JEV infection induces broad metabolic reprogramming in mouse brain

To gain biological insights into the reprogramming of key host metabolic pathways induced by JEV-infection, six-week-old female C57BL/6 mice were intracranially injected with 200 PFU of JEV P3 strain or an equal volume DMEM as a control. At 5 dpi, the mouse brains were harvested and then subjected to metabolic analysis using LC-MS (Fig 1A). Principal component analysis (PCA) showed distinct clustering between the mock-infected controls and the JEV-infected mouse brains, indicating substantial metabolic differences between the two groups (S1A Fig). Metabolomic profiling of brain tissues revealed profound metabolic remodeling upon JEV infection, with 260 metabolites exhibiting significant alterations (VIP > 1, p value<0.05, |Log₂FC| > 1) (S1B and S1C Fig, S3 Table). Among these, 204 metabolites were found to be upregulated, while 56 were downregulated in the brains of JEV-infected mice compared to the mock-infected controls (S1B and S1C Fig). The Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis identified the top 25 significantly enriched metabolic pathways, including nucleotide metabolism (encompassing purine and pyrimidine metabolism), one-carbon (1C) metabolism, glycolysis, and amino acid metabolism (Fig 1B). Heatmap visualization further revealed global metabolic shifts associated with these enriched pathways, characterized by marked accumulation of glycolytic intermediates and significant alterations in nucleotide precursors (Fig 1C).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. JEV infection induces significant metabolic reprogramming in mice brains.

(A) Schematic diagram of the experimental design. Six-week-old female C57BL/6 mice were intracranially injected with 200 PFU of JEV P3 strain or DMEM (n = 5). Brains were harvested at 5 dpi for LC-MS-based metabolomic analysis. Created in BioRender. Gu, J. (2026) https://BioRender.com/f2uz4b3. (B) KEGG pathway enrichment analysis of significantly altered metabolites shows the top 25 enriched metabolic pathways. (C) Heatmap of Z-scores showing the relative abundance of key metabolites from the top 25 enriched pathways in mock and JEV-infected brains (n = 5). (D) Integrated metabolic network analysis. Schematic representation of key metabolic pathways, including glycolysis, TCA cycle, pentose phosphate pathway, one-carbon metabolism, nucleotide metabolism, and amino acid metabolism, with metabolite abundance changes overlaid. Data are presented as mean ± s.e.m.; statistical significance was determined using two-tailed unpaired Student’s t-tests (n = 5 mice per group, *p < 0.05, **p < 0.01, ***p < 0.001). (E) Transcriptomic analysis of enzyme-encoding genes in key metabolic pathways. The heatmap displays transcriptional changes of genes involved in indicated pathways in mouse brain between JEV-infected and mock-infected groups. Statistical significance was determined using two-tailed unpaired Student’s t-tests (n = 3 mice per group, *p < 0.05, **p < 0.01, ***p < 0.001).

https://doi.org/10.1371/journal.ppat.1014335.g001

To visualize the metabolic crosstalk and identify critical nodes, we constructed an integrated network map synthesizing glycolysis, TCA cycle, PPP, one-carbon metabolism, nucleotide metabolism, and key amino acid branches, overlaying metabolite abundance (Fig 1D, S3 Table). In parallel, to explore potential regulatory mechanisms underlying the observed metabolic alterations, we integrated these metabolomic findings with our previous transcriptomic data [27], focusing specifically on enzyme-encoding genes within enriched pathways (Fig 1E). This integrated analysis revealed a profound reprogramming of CCM characterized by the accumulation of glycolytic intermediates, such as glucose-6-phosphate, fructose-6-phosphate, fructose-1,6-bisphosphate, glyceraldehyde-3-phosphate, and phosphoenolpyruvate, despite significant glucose depletion and stable lactate levels (Fig 1D). This metabolic rewiring coincided with the transcriptional upregulation of the rate-limiting enzyme Hk2, indicating Hk2-driven substrate channeling that enhances the glycolytic pathway (Fig 1E). Analysis of TCA cycle metabolites revealed a significant accumulation of succinate, while citrate, isocitrate, and malate exhibited no significant changes (Fig 1D). Concurrently, transcriptomic profiling showed a significant downregulation of core TCA enzymes: Ogdh, Sdhc, Pcx, Idh2, Idh3g, Aco2 and Mdh2 (Fig 1E). Among PPP metabolites analyzed, erythrose-4-phosphate (E4P) was uniquely downregulated, while other intermediates showed no significant alterations (Fig 1D). Concurrently, transcriptional upregulation of G6pdx and Pgd was observed (Fig 1E). This coordinated enhancement of oxidative PPP enzymes, along with E4P depletion suggests amplified flux through the oxidative branch and attenuated activity in the non-oxidative branch in infected mouse brains. Depleted aspartate and glutamate abundances, alongside elevated glutamine levels (Fig 1D), aligned with transcriptional alterations of Asns, Asrgl1, Glud, and Glul (Fig 1E). Furthermore, the observed upregulation in the abundance of 3-phosphoserine and serine indicates enhanced de novo serine synthesis in the JEV-infected mouse brain (Fig 1D). We also found increased transcriptional levels of key 1C metabolism enzymes, including Dhfr, Shmt2, Mthfd2, Mthfd1l, and Aldh1l1/2 (Fig 1E). Notably, the upregulation of Shmt2, which catalyzes the direct conversion of serine to glycine while generating 5,10-methylene-THF, aligns with the observed increase in glycine abundance (Fig 1D and 1E). This further corroborates enhanced 1C metabolism. However, the levels of 10-Formyl-THF were significantly decreased (Fig 1D). This apparent contradiction may be attributed to the heightened consumption of 10-formyl-THF, potentially driven by hyperactive purine biosynthesis which utilizes this cofactor as a critical formyl donor.

In our metabolomic analysis, intermediates of nucleotide metabolism exhibited deeply pronounced alterations (Fig 1D). Though transcriptional analysis showed no significant changes in the expression of critical de novo pyrimidine biosynthesis enzyme genes Dhodh and Cad, key intermediates of de novo pyrimidine biosynthesis, including orotate and UMP were significantly upregulated compared to uninfected controls (Fig 1E). This seems to be distinct from the brain’s generally limited capacity for de novo nucleotide synthesis under normal physiological conditions. Concurrently, we observed a significant decrease in the levels of uracil and uridine, whereas cytosine, thymine, thymidine, deoxyuridine, and deoxycytidine were significantly increased (Fig 1D). This reduction of uracil and uridine levels correlated with upregulated transcription of key salvage pathway enzyme genes Uck2 and Upp1 (Fig 1E). These findings suggest that JEV infection enhances both de novo pyrimidine biosynthesis and salvage pathways, ensuring sufficient nucleotide supply.

