Abstract

   The browning reaction during the flue-curing process of tobacco
   (Nicotiana tabacum L.) dramatically affects leaf quality and value,
   leading to significant economic losses. However, the molecular
   regulatory mode of the browning in flue-cured tobacco is still unclear.
   The two tobacco cultivars, Dabaijin and Yunyan87, with browning
   differences in upper, middle, and lower leaves during the flue-curing
   process, were characterized through transcriptomic and metabolomic
   analyses. The results showed that the significant browning in Dabaijin
   was mainly due to the preferential accumulation of products of amino
   acid metabolism (L-arginine, L-ornithine, L-leucine, and L-valine) and
   amino sugar and nucleotide sugar metabolism (L-fucose) which promoted
   the Maillard reaction. Meanwhile, riboflavin metabolism (downregulation
   of PYRP and RIBF expression) impaired the function of the antioxidant
   system, and imbalanced inositol phosphate metabolism affected membrane
   homeostasis through PIP2. However, the consumption rate of reduced
   hydrogen by oxidative phosphorylation, as well as enhanced sulfur
   metabolism, and taurine and hypotaurine metabolism for scavenging ROS
   was mainly increased in Yunyan87, ultimately reducing the accumulation
   of ROS and slowing membrane damage during the flue-curing process. This
   research revealed that the difference in browning between Yunyan87 and
   Dabaijin cultivars was mainly regulated by the accumulation of Maillard
   substrates and the efficiency of the antioxidant system, which provides
   a theoretical basis for improving tobacco cultivars resistant to
   browning and optimizing processing techniques.

Supplementary Information

   The online version contains supplementary material available at
   10.1186/s12870-025-07150-0.

   Keywords: Leaf browning, Metabolome, Transcriptome, Maillard reaction,
   Antioxidant system, Tobacco

Introduction

   Browning is a phenomenon that occurs when a series of reactions produce
   brown polymers during the storage and processing of flue-cured tobacco
   and the value of the flue-cured tobacco is closely related to its
   degree of browning, which increases dramatically, especially during
   intense flue-curing. Flue-cured tobacco is a valuable commodity because
   of its unique aroma [[42]1–[43]3]. The flue-curing process includes
   stages of yellowing, color-fixing, and stem-drying. Browning takes
   place during the yellowing and color-fixing stages, gradually
   transforming the leaf’s color from green to brown while imparting a
   distinctive aroma [[44]4]. However, excessive browning can lead to
   black spots, reducing the value and industrial usability of tobacco
   leaves [[45]5, [46]6].

   Browning reactions can be categorized into non-enzymatic and enzymatic.
   Non-enzymatic browning is a reaction of carbonyl compounds (reducing
   sugars) and amino compounds (amino acids) that does not involve enzymes
   during heating or storage, of which the Maillard reaction is the most
   important non-enzymatic browning [[47]7]. In contrast, enzymatic
   browning is the oxidation of polyphenols to brown polymers by the
   action of polyphenol oxidase (PPO) [[48]8]. Under high-temperature
   flue-curing stress, the chloroplast and mitochondrial electron
   transport chains accumulate excess reactive oxygen species (ROS) such
   as superoxide anion, hydrogen peroxide, and hydroxyl radicals [[49]9,
   [50]10]. ROS peroxidize lipid membranes and disrupt cellular
   compartmentalization that causes the localization of PPOs and
   substrates in the plastid, ultimately leading to enzymatic browning
   [[51]11]. In addition, the integrity of the plant biofilm system under
   stress directly determines the occurrence of non-enzymatic and
   enzymatic browning [[52]12]. The plant biofilm system acts as a vital
   structural barrier, compartmentalizing various substances within living
   organisms. The stress process induces injuries, disrupting the
   stability of the membrane and resulting in the leakage of cellular
   contents. This is conducive to the occurrence of browning reaction
   [[53]13–[54]16].

   With the application of analysis methods of the transcriptomics and
   metabolomics, the identification and analysis of key genes and
   metabolites in plants under stress are increasing, making it possible
   to elucidate the molecular mechanisms of browning [[55]11]. These
   approaches have been successfully applied to investigate the molecular
   mechanisms of browning in crops such as Malus domestica [[56]17],
   Prunus salicina [[57]18], Glycyrrhiza uralensis [[58]19], and Lactuca
   sativa [[59]20]. In tobacco, metabolomics reveals that the accumulation
   of amino acids mainly causes the post-harvest browning of cigar tobacco
   [[60]5]. In addition, transcriptomic and metabolomic analyses of
   tobacco leaf with stem show that inhibiting membrane lipid degradation
   and increasing the supply of reduced hydrogen reduce browning [[61]4].
   However, the molecular mechanisms of browning in flue-cured tobacco are
   still not fully understood.

   In this study, the transcriptome and metabolome of the two tobacco
   cultivars Yunyan87 and Dabaijin were analyzed to determine their
   changes during the browning process and to dissect the molecular
   mechanism underpinning the difference in the degree of browning of the
   upper, middle, and lower leaves during the flue-curing process
   (yellowing stage and color-fixing stages). The results of this study
   will provide a more hierarchical insight into the molecular mechanisms
   of browning and provide theoretical basis for the further optimization
   of the flue-curing process.

Materials and methods

Plant materials and experimental treatments

   The tobacco cultivars Yunyan87 (Y) and Dabaijin (D) were grown
   according to the National Standard of China Tobacco Industry (GB/T
   23221–2008). Mature tobacco plants in uniform developmental stage and
   of high-quality were selected for analysis. Each tobacco leaf of
   uniform quality was randomly collected from the upper (U), middle (M),
   and lower (L) positions in the stem. After harvesting, the different
   leaf sets were separately flue-cured following the technical
   specification for flue-cured tobacco [[62]21]. Three special periods
   were set: the mature stage (M), the yellowing stage (Y), and the
   color-fixing stage (C). There were 18 treatment groups in total, and
   each treatment group was subjected to three biological replicates. The
   collected tobacco leaves were frozen in liquid nitrogen and kept at
   −80 °C for the downstream analyses.

Determination of browning index

   The browning index is measured by assessing the degree of browning on
   the leaf surface. The browning range index of tobacco leaves was
   obtained by calculating the proportion of the browning area of the leaf
   to the total leaf area using the Photoshop pixel method. The
   methodology was adapted from a previous study with minor adjustments
   [[63]22]. An average of five tobacco leaves was taken for each set of
   results.

