Reverse Genetics Recovery of Lujo Virus and Role of Virus RNA Secondary Structures in Efficient Virus Growth

Abstract

Arenaviruses are rodent-borne viruses with a bisegmented RNA genome. A genetically unique arenavirus, Lujo virus, was recently discovered as the causal agent of a nosocomial outbreak of acute febrile illness with hemorrhagic manifestations in Zambia and South Africa. The outbreak had a case fatality rate of 80%. A reverse genetics system to rescue infectious Lujo virus from cDNA was established to investigate the biological properties of this virus. Sequencing the genomic termini showed unique nucleotides at the 3′ terminus of the S segment promoter element. While developing this system, we discovered that reconstructing infectious Lujo virus using the previously reported L segment intergenic region (IGR), comprising the arenaviral transcription termination signal, yielded an attenuated Lujo virus. Resequencing revealed that the correct L segment IGR was 36 nucleotides longer, and incorporating it into the reconstructed Lujo virus restored the growth rate to that of the authentic clinical virus isolate. These additional nucleotides were predicted to more than double the free energy of the IGR main stem-loop structure. In addition, incorporating the newly determined L-IGR into a replicon reporter system enhanced the expression of a luciferase reporter L segment. Overall, these results imply that an extremely stable secondary structure within the L-IGR is critical for Lujo virus propagation and viral protein production. The technology for producing recombinant Lujo virus now provides a method to precisely investigate the molecular determinants of virulence of this newly identified pathogen.

INTRODUCTION

Lujo virus (LJV) is a newly discovered arenavirus associated with a viral hemorrhagic fever (VHF) outbreak in southern Africa in September 2008, with an unusually high case fatality rate of 80% (8, 29). The index case was a female travel agent living on the outskirts of Lusaka, Zambia, who was evacuated to Johannesburg, South Africa (generating the virus name Lujo [from “Lusaka-Johannesburg”) and initiated a chain of person-to-person transmission. The index patient and 3/4 secondary cases had fatal outcomes.

All arenaviruses (with the exception of Tacaribe virus) are maintained in nature by persistently infected rodent hosts. Contact with contaminated rodent excreta, blood, or infectious aerosol is the origin of disease outbreaks, which can subsequently involve person-to-person transmission via direct and indirect contact with infected individuals (10). It is unclear how the initial Lujo HF patient became infected, but a rodent reservoir is considered likely.

Arenaviruses are grouped in 2 geographically and genetically distinct complexes: the New World and Old World arenaviruses. Both complexes contain viruses which can cause VHF in immunocompetent individuals, with typical case fatality rates of 5% to 35% (10). These pathogens include the New World clade B arenaviruses (Junín, Machupo, Guanarito, Sabiá, and Chapare viruses) and the Old World arenavirus Lassa virus. Phylogenetic comparison of Lujo virus sequences with those of other arenaviruses showed that Lujo virus is genetically very divergent from other known Old World and New World arenaviruses, highlighting the distinct character of this highly pathogenic arenavirus (8).

The arenaviral genome comprises 4 genes arranged in ambisense orientation in pairs on the small (S) and large (L) RNA segments. The S segment encodes the glycoprotein precursor (GPC) and nucleoprotein (NP), while the L segment encodes the RING finger Z protein (Z) and L RNA-dependent RNA polymerase (L-RdRp). Electron micrographs show encapsidated arenavirus genomes as closed circular filaments (28, 36). The circular nature of the viral genome is likely maintained by complementary inverted sequences at the 5′ and 3′ ends of the viral RNA segments, which form a panhandle-like RNA structure (4, 5, 11). The 19 terminal nucleotides (nt), which may contribute to formation of the panhandle structure on both segments, are strictly conserved (27) and critical for replication (17, 30).

The encapsidated viral RNA serves as the template for L-RdRp to initiate viral gene transcription and genome replication. Transcription is initiated by a capped primer RNA snatched from cellular mRNA by the endonuclease located at the N terminus of L-RdRp (26). Since arenaviral open reading frames (ORFs) are in opposite orientations on viral RNA segments, viral mRNA transcription is initiated at the 3′ end of the genome and antigenome-encapsidated templates. Viral mRNA transcription terminates in a noncoding intergenic region (IGR) composed of 1 or 2 strong RNA stem-loop structures (18, 31, 32). The stability of these RNA elements is apparently critical for terminating L-RdRp transcription (22), reminiscent of Rho-independent transcription termination in bacteria (37).

