Analysis of the Function of the Lymphocytic Choriomeningitis Virus S Segment Untranslated Region on Growth Capacity In Vitro and on Virulence In Vivo

Lymphocytic choriomeningitis virus (LCMV) is a prototypic arenavirus. The function of untranslated regions (UTRs) of the LCMV genome has not been well studied except for the extreme 19 nucleotide residues of both the 5′ and 3′ termini. There are internal UTRs composed of 58 and 41 nucleotide residues in the 5′ and 3′ UTRs, respectively, in the LCMV S segment. Their functional roles have yet to be elucidated. In this study, reverse genetics and minigenome systems were established for LCMV strain WE and the function of these regions were analyzed. It was revealed that nucleotides 20–40 and 20–38 located downstream of the 19 nucleotides in the 5′ and 3′ termini, respectively, were involved in viral genome replication and transcription. Furthermore, it was revealed that the other internal UTRs (nucleotides 41–77 and 39–60 in the 5′ and 3′ termini, respectively) in the S segment were involved in virulence in vivo, even though these regions did not affect viral growth capacity in Vero cells. The introduction of LCMV with mutations in these regions attenuates the virus and may enable the production of LCMV vaccine candidates.


Introduction
Lymphocytic choriomeningitis virus (LCMV) belongs to the genus Mammarenavirus, family Arenaviridae. There are two groups of Mammarenavirus, new-world and old-world arenaviruses [1]. LCMV belongs to the old-world arenavirus group, as does Lassa virus (LASV), the causative agent of Lassa fever. LCMV infects humans, causing flu-like fever, nausea, neck stiffness, headache,

Immunofocus Assay and Immunofluorescence Assay (IFA)
The infectious dose of LCMV was determined using a viral immunofocus assay. Briefly, after absorption of virus solution into Vero cells cultured in 12-well plates, cells were further cultured for 120 h at 37 • C in DMEM supplemented with 1% FBS and 100 µg/mL penicillin-streptomycin (DMEM-1FBS) with agarose (1%). The cell monolayers were then fixed with 10% formalin in phosphate buffered saline (PBS), permeabilized by incubating with 0.2% Triton X-100 in PBS, and stained with anti-LCMV-WE recombinant NP immunized rabbit serum and horseradish peroxidase (HRP)-goat anti-rabbit IgG (H+L) DS Grd (lot: 917439A, Life Technologies) [31]. Cells were then stained with Peroxidase Stain DAB Kit (Nacalai, Kyoto, Japan), and the number of stained foci was counted. For immunofluorescence assay (IFA), Alexa Fluor 488 goat anti-rabbit IgG (H+L) (Life Technologies) was used as the secondary antibody. The cells were observed to determine if they were LCMV-positive or -negative under a fluorescence microscope (BZ-9000, Keyence, Osaka, Japan).

of 24
Plasmids for the expression of LCMV-WE NP, L, Z, and GPC (referred to as pC-NP, pC-L, pC-Z, and pC-GPC, respectively) were constructed by cloning PCR amplicons of the NP, Z, and GPC genes flanked by EcoRI and NheI sites and of the L gene flanked by KpnI and NheI sites into their respective restriction enzyme sites in the pCAGGS plasmid.
A plasmid expressing GFP (referred to as pC-GFP) was constructed by cloning PCR amplicons encoding GFP genes flanked by EcoRI restriction sites into a compatible site in pCAGGS plasmid. A plasmid expressing firefly luciferase (referred to as pC-Fluc) was kindly provided by Hideki Aizaki of the National Institute of Infectious Diseases (Tokyo, Japan).

Transfection, Minigenome, and Reverse Genetics System
For the minigenome assay, 1.0 × 10 5 BHK-21 cells were seeded into wells of 24-well plates and grown to approximately 80% confluence. Cells were transfected with minigenome plasmids (0.24 µg of SMG-luc or SMG-GFP), 0.24 µg of pC-NP or pCAGGS vector (empty vector), 0.3 µg of pC-L or empty vector, and 0.03 µg of pC-Fluc; transfections were conducted using TansIT-LT1 DNA transfection reagent (Mirus Bio, Madison, WI, USA). The total amount of DNA used in transfections was maintained at 0.81 µg, and the amount of transfection reagent was maintained at a constant ratio of three volumes (µL) per amount of DNA added (µg). Transfected cells were incubated for two days at 37 • C. GFP expression was detected by fluorescence microscope; and renilla and firefly luciferase activities were measured using a Renilla Luciferase Assay System (Promega, Fitchburg, WI, USA) and a Bright-Glo Luciferase Assay System (Promega), respectively. The detailed methodology is described in the luciferase assay section.
For the reverse genetics system, 3.0 × 10 5 BHK-21 cells were seeded into 6-well plates to reach approximately 80% confluence. Cells were transfected using TansIT-LT1 DNA transfection reagent with 0.8 µg pRF-WE-SRG, 1.4 µg pRF-WE-LRG, 0.8 µg pC-NP, and 1.0 µg pC-L. The total amount of DNA transfected was 4.0 µg, and the amount of transfection reagent used was 12.0 µL. Under these transfection conditions, the cells were incubated for at least 3 days at 37 • C. The supernatant was harvested, and the infectious dose of recombinant virus was measured using an immunofocus assay. The RNA of each recombinant LCMV (rLCMV) was extracted, and viral genome sequences of all rLCMVs were confirmed to be as intended using the Sanger Sequence method.

