Progressive salinity drives flavonoid branch reprogramming in Anoectochilus roxburghii

Abstract

Flavonoids play critical roles in plant adaptation to abiotic stress; however, how salt stress modulates metabolic flux distribution within flavonoid branches remains poorly understood, particularly in non-model medicinal plants. Here, we integrated targeted metabolomics, transcriptomics, and proteomics to examine flavonoid regulation in Anoectochilus roxburghii under 0, 50, 100, and 200 mmol·L^− 1 NaCl. Metabolite profiling showed that salinity reshaped flavonoid composition rather than uniformly increasing flavonoid abundance. A metabolite-derived branch bias index (MI), representing the balance between reductive branch metabolites and flavonol products, increased under salt treatment, peaked at 100 mmol·L^− 1 NaCl, and declined at 200 mmol·L^− 1, indicating maximal branch bias under moderate stress followed by partial rebalancing under severe stress. Transcriptomic analysis showed induction of upstream phenylpropanoid and flavonoid entry genes, including PAL, 4CL, and CHS, whereas F3H was suppressed and FLS showed no induction. Furthermore, several short-chain dehydrogenase/reductase homologs (IFR-like SDR homologs) were upregulated, and the transcript-derived reductive branch index (EI) increased progressively across the salt gradient. EI was positively associated with MI, although the relationship was not strictly proportional under severe stress (200 mmol·L^− 1 NaCl). Proteomic profiling further provided supportive evidence for sustained activation of upstream flavonoid biosynthesis, such as salt-induced accumulation of chalcone synthase (CHS) protein, complementing the transcriptomic and metabolomic datasets. Together, these results indicate that salt stress reorganizes flavonoid metabolism in A. roxburghii through persistent upstream activation and branch-specific regulation, favoring the reductive branch under moderate salinity.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12870-026-09559-7.

Highlight

Salt stress reprograms flavonoid branch reprogramming in Anoectochilus roxburghii, coupling sustained upstream activation with repression of the flavonol node and preferential engagement of the reductive branch.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12870-026-09559-7.

Introduction

Salt stress is one of the major abiotic constraints limiting plant growth, metabolic homeostasis and agricultural productivity worldwide. High concentrations of Na⁺ and Cl⁻ ions disrupt osmotic balance, impair photosynthesis, induce oxidative stress, and trigger extensive reprogramming at transcriptional, metabolic, and proteomic levels [1–3]. Understanding the multi-layer regulatory mechanisms underlying salt tolerance is therefore essential for crop improvement [4]. Salinity severely restricts plant growth by disrupting ion and osmotic homeostasis, impairing photosynthesis, and promoting excessive ROS accumulation [5, 6] In response, plants undergo extensive metabolic reprogramming involving both primary and secondary metabolism, especially flavonoids, which were frequently induced to participate in antioxidant defense and redox homeostasis.

Anoectochilus roxburghii (Wall.) Lind., commonly known as ‘Jinxianlian’, is a widely used Chinese medicinal herbs belonging to the Orchidaceae family [7]. Its bioactive components, especially flavonoids such as rutin, quercetin, kaempferol, isorhamnetin, and diosmetin, exhibit antioxidant, anti-inflammatory, hepatoprotective, and antiviral activities. Recent studies have highlighted the ability of these compounds to scavenge reactive oxygen species, chelate metal ions, and modulate stress-related enzymatic pathways, thereby enhancing plant tolerance to adverse conditions [8–10]. Understanding how environmental stress modulates flavonoid in this species is important not only for elucidating stress adaptation mechanisms but also for optimizing medicinal quality through controlled cultivation strategies [11]. The potential adaptive significance of flavonoid branch reprogramming lies in its capacity to tailor antioxidant defense profiles to specific stress intensities. However, the regulatory logic governing flavonoid branch distribution under salt stress in A. roxburghii has not been systematically investigated.

Flavonoids constitute a structurally diverse group of phenolic compounds synthesized through the phenylpropanoid pathway [12]. Beyond their well-established roles in UV protection and pigmentation, flavonoids function as potent ROS scavengers, metal chelators, and modulators of redox signaling. Flavonoid biosynthesis branches into multiple sub-pathways, including flavonols, flavan-3-ols, anthocyanins, and proanthocyanidins [13]. These branches compete for shared precursors, and their relative allocation determines the qualitative composition of flavonoid pools. Recent reviews emphasize that flavonoid metabolism under stress is not merely a passive accumulation but a highly dynamic process governed by sophisticated signaling networks, involving phytohormones (ABA, SA), protein kinases (SnRK2s, MAPKs), and transcriptional reprogramming [14, 15]. Under abiotic stress, flavonoid accumulation is frequently reported to increase [16, 17]. However, most studies focus on total flavonoid content or single metabolites, leaving unresolved how carbon flux is redistributed among competing branches under progressive stress intensities [18–20]. Recent advances highlight that such branch-specific redistribution, rather than mere quantitative changes, is critical for plant environmental stress responses and tolerance.

