R2R3 MYB-dependent auxin signalling regulates trichome formation, and increased trichome density confers spider mite tolerance on tomato.

Unicellular and multicellular tomato trichomes function as mechanical and chemical barriers against herbivores. Auxin treatment increased the formation of II, V and VI type trichomes in tomato leaves. The auxin response factor gene SlARF4, which was highly expressed in II, V and VI type trichomes, positively regulated the auxin-induced formation of II, V and VI type trichomes in the tomato leaves. SlARF4 overexpression plants with high densities of these trichomes exhibited tolerance to spider mites. Two R2R3 MYB genes, SlTHM1 and SlMYB52, were directly targeted and inhibited by SlARF4. SlTHM1 was specifically expressed in II and VI type trichomes and negatively regulated the auxin-induced formation of II and VI type trichomes in the tomato leaves. SlTHM1 down-regulation plants with high densities of II and VI type trichomes also showed tolerance to spider mites. SlMYB52 was specifically expressed in V type trichomes and negatively regulated the auxin-induced formation of V type trichome in the tomato leaves. The regulation of SlARF4 on the formation of II, V and VI type trichomes depended on SlTHM1 and SlMYB52, which directly targeted cyclin gene SlCycB2 and increased its expression. In conclusion, our data indicates that the R2R3 MYB-dependent auxin signalling pathway regulates the formation of II, V and VI type trichomes in tomato leaves. Our study provides an effective method for improving the tolerance of tomato to spider mites.

Introduction

Pests have threatened crop yield since the commencement of plant domestication. At present, pest invasion ruins approximately 13% of crop production globally (Hamza et al., 2018). Plants tactically developed response mechanisms against pest attacks. They developed mechanical and chemical barriers in the form of trichomes. Differentiated epidermal cells from leaves, stems and floral organs produce plant trichomes, which can be glandular or nonglandular, unicellular or multicellular, and branched or unbranched (Yang and Ye, 2013).

Trichomes are nonglandular, unicellular and branched in Arabidopsis. Trichome formation passes through three stages: cellular determination, specification and morphogenesis of epidermal cell (Yang and Ye, 2013). A trimeric transcription factor (TF) complex formed by R2R3 MYB, WD40 repeat and basic helix–loop–helix (bHLH) proteins positively regulates trichome formation (Ishida et al., 2008). The WD40–bHLH–MYB complex directly induces the expression of GLABRA2 and cell cycle‐related SIAMESE (Morohashi and Grotewold, 2009). R3 MYBs compete with R2R3 MYBs and form a repressor complex that inhibits trichome formation (Wang et al., 2007).

Tomato has several types of trichomes: II, III, V and VIII type, which are nonglandular trichomes, and I, IV, VI and VII type, which are multicellular glandular trichomes (Glas et al., 2012). Glandular trichomes produce volatile organic compounds (VOCs) to repel or eradicate pests (Turlings et al., 1995). Tomato’s jasmonic acid (JA) signal transduction COI1 regulates VI type glandular trichome formation (Li et al., 2004). Tomato Woolly (Wo), a homeodomain‐leucine zipper (HD‐Zip) TF, interacts with B‐type cyclin SlCycB2 protein, which are essential for the formation of type I trichomes in tomato (Yang et al., 2011; Yang and Ye, 2013). The overexpression of Wo in potato and tobacco can promote the production of multicellular trichomes and increase tolerance to aphids (Yang et al., 2015). A single cysteine2–histidine2 (C2H2) zinc‐finger TF Hair (H) gene, which interacts with Wo, regulates the formation of type I multicellular trichomes (Chang et al., 2018). SlMYC1, a basic bHLH TF, plays important roles in the formation of VI type glandular trichomes and terpene biosynthesis in tomato glandular cells (Xu et al., 2018). The overexpression of the MYB gene GLABROUS1 involved in unicellular trichome formation in Arabidopsis does not affect multicellular trichome formation in Nicotiana, a Solanaceae species (Payne et al., 1999). The formation of multicellular trichomes in Solanaceae may be controlled by a pathway that is different from that of unicellular trichomes in Arabidopsis (Payne et al., 1999). However, the regulatory mechanism involved in nonglandular and glandular trichome formation in Solanaceae is rarely studied in tomato.

Auxin plays important roles in several physiological processes in plants. Short‐lived auxin/indole acetic acid (Aux/IAA) proteins and auxin response factors (ARFs) are involved in auxin‐dependent transcriptional regulation. At low auxin levels, Aux/IAA proteins interact with ARFs to repress ARF transcription activity by recruiting the co‐repressor TOPLESS (Szemenyei et al., 2008). At high auxin levels, Aux/IAAs bind to the SCFTIR1/AFB complexes degraded by the 26S proteasomes and release ARFs, which regulate the transcription of auxin response genes (Wang and Estelle, 2014). ARFs act as the transcriptional repressors or activators of auxin‐responsive genes (Guilfoyle and Hagen, 2012; Ren et al., 2011). Most ARF proteins have an N‐terminal DNA‐binding domain (B3) for the regulation of the transcription of auxin response genes, a middle region as an activation or repression domain, and a C‐terminal dimerization domain (Aux/IAA) for the formation of homo‐ or heterodimers (Krogan and Berleth, 2012; Zouine et al., 2014). ARFs play important roles in various plant developmental processes (Ckurshumova et al., 2014). SlARF4 is a repressor of auxin response and regulates the accumulation of chlorophyll and starch in tomato fruits (Sagar et al., 2013).

The tomato Aux/IAA family gene SlIAA15 is involved in the formation of glandular and nonglandular trichomes (Deng et al., 2012). However, the mechanism of auxin‐mediated transcriptional regulation of trichome formation in tomato still needs to be elucidated. In this study, auxin treatment increased the densities of II, V and VI type trichomes in tomato. SlARF4 positively regulated the auxin‐induced formation of II, V and VI type trichomes, and the regulation was dependent on R2R3 MYB TFs SlTHM1 and SlMYB52. Transgenic plants with high densities of trichomes because of the overexpression of SlARF4 or down‐regulation of SlTHM1 increased trichome density and conferred spider mite tolerance. Overall, our results showed that an auxin‐mediated regulatory pathway regulated formation of nonglandular and glandular trichomes and provided an effective method for improving the tolerance of tomato plants to spider mites.

Results

Auxin treatment induced the formation of II, V and VI type trichomes in leaves

Different type of trichomes was first identified by scanning electron microscopy (SEM) on the adaxial leaf surfaces of ‘Micro‐Tom’ tomato cultivar. It is clear that II, V and VI type trichomes were abundant and easily recognizable (Glas et al., 2012). Therefore, II, V and VI type trichomes on the adaxial leaf surfaces were studied. The influence of auxin on tomato trichome formation was investigated by treating tomato seedlings with various concentrations of IAA (Indole acetic acid). The densities of II, V and VI type trichomes increased with IAA concentration (from 1 mg/L to 30 mg/L), and the most obvious effects on trichome densities was 30 mg/L IAA (Figure 1a). Thus, this concentration was used in subsequent analysis. Given that trichomes are initiated from the epidermal pavement cells of leaves, the effects of auxin treatment on the densities of epidermal pavement cells were analysed. IAA treatment had no effect on the density of epidermal pavement cells (Figure S1). Our result indicated that auxin induced the formation of type II, V and VI trichomes in the tomato leaves.

