Mutations in the LUMINIDEPENDENS ( LD) gene of Arabidopsis thaliana (L.) Heynh. (Arabidopsis) confer a late-flowering phenotype, indicating that LD normally functions to promote the floral transition. RNA and protein blot analyses, along with the analysis of transgenic plants containing a fusion between a genomic fragment of LD and the reporter gene uidA ( GUS), indicate that LD is expressed primarily in apical proliferative regions of the shoot and root, including the shoot apical meristem and leaf primordia. Subcellular localization studies indicate that LD is a nuclear protein, consistent with its previously proposed transcriptional regulatory role. We have also found that in an apetala1 cauliflower ( ap1 cal) background the ld mutation converts the reproductive shoot apex to a more vegetative state, a phenotype that is similar to that seen for the leafy ( lfy) mutant. Furthermore, in situ hybridization analysis indicates that LFY levels are drastically reduced at the apex of ld ap1 cal plants after bolting. These data are consistent with the idea that at least one function of LD is to participate in the regulation of LFY.
The transition of the shoot apex from a vegetative to a reproductive mode of growth is a critical developmental switch in the life cycle of a plant. Prior to this floral transition, the shoot apical meristem (SAM) primarily forms leaves, whereas afterwards the SAM produces floral primordia that differentiate into flowers. In order to ensure that flowering occurs at a proper time and thus maximize their reproductive success, many species have evolved mechanisms to regulate the timing of the floral transition in response to developmental cues and certain environmental stimuli ( Lang 1965; Napp-Zinn 1987; Poethig, 1990). The complexity of these floral timing mechanisms has become increasingly clear from genetic studies which reveal that many genes are involved in the regulation of flowering time ( Koornneef et al. 1998 ; Weller et al. 1997 ).
Arabidopsis thaliana (L.) Heynh. (Arabidopsis) is a facultative long-day plant which responds to long days by flowering earlier than when grown in short days ( Koornneef et al. 1998 ). More than 20 genes that control flowering time in Arabidopsis have been identified through the analysis of both late and early flowering mutants. The flowering-time mutants have been grouped into several different classes based upon the response of each mutant to changes in photoperiod and temperature ( Koornneef et al. 1998 ). One class of Arabidopsis flowering-time mutants displays a reduced response to changes in photoperiod when compared to wild type and it has been proposed, therefore, that the corresponding genes participate in a photoperiod-regulated pathway. A second class of mutants, while displaying an altered flowering time, are nonetheless unaffected in their response to changes in photoperiod; the genes corresponding to these mutants have thus been placed in an autonomous pathway. A third class, and perhaps a third pathway, is represented by a single mutant, ga1, which is deficient in gibberellin and is more responsive to changes in photoperiod than wild-type plants ( Wilson et al. 1992 ).
The protein product of the CO gene ( Putterill et al. 1995 ) displays similarity to the zinc finger class of transcriptional activators and, therefore, promotion of flowering by CO is probably carried out by transcriptional activation of one or more downstream target gene(s). One possible candidate for this is the floral meristem-identity gene LEAFY (LFY), which is expressed in primordia arising on the flanks of the shoot apical meristem (SAM) ( Weigel et al. 1992 ). Meristem-identity genes such as LFY and APETALA1 (AP1) and CAULIFLOWER (CAL) function to switch these primordia from a vegetative to a floral state ( Bowman et al. 1993 ; Weigel et al. 1992 ). LFY is upregulated as the plant approaches the floral transition, and it is thought that the level of LFY reaches some critical threshold at which point the vegetative-to-floral transition occurs ( Blazquez et al. 1997 ). Evidence for regulation of LFY by CO was provided by Simon et al. (1996) who showed that LFY transcription is rapidly initiated in response to CO expression. It appears, then, that promotion of flowering by CO in wild-type plants is accompanied by upregulation of LFY at the shoot apex.
The mode of regulation employed by the gene products in the autonomous flowering pathway is less clear. The FCA gene product ( Macknight et al. 1997 ) was found to be homologous to a class of RNA-binding proteins, suggesting that FCA may promote flowering via a post-transcriptional mechanism. The LD gene product ( Lee et al. 1994 ) shows no strong similarity to other proteins, but does have two consensus bipartite nuclear-localization domains ( Dingwall & Laskey 1991), implying that it is a nuclear protein. LD also contains a glutamine-rich region at the carboxy terminus that resembles the glutamine-rich domains found in several transcription factors ( Mitchell & Tjian 1989). These two features, along with a putative divergent homeodomain ( Aukerman & Amasino 1996), suggest that LD may be a transcriptional regulatory protein.