Furthermore, the transcriptomic analysis revealed that JEV infection significantly enhances purine catabolism, as evidenced by the elevated expression of key enzyme genes governing nucleotide breakdown and interconversion, including Pnp, Ada, Ampd3, Nt5c2, and crucially, Xdh (Fig 1E). This indicates an accelerated degradation of nucleotides and nucleosides toward terminal metabolites. Metabolomic profiling corroborated this shift, showing a marked accumulation of xanthine/urate (Fig 1D). Despite this catabolic bias, the upregulation of salvage enzyme APRT and the interconversion hub NT5C2 suggests a compensatory mechanism aimed at recycling residual purine bases and nucleosides. This salvage effort likely contributes to stabilizing cellular AMP/GMP pools amid heightened purine demand during infection. While de novo purine biosynthesis enzymes ADSL and ATIC were transcriptionally induced (Fig 1E), the early precursor 5-phosphoribosylamine decreased (p < 0.01) without detectable expansion of the IMP pool (Fig 1D), implying substrate consumption exceeding synthetic capacity. Finally, induction of mitochondrial Ak2 likely preserved energy homeostasis by dynamically balancing ATP/ADP/AMP ratios (Fig 1E), thereby supporting the high-flux purine metabolic state during JEV infection. Collectively, these data demonstrate that JEV systematically hijacks host metabolic networks in the brain, orchestrating a multi-layered reprogramming of central carbon and nucleotide metabolism.

Targeted metabolomics reveals rewiring of central carbon metabolism in JEV-infected neurons

To validate the widespread perturbations across multiple metabolic networks identified by the untargeted metabolomic analysis, we performed a targeted metabolomic analysis in mock and JEV-infected N2a, a mouse neuronal cell line (Fig 2A). This analysis specifically profiled multiple core intermediates of CCM, which integrates glycolysis, TCA cycle, and PPP (Fig 2B), serving dual roles in energy production and in supplying precursors for biosynthesis [28]. The reproducibility of the metabolomics dataset was confirmed by PCA, which showed tight clustering of biological replicates and clear separation between mock‑ and JEV‑infected groups (S1D Fig). Targeted metabolomic profiling revealed extensive remodeling of CCM induced by JEV infection, involving a substantial number of differentially regulated core intermediates (Fig 2C, S4 Table). The TCA cycle exhibited specific dysregulation, characterized by a coordinated accumulation of downstream intermediates (including malate, cis-aconitate, and α-ketoglutarate), while acetyl-CoA levels remained unchanged (Fig 2D). Concurrently, there was an accumulation of several glycolytic intermediates, including G6P, G3P, and 2-phosphoglycerate (2PG) (Fig 2E), indicating an enhancement of glycolysis. Among these intermediates, G3P can be converted into ribose-5-phosphate (R5P) via carbon-shuffling reactions involving sedoheptulose-7-phosphate (S7P) in the non-oxidative PPP (non-oxPPP) [29], thereby directly fueling nucleotide synthesis. In addition, the PPP in JEV-infected N2a cells showed phase-specific activation, characterized by an increase in the levels of ribulose-5-phosphate (Ru5P) and non-oxidative branch metabolites such as E4P and xylulose-5-phosphate (Xu5P) (Fig 2F). Notably, the abundance of serine, which potentially contribute to folate-mediated one-carbon metabolism to support nucleotide biosynthesis, was also upregulated (Fig 2G). This aligns with the pronounced accumulation of purine pathway intermediates, including IMP, AMP, and other purine nucleosides and bases (Fig 2H). These findings reveal that JEV significantly disrupts CCM networks associated with nucleotide biosynthesis in N2a cells.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Targeted metabolomics identifies JEV-induced rewiring of central carbon metabolism in N2a Cells.

(A) Schematic of the experimental workflow for targeted metabolomic analysis. N2a cells were infected or mock-infected with JEV P3 strain at an MOI of 5. The cells were collected at 24 hpi for CCM-targeted metabolomic analysis. Created in BioRender. Gu, J. (2026) https://BioRender.com/f2uz4b3. (B) Schematic representation of integrated central carbon metabolism pathways. Metabolites that were significantly upregulated in JEV-infected N2a cells compared to mock-infected controls are highlighted in red. No metabolites were significantly downregulated in this targeted dataset. (C) Heatmap visualization of detected cellular metabolites in JEV-infected vs. mock-infected N2a cells at 24 hpi (n = 3). (D-H) Comparative analysis of metabolic alterations in key pathways, including TCA cycle (D), glycolysis (E), pentose phosphate pathway (PPP) (F), amino acid metabolism (G), and nucleotide metabolism (H). Data are presented as mean ± s.e.m. (n = 3). Statistical significance was determined using two-tailed unpaired Student's t-tests (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001).

https://doi.org/10.1371/journal.ppat.1014335.g002

[U-¹³C] Glucose tracing reveals an enhancement of de novo nucleotide biosynthesis flux induced by JEV infection in neurons

To address the dynamic rewiring of CCM implied by our static metabolomic profiles, we conducted [U-¹³C] glucose tracing to track isotopic enrichment across metabolites of CCM in JEV-infected N2a cells at 24 hpi (Fig 3A). Following incubation with [U-¹³C] glucose, isotopic equilibration through multiple turns of TCA cycle resulted in diverse labeling patterns in cycle intermediates, as evidenced by the co-occurrence of M + 2 to M + 5 and fully labeled M + 6 isotopologues (S2A Fig). This dynamic carbon cycling reflects the iterative entry of ¹³C-acetyl-CoA and the anaplerotic replenishment of oxaloacetate pools. Our isotope tracing analysis revealed that JEV-infected N2a cells displayed elevated rates of [U-¹³C] glucose progressive incorporation into all detected TCA intermediates, including citrate, aconitate, α-ketoglutarate, succinate and malate (S2B Fig, S5 Table), confirming that JEV infection significantly enhanced TCA cycle flux in host cells. Interestingly, acetyl-CoA flux remained stable despite the elevated activity of TCA cycle driven by JEV, indicating JEV’s engagement of alternative pathways to sustain this increased TCA cycle activity. Concurrently, an increase in glycolytic metabolic flux was evidenced by the incorporation of ¹³C into M + 3 pyruvate and M + 3 lactate, suggesting a virally hijacked Warburg-like metabolic reprogramming (S2C Fig).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. [U-¹³C] Glucose tracing reveals JEV-induced enhancement of de novo nucleotide biosynthesis flux in N2a cells.

