RNA Sequencing Reveals Transcriptomic Changes in Tobacco (Nicotiana Tabacum) Following NtCPS2 Knockout

Background: Amber-like compounds form in tobacco (Nicotiana tabacum) during leaf curing and impact aromatic quality. In particular, cis-abienol, a polycyclic labdane-related diterpenoid, is of research interest as a precursor of these compounds. Glandular trichome cells specically express copalyl diphosphate synthase (NtCPS2) at high levels in tobacco, which, together with NtABS, are major regulators of cis-abienol biosynthesis in tobacco. Results: To identify the genes involved in the biosynthesis of cis-abienol in tobacco, we constructed transgenic tobacco lines based on an NtCPS2 gene-knockout model, using CRISPR/Cas9 genome-editing technology to inhibit NtCPS2 function in vitro. In mutant plants, cis-abienol and labdene-diol contents decreased, whereas gibberellin and abscisic acid (ABA) contents increased compared with those in wild-type tobacco plants. RNA sequencing analysis revealed the presence of 9,514 differentially expressed genes (DEGs; 4,279 upregulated, 5,235 downregulated) when the leaves of wild-type and NtCPS2-knockout tobacco plants were screened. Among these DEGs, the genes encoding cis-abienol synthase, ent-kaurene oxidase, auxin/ABA-related proteins, and transcription factors were found to be involved in various biological and physiochemical processes, including diterpenoid biosynthesis, plant hormone signal transduction, plant-pathogen interactions, etc. Conclusions: Our ndings provide clues to the molecular regulatory mechanism underlying NtCPS2 activity, allowing for a better understanding of the interactions among related genes in tobacco.

from the angiosperm Cistus creticus subsp. creticus was rst analyzed through prokaryotic expression and dephosphorylation. Then, gas chromatography-mass spectrometry (GC-MS) analysis revealed that CPS2 catalyses the formation of 13(E)-labden-8-ol-15-diphosphate, implying that CPS2 is involved in the biosynthesis of cis-abienol [14]. In gymnosperms, cis-abienol synthase (ABS/KS) contains both class I and class II functional domains, as shown by cloning and characterizing the gene from balsam r (Abies balsamea) via transcriptome sequencing [15]. Sallaud et al. [16] cloned NtCPS2 and NtABS from tobacco and showed that both genes are involved in the biosynthesis of cis-abienol, which involves two steps. First, CPS-like catalytic activity yields 8-hydroxy-copalyl diphosphate with a normal con guration, which can then be converted to cis-abienol by the NtABS product [16][17][18]. No other diterpenoid synthase has been reported to be able to use 8-hydroxy-copalyl diphosphate as a substrate in dicotyledons to date. In addition, promoter analysis of NtCPS2 showed that it could drive the expression of the GUS gene in glandular hairs [16,19,20]. The identi cation of NtCPS2 and NtABS is of great signi cance for breeding high-quality tobacco and future microbial metabolic engineering. From this knowledge base, other diterpenoid-synthesising genes can be cloned and identi ed.
Among tobacco types, cis-abienol accumulates at different levels. It is mainly found in oriental and cigar tobacco but not in ue-cured tobacco, Burley tobacco, or Maryland tobacco [1,16,21]. To study the variation in cis-abienol content among different types of cultivated tobacco, 157 varieties of tobacco with or without cis-abienol were selected, and the expression levels of NtCPS2 and NtABS were analyzed [16]. NtABS cDNA sequences did not differ among tobacco varieties, but two distinct polymorphisms were found in NtCPS2 cDNA: an 8-bp insertion at position 275 and a G-T transversion at position 292 of NtCPS2. Both of these result in a stop codon, which leads to early termination and shortening of the encoded peptide chain. Because the encoded protein loses its active site, it also loses its original function [16]. Thus, NtCPS2 is key for cis-abienol biosynthesis. However, the mechanism by which the metabolic pathway of labdanoid diterpenoids regulate NtCPS2 in tobacco and the effects of NtCPS2 knockout on other metabolic pathways are still unknown.
In this study, we used CRISPR/Cas9 gene-editing technology to knock out NtCPS2. The CRISPR/Cas9 NtCPS2 expression vector was constructed from the high-aroma strain 8306, and transformed NtCPS2knockout plants were obtained. A high-throughput RNA sequencing (RNA-seq) technique was used to compare expression pro les between mutant and 8306 plants. Sequencing results were veri ed using uorescence quantitative polymerase chain reaction (PCR), and physiological changes and transcriptional inheritance were analyzed. By elucidating the function of the NtCPS2 gene and the molecular mechanisms underlying the regulation of its related genes, high-aroma tobacco varieties can be cultivated.

