RNA-seq analyses of adult and weanling brain microvessels identify candidate gene regulators of LACV infection
We first sought to identify candidate gene regulators that contribute to age-dependent susceptibility to LACV. Weanling (W) and adult (A) mice were treated with either poly I:C (PI) for 3 h (as a surrogate for a viral RNA stimulus) or vehicle control (V). Whole brain microvessel fragments isolated from these mice were subjected to RNA-seq analysis to identify both intrinsically age dependent or Poly I:C stimulation induced genes (Fig. 1a). Gene expression was compared for basal level age-related differences comparing weanling vehicle (WV) versus adult vehicle (AV) groups. Activation comparisons were done by comparing Weanling poly I:C stimulated (WPI) versus WV, Adult poly I:C stimulated (API) versus AV as well as WPI versus API. Differential gene expression volcano plots and pathway enrichment analyses (Ingenuity Pathway Analysis, IPA), from the above comparisons highlight the cellular processes that distinguish the different treatment and age groups (Supplementary Fig. 1a–d). Specific genes for further evaluation were selected based on differential expression between groups with a log2 difference of ≥ 1, p < 0.05 and base-mean ≥ 100 (Supplementary Table 1, Group 1). To evaluate the likely cell composition in the isolated microvessel fragments, we assessed expression of cell type-specific genes from the RNA seq data (Supplementary Table 2). This showed that BCEC-specific genes were highly enriched, there was moderate/low expression of pericyte markers, but very little expression of genes characteristic of astrocytes, neurons, and smooth muscle cells. This confirmed that the method used for microvessel fragment preparation resulted in highly specific BCEC enrichment, as described previously18,19,20.
As regional differences in BBB integrity are observed following LACV infection and virus infection is first observed in the OB region in weanling mice16, we also completed an RNA-seq analysis on microvessel fragments isolated from different regions of LACV-infected (L) and mock-inoculated (M) weanling mouse brains at 3 dpi. Genes from this comparison were selected based on two differential gene expression criteria: either a difference in expression between microvessel fragments from the OB of LACV-infected mice (LOB) versus the OB of mock-infected mice (MOB), or a difference in gene expression between LOB versus microvessel fragments from CT of LACV-infected mice (LCT) (Fig. 1b). These criteria were used to select additional genes for further analysis (Supplementary Table 1, Group 2). Differential gene expression and IPA analyses from these comparisons highlight the expected enrichment of interferon and neuroinflammation following LACV infection of the OB, and differences in Gap Junction and Ephrin signaling in the OB/CT comparison (Supplementary Fig. 1e, f). Finally, in addition to the selected Group 1 and 2 genes, we also included genes known to affect BBB function and the host immune response to LACV, but not necessarily differentiated in the above RNA seq analysis (Supplementary Table 1, Groups 3 and 4, respectively).
To further confirm expression differences of the selected genes, we analyzed expression by qPCR following LACV infection. RNAs were extracted from microvessel fragments obtained from LACV-infected adult (AL) or weanling (WL) mice at 3 dpi as well as mock-inoculated weanling (WM) and adult (AM) mice, while additional microvessel fragments were extracted from OB and CT regions of LACV infected (3 dpi) weanling and adult mice. Real-time PCR analyses on ex vivo isolated microvessel fragments showed genes falling into four main categories, adult-enhanced, weanling enhanced, LACV-induced and region specific (OB-CT) (Fig. 1c–f). For example, ephrinA2 (Efna2; an angiogenic molecule) and H2q6 (belonging to MHC class I family) were expressed at higher levels in adult microvessel fragments, regardless of virus infection (Fig. 1c). Bone marrow stromal cell antigen 1 (Bst1; an immune regulator) and matrix metallopeptidase 25 (Mmp25) had higher basal levels in weanling mice (Fig. 1d). C-type lectin domain family 4, member e (Clec4e; a molecule that recognizes LACV and plays a limited role in early antiviral responses against LACV21) and claudin 1 (Cldn1; a TJ protein present in endothelial/epithelial cells) were increased by LACV infection in adult mice, but not in weanling mice (Fig. 1e). Only a few genes were differentially expressed between OB and CT microvessel fragments. Among these, two homeostasis maintenance genes aquaporin 1 (Aqp1; a passive transporter) and transthyretin (Ttr; a carrier protein) were increased in CT microvessel fragments from both weanling (WCT) and adult (ACT) mice compared to OB microvessel fragments obtained from weanling (WOB) and adult (AOB) mice. The pairwise comparison between the OB and CT microvessel fragments are shown for each age group (Fig. 1f). Thus, we identified genes by RNA seq (Fig. 1a, b) and qPCR analyses (Fig. 1c–f) that were differently expressed in microvessel fragments by age, infection, or region for further study.