Quasi-targeted metabolomics analysis

   Each sample was ground thoroughly in liquid nitrogen and 100 mg of the
   resulting powder was extracted in 80% methanol and centrifuged at
   15,000 g for 20 min at 4 °C. Subsequently, the supernatant was diluted
   with ddH[2]O to a final concentration of 53% methanol. Subsequently,
   the sample was centrifuged at 15,000 g for 20 min at 4 °C. The
   supernatant was filtered (0.22 μm, ANPEL, China) and then analyzed by
   LC-MS/MS (Novogene Co. Ltd., Beijing, China). Each biological replicate
   was subjected to two technical replicates in LC-MS/MS. The detection of
   the experimental samples using MRM (Multiple Reaction Monitoring) were
   based on novogene in-house database. The Q3 were used to the metabolite
   quantification. The Q1, Q3, RT (retention time), DP (declustering
   potential) and CE (collision energy) were used to the metabolite
   identification. These metabolites were annotated using the KEGG
   ([64]http://www.genome.jp/kegg/), HMDB ([65]http://www.hmdb.ca/) and
   Lipidmaps ([66]http://www.lipidmaps.org/).LC-MS/MS analyses were
   performed on an ExionLC™ AD system (SCIEX, US) and a QTRAP^®
   6500 + mass spectrometer (SCIEX, US). MRM and QQQ-MS were used for
   metabolite quantification. Subsequently, accumulated metabolites in the
   same leaf positions of the same tobacco cultivar were compared across
   different flue-curing treatments (e.g. DUY vs. DUM, DUC vs. DUM; YUY
   vs. YUM, YUC vs. YUM). Metabolites with |log[2] (fold change)| > 1,
   p-value < 0.05, and VIP > 1 were considered as differentially
   accumulated metabolites (DAMs).

RNA extraction, library construction, and transcriptome analysis

   The total RNA from tobacco leaves of 18 treatment groups was extracted
   using StarSpin Plant RNA Kit (GenStar, Beijing, China) according to the
   manufacturer’s protocol. Genomic DNA contamination was removed using
   DNase I (Takara, Beijing, China). Subsequently, the 2100 Bioanalyzer
   (Agilent Technologies Ltd, California, USA) and ND-2000 (NanoDrop
   Technologies, Thermo Fisher Corporation, Madison, USA) were used to
   assess RNA integrity and quality. High-quality RNA transcriptome
   libraries were prepared using the TruSeqTM RNA Sample Prep Kit
   (Illumina, Inc., San Diego, California, USA). Subsequently, RNA-seq
   sequencing was performed using the Illumina NovaSeq 6000 system
   (Novogene Co.). Clean reads were mapped to the Nicotiana tabacum L.
   genome using HISAT2
   ([67]https://solgenomics.net/ftp/tobacco_genome/edwards_et_al_2017/anno
   tation/). NR ([68]https://www.ncbi.nlm.nih.gov/protein/), KEGG, GO
   ([69]http://geneontology.org/), and other free public databases were
   used for gene function annotation. Furthermore, expressed genes in the
   same leaf positions of the same tobacco cultivar were compared across
   different flue-curing treatments (e.g. DUY vs. DUM, DUC vs. DUM; YUY
   vs. YUM, YUC vs. YUM). Differentially expressed genes (DEGs) between
   two groups were screened using DESeq2
   ([70]http://bioconductor.org/packages/stats/bioc/DESeq2/) based on
   |log2 (fold change)| >1 and p[adj] <0.05.

Quantitative real-time PCR (qRT-PCR) validation

   Eight DEGs were randomly selected, and their RNA-seq data was verified
   using qRT-PCR. Based on the sequences of the screened transcripts,
   Primer Premier 6.0 software was used to design primers to amplify the
   screened genes (Table [71]S1). The cDNA was synthesized using FastKing
   gDNA dispelling RT SuperMix (TianGen, Beijing, China). qRT-PCR
   validation was performed using Talent qPCR PreMix (TianGen, Beijing,
   China) on Roche LightCycler 480. Actin was used as an internal
   reference gene. Relative gene expression levels were calculated using
   the 2^-ΔΔCt method.

Identification of hub genes using WGCNA

   WGCNA package (version 1.71) based on R (version 4.2.2) was used to
   analyze the DEGs and corresponding DAMs in the yellowing and
   color-fixing stages for each leaf position obtained from Yunyan87 and
   Dabaijin cultivars. The DAMs significantly correlated with the
   up-regulation module in each site’s yellowing and color-fixing stages
   were extracted, and the first ten were taken from each site. Finally,
   core DAMs were identified based on the correlation and significance
   based on literature. The core DAMs were analyzed using the WGCNA
   package for co-expression network analysis of unique DEGs in Dabaijin
   and Yunyan87 to screen for genes associated with browning in flue-cured
   tobacco. The correlation between DEGs and DAMs was calculated using
   Pearson’s weighted correlation coefficient values. Hierarchical
   clustering trees were constructed using TOM-based dissimilarity
   degrees. ‘minModuleSize’ was selected as 250 for Dabaijin and Yunyan87,
   and the rest of the parameters were set to default.

Accession numbers

   The raw metabolome and transcriptome data have been submitted to the
   China National Center for Bioinformation ([72]https://www.cncb.ac.cn/),
   with accession numbers OMIX009385 and CRA023683, respectively. The raw
   transcriptome data has also been submitted to the National Center for
   Biotechnology Information ([73]https://www.ncbi.nlm.nih.gov/), with
   accession number PRJNA1293734.

Statistical analysis

   The means and standard deviations of the browning index and qRT-PCR
   values were analyzed using Excel 2016. Statistical differences between
   sample groups were analyzed by ANOVA using SPSS 25 and their
   significance was defined as * P < 0.05, ** P < 0.01, and *** P < 0.001.
   The R language online platform was used for graphics
   ([74]https://www.chiplot.online/).

Results

Phenotypic characteristics of browning

   After the Yunyan87 and Dabaijin tobacco leaves had undergone the
   mature, yellowing, and color-fixing stages of flue-curing, gradual
   changes in tobacco leaves were observed. The leaves shrank and
   shriveled, their color turned green to yellow, and browning occurred in
   some areas (Fig. [75]1A). Among them, the degree of browning of the
   leaves in all parts of Dabaijin was more significant than that of
   Yunyan87 (Fig. [76]1A, B). Moreover, the degree of browning in leaves
   of Dabaijin at the color-fixing stage was higher than that at the
   yellowing stage (Fig. [77]1B). The browning index (BI) of the upper and
   middle leaves of Dabaijin at color-fixing stage was higher than that of
   Yunyan87, reaching 36.84% and 34.95%, respectively (Fig. [78]1B).

Fig. 1.

   [79]Fig. 1
   [80]Open in a new tab

   Browning phenotypes (A) and browning indexes (B) of the upper, middle,
   and lower leaves of Yunyan87 and Dabaijin tobacco cultivars at the
   mature, yellowing, and color-fixing stages

Metabolomic analysis during the flue-curing process

   A quasi-targeted metabolomics analysis was performed for the 18 samples
   of Yunyan87 and Dabaijin to profile the metabolites during the
   flue-curing process. A total of 1207 metabolites were identified by
   LC-MS/MS, which were divided into 20 categories. The top eight
   categories were: amino acid and its derivatives, flavonoids, organic
   acid and its derivatives, carbohydrates and its derivatives, lipids,
   organoheterocyclic compounds, nucleotide and its derivatives, and
   phenolic acids, each accounting for more than 5% of the total
   metabolites (Fig. [81]2A). In addition, 767 of the identified
   metabolites were classified into 10 super-classes according to the HMDB
   4.0 database, of which organic acids and derivatives were the most
   abundant (Fig. [82]2B). Principal component analysis (PCA) showed that
   the 18 flue-cured tobacco samples had good intra-group repeatability
   (Fig. [83]2C). The metabolomic profiles of the yellowing stage and the
   color-fixing stage were overlapping but they appeared different from
   that of the mature stage, indicating that significant metabolite
   changes occurred during the curing process.

