Mutational Analysis of the Rift Valley Fever Virus Glycoprotein Precursor Proteins for Gn Protein Expression

The Rift Valley fever virus (RVFV) M-segment encodes the 78 kD, NSm, Gn, and Gc proteins. The 1st AUG generates the 78 kD-Gc precursor, the 2nd AUG generates the NSm-Gn-Gc precursor, and the 3rd AUG makes the NSm’-Gn-Gc precursor. To understand biological changes due to abolishment of the precursors, we quantitatively measured Gn secretion using a reporter assay, in which a Gaussia luciferase (gLuc) protein is fused to the RVFV M-segment pre-Gn region. Using the reporter assay, the relative expression of Gn/gLuc fusion proteins was analyzed among various AUG mutants. The reporter assay showed efficient secretion of Gn/gLuc protein from the precursor made from the 2nd AUG, while the removal of the untranslated region upstream of the 2nd AUG (AUG2-M) increased the secretion of the Gn/gLuc protein. Subsequently, recombinant MP-12 strains encoding mutations in the pre-Gn region were rescued, and virological phenotypes were characterized. Recombinant MP-12 encoding the AUG2-M mutation replicated slightly less efficiently than the control, indicating that viral replication is further influenced by the biological processes occurring after Gn expression, rather than the Gn abundance. This study showed that, not only the abolishment of AUG, but also the truncation of viral UTR, affects the expression of Gn protein by the RVFV M-segment.


Introduction
Rift Valley fever (RVF) is a mosquito-borne zoonotic disease affecting humans and ruminants. The disease was originally endemic to sub-Saharan Africa, but it has since spread to Egypt, Madagascar, Saudi Arabia, and Yemen [1,2]. RVF causes a high-rate of abortion in sheep, cattle, and goats, and hemorrhagic fever, encephalitis, or blindness in humans [3]. The mortality rate of RVF patients is considered to be less than 0.5% to 1% [3,4]. However, RVF outbreaks have been known to involve a large number of patients: For example, 20,000 to 200,000 infections and 600 deaths in Egypt in [1977][1978]. Floodwater Aedes mosquitoes can transovarially transmit RVFV [5]. These eggs are resistant to draught, and flooding, due to heavy rainfall, facilitates the hatching of infected eggs [1]. RVF is an important public health and agricultural concern, and vaccination of susceptible animals is important to minimize the spread of disease [6]. Though live-attenuated RVF vaccines are available for veterinary  [14,16,28]. The 78 kD protein and Gc are generated from the 78 kD-Gc precursor, while NSm, Gn, and Gc are made from the NSm-Gn-Gc precursor; (B) the pre-Gn region was fused to the gLuc ORF lacking the intrinsic signal peptide, which allows for secretion from the cell via Gn signal peptide. The Gn/gLuc precursor-1 makes a chimeric protein consisting of the pre-Gn region and gLuc, while the Gn/gLuc precursor-2 generates NSm and the Gn/gLuc fusion protein.
Little information is available about the consequence of pre-Gn region mutagenesis in terms of viral phenotypes, other than the expression of 78 kD or NSm/NSm'. We hypothesized that an optimization of the M-segment pre-Gn region increases the secretion of Gn, which in turn, will lead to an increase in viral titer. We aimed to analyze the consequences of the abolishment of the 1st, 2nd, 3rd, 4th, and/or 5th AUG, in terms of Gn secretion and viral phenotype. We established a reporter assay, in which Gaussia luciferase (gLuc), fused to the N-terminal region of Gn, is expressed from the pre-Gn region of the RVFV M-segment. In the reporter assay, the Gn/gLuc fusion protein is cleaved from the Gn/gLuc precursor proteins made from either the 1st or 2nd AUG, while the 3rd, 4th, or 5th AUG can serve as surrogates for precursor production, as shown in Figure 1B. Normally, gLuc  [14,16,28]. The 78 kD protein and Gc are generated from the 78 kD-Gc precursor, while NSm, Gn, and Gc are made from the NSm-Gn-Gc precursor; (B) the pre-Gn region was fused to the gLuc ORF lacking the intrinsic signal peptide, which allows for secretion from the cell via Gn signal peptide. The Gn/gLuc precursor-1 makes a chimeric protein consisting of the pre-Gn region and gLuc, while the Gn/gLuc precursor-2 generates NSm and the Gn/gLuc fusion protein.
