Expression of CENTRORADIALIS (CEN) and CEN-like Genes in Tobacco Reveals a Conserved Mechanism Controlling Phase Change in Diverse Species

Transcription

1 The Plant Cell, Vol. 11, , August 1999, American Society of Plant Physiologists Expression of CENTRORADIALIS (CEN) and CEN-like Genes in Tobacco Reveals a Conserved Mechanism Controlling Phase Change in Diverse Species Iraida Amaya, a Oliver J. Ratcliffe, b and Desmond J. Bradley a,1 a Sainsbury Laboratory, John Innes Centre, Colney Lane, Norwich NR4 7UH, United Kingdom b Genetics Department, John Innes Centre, Colney Lane, Norwich NR4 7UH, United Kingdom Plant species exhibit two primary forms of flowering architecture, namely, indeterminate and determinate. Antirrhinum is an indeterminate species in which shoots grow indefinitely and only generate flowers from their periphery. Tobacco is a determinate species in which shoot meristems terminate by converting to a flower. We show that tobacco is responsive to the CENTRORADIALIS (CEN) gene, which is required for indeterminate growth of the shoot meristem in Antirrhinum. Tobacco plants overexpressing CEN have an extended vegetative phase, delaying the switch to flowering. Therefore, CEN defines a conserved system controlling shoot meristem identity and plant architecture in diverse species. To understand the underlying basis for differences between determinate and indeterminate architectures, we isolated CEN-like genes from tobacco (CET genes). In tobacco, the CET genes most similar to CEN are not expressed in the main shoot meristem; their expression is restricted to vegetative axillary meristