Bamboo mosaic virus-induced metabolic reprogramming engages mitochondrial function to regulate redox homeostasis and defense responses in Nicotiana benthamiana

Abstract

Plants possess a remarkable capacity to reprogram their metabolism in response to pathogen attacks. However, how virus-induced metabolic shifts intersect with redox dynamics and defense signaling remains incompletely understood. In this study, we leveraged a multifaceted omics approach to investigate the metabolic shifts induced by Bamboo mosaic virus (BaMV), a positive-sense single-stranded RNA virus, in Nicotiana benthamiana—a model host widely used for dissecting plant–virus interactions due to its high susceptibility and amenability to genetic manipulation. Metabolic profiling revealed the accumulation of hexose phosphates and Krebs cycle intermediates following BaMV infection, while fluxomic analysis uncovered an orchestrated redirection of carbon flux toward glycolysis and the Krebs cycle. Proteomic data further highlighted a concerted upregulation of mitochondrial enzymes, with three mitochondrial proteins showing markedly increased accumulation in BaMV-infected tissues. Together, these integrated omics analyses indicate that BaMV infection induces a metabolic shift toward energy-generating pathways, possibly to meet the elevated metabolic demands of infection. Notably, functional analysis revealed that silencing mitochondrial NAD⁺-dependent malic enzyme 1 significantly enhanced BaMV accumulation, accompanied by alterations in cytoplasmic NADH-to-NAD⁺ ratio and changes in the landscape of defense gene expression. Collectively, our findings underscore the pivotal role of mitochondrial metabolism in governing cytoplasmic redox balance and finely tuning defense responses during BaMV infection.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12870-025-07996-4.

Introduction

In the intricate arms race between plants and viruses, both organisms have evolved sophisticated strategies to gain an advantage. Viruses, as obligate intracellular parasites, hijack host cellular machinery to ensure their replication, movement, and survival. In turn, plants have evolved multilayered immune responses, ranging from RNA silencing to hormone-mediated defense signaling, aimed at recognizing and counteracting viral invasion [1–7]. This dynamic interplay shapes the outcome of infection and often centers around the host's ability to remodel cellular physiology in response to viral cues.

Viruses extensively reprogram host metabolism to create a cellular environment conducive to their replication, movement, and persistence. During infection, they manipulate the host’s metabolic network not only to secure energy and building blocks for replication but also to subvert defense signaling pathways that rely on tightly regulated redox and metabolic homeostasis [8–10]. A growing body of evidence suggests that this virus-driven metabolic rewiring is not a passive consequence of infection but a deliberate strategy that shapes the outcome of host–virus interactions [8–13]. Upon viral invasion, host primary metabolism—including sugar and amino acid metabolism, redox balance, and organelle function—is extensively altered. For instance, many plant viruses cause significant shifts in carbohydrate metabolism, leading to the accumulation or depletion of sugars and sugar-phosphates that can influence defense responses [14–16]. Similarly, amino acid pools are reshaped during infection, potentially affecting the synthesis of defense-related compounds or viral components [8, 17]. Viruses such as Tobacco rattle virus (TRV) benefit from elevated amino acid levels, yet specific metabolic perturbations—like the loss of DARK INDUCIBLE4—can restrict viral replication, illustrating a tight coupling between metabolic flux and susceptibility [18].

Of particular interest is the impact of viral infection on the host redox system. Viruses target key redox-active organelles—mitochondria and chloroplasts—whose metabolic activities control reactive oxygen species (ROS) production and redox signaling. Mitochondrial enzymes, including components of the Krebs cycle and the electron transport chain, are known to contribute to immune signaling by modulating ROS generation and salicylic acid (SA) pathways [19, 20]. Simultaneously, chloroplast impairment caused by viral infection disrupts photosynthetic electron flow and ROS-mediated signaling [21–23], both of which are crucial for immune responses.

Viruses thus exploit the metabolic and redox plasticity of host cells, while the host simultaneously attempts to reclaim control over these pathways to mount an effective defense [24–26]. Central redox mediators such as glutathione (GSH/GSSG) and NAD(H) are emerging as critical nodes at the interface of metabolism and immunity [24, 27–29]. These redox couples regulate not only antioxidant responses but also the activity of redox-sensitive immune regulators such as nonexpressor of pathogenesis-related genes 1 (NPR1), which integrates oxidative cues into transcriptional activation of defense genes [30, 31]. Although the immunomodulatory role of NAD has been recognized in bacterial and fungal contexts [24], its specific function during viral infection have not been investigated yet.

Bamboo mosaic virus (BaMV) is a well-characterized positive-sense, single-stranded RNA virus of the Potexvirus genus. Its genome encodes five open reading frames [32] responsible for replication [33–36], movement [37–39] and encapsidation [40–42]. To complete its infection cycle, BaMV exploits numerous host factors, including metabolic enzymes that play either proviral or antiviral roles. For instance, BaMV hijacks chloroplast phosphoglycerate kinase (PGK) to promote viral RNA targeting to chloroplasts, thereby facilitating viral accumulation, while glyceraldehyde 3-phosphate dehydrogenase (GAPDH) interferes with BaMV replication by directly binding to viral RNA [43–46]. Successful BaMV accumulation is tightly linked to chloroplasts and mitochondria—organelles that serve as central hubs for plant metabolism. By hijacking cytosolic enolase and mitochondrial voltage-dependent anion channel (VDAC), BaMV assembles a metabolon-like replication complex that bridges chloroplasts and mitochondria, potentially enabling access to localized metabolic and energy resources required for efficient viral replication [44, 47, 48].

Although numerous studies have revealed the association between plant metabolic processes and BaMV accumulation, most have focused on how BaMV exploits host factors to facilitate its replication. How virus-induced metabolic rewiring intersects with redox dynamics and defense signaling remains insufficiently investigated. In this study, we explored the potential contribution of BaMV-induced metabolic reprogramming to redox modulation and defense-associated responses in N. benthamiana plants. By integrating metabolic profiling, fluxomics, and proteomics, we demonstrated that BaMV infection reprograms plant metabolism by re-directing metabolic flux toward glycolysis, Krebs cycle and amino acid biosynthesis. Notably, we observed that mitochondrial NAD⁺-dependent malic enzyme 1 (NAD-ME1) is involved in cytoplasmic redox homeostasis and defense signaling. Our findings highlight the central role of mitochondrial metabolism in orchestrating the defense response, shedding light on how the balance between host metabolic fluxes and redox states can influence the outcome of viral infections.

