BolTLP1, a Thaumatin-like Protein Gene, Confers Tolerance to Salt and Drought Stresses in Broccoli (Brassica oleracea L. var. Italica)

Plant thaumatin-like proteins (TLPs) play pleiotropic roles in defending against biotic and abiotic stresses. However, the functions of TLPs in broccoli, which is one of the major vegetables among the B. oleracea varieties, remain largely unknown. In the present study, bolTLP1 was identified in broccoli, and displayed remarkably inducible expression patterns by abiotic stress. The ectopic overexpression of bolTLP1 conferred increased tolerance to high salt and drought conditions in Arabidopsis. Similarly, bolTLP1-overexpressing broccoli transgenic lines significantly improved tolerance to salt and drought stresses. These results demonstrated that bolTLP1 positively regulates drought and salt tolerance. Transcriptome data displayed that bolTLP1 may function by regulating phytohormone (ABA, ethylene and auxin)-mediated signaling pathways, hydrolase and oxidoreductase activity, sulfur compound synthesis, and the differential expression of histone variants. Further studies confirmed that RESPONSE TO DESICCATION 2 (RD2), RESPONSIVE TO DEHYDRATION 22 (RD22), VASCULAR PLANT ONE-ZINC FINGER 2 (VOZ2), SM-LIKE 1B (LSM1B) and MALATE DEHYDROGENASE (MDH) physically interacted with bolTLP1, which implied that bolTLP1 could directly interact with these proteins to confer abiotic stress tolerance in broccoli. These findings provide new insights into the function and regulation of bolTLP1, and suggest potential applications for bolTLP1 in breeding broccoli and other crops with increased tolerance to salt and drought stresses.


Introduction
Thaumatin-like proteins (TLPs) are widely distributed in plants, animals and fungi [1], and are named for their high sequence similarity to the thaumatin protein initially identified in shrub Thaumatococcus daniellii Benth [2]. Most TLPs contain about 200 amino acids and have a molecular mass ranging from 21 to 26 kDa, except the poplar TLPs, most of which are 24 to 34 kDa in size [1,3,4]. In plants, TLPs are classified as pathogenesis-related protein family 5 (PR-5), one out of the 17 defense-related PR protein families [5,6]. Examination of genome databases revealed at least 37 TLPs in Oryza sativa, 33 in Zea mays, 18 in Vitis vinifera, 25 in Arabidopsis thaliana, 28 in Prunus persica and 42 in Populus trichocarpa [4,7]. However, the functions of the majority of TLPs are still unknown. Previous reports demonstrated that a few plant TLPs play important roles in defending against pathogen infection, particularly fungal infection. A 20 kDa TLP from the French bean legumes displays considerable resistance against Fusarium oxysporum, Pleurotus ostreatus and Coprinus comatus [8]. In Cassia didymobotrya, a 23 kDa TLP was identified, which exerts antifungal activity toward some Candida species [9]. TaLr35PR5, a TLP from wheat, was involved in Lr35-mediated adult wheat defense in response to leaf rust attack [10]. The recombinant expression of Mald2, a TLP identified in apple, conferred antifungal activity against Fusarium oxysporum and Penicillium expansum in Nicotiana benthamiana [11]. In tobacco, the ectopic overexpression of PpTLP from Pyrus pyrifolia or AdTLP from Arachis diogoi can enhance resistance to Rhizoctonia solani [12,13]. Transgenic canola, wheat and banana overexpressing TLPs from rice all exhibit increased resistance to diverse pathogens [14][15][16]. In addition to resistance against biotic stress, some TLPs in plants also respond to abiotic stress. Transgenic tobacco plants ectopically expressing AdTLP not only have enhanced resistance to Rhizoctonia solani, but also have improved tolerance to salt and oxidative stresses [13]. The ectopic expression of ObTLP1 isolated from Ocimum basilicum enhances tolerance to dehydration and salt stresses in Arabidopsis [17]. Under drought stress, TLPs showing inducible expression patterns were also identified in carrot and tea, suggesting that these TLPs could be involved in drought response [18,19]. In addition, TLPs have also been demonstrated to play roles in regulating plant growth and development including fruit maturation, lentinan degradation and seed development [20][21][22][23][24]. These investigations indicate that TLPs have undergone functional diversification in plant species, and play pleiotropic roles in response to biotic and abiotic stresses as well as in the regulation of growth and development [4,7].
Broccoli (Brassica oleracea L. var. italica) is one of the major vegetables among the Brassica oleracea varieties, and is widely planted in Asia, Europe and North America [25]. Soil salinization and drought are two potential threats in planting broccoli, which can result in curd yield reduction or even no harvest [26]. Breeding broccoli varieties with high tolerance to drought and salt is an important target for broccoli breeders. Several studies related to the responses to these two abiotic stresses were reported in broccoli and other species of Brassica. Broccoli pre-treating with SA and chitosan showed the highest drought stress recovery in a dose-dependent manner [27]. Drought-tolerant broccoli cultivars presented higher levels of methionine and ABA than that of the drought-sensitive cultivars [28]. 42 putative known and 39 putative candidate miRNAs displayed differentially expressed patterns between control and salt-stressed broccoli [29]. Bra-botrytis-ERF056 from cauliflower was demonstrated to increase tolerance to salt and drought stresses in overexpressing bra-botrytis-ERF056 Arabidopsis [30]. BrEXLB1 is involved in drought tolerance in Brassica rapa [31]. BnaABF2 from Brassica napus confers enhanced drought and salt tolerance in transgenic Arabidopsis [32]. BolOST1 was dramatically induced by drought and high-salt stress, and the ectopic expression of BolOST1 restored the drought-sensitive phenotype of ost1 [33]. However, natural response to salt and drought stresses in broccoli is still largely unknown. In our previous transcriptome data analysis, a TLP gene named bolTLP1 was identified in broccoli which exhibited a significant positive response to salt stress. However, the role of bolTLP1 is unknown. In the present study, the full-length cDNA of bolTLP1 was cloned, and its expression patterns under salt stress and drought stress were analyzed. Transgenic Arabidopsis and broccoli overexpressing bolTLP1 were generated. The phenotypes of these bolTLP1 overexpressing transgenic plants under salt and drought stresses were observed. The genes and regulatory processes involved in bolTLP1-mediated tolerance to abiotic stress were explored.