Fig. 2.

   [84]Fig. 2
   [85]Open in a new tab

   Metabolome analysis of Yunyan87 and Dabaijin tobacco leaves during the
   flue-curing process. A Classification of Class I metabolites
   identified. B HMDB database classification of metabolites. C PCA
   analysis of metabolites. Samples were labeled with three letters
   pertaining to cultivar (Yunyan87, Y; Dabaijin, D), leaf position
   (lower, L; middle, M; upper, U), and curing stage (mature stage, M;
   yellowing stage, Y; color-fixing stage, C), respectively. D Number of
   differentially accumulated metabolites (DAMs) in different comparison
   groups. E Screening of unique DAMs in the upper, middle, and lower
   leaves of Yunyan87 (blue box) and Dabaijin (red box). Enriched KEGG
   pathways in unique DAMs in Dabaijin (F) and Yunyan87 (G)

   DAMs were screened during the flue-curing process, and most of the DAMs
   were significantly upregulated at the yellowing and color-fixing stages
   compared to the mature stage (Fig. [86]2D). In the upper leaves, the
   number of DAMs identified at yellowing stage was 219 (203 up and 16
   down) in Dabaijin (DUY vs. DUM) and 350 (331 up and 19 down) in
   Yunyan87 (YUY vs. YUM); at color-fixing stage, the DAMs were 361 (325
   up and 36 down) in Dabaijin (DUC vs. DUM) and 461 (439 up and 22 down)
   in Yunyan87 (YUC vs. YUM). In the middle leaves, the DAMs at yellowing
   stage were 378 (352 up and 26 down) in Dabaijin (DMY vs. DMM) and 523
   (33 up and 490 down) in Yunyan87 (YMY vs. YMM); the DAMs at
   color-fixing stage were 452 (26 up and 426 down) in Dabaijin (DMC vs.
   DMM) and 546 (41 up and 505 down) in Yunyan87 (YMC vs. YMM). In the
   lower leaves, there were 251 (229 up and 22 down) DAMs in Dabaijin (DLY
   vs. DLM) and 349 (335 up and 14 down) DAMs in Yunyan87 (YLY vs. YLM) at
   yellowing stage, while there were 387 (369 up and 18 down) DAMs in
   Dabaijin (DLC vs. DLM) and 543 (528 up and 15 down) DAMs in Yunyan87
   (YLC vs. YLM) at color-fixing stage (Fig. [87]2D).

   The DAMs unique to Dabaijin and to Yunyan87 across leaf positions and
   flue-curing stages (294 and 528, respectively) were determined
   (Fig. [88]2E). K-means clustering of those unique DAMs resulted to five
   and seven clusters in Dabaijin and Yunyan87, respectively, each cluster
   exhibited a different expression pattern (Fig. [89]S2A, B). To better
   understand the affected metabolic pathways in which the unique DAMs
   were involved, KEGG enrichment analysis was performed. The top 30 KEGG
   pathways significantly enriched in Dabaijin-specific DAMs include amino
   sugar and nucleotide sugar metabolism; glutathione metabolism; alanine,
   aspartate and glutamate metabolism; arginine and proline metabolism;
   and ascorbate and aldarate metabolism among others (Fig. [90]2F). On
   the other hand, the enriched pathways in Yunyan87-specific DAMs include
   flavone and flavonol biosynthesis, phenylpropanoid biosynthesis, folate
   biosynthesis, lysine biosynthesis, and nicotinate and nicotinamide
   metabolism (Fig. [91]2G).

Transcriptomic analysis during the flue-curing process

   We performed transcriptome sequencing on 18 flue-cured tobacco samples
   to comprehensively analyze the potential molecular mechanisms of
   browning in Yunyan87 and Dabaijin. A total of 2.48 billion clean reads
   were obtained from 54 cDNA libraries, with an average of 45.96 million
   reads per cDNA library. The number of clean reads per library ranged
   from 39.04 million to 55.23 million, and the Q30 value ranged from
   91.98 to 94.98% (Table [92]S2). These results indicated that the
   quality and quantity of the sequencing data were suitable for further
   analysis. PCA showed that the 18 samples had good intra-group
   repeatability, with more significant differences between the mature
   stage and the yellowing stage and color-fixing stage, indicating the
   changes in gene expression during the flue-curing process
   (Fig. [93]3A).

Fig. 3.

   [94]Fig. 3
   [95]Open in a new tab

   Transcriptome analysis of Yunyan87 and Dabaijin tobacco leaves during
   the flue-curing process. A PCA analysis of metabolites. Samples were
   labeled with three letters pertaining to cultivar (Yunyan87, Y;
   Dabaijin, D), leaf position (lower, L; middle, M; upper, U), and curing
   stage (mature stage, M; yellowing stage, Y; color-fixing stage, C),
   respectively. B Number of DEGs in different comparison groups. C
   Screening of unique DEGs in the upper, middle, and lower leaves of
   Yunyan87 (blue box) and Dabaijin (red box). GO terms enriched in unique
   DEGs in Dabaijin (D) and Yunyan87 (E). KEGG pathways enriched in unique
   DEGs in Dabaijin (F) and Yunyan87 (G)

   DEGs were identified by comparing the transcriptome data. In the upper
   leaves, the number of DEGs identified at yellowing stage was 9152 (3999
   up and 5153 down) in Dabaijin (DUY vs. DUM) and 11,552 (5176 up and
   6376 down) in Yunyan87 (YUY vs. YUM); at color-fixing stage, the DEGs
   were 12,189 (5642 up and 6547 down) in Dabaijin (DUC vs. DUM) and
   12,869 (6170 up and 6699 down) in Yunyan87 (YUC vs. YUM). In the middle
   leaves, the DEGs at yellowing stage were 14,504 (7228 up and 7276 down)
   in Dabaijin (DMY vs. DMM) and 10,515 (7361 up and 7805 down) in
   Yunyan87 (YMY vs. YMM); at color-fixing stage, the DEGs were 15,166
   (7361 up and 7805 down) in Dabaijin (DMC vs. DMM) and 11,534 (6241 up
   and 5293 down) in Yunyan87 (YMC vs. YMM). In the lower leaves, the DEGs
   at yellowing stage were 9335 (4574 up and 4761 down) in Dabaijin (DLY
   vs. DLM) and 10,196 (5621 up and 4575 down) in Yunyan87 (YLY vs. YLM);
   at color-fixing stage, the DEGs were 13,560 (7116 up and 6444 down) in
   Dabaijin (DLC vs. DLM) and 12,537 (6245 up and 6292 down) in Yunyan87
   (YLC vs. YLM) (Fig. [96]3B).