(A) Schematic of the experimental design for [U-¹³C] glucose tracing assay. N2a cells were infected with JEV at a MOI of 5, followed by incubation in glucose-deficient medium supplemented with dialyzed serum and [U-¹³C] glucose for 24 h. Created in BioRender. Gu, J. (2026) https://BioRender.com/f2uz4b3. (B) Pathways of de novo and salvage nucleotide biosynthesis showing the incorporation of glucose-derived carbon. The ribose sugar moiety derived from glucose contributes five ¹³C atoms (M + 5) to nucleotides, while the nucleobases incorporate ¹³C from various precursors (e.g., glycine, formate, aspartate, and glutamine) leading to intermediate (0 < M < 5) or fully labeled (M > 5) species when combined with labeled ribose. (C–E) Normalized peak areas of isotopologues for NTPs (C), NDPs (D), and NMPs (E) in mock- and JEV-infected N2a cells. Data are presented as mean ± s.e.m. (n = 3). Statistical significance was determined using two-tailed unpaired Student’s t-tests (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001).

https://doi.org/10.1371/journal.ppat.1014335.g003

Nucleotide synthesis encompasses both de novo synthesis and the salvage synthesis pathway. In general, proliferating cells predominantly depend on de novo nucleotide biosynthesis to satisfy the heightened demand for DNA/RNA precursors during rapid proliferation [30,31]; whereas neurons primarily rely on nucleotide salvage pathways due to their higher levels of salvage enzymes like HPRT but lack key de novo synthesis machinery [22,32], making salvage pathways essential for maintaining nucleotide pools. To investigate whether the rapid replication of JEV influences the nucleotide biosynthesis pathway in neurons, we performed metabolic flux analysis by tracking the incorporation of [U-¹³C] glucose-derived carbon into nucleotide mono-, di-, and triphosphates, as well as 5-phosphoribosyl-1-pyrophosphate (PRPP) (Fig 3B). Pre-existing unlabeled nucleotide pools within cells contribute to the generation of M + 0 isotopologues, reflecting the incorporation of endogenous, non-¹³C-labeled precursors during nucleic acid turnover. Since the ribose moiety of nucleotides typically incorporates five ¹³C atoms from [U-¹³C] -glucose [30], M + 5 labeled nucleotides may originate from either de novo synthesis requiring PRPP-dependent assembly of small-molecule precursors, or salvage pathways involving attachment of a labeled ribose to pre-existing unlabeled nucleobases (Fig 3B) [33]. The ¹³C-labeling patterns of nucleobases illustrate the integration of carbon precursors from multiple pathways. Glycine, formate, glutamine, and aspartate collectively contribute to the progressive labeling, generating 0 < M < 5 intermediates initially, before reaching fully labeled states (M > 5 when combined with the labeled ribose moiety). Despite isotopic labeling for 24h, substantial unlabeled nucleotide fractions (M + 0) persisted in both the mock- and JEV-infected groups, indicating a continuous turnover of pre-existing nucleotide pools (Fig 3C–3E). In terms of nucleotide triphosphates (NTPs), GTP exhibited comprehensive upregulation across multiple isotopologues, while CTP and UTP demonstrated increases only in the M + 6 and M + 3 fractions, respectively (Fig 3C). Although the relative distribution of ATP isotopologues shifted, with M + 1 to M + 5 fractions increasing and M + 6 to M + 10 fractions decreasing (S2D Fig), the normalized peak area data showed no significant change in either individual isotopologue abundance following JEV infection (Fig 3C), indicating that JEV infection does not grossly alter ATP metabolic flux. However, JEV infection widely increased the levels of detected isotopologues of nucleotide diphosphates (NDPs) and nucleotide monophosphates (NMPs) compared to the mock-infected controls (Fig 3D and 3E). The increased dominant M + 5 fractions of these NMPs may indicate coordinated activation of both de novo synthesis and salvage pathways. In contrast, only the M + 4 fraction of IMP was elevated in JEV-infected cells (Fig 3E). Further analysis of the distributions of 0 < M < 5 and M > 5 fractions of these NMPs revealed a significant upregulation of both partially labeled (M + 1 to M + 4) and fully labeled (M + 6 to M + 10) NMP species during JEV infection in N2a cells (S2E Fig). This metabolic reprogramming was further corroborated by the elevated levels of M + 5 labeled PRPP, an essential gateway metabolite for de novo purine and pyrimidine synthesis (S2F Fig). The coordinated upregulation of PRPP and downstream labeled NMPs suggests that JEV strategically hijacks the host’s de novo nucleotide biosynthesis by augmenting PRPP supply through the PPP and accelerating the integration of carbon-nitrogen into nucleobases. Moreover, the limited detection of M + 1 to M + 4 nucleotide fractions and the high levels of M > 5 nucleotides observed in the results likely reflects the slower kinetics of ¹³C incorporation into nucleobases compared to ribose moieties (S2E Fig). Taken together, these findings demonstrate that JEV infection profoundly reprograms the CCM flux in N2a cells, with particularly striking alterations in nucleotide biosynthetic pathways. Our isotopic tracing data provide direct evidence that this rewiring of nucleotide biosynthesis is mediated, at least in part, by enhanced de novo nucleotide synthesis.

JEV hijacks de novo purine synthesis for its replication in neurons

Our metabolic analysis suggested that de novo nucleotide pathway was activated in N2a cells during JEV infection. Given that the genome of JEV P3 strain exhibits a purine-biased nucleotide composition (A: 27.61%, G: 28.48%, C: 22.96%, U: 20.95%), with purines (A + G, 56.09%) significantly outnumbering pyrimidines (C + U, 43.91%) (S3A and S3B Fig), this likely reflects a higher metabolic demand for purine precursors during viral replication. Consequently, we prioritized the investigation of JEV's manipulation of host purine biosynthesis over pyrimidine pathways. DNPB requires ten enzymatic steps to synthesize IMP from PRPP. Subsequently, IMP undergoes divergent metabolic pathways catalyzed by IMPDH/GMPS or ADSS/ADSL, leading to the production of GMP and AMP, respectively. To deepen our understanding of how JEV modulates de novo purine biosynthesis (DNPB), we measured the mRNA levels of relevant enzymes in JEV infected N2a cells and primary neurons. The results revealed a significant transcriptional upregulation of all core components of the DNPB pathway in N2a cells infected with 5 MOI of JEV at 24 and 36 hpi (S4A Fig), or 1 MOI of JEV at 48 hpi (S4B Fig). A similar transcriptional upregulation of these enzymes was also observed in mouse primary neurons and TM4 cells (a mouse Sertoli cell line) (S4C and S4D Fig). Subsequently, we assessed the corresponding changes at the protein level using Western blot analysis. Consistent with the qPCR results, the protein levels of PFAS, GART, IMPDH2, ADSS, and GMPS were significantly upregulated at 48 h post infection of JEV at 1 MOI (Fig 4A and 4B). Collectively, these results demonstrate a widespread upregulation of de novo purine biosynthetic enzymes in JEV-infected neurons.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 4. JEV hijacks de novo purine synthesis for its replication in neurons.

