Overexpression of SlMBP22 in Tomato Affects Flower Morphology, Fruit Set and Development

MADS-domain transcription factors have been claried as key regulators involved in proper ower and fruit development in angiosperms. B s genes, as members of the MADS-box subfamily, have been suggested to play an important role during the evolution of the reproductive organs in seed plants. Our knowledge about their effects on reproductive development in fruit crops like tomato (Solanum lycopersicum), however, is still unclear. Here, we found that the overexpression of SlMBP22 (SlMBP22-OE) resulted in considerable alterations regarding oral morphology, and affected the expression levels of several oral homeotic genes. Further analysis by yeast-two-hybrid assays demonstrated that SlMBP22 could form dimers with class A protein MACROCALYX (MC) and with SEPALLATA (SEP) oral homeotic proteins TM5 and TM29, respectively. In addition, pollen viability and cross-fertilization assays suggested that the defect in female reproductive development was responsible for infertility phenotype observed in the strong overexpression transgenic plants. The mild overexpression transgenic fruits were reduced in size, as a result of reduced cell expansion, rather than impaired cell division. Additionally, overexpression of SlMBP22 in tomato not only affected proanthocyanidin (PA) accumulation but also altered seed dormancy. Taken together, these ndings may provide new insights into the knowledge of B s MADS-box genes in ower and fruit development in tomato.

Floral development is a complex biological process and highly regulated by both the genetic background of plants and environmental factors (Fornara, et al. 2010). Previous studies have revealed that oral homeotic genes determining reproductive oral organ identities can be well understood via ABCDE model. (Coen and Meyerowitz 1991;Colombo, et al. 1995;Theissen 2001;Theissen and Saedler 2001).
Interestingly, all genes thus far identi ed in this model, except for APETALA2 (AP2), encode MIKC C -type proteins and belong to MADS-box transcription factor family (Parenicova, et al. 2003). With the deepening of research, more and more MADS-box genes have been identi ed and characterized as key regulators of tomato ower development. For instance, tomato class A gene MC is involved in the in orescence determinacy and the sepal development (Vrebalov, et al. 2002). Overexpression of FYFL in tomato presents longer sepals than wild-type (Xie, et al. 2014). Tomato class C gene TOMATO AGAMOUS 1 (TAG1), as the cognate homologs of Arabidopsis AGAMOUS (AG), participants in the regulation of stamen and carpel development as well as oral meristem determinacy (Pan, et al. 2010;). TOMATO AGAMOUS-LIKE1 (TAGL1), is the tomato ortholog of duplicated SHATTERPROOF of Arabidopsis, and also is the most closely related gene to TAG1, playing a key role in regulating carpels development (Vrebalov, et al. 2009). Tomato plants with up-regulated mRNA level of D-class MADS-box gene, Sl-AGL11, display carpel-like sepals ). SlMBP3 is the most closely related paralog of Sl-AGL11, and is notably expressed in the pistils. Silencing of SlMBP3 affects the development of seeds/placenta, suggesting that this gene speci es carpel/ovule identity (Zhang, et al. 2019). Two Eclass MADS-box genes, TM5 and TM29, have predominant functions in the development of oral organs and the determination of oral meristem identity in tomato.
Generally, early fruit development undergoes three major phases, namely, fruit set, cell division and cell expansion. Fruit development involves complex spatial and temporal regulation by the interplay of numerous biotic and abiotic factors, such as plant hormones, transcription factors, elongation factors, microRNA, RNA-binding proteins, ubiquitin-proteasome and so on (Hussain, et al. 2020). In addition to researches describing the in uence of MADS-box proteins on ower development, there have also been numerous studies highlighting the roles of MADS-box transcription factors in mediating various fruit morphologies, ripening and seed dispersal. Besides the role of tomato SHATTERPROOF 1 and 2 (SHP1, 2) ortholog TAGL1 in oral organ identities, this gene also functions in the fruit expansion and ripening process. TAG1 and TAGL1 are paralogous genes, the TAG1 silenced plants display smaller fruits than wild-type, which may be related to the reduction of pericarp thickness (Gimenez, et al. 2016;Vrebalov, et al. 2009). Transgenic plants with reduced tomato SEP1, 2 ortholog TM29 expression levels show phenotype with parthenocarpic fruits (Ampomah-Dwamena, et al. 2002). Class D gene Sl-AGL11, a close paralog of SlMBP3, is involved in the regulation of fruit quality and productivity ).
The phylogenetic sister clade of the class B genes has been termed B sister (B s ) (Becker, et al. 2002).
Relatively few members of this subfamily involved in the plant vegetative and reproductive growth have so far been characterized in different angiosperm species. It has been considered that B s genes makeavaluablecontribution to the regulation of reproductive development in seed plants. For instance, the B s MADS-box transcription factor, GORDITA (GOA), controls fruit size largely by modulating cell expansion in Arabidopsis (Prasad, et al. 2010). ABS/TT16/AGL32, the closest relative of GOA, is involved in the regulation of seed coat pigmentation and proanthocyanidin (PA) accumulation in the inner endothelial cell of the developing seeds in Arabidopsis (de Folter, et al. 2006;Nesi, et al. 2002;Xu, et al. 2017). It has been shown that ABS acts redundantly in the formation of endothelium with the D-class MADS-box protein SEEDSTICK (STK). The very few seeds observed in the Arabidopsis abs stk double mutant caused by the reduction of the number of fertilized ovules and the seed abortions (Mizzotti, et al. 2012). Similarly, FLORAL BINDING PROTEIN 24 (FBP24) is necessary for proper endothelium development in petunia (Petunia hybrida). Nevertheless, a mutant complementation experiment demonstrates that FBP24 fails to replace ABS/TT16, suggesting that there are functional conservation and divergence of the supposed orthologous genes in different angiosperm species (Becker, et al. 2002).
OsMADS30 T-DNA insertion plants display the alterations of plant size and architecture in rice, indicating that OsMADS30 may have evolved a new function and therefore is not a canonical B s gene (Schilling, et al. 2015).
Although we previously identi ed a tomato B s gene SlMBP22, homologous to Arabidopsis ABS/TT16 and petunia FBP24, participated in regulating tomato growth and tolerance to drought stress ), its role in reproductive development has not been fully explored. In this study, we found that transgenic tomato plants overexpressing SlMBP22 exhibited phenotypes related to defects in oral architecture, fruit set and development. Moreover, the underlying causes for these phenotypes were respectively analyzed at the morphological, statistical and molecular levels. Our data further expand the understanding of the functions of B s MADS-box proteins in the regulation of plant reproductive development.