Targeted siRNA screen elucidates potential LACV susceptibility or resistance factors in endothelial cells
To directly examine whether the selected genes may influence viral pathogenesis, we utilized an in vitro model of endothelial cellLACV infection. We previously found that, although infected endothelial cells are not readily detectable in vivo, they are infected in an age-dependent manner in vitro and this infection is associated with direct and bystander damage to endothelial cells17. To establish a gene-perturbation cell-based screening assay, we chose a mouse endothelioma cell line, bEnd.3, as these cells could be infected with LACV and were readily amenable to siRNA transfection. To optimize siRNA delivery, we employed lethal and nontargeting (NT) control siRNAs to test different transfection reagents and identify conditions with maximum cell killing with lethal siRNA while maintaining maximum viability with the si-NT control. This identified an optimal delivery protocol using 50 nM siRNA with the Transit TKO transfection reagent (Supplementary Fig. 2a). We then established three control conditions for the screen: a) a baseline condition for LACV infection with NT siRNAs, b) a control for known restriction factors using a combination of siRNA to Ifnar1 and Ifnar2 (si-Ifnar or si-Upreg. Control; upregulated LACV infection) and c) a control for known susceptibility factors using a combination of siRNAs to Clathrin heavy chain and Rab5a (si-CltcRab or si-Downreg. Control; downregulated LACV infection) (Supplementary Fig. 2b, c).
We then screened the 35 selected genes shown in Supplementary Table. 1 along with the three control conditions at three different phases of LACV infection on BCEC monolayers at room temperature. The first was the early phase (6 hpi; Fig. 2a), which would suggest an effect on virus entry. The second was the medial infectious phase (24 hpi; Fig. 2b), suggesting an effect on virus replication and spread through the endothelial cell culture. These kinetic phases have been similarly observed for LACV infection in a different cell line22. The cell-based assay for these first two phases used an anti-LACV antibody to measure viral fluorescence intensity on a per-cell basis. This was determined by measuring LACV sum-intensity (which is impacted by both the number of infected cells as well as the intensity of infection) and then quantitating the ratio to Hoechst area (which normalizes the intensity value to total cell number). The last, late phase assay (72 hpi; Fig. 2c) measured cell loss through LACV-induced cell death. Using these criteria, knockdown of putative restriction factors would be expected to give increased LACV intensity and reduced cell viability, while knockdown of putative susceptibility factors would lead to reduced LACV intensity and higher cell viability. We identified two clusters of genes that showed putative resistance or susceptibility phenotypes (color highlighted in Fig. 2a–c), with representative images of virus upregulation or downregulation shown in Supplementary Fig. 3.
