Genetic Changes at the Glycoprotein Editing Site Associated With Serial Passage of Sudan Virus

Abstract

Sudan virus (SUDV), like the closely related Ebola virus (EBOV), is a filovirus that causes severe hemorrhagic disease. They both contain an RNA editing site in the glycoprotein gene that controls expression of soluble and full-length protein. We tested the consequences of cell culture passage on the genome sequence at the SUDV editing site locus and determined whether this affected virulence. Passage resulted in expansion of the SUDV editing site, similar to that observed with EBOV. We compared viruses possessing either the wild-type or expanded editing site, using a nonhuman primate model of disease. Despite differences in virus serum titer at one time point, there were no significant differences in time to death or any other measured parameter. These data imply that changes at this locus were not important for SUDV lethality.

Sudan virus (SUDV; formerly known as Sudan ebolavirus), like the closely related Ebola virus (EBOV; formerly known as Zaire ebolavirus), can cause severe hemorrhagic disease with high case-fatality rates. Filoviruses are negative-sense, single-stranded, enveloped RNA viruses; no approved vaccines or therapies exist for filovirus infections. Evidence suggests that fruit bats may be the natural reservoir for filoviruses, but they can also replicate in other hosts, including humans, nonhuman primates (NHPs), and pigs. Filoviruses also show great potential for emergence in new geographic regions and hosts and for transmission by new modes [1–3]. Indeed, the 2014 West African outbreak highlights the capacity for these viruses to emerge and cause a global threat [4].

It has long been recognized that passage of virus results in accumulation of genetic changes as the virus adapts to the cultured cells [5–7]. In EBOV, one such genetic change is in the glycoprotein (GP) RNA editing site [8]. EBOV and SUDV both contain an RNA editing site in the GP region. The genome contains a stretch of 7 uridines (U), and the majority of messenger RNA (mRNA) transcripts consequently contain 7 adenosines (A). The resulting product is soluble GP (sGP), a truncated form of GP. However, when RNA editing occurs, an additional A is incorporated into the mRNA transcript, resulting in the translation of full-length GP [9–11]. Others have previously observed that cell culture passage of EBOV in Vero E6 cells can cause a change in the genomic sequence at the editing site such that 8 U are present instead of 7 U [8, 12]. During high-multiplicity infection, these changes have been seen as early as 24 hours after infection [13]. The effect of cell culture adaptation on the SUDV editing site is unclear.

While the West African outbreak is revealing novel information about EBOV pathogenicity, much still remains unknown. The different forms of GP and the different ratios produced during infection are thought to play a major role in pathogenicity [14, 15]. However, all of the functions are not understood, and it remains unclear which forms influence pathogenicity [16–19]. Several researchers have begun to characterize the importance of EBOV GP during infection, but SUDV GP RNA editing is still poorly understood. As it seems likely this region is significant during SUDV infections, further information is critical. Herein, we tested the consequence of serial passages of SUDV on the editing site and determined whether this had an effect on virulence in the Macaca fascicularis model of SUDV infection.

METHODS

Ethics Statement

Animal research was conducted under a Texas Biomedical Research Institute (TBRI) Institutional Animal Care and Use Committee (IACUC)–approved protocol (protocol 1381) in compliance with the Animal Welfare Act. The facility where this research was conducted is accredited by the Association for Assessment and Accreditation of Laboratory Animal Care International and adheres to principles stated in the Guide for the Care and Use of Laboratory Animals [20]. The studies were conducted in the animal biosafety level 4 facility at the TBRI in compliance with the TBRI IACUC and the Guide for the Care and Use of Laboratory Animals [20]. Animals were housed individually and fed monkey biscuits. Enrichment included visual stimulation and commercial toys. During blood collection, animals were anesthetized using tiletamine/zolazepam, and euthanasia criteria were developed to minimize undue pain and distress. When animals became moribund, they were euthanized with an intravenously administered overdose of sodium pentobarbital.

Cells and Virus

Vero E6 cells (VERO V2008; catalog number NR-596; BEI resources) were grown in minimum essential medium (Life Technologies) containing 2 mM l-glutamine and 1 mM sodium pyruvate (henceforth, referred to as “normal growth medium”) with 10% heat-inactivated fetal calf serum (FCS; Life Technologies) at 37°C with 5% CO₂. The starting material for viral passage experiments consisted of: Ebola virus (Kikwit), 199 510 621 ZAIRE, passage number 2 on Vero E6 cells; and Sudan virus, 200 011 676 GULU, passage number 2 on Vero E6 cells; both were acquired from Dr T. Ksiazek at the University of Texas Medical Branch. Virus was amplified using the following methods. Vero E6 cells were infected at a multiplicity of infection (MOI) of 0.001 plaque-forming units (PFU)/cell in normal growth medium containing 2% FCS. Viral supernatant was harvested when the cells exhibited 3+ cytopathic effects.