Materials and methods

Plant material and growth conditions

The N. benthamiana plants were kindly provided by Dr. Yau-Heiu Hsu (Graduate Institute of Biotechnology, National Chung Hsing University, Taichung, Taiwan). Plants were grown in a climate chamber equipped with regular white light tubes. The light intensity was set as 120 μmol photons m⁻² s⁻¹ with an 8-h dark/16-h light regime, and the temperature was set as 28 °C. The fourth leaves of 21-day-old plants were used for further analyses.

Construct generation

To monitor the subcellular localization of NAD-ME1, the construct, pBIN61-NAD-ME1-eGFP-HA, was generated. The full-length coding region of NAD-ME1 was amplified from the cDNA library of N. benthamiana plants. The NAD-ME1 amplicons were digested with XhoI, and ligated into the pBIN61-eGFP-HA vector linearized with XhoI and PmlI. To generate the pBIN61-mito-mCherry clone for mitochondrial visualization, the signal peptide of an ATPase gene [49] was amplified from the cDNA library of Arabidopsis thaliana. The amplicons were digested with XbaI and BamHI and ligated into the pBIN61-3HA-mCherry vector linearized with the same enzymes. The construct, pBIN61-NbNPR1-eGFP-HA, was generated to test NPR1 protein stability. The full-length coding region of NbNPR1 was amplified from the cDNA library of N. benthamiana plants. The NbNPR1 amplicons were digested with XhoI and ligated to the pBIN61-eGFP-HA vector linearized with XhoI and PmlI. For gene silencing assays, the clones, pANDA-NAD-ME1, pANDA-mtLPD1, pANDA-COX6b-1 and pANDA-NbNPR1, were generated. Fragments of approximately 200- 300 bp of the above genes were amplified from the cDNA library of N. benthamiana and cloned into the pCR™8/GW/TOPO™ vector (Invitrogen™) to generate the subclones, pCR8-NAD-ME1, pCR8-mtLPD1, pCR8-COX6b-1 and pCR8-NbNPR1. These subclones were further incorporated into the destination vector pANDA [50], using Gateway™ LR Clonase™ (Invitrogen™). All primer pairs are listed in Table S1.

Agroinfiltration

Agrobacterium tumefaciens strain C58C1 and EHA105 was used for transient gene expression and silencing in this study. Bacterial strains were revived from glycerol stocks and streaked onto LB agar plates containing appropriate antibiotics. Following overnight incubation at 28 °C, cells were harvested by centrifugation at 5,000 × g for 5 min at room temperature and resuspended in MMA buffer (10 mM MgCl₂, 10 mM MES-KOH, 200 μM acetosyringone). Bacterial suspensions were adjusted to the desired optical density (OD₆₀₀ = 0.1) and infiltrated into the fourth true leaves of 4-week-old N. benthamiana plants using a 1 mL needleless syringe.

RNA extraction and RT-qPCR

The N. benthamiana leaves were frozen in liquid nitrogen and ground into a fine powder using a mortar and pestle. Approximately 100 mg of leaf powder was used for RNA extraction by utilizing TRIzol™ Reagent (Invitrogen™) according to the manufacture’s instruction. The RNA concentration was quantified by using a NanoDrop spectrometer (ND-2000, Thermo Fisher Scientific). One microgram of total RNA was reverse-transcribed into cDNA using TOOLSQuant II Fast RT Kit (KRT-BA06, TOOLS). The synthesized cDNA samples were diluted 10 times with nuclease-free water, and 4.5 μL of diluted cDNA samples was mixed with 0.25 μL each of 10 μM forward and reverse primers and 5 μL of Power SYBR™ Green PCR Master Mix (Applied Biosystems™). The qPCR reaction was conducted in the thermocycler (12 k Flex, QuantStudio™ 3, Applied Biosystems™) with the following cycling conditions: 95 °C for 3 min, followed by 40 cycles of 95 °C for 15 s and 60 °C for 30 s. Gene expression levels were calculated using the 2^−(ΔΔCt) method [51, 52], with elongation factor 1α (EF1α; Niben101Scf08618g01012.1) as the reference gene for normalization [53]. All qPCR primer pairs are listed in Table S1.

Starch and soluble sugar quantification

Twenty milligrams of pulverized leaf tissue were mixed with 500 μL of 80% (v/v) ethanol and incubated at 90 °C for 30 min. The mixture was centrifuged at 20,000 X g for 15 min at ambient temperature. The supernatant was transferred into a microtube, while the pellet was resuspended in 250 μL of 50% (v/v) ethanol and incubated at 90 °C for 30 min. After centrifugation under the same conditions, the supernatant was transferred into the previous microtube and used for the quantification of soluble sugars, while the pellet was dehydrated and used for the quantification of starch. The detailed detection method was described previously [54].

Metabolomic analysis

One hundred milligrams of pulverized leaf tissue was mixed with 500 μL of methanol containing 5 ppm of succinate-2,2,3,3-d4 as internal standard. The mixture was incubated at 4 °C for 10 min followed by the addition of 200 μL of chloroform and 500 μL of double-distilled water. After vigorous vortex at 4 °C for 5 min, the mixture was centrifuged at 20,000 X g for 15 min at 4 °C. The aqueous phase was transferred into a microtube followed by dehydration using a vacuum centrifugal concentrator (ScanSpeed MaxiVac, ScanVac). The pellet was resuspended in 100 μL of 50% methanol.