Reverse genetics systems provide invaluable tools for studying viral replication and pathogenesis and for developing potential attenuated vaccine strains (3, 13). Since Lujo virus is highly pathogenic and its genome appears unique, we endeavored to develop a reverse genetics system for recovering Lujo virus from cloned cDNA.

Here, we report the successful generation of infectious Lujo virus variants by reverse genetics. We also demonstrate that the Lujo virus L segment IGR is 36 nt longer than previously reported (8) and is necessary for optimal propagation of authentic Lujo virus. In addition, the 3′ end of the Lujo virus S RNA segment, which represents the replication and transcription promoter, is surprisingly shown to be unique relative to those of the other arenaviruses.

MATERIALS AND METHODS

Viruses and antibodies.

The prototype Lujo virus was isolated from the blood of the second Lujo HF case, a paramedic who died in Johannesburg, South Africa. This Lujo virus strain was passaged no more than 5 times on Vero-E6 cells. Lujo virus antiserum produced in hamsters and goat anti-hamster Alexa Fluor 488 secondary antibody (Invitrogen, Grand Island, NY) were used for indirect immunofluorescence assays (IFAs). All work with infectious Lujo virus and attempts to rescue virus from cDNA were performed in a biosafety level 4 (BSL4) facility.

Sequencing virus genomic RNA termini and L segment IGR.

The 5′ and 3′ termini of Lujo virus S and L genomic segments were determined by rapid amplification of cDNA ends (RACE) protocols as described earlier (1, 14) and by the use of commercial RACE kits (Roche Diagnostics, Indianapolis, IN). The 140-nt L-IGR sequence was obtained by reverse transcriptase PCR (RT-PCR) using a Qiagen OneStep RT-PCR kit (Qiagen, Valencia, CA). The resulting RT-PCR product was cloned into a modified V(0.0) backbone (2), and 78 individual clonal sequences were obtained. Of the 78 clones, only 2 had sequences matching previously reported the IGR of 104 nt. The remaining 76 clones had longer sequences from which a new L-IGR consensus sequence was derived and defined as L-IGR-140.

Plasmids.

With the exception of the last 19 nt at the L and S RNA termini, the Lujo virus isolate genomic sequence was previously reported (8). Viral RNA from this isolate was therefore used as the template to generate all the Lujo virus cDNA clones. Full-length Lujo virus antigenomic segments were cloned into the same modified V(0.0) T7 expression vector used to rescue Lassa virus from cDNA (2). [The detailed cloning strategy of the L and S segments and the modified V(0.0) vector sequence are available upon request.] Briefly, the S and L segments were amplified by a 2-step RT-PCR protocol using Thermoscript RT (Invitrogen) and Phusion PCR (New England BioLabs, Ipswich, MA) enzymes. Since this approach produced no PCR products for the region encompassing L-IGR, we used a SuperScript III One-Step RT-PCR system with Platinum Taq High Fidelity (Invitrogen) to amplify this problematic region. The complete L segment with a 104-nt IGR (JX017361) was assembled from 2 L fragments, and the S segment (JX017360) was assembled from a single and complete S segment RT-PCR product. The L and S RT-PCR products were assembled and cloned into the modified V(0.0) vector using InFusion PCR cloning (Clontech, Mountain View, CA) to generate the pLJL construct, here referred to as pLJL IGR-104, and the pLJS construct. All full-length clones matched previously published genome and terminus sequences, including the initial G and terminal C, which are required for efficient initiation of T7 RNA polymerase antigenome transcription and cleavage of the 3′ end by hepatitis delta ribozyme (20).

The majority of the Z protein open reading frame (ORF), including the initiation codon, was deleted from pLJL IGR-104 to generate the construct pLJL-ΔZ. The L segment reporter plasmid was obtained by replacing the Z protein ORF of pLJL IGR-104 with Gaussia luciferase (pLJL GLuc-IGR 104). The mutation of SDD₁₃₅₃ to SAA was introduced into pLJL-GLuc IGR-104 to inactivate L-RdRp (pLJL-Gluc SDD) (1, 2, 7, 19, 35). pLJL IGR-140 and pLJL-GLuc IGR-140 were obtained by replacing the IGR of pLJL IGR-104 or pLJL-GLuc IGR-104 with L-IGR-140 from a clone comprising the new consensus. The pLJS-CLuc S segment reporter plasmid was obtained by replacing the GPC gene with the Cypridina luciferase gene (CLuc).

