INTRODUCTION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIAL AND METHODS LITERATURE CITED

The development of highly specialized plant organs from undifferentiated meristematic cells is a complex process and requires a cascade of regulatory genes controlling e.g. the differentiation of distinct flower organs from the apical meristem (for review, see Levy and Dean, 1998). With the recent discovery of individual genes that, when deregulated, cause homeotic transformation of flower organs, underlying regulatory mechanisms have started to be illuminated. Many of these genes code for MADS box transcription factors, acting at early stages in the organ developmental program (Riechmann and Meyerowitz, 1997; Theißen et al., 2000). Since the isolation of the first plant MADS box transcription factor genes, AGAMOUS and DEFICIENS, about 10 years ago (Sommer et al., 1990; Yanofsky et al., 1990), numerous MADS box genes have been isolated from various mono- and dicotyledonous flowering plants, but also from ferns and fungi (Krüger et al., 1997; Münster et al., 1997). MADS box proteins bind to DNA at specific binding sites (CarG boxes) as homo- and/or heterodimers regulating their own transcription and that of target genes (see West et al., 1998, and references therein).

Intensive studies on mutant plants clearly demonstrated the essential, homeotic role of MADS box proteins in the development of the four distinct flower organs (sepals, petals, stamen, and carpels) and led to the formulation of the ABC model (Weigel and Meyerowitz, 1994). Because it was demonstrated that the petunia (Petunia hybrida) MADS box gene FBP11 is exclusively expressed in whorl 4 and induces ovule development on sepals when ectopically expressed, this model has been extended to the ABCD model (Colombo et al., 1995).

Detailed analyses of AGL2, 4, and 9 (renamed SEPALLATA 1, 2, and 3) recently showed that these genes represent a novel class of organ identity genes (class E). It was demonstrated that SEP3 interacts with ABC function proteins and that ternary and quartary complexes are probably the molecular basis for regulation of flower development (Pelaz et al., 2000; Honma and Goto, 2001; Theißen and Saedler, 2001). Before ABCDE genes determine organ identity of the distinct whorls, meristem identity genes regulate the transition of vegetative meristems into inflorescence and flower meristems. A third group of genes is expressed after the onset of meristem identity genes but before organ identity genes are detectable and have been referred to as intermediate or identity-mediating genes (for review, see Gutierrez-Cortines and Davis, 2000).

Functional analyses of MADS box genes have been performed mainly with plants possessing bisexual flowers, e.g. Arabidopsis and tobacco (Nicotiana sp.), for which efficient transformation systems and numerous mutants are available. Comparably few studies have been performed with plants developing unisexual flowers, e.g. maize (Zea mays). During maize ear and tassel development, male and female organs are initiated, but stamen in ear spikelets and the gynoeceum in tassel spikelets do not reach maturity (for review, see Cheng et al., 1983). Some maize MADS box genes have been isolated and exclusive expression in developing ears has been shown for ZAG2, where expression is largely restricted to developing carpels (Schmidt et al., 1993). Other maize MADS box genes are expressed in developing male and female inflorescences (Schmidt et al., 1993; Fischer et al., 1995; Mena et al., 1995, 1996; Cacharrón et al., 1999).

Plant MADS box genes are also expressed in mature flowers where they have been detected for example in the stigma, style, and ovules (Flanagan et al., 1996; e.g. Colombo et al., 1997). In addition, expression in female and male gametophytes, i.e. embryo sac and pollen, have been reported (e.g. Perry et al., 1996; Heuer et al., 2000, and references therein). MADS box gene expression in all organs and cell types participating in the fertilization process indicate that they might regulate expression of genes involved in pollen-stigma interaction, pollen tube growth/guidance, embryo sac maturation, and the onset of gene expression after fertilization. In addition, expression in zygotic and somatic embryos as well as in endosperm have been described, so that participation of MADS box proteins in regulatory processes concerning embryo and endosperm development can also be assumed (Montag et al., 1995; Filipecki et al., 1997; Perry et al., 1999; Alvarez-Buylla et al., 2000).

We are interested in the double fertilization process of higher plants and addressed the question of whether MADS box genes are expressed in the cells of the female gametophyte and at earliest stages of zygote and embryo development. Here, we present two novel MADS box genes of maize of which the expression has been studied in detail in reproductive as well as in vegetative tissues. To elucidate the function of these genes, we have ectopically expressed one gene in maize and discuss the obtained phenotype.