Putative restriction factors identified in the primary screen included Efna2, Clec4e, Gja1 (expressed as Connexin43 protein, abbreviated as Cx43), interferon induced transmembrane protein 3 (Ifitm3, an RNA viral restriction factor), lymphocyte antigen 6 complex locus C2 (Ly6c2) and H2q6. For H2q6, there was a variation between 6 and 24 hrs in phenotype, with increased viral intensity only observed at the early time point (Fig. 2a, b). Clusters of syncytia-like aggregation and multinucleated cells were observed in H2q6 knockdown cultures, which might explain the limitation of the increased viral intensity phenotype to the early time point (white arrows, Supplementary Fig. 3a, b). Gja1 knockdown caused dissociation of the cellular monolayer (white arrow, Supplementary Fig. 3b), consistent with its established role in gap junction integrity. Putative susceptibility factors included vascular endothelial growth factor A (Vegfa), actin-cytoskeleton remodeling protein gelsolin (Gsn), claudin 2 (Cldn2), claudin 5 (Cldn5) and tight junction protein 2 (Tjp2/ZO2) (Fig. 2a–c). Thus, multiple genes were identified that either enhanced or reduced LACV in vitro infection of bEnd.3 endothelial cells.
Putative target genes control susceptibility of endothelial cells to LACV infection in different phases of viral infection
Based on the targeted siRNA screen, we chose seven putative restriction factors (Group1 genes: Bst1, Efna2, H2q6, Clec4e, Gja1, Ifitm3 and Ly6c2) for further validation in primary adult and weanling BCECs. Interestingly, knockdown of all of these genes increased susceptibility of adult primary BCECs to LACV infection (Fig. 3a), while only Ifitm3 knockdown showed a significantly increased infection level in primary weanling BCECs (Fig. 3b), supporting the premise that these genes may contribute to the increased resistance to LACV in adult BCECs.
We also tested putative susceptibility factors for validation in primary culture (Group2 genes: Cldn2, Tjp2, Cldn5, Vegfa, Mmp8, Mmp25 and Gsn). Knockdown of Mmp25, a weanling enhanced gene, led to reduced LACV intensity in both adult and weanling primary BCECs (Fig. S4), suggesting this host gene may facilitate LACV infection. However, only marginal effects were observed for the other putative susceptibility factors in primary BCECs. Since we observed a higher frequency of hit validation for Group1 genes in primary cells, we focused on genes from this group for further analysis.
In an effort to dissect mechanisms of virus inhibition, we next utilized the bEnd.3 cells to examine the stages at which perturbation of the putative restriction factors may limit virus replication. Attachment / entry assays (inducing viral binding at low temperature, see methods section) showed that two genes, Ly6c2 and Ifitm3, restricted viral entry (Fig. 3c). Compared to si-Ly6c2, si-Ifitm3 knockdown caused smaller upregulation of infected cell count as well as LACV intensity/cell (Fig. 3c, d). We then conducted plaque assays and found that perturbation of Ly6c2, Gja1 and Efna2 increased virus production at 24 hpi (Fig. 3e).
In our primary screen, we had noted that knockdown of the H2q6 gene also induced syncytia-like aggregation formation in bEnd.3 cells (Supplementary Fig. 3a, b). To investigate this further, control and H2q6-perturbed bEnd.3 cells were co-stained for both LACV and F-actin. It has been reported previously that actin cytoskeletal proteins are altered in OB BCECs following LACV infection16. We observed that LACV colocalized with the actin network in control cells, specifically at the cellular periphery and lamellipodial extensions (thin arrow, upper panel, Fig. 3f). In the H2q6 knockdown condition however, specifically in the cells having two or more nuclei (forming syncytia-like aggregation), we observed disruption of regular, filamentous actin staining and granular staining of actin and LACV located around cellular nuclei (thick arrow, bottom panel, Fig. 3f) suggesting a potential alteration of virus trafficking. Thus, in vitro validation of the putative viral restriction factors (Efna2, H2q6, Clec4e, Gja1, Ifitm3) identified from our focused siRNA screen suggests possible roles for select host genes in restricting or altering viral entry, trafficking, or viral particle release in LACV-infected endothelial cells.