Targeted Mutagenesis of NtCPS2 by CRISPR/Cas9 in Tobacco
In this study, Cas9 gene was optimised to edit the tobacco genome. To generate Cas9-induced mutations in NtCPS2, a vector was designed that harboured chimeric guide RNA (gRNA) to guide Cas9 to target sequences where it bound and cleaved genomic DNA to generate double-strand breaks [22]. Two target sites of CPS2 were selected ( Supplementary Fig. 1). The gRNA for each target site, which was generated by overlap-extension PCR. Cas9 were subcloned into a single expression vector [23]. The Cas9 and gRNA expression cassette was located in one expression vector (pRGEB32-Cas9-NPT II-CPS2-gRNA). Through the Agrobacterium tumefaciens-mediated method, 36 transformed regenerated plants in the T 0 generation were obtained. After ampli cation with target-speci c primers, all positive samples were sequenced to assess the mutation e ciency. Of 36 plants, eight were transgenic lines. Most of the transgenic lines had a single-base insertion of A, C, or T at Target 2. Thus, as the peptide chain was being formed, the stop codon was encountered early in the process, and the translated amino-acid chain was greatly shortened. To test the heritability of the mutations, homozygous transgenic plants in the T 0 , T 1 , and T 2 generations were analyzed. Detailed information about the homozygous T 2 plants (M1-M9) is shown in Fig. 1A, and these plants were used for the following experiments.

NtCPS2 Knockout Affects cis-Abienol Content
To verify whether the gene mutations caused changes in gene expression, quantitative real-time PCR (qRT-PCR) was used to detect expression levels of NtCPS2 in the leaves of mutant and wild-type (8306) plants. The results showed that NtCPS2 expression decreased signi cantly in transgenic plants compared to wild-type plants (Fig. 1B). To detect changes in cis-abienol content in the leaves, exudates were collected from the mutant plants and analyzed using GC-MS. Contents of cis-abienol also decreased signi cantly in mutant plants compared to wild-type plants (Fig. 1C). The results indicate that NtCPS2 is one of the key genes regulating the cis-abienol biosynthesis pathway, and NtCPS2 knockout results in low levels of cis-abienol biosynthesis and accumulation. NtABS is another key gene involved in cis-abienol biosynthesis [16]. A previous study reported that cis-abienol was detected in plants expressing both NtCPS2 and NtABS but not in plants expressing just one of the two genes [16]. NtABS expression was weak in the mutant plants compared to the wild-type plants, implying that NtCPS2 knockout negatively in uenced NtABS expression. This is possibly because NtCPS2 is located upstream of NtABS in the cisabienol biosynthesis pathway.

NtCPS2 Has a Minor Effect on the Development of Glandular Trichomes in Tobacco
Agronomic characteristics were analyzed to assess the mutant phenotypes ( Fig. 2 and Supplementary  Fig. 2). Differences in plant height, internode length, number of leaves, and stem girth between mutant and wild-type plants did not exhibit the same trend. T2-2 mutants had longer internodes and wider stems than other mutants and wild-type plants, whereas all mutants except for T2-1 had shorter plant heights than the wild-type plants ( Supplementary Fig. 2). These results indicate that NtCPS2 expression does not strongly affect tobacco plant morphology. As NtCPS2 is speci cally expressed in glandular cells [16], the morphology of the glandular trichomes on the largest leaf of each plant was examined. Both the length and width of the largest leaf were signi cantly shorter in mutant plants compared to wild-type plants. The average diameter of glandular trichomes was smaller in mutant plants, especially T2-1, whereas both longer and shorter glandular trichomes were observed in mutant plants compared to wild-type plants ( Fig. 2). Other trichome characteristics, such as numbers of long and short trichomes, did not differ signi cantly between mutant and wild-type plants (data not shown). Thus, in the absence of NtCPS2 expression in tobacco plants, the diameter of glandular cells and the area of the largest leaf decrease, but not the length of glandular trichomes. The T2-1 line was selected and used to pro le transcriptomic changes after NtCPS2 knockout in tobacco 8306.