In this report, we have analyzed the spatial expression of LD by RNA and protein blots, and by transgenic plant analysis. We have found that LD is expressed primarily in regions of the shoot and root apex that contain dividing cells, including the apical meristems. In addition, we demonstrate that the LD gene product is targeted to the nucleus, consistent with its proposed transcriptional regulatory function. Finally, we have found that ld ap1 cal triple mutants display a drastic reduction of LFY expression, leading to the complete inability of these plants to form floral structures. These results suggest that one function of LD is to participate in the regulation of the floral meristem-identity gene LFY. Our studies also provide support for the notion that both the autonomous and photoperiod-regulated flowering pathways are involved in the regulation of LFY.
Analysis of LD transcript and protein accumulation
We initially determined the spatial distribution of the LD transcript by extracting RNA from various organs of wild-type Arabidopsis and performing RNA blot analyses utilizing an LD-specific probe. Figure 1 shows that LD is expressed in both seedlings and mature plants. The LD transcript is most abundant in shoot apices, inflorescence stems, floral buds and roots, and less abundant in cotyledons and leaves.
To determine whether the LD protein exhibits a similar pattern of expression, we utilized a transgenic Arabidopsis line containing a genomic copy of LD with six copies of a c-myc epitope inserted within the coding region of LD ( Fig. 2a, see Experimental procedures for details). This LD-myc construct rescues the late-flowering phenotype when transformed into the ld-2 mutant background (data not shown), which demonstrates that the LD-myc protein is fully functional, and further suggests that the LD promoter in the LD-myc construct is driving the expression of the LD gene product in a pattern identical to that seen for the endogenous LD promoter. Due to the relatively low levels of LD protein present in wildtype Arabidopsis (M. Aukerman, unpublished observations), it was necessary to perform a two-step, immunoprecipitation/immunoblot analysis of the LD protein. Proteins extracted from various organs of transgenic plants containing the LD-myc construct were immunoprecipitated with an LD-specific antibody, and the precipitate was resuspended and analyzed by standard immunoblotting using an antibody specific for the myc epitope. This analysis ( Fig. 2b) indicates that, as seen for the LD transcript, the LD protein accumulates to the highest levels in vegetative shoot apices, roots and floral buds, and is present at much lower levels in leaves.
Analysis of LD-GUS expression
To confirm and extend our findings on the spatial expression of LD, we constructed a fusion between a 5 kb genomic fragment of LD and the E. coli gene uidA which encodes β-glucuronidase (GUS) ( Jefferson 1987). As shown in Fig. 2(a), the LD-GUS fusion construct consists of 2 kb of the LD promoter region and 3 kb of the LD coding region fused in-frame to the GUS gene. This construct generates a protein consisting of the N-terminal 451 amino acids (out of a total of 953) of LD fused to the GUS protein. Staining of transgenic plants containing this construct with X-gluc revealed an accumulation of GUS enzyme in both root and shoot apices ( Fig. 3a), indicating a preferential expression of the LD gene in those regions. LD is expressed strongly throughout the SAM ( Fig. 3d), and young leaf primordia also stain strongly ( Fig. 3b–d). Consistent with our RNA and protein analyses, cotyledons and fully expanded leaves display very little GUS staining ( Fig. 3a,c). Thus, rapidly proliferating tissues, including the shoot and root meristems and young leaf primordia, express LD at the highest levels.
LD expression remains high after flowering has occurred, as indicated by staining inflorescence sections of transgenic plants ( Fig. 3e). LD is expressed throughout the reproductive SAM and in all whorls of younger flower primordia, for example, the stage 3 flower in Fig. 3(e) (see Bowman 1994 for discussion of the flower stages). As individual flowers develop, the LD expression pattern becomes more restricted, so that at more mature stages only the more recently developed inner whorls corresponding to carpel and stamen primordia express the LD gene product. A final restriction of LD expression to the inner whorl occurs such that prior to the opening of the flower bud, expression of LD appears to be confined primarily to the developing ovules ( Fig. 3f). Therefore, LD expression in the reproductive stage is similar to its expression in the vegetative stage, in that rapidly proliferating and less mature tissues display the highest levels of LD expression. We have analyzed three independent transgenic lines containing the LD-GUS construct, and all three display the expression patterns detailed above (I. Lee and M. Aukerman, unpublished observations).