Viral Passage

Virus was serially passaged 18 times in Vero E6 cells. The initial amplification was performed as described above. Subsequent experiments designed to test various MOIs were performed identically to the 0.001 passages and used the MOI indicated in the “Results” section. After each round of amplification, the viral titer was determined. After titration, a portion of the virus was used for a subsequent round of amplification.

Determination of Viral Titers

Viral titers were determined by a plaque assay, using agarose and neutral red [21].

Deep Sequencing and Sample Preparation

An aliquot of viral supernatant was diluted in TRIzol LS Reagent (Life Technologies) and transferred to a biosafety level 2 laboratory, where RNA was harvested following the manufacturer's instructions. Following RNA harvest, DNA, ribosomal RNA, and mRNA were removed as previously described [22]. RNA libraries were then prepared using Illumina's TruSeq Total RNA sample preparation kit, according to the manufacturer's instructions. The initial RNA purification and fragmentation step was performed following an optional manufacturer-suggested modification that omitted the polyA RNA purification step. The final complementary DNA library was sequenced using sequencing by synthesis (Illumina) using the 150-bp paired end format on an Illumina MiSeq. The equivalent technologies have been shown previously not to yield artifacts at the Ebola virus editing site [13]. Data analysis began by using the Illumina pipeline to generate a FASTQ file containing all the reads, which was then mapped to the parental virus sequence by using Lasergene SeqMan NGen (DNASTAR). Quality trimming and trimming to mer were performed on reads with a minimum similarity of 93%. The abundance of each nucleotide at each position in the genome was determined and compared to that in a reference sequence. Changes were then presented as the percentage of single-nucleotide polymorphisms. To identify changes at the editing site, reads were extracted that mapped to the editing site and included >10 nucleotides of matching of flanking sequence on one side of the editing site plus one nucleotide of matching flanking sequence on the other side. The length of the editing site was determined from these data, and the abundance of each form of the editing site was then quantified.

To determine whether our sequencing procedures introduced artifacts at homopolymeric sequences, we analyzed an independent run of 7 U in the viral genome, located in the L gene, using the procedures described above. In each of the 5 sequencing runs that were analyzed, we observed <1% of the total reads carrying an 8 U genotype. This suggests that our sequencing procedure incorporates an additional base during a run of 7 U at a frequency of <1% (determined as the combined values from sequencing errors and genomes that may actually carry 8 U at this locus).

Experimental Inoculation of NHPs

Eight male M. fascicularis aged 2–4 years and weighing >4 kg were used. Animals were acquired from Covance and underwent serum testing to ensure no reactivity to filovirus antigen prior to purchase. Animals were shipped directly to the TBRI for a standard quarantine period. This quarantine period allowed for evaluation of the animals' health status and for their acclimation to caging and diet. Eight animals were intramuscularly injected in the deltoid muscle with 100 PFU of one of 2 stocks of SUDV. Four animals received virus that had been passaged in cell culture 3 times (P3 virus) with a 7-U genotype, and 4 animals received virus that had been passaged in cell culture 13 times (P13 virus) with an 8-U genotype. Test subjects were then observed at least twice daily.

Hematologic, Coagulation, and Blood Chemistry Analyses

Biochemical analysis was performed using the mammalian liver enzyme profile rotor on a Vet Scan analyzer (Abaxis). Complete blood counts were performed using a Vet HM2 machine (Abaxis). Coagulation times were determined using the IDEXX Coag Dx Analyzer (IDEXX Laboratories).

Statistical Analysis

The log-rank Mantel–Cox test was used to analyze the survival curves. Two-way analysis of variance (ANOVA) with the Bonferroni post hoc test was used to analyze serum titers.

