Simultaneous Knockout of Tk1‐SST and Tk1‐FFT via CRISPR/Cas9 Enhances the Natural Rubber Accumulation in Taraxacum kok‐saghyz

ABSTRACT

Taraxacum kok‐saghyz (TKS) synthesises natural rubber (NR) and inulin using sucrose as a carbon source. However, molecular mechanisms regulating inulin and NR accumulation remain largely unclear. Here, we report the generation of double‐gene homozygous mutants, 1‐sst1‐fft, by simultaneously knocking out two key genes responsible for inulin biosynthesis (Tk1‐SST and Tk1‐FFT) using CRISPR/Cas9 technology. The 1‐sst1‐fft mutants exhibited significant increases in rosette leaf number, flower number, leaf area, whole‐plant biomass and seed set. Moreover, inulin biosynthesis was abolished in 1‐sst1‐fft, leading to significant changes in sugar composition, particularly a marked increase in sucrose levels. Notably, NR accumulation more than doubled, with no significant change in molecular weight and most terpenoid accumulation also increased in 1‐sst1‐fft, both being positively correlated with sucrose levels. For the first time, this study reports the generation of an inulin synthesis‐deficient mutant in plants, emphasising the essential roles of 1‐SST and 1‐FFT in regulating carbon partitioning and, consequently, modulating important traits and metabolite accumulation. Transcriptomic analysis revealed fundamental genes involved in sucrose metabolism, sugar signalling and transport and NR elongation were significantly upregulated in 1‐sst1‐fft; accompanied by enhanced enzymatic activities of sucrose–phosphate synthase and invertase. These findings demonstrate that blocking inulin biosynthesis in 1‐sst1‐fft redirects sucrose towards NR biosynthesis, highlighting the dual role of sucrose as both a carbon source and signalling molecule in modulating plant growth and development and metabolite synthesis. Our study provides a successful approach to enhancing NR accumulation by modifying carbon allocation in TKS, offering novel insights into high‐yield breeding strategies.

1. Introduction

Inulin, a fructose‐based polysaccharide classified within the fructan family, consists exclusively of (2,1)‐linked fructosyl units and a terminal glucose residue (Van Laere and Van Den Ende 2002). It is predominantly synthesised in storage organs of dicotyledonous plants from families such as Asteraceae, Campanulaceae and Boraginaceae (Hendry 1993). Inulin functions as a reserve carbohydrate, supporting energy demands during germination and regrowth. Beyond its role as a carbon sink, inulin contributes to osmotic regulation, thereby enhancing plant tolerance to abiotic stresses such as drought and cold (Ritsema and Smeekens 2003; Van den Ende 2013). Inulin biosynthesis is catalysed by two key enzymes: sucrose: sucrose 1‐fructosyltransferase (1‐SST) (EC 2.4.1.99) and fructan:fructan 1‐fructosyltransferase (1‐FFT) (EC 2.4.1.100) (Jefford 1968; Lüscher et al. 1996). Degradation of inulin into free sucrose and fructose is catalysed by fructan 1‐exohydrolase (1‐FEH) (EC 3.2.1.80) (De Roovera et al. 1999; Jefford 1968). Simultaneous heterologous expression of 1‐SST and 1‐FFT genes from globe artichoke ( Cynara scolymus ) has been shown to synthesise the full spectrum of inulin molecules in potato (Hellwege et al. 2000). Transgenic ryegrass lines expressing heterologous 1‐SST and 6G‐FFT/1‐FFT genes from onion exhibited a significant increase in fructan accumulation with no morphological changes (Gadegaard et al. 2008). Transgenic plants overexpressing 1‐SST and 1‐FFT genes exhibit enhanced fructan production, often accompanied by increased stress tolerance (Kawakami et al. 2008; Diedhiou et al. 2012; Hellwege et al. 2000; Van den Ende 2013). Combinational transformation of three wheat genes encoding fructan biosynthesis enzymes increased fructan content and abiotic stress tolerance in tobacco, with tolerance levels correlating to gene combinations and fructan/soluble sugar contents in the transgenic plants (Bie et al. 2012). 1‐FFT is responsible for chain elongation, and mutations involving specific amino acids can significantly reduce 1‐FFT enzymatic activity, leading to decreased inulin polymerisation in Viguiera discolour (De Sadeleer et al. 2015). Species within the Asteraceae family store species‐specific fructan patterns, characterised by varying degrees of polymerisation, which are associated with differential affinities of 1‐FFTs for sucrose, fructose and 1‐kestose as acceptor substrates (Vergauwen et al. 2003). Salicylic acid (SA) and jasmonic acid methyl ester (MeJA) drought stress induced fructan synthesis by upregulating expression levels of 1‐FFT and 1‐SST in Agave tequilana , A. inaequidens and Cichorium (De Roover et al. 2000; Suarez‐Gonzalez et al. 2014). In Grey guayule ( Parthenium argentatum Grey), 1‐SST expression and fructan synthesis occur under abiotic stress conditions (Niu et al. 2017). Ectopic expression of Allium cepa Ac1‐SST in cotton improves drought tolerance and reduces yield loss under drought stress in the field (Liu et al. 2021). Transgenic rice lines expressing wheat‐derived fructosyltransferase genes accumulated high levels of fructan in mature leaves, showed enhanced chilling tolerance in seedlings, and exhibited no phenotypic abnormalities or fertility issues (Kawakami et al. 2008). Despite the established roles of 1‐FFT and 1‐SST in inulin synthesis, the molecular mechanisms linking these genes to carbon supply, metabolite synthesis and plant growth and development processes remain poorly understood due to the lack of relevant mutants.

Russian dandelion ( Taraxacum kok‐saghyz Rodin, TKS) is a perennial herbaceous plant belonging to the Composite family, originating from the Tian Shan Mountain region in China and Kazakhstan (Kirschner et al. 2013). The roots of TKS can synthesise high‐molecular‐weight natural rubber (NR), the structure and properties of which are comparable to those of NR from rubber trees ( Hevea brasiliensis ) (Schmidt et al. 2010). Moreover, TKS possesses numerous advantages, including wide adaptability, a short growth cycle and compatibility with mechanised planting and harvesting, rendering it a promising alternative crop for NR production. However, the low NR yield is the crucial factor limiting commercial TKS cultivation. NR and inulin are the two main metabolites in the roots of TKS, and sucrose is used as a carbon source for the synthesis of both these metabolites. The basic carbon skeleton of NR is the five‐carbon building unit isopentenyl pyrophosphate (IPP), which is produced via the mevalonic acid (MVA) pathway in the cytosol and the methylerythritol (MEP) pathway in plastids (Cherian et al. 2019). Inulin is the major reserve carbohydrate that is synthesised and stored in TKS roots, and its content can reach 25%–40% of the root dry weight (DW) (Arias et al. 2016; Ende et al. 2000). In contrast, NR is primarily synthesised in the laticifer of the root and accounts for 3%–28% of the DW, which is much lower than the inulin content. According to some studies, NR accumulation in TKS seems to be negatively correlated with inulin (Kreuzberger et al. 2016; Stolze et al. 2017). Overexpression of Tk1‐FEH promoted the degradation of inulin to fructose and sucrose, which was transported into the laticifer cells and used as a precursor for the synthesis of IPP, and subsequently increased the NR content in the roots of TKS (Stolze et al. 2017). Therefore, inhibiting inulin synthesis may be an effective way to improve the NR yield in TKS. Tk1‐SST and Tk1‐FFT, which are responsible for inulin synthesis, could be good candidate genes for suppressing inulin synthesis and increasing the NR yield in TKS. Although single‐gene mutants of Tk1‐FFT have been generated in TKS (Ariyaratne et al. 2023; Iaffaldano et al. 2016), the creation and characterisation of double‐gene homozygous Tk1‐SST and Tk1‐FFT mutants, as well as their regulatory effects on NR accumulation and plant growth and development, have not yet been reported. Therefore, elucidating the underlying molecular mechanisms of Tk1‐SST and Tk1‐FFT in regulating inulin and NR accumulation could greatly contribute to the domestication of TKS. In this study, we generated homozygous double‐knockout mutants, 1‐sst1‐fft, by simultaneously deleting Tk1‐SST and Tk1‐FFT using the CRISPR/Cas9 system. By comparing and analysing the differences in metabolite accumulation and transcriptome between the 1‐sst1‐fft mutants and wild‐type (WT), we systematically investigated the regulatory effects of these two genes on the accumulation of inulin, NR and other metabolites, as well as their molecular regulatory network.

2. Results

2.1. Knockout of Tk1‐SST and Tk1‐FFT via CRISPR/Cas9 and Generation of Double‐Gene Homozygous Mutants 1‐sst1‐fft

Tk1‐SST and Tk1‐FFT, which are responsible for inulin synthesis, were selected as targets for CRISPR/Cas9‐mediated editing for improving NR yield in our study. First, the coding sequence (CDS) of Tk1‐SST and Tk1‐FFT was cloned from the receptor line 1152 and verified. The results indicated that both Tk1‐SST and Tk1‐FFT contain seven exons and six introns (Figure 1a). Multiple sequence alignment of the two genes from different TKS lines revealed that the third exon was the longest exon with the highest sequence conservation (Figure S1), which was selected for target design in Tk1‐SST and Tk1‐FFT. Two single guide RNAs (sgRNAs) were designed to target the third exon of Tk1‐SST and Tk1‐FFT (Figure 1a). The CRISPR/Cas9 constructs, in which Cas9 was driven by the ubiquitin promoter and the sgRNAs with target sequences (target‐sgRNAs) were driven by the Arabidopsis U3 and U6 promoters (Figure S2), were transformed into TKS to generate mutants.

FIGURE 1.

Generation and validation of double‐gene homozygous mutants 1‐sst1‐fft. (a) The gene structure of Tk1‐SST and Tk1‐FFT, and their sgRNAs. (b) Identification of mutagenesis in Tk1‐SST and Tk1‐FFT in double‐gene homozygous mutant lines using Hi‐TOM sequencing. The target sites are underlined; PAM sites are indicated with a yellow background; numbers represent the reads of sequences; D, deletion; LD, Large_deletion; I, insertion. Deleted nucleotides are represented by ‘–’ in red, and inserted nucleotides are highlighted in red. (c and d) Relative expression levels of Tk1‐SST (c) and Tk1‐FFT (d) in the homozygous mutant 1‐sst1‐fft plants. The different lowercases indicate a significant difference with p < 0.05. (e and f) Western blot analysis of Tk1‐SST and Tk1‐FFT protein expression in the roots of 1‐sst1‐fft mutants and WT. The roots of the 1‐sst1‐fft mutants and WT at ~200 DAC were used for detection. For western blot analysis, total proteins were extracted and probed with anti‐Tk1‐SST and anti‐Tk1‐FFT polyclonal antibody, respectively. β‐Actin was used as the internal reference. Each lane was loaded with an equal quantity of protein (50 μg).

