Overexpression of ROSMARINIC ACID SYNTHASE enhances the hepatoprotective potential of Dracocephalum heterophyllum hairy root cultures Authors: Lu Cao, Zhenying Wu, Yingping Cao, Lichengcheng Ren, Na Hu, Yaling Liu, Yuchen Liu, Shaowei Yu, Jingzhe Sun, Meifeng Liu, Feng Yuan, Honglun Wang, Chunxiang Fu Published: 2026 ABSTRACT AbstractDracocephalum heterophyllum Benth., a widely used Tibetan medicinal herb in the Lamiaceae family, accumulates bioactive secondary metabolites such as rosmarinic acid (RosA) and exhibits strong hepatoprotective, antioxidant, and anti-inflammatory activities. The sustainable production of this herb is limited by the extreme conditions required for its growth. To address this limitation, we developed an efficient axillary bud propagation system for D. heterophyllum, achieving a multiplication coefficient of 4.8. We then generated hairy roots from propagated shoot explants using Agrobacterium rhizogenes–mediated transformation. Among the tested A. rhizogenes strains, LBA9402 showed the highest induction efficiency on stem explants, reaching 23.3% ± 1.0. Heterologous expression of the C. blumei gene encoding rosmarinic acid synthase (RAS) in hairy roots resulted in a 1.2- to 2.1-fold increase in RosA levels compared with those in empty vector control roots. Consistent with the elevated RosA levels, the RAS-overexpressing hairy roots exhibited markedly enhanced lipid-lowering and hepatoprotective effects in oleic acid–induced HepG2 cells. These results establish bioengineered D. heterophyllum hairy root cultures as a scalable and sustainable source of bioactive compounds and as an alternative to wild-harvested plant material.Keywords: Dracocephalum heterophyllum Benth., Hairy root, Hepatoprotective effect, Propagation, Rosmarinic acid Introduction Dracocephalum heterophyllum Benth., a perennial herb in the Lamiaceae family, has considerable medicinal and economic value. Broadly distributed in the high-altitude regions of Tibet, Western Sichuan, Xinjiang, Qinghai, Gansu, Inner Mongolia, and Shanxi, China [1], D. heterophyllum synthesizes and accumulates diverse secondary metabolites associated with hepatoprotective, antioxidant, and anti-inflammatory activities [2,3]. D. heterophyllum has long been used in the ethnomedicinal practices of Northwest China and in the Amchi system of medicine in the Ladakh region of the Himalayas. Modern pharmacological studies have demonstrated that ethyl acetate extracts of D. heterophyllum have strong anti-hepatitis activity and contain nine major compounds, including rosmarinic acid (RosA), methyl rosmarinate, and pomolic acid [4]. Ethanol-ethyl acetate extracts of D. heterophyllum have significant anti-proliferative effects on autoreactive CD4+ T cells and significantly slow the progression of experimental autoimmune uveitis [5]. Among the compounds extracted from D. heterophyllum, RosA has been identified as the predominant bioactive component [4,5]. The phenylpropanoid ester RosA is widely distributed across the plant kingdom, with particularly high concentrations in members of the Lamiaceae family [6]. RosA exhibits a broad range of pharmacological and biological activities, including hepatoprotective, neuroprotective, antioxidant, anti-diabetic, and anti-inflammatory effects [7,8]. Conventional RosA extraction methods, including maceration, heat reflux, and ultrasound-assisted extraction, require large amounts of plant material [9]. Enhancing endogenous RosA biosynthesis could provide a practical strategy for increasing production of this metabolite. In plants, RosA biosynthesis is thought to proceed de novo from phenylalanine. Rosmarinic acid synthase (RAS) catalyzes the condensation of 4-coumaroyl-CoA and 3,4-dihydroxyphenyllactic acid, leading to the formation of RosA. Overexpression of RAS in Salvia miltiorrhiza significantly increased RosA content, while knockout of RAS reduced RosA accumulation [10,11]. These observations support the genetic regulation of RAS as a promising strategy to increase RosA accumulation and thereby improve the overall quality of D. heterophyllum. Recent advances in genomics and synthetic biology have markedly accelerated research aimed at improving and harnessing the potential of medicinal plants [12]. In addition to identifying the biosynthetic and regulatory mechanisms underlying secondary metabolism, the efficient production of natural bioactive compounds requires suitable biological chassis [13]. Hairy roots have emerged as a valuable specialized tissue system for studying plant secondary metabolism, identifying and characterizing key metabolic genes, and elucidating metabolic pathways. Furthermore, hairy roots complement mature plant organs and undifferentiated callus cultures in diverse biotechnological applications [14,15]. The successful use of hairy roots as biosynthetic platforms for the production of secondary metabolites such as