Transcription Factor SmSPL2 Inhibits the Accumulation of Salvianolic Acid B and Influences Root Architecture

The SQUAMOSA PROMOTER-BINDING PROTEIN-LIKE (SPL) transcription factor play vital roles in plant growth and development. Although 15 SPL family genes have been recognized in the model medical plant Salvia miltiorrhiza Bunge, most of them have not been functionally characterized to date. Here, we performed a careful characterization of SmSPL2, which was expressed in almost all tissues of S. miltiorrhiza and had the highest transcriptional level in the calyx. Meanwhile, SmSPL2 has strong transcriptional activation activity and resides in the nucleus. We obtained overexpression lines of SmSPL2 and rSmSPL2 (miR156-resistant SmSPL2). Morphological changes in roots, including longer length, fewer adventitious roots, decreased lateral root density, and increased fresh weight, were observed in all of these transgenic lines. Two rSmSPL2-overexpressed lines were subjected to transcriptome analysis. Overexpression of rSmSPL2 changed root architectures by inhibiting biosynthesis and signal transduction of auxin, while triggering that of cytokinin. The salvianolic acid B (SalB) concentration was significantly decreased in rSmSPL2-overexpressed lines. Further analysis revealed that SmSPL2 binds directly to the promoters of Sm4CL9, SmTAT1, and SmPAL1 and inhibits their expression. In conclusion, SmSPL2 is a potential gene that efficiently manipulate both root architecture and SalB concentration in S. miltiorrhiza.


Introduction
Salvia miltiorrhiza Bunge is a renowned model medical plant species [1]. Its dried root is a traditional Chinese medicine (referred to as Danshen in Chinese), which has been extensively utilized for the remedy of cardiovascular and cerebrovascular diseases [2]. Water-soluble salvianolic acid B (SalB) is one of the main bioactive constituents of Danshen and has a wide range of pharmacological activities, which include antioxidant, antiinflammatory, antiapoptotic, antifibrotic effects; promoting stem cell proliferation and differentiation; and more [3]. The SalB content in Danshen has been listed as an indicator of quality control [4].
Recently, several strategies, such as treatment with an elicitor, manipulation of the key enzyme genes in the SalB biosynthesis pathway, and the application of transcription factors (TFs), have been employed to improve the availability of SalB in vivo [5]. Among these strategies, TFs are considered as an effective tool to control the biosynthesis of the main bioactive constituents in S. miltiorrhiza [6]. A few families of TFs, such as R2R3-MYBs (R2R3-type myeloblastosis proteins), bHLHs (basic helix-loop-helix protein), WKRYs, AP2/ERFs (APETALA2/ETHYLENE RESPONSE FACTORs), GRAS (named after the first three identified proteins within this family, GAI, RGA, and SCR), and LBDs (lateral organ boundary domain proteins), have been reported to regulate SalB biosynthesis and influence its accumulation [6].
Since the root of S. miltiorrhiza is the medicinal part, its morphology and biomass is an important agronomic trait. Although some agricultural production measures, such as proper planting densities and fertilization combinations, have been used to improve root production [7], genetic manipulation strategies have not been reported to date. For S. miltiorrhiza breeders, the ultimate goal is to acquire high-quality S. miltiorrhiza resources, which possess not only a high active ingredient content, but also high root production. Identifying potential genes, which participate in the regulation of root morphology and biomass, is the prerequisite of the molecular breeding of S. miltiorrhiza.
The SQUAMOSA PROMOTER-BINDING PROTEIN-LIKE (SPL) TFs are exclusive to plants and play vital roles in plant growth and development, such as controlling plant leaf development, nutritional phase transition, flowering time, fruit development, root development, anthocyanin accumulation, and stress responses [8]. Among the 16 SPL genes of Arabidopsis thaliana, ten are targeted by miR156 and clustered into five clades in accordance with the amino acid sequences of the conserved DNA binding domains-AtSPL2/AtSPL10/AtSPL11, AtSPL9/AtSPL15, AtSPL3/AtSPL4/AtSPL5, AtSPL6, and At-SPL13 [9]. AtSPL2 affects plant fertility and floral organ development due to the direct activation of ASYMMETRIC LEAVES 2 [10]. AtSPL3, AtSPL9, and AtSPL10 participate in the inhibition of lateral root development in Arabidopsis, with AtSPL10 playing a predominant character [8]. AtSPL9 interacts with AtPAP1 (Production of Anthocyanin Pigment 1, a well-known R2R3-MYB member in Arabidopsis) to negatively regulate anthocyanin biosynthesis [11]. AtSPL2, AtSPL10, and AtSPL11 bind directly to the promoters of AP2/ERFs and inhibit wound-induced auxin biosynthesis to suppress root regeneration [12].
For S. miltiorrhiza, 15 SPL genes have been found in its genome; eight of which, including SmSPL2/3/5/6/8/11/14/15, have been validated to be the targets of miR156 [13]. To date, only two SPL genes, SmSLP6 and SmSPL7, have been functionally characterized in S. miltiorrhiza [14,15]. The overexpression of SmSPL6 promotes the biosynthesis of phenolic acids and reduces the production of anthocyanin [14]. SmSPL7 has been proved to be a positive regulator for anthocyanin accumulation while a negative regulator for phenolic acid in S. miltiorrhiza [15]. The functions of the other SmSPL genes have not been characterized to date.
SmSPL2 and AtSPL2/10/11 were clustered in the same subfamily based on phylogenetic tree analysis [13]. We speculated that SmSPL2 may affect root growth and the production of bioactive constituents in S. miltiorrhiza. In the current work, we studied the function of SmSPL2 by analyzing its expression profiles and gain-of-function phenotype. The overexpression of SmSPL2 (SmSPL2-OE) and miR156-resistant SmSPL2 (rSmSPL2-OE) in S. miltiorrhiza resulted in obvious changes in the adventitious root morphology, and the transcriptome of rSmSPL2-OE lines was further analyzed. SmSPL2 was proved to be a negative regulator for SalB biosynthesis by directly suppressing the transcription of several enzyme genes for the pathway of phenolic acid biosynthesis. The present study confirmed that SmSPL2 was a strong regulator for both SalB accumulation and root development in S. miltiorrhiza.

