malaria are common. MCM evaluation and pathogenesis studies for LCMV and other arenavirus infections should also be performed in immunocompromised populations.

3. There exist only limited data on survival models of OW or NW mammarenaviruses.

4. Not all isolates are lethal in NHP despite being from human fatal cases. Mechanisms that determine virulence are unclear.

ALTERNATIVES TO WORK WITH BSL-4 PATHOGENS

Handling most arenaviruses of public health priority requires high containment laboratories. Potential solutions to high containment work include research with virus-like particle (VLP) systems. However, VLPs provide an incomplete picture of events of infection after entry or host responses to productive infection. Pseudotyped viruses are also useful for study of entry and related MCM development, but not reflective of the pleomorphic shape of most arenavirus and do not undergo postentry replication steps. Surrogate virus infection models (ie, Pichinde virus in hamsters, LCMV in mice and guinea pig) may be useful in predicting viral entry, but lack key pathogenic features only found in authentic isolates. Surrogate viruses also have limited value where antigen is central to countermeasure efficacy or understanding pathogenesis.

Knowledge Gap

• 1. Surrogate virus systems, particularly those that can utilize murine models at BSL-2, have not been a research priority and are underdeveloped.

DIAGNOSTICS

Lassa fever is difficult to recognize on clinical grounds alone, especially in the early stages. Prompt laboratory diagnosis is therefore essential for treatment to begin as early as possible. While still limited to some central laboratories, Lassa testing has become more readily available in endemic regions of West Africa. Laboratory-made and commercial polymerase chain reaction (PCR) assays are available. Likewise, LASV antigen-capture and IgM- and IgG-capture ELISA using recombinant LASV proteins have been produced and characterized [86]. Immunoglobulin M antibodies appear not to be a reliable indicator of recent LASV infection, as this class of anti-LASV antibody can persist for months or years after infection [87]. A lateral flow immunoassay for LASV antigen can reliably detect LASV antigen from a drop of blood of patients obtained by a simple fingerstick. In principle, rapid tests could be disseminated to outlying clinics and laboratories and reliably and safely performed by persons with limited laboratory training.

Knowledge Gaps

1. Diagnostics for NW mammarenaviruses are not generally available.

2. The extent of antigenic cross-reactivity between proteins of LASV and proteins of other OW arenaviruses in Africa remains to be determined. Exposure to non-LASV arenaviruses could influence seroreactivity.

IMMUNOTHERAPEUTICS

Humoral immune responses play a minor role, if any, in recovery from LASV infection. Nevertheless, immunotherapy appears to provide a significant therapeutic benefit. It is well established that convalescent plasma from survivors of Argentine hemorrhagic fever, caused by JUNV, can increase survival of people with acute infection. Convalescent plasma is the only approved treatment for any arenavirus infection [88, 89]. Plasma from human Lassa fever survivors has also been used to treat Lassa fever in Cynomolgus macaques [90, 91]. In contrast, convalescent plasma was less effective than ribavirin in treating acute Lassa fever [92]. Although neutralizing antibodies against LASV GPC are rare, 16 neutralizing human monoclonal antibodies (mAbs) from survivors, some of whom had long-term, multiple LASV exposure, have been cloned and expressed. Three of these antibodies, 8.9F, 12.1F, and 37.2D, have potent and broad neutralization activities against multiple LASV lineages [26]. A monoclonal antibody cocktail, termed arevirumab-3, comprising these 3 antibodies, provides complete protection in Cynomolgus macaque models, even when delivered in late-stage Lassa fever (8 days postinfection with death typically occurring on day 11) and at low doses [93, 94]. Arevirumab-3 also protects Cynomolgus macaques after challenge with LASV lineages II and III (unpublished, Cross et al). Work is underway to validate this treatment against other LASV lineages.

All neutralizing antibodies characterized to date target prefusion LASV GP and nearly all bind to quaternary epitopes that span the GP1 and GP2 subunits. These antibodies have various mechanisms of action, including likely competition with cellular entry factors and prevention of the LAMP1 receptor switch or fusion of viral and host cell membranes (Li, Hastie, and Saphire, unpublished observations). In contrast, thus far, all neutralizing antibodies targeting NW viruses bind to the GP1 subunit alone and all function by direct competition with TfR1. Furthermore, elicitation of neutralizing antibodies against JUNV is possible through vaccination with the GP1 subunit alone [95].