Nuclear localization of LD-GUS protein
Closer inspection of the root tips of stained transgenic plants expressing LD-GUS reveals a punctate pattern of X-gluc accumulation ( Fig. 4, far left), suggesting that the LD-GUS fusion protein resides in the nucleus. Under longer incubation times, the root hairs of LD-GUS plants also stain, and the two LD-GUS panels of Fig. 4 show a close-up of a root hair cell from an LD-GUS plant stained with X-gluc. The staining pattern demonstrates that the LD-GUS fusion protein accumulates exclusively in the nucleus, as indicated by counterstaining the sample with 4′,6-diamidino-2-phenylindole (DAPI). The wild-type GUS protein normally accumulates in the cytoplasm, as indicated by the X-gluc and DAPI staining patterns of a root hair of a transgenic plant expressing GUS alone ( Fig. 4, GUS panels). Because the LD-GUS fusion construct contains the 5′ half of the LD coding region fused to GUS, it appears that amino acid sequences within the N-terminal half of LD are sufficient to target the LD-GUS protein to the nucleus. When the LD-GUS fusion protein is expressed in tobacco, X-gluc staining is also confined to the nucleus ( Fig. 4, far right), indicating that the nuclear localization signals present in the LD protein can function in a heterologous species. We have also fused the entire LD coding region to GUS and shown that this full-length LD-GUS protein is also localized to the nucleus in transgenic tobacco (M. Aukerman, unpublished results).
Phenotype of the ld mutant in an ap1 cal background
The expression pattern of LD in the shoot apex overlaps with that of the LEAFY (LFY) gene, which is expressed in both vegetative and reproductive organ primordia arising on the flanks of the SAM ( Blazquez et al. 1997 ). The floral initiation process is regulated by the LFY gene product in combination with the products of other floral meristem-identity genes such as APETALA1 (AP1) and CAULIFLOWER (CAL). Double and triple mutant combinations of lfy with ap1 and with ap1 cal produce a more severe phenotype than that seen in the single mutants ( Bowman et al. 1993 ), indicating that LFY and AP1 have distinct but overlapping functions. The similarities in the expression patterns of LD and LFY suggest that ld might, like lfy, display genetic interactions with ap1 and cal. We tested this idea by crossing the ld mutant to the ap1 cal double mutant and comparing the phenotype of the resultant ld ap1 cal plants to that of ap1 cal plants. As reported by Bowman et al. (1993) , the reproductive shoot apex of an ap1 cal mutant plant grown in long days consists of a proliferation of undifferentiated inflorescence meristems, with occasional leaf-like structures emerging ( Fig. 5a). Eventually, flowers with a typical ap1 phenotype will form in lateral positions ( Bowman et al. 1993 ). In contrast, the reproductive shoot apex of the ld ap1 cal triple mutant grown in long days is almost entirely converted to leaf-like structures ( Fig. 5b); furthermore, flowers or flower-like organs never formed on ld ap1 cal plants, even after several months of growth. In ld ap1 cal plants, lateral inflorescence branches develop in the positions on the main stem where flowers would normally develop. The leaves which proliferate at the shoot apex ( Fig. 5b) are associated with these lateral branches, and thus they become separated from the shoot apex as elongation of the main stem occurs (M. Aukerman, unpublished observations).
To establish whether the ld ap1 cal phenotype is specifically caused by the absence of LD function or by a more general delay of flowering, we grew ap1 cal plants in short-day conditions (8 h light, 16 h dark), causing them to bolt at approximately the same time as ld ap1 cal mutants grown in long days. As seen in Fig. 5(c), the shoot apex of an ap1 cal plant grown in short days consists mostly of the proliferating inflorescence meristem also seen in ap1 cal grown in long days ( Fig. 5a), but with a slight increase in the number of leaf-like structures emerging from it. This increase in vegetative character in short-day grown ap1 cal plants is very mild in comparison to the almost complete conversion to vegetative growth seen in long-day grown ld ap1 cal plants ( Fig. 5b). Furthermore, unlike long-day grown ld ap1 cal plants, short-day grown ap1 cal plants will eventually form flowers (M. Aukerman, unpublished observations). This experiment indicates that the phenotype conferred to ap1 cal plants by the ld mutation cannot be mimicked by a general delay of flowering, and therefore the phenotype is more likely due to the loss of a specific function associated with the LD gene product.