Rescue of LCMV RNA Analogs into LCMV-Like Particles
The rescue of LCMV RNA analogs into VLPs was carried out as previously reported [11], with 3.0 × 10 5 BHK-21 cells in 2.0 mL of DMEM-5FBS seeded into 6-well plates to reach approximately 80% confluence. The cell culture medium was changed to DMEM with 2% FBS and 100 µg/mL penicillin-streptomycin (DMEM-2FBS). The cells were transfected using TansIT-LT1 DNA transfection reagent with 1.0 µg SMGs-luc or -GFP (SMG-luc or -GFP or various mutated SMGs-luc or -GFP), 0.8 µg pC-NP, 1.0 µg pC-L, 0.1 µg pC-Z, and 0.3 µg pC-GPC. As a background control for the rescue of LCMV RNA analogs into VLPs, cells transfected with 1.0 µg SMG, 0.8 µg pC-NP, and 1.0 µg pC-L, which were not expected to produce VLPs without the VLP expression plasmids (pC-Z and pC-GPC), were used (SMG-luc, (no VLPs); SMG-GFP, (no VLPs)). Cells transfected with 1.0 µg pC-GFP were used to control for transfection efficiency. The amount of transfection reagent was kept to three volumes (µL) per amount of DNA added (µg). After incubation for 48 h at 37 • C with 5% CO 2 , the expression of renilla luciferase or GFP was examined and then 1.2 mL of supernatant from each well was harvested. Fresh monolayers of BHK-21 cells seeded into 6-well plates were infected with the supernatants. Cells were incubated with the supernatants for 4 h at 37 • C and infected with helper LCMV at a multiplicity of infection (MOI) of 2 focus-forming units (FFU)/cell. Ninety hours postinfection, the cells were examined for renilla luciferase or GFP expression.

RNA Secondary Structure Prediction
To predict RNA secondary structures, the web application CENTROIDFOLD was used (https: //www.ncrna.org) [32]. The RNA sequences of the 5 and 3 termini UTRs and 50 nt lengths of ORF (GPC or NP ORF) regional RNA sequences, which were directly downstream of the UTRs, were linked and sent to the CENTROIDFOLD server. The CONTRAfold model (weight of base pairs: 2 2 ) was used to calculate base-pairing probabilities. The results of RNA secondary structure prediction were referenced in the generation of rLCMVs, in which mutations were introduced into the UTRs.

Luciferase Assay
Cells were washed with PBS and lysed with 100 µL (24-well plates) or 500 µL (six-well plates) Renilla Luciferase Assay System Lysis Buffer (Promega). To measure renilla luciferase activity, cell lysates (20 µL) were mixed with 100 µL Renilla Luciferase Assay System Substrate (Promega). To measure firefly luciferase activity, cell lysates (2 µL) were mixed with 18 µL Renilla Luciferase Assay System Lysis Buffer and 50 µL Bright-Glo Luciferase Assay System Substrate (Promega). The luminescence level was measured, and the relative light units (RLUs) of luciferase were determined using a GloMax 96 luminometer (Promega). To analyze the efficiency of viral genome packaging, the RLUs acquired from cells infected with VLPs were divided by the RLUs acquired from cells transfected with corresponding minigenome plasmids. Bar graphs were drawn using GraphPad Prism 7 software (GraphPad Software, Inc.) and statistically analyzed using an unpaired two-tailed t-test.

Viral Growth Kinetics
Viral growth kinetics of wtLCMV, recombinant non-mutated wildtype LCMV (rwtLCMV), and rLCMV with various mutations in Vero or A549 cells were analyzed. Briefly, confluent monolayers of Vero or A549 cells cultured in 12-well plates were infected with each LCMV at an MOI of 0.001 per cell. Cells were washed 3 times with DMEM-2FBS after a one-hour adsorption period at 37 • C, and 1 mL DMEM-2FBS was added to each well. Supernatant samples were collected at 0, 24, 48, 72, 96, and 120 h postinfection. The supernatants were centrifuged at 8000× g for 5 min to remove cell debris and were stored at −80 • C until the infectious dose was measured using a viral immunofocus assay. Viral growth curves of LCMVs were drawn using GraphPad Prism 7 software and statistically analyzed using two-way ANOVA.