BolTLP1 Was Identified in Broccoli
In previous comparative transcriptome sequence analysis to identify genes involved in salt stress response in broccoli, bolTLP1 was found to be rapidly induced under salt stress condition ( Figure 1A). In the present study, the full-length cDNA of bolTLP1 was identified based on the ESTs of this gene from broccoli transcriptome data. Sequence analysis indicated that the coding region of bolTLP1comprises 696 bp and encodes a predicted 231-amino acid protein, which contained a conserved TLP domain. The phylogenetic analysis indicated that bolTLP1 and 1 out of the 29 TLP genes identified in Brassica oleracea var. oleracea (XP_013621285), are classified into a separate group, and they appear to show far genetic distance with other TLP genes ( Figure 1B).

BolTLP1 Positively Responded to Salt and Drought Stresses in Broccoli
The expression pattern of bolTLP1 under salt treatment in broccoli was further confirmed by qRT-PCR. Consistent with the expression trend identified by comparative transcriptome sequence analysis, the expression level of bolTLP1 was increased significantly in 4 h after exposure to salt stress ( Figure 1C). In addition, the expression pattern of bolTLP1 in the leaves of broccoli treated with 300 mM mannitol, which could mimic drought condition, was explored. Similar to observations under salt stress, the expression level of bolTLP1 was significantly higher under drought stress than that observed under control condition (0 h) ( Figure 1D).

