SlMYB72 affects pollen development by regulating autophagy in tomato

Abstract The formation and development of pollen are among the most critical processes for reproduction and genetic diversity in the life cycle of flowering plants. The present study found that SlMYB72 was highly expressed in the pollen and tapetum of tomato flowers. Downregulation of SlMYB72 led to a decrease in the amounts of seeds due to abnormal pollen development compared with wild-type plants. Downregulation of SlMYB72 delayed tapetum degradation and inhibited autophagy in tomato anther. Overexpression of SlMYB72 led to abnormal pollen development and delayed tapetum degradation. Expression levels of some autophagy-related genes (ATGs) were decreased in SlMYB72 downregulated plants and increased in overexpression plants. SlMYB72 was directly bound to ACCAAC/ACCAAA motif of the SlATG7 promoter and activated its expression. Downregulation of SlATG7 inhibited the autophagy process and tapetum degradation, resulting in abnormal pollen development in tomatoes. These results indicated SlMYB72 affects the tapetum degradation and pollen development by transcriptional activation of SlATG7 and autophagy in tomato anther. The study expands the understanding of the regulation of autophagy by SlMYB72, uncovers the critical role that autophagy plays in pollen development, and provides potential candidate genes for the production of male-sterility in plants.


Introduction
Pollen formation and development are necessary for the reproduction of f lowering plants and consist of several distinct stages [1]. Pollen grains are developed from microspore mother cells in the microsporangium of an anther. Each microspore mother cell can form a microspore tetrad (four haploid microspores) through meiotic division. When the anther matures, the microspores dissociate and develop into pollen grains [2]. The development of pollen could be inf luenced by many factors, such as tapetum irregularity [3][4][5][6], cytoskeleton alteration [7,8], auxin metabolism aberration [9,10], altered sugar utilization [6,11], and reactive oxygen species [12,13]. Among all these factors the tapetum plays an important role in the regulation of pollen development [14].
Tapetum is the innermost layer of the anther and adheres to the microspores. The tapetum provides various enzymes, proteins, lipids, starch, sporopollenin, and other molecules required for pollen development [1,15,16]. An adequately programmed cell death (PCD) is necessary for the degradation of tapetum, which for pollen development and microspores release [17,18]. Many transcription factors (TFs) affect pollen development by regulating the tapetum PCD. The knocking out of the AtMYB103 gene in Arabidopsis leads to defective tapetal cell wall degradation, pollen development, and pollen exine formation [19]. Mutants of Arabidopsis MYB33, MYB65, and rice GAMYB genes result in a loss of tapetum PCD and abnormal pollen development [20][21][22]. Arabidopsis MYB80 regulates the tapetal PCD and pollen development by direct binding to pectin methylesterase, glyoxal oxidase, and aspartic protease genes [3]. A putative bHLH gene affects the tapetum development and pollen formation in tomatoes [23].
Macroautophagy (autophagy) is involved in plant growth and development and plays critical roles in many abiotic and biotic stress responses [24]. In normal circumstances, autophagy remains at a base level to degrade cytoplasmic components and maintain homeostasis in cells. Previous research has well documented that autophagy loss reduces seed yields and causes the senescence of premature leaves [25][26][27], but an increased level of autophagy promotes growth, seed yields, and nitrogen remobilization [28,29]. Autophagy is upregulated and recycles cellular material and nutrients to promote plant survival during senescence or stress response [24,30,31]. Autophagy is involved in cell death during tapetum degradation [32]. In tobacco, the overexpression of Arabidopsis ATG6/Beclin1 genes in tapetum results in sterility, and excessive autophagy increases the PCD and abortion of microsporogenesis [32,33]. The rice autophagy mutant OsATG7 exhibits delay of the tapetum layer's degradation and leads to sporophytic male sterility [34,35]. High temperatures lead to tapetal PCD abortion and damage pollen development, but autophagy mitigates the injury in pollen development in Arabidopsis [36]. The R2R3MYB is one of the most prominent families of TFs and is involved in plant development, metabolism, and stress responses. We have reported that SlMYB72 is a typical R2R3 MYB family TF and is involved in the accumulation of pigment in tomato fruits [37]. In the present study, we found that SlMYB72 affected the tapetum degradation and pollen development in tomatoes. SlMYB72 affected SlATG7 expression and autophagy process in tomato anthers. SlATG7 also regulated the tapetum degradation and pollen development in tomatoes. The results strongly suggest that SlMYB72 affects pollen development by regulating autophagy in tomatoes and uncovers the important role of autophagy in tomato development.