Virus rescue.

BSRT7/5 cells were cultured as previously described (1). Recombinant Lujo virus (rLJV) was rescued by transfecting a 10-cm² well of subconfluent BSRT7/5 cells with 1.5 μg pLJS, 1.5 μg pLJL, and 9 μl Mirus LT1 transfection reagent (Mirus Bio LLC, Madison, WI). The viruses were harvested from cell supernatants 4 days posttransfection, and infectious titers were obtained by the use of a 50% tissue culture infective dose (TCID₅₀) in Vero-E6 cells. Since Lujo virus did not produce an overt cytopathic effect in Vero-E6 cells, we determined infectious titers by IFA 2 to 3 days postinfection with Lujo virus-specific antisera produced in hamsters.

RESULTS

Sequencing Lujo virus RNA genome termini.

Since the terminal sequences of Lujo virus S and L segments were not available, we used 5′ and 3′ RACE analyses to determine these critical genome features. For all previously characterized arenaviruses, the terminal 19 nt of the S and L segments are completely conserved, with the exception of a nontemplate 5′ pppG at the ends of the lymphocytic choriomeningitis virus (LCMV) and Tacaríbe virus (14, 34) genomes. When we amplified the 5′ ends of the Lujo virus genome by RACE, we also noted the possible presence of a nontemplated 5′ pppG on both the S and L genomic segments (Fig. 1A). Interestingly, the Lujo virus L segment 3′ end terminated with an extra C following the arenavirus consensus sequence (1, 27). Similarly, the terminal 3′ nucleotide on the S segment was a mix of a C and A (Fig. 1B); consequently, it appears that at least a fraction of the S segment also contains an extra C. As arenavirus 5′ and 3′ segment ends possibly form complementary panhandle structures, these sequences indicate that the Lujo virus L segment 5′ pppG might pair with the 3′ C, forming an RNA duplex that covers 20 nt instead of the 19 nt conserved across other arenaviruses. In addition, the S segment panhandle region might be 20 nt or 19 nt with an overhanging 5′ pppG, depending on whether the corresponding 3′ terminal nucleotide is C or A. More striking was the presence of nonconserved nucleotides corresponding to positions 6 and 8 of the arenavirus S segment consensus (Fig. 1B). We concluded from these experiments that the termini of Lujo virus genome RNA segments are distinct from those of previously characterized arenaviruses.

Fig 1.

Sequences of Lujo virus genomic termini. (A) 5′ and 3′ sequence analyses were performed as described in Materials and Methods. Chromatograms of the rapid amplification of cDNA ends (RACE) terminus products comprise a homopolymeric A tail incorporated after first-strand cDNA synthesis (5′ tailing) or directly on the 3′ end of the viral genome (3′ tailing). The arenavirus terminus consensus sequence is aligned above the Lujo virus sequences; red arrows indicate the first and last nucleotides of each segment. Note that the last S segment nucleotide (3′) is heterogeneous. Internal nucleotides diverging from the arenavirus consensus are boxed in red. WT, wild type. (B) Predicted base pairing of the 5′ and 3′ termini. Nucleotides not compatible with the current consensus are shown in red and possible nontemplated nucleotides in bold.

Reverse genetics recovery of rLJV.

Using the derived RNA terminus sequences and published Lujo virus consensus sequence (S accession number, NC_012776; L accession number, NC_012777), a strategy was developed to produce rLJV. Based on our previous experience with arenavirus reverse genetics systems, we cloned full-length Lujo virus S and L segments into a vector designed to transcribe antigenome RNA copies in cells expressing the bacteriophage T7 RNA polymerase (T7pol) (9). These plasmids were sequenced to verify that the termini were correct and that the overall genome sequences were faithful copies of the Lujo virus sequences published in GenBank. Full-length antigenomic RNAs transcribed from these plasmids should serve as uncapped mRNA for synthesizing L-RdRp and NP (Fig. 2). As arenaviruses require only the L-RdRp and NP to reconstitute replication-competent ribonucleoprotein particles (RNPs), we attempted to recover recombinant Lujo virus by transfecting T7pol-expressing BSRT7/5 cells with pLJS and full-length pLJL or the negative-control pLJL-ΔZ plasmid containing a Z gene deletion. As the Z protein is not required for virus transcription or replication (12, 23), this deletion should prevent only virus assembly and budding. Four days posttransfection, the BSRT7/5 cell monolayers were examined for the presence of Lujo virus antigens. Consistent with the lack of viral spread, Lujo virus proteins were detected in only a few of the cells transfected with pLJS and pLJL-ΔZ (Fig. 2). In contrast, nearly all cells transfected with pLJS and wild-type pLJL contained viral proteins, suggesting that rLJV can be recovered by expressing the wild-type S and L antigenomes. Lujo virus was recovered from the cell supernatants with a titer of 3 × 10⁵ TCID₅₀/ml 4 days postinfection, confirming successful virus rescue.