flavonoids highlights their potential value in developing traditional Chinese medicinal plant resources [16]. However, hairy root systems and other in vitro systems, such as tissue-cultured seedlings or callus cells, have not been reported for D. heterophyllum. Therefore, establishing tissue culture and genetic engineering–based synthetic biology systems could greatly enhance the conservation and sustainable use of D. heterophyllum resources. Metabolic dysfunction–associated steatotic liver disease (MASLD), the most common liver disease worldwide, is estimated to affect at least 30% of adults [17]. This disease is characterized by excessive lipid accumulation in the liver and can progress to cirrhosis and hepatocellular carcinoma [18]. Plant-derived natural compounds enriched with polyphenols, flavonoids, polysaccharides, terpenes, and alkaloids show hepatoprotective potential by ameliorating or reversing liver damage, often with minimal side effects [19]. Oleic acid (OA)-induced HepG2 cells are widely used as an in vitro model of hepatic steatosis for evaluating bioactive compounds in MASLD studies [20,21]. This model has been instrumental in evaluating the therapeutic efficacy of plant-derived bioactive compounds such as kaempferol and kaempferide from Hippophae rhamnoides and caffeic acid and rosmarinic acid-3-O-glucoside from Mesona chinensis [22,23]. Therefore, evaluating RosA-rich D. heterophyllum materials using the OA-induced HepG2 model could clarify the potential of this compound to alleviate hepatic steatosis and dysfunction. In this study, we established an efficient and rapid propagation system for D. heterophyllum to generate sufficient explants for hairy root induction. We then developed an A. rhizogenes–mediated genetic transformation system and introduced the RAS gene from Coleus blumei into D. heterophyllum, resulting in hairy root lines with markedly increased RosA accumulation. Finally, we evaluated the potential therapeutic efficacy of extracts from these high-RosA hairy roots in an OA-induced HepG2 model of MASLD. The efficient micropropagation and hairy root transformation systems established in this study provide a biotechnological platform for both the conservation and sustainable utilization of D. heterophyllum resources. Results Establishment of a rapid propagation system via tissue culture To obtain sufficient plant material from D. heterophyllum, we evaluated two in vitro regeneration methods: axillary clustered bud induction and embryogenic callus differentiation. Mature seeds harvested from wild D. heterophyllum plants (Fig. 1A and B) were surface-sterilized and germinated to produce seedlings for subsequent procedures (Fig. 1C). Tender axillary buds were excised from the seedlings for adventitious bud induction, and the remaining leaves and petioles were used as explants for callus induction (Fig. 1D and E). Medium containing the cytokinins 6-benzyladenine (BA; 0.5 mg/L) and kinetin (KT; 2.0 mg/L) induced the emergence of up to five new buds within three weeks. After transfer to medium containing the auxins 1-naphthaleneacetic acid (NAA; 0.5 mg/L) and indole-3-butyric acid (IBA; 0.5 mg/L), approximately 85.3 ± 5.3% of the buds formed roots and developed into seedlings, resulting in an axillary bud propagation system with a multiplication rate of approximately 4.8 over two months (Fig. 1D–F). Establishment of highly efficient micropropagation systems for D. heterophyllum. A Wild D. heterophyllum plants. B Mature seeds of D. heterophyllum. C Seedling from a sterilized D. heterophyllum seed. D Micropropagation of D. heterophyllum plantlets from axillary buds. E Micropropagation of D. heterophyllum plantlets from embryogenic calli. F Multiplication rates of D. heterophyllum plantlets from axillary buds (AB) and embryogenic callus (EC). G Callus induction (CI) and callus regeneration (CR) efficiencies of D. heterophyllum. Values are means ± SE (n = 3). Student's t-test, ∗∗P < 0.01. For the callus-based approach, embryogenic calli formed on callus induction medium within approximately three weeks, achieving an induction rate of approximately 80.7% (Fig. 1E–G). After a two-month differentiation phase on regeneration medium, regenerative buds formed at a rate of about 30.7% (Fig. 1G). These buds were then rooted to produce regenerated seedlings, resulting in the establishment of a propagation system with a multiplication coefficient of approximately 1.3 within four months (Fig. 1E and F). Overall, axillary bud induction was faster and more efficient than callus-based regeneration, making it the preferred method for the rapid in vitro propagation of D. heterophyllum. Induction of hairy roots in D. heterophyllum To develop a hairy root induction and culture system for D. heterophyllum, we infected leaves, petioles, and stem cuttings from regenerated seedlings with two A. rhizogenes strains: K599 and LBA9402. Putative