Metabolic networks operate not only by altering total pathway output but also by dynamically reallocating carbon flux among branches [21]. Such reallocation allows plants to optimize metabolic efficiency, redirect resources toward compounds with higher functional relevance, and balance energy costs. In the flavonoid pathway [13, 22], seen as Fig. 6, key node enzymes such as flavanone 3-hydroxylase (F3H) and flavonol synthase (FLS) determine whether carbon is directed toward flavonols or reductive products such as proanthocyanidins. Downstream reductive enzymes, including dihydroflavonol 4-reductase (DFR) and anthocyanidin reductase (ANR), further shape branch output. Despite this structural complexity, branch reprogramming under salt stress remains poorly characterized, particularly in non-model medicinal plants. Whether salt stress triggers uniform pathway activation or dynamic branch-specific reprogramming is largely unknown. Specifically, the precise regulatory nodes controlling the shift between flavonol and reductive branch metabolites under progressive salinity warrant systematic investigation.

Fig. 6.

Proposed model of salt-induced flavonoid branch reprogramming in A. roxburghii. Proposed model of flavonoid branch reprogramming under progressive salt stress in A. roxburghii. Salt stress induces sustained activation of upstream phenylpropanoid and flavonoid entry enzymes, including PAL, 4CL, and CHS, thereby maintaining precursor supply. Under moderate salinity, downregulation of the flavonol node (F3H/FLS) together with induction of reductive branch-associated IFR-like SDR homologs promotes preferential metabolite redistribution toward reductive branch metabolites, including epicatechin and procyanidin B2, resulting in maximal branch bias. Under severe salinity, upstream pathway activation persists, but branch reprogramming becomes partially rebalanced, suggesting metabolic constraints or regulatory feedback under extreme stress. Blue indicates the flavonol branch, and orange indicates the reductive branch

Therefore, in this study, we integrated targeted metabolomics and transcriptomics, complemented by proteomics, to determine whether progressive salinity (0, 50, 100, and 200 mmol·L^{− 1} NaCl) reorganizes flavonoid branch preference and regulates flavonoid branch metabolism in A. roxburghii. We hypothesized that progressive salt stress reorganizes branch reprogramming within the flavonoid pathway, shifting flux away from flavonol production and toward the reductive branch under moderate salinity rather than uniformly enhancing overall flavonoid biosynthesis. To evaluate this process, we developed complementary metabolite- and transcript-based indices (MI and EI) to quantify branch reprogramming at two regulatory levels [23–25]. Our results show that salinity maintains upstream pathway activation while altering downstream branch selection.

Materials and methods

Plant materials and salt treatments

Anoectochilus roxburghii (Wall.) Lind. (Fujian “Dayehongxia” cultivar) was cultivated in a controlled greenhouse under the following conditions: 25 ± 2 °C, 70% relative humidity, and a 12 h light/12 h dark photoperiod. Plants were grown in soil substrate under uniform management conditions. Salt treatments were applied using NaCl solutions at four concentrations: 0 mmol·L^− 1 (control), 50 mmol·L^− 1 (low salt), 100 mmol·L^− 1 (moderate), and 200 mmol·L^− 1 (high salt). To minimize osmotic shock, salt concentration was gradually increased during the first week. Plants were irrigated every three days (250 mL per pot 45 cm*25 cm*15 cm) beginning with one-third of the final concentration and progressively adjusted to the target concentration over 7 days. After 30 days of treatment, fully expanded leaves at the same developmental stage (third leaf below the flag leaf) were harvested. Three independent biological replicates were collected per treatment, each replicate consisting of pooled tissue from multiple plants. All the following analyses were performed using the same leaf tissues collected from independent biological replicates. Samples were immediately frozen in liquid nitrogen and stored at − 80 °C until further analysis.

No wild plant materials were collected in this study. All experimental seedlings of the A. roxburghii “Dayehongxia” cultivar were artificially propagated seedlings, which were introduced from Nanjing County, Zhangzhou City, Fujian Province, China by Associate Researcher Jiangbo Lin (Institute of Subtropical Agriculture, Fujian Academy of Agricultural Sciences). Formal species identification of the experimental materials was completed and confirmed by Jiangbo Lin. The germplasm of “Dayehongxia” cultivar has been permanently preserved in the germplasm repository of our institution under the accession number HJ2015.All plant cultivation, stress treatment, and sample collection procedures were performed in accordance with institutional rules and national relevant research guidelines.

Targeted flavonoid metabolite profiling

Targeted quantification of flavonoids was performed using UPLC–MS/MS in multiple reaction monitoring (MRM) mode. Briefly, approximately 100 mg of freeze-dried leaf tissue was extracted with 80% methanol, followed by vortexing, sonication at 4 °C, and centrifugation. The supernatant was evaporated to dryness and reconstituted in 80% methanol containing an internal standard prior to analysis.