   The unique DEGs in Dabaijin and Yunyan87 were collected (10792 and
   9088, respectively) (Fig. [97]3C) and analyzed by K-means clustering
   which generated two clusters for each dataset (Fig. [98]S2C). GO and
   KEGG enrichment analyses showed significant differences between
   Dabaijin and Yunyan87 (Fig. [99]3D-G). The unique DEGs in Dabaijin were
   significantly enriched in GO terms such as tubulin binding, ion channel
   activity, DNA repair, cellular response to stress, and oxoacid
   metabolic process, etc. (Fig. [100]3D), as well as in KEGG pathways
   such as valine, leucine and isoleucine biosynthesis; pantothenate and
   CoA biosynthesis; starch and sucrose metabolism; and biosynthesis of
   cofactors and arginine biosynthesis (Fig. [101]3F). However, in
   Yunyan87 that has lower browning degree, the unique DEGs were
   significantly enriched in GO terms such as coenzyme biosynthetic
   process, intracellular transport, intracellular protein transport,
   secretion, and oxidoreduction coenzyme metabolic process
   (Fig. [102]3E), and in KEGG pathways such as phosphatidylinositol
   signaling system, N-glycan biosynthesis, mismatch repair, pentose
   phosphate pathway, and MAPK signaling pathway (Fig. [103]3G).

qRT–PCR validation

   Eight DEGs were randomly selected (including Nitab4.5_0003858g0010,
   Nitab4.5_0000694g0080, Nitab4.5_0001223g0010, Nitab4.5_0000356g0010,
   Nitab4.5_0000263g0190, Nitab4.5_0000273g0080, and
   Nitab4.5_0000249g0160, Nitab4.5_0000407g0130), and their expression
   patterns were verified by qRT–PCR. The expression data obtained by
   RNA-Seq and qRT–PCR were compared (Fig. [104]4A). These genes had the
   same trends as the transcriptome results. Significant positive
   correlations were observed between the qRT–PCR and transcriptome data
   of the eight DEGs (R = 0.43), which indicated that the RNA-Seq data
   were reliable (Fig. [105]4B).

Fig. 4.

   [106]Fig. 4
   [107]Open in a new tab

   qRT-PCR validation of the eight genes in the transcriptome. A The
   relative expression levels of eight DEGs were analysed by qRT-PCR and
   Vactin was used as reference gene. The primary coordinate axis
   represents the relative expression, and the secondary coordinate axis
   represents the value of LOG[2] (FPKM + 1). B Correlation plot of the
   RNA-Seq results and qRT-PCR results. Results were calculated using
   log[2]FC variation measurements. The R^2 value represents the
   correlation between the RNA-seq and qRT-PCR results

Combined metabolomic and transcriptomic analyses

   In this study, the unique DEGs and DAMs in Dabaijin and Yunyan87 were
   enriched in several KEGG pathways (Fig. [108]5). In the two omics data
   of Daibaijin, the valine, leucine and isoleucine biosynthesis (ko00290)
   and degradation (ko00280), arginine biosynthesis (ko00220), riboflavin
   metabolism (ko00740), inositol phosphate metabolism (ko00562), and
   amino sugar and nucleotide sugar metabolism (ko00520) were enriched
   altogether. In the two omics data of Yunyan87, the galactose metabolism
   (ko00052), lysine biosynthesis (ko00300), taurine and hypotaurine
   metabolism (ko00430), and sulfur metabolism (ko00920) were enriched.
   Comprehensively, these distinct pathways between Dabaijin and Yunyan87
   might underlie the significant difference in the browning reaction
   during the flue-curing process.

Fig. 5.

   [109]Fig. 5
   [110]Open in a new tab

   Conjoint analysis of DEGs and DAMs. KEGG analysis of the metabolome and
   transcriptome in Dabaijin (A) and Yunyan87 (B). The green dotted line
   and the red dotted line are -lg (0.05) and -lg (0.01), respectively

DAMs and DEGs related to amino acid metabolism in Dabaijin

   By comparing the transcriptome and metabolome in Dabaijin leaves during
   the flue-curing process, three common KEGG pathways related to amino
   acid metabolism were identified, including the valine, leucine and
   isoleucine biosynthesis and degradation, and arginine biosynthesis
   (Fig. [111]5). Metabolome analysis found that L-leucine, L-valine,
   L-arginine, and L-ornithine, which contain free amino groups, were
   upregulated in Dabaijin during the flue-curing process at all leaf
   positions; however, L-glutamate was downregulated during the
   flue-curing process at all leaf positions. Transcriptome analysis
   revealed 27 DEGs in Dabaijin during the flue-curing process, including
   acetolactate synthase small subunit 1 (LOC107770361), ketol-acid
   reductoisomerase (LOC107759104), 3-isopropylmalate dehydrogenase
   (IMDH3), branched-chain-amino-acid aminotransferase 2 (BCAT2),
   2-oxoisovalerate dehydrogenase subunit beta 1 (BCDH),
   methylcrotonoyl-CoA carboxylase beta chain (MCCB),
   hydroxymethylglutaryl-CoA lyase (HMGCL), probable enoyl-CoA hydratase 1
   (ECHIA), 3-hydroxyisobutyryl-CoA hydrolase-like protein 5 (CHY5),
   aspartate aminotransferase, chloroplastic (ASP), glutamate
   dehydrogenase B-like (GDH), glutamine synthetase-like (GLN), amino-acid
   N-acetyltransferase (NAGS), N-acetyl-gamma-glutamyl-phosphate reductase
   (LOC107766214), acetylornithine deacetylase-like (LOC107780733),
   glutamate N-acetyltransferase (LOC107783830), arginase 1,
   mitochondrial-like (ARGAH), and ornithine carbamoyltransferase (OTC)
   genes (Fig. [112]6A). While in various parts of Dabaijin, the remaining
   DEGs showed a downregulated trend in all leaf positions during the
   flue-curing process, except for Nitab4.5_0001646g0050 (BCAT2),
   Nitab4.5_0000174g0040 (BCDH), Nitab4.5_0000252g0120 (MCCB),
   Nitab4.5_0000177g0080 (ECHIA), and novel.3761 (CHY5) which were
   upregulated in the lower and upper leaf.

Fig. 6.

