The indeterminate floral apex1 gene regulates meristem determinacy and identity in the maize inflorescence

Plant development depends largely on the activities of meristems, pools of undifferentiated cells that divide to maintain the meristem and to provide cells for lateral organs, branches, and stems (Steeves and Sussex, 1989). Indeterminate meristems continuously produce organs until senescence, whereas determinate meristems cease producing new cells after a fixed number of primordia initiate. Inflorescence meristems often exemplify indeterminate meristems by producing an indefinite number of floral meristems. Flower meristems are usually determinate; after initiating four whorls of floral organs, they are consumed in the production of carpels. Maize and other related grasses provide an excellent system to study the transition from indeterminate to determinate meristem owing to the presence of intermediate branching steps prior to the formation of flowers.

The arrangement of floral meristems on the inflorescence varies broadly. In some species, floral meristems are born directly on the inflorescence and in others they arise from lateral branches. The spacing and phyllotactic pattern of the floral meristems can also provide great diversity in form. Many species change from one phyllotactic pattern in the vegetative stage to a different pattern in the production of flowers. The inflorescence meristem of maize produces primordia in a polystichous (or multi-rowed) pattern. The early arising primordia develop into branch meristems and the later formed primordia become spikelet pair meristems (Fig. 1). The branch meristem (BM) is similar to the inflorescence meristem in the continuous production of spikelet pair meristems (SPM) but differs in the pattern of initiation, which is distichous, or two-rowed. SPMs form two spikelet meristems (SM), each of which forms two floral meristems (Fig. 1A). As a monoecious plant, maize has two types of inflorescence meristems, a terminal male inflorescence (tassel, Fig. 1D) and a lateral female inflorescence (ear, Fig. 1E) (Cheng et al., 1983; McSteen et al., 2000). The tassel is formed by conversion of the vegetative meristem into an inflorescence, whereas, the ear arises from an axillary meristem of a vegetative leaf. Both the tassel and the ear produce SPM but the production of BM is suppressed in the ear.

Like the floral meristem, both SPM and SM are determinate, producing a fixed number of derivatives. A number of mutations in maize affect the determinacy of these meristems. In ramosa 1 (ra1) mutants, SPM on the main axis of the tassel become indeterminate, producing multiple spikelets (Postlethwait and Nelson, 1964; Veit et al., 1993). SM themselves are affected in a number of mutations including indeterminate spikelet1 (ids1), Tasselseed6 (Ts6), and reverse germ orientation1 (rgo1) (Irish, 1997; Chuck et al., 1998; McSteen et al., 2000). In ids1, Ts6 and rgo1, additional flowers are produced, suggesting that the SM is indeterminate. Determinacy of the floral meristem itself is affected in the maize AGAMOUS homolog, zea agamous1 (zag1), knotted1 (kn1), and thick tassel dwarf1 (td1) (Mena et al., 1996; Kerstetter et al., 1997; McSteen et al., 2000). In each of these three mutants, more carpels are produced.

Several mutations in maize also affect the identity of meristem products. In tasselseed4 (ts4) mutants, SPM initiate SPMs and thus have the identity of BMs (Irish, 1997). In branched silkless1 (bd1) mutant ears, SM identity is altered and floral meristems are not produced (Kempton, 1934; Colombo et al., 1998). In indeterminate1 (id1), the SM is converted into a vegetative meristem after producing two flowers (Galinat and Naylor, 1951) (D. L. C. and J. Colasanti, unpublished).

The mutations previously described affect only one particular type of meristem. Here we describe a new mutation indeterminate floral apex1 (ifa1) that affects all determinate meristems of the maize inflorescence; the number of spikelets and flowers is increased, and ovule primordia proliferate. Double mutants with zag1 and ids1 reveal an additional role for ifa1 in meristem identity. The specification of meristem identity is distinct in male and female flowers.