Overexpression of bolTLP1 Increased Resistance to Salt and Drought Stresses in Arabidopsis
To further uncover the roles of bolTLP1, a transformation vector containing the bolTLP1 under the control of the enhanced CaMV 35S promoter was constructed, and used to transform Arabidopsis. A total of 26 independent 35S::bolTLP1 transgenic lines were obtained ( Figure S1). Under normal growth conditions, there were no phenotypic differences between the 35S::bolTLP1 transgenic lines and the vector controls, thereby indicating that the overexpression of bolTLP1 did not affect the growth and development of the transgenic lines in Arabidopsis (Figure 2A). To determine if overexpression of bolTLP1 could improve resistance to salt and/or drought stress, 35S::bolTLP1 Arabidopsis lines were planted in soil, and the 25-day-old seedlings were subjected to 125 and 200 mM NaCl treatments, respectively. At 12 days after 200 mM NaCl treatment, the aerial organs of the vector controls were wilted, growth was retarded, and leaves were yellow and senescent. Compared with the vector controls, the 35S::bolTLP1 Arabidopsis was only slightly affected by the high salt condition ( Figure 2B). At 18 days after salt stress, the growth of the remaining vector control plants was substantially inhibited. The survival rate of 35S::bolTLP1 Arabidopsis was significantly higher than that of the vector controls, and the aerial organs of the 35S::bolTLP1 Arabidopsis still showed normal growth. (Figure 2C,G). Under 125 mM NaCltreatment, although over 45% of the vector control plants died, almost all individual plants of the 35S::bolTLP1 Arabidopsis survived ( Figure 2G). At both low (125 mM) and high (200 mM) NaCl treatments, the primary and lateral roots of the 35S::bolTLP1 Arabidopsis were stronger than those of the vector controls ( Figure 2D-F). The 35S::bolTLP1 Arabidopsis plants were treated with water deficit. At 15 days after the water deficit, the leaves of the vector controls were seriously withered, and plant growth was inhibited. The growth and development of the 35S::bolTLP1 Arabidopsis were much better than those of the vector controls. In brief, the aerial organs of the 35S::bolTLP1 Arabidopsis were stronger than those of the vector controls, and their leaves were normal, although some of them also became senescent. The survival rate of 35S::bolTLP1 Arabidopsis was over 75%, whereas only 10% of the vector control plants survived after rewatering ( Figure 3A-C,G). To further confirm the roles of bolTLP1 under drought stress, the 35S::bolTLP1 Arabidopsis plants were treated with different concentrations of mannitol. At 200 mM mannitol, the aerial organs of the 35S::bolTLP1 Arabidopsis were larger than those of the vector controls. Similarly, at 300 mM mannitol, the growth and development of the aerial organs of the 35S::bolTLP1 Arabidopsis were still stronger than those of the vector controls ( Figure 3D,E). Nevertheless, the roots, especially the primary roots, were not significantly affected by the mimic drought stresses ( Figure 3F).

Transgenic Broccoli Overexpressing bolTLP1 Exhibited High Salt and Drought Tolerance
BolTLP1 was also overexpressed in broccoli. A total of 16 independent 35S::bolTLP1 broccoli transgenic lines were obtained ( Figure S1). Similar to the 35S::bolTLP1 Arabidopsis, the growth and development of the 35S::bolTLP1 broccoli were unaffected under the normal growth conditions ( Figure 4A). The 30-day-old 35S::bolTLP1 broccoli seedlings were irrigated with 200 mM NaCl. At 13 days after the salt treatment, the growth of the 35S::bolTLP1 broccoli was normal. Under these conditions, the vector controls grew slower and the first leaves showed signs of senescence ( Figure 4B). As time went on, the 35S::bolTLP1 broccoli showed slight salt damage, but their growth vigor was much better than that of the vector controls ( Figure 4C-E). The growth and development of the 35S::bolTLP1 broccoli under drought stress were also examined. The phenotypic data demonstrated that at 12 days after the water deficit, the aerial organs of the vector controls became withered, and their first and second leaves were senescing. Interestingly, at this same stage of water deficit, the growth of the 35S::bolTLP1 broccoli was not considerably affected, and their aerial organs were larger and stronger than those of the vector controls. At 15 days after the water deficit, although the leaves of the 35S::bolTLP1 broccoli and vector controls both showed withering, the growth vigor of the vector controls was inhibited to a greater degree than the 35S::bolTLP1 broccoli. After rewatering, the majority of the 35S::bolTLP1 broccoli (over 78%) survived, whereas almost all vector controls died ( Figure 4F-J). (Student's t-test, ** p < 0.01; data are means ± SD (n ≥ 15)). L3, L5 and L9 indicated the three independent 35S::bolTLP1 broccoli lines. CK indicated the vector control plants.

DEGs Were Confirmed between the 35S::bolTLP1 Broccoli and Vector Control
Comparative transcriptome analysis was conducted to identify the DEGs between the 35S::bolTLP1 broccoli and the vector controls. In total, 34,258 genes showing transcriptional expression were detected, among which 3284 genes showed significantly differential expression levels (corrected p-value < 0.01). The expression levels of 2202 out of the 3284 DEGs were up-regulated in the 35S::bolTLP1 broccoli, and the other approximately 33% of the DEGs displayed lower expression levels in the 35S::bolTLP1 broccoli compared to the vector control plants (Table S1). GO functional annotations indicated that these DEGs were mapped to 933 GO terms in the biological process, 154 GO terms in the cellular component and 343 GO terms in the molecular function, among which 33, 1 and 1 GO terms were significantly enriched in biological process, molecular function and cellular component, respectively (corrected p-value < 0.01) ( Figure 5, Table S2). The significantly enriched GO terms in the biological process were mainly involved in stress responses, glycosinolate and sulfur compound synthetic and metabolic process, and peptidase, proteolysis and hydrolase activity. The GO terms involved in oxidoreductase activity and chromatin organization were significantly enriched in the molecular function and the cellular components, respectively ( Figure 5, Table S2).  Table S2.

