Arenaviruses are enveloped viruses with a bisegmented negative-strand (NS) RNA genome and a life cycle restricted to the cell cytoplasm [2].

Each viral RNA (vRNA) segment, the small (S) segment or the large (L) segment, uses an ambisense coding strategy to direct the expression of two gene products in opposite orientation and separated by a non-coding intergenic region (IGR) (Figure 2A). The S segment (3.5 kb) encodes the viral nucleoprotein (NP) and the glycoprotein precursor (GPC), which is post-translationally processed to yield the peripheral virion attachment protein GP1 and the fusion-active transmembrane protein GP2. Trimers of GP1/GP2 (GP) form spikes that decorate the virus surface and mediate cell entry via receptor-mediated endocytosis (Figure 2B). The L segment (7.2 kb) encodes the L protein, an RNA‑dependent RNA polymerase, and the small zinc finger protein Z that is the viral matrix protein (Figure 2A,B).

Consistent with the broad host range and cell-type tropism, the highly conserved and widely expressed cell-surface protein α-dystroglycan (α-DG) has been identified as the main receptor for LCMV, LASV, and several other arenaviruses [20], whereas the human transferrin receptor 1 (TfR1) was identified as the primary receptor used by pathogenic clade B NW arenaviruses [21] (Figure 3). Recently, the TAM family (Axl and Tyro3) and C-type lectin family (DC-SIGN and LSECtin) were also reported to function as receptors for LASV [22] (Figure 3). Following receptor binding, the virions are internalized through the clathrin-independent or clathrin-dependent endocytotic pathway for OW or NW arenaviruses, respectively [23,24]. Interestingly, entry of OW LCMV and LASV into the cell is independent of caveolin, dynamin, actin, and the small GTPases, Rab5 and Rab7, but dependent on cholesterol [23,24]. Following release of the viral ribonucleoprotein (vRNP) into the cytoplasm of infected cells, the vRNP-associated viral RNA polymerase initiates viral RNA replication and gene transcription. Release of infectious progeny virions from the infected cell involves association of the vRNP core and surface GP complex along with viral matrix protein Z (Figure 3).

NP and L have been shown as the minimal components for efficient vRNA synthesis using mini‑genome (MG)-based assays [25]. Z has been shown to inhibit both viral RNA replication and transcription using arenaviral MG systems [26,27,28,29,30,31]. The function of Z in the replication and transcription were described in recent reviews in detail [18,32]. In brief, the Really Interesting New Gene (RING) domain of Z was strictly required, but not sufficient, for this inhibitory effect [26] (Figure 4A). The RING domain of Z mediates an interaction with some cellular proteins, including the promyelocytic leukemia protein PML and the translation initiation factor eIF4E. The interaction between LASV Z and PML inhibited the function of eIF4E by decreasing the binding affinity of eIF4E to the m(7)G cap ligand [33]. However, the biological roles of these interactions on arenavirus replication are still largely unknown. It might be worth to note that overexpression of PML had little effect on LCMV replication [34]. Recently, viral RNA synthesis and gene expression of MACV was reconstituted in vitro. The study using this system showed that the binding of Z to L impaired the catalytic activity of L [35]. In addition to the inhibitory role of Z in transcription and gene expression, NW arenavirus Z have been shown to bind to RIG-I and inhibit IFN-β activation [36].

NMR structure of the LASV Z RING domain showed that Z interacts directly with eIF4E [33]. It has been shown that type I IFN production by plasmacytoid dendritic cell (pDC) plays an important role on innate immune responses and interferon regulatory factor 7 (IRF7) regulates this IFN production. Since the translation of IRF7 in pDC is influenced by the eIF4E-BP/eIF4E ratio, it is of interest to examine the effect of Z on the IFN production in pDC [37].

As described above, Z protein is a viral matrix protein possessing bona fide budding activity. The Z protein consists of three domains, i.e., the N-terminal domain, central RING domain, and the C‑terminal domain (Figure 4A). The function of the N-terminus has not been well characterized. In all the arenavirus Z proteins, glycine at position 2 (G2) is myristoylated and is important for attaching to the plasma membrane (PM) (Figure 4A). The role of Z protein myristoylation is also described in section 5 (“Roles of phosphatidylinositol (PI) at the PM, myristoylation of viral matrix protein, and PI3K on viral budding”). The central region of Z protein contains the RING domain, which is well‑known to function in E3 ligase as well as protein-protein interactions [38,39]. As described above, the best characterized function of the RING domain in Z is the regulation of viral gene expression [26,27,35].

