Development and application of base editing systems in Salvia miltiorrhiza for precise metabolic engineering

Salvia miltiorrhiza (Danshen), a key traditional medicinal plant for treating cardiovascular, inflammatory and oncological diseases (Zheng et al., 2023). While CRISPR/Cas-mediated knockout enables gene disruptions in Danshen, precision base editing remains limited (Hsu et al., 2024). This study aimed to develop highly efficient adenine (A) and cytosine (C) base editors (ABEs and CBEs) for A-to-G or C-to-T conversions. By fusing a mutated N-methylpurine DNA glycosylase (mMPG) to ABE, we further expand its editing capabilities to include A-to-T and A-to-C substitutions (Anzalone et al., 2020). To investigate the feasibility of base editing in Danshen, we initially constructed first-generation editors: SmCBEmax-01 (CBE, using Arabidopsis-optimized Anc689-nCas9(D10A)-UGI with AtU6-driven sgRNAs) and SmABE8e-01(ABE, using TadA8e-nCas9(D10A) with AtU6-driven sgRNAs and CaMV 35S promoter-expressed editors) (Figure 1a, Figure S1). Seven sgRNAs targeting SmMYB1, SmMYB36, SmMYB76, SmMYB39, SmbZIP2, SmJAZ3, SmbHLH3 (CBE) and five sgRNAs targeting SmMYB1, SmC4H, SmKSL2, SmHMGR1, SmCPS2 (ABE) were designed (Figure 1b, Tables S1, S2) (Zheng et al., 2023). Agrobacterium (C58C1)-mediated hairy root transformation generated 281 CBE and 397 ABE hygromycin-resistant lines. Sanger sequencing revealed no detectable editing in CBE, while ABE exhibited 18.3% A-to-G editing efficiency (22/120 sequenced lines) at SmMYB1-sgRNA2 A5 position. Limited efficiency was attributed to genome complexity, chromatin accessibility, and sgRNA/editor expression bottlenecks (Jiang et al., 2020). To optimize, we tested promoter combinations: native SmU6 or 35SEN (35S enhancer + CmYLCV + truncated U6-26) for sgRNAs (Jiang et al., 2020), and endogenous SmEF1α or SmRps5A (shown to enhancing precision in other plants) for editors (Niu et al., 2022). This yielded second-generation editors: SmCBEmax-02/03 and SmABE8e-02/03, with SmU6-SmEF1α and 35SEN-SmRps5A combinations, respectively (Figure 1a, Figure S1). To rapidly assess the performance of optimized CBEs and ABEs, we selected SmMYB1-sg1, SmMYB76-sg1 and SmMYB36-sg1 (CBE) and SmMYB1-sg2, SmKSL2-sg1 and SmHMGR1-sg1 (ABE). Stable T0 plants (134 CBE, 295 ABE) were generated via Agrobacterium tumefaciens only 1 heterozygous C-to-T edit in CBEs (SmMYB36-sg1, SmCBE-max-03; 1/72, 1.4%; Table S1, Figure S2). In contrast, optimized ABEs showed marked improvements: SmABE8e-02 achieved 36.6%–22.5% A-to-G editing efficiencies at SmMYB1-sg2, SmHMGR1-sg1, SmKSL2-sg1; while SmABE8e-03 displayed 58.3%-100.0% even with homozygous edits (Figure 1b,c). Further analysis of SmABE8e-03 exhibited position-dependent editing: A5 (100%) > A4 (58.3%); consecutive AA dinucleotides reduced editing (e.g. SmKSL2-sg1: A5-A6) showed reduced editing at the second adenine. Editing window spanned A4-A9, with rare A9 conversion (unlike Arabidopsis/tomato) (Figure 1d,e) (Niu et al., 2022). To explore multiplex editing, SmABE8e-03-Dual (bidirectional sgRNAs) targeting SmKSL2-sg1 and SmHMGR1-sg1 was constructed (Figure 1a) (Fan et al., 2024). Among 177 transgenic plants, editing efficiencies decreased to 19.2% (34/177) and 3.4% (6/177) for individual targets, yet 3.4% (6/177) showed simultaneous editing, confirming feasibility (Figure 1b,c). These results validate SmABE8e-03 as highly efficient for A•T-to-G•C base editing in Danshen. MYB-family transcription factors positively regulate their biosynthesis. Previous studies showed disrupting upstream ATG (uATG) or upstream open reading frames (uORFs) enhances protein translation efficiency through post-translational regulation (Shen et al., 2024). We identified a 9-amino-acid uORF (MGKVCVGR*) in the 5′ UTR of SmMYB1 and designed a target to convert uATG to uGTG, thereby disrupting both the uATG and uORF (Figure 