VEEV TC-83 infection induces chronic weight loss, memory dysfunction, and reduced anxiety-like behavior in mice

To characterize a model of neurological consequences of viral infection, C57BL/6 mice were intranasally infected with 2 x 10⁷ plaque forming units (PFU) of VEEV TC-83. Mice labeled “Mock” are uninfected/control animals that have received intranasal PBS. A subset of mice was euthanized at 7 days post-infection (DPI; n = 10 VEEV; n = 10 mock/uninfected) to evaluate acute effects of infection and the remaining mice were followed to 106 DPI (n = 36–40 VEEV, n = 10 mock/uninfected) (Fig 1A). Mice were monitored and weighed at least daily for the first 4 weeks and then weekly until the end of the study. Mice that presented with neurological symptoms beyond 7 DPI are noted as VEEV + N and those without neurological symptoms are noted as VEEV. Mice were classified as VEEV + N if they had persisting (≥ two days) neurological symptoms (e.g., altered gait, imbalance, tremors, head tilt, head pressing, circling, seizure-like activity, partial paralysis, and recumbency). 100% of Mock infected animals survived (n = 10), 100% of VEEV infected (n = 18/18), and 81.81% of VEEV + N mice survived (n = 18/22) (Fig 1B). Neurological symptoms were observed primarily between 7 and 12 DPI; therefore, none of the 7 DPI animals were classified as having neurological symptoms. Mice infected with VEEV TC-83 lost significant weight and displayed clinical symptoms including changes in appearance, mobility, attitude, and body condition starting at day six post-infection (Fig 1D and 1E). Surviving mice slowly begin to regain weight and have little to no clinical symptoms by 15–16 DPI. Mice with neurological symptoms have increased clinical scores and weight loss as compared to non-neurological presenting VEEV-infected mice.

Download:

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

Fig 1. VEEV induces chronic weight loss, memory dysfunction, and reduced anxiety-like behavior in mice.

A) Study timeline indicating collection timepoints (red) and behavioral assessments (blue). B) Neurological symptoms observed. C) Survival curves comparing mock-infected (Mock), VEEV-infected (VEEV), and VEEV-infected + neurological symptoms (VEEV + N). D) Clinical scores across groups. E) Percentage of weight maintained over time, relative to baseline weight. F) Schematic of the Elevated Plus Maze (EPM) test. G) Percentage of time spent in the open (unwalled) arms of the EPM at 30, 60, and 90 days post-infection (DPI). H) Schematic of the Novel Object Recognition (NOR) test. I) Percentage of time spent exploring the novel object versus the familiar object. J) Total distance traveled within the NOR arena. K) Results from SHIRPA neurological assessments. Data represents combined findings from two independent experiments: Mock (N = 10), VEEV (N = 18), and VEEV + N (N = 18). All Statistics shown are One-Way ANOVA with mean ± SEM error bars. The schematics in panels F and H were created in BioRender. Kehn-hall, K. (2026) https://BioRender.com/pb0mn1n.

https://doi.org/10.1371/journal.ppat.1014115.g001

To assess the long-term impacts of VEEV infection, a series of behavioral and neuromuscular testing was performed at 30, 60, and 90 DPI. The elevated plus maze (EPM) is a standard assay of anxiety-like behavior in rodents [40,41] that can provide insight into the emotional changes observed following VEEV infection in humans [42]. Mice that spend more time in the enclosed (e.g., walled/closed) arms versus the open (e.g., unwalled/open) arms of the maze can indicate anxiety-like phenotypes (Fig 1F). In the EPM test, VEEV-infected mice showed reduced open arm dwelling time at 60 and 90 DPI compared to mock-infected controls (Fig 1G). A similar reduction in open-arm dwelling time was observed in VEEV + N mice at all timepoints (30, 60, and 90 DPI) further indicating alterations in behavior. The novel objection recognition (NOR) test assesses recognition memory, as memory deficits have previously been reported in humans infected with VEEV [42]. Prior to testing, naïve mice underwent preference testing to ensure no bias between the two objects (S1D Fig). For memory experiments, two identical objects (referred to as familiar objects) were placed in the arena, and mice were allowed to explore them for 5 minutes (Fig 1H). After a 60–90 minute retention interval, one familiar object and one novel object were placed in the arena, and the total time spent interacting with the objects were recorded. In this task, mice that spend greater time with the novel object is indicative of a potentially intact recognition memory. At 30 DPI, VEEV + N, but not VEEV mice, spent less time exploring the novel object compared to mock controls (Fig 1I). VEEV + N mice demonstrated increased distance traveled at 60 and 90 DPI compared to controls (Fig 1J), suggesting impairment of short-term memory and increased locomotion. To assess neuromuscular function, a modified and standardized version of the semiquantitative SHIRPA (SmithKline, Harwell, Imperial College, Royal London Hospital, Phenotype Assessment) testing, which has been previously used for to study VEEV infection in mice [43], was employed to evaluate the physical, behavioral, and neuromuscular phenotypes of mice following VEEV infection (Fig 1K). A score of 2 indicates normal responses, whereas values above or below 2 reflect hyperactivity or reduced performance, respectively. VEEV + N mice showed alterations in 1/6 of the parameters at 30 DPI, and 3/6 at 60 and 90 DPI. VEEV infected mice exhibited reduced responses in tests in 1/6 tests at 90DPI. Collectively, VEEV-infected mice exhibited impairments in visual-motor responses and limb strength consistent, but potentially less severe, than prior reports. Together, these results demonstrate that neuromuscular and behavioral performance are altered following VEEV infection, with mice that demonstrate neurological symptoms having decreased performance.