Figure 1

Trichome density with auxin treatment, expression analysis of SlARF4 and trichome density in SlARF4 overexpression and RNAi plants. a, Densities of II, V and VI type trichomes of tomato leaves with IAA treatment. b, d and f, Expression pattern of SlARF4 in trichome tissues was explored by GUS reporter gene driven by the SlARF4 promoter. Scale bars = 0.2 mm. b, II type trichome of pSlARF4‐GUS lines. d, V type trichome of pSlARF4‐GUS lines. f, V type trichome of pSlARF4‐GUS lines. c, e and g, Gus staining of trichome tissues of WT. Scale bars = 0.2 mm. c, type II trichome of WT. e, V type trichome of WT. g, VI type trichome of WT. h, qRT‐PCR analysis of SlARF4 expression in overexpression and down‐regulation lines. Each value indicates the mean ± standard errors (SE) of four biological replicates. i, Scanning electron micrographs of the leaf surface. Scale bars = 60 mm. j, Density analysis of II, V and VI type trichomes in the leaves of WT and SlARF4 lines without IAA treatment. k, Number of trichomes per epidermal cell of the tomato leaves of WT and SlARF4 lines. l, Density analysis of II, V and VI type trichomes in the leaves of WT and SlARF4 lines with IAA treatment. m, Trichome density ratio of IAA treatment to no‐IAA treatment. WT, wild‐type plants; OE‐28 and ‐31, SlARF4 overexpression plants; RNAi‐8 and ‐18, SlARF4 RNAi plants. The number of II type trichomes in an area of 0.5 cm2 and the number of V and VI type trichomes in an area of 2.2 mm2 were calculated under a light microscope. Each value indicates the mean ± SE of three biological replicates. * and ** indicate significant difference between WT and transgenic leaves with P < 0.05 and P < 0.01, respectively, as determined by t‐test.

SlARF4 positively regulated the auxin‐induced formation of II, V and VI type trichomes in the leaves

qRT‐PCR was performed to assess the expression pattern of SlARF4 (Figure S2).SlARF4 is highly expressed in leaf trichomes. The expression pattern of SlARF4 in leaves was further analysed by GUS staining. A transcriptional fusion was generated between SlARF4 promoter and GUS gene (pSlARF4‐GUS), and GUS staining was detected in II, V and VI type trichomes (Figures 1b, d and f). No GUS staining was detected in the trichomes of WT plants (Figures 1c, e and g). This result indicated that SlARF4 was expressed in II, V and VI type trichomes in tomato leaves.

The role of SlARF4 in trichome formation was elucidated using transgenic technology. The expression level of SlARF4 decreased in the RNAi‐SlARF4 lines and increased in the OE‐SlARF4 lines (Figure 1h). Without IAA treatment, the OE‐SlARF4 lines displayed increases in the densities of II, V and VI type trichomes in the leaves, whereas the RNAi lines exhibited decreases in the densities of II, V and VI type trichomes compared with WT plants (Figures 1i and j). The OE‐SlARF4 lines displayed increase in the density of epidermal pavement cells in the leaves, whereas RNAi lines exhibited decreased density compared with WT plants (Figure S3). Then, the number of trichomes per epidermal cell in the OE‐SlARF4 and RNAi‐SlARF4 lines was calculated. An apparent increase in the number of trichomes per epidermal cell in the overexpressed lines and decrease in that of the RNAi lines were observed (Figure 1k), indicating that SlARF4 positively regulated the formation of II, V and VI type trichomes. IAA treatment was performed to study the response of SlARF4 transgenic lines to auxin. At 30 mg/L IAA concentration, the densities of II, V and VI type trichomes increased in the leaves of OE‐SlARF4, RNAi‐SlARF4 and WT plants (Figure 1l). The trichome density ratio between IAA treatment and no‐IAA treatment was calculated in studying the effect of auxin treatment on trichome formation. The ratio of OE‐SlARF4 plants was similar to that of the WT plants, but RNAi‐SlARF4 plants had lower ratio than the OE‐SlARF4 and WT plants (Figure 1m). These results indicated that the down‐regulation of SlARF4 decreases the effect of IAA treatment on the formation of II, V and VI type trichomes in tomato leaves and that the inducement of auxin on the formation of II, V and VI type trichomes is dependent on SlARF4. Given the fact that SlARF4 was expressed in tomato fruits, the fruit trichome density was analysed. There was no significant difference between WT and SlARF4 transgenic lines (Figure S4).

The function of SlARF4 in trichome formation was validated in Slarf4 CRISPR/Cas9 plants. The trichome density and the number of trichomes per epidermal cell were consistent with those in the RNAi‐SlARF4 plants (Figure S5), indicating SlARF4 positively regulates the formation of II, V and VI type trichomes in tomato leaves.

Overexpression of SlARF4 conferred spider mite tolerance to tomato

The response of SlARF4 transgenic plants to spider mite was analysed through spider mite bioassays. Pest preference experiment was used in assessing the relative preference of herbivores for WT, OE‐SlARF4 and RNAi‐SlARF4 plants. The number of spider mites that preferred the leaves of the OE‐SlARF4 plants was lower than that of the spider mites that preferred the WT plants (Figures 2a and b). The number of spider mites that preferred the leaves from the RNAi‐SlARF4 plants was higher than that of the spider mites that preferred the WT plants (Figures 2c and d). Spider mite inoculation assay was performed. The 10‐ and 25‐day trials with adult female mites resulted in the localized collapse of the leaves of the RNAi‐SlARF4 lines and in small chlorotic lesions, which are indicative of mite feeding, in the leaves of the OE‐SlARF4 lines (Figure 2e). After 45 days of assays, severe damages were observed in the RNAi‐SlARF4 and WT plants and some plants died. By contrast, the inoculated leaves of OE‐SlARF4 lines remained green and had relatively few signs of macroscopic damage (Figure 2e). The ability of the spider mites to colonize the host was examined by measuring the fecundity of the female mites. The egg number laid on the OE‐SlARF4 plants was lower than that on the WT plants, whereas the number of eggs laid on the RNAi‐SlARF4 plants was higher than that on the WT plants (Figure 2f). The overexpression of SlARF4 in tomato increased spider mite tolerance, and the down‐regulation of SlARF4 increased the sensitivity of the tomato plants to spider mites.

Figure 2

Tolerance of SlARF4 plants to two‐spotted spider mites and SlARF4 targets SlTHM1 and SlMYB52. a–d, Preference experiment to analyse the preference of spider mites for WT and SlARF4 plants. Ten adult female spider mites were positioned in an area that was equidistant from WT and SlARF4 leaflets. The number of mites that moved to different leaflets and those that failed to make a choice (nc) were counted 1 h after initiating the assay. The data represent the mean values ± SE based on 16 experimental repetitions (total of 160 mites). e, Inoculation of WT and SlARF4 plants with two‐spotted spider mites for 45 days. Fifteen adult female mites were transferred to a single leaf on 15‐day‐old WT and SlARF4 plants. f, Fecundity of two‐spotted spider mites in the leaves of WT and SlARF4 plants. Five adult female mites moved to the leaf discs (12 mm) of WT and SlARF4 plants. Eggs were counted via a microscope at 24‐h intervals for 4 days. WT, wild‐type plants; OE‐28 and ‐31, SlARF4 overexpression plants; RNAi‐8 and ‐18, SlARF4‐down‐regulated plants. The data represent the mean values and SE from 12 independent experiments. * and ** indicate significant difference between WT and SlARF4 leaves with P < 0.05 and P < 0.01, respectively (t test). g and h, EMSA showing the direct binding of SlARF4 to the promoters of SlTHM1 and SlMYB52. Biotin‐labelled DNA probes from original promoter or mutants were incubated with GST‐SlARF4 protein, and then, DNA–protein complexes were separated on 6% native polyacrylamide gels. + or +++ indicates increasing amounts of unlabelled probes for competition. i and j, ChIP‐qPCR assay for the direct binding of SlARF4 to the promoters of SlTHM1 and SlMYB52. Values represent the percentage of DNA fragments that co‐immunoprecipitated with anti‐SlARF4 antibodies or nonspecific antibodies (anti‐IgG) comparative to the input DNA. The data are means ± SE from qPCR of four biological replicates. k, Diagrams of the reporter and effector vectors in the dual‐luciferase reporter assay. l, SlARF4 suppresses the transcription of SlTHM1 and SlMYB52. The activities of firefly LUC and REN in tobacco leaves were measured and LUC/REN ratio was analysed after infiltration with A. tumefaciens carrying the reporter plasmid and different combinations of effector plasmids. Each value represents the mean of six biological replicates, and vertical bars represent the SE.