RESULTS

Changes to the Editing Site Genotype as SUDV Is Passaged in Vero E6 Cell Culture

It has been demonstrated that cell culture passage of EBOV in Vero E6 cells correlates with a change in the genomic sequence at the GP RNA editing site [8, 12]. To determine whether SUDV undergoes a similar change, we serially passaged SUDV in Vero E6 cells at a MOI of 0.001 PFU/cell. We used ultra-deep sequencing to investigate genomic changes. As a control, we performed a parallel experiment with EBOV and demonstrated that expansion of the editing site occurred similarly to previously published observations (unpublished data). Figure 1 shows the relative abundance of 6, 7, 8, 9, or 10 U at the editing site in SUDV. The percentages are calculated from the number of sequence reads obtained during the sequencing run (mean coverage across the editing site ranged from 363X to >40 000X coverage). The first 5 serial passages were performed twice, and results from each replicate are displayed individually to illustrate the differences. Overall, a similar pattern emerged during both experiments; cell culture passage of SUDV in Vero E6 cells correlated with a shift from 7 U to 8 U in the genomic sequence at the GP editing site. However, during early passages the percentage of 7 U versus 8 U sequence reads did not exhibit an obvious trend. For example, in the first replicate the abundance of 7 U declined initially before increasing again at passage 5, and a steady decline was not observed until after the eighth passage. The consensus genotype shifted to 8 U after the 11th passage. The EBOV editing site genotype is not as variable during early passages, but the consensus genotype shift from 7 U to 8 U occurred at passage 10, similar to that observed with SUDV.

Figure 1.

Editing site genotype of Sudan virus (SUDV) after serial passage in Vero E6 cells. SUDV was serially passaged in Vero E6 cells at a low multiplicity of infection (MOI) of 0.001 plaque-forming units (PFU)/cell. Ultra-deep sequencing was used to investigate the relative abundance of uridine (U) at the editing site after each passage. A, SUDV was passaged 18 times. B, The first 5 passages were repeated at the same MOI. C, The first 5 passages were repeated at different MOIs (0.01, 0.1, and 1.0 PFU/cell).

We also saw an accumulation of sequence reads containing 9 U, and rarely a small percentage of sequencing reads contained 6 or 10 U. As passaging continued, the abundance of 9 U appeared to exhibit a moderate increase. The appearance and potential significance of these other editing site lengths has not been well described in the literature, although when 9 A are present in EBOV mRNA, a +2 shift in the open-reading frame occurs and the small sGP (ssGP) is generated [11, 14].

We hypothesized that the MOI used during cell culture passage might affect the genotypic changes seen at the SUDV editing site. To test this, we serially passaged SUDV 5 times in Vero E6 cells at 3 MOIs higher than the one used in previous experiments (0.01, 0.1, and 1.0 PFU/cell). As seen in Figure 1C, it appeared that increasing the MOI does not result in the same variable yet fairly rapid decline in 7 U at the editing site. At all 3 higher MOIs, we did not observe an accumulation of 8 U genomes during the first 5 passages in Vero E6 cells. The abundance of reads containing 6, 9, or 10 U was also much lower; as the MOI increased, the abundance of these alternative editing site lengths declined.

Editing Site Genotype of SUDV During NHP Infection

It had been previously observed that when EBOV with an 8 U consensus genotype was used to infect NHPs, genotype reversion from 8 U to 7 U occurred [8]. During similar studies at our facility, we have also observed a reversion to the 7-U genotype (unpublished observations). We hypothesized that the SUDV genome would undergo a similar reversion in vivo. To test this, 8 M. fascicularis were challenged intramuscularly with 100 PFU of one of 2 SUDV stocks. Four animals received virus that had only been passaged in Vero E6 cell culture 3 times and had a 7-U genotype (P3 virus). The other 4 received virus that had been passaged in Vero E6 cell culture 13 times and had an 8-U genotype (P13 virus). When moribund, animals were humanely euthanized and necropsied, and blood samples were collected. Nucleic acid was isolated from serum specimens, and ultra-deep sequencing was used to determine the sequence of the editing site in the viral RNA (Figure 2A). Viral RNA isolated from the 4 animals inoculated with P3 virus having a 7-U genotype maintained a 7-U genotype at the end of the project. Interestingly, the progeny viral population appeared to contain even more 7-U genomes than the parent virus. However, in contrast to what is seen during EBOV infections, viral RNA isolated from animals inoculated with P13 SUDV having an 8-U genotype did not show uniform reversion back to the 7-U genotype. Viral RNA found in the serum of one animal still exhibited an 8-U genotype. All animals succumbed to viral infection. The median time to death for P3 virus–challenged animals was 8.5 days, and the median time to death for P13 virus–challenged animals was 10 days; no statistically significant difference was observed between survival of the 2 groups when using the log-rank Mantel–Cox test. When comparing serum titers of P3 and P13 virus–challenged animals, no significant differences were found on day 3, day 5, or at the time of death (P > .05, by 2-way ANOVA with the Bonferroni post hoc test); only day 7 serum titers exhibited a significant difference (P < .01; Figure 3). Rectal temperature was recorded on days 0, 3, 5, 7, and 10 after infection (Figure 3). All animals developed increased temperatures as the study progressed and then exhibited temperature declines with the need for humane euthanasia. Blood specimens were collected on days 0, 3, 5, 7, and 10 for blood chemistry analysis, determination of coagulation times, and hematologic analysis. Hematologic data showed a decrease in platelets, an increase in neutrophils, and lymphocytopenia in all animals over the course of infection (Figure 4). Prolonged coagulation times and decreased albumin levels were also detected (Figure 4). All animals exhibited increased activated partial thromboplastin times (aPTTs) as the study progressed. Animal 960 had an aPTT that was out-of-range high (>350 seconds) on the day of euthanasia (day 7 after infection). Animal 962 also had an out of range-high value on day 7 post infection, but the values on day 10 were within range and similar to those for the other remaining animals. As seen in Figure 5, all animals exhibited increases in levels of serum alanine aminotransferase, alkaline phosphatase, γ glutamyl transferase, and blood urea nitrogen. These observations are consistent with filovirus infection in experimentally infected NHPs, as previously described [23, 24].

