Loss of type I IFN signaling drives echovirus-associated mortality and liver infection

To investigate how type I and type III IFNs differentially protect the neonatal liver from echovirus-induced disease, we employed a panel of four transgenic mouse models to capture early acute infection. Building on our prior work showing that adult hFcRn-expressing mice lacking type I IFN signaling (hFcRn^Tg32 Ifnar1^⁻/⁻) are highly susceptible to echovirus infection with hepatocyte-specific viral tropism [31], we extended our analysis to include mice lacking type III IFN signaling (hFcRn^Tg32 Ifnlr1^-/-), as well as a newly generated double knockout strain lacking both type I and type III IFN receptors (hFcRn^Tg32 Ifnar1^-/-Ifnlr1^-/-). For simplicity, we refer to these strains hereafter as Ifnar1^-/- Ifnlr1^-/-, and Ifnar1^-/-Ifnlr1^-/-. Neonatal pups were infected intraperitoneally with echovirus 5 (E5), a representative strain used due to its reliance, exclusively, on hFcRn for infection and not DAF as an additional attachment factor [32]. Based on our prior findings that loss of type I IFN signaling results in dramatically increased mortality [24,31], we optimized a new infection model in which Ifnar1^-/- and Ifnar1^-/-Ifnlr1^-/- mice were infected with 50 PFU of E5, whereas wild-type (Wt), and Ifnlr1^-/- mice received a higher dose of 10⁶ PFU. As expected, Ifnar deficient strains (Ifnar1^-/- and Ifnar1^-/-Ifnlr1^-/-) succumbed to infection by 96 hours post infection (hpi) whereas Wt and Ifnlr1^-/-, mice survived beyond this early acute timepoint (Fig 1A). To capture which liver cell types were participating in the early-acute response to echovirus infection we selected a 40-hour harvest timepoint. At this stage, animals across all four genotypes remained alive, allowing direct comparison of innate immune activity among models. Consistent with survival dynamics, viral titers at 40hpi in the livers and blood of Ifnar1 deficient strains were significantly higher than those observed in Wt or Ifnlr1^-/- strains (Fig 1B). Viral titers did not differ between sex across any of the genotypes tested (S1A Fig). Notably, viral titers were comparable between Ifnar1^-/- and Ifnar1^-/- Ifnlr1^-/- mice, indicating that type III IFN signaling does not provide additional protection in the absence of type I IFNs during echovirus infection.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Type I IFN deficiency increases viral burden and mortality during echovirus infection.

A) Early acute survival curve of four interferon knockout neonatal mouse models infected with echovirus. All mice are on a human neonatal Fc receptor background (hFcRcn^Tg32): green line represents immunocompetent (Wt) genotype, orange lacks the type III IFN receptor (hFcRcn^Tg32Ifnlr1^-/-), blue lacks the type I IFN receptor (hFcRcn^Tg32 Ifnar1^-/-), and the red line represents the double knockout genotype which lacks both type I and type III IFN receptors (hFcRcn^Tg32 Ifnar1^-/- Ifnlr1^-/-). Wt and Ifnlr1^-/-pups were challenged with 10⁶pfu of echovirus5 (E5) via intraperitoneal (IP) infection. Ifnar1^-/- and Ifnar1^-/- Ifnlr1^-/- pups were challenged with 50pfu of E5 via an IP route. B) Viral titers from blood and liver of infected mice, collected 40hpi. Asterisks represent samples with statistically significant differences in viral titers across genotypes as determined by one-way ANOVA tests (p < 0.0001). Limit of detection for the assay is marked by a dashed black line. C) H&E images (scale bar, 50μm) of infected livers at 40hpi across all four genotypes. Ifnar1^-/-and Ifnar1^-/- Ifnlr1^-/- mice have more immune cell infiltrates and hepatocyte damage than Wt and Ifnlr1^-/- pups as indicated by yellow arrows. D) Confocal microscopy images (scale bars, 150μm and 100μm) of echovirus RNA in infected neonatal mouse livers. Hybridization chain reaction (HCR) was used to detect E5 RNA (red) in liver sections from neonatal mice at 40hpi. Ifnar1^-/- and Ifnar1^-/- Ifnlr1^-/- livers showed higher levels of E5 RNA compared to wild-type and Ifnlr1^-/- livers at the same time point. Nuclei are stained with DAPI (grey). E) Luminex protein quantification of type I (Ifnα and Ifnβ) and III (Ifnl2) IFNs from whole liver lysate – same samples used for titer data generation. Values are shown as pg/mL. Asterisks represent samples with statistically significant differences in viral titers across genotypes as determined by one-way ANOVA tests (* p < 0.05, ** p < 0.005, *** p < 0.0005).

https://doi.org/10.1371/journal.ppat.1013891.g001

We next determined the impact of E5 infection on liver structure and integrity across genotypes. Using hematoxylin and eosin (H&E) staining we observed robust immune cell infiltration and hepatocyte destruction in Ifnar1^-/- and Ifnar1^-/- Ifnlr1^-/- models at 40hpi following echovirus infection (Fig 1C). In contrast, infected Wt and Ifnlr1^-/- livers were comparable to uninfected controls (S1B Fig). Finally, we used hybridized chain reaction (HCR) to localize echovirus replication in the liver. Consistent with viral titer data, we observed very low viral RNA in Wt and Ifnlr1^-/- strains (Fig 1D). However, there was high viral RNA burden in both the Ifnar1^-/- and Ifnar1^-/- Ifnlr1^-/- strains.

Finally, to quantify and characterize the IFN response across genotypes during early-acute echovirus infection, we performed Luminex protein analysis on whole liver lysates from the same samples used to generate titer data. All genotypes induced both type I (Ifnα and Ifnβ) and type III (Ifnl2) IFNs (Fig 1E). IFN levels were significantly higher in Ifnar ⁻ / ⁻ mice compared with Ifnar ⁺ / ⁺ controls, reflecting the increased viral replication and consequent accumulation of pathogen-associated molecular patterns (PAMPs) that drive heightened IFN induction in the absence of functional signaling. Together, these findings demonstrate that type I, but not type III, IFN signaling is essential for survival during echovirus infection and for limiting viral replication in the liver.

Single-cell transcriptomics defines echovirus tropism and IFN responses in the liver

To comprehensively define liver cell types susceptible to echovirus infection and characterize their transcriptional responses, we performed scRNA-seq on the livers from neonatal mice across all four genotypes (Wt, Ifnlr1^-/-, Ifnar1^-/-, and Ifnar1^-/- Ifnlr1^-/-). For each genotype, we collected livers from three uninfected and three E5-infected littermates at 40 hpi, resulting in eight experimental conditions. Livers were manually dissociated into single-cell suspensions, and samples from each mouse were processed and sequenced individually. Reads were mapped to a custom reference genome constructed according to the 10X Genomics assembly rules [33]. The E5 genome and the human neonatal Fc receptor (hFcRn) were added as extra chromosomes to the mm10 reference genome from 10X Genomics [34]. A total of 219,980 cells passed quality control thresholds and were visualized using UMAP projections (Fig 2A). To account for technical variation between samples, we applied Harmony for batch correction during integration. In addition, sex, number of detected features (nFeature_RNA), total UMI counts (nCount_RNA), and percentage of mitochondrial transcripts (%mt) were regressed out during data normalization. We then examined cluster composition across genotypes and infection conditions (Fig 2B). Cluster proportions and total cell numbers did not vary significantly by infection status, individual mouse, or sex (S2A, S2B, and S2D Fig), confirming the robustness of the dataset for downstream analyses.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Loss of type I IFN signaling broadens echovirus tropism in the neonatal liver.

