Thinking out of the box : MADS-box genes and maize spikelet development

Due to the importance of maize as an agricultural crop and its stature as an ideal model plant for the study of developmental biology in monocots, it is natural that research into its genetic structure has gained worldwide attention. Unfortunately, although much progress has been made in our understanding of the genetic control of the maize spikelet over the last decade, the depth of research in this field still lags behind that of dicots. Here, we review the developmental features of the maize spikelet and the characterization and function of MADS-box genes with the hope of stimulating further research in this area.


INTRODUCTION
Among the multitude of developmental phases in plants, flower formation is the most exciting and complex one.About 20 years ago, the ABC model for the genetic control of flower development was proposed and was initially based on the analysis of floral homeotic mutants in Arabidopsis thaliana and Antirrhinum majus.Later, the genetic regulatory network of flowering and floral meristem also began to be elucidated upon.Among these genes, transcription factors, especially MADS-box genes, play crucial roles in the whole regulation network (Krizek and Fletcher, 2005).Poaceae (grasses) is one of the most species-rich flowering plant families and includes many economically important crops.Flowers of grasses are arranged in spikelets, in which glumes, lemma, palea and lodicules are characteristic organs and serve as a basis for the classification of grasses (Schmidt and Ambrose, 1998).However, it still remains controversial whether they are equivalent to the bract, sepal and petal of dicot flowers.Thus, more research on the development of grass spikelets will provide a further understanding of the genetic control of monocot flower development and the molecular evolution of the grass-specific floral organs.*Corresponding author.E-mail: hgli20108@gmail.com.
# These authors contributed equally to this study.
Author(s) agree that this article remains permanently open access under the terms of the Creative Commons Attribution License 4.0 International License As one of the most important agricultural crops in the world, maize (Zea mays L. ssp.mays) is an important model plant for grass developmental biology because it possesses all of the following: a rich genetic history and abundant developmental mutants, a fully sequenced genome, and a good synteny with other grass species.Additionally, unlike rice producing hermaphrodite flowers, maize is a monoecious plant.Thus, research on the development of maize spikelets will contribute largely to understanding monocot plant flower development, the regulation mechanism of sex determination, and molecular evolution of the grass-specific floral organs.In this review, we summarize recent findings concerning maize spikelet development, mainly focusing on the characterization and functional study of MADS-box genes in maize flower organ development.

THE DEVELOPMENTAL STRUCTURE OF MAIZE SPIKELET
Maize is a monoecious plant that forms male and female inflorescence on a terminal tassel and on lateral ears, respectively (Figure 1A).At an early developmental stage, tassel and ear both initiate bisexual spikelets.Each spikelet contains two florets, the upper and the lower floret, which are subtended by two glumes.Each floret consists of a lemma, a palea, two lodicules, three stamens and a pistil (Figure 1B).Later in development, pistils cease to develop in the floret of male spikelets, and two unisexual florets form as a result of the pistil abortion; while in female spikelets, the lower floret and the stamens in the upper floret abort, resulting in each ear spikelet bearing a single female floret (Figure 1C).Therefore, unisexual flowers in maize are achieved by the process of selective arrest and abortion of the pistil or stamen primordium within a bisexual floret at the appropriate time (Thompson et al., 2009).

THE CHARACTERISTICS OF MADS-BOX GENES
MADS-box genes encode a family of transcription factors that play crucial roles in higher eukaryotes, especially in the regulation of floral development in flowering plants (Ciaffi et al., 2011).Previous phylogeny reconstructions revealed that the MADS-box gene family is composed of several defined gene clades.Almost all the plant MADSbox genes that are currently known are members of a monophyletic superclade of genes with a conserved structural organization, including a MADS (M-), intervening (I-), keratin-like (K-) and C-terminal (C-) domain, so called MIKC-type domain structure (Münster et al., 2002) (Figure 2).Some conserved motifs in the Cterminal domain play an important role in the formation of the MADS-box protein complex and transcription activation (Theissen et al., 1996).Many MIKC-type MADS-box genes that regulate floral development have been identified in angiosperms and can be divided into 12 major subfamilies by phylogeny reconstructions, including AG, AGL6, AGL12, AP3/PI, GGM13 (B s ), STMADS11, TM3, AGL2, AGL17, AP1/SQUA, AGL15 and FLC  (Becker and Theissen, 2003).So far, at least 32 MADSbox genes, belonging to 9 subfamilies, have been reported from maize and are widely involved in the regulation of floral organ identity, determinacy of the floral meristem, flowering time and the development of seeds (Table 1) (Zhao et al., 2011).Some maize MADS-box genes are functionally conservative with their orthologous genes in dicot plants, while the functions of others still await verification.