VEEV infection significantly impacts single cell gene expression within the hippocampus at 7 DPI, including multiple immune and antiviral genes

To assess the impact of viral infection on critical cell populations related to memory and anxiety, we performed single-cell RNA sequencing (scRNA-seq) of pooled hippocampus from n = 5 animals for all groups. The number of sequenced cells was 4603 for mock mice and 7228 for VEEV 7 DPI mice. Cell populations critical for brain function were annotated using gene expression markers from published literature [44,45] for excitatory neurons (EXC), inhibitory neurons (INH), microglia (MG), astrocytes (AST), pericytes (PER), oligodendrocytes (OD), oligodendrocyte progenitor cells (OPC), endothelial cells (END), macrophages (MAC), monocytes (MON), neutrophils (NEU), B cells, T cells, proliferating T cells (PRO_T) and natural killer (NK) cells (S2A and S2B Fig). Total recovered cells were impacted by viral infection as indicated by cell type abundance (Fig 2A). Calculations were performed to assess the expected assortment of recovered cells as a result of viral infection using a traditional cell expectation calculation overlaid on the fractions of cells recovered in each group where the expectation = number of VEEV cells/ (number of mock cells + number of VEEV cells). VEEV infection substantially altered both the total number of cells (Fig 2A) and clustering diversity identified through Uniform Manifold Approximation and Projection (UMAP) plot showing differential annotated cell clusters in the hippocampus. The most striking differences were in the immune and microglia cell populations, with the majority of immune cells and microglia identified in VEEV-infected animals (p < 1e-3, Fisher’s Exact Test). Utilizing a UMAP to take multidimensional transcriptomic data and create a 2D clustering map, which presents more similar datapoints clustered together, and less similar datapoints to be further apart, we can highlight broad transcriptomic changes associated with VEEV infection. Cells recovered from VEEV-infected mice (Fig 2B, blue), displayed highly differential expression in several cell types, most notably immune cells and microglia as compared to mock (Fig 2B, red). Lymphoid immune cell clusters for both 7 DPI and 106 DPI were further analyzed and identified to be predominantly T-cells with few NK cells, proliferating T cells, and B cells (S3A- S3D Fig). Myeloid immune cell clusters were further analyzed and identified to be microglia, macrophages, monocytes and neutrophils (S3E-S3H Fig).

Download:

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

Fig 2. Single-cell RNA sequencing (scRNA-seq) analysis reveals transcriptomic alterations in the hippocampus at 7 DPI.

A) Changes in cell abundance in response to viral infection. Statistical significance was determined by Fisher’s exact test,* = p-value≤0.05,*** = p-value ≤ 1e-10. B) Uniform Manifold Approximation and Projection (UMAP) plot showing differential annotated cell clusters in the hippocampus. C) UMAP identifying VEEV-infected cells detected via probe-based scRNA-seq. “VEEV NEG” represents viral probe negative cells sequenced from the VEEV 7 DPI dataset while “VEEV POS” represents cells which are probe positive from the same dataset. D) Top altered canonical pathways identified through Ingenuity Pathway Analysis (IPA). E) Venn diagram of differentially expressed genes of VEEV POS and VEEV NEG. F) Dot plot showing the expression of probes targeting VEEV viral RNA. Dot plot of genes from G) Neuroinflammation signaling pathway, H) Synaptogenesis signaling pathway, and I) Pathogen-induced cytokine storm pathway. EXC = excitatory neurons, INH = inhibitory neurons, MG = microglia, AST = astrocytes, PER=pericytes, OD = oligodendrocytes, OPC = oligodendrocyte progenitor cells, END = endothelial cells, MAC = macrophages, MON = monocytes, NEU = neutrophils, T = T cells, PRO_T = proliferating T cells, NK = natural killer cells.

https://doi.org/10.1371/journal.ppat.1014115.g002

To elucidate the role of active VEEV infection in cells, rather than bystander effects from neighboring cells, we introduced a probe-based sequencing approach with a series of 11 probes specific for identifying different portions of the viral genome. Cells that were found to express all 11 of these probes, were recorded as VEEV positive (POS) whereas cells which do not express all 11 probes were recorded as VEEV negative (NEG). No VEEV POS cells were recovered in the mock mice, while 17% of cells in the VEEV-infected group were positive for all 11 probes (n = 1194/7228) and 83% did not contain all probes (n = 6034/7228). VEEV POS cells were most abundant in excitatory neurons (28%, n = 352/1259), while other cell types had between 1% and 21% infection rates (END 21% (n = 64/309); PER 21% (n = 35/170); AST 20% (n = 173/885); MG 19% (205/1067); OPC 18% (n = 43/238); INH 18% (n = 84/469); NEU 18% (5/28); OD 13% (n = 110/830); MAC 11% (82/719); MON 7% (17/260); B cells 6% (1/16); PRO_Ts 4% (6/148); T cells 2% (16/641); and NK 1% (2/191)) (Fig 2C). In addition, excitatory neurons had the highest average expression of all the viral probes (Fig 2C and 2F), indicating robust viral replication within this cell type. As expected, probes targeting the region of the genome encoding the structural proteins (Probe 1–6) had the highest average expression compared to probes targeting the nonstructural region (Probes 7, 9–12) (Fig 2F). These results confirm that neurons, specifically excitatory neurons, are the primary, but not only target, for active VEEV infection.