Fig 2.

Rescue of recombinant Lujo virus. (A) Schematic representation of the primary antigenomic templates produced by the pLJL and pLJS constructs. (B) Strategy for recovering recombinant Lujo virus from cDNA. Transfecting pLJL and pLJS into T7 RNA polymerase-expressing cells generated antigenomic transcripts that were translated, respectively, into RNA-dependent RNA polymerase (L-RdRp) and nucleoprotein (NP), the minimal proteins required for the replication of T7-derived antigenomic templates and the production of recombinant Lujo virus (rLJV). (C) Immunostaining of Lujo virus antigens (green) from BSRT7/5 cells transfected with the indicated combination of plasmids; nuclei were counterstained with propidium iodide (red). (D) Propagation of rLJV compared to the parental strain isolated from a patient (LJV). Vero-E6 cells were initially infected with an MOI of 0.5 50% tissue culture infective dose (TCID₅₀)/cell.

We compared rLJV and parental Lujo virus isolate growth kinetics by infecting Vero-E6 cells with 0.5 TCID₅₀/cell of each virus; titers were determined daily for 5 days. rLJV reached a maximum titer of 1.7 × 10⁶ TCID₅₀/ml 5 days postinfection, while the parental Lujo virus titers peaked 3 days after infection at 1.6 × 10⁷ TCID₅₀/ml (Fig. 2). These data indicate that while recombinant Lujo virus was successfully rescued, its replication in cell culture was impaired relative to that of the parental Lujo virus isolate.

Revising the Lujo virus L-IGR sequence.

To investigate the basis for the impaired growth of rLJV in tissue culture, we reanalyzed the complete genome sequences of rLJV and the parental isolate. The sequence of rLJV perfectly matched the Lujo virus GenBank sequence entry. However, the parental Lujo virus sequence contained discrepancies in the L segment hairpin IGR. Standard PCR produced abnormally low yields of the PCR product encompassing the parental L-IGR, and obtaining a clear sequence was difficult (data not shown). Thus, we revisited the published L-IGR sequences.

Arenavirus IGRs fold into highly stable structures that signal termination of L-RdRp transcription (22, 31). These GC-rich regions are notoriously difficult to amplify by RT-PCR, and we began to suspect that polymerase errors may have occurred in Lujo virus L-IGR during PCR amplification. We used a one-step RT-PCR kit from another vendor (Qiagen) and found that the majority of the L-IGR clones were longer (>104 nt), but no clear PCR product sequence was obtainable. The PCR product was cloned, and its consensus sequence indicated that the L-IGR was 140 nt in length (Fig. 3A), 36 nt longer than previously reported. As suspected, these additional nucleotides were almost exclusively G and C and were predicted to increase the free energy of the L segment genome and antigenome stem-loop III and stem-loop II structures by ∼2.5-fold (Fig. 3B). To confirm that this revised longer L-IGR was not a feature selected during virus passage in tissue culture, this region was also directly amplified by RT-PCR from RNA extracted from an original patient liver specimen. The sequence from the liver specimen matched exactly the 140-nt L-IGR from the virus isolate (Fig. 3A).

Fig 3.

Sequence and structure prediction of L segment intergenic regions. (A) Comparison of the newly derived 140-nt L-IGR with the 104-nt L-IGR. (B) The previous (L-IGR 104) and new (L-IGR 140) intergenic region (L-IGR) structures and free energy (ΔG) were predicted using the CLC Main Workbench 6 RNA prediction package (CLC Bio, Aarhus, Denmark). The predicted hairpin structures (I, II, and III) are annotated according to their locations with respect to the 5′ end. Additional nucleotides from L-IGR 140 not seen in L-IGR 104 are shown in green. ΔG of hairpin II and ΔG of hairpin III are indicated under the corresponding structures.

Lujo virus 140-nt L-IGR increases reporter gene expression.