It is not known whether this RING domain contributes to Z-mediated budding by itself. It has been shown that the YLCL motif in the Mopeia (MOPV) Z RING domain recognizes ALIX/AIP1 and recruits NP [40,41]. The best characterized domain in Z is the C-terminus where the L-domain is located (Figure 4A,B). The L-domain is a short amino acid motif, which plays an essential role in virus budding, and three motif sequences, PT/SAP, PPXY, and YPXnL (or YXXL), have been characterized [42]. In addition, other sequences including FPIV were also reported to act as an L-domain [43]. Although the cellular host factors that interact with many of these L-domains have been identified [42,44,45], the cellular partner of the FPIV sequence has not been found (Figure 4B).

Three L-domain motifs, PT/SAP, PPXY, and YPXnL (or YXXL), interact with the cellular proteins, Tsg101, Nedd4-like E3 ligase, and ALIX/AIP1, respectively [42]. Tsg101 and ALIX/AIP1 are components of the Endosomal Sorting Complex Required for Transportation (ESCRT) complex. The ESCRT machinery is involved in protein sorting, cytokinesis, and autophagy, and all of which are key mechanisms for cellular homeostasis [46]. We describe the interactions between L-domains and cellular factors in detail in section 6 (“Involvement of ESCRT machinery in the budding of enveloped viruses”).

As shown in Figure 4A,B, the Z proteins of OW arenaviruses possess PT/SAP (or similar to PT/SAP) and PPPY, but LUJV possesses only the PTAP motif. Some NW arenaviruses also possess only the PT/SAP motif. PICV and WWAV possess overlapping PT/SAPPY, very similar to that of Ebola virus (EBOV) VP40 [47]. Tacaribe virus (TCRV) Z does not possess a canonical L-domain, but rather possesses an L-domain-like motif. We and others previously showed that the TCRV Z ASAP motif (similar to the PT/SAP motif) do not contribute to Z-mediated budding [48,49]. It will be important to assess the different budding modes among arenaviruses and their functional mechanisms.

Viral matrix proteins of most RNA viruses, including retroviruses, possess a bona fide budding activity. That is, sole expression of viral matrix protein can produce virus-like particles (VLPs). This makes it possible to examine the molecular mechanism of virus budding, especially for BSL-4 agents, such as LASV. In this regard, HIV-1 Gag is the best characterized matrix protein. Initially, the p6 domain of HIV-1 Gag was reported to play a critical role in HIV-1 budding [50,51,52]. Other retrovirus Gag proteins have also been characterized in detail, including equine infectious anemia virus (EIAV) [53], Moloney murine leukemia virus (Mo-MLV) [54], Mason-Pfizer monkey virus (M-PMV) [55], Rous sarcoma virus (RSV) [56,57], prototypic foamy virus (PFV) [58], and human T-cell leukemia virus type 1 (HTLV-1) [59]. Vesicular stomatitis virus (VSV) and Rabies virus (RV) M and EBOV and Marburg virus (MARV) VP40 also function as viral matrix proteins [47,60,61,62,63,64,65,66,67,68,69]. The M proteins of paramyxoviruses, which includes measles virus [70,71], Newcastle disease virus (NDV) [72], SV5 [43], human parainfluenza virus (hPIV) [73], Sendai virus [74], and Nipha virus [75,76,77], have also been shown to have budding activity [78,79] (Figure 4B). In the case of arenaviruses, we and others have shown that LASV and LCMV Z are viral matrix proteins [80,81,82,83]. Z proteins of JUNV, MACV, TCRV, and MOPV have also been reported to be viral matrix proteins [40,48,49,84,85]. All Z proteins of these viruses possess G2, central RING domain, and L-domain at the C-terminus [18] (Figure 4A).

Although Z plays a central role in arenavirus assembly and budding, little is known about the domain(s) required for this process. The best studied domain is the L-domain, which is described above. In addition, G2 of Z has been shown to be myristoylated similar to HIV-1 Gag, which facilitates its membrane anchorage (Figure 4A). Several studies have indicated the importance of phosphatidylinositol for the association of HIV-1 Gag with PM [86,87]. It is not yet known whether the same mechanism is crucial for arenavirus budding. Capul et al. demonstrated that the highly conserved amino acid residues among the arenavirus Z proteins are important for incorporation of vRNP into the virion, but not budding itself [88].