Overview of Transcriptome Sequencing
To pro le gene expression after NtCPS2 knockout, RNA-seq libraries were constructed for the mutant and wild-type plants. Six samples of each line were sequenced, and 41.64 G of clean data was obtained. In total, 6.70-7.02 G of effective data was collected from each sample, with a Q30 distribution of 94.29-94.91%, and an average GC content of 43.41%. More than 95.58% of the clean reads had quality scores that met the Q30 criterium (probability of base-calling error = 0.1%) [24]. Furthermore, the GC content ranged from 43.15-43.66%. The sequencing data are summarised in Table 1.

Analysis of Differentially Expressed Genes (DEGs) and Their Functions
Volcano plots were used to assess the variation in gene expression between mutant and wild-type plants ( Fig. 3A). In total, 9,514 DEGs were detected. Among them, 4,279 were upregulated and 5,235 were downregulated in the transgenic tobacco plant compared to 8103 using the thresholds p < 0.05 and |log 2 (fold change [FC])| > 1 (Fig. 3B).
Kyoto Encyclopedia of Genes and Genomes (KEGG) and Gene Ontology (GO) pathway analyses of the differentially expressed mRNAs were performed to determine the functions of the DEGs. The 20 most signi cantly enriched pathways (lowest q values) according to KEGG metabolic pathway annotation were examined in detail (Fig. 4A). Based on GO analysis, the DEGs were most likely to be associated with biological processes (Fig. 4B) and cellular components (Fig. 4C). A large percentage of the DEGs was assigned to the categories metabolic process, cellular process, catalytic activity, binding, and singleorganism process, with only a few genes assigned to channel regulator activity, cell killing, and protein tag. The DEGs involved in the pathways for diterpenoid biosynthesis, plant hormone signal transduction, and plant-pathogen interactions were analyzed in detail.

Validation of Selected DEGs Using qRT-PCR
To validate the RNA-seq data, 12 DEGs, including genes involved in cis-abienol and gibberellin (GA) biosynthesis as well as genes related to plant-pathogen interactions and other hormone signalling pathways, were selected randomly for qRT-PCR analysis. The gene-expression patterns as determined using qRT-PCR were consistent with those determined via transcriptome sequencing (Fig. 5). FC values differed between qRT-PCR and RNA-seq, possibly due to differences in the sensitivity of each method or because different samples were used for qRT-PCR and RNA-seq.

Expression Levels of Genes Related to cis-Abienol Biosynthesis Decreased Signi cantly in Mutant Plants
NtCPS2 (Nitab4.5_0001630g0010) was identi ed as a DEG via RNA-seq, and its expression level was 9.27-fold lower in the mutant compared to the wild type. The expression level of another key gene related to cis-abienol biosynthesis, NtABS (Nitab4.5_0015240g0010), also decreased 2.43-fold in the mutant. NtCPS2 and NtABS operate in succession to synthesise cis-abienol [16]. When both NtCPS2 and NtABS are expressed, cis-abienol is synthesised and can be detected in plants. However, cis-abienol synthesis does not occur in plants that express only one of these genes [16]. NtCPS2 encodes 8-hydroxy-copalyl diphosphate synthase, which synthesises 8-hydroxy-copalyl diphosphate, and NtABS encodes a kaurene synthase-like (KSL) protein, abienol synthase, which uses 8-hydroxy-copalyl diphosphate to produce cisabienol. Our results indicate that weak expression of NtCPS2 directly or indirectly results in a decrease in the expression level of NtABS and consequently, low cis-abienol contents. Another putative cis-abienol synthase (Nitab4.5_0008024g0010) was also found to be downregulated in the mutant, indicating that this enzyme may have the same substrate as NtABS and thus be involved in the cis-abienol biosynthesis pathway. By contrast, other putative cis-abienol synthases, including Nitab4.5_0004164g0070 and Nitab4.5_0004164g0010, were found to be upregulated in the mutant. These two enzymes may have other functions in tobacco. Other DEGs involved in the cis-abienol biosynthesis pathway were also identi ed (based on KEGG analysis) and had lower expression levels in the mutant (Fig. 4). This included KSL4 (Nitab4.5_0000029g0200, FC = 2.43) and genes predicted to encode ent-kaur-16-ene synthase (Nitab4.5_0002280g0060, FC = 5.61; and Nitab4.5_0002862g0030, FC = 1.57). As with KSL4, NtABS is a KSL gene (Table 2). Hence, KSL4 and genes that putatively encode ent-kaur-16-ene synthase may be involved in cis-abienol biosynthesis. This needs to be veri ed in future work.