Viral Genetic Stability Test
Confluent monolayers of Vero cells cultured in 12-well plates were infected with each LCMV at an MOI of 0.001 per cell, and infected cells were cultured at 37 • C for 3 days. Then, 200 µL of the supernatant was serially passaged to Vero cells freshly prepared in 3-or 4-day intervals. Serial passage was conducted 10 times; RNA of each rLCMV was extracted from the 10-times passaged supernatant; and viral genome sequences of all rLCMVs were analyzed by Sanger Sequencing of viral cDNAs generated by RT-PCR method [33].

Animal Experiments
Animal experiments were performed in an Animal Biosafety Level 3 laboratory. All experiments were performed in accordance with the Guidelines for Animal Experimentation of National Institute of Infectious Diseases (NIID), Japan, under the approval of the Committee on Experimental Animals at NIID (No. 116038). Specific pathogen-free 7-week-old female CBA/NSlc mice and specific pathogen-free 8-week-old female DBA/1JJmsSlc mice were purchased from Japan SLC, Inc. (Shizuoka, Japan). Mice were intraperitoneally (i.p.) infected with 1.0 × 10 2 FFU wtLCMV, rwtLCMV, mutated rLCMVs, or medium alone. Mice were monitored daily for 21 days for clinical symptoms, body weight, and survival. Mice showing more than 20% weight loss were euthanized out of ethical consideration. Blood was collected from the caudal vein of each mouse that survived to 37 days postinfection (d.p.i.) Viruses 2020, 12, 896 7 of 24 (CBA/NSlc mice) or 40 d.p.i. (DBA/1JJmsSlc mice). Following the first infection, mice that survived and showed no apparent symptoms at 40 d.p.i. were further i.p. inoculated with 1.0 × 10 3 FFU of wtLCMV and again monitored daily for 21 days for clinical symptoms, body weight, and survival. After 21 days postinfection, mice were euthanized under isoflurane deep anesthesia. Survival curves were drawn using GraphPad Prism 7 software and statistically analyzed using the log-rank (Mantel-Cox) test. The curves of body weight changes were drawn using GraphPad Prism 7 software and statistically analyzed using multiple t-tests. Discovery was determined using the two-stage linear step-up produced by Benjamini, Krieger, and Yekutieli, with Q = 5% [34].

Neutralization Assay
Blood was collected from the caudal vein of each mouse at either 37 d.p.i. or 40 d.p.i. using BD Microtainer blood collection tubes (BD, Franklin Lakes, NJ, USA) and centrifuged at 8000 × g for 5 min. Separated plasma was inactivated by heat treatment at 56 • C for 30 min. Plasma was serially diluted with DMEM-2FBS from 1:20 to 1:160 (two-fold serial dilution) and mixed with an equal volume of DMEM-2FBS containing 40-70 FFU/50 µL wtLCMV. The mixtures were incubated for 1 h at 37 • C. Vero cells cultured in 12-well plates were inoculated with 100 µL of each mixture and cultured at 37 • C for 1 h for adsorption. The cells were overlaid with 1 mL maintenance medium (Eagle's minimal essential medium containing 1% methylcellulose, 2 mM L-glutamine, 0.22% sodium bicarbonate, and 2% FBS). The plates were incubated at 37 • C in 5% CO 2 for 120 h. Cells were fixed with 10% formaldehyde, permeabilized by incubation with PBS containing 0.2% Triton X-100 (Sigma-Aldrich, St. Louis, MO, USA), and then stained with anti-LCMV-WE recombinant NP immunized rabbit serum and HRP-goat anti-rabbit IgG (H+L) DS Grd (Life Technologies) [31]. The cells were then stained with Peroxidase Stain DAB Kit (Nacalai). The number of stained foci was counted as described above, and the 50% focus reduction neutralization titer (FRNT 50 ) was measured. The FRNT 50 titers were defined as the reciprocal of the serum dilution level where the focus number became less than 50% of the control, and the focus number of the wells were inoculated with the mixture of non-plasma containing DMEM-2FBS and DMEM-2FBS containing 40-70 FFU/50 µL of wtLCMV.