Chromatographic separation was conducted using a Waters Acquity UPLC system equipped with an Acquity UPLC HSS T3 column (1.8 μm, 2.1 × 100 mm). The column temperature was maintained at 40 °C, and the flow rate was set to 0.300 mL·min⁻¹. The gradient increased from 10% to 90% acetonitrile over 14 min, followed by re-equilibration, with a total run time of 18 min. Mass spectrometric detection was performed on an AB SCIEX 5500 QTRAP mass spectrometer equipped with an electrospray ionization (ESI) source operating in multiple reaction monitoring (MRM) mode. Source parameters were optimized for flavonoid detection. Compound identification was based on retention time and specific MRM transitions confirmed using authentic standards.

Quantification was performed using external calibration curves, and metabolite concentrations were normalized to internal standards and expressed as ng·g⁻¹ dry weight. Three independent biological replicates were analyzed per treatment. Compounds included flavonols (kaempferol, isorhamnetin), flavan-3-ols (epicatechin), and proanthocyanidin-related metabolites (procyanidin B2), among others. All metabolite measurements were performed in triplicate biological replicates. (Supplementary Dataset S6)

RNA extraction, sequencing, and functional annotation

Total RNA was extracted from frozen leaf tissues using TRIzol reagent (Life technologies) following the manufacturer’s instructions. RNA integrity was assessed using an Agilent 2100 Bioanalyzer and agarose gel electrophoresis, and RNA purity and concentration were determined with the Nanodrop micro-spectrophotometer (Thermo fisher). Full-length transcriptome sequencing was performed using the PacBio single-molecule real-time (SMRT) platform. Briefly, mRNA was enriched by oligo(dT) magnetic beads and reverse-transcribed into cDNA using the Clontech SMARTer PCR cDNA Synthesis Kit. PCR cycle optimization was conducted to determine the optimal amplification cycle number. Double-stranded cDNA was generated and subjected to size selection (> 5 kb) using the BluePippin™ Size-Selection System. Size-selected and non-size-selected cDNA fractions were pooled for large-scale PCR amplification. The amplified cDNA was end-repaired, ligated to SMRTbell adapters, and sequenced on the PacBio Sequel II platform by Gene Denovo Biotechnology Co. (Guangzhou, China).

Raw sequencing data were processed using the isoform sequencing (Iso-Seq) pipeline supported by Pacific Biosciences [26]. Firstly, high quality circular consensus sequences (CCS; HiFi reads) were extracted out of subreads BAM file, and full-length non-chimeric (FLNC) reads were identified based on the presence of 5’ primer, 3’ primer, and polyA tail signals. And FLNC reads were clustered to generate the entire isoform. Similar FLNC reads were used minimap2 to cluster hierarchically to get the consistency sequence (unpolished consensus isoforms). Then the quiver algorithm was used to further correct the consistency sequence. According to the results, the high quality isoforms (prediction accuracy is ≥ 0.99) were used to do sequent analysis.

For downstream quantitative analyses, transcript abundance was estimated using FPKM values derived from RNA-seq expression profiling. Expression data were log₂-transformed [log₂ (FPKM + 1)] prior to statistical testing.

High-quality isoforms obtained from SMRT sequencing were functionally annotated through sequence similarity searches against major public protein and pathway databases using BLAST with a significance threshold of E-value ≤ 1🞨e^− 5. Gene Ontology (GO) annotation was assigned using Blast2GO software, and functional classification was conducted using WEGO software. Gene Ontology terms were assigned based on homologous annotations, and KEGG pathway mapping was used to identify transcripts associated with phenylpropanoid and flavonoid biosynthesis. Open reading frames (ORFs) were predicted to obtain coding sequences (CDs) and corresponding protein translations by using ANGEL software. Protein Domain Prediction was performed by aligning protein sequences of isoforms to Pfam database (version 26.0) by Pfam_Scan program and SMART database (version 06/08/2012) by HMMER-profile hidden Markov models for biosequence analysis (http://hmmer.org/) program to obtain protein domain annotations. Transcription factor families were predicted using hmmscan to Plant TFdb (http://planttfdb.cbi.pku.edu.cn/).

These procedures enabled comprehensive structural and functional characterization of the A. roxburghii transcriptome for downstream pathway and regulatory analyses.

Proteomic analysis

Proteomic profiling was performed using data-independent acquisition (DIA) mass spectrometry. Leaf samples from control, 50 mmol·L^− 1 NaCl, and 100 mmol·L^− 1 NaCl treatments (three biological replicates per group) were subjected to label-free quantitative analysis. Due to resource limitations, proteomic data were not obtained for the 200 mmol·L^− 1 NaCl treatment.

Total proteins were extracted using a phenol-based extraction method, followed by acetone precipitation and resuspension in lysis buffer. Protein concentration was determined using a bicinchoninic acid (BCA) assay, and protein integrity was verified by SDS-PAGE. Equal amounts of protein from each sample were reduced, alkylated, and digested with trypsin overnight at 37 °C. Peptides were desalted using solid-phase extraction prior to LC-MS/MS analysis.

