MBW complexes impinge on anthocyanidin reductase gene regulation for proanthocyanidin biosynthesis in persimmon fruit

MBW protein complexes containing MYB, bHLH and WD40 repeat factors are known transcriptional regulators of secondary metabolites production such as proanthocyanidins and anthocyanins, and developmental processes such as trichome formation in many plant species. DkMYB2 and DkMYB4 (MYB-type), DkMYC1 (bHLH-type) and DkWDR1 (WD40-type) factors have been proposed by different authors to take part of persimmon MBW complexes for proanthocyanidin accumulation in immature fruit, leading to its characteristic astringent flavour with important agronomical and ecological effects. We have confirmed the nuclear localization of these proteins and their mutual physical interaction by bimolecular fluorescence complementation analysis. In addition, transient expression of DkMYB2, DkMYB4 and DkMYC1 cooperatively increase the expression of a persimmon anthocyanidin reductase gene (ANR), involved in the biosynthesis of cis-flavan-3-ols, the structural units of proanthocyanidin compounds. Collectively, these data support the presence of MBW complexes in persimmon fruit and suggest their coordinated participation in ANR regulation for proanthocyanidin production.

employed by plants to drive frugivore-dependent dispersal of seeds, and to improve deastringency treatments and crop management with the aim to reduce costs and increase the sustainability of persimmon production.
The pathway of PAs biosynthesis has been genetically dissected by the analysis of different seed mutants in Arabidopsis thaliana, barley and maize among other species 15 . PAs are formed by condensation of trans-and cis-flavan-3-ols units, synthesized respectively by stereospecific leucoanthocyanidin reductase (LAR) and anthocyanidin reductase (ANR) enzymes 16,17 . These PA biosynthetic activities and genes are essentially conserved in persimmon, with some regulatory particularities 18 . In persimmon fruit, ANR gene is much more expressed than its counterpart LAR, consistent with the higher content of cis-flavan-3-ols stereoisomers in PA composition 19 . In addition, ANR is strongly repressed in advanced steps of fruit maturation, concomitantly with tannin decrease, which points to a role of ANR as a major integrative target of PA regulatory pathways in persimmon.
In Arabidopsis, regulation of the ANR orthologous gene BANYULS (BAN) and PA accumulation in seed coat requires the concerted action of TRANSPARENT TESTA2 (TT2), TRANSPARENT TESTA8 (TT8) and TRANSPARENT TESTA GLABRA1 (TTG1), encoding respectively a R2R3-MYB transcription factor, a basic helix-loop-helix (bHLH) transcription factor and a WD40-repeat (WDR) protein 20 . These regulatory factors form a ternary complex named MYB-bHLH-WD40 (MBW) that may invoke the participation of alternative MYB and bHLH components for the regulation of particular steps of PA and anthocyanin biosynthetic pathways 21,22 . MBW complexes also contribute to PAs production in edible fruits of crop plants such as grapevine 23 , strawberry 24 , apple 25 and persimmon 26 .
In persimmon, DkMYB2 and DkMYB4 genes cause an altered pattern of PAs accumulation and expression of biosynthetic enzymes when misexpressed, and increase ANR promoter transcriptional activity in transient reporter assays when combined with a heterologous bHLH factor from Arabidopsis 27,28 . Interestingly, DkMYB2 and DkMYB4 specifically recognize different MYB-binding cis-elements by electrophoretic mobility shift assays, arguing for certain degree of subfunctionalization 28 . On the other hand, a persimmon bHLH gene named DkMYC1 is underexpressed in non-astringent cultivars, following an expression pattern in fruit similar to DkMYB4 29 . DkMYC1 protein interacts with DkMYB2 and DkMYB4, and these, in turn, interact with the WD40-repeat protein DkWDR1 by two-hybrid analysis, suggesting the conserved participation of MBW complexes in PA synthesis regulation in persimmon fruit 26 . In this study, we go further on the cytological and molecular characterization of MBW complex in persimmon by approaching the subcellular localization, protein interaction and transcriptional regulation effects of these MYB (DkMYB2 and DkMYB4), bHLH (DkMYC1) and WD40 (DkWDR1) components.