	RESULTS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIAL AND METHODS LITERATURE CITED

ZmMADS1 and ZmMADS3 Represent Putative MADS Box Transcription Factors

Two novel maize MADS box cDNAs, ZmMADS1 and ZmMADS3, were isolated after screening cDNA libraries of maize egg cells (ECs) and mature pollen under medium stringent conditions with the conserved MADS box region of various maize MADS box genes as a probe. Predicted amino acid (AA) sequences are illustrated in Figure 1 and are accessible at the EMBL and GenBank databases (accession no. AF112148, ZmMADS1; and accession no. AF112150, ZmMADS3). Both cDNAs encode proteins possessing the motifs typical for MIKC-type MADS box proteins (MADS box, I region, K box, and less conserved C-terminal end). A putative bipartite nuclear localization signal (KR-[X]₁₂KRR) can be outlined in the MADS box of both proteins (Fig. 1, A and B; for review, see Dingwall and Laskey, 1991). According to a SWISS-MODEL protein structure prediction (Guex and Peitsch, 1997), ZmMADS1 and ZmMADS3 proteins form an N-terminal -helical structure (N₁₃-C₃₉) and two, C-terminal adjacent -sheets (1, E₄₂-F₄₈; loop, S₄₉-K₅₃; 2, L₅₄-A₅₈; data not shown).

View larger version (95K): [in this window] [in a new window]   Figure 1.   Predicted AA sequence of ZmMADS1 and ZmMADS3, and alignment to MADS box proteins with high AA identity. The MADS domain of ZmMADS1 (A) and ZmMADS3 (B) is illustrated in light-gray and the K domain in dark-gray boxes. Conserved C-terminal regions are boxed. Gaps () were introduced to improve alignment and identical AAs are indicated by asterisks. A putative nuclear localization signal is indicated by bold, italic letters. AAs highly conserved among MADS box proteins are indicated by plus signs. Positions where conservative AA substitutions occur are indicated by dots. A Q-rich region within the C-terminal end of ZmMADS3 is indicated by bold letters.

ZmMADS1 and ZmMADS3 Belong to Distinct MADS Box Subfamilies

Comparison of ZmMADS1 and ZmMADS3 protein sequences with other MADS box proteins revealed that ZmMADS1 can be classified as a member of the TM3 subfamily of MADS box proteins, whereas ZmMADS3 belongs to the SQUAMOSA subfamily (Fig. 2). Alignments with the most homologous proteins (for accession nos., see "Materials and Methods") are illustrated in Figure 1. For ZmMADS1, AA identity is highest to the rice (Oryza sativa) clone S11905 (75%). Within the C-terminal end, a highly conserved region can be outlined in all aligned proteins (Fig. 1A). ZmMADS3 exhibits 95% overall AA identity to the maize MADS box protein ZAP1 (=MADSD; Mena et al., 1995). Substitutions are mainly conservative and both proteins additionally share Glu (Q)-rich clusters. At the very C-terminal end a cluster of nine AAs is highly conserved among aligned proteins (Fig. 1B). Using two recombinant inbred (RI) families of maize (TxCM and COxTx), ZAP1 was mapped to the long arm of chromosome 2 (2L193). We have used the same RI families and have mapped ZmMADS3 to the short arm of chromosome 7 (7S000).

View larger version (30K): [in this window] [in a new window]   Figure 2.   ZmMADS1 and ZmMADS3 belong to different MADS box subfamilies. A homology search was performed with ZmMADS1 and ZmMADS3 full-length cDNA sequences to identify MADS box genes with sequence homology. Subsequent multiple alignments were performed with protein sequences of ZmMADS1 and ZmMADS3 (gray boxes), most homologous proteins, and representatives of the MADS box subfamilies. Note that ZmMADS1 is the only maize protein within the TM3 subfamily. ZmMADS3 is a member of the SQUAMOSA subfamily and most similar to ZAP1. Names of subfamilies are given at the junctions. Bar represents 10% AA substitution per site. The tree is unrooted, bootstrap is 1,000.

ZmMADS1 and ZmMADS3 Are Expressed in ECs, Zygotes, and Somatic Embryo-Forming Cells As Well As in Stem Nodes during Vegetative Development

Single-cell reverse transcriptase (RT)-PCR analyses showed that ZmMADS1 and ZmMADS3 are both expressed in maize ECs as well as in in vivo and in vitro zygotes (Fig. 3). In contrast to ZmMADS3, ZmMADS1 transcripts are additionally detectable in synergids, central cells, and antipodals. Zmcdc2, amplified as a positive control, was detectable in all cells analyzed (Fig. 3). Northern-blot analyses (Fig. 4) revealed expression of ZmMADS1 and ZmMADS3 also in immature pistils as well as in non-pollinated and pollinated mature pistils (2 and 5 d after pollination [DAP]). However, expression of both genes is undetectable in isolated immature (stage 2) and mature embryos (Fig. 4). Analyses of distinct maize in vitro culture systems indicated ZmMADS1 expression in embryogenic suspension cultures and embryogenic type II callus (Fig. 4). More sensitive RT-PCR analyses showed that ZmMADS1 is also expressed in embryogenic type I callus and confirmed lack of expression in nonembryogenic suspension cultures. Expression of ZmMADS3 was not detectable by northern-blot analysis (Fig. 4), but RT-PCR studies showed a similar although weaker expression pattern than that of ZmMADS1 in all embryogenic cultures analyzed and expression was undetectable in nonembryogenic suspension cells (data not shown). Type II callus and suspension cultures were analyzed in more detail by RNA in situ hybridization (Fig. 5). Experiments were performed with competent type II callus, which consists of a central area with large, highly vacuolated cells and a peripheral part consisting of smaller, less vacuolated cells (Fig. 5A). In this type of callus, ZmMADS1 transcripts are mainly detectable in the peripheral zone (Fig. 5B). At 7 d after the induction of somatic embryogenesis on hormone-free medium, ZmMADS1 transcripts accumulate in developing globular structures (Fig. 5, D and E). When somatic embryo and scutellar-like structures were further differentiated, ZmMADS1 transcripts centralized to the embryo axis and outer cell layers (Fig. 5F). RNA in situ analyses of embryogenic suspension cultures showed that ZmMADS1 transcripts accumulate in sub-peripheral cell layers, most likely constituted from cells with embryogenic potential (Fig. 5H). No expression was detectable in the central part and the outermost cell layers of the cell aggregates as well as in nonembryogenic callus (Fig. 5I). Hybridization of the samples with a ZmMADS1 sense probe never gave any signal (Fig. 5C). ZmMADS3 transcripts were not detectable by in situ hybridization due to the low expression level already pointed out above.

