Powassan Viruses Spread Cell to Cell during Direct Isolation from Ixodes Ticks and Persistently Infect Human Brain Endothelial Cells and Pericytes

ABSTRACT

Powassan viruses (POWVs) are neurovirulent tick-borne flaviviruses emerging in the northeastern United States, with a 2% prevalence in Long Island (LI) deer ticks (Ixodes scapularis). POWVs are transmitted within as little as 15 min of a tick bite and enter the central nervous system (CNS) to cause encephalitis (10% of cases are fatal) and long-term neuronal damage. POWV-LI9 and POWV-LI41 present in LI Ixodes ticks were isolated by directly inoculating VeroE6 cells with tick homogenates and detecting POWV-infected cells by immunoperoxidase staining. Inoculated POWV-LI9 and LI41 were exclusively present in infected cell foci, indicative of cell to cell spread, despite growth in liquid culture without an overlay. Cloning and sequencing establish POWV-LI9 as a phylogenetically distinct lineage II POWV strain circulating in LI deer ticks. Primary human brain microvascular endothelial cells (hBMECs) and pericytes form a neurovascular complex that restricts entry into the CNS. We found that POWV-LI9 and -LI41 and lineage I POWV-LB productively infect hBMECs and pericytes and that POWVs were basolaterally transmitted from hBMECs to lower-chamber pericytes without permeabilizing polarized hBMECs. Synchronous POWV-LI9 infection of hBMECs and pericytes induced proinflammatory chemokines, interferon-β (IFN-β) and proteins of the IFN-stimulated gene family (ISGs), with delayed IFN-β secretion by infected pericytes. IFN inhibited POWV infection, but despite IFN secretion, a subset of POWV-infected hBMECs and pericytes remained persistently infected. These findings suggest a potential mechanism for POWVs (LI9/LI41 and LB) to infect hBMECs, spread basolaterally to pericytes, and enter the CNS. hBMEC and pericyte responses to POWV infection suggest a role for immunopathology in POWV neurovirulence and potential therapeutic targets for preventing POWV spread to neuronal compartments.

IMPORTANCE We isolated POWVs from LI deer ticks (I. scapularis) directly in VeroE6 cells, and sequencing revealed POWV-LI9 as a distinct lineage II POWV strain. Remarkably, inoculation of VeroE6 cells with POWV-containing tick homogenates resulted in infected cell foci in liquid culture, consistent with cell-to-cell spread. POWV-LI9 and -LI41 and lineage I POWV-LB strains infected hBMECs and pericytes that comprise neurovascular complexes. POWVs were nonlytically transmitted basolaterally from infected hBMECs to lower-chamber pericytes, suggesting a mechanism for POWV transmission across the blood-brain barrier (BBB). POWV-LI9 elicited inflammatory responses from infected hBMEC and pericytes that may contribute to immune cell recruitment and neuropathogenesis. This study reveals a potential mechanism for POWVs to enter the CNS by infecting hBMECs and spreading basolaterally to abluminal pericytes. Our findings reveal that POWV-LI9 persists in cells that form a neurovascular complex spanning the BBB and suggest potential therapeutic targets for preventing POWV spread to neuronal compartments.

INTRODUCTION

Powassan virus (POWV) is a neurovirulent tick-borne flavivirus (FV) that was first isolated in 1958 from a fatal encephalitic case of human disease in Powassan, Ontario, Canada (1). POWV has recently emerged in ticks in the northeastern United States, with host reservoirs in white-footed mice, deer, skunks, woodchucks, and squirrels (2–5). In humans, POWV enters the central nervous system (CNS), lytically infects neurons, and causes severe encephalitis, with an ∼10% mortality rate and long-term neurologic sequelae occurring in 50% of infected patients (6–9). POWV awareness and diagnostic testing of patients have revealed an increasing number of human POWV encephalitis cases that suggest a high incidence of previously undiagnosed POWV disease, especially in areas where Lyme disease is endemic. Approximately 2% of deer ticks (Ixodes scapularis) in the highly populated Long Island (LI), NY, area are POWV positive (6). Currently there are no approved vaccines or therapeutics for POWV.

The presence of POWV in tick saliva accounts for POWV transmission to hosts or humans in as little as 15 min after a tick bite (8–10). POWV transmission is followed by a 1- to 5-week incubation period, a week of acute febrile illness, and weeks to months of encephalitic manifestations of fever, headache, and confusion that worsen into meningoencephalitis with seizures, impaired consciousness, coma, and respiratory failure (9). In autopsies of POWV patients, neurons in the central nervous system (CNS) were infected with POWV antigen and RNA within lesions in the cerebellum, brainstem, basal ganglia, and thalamus (11). The emergence of neurovirulent POWV in highly populated areas, rapid tick-to-human transmission, and the severity of POWV disease provide urgency for understanding how POWVs spread to neuronal compartments and defining potential therapeutic targets.

POWV is the only North American tick-borne flavivirus, and POWVs are distantly related to tick-borne encephalitis virus (TBEV), the leading cause of arthropod-borne encephalitis in Europe and Asia (12). There are two genetic lineages of POWV: lineage I includes the LB strain derived from the 1958 human encephalitis case, and lineage II strains that were sequenced or isolated from North American deer ticks. POWV I and II strains share 94% amino acid identity and represent POWV genotypes carried by discrete ticks and mammalian hosts (13). Prototype lineage I and II POWVs were first isolated following intracranial inoculation of mice and subsequent passage in Vero cells (1, 8, 14, 15). However, there are no studies of POWV infection of cells that comprise the blood-brain barrier (BBB) and normally prevent viral access to neuronal compartments.

How POWV spreads from a tick bite site to the CNS, persists to cause a late acute febrile illness, or enters protected neuronal compartments remains to be determined. Reservoir cells where POWV persists following infection and mechanisms by which POWV crosses the BBB to enter the CNS remain enigmatic. There are no reports of POWV infection of human brain microvascular endothelial cells (hBMECs) or pericytes or cellular responses to POWV or TBEV infections. TBEV was found to infect only 2 to 5% of hBMECs, was primarily apically released, and failed to lyse ECs, alter EC barrier functions, or spread within EC monolayers (16). In mice, TBEV reportedly enters the CNS without permeabilizing the BBB, and at late stages of infection induces proinflammatory cytokine and chemokine responses when high viral loads are already present in the brain (17).

hBMECs have intimate basolateral contacts with pericytes, forming a neurovascular unit that plays a critical role in regulating BBB permeability and immune cell recruitment. POWV infection of hBMECs and brain pericytes has not been investigated. In this study, we isolated POWVs from Long Island deer ticks and evaluated their ability to replicate, persistently infect, and spread in VeroE6 cells, primary hBMECs, and brain pericytes. We collected 438 deer ticks from Long Island, and from 44 pools of 10 ticks, we found 3 POWV-positive pools by quantitative reverse transcription-PCR (qRT-PCR). Tick homogenates inoculated into 44 wells of VeroE6 cells revealed 2 wells (9 and 41) with POWV antigen-positive cells 7 days postinfection (dpi), both coinciding with PCR positivity. We observed that direct inoculation of VeroE6 cells with LI tick homogenates resulted in large POWV-infected cell foci consistent with cell-to-cell spread, despite the absence of restrictive overlays. POWV-LI9 and POWV-LI41 were propagated from supernatants, and POWV-LI9 was cloned and sequenced, identifying it as a distinct lineage II POWV strain.