Results
Subcellular localization of MBW complex components. As transcription factors belonging to a hypothetical regulatory protein complex, DkMYB2, DkMYB4, DkMYC1 and DkWDR1 are expected to co-localize temporarily in the cell nucleus where they interact and perform their regulatory role within the framework of a developmental programme. The subcellular localization of these four putative MBW components in persimmon has been elucidated by transient expression in Nicotiana benthamiana leaves of the four corresponding genes fused to enhanced green fluorescent protein (eGFP) gene. According to the localization of specific nucleus and nucleolus markers, DkMYB2, DkMYB4 and DkMYC1 show differential nucleus/nucleolus partitioning when eGFP fusion is either at the N-terminal (Nt) or C-terminal (Ct), suggesting that eGFP position affects protein targeting (Fig. 1). Protein fusions with higher abundance in the nucleolus are DkMYB2-eGFP, DkMYB4-eGFP and eGFP-DkMYC1, whereas DkWDR1-eGFP shows appreciable presence in the cytoplasm in spite of its predominant localization in the nucleus. Overall, these four transcription factors show preferential localization in the nuclear compartment, and consequently physical interactions at the protein level occur most plausibly in the nucleus.
In vivo interaction of DkMYB2, DkMYB4, DkMYC1 and DkWDR1. Physical protein interactions among members of MBW complex involved in PAs accumulation in persimmon fruit has been only previously tested by the two-hybrid system in the yeast model, however no additional in planta evidences have been obtained on the formation of this complex. To asses this issue, we have assayed pair-wise interactions between DkMYB2, DkMYB4, DkMYC1 and DkWDR1 factors by bimolecular fluorescence complementation (BiFC) and transient expression in N. benthamiana. The Nt and Ct fragments of the yellow fluorescent protein (YFP), required for reassembling of the fluorescent reporter, have been tested on both Nt and Ct sides of each transcription factor. Positive protein interactions by BiFC are shown in Fig. 2a. According to these results, DkMYB4 and DkMYC1 are able to interact with the rest of the factors and with themselves, and DkWDR1 does not reconstitute YFP fluorescence when paired with DkMYB2 and itself. This reproduces previous yeast two-hybrid results with few exceptions (Fig. 2b). Particularly, DkMYC1 homodimerization was not observed, and DkMYB2 self-interaction was not assayed due to autoactivation issues in former yeast-two hybrid experiments. Moreover, DkMYB2-DkWDR1 and DkMYC1-DkWDR1 interactions have been exclusively detected by yeast-two hybrid and BiFC analysis, respectively. A western analysis of transiently transformed leaves confirms that DkWDR1 and DkMYB2 were successfully co-expressed in different construct combinations and hence, the absent interaction of DkWDR1 with itself and DkMYB2 are not due to deficient protein synthesis or accumulation ( Supplementary Fig. S1). Overall, these data support the biochemical ability of these factors to associate in a putative MBW complex in planta.
DkANR expression correlates well with PA accumulation and astringency. PA content has been measured at different points of fruit development in 'Hachiya' cultivar, starting in July and finishing in September, after external colour change has been initiated and before ripening leads to fruit softening and natural deastringency (Fig. 3a). During that period fruit average weight increases three-fold and the percentage of PA decreases concomitantly (Fig. 3a). Fruit soluble PA content is a balance between PA biosynthesis and insolubilization. PA insolubilization is mediated by acetaldehyde accumulation as a result of ripening and deastringency treatments 10,11 . Acetaldehyde content in these 'Hachiya' fruit samples reaches values around 0.1 mg per 100 ml of juice (Fig. 3b), which is by far lower than acetaldehyde produced in stored fruit and fruit treated for deastringency 3,30 . Thus, PA insolubilization due to acetaldehyde accumulation is not expected to contribute significantly to reduce soluble PA level in our samples, and PA content is mostly dependent on its biosynthesis rate. When representing the total estimated amount of PAs per fruit instead of its relative percentage, PA amount remains almost unchanged during the whole interval, with the exception of an initial increase in July samples (Fig. 3b). As we www.nature.com/scientificreports www.nature.com/scientificreports/ consider that PA reduction by acetaldehyde-dependent insolubilization is relatively low, the observed decrease in relative PA content in Fig. 3a must be mostly due to a growth dilution effect, and the rate of PA biosynthesis is expected to be also low in this period.
