Figure 1

Figure 1. Infection of cells with vesicular stomatitis virus (VSV), Herbert virus (HEBV), and Gouléako virus (GOLV). A) Porcine kidney 15 cells; B) human embryonic kidney cells; C) human hepatocellular 7 cells. Cells...

We first extended our recent cell culture studies to include porcine kidney 15 and human embryonic kidney 293 cells, which were the type of cells used by Chung et al. (9). Human hepatocellular 7 carcinoma cells were also included because they are highly susceptible to virus infection, as are Vero cells and several other cell lines we used in earlier studies (5,6). Infections with GOLV and HEBV were performed at multiplicities of infection of 1 in doublets in all cell lines. Vesicular stomatitis virus was used as a positive control at multiplicity of infection 1. Cell culture supernatants were analyzed for viral RNA after 0, 3, and 6 days by real-time reverse transcription PCR (RT-PCR) (5,6). No replication of GOLV and HEBV was detected, whereas vesicular stomatitis virus replicated to high concentrations (Figure 1). Three blind passages on fresh cells failed to yield virus.

Because cell culture experiments may not show the full host range of a specific virus, we tested serum samples collected in 2008 from Sus scrofa domestica pigs in Gouléako, the rural village where GOLV and HEBV were first isolated from mosquitoes in Côte d’Ivoire (5,6). The 28 tested samples represented nearly all the pigs kept in Gouléako at that time, all of which were constantly exposed to mosquitoes. We also tested 108 serum samples collected in 2011 from mosquito-exposed swine in Kumasi, Ghana, where mosquitoes were found to be infected with HEBV (6) and GOLV (S. Junglen, unpub. data).

All samples were tested for virus by real-time RT-PCR (5,6) and tested for antibodies against GOLV and HEBV nucleocapsid proteins by recombinant immunofluorescence assay (10). All samples were negative for the viruses (Technical Appendix Table). Technical Appendix Figure 1 shows antigen controls and results from 1 representative swine serum sample.

Figure 2

Figure 2. Maximum-likelihood phylogenetic analyses of Gouléako virus (GOLV) and Herbert virus (HEBV) strains from mosquitoes in Côte d’Ivoire, 2004, and Ghana, 2011, and virus strains detected by Chung el al. (

To compare the viruses found in pigs in South Korea with viruses found in mosquitoes in Africa, we replicated methods used by Chung et al. (9) and amplified a region of the GOLV glycoprotein precursor gene from 27 GOLV strains in mosquitoes (Technical Appendix). Nucleotide sequence distance among mosquito strains was as high as 9.0%. The viruses found in the pigs fell within the genetic diversity of viral strains of GOLV and HEBV and did not constitute phylogenetic outliers (Figure 2, panel A). The analyzed fragment had 6 aa exchanges, but they were insufficient for drawing conclusions about protein function because the fragment did not include domains putatively relevant for receptor binding (Technical Appendix Figure 2).

Small RT-PCR fragments from the RNA-dependent RNA polymerase (RdRp) gene were presented by Chung et al. for HEBV. We performed phylogenetic analyses to compare these swine-derived sequences with sequences from all mosquito-derived viruses from which we could sequence the corresponding genome region (Figure 2, panel B). Comparison of swine-derived sequences with the phylogeny of mosquito-derived HEBV strains, constructed on the basis of the third conserved region of the RdRp (Figure 2, panel C), showed that the strains from South Korea fell within the phylogenetic diversity of HEBV strains identified in West Africa. Technical Appendix Figure 3 shows nucleotide- and amino acid–based alignments.

Our results contrast with those of Chung et al. (9) for several possible reasons. First, the viruses infecting swine in South Korea may constitute variants of GOLV and HEBV that can infect vertebrates. The presence of an NSs protein in phleboviruses and orthobunyaviruses provides interferon resistance required to infect vertebrates efficiently (3,4). Because full genome sequences from swine viruses detected by Chung et al. are not available, we have no information on the presence of NS proteins in these viruses. Furthermore, our detection assays might have failed to detect variant viruses. However, our RT-PCR assays have been shown to detect variant viruses, have been validated for sensitivity (≈100 viral genome copies per mL in liquid specimens), and provide high specificity by probe detection (5,6). A concern regarding the results of Chung et al. is the use of RT-PCR assays based on SYBR Green (Thermo Fisher Scientific, Lithuania) product detection, which, from our experience, is prone to yield nonspecific results because no probe is used in this assay. Nevertheless, RT-PCR products in Chung et al. have been confirmed by sequencing. Some sequences presented by these researchers contained stop codons in the HEBV RdRp and the GOLV glycoprotein precursor genes, making it unlikely that these sequences represent replicating viruses. Besides technical explanations, these sequences could represent viral genome fragments integrated in genomes of organisms, such as insects, that are eaten by pigs in the region. Integration of RNA virids derived from flaviviruses into the host genome has been described in insects (11). Testing food eaten by swine for insect DNA or viral RNA could yield insight. In addition, we may have collected serum when no active virus infections occurred in tested animals. However, past infections would have been shown by antibody tests. Because bunyaviruses from all vertebrate-infecting genera induce antibodies against the nucleoprotein (12–14), we are confident about our choice of antigen in our assays. Chung et al. presented no serologic results to support virus detections (9).

Several technical issues in the study by Chung et al. should be clarified further. First, RNA concentration in tissue, as determined by RT-PCR, did not correlate with the success of probe-based immunohistochemistry in several organ samples (9). Second, supernatants from the virus isolate from South Korea showed high cytopathogenic activity in cell culture (10³–10⁵ cytopathogenic units/mL) but low levels of concomitant viral RNA by RT-PCR. Because no antigen detection in cells was attempted, the cytopathogenic effect could have been caused by any other virus blindly isolated. One of the most infectious and deadly swine pathogens, the porcine reproductive and respiratory syndrome virus (15), was co-detected in lung samples of dead pigs in South Korea (9).

The finding of genome fragments of GOLV and HEBV in swine in South Korea needs to be more fully explored. However, with no further independent proof of infection of swine or other vertebrates, HEBV and GOLV should not be considered epizootic pathogens or arboviruses.