Antagonism between abscisic acid and gibberellin regulates starch synthesis and corm development in Gladiolus hybridus

Understanding corm development in flower bulbs is of importance for securing the quality of cut flowers and propagation of commercial stocks. Gladiolus is one of the most popular bulb plants worldwide. Its corm development is characterized by starch accumulation. Previous research has shown that phytohormones (especially gibberellin (GA)) are involved in tuber development. However, the relationship between abscisic acid (ABA)/GA and starch during corm development remains unclear. To gain deeper insights into the biological process of corm development, we performed a detailed anatomical characterization of different stages of corm development and analyzed phytohormone levels. Our study showed that corm development is linked to hormones (ABA and GA) and carbohydrates (sucrose and starch). Exogenous hormone treatment and silencing of endogenous hormone biosynthesis genes indicated that ABA positively regulates corm development, while GA acts as an antagonist of ABA function. A sucrose synthase gene (GhSUS2) was shown to be involved in the antagonism between ABA and GA. GhSUS2 was upregulated by ABA and downregulated by GA. The increase in the transcript level of GhSUS2 coincided with the development of corm/cormels. Silencing of GhSUS2 repressed corm development and starch accumulation. In conclusion, we propose that GhSUS2, an essential enzyme in sucrose degradation, is differentially regulated by ABA and GA and controls corm development in Gladiolus.