Peptide separation was performed on a reversed-phase C18 column using a binary gradient of (A) 0.1% aqueous formic acid and (B) acetonitrile/water/formic acid (80:20:0.1, v/v/v) as the mobile phases. The liquid chromatography system was coupled online to a hybrid trapped ion mobility spectrometry-quadrupole time-of-flight mass spectrometer (timsTOF Pro, Bruker). Eluted peptides were analyzed in data-independent acquisition (DIA) mode on the high-resolution mass spectrometer. Indexed retention time (iRT) peptides (Biognosys, Thermo Fisher Scientific) were spiked into each sample for retention time calibration. Raw DIA data were processed using DIA-NN software for peptide identification and protein quantification. Carbamidomethylation of cysteine was set as a fixed modification, and methionine oxidation and protein N-terminal acetylation were specified as variable modifications. Peptide and protein identifications were filtered at a false discovery rate (FDR) of 1%. Quantification was based on MS2-level signal intensities.

Differentially abundant proteins between treatment groups were identified using unpaired t-tests with significance defined as P < 0.05 and Fold change (FC) ≥ 1.2 or ≤ 1/1.2.

Construction of metabolite-derived branch bias index (MI) and transcript-derived reductive branch expression index (EI)

To quantitatively describe branch reprogramming, we further constructed a metabolite-derived branch bias index (MI) and a transcript-derived reductive branch expression index (EI) (Fig. 1, Supplementary Datasets S3 and S4).

Fig. 1.

Experimental design and overview of the workflow. Experimental design and overview of the workflow used to investigate salt-induced flavonoid branch reprogramming in Anoectochilus roxburghii. Plants were subjected to four NaCl treatments (0, 50, 100, and 200 mmol·L^-1) for 30 d, and leaves at the same developmental stage were collected for targeted flavonoid metabolomics and transcriptome analysis. Proteomic profiling was performed for control, 50 mmol·L^-1, and 100 mmol·L^-1 treatments. Targeted metabolomics was used to quantify branch-associated flavonoids, transcriptomics was used to characterize the expression of structural genes and reductive branch-associated SDR homologs, and proteomics was used to validate protein-level changes in upstream flavonoid biosynthesis. Metabolite-derived branch bias index (MI) and transcript-derived reductive branch expression index (EI) were further constructed to evaluate branch reprogramming patterns under progressive salt stress

MI was constructed to estimate the relative metabolic redistribution between the flavonol branch and the proanthocyanidin-related reductive branch. In the flavonoid pathway, flavonols such as kaempferol and isorhamnetin represent products of the flavonol synthase (FLS) branch, whereas flavan-3-ols and their oligomers, including epicatechin and procyanidin B2, are representative metabolites of the reductive branch derived from dihydroflavonol reduction. Based on these biochemical relationships, MI was calculated as:

The value of MI reflects the relative metabolic bias toward reductive branch metabolites compared with flavonol-derived products. Higher MI values indicate a stronger metabolic shift toward the reductive branch. MI values were calculated for each biological replicate using absolute metabolite concentrations obtained from targeted flavonoid metabolite analysis.

EI was constructed to evaluate transcriptional regulation associated with flavonoid branch reprogramming. Because canonical reductive enzymes such as dihydroflavonol reductase (DFR) and anthocyanidin reductase (ANR) belong to the short-chain dehydrogenase/reductase (SDR) superfamily, transcripts annotated as IFR-like SDR homologs showing significant responses to salt stress were selected as candidate reductive branch enzymes. The EI index was defined as the ratio between the summed expression levels of selected reductive branch SDR homologs and the expression level of flavonol synthase, a key enzyme directing flux toward flavonol biosynthesis:

Transcript abundance values were derived from RNA-seq data and log₂-transformed [log₂ (FPKM + 1)] prior to calculation. EI values were computed for each sample to estimate transcriptional bias toward the reductive branch relative to the flavonol pathway.

Statistical analysis

All experiments were performed using three independent biological replicates unless otherwise stated. Statistical analyses were conducted using R software (version 4.4.3). For comparisons among different salinity treatments, one-way analysis of variance (ANOVA) was performed, followed by Tukey’s HSD multiple comparisons to determine significant differences among treatment means. Differences were considered statistically significant at P < 0.05. Pearson’s correlation analysis was performed to evaluate the relationship between MI and EI across salinity treatments. Correlation coefficients (r) were calculated using calculated using R (version 4.4.3).

Results

Progressive salt stress reshapes flavonoid metabolic profiles and branch reprogramming in A. roxburghii

Targeted flavonoid profiling showed that salt treatment substantially altered the composition of the flavonoid pool in A. roxburghii leaves (Fig. 2A, Supplementary Dataset S1). Rather than a uniform increase across all detected metabolites, salinity induced clear branch-specific changes. Under 50 mmol·L^− 1 NaCl, several flavonoids showed elevated accumulation relative to the control, indicating early activation of flavonoid-associated stress responses. Under 100 mmol·L^− 1 NaCl, the metabolic response became more selective, with stronger enrichment of reductive branch-related metabolites and clear reduction of representative flavonol branch products. Under 200 mmol·L^− 1 NaCl, multiple metabolites remained altered relative to the control, but the branch bias was weaker than that observed at 100 mmol·L^− 1, suggesting partial metabolic rebalancing.

Fig. 2.

