Interactions of OsMADS1 with Floral Homeotic Genes in ... - Cell Presshttps://www.researchgate.net/.../y-Interactions-of-OsMADS1-with-Floral-Homeotic-Gene...

7 downloads 0 Views 5MB Size Report
Apr 24, 2015 - Yun Hu1, Wanqi Liang1, Changsong Yin1, Xuelian Yang1, Baozhe Ping1, Anxue Li2, ... *Correspondence: Zheng Yuan ([email protected]).
Molecular Plant Research Article

Interactions of OsMADS1 with Floral Homeotic Genes in Rice Flower Development Yun Hu1, Wanqi Liang1, Changsong Yin1, Xuelian Yang1, Baozhe Ping1, Anxue Li2, Ru Jia1, Mingjiao Chen1, Zhijing Luo1, Qiang Cai1, Xiangxiang Zhao3, Dabing Zhang1,4 and Zheng Yuan1,* 1

State Key Laboratory of Hybrid Rice, Shanghai Jiao Tong University–University of Adelaide Joint Centre for Agriculture and Health, School of Life Sciences and Biotechnology, Shanghai Jiao Tong University, Shanghai 20040, China

2

Shanghai Ocean University, Shanghai 201306, China

3

Jiangsu Collaborative Innovation Center of Regional Modern Agriculture and Environmental Protection, Huaiyin Normal University, Huaian 223300, China

4

School of Agriculture, Food and Wine, University of Adelaide, Waite Campus, Urrbrae, SA 5064, Australia

*Correspondence: Zheng Yuan ([email protected]) http://dx.doi.org/10.1016/j.molp.2015.04.009

ABSTRACT During reproductive development, rice plants develop unique flower organs which determine the final grain yield. OsMADS1, one of SEPALLATA-like MADS-box genes, has been unraveled to play critical roles in rice floral organ identity specification and floral meristem determinacy. However, the molecular mechanisms underlying interactions of OsMADS1 with other floral homeotic genes in regulating flower development remains largely elusive. In this work, we studied the genetic interactions of OsMADS1 with B-, C-, and D-class genes along with physical interactions among their proteins. We show that the physical and genetic interactions between OsMADS1 and OsMADS3 are essential for floral meristem activity maintenance and organ identity specification; while OsMADS1 physically and genetically interacts with OsMADS58 in regulating floral meristem determinacy and suppressing spikelet meristem reversion. We provided important genetic evidence to support the neofunctionalization of two rice C-class genes (OsMADS3 and OsMADS58) during flower development. Gene expression profiling and quantitative RT-PCR analyses further revealed that OsMADS1 affects the expression of many genes involved in floral identity and hormone signaling, and chromatin immunoprecipitation (ChIP)–PCR assay further demonstrated that OsMADS17 is a direct target gene of OsMADS1. Taken together, these results reveal that OsMADS1 has diversified regulatory functions in specifying rice floral organ and meristem identity, probably through its genetic and physical interactions with different floral homeotic regulators. Key words: OsMADS1, floral homeotic genes, floral organ identity, floral meristem, regulatory network Hu Y., Liang W., Yin C., Yang X., Ping B., Li A., Jia R., Chen M., Luo Z., Cai Q., Zhao X., Zhang D., and Yuan Z. (2015). Interactions of OsMADS1 with Floral Homeotic Genes in Rice Flower Development. Mol. Plant. 8, 1366– 1384.

INTRODUCTION Flowers are characteristic features of angiosperms, which display a huge morphological diversity and play a pivotal role in life-cycle succession and seed production (Theissen and Melzer, 2007). Genetic and molecular investigations in model eudicot plants, such as Arabidopsis thaliana and Antirrhinum majus, led to the ABC model of flower development for the interpretation on the genetic control of floral organ specification (Coen and Meyerowitz, 1991), which later was extended into the ABCDE model (Colombo et al., 1995; Pelaz et al., 2000; Theissen and Saedler, 2001). In Arabidopsis, two A-class genes (APETALA1, AP1; APETALA2, AP2) and one C-class gene (AGAMOUS, AG)

specify the identities of sepals and carpel in whorls 1 and 4, respectively. Combination of A-class genes and B-class genes (APETALA3, AP3; PISTILLATA, PI) determines the identity of petals in whorl 2, and B-class and C-class genes together control the stamen identity in whorl 3 (Coen and Meyerowitz, 1991). In addition, D-class genes SEEDSTICK (STK), SHATTERPROOF1 (SHP1), and SHP2 specify ovule identity (Pinyopich et al., 2003), and E-class genes (SEPALLATA1/2/3/4, SEP1/2/3/4) redundantly determine the floral organ identity of all four whorls and maintain

Published by the Molecular Plant Shanghai Editorial Office in association with Cell Press, an imprint of Elsevier Inc., on behalf of CSPB and IPPE, SIBS, CAS.