GA Biosynthesis Increased Signi cantly in Mutant Plants
According to diterpenoid biosynthesis pathways, the same substrate, geranylgeranyl pyrophosphate (GGPP), is used for cis-abienol and GA synthesis. In this study, most of the DEGs involved in GA biosynthesis were strongly upregulated in the mutant, including KAO2 (Nitab4.5_0001476g0100, FC = 16.67), KS1 (Nitab4.5_0004164g0010, FC = 2.76), and CPS1 (Nitab4.5_0010312g0010, FC = 12.14) ( Table 2). From these genes, ent-copalyl diphosphate synthase 1 (encoded by CPS1) and ent-kaurene synthase (encoded by KS1) were found to separately catalyse the synthesis of ent-kaurene from GGPP. However, KO (encodes ent-kaurene oxidase, which converts ent-kaurene to kaur-16-en-18-oate) expression was downregulated in the mutant. DEGs participating in the latter stages of the pathway, such as KAO2 and GA20 OX2 , were upregulated compared to the wild type. KO and KAO belong to the CYP701A, P450, and CYP88A clade. Accordingly, KAO is localised in the endoplasmic reticulum, whereas KO is localised in both the endoplasmic reticulum and plastid envelope [25]. The differential expression of KO1 and KAO2 in response to NtCPS2 knockdown was explored further. GA contents in mutant plants were also analyzed via GC-MS. The results showed that the GA contents in transgenic plants were signi cantly lower than those in wild-type plants (Fig. 6). GA12 is considered the precursor of all GAs in plants [26], and other GA forms are produced through oxidative steps catalysed by GA12. Genes involved in the production of these GA forms were up-and downregulated in the mutant.

Changes in Abscisic Acid (ABA) Biosynthesis and Signal Transduction in Mutant Plants
In carotenoid biosynthesis pathways, GGPP is also a substrate for ABA synthesis. RNA-seq analysis showed that four PSY genes (encoding phytoene synthases) were upregulated at the rst step, which involves GGPP, in the mutant compared to the wild type. PSY is a transferase enzyme that is involved in the biosynthesis of carotenoids. It catalyses the conversion of GGPP to phytoene. Two genes encoding LCYs (lycopene epsilon cyclases) were also upregulated in the mutant at the next step. These results indicate that NtCPS2 knockout positively regulates ABA synthesis, likely because substrate competition decreases. In addition, two ABA 8'-hydroxylases, which are involved in ABA degradation, were downregulated in the mutant. In the ABA signal transduction pathway, ve of six ABA receptors (PYLs), which inhibit the expression of protein phosphatase 2C, were upregulated in the mutant. At the next step, serine/threonine-protein kinase expression was upregulated in the mutant. This might have been related to stress responses and stomatal opening and closure in tobacco leaves.

Transcriptomic Analysis of Genes Involved in Plantpathogen Interactions
In plants, cis-abienol may participate in insect resistance and disease resistance [27]. Plant resistance to pathogen attack can induce the accumulation of pathogenesis-related proteins (PRs) that contribute to systematically acquired resistance. In this study, PRs were identi ed through RNA-sEq. Of 17 PRs, 14 (82.35%) were signi cantly upregulated in the mutant than in the wild type, including genes that encode PR proteins 1A, B, and C (Table 3). Among the 17 families of PRs, PR 1-5, 9-11 and 17, were related to the acquisition of defence against the pathogen infections. In addition, calcium is involved in regulating diverse physiological processes as a second messenger [28]. Results of transcriptomic analysis revealed that, 15 of 19 CDPKs and most CAM/CML were signi cantly downregulated upon NtCPS2 knockout and low content of cis-abienol, which disturbed the balance among active oxygen species, including rubidium hydroxide, reactive oxygen species, and nitric oxide synthase.