SlARF4 bound to the AuxRE and TGA motifs of the SlTHM1 and SlMYB52 promoters, respectively, and inhibited their expression

The differentially expressed genes (DEGs) between the WT and SlARF4 RNAi lines were identified through RNA sequencing (RNA‐Seq), using the leaves of 45‐day‐old tomato seedlings. A total of 398 DEGs, including 275 up‐regulated and 123 down‐regulated genes, were identified (Table S1). Gene Ontology (GO) function and pathway enrichment analyses showed that the down‐regulation of SlARF4 influenced several metabolic pathways (Table S2, Figure S6 and S7a). Forty‐nine DEGs belong to 16 TF families (Table S3). The dominant families included WRKY (nine DEGs), MYB (six DEGs), ERF (six DEGs), MIKC_MADS (five DEGs) and C2H2‐type zinc‐finger families (five DEGs; Figure S7b). MYBs play an important role in trichome formation (Machado et al, 2009; Wang et al, 2004). SlTHM1 (Solyc08g081500) and SlMYB52 (Solyc03g093890) contained the AuxRE (TGTCTC) motif and auxin response element (TGA motif, AACGAC), respectively, which are the binding motifs of ARF proteins.

Both SlTHM1 and SlMYB52 belong to R2R3MYB transcription factors family that plays regulatory roles in plant development and defence responses as one of the largest transcription factor families in tomato (Li et al, 2016; Zhao et al, 2014). SlTHM1 has an open reading frame (ORF) of 762 bp that encodes a protein containing 253 amino acid residues while SlMYB52 has an open reading frame (ORF) of 996 bp that encodes a protein containing 331 amino acid residues. Phylogenetic analysis of tomato R2R3MYB family proteins revealed that SlTHM1 along with SlMYB56, SlMYB67 and SlMYB26 belong to S15 subgroup; SlMYB52 along with SlMYB51, SlMYB95 and SlMYB105 belong to S6 subgroup (Zhao et al, 2014). Amino acid sequence analyses of SlTHM1 and SlMYB52 were conducted with known MYB proteins in the same subgroup, respectively (Figure S10). Both SlTHM1 and SlMYB52 proteins contain an R2R3 domain in the N‐terminus (Figure S10). In addition, phylogenetic trees of SlTHM1 and SlMYB52 with homologous proteins from other species were also generated, respectively (Figure S11). SlTHM1 is closely related to SbMYB21 while SlMYB52 is closely related to SpMYB15, indicating possible functional similarity among them.

The direct binding of SlARF4 protein to SlTHM1 and SlMYB52 genes was verified by electrophoretic mobility shift assay (EMSA). Purified recombinant truncated SlARF4 and glutathione S‐transferase (GST) fusion protein (GST‐tSlARF4) were successfully obtained (Figure S8). The GST‐tSlARF4 fusion protein bound to biotin‐labelled probes containing AuxRE and TGA motifs that were derived from SlTHM1 and SlMYB52 promoters, respectively, and caused a mobility shift that was effectively abolished in a dose‐dependent manner when unlabelled SlTHM1 and SlMYB52 promoter fragments were added as competitors (Figures 2g and h). Mobility shift was not observed when biotin‐labelled probes were incubated with GST only (Figures 2g and h). This finding indicated that the specific targets of SlARF4 were SlTHM1 and SlMYB52. Chromatin immunoprecipitation coupled with quantitative polymerase chain reaction (ChIP‐qPCR) was used in confirming the interaction between SlARF4 and the SlTHM1 and SlMYB52 genes in vivo. The promoter regions of SlTHM1 and SlMYB52 were specifically enriched when FLAG antibodies were used instead of nonspecific antibodies (IgG; Figures 2i and j). SlARF4 possessed transcriptional repression activity and directly targeted the promoters of SlTHM1 and SlMYB52. Hence, the transcriptional repression of SlTHM1 and SlMYB52 by SlARF4 were determined through transient dual‐luciferase assays. The overexpression of SlARF4 remarkably reduced luciferase activity driven by the promoters of SlTHM1 and SlMYB52 compared with the vector control (pEAQ; Figures 2k and l). qRT‐PCR showed that the expression levels of SlTHM1 and SlMYB52 increased in the RNAi‐SlARF4 plants but decreased in the OE‐SlARF4 plants (Figure S9). The results indicated that SlARF4 targeted SlTHM1 and SlMYB52 repressed their transcription.

SlTHM1 negatively affected the auxin‐induced formation of II and VI type trichomes in the leaves

qRT‐PCR was performed to assess the expression pattern of SlTHM1 (Figure S12).SlTHM1 was expressed in all tissue tested with the highest expression in leaf trichomes. GUS staining showed that SlTHM1 was specifically expressed in II and VI type trichomes but not in V type trichome (Figure 3a). SlTHM1‐green fluorescent protein (GFP) was transiently expressed in the leaf epidermal cells of tobacco for subcellular localization, and fluorescence signals were detected in the nucleus, indicating that the SlTHM1 was localized in the nucleus of plant cells (Figure 3b). The transcriptional activity of SlTHM1 in tobacco leaves was analysed using a dual‐luciferase reporter system. SlTHM1 increased the LUC/REN ratio compared with the empty vector pBD, revealing that SlTHM1 possesses transcriptional activation activity (Figure S13).

Figure 3

Expression analysis of SlTHM1, subcellular localization analysis of SlTHM1 protein, transcriptional activation and trichome density of the leaves of RNAi‐SlTHM1 plants. a, Expression pattern of SlTHM1 in trichome tissues explored by GUS reporter gene expression driven by the SlTHM1 promoter. Scale bars = 0.2 mm. b, Subcellular localization analysis of SlTHM1. Tobacco leaves were transiently transformed with empty GFP vector and SlTHM1‐GFP construct via A. tumefaciens transfection. A fluorescence microscope was used to observe GFP fluorescence. 35S:GFP was used as positive control. Scale bars = 15 μm. c, qRT‐PCR analysis of the expression of SlTHM1 gene in RNAi‐SlTHM1 plants. The data are the means ± SE from qPCR of four biological replicates. d, Scanning electron micrographs of the leaf surface. Scale bars = 60 mm. e, Density analysis of II, V and VI type trichomes from the leaves of WT and SlTHM1 RNAi plants without IAA treatment. Scale bars = 0.2. f, Number of trichomes per epidermal cell of the leaves of WT and RNAi‐ SlTHM1 tomato plants without IAA treatment. g, Density analysis of II, V and VI type trichomes in the leaves of WT and SlTHM1 lines with IAA treatment. h, Trichome density ratio of IAA treatment to no‐IAA treatment. WT, wild‐type plants; RNAi‐3 and RNAi‐19, SlTHM1‐down‐regulated plants. The number of II type trichomes in an area of 0.5 cm2 and the number of V and VI types trichomes in an area of 2.2 mm2 were calculated under a light microscope. All experiments were replicated three times. * and ** indicate significant difference between WT and transgenic leaves with P < 0.05 and P < 0.01, respectively, as determined by t‐test.