SlMYB72 highly expresses in pollen and tapetum of tomato
Our previous study has shown that SlMYB72 is a typical R2R3 MYB family transcription factor. In this study, the SlMYB72 promoter sequence was submitted to a public database (http://www.dna. affrc.go.jp/PLACE/signalup.html) to search for the cis-acting element. The SlMYB72 promoter contains pollen-specific elements POLLEN1LELAT52 (AGAAA) and GTGANTG10 (GTGA) (Fig. 1a). This result indicated that SlMYB72 might be involved in the pollen formation.
The expression level of SlMYB72 in f lower tissue was precisely analysed. RT-qPCR data showed that the SlMYB72 gene was expressed weekly in the calyx, petal, carpel, and pistil, but expressed strongly in the stamen (Fig. 1b). Glucuronidase (GUS) staining of plant tissues was used to analyse the SlMYB72 expression pattern. As shown in Fig. 1c, expression of the GUS gene driven by the SlMYB72 promoter was found in the stamen, especially in pollen and tapetum. These results indicated that SlMYB72 has a tissue-specific expression in pollen and tapetum.

Downregulation of SlMYB72 affects the pollen development in tomatoes
Our previous study generated six SlMYB72 downregulated transgenic plants (Lines 3, 5, 8, 9, 10, and 11). The RNAi-SlMYB72 lines 9, 10, and 11 exhibited a 50-60% decrease in SlMYB72 expression and about a 50% decrease in seed count compared with the WT plant, but Lines 3, 5, and 8 with the 80-90% decrease in SlMYB72 expression levels did not produce seeds (Fig. S1, see online supplementary material). This phenotype indicated that downregulation of the expression of SlMYB72 inhibited seed formation.
To examine whether the downregulation of SlMYB72 affected female or male fertility, a cross-assay was performed. When we used the RNAi-SlMYB72 plant as the female parent to cross with the male parent from the WT plant, the seed count of the hybrid was similar to the self-pollinating WT plant, but the seed number was reduced by half when we used the RNAi-SlMYB72 plant as the male parent to cross with the WT female parent (Fig. 2a). This result indicated that reducing the number of seeds might be due to decreased pollen viability in RNAi-SlMYB72 f lowers. Then the pollen viability was tested by I 2 -KI and TTC staining. About 90% of the pollen in WT exhibited viability, while about 40%-50% of the pollen grains had viability in RNAi-SlMYB72 plants ( Fig. 2b and c). We further investigated the germination and growth of pollen in RNAi-SlMYB72 and WT plants. The germination rate of WT pollen was 80%, but only about 52%-56% in RNAi-SlMYB72 plants (Fig. 2d). The RNAi-SlMYB72 plants also showed decreased pollen growth compared with the WT plants ( Fig. 2e and f). Moreover, pollen growth in style was observed after pollination. When the RNAi-SlMYB72 plant as the female parent was used to cross with the male parent from the WT plant, the f luorescent signal of pollen tubes in the style was clearly observed, which was similar to the self-pollinating WT plants, but the f luorescent signal of pollen tubes was not clearly observed when we used the RNAi-SlMYB72 plant as the male parent to cross with the WT female parent (Fig. 2g). The data indicated that down-regulation of SlMYB72 affects the pollen development in tomatoes.
The morphology of mature pollen was analysed by using scanning electron microscopy (SEM). The pollens of RNAi-SlMYB72 plants were shriveled and irregular in shape compared with the WT pollens (Fig. 3a). A semi-thin section was used to investigate the anatomic structure of mature anthers. The tapetum of the RNAi-SlMYB72 plant was the same as the WT plant at the Pollens were incubated on germination media at 22 • C for 1 h. Scale bar = 50 μm. e, f In vitro pollen tube growth. Pollens were incubated on germination media at 22 • C for 2 and 6 h. Data are means ± SD (ten biological replicates). Asterisks represent significant differences between WT and RNAi-SlMYB72 plants (Student's t-test, * P < 0.05 and * * P < 0.01). Scale bar = 50 μm. g Fluorescence observation of pollen tube growth in the style of pistil after pollination (2 DAP). Scale bar = 50 μm.
uninucleate microspore stage. The WT tapetum has disappeared completely, but the tapetum of the RNAi-SlMYB72 plant still had a complete cell morphology at the binucleus microspore stage (Fig. 3b). The expression of MYB72 gene in transgenic lines and WT at the different stage was examined by RT-qPCR. The expression level of SlMYB72 was significantly lower than that of the control group in the EUM and MLUM stages, but there was no significant difference in the BP stage (Fig. 3c). This result indicated that downregulation of SlMYB72 delayed the tapetum degeneration in tomato anthers. RNAi-SlMYB72 f lower at full bloom stage. Scale bars = 10 μm. b Anatomical analysis of WT and RNAi-SlMYB72 anthers at different development stages. EUM, MLUM, and BP represent the stages of early uninucleate microspore, middle, and later uninucleate microspore, and binucleate pollen. Arrows indicate tapetum. Scale bar = 30 μm. c RT-qPCR analysis of the expression levels of SlMYB72 at different development stages. The data are means ± SD (four biological replicates). Asterisks represent significant differences between WT and RNAi-SlMYB72 plants (Student's t-test, * P < 0.05 and * * P < 0.01). Scale bar = 50 μm.