hairy roots emerged approximately two weeks after inoculation, mainly at the wounded sites of petioles and stem cuttings, whereas no roots formed on leaves (Fig. 2A–D). With strain K599, root induction rates were 4.4 ± 0.4% in petioles and 3.0 ± 0.9% in stem cuttings. In contrast, LBA9402 induced root formation at rates of 5.6 ± 0.9% in petioles and 23.3 ± 1.0% in stem cuttings (Fig. 2E). Hairy root induction and identification in D. heterophyllum. A The hairy root induction process. B Leaf explants after induction treatment. C Hairy root induced from petiole explants. D Hairy roots induced from stem explants. E Hairy root induction efficiencies of A. rhizogenes strains K599 and LBA9402 across different explant types. Values are means ± SE (n = 3). Different letters above the error bars indicate significant differences determined by one-way ANOVA (P < 0.05, Tukey's multiple-range test). F PCR analysis of DNA from A. rhizogenes-infected hairy root lines derived from stem explants. M: 2 kb plus DNA marker; PC: positive control, DNA from LBA9402; WR: DNA from untransformed wild-type roots; 1–15: putative hairy root lines. The rolB and virG amplicons were 194 bp and 529 bp long, respectively. To confirm that the newly formed roots were true hairy roots, we individually cultured all 15 root lines derived from LBA9402-infected stem cuttings for four weeks and tested them by PCR for the genomic integration of the Ri plasmid. None of the roots harbored virG, whereas 12 lines tested positive for rolB. The remaining three lines grew slowly and were considered potential pseudo-positives (Fig. 2F; Fig. S1). Among the tested conditions, stem cuttings infected with A. rhizogenes strain LBA9402 had the highest hairy root induction rate in D. heterophyllum. Generation and identification of RAS_OE hairy root lines To develop novel D. heterophyllum germplasm with increased RosA content, we constructed the binary expression vector pEarlyGate100-CbRAS, which harbors the RAS gene from C. blumei fused with eGFP (Fig. 3A). We infected 60 tender stem cuttings of D. heterophyllum with A. rhizogenes strain LBA9402, resulting in 15 candidate hairy root lines after a two-week induction. Screening for green fluorescence from the eGFP marker identified seven potential RAS heterologous overexpression (RAS_OE) hairy root lines (Fig. 3B and C). PCR amplification to detect the presence of CbRAS, eGFP, virG, and rolB in the 15 candidate hairy root lines revealed a co-transformation efficiency of approximately 53.9% for both the Ri and Ti plasmids (Fig. 3D). qRT-PCR analysis further showed that the seven positive co-transformed hairy root lines exhibited significantly higher RAS transcript levels than the control lines (Fig. 3E), confirming that we successfully generated hairy root lines heterologously expressing CbRAS. Expression of the eGFP gene in co-transformed hairy roots of D. heterophyllum. A Schematic illustration of the modified pEarlyGate 100 vector construct used in this study. B Hairy root under direct light. C Hairy root under excitation light. D Identification of co-transformed hairy root lines of D. heterophyllum by PCR screening. M: 2 kb DNA plus marker; PC: positive control, DNA from LBA9402; WR: DNA from untransformed wild-type roots. 1–15: putative transgenic hairy root lines. The rolB, virG, eGFP, and RAS amplicons were 194 bp, 529 bp, 280 bp, and 1572 bp long, respectively. E Relative RAS expression levels in EV control and RAS_OE hairy roots. OE-n: putative hairy root lines. Values are means ± SE (n = 3). One-way ANOVA (Dunnett's multiple-range test), ∗∗P < 0.01. RosA accumulation and growth analysis of RAS_OE hairy root lines After subculturing the seven positive lines for two weeks, we analyzed their metabolite profiles and RosA contents using HPLC. RosA (Peak 3) was the dominant phenylpropanoid metabolite in the hairy roots, whereas only trace amounts of putative luteoloside (Peak 1) and an unidentified compound (Peak 2) were detected (Fig. 4A; Fig. S2). Notably, RAS overexpression further increased RosA contents in the RAS_OE lines, which were 1.2- to 2.1-fold higher than in the empty vector control (EV) lines. Among these lines, OE-1 exhibited the highest RosA concentration, reaching 16.9 mg/g dry weight (Fig. 4B). Rosmarinic acid content of hairy roots. A HPLC chromatograms of methanolic extracts from EV control and RAS_OE lines. B RosA (Peak 3) contents in methanolic extracts from hairy roots of EV control and RAS_OE lines. Values are means ± SE (n = 3). One-way ANOVA (Dunnett's multiple-range test), ∗∗P < 0.01. To evaluate the growth of hairy roots with elevated RosA contents, we subcultured 0.5 g root cultures from EV, OE-1, OE-2, and OE-13 transgenic lines for four weeks (Fig. 5A and B). The average fresh weight of control EV hairy roots was 3.2 ± 0.06 g, whereas that of OE-1, OE-2, and OE-13 was 1.6 ± 0.06 g, 1.3 ± 0.10 g, and 1.9 ± 0.22 