It was shown that myristoylation of TCRV Z at G2 is important for self-assembly of Z proteins [89]. This suggests that self-assembly of Z occurs once Z interacts with the PM through its myristoylation. In addition, R37, N39, W44, L50, and Y57 (Zinc domain I) in TCRV Z were shown to be critical for the interaction with viral L protein [90]. TCRV Z also interacts with NP, and this role in viral budding is described in section 4.1. (“Transport of vRNA and NP to the budding site”) [49,89]. LCMV and LASV Z interact with the C-terminus of NP, which does not overlap with the self-assembly domain and the functional domain as an interferon (IFN) antagonist [91,92]. These results indicated that this Z‑NP interaction recruits vRNP complex and induces uptake into infectious progeny virions. Interestingly, it was also shown that there is some specificity of NP-Z interaction between LASV and LCMV [91]. Briefly, LCMV NP interacts with both LASV and LCMV Z, but LASV NP interacts only with LASV Z. A previous study showed that JUNV Z L79, which is conserved among almost all arenaviruses, is important to recruit TCRV NP into Z-mediated VLP and its interaction facilitates GP incorporation into VLP [85] (Figure 4A).

Previously, it was reported that the requirement of L-domain of HIV-1 Gag for virus propagation was cell type dependent [93]. Targeting of HIV-1 Gag to PM or Multi-vesicular body (MVB) and the L-domain requirement for virus budding seem to be cell type dependent [93,94,95,96]. In the case of RV, the PPXY motif is critical for its replication and pathogenesis in a mouse model [97]. In the case of VSV, which belongs to the Rhabdoviridae along with RV, virus budding in BHK-21 cells is dependent on the PPPY motif, but not PSAP, within M protein [98]. The PPPY motif in VSV M was also shown to be important for VSV replication on HEK293T cells [99]. These studies also implied that Vps4A/B requirements may be different among cell types for VSV replication. In the case of EBOV, the L-domain is not essential for replication and propagation in cell culture, although the L-domain mutant virus showed a 2–3 log reduced in viral titer compared to wild-type (WT) virus in Vero cells [100]. It has not been determined whether the role of L‑doomain on arenavirus replication and budding are cell type dependent. However, the data from other enveloped viruses containing multiple L-domain motifs suggest that the requirement for the L‑domain located in the C-terminus of Z for efficient replication and budding may depends on cell type.

All the virion components must be concentrated at the site of budding, but the mechanism underlying this process is largely unknown for arenaviruses. The presence of IGR in vRNA, which is a hairpin structure aligned between two viral coding genes in both L and S segments, is one of the features of arenavirus [2] (Figure 2A). IGR plays a critical role in LCMV genome incorporation [101]. Previous studies showed that Z interacts with NP, L, and GP [40,49,85,89,90,91,102]. Through the interaction between Z and NP/L, vRNP may be recruited into the virion (Figure 8).

The contribution of NP to LCMV and LASV Z-mediated budding has not been known, although LCMV and LASV NP interact with LCMV and LASV Z, respectively. On the other hand, some arenaviruses’ NP were reported to contribute to the assembly and budding processes. One example is MOPV NP. ALIX/AIP1, one of the ESCRT components, can interact with both MOPV NP and Z, and therefore bridges the interaction between NP and Z [41] (Figure 4A). This interaction facilitates MOPV Z-mediated budding. Another example is TCRV NP. TCRV Z-mediated budding is enhanced by the co-expression of TCRV NP, which was not observed with JUNV Z and NP [49]. These observations suggest that arenavirus Z protein may utilize a different mechanism to facilitate budding.

Although we previously showed that Tsg101 is important for MARV VP40-induced VLP budding and the PPPY motif located in VP40 is crucial for Tsg101-mediated budding and incorporation into VP40-induced VLP [66], it is unlikely that MARV VP40-Tsg101 interaction is direct [103] (Figure 4B). MARV NP was reported to interact with Tsg101, and facilitate VP40-induced VLP production [104]. Based on these data, it is possible that viral proteins can recruit ESCRT components with several different mechanisms and facilitate budding.