Expression Profiles of Genes Involved in ABA, Ethylene and Auxin-Mediated Signaling Pathways Were Significantly Altered by Overexpression of bolTLP1
According to the functional annotations of all DEGs by GO analysis, genes involved in the biotic and abiotic stress response processes were obviously predominant in the DEGs. At least 685 out of the 3284 DEGs (>21%) were functionally associated with stress response processes. Among the stress response-associated DEGs, approximately 70% of them displayed higher expression levels in the 35S::bolTLP1 broccoli than those in the vector controls (Table S3). A large proportion of these DEGs were key regulators in phytohormone-mediated signaling pathways. In brief, twenty-six DEGs were involved in the auxin signaling pathway. Twenty-four out of the 26 DEGs displayed higher expression levels in the 35S::bolTLP1 broccoli than those in the vector controls. Thirty-one genes involved in the ABA signaling pathway displayed differential expression patterns between the 35S::bolTLP1 broccoli and the vector controls, among which 14 genes, such as RD22, HVA22, RCI2A and ABSCISIC ACID-INSENSITIVE 5 (ABI 5), displayed significantly upregulated expression patterns in the 35S::bolTLP1 broccoli. At least 11 genes involved in the ethylene signaling pathway displayed differential expression patterns between the 35S::bolTLP1 broccoli and the vector controls ( Figure 6A-C). The expression levels of some of these DEGs were further verified by qRT-PCR (Figure 7).

Genes Involved in Sulfur Compound Synthesis and Hydrolase/Oxidoreductase Activity Were Significantly Inductively Expressed by the Overexpression of bolTLP1
Besides the DEGs involved in stress response, at least 62 DEGs were confirmed to map the GO terms associated with the glycosinolate and sulfur compound synthesis and metabolism, which displayed significant enrichment in the biological process. Forty-seven out of the 62 DEGs, such as genes encoding cytochrome P450 81F1, epithiospecifier protein, S-alkyl-thiohydroximate lyase, adenylyl-sulfate kinase, nitrile-specifier protein, flavincontaining monooxygenase, 5 -adenylylsulfate reductase and cytochrome P450 83B1, had significantly higher expression levels in the 35S::bolTLP1 broccoli compared to the vector control plants. The majority of these genes play crucial roles in sulfur compound synthesis and metabolism ( Figure 6D, Table S3). The expression patterns of genes related to significantly enriched GO terms involved in the activity of diverse enzymes, such as the proteolysis, peptidase, hydrolase and oxidoreductase, were also analyzed. A total of 12 genes including genes encoding kunitz-type serine protease inhibitor, cysteine proteinase inhibitor, trypsin inhibitor and 4-hydroxy-4-methyl-2-oxoglutarate aldolase were associated with the activity of proteolysis and hydrolase, all of which exhibited significantly increased expression levels in the 35S::bolTLP1 broccoli ( Figure 6E). At least 192 DEGs were confirmed to be associated with the oxidoreductase activity. The majority of them (146 out of 192) also exhibited higher expression levels in the 35S::bolTLP1 broccoli than those in the vector controls (Table S3). Moreover, 15 genes associated with the significantly enriched GO terms involved in chromatin organization, especially histone variants, were identified. All these genes showed increased expression levels in the 35S::bolTLP1 broccoli ( Figure 6F, Table S3).

Five Proteins Involved in Abiotic Stress Responses Were Confirmed to Interact with bolTLP1
Yeast two-hybrid screening was conducted to identify proteins that could interact with bolTLP1. In total, 266 positive clones were identified and sequenced. These sequences were annotated as 20 diverse genes, among which RD2, RD22, VOZ2, LSM1B and MDH were functionally associated with stress responses in plants. The interaction of these five proteins with bolTLP1 was further confirmed by using bolTLP1 as the prey and RD2, RD22, VOZ2, LSM1B and MDH as the baits in the yeast two-hybrid assay ( Figure 8A, Table S4). Quantitative expression analysis revealed that the expression levels of these stress response-associated genes were significantly higher in 35S::bolTLP1 broccoli than those in the vector controls ( Figure 8B).

