Molecular genetic basis of pod corn (Tunicate maize)

Abstract

Pod corn is a classic morphological mutant of maize in which the mature kernels of the cob are covered by glumes, in contrast to generally grown maize varieties in which kernels are naked. Pod corn, known since pre-Columbian times, is the result of a dominant gain-of-function mutation at the Tunicate (Tu) locus. Some classic articles of 20th century maize genetics reported that the mutant Tu locus is complex, but molecular details remained elusive. Here, we show that pod corn is caused by a cis-regulatory mutation and duplication of the ZMM19 MADS-box gene. Although the WT locus contains a single-copy gene that is expressed in vegetative organs only, mutation and duplication of ZMM19 in Tu lead to ectopic expression of the gene in the inflorescences, thus conferring vegetative traits to reproductive organs.

Maize (Zea mays ssp. mays) is an important cereal crop with a history of at least 8,000–9,000 y of human domestication, which has resulted in maize being an important resource of 21st century agriculture. During this long history, some traits may have been chosen for different reasons than favorable agricultural properties, as is most likely the case for pod corn. Pod corn plants are mutant at the Tunicate (Tu) locus and show a striking phenotype. The predominant phenotypic feature of Tu maize is a foliaceous elongation of the glumes, which cover the kernels in the ears (1–4), different from other maize varieties in which glumes are not present or are invisible in the mature ear (Fig. 1). Depending on gene dosage, the Tu mutation may also cause other strong phenotypic features in male (tassel) and female (ear) inflorescences of maize, including branching and development of lower florets in the ears and development of seeds in the tassel (5). Because of its bizarre phenotype, pod corn has been of religious significance for certain native tribes of American Indians since pre-Columbian times, who believed it to have magical and curative properties (5–7). This presumably led to the propagation of a mutant by medicine men that might otherwise have been discarded as worthless (7). Pod corn was described as “Zea Mais var. tunicata” almost 2 centuries ago by the French naturalist Saint-Hilaire, who proposed that pod corn represents the natural state of maize (8). This raised a considerable, long-lasting scientific interest in pod corn, as is evident from the extensive literature on the topic (1–11). Despite this, the molecular mechanism responsible for the Tu phenotype remained elusive.

Fig. 1.

Morphology of an ear from pod corn (Lower) and WT (Upper) maize. The pod corn ear is from a heterozygous Tu/+ plant; homozygous plants have an even stronger phenotype.

MADS-box genes encode transcription factors involved in plant development. One subgroup, formed by the MIKC^c-type genes, is best known for its important role in the control of flower development (reviewed in 12). Some MIKC^c-type MADS-box genes, however, have functions outside of flower development and are mainly expressed in vegetative organs, for example, the STMADS11-like genes (13–15). MADS-box gene functions are frequently conserved between plant species and confer comparable organ identities in different species (16). In other cases, the role of MADS-box genes has diversified during evolution and changes in gene functions have given rise to morphological novelties (16).

We studied MADS-box genes in maize to identify the roles of these genes in development and evolution of a grass species. Maize is a long-standing genetic model plant with a rich resource of classic morphological mutants. We argued that knowing the chromosomal location of maize MADS-box genes would open the possibility to link individual genes to classic morphological phenotypes (17). We mapped the STMADS11-like MADS-box gene ZMM19 close to the Tu locus, which thus became a candidate gene for this locus (17).

The present study provides several lines of evidence indicating that the bizarre Tu phenotype is attributable to ectopic expression of the developmental control gene ZMM19 in the maize ear, a gene that is normally expressed only in vegetative tissue. Moreover, we show that a change in the 5′-upstream region of the gene and gene duplication are the most likely causes for the mutant phenotype of Tu.

Results

Restriction Fragment Length Polymorphism Mapping of ZMM19.

To establish the closeness of linkage between ZMM19 and the Tu locus, we performed a restriction fragment length polymorphism (RFLP) analysis of a population of 93 individuals, segregating for mutant (Tu/+) and WT plants (+/+) using a ZMM19 hybridization probe. No recombination between the Tu phenotype and the ZMM19 locus was found (Table S1), which suggests a mapping distance of 0 cM (95% confidence interval: 0–3.8 cM). This tight linkage between the ZMM19 gene and the Tu locus supports the hypothesis that ZMM19 represents the Tu locus itself.