To elucidate the impacts of active viral infection state on gene expression, we investigated transcriptomic alterations using Ingenuity Pathway Analysis (IPA), a bioinformatics software effective for determining gene signatures associated with canonical pathways and upstream regulators [46–48]. Canonical pathway analysis, shown in a Z-score scale where orange indicates predicted activation and blue indicates predicted inhibition, identified activation of immune signaling in nearly all cell types, regardless of infection status (Fig 2D). VEEV POS EXC had the least predicted activation of immune signaling, which correlated with the highest level of viral probes detected. Signal transduction pathways including RAF/MAP kinase cascade and RHO GTPase signaling were inhibited in EXC, whereas the INH and AST displayed activation of the majority of the altered signal transduction pathways. Likewise, multiple neurological signaling pathways including Serotonin Receptor Signaling, Synaptogenesis Signaling, and Huntington’s Disease Signaling, were inhibited in EXC. MG had a mixed transcriptomic profile having predominantly activated pathways in both VEEV POS and NEG cells, but a few neurological signaling pathways were only inhibited in VEEV NEG cells. Neuroinflammation Signaling was activated in all cells types regardless of infection status. While the majority of the altered canonical pathways were similarly regulated between VEEV POS and VEEV NEG cells, there were some notable differences in DEGs between these cell populations. Greater than 45% of upregulated differential expressed genes (DEGs) were shared between the VEEV POS and VEEV NEG cells in EXC, INH, AST, and MG. In contrast, less than 31% of downregulated DEGs were shared in MG, INH and EXC, indicating that these cell populations have cell-type-specific responses in cells containing viral RNA (Fig 2E).

Among the most significantly altered canonical pathways were Neuroinflammation Signaling (Fig 2G), Synaptogenesis Signaling (Fig 2H), and Pathogen-Induced Cytokine Storm Signaling (Fig 2I). Differentially expressed genes within each pathway were compared across VEEV POS, VEEV NEG, and Mock samples. Across all cell types, key gene signatures, particularly those linked to cytokine production and inflammatory responses, including Cxcl10, Ifnγ, Ccl2, Ccl5, and Stat1, were consistently upregulated in the Neuroinflammation Signaling and Pathogen-Induced Cytokine Storm pathways (Fig 2G and 2I). These results are consistent with other published bulk RNAseq and scRNA-seq studies in models of VEEV infection [32,49], showing robust innate immune and antiviral gene responses during acute infection. Stat1, Irf7, Ifnγ, and Tnf were identified as significant upstream regulators via IPA. Both Stat1 and Irf7 were significantly upregulated in AST, MG, INH, and EXC in both VEEV POS and NEG cells (S4 Fig). In contrast, Tnf and Ifnγ, had cell-specific expression, with MG VEEV POS and NEG cells expressing Tnf and INH and EXC VEEV POS and NEG cells expressing Ifnγ. In contrast, genes within the Synaptogenesis Signaling pathway were predominantly downregulated, a trend that was consistent across both VEEV POS and VEEV NEG groups (Fig 2H), suggesting a virus-associated disruption of synaptic remodeling independent of direct viral detection. These findings indicate that VEEV infection broadly disrupts immune and synaptic signaling, marked by strong upregulation of inflammatory cytokine pathways and consistent downregulation of synaptogenesis-related genes in both probe-positive and probe-negative cells, highlighting a substantial bystander effect driven by infected cells.

VEEV infection significantly impacts single cell gene expression within the hippocampus at 106 DPI, including the Glutaminergic Receptor Signaling Pathway

To determine the long-term impacts of VEEV infection on gene expression in the hippocampus, scRNA-seq was performed on n = 5 pooled animals for all groups, Mock, VEEV, and VEEV + N, at 106 DPI. The number of sequenced cells was 3941 for mock, 4483 for VEEV, and 5839 for VEEV + N samples. Cell populations were again clustered into the cell populations (Figs 3C and S3) as described for the 7 DPI samples. VEEV probes were also included for these samples, but no probe-positive cells were detected, indicating that mice had cleared the infection within the hippocampus by 106 DPI. At 106 DPI, the total recovered cells were altered, with the most dramatic increase in immune cells in VEEV and VEEV + N mice (Fig 3A and 3B). The top altered canonical pathways across the cell types include pathways involved in neurological signaling, immune response regulation, and signal transduction (Fig 3D). Interestingly, the altered pathways at 106 DPI were largely predicted to be inhibited in AST based on their z-score as compared to 7 DPI where the majority of the pathways were active (compare Figs 2D and 3D). In contrast, altered pathways in MG, EXC, and INH had largely positive z-scores in VEEV mice, but predominately negative z-scores in VEEV + N mice, indicating differential regulation of these pathways within the two groups of mice. To compare upregulated and downregulated DEGs across EXC, INH, MG, and AST cell types, we analyzed the overlap between VEEV + N vs Mock (blue) and VEEV vs Mock (grey), with shared genes represented in the center (Fig 3E). At 106 DPI, fewer than 20% of downregulated DEGs were shared between the VEEV + N and VEEV groups. In contrast, a greater proportion of upregulated DEGs (ranging from 18% to 35%) were shared between the two groups, with the highest overlap observed in AST (35.1%) and the lowest in MG (18.1%). These results highlight differentially regulated genes present in VEEV-infected mice with neurological symptoms, particularly driven by alterations in excitatory neuron genes, and to a lesser extent, inhibitory neuron genes, pointing to a potential role for excitatory circuits in the manifestation of disease-related transcriptional changes.

Download:

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

Fig 3. scRNA-seq analysis reveals transcriptomic alterations in the hippocampus at 106 DPI.

A-B) Changes in cell abundance, between A) Mock vs VEEV and B) Mock vs VEEV + N. Statistical significance was determined by Fisher’s exact test,* = p-value≤0.05,*** = p-value ≤ 1e-10. C) UMAP plot showing differential annotated cell clusters in the hippocampus. D) Top altered canonical pathways identified through IPA. E) Venn diagram of differentially expressed genes of VEEV + N and VEEV. Dot plot of genes from F) Glutaminergic receptor signaling pathway and G) Synaptogenesis signaling pathway. EXC = excitatory neurons, INH = inhibitory neurons, MG = microglia, AST = astrocytes, PER=pericytes, OD = oligodendrocytes, OPC = oligodendrocyte progenitor cells, END = endothelial cells, MAC = macrophages, MON = monocytes, NEU = neutrophils, T = T cells, PRO_T = proliferating T cells, NK = natural killer cells.