For the targeted metabolomic analysis, the extract was separated at 30 °C using an Agilent InfinityLab Poroshell 120 Hydrophilic Interaction Liquid Chromatography (HILIC) column (2.7 μm, 2.1 × 100 mm) connected to a Thermo Scientific Vanquish Horizon Ultra-High Performance Liquid Chromatography (UHPLC) System and subjected to a Thermo Scientific Orbitrap Fusion Lumos Tribrid Mass Spectrometer. The mobile phases were composed of eluent A (50% [v/v] acetonitrile) and eluent B (90% [v/v] acetonitrile). Both eluents were supplemented with 20 mM ammonium acetate (pH 9.0). The flow rate was set as 500 μL/min. The mass spectrometer was equipped with an electrospray ionization (ESI) source operating in negative full-scan ion mode to collect signals ranging from 70 to 1000 (m/z). The chromatogram acquisition, detection of mass spectral peaks, and their waveform processing were performed using Xcalibur™ software (Thermo Fisher Scientific).

For the metabolic flux analysis, the metabolite extract was separated at 40 °C using an ACQUITY UPLC BEH amide column (1.7 μm, 2.1 × 100 mm, Waters Corp., Milford, MA, USA) connected to an Agilent 1290 II Infinity Ultra-High-Performance Liquid Chromatography system (Agilent Technologies, Palo Alto, CA, USA) and subjected to an Agilent 6545XT quadrupole time-of-flight (Q-TOF) mass spectrometer (Agilent Technologies, Palo Alto, CA, USA). The mobile phases were composed of eluent A (deionized water) and eluent B (90% [v/v] acetonitrile). Both eluents were supplemented with 15 mM ammonium acetate and 0.3% NH₃.H₂O. The flow rate was set as 300 uL/min. The mass spectrometer was equipped with an Agilent Dual AJS ESI source operating in positive and negative full-scan ion mode to collect signals ranging from 60 to 1500 (m/z). The chromatogram acquisition, detection of mass spectral peaks, and their waveform processing were performed using Agilent Qualitative Analysis 10.0 (Agilent, USA). Isotopologue extraction and the correction of natural isotope abundance was performed using Agilent Profinder 10.0 (Agilent, USA). The mass tolerance and retention time tolerance were set to ± 10 ppm and ± 0.5 min, respectively.

Protein extraction and proteomic analysis

Approximate 25 mg of leaf powder was mixed with 250 μL of extraction buffer (20 mM Tris–HCl, pH 7.5, 5 mM EDTA, 10 mM NaCl, 8 M Urea, 10 mM DTT) and incubated for 10 min at 37 °C with mild shaking. The extract was centrifuged at 20,000 X g for 10 min at 4 °C to remove leaf debris. Two hundred microliter of supernatant was mixed with 4 volumes of absolute acetone and centrifuged at 20,000 X g for 10 min at 4 °C to precipitate protein. The protein pellet was washed with 80% acetone and dehydrated under a chemical hood, then resuspended in 50 μL of extraction buffer. The protein concentration was determined using Pierce™ 660 nm Protein Assay Reagent (Thermo Scientific™).

Protein digestion in the S-Trap microcolumn was performed according to the protocol of S-Trap-IMAC with minor modifications [55, 56]. In brief, 10 µg of protein was reconstituted in the buffer containing 4 M urea and 5% (v/v) sodium dodecyl sulfate. The protein sample was reduced and alkylated using 10 mM Tris(2-carboxyethyl)phosphine (TCEP) and 40 mM chloroacetamide (CAA) at 45 °C for 15 min, respectively. A final concentration of 5.5% (v/v) phosphoric acid (PA) was applied to the protein sample followed by mixing with a six-fold volume of binding buffer containing 90% (v/v) methanol in 100 mM of triethylammonium bicarbonate (TEAB). After moderate vortexing, the solution was loaded into an S-Trap microcolumn and centrifuged at 4,000 X g for 1 min. The column was washed with 350 µL binding buffer three times. Finally, 20 µL of digestion solution (0.2-unit Lys-C and 200 ng trypsin in 50 mM TEAB) was added to the column and incubated at 47 °C for 2 h. Each digested peptide was eluted using 40 µL of three buffers consecutively: (1) 50 mM TEAB, (2) 0.2% (v/v) formic acid in H₂O, (3) 50% (v/v) acetonitrile. Eluted peptides were collected, dehydrated under vacuum and resuspended in 50 µL H₂O. Five micrograms of peptides were desalted using Evotip and dehydrated under vacuum.

Peptide samples were loaded onto Evotips Pure and eluted into an EV1137 performance column (15 cm × 150 µm ID, 1.5 µm, Evosep Biosystems) on the Evosep One LC system (Evosep Biosystems) connected to a timsTOF HT mass spectrometer (Bruker Daltonics). The mass spectrometer was set to PASEF scan mode for data-dependent acquisition. All spectra were acquired over an m/z range of 100 to 1,700 with 10 PASEF ramps. The TIMS ranges were initially set from 0.6 to 1.6 1/K0 [V·s/cm²] with settings of 100 ms ramp and accumulation time (100% duty cycle) and a ramp rate of 9.43 Hz, resulting in a total cycle time of 1.17 s. Singly charged precursors were excluded based on their position in the m/z-IM plane using a polygon shape, and linear precursor repetitions were set at a target intensity of 20,000 with a threshold of 2,500. Active exclusion was enabled with a 0.4 min release time. The collision energy remained at default, with a base of 1.6 1/K0 [V·s/cm²] set at 59 eV and 0.6 1/K0 [V·s/cm²] set at 20 eV. Isolation widths were set to 2 m/z at < 700 m/z and 3 m/z at > 800 m/z. The TIMS ranges were initially set from 0.6 to 1.6 1/K0 [V·s/cm²].

These raw files were searched with SpectroMine™ (version SpectroMine 4.2.230428.52329, Biognosys) against the established proteome database of N. benthamiana [57] using the following settings: fixed modification: carbamidomethyl (C); variable modifications: acetyl (protein N-term), oxidation (M); enzyme: trypsin and LysC with up to two missed cleavages. Mass tolerances were automatically determined by SpectroMine, and other settings were set to default parameters. Identified results were filtered by a 1% FDR at the precursor, peptide, and protein levels.