Introduction
A storage organ is a portion of a plant specifically modified as a reserve of energy or water. This is an evolutionary strategy that maintains plant survival and propagation from generation to generation. The storage organs of geophytes are often found underground and are modified from the root, leaf, stem, or hypocotyl. They include bulbs, tubers, corms, rhizomes, and root tubers. Storage organs serve as sources of food (e.g., potato), commercial goods (e.g., Fritillaria spp.), and decorations for gardens (e.g., Gladiolus spp.). To date, the development of storage organs is poorly understood, except for potato. For wild potato species, tuberization is regulated by environmental factors, such as ambient temperature, photoperiod, and nitrogen 1,2 . Tuberization is stimulated under short-day conditions and promoted by low night temperatures 1 . Under long-day or hightemperature conditions, inhibition of tuberization is mediated by the CONSTANTS/FLOWERING LOCUS T (CO/FT) protein 3,4 . In hydroponically cultivated potatoes, tuberization is induced by withdrawing nitrogen from the nutrient solution, indicating that nitrogen has an inhibitory effect on tuberization 5 . In the context of phytohormones, GA dominates tuberization 6 . The GA level decreases when the stolon tip starts to swell, and excess GA inhibits the transition from stolon to tuber 7,8 . GA metabolism genes (StGA20ox1 and StGA3ox2) have been shown to be involved in tuberization 9,10 . Cytokinins and jasmonic acid (JA) positively induce tuberization 11,12 . Aside from insights into the regulation of tuberization in potato, the physiological and molecular mechanisms of storage organ development in other species are poorly studied. Recently, by transcriptome and physiological analyses, a GA inhibitor was shown to stimulate the shoot-to-bulblet transition of lily in vivo, and starch was found to be a fundamental compound in this process 13 . However, direct genetic evidence of the role of hormones in bulb development remains elusive.
As the principal carbohydrate in storage organs, starch plays crucial roles in the development of storage organs, such as tubers and seeds [14][15][16] . Starch is an insoluble glucan comprised of amylopectin, amylose, and two polymers of glucose 17 . Sucrose is loaded from photosynthetic organs (e.g., leaves) and unloaded in nonphotosynthetic organs (i.e., sink organs) where sucrose is converted to starch for long-term storage in amyloplasts. Upon arriving in sink organs, sucrose is transported into sink cells via transporters or plasmodesmata 18,19 . In the cytosol, sucrose can be hydrolyzed by cytosolic invertase (CIN) to fructose and glucose or cleaved by sucrose synthase (SUS) to fructose and uridine diphosphate glucose (UDP-G). In the latter pathway, UDP-G can be further transformed to adenosine diphosphate glucose (ADP-G) by UGPase and AGPase 20 . ADP-G is the main substrate for starch biosynthesis in angiosperms. The glucosyl moiety is added to existing glucan chains by starch synthases 17 . Starch synthesis is regulated by phytohormones. In maize and potato, GA inhibits starch accumulation, while its inhibitor (chlorocholine chloride) significantly increases starch content 21,22 . In rice internodes, overexpression of ethylene receptors (ETHYLENE RESPONSE2/ETR3) induces the accumulation of starch granules 23 .
SUS is a glycosyltransferase that catalyzes the reversible transfer of a glucosyl moiety between fructose and UDP-G 24 . UDP-G is used to biosynthesize starch or cellulose in plants 25 . Selective phosphorylation of sucrose synthase isoforms results in enhanced sucrose degradation 26 . SUS promotes vegetative growth, early flowering, plant biomass accumulation, and the response to low oxygen [27][28][29] . Moreover, SUS decomposes sucrose in sink organs, generating a sucrose gradient flow from source to sink, i.e., the flow provides pressure for the transport of sucrose from the phloem to the sink and ensures the continuous supply of sucrose to the sink in potato. Reducing SUS activity in potato results in reduced tuber dry weight and a lower content of starch 30 .
Although much is known about sink organs in model plants, it has become clear that research on model plants alone will not provide adequate information for the improvement of starch accumulation in all plant species. Indeed, factors controlling starch metabolism differ among species and organs 17 . Therefore, it is worthwhile to study the role of starch in the development of storage organs in nonmodel plants. Gladiolus is one of the most widely planted geophytes worldwide, and its corm is the only plant stock and propagation tissue used for commercial cultivation. Corm degradation, which leads to decreased cut flower production and propagation yield, is one of the most severe problems in Gladiolus cultivation. The gladiolus corm is a specialized underground organ consisting of an enlarged stem axis with distinct nodes and internodes and enclosed by dry, scale-like leaves (tunics). A new corm is generated each growing season over the mother corm. Meanwhile, cormels are produced at the tips of branched stolons that develop from buds located at the base of the new corm 31 . Corm development is tightly related to carbohydrates. During the rapid expansion of corm and cormels, starch and sugar levels increased sharply 32 . Recently, we have shown that induction of starch metabolic genes (the small/large subunit of ADP-glucose pyrophosphorylase; GhAGPS1/ GhAGPL1) stimulates corm development and cormel numbers 33,34 . Additionally, JA could promote corm expansion, and the expression pattern of GhLOX1 (LIPOXYGENASE), a key JA biosynthesis gene, coincides with corm development 35 . Whether other hormones participate in corm development and how hormones regulate carbohydrates in corms are still not clear.
Here, we show that the levels of starch and endogenous ABA increase sharply during the transition from stolon to cormel, while the GA 3 level decreases. Interestingly, exogenous ABA promotes corm weight and cormel numbers, and GA 3 represses the stolon-to-cormel transition. Furthermore, decreasing the ABA level in the corm by silencing GhNCED dramatically inhibited starch accumulation and the expansion of corms, which was the opposite of the phenotype observed in GhGA20oxsilenced corms. We show that GhSUS2 expression is induced by ABA but inhibited by GA 3 . GhSUS2 is expressed mainly in sink organs (corms and cormels) and upregulated during corm development. Silencing GhSUS2 in corms impairs corm development. Hence, we propose that ABA and GA play antagonistic roles in corm development by differentially regulating GhSUS2 expression and starch synthesis in Gladiolus.