LFY expression in the ld ap1 cal mutant
It has been shown that LFY expression in the inflorescence of the ap1 cal mutant is partially reduced but not absent, and remains at wild-type levels in those primordia that are likely to develop into flowers ( Bowman et al. 1993 ). The fact that no flowers or flower-like structures were ever observed in ld ap1 cal plants suggests that LFY levels in these plants might be even lower than those seen in ap1 cal alone. We tested this by examining LFY expression by in situ hybridization of inflorescences of both ld mutant plants and ld ap1 cal mutant plants. As shown in Fig. 6(a), LFY expression in ld mutants within the reproductive apex is normal, reaching high levels in stage 1 and 2 flower primordia. In contrast, the ld ap1 cal triple mutant displayed no detectable expression of LFY in the inflorescence ( Fig. 6b). Because LFY levels are only partially reduced in ap1 cal alone ( Bowman et al. 1993 ), this result indicates that the ld mutation causes a further reduction of LFY expression in this background.
Using several independent methods, we have described the spatial expression pattern of the flowering-time gene LUMINIDEPENDENS. RNA and protein blot analyses indicated that the LD gene product accumulates to higher levels in shoot apices, roots and floral buds than in mature leaves. Further work utilizing an LD-GUS transgenic line allowed us to observe high levels of LD expression within the SAM, the root apex, young leaf primordia, and the inflorescence. In general, LD expression is highest in younger tissues where cells are still rapidly dividing. As these tissues mature and differentiate, LD expression declines. Interestingly, the pattern of LD expression in the shoot apex closely mirrors that of the photoperiod response pathway gene CO ( Simon et al. 1996 ), which is consistent with the idea that the autonomous and photoperiod responsive flowering pathways converge at the shoot apex. Although the expression patterns of LD and CO overlap, to date there is no evidence that the LD and CO gene products directly interact.
The late-flowering phenotype of ld ( Lee et al. 1994 ; Redei 1962) indicates that the LD gene plays an important role in the floral transition, and our studies on the LD expression pattern have provided some clues as to what that role might be. Classical physiological studies have indicated that diffusible signals, thus far uncharacterized, travel from the leaves to the shoot apex to stimulate the flowering response ( Zeevaart 1984). Given that LD is expressed at high levels in the SAM, one possibility is that LD functions in the SAM to somehow promote competency to flower in response to a stimulus from the leaves. LD is also expressed in leaf primordia and, in light of recent studies in maize indicating that leaf primordia are sufficient to provide the flowering signal to the apex ( Colasanti et al. 1998 ; Irish & Jegla 1997), it is also possible that LD plays a role in generating or transmitting the flowering signal. There does not appear to be any role for LD in leaf development itself, however, since ld mutants do not display any obvious leaf abnormalities. Likewise, although we observe expression of LD in the root apex and in developing ovules, the LD gene product does not appear to play a prominent role in the development of those organs because ld mutations do not produce any obvious root phenotype or adversely affect fertility (S. Sanda and M. Aukerman, unpublished observations). Nonetheless, we cannot conclude from the experiments described herein that the LD gene product in roots plays no role in flowering. Although it seems unlikely that LD produced in the root could affect the vegetative-to-reproductive conversion in the SAM, it has been suggested that certain flowering signals emanate from the roots ( McDaniel et al. 1996 ). We are currently attempting both grafting experiments and transgenic experiments to determine whether the LD gene product expressed in roots can affect flowering at the shoot apex.
The LFY gene is expressed in a subset of the tissues in which LD is expressed and, given the central role that LFY plays in flower initiation, it seemed likely that LD might participate in its regulation. This indeed appears to be the case, as the apex of the ld ap1 cal triple mutant proliferates leaf-like structures ( Fig. 5b) that are similar to those seen for lfy ap1 cal ( Weigel et al. 1992 ). This phenotype is probably due to a specific loss of LD function rather than to a general late-flowering effect because ap1 cal plants grown in non-inductive short days do not display this phenotype ( Fig. 5c). Furthermore, LFY levels are drastically reduced in ld ap1 cal plants ( Fig. 6b) and the reduction appears more severe than that reported for ap1 cal mutants ( Bowman et al. 1993 ). This last observation is consistent with the proposed role of LFY in flower initiation because ld ap1 cal plants never form any floral structures, whereas ap1 cal plants do. Nevertheless, the loss of LFY activity alone cannot account for the severity of the phenotype seen in ld ap1 cal because lfy ap1 cal plants make a limited number of floral structures ( Bowman et al. 1993 ; Weigel et al. 1992 ). This suggests that LD regulates other genes in the floral meristem pathway in addition to LFY, for example APETALA2 or UNUSUAL FLORAL ORGANS ( Bowman et al. 1993 ; Lee et al. 1997 ). It is important to emphasize, however, that because ld single mutant plants do not display a meristem-identity phenotype, regulation of the meristem-identity pathway by LD appears to be masked by the more prominent meristem-identity functions of genes such as AP1 and CAL.