   [113]Fig. 6
   [114]Open in a new tab

   Integrated analysis of the transcriptome and metabolome in Dabaijin
   leaves during the flue-curing process. A Amino acid metabolism-related
   (valine, leucine, and isoleucine biosynthesis and degradation; arginine
   biosynthesis) DAMs and DEGs. B Riboflavin metabolism-related DAMs and
   DEGs. C Amino sugar and nucleotide sugar metabolism-related DAMs and
   DEGs. D Inositol phosphate metabolism-related DAMs and DEGs. The
   numbers in each box represent the log[2]FC values of DAMs or DEGs.
   Samples were labeled with three letters pertaining to cultivar
   (Yunyan87, Y; Dabaijin, D), leaf position (lower, L; middle, M; upper,
   U), and curing stage (mature stage, M; yellowing stage, Y; color-fixing
   stage, C), respectively. Red stars represent the core DAMs of combined
   analysis in the figure

DAMs and DEGs related to riboflavin metabolism in Dabaijin

   Riboflavin metabolism was an enriched KEGG pathway shared by both
   transcriptome and metabolome of Dabaijin leaves (Fig. [115]5). Ribitol
   and vitamin B2 were upregulated in various Dabaijin leaf positions, and
   six related DEGs encoding riboflavin-biosynthesis protein pyrimidine
   reductase PYRR (PYRR), 5-amino-6-(5-phospho-D-ribitylamino) uracil
   phosphatase (PYRP), 6,7-dimethyl-8-ribityllumazine synthase
   (LOC107759548), riboflavin synthase (LOC107796682), purple acid
   phosphatase (PAP), and FAD synthetase (RIBF) were identified
   (Fig. [116]6B). PYRR, PYRP, LOC107759548, LOC107796682, and RIBF were
   downregulated in all leaf positions, with FHY being downregulated to a
   greater extent. In addition, PAP was upregulated in the lower and
   middle leaves.

DAMs and DEGs related to amino sugar and nucleotide sugar metabolism in
Dabaijin

   The amino sugar and nucleotide sugar metabolism KEGG pathway appeared
   in both the transcriptome and metabolome of Dabaijin leaves
   (Fig. [117]5). UDP-D-glucose, L-fucose, and D-glucose 1-phosphate were
   implicated in the browning of Dabaijin during the flue-curing process.
   Among them, L-fucose was mainly upregulated in middle leaves and
   D-glucose 1-phosphate was mainly downregulated in upper leaves
   (Fig. [118]6C). Meanwhile, UDP-D-glucose was downregulated in the
   middle and lower leaves. On the other hand, there were 34 related DEGs
   in Dabaijin during the flue-curing process including endochitinase A
   (CHN-A), beta-hexosaminidase 3 (HEXO3), hexokinase-2 (HXK2),
   L-arabinokinase (ARA), glucosamine-phosphate N-acetyltransferase (GNA),
   glutamine fructose-6-phosphate aminotransferase (GFAT),
   mannose-6-phosphate isomerase 2 (PMI2), galacturonokinase (GALAK),
   bifunctional fucokinase (FKGP), UDP-N-acetylglucosamine diphosphorylase
   1 (GLCNAC1PUT1), alpha-L-arabinofuranosidase 1 (ASD1),
   UTP-glucose-1-phosphate uridylyltransferase (UGP), mannose-1-phosphate
   guanylyltransferase 1 (CYT1), glucose-1-phosphate adenylyltransferase
   (ADG2), UDP-arabinose 4-epimerase (MUR4), UDP-glucuronate decarboxylase
   (UXS1), UDP-glucose 6-dehydrogenase (EC:1.1.1.22), UDP-glucuronate
   4-epimerase 1 (GAE1), GDP-L-fucose synthase (GER2), beta-xylosidase
   (BXL2), UDP-glucose 4,6-dehydratase (RHM),
   alpha-1,4-galacturonosyltransferase (GAUT), and cytochrome-b5 reductase
   (CBR1). Among them, ARA (Nitab4.5_0000970g0150 and
   Nitab4.5_0001942g0140), ASD1, EC:1.1.1.22 (Nitab4.5_0000154g0320),
   GAUT, HXK2, MUR4, PMI2, and RHM showed an increasing trend from
   yellowing to color-fixing stage in the lower and upper leaves of
   Dabaijin. In contrast, GAE1 and UXS1 (novel.4196) were upregulated in
   the middle and upper leaves. The remaining genes showed a decreasing
   trend in all leaf positions. It is worth noting that except for ASD1,
   GAE1, PMI2, and UXS1 (novel.4196), all genes were downregulated in the
   middle leaf at yellowing stage compared to the mature stage
   (Fig. [119]6C).

DAMs and DEGs related to inositol phosphate metabolism in Dabaijin

   The KEGG pathway inositol phosphate metabolism was also shared by the
   transcriptome and metabolome of Dabaijin leaves. The dihydroxyacetone
   phosphate was downregulated in the upper leaves at the color-fixing
   stage compared to the mature stage. During the flue-curing process, 34
   DEGs related to inositol phosphate metabolism were identified in
   Dabaijin leaves which include multiple inositol polyphosphate
   phosphatase 1 (EC:3.1.3.62), 1-phosphatidylinositol-3-phosphate
   5-kinase (FAB1C), myo-inositol-1(or 4)-monophosphatase (IMPL1),
   myo-inositol-1-phosphate synthase (IPS), inositol-tetrakisphosphate
   1-kinase 3 (ITPK), phosphatidylinositol-3-phosphatase myotubularin-1
   (MTM1), phosphatidylinositol 3-kinase (PI3K), phosphatidylinositol
   4-phosphate 5- kinase 6 (PIP5K6), phosphoinositide phospholipase C
   2-like (PLC2), non-specific phospholipase C4 (PLC4),
   phosphatidylinositol-3,4,5-trisphosphate 3-phosphatase,
   dual-specificity protein phosphatase PTEN (PTEN), and triosesephosphate
   isomerase (TIM). Of these, EC:3.1.3.62, IMPL1 (Nitab4.5_0003234g0080),
   MTM1, PI3K, PIP5K6 (Nitab4.5_0000494g0010), PLC4, and PTEN were found
   to be upregulated, while the FAB1C, IMPL1 (Nitab4.5_0002056g0080), IPS,
   PLC2 (Nitab4.5_0000119g0150), and TIM were found to be downregulated in
   various leaf positions of Dabaijin during the flue-curing process
   (Fig. [120]6D).

DAMs and DEGs related to amino acid metabolism in Yunyan87

   In this study, lysine biosynthesis, as a branch of amino acid
   metabolism, was also shared in transcriptome and metabolome profiles of
   Yunyan87. Metabolomics analysis revealed that L-saccharopine,
   L-2-aminoadipic acid, and 2-oxoadipic acid were upregulated, while
   L-aspartic acid was downregulated in various parts of the leaves during
   the flue-curing process. L-homoserine showed varying trends in various
   parts. Transcriptome analysis revealed four related DEGs, including the
   genes encoding for diaminopimelate decarboxylase 1 (LYSA1),
   4-hydroxy-tetrahydrodipicolinate reductase 1 (DAPB1),
   aspartate-semialdehyde dehydrogenase (EC:1.2.1.11), and diaminopimelate
   epimerase (DAPF). LYSA1 and EC:1.2.1.11 were upregulated in the lower
   and middle leaves, respectively, while DAPB1 and DAPF were
   downregulated in all leaf positions (Fig. [121]7A). The above results
   indicated that the DAMs and DEGs related to amino acid metabolism were
   markedly different between Dabaijin and Yunyan87.

Fig. 7.