Discussion
Aromatic characteristics of tobacco are improved by cis-abienol, which belongs to the labdane diterpenoid family. Although the genes encoding the enzymes participating in the two steps of cis-abienol biosynthesis have been cloned in tobacco [16], the function and regulatory mechanism of NtCPS2 are less well-understood. By knocking out NtCPS2, whose product catalyses the rst reaction in the cisabienol biosynthesis pathway, we were able to examine how cis-abienol biosynthesis and other related metabolic pathways are controlled. The regulatory network is shown in Fig. 7.

NtCPS2 Plays a Limited Role in the Biosynthesis of cis-Abienol and Other Terpenoids
Mutations in NtCPS2 were previously reported to be strongly correlated with the absence of cis-abienol and labdene-diol in tobacco or a decrease in their levels [16]. In N. sylvestris, cis-abienol is accumulated when both NtCPS2 and NtABS are expressed [16]. In this study, we generated NtCPS2-knockout tobacco lines using the CRISPR-Cas9 method. In mutant plants that weakly express NtCPS2, the levels of cisabienol produced decreased (Fig. 1). NtABS is involved in the second step of cis-abienol biosynthesis, and its expression levels also decreased (Fig. 5A). The decreased expression of both these genes might have resulted in low levels of the intermediate 8-hydroxy-copalyl diphosphate being accumulated and the cessation of cis-abienol production downstream (Fig. 1C). The results indicate that NtCPS2 plays a key role in cis-abienol biosynthesis; thus, downregulated gene expression leads to an inactivation of the cisabienol biosynthesis pathway.
The precursor GGPP, which participates in the rst step of the pathway, is a common precursor for the biosynthesis of not only diterpenoids (including cis-abienol and labdene-diol) but also GA, carotenoids (including ABA), and the phytolchain of chlorophyll [29]. When NtCPS2 is absent, GGPP is not catalysed to produce 8-hydroxy-copalyl diphosphate, and other reactions that use GGPP as a substrate are enhanced. During GA biosynthesis in Arabidopsis, GGDP is converted to ent-kaurene in a two-step reaction catalysed by CPS and KS, which are encoded by AtCPS and AtKS, respectively [30,31]. In this study, NtCPS1 and NtKS1 expression levels were upregulated after NtCPS2 knockout, and GA production, which occurs downstream, increased in the leaves of mutant plants (Fig. 6). In terms of carotenoid biosynthesis, genes (including phytoene synthase 2 and lycopene epsilon cyclase) involved in converting GGPP to phytoene were upregulated after NtCPS2 knockout. Overall, reactions that consume GGPP as a substrate were enhanced. The results indicate that NtCPS2 knockout also contributes to the biosynthesis of other terpenoids depending on the same substrate. Future studies can verify this hypothesis by overexpressing NtCPS2 and/or NtABS.

Plants with Mutations in NtCPS2 Still Exhibit Wild-type Morphology
Diterpenoids such as cembranoid diterpenes and labdanoid diterpenes from tobacco-leaf exudates signi cantly in uence cigarette-smoke characteristics and avour pro les [2,3]. To our knowledge, no previous studies on the effects of cis-abienol on the growth and development of tobacco plants have been reported. We found that the mutant and wild-type morphology did not differ much, except for the diameter of glandular trichomes. This indicates that NtCPS2 knockout and the subsequent decrease in cis-abienol do not affect tobacco plant morphology. However, the contents of other chemical substances (including GA and ABA) may change in mutants. Mutants had lower levels of GA and did not exhibit the GA-overdose morphology. In Arabidopsis, CPS-and/or KS-overexpressing mutants also did not exhibit the GA-overdose morphology [32]. This suggests that levels of bioactive GA in these plants likely did not change. Transcriptomic analysis showed that the expression of GA20 OX2 was upregulated, whereas that of GA2 OX4 , GA2 OX2 , and GA2 OX1 was downregulated ( both exhibited aspects of GA-overdose morphology [33]. The differential regulation of GA20 OX , GA2 OX4 , GA2 OX2 , and GA2 OX1 might resulted in different types of GA being accumulated at different levels; thus, overall levels of bioactive GA may not change much, and plants may not exhibit a GA-underdose morphology.