Design of Plasmids for Recombinant LCMV with Mutations in the S Segment UTRs and Prediction of Their RNA Secondary Structure
The genome structure of wtLCMV S segment is shown in Figure 1. We used the CENTROIDFOLD server to predict secondary structures of the S segment UTRs ( Figure 2). In addition to the previously identified 19 nt forming the terminal panhandle structure, RNA produced from the wtLCMV genome segment (pRF-WE-SRG) also formed a double-stranded region comprising 45 and 42 nt at the 5 and 3 termini, respectively. In the terminal panhandle structures, nt 28-33 and 26-31 at the 5 and 3 termini, respectively, formed base pairs with the highest base-pairing probability in addition to the 19 base pairs at the termini. Focusing on the panhandle structure and in particular on the base pairs comprising residues 28-33 at the 5 terminus and 26-31 at the 3 terminus, we designed pRF-WE-SRGs with various mutations and predicted the secondary structures of their RNA products ( Figure 1, Supplementary Figure S1, and Tables 1 and 2). The predicted RNA secondary structure of RNA products from pRF-WE-SRG-5UTR∆20-40, pRF-WE-SRG-5UTR∆41-60, pRF-WE-SRG-5UTR∆60-77, pRF-WE-SRG-3UTR∆20-38, and pRF-WE-SRG-3UTR∆39-60 are shown in Figure S1 and Table 3. The RNAs produced from pRF-WE-SRG-5UTR∆41-60, pRF-WE-SRG-5UTR∆60-77, and pRF-WE-SRG-3UTR∆39-60 formed panhandle structures that comprised more than 36 and 33 nt at the 5 and 3 termini, respectively. The base pairs comprising the 28th to the 33rd nt at the 5 terminus and the 26th to the 31st nt at the 3 terminus showed the highest base-pairing probabilities. Only the extreme terminal panhandle of the 19 base pairs in the termini was present in the RNAs produced from pRF-WE-SRG-5UTR∆20-40 and pRF-WE-SRG-3UTR∆20-38.
Viruses 2020, 12, x FOR PEER REVIEW 9 of 27 comprised more than 36 and 33 nt at the 5′ and 3′ termini, respectively. The base pairs comprising the 28th to the 33rd nt at the 5′ terminus and the 26th to the 31st nt at the 3′ terminus showed the highest base-pairing probabilities. Only the extreme terminal panhandle of the 19 base pairs in the termini was present in the RNAs produced from pRF-WE-SRG-5UTR∆20-40 and pRF-WE-SRG-3UTR∆20-38.
To further investigate the effect of the base pairs comprising the 28th to the 33rd nt at the 5′ terminus and the 26th to the 31st nt at the 3′ terminus and their peripheral regions, various plasmids with mutations or deletions in the regions of interest (nt 20-40 at the 5′ terminus and 20-38 at the 3′ terminus) were generated ( Figure 1). The results of RNA secondary structure prediction of the RNA products of pRF-WE-SRG-UTR-comple, pRF-WE-SRG-UTR 5-3 change, pRF-WE-SRG-∆26-40, pRF-WE-SRG-∆20-25, pRF-WE-SRG-∆20-30, and pRF-WE-SRG-∆31-40 are shown in Supplementary Figure S2. A summary of the RNA secondary structure predictions is shown in Table 3. Briefly, the RNAs produced from pRF-WE-SRG-UTR-comple, pRF-WE-SRG-UTR 5-3 change, pRF-WE-SRG-∆20-25, and pRF-WE-SRG-∆20-30 formed panhandle structures with 5′ and 3′ termini, which were composed of the base pairs with high base-pairing probability, in addition to the 19 base pairs in the termini. However, the RNAs produced from pRF-WE-SRG-∆26-40 and pRF-WE-SRG-∆31-40 formed panhandle structures with 5′ and 3′ termini, but there were no base pairs with high base-pairing probability except for the extreme 19 base pairs in the termini.   The predicted RNA secondary structure of LCMV-WE S segment UTR derived from pRF-WE-SRG. The location of the 19 base pairs in the termini and the base pairs comprising the 28th-33rd nt in the 5′ terminus and the 26th-31st nt in the 3′ terminus are shown. RNA sequences of LCMV-WE S segment genome 5′-terminal and 3′-terminal UTRs and 50 nt of open reading frame (ORF) regional RNA sequences that were directly downstream of the UTRs were linked, sent to the CENTROIDFOLD server, and analyzed using the CONTRAfold model (weight of base pairs: 2 2 ). Each predicted base pair is colored with heat-color gradation from blue to red, corresponding to the basepairing probability from 0 to 1. The labels "5" and "3" indicate RNA 5′ and 3′ termini, respectively. The predicted RNA secondary structure of LCMV-WE S segment UTR derived from pRF-WE-SRG. The location of the 19 base pairs in the termini and the base pairs comprising the 28th-33rd nt in the 5 terminus and the 26th-31st nt in the 3 terminus are shown. RNA sequences of LCMV-WE S segment genome 5 -terminal and 3 -terminal UTRs and 50 nt of open reading frame (ORF) regional RNA sequences that were directly downstream of the UTRs were linked, sent to the CENTROIDFOLD server, and analyzed using the CONTRAfold model (weight of base pairs: 2 2 ). Each predicted base pair is colored with heat-color gradation from blue to red, corresponding to the base-pairing probability from 0 to 1. The labels "5" and "3" indicate RNA 5 and 3 termini, respectively. To further investigate the effect of the base pairs comprising the 28th to the 33rd nt at the 5 terminus and the 26th to the 31st nt at the 3 terminus and their peripheral regions, various plasmids with mutations or deletions in the regions of interest (nt 20-40 at the 5 terminus and 20-38 at the 3 terminus) were generated ( Figure 1). The results of RNA secondary structure prediction of the RNA products of pRF-WE-SRG-UTR-comple, pRF-WE-SRG-UTR 5-3 change, pRF-WE-SRG-∆26-40, pRF-WE-SRG-∆20-25, pRF-WE-SRG-∆20-30, and pRF-WE-SRG-∆31-40 are shown in Supplementary Figure S2. A summary of the RNA secondary structure predictions is shown in Table 3. Briefly, the RNAs produced from pRF-WE-SRG-UTR-comple, pRF-WE-SRG-UTR 5-3 change, pRF-WE-SRG-∆20-25, and pRF-WE-SRG-∆20-30 formed panhandle structures with 5 and 3 termini, which were composed of the base pairs with high base-pairing probability, in addition to the 19 base pairs in the termini. However, the RNAs produced from pRF-WE-SRG-∆26-40 and pRF-WE-SRG-∆31-40 formed panhandle structures with 5 and 3 termini, but there were no base pairs with high base-pairing probability except for the extreme 19 base pairs in the termini.
The focus morphologies of wtLCMV and rLCMVs generated are shown in Figure 3e. The size of foci was concordant with viral growth capacity features. The focus sizes of rLCMV-5UTR∆41-60, rLCMV-5UTR∆60-77, and rLCMV-3UTR∆39-60 were equivalent to those of wtLCMV and rwtLCMV. The focus sizes of rLCMV-UTR-comple and rLCMV-∆26-40 were smaller than those of wtLCMV and rwtLCMV, while the focus size of rLCMV-UTR 5-3 change was the smallest.   rLCMV-∆31-40 None ND ND ND # Viral growth efficiency of rwtLCMV in Vero cells is equal that of wtLCMV. ## Viral growth efficiency of rLCMV-3UTR∆39-60 in Vero cells is equal that of wtLCMV. * One DBA/1JJmsSlc mouse infected with rLCMV-UTR-comple showed ruffled fur and weight loss. ** All DBA/1JJmsSlc mice showed ruffled fur and weight loss. *** All DBA/1JJmsSlc mice showed ruffled fur and weight loss, and one of five mice died.

