Conservation and molecular dissection of ROUGH SHEATH2 and ASYMMETRIC LEAVES1 function in leaf development

Abstract

Maize ROUGH SHEATH2 (RS2) and Arabidopsis ASYMMETRIC LEAVES1 (AS1) are orthologous Myb-related genes required for leaf development and act as negative regulators of class 1 KNOTTED1-like homeobox (KNOX) genes in leaf primordia. Expression of RS2 in Arabidopsis fully complements as1 leaf phenotypes and represses the expression of the KNOX gene KNAT1 in leaves. Whereas loss of AS1 function in Arabidopsis results in rounded, lobed leaves with shorter and wider petioles, overexpression of either RS2 or AS1 results in longer and narrower leaves with longer petioles than wild type. A conserved C-terminal domain (CTD) mediates homodimerization of both RS2 and AS1 and modulates leaf shape when expressed independently of the Myb domain in Arabidopsis. Homodimerization is not absolutely required for KNAT1 repression. RS2:GFP fusion protein is biologically active, localized in discrete dynamic subnuclear foci and associates with DNA during cell division.

Plants have evolved distinctive and diverse leaf morphologies to adapt to their environments. However, beneath this variation in design are shared developmental regulatory pathways. Leaves are made continuously from founder cells recruited from meristems (groups of pluripotent cells found at the growing tips of the plants). This transition in cell fate from meristem to leaf requires extensive reprogramming of gene expression.

One pathway involved in reprogramming gene expression during leaf development involves the maize ROUGH SHEATH2 (RS2), Arabidopsis ASYMMETRIC LEAVES1 (AS1) and the Antirrhinum PHANTASTICA (PHAN) genes. RS2, AS1, and PHAN genes are members of a small unique MYB-related gene family that are required for repressing expression of certain KNOX (KNOTTED1-like homeobox) genes in leaves (1–6). By using the rice genome to represent the fellow grass, maize, RS2 is shown to be the most homologous gene to AS1 in Arabidopsis (data not shown).

KNOX genes are expressed in overlapping domains within meristems and turned off in developing leaves (7–14). Various KNOX genes have been implicated in meristem maintenance, the transition from meristem to leaf, and the regulation of inflorescence and leaf and stem architecture (11, 15–18). KNOX genes may function by regulating hormone levels: the tobacco KNOX gene NTH15 and the Arabidopsis KNOX gene STM have been shown to be directly involved in regulating the gibberellin biosynthetic enzyme GA 20-oxidase (19, 20). The regulation of KNOX gene expression seems to be complex and likely occurs at the level of transcription, translation, alternate splicing, and extracellular trafficking (7, 21–24).

The RS2/AS1/PHAN genes are expressed in leaf primordia and condition aberrant leaf developmental phenotypes similar to KNOX overexpression when mutated (1–5). In maize, loss of RS2 function results in plants with twisted leaves, disorganized differentiation of the blade-sheath boundary, and aberrant vascular patterning (2–4). The KNOX genes rs1 and lg3 are misexpressed in leaves of rs2 mutants (2–4). In Arabidopsis, loss of AS1 function results in leaves with a much wider, rounder, and occasionally lobed leafblade, a shortened petiole, and broader midvein (5, 6, 12, 25). The KNOX genes KNAT1 and KNAT2 are misexpressed in leaves of as1 loss of function mutants, whereas AS1 may itself be negatively regulated by the KNOX gene STM (5, 12, 26). Antirrhinum phan mutants and a recently found as1 allele are unique in that the adaxial–abaxial leaf axis is affected (1, 6, 27). The differences in the phenotypes of phan, as1, and rs2 mutants suggest that, although the RS2/AS1/PHAN regulatory pathway is conserved between monocots and dicots, the consequences of this conserved regulation are not the same.

Beyond the inference that RS2/AS1/PHAN regulate the expression of certain genes, nothing is known of the molecular mechanisms by which these genes control leaf development. We are interested in elucidating the role of RS2/AS1/PHAN genes in leaf development and in understanding the molecular mechanisms by which they function.

Materials and Methods

Arabidopsis Genetic Stocks and Growth Conditions. The wild type used in all experiments was Col-1 or glabrous Col-0. The as1-1 mutant line accession number CS3774 (Col-1 background) was obtained from the Arabidopsis biological resource center. The wild-type and as1 mutant KNAT1:GUS reporter lines were a gift from S. Hake (University of California, Berkeley) and have been described (12). The wild-type Col-0 line was a gift from R. Sung (University of California, Berkeley).