(A–B) Impact of JEV infection on protein expression of key DNPB enzymes. N2a cells were infected or mock-infected with JEV at an MOI of 1. The expression of PFAS, GART, IMPDH1, IMPDH2, ADSS, GMPS, and β-Tubulin proteins were detected by immunoblotting, and representative blots from 3 independent experiments are presented (A). Protein levels were quantified by using Image J software, normalized to β-Tubulin, and presented relative to the mock-infected control (n = 3) (B). (C–E) Inhibitory effect of MPA on JEV replication. N2a cells were infected with JEV at an MOI of 5, followed by treatment with either 50 nM MPA or an equivalent volume of DMSO at 1.5 hpi. The levels of intracellular viral RNA (C) and E protein (D), as well as the infectious virus production in the supernatant (E) at 6, 12, 18, and 24 hpi, were determined using qRT-PCR, immunoblotting, and plaque assay, respectively. (F–H) The effect of guanosine (Gua) treatment on MPA-mediated inhibition of JEV replication. N2a cells infected with JEV were treated with either MPA or DMSO, with or without addition of 100 μM Gua. The levels of intracellular viral RNA (F) and E protein (G), as well as the production of infectious virus in the supernatant (H), were determined using qRT-PCR, immunoblotting, and plaque assay, respectively, at 18 hpi. (I–J) Inhibitory effect of Ag2037 on JEV replication. N2a cells were infected with JEV at an MOI of 5, followed by treatment with either 0.5 μM Ag2037 or an equivalent volume of DMSO at 1.5 hpi. The levels of intracellular viral RNA (I) and viral titers in the supernatant (J) were determined at 6, 12, 18, and 24 hpi. (K–L) The effect of hypoxanthine (Hyp) treatment on Ag2037-mediated inhibition of JEV replication. N2a cells infected with JEV were treated with either Ag2037 or DMSO, with or without addition of Hyp at 1.5 hpi. The levels of intracellular viral RNA (K) and viral titers in the supernatant (L) were measured at 18 hpi. Data are presented as mean ± s.e.m. from 3 independent biological replicates (n = 3). Significance was determined using two-tailed unpaired Student’s t-test (B), one-way (F, H, K and L) or two-way ANOVA (C, E, I and J) (*p < 0.05, **p < 0.01, ***p < 0.001; ****p < 0.0001; ns, not significant).

https://doi.org/10.1371/journal.ppat.1014335.g004

Additionally, we analyzed our previous scRNA-seq datasets from JEV-infected mouse brains [34], focusing specifically on transcriptional expression of DNPB enzymes in neuronal populations. Based on the levels of viral gene expression, we classified all neuronal cells into five distinct groups: the mock-infected group (Mock), JEV-exposed but viral genome-negative neurons (JEV-N, JEV expression = 0), viral genome-low neurons (JEV-L, log2[JEV expression] < 3), viral genome-medium neurons (JEV-M, 3 ≤ log2[JEV expression] ≤ 8), and viral genome-high neurons (JEV-H, log2[JEV expression] > 8), establishing a comprehensive gradient model to investigate neuronal responses across the entire spectrum of infection intensity, ranging from viral exposure to high viral burden. As JEV infection progressed, we observed a striking divergence in the transcriptional dynamics of DNPB enzymes (S4E Fig). The mRNA levels of PAICS, PFAS, GMPS, and ADSS showed a sustained decrease with increasing viral burden. Notably, these enzymes catalyze ATP- and GTP-consuming steps, suggesting that JEV infection suppresses the energy-intensive branches of DNPB to reduce host metabolic expenditure. In contrast, the virus appeared to selectively hijack the pathway’s rate-limiting nodes PPAT and IMPDH1 transcripts were downregulated at low infection levels but markedly upregulated in heavily infected neurons, highlighting a biphasic regulation pattern. Given the isoform-specific expression of IMPDH1 and IMPDH2 in different tissues, our data further suggest a possible isoform switch during infection, with IMPDH1 induction compensating for the decline of IMPDH2 to sustain GTP availability under high viral load. Interestingly, in low and medium level infected neurons, GART, ADSL, IMPDH2, and ATIC were transiently upregulated, peaking in the low-infection groups before declining at higher viral burdens. Thus, our single-cell transcriptomic profiling of infected mouse brain neurons reveals a more dynamic and refined transcriptional landscape that is induced by JEV infection. This landscape includes the sustained downregulation of several ATP- or GTP-consuming enzymes alongside the biphasic upregulation of key rate-limiting enzymes such as PPAT and IMPDH1 in heavily infected neurons.

To investigate the role of DNPB in JEV replication, we next pharmacologically inhibited IMPDH, the rate-limiting enzyme in this pathway, using MPA. The results demonstrated that treatment with 50 nM MPA, which maintained over 90% cell viability (S5A Fig), significantly suppressed JEV RNA replication (Fig 4C), reduced viral E protein expression (Fig 4D), and decreased viral titers (Fig 4E) compared to the DMSO-treated group. To confirm pathway specificity, we rescued MPA-mediated inhibition by supplementing guanosine (Gua), the downstream product of IMPDH. Remarkably, addition of 100 μM Gua fully restored viral RNA levels (Fig 4F), E protein expression (Fig 4G), and infectious particle production (Fig 4H). These findings establish the critical dependence of JEV replication on de novo purine biosynthesis. To systematically investigate the impact of DNPB upstream pathways on JEV replication, we treated the N2a cell with 0.5 μM Ag2037 (S5B Fig), an inhibitor of the GART enzyme. Similar results were obtained with Ag2037 inhibition of JEV replication (Fig 4I and 4J). Furthermore, the antiviral effect of Ag2037 could be completely rescued by exogenous supplementation with hypoxanthine, a critical downstream metabolite of the DNPB pathway (Fig 4K and 4L). Collectively, these findings indicate that JEV hijacks de novo purine synthesis for its replication in host neuronal cells.

JEV reprograms the pentose phosphate pathway to fuel nucleotide biosynthesis via non-oxidative branch dependency in neurons

PPP serves as the primary source of ribose-5-phosphate (R5P), an essential precursor for PRPP synthesis required by DNPB. The PPP comprises both oxidative (generating NADPH and Ru5P) and non-oxidative (producing R5P) branches (Fig 5A) [35]. Metabolic flux analysis revealed that JEV infection induces a pronounced divergence between the oxidative and non-oxidative branches of the PPP. Specifically, M + 7 labeled sedoheptulose-7-phosphate (S7P) in the non-oxidative branch exhibited increased flux (Fig 5B), whereas intermediates of the oxidative branch, such as M + 6 6-phosphogluconic acid, remained unchanged (Fig 5C). To characterize the modulation of PPP enzyme expression in neurons by JEV infection, we first conducted qRT-PCR on JEV-infected N2a cells, primary neurons and TM4 cells, which confirmed a consistent upregulation pattern for enzymes in the PPP (S4F-S4H Fig). Subsequently, we assessed the protein expression levels of two key PPP enzymes, G6PD and TKT. Western blot analysis revealed that both enzymes were upregulated following JEV infection in N2a cells (Fig 5D and 5E). The scRNA-seq analysis also demonstrated that neurons from JEV-infected mice exhibited significantly elevated transcription of PPP-related enzymes in neurons with low and medium viral burden neurons compared to the Mock group (S4I Fig).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 5. JEV reprograms the PPP to fuel nucleotide biosynthesis through non-oxidative branch dependency in neurons.