1366 Molecular Plant 8, 1366–1384, September 2015 ª The Author 2015.

Interactions of OsMADS1 with Floral Homeotic Genes

Molecular Plant

floral meristem by regulating expression of floral organ identity genes (Pelaz et al., 2000; Honma and Goto, 2001; Theissen and Saedler, 2001; Pelaz et al., 2001a, 2001b; Favaro et al., 2003; Pinyopich et al., 2003; Ditta et al., 2004; Immink et al., 2009; Kaufmann et al., 2009). Furthermore, genetic, molecular, and evolutionary studies revealed that this model is partially conserved in angiosperms even though it varies in different species in line with diverse morphology of flowers (Soltis et al., 2007; Theissen and Melzer, 2007; Rijpkema et al., 2010).

(Jeon et al., 2000; Prasad et al., 2001; Agrawal et al., 2005; Chen et al., 2006; Wang et al., 2010; Khanday et al., 2013), suggesting its possible role in specifying the lemma and palea identity as well as floral meristem determinacy. In agreement with the ABCDE model of flower development, OsMADS1 as a E-class protein physically interacts with SEP-like proteins OsMADS7, OsMADS8, AP1-like proteins OsMADS14, OsMADS15, and AGL6 protein OsMADS6 (Moon et al., 1999; Lim et al., 2000; Cui et al., 2010). Genetically, OsMADS1 ensures rice sexual reproduction together with OsMADS15 (Wang et al., 2010), given that osmads1 osmads15 double mutants exhibit new plantlets instead of flowers. Moreover, the developmental regulation of all floral whorls and meristem fate specification is accomplished by the interaction between OsMADS1 and OsMADS6 (Ohmori et al., 2009; Li et al., 2010, 2011a), and the direct repression of OsMADS34 expression by OsMADS1 mediates the spikelet-to-floret meristem transition, meristem maintenance, determinacy, and lateral organ development through multiple signaling pathways (Khanday et al., 2013). More recently, we showed that OsMADS1 expression is directly activated by OsMYC2, a crucial activator of jasmonic acid response genes, to regulate rice flower development (Cai et al., 2014). Despite the finding that OsMADS1 specifies floral meristem and organ identities, mechanisms on whether and how OsMADS1 interplays with other floral homeotic genes remain largely unknown.

Rice is a monocot plant that has evolved a specialized spikelet, which is the basic unit of the inflorescence and a determinant to grain yield (Zhang and Yuan, 2014). Each rice spikelet consists, from the outside inwards, of two rudimentary glumes (also called reduced glumes) and two sterile lemmas (also called empty glumes) arranged in an alternate phyllotaxis, and one fertile floret (Kellogg, 2001; Yuan et al., 2009). The rice floret, following the alternate phyllotaxis of external glumes, contains two bract-like structures (a lemma and a palea) hypothesized to be whorl-1 organs, two fleshy organs (called lodicules) considered to be equivalent to eudicot petals at the lemma side of whorl 2, six stamens (whorl 3), and one pistil (whorl 4) (Zhang et al., 2013a). Many rice BCDE genes exhibit their conserved or semi-conserved function in flower morphogenesis (Kyozuka et al., 2000; Kater et al., 2006; Preston and Kellogg, 2006; Dreni et al., 2007; Prusinkiewicz et al., 2007; Reinheimer and Kellogg, 2009; Zhang and Wilson, 2009; Ciaffi et al., 2011; Dreni et al., 2011; Yoshida and Nagato, 2011; Zhang et al., 2013a). However, a few rice genes, especially A-class and E-class genes, have novel functions in flower development compared to Arabidopsis. For example, the AP1/ FRUITFULL (FUL)-like genes, OsMADS14, OsMADS15, and OsMADS18, specify inflorescence meristem identity by interacting genetically with one of the E-class genes OsMADS34/ PAP2 (Kobayashi et al., 2012). In addition, rice AGAMOUS-LIKE6 (AGL6) gene OsMADS6 shows similar function as E-class genes in flower organ identity specification and flower meristem determinacy, suggesting that the E-function extends to genes from the AGL6-like subfamily besides the SEP subfamily (Ohmori et al., 2009; Li et al., 2010, 2011a). In flowering plants, the first gene duplication event occurred within the SEP subfamily prior to the origin of the extant angiosperms (Zahn et al., 2005). Rice genome contains five SEP-like genes, OsMADS1/LEAFY HULL STERILE1 (LHS1), OsMADS5, OsMADS7, OsMADS8, and OsMADS34 (Malcomber and Kellogg, 2004; Agrawal et al., 2005; Zahn et al., 2005). Knockdown of four of them, i.e. OsMADS1, OsMADS5, OsMADS7, and OsMADS8, resulted in the homeotic transformation of all floral organs into leaf-like organs with the exception of lemma, indicating that SEP-like genes play a conserved role of ‘‘E’’-function in specifying floral determinacy and organ identities (Cui et al., 2010). Unlike the subtle or undetectable phenotype of single mutant of Arabidopsis SEP-like genes (Pelaz et al., 2000; Ditta et al., 2004), osmads34 mutants exhibit obvious defects in both inflorescence and spikelet morphology, suggesting its pivotal function in controlling rice flower development (Gao et al., 2010; Kobayashi et al., 2010). Different from osmads34, osmads1 mutant phenotypes include reiterative formation of glumes and defective floral organ identity, or the formation of an extra spikelet within the spikelet