Plant materials and growth conditions
Tomato (Solanum lycopersicum Mill. cv. Ailsa Craig) wild-type (WT) and transgenic plants were grown under normal greenhouse conditions (16-h-day/8-h-night cycle, 25°C/18°C day/night temperature, and 250 μmol m -2 s -1 light intensity). For gene expression analysis, owers were harvested at different developmental stages: -4D (4 d ahead of anthesis), -2D (2 d ahead of anthesis), anthesis and 2D (2 d after anthesis), and four-whorl mature oral organs (sepals, petals, stamens and pistils) were harvested at the anthesis stage. All samples were collected and promptly frozen in liquid nitrogen and then stored at -80°C until required.

Vector construction and plant transformation
For the construction of the SlMBP22-overexpressing (SlMBP22-OE) vector, the SlMBP22 full-length coding region was ampli ed by PCR with primers SlMBP22-F and SlMBP22-R, adding the Xba I and Sac I site to the 5´end and 3´end, respectively (Supplementary Table S1). The ampli ed SlMBP22 products were digested with Xba I and Sac I and then ligated into the plant binary vector pBI121 placed under the control of CaMV 35S promoter. The resulting binary vectors were transferred into S. lycopersicum variety AC ++ cotyledons, according to the transformation and regeneration methods as previously reported ).

Gene expression analysis
Total RNA was isolated using RNAiso Plus (Takara). The cDNA was synthesized using M-MLV Reverse Transcriptase Kit (Promega). Gene expression levels in different organs were evaluated by quantitative real-time PCR (qRT-PCR) using a CFX96 Touch™ Real-Time PCR Detection System (Bio-Rad). SlCAC (Solyc08g006960), a tomato housekeeping gene, was used as an internal control (Exposito-Rodriguez, et al. 2008). The analysis of relative expression levels was conducted using the 2 −ΔΔCT method (Livak and Schmittgen 2001). All primers for qRT-PCR are presented in Supplementary Table S1. Yeast two-hybrid assays The ORFs of SlMBP22, MC, TAGL1, SlMBP3, TM5 and TM29 were ampli ed by PCR using primers (Supplementary Table S1). The PCR fragments of SlMBP22 were cloned into the pGBKT7 vector to generate a bait construct BD-MBP22, and MC, TAGL1, SlMBP3, TM5 and TM29 were linked into pGADT7 vector to obtain prey constructs, namely, AD-MC, AD-TAGL1, AD-SlMBP3, AD-TM5 and AD-TM29, respectively. Different combinations of bait and prey vectors were co-transformed into Y2Hgold. The yeast two-hybrid assays were conducted as described previously (Tang, et al. 2020). The experiments were repeated three times.