As anthocyanidin reductase encoded by ANR gene has been postulated to perform a key role in PA biosynthesis, we have measured DkANR expression in 'Hachiya' fruit samples by qRT-PCR. DkANR transcript sharply decreases until a 0.01-fold change during the whole period (Fig. 3c), which cannot be explained by just growth dilution effects. On the contrary, it indicates a strong transcriptional repression in advanced developmental stages. The higher expression of DkANR in initial samples is in close agreement with the concurrent increase in total PA content per fruit (Fig. 3b). Then PA content remains steady in concordance with DkANR down-regulation (Fig. 3b,c). Such a positive correlation between DkANR expression and astringent-responsible PAs is confirmed in the low-PA non-astringent cultivar 'Hana Fuyu' , which shows a constantly low DkANR expression during fruit development stages (Fig. 3c).
Other PA biosynthetic regulators, such as DkMYB4 and DkMYC1, reproduce well this low expression profile in 'Hana Fuyu' and other non-astringent cultivars, as shown in previous studies 26,27,29 . However, DkMYB2 relative expression decays in a similar way in both astringent and non-astringent cultivars (Fig. 3d), suggesting a common DkMYB2 transient expression in N. benthamiana leaves increases ANR promoter transcriptional activity three-fold, whereas DkMYB4, DkMYC1 and DkWDR1 do not modify it significantly (Fig. 4a). Interestingly, any combination of two or three elements containing both DkMYC1 and a MYB gene (DkMYB2 or DkMYB4), strongly increases LUC/REN ratio, being highest when DkMYC1 and DkMYB2 are co-expressed. On the contrary, DkWDR1 does not improve ANR promoter expression under any gene combination.
As DkMYB4 and DkMYC1 genes are similarly down-regulated during fruit development concomitantly with PA reduction, and are differentially expressed in a non-astringent cultivar 26,27,29 , their promoters have been also cloned and tested by dual luciferase assays in N. benthamiana. MBW factors assayed in the experiment do not increase the activity of DkMYB4 nor DkMYC1 promoters, however DkWDR1 transient expression associates with a significant decrease in LUC expression driven by DkMYB4 promoter (Fig. 4b,c).

Discussion
DkMYB2, DkMYB4, DkMYC1 and DkWDR1 have been postulated to co-regulate the expression of PA biosynthesis genes in persimmon as a complex 26 , and hence a coordinated nuclear co-localization of them is expected. Related components of MBW complexes in other species have been found located in the nucleus (TT2) 31 , or partitioned in nucleus and cytoplasm (TTG1 and VvMYC1) 23,32 . In this study, DkMYB2, DkMYB4, DkMYC1 and DkWDR1 have been mostly localized in the cell nucleus, but DkMYB2, DkMYB4, DkMYC1 show a differential nucleus-nucleolus partitioning depending on the Nt or Ct position of the eGFP fusion. To our knowledge, this has not been previously observed in other MBW factors, although in most of these cases, only one Nt or Ct fusion is assayed. Indeed, a systematic approach in yeast has shown that a high percentage of proteins display different subcellular localization when GFP is tagged at either the Nt or Ct 33 .