A total of 119 T₀ plants were screened via Hi‐TOM sequencing, of which 66 contained effective mutations (deletions or insertions). The mutation frequency was 55.46%, including seven genotypes, namely, SsFF, SSFf, SSff, SsFf, Ssff, ssFf and ssff (Table S1; uppercase indicates the wild‐type, and lowercase indicates the mutant genotype, respectively). The mutation frequency of the Tk1‐SST gene was 51.26%, and that of the Tk1‐FFT gene was 43.70%. Five plants with the double‐gene homozygous mutation 1‐sst1‐fft (ssff‐10, ssff‐22, ssff‐36, ssff‐43 and ssff‐46) were obtained, and the frequency of the homozygous mutation was 4.20% (Figure 1b, Table S1 and Data S1). Since ssff‐10 and ssff‐17 contained the same mutation site, likely resulting from the same mutational event, they were classified as a single type. Owing to the incompatibility of TKS, the homozygous mutants were propagated via tissue culture. Due to the insufficient number of tissue culture seedlings derived from ssff‐22, as well as the loss of ssff‐43 and ssff‐46, most of the tissue culture seedlings used for subsequent analysis were obtained from the propagation of ssff‐10, ssff‐17 and ssff‐36. All individual plants obtained from tissue culture were verified by Hi‐TOM sequencing; and the confirmed mutant plants were subsequently used for further study (Table S2).

2.2. Tk1‐SST and Tk1‐FFT Expression and Their Protein Translation Were Suppressed in the Homozygous Mutants 1‐sst1‐fft

The Cas9 expression level was analysed in plants with different genotypes. The results demonstrated that Cas9 was highly expressed in all positively transformed plants, although its expression level varied greatly across different lines (Figure S3). Furthermore, we determined the expression levels of target genes in homozygous mutants and found that both Tk1‐SST and Tk1‐FFT were severely inhibited in the double‐gene homozygous mutants 1‐sst1‐fft compared to WT (Figure 1c,d). Among the 14 homozygous mutant plants tested, Tk1‐SST expression was markedly reduced to between 0.41% and 9.1% of WT levels. Similarly, Tk1‐FFT expression in the 1‐sst1‐fft mutant plants mirrored that of Tk1‐SST, decreasing to a range of 0.17% to 6.3% of WT levels. Moreover, western blot analysis of whole protein extracted from the roots confirmed strong expression of Tk1‐SST and Tk1‐FFT protein in the WT, whereas no corresponding signals were detected in any of the mutant lines (ssff‐10, ssff‐17 and ssff‐36) (Figure 1e,f). The absence of Tk1‐SST and Tk1‐FFT protein expression conclusively demonstrates successful gene knockout and translational inhibition in the 1‐sst1‐fft mutants, confirming the successful generation of double‐gene homozygous mutants targeting Tk1‐SST and Tk1‐FFT in TKS.

2.3. The Mutation Sites of 1‐sst1‐fft Were Stably Heritable Across Generations

To assess the genetic stability of the mutation site in the offspring, T₀ plants of the 1‐sst1‐fft mutants were crossed with two WT lines, 1153 and 1154, to generate T₁ progeny. The mutation types and mutation frequencies in the T₁ plants were subsequently confirmed via Hi‐TOM sequencing. In total, 107 T₁ plants derived from four individual plants of 1‐sst1‐fft were analysed (Table S3 and Data S2). The results revealed that the mutation sites of 1‐sst1‐fft could be stably inherited by T₁ plants, regardless of whether the WT was used as the female or male parent. Additionally, a few plants exhibited new mutation sites resulting from the integration of CRISPR/Cas9 constructs into the genome and/or the exclusion of mutations present at frequencies below 5% in the Hi‐TOM sequencing data from their T₀ plants. In terms of genotypes, all tested T₁ plants were SsFf heterozygotes, consistent with Mendelian inheritance patterns. Moreover, backcrossing was performed between the T₁ plants and the homozygous T₀ lines, resulting in the generation of T₂ plants. A total of 39 individual plants of the T₂ generation were obtained. Hi‐TOM sequencing analysis revealed that 20 plants (51.28%) exhibited the homozygous genotype of ssff, 15 plants (38.46%) displayed the genotype SsFf, and four strains (10.26%) possessed the genotype Ssff (Table S3 and Data S2). The Tk1‐SST and Tk1‐FFT gene loci exhibit tight genetic linkage, as evidenced by their high co‐segregation frequency (51.28% of ssff and 38.46% of SsFf). We successfully introgressed the double homozygous mutants 1‐sst1‐fft into different genetic backgrounds, establishing stable lines for subsequent breeding and applications. These results demonstrate that the mutation sites in the 1‐sst1‐fft mutant plants are stably inherited and can be transmitted across generations in TKS.

2.4. Analysis of CRISPR/Cas9 Off‐Target Activity in the 1‐sst1‐fft Mutants

In silico analysis of the targets of Tk1‐SST and Tk1‐FFT in the reference genome of TKS line 1151 (GWHBCHF00000000) revealed that no potential off‐target loci with up to 3 base pair mismatches were identified for SST‐T2 and FFT‐T2. However, one potential off‐target locus with three mismatches was identified for both SST‐T1 and FFT‐T1, designated as SST‐T1‐OFF1 and FFT‐T1‐OFF, respectively. SST‐T1‐OFF1 contained three mismatches and was located within the coding region of a predicted gene encoding an F‐box family protein (TkAUNG097010). FFT‐T1‐OFF1 also contained three mismatches but was situated within a non‐coding region of the TKS genome. No off‐target sites with fewer than three mismatches were found. Furthermore, we amplified both SST‐T1‐OFF1 and FFT‐T1‐OFF1 in the 1‐sst1‐fft mutant plants by PCR, and their products were verified by Hi‐TOM sequencing. No mutations were detected at SST‐T1‐OFF1 or FFT‐T1‐OFF1 in the 1‐sst1‐fft mutants, compared with the WT sequence (Table S4). To expand the analysis, we increased the maximum number of mismatches to 5 bp and performed a search against the reference genome. Among 4–5 mismatched bases, 310 potential off‐target sites were identified within predicted genes. Simultaneously, whole‐genome sequencing (WGS) analysis was conducted on the mutants (ssff‐10, ssff‐17 and ssff‐36) and WT. By integrating these potential off‐target sites with WGS‐identified genes harbouring insertions/deletions (InDels), an overlap was observed in 28 genes (Data S3). Further investigation revealed that no mutations occurred at these sites between the mutants and WT, but rather discrepancies compared to the reference genome. Collectively, these results confirmed the absence of off‐target effects in the 1‐sst1‐fft mutants.

2.5. The Biomass and Fertility of Plant Were Significantly Enhanced in the 1‐sst1‐fft Mutants

The phenotype of the 1‐sst1‐fft mutants differed significantly from that of WT, and these differences were observed across various developmental stages (Figures 2a,b and S4). Consequently, the phenotypic differences between 1‐sst1‐fft and WT plants were analysed at 150–180 days after culture (DAC), when plant phenotypes were relatively stable. Knockout of Tk1‐SST and Tk1‐FFT genes markedly increased aboveground biomass in the plants. As shown in Figure 2, the leaf length, leaf width and leaf area of the mutant plants were 11.0% to 32.1%, 45.3% to 52.9%, and 74.1% to 89.0% greater than those of the WT plants, respectively (Figure 2c). The number of rosette leaves in the 1‐sst1‐fft mutant plants was 1.77 to 1.96 times greater than that in the WT (Figure 2d). Except for the leaf length of ssff‐10, significant differences were observed in the leaf characteristics between 1‐sst1‐fft and WT plants (p < 0.05). Furthermore, root characteristics were analysed in the mutants and WT plants (Figure S4a). The number of lateral roots per plant in 1‐sst1‐fft was reduced (Figure 2e). Given that most plants lacked prominent main roots, the diameter of the primary root was measured within the first 0.5 cm of the underground portion. No significant difference in primary root diameter was observed between 1‐sst1‐fft and WT, except for a decrease in ssff‐36 (Figure 2f). Root length in the mutant plants was significantly increased (p < 0.05), except for ssff‐17 (Figure 2g). The whole‐plant fresh weight of the 1‐sst1‐fft mutant plants was significantly greater (81.6% to 119.4%) than that of WT plants (p < 0.05) (Figure 2h). Specifically, the aboveground biomass of 1‐sst1‐fft was 2.62 to 3.09 times greater than that of WT (p < 0.05), while no significant difference was observed in underground biomass (Figure 2h). The root‐to‐shoot biomass ratio (RSR) in the mutants was significantly decreased (p < 0.05) (Figure 2i). Additionally, flowering and fertility characteristics were analysed. Both the mutant and WT plants bloomed twice within 6–7 months after cultivation, with an approximate two‐month interval between the first and second flowering under greenhouse conditions (Figure 2a,b). There was no difference in the number of flowers per plant during the first flowering stage; however, the number of flowers per plant in the mutants was 2.29 to 2.88 times greater than that in WT during the second flowering stage (Figure 2j). Compared to WT plants, the fertility of the 1‐sst1‐fft mutant plants was significantly improved, with the number of normal seeds per flower and seed setting rate being 1.76–2.45 times and 1.43–1.69 times greater, respectively (Figure 2k). These findings demonstrate that simultaneous deletion of Tk1‐SST and Tk1‐FFT leads to substantial alterations in the phenotype, growth and development of TKS. Notably, this genetic modification significantly enhances the growth of the aboveground structures and improves seed set.

FIGURE 2.

Comparative analysis of the phenotypic traits and fertility between the 1‐sst1‐fft mutants and WT. The photographs were taken at the first flowering stage (~50 DAC) (a) and the second flowering stage (~150 DAC) (b). Bar = 2.0 cm. The leaf length, leaf width and leaf area (c); and the number of rosette leaves (d); the lateral root number (e); the diameter of the primary root (f); root length (g); the whole‐plant fresh weight; the fresh weight of the aboveground and underground (root) (h) of the plant and the root/shoot ratio (RSR) (i) were measured at 150–180 DAC. The number of flowers per plant at the first and second flowering stages (j), the number of normal seeds and abortive seeds per flower, and the seed setting rate (k) of the 1‐sst1‐fft mutants and WT plants were determined. Each value was calculated from the means of 4–6 individual plants. The different lowercases indicate a significant difference with p < 0.05. n.s. means no significant difference.