Little information is available about the consequence of pre-Gn region mutagenesis in terms of viral phenotypes, other than the expression of 78 kD or NSm/NSm'. We hypothesized that an optimization of the M-segment pre-Gn region increases the secretion of Gn, which in turn, will lead to an increase in viral titer. We aimed to analyze the consequences of the abolishment of the 1st, 2nd, 3rd, 4th, and/or 5th AUG, in terms of Gn secretion and viral phenotype. We established a reporter assay, in which Gaussia luciferase (gLuc), fused to the N-terminal region of Gn, is expressed from the pre-Gn region of the RVFV M-segment. In the reporter assay, the Gn/gLuc fusion protein is cleaved from the Gn/gLuc precursor proteins made from either the 1st or 2nd AUG, while the 3rd, 4th, or 5th AUG can serve as surrogates for precursor production, as shown in Figure 1B. Normally, gLuc encodes an intrinsic signal sequence, which allows it to be secreted from expressed cells [29]. In our reporter assay, we deleted the intrinsic signal sequence for gLuc so that the cleavage of Gn/gLuc fusion proteins occurs only through the Gn signal sequence, and subsequently, the Gn/gLuc fusion proteins are secreted from expressed cells. The Gn/gLuc fusion proteins produced from this construct do not accumulate in the Golgi, due to a lack of the Golgi retention signal at the C-terminus of Gn [30], and the secretion occurs through the endocytic recycling pathway from the Golgi to plasma membrane (e.g., Rab8-positive vesicles) [31]. Although the level of Gn/gLuc fusion protein secretion does not predict virion release efficiency, as RVFV Gn/Gc accumulates in the Golgi, we aimed to measure relative expression level of Gn protein from precursor proteins made from RVFV M-segment. We analyzed the relative secretion of the Gn/gLuc fusion proteins into the culture supernatant among different AUG (Met) to CUC (Leu) substitution mutants. Furthermore, recombinant MP-12 (rMP-12) encoding the AUG mutations, which can secrete distinct levels of Gn, were rescued by reverse genetics, and their phenotypes were characterized. Our study provides fundamental information for the consequences of mutagenesis in the pre-Gn region of the M-segment, and will support the understanding of current and future studies using NSm or 78 kD knockout mutants.

Plasmids
The pCAGGS-PreGn-gLuc-SF plasmid was made as follows: Synthetic DNA was created (gBlocks, Integrated DNA Technologies, Coralville, IA, USA) for the RVFV MP-12 M-segment (nt. 1-533) fused with gLuc (lacking nt.1-51 to remove an intrinsic signal sequence for gLuc), and two tandem Strep-tags, and a Flag-tag. The DNA fragment was cloned into pCAGGS plasmid by Gibson assembly mastermix (New England BioLabs, Ipswich, MA, USA). To introduce mutations in pCAGGS-PreGn-gLuc-SF plasmid, the DNA fragment (nt. 1-830) was first transferred to a pProT7 plasmid. Then, site-directed mutagenesis was performed using the pProT7 plasmid back-bone, before cloning the insert sequence into a pCAGGS plasmid. Corresponding AUGs were replaced with CUC (Leu) to abolish the AUG codon.