Pollen viability assay
Pollen viability assay was tested by TTC staining method described previously ). Brie y, soaked pollen grains of the fully opened ower from the wild-type and transgenic plants in a 0.1% 2,3,5-triphenyl-2 h-tetrazolium chloride (TTC) solution at room temperature for 15 min, and then, stained pollen grains were observed and photographed under the microscope (Nikon E100). The experiments were repeated three times.

Cross assay
A cross assay was conducted according to the method described in a previous report (Shen, et al. 2019). The anthers of the WT and three OE plants (lines 18, 2 and 17) were carefully removed at 2 days ahead of anthesis, and the emasculated owers were labeled and bagged with small plastic bags to prevent natural pollination. Two days after emasculation, mature pollen grains of newly opened owers from the two OE lines, OE18 and OE2, were collected and brushed to the styles of the emasculated owers from the WT plants. Meanwhile, we brushed mature pollen grains from the newly opened WT owers to the styles of the emasculated owers of the transgenic lines: OE 18, 2 and 17, respectively.

Histological analysis
Histological analyses of sepal at the anthesis stage and fruit pericarp at the breaker stage from the WT and transgenic plants were processed following the method as described in our previous report ). The morphological observations of the para n sectioning were performed under a light microscope (Nikon E100). The number of cell layers and the mean mesocarp cell size were estimated according to our previous report ).

Vanillin Assay and PA extraction
Vanillin staining of mature tomato seeds was carried out using vanillin reagent as previously described (Chen, et al. 2013;Debeaujon, et al. 2000). Brie y, fresh seeds of the three independent mild transgenic OE lines and the WT plants at the 42d post anthesis (DPA) stage were harvested and soaked in a 1% vanillin solution at room temperature for 30 min. PAs of tomato seeds were determined by using the Vanillin-HCl method according to a previous report (Gao, et al. 2018;Mitsunaga, et al. 1998).

Seed germination assays
The seed germination assays were performed as previously described (Zhou, et al. 2019). In brief, the WT and T2 homozygous transgenic seeds were sown onto MS medium after surface sterilization, and then germinated in the dark at 25°C for 7 days. Seed germination rates were recorded on days 3, 5 and 7 respectively. The experiments were repeated three times.

Statistical analysis
Data presented in this report are means ± standard deviation from three independent repeats. The signi cant difference was assessed using Student's t-test (P < 0.05).

Results
Expression patterns of SlMBP22 B s MADS-box transcription factors have important regulatory functions during the evolution of the reproductive organs in seed plants (Becker, et al. 2002). Our previous report indicated that SlMBP22 may play essential roles in tomato owers, fruits and roots development based on its expression pattern analysis ( Supplementary Fig. S1) . To further explore the potential functions of SlMBP22 in tomato, we further evaluated its relative expression levels in owers at different development stages and in four-whorl oral organs at the anthesis stage by qRT-PCR analysis. The results showed that the transcript abundances of SlMBP22 were higher in four days ahead of anthesis and anthesis owers than that in two days before and after anthesis owers (Fig. 1a). In addition, SlMBP22 transcripts were mainly abundant in the pistils of oral organs, consistent with other plant B s MADS-box genes (Chen, et al. 2012) ( Fig. 1b). These results hinted that SlMBP22 may participate in the regulation of oral organ development, especially female reproductive organ development in tomato.