Pollen from plants homozygous for ifa1-r (r for reference) was crossed onto the standard set of waxy-marked reciprocal translocation lines (Laughnan and Gabay-Laughnan, 1993). Segregation distortion using T1-9c showed evidence of linkage, suggesting that ifa1-r mapped to chromosome 1. Several RFLP markers were used to further define the location of ifa1 on chromosome 1S. ifa1-r mapped 3 cM from umc76 and less than 1 cM from zmm14 (zero recombinants in 110 chromosomes). ifa1-r has been introgressed four to six generations into several inbred lines including B73, Mo17, W23, A188, A619 and A632.

zag1-mum1 was a gift from R. Schmidt (University of California, San Diego). zag1 and ifa1 double mutants were generated by crossing pollen from plants homozygous for ifa1-r onto zag1-mum1/+ ears. F₁ plants were self-pollinated and F₂ progenies were scored for the segregation of both ifa1-r and zag1 phenotypes in the family. The genotype of the double mutants was verified by crossing putative double mutant plants as male to plants heterozygous for zag1 and ifa1-r or by scoring for diagnostic RFLP markers associated with each mutation. Double mutants of ifa1-r with ids1-mum1 (Chuck et al., 1998) were constructed in a similar fashion. The alleles used in constructing double mutants are likely to be null based on the fact that ifa1-r/– plants are similar to ifa1-r/ifa1-r plants when compared in the same background, and that gene-specific transcripts are absent in ids1-mum1 and zag1-mum1 (Mena et al., 1996; Chuck et al., 1998).

Young ear primordia used for RNA in situ experiments and for scanning electron microscopy were harvested from plants grown in the greenhouse. In situ hybridization was performed as described (Jackson et al., 1994) using kn1 or zag2 cDNA as probes. A 650 bp fragment downstream of the MADS box domain was amplified from a zag2 cDNA subclone using the following primer pairs: 5′-TAA TAC GAC TCA CTA TAG GCC TTA ACC TGA TCA CAG CCT-3′ and 5′-ATT TAA CCC TCA CTA AAG GCA TCT CTA AGA TCA GGG-3′, where the underlined sequences are the T7 and T3 promoter sequences, respectively. The probes were generated by in vitro transcription using T7 polymerase to generate the antisense strand and the T3 polymerase to generate the sense strand. Young ear primordia were fixed in FAA (50% ethanol, 5% acetic acid, 3.7% paraformaldehyde) at 4°C overnight and dehydrated in a graded ethanol series to 100%. The samples were critical point dried and sputter-coated with palladium. Samples were viewed on an ISI 30 scanning electron microscope at 10 kV accelerating voltage.

ifa1-r (r for reference) was identified in a maize line carrying the Ac transposon but is unlinked to Ac (data not shown). We mapped it to chromosome 1 using waxy translocation lines (Laughnan and Gabay-Laughnan, 1993) and showed that it behaves as a single recessive mutation. Crosses of ifa1-r to B-A translocation lines that result in chromosomal deficiencies (Beckett, 1993) support its position on the short arm of chromosome 1. Further mapping using RFLP markers placed ifa1 within 3 cM of umc76 and less than 1 cm from zmm14.

The phenotype of plants carrying the ifa1-r allele uncovered by a deficiency is similar to ifa1-r homozygotes suggesting that ifa1-r is a null mutation. ifa1-r has been backcrossed to several inbred lines and is fully penetrant. Aspects of the phenotype are more expressive in the original lines in which it was identified than after introgressing into a given inbred. Additional alleles have been recovered by screening F₁ populations for transposon and EMS induced alleles. The double heterozygotes are not as severe as the ifa1-r homozygotes. This paper reports the phenotype of ifa1-r.

The inflorescence meristems of the tassel and ear produce spikelet pair meristems (SPM) in a polystichous branching pattern (Fig. 1D,E). SPMs produce a pair of spikelet meristems (SM). In the tassel, the upper spikelet, which has a longer pedicel, is referred to as the pedicellate spikelet, and the lower spikelet is sessile (Fig. 2A). Each SM produces four leaf-like organs in a distichous pattern. The first two, also known as glumes, are sterile. The next two (called lemma) are fertile and contain floral meristems in their axil (Fig. 1B,C). The lemma and products of the floral meristem constitute the floret. Thus, a spikelet is composed of a pair of glumes that enclose two florets (Fig. 1B, Fig. 2B).