Genetic Stability of rLCMVs
To investigate the genetic stability of rLCMVs, the supernatants of rLCMV-infected cells were serially passaged in Vero cells 10 times and the genome sequence of the 10-times passaged (P10) rLCMVs were compared with the original (P0) rLCMVs. The results of genome sequence analysis of P10 rLCMVs are summarized in Table 5. Briefly, rLCMV-5UTR∆60-77, rLCMV-UTR-comple, rLCMV-UTR 5-3 change, and rLCMV-∆26-40 obtained mutations through serial passage. Most mutations accumulated in S segment UTRs or L protein ORF. One mutation which caused synonymous amino acid change was found in GPC of rLCMV-UTR 5-3 change. No mutations were observed in nucleic acid sequences of rwtLCMV, rLCMV-5UTR∆41-60, and rLCMV-3UTR∆39-60. * P0 indicates the nucleic acid sequence of original rLCMVs. ** P10 indicates the nucleic acid sequence of serially passaged rLCMVs. *** Sequences represented by "x" and "y" in the P10 column indicate that the nucleic acid sequence of this nt position was a mixture of "x" and "y" and the major nucleic acid was "x".

Acquired Immunity against LCMV in Mice Induced by Infection With rLCMVs
Neutralization titers against LCMV in sera collected from CBA/NSlc mice at 37 days after first infection and from DBA/1JJmsSlc mice at 40 days after the first infection were evaluated. The 53 sera specimens collected from rLCMV-infected mice all showed negative reactions in the focus reduction  Figure 4a,c indicate standard errors of the mean. Asterisks indicate that significant differences were observed between the mean body weight of mice infected with medium alone and that of mice infected with LCMVs, displayed in the same colors.

Acquired Immunity against LCMV in Mice Induced by Infection With rLCMVs
Neutralization titers against LCMV in sera collected from CBA/NSlc mice at 37 days after first infection and from DBA/1JJmsSlc mice at 40 days after the first infection were evaluated. The 53 sera specimens collected from rLCMV-infected mice all showed negative reactions in the focus reduction neutralization test with the exception of one specimen from an rLCMV-3UTR∆39-60-CBA/NSlc mouse. The FRNT 50 of the positive sample was 40.

Minigenome Assay and VLP Assay
The effect of various mutations or deletions in the UTRs on viral genome transcription and replication was assessed with the established minigenome assay. The function of UTRs with mutated SMGs was also analyzed. To construct SMG-luc or SMG-GFP, cDNA fragments containing the S 5′ UTR, S IGR, GFP, or renilla luciferase ORFs in an antisense orientation with respect to the 5′ UTR, and the S 3′ UTR were cloned between the murine pol I promoter and terminator of the pRF vector system (Figure 6a). Replication, transcription, and translation of SMG-luc and SMG-GFP are shown in Figure 6a. The expression of luciferase or GFP was only observed in BHK-21 cells when these were