Overexpression of SlMBP22 alters tomato ower morphology and affects oral organ identity genes
We successfully generated ve independent transgenic OE lines that were used for further study ( Supplementary Fig. S2) , and observed that all of these lines displayed aberrant characteristics related to reproductive parts. The most evident phenotype was that all the SlMBP22 overexpression plants displayed smaller owers, especially the strong overexpression transgenic lines OE18 and OE2 (Fig. 1c, d). The measurements of the lengths of the four types of oral organs (sepals, petals, stamens and pistils) indicated signi cant reductions in the strong overexpression transgenic plants than those in the wild-type plants (Fig. 1e). Of particular note, the strong overexpression transgenic sepals were extremely abnormal in development, their color was a lighter green, the size was reduced by approximately 61% to 63% and then could not wrap the petals, when compared to the equivalent organs in the wild-type plants. Also, the petals of the SlMBP22-OE plants were more yellow than those of the WT, and had curly edges (Fig. 1c, d). To determine whether the light green sepal phenotype represented a change in total chlorophyll content, we extracted chlorophyll from sepals of fully opened owers and observed that the WT plants possessed higher chlorophyll levels compared with the strong SlMBP22-OE lines (Fig. 1f). Furthermore, the expression levels of genes related to chlorophyll biosynthesis and degradation, CHLH, CHLM, CAO1 and SGR1 (Hu, et al. 2011), were examined in sepals of both the wildtype and SlMBP22-OE transgenic plants by qRT-PCR analysis. The results showed that these genes were dramatically down-regulated in the transgenic plants ( Supplementary Fig. S3).  (Vrebalov, et al. 2002), was sharply downregulated in both sepals and pistils in the SlMBP22 overexpression plants compared to the WT. TAG1 and TAGL1, two tomato C-class oral organ identity genes, are orthologs of AGAMOUS (AG) and SHATTERPROOF1/2 (SHP1/SHP2) genes of Arabidopsis, respectively (Gimenez, et al. 2010 ;Vrebalov, et al. 2009). In the overexpression transgenic plants, TAG1 expression was dramatically upregulated in the pistils, but the TAGL1 expression was greatly downregulated in both the stamens and pistils when compared with the WT (Fig. 2b, c). The transcript level of SlMBP3, one member of the class D MADS-box genes and speci es carpel/ovule identity, was evidently increased in the transgenic pistils compared to the wild-type (Fig. 2d). Two E-class genes, TM5 and TM29, participate in the maintenance of oral meristem identity and the regulation of oral organ development (Ampomah-Dwamena, et al. 2002;). Relative to WT, the TM5 showed increased expressions in the transgenic sepals, stamens and pistils (Fig. 2e). Fig. 2f showed that the transcript for the TM29 was much lower in the transgenic pistils than that in the wildtype. Our results suggested that overexpression of SlMBP22 leading to the morphological alterations of owers might be attributed to the changes in the expressions of the oral organ identity genes.

