Tissue-specific accumulation of diterpene alkaloids in Aconitum japonicum

Specialized metabolites accumulate in specific tissues to fulfill distinct functions under various developmental and stress conditions [29, 47]. To capture diversity and distribution of phytochemical in A. japonicum, we performed untargeted metabolite profiling using a high-resolution UPLC–QTOF-MS/MS for four tissues, namely, leaf, mother root, daughter root, and rootlet (Fig. 1, Figure S1). Metabolite profiling was performed in MS1 and data-dependent MS/MS mode, and the acquired raw data were pre-processed and used for peak detection and alignment using MS-DIAL as described previously [34, 35]. A total of 2,045 mass features were detected in the positive-ion mode, enabling multivariant data analysis and MS/MS-based structure prediction (Table S1). Unsupervised principal component analysis (PCA) revealed tissue-wise classification for metabolite content of four tissues of A. japonicum along the PC1 and PC2 axes. Notably, leaf samples exhibited separation from the other tissues along the PC1-axis, while different root types were distinguished along the PC2-axis (Fig. 2A). These findings suggest that the metabolomes of the selected tissues effectively encompass a diverse and distinct array of bioactive compounds across its tissues, thereby underpinning the therapeutic potential possessed by the different tissues of A. japonicum.

Fig. 1

Overview of the experimental design and opted multi-omics analysis pipeline to establish high-quality omics resources for Aconitum japonicum. (A) Four tissues of A. japonicum were used for integrative omics study, including leaf (a), mother root (b), daughter root (c), and rootlet (d). The scale bar represents 1 cm. (B) Schematics of the opted multi-omics analysis pipeline used for the metabolome, transcriptome, and their integration analysis to identify potential candidate genes associated with diterpene alkaloids biosynthesis in A. japonicum

Fig. 2

Metabolome resource of Aconitum japonicum established using untargeted metabolomic approach. (A) Unsupervised principal component analysis (PCA) using the metabolome datasets acquired for the four tissues of A. japonicum. (B) Top ten KEGG pathways based on the number of assigned metabolites. The mass features were mapped to the KEGG database, and the chemical identities were assigned based on the m/z similarity with a mass-error window of ± 10 ppm. Further, the KEGG pathways related to the assigned metabolites were extracted, the top ten of which are shown here. Abbreviations: LF (leaf), MR (mother root), DR (daughter root), RT (rootlet)

In an attempt to associate the metabolome datasets with comprehensive chemical databases and biochemical pathway information, we performed mapping using the KNApSAcK [48] and KEGG [49] databases for the acquired mass features. We assigned 533 and 123 mass features to KNApSAcK and KEGG databases, respectively (Table S2). The top ten pathways, ranked by the number of mapped mass features, are shown in Fig. 2B. Notably, the top three KEGG pathways included “Biosynthesis of Plant Secondary Metabolites”, “Biosynthesis of Alkaloids Derived from Shikimate Pathway”, and “2-Oxocarboxylic Acid Metabolism”. The untargeted metabolome profiling and MS/MS-based metabolite annotation based on pathway mapping unveiled a plethora of metabolites, including flavonoids, phenylpropanoids, and anthocyanins, in addition to the known diterpene alkaloids within the metabolite space of A. japonicum. Comparison of our findings with established databases like KEGG and KNApSAcK revealed a significant overlap between the detected metabolites and known Aconitum secondary metabolites, thereby validating our analytical approach. Recently, a comprehensive untargeted metabolome profiling for over 150 plant species was published as RefMetaPlant, which included A. carmichaelii as one of the plant species [50]. The database reported 1,102 metabolites from A. carmichaelii. For A. japonicum, we identified over 2,000 mass features, with over 500 mass features being mapped to the KNApSAcK database. The challenges in terms of assigning chemical identities to a given mass feature is limited availability of MS/MS spectral datasets; nevertheless, the identified metabolites in this study were relative to what has been previously reported for other plant species. Comparing spectral profiles using the metabolome of A. japonicum will be useful to identify intermediates associated with aconitine biosynthesis. Additionally, tissue-wise accumulation trends offer key insights on the function of a metabolite toward plant physiology, which would be possible using the established metabolome resource of A. japonicum from this study. Moreover, the metabolite diversity reported in this study lay the groundwork for future investigations focused on discovery and structural validation of the bioactive metabolites and their corresponding pathways in A. japonicum.

