A Lassa Virus Live-Attenuated Vaccine Candidate Based on Rearrangement of the Intergenic Region.

Lassa virus (LASV), the causative agent of Lassa fever, infects several hundred thousand people in Western Africa, resulting in many lethal Lassa fever cases. No U.S. Food and Drug Administration-licensed countermeasures are available to prevent or treat LASV infection. We describe the generation of a novel LASV live-attenuated vaccine candidate rLASV(IGR/S-S), which is based on the replacement of the large genomic segment noncoding intergenic region (IGR) with that of the small genome segment. rLASV(IGR/S-S) is less fit in cell culture than wild-type virus and does not cause clinical signs in inoculated guinea pigs. Importantly, rLASV(IGR/S-S) protects immunized guinea pigs against an otherwise lethal exposure to LASV.

strain 13, vaccine, VHF, viral hemorrhagic fever, intergenic region, IGR, reverse genetics L assa fever (LF) was first described in Lassa, Nigeria, in 1969 (1). The causative agent of LF, Lassa virus (LASV), was subsequently discovered and isolated from LF patients and its natural host reservoir, the Natal mastomys (Mastomys natalensis) (2,3). LF poses a significant public health burden in the areas where it is endemic, mainly sub-Saharan Western Africa (e.g., Guinea, Liberia, Nigeria, and Sierra Leone). A single longitudinal study conducted more than 30 years ago in Sierra Leone estimated that hundreds of thousands of LASV infections occurred annually (4). Among the people infected, 80% are asymptomatic, whereas the other 20% develop disease that usually requires hospital admission. Nevertheless, LASV infection was largely ignored as a public health threat for extended periods of time (5), and the true incidence of LASV infections and the public health burden of LF in Western Africa remain unknown but are likely significant. Case fatality rates (CFRs) of 15% to 20% have been reported among hospitalized LF patients (6). Recently, the occurrence of a high number of LF cases associated with high (up to 60%) case fatality rates in Nigeria (6, 7) triggered reevaluation of LF risk for global health security (8). Moreover, LASV expansion outside its traditional areas of endemicity and increased travel have resulted in exportation of LF cases from Western African countries where the disease is endemic to countries where it is nonendemic (9). To date, no U.S. Food and Drug Administration (FDA)-licensed countermeasures are available to prevent or treat LASV, and current anti-LASV therapy is limited to the use of ribavirin, which is only partially effective and can cause significant side effects (10,11). Due to the impact of LF on human health and limited existing countermeasures to combat LF, the World Health Organization (WHO) included LF on the revised list of priority diseases for development of effective vaccines (12).
Like other mammarenaviruses (Bunyavirales: Arenaviridae), LASV is an enveloped virus with a bisegmented, single-stranded RNA genome (13,14). Each viral genome segment uses an ambisense coding strategy to direct the synthesis of two viral proteins from open reading frames separated by noncoding intergenic regions (IGRs) (14). The large (L) segment encodes the large (L) protein that functions as a viral RNA (vRNA)directed RNA polymerase and RING finger protein Z, which functions as a matrix-like protein for virion assembly and budding (15)(16)(17)(18). The small (S) segment encodes the nucleoprotein (NP) and the glycoprotein (GP) precursor (GPC) (19)(20)(21). NP and L, together with the vRNA, form the viral L and S ribonucleoprotein complexes (vRNPs) responsible for directing replication and transcription of the viral RNA genome. NP is also involved in counteracting innate immune responses during viral infection (22)(23)(24)(25). GPC is cotranslationally processed by cellular signal peptidase to generate a stable signal peptide (SSP) and a GPC precursor that is posttranslationally cleaved by the cellular proprotein convertase and subtilisin kexin isozyme-1/site-1 protease (SKI-1/S1P) to generate GP1 and GP2 subunits. The GP1 and GP2 subunits, together with the SSP, form mature GP peplomers on the surface of the virion envelope that mediate virion cell entry via receptor-mediated endocytosis (26)(27)(28).