Efna2 as a regulator of LACV susceptibility in vivo
One of the primary genes that significantly affected LACV infection of endothelial cells was Efna2, which is intrinsically abundant in adult BCECs. The encoded EFNA2 protein is an angiogenic factor23 that carries out signaling responses via EphrinA (EphA) class receptors24. To further examine EFNA2, we tested whether recombinant mouse EFNA2 (rec-EFNA2) protein could restrict LACV infection in weanling and adult primary BCECs. Weanling and adult primary BCECs were cultured, infected with LACV (10 multiplicities of infection; abbreviated as MOI), then maintained in EFNA2 conditioned or control medium for up to 72 h post infection. Two different EFNA2 concentrations, low (2ug/ml) or high (20 ug/ml) were used. Upon rec-EFNA2 treatment, weanling BCECs showed reduced levels of LACV infection at 24 hpi, which was significant for the high concentration of EFNA2 (Fig. 4a). The adult BCECs, already mostly resistant to LACV infection, showed partial but not significant reduction in infection. Thus, exogenously added EFNA2 can restrict in vitro LACV replication in weanling BCECs (Fig. 4a).
To examine whether Efna2 has a role in restricting LACV infection of the CNS in vivo, we analyzed the development of neurological disease in normally resistant adult mice deficient in Efna family members. We initially tested >6weeks old mice that had mixed deficiency genotypes for Efna2, Efna3 and Efna5 (Efna2−/− mixed; abbreviated as Efna2−/− (m)), as detailed in Supplementary Table 3. All mice were distributed in two groups: homozygous for Efna2 deficiency with mixed +/+, +/− or −/− genotypes for Efna3 and Efna5 (Efna2−/− (m) or heterozygous for Efna2 deficiency with mixed +/+, +/− or −/− genotypes for Efna3 and Efna5 (Efna2+/− (m)) as well as wildtype mice with Efna2+/+3+/+5+/+ genotype (wildtype, abbreviated as WT). Following inoculation of 105 PFU/mouse, approximately 50% of Efna2−/− (m) mice developed neurologic disease, regardless of their Efna3 or Efna5 genotype (Supplementary Table 3). In contrast, most of the WT or Efna2+/− (m) mice (~91%) did not show any signs of neurological disease (Fig. 4b). We also observed high levels of virus in the Efna2−/− (m) mice with clinical (C) neurological disease versus non-clinical (NC) mice (Fig. 4c). Thus, Efna2 appears to have an inhibitory role in LACV pathogenesis in vivo (Fig. 4b, c).
To determine if Efna2 deficiency affects BBB permeability and LACV neuroinvasion, we examined mice at 3 dpi, four days prior to the onset of clinical signs. LACV RNA was detected in approximately ½ of the Efna2 −/− (m) mice, but in only one mouse from the WT and Efna2+/− (m) group (Fig. 4d). These LACV-infected Efna2−/− (m) mice were also injected with ~100 nm red FluoSphere beads (virus-sized particle; red staining) at 3 dpi to monitor vascular leakage. Fluorospheres were primarily detected within CNS blood vessels in WT mice (thick white arrows). However, in one of the four Efna2−/− mice, fluorospheres were detected within the brain parenchyma (thin white arrows, Fig. 4e) suggesting early vascular leakage. Thus, the 25–50% increased disease in Efna2−/− (m) mice correlated with increased incidence of early viral RNA detection in the CNS and detection of vascular leakage in a subset of these mice. These data indicate that Efna2 may be an important restriction factor in adult mice to prevent LACV neuroinvasion.