As a proof of concept to explore the established metabolome resource of A. japonicum, we used a combination of manual inspection and computational cheminformatics approach using MS-FINDER [51], and confirmed 160 metabolites using their MS/MS fragmentation spectra (Table S3). Among these annotated metabolites, we identified 51 diterpene alkaloids, with 40 belonging to the C19-type, known to be the predominant alkaloids of the toxic Aconitum species. The accumulation-based correlation analysis across four tissues of the MS/MS confirmed alkaloids revealed the presence of four distinct clusters (Fig. 3A). Cluster 1, the biggest cluster with 36 diterpene alkaloids, exhibited high accumulation in the mother root and included the majority of the C19-type alkaloid, such as aconitine, mesaconitine, jesaconitine, and hypaconitine, among others (Fig. 3B). The medicinal efficacy of A. japonicum can be directly linked to the presence of C19-type diterpene alkaloids. Therefore, the enhanced accumulation of these metabolites in the mother root, the tissue primarily used for medicinal purposes, reinforces the robustness of our metabolome data and the analysis pipeline. Cluster 2, comprising two alkaloids, namely macrocentrine and piepunensine A, were highly accumulated in rootlets, followed by the mother root (Fig. 3C). Cluster 3 represented ten diterpene alkaloids, primarily of C20-type, including dolaconine, delelatine, and yunaconitine, among others, exhibited high accumulation in the daughter root (Fig. 3D). It is hypothesized that C20-type diterpene alkaloids are the precursor for the C19- and C18-type alkaloid found in Aconitum species [52]. Therefore, the high accumulation of C20-type diterpene alkaloids in the daughter root may signal its readiness for detachment, poised to serve as the mother root for the next propagation. Cluster 4, comprising three metabolites, namely dehydroacosanine, 10-hydroxytalatizamine, and 9-deoxyglanduline, exhibited the highest accumulation in the leaves of A. japonicum (Fig. 3E). While 10-hydroxytalatizamine and 9-deoxyglanduline have been previously isolated from the aerial parts of plants from the Ranunculaceae family [53, 54], dehydroacosanine has been reported to be present in the whole plant metabolic extract of A. barbatum [55]. Therefore, the high accumulation of metabolites from cluster 4 in the leaves, as identified in our study, aligns with prior reports that had hinted at their presence in the aerial part. This consistency in our results with other plants from the Ranunculaceae plants demonstrates the potential role of these phytochemicals in plant physiology, including their potential effects on predators.

Fig. 3

Correlation analysis and accumulation pattern for MS/MS fragmentation based annotated alkaloids in Aconitum japonicum. Pearson correlation coefficients were calculated between chemically assigned alkaloids using accumulation levels across four tissues of A. japonicum, and correlation scores are plotted as a heatmap with corresponding alkaloid names represented along the X- and Y-axes. Hierarchical clustering based on correlation scores for alkaloids formed four distinct groups, labeled as 1–4, and the accumulation levels are shown as metabolite clusters (B–E) across four tissues of A. japonicum. Abbreviations: LF (leaf), MR (mother root), DR (daughter root), RT (rootlet)

Our metabolome resource data, thus established, effectively captured the overall metabolite diversity in A. japonicum, with diterpene alkaloids emerging as the key metabo-constituents. The results reaffirmed the abundance of C19-type diterpene alkaloids in toxic Aconitum species. Additionally, we observed high accumulation of diterpene alkaloids in both the mother and daughter roots, consistent with these tissues being primarily used for medicinal purposes [5]. The tissue-specific accumulation of key metabolites in A. japonicum suggests tissue-wise biosynthesis and a higher order and unexplored transport mechanisms for phytochemical as key player in shaping this metabo-phenotype. Therefore, tissue-wise expression analysis and gene-metabolite association analysis could reveal candidates involved in the biosynthesis of diterpene alkaloids in A. japonicum. Furthermore, results from this study, along with the comprehensive metabolome established herein, will undoubtedly help in further investigations and structural validation of metabolites of interest in Aconitum species.