Expression Patterns of ZMM19 in WT and Tu Plants.

Inspired by data about homeobox genes from maize, such as Knotted1 (18), Gnarley1 (19), and Rough Sheath1 (20), as well as Hooded from barley (21), we reason that the codominance (2, 11), cell autonomy (22), and gain-of-function characteristics (4) of Tu could be based on the ectopic expression of a transcription factor encoded by ZMM19 in those organs of maize that show a mutant phenotype. We tested this hypothesis by RNA gel blot analyses of our segregating population and found that ZMM19 is expressed in vegetative leaf blades and husk leaves in WT plants (Fig. 2). In all Tu mutant plants investigated (15 of 30 individuals), expression of ZMM19 in vegetative and husk leaves was about the same as in WT plants, indicating that the Tu mutation does not significantly change the expression of ZMM19 in vegetative organs (Fig. 2 and Fig. S1). In addition, however, in all mutant plants, a strong ectopic expression in male inflorescences and a very strong ectopic expression in female inflorescences were found (Fig. 2 and Fig. S1). Thus, ZMM19 is ectopically expressed in exactly those structures that show morphological changes in the mutant. We corroborated this finding by in situ hybridization studies, which revealed a somewhat patchy pattern of ZMM19 expression throughout spikelets of mutant (Tu/+) ears from early to late stages of development (Fig. 3, row A). The expression is initially stronger in lateral regions, which give rise to the glumes; however, in later stages, it also involves larger parts of lower and upper spikelet primordia. The ZMM19 expression signal differs considerably in both space and time from that of GAPDH (Fig. 3, row C) and the homeobox gene KNOTTED1 (Fig. 3, row D), thus documenting the specificity of the ZMM19 signal. ZMM19 expression appears to be completely absent in WT inflorescences comprising spikelets (Fig. 2). Female inflorescences show a stronger phenotype in Tu/+ plants than male inflorescences. The strength of phenotype and expression of ZMM19 are thus positively correlated, underpinning the hypothesis of the identity of ZMM19 and Tu. There are at least four different STMADS11-like genes in the maize genome (12). However, no strong ectopic expression in Tu mutants was detected for the genes of this subfamily that we tested: ZMM20 and ZMM26 (Fig. 2). This makes a mutation of a cis-regulatory element of ZMM19 a likely candidate for the cause of the ectopic expression of ZMM19.

Fig. 2.

Hybridization signals of ZMM19-, ZMM20-, and ZMM26-specific probes to an RNA gel blot containing total RNA from vegetative leaf blades (L), husk leaves (H), tassels (T), and ears (E) of WT plants (+/+) and heterozygous Tu mutant plants (Tu/+). ZMM20 and ZMM26 are two of the other STMADS11-like genes in the maize genome (17). The rRNA signals represent loading controls.

Fig. 3.

Expression patterns of ZMM19 in longitudinal sections through ears of heterozygous Tu mutant (Tu/+) plants as revealed by in situ hybridization. Early (Left), intermediate (Center), and late (Right) stages, respectively, of spikelet development. (Row A) Utilization of a ZMM19 antisense probe reveals strong ectopic expression of ZMM19 in a somewhat patchy pattern throughout the spikelets of mutant (Tu/+) ears. (Row B) Negative controls using a ZMM19 sense probe. (Row C) Positive controls using an antisense probe of the cytosolic GAPDH gene. (Row D) Positive controls using an antisense probe of the maize homeobox gene KNOTTED1 (18). GAPDH and KNOTTED1 signals are distinct, and hence gene-specific. Hu, husk leaf; loGl, lower glume; SpP, spikelet pair primordium; uFP, upper floret primordium. (Scale bar = 100 μm.)

Sequence Analysis of Different ZMM19 Alleles.