A) UMAP of integrated liver datasets from all four IFN knockout models at an early acute timepoint (40hpi). Each genotype is represented by three uninfected and three infected littermates for a total of 8 individual datasets. Cell types are annotated as follows: erythroblasts (greens), hepatocytes (dark maroon), mast cell and basophils (mast.baso- light maroon), monocytes and macrophages (mono.mac- dark blue), neutrophils (purples), natural killer cells (NK- yellow), dendritic cells (DC- orange), red blood cells (RBC- red), T-cells (light blue), and B-cells (brown). Cell types that contained multiple clusters are identified by “_number” after the name. B) Bar plot demonstrating the proportion of each cell cluster within the dataset, split by genotype (top) and infection status (bottom). C) Dot plot of key markers for each cell type in the combined dataset, E5 receptors (hFcRn and B2m) are included. Values are scaled with 1 (red) representing the highest, normalized, expression of a given gene and 0 (white) representing the lowest normalized expression. Dot size indicates the percentage of total cells in a cluster that express the gene of interest. D) Feature plots of normalized expression of E5 transcript (red) in infected Wt and Ifnlr1^-/- genotypes across the whole liver. E) Dot plots of scaled E5 transcript in uninfected and infected Wt and Ifnlr^-/- genotypes. E5 transcript maps strongly to the Mono.mac cluster in both genotypes as indicated by high mean expression (red) and the percent cells that have transcript E5 mapping to them (dot size). F) Feature plots of E5 transcript (red) in infected Ifnar1^-/- and Ifnar1^-/- Ifnlr1^-/- genotypes across the whole liver. G) Dot plots of scaled E5 transcript in uninfected and infected Ifnar1^-/- and Ifnar1^-/- Ifnlr1^-/- genotypes. E5 tropism in the liver expands in the absence of Ifnar1 expression as highlighted the percentage of cells in each cluster that contain mapped E5 transcript (dot size).

https://doi.org/10.1371/journal.ppat.1013891.g002

We identified 10 distinct cell types across the 13 transcriptional clusters, based on expression of established gene markers (Fig 2C and S1 Table) [22,25,28,35,36]. Among these clusters, we detected well-characterized liver-resident populations, including hepatocytes (high expression of Alb, Afp, and Ttr) and a heterogeneous monocyte/macrophage cluster (high expression of Cd68, Cx3cr1, and Adgre1). We also identified multiple populations of circulating professional immune cells, including dendritic cells (Siglech, Tcf4, and Irf8 high), a mixed mast cell/basophil population (Mcpt8, Gata2, and Cd200r3 high), a natural killer (NK) cell population (Nkg7, Cd7, and Klrd1), B-cells (Cd19, Rag1, and Cd79a), and T-cells (Ccr7, Tcf7, and Themis high). In addition, three transcriptionally distinct neutrophil clusters were identified (marked by Ly6g, Csf3r, and Chil3). Finally, three of the 13 clusters represented erythroid lineage cells, including red blood cells (Alas2, Gypa) and erythroblasts (Gata1, Epor, Tfrc). Given the diversity of identified cell types and the consistent representation of clusters across genotypes and conditions, we next focused on determining which of these cell types harbored echovirus transcripts.

While hFcRn has been detected in various liver cell types in both human tissue and transgenic mouse models, its expression at single-cell resolution remains poorly defined. Notably, IHC analysis in Tg32 mice demonstrates an hFcRn expression pattern similar to that observed in human liver, including localization to Kupffer cells and hepatocytes [37]. To identify candidate cell types likely to support echovirus entry, we analyzed the expression of the two genes encoding the echovirus receptor heterodimer: hFcRn and beta-2-microglobulin (B2m), across the dataset. Six clusters showed high expression of both genes including hepatocytes, monocytes/macrophages, dendritic cells, mast cells/basophils, and two neutrophil populations (Neutrophil_1 and Neutrophil_2) (Fig 2C). These clusters were thus predicted to be the most susceptible to echovirus infection. We next examined the presence of E5 viral RNA across the datasets to determine how infection varied by IFN signaling status. Notably, this dataset allows us to identify cell types that harbor echovirus RNA but does not distinguish whether individual cells are productively infected. However, when integrated with viral titer data (Fig 1B), the results indicate that the livers of these mice are productively infected overall. To localize viral RNA, we separated the dataset by genotype and analyzed only the infected samples. In Wt and Ifnlr1^-/- mice, viral reads were primarily detected in the monocyte/macrophage (Mono.mac) cluster, with lower-level expression also observed in hepatocytes (Figs 2D–2E and S2C). These findings are consistent with survival and viral titer data, indicating that intact type I IFN signaling restricts viral spread. In contrast, loss of type I IFN signaling markedly expanded the range of infected cell types. In both Ifnar1^-/- and Ifnar1^-/- Ifnlr1^-/- mice, echovirus RNA was detected broadly across nearly all clusters. Despite this widespread dissemination, the hepatocyte and Mono.mac clusters still exhibited the highest levels of viral RNA by mean gene expression (Figs 2F, 2G and S2E), highlighting their role as major targets of infection in the absence of type I IFN-mediated control. These data demonstrate that type I IFN signaling is essential for restricting echovirus infection to a narrow set of target cells in the liver, and that in its absence, both viral tropism and burden expand dramatically, particularly within hepatocytes and monocyte/macrophage populations.

Coordinated IFN signaling by myeloid cells shapes the hepatic inflammatory response to echovirus

To control damage during early acute virus infections, the liver canonically induces a broad IFN response to defend against invading pathogens [38]. However, the specific cell types in the liver that are responsible for producing an IFN response during echovirus infection are unknown. To address this, we next examined how distinct cellular clusters activated and upregulated inflammatory and antiviral pathways in response to infection. We first analyzed the induction of IFNs (type I, type II, or type III) between infected and uninfected samples across all four genotypes using normalized gene counts (S2 Table). To visualize these patterns, we generated heatmaps of log₂-transformed expression values across the eight conditions. We found that Ifna2, Ifnb1, and Ifnl2 were each induced to varying degrees in response to infection (Fig 3A–3C). Among these, Ifnb1 stood out for its robust induction across all four genotypes, particularly in key cell populations including hepatocytes, monocyte/macrophages, dendritic cells, and neutrophils (Fig 3B).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Myeloid cell–driven interferon responses mediate inflammation during acute echovirus infection.