The SlTHM1 function on trichome formation was analysed in using RNAi strategy. The homozygous transgenic lines of RNAi‐SlTHM1 showed a substantially low degree of SlTHM1 transcript accumulation (Figure 3c). Without IAA treatment, RNAi‐SlTHM1 plants exhibited increases in the densities of II and VI type trichomes in leaves and no change in V type trichome compared with WT plants (Figures 3d and e). The density of epidermal pavement cells was analysed, and results showed that the down‐regulation of SlTHM1 increased the density of the epidermal pavement cells of the tomato leaves (Figure S14). An apparent increase in the number of II and VI type trichomes per epidermal cell in RNAi‐SlTHM1 lines was observed (Figure 3f). These results indicated that SlTHM1 negatively affected the formation of II and VI type trichomes. RNAi‐SlTHM1 plants with IAA treatment showed obvious increases in the densities of II, V and VI type trichomes in the tomato leaves (Figure 3g). The trichome density ratio between IAA treatment to no‐IAA treatment was calculated, showing that the density ratios of II and VI type trichomes on the RNAi‐SlTHM1 plants were higher than those of the WT plants and that the density ratio of V type trichome was similar to that of the WT plants (Figure 3h). The down‐regulation of SlTHM1 increased the effect of IAA treatment on the formation of II and VI type trichomes. Our results demonstrated that SlTHM1 regulated the auxin‐induced formation of II and VI type trichomes in tomato leaves.

Down‐regulation of SlTHM1 conferred spider mite tolerance to tomato

Spider mites were used in analysing the response of the RNAi‐SlTHM1 lines to herbivores. Spider mite preference experiment showed that the number of spider mites that preferred RNAi‐SlTHM1 leaves was lower than that of the mites that preferred the leaves of the WT plants (Figures 4a and b). Spider mite inoculation assay was carried out. The localized collapse of the WT leaves was obvious, but only small chlorotic lesions indicative of mite feeding was apparent on the RNAi‐SlTHM1 leaf tissues after 25 days (Figure 4c). After 45 days, spider mites caused severe damage and nearly killed the WT plants, but the RNAi‐SlTHM1 plants survived and retained some green leaves (Figure 4c). Furthermore, the number of eggs laid on the RNAi‐SlTHM1 leaves was remarkably lower than that on the WT leaves during the 4‐day assay (Figure 4d). The results indicated that the down‐regulation of SlTHM1 in tomato improved tolerance to spider mites.

Figure 4

Tolerance of SlTHM1 RNAi plants to two‐spotted spider mites. a and b, Spider mite preference experiment to measure the preference of spider mites for WT and SlTHM1 plants. The data represent the mean values and SE based on 16 experimental repetitions (total of 160 mites). c, Inoculation of WT and SlTHM1 RNAi plants with two‐spotted spider mites for 45 days. Photographs of the infested leaves and whole plants were obtained after treatment. d, Fecundity of two‐spotted spider mites on WT and SlTHM1 plants. The data are the mean values and SE from 12 independent experiments. WT, wild‐type plants; RNAi‐3 and RNAi‐19, RNAi‐SlTHM1 plants.

SlMYB52 negatively regulated the auxin‐induced formation of V type trichome in tomato leaves

qRT‐PCR and GUS staining were used to explore the expression pattern of SlMYB52 (Figures 5a, S15). GUS staining showed that SlMYB52 was specifically expressed in the V type trichome of leaves (Figure 5a). Subcellular localization assay of SlMYB52 showed that SlMYB52 was localized in the nucleus of plant cells (Figure 5b). RNAi‐SlMYB52 transgenic lines were generated to analyse the functions of SlMYB52 on trichome formation in tomato. The homozygous transgenic lines of RNAi‐SlMYB52 showed a substantially low degree of SlMYB52 transcript accumulation and obvious increase in the density of V type trichome (Figures 5c–e). The down‐regulation of SlMYB52 increased the density of epidermal pavement cells of tomato leaves (Figure S16). Then, the calculated number of trichomes per epidermal cell of RNAi‐SlMYB52 lines indicated an apparent increase in the number of V type trichomes per epidermal cell in the leaves (Figure 5f). Our data indicated that SlMYB52 negatively affected the formation of V type trichome in the tomato leaves. RNAi‐SlMYB52 plants with IAA treatment showed obvious increases in the densities of II, V and VI type trichomes in the leaves (Figure 5g). The calculated trichome density ratio between the IAA treatment and no‐IAA treatment of V type trichomes in the RNAi‐SlMYB52 plants were higher than that in the WT plants, and the ratios of II and VI type trichomes in the RNAi‐SlMYB52 plants were similar with the ratio in the WT plants (Figure 5h). The down‐regulation of SlMYB52 increased the effect of IAA treatment on the formation of V type trichomes in the tomato leaves, and SlMYB52 was involved in the auxin‐induced formation of V type trichome in tomato leaves.

Figure 5

Expression analysis of SlMYB52, subcellular localization analysis of SlMYB52 protein and trichome density of leaves in SlMYB52 plants. a, Expression pattern of SlMYB52 in trichome tissues explored by GUS reporter gene driven by the SlMYB52 promoter. Scale bars = 0.2 mm. b, Subcellular localization analysis of SlMYB52 in tobacco leaves. 35S: GFP was used as positive control. Scale bars = 15 μm. c, qRT‐PCR analysis of the expression of SlMYB52 in RNAi plants. d, Scanning electron micrographs of the leaf surface. Scale bars = 60 mm. e, Density analysis of II, V and VI type trichomes from the leaves of WT and SlMYB52 RNAi plants without IAA treatment. Scale bars = 0.2 mm. f, Number of trichomes per epidermal cell of tomato leaves from WT and RNAi‐SlMYB52 plants without IAA treatment. g. Density analysis of II, V and VI type trichomes in the leaves of WT and SlMYB52 lines with IAA treatment. h, Trichome density ratio of IAA treatment to no‐IAA treatment. Number of II type trichomes in an area of 0.5 cm2 and number of V and VI type trichomes in an area of 2.2 mm2 were calculated under a light microscope. WT, wild‐type plants; RNAi‐1, RNAi‐3 and RNAi‐9, RNAi‐SlMYB52 plants. All experiments were replicated three times. ** represent significant difference between WT and transgenic leaves with P < 0.01 (t‐test).

Spider mite inoculation assay, preference assay and fecundity assay were conducted to analyse the response of RNAi‐SlMYB52 plants to spider mites. RNAi‐SlMYB52 plants showed no change in the response to spider mite compared with WT plants (Figure S17a, b, c).