Minigenome Assay and VLP Assay
The effect of various mutations or deletions in the UTRs on viral genome transcription and replication was assessed with the established minigenome assay. The function of UTRs with mutated SMGs was also analyzed. To construct SMG-luc or SMG-GFP, cDNA fragments containing the S 5 UTR, S IGR, GFP, or renilla luciferase ORFs in an antisense orientation with respect to the 5 UTR, and the S 3 UTR were cloned between the murine pol I promoter and terminator of the pRF vector system (Figure 6a). Replication, transcription, and translation of SMG-luc and SMG-GFP are shown in Figure 6a. The expression of luciferase or GFP was only observed in BHK-21 cells when these were co-transfected with SMG-luc (or SMG-GFP) and both pC-NP and pC-L (Figure 6b and Supplementary Figure S3). The signal-to-noise ratio for the established minigenome assay was 7.9 × 10 2 .
We also evaluated the packaging efficiency of the viral genome RNA analogs into VLPs ( Figure  7c and Supplementary Figure S4b). The luciferase expression in cells infected with VLPs, which encapsulated RNA products derived from SMG, SMG-5UTR∆41-60-luc, SMG-5UTR∆60-77-luc, or SMG-3UTR∆39-60-luc, were approximately equal to each other. Luciferase expression in the cells infected with VLPs that encapsulated RNA products derived from SMG-UTR-comple-luc was significantly lower than that in the cells infected with VLPs that encapsulated RNA products derived from SMG-luc but significantly higher than that of the cells treated with a supernatant derived from cells transfected with SMG-luc without VLP expression plasmids (pC-Z and pC-GPC) (p = 0.0003) (Figure 7c).
(a) Cont. Figure 6.  (NP). vRNAs are produced in the presence of L protein (L). The encapsidated cRNA is transcribed into reporter gene mRNA. Finally, transcribed mRNA is translated by host-cell machinery to produce reporter proteins (luciferase or GFP). (b) BHK-21 cells were transfected with minigenome plasmids SMG-luc, either pC-NP or pCAGGS, either pC-L or pCAGGS, and pC-Fluc. The transfected cells were incubated for 2 days at 37 °C, and then renilla luciferase activity was measured using the Renilla Luciferase Assay System and firefly luciferase activity was measured using the Bright-Glo Luciferase Assay System. (**** p < 0.0001) Error bars indicate standard deviations. The experiments were performed three times independently. (a) Cont. Figure 7.

Discussion
Here, we described an LCMV-WE reverse genetics system and a minigenome system (Figures  3a and 6). A polymerase-I-driven LCMV-ARM reverse genetics system has been previously reported with analyses of LCMV-ARM [13,15]. The identities of the nucleotide sequences between the S and L segments of LCMV-WE and LCMV-ARM are 85% and 82%, respectively, and several reports have We also evaluated the packaging efficiency of the viral genome RNA analogs into VLPs (Figure 7c and Supplementary Figure S4b). The luciferase expression in cells infected with VLPs, which encapsulated RNA products derived from SMG, SMG-5UTR∆41-60-luc, SMG-5UTR∆60-77-luc, or SMG-3UTR∆39-60-luc, were approximately equal to each other. Luciferase expression in the cells infected with VLPs that encapsulated RNA products derived from SMG-UTR-comple-luc was significantly lower than that in the cells infected with VLPs that encapsulated RNA products derived from SMG-luc but significantly higher than that of the cells treated with a supernatant derived from cells transfected with SMG-luc without VLP expression plasmids (pC-Z and pC-GPC) (p = 0.0003) (Figure 7c).