For a given mammarenavirus, S-IGRs and L-IGRs differ in both sequence and predicted secondary structures. The IGRs play critical roles in mammarenavirus transcriptional control and production of infectious particles (29). The L-IGR and S-IGR of lymphocytic choriomeningitis virus (LCMV) have distinct functional roles in posttranscriptional control of viral gene expression (30,31). The importance of the IGRs in viral multiplication has also been documented for other arenaviruses. For instance, replication of a recombinant Lujo virus with a 36-nucleotide deletion in the L-IGR was impaired in cultured cells compared to its parental virus (32). A Machupo virus with a 35-nucleotide deletion within the L-IGR was only moderately less fit in cell culture than the wild-type (WT) virus but highly attenuated in vivo (33).
Importantly, a recombinant LCMV containing the S-IGR in both the S and L genome segments [rLCMV(IGR/S-S)] was only modestly less fit in cultured cells than rLCMV-WT.
However, rLCMV(IGR/S-S) was highly attenuated in a laboratory mouse model of LCMV infection, but mice immunized with rLCMV(IGR/S-S) were fully protected against an otherwise lethal infection with rLCMV-WT (30).
On the basis of these promising results, we examined whether reorganization of IGRs could represent a general molecular strategy for mammarenavirus attenuation that could be applied to LASV. In line with prospective epidemiological studies in Western Africa, live-attenuated vaccines (LAVs) represent the most feasible approach to control LF, as LAVs induce long-term robust cell-mediated responses following single immunizations (34). Here, we report the generation of a recombinant LASV that contains the LASV S-IGR in both the S and L genome segments, rLASV(IGR/S-S). In vitro, the viral fitness of rLASV(IGR/S-S) was moderately decreased compared to rLASV-WT. In vivo, rLASV(IGR/S-S) was fully attenuated in both strain 13 and Hartley guinea pigs but provided full protection, upon a single subcutaneous (s.c.) low-dose administration, against an otherwise lethal exposure to LASV-WT. Our data support further development of this strategy as a general mechanism for the development of LAVs to combat human mammarenavirus infections.
Characterization of rLASV(IGR/S-S). rLASV-WT and rLASV(IGR/S-S) growth kinetics were compared by inoculating interferon (IFN)-competent (A549) and IFN-deficient (Vero) cells with these viruses at different multiplicities of infection (MOIs) (Fig. 2). Viral titers were measured daily for 4 days postexposure (p.e.). rLASV(IGR/S-S) replicated efficiently in both cell types, although viral peak titers in both cell lines were up to 2 orders of magnitude lower than those of rLASV-WT ( Fig. 2A and B). The plaque sizes produced by rLASV(IGR/S-S) (0.60 Ϯ 0.21 mm in diameter) were significantly smaller than those caused by rLASV-WT (2.12 Ϯ 0.74 mm; n ϭ 25, P Ͻ 0.0001) (Fig. 2C). Viral spread was analyzed by immunofluorescence assay in A549 and Vero cells infected with rLASV-WT or rLASV(IGR/S-S). Both GP-positive and NP-positive cell numbers were lower in rLASV(IGR/S-S)-infected cells than in rLASV-WT-infected cells at 48 h and 72 h p.e. (Fig. 2D and E; see also Fig. S1 in the supplemental material). Reduced GP and NP expression was also observed in rLASV(IGR/S-S)-infected A549 cells by Western blotting (Fig. 2F). Taking the results together, the reorganization of IGR in LASV genome modestly reduced fitness in cultured cells.
Attenuation of rLASV(IGR/S-S) in strain 13 guinea pigs. Next, we examined whether replacement of L-IGR with S-IGR in the L segment would have an appreciable impact on LASV virulence in vivo. Since strain 13 guinea pigs are highly sensitive to LASV infection, with 100% lethality following s.c. exposure to 10 3 PFU (37), the safety profile of rLASV(IGR/S-S) was evaluated in strain 13 guinea pigs.