Here, we studied interactions of OsMADS1 with OsMADS16 (a B-class gene), OsMADS3, OsMADS58 (C-class genes), and OsMADS13 (a D-class gene), respectively, in rice flower development. Yeast two-hybrid and in vitro GST pull-down assays demonstrated that OsMADS1 physically interacts with these B-, C-, and D-class proteins. Genetic studies revealed that neofunctionalization occurred in rice C-class genes. OsMADS1 interacts genetically with OsMADS3 to maintain the floral meristem activity, but interacts genetically with OSMAD58 to suppress spikelet meristem reversion, besides reported roles in specifying the reproductive organs identity and regulating floral meristem determinacy. In addition, gene expression profiling of the OsMADS1 null allele mutant osmads1-z revealed that expression of numerous genes, such as floral homeotic and hormone signaling genes, were affected during the flower initiation. Further studies discovered that OsMADS1 directly binds to a CArG box in the promoter of an AGL6-like gene OsMADS17. Therefore, our results demonstrated the diversified regulatory functions of OsMADS1 in specifying rice floral organ and meristem identity.

RESULTS OsMADS1 Regulates Floral Meristem Identity and Activity To comprehensively understand the function of OsMADS1, we reanalyzed the previously identified null allele osmads1-z mutant (Gao et al., 2010) (for mutant information, please see Supplemental Figure 1). Consistent with previous reports (Jeon et al., 2000; Agrawal et al., 2005; Prasad et al., 2005; Chen et al., 2006; Gao et al., 2010; Khanday et al., 2013), defective floral organs were observed in whorls 1 and 2 of osmads1-z flowers, including elongated leafy lemmas, palea-like and glume-like lodicules (Figure 1A–1E). According to phenotypes of

Molecular Plant 8, 1366–1384, September 2015 ª The Author 2015. 1367

Molecular Plant

Interactions of OsMADS1 with Floral Homeotic Genes

Figure 1. Phenotype Analysis of Wild-Type and osmads1-z Mutant Flower. (A and B) Wild-type rice flower. (C) Cross section of the wild-type spikelet. Asterisks indicate stamens. (D and E) osmads1-z flower showing leafy lemma and palea (D) and lemma/palea-like organs inside lemma and palea (E). Arrows in (E) indicate lemma/ palea-like organs. (F–J) Four different types of mutant phenotypes of osmads1-z. (F) Type I osmads1-z flower showing weak mutant phenotype with glume-like organs and decreased number of stamens. Arrow indicates pistil. (G) Cross section of the type I osmads1-z spikelet. Arrows indicate lemma/palea-like organs. Arrowheads indicate glume-like organs. Asterisks indicate stamens. (H) Type II osmads1-z flower showing no stamens and pistil in the center of the flower. (I) Type III osmads1-z flower showing twin flowers in one osmads1-z spikelet. Each arrow in (I) indicates one flower. (J) Type IV osmads1-z flower showing an ectopic pedicel in the center of the flower. Arrow indicates ectopic pedicel. (K) Expression patterns of OSH1 in wild-type flower at stage Sp4, Sp6, and Sp8. Arrowhead in the middle panel indicates floral meristem. Arrowhead in the right panel indicates signals that could not be detected after carpel developed. (L) Expression patterns of OSH1 in osmads1-z flower at stages Sp4 and Sp8. Arrowhead in Sp8 indicates floral meristem. ca, carpel; fm, floral meristem; gll, glume-like organ; le, lemma; lo, lodicule; osm, OsMADS; pa, palea; pi, pistil; st, stamen; wt, wild-type. Scale bars: 1 mm in (A, B, D–F, H–J); 100 mm in (C, G); 50 mm in (K, L).

their inner floral organs, osmads1-z flowers were grouped into four different types (Table 1): type I (71.99%, n = 432) with altered number of stamens and pistils (Figure 1F and 1G; Table 2); type II (11.81%, n = 432) with inner floral organs completely replaced by glume-like organs inside the lemma and the palea (Figure 1H); type III (9.72%, n = 432) with twin flowers in each spikelet (Figure 1I); and type IV (6.48%, n = 432) with a new spikelet consisting in an elongated pedicel at the flower center surrounded by multiple layers of glume-like structures (Figure 1J). These observations suggested that OsMADS1 plays a key role in specifying floral organ identity and meristem

determinacy. To further decipher the role of OsMADS1 in floral meristem activity, we analyzed the expression pattern of Oryza sativa homeobox1 (OSH1), a marker of shoot apical and reproductive meristem cells (Sato et al., 1996; Yamaguchi et al., 2004). Although wild-type and osmads1-z flower primordia displayed a similar and uniform expression of OSH1 at stage Sp4 (Figure 1K and 1L), a prolonged expression of OSH1 was noticed in osmads1-z flowers at stage Sp8, while no signals were detected in wild-type flowers (Figure 1K and 1L). Consistently, scanning electron microscopy (SEM) revealed the abnormality on flower organ identity specification and meristem