cis-Abienol May Participate in Tobacco Disease Resistance
Labdanoid diterpenes may exhibit defence-related activities such as antifungal [34] and insecticidal [7,8] activities [27,35]. The application of cis-abienol to the roots of tobacco, tomato, and Arabidopsis at a concentration of 100 µmol/L can induce the expression of resistance genes and inhibit bacterial wilt disease [27]. In vitro experiments showed that concentrations of cis-abienol and related diterpenoids in the range 0.01-100 mg/kg can inhibit the growth of Phytophthora nicotianae in tobacco. However, upon examination of the cis-abienol biosynthesis pathway, we conclude that the accumulation of cis-abienol is not related to disease resistance in tobacco plants. Under eld conditions, cis-abienol does not have an effect on diseased leaves in tobacco. Therefore, it is unclear whether genes related to cis-abienol synthesis contribute to resistance against P. nicotianae in tobacco. A key defence response to pathogen attack in plants is the induction and accumulation of various PR proteins, which also contribute to systematically acquired resistance [36,37]. The PR-1, PR-2 [38], PR-3, PR-4, PR-5 [39], PR-9 [40], PR-10 [41], PR-11 [42], and PR-17 [43] families are associated with acquired resistance to pathogen infections.
Among the genes encoding these PR proteins, PR-1 is generally considered a marker gene for disease resistance [44]. In this study, PR-related genes were both signi cantly up-and downregulated in the mutant plants (Table 3), implying that cis-abienol may participate in TbCSV resistance in tobacco plants. Future research could assess disease resistance in NtCPS2-knockout and -overexpressing mutants to clarify the contribution of cis-abienol to tobacco disease resistance.

Conclusions
In this study, a genome-wide transcription pro le was obtained for NtCPS2-knockout tobacco plants edited using CRISPR-Cas9. NtCPS2 is a key gene for cis-abienol biosynthesis in tobacco. Genes involved in the biosynthesis of cis-abienol, early metabolites of GA, and carotenoids (including ABA) were signi cantly differentially expressed after NtCPS2 knockout. The expression of PR-related genes also changed in response to low cis-abienol contents. Our ndings may be useful for further investigations of the molecular mechanisms associated with NtCPS2 gene function and those underlying the regulation of related genes. Additionally, our results can contribute to the development of high-aroma tobacco varieties.

Tobacco Plant Culture
The tobacco plant variety (N. tabacum cv. 8306) used in this study produces high-aroma, ue-cured tobacco with high levels of cis-abienol. Plants were grown on a farm with loamy tidal soil near Henan Agricultural University, Zhengzhou City, China (136º56′6″E, 35º9′5″N)

Vector Construction
Based on the mRNA sequences and corresponding genome sequences, two CRISPR target sites (Supplementary Table 1) were designed to improve gene-targeting e ciency. Target primers for PCR (Supplementary Table 2) were designed and synthesised. After primer synthesis, fragments containing the target sites were ampli ed using overlap-extension PCR. The ampli ed fragments were cloned into a CRISPR expression vector using a recombinant enzyme from Nanjing Novozan Biotechnology Co., Ltd.
(Nanjing, China). The CRISPR vector was electroporated into Escherichia coli, and positive clones were screened using colony PCR for Agrobacterium tumefaciens-mediated transformation and tobacco gene transformation.

Agrobacterium-mediated Transformation
Agrobacterium tumefaciens-mediated transformation was performed as follows: 5 µL of recombinant plasmid was mixed with 50 µL of competent Agrobacterium tumefaciens cells on ice for 30 min. Blank YEB medium was added, and the mixture was incubated at 28 °C for 12-13 h. Then, the mixture was transferred to YEB solid medium containing 50 mg/L of kanamycin and incubated at 28 °C for 36-48 h.
Mature tobacco seeds were sterilised by washing with 75% alcohol and 10% sodium hypochlorite and placed into a germination medium. The seeds were then grown under light for 45 days. Samples with a diameter of 0.5 cm were taken from leaves with a hole punch, transferred to a pre-culture medium, and incubated for 2 days under light. Agrobacterium tumefaciens was activated in the medium containing 50 mg/L of kanamycin. Leaf discs were infected with Agrobacterium tumefaciens, transferred to a coculture medium, and left for 3 days. Thereafter, the leaf discs were washed with sterilised distilled water and antibiotics in an aqueous solution. After the leaves were dried, they were transferred to a screening medium and cultured under light. After differentiation, they were transferred to a rooting medium.
Transformed plants were obtained via rooting culture and transplanted into soil after 1 month.