MADS-box
Numerousresearches have demonstrated that MADS-box transcription factors carry out their functions in ower development by forming dimers or higher-order complexes (Tonaco, et al. 2006). Subsequently, a yeast two-hybrid assay was performed to assess the self-activation of pGBKT7-SlMBP22 and to con rm the interactions between the SlMBP22 and other oral homeotic MADS-box proteins. As shown in Fig. 2g, no autoactivation activity was detected on SD/-Leu-Trp-Ade-His and SD/-Leu-Trp-Ade-His containing X-α-Gal plates. Besides, SlMBP22 could physically interact with MC, TM5 and TM29 but not with TAGL1, SlMBP3 in yeast. These results suggested that SlMBP22 may carry out its role in ower development by forming dimers with MC, TM5 and TM29, respectively.
Overexpression of SlMBP22 results in reduced fecundity in tomato The relatively high mRNA accumulation of SlMBP22 in tomato fruit suggested the possibility of additional functions in fruit development ( Supplementary Fig. S1) ). Thus, the effects of the overexpression of the SlMBP22 on fruit development were then investigated in the transgenic OE lines. We found another remarkable phenotype was that the strong overexpression lines (OE18 and OE2) could not bear fruit, whereas the mild overexpression transgenic lines (OE17, OE12 and OE14) showed reduced fruit size and produced fewer seeds. Occasionally, the strong overexpression line OE2 produced a much smaller fruit, while could not expand as normally as the wild type (Fig. 3a, b).
To further compare fruit development, some parameters were measured, including fruit weight, fruit volume, fruit diameter, pericarp thickness and the number of seeds in the B4 (4 d after breaker) stage fruits. As shown in Fig. 3c-e, the mild overexpression of SlMBP22 resulted in signi cant reductions in fruit weight, fruit volume and fruit diameter when compared with those in the control fruits. Moreover, the pericarp thickness of fruits was also measured, and the result showed that the OE lines showed thinner pericarp tissues than WT plants (Fig. 3f). Additionally, compared to WT, the seed number per fruit of the mild overexpression transgenic plants were reduced by approximately 62% to 68% (Fig. 3g). Previous studies indicate that there are close relationships between fruit size and seed number per fruit (Hussain, et al. 2020), and then we speculate that the notable differences in fruit size between the mild SlMBP22-OE lines and WT plants might be attributed to signi cantly reduced pericarp thickness and seed yield.
Additionally, to further investigate whether the reduced fertility of transgenic owers is a result of the defects in either the male or the female parts, cross-pollination experiments were conducted between wild-type and transgenic plants. The results showed that failed fertilization occurred by crosses between the wild-type pollen and strong transgenic pistils (data not shown), while seeds were produced successfully when mild overexpression transgenic owers and wild-type owers were respectively crossed with wild-type pollen and strong overexpression transgenic pollen (Fig. 3h). Besides, pollen viability was detected by TTC staining, and the result suggested that the strong overexpression transgenic pollen stained similarly to the WT pollen (Fig. 3i), hinting that the pollen viability may not be affected in the transgenic plants. Moreover, less viable pollen grains were observed in the strong overexpression transgenic owers than in the WT owers (data not shown). Subsequently, the relative transcript accumulation of pollen development-related genes SlCRK1 (Kim, et al. 2014), SlPRALF (Covey, et al. 2010), LePRK3 (Kim, et al. 2002), SlPMEI (Kim, et al. 2013) were also examined by qRT-PCR assay. Intriguingly, all of these four gene transcripts were consistently reduced in the strong SlMBP22overexpression lines compared to those in the non-transgenic plants (Fig. 3j-m), which were likely to be associated with the reduction of pollen grains in the transgenic lines with a strong SlMBP22 overexpression. Thus, we propose that the infertility phenotype observed in the SlMBP22-OE transgenic tomato plants is probably attributed to low pollen production and defect in female reproductive development.
Overexpression of SlMBP22 affects auxin signalling-related genes Auxin plays critical roles in regulating fruit development, including fruit set and growth, ripening and abscission (Pattison, et al. 2014). Recently, we demonstrated that the mild overexpression of SlMBP22 led to reduced plant height by affecting gibberellin (GA) and auxin homeostasis. . In this study, the arti cial enhancement of SlMBP22 resulted in infertility phenotype in the strong overexpression lines, and then we speculated that it was also related to the alteration of auxin signalling. Subsequently, qRT-PCR assay was conducted to further investigate the expression levels of auxin pathway-related genes in the wild-type and strong overexpression transgenic ovaries. It was found that the transcripts of an auxin biosynthesis gene ToFZY5 (Exposito-Rodriguez, et al. 2011), an auxin response gene (ARF3) (Zouine, et al. 2014), three Aux/IAA transcription factor genes (IAA13, IAA14 and IAA29) (Audran-Delalande, et al. 2012), were greatly increased in the SlMBP22 strong overexpression transgenic ovaries at the anthesis stage, compared with the WT ovaries (Fig. 4a-e). By contrast, transcripts of an AUX/LAX gene LAX1, and three PIN genes (PIN1, PIN2 and PIN4) (Pattison and Catala 2012), respectively encoding auxin in ux and e ux transport proteins, were sharply decreased in the transgenic ovaries than in the WT (Fig. 4f, g and I). Relative to WT, PIN2 gene expression was upregulated in the transgenic ovaries (Fig. 4h). Based on the results described above, we inferred that the overexpression of SlMBP22 may alter tomato productive development via disturbing auxin signaling.