SlARF4 depended on SlTHM1 to regulate type II and VI trichome formation

The OE‐SlARF4 and RNAi‐SlARF4 lines were separately crossed with the RNAi‐SlTHM1 line to further explore the interaction between SlARF4 and SlTHM1 in the formation of type II, V and, VI trichomes. The densities of II and VI type trichomes in RNAi‐SlTHM1 and OE‐SlARF4 crossed (OE‐SlARF4♂RNAi‐SlTHM1♀) lines without IAA treatment were similar to those in RNAi‐SlTHM1 and OE‐SlARF4 lines and higher than those in the WT plants (Figure 6a). The densities of V type trichomes on OE‐SlARF4♂RNAi‐SlTHM1♀ lines increased compared with those on RNAi‐SlTHM1 and WT lines and were similar with those in the OE‐SlARF4 lines. Moreover, the densities of II and VI type trichomes in RNAi‐SlTHM1 and RNAi‐SlARF4 crossed (RNAi‐SlARF4♂RNAi‐SlTHM1♀) lines were increased compared with those in the RNAi‐SlARF4 and WT lines and were similar to those in the RNAi‐SlTHM1 and OE‐SlARF4 lines. The densities of V type trichome in the RNAi‐SlARF4♂RNAi‐SlTHM1♀ lines were lower than those in the RNAi‐SlTHM1 and WT lines and similar to those in the RNAi‐SlARF4 lines (Figure 6a). Hybrid experiments showed that the down‐regulation of SlTHM1 specifically increased II and VI type trichomes in the SlARF4 down‐regulation background. SlARF4 induced the formation of II and VI type trichomes and was dependent on SlTHM1.

Figure 6

Density analysis of II, V and VI type trichomes from the leaves of crossed plants among OE‐SlARF4, RNAi‐SlARF4, RNAi‐SlTMH1 and RNAi‐SlMYB52 plants. a, Trichome densities of crossed plants between OE‐SlARF4 or RNAi‐SlARF4 and RNAi‐SlTMH1 plants without IAA treatment. b, Trichome densities of crossed plants between OE‐SlARF4 or RNAi‐SlARF4 and RNAi‐SlTHM1 plants with IAA treatment. c, Trichome density ratio of IAA treatment to no‐IAA treatment. d, Trichome densities of crossed plants between OE‐SlARF4 or RNAi‐SlARF4 and RNAi‐SlMYB52 plants without IAA treatment. e, Trichome densities of crossed plants between OE‐SlARF4 or RNAi‐SlARF4 and RNAi‐SlMYB52 plants with IAA treatment. f, Trichome density ratio of IAA treatment to no‐IAA treatment. Number of II type trichomes in an area of 0.5 cm2 and number of V and VI type trichomes in an area of 2.2 mm2 were calculated under a light microscope. All experiments were replicated three times. Significant differences (Tukey’s multiple range test, P < 0.05) are indicated in lowercase.

The densities of II, V and VI type trichomes in the OE‐SlARF4♂RNAi‐SlTHM1♀ and RNAi‐SlARF4♂RNAi‐SlTHM1♀ lines with IAA treatment increased compared with the plants without IAA treatment (Figure 6b). The trichome density ratios between IAA treatment and no‐IAA treatment were calculated. The density ratios of II and VI type trichomes on OE‐SlARF4♂RNAi‐SlTHM1♀ and RNAi‐SlARF4♂RNAi‐SlTHM1♀ lines were similar to those in the RNAi‐SlTHM1 plants and higher than those in the OE‐SlARF4, RNAi‐SlARF4 and WT plants (Figure 6c). The density ratios of V type trichome in OE‐SlARF4♂RNAi‐SlTHM1♀ were similar to those in the OE‐SlARF4 plants, and the density ratios of V type trichomes in the RNAi‐SlARF4♂RNAi‐SlTHM1♀ lines were similar to those in the RNAi‐SlARF4 plants (Figure 6c). The down‐regulation of SlTHM1 gene specifically increased the IAA‐induced formation of II and VI type trichomes in the SlARF4 down‐regulation background, and the regulation of SlARF4 in the auxin‐induced formation of II and VI type trichomes was dependent on SlTHM1 in tomato.

SlARF4 depended on SlMYB52 to regulate V type trichome formation

OE‐SlARF4 and RNAi‐SlARF4 lines were separately crossed with RNAi‐SlMYB52 lines. The densities of V type trichomes on RNAi‐SlMYB52 and OE‐SlARF4 crossed (OE‐SlARF4♂RNAi‐SlMYB52♀) lines without IAA treatment were similar to those in the RNAi‐SlMYB52 lines and higher than those in the WT plants (Figure 6d). The densities of II and VI type trichomes in the OE‐SlARF4♂RNAi‐SlMYB52♀ lines increased compared with those in the RNAi‐SlMYB52 and WT lines and were similar to those in the OE‐SlARF4 lines (Figure 6d). Moreover, the densities of V type trichomes in RNAi‐SlMYB52 and RNAi‐SlARF4 crossed (RNAi‐SlARF4♂RNAi‐SlMYB52♀) lines increased compared with those in the RNAi‐SlARF4 and WT lines and were similar to those in the RNAi‐SlMYB52 and OE‐SlARF4 plants (Figure 6d). The densities of II and VI type trichomes in the RNAi‐SlARF4♂RNAi‐SlMYB52♀ lines were lower than those in the RNAi‐SlMYB52, OE‐SlARF4 and WT lines and similar to those in the RNAi‐SlARF4 plants (Figure 6d). Cross experiments showed that the down‐regulation of SlMYB52 gene specifically increased the formation of V type trichome in the SlARF4 down‐regulation background. The regulation of SlARF4 in the formation of V type trichomes in tomato was dependent on SlMYB52.

The densities of II, V and VI type trichomes in the OE‐SlARF4♂RNAi‐SlMYB52♀ and RNAi‐SlARF4♂RNAi‐SlMYB52♀ lines with IAA treatment obviously increased compared with those in plants without IAA treatment (Figure 6e). The trichome density ratios between the IAA treatment and no‐IAA treatment were calculated. The density ratios of V type trichomes in the OE‐SlARF4♂RNAi‐SlMYB52♀ and RNAi‐SlARF4♂RNAi‐SlMYB52♀ lines were similar to those in the RNAi‐SlMYB52 plants and higher than those in the OE‐SlARF4, RNAi‐SlARF4 and WT plants (Figure 6f). The density ratios of II and VI type trichomes in the OE‐SlARF4♂RNAi‐SlMYB52♀ plants were similar to those in the OE‐SlARF4 plants, and the ratios of II and VI type trichomes in the RNAi‐SlARF4♂RNAi‐SlMYB52♀ lines were similar to those in the RNAi‐SlARF4 plants (Figure 6f). The down‐regulation of SlMYB52 gene specifically increased the IAA‐induced formation of V type trichome in the SlARF4 down‐regulation background, and the regulation of SlARF4 in the auxin‐induced formation of V type trichomes was dependent on SlMYB52 in tomato.

SlTHM1 and SlMYB52 positively regulated the expression of SlCycB2

SlCycB2, a B‐type cyclin gene, plays key roles in trichome initiation (Gao et al., 2017). The promoter of SlCycB2 contains an MYB‐binding motif (AC‐rich). The expression levels of SlCycB2 decreased in the SlTHM1 and SlMYB52 RNAi plants (Figure S18). The direct bindings of SlTHM1 and SlMYB52 to SlCycB2 promoter were analysed by EMSA. Purified recombinant truncated SlTHM1, SlMYB52 and GST fusion proteins (GST‐tSlTHM1 and GST‐tSlMYB52) were obtained (Figure S19). GST‐tSlTHM1 and GST‐SlMYB52 fusion proteins bound to biotin‐labelled probes containing the AC‐rich motifs of the promoters of SlCycB2 (Figures 7a and b), indicating the specific targets of SlTHM1 and SlMYB52 to the promoters of SlCycB2. Transient dual‐luciferase assay was used in determining the transcriptional regulation of SlCycB2 by SlTHM1 and SlMYB52. The overexpression of SlTHM1 and SlMYB52 remarkably increased luciferase activity driven by SlCycB2 promoters compared with the empty control vector (pEAQ) (Figures 7c and d), and SlTHM1 and SlMYB52 increased the transcription of SlCycB2.