Molecular Techniques. Standard methods for DNA manipulation were used. For all of the plasmid constructs, Taq polymerase was used for PCR amplifying RS2, and pfu polymerase was used to amplify AS1.

Yeast Two-Hybrid Analysis and Constructs. The yeast two-hybrid analysis was performed in the following manner. pGBDRS2 and pGADRS2 plasmids were created by amplifying the RS2 coding sequence using PCR primers GCGATGGCCTTTGCAGGAATTCATGAAGGAGCGACAG and GCGGCTCGTTTGGATCCAGTTCAACAGTTC and subsequently cleaved and ligated into the EcoRI and BamHI sites of the two-hybrid plasmids pGBT9 and pGAD424 (CLONTECH) to create inframe translational fusions of RS2 to the GAL4 DNA binding domain and the GAL4 activation domain. The coding sequence of AS1 was similarly amplified and cloned from Arabidopsis genomic DNA by using primers GA AT TCATGA A AGAGAGACAACGTTGGAGTGGTGAA and GGATCCTCAGGGGCGGTCTAATCTGCAACCCATTTG.

The GAL4 binding domain RS2 deletion derivatives were created by cleaving and religating plasmid pGBDRS2 by using different combinations of restriction enzymes. The BD-RS2 conserved C-terminal domain (CTD) fusion was created by using PCR primers GTCTGTGGACGGGGAATTCCAGGCGCTGGCGG and GGAGCAGTCGACCCCCCCCATCTGCTCG to amplify the RS2 CTD and clone into pGBT9 to create an in-frame fusion of the GAL4 DNA binding domain and amino acids 245–357 of RS2. The AD-RS2 conserved CTD fusion plasmid was isolated as an interactor in an RS2 two-hybrid screen (G.T., unpublished data) and contains an in-frame translational fusion of the GAL4 activation domain to amino acids 224–370 of RS2.

For the deletion analysis of the AS1 CTD the following primers were used to amplify different C-terminal deletion fragments, which were cloned in-frame into the EcoRI and SalI site of pGBT9: for amino acids 252–356, GCAGCCCGAATTCTCAGAGCTTGTGGAGTGTTGTAGAGT and CTAGCACCTGTCGACAAGAAACTTGGTGAGTCTGATATGCC; for amino acids 252–341, GCAGCCCGAATTCTCAGAGCTTGTGGAGTGTTGTAGAGT and GTCGACAGGTGCTAGCCAGTTTCTGGTCTTTGGCCTCTGCG; for amino acids 268–356, GCAGCCCGAATTCTGGGCAGACCATAAGAAAGAGGCTGCATGG and CTAGCACCTGTCGACAAGAAACTTGGTGAGTCTGATATGCC. Standard yeast genetic and transformation techniques were used (28, 29). The yeast strain Y116 was used for the two-hybrid analysis (30). The β-galactosidase assays were performed as described (28, 29).

Transgenic Arabidopsis Methods and Constructs. The binary vector pCAM3301 was used to create the 35S AS1 and RS2 overexpression lines, and the vector pCAM1302 was used to create the RS2:GFP fusion (31).

The following primers were used to create the constructs: for the 35S:RS2 construct, PCR primers ATGGTAGATCTGATGAAGGAGCGACAGCGCTGGCGGCCTGAG and GTGGTCACCAGTTCAACAGTTCATGTCGGTGGCAGAG were used to amplify the RS2 coding sequence; for the 35S:AS1 construct, PCR primers CCATGGAAGAGAGACAACGTTGGAGTGGTGAAGAAG and CACGTGTCAGGGGCGGTCTAATCTGCAACCCATTTGTTG were used to amplify the AS1 coding sequence; for the 35S:ΔMYB construct, PCR primers AGATCTGGGCGGCAAGTGGTGGGAGGTGTTCAAGGAGAAG and GTGGTCACCAGTTCAACAGTTCATGTCGGTGGCAGAG were used to amplify the RS2 coding sequence beginning downstream of the Myb domain; for the 35S:ΔCTD construct, PCR primers ATGGTAGATCTGATGAAGGAGCGACAGCGCTGGCGGCCTGAG and GCGCTGCGGTCACCTCACAGCGCCTGCCCGTCCACGACCGCCATGCC were used to amplify the RS2 coding sequence; for the 35S:RS2:GFP construct, PCR primers ATGGTAGATCTGATGAAGGAGCGACAGCGCTGGCGGCCTGAG and cagatctgcacagttcatgtcggtggcagaggaccatg were used to amplify the RS2 coding sequence, which was cloned into pCAM1302.