(A) Schematic of the pentose phosphate pathway, highlighting the oxidative branch and the non-oxidative branch. (B–C) Metabolic flux analysis using [U-¹³C] glucose tracing. The flux in the non-oxidative branch intermediate sedoheptulose-7-phosphate (M + 7) (B) and the oxidative branch intermediate 6-phosphogluconate (M + 6) (C) were presented. (D–E) Expression of the key PPP enzymes G6PD and TKT in N2a cells infected or mock-infected with JEV was detected using immunoblotting at 36 hpi, and the representative blots were presented (D). The relative expression levels of these proteins were calculated (E). (F–J) Effect of the oxidative PPP branch inhibitors on JEV replication. N2a cells were infected with JEV at an MOI of 5, followed by treatment of 250 nM RRX-001 at 1.5 hpi. The relative viral RNA levels at indicated time points were measured (F). Additionally, increasing concentrations of RRX-001 (G) or 6-AN (H) were used to treat the JEV-infected N2a cells, and the viral RNA levels were determined. (I–J) The regulatory role of G6PD in JEV replication in N2a cells. The G6pdx knockout (KO) N2a cell line and the control cells (NC) were infected with JEV at an MOI of 5. The expression of G6PD and JEV E proteins (I) and the viral RNA levels (J) were assessed at 24 hpi. (K–L) The regulatory role of TKT in JEV replication in N2a cells. Tkt-deficient (sg1 and sg2) and negative control (NC) N2a cells were infected with JEV at an MOI of 5. The expression of TKT and JEV E proteins (K) and the viral RNA levels were determined (L). Data are presented as mean ± s.e.m. from 3 independent biological replicates (n = 3). Significance was determined by two-tailed unpaired Student’s t-test (B-C, E, J and L), one-way ANOVA (G-H), or two-way ANOVA (F) (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001; ns, not significant).

https://doi.org/10.1371/journal.ppat.1014335.g005

To investigate whether the PPP contribute to JEV replication in neurons, we employed G6PD inhibitors, RRx-001 (S5C Fig) and 6-AN (S5D Fig), to suppress the oxidative branch during JEV infection in N2a cells. However, treatment with these inhibitors failed to suppress viral replication (Fig 5F–5H), and a high dose of RRx-001 even paradoxically enhanced JEV RNA accumulation (Fig 5G). This phenotype was further confirmed in G6pdx knockout (KO) cells, where viral RNA and protein expression also significantly increased compared to negative control (NC) N2a cells (Fig 5I and 5J). We subsequently assessed the role of the non-oxidative PPP in JEV replication. We found that deletion of the non-oxidative branch enzyme gene Tkt resulted in a significant reduction of JEV RNA and protein levels (Fig 5K and 5L), confirming its critical role in sustaining infection. These results highlighted a critical dependency of JEV replication on the non-oxidative branch of the PPP.

MTHFD2-dependent one-carbon metabolism sustains JEV replication in neurons

The DNPB pathway requires 1C units as essential building blocks for purine ring assembly (Fig 6A) [36]. Our integrated bulk RNA-seq and metabolomic analysis of JEV-infected mouse brains revealed marked transcriptional upregulation of key one-carbon metabolic enzymes and significant depletion of 10-formyl-THF (Fig 1D and 1E), implying that the virus hijacks 1C units through remodeling host one-carbon metabolism. Therefore, we speculated that folate-driven one-carbon metabolism supplies the necessary 1C units for purine synthesis, and the dynamic regulation of its enzymatic machinery may serve as a critical node for JEV replication. As expected, the upregulation of both mRNA and protein levels of key one-carbon metabolic enzymes was validated in JEV-infected N2a cells using qRT-PCR (S4J Fig) and immunoblotting (Fig 6B and 6C), respectively. Except Aldh1l1, all detected enzymes increased significantly in their transcriptional levels in JEV-infected primary neurons (S4K Fig). Consistent with these neuronal findings, MTHFD1 and MTHFD2 mRNA expression was also upregulated in TM4 cells (S4L Fig). Intriguingly, the scRNA-seq analysis at single-neuron resolution revealed a compartmentalized shift in enzyme expression: mildly infected neurons (JEV-L) predominantly expressed cytosolic enzymes such as SHMT1 and MTHFD1, whereas severely infected neurons (JEV-H) exhibited elevated expression of mitochondrial enzymes like MTHFD2 and SHMT2 (S4M Fig).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 6. MTHFD2-dependent mitochondrial one-carbon metabolism is essential for JEV replication in neurons.

(A) Schematic diagram illustrating the metabolic origins of purine ring atoms, highlighting the contribution of key metabolites. One-carbon (1C) units are specifically required for purine ring assembly at the C2 and C8 positions. (B–C) Expression of key 1C metabolic enzymes in JEV-infected N2a cells. N2a cells were infected or mock-infected with JEV at an MOI of 5. The indicated enzymes were detected by using Western blotting at 36 hpi (B), and the relative protein levels were calculated (C). Representative blots from 3 independent experiments are presented. (D–E) Effect of MTHFD2 inhibitor on JEV replication. N2a cells were infected with JEV at an MOI of 5, followed by treatment with increasing concentrations of DS18561882 at 1.5 hpi. The viral RNA levels (D) and titers (E) were assessed at 18 hpi. (F–G) Effect of sodium formate and hypoxanthine on DS18561882-mediated inhibition of JEV replication. JEV-infected N2a cells were treated with DS18561882, supplemented with or without sodium formate or hypoxanthine. The viral RNA levels (F) and titers (G) were determined at 18 hpi. (H–I) Role of MTHFD2 on JEV replication in N2a cells. Mthfd2 KO N2a cells line and the negative control cells were infected with JEV at an MOI of 5. The viral RNA accumulation (H) and E protein expression (I) were measured. Data are presented as mean ± s.e.m. from 3 independent biological replicates (n = 3). Significance was determined by two-tailed unpaired Student’s t-test (C and I) or one-way ANOVA (D-G). (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001; ns, not significant).

https://doi.org/10.1371/journal.ppat.1014335.g006

To investigate the impact of 1C metabolism on JEV replication in neurons, we treated N2a cells with DS18561882 (S5E Fig), a selective inhibitor of MTHFD2, to obstruct the 1C metabolic pathway during JEV infection. Our results demonstrated that DS18561882 effectively inhibited JEV RNA replication and significantly reduced viral titers in a dose-dependent manner (Fig 6D and 6E). Notably, the supplementation of sodium formate or hypoxanthine remarkably rescued JEV replication efficiency (Fig 6F and 6G), confirming that DS18561882 specifically suppresses viral replication through MTHFD2 inhibition and that JEV replication depends strictly on functional one-carbon metabolism to furnish essential precursors for purine nucleotide synthesis. To further validate the essential role of MTHFD2 in JEV replication, we generated a Mthfd2 KO N2a cell line. The Mthfd2 deficiency significantly suppressed both JEV RNA and protein synthesis (Fig 6H and 6I). These findings collectively establish mitochondrial 1C metabolism, particularly MTHFD2-mediated folate cycling, as a critical metabolic vulnerability for JEV replication in neuronal cells.