SDS-PAGE and immunoblotting

Protein extraction followed the approach mentioned above. For non-reducing conditions, DTT was omitted in the extraction buffer. The protein samples were assayed by SDS-PAGE using 8% (v/v) acrylamide gel and immunoblotting. The signals of actin and NPR1-GFP were detected using commercial actin antibody (A0408, Sigma-Aldrich) and GFP-tag antibody (CPA9023, Cohesion Bioscience), respectively.

Viral titer detection

BaMV RNA levels were quantified by RT-qPCR as described above using primer pairs targeting the coat protein (CP) coding region of the BaMV genome (Table). To detect BaMV CP accumulation, total protein extracts were analyzed by SDS-PAGE followed by immunoblotting using a home-made polyclonal antibody raised against the BaMV CP. Actin was detected using an anti-actin antibody (A0480, Sigma-Aldrich) and served as a loading control.

Confocal laser scanning microscopy

Leaf discs of N. benthamiana plants were placed onto a glass slide together with a few droplets of double distilled water and covered with a coverslip. The fluorescence signals were monitored using a confocal microscope (Olympus FV3000). The excitation wavelengths were set at 488 nm for eGFP, 561 nm for mCherry and 405 nm for tSapphire (NADH biosensor). The emitted signals of eGFP, mCherry and tSapphire were collected at 510 ± 10 nm, 610 ± 10 mn and 515 ± 10 nm, respectively.

Accession numbers

Sequence information in this article can be retrieved in the data library of Sol Genomic Network and QUT under the below accession numbers: NAD-ME1: Nbv5.1tr6319150; mtLPD1: Niben101Scf01334g06004.1; COX6b-1: Nbv5.1tr6218676; NPR1: Niben101Scf14780g01001.1. The curated sequences are listed in Supplementary Table.

Results

N. benthamiana accumulate high levels of hexose phosphates and Krebs cycle intermediates in response to BaMV infection

To determine the optimal timepoint for downstream omics analyses, we first performed a time-course assay following BaMV inoculation on N. benthamiana leaves. As shown in Fig. S1a, mosaic symptoms became visible at 3 days post-inoculation (dpi) and intensified thereafter. Immunoblot analysis revealed that BaMV CP was detectable as early as 1 dpi, with a marked increase at 2 and 3 dpi. At a later time point beyond 3 dpi, CP levels remained elevated but did not show a further pronounced increase (Fig. S1b). These results indicate that by 3 dpi, BaMV had accumulated to high levels in the inoculated leaf tissue, reflecting a robust and widespread infection. Given the strong CP accumulation at 3 dpi and the lack of severe mosaic symptoms, this time point was selected for downstream omics analyses to capture early virus–host interactions in infected leaf tissue.

To assess the impact of BaMV infection on host metabolic processes, we used liquid chromatography coupled with tandem mass spectrometry (LC–MS/MS) to quantify primary metabolites in BaMV-infected leaves. The results revealed significant increases in the levels of Glc-6-phosphate (G6P), Fru-6-phosphate (F6P) and Glc-1-phosphate (G1P) at 3 dpi (Fig. 1 and Table S2). Following BaMV infection, the abundances of several Krebs cycle intermediates, including citrate, α-ketoglutarate (α-KG), and succinate, were significantly increased, while the levels of fumarate and malate were moderately decreased (Fig. 1 and Table S2). Given the close link between glycolytic and Krebs cycle intermediates and amino acid biosynthesis, we examined the levels of individual amino acids in BaMV-infected leaves. The infection induced changes in amino acid abundance, with notable increases in glutamate (Glu) and arginine (Arg), while valine (Val), aspartate (Asp), and threonine (Thr) levels were moderately reduced (Fig. 1 and Table S2). Other amino acids showed only minor changes in response to BaMV infection.

Fig. 1.

The changes of glycolytic, Krebs cycle intermediates and amino acids in BaMV-infected leaves. Leaves of N. benthamiana plants were inoculated with 1 μg BaMV virions and harvested at 3 day-post-inoculation. Metabolite levels were measured by LC–MS/MS, and heatmaps show the log₂ fold changes of metabolite levels in comparison to mock-infected leaves. Results are from 3 independent experiments, each with 4 to 5 biological replicates. Statistical analysis was conducted using unpaired t test with Holm-Šídák method (*P < 0.05). The symbol “Sig.” indicated significant changes, while “Nonsig.” Indicated non-significant changes

To determine whether the changes in glycolytic and Krebs cycle metabolites were associated with carbohydrate metabolism, we further analyzed the accumulation of starch and major soluble sugars. In BaMV-infected leaves, the starch level was comparable to the mock level, while the levels of Glc, Suc and maltose were significantly increased at 3 dpi (Table S3), potentially reflecting altered carbon allocation. Thus, the increase in carbon pool in BaMV-infected leaves can further facilitate glycolysis and Krebs cycle.

BaMV infection redirects metabolic flux toward glycolysis and the Krebs cycle and facilitates amino acid biosynthesis

Given that steady-state metabolite levels represent a balance between biosynthesis and utilization rather than the directional metabolic activity [58], we performed a metabolic flux analysis using ¹³C-labeled glucose to directly assess the dynamic behavior of glycolytic and mitochondrial pathways during BaMV infection. At 3 dpi, we infiltrated BaMV-infected leaves with ¹³C-labeled Glc and analyzed ¹³C incorporation into metabolites using Liquid chromatography/quadrupole time-of-flight mass spectrometry (LC/Q-TOFMS). The ¹³C incorporation into glycolytic metabolites (G6P, F6P, G1P and pyruvate), as well as into Krebs cycle intermediates (citrate, α-KG, succinate and malate), were significantly increased in BaMV-infected leaves compared to mock-treated leaves (Fig. 2a and Table S4). Notably, despite this elevated flux, steady-state metabolite profiling revealed an increased accumulation of succinate along with reductions in fumarate, malate, and several downstream amino acid derivatives (e.g., Asp, Met) (Fig. 1), a pattern that may suggest a bottleneck at the succinate dehydrogenase step. However, the increased ¹³C labeling of both succinate and malate argues against a flux-limiting block and instead points to accelerated turnover or rerouting of these intermediates into biosynthetic or redox-related pathways. In addition, the ¹³C incorporation into many amino acids including Ser, Gly, Phe, Ala, Val, Glu, Gln and Thr were also elevated upon BaMV infection (Fig. 2b and Table S4). These results indicate that BaMV infection reprograms metabolic flux to enhance glycolysis, the Krebs cycle, and associated amino acid biosynthesis.