Strain 13 guinea pigs were inoculated s.c. with 10 5 PFU of rLASV(IGR/S-S) (n ϭ 5), rLASV-WT (n ϭ 4), or LASV (n ϭ 5). All animals inoculated with rLASV(IGR/S-S) survived to the end of the study (day 42 p.e.) (Fig. 3A) without developing any overt clinical signs of disease (Fig. 3B), including weight loss (Fig. 3C) or elevated temperature (Fig. 3D). In contrast, all animals inoculated with rLASV-WT or LASV developed several clinical signs of infection (e.g., ruffled coat, labored respiration), body weight loss, and transient  elevated temperatures (Fig. 3B, C, and D, respectively). Two of 4 rLASV-WT-exposed and 4 of 5 LASV-exposed guinea pigs eventually met euthanasia criteria. Viral loads in blood at the different times p.e. (Fig. 3E) and in the indicated tissues at the end of the study (Fig. 3F) were measured by reverse transcription-quantitative PCR (RT-qPCR). vRNAs were not detected in the blood collected at days 7, 14, 21, 28, 35, and 42 p.e. from the majority of rLASV(IGR/S-S)-inoculated guinea pigs. Low concentrations of viral RNA FIG 3 rLASV(IGR/S-S) is attenuated in strain 13 guinea pigs. Strain 13 guinea pigs were s.c. exposed to 10 5 PFU of rLASV(IGR/S-S) (n ϭ 5, blue), rLASV-WT (n ϭ 4, purple), or LASV (n ϭ 5, red). (A to D) Survival (A), clinical scores (B), body weight loss (C), and temperature changes (D) were monitored daily for 42 days. (E and F) Viral loads in the blood at different times p.e. (E) and viral loads in the indicated tissues at necropsy (F) were measured by RT-qPCR. (G) Anti-LASV IgG plasma titers were determined by ELISA at the indicated days p.e. LLOQ, lower limit of quantification (panels E and F); LOD, limit of detection (panel G); ND, not detected; Nec, necropsy date; d, day.
Significant histopathological lesions were not observed in the examined tissues of rLASV(IGR/S-S)-inoculated guinea pigs (Fig. 4A), and all examined tissues from rLASV(IGR/S-S)-inoculated guinea pigs were negative for LASV antigen (Fig. 4). In contrast, typical acute LF lesions, including those resulting from interstitial pneumonia (data not shown), hepatic degeneration and necrosis (Fig. 4A, white arrow), endocar- A LASV Vaccine Based on IGR Rearrangement ® ditis (data not shown), and splenic lymphoid depletion (data not shown), were observed in all guinea pigs that succumbed to rLASV-WT or LASV infection. Positive LASV antigen staining was observed mainly in macrophages, epithelial cells, and/or arterial endothelial cells. Consistent with previous findings (38), mild to moderate systemic, lymphoplasmacytic, and histiocytic perivasculitis (Fig. 4B, black arrow) was observed in guinea pigs that survived rLASV-WT or LASV infection. Positive LASV antigen staining was noted in the smooth muscle cells in the tunicae mediae of large renal arteries (Fig. 4B). Overall, these results demonstrated the attenuation of rLASV(IGR/S-S) in strain 13 guinea pigs.
Protective efficacy of rLASV(IGR/S-S) against a typically lethal LASV exposure in strain 13 guinea pigs. To test whether a single dose of rLASV(IGR/S-S) could provide protection to strain 13 guinea pigs against a typically lethal LASV exposure, strain 13 guinea pigs were immunized with 10 5 PFU of rLASV(IGR/S-S) s.c. (n ϭ 4) or with phosphate-buffered saline (PBS) as an immunization control (n ϭ 5) (Fig. 5). At 30 days postimmunization (p.i.), these guinea pigs were exposed to 10 5 PFU of LASV s.c., and animals were monitored daily for clinical signs of infection throughout the study. Like the previous observation (  (Fig. 5C), or elevated temperatures (Fig. 5D). In contrast, all mock-immunized guinea pigs developed classic disease signs of LASV infection (Fig. 5B), began losing weight at day 8 p.e. (Fig. 5C), and had elevated (Ͼ40°C) temperatures at day 9 p.e. (Fig. 5D). In the mock-immunization group, 3 of 5 guinea pigs succumbed to LASV infection on days 15 to 16 p.e., one guinea pig developed mild to moderate clinical signs until the study endpoint, and one guinea pig had mild clinical signs and recovered from infection ( Fig. 5A and B).