GP is a viral surface glycoprotein that is responsible for binding to the cellular receptor and for the subsequent fusion event (Figure 3). LASV, LCMV [20], as well as clade C of NW arenaviruses [105] use α-DG as the primary receptor, whereas clade B of NW arenaviruses utilize TfR1 [21,106] (Figure 3). Therefore, the incorporation of GP into the virion is a critical process to produce infectious progeny virions. GPC is cleaved by a cellular site 1 protease (S1P/SKI-1) at either ER or Golgi [107,108,109,110], to form GP trimer complex [108,111] (Figure 5). Interestingly, this GP cleavage by S1P/SKI-1 is necessary to incorporate GP into virions, but not transportation to the cell membrane [23]. Recently, it was shown that the S1P/SKI-1 processing is different between GPC and cellular substrates, such as SREBP2 and ATF6, based on the dependency on S1P/SKI-1 autoprocessing [110]. This S1P/SKI-1 cleavage is also important for GP oligomerization [112]. Thus GPC cleavage by S1P/SKI-1 is an attractive target for arenavirus therapy [113,114,115,116,117]. Indeed, we and others have reported that the small chemical compound PF-429242, which targets S1P/SKI-1, can be a leading compound to combat arenavirus [115,118]. In addition, stable signal peptide (SSP) located at the N-terminus of GPC was shown to play a critical role in GPC trafficking of LCMV and JUNV [119,120] (Figure 5), as well as its own expression, processing, cis-acting, and fusion activity [121,122]. SSP was also showed to interact with myristoylated Z [102]. The N-glycosylation of GP is also important for its expression, cleavage, and fusion activity, and these influence virion infectivity [123,124].

IFN is a well-characterized host immune molecule that triggers expression of hundreds of so-called interferon stimulated genes (ISGs). In general, many of the viruses encode specific proteins that antagonize innate immune responses [159,160]. The tripartite motif containing 5α (TRIM5α) has been shown to be induced by IFN-α, and is well known as a HIV-1 entry host restriction factor [161,162,163]. In addition, TRIM5α has been shown to regulate HIV-1 budding [164,165]. Recent reports have suggested that not all of the ISGs show anti-viral activity, but rather exhibit specificity among viruses [166,167]. A wide scale experiment was performed to determine the antiviral activities of ISGs against hepatitis C virus, yellow fever virus, West Nile virus, chikungunya virus, Venezuelan equine encephalitis virus, and HIV-1. The authors screened more than 380 ISGs independently to determine their antiviral activities against these selected viruses, and demonstrated the diversity of these functions [166]. A total of 288 ISGs were then tested against VSV and murine gammmaherpes virus 68 (MHV-68) [167]. It is not clear whether these ISGs contribute to inhibition of arenvirus assembly, budding, and release. Therefore further analysis is required.

Ubiquitin is a molecule of 76 amino acids with multiple functions. Ubiquitin binds to the substrate, normally at lysine (K), and the substrates are either mono- or poly-ubiquitinated. Monoubiquitination is involved in intracellular trafficking of the substrates. On the other hand, poly-ubiquitination leads to substrate degradation via the ubiquitin-proteasome pathway to maintain cell homeostasis. Although ubiquitin is not an IFN-inducible protein, one of the ubiquitin-like proteins (UBL), ISG15 has been shown to regulate viral assembly and budding. Previously, it was reported that HIV-1 Gag and EBOV VP40 are tagged with ISG15, and regulates the budding process [168,169]. In addition, ISG15 was reported to be tagged to ESCRT III components, CHMP2A, CHMP4B, CHMP5, and CHMP6 [170,171,172]. Interaction of Tsg101 with ubiquitinated HIV-1 Gag was reported to be 10-times stronger than that with non-ubiquitinated Gag [173]. Therefore, ubiquitin and ISG15 have been suggested to be important for viral budding, but the precise role of these in arenavirus budding is unclear [174]. Some studies utilized proteasome inhibitors to assess the importance of ubiquitin, but this treatment seems to affect multiple cellular processes, not only viral protein ubiquitination and viral budding [174].

As described above, some interferon-induced proteins clearly show antiviral activity specifically at the late stage of virus replication, which included ISG15 and TRIM5α [164,165,168,169]. Recently, one of these ISGs, named Tetherin (also known as BST-2, CD317, or HM1.24), has been reported as an inhibitory cellular factor against HIV-1 [175,176]. We and other groups reported that Tetherin shows antiviral activity against other retroviruses, filoviruses, and arenaviruses [177,178]. Tetherin is constitutively expressed in terminally differentiated B cells, bone marrow stromal cells, and pDC, and is also broadly induced by treatment with Type I and Type II IFNs in various cell types [176]. Therefore, Tetherin is thought to be involved in antiviral host defense as one of the innate immunity mechanisms. In addition, Tetherin expression has been shown to be increased in multiple myeloma [179], endometrial cancer [180], primary lung cancer [181], and glioma cells [182]. Therefore, several groups have attempted immunotherapy using anti-Tetherin antibodies, which mediate antibody-dependent cellular cytotoxicity [181,183].