Homology-based functional characterization for de novo transcriptome assembly for Aconitum japonicum

In the absence of genomic resources of A. japonicum, we performed deep coverage RNA-seq-based transcriptome profiling using the same tissues as used for metabolome analysis. Total RNA was extracted from the four tissues, and the corresponding cDNA libraries were sequenced using Illumina HiSeq™ 2000 platform. Subsequently, the sequencing reads were pre-processed to remove adapter sequences, low-quality reads, and shorter reads (< 50 base pairs), resulting in 44,107,721 clean paired-end reads. The average Phred score, which is a measure of the quality of the sequence reads, for each library exceeded 36. Preprocessing RNA-seq datasets using Trimmomatic program [38] resulted in < 1% of the total sequence data being dropped, indicating the suitability of our RNA-sequencing for deriving de novo transcriptome assembly (Table S4). Subsequently, pre-processed RNA-seq reads, thus obtained from individual tissue, were pooled together to derive de novo transcriptome assembly using Trinity program v.3.0 [39], followed by application of CD-HIT-EST [40] to remove sequence redundancies. The resulting de novo transcriptome assembly of A. japonicum consists of 129,760 transcripts, with an average length and median length of 600 and 368 bps, respectively, and an N50 value of 811bps (Table 1). The length of the assembled transcripts varied from 224 to 15,777 bps, with sequence lengths of 19,288 and 1,261 unigenes greater than 1000 and 3000 bps, respectively (Figure S2A). SSRs’ identification driven by transcriptome assemblies has been successfully used to validate polymorphisms across multiple genotypes in different plant species [56,57,58]. We used de novo transcriptome assembly of A. japonicum and identified 18,572 SSRs candidates, with 2,092 transcripts containing more than one SSRs (Figure S2B, Table S5). The results would be useful to develop EST-SSR markers of A. japonicum for determining genetic variations in establishing A. japonicum genotypes for medicinal purposes and its sustainable production.

Table 1 Transcriptome assembly statistics for A. japonicum

The de novo transcriptome assembly of A. japonicum were subjected to a Blastx [59] search against the NCBI non-redundant (nr) database (http://www.ncbi.nlm.nih.gov, 8 May 2020, date last accessed; formatted on February 2022) for sequence homology-based annotation and characterization. Blastx results showed the majority of the transcripts having significant homology with their corresponding matched sequences in the database (Figure S3A, Table S6). The top-hit obtained for each query was used for the transcripts’ annotation. In total, 59,691 transcripts were annotated, with 43,152 transcripts sharing above 70% sequence similarity with their corresponding Blastx search hit (Figure S3B, Table S6). Further, 54% of the transcripts of A. japonicum remained unannotated, which potentially include novel, uncharacterized genes, or isoforms lacking closely related sequences in the database (Figure S3C, Table S6). Species distribution plot based on Blastx-top-hit identified Vitis vinifera, Populus trichocarpa, and Ricinus communis as the top three species exhibiting high-sequence similarity with the assembled transcripts of A. japonicum (Figure S3D). Coptis japonica, a member of Ranunculaceae family as A. japonicum, was also among the top-hit species showing high-sequence similarity with the assembled transcripts. Annotation distribution results further highlight the limited genomic availability for plants producing similar metabolite classes as Aconitum.