Fig. 2.

¹³C incorporation into glycolytic, Krebs cycle intermediates and amino acids in leaves of mock- and BaMV-infected N. benthamiana plants. Leaves of N. benthamiana plants were inoculated with 1 μg BaMV virions. At 3 day-post-inoculation, the infected leaves were infiltrated with 15 mM of ¹³C-labeled glucose for 60 min, and then harvested for metabolite extraction. The ¹³C incorporation of each metabolite were analyzed using LC/Q-TOFMS, and the percentage of.¹³C incorporation was calculated as the intensity of labeled metabolites divided by the total intensity (labeled + unlabeled) for a) glycolytic, Krebs cycle metabolites and b) amino acids. Results are the mean with SE from three to five biological replicates. Statistical analysis was conducted using unpaired t test with Holm-Šídák method (*P < 0.05)

It is worth noting that although BaMV has been shown to interact with host enzymes such as pyruvate decarboxylase 1 (Pdc1) and alcohol dehydrogenase 1 (Adh1), and silencing of these genes reduces BaMV accumulation [59], our current data do not support activation of the canonical fermentation pathway. None of the major fermentation intermediates (e.g., acetaldehyde, ethanol) were detected, and lactate levels — another potential endpoint of anaerobic metabolism — remained unchanged with no detectable ¹³C labeling (Table S4). This suggests that BaMV may hijack these enzymes for non-canonical functions, such as scaffolding, spatial organization, or metabolite channeling within replication complexes.

BaMV infection dramatically downregulates photosynthetic proteins while upregulates mitochondrial enzymes

To identify host factors involved in promoting glycolysis and mitochondrial metabolism during BaMV infection, we conducted a time-resolved proteomic analysis in BaMV-infected leaves. We harvested mock- and BaMV-infected leaves for protein extraction and LC–MS/MS analysis at 1, 2, 3 dpi. Using this approach, we identified a total of 2,941 proteins. We considered proteins detected in at least three biological replicates and containing at least two unique peptides as valid targets. Applying this criterion, we selected 1,225 protein targets for further analysis. To obtain well-annotated protein information, we performed a homology search against the Arabidopsis proteome database. The principal component analysis (PCA) plot shows that mock samples (mock_1dpi, mock_2dpi and mock_3dpi) closely clustered to each other, while BaMV-infected samples (BaMV_1dpi, BaMV_2dpi and BaMV_3dpi) progressively shifted away from the mock cluster along the Dim2 axis. The BaMV_3dpi samples occupied the farthest position from the mock group (Fig. a), indicating that BaMV infection induces progressive and substantial changes in the host proteome over time.

To further evaluate the impact of BaMV infection on protein levels, we calculated the fold changes in abundance for each identified protein between BaMV-infected and mock samples. Proteins that exhibited statistically significant changes (p < 0.05) at any time point were retained for downstream analysis. Based on these criteria, we identified 328 proteins that showed statistically significant changes in abundance at one or more time points following BaMV infection (Table S5). To illustrate the temporal distribution of these differentially accumulated proteins, we generated a Venn diagram indicating the specific time point(s) at which each protein showed a significant change. The Venn diagram analysis indicated that the non-overlapping protein sets at 1, 2, and 3 dpi sample sets accounted for 25%, 24%, and 37% of the 328 differentially expressed proteins, respectively. Notably, 3% or 5% of the proteins overlapped between any of two time points, and only four proteins showed significant changes at all three time points (Fig. S2b). These results suggest that plant proteomes exhibit distinct and dynamic changes throughout the course of BaMV infection.

Given that the 3-dpi sample set exhibited the most pronounced proteomic changes, we further analyzed those 156 differentially expressed proteins identified from this time point according to their biological functions (Table S5). We identified 38 proteins associated with photosynthetic activity, 4 with cellular respiration, 9 with amino acid metabolism, 15 with other metabolic processes, 6 with stress response, 7 with redox homeostasis, and 17 with various other cellular functions. Furthermore, 45 proteins were classified under DNA, RNA and protein-related processes, while 15 remained uncharacterized with unknown function. Within the photosynthesis category, the majority of identified proteins showed reduced abundance upon BaMV infection. However, five proteins—6-phosphogluconate dehydrogenase (6PGDH), dihydrolipoyl dehydrogenase 2 (LPD1), oxygen-dependent coproporphyrinogen-III oxidase (LIN2), glutamate–glyoxylate aminotransferase 2 (AOAT2), and aminomethyltransferase—were notably upregulated. A similar trend of decreased protein abundance was observed in the redox homeostasis category (Fig. 3a and Table S5). These results suggest that BaMV infection is associated with a general downregulation of proteins involved in photosynthesis and redox homeostasis, although a subset of proteins were upregulated, potentially reflecting compensatory or stress-responsive adaptations. In addition, the differential accumulation of enzymes, including Glu dehydrogenase (GDH1), Glu decarboxylase (GAD4), Gln synthetase (GSR1 and GSR2), and argininosuccinate synthase (ASS), may account for the accumulation of Glu and Arg in BaMV-infected leaves (Fig. 1 and Table S5). Considering that BaMV infection leads to the upregulation in Krebs cycle (Figs. 1 and 2), we next examined the changes in mitochondrial proteins. In BaMV-infected leaves, three mitochondrial enzymes—cytochrome c oxidase subunit 6b (COX6b-1), mitochondrial lipoamide dehydrogenase 1 (mtLPD1), and NAD-malic enzyme 1 (NAD-ME1)—were increased to 1.6-, 1.9-, and 2.1-fold relative to mock levels, respectively. In contrast, the level of pyruvate dehydrogenase E1 component subunit beta-1 was reduced to approximately half of the mock level (Fig. 3b and Table S5). These results suggest that BaMV infection alters the abundance of key mitochondrial enzymes, potentially reflecting a reconfiguration of mitochondrial metabolism during infection.