Attenuation and protective efficacy of rLASV(IGR/S-S) against a typically lethal GPA-LASV exposure in Hartley guinea pigs. Next, a low dose of rLASV(IGR/S-S) was evaluated for efficacy to protect against LASV infection. Hartley (outbred) guinea pigs are less sensitive than strain 13 (inbred) guinea pigs to LASV infection, with 30% lethality following s.c. administration of 2.4 ϫ 10 4 PFU (37). However, intraperitoneal (i.p.) administration of 10 4 PFU of guinea pig-adapted LASV isolate Josiah (GPA-LASV) in Hartley guinea pigs results in Ն80% lethality (reference 39 and unpublished data). Given the limited availability of strain 13 guinea pigs, the Hartley model is attractive for testing the efficacy of LASV therapeutic and vaccine candidates (39,40). We immunized three groups of Hartley guinea pigs s.c. with 10 2 PFU (low dose, n ϭ 8) or 10 4 PFU (high dose, n ϭ 8) of rLASV(IGR/S-S) or with PBS (n ϭ 7). Hartley guinea pigs immunized with rLASV(IGR/S-S) did not develop any clinical signs of disease from immunization alone (Fig. 6A), including abnormal weight loss (Fig. 6B), although transient elevated tem- peratures were observed in several rLASV(IGR/S-S)-immunized animals (Fig. 6C). These data demonstrate the attenuation of rLASV(IGR/S-S) in Hartley guinea pigs.
At 30 days postimmunization (p.i.), all immunized guinea pigs were exposed to a typically lethal dose of GPA-LASV (10 4 PFU) i.p. and monitored daily for clinical signs of infection, including body weight and temperature changes, throughout the study. All guinea pigs immunized with either 10 2 or 10 4 PFU of rLASV(IGR/S-S) were protected against GPA-LASV-associated disease without having any clinical signs of disease [P ϭ 0.0442 for both doses of rLASV(IGR/S-S)] (Fig. 7A to D). In contrast, all mockimmunized guinea pigs developed clinical signs of disease, became febrile (Ͼ40°C), and began losing weight at day 8 p.e. (Fig. 7B to D). Three of seven guinea pigs from the control group succumbed to GPA-LASV infection at 13 to 16 days p.e., whereas the other four animals recovered from infection.
Viremia was assessed by RT-qPCR on days Ϫ16 and Ϫ2 preexposure [days 14 and 28 after rLASV(IGR/S-S) immunization]. At day Ϫ16 preexposure (day 14 p.i.), vRNAs were detected in 3 of 8 and 2 of 8 animals that were immunized with 10 2 and 10 4 PFU of rLASV-IGR(S-S), respectively. At day Ϫ2 preexposure (day 28 p.i.), vRNAs were not detected in blood of any of the rLASV(IGR/S-S)-immunized animals (Fig. 7E). Importantly, at day 12 after GPA-LASV exposure, vRNAs were detected in all mock-immunized guinea pigs (average 10 7 vRNA copies/ml) but not in any rLASV(IGR/S-S)-immunized animals (Fig. 7E). Likewise, high vRNA loads were detected in most of the tissues collected from mock-immunized animals that had succumbed to GPA-LASV infection (Fig. 7F, red). vRNAs were also detected in different tissues collected from mockimmunized animals that had recovered from GPA-LASV infection (Fig. 7F, orange). However, vRNAs were not detected in most tissues from rLASV(IGR/S-S)-immunized guinea pigs. The exception was a very low concentration of vRNAs (9.3 ϫ 10 vRNA copies/mg tissue, i.e., below the lower limit of quantification [ϽLLOQ]) that was detected in the spleen of one guinea pig that had been immunized with 10 2 PFU of rLASV(IGR/S-S) (Fig. 7F, blue and green). No significant histopathological lesions or positive LASV antigen staining was detected in the tissues from rLASV(IGR/S-S)immunized guinea pigs (data not shown).
Anti-LASV IgG titers were detected in rLASV(IGR/S-S)-immunized guinea pigs 2 weeks p.i. (day Ϫ16 preexposure to GPA-LASV) and continued to increase after 4 weeks (day Ϫ2 preexposure to GPA-LASV). However, anti-LASV IgG antibody titers were not significantly boosted after GPA-LASV exposure (Fig. 7G). Anti-LASV neutraliz- ing antibodies were detected in the serum of 8 of 16 guinea pigs immunized with rLASV(IGR/S-S) and exposed to GPA-LASV, suggesting that neutralization antibodies do not play an important role in the protection provided by rLASV(IGR/S-S) (Fig. 7H). Taken together, these data demonstrate that Hartley guinea pigs were completely protected from LASV infection after a single administration of a low rLASV(IGR/S-S) dose.