MAIZE HOMEOTIC MADS-BOX GENES
Studies on the two model eudicot plants Arabidopsis and Antirrhinum have led to the classic genetic ABC model that explains how three classes of genes (A, B and C) work together to specify floral organ identity.This model holds that A-class genes specify sepal fate in the first flower whorl, A plus B genes specify petals in the second whorl, B plus C genes give rise to stamens, and C genes alone are needed for carpel development in the fourth whorl (Coen and Meyerowitz, 1991).Later, the model has been expanded to incorporate D class genes, which are responsible for the development of ovules (Angenent et al., 1995), and E-class genes, which are necessary for the normal expression of A, B , C and D class genes and for the formation of functional complexes (Pelaz et al., 2000;Ditta et al., 2004).D class genes specify ovules, which will develop into seeds after pollination.Thus, some scholars define the model as -ABCDE model‖ (Goto et al., 2001).With the exception of the A-class gene AP2, all of those genes are members of the MIKCtype MADS-box family of transcription factors and they act by forming dimers and complexes of higher order.In Arabidopsis, there are two different class A genes, APETALA1 (AP1) and APETALA2 (AP2), two class B genes, APETALA3 (AP3) and PISTILLATA (PI), one class C gene, AGAMOUS (AG) (Theissen et al., 2000), and one class D gene, SEEDSTIK (STK).The class E genes in Arabidopsis are represented by SEP-like genes (also known as AGL2 subfamily), namely SEPALLATA1, 2, 3, 4 (SEP1, 2, 3, 4) (Ditta et al., 2004).Mutations of these genes will result in homeotic conversions of the regulated floral organs into organs of adjacent floral whorls, so these genes are also called homeotic genes.At present, the candidate class A, B, C, D and E genes from maize have been obtained by cDNA cloning, but it is still unclear whether the function of these genes is fully conservative with the homologous genes of dicotyledonous plants (Münster et al., 2002).Despite this, some important indications have been gained from the comparative functional analysis of rice orthologs (Cui et al., 2010;Kobayashi et al., 2010;Wang et al., 2010).

Putative A-class genes in maize
By cDNA isolation and phylogenetic sequence analysis, Münster et al. (2002) have found an orthologous relationship among ZAP1 in maize, AP1 in Arabidopsis and OsMADS15 in rice.OsMADS15 has been confirmed to control the differentiation of lemma and palea (Wang et al., 2010), two floral organs positionally orthologous to sepals, whereas until now only northern blot analysis gives us a limited clue of ZAP1 as a putative A class gene that ZAP1 was expressed in lemma, palea and lodicules, but not in anthers and carpel, similar with the expression pattern of AP1 in Arabidopsis (Münster et al., 2002).Additionally, two other MADS-box genes, ZMM4 and ZMM15, are orthologous genes of AP1 and OsMADS14 (Itoh et al., 2005).Because of the lack of the corresponding mutants, more supporting data is required before it can be determined whether or not these mentioned maize MADS-box genes have strict class A functions.

Putative B-class genes in maize
Compared to class A genes, the functions of class B genes have been proved to be conservative between monocots and eudicots.Silky1 (Si1) in maize is an orthologous gene of AP3 in Arabidopsis.Si1 was expressed in anthers and lodicules.Compared to wildtype plants, si1 mutants showed homeotic conversion of stamens to carpels and lodicules to palea/lemma-like structures, which suggest that palea/lemma in maize are homologous organs to sepals and lodicules are homologous to petals (Ambrose et al., 2000).Additionally, there are at least three class B genes in maize, ZMM16, ZMM18 and ZMM29, which are orthologous genes of PI in Arabidopsis (Whipple et al., 2004).Whipple et al. (2004) have demonstrated that ZMM16 is capable of interacting in vitro with Si1, as well as with the orthologous  (Whipple et al., 2004).Though there is still lack of studies on maize PI-genes related mutants, several lines of evidence from rice give us clues about their possible functions.Using RNAi and yeast two-hybrid strategy, OsMADS2, rice orthologous gene of ZMM16, has been proved to play an important role in lodicule and stamen development; in contrast, OsMADS4, orthologue of ZMM18 and ZMM29, mainly takes part in stamen development (Prasad and Vijayraghavan, 2003;Yoshida et al., 2007;Yao et al., 2008).Thus, at least to some extent, we can make an easy speculation that, as putative class B gene, duplicated PI clade MADS-box genes may function redundantly in maize stamen and lodicule development.

