The Control of Spikelet Meristem Identity by the branched silkless1 Gene in Maize

See allHide authors and affiliations

Science  08 Nov 2002:
Vol. 298, Issue 5596, pp. 1238-1241
DOI: 10.1126/science.1076920


Most of the world's food supply is derived from cereal grains that are borne in a unique structure called the spikelet, the fundamental unit of inflorescence architecture in all grasses.branched silkless1 (bd1) is a maize mutation that alters the identity of the spikelet meristem, causing indeterminate branches to form in place of spikelets. We show that bd1encodes a putative ERF transcription factor that is conserved in different grasses and is expressed in a distinct domain of the spikelet meristem. Its expression pattern suggests that signaling pathways regulate meristem identity from lateral domains of the spikelet meristem.

Development in plants depends primarily on the activity of meristems, indeterminate cell populations whose derivatives form lateral organs. Meristems can be considered determinate or indeterminate, depending on whether they are consumed in the production of lateral primordia. Grass species have evolved unique meristem fates that are central to orchestrating the diverse inflorescence architecture found in this extensive family (1, 2).

Both the male and female inflorescence meristems of maize (tassel and ear, respectively) produce spikelet meristems, determinate branches that produce two glumes and two florets before undergoing sexual specialization (2). bd1 mutants show a loss of determinacy in the maize ear,with branches found in place of female spikelets (3, 4) (Fig. 1; compare A and B). In thebd1-N2355 (bd1-N) tassel, spikelets appear indeterminate and produce a series of lateral spikelets (Fig. 1; compare C and D). All five bd1 alleles examined displayed similar tassel phenotypes but differed in severity in the ear. The weakbd-mum::20250 allele initiated fewer branches in the ear, compared with the other alleles, and was able to form a few fertile florets, whereas the most severe alleles converted almost all the initial spikelets to indeterminate branches.

Figure 1

Wild-type and bd1-N mutant tassel and ear spikelets. (A) Wild-type OH43 ear with initiating pistil primordia. The ear is normally unbranched. (B)bd1 ear showing the development of multiple branches (arrowhead). (C) Single wild-type tassel spikelet. (D) Single bd1 tassel spikelet structure. Arrowheads mark spikelets that branch laterally from other spikelets. (E) Schematic of branching in wild-type maize tassels and ears. (F) SEM of wild-type tassel primordium. (G to J) SEM of bd1mutant ear. (G) Presumptive SM (marked with red star) is initiating primordia (arrowhead). (H) The SM is initiating primordia distichously. (I) The initiated primordia have SPM activity. (J) The SM appears similar to a tassel branch. (K to N) SEM of bd1 mutant tassel. (K) Early stage of SM development. Arrowhead notes axillary branching event. (L) Later stage showing development of an additional meristem (SM) in the axil of a glume. (M) Side view showing distichous phyllotaxy of initiated SM. Numbers correspond to order of initiation. (N) Dissected SM with outer glume removed to show floral organ initiation. OG, outer glume; St, stamen; C, carpel; Si, silk. Scale bar is ∼100 μm.

To establish identity of the altered meristems in bd1mutants, we carried out scanning electron microscopy (SEM). In wild type, the first 5 to 15 lateral primordia initiated by the tassel inflorescence meristem (IM) are called branch meristems (BMs) (2) (Fig. 1, E and F). The tassel IM then produces spikelet pair meristems (SPMs), which form two spikelet meristems (SMs). BMs also produce SPMs, but in a distinct two-ranked (distichous) phyllotaxy. SMs initiate an outer and inner glume, as well as two florets (fig. S1). The ear undergoes these same branching processes, except that BMs are not initiated. Thus, the identity of the meristems can be defined based on their activities: BMs form SPMs in a distichous phyllotaxy; each SPM forms two SMs; and SMs initiate a pair of glumes before forming two floral meristems.

Both the IM and SPM appear normal in bd1-N tassels and ears. Defects are first seen in the mutant after SMs are made. Inbd1-N ears, the presumptive SMs often fail to initiate an outer glume and, instead, indeterminately produce SPMs in a distichous pattern (Fig. 1, G and H). These SPMs form two SMs, each of which initiates outer glumes (Fig. 1I) and then ceases development. Given that the presumptive SMs initiate SPMs in a distichous phyllotaxy, we propose that their identity is converted to a BM identity (Fig. 1; compare F with J). From this phenotype, we hypothesize that one role of BD1 is to repress BM identity within the SMs of the ear.