A-C) Heatmap to visualize Ifna2, Ifnb1, and Ifnl2 transcript from each sample, grouped by infection status (x-axis), genotype (top) and cell type (right). Values represent the Log₂ (normalized counts) of the transcripts for each IFN with red indicating the highest expression of each gene in the dataset and dark blue reflecting no expression. D-G) Feature plots mapping the ISG score results to all cell types across infected Wt, Ifnlr1^-/-, Ifnar1^-/-, and Ifnar1^-/- Ifnlr1^-/- samples. Red dots represent the highest expression of the ISG score in a cell, purple reflects a moderate expression of the score, and grey reflects no expression. The ISG score was generated using the AddModuleScore package applied to a panel of 11 ISGs (Ifit3, Ifit3b, Siglec1, Apol9b, Ifi44, Ifit1, Oas1g, Apod, Phf11d, Sct, and Ifit1bl1) differentially expressed between both uninfected and infected Wt conditions and infected Wt vs infected Ifnar1^-/- conditions. H-K) Feature plots mapping the ISG score results to all cell types across uninfected Wt, Ifnlr1^-/-, Ifnar1^-/-, and Ifnar1^-/- Ifnlr1^-/- samples. Gene expression scales match the infected samples above. L-N) Dot plots of mean RNA expression of each of the individual ISGs that contribute to the ISG score across key IFN producing cell types (Mono.macs, DCs, hepatocytes, respectively). Dot plots are grouped by genotypes and infection status (y-axis) with mean expression and percent cells for each cluster as the legend on the right.

https://doi.org/10.1371/journal.ppat.1013891.g003

Next, to assess how antiviral IFNs (type I and type III) shape the inflammatory landscape of the liver across genotypes, we calculated an IFN-stimulated gene (ISG) score using Seurat’s package AddModuleScore. This approach computes the average expression of a predefined gene set and subtract background expression from control genes, enabling a direct comparison of ISG expression across cell populations and genotypes. To generate a list of candidate ISGs, we performed differential gene expression analysis using DeSeq2. We first identified genes that were significantly differentially expressed between Wt-infected and Wt-uninfected livers. To focus on the most robust changes, we applied a cutoff of log₂ fold change > 4 and p < 0.05. The same criteria were then used to identify significantly differentially expressed genes between Wt-infected and Ifnar1 ⁻ / ⁻ Ifnlr1 ⁻ / ⁻ -infected livers (S3 Table). By intersecting these two gene sets, we identified 11 shared ISGs that met our criteria: Ifit3, Ifit3b, Siglec1, Apol9b, Ifi44, Ifit1, Oas1g, Apod, Phf11d, Sct, and Ifit1bl1. This final gene set represents antiviral IFN dependent genes that are induced during echovirus infection and was used to calculate an ISG score, which we visualized using FeaturePlots across infected and uninfected conditions in all genotypes (S3 Table). Consistent with the survival and low titers observed in Wt and Ifnlr1^-/- pups, these genotypes mounted a robust inflammatory response to echovirus infection, as indicated by elevated ISG scores (Fig 3D, 3E, 3H, 3I). In contrast, the inflammatory response was markedly diminished in Ifnar1^-/- and Ifnar1^-/- Ifnlr1^-/- livers, with the notable exception of neutrophil clusters in the Ifnar1^-/- samples, which retained partial ISG induction (Fig 3F, 3G, 3J, 3K).

To assess how Ifnar ablation affected the inflammatory profile of clusters identified as type I IFN producers, we generated dot plots displaying expression of the 17 ISGs used to calculate the ISG score. In the Mono.mac and DC clusters, Ifnar ablation impaired the induction of 13 of the 17 ISGs in response to infection (Fig 3L, 3M). In contrast, Wt and Ifnlr1^-/- mice exhibited strong induction of all 17 ISGs. Hepatocytes are known to heavily rely on type III IFN signaling during viral infection [18,39], a pattern reflected in their ISG induction following echovirus infection. Consistent with this, Wt, Ifnlr1^-/-, and Ifnar1^-/- hepatocytes each upregulated antiviral ISGs in response to infection (Fig 3N). In contrast, hepatocytes from Ifnar1^-/- Ifnlr1^-/- mice failed to upregulate ISGs to the same extent, indicating that the antiviral response in this population is entirely dependent on IFN signaling. Together, these analyses identified four major clusters (Mono.mac, DC, hepatocytes, and neutrophils) that upregulated type I IFNs in response to echovirus infection across the four genotypes. They also demonstrate that type I IFNs are essential for driving a protective inflammatory response, particularly in Ifnar1^-/- models Given the cellular heterogeneity within these broad clusters, we next sought to resolve which specific subpopulations are responsible for IFN production across these four major clusters. Because the Mono.mac cluster harbored the highest number of viral reads in immunocompetent genotypes, we begin our subcluster analysis there, before moving on to the dendritic cell, hepatocyte, and neutrophil clusters.

Kupffer cells expand and serve as the primary source of type I IFNs during echovirus infection

To improve resolution and better define the cellular composition of key clusters across the eight experimental conditions, we performed sub-clustering of select populations. We began by focusing on the monocyte/macrophage (Mono.mac) cluster due to its robust type I IFN production and ISG responsiveness. To improve resolution, we extracted all Mono.mac cells across the eight experimental conditions and re-performed integration and clustering. To ensure the specificity of this subset, we screened for expression of lineage-defining markers (Adgre1, Cd68, Siglech, Cd200r3, Nkg7, Gata1) and removed three contaminating clusters identified as NK cells, a mast cell/basophil mix, and a DC population. We then re-ran clustering on the dataset, yielding 16,527 cells (Fig 4A). Based on canonical monocyte and macrophage markers [22,36], we identified six distinct cell types: proliferating monocytes (Top2a high), classic monocytes (Ccr2 and Ly6c2 high), transitioning monocyte.macrophages (Ccr2, Adgre1, Cd68 high), patrolling monocyte- macrophages (Spn, Itgam high), inflammatory macrophages (Card11 high), and liver resident Kupffer cells (Marco, Clec4f, Folr2, Timd4 high) (Fig 4B and S4 Table). Expression of hFcRn and B2m, which are necessary for echovirus infection, were highly expressed across all identified clusters, highlighting the potential for multiple liver cell types to serve as viral targets (Fig 4D). All six monocyte/macrophage subtypes were present across the eight experimental conditions. However, when we examined changes in cell type proportions, we found that the KC cluster uniquely expanded in response to infection, showing increased abundance across all four genotypes (Figs 4C, 4F, and S3A and S5 Table). This expansion was not observed in any other monocyte or macrophage subset and was specific to infected conditions. Together, these results implicate Kupffer cells as key responders to echovirus infection in the liver.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Kupffer cells harbor echovirus and are the main producer of type I Ifns during infection.