Putative C-class genes in maize
The typical class C gene of Arabidopsis AG is involved in controlling of floral determinacy and specification of carpel and stamen identity (Yanofsky et al., 1990;Coen and Meyerowitz, 1991).Such C gene activity may be diversified in maize because of gene duplication event.
Maize contains two subclades of putative C-lineage genes, ZAG1 and ZMM2/ZMM23 (Kramer et al., 2004;Zahn et al., 2006;Dreni et al., 2007).The speculation about maize C function diversification is supported by the observations from their rice orthologous genes.A knockout line of OsMADS3 (orthologous to ZMM2/ZMM23) and OsMADS58 (orthologous to ZAG1) showed that both gene function as C-class genes.However, OsMADS3 had a stronger role in specifying stamen identity and OsMADS58 was more relevant in conferring floral meristem determinacy and in regulating morphogenesis (Yamaguchi et al., 2006).Similarly, in the maize zag1 mutants, floral meristem determinacy was partially lost, whereas stamens were almost normal in male flowers, suggesting that other class C genes, such as ZMM2 and ZMM23, may be responsible for stamen specification.The expression pattern of ZAG1 and ZMM2 was consistent with this hypothesis; ZAG1 is mainly expressed in carpels, while ZMM2 is mainly expressed in anthers (Schmidt et al., 1993;Theissen et al., 1995;Mena et al., 1996).On the basis of these indications, ZAG1 probably specifies floral meristem determinacy and ZMM2/ZMM23 may specify organ identity of stamens, though mutants of ZMM2/ZMM23 have not yet been identified.

Putative D-class genes in maize
Arabidopsis contains only one D class gene STK, which is involved in ovule development and seed dispersal (Pinyopich et al., 2003).By contrast, maize has three duplicated D class genes: ZMM1 and ZAG2, together with rice OsMADS13, Brachypodium BdMADS2 and wheat TaAG-3, belongs to one subclade, and ZMM25 lies in another subclade with OsMADS13, BdMADS2 and TaAG-4, according to phylogenetic analysis of D-lineage gene among grasses (Pinyopich et al., 2003;Paolacci et al., 2007;Ciaffi et al., 2011;Wei et al., 2013).Similar to STK gene, ZAG2 primarily expressed in carple and ovule (Schmidt et al., 1993;Lopez-Dee et al., 1999).However, till now no more information have been gained about maize D class genes.Fortunately, recent studies on their grass counterparts may provide some interesting clues.Besides, expressing with a similar pattern with ZAG2, the Li et al. 4677 osmads13 knock-out mutant was completely female sterile, and its ovules were converted into a reiteration of ectopic carpels or into more amorphous structures with carpel identity (Dreni et al., 2007).Interestingly, knock-out of osmads21 had a normal phenotype; moreover, osmads21 could not modify the osmads13 phenotype (Dreni et al., 2007).These data suggest that OsMADS13 plays a role in ovule identity determination and floral meristem determinacy, while OsMADS21 has probably lost this function during evolution (Dreni et al., 2007).In line with this, expression pattern and functional divergence have also been revealed among D lineage genes in wheat and Brachypodium (Paolacci et al., 2007;Wei et al., 2013).More or less, such divergence may also lie in the putative maize D class genes.