View larger version (37K): [in this window] [in a new window]   Figure 3.   Expression of ZmMADS1 and ZmMADS3 in female gametophytic cells and zygotes. Single-cell RT-PCR analysis was performed with individual maize egg cells (ECs), synergids (SYs), central cells (CCs), and antipodal cells (APs), with primers specific for ZmMADS1 (A) and ZmMADS3 (B), respectively. Zygotes (Z) were analyzed at 24 h after pollination (hap; in vivo zygotes) and 14 to 29 h after in vitro fertilization (haf; in vitro zygotes), respectively. Maize suspension cells (BMS) served as a control for vegetative gene expression. Cells were washed four times after isolation and washing buffer (WB) of the last wash step served as control for contamination with cytoplasm of burst cells of the embryo sac, nucellus, or integument cells. Multiplex RT-PCR was performed with Zmcdc2-specific primers as a control for successful RT-PCR. DNA fragments were blotted after gel electrophoresis and hybridized to ZmMADS1-, ZmMADS3-, and Zmcdc2-specific probes. Size of DNA fragments and gene names are indicated.

View larger version (47K): [in this window] [in a new window]   Figure 4.   Temporal and spatial ZmMADS1 and ZmMADS3 expression. RNA gel-blot analyses were performed with 10 µg of total RNA of the tissues indicated and hybridized to ZmMADS1- and ZmMADS3-specific probes. As a loading control, filters were hybridized to an 18S rRNA probe. Relative RNA amounts were determined with a phosphor imager. Size of RNA fragments is indicated.

View larger version (121K): [in this window] [in a new window]   Figure 5.   Expression of ZmMADS1 during somatic embryogenesis. RNA in situ hybridization experiments were performed with type II callus before (A-C) and after (D-F) induction of embryogenesis, with a maize embryogenic (G and H) and nonembryogenic suspension culture (I). Samples were hydridized to ZmMADS1 specific probes in antisense (B, E, F, H, and I) and sense orientation (representative experiment shown in C). Arrows point at hybridization signals in developing globular structures (E), and at later stages to a cluster of cells at the embryo axis and outer cell layer (F). In embryogenic suspension cultures, signals are restricted to subperipheral cell layers (arrows in H). A signal was never obtained in nonembryogenic suspension cultures (I). Bars represent 100 µm.

Expression of ZmMADS1 and ZmMADS3 during Flower Development

The northern-blot analyses performed further revealed that ZmMADS1 and ZmMADS3 are co-expressed during ear and tassel development (Fig. 4). Maize plants develop ear primordia at several stem nodes, although depending on the variety, only one or a limited number of ears reach maturity. Analyses of immature ears isolated from nodes 5 through 7 showed that ZmMADS1 and ZmMADS3 expression is highest in the ear isolated from node 7 (Fig. 4). This corresponds to the most advanced stage of development among the ears analyzed and to the node where the fully developed ear generally appears in inbred line A188. More detailed in situ hybridization analyses of female flower development showed that transcripts of both genes are first detectable after two spikelet primordia are differentiated from the female inflorescence meristem (stage D; Fig. 6A) but not at earlier stages (stage A/C, data not shown; for comparison of flower developmental stages, see Cheng et al., 1983). Within single spikelet primordia, transcripts were detectable in the upper and the lower floret as well as in glumes (Fig. 6, B, C, and F). This pattern persisted throughout further development and transcripts were detectable in all flower organs, including the stamen primordia, which later abort in the developing ear (Fig. 6, D and G). At more advanced developmental stages, when the silk can be clearly distinguished (stage I/J), ZmMADS1 and ZmMADS3 transcripts were no longer detectable (Fig. 6E). No signals were obtained after hybridization with ZmMADS1 and ZmMADS3 corresponding sense probes (Fig. 6, H and K). Analyses of gene expression during tassel development indicated that ZmMADS1 and ZmMADS3 are not expressed in tassel primordia at very early stages of development (stage A/C; data not shown). In tassels more advanced in development (after stage G/H), ZmMADS1 and ZmMADS3 transcripts are most abundant in developing stamen (Fig. 6, I and J). As reported earlier, ZmMADS1 is expressed throughout pollen development with the highest transcript abundance in microspores (Heuer et al., 2000). Expression of ZmMADS3 was undetectable in the pistil primordia, which is still present in developing maize tassels at the stage analyzed.