Constructs were introduced into the Agrobacterium strain GU3101(pMP90) by electroporation and then introduced into Arabidopsis plants by the floral dip method (32). Transformants were selected for Basta resistance by spraying plants with a solution of 1.75% Finale (Farnam, Phoenix, AZ) in water with 0.035% Silwet L-77 (Lehle Seeds, Round Rock, TX) 7–10 days after germination and repeating the spraying after a 1-wk interval. Presence of the transgenes was confirmed by using PCR with primers used to generate the constructs. Transformants were taken to the next generation, and segregation of Basta resistance was tested by applying 5 μl of Basta solution to plant leaves. For the GFP fusion constructs, transgenic plants were selected on 0.5× Murashige and Skoog (MS) media containing 50 mg/liter hygromycin. For each construct, at least eight independent plants were transformed, and multiple transformants were examined from each plant.

Microscopy. The localization pattern of RS2:GFP was analyzed by examining leaf primordia, young leaves, and root tips of 3- to 5-day-old RS2:GFP transgenic seedlings by using a Zeiss LSM510 confocal microscope at the Biological Imaging Facility at University of California, Berkeley. Nuclei were visualized by staining live tissue with 1 mg/ml Hoechst stain in 20% DMSO and 50% glycerol. GFP fluorescence was excited by an argon ion laser at 488 nm, and emissions were detected between 505 to 530 nm. Hoechst fluorescence was excited by an argon ion laser at 353 nm and detected between 480 to 500 nm.

Biochemical Methods. Full-length RS2 was cloned into the pTOPO cloning vector such that the translation start site was downstream of the T7 promoter in the vector. One microgram of this construct was used in a 50-μl reaction with a 2-μl aliquot of [³⁵S]methionine and was subjected to in vitro transcription and translation with the Promega TNT Coupled Reticulocyte Lysate System following the manufacturer's protocol. The [³⁵S]methionine-labeled RS2 was resolved by SDS-10% PAGE with or without pretreatment with 100 mM DTT and 2-mercaptoethanol.

GUS Staining. GUS staining was carried out as described (12).

Results and Discussion

The RS2/AS1 Pathway Is Functionally Conserved Between Monocots and Dicots. Given that mutations in RS2 and AS1 affect leaf morphology and KNOX gene expression in two distantly related plants, maize and Arabidopsis, we wanted to test the functional conservation of this pathway. To determine whether expression of maize RS2 could complement the as1 phenotype and to examine the effects of AS1 and RS2 overexpression on leaf development, we generated transgenic Arabidopsis plants expressing AS1 and RS2 from the 35S promoter of cauliflower mosaic virus. To monitor the effects on the expression of the knox gene KNAT1, the constructs were introduced into lines containing a fusion of the promoter of the KNAT1 to the reporter gene GUS (12). In a wild-type line, GUS expression is limited to the meristem, whereas in an as1 mutant, ectopic GUS expression is found in the petiole and leaf veins (12). Expression of either RS2 or AS1 in an as1 knockout was able to complement the as1 leaf development phenotype (Fig. 1, compare C and E with A and K) and abolished ectopic GUS expression in leaves (Fig. 1, compare D and F with B and L), suggesting that RS2 and AS1 are functionally interchangeable and that this pathway is functionally conserved over 200 million yr of plant evolution.

Fig. 1.

Phenotype (A, C, E, G, I, K, N, Q, T, and W) and KNAT1:GUS (B, D, F, H, J, L, N, P, R, and T) expression in as1 mutant (A–J) and wild type (K–T) plants containing RS2 and AS1 transgenes. (A and B) as1 mutant; (C and D) as1 mutant + 35S:AS1;(E and F) as1 mutant + 35S:RS2;(G and H) as1 mutant + RS2ΔMYB;(I and J) as1 mutant + RS2ΔCTD; (K and L) wild type; (M and N) wild type + 35S:AS1; (O and P) wild type + 35S:RS2; (Q and R) wild type + RS2ΔMYB; (S and T) wild type + RS2ΔCTD.