https://doi.org/10.1371/journal.ppat.1014115.g003

To further characterize these genes, we identified substantial changes in the Glutaminergic Receptor Signaling Pathway (Fig 3F) and Synaptogenesis Signaling Pathway (Fig 3G), two highly related pathways essential for hippocampal function and neuronal activity because they underlie critical processes such as synaptic plasticity, learning, memory, and communication [50–52]. A large number of genes within these pathways were downregulated in all cell types including Grin2b (Fig 3G), a gene encoding a subunit of the NMDA glutamate receptor GluN2b; Slc1a2, the excitatory amino acid transporter 2 (Eaat2) protein-encoding gene, which is critical for removal of glutamate from neuronal synapses (Fig 3F) [53]; Camk2a, Calcium/Calmodulin Dependent Protein Kinase II Alpha, which regulates calcium signaling at the synapse [54] (Fig 3G); and Mapk1, mitogen activated protein kinase 3 (also known as Erk1; extracellular signal-regulated kinase) (Fig 3F). These changes correspond with VEEV + N mice having predicted inhibition of related downstream signaling pathways including cAMP and CREB signaling cascades [55]. In contrast, VEEV INH and EXC had predicted activation of these pathways, again highlighting transcriptional differences in these mice at 106 DPI. These data indicate that Glutaminergic Receptor Signaling and related downstream signaling pathways are suppressed in VEEV + N mice, correlating to behavioral alterations observed at 106 DPI.

VEEV activates astrocytes and microglia in the hippocampus at 7 and 106 DPI

To assess changes in microglia and astrocytes during acute VEEV infection, we conducted immunohistochemical analysis of key cell populations in the hippocampus, which are critical for the structure and function of these brain regions. To evaluate the localization of VEEV in the hippocampus, the envelope glycoprotein (E2), a marker of VEEV infection, was utilized to observe active infection. VEEV, shown in green, was found to readily infect the hippocampus at 7 DPI (Fig 4B), in agreement with other studies [32,56]. In mock infected mice, astrocytes (Glial fibrillary acidic protein (GFAP); red), and microglia/macrophages (Ionized calcium binding adaptor molecule 1 (Iba1); white) were found at a regular distribution with highly branched and ramified morphological structures, indicating a resting state overlapping with cellular nuclei (DAPI; blue) controls throughout the hippocampus (Fig 4A and 4C). A marker of astrogliosis, GFAP, shown in red, is increased in VEEV-infected mice at 7 DPI (Fig 4B). Morphological changes including the formation of ameboid-like structures indicating activated microglia, were observed in VEEV-infected mice, consistent with activation of the innate immune system and VEEV-induced neuroinflammation (Iba1; white) (Fig 4B). In the hippocampus, the hilar region and the dentate gyrus (DG) were examined in greater detail because of their critical functions. The hilar region plays a key role in regulating neuronal signaling pathways and maintaining network dynamics within the hippocampus, while the DG is essential for episodic memory formation, acting as a gateway for processing and encoding new information [57,58]. VEEV infection was localized in the hilus and the arc of the DG (Fig 4D). Formation of VEEV infection along the DG suggests this may be the site of entry of VEEV into the hilar region of the hippocampus. In mock infected mice, microglia and astrocytes were uniform across the DG and the hilus of the hippocampus (Fig 4C). Microscopic examination reveals microglia activation and increased colocalization around VEEV-infected cells, which may indicate the presence of either infected microglia or microglia engulfing infected cells (Fig 4E). In the hilus, VEEV infection was primarily found in the microglia, and substantial astrocyte activation was visible surrounding infected microglia (Fig 4E). In the DG, VEEV infection did not appear to overlap with microglia and astrocytes (Fig 4F).

Download:

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

Fig 4. VEEV activates astrocytes and microglia in the hippocampus at 7 and 106 DPI.

A and B) Wide-view images of the hippocampus (hippo) regions of the brain in mock (A) and VEEV infected (B) mice at 7 DPI. Sections were stained for VEEV infected cells (E2 glycoprotein, green), astrocytes (GFAP, red), and microglia/macrophages (Iba1, white), and a cell nuclei control (DAPI, blue). White boxes indicate regions further analyzed in subsequent panels. C and D) High magnification images of the hippocampus in C) mock-infected and D) VEEV infected animals. VEEV infected cells (E2 glycoprotein; green); Astrocytes (GFAP; red); Microglia/macrophages (Iba1; white). E and F) High magnification images of the hilus (E) and dentate gyrus (DG) (F) of the hippocampus demonstrating Iba1 + microglia/macrophages in an activated, amoeboid morphology in VEEV infected tissue and demonstrating increased activation of astrocytes in VEEV+ tissue through hypertrophied morphology. G and H) Representative wide-view confocal images of hippocampus (hippo) at 106 DPI with VEEV Envelope protein (VEEV E2; green), astrocyte (GFAP; red), microglia/macrophages (IBA1; white) expression in G) Mock and H) VEEV + N infected animals at 106 DPI. White boxes indicate areas investigated further in later panels for the hippocampus (I and J). Astrocyte and microglia/macrophages in Mock infected animals have healthy, branched Iba1 + cells and fewer astrocytes in the hippocampus (I) whereas prolonged activation of astrocytes and Iba1 + cells, indicated by a circled regions contain activated ameboid microglia/macrophage structures, and more astrocyte coverage in the hippocampus are seen in VEEV + N 106 DPI samples (J). IMARIS quantification of K) GFAP average cell area, L) Iba1 average cell area, M) Iba1 average cell volume, and N) Iba1 average cell sphericity of VEEV + N 106 DPI vs Mock infected animals. n = 4 for Mock mice and n = 5 for VEEV + N mice. Statistical significance was determined by Student’s t- test,* = p-value<0.05.