We found that POWV-LI9 and POWV-LI41 strains productively and persistently infected VeroE6 cells, hBMECs, and human pericytes nonlytically, reaching respective POWV-LI9 titers of 8 × 10⁵, 1 × 10⁴, and 5 × 10⁴ focus-forming units (FFU)/mL. POWV-LI9 failed to permeabilize polarized hBMECs or VeroE6 cells, but was preferentially released basolaterally from cells (3- to 12-fold) to lower-chamber pericytes in a BBB-like Transwell model. Prototypic POWV-LB acquired from the ATCC has been passaged extensively in Vero cells since 1958 (1), and similar analysis of POWV-LB infection of VeroE6 cells resulted in a few small clusters of infected cells but primarily spread uniformly within the monolayer. However, like POWV-LI9, we found that POWV-LB also productively infected hBMECs and pericytes, and POWV-LB was basolaterally released from polarized hBMECs (3-fold over apical titers). These findings suggest a potential mechanism for lineage I and II POWVs to infect hBMECs and pericytes and cross the BBB.

Lectin pretreatment of hBMECs permitted POWV-LI9 to synchronously infect hBMECs (>90%), and we used this approach to evaluate hBMEC and pericyte responses to POWV infection. POWV induced interferon beta (IFN-β), CCL5 (>3,000-fold), an array of inflammatory chemokines, and IFN-stimulated genes (ISGs) 1 to 3 dpi, with conserved and novel responses between POWV-infected hBMECs and pericytes. IFN-β induction and secretion were restricted in POWV-infected pericytes, but not hBMECs, 1 dpi, but IFN-β was highly secreted by both hBMECs and pericytes 2 dpi. Pretreatment of VeroE6 cells, hBMECs, or pericytes with type I IFN prevented POWV infection. Despite IFN-β induction and sensitivity, a subset of hBMECs and pericytes were persistently infected with POWV-LI9, 12 to 30 dpi and through cell passage. These findings differ from Zika virus (ZIKV), which spreads and replicates continuously in hBMECs in the absence of secreted IFN-β (18). Our results reveal novel POWV-LI9 regulation of hBMEC and pericyte responses and suggest that hBMEC-pericyte neurovascular complexes may serve as POWV reservoirs and conduits for POWV to spread basolaterally across the BBB and into neuronal compartments.

RESULTS

POWV-LI9 isolation from Ixodes scapularis ticks.

In November 2020, we collected 438 I. scapularis adult ticks, in Suffolk County, Long Island, NY. Homogenates of 44 pools of 10 ticks were prepared in phosphate-buffered saline (PBS) by Dounce homogenization, and following centrifugation, supernatants were inoculated into VeroE6 cells and hBMECs in 24-well plates and screened for POWV RNA by qRT-PCR. Three tick pool inocula were PCR positive for POWV RNA. One week after inoculation, infected cell supernatants were harvested, and cell monolayers were immunoperoxidase stained using anti-POWV hyperimmune mouse ascites fluid (HMAF) antibody (ATCC). Coincident with RNA-positive tick pools, VeroE6 cell wells 9 and 41 had POWV antigen-positive infected cells 7 dpi (Fig. 1A).

FIG 1.

POWV-LI9 isolation from Ixodes scapularis ticks. (A) I. scapularis ticks were collected in Long Island, NY. Tick homogenates from groups of 10 ticks were added to wells of 24-well plates containing VeroE6 cells. After 7 days, cells were fixed and immunoperoxidase stained using a specific anti-POWV HMAF (1:10,000). Wells 9 and 41 were positive for POWV antigen and formed distinct infected cell foci. First passage (P1) virus was inoculated into a well of a 24-well plate, and cells were fixed and immunoperoxidase stained using a specific anti-POWV HMAF (1:10,000) after 7 days (left panels). The right panels show the magnification of the cell foci from initial viral isolation and P1 infection of VeroE6 cells. Bars represent 50 μm. (B) VeroE6 cells were infected with POWV-LI9 (P3) at an MOI of 5 or mock infected, and the cells were fixed at 1, 3, 5, 8, and 10 dpi and immunoperoxidase stained using anti-POWV HMAF (1:10,000). Bars represent 50 μm. (C) Viral titration from POWV-infected VeroE6 supernatants was determined by FFU assay at 1 dpi in VeroE6 cells. (D) VeroE6 cells were infected with POWV-LI41 (P2) at an MOI of 5, and cells were fixed at 1, 2, 3, and 6 dpi and immunoperoxidase stained using anti-POWV HMAF (1:10,000). Bars represent 50 μm. (E) VeroE6 cells were infected with POWV-LB at an MOI of 5, and cells were fixed at 1, 2, 3, and 6 dpi and immunoperoxidase stained using anti-POWV HMAF (1:10,000). The inset shows a cluster of infected cells without a discrete focal phenotype. Bars represent 50 μm. (F) Viral titers were determined by FFU assay 1 dpi in VeroE6 cells.

Despite growth in liquid culture, POWV antigen-positive VeroE6 cells were present in large infected cell foci, consistent with cell-to-cell spread, without apparent cell lysis (Fig. 1A). Infection of VeroE6 cells with passage 1 POWV-LI9 or POWV-LI41 stocks resulted in smaller, more dispersed infected cell foci (Fig. 1A). A time course of POWV-LI9 spread in VeroE6 cells from 1 to 10 dpi shows that virus spread from initial infected cell foci 3 to 5 dpi to uniformly infect monolayers 6 to 10 dpi (Fig. 1B). There was no evidence of cytopathic effect (CPE) in POWV-LI9-infected VeroE6 cells, with ∼100% of VeroE6 cells still infected 10 to 13 dpi (Fig. 1B). Supernatant titers of POWV-LI9 in VeroE6 cells reached 8 × 10⁵ focus-forming units (FFU)/mL (Fig. 1C), and VeroE6 cells remained highly infected following cellular passage (>10⁵ FFU/mL). Similar initial foci and later spread were observed when POWV-LI41 infected VeroE6 cells (Fig. 1D). Infection of VeroE6 cells with extensively passaged lineage I POWV-LB (ATCC 1958) resulted in some small clusters of 3 to 12 infected VeroE6 cells 1 dpi (Fig. 1E) and uniform infection of VeroE6 cell monolayers with titers of 1 × 10⁶ FFU/mL (Fig. 1E and F). As POWV spread in tissue culture cells following direct tick inoculation has not been previously examined, it remains to be seen if initial cell-to-cell spread is unique to POWV-LI9/LI41 or a fundamental attribute of direct tick inoculation of POWVs in mammalian cells. Overall, these findings demonstrate the direct isolation and productive replication of POWV-LI9 and POWV-LI41 from LI ticks in VeroE6 cells.

POWV-LI9 is a discrete lineage II strain of POWV present in LI deer ticks.

The POWV-LI9 isolate was passaged in VeroE6 cells for 7 to 10 days, with supernatants collected for viral stocks, and we cloned and sequenced the POWV-LI9 genome from RNA isolated from passage 3 POWV-LI9-infected VeroE6 cell lysates. Randomly primed cDNAs were amplified into ∼2-kb fragments, using conserved POWV primers, and cloned into pMiniT 2.0. Clones encompassing the complete POWV genome were sequenced with 5′ random amplification of cDNA ends (5′ RACE), and a 3′-end oligonucleotide was used to amplify noncoding POWV-LI9 genome termini (GenBank accession no. MZ576219) (19).

Sequences of POWV-LI9 were phylogenetically compared to 15 full-length POWV lineage II sequences, 10 POWV lineage I sequences, and the sequence of TBEV (20, 21). Phylogenetic tree analysis revealed that the POWV-LI9 isolate is a lineage II POWV strain that forms a distinct clade by comparison with other northeastern New York and Wisconsin POWV lineage II sequences (Fig. 2A) (20–22). Sequence alignment of POWV strains with POWV-LI9 revealed an average 99% nucleotide and amino acid identity with POWV genomes collected in the northeastern region (Fig. 2B, group A), 93.5% nucleotide and 97% amino acid identity with POWV collected in Wisconsin (Fig. 2B, group B), and 85% nucleotide and 94.7% amino acid identity with POWV lineage I (Fig. 2B, group C). Novel POWV-LI9 residues include envelope (Env) protein residues S119 and K151 and NS2B (F125), NS3 (D402), and NS5 (T282) residues. The POWV-LI41 isolate is derived from discrete tick homogenates but has yet to be sequenced to determine if it differs from POWV-LI9.