Salt stress reshapes flavonoid metabolite profiles and branch reprogramming. Salt-induced changes in flavonoid metabolite accumulation in A. roxburghii leaves. A Heatmap showing relative accumulation patterns of targeted flavonoids across Control (0 mmol·L^-1 NaCl), Low salt (50 mmol·L^-1 NaCl), Moderate (100 mmol·L^-1 NaCl), and High salt (200 mmol·L^-1 NaCl) treatments. Metabolite abundances were normalized by row and visualized using Z-scores. B Relative abundance of representative flavonol branch metabolites (kaempferol and isorhamnetin) and reductive branch metabolites (epicatechin and procyanidin b2) under progressive salt stress. C Metabolite-derived branch bias index (MI), calculated as (epicatechin + procyanidin B2)/ (kaempferol + isorhamnetin), across the four treatment groups. MI increased under salt stress, peaked at 100 mmol·L^-1 NaCl, and declined at 200 mmol·L^-1, indicating maximal reductive branch reprogramming under moderate salinity followed by partial rebalancing under severe stress. Values are shown as mean ± SD (n = 3 biological replicates). Different letters indicate significant differences at P < 0.05. (Supplementary Dataset S3)

The behavior of representative metabolites supported this interpretation (Fig. 2B). The flavonol-associated compounds kaempferol and isorhamnetin were relatively reduced under moderate salinity, whereas l-epicatechin and procyanidin B2, which represent the reductive branch, were markedly elevated under salt treatment.

To quantify branch reprogramming, we calculated the metabolite-derived branch bias index (MI). MI increased under salt treatment relative to the control, indicating a shift in metabolite reprogramming toward the reductive branch (Fig. 2C). The highest MI was observed at 100 mmol·L⁻¹NaCl, demonstrating that moderate salinity produced the strongest branch bias. At 200 mmol·L⁻¹NaCl, MI declined but remained above the control level, consistent with partial metabolic rebalancing at high salinity.

These data indicate that salinity alters flavonoid branch reprogramming in a nonlinear manner, with maximal metabolite-level bias under moderate stress.

Upstream phenylpropanoid and flavonoid entry genes remain activated under salinity

To determine whether branch remodeling was accompanied by changes in pathway entry, we examined representative upstream genes involved in phenylpropanoid activation and flavonoid biosynthesis. Phenylalanine ammonia-lyase (PAL) and 4-coumarate-CoA ligase (4CL) isoforms showed substantial induction under salt treatment, particularly at 100 and 200 mmol·L⁻¹NaCl (Fig. 3A), indicating enhanced upstream precursor activation.

Fig. 3.

Upstream flavonoid biosynthesis remains activated under salt stress. Transcript and protein responses of upstream flavonoid biosynthetic enzymes under salt stress. A Differential transcript responses of representative upstream genes involved in phenylpropanoid activation, including PAL and 4CL isoforms, under 50, 100, and 200 mmol·L^-1 NaCl treatments. Values are shown as mean ± SD (n = 3 biological replicates). Different letters indicate significant differences at P < 0.05. B Heatmap showing differential expression patterns of CHS isoforms across salt treatments, highlighting strong induction under 200 mmol·L^-1 NaCl. Data are presented as mean ± SD. C CHS protein abundance under control, 50 mmol·L^-1, and 100 mmol·L^-1 NaCl treatments. Full-length CHS protein (Accession A0A5B8IM30; 5 unique peptides) was not detected in control samples but was readily detected in both salt-treated groups, indicating salt-induced accumulation of CHS at the protein level

Among upstream structural genes, chalcone synthase (CHS) showed the strongest transcriptional response. Heatmap analysis of four representative CHS isoforms showed robust induction under salt treatment, with especially strong upregulation at 100 and 200 mmol·L⁻¹NaCl (Fig. 3B). This pattern suggests that entry into the flavonoid pathway remained active throughout the salinity progression. Proteomic evidence further supported these transcript trends. CHS protein was not detected in the control samples but was detected in both 50 and 100 mmol·L^− 1 NaCl treatments (Fig. 3C), indicating salt-induced accumulation of CHS at the protein level. Together, these results show that salt stress does not shut down the flavonoid pathway at the entry stage; instead, upstream phenylpropanoid and flavonoid entry steps remain activated.

The flavonol node is suppressed under moderate salt stress

Although upstream genes were induced, the flavonol node genes showed a contrasting response pattern. F3H expression decreased under salt treatment, with significant downregulation under 50 and 100 mmol·L⁻¹NaCl, while remaining lower than the control under 200 mmol·L^− 1 NaCl (Fig. 4A). FLS did not show salt-induced activation and instead remained relatively low across treatments (Fig. 4A). This differential response suggests that branch regulation may occur primarily through modulation of substrate entry into the flavonol node rather than coordinated induction of all downstream enzymes.

Fig. 4.