Transient expression in N. benthamiana and BiFC analysis support the ability of DkMYB2, DkMYB4, DkMYC1 and DkWDR1 proteins to interact with each other in vivo, which in fact reinforces previous two-hybrid data in the yeast Saccharomyces cerevisiae 26 . The only combinatorial interactions not confirmed by BiFC are DkMYB2-DkWDR1 and DkWDR1 with itself. DkMYB2-DkWDR1 interaction was observed in a previous two-hybrid study, thus only homodimerization of DkWDR1 is not sustained on experimental evidences. These BiFC and yeast two-hybrid interaction data are compatible with a multitude of possible combinations and sizes These BiFC and subcellular localization results strongly support the formation in vivo of protein complexes comprising at least several of these factors, but conclusive functional evidences about their coordinated recruitment to modify the expression of PA biosynthetic genes in persimmon fruit are scarce. Anthocyanidin reductase is the main enzyme specifically involved in PA production in persimmon, and its coding ANR gene is considered a major target of transcriptional regulation 19 , being consequently a proper candidate gene for studying PA biosynthesis regulation. Previous dual luciferase assays in N. benthamiana have shown that DkMYB2 and DkMYB4 increase the activity of ANR promoter when co-expressed with Arabidopsis AtEGL3 gene coding for a bHLH protein involved in regulation of the flavonoid pathway, but not when expressed individually 28 . On the contrary, in our hands, DkMYB2 is able to increase ANR promoter activity three-fold in the absence of other factors, which suggests that DkMYB2 does not require a complete MBW complex to enhance, at some level, ANR expression and consequently improve PA production (Fig. 5). Interestingly, the ectopic expression of DkMYB2 in kiwifruit calluses induces PA accumulation without additional MBW components 28 . DkMYB2 expression is regulated by fruit maturation factors in persimmon that markedly reduce it in advanced stages of development, similarly to ANR, DkMYB4 and DkMYC1, but in contrast to these genes it seems not to be impaired in ast non-astringent mutants. Unexpectedly, DkMYB2 has only a minor contribution to PA accumulation in these mutants, which is most likely due to its low expression level in fruit in comparison with DkMYB4 27 .
We have employed for the first time the persimmon bHLH (DkMYC1) and WD40 (DkWDR1) components in transient expression assays in combination with MYB factors, for the elucidation of the function of MBW complexes in the expression of PA responsive genes in persimmon fruit. DkMYC1 does not affect significantly ANR expression on its own, but consistently intensifies the effect of DkMYB2 and DkMYB4. In Arabidopsis, there is a similar synergistic effect of TT2 (MYB) and TT8 (bHLH) on the expression of BAN that responds to a stronger cooperative binding of the pair TT2-TT8 to BAN promoter 20 . Thus, the low expression of ANR gene from early stages of fruit development in the ast non-astringent cultivar 'Hana Fuyu' (Fig. 3c) seems to be caused by the concomitant defective expression of DkMYB4 and DkMYC1 genes in this mutant 26 (Fig. 5). Contrarily to the positive effect of TTG1 (WD40) gene overexpression, and the negative effect of TTG1 silencing on BAN expression in Arabidopsis 20 , DkWDR1 expression does not affect ANR promoter activity in our transient expression experiments. The strawberry ortholog of TTG1 also increases Arabidopsis BAN expression in combination with its bHLH and MYB co-interactors 24 , which suggests the presence of certain functional particularities in persimmon DkWDR1 or perhaps regulatory differences between ANR and BAN promoters. On the other side, WD40 proteins act as structural platforms for facilitating protein-protein interaction, and consequently its effect on the positive transcriptional activity of the complex could be shaded by the ectopic overexpression of components of the complex and the presence of endogenous similar factors in N. benthamiana cells.