2.6. Inulin Biosynthesis Was Inhibited, While Available Sucrose Levels Were Markedly Enhanced in the 1‐sst1‐fft Mutants

Furthermore, we investigated whether the knockout Tk1‐SST and Tk1‐FFT would alter the inulin accumulation in TKS. The results demonstrated that there was a sharp decrease in inulin accumulation across all the 1‐sst1‐fft mutant lines. Specifically, the inulin contents in the ssff‐10, ssff‐17 and ssff‐36 plants were 1.75 mg/g DW, 2.43 mg/g DW and 1.47 mg/g DW, respectively, corresponding to only 0.97%, 1.34% and 0.81% of that in WT plants (180.96 mg/g DW) (Figure 3a). Additionally, the mean degree of polymerisation (DP) in the 1‐sst1‐fft mutants sharply declined to values between 2.1 and 3.3, compared with 11.5 in WT (Figure 3b). Moreover, inulin accumulation was analysed at different developmental stages: the end of the first flowering stage (~120 DAC), the end of the second flowering stage (~200 DAC), and during dormancy (~300 DAC). Results indicated that a dramatic reduction in inulin content in the mutant at all stages, reaching only 1.40%, 0.81% and 1.72% of that in WT plants at approximately 120 DAC, 200 DAC and 300 DAC, respectively (Figure 3c). Similarly, the mean DP also decreased significantly in ssff‐36 at these stages (Figure 3d). In addition, mixed samples from different lines were analysed for inulin content and mean DP, which were used for subsequent transcriptome, sugar and terpenoid analyses (Figure S4b). The results showed a significant reduction in both inulin content and mean DP in 1‐sst1‐fft, with the average inulin level being only 4.55% of that in WT plants (Figure S5a). These findings were consistent with those observed in individual strains and across different developmental periods. The small amount of inulin detected in the mutants is likely due to the enzymatic hydrolysis of other oligosaccharides by inulinase. Collectively, these results demonstrate that the knockout of inulin synthetase genes Tk1‐SST and Tk1‐FFT abolishes inulin synthesis in TKS roots.

FIGURE 3.

Analysis of inulin and free sugar levels in the 1‐sst1‐fft mutants and WT plants. Inulin content (a) and mean degree of polymerisation (DP) (b), as well as the contents of starch (e), free fructose (f), glucose (g) and sucrose (h), were examined in different mutant lines and WT at ~200 DAC. Inulin content (c) and DP (d), and the contents of free fructose (i), glucose (j) and sucrose (k) were compared between the mutant line ssff‐36 and WT at different developmental stages. The different lowercases indicate a significant difference with p < 0.05. *, **, *** indicate a significant difference with p < 0.05, p < 0.01, p < 0.001, respectively. n.s. means no significant difference.

Starch is also a major form of carbohydrate storage in plants; therefore, changes in starch content in the mutant plants were analysed. As shown in Figure 3e, starch accumulation was much lower than that of inulin in the roots of TKS, indicating that inulin was the primary form of carbon storage in TKS roots. Interestingly, except for ssff‐10, the starch content in 1‐sst1‐fft was significantly lower (by 14.5%–25.7%) compared to WT plants (p < 0.05). Subsequently, the contents of three free sugars—sucrose, fructose and glucose—were analysed. Sucrose was the most abundant sugar in TKS roots, followed by fructose, with glucose being the least abundant. The fructose content decreased significantly in 1‐sst1‐fft (Figure 3f). Compared to WT, glucose content in the mutants was increased, with only the ssff‐36 strain reaching a significant level (p < 0.05) (Figure 3g). The sucrose contents in all three mutant lines were markedly elevated, being 1.16‐, 2.25‐ and 1.33‐fold higher than that in WT, respectively (p < 0.05) (Figure 3h). Further analysis of the three sugars in 1‐sst1‐fft and WT plants at different developmental stages revealed that the fructose content in the mutant was significantly decreased at approximately 120 DAC and 200 DAC, but no significant difference was observed at the later stage (~300 DAC) (Figure 3i). Conversely, glucose content showed an opposite trend, with the mutant exhibiting significantly higher glucose levels than WT from approximately 200 DAC onwards (Figure 3j). Compared to WT, sucrose content in the mutant plants was significantly increased by more than 33% at all stages (Figure 3k). These findings suggest that inhibition of inulin synthesis in 1‐sst1‐fft mutant plants resulted in substantial changes in free sugar levels, particularly a continuous increase in sucrose content.

2.7. NR Accumulation Was Substantially Elevated in the 1‐sst1‐fft Mutants

Moreover, we investigated whether the inhibition of inulin synthesis in the roots of the mutant plants affected NR accumulation, which is synthesised using sucrose as a carbon source. Cross‐sectional analysis of the dry roots revealed that the roots of the 1‐sst1‐fft mutant plants clearly contained more rubber filaments compared to those of the WT (Figure 4a). Laticifer cells, specialised cells responsible for NR biosynthesis and storage in TKS, were analysed in transverse sections of 1‐sst1‐fft and WT roots. The results indicated that laticifer cells in the 1‐sst1‐fft mutants were thicker and denser than those in the WT (Figure 4b,c), suggesting enhanced capacity for NR synthesis. As shown in Figure 4d, there was a significant increase in NR accumulation across all three 1‐sst1‐fft mutant lines (p < 0.05); specifically, NR content in ssff‐10, ssff‐17 and ssff‐36 was 2.02‐, 1.65‐ and 2.51‐fold higher than that in the WT plants, respectively. The highest NR content was observed in ssff‐36, reaching 123.95 mg/g DW. On average, NR content in the mutant plants was 2.06 times greater than that in the WT plants. Additionally, the molecular weight of NR, including average molecular weight (M _w) and number average molecular weight (M _n), was analysed in the 1‐sst1‐fft mutants and WT plants. While both M _w and M _n values in ssff‐10 and ssff‐17 were higher than those in WT, no statistically significant difference was observed between the mutants and WT plants (Figure 4e). Furthermore, NR accumulation was assessed in mixed samples from different lines (Figure S4b), used for subsequent transcriptome, sugar and terpenoid analysis. The results showed that NR content in 1‐sst1‐fft plants was 2.85 times greater than that in WT (Figure S5b), consistent with findings from individual mutants. These findings suggest that the simultaneous deletion of Tk1‐SST and Tk1‐FFT in TKS suppresses inulin synthesis while increasing the availability of free sucrose as a carbon source. This elevated carbon availability, together with other factors, promotes laticifer formation and thereby supports greater NR biosynthetic capacity. More importantly, the increased supply of carbon source directly accelerates NR biosynthesis. Overall, these combined effects lead to a substantial increase in NR accumulation in TKS roots.

FIGURE 4.

The 1‐sst1‐fft mutants significantly promote laticifer differentiation and increase NR content. (a) Rubber filaments were observed on the dry root cross‐sections of 1‐sst1‐fft and WT. Bar = 2.5 mm. (b) Laticifer cells in the fresh root cross‐sections of 1‐sst1‐fft and WT. The laticifer cells (white arrowheads) were coloured in deep red by Oil Red O. Images are representative of three independent plants. Bars = 500 μm. (c) The laticifer cell index was calculated using ImageJ software. (d) NR content was measured via FTIR spectroscopy. (e) Weight average molecular weight (M _w) and number average molecular weight (M _n) were measured via HPLC‐GPC analysis. Root samples collected from 1‐sst1‐fft and WT plants at ~200 DAC were used for these analyses. Each value represents the means of biological triplicates. The different lowercases indicate a significant difference with p < 0.05.

2.8. Elimination of Inulin Biosynthesis in the 1‐sst1‐fft Mutants Resulted in Significant Alterations in the Expression of Genes Involved in Secondary Metabolic Pathway, Carbon Sources Metabolism and Plant Growth and Development Process

To elucidate the molecular mechanisms by which Tk1‐SST and Tk1‐FFT regulate inulin and NR accumulation, as well as plant growth and development in TKS, transcriptome sequencing was conducted on the roots of the 1‐sst1‐fft mutants and WT plants. Three biological replicates of both the mutant and WT plants were sequenced using the Illumina sequencing platform. Differentially expressed genes (DEGs) were identified by comparing the 1‐sst1‐fft mutants and WT. A total of 8335 DEGs were identified in the 1‐sst1‐fft (sf) vs. WT comparison; among these, 4340 DEGs were upregulated, and 3995 DEGs were downregulated (Figure 5a,b, Data S4). Moreover, Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway and Gene Ontology (GO) term enrichment analysis were performed on the DEGs (Figure 5c,d, Data S5). KEGG pathways significantly enriched in upregulated DEGs primarily included phenylpropanoid biosynthesis, starch and sucrose metabolism and propanoate metabolism. In contrast, KEGG pathways significantly enriched in downregulated DEGs predominantly encompassed terpenoid metabolism‐related pathways, phenylpropanoid biosynthesis and fructose and mannose metabolism. In the GO categories significantly enriched in upregulated DEGs, the molecular function category was mainly associated with acyl‐CoA dehydrogenase activity, sucrose transmembrane transporter activity, glucosidase activity, sugar transmembrane transporter activity and carbohydrate transmembrane transporter activity. The biological process category was primarily related to cell differentiation, endoderm development, regulation of root morphogenesis and reproductive system development. For the GO categories significantly enriched in downregulated DEGs, the molecular function category was mainly associated with oxidoreductase activity, monooxygenase activity and phosphotransferase activity. The biological process category was primarily related to secondary metabolite biosynthetic processes, triterpenoid metabolic processes, negative regulation of cell growth and negative regulation of growth. Thus, simultaneous knockout of Tk1‐SST and Tk1‐FFT resulted in substantial changes in the gene expression profiles of TKS, particularly affecting genes involved in secondary metabolic pathways, carbon, starch and sucrose metabolic pathways, as well as plant growth and development processes.

FIGURE 5.

DEGs between the 1‐sst1‐fft mutants and WT and their enriched KEGG pathways and GO terms. (a) DEGs identified via comparing the root transcriptome of 1‐sst1‐fft and WT. Red dots represent upregulated DEGs, blue dots represent downregulated DEGs, and grey dots represent no DEGs. Three biological replicates of 1‐sst1‐fft mutants (sf‐1, sf‐2 and sf‐3) and WT (WT‐1, WT‐2 and WT‐3) plants at ~200 DAC were used for analysis. (b) Heatmap of differentially expressed genes in 1‐sst1‐fft and WT. The top 20 enriched KEGG pathways (c) and GO terms (d) for the DEGs that were upregulated and downregulated in the 1‐sst1‐fft mutants, respectively.