In ifa1-r plants, the SPM and SM become less determinate, producing more spikelets and florets (Fig. 2). The extra spikelets, as with normal spikelets, contain florets enclosed by glumes. The additional spikelet(s) in the tassel are pedicellate and initiated consecutively. Homozygous ifa1-r mutants can make more than two florets per spikelet, which like normal florets, are separated by short internodes and are contained within two sterile glumes (Fig. 2C). Extra spikelets and florets are not as frequent after repeated back-crossing into a single inbred and are more likely to be found towards the tip of the main tassel and the tips of the lateral branches. The formation of additional spikelets and florets in ifa1 mutants suggests that ifa1 regulates determinacy of the SPM and SM, respectively.

The initial steps of male and female floral development are similar, differentiating into unisexual flowers at later stages. The first organ formed from the floral meristem is the palea, a membranous leaf-like organ, followed by three lodicules (one suppressed), three stamens and a pistil (Fig. 1B,C). Lodicules are considered rudimentary petals based on expression of marker genes (Ambrose et al., 2000). The pistil is formed from three carpels (one rudimentary and two fused) that elongate to form the silk (Fig. 3A) (Nickerson, 1954). Within the fused carpels is a single ovule that consists of integuments and nucellar tissue surrounding the female gametophyte (Fig. 3B). The pistil aborts early in floret development in the tassel (Fig. 1B). In the ear, all stamens as well as the lower floret of each spikelet abort, thus producing a single female flower per spikelet (Fig. 1C).

In ifa1 mutants, the floral meristem becomes less determinate. Female flowers are characterized by short silks and undeveloped carpels that surround a mass of protruding nucellus-like tissue (Fig. 3C). The identity of the tissue as a product of the floral meristem was determined by its placement within the whorl of the aborted stamens (Fig. 3D). The severity of the protruding nucellus phenotype is greatest at the tip of the ear. Occasionally, more than one protruding nucellar mass is seen (Fig. 3D) or the ovule primordium appears to be transformed into an inflorescence structure (Fig. 3E). These transformations were seen in the original Ac transposon lines and occur at less than 0.1% of the time in any introgressed material. The protruding nucellus phenotype is 100% penetrant when ifa1-r is introgressed into all inbreds tested.

A similar phenotype was observed in florets of male spikelets. In normal flowers, three stamens are found at maturity inside the enclosing palea and lemma (Fig. 3F). In ifa1 mutants, an enlarged nucellar-like mass sometimes forms within the whorl of the three stamens (Fig. 3G). In florets closer to the tip of the tassel or tassel branches, extra stamens form in the center of the flower instead of the enlarged nucellar mass (Fig. 3H). Each of the extra stamens appear to develop in the axil of a lemma-like structure (Fig. 3I) suggesting that the extra stamens are not due to the formation of an additional whorl of stamens but rather arise from a sustained floral meristem. The nucellar mass or extra stamens is found in both the upper and lower male florets. Thus, both male and female floral meristems continue to proliferate. This result suggests that ifa1 regulates determinacy of the floral meristem.

In order to determine when ifa1-r flowers first deviate from normal, we examined developing female inflorescences from mutants (Fig. 4F-I) and their normal siblings (Fig. 4A-D) using scanning electron microscopy (SEM). Because formation of the stamen whorl is not affected in ifa1-r, we used the stage of stamen development as a standard to compare developing florets. In Fig. 4A,F, both upper and lower floral meristems have initiated. Although the lower floral meristem is initiated first (Cheng et al., 1983), it lags in development. Normal and ifa1-r floral meristems appear similar, both in size and morphology, as stamen primordia initiate. Normal florets form a ridge of carpel tissue that arises on the adaxial side (relative to the inflorescence meristem) of the floral meristem. The carpels surround the floral meristem, which at this stage terminally differentiates into an ovule primordium (Fig. 4B,C) and elongates to make the silk (Fig. 4D).