https://doi.org/10.1371/journal.ppat.1014115.g004

To evaluate more chronic alterations in glia structure and activation states as a result of VEEV infection, we performed immunohistochemistry analysis of the hippocampus from 106 DPI groups. For these analyses, only mice that presented with persisting neurological symptoms (VEEV + N) were selected to be compared to mock mice. VEEV E2 was not detected in VEEV-infected mice at 106 DPI (Fig 4H), which is consistent with the lack of probe positive cells detected in the scRNA-seq studies. Consistent with the 7 DPI mock mice, microglia/macrophages in 106 DPI mock mice were relatively sparse with a healthy distribution and resting phenotype in both the wide view images (Fig 4G) and the zoomed-in representative images of the hippocampus (Fig 4I) at 106 DPI. In VEEV + N 106 DPI samples, an increase in the proportion of astrocytes and increased ameboid-activated microglia/macrophages were observed in the wide-view images (Fig 4H) and the zoomed-in portions of the hippocampus (Fig 4J). Morphology analysis was performed utilizing Interactive Microscopy Image Analysis Software (IMARIS) to assess the morphology of Iba1 and GFAP-expressing cells (Fig 4K–4N). In VEEV + N 106 DPI mice there was a significant increase in the average cell area of GFAP+ cells in the hippocampus compared to mock 106 DPI mice (Figs 4K and S5), which is consistent with hypertrophic activated astrocytes. Likewise, VEEV + N mice had a significant increase in both the average cell area and average cell volume of Iba1 + cells (Figs 4L, 4M and S5), which is consistent with an activated ameboid microglia shape. In further support of this, Iba1 + cells in VEEV + N mice had an increased average cell sphericity (Figs 4N and S5). Collectively, VEEV-infected mice showed acute and prolonged activation of microglia and astrocytes, consistent with previous reports in VEEV mouse models [3,32] and a persisting inflammatory response observed in multiple neurodegenerative diseases [59,60].

VEEV infection results in hippocampal hilar interneuron loss

To characterize neuronal changes in the hilus of the hippocampus, we quantified the total number of neurons using NeuN, a pan-neuronal marker, as well as Reelin (RELN) as a marker of interneurons (Fig 5A). The hilus of the hippocampus is predominantly composed of inhibitory interneurons [61,62]; therefore, an overall reduction in NeuN-positive cells (all neurons) is primarily indicative of interneuron loss in this specific region. VEEV-infected mice displayed a loss of interneurons in the hilus of the hippocampus at 7 DPI (Fig 5B and 5E), as indicated by reduced NeuN. This phenotype persisted through 106 DPI (Fig 5D and 5G). Reduced expression of Reelin-expressing interneurons was also observed at 7 DPI (Fig 5B) and 106 DPI (Fig 5D). Non-biased stereological quantification confirmed the loss of reelin-expressing neurons at 7 (Fig 5F) and 106 (Fig 5H) DPI. Transcriptomically, we observed an increase in apoptotic, pyroptotic, and immunogenic signaling at 7 DPI which may contribute to inhibitory neuron loss; however, at 106 DPI these cell death pathways are minorly observed in inhibitory neurons (Fig 5I) and downregulated in excitatory neurons (Fig 5J), which may be altered by the loss of inhibitory neurons, or excitatory neurons attempting to regulate signaling through inhibitory action (Fig 5I). Dot plot analysis of genes associated with Reelin signaling in neurons were analyzed across the 4 predominant cell types of interest, where a consistent downregulation was observed in nearly all cell types and conditions compared to mock. Reelin-pathway genes, including Reln, Dab1, Camk2b demonstrated negative fold changes supporting suppression of these genes. Additional neuronal signaling genes, including Grin2a and Mapk3, followed similar patterns, supporting a decrease in synaptic signaling as a whole including reelin-signaling pathways. By contrast, Apoe was both enriched in microglia and upregulated, supporting a disease-associated transcriptional response often associated with synaptic remodeling and injury response. Taken together, these results suggest that VEEV induced restructuring of the hippocampus as evidenced by loss of total hilar neurons (NeuN) and reelin-positive inhibitory neurons. Furthermore, these results are unexpected, as Reelin is traditionally associated with inhibitory interneurons, suggesting a shift in pathway activity toward excitatory neurons during infection.

Download:

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

Fig 5. VEEV causes loss of neurons (NeuN) and inhibitory RELN-expressing neurons.

A) Stereological quantification of NeuN (neuron nuclear marker; pink) and RELN (inhibitory neuron marker; green) in the hilus of the hippocampus at 7 DPI in A) Mock and B) VEEV and 106 DPI in C) Mock and D) VEEV. Quantified graph showing a significant decrease in E) NeuN+ cells and F) decrease in co-expressing of NeuN + RELN expressing cells in Mock vs. VEEV at 7 DPI. A loss of G) NeuN+ and H) NeuN + RELN expressing neurons persists to 106 DPI. I) Cell Death Pathways Identified via IPA of Neurons (INH and EXC). K) Dot plot of genes from Reelin signaling in neurons. All statistics performed with One-way ANOVA and Bonferroni post hoc correction; n = 5/group, ns = not significant, *p < 0.05; **p < 0.01, ***p < 0.001, ****p < 0.0001. Scale = 20µm. EXC = excitatory neurons, INH = inhibitory neurons, MG = microglia, AST = astrocytes.

https://doi.org/10.1371/journal.ppat.1014115.g005

VEEV elevates c-FOS–mediated neuronal stress without affecting neurogenesis in the hippocampus