In the bd1-N mutant tassel primordia, SMs initiate an outer glume (Fig. 1K), initiate a meristem in the axil of that glume, and then continue this pattern indeterminately, as revealed by SEM (Fig. 1, L and M). Later, each of these axillary meristems initiates floral organs (Fig. 1N). The identity of the mutant meristem can be considered BM-like because it initiates a series of spikelets indeterminately, although the spikelets are not in pairs. We conclude that BD1 functions to repress indeterminate branching in both inflorescences, but its absence has different consequences, possibly owing to other genetic factors.

The BD1 gene was cloned by transposon tagging withMutator (Mu) (5). The sequence flanking a cosegregating Mu1 element revealed an open reading frame that encodes a putative protein of 315 amino acids with similarity to the ethylene-responsive element–binding factor (ERF) class of transcription factors (Fig. 2A). Analysis of mutant alleles revealed a frameshift mutation upstream of the ERF domain that resulted in a premature stop codon inbd1-2, a transition mutation within the ERF domain that introduces a premature stop codon in bd1-N, and a deletion of the 3′ end of the gene that is predicted to remove the last 10 amino acids in bd1-3 (fig. S1).

Figure 2

Sequence and expression of bd1. (A) ClustalW alignment of the ERF domains of BD1, the duplicate locus BD1B, putative orthologs from other grasses, and theArabidopsis LEAFY PETIOLE and ESR1 proteins. (B) Northern blot with 2 μg of poly(A)+ RNA from maize inbred OH43 root (lane 1), leaf (lane 2), embryo (lane 3), shoot (lane 4), tassel (lane 5), and ear (lane 6) tissue probed with the 3′ UTR of thebd1 cDNA. (Bottom) Hybridization with the maizeubiquitin cDNA. (C) Northern blot with 1 μg of poly(A)+ RNA from 0.5-cm ears of OH43 (lane 1), bd1-2 (lane 2),bd1-mum::20250 (lane 3), bd1-N2355(lane 4), bd1-ref (lane 5), and bd1-3 (lane 6). (Bottom) Hybridization with the zap1a cDNA (24). (D) Maximum likelihood tree of the grass bd1-like genes excluding regions of ambiguous alignment. Numbers above branches are branch lengths (substitutions per site); numbers below branches are bootstrap values.

A related gene, bd1b, was cloned from a bacterial artificial chromosome (BAC) library and mapped to chromosome 2L, which is likely to be a segmental duplication of chromosome 7L where bd1maps (6); this suggests that it is a duplicate locus. Similar bd1 genes were cloned from rice and sorghum BAC libraries, and from Setaria italica (foxtail millet),Panicum miliaceum (common millet), Avena sativa(oats), and Eleusine coracana (finger millet) by polymerase chain reaction (PCR) with degenerate primers. All members of thebd1 gene family in grasses have complete amino acid identity within the ERF domain and 45 to 75% identity overall (Fig. 2A). A maximum likelihood tree (Fig. 2D) indicates that the genes have evolved largely in accordance with the species tree (7), which is consistent with a hypothesis of conserved function. In contrast, bd1b shows an accelerated rate of evolution, retaining an open reading frame and the conserved ERF domain, but elsewhere diverging in sequence from bd1. This rapid accumulation of mutations suggests a change or loss of function inbd1b.

The ERF domain is a plant-specific DNA binding motif (8). ERF containing transcription factors function in ethylene-mediated pathogen response as well as cold and abiotic stress responses (9). In Arabidopsis, two dominant developmental mutants result from ectopic expression of ERF genes. One is leafy petiole, which causes the leaf blade to extend down the petiole (10), and the other is tiny, which causes severe dwarfism (11). Overexpression of another relatedArabidopsis ERF gene, ESR1 (12), confers cytokinin-independent regeneration of shoots from callus tissue, which indicates that some of these genes operate in hormone-related pathways.

To determine the tissue-specific expression pattern of bd1, the 3′ UTR of the cDNA, which shares no sequence identity withbd1b, was used to probe RNA blots (Fig. 2B). A single transcript was detected only in ear and tassel tissue. Nobd1 transcript was detected in the bd1-N,bd1-2, bd1-3, and bd1-refalleles (Fig. 2C). The weak bd1-mum::20250 allele displayed reduced transcript of a slightly smaller size, consistent with studies of Mu1 insertions into 5′ untranslated leaders (13).