Discussion
Here, we described an LCMV-WE reverse genetics system and a minigenome system (Figures 3a  and 6). A polymerase-I-driven LCMV-ARM reverse genetics system has been previously reported with analyses of LCMV-ARM [13,15]. The identities of the nucleotide sequences between the S and L segments of LCMV-WE and LCMV-ARM are 85% and 82%, respectively, and several reports have described characteristic differences between these strains [19,22,26,27]. Our LCMV-WE reverse genetics system will enable us to elucidate the mechanistic significance of these differences from a virological perspective in future studies. For instance, it became possible to generate chimeric LCMV between LCMV-WE and LCMV-ARM and to determine which viral proteins were responsible for the pathogenic difference in mice between LCMV-WE and LCMV-ARM. Furthermore, the infection of NHPs with rLCMVs might provide further understanding of the arenavirus-mediated pathogenic mechanisms associated with viral hemorrhagic fever [23][24][25][26][27].
In this study, we focused on the S segment UTRs and performed RNA secondary structure prediction for these ( Figure 2). The 28th to 33rd nt in the 5 terminus and the 26th to 31st nt in the 3 terminus formed base pairs with the highest base-pairing probability ( Table 3). The base pairs comprising the 20th to the 40th nt in the 5 terminus and the 20th to the 38th nt in the 3 terminus, which included nt 28-33 in the 5 terminus and 26-31 in the 3 terminus, greatly affected viral propagation (Supplementary Figure S2, Figure 3, and Table 4). The successful generation of rLCMV-UTR-comple, rLCMV-UTR 5-3 change, and rLCMV-∆26-40 and the failure to generate rLCMV-∆20-25, rLCMV-∆20-30, and rLCMV-∆31-40 suggested that both the conformation of the panhandle structure and the specific nucleotide sequences of these base pairs may be important for the recognition by the L protein. However, further studies will be needed to confirm or refute this hypothesis. The predicted RNA secondary structure of the rLCMV-∆26-40 S segment lacked base pairs with high base-pairing probability with the exception of the 19 base pairs in the extreme termini. The reason why rLCMV-∆26-40 successfully propagated was not clarified in this study, but the conservation of base pairs comprising nt 20-25 in both termini and the RNA conformation may be advantageous to propagation.
The viral growth capacity and focus size of rLCMV-5UTR∆41-60, rLCMV-5UTR∆60-77, and rLCMV-3UTR∆39-60 in Vero cells were equivalent or similar to those of wtLCMV and rwtLCMV (Figure 3b,e and Table 4). This suggests that the conservation of the base pairs comprising the 20-40 nt region in the 5 terminus and the 20-38 nt region in the 3 terminus was not essential but did have a considerable effect on viral propagation, though the other UTRs (nt 41-77 in the 5 terminus and nt 39-60 in the 3 terminus) did not affect viral propagation in vitro.
The mutations or deletions in UTRs of the S segment of LCMV attenuated virulence in vivo ( Figure 4 and Table 4). In particular, mutations in the base pairs of the 20-40 nt region in the 5 terminus and the 20-38 nt region in the 3 terminus (rLCMV-UTR-comple, rLCMV-UTR 5-3 change, and rLCMV-∆26-40) caused major in vivo attenuation of LCMV. Low efficiency of viral propagation in vitro may lead to viral attenuation in vivo. Significant differences in the survival curves were observed between rwtLCMV-CBA/NSlc mice and rLCMV-3UTR∆39-60-CBA/NSlc mice, between rwtLCMV-DBA/1JJmsSlc mice and rLCMV-5UTR∆41-60-DBA/1JJmsSlc mice, and between rwtLCMV-DBA/1JJmsSlc mice and rLCMV-5UTR∆60-77-DBA/1JJmsSlc mice. There were no significant differences in the survival curves between rwtLCMV-CBA/NSlc mice and rLCMV-5UTR∆60-77-CBA/NSlc mice or rLCMV-5UTR∆41-60-CBA/NSlc mice and between rwtLCMV-DBA/1JJmsSlc mice and rLCMV-3UTR∆39-60-DBA/1JJmsSlc mice. However, the hazard ratios obtained in this study suggested that rLCMV-5UTR∆41-60, rLCMV-5UTR∆60-77, and rLCMV-3UTR∆39-60 possessed less virulence compared with wtLCMV and rwtLCMV. Reports have suggested that UTRs in other viruses were involved in virulence, where the variable region of the 3 UTR of the tick-borne encephalitis virus genome was associated with virulence in mice [35,36]. Additionally, UTRs of the picornavirus genome were reported to affect viral infection and host innate immunity [37]. In our study, the role of nt 41-77 in the 5 UTR and nt 39-60 in the 3 UTR was not elucidated. However, the results of the mouse experiments indicate these UTRs may affect host innate immunity or other factors in vivo. CBA/NSlc mice are widely used in infectious disease and immunity research [38][39][40]. A single amino acid mutation in Bruton's tyrosine kinase reduces the function of B cells in CBA/NSlc mice [41,42]. On the other hand, DBA/1JJmsSlc mice are used in arthritis and immunity research and their B cell function is competent [43][44][45]. Although one of five rLCMV-3UTR∆39-60-DBA/1JJmsSlc mice lost body weight and died, none of the rLCMV-3UTR∆39-60-CBA/NSlc mice showed any apparent symptoms and survived the first infection. These results suggest the 39-60 nt region in the 3 UTR may affect innate immunity; the low growth efficiency of rLCMV-3UTR∆39-60 in A549 cells, in which the type I interferon pathway is intact, supports this hypothesis (Figure 3c). However, further investigation is required to make a firm conclusion about this.
The results of the secondary infection with a lethal dose of wtLCMV in mice that survived first infection suggest that protective immunity was induced in these mice ( Figure 5 and Table 4). rLCMV-UTR-comple-, rLCMV-UTR 5-3 change-, and rLCMV-∆26-40-DBA/1JJmsSlc mice developed symptoms following the secondary infection, and one mouse died at 8 d.p.i. Conversely, the rLCMV-5UTR∆41-60-, rLCMV-5UTR∆60-77-, or rLCMV-3UTR∆39-60-DBA/1JJmsSlc mice lacked any clinical symptoms. These results indicate that mice infected with rLCMVs that propagate in a similar way to wtLCMV induced higher protective immunity than mice infected with rLCMVs that have a low capacity for propagation. Taking into consideration the reduced function of B cells in CBA/NSlc mice, neutralizing antibodies against LCMV were still not induced in most mice. This is consistent with previous studies, where the generation of anti-LCMV neutralizing antibodies was not detectable between 60 and 120 d.p.i. following LCMV-WE infection in mice and where CD8 + T-cell-mediated cytotoxicity played a key role in LCMV-WE infection [46].
Deletions in the UTRs of the Bunyamwera virus (BUNV) can attenuate viral growth properties, but these were recovered following BUNV serial passage in vitro [47]. Amino acid changes were found in the C-terminal domain of the BUNV L protein after the serial passage, indicating that these changes may be involved in the evolution of the L polymerase, which enabled more efficient recognition of the deleted UTRs. In our study, deletions or other mutations in the base pairs comprising the 20-40 nt region in the 5 terminus and the 20-38 nt region in the 3 terminus greatly hampered viral genome transcription and replication (Figure 7a and Supplementary Figure S4a). This may be the cause of the non-production of rLCMVs derived from pRF-WE-SRG-5UTR∆20-40 and pRF-WE-SRG-3UTR∆20-38. These results supported the notion that the panhandle structure composed of the base pairs targeted in these regions was involved in the recognition site of the L protein. In the genetic stability test, amino acid changes were mainly accumulated in the RdRp region and PB2-like region of the L protein and in the UTRs of the rLCMVs that showed low growth efficiency in Vero cells (rLCMV-UTR-comple, rLCMV-UTR 5-3 change, and rLCMV-∆26-40). Therefore, the L protein of these rLCMVs may have adapted its conformation to the mutated UTRs as indicated in the BUNV research. Deletions in the other UTRs (the 5 UTR 41-60 region, the 5 UTR 60-77 region, and the 3 UTR 39-60 region) did not affect viral genome transcription and replication efficiency in the minigenome assay or viral genome packaging efficiency in the VLP assay (Figure 7 and Supplementary Figure S4). These results suggest that UTRs, except for those base pairs composed of 40 nt in the 5 terminus and 38 nt in the 3 terminus, were not involved in viral genome transcription, replication, or packaging. Furthermore, the minigenome assay data suggest a low possibility for the conformational adaptation of L protein to the mutated UTRs in rLCMV-5UTR∆41-60, rLCMV-5UTR∆60-77, and rLCMV-3UTR∆39-60. In fact, no genetic mutations were observed in rLCMV-5UTR∆41-60 and rLCMV-3UTR∆20-38 and only one mutation was found in rLCMV-5UTR∆60-77 after serial passage.
Although luciferase expression levels from SMG-3UTR∆20-38-luc were higher than those from SMG-UTR-comple-luc, the packaging efficiency of the RNA products from SMG-UTR-comple-luc was significantly higher compared with that of SMG-3UTR∆20-38-luc (Figure 7a,c). These results suggest that RNA products derived from SMG-UTR-comple-luc were certainly packaged into VLPs and carried to the next cells but at low efficiency and that the sequence and/or conformation of UTRs also affected viral genome packaging efficiency.
In summary, we found that, in addition to the 19 nucleotide base pairs in both termini of the S segment, the base pairs between the 20-40 nt in the 5 terminus and the 20-38 nt in the 3 terminus of the S segment were predicted to form panhandle structures with high base-pairing probabilities. These regions affected viral propagation as they were heavily involved in viral genome transcription and replication in terms of their conformation and/or nucleotide sequence. Furthermore, our findings suggest that the other UTRs were involved in viral pathogenicity in vivo, though they did not affect the efficiency of viral genome transcription, replication, and packaging. The mechanism of attenuation of the virus in vivo remains to be determined, but the attenuation of LCMV without amino acid changes in component proteins might help us develop new vaccines.

