Efficient Rescue of Recombinant Lassa Virus Reveals the Influence of S Segment Noncoding Regions on Virus Replication and Virulence

Abstract

Lassa virus (LASV), is a significant cause of severe, often fatal, hemorrhagic fever in humans throughout western Africa, with an estimated 100,000 infections each year. No vaccines are commercially available. We report the development of an efficient reverse genetics system to rescue recombinant LASV and to investigate the contributions of the long 5′ and 3′ noncoding regions (NCRs) of the S genomic segment to in vitro growth and in vivo virulence. This work demonstrates that deletions of large portions of these NCRs confer an attenuated phenotype and are a first step toward further insights into the high virulence of LASV.

Lassa virus (LASV), family Arenaviridae, is an enveloped RNA virus (8) which can cause a severe hemorrhagic syndrome with case fatalities approaching 10 to 20% (13, 20). The public health impact of LASV is enormous, with an estimated 100,000 to 300,000 cases per year in western Africa (19, 20). Human infections typically occur following contact with contaminated excrement or inhalation of infectious aerosol from the natural rodent reservoir (Mastomys spp.) (23). Nosocomial transmission has been documented in resource-poor settings (11). LASV is a category A select agent requiring biosafety level 4 facilities for safe handling. Treatment of Lassa fever with ribavirin has proven efficacious when it is administered early in the course of infection, but prompt detection of cases is challenging (5, 21). Despite the public health importance of LASV infection, effective vaccines are not currently available.

The LASV genome is composed of two ambisense single-stranded RNA segments, S and L (8). The S segment (∼3.4 kb) encodes the glycoprotein precursor (GPC; ∼75 kDa) and the nucleoprotein (N; ∼65 kDa); while the L segment (∼7.2 kb) encodes the matrix protein (Z; ∼11 kDa) and the viral polymerase (L; ∼200 kDa). The coding regions of the segments are separated by an intergenic (IG) region with strong secondary structure, where transcription termination occurs (15, 17, 22, 26). Interestingly, analyses of the lymphocytic choriomeningitis virus (LCMV) IG region revealed functions in both transcription termination and assembly or budding of virus particles (26).

The arenavirus RNA genome termini are highly conserved, and their 5′ and 3′ termini are complementary, enabling panhandle structures to form (4). These RNA panhandles contain essential promoter elements for both virus mRNA transcription and RNA genome replication (8). In a closely related arenavirus, LCMV, the 5′- and 3′-terminal 19 nucleotides (nt) were found to be essential for maintaining promoter function, as mutation of these residues (nt 1 to 19) severely reduced or abolished promoter activity (24). However, deletions internal to the predicted panhandle structure (positions 20 to 23 or 24 to 33) did not reduce LCMV promoter activity. A similar report (14) identified two subregions contained within the 19-nt promoter element of LASV. Terminal residues 1 to 12 were strictly required, in a sequence-specific manner, for virus replication. However, the internal subregion (positions 12 to 19) could be mutated without a significant impact on virus promoter activity, as long as RNA terminal complementarity was maintained by corresponding changes in the opposite terminus.

Reverse genetics techniques have allowed the precise manipulation of virus genomes to study critical features of the virus life cycle, including mechanisms of transcription, replication, and persistence and the function of viral proteins (9, 13). Recently, reverse genetics platforms have been reported for several arenaviruses, including LCMV (12, 27), Junin virus (1), and Pichinde virus (16), allowing the development of new vaccine candidates (2, 6, 25) and tools to allow high-throughput screening of antiviral compounds (10, 28).

We report here a highly efficient and robust two-plasmid reverse genetics system to rescue recombinant LASV. We further utilize this system to study the role of the genomic 5′ and 3′ noncoding regions (NCRs) of LASV in in vitro growth and in vivo virulence. This system relies on the transcription of full-length viral complementary genomes (antigenomes) on a T7 RNA polymerase-based platform that has been well described previously (1, 7, 18). Briefly, we amplified the S and L genomic RNAs of LASV (Josiah strain) by reverse transcription (RT)-PCR and cloned them into a T7 transcription vector. For virus rescue, two plasmids encoding full-length virus complementary copies of the S and L segments, pLasS and pLasL (Fig. 1A), were transfected into BSRT7/5 cells and incubated at 37°C for 5 days. After this incubation, the supernatant was clarified and passed onto a fresh monolayer of VeroE6 cells. After 3 days of incubation, the cells were fixed and the presence of virus was detected by indirect fluorescent-antibody assay (IFA) (Fig. 1B). As a negative control, we combined the wild-type S (wtS)-containing plasmid (pLasS) with a plasmid encoding an inactive viral polymerase (pLasL-ΔSDD) in which the three essential amino acids of the catalytic core motif were replaced with three alanines. As a further control, we combined the wild-type L (wtL)-containing plasmid (pLasL) with a plasmid (pLasS-ΔSKI) encoding a GPC defective in G1-G2 cleavage via disruption of the natural SKI-I protease recognition site. As shown in Fig. 1B, only the combination of wtS and wtL allowed the rescue of recombinant LASV. As a genetic marker, two silent nucleotide changes were introduced into the pLasL plasmid for the differentiation of recombinant and wild-type viruses. The S segment plasmid (pLasS) contains a perfect copy of the wild-type LASV (wtLASV) Josiah strain (Fig. 1C). Rescue of recombinant LASV using this system was highly efficient and resulted in 100% success in at least 5 independent rescue replicates.