g, respectively (Fig. 5C). In contrast, untransformed wild-type roots displayed severely retarded growth, minimal biomass accumulation over 28 d, and partial browning (Fig. 5A–C; Fig. S1D). Together, RAS overexpression improved RosA production but reduced root growth, implying a trade-off between yield and biomass. Growth of D. heterophyllum hairy roots. A-B Hairy roots grown in solid medium for 0 and 28 d, respectively. C Biomass of root cultures. WR, wild-type roots; EV, empty vector control hairy roots; OE-1, 2, and 13, RAS_OE lines. Values are means ± SE (n = 3). Different letters above the error bars indicate significant differences determined by one-way ANOVA (P < 0.05, Tukey's multiple-range test). Analysis of the hepatoprotective capacity of extracts from RAS_OE hairy root lines To assess the potential efficacy of hairy root lines with elevated RosA contents, we established an OA-induced lipid accumulation model in HepG2 cells and used it to evaluate extracts from wild-type plants (WP), transgenic hairy roots, and non-transgenic hairy roots. We first measured changes in triglycerides (TG), total cholesterol (TC), alanine aminotransferase (ALT), aspartate aminotransferase (AST), malondialdehyde (MDA) levels. Additionally, we performed dose-response cell viability assays to examine the effects of RosA and the extracts on HepG2 cell viability. RosA concentrations of 0–10 μM had no significant effect on cell viability, whereas 20 μM RosA significantly reduced cell viability (Fig. S3A). Similarly, extracts had negligible effects on cell viability at concentrations of 0–10 μg/mL but significantly reduced viability at 20 μg/mL (Fig. S3B). Based on these results, we selected 10 μM RosA and 10 μg/mL extract for subsequent experiments to assess their effects on TG, TC, ALT, AST, and MDA levels. The RosA concentrations in 10 μg/mL extracts from WP, EV, OE-1, 2, and 13 were 0.2 μg/mL, 0.5 μg/mL, 1.2 μg/mL, 0.8 μg/mL, and 0.6 μg/mL, respectively. Moreover, 10 μg/mL extracts from both WP and EV hairy roots significantly suppressed the OA-induced increases in TG, TC, ALT, AST, and MDA contents, with inhibitory effects comparable to those of 20 μM orlistat and 10 μM RosA (Fig. 6). Compared with the EV control, extracts from OE-1, OE-2, and OE-13 significantly reduced TG and TC levels in OA-induced HepG2 cells (Fig. 6A and B). By contrast, the effects of extracts from these RAS_OE hairy root lines on the other indicators (ALT, AST, and MDA contents) were inconsistent (Fig. 6C–E). These results demonstrate that increasing RAS expression through genetic engineering increases RosA contents in D. heterophyllum hairy roots and strengthens their potential hepatoprotective effects in the OA-induced HepG2 cell model. Effects of different extracts on lipid accumulation and hepatoprotective effects in HepG2 cells. A Triglycerides (TG) levels. B Total cholesterol (TC) levels. C Alanine aminotransferase (ALT) levels. D Aspartate aminotransferase (AST) levels. E Malondialdehyde (MDA) levels. WP: wild-type D. heterophyllum plants; EV: empty vector control lines; OE: RAS_OE lines. The RosA concentrations of WP, EV, OE-1, 2, and 13 extracts (10 μg/mL) were 0.2 μg/mL, 0.5 μg/mL, 1.2 μg/mL, 0.8 μg/mL, and 0.6 μg/mL, respectively. Values are means ± SE (n = 3). Different letters above the error bars indicate significant differences determined by one-way ANOVA (P < 0.05, Tukey's multiple-range test). Discussion Medicinal plants, which are prized for their unique therapeutic uses, have long been a key focus for resource exploration and utilization. However, environmental changes and human overexploitation have led to a gradual decline in high-quality plant materials, placing some species at risk of endangerment. D. heterophyllum, a medicinal plant primarily found on the Qinghai-Tibet Plateau, has been extensively described in traditional Tibetan medical literature due to its medicinal properties, underscoring its significant potential for development. To date, most research on D. heterophyllum has focused on its metabolite composition and physiological activities [24]. In this study, we established a rapid in vitro propagation technique for D. heterophyllum, developed a hairy root induction and cultivation system, and used genetic engineering to increase the levels of its key active compound, RosA. Collectively, our results confirm the feasibility of enhancing D. heterophyllum through biotechnological approaches and underscore its therapeutic potential for MASLD. Tissue culture-based rapid propagation techniques offer an effective, practical solution for addressing the shortage of medicinal plant resources. Many valuable medicinal plants have been successfully propagated on a large scale via tissue culture, including Rhodiola dumulosa and Fritillaria cirrhosa [25,26]. Plant regeneration via callus induction tends to have a longer cycle, whereas direct regeneration through