A) UMAP highlighting the Mono.mac cluster in the original whole-liver dataset. This cluster is made up of 16,527 cells across all 8 experimental conditions. B) UMAP of the reclustered Mono.mac population, new cell identities were assigned based on key marker expression. Cell types are annotated as follows: Kupffer cells (red), patrolling monocytes and macrophages (Patrol.mono.macs, yellow), classic monocyte (pink), inflammatory macrophages (Inflam.macs, purple), transitioning monocytes and macrophages (Trans.mono.macs, blue), and proliferating monocytes (Prolif.monocytes, green). C) Bar plot showing the proportion of each cell cluster within the subsetted dataset, split by genotype (top) and infection status (bottom). D) Key markers (genes on x-axis) for each cell type (y-axis) including E5 receptors are shown as a Dot plot. Scaled mean expression is shown as a color scale and the percent of cell expressing each gene is demonstrated by the size of the dot. Labels on top of the graph indicate gross groupings of key genes. E) Slingshot trajectories superimposed and labeled (Inflam or KC) on the reclustered Mono.mac UMAP (left panel). Feature plots of the pseudotime values (0-100) assigned to each cell along each trajectory (right panels). The color gradient (blue to red) and arrow indicate the directionality of cell differentiation along these pathways. F) UMAP of the reclustered Mono.mac population split by infection status to highlight the difference in abundance of the Kupffer cell cluster (red). G) Dot plot showing the average expression and proportion of cells expressing Ifnb1 across the different cell types (y-axis) separated by genotype (top) and infection status (x-axis). H) Merged Feature plots of infected Wt (left) and Ifnar1^-/- (right) samples demonstrating expression of a key Kupffer cell marker Marco (green) and E5 (red). Yellow indicates that a given cell expresses both Marco and E5. I) Merged Feature plots of infected Wt (left) and Ifnar1^-/- (right) samples showing expression of Ifnb1 (blue) and E5 (red). Purple indicates that a given cell co-expresses Ifnb1 and E5. J and K) Confocal microscopy images (scale bars, 150 µm and 100 µm) showing expression of the Kupffer cell marker Adgre1 and echovirus RNA in uninfected and infected Wt neonatal mouse livers, respectively. RNAscope was used to detect Adgre1 (green) and E5 viral RNA (red) in liver sections from uninfected or 40 hpi mice. Uninfected livers display a uniform distribution of Adgre1 ⁺ cells without detectable echovirus signal, whereas infected livers show Adgre1 ⁺ Kupffer cells throughout the tissue, with localized “hotspots” of cells co-expressing E5 RNA. Nuclei are stained with DAPI (gray). L) 60x confocal images of the slides described above. Echovirus (E5) positive cells in red, Adgre1 (macrophage) positive cells are in green, nuclei are in grey (DAPI) Scale bars (10μm).

https://doi.org/10.1371/journal.ppat.1013891.g004

To confirm that the expanding Kupffer cell cluster represented a distinct population from the infiltrating inflammatory macrophage cluster, we performed trajectory analysis on the Mono.mac dataset using Slingshot. This approach infers a pseudotemporal ordering of cells, allowing us to model how cell types mature or transition over time. Three trajectories were predicted using the Slingshot algorithm, all originating from the Prolif.monocytes cluster (Figs 4E and S3B). Trajectory analysis revealed three distinct developmental pathways: one terminating in the Kupffer cell cluster (top right panel), a second terminating in the inflammatory macrophage (Inflam.mac) cluster (bottom right panel), and a third looping back to the proliferating monocyte cluster (S3C Fig, right panel). The separation of Kupffer cells and inflammatory macrophages into distinct trajectories suggests that the expansion of the Kupffer cell cluster during infection is consistent with differentiation from circulating monocyte precursors, rather than reflecting a direct transition from inflammatory macrophages into the resident Kupffer cell niche, in line with previous descriptions of monocyte-derived Kupffer cell replenishment in the liver.

After identifying the Mono.mac cluster as a major source of type I IFNs at the whole-liver level, we next sought to pinpoint which specific cell types within this heterogeneous population were responsible for IFN induction during echovirus infection. To do this, we examined Ifnb1 expression as a representative type I IFN. When we mapped Ifnb1 transcripts across infected and uninfected samples from all genotypes, we found that Kupffer cells were the primary source of Ifnb1 expression in Wt, Ifnar1^-/-, and Ifnar1^-/-Ifnlr1^-/- mice, based on mean expression levels. (Figs 4G and S3C). Notably, Ifnar1^-/- and Ifnar1^-/-Ifnlr1^-/- genotypes showed broader Ifnb1 induction across multiple Mono.mac subtypes, suggesting that in the absence of downstream IFN signaling, cells may sustain or amplify Ifnb1 production in a compensatory attempt to control infection.

Next, we examined the distribution of echovirus transcripts across cell types and genotypes to determine whether IFN-producing cells were also sites of viral infection. Using Seurat’s blend visualization to co-map echovirus and cell-type–specific markers, we found that in Wt and Ifnlr1^-/- livers, viral transcripts localized predominantly to Kupffer cells, as indicated by co-expression with Marco (Figs 4H and S3D, S3E - left panel). In contrast, echovirus transcripts in Ifnar1^-/- and Ifnar1^-/-Ifnlr1^-/- mice were broadly distributed across multiple cell types, though they remained most concentrated in the Kupffer cell and inflammatory macrophage clusters (Figs 4H- right panel and S3D, S3E- right panel). To determine whether Ifnb1-producing Kupffer cells were themselves infected, rather than merely responding to signals from neighboring cells, we mapped co-expression of echovirus transcripts (red) and Ifnb1 (blue). In Wt and Ifnar1^-/- mice, approximately 40% of virus-positive cells also expressed Ifnb1 (137/336 and 81/217 cells, respectively; Fig 4I and S6 Table). In contrast, Ifnar1^-/- Ifnlr1^-/- mice exhibited a reduced frequency of Ifnb1 induction, with only 20% (197/1,020) of virus-harboring cells expressing Ifnb1 (S3F Fig).

Finally, to visualize echovirus localization relative to Kupffer cells in the liver, we performed RNAscope on liver sections from Wt pups. As expected, viral RNA was absent in uninfected Wt livers despite robust expression of the canonical Kupffer cell marker Adgre1 (Fig 4J). In contrast, infected Wt livers showed colocalization of echovirus RNA with Adgre1 in a small subset of liver-resident macrophages (Fig 4K and 4L), consistent with our scRNA-seq findings. Echovirus-positive cells that also expressed the macrophage marker Adgre1 were quantified using unbiased image analysis. Ten 40 × fields containing echovirus-positive cells were randomly captured, and post-acquisition analysis was used to determine co-expression of Adgre1. Using this approach, we found that ~96% of echovirus-positive cells (24/25) in the livers of immunocompetent mice were also Adgre1-positive. This observation is consistent with our single-cell data. Together, these results suggest that Kupffer cells act as early responders during echovirus infection, directly sensing viral presence and initiating a type I interferon response to alert neighboring cells.