Ag2037 treatment reduces viral loads and attenuates neuroinflammation in mouse brains during JEV infection

To further assess the role of DNPB in JEV replication in vivo, we employed the GART inhibitor Ag2037 in a JEV-infected mouse model. To minimize its inhibitory effects on DNPB in peripheral tissues, mice that either infected or mock-infected with JEV were administered with Ag2037 intravenously at 3 and 6 dpi (Fig 7A), considering that CNS invasion of JEV has been reported to occur at 3 dpi [37]. Our findings revealed that JEV-infected mice treated with Ag2037 exhibited milder symptoms and improved survival outcomes compared to the vehicle-treated group, as evidenced by an extended time-to-death (median survival of 11.5 days vs 8.5 days) and a reduced mortality rate (Fig 7B). Consistent with these survival benefits, Ag2037 significantly suppressed viral RNA accumulation and E protein expression in the mouse brain at 6 dpi as demonstrated by RT-PCR and immunohistochemistry analysis (Fig 7C–7E). As anticipated, a significant reduction in JEV viral RNA copies was observed exclusively in the brain tissues of Ag2037-treated mice, rather than in other peripheral tissues (Fig 7C), indicating that the effect of Ag2037 was limited to the central nervous system under the conditions of this study.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 7. Ag2037 treatment reduces viral load and attenuates neuroinflammation in JEV-infected mice.

Six-week-old C57/BL6 mice were intravenously infected or mock-infected with 10⁵ PFU JEV, followed by i.v. injected with 15mg/kg Ag2037 at 3 and 6 dpi (A). Created in BioRender. Gu, J. (2026) https://BioRender.com/f2uz4b3. The survivals of treated mice were monitored until 21 dpi (n = 20 mice/group) (B). The JEV RNA copies and the expression of E protein in the brains were determined using RT-qPCR (n = 3 mice/group) (C) and immunohistochemical staining (D) at 6 dpi. The integrated optical density (IOD) of E protein signals was calculated with 3 sections for each mouse (n = 3 mice/group) using Image-Pro Plus 6.0 software (E). Additionally, the mRNA levels of type I interferon and pro-inflammatory cytokines (IFN-β, TNF-α, CCL-2, CCL-5, IL-6, IL-1β) in brain tissues were measured using RT-qPCR at 6 dpi (n = 3 mice/group) (F). Moreover, the brain sections were subjected to H&E staining (G) and immunohistochemistry staining of IBA-1 (H), with the IOD of IBA signals being calculated (I) (n = 3 mice/group). Data are presented as mean ± s.e.m.. Statistical significance was determined using log-rank test (B), two-tailed unpaired Student’s t-test (C and F), or one-way ANOVA (E and I) (*p < 0.05, **p < 0.01, ***p < 0.001, ****p<0.0001).

https://doi.org/10.1371/journal.ppat.1014335.g007

Considering the reduced viral loads observed in the mouse brain, which may influence the neuroinflammation caused by JEV, we next investigated the effect of Ag2037 treatment on the inflammatory response in the brains of JEV-infected mice. As shown in the results, Ag2037 treatment significantly downregulated the mRNA levels of type I interferon and pro-inflammatory cytokines, including IFN-β, TNF-α, CCL-2, CCL-5, IL-6, and IL-1β, compared to vehicle treatment (Fig 7F). Histopathological examination of brain sections via H&E staining further revealed that vehicle-treated brains displayed prominent perivascular inflammatory cuffing, indicative of lymphocyte infiltration and blood-brain barrier disruption, whereas Ag2037-treated tissues displayed alleviative pathological changes (Fig 7G). To dissect the cellular mechanisms underlying this attenuated inflammatory response, we assessed microglial activation through IBA-1 immunohistochemistry staining. Ag2037 treatment significantly suppressed microglial hyperactivation, as evidenced by a reduction in IBA-1 integrated optical density (IOD) compared to vehicle-treated controls (Fig 7H and 7I). Taken together, our data demonstrate that the administration of Ag2037 inhibits JEV replication in the CNS and attenuated the neuroinflammatory response, highlighting its therapeutic efficacy in the context of Japanese encephalitis.

Ethics statement

Female C57BL/6 mice were obtained from the Animal Center of Huazhong Agricultural University, Wuhan, China. All mouse experiments were conducted in accordance with the approved protocols of the Animal Care and Ethics Committee of Huazhong Agricultural University, under the reference number HZAUMO-2025–0082. The protocols followed the guidelines outlined in the Guide for the Care and Use of Laboratory Animals.

Cells, viruses, and viral titration

Primary neuron cultures were prepared from the cerebral cortices of C57BL6J mice at embryonic day 16, as previously reported [61]. Briefly, the isolated primary cells were seeded on poly-lysine-coated (100 mg/mL) plates at a density of 6 × 10⁵ cells per milliliter in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 5% fetal bovine serum (FBS). At 6 hours (h) post-seeding, the culture medium was replaced with neurobasal medium supplemented with 2% B-27. The cells were used for subsequent experiments after culture for 7 days with half of the medium replaced by fresh medium every 3.5 days.

N2a cells (mouse neuroblastoma cell line), HEK-293T cells (human embryonic kidney cell line), TM4 cells (mouse Sertoli cell line) and BHK-21 cells (baby hamster kidney cell line) were cultured and maintained in DMEM supplemented with 10% FBS, 100 U/ml penicillin, and 100 µg/ml streptomycin sulfate.

The JEV P3 strain was propagated in the brains of suckling mice and BHK-21 cells for in vivo and in vitro experiments, respectively. The viral titer was determined using a plaque assay on BHK-21 cells. In brief, supernatants from infected cells were harvested and serially diluted 10-fold in DMEM, followed by inoculation onto BHK-21 cells. After incubation at 37 °C for 1.5 h, the supernatants were discarded, and the cells were incubated for 4 days in DMEM containing 2% FBS and 1.5% sodium carboxymethyl cellulose at 37 °C. The cells were then fixed with 10% formaldehyde overnight and stained with crystal violet solution for 2 h. The final viral titers were calculated based on the numbers of visible plaques counted.