Western Blotting
Cells were suspended in 2x SDS sample buffer, and boiled for 10 min. Samples were separated by SDS-polyacrylamide gel electrophoresis (SDS-PAGE) under reducing conditions. Western blot was performed as described previously [35]. Anti-Flag M2 antibody (Sigma-Aldrich, St. Louis, MO, USA), and anti-actin antibody (I-19, Santa Cruz Biotechnology, Inc., Dallas, TX, USA) were used.

Measurement of Plaque Sizes
Plaque images of rMP-12, ∆2 + 3, or the AUG2-M mutants, formed in VeroE6 cells in 6-well plates, were incorporated by a scanner, and the diameters (mm) of small and large plaques (n = 10 each) were measured using Adobe Photoshop Element version 7.0 [36]. The average and standard errors were plotted onto the graph using GraphPad Prism version 6.05 [37].

Statistical Analysis
Statistical analysis was performed using GraphPad Prism version 6.05. For the gLuc/cLuc values normalized to parental construct value in Figures 2 and 3 or virus titers in Figure 4, arithmetic means of log 10 values were analyzed by one-way ANOVA followed by Tukey's multiple comparisons test.

Measurement of Plaque Sizes
Plaque images of rMP-12, Δ2 + 3, or the AUG2-M mutants, formed in VeroE6 cells in 6-well plates, were incorporated by a scanner, and the diameters (mm) of small and large plaques (n = 10 each) were measured using Adobe Photoshop Element version 7.0 [36]. The average and standard errors were plotted onto the graph using GraphPad Prism version 6.05 [37].

Statistical Analysis
Statistical analysis was performed using GraphPad Prism version 6.05. For the gLuc/cLuc values normalized to parental construct value in Figures 2 and 3, or virus titers in Figure 4, arithmetic means of log10 values were analyzed by one-way ANOVA followed by Tukey's multiple comparisons test.   The ratio of gLuc to cLuc (control plasmid) was normalized to that of parental pCAGGS-PreGn-gLuc-SF. The graph represents the mean + the standard error of three independent experiments. Asterisks represent statistically significant differences (one-way ANOVA followed by Tukey's multiple comparisons test, * p < 0.05, ** p < 0.01). Asterisks shown on error bars represent the comparison with compared to the Δ4 + 5 mutant.  (B) the extracellular Gn/gLuc fusion proteins were measured using the culture supernatant of transfected cells. The ratio of gLuc to cLuc (control plasmid) was normalized to that of parental pCAGGS-PreGn-gLuc-SF. The graph represents the mean + the standard error of three independent experiments. Asterisks represent statistically significant differences (one-way ANOVA followed by Tukey's multiple comparisons test, * p < 0.05, ** p < 0.01). Asterisks shown on error bars represent the comparison with compared to the ∆4 + 5 mutant. The ratio of gLuc to cLuc (control plasmid) was normalized to that of parental pCAGGS-PreGn-gLuc-SF. The graph represents the mean + the standard error of three independent experiments. Asterisks represent statistically significant differences (one-way ANOVA followed by Tukey's multiple comparisons test, * p < 0.05, ** p < 0.01). Asterisks shown on error bars represent the comparison with compared to the Δ4 + 5 mutant.