To assess whether VEEV disrupts neurogenesis and contributes to neuronal loss, we investigated doublecortin (DCX)-positive cells in the hippocampus across five groups: Mock 7 DPI (Fig 6A), VEEV 7 DPI (Fig 6B), Mock 106 DPI (Fig 6C), VEEV 106 DPI (S6C Fig), and VEEV + N at 106 DPI (Fig 6D), and found no significant changes in both the hilus (Fig 6E–5F) and DG (Fig 6G–6H), indicating that neurogenesis remains unaltered after viral infection. To assess whether loss of neurons in the hippocampus correlated with neuronal activity, we quantified c-Fos expression, an immediate early gene induced in response to neuronal activity, stress, and learning tasks [63–65]. An increase in c-Fos expression was observed at both 7 and 106 DPI in VEEV infected mice (Fig 6I–6L). Quantification showed that cFos was significantly elevated in the hilus and DG at 7 DPI (Fig 6M and 6O), but only within the hilus at 106 DPI (Fig 6N and 6P). We observed colocalization of c-Fos with neurons (NeuN) and astrocytes (GFAP) (S6G and S6H Fig). At 7 DPI, c-Fos colocalized with neurons (white arrows) and some c-FOS did not localize with neurons (yellow arrows), but may overlap with other cells such as activated astrocytes. At 106 DPI, c-Fos only colocalized with NeuN-expressing neurons (S6G and S6H Fig). Utilizing scRNA-seq data from the hippocampus, we observed an increase in c-Fos expression in VEEV POS and VEEV NEG astrocytes, microglia, inhibitory neurons, and excitatory neurons at 7 DPI (Fig 6Q). At 106 DPI, c-Fos expression was primarily observed in inhibitory and excitatory neurons, with mock and VEEV + N mice having similar expression levels (Fig 6R). The differences in c-Fos expression between the IHC and scRNA-seq data observed at 106 DPI could be reflective of differences in protein versus RNA levels in addition to the potential attempt of excitatory neurons to re-regulate c-Fos signaling post-infection. Collectively, these VEEV-induced alterations in c-Fos, most evident in the hilus and DG of the hippocampus, indicate a prolonged stressed state within the hippocampus and a potential reduction in synaptic function.

Download:

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

Fig 6. VEEV induces persistent neuronal stress (c-FOS) associated with memory loss in the hippocampus.

A-D) High magnification images of DCX (green) in the hilus and DG of the hippocampus. A) Mock 7DPI compared to B) VEEV 7 DPI, quantified in the E) hilus and G) the DG of the hippocampus. C) Mock 106 DPI compared to D) VEEV + N 106 DPI, quantified in the F) hilus and H) the DG of the hippocampus. I-L) High magnification images of neuronal activity marker, c-Fos (green) expression in the hilus and DG of the hippocampus with DAPI (blue) as a cell nuclei control. c-Fos is lowly abundant in I) Mock 7DPI compared to J) VEEV 7 DPI, quantified in the M) hilus and O) the DG of the hippocampus. At 106 DPI K) Mock c-Fos expression is low compared to L) VEEV infected quantified in the N) hilus and P) DG. Analysis of Fos expression via scRNA-seq at Q) 7 DPI and R) 106 DPI. All statistics performed with One-way ANOVA and Bonferroni post hoc correction; n = 5/group, ns = not significant, *p < 0.05; **p < 0.01, ***p < 0.001, ****p < 0.0001. Scale = 20µm. EXC = excitatory neurons, INH = inhibitory neurons, MG = microglia, AST = astrocytes.

https://doi.org/10.1371/journal.ppat.1014115.g006

Ethics statement

All studies were reviewed and approved by both the Virginia Tech Institutional Animal Care and Use Committee (IACUC; #23–041) and Defense Threat Reduction Agency (DTRA) Animal Research Oversight Animal Care and Use Review Office (ACURO; CB11147.e001) before animal studies were started.

Infection

Male 10–12-week-old wild-type C57BL/6 mice were ordered from The Jackson Laboratory and housed in an ABSL-2 laboratory at Virginia Tech. Mice were housed in groups of n = 5/cage and provided with food and water ad libitum. Standard 12 h light and 12 h dark cycles were used. Mice were anesthetized with vaporized isoflurane and then infected via intranasal inoculation with 2E7 PFU of VEEV TC-83 per mouse diluted in sterile phosphate buffered saline (PBS). Control mice were inoculated with PBS. Mice were monitored 1-2x daily for the first 28 days after infection in which appearance, mobility, attitude, and body condition on a scale of 0–3 or 0–4 per criteria (Table 1). Any mouse scoring the highest score in appearance, mobility and attitude were immediately euthanized. Attitude was scored as an observational index of the animals baseline responsiveness to being handled and lethargic, ranging from fully alert showing normal movement upon opening the cage and interacting with the animal (0), slowed or reduced exploration during handling, but aware [1], reduced responsiveness and alertness, minimal response to handling [2], and no observable reaction to handling, including static position and lack of movement [3]. Weight loss per se was not considered a feature requiring euthanasia unless weight loss exceeded 30% of the animal starting weight or was accompanied with additional severe illness or visual emaciation. Euthanized animals were counted as succumbing to viral infection when calculating mortality. Total score will determine whether additional monitoring or intervention will be given. A score of 0–5 will result in no change of monitoring (normal 1x daily), 6–10 will increase monitoring to 2x daily, and at ≥11 the animal is considered moribund and euthanized immediately. Additionally, any animal achieving the highest score in any category was immediately euthanized. Mice were also scored on a binary scale (0 = absent, 1 = present) for neurological symptoms including circling, head tilt, head pressing, altered gait, imbalance, and seizure activity as well as moribund symptoms including labored breathing, tremors, eye crust, lateral recumbency, and paralysis.

Download:

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

Table 1. Animal health and wellness score guide.

https://doi.org/10.1371/journal.ppat.1014115.t001

Behavioral experiments

The elevated plus maze and novel object recognition tests were performed in a dark environment inside of secondary containment (Uline Rubbermaid cube truck; #H-8195) and cube truck lid; #H-8197). Tests were performed at approximately 30, 60, and 90 DPI in a consistent order (elevated plus maze, SHIRPA, novel object recognition) with one test being performed each day and at least one day off between tests. All mazes, tools, and surfaces were cleaned with 70% ethanol between each mouse and allowed to dry for at least 20 seconds between trials. A transport cage was cleaned between each cage of mice.