35S:RS2 and 35S:AS1 transformants were recovered at approximately the same frequency as vector alone, suggesting that increased expression of either is not lethal. A range of phenotypes was observed among the transformants from partially to fully complemented lines.

In both wild-type and mutant plants, expression of RS2 and AS1 from the 35S promoter, which should both increase AS1/RS2 expression in leaves and result in ectopic expression in meristems, resulted in dominant leaf phenotypes. Whereas loss of AS1 resulted in plants with shorter petioles and wider leaves (increased width/length ratio), 35S expression of either AS1 or RS2 resulted in the opposite effect: a greater percentage of the leaf length was composed of petiole and the leaves were narrower (width/length ratio was decreased; Fig. 2D). This phenotype was most apparent in young plants (Fig. 2, compare B and C with A) but was also observed in mature plants (Fig. 1, compare C, E, M, and O with K).

Fig. 2.

Phenotypic consequences of overexpression of AS1 and RS2. Juvenile rosette leaves of wild-type plants transformed with vector alone (A), 35S:AS1 (B), and 35S:RS2 (C). (D) Graph of leaf dimensions of wild-type, as1 mutant, and 35S:AS1 plants. The left bars represent the percentage of leaf length comprised of petiole and the right bars represent the leaf width/length ratio.

The C-Terminal Conserved Domain of RS2 and AS1 Mediates Homodimerization. To date, the nature of how RS2/AS1/PHAN function on the molecular level has not been ascertained. The proteins encoded by these genes contain two extensive domains of high homology. The Myb domain, presumed to be involved in nucleic acid binding, is at the N terminus. The Myb domain is highly diverged compared with that in other Myb proteins, and the DNA recognition helix in Myb repeat R3 is completely unique. Despite considerable effort, we have been unable to demonstrate that RS2 binds DNA in vitro (data not shown) and believe that cofactors are required for nucleic acid interaction.

In addition to the Myb domain, RS2/AS1/PHAN proteins also contain a conserved domain of unknown function at the C terminus. Using a yeast two-hybrid assay, we found that one of the functions of this CTD is to mediate homodimerization of RS2 and AS1 (Fig. 3). RS2 and AS1 also form RS2/AS1 dimers (Fig. 3). Furthermore the majority of RS2 interactors identified in a two-hybrid screen encoded RS2 and RS2 deletion derivatives (data not shown). Deletion analysis delineated the minimal dimerization domain to amino acids 245–347 in RS2 and to amino acids 252–356 in AS1. This domain is highly conserved within the RS2/AS1/PHAN family but does not display homology to any other known protein sequence. This domain is predicted to form an alpha helix, is rich in highly charged amino acids, and contains two conserved short potential leucine zipper-like motifs. Deleting 17 aa from the N terminus or 16 aa from the C terminus of the minimal dimerization region of AS1 results in loss of interaction.

Fig. 3.

(A) Two-hybrid deletion analysis of RS2 and AS1 homodimerization. Interaction is observed only between bait and prey plasmids containing the conserved full-length RS2 or AS1 or the conserved CTD. β-Galactosidase activity is expressed as nmol of o-nitrophenyl-d-galactoside (ONPG) hydrolyzed min^-1·mg^-1 total protein. (B) SDS/PAGE of ³⁵S-labeled RS2 protein under reducing and nonreducing conditions. Lane 1, RS2 after reduction with 100 mM DTT; lane 2, RS2 without treatment with a reducing agent.

Homodimers of RS2 were also observed biochemically. [³⁵S]Methionine-labeled RS2 synthesized by in vitro transcription and translation was subjected to SDS/PAGE under reducing or nonreducing conditions. Under nonreducing conditions, two bands of approximately equal intensity are seen, one corresponding to the monomer (41,900 Da) and a band that runs at approximately the predicted size of an RS2 homodimer (83,800 Da; Fig. 3B, lane 2). After treatment with DTT or SDS/heating, RS2 molecules were fully resolved to monomeric size (Fig. 3B, lane 1).