FIG 2.

Genetic analysis of POWV-LI9 isolates. (A) Maximum likelihood tree analysis between POWV-LI9- and POWV-related strains using complete genome sequences (84). Branch numbers are bootstrap confidence estimates based on 100 replicates. Tick-borne encephalitis virus (TBEV) is shown as an outgroup. (B) Multiple-sequence analysis comparison using Clustal Omega (87) between POWV-LI9 and other POWV sequences deposited in GenBank. POWV genomes were divided into three groups based on the genetic identity compared with POWV-LI9: sequences from the viruses collected in the northeastern United States in group A, those from Wisconsin in group B, and sequences from lineage I POWV in group C. Black bars within the sequences represent nucleotide differences compared to POWV-LI9. (C) Among 5 residue differences with northeastern POWVs, there are 2 not-conserved residues in the envelope (Env), including domain II residues S119 and K151 in POWV-LI9 and additional unique residues in the NS2B (F125), NS3 (D402) and NS5 (T282) proteins of POWV-LI9. (D) C57BL/6 mice (Jackson Laboratory, 10 weeks old; n = 3 to 4) were anesthetized via intraperitoneal injection with 100 mg/mL ketamine and 20 mg/mL xylazine per kg of body weight. Mice were footpad inoculated with 20 μL of 1 × 10³ FFU POWV-LI9 or mock infected with saline. Sera were collected 12 dpi and analyzed by Western blotting for detection of envelope proteins from POWV-LI9-, POWV-LB-, or mock-infected VeroE6 cell lysates (3 dpi) compared to anti-POWV HMAF antibody (ATCC). (E) Neutralizing antibodies present in POWV-LI9-, POWV-LB-, or mock-infected mouse serum were determined by serial dilution and addition to ∼500 FFU of POWV-LI9 or POWV-LB prior to adsorption to VeroE6 cells. Inoculated VeroE6 cells were washed, and 1 dpi, cells were methanol fixed and immunoperoxidase stained, and infected cells were quantitated.

In initial experiments, POWV-LI9 (10³ FFU) was inoculated into the footpad of C57BL/6 mice, and sera collected 12 dpi were found to elicit antibody responses that detect envelope proteins from POWV-LI9- and POWV-LB-infected VeroE6 cells by Western blotting (Fig. 2D) and neutralize POWV-LI9 and POWV-LB viruses (Fig. 2E). Collectively, these findings establish that POWV-LI9 is a lineage II POWV strain newly discovered in Ixodes scapularis ticks on Long Island that infects mice and elicits neutralizing antibody responses to lineage I and II POWVs.

POWVs productively infect primary hBMECs and pericytes.

Mechanisms by which POWVs enter neuronal compartments in order to lytically infect neurons have not been investigated. Here, we assessed the ability of POWV-LI9 to infect, replicate and persist in primary human BMECs and pericytes that form a neurovascular complex within the BBB (23–26). We found that POWV-LI9 productively infects hBMECs 1 to 8 dpi, with low viral titers and a subset of hBMECs remaining persistently infected 6 to 8 dpi (Fig. 3A and B). Infection of hBMECs with POWV-LI9 at a multiplicity of infection (MOI) of 5 only marginally increased the number of infected hBMECs 1 dpi (10 to 15% maximum), suggesting that entry was restricted to a subset of hBMECs.

FIG 3.

POWV-LI9 productively and persistently infects human BMECs and pericytes. (A) hBMECs were infected with POWV-LI9 (P3) at an MOI of 5 or mock infected, and cells were fixed at 1, 2, 3, 6, and 8 dpi and immunoperoxidase stained using anti-POWV HMAF (1:10,000). Bars represent 50 μm. (B) Viral titration from POWV-infected hBMEC supernatants was determined by FFU assay at 1 dpi in VeroE6 cells. (C) POWV-LI9 was pretreated with WGA (1 μg/mL) for 5 min. Control mock-WGA-treated medium (1 μg/mL) or WGA-treated POWV was adsorbed to hBMECs for 1 h, and subsequently cells were PBS washed and resupplemented with EBM2 to 2% FBS medium. Cells were fixed and immunoperoxidase stained using anti-POWV HMAF (1:10,000) at 1 dpi. (D) Human brain pericytes (hPCs) were infected with POWV-LI9 (P3) at an MOI of 5 or mock infected, and cells were fixed at 1, 2, 3, 6, and 8 dpi and immunoperoxidase stained using a specific anti-POWV HMAF (1:10,000). Bars represent 50 μm. (E) Viral titers from POWV-infected pericyte supernatants were determined by FFU assay at 1 dpi in VeroE6 cells. (F) hBMECs and hPCs were infected with POWV-LB at an MOI of 5 or mock infected, and cells were fixed at 1, 2, 3, and 6 dpi and immunoperoxidase stained using anti-POWV HMAF (1:10,000). Bars represent 50 μm. (G) Viral titration from POWV-infected hBMECs supernatants was determined by FFU assay at 1 dpi in VeroE6 cells. (H) Viral titration from POWV-infected hPCs supernatants was determined by FFU assay at 1 dpi in VeroE6 cells. (I) hBMECs were grown for 5 days in Transwell inserts and apically infected with POWV-LB (MOI, 5) in triplicate. Cells were washed, and 3 dpi, viral titers from both apical and basolateral compartments were determined in VeroE6 cells.

To determine if attachment limited hBMEC infection, we treated POWV-LI9 with wheat germ agglutinin (WGA; 1 μg/mL, 5 min) prior to infecting hBMECs, a method previously used to direct efficient HIV infection of ECs (27–29). Pretreatment of POWV with WGA prior to adsorbing POWV-LI9 to cells resulted in a synchronous ∼95% infection of hBMECs (1 dpi) (Fig. 3C). The ability of WGA-treated POWV to efficiently direct the infection of hBMECs suggests that POWV attachment and entry, rather than postattachment replication, are restricted to a subset of hBMECs and permits analysis of hBMEC responses to a synchronous POWV infection.

In BBB capillaries, human brain pericytes are attached to the basolateral side of hBMECs and form abluminal contacts with astrocytes and neurons (23–25, 30, 31). Pericyte responses and interactions with hBMECs control the integrity and immune cell recruiting functions of the BBB (31). We found that POWV-LI9 infected primary human brain pericytes, spreading from 1 to 3 dpi to infect ∼80% of pericytes (Fig. 3D) and produce titers of 5 × 10⁴/mL (Fig. 3E) without apparent cytopathology. From 3 to 8 dpi, POWV titers were reduced in pericyte supernatants, but a high percentage of pericytes remained POWV infected 8 to 30 dpi and after pericyte passage. Unlike POWV-LI9 infection of VeroE6 cells, we found that POWV-LI9 infection of pericytes failed to form infected cell foci and instead infected single cells that spread ubiquitously. Similar to POWV-LI9 findings, we found that POWV-LI41 (data not shown) and POWV-LB also productively infect hBMECs and pericytes (Fig. 3F to I). Collectively, these findings demonstrate that lineage I and II POWVs infect hBMECs and pericytes, which form neurovascular complexes within the BBB, and thereby suggest potential mechanisms for POWVs to enter neuronal compartments.

Preferential basolateral release of POWVs from hBMECs directs pericyte infection.

hBMECs form polarized monolayers with apical and basolateral surfaces that mimic luminal and abluminal capillary surfaces (24). The ability of POWV-LI9 to infect hBMECs prompted us to determine whether POWV spreads basolaterally across polarized hBMECs. Neither POWV-infected hBMECs, pericytes, nor VeroE6 cells showed signs of cytopathic effects, and 3 dpi, POWV-LI9-infected cells were nearly all viable by calcein AM staining (green indicates live cells) with little propidium iodide (PI) uptake (red indicates dead cells) (Fig. 4A). Consistent with this, we found no apparent disruption of tight junction protein ZO-1 in cell-to-cell contacts of POWV-LI9-infected hBMECs (Fig. 4B).