Conventional dendritic cells and kupffer cells act in concert to induce type I IFN responses

In addition to the Mono.mac cluster, we identified a heterogeneous DC cluster that also induced a type I IFN response across all genotypes during echovirus infection. To investigate whether a distinct cell type within the DC-like population, similar to Kupffer cells, was responsible for viral sensing and IFN induction, we applied the same analytical approach described previously. First, we selected and subsetted the DC cluster from the whole liver dataset, which consisted of 4,622 cells across all eight conditions (Fig 5A). Next, we performed PCA and re-clustering on the DC population after confirming the absence of contaminating cell types based on lineage marker expression. Six DC clusters were identified based on expression of canonical gene markers; four plasmacytoid clusters (pDC-1 – 4) due to high Siglech, Ptprc, and Irf4 expression, two conventional dendritic cell 1-like clusters (cDC1–1 and 2) based on Ccr7, Batf3, Id2, and Zbtb46 expression, and a single monocyte derived DC cluster (moDC) as determined by Klf4, Cx3cr1, and Mrc1 expression (Fig 5B and S7 Table). All six clusters highly expressed hFcRn and B2m (Fig 5D).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. A conventional dendritic cell 1–like population produces Type I IFNs in response to infection.

A) UMAP highlighting the dendritic cell (DC) cluster to be subsetted from the whole-liver dataset. This cluster is made up of 4,622 cells across all 8 experimental conditions. B) UMAP of the reclustered DC population, new cell identities were assigned based on key marker expression. Cell types are annotated as follows: plasmacytoid dendritic cells (pDC-1, orange), pDC-2, yellow), pDC-3- brown, pDC-4, red) conventional dendritic cells type 1-like (cDC1-1, dark blue, cDC1-2, light blue), and monocyte derived dendritic cells (moDC- purple). C) Bar plot showing the proportion of each cell cluster within the subsetted dataset, split by genotype (top) and infection status (bottom). D) Key markers (x-axis) for each cell type (y-axis) including E5 receptors are shown as a Dot plot. Scaled mean expression is represented by a color scale and percent of cell expressing each gene is demonstrated by the size of the dot. Labels on top of the graph indicate gross groupings of key genes. E) UMAP of the reclustered dendritic cell population split by infection status to highlight the difference in abundance of the cDC-1 cluster (dark blue). F) Merged Feature plots of infected Wt (left) and Ifnar1^-/- (right) samples showing the expression of a key cDC1 marker Ccr7 (green) and E5 (red). Yellow dots indicate that a given cell expresses both Ccr7 and E5 (co-expression). G) Dot plot showing the average expression and proportion of cells expressing Ifnb1 across the different cell types (y-axis) separated by genotype (top) and infection status (x-axis). H) Merged Feature plots of infected Wt (left) and Ifnar1^-/- (right) samples showing the expression of Ifnb1 (blue) and E5 (red). Purple indicates that a given cell expresses both Ifnb1 and E5.

https://doi.org/10.1371/journal.ppat.1013891.g005

To assess whether any dendritic cell populations responded to echovirus infection similarly to Kupffer cells, we examined changes in cell-type abundance across all eight conditions using bar plots. This analysis identified a single cluster, cDC1–1, that consistently expanded in response to infection across all four genotypes (Figs 5C and S4A). We confirmed this expansion by comparing the distribution of cDC1–1 cells between infected and uninfected samples using UMAP visualization (Fig 5E and S8 Table).

To explore whether this expanding cDC1–1 population was directly infected, we examined the localization of echovirus transcripts in infected samples. Using blended feature plots of echovirus (red) and the cDC marker Ccr7 (green), we found clear co-localization in the cDC1–1 cluster across all genotypes (Figs 5F and S4E). However, viral transcripts were also detected in multiple other clusters, with broader distribution in Ifnar1^-/- and Ifnar1^-/-Ifnlr1^-/- mice, suggesting that viral abundance and cell-type targeting are shaped by IFN signaling status (S4B, S4C Fig). These findings differ from earlier observations in immunocompetent mice, where echovirus RNA was detected almost exclusively in KCs. In contrast, low levels of viral RNA were observed within DCs, particularly the cDC1–1 subset, suggesting that these cells may acquire viral material in the setting of impaired Kupffer cell control.

To determine whether type I IFN production was induced by a single cluster or more ubiquitously throughout the DC dataset, we analyzed the mean Ifnb1 expression across infected and uninfected genotypes. Like the Mono.mac dataset, Ifnb1 expression in Wt pups mapped to a single cluster, cDC1–1 (Figs 5G and S4D). In Ifnar1^-/- and Ifnar1^-/-Ifnlr1^-/- pups, Ifnb1 induction extended across a broader range of dendritic cell populations including cDC1–1, cDC1–2, pDC-4, and moDC, highlighting the failure of these genotypes to effectively respond to type I IFNs during infection. Finally, we examined whether cDC1–1 cells that harbored echovirus also expressed Ifnb1, as observed in Kupffer cells. Using FeaturePlots, we assessed co-localization of echovirus and Ifnb1 transcripts and quantified double-positive cells. In Wt samples, 40% of echovirus-positive cDC1–1 cells also expressed Ifnb1 (Figs 5H and S4F and S9 Table). Strikingly, in Ifnar1^-/- and Ifnar1^-/-Ifnlr1^-/- samples, 100% of virus-positive cDC1–1 cells also expressed Ifnb1, indicating sustained or amplified cytokine production in the absence of downstream signaling. These findings underscore the role of cDC1–1 cells in both sensing echovirus and initiating type I IFN responses.

Hepatocytes act as key responders to type I IFN signals during infection

Hepatocytes are highly susceptible to many viral infections [14,17,30,31,40–43], make up the bulk of liver mass (~80%), and serve as the liver’s major metabolic cell type [44]. To better understand the role of hepatocytes in innate immune sensing and response during echovirus infection, we determined whether their ability to respond to type I IFNs is essential for host survival. We subset the hepatocyte cluster from the whole liver dataset and then mapped viral RNA expression across hepatocyte clusters (Fig 6A, 6B). Using dot plots we found that in Wt and Ifnlr1^-/- mice, echovirus transcripts localized primarily to the Hepatocyte-3 cluster (~50% of cells) and to a lesser extent the Hepatocyte-1 cluster (~20%) (S5B Fig). In contrast, Ifnar1^-/- and Ifnar1^-/-Ifnlr1^-/- livers showed widespread infection, with 80–100% of cells in all four hepatocyte clusters harboring virus (S5C Fig). We next determined whether infected hepatocytes were producing IFNs. In Wt and Ifnlr1^-/- livers, expression of both type I and type III IFNs was nearly undetectable (Figs 6E and S5D, S5E). However, in Ifnar1^-/- and Ifnar1^-/-Ifnlr1^-/- mice, we observed robust induction of Ifnb1 and Ifnl2 in the Hepatocyte-2 cluster, despite this cluster not harboring high levels of viral RNA. These findings suggest that in the absence of IFN receptor signaling, infected hepatocytes fail to control viral replication and instead attempt to compensate by producing their own type I and III IFNs. To confirm that these clusters represented bona fide hepatocytes, we examined expression of canonical hepatocyte markers (Alb, Afp, Ttr) and proliferation markers (Top2a, Mki67, Rps12) (Fig 6D and S10 Table). We also examined cluster proportions across genotypes and infection conditions. Consistent with the non-migratory nature of hepatocytes, the proportions of each cluster remained stable across all eight experimental conditions and genotypes (Figs 6C and S5A).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Hepatocytes are major responders to type I interferon during infection.