Since IGRs are transcription termination structures and since the stability of the IGR RNA stem-loop is important in arenavirus transcription (22, 31), we hypothesized that the 140-nt L-IGR might terminate viral gene transcription more efficiently than the 104-nt L-IGR (Fig. 2). To test potential differences between the 104-nt and the 140-nt L-IGRs, we used a replicon system where the S segment GPC was replaced with CLuc and the L segment Z with GLuc (Fig. 4A). Since Z and GPC were absent, the expression of reporter plasmids should have produced active RNPs but no viral particles. Cotransfection of the reporter plasmids yielded CLuc and GLuc expression proportional to the virus protein production associated with the S and L segments, respectively. To confirm the reporter signal specificity, a catalytically inactive L-RdRp (SAA) was used as a negative control. pLJL-GLuc-SAA, pLJL-GLuc IGR-104, or pLJL-GLuc IGR-140 was cotransfected with pLJS-CLuc, and luciferase activity was determined. CLuc reporter activity (indicating S segment expression) was the same regardless of whether the 104-nt or the 140-nt L was used (Fig. 4B). In contrast, pLJL IGR-140 approximately tripled GLuc activity compared to that of pLJL IGR-104, indicating that the 140-nt L-IGR specifically enhances L segment protein expression.

Fig 4.

Effect of L-IGR on reporter gene expression. (A) Schematic representation of antigenomic T7 RNA polymerase transcripts derived by transfecting the indicated plasmids. The L segment Z protein and S segment glycoprotein precursor (GPC) genes were substituted with Gaussia luciferase (GLuc) and Cypridina luciferase (CLuc) reporter genes, respectively. Subconfluent BSRT7/5 cell monolayers plated in 2-cm² wells were transfected using 1.5 μl of Mirus LT1 (Mirus Bio), 0.2 μg of pLJL-SDD, pLJL-GLuc IGR-104, or pLJL-GLuc IGR-140, 0.2 μg of pLJS-CLuc, 0.05 μg of pGL3-control luciferase (Promega, Fitchburg, WI), and 0.040 μg of pCAGGS-RVFV-NSs to suppress any cell background transcription of the reporter genes (21). At 48 h posttransfection, GLuc and CLuc activities were quantified by processing 25 μl of cell supernatant with BioLux Gaussia and Cypridina luciferase assay kits. Light emitted by the luciferase reactions was measured with a Synergy 4 plate reader (Biotek, Winooski, VT). GLuc and CLuc activities were normalized to firefly luciferase activity in the cell lysates by the use of a Bright-Glo luciferase assay system (Promega) per the manufacturer's instructions. Levels of CLuc (B) and GLuc (C) luciferase activity were measured in arbitrary units (AU). Error bars indicate standard deviations (n = 3).

Recombinant Lujo virus with the 140-nt L-IGR grows like the parental virus.

Based on the reporter system results, we hypothesized that the 140-nt L-IGR represents the authentic Lujo virus L-IGR and that replacing the 104-nt L-IGR with the 140-nt L-IGR would correct the original rLJV growth defect. We constructed the 140-nt L-IGR version of the L segment-transcribing plasmid, and this recombinant virus was successfully rescued (for S, JX017360; for L, JX017362) (Fig. 5). We found that while rLJV-104 virus grew to lower titers than the parental Lujo virus isolate (Fig. 5B), rLJV-140 grew equivalently to the parental Lujo virus, confirming that the 140-nt L-IGR is critical for efficient viral replication and growth. In summary, these data indicate that the previously reported Lujo virus sequence contains a truncated version of the L-IGR which, if engineered into infectious virus, leads to reduced viral growth.

Fig 5.

Effect of L-IGR on viral propagation. (A) Schematic representation of the antigenomic T7 RNA polymerase transcripts derived from plasmids used to recover recombinant Lujo virus containing a 104-nt or 140-nt L-IGR. (B) Vero-E6 cells were initially infected at an MOI = 0.1 TCID₅₀/cell, and propagation of recombinant Lujo virus variants with the long L-IGR (rLJV-L-IGR140) or short L-IGR (rLJV-L-IGR104) was compared to propagation of the wild-type parental LJV isolate.

DISCUSSION

The successful development of a Lujo virus reverse genetics system is a major technical achievement that will likely aid in the precise assessment of Lujo virus pathogenesis. We first utilized this system to determine that the Lujo virus L-IGR encompasses a longer and more stable hairpin than previously reported and that this highly stable RNA stem-loop was critical for efficient Lujo virus propagation in cell culture.