Suppression of the flavonol node and induction of reductive branch-associated SDR homologs. Transcriptional regulation of branch-associated genes under progressive salt stress. A Expression patterns of flavonol node genes, including F3H and FLS, across control, 50, 100, and 200 mmol·L^-1 NaCl treatments. Expression values are shown as log2(FPKM+1). Bars represent mean ± SD (n = 3 biological replicates). Different letters indicate significant differences at P < 0.05. F3H showed significant downregulation under 50-100 mmol·L^-1 NaCl, while FLS displayed a mild downward trend. B Heatmap showing expression responses of all detected IFR-like SDR homologs under salt stress. Values represent log2FC relative to the control based on log2(FPKM+1)-transformed expression values

This transcriptional pattern was consistent with the metabolite data. The relative reduction of kaempferol and isorhamnetin under moderate salinity coincided with the suppression of F3H/FLS and the peak of MI at 100 mmol·L⁻¹NaCl. These data suggest that salinity suppresses the flavonol node, thereby limiting carbon flow toward flavonol production and contributes to branch redistribution.

A subset of IFR-like SDR homologs is induced under salt stress

We screened the transcriptome for IFR-like short-chain dehydrogenase/reductase (SDR) homologs as candidate reductive branch-associated genes, due to the lack of canonical DFR and ANR isoforms annotated in the current transcriptome set. Among the detected candidates, a subset showed clear induction under salt treatment, especially at 100 and 200 mmol·L^− 1 NaCl. Heatmap visualization of the selected IFR-like SDR homologs showed that these transcripts were induced under salt treatment, with stronger upregulation under 100 and 200 mmol·L^− 1 NaCl than under 50 mmol·L^− 1 NaCl (Fig. 4B, Supplementary Dataset S2). It should be emphasized that the SDR homologs identified in this study were annotated based on sequence similarity. Their precise catalytic functions remain unknown and require future biochemical characterization. Therefore, our interpretation is limited to their association with reductive branch-associated homologs rather than confirmed enzymatic activity.

To summarize the coordinated response of branch-associated genes, we integrated the behavior of the flavonol node and IFR-like SDR homologs into a simplified branch-shift scheme (Fig. 6). In this model, suppression of F3H and FLS reduces flux toward the flavonol branch, whereas induction of IFR-like SDR homologs promotes the reductive branch. The contrasting metabolite patterns of kaempferol/isorhamnetin versus epicatechin/procyanidin B2 were consistent with this direction of reallocation. Thus, transcriptomic and metabolomic evidence together support a salt-induced shift away from flavonol production and toward reductive branch-associated pathways.

Transcript-derived reductive branch index (EI) increases under salt treatment

To quantify transcriptional regulation of branch reprogramming, we constructed the transcript-derived reductive branch expression index (EI) using the best-performing IFR-like SDR homolog model relative to FLS expression (Fig. 5A). EI increased progressively from the control to 50, 100, and 200 mmol·L^− 1 NaCl, indicating an increasingly strong reductive transcriptional program with greater salt intensity.

Fig. 5.

Concordance between transcript-derived and metabolite-derived branch indices. Relationship between transcript-level and metabolite-level indicators of flavonoid branch reprogramming. A Transcript-derived reductive branch expression index (EI), calculated as the summed abundance of selected IFR-like SDR transcripts divided by FLS transcript abundance, across control and salt treatments. B Relationship between transcript-level reductive branch activation (EI) and metabolite-level branch bias (MI) across all samples. Each point represents one biological replicate. The positive but modest association (Pearson’s r = 0.313, P = 0.321) suggests that transcriptional activation of the reductive branch generally tracked metabolite redistribution, although the two indices were not strictly proportional under severe salt stress. C Comparative trends of transcript-derived reductive branch index (EI) and metabolite-derived branch bias index (MI) under progressive salt stress. MI is shown on the left y-axis and EI on the right y-axis. Values represent mean ± SD (n = 3 biological replicates), Different letters indicate significant differences at P < 0.05. EI increased progressively with salt intensity, whereas MI peaked at 100 mmol·L^-1 NaCl and declined at 200 mmol·L^-1, indicating that transcript-level reductive branch activation and metabolite-level branch bias were broadly coordinated but not strictly proportional under severe stress

When EI and MI were compared across all samples, EI showed a positive but modest association with MI (Fig. 5B), indicating that transcript-level reductive branch activation generally tracked metabolite-level branch redistribution, with the exception of one outlier point at 200 mmol·L^− 1 NaCl. We then plotted EI and MI across the salt gradient (Fig. 5C). EI increased steadily with salt intensity, whereas MI peaked at 100 mmol·L^− 1 NaCl and declined at 200 mmol·L^− 1. This divergence indicates that transcript-level reductive branch activation continued under severe stress, while metabolite-level branch bias was partially reduced.

Discussion

Salt stress reorganizes flavonoid composition rather than simply increasing flavonoid abundance

Flavonoids are central components of plant responses to abiotic stress because of their antioxidant, redox-buffering, and signaling functions [9]. Most existing studies characterize salt stress primarily as inducing an overall increase in total flavonoid accumulation. However, our findings reveal that the flavonoid response in A. roxburghii under salt stress entails a branch-specific metabolite redistribution across distinct flavonoid biosynthetic branches, rather than a global upregulation of flavonoid production.