The Gn/gLuc Precursor-1 Does Not Efficiently Generate Gn/gLuc Fusion Proteins
293 cells were co-transfected with pSV40-CLuc (transfection control), and pCAGGS-PreGn-gLuc-SF or the AUG mutant plasmids. At 36 h post transfection, cell lysates were collected and the intracellular expression of Gn/gLuc proteins or precursor proteins was analyzed by Western blot using anti-Flag antibody. Culture supernatants were used to measure the level of secreted extracellular Gn/gLuc fusion proteins by reporter assay. The cLuc protein encodes an intrinsic signal peptide and is secreted into the secretory pathway after expression. Thus, secreted cLuc served as a control to measure the secretion of the Gn/gLuc fusion proteins from transfected cells. Western blot using anti-Flag antibody was performed to confirm the expression of precursor proteins and Gn/gLuc fusion proteins ( Figure 2A). As expected, the Gn/gLuc precursor-1 disappeared after the abolishment of the 1st AUG. Similarly, the Gn/gLuc precursor-2 disappeared when the 2nd AUG was abolished. When the 2nd AUG was abolished, the Gn/gLuc precursor-3 appeared (i.e., ∆2 or ∆1 + 2). The Gn/gLuc precursor-4 and -5 were indistinguishable from the cleaved Gn/gLuc fusion protein based on band migrations. When the 1st, 2nd, 3rd, 4th, and 5th AUGs were abolished, no Gn/gLuc fusion proteins were detected. On the other hand, when the 2nd, 3rd, 4th, and 5th AUGs were abolished, the band intensity of the Gn/gLuc fusion protein was largely decreased but still detectable in the Western blot. Next, both gLuc and cLuc activities in culture supernatants were measured. We calculated the ratio of gLuc to cLuc, and the value was normalized to that of parental pCAGGS-PreGn-gLuc-SF ( Figure 2B). When the 1st, 2nd, 3rd, 4th, and 5th AUGs were all abolished, no gLuc activity (0.79% compared to parental 100%) could be measured in the supernatant. On the other hand, when the 2nd, 3rd, 4th, and 5th AUGs were abolished, a decreased level of gLuc activity (7.7% compared to parental 100%) was detected in the supernatant. These results indicated that the precursor protein synthesized from the 1st AUG can also generate Gn, though at a decreased level.

The Viral Untranslated Region Sequence, Upstream of the 4th or 5th AUG, Affects Efficient Generation of Gn/gLuc Fusion Proteins
Spik et al. previously showed that the cloned open reading frame (ORF) of Gn-Gc (starting from the 4th AUG), without an upstream viral untranslated region (UTR), expresses slightly higher Gn/Gc than the cloned ORF of the NSm-Gn-Gc precursor (starting from the 2nd AUG) [38]. Thus, we suspected that the decrease in the gLuc activity from the Gn/gLuc precursor-4 or -5 occurred due to the presence of the UTR upstream of the 4th or 5th AUG. Therefore, we truncated the upstream viral UTR sequence from pCAGGS-PreGn-gLuc-SF ( Figure 3A). The plasmids, AUG2-M, AUG3-M, AUG4-M, or AUG-5-M, generate the Gn/gLuc precursor-2, -3, -4, or -5, respectively, and share the common UTR sequence (5 1 -ACACAAAGACGGUGCACGAGAUG (initiation codon is underlined)). Each plasmid also lacks downstream initiation codons, to prevent the generation of additional precursor proteins. This abolishment allowed us to analyze the role of a single precursor protein in the production of the Gn/gLuc fusion protein. Using those plasmids, we analyzed the secretion of the Gn/gLuc fusion proteins into the culture supernatant. Surprisingly, AUG2-M had 282% gLuc activity, which was significantly higher than that of AUG3-M, AUG4-M, or AUG5-M. On the other hand, the AUG3-M had significantly lower gLuc activity (66%) than that of the AUG4-M (116%) or the AUG5-M (92%). The difference of gLuc activity between AUG4-M and AUG5-M was marginally significant. The results indicated that precursor-3, produced from the AUG3-M, is less efficient than precursor-4 from the AUM4-M in the secretion of the Gn/gLuc fusion protein. Since the gLuc activities of ∆1 + 2 + 3, and ∆1 + 2 + 3 + 4 mutants were 13%, and 15% ( Figure 2B), respectively, the viral UTR upstream of the 4th or 5th AUG affects the translation efficiency of precursor-4 or 5. Furthermore, the AUG2-M mutant increased the secretion of the Gn/gLuc fusion protein, compared to the ∆1 mutant, indicating that viral UTR upstream of the 2nd AUG also affects the translation efficiency of precursor-2.