In ifa1 mutants, the carpel ridge does not enclose the ovule primordium (Fig. 4G). The enlarged and misshapen ovule primordium produces additional organs in a distichous pattern (Fig. 4H,I). These organs are presumed to be carpels because of their thickened ridge. The lower floral meristem is also affected in ifa1 mutants. In Fig. 4I, the lower floral meristem has produced one carpel and is producing a second one in an opposite position.

We also examined the origin of the extra spikelets in ifa1 mutants using SEM. As in normal plants, the SPM divides unequally to form two primordia. In ifa1-r, the larger primordium divides again into two unequal primordia one of which develops into an extra spikelet (data not shown). Fig. 4E shows a spikelet pair from a normal individual. The upper floral meristems have already initiated stamens and pistils. In the ifa1 mutant, an additional spikelet appears between a spikelet pair (Fig. 4J). Extra spikelets were always found between two normally placed spikelets (data not shown). This extra spikelet is in an opposite orientation to the other spikelets as revealed by the orientation of the carpel ridge on the upper floret (arrow in Fig. 4J). The carpel ridge, which is normally adaxial in position, is now abaxial. The extra spikelet appears to be slightly delayed in development as compared to the other spikelets. This result suggests that branching, rather than subdivision, produces the extra spikelet.

Thus, SEM reveals that many aspects of normal development persist in the ifa1 mutant. The SPM makes SMs, which make FMs, which initiate the correct complement of organs. In each case, however, the determinate meristems produce more products than normal. The distichous patterns of initiation suggest that the defect is one of determinacy rather than meristem size.

In order to evaluate the determinacy of ifa1 meristems at the molecular level, we carried out in situ hybridization with the homeobox gene, knotted1 (kn1) (Smith et al., 1992; Jackson et al., 1994). kn1 is expressed in all shoot meristems and shoot axes and is down regulated as lateral organs form. It is expressed in spikelet pair, spikelet and floral meristems but not in glumes or other floral primordia. As shown in Fig. 5A, kn1 expression in these spikelets recedes from the upper floral meristem as the ovule primordium develops. At the same time, kn1 expression is still strong in the lower floral meristem, which is in an abaxial position. Later in development, neither the silk nor the ovule expresses kn1, although the base of the floral apex and undifferentiated lower floral meristem continue to do so (Fig. 5D).

The expression pattern of kn1 in a longitudinal section of an ifa1-r inflorescence reveals the presence of an extra spikelet (* in Fig. 5B). Spikelet pairs develop side by side in a normal inflorescence, thus, only one column of spikelets can be observed per longitudinal section (Fig. 5A). The presence of an extra spikelet can be observed in a longitudinal section of an ifa1-r inflorescence because a portion of the extra spikelet overlaps with the normal spikelet pair during development (Fig. 5B). A young floral meristem is branching from one of these extra spikelets. Based on its developmental stage, this adaxially located meristem is the lower floral meristem. This observation is consistent with our SEM of an ifa1-r inflorescence (Fig. 4J), indicating that the orientation of the extra spikelet is opposite from that of normal spikelets.

We examined kn1 expression in ovule primordia of ifa1-r flowers. In wild-type floral meristems, kn1 is expressed until the ovule primordia initiate (Fig. 5D). At the stage in which the carpel ridge begins to elongate, kn1 expression forms a sharp line between the ovule primordium and the axis of the floral meristem. In ifa1 mutants, kn1 expression is also restricted to shoot meristems and not found in floral organs. Expression disappears from the ovule primordium as in normal siblings, however, expression returns in a few cells in the center of the enlarging ovule-like structure. A serial section through an ifa1-r ovule primordium reveals a small group of centrally located cells that express kn1 (Fig. 5E-H). This pattern of expression was observed in several sectioned ovules (data not shown). Thus, ifa1 functions directly or indirectly, to negatively regulate kn1 in a subset of cells at the center of the ovule primordium.