rLASV(IGR/S-S) is genetically stable during serial passages in cultured cells.
To develop any LAV, evaluation of the genetic stability of LAV during multiplication in infected cells is critical. Therefore, the genetic stability of rLASV(IGR/S-S) in vitro was investigated by serial passaging in Vero cells. To that end, Vero cells were inoculated (MOI ϭ 0.01) with rLASV(IGR/S-S) and at 72 h p.e., TCS were collected (passage 1 [P1]), and viral titers were determined by plaque assay. Fresh Vero cells were inoculated (MOI ϭ 0.01) with rLASV(IGR/S-S) P1, and this process was serially repeated for a total of 15 passages (P15). Endpoint titers of each passage were in the range from 10 5 to 10 7 PFU/ml (Fig. 8A). The full-length viral genome sequences collected at P1, P5, P10, and P15 were analyzed by next-generation sequencing (NGS), and single nucleotide polymorphisms (SNPs) with frequencies greater than 20% were recorded (Fig. 8B). All SNPs detected within the S-IGR sequence in the L segment remained at a low level (2.33% Ϯ 1.95) during serial passages in Vero cells. Several mutations occurred in the remainder of the L segment (e.g., T1856C, T2023C, and T4241A) and in the S segment (e.g., C1915T), which undulated in frequency during passages, suggesting that these mutations may not confer significant selective advantages. A single nucleotide polymorphism (SNP), A1691T, accumulated in the S segment during P10 to P15. At P15, SNPs T239A (Y62F) in the GPC gene of the S segment and T5350A (K580N) in the L gene were present at low frequency. To test the impact of these SNPs on viral fitness, the growth kinetics of rLASV(IGR/S-S) in Vero cells at an MOI of 0.01 were compared at P0, P5, P10, and P15. rLASV(IGR/S-S) from these passages had similar growth kinetics in Vero cells, and the viral peak titers were not statistically significantly different (P Ͼ 0.05), suggesting that the mutations described above did not impact viral fitness in cell culture (Fig. 8C). These results demonstrate that rLASV(IGR/S-S) was genetically stable up to 15 passages in Vero cells.

DISCUSSION
Fifty years after its first description, LF still poses a major public health burden in the regions where it is endemic in Western Africa. The recent LF outbreaks in Nigeria (41,42), together with the lack of licensed medical countermeasures to combat LASV infection, underscore the urgent need for LASV vaccine development (43). Epidemiological studies and nonhuman primate studies provided evidence supporting the concept that LASV-specific cell-mediated immune responses are associated with effective control of virus replication and subsequent recovery (44,45). Although passive transfer of a cocktail of engineered human neutralization antibodies (nAbs) provided protective efficacy in LASV animal models (46)(47)(48), LASV nAbs appeared at low titers in convalescing patients several months after initial infection (49,50). These discrepant results indicate that nAbs probably do not contribute to viral control and recovery from acute infection.
The mammarenavirus IGRs play critical roles in viral gene expression and infectious particle production (29). We previously demonstrated the feasibility of replacement of the L-IGR in the L segment with the S-IGR in LCMV, generating a recombinant rLCMV/IGR(S-S) (30). Although this rLCMV(IGR/S-S) grew to a relatively high titer in cultured cells in vitro, the virus was highly attenuated in vivo and provided complete protection to immunized laboratory mice against typically lethal LCMV-WT exposure (30). In the present study, we used the same attenuation strategy to generate rLASV(IGR/S-S), in which the L-IGR in the L segment was replaced with the S-IGR. Although rLASV(IGR/S-S) exhibited a moderate decrease in viral fitness in cultured cells compared to rLASV-WT, its viral titer still reached 1 ϫ 10 6 PFU/ml in Vero cells. Such a high viral titer response is important to develop a cost-effective LAV vaccine. Moreover, rLASV(IGR/S-S) was highly attenuated in both strain 13 and Hartley guinea pigs. Importantly, immunization with only 10 2 PFU of rLASV(IGR/S-S) was able to protect these guinea pigs against an otherwise lethal exposure to LASV. Consistent with results from other LASV vaccine platform studies (34,39,58,65), protection against an otherwise lethal dose of LASV in guinea pigs immunized with rLASV(IGR/S-S) occurred in the absence of detectable titers of LASV-specific neutralizing antibodies.