Figure 7

SlTHM1 and SlMYB52 target SlCycB2 and working model for R2R3 MYB‐dependent auxin signalling pathway. a and b, EMSA showing the direct binding of SlTHM1 and SlMYB52 to SlCycB2 promoters. + or +++ indicates increasing amounts of unlabelled probes for competition. c, Diagrams of reporter and effector vectors in the dual‐luciferase reporter assay. d, SlTHM1 increases the transcription of SlCycB2. e, SlMYB52 increases the transcription of SlCycB2. Each value represents the mean of six biological replicates, and vertical bars represent the SE. f, Working model. Auxin induces the expression of SlARF4 gene, and SlARF4 protein inhibits the expressions of SlTHM1 and SlMYB52 genes. The inhibited SlTHM1 levels reduce the SlCycB2 expression, which promotes the formation of II and VI type trichomes. The inhibited SlMYB52 levels reduce the SlCycB2 expression, which promotes the formation of V type trichome.

Discussion

Increased trichome density confers spider mite tolerance

The two‐spotted spider mite is a ruthless pest that damages more than 140 plant families and 1100 plant species, including tomato (Dermauw, et al., 2013). Spraying synthetic acaricides is primary performed to inhibit spider mite infestation. The drawback to the use of synthetic acaricides is that spider mites have the ability to upsurge resistance to acaricides (Dermauw, et al., 2013; Van Leeuwen et al., 2004). An imperative substitute approach to beating spider mites is the breeding of tomato cultivars with resistance to spider mites (Johnson, 1992).

Elevated trichome density hampers pest feeding and migration and diminishes herbivore populations (Handley et al., 2005; Horgan et al., 2009). VOCs by glandular trichomes fend off or destroy pests (Schilmiller et al., 2010). In this study, the overexpression of SlARF4 amplified the densities of II, V and VI type trichomes and the tolerance of tomato to spider mites (Figures 1 and 2). Furthermore, the down‐regulation of SlTHM1, a repressor of trichome formation, increased the densities of II and VI type trichomes and tolerance to spider mites (Figures 3 and 4). However, the RNAi‐SlMYB52 plants with increased density of V type trichome did not exhibit increase in tolerance to spider mite compared with the WT plants (Figure S17). Our study confirmed the positive correlation between trichome density and spider mite tolerance and demonstrated that the regulation of trichome densities by genetic engineering is an effective strategy for increasing plant resistance to herbivores. In addition, the major agricultural traits of the transgenic plants were also analysed. No significant difference was detected between transgenic and WT plants (Figures S20, S21 and S22), indicating that the R2R3 MYB‐dependent auxin signalling may not be involved in the regulation of some other agricultural traits.

Auxin triggers the formation of unicellular and multicellular trichomes via SlARF4

Hormones regulate trichome formation, and different hormones stimulate different type of trichomes in tomato (Maes and Goossens, 2010). The utilization of JA application promotes the creation of multicellular I and VI type trichomes and unicellular V type trichome (Maes et al., 2010), whereas cytokinin and gibberellin cause the creation of I type trichomes in tomato (Maes et al., 2010). Given the fact that both auxin and JA play key roles in trichome formation, the expression levels of SlARF3, SlARF4, SlTHM1, SlMYB52 and SlCycB2 in response to auxin and JA treatments were analysed by qRT‐PCR (Figures S23, S24). All five genes showed response to IAA treatment, with SlARF4 showing the highest up‐regulation whereas SlARF3 displayed slight up‐regulation (Figure S23). On the other hand, SlTHM1, SlMYB52 and SlCycB2 were down‐regulated in response to IAA treatment (Figure S23). Moreover, SlARF3, SlARF4, SlTHM1 and SlMYB52 were barely responsive to JA treatment while SlCycB2 manifested mild down‐regulation (Figure S24). In this study, auxin promoted the formation of unicellular V type trichome and multicellular II and VI type trichomes. SlARF4 positively modulates the formation of II, V and VI type trichomes, and the down‐regulation of SlARF4 decreases the effect of IAA treatment on the formation of II, V and VI type trichomes (Figure 1). Our results demonstrated that auxin induced the formation of II, V and VI type trichomes in tomato leaves by promoting SlARF4 expression. In addition, IAA treatment also increased the density of trichomes in RNAi‐SlARF4 and Slarf4 plants, which suggested that, other than SlARF4, there might be another regulatory pathway in controlling trichome formation. Further research will be needed to examine trichome formation in response to IAA treatment.

It is noteworthy that another auxin‐responsive gene, SlARF3, is involved in the formation of epidermal cells and trichomes (Zhang et al., 2015). The relation between SlARF3 and SlARF4 in regulating trichome formation was explored by multiple approaches. The expression level of SlARF3 was first assessed in SlARF4 transgenic lines (Figure S25). qRT‐PCR results revealed that there was no significant change of the expression level of SlARF3 in SlARF4 transgenic lines, suggesting that SlARF3 and SlARF4 could function independently in regulating trichome formation. Next, further examination of both proteins reveals that, SlARF4 protein has typical domains (B3, ARF and Aux/IAA) while SlARF3 only contains two conserved domains, B3 and ARF (Zhang et al., 2015). Furthermore, we generated SlARF3/4 double knockdown tomato plants by crossing RNAi‐SlARF3 and RNAi‐SlARF4 transgenic lines. These plants exhibited significant lower trichome density, epidermal cell density and number of trichomes per epidermal cells compared with the parent lines (Figures S26, S27 and S28). Finally, RNA‐Seq analyses of RNAi‐SlARF3 and RNAi‐SlARF4 plants revealed that SlARF3 and SlARF4 regulate the different set of down‐stream genes (data not shown). These results demonstrate that SlARF3 and SlARF4 play different regulatory roles in regulating trichome formation.

SlARF4 regulates the auxin‐induced formation of unicellular and multicellular trichomes through the transcriptional inhibition of SlMYB52 and SlTHM1

Numerous overexpressed genes linked to unicellular trichomes in Arabidopsis negate the same result for Solanaceae species with multicellular trichome formation (Payne et al., 1999). In Arabidopsis, the overexpression of MIXTA and AmMYBML1, which are multicellular trichome‐related genes, did not prompt the formation of unicellular trichomes (Glover et al., 1998), whereas the interaction between Wo and SlCycB2 regulated trichome formation in tomato (Yang et al., 2011). The promotion of trichome formation was not observed in Arabidopsis with overexpressed Wo gene (Yang and Ye, 2013). These reported data implied tomato and tobacco multicellular trichomes and Arabidopsis unicellular trichomes employ different regulatory pathways (Yang and Ye, 2013).