1368 Molecular Plant 8, 1366–1384, September 2015 ª The Author 2015.

Molecular Plant

Interactions of OsMADS1 with Floral Homeotic Genes

Mutant plants

Weak mutant (type I)

‘‘Empty hull’’ (type II)

osmads1-z

311 (71.99%)

51 (11.81%)

42 (9.72%)

28 (6.48%)

432

osmads1-z spw1-1

225 (72.12%)

27 (8.65%)

15 (4.81%)

45 (14.42%)

312

46 (12.14%)

283 (74.67%)

5 (1.32%)

45 (11.87%)

379

26 (6.45%)

65 (16.13%)

6 (1.49%)

306 (75.93%)

403

300 (81.97%)

15 (4.10%)

39 (10.66%)

12 (3.28%)

366

osmads1-z osmads3-4 osmads1-z osmads58 osmads1-z osmads13-3

‘‘Twin flowers’’ (type III)

‘‘New spikelet’’ (type IV)

No. of spikelets examined

Table 1. The Number and Percentage of Four Phenotypes in Mutant Flowers. The main phenotype of mutant flowers is underlined.

determinacy in osmads1-z (Supplemental Figure 2), confirming the role of OsMADS1 in the determinacy of floral meristem activity.

OsMADS1 RNAi transgenic plants was initially interpreted by reduced expression of B-class genes, including OsMADS2, OsMADS4, and OsMADS16 (Jeon et al., 2000; Nagasawa et al., 2003; Agrawal et al., 2005; Prasad et al., 2005; Chen et al., 2006; Yao et al., 2008; Yoshida, 2012). Given that spw1-1 is the only available mutant among rice B-class genes (Figure 3A and 3B) (Nagasawa et al., 2003), genetic interaction was investigated between OsMADS1 and OsMADS16 by generating double mutant osmads1-z spw1-1. Additive phenotypes appeared in osmads1-z spw1-1 double mutants (Tables 1 and 2), showing that the lemma, palea, and lodicules of most of the double mutants were homeotically converted into leaf-like structures, mimicking the osmads1-z flower, and the stamens were transformed into carpel-like structures reminiscent of spw1-1 phenotypes (Figure 3C–3H and Table 2). These phenotypes suggested that no direct genetic interaction exists between OsMADS1 and OsMADS16.

The Interactions of OsMADS1 with B-, C-, and D-Class Proteins SEP proteins are thought of as a glue to mediate MADS box complex formation by interacting with A-, B-, C- and D-class proteins to determine floral context (Malcomber and Kellogg, 2005; Immink et al., 2009). In rice, previous studies have shown that OsMADS1 interacts with OsMADS14 and OsMADS15 (A-class), OsMADS6 (AGL6-class), and OsMADS7 and OsMADS8 (E-class), respectively (Moon et al., 1999; Lim et al., 2000; Cui et al., 2010). To investigate whether OsMADS1 physically interacts with B-, C-, and D-class proteins, yeast two-hybrid analysis was performed. The data showed that OsMADS1 could form heterodimers with OsMADS16, OsMADS3, OsMADS58, and OsMADS13, respectively (Figure 2A). Physical interactions of OsMADS1 with OsMADS16, OsMADS3, OsMADS58, and OsMADS13 were further confirmed by in vitro GST pull-down assays (Figure 2B–2E). Next, we decided to take genetic approaches to explore possible in vivo biological functions of these protein–protein interactions.

Moreover, in situ hybridization analysis of OsMADS1 and OsMADS16 was carried out to look into spatial and temporal regulatory relationship between these two genes. As previously reported (Prasad et al., 2001), we observed that OsMADS1 was exclusively expressed in the whole wild-type spikelet meristem at stage Sp2, then in lemma and palea primordia, but not in lodicules and stamens at stage Sp6, and in lemma, palea, and carpel at stage Sp8 (Figure 3J). In spw1-1 flower, OsMADS1 transcripts were detected in the lemma and palea, as well as ectopic carpels at stages Sp6 and Sp8 (Figure 3K), similar to

OsMADS1 Specifies Lodicule and Stamen Identity Partially Independent of OsMADS16-Regulated Pathway The phenotype of homeotic transformation of two lodicules into palea-/lemma-like structures in osmads1 mutant alleles and in

Plants (no. of examined spikelets)

Lemma/Palea

Glume-like organ

Normal lodicule

Normal stamen

Lodicule– stamen

Carpel

Stigma

Wild-type (52)

2

0

2

6

0

1

2

osmads1-z (311)

3.31 ± 0.8

2.48 ± 1.4

0.81 ± 1.0

3.90 ± 1.0

0

1.11 ± 0.3

2.22 ± 0.6

spw1-1 (78)

2

2.61 ± 0.5

0

0

0

6.22 ± 0.7

12.44 ± 1.4

osmads1-z spw1-1 (225)

3.20 ± 1.1

3.31 ± 1.2

0

0

0

4.22 ± 1.0

8.44 ± 2.0

osmads3-4 (164)