Genetic stability is a critical feature of an LAV candidate. A common concern about viral LAV is their potential for reversion to a more virulent phenotype, which is particularly important for RNA viruses because their error-prone replication machinery enables rapid evolution. On the basis of our findings with LCMV (30,31), we expect that the underlying mechanism of attenuation of rLASV(IGR/S-S) involves altered control of virus gene expression caused by a well-defined genetic determinant, namely, the replacement of the L-IGR by the S-IGR. This attenuated virus exhibited high genetic stability during serial passages in Vero cells, indicating that the emergence of rLASV(IGR/S-S) variants with increased virulence is unlikely.
Our previous findings obtained with LCMV (30,31), together with those we report here for LASV, indicate that replacement of the L-IGR by the S-lGR could represent a general molecular strategy for mammarenavirus attenuation. This approach has the advantage that the corresponding recombinant virus retains the same antigenic composition and, therefore, immunogenicity potential as the parental pathogenic strain.
The development of LASV reverse genetics has provided us a novel powerful tool to manipulate the LASV genome. Combined with the safety features provided by the manipulation of IGRs with other attenuation strategies such as codon-deoptimization (CD) (66,67), the generated LASV LAV could provide broad cross-protection against isolates from different LASV lineages while exhibiting optimal safety profiles. mesticated guinea pigs (Rodentia: Caviidae: Cavia porcellus Linnaeus, 1758) were housed in an animal BSL-4 (ABSL-4) laboratory, monitored daily for signs of disease, including terminal signs, and humanely euthanized at terminal stages or at study endpoint (day 42 or day 47 p.e.), and necropsies were performed as previously described (65). To reduce the numbers of guinea pigs used, animal studies performed in the present work used the same control guinea pigs as those used as indicated in Fig. 3 to 6 of studies done previously to characterize an rLASV expressing a codon-deoptimized (CD) glycoprotein precursor (GPC) gene (rLASV-GPC/CD) (65).

MATERIALS AND METHODS
Safety evaluation of rLASV(IGR/S-S) in strain 13 guinea pigs. Male and female strain 13 guinea pigs (6 to 16 weeks of age) obtained from the IRF-Frederick breeding colony were divided into three groups of 4 or 5 animals. Because of limited animal availability, the distributions into the groups were not proportional by age and sex. Guinea pigs were inoculated subcutaneously (s.c.) with 10 5 PFU of rLASV(IGR/S-S) (n ϭ 5), rLASV-WT (n ϭ 4), or LASV (n ϭ 5). At days 7, 14, 21, 28, 35, and 42 p.e., blood samples from the cranial venae cavae were collected as described previously (65) to determine viral loads and Ab responses.
Efficacy evaluation of rLASV-IGR(S-S) in Hartley guinea pigs. Twenty-three Hartley guinea pigs (Charles River Laboratories, Wilmington, MA, USA) (aged 6 to 7 weeks; male and female) were divided into three groups, distributed proportionally by age and sex. Groups were immunized s.c. with 10 2 or 10 4 PFU of rLASV(IGR/S-S) (n ϭ 8 per dose, n ϭ 16 total) or were mock-immunized with PBS (n ϭ 7). At 30 days p.i., all animals were injected intraperitoneally (i.p.) with 10 4 PFU of GPA-LASV. Blood samples were collected at day 5 before immunization, day 14 and 28 p.i., and day 12 p.e. to GPA-LASV. Viral load measurement by RT-qPCR. Whole-blood and whole-tissue samples collected at necropsy were inactivated by the use of TRIzol LS (Thermo Fisher Scientific). Total RNA was isolated using a viral RNA minikit (Qiagen, Germantown, MD, USA). Briefly, 70 l of TRIzol LS-inactivated sample was added to 280 l of Buffer AVL (Qiagen) containing carrier RNA. After the binding and washing steps, the sample was eluted into 70 l of Buffer AVE (Qiagen). Viral loads in the sample were measured using RT-qPCR as previously described (38,71). The standard curve spanned 10 8 copies/reaction (upper limit of quantification [ULOQ]) through 10 copies/reaction (lower limit of quantification [LLOQ]). Transformed data from the whole-blood and whole-tissue samples were plotted in viral RNA copies (log 10 ) per milliliter and viral RNA copies (log 10 ) per milligram of tissue, respectively.