In all of our Lujo virus rescue attempts (n = 12), simply transfecting 2 plasmids encoding Lujo virus S and L segments was sufficient to generate wild-type and mutant recombinant Lujo viruses, demonstrating system robustness and efficiency. To obtain a recombinant Lujo virus, we focused our initial efforts on sequencing the genome termini, as these are essential cis elements of viral promoters (6, 15, 17, 30). The sequences of the genomic RNA termini of Pichínde virus (33), Tacaríbe virus (34), and LCMV (14) implied the presence of a nontemplated pppG originating from internal priming by a 5′ dinucleotide triphosphate (5′ pppGpC) and realigning the primer with the terminal 3′ G. In contrast to other arenavirus results, we detected not only a possible nontemplated 5′ pppG in Lujo virus but also an additional 3′ C at the termini of both segments, albeit a significant proportion of S termini end with 3′ G, as in other arenaviruses (1, 2, 5, 14, 27, 34). Our data suggest that Lujo virus genome termini differ significantly from those of other arenaviruses. More-detailed analyses using purified RNPs are necessary to confirm the exact Lujo virus genome terminal nucleotides. Nevertheless, these differences suggest that the mechanism by which Lujo virus replicates may differ from the prime-and-realign model proposed for Tacaríbe virus (16).

Unexpectedly, we also found 2 nucleotides within the 19-nucleotide region of the S segment 3′ termini that were not conserved compared to the arenavirus consensus previously considered invariable. Lujo virus contains an adenine and a uracil, respectively, in positions 6 and 8 (A₆ and U₈; Fig. 1), while other arenaviruses contain U₆ and A₈. Importantly, these nucleotides are located in the double-stranded RNA (dsRNA) panhandle structure central to virus transcription and replication, but unlike the other first 19 nt, nucleotides 6 and 8 are not predicted to canonically base pair with the corresponding 5′ end residues (Fig. 1). The dsRNA panhandle encompasses essential cis-acting elements that regulate viral transcription and replication. An extensive mutagenesis study of the Lassa virus promoter (dsRNA panhandle) showed that, within the first 18 nt of the arenavirus consensus, only nucleotides 6 and 8 are efficiently substituted by any of the other 3 bases, suggesting that these unpaired positions are unimportant for the replication of a model minigenome (17). In addition, these residues are dispensable for in vitro binding of Machupo virus L-RdRp at the 3′ end (19). Together, these data suggest that the nucleotides at positions 6 and 8 do not contribute to recognition, binding, or replication of arenavirus genomes.

Recent studies (24, 25) suggest that unpaired pppG and mismatched bases in the panhandle region might allow arenaviruses to escape detection by the innate immune response by preventing the activation of RIG-I, an important immune sensor of LCMV infection (38). The panhandle configuration of the Lujo virus termini and the nature of nucleotide mismatches appear to be unique (Fig. 1), leading to speculation that the predicted panhandle configuration contributes to Lujo virus replication efficiency and pathogenesis.

Our data show that the 140-bp L-IGR is the authentic Lujo virus IGR and that the originally reported 104-nt hairpin likely resulted from a technical artifact or amplification of a minor quasispecies. The main function of arenavirus IGRs is to terminate mRNA transcription, and this process requires an RNA hairpin (22, 31). Our data indicate that the length and stability of the L-IGR hairpin determine not only the level of viral reporter protein expression (Fig. 4) but also the efficiency of virus propagation (Fig. 5). According to the current arenavirus transcription termination model, the smaller, less stable 104-nt L-IGR hairpin probably attenuated Lujo virus growth by reducing viral protein expression. Most likely, the 104-nt L-IGR resulted in aberrant or inefficient transcriptional termination of at least the Z gene, as suggested by decreased L segment reporter gene expression (Fig. 4). Extrapolating from these data, we propose that the attenuation of Lujo virus containing the 104-nt L-IGR resulted from decreased Z protein expression and a subsequent defect in virus assembly.

In conclusion, we describe the development of a new reverse genetics system for generating Lujo virus. Using this novel system, we provide compelling evidence that the Lujo virus L-IGR is longer than previously reported and that the longer L-IGR is required for efficient viral propagation. This technology offers a unique opportunity to engineer Lujo viruses with defined genomes and, consequently, to determine the underlying molecular determinants of Lujo virus pathogenesis.