Results
Endogenous ABA and GA levels, along with starch accumulation, change dramatically during Gladiolus cormel development Early cormel development in Gladiolus can be divided into 5 stages (Fig. 1A): stolons, with a tip that is not swollen (I); pale cormels (II; 0-5 mm in diameter); yellow cormels (III; 5-7 mm in diameter); expanded cormels (IV; 7-9 mm in diameter) and cormels that become planting stocks in commercial cultivation (V; approximately 1 cm in diameter). To characterize the different stages of cormel development (I-V), we observed cells and sugar distribution by periodic acid-Schiff (PAS) staining. The results showed that parenchyma cells and vascular cells in stolons (stage I; Fig. 1B) accumulated or transported fewer sugars than formed cormels (stages II-V). After cormels formed at the tips of stolons, parenchyma cells started to accumulate sugars that were delivered from the mother corm (Fig. 1B). The distribution of sugars in cormels was not equal, with the bottom of cormels accumulating more sugars (Fig. 1B). At later stages (IV to V), the parenchyma cells were much larger, and the vascular cells accumulated sugars at the tip (Fig. 1B).
Next, we analyzed the starch and sucrose levels at the different developmental stages. The levels of both starch and sucrose were low in stolons but increased gradually in cormels from stage II to V (Fig. 1C, D).
As phytohormones regulate plant development, including germination, vegetative development, and reproductive development 36 , we quantified phytohormone levels during cormel development. When corms formed at the tips of cormels (transition from I to II), the endogenous ABA level increased sharply, while the GA 3 level decreased ( Fig. 1E-F). When cormels developed from III to V, endogenous ABA was maintained at a relatively high level (although not significantly higher than the level at stage I), while GA 3 remained at a relatively stable lower level than the level at stage I ( Fig. 1E, F). These results indicate a dramatic increase in sucrose and starch levels during corm development, accompanied by a transient increase in the ABA level and a decrease in the GA level.

Exogenous ABA and GA affect the development of Gladiolus corm and cormels
Given the opposite changes in ABA and GA 3 levels during the transition from stage I to II (Fig. 1E, F), we hypothesized that these hormones play opposite roles in Gladiolus cormel formation and enlargement. To test our hypothesis, we treated plants with ABA and GA 3 at 10 weeks after planting (WAP). Usually, cormels are largely formed at 14 WAP. After 6 weeks of treatment, plants under ABA treatment had larger mother corms ( Fig. 2A-C). Moreover, ABA also significantly promoted the formation and development of cormels, resulting in more cormels and higher yield (Fig. 2D, E). Plants subjected to GA 3 treatment had a slightly smaller circumference, but their corms were much heavier ( Fig. 2B-C). GA 3 played a negative role in corm formation by reducing the number of cormels (Fig. 2D). Taken together, the results show that ABA plays a positive role in the formation and development of cormels and corm development, while GA negatively affects the formation of cormels.