axillary bud induction produces a higher multiplication rate within the same period [27]. This pattern is exemplified by the rapid in vitro propagation of Vanilla planifolia [28]. In this study, we compared two regeneration methods for D. heterophyllum: axillary bud induction and callus induction. Axillary bud regeneration achieved a multiplication rate of 4.8 in approximately two months, whereas callus induction required more than four months and resulted in a lower multiplication coefficient of 1.3. Seedlings obtained from a single callus regeneration cycle accounted for only ∼26.3% of those produced through axillary buds, highlighting the greater efficiency of axillary bud regeneration. Although repeated callus subculture could eventually produce large numbers of D. heterophyllum seedlings, axillary bud regeneration is more practical when explants are limited and the aim is rapid, high-efficiency propagation with minimal hormone usage. Hairy roots are valuable platforms for the synthesis of plant secondary metabolites and play a central role in plant synthetic biology research. As valuable platforms for plant secondary metabolite synthesis and synthetic biology research, hairy roots offer rapid growth, phytohormone autotrophy, the ease of generating single lines, and high secondary metabolite production over conventional plant cell suspension cultures, untransformed roots, and whole plants [29]. The efficiency of hairy root induction depends on multiple factors, such as the plant species, explant type, A. rhizogenes strain, infection duration, co-culture duration, and bacterial concentration. Among these factors, strain-explant compatibility is crucial for achieving the highest induction rates [29]. The susceptibility to A. rhizogenes infection varies across plant species.For example, using A. rhizogenes strain ATCC15834, hairy root induction was efficient in Salvia corrugata and Salvia officinalis, but poor in Salvia wagneriana [30,31]. Conversely, LBA9402 strain induced roots poorly in S. corrugata, but efficiently in Callerya speciosa and Amaranthus tricolor [30,32]. Moreover, different explant types, such as cotyledons, leaves, petioles, stems, and hypocotyls, vary in their susceptibility to different strains, with no consistent pattern, making multiple tissues suitable for hairy root induction [33]. In this study, we examined the ability of two A. rhizogenes strains, LBA9402 and K599, to infect D. heterophyllum. LBA9402 was more effective than K599. Neither strain successfully induced hairy roots in leaves; however, both achieved successful infection in petioles and stems, with stems showing the highest infection rate (23.3%). Compared with S. corrugata, S. officinalis, C. speciosa, and A. tricolor, the hairy root induction rate of D. heterophyllum was relatively low. To further enhance hairy root induction in D. heterophyllum, additional optimization of infection conditions or the identification of additional hypervirulent strains is needed. To construct overexpressing hairy roots, we employed a binary vector transformation system with a GFP visualization tag. This system enabled more rapid and efficient identification of single overexpression hairy root lines at early stages compared to traditional PCR-based methods. The resulting D. heterophyllum hairy roots demonstrated a remarkable increase in biomass compared with untransformed wild-type roots. These hairy roots have significant application potential and provide valuable resources for the development and utilization of D. heterophyllum. RAS is the rate-limiting enzyme that catalyzes the final step in RosA biosynthesis [34]. The functional importance of RAS has been confirmed in several studies. In C. blumei, RNA interference (RNAi)-mediated knockdown of HPPR and RAS in hairy roots resulted in a 0.14% and 0.87% reduction in RosA contents relative to the dry weight of hairy roots, respectively, indicating that decreasing RAS expression had a more pronounced effect on RosA levels. Similarly, correlation analyses between C4H and RAS expression and RosA content in Origanum vulgare demonstrated that RAS plays a more critical role in regulating RosA biosynthesis than HPPR [35]. Although we attempted to isolate the endogenous RAS sequence from D. heterophyllum, this proved difficult because its genome sequence is not available. Therefore, we selected RAS from C. blumei, a functional ortholog from another Lamiaceae species, for overexpression to generate transgenic hairy roots. However, heterologous overexpression of RAS may lead to different feedback regulation of the RosA biosynthetic pathway in D. heterophyllum. Thus, the functions of endogenous D. heterophyllum RAS genes merit further investigation. Most RAS_OE lines with high RAS expression levels generated in this study exhibited correspondingly high RosA contents. However, OE-8, with a high RAS expression level, accumulated