   [122]Fig. 7
   [123]Open in a new tab

   Integrated analysis of the transcriptome and metabolome in Yunyan87
   leaves during the flue-curing process. A Amino acid metabolism-related
   (lysine biosynthesis) DAMs and DEGs. B Sulfur metabolism-related DAMs
   and DEGs. C Taurine and hypotaurine metabolism-related DAMs and DEGs. D
   Galactose metabolism-related DAMs and DEGs. Samples were labeled with
   three letters pertaining to cultivar (Yunyan87, Y; Dabaijin, D), leaf
   position (lower, L; middle, M; upper, U), and curing stage (mature
   stage, M; yellowing stage, Y; color-fixing stage, C), respectively

DAMs and DEGs related to sulfur metabolism and taurine and hypotaurine
metabolism in Yunyan87

   In this study, the sulfur metabolism was shared in transcriptome and
   metabolome profiles of Yunyan87. Succinic acid, 2-aminoethanesulfonic
   acid, L-homoserine, and L-cysteine involved in sulfur metabolism were
   especially changed during browning in Yunyan87. Among them, L-cysteine
   was downregulated in the lower and upper leaves. Meanwhile,
   2-aminoethanesulfonic acid was upregulated during the flue-curing
   process. In addition, L-homoserine and succinic acid were upregulated
   in the lower leaves and downregulated in the upper leaves. Relevant to
   these metabolites were the 11 relevant DEGs during the flue-curing
   process in Yunyan87 leaves, including ATP sulfurylase 2 (APS2),
   histidine triad nucleotide-binding protein 3 (HINT3),
   3’-phosphoadenosine 5’-phosphosulfate synthase (APK),
   5’-adenylylsulfate reductase 3 (APR3), PAP-specific phosphatase HAL2
   (AHL), serine acetyltransferase 1 (SAT1), and cysteine synthase (CYS).
   Among them, HINT3 and APK (Nitab4.5_0002119g0120 and
   Nitab4.5_0000394g0020) were upregulated during the flue-curing in all
   leaf positions. Conversely, APK (Nitab4.5_0001079g0180), AHL, and CYS
   were downregulated in all leaf positions during the flue-curing
   process. Furthermore, APR3 and SAT1 exhibited different expression
   trends during the flue-curing of different leaves (Fig. [124]7B).

   Taurine and hypotaurine metabolism, an important branch of sulfur
   metabolism, was also notable in Yunyan87. The metabolites L-alanine,
   L-cysteinesulfinic acid, guanidinoethyl sulfonate, and
   2-aminoethanesulfonic acid as well as the genes GGT4 and GAD1 were
   upregulated during the flue-curing process. Conversely,
   Nitab4.5_0000783g0240 was downregulated in the lower and upper leaves
   and upregulated in the middle leaves (Fig. [125]7C).

DAMs and DEGs related to galactose metabolism in Yunyan87

   During the flue-curing process in Yunyan87 leaves, the accumulation of
   raffinose, stachyose, dulcitol, D-fructose 6-phosphate, lactose,
   galactinol, sucrose, D-glucose, and D-sorbitol was observed. Among
   them, the non-reducing sugars are raffinose, stachyose, galactinol,
   sucrose, D-sorbitol, and dulcitol. The DEGs associated with these
   metabolites include alpha-galactosidase 3 (AGAL3), aldose 1-epimerase
   (GAL), UTP-glucose-1-phosphate uridylyltransferase (UGP),
   phosphoglucomutase (PGMP), phosphoglucomutase (HXK2),
   beta-fructofuranosidase (CWINV1), alpha-galactosidase 3 (AGAL3),
   ATP-dependent 6- phosphofructokinase 6 (PFK6). AGAL3, PGMP, HXK2, and
   AGAL3 were downregulated in the upper leaves while GAL was
   downregulated in all leaf positions. In contrast, PFK6, UGP, and CWINV1
   (Nitab4.5_0001780g0120) were upregulated in the lower leaf
   (Fig. [126]7D).

Identification of key KEGG pathways related to Browning by WGCNA

   The WGCNA screened eight core DAMs associated with the browning process
   in Dabaijin, namely, L-glutamine (Com_412_neg), O-acetylserine
   (Com_7_neg), homocysteic acid (Com_592_pos), valine (Com_ 686_pos),
   naringerin (Com_796_pos), Vitamin A (Com_1029_pos), D-xylonic acid
   (Com_466_neg), and vanillic acid (Com_276_neg) (Fig. S4, Table S3). The
   eight core DAMs were then correlated with the unique DEGs in Dabaijin
   and Yunyan87, generating six and five gene modules, respectively (Fig.
   S5). The turquoise module eigenvalues of both Dabaijin and Yunyan87
   were significantly and positively correlated with all eight core DAMs
   (Fig. [127]8A, B). The turquoise modules of Dabaijin and Yunyan87
   contained 2869 and 2255 DEGs, respectively (Fig. [128]8A-D). The DEGs
   in the turquoise module of Dabaijin were enriched in nucleotide
   excision repair, inositol phosphate metabolism, arginine and proline
   metabolism, cysteine and methionine metabolism, and oxidative
   phosphorylation, among other KEGG pathways (Fig. [129]8E). Conversely,
   the DEGs in the turquoise module of Yunyan87 were enriched in oxidative
   phosphorylation, MAPK signaling pathway, phosphatidylinositol signaling
   system, ABC transporters, and plant hormone signal transduction, among
   other KEGG pathways (Fig. [130]8F).

Fig. 8.

   [131]Fig. 8
   [132]Open in a new tab

   WCGNA analysis of unique DEGs in Dabaijin and Yunyan87 with eight core
   DAMs. A Module–trait relationships of Dabaijin. B Module–trait
   relationships in Yunyan87. The color and size of circles in A and B
   represent the correlation value and P-value. Significantly upregulated
   turquoise module in Dabaijin (C) and in Yunyan87 (D). KEGG pathway
   enrichment analysis of turquoise module genes in Dabaijin (E) and in
   Yunyan87 (F). Samples were labeled with three letters pertaining to
   cultivar (Yunyan87, Y; Dabaijin, D), leaf position (lower, L; middle,
   M; upper, U), and curing stage (mature stage, M; yellowing stage, Y;
   color-fixing stage, C), respectively

Identification of hub genes associated with oxidative phosphorylation by
WGCNA

   Based on the above WGCNA results, the oxidative phosphorylation pathway
   shared by Dabaijin and Yunyan87 was further analyzed to screen for
   potential key hub genes that might be implicated for the browning
   differences between the two tobacco cultivars. The results showed that
   14 DEGs in Dabaijin were notable during the flue-curing process, five
   of which encode for NADH dehydrogenases (ND5, Ndufa1, Ndufa9, Ndufab1
   and Nitab4.5_0001170g0160), four encode for cytochrome c oxidases
   (COX10, COX17, Nitab4.5_0000999g0020, and Nitab4.5_0002546g0050) and
   five encode for ATP synthases (ATPB, Nitab4.5_0001584g0020, VHA-A,
   VHA-E1, PPA). Except for ND5 which was upregulated higher in Dabaijin
   than in Yunyan87 middle leaf, the rest of the DEGs in Dabaijin were not
   significantly different from those in all leaf positions in Yunyan87
   (Fig. [133]9).

Fig. 9.