Discussion
TLPs are present as a large group of PR-5 gene family in plants, and most TLPs were demonstrated to play a role in defending against pathogen [1,9,11,14,16]. Only a few TLPs have been demonstrated to function in response to abiotic stress. Nevertheless, in the present study, bolTLP1, a member of TLP subfamily, was first identified in broccoli. The expression profiles assay confirmed that bolTLP1 displayed inducible expression patterns under salt and drought stresses. Overexpression of bolTLP1 in broccoli, as well as in Arabidopsis, significantly increased the salt and drought tolerance of transgenic plants. These findings demonstrated that bolTLP1 is an important positive response factor that plays a role in combatting salt and drought stress in broccoli, and suggested the potential applications of bolTLP1 in breeding crops with high salt and drought tolerance via genetic engineering.
The molecular mechanisms by which the TLPs respond to abiotic stress remain largely unknown, although investigations have confirmed the roles of several TLP genes in cold temperature, salt and drought stresses [13,[17][18][19]. In the present study, comparative transcriptome analysis revealed that the DEGs detected between the 35S::bolTLP1 broccoli and the vector controls were mainly involved in the regulatory processes including "stress responses", "glycosinolate and sulfur compound metabolic process", "peptidase, proteolysis and hydrolase activity", "oxidoreductase activity" and "chromatin organization" ( Figure 5, Table S2). These results suggested that these regulatory processes should participate in bolTLP1-mediated tolerance to salt and drought stresses. Furthermore, the functional annotation analysis of the DEGs enriched in these regulatory processes confirmed that over 21% DEGs were involved in stress response processes, which represented the largest gene group in the DEGs. Remarkably, a large proportion of these DEGs were associated with phytohormone-mediated signaling pathways. For example, RD2, RD22, ABI5, WRKY15/17/20/33/40/46/53/62/70, abscisic acid 8 -hydroxylase 2, annexin, PUB23, HVA22, RCI2A, RCI2B and PYL4/5/8/9 are the important members of ABA-mediated signaling pathway, which all displayed significantly differential expression levels between the 35S::bolTLP1 broccoli and the vector controls ( Figure 6B). Most of these genes were confirmed to play crucial roles in response to abiotic stresses in diverse plant species [34][35][36][37][38]. Interestingly, RD2 and RD22 were confirmed to directly interact with bolTLP1 ( Figure 8). RD2 and RD22 are up-regulated by drought stress, salinity stress and exogenously supplied ABA. Their inducible expression has been used as a marker for abiotic stress [39,40]. In addition, at least 26 DEGs, such as auxin response factor (ARF) 4/5/8/11/14/18/19, auxin transporter 2, IAA3/9/12/19/26/29, inositol-3-phosphate synthase, indole-3-aceticacidamido synthetase GH3, calmodulin-binding transcription activator 1 (CAMTA 1) and auxininduced gene X15 and SAUR32 are involved in auxin-mediated regulatory processes ( Figure 6A). Similarly, some of these genes such as CAMTA 1 [41,42] and inositol-3-phosphate synthase [43,44] were identified to function in response to drought or salt stress. On the other hand, a few DEGs, such as Ethylene receptor 2 and ethylene responsive factors (ERFs) including ERF04, ERF05, ERF09, ERF12, RAP2-3, RAP2-10, ERF11, ERF104, TEM1 and TOE3, are involved in ethylene signaling pathway ( Figure 6C). Inducible expression of RAP2-3, as well as RAP2-12 and RAP2-2, confers tolerance to anoxia, oxidative and osmotic stresses [45]. The ERF6 overexpressing transgenic lines are hypersensitive to osmotic stress, while the growth of erf5erf6 loss-of-function mutants is less affected by stress [46]. These results demonstrated that the overexpression of bolTLP1 significantly affects the expression levels of a set of genes involved in the ABA, ethylene and auxin signaling pathways. Moreover. The homologues of a large proportion of these genes have been demonstrated to function in defending against abiotic stress in diverse plants. Our results also support the role of these genes in the bolTLP1-mediated tolerance to salt and drought stresses. This finding provided new insight into the roles of phytohormone-associated genes in response to abiotic stress in broccoli. Moreover, beside RD2 and RD22, three other proteins VOZ2, LSM1B and MDH were confirmed to directly interact with bolTLP1, and their genes displayed