less RosA than other lines with high RAS expression levels, such as OE-1 and OE-2, indicating that transcript abundance alone does not necessarily determine the final concentration of the synthesized product. The lack of complete correspondence between gene expression and metabolite accumulation has also been observed in S. miltiorrhiza RAS overexpression studies [10,36]. This discrepancy may arise because secondary metabolite accumulation is a gradual, dynamic process, while gene expression levels reflect a more transient state [37]. Additionally, post-transcriptional regulation and post-translational modifications can further decouple gene expression from enzymatic activity. Overall, the combined influence of these various metabolic regulatory mechanisms likely leads to a discrepancy between gene expression levels and the accumulation of metabolic products [38]. In the present study, we also noted that RAS_OE lines with high RosA contents grew more slowly than the EV control. This reduced growth may result from the accumulation of RosA imposing physiological stress on hairy roots. Plants maintain a balance between growth/development and defense mechanisms: the activation of secondary metabolic pathways enhances the synthesis of secondary metabolites that confer defensive functions but often suppresses growth and reduces biomass. Similarly, excessive accumulation of these metabolites can induce cytotoxicity, oxidative stress, and metabolic imbalances, ultimately hindering the growth of hairy roots [39,40]. For instance, overexpressing SmMYB52 promoted salvianolic acid B biosynthesis while markedly inhibiting hairy root proliferation in S. miltiorrhiza [41]. In the current study, we successfully generated genetically engineered D. heterophyllum lines with elevated RosA levels. However, for large-scale cultivation and further medicinal applications, it is essential to develop additional lines that achieve an optimal balance between biomass production and RosA accumulation using a stable genetic transformation system. MASLD, formerly known as non-alcoholic fatty liver disease (NAFLD), is a prevalent chronic liver disease worldwide [17,18]. The development of this disease is influenced by multiple mechanisms. Several drugs have demonstrated clinical benefits for treating MASLD, including insulin sensitizers, statins, ursodeoxycholic acid, and betaine [42]. Orlistat, a gastrointestinal lipase inhibitor, can help alleviate MASLD symptoms by reducing fat accumulation and protecting liver cells from oxidative stress-induced damage [43,44]. Additionally, RosA has been shown to ameliorate MASLD by regulating lipid metabolism, combating oxidative stress, and minimizing liver damage [45]. Certain traditional Chinese medicinal plants abundant in RosA, such as Orthosiphon aristatus and Melissa officinalis, have shown notable theoretical and practical efficacy for the treatment of MASLD [46,47]. Extracts from the hepatoprotective herb D. heterophyllum containing high levels of RosA have also exhibited therapeutic effects against MASLD [24]. In this study, we evaluated extracts from WP, EV, and RAS_OE hairy roots of D. heterophyllum in HepG2 cells treated with OA. All extracts reduced lipid accumulation and markers of cellular injury, indicating hepatoprotective activity. The hepatoprotective effects of EV extracts were similar to those of WP. However, this comparison should be interpreted cautiously because WP and hairy roots differ in tissue type and culture conditions. Notably, RAS_OE hairy root extracts were more effective than EV hairy roots at reducing OA-induced TG and TC levels. These observations suggest that increased RosA accumulation enhances the lipid-lowering and hepatoprotective effects of D. heterophyllum hairy roots. Materials and methods Plant materials Wild-type D. heterophyllum plants were collected in Tongren city, Qinghai Province, China (101°58′33.24″ E, 35°30′29″ N) in August 2023 and identified by Prof. Qingbo Gao at Northwest Plateau Biology Institute, CAS. To obtain sterile seedlings, robust D. heterophyllum seeds were sanitized by soaking in 75% (v/v) aqueous ethanol (EtOH) for 30 s, followed by 10% (v/v) sodium hypochlorite (NaClO) for 10 min, and rinsed 3-5 times until no odor from NaClO was detected. The seeds were treated with 100 mg/L GA for 12 h, plated on seed germination medium (Table S1), and incubated under a 16-h light/8-h dark photoperiod at 25 °C for one month. Tissue culture and plant regeneration of D. heterophyllum In vitro D. heterophyllum cultures were established using two approaches: (1) rapid propagation through axillary bud induction, and (2) callus formation followed by differentiation. For axillary bud induction, robust branches from sterile seedlings were sectioned to include one or two pairs of opposite leaves and placed on induction medium supplemented with 2.0 mg/L kinetin (KT) and 0.5 mg/L