Endpoint anti-LASV IgG enzyme-linked immunosorbent assay. To measure LASV-specific antibody titers, an IgG ELISA was developed in-house. The LASV antigens used in this assay were crude cell extracts generated from LASV-infected Vero cells. These extracts were lysed with radioimmunoprecipitation (RIPA) buffer (Cell Signaling Technology) and subjected to gamma irradiation (50 kSv) to inactivate viable virus before removal from the BSL-4 laboratory. Plates were coated with LASV-infected cell extracts diluted in coating buffer (Biolegend, San Diego, CA, USA) at a concentration of 50 ng/well, and the plates were stored at 4°C. The plates were washed six times with PBST (PBS plus 0.2% Tween 20 [Sigma-Aldrich]), and 300 l of blocking buffer (PBST plus 3% normal chicken serum [Abcam] plus 2% milk [Thermo Fisher Scientific]) was added to each well. After incubation at 37°C for 2 h, heat-inactivated irradiated plasma that had been serially diluted 2-fold was added to the plates, and the plates were kept at 4°C overnight. After the plates were washed six times with PBST, goat anti-guinea pig IgG-horseradish peroxidase (Sigma-Aldrich) was added. The plates were incubated at 37°C for 1 h and washed again with PBST. Antibody-antigen complexes were revealed by adding 3,3=,5,5=-tetramethylbenzidine substrate (Thermo Fisher Scientific) for 10 min at room temperature, and the reaction was stopped with stop solution. The absorbance was read at 450 nm on an Infinite M1000 plate reader (Tecan, Morrisville, NC, USA). The average signal from normal guinea pig plasma plus 3ϫ standard deviations was set as the cutoff value for endpoint titer measurement (72). Reciprocal serum dilutions corresponding to minimal binding were used to calculate titers.
Virus neutralization assays. Antibody neutralization titers were determined using a previously described fluorescence-based neutralization assay (35).
Histopathology and immunohistochemical staining. Tissue samples collected at necropsy were fixed in 10% NBF (formalin solution, neutral buffered) for at least 72 h before removal from the BSL-4 laboratory. Tissues were then embedded in paraffin, sectioned, mounted on glass slides, and stained with hematoxylin and eosin (H&E) following standard procedures. Immunohistochemistry (IHC) staining was performed with an anti-LASV-NP MAb (catalog no. 01-04-0104; Cambridge Biologics), followed by secondary and tertiary antibodies, and slides were examined and imaged as previously described (65).
Genetic stability assessment of rLASV(IGR/S-S) during serial passages in cell culture. To evaluate the genetic stability of rLASV(IGR/S-S) in cell culture, rLASV(IGR/S-S) was serially passaged 15 times in Vero cells. Briefly, Vero cells were infected with rLASV(IGR/S-S) in a 75-cm 2 flask (defined as passage 0 [P0]) at MOI ϭ 0.01. At 72 h p.i., TCS (P1) were collected, and virus titer was measured by plaque assay. Then, fresh Vero cells were inoculated with P1 TCS (MOI ϭ 0.01) to generate P2. This process was repeated to generate P15. P1, P5, P10, and P15 of rLASV(IGR/S-S) were inactivated and vRNA was extracted using a PureLink RNA minikit (Thermo Fisher Scientific). Each library from purified vRNA was prepared using SMARTer stranded total RNA-Seq kit v2-Pico input mammalian (TaKaRa Bio USA, Mountain View, CA, USA) according to the manufacturer's protocol and then sequenced on a HiSeq 3000 system (100-bp paired end). CLC Genomics Workbench 12 (Qiagen Digital Insights, Redwood City, CA,