As DkMYB4 and DkMYC1 show a development and cultivar dependent expression profile highly similar to ANR 26,27,29 (Fig. 3c), we considered plausible a self-regulatory loop in the expression of these genes, similar to the positive feedback mechanism operating in TT8 from Arabidopsis 35 . However, DkMYB4 and DkMYC1 promoters are not activated by any MBW element utilized in this study. On the contrary, DkWDR1 reduces the expression of DkMYB4 by itself and in combination with MYB genes. This repressive effect of DkWDR1 resembles the activity of MBW complexes containing MYB proteins showing at the C-terminal end an ethylene-responsive element binding factor-associated amphiphilic repression (EAR) motif 36 . Thus, binding of specific repressive MYB www.nature.com/scientificreports www.nature.com/scientificreports/ proteins has potentially the ability to turn MBW complex into a repressive factor. In light of these data we cannot discard a repressive role of DkWDR1 on the expression of DkMYB4 and other genes, which could depend on the binding of distinct MYB of bHLH elements with specific activating or repressive domains.
Gene isolation. The Diospyros kaki (cv. 'Hachiya') genes DkMYB2 (AB503699.1), DkMYB4 (KR057233.1), DkMYC1 (KR057234.1) and DkWDR1 (KR057229.1) were obtained from pGADT7 plasmids described in a previous study 26 . In order to identify DkMYB4 and DkMYC1 promoter sequences, a manual assembly of D. lotus cv. Kunsenshi genome reads stored in the Sequence Read Archive (SRA) database (ID: SRP045872) was performed 40 . For amplifying ANR promoter, we designed primers (Supplementary Table S1) from the previously published sequence of D. lotus gene (AB504523.1). ANR promoter amplification was not possible in D. kaki, and therefore the D. lotus promoter was used in the analysis. DkMYB4 and DkMYC1 promoters were amplified from D. kaki cv. 'Hachiya' genomic DNA (Supplementary Table S1). The DNA of both D. kaki and D. lotus was extracted from fresh leaves following a standard CTAB DNA extraction protocol 41 . plasmid construction. For the construction of subcellular localization and BiFC vectors, each of the four genes were amplified (Supplementary Table S1) with a 15 bp target vector residue at the 5′ end needed for recombination with the In-Fusion HD Cloning kit (TAKARA BIO, Otsu, Japan). Fragments were purified and cloned into ampicillin resistant pSK + 35S-eGFP-PoPit vectors 42 and two cassettes for each protein were made for subcellular localization. Four cassettes for each protein were made with ampicillin resistant pSK + 35S-(N-YFP or C-YFP)-PoPit vector 43 for BiFC analysis. The fragments containing the expression cassette from pSK vectors were digested with HindIII and subcloned into the kanamycin resistant pMOG800 vector 44 .
In order to construct the vectors for transient expression, the four genes contained in pGADT7 plasmids 26 were digested with SacI and XhoI to release the insert. The fragments were purified and inserted in a pGreenII-62-SK kanamycin resistant vector 45 , previously linearized with SacI/XhoI. The promoter sequences were amplified from vectors using specific primers with restriction enzyme sequences tails at 5′ (Supplementary Table S1). The purified PCR products were digested (HindIII/PstI for ANR and DkMYC1 promoters and HindIII/NcoI for DkMYB4 promoter), purified and cloned into linearized pGreenII-0800-LUC kanamycin resistant vector 45 .
All the described vectors were provided by Dr. J. A. Sánchez-Navarro (Instituto de Biología Molecular y Celular de Plantas "Primo Yúfera", Valencia, Spain). Plasmids containing gene and promoter constructs were finally transferred to Agrobacterium tumefaciens strain C58 by electroporation. For pMOG800 vectors, transformation was carried out in bacteria containing the virulence helper plasmid pCH32 46 . All DNA constructions were verified by plasmid DNA sequencing.

Subcellular localization of MBW complex proteins in vivo.