2

0

3.02 ± 1.0

5.05 ± 0.9

2.31 ± 0.7

1.33 ± 0.5

2.66 ± 1.0

osmads1-z osmads3-4 (46)

3.05 ± 1.0

2.26 ± 1.1

3.64 ± 1.2

1.39 ± 1.2

2.19 ± 1.4

1.29 ± 0.5

2.58 ± 1.0

osmads58 (47)

2

0

2

6

0

1

2

osmads1-z osmads58 (26)

2.89 ± 0.7

2.60 ± 1.1

0.84 ± 0.7

2.09 ± 1.0

0

1.09 ± 0.3

2.18 ± 0.6

osmads13-3 (83)

2

0

2

6

0

1

2.57 ± 0.5

osmads1-z osmads13-3 (300)

2.87 ± 0.6

2.45 ± 1.2

0.94 ± 0.8

4.17 ± 1.0

0

1.12 ± 0.3

3.26 ± 1.2

Table 2. The Number of Floral Organs in Wild-Type and Mutant plants.

a

a

The number of floral organs was examined in type I phenotype of osmads1-z, osmads1-z spw1-1, osmads1-z osmads3-4, osmads1-z osmads58, and osmads1-z osmads13-3 mutant flowers, and the average number is shown as mean ± SD.

Molecular Plant 8, 1366–1384, September 2015 ª The Author 2015. 1369

Molecular Plant

Interactions of OsMADS1 with Floral Homeotic Genes Figure 2. Protein–Protein Interactions of OsMADS1 with OsMADS16, OsMADS3, OsMADS58, and OsMADS13. (A) Yeast two-hybrid assays show the interactions between OsMADS1 and OsMADS16, OsMADS3, OsMADS58, and OsMADS13, respectively. Serial dilutions of 103–101 transformed yeast cells were grown on SD -Trp/-Leu/His/-Ade/+X-a-gal medium containing 2.5 mM 3-AT for groups with OsMADS16 as bait and SD -Trp/-Leu/-His/-Ade/+X-a-gal medium for other groups. The transformants co-transformed with plasmids encoding OsMADS1 and OsMADS7, OsMADS1, and OsMADS8 were used as positive controls. Empty vectors pGADT7, pGBDT7 were used as negative controls. (B–E) In vitro GST pull-down assays with OsMADS1 and OsMADS16, OsMADS3, OsMADS58, and OsMADS13 proteins, respectively. His-tagged OsMADS1 produced by in vitro translation was incubated with immobilized GST or GST-OsMADS16 (B), GST-OsMADS3 (C), GST-OsMADS58 (D), and GST-OsMADS13 (E). Input, in vitro translation product. GST tag alone was used as a negative control. Western blot analysis was performed using anti-OsMADS1 antibody.

the wild-type. On the other hand, although expression levels of OsMADS16 in osmads1-z mutant lodicule and stamen primordia were comparable with those in wild-type flowers during stages Sp4 and Sp7 (Figure 3L), expression signals could not be detected until stage Sp8 when the primordium of a leaf-like organ emerged in the lodicule position (Figure 3M). This observation suggests that the expression of OsMADS16 is partially independent of OsMADS1, and other regulators probably are required for the activation of OsMADS16 expression during early flower development. Therefore, we propose that OsMADS1 and OsMADS16 might specify the lodicule and stamen identity in partially independent genetic pathways.

OsMADS1 Interacts with OsMADS3 in Promoting Flower Development The rice genome carries two duplicated class C genes, namely OsMADS3 and OsMADS58, with a conserved and partially redundant role in specifying the identity of inner reproductive floral organs and floral meristem determinacy (Kyozuka and

Shimamoto, 2002; Yamaguchi et al., 2006; Dreni et al., 2011; Hu et al., 2011). To investigate the genetic interaction between OsMADS1 and OsMADS3, we generated double mutants using our previously identified intermediate mutant of OsMADS3, osmads3-4 (Hu et al., 2011). The osmads3-4 flowers grew normal lemma and palea displaying altered number of inner floral organs (with average numbers of 3.02, 5.05, and 1.33 [n = 164] for lodicule, stamen, and carpel) and developed lodicule–stamen mosaic organs (Figure 4A and 4B; Table 2). About 12.14% of osmads1-z osmads3-4 flowers (n = 379) had elongated lemma and palea similar to osmads1-z, and abnormal lodicules, stamens, and carpel mimicking those of osmads3-4 (Figure 4C–4F and Table 2), while 74.67% flowers (n = 379) of double mutants had no inner floral organs but only extra glume-like structures at the flower center (type II flowers; Figures 1H, 4G, and 4H; Table 1). Consistent with these phenotypes, no obvious expression signals of OsMADS16, OsMADS58, and OsMADS13 were detected in type II flowers at stages Sp4–Sp8 (Figure 4I–4K). These results suggested that the combination of OsMADS1 and OsMADS3 is required for floral meristem activity maintenance and the development of inner floral organs. Moreover, normal expression of the meristematic marker gene OSH1 in the osmads1-z osmads3-4 flowers at stage Sp4 and loss of expression at later stages in those flowers without stamens and carpel developing (Figure 4L) suggested that the floral meristem activity within double mutants may be terminated earlier than in the wild-type. To elucidate the interactions between OsMADS1 and OsMADS3, mutual expression analysis