Fig. 3.

The proteome landscape in BaMV-infected leaves harvested at 3 day-post-inoculation (dpi). Leaves of N. benthamiana plants were inoculated with 1 μg BaMV virions and harvested at 3 dpi for protein extraction followed by a proteomic analysis. a) Functional categories of proteins showing significant changes in their abundance during BaMV infection. b) Volcano plot indicating significantly increased mitochondrial targets upon BaMV infection. COX6b-1: cytochrome C oxidase 6b; mtLPD1: mitochondrial lipoamide dehydrogenase 1; NAD-ME1: NAD-malic enzyme 1. Results are from 3 to 4 biological replicates. Statistical analysis was conducted using unpaired t test (*P < 0.05)

Silencing mtLPD1 and NAD-ME1 increases BaMV accumulation

To investigate the functional roles of identified mitochondrial components in the response to BaMV infection, we adopted an RNA interference approach to downregulate the expression of COX6b-1, mtLPD1 and NAD-ME1 in N. benthamiana leaves (Miki and Shimamoto, 2004). In the silenced leaves (hpCOX6b-1, hpmtLPD1 and hpNAD-ME1), the transcript levels of COX6b-1, mtLPD1 and NAD-ME1 were reduced to 12%, 28% and 16% of those in control leaves (EV; Fig. 4d-f), respectively, indicating effective gene silencing. When examining BaMV accumulation, silencing COX6b-1 did not affect BaMV accumulation (Fig. 4a). However, silencing mtLPD1 and NAD-ME1 significantly increased BaMV titers to 1.5- and twofold of the levels in control leaves, respectively (Fig. 4b, c). These findings imply that mtLPD1 and NAD-ME1 may participate in host defense processes that help limit BaMV accumulation.

Fig. 4.

Effects of silencing mitochondrial targets on BaMV accumulation. Leaves of N. benthamiana were infiltrated with Agrobacterium harboring an empty vector (EV) or hairpin RNA silencing clones. At 3 day-post-infiltration, leaves were inoculated with 1 μg BaMV virions and harvested at 27-h-post-inoculation. BaMV accumulation and target gene expression were measured by RT-qPCR in silencing plants for a, d) COX6b-1, b, e) mtLPD1 and c, f) NAD-ME1, with EF1α as the reference gene. Results are from 12 to 16 biological replicates. Statistical analysis was conducted using unpaired t test (*P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001)

Silencing NAD-ME1 increases the cytoplasmic NADH-to-NAD⁺ ratio and alters the expression patterns of defense-related genes

Among the three mitochondrial candidates identified, NAD-ME1 exhibited the highest increase in protein abundance during BaMV infection (Table S5), and its silencing caused the strongest enhancement of viral accumulation (Fig. 4c). Therefore, we prioritized NAD-ME1 for further investigation. In Arabidopsis, potato, Crassula argentea, and other plant species, NAD-ME1 has been consistently identified as a mitochondrial enzyme that catalyzes the decarboxylation of malate to produce pyruvate and NADH within the mitochondrial matrix [60–65]. To investigate whether the homolog of NAD-ME1 in N. benthamiana is also present in mitochondria, we fused N. benthamiana NAD-ME1 to an enhanced green fluorescence protein (eGFP) and co-expressed it with a mitochondrial marker (mito-mCherry), in which the mitochondrial targeting signal peptide of an Arabidopsis ATPase was fused to mCherry [49]. Both constructs were transiently expressed in N. benthamiana leaves via Agrobacterium infiltration. The confocal microscopic images showed that the signal of NAD-ME1-eGFP colocalized with the mitochondrial marker mito-mCherry (Fig. S3), indicating that NAD-ME1 is a mitochondrial protein in N. benthamiana.

We next examined whether silencing NAD-ME1 alters the levels of glycolytic and Krebs cycle intermediates during BaMV infection. In NAD-ME1–silenced leaves, the levels of G6P and F6P/G1P were reduced to approximately 75% and 70% of control levels, respectively, while pyruvate levels remained unchanged (Table S6). Among Krebs cycle intermediates, most showed no significant differences; however, succinate levels were decreased to 50% of control in the silenced leaves (Table S6). These results suggest that NAD-ME1 silencing may partially interfere Krebs cycle activity during BaMV infection. Given that mitochondrial metabolism is tightly associated with the balance of NAD(H) redox couple in the cytoplasm [66], we further assess the impact of NAD-ME1 silencing on cytoplasmic NADH homeostasis. We employed a genetically encoded NADH biosensor comprising a tSapphire (tS) fluorescent protein flanked by two NAD(H)-binding domains, fused to mCherry (mC) as an internal control. NADH binding enhances tSapphire fluorescence, while NAD⁺ binding reduces it; thus, the cytosolic NADH-to-NAD⁺ ratio can be inferred by calculating the tSapphire/mCherry fluorescence ratio [67]. In both mock-treated and BaMV-infected leaves, silencing NAD-ME1 (hpNAD-ME1) caused a modest but statistically significant increase in the cytoplasmic NADH-to-NAD⁺ ratio compared to control leaves (EV) at 3 dpi (Fig. 5a, b). While the increase was moderate, the consistent trend across conditions suggests a role for NAD-ME1 in buffering cytosolic redox balance, likely through its influence on Krebs cycle. Notably, although BaMV infection increased the level of NAD-ME1, it did not significantly alter the cytosolic NADH-to-NAD⁺ ratio compared to mock-treated leaves (Fig. 5b). This suggests that BaMV-induced NAD-ME1 upregulation may serve a buffering role in maintaining NAD(H) redox balance under viral stress.Cellular redox balance can affect defense signaling during pathogen infection [24–26]. To investigate whether the redox imbalance resulting from the disruption of NAD-ME1 expression can affect defense responses during viral infection, we examined the expression profile of defense marker genes [68–70]. In BaMV-infected leaves, the transcript levels of pathogenesis-related protein 2b (PR2b), NPR1 and respiratory burst oxidase homolog protein B (RBOHB) were significantly increased to 1.38-, 1.24-, and 1.23-fold of the levels in mock-treated leaves (Fig. 6a, b, f), respectively. In contrast, the levels of lipoxygenase (LOX), plant defensin 1.2 (PDF1.2), ethylene response factor 1 (ERF1) remained comparable to mock-treated leaves (Fig. 6c-e). Interestingly, silencing NAD-ME1 changed the expression pattern of these defense genes. In mock-treated leaves, silencing NAD-ME1 up-regulated the expression of PR2b to 1.47-fold of control levels (Fig. 6a) while down-regulated the expression of NPR1, LOX and ERF1 to 0.73-, 0.60- and 0.66-fold of control levels (Fig. 6b, c, e), respectively. Upon BaMV infection, silencing NAD-ME1 further elevated the levels of PR2b, PDF1.2 and RBOHB to 1.86-, 1.67- and 1.38-fold of control levels (Fig. 6a, d, f), respectively. Notably, NPR1 expression remained suppressed (0.86-fold of control) in NAD-ME1–silenced leaves during infection (Fig. 6b). Although BaMV infection induced the expression of PR2b, NPR1 and RBOHB in BaMV-infected leaves, their corresponding proteins were not detected in our proteomic dataset (Table S5), a discrepancy likely attributable to translational suppression, rapid protein turnover, and/or temporal or spatial expression patterns that escaped detection. Given this limitation, the biological significance of these modest transcript changes remains uncertain in the absence of corresponding protein data. Nonetheless, the consistent transcriptional shifts observed upon NAD-ME1 silencing suggest that perturbation of cytosolic NAD(H) redox balance may influence the expression of defense-related genes during BaMV infection.