A) UMAP highlighting the hepatocyte cluster to be subsetted from the whole-liver dataset. This cluster is made up of 6,401 cells across all 8 experimental conditions. B) UMAP of the reclustered hepatocyte population, new cell identities were assigned based on key marker expression. Cell types are annotated as follows: hepatocytes-1 (yellow), hepatocytes-2 (orange), hepatocytes-3 (brown), proliferating hepatocytes (prolif-hepatocytes, blue). C) Bar plot showing the proportion of each cell cluster contributes to the total cell composition of each experimental condition. Annotations include genotype (top), infection status (bottom), and total proportion (right). D) Dot plot of 6 key hepatocyte markers (x-axis) for each cluster (y-axis), including the two components of the E5 receptor complex. Scaled mean expression is represented by a color scale and the percent of cell expressing each gene is demonstrated by the size of the dot. Labels on top of the graph indicate gross groupings of key genes. E) Dot plot of the average expression and proportion of cells expressing Ifnb1 across the different cell types (y-axis) separated by genotype (top) and infection status (x-axis). F) Survival curve of six-week-old control and experimental mouse genotypes infected with 10⁴ echovirus5, intraperitoneally. Animals that survived past one week post infection (1wpi) were euthanized. Genotypes are as follows: negative control hFcRcn^Tg32Ifnar1^fl/fl (grey), positive control hFcRcn^Tg32 Ifnar1^-/- (blue), and experimental cKO hFcRcn^Tg32 Ifnar1^fl/fl Alb^Cre (red). G) Viral titers collected from key sites of infection (blood, liver, brain, pancreas) at 40hpi, the black dashed line indicated the limit of detection for this assay. Asterisks represent samples with statistically significant differences in viral titers across genotypes as determined by one-way ANOVA tests (* p < 0.0113,**** p < 0.0001, ns: not significant).

https://doi.org/10.1371/journal.ppat.1013891.g006

To test whether hepatocyte responsiveness to type I IFNs is required to control echovirus infection and promote survival, we generated a conditional knockout (cKO) mouse in which Ifnar1 was selectively deleted in hepatocytes (Ifnar1^fl/fl Alb^Cre) on the hFcRn^Tg32 background (S11 Table). Because Albumin-Cre activity is inefficient in neonatal mice [45], we conducted the following experiments in 6-week old adults. At this age, Cre-mediated recombination is robust enough to observe hepatocyte-specific recombination events. Hepatocyte-specific Ifnar1 knockout mice were challenged with a previously determined, sub-lethal, dose of echovirus (10⁴ PFU, IP) and compared to both negative and positive control genotypes: hFcRn^Tg32 Ifnar1^fl/fl (no Cre activity; IFN-responsive liver) and hFcRn^Tg32 Ifnar1^-/- (whole-body Ifnar knockout). Following infection, all Ifnar1^-/- mice exhibited mortality by 3 dpi, while all hepatocyte-specific Ifnar1 cKO mice succumbed by 6 dpi (Fig 6F). In contrast, 60% of Ifnar1^fl/fl control mice survived beyond 7 dpi, demonstrating that the absence of Ifnar1 specifically in hepatocytes is sufficient to cause mortality in an otherwise immunocompetent background. To confirm the specificity and consequences of the cKO, we quantified viral titers in key organs. In the liver, Ifnar1^fl/fl Alb^Cre mice showed significantly higher viral loads than control mice (Figs 6G and S5F), confirming both the efficiency of hepatocyte-specific Ifnar1 deletion and the critical role of hepatocyte IFN signaling in viral restriction. In the brain, viral titers were markedly lower in cKO mice compared to whole-body Ifnar1^-/- animals, indicating that the brain retained IFN responsiveness and could still mount a protective response. Pancreatic viral loads were comparable across all genotypes (S5F Fig), consistent with this organ being broadly permissive to echovirus infection regardless of IFN status. Blood viral titers (S5F Fig) confirmed systemic dissemination and excluded impaired access as the explanation for reduced brain infection. These data demonstrate that hepatocyte-specific loss of Ifnar1 is sufficient to recapitulate the mortality phenotype observed in whole-body knockouts following echovirus challenge. Our findings establish hepatocytes as critical recipients of protective type I IFN signals during infection.

Ethics statement

All animal procedures were performed in accordance with the recommendations in the National Institutes of Health Guide for the Care and Use of Laboratory Animals and were approved by the Duke University Institutional Animal Care and Use Committee (IACUC).

Virus growth and purification

Echovirus 5 (Noyce) was purchased from the ATCC. Echovirus stocks were amplified in HeLa cells cultured in DMEM supplemented with 10% FBS. Cells were infected at ~80% confluency and incubated for 48 hours until cytopathic effect was observed in ~90% of the monolayer. Infected cultures were harvested and subjected to freeze-thaw lysis, followed by detergent treatment (10% Triton X-100) and clarification by ultracentrifugation (7200 RPM, 20 min, 16°C, SW28 rotor). Virus-containing supernatants were further purified by ultracentrifugation over a 30% sucrose cushion (27,000 RPM, 3 hr, 16°C) and resuspended in PBS. Final viral stocks were aliquoted and stored at –80°C.

Virus quantification and sequencing

Viral titers were determined by standard plaque assay on HeLa cells. Briefly, 12-well plates were seeded with 1.8 × 10⁵ cells per well and incubated overnight. The following day, 10-fold serial dilutions of viral stocks were prepared and 100 μL was added per well. After 1 hour of adsorption at room temperature on a plate shaker, cells were overlaid with 1% agarose in 2 × MEM supplemented with 4% FBS, 2% penicillin/streptomycin, and 2% non-essential amino acids. Plates were incubated at 37°C for 48 hours, then fixed with 0.05% crystal violet in methanol following plug removal. Viral titers (PFU/mL) were calculated based on plaque counts and dilution factors. Viral genome identity was confirmed by unbiased deep sequencing performed by Alex Greninger (University of Washington). Viral RNA was extracted from purified stocks, converted to cDNA, and subjected to metagenomic next-generation sequencing (NGS) using the Illumina platform. Reads were quality-filtered, aligned to the reference echovirus genome, and analyzed for sequence identity and contamination. Consensus sequences were >99.9% identical to the input strain, confirming purity and genomic integrity of the viral stocks.