N6-benzyladenine (6-BA). The cultures were maintained at 25 °C under a 16-h light/8-h dark photoperiod to promote axillary bud development. Axillary shoots approximately 2 cm long were transferred to rooting medium supplemented with 0.5 mg/L naphthaleneacetic acid (NAA) and 0.5 mg/L indolebutyric acid (IBA). For callus induction, the remaining leaves and stem explants from sterile seedlings were cultured in the dark at 25 °C for three weeks on medium containing 5.0 mg/L 2,4-dichlorophenoxyacetic acid (2,4-D) and 0.15 mg/L 6-BA. Fresh, actively growing calli were transferred to differentiation medium supplemented with 0.5 mg/L KT and 2.0 mg/L 6-BA. After approximately two months, adventitious shoots regenerated from calli were subcultured on rooting medium to induce root formation. The types and compositions of media used in this study are listed in Table S1. Induction of hairy roots Two Agrobacterium rhizogenes strains, K599 and LBA9402, were used to infect explants from D. heterophyllum seedlings for hairy root induction [48]. Briefly, 200 μL of A. rhizogenes culture stored at −80 °C was cultured in 50 mL of liquid LB medium supplemented with antibiotics (50 mg/L streptomycin for the A. rhizogenes strain K599, 50 mg/L rifampicin for LBA9402). The cultures were shaken at 200 rpm at 28 °C until the OD600 reached 0.6–0.8. Acetosyringone was added to a concentration of 100 μM, and the cultures were shaken for an additional 2 h. The A. rhizogenes cells were concentrated by centrifugation at 1400 × g for 15 min at 25 °C and resuspended in infection solution (Table S1). The OD600 of the resuspended Agrobacterium cultures was adjusted to 0.3–0.4 for inoculation. Sterile explants (leaf, petiole, and stem sections; 100 of each type) were immersed in infection solution and vacuum-infiltrated at 0.1 MPa pressure for 25 min. Following infection, the explants were transferred to co-cultivation medium and incubated in the dark at 25 °C for two days. After co-cultivation, the explants were transferred to hairy root induction medium consisting of 1/2 MS basal medium supplemented with 300 mg/L Timentin and cultured under the same conditions. The emergence of new hairy roots was first observed at wound sites within approximately two weeks. The hairy root induction efficiency for A. rhizogenes strains K599 and LBA9402 on different explants was calculated using the following formula: Hairy root induction rate (%) = (Number of explants that produce hairy roots/total number of infected explants) × 100%. After four weeks of culture, genomic DNA was extracted from putative hairy roots using the CTAB method [49], and PCR amplification of the rolB and virG genes was performed to confirm successful transformation. Generation of RAS-overexpressing (RAS_OE) hairy root lines The full-length coding sequence of RAS from C. blumei (GenBank accession no. AM283092.1) [50] was cloned into the pEarleyGate100 overexpression vector fused with the green fluorescent protein (GFP) gene via attL-attR recombinant reactions. The resulting fusion plasmids were used to transform A. rhizogenes strain LBA9402. RAS_OE hairy root lines were generated using the induction method described above. Putative transformants were initially identified based on GFP fluorescence, and the fluorescent hairy roots were transferred to fresh hairy root induction medium for further culture. After four weeks of growth, PCR amplification was performed to detect the presence of the rolB, eGFP, virG, and RAS genes, thereby confirming successful transformation and identifying overexpression lines. The co-transformation rate was calculated as: Co-transformation rate (%) = (Number of roots containing rolB, eGFP, and RAS genes/total number of roots with the rolB gene) × 100%. Total RNA was extracted from the hairy root lines using a TRIzol Kit (TransGen Biotech, Beijing, China) and reverse transcribed into cDNA using a PrimeScript™ RT Kit (TransGen Biotech, Beijing, China). The cDNA was diluted five-fold and subjected to qRT-PCR in triplicate using gene-specific primers to analyze the expression levels of RAS. The housekeeping gene Actin was amplified using primers previously utilized in a study of Dracocephalum moldavica [51]. All primers used in this study are listed in Table S2. Quantification of RosA and analysis of hairy root growth After two weeks of culture, fresh, vigorous D. heterophyllum hairy roots were harvested, lyophilized, and ground into a fine powder. To quantify RosA, the ground samples were extracted in 75% methanol via ultrasonication for 30 min, and the extracts were analyzed by high-performance liquid chromatography (HPLC) as described previously [52]. Chromatographic separation was carried out on an Ultimate LP-C18 column (3.0 mm × 150 mm, 2.7 μm) at 30 °C, with the mobile phase consisting of 0.1% formic acid in water (A) and acetonitrile (B). The gradient elution program was as follows: 0–2 min, 