We next investigated the nature of this putative mutation and sequenced ZMM19 alleles from Tu/+ (pod corn) and +/+ (WT) plants. We detected a rearrangement over at least 1.5 kb in the putative promoter region of ZMM19 (Fig. S2), which might cause or contribute to the ectopic expression of the gene, and thus to the pod corn phenotype. The deviant upstream sequence shows similarity to mudrA, encoding the transposase of a MuDR-like transposable element. This aberrant putative promoter region was present in ZMM19 alleles of all 11 pod corn accessions tested, as revealed by DNA blot hybridization (Fig. S3) and sequence analysis (Figs. S2 and S4). In contrast, no such deviant promoter was found in any WT maize accession (13 tested), including a number of primitive maize lines (Fig. S5). This survey on the allelic variation in all available pod corn accessions and several WT accessions supports the identity of Tu and ZMM19. Although the perfect correlation between an aberrant allele structure and a mutant phenotype in all accessions tested represents considerable evidence in itself that mutant structure and phenotype are causally linked (21), these findings also imply that there are no independent alleles that could be used to corroborate further our conclusions of ZMM19 being Tu.

DNA Gel Blot Analysis Reveals a Compound ZMM19 Locus in Tu Plants.

Previous genetic analyses suggested that the Tu locus is compound (1–3). Two lines (“half tunicates”; Tu-md and Tu-l) exhibiting similar but weaker phenotypes than Tu mutants were isolated several times independently, and could be recombined again subsequently to reconstitute the “full” Tu (2). However, another recombinant version of Tu was isolated and called Tu-d (2). To disclose the structure of the Tu locus, we isolated and sequenced genomic clones containing the ZMM19 locus from WT plants and the different mutants. By means of comparative sequence analyses, we identified conceptual RFLPs that differentiate between the WT (wt ZMM19) and two different putative mutant copies of the ZMM19 locus, which we named ZMM19/Tu-A and ZMM19/Tu-B. The two mutant copies differ by an insertion of a Jare transposable element into the first intron of ZMM19/Tu-A and a number of point mutations (Fig. S2). DNA gel blot hybridization using suitable restriction enzymes revealed bands specific for each wt ZMM19, ZMM19/Tu-A and ZMM19/Tu-B, in Tu/+ plants (Fig. 4B). This indicates that ZMM19 is duplicated in the case of the mutant Tu allele, comprising copies ZMM19/Tu-A and ZMM19/Tu-B, as predicted by the classic genetic analyses (2). We have determined 5.8 kb of DNA sequence upstream of the translation start site of both ZMM19/Tu-A and ZMM19/Tu-B (published in sequences AJ850302 and AJ850303, respectively) without finding homology to any of the ZMM19 region, and thus conclude that the two copies must be at least 5.8 kb apart. DNA gel blot analysis (Fig. 4B) revealed that Tu-md, Tu-l, and Tu-d have indeed lost one of the mutant copies, either ZMM19/Tu-A (Tu-d and Tu-l) or ZMM19/Tu-B (Tu-md).

Fig. 4.

Correlation of mutant gene dosage with the amount of ectopic expression of ZMM19 in ears and with the strength of mutant phenotype. (A) Phenotypes of cobs of a WT (Left), Tu/+ (Center), and Tu-d/+ (Right) plant, respectively. (B) Number of ZMM19 copies in WT, Tu/+, Tu-d/+, Tu-md/+, and Tu-l/+ plants as revealed by DNA blot analysis. The positions of the bands representing the different ZMM19 genes are indicated. In addition to one WT allele of ZMM19, Tu/+ plants have two mutant gene copies in cis (ZMM19/Tu-A and ZMM19/Tu-B), whereas Tu-d/+, Tu-l/+, and Tu-md/+ plants have only one mutant gene copy (either ZMM19/Tu-A or ZMM19/Tu-B). Numbers on the left side indicate band lengths (kb). (C) RT-PCR analysis of ZMM19 expression in female inflorescences (ears). Columns show normalized expression levels, with the expression levels of ZMM19 in Tu/+ plants set to 1. Error bars represent SD.

Mutant Phenotype, Number of Mutant Components, and Strength of Ectopic Expression Are Correlated.

Classic genetic analyses demonstrated a strong correlation between the number of mutant components of Tu (mutant “gene dosage”) and the expression of the phenotype (1). We thus analyzed the correlation between the amount of ectopic expression of ZMM19 in ears (Fig. 4C) and the strength of mutant phenotype (1) (Fig. 4A), and found a good agreement. The correlation of these features to the number of mutant copies of ZMM19 (Fig. 4B) was less strict, presumably attributable to differences in the genetic backgrounds of the mutant lines. Ears of heterozygotes Tu-d/+ and Tu-l/+, having only one mutant gene copy per diploid genome, showed only a mild phenotype compared with heterozygote Tu/+ with two mutant gene copies (1) (Fig. 4A). In line with this, the level of ectopic expression of ZMM19 in ears of Tu-d and Tu-l plants was significantly reduced compared with that of Tu (Fig. 4C). Tu-md/+ plants, however, despite also having only one mutant gene copy, showed quite a strong phenotype under our growth conditions and a level of ectopic expression of ZMM19 very similar to that of Tu/+ plants (Fig. 4C).