1370 Molecular Plant 8, 1366–1384, September 2015 ª The Author 2015.

Interactions of OsMADS1 with Floral Homeotic Genes

Molecular Plant

Figure 3. Phenotype Analysis of spw1-1 and osmads1-z spw1-1 Mutant Flower and In Situ Hybridization Analysis of OsMADS1 and OsMADS16. (A and B) spw1-1 flower showing normal lemma and palea (A), and glume-like organs and carpel-like organs (B). Arrowhead in (B) indicates stigma. (C and D) osmads1-z spw1-1 flower showing leafy lemma and palea (C) and carpel-like organs (D). Arrowhead in (D) indicates stigma. (E) Transverse paraffin section of osmads1-z spw1-1 flower. Arrow indicates glume-like organ. (F) Longitudinal paraffin section of osmads1-z spw1-1 flower. Arrowheads indicate carpels. (G) Scanning electron micrograph of osmads1-z spw1-1 mutant gynoecium. Arrowhead indicates stigma. (H) Scanning electron micrograph of osmads1-z spw1-1 spikelet meristem at stage Sp7. (I) osmads1-z spw1-1 spikelet shows that stamens were replaced by gynoecium in the center of twin flowers (arrowheads). (J) Expression patterns of OsMADS1 in wild-type flower at stages Sp2, Sp6, and Sp8. Arrowhead in Sp8 indicates carpel. (K) Expression patterns of OsMADS1 in spw1-1 flowers at stages Sp2, Sp6, and Sp8. Arrowheads indicate carpel. (L) Expression patterns of OsMADS16 in wild-type flowers at stages Sp4 and Sp7. In Sp4, arrows indicate stamen primordium and arrowhead indicates lodicule primordium. Arrowhead in Sp7 indicates lodicule. (M) Expression patterns of OsMADS16 in osmads1-z flowers at different developmental stages. Arrow in the middle panel indicates stamen primordium and arrowhead in the middle panel indicates lodicule primordium. Arrowhead in the right panel indicates lodicule. ca, carpel; fm, floral meristem; gll, glume-like organ; le, lemma; osm, OsMADS; pa, palea; spm, spikelet meristem; st, stamen; sti, stigma; wt, wild-type. Scale bars: 1 mm in (A–D, I); 400 mm in (E, F); 200 mm in (G); 50 mm in (H–M).

Molecular Plant 8, 1366–1384, September 2015 ª The Author 2015. 1371

Molecular Plant

Interactions of OsMADS1 with Floral Homeotic Genes

Figure 4. Phenotype Analysis of osmads3-4 and osmads1-z osmads3-4 Mutant Flower and In Situ Hybridization Analysis of OsMADS1 and OsMADS3. (A and B) osmads3-4 flower showing normal lemma and palea (A) and lodicule–stamen mosaic organ (arrowhead) (B). (C and D) osmads1-z osmads3-4 flower showing leafy lemma and palea (C) and a lodicule–stamen mosaic organ (arrowhead) (D). (E) Scanning electron micrograph of osmads1-z osmads3-4 lodicule–stamen mosaic organ (arrowhead). (F) Longitudinal paraffin section of the osmads1-z osmads3-4 flower. Arrowhead indicates lodicule–stamen mosaic organ. (G) Transverse paraffin section of the osmads1-z osmads3-4 flower showing no normal floral organ inside. (H) Scanning electron micrograph of osmads1-z osmads3-4 spikelet meristem at different developmental stages showing that no normal floral organs develop in the flower. Arrow in the right panel indicates floral meristem. (I–K) Expression patterns of OsMADS16, OsMADS58, and OsMADS13 in type II osmads1-z osmads3-4 flowers, respectively. Arrowheads indicate no expression signals in the spikelet meristem of type II osmads1-z osmads3-4 flower. (L) Expression patterns of OSH1 in type II osmads1-z osmads3-4 flower at stages Sp4, Sp6, and Sp8. Arrowheads in Sp6 and Sp8 indicate no expression signals in the spikelet meristem of the type II osmads1-z osmads3-4 flower. (M) Expression patterns of OsMADS1 in osmads3-4 flowers at stages Sp2, Sp6, and Sp8. Arrowhead in Sp8 indicates carpel. (N) Expression patterns of OsMADS3 in wild-type flowers at stages Sp4, Sp6, and Sp8. Arrows indicate stamen primordium and arrowhead indicates ovule primordium. (O) Expression patterns of OsMADS3 in osmads1-z flowers at different spikelet meristem stages. Arrows in second panel from left indicate stamen primordium. Arrowhead in rightmost panel indicates ovule primordium. fm, floral meristem; gll, glume-like organ; le, lemma; osm, OsMADS; pa, palea; spm, spikelet meristem; st, stamen; wt, wild-type. Scale bars: 1 mm in (A–D); 400 mm in (F, G); 200 mm in (E); 50 mm in (H–O).