Modified SHIRPA.

SHIRPA testing of VEEV-infected mice has been previously described by other groups [3,66,108]. We performed a modified SHIRPA testing which included tests for body tone, reach touch, trunk curl, forelimb place, palpebral reflex, and grip test. Each test is briefly described below and score sheet for each test is displayed in (Table 2). Body tone test was performed by holding the mouse by the base of the tail on clean lab mat and gently depressing on the mouse’s spine using two fingers. Reach touch is performed by holding the mouse from the base of the tail approximately 6 inches high from the base of the biosafety cabinet and lowering the mouse towards an empty cage until they either touch the base of the cage or extend forelimbs and reach towards the cage. Trunk curl is performed by holding the mouse from the base of the tail at least 6 inches above the base of the biosafety cabinet and assessing the mouse ability to perform a trunk curl. For this test, mice were provided three attempts to exceed 90-degree trunk curl (if necessary) and provided 30 seconds between each test to rest. Palpebral reflex was performed by holding the base of a mouse tail and approaching it from 3 directions, in front of the animal (between the eyes), and from the left, and right sides of the animal towards the sides of the skull. Animals were approached until a response could be recorded (blinking, movement, etc.) or until the q-tip is approximately 2 mm from the animal to ensure they are not contacted by the q-tip. The grip test was performed by allowing a mouse to grab a metal grate and then inverting them at 10 inches above the base of the biosafety cabinet and recording the grip duration up to 60 seconds. For this test, mice were provided three attempts to exceed 60 seconds (if necessary) and provided 30 seconds between each test to rest.

Download:

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

Table 2. Modified SmithKline, Harwell, Imperial College, Royal London Hospital, Phenotype Assessment (SHIRPA) score guide for neuromuscular function assessment.

https://doi.org/10.1371/journal.ppat.1014115.t002

IHC Perfusion fixation and organ dissection

Mice were euthanized via CO₂ and blood was collected via cardiac puncture. Dissection was performed to expose the heart, and then 20 mL heparin diluted in ice-cold sterile PBS (2000 units/kg) was injected via the apex of the heart into the left ventricle. A small incision in the right atrium was made, and 50 mL of ice-cold 4% paraformaldehyde and 3% sucrose in PBS solution (pH = 7.4) was slowly injected into the same puncture site at a rate of 5 mL/minute to flush blood from the circulatory tract. Brains were dissected from the mice and placed in the 4% paraformaldehyde and 3% sucrose in PBS solution for 4 hours on ice. Brains were then washed with PBS for 5 minutes (~20mL) and then preserved in a sucrose gradient until brains sunk to the bottom of a 50 mL conical tube (20% for ~48h; 25% ~ 48h, all at 4°C) and then molded in OCT solution (30% OCT + 1:1 ratio of 60% sucrose and PBS) and stored at -80°C before cryosectioning.

Brain molding, sectioning, and slide preparation

Following sucrose gradients, brains were molded and frozen in a solution containing 30% OCT and 60% sucrose dissolved in PBS and slow frozen for 30 minutes on dry ice before storage at -80°C for sectioning. The molds were filled at least 2mm beyond the brain in the molds with OCT (Tissue-Plus O.C.T. Compound, # 23-730-571, Fisher Scientific Waltham, MA, USA) and slow frozen for 30 minutes on dry ice. Brains were then placed into the –80°C for storage until cryosectioning. The brains were trimmed through the coronal plane of the brain through the center of the cerebellum which was then mounted on a microtome. Cryosectioning was performed via a cryostat (CryoStar NX50, #957130, Thermo Scientific, Waltham, MA, USA) from -1.1 to -2.6 mm posterior from bregma at -20° and 30μM coronal sections were obtained with 5 sections per slide, spaced 450μM apart. Sections were dried to adhere to the slide at 40°C for at least 1 hour and subsequently stored at -80°C for staining.

Immunohistochemistry analysis

Immunohistochemistry analysis was performed to assess Microglia (Iba1), Neurons (NeuN), inhibitory neurons (RELN), Astrocytes (GFAP), Neural activity (c-FOS), and viral presence (VEEV E2) in the hippocampus using the antibodies and dilutions shown in Table 3. Wide-view images of the hippocampus were taken on a Zeiss LSM 880 (ZEISS, Oberkochen, Germany) with a Zeiss FLUAR 5x/0.25 lens (#440125, ZEISS, Oberkochen, Germany) and Zeiss 63X LD C-Apochromat 63x/1.15 water emersion lens (#421887, ZEISS, Oberkochen, Germany).

Download:

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

Table 3. Antibodies used for IHC.

https://doi.org/10.1371/journal.ppat.1014115.t003

Quantitative analyses and representative images of the dentate gyrus and hilar regions of the hippocampus were collected by a blinded investigator. Five coronal sections were observed using the Optical Fractionator probe in Stereo Investigator software (MBF Bioscience, version 2017.03, Williston, VT, USA) with an Olympus BX51TRF motorized microscope (Olympus America, Center Valley, PA, USA), following previously established protocols [109–113]. Regions of interest, including the DG and hilus, were outlined, and grid dimensions for the optical fractionator were set to 150 × 150 µm with a 75 × 75 µm counting frame. The estimated cell count was normalized to the contoured volume using planimetry, providing the number of cells per mm³.