3.4. The rMP-12 Encoding the AUG2-M Mutation or the rMP-12 Encoding the ∆2 + 3 Mutation Replicate Less Efficiently Than Parental rMP-12 We, next, characterized the viral phenotypes caused by modification of the M-segment precursors. The reporter assay results indicated that the AUG2-M mutant plasmid generates Gn proteins efficiently, due to a lack of upstream viral UTR sequence. Thus, we aimed to test whether the recombinant MP-12 encoding the AUG2-M mutations in the preglycoprotein coding region can replicate more efficiently than other mutants. However, the modification of preglycoprotein region also affects the expression of the 78 kD, NSm, or NSm'. The AUG2-M mutant does not encode the 78 kD and NSm' proteins, but encodes the NSm protein with 3 mutations (Met to Leu, at the 3rd, 4th, and 5th AUGs). For comparison, we also analyzed the ∆1 mutant (lacking the 78 kD, but still encoding NSm and NSm'), ∆1 + 2 mutant (lacking 78 kD and NSm, but encoding NSm'), and the ∆2 + 3 mutant (encoding 78 kD, but lacking NSm and NSm'). Those constructs were predicted to express Gn protein less efficiently than the AUG2-M mutant construct, based on reporter assay result. As a control, the ∆4 + 5 mutant was analyzed, which produces the default precursors from the 1st and 2nd AUGs.
We, first, analyzed the replication of rMP-12 mutants (the AUG2-M, ∆1, ∆1 + 2, ∆2 + 3, or ∆4 + 5) at a multiplicity of infection (MOI) of 0.15. All the rMP-12 mutants replicated efficiently in Vero cells, while the rMP-12 encoding the ∆2 + 3 mutation, or that encoding the AUG2-M mutation replicated slightly more slowly than the others ( Figure 4A). The arithmetic means of log 10 titers of AUG2-M mutant at 48 and 72 hpi were 7 and 2 times lower than those of the ∆4 + 5 mutant at 48 and 72 hpi, respectively (p < 0.05). On the other hand, the arithmetic means of log 10 titers of ∆2 + 3 mutant were 6 and 2 times lower than those of the ∆4 + 5 mutant at 48 and 72 hpi, respectively, and the differences were not statistically significant. These two mutants, AUG2-M and ∆2 + 3, were further analyzed at a higher MOI infection (1 MOI). After virus infection at 37˝C for 1 h, Vero cells were washed 6 times with media, and the cell lysates were collected at 3, 4, 5, 6, 7, and 8 hpi. Western blot analysis showed that parental rMP-12 generated a detectable level of 78 kD at 6, 7, and 8 hpi, and NSm-Gn at 7 and 8 hpi ( Figure 4B). On the other hand, 78 kD was not synthesized from the AUG2-M mutant, and the NSm-Gn was not made from the ∆2 + 3 mutant. Viral RNA replication most likely started between 5 and 6 hpi, as there was an increase of all viral proteins at 6 hpi. The parental rMP-12 made plaques with heterogeneous sizes (2.6 to 5.7 mm in diameter). The ∆2 + 3 mutant made smaller plaques than parental rMP-12 ranging from 0.9 to 2.3 mm in diameter ( Figure 4C). On the other hand, the AUG2-M made the intermediate sized plaques between the ∆2 + 3 and parental rMP-12 (1.7 to 4.5 mm in diameter). Taken together, these results indicate that rMP-12 encoding the AUG2-M mutation replicates less efficiently than ∆4 + 5 mutant, though the NSm-Gn-Gc precursor made from the AUG2-M mutant was predicted to generate more Gn than ∆4 + 5 mutant. Thus, the discrepancy of reporter assay and virus replication efficiency suggested a role of 78 kD, NSm, or NSm' in the downstream expression of Gn.