15% B; 2–5 min, 15%–25% B; 5–20 min, 25%–35% B; 20–21 min, 35%–95% B; 21–30 min, 95% B. The flow rate was 0.2 mL/min, and the injection volume was 1 μL. The detection wavelength was 325 nm. RosA was identified based on its UV-visible spectra and comparison with an authentic standard compound (Shanghai Yuanye Bio-Technology Co. Ltd., Shanghai, China). Control hairy roots carrying the empty vector (EV) and three RAS_OE lines exhibiting high RosA contents were selected to generate growth curves. Each 0.5 g sample of robust hairy roots was inoculated into 40 mL hairy root-inducing medium, and fresh weight was recorded after 28 d of cultivation. All experiments were performed in triplicate. Analysis of lipid reduction and hepatoprotective activity assay of hairy root extracts Human liver cancer cells (HepG2) were obtained from the Cell Resource Center of the Shanghai Institute of Life Sciences, Chinese Academy of Sciences (Shanghai, China). The cells were cultured in Dulbecco's Modified Eagle Medium (DMEM) supplemented with 1% penicillin-streptomycin solution and 10% (v/v) fetal bovine serum and incubated at 37 °C in a humidified atmosphere containing 5% CO2. To establish a lipid accumulation model, the HepG2 cells were inoculated into 96-well plates at a density of 1 × 105 cells/mL and treated with OA for 24 h. The D. heterophyllum extracts used in this experiment were prepared by ultrasonic extraction in 75% methanol, followed by freeze-drying, and were stored at 4 °C until use. The experimental design included four treatment groups: (1) the blank control, (2) a model group treated with 0.5 mM OA, (3) the positive control treated with 0.5 mM OA plus 20 μM Orlistat or 10 μM RosA, and (4) experimental groups treated with 0.5 mM OA and extracts derived from wild-type plants (WP), EV control, and RAS_OE (OE-1, OE-2, OE-13) hairy root lines, each at concentrations of 10 μg/mL. The extracts were dissolved in dimethyl sulfoxide (DMSO) solution by sonication and diluted in DMEM to the required concentration. After 24 h of incubation, triglyceride (TG), total cholesterol (TC), aspartate aminotransferase (AST), alanine aminotransferase (ALT), and malondialdehyde (MDA) levels were determined using biochemical assay kits [53]. The effects of RosA and the extracts on HepG2 cell viability were evaluated using a 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-2H-tetrazolium bromide (MTT) assay [53]. All experiments were performed in triplicate. Statistical analysis Statistical analysis was performed using GraphPad Prism (GraphPad Software, San Diego, CA, USA, Version 9.5.1). Each experiment was conducted independently three times, and the data were reported as mean ± standard error of the mean (SEM). Data from independent samples were subjected to Student's t-test or one-way analysis of variance (ANOVA). A value of P < 0.05 indicates statistical significance. CRediT authorship contribution statement Lu Cao: Writing – original draft, Methodology, Investigation, Data curation. Zhenying Wu: Writing – original draft, Supervision, Data curation, Conceptualization. Yingping Cao: Writing – original draft, Methodology, Investigation, Data curation. Lichengcheng Ren: Methodology, Investigation, Data curation. Na Hu: Supervision, Resources, Funding acquisition, Data curation. Yaling Liu: Project administration, Data curation. Yuchen Liu: Methodology, Investigation, Data curation. Shaowei Yu: Methodology, Investigation. Jingzhe Sun: Methodology, Investigation, Funding acquisition. Meifeng Liu: Project administration, Methodology, Funding acquisition. Feng Yuan: Project administration. Honglun Wang: Writing – review & editing, Supervision, Funding acquisition, Conceptualization. Chunxiang Fu: Writing – review & editing, Supervision, Funding acquisition, Conceptualization. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgements This work was supported by the National Technology Innovation Center for Prataculture Special Fund for Innovation Platform Construction (CCPTZX2024QN13), Qinghai Provincial Science and Technology Major Project (2023-SF-A5), the National Natural Science Foundation of China (No. 32570422), the Postdoctoral Innovation Program of Shandong Province (SDCX-ZG-2024 00118), China Postdoctoral Science Foundation (2025M774013), the Taishan Scholar Program of Shandong, and Huzhou City Key Research and Development Plan Project (2023ZD2035). Footnotes Supplementary data to this article can be found online at https://doi.org/10.1016/j.abiote.2026.100087. Contributor Information Honglun Wang, Email: hlwang@nwipb.cas.cn. Chunxiang Fu, Email: fucx@qibebt.ac.cn. Supplementary data The following is the Supplementary data to this article: Multimedia component 1 Data availability Data will be made available on request. References Associated Data Supplementary Materials Multimedia component 1 Data Availability Statement Data will be made available on request.