2.9. Upregulated Expression of Genes Involved in Carbohydrate Metabolism and Sugar Signalling and Transport Increased the Efficiency of Sucrose Synthesis, Transport and Utilisation in 1‐sst1‐fft

As shown in Figure 6a, the inhibition of inulin synthesis in the 1‐sst1‐fft mutants resulted in significant changes in the expression of genes involved in sugar metabolism, signalling and transport. In the inulin metabolism pathway, the expression levels of Tk1‐SST and Tk1‐FFT were markedly suppressed in 1‐sst1‐fft, being only 0.25 and 0.3 times that of the WT, respectively. This result was consistent with the qRT–PCR analysis of individual mutant plants (Figure 1c,d). In the starch and sucrose metabolism pathway, 71 DEGs were upregulated, while 57 DEGs were downregulated (Data S6). Among these, 8 DEGs encoding the structural enzyme of sucrose metabolism, 4 DEGs involved in sugar signalling pathways, 7 DEGs encoding Sugar Will Eventually be Exported Transporters (SWEET) and 6 DEGs encoding the structural enzymes of starch metabolism were identified (Figure 6a). Notably, the transcript abundance of the sucrose–phosphate synthase (SPS) gene and invertase (INV) gene was significantly upregulated in the mutants, whereas the sucrose synthase (SUS) gene and fructokinase (FK) gene were downregulated in 1‐sst1‐fft. In the sugar sensing and signalling pathways, the transcript levels of genes encoding hexokinase (HXK1), Snf1‐related kinase 1 (SnRK1‐3) and the plant target of rapamycin (TOR) kinase were significantly increased in the 1‐sst1‐fft mutants. Regarding trehalose synthesis, two genes encoding trehalose‐6‐phosphate synthases (TS1 and TS2) and one gene encoding trehalose‐6‐phosphate phosphatases (TPP) were differentially expressed in 1‐sst1‐fft. Interestingly, seven DEGs encoding SWEET transporters were identified, all of which were upregulated in 1‐sst1‐fft (Figure 6a), suggesting that enhanced sucrose transport mediated by SWEET transporters occurred in the 1‐sst1‐fft mutants. Moreover, the expression of all DEGs encoding structural enzymes of sucrose metabolism and sugar signalling and transport was verified by quantitative real‐time PCR (qRT–PCR), indicating that, except for TPP, the variation in gene expression between the mutants and WT was consistent with the transcriptomic data (Figure 7a,b). Additionally, the activities of SPS and INV were determined in 1‐sst1‐fft and WT plants. The results demonstrated that the activities of SPS, neutral INV and acid INV in both the leaves and roots of 1‐sst1‐fft were significantly higher than those in WT plants (Figure 7c–e), suggesting that sucrose synthesis and hydrolysis are more active in 1‐sst1‐fft compared to WT.

FIGURE 6.

Identification of DEGs in the carbon metabolic pathway and NR biosynthesis pathway, and differentially accumulated metabolites. (a) The schematic of the carbon metabolic pathway and NR biosynthesis pathway and expressions of DEGs encoding structural enzymes of these pathways, sugar signalling and transport in 1‐sst1‐fft and WT. Expression levels for each gene in 1‐sst1‐fft and WT were shown by heatmap using FPKM. Each value represents the mean of three biological replicates. FPKM, fragments per kilobase of transcript per million fragments mapped. (b) Heatmaps of carbohydrate accumulation in the roots of 1‐sst1‐fft and WT. Three biological replicates of 1‐sst1‐fft mutants (sf‐1, sf‐2 and sf‐3) and WT (WT‐1, WT‐2 and WT‐3) plants at ~200 DAC were used for analysis. The normalised signal intensities of sugar contents (unit variance scaling) were visualised as a colour spectrum and presented as heatmaps using hierarchical cluster analysis. Upregulated sugars are indicated by red asterisks, and downregulated sugars are indicated by blue asterisks. Ribose: D‐ribose; GlcNAc: 2‐Ace‐2‐Deo‐D‐glucosamine; Raf: Raffinose; Fru: Fructose; Cel: Cellobiose; Suc: Sucrose; Lac: Lactose. (c) Heatmaps of differentially accumulated terpenoids in the roots of 1‐sst1‐fft mutants and WT. Differentially accumulated metabolites were determined by VIP > 1, fold change ≥ 3, and p < 0.01. Monoterpenoids, sesquiterpenoids, diterpenoids, triterpenes and terpenes are highlighted in blue, green, yellow, orange and black font, respectively.

FIGURE 7.

Validation of expression levels of DEGs in the carbon metabolic pathway and key enzyme activities. (a, b) Validation of DEGs encoding structural enzymes of sugar and inulin metabolism (a) and sugar signalling and transport (b) in the 1‐sst1‐fft mutants and WT by qRT–PCR analysis. (c–e) The activities of SPS (c), neutral INV (d) and acid INV (e) were measured in the leaves and roots of 1‐sst1‐fft and WT plants. *, **, *** indicates p < 0.05, p < 0.01, p < 0.001, respectively.

Furthermore, carbohydrate profiling in the roots of 1‐sst1‐fft and WT plants was conducted using a gas chromatography mass spectrometry (GC/MS) platform. A total of 27 sugars were identified in TKS roots (Data S9). As shown in Figure 6b, eight sugars exhibited differential accumulation between 1‐sst1‐fft and WT. Specifically, the levels of sucrose, ribose, 2‐Ace‐2‐Deo‐D‐glucosamine, inositol and raffinose were significantly increased in the mutants, while the D‐cellobiose, lactose and fructose levels were markedly decreased. Notably, raffinose was undetectable in WT plants but accumulated significantly in the mutants. Conversely, lactose was present in WT plants but absent in the mutants. In the 1‐sst1‐fft mutants, the contents of D‐cellobiose and fructose were reduced to 0.05 and 0.11 times those in WT, respectively. The overall sugar content in TKS roots was dominated by sucrose, followed by fructose and inositol. In the 1‐sst1‐fft mutants, sucrose content was 987 times and 258 times higher than that of fructose and inositol, respectively, and 9.45 times greater than in WT (Data S9). These findings suggest that the deletion of Tk1‐SST and Tk1‐FFT upregulated the expression of genes and enzymatic activities involved in sucrose metabolism, leading to elevated sucrose levels in 1‐sst1‐fft. Additionally, the expression of genes encoding central sensors and regulators of sugar signalling, such as HXK1, TOR and SnRK1‐3, as well as sugar transporters (SWEET), was also increased. These changes enhanced the efficiency of sugar transport and utilisation in the 1‐sst1‐fft mutants.

2.10. NR Biosynthesis in the 1‐sst1‐fft Mutants Was Enhanced by the Upregulation of Key Genes Involved in NR Elongation

NR is composed of isopentenyl monomers derived from IPP, which is produced by the MVA and MEP pathways, using sucrose as a carbon source. By comparing the root transcriptomes of the 1‐sst1‐fft mutants and WT plants, 20 DEGs involved in NR biosynthesis were identified (Figure 6a and Data S7). In the MVA pathway, except for one mevalonate kinase (MVK2) gene, the other 4 DEGs, including three HMG‐CoA reductase (HMGR2, HMGR4 and HMGR5) genes and MVK5, were downregulated in the mutants. In the MEP pathway, the expression of genes encoding 4‐diphosphocytidyl‐2‐C‐methyl‐D‐erythritol synthase 2 (CMS2) and 1‐hydroxy‐2‐methyl‐2‐(E)‐butenyl 4‐diphosphate reductase (HDR5) was upregulated, while HDR2 expression was downregulated in the mutants. Four DEGs involved in initiator synthesis were identified. Among these, genes encoding geranylgeranyl diphosphate synthase (GGPS4 and GGPS5) were significantly upregulated in 1‐sst1‐fft, with expression levels 3.95‐ and 2.55‐fold higher than those in WT, respectively. Conversely, genes encoding isopentenyl diphosphate isomerase (IPI3) and geranyl diphosphate synthase (GPS5) were downregulated in the mutants. In the rubber elongation process, cis‐prenyltransferase (CPT) is the key enzyme for NR elongation, and CPT‐like protein (CPTL) is an essential protein that interacts with and activates CPT. A total of eight DEGs involved in rubber elongation were identified (Figure 6a). Notably, CPT4, which exhibited the highest transcript abundance, was markedly upregulated in 1‐sst1‐fft, with an expression level 8.13 times greater than that in WT. Similarly, CPTL1, which also showed high expression, was clearly upregulated in the mutants, with an expression level 2.06 times greater than that in WT. Small rubber particle protein (SRPP) and rubber elongation factor (REF), which function as structural proteins stabilising rubber particles, are crucial for NR biosynthesis. Five SRPPs were differentially expressed between 1‐sst1‐fft and WT plants. Among these, SRPP4, with the highest transcript abundance, was significantly upregulated in the mutants. Additionally, SRPP10 and SRPP11, although expressed at relatively low levels, were also upregulated in the mutants. Interestingly, 26 DEGs involved in sesquiterpenoid and triterpenoid biosynthesis, which use IPP as a substrate, were identified. Among these, 8 DEGs were upregulated, while 18 DEGs were downregulated (Data S8). Specifically, six DEGs encoding terpenoid biosynthesis enzymes (TPS1‐TPS6) were identified, with two TPSs upregulated and four TPSs downregulated in the 1‐sst1‐fft mutants (Figure 6a). Furthermore, the expression levels of DEGs encoding structural enzymes involved in MVA, MEP, NR and terpenoid biosynthesis were verified by qRT–PCR. The results indicated that, except for MVK5, the variation in all other genes between the mutants and WT plants was consistent with the transcriptomic data (Figure 8a–c). We further quantified the enzyme activities of the key enzymes in the MVA and MEP pathways. Results indicated that, compared with WT, there were no significant differences in the enzyme activities of HMGR and HDR between the 1‐sst1‐fft mutants and WT plants (Figure 8d,e). These findings demonstrate that the knockout of Tk1‐SST and Tk1‐FFT upregulated the expression of genes involved in NR elongation, thereby promoting enhancement of NR biosynthesis in 1‐sst1‐fft.

FIGURE 8.

Validation of expression levels of DEGs in the NR biosynthesis pathway and key enzyme activities. (a, b) Validation of DEGs encoding structural enzymes of the MVA and MEP pathways (a), terpenoid biosynthesis (b), initiator synthesis and NR elongation (c) in the 1‐sst1‐fft mutants and WT by qRT–PCR analysis. (d, e) The activities of HMGR (d) and HDR (e) were measured in the leaves and roots of 1‐sst1‐fft and WT plants. *, **, *** indicates p < 0.05, p < 0.01 and p < 0.001, respectively. n.s. means no significant difference.

Since NR and terpenoids are compounds derived from isopentenyl monomers originating from IPP, the accumulation of terpenoids was evaluated in the roots of the 1‐sst1‐fft mutants and WT plants. A total of 460 terpenoids were identified in TKS roots (Data S10). Among these, 16 terpenoids (including 9 sesquiterpenoids, 3 monoterpenoids, 2 triterpenes, 1 terpene and 1 diterpenoid) were upregulated, while one sesquiterpenoid was downregulated in 1‐sst1‐fft compared to WT (Figure 6c). Notably, the accumulation of five terpenoids in the 1‐sst1‐fft mutants was more than 5 times greater than that in the WT. These terpenoids included Wasn002690 (villosolside), MWSmce630 (atractyloside A), Wcfp003018 (5‐(hydroxymethyl)‐2,7‐dimethyl‐6‐(prop‐1‐en‐1‐yl)‐1H‐inden‐1‐ol), Wbsp006047 (serratane‐3α,14α,20β,21β,24,29‐heptol) and Zjyp102937 (oxyphyllone C) (Data S11). Furthermore, to investigate the relationship between carbon sources and metabolite accumulation, we conducted a correlation analysis involving NR and inulin, as well as differentially accumulated sugars and terpenoids in the 1‐sst1‐fft mutants and WT plants. Results demonstrated that inulin accumulation was negatively correlated with NR levels (p < 0.001) and the 16 upregulated terpenoids (p < 0.01 or p < 0.001), while sucrose levels exhibited a positive correlation with NR levels (p < 0.01) and the 16 upregulated terpenoids (p < 0.05 or p < 0.01). Additionally, the contents of inositol and raffinose were positively correlated with NR levels and the 16 upregulated terpenoids (Figure S5c). These results indicate that the deletion of Tk1‐SST and Tk1‐FFT in TKS not only inhibited inulin biosynthesis but also led to significant changes in the accumulation of other carbohydrates. Notably, a sharp increase in sucrose content was observed, enhancing the supply of carbon sources for NR and terpenoid biosynthesis in the 1‐sst1‐fft mutants.