Discussion
A live-attenuated RVFV vaccine, MP-12, is conditionally licensed for veterinary use in the U.S. Though the MP-12 vaccine is safe and efficacious [19][20][21][22]39], the vaccine lacks a marker for DIVA (differentiation of infected from vaccinated animals). Reverse genetics is a useful tool to generate RVFV lacking either the 78 kD or NSm proteins. The ∆78 kD or ∆NSm mutant can be made by the abolishment of the 1st or the 2nd AUG, respectively. An introduction of a DIVA marker in the M-segment can be made by truncating the 78 kD/NSm coding region ranging from the 1st AUG to the downstream of the 3rd AUG, while leaving a short UTR upstream of the 4th AUG. The rMP12-∆NSm21/384 (similar to our AUG4-M mutant plasmid), which lacks both 78 kD and NSm expression, showed a similar immunogenicity and efficacy with parental MP-12 [40,41]. However, without knowing the role of each AUG or UTR in the Gn/Gc expression or virion productions, the impact of alterations of the preglycoprotein coding region on the virological phenotype cannot be predicted. In this study, we analyzed the effect of AUG abolishment or an in-frame deletion of viral UTR sequence upstream of the AUG on the Gn expression levels using a quantitative reporter assay system. Subsequently, we also characterized the virological phenotypes of representative AUG mutants. Initially, we hypothesized that increased expression levels of Gn/Gc would increase progeny virus titer. However, that assumption was not correct, and the results showed an unexpectedly complex regulation of viral progeny production through Gn and Gc. As increased production of Gn did not lead to an increase in virus production, regulation of viral production may be regulated at a later step in the viral life cycle.
The reporter assay expressing the Gn/gLuc fusion protein downstream of the RVFV preglycoprotein coding region is useful to measure the level of cleaved Gn/gLuc from the precursor proteins translated from the 1st, 2nd, 3rd, 4th, or 5th AUG. Since the Gn/gLuc fusion protein does not encode the Golgi retention signal, the protein is secreted out from transfected cells without accumulation in the Golgi. Thus, the relative reporter activities in the culture supernatants indicate the efficiency of precursor expression, and subsequent cleavage of the precursor protein. We confirmed that the plasmid lacking the 2nd, 3rd, 4th, and 5th AUGs (∆2 + 3 + 4 + 5) still generates a small amount of the Gn/gLuc fusion protein and a detectable gLuc activity in the culture supernatant. The 78 kD protein encodes a signal sequence at the N-terminus [28], and the 78 kD-Gc precursor is synthesized in the ER membrane. The second signal sequence for 78 kD protein, which is located between the NSm and Gn coding region, may not be efficiently recognized by signal peptidase, which may be hindered by the folding of ectodomain in the ER lumen. Indeed, we could not rescue the rMP-12 encoding the ∆2 + 3 + 4 + 5 mutation in the M-segment, most likely due to such low expression of Gn from the 78 kD-Gc precursor. The ∆2 showed decreased secretion of the Gn/gLuc fusion proteins (32%) in culture supernatants. However, the ∆1, ∆3, or ∆4 mutant still efficiently secreted the Gn/gLuc fusion proteins (121%, 85%, or 72%, respectively). In addition, in the presence of the 1st and 2nd AUG, the abolishment of the 3rd, 4th, or 5th AUG (∆3 + 4 + 5) did not affect the gLuc activity (104%). Thus, the precursor-2, produced by AUG 2, plays a role in the production of the Gn/gLuc fusion protein. Since the ∆1 + 2 mutant still expresses relatively high gLuc activity (74%), compared to the ∆1 + 2 + 3 mutant (13%), the precursor-3, but not precursor-4, or 5, serves as an efficient surrogate of precursor-2 in the production of the Gn/gLuc fusion protein. Thus, our results indicated that the NSm-Gn-Gc precursor plays a default role, and the NSm'-Gn-Gc precursor plays a surrogate role, in the expression of the Gn protein.
