Identification of Single Amino Acid Changes in the Rift Valley Fever Virus Polymerase Core Domain Contributing to Virus Attenuation In Vivo

Abstract

Rift Valley fever (RVF) is an arboviral zoonotic disease affecting many African countries with the potential to spread to other geographical areas. RVF affects sheep, goats, cattle and camels, causing a high rate of abortions and death of newborn lambs. Also, humans can be infected, developing a usually self-limiting disease that can turn into a more severe illness in a low percentage of cases. Although different veterinary vaccines are available in endemic areas in Africa, to date no human vaccine has been licensed. In previous works, we described the selection and characterization of a favipiravir-mutagenized RVFV variant, termed 40Fp8, with potential as a RVF vaccine candidate due to the strong attenuation shown in immunocompromised animal models. Compared to the parental South African 56/74 viral strain, 40Fp8 displayed 7 amino acid substitutions in the L-protein, three of them located in the central region corresponding to the catalytic core of the RNA-dependent RNA polymerase (RdRp). In this work, by means of a reverse genetics system, we have analyzed the effect on virulence of these amino acid changes, alone or combined, both in vitro and in vivo. We found that the simultaneous introduction of two changes (G924S and A1303T) in the heterologous ZH548-RVFV Egyptian strain conferred attenuated phenotypes to the rescued viruses as shown in infected mice without affecting virus immunogenicity. Our results suggest that both changes induce resistance to favipiravir likely associated to some fitness cost that could be the basis for the observed attenuation in vivo. Conversely, the third change, I1050V, appears to be a compensatory mutation increasing viral fitness. Altogether, these results provide relevant information for the safety improvement of novel live attenuated RVFV vaccines.

Keywords: RVFV; attenuation; live vaccines; mutagenic drugs; reverse genetics; viral polymerase.

Figure 1

Kinetics of growth in mammalian cells: Vero (A) and HEK293T (B) cells were infected at a MOI of 0.01 with the indicated rRVF viruses ( Table 1 ). Supernatants collected at 48 and 72 hpi were titrated onto Vero cells by a plaque assay. Experiments were performed in duplicate. Data shown correspond to a single representative assay. The range of final yield between both wild-type polymerase-control viruses C1 and G1 has been shaded. (C) Nucleotide diversity bars per sequenced sample (left) and total number of variants annotated per sample (right). No data could be retrieved for DM#6.

Figure 2

Growth in cultured cells of RVFVs rA2 and rB3 carrying substitutions G924S & A1303T in the RdRp. Viral yield at 24, 48 and 72 hpi after infection at a MOI of 0.01 in (A) Vero and (B) HEK293T cells. Although coming from an independent experiment, a shade has been depicted to allow for a comparison with results shown in Figure 1 . In HEK293T infections, rZH548-P82L virus 2B7 was included for a comparison. Experiments were performed in duplicate. (C) Titer of supernatants collected at 72 hpi in serial passages (1 to 5) on Vero cells at a MOI of 0.01. Viral yields were determined by a plaque assay on Vero cell monolayers. (D) Sequence analysis of the RT-PCR product of viral RNAs recovered after 5 passages on Vero cells. The three nucleotide positions under study, 2788, 3166 and 3925, are shown. In the upper line, the wt sequence (rZH548) on the codons corresponding to residues 924, 1050 and 1303 is highlighted.

Figure 3

Susceptibility of rRVFVs to mutagenic drugs in Vero cells. Reduction of viral yield in the presence of different concentrations of the mutagenic drugs favipiravir (A) and ribavirin (B). Vero cells were infected by duplicate at a MOI of 0.1 with the rZH548_L mutant viruses under study rA2 and rB3, and also with virus 2B7 displaying the substitution NSs[P82L] also present in rA2, and with the corresponding wt virus rZH548. Supernatants were collected at 72 hpi and titrated by plaque assay, and percentages of viral titers upon treatment were calculated. Geometric mean of the values obtained in at least two independent experiments is represented. Percentages of viral yield are represented as log10 in order to highlight the differences at the highest concentrations tested. The depicted dot-line corresponds to a reduction of 90% in viral yield. For statistical analysis each L-mutant RVFV was compared to its corresponding L-wt virus: rA2 vs 2B7 and rB3 vs rZH548. *p < 0,05; **p < 0,005 (multiple T-test, Holm-Sidak method).

Figure 4

Analysis of the in vivo pathogenicity of the rZH548-L mutant viruses in IFNAR KO mice. 15-week-old female mice (n = 5) were inoculated with 10 or 500 plaque-forming units (pfu) of the RVFVs rA2 and rB3. G1 (rZH548ΔNSs::GFP) was included as reference control. (A) Survival rates. (B) Viremia determined by RT-qPCR on EDTA-blood samples collected at day 3 pi. Samples giving a Cq (quantification cycle) value under the detection level of the assay (37) are arbitrarily represented as 45. The correlation of Cq data with pfu equivalents (Moreno et al., 2020) is indicated in the right Y axis. Statistical analysis was only performed on results obtained from the groups inoculated with 500 pfu. The corresponding survival curves analyzed by log-rank (Mantel-Cox) rendered p values of p = 0.0047 (A2 vs G1); p = 0.0016 (B3 vs G1) and p = 0.0145 (A2 vs B3). *p < 0,05; **p < 0,005.

Figure 5

Analysis of the immunogenicity and vaccine efficacy of the rZH548-L mutant viruses in 129 (wt) mice. Groups of 4-month-old mice (n = 6, including male and female equally distributed) were inoculated with 500 pfu of rA2 and rB3 viruses. C1 (rZH548) and 2B7 (rZH548-P82L) were included for a comparison. 3 weeks later survivors were challenged with a lethal dose of 1000 pfu of C1 (rZH548). (A) Survival rates after first inoculation. (B) Percentage weight variation, geometric mean per group. The area between 97.5 and 102.5% has been shaded. (C) Viremia determined by RT-qPCR on EDTA blood samples collected at day 3 pi. Samples giving a Cq (quantification cycle) value under the detection level of the assay (37) are arbitrarily represented as 45. The correlation of Cq data with pfu equivalents (Moreno et al., 2020) is indicated in the right Y axis. (D) Neutralizing antibodies at day 12 pi as determined by a plaque reduction assay. (E) Viremia determined by RT-qPCR on EDTA blood samples collected at day 3after challenge. (F) Antibodies to N protein by ELISA both pre- and post-challenge (small and large symbols respectively).

Figure 6

Location of the three positions under study in the representation of RVFV L-protein by PyMOL, as defined by the cryo-EM model (accession code 7EEI, (Wang et al., 2021).