   [134]Fig. 9
   [135]Open in a new tab

   WGCNA identifies DAMs and DEGs related to oxidative phosphorylation.
   Red stars represent the selected key genes in oxidative phosphorylation
   based on Log[2]FC values. Samples were labeled with three letters
   pertaining to cultivar (Yunyan87, Y; Dabaijin, D), leaf position
   (lower, L; middle, M; upper, U), and curing stage (mature stage, M;
   yellowing stage, Y; color-fixing stage, C), respectively

   In Yunyan87 leaves, 27 relevant DEGs were identified, comprising of 13
   DEGs that encode for NADH dehydrogenases (ND1, ND2, ND5, ND6, Ndufa9,
   Nitab4.5_0000127g0050, and Nitab4.5_0001034g0060), one DEG that encodes
   for a cytochrome c reductase (Nitab4.5_0000120g0110), six DEGs that
   encode for cytochrome c oxidase (COX1, COX6B, Nitab4.5_0001057g0180,
   Nitab4.5_0003120g0030 and Nitab4.5_0003013g0040), and seven DEGs that
   encode for ATP synthases (ATPA, ATPB, VHA-E1, AHA,
   Nitab4.5_0000435g0090 and Nitab4.5_0000172g1000). The upregulation of
   the above DEGs in all Yunyan87 leaf positions during the flue-curing
   process was generally much higher than those of Dabaijin (Fig. [136]9).
   Meanwhile, the number of DEGs enriched in oxidative phosphorylation was
   also higher in Yunyan87 than in Dabaijin, indicating that oxidative
   phosphorylation was the most important factor influencing the browning
   differences between the two tobacco cultivars during the flue-curing
   process. Based on the kWithin values of Yunyan87 genes and their
   expression within the module (Table S4), 12 hub genes were identified
   for subsequent studies, including ND5 (Nitab4.5_0000085g0160 and
   Nitab4.5_0001152g0250), ND6 (Nitab4.5_0001152g0030,
   Nitab4.5_0001127g0280, and Nitab4.5_0000172g0300), E 7.1.1.8
   (Nitab4.5_0000120g0110), COX1 (Nitab4.5_0000161g0030), E 7.1.2.2
   (Nitab4.5_ 0000435g0090 and Nitab4.5_0000172g1000), and the genes
   encoding for the unknown enzymes E 7.1.1.2 (Nitab4.5_0000127g0050 and
   Nitab4.5_0000134g0060) and E 7.1.1.9 (Nitab4.5_ 0003013g0040)
   (Fig. [137]9, Table S4).

Discussion

Accumulation of Maillard reaction substrates promoted Browning

   Browning affects the quality of flue-cured tobacco leaves which causes
   significant economic losses [[138]2, [139]23]. Studies have shown that
   the accumulation of melanoidins, a product of the Maillard reaction, is
   one of the significant causes of browning formation during flue-curing
   and processing [[140]24]. In this study, Dabaijin and Yunyan87 tobacco
   cultivars had different accumulation patterns of amino acid metabolites
   during the flue-curing process, such that the amino acids were more
   abundant in Dabaijin than in Yunyan87. In addition, the amino acids
   accumulated in Dabaijin due to the downregulation of genes involved in
   amino acid metabolism (Figs. [141]6A and [142]7A). L-arginine,
   L-ornithine, L-leucine, and L-valine accumulated in Dabaijin, but
   L-saccharopine and L-2-aminoadipic acid accumulated in Yunyan87.
   Previous studies have shown that the type of amino acids in the
   Maillard reaction influences the degree of browning [[143]25, [144]26].
   In cigars, L-glutamate, L-arginine, and L-ornithine are the primary
   amino acids that promote the browning reaction in tobacco [[145]5].
   Consistently, amino sugar and nucleotide sugar metabolism were enriched
   in Dabaijin. Our results showed that L-fucose accumulated, while
   UDP-glucose and its D-glucose 1-phosphate were depleted in the lower
   leaves during the flue-curing process. Notably, most of the genes in
   amino sugar and nucleotide sugar metabolism, such as CBR1, CHN-A, and
   HEXO3, were downregulated in overall, indicating that the accumulation
   of DAMs may be a combination of the downregulation of related genes and
   the occurrence of flue-curing stress (Fig. [146]6C). Studies have shown
   that L-fucose and glucose can be involved in the Maillard reaction to
   promote melanoidin production [[147]27, [148]28]. Likewise, UDP-glucose
   and the glucose product of its D-glucose 1-phosphate hydrolysis were
   also involved in the Maillard reaction. In addition, we speculated that
   the accumulation of Maillard substrates was also a manifestation of the
   imbalance of sugar and amino acid metabolism. This metabolic imbalance
   may indirectly weaken the ability of cells to maintain energy and redox
   homeostasis. Hence, these types of amino and nucleotide sugars and
   amino acids accumulated in Dabaijin leaves and provided abundant
   Maillard substrates, promoting the browning reaction during the
   flue-curing process.

   In addition, our results showed that galactose metabolism was enriched
   in Yunyan87 leaves, with the accumulation of raffinose, stachyose,
   dulcitol, D-fructose 6-phosphate, lactose, galactinol, sucrose,
   D-glucose, and D-sorbitol in each leaf position during the flue-curing
   process. The occurrence of more non-reducing sugars and derivatives
   (i.e., raffinose, stachyose, galactinol, sucrose, D-sorbitol, dulcitol)
   (Fig. [149]7D) might have caused the lower degree of browning in
   Yunyan87, as non-reducing sugars have high stability and unique
   properties that buffer the Maillard reaction [[150]29, [151]30].

The antioxidant system plays an important role in Browning

   Plants have developed several effective antioxidant defense mechanisms
   to scavenge ROS that are induced by adverse environmental conditions.
   These include enzymatic and non-enzymatic systems such as the
   ascorbate-glutathione (AsA-GSH) cycle and thioredoxin reductase (Trx)
   [[152]31]. Riboflavin metabolism was enriched in Dabaijin leaves, and
   ribitol and vitamin B2 were upregulated. Conversely, the FAD and
   FMN-related synthesis genes PYRP and RIBF have decreased expression,
   while the FMN metabolic gene PAP showed induced expression
   (Fig. [153]6B). Concurrently, the AsA-GSH precursor D-glutamate showed
   decreased expression (Fig. [154]6A). Studies have shown that FAD is a
   cofactor for antioxidant enzymes such as glutathione reductase (GR) and
   TrxR which remove ROS by maintaining the reduced state of GSH and Trx
   [[155]32, [156]33]. The reduced levels of GSH and Trx in Dabaijin
   leaves resulted in the accumulation of ROS during the flue-curing
   process. In Yunyan87, however, sulfur metabolism as well as taurine and
   hypotaurine metabolism were enhanced, in conjunction with a decrease in
   L-cysteine level and accumulation of L-cysteine derivatives
   L-cysteinesulfinic acid, guanidinoethyl sulfonate, and
   2-aminoethanesulfonic acid (taurine). Concurrently, the key taurine
   synthesis gene GAD1 was upregulated in all leaf positions
   (Fig. [157]7C). Taurine has been shown to have potent antioxidant
   activity, which can reduce the accumulation of ROS by directly
   scavenging hydroxyl radicals and inhibiting PPO-mediated enzymatic
   browning [[158]34]. Therefore, the accumulation of taurine in Yunyan87
   leaves during the flue-curing process likely contributed to the
   modulation of browning by scavenging ROS.