FIG 4.

POWV-LI9 is basolaterally released from cells without barrier disruption. (A) Polarized hBMECs and VeroE6 cells grown for 5 days in Transwell inserts were apically infected with POWV-LI9 (MOI, 5). Cells were visualized by phase-contrast microscopy for CPE and costained with calcein AM (green indicates live cells) and propidium iodide (red indicates dead cells) to visualize cell viability 3 dpi. (B) Confocal immunofluorescence. VeroE6 cells grown on microslides were infected with POWV-LI9 (MOI, 5). Cells were fixed 2 dpi, immunostained with anti-POWV HMAF and anti-ZO-1 antibodies, DAPI counterstained, and visualized by confocal microscopy. Experiments were done in triplicate, repeated at least three times, and representative data are presented. Bars represent 10 μm. (C) Polarized hBMECs and VeroE6 cells grown for 5 days in Transwell inserts were apically infected with POWV-LI9 (MOI, 5) in triplicate, and TEER was measured 1 to 3 dpi. (D) hBMECs and (E) VeroE6 cells, grown for 5 days in Transwell inserts, were apically infected with POWV-LI9 (MOI, 5) in triplicate. Cells were washed, and viral titers from both apical and basolateral compartments were determined 3 dpi. (F) Schematic representation of hBMECs and the human pericyte coculture model (left). hBMECs were seeded on the apical side of Transwell inserts and grown for 7 days confluent by TEER resistance. hBMECs were mock or POWV infected from the apical side for 1 h, cells were washed, and 1 dpi, Transwell inserts were transferred to 24-well plates containing monolayers of primary human pericytes. Pericytes were immunoperoxidase stained for POWV antigen 3 dpi using anti-POWV HMAF (1:10,000). Bars represent 50 μm.

To determine whether POWVs permeabilize hBMECs or VeroE6 cells, we analyzed polarized cells grown on Transwell inserts for changes in transendothelial/epithelial electrical resistance (TEER). We found no significant change in TEER of POWV-infected VeroE6 cells or hBMECs 1 to 3 dpi versus uninfected cells, indicating that POWV infection did not permeabilize infected hBMEC monolayers (Fig. 4C). Infection of hBMECs and VeroE6 cells on polarized Transwell inserts permitted analysis of apical and basolateral titers following POWV infection. We found that POWV-LI9 titers in basolateral supernatants were enhanced >3-fold in VeroE6 cells and >12-fold in hBMECs over apically released virus (Fig. 4D and E). Analysis of POWV-LB found that POWV titers were enhanced 3-fold in basolateral versus apical supernatants (Fig. 3D).

POWV basolateral release from hBMECs suggested the potential for POWV to directly infect pericytes. We POWV-LI9 infected polarized hBMECs on Transwell inserts, transferred inserts to wells containing lower-chamber pericytes (1 dpi), and found that pericytes were infected by basolaterally released POWV (Fig. 4F). Preferential basolateral release of POWV from hBMECs suggests a potential mechanism for POWVs to spread across the BBB to abluminal pericytes that reside within neuronal compartments.

Interferon pretreatment blocks POWV infection and spread.

To determine if POWV infections are IFN sensitive, we pretreated VeroE6 cells, hBMECs, and pericytes with type I IFN (IFN-α) 2 to 18 h prior to POWV-LI9 infection and quantitated infected cells 1 dpi. We found that pretreatment of cells with IFN-α nearly completely abolished POWV infection (Fig. 5A), indicating that preexisting or induced type I IFN prevents POWV infection and spread in hBMECs and pericytes. We analyzed POWV-infected hBMEC and pericyte supernatants for IFN-β by enzyme-linked immunosorbent assay (ELISA) and found that POWV-infected hBMECs highly secreted IFN-β 1 dpi, while IFN-β secretion by POWV-infected pericytes was not apparent until 2 dpi (Fig. 5B and C). Secreted IFN-β responses are consistent with the early restriction of POWV-LI9 spread in hBMECs and continued POWV spread in pericytes 1 to 3 dpi (Fig. 5D). In both cell types, IFN-β secretion is dramatically reduced 3 dpi, suggesting a potential mechanism for POWVs to inhibit late IFN expression and persist in infected hBMECs and pericytes.

FIG 5.

POWV-LI9 is susceptible to exogenous interferon addition. (A) VeroE6 cells, hBMECs, and human brain pericytes (hPCs) were pretreated with IFN-α (1,000 U/mL) for 2 or 18 h and before POWV infection (MOI, 5). One day postinfection, cells were methanol fixed and immunoperoxidase stained using anti-POWV HMAF (1:10,000), and the number of infected cells was quantitated and compared to that in untreated controls. (B) hBMECs and (C) hPCs were mock or POWV-LI9 infected (MOI, 5), and 1 to 3 dpi, supernatants of cells were assayed for secreted IFN-β by ELISA (R&D Systems).

POWV-infected hBMECs and pericytes elicit immune-enhancing responses.

hBMECs and basolateral pericytes form a functional unit that determines the integrity of the BBB through synergistic responses that regulate vascular permeability, cell proliferation, and immune cell transit across the endothelium (30, 31). To evaluate transcriptional changes of POWV-infected hBMECs, we WGA treated POWV to synchronously infect 85 to 90% of cells 1 dpi and compared transcriptional responses to those of WGA-treated, mock-infected, hBMECs. RNAs from control and POWV-infected hBMECs were analyzed by Affymetrix arrays 1 to 3 dpi. POWV infection of hBMECs resulted in the dramatic induction of chemokines (CCL5, CXCL10, CCL20, and CXCL11), IFN-β, IFN-stimulated gene family proteins (ISGs: RSAD2, OASL, IFIT2/3, MX2, ISG15), IFN response factors (IRFs 1/2/7), transcription factors (ATF3 and EGR1), and coagulation factor B over mock-infected hBMEC controls (Fig. 6A). Selected POWV-directed hBMEC transcriptional changes are presented in Fig. 5A, and complete data are available in the NCBI GEO database (accession no. GSE176251). POWV induction of high levels of IFN-β, CCL5, and ISG15 in hBMECs was verified by qRT-PCR (Fig. 6B).

FIG 6.

Global transcriptional changes in POWV-LI9-infected hBMECs and pericytes. (A) POWV-LI9 was pretreated with WGA (1 μg/mL) for 5 min. Control mock-WGA-treated medium (1 μg/mL), or WGA-treated POWV was adsorbed to hBMECs for 1 h, and subsequently, cells were PBS washed, resupplemented with EBM2 to 5% FBS medium, and grown for 1 to 3 days. Total cellular RNA from mock-WGA-treated and WGA-treated POWV-infected hBMECs was harvested in RLT buffer, purified, DNase treated, and subjected to Affymetrix array analysis. Log₂-induced genes from POWV-infected samples were compared to respective mock-infected time points, generating a heat map with 0- to 100-fold induction scale, where induced genes above 100-fold are in black. Genes were separated into functional categories as those encoding chemokines, interferons (IFN), ISGs, IRFs, cytokines (Cytok.), or transcriptional factors (TFs) and apoptosis-related (Apop.) and ISGylation-related (ISGyl.) genes. (B) Selected genes from the Affymetrix array analysis were confirmed by qRT-PCR. (C) Total RNA from mock- and POWV-infected hPCs was isolated at 1 to 3 dpi and subjected to Affymetrix array analysis as in panel A. Log₂-induced genes from POWV-infected samples were compared against their respective mock-infected time point, generating a heat map with a 0- to 500-fold (left graph) or 100-fold (right graph) induction scale, where genes induced above 500- or 100-fold are represented as black. Genes were separated in functional categories as those encoding chemokines, interferons, ISGs, IRFs, cytokines, or transcriptional factors and apoptosis-related, ISGylation-related, and complement-related genes. (D) Selected genes induced by POWV infection of pericytes by Affymetrix array were confirmed by qRT-PCR. (E) Gene Ontology (GO) enrichment analysis was conducted by querying the upregulated genes from infected hBMECs or hPCs 1, 2, and 3 dpi using STRING v11 (32). GO terms from selected biological processes are presented.