In situ hybridization experiments localized the bd1transcript to a semicircular domain at the glume and meristem junction in wild-type ear and tassel inflorescences (Fig. 3, A and B). Expression initiates between the SM and the outer glume and is then detected between the SM and inner glume (Fig. 3D). No expression was observed in the IMs, SPMs, or floral meristems. Whole-mount in situ hybridizations show thatbd1 expression begins soon after the first SM initiates from the SPM and becomes localized to a semicircular domain at the base of the SM above the outer glume (Fig. 3C). Later, this expression domain persists at the base of the developing florets (Fig. 3E). As controls, the sense probe used with wild-type tissue showed no signal (5), and the antisense probe on bd1-2 ears also gave no signal (Fig. 3F). We observed similar expression patterns in the SMs of sorghum and rice using the 3′ ends of their respectivebd1 genes as antisense probes (Fig. 3, G and H).

Figure 3

RNA in situ hybridization of the bd1gene. (A to F) maize bd1 antisense. (A) Longitudinal section of a wild-type tassel primordium showing expression at the outer glume/SM junction. Arrowheads denote the outer glume. (B) Longitudinal section of a wild-type ear primordium. (C) Whole mount in situ of an SPM from a wild-type ear primordium. (D) Radial section of a wild-type tassel SM showing an arc of bd1 expression at the inner glume/SM junction. (E) Wild-type ear spikelet showingbd1 expression at the base of the floral meristems. (F) bd1-2 ear. (G) Sorghum SM probed with sorghum bd1 antisense. (H) Rice SM probed with rice bd1 antisense. uf, upper floret; lf, lower floret.

We propose that BD1 functions to specify SM identity by repressing indeterminate branch fates within the lateral domain of the SM. Like other class two ERF proteins of Arabidopsis (8), BD1 may accomplish this task by acting as a transcriptional repressor. In one scenario, the zone of bd1 expression may form a boundary that prevents the ectopic expression of other meristem identity genes in the SMs. In another scenario, BD1 may repress the axillary meristem of the glume, which can secondarily alter the fate of the SMs when de-repressed.

None of these models adequately address the fact that thebd1 SM has different fates in the tassel and ear. It is unlikely that bd1b partially compensates for the loss of BD1 in the tassel, as proposed for the zag1/zmm2 duplication in maize (14), because we have been unable to detectbd1b transcript in any tissues (15). It is possible that a different tassel-specific factor may function redundantly with bd1. Given the intense selective pressure on the maize ear, it is not surprising that the ear and tassel are genetically distinct.

The expression pattern and mutant phenotype of bd1 show similarities to the FIMBRIATA/UFO genes ofAntirrhinum and Arabidopsis, respectively (16, 17). Both genes are expressed in a ring at the base of the floral meristem adjacent to the sepals, and theAntirrhinum mutant shows a partial loss of lateral determinacy within the meristem. In the case of UFO, the basal floral meristems may be replaced with coflorescence branches (18). In Arabidopsis, the UFOand LEAFY genes have been proposed to be coregulators of floral meristem identity (19). Therefore, BD1 may interact with other SM identity factors to impose determinate meristem fates. As in wild type, the maize LEAFY ortholog is expressed in the SPMs and SMs of bd1 mutants (5). However, the genetic interaction between bd1 and leafy is unknown and awaits identification of leafy mutants in maize.

To date, bd1 is the only maize mutant that specifically displays altered SM identity. Several maize mutants that affect SM determinacy have been described, such as Tasselseed6(20) and indeterminate spikelet1(21). Both these mutants display SMs that initiate more than two florets per spikelet, and interestingly, both show normal patterns of bd1 expression in the SM (fig. S1). The latter result indicates that SM identity is acquired before SM determinacy. Recently, it has been shown that SM identity and determinacy are interdependent, as two genes that control SM determinacy,indeterminate spikelet1 and indeterminate floral apex1, also show SM identity defects as a double mutant (22).

The grass spikelet is conventionally interpreted as a strongly contracted branch system—literally, a little spike (23). If this interpretation is correct, then genes should exist that, when mutated, cause the spikelet to revert to a branchlike structure. We have identified a gene that regulates spikelet versus branch meristem fates within the inflorescence of maize, and whose sequence and expression are conserved in other grasses such as rice and sorghum. Our data suggest that the expression of bd1 is fundamental to grass spikelet formation and may have played a role in the origin of this evolutionary novelty.

Supporting Online Material

Materials and Methods

Fig. S1

  • * To whom correspondence should be addressed. E-mail: gchuck{at}


View Abstract

Navigate This Article