SlARF4 regulated the formation of unicellular and multicellular trichomes. SlARF4 negatively regulated SlTHM1 and SlMYB52 expression by binding to AuxRE and TGA motifs, respectively (Figure 2). SlTHM1 functioned as a repressor in the regulation of the formation of II and VI type trichomes, and SlMYB52 acted as a repressor during V type unicellular trichome formation. qRT‐PCR showed that the expression levels of SlTHM1 and SlMYB52 increased in the RNAi‐SlARF4 plants and decreased in the OE‐SlARF4 plants (Figure S9). Hybrid experiments showed down‐regulation of SlTHM1 specifically increased the formation of II and VI type trichomes, and SlMYB52 specifically increased the formation of V type trichome in the SlARF4 down‐regulation background (Figure 6). SlARF4 modulated the formation of II, V and VI type trichomes depending on SlTHM1 and SlMYB52. II and V type trichomes are nonglandular, whereas V type is glandular (Deng et al., 2012). SlARF4 positively regulated the formation of unicellular and multicellular trichomes through the direct transcriptional inhibition of SlMYB52 and SlTHM1. Our study demonstrated the presence of consistent but different characteristics in the regulation of unicellular and multicellular trichome formation in tomato leaves.

Compared with SlUBI expression, real expression levels of SlTHM1 and SMYB52 are not very high in the tomato leaf tissue (Figures 3c,5c). Furthermore, SlTHM1 had highest expression level in leave trichomes (Figure S12) and SMYB52 had relatively high expression level in leaves (Figure S15). These facts indicate that there might be other regulatory mechanisms in regulation trichome formation. Protein translation regulation and post‐translation regulation also impact the actual protein levels of SlTHM1 and SMYB52 in regulating trichome formation. In addition, besides R2R3 MYB‐dependent auxin signalling, there should exist other regulatory pathways in regulating trichome formation, such as JA signalling pathway.

The down‐regulation of SlTHM1 increased the effect of IAA treatment on the formation of II and VI type trichomes, and the down‐regulation of SlMYB52 increased the effect of IAA treatment on the formation of V type trichome (Figures 3 and 5). SlTHM1 and SlMYB52 were involved in the auxin‐induced formation of II, V and VI type trichomes in tomato leaves. Our results demonstrated the important roles of SlARF4, SlTHM1 and SlMYB52 in the auxin‐mediated transcriptional regulation of unicellular and multicellular trichome formations in tomato leaves.

SlTHM1 and SlMYB52 regulate trichome formation by directly binding to SlCycB2

Cyclins are involved in the transition between the phases of the cell cycle in eukaryotes and function as the positive regulators of cell proliferation (Meijer and Murray, 2001). B‐type cyclins play important roles in G2/M transition (Fobert et al., 1994; Hirt et al., 1992). In Arabidopsis, the specific expression of B‐type cyclin B1:2 induces the formation of multicellular trichomes; hence, B‐type cyclins play important roles in unicellular and multicellular trichome formation (Schnittger et al., 2002). In tomato, Wo protein interacts with SlCycB2, which is essential for type I trichome formation (Yang et al., 2011). The overexpression of SlCycB2 decreases the levels of glandular I and VI type trichomes and all nonglandular trichomes. On the other hand, the suppression of SlCycB2 promotes the formation of nonglandular III and V type trichomes on Ailsa Craig tomato (Gao et al., 2017). Hence, SlCycB2 may function as an inhibitor in the formation of glandular and nonglandular trichomes (Gao et al., 2017). In the present study, SlTHM1 and SlMYB52 functioned as repressors in the formation of multicellular and unicellular trichomes (Figures 3 and 5). SlTHM1 and SlMYB52 directly targeted the SlCycB2 and activated its expression (Figure 7). We anticipated that SlTHM1 and SlMYB52 regulate trichome formation through the activation of SlCycB2 that may act as a repressor in the formation of unicellular and multicellular trichomes in ‘Micro‐Tom’ tomato cultivar. Future study could focus on functional analysis of SlCycB2 in ‘Micro‐Tom’ using CRISPR/Cas9 technology.

We proposed a model of how auxin induces the formation of II, V and VI type trichomes in tomato leaves. Auxin induces the expression of SlARF4, and SlARF4 protein inhibits the expressions of SlTHM1 and SlMYB52. Decreased SlTHM1 levels reduce SlCycB2 expression, which promotes the formation of II and VI type trichomes. Decreased SlMYB52 levels inhibit SlCycB2 expression, resulting in the promotion of the formation of V type trichomes (Figure 7). The SlTHM1/SlMYB52‐dependent auxin signalling pathway modulates the formation of unicellular and multicellular trichomes in tomato, and increasing trichome density is an effective method to improve the tolerance of tomato to spider mites.

Methods

Plant material and growth conditions

Tomato (Solanum lycopersicum ‘Micro‐Tom’) plants were used. ‘Micro‐Tom’ is a typical laboratorial tomato cultivar because of its short life cycle and efficient genetic transformation system. Standard greenhouse conditions are 14‐h day/10‐h night cycle, 25°C/20°C day/night temperature, 80% relative humidity (RH) and 250 mol/m2/s intense luminosity. Major agricultural traits were measured as previously described by Lovelli et al., 2012.

Sequence analysis

Sequence analysis was performed according to Zhang et al. (2015). GenBank accession numbers for the alignment as well as phylogenetic analysis are presented in supplementary materials.

Trichome counts and phenotyping

Forty‐five‐day‐old tomato seedlings were used for trichome counts. Fully expanded leaves were collected from the fifth internode counted from the shoot tip. Samples were dissected from midway between the margin and midrib in 10 mm × 4 mm strips covering the whole leaf blade (avoiding the primary veins). II, V and VI type trichomes on the adaxial leaf surfaces were analysed under a JNOEC JSZ5B stereo microscope and a HITACHI TM400 plus scanning electron microscope. The numbers of type II trichomes in an area of 0.5 cm2 and the numbers of V and VI type trichomes in an area of 2.2 mm2 were calculated. Adaxial epidermal pavement cells were analysed through colourless nail polish printing mark method.

Auxin and JA treatment

Indole acetic acid (IAA) and jasmonate acid (JA) were purchased from Sigma company. Fifteen‐day‐old tomato seedlings for auxin treatment were sprayed with IAA solution (0, 1,10, 30, 50 mg/L) every 2 days for 1 month. Four tomato seedlings were used for each group, and three groups were used for each treatment. Trichome analysis was conducted 1 month after the first spray. IAA solution (30 mg/L) was used in subsequent experiments in this study. For qRT‐PCR analysis, JA solution (100 µm) and IAA solution (30 mg/L) were used. Chemical induction of 3‐week‐old tomato plants was conducted by dipping the tomato leaf in a solution containing either IAA or JA. Leaves from four WT were collected at three time points (0 h, 2 h and 6 h) during IAA and JA treatments, and frozen immediately in liquid nitrogen for RNA extraction.

Subcellular localization of SlTHM1 and SlMYB52

Subcellular localization assays were conducted in tobacco (Nicotiana benthamiana) leaves. The coding regions of SlTHM1 and SlMYB52 without a stop codon were cloned into the pCX‐DG vector in frame with the GFP sequence and cauliflower mosaic virus (CaMV) 35S promoter. Agrobacterium tumefaciens strain GV3101, which carried the fusion constructs and the control GFP vector (pCX‐DG frame), were infiltrated into the abaxial air space of 4‐ to 6‐week‐old tobacco plants, using a needleless 2‐mL syringe. GFP fluorescence was observed using a laser scanning confocal microscope. All transient expression assays were repeated at least three times. The primers used are listed in Table S4.