View larger version (111K): [in this window] [in a new window]   Figure 6.   ZmMADS1 and ZmMADS3 expression during spikelet development. RNA in situ hybridization experiments were performed with ZmMADS1- (A-E and J) and ZmMADS3- (F, G, and I) specific RNA probes in antisense orientation. Representative sense control experiments are shown in H and K. ZmMADS1 is expressed in meristems of upper (uf) and lower (lf) ear florets and in glume (gl) primordia at developmental stage D (A and B). At stage G and H (C and D), ZmMADS1 expression is additionally detectable in developing lemmas (l), stamen (st), and the gynoeceum (gyn; gr, gynoecial ridge). At stage I/J (E), ZmMADS1 is no longer detectable. ZmMADS3 is expressed in an identical temporal and spatial pattern but signals were always less intense (F and G). During tassel development, ZmMADS1 is expressed in lodicules (lo), glumes, lemmas, and stamen (J). The ZmMADS3 expression pattern is identical but signals were not obtained in gynoeceum primordia (I).

Within vegetative organs, ZmMADS1 is most abundant in leaves (Fig. 4). Low level of expression additionally was found in root tips and internodes (data not shown). Whereas ZmMADS1 is expressed at a low level only in nodes 5 and 6 (counted from the first node above ground), ZmMADS3 is detectable in all nodes analyzed, displaying a gradient with the highest expression found in the last stem node immediately adjacent to the tassel (node 12; Fig. 4). Preliminary results from in situ hybridization experiments performed with transverse and longitudinal node sections indicate that ZmMADS3 is not expressed in vascular and parenchymatic cells, but in cell layers consisting of small, non-vacuolated cells probably representing meristematic cells (data not shown).

Ectopic ZmMADS3 Expression Affects Plant Height and Male Spikelet Development

To gain insight into putative functions of ZmMADS3, immature maize embryos were transformed with a full-length ZmMADS3 sense construct and a ZmMADS3 antisense construct under the control of the constitutive rice actin promoter. Taking the high sequence identity of ZmMADS3 and ZAP1 into account, the antisense construct used for these experiments encompassed only the 3'-untranslated region of the ZmMADS3 cDNA. Plants regenerated from these experiments were transferred into the greenhouse for further cultivation and were monitored by Southern- and northern-blot analyses until the F₃ generation. Transgenic plants that integrated the antisense construct did not display a phenotype over the generations analyzed, which might be due to the short length of the antisense construct, and were not analyzed further. The plant T0#12 (Fig. 7A) contained five copies of the sense construct and full-length transgene expression at a low level was determined by northern-blot analyses of leaves, where ZmMADS3 is not detectable in WT plants (data not shown). The transgenic plant was strongly reduced in height and developed no ear, whereas the basal region of the apical tassel developed into ear-like structures (Fig. 7A). The apical region of the tassel showed no differentiation into male spikelets. Seeds could not be obtained after pollination of the female spikelets located at the tassel with WT pollen, which prevented analyses of the progeny of this plant. The plant T0#6 (Fig. 7B) contained two integrated copies of the transgene and expression in leaves was higher than in T0#12 (data not shown). The plant was male sterile and strongly reduced in height compared with control plants transformed with the selection marker only (Fig. 7B, left). Leaf development was not affected (Fig. 7B). After pollination with WT pollen, only 11 kernels developed that germinated normally. The phenotype observed in T0 was confirmed in the progeny: Tassels of representative plants of the T2 and T3 generation are presented in Figure 7C. Progeny plants that lost the transgene due to segregation were always cultivated as control plants and developed normally (Fig. 7C, left). Plants ectopically expressing ZmMADS3 showed different levels of female and male sterility and were reduced in height (Fig. 7, B and C). Seed set was reduced, but the grains obtained after self-pollination and pollination with A188 pollen germinated normally. The reduction in height reflected a reduced number of nodes because transgenic plants developed only eight to nine nodes, in contrast to 12 nodes generally developed by WT plants in the greenhouse. Tassels of transgenic plants were smaller with a reduced number of branches in comparison with control plants (Fig. 7, C-E). More detailed analyses of the tassel of transgenic plants showed that the outer glume appeared normal (Fig. 7, E, F, and H), whereas the inner glume was reduced to a small, leaf-like structure (Fig. 7, F and H). No differentiation of lemma, stamen, lodicules, and palea was apparent in the lower and the upper male floret of transgenic plants (Fig. 7H).