Silencing ABA and GA biosynthesis genes affects Gladiolus cormel development
To further address the role of ABA and GA in cormel development, we silenced ABA and GA biosynthesis genes in cormels. Silencing GhNCED (NINE-CIS-EPOXYCAROTENOID DIOXYGENASE), a key enzyme in ABA biosynthesis in Gladiolus cormels, led to reduced ABA content (Fig. S1), increased GA 3 content (Fig. S1) and early sprouting 37 . Here, we found that silencing GhNCED in corms before planting also reduced the starch content in leaves (Fig. 3A) and led to the generation of smaller and lighter corms (Fig. 3B-D). Conversely, silencing the GA biosynthesis gene GhGA20ox decreased GA 3 levels and had the opposite effect, promoting the development of enlarged and heavier corms (Fig. 3B-D;  Fig. S1). Furthermore, the starch content in GhNCEDsilenced corms was dramatically lower than that in the control, while it was much higher in GhGA20ox (gibberellin 20-oxidase)-silenced corms (Fig. 3A, E, F). The sucrose levels did not change as obviously as those of starch in the silenced corms. These results, taken together with the previous results ( Figs. 1 and 2), led us to conclude that ABA has positive effects, while GA has negative effects, on corm development. The effect of ABA and GA on corm development may be correlated with starch.
Expression of GhSUS2 is correlated with Gladiolus corm development and regulated by ABA and GA Previous work has shown that SUSy, not INV, is the dominant active enzyme in actively growing sink organs in geophytes, such as potato tuber and cassava roots 19,38,39 . Moreover, SUSy is essential for sink strength, especially in starch-accumulating organs 19 . To identify corm-expressed GhSUSs, we screened our cormel transcriptome database 40 . We found six unigenes expressed in cormels, of which GlaUn069031 was the most abundant ( Fig. S2A). GlaUn069031 has a sequence similar to those of the homologous genes in Arabidopsis (AtSUS2) and rice (OsSUS2), so we named it GhSUS2 (Fig. S3A). GhSUS2 shared a conserved Ser-phosphorylation site at the N-terminus of the amino acid sequence (Fig. S3B).
Given that ABA and GA affected starch synthesis in Gladiolus (Fig. 3), we investigated the relationship between these two hormones and GhSUS2. First, we explored the expression pattern of GhSUS2 in corms treated with ABA and GA 3 . Quantitative RT-PCR results revealed that GhSUS2 expression was induced by ABA and slightly repressed by GA 3 , although the change was not statistically significant (Fig. 4A). We then tested the transcript level of GhSUS2 in GhNCED-and GhGA20oxsilenced corms. The results showed that GhSUS2 expression was decreased in GhNCED-silenced corms but The experiments were performed with three biological replicates (10 corms per biological replicate). The significant differences between the mock and treatments were determined by Tukey's multiple comparison test. The P value is indicated above the black line increased in GhGA20ox-silenced corms (Fig. 4B). Altogether, our data indicate that GhSUS2 expression is induced by ABA and repressed by GA.
To test the role of GhSUS2 in cormel development, we analyzed its expression pattern in different stages of cormel development (I-V) and in different organs. GhSUS2 was highly expressed in sink organs (corms and cormels), and its expression level gradually increased with cormel development (Fig. 4C, D). In addition, we also tested GhSUS2 expression in cormels at different growth points. The cormel was formed starting from 10 weeks after planting (WAP), developed until 26 WAP, and dried at room temperature for 4 additional weeks. The expression level of GhSUS2 closely matched the level of sucrose accumulation in cormels (Fig. S2B, C).
Plant sucrose synthase isozymes are mainly located in the cytosol or adjacent plasma membrane 19 . To trace the protein localization, GFP was fused in frame with GhSUS2 (pSuper: GFP-GhSUS2). A green fluorescence signal of GFP-GhSUS2 was observed in the cytosol (Fig. 4E), showing similar subcellular localization as its homologs in potato 25 . This finding suggests that GhSUS2 is a cytosolic SUS.

Silencing of GhSUS2 represses Gladiolus corm development by decreasing starch content
As the expression of GhSUS2 is correlated with starch synthesis and corm development, we reasoned that corm development should be regulated by GhSUS2. To test our hypothesis, we silenced GhSUS2 in corms before planting them. After four months of growth, the silenced plants were dug out, and the corm size, corm weight, and cormel frequency were measured. Notably, the corm diameter in GhSUS2-silenced plants was significantly smaller than that in the control (Fig. 5A, B). The fresh weight of silenced corms was also smaller than that of the control (Fig. 5C). In addition to its effects on corm development, GhSUS2 regulated cormel formation. The cormel   Silencing of GhSUS2 in Gladiolus reduced the corm diameter (B) and corm fresh weight (C). D. The cormel frequency in GhSUS2-silenced plants was lower than that in the control. B to D Averages of four biological repeats, ± SDs (n = 24). Significant differences were determined by the two-sided ttest. The P value is indicated above the black line frequency in the silenced plants was lower than that in the control (Fig. 5D).
To better understand the role of GhSUS2 in corm development, we determined the expression pattern of key genes in starch biosynthesis in GhSUS2-silenced corms (Fig. 6A). Cytosolic invertase degrades imported sucrose to glucose and fructose in the cytosol 41 . Here, we found that the transcript level of GhCIN was sharply increased in the silenced corms, and accordingly, invertase activity was increased (Fig. 6B, H). The glucose content was higher in the GhSUS2-silenced corms, but the sucrose content was lower (Fig. 6D, E). In previous research, GhAPS has been shown to play a positive role in regulating corm development and starch synthesis 33 . qRT-PCR showed that GhAPS was downregulated when GhSUS2 was silenced (Fig. 6C). Additionally, both sucrose synthase and soluble starch synthase had lower activity in GhSUS2-silenced corms than in the control (Fig. 6G, I). For these reasons, the starch content in GhSUS2-silenced corms was lower than that in the control (Fig. 6F). Overall, we concluded that GhSUS2 promotes corm development by modulating starch synthesis.