Conclusions
In addition to the 19 nucleotide base pairs in both termini of the LCMV S segment, the base pairs between the 20-40 nt in the 5 terminus and the 20-38 nt in the 3 terminus of LCMV S segment were predicted to form panhandle structures and were heavily involved in viral genome transcription and replication. Furthermore, the other LCMV S segment UTRs were involved in viral pathogenicity in vivo, though they did not affect the efficiency of viral genome transcription, replication, and packaging. The mechanism of attenuation of the virus in vivo remains to be determined, but the attenuation of LCMV might help us develop new vaccines.
Supplementary Materials: The following are available online at http://www.mdpi.com/1999-4915/12/8/896/s1, Figure S1: Prediction of RNA secondary structures of mutated LCMV strain WE (LCMV-WE) S segment UTRs which have 18-22 nt deletions in the 5 or 3 termini., Figure S2: Prediction of RNA secondary structures of mutated LCMV strain WE (LCMV-WE) S segment UTRs which have deletions or mutations between the 20th-40th nt in the 5 terminus or between the 20th-38th nt in the 3 terminus. Figure S3: Establishment of a GFP expressing minigenome system for LCMV strain WE (LCMV-WE). Figure S4: Evaluation of the efficiency of viral genome transcription, replication, and packaging in virus-like particles (VLPs).