3. Discussion

1‐SST and 1‐FFT are two key genes essential for inulin biosynthesis in plants; however, their roles in growth and development remain poorly understood due to the lack of mutant lines. While genome editing technology has been extensively applied in various plant species, its application in TKS is limited. This limitation is partly attributed to the self‐incompatibility of TKS, which complicates the generation of homozygous mutants through backcrossing. A single‐gene mutant Tk1‐fft was generated. A decrease in inulin content and an increase in NR content were observed only in the K599‐FFT plants, while there was no significant change noted in EHA105‐FFT plants (Ariyaratne et al. 2023). The heterozygosity of 1‐FFT in the mutant may be the primary reason for the insignificant change in inulin and NR content observed in EHA105‐FFT. The efficiency of generating homozygous mutants in T₀ plants varies across species and is influenced by factors such as transformation methods, regeneration systems and CRISPR delivery timing. In rice, CRISPR/Cas9 systems driven by strong promoters (e.g., OsUbi) achieve 10%–50% of homozygous/biallelic mutations in T₀ plants (Ma et al. 2015). A multiplex CRISPR system targeting three homoeologs of TaMLO in hexaploid wheat achieved 5.3% homozygotes in T₀ plants (Wang et al. 2014). For the first time, we reported the generation of double‐gene knockout homozygous mutants (1‐sst1‐fft) which abolish inulin biosynthesis. This allowed us to reveal the important functions of inulin and the genes Tk1‐SST and Tk1‐FFT in plant growth and development. The frequency of the homozygous mutation was 4.20% (Table S1), which was lower than that observed in wheat and rice, likely due to the use of root segments for transformation in this study. The identification of mutation sites in the T₁ and T₂ generation plants demonstrated that the mutant loci 1‐sst and 1‐fft were stably inherited by the offspring. All T₁ generation lines exhibited the heterozygous SsFf genotype, consistent with Mendel's laws of inheritance. However, a biased segregation pattern was observed in the T₂ generation lines obtained through backcrossing T₁ with T₀ plants (Table S3 and Data S2), which may be attributed to the complex nature of the mutation sites in the T₀ plants and the limited population size. Additionally, Tk1‐SST (GWHBCHF00000003:150859377..150863629) and Tk1‐FFT (GWHBCHF00000003:150893366..150897334) showed a high frequency of co‐segregation (51.28% of ssff and 38.46% of SsFf), likely due to their relatively close physical distance (29.74 kb). Nevertheless, through hybridization and backcrossing, we successfully introduced the double‐gene mutation sites (1‐sst1‐fft) into different genetic backgrounds, ensuring each gene carried only a single type of mutation (i.e., ssff‐10×T1‐4, ssff‐17×T1‐1, Data S2). Recently, Tkcpt1/2 double mutant was generated using CRISPR/Cas9, but the mutation frequency was not mentioned (Xu et al. 2025). Unlike this study, the TkCPT1 and TkCPT2 sequences are highly homologous (96.10% sequence identity), allowing simultaneous knockout with the same target and reducing the difficulty of creating and screening double mutants. This study marks the first successful report in TKS of obtaining homozygous mutants by concurrently knocking out two distinct gene loci via CRISPR/Cas9 system and subsequently introgressing these mutated loci into different genetic backgrounds, which expedites the breeding and application of the 1‐sst1‐fft mutants in TKS.

Given that inulin serves as a reserve carbohydrate, it plays a crucial role in maintaining essential root functions during winter dormancy and protecting plants against abiotic stress by stabilising cell membranes (Kreuzberger et al. 2016; van Arkel et al. 2012; Van den Ende 2013; Wei et al. 2017). During cultivation, both WT and 1‐sst1‐fft mutants entered a period of slow growth, referred to as dormancy, following flowering. Despite the absence of inulin accumulation in the 1‐sst1‐fft mutants, their aboveground biomass remained greater than that of WT plants even during the slow growth phase after the second flowering (Figure S4b). Notably, the increase in leaf area and rosette leaf number (Figure 2c,d) enhanced photosynthetic productivity in the mutants, ensuring sufficient carbon sources for post‐flowering growth in the 1‐sst1‐fft mutants. Sugar signalling integrates metabolic and developmental processes in plants, coordinating growth in response to energy availability. Providing elevated levels of sucrose to harvested sinks represents a promising strategy for enhancing crop yields (Flavell 2023; Nuccio et al. 2015). For instance, overexpression of SoSPS1 in sugarcane increased sucrose accumulation, plant height and stalk number, leading to increased sugarcane yield (Anur et al. 2020). Similarly, overexpression of maize SPS in tomato plants under high CO₂ conditions enhanced leaf photosynthesis and sucrose synthesis, resulting in an increased number of inflorescences and fruits, as well as a significant rise in total fruit DW (Micallef et al. 1995). In Arabidopsis, HXK1 interacts with photosynthesis‐related genes (RBCS) to enhance carbon assimilation and cell expansion in roots, leaves and inflorescence (Jang et al. 1997; Moore et al. 2003). Glucose‐TOR signalling regulates key gene sets involved in central metabolism, cell cycle, transcription, signalling, transport and protein folding (Xiong et al. 2013). In this study, HXK1 and TOR expression (Figure 6a), as well as glucose and sucrose content (Figure 3g,h), were significantly upregulated in the 1‐sst1‐fft mutants, thereby promoting aboveground plant growth and development. This was evidenced by increases in leaf area, rosette leaf number, flower number, root length, aboveground biomass and whole‐plant biomass (Figure 2). Based on these findings, we speculate that the glucose‐TOR signalling pathway plays a pivotal regulatory role in modulating the phenotypic variations observed in the 1‐sst1‐fft mutants. Overexpression of A. cepa Ac1‐SST in cotton enhanced drought tolerance and agronomic traits, with increased proline and soluble carbohydrates (especially 1‐kestose), but reduced sucrose content under drought stress (Liu et al. 2021). In contrast, both sucrose content and plant growth were elevated in the 1‐sst1‐fft mutants with Tk1‐SST deletion in this study. These findings collectively highlight the critical role of 1‐SST in regulating sugar metabolism. The opposing effects of 1‐SST on plant growth in TKS and cotton may stem from differences in the integrated regulatory mechanisms involving sugar signalling as well as other factors, such as hormone signalling (to be discussed later). Additionally, significant variations were observed among different mutant lines in terms of lateral root number, root length and NR content (Figures 2e,g and 4d). These differences may primarily be attributed to variations in glucose, fructose and sucrose levels across the mutant lines (Figure 3f–h). Alternatively, the different types of mutations present in the mutant lines, as well as epigenetic alterations arising during the tissue culture process, could also contribute to these phenotypic discrepancies. Previous studies have demonstrated that epigenetic modifications can be induced during plant regeneration through tissue culture (Kaeppler et al. 2000). Despite these variations, the overall trends observed in the 1‐sst1‐fft mutants remain significant when compared to WT plants. Moreover, transcriptome comparative analysis revealed that DEGs involved in cell differentiation, endoderm development, regulation of root morphogenesis, plant organ morphogenesis and reproductive system development were upregulated in the mutants. Conversely, DEGs associated with negative regulation of cell growth and overall growth were downregulated (Figure 5d). For instance, TkA04G055070, which encodes squamosa promoter‐binding‐like protein 9 (SPL9), a core transcription factor in the miR156 regulatory network (Bao et al. 2019; Schwarz et al. 2008), was significantly upregulated in the mutants (Data S4). In addition, hormone signalling pathway genes exhibited differential expression patterns in the mutants. Specifically, 35 DEGs related to auxin response were upregulated, 6 DEGs associated with auxin homeostasis were downregulated, 11 DEGs involved in brassinosteroid‐mediated signalling were upregulated, 19 DEGs linked to brassinosteroid response were upregulated and all 37 DEGs associated with ethylene response were downregulated (Data S5). These findings suggest that Tk1‐SST and Tk1‐FFT function synergistically to regulate not only inulin biosynthesis but also plant growth and development by modulating carbon source availability, sugar signalling and hormone signalling, thereby regulating downstream genes associated with growth and development.