Fig. 1.

(A) Schematic of plasmids used to generate fully infectious LASV. (B) Recombinant (rec.) viruses were generated in BSRT7/5 cells and passaged onto VeroE6 cells as described before (1). Infected cells were fixed at 4 dpi, and LASV proteins were detected with anti-LASV rabbit serum and anti-rabbit Alexa Fluor 488. Fluorescent photomicrographs were taken using a specific wavelength filter. (C) Minor changes in the sequence of the L RNA segment were maintained as selective markers to differentiate the recombinant from the wild-type virus.

After the successful generation of recombinant wtLASV (rLASV), we utilized this robust system to study the role of the 5′ and 3′ NCRs in the virus replication cycle. The composition of the minimum promoter element in the S segment of LCMV and LASV has been studied previously (14, 24), and a critical role for the terminal 19 nt has been shown. An analysis of the predicted secondary structure of the LASV S segment RNA (Mfold [29]) revealed this well-described 19-nt terminal panhandle, but interestingly, a second high-energy panhandle structure could involve base pairing up to position 33 (Fig. 2A). Further analyses revealed that the lengths of the 3′ and 5′ NCRs of the LASV S and L segments vary significantly (Fig. 2B, top). For example, the Josiah strain 5′ NCRs are 54 and 65 nt, while the 3′ NCRs were 100 and 157 nt, for S and L, respectively. This contrasts sharply with the S and L segment NCR lengths of other LASV strains, e.g., ACAR (70 to 113), Pinneo (70 to 114), and Weller (69 to 156), respectively. While they are highly variable individually, a consistent trend exists among all of the characterized LASV strains in that all of their 3′ NCRs are longer than their 5′ NCRs.

Fig. 2.

(A) RNA secondary structure of the full-length LASV S RNA segment predicted by Mfold (29). The box on right depicts a magnification of the 5′-3′-terminal region and the predicted panhandle structure. (B) The lengths of the 5′ and 3′ NCRs of S RNAs from representative LASV strains are shown at the top; below are the available L RNA NCR lengths. The lengths of NCRs of S RNAs from LCMV strains are shown in the middle, and those from three representative phleboviruses are shown at the bottom.

Similar NCR length variation can be found in LCMV (Fig. 2B, middle). The 5′ S segment NCR of the prototype Armstrong strain and 13 others is 77 nt long, while it is only 60 nt long in the naturally occurring CA2003 strain (3). In comparison, the phlebovirus S segment, which is also ambisense, contain complete promoter elements in even more compressed NCRs (Fig. 2B, bottom).

The observed length polymorphisms led to a central question regarding the function of these NCRs in the virus life cycle. To test whether replication could be modulated by mutations on the NCR outside of the classically recognized terminal panhandle, we constructed an array of S clones with precise deletions of various lengths in the S segment NCRs (Fig. 3, left). To ensure that basal RNA-dependent RNA polymerase transcription and replication promoter elements and translation were preserved, we left intact the 5′ and 3′ 19-nt NCR and the original Kozak motif flanking the start codon of each open reading frame.

Fig. 3.

(A) Schematic of plasmids used to generate recombinant LASV mutants carrying single or double deletions in the 5′ and/or 3′ NCR. (B) Recombinant viruses were generated in BSRT7/5 cells and passaged twice on VeroE6 cells. Viral titers (right) were determined by standard TCID₅₀ assay.

