The DELLA proteins interact with MYB21 and MYB24 to regulate filament elongation in Arabidopsis

Background Gibberellin (GA) and jasmonate (JA) are two essential phytohormones for filament elongation in Arabidopsis. GA and JA trigger degradation of DELLAs and JASMONATE ZIM-domain (JAZ) proteins through SCFSLY1 and SCFCOI1 separately to activate filament elongation. In JA pathway, JAZs interact with MYB21 and MYB24 to control filament elongation. However, little is known how DELLAs regulate filament elongation. Results Here we showed that DELLAs interact with MYB21 and MYB24, and that R2R3 domains of MYB21 and MYB24 are responsible for interaction with DELLAs. Furthermore, we demonstrated that DELLA and JAZ proteins coordinately repress the transcriptional function of MYB21 and MYB24 to inhibit filament elongation. Conclusion We discovered that DELLAs interact with MYB21 and MYB24, and that DELLAs and JAZs attenuate the transcriptional function of MYB21 and MYB24 to control filament elongation. This study reveals a novel cross-talk mechanism of GA and JA in the regulation of filament elongation in Arabidopsis.


Background
Stamen, comprising a filament and an anther, is one of the plant reproductive organs. Disruptions in stamen development, such as in filament elongation, anther dehiscence, or pollen maturation, can cause male sterility [1,2]. Numerous studies revealed that these processes are controlled by phytohormones, including jasmonate (JA) and gibberellin (GA) [3,4].
GAs activate the expression of MYB21, MYB24, and MYB57 via suppressing DELLA proteins and upregulating the expression of JA biosynthetic genes DAD1 and LOX1 and JA biosynthesis to mediate filament elongation [52]. Here, in this study we further demonstrated that MYB21 and MYB24 are the direct targets of DELLAs, and act as a necessary node for GA-JA synergism in filament elongation. We showed that DELLAs interact with MYB21 and MYB24 via R2R3 domains, and that DELLA and JAZ proteins coordinately repress the transcriptional function of MYB21 and MYB24 to inhibit filament elongation.

MYB21 and MYB24 interact with DELLA proteins
We fused MYB21 with LexA DNA binding domain (BD), and found that BD-MYB21 showed strong autoactivation in yeast (Additional file 2: Figure S1a). We further truncated MYB21 into MYB21NT containing R2R3 DNA binding domain and MYB21CT including NYW G / S M / V DD I / L W S / P motif (Fig. 1a), and found that MYB21NT lost strong auto-activation (Additional file 2: Figure S1a). MYB21NT was used as bait to screen MYB21 interaction proteins in Arabidopsis cDNA library in Y2H system. The DELLA protein RGA is one of the putative interaction clones.
Full-length of DELLAs with BD domain showed strong auto-activation in yeast, therefore, we truncated DELLAs into L and R domains (RGA-L and RGL2-L; RGA-R, GAI-R, RGL1-R, RGL2-R, and RGL3-R) (Fig. 1c), and R domains lost strong auto-activation. MYB21 and MYB24 were respectively fused with AD domain, and showed no auto-activation in yeast (Additional file 3: Figure S2b). The Y2H results in Fig. 1d showed that RGA-R, GAI-R, RGL1-R, RGL2-R, and RGL3-R interact with MYB21 and MYB24 respectively in yeast.
We next performed pull-down assay to verify the interactions of MYB21 and MYB24 with DELLAs in vitro. Maltose binding protein (MBP), MBP-fused MYB21 (MBP-MYB21) and MYB24 (MBP-MYB24) were expressed in Escherichia coli and purified by amylose resin. DELLA proteins RGA, RGL1 and RGL2 were extracted from transgenic Arabidopsis expressing tandem affinity purification (TAP) tag-fused DELLA proteins (TAP-RGA, TAP-RGL1 and TAP-RGL2) [41]. MBP, MBP-MYB21 and MBP-MYB24 were incubated with TAP-RGA extracts expressing TAP-RGA, and then separated on SDS-PAGE for immunoblotting with anti-cmyc antibody. The result showed that MBP-MYB21 and MBP-MYB24 could efficiently pull down TAP-RGA, but the negative control MBP could not (Fig. 1e), indicating that RGA interacts with MYB21 and MYB24. In addition, we also observed that DELLA proteins RGL1 and RGL2 interact with MYB21 and MYB24 in pull -down assay ( Fig. 1f and g).
Taken together, the Y2H assay, and pull-down assay consistently demonstrate that DELLAs interact with the R2R3 MYB transcription factors (MYB21 and MYB24), implying that these two transcription factors function as direct targets of DELLA proteins.