Other investigators who have constructed mutant combinations between flowering-time mutants and meristem-identity mutants have reported a similar enhancement of meristem-identity mutant phenotypes by flowering-time mutations ( Putterill et al. 1995 ; Ruiz-Garcia et al. 1997 ). From these studies, it is becoming clear that different flowering-time genes can have effects on different genes within the meristem-identity pathway. CO, for example, has been demonstrated to control LFY expression ( Simon et al. 1996 ), whereas activation of AP1 appears to require an additional pathway. This additional pathway is likely to contain the flowering-time genes FT and FWA, since these seem to primarily regulate AP1 and CAL instead of LFY (Ruiz-Garcia et al. 1997). Furthermore, recent studies by Nilsson et al. (1998) support the idea that CO operates upstream of LFY, whereas FT and FWA primarily affect the response to LFY activity. The observation that LFY expression is normal in ld single mutants could be interpreted as evidence that LD acts in parallel with LFY or downstream of LFY to affect floral meristem identity, in a manner similar to FT and FWA. However, the loss of LFY expression in ld ap1 cal plants suggests instead that LD regulates LFY, and thus is more similar to CO in terms of its regulatory properties. Interestingly, CO and LD belong to different environmental response classes (photoperiod response and autonomous, respectively), and yet they share a common regulatory target, LFY. This is consistent with the idea that LFY expression is not only regulated by photoperiod signals, as has been demonstrated previously ( Blazquez et al. 1997 ), but also by autonomous signals, including gibberellins ( Blazquez et al. 1998 ).
Although our work suggests that one function of the LD gene product is to regulate LFY expression at the shoot apex, it is still not clear how many transduction steps lie between LD and LFY, or what biochemical mode of action LD employs in regulating downstream target genes. One approach to obtain clues about the biochemical function of a protein is to determine its subcellular localization and, in this report, we have demonstrated that an LD-GUS fusion protein is localized to the nucleus thus indicating that LD itself is a nuclear protein. The N-terminal half of LD ( Fig. 2a, LD-GUS), which contains one bipartite nuclear localization consensus motif, is sufficient for nuclear localization. There does not appear to be any tissue-specificity to the nuclear localization or any regulation by photoperiod because the LD-GUS fusion protein accumulates in the nuclei of both the shoot apex and the root apex and under various photoperiodic treatments (M. Aukerman and I. Lee, unpublished observations).
The observation that LD is nuclear localized has spurred us to investigate the biochemical nature of its role within the nucleus. Two clues regarding LD biochemical function have come from inspection of the primary sequence ( Lee et al. 1994 ). First, there is a glutamine-rich region towards the C-terminus of LD which, by analogy to other proteins that contain such a region, could function in transcriptional regulation. Second, a region near the N-terminus has limited similarity to homeodomains which suggests that LD may be a DNA-binding protein ( Aukerman & Amasino 1996). Although we have observed non-sequence-specific binding by LD to calf thymus DNA in vitro, we have been unable to detect sequence-specific DNA binding by LD (M. Aukerman and Y. Noh, unpublished observations). It is possible that the LD homeodomain-like region serves as an RNA-binding motif, similar to the homeodomain in the Drosophila protein Bicoid ( Dubnau & Struhl 1996). This possibility is especially intriguing since the FCA gene encodes an RNA-binding protein and, like LD, is a member of the autonomous flowering pathway ( Macknight et al. 1997 ). It may also be that LD requires other proteins in addition to itself to form a specific complex with DNA or RNA. In this regard, it is tantalizing to speculate that LD directly interacts with the protein product(s) specified by one or more of the other flowering-time genes in the autonomous pathway. Such putative interactions can be more easily investigated as more flowering-time genes are cloned and their protein products become amenable to study.