To characterize the subcellular localization of the MBW complex components, each protein was fused at either the Nt or the Ct of eGFP and transiently expressed in planta. For a better visualization of the fluorescence signal, all proteins were co-expressed with the silencing suppressor HC-Pro protein from the Tobacco Etch Virus 43 . A. tumefaciens C58 strains were grown overnight in LB media supplemented with kanamycin and rifampicin, at 28 °C. Cultures were centrifuged 5 min at 4,000 × g, and pellets were resuspended in infiltration media (MgCl 2 10 mM + MES 10 mM pH 5.6) to an OD 600 of 0.5 for each construct and an OD 600 of 0.1 for the HC-Pro. N. benthamiana young plants (2 pairs of leaves) were agroinfiltrated as previously described 47 . Plants remained in a greenhouse at 24 °C (day) and 18 °C (night) with a 16 h light photoperiod. Three days after infiltration, leaf samples were collected and mounted in a microscope slide with a drop of water. Observation of the fluorescence in the underside epidermis was performed with a LEICA TCS SL confocal microscope (λ exc = 488 nm; λe m = 492-533 nm for eGFP). For the nucleus and nucleolus subcellular colocalization, the proteins were coinfiltrated with cultures (OD 600 0.1) expressing the NLS of SV40 large T antigen fused to the red fluorescent protein and the fibrillarin fused to the cherry fluorescent protein, respectively (λ exc = 561 nm; λe m = 588-634 nm).

Bimolecular fluorescence complementation assays (BiFC).
In the BiFC assay 48 , addressed to characterize the interaction between the components of the hypothetical MBW complex, all the possible two-by-two combinations of homodimers and heterodimers were assayed in planta. Chimeric proteins were transiently co-expressed in N. benthamiana using A. tumefaciens (strain C58) cultures (OD 600 = 0.4) transformed with the corresponding binary plasmids pMOG800, as previously described 47 . To increase the expression of the different proteins, we included an A. tumefaciens culture (OD 600 = 0.1) expressing the HC-Pro. At 3 days post-infiltration, the fluorescence reconstitution was monitored in the confocal LEICA TCS SL (λ exc = 488 nm; λ em = 492-533 nm).

Dual luciferase assay.
To determine the effects of the hypothetical MBW protein complex on the promoters of ANR, DkMYB4 and DkMYC1 genes, a dual luciferase assay was performed. The effect of the homo and heterodimers formed by two and three proteins was assayed by the co-expression of the promoter and protein vectors as previously described in N. benthamiana with A. tumefaciens strain C58. HC-Pro was also used for enhancing the transient expression of the different proteins. After 3 days, 30 mg of agroinfiltrated leaves from 3 biological replicates, were sampled for each combination. Samples were immediately frozen in liquid nitrogen. For measuring promoter activity, samples were ground to powder and then Firefly (LUC) and Renilla (REN) luciferase activity was measured following the Dual-Luciferase Reporter Assay System (PROMEGA, Madison, WI, USA) with the aid of a PROMEGA GloMax Multi Microplate Reader luminometer. Promoter activity was measured as the quotient between the LUC/REN ratio of promoter plus transcription factors samples and the LUC/REN ratio of promoter without additional factors.
isolation of RnA and quantitative real-time Rt-pcR (qRt-pcR). Total RNA was isolated from 150 mg of fruit flesh using a cetyltrimethylammonium bromide (CTAB)-based procedure 50 . Genomic DNA was removed with the RNase-Free DNase Set (QIAGEN, Hilden, Germany) according to manufacturer's instructions. Purified RNA was reverse transcribed with PrimeScript RT Reagent Kit (TAKARA BIO). qRT-PCR was performed in a StepOnePlus Real-Time PCR System (LIFE TECHNOLOGIES, Carlsbad, CA, USA), using 1-2 µl of 10X diluted cDNA, SYBR premix Ex Taq (Tli RNaseH plus) (TAKARA BIO) and primers shown in Supplementary Table S1, in a total volume of 20 µl. The PCR protocol consisted of 10 min at 95 °C, followed by 40 cycles of 15 s at 95 °C, and 1 min at 60 °C. PCR specificity was confirmed by the presence of a single peak in the dissociation curve and by agarose electrophoresis. We used DkActin as reference gene 26,27 . A relative standard curve procedure was employed for measuring relative expression. Results were the average of two independent biological replicates with 2-3 technical replicates each.