differently expressed pattens between the 35S::bolTLP1 broccoli and vector controls (Figure 8). The investigations demonstrated that VOZ transcription factors act as positive regulators of salt tolerance in Arabidopsis [47]. LSM1-7 are the components of the 5 -3 mRNA decay pathway, which play a crucial role in regulating ABA signaling and modulating ABA-dependent expression of stress related transcription factors from the AP2/ERF/DREB family [48]. Overexpression of plastidic maize NADP-malate dehydrogenase (ZmNADP-MDH) increases tolerance to salt stress in Arabidopsis [49]. It suggested that these three abotic-stress associated genes could also play important roles in bolTLP1-mediated regulatory network.
Another remarkably overrepresented group of DEGs was associated with various enzymes, particularly proteolysis, hydrolase and oxidoreductase. The DEGs involved in proteolysis and hydrolase activity mainly function as lectin and kunitz-type protease inhibitors such as serine protease inhibitor, cysteine proteinase inhibitor and trypsin inhibitor ( Figure 6E). Plant lectins and protease inhibitors constitute a class of proteins which play a crucial role in plant defense against insect and pathogen attacks [50,51]. Previous investigations confirmed that protease inhibitors also function in abiotic stress. The constitutive expression of a trypsin protease inhibitor confers tolerance to multiple stresses in transgenic tobacco [52]. Oryzacystatin-I (OCI) is a rice cysteine proteinase inhibitor. Ectopic overexpression of OCI can enhance drought stress tolerance in soybean and Arabidopsis [53]. AtCYSa and AtCYSb are two cysteine proteinase inhibitors from Arabidopsis. Overexpression of these inhibitors in transgenic yeast and Arabidopsis increases their resistance to high salt, drought, oxidative and cold stresses [54]. The present study confirmed that all these DEGs involved in proteolysis and hydrolase activity displayed significantly up-regulated expression in 35S::bolTLP1 broccoli ( Figure 6E). These findings indicated that the increased expression levels of these genes could also play important roles in promoting salt and drought tolerance of the bolTLP1 transgenic plants.
A series of histone variants including H1.1, H2A.1, H2A.2, H2A.5, H2B.3, H2A.7, H2B.8 and DDM1 were also significantly overrepresented in the DEGs ( Figure 6F). These genes are important epigenetic regulators, and display significantly higher expression levels in the 35S::bolTLP1 broccoli than those in the vector controls ( Figure 6F). TaH2A.7 in wheat can enhance drought tolerance and promote stomatal closure when overexpressed in Arabidopsis [55]. In Arabidopsis, DDM1 is an epigenetic link between salicylic acid metabolism and heterosis, in which salicylic acid can protect plants from pathogens and abiotic stress [56]. Consistently, the mutants of DDM1 show higher sensitivity to NaCl stress than the wild type plants [57]. In addition, the data confirmed that over 60 genes involved in glycosinolate and sulfur compound synthetic and metabolic processes showed significant differential expression patterns in the bolTLP1 overexpressing transgenic plants and the vector controls. Genes associated with sulfur compound synthesis were up-regulated expression in 35S::bolTLP1 broccoli ( Figure 6D). Sulfur compound such as glucosinolates are best known for their roles in plant defense against herbivores and pathogens as well as their cancer-preventive properties [58][59][60]. A few investigations confirmed that glucosinolates also contributed to abiotic stress [61][62][63][64]. These results suggested that the histone variants and these sulfur compound-associated genes may also play an important role in bolTLP1mediated tolerance to abiotic stress. , and three such replicated treatments were conducted. Similarly, a total of 48 individual plants from three independent 35S::bolTLP1 broccoli lines were divided into three groups, and unirrigated to evaluate their drought tolerance. In addition, at least three 35S::bolTLP1 Arabidopsis lines in 1/2 MS medium or in soil were treated with 125 and 200 mM NaCl to mimic salt stress. Three 35S::bolTLP1 Arabidopsis lines in 1/2 MS medium were treated with 200 and 300 mM mannitol to mimic drought stress. The 35S::bolTLP1 Arabidopsis planted in soil were also unirrigated to further evaluate their drought tolerance. To validate the reliability of the assessments of salt and drought tolerance in 35S::bolTLP1 Arabidopsis, three replicates were performed for each treatment.