Downregulation of the SlMYB72 gene inhibits the autophagy in tomato anthers
It has been reported that autophagy is involved in tapetum degeneration and pollen development [35,38]. We further analysed the autophagy process in the anther of RNAi-SlMYB72 plant in current research. Fluorescent dye monodansyl cadaverine (MDC) staining was used to analyse autophagosome occurrence in RNAi-SlMYB72 anthers. Compared with WT anther, the number of punctate f luorescent signals in RNAi-SlMYB72 anther was significantly reduced when autophagy was induced with exogenous rapamycin (Fig. 4a). To further confirm the result of MDC staining, western blotting (WB) was applied to explore the formation of ATG8-phosphatidylethanolamine (ATG8-PE) conjugates as autophagy marker with an anti-ATG8a antibody. The ATG8-PE band was less abundant in RNAi-SlMYB72 anthers than that of WT anthers, especially at the early uninucleate microspore and binucleate stages (Fig. 4b). Then transmission electron microscopy (TEM) was used to study the autophagosome occurrence in RNAi-SlMYB72 anthers. Some autophagosome vesicles in the WT anthers were detected, while the autophagosome vesicles of RNAi-SlMYB72 anthers were hardly detected under the conditions of inducing autophagy (Fig. 4c). The results demonstrated that the downregulation of SlMYB72 inhibited the autophagy process in tomato anthers.
The expression levels of autophagy-related genes (ATGs) were further analysed in RNAi-SlMYB72 anthers. RT-qPCR data showed seven ATG genes' expression levels, including SlATG1, SlATG6, SlATG7, SlATG8D, SlATG8F, SlATG8H, SlATG9, and SlATG10 were decreased in the anthers of RNAi-SlMYB72 plants (Fig. 4d). The tissue was induced with exogenous rapamycin. The autophagosomes stained with MDC are shown in green f luorescence. Scale bars = 5 μm. b ATG8 protein levels in RNAi-SlMYB72 and WT anthers. ATG8-PE and ATG8 are the lipidated and nonlipidated ATG8 forms, respectively. Internal control is actin. EUM, MLUM, and BP represent the stages of early uninucleate microspore, middle, and later uninucleate microspore, and binucleate pollen. c TEM observation of autophagosomes in the tapetum of RNAi-SlMYB72 and WT plants. White arrows indicate autophagosomes. Scale bars = 1 μm. d Expression of ATGs in the WT and RNAi-SlMYB72 anthers during MLUM stage. RT-qPCR was used to analyse the expression levels. Data are means ± SD (four biological replicates). Asterisks represent significant differences (Student's t-test, * P < 0.05 and * * P < 0.01).