Putative E-class genes in maize
In Arabidopsis, class E genes function as cofactors with class A, B, and C genes, and in the absence of all four SEP genes, floral organs are transformed into leaf-like structures (Pelaz et al., 2000;Ditta et al., 2004).At least eight class E genes have been identified in maize and five in rice (Zahn et al., 2005;Arora et al., 2007;Cui et al., 2010;Ciaffi et al., 2011).Sequence analysis showed that ZMM8 and ZMM14 are orthologous to OsMADS1 (LHS1) in rice (Cacharrón et al., 1999).ZMM3 is orthologous to OsMADS5, ZMM6 is orthologous to OsMADS7, ZMM27 is orthologous to OsMADS8, and ZMM24 and ZMM31 are orthologous to OsMADS34 (PAP2) (Kobayashi et al., 2010).ZMM8 and ZMM14 are found expressed in the all floral organs of the mature upper floret, but not in the lower floret.Cacharrón claimed that ZMM8 and ZMM14 work as selector genes to distinguish the upper from the lower floret during spikelet development (Cacharrón et al., 1999).In other words, ZMM8 and ZMM14 may be involved in conferring the identity or determinacy of the upper floret meristem, or they may prevent the conversion of the floret meristem into a spikelet meristem.In situ hybridization experiments revealed that ZMM6 and ZMM27 are not expressed during the vegetative growth period of maize, weakly expressed in the development of inflorescence, and strongly expressed during maize kernel development.But neither single mutant nor the zmm6 zmm27 double mutant displays any obvious abnormities in kernel or flower development, suggesting that other SEP-like genes may provide functional redundancy with ZMM6 and ZMM27 (Lid et al., 2004) .Rice has five SEP-like genes, OsMADS1, OsMADS5, OsMADS57, OsMADS8 and PAP2/OsMADS34, whose functions are similar to SEP-like genes in Arabidopsis (Agrawal et al., 2005;Prasad et al., 2005;Cui et al., 2010;Kobayashi et al., 2010).However, it is still unclear whether the precise function of SEP-like genes in maize is conservative to the orthologous genes in Arabidopsis or rice (Malcomber and Kellogg, 2004).Some research shows that the SEP-like genes in grass have a complex genetic lineage and a variety of expression patterns.These findings indicate that the SEP-like genes in grass are likely to undertake more function than their orthologous genes in Arabidopsis (Malcomber and Kellogg, 2004).However, it is still unclear whether the precise function of SEP-like genes in maize is conservative to the orthologous genes in Arabidopsis or rice (Malcomber and Kellogg, 2004).Some research shows that the SEP-like genes in grass have a complex genetic lineage and a variety of expression patterns.These findings indicate that the SEP-like genes in grass likely have additional functions to their orthologous genes in Arabidopsis (Malcomber and Kellogg, 2004).

PROSPECTS
Because of the agricultural and biological importance of maize, research on the roles of MADS-box genes in maize spikelet development has attracted worldwide attention.Although, research indicating that MADS-box genes regulate maize spikelet development has made great progress, it is still in the early stages.Many maize MADS-box genes were obtained by homologous cloning, but their expression patterns and function still remain unclear.The accumulating data from studies on MADSbox genes in dicots provide a guideline for the research of MADS-box genes in maize spikelet development.However, because of the unique floral structure of maize, whether or not the research of MADS-box genes in Arabidopsis can be applied to maize and other important crops needs to be verified by more powerful and diverse technologies.These could include isolation of maize mutants in related to MADS-box related genes, expression profiling of MADS-box genes, and the comparative study of their regulatory networks, and implementation of the maize Floral Genome Project (FGP).All of these studies will help to explore the function of MADS-box genes in maize and broaden our understanding of the molecular development and evolution of maize and other grass spikelets.Acquiring more understanding of MADS-box genes in maize spikelet development will promote awareness of the floral developmental mechanism of maize and other monocot plants.

Figure 1 .
Figure 1.Maize plant and floral development.(A) Mature maize plant.(B) Maize floral development.At an early developmental stage, the spikelet is bisexual.(C) Mature spikelet.Later in development, pistils arrest in the floret of male spikelet (up) and the stamens arrest in the floret of female spikelet (below).Pi, pistil; st, stamen; lo, lodicules; pa palea, le lemma; gl, glume.
AP3.They have also shown that maize B-class genes are capable of rescuing the corresponding Arabidopsis Bclass mutants, providing additional evidence of class B conservation