Discussion
Starch serves as the main storage compound in higher plants and accumulates in sink organs (e.g., seeds, bulbs, roots, and buds) before the end of each life cycle. Transcriptome analysis in potato and Lilium suggests that sucrose and starch metabolism is involved in the tuberization and the shoot-to-bulblet transition 13,20,[42][43][44] . In temperate zones, most plants enter dormancy during winter. ABA promotes seed dormancy, bud dormancy and bulb dormancy. Indeed, the ABA level gradually increases during the development of dormant organs (before they become dormant) [45][46][47] . However, the link between ABA and starch at this developmental stage is less well understood. Here, we found that ABA plays a positive role in corm development and cormel development. While ABA promotes starch synthesis by upregulating GhSUS2 expression, GA has the opposite effect. Furthermore, we showed that starch synthesis positively regulates corm H The enzyme activity of invertase was increased in GhSUS2-silenced corms. Averages of three to five biological replicates ± SDs are shown. Significant differences were determined by the two-sided t-test. The P value is indicated above the black line development. We propose that ABA promotes the development of corms in Gladiolus by increasing starch synthesis through upregulation of GhSUS2, while GA has the opposite effect (Fig. 7).
ABA and GA are two classic hormones in plants that antagonistically regulate several plant developmental processes, including seed maturation, seed dormancy and germination, primary root growth, and flowering time [48][49][50] . However, GA does not always function as an antagonistic hormone to ABA, e.g., GA inhibits corm dormancy release in the early stage of Gladiolus dormancy release, which is similar to the role of ABA 45,51 . Here, we provide evidence that ABA and GA antagonize corm development in Gladiolus: (i) during the transition from stolon to cormel, ABA production was induced and GA production was decreased (Fig. 1); (ii) opposite phenotypes of corm circumference and cormel numbers were observed following treatment with ABA or GA (Fig. 2); (iii) silencing GhNCED or GhGA20ox had opposite effects on corm development (Fig. 3). Notably, we found that GA 3 could increase corm weight and the uniformity of cormels (Fig. 3). This might have been caused by cell expansion under GA treatment 52 . GA was shown to promote stolon elongation and reduce the activity of ADP-G 1,53 . Moreover, GA could also regulate tuber formation by mediating the FT/CO pathway 3 . However, JA and ABA do not seem to be effective hormones in tuberization 1 . In a recent study, ectopic expression of AtABF4 (ABRE BIND-ING FACTOR 4 in Arabidopsis) in potato caused GAdefective phenotypes and significantly increased the number and weight of the tubers obtained, suggesting that ABA/ GA cross-talk may be involved in tuberization 6 . Although potato tuberization is regulated by the crosstalk between ABA and GA, the effects of ABA are always mediated by a change in the levels of GA, which is ultimately the master regulator of tuberization 6,54,55 . Previously, JA was reported to promote corm expansion in Gladiolus 35 , suggesting that the effects of hormones may be species dependent.
In this study, we showed that ABA increases the starch content in corms and promotes corm development and stolon-to-cormel transition. Typically, discourse on tuber development revolves around gibberellins. In the regulation of Gladiolus corm development and starch/sucrose levels, the effect of ABA was more obvious than that of GA, suggesting that ABA plays a more prominent role in corm development. GhNCED is most abundant in the stolon at the flowering/blooming stage when corms start to emerge 37 . This is in agreement with the fact that during the cormel transition, the levels of ABA and starch sharply increased (Fig. 1). NCED is a key enzyme in ABA biosynthesis and is widely involved in plant development and abiotic stress 56,57 . Unlike its role in model plants such as Arabidopsis, rice, and tomato, the role of NCED in modified organs is less well known. Here, we characterized the role of GhNCED in sink organs (corm and cormels) and starch synthesis (Fig. 3). Moreover, we found roles for ABA in promoting starch accumulation after stolon swelling, in cormel transition and in corm dormancy 37,45 .