The metabolite-derived branch bias index (MI) clearly illustrates this pattern. MI increased under salt treatment, reached its maximum at 100 mmol·L^{− 1} NaCl, and decreased at 200 mmol·L^{− 1} while remaining higher than the control level. This response indicates that flavonoid metabolism under salinity is not simply increase but dynamically reorganized according to stress intensity. Moderate salinity produced the strongest bias toward reductive branch metabolites, whereas severe salinity leads to partial rebalancing within branch reprogramming. It shifts the interpretation of salt-induced flavonoid responses from “more flavonoids” to “different flavonoid composition,” which is more mechanistically informative for both stress physiology and medicinal quality. This perspective aligns with recent frameworks suggesting that salt stress triggers specific qualitative adjustments in secondary metabolism rather than indiscriminate quantitative enhancements [15].

Sustained upstream activation maintains precursor supply under salinity

A prominent feature of our dataset is the pronounced activation of upstream phenylpropanoid and flavonoid biosynthetic entry genes. PAL and 4CL increased under salt treatment, and CHS exhibited especially strong induction at the transcript level. Proteomic data further corroborated these findings, revealing that CHS protein was detectable only in salt-treated samples. Together, these observations indicate that salinity does not repress pathway entry. These results suggest that severe salt stress does not suppress the overall flavonoid pathway. Instead, precursor supply remains active, allowing downstream branch reprogramming to remain flexible. In other words, the pathway entry steps appear to provide a metabolically active background on which branch-specific regulation can occur. Such a configuration would allow the plant to maintain a responsive phenylpropanoid reservoir, from which metabolites can be selectively directed toward the most functionally relevant branch.

Repression at the flavonol node and induction of IFR-like SDR homologs

In contrast to the strong activation of upstream genes, F3H and FLS showed no induction. F3H decreased under salt treatment, with the clearest suppression at 50 and 100 mmol·L^− 1 NaCl, while FLS showed no induction and remained relatively low across treatments, which were coincided with the reduction of representative flavonol metabolites and with the maximal MI observed at 100 mmol·L^− 1 NaCl. Mechanistically, the lack of FLS induction is likely a consequence of substrate deprivation; since F3H catalyzes the committed step converting dihydroflavonols from the central flux, its suppression effectively starves FLS of its substrates. This regulatory strategy effectively restricts carbon flux into the flavonol branch, thereby favoring the competing reductive branch mediated by IFR-like SDR homologs.

Kaempferol and isorhamnetin are typical flavonol branch products [27], their relative decline under moderate salt treatment is consistent with reduced flux through this route. These behavior in A. roxburghii indicate that the flavonol node acts as a key control point in branch reprogramming. Suppression of F3H/FLS limits carbon flow toward flavonol production and thereby favors alternative downstream reprogramming.

Canonical DFR and ANR isoforms were not explicitly recovered from the current annotation set, which is not unusual in de novo transcriptome datasets from non-model medicinal species, where incomplete gene models and limited pathway-specific annotation can hinder the recovery of specialized-metabolism genes [28–30]. However, we therefore re-screened the transcriptome and identified a subset of IFR-like SDR homologs that showed clear salt-responsive induction, especially under 100 and 200 mmol·L^{− 1} NaCl. We interpret these genes as reductive branch-associated SDR candidates, rather than assigning them as canonical reductive enzymes (DFR or ANR). The selected IFR-like SDR homologs are induced under salt stress, together with the accumulation of epicatechin and procyanidin B2, strongly suggests that salt treatment activates a transcriptional program favoring the reductive branch. It is noteworthy that the identification of these proteins relied on sequence annotation. Their exact enzymatic functions remain to be unequivocally confirmed through dedicated biochemical assays and phylogenetic investigations. Likewise, MI and EI should be regarded as empirical indicators describing coordinated changes in metabolite composition and transcript abundance rather than direct measurements of metabolic flux. Future isotope-labeling studies will be required to validate the inferred branch preferences.

Additionally, EI and MI showed a positive but modest association rather than a strict linear relationship, with the largest deviation occurring under 200 mmol·L^{− 1} NaCl (Fig. 5B). This divergence under severe salinity may reflect additional constraints beyond transcript abundance, including redox limitation, altered substrate partitioning, post-transcriptional control, enzyme-level flux regulation, and broader metabolic instability under extreme stress [5, 31, 32]. Moderate salinity may promote antioxidant-related metabolite accumulation, whereas severe salinity likely imposes metabolic and oxidative constraints, resulting in partial attenuation of branch bias. Therefore, the 200 mmol·L^{− 1} response indicates a situation in which the transcriptional program remains active, but the metabolic system can no longer convert that activation into proportional branch bias.

The different responses observed under moderate and severe salinity may reflect distinct physiological demands during stress adaptation. Moderate salinity was associated with increased accumulation of epicatechin and procyanidin B2, both of which possess strong antioxidant properties and may contribute to maintaining cellular redox homeostasis. In contrast, although transcript-derived branch preference continued to increase under 200 mmol·L^− 1 NaCl, the metabolite-derived branch bias declined, suggesting that severe salinity may impose metabolic and oxidative constraints that limit further accumulation of reductive branch metabolites.