Sucrose serves as a central carbon source and signalling molecule governing plant growth, development, defence and metabolic processes (Skala et al. 2022; Yoon et al. 2021). In plants, sugar transporters, sucrose metabolic enzymes and regulatory systems like trehalose 6‐phosphate (T6P) signalling—interacting with SnRK1 and TOR—are key determinants of tissue sucrose levels (Miret et al. 2024; Morales‐Herrera et al. 2023). In the sucrose metabolism pathway, SPS is a key regulator of sucrose synthesis, while INV plays a crucial role in determining sink strength by hydrolysing sucrose into glucose and fructose (Ruan et al. 2010). In this study, the inhibition of inulin synthesis in the 1‐sst1‐fft mutants resulted in significant alterations in sugar profiles (Figure 6b). Notably, there was a continuous and substantial increase in sucrose content in the 1‐sst1‐fft mutants (Figures 3h,k and 6b). In the 1‐sst1‐fft mutants, most sucrose metabolism‐related genes (except FK) were upregulated, accompanied by increased SPS and INV expression and enzyme activity in both leaves and roots (Figures 6a and 7a,c–e), suggesting enhanced sucrose synthesis and hydrolysis to optimise carbon flux. TS (encoding trehalose‐6‐phosphate synthase) and TPP, which mediate T6P synthesis (Lyra et al. 2021), were upregulated in the mutants (Figure 7a). Furthermore, key regulators involved in sugar signalling and energy status, such as HXK1, SnRK1 and TOR, as well as SWEET family transporters, exhibited significantly upregulated transcript levels in 1‐sst1‐fft (Figure 6a), indicating that the carbon source metabolism and transport in the mutants mediated by these key regulators is more active. SWEET proteins play a critical role in modulating the allocation of sucrose, glucose and fructose in plants (Chen et al. 2012). In Arabidopsis, phosphorylation of SWEET11 and SWEET12 by SnRK2 enhances sucrose transport activity, leading to increased sucrose content in roots and improved root growth under drought stress, thereby resulting in an enhanced root‐to‐shoot biomass ratio and improved drought resistance (Chen et al. 2022). A previous study indicated that the root DW did not differ significantly between the single‐gene mutant Tk1‐fft and WT in TKS (Ariyaratne et al. 2023). Despite elevated root sucrose and total biomass in the 1‐sst1‐fft mutants, the root‐to‐shoot ratio declined (Figure 2i), implying insufficient sucrose conversion into root biomass. In Arabidopsis, lateral root growth regulation involves T6P downstream of auxin signalling in pericycle lateral root founder cells and regulation of SnRK1 and TOR (Morales‐Herrera et al. 2023). This highlights the need for additional factors to regulate root biomass formation for NR yield improvement in TKS. Sucrose dynamics are stage‐dependent, balancing growth and secondary metabolism. During vegetative growth, SPS and INV regulate sucrose partitioning to roots and leaves (Winter and Huber 2000). At reproductive stages, SWEET11, 12 and 15 mediated sucrose efflux from seed coat to embryo (Chen et al. 2015), while senescence involves autophagy‐driven nutrient recycling (Avila‐Ospina et al. 2014) and T6P‐mediated sink coordination (Miret et al. 2024). Field studies have shown that NR and inulin concentration in the roots of TKS exhibit seasonal dependency: NR peaks at 12–14 months (at the end of flowering during the second season), while inulin peaks later at 16 months (Kreuzberger et al. 2016). In this study, root sucrose content peaked at approximately 300 DAC (slow growth phase) and was lowest at approximately 200 DAC (post‐second flowering) (Figure 3k), suggesting sink competition between reproductive tissues and root metabolites. Moreover, sucrose levels correlated positively with NR and terpenoid accumulation (Figure S5c), highlighting its dual role in development and metabolite synthesis in TKS. Collectively, the 1‐sst1‐fft mutants elevate sucrose metabolism and transport, thereby modulating plant growth and development as well as metabolite accumulation by promoting carbon source provision and regulating sugar signalling pathways.

Enhanced sucrose allocation to storage organs (e.g., grain, seed or tuber) is critical for achieving high yields in major crops. To improve photosynthesis and increase sucrose production, it is necessary to integrate enhanced photosynthesis into the source‐sink dynamics and sucrose homeostasis system, thereby boosting crop yields (Miret et al. 2024). In the NR‐producing plant H. brasiliensis , sucrose is considered to be the exclusive precursor of IPP and is actively translocated into the laticifer cells (in the stem) via sucrose transporters (SUT), which is crucial for latex yield formation (Dusotoit‐Coucaud et al. 2009, 2010; Tang et al. 2010). Guayule ( P. argentatum Grey) synthesises NR in stem epithelial cells and fructan in roots, with sucrose as the main carbon source (Kajiura et al. 2018). NR and inulin are two of the most important metabolites in the roots of TKS, and there is likely competition for carbon sources between their biosynthetic pathways (Kreuzberger et al. 2016; Stolze et al. 2017). Although sucrose is the main carbon source for NR synthesis in three NR‐producing plants, NR synthesis and storage occur in distinct tissues and cells. Therefore, enhancing the utilisation of carbon sources to promote NR biosynthesis represents a critical strategy for improving yields in NR‐producing plants. 1‐SST and 1‐FFT are responsible for inulin biosynthesis, while 1‐FEH is involved in inulin degradation. Downregulation of 1‐SST reduced fructan levels but did not enhance NR production, likely due to limited carbon flux plasticity in guayule (Niu et al. 2017). Although editing the Tk1‐fft mutant and overexpressing Tk1‐FEH in TKS increased NR content, substantial inulin synthesis still occurs, particularly in the Tk1‐fft mutant (Ariyaratne et al. 2023; Stolze et al. 2017). These studies suggest that the impact of 1‐sst or 1‐fft single‐gene mutants on NR synthesis is relatively limited. This study achieved a significant breakthrough for the first time by simultaneously knocking out Tk1‐SST and Tk1‐FFT, thereby abolishing inulin synthesis and developing a new variety with NR accumulation more than doubled, while maintaining the molecular weight of NR (M _n and M _w ) (Figure 4d,e). This provides a novel perspective for the domestication of TKS. IPP, the precursor of NR, is synthesised via the MVA and/or MEP pathways using sucrose as the carbon source. IPP is also used in the biosynthetic pathways for sterols, ubiquinone, triterpenoids, dolichols and polyprenols (Pulido et al. 2012). The MEP and MVA pathways are known to be subject to a “multi‐layered” form of regulation (including transcriptional regulation, post‐translational modification, allosteric regulation and feedback mechanisms), which allows plants to modulate enzymatic activity in response to changes in precursor supply, product demand, physiological conditions and environmental stimuli (Bergman et al. 2024; Nieto et al. 2009; Schulte et al. 2000). In the 1‐sst1‐fft mutants, most DEGs involved in the MVA pathway were downregulated, while two genes involved in the MEP pathway were upregulated but with very low abundance (Figures 6a and 8a). Although transcript levels of key enzymes in the MVA and MEP pathways were altered in the 1‐sst1‐fft mutants, the enzymatic activities of HMGR and HDR remained comparable to those in WT plants (Figure 8d,e). Additionally, the accumulation of most differential terpenoids increased in the 1‐sst1‐fft mutants, while the expression of most TPS genes involved in terpenoid biosynthesis was downregulated (Figure 6). These findings suggest that the transcripts of genes involved in the MVA and MEP pathways and terpenoid biosynthesis are regulated by feedback mechanisms mediated by downstream products, including farnesyl diphosphate (FPP), geranyl diphosphate (GPP), geranylgeranyl diphosphate (GGPP) and terpenoids. This regulation dynamically balances precursor supply with product demand in the 1‐sst1‐fft mutants.

Studies in several NR‐producing plants demonstrated that NR initiator synthesis and NR elongation, rather than IPP synthesis, are crucial determinants of NR quality and yield (Epping et al. 2015; Hillebrand et al. 2012; Kwon et al. 2023). Comparative genomic analysis revealed higher IPP/dimethylallyl diphosphate (DMAPP) levels in the latex of non‐NR‐producing T. mongolicum (TM5) than in NR‐producing TKS (TK1151). However, TK1151 showed expanded CPT/SRPP gene families and enhanced NR initiator synthesis activity (Lin et al. 2022). Both CPT and CPTL (serving as a rubber transferase activator) are essential for NR biosynthesis. CRISPR/Cas9‐mediated deletion or RNAi‐mediated knockdown of CPT/CPTL abolished NR biosynthesis (Epping et al. 2015; Kwon et al. 2023; Niephaus et al. 2019; Post et al. 2012). The latest study revealed that knockout of TkCPT1 resulted in significant suppression of NR biosynthesis in TKS, while simultaneous knockout of both TkCPT1 and TkCPT2 genes or single knockout of TkCPTL1 completely abolished NR biosynthesis and rubber particle ontogeny (Xu et al. 2025). In T. brevicorniculatum, silencing CPT also significantly reduced NR biosynthesis and HMGR activity while increasing levels of triterpenes and inulin (Post et al. 2012). Knockdown of a rubber transferase activator (TbRTA or CPTL) in dandelion eliminated rubber biosynthesis without affecting dolichol accumulation, indicating its independent function from dolichol synthesis (Epping et al. 2015). In this study, key genes involved in NR elongation (CPT, CPTL and SRPP), particularly CPT4 and CPTL1 with high expression levels, were significantly upregulated in the 1‐sst1‐fft mutants (Figures 6a and 8c), thereby ensuring the efficient synthesis of NR. Additionally, laticifer cells are specialised for NR biosynthesis and storage in rubber trees and TKS. Laticifer differentiation can be induced by exogenous jasmonic acid (JA) and linolenic acid, a precursor of JA biosynthesis (Hao and Wu 2000). A recent study demonstrates that heterologous overexpression of HbPSK5, which encodes the small‐peptide hormone phytosulfokine, along with its transcriptional activator HbMYC26 significantly increased rubber content by promoting laticifer differentiation in TKS (Chao et al. 2023). In this study, laticifer differentiation was enhanced in the 1‐sst1‐fft mutants (Figure 4b,c), where eight DEGs involved in JA biosynthesis were upregulated (Data S5). Additionally, two DEGs (TkA01G531370 and novel.22764) encoding PSK were differentially expressed in the mutants (Data S4), suggesting a potential connection between sugar signalling and both PSK and JA signalling pathways. Notably, sugar levels, particularly sucrose, exhibited a significant positive correlation with the accumulation of NR and terpenoids (Figure S5c), further emphasising the critical role of carbon sources in the biosynthesis and accumulation of these metabolites. Collectively, in the 1‐sst1‐fft mutants, the suppression of inulin synthesis results in a substantial increase in root sucrose supply. This increased carbon availability, in conjunction with other signalling pathways (JA and/or PSK), promotes laticifer differentiation and thereby enhances the capacity for NR biosynthesis. Concurrently, the upregulation of genes involved in sucrose metabolism and transport facilitates the delivery of sucrose to the laticifer cells, thereby promoting precursor synthesis. Furthermore, core genes associated with NR elongation play a crucial role in elevating NR biosynthesis and accumulation in the mutants. Our study proposes an efficient breeding strategy for developing NR‐rich TKS cultivars by redirecting sucrose allocation towards NR synthetic tissues, offering a novel approach for other NR‐producing plants to overcome the NR yield bottleneck. These novel varieties eliminate inulin production, increase root NR content, simplify NR extraction by reducing metabolic interference and accelerate the domestication of TKS as an NR‐producing crop.

4. Conclusions

In this study, we reported for the first time the generation of double‐gene homozygous mutants 1‐sst1‐fft in TKS by simultaneously knocking out Tk1‐SST and Tk1‐FFT. The 1‐sst1‐fft mutants eliminated inulin biosynthesis while doubling NR accumulation without altering the molecular weight of NR. We propose that the deletion of Tk1‐SST and Tk1‐FFT enhances sucrose synthesis and the supply of photoassimilates to sink tissues through the upregulation of genes involved in sucrose metabolism and sucrose transport, leading to increased sucrose levels. These elevated sucrose levels subsequently modulate plant growth and development directly or indirectly via the sugar signalling pathway. Moreover, the increased sucrose levels contribute to higher carbon source availability for NR biosynthesis, coupled with enhanced expression of key genes associated with NR elongation, thereby improving NR biosynthesis and accumulation in the 1‐sst1‐fft mutants. Consequently, our findings provide a novel strategy to enhance NR accumulation in TKS by altering carbon source allocation, which may be applicable to other NR‐producing plants. Moreover, the transcriptomic comparison between 1‐sst1‐fft and WT plants represents a significant advancement in understanding sugar metabolism related to NR biosynthesis, elucidating new functions of Tk1‐SST and Tk1‐FFT.