Overexpression of SlMBP22 affects owers and fruits size mainly by inhibiting cell expansion
Plant organ growth is controlled by multiple regulatory factors that coordinate cell proliferation and cell expansion (Anastasiou and Lenhard 2007;Horiguchi, et al. 2006). In our work, the SlMBP22-OE tomato plants exhibited reduced ower size, particularly rst-whorl sepals, and smaller fruits with thinner fruit pericarp (Fig. 1c-e and Fig. 3b, f). Therefore, anatomical analyses were performed to investigate the cytological differences between the WT and transgenic sepals and fruit pericarps. Obviously, the cells in the sepal and pericarp respectively from the strong and mild SlMBP22 overexpression transgenic lines were much smaller than those in the wild-type plants (Fig. 5a, b). Compared with the WT, the transgenic fruit pericarps contained slightly reduced cell layers, while the reduction in mean cell size of the mesocarp cell of the transgenic fruits reached up to a 60 % difference (Fig.5c, d). Furthermore, we analyzed the transcript levels of genes associated with plant cell division and cell expansion in the in orescences from the WT as well as the strong SlMBP22-OE plants by qRT-PCR. CDKA1, involved in the progression of the cell cycle (Czerednik, et al. 2015;Czerednik, et al. 2012), showed no signi cant difference in the mRNA accumulation between the transgenic and non-transgenic plants (Fig. 6a). The transcripts for three cyclin genes, SlCycA3;1, SlCycB1;1 and SlCycB2;1 were not clearly affected (Fig. 6b-d). In contrast, the expression levels of cell expansion-related genes, SlEXP1 (Perini, et al. 2017), LeEXP2 (Caderas, et al. 2000) and LeEXP8 (Chen and Bradford 2000), were distinctly repressed in the SlMBP22-overexpressing plants ( Fig. 6e-g). FUL2, as a member of the MADS-box transcription factor family, affects style abscission and cell expansion ). In the SlMBP22-OE plants, FUL2 was strongly upregulated when compared with its respective expression in the WT (Fig. 6h). Overall, these ndings support the possibility that SlMBP22 up-regulation leads to the alterations in tomato owers and fruit size are, atleastinpart, due to the reduced cell expansion, rather than impaired cell division.
Mild overexpression of SlMBP22 promotes proanthocyanidin accumulation and affects seed germination The tt16 seeds are yellow in color and the PA accumulation was restricted to the chalazal bulb and the micropylar end in the mutant seed coat, while the ectopic expression of TT16 produced brown seeds as a result of ectopic PA biosynthesis (Nesi, et al. 2002). According to our observations, the SlMBP22 overexpression transgenic seeds had a dark brown color (Fig. 7a). To further explain the phenotype regarding the pigmentation of the seed coat, a vanillin assay was conducted and indicated that the mild OE transgenic seeds may accumulate more PA than WT, and then we decided to measure the PA content.
As expected, the OE plant seeds possessed higher PA levels than wild-type plants (Fig. 7b). The overaccumulation of pigments in the seed coat has a negative effect on seed germination (Debeaujon, et al. 2000). Subsequently, a seed germination assay was performed to try to detect the germination energy of the transgenic tomato seeds. The results exhibited lower seed germination rates in seeds from transgenic lines compared with those from the non-transgenic plants (Fig. 7c, d), implying that mild overexpression of SlMBP22 may inhibit the germination ability of the transgenic tomato seeds.  ). Furthermore, we found that the overexpressing SlMBP22 transgenic tomato plants exhibited defects in reproductivegrowth and development, including morphological alterations of owers, reduced fruit set and growth, and abnormal seed color. These data indicate that the functional conservation and diversity of TT16 genes in different plant species.