We also analyzed human brain pericyte transcriptional responses following synchronous POWV-LI9 infection (MOI f 5) compared to mock-infected pericytes (NCBI GEO accession no. GSE176252). ISGs were the most highly induced responses observed in POWV-infected pericytes 1 dpi, with 100- to 1,700-fold increases over a wide array of ISGs (Fig. 6C). Despite high ISG responses, IFN-β was only induced 9-fold 1 dpi, but increased dramatically to 463-fold by 2 dpi and 281-fold 3 dpi. Chemokines CCL5 and CXCl10 were moderately induced 93- to 156-fold 1 dpi but were induced 1,800- to 3,300-fold 2 to 3 dpi (Fig. 6C). Complement-activating factors (CFB, C3, C1S, C1R, and C7), IL-6 (111- to 256-fold 2 to 3 dpi), and IL-1β (10- to 19-fold) were highly induced by POWV infection of pericytes 2 to 3 dpi. Despite high-level IFN-β and ISG responses, POWV productively replicated and spread in pericytes 1 to 3 dpi, with persistent infection apparent >8 dpi (Fig. 3D and E). The induction of selected chemokines, cytokines, and ISGs by POWV infection was validated by qRT-PCR (Fig. 6D). Our findings indicate that POWV infection of human pericytes transcriptionally induces chemotactic cytokine and complement responses that are known to enhance inflammation and immune cell recruitment to the endothelium. How POWV spreads and persists in pericytes in the presence of high-level IFN and ISG responses at later times after infection remains to be investigated.

Pathway Gene Ontology (GO) analysis (32) of hBMECs and pericyte transcriptomes showed similar GO enriched responses related to cytokine production, leukocyte migration, type I IFN signaling, ISG15-protein conjugation, cell death and apoptosis regulation, angiogenesis, and EC migration (Fig. 6E). Regulation of prosurvival extracellular signal-regulated kinase 1/2 (ERK1/2) signaling, which was recently shown to be required for ZIKV persistence in hBMECs (33), was specifically associated with POWV-induced hBMEC transcriptomes 2 to 3 dpi (Fig. 6E, left panel). ERK1/2 pathway activation is consistent with CCR3/CCR5 activation by high levels of CCL5 induced by POWV-infected hBMECs and pericytes (33). Pericyte-specific responses included cell cycle, NF-κB signaling, antigen processing and presentation, and complement activation 2 to 3 dpi (Fig. 5E, right panel).

Collectively, chemokine responses elicited by POWV-infected hBMECs appear to be synergistic, with robust chemokine and IFN responses induced by POWV-LI9-infected pericytes. These findings indicate that POWV infection of pericytes is likely to enhance immune cell recruitment to hBMECs that may contribute to immune-enhanced pathogenesis. In contrast to IFN secretion 1 dpi by hBMECs, POWV-infected pericytes inhibit IFN secretion 1 dpi, yet both cell types induce late IFN and ISG responses that fail to prevent POWV persistence. These findings suggest that POWV differentially regulates early IFN responses in pericytes and hBMECs and suggest that at later times, a subset of POWV-infected hBMECs and pericytes may become resistant to IFN and ISG responses.

DISCUSSION

POWV is transmitted to humans in tick saliva during a short ∼15-min tick bite (34). POWVs cause a severe neurovirulent disease that is fatal in 10% of cases and results in long-term neurologic deficits in 50% of patients (8). The rapid transmission of POWV is unlike Lyme disease, which requires long-term tick attachment and for which antibiotics are available to clear the infection. For POWV, there are no therapeutic treatments or approved vaccines to prevent neurovirulent infection and disease. POWV-carrying ticks and human POWV infections have increased in the northeastern United States (22, 35, 36). Climate change is expected to exacerbate future POWV incidence by expanding the range and density of I. scapularis and mammalian disease reservoirs, in addition to increasing tick survival and seasonal activity (37). A recent study found a 2% prevalence of POWV in Ixodes scapularis ticks on Long Island, NY (6), and the spread of POWV and the severity of disease rationalized evaluating POWVs circulating in this densely populated area. In this report, we isolated POWVs from I. scapularis ticks on LI and demonstrate that POWVs (LI9, LI41, and LB) infect human brain microvascular endothelial cells and pericytes that form a neurovascular complex that spans the BBB. Our findings suggest the potential for persistently infected hBMECs and pericytes to be reservoirs of POWV spread and suggest a mechanism for POWVs to cross the BBB and enter neuronal compartments.

We isolated POWVs (LI9 and LI41) from 3/44 PCR-positive LI tick pools by directly inoculating tick homogenates into VeroE6 cells. Cloning and sequencing of the POWV-LI9 tissue culture isolate revealed it to be a phylogenetically distinct lineage II POWV strain present in LI Ixodes scapularis ticks. The POWV-LI9 isolate is 99% identical to POWV genomes collected in Rhode Island and Maine (Fig. 2A) (20–22) and shares 95% amino acid identity with the prototypic 1958 lineage I POWV-LB strain (Fig. 2B) (20–22). However, other than sequencing, the isolation and tissue culture analysis of lineage II northeastern POWV strains have yet to be reported.

We isolated POWV-LI9 and POWV-LI41 by directly infecting VeroE6 cells with tick homogenates and observed large POWV-infected cell foci in liquid medium 7 dpi. In this setting, viruses are normally dispersed in cell supernatants and spread to single cells throughout the monolayer. The observation that POWV-LI9 and -LI4 form foci indicates that these POWVs spread cell to cell. This is novel and had not been previously evaluated or reported for any POWV (38). In subsequent POWV-LI9 passages, focal spread observed in passages 1 and 2 changed to a more uniform infection of VeroE6 cells. Whether cell-to-cell spread is a fundamental feature of POWVs directly inoculated into mammalian cells has yet to be considered and remains to be determined as there are no studies of POWV spread following direct tissue culture isolation from ticks. Prototypic POWV-LB and POWV-SP strains were isolated by intracranial inoculation of mice prior to extensive passage in Vero cells (1, 8, 14, 15). Prior studies of POWV-LB-infected VeroE6 cells did not report infected cell foci, but also lacked analysis of POWV spread in infected cells. We found that POWV-LB uniformly infected VeroE6 cells, with a few grouped infected cells (Fig. 1E) but no large infected cell foci, indicative of cell-to-cell spread.

Initial cell-to-cell spread of POWV strains remains an enigma that requires analysis of more POWVs by direct isolation in tissue culture and the analysis of potential POWV genetic changes during isolation and early passage. These findings also raise the question of whether POWVs are initially restricted to cell-to-cell spread following a tick bite and whether this contributes to the long prodrome of POWV disease. These findings are fundamental to initial POWV spread and may need to be considered in studies of POWV spread and neuropathogenesis in murine models that are uniformly based on POWVs that were first passaged in mouse brain.

The ability of POWV-LI9 and -LI41 strains to form infected foci represents new and novel observations, as there are no reports of cell-to-cell spread for other mosquito- or tick-borne flaviviruses (38). Focal cell-to-cell spread has been shown for several viruses, including herpesvirus, rhabdovirus, vaccinia virus, HIV-1, and the flavivirus hepatitis C virus (HCV) (38–44). Neurovirulent measles virus forms infectious foci in primary human airway epithelial cells, with viral spread dependent on intercellular adherens junction-localized pores (42). Similar to POWV, measles virus foci form without cytopathology, cell-to-cell fusion, or syncytium formation (42). Additional experiments that differentiate POWV cell-to-cell spread, basolateral release, or spread via intercellular membrane pores are required to understand how POWV-LI9 forms foci and to define mechanisms of POWV spread from infected hBMECs to the CNS.