1f). Using SmABE8e-03, we obtained two homozygous A5-to-G5 conversions (Figure 1b). Transcript analysis showed no significant difference (Figure 1g) UPLC-TQ-MS revealed significant metabolic enhancements: tanshinone I/IIA, dihydrotanshinone I (1.6/2.5/1.9-fold), cryptotanshinone, salvianolic acid A/B, rosmarinic acid (2.1/1.4/1.5/3.3-fold) (Figure 1h). SmMYB-uGTG lines exhibited normal growth despite metabolic boosts (Figure 1i). To further expand editable base conversions (A•T-to-T•A/A•T-to-C•G) (Li et al., 2024), we constructed SmAKBE-03 by fusing plant-codon-optimized mMPG to SmABE8e-03 (Li et al., 2024; Wu et al., 2023), and endogenous SmPolη (incorporates AP sites to enhance A-T editing) to the N-terminus of HPTII (Figure 1a, Table S4). Evaluating SmAKBE-03 at SmMYB1-sg2 and SmKSL2-sg1 via GV3101 transformation generated 39 and 53 transgenic lines. Sequencing revealed 71.7% (38/53) A-to-G, 7.5% (4/53) A-to-T, and 3.8% (2/53) A-to-C edits within the A3-A11 window (PAM positions 21–23) for SmKSL2-sg1; enables while SmMYB1-sg2 achieved 53.8% (21/39) A-to-G efficiency with persistent A11 editing but no A-to-T/C products (Figure 1b,d). These findings confirm SmAKBE-03 enables diversified base conversions in Danshen. SmKSL1, catalyzing of copalyl diphosphate (CPP) into miltiradiene. Its homolog SmKSL2 competitively depletes GGPP (CPP precursor), diverting metabolic flux (Zheng et al., 2023). The AKBE-SmKSL2-sg1 introduced A9-to-G9 substitution, converting the premature stop codon TGA (*252) to CGA (arginine, R), disrupting its structure (Figure 1j). Two edited lines harbored the A9-G9 mutation (Figure 1b,j). Transcript analysis revealed no significant changes in SmKSL2 expression (Figure 1k). UPLC-TQ-MS quantification demonstrated metabolite elevation: AKBE-SmKSL2#9 increased tanshinone I/IIA, salvianolic acid A (1.4/2.3/1.1-fold); AKBE-SmKSL2#35 enhanced tanshinone I, cryptotanshinone, rosmarinic acid (1.6/3.2/1.1-fold) (Figure 1l). Both lines maintained normal growth (Figure 1m). Metabolic variability likely stems from a downstream TGA (*254) preserving partial SmKSL2 protein integrity (Figure 1j). Though SmAKBE-03 shows low A-to-T/C efficiencies and no A-to-Y edits in several sites (e.g. SmMYB1-sg2), its broad editing window (A3-A11) and A•T-to-T•A/C•G conversions highlight its potential. To date, we have successfully established high-efficiency ABE, ABE-dual and AKBE editors in Danshen, surpassing prior systems with a more complete toolkit and improved efficiency (Han et al., 2024). SmABE8e-03 achieves 100.0% editing efficiency and multiplex editing; while SmAKBE-03 expanded the ABE window to A3-11 and enables A-to-T/C conversion. Future, we will refine these systems for precise editing of key metabolic genes (e.g. SmCPS1, SmCYP76AH1), laying a robust foundation for next-generation metabolic engineering in S. miltiorrhiza. This study was supported by the Natural Science Foundation of China (82474040), the Scientific and Technological Innovation Project of CACMS (CI2024C005YN, CI2023E002), Fundamental Research Funds for the Central public welfare research institutes (ZZ15-YQ-060, ZZXT202417) and a key project at the central government level: the ability establishment of sustainable use for valuable Chinese medicine resources (2060302). We thank Xiaokai Gao and Shuai Liu from Shunfeng Biotechnology (Hainan) for S. miltiorrhiza transformation. Q.Y. and H.Z. proposed and designed the research; Q.Y., Y.Y. and Q.L. performed experiments; Q.Y. and L.L. analysed the data. X.L., H.L., M.Y., Y.Y., Y.T. and J.Z. performed transformation; Q.Y., H.Z. and L.-Q.H. wrote and revised the manuscript. All authors read and approved the final manuscript. All data generated or analysed during this study are included in this manuscript and its supplementary information files. Figures S1–S2. Tables S1–S4. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.