BolTLP1 Cloning and Phylogenetic Analysis
The ESTs of bolTLP1 were identified from the broccoli transcriptome data, and the specific primers (bolTLP1-F/bolTLP1-R) were designed to amplify the full coding sequences of bolTLP1 based on these expressed sequence tags (ESTs) ( Table S5). Total RNAs were isolated from the 25-day-old broccoli seedlings by using TRIzol reagent (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's instructions. The first-strand cDNA was then synthesized using M-MLV reverse transcriptase (Promega, Madison, WI, USA), and the coding sequences of bolTLP1 were amplified by PCR. Because the genome data of broccoli is still limited, to elucidate the sequence homology of bolTLP1 with other TLP genes, those genes containing a TLP domain were identified from the genome database of Brassica oleracea var. oleracea, which is genetically closely related with broccoli (Table S6). The deduced amino acid sequences of bolTLP1 and its homologous genes from B. oleracea var. oleracea were used to construct the phylogenetic tree using the neighbor-joining method by the MEGA 6.0 program with the following parameters: bootstrap value of 1000, poisson correction and pairwise deletion [65].

BolTLP1 Expression Profile Analysis
Total RNAs from the leaves of individual plants, which were harvested at different time points after imposing salt or drought stress, were isolated and reverse transcribed to cDNAs. Specific primers of bolTLP1 (qbolTLP1-F/qbolTLP1-R) was designed by Primer premier 5.0 software (Table S5). The bolActin gene from broccoli was selected as the internal control and Faststart Universal SYBR Green Master (Roche, Basel, Switzerland) was used in quantitative real-time RT-PCR (qRT-PCR) assay. The relative expression levels of bolTLP1 under salt or drought stress were calculated by the comparative 2 −∆∆CT method. Three biological replicates and three technological replicates were performed to ensure the reliability of quantitative analysis.

Expression Vector Construction and Genetic Transformation
The full coding sequence of bolTLP1 with incorporated Nco I and BstE II restriction sites was amplified (Table S5) and cloned into the pEASY-T1 vector. The insert released from the pEASY-T1 vector with Nco I/BstE II double digestion was sub-cloned into the pCAMBIA3301 binary vector. The recombinant 35S::bolTLP1 plasmids and vectors without exogenous gene insertion (empty vectors) were transformed into A. tumefaciens strain LBA4404. Agrobacterium-mediated Arabidopsis genetic transformation was performed via the floral dip method. The transgenic Arabidopsis lines were screened by spraying 1:10,000 dilute Basta solution and further identified by PCR using specific primers (bolTLP1-F/bolTLP1-R) and combined primers (35S-F/bolTLP1-R). Simultaneously, the plants with empty vector were identified by PCR using specific primer pair (35S-F/NOS-R). The transgenic Arabidopsis plants were further verified by qRT-PCR using unique primers (qbolTLP1-F/qbolTLP1-R) and atActin as an internal control (Table S5). The 35S::bolTLP1 plasmids and empty vectors were also transformed into broccoli using Agrobacteriummediated method. In brief, broccoli seeds were rinsed with 75% ethanol for 2 min; 2% NaClO (Tianjin Fengchuan Chemical Reagent Co., Ltd., Tianjin, China) was then further used for surface sterilization of the seeds for 10 min. The sterilized seeds were planted on Murashige and Skoog (MS) medium with a 16 h/8 h light/dark cycle at 22 • C. The hypocotyls of 7-day-old broccoli seedlings were cut to 0.5 cm and pre-cultured on MS medium containing 1.5 mg/L 6-BA and 0.15 mg/L NAA for 2 days. Then, the truncated hypocotyls were dipped into a suspension of Agrobacterium containing the 35S::bolTLP1 expression vector or an empty vector for 1 min. The infected explants were transferred onto co-cultivation medium with 1 mg/L 6-BA, 0.1 mg/L NAA and 100 µmol/L acetosyringone, and cultured for 2 days in the dark. Subsequently, the explants were washed with sterile water and transferred onto MS medium containing 1.6 mg/L 6-BA, 0.2 mg/L NAA and 200 mg/L cefotaxime for 10 days for callus induction. The differentiated explants were transferred into screening medium supplemented with 3 mg/L Basta for 2 weeks. Finally, buds that continued to differentiate were transferred to the 1/2 MS medium containing 1 mg/L IBA for root initiation. The transgenic broccoli plants were further verified by PCR using prime pair (35S-F/bolTLP1-R) and qRT-PCR using unique primers (qbolTLP1-F/qbolTLP1-R) (Table S5).