Discussion
B sister genes are predominantly expressed in female reproductive organs, suggesting that this subfamily is involved in the evolution of the reproductive organs in seed plants (Becker, et al. 2002). Here, owers of the SlMBP22 overexpression plants showed considerable changes regarding oral organs size, sepal and petalcolor as compared with the wild-type, suggesting that the overexpression of SlMBP22 in tomato may affect the development of ower, following what would be expected for a typical B sister gene. To obtain further insights into the potential molecular regulation mechanism explaining the phenotypes associated with oral organ development, several MADS-box genes related to ower development were tested by qRT-PCR analysis in mature oral organs of both the WT and overexpression plants. These results revealed that overexpression of SlMBP22 caused alterations in the expression levels of these genes such as the close tomato of SHP1/SHP2 (TAGL1), SEP1/2 (TM29), AG (TAG1) and SEP3 (TM5) AP1 (MC) and STK (SlMBP3). These results suggest that the impacts of SlMBP22 on ower development may be associated with other transcription factors. It has been proposed that MADS-box transcription factors carry out their functions in oral organ formation and identity or other developmental processes by a complex network of protein-protein and protein-DNA interactions (Tonaco, et al. 2006). In the case of Arabidopsis, ABS/TT16 can form dimers with SEPALLATA (SEP) oral homeotic proteins and form higher-order complexes that also include the SEEDSTICK (STK) or SHATTERPROOF1/2 (SHP1, SHP2), which are veri ed by yeast-two-hybrid and three-hybrid assays, respectively (Kaufmann, et al. 2005).
Hence, the formations of dimers and higher-order complexes may have a key role in regulating plant growth and development among MADS-box genes. In our study, a yeast two-hybrid experiment showed that SlMBP22 could physically interact with a MADS-box transcription factor, MC, which is known as a regulator of sepal size (Vrebalov, et al. 2002), indicating that SlMBP22 and MC can potentially form a dimer and then may explain the sepals with extremely reduced size observed in the SlMBP22-OE lines compared with WT. Meanwhile, protein-protein interactions were also observed in the yeast two-hybrid assay between SlMBP22 and another two SEP MADS-box proteins, TM5 and TM29, which are previously veri ed to mediate organ differentiation of the inner three whorls of tomato owers (Ampomah- Dwamena, et al. 2002;. It is possible that overexpressing SlMBP22 can affect ower morphology at least partly be interpreted as the consequence of forming dimers, trimers, or even tetramers with other oral homeotic proteins in tomato plants. The phases of fruit initiation and development have been considered to be the continuation of the oral developmental program, and MADS-box transcription factors play considerable and multiple functions during ower, fruit and seed development (Busi, et al. 2003). For instance, transgenic tomato plants with reduced TAG1 expression levels exhibit defects in stamen and carpel identity, while overexpression lines display alteration in the rst whorl, male and female sterility, and parthenocarpic fruit ). Another C class homeotic gene TAGL1 plays an important role in both regulating eshy fruit expansion and ripening processes. (Vrebalov, et al. 2009). The down-regulation of SEP homolog TM29 leads to infertile stamens and ovaries, parthenocarpic fruit, and green-colored petals and stamens which suggest a partial conversion of these organs into sepals (Ampomah-Dwamena, et al. 2002) Besides modifying the ower morphology, we also found that the arti cial enhancement of SlMBP22 affected fruit set and growth, and the strong SlMBP22 overexpression transgenic plants bore no fruit, while the smaller fruits with fewer seeds were observed in the mild overexpression transgenic lines compared to those in the WT.
The strong SlMBP22-OE transgenic owers were manually cross-pollinated with wild-type pollen, which failed to produce fruit, whereas normal seeds could develop when strong overexpression transgenic pollen was used to pollinate WT styles. In addition, TTC staining and qRT-PCR assays were carried out and the results showed that the pollen viability was not affected in SlMBP22-OE owers, but the strong overexpression of SlMBP22 led to reduced viable pollen grains. These data indicated that the strong overexpression of SlMBP22 caused female sterility and disturbed mature pollen formation, thus resulting in reduced fertility.
The nal size of an organ is controlled by two phases of growth, namely, cell proliferation and subsequent cell expansion (Anastasiou and Lenhard 2007;Horiguchi, et al. 2006). SlMBP22 overexpressing plants showed smaller owers and fruits with thinner pericarp and fewer seeds than those of the WT. qRT-PCR analysis was conducted and the results revealed that the expression of several genes involved in cell expansion, including SlEXP1, LeEXP2 and LeEXP8, was greatly down-regulated in the OE plants, but the transcripts for cell cycle genes such as CDKA1, SlCycA3;1, SlCycB1;1 and SlCycB2;1 displayed no obvious differences, when compared to their respective expression in the WT. These expression data agreed with the ndings of the microscopic analysis, implying that the overexpression of SlMBP22 affects ower and fruit size is likely to originate from the reduced