Genotype generation

To generate the hFcRnTg32 Ifnar1^-/- Ifnlr1^-/- line, we crossed previously established hFcRn^Tg32 Ifnar1^-/- and hFcRn^Tg32 Ifnlr1^-/- mice [31,55]. Offspring were genotyped using a custom Transnetyx panel (available upon request), and pups homozygous for both Ifnar1 and Ifnlr1 deletions and positive for the hFcRn^Tg32 transgene were selected and interbred until litters were uniformly homozygous. To generate hepatocyte-specific Ifnar1 knockout mice (hFcRn^Tg32 Ifnar1^fl/flAlb^Cre), Ifnar1^fl/fl mice (JAX #028256) and Albumin-Cre mice (JAX #003574) were each first crossed to hFcRn^Tg32 mice to generate hFcRn^Tg32 Ifnar1^fl/fl and hFcRn^Tg32 Alb^Cre lines. These lines were then intercrossed and backcrossed, three times, to obtain mice homozygous for hFcRn^Tg32, Ifnar1^fl/fl, and Alb^Cre. Genotyping was performed at each generation using Transnetyx panels (available upon request). All animal procedures were approved by the Institutional Animal Care and Use Committee (IACUC) at Duke University and conducted in accordance with NIH guidelines for the care and use of laboratory animals.

Mouse infections and virus quantification

Unless otherwise specified, the experimental unit was the individual mouse, and all analyses were performed at the level of the individual animal. Neonatal (P7) and adult (6-week-old) mice were infected intraperitoneally (i.p.) with 50 μL of echovirus diluted in PBS using 27G (neonates) or 25G (adults) needles. Based on prior work, inoculum doses were pre-specified and differed by genotype to capture early-acute survival windows: hFcRn^Tg32 Ifnar1^-/- and hFcRn^Tg32 Ifnar1^-/- Ifnlr1^-/- neonates received 50 PFU, hFcRn^Tg32 and hFcRn^Tg32 Ifnlr1^-/- neonates received 10⁶ PFU, and adult hFcRn^Tg32 Ifnar1^-/-, hFcRn^Tg32Ifnar1^fl/fl, and hFcRn^Tg32Ifnar1^fl/flAlb^Cre received 10⁴ PFU. Following infection, neonates were transferred to surrogate dams to prevent maternal transmission. These surrogate dams were C57BL/6 females lacking the hFcRn transgene, which renders them resistant to echovirus infection and thus limits the risk of viral shedding or re-infection of pups. The number of mice per genotype ranged from 8 to 12 individuals, depending on litter size, as two separate litters were used for each experimental condition (survival and titer). Litters with fewer than four pups were excluded a priori. No animals were excluded from survival or titer analyses after enrollment. For single-cell RNA-sequencing (scRNA-seq), cell-level quality control criteria are described below. Within each genotype, pups were assigned to experiments using block randomization by litter to balance potential litter effects across experimental days and cages. Analyses using the mouse as the experimental unit accounted for litter as a random effect. Primary outcomes for model characterization included (i) survival to 7 days post-infection (dpi)—all Wt and Ifnlr ⁻ / ⁻ mice survived to this time point and were euthanized—and (ii) viral titers in blood, liver, pancreas, and brain at 40 hours post-infection (hpi). All experimental cages were monitored twice daily for signs of distress, and no unexpected adverse events were observed. Blinding was not feasible given the experimental design; however, primary endpoints were objective and not influenced by blinding status.

For survival analyses, animals were monitored daily and euthanized upon reaching humane endpoints (e.g., ≥ 25% weight loss or predetermined signs of morbidity). At euthanasia, tissues were collected for viral quantification. Blood was harvested into K₂EDTA microtainer tubes (BD, Ref #365974), and livers were collected into bead-beating tubes (Omni International, Cat #19–628; neonates) or gentleMACS M tubes containing 6 mL DMEM + 10% FBS (Miltenyi Biotec, Cat #130-093-236; adults). Brain and pancreas were collected in 4 mL DMEM in gentleMACS tubes and homogenized using the gentleMACS Dissociator (Programs RNA_02.01 or RNA_01.01). Tail snips were collected for genotyping (Transnetyx). All samples were snap-frozen and stored at –80°C until processing.

Frozen tissues were thawed at room temperature. Neonatal livers were homogenized in 500 μL DMEM + 10% FBS using a Bead Ruptor Elite (Omni International; 4 m/s, 20 s). After centrifugation (5 min, 12,000 rpm), 400 μL of supernatant was collected. Adult tissue homogenates were similarly prepared, and 1 mL aliquots were reserved for viral titering.

For viral titration of infected tissues, serial 10-fold dilutions were prepared in DMEM + 10% FBS. For blood, 1.2 μL was added to 119 μL media for the first dilution; for tissues, 12 μL homogenate was added to 108 μL media. Dilutions (100 μL) were transferred to 96-well plates pre-seeded with HeLa 7B cells at ~60% confluency and incubated at 37°C for 5 days. Plates were developed by replacing media with 0.05% crystal violet in 10% methanol for 15 minutes, rinsing briefly in water, and air-drying overnight.

For scRNA-seq experiments, the experimental unit was the individual liver, encompassing all cells isolated from that organ. All mice were euthanized at 40 hours post-infection (hpi), and livers were collected and processed for single-cell sequencing. Within each genotype, litters were randomly divided into two groups with an equal distribution of males and females: one group was designated for infection, and the other served as the uninfected control. Primary scRNA-seq outcomes included cell type composition, viral RNA detection and mapping, interferon and ISG induction scores, and transcriptional trajectory analyses.

Luminex protein quantification

For protein quantification, whole-liver lysate samples previously collected for viral titer measurements were thawed and centrifuged at 10,000 × g for 5 min to pellet residual cellular debris. Supernatants (100 µL per sample) were assayed according to the manufacturer’s instructions using a custom IFN multiplex consisting of the Invitrogen ProcartaPlex Mouse IFN-α/IFN-β Panel 2-Plex (Cat. #EPX02A-22187-901) and the ProcartaPlex Mouse IL-28 Simplex (Cat. #EPX01A-26028-901). Protein concentrations (pg/mL) were used directly for subsequent analyses and figure generation.

scRNA sequencing

Livers were harvested from uninfected and echovirus 5–infected neonatal mice (n = 3 per condition and genotype) at 40 hours post-infection. Each liver was dissected into ice-cold DPBS supplemented with calcium and magnesium, minced using sterile scalpels, and mechanically dissociated by syringe aspiration. Tissue suspensions were filtered through 100 μm strainers and centrifuged at 50 × g for 5 minutes at 4°C to enrich hepatocytes. The resulting supernatant containing non-parenchymal cells (NPCs) was saved, and hepatocyte pellets were treated with red blood cell lysis buffer (Roche, Cat #11814389001) for 1 minute at room temperature. Lysis was quenched with ice-cold PBS + 1% FBS, and samples were passed through 70 μm filters. Hepatocyte and NPC fractions were combined, and viable cell counts were determined using trypan blue exclusion on an automated cell counter (Bio-Rad TC20). Up to 2 × 10⁷ cells per sample were processed using the Dead Cell Removal Kit (Miltenyi Biotec, Cat #130-090-101) according to the manufacturer’s instructions. Final cell suspensions were resuspended in PBS + 1% FBS and kept on ice prior to single-cell RNA-seq library preparation, which was performed at the Duke Sequencing and Genomic Technologies Shared Resource.