Leaf Characteristics Are Promoted Under ZMM19 Ectopic Expression.

ZMM19 is mainly expressed in vegetative tissue (17), and we hypothesized that ectopic expression could promote vegetative characteristics. To test this, we conducted morphological analysis of glumes from Tu/+ and WT plants and found the glumes from Tu/+ plants highly populated with trichomes, a leaf sheath-like feature, in comparison to glumes from WT plants, which had nearly no trichomes (Fig. 5). Moreover, on heterologous expression of ZMM19 cDNA under control of the cauliflower mosaic virus (CaMV) 35S promoter in Arabidopsis, sepals are considerably enlarged and develop vegetative characteristics (Fig. 6). The comparative morphological analyses of WT, mutant, and transgenic plants suggest that ZMM19 expression promotes the development of leaf features, thus providing a clue concerning the function of ZMM19 in WT maize.

Fig. 5.

Abaxial epidermal surface structure of glumes from WT (Left) and Tu/+ plants (Center) compared with the surface structure of leaf sheaths from WT plants (Right). The images were obtained by SEM. C, carpel; loGl, lower glume; T, trichome; upGl, upper glume. (Scale bar = 100 μm.)

Fig. 6.

Phenotype of an Arabidopsis flower in which the ZMM19 cDNA is expressed under control of the 35S promoter of the CaMV (35S::ZMM19), compared with a WT flower. Note the enlarged, leaf-like sepals with trichomes in the transgenic plant.

Discussion

Evidence That Tu Is a Mutant Allele of ZMM19.

The agreements documented here between molecular changes at the ZMM19 locus, changes in ZMM19 expression in both space and intensity, and the pod corn (Tu) phenotype in quantitative terms provide compelling evidence that Tu is a mutant allele of ZMM19. Moreover, even though we cannot provide an analysis of multiple independent mutants, we argue that the fact that several partial phenotypic revertants of the pod corn phenotype (Tu-d/+, Tu-l/+, and Tu-md/+ plants) all show loss of one of the duplicate mutant copies (Fig. 4B) comes quite close to such kind of evidence.

The difference in the expression pattern between ZMM19 and the closely related paralogs suggests that a mutation of a cis-regulatory element of ZMM19 is responsible for the ectopic expression of ZMM19 in Tu mutants. RFLP and sequence analysis of genomic DNA from different Tu accessions from several maize collections show nearly identical aberrant promoter structures for the ZMM19 gene (Figs. S3 and S4). We thus suggest that pod corn has originated only once so far in the history of maize. In line with this, a spontaneous origin of pod corn has never been reported, despite the millions of acres of corn that are grown each year.

Structural Evolution of the Tu Locus.

We detected a rearrangement at least 1.5 kb long, and possibly longer than 5.3 kb, of the 5′-upstream promoter region of the ZMM19 allele from a Tu plant (Fig. S2). The deviant upstream sequence shows similarity to mudrA (Fig. S2), suggesting that illegitimate recombination facilitated by a transposon-derived sequence was involved in the promoter rearrangement. Similar DNA rearrangements have been described for other regulatory loci in maize encoding diverse transcription factors (23, 24). The requirement for a complex change in the promoter region may explain why no independent alleles of Tu ever occurred.

DNA gel blot analyses (Fig. 4B) revealed that the Tu locus is composed of two ZMM19 copies ZMM19/Tu-A and ZMM19/Tu-B. A compound locus was already predicted by the classic genetic analyses almost 50 years ago (2). However, ZMM19 appears to be a single-copy gene rather than a compound locus in WT maize (Fig. 4B), in contrast to previous assumptions (2).

Comparison of sequences of ZMM19/Tu-A and ZMM19/Tu-B to that of the WT copy of ZMM19 allowed us to retrace the complex mutational history of the Tu locus, first involving a rearrangement in the promoter region and then gene duplication, followed by sequence divergence of both copies, including the insertion of a Jare element (Fig. 7).