A. roxburghii flavonoid branch reprogramming acts in a three-stage model

Based on the integrated metabolomic, transcriptomic, and proteomic data, we propose a three-stage model of flavonoid branch reprogramming under salt stress in A. roxburghii, as shown in Fig. 6. In Stage I (50 mmol·L^− 1 NaCl), the pathway shows general activation, as reflected by increased flavonoid accumulation and induction of upstream biosynthetic genes (such as PAL, 4CL, CHS et al.). This stage likely represents an early protective response that enhances antioxidant buffering. In Stage II (100 mmol·L^− 1 NaCl), reductive branch reallocation becomes pronounced. At this stage, MI reaches its maximum and IFR-like SDR homologs are induced, whereas F3H/FLS are suppressed. This stage reflects the process by which metabolic resources are reprogrammed to favor reductive branch metabolites. In Stage III (200 mmol·L^− 1 NaCl), upstream pathway activation persists and EI continues to increase, but MI declines relative to 100 mmol·L^− 1. This indicates that severe salinity induces partial metabolic rebalancing, while the system remains transcriptionally activated.

Phenotypic observations further demonstrated progressive reductions in biomass and survival (Supplementary Dataset S5), together with increased leaf chlorosis and wilting severity as salinity intensified, confirming the overall physiological impact of salt stress. Notably, the strongest metabolite-derived branch bias occurred under moderate salinity, whereas severe salinity was characterized by pronounced physiological impairment and reduced branch bias despite continued transcriptional responses. Our results suggest that flavonoid branch reprogramming may contribute to stress acclimation under moderate salinity, while prolonged or severe salt stress imposes broader physiological constraints that cannot be overcome solely through transcriptional regulation. Nevertheless, additional physiological analyses, including ion homeostasis, oxidative stress markers, and photosynthetic performance, will be required to further clarify the functional significance of these metabolic responses.

The metabolic profile and phenotypic observations after 30 days of treatment likely reflect a stabilized acclimation state. However, flavonoid metabolism is highly dynamic, and branch-specific regulation may vary during different stages of stress progression. Early salt exposure may preferentially activate rapid antioxidant responses, whereas prolonged stress duration may have triggered feedback inhibition or depleted cellular energy reserves, thereby constraining the initial branch bias observed at Stage II. This temporal dimension suggests that the redistribution mechanism is not static but evolves dynamically as the plant transitions from acute stress response to chronic stress adaptation. However, substantial reductions in biomass and survival, together with increased leaf chlorosis and wilting, were observed under all salt treatments, particularly at 100 and 200 mmol·L^− 1 NaCl (Supplementary Dataset S5). These results indicate that although flavonoid branch reprogramming represents a sustained metabolic response to salinity, it was insufficient to prevent progressive physiological deterioration under prolonged or severe salt stress. Therefore, the observed branch-specific responses should be interpreted as one component of the overall stress response rather than evidence of effective long-term salt tolerance.

Future time-course studies under a wider range of salinity conditions will be required to evaluate the duration and physiological consequences of this branch-specific response, and determine how long branch reprogramming can be maintained and whether it contributes to improved growth, survival, or recovery under moderate stress conditions.

Conclusion

A. roxburghii is valued for its flavonoid-rich medicinal properties, salinity-induced flavonoids branch reprogramming may affect both stress adaptation and metabolite quality, which could influence both antioxidant properties and medicinal quality.

Branch reorganization proposed a strategy for optimizing metabolite composition under different stress levels. The observed shift in our study may reflect an adaptive strategy in which A. roxburghii preferentially deploys metabolite classes that are more effective under a given oxidative or osmotic environment. Our findings suggest that controlled salinity could potentially be used to reshape flavonoid composition in cultivation systems, although further pharmacological validation would be required.

Future work integrating functional validation of SDR candidates, enzyme assays, and direct flux measurements will be helpful to refine the current model. Nevertheless, the present study provides a robust framework for understanding how salt stress reorganizes flavonoid branch reprogramming in a non-model medicinal species.

Abbreviations

MI

Metabolite-derived branch bias index

EI

Transcript-derived reductive branch index

PAL

Phenylalanine ammonia lyase

CHS

Chalcone synthase

FLS

Flavonol synthase

CHI

Chalcone isomerase

DFR

Dihydroflavonol 4-reductase

ANR

Anthocyanidin reductase

SDR

Short-chain dehydrogenase/reductase

Authors’ contributions

H. Huang designed the experiments. H. Huang, W. Bao, J. Hong, and J. Lin carried out the experiments. H. Huang, J. Hong, X. Yu, J. Lin, and Y. Dai analyzed the data. H. Huang wrote the manuscript. All the authors read and approved the final manuscript.

Funding

This work was supported by the Fujian Provincial Department of Science and Technology (Grant No. 2024R1029003).

Data availability

The datasets analysed during the current study are available in the NCBI GenBank repository, accession number: PRJNA1420432. http://www.ncbi.nlm.nih.gov/bioproject/1420432 (accessed on 9 February 2026). All data and materials supporting the findings of this study are available in the main article and its supplementary information files.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.