As kn1 expression disappears in incipient ovule primordia, carpel and ovule specific genes, such as zea agamous2 (zag2), are expressed (Schmidt et al., 1993). We utilized zag2, as an in situ hybridization probe to determine whether the ifa1-r primordium gained ovule identity. As in wild-type female flowers, zag2 mRNA was detected in the ovule primordium and inner carpel wall of the ifa1-r pistil (Fig. 6A,B). Although the ifa1-r primordium is greatly enlarged compared to normal ovules, zag2 expression is still visible. We also observed megaspore mother cells in some ifa1-r ovule primordia (data not shown). These results indicate that the floral meristem in ifa1 mutants has acquired ovule-like identity. Thus, the expression of kn1 in a few cells of the ifa1-r ovule primordium is not simply due to a failure to differentiate into ovule tissue. It appears that a subset of ifa1-r ovule primordium cells regain, or retain meristematic identity.

To gain further insights into the function of ifa1 in meristem determinacy, we constructed double mutants between ifa1-r and two other mutations that affect floral and spikelet meristem determinacy, zag1-mum1 and ids1-mum1. Mutations in zag1, a maize homologue of AGAMOUS (Yanofsky et al., 1990) in Arabidopsis and PLENA in Antirrhinum (Bradley et al., 1993), result in a loss of determinacy of the female floral meristem (Mena et al., 1996). zag1 mutant flowers form extra carpels and extra silks (Fig. 7B), and occasionally form a mass of tissue within the carpel whorl. Mutations in zag1 do not affect male flower morphology.

Female flowers of ifa1 zag1 double mutants reveal a phenotype not found in either single mutant (Fig. 7A,B). All female flowers in plants defective for ifa1 and zag1 develop a branching structure at the center of the flower (Fig. 7C,D). The branching structures form spikelet pairs in a distichous pattern and therefore are considered BM. These secondary BM arise within the whorl of stamens (data not shown). Flowers produced by the secondary BM reiterate the defective phenotype, giving rise to tertiary branches. Consistent with this indeterminate phenotype, kn1 is expressed in the SPM that arise from this ectopic BM (Fig. 5C). The phenotype is completely penetrant and very expressive in both inbred and mixed backgrounds (data not shown).

The tassels of ifa1 zag1 double mutants (Fig. 7F) are clearly distinct from zag1-mum1 tassels, which display normal morphology, and ifa1-r tassels, which are slightly bushier (Fig. 7E,F). Dissection of a floret from a double mutant tassel, shown in Fig. 7G, reveals an ectopic spikelet within the stamen whorl. This spikelet is in the position of the pistil, which normally aborts in male flowers. The ectopic spikelet has leaf-like glumes not observed in normal spikelets. When the leaf-like glumes of the ectopic spikelet are opened, several stamens enclosed by additional leaf-like organs are revealed (Fig. 7H). This phenotype suggests that the ectopic spikelet made multiple florets. Thus ifa1 and zag1 independently regulate floral meristem determinacy, but act redundantly to regulate the identity of the floral meristem. When both ifa1 and zag1 genes are mutated, the floral meristem converts to a branch meristem in the ear and to a spikelet meristem in the tassel.

indeterminate spikelet1 (ids1), which encodes an AP2-like protein, controls spikelet meristem determinacy in maize (Chuck et al., 1998). The ids1-mum1 allele results in multiple florets forming from the spikelet meristem. This phenotype is similar to one aspect of the ifa1-r phenotype (Fig. 2C). Both female and male florets are affected in ids1 mutants. Florets in the ear make silks (Fig. 8A) but are usually infertile (Chuck et al., 1998).

Ears of the ifa1 ids1 double mutant show a phenotype that is distinct from either single mutant (Fig. 8A-D). A highly branched structure arises from within the spikelets. Dissecting a spikelet pair (Fig. 8E) and spikelet (Fig. 8F) reveals the basis of the highly branched phenotype; multiple florets and a branching structure form within the enclosing glumes of the spikelet (Fig. 8F). This branched structure forms spikelet pairs in a distichous pattern and thus is considered to be a branch meristem. Florets on the secondary branch reiterate the ifa1-r floral phenotype. Both spikelets are affected (Fig. 8E).