How POWVs enter the CNS remains unknown, and there is no information on cellular targets of POWV infection or the ability of POWVs to infect and replicate in cells of the BBB. The BBB is a highly selective semipermeable barrier that separates the circulating blood from the CNS and prevents pathogens from entering the brain (45). The BBB is composed of brain microvascular endothelial cells, pericytes, astrocytes, and microglia (31, 45, 46). Pericytes form intimate basolateral contacts with both BMECs and astrocytes, forming a neurovascular unit that regulates functions of luminal endothelial cells and abluminal pericytes and astrocytes within the neuronal compartment (25, 47).

There are no prior studies of POWV infection of hBMECs and pericytes. We found that POWV-LI9, POWV-LI41, and POWV-LB infect hBMECs, resulting in single infected cells with little cell-to-cell spread after initial infection. This differs from POWV-LI9 infection of VeroE6 cells, and it remains to be determined if these differences are the result of hBMEC receptors or IFN-restricted spread that is absent in VeroE6 cells. POWVs productively infected hBMECs without apparent cell lysis or permeabilizing effects (Fig. 4C). However, a comparison of POWV titers from apical and basolateral sides of polarized VeroE6 or hBMECs in Transwell plates demonstrated that POWV-LI9 is preferentially released from the basolateral side (3- to 12-fold) (Fig. 4D and E). Similar analysis of POWV-LB found both apical and basolateral spread from infected hBMECs, with a 3-fold increase in basolateral spread (Fig. 3I). Only a few studies demonstrate preferential basolateral viral spread (48–50). Vesicular stomatitis virus (VSV) and Marburg virus (MARV) bud from the basolateral side of polarized MDCK cells (48–50), and arenaviruses are basolaterally released from airway epithelial cells (51). Basolateral spread of POWVs suggests a mechanism for POWVs to cross the BBB and spread abluminally to neuronal compartments.

In the brain, pericytes form a neurovascular complex with hBMECs spanning the BBB and contacting astrocyte end feet, microglia, and neurons (30, 31, 52). POWV infection of hBMECs prompted us to analyze whether pericytes are infected and whether hBMECs convey POWV infection to pericytes. We found that POWV-LI9, -LI41, and -LB highly infect primary human brain pericytes (Fig. 3). In a BBB-like in vitro model, POWV-LI9-infected hBMECs also basolaterally transmitted POWV to pericytes in the lower chamber (Fig. 4F). Our findings demonstrate that POWVs productively and persistently infect hBMECs and pericytes and suggest hBMEC-pericyte complexes as potential POWV reservoirs and conduits for viral spread to neuronal compartments.

Neurotropic flaviviruses, including ZIKV, Japanese encephalitis virus (JEV), West Nile virus (WNV), and TBEV, are reportedly transmitted across polarized ECs without permeabilizing or lysing cells (16, 18, 53, 54). In TBEV-infected mice, BBB permeability was not required for viral entry into the CNS, and permeability was only observed at late stages of infection and after high viral loads and inflammatory damage were evident in the brain (17). In innate immunity-compromised mice, neurotropic flaviviruses invade the CNS during early viremic stages (55), with infection of the BBB one of the most likely routes of neuronal invasion (45).

Many murine POWV pathogenesis models use POWV-LB and -SP strains that were first passaged in mouse brains (14, 56) and IFNAR antibody blockades to promote lethal outcomes (57, 58). Initial inoculation of POWV-LI9 strain in wild-type WT C57BL/6 mice resulted in murine seroconversion and antibodies that neutralized both POWV-LI9 and POWV-LB (Fig. 2D and E). It remains to be determined if POWV-LI9 and -41 strains are neurovirulent in mice. However, analysis of POWV-LI9 viremia, CNS spread, pathology, and neurovirulence in mice, at early or late passage, is beyond the scope of this initial study of newly isolated POWV strains circulating on LI.

POWV-LI9’s persistence in hBMECs and pericytes is consistent with the delayed onset of neurologic disease in POWV patients and the long-term pathology of POWV infection (4, 8, 59). However, POWV persistence is very different from the persistent infection of hBMECs by ZIKV (18, 33). ZIKV replicates and spreads persistently in hBMECs without permeabilizing or lysing cells, and persistence continues through serial cellular passage (18). Recent findings indicate that ZIKV persistence requires the suppression of IFN secretion and the high-level secretion of CCL5, which acts as an autocrine activator of extracellular signal-regulated kinase (ERK1/2) survival signaling responses (33).

In contrast to ZIKV, POWV infection of hBMECs does not result in the constitutive spread of POWV throughout the monolayer. ZIKV, dengue virus (DENV), and POWV are inhibited by pretreating cells with type I IFN (18, 60). However, ZIKV bypasses this restriction by blocking IFN-β secretion from infected hBMECs, while DENV is cleared from infected ECs by IFN-β and fails to establish a persistent infection (18, 60). Although POWV spread is restricted by secreted IFN-β, a subset of infected hBMECs remain persistently infected, viable, and POWV antigen positive. Curiously, POWV inhibits pericyte secretion of IFN-β 1 dpi, permitting early viral spread within pericytes. This suggests novel POWV regulation of early IFN responses in pericytes and suggests that POWV resistance to IFN secreted 2 to 3 dpi permits POWV to persistently infect cells. It remains unknown whether POWV proteins, generated early in infection, permit POWV persistence or resistance to IFN/ISG responses hBMECs or pericytes or how secreted IFN-β levels 3 dpi are restricted, despite highly induced IFN-β transcripts. Collectively these findings indicate that POWV persists in hBMECs and pericytes without a total blockade of IFN-β secretion observed during ZIKV infection (18).

BMECs basolaterally secrete platelet-derived growth factor (PDGF), which recruits and stabilizes pericyte binding to BMECs, and reciprocally, pericytes secrete factors that act on BMEC receptors, which regulate barrier integrity, angiogenesis, and immune cell recruitment (25, 47). Pericytes regulate the BBB, and targeting PDGFs or pericyte-BMEC interactions using imatinib or nintedanib has been suggested as a potential therapeutic approach to control damage of this neurovascular complex (25, 47, 61). Our analysis of POWV-LI9-directed changes in hBMEC and pericyte transcriptional responses included finding the induction of several prosurvival genes, including those coding for EGR1, ATF3, and BIRC3 (62–64), as well as the gene coding for chemokine CCL5, a cell survival response required for ZIKV persistence in hBMECs (33). Highly induced CCL5 activates ERK1/2 cell survival pathways in ECs (33) that may regulate antiapoptotic and proliferative responses (65) and contribute to the viability of POWV-infected hBMECs and pericytes.

Several chemokines were upregulated in both cell types; however, CCL5 and CXCL10 were the most highly induced, especially by POWV-infected pericytes (Fig. 6). CXCL10 is prominently upregulated in the cerebrospinal fluid (CSF) of human patients and mouse models during TBEV and WNV infections (66–70), and higher CXCL10 expression in the brain is correlated with more severe TBEV disease in mice (71). CCL5 and CXCL10 also play an important role in recruiting activated T cells into sites of tissue inflammation (72), and these chemokines have been implicated in both viral clearance and the neuropathology of neurotropic flavivirus infections (73). CCL5 is produced by several cell types that impact inflammation in the CNS, including hBMECs, pericytes, astrocytes, microglia, and neurons (73, 74). Increased CCL5 has been detected in the CSF, but not in the serum, of TBEV patients and TBEV-infected mice (68, 75). During TBEV infections, CCL5-directed neuroinflammation may contribute to brain pathology as treatment of mice with CCL5 antagonists prolonged survival and reduced brain inflammation after TBEV infection (76). Similarly, the CCL5 receptor antagonist maraviroc reduced JEV-induced CNS inflammation and increased the survival of JEV-infected mice (77). CCR5 knockout mice were also protected from fatal WNV or JEV encephalitis (70, 78). ZIKV persistence in hBMECs was recently shown to require high-level CCL5 secretion to support prosurvival responses, and CCL5 receptor inhibitors, which prevent ZIKV persistence in hBMECs, are suggested as therapeutics that clear ZIKV infections (33).