Overexpression of the SlMYB72 gene results in abnormal pollen development and promoted tapetum degradation in tomato anthers
To further analyse the function of SlMYB72 in pollen development, two SlMYB72 overexpression lines (OE-SlMYB72) corresponding to independent transformation events were produced. RT-PCR analysis showed that the SlMYB72 gene expression level was increased in the OE-SlMYB72 plants ( Fig. 5a; Fig. S3, see online supplementary material). The pollen viability in OE-SlMYB72 f lowers was analysed by I 2 -KI stain. About 90% of the pollen in WT exhibited viability, while about 50%-60% of the pollen grains had viability in OE-SlMYB72 plants (Fig. 5b and c). The germination rate of pollen and the growth of pollen tubes in the OE-SlMYB72 plants were further detected (Fig. 5d-f). The pollen germination rate was about 70% in the WT plants, while only 20% in OE-SlMYB72 plants (Fig. 5e). The pollen tube length was tested after culturing for 2 hours. The length of the pollen tube reached 0.45 mm in the WT plants, while the length of the pollen tube was only 0.16 mm in OE-SlMYB72 plants (Fig. 5f). The results indicated that the pollen viability and pollen germination rate were significantly affected in the OE-SlMYB72 plants, which are similar to RNAi-SlMYB72 plants.
SEM was performed to analyse the morphology of mature pollen. Most of the pollen had intact morphology in the WT plants, while the pollen collapsed and was shrunken in the OE-SlMYB72 plants (Fig. 5g). The anatomic structure of mature anthers was analysed by a semi-thin section. The OE-SlMYB72 tapetum has disappeared completely, but the tapetum of the WT still had a lot of residuals at the middle and later uninucleate microspore stage (Fig. 5h). This result indicated that overexpression of SlMYB72 triggered premature tapetum degradation and pollen abortion. The expression levels of autophagy-related genes in OE-SlMYB72 plants were further investigated. The result of RT-qPCR indicated that four ATG genes' expression levels were significantly RT-qPCR was used to analyse the expression levels. Data are means ± SD (four biological replicates). Asterisks represent significant differences (Student's t-test, * * P < 0.01). b, c Pollen viability. Pollens were stained with I2-KI. Viable pollens were stained dark with I 2 -KI staining. Scale bar = 50 μm. d, e, f In vitro pollen germination and pollen tube growth. Pollens were incubated on germination media at 22 • C for 2 h. Scale bar = 50 μm. Data are means ± SD (ten biological replicates). g Scan electron microscope observation of pollens in WT and OE-SlMYB72 f lower at full bloom stage. Scale bars = 10 μm. h Anatomical analysis of WT and OE-SlMYB72 anthers at MLUM (middle and later uninucleate microspore) stages. Scale bar = 50 μm. i Expression of ATGs in the WT and OE-SlMYB72 anthers during MLUN stage. RT-qPCR was used to analyse the expression levels. Data are means ± SD (four biological replicates). Asterisks represent significant differences (Student's t-test, * P < 0.05 and * * P < 0.01).

SlMYB72 directly targets SlATG7 genes and increases their expression
OsATG7 has been reported to regulate tapetum degeneration and pollen development in rice [35]. RT-qPCR was performed to analyse the expression of SlATG7 in MLUM stage. The expression of SlATG7 in the RNAi-SlMYB72 plants was significantly lower than that in the WT plants, while it was significantly higher in the OE-SlMYB72 transgenic plant than that in the WT plants ( Fig. 6a; Fig. S3, see online supplementary material). Promoter analysis found that the SlATG7 promoter has an AC-rich element, an R2R3-MYB binding motif. An electrophoretic mobility shift assay (EMSA) was carried out to explore the direct binding of the SlMYB72 protein to the SlATG7 gene. The recombinant proteins of SlMYB72 and GST (GST-SlMYB72) were successfully purified. The GST-SlMYB72 recombinant protein targeted the biotin-labeled probes which contained an AC-rich motif derived from the promoter of the SlATG7 gene and produced a mobility shift. Unlabeled fragment of SlATG7 promoter as a competitor abolished the gel mobility shift. But no mobility shift bands were detected when the probes were incubated with only GST (Fig. 6b). The EMSA result revealed that the SlMYB72 directly targeted the AC-rich element of the SlATG7 promoter. Chromatin immunoprecipitation (ChIP) qPCR was carried out to verify the Agrobacterium tumefaciens carrying the effector and reporter vectors were infiltrated into tobacco leaves. LUC and REN activities were analysed. Data are means ± SD (six biological replicates). Asterisks represent significant differences (Student's t-test, * P < 0.05 and * * P < 0.01).
interaction between SlMYB72 and SlATG7 promoter in vivo. SlATG7 promoter region containing the AC-rich motif was enriched when FLAG antibodies were used, but not enriched when nonspecific antibodies (IgG) were used (Fig. 6c), which indicated that the SlMYB72 targeted the SlATG7 promoter. A transient dual-luciferase reporter assay was performed to determine the regulation of SlATG7 by SlMYB72. Overexpression of the SlMYB72 effectively increased the activity of luciferase driven by the SlATG7 promoter, compared with the control (pEAQ) (Fig. 6d and e). The results further proved that SlMYB72 directly targeted the SlATG7 promoter and increased gene expression.