Deletion Analysis of RS2 Function. To determine the significance of the homodimerization and Myb domains to RS2 function, we created constructs expressing the RS2 coding sequence with the Myb domain deleted (35S:RS2ΔMYB) or with the carboxyl terminal homodimerization domain deleted (35S:RS2ΔCTD). The rationale for these experiments was that if homodimerization were essential for function then RS2ΔCTD-encoded protein should be nonfunctional and would not be expected to complement the as1 mutant. Moreover, because RS2 homodimerized with itself and formed dimers with AS1, expression of RS2ΔMYB in wild-type Arabidopsis plants might result in the incorporation of wild-type AS1 protein into nonactive dimers with the Myb deletion derivative and would result in a dominant negative as1 null phenotype. It is important to note that putative nuclear localization signals are found in both the myb domain and the homodimerization domains (see next section), suggesting that both the RS2ΔMYB and RS2ΔCTD protein products would be predicted to reside in the nucleus.

The 35S:RS2ΔCTD construct had no discernable effect on leaf morphology or KNAT1 expression when introduced into wild-type lines (Fig. 1 S and T). In as1 mutant lines, RS2ΔCTD decreased ectopic expression of KNAT1 but not to wild-type levels (Fig. 1, compare J with B and L) and partially suppressed the as1 leaf development phenotype (Fig. 1, compare I with A and K; Fig. 4, compare B with A and D). These results suggest that the CTD, and by extension homodimerization, are not absolutely required for KNAT1 repression. Supporting this idea, Waites and coworkers (1) identified an allele of phantastica that contained an insertion into the caroxyl-terminal domain of PHAN. This allele, phan-249G, had the potential to encode a novel protein in which the 40 C-terminal amino acids were replaced by 34 aa encoded by the TAM2 transposon (1). The phan-249 allele conditions an intermediate phenotype, suggesting that the protein may be partially functional (1). Based on our two-hybrid analysis of RS2 and AS1, the phan-249G-encoded protein would not be expected to homodimerize. This protein, like RS2ΔCTD, may be weakly functional.

Fig. 4.

Suppression of the as1 leaf phenotype by overexpression of RS2ΔCTD and RS2ΔMYB. Representative leaves from as1 mutant plants transformed with vector alone (A), 35S:RS2ΔCTD (B), and 35S:RS2ΔMYB (C) and a wild-type plant (D) are shown.

When introduced into wild-type Arabidopsis lines, the 35S:RS2ΔMYB construct did not alter KNAT1 expression (Fig. 1, compare R with L) and did not confer an as1 leaf phenotype (Fig. 1, compare Q with A; Fig. 5, compare F with G). Instead, the construct conferred unexpected dominant leaf phenotypes (Fig. 1, compare Q with K; Fig. 5, compare D with C and F with E). One of the RS2ΔMYB overexpression phenotypes, having longer and narrower leaf dimensions (Fig. 5, compare B with A and F with E, is similar to the phenotype of AS1 and RS2 overexpression (Fig. 2, compare B and C with A). The leaf serrations of RS2ΔMYB plants are larger and more lobe-like than those of wild-type plants and are reminiscent of, although not as conspicuous as, lobes seen in as1 mutant leaves (Fig. 5, compare F with G). RS2ΔMYB-overexpressing plants also have a novel leaf phenotype: the leaf shape is altered such that the transition from petiole to leaf blade is more gradual and the petiole/leaf blade boundary is more diffuse (Fig. 5, compare F with E).

Fig. 5.

Phenotypic consequences of overexpression of RS2ΔMYB on leaf development. Shown are the first true leaves of a wild-type sib (A) and two 35S:RS2ΔMYB (B) plants. (C and D) Rosette phenotypes of older mature plants: wild type transformed with vector alone (C) and 35S:RS2ΔMYB (D). Adult leaf morphology of wild type (E), wild type + 35S:RS2ΔMYB (F), as1 (G), and as1 + 35S:RS2ΔMYB (H) is shown.

Expression of 35S:RS2ΔMYB in the as1 mutant line did not alter the ectopic expression of KNAT1 (Fig. 1, compare H with B) but partially suppressed the as1 leaf development phenotype (Fig. 1, compare G with A; Fig. 4, compare C with A and D). The leaf shape of suppressed plants was closer to the spatulate shape of wild-type leaves than to the rounded shape of as1 (Fig. 1, compare G with A; Fig. 4, compare C with A and D).