Tetherin is a type II transmembrane protein, consisting of four domains, i.e., an N-terminal cytoplasmic domain (CD), a single transmembrane domain (TM), an extracellular domain, and a putative C-terminal glycosyl phosphatidylinositol (GPI) anchor (Figure 7A–C), which is present on the cell surface and in perinuclear compartments. Tetherin is anchored in the cell membrane at both ends, and combined high-resolution crystallography and small-angle X-ray scattering-based modeling indicated that the Tetherin ectodomain forms a parallel coiled-coil homodimer [184,185,186,187,188]. Therefore, it has been suggested that Tetherin inhibits progeny virus release by directly tethering virions to cells, briefly by anchoring one end of the molecule on the cell membrane and the other end on the viral envelope. Progeny virions released from cells could also be directly tethered to each other by Tetherin. Electron microscopy was performed to confirm this model, and based on the distances between tethered virions and cells, the membrane-spanning model was proposed as the model of action of Tetherin [176,189,190]. Interestingly, an artificial Tetherin-like protein, composed by domain replacement of TfR at the N-terminus, dystrophia myotonica protein kinase (DMPK) at the ectodomain, and urokinase plasminogen activator receptor (uPAR) at the C-terminus was also shown to have antiviral activity similar to that of Tetherin, suggesting that this typical conformation or configuration acts to tether virion progeny at the cell surface [189].

Some viruses encode IFN antagonists to avoid or inhibit the innate immune response, such as arenavirus NP, EBOV VP35, and influenza virus NS1 [159,191,192]. In general, viral antagonists inhibit the innate immune signaling cascades. In addition to these IFN antagonists, some viruses encode specific viral proteins that antagonize the antiviral activity of Tetherin [178,193]. HIV-1 Vpu is the best characterized Tetherin antagonist. HIV-2 Env, SIV Env, SIV Nef, KSHV K5, and EBOV and MARV GP have also been reported as Tetherin antagonists, but their mechanisms of action seem to differ among these viral proteins [178,193,194,195].

Initially, we reported that Tetherin inhibits LASV Z-mediated VLP release, and this restriction was not overcome by LASV GP, but was overcome by HIV-1 Vpu expression [177]. In addition, we showed that the glycosylation of Tetherin at 65 and 92 asparagine (N) is dispensable for its activity [177]. Moreover, we showed that Tetherin dimerization through its C-C interaction had little effect on its inhibitory activity against Lassa VLP release, but was not essential for its antiviral activity [196]. It is likely that tethering of virions by Tetherin dimers is stronger than that by Tetherin monomers because of the stronger association with the membrane. Therefore, Tetherin appears to inhibit release of a wide variety of enveloped viruses from host cells by a similar mechanism, as Tetherin shows antiviral activities against many retroviruses, filoviruses, arenaviruses, and KSHV. However, there is controversy regarding the importance of glycosylation and dimerization of Tetherin for its antiviral activity.

Recent studies using mouse models have assessed the importance of Tetherin with regard to its antiviral activity against Mo-MLV [197], mouse mammary tumor virus (MMTV) [198], VSV, and influenza B virus [199]. In case of Mo-MLV and MMTV, knock out (KO) or knock down (KD) of Tetherin exhibited higher titer of these viruses in vivo, suggesting Tetherin exhibited antiviral activity against these viruses. In addition, HIV-1 Vpu, which antagonizes Tetherin, has been shown to contribute to the efficient spread of HIV-1 using the humanized mouse model transplanted human hematopoietic stem cells [200]. On the other hand, KO of Tetherin showed lower titer of VSV and influenza virus B [199]. These results indicated that the antiviral effect of Tetherin in vivo is complicated and showed different phenotype depending on the virus species.

In the case of arenaviruses, one study demonstrated that the infectious LASV was inhibited by human Tetherin [84]. In addition, this report indicated that LASV does not possess any Tetherin antagonist [84]. It is not yet clear whether Tetherin has any effect on arenavirus propagation and pathogenesis in vivo.