We were surprised to find that complement-related factors CFB (149-fold), C3 (27-fold), C1r (30-fold), and C1s (14-fold) were highly induced by POWV-infected pericytes. Complement proteins in the brain can be induced by astrocytes and microglia in response to injury (79), but little is known about brain pericytes actively producing complement proteins. WNV and HIV-1 induce complement activation within the CNS (80, 81), and elevated C1q levels were found in the CSF of WNV- and TBEV-infected patients (82) and were suggested to contribute to neurocognitive impairment in recovering patients (83). Our findings suggest that POWV-infected pericytes might enhance complement activation and C3a and C5a production, which contribute to immune cell chemotaxis, and T cell priming responses involved in both viral clearance and POWV neuropathology.

In this study, we demonstrate that POWVs productively infect and are basolaterally released from brain endothelial cells, suggesting a potential mechanism by which POWVs infect pericytes and cross neurovascular complexes within the BBB. A subset of hBMECs and pericytes are persistently infected by POWV for 8 to 30 days, providing a novel location for the replication and spread of a neurovirulent virus to the CNS. The abluminal location of pericytes in association with BBB hBMEC makes them ideally suited to convey POWV to the CNS and recruit immune cells to the BBB that contribute to neuronal immunopathogenesis. The ability of POWV to persist in hBMECs and pericytes also suggests that these cellular niches are potential therapeutic targets for preventing POWV neurovirulence.

MATERIALS AND METHODS

Cells.

VeroE6 cells (ATCC CRL 1586) were grown in DMEM (Dulbecco’s modified Eagle’s medium) supplemented with 5% fetal bovine serum (FBS) and penicillin (100 μg/mL), streptomycin sulfate (100 μg/mL), and amphotericin B (50 μg/mL; Mediatech) at 37°C and 5% CO₂. hBMECs (passages 4 to 10) were purchased from Cell Biologics (H-6023), grown in EC basal medium-2 MV (EBM-2 MV; Lonza) supplemented with EGM-2 MV SingleQuots (Lonza), and incubated at 37°C and 5% CO₂. Human brain vascular pericytes (hPCs; passage 1) were purchased from ScienCell (no. 1200), grown in pericyte medium (PM; no. 1201) supplemented with pericyte growth supplement (no. 1252), and incubated at 37°C and 5% CO₂.

POWV-LI9 isolation from Ixodes scapularis ticks.

Ticks were collected during the fall season peak (October to November) throughout Suffolk County, New York, in 2020. Questing ticks were collected from vegetation along trails by flagging a 1-m² cotton flannel fabric attached on both ends to a wooden pole between 10:00 and 14:00 h during sunny days. A total of 438 adult Ixodes scapularis ticks were identified morphologically on a dissecting microscope for taxonomic keys. Pools of 10 ticks were homogenized using an Eppendorf tube mortar and pestle in 500 μl of phosphate-buffered saline (PBS), and debris was removed by centrifugation for 5 min at 12,000 × g.

qRT-PCR for tick screening.

Tick pool homogenates were tested for POWV RNA with a multiplex one-step quantitative reverse transcription-PCR (qRT-PCR) using a forward primer (GTGATGTGGCAGCGCACC), a reverse primer (CTGCGTCGGGAGCGACCA), and a probe (CCTACTGCGGCAGCACACACAGTG) (6). The final reaction mixture contained 5 μL of template and 20 μL of master mix. The master mix contained 0.2 μM forward primer, 0.2 μM reverse primer, 0.1 μM probe, 1.25 μL of RNA UltraSense enzyme mix, and 5 μL RNA UltraSense 5× reaction mix. The reverse transcription step was performed at 55°C for 15 min, followed by incubation at 95°C for 10 min. The PCR consisted of 40 cycles at 95°C for 15 s and 60°C for 30 s. The qPCR was performed on a Bio-Rad C1000 Touch system with a CFX96 optical module (Bio-Rad) using the RNA UltraSense one-step quantitative RT-PCR system protocol (Invitrogen).

Inoculation of mammalian cells.

Tick pool homogenates were inoculated onto individual 24-well or 6-well plates of VeroE6 cells (150 μl/well) in DMEM (supplemented with 2% FBS, 100 μg/mL penicillin; 100 μg/mL streptomycin sulfate, and 50 μg/mL amphotericin B). Plates were transferred to the Stony Brook biosafety level 3 (BSL3) facility for growth at 37°C and 5% CO₂. At 7 to 10 dpi, cell supernatants were harvested, cell monolayers were 100% methanol fixed, and POWV-infected cells were detected by immunoperoxidase staining with anti-POWV HMAF sera (ATCC). Supernatants positive for POWV RNA by qRT-PCR and POWV antigen staining in infected cells were inoculated into T25 flasks containing VeroE6 cells and propagated for 7 days in DMEM with 2% FBS.

POWV-LI9 infection and immunoperoxidase staining.

POWV-LI9 was adsorbed to ∼60% confluent hBMEC or human pericyte monolayers for 1 h. Following adsorption, monolayers were washed with PBS and grown in supplemented EBM-2 MV or PM with 5% FBS, respectively. VeroE6 cells were similarly infected, washed, and supplemented with DMEM with 8% FBS. POWV-LI9 titers were determined by serial dilution and infection of VeroE6 cells, quantifying infected cell foci at 24 hpi by immunoperoxidase staining with anti-POWV hyperimmune mouse ascites fluid (HMAF; 1:5,000 [ATCC]), horseradish peroxidase (HRP)-labeled anti-mouse IgG (1:2,000; KPL-074-1806), and 3-amino-9-ethylcarbazole (AEC) staining (60). For the IFN-α inhibition assay, medium was supplemented with 1,000 U/mL IFN-α (Sigma-Aldrich) at indicated times, and cells were incubated at 37°C and 5% CO₂.

POWV-LI9 cloning and sequencing.

Total RNA was purified using RNeasy (Qiagen) from early-passage POWV-infected VeroE6 cells (passage 3). cDNA synthesis was performed using a Transcriptor first-strand cDNA synthesis kit (Roche) using random hexamers as primers (25°C for 10 min, 50°C for 60 min, and 90°C for 5 min). The POWV genome was amplified using primer pairs to regions located at or near the junctions between capsid (forward, AGGAGAACAAGAGCTGGGAGTGGTC) and envelope (reverse, TGCTCCGACTCCCATTGTCATCAT), NS1 (forward, GACTATGGATGTGCAGTTGATCC) and NS2B (reverse, TCTGCGTGCTGATGAGAAGA), NS3 (forward, ACTGACCTTGTATTCTCAGGG) and NS4B (reverse, GTGCATGAGTTCAACCGTTG), and NS5 (forward, CTAGAAGGGGTGGAGCAGAGG; reverse, TTAGATTATTGAGCTCTCTAGCTTG). The 5′ untranscribed region (5′ UTR) was obtained using the Template switching RT enzyme mix (NEB no. M0466) following the manufacturer’s protocol. The 3′ UTR was obtained using the reverse primer (GCGGGTGTTTTTCCGAGTCACACAC). PCR fragments were obtained using Phusion polymerase (NEB no. M0530L) with 37 cycles at 98°C for 30s, 68°C for 30 s, and 72°C with 30 s per kb. PCR fragments were gel purified (NEB no. T1020L) and cloned into pMiniT 2.0 vector (NEB no. E1202S). The POWV-LI9 genome was sequenced using vector and internal primers, assembling both strands at least twice (GenBank accession no. MZ576219).

