Abietane diterpenoids, including tanshinones and carnosic acid, are the most abundant type of tricyclic terpenoids, the biosynthetic pathway of which has been best investigated [45, 63, 64]. Tanshinones, carnosic acid and its derivatives are present in several species of Lamiaceae, in various combinations. It is known that the family of CYP450s plays an essential role in decoration of diterpene skeletons, providing for various types of structural oxidative modification reaction, such as: hydroxylation, carbonylation and heterocyclization [48, 49, 52, 53, 57]. Additionally, it has been shown that the catalytic promiscuity of CYP450s creates a metabolic grid for tanshinone biosynthesis [52, 61]. Nonetheless, despite numerous experimental strategies to elucidate the tanshinone biosynthesis pathway has been implemented to date, including: investigation of a spatial organization of the diterpenoid biosynthetic gene cluster [65], comparative metabolomic, transcriptomics and proteomic analysis [33, 40, 44, 66], targeted mutagenesis approaches using RNA interference and CRISPR/Cas9 system [41, 67, 68], hairy roots cultures and elicitor treatments [31, 69–75], or endophytic fungi infection [76], several steps from the pathway remain unresolved. Particularly, the most intriguing transformation – the hypothetical conversion of 11,20-dihydroxyferruginol to miltirone, which requires loss of C20 (directly via demethylation or via decarboxylation of carnosic acid as an intermediate) and aromatization of the “B” ring, has not been determined yet. Also, downstream transformations involving modification at the C4 of the miltirone backbone and leading to the formation of the metabolic grid of tanshinones remain undetermined as well as the direct precursor and transformations leading to the biosynthesis of isograndifoliol [77] and trilobinol [78].
Utilising two closely related species of slightly distinct diterpenoid profile, the metabolomics-guided transcriptomic approach allowed to select candidate genes, which expression might have led to observed chemical differences. Firstly, we have used the UHPLC-QTOF-MS analysis to follow the content of diterpenoids in the course of vegetation season (Table 1). Eight compounds were estimated quantitatively, out of which carnosol was the most abundant compound in the leaves of both species while in roots, the profile of abietane diterpenoid compounds was clearly dominated by cryptotanshinone (11). Interestingly, several compounds were found to be unique for one or the other species. Sugiol (9) was detected only in the S. yangii leaves, while trilobinol (10) was unique for S. abrotanoides. Didehydrotanshinone IIA (18), acetyloxycryptotanshinone (14) and ketoisograndifoliol (16) were unique for the roots of S. yangii. Also, the levels of isograndifoliol (15) and OH-tanshindiol A (17) were significantly higher in the roots of S. yangii than in S. abrotanoides. This analysis has confirmed our previous observations of a diverse diterpenoid profiles of S. abrotanoides and S. yangii, although it became clear that the difference in the content of isograndifoliol is quantitative, not qualitative, as previously thought [28]. Current observations corroborate with those made earlier in the roots of S. yangii, where the profile of abietane diterpenoid compounds was also dominated by the cryptotanshinone, followed by 1,2-didehydrotanshinone, miltirone and low amounts of tanshinone IIA [79]. Although the presence of tanshinone IIA was detected in roots of both S. abrotanoides and S. yangii [28], its low amounts did not allow to analyse its content quantitatively in this work.
As metabolic differences are reflected in different transcript profile of a species or tissues, we performed comparative analysis of S. abrotanoides and S. yangii transcriptomes. Using an RNA-sequencing technology we have sequenced and de novo assembled transcriptomes of leaves and roots of S. abrotanoides and S. yangii. As a result, 134,443 transcripts were annotated by UniProt and 56,693 of them were assigned as Viridiplantae. To our knowledge, this is first transcriptomic data reported for these species. In order to seek for differences between sequenced transcriptomes, the differential expression analysis was performed, which revealed that 463, 362, 922 and 835 genes indicated changes in expression in four comparisons: S. abrotanoides roots versus S. yangii roots, S. abrotanoides leaves versus S. yangii leaves, S. yangii roots versus leaves; and S. abrotanoides roots versus leaves, respectively (Fig. 2a). GO enrichment analysis of selected DEGs has found many DEGs to be enriched in the cellular metabolic term, which suggested significant variations in the secondary metabolic pathways in roots and leaves of S. abrotanoides and S. yangii. Additionally, related to the upstream terpenoid biosynthesis pathway, the ISPF gene (orthologous to the SmMCS), was found to be highly enriched in multiple GO terms, which points out towards lower steps of the diterpenoid pathway, currently not present in the databases. The homology and expression of two gene families associated with downstream steps of tanshinone and carnosic acid biosynthesis were studied, namely: cytochromes P-450 and 2-oxoglutarate-dependend dioxygenases. 44 CYP450s were identified among DEGs in this study (Supplementary Table S5), out of which seven are thought to be related to the abietane diterpenoid biosynthesis (Table 2). A total of 16 transcripts homologous to the 2-ODDs of S. miltiorrhiza were detected in transcriptomes of S. abrotanoides and S. yangi (Supplementary Table S6) with one being homologous to the Sm2-ODD14 designated as the tanshinone IIA synthase (Table 2). Additional BLAST analysis revealed existence of 39 different transcripts related to abietane diterpenoid biosynthesis in transcriptomes of S. abrotanoides and S. yangii, out of which 17 transcripts were found to be related to the downstream steps of biosynthesis of abietanoids and nor-abietanoids (Table 2) and eight formed a cluster of root-specific genes, when clustered according to their expression profile (Fig. 5b). Finally, eleven candidate genes from the downstream biosynthesis pathway were selected for extended analysis of their expression during the vegetation season.