cell expansion but not the impaired cell division. There have been reports highlighting the in uence of phytohormones especially auxin and GA ( shown that the transcript levels of FBP24/ SlMADS29 (renamed here as SlMBP22) were greatly reduced after 2,4-D treatment in tomato ovaries by qRT-PCR analysis (Hu, et al. 2018;Tang, et al. 2015), indicate that SlMBP22 may participate in the regulation of fruit development by mediating auxin signalling. Indeed, our qRT-PCR assay showed that the mRNA accumulations of several genes related to auxin signalling were altered in the transgenic ovaries. In addition, auxin signalling is also important to regulate oral organ size. For instance, tomato MADS-box gene SlMBP21 negatively regulates cell expansion to affect sepal size, which was mediated by ethylene and auxin ). Thus, it is likely that the overexpression of SlMBP22 affected tomato reproductive development including floral organ size, fruit initiation and growth via disturbing auxin signalling, which is consistent with its effect on tomato plant vegetative development ).
Proanthocyanidins (PAs), as one of the main avonoids in Arabidopsis seeds (Routaboul, et al. 2012), play a crucial role in modulating seed dormancy and longevity during storage (Debeaujon, et al. 2000).
Once oxidized, PAs confer brown pigmentation in the mature seed coat (Devic, et al. 1999). Arabidopsis ttg1 mutant seeds have a yellow color, and appear reduced seed dormancy can be ascertainedby a highergermination rate (Debeaujon, et al. 2000). Arabidopsis TT16 gene participates in the control of seed coat pigmentation and proanthocyanidin (PA) accumulation in the endothelium of developing seeds (Deng, et al. 2012;Nesi, et al. 2002;Xu, et al. 2017). In our study, SlMBP22-OE transgenic seeds were dark brown in color, and with more PA accumulation and exhibited lower germination than non-transgenic seeds, indicating that the overexpression of SlMBP22 in tomato not only affected biosynthesis of PAs but also altered seed dormancy. As reported, endogenous plant avonoids not only affect seed germination and dormancy, but also play an important role in regulating cellular auxin e ux and consequent polar auxin transport. Auxin transport is elevated in tt4 (no avonoid production) but reduced in tt7 and tt3 mutants (which accumulate excess avonols) (Peer, et al. 2004). Additionally, previous reports indicate that seeds could produce or deliver auxins to the surrounding fruit tissues and then promote fruit expansion (Ariizumi, et al. 2013;Gillaspy, et al. 1993). The agl62 mutant shows impaired transport of auxin was responsible for seed abortion (Figueiredo, et al. 2016). Our data demonstrated that overexpression of SlMBP22 resulted in altered expression of a series of genes related to auxin transport. Based on the results described above, we infer that SlMBP22 may act as a key regulator that affects ower development, seed development, fruit set and growth via affecting auxin signalling.
In summary, we aimed to investigate the effects of SlMBP22 overexpression on the tomato reproductive development regarding owers, fruits and seeds in this study. Our data demonstrate that the overexpression of SlMBP22 not only results in morphological alterations of owers but also affects fruit set, fruit size and seed pigmentation. Morphological, physiological,andmolecularanalyses have been carried out to preliminarily elucidate the causes related to the defects of  replicates. Asterisks indicate signi cant differences compared with WT (P < 0.05). g Yeast two-hybrid assay for the SlMBP22 with MC, TAGL1, SlMBP3, TM5 and TM29. The protein interaction was detected on synthetic de ned quadrupledropout (SD/-Leu-Trp-Ade-His) medium (middle) and which also containing X-α-Gal (SD/-Leu-Trp-Ade-His & X-α-Gal) medium (right) after the yeast cells had been screened and had positive growth on SD double dropout (SD/-Leu-Trp) medium (left). pGADT7-T and pGBKT7-53, positive control. pGADT7-T and pGBKT7-Lam, negative control. The three columns (from left to right) correspond to three concentration gradients (10-1, 10-2, and 10-3). Combinations of empty bait and empty prey vector, and autoactivation assay with no yeast growth on SD/-Leu-Trp-Ade-His medium and SD/-Leu-Trp-Ade-His & X-α-Gal medium.  Overexpression of SlMBP22 affects auxin signalling-related genes. The relative expression levels of genes related to auxin biosynthesis (a), auxin response (b), Aux/IAA gene family (c-e), auxin in ux (f) and e ux transport (g-i). Error bars indicate the standard error between three replicates performed. Asterisks indicate signi cant differences (P < 0.05).  Asterisks indicate signi cant differences between WT and OE plants (P < 0.05).

Figure 7
Phenotypes of the wild-type and SlMBP22-OE transgenic seeds. a A comparison of the seeds treated with vanillin reagent (bottom) and untreated seeds (upper). b Contents of proanthocyanidins of seeds from the wild-type and SlMBP22-OE plants. Results represent mean ± SE from three biological replicates.
Asterisks indicate statistically signi cant differences (P < 0.05). c Germination phenotype of WT and SlMBP22-OE transgenic seeds. d Germination rates of seeds on MS medium. Seed germination was scored on the third, fth and seventh day after sowing. The data are the means ± SE from three independent experiments with about 30 seeds per replicate.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.