5. Experimental Procedures

5.1. Plant Materials and Cultivation

TKS plants were cultivated in greenhouses under a photoperiod of 16 h light/8 h dark at 23°C. Tissue culture‐generated plants of three wild‐type lines were utilised: line 1152 served as an acceptor material for transformation, while lines 1153 and 1154 were crossed with T₀ plants to generate T₁ progenies. Homozygous mutants 1‐sst1‐fft and WT plants were propagated via tissue culture as described in a previous study (Zhang et al. 2015). Root samples were collected at three developmental stages to assess sugar dynamics: ~120 days after cultivation (DAC; the recovery phase following the first flowering), ~200 DAC (post‐second flowering completion), and ~300 DAC (the slow growth phase after two flowering events). Roots harvested at ~200 DAC were further utilised for the determination of NR, inulin, sugar and terpenoid levels, as well as for RNA‐Seq analysis.

5.2. Vector Construction and Genetic Transformation

The CDSs of Tk1‐SST and Tk1‐FFT were cloned using the cDNA from TKS line 1152 as the template. The verified sequences were subsequently aligned with the TKS reference genome (GWHBCHF00000000). The exon–intron structure was determined using the GSDS 2.0 online tool (http://gsds.cbi.pku.edu.cn/index.php). sgRNAs for Tk1‐SST and Tk1‐FFT were designed using the CRISPR‐GE online tool (http://skl.scau.edu.cn/) (Xie et al. 2017), based on their target sequences. DNA oligonucleotides for the four sgRNAs were synthesised by IGE Biotechnology (China), annealed to form dimers and subsequently integrated into AtU3b, AtU3d, AtU6‐1 and AtU6‐29 sgRNA expression cassettes via overlap PCR (primers listed in Table S7). The resulting expression cassettes were cloned into the pYLCRISPRCas9Pubi‐B plasmid via the Golden Gate cloning method (Ma et al. 2015). In the final binary vector, Cas9 expression was driven by the ubiquitin promoter; SST‐T1 by the AtU6‐1 promoter and SST‐T2 by the AtU6‐29 promoter targeting the third exon of Tk1‐SST and FFT‐T1 by the AtU3b promoter and FFT‐T2 by the AtU3d promoter targeting the third exon of Tk1‐FFT. The final binary vector was introduced into Agrobacterium tumefaciens strain AGL1 for transformation. Roots of tissue culture‐generated seedlings of TKS line 1152 were isolated, and A. tumefaciens ‐mediated transformation was performed as described in a previous study (Iaffaldano et al. 2016).

5.3. Detection of the Targeted Genome Editing Events

Genomic DNA was extracted from the leaves of the transgenic TKS plants using the DNA quick Plant System (TIANGEN, China). The target region was amplified from genomic DNA using site‐specific primers in the first round of PCR. In the next round of PCR, both forward and reverse barcodes were added to the ends of the PCR products via barcode primers. The resulting PCR products were subsequently subjected to next‐generation sequencing (NGS) at the State Key Laboratory of Rice Biology and Breeding using the Hi‐TOM platform (China National Rice Research Institute, Chinese Academy of Agricultural Sciences, Hangzhou). Mutations were identified through Hi‐TOM analysis (Liu et al. 2019; Sun et al. 2024).

5.4. CRISPR‐Mutagenesis Off‐Target Site Identification and Analysis

Potential off‐target effects of each sgRNA were evaluated against the TKS genome (GWHBCHF00000000) using Cas‐OFFinder (http://www.rgenome.net/cas‐offinder/). Previous studies have demonstrated that the potential of off‐target mutations can be eliminated completely when the number of mismatched bases is more than 3 (Brant et al. 2024; Kim et al. 2017; Sturme et al. 2022). Therefore, sites containing three or fewer mismatches to the sgRNA sequence, along with an NGG PAM, were selected for off‐target analysis. Primers were designed to amplify 200–300 bp regions flanking each potential off‐target site in the mutants and WT plants (Table S7), and the resulting PCR products were subjected to Hi‐TOM analysis. Furthermore, the maximum number of mismatches was increased to 5 bp, and a comprehensive search was conducted against the reference genome. Potential off‐target sites identified within predicted genes were selected for further evaluation. Simultaneously, WGS analysis was performed on the genomic DNA of the mutants (ssff‐10, ssff‐17 and ssff‐36) and WT by Biomarker Technologies Co. Ltd. (Beijing, China). After filtering out low‐quality reads and adapter sequences, clean reads were mapped to the TKS reference genome (GWHBCHF00000000), and single nucleotide polymorphisms (SNPs) and InDels were identified using SAMtools (Danecek et al. 2021) and Genome Analysis Toolkit (GATK) (McKenna et al. 2010). The raw sequencing data have been deposited in the Genome Sequence Archive in the National Genomics Data Center (GSA accession: CRA026283) (https://ngdc.cncb.ac.cn/gsa). Due to the high heterozygosity of the TKS genome resulting from its self‐incompatibility, substantial genetic variation exists among different varieties. Therefore, only the InDels located within predicted coding genes were selected for further analysis. By integrating potential off‐target sites with WGS‐identified genes harbouring InDels, overlapping genes and mutation sites were selected for off‐target activity analysis.

5.5. Western Blot Analysis

Total proteins were extracted from the roots of the 1‐sst1‐fft mutants and WT plants at ~200 DAC using a plant root protein extraction kit (Bestbio, China). Equal amounts of proteins (50 μg) were separated by SDS–PAGE and transferred to PVDF membranes. For immunoblot analysis, the membranes were incubated with the anti‐Tk1‐SST polyclonal antibody (produced in rabbits by Mabstar Antibody Technology Company, China) at a 1:2000 dilution and the anti‐Tk1‐FFT polyclonal antibody (produced in rabbits by Zoonbio Biotechnology Company, China) at a 1:500 dilution, respectively. The secondary antibodies (anti‐rabbit horseradish peroxidase‐conjugated antibodies, Cat. No. C31460100; Invitrogen) were diluted 1:5000. Detection was performed using an Ultra‐sensitive Enhanced Chemiluminescence Substrate Kit (Cat. No. AR1191; Boster Biological Technology, China) according to the manufacturer's instructions. Chemiluminescence was imaged via ImageQuant LAS4000 system (GE).

5.6. Phenotypic Characterisation

The morphological characteristics, including the number of rosette leaves, leaf length, leaf width, leaf area, the lateral root number, the diameter of the primary root, root length, the fresh weight of the whole plant, the aboveground and underground (root) biomass and the root‐to‐shoot ratio (RSR), were investigated during a stable period of plants at 150–180 DAC. The number of flowers per plant at the first and second flowering stages was determined. At the mature‐seed stage, the number of normal seeds and abortive seeds per flower was counted. For each line, measurements were taken from 4 to 6 individual plants.

5.7. Measurement of NR Contents and Laticifer Cell Observation in the Roots of TKS

Root samples from ~200 DAC plants were collected for the determination of NR content and root staining. These analyses were performed according to previously reported methods with minor modifications (Lin et al. 2022). Briefly, 100 mg of dry root powder was sequentially washed with H₂O and acetone for 3–4 times to remove water‐soluble impurities, proteins and other impurities. The samples were then centrifuged; the supernatant was discarded, and the precipitate (NR extracted) was dried at 80°C and then thoroughly dissolved in 5 mL of petroleum ether. To this solution, 100 μL of a 5 mg/mL polystyrene internal standard (dissolved in petroleum ether) was added. The resulting mixture was analysed using NICOLET iS50 FTIR spectroscopy (Thermo Scientific, USA). The peak areas at 830 cm⁻¹ for rubber (R) and 699 cm⁻¹ for polystyrene (S) were used to calculate the R/S ratio for NR contents (Banigan et al. 1982; Liu et al. 2017). For molecular weight analysis, the extracted NR samples were dissolved in THF and subjected to calculate weight average molecular weight (M _w) and number average molecular weight (M _n) by high‐performance liquid chromatography and gel‐permeation chromatography (HPLC‐GPC) analysis using a Waters 1515 system (Waters). THF was used as a mobile phase at a flow rate of 1 mL/min. The HPLC‐GPC was connected to a Waters 2414 Refractive Index Detector (Waters) with a WAT054460 column. A standard curve was developed using polystyrene (PS) standards (50 000–2 300 000 Da; PSS polymer, Waters). Fresh root samples were fixed in 80% (v/v) ethanol for 24 h, sectioned into 100 μm slices using a vibratome (Leica, VT1200S, Germany), stained with Oil Red O (Sigma–Aldrich, USA) and mounted in 60% (v/v) glycerol. Bright field observations were performed using a microscope (LeiCa, M205FA). The laticifer cell index was calculated as the total area of the Oil Red O‐stained laticifer cells divided by the total area of the whole root section using ImageJ software.

5.8. Measurement of Inulin and Starch Contents in the Roots of TKS

The inulin content and DP were determined via HPLC as described previously (Stolze et al. 2017). Inulin was extracted from the dry root powder, and the resulting inulin‐containing solution was divided into two groups: one undigested and the other digested with inulinase (from Aspergillus niger, Sigma–Aldrich). To determine inulinase activity, 10 mg/mL inulin (from chicory) was also digested under the same conditions. The contents of fructose, glucose and sucrose in both undigested and digested samples were analysed by HPLC. Chromatographic analysis was performed using an Asahipak NH2P‐50 4E column (Shodex, Germany) as the stationary phase at 30°C. The mobile phase, acetonitrile with HPLC‐grade water (75:25, v/v), was used with a flow rate of 1 mL/min. Detection was carried out using a refractive index detector (G7162A) on the Agilent 1260 Infinity II system. The peak areas of the 0.1–50 mg/mL fructose, glucose and sucrose standards were used for calibration and to establish standard curves. The inulin concentration was determined by comparing the fructose, glucose and sucrose contents between undigested and digested samples of the same inulin extract. The mean degree of polymerisation (DPn) was calculated as previously reported (Hahn et al. 2016). The starch content was measured with a starch content assay kit (Solarbio, China) following the manufacturer's instructions.