Tassels of ifa1 ids1 double mutants show a highly branched phenotype in which spikelets form within the spikelets (Fig. 8G). Fig. 8I shows a single spikelet that contains spikelets emerging from within the encircling glumes. The presence of spikelets arising within spikelets is found in both the pedicellate and sessile spikelet as well as in extra spikelets that form (Fig. 8H). The ectopic spikelets develop singly and consecutively, suggesting that they arise from a spikelet pair meristem. Thus, ifa1 and ids1 act independently to maintain the determinacy of the SM and redundantly to maintain its identity. The fate of the SM depends on whether it is born on a male or female inflorescence. In the double mutant, the ear SM becomes a BM, whereas, in the tassel, the SM becomes a SPM.

SEM was used to follow the meristems as they initiate in the double mutants (Fig. 9). As shown in Fig. 9B, ids1 mutants make extra florets in a spikelet, which arise in a distichous branching pattern. The extra floret is between the upper and lower floret. In ifa1 ids1 double mutants, a bulging meristem forms between the upper and lower floral meristem (Fig. 9C). The spikelet on the right (Fig. 9C) is slightly more advanced and reveals an elongating meristem that appears to be initiating spikelets. The younger spikelet on the left clearly shows that this meristem arises between the two florets. SEM of the ifa1 zag1 double mutant shows an abnormal upper and lower floral meristem that has initiated a few ridges (arrows in Fig. 9D). No meristems arise between the upper and lower flower. These results emphasize the fact that the secondary inflorescences in ifa1 zag1 or ifa1 ids1 mutants arise from different locations.

We also examined expression of both zag1 and ids1 in ifa1 mutants and found that their expression patterns were indistinguishable from wild type (data not shown). These results support the genetic analysis that failed to show an epistatic interaction.

We have discovered a gene in maize that regulates meristem determinacy in the inflorescence. Maize has three types of determinate meristems, the spikelet pair meristem (SPM), the spikelet meristem (SM) and the floral meristem (FM). The ifa1 mutation affects the floral meristem most severely, but also affects the other two determinate meristems. In an ifa1 single mutant, each determinate meristem produces additional subsequent meristems or organs that are normally specified. However, in a background in which either zag1 or ids1 is mutated, each affected meristem produces its normal organs and then converts to earlier meristem types. Which meristem converts, and to what, is genetically controlled. The conversions are distinct without gradients of transformation.

Flower meristems produce four whorls of organs in most Angiosperm flowers, starting with sepals as the outer whorl, followed by petals, stamens and lastly, carpels. Carpels produce ovules, which contain the female gametophyte. Ovules are regarded by some as a fifth whorl (Angenent et al., 1994; Ray et al., 1994). In maize, the fused carpels constitute the pistil, which contains a single ovule that is the terminal product of the floral meristem (Sharman, 1945; Barnard, 1964). In male flowers, the pistil aborts and in female flowers, the stamens are arrested in growth (Cheng et al., 1983; DeLong et al., 1993).

ifa1 mutants appear normal at early stages of floral development with properly specified floral organs. They begin to differ from wild type as the carpels form and the ovule primordium is initiated. The carpels fail to enclose the ovule primordium, which continues to expand. The expanding ovule primordium has many features of a meristem although it still expresses the carpel/ovule marker, zag2 (Schmidt et al., 1993). Male flowers of ifa1 mutants are also defective at the floral apex. In ifa1 mutants, a protruding nucellar-like mass or additional male florets are occasionally found in the position normally occupied by the aborted pistil. The frequency of indeterminacy is not as high in male flowers as it is in female flowers, but dramatically increases when the pistil does not undergo abortion. Such a condition occurs in mutants such as tasselseed2 (ts2) and Tasselseed5 (Ts5) that have feminized flowers on the tassel (Irish and Nelson, 1989; DeLong et al., 1993; Veit et al., 1993). In ifa1 ts2 or ifa1 Ts5 double mutants, all the tassel florets have a protruding nucellus-like mass within the stamen whorl (data not shown). Thus the indeterminate nature of the floral meristem may override abortion of the pistil in the male flower and is enhanced when pistil abortion is suppressed.