To clarify whether Efna2 itself was the primary mediator of viral restriction, we compared LACV susceptibility in adult (> 6 weeks old) WT and Efna2−/−3+/+5+/+ mice (Efna2 single knockout (KO); abbreviated as Efna2−/−(s)). As expected, the adult WT mice were not susceptible to a 105 PFU/mouse LACV dose (~96% non-neurologic). In contrast, approximately 38% of adult Efna2−/− (s) mice developed neurological disease in response to LACV infection (Fig. 5a). Furthermore, microvessel fragments isolated from Efna2−/−(s) mice had higher levels of LACV infection compared to WT at 24 and 48 hpi (Fig. 5b, d) and reduced survival at 72 and 96 hpi (Fig. 5c). We also examined vascular leakage in vivo using fluorospheres in WT and Efna2−/− (s) mice using a slightly later timepoint of 5 dpi. WT adult mice showed no vascular leakage, while Efna2−/− (s) mice had consistent detection of 1–3 fluorescent bead foci in the brain parenchyma (Fig. 5e, 5 out of 7 mice). Thus, Efna2−/− (s) mice were more susceptible to LACV-induced neurological disease, which correlated with an increase in vascular leakage in the CNS prior to disease onset. This also correlated with increased virus infection and decreased cell survival in Efna2−/− (s) BCECs compared to WT BCECs (Fig. 5b, c). Collectively, these data indicate an important role for Efna2 in BBB integrity during LACV infection.
Small molecule-based activation of Cx43 protein delays LACV-induced neurological endpoint in weanling animals
Since knockdown of Gja1 (which expresses the Cx43 protein) impacted LACV infection in vitro, we hypothesized that activating or upregulating Cx43 might reduce virus infection and pathogenesis. 4-Phenylbutyric acid (4-PBA) is a known activator and channel formation enhancer for several connexins, including Cx43. We infected bEnd.3 cells with LACV + /− 4-PBA and immunostained for LACV and Cx43 at different time points. 4-PBA treatment significantly reduced LACV infection and increased host cell survival at 96 hpi (Fig. 6a). Using confocal microscopy, we observed that the reduction of LACV infection (LACV: green) at 96 hpi correlated with enhancement of Cx43 puncta formation (Cx43: red). The increase in Cx43 puncta formation (white arrows; denoted by granular or laminar Cx43 staining) or localization toward the cell surface was 4-PBA dose dependent (Fig. 6b)25.As observed in previous studies25, the upregulation of Cx43 protein level was confirmed by western blot (Fig. 6c). Thus, 4-PBA-induced upregulation in Cx43 protein levels or puncta formation correlated with inhibited of LACV infection in vitro.
To examine if 4-PBA could inhibit LACV-induced neurological disease, LACV infected weanling mice (~21–23 days old) were treated daily with 500 mg/kg 4-PBA from 0 to 5 dpi (IP injection, LP). LACV infected and vehicle treated mice (LV) were analyzed in parallel. The LV mice exhibited neurological disease between 6 and 8 dpi. In contrast, LP mice had delayed development of neurological disease (Fig. 6d). At the neurological endpoint, virus levels and Gja1 mRNA expression levels were similar in vehicle and 4-PBA treated groups (Supplementary Fig. 5). However, earlier analysis at 3 dpi, when virus first invades the CNS, demonstrated that LP animals had reduced virus levels, which correlated with an increase in Gja1 mRNA expression (Fig. 6e). At 4 dpi, brains were isolated from LV and LP mice and co-stained for the endothelial marker ZO1 and LACV. LV mice showed persistent and prominent presence of LACV staining in the OB and OT region of the brain (Fig. 6f; upper panels). In contrast, 4-PBA-treated mice had either no detectable LACV (3 of 4 mice) or much more limited viral staining (1 out of 4 mice, (Fig. 6f; lower panels)) compared to vehicle controls. Higher magnification images showed prominent LACV infection in brain parenchyma in a highly vascularized area in LV mice, whereas LACV infection was restricted primarily to the vascular lumen at 4 dpi in LP mice (Fig. 6f, right side panels). Notably, virus particles were restricted inside the vessels without establishing a sufficient leakage inside brain parenchyma at 4 dpi. Thus, 4-PBA treatment induced upregulation of Cx43 and could partially restrict viral infection and dissemination in the OB/OT region at the early stage of LACV infection.
Read more here: Source link