5.9. Measurement of Sugar Contents in the Roots of TKS via GC/MS

Three biological replicates of the roots from 1‐sst1‐fft (sf‐1, sf‐2 and sf‐3) and WT (WT‐1, WT‐2 and WT‐3) were used to analyse sugar contents. Each replicate comprised the roots of four individual plants. Specifically, sf‐1 included ssff‐27, ssff‐28, ssff‐33 and ssff‐40; sf‐2 included ssff‐1, ssff‐21, ssff‐22 and ssff‐24; and sf‐3 included ssff‐10, ssff‐17, ssff‐31 and ssff‐36. A total of 20 mg of freeze‐dried material was diluted to 500 μL with methanol:isopropanol:water (3:3:2, V/V/V), vortexed for 3 min and ultrasonicated for 30 min. The extract was centrifuged at 14 000 rpm at 4°C for 3 min. Fifty microlitres of the supernatant was mixed with 20 μL of internal standard (1000 μg/mL) and evaporated under a nitrogen gas stream. The evaporated sample was transferred to a lyophilizer for freeze‐drying. The residue was mixed with 100 μL of a solution of methoxyamine hydrochloride in pyridine (15 mg/mL). The mixture was incubated at 37°C for 2 h. Then, 100 μL of bistrifluoroacetamide (BSTFA) was added to the mixture and incubated at 37°C for 30 min after vortex mixing. The mixture was examined by MetWare (China) on the Agilent 8890‐5977B GC/MS platform, and the parameters were set as described in previous studies. Significant differences in sugar accumulation between groups were determined by applying a fold change ≥ 2 or a fold change ≤ 0.5 as thresholds, with p < 0.05. The hierarchical cluster analysis (HCA) results of the samples and metabolites are presented as heatmaps with dendrograms. For HCA, normalised signal intensities of metabolites (unit variance scaling) are visualised as a colour spectrum.

5.10. Measurement of Terpenoid Contents in the Roots of TKS via Ultra Performance Liquid Chromatography Tandem Mass Spectrometry (UPLC–MS/MS)

Fifty milligrams of the freeze‐dried roots of 1‐sst1‐fft and WT plants were weighed from the three biological replicates mentioned above, and 1200 μL of 70% methanolic aqueous internal standard extract precooled at −20°C was added. The mixture was vortexed once every 30 min for 30 s, and this process was repeated 6 times. After centrifugation (12 000 rpm, 3 min), the supernatant was aspirated and then filtered through a microporous membrane (0.22 μm). The sample extracts were examined by MetWare via a UPLC–ESI–MS/MS system and a tandem mass spectrometry system. The analytical conditions were as follows: UPLC: column, Agilent SB‐C18 (1.8 μm, 2.1 mm × 100 mm). The mobile phase consisted of solvent A, pure water with 0.1% formic acid and solvent B, acetonitrile with 0.1% formic acid. Sample measurements were performed with a gradient programme using starting conditions of 95% A and 5% B. This was followed by a linear gradient to 5% A and 95% B in 9 min, and this composition was maintained for 1 min. Subsequently, the composition was changed to 95% A and 5.0% B within 1.1 min and maintained for 2.9 min. The flow velocity was set as 0.35 mL per minute; the column oven was set to 40°C; and the injection volume was 2 μL. The effluent was alternatively connected to an ESI‐triple quadrupole‐linear ion trap (QTRAP)‐MS. The ESI source operation parameters were as follows: source temperature, 500°C; ion spray voltage, 5500 V (positive ion mode)/−4500 V (negative ion mode); ion source gas I, gas II and curtain gas set at 50, 60 and 25 psi, respectively; and collision‐activated dissociation, high. QQQ scans were acquired via MRM experiments with the collision gas (nitrogen) set to medium. A specific set of MRM transitions was monitored for each period according to the metabolites eluted within the period. Differentially accumulated metabolites were determined by applying the thresholds variable importance in projection (VIP) > 1, fold change ≥ 3, and p < 0.01. VIP values were extracted from the orthogonal partial least square discriminant analysis (OPLS‐DA) results, which included score plots and permutation plots, generated via the R package MetaboAnalystR. The data were log transformed (log2) and mean centred before OPLS‐DA. To avoid overfitting, a permutation test (200 permutations) was performed.

5.11. RNA‐Seq and Data Analysis

The same samples mentioned above for sugar and terpenoid determination were also used for RNA‐Seq analysis. The roots of 1‐sst1‐fft and WT plants were collected for RNA extraction and transcriptome sequencing in three biological replicates. Total RNA was extracted with Trizol reagent (Invitrogen, USA), and cDNA library preparation and transcriptome sequencing were carried out via the Illumina HiSeq platform by MetWare. Reads containing adapters, sequences with more than 10% unknown nucleotides (N) and a quality rating less than 50% (Q ≤ 20) were first removed from the dataset. The clean reads were mapped to the TKS reference genome (GWHBCHF00000000) with HISAT2 using the default parameters. The expression levels were calculated in terms of fragments per kilobase of transcript per million fragments mapped (FPKM). The DEGs were determined via the DESeq2 R package (1.22.1), with |log2 fold change| > 1 and the adjusted p < 0.05 applied as thresholds. To infer the putative functions of the DEGs, we conducted GO function and KEGG pathway analyses via the OmicShare tools (http://www.omicshare.com/tools) with the default parameters.

5.12. RNA Extraction and qRT–PCR Analysis

Total RNA was isolated from TKS leaves and roots via an RNAprep Pure Plant Plus Kit (TIANGEN). cDNA was synthesised by HiScript III RT SuperMix for qPCR (+gDNA wiper) (Vazyme, China) using 1 μg of total RNA as the template. Additionally, to validate the DEGs, qRT–PCR analyses were conducted using mRNAs extracted from the same samples as those used in the aforementioned RNA‐Seq study. qRT–PCR was performed using the ChamQ Universal SYBR qPCR Master Mix (Vazyme) according to the manufacturer's protocol. The primer sequences are listed in Table S7. Actin was used as a reference gene in the qRT–PCR assay, and the 2^−ΔΔCt method was used to calculate the relative fold change values of the target genes (Livak and Schmittgen 2001).

5.13. Enzyme Activity Determination in TKS Leaves and Roots

Enzyme activity assays were conducted using crude enzyme extracts prepared from 0.1 g of fresh leaves or roots of TKS according to the manufacturer's recommended protocols. The following assay kits from Boxbio were utilised: sucrose phosphate ynthase (SPS) Activity Assay Kit, Plant Sucrase (P‐SC) Activity Assay Kit, Neutral Invertase (NI) Activity Assay Kit and Acid Invertase (AI) Activity Assay Kit. For SPS activity determination, absorbance measurements at 480 nm were interpolated into a sucrose standard curve, with enzyme activity expressed as μg sucrose g⁻¹ FW min⁻¹. For sucrase, neutral invertase, and acid invertase activities, absorbance measurements at 540 nm were interpolated into a glucose standard curve, with enzyme activities expressed as μg glucose g⁻¹ FW min⁻¹. HMGR and HDR activity were measured using ELISA kits (Plant HMGR ELISA Research Kit and Plant HDR ELISA Research Kit, MEIMIAN, China) following the manufacturer's instructions. Briefly, assay plates were coated with mouse anti‐HMGR or anti‐HDR antibody, blocked with 5% skim milk at 25°C for 2 h, and incubated with diluted crude enzyme extracts (1:4 v/v) at 37°C for 30 min. After washing, biotin‐labelled antibodies and HRP‐conjugated avidin were added for detection. Colour development was performed with 3,3′,5,5′‐tetramethylbenzidine (TMB) substrate and absorbance was measured at 450 nm. Standard curves were generated using two enzyme standards at different concentrations, and activities were expressed as U/L of crude enzyme extract.

5.14. Statistical Analysis

The data were analysed with SPSS 24.0 (IBM Corporation, USA), and p < 0.05 was considered indicative of statistical significance. Analysis of variance (one‐way ANOVA) was used to compare significant differences in various traits between the mutants and WT plants. The histograms were drawn with OriginPro 2018 (Origin Lab Corporation, USA).

Author Contributions

Bi Qin and Shizhong Liu designed the research and wrote the manuscript. Bi Qin, Chuang Li, Xiaoxiao Wang and Pin Lin performed Agrobacterium‐mediated transformation, mutation identification and propagation. Xue Hou, Songle Fan, Qiuhui Chen, Yushuang Yang, Liang RuiNan Lu and Canni Fu performed the experiment. Bi Qin, Songle Fan and Chuang Li analysed the data. Ning Yang, Qiaoli Gao and Shanshan Zheng revised the manuscript. All authors read and approved the final manuscript.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Figure S1: DNA sequence variation analysis of Tk1‐SST and Tk1‐FFT. Multiple sequence alignment of Tk1‐SST (a) and Tk1‐FFT (b) from different germplasms.

Figure S2: CRISPR/Cas9 recombinant expression vector used in this study. The structure (a) and sequence (b) of the CRISPR/Cas9 recombinant expression vector.

Figure S3: Analysis of Cas9 gene expression in different mutant lines. Relative expression of the Cas9 gene in different genotype plants (a) and different homozygous mutants (b). Cas9 transcript levels were quantified relative to the expression level of the housekeeping gene Actin. n.d. indicates no detection. The different lowercases indicate a significant difference with p < 0.05.

Figure S4: Mutants and WT plants for agronomic trait, transcriptome and metabolite analysis. The phenotype of 1‐sst1‐fft and WT plants at ~170 DAC for assessing plant biomass and root characteristics (a), and plants used for analyses of transcriptome, carbohydrate and terpenoid accumulation in the roots at ~200 DAC. Bar = 5.0 cm.

Figure S5: Analysis of inulin and NR content and their correlation with differentially accumulated metabolites. The average contents of inulin and DP (a) and NR (b) were determined in mixed samples from different lines, which were used for transcriptome, sugar and terpenoid analysis. ** indicates a significant difference with p < 0.01. (c) Pearson correlation map of NR, inulin and differentially accumulated carbohydrates and terpenoids in the 1‐sst1‐fft mutants and WT. *, **, *** indicates p < 0.05, p < 0.01 and p < 0.001, respectively; the ellipses represent the correlations, with finer characters and deeper colours indicating stronger correlations.

Data S1: The mutation types and mutation ratios for Tk1‐SST and Tk1‐FFT in all the T₀ plants.

Data S2: The mutation types and mutation ratios for Tk1‐SST and Tk1‐FFT in T₀, T₁ and T₂ generation.

Data S3: Genes and their corresponding loci harbouring potential off‐target sites and InDel mutations.

Data S4: Annotation of differentially expressed genes identified in the sf vs. WT comparisons.

Data S5: Summary of KEGG and GO enrichment of DEGs.

Data S6: DEGs involved in the starch and sucrose metabolism pathway, sugar signalling, and sugar transport.

Data S7: DEGs involved in MVA, MEP and NR biosynthesis.

Data S8: DEGs involved in sesquiterpenoid and triterpenoid biosynthesis.

Data S9: Sugar levels determined in the roots of 1‐sst1‐fft mutants and WT plants using GC/MS.

Data S10: Terpenoids identified by UPLC–MS/MS and annotated.

Data S11: The differentially accumulated terpenoids identified in the sf vs. WT comparisons.

Table S1: Genotypes of T₀ plants.

Table S2: All the homozygous mutant plants used in this study and their mutation sites.

Table S3: The mutation types and segregations for Tk1‐SST and Tk1‐FFT in T₁ and T₂ generations.

Table S4: CRISPR‐mutagenesis off‐target site identification and analysis.

Table S5: The statistics of RNA‐Seq data.

Table S6: Statistics of data mapping to the TKS reference genome.

Table S7: Primers for target gene amplification, vector construction, mutation detection, off‐target site identification and qRT–PCR analysis.

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.