R2R3 domains of MYB21 and MYB24 are involved in interactions with DELLAs
We further investigated interactions of DELLA R fragments with MYB21NT, MYB24NT, MYB21CT, and MYB24CT. As shown in Fig. 1d, RGA-R, GAI-R, RGL1-R, RGL2-R, and RGL3-R interacted with MYB21NT and MYB24NT, but not with MYB21CT and MYB24CT, whereas no interaction was detected in negative controls ( Fig. 1d and Additional file 3: Figure S2b), indicating that DELLAs interact with R2R3 DNA binding domains of MYB21 and MYB24.
We also examined whether L or R domains of DELLA proteins interact with MYB21NT and MYB24NT. Y2H results suggested that MYB21NT and MYB24NT interact with both L parts and R parts of RGA and RGL2 (Fig. 1b), while no interaction was detected in negative controls ( Fig. 1b and Additional file 3: Figure S2a), indicating that both N-terminus Having shown that DELLAs interact with MYB21 and MYB24, we then performed an Arabidopsis protoplast transient expression assay [53] using the GAL4 DNA binding domain (GAL4DB) and its binding sites [GAL4(4X)-D1-3(4X)] to test whether DELLAs could influence the transcriptional function of MYB21 and MYB24.
MYB21 and MYB24 were respectively fused with GAL4DB vector and served as effectors. The reporter was the GUS (β-glucuronidase) gene controlled by four copies of GAL4 DNA binding site [GAL4(4x)-D1-3(4x)], and the internal control was 35S promoter-driven firefly luciferase (LUC) gene (Fig. 2a). The DELLA genes RGA and RGL2 were respectively cloned into pGreen62 vector (Fig. 2a). As shown in Fig. 2b, expression of GAL4DB-MYB21 together with the GUS reporter could enhance the GUS/LUC ratio, while coexpression of RGA or RGL2 with GAL4DB-MYB21 repressed the transcriptional function of MYB21. We also observed that RGA and RGL2 inhibit the transcriptional function of MYB24 (Fig. 2c).
As both JAZs and DELLAs interact with MYB21 and MYB24 (Fig. 1) [32], and DELLAs could repress these two factors transcriptional function (Fig. 2b and c), we further investigated whether JAZs or DELLAs and JAZs synergistically regulate the transcriptional function of MYB21 and MYB24. We used MYB21 as the representative (Fig. 2d) and found that RGL2 or JAZ1 alone repressed the transcriptional activity of MYB21 (Fig. 2e). We further discovered that coexpression of RGL2 and JAZ1 exhibited a much stronger inhibition of the transcriptional activity of MYB21 compared with RGL2 or JAZ1 alone (Fig. 2e). These results revealed that DELLAs and JAZs coordinately repress the transcription activity of MYB21.
We also explored a cell-free assay system [54] to discuss whether the MYB21 protein level is regulated by GA or JA through analysis of N. benthamiana leaves transiently expressed myc-taged MYB21 (myc-MYB21). As the results shown in Fig. 2f and h, the myc-MYB21 protein level was decreased to~33% after 2 h treatment with mock, while JA treatment promoted the degradation of myc-MYB21 protein (~24% of the myc-MYB21 level without treatment). However, GA could delay the degradation of myc-fused MYB21 (~46% of the myc-MYB21 level without treatment) (Fig. 2g and i). These results indicated that GA and JA may play an opposite role in regulating the stability of MYB21 protein.