View larger version (64K): [in this window] [in a new window]   Figure 7.   Ectopic expression of ZmMADS3 in transgenic maize plants. Immature maize embryos were transformed with a pAct1::ZmMADS3::nosT full-length sense construct. Transgenic plants of the T0 generation with five (plant T0#12 shown in A) and two integrated copies of the transgene (plant T0#6 shown in B, left) were reduced in height in comparison with wild-type (WT) plants (B, right) and were male sterile. Progenies of plant T0#6 were reduced in height and developed small, completely (C, left: plant T2#6.6.8), or partially sterile tassels (C, middle: plant T2#6.7.2). A progeny plant without the transgene is shown for comparison (C, right). The phenotype was confirmed in the T3 generation: No anthers dehisced from sterile spikelets (E) when control plants were at anthesis (D). Sterile transgenic spikelets developed an outer glume (og), whereas the inner glume (ig) appeared as a small leaf like structure (arrows in F). Longitudinal 2-µm sections of the regions indicated in D and F were stained with Toluidine blue (G and H). In spikelets of WT plants, lemmas (le), lodicules (lo), palea (p), and part of the filaments (f) and anthers (a) are visible (G). In transgenic spikelets, only the outer glume was differentiated, whereas the other organs were missing or not fully differentiated and leaf-like structures (arrows) developed instead (H). Bars represent 2 mm (D-F) and 300 µm (G and H), respectively.

	DISCUSSION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIAL AND METHODS LITERATURE CITED

ZmMADS3 Is Allelic to ZAP1 and Represents the ZAP1b Gene

We have characterized two novel maize MADS box cDNAs, ZmMADS1 and ZmMADS3, members of the TM3 and SQUAMOSA subfamiliy of MIKC-type MADS box proteins (Theißen et al., 2000), respectively. The high conservation of functional/structural units within the MADS and K box of ZmMADS1 and ZmMADS3 suggests that both proteins are located within the nucleus, that they can bind to DNA, and that they are capable of dimer formation. As was determined for the human SRF core homodimer, the primary DNA-binding element is an antiparallel coiled coil of two amphipathic -helices, one from each monomer (Pellegrini et al., 1995). Dimerization of the monomers is permitted by interaction of the -sheets forming a four-stranded antiparallel -sheet. These structural domains are conserved in the ZmMADS1 and ZmMADS3 proteins.

ZmMADS3 exhibits 95% AA identity to the maize MADS box protein ZAP1, which has been mapped at 2L193. A duplicated gene of ZAP1 (ZAP1b) has been predicted based on RFLP mapping analyses (Mena et al., 1995). We have mapped ZmMADS3 on 7S000, the same position determined for ZAP1b. Therefore, we propose that ZmMADS3 represents the ZAP1b gene. It was shown by Mena et al. (1995) that ZAP1 expression is excluded from mature stamen and carpels that clearly distinguishes ZAP1 from ZmMADS3, which is detectable in mature pistils. ZAP1 is not represented in the cDNA library of ECs as was determined by PCR with ZAP1-specific primers (data not shown). As a consequence of its ancestral allotetraploid origin (Leitch and Bennett, 1997), other maize MADS box genes are reported to represent duplicated genes, namely ZAG1/ZMM2, ZAG2/ZMM1, ZAG3/ZAG5, and ZMM8/ZMM14, and likewise have developed distinct expression patterns (Mena et al., 1995; Theißen et al., 1995; Cacharrón et al., 1999).

ZmMADS1 and ZmMADS3 Expression Pattern Implies a Function during Fertilization and Early Embryogenesis

Many of the MADS box genes described so far have important functions during inflorescence development and flower organ differentiation, and only relatively few data are available for MADS box gene expression in mature flowers. Transcripts of some MADS box genes have been detected in mature ovules (for review, see Riechmann and Meyerowitz, 1997), but so far AGL15 and AGL18 are the only MADS box genes shown to be expressed in the cells of the embryo sac, without further specification of the cell type (Perry et al., 1996, 1999; Alvarez-Buylla et al., 2000). Therefore, ZmMADS1 and ZmMADS3 represent the first MADS box genes for which an expression in plant ECs and zygotes has been shown. Tight temporal regulation of cell cycle regulatory genes (cyclins) in maize zygotes demonstrated de novo gene transcription before the first cell division of the zygote takes place (Sauter et al., 1998). Changes of transcript abundance in cDNA populations derived from maize in vitro zygotes additionally has been shown for distinct genes expressed in maize ECs (Dresselhaus et al., 1999). These analyses showed that zygotic gene activation in plants occurs already at the one-cell stage and therefore earlier than in animals. Transcription factors accordingly must be present regulating this transcription activity. Co-expression of ZmMADS1 and ZmMADS3 in ECs and zygotes theoretically facilitates heterodimerization/interaction of the proteins (provided that RNA and proteins are co-expressed). However, exclusive expression of ZmMADS1 in the CC, SYs, and APs suggests ZmMADS1 interaction with yet unidentified partners.