GhSUS2 positively regulates corm development
Sucrose synthase was discovered in beets, sweet sorghum, and pea seeds in 1955 58 . Most SUS members are found in the cytosol, and some are associated with the plasma membrane. A few SUS isoforms are found in the cell wall and other organelles (e.g., vacuole membrane, cytoskeleton, mitochondria, and Golgi apparatus) 19,[59][60][61][62] . SUS is the primary active enzyme involved in the breakdown of sucrose in sink organs, enhancement of sink strength, and active growth of sink organs 19 . SUS is highly expressed in storage organs such as seeds, fruits, and taproots and is often positively correlated with starch content and fruit size [63][64][65] . In potato, reducing SUS by an antisense technique markedly decreased the starch content and tuber yield 63 . Here, we found that GhSUS2 is localized in the cytosol and that the expression of GhSUS2 is positively related to the early developmental stage of cormels and starch accumulation (Fig. 1C, D, E; Fig. S2B). Silencing of GhSUS2 in corms resulted in lower levels of GhAPS, sucrose and starch and reduced SSS enzyme activity (Fig. 6). Intriguingly, we found that INV activity and glucose content were increased in GhSUS2-silenced corms (Fig. 6). As SUS activity was reduced in the silenced corm, INV might have taken over sucrose degradation, resulting in increased glucose level. It has been shown that a high concentration of sucrose can induce tuberization in potato and bulb formation in onion 7,66 . In our GhSUS2silenced corms, the sucrose level was also lower than that in the control (Fig. 6). Moreover, ectopic expression of GhSUS2 in Arabidopsis could promote root elongation on sucrose-containing media (Fig. S4), suggesting that GhSUS2 could accelerate sucrose degradation in cells. In GhSUS2-silenced corms, there was a small decrease in starch content, suggesting that silencing GhSUS2 may not only reduce starch biosynthesis but also slow starch degradation or conversion. Given this evidence, we speculate that GhSUS2 is involved in degrading sucrose in the cytosol and in converting sucrose to starch to promote the sucrose flow from source organs to sink organs.
In plants, the relationship between ABA and sucrose synthase is variable depending on the species. In pea, ABA induces a decline in nitrogen fixation in a manner that is independent of SUS 67 . In rice, ABA treatment shows a positive relationship between SUS and grain filling 68 . How does ABA relate to SUS in Gladiolus corm development? We utilized both exogenous ABA treatment and silencing of the ABA synthesis gene GhNCED to investigate the effects of ABA. Both assays showed that ABA increases the transcription of GhSUS2 and starch in corms. Recently, a bZIP family member, AtABF4, was demonstrated to increase tuber yield through ABA-GA crosstalk regulation 49 . It will be interesting to investigate how ABA signaling-related transcription factors are involved in regulating GhSUS2 in Gladiolus in future work.
In conclusion, we showed that GhSUS2, an essential enzyme in the starch biosynthesis pathway that mediates the antagonism of ABA and GA, plays an important role in corm development in Gladiolus.

Plant materials and treatments
The Gladiolus cultivar 'Rose Supreme' was planted in the Science Research Garden at China Agricultural University. For tracking corm development, stolons and cormels (Φ = 0-5, 5-7, 7-9, 9-11 mm) were sampled at 10 WAP. For exogenous hormone treatment, uniform corms (6 cm in circumference and 3.0-3.8 g) from plants at 10 WAP were irrigated with ABA (0.5 mg/L; Solarbio, Beijing, China), GA 3 (0.5 mg/L; Solarbio, Beijing, China), or water (the control) every 3 days. The plants were lifted after six weeks of treatment. Three biological replicates were examined (10 corms per biological replicate). For counting cormels, the expanded stolons, which were empty inside and not real cormels, were not included.

Measurement of starch, glucose, and sucrose
Gladiolus corms or stolons (50 mg) were ground in powder by liquid nitrogen. The extraction and quantification of starch, glucose, and sucrose were performed as described by Fan et al. 69 .

Measurement of ABA and GA 3
Gladiolus corms or stolons were collected (50 mg) and ground to a powder with liquid nitrogen. The extraction procedure was performed as described by Wu et al. 37 . d 6 -ABA and d 2 -GA 3 were used as internal standards. The extracts were analyzed by HPLC-MS/MS 70 . Three biological replicates were examined.