SmSPL2 Expression Patterns
We detected the expression profiles of SmSPL2 in different tissues of one-year-old S. miltiorrhiza by quantitative reverse transcription PCR (qRT-PCR) analysis. The qRT-PCR results revealed that SmSPL2 was expressed in various tissues of S. miltiorrhiza, and the highest transcriptional level was found in the calyx, followed by the pistil ( Figure 1A).
To further detect the temporal and spatial expression profiles of SmSPL2, we generated transgenic Arabidopsis plants in which the GUS gene was driven by SmSPL2 promoter (ProSmSPL2::GUS). The histochemical GUS staining was performed on T2 generation transgenic Arabidopsis at various developmental periods. The result showed that the GUS signal was observed in both vegetative and reproductive stages of Arabidopsis ( Figure 1). A strong GUS signal was visible in the whole seedlings after germination at two days ( Figure 1B), five days ( Figure 1C), and nine days ( Figure 1D), with stronger signals in the older leaves than the younger leaves. During the anthesis stage of Arabidopsis, the GUS signal was observed in stem leaf ( Figure 1E), rosette leaf ( Figure 1F), stem ( Figure 1G), flower ( Figure 1H), and root ( Figure 1I), with stronger signals in the tap root and calyx. The blue signal was obvious at both ends of the pod ( Figure 1J). Our results suggested that the SmSPL2 promoter drove the expression of genes in almost all plant components at various developmental periods.

SmSPL2 Resides in the Nucleus and Exhibits Transactivation Activity
To check the subcellular localization of SmSPL2, we constructed recombinant vector 35Spro::SmSPL2-GFP, which was instantaneously expressed in the onion epidermis. Our results revealed that the signal of green fluorescent GFP in the positive control was uniformly distributed throughout the cell, while that of the SmSPL2-GFP fusion protein was specifically located in the nucleus (Figure 2A), which indicated that SmSPL2 was nuclear protein.
Meanwhile, the recombinant pGBKT7-SmSPL2 construct was used for the transactivation assay of SmSPL2 in transformed yeast AH109. AH109 cells containing pGBKT7-SmSPL2 or pGBKT7 plasmid grew normally on SD/-Trp solid medium; nevertheless, only cells containing pGBKT7-SmSPL2 showed normal growth and turned blue on SD/-Trp/-Ade/-His/X-α-gal yeast medium ( Figure 2B). These results revealed that SmSPL2 had a transcriptional activation function.