Both genes are, although at highly different levels of transcript abundance, expressed also during somatic embryogenesis of distinct in vitro culture systems analyzed. Before somatic embryos develop from callus, ZmMADS1 is expressed in cells in the periphery of the callus and is subsequently detectable in developing embryos, where transcripts are finally restricted to specific cells at the periphery and the embryo axis. This expression pattern is distinct from that observed for other MADS box genes, which are expressed in external cell layers of the radicular part in heart stage somatic embryos (CUS1) or are not restricted to specific regions (AGL15), respectively (Filipecki et al., 1997; Perry et al., 1999). Neither ZmMADS1 nor ZmMADS3 transcripts are detectable in mature zygotic embryos indicating a specific function during early stages of embryo development. Because transgenic seeds germinated normally, ZmMADS3 overexpression has no obvious effect on zygotic embryo and early seedling development.

ZmMADS1 and ZmMADS3 Are Expressed at Intermediate Stages of Flower Development

ZmMADS1 and ZmMADS3 are also co-expressed during flower development, where expression was detectable in the upper and the lower floret only at intermediate stages of development. This expression pattern is distinct from that of other maize MADS box genes. ZMM8 and ZMM14 are exclusively expressed in the upper floret of maize ear spikelets, whereas ZAG1 and ZAG2 expression is restricted to reproductive organ primordia (Schmidt et al., 1993; Cacharrón et al., 1999).

At later stages of flower development, ZmMADS1 and ZmMADS3 become undetectable, but are again expressed in mature pistils. Based on the signal intensity observed in northern-blot analyses, we assume that ZmMADS1 and ZmMADS3 are not exclusively expressed in the cells of the embryo sac, but also in surrounding nucellus and/or integument tissues. This expression pattern is similar to that of SEP1 (AGL2, Flanagan and Ma, 1994) and largely identical to that of DEFH200 and DEFH72 (Davies et al., 1996). These genes are expressed in all four whorls of floral meristems at intermediate stages, and later in development in ovules (DEFH200 and DEFH72), developing embryos, and the seed coat (SEP1), respectively. In analogy to ZmMADS1 and ZmMADS3, transcription of DEFH200 and DEFH72 is overlapping (Davies et al., 1996).

In transgenic maize plants ectopically expressing ZmMADS3, organ differentiation processes in male spikelets are prevented (except glumes), but the individual whorls are distinguishable. This phenotype suggests normal function of meristem identity genes, but absence of organ identity gene function. In cosuppression and antisense mutants of the intermediate genes FBP2 from petunia and TM5 from tomato (Lycopersicon esculentum), respectively, organ development was not prevented, but organs were phenotypically abnormal and floral meristems undetermined (Angenent et al., 1994; Pnueli et al., 1994). Functional analyses of intermediate Arabidopsis MADS box genes recently showed that SEP1/2/3 triple mutant flowers develop sepals in all whorls of indeterminate flowers (Pelaz et al., 2000), and that overexpression of SEP3 in combination with ABC function genes leads to the transformation of vegetative organs into petaloid and staminoid organs, respectively (Honma and Goto, 2001). These analyses showed that class E SEP genes interact with ABC organ identity genes. Lack of organ differentiation in plants ectopically expressing ZmMADS3 therefore might suggest that proper ternary and quartary complex formation is prevented. In an alternate manner, absence of ZmMADS3 expression at a certain developmental stage might be necessary for the function of organ identity genes. This hypothesis is supported by the finding that ZmMADS3 expression is absent during intermediate stages of flower development in WT plants.

ZmMADS1 and ZmMADS3 Are Specifically Expressed in Stem Nodes

A remarkable characteristic of ZmMADS1 and ZmMADS3 is their expression in nodes. MADS box gene expression in the stem has been reported frequently (e.g. Ma et al., 1991; Mandel and Yanofsky, 1995), and recently the expression of the barley MADS box gene BM1 in the meristematic cell layer of stem nodes and the vascular system was reported (Schmitz et al., 2000). More detailed analysis has so far been performed only with STMADS16, a MADS box gene from potato (Solanum tuberosum) that is exclusively expressed in vegetative tissue (García-Maroto et al., 2000; see below).

ZmMADS1 and ZmMADS3 expression overlap in stem node 5 and 6, but not in the more apical nodes (7-12). Furthermore, ZmMADS3 displays a gradient between the nodes and reaches an expression maximum in the uppermost node. Because expression is highest in nodes where no ear primordia is present (nodes 8-12), a node-specific function of ZmMADS3 can be assumed. The reduced number of stem nodes observed in transgenic plants indicates that ZmMADS3 overexpression influences node development. A similar phenotype was observed in 35S:STMADS16 transgenic tobacco plants, which also developed a reduced number of nodes, although plants were not reduced in height due to an increased number of internode cells (García-Maroto et al., 2000). However, the number of inflorescence branches was increased in 35S:STMADS16 plants (under long-day conditions), whereas number and size of tassel branches were reduced in most of the ZmMADS3 transgenic plants analyzed.