Reagents and antibodies

Mycophenolic acid (MPA, HY-108325), Ag2037 (HY-14530), DS18561882 (HY-130251), RRx-001 (HY-16438) and 6-Aminonicotinamide (6-AN, HY-101966) were purchased from MedChem Express. Sodium formate (71539), uridine (U3003), guanine (G6264), and hypoxanthine (H9636) were obtained from SigmaAldrich as exogenous 1C units and nucleotides. [U-¹³C] glucose was purchased from Cambridge Isotope Laboratories, Inc. (CLM-1396-1). Mouse anti-JEV-E monoclonal antibody (mAb) was prepared in our laboratory. Mouse anti-β-Actin mAb (HRP-66009), Rabbit anti-IMPDH1 polyclonal antibody (pAb, 22092-1-AP), anti-PFAS pAb (24716-1-AP), anti-TKT pAb (11039-1-AP), and anti-SHMT1 pAb (30192-1-AP) were purchased from Proteintech. Rabbit anti-GART pAb (A3876), anti-ADSS mAb (A20945), anti-IMPDH2 mAb (A9208), anti-GMPS pAb (A6606), anti-G6PD pAb (A13983), anti-MTHFD1 pAb (A8661), anti-SHMT2 pAb (A1215), anti-MTHFD2 mAb (A22653), and anti-β-Tubulin mAb (A12289) were purchased from ABclonal Technology. HRP conjugated goat anti-mouse IgG (BA1051) and anti-rabbit IgG (BA1055) were purchased from Boster. All commercial antibodies were used at concentrations recommended by the manufacturers.

RNA extraction and quantitative reverse transcription-PCR (qRT-PCR)

Total RNA was extracted from treated samples using TRIzol reagent (Magen), according to the manufacturer’s instructions. Subsequently, 1 µg of RNA was used to synthesize cDNA with the ABScript II cDNA First-Strand Synthesis Kit (ABclonal Technology). Quantitative real-time PCR was performed using the 2 × University SYBR Green Fast qPCR Mix (ABclonal Technology) and a ViiA 7 Real-time PCR System (Applied Biosystems). The relative expression levels of the targeted mRNA were normalized to that of endogenous control β-actin within each sample, employing the 2^−ΔΔCt method. To determine the RNA copies of JEV in mouse brains, Taqman qRT-PCR was performed with the primers JEV-E-F, JEV-E-R, and the corresponding probe. A standard curve was generated with a 10-fold serial dilution of the plasmid encoding the JEV E gene. All primers and probes utilized are listed in S1 Table.

Western blotting

Cells were lysed in RIPA buffer (Beyotime, P0013) supplemented with protease inhibitors (TargetMOI, C0001). Protein concentrations were determined using the Enhanced BCA Protein Assay Kit (Beyotime, P0010S). Equal amounts of protein were separated by SDS-PAGE and transferred onto a polyvinylidene fluoride (PVDF) membrane (Millipore) using a Mini Trans-Blot Cell (Bio-Rad). The membranes were blocked by incubation in TBS-T containing 2% BSA for 2 h, followed by incubation with the relevant primary antibodies at 4°C overnight. After washing three times with TBS-T, the membranes were then incubated with appropriate secondary antibodies at room temperature for 45 min and visualized using enhanced chemiluminescent (ECL) reagent (Tanon, China).

Gene editing

All sgRNA sequences for gene editing were designed using CRISPOR [62]. The sgRNA oligos were annealed and subsequently cloned into the sgRNA/Cas9 expression vector (LentiCRISPRv2). A total of 800 ng of the LentiCRISPRv2 vector, 800 ng of the packaging plasmid pSPAX2, and 400 ng of the pMD2.G were co-transfected into HEK 293T cells using FuGENE HD Transfection Reagent (Promega, E2311), following the manufacturer's protocol. At 48 h post-transfection, viral supernatants were harvested and added to 4 × 10⁵ N2a cells. For Tkt gene silencing in N2a cells, the transduced cells were treated with 1.5 μg/ml puromycin for 7 days. To generate Mthfd2 and G6pdx knockout (KO) N2a cell lines, puromycin selection (1.5 μg/ml) was performed followed by limiting dilution. Single-cell clones were expanded in 96-well plates for 7 days. The corresponding protein expression after gene editing was detected by immunoblotting. The sequences of the sgRNAs used in this study were presented in S2 Table.

Untargeted metabolomics of mouse brain tissues

For untargeted metabolomics with targeted MRM-based quantification, six-week-old female C57BL/6 mice were intracranially injected with 200 PFU of JEV P3 strain or equal volume DMEM. At 5 days post-infection, the mouse brains were harvested and then subjected to metabolic analysis by using LC-MS. Briefly, each mouse brain was multi-point sampled and accurately weighed to 20 mg and then homogenized in ice-cold 80% LC-MS-grade methanol in water and vortexed for 5 minutes (min). Insoluble material was pelleted by centrifugation at 12000 × g for 10 min. The metabolite extract from the supernatants was subsequently dried down under nitrogen gas for LC-MS analysis.

Chromatographic separation was performed on a Waters ACQUITY UPLC HSS T3 C18 column (1.8 μm, 2.1 mm × 100 mm) using an ExionLC AD UPLC system coupled to a TripleTOF 6600 mass spectrometer (AB SCIEX). The mobile phases consisted of water containing 0.1% formic acid and acetonitrile containing 0.1% formic acid. Mass spectra were acquired in information-dependent acquisition (IDA) mode under both positive and negative electrospray ionization conditions. Qualitative analysis was performed based on retention time (RT), precursor ion information, and MS/MS fragmentation spectra of the detected metabolites. Metabolite identification was achieved using a self-built database together with public databases, including HMDB, KEGG, MoNA, and MassBank. For quantitative analysis, multiple reaction monitoring (MRM) was performed using a QTRAP LC-MS/MS system (AB SCIEX) operating in triple quadrupole mode. QC samples prepared by pooling all sample extracts were injected periodically throughout the analytical run to monitor instrument stability. Chromatographic peak areas were integrated using MultiQuant software (AB SCIEX) and normalized to sample weight. The normalized metabolite abundance data are provided in S3 Table.

Targeted metabolomics of central carbon metabolism in N2a cells

N2a cells, cultured in 10 cm dishes for intracellular metabolite extraction, were washed twice with ice-cold PBS and frozen in liquid nitrogen to halt metabolic enzyme activity. The intracellular metabolites were then extracted with ice-cold 80% methanol. The extracts were collected by vortexing and shaking for 15 min and centrifuging at 12,000 rpm at 4°C for 10 min. Metabolites were detected by MetWare (Wuhan, China) using the AB Sciex QTRAP 6500 LC-MS/MS platform.