Generation of SmSPL2/rSmSPL2-Overexpressed Transgenic S. miltiorrhiza
To analyze the roles of SmSPL2 in S. miltiorrhiza, we obtained transgenic S. miltiorrhiza plants that overexpressed SmSPL2. Since SmSPL2 is a target of miR156 [13], we also generated rSmSPL2-overexpressed (OE) S. miltiorrhiza plants. Compared with the control, the morphology of adventitious roots (AR) changed significantly in both SmSPL2-OE and rSmSPL2-OE lines, exhibiting longer root lengths and a reduced number of AR and lateral roots ( Figure 3A-C). The qRT-PCR was used to analyze the transcriptional abundance of SmSPL2 in the transgenic lines. Our results revealed that the transcriptional abundance of SmSPL2 was dramatically increased in SmSPL2-OE and rSmSPL2-OE lines, in contrast to the control ( Figure 3D). Among the transgenic lines, the expression levels of SmSPL2 in rSmSPL2-OE1, rSmSPL2-OE2, and rSmSPL2-OE7 were 83.17-, 53.07-, and 42.66-fold of that of the control, respectively.
The phenotypic changes in the roots in the SmSPL2-OE1, SmSPL2-OE2, SmSPL2-OE4, rSmSPL2-OE1, rSmSPL-2OE2, and rSmSPL2-OE7 lines were obvious. Therefore, we measured AR number, AR length, and fresh weight of roots for the two-month-old transgenic lines. Compared with the control, the AR length was significantly increased in all these lines ( Figure 3E), while the AR number was considerably decreased ( Figure 3F). The fresh weight of the root was significantly increased in SmSPL2-OE1, SmSPL2-OE2, SmSPL2-OE4, and rSmSPL2-OE2 ( Figure 3G). These results revealed that the overexpression of SmSPL2 or rSmSPL2 strongly affects the S. miltiorrhiza root phenotype. , AR numbers (F), fresh weight of root (G) of two-month-old transgenic lines. All data are mean values of three biological replicates, with the error bar representing SD. * and ** indicate significant differences from the control at p < 0.05 and p < 0.01 level, respectively, by Student's t-test.

Transcriptome Analysis of Overexpressed rSmSPL2 and S. miltiorrhiza Control Plants
To further investigate the potential molecular mechanisms of SmSPL2 on the phenotype changes in S. miltiorrhiza plants, we selected rSmSPL2-OE2, rSmSPL2-OE7, and a control for RNA-seq. The Illumina deep sequencing obtained~42.59, 41.59, and 44.06 million high-quality clean reads for rSmSPL2-OE2, rSmSPL2-OE7, and the control, respectively. The Q30 value (percentage of sequences with a sequencing error rate <0.1%) was 92.75% for rSmSPL2-OE2, 93.42% for rSmSPL2-OE7, and 93.31% for the control.
We listed 20 significantly enriched metabolic pathways for those DEGs by KEGG analysis (Figures S3 and S4). The rSmSPL2-OE2 and rSmSPL2-OE7 shared nine significantly enriched pathways, including phenylpropanoid biosynthesis, flavonoid biosynthesis, starch and sucrose metabolism, zeatin biosynthesis, carotenoid biosynthesis, and more ( Figures  S3 and S4). Among the significantly enriched metabolic pathways, phenylpropanoid biosynthesis was the most significant pathway in both rSmSPL2-OE2 and rSmSPL2-OE7, which indicated that SmSPL2 was involved in phenylpropanoid biosynthesis.