Although we introduced mutations to abolish specific AUGs, this approach did not address the effect of long viral UTR upstream of AUG in the translation efficiency of Gn. Relative gLuc activities of the ∆1 + 2 + 3 or ∆1 + 2 + 3 + 4 mutants were low. To address this concern, we generated additional reporter constructs encoding the 2nd, 3rd, 4th, or 5th AUG, without upstream viral UTR sequences (AUG2-M, AUG3-M, AUG4-M, or AUG5-M, respectively). We also abolished downstream AUGs, to prevent the expression of more than one precursor protein. The AUG4-M (116%), and AUG5-M (92%) constructs showed increased gLuc activity, compared to ∆1 + 2 + 3 (13%), and ∆1 + 2 + 3 + 4 (15%) mutant plasmids. Thus, the deletion of the UTR upstream of the 4th or 5th AUG improved the secretion of the Gn/gLuc fusion protein from precursor-4, or 5, respectively. Those results indicated that an in-frame deletion of UTR sequence increases the expression of Gn/Gc from the precursor made from the 4th or 5th AUG.
A limitation of this reporter assay is the lack of natural viral assembly and budding from the Golgi. The results from reporter assay predict the expression levels of Gn proteins from mRNA. However, RVFV Gn encodes a Golgi retention signal at the C-terminus, and co-localizes with Gc to form heterodimers [30,42,43]. Gn and Gc are assembled with the viral ribonucleocapsid, bud from the Golgi, and out of the cell. On the other hand, little is known about the functions of 78 kD, NSm, or NSm' in the assembly process. We generated recombinant rMP-12 encoding mutations in the preglycoprotein coding region (∆1, ∆1 + 2, ∆2 + 3, ∆4 + 5, or AUG2-M). Our study showed that the rMP-12 encoding the AUG2-M mutations replicated slightly less efficiently than parental rMP-12 in Vero cells, despite having increased Gn expression. The AUG2-M does not make 78 kD and NSm', but encodes NSm having Met-to-Leu substitutions at the 3rd, 4th, and 5th AUGs. The Met-to-Ala substitution at the 3rd AUG is known to decrease the migration of NSm protein [18], indicating the occurrence of posttranslational modification of NSm. We assume that NSm and NSm' play a role at a later step of viral protein synthesis: For example, viral assembly process. The ∆2 + 3 mutant encode neither NSm nor NSm' but still generates the 78 kD protein [18]. In our study, this mutant also showed relatively inefficient replication kinetics in Vero cells, although it was not statistically significant. The NSm of Bunyamwera virus (BUNV: genus Orthobunyavirus) serves as a scaffold to form a "viral tube" structure to facilitate the assembly of the ribonucleocapsid with the Gn/Gc complex at the peripheral Golgi, and a lack of NSm reduce virus production by 10 to 100-fold [44]. The 78 kD protein may also play a role in viral replication. Kreher et al. showed that the AUG (Met) to GCG (Ala) mutation at the 1st AUG (∆1), at the 1st and 2nd AUGs (∆1 + 2), or at the 1st, 2nd, and the 3rd AUGs (∆1 + 2 + 3) led to the emergence of RVFV mutant encoding a new AUG upstream of the original 1st AUG, during 5 serial passages in mammalian cells [18]. This new AUG can generate a 78 kD-Gc-like precursor protein, in addition to the NSm-Gn-Gc precursor. Previous studies indicated that the 78 kD plays a major role in viral dissemination in mosquito vectors, while NSm and NSm' contribute to the RVFV propagation in vertebrate cells [18]. However, little is known about the role of 78 kD, NSm, or NSm' proteins in viral replication, and further studies will be required to elucidate the mechanisms.

Conclusions
This study showed that, not only the abolishment of AUG, but also the truncation of viral UTR, affects the expression of Gn protein by the RVFV M-segment. Increased production of Gn did not lead to an increase in virus production, and thus, regulation of viral production may be further regulated at a later step in the viral life cycle.