Introduction
Powassan virus (POWV) is a neurotropic tick-borne
Flavivirus (TBFV) responsible for life-threatening meningoencephalitis with necrotizing inflammation and lymphocytic infiltrations in humans (Gholam et al.
1999; McLean and Donohue
1959; Piantadosi et al.
2015). The virus is closely related to deer tick virus (DTV or POWV lineage II), and both belong to the tick-borne encephalitis virus (TBEV) serogroup. POWV was initially described in 1958 in a fatal case involving a 5-year-old boy in the small town of Powassan, Ontario, Canada (McLean and Donohue
1959). Although POWV infections in the USA are sporadic, cases have been reported in several states, such as Connecticut, New York State, New Hampshire, and Massachusetts (Deardorff et al.
2013; Ebel
2010; Hermance and Thangamani
2017; Leonova et al.
2009; Lindsey et al.
2015; Pastula et al.
2016; Piantadosi et al.
2015; Simon et al.
2013). In recent months, additional cases of POWV encephalitis have been reported in the Northeastern United States and some were fatal. POWV is also a significant cause of illness in eastern regions of Russia (Deardorff et al.
2013). The incidence of human cases appears to be on the increase (Deardorff et al.
2013; Ebel
2010; Hinten et al.
2008; Piantadosi et al.
2015). An effective vaccine against the TBEV serogroup is available mainly in Europe, but up to 15,000 new infections continue to be recorded annually, leading to death in close to 40% of infected cases depending on the TBEV strain (Dobler
2010; Heinz and Kunz
2004; Heinz et al.
2013).
In almost all cases, the ticks that have been shown to transmit TBFVs including POWV belong to the
Ixodes genus. These ticks typically feed on small-to-medium-sized mammals, such as white-footed mice or
Peromyscus leucopus, striped field mice (
Apodemus agrarius), skunks (
Mephitis mephitis), and woodchucks (
Marmota monax) (Dupuis II et al.
2013; Ebel
2010; Kim et al.
2008; Main et al.
1979; Mlera et al.
2014; Perkins et al.
2003). Evidence implicating these mammals as reservoirs, bridge, or amplification hosts for TBFVs is inconsistent and includes the isolation of the TBEV strain A104 from the brains of wild-caught yellow-necked mice (
Apodemus flavicollis) in Austria (Frey et al.
2013). The TBEV strains Oshima 08-As and Oshima A-1 were isolated from spleens of captured
Apodemus speciosus, and the Oshima C-1 strain from the gray-backed vole
Clethrionomys rufocanus in Japan (Kentaro et al.
2013; Takeda et al.
1999). A report from South Korea describes PCR detection and TBEV isolation from lung and spleen tissue obtained from wild
Apodemus agrarius mice (Kim et al.
2008). In addition, scientists in Finland detected TBEV RNA in mouse brains, but some mice that had viral RNA were seronegative (Tonteri et al.
2011). Indirect serological evidence suggesting exposure to TBFVs in wild rodents includes the detection of anti-POWV antibodies in wild-caught
Peromyscus truei and
Peromyscus maniculatus in New Mexico,
Myodes rutilus in Siberia and Alaska, and
Myodes gapperi in Southern Alaska (Deardorff et al.
2013). Recent surveys have also found evidence of antibodies against POWV in 4.2% of Eastern US white tail deer, suggesting that virus-infected ticks might also be feeding on this large mammal (Pedersen et al.
2017). In addition to transmitting virus to the mammalian host, infected ticks can also rapidly infect naïve ticks feeding in proximity via a process called “co-feeding” (Labuda et al.
1993a,
1997,
1993b). Thus, the nature of the interaction between arthropod hosts, potential reservoir species, and virus remains uncertain.
Mice belonging to the
Peromyscus genus, particularly
P. leucopus and
P. maniculatus species, are the most abundant mixed-forest rodents in the USA (Bedford and Hoekstra
2015). However, very little is known about the specific role that these mice play in POWV biology. Our initial effort to decipher this role was the development of a
P. leucopus model of POWV infection, which is characterized by a lack of overt clinical signs of disease following intraperitoneal or intracranial challenge (Mlera et al.
2017). However, intracranial challenge of
P. leucopus mice leads to mild subclinical encephalitis at early time points (5 to 15 days post infection (dpi)), but the inflammation resolves by 28 dpi (Mlera et al.