PAS staining
Gladiolus corms and stolons were fixed in formalinaceto-alcohol (FAA) solution. The dehydration and embedding procedures were performed according to a previously published protocol 71 . Paraffin sections (8 μm) were obtained using a slicing machine. The slides were stained with PAS 72 and photographed by a light microscope (Olympus BX51, Tokyo, Japan). Sugars were stained in red by PAS.

Virus-induced gene silencing in Gladiolus
Silencing of the target gene (GhSUS2, GhNCED, or GhGA20ox) by VIGS was performed as described by Zhong et al. 73 with some modifications. Briefly, a 200-500 base pair (bp) fragment specific to the target gene was generated and cloned into the pTRV2 vector (primers are listed in Table S1). The TRV1, TRV2, and TRV2 target genes were transformed individually into the Agrobacterium GV3101 strain. The transformed colonies were cultured overnight in LB medium containing 50 mg/L kanamycin and 50 mg/L rifampicin. Then, bacteria were collected and resuspended in infiltration buffer (10 mM MgCl 2 , 200 mM acetosyringone, and 10 mM 2-(N-morpholino) ethanesulfonic acid (pH 5.6)) to a final OD 600 of 1.8. Equal volumes of TRV1 and TRV2 (the control), as well as the TRV1 and TRV2 target genes, were mixed together and kept in the dark for 3 h at 25°C before vacuum infiltration. Corms were submerged in infiltration buffer and infiltrated under 0.9 MPa for 30 min. Ultimately, the corms were planted in pots and grown in a green chamber at 22°C under 16/8 h light/dark for four months.

Iodine staining of starch
Gladiolus leaves were submerged in 75% (v/v) ethanol at 80°C until chlorophyll was eliminated. Then, the leaves were washed with ddH 2 O and stained with Lugol's solution [0.06% I 2 (w/v), 0.1% KI (w/v), and 4 mM HCl] for 10 min. Finally, the stained leaves were rinsed with ddH 2 O for 15 min 33 .

RNA extraction and qRT-PCR
Total RNA from Gladiolus samples was extracted using the Tiangen RNA Extraction Reagent Kit (Tiangen, Beijing, China) and reverse transcribed with the M-MLV Reverse Transcriptase Kit (TaKaRa, Shiga, Japan). Approximately 400 ng of cDNA was used as the template for qRT-PCR and analyzed by using the Applied Biosystems StepOnePlus TM real-time PCR system with the Takara qRT-PCR kit. The Gladiolus actin gene acted as the reference gene 45 . The PCR procedures used were based on the manufacturer's instructions. All primers are listed in Table S1, and the data were analyzed with the 2 -ΔΔT method 74 .

Subcellular localization of GhSUS2-GFP
The coding sequence of GhSUS2 was cloned into pCAMBIA1300-GFP with the SalI and KpnI restriction sites (pSuper: GFP-GhSUS2). Both the fusion construct (GFP-GhSUS2) and the control (empty vector; GFP) were transiently transformed into onion epidermal cells by particle bombardment. After incubation at 25°C in the dark overnight, the cells were visualized using confocal microscopy (Zeiss LSM 710, Baden-Württemberg, Germany; 488 nm excitation and a 515-535 nm bandpass filter).

Activities of SUS, INV, and SSS
Gladiolus corms were extracted by grinding tissue in liquid nitrogen before adding 1 ml of extraction buffer [25 mM HEPES-KOH (pH 7.3), 5 mM ethylenediamine tetraacetic acid (EDTA), 0.1% (w/v) polyvinyl pyrrolidone (Mr 4000), 1 mM dithiothreitol, 0.01 mM leupeptin and 1 mM phenylmethylsulfonyl fluoride]. The extraction procedures were performed as described by Fan et al. 69 . The supernatants were used for the determination of cytosolic invertase activity as described by Wang et al. 75 . SUS activity assays, in the direction of sucrose breakdown, were carried out as described by Wang et al. 76 . Starch synthase activity was measured using the Soluble Starch Synthase Activity Assay Kit (Solarbio, Beijing, China).