Fig. 7.

Presumed mutational history of the Tu locus. The origin of the mutant Tu locus, causing pod corn, by promoter rearrangement, gene duplication, and sequence diversification is schematically depicted. ZMM19 exons, labeled with E plus a number, are shown as boxes, with open boxes representing 5′- and 3′-UTRs and filled boxes representing the coding region. Upstream, downstream, and intron sequences are depicted as lines. Breaks in the lines indicate an omission of sequence; red flashes indicate the positions of major molecular events, question marks and the red zig-zag line indicate the foreign DNA in the ZMM19/Tu promoters, and the open triangle highlights the insertion locus of a “Jare” transposon (blue line) into the first intron of the ZMM19/Tu-A gene.

Three half tunicate lines (Tu-md, Tu-d, and Tu-l) exhibiting similar but weaker phenotypes than full Tu mutants were isolated after recombination occurred within the Tu locus (1, 2, 4). Genetic separation of Tu into three predicted components (m, d, and l), however, was never achieved (1, 4). This is now explained by our molecular data, which reveal just two copies of ZMM19 at the mutant Tu locus.

The origin of half tunicates and the reconstitution of full tunicates from half tunicates by recombination events at considerable high frequencies were striking observations of the classic genetic experiments (1, 2, 4), which the DNA gel blot hybridization experiment (Fig. 4B) now helps us to understand better. In the case of Tu-d and Tu-l, the band representing the WT copy of ZMM19 is about twice as intense as that of the mutant copy (ZMM19/Tu-B), whereas in the case of Tu-md, the bands of the mutant (ZMM19/Tu-A) and WT copy are of equal intensity (Fig. 4B). This observation suggests that in the process leading to half tunicates, recombination occurred between the single WT copy on one chromatid and the duplicate mutant copies on the other chromatid during meiosis in Tu/+ plants. This would have created two gamete types, either with single mutant copies or with chimeric duplicate loci comprising a wt ZMM19 and a mutant copy (ZMM19/Tu-B in observed cases) in cis. Fertilization of WT gametes may then have generated plants with two WT copies of ZMM19 (in trans) and a mutant copy (ZMM19/Tu-B) in the case of Tu-d and Tu-l, and a WT copy and mutant copy (ZMM19/Tu-A) in the case of Tu-md. It is quite likely that reconstitution of a full tunicate from half tunicates may have involved recombination between chimeric duplicate loci composed of WT and mutant copies of ZMM19.

ZMM19 Expression May Promote the Development of Leaf Features.

Analysis of trichome formation, texture, stomatal development, vascular system, and bundle structure revealed that the glumes of Tu plants have all the characteristics of typical leaf sheaths rather than of WT glumes (3) (Fig. 5). Because ZMM19 is strongly expressed in leaves in WT plants and ectopically expressed in glumes in pod corn, ZMM19 expression may promote the development of leaf features. On heterologous expression of ZMM19 cDNA under control of the CaMV 35S promoter in Arabidopsis, sepals are considerably enlarged and develop vegetative characteristics (25) (Fig. 6), indicating that inflationary growth of leaf-like organs (sepals and glumes) surrounding reproductive organs is a conserved response to ectopic expression of ZMM19. Similarly, ectopic expression of closely related homologs (STMADS11-like genes) of ZMM19 from Arabidopsis (SVP), potato (STMADS16), and rice (OSMADS22 and OsMADS47) in the flowers of Arabidopsis or tobacco, respectively, yields very similar reactions (14, 26–28), and heterotopic expression of the ZMM19 ortholog MPF2 brings about the “inflated-calyx-syndrome” of the “Chinese lantern” of Physalis (27), all suggesting that ZMM19-like genes promote vegetative development. Detailed morphological studies even indicated that the sepals of tobacco flowers expressing STMADS16 are transformed into vegetative leaves (14).

Three STMAD11-like genes are present in rice. They seem to encode negative regulators of brassinosteroid responses and induce vegetative growth via this pathway (29–31). Most relevantly, ectopic expression of the putative ZMM19 ortholog OsMADS22 of rice in transgenic rice plants leads to the elongation of glumes (31), somewhat mimicking the Tu phenotype and thus corroborating that ZMM19 is Tu.

