Detection of gene expression in WNV infected human neuroblastoma cells
To determine the optimum time post-infection for induction of inflammation associated genes in infected neuroblastoma cells (SK-N-SH line), the cells were infected with the prototype WNVKUN strain (MRM16) and the cells harvested at different time points post-infection (days 0, 1, 2, 3 and 4). Quantitative RT-PCR confirmed that the cells were infected (Additional file 1). Gene expression of 249 inflammation associated genes was measured using a nCounter panel (NanoString). To test that the cells were capable of responding to immuno-stimulants, poly(I:C) and LPS were used and gene expression of innate immune responses measured (Additional file 3). Poly(I:C) induced gene expression more strongly (> 10-fold induction: C1S, CCL5, CFB, CXCL9, IFI44, IFIT1, IFIT3, MX1, OAS2 and OASL) than LPS (> 10-fold induction: IL8) at the respective concentrations used. WNVKUN (MRM16) infected cells produced strong inflammation gene expression relative to uninfected cells at the same time points. The expression of the 10 most strongly-induced genes (IFNB1, CCL5, OASL, CXCL10, IFIT3, IFI44, IFIT1, IFIT2, PTGS2 and OAS2) are presented as a line-plot (Fig. 1 and Additional file 3). These genes have all previously been shown to be induced by WNV infection [17,18,19,20] indicating an authentic response. The expression of these genes was maximal at day 2 post-infection; therefore, in subsequent experiments cells were harvested at this time-point.
To determine the response to WNVKUN, SK-N-SH cells were infected with 10 different isolates in separate culture wells (Additional file 2: Table S1). These included two virulent isolates which included one isolated from a horse in 1984 (Boort), and another collected from mosquitoes which was closely related to the strain which caused an outbreak in Australian horses in 2011 (NSW2012) [7]. An isolate of Murray Valley encephalitis virus (MVEV) was included to enable a comparison with another Australian pathogenic flavivirus. Gene induction was measured for each isolate at the 48 h optimized time-point post-infection relative to uninfected cells. To measure changes in gene expression, infected cells were compared to uninfected controls. For those genes included on the panel, the response to WNVKUN infection was one of induction rather than down-regulation. The expression of some genes was reduced but these were not statistically significant. Seven genes (FOS, CEBPB, RELB, JUN, PTGFR, IFNB1 and DDIT3; Fig. 2a) were determined to be significantly induced (BY adjusted p-value < 0.05). The most strongly-induced gene was FOS, also referred to as the proto-oncogene c-fos. The gene product forms a heterodimer with the product of another proto-oncogene JUN, or c-jun, in the activator protein complex 1 (AP-1). Interestingly, JUN was also strongly induced in WNVKUN infected SK-N-SH cells. AP-1 has been linked to activation of apoptosis; therefore, induced FOS and JUN expression might also indicate initiation of programmed cell death by WNV infection. Related to this, another induced gene, DNA damage-inducible transcript 3 (DDIT3), is a pro-apoptotic protein which is induced in primary human cell cultures by WNV infection [20]. DDIT3 can interact with FOS and JUN [21], hence its induction further suggests activation of apoptosis by infection.
At least two other transcription factors were significantly induced by WNVKUN infection (Fig. 2a). The first, CCAAT/enhancer-binding protein beta (CEBPB), is a transcription factor which can regulate a large number of genes including c-fos and cytokines including IL6, 8 and 12, and TNF-α (reviewed in [22]), and was previously shown to be upregulated in the brains of WNV infected mice [19, 23]. CEBPB is known to interact with DDIT3 [24], suggesting that they may be part of an apoptotic protein complex (or complexes) which also includes FOS and JUN. The second, RELB, is a member of a sub-family of the NF-κB transcription factors, which can itself form a complex with NF-κB [25]. NF-κB is downstream of the RIG-1 and TLR3/7 pattern recognition receptor (PRR) activated pathways, and is imported into the nucleus to activate IFNα/β and interferon stimulate genes (ISGs) in infected cells. IFNB1 (IFNβ gene) was also significantly and strongly induced by WNVKUN infection consistent with PRR pathway activation. Finally, prostaglandin F receptor (PTGFR), which has several functions in the cell was induced, but its relevance to WNV infection is currently unclear.
To examine the innate immune gene induction response to WNVKUN infection, the 249 inflammation-associated genes on the panel were analyzed using principal component analysis (PCA). The results of the three separate infections showing variation in gene induction across two axes are shown (Fig. 3a-c). The plots of isolates varied among experiments, with no distinct clustering. The position of the virulent NSW2012 isolate was consistently distant relative to the positions of the other isolates in all three experiments. However, that was not the case for the virulent Boort isolate suggesting the nature of its virulence may be different to NSW2012. Gene induction by NSW2012 infection was then compared with the isolates (excluding Boort). Two genes, IL8 and the chemokine CCL2 (also known as MCP-1), were clearly significantly differentially induced (Fig. 2b). When gene induction of the Boort infected cells was compared with gene inductions resulting from infections with other WNVKUN isolates (excluding NSW2012), no significant gene expression differences were observed (Additional file 4: Figure S1). Overall, the above findings suggested that the expression of the induced genes may be different for the NSW2012 isolate compared to the others, and may possibly relate to its greater virulence.
To further explore the possibility that the differential gene induction may reveal markers of virulence, a pathway analysis was conducted. Scores were calculated for 126 different inflammation-associated pathways for the SK-N-SH cells infected with the WNVKUN isolates and plotted as a heat-map (Fig. 3d and Additional file 3). Between the three separate infections, WNVKUN infected cells generally had higher pathway scores than uninfected cells indicative of virus infection. NSW2012 demonstrated the highest and most consistent (all three experiments) pathway scores among the isolates including pathways Iκ-B kinase/NF-κB cascade, cytokine activity, chemokine activity, G-protein coupled receptor binding, JAK-STAT cascade, response to virus, inflammatory response, and programmed cell death. Strain 18658C showed higher pathway scores than the other WNVKUN isolates, albeit with lower scores than NSW2012, in two out of three experiments. Collectively, the above data suggest that the NSW2012 isolate has a more inflammatory phenotype than the other isolates when infecting the neuroblastoma cell line SK-N-SH. By contrast, cells infected with MVEV did not show statistically significant induction of individual genes (Additional file 4: Figure S1). In addition, MVEV showed consistently generally lower pathway scores than WNVKUN. This may have been due to different growth kinetics for MVEV replication, as the assay was optimized for WNVKUN, resulting in gene induction kinetics that were not directly comparable.
Gene induction in WNV-infected, iPSC-derived neuronal cultures
Continuously cultured cell lines such as SK-N-SH have a more metabolically active cellular environment than primary cell cultures or cells from tissue. This would be expected to influence signaling pathways and, as a result, gene expression. Hence, infection experiments were conducted in neuronal cultures derived from human induced pluripotent stem cell (iPSC) derived brain organoids to determine whether the phenotypic differences observed between isolates in the SK-N-SH cells could be recapitulated in this cell system. A smaller subset of WNVKUN isolates was used in these experiments and included isolates MRM16 and K68967, the virulent isolates Boort and NSW2012, and MVEV for comparison. Cells were harvested at 48 h post-infection to be consistent with earlier experiments.
As the neuronal cultures were stem-cell derived, it was firstly necessary to determine if they were permissive to WNVKUN infection and replication. Four separate infections were performed, and infected neuronal cells were found to have significantly lower cycle threshold values by qRT-PCR assay, with ΔCT values (day 0 – day 2 post-infection) typically > 9 (Additional file 1) compared to non-infected cells. Hence, the cells were permissive to WNVKUN (and MVEV) infection. RNA from the cells was then used to measure gene induction as previously. Similarly to SK-N-SH cells, infection resulted primarily in induction (as opposed to down-regulation) of those genes included on the panel, for both WNVKUN and MVEV (Fig. 4 and Additional file 5). There were more induced genes that had lower probability (i.e. statistically significant) scores in the neuronal cultures than the SK-N-SH cells. This may have been a result of the less variable baseline of expression in the uninfected cells. There were many induced genes shared in common between cells infected with these two viruses. Similarly to SK-N-SH cells, WNVKUN infected cells showed induction of IFNB1, RELB, and JUN genes. Of these, only the latter suggested an apoptotic response. There were many other inflammation-associated genes that were found to have significantly higher induction than uninfected cells (Fig. 4a). They included members of the IFIT group, chemokines, cytokines, interferon regulatory factors, NF-κB, OASL, MyD88 and STAT genes. Interestingly, another gene associated with prostaglandin function (prostaglandin-endoperoxide synthase 2 or PTGS2) was also induced and, along with a similar observation in infected SK-N-SH cells, suggests that this gene may play a role in the innate response to WNV. In the neuronal cultures, the gene expression induced by infection with Boort and NSW2012 strains was not significantly different to that of the other strains (Additional file 4: Figure S1).
Pathway scores were calculated using the gene induction data and are presented as a heat-map (Fig. 5 and Additional file 3). As with the SK-N-SH cells, uninfected neuronal cells had consistently lower pathway scores than infected cells. A characteristic of the neuronal cells was that gene induction in the WNV infected cells was less variable across isolates and replicates. However, there were no visibly obvious differences that were consistent among all 4 replicate infections between the more and less virulent WNVKUN isolates. In contrast to the SK-N-SH cell infections, MVEV showed a more activated phenotype having consistently higher scores in such pathways as G protein coupled receptor protein signaling pathway, G couple protein receptor binding, chemokine activity, chemokine receptor binding, and cytokine activity. The above data show that the cell environment is an important determinant in the innate immune response to viruses.