The precise function of ZmMADS3 cannot be determined by ectopically expressing the gene in maize and our future experiments therefore will concentrate on the study of loss of gene function after screening for ZmMADS1 and ZmMADS3 insertion mutants. Again, a transgenic antisense approach will not be a valuable tool due to the high sequence identity of the ZAP1 and ZmMADS3 genes, and an even higher gene redundancy within the ZmMADS1 gene group (data not shown). Further experiments will focus on the determination of dimerization properties of ZmMADS1 and ZmMADS3 proteins and the identification of target genes. It will be of particular interest to further characterize the role of ZmMADS1 and ZmMADS3 during the earliest events of fertilization and embryo development.

	MATERIAL AND METHODS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIAL AND METHODS LITERATURE CITED

Screening of cDNA Libraries, Sequence Analyses, and Gene Mapping

cDNA libraries of maize (Zea mays) ECs (Dresselhaus et al., 1994) and mature pollen (Heuer et al., 2000) were screened with the MADS box region of maize MADS box genes as described by Heuer et al. (2000). cDNA isolation and FASTA homology search with ZmMADS1 and ZmMADS3 full-length cDNA sequences were performed as described therein. Alignment of ZmMADS1 and ZmMADS3 homologous MADS box genes, MIKC-type maize MADS box genes, and representatives of MADS box gene subfamilies subsequently were performed at the protein level with ClustalX version 1.8 (Thompson et al., 1997) and graphically illustrated with TREEVIEW (Page, 1996). GenBank and EMBL accession nos. of proteins aligned with ZmMADS1 (accession no. AF112148) and ZmMADS3 (accession no. AF112150) are as follows: AG, X53579; AGL17, U20186; AGL20 (SOC1), T00879; ANR1, Z97057; AP1, Z16421; BpMADS3, X99653; DEF, X52023; DEFH125, Y10750; FDRMADS8, AF141965; GLO, X68831; HvM5, AJ249144; HvM8, AJ249146; LtMADS1, AF035378; LtMADS2, AF035379; OsFDRMads6, AF139664; OsMADS14, AF058697; OsMADS15, AF058698; OsRAP1B, AB041020; OsS11905, AB003328; PrMADS5, U90346: SaMADSa, U25696; SbMADS2, U32110; SEP1 (AGL2), M55551; SEP2 (AGL4), M55552; SEP3 (AGL9), AF015552; SILKY1, AF181479; SQUA, X63071; TaMADS11, AB007504; TM3, Pnueli et al., (1991); TobMADS1, X76188; ZAG1, L18924; ZAG2, L18925; ZAG3, L46397; ZAG5, L46398; ZAP1, L46400; ZMM1, X81199; ZMM2, X81200; and ZmMADS2, AF112149. The 3'-untranslated region of ZmMADS3 was amplified by PCR using the primers below and used as a probe in DNA gel blots to identify RFLPs between the parents of the inbred mapping populations CO159 × TX303 and CM37 × T232A (Burr and Burr, 1991). The resulting polymorhisms were scored within the corresponding loci placed on the Brookhaven National Laboratory map using the Map-Maker program.

Northern-Blot and Single-Cell RT-PCR Analyses

Plant material for northern-blot analyses was collected in the greenhouse from maize inbred line A188. Node samples include the complete node section plus approximately 0.5-cm apical and basal adjacent internode regions. Immature tassels were approximately 1 to 2 cm in size. Root tips were isolated from seedlings cultivated under sterile conditions in a growth chamber. RNA was isolated with TRIzol (Gibco-BRL, Karlsruhe, Germany) according to the manufacturer's specification. Northern-blot analyses were performed according to Heuer et al. (2000) with probes specific for the 3'-end of ZmMADS1 and ZmMADS3, respectively. Filters subsequently were stripped before hybridization with an 18S-rRNA probe. Relative RNA amounts were quantified with a bio-imager system (BAS-1000, Fuji, Tokyo).

ECs, SYs, CCs, APs, and in vitro zygotes were isolated from maize inbred line A188 (Green and Phillips, 1975) according to the protocols of Kranz et al. (1991, 1995). In vivo zygotes were isolated as described by Cordts et al. (2000). Multiplex RT-PCR analyses of individual cells were performed with specific primers for the 3' end of ZmMADS1 (5'-GAAGGACGACGGGATGGA-3'; 5'-CACACAACGCGATATCACAT-3') and intron-spanning primers specific for the 3' end of ZmMADS3 (5'-CTGAAGCACATCAGATCAAGA-3' and 5'-AGAGGTTTTATTCATG-CATCC-3') as described by Cordts et al. (2000). Specific amplification of Zmcdc2 served as control for successful RT-PCR (Cordts et al., 2000).