For a comprehensive characterization of the transcriptome assembly, annotations from Blastx search against NCBI-nr database were further analyzed using OmicsBox software v3.1.2 (https://www.biobam.com/) to assign functional descriptions, Enzyme Commission (EC) numbers, Gene Ontology (GO) terms, and KEGG pathways. GO terms were assigned to 48,756 transcripts across three broad categories: biological processes, metabolic processes, and cellular components. Within the biological process GO category, the top GO terms included organic substance metabolic process, primary metabolic process, cellular metabolic process, biosynthetic process, and nitrogen compound metabolic process (Fig. 4A, Table S6). The molecular function category comprised of heterocyclic compound binding, organic cyclic compound binding, transferase activity, small molecule binding, and hydrolase activity as the top five GO terms. In the cellular component category, intracellular, intracellular part, intracellular organelle, membrane-bound organelle, and cell periphery represented the most abundant GO term. To gain insights into the metabolic pathways being present in A. japonicum, the assembled transcripts were mapped to the KEGG database, resulting in 8,349 transcripts assigned to 154 KEGG pathways. The top ten pathways based on the number of assigned transcripts are shown in Fig. 4B. Among these pathways, 1136 transcripts were mapped to starch and sucrose metabolism, 970 transcripts to purine metabolism, and 515 transcripts were mapped to cysteine and methionine metabolism. Additionally, 271 transcripts were assigned to isoquinoline alkaloid biosynthesis, while 128 and 59 transcripts were mapped to terpenoid backbone biosynthesis and diterpenoid biosynthesis, respectively.

Fig. 4

Transcriptome analysis across four tissues of Aconitum japonicum (A) Gene ontology (GO) annotation for A. japonicum transcriptome assembly into three major classifications, namely, BP (biological processes), MF (molecular functions), and CC (cellular components) based on Blast2GO-based analysis. (B) Top ten KEGG pathways based on the number of transcripts being assigned. (C) Unsupervised principal component analysis using transcriptome datasets generated for four tissues of A. japonicum. (D) Venn diagram representing the overall distribution of transcripts (FPKM > 0) across four tissues of A. japonicum. Abbreviations: LF (leaf), MR (mother root), DR (daughter root), RT (rootlet)

GO analysis categorized a significant proportion of the assembled transcripts across well-established functional groups, offering important insights into the biological processes, molecular functions, and cellular components being represented within the transcriptome of A. japonicum. Further, KEGG pathway mapping revealed extensive coverage of key metabolic pathways being expected, thus, highlighting the capability of de novo transcriptome assembly of A. japonicum in capturing its essential molecular mechanisms. These results demonstrate that the quality of the A. japonicum transcriptome is highly comparable to that of other non-model plant species [11, 32, 33, 37], thus providing a solid foundation to explore its genetic and metabolic landscapes.

Expression analysis identifies potential candidate genes associated with diterpene alkaloid biosynthesis in Aconitum japonicum

Within a plant tissue, transcriptionally active genes encompass both shared transcripts across tissues responsible for general growth and development, and distinct sets of active transcripts exclusive to that tissue, which play a pivotal role in facilitating specialized tissue functions, including biosynthesis, regulation, and transportation of specialized metabolites [37]. To gain a comprehensive view of active biological processes and the associated transcripts within and across the four tissues of A. japonicum, we utilized the RSEM program [42] to determine transcript expression. Subsequently, we performed PCA using expression abundance dataset to elucidate overall relationship between four tissues of A. japonicum based on active transcripts expression levels. PCA plot showed a distinct separation of the four tissues into well-defined groups. While the leaf and different root types showed separation along the PC1-axis, accounting for 79% variance, different root types were separated along the PC2-axis with 15% variance (Fig. 4C). Further, PCA for transcriptome datasets showed a similar clustering pattern as that of for metabolome datasets, suggesting a strong correlation between the transcript’s expression and metabolite accumulation. PCA analysis for four tissues of A. japonicum, therefore, clearly suggests the transcripts abundance dataset consists of tissue-specific signature transcripts, and the overlap of transcripts expression across tissues can be related to the tissue-type under investigation. Among the four tissues of A. japonicum, leaf, mother root, daughter root, and rootlet, we identified 96,788, 106,960, 98,997, and 79,662 transcriptionally active transcripts (FPKM > 0), respectively (Fig. 4D). While 55,560 transcriptionally active transcripts were shared across the four tissues, 6,466, 3,514, 2,831, and 3,376 transcripts were tissue-specifically and expressed in the leaf, mother root, daughter root, and rootlet, respectively (Table S7). Moreover, a total of 64,295 active transcripts were shared between different root types. In particular, the mother root and daughter root, the two most important tissues of A. japonicum for its medicinal properties and commercial values, exhibited a significant overlap with 88,480 active transcripts shared between them. KEGG pathway enrichment analysis using highly expressed transcripts (FPKM > 100) in these two medicinally significant tissues revealed several of the KEGG pathways attributed to primary metabolism, including starch and sucrose metabolism, tryptophan metabolism, purine metabolism, and pyrimidine metabolism, being enriched (Figure S4). Interestingly, the terpenoid backbone biosynthesis, which synthesizes the precursor GGPP for the diterpene alkaloid biosynthesis, was also enriched in both the medicinally valuable tissues.