Figure 2.

Editing site genotype and phenotype of Sudan virus (SUDV) after infection in Macaca fascicularis. Eight M. fascicularis were challenged intramuscularly with 100 plaque-forming units of one of 2 SUDV stocks. Four animals received virus that had only been passaged in Vero E6 cell culture 3 times and had a 7-uridine (U) genotype (P3 virus). The other 4 received virus that had been passaged in Vero E6 cell culture 13 times and had an 8-U genotype (P13 virus). A, Nucleic acid was isolated from serum specimens, and ultra-deep sequencing was used to determine the sequence of the editing site in the viral RNA. B, Survival proportions of animals challenged with P3 vs P13 virus. No statistical difference was observed using the log-rank Mantel–Cox test. Abbreviation: NHPs, nonhuman primates.

Figure 3.

Serum titers and rectal temperatures in Macaca fascicularis experimentally infected with P3 or P13 Sudan virus (SUDV). Eight M. fascicularis were challenged intramuscularly with 100 plaque-forming units (PFU) of one of 2 SUDV stocks; 4 animals received virus that had a 7-uridine genotype (P3 virus), and 4 animals received virus that had an 8-uridine genotype (P13 virus). A, Blood samples were collected on days 0, 3, 5, 7, and 10 and at the time of euthanasia for viral load determination. No significant differences were found on days 3 or 5 or at time of death (P > .05, by 2-way analysis of variance with the Bonferroni post hoc test); only day 7 serum titers exhibited a significant difference (P < .01). B, At each scheduled time of blood sample collection in the morning, rectal temperature was recorded. All animals initially exhibited increases in temperature, followed by a decline. Abbreviation: NHP, nonhuman primate.

Figure 4.

Coagulation times and hematology in Macaca fascicularis experimentally infected with P3 or P13 Sudan virus (SUDV). Eight M. fascicularis were challenged intramuscularly with 100 plaque-forming units of one of 2 SUDV stocks; 4 animals received virus that had a 7-uridine genotype (P3 virus), and 4 animals received virus that had an 8-uridine genotype (P13 virus). Blood specimens were collected on days 0, 3, 5, 7, and 10 for analysis of coagulation times and hematologic testing. A, Activated partial thromboplastin time (aPTT). No data are available for animals 960 and 962 on day 7 after infection because samples did not produce readings within range for the test. Coagulation data on day 5 after infection were not recorded for animal 962. B, Percentage of neutrophils and lymphocytes. C, Number of platelets. D, Albumin levels. Abbreviation: NHP, nonhuman primate.

Figure 5.

Findings of blood chemistry analysis of Macaca fascicularis experimentally infected with P3 or P13 Sudan virus (SUDV). Eight M. fascicularis were challenged intramuscularly with 100 plaque-forming units of one of 2 SUDV stocks; 4 animals received virus that had a 7-uridine genotype (P3 virus), and 4 animals received virus that had an 8-uridine genotype (P13 virus). Blood was collected on days 0, 3, 5, 7, and 10 for blood chemistry analysis. A, Alanine aminotransferase (ALT) levels. B, Alkaline phosphatase (ALP) levels. C, γ glutamyl transferase (GGT) levels. D, Blood urea nitrogen (BUN) levels. Abbreviation: NHP, nonhuman primate.