No evidence yet exists that NW arenaviruses employ receptor switching in the endosome, nor does structural analysis of several NW GP1 subunits suggest that conformational changes occur upon exposure to low pH or separation from GP2 [24]. Hence, the difference in epitope requirements for neutralizing antibodies may reflect the particular entry strategy employed by NW viruses versus LASV. Variations in the entry pathways may also be an underlying reason for the disparity in the effectiveness of convalescent plasma between the 2 groups of viruses.

Although neutralizing antibodies against NW arenaviruses appear easier to elicit, their breadth is limited. Of anti-JUNV antibodies identified to date, most are not cross-reactive against MACV [96] and none neutralize other clade A viruses [97]. The primary underlying cause of the narrow activity for these antibodies is high sequence variability in the receptor binding site of NW viruses. For example, MACV GP1 contains an insertion in a loop that interacts with TfR1 yet blocks binding by an anti-JUNV GP1 neutralizing antibody. Similarly, other NW arenaviruses have additional N-linked glycans that may occlude access by antibodies [98].

Therapeutic antibodies are available or in development for JUNV [99] and MACV [100]. All NW mammarenavirus neutralizing antibodies identified to date are against the GP1 subunit and block TfR1 binding. It appears that exposure to the GP1 subunit of NW mammarenaviruses alone is capable of neutralizing antibodies. As the GP1 subunit may be presented authentically, no matter the conformation of the GP2, it may be more straightforward for an infected or immunized person to develop anti-GP1 neutralizing antibodies. Furthermore, such antibodies may be easier to detect as the GP1 subunit in isolation is easier to make recombinantly and is not subject to the conformational heterogeneity of a complete GPC. The formal possibility remains that there could be other types of neutralizing antibodies against NW arenaviruses that involve GP2-containing epitopes. In the absence of well-formed GPC maintained in the prefusion state through the use of molecular clamp technology, such antibodies are difficult to identify for NW mammarenaviruses.

Knowledge Gaps

1. It is unclear whether the sequence variability in receptor-binding domains of NW mammarenaviruses is a barrier to cross-protective vaccines and/or Ab-based therapeutics. It remains to be determined if receptor use is related to the types of neutralizing antibodies elicited or to the antibody repertoire needed to generate a protective response.

2. It is unknown whether neutralizing antibodies against GP2-containing quaternary epitopes for OW or NW mammarenaviruses exist.

3. mAbs provide stunning late-stage and low-dose protection against LASV in NHPs. However, clinical trials of LASV and NW mammarenavirus mAbs in humans have not yet been conducted. Opportunities for more broadly cross-reactive neutralizing antibodies against the NW or OW arenaviruses may be available.

4. Beyond LASV, no human antibodies have yet been described for LCMV or LUJV. Some neutralizing antibodies from mice have been described for LCMV, but none for LUJV. Some neutralizing antibodies against LASV can neutralize LCMV, but none neutralize authentic LUJV (Cross, unpublished).

5. It is unclear whether similar antibodies to the GPC-specific antibodies against LASV exist for NW or other OW mammarenaviruses.

6. While immunotherapeutics are very promising, most arenaviruses have neurotropic tendencies. Monoclonal antibodies are unlikely to penetrate the blood-brain barrier. Options available to treat the infected central nervous system may include small molecules or nanobodies.

VACCINES

The live-attenuated Candid#1 strain of JUNV is the only approved arenavirus vaccine [101]. Candid#1 is licensed for use in Argentina; however, due in part to the possibility of genetic reversion to virulence of its single GP2 attenuating mutation, Candid#1 is not licensed in the United States [102]. Multiple platforms have been evaluated as potential Lassa fever vaccines, including a ML29 MOPV/LASV live reassortant (L segment from Mopeia virus, S segment from LASV) [103], recombinant vesicular stomatitis viruses in which the glycoprotein gene is replaced by the coding sequence for LASV GPC [104, 105], plasmid DNA encoding codon optimized GPC [106], recombinant measles virus [107], and a vaccinia virus recombinant vaccine (modified vaccina virus Ankara-virus-like particle) expressing LASV GPC + Z [108]. Additional vaccine candidates include a rabies virus recombinant expressing LASV GPC [109], adenovirus vectored vaccines [110, 111], LASV VLP [112, 113], and a virus replicon particle vaccine [114]. Important considerations for LASV vaccines stem from their use in Africa. They should be cost-effective, stable without extensive cold chain requirements, and require a minimal number of vaccinations. Efforts to develop vaccines for other arenaviruses beyond JUNV and LASV have been limited.