   ROS is released along with hydrogen carriers through the electron
   transport chain, and oxidative phosphorylation is an important energy
   production pathway that consumes reduced hydrogen and e^-, and produces
   ATP [[159]35, [160]36]. In this study, the number and upregulation of
   NADH dehydrogenase, cytochrome c reductase, cytochrome c oxidase, and
   ATP synthase-related DEGs were higher in Yunyan87 than in Dabaijin
   (Fig. [161]9). Studies have shown that GR24 treatment accelerates the
   browning of the pericarp of Litchi chinensis Sonn, and it is positively
   correlated with the downregulation of oxidative phosphorylation-related
   genes [[162]37]. Exogenous AT-MAP treatment affects the expression of
   specific genes in the ETS pathway, especially upregulating the
   expression of ATP synthase, thereby reducing postharvest browning of
   Agaricus bisporus [[163]38]. In addition, browning in Lactuca sativa L
   [[164]39]., Flammulina filiformis [[165]40], and Vitis vinifera
   [[166]41] was correlated with the expression of genes/proteins related
   to oxidative phosphorylation. The upregulation of oxidative
   phosphorylation expression helps to consume more reduced hydrogen
   through the respiratory chain and reduces the chance of electron
   leakage to produce ROS [[167]35, [168]42, [169]43]. The excess reduced
   hydrogen and e^- consumed through oxidative phosphorylation in Yunyan87
   during the flue-curing process, might decrease ROS accumulation and
   thus, reduce browning.

   We also observed that the MAPK signaling pathway and
   phosphatidylinositol signaling system were enriched in unique DEGs
   (Fig. [170]3G) and in WGCNA gene modules (Fig. [171]8F) in Yunyan87.
   MAPK signaling pathway can regulate growth and development, hormonal
   response, and stress responses by an extremely conserved network
   [[172]44, [173]45]. Studies have shown that after being activated by
   stress, the MAPK signaling pathway activates downstream antioxidant
   defense systems (GSH, Trx, superoxide dismutase, etc.) through a
   cascade reaction that reduces ROS accumulation, thereby inhibiting cell
   damage and browning caused by oxidative stress [[174]46–[175]48].
   Concretely, MEKK1, MPK4, and MPK6 can regulate the expression of CAT1
   and CAT2 to overcome oxidative stress, indicating that MAPKs could
   control ROS and arbitrate oxidative stress in plants [[176]49,
   [177]50]. Phosphatidylinositol signaling system can be involved in
   membrane lipid metabolism and damage repair, maintaining cell membrane
   integrity and reducing browning [[178]51, [179]52]. In addition,
   phosphatidylinositol signaling may also stabilize the membrane
   structure by affecting calcium signaling or specific membrane lipid
   remodeling [[180]53–[181]56]. Thus, the MAPK signaling pathway and
   phosphatidylinositol signaling system in Yunyan87 were firstly
   activated during the flue-curing process which then triggers the
   antioxidant defense system that maintains membrane homeostasis, thereby
   slowing browning.

   ROS accumulation can lead to biofilm peroxidation, change the
   localization of the Maillard reaction substrates, PPO, and oxidizing
   enzymes in the plastid and other organelles, and cause non-enzymatic
   and enzymatic browning reactions [[182]57–[183]59]. In this study,
   inositol phosphate metabolism was enriched in Dabaijin leaves. Among
   these, PI3K, MTM1, and PIP5K6 which were involved in the synthesis of
   PI/PIP2, were upregulated, whereas, PLC2 which was involved in the
   decomposition of PIP2, was downregulated (Fig. [184]6D). Studies have
   shown that PI/PIP2 is an essential component for maintaining the
   integrity of the biofilm, i.e., its metabolism affects the stability of
   the biofilm [[185]60, [186]61]. Furthermore, PIP2 can be hydrolyzed by
   PLC to generate IP[3] and DAG, with IP[3] activating calcium channels
   to release intracellular Ca^2+, leading to an increase in Ca^2+
   concentration and inhibiting PPO to accelerate enzymatic browning
   [[187]62–[188]65]. We speculated that the accumulation of ROS during
   the flue-curing process accelerated the destruction of the membrane
   stability in Dabaijin leaves, and that the downregulation of PLC2
   reduced the Ca^2+ concentration which also accelerated the enzymatic
   browning.

Conclusions

   This study investigates the transcriptional and metabolomic mechanisms
   underlying the contrasting browning responses observed in two
   flue-cured tobacco cultivars, Yunyan87 and Dabaijin. The significant
   accumulation of Maillard reaction substrates (L-arginine, L-ornithine,
   L-leucine, L-valine, and L-fucose) in Dabaijin leaves during the
   flue-curing process promoted non-enzymatic browning. Concurrently,
   membrane homeostasis and antioxidant system function were impaired,
   further weakening the ability of the leaf to remove ROS which, in turn,
   promoted enzymatic browning. Conversely, the MAPK signaling pathway and
   phosphatidylinositol signaling were activated in Yunyan87 to regulate
   the antioxidant system and slow down membrane damage. Moreover, the
   efficiency of oxidative phosphorylation was enhanced in Yunyan87, which
   likely consumed the reduced hydrogen and electrons thereby decreasing
   ROS accumulation. Taken together, the accumulation of Maillard
   substrates and the efficacy of the antioxidant system and membrane
   homeostasis underlie the differential regulation of the browning
   process between Yunyan87 and Dabaijin tobacco cultivars (Fig. [189]10).
   These findings provide a key theoretical basis for the targeted
   improvement of tobacco processing technology and for breeding tobacco
   cultivars with improved resistance to browning.

Fig. 10.

   [190]Fig. 10
   [191]Open in a new tab

   Model of browning mechanisms during the flue-curing process of Yunyan87
   and Dabaijin tobacco leaves. In Dabaijin, Maillard reaction substrate
   accumulation promotes non-enzymatic browning. Concurrently, impaired
   membrane homeostasis and antioxidant function reduce ROS -scavenging
   capacity, exacerbating enzymatic browning. Conversely, Yunyan87
   activates the MAPK and phosphatidylinositol signaling pathways,
   enhancing antioxidant defense and mitigating membrane damage. Further,
   upregulated oxidative phosphorylation consumes reducing equivalents
   (hydrogen/electrons), limiting ROS accumulation. Therefore, the
   combined effects of Maillard substrate levels and the efficacy of
   antioxidant systems determine the distinct browning outcomes between
   Dabaijin and Yunyan87 cultivars

Supplementary Information

   [192]Supplementary Material 1.^ (31.2KB, docx)
   [193]Supplementary Material 2.^ (827.4KB, docx)

Acknowledgements