SlATG7 is involved in pollen development and seed formation
SlATG7 protein contains 716 amino acid residues including an N-terminal domain of ubiquitin-like modifier-activating enzyme ATG7 (ATG7-N) and a NAD/FAD-binding domain (ThiF) (Fig. S2a, see online supplementary material). Phylogenetic analysis revealed that the SlATG7 was close to NtATG7 and AtATG7 (Fig. S2b, see online supplementary material). The expression pattern of the SlATG7 gene was analysed using a public database (http://tomexpress.toulouse.inra.fr/query). The SlATG7 gene was expressed in all of the tomato plants, but was highest in the tomato f lower tissues (Fig. S4, see online supplementary material). Then RT-qPCR was performed to obtain further insights into the expression pattern, and the result showed that SlATG7 was highly expressed in stamens and carpels in tomatoes (Fig. 7a).
Ten downregulated transgenic plants were obtained to study the functions of the SlATG7 gene in tomatoes. The RT-qPCR expression analysis showed the SlATG7 decreased significantly in lines 4, 25, and 37, which were further analysed (Fig. S5, see online supplementary material). The homozygous transgenic lines (RNAi-SlATG7) showed a noticeable decrease in seed number compared with the WT plants (Fig. 7b).
Pollen viability of RNAi-SlATG7 plants was tested using I 2 -KI staining, and results revealed that downregulation of SlATG7 decreased the pollen viability in tomato anthers (Fig. 7c). Then the pollen germination was investigated. As shown in Figure 7d, the germination rate of WT pollen was about 90%, while only 60%-70% in transgenic plants.
The SEM analysis showed that the RNAi-SlATG7 pollens were shriveled and irregular in shape compared with the WT pollens (Fig. 7e). Analysis of the semi-thin section indicated the RNAi-SlATG7 tapetum still had a complete cell morphology at the Asterisks represent significant differences (Student's t-test, * P < 0.05 and * * P < 0.01).
binuclear microspore stage (Fig. 7f). However, the WT tapetum disappeared utterly at the binuclear microspore stage (Fig. 7f). The combined results indicated that downregulation of SlATG7 affected the tapetum degeneration in tomato anthers.

Downregulation of SlATG7 inhibits autophagy in the tapetum
We further analysed the autophagy process in the anthers of RNAi-SlATG7 plants. The autophagy-related genes were Figure 8. Analysis of autophagy in RNAi-SlATG7 and WT anther. a Expression levels of ATGs in RNAi-SlATG7 and WT anthers. The expression levels were determined by RT-qPCR. Data are means ± SD (six biological replicates). b Autophagosomes stained with MDC in the tapetum of RNAi-SlATG7 and WT plants. The tissue was induced with exogenous rapamycin. Scale bars = 5 μm. c ATG8 protein levels in RNAi-SlATG7 and WT anthers. ATG8-PE and ATG8 are the lipidated and nonlipidated ATG8 forms, respectively. Internal control is actin. EUM, MLUM, and BP represent the stages of early uninucleate microspore, middle, and later uninucleate microspore, and binucleate pollen. d TEM observation of autophagosomes in the tapetum of RNAi-SlATG7 and WT plants. The autophagosomes are indicated by white arrows. Scale bars = 1 μm. Asterisks represent significant differences (Student's t-test, * P < 0.05 and * * P < 0.01).
analysed by RT-qPCR, and most of the autophagy-related genes were repressed in RNAi-SlATG7 lines compared with the WT plants (Fig. 8a). Then the MDC staining was used to detect autophagic activity in the RNAi-SlATG7 anthers. Low f luorescent signals were observed in the RNAi-SlATG7 tapetum, while strong f luorescent signals were observed in the WT plants (Fig. 8b).
To prove this result, WB was carried out to analyse the ATG8-PE formation in anthers. Accumulation of ATG8-PE was detected in WT anthers using the anti-ATG8a antibody, but the ATG8-PE band was hardly detected in RNAi-SlATG7 anthers (Fig. 8c). Then TEM was used to study the occurrence of autophagosomes in RNAi-SlATG7 anthers under the rapamycin treatment. The autophagosome vesicles of RNAi-SlATG7 anthers were hardly detected, while the autophagosome vesicles in the WT anthers were easily recognized (Fig. 8d). The data strongly supported that the downregulation of the SlATG7 gene inhibited the autophagy in tomato anthers.