Knowledge Gaps

1. Most advanced experimental vaccines target GPC, but it is unclear whether other viral targets, such as N, can be exploited. While the majority of these platforms demonstrated a robust immune response in laboratory animals, many have failed to fully protect against lethal disease in relevant disease models. Both T and B cell responses may be required for optimal protection against OW or NW mammarenaviruses.

2. It is unclear whether vaccination by one lineage of LASV can protect against other lineages. Most experimental vaccines have used the Josiah strain of LASV as platform and only a few have been assessed against a nonhomologous LASV challenge [115]. It is possible that both cross-reactive T and B cell epitopes are needed for broad protection.

3. There has been little MCM development for LUJV, LCMV, and other NW arenaviruses.

4. It is not known whether existing arenavirus vaccines can be improved. JUNV Candid#1 is approved in endemic areas, but evidence of reversion or neurovirulence exists. The durability of protection for existing and experimental vaccines has not been determined.

5. Several promising recombinant live-attenuated vaccines are under evaluation in humans, but more benign platforms (eg, VLP, subunit, inactivated vaccines) may be required for immune-compromised populations.

PROTOTYPE PATHOGENS

Viruses are organized into families based on shared properties. In-depth knowledge of prototype viral pathogens within each family can inform development of MCM for other family members. The Arenaviridae family is genetically diverse, and it is unlikely that a single arenavirus prototype will be sufficient for this purpose. A representative from both OW and NW mammarenaviruses will likely be necessary. LASV is the ideal OW mammarenavirus to serve as a prototype pathogen from a public health standpoint due to its high incidence of disease, but careful consideration of phenotypic differences across phylogenetic lineages must be taken into account. Furthermore, high seroprevalence rates suggest asymptomatic or avirulent LASV (or closely related OW arenaviruses) are circulating. For MCM development, a highly virulent isolate may be preferred; however, there is value in understanding the protection conferred by “avirulent” strains. JUNV could be the ideal NW arenavirus for similar issues related to incidence; however, there have been reports of isolate-specific outcomes where some are more neurotropic, others more viscerotropic, and others a mix of both.

Knowledge Gaps

1. It is unclear which isolates of LASV and JUNV are most representative of contemporary natural infections or whether a single prototype is appropriate for either virus.

2. Most “prototype” isolates of LASV and JUNV were isolated between 30 and 50 years ago. It will be important to understand how much currently circulating isolates have changed and how the differences will impact MCM efficacy. A challenge will be to access recent isolates, which is heavily dependent on active surveillance systems with cooperative international collaboration.

3. While there is an appropriate focus on the most pathogenic members of the Arenaviridae family, treatment options for other arenaviruses, except JUNV, remain limited. For example, immunocompromised patients infected with LCMV do not have access to specific antivirals.

CONCLUSIONS

Improved, though still limited, clinical and laboratory research infrastructure in the Lassa endemic zone of West Africa has created a pathway to development of effective countermeasures, including small-molecule drugs, immunotherapeutics, and vaccines for LASV. Both LASV vaccines and drugs or drug combinations are urgently needed. Development for MCM for other OW arenaviruses, in particular LCMV, should also be a priority. Moreover, while development of MCM for JUNV preceded that for LASV by decades, development of countermeasures against other NW arenaviruses still represents an important gap. Designation of LASV and JUNV as prototype pathogens can facilitate progress in addressing public health challenges posed by members of the Arenaviridae family.

Notes

Author contributions. All authors wrote the original draft, and reviewed and edited the article.

Acknowledgments. The authors gratefully acknowledge the Viral Hemorrhagic Fever Consortium, the Viral Hemorrhagic Fever Immunotherapeutic Consortium, and patients and families affected by arenavirus infections.

Financial support. This work was supported by National Institutes of Health (NIH) (grant numbers R21 AI137809 to K. M. H.; U01AI51801 to R. W. C.; U19 A142790, R01AI132244, R01AI141251, and U19 AI142790 to R. W. C., E. O. S., and R. F. G.; U19AI135995, U01AI151812, and UL1TR002550 to K. G. A.; U54CA260581, U54HG007480, OT2HL158260, U19AI135995, and U01AI151812 to R. F. G.); and the Coalition for Epidemic Preparedness Innovation, the Wellcome Trust Foundation, Gilead Sciences, and the European and Developing Countries Clinical Trials Partnership Programme (to R. F. G).

Supplement sponsorship . This article appears as part of the supplement “Pandemic Preparedness at NIAID: Prototype Pathogen Approach to Accelerate Medical Countermeasures—Vaccines and Monoclonal Antibodies,” sponsored by the National Institute of Allergy and Infectious Diseases, NIH, Bethesda, MD.