Perhaps the simplest model to explain the phenotype of RS2ΔMYB is that it has activity unrelated to the normal function of the gene, perhaps due to altered protein stability or localization, and that this activity confers the dominant phenotype in wild type and masks the as1 mutant phenotype. However, RS2ΔMYB overexpression affects morphological traits, namely leaf elongation, petiole/leafblade development, and leaf serration, which are also effected by AS1/RS2 overexpression and loss of as1. The fact that those traits are altered in novel ways lends support to more complex models in which the RS2ΔMYB protein interacts with and possibly titrates factors that normally interact with AS1 to modulate leaf morphology. In wild type, the leaf phenotype may involve dimerization between wild-type AS1 and the RS2ΔMYB deletion or may result from an additive effect of overexpression of RS2ΔMYB in the presence of AS1. These data suggest a model for RS2 function in which the CTD functions to regulate interactions between the Myb domain and other factors involved in modulating leaf shape.

Nuclear Localization of RS2. Although RS2/AS1 have been shown to be expressed in leaf primordia, their subcellular localization has not been examined. PSORT, a computer algorithm used to predict protein localization sites, predicted nuclear localization for RS2 and found three bipartite nuclear localization sites in the RS2 protein sequence starting at amino acids 82 (KKIAAEVPGRTAKRLGK), 261 (RRAWAAHRREAAWRLKR), and 276 (KRVEQQLEMEREMRRRE) (33). To conclusively determine the subcellular localization of RS2, we generated a fusion construct in which the RS2 coding sequence was fused to GFP and expressed from the 35S promoter. When expressed in an as1 mutant line, the RS2:GFP fusion was able to fully complement the as1 phenotype, indicating that the fusion protein is biologically active (data not shown). Using confocal laser scanning microscopy, we examined the subcellular localization of the RS2:GFP fusion protein. As expected, a GFP-only control was localized to both the cytoplasm and the nucleus (data not shown). The RS2:GFP fusion was exclusively localized in nuclei, confirming that RS2 is a nuclear protein. In addition, fusion protein was localized in discrete subnuclear bodies (Fig. 6 A and B). These bodies were observed throughout the nucleoplasm (Fig. 6B) and often around the periphery of the nucleolus (Fig. 6A; the nucleolus is marked by an arrow). The number of bodies varied from nucleus to nucleus, and in some nuclei the bodies were absent altogether. The bodies were dynamic, able to move around the nucleus, dissipate, and coalesce within 2 min. To observe the behavior of the bodies during cell division, we examined root tips of seedlings 1–3 days after germination. In dividing cells, the bodies dissociated, but the fusion protein seemed to be associated with chromosomes as can be seen in Fig. 6 C and D (a magnified view of C).

Fig. 6.

Subcellular localization of RS2:GFP. Nuclei are visualized with Hoechst stain. (A and B) Localization of RS2:GFP fusion protein in young leaf cells. The subnuclear bodies are present in the nucleoplasm (B) and around the periphery of the nucleolus, which is marked by an arrow in A.(C) Confocal image of a root tip showing a dividing cell marked by arrows. (D) A close-up showing localization of fusion protein with dividing chromosomes.

However, it should be noted here that RS2:GFP is overexpressed under control of the 35S promoter. Hence, there remains the possibility that overaccumulation of the fusion protein at nonphysiological concentration could contribute to the observed subnuclear localization. If the bodies are in fact artifactual protein crystals, one might expect the crystals to remain intact during cytokinesis, whereas authentic nuclear structures might be dissembled and reassembled. We found that the bodies are not present in dividing cells. Moreover, the protein seems to be associated with chromosomes during cell division, suggesting an association with DNA (Fig. 6 B and C).

The finding that RS2:GFP is localized in subnuclear bodies is intriguing because a growing body of evidence suggests that proteins involved in many nuclear processes, including DNA replication, chromatin remodeling, transcriptional regulation, and RNA processing, accumulate in discreet domains or foci (34–37).

Conclusion

In conclusion, we found that expression of the maize RS2 gene is able to fully complement the Arabidopsis as1 mutation, indicating a high degree of functional conservation between AS1 and RS2 consistent with their sequence homology. This conservation suggests that the differences between leaf shape and vasculature between Arabidopsis and maize have arisen by the evolution of downstream pathways over the past 200 million yr. Our results are an example of complementation of upstream developmental regulatory genes between monocots and dicots. Moreover, these data suggest that some developmental pathways in plants are functionally interchangeable between mononcots and dicots and are consistent with animal studies that suggest that fundamental developmental networks are functionally conserved between animals as distantly related as arthropods and vertebrates (38). The results of our molecular analysis, including subnuclear localization, homodimerization, and deletion analysis, provide insight into how this evolutionarily conserved developmental pathway functions on the molecular level.