DELLAs and JAZs converge on MYB21 and MYB24 to regulate filament elongation
As both JAZs and DELLAs target MYB21 and MYB24, we further explore the cross-talk of JA and GA in the regulation of filament elongation.
Filaments in JA-deficient mutant opr3 at floral stage 13 are much shorter than that in wild type, indicating that stabilized JAZ proteins attenuate the function of MYB21 and MYB24 to repress filament elongation, and JA treatment could restore the filament elongation of opr3 ( Fig. 3a and b). Previous studies showed that GA could induce degradation of DELLA proteins [46,55]. We tested whether GA treatment could release DELLAtargeted MYB21 and MYB24 to enhance filament growth in opr3. We further measured the ratio of filament to pistil length of opr3 treated with GA. As shown in Fig. 3a and b, GA treatment could only very slightly rescue the filament/pistil ratio of opr3, indicating that majority of MYB21 and MYB24 are inhibited by JAZs in opr3, and that the released MYB21 and MYB24 by GA treatment were not enough to rescue the filament elongation of opr3, whereas treatment with both JA and GA recovered filament elongation of opr3 ( Fig. 3a and b).
RGL2 protein alone could repress the function of MYB21 and MYB24 to control filament elongation in Q3 (ga1-3 gai-t6 rgl1-1 rga-t2, wild type for RGL2 and RGL3) [52]. Application of GA could induce RGL2 degradation and restored the filament elongation of Q3 ( Fig.  3c and d). We further explored whether degradation of JAZs by JA treatment could release enough MYB21 and MYB24 to rescue the filament growth of Q3. As shown in Fig. 3c and d, JA treatment could very weakly recover the ratio of filament and pistil length of Q3, indicating that RGL2 alone in Q3 is able to dominantly inhibit the function of MYB21 and MYB24 to repress the filament elongation. Exogenous application of both GA and JA could also restore the filament elongation of Q3 ( Fig. 3c and d).

Discussion
Both GA and JA act antagonistically to regulate hypocotyl elongation and root growth [56,57], and synergistically to mediate stamen development, trichome development and sesquiterpene production [43,52,58]. In this work, we further present a new model that provides a synergistic mechanism for GA and JA signaling in filament elongation. In this model, both DELLAs and JAZs interact with R2R3 MYB transcription factors MYB21 and MYB24 to repress their transcription activity, while GA and JA induce the degradation of DELLAs and JAZs, respectively, to release and coordinately activate MYB21 and MYB24, and synergistically regulate filament elongation (Fig. 3e).
Consistent with this synergistic regulatory mechanism, accumulation of JAZs in the JA biosynthesis mutant (opr3), and the DELLA protein RGL2 in GA-deficient Asterisks represent Student's t-test significance between pairs indicated with brackets (*, P < 0.05; **, P < 0.01). The abundance of myc-MYB21 with mock treatment for 0 h was set to 1.0 mutant Q3, respectively interact with MYB21 and MYB24, and repress filament elongation. Exogenous GA treatment of opr3 or JA treatment of Q3 could not significantly recover the filament elongation (Fig. 3a-d), suggesting that JAZs in opr3 and RGL2 in Q3 are efficient to inhibit the function of MYB21 and MYB24 to control filament elongation.
A recent study revealed that IIIe bHLH transcription factors (MYC2, MYC3, MYC4 and MYC5) act as JAZstargeted proteins to redundantly regulate filament elongation, and that these four factors interact with MYB21 and MYB24 to form a MYB-MYC complex to regulate filament elongation [38]. Having shown that DELLAs interact with and repress the transcription activity of the MYB components of the MYB-MYC complex (Figs. 1 and 2), it will be interesting to investigate whether DEL-LAs are able to interact with and attenuate the transcriptional function of the bHLH components (MYC2, MYC3, MYC4 and MYC5) of the MYB-MYC complex to regulate filament elongation in Arabidopsis.
Previous studies revealed that JA induces the expression of MYB21 and MYB24 [33,34], and that GA promotes the JA biosynthesis to activate the mRNA level of MYB21 and MYB24 [52]. We also explored whether JA and GA affect the protein stability of MYB21. As shown in Fig. 2f-i, JA and GA respectively could promote, or delay the degradation of MYB21. It indicated that the post-transcription regulation mechanism of MYB21 is sophisticated, which would contribute to the delicate dynamic stamen development and need further investigation.