Fig. 5.

Effect of silencing NAD-malic enzymes 1 (NAD-ME1) on cytoplasmic NADH-to-NAD ratio. Leaves of N. benthamiana plants were infiltrated with Agrobacterium harboring either empty vector (EV) or NAD-ME1 silencing clones, along with the NADH biosensor and BaMV infectious clone. At 3 day-post-infiltration, leaves were imaged by confocal microscopy. The florescence of tSapphire (tS, NADH binding module) and mCherry (mC, control module) were recorded sequentially. The intensity of tS was divided by the intensity of mC to yield the ratio representing NADH-to-NAD ratio. a) Representative confocal images showing cellular NADH-to-NAD ratios. b) Qualitative analysis of the cellular NADH-to-NAD ratio in relative to the mock samples. Results are from 18 biological replicates. Statistical analysis was conducted using one-way ANOVA with Tukey’s post-hoc test (P < 0.05)

Fig. 6.

Expression profiles of defense genes in NAD-ME1 silencing plants. Leaves of N. benthamiana plants were infiltrated with Agrobacterium harboring hairpin RNA silencing clones. At 3 day-post-infiltration, leaves were inoculated with 1 μg BaMV virions and harvested at 27-hour-post-inoculation. The expression of defense-related genes in a) PR2b, b) NPR1, c) LOX, d) PDF1.2, e) ERF1 and f) RBOHB were measured by RT-qPCR with EF1α as the reference gene. Results are from 12 to 16 biological replicates. Statistical analysis was conducted using one-way ANOVA with Tukey’s post-hoc test (P < 0.05)

Since NPR1 activation involves redox-dependent monomerization, nuclear translocation, and the induction of defense-related gene expression, and is tightly associated with cytoplasmic redox dynamics [30, 31], we examined whether redox imbalance caused by NAD-ME1 silencing affects NPR1 protein homeostasis. In the NAD-ME1-silenced leaves, the levels of exogenously expressed NPR1-GFP oligomers and monomers were dramatically reduced compared to the control leaves, as observed under non-reducing conditions (without DTT; Fig. S4a, b). Under reducing conditions (with DTT; Fig. S4a, b), the total NPR1-GFP abundance was also decreased in the NAD-ME1-silenced leaves. These findings suggest that fluctuation of NAD(H) redox balance may compromise NPR1 stability or interferes with its redox-regulated processing. Although our data do not directly assess other NPR1 activation events such as nuclear localization, degradation or PR gene induction, the observed reduction in NPR1 protein levels implies that NAD-ME1–dependent redox homeostasis plays a role in maintaining cellular NPR1 protein pools. Furthermore, silencing NPR1 significantly increased BaMV accumulation (Fig. S4b). The observed reduction in NPR1 protein levels in NAD-ME1–silenced plants suggest a link between redox imbalance and NPR1 protein homeostasis, which could have downstream consequences for antiviral defense.

Discussion

Viral infection often induces substantial alterations in host metabolism, redirecting cellular resources and modifying the intracellular environment to support viral replication. Simultaneously, these metabolic changes can activate host defense responses. Nevertheless, how virus-induced metabolic rewiring intersects with redox dynamics and defense signaling remain poorly understood. In this study, we employed multi-omics profiling combined with functional validation to identify redox-associated metabolic components that influence viral accumulation and host defense responses in N. benthamiana. Our metabolomic and fluxomic analyses revealed that BaMV infection redirected metabolic flux toward glycolysis, Krebs cycle, and amino acid biosynthesis (Figs. 1 and 2). Proteomic data showed a reduction in the abundance of photosynthetic proteins, accompanied by increased abundance in a subset of mitochondrial proteins, such as COX6b-1, mtLPD1, and NAD-ME1 (Fig. 3), implicating mitochondrial involvement in coordinating virus-induced metabolic reprogramming. Notably, while most photosynthesis-related and redox-regulating proteins were downregulated, several plastid-localized enzymes were markedly induced during BaMV infection. Among them, 6PGDH, a key enzyme of the pentose phosphate pathway (PPP), exhibited the highest fold increase. As the PPP is a major source of NADPH in plastids [71], this strong upregulation of 6PGDH may reflect a compensatory response to virus-induced oxidative stress and suppressed photosynthetic activity. Importantly, silencing of NAD-ME1 resulted in increased cytoplasmic NADH/NAD⁺ ratio, altered defense gene expression, and elevated BaMV accumulation (Figs. 4, 5 and 6). Together, these findings indicate that NAD-ME1 participates in redox homeostasis and defense regulation during BaMV infection, even though it is likely one component of a broader metabolic network.