Single-cell RNA-seq library preparation and data analysis

Libraries were prepared from 10,000 total cells per sample using the 10x Genomics Single Cell 3’ Reagent Kit v2 (Manual Part #CG00052). Sequencing was performed on an Illumina NovaSeq 6000 (Illumina, San Diego), using an S2 flow cell for Wt and Ifnlr1^-/- samples, and an S4 flow cell for Ifnar1^-/- and Ifnar1^-/- Ifnlr1^-/- samples, achieving approximately 60,000 reads per cell. FASTQ files were processed using Cell Ranger v7.1.0 via the 10x Genomics Cloud Analysis pipeline, using a custom Mus musculus reference genome that included the Echovirus 5 genome and the human FcRn transgene (hFcRn). Filtered feature-barcode matrices were downloaded via command line and analyzed in R (v4.3.1) using a Seurat-based pipeline. The following packages were used: Seurat (v4.4.0), SeuratObject (v4.1.4), SeuratWrappers (v0.3.19), devtools (v2.4.5), dplyr (v1.1.3), harmony (v1.1.0), patchwork (v1.1.3), CellChat (v1.6.1), DESeq2 (v1.40.2), EnhancedVolcano (v1.18.0), and ComplexHeatmap (v2.16.0). Filtered matrices were merged within each genotype, and cells were retained if they contained 500–7,500 detected genes, fewer than 75,000 total counts, and less than 10% mitochondrial transcripts. X- and Y-linked genes were annotated using Ensembl BioMart (mmusculus_gene_ensembl), and cells expressing red blood cell markers were excluded. SCTransform normalization was applied to each dataset, regressing out mitochondrial content, ribosomal content, gene/unique molecular identifier (UMI) counts, and sex-linked gene expression. Datasets were integrated using the SCTransform workflow (SelectIntegrationFeatures, PrepSCTIntegration, FindIntegrationAnchors, and IntegrateData), followed by PCA and UMAP dimensional reduction using the top 50 principal components and clustering with FindNeighbors and FindClusters. ALRA-based imputation (RunALRA) was performed prior to marker identification. All datasets were also normalized and scaled (NormalizeData, ScaleData) before Harmony-based integration to align cells by cell type rather than sample origin. Harmony embeddings were used for final dimensional reduction (PCA and UMAP, 40 PCs), and clustering was performed at resolution 0.2. Marker genes were identified using FindAllMarkers on ALRA-imputed data (minimum log fold change = 0.25, expressed in ≥25% of cells). For focused analysis of specific clusters, subsets were reprocessed with SCTransform and Harmony to resolve intracluster heterogeneity. Differential expression analyses across infection status were performed using DESeq2, and results were visualized using ComplexHeatmap. Pseudotime trajectories of infected cells were inferred using Slingshot, specifying a starting cluster and allowing the endpoint to be inferred. Additional analysis code is available on the CoyneLab GitHub repository (https://github.com/CoyneLabDuke/Echovirus-infected-neonatal-mouse-livers). All raw files have been deposited on SRA under the following BioProjectID: PRJNA1303135.

Histology, hybridization chain reaction (HCR), and RNA scope imaging

Fresh liver lobes were collected and fixed overnight in 10% neutral buffered formalin (NBF), then rinsed and stored in 70% ethanol until processing. Fixed tissues were submitted to Histowiz for paraffin embedding, sectioning (horizontal orientation to maximize cross-sectional area), and hematoxylin and eosin (H&E) staining. Digitized H&E images were accessed via the Histowiz user portal.

For HCR, paraffin-embedded slides were deparaffinized through sequential xylene and ethanol washes, followed by heat-induced antigen retrieval in 1 × citrate buffer (pH 6.0) using a pressure cooker for 20 minutes. Slides were cooled to room temperature, rinsed in deionized water, and sample areas circled with a hydrophobic barrier pen. Sections were treated with 10 μg/mL proteinase K for 10 minutes at 37°C, then incubated with HCR hybridization buffer (Molecular Instruments) for 10 minutes at 37°C. Custom probes targeting Echovirus 5 RNA (previously described [64]) were applied in hybridization buffer and incubated overnight at 37°C in a humidified chamber. The following day, slides were washed in probe wash buffer and incubated in 5 × SSCT at room temperature for 3 hours. HCR amplification was initiated by incubating slides in amplification buffer for 30 minutes. Snap-cooled fluorescent HCR hairpins were prepared by heating to 95°C for 90 seconds, then allowed to equilibrate at room temperature in the dark for 30 minutes. Hairpins diluted in amplification buffer were applied to the slides and incubated overnight at room temperature. On Day 3, slides were washed in 5 × SSCT, counterstained with DAPI in PBS for 5 minutes, and mounted using Vectashield (Vector Laboratories, Cat #H-1200–01). Coverslips were sealed with nail polish, and slides were imaged using an Olympus FV3000 confocal laser scanning microscope and images were prepared using FIJI.

For RNAScope. Day 1, samples were treated as previously described [65]. Formalin-fixed paraffin-embedded (FFPE) liver sections were baked at 60°C for 1 hr, deparaffinized, and allowed to dry for 5 min at 60°C. Hydrogen peroxide was applied for 10 min at room temperature (RT), followed by a brief water rinse. Slides were then placed in target retrieval buffer and incubated at >99°C for 15 min (InstantPot method). After a brief water rinse, slides were immersed in 100% ethanol for 3 min at RT, dried for 5 min at 60°C, and hydrophobic barriers were drawn around the tissue sections. Protease Plus was added, and slides were incubated at 40°C for 30 min. During this step, probe solutions were prepared by diluting Channel 2 (C2) or Channel 3 (C3) probes 1:1 in the Channel 1 (C1) probe solution or probe diluent. Probes were prewarmed at 40°C for 10 min and cooled to RT for 20 min before use. Slides were washed and hybridized with the diluted probe solution for 2 hr at 40°C, washed in wash buffer, and stored overnight in 5 × SSC buffer. On day 2, slides were sequentially incubated at 40°C with AMP1, AMP2, and AMP3 reagents for 30 min, 30 min, and 15 min, respectively, with washes between each step. HRP reagent was then applied for 15 min at 40°C, followed by incubation with a diluted fluorophore (TSA Vivid 520 or 650, 1:1500 in TSA buffer) for 30 min at 40°C. Slides were washed, treated with HRP blocker for 15 min at 40°C, and this process was repeated for each detection channel with a distinct fluorophore. After hybridization and detection, slides were incubated with an autofluorescence quenching reagent (Vector Labs, Cat. #SP-8400) for 5 min at RT, counterstained with DAPI for 30 sec, and mounted in Vectashield antifade medium (Cat. #H-1200–01). Images were acquired using an Olympus FV3000 confocal laser scanning microscope and processed with FIJI software. A detailed version of the RNAscope protocol is available from ACDbio.

Statistics and reproducibility

All experiments were independently reproduced using tissue samples from multiple litters. Statistical analyses were performed using Prism (GraphPad Software), and data are presented as mean ± standard deviation (SD) unless otherwise noted. Statistical tests used for each experiment are detailed in the corresponding figure legends. One-way ANOVA was used to assess differences in viral titers across multiple genotypes. A p value of <0.05 was considered statistically significant, with exact p values provided in the figure legends.