DEGs Involved in Cytokinin Biosynthesis and the Signal Transduction Pathway
The cytokinin class of phytohormones has been implicated in various aspects of plant growth and development, including the inhibition of root growth and lateral root formation [16]. Since the DEGs in rSmSPL2-OE2 and rSmSPL2-OE7 were significantly enriched in the KEGG pathway of zeatin biosynthesis ( Figures S3 and S4), we further analyzed all the DEGs involved in cytokinin biosynthesis and signaling pathway. As shown in Figure 4A,B, five DEGs encoding putative IPT (isopentenyltransferase), CYP735A, and LOG (LONELY GUY) for cytokinin biosynthesis and eight DEGs involved in cytokinin signal transduction were all up-regulated in rSmSPL2-OE2. For rSmSPL2-OE7, although two of the four DEGs encoding enzymes for cytokinin biosynthesis were downregulated, two DEGs involved in cytokinin signal transduction were up-regulated. Our transcriptome data indicated that the overexpression of rSmSPL2 robustly triggered cytokinin signaling in S. miltiorrhiza roots.
Furthermore, we examined cytokinin levels in the roots of one-month-old rSmSPL2-OE2, rSmSPL2-OE7, and the control. Consistent with the transcriptome data, cytokinin levels in rSmSPL2-OE2 and rSmSPL2-OE7 were significantly upregulated, compared to the control ( Figure 4C).

DEGs Involved in Auxin Biosynthesis and the Signal Transduction Pathway
It is well documented that the crosstalk between auxin and cytokinin signaling plays an important role in the regulation of root growth [17]. We further analyzed the DEGs involved in auxin biosynthesis and the signal transduction pathway ( Figure 4D,E). Among the DEGs that encoded putative enzymes for auxin biosynthesis, including TAA (Tryptophan aminotransferase of Arabidopsis), AIM (Indole-3-acetamide hydrolase), and AAO (Aldehyde oxidase), one (TAA1) was down-regulated and two (TAA2 and AIM1) were up-regulated in rSmSPL2-OE2, while two (TAA1 and AAO1) were down-regulated, and one (TAA2) was up-regulated in rSmSPL2-OE7. The DEGs involved in auxin signal transduction, which encode the putative IAA (AUXIN (AUX)/INDOLE-3-ACETIC ACID) protein, GH3 (Gretchen Hagen 3), and SAUR (small auxin up RNA), were all down-regulated in rSmSPL2-OE2 (15 DEGs) and rSmSPL2-OE7 (five DEGs). The transcriptome data revealed that the overexpression of rSmSPL2 strongly inhibited auxin signaling in the roots.
Additionally, we determined auxin levels in the roots of one-month-old rSmSPL2-OE2 and rSmSPL2-OE7. As shown in Figure 4F, the auxin concentrations in rSmSPL2-OE2 and rSmSPL2-OE7 were only 9.8% and 20.1% of the control, respectively, suggesting overexpression of rSmSPL2 significantly suppressed auxin biosynthesis.

DEGs Involved in SalB Biosynthesis Pathway
To evaluate whether the decreased accumulation of SalB was the result of the altered transcription levels of genes in the salvianolic acid biosynthesis pathway, we examined all the possible enzyme genes (SmTATs, SmHPPRs, SmPALs, SmC4Hs, Sm4CLs, SmRASs, and SmCYP98A14) identified in the transcriptome data ( Figure 5A). Among the 24 detected genes, 10 DEGs were found in rSmSPL2-OE2 with 9 being down-regulated (SmTAT1, SmPAL1, SmPAL3, SmC4H1, Sm4CL1, Sm4CL3, Sm4CL10, SmRAS2, and SmRAS3). For the six DEGs in rSmSPL2-OE7, five (SmTAT1, SmPAL3, Sm4CL9, SmRAS2, and SmRAS3) were down-regulated. To inspect the accuracy of the transcriptome data, we selected six enzyme genes (SmTAT1, SmPAL1, SmC4H1, Sm4CL1, SmCYP98A14, and SmRAS1) for qRT-PCR analysis. Our qRT-PCR results revealed that all six tested genes were significantly suppressed in rSmSPL2-OE2 and rSmSPL2-OE7, which was relatively aligned with the RNA-seq data ( Figure 5B). Our qRT-PCR results testified that the RNA-seq data was statistically reliable and the overexpression of SmSPL2 suppressed the salvianolic acid biosynthesis pathway.