zag1 also regulates floral meristem determinacy (Mena et al., 1996). It is first expressed in floral meristems and then in stamens and carpels, similar to the expression of AG in Arabidopsis (Yanofsky et al., 1990). Female florets of zag1 mutants have an indeterminate floral meristem in which extra carpels are produced (Mena et al., 1996). This defect is similar to one aspect of ifa1 mutants. In ifa1 zag1 double mutants, the floral meristem is converted to a branch meristem in female flowers and converted to a spikelet in male flowers after production of normal whorls of floral organs (Fig. 10). This phenotype reveals additional roles for both zag1 and ifa1 in female and male flowers. Each gene functions independently to maintain determinacy of the female floral meristem but functions redundantly to maintain its identity. Although zag1 plays no apparent role in male flowers, both zag1 and ifa1 function redundantly to maintain identity of the floral meristem in male flowers as well.

The penetrance of the large nucellus phenotype in ifa1 single mutants is 100%, as is the penetrance of the branch meristem phenotype in ifa1 zag1 double mutants. Occasionally, a single kernel on an ifa1 mutant ear may show the branching phenotype. While this finding could suggest that ifa1 alone can regulate identity, it is also possible that zag1, or genes downstream of zag1, fail to function at some stochastic frequency.

A synergistic interaction also occurs in ifa1 ids1 double mutants. ids1 single mutants have extra florets in both male and female spikelets (Chuck et al., 1998). ids1 is expressed in immature floral organs, spikelet pair and spikelet meristems. In the ifa1 ids1 double mutant, the SM in the ear produces floral meristems as usual but then is converted to a branch meristem. The SM in the tassel is also converted after producing floral meristems. This male meristem produces a series of spikelets that are initiated singly and thus may be considered a SPM.

Despite the effect of ifa1-r on all determinate meristems, ids1-mum1 and zag1-mum1 limit the conversion seen in double mutants to a single meristem type. ids1 plays a role in SM determinacy and thus the double mutant affects conversion of the SM. zag1 plays a role in floral meristem determination and thus the double mutant affects floral meristem conversion (Fig. 10). Whether the meristems of the double mutants revert to branch, spikelet pair or spikelet, the reversions are complete and not a mixture of meristem types. This result suggests that the identity of each meristem type is distinctly defined. This finding is intriguing given that ifa1 affects all meristems and ids1 is expressed at all stages, yet their dual absence affects one meristem type most significantly. It suggests that some other unidentified genes must function only at the SM stage.

Male and female conversions are distinct in both double mutants (Fig. 10). In each case, the converting female meristem becomes a branch meristem and the converting male meristem becomes a spikelet or spikelet pair meristem. We also found that the severity of the ifa1-r phenotype differed between male florets that were feminized or not. Other maize mutants are also more severe in the ear than the tassel. For example, bd1 mutant ears have SM that produce BM and never reach the stage of floral meristem. bd1 mutant tassels, on the other hand, are capable of producing floral meristems (G. Chuck, S. H. and R. Schmidt, unpublished) (Colombo et al., 1998). The difference between male and female inflorescence may result from different hormone levels or different sensitivities to hormones due to the different positions on the plant (Dellaporta and Calderon-Urrea, 1994; Irish, 1996).

A classic example of meristem conversion is found in Impatiens balsamina (Battey and Lyndon, 1990). Impatiens requires constant short days for flowering. If plants are returned from inductive short days to long days, the floral meristem begins producing vegetative leaves from the floral meristem. Chimaeric organs that are part petal and part leaf sometimes form. A switch back to short days results in the immediate production of petals again. Reversion to vegetative development is also seen in the indeterminate1 (id1) mutation of maize (Colasanti et al., 1998). id1 mutants fail to flower under most normal growing seasons and produce two to three times more than the usual number of leaves. Eventually, the shoot apical meristem is converted to a tassel that is highly abnormal. Ears do not form on these plants. Dissection of an id1 tassel shows that vegetative seedlings are initiated from the spikelet meristem after initiation of two florets (Galinat and Naylor, 1951) (D. L. C. and J. Colasanti, unpublished). Thus, the id1 spikelet meristem has converted to a vegetative meristem.