Transcriptome Sequencing and Data Analysis
Leaves from six 30-day-old individual plants of per independent 35S::bolTLP1 broccoli line were equally mixed and used to perform three batches of independent RNA isolation. Equal amounts of RNA from each independent 35S::bolTLP1 broccoli line were then used to construct sequencing library. In total, two such sequencing libraries from two independent 35S::bolTLP1 broccoli lines and one sequencing library from the vector controls were constructed. The sequencing reaction was conducted by the Illumina HiSeq TM 2500 sequencing platform (Beijing Genomics Institute, Shenzhen, China) with three technolog-ical replicates. The clean reads were annotated and mapped to the reference genome of B. oleracea var. oleracea. The expression levels of genes were calculated by the fragments per kilobase of transcript sequence per million base pairs sequenced (FPKM). The significantly expression levels of genes between the 35S::bolTLP1 broccoli and the vector controls were identified based on the thresholds: |log 2 (fold-change (35S::bolTLP1 broccoli/vector control))| > 1 and corrected p-value < 0.05. Gene Ontology (GO) analysis of the differentially expressed genes (DEGs) was performed by the agriGO platform. Available online: http://bioinfo.cau.edu.cn/agriGO/ (accessed on 4 May 2018).

Differentially Expressed Gene Identification by qRT-PCR
The gene expression profiles detected by comparative transcriptome analysis were verified by qRT-PCR by using the specific primer pairs (Table S5). Three independent 35S::bolTLP1 broccoli lines (Line 3, Line 5 and Line 9) were used to perform three biological replicates. Similar to RNA isolation in transcriptome sequencing, RNAs from leaves of six randomly selected 30-day-old plants per 35S::bolTLP1 broccoli line were isolated and reverse transcribed to cDNAs. The comparative 2 −∆∆CT method was conducted to calculate the relative expression levels of genes using bolActin as an internal control. To further ensure the reliability of qRT-PCR, three technological replicates were carried out.

Yeast Two-Hybrid Screening and Assays
The proteins which interact with bolTLP1 were screened using Matchmaker TM Gold Yeast Two-Hybrid System (Clontech, Mountain View, CA, USA) according to the manufacturer's instructions. In brief, the full-length coding sequences of bolTLP1 with the Nco I and BamH I restriction enzyme sites were constructed into the bait vector pGBKT7 (Table S5). The recombinant bait vector pGBKT7-bolTLP1 was transformed into the Y2H Gold yeast train, and the autoactivation and toxicity of bolTLP1 were detected. The positive pGBKT7-bolTLP1 yeast then was mixed and mated with the universal Arabidopsis Mate & Plate library (Clontech, Mountain View, CA, USA). The positive mating yeast was screened on SD/-Trp-Leu-His-Ade medium with 0.5 mM 3-Amino-1, 2, 4-triazole (3-AT), which could effectively inhibit the autoactivation of bolTLP1. The plasmids from the mating yeast were isolated and used as a template to amplify the candidate genes using the universal primers (T7-F/3 AD-R) (Table S5). To further confirm the relationship of bolTLP1 and its candidate interacting proteins, the broccoli homologs of the positive Arabidopsis genes were cloned and inserted into the bait vector pGBKT7. In this case, bolTLP1 was inserted into the prey vector pGADT7 (Table S5). Subsequently, the yeast two-hybrid experiments were conducted according to the manufacturer's instructions. Transcript expression levels of the broccoli genes encoding proteins interacting with bolTLP1 were analyzed by qRT-PCR as mentioned above (Table S5).

Conclusions
Overexpression of bolTLP1 significantly increased the salt and drought tolerance in both Arabidopsis and broccoli. BolTLP1 may directly interact with stress responseassociated proteins RD2, RD22, VOZ2, LSM1B and MDH to regulate a series of genes involved in phytohormone (ABA, ethylene and auxin)-mediated signaling pathways, hydrolase/oxidoreductase activity, sulfur compound synthesis, and histone variants, which could play important roles in bolTLP1-mediated tolerance to salt and drought stresses in broccoli ( Figure 9). BolTLP1 is a potential candidate gene in breeding crops with high tolerance to abiotic stress via genetic engineering.