IMARIS analysis

Three-dimensional image analysis was performed using the z-stack confocal images and IMARIS software version 9.8.2 (Oxford Instruments, Concord, MA, USA). For all analyses, surfaces were generated with consistent parameters across images, using a smoothing factor of 0.625 and a background subtraction (local contrast) defined by the diameter of the largest sphere that fits into the object (2.58 µm). For individual cell morphology analyses, cells were rendered using the surface creation wizard with split touching object enabled to separate adjacent cells with a minimum object separation of 6.5 µm. Only well-rendered cells fully contained within the hilus were included for analysis; incomplete or poorly rendered cells were excluded following visual inspection. Three brain sections were analyzed per mouse with 4–17 well-rendered cells being retained per image for morphology analysis for both Iba1-labeled microglia and GFAP-labeled astrocytes. Cell volume, area, and sphericity were extracted for Iba+ cells and cell area for GFAP+ cells. N = 5 per group were analyzed. However, one Mock mouse had substandard tissue quality resulting in suboptimal staining and was excluded from the final analysis, resulting in N = 4 for Mock and N = 5 for VEEV + N.

scRNA-seq sample collection

Mice were euthanized via CO₂ asphyxiation and secondarily euthanized via cardiac exsanguination. Mice were perfused with 10mL heparin (2000 units/kg) or 1% EDTA in phosphate buffered saline (PBS) via the apex of the heart into the left ventricle followed by 50mL of PBS. Brains were dissected and left hemisphere was placed into 10% formalin. From the right hemisphere, the hippocampus was dissected and placed into 10x genomics fixation buffer solution (Chromium Next GEM Single Cell Fixed RNA Sample Preparation Kit #1000414, 10x Genomics, Pleasanton, CA). N = 5 pooled hippocampus was minced in a petri dish until they could be pipetted through a wide bore pipette tip and weighed to achieve 30mg total fixed tissue. After 24-hour fixation, the pieces were spun down and washed with PBS following 10X sample fixation protocol and stored at 4°C prior to sending samples out for sequencing.

scRNA-seq sample purification, processing, and analysis

scRNA-sequencing was performed by MedGenome (Foster City, CA). Fixed and quenched tissues were dissociated for 10 minutes at 37°C using an Miltenyi Octo Dissociator (Mitenyi Biotec, Gladbach, Germany #130–096). Dissociated tissue were filtered through a 30μm filter to remove debris and undissociated tissue pieces, and centrifuged at 850 rcf for 5 minutes. The cell pellet was resuspended in chilled quenching buffer. Cells were stained with ethidium homodimer-1 and counted using countess II FL. The cell concentration was determined based on the reactive fluorescent protein positive concentration. For custom probe spike in, single cell gene expression flex probe configuration was used with sequences (Table 4) ordered from Integrated DNA Technologies (Integrated DNA Technologies, Coralville, IA). Each probe represents 25 bp sequences homologous to the target VEEV viral RNA (VEEV genome identified from NCBI GenBank: L01443.1). Pool 1 (Probe pairs 1–6) target the VEEV structural polyprotein gene and Pool 2 (Probe pairs 7–12) target the VEEV nonstructural polyprotein gene. Note that Probe 8 resulted in no captured sequences from any sample and therefore this probe was excluded from the study. One – two million cells were hybridized with a modified probe hybridization mix containing mouse WTA probes BC001 and custom probes, each probe at 40 nM for 16–24 h at 42°C. Each Left-hand probe (LHS) contained a linker region 5’-CCTTGGCACCCGAGAATTCCA-target_LHS-3’ and right-hand probe/5Phos/-target_RHS-ACGCGGTTAGCACGTANNACTTTAGGCGGTCCTAGCAA-3’. Following probe hybridization, gel beads in emulsions were generated for 6000 cells targeted recovery using 10x genomics chromium fixed RNA profiling reagent kit (10X Genomics, Pleasanton, CA, #1000414). The libraries were generated and sequenced in NovaSeq X for 120M paired reads.

Download:

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

Table 4. Probe Identifiers for scRNA-seq. All sequences listed 5’- > 3’.

https://doi.org/10.1371/journal.ppat.1014115.t004

The raw sequencing reads were trimmed and aligned to the mouse reference genome (mm10) using Cell Ranger (10x genomics, v.8.0.1) and quantified using ‘Cell Ranger count’. The resulting count matrix was further processed using the R package Seurat (v.5.1.0) for quality control and downstream analysis [114]. The following filtering steps were applied to remove low-quality cells, cells with the number of expressed genes (denoted as nFeature_RNA) over 200 and cellular mitochondrial genes detected < 10% were retained. Based on the 2,000 most variable genes, the normalized scRNA-seq count matrix was used for subsequent PCA analysis. The first 20 PCs were used for uniform manifold approximation and projection (UMAP) and the shared nearest neighbor (SNN) computation. The harmony package (v1.2.1) was adopted to remove batch effects among samples [115]. The gene expression comparison between samples was performed by FindMarkers function in Seurat package based on a Wilcoxon Rank Sum test with parameter “min.pct=0.05”. The non-parametric statistical method, Wilcoxon test compares the ranks of gene expression values between two groups without assuming normal distribution, making it robust for sparse scRNA-seq data. It tests whether the expression distributions are stochastically equal, with significant p-values indicating differential expression. Gene expression violin plots were generated, with asterisks indicating statistical significance levels as: *p < 0.05, **p < 0.01, and *p < 0.001 (ns = not significant, p ≥ 0.05). Cell type markers were identified utilizing published literature as shown in Table 5.

Download:

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

Table 5. Markers used for cell annotation.

https://doi.org/10.1371/journal.ppat.1014115.t005

Ingenuity pathway analysis

Filtering was performed to P-value cutoff p < 0.0001. Z-scores that are not a number (pathway activation or inhibition are unable to be predicted) are displayed as zero on the heatmaps.

Statistical analysis

Statistical analysis for the scRNA-seq data is described above. For all other data, unless otherwise specified, statistical analyses involving two or more groups were conducted using a one-way ANOVA, with asterisks denoting significance levels as follows: p > 0.05, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. For comparisons between two experimental groups, a Student’s two-tailed t-test was applied. All statistical analyses were performed using GraphPad Prism Version 10.

deposited in GitHub: https://github.com/Gavin-Yinld/VEEV_scRNA-seq