Tapetum degradation affects pollen development in anther
The early or delayed tapetum degradation will cause abnormal growth of pollen [1,15,16]. Studies have found that several MYB transcription factors participate in pollen development by affecting the degradation process of tapetum [3,6,20]. Our study found that downregulation of the SlMYB72 gene inhibited the degradation of the anther tapetum in tomatoes. The pollen of RNAi-SlMYB72 plants showed abnormal morphology and reduced viability (Fig. 2). Overexpression of the SlMYB72 gene promoted the tapetum degradation and led to abnormal pollen development (Fig. 5). Downregulation of the SlATG7 gene delayed the anther tapetum's degradation, resulting in abnormal pollen development and reduced pollen viability (Fig. 3). The results prove that the SlMYB72 and SlATG7 genes can affect the tapetum degradation and pollen development, and the tapetum degradation plays important role in pollen development. The inhibition of the tapetum degradation will seriously interfere with the normal development of pollen.

Autophagy is essential for pollen development in tomatoes
Studies have provided evidence that autophagy plays an important role in the degradation of anther tapetum and pollen development [32,[34][35][36]. This study showed that downregulation of the SlMYB72 gene in tomatoes inhibited the expression of autophagy genes in tomato anthers. MDC staining and WB revealed that the autophagy in anthers of RNAi-SlMYB72 plants was significantly inhibited during the mononuclear microspore stage. The number of autophagosomes was decreased in the tapetum cells of the transgenic plants (Fig. 4). The results demonstrate the autophagy defect results in delayed tapetum degradation and abnormal pollen development in RNAi-SlMYB72 anthers, which are consistent with the results of Kurusu et al. [35]. In the OsATG7 mutant, the tapetum autophagy is inhibited, and the degradation of the tapetum layer is hindered, which eventually leads to pollen abortion.
We studied the function of SlATG7, which is homologous to OsATG7, in tomato pollen development. The MDC, WB, and anatomy results showed that downregulation of the SlATG7 gene in tomatoes inhibited the autophagy occurrence and delayed tapetum degradation in anther. The RNAi-SlATG7 plants exhibited abnormal pollen morphology and reduced pollen viability (Fig. 7).
The results indicate that the functions of the tomato SlATG7 gene in pollen development are relatively consistent with that of rice OsATG7 reported from Kurusu et al. [35]. The autophagy gene ATG7 in rice and tomato affects anther tapetum degradation and pollen development under normal conditions. Arabidopsis ATG mediates pollen development at high temperatures but isn't functional under normal conditions [36]. The Arabidopsis ATG7 mutant undergoes normal embryogenesis, germination, and seedling development under normal conditions, demonstrating that the ATG7 gene is not essential in the pollen development of Arabidopsis [38]. In tobacco, the downregulation of ATG7 affects the autophagy process in the anther but does not affect pollen development. These results indicate that the function of ATG7 in pollen development has functional differences among different species. The precise genomic editing of the ATG genes in different species will be carried out to identify the gene functions in tapetum degradation and pollen development in future research.