DISCUSSION

EBOV and SUDV both contain an RNA editing site in the GP region. This site regulates production levels of the different forms of GP [10, 11]. The various forms of GP and the different ratios produced during infection are thought to play a major role in pathogenicity [14, 15, 25]. Full-length GP is involved in cell attachment and membrane fusion, although it can also be cleaved and shed from infected cells. The nonstructural, sGP is released from virus-infected cells. High levels of both forms have been detected circulating in the blood of infected humans and animals [9, 15]. Full-length GP appears to play a major role in the immune dysregulation and vascular instability seen during EBOV infection [15, 25, 26]. It also seems to cause differential expression of host cell surface proteins [19, 27, 28]. Furthermore, full-length GP has been described as more cytotoxic than sGP, although the importance of this remains unclear [19]. The shed form of full-length GP has been shown to activate certain immune cells and induce the secretion of cytokines. sGP is less understood but is also thought to play a role in pathogenesis, possibly through its ability to interact with the immune system [14, 26]. sGP has also been hypothesized to have antiinflammatory properties [16, 29]. As a result of the assorted functions, it seems that the GP RNA editing site is critical in modulating expression levels of the different forms of GP. The genome sequence is likely to be highly involved in this process [8, 11, 26]. While changes to the EBOV RNA editing site have been observed following in vitro passage and in vivo challenge [8], the SUDV GP RNA editing site remains poorly characterized and not well understood. Herein, we tested the consequence of serial passage in cell culture on the SUDV editing site to determine whether this had an effect on virulence in the M. fascicularis model of SUDV infection.

Our in vitro results indicate that the SUDV GP editing site does mutate during Vero E6 cell culture passage. Vero E6 cell culture passage of SUDV correlated with a shift from 7 to 8 U in the genomic sequence at the GP editing site, similar to that observed with EBOV. However, SUDV exhibited more variability than EBOV at the editing site during early passages. The SUDV editing site also accumulated sequence reads containing 6, 9, and 10 U. These variations have also been described for EBOV and may have biological significance. For example, during EBOV infection, full-length GP appears to exhibit differential binding and activation of immune cells than sGP [15, 16]. Additionally, the ssGP produced from mRNA carrying the 9-A sequence seems structurally similar to sGP but has different effects on immune and endothelial cells [14, 16, 29]. Overall, the appearance and potential significance of these other quantities of U has not been well described in the literature, and their influence on SUDV pathogenesis requires more study.

After completing the initial passaging experiments, we hypothesized that the MOI might play a role in the genotypic changes seen at the editing site. When passaging viruses in cell culture, selection of the MOI is an important consideration. A high MOI may minimize the number of rounds of replication the virus undergoes, while a low MOI may result in a bottleneck, and genome diversity might not be maintained. To test the influence of the MOI on the editing site genotype, virus was serially passaged at various MOIs. Higher MOIs did not result in the same rapid shift from 7 U to 8 U in the genomic sequence at the GP editing site. The increased rounds of replication that the virus undergoes at the lowest MOI seem to lead to the increased accumulation of 8 U. This suggests that the MOI does have an influence on the SUDV editing site genotype.

During in vivo challenge with SUDV, we did not see the 100% reversion to the 7-U genotype that is seen during EBOV infections; however, it is important to note that these apparent differences are based on studies using limited numbers of animals. When P13 SUDV with an 8-U genotype was used to challenge M. fascicularis, the RNA editing site exhibited reversion to wild-type 7 U. However, this reversion was not seen in all animals. Others have hypothesized that the GP editing site genomic adaptations seen in vitro and in vivo suggest that selection pressures dictate the amount of full-length versus sGP that are desirable. For example, if full-length GP causes more cytotoxicity, the in vivo adaptation in which expanded 8 U reverts back to wild-type 7 U indicates that overexpression of full-length GP in vivo is detrimental [8, 11, 26]. This, coupled with our data, suggests that in vivo selection pressures may differentially affect EBOV and SUDV. For example, previous research suggests that cytokine expression varies between SUDV disease and EVD, possibly because EVD develops more rapidly [30]; exposure to different cytokines could be partially responsible to different GP expression patterns. Furthermore, while GP expression is likely significant during SUDV disease, our data indicate that pathogenicity also may be influenced by other factors. We found that SUDV can be highly pathogenic in vivo, but pathogenicity seems independent of the GP editing site genotype. We observed that, during infection of M. fascicularis with SUDV, there was no difference in survival during infection with a low-passage 7-U genotype virus versus infection with a high-passage 8-U genotype virus. This provides further evidence that the SUDV GP editing site genotype may have limited impact on pathogenicity. SUDV may be less reliant on GP expression modulation to regulate pathogenicity. Alternatively, SUDV may have evolved different mechanisms to modulate GP expression levels, rather than relying on the editing site.

Overall, we observed that the SUDV editing site undergoes similar changes to EBOV when it is passaged in vitro or in vivo, but these changes are more variable than what is seen in EBOV. This may indicate that future experiments should focus specifically on SUDV, rather than relying on extrapolations from EBOV data.