Plant strains, genetics and growth conditions
The Arabidopsis ecotype Wassilewskija (WS) was used for RNA analyses. Plant transformations were performed either with the ld-2 mutant (in the WS background) or with the RLD ecotype, as described below. To generate the ld ap1 cal triple mutant, the ld-3 mutant was crossed to the ap1-1 cal-1 mutant ( Bowman et al. 1993 ), and late-flowering F2 plants that displayed the enhanced floral meristem phenotype were selected. Growth conditions in all experiments consisted of continuous light from a mixture of fluorescent and incandescent bulbs, unless otherwise noted.
RNA and protein analyses
For RNA blot analysis, plant tissue was ground in liquid nitrogen and RNA was extracted using an RNeasy Plant RNA kit (Qiagen). Twenty μg of each RNA was electrophoresed and blotted, and the blot was hybridized as described previously ( Aukerman et al. 1991 ), using a 0.9 kb Pst fragment of the LD cDNA as a probe. Staining of the gel before transfer indicated that equal amounts of RNA were loaded in each lane (data not shown). Immunoblot analysis utilized an LD-myc transgenic line that was generated as follows. Specific oligonucleotides were used in a PCR reaction to amplify the insert from plasmid pMT6, which contains six tandem copies of a DNA sequence encoding a c-myc epitope ( Roth et al. 1991 ). The six myc copies were inserted into a BamHI site that was introduced by site-directed mutagenesis at codon 942 in a genomic DNA fragment containing the LD gene. The LD-myc genomic fragment was inserted into the plant transformation vector pCGN1547 ( McBride & Summerfelt 1990) and this construct was introduced into ld-2 mutant plants by standard Agrobacterium-mediated transformation as described by Lee et al. (1994) . Transgenic plants containing this construct were analyzed by a combination of immunoprecipitation and immunoblotting as follows. Transgenic LD-myc plant tissue was ground in liquid nitrogen, transferred to a centrifuge tube and proteins were immunoprecipitated with an LD-specific antibody essentially as described by Vierstra & Quail (1982), except that protein A-Sepharose (Pharmacia) was used for the immunoprecipitation step. The polyclonal LD-specific antibody used for immunoprecipitation was raised against an LD fusion protein expressed in E. coli. The immunoprecipitate was loaded onto a 6% SDS/PAGE gel, electrophoresed and transferred to Hybond ECL membrane (Amersham), following standard protocols ( Harlow & Lane 1988). Subsequent immunodetection followed the ECL protocol (Amersham) and utilized the 9E10 monoclonal antibody to c-myc (a gift from M. Sussman, University of Wisconsin) and a goat anti-mouse secondary antibody conjugated to horseradish peroxidase (Amersham).
LD-GUS construction, transformation and staining
To generate the LD-GUS fusion construct, a 5 kb HindIII/BamHI genomic fragment of LD was inserted into a shuttle vector containing the uidA gene (GUS) from E. coli ( Jefferson 1987) and a nopaline synthase terminator. The resulting LD-GUS construct contains 3 kb of LD promoter and 2 kb of LD coding region fused to the GUS gene, and generates a predicted polypeptide that consists of the N-terminal 451 amino acids of LD fused to the N-terminus of the GUS polypeptide (see Fig. 2a). A HindIII fragment containing this entire LD-GUS fusion construct was excised from the shuttle vector and inserted into the HindIII site of pCGN1578. This construct was introduced into the Arabidopsis ecotype RLD and Nicotiana tabacum by Agrobacterium-mediated transformation ( Lee et al. 1994 ). Fixation, staining and clearing of LD-GUS transgenic plants with X-gluc was performed as described previously ( An et al. 1996 ). Stained plants were viewed and photographed through a Leica MZ6 dissecting microscope. For nuclear localization, stained roots were mounted on a slide in 50 m m phosphate buffer, pH 7.2, 20 μg ml–1 4′,6-diamidino-2-phenylindole (DAPI) and photographed through a Nikon microscope using Nomarski optics. For the analysis of LFY expression in ld and ld ap1 cal plants, in situ hybridization was performed as previously described ( Weigel et al. 1992 ).
We are grateful to Manorama C. John for technical assistance, and to the laboratory of Joel Rothman for assistance with microscopy. This research was supported by the College of Agricultural and Life Sciences of the University of Wisconsin, and by a grant to R.M.A. from the National Science Foundation (98070843). I.L. was supported by the S.N.U. Research Fund and Special Program for the Promotion of Graduate Studies, Ministry of Education and the Korea Science and Engineering Foundation through the Research Center for Cell Differentiation. M.J.A. was supported by a Postdoctoral Fellowship (GM15683–02) from the National Institutes of Health.