Phylogenetic analysis.

A phylogenetic tree of POWV sequences was generated by the neighbor-joining method (84) with a maximum composite likelihood model and bootstrap test of phylogeny with 100 replications (85) in MEGA version X (86). Phylogenetic analysis was performed with the POWV-LI9 sequence and reference full genome sequences representing POWV lineages I and II, with TBEV as the outgroup, obtained from NCBI, as follows (with accession numbers in parentheses): for lineage I, LB (L06436), LEIV-5530 (KT224351), Ternay (HQ231415), Pow-24 (MG652438), Ulysses (HQ231414), Spassk-9 (EU770575), Nadezdinsk-1991 (EU670438), LEIV-3070Prm (KT224350), POWPa06 (EU543649), and POWANY64-7062 (HM440563); for lineage II, RTS81 (MG647779), RTS82 (MG647780), RTS84 (MG647783), RTS92 (MG647781), RTS96 (MG647782), NSF001 (HM440559), MeW17-228 (MK309362), P0375 (KU886216), LI-1 (KJ746872), DTVMeC17 (MK104144), CTB30 (AF311056), DTVWi99 (HM440558), DTVWiA08 (HM440560), DTVWiB08 (HM440561), DTVWiC08 (HM440562), and TBEV (NC_001672).

Murine inoculation.

C57BL/6 mice (10 weeks old) were purchased from Jackson Laboratory. Mice were anesthetized via intraperitoneal injection with 100 mg/mL ketamine and 20 mg/mL xylazine per kg of body weight. Mice were footpad inoculated with 20 μL of 1 × 10³ FFU POWV-LI9 or mock infected with saline. Sera were collected 12 dpi and analyzed by Western blotting for detection of envelope proteins from POWV-LI9 and POWV-LB and for neutralizing antibodies by serial dilution of sera with 500 FFU of POWV-LI9 or POWV-LB.

Viability assay.

Uptake of propidium iodide (PI; Calbiochem) and calcein AM (Invitrogen) was used to evaluate VeroE6 and hBMEC viability as previously described (18). Briefly, POWV-infected cells (MOI, 5) or mock-infected cells were seeded into 24-well plates and costained with the membrane-permeable dye calcein AM (3 μM; green fluorescence in live cells) and 2.5 μM propidium iodide (red fluorescent DNA stain to detect dead cells) 3 dpi. Images of calcein AM-positive versus PI-positive cells were resolved using an Olympus IX51 microscope and Olympus DP71 camera.

Confocal immunofluorescence.

POWV-LI9 (MOI, 5) was adsorbed to 80% confluent VeroE6 monolayers cultured in Lab-Tek II chambers (Nunc), washed with PBS, and grown in DMEM with 8% FBS during 48 h. Cells were washed with PBS, fixed for 10 min with 4% paraformaldehyde–PBS, and permeabilized with 0.1% Triton X-100 in PBS for 10 min. Cells were blocked using 5% bovine serum albumin in PBS for 2 h and incubated with anti-POWV HMAF (1:10,000) and anti-ZO1 rabbit monoclonal antibody (1:100) in blocking solution for 18 h at 4°C. Cells were washed and incubated for 2 h with Alexa 488-conjugated goat anti-mouse IgG antibody (Invitrogen) and with Alexa 546-conjugated goat anti-rabbit IgG antibody (Invitrogen) diluted 1:400 in blocking solution at room temperature. hBMECs were subsequently incubated with 5 μM 4′,6-diamidino-2-phenylindole (DAPI; Sigma) for 5 min at room temperature. Slides were mounted using ProLong antifade (Thermo Fisher) solution and observed using a Zeiss LSM 510 META/NLO confocal microscope.

Affymetrix gene array analysis.

Primary hBMECs or human pericytes (passages 4 to 10) were synchronously infected with POWV-LI9 (MOI, 5) or mock infected. For hBMECs only, POWV-LI9 or equivalent medium was incubated with WGA (1 μg/mL) for 5 min prior to adsorbing POWV to cells for 1 h or mock infecting cells with equivalent amounts of WGA as controls. Following adsorption, cells were washed with PBS, and medium was replaced. Total RNA was purified 1 to 3 dpi from mock- or POWV-infected hBMECs or hPCs using RLT lysis buffer and RNeasy columns (Qiagen). Purified RNA was quantitated, and transcriptional responses were detected on Affymetrix Clariom-S chip arrays in the Stony Brook Genomics Core Facility. POWV-infected cell transcriptional responses were compared to those in mock-infected cells harvested at each time point, and fold changes in POWV versus control transcripts were analyzed using Affymetrix TAC software. Data obtained from these studies were submitted to the NCBI Gene Expression Omnibus database.

qRT-PCR analysis.

Quantitative real-time PCR was performed on purified cellular RNAs derived from mock- or POWV-infected hBMECs/hPCs as described above. cDNA synthesis was performed using a Transcriptor first-strand cDNA synthesis kit (Roche) using random hexamers as primers (25°C for 10 min, 50°C for 60 min, and 90°C for 5 min). qRT-PCR primers for specific genes were designed according to the NCBI gene database, with 60°C annealing profiles (provided by Operon). Genes were analyzed using PerfeCTa SYBR green SuperMix (Quanta Biosciences) on a Bio-Rad C1000 Touch system with a CFX96 optical module (Bio-Rad). Responses were normalized to internal glyceraldehyde-3-phosphate dehydrogenase (GAPDH) mRNA levels, and the fold induction was calculated using the threshold cycle (2^−ΔΔCT) method for differences between mock- and POWV-infected RNA levels at each time point.

Transendothelial/epithelial electrical resistance.

hBMECs or VeroE6 cells were plated on Costar Transwell inserts (3-μm pore size; Corning) at high density, and 7 days postseeding, monolayers were analyzed for transendothelial/epithelial electrical resistance (TEER) (EVOM2, STX3; World Precision Instruments, Inc.). Confluent Transwell cultures were infected with POWV-LI9 (MOI, 5) or mock infected, and TEER values for POWV-infected versus mock-infected cells were compared at 1 to 3 dpi. For analysis of apical and basolateral release of POWV-LI9, hBMECs and VeroE6 cells were similarly infected 3 dpi prior to analyzing viral titers in apical and basolateral supernatants.

POWV transcytosis assay.

hBMECs were plated on Costar Transwell inserts (3-μm pore size; Corning) at high density, and 7 days postseeding, hBMECs were mock or POWV infected in the upper chambers for 1 h, the cells were washed with PBS, and the inserts were transferred to 24-well plates containing 70% confluent primary human brain vascular pericytes.

ELISA.

IFN-β in the supernatants of mock- and POWV-infected hBMECs or hPCs at 1, 2, and 3 dpi was measured using a DuoSet ELISA (R&D Systems). ELISA plates (Immunolon 2, U-bottom; Dynatech Laboratories) were coated with anti-IFN-β capture antibody according to the manufacturer. Viral supernatants were incubated with antibody-coated plates (2 h), washed with PBS (0.1% Tween 20), and bound protein was detected with IFN-β-specific antibodies conjugated to streptavidin-HRP and developed using tetramethylbenzidine. Levels of secreted IFN-β in samples were compared to a purified IFN-β (R&D Systems) standard curve using a BioTek EL312e microplate reader (450 nm).

Statistical analysis.

The results shown in each figure were derived from two to three independent experiments with comparable findings; the data presented are means ± standard errors of the means (SEM), with the indicated P values of <0.01 and <0.001 considered significant. Two-way comparisons were performed by two-tailed analysis of variance and an unpaired Student's t test. All analyses were performed using GraphPad Prism software version 9.0.

Data availability.

Sequencing data from POWV-LI9 were submitted to GenBank (accession no. MZ576219). Microarray data obtained from these studies were submitted to the NCBI Gene Expression Omnibus database (accession no. GSE176251 for hBMECs and GSE176252 for hPCs).