Overexpressed rSmSPL2 Inhibits RA and SalB Biosynthesis
Since SalB is synthesized via the pathways of tyrosine and phenylpropanoid metabolism [18], we speculated that SmSPL2 may be involved in SalB biosynthesis, where rosmarinic acid (RA) is believed to be the SalB precursor [19]. We detected the RA and SalB in the roots of control, rSmSPL2-OE2, and rSmSPL2-OE7 via high-performance liquid chromatography (HPLC). As shown in Figure 5C, the retention time of RA and SalB was 31.37 ± 0.02 and 40.59 ± 0.02 min, respectively. Compared with the control, the RA and SalB contents significantly decreased in the rSmSPL2-OE transgenic lines. The RA concentrations in rSmSPL2-OE2 and rSmSPL2-OE7 were 59% and 78% of the control, respectively ( Figure 5D). For SalB, the concentrations in rSmSPL2-OE2 and rSmSPL2-OE7 were only 20% and 36% of the control, respectively ( Figure 5D). These results suggested that the overexpression of rSmSPL2 suppressed RA and SalB biosynthesis in the roots of S. miltiorrhiza.

SmSPL2 Binds to and Represses Sm4CL9, SmTAT1, and SmPAL1 Promoters
SPL transcription factor can bind to the GTAC motif of target genes to regulate their expression [20]. We found that the promoter regions of quite a few genes in the salvianolic acid biosynthesis pathway (Sm4CL9, SmTAT1, SmPAL1, and SmCYP98A14) contain the GTAC motif ( Figure 6A). To test whether SmSPL2 directly binds to their promoters, yeast one-hybridization (Y1H) experiment was carried out. The Y1H results revealed that SmSPL2 binds to the promoter regions of the three genes (Sm4CL9, SmTAT1, and SmPAL1) ( Figure 6B). Additionally, we performed dual luciferase transient transcriptional assays (TTA) in tobacco to investigate whether SmSPL2 might inhibit their expression. The TTA results revealed that SmSPL2 inhibited the expression of Sm4CL9, SmTAT1, and SmPAL1 ( Figure 6C,D). Overall, our results indicated that SmSPL2 directly suppressed transcription of Sm4CL9, SmTAT1, and SmPAL1 by binding to their promoters.

Role of SmSPL2 in the Biosynthesis of Phenolic Acids
Because of the wide-ranging pharmacological effects of SalB, improving its production in vivo through biotechnology is one of the research hotspots for S. miltiorrhiza [21]. The most direct strategy is to change the expression levels of enzyme genes in the SalB biosynthesis pathway. For example, overexpressing SmRAS and SmCYP98A14, which encode the key enzymes for SalB biosynthesis, significantly increased the content of SalB in S. miltiorrhiza hairy roots [22]. Several TFs, such as SmMYB97 [23], SmMYB111 [24], SmMYC2 [25], SmERF115 [26], SmbZIP1 [27], and SmNAC1 [28], positively regulate SalB biosynthesis by directly targeting the promoter regions of some key enzyme genes of phenolic acid biosynthesis and activating their expression in S. miltiorrhiza, while other TFs, such as SmLBD50 [29], SmbZIP2 [30], SmGRAS1/2/3 [31], SmERF1L1 [32], SmMYB36 [33], and SmbHLH37 [34], have been verified to be the negative regulators for SalB biosynthesis.
Among the 16 SPL family members in S. miltiorrhiza, SmSPL6 and SmSPL7 have been experimentally proven to participate in the accumulation of SalB [14,15]. SmSPL6 promotes SalB biosynthesis by directly activating the transcription of SmCYP98A14 and Sm4CL9 [14]. SmSPL7 suppresses the biosynthesis of SalB by directly inhibiting the transcription of SmTAT1 and Sm4CL9 [15]. In the present study, we confirmed that SmSPL2 was a negative regulator for SalB biosynthesis and that overexpression of rSmSPL2 dramatically decreases the concentrations of SalB and RA in S. miltiorrhiza roots ( Figure 5). Y1H and dual luciferase assays revealed that SmSPL2 inhibited SalB biosynthesis by directly suppressing the transcription of multiple genes, including SmTAT1, Sm4CL9, and SmPAL1 ( Figure 6).