Viral rescue was attempted as described above, by cotransfection of BSRT7/5 cells with the mutated pLasS-derived plasmids and pLasL, followed by two blind passages in VeroE6 cells and subsequent titration. As shown in Fig. 3 (right), the rescue of the wild type and each mutant virus was successful. Interestingly, after 2 passages on VeroE6 cells, mutant viruses containing deletions in the 3′ NCR only [rLASV-S(3′Δ24), rLASV-S(3′Δ48), and rLASV-S(3′Δ74)] were rescued with final titers (2 × 10⁵ to 4 × 10⁵ 50% tissue culture infective doses [TCID₅₀]/ml) similar to that of full-length rLASV. In contrast, viruses with 5′ NCR deletions only [rLASV-S(5′Δ25)] or 5′ and 3′ double deletions [rLASV-S(5′Δ25/3′Δ24), rLASV-S(5′Δ25/3′Δ48), and rLASV-S(5′Δ25/3′Δ74)] grew to lower titers (1 × 10⁴ to 4 × 10⁴ TCID₅₀/ml), indicating that while neither region is absolutely required for virus replication, deletions in the 5′ NCR are more deleterious to virus growth efficiency than 3′ NCR deletions.

To determine the potential significance of these minor differences in initial virus titers, we examined the in vitro growth kinetics of rLASV and three single or double deletion mutants [rLASV-S(3′Δ74), rLASV-S(5′Δ25), and rLASV-S(5′Δ25/3′Δ74)]. Wild-type and mutant viruses reached peak titers at 3 days postinfection (dpi) (Fig. 4). Interestingly, rLASV grew to 5 × 10⁷ TCID₅₀/ml, while the rLASV-S(3′Δ74), rLASV-S(5′Δ25), and rLASV-S(5′Δ25/3′Δ74) mutants grew to lower titers of 5 × 10⁶, 2 × 10⁵, and 4 × 10⁴ TCID₅₀/ml, respectively. These data clearly demonstrated that while these NCRs were not absolutely required for virus growth in cell culture, the deletions did impact virus growth efficiency.

Fig. 4.

(A) Growth curves generated by infecting VeroE6 cells with a multiplicity of infection of 0.01 and collecting supernatants at 24-h intervals. Virus titers were determined in a TCID₅₀ assay. Peak titers reached at 3 dpi are shown on the right. (B). Replication phenotype in the mouse model. Four groups of 14-day-old mice were infected intracranially with 500 TCID₅₀ of authentic rLASV or a single or double deletion mutant virus. Brains and spleens were collected at 7 dpi, and LASV RNA was detected with a quantitative RT-PCR assay. The number of positive (pos) detections and the average C_T (threshold cycle) value are shown. LASV RNA was normalized between specimens using a multiplexed primer-probe set for glyceraldehyde 3-phosphate dehydrogenase (GADPH) mRNA (ABI).

Given the modest growth differences in vitro, we investigated the impact of NCR deletions on virus virulence. Fourteen-day-old weanling mice (10/group) were injected intracranially with 500 TCID₅₀ of rLASV (a known lethal dose) or 500 TCID₅₀ of rLASV-S(3′Δ74), rLASV-S(5′Δ25), or rLASV-S(5′Δ25/3′Δ74). At 7 dpi, all animals were humanely euthanized, and brain and splenic tissues were collected for virus detection using a quantitative RT-PCR assay (Fig. 4B). Interestingly, in animals inoculated with full-length rLASV, rLASV-S(3′Δ74), or rLASV-S(5′Δ25), similar levels of viral RNA were detected in the brain (n = 10 of 10; C_T = 20 to 23) and spleen (n = 8 to 9 of 10; C_T = 24 to 25). In contrast, animals infected with the double deletion mutant rLASV-S(5′Δ25/3′Δ74) were found to have 10- to 1,000-fold lower viral RNA copies than those infected with full-length rLASV (brain, n = 9 of 10; C_T = 28; spleen, n = 3 of 10; C_T = 29). These initial results suggest that significant differences exist between the virulence of full-length rLASV and that of the double deletion virus rLASV-S(5′Δ25/3′Δ74). Apart from lower viral RNA titers, there also appears to be a reduction in the ability of rLASV-S(5′Δ25/3′Δ74) to pass across the blood-brain barrier and cause a disseminated infection. However, due to the inherent limitations of the suckling mouse model, further experiments are under way to investigate the overall contribution of these UTR deletions to in vivo lethality and disease pathogenesis in a more relevant nonhuman primate model.

In summary, we describe herein a robust reverse genetics system to generate fully infectious rLASV. We further report the generation of recombinant mutant viruses carrying specific 5′ and 3′ NCR deletions and show that these are attenuated to various degrees in replication and virulence. Further work is required to fully elucidate the role of the LASV NCRs in the virus life cycle and, most importantly, virus virulence.

Nucleotide sequence accession numbers.

The sequences determined in this study have been submitted to the GenBank database and assigned the following accession numbers: wtLASV, HQ688672 and HQ688674; rLASV, HQ688673 and HQ688675.