BaMV-infected leaves exhibited elevated levels of soluble sugars, glycolytic intermediates, and Krebs cycle metabolites (Fig. 1), reflecting increased carbon availability during BaMV infection. Fluxomic analysis revealed a redirection of carbon flux toward glycolysis and the Krebs cycle (Fig. 2), a shift that may represent the host’s efforts to meet increased energy and biosynthetic demands under viral challenge. This is reminiscent of earlier work on cucumber mosaic virus (CMV) in Cucurbita pepo, where Tecsi et al. spatially mapped virus-induced metabolic transitions and demonstrated an increase in glycolysis and mitochondrial respiration at the lesion core while photosynthetic activity declined [72]. While such metabolic changes can benefit viral replication, several lines of evidence suggest they also contribute to defense activation. Invertase-mediated sugar accumulation is associated with enhanced resistance to potato virus Y in transgenic tobacco plants [16], and virus-resistant tomato cultivars accumulate more amino acids than susceptible ones upon the infection of Tomato yellow leaf curl virus [73]. Furthermore, virus-triggered nitrogen reallocation also appears to intersect with antiviral defense. In BaMV-infected leaves, Glu and Arg levels were elevated along with differential expression of related enzymes such as GDH1, GAD4, GSR1/2, and ASS (Fig. 1 and Table S5). Both amino acids are increasingly recognized for their signaling roles in defense [74–77]. Indeed, our results demonstrate that silencing NAD-ME1 alters both cytosolic redox status and the expression of defense-related genes during BaMV infection, suggesting a potential regulatory link between mitochondrial metabolism and defense signaling (Figs. 4, 5 and 6).

Among the mitochondrial enzymes upregulated during BaMV infection, NAD-ME1 was of particular interest due to its role in coordinating mitochondrial carbon flux and cytoplasmic redox status. Silencing NAD-ME1 decreased the levels of hexose phosphates and succinate and increased cytoplasmic NADH/NAD⁺ ratio (Fig. 5 and Table S6), which coincided with elevated BaMV accumulation and altered defense gene expression (Figs. 4 and 6). NAD-ME1 catalyzes the decarboxylation of malate to pyruvate, facilitating NAD⁺ regeneration and supporting mitochondrial respiration [60, 78]. Furthermore, pharmacological inhibition of mitochondrial NADH consumption has been shown to elevate cytoplasmic NADH and NADPH levels, further supporting the direct influence of mitochondrial metabolism on cytosolic redox status [66]. In this context, disruption of NAD-ME1 function likely impairs mitochondrial activity and shifts the cytoplasmic redox environment toward a more reduced state. This altered redox balance may, in turn, influence redox-sensitive defense regulators such as NPR1. Under basal conditions, NPR1 exists as cytosolic oligomers stabilized by disulfide bonds, and reduction by thioredoxins allows its monomerization and subsequent nuclear translocation [30, 31]. The moderately enhanced cytoplasmic reduction observed in NAD-ME1-silenced plants may disrupt this tightly regulated redox cycle, potentially affecting NPR1 protein stability or availability. Although we did not directly assess NPR1 nuclear translocation or transcriptional activity, the reduced NPR1 protein levels observed in NAD-ME1–silenced leaves, together with altered defense gene expression and enhanced BaMV accumulation, suggest that redox imbalance may compromise NPR1-associated defense signaling during BaMV infection. Our findings extend previous work [79, 80], suggesting that mitochondria may serve as signaling organelles that link ROS to defense gene expression and influences SA-dependent antiviral immunity. While earlier studies implied a role for mitochondrial redox signaling, our results pinpoint a specific metabolic enzyme—NAD-ME1—and demonstrate that its activity is important for maintaining the redox environment critical for defense gene expression. In addition, it is well-documented that reduced protein forms are more susceptible to irreversible oxidative damage and proteolytic degradation under stress conditions [81]. Supporting this notion, we observed a marked reduction in the accumulation of exogenously expressed NPR1-GFP in NAD-ME1-silenced leaves (Fig. S3a), further implicating the redox imbalance in compromising the stability of redox-sensitive defense components. It is worth noting that mitochondrial NAD-malic enzyme (NAD-ME) typically functions as a heteromeric complex composed of ME1 and ME2 subunits, with enzymatic activity derived from both homo- and heterodimers [60, 65]. While both NAD-ME1 and ME2 were detected in our proteomic analysis, only NAD-ME1 exhibited increased abundance during BaMV infection (Table S5), suggesting that ME1 is the more dynamically regulated subunit of the NAD-ME complex under these conditions. In addition to compromising immune responses, this altered redox environment may create conditions favorable for viral replication. BaMV benefits from a more reduced cytoplasmic state [82] and co-opts mitochondrial and chloroplast proteins, such as VDAC and PGK to assemble replication complexes [47, 48], highlighting the close interplay between viral replication and host metabolic reprogramming.

Conclusion

Our study demonstrates that BaMV infection induces extensive host metabolic reprogramming, which not only fulfills increased energetic and biosynthetic demands but also intersects with redox-sensitive defense signaling pathways. NAD-ME1 thus emerges as a putative metabolic hub linking mitochondrial metabolism to cytoplasmic redox balance and defense-associated transcriptional reprogramming. These findings provide deeper insight into how metabolic networks shape antiviral responses and highlight redox balance mediated by mitochondrial metabolism as a central node in plant-BaMV interactions.

Authors’ contributions

NSL and LYH conceived the project. NSL and LYH designed the research. LYH performed the experiments. NSL and LYH analyzed the data. NSL and CHW supervised the experiments. NSL, CHW and LYH wrote and edited the manuscript.

Funding

This work was supported by Academia Sinica Investigator Award (AS-IA-108-L05) and Academia Sinica Postdoctoral Scholar Program.

Data availability

The processed omics data that support the findings of this study are available in the supplemental information. The raw data are available in Zenodo (10.5281/zenodo.17521578).

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.