SmSPL2 Influences S. miltiorrhiza Root Architecture
In Arabidopsis, AtSPL2, AtSPL10, and AtSPL11 inhibit root regeneration by impeding wound-induced auxin biosynthesis [12]. Phylogenetic analysis indicated that SmSPL2 tended to cluster with AtSPL2/10/11 [13]. We wondered whether SmSPL2 was involved in root growth and development. In this study, obvious morphological changes in the roots, including longer length, fewer adventitious roots, decreased lateral root density, and increased fresh weight, were observed in SmSPL2 or rSmSPL2-overexpressed transgenic S. miltiorrhiza (Figure 3). We used RNA-seq to investigate the likely molecular mechanisms that overexpressed rSmSPL2 employed to influence root architectures. Transcriptome analysis showed that the DEGs in rSmSPL2-OE2 and rSmSPL2-OE7 were significantly enriched in the KEGG pathway of zeatin biosynthesis (Figures S3 and S4).
Cytokinin phytohormones have been known to inhibit root growth, since they were first discovered [35]. Cytokinin was observed to inhibit the adventitious roots formation near the cut end of cucumber hypocotyls [36]. The exogenous application of cytokinin inhibits the initiation of lateral roots and results in fewer lateral roots in Arabidopsis [37]. Mutants with increased cytokinin signaling have fewer lateral roots [38], whereas those with decreased cytokinin sensitivity have an increased number of lateral roots [39]. Transcriptome data indicated that cytokinin biosynthesis and the signal transduction pathway was significantly activated in transgenic lines (Figure 4), which is potentially one the most important reasons for morphological changes in the roots.
It is well documented that cytokinin and auxin interact antagonistically to regulate the size of the root apical meristem and the formation of lateral roots [40]. Auxin plays critical role in the control of the development and architectures of root systems and is a positive regulator of lateral root formation [41]. During lateral root development, auxin regulates cytokinin biosynthesis and signaling pathways, while cytokinin affects the pathways of auxin biosynthesis and transport [40]. Auxin biosynthesis, polar transport, and signal transduction are pivotal processes for lateral and adventitious root development [42]. Transcriptome analysis hinted that the overexpression of rSmSPL2 strongly inhibited auxin signaling in the roots of transgenic lines (Figure 4). Additionally, the endogenous cytokinin levels in rSmSPL2-OE2 and rSmSPL2-OE7 was significantly increased, while the endogenous auxin levels were dramatically decreased when compared with the control. Taken together, we speculated that the morphological changes of the roots in transgenic lines were to a large extent attributed to the alteration of auxin and cytokinin biosynthesis and signaling.

Experimental Materials and Hormone Treatments
S. miltiorrhiza seeds were collected in the experimental field at Shaanxi Normal University. The seeds were sterilized, followed by germination on the MS medium to obtain sterile seedlings, which were used for genetic transformation, as previously described [43]. S. miltiorrhiza, A. thaliana ecotype Columbia-0, Nicotiana tabacum were planted under the following conditions: 22 • C under extended daylight conditions (16 h light: 8 h dark).
One-year-old S. miltiorrhiza plants growing in the experimental field at the flowering stage were collected and its different parts, including tap roots, lateral roots, stems, leaves, pistils, stamens, corollas, and calyxes, were separately frozen in liquid nitrogen for RNA extraction.

RNA Extraction and qRT-PCR
Total RNA was extracted using Tissue Total RNA Isolation Kit (Vazyme, Nanjing, China), followed by a reverse transcription using HiScript II Reverse Transcriptase (Vazyme, Nanjing, China). SYBR green qPCR Mix (Vazyme, Nanjing, China) was used for the qRT-PCR experiment on the LightCycler ® 96 real-time PCR assay system (Roche, Switzerland). SmUbiquitin served as the reference gene.

SmSPL2 Promoter Analysis in Arabidopsis
The SmSPL2 promoter region, 1380 bp fragment upstream of the translation initiation site, was cloned and inserted into the pCAMBIA1391Z to drive the expression of GUS. The Arabidopsis-expressing ProSmSPL2::GUS was acquired via the reported method [44]. GUS histochemical staining was performed on T2 generation transgenic Arabidopsis at various development periods, according to a previously described protocol [45].