in both single mutants was performed. In the osmads3-4 flowers, expression of OsMADS1 is concentrated in the meristem at stage Sp2 and later in the lemma, palea, and carpel (Figure 4M). On the other hand, expression of OsMADS3 in osmads1-z flowers was observed in stamen primordium, stamens, carpel primordium, and ovule primordium (Figure 4O). These phenotypes in either mutant are comparable with the

ones in the wild-type (Figures 3J and 4N). However, consistent with the delayed flower development in osmads1-z mutants (Supplemental Figure 2), the initiation of OsMADS3 expression, like OsMADS16, is later than in the wild-type (Figure 4N and 4O, left panels), suggesting that OsMADS1 might be important for the activation of OsMADS3 expression before stamen initiation through unknown mechanisms.

1372 Molecular Plant 8, 1366–1384, September 2015 ª The Author 2015.

Interactions of OsMADS1 with Floral Homeotic Genes

Molecular Plant

Figure 5. Phenotype Analysis of osmads58 and osmads1-z osmads58 Mutant Flower and In Situ Hybridization Analysis of OsMADS1 and OsMADS58. (A and B) osmads58 rice flower. Arrow in panel B indicates pistil. (C and D) osmads1-z osmads58 flower showing leafy lemma and palea (C), and an ectopic spikelet with pedicel (arrowhead) inside the flower (D). (E) Longitudinal paraffin section of the osmads1-z osmads58 flower. Arrowhead indicates an ectopic spikelet with pedicel. Arrow indicates pistil. (F) Scanning electron micrograph of the osmads1-z osmads58 spikelet meristem showing that the floral meristem does not differentiate into inner floral organ primordium in the flower. (G and H) Expression patterns of SNB in the wild-type flower at incipient emerging spikelet meristem (G) and stage Sp2 (H). (I) Expression patterns of SNB in the osmads1-z osmads58 flower. Arrowhead indicates floral meristem. (J) Expression patterns of OSH1 in the osmads1-z osmads58 flower. Arrowhead indicates floral meristem. (K) Expression patterns of OsMADS58 in the wild-type flower at stages Sp4, Sp6, and Sp8. Arrowhead in Sp8 indicates carpel. (L) Expression patterns of OsMADS58 in the osmads1-z flower at stages Sp4 and Sp8. Arrowhead indicates no expression signals in the floral meristem. (M) Expression patterns of OsMADS1 in osmads58 flowers at Sp2 and Sp8. Arrowhead indicates carpel. ca, carpel; eg, empty glume; fm, floral meristem; le, lemma; lo, lodicule; osm, OsMADS; pa, palea; pi, pistil; spm, spikelet meristem; st, stamen; wt, wild-type. Scale bars: 1 mm in (A–D); 400 mm in (E); 50 mm in (F–M).

OsMADS1 Interacts with OsMADS58 to Determine Floral Meristem Activity and Suppress Spikelet Meristem Reversion OsMADS58 has been reported in previous research to play a key role in regulating the development of reproductive organs and meristem determinacy (Yamaguchi et al., 2006; Dreni et al., 2011). Although osmads58 developed normal flowers (Figure 5A and 5B), osmads3 osmads58 double mutants have been previously reported to completely lose stamen and carpel identity and develop multiple whorls of lodicule-like organs at the flower center (Dreni et al., 2011). We found only 6.48% (n = 432) of flowers are type IV showing ectopic spikelet-like structures inside the lemma in osmads1-z single mutants. However, the majority of osmads1-z osmads58 double mutant flowers are type IV phenotypes (75.93%, n = 403) (Figure 5C–5E). SEM

also revealed that osmads1-z osmads58 floral meristem activity was prolonged without normal inner floral organ primordium formation inside the lemma and palea (Figure 5F). To confirm the synergistic role of OsMADS1 and OsMADS58 in determining the floral meristem activity, expression of SUPERNUMERARY BRACT (SNB), a gene promoting the transition from spikelet meristem to floral meristem, was examined (Lee et al., 2007). In wild-type flowers, SNB was only expressed in the incipient emerging spikelet meristems (Figure 5G) and in the primordia of rudimentary and empty glumes (Figure 5H), consistent with previous observations (Lee et al., 2007; Lee and An, 2012). However, in the osmads1-z osmads58 flowers, ectopic SNB expression was detected in the floral center (Figure 5I). Furthermore, expression of OSH1 continued in the flower center of double mutants after stage Sp8 (Figure 5J). These

Molecular Plant 8, 1366–1384, September 2015 ª The Author 2015. 1373

Molecular Plant

Interactions of OsMADS1 with Floral Homeotic Genes

results suggested that floral meristem determinacy was significantly reduced in double mutants.

at stages Sp4 and Sp7–Sp8, respectively (Supplemental Dataset 1).