Diterpene alkaloids, majorly consisting of the C19-type, are pivotal constituents of the A. japonicum metabolome, attributing to its medicinal properties [1]. As such, diterpene alkaloids remains at the primary focus for enhancing value addition and species improvement for medicinal purposes in Aconitum species. While the biosynthetic pathway of diterpene alkaloids remains largely unknown, the precursor molecules, GGPP, are derived from isopentenyl diphosphate (IPP) biosynthesized via both mevalonate (MVA) and methylerythritol (MEP) pathway [11, 15]. We used a combination of homology-based annotation, KEGG pathway and EC classifications, and strict criterion (length > 500 bps and similarity > 70%) to annotate 13 and 18 transcripts predicted to correspond to the known enzymes from the MVA and MEP pathway, respectively. Subsequently, we explored qualitative expression of transcripts annotated as enzymes associated with MVA or MEP pathway across four tissues. Most transcripts encoding enzymes from the MVA pathway were highly expressed in the daughter root (Fig. 5), while homologs of enzymes 3-hydroxy-3-methylglutaryl coenzyme A reductase (HMGR), mevalonate kinase (MVK), and diphosphomevalonate decarboxylase (MVDD) showed relatively higher expression in the mother root. HMGR has long been considered as a rate-limiting enzyme for the MVA pathway, and its association with terpenoid biosynthesis has been widely reported [12, 60], which may suggest its critical role in the biosynthesis of diterpene alkaloids in A. japonicum. As expected for the plastidial MEP pathway, while all enzymes’ homologs were highly expressed in the leaf, the expression level of homologs associated with bottleneck enzyme, 1-deoxy-D-xylulose-5-phosphate synthase (DXS), and isopentenyl diphosphate isomerase (IPPI) showed relatively higher expression in the mother root and daughter root [12], suggesting subtle regulatory patterns in this pathway.

Fig. 5

Schematics for the putative biosynthetic pathway of diterpene alkaloids in Aconitum japonicum and the tissue-wise expression for the assigned candidate genes to the biosynthesis pathways. Abbreviations: MVA pathway (mevalonate pathway), MEP pathway (methylerythritol pathway), LF (leaf), MR (mother root), DR (daughter root), RT (rootlet), AACT (acetoacetyl-CoA thiolase), HMGR (3-hydroxy-3-methylglutaryl-CoA reductase), HMGS (3-hydroxy-3-methylglutaryl-CoA synthase), MVK (mevalonate kinase), PMK (phosphomevalonate kinase), MVDD (mevalonate diphosphate decarboxylase), DXS (1-deoxy-d-xylulose 5-phosphate synthase), DXR (1-deoxy-d-xylulose 5-phosphate reductoisomerase), ISPD (2-C-methyl-d-erythritol 4-phosphate cytidylyltransferase), ISPE (4-(cytidine-5′-diphospho)-2-C-methyl-d-erythritol kinase), ISPF (2-C-methyl-d-erythritol 2,4-cyclodiphosphate synthase), ISPG ((E)-4-hydroxy-3-methylbut-2-enyl diphosphate synthase), ISPH ((E)-4-hydroxy-3-methylbut-2-enyl diphosphate reductase), IPPI (isopentenyl diphosphate isomerase), GGPPS (geranylgeranyl pyrophosphate synthase), CDPS (ent-copalyl diphosphate synthase), and KS (kaurene synthase)