In transcriptomes of S. abrotanoides and S. yangii, we have detected three cytochromes from the CYP76AH clade, namely: CYP76AH89, CYP76AH90 and CYP76AH91 (Supplementary Table S5 and Fig. 5). RNA-seq showed very high transcript levels of CYP76AH89 in root transcriptomes (at MOS), which was confirmed by qRT-PCR analysis (Table 2 and Fig. 7). Additionally, qRT-PCR analysis revealed a very high expression of CYP76AH89 in leaf transcriptomes, at the beginning of the season, which suggests the CYP76AH89 being a key enzyme in the biosynthesis of both tanshinones and carnosic acid, with notably different expression pattern in leaves and roots. Homology analysis showed that CYP76AH89 exhibits the highest similarity with the CYP76AH22 and CYP76AH23 from S. rosmarinus as well as with the SmCYP76AH3. In fact, all these enzymes possess the same amino acid residues in their active sites, which are: E301, E306, M395, F479 (Fig. 6). The other two genes from the CYP76AH clade, CYP76AH90 and CYP76AH91, also showed high similarity with ferruginol synthases-like genes from various Salvia species, however, their expression levels were much lower than those of the CYP76AH89, proving their minor involvement in the abietane biosynthesis. Interestingly, no obvious and significantly expressed homolog of the SmCYP76AH1 has been found in transcriptomes of S. abrotanoides and S. yangii. Therefore, we postulate, that the CYP76AH89 may accept both miltiradiene and ferruginol as substrates and catalyses both C11 and C12 hydroxylation, similarly as it has been proved for CYP76AH22-24 from S. rosmarinus [26, 58], as well C7 oxidation of the miltiradiene backbone, catalyzed by SmCYP76AH3. The close phylogenetic relationship of S. abrotanoides and S. yangii with S. rosmarinus was demonstrated by us earlier [28]. On the other hand, it was demonstrated that, although CYP76AH1 and CYP76AH3 catalyze two consecutive reactions in planta [49, 52], in vitro studies proved them having the same catalytic functions but different catalytic efficiencies. It was found that when ferruginol, 11-hydroxy ferruginol, and sugiol were used as substrates for in vitro enzymatic reactions, CYP76AH1 also catalyzed the hydroxylation of C11 and oxygenation of C7 sites, though at very low catalytic efficiency. Similarly, CYP76AH3 also catalyzed the production of trace amounts of ferruginol from miltiradiene. In the same work, a series of modeling-based mutational variants of CYP76AH1 were designed to integrate the functions of CYP76AH1 and CYP76AH3. The mutant CYP76AH1D301E,V479F, which integrated the functions of CYP76AH1 and CYP76AH3, was found to efficiently catalyze C11 and C12 hydroxylation and C7 oxidation of miltiradiene substrates [62]. We hypothesize, that in S. abrotanoides and S. yangii, the CYP76AH89 might have similarly integrated the two catalytic activities of CYP76AH1 and CYP76AH3 and became a multifunctional enzyme that catalyzes the formation of ferruginol from miltiradiene and further oxidizes the C7 and C11 positions of ferruginol to form sugiol, 11-hydroxy ferruginol, and 11-hydroxy sugiol. The other two CYP76AHs transcripts detected in S. abrotanoides and S. yangii, CYP76AH90 and CYP76AH91, possessed following predicted amino acid residues in their active site: E301, S306, I395, L479 and E301, S306, I395, I479, respectively (Fig. 6). Whether they possess redundant functions to the CYP76AH89 or contribute to the abundance of different abietane-diterpenoids in S. abrotanoides and S. yangii, it would need involvement of functional in vitro studies, however, the relatively low expression of CYP76AH90 and CYP76AH91 throughout the season (Table 2 and Fig. 7) suggests they minor role in the general biosynthesis pathway of abietanoids.