DNA Extraction and Sequencing for Detecting Mutations in the Target Gene
Lea ets were collected from each plant, and genomic DNA was extracted using a standard cetrimonium bromide protocol. NPTII speci c primers were used to detect successfully transformed plants via PCR (Supplementary Table 3). After con rming that the exogenous DNA fragment had been inserted, the primer 17KN48 was designed based on the NtCPS2 gene sequence and target-site location to detect positive plants using PCR. PCR ampli cation was performed in the following reaction volume: 1 µL DNA,

Measurements of Morphological Characteristics of Transgenic Tobacco Plants
Homozygous T 2 tobacco plants were selected, and morphological parameters including plant height, number of leaves, stem girth, internode length, and length and width of the largest leaf were measured at a leaf age of 60 days. The morphology of leaf glandular trichomes was also characterised. The largest leaves of each plant with the same age were detached, and the epidermis at the centre of each leaf was peeled off to examine the glandular trichomes using an Axioplan 2 microscope (Carl Zeiss, Oberkochen, Germany). The numbers of long and short glandular trichomes were counted and their lengths and diameters were measured. Each seedling had an average of approximately 100 glandular trichomes.

Analysis of Diterpenoids in Leaf Exudates Using GC-MS
Leaf exudates were sampled from fresh tobacco leaves, and 1:1 portions of the samples were directly injected into a 6890 N gas chromatograph coupled to a 5973 N mass spectrometer (Agilent Technologies, Santa Clara, CA, USA) for GC-MS analysis. Tobacco diterpenoids were identi ed based on their mass spectra.

RNA-seq Analysis
Total RNA was extracted from frozen leaf samples using TRIzol reagent (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's protocol. RNA integrity was assessed using agarose gel electrophoresis, and RNA purity was checked using a NanoPhotometer® spectrophotometer (Implen, Munich, Germany).
RNA concentration was quanti ed with a Qubit®2.0 Flurometer using a Qubit® RNA Assay kit (Life Technologies, Carlsbad, CA, USA  [47], and an absolute log 2 (FC) value > 1 and a corrected p-value < 0.05 were set as the thresholds for DEGs for subsequent analysis.
DEGs were further annotated using GO functional enrichment analysis. GO terms with corrected p-values < 0.05 were considered to be signi cantly enriched for a given DEG. Clusters of orthologous groups and pathway analyses were performed using KEGG (http://www.genome.jp/kegg) analytical tools. We used the clusterPro ler R package [48] to test the statistical enrichment of KEGG pathways for the DEGs. Data was showed in Supplementary Table 5.
Validation of DEGs Using qRT-PCR The differential expression of 30 genes between wild-type and transgenic tobacco leaf samples was con rmed using qRT-PCR analysis with three biological replicates per sample. Primer sets for the DEGs were designed using Primer Premier 5.0 (Premier Biosoft, San Francisco, CA, USA) and synthesised by Invitrogen Trading (Shanghai) Co., Ltd. (China). All primer sequences are listed in Supplementary Table 4. RNA isolation, cDNA synthesis, qRT-PCR, and statistical analyses were performed as previously described [45]. Expression levels of the DEGs were normalised to that of the internal control gene L25 [46].

Statistical Analyses
Data are presented as the means ± standard deviations. Two-sample t-tests were used to compare the means between two treatments. Comparisons across multiple treatments were performed using one-way analysis of variance followed by Tukey's honestly signi cant difference post-hoc test with SPSS v19   Differentially expressed genes (DEGs) were screened using the criteria fold change ≥ 2.0 and false discovery rate < 0.05. Signi cant differences in expression were observed for 9,514 genes, as represented by a volcano plot (A) and heat map (B).   Gibberellin contents in the leaves of NtCPS2-knockout (M1 and M2) and wild-type (8306) tobacco.
Different lowercase letters denote signi cant differences among plant strains (p < 0.05).