In the initial stage of diterpene alkaloid biosynthesis, three molecules of IPP undergo condensation reaction catalyzed by the enzyme geranylgeranyl pyrophosphate synthase (GGPPS), to form GGPP. GGPP then undergoes bicyclization in the presence of terpene synthases- copalyl diphosphate synthase (CDPS), and kaurene synthase or kaurene synthase like (KS/KSL) enzymes to form kaurene or atisene. Further, kaurene or atisene molecules are oxidized and hydroxylated in the subsequent steps as parallel reactions and form the atisine skeleton via incorporation of β-ethanolamine moiety [26]. Kumar et al. [61] previously reported that kaurene undergoes oxidation and hydroxylation in the presence of enzyme kaurene oxidase (KO) and kaurene hydroxylase (KH), respectively, to form steviol, which through incorporation of β-ethanolamine moiety leads to the formation of atisine. While we identified 19 transcripts predicted to be annotated as known enzymes from diterpene alkaloid biosynthesis pathway, namely GGPPS, CDPS, KS, and KO, homologous transcripts corresponding to kaurene hydroxylase (KH), a key enzyme implicated in the hydroxylation of kaurene to produce steviol, were conspicuously absent in our transcriptome dataset of A. japonicum. Intriguingly, the artificial sweetener steviol, resulting from this enzymatic process, was not only absent in our metabolome dataset but has also not been previously reported in the toxic Aconitum species. The expression levels for both the homologs corresponding to KS, class I diterpene synthase, were highly expressed in the mother root together with KO (Fig. 5, Table S8). The significantly high expression level of KS and KO, together with the absence of KH, may suggest that while the oxidation reaction toward biosynthesis of atisenol from atisene may be catalyzed by KO, the hydroxylation reaction is catalyzed by an enzyme other than KH, which remains unknown.

Comparative transcriptome analysis to explore orthologous relationships among three Aconitum species

Comparative transcriptome analyses of closely related species have been employed to identify conserved transcripts involved in the biosynthesis of specialized metabolites in various plant species with distinct metabolic profiles, including Panax [37], and Ilex [62], among others. While plants from Aconitum genus are generally recognized for their high toxicity, Aconitum heterophyllum, primarily found in the Himalayan region, are significantly less toxic. This can be attributed to the predominance of C20-type diterpene alkaloids in A. heterophyllum in contrast to toxic Aconitum species such as A. japonicum and A. carmichaelii accumulating C19-type diterpene alkaloids as their signature metabolites [1, 61]. We used OrthoFinder for comparative transcriptome analyses using de novo transcriptome assembly for A. japonicum (this study), previously published transcriptome assembly for A. carmichaelii, and de novo transcriptome assembly (re-established in this study) using RNA-seq data from a previously published study for two tissues of A. heterophyllum were used for this analysis [11, 15].

OrthoFinder based comparative transcriptome analysis identified 38,828, 38,339, and 24,433 orthogroups in A. japonicum, A. carmichaelii, and A. heterophyllum, respectively, with 17,787 orthogroups containing at least one transcript from all three species (Figure S5A, Table S9). Within the subset of toxic Aconitum species transcriptome, 17,467 orthogroups represented at least one transcript from A. japonicum and A. carmichaelii while none from A. heterophyllum. In total, we identified 20,943 transcripts specific to the toxic species (Table S9), which include transcripts vital for the biosynthesis and regulation of specialized metabolites that are associated with the characteristic toxic biochemicals. GO term enrichment analysis identified defense response as one of the significant GO terms being enriched (Figure S5B). When comparing the transcripts of A. japonicum, annotated as enzymes involved in the diterpene alkaloid biosynthesis, with those of A. carmichaelii, we observed that nearly all transcripts identified from the A. japonicum transcriptome dataset had significant matches in the A. carmichaelii dataset. Among the 41 identified transcripts of A. japonicum predicted to encode enzymes involved in the diterpene alkaloid biosynthesis pathway leading to the kaurene/atisene core structure (Fig. 5, Table S8), 34 were included in the orthogroups shared with A. carmichaelii (Table S10). Moreover, the trend of the expression for most of the shared transcripts between root and leaf tissues was similar between these two plant species.