Generation of transgenic plants

The open reading frame sequence of SlARF4 was amplified and cloned into plant binary vector pLP100 to obtain SlARF4 overexpression vector. RNAi vector was constructed by cloning the target sequences of SlARF4, SlTHM1 and SlMYB52 into pCAMIBA2301. GUS staining vector was constructed by cloning the 2 kb promoter sequences of SlTHM1 and SlMYB52 into the pLP100 vector that contain GUS reporter gene. Transgenic plants were obtained via A. tumefaciens‐mediated method based on Zhang et al., 2015. All experiments were conducted using homozygous lines from T3 generations. The primers used are listed in Table S4.

qRT‐PCR

Total RNA was extracted using the RNeasy plant mini kit (Qiagen), and qRT‐PCR was conducted using All‐in‐One™ qPCR mix (GeneCopoeia) according to Deng et al. (2012). The relative expression level for each gene was evaluated using the ΔΔCt values with SlActin as internal control. The primer sequences used are listed in Table S4.

GUS staining

GUS staining was conducted according to Yuan et al.,(2019).

Spider mite bioassays

General procedures for treating two‐spotted spider mites (Tetranychus urticae) were performed according to Li et al. (2004). Inoculation assay was done by transferring 15 adult female mites to a single leaf using one‐month‐old seedlings. Preference assays were performed by placing 10 mites in a 1 cm circle located equidistantly (1 cm) between the single leaflets derived from 5‐week‐old WT and transgenic plants. Spider mites were counted at 1 h after initiating the trial. Fecundity assays started with transferring five adult female mites to leaf discs (12 mm) of WT and transgenic plants that were placed on wetted cotton (Rodriguez et al., 1971; Rodriguez et al., 1972). Eggs were counted using a microscope at 24‐h intervals for four days.

RNA‐Seq analysis

RNA‐Seq was conducted by Shanghai Majorbio Biopharm Technology Co., Ltd. Total RNA of the leaf tissue was isolated with the DNeasy plant mini kit (Qiagen). RNA‐Seq was performed according to conditions as previously reported (Zhang et al., 2015).

Promoter analysis and dual‐luciferase transient expression assay

Motif promoter sequences were analysed using PLACE Signal Scan search software (http://www.dna.affrc.go.jp/PLACE/signalscan.html). Dual‐luciferase transient expression assay was performed by amplifying and cloning the sequence of gene‐coding regions into the pGreenII 62‐SK vector (Hellens et al., 2005). GAL4 sequence or gene promoter sequences were cloned into pGreenII 0800‐LUC vector (Hellens et al., 2005). The activities of LUC and REN luciferases were measured by Luminoskan Ascent microplate luminometer using a dual‐luciferase assay kit (Promega, Madison, WI, USA). Six biological repeats were used for each pair of vectors. All primers for this assay are listed in Table S4.

Protein expression and EMSA

Truncated gene‐coding regions were cloned into pGEX‐4T‐1 bacterial expression vectors (GE Healthcare Life Science, China) and expressed in Escherichia coli strain BM Rosatta (DE3). Recombinant proteins were induced with 0.5 mm isopropyl‐β‐D‐thiogalactopyranoside for 16 h at 28 °C and purified through a GST‐tagged protein purification kit (Clontech, Polo Alto, CA, USA). LightShift chemiluminescent EMSA kit (Thermo Fisher Scientific, Rockford, IL, USA) was used according to Han et al. (2016). EMSA assay was conducted according to Yuan et al.,(2019). All primers designed for EMSA are listed in Table S4.

ChIP‐qPCR assay

ChIP‐qPCR assay was conducted as described by Qin et al. (2012). All primers designed for ChIP‐qPCR analysis are shown in Table S4.

Statistics

Unpaired, two‐tailed Student’s t‐tests were performed to compare individual lines with their relevant controls. Univariate ANOVA followed by the post hoc Tukey test of multiple pairwise comparisons was used for the comparison among the measurements of multiple experiment designs. P < 0.05 was considered a significant difference, and P < 0.01 was considered a highly significant difference.

Conflicts of interest statement

The authors declare no conflicts of interest.

Author contributions

W.D. and Z.G.L. conceived the research. Y.J.Y., X.X., Y.Q.L., Z.H.G., X.W.H., M.B.W., Y.D.L., F.Y., X.L.Z., W.F.Z., Y.W.T. and B.H.F performed experiments. W.F.Z. analysed the data. W.D. and Y.J.Y. wrote the manuscript. C.Z.J. revised the manuscript.

Figure S1 Density analysis of epidermal pavement cells in the leaves of IAA‐treated tomato plants.

Figure S2 qRT‐PCR analysis of SlARF4 expression levels

Figure S3 Density analysis of epidermal pavement cells in the leaves of OE‐SlARF4 and RNAi‐SlARF4 plants.

Figure S4 Trichomes densities of fruits in SlARF4 transgenic plants.

Figure S5 Trichomes densities of leaves in Slarf4 CRISPR‐Cas9 mutants.

Figure S6 RNA‐Seq analysis of RNAi‐SlARF4 plants.

Figure S7 RNA‐Seq analysis of RNAi‐SlARF4 plants.

Figure S8 SDS‐PAGE gel stained with Coomassie brilliant blue demonstrating the affinity purification of recombinant GST‐tSlARF4 protein used for EMSA.

Figure S9 qRT‐PCR analysis of the expression levels of SlTHM1 and SlMYB52 genes in RNAi‐SlARF4 and OE‐SlARF4 plants.

Figure S10 Sequence alignments of SlMYB52 and SlTHM1 with known proteins from the same subgroup

Figure S11 Phylogenetic trees of SlTHM1 and SlMYB52 with homologous proteins from other species

Figure S12 qRT‐PCR analysis of SlTHM1 expression levels.

Figure S13 Diagrams of the reporter and effector vectors and transcriptional activation activity assay of SlTHM1

Figure S14 Density of epidermal pavement cells in RNAi‐SlTHM1 plants

Figure S15 qRT‐PCR analysis of SlMYB52 expression levels.

Figure S16 Density of epidermal pavement cells in RNAi‐SlMYB52 plants.

Figure S17 Spider mite bioassay of SlMYB52 RNAi plants

Figure S18 qRT‐PCR analysis of the expression of SlCycB2 in leaves of SlTHM1 and SlMYB52 transgenic lines.

Figure S19. SDS‐PAGE gel stained with coomassie brilliant blue demonstrating affinity purification of the recombinant GST‐tSlTHM1 and GST‐tMYB52 used for the EMSA assay

Figure S20 Agricultural trait analysis of the SlARF4 transgenic plants.

Figure S21 Agricultural trait analysis of the SlTHM1 transgenic plants.

Figure S22 Agricultural trait analysis of the SlMYB52 transgenic plants.

Figure S23 Expression patterns of SlARF3, SlARF4, SlTHM1, SlMYB52 and SlCycB2 genes in response to IAA treatments

Figure S24 Expression patterns of SlARF3, SlARF4, SlTHM1, SlMYB52 and SlCycB2 genes in response to JA treatments.

Figure S25 qRT‐PCR analysis of the expression of SlARF3 in leaves of SlARF4 transgenic lines

Figure S26 Density analysis of II, V and VI type trichomes of RNAi‐SlARF3 and RNAi‐SlARF4 crossed plants.

Figure S27 Number of trichomes per epidermal cell of RNAi‐SlARF3 and RNAi‐SlARF4 crossed plants.

Figure S28 Density of epidermal pavement cells of RNAi‐SlARF3 and RNAi‐SlARF4 crossed plants.

Click here for additional data file.

Table S1 All DEGs in RNAi‐SlARF4 plants

Click here for additional data file.

Table S2 GO function and pathway enrichment analyses of DEGs in RNAi‐SlARF4 plants

Click here for additional data file.

Table S3 DEGs encoding TFs in RNAi‐SlARF4 plants.

Click here for additional data file.

Table S4 Primers used in this study.

Click here for additional data file.