Pod Corn Is Not an Ancestral Form of Maize.

Based on archaeological data, pod corn was long suspected to represent the natural state of maize (8) and the Tu locus was suspected to be responsible for controlling the switch from hard to soft glumes during maize domestication (5, 6, 9). However, the hypothesis that Tu was involved in maize evolution has been refuted by chromosomal mapping data (10). In line with this, we did not find the deviant promoter structure that is typical for Tu alleles of ZMM19 in diverse close relatives of maize collectively termed “teosintes,” among which is found the putative direct ancestor of maize, Zea mays ssp. parviglumis (Fig. S5).These findings strongly support the view that pod corn does not represent an ancestral form of maize but traces back to a unique mutational event in the promoter of ZMM19 that probably occurred only once after the domestication of maize. Except for maize, whose glumes, palea, and lemma are so reduced that the mature grain emerges naked above them, almost all other grasses (including all cereals) have their kernels enclosed in at least one kind of these organs. It is clear from our data, however, that pod corn just phenocopies this trait rather than representing an ancestral state. In line with this, it has been shown that changes at the TEOSINTE GLUME ARCHITECTURE1 locus, encoding an SBP-domain transcription factor, and thus not mutations at the Tu locus, controlled the critical changes in glume structure during the domestication of Zea mays (32).

Copy Number Variation as an Underestimated Source of Genetic Diversity in Domestication.

Little is known about the role of tandem gene duplications and loss of duplicate copies in domestication processes, but our findings concerning the Tu locus are not unparalleled. Strikingly similar cases for copy number variation are provided by the Agouti loci controlling coat color in sheep and goats (33, 34). The typical coat color of WT sheep is dark-bodied with a pale belly, but artificial selection for white fibers during domestication led to a high frequency of the white coat phenotype in certain breeds of sheep (33). Recently, it could be determined that a 190-kb tandem duplication, including the ovine agouti signaling protein gene (ASIP), brought a second copy of the ASIP gene under the control of a nearby promoter, which led to high levels of deregulated expression of ASIP and, consequently, a dominant white color phenotype (33). Moreover, reciprocal deletions generating mutant single-copy loci of ASIP were observed, which very much reminds us of the selective loss of one of the duplicate copies at the Tu locus during the origin of half tunicates (Fig. 4B). Quite similar to what we suggest for the Tu locus, a process involving recombination between duplicated copies has also been suggested as a likely route of mutant copy loss at the Agouti locus (33).

There is evidence that a similar mechanism to that of sheep is also responsible for the white coat color in goats (34). The combination of gene duplication, promoter rearrangement and deregulated expression, and duplicate mutant copy loss, as exemplified by pod corn (this work) and white sheep and goats (33, 34), may be of more general importance in domestication, and possibly also in natural evolution, than has previously been assumed.

Materials and Methods

Plant Material and Growth Conditions.

A description of all maize and teosinte accessions used is given in Table S2. Different accessions from the Maize Genetics Stock Center with Tu phenotype (e.g., Tu, Tu-md, Tu-d, Tu-l accessions) were used to compare expression levels and morphological features; however, they were in undefined genetic backgrounds. A population segregating for the Tu phenotype comprised 93 individuals and was generated by crossing a heterozygous Tu/+ plant with a WT (+/+) plant. Thirty of these plants were used for the RNA blot analyses. All plants were cultivated under standard greenhouse conditions.

DNA and RNA Blot Analysis.

Maize genomic DNA was extracted from young leaf material using a diethyldithiocarbamate sodium-based protocol (35). DNA blot analysis was performed following described methods (36). Hybridizations with radioactive-labeled DNA probes were always performed under conditions of high stringency [hybridization: 68 °C in 5× SSC, 5× Denhardt’s solution, 0.5% SDS, 1 mg/mL herring sperm DNA; washing: 68 °C in 0.1× SSPE (3.0 M sodium chloride, 0.2 M sodium hydrogen phosphate, 0.02 M EDTA, pH 7.4), 0.1% SDS].

DNA restricted with enzyme BstF5I and a probe derived from the region −11 to −327 upstream of the ATG codon of the ZMM19 gene of WT maize line T232 were used for DNA blot analysis distinguishing between different ZMM19 alleles. The same probe was used with SacI-digested DNA to screen six additional pod corn accessions and for the RFLP analysis of the 93 plants segregating for mutant (Tu/+) and WT (+/+).