Subcellular Localization Assay of SmSPL2
To study the subcellular localization of the SmSPL2, the open reading frame (ORF) of SmSPL2 without termination codon were recombined into vector pEarleyGate103 to construct pEarleyGate103-SmSPL2 by Gateway cloning technology [46]. The pEarleyGate103 and the recombinant vector pEarleyGate103-SmSPL2 were transformed into onion epidermal cells, respectively, and the GFP signal was recorded, as we previously described [25].

SmSPL2 Transcription Activation Assays
The full-length ORF of SmSPL2 was recombined into the pGBKT7 vector to generate recombinant vector pGBKT7-SmSPL2 through the gateway cloning system [46]. The pGBKT7-SmSPL2 was transformed into yeast strain AH109 using a lithium acetate mediated method [47]. The transformed cells were grown on SD/-Trp, and further cultured on SD/-Trp/-Ade/-His/X-α-gal medium to determine the transactivation activity of SmSPL2, as we previously described [25].

Generation of Transgenic S. miltiorrhiza Plants
The miR156-resistant SPL2 (rSmSPL2) was obtained through a three step PCR method (as described previously) without changing the protein sequence [48]. The primers designed for site-directed mutagenesis are listed in Table S1. For the SmSPL2 and rSm-SPL2-overexpression vectors, their full-length ORFs were amplified and recombined into pEarleyGate202 to generate pEarleyGate202-SmSPL2 and pEarleyGate202-rSmSPL2 vectors by the gateway cloning system [46]. Transgenic S. miltiorrhiza plants that overexpressed SmSPL2 or rSmSPL2 were obtained by Agrobacterium-mediated transformation [43].

Transcriptome Analysis
Total RNAs were extracted from the roots of two-month-old rSmSPL2-OE2, rSmSPL2-OE7, and control plants using the HiPure HP Plant RNA Mini Kit (Guangzhou Meiji Biotechnology Co., Ltd., Guangzhou, China). Three plants of each line (rSmSPL2-OE2, rSmSPL2-OE7, and control) were mixed for RNA-seq by Novogene (Beijing, China). Genes with a Q value of < 0.05, a false discovery rate (FDR) < 0.001, and an absolute log 2 foldchange (log 2 FC) > 1.0 were identified as DEGs. We mapped all DEGs to the KEGG database terms to search for significantly overrepresented metabolic or signal transduction pathways.

Determination of Cytokinin and Auxin Concentrations
The fresh roots of one-month-old transgenic lines (rSmSPL-OE2 and rSmSPL-OE7) and control plants were used to calculate endogenous cytokinin and auxin concentrations by the Plant CTK/IAA ELISA Kit (mlbio, Shanghai, China).

Y1H Assays
The specific primers containing SmaI and SacI sites were used to amplify SmSPL2 and insert it into pGADT7-Rec2 to obtain pGADT7-SmSPL2. The pHIS2-pTAT1, pHIS2-pPAL1, and pHIS2-p4CL9 vectors have been obtained in our previous studies [49]. The recombinant vectors were co-transferred into yeast strain Y187 using the lithium acetate mediated method [47]. The interaction between the SmSPL2 and the promoter region was tested, according to the method previously reported [15].

Transient Transcriptional Activity (TTA) Assay in Tobacco
The full-length ORF of SmSPL2 were inserted into the SmaI and KpnI sites of pGreenII-62SK to obtain the effector vector pGreenII-SmSPL2. The pSm4Cl9-LUC, pSmTAT1-LUC, and pSmPAL1-LUC reporter vectors were obtained in our previous study [50]. The pair of effector and reporter vector were co-transferred into tobacco leaves, then monitor the transient expression of firefly luciferase (LUC) and renillia luciferase (REN), according to the previous protocol [34].

Conclusions
The overexpression of SmSPL2 or rSmSPL2 not only significantly reduced the accumulation of SalB but also strongly influenced root architectures, with longer lengths, fewer adventitious roots, decreased lateral root densities, and increased fresh weights. Transcriptome analysis indicated that the overexpression of rSmSPL2 influenced root morphologies by triggering cytokinin biosynthesis and signal transduction, while inhibiting auxin biosynthesis and signaling. SmSPL2 negatively regulates SalB and RA biosynthesis by directly binding to the promoters of Sm4CL9, SmTAT1, and SmPAL1 and suppressing their expression. SmSPL2 is a potential gene for the genetic engineering breeding of S. miltiorrhiza.