Chromatographic separation was performed on a Waters ACQUITY H-Class UPLC system with an ACQUITY UPLC BEH Amide column (1.7 μm, 2.1 × 100 mm). Mobile phase A was water containing 10 mM ammonium acetate and 0.3% ammonium hydroxide; mobile phase B was 90% acetonitrile. The gradient was: 0–1.2 min, 95% B; 8 min, 70% B; 9–11 min, 50% B; 11.1–15 min, 95% B. Flow rate was 0.4 mL/min, column temperature 40°C, injection volume 2 μL. Detection was performed on an AB Sciex QTRAP 6500 + mass spectrometer with an ESI Turbo Ion-Spray source in both positive and negative modes. Source parameters: temperature 550°C; ion spray voltage 5500 V (positive) / −4500 V (negative); curtain gas 35 psi; GS1/GS2 50 psi. Data were acquired in MRM mode with optimized declustering potential and collision energy for each transition. Analyst 1.6.3 and MultiQuant 3.0.3 software were used for data acquisition and peak integration.

Metabolites were annotated based on retention time and MRM transitions using authentic standards and the MetWare database. Quantification was performed using external standard calibration curves. QC samples, prepared by pooling equal volumes of all sample extracts, were injected every ten runs. Metabolite concentrations were normalized to sample protein content. The detected metabolites spanning central carbon and nucleotide metabolism are listed in S4 Table.

[U-¹³C] Glucose isotope tracing

N2a cells were seeded in 6 cm dishes at a density of 1.5 × 10⁶ cells/dish. Following 18 h of cell adherence, the cells were infected with JEV at a multiplicity of infection (MOI) of 5 for 1.5 h. After three washes with PBS, the cells were subsequently cultured in glucose-free DMEM medium (Gibco) supplemented with 25 mM [U-¹³C] glucose for 24 h. Then, cells were gently washed three times with PBS and lysed with pre-chilled 80% LC-MS grade methanol at -80 °C for 2 h. The metabolite-containing supernatant, collected after centrifugation at 14,000 × g for 20 min at 4 °C, was dried under a gentle stream of nitrogen at room temperature. The incorporation of ¹³C into CCM was determined using Ultra High-Performance Liquid Chromatography -High-Resolution Mass Spectrometry (UHPLC-HRMS). Chromatographic separation was performed on an Ultimate 3000 UHPLC system (Thermo Fisher, USA) equipped with a Waters BEH C18 column (2.1mm × 100 mm, 1.7 μm). Simultaneously, the mass spectrometry was conducted using a Q Exactive Hybrid Quadrupole-Orbitrap Mass Spectrometer (Thermo Fisher, USA). Metabolite levels in each sample were normalized to their corresponding protein concentrations. The data from the [U-¹³C] Glucose metabolic flux analysis have been deposited in S5 Table.

Analysis of bulk RNA-sequencing and single-cell RNA-sequencing data

The bulk RNA-sequencing and single-cell RNA-sequencing (scRNA-seq) data used in this study were sourced from our previous research. The datasets generated by Li et al. [27] and Yang et al. [34] are publicly accessible in the Gene Expression Omnibus database under accession numbers GSE94789 and GSE237915, respectively. For our analysis, we retrieved the expression data directly from the original publication and the GEO repository. Subsequently, the expression levels of metabolic enzyme genes were plotted in a heatmap created using the ComplexHeatmap R package (version 2.24.1).

Inhibitor treatment in vitro

N2a cells were infected with JEV at a MOI of 5 in serum-free medium. After 1.5 h for virus absorption, the cell supernatant was replaced with fresh DMEM containing 2% FBS. The inhibitor at an appropriate dosage, determined by a CCK-8 cytotoxicity assay with SuperKine Maximum Sensitivity Cell Counting Kit-8 (Abkkine, BMU106), was added to the maintenance medium of the cell culture. Supernatants and cells were collected at specific time points. The viral titer and RNA abundance were determined using plaque assays and qRT-PCR, respectively.

Hematoxylin-eosin and immunohistochemistry staining

The harvested mouse brains were fixed overnight in 4% paraformaldehyde (PFA) at 4 °C. A minimum of three biological replicates per experimental group were paraffin-embedded and sectioned to a thickness of 5 μm. The pathological changes in the tissues were evaluated using hematoxylin and eosin (H&E) staining, following standard protocols.

For the detection of JEV E protein expression and microglial activation via immunohistochemistry (IHC) staining, sections were incubated with 3% hydrogen peroxide (H₂O₂) for 30 min to eliminate endogenous peroxidase activity. Subsequently, heat-mediated epitope recovery was performed in citrate buffer (pH 6.0) at 96°C for 30 min. The sections were then blocked with 5% bovine serum albumin (BSA) for 1 h at room temperature, followed by an overnight incubation at 4°C with a mouse anti-IBA-1 mAb (Servicebio, GB15105) or anti-JEV-E mAb (prepared in our lab). After washing with PBS-T, the sections were incubated with horseradish peroxidase-conjugated sheep anti-mouse IgG secondary antibody (Servicebio, GB23301) for 45 min. Finally, signal visualization was achieved using a 3,3’-diaminobenzidine substrate with hematoxylin counterstaining.

All stained sections were digitally scanned with a Leica Apero CS2 slide scanning system for high-resolution analysis of neuropathological features.

Ag2037 administration in mice

Six-week-old female C57BL/6 mice were randomly assigned to four groups: group 1 consisted of mice injected intravenously with 100 µL DMEM, followed by intravenous injection with 100 µL of 10% DMSO and 90% (20% SBE-β-CD) in saline at 3 and 6 days post-mock infection (Mock+Vehicle, n = 20); group 2 included mock-infected mice i.v. injected with 15 mg/kg of Ag2037 diluted in 100 µL of 10% DMSO and 90% (20% SBE-β-CD) in saline at 3 and 6 days post-mock infection (Mock+Ag2037, n = 20); group 3 received an intravenous injection of 10⁵ PFU of the JEV P3 strain in 100 μL of DMEM, followed by intravenous injection of the vehicle at 3 and 6 days post-infection (dpi) (JEV + Vehicle, n = 20); group 4 consisted of mice injected with 10⁵ PFU of JEV, followed by treatment with 15 mg/kg of Ag2037 (JEV + Ag2037, n = 20). Mice were monitored daily for 21 days post-infection for changes in weight, symptoms and mortality.

Statistical analysis

All experiments in this study were conducted at least three times under similar conditions. Data analysis was performed using GraphPad Prism, version 9 (GraphPad Software, San Diego, USA). Results are presented as the mean ± standard error of the mean (SEM). Statistical differences between experimental groups were performed using two-tailed Student’s t-test, one-way or two-way ANOVA. For all statistical significance indications in this manuscript, ****, p < 0.0001; ***, p < 0.001; **, p < 0.01; *, p < 0.05, and ns, no significance.