In Vitro Culture Systems

For the induction of type I callus (low embryogenic potential), zygotic maize embryos derived from crosses of maize inbred lines H99 (D'Halluin et al., 1992) and A188 were isolated 11 to 13 DAP and cultivated on modified N6⁺ medium according to Brettschneider et al. (1997). To obtain competent type II callus, immature embryos (11-13 DAP) were isolated from inbred line B73 (Iowa State University, Ames), pollinated with A188 pollen, and cultivated on N6.1.100.25 medium (Songstad et al., 1992). Calli were sub-cultivated every 2 weeks as described by Brettschneider et al. (1997) for 6 months. Somatic embryo development from type II callus was initiated by transferring calli to Murashige and Skoog medium without hormones. Embryogenic and nonembryogenic suspension cultures were started from competent type II callus and cultivated in callus maintenance medium (Emons and Kieft, 1991).

RNA in Situ Hybridization Experiments

Male and female flowers at various developmental stages were collected from maize inbred line A188 and B73. The in situ hybridization procedure basically followed the protocol provided by Dr. L. Colombo (personal communication). Samples were fixed in ethanol-acetic acid-formaldehyde medium (50% [v/v] ethanol, 5% [v/v] acetic acid, and 4% [w/v] paraformaldehyde) and embedded in paraffin (Paraplast Plus, Sigma, Taufkirchen, Germany). Eight- to 10-µm sections were digested with 1 µg mL¹ Proteinase K (Roche, Mannheim, Germany) for 30 min at 37°C. Further treatment and hybridization to gene-specific probes was performed as described by Cañas et al. (1994). In vitro culture tissues were embedded in butyl-methyl methacrylat (BMM) according to the protocol of Gubler et al. (1989). Material was fixed for 2 h in 4% (w/v) paraformaldehyde in PBS buffer (Sambrook et al., 1989) with 3- × 20-min vacuum infiltration. After washing in PBS buffer (4 × 30 min), material was dehydrated in an ethanol series (10%, 30%, and 50% [v/v] ethanol, 30 min each) at room temperature and incubated in 70% (v/v) ethanol overnight at 4°C. The material was further dehydrated in 90%, 96%, and 3× 100% (v/v) ethanol (1 h each at room temperature). BMM (40 mL of butyl-methacrylate, 10 mL of methyl-methacrylate, 250 mg of ethylbenzoine, and 10 mM dithiothreitol) infiltration was performed at room temperature with 5:1, 3:1, 1:1, 1:3 ethanol:BMM (v/v) for 2 h each step and in 100% BMM overnight before probes were transferred to Beem capsules with fresh BMM solution. BMM polymerization was performed at 20°C under long-wave UV light (8 W, TW6; N.V. Philips, Eindhoven, The Netherlands; at ±15-cm distance) for 48 h. Sections (7-9 µm) of BMM-embedded material were transferred to Super-Frost-Plus slides and BMM was removed with acetone (10 min 100% [acetone] and 5 min 50% [acetone] in water [v/v]). After washing in water and 0.05 M Tris-HCl (pH 7.6), probes were digested with 1 µg mL¹ Proteinase K (Roche) in 0.05 M Tris-HCl (pH 7.6) for 20 min at 37°C. Reactions were stopped with cold water and probes were washed three times with water and dehydrated in 70% and 100% (v/v) ethanol before hydridization to gene-specific probes as described above. Digoxigenin-labeled RNA probes were synthesized from ZmMADS1 and ZmMADS3 gene-specific 3' ends cloned into pGEM-T-vector (Promega, Mannheim, Germany). Probes were synthesized from 1 µg of plasmid at 37°C for 3 to 4 h in 40-µL assays (40 units of T7 or Sp6 RNA polymerase, Roche), 4 µL of NTP labeling mix (Roche), and 20 units of RNasin (Promega) according to the manufacturer's protocol (Roche).

Biolistic Transformation and Analyses of Transgenic Maize Plants

Full-length ZmMADS3 cDNA was cloned in sense orientation into the SmaI and KpnI restriction sites in the polylinker of the pAct1.cas vector (McElroy et al., 1995). Immature embryos from maize inbred line A188 were isolated 12 d after hand pollination and cobombarded with pAct1::ZmMADS3::nosT and p35S::pat::35ST (P. Eckes, unpublished data; Aventis, Frankfurt) containing phosphinotrycin-acetyl-transferase as the selection marker. Experimental procedures followed the protocol of Brettschneider et al. (1997), except that embryos were bombarded twice with 28 Hg inch vacuum and 36.46 ng of each plasmid. Cultivation and plant regeneration was carried out as described by Brettschneider et al. (1997). Sections of male spikelets for microscopic analyses were prepared as follows: after prefixation in 0.5% (v/v) glutaraldehyde in 0.1 M cacodylate buffer at pH 7.1 overnight at 4°C, spikelets were fixed in 2.5% (v/v) glutaraldehyde in 0.1 M cacodylate buffer at pH 7.1 for 2 h followed by six buffer rinses. The spikelets were then postfixed overnight at 4°C with 1% (w/v) OsO₄ in 0.1 M cacodylate buffer followed by four buffer rinses, dehydrated in an acetone series, and embedded in Spurr resin. Semithin sections of 2 µm were stained with 0.1% (w/v) Toluidine blue in 2% (w/v) sodium tetraborate buffer.