To further clarify the molecular interaction between OsMADS1 and OsMADS58, in situ hybridization analysis was used to examine expression of these two genes. As previously reported (Yamaguchi et al., 2006; Dreni et al., 2011), OsMADS58 was expressed in the wild-type floral meristem after the initiation of lemma and palea primordia, and subsequently in emerging stamens and carpel primordia (Figure 5K). However, the osmads1-z mutants had no detectable expression signals of OsMADS58 in the floral meristem, but only later in the carpel (Figure 5L). On the other hand, no obvious expression changes of OsMADS1 were observed in osmads58 on early flower and carpel primordium (Figures 3J and 5M), suggesting that OsMADS1 is required for the activation of OsMADS58 expression in determining floral meristem activity and regulating reproductive organ formation.

In agreement with the function of OsMADS1 in flower development, gene ontology (GO) function enrichment analysis showed that the most significant enrichment of down-regulated genes at stage Sp4 was associated with floral organ identity specification, such as a C-class gene (OsMADS58), E-class genes (OsMADS7, OsMADS8), and AGL6-like genes (OsMADS6, OsMADS17). At stage Sp7–Sp8, down-regulated genes were mainly related to hormone signaling pathways such as jasmonic acid, ethylene, and abscisic acid. Furthermore, expression of some inflorescence development-related genes, such as DENSE AND ERECT PANICLE1 (DEP1), LAX PANICLE1 (LAX1), ABERRANT PANICLE ORGANIZATION2 (APO2), LONELY GUY (LOG), DEPRESSED PALEA1 (DP1), and OPEN BEAK (OPB), were significantly changed at stage Sp7–Sp8 (Table 3, Figure 7, and Supplemental Datasets 1 and 2).

OsMADS1 Regulates Floral Meristem Determinacy Partially Independent of OsMADS13

Compared with the recent reported microarray data obtained from 5–40-mm OsMADS1-RNAi inflorescences (Khanday et al., 2013), which covered later stages of floral organ morphogenesis, we found that expression of 680 genes differed significantly in both experiments (inflorescence in length of 5–7 mm in this study, Figure 7). The discrepancy of these two microarray data probably came from the difference in genetic background and/or research materials from different developmental stages. Among these 680 genes, 223 were up-regulated and 247 were downregulated significantly. GO analysis further revealed that hormone-related gene categories (jasmonic acid, ethylene, abscisic acid) were specifically down-regulated during late developmental stages (Figure 7 and Supplemental Dataset 2), suggesting a highly linked relationship between these hormones and flower morphogenesis. Furthermore, we used motif prediction and expression profiling assay for further analysis, and found that 16 genes might be directly regulated by OsMADS1 due to their having expression patterns similar to those of OsMADS1 (microarray data from the Rice Oligonucleotide Array Database) (Figure 7E), and the presence of ‘‘CArG’’ motifs, the consensus binding site of MADS domain proteins, within their proximal promoter regions (Supplemental Figure 3).

OsMADS13 belongs to AGL11 lineage and controls ovule identity and floral meristem determinacy (Lopez-Dee et al., 1999; Dreni et al., 2007, 2011; Li et al., 2011b). osmads13 mutant flowers developed malformed ovaries with extra stigmas, resulting from the transformation of ovules into carpelloid structures (Figure 6A and 6B) (Dreni et al., 2007; Li et al., 2011b). To investigate the genetic relationship between OsMADS1 and OsMADS13, we generated osmads1-z osmads13-3 double mutants. Most of the osmads1-z osmads13-3 flowers (81.97%, Table 1, n = 366) showed additive phenotypes on flowers compared to single mutants: the outer whorl organs mimicked those of osmads1-z while double mutants displayed reproductive organs similar to those from osmads13-3 flowers (Figure 6C–6F). Furthermore, normal spatial expression of OsMADS13 in osmads1-z ovules (Figure 6G and 6H) and unchanged expression of OsMADS1 in osmads13-3 flowers were found when compared with the wild-type (Figures 3J and 6I). It suggested that there is no direct genetic interaction between OsMADS1 and OsMADS13 in specifying floral meristem determinacy. Our in situ hybridization analysis showed normal expression of DROOPING LEAF (DL), a marker gene in the carpel (Yamaguchi et al., 2004; Dreni et al., 2007), in osmads1-z osmads13-3, compared with that in osmads1-z and osmads13-3 flowers (Figure 6J–6M), further supporting that OsMADS1 and OsMADS13 regulate floral meristem determinacy partially independently.

OsMADS1 Regulates OsMADS17 Expression To further characterize the role of OsMADS1 in specifying floral meristem identity and determinacy, we performed whole genome transcriptome analysis. Total RNA were isolated for Rice Affymetrix GeneChip analysis from inflorescences of 2 mm in length at stage Sp4 during the initiation of whorl-1 organs, and from 5–7 mm in length at stage Sp7–Sp8 when the flower structure formation was completed. Our data revealed that osmads1-z had 654 (451 up-regulated; 203 down-regulated) and 1659 (828 up-regulated; 831 downregulated) genes with at least two-fold expression change compared with wild-type (false discovery rate [FDR]