A number of mutations affect determinacy and identity of the flower meristem in Arabidopsis. AGAMOUS (AG) plays a crucial role in floral meristem determinacy in Arabidopsis (Bowman et al., 1989). Under certain conditions, the ag floral meristem produces an inflorescence meristem at the center of the carpel whorl similar to ifa1 zag1 double mutants. Okamura and colleagues showed that agamous homozygotes or leafy (lfy) heterozygotes in short day conditions produce shoots from within the carpel whorl (Okamuro et al., 1993). Similar results were obtained using ag constans double mutants (Mizukami and Ma, 1997). CONSTANS regulates LFY (Simon et al., 1996), which itself is an activator of AG (Busch et al., 1999), thus explaining why limiting amounts of LFY might result in the same phenotype. Short day conditions delay the activation of LFY (Busch et al., 1999), and so perhaps also reduce the amount of AG.

Recently, two groups have shown that LFY together with WUSCHEL, a homeodomain protein that is required for meristem function (Lenhard et al., 2001; Lohmann et al., 2001), positively regulate AG and that AG negatively regulates WUS. WUS expression persists in the indeterminate meristems of ag flowers (Lohmann et al., 2001). We have shown that kn1 is expressed in ovule primordia of ifa1 mutants, suggesting that IFA1 negatively regulates KN1. Given the parallels between ifa1 zag1 and weak ag alleles of Arabidopsis, it is likely that IFA1 also negatively regulates a WUS ortholog.

Different models have addressed the relationships of the branch meristems in maize. The conversion model proposes that the SPM converts to an SM after initiation of the sessile spikelet, and the SM converts to the upper floral meristem after initiating the lower floret (Irish, 1997). With this model, the ifa1-r phenotype could be explained as a delay in the conversion of one meristem type to another. This delay would result in the initiation of more meristems before conversion. The SPM would initiate more than one SM and the SM would initiate more than one floral meristem before converting. Perhaps, if an SM can convert to a floral meristem, then it can also convert back to a SPM, depending on amounts of gene product. The other model proposes that a residual SM is present after production of both floral meristems (Chuck et al., 1998; Irish, 1998). In mutants such as ids1-mum1 or ifa1-r, the residual meristem continues to produce lateral organs. Our SEM analysis of ifa1 ids1 lends support to the residual meristem model, in which a meristem is clearly present between the two floral meristems.

The spikelet is a unifying feature of all grasses, yet the branches upon which the spikelets are born vary considerably, depending on the species (Clifford, 1987). Spikelets may arise directly on the primary inflorescence, or on secondary or tertiary branches, and they may produce one, two or many florets. The Andropogoneae, a subgroup of the grasses that include maize, have two florets in the spikelet and two spikelets in a modified branch called the spikelet pair. The unique structure of the Andropogoneae is considered to be evidence of monophyly of the subfamily (Kellogg and Campbell, 1987). ifa1 and a handful of other genes are responsible for determinacy in the SPM and SM. It is interesting to speculate whether ifa1 was recruited from floral meristems to branch meristems and thus responsible for the determinacy of the SPM and SM. An investigation of the expression and function of branch meristem determinacy genes in different grasses may answer such questions and lead to an understanding of the evolution of grass morphology.

We thank members of the Hake laboratory for assistance in the field, interesting discussion and input on this paper, R. Schmidt for the zag2 cDNA clone, Tracy Yamawaki for mapping ifa1-r relative to zmm14, and E. Vollbrecht for providing an allele of ifa1 from his Mu field. Special thanks to George Chuck, Nick Kaplinsky, Yuki Mizukami, Leonore Reiser and Mark Running for critical reading of the manuscript. We are grateful to David Hantz for taking care of our plants in the greenhouse and for Manuel Eisenberg for help in screening Mu fields for ifa1 alleles. This work was supported by the United States Department of Agriculture, CRIS 5335-21000-018-00D. D. L. C. was supported by an NSF postdoctoral fellowship (Award #9406828).