Total RNA was isolated, separated by electrophoresis, and transferred to positively charged nylon membranes (Pall) as described (35). The following tissues were used for the RNA preparation: leaf blades of 6-wk-old plants, husk leaves covering 0.3- to 2.5-cm immature ears, tassels of 0.5–2.5 cm, and immature ears 0.3–2.5 cm in length. Hybridization under stringent conditions [68 °C, DIG Easy Hybridization Solution (Roche)] was performed using digoxigenin-labeled riboprobes representing the C-regions and the 3′-UTRs of ZMM19, ZMM20, and ZMM26 cDNA sequences, respectively.

Construction of Genomic DNA Libraries.

Genomic libraries of heterozygote Tu/+, Tu-d/+, Tu-md/+, and Tu-l/+ plants were produced in Lambda EMBL3 phages [primary titers were about 1.6 × 10⁶ pfu/mg (phage arms)]. The genomic DNA was prepared as described, but an additional CsCl-gradient purification step was carried out (37). BamHI-digested phage arms and Gigapack III XL packaging extracts were supplied by Stratagene. The cloning, packaging, and reamplification procedures followed the manufacturer’s instructions. In addition, the Sau3a-digested DNA fragments were size-fractionated in a sucrose gradient (36). A maize T232 genomic library cloned in Lambda DASH2 had been made previously in a similar way.

Cloning and Sequencing of ZMM19 Genomic Loci.

Radioactive-labeled probes representing different regions of the ZMM19 cDNA were used to screen the genomic libraries. Plaque hybridizations were performed under stringent conditions to avoid cross-hybridization with the closely related ZMM26 locus (hybridization: 65 °C in 5× SSC, 5× Denhardt’s solution, 0.5% SDS, 1 mg/mL herring sperm DNA; washing: 65 °C in 0.1× SSPE, 0.1% SDS). The following probes were used: (i) a probe representing the complete ZMM19 cDNA to screen the T232 library; (ii) a probe representing a small part of the 5′-UTR, the MADS-box, and about 300 bp of the first intron of the ZMM19 T232 allele to screen the Tu-, Tu-d, Tu-md, and Tu-l libraries (each made of the DNA of heterozygous plants); and (iii) a probe representing the region between the seventh exon and a region immediately downstream of the 3′-UTR of the ZMM19 T232 allele for an additional screen of the Tu library. The isolated clones were preliminarily characterized by means of PCR assay using different pairs of ZMM19-specific primers, and selected clones were sequenced via primer walking.

Promoter Isolation by RAGE.

Promoter analyses were done using the Genome Walker Kit (Stratagene), following the instructions in the manual, and sequencing. Sequences of ZMM19-specific primers are available on request.

In Situ Hybridization Studies.

In situ hybridization studies were done as generally described with digoxigenin-labeled riboprobes (35) using ZMM19 antisense cDNA. Hybridization with a ZMM19 sense probe is shown as a negative control. Hybridizations with antisense probes of GAPDH and KNOTTED1 are shown as positive controls.

Generation and Analysis of Transgenic Arabidopsis Plants.

Plants of Arabidopsis thaliana ecotype Columbia were transformed using the plant binary vector pBAR-A harboring the ZMM19 coding sequence under control of the constitutive CaMV 35S promoter. Transformation and characterization of the transgenic plants were carried out as described by He and Saedler (27).

SEM.

SEM of glumes and husk leaves of young cobs (1 cm long) of Tu/+ and WT plants was done essentially as described (37).

RT-PCR Analysis.

For RT-PCR analysis, single-stranded cDNA of equal amounts of total RNA samples of the tissue materials described above were generated. Pairs of ZMM19- and actin-specific oligonucleotide primers were used for the amplification of DNA fragments of 297 and 422 bp, respectively. Samples were taken after 28, 30, and 32 cycles and separated on agarose gels. The intensities of the obtained signals were quantified using the ImageQuant software package (Molecular Dynamics). Quantities of ZMM19 amplification products were first standardized for the signals of actin controls and then to the respective expression in ears of Tu/+ plants.

Statistics.

Mapping distance and the conservative estimate of a 95% confidence interval were calculated as described (38).