Transcriptional regulation of SlATG7 by SlMYB72
Under abiotic stress, the autophagy and ATG genes are transcriptionally regulated in many plants, such as tomato, pepper, rice, wheat, and Arabidopsis [39][40][41][42][43][44][45][46][47]. Some TFs have been reported to transcriptionally regulate autophagy and ATG genes. Heat shock factor 1A (HsfA1a) is involved in drought tolerance through regulating autophagy by directly transcriptional regulation of the ATG10 and ATG18f genes in tomatoes [44]. WRKY DNA-binding protein 33 (WRKY33) functions as a possible autophagy modulator by regulating several ATG genes under heat stress in Arabidopsis [48]. TGA motif-binding protein 9 modulates the autophagy and ATG gene expression by direct binding to their promoters under the stress conditions of sucrose starvation and osmosis in Arabidopsis [49]. Until now, the transcriptional regulation of the ATG gene by TF is found only in abiotic stress conditions. In the present study, downregulation of SlMYB72 decreased the expressions of ATG genes, while overexpression of SlMYB72 increased the expression of some ATG genes (Figs 4 and 5). Downregulation of SlMYB72 also inhibited the autophagy in tomato anther tapetum. The results of RT-qPCR showed that SlMYB72 and SlATG7 had similar expression patterns in anther developmental stage (Figs 3 and  6), indicating that SlMYB72 may directly regulate the expression of SlATG7. The EMSA and ChiP-qPCR confirmed that SlMYB72 could directly target the ACCAAC/ACCAAA motif of the SlATG7 gene. Our research with the transient dual-luciferase assay found that SlMYB72 can transcriptionally activate the SlATG7 expression (Fig. 6). Downregulation of SlATG7 affected the tapetum degradation and pollen development in tomatoes. Our data indicate that SlMYB72 transcriptionally regulates the autophagy and SlATG7 gene during pollen development in tomatoes. Our study provides a new cue that MYB TF can transcriptionally regulate autophagy and ATG genes during plant development. Further study directly targeting SlMYB72 to other ATG genes is necessary to understand the transcriptional regulation of autophagy in the plant.
In summary, both SlMYB72 and SlATG7 affect tapetum degradation and pollen development in tomatoes. SlMYB72 affects the autophagy process in the pollen tapetum of tomatoes. SlMYB72 directly targets the SlATG7 gene and actives its expression. SlATG7 also affects the autophagy process, tapetum degradation, and pollen development in tomatoes. The present study demonstrates that SlMYB72 affects the tapetum degradation and pollen development via transcriptional activation of SlATG7 and autophagy in anthers. The results explore the roles of autophagy in tapetum degradation and pollen development and provide valuable candidate genes for the production of male sterility.

Plant growth conditions
Tomato cultivar Micro-Tom (Solanum lycopersicum cv Micro-Tom) was used in this study. Seeds of Micro-Tom tomato were grown on soil in a standard greenhouse at 25 ± 2 • C, 16-h-light/8-h-dark cycle, 250 mol min −2 s −1 intense light, and 60% humidity.

Pollen viability and germination assays
Pollen assay was carried out as described previously [50]. Pollen grains were collected from the anthers at full bloom f lower and transferred to 2% 2,3,5-triphenyl-2 h-tetrazolium chloride (TTC) solution in the dark at 37 • C for 15 min. Then the pollens were analysed with a light microscope. Inviable pollens were unstained, while viable pollens were stained in red. The pollen grains were stained with I 2 -KI solution for 5 min and examined under a light microscope. Viable pollen grains were stained in the dark. The staining experiments were repeated at least three times.
The germination experiment was performed according to the report from Gan et al. [51]. Mature pollens were placed on slides with a germination medium containing 0.015% w/v boric acid, 13% w/v sucrose, and 1% w/v phytagel in the dark at 25 • C for 2 and 6 h.

Fluorescence observation of pollen tube growth
Pistils (2 DAP) were fixed with 3:1 ethanol:glacial acetic acid overnight. The tissues were softened with 8 M sodium hydroxide for 2 d and stained for 3 h with 0.05% aniline blue. The tissues were mounted in a drop of 50% glycerin and observed with a Leica DMRXA epif luorescence microscope.

Plasmid construction and plant transformation
For construction of the SlMYB72 overexpression vector, the ORF sequence of the SlMYB72 gene was inserted into pLP100 containing the caulif lower mosaic virus (CaMV) 35S promoter, which has been described in Wu et al. [37]. To construct the RNAi-SlATG7 vector, a 300-bp conserved sequence of SlATG7 was cloned into pCAMIBA2301 containing the 35S promoter. The construction was transferred into Agrobacterium tumefaciens GV3101 and transgenic tomato plants were obtained using A. tumefaciens-mediated method as previously described [37]. Transgenic lines of T3 generations were screened by 1/2 Murashige & Skoog (MS) medium containing 80 mg·L −1 kanamycin. In this study, sequences of primers are shown in Supplemental Table S1 (see online supplementary material).