Conclusions
In conclusion, we found that MYB21 and MYB24 are the direct targets of DELLAs, and that DELLA and JAZ proteins synergistically repress the transcription activity of MYB21 and MYB24 to repress filament elongation.

Plant materials and growth conditions
The Arabidopsis thaliana mutants opr3 and Q3 were previously described [29,52], and respectively donated by professor John Browse (Washington State University) and professor Jinrong Peng (Zhejiang University). The Arabidopsis thaliana transgenic plants TAP-RGA, TAP-RGL2 and TAP-RGL1 were previously described [41], and donated by professor Xingwang Deng (Peking University). Arabidopsis seeds were surface sterilized with 20% bleach for 10 min, plated on Murashige and Skoog (MS) medium (Sigma-Aldrich) supplied with 3% sucrose, kept at 4°C for 3 days in the dark, and then plated to growth house with 16 h-light (22°C to 24°C)/8 h-dark (16°C to 19°C) photoperiod. Seeds of Q3 were immerged in 100 μM GA 3 at 4°C for 7 days before sowing. Nicotiana benthamiana seeds were donated by professor Yule Liu (Tsinghua University), and cultivated at 22°C-28°C under 16 h-light/8 h-dark photoperiod.

Y2H screening and Y2H assay
The Y2H screening method and Y2H assay was previously described [32]. For Y2H assay, MYB21, MYB24, RGA, GAI, RGL1, RGL2, RGL3, and their related domains were inserted into pLexA or pB42AD vectors respectively. The indicated construct pairs were cotransformed into yeast strains EGY48, and plated on SD agar medium with -His/−Trp/−Ura DO supplement (Clontech) at 30°C for 4-5 days. The yeast transformants was resuspended with SD/−His-Trp-Ura liquid medium and cultured at 30°C for 24 h, subsequently harvested and resuspended with distilled water. 5 μl of the indicated suspension was dropped into 96-well plates containing 2% Gal/1% raffinose/SD/−Ura/−His/−Trp/−Leu/X-β-Gal medium. Y2H images were taken 2 days after incubation at 30°C. The primers for the vector construction are listed in Additional file 1: Table S1.

Pull-down assay
The coding region of MYB21, and MYB24 were cloned into the pMAL-c5x (NEB) vector to generate MBP fused MYB21 and MYB24 respectively. The MBP, MBP-fused MYB21 and MYB24 were expressed in E. coli and purified by amylose resin beads.

Protoplast transfection assay
For transient expression assay, the Arabidopsis MYB21 and MYB24 were amplified and fused with the GAL4DB vector through SmaI and SalI sites, and RGA, RGL2 and JAZ1 were respectively cloned into the pGreenII 62-SK [59]. The GUS, and the firefly LUC gene was respectively controlled by four copies of upstream GAL4DB binding sites (GAL4(4x)-D1-3(4x)) and 35S promoter, and acted as a reporter and the internal control respectively [53]. Arabidopsis mesophyll protoplasts preparation and transfection were described as previously [60]. Primers used for plasmid construction are listed in Additional file 1: Table S1.
Protein degradation assay N. benthamiana leaves were infiltrated with Agrobacterium strains harboring myc-MYB21. 50 h after infiltration, 3 g of Agrobacterium-infiltrated leaves was collected. The total protein was extracted using RB buffer and incubated at 22°C for indicated time periods without or with 100 μM methyl-jasmonate (MeJA), 100 μM GA 3 , then separated by SDS-PAGE, transferred to PVDF membrane, and detected with anti-c-myc antibody (Abmart). The protein level of MYB21 was quantified by the software from FluorChem M MultiFluor System (Alpha). The experiment was repeated for three independent biological replicates.

Measurement of pistil and filament length
For this experiment, the pistil and one of the four longer filaments of 10 flowers at floral stage 13 [61] of each genotype were collected and measured under the microscope. Young flowers buds of opr3 and Q3 were treated with mock, 100 μM methyl jasmonate, 100 μM GA 3 , or 100 μM methyl jasmonate plus 100 μM GA 3 twice a day for detecting the ratio of filament and pistil length. The experiment was repeated for three independent biological replicates.