Next genes possibly encoding for abietane diterpenoid pathway enzymes has been found to be represented by following homologs: CYP76AK25, CYP76AK-fragment1, CYP71D754, 2-ODD14 and CYP71BE213.
Both CYP76AKs genes cluster together with other root-specific genes found by our RNA-seq analysis (Fig. 5b and Table 2), but the qRT-PCR has additionally revealed their high expression levels in leaves, at the beginning of the season (Fig. 7). Again, this would suggest involvement of CYP76AK25 and CYP76AK-fragment1 in the biosynthesis of both tanshinones and carnosic acid, in, however, different temporal and spatial manner. Homology analysis indicate that CYP76AK25 might catalyze biosynthesis of 11,20-dihydroxysugiol and 11,20-dihydroxyferruginol, while CYP76AK-fragment1 possibly catalyzes three sequential C20 oxidations for the conversion of 11-hydroxyferruginol to carnosic acid in S. abrotanoides and S. yangii (Fig. 7A) as it was demonstrated for RoCYP76AK6-8 [58].
Given the high homology of CYP71D754 with SmCYP71D411 and its very high expression in roots of S. abrotanoides and S. yangii, we hypothesize that CYP71D754 may possibly accept the 11-dihydroxysugiol and 11-dihydroxyferruginol as substrates and produce the 11,20-dihydroxysugiol and 11,20-dihydroxyferruginol in these species (Table 2 and Fig. 7). That would make its catalytic activity redundant with the CYP76AK25, however, cytochrome P450s playing redundant roles in plants have already been described [26, 60, 61]. Whether the CYP71D754 is also able to catalyze the heterocyclization of the D-ring in S. abrotanoides and S. yangii to produce the cryptotanshinone, remains unknown. On the other hand, given a high level of cryptotanshinone in roots of S. abrotanoides and S. yangii, an enzyme catalyzing the direct upstream hydroxylation of C16 and 14,16-ether (hetero)cyclization to form the D-ring (as SmCYP71D375) should have a clear root-specific expression profile and in that perspective the CYP71D754 seems to be a good candidate.
2-ODD14 was found to share high similarity with the Sm2-ODD14 and had its expression level only slightly higher in roots than in leaves (Table 2 and Fig. 7). Investigation of other isoforms with more apparent root-specific expression would certainly be beneficial, although, the hardly detectable amount of tanshinone IIA in roots of S. abrotanoides and S. yangii does not allow for expecting a high and root-specific transcript levels of the tanshinone IIA-producing enzyme.
The role of the CYP71BE213 in the biosynthesis of abietane diterpenoids in S. abrotanoides and S. yangii could only be speculated, but its homology with salviol-producing CYP71BE52 from S. pomifera and relatively high expression in roots and leaves of S. abrotanoides and S. yangii (Table 2 and Fig. 7) suggest potential involvement of CYP71BE213 in other abietane diterpenoid transformations, possibly through the substrate promiscuity of this homolog.
Up to date, the commercial supply of tanshinones relies on extraction from S. miltiorrhiza. Originally harvested from the wild, S. miltiorrhiza is now field cultivated, which has become the main source of Danshen. With good agricultural practices the drug is relatively uniform in quality, but can contain heavy metals and pesticide residues [45]. Therefore, there is a high demand for searching for alternative sources of tanshinones. One of them could be hairy root and cell cultures of S. miltiorrhiza, which produce and secrete tanshinones, but the yields from such cultures also does not yet appear to be economically viable [31]. Also, endophytic fungi strains isolated from native S. abrotanoides have been reported to be able for ex planta biosynthesis of cryptotanshinone [80], while cultured endophytic fungi from S. miltiorrhiza turned out to be capable of tanshinone I and tanshinone IIA production [81, 82]. Unfortunately, due to the unknown reasons, endophytic fungi cultures often tend to lose their ability for plant-derived metabolites production [83]. In this view, another tanshinone-producing and easy to grow plant species would serve as a valuable alternative.
Therefore, by using the comparative transcriptomic approach, we have generated a dataset of candidate genes which provides a valuable resource for further elucidation of tanshinone biosynthesis. In a long run, our investigation may lead to optimization of diterpenoid profile in S. abrotanoides and S. yangii through genetic engineering, which may become an alternative source of tanshinones for further research on their bioactivity and pharmacological therapy.