GUS staining and RT-qPCR
For GUS staining, different tissues from SlMYB72 promoter-GUS plant were placed in a GUS staining solution containing 10 mM EDTA and 0.1 M sodium phosphate buffer at 37 • C for 24 h. Samples were fixed overnight in FAA solution, washed by ethanol, embedded in paraffin, and sectioned at 5 μm with a microtome. The sections were observed under a microscope.
The expression levels of SlATG7 in different tissues were analysed with a public tomato database (http://tomexpress. toulouse.inra.fr/). For RT-qPCR analysis of SlMYB72 and SlATG7 expression, RNA was extracted from various tissues using RNeasy Plant Mini Kit (Qiagen, Chongqing, China). One μg of RNA was reverse transcribed using HiScriptII Q Select RT SuperMix (Vazyme Biotech). RT-qPCR was performed using the SYBR Premix ExTaq kit (TaKaRa). Gene expression levels were calculated from the t values. Ubiquitin and actin genes were used as the internal standard to normalize the expression. The RT-qPCR was carried out with four biological replicates. Sequences of primers are shown in Table S1 (see online supplementary material).

Electrophoretic mobility shift assays
A full-length coding sequence of SlMYB72 was inserted into a pGEX-4 T-1 bacterial expression vector containing a glutathione S-transferase (GST) tag and the expression vector was transformed into Escherichia coli (E. coli) Rosetta (DE3) strain. The fusion protein was induced at 20 • C by 0.5 mM isopropyl-β-Dthiogalactopyranoside (IPTG) in the dark and purified through a GST-tagged purification kit (Clontech). Probes containing the AC-rich motif sequence from the gene's promoter were labeled using a lightshift chemiluminescent EMSA kit (Thermo Fisher Scientific). The unlabeled probe was used as a competitor and the AC-rich motif sequence was changed to AAAAAA sequence as the mutant probe in this experiment. The recombinant protein and probes (biotin-labeled, competitor, and mutated probes) were used to carry out the EMSA reaction according to the previously described [37]. The SlMYB72-GST bound probes were separated from the unbound probes by polyacrylamide gel electrophoresis.

Dual-luciferase transient expression assay
SlATG7 promoter was analysed using PlantCARE database (http://bioinformatics.psb.ugent.be/webtools/plantcare/html/). The dual-luciferase transient expression assay was carried out to analyse the interaction between SlMYB72 and SlATG7 genes. The full-length coding sequence of SlMYB72 was inserted into pGreenII 62-SK as an effector vector. Promoter sequence of SlATG7 was inserted into pGreen 0800-LUC as a reporter vector. The reporter and effector plasmids were cotransfected into leaves of tobacco by A. tumefaciens-mediated transient transformation. LUC and REN activities were determined using a dual luciferase assay kit (Promega). The experiments were biologically repeated at least six times. Sequences of primers are shown in Table S1 (see online  supplementary material).

ChIP-qPCR assays
The ChIP-qPCR assay was performed according to the previous study [37]. Transgenic plants overexpressing 35S-SlMYB72-FLAG were used for the assay in this study. The immunoprecipitated DNA fragments were determined using RT-qPCR for examining the relative enrichment of the promoter fragment. The specific primers used are listed in Table S1 (see online supplementary material).

Light microscopy and electron microscopy
For semi-thin sections, f lowers at different development stages were fixed overnight in 4% glutaraldehyde, dehydrated in ethanol, and embedded in epoxy resin. The samples were sectioned at 0.5 μm, and analysed under light microscopy. For SEM analysis, pollens were isolated from f lowers and observed under a Hitachi TM-1000. For TEM analysis, a FEI Tecnai T12 twin TEM was used to examine the anther tissues according to the method described by Yuan et al. [52].

MDC staining
For MDC staining, tissues were induced with exogenous rapamycin, fixed in 4% paraformaldehyde, and washed in PBS buffer by three times. Tissues were stained with 0.2 mM monodansylcadaverine (MDC) for 30 min and sectioned with a freezing microtome. Sections were analysed under a Leica TCS SP2 laser confocal microscope (excitement at 405 nm and detection at 445 to 465 nm).

Protein extraction and western blotting
The anther tissues of transgenic plants were ground in RIPA buffer and placed on ice for 35 min for protein extraction. The extracted proteins were added to the protein loading buffer, heated at 95 • C, and separated on a 15% SDS-PAGE gel for WB. Agrisera rabbit anti-Atg8 antibody and Abcam anti-actin antibody were used in this assay.

Data availability statement
All relevant data and figures in this study can be found within the article and its supporting materials.