The comparative transcriptome analysis, presented here, provides a comprehensive perspective on the conserved sequences across two toxic Aconitum species, and contributes valuable insights into the shared genetic underpinnings of diterpene alkaloid biosynthesis. These transcripts could serve as potential candidates for further functional characterization and to explore their contribution to the biosynthesis of specialized metabolites. While the comparative transcriptome analyses yielded significant insights into the orthogroups’ distributions among Aconitum species, it is important to note that the use of de novo transcriptome assemblies presents certain limitations. These assemblies are inherently uneven and may not represent the entirety of the genome, leaving potential room for partially reconstructed transcripts. Further, the absence of transcripts in A. heterophyllum could well be due to fewer tissues used for constructing its de novo transcriptome assembly, which needs to be taken into consideration while selecting genes for further analysis. Analyzing these results with integrative multi-omics data is essential to identify strong candidate genes with a role in the biosynthesis of diterpene alkaloids. Consequently, further validation through complementary genomic or proteomic approaches is necessary to corroborate these findings.

Data-driven integration of metabolome and transcriptome to decipher the biosynthesis of diterpene alkaloids in Aconitum japonicum

Only the initial steps of the diterpene alkaloids biosynthesis have been known in Aconitum species. Therefore, the utilization of the advanced biotechnological toolkits, which relies on the complete knowledge of the biosynthetic pathway, for enhancing the production of bioactive metabolites is severely limited. Several studies in the past have highlighted the importance of co-expression-based integration analysis of metabolome and transcriptome datasets as a valuable tool to identify the functionally active genes across different plant species [63]. Therefore, in this study, we used a gene-metabolite integration approach to get insights into the known and unknown molecular components involved in the biosynthesis of diterpene alkaloids in A. japonicum. Pearson-based correlation analysis was performed for MS/MS-validated metabolites (Table S3) and annotated transcripts that were highly expressed (FPKM > 5) in at least one of the four tissue of A. japonicum (Table S11). Results of integrative omics analysis identified metabolite–transcripts relationships in the form of two distinct large clusters, cluster 1 and cluster 2, respectively (Figure S6A, Table S11). While cluster 1 was dominated by metabolites annotated as diterpene alkaloids, cluster 2 majorly contained metabolites annotated as phenylpropanoids and anthocyanins. Since the primary objective of this study was to identify candidate molecular components associated with the biosynthetic pathway of diterpene alkaloid in A. japonicum, we next conducted an in-depth investigation of the diterpene alkaloid cluster (cluster 1) to understand the type of enzymes being represented by the transcripts included in the cluster 1. Cluster 1 included 2,517 transcripts together with 47 metabolites (Table S11). The expression level of all the transcripts included in cluster 1 was highest in mother root (Figure S6B). Similarly, the accumulation levels of diterpene alkaloids present in cluster 1 were highly accumulated in the mother root, followed by the daughter root of A. japonicum (Figure S6C). Interestingly, transcripts encoding several of the important enzyme classes, such as CYPs, KS, KO, transcription factors, acyltransferase, and methyltransferase, were co-expressed with these metabolites and may suggest critical enzymes participating in the biosynthesis and diversification of these diterpene alkaloids. Further utilizing phylogenetic analysis of the identified candidate genes of A. japonicum with functionally characterized candidate genes will provide valuable insights into their respective functions.

Phylogenetic analysis of transcripts annotated as cytochrome P450 and co-clustered in the diterpene alkaloid cluster

Terpenoids represent the most abundant group of specialized plant metabolites, with tens of thousands of known structures spread across the entire plant kingdom [64]. The structural diversity of terpenoids relies on the modification of specific chemical groups, rearrangements of the scaffold sk