2017). These observations starkly contrasted with inbred laboratory mouse strains C57BL/6 and BALB/c, which succumb to severe and fatal neurological disease upon intracranial inoculation, recapitulating some aspects of human disease (Hermance and Thangamani
2015; Mlera et al.
2017; Santos et al.
2016).
To extend our studies and to characterize the mild encephalitic response in P. leucopus mouse brains, we used RNA sequencing (RNA-Seq) to profile the differential transcriptome changes associated with POWV infection. Our results indicate that POWV induces the differential expression of many genes and the P. leucopus mice mount a robust interferon response against the virus in a tightly regulated manner. These results will be useful for identification of factors that that have a role in restricting POWV.
Discussion
Powassan virus causes life-threatening encephalitis in humans. This virus is transmitted by hard-bodied
Ixodes tick species, and it seems very likely that human infection with POWV is on the increase (Ebel
2010; Piantadosi et al.
2015). The
Ixodes ticks obtain blood meals from small-to-medium-sized mammals, which include
P. leucopus mice, but very little is known regarding the relationship between these animals and POWV. To improve knowledge in this subject matter, we developed a
P. leucopus mouse model of POWV infection in which the virus does not cause obvious clinical signs of disease, but intracranial inoculation resulted in mild inflammation. Evidence of virus replication was restricted in time and space mainly to the olfactory bulb (Mlera et al.
2017). To further characterize the
P. leucopus mouse brain response to POWV, we used RNA-Seq to profile the transcriptome of intracranially inoculated
P. leucopus mice and compared it to mock-inoculated mice.
Results of our RNA-Seq study showed that early after infection, there was a progressive increase in the number of genes that were significantly differentially expressed. The total number of significantly differentially expressed genes peaked to 232 (± 2-fold change in expression levels) at 7 dpi. By 28 dpi, the number of significantly differentially expressed genes had declined to only 28, suggesting a well-controlled transcriptome response to POWV. These results were consistent with our previous observations that infectious POWV was only detectable by culture in the first 7 days of infection (Mlera et al.
2017). It is also interesting to note that intraperitoneally or intracranially inoculated 4-week-old BALB/c mice succumb to POWV disease within the first week of infection, suggesting that there are unique factors of the
P. leucopus mouse response that are expressed early and critical for restricting POWV without extensive pathology in the host.
The emergent theme in the transcriptome profile was that
P. leucopus mouse brains mounted a robust and well-controlled interferon response (Figs.
2 and
3, Tables
3 and
4) that was effective in controlling the spread of infection. This is exemplified by a change in the landscape of IFN-α signaling, which starts with just nine upregulated IFN-stimulated genes at 1 dpi, followed by a dramatic increase to 54 upregulated ISGs by 7 dpi (Fig.
3). Indeed, the IFN response is a well-known potent antiviral host response system, which leads to the transcription of hundreds of virus-incapacitating ISGs (Raftery and Stevenson
2017; Wang et al.
2017). Thus, it was not surprising to see the increase in the genes affected by IFN-α. The IFN response is also activated in the brains of Swiss Webster mice challenged with the mosquito-borne West Nile virus (WNV) or Japanese encephalitis virus, and in the brains of C57BL/6 mice challenged with WNV (Clarke et al.
2014; Kumar et al.
2016). Importantly, unlike
P. leucopus mice infected with POWV, the C57BL/6 and Swiss Webster mice develop widespread infection and neurological symptoms and succumb from disease. A recent report has shown that IFN signaling is associated with restriction of POWV replication in vitro in adult and embryonic
P. leucopus fibroblasts, in comparison to
Mus musculus fibroblast cells (Izuogu et al.
2017). In this report, the authors found that both
P. leucopus and
M. musculus fibroblast cells secrete IFN upon challenge with the tick-borne Langat virus and that knockdown of STAT1 or IFNAR1 increased viral replication in
P. leucopus fibroblasts (Izuogu et al.
2017). Although we were not able perform direct
P. leucopus and
M. musculus comparisons in vivo, results from the WNV studies also suggest that the IFN response is, in and of itself, insufficient to control virus replication. This implies that there may be additional factors that restrict POWV replication and disease induction not typically associated with the control/regulation of IFN signaling or, perhaps indeed, even novel factors completely unrelated to the classical IFN antiviral system. We are currently pursuing studies aimed at decoding the ~ 40% RNA-Seq reads that we were unable to map to a reference genome as a way of deciphering these factors.
The specific ISGs common to early POWV replication included
DDX60,
GBP1,
GBP4,
GBP6, and
MX2; the products of all these genes are antiviral. DDX60 mediates its antiviral properties by binding to RIG-I to promote RIG-I-like signaling (Miyashita et al.
2011). The guanylate-binding proteins (GBPs) are IFN inducible, and GBP1 has been shown to be upregulated with an inhibitory effect during dengue virus infection (Pan et al.
2012). Our results showed that in addition to GBP1, GBP4 and GBP6 are also upregulated during POWV infection. Similar results have been reported for WNV infection in Swiss Webster mice (Clarke et al.
2014).
The use of IPA enables us to observe that TRIM24 was inhibited over the entire course of POWV infection. This was an interesting finding because TRIM24 is proposed to be a negative regulator of IFN signal transducers and activators through retinoic acid receptor alpha (RARA). In TRIM24 knockout cells, many ISGs, genes such as
IFIT2,
IFIT3, and
IFIH1, were found to be TRIM24-dependent (Tisserand et al.
2011). POWV does not replicate to high titers in
P. leucopus mouse brains, and the suppression of TRIM24 suggests that this molecule may be a crucial factor in controlling the IFN response and, subsequently, POWV replication. Thus, further experiments to explore the role of TRIM24 in POWV replication or host response are warranted.
In contrast to TRIM24, genes that code for several other TRIMs were upregulated in expression and these included
TRIM19,
TRIM21,
TRIM25,
TRIM30A, and
TRIM34A (Fig.
2d). There is a cornucopia of TRIMs, and they have varied functions, including cell proliferation, differentiation, as well as antiviral activity (Nisole et al.
2005; Rajsbaum et al.
2008). TRIM79α is rodent-specific and was shown to inhibit the tick-borne Langat virus by targeting the RNA-dependent RNA polymerase (NS5) for lysosomal degradation (Taylor et al.
2011). The precise role of the TRIMs we identified is unclear, and it will be interesting to delineate the specific role played by each one of the TRIMs identified in our study with POWV. In vitro modeling with
P. leucopus cells may prove a more tractable system for elucidating these complicated networks, and we are currently undertaking this work.
The 28-dpi differential gene expression profile was different from the rest of the time points we analyzed. This was not unexpected, considering that the times were far apart and that the persistent POWV RNA was no longer associated with any infectious POWV (Mlera et al.
2017). However, it was interesting that we did not observe any signs of inflammation at 28 dpi by histology (Mlera et al.
2017), but the IPA suggested that some elements of the acute-phase response signaling remained active, indicative of inflammation. In addition, the IPA of the 28-dpi gene set indicated the activation of the LXR/RXR and FXR/RXR systems (Table
3). LXR is a heterodimeric transcription factor involved in cholesterol metabolism, and it also has anti-inflammatory activity (Tall and Yvan-Charvet
2015). LXR has been reported to have antiviral activity for several viruses. For example, stimulation of the LXR with LXR agonists resulted in potent inhibition of HIV replication in a humanized mouse model (Ramezani et al.
2015). Nakajima et al. showed that neoechinulin B inhibits LXR and subsequently inhibits HCV replication because it reduced double-membrane vesicles in which HCV replication occurs (Bocchetta et al.
2014; Nakajima et al.
2016; Zeng et al.
2012). Langat virus and the mosquito-borne Zika virus also cause an expansion of membrane-bound vesicles in the endoplasmic reticulum (Offerdahl et al.
2012,
2017), suggesting that blocking LXR could inhibit POWV replication, a hypothesis that needs further study. In contrast, the LXR/RXR genes seem to have a pro-viral effect in the case of Coxsackie virus B3 (CVB3), and it does not reduce cardiac inflammation in vivo, predisposing mice to mortality upon infection (Papageorgiou et al.
2015). At present, we are uncertain of the role of LXR/RXR and FXR/RXR activation during POWV infection and this merits further study.
In summary, we determined the brain transcriptome profile of P. leucopus mice following intracranial inoculation with POWV and compared the results to those of mock-inoculated animals. There was an increase in the number of genes that were significantly differentially expressed from 1 to 7 dpi, followed by a decline at 15 and 28 dpi. The IPA of the genes at 1 to 15 dpi indicates that P. leucopus mice infected with POWV mount a robust IFN response, which is characterized by the upregulation of many antiviral genes. Some of the induced genes include GBP1, GBP4, GBP6, several TRIMs, and MX2. As mentioned earlier, further studies and development of in vitro systems will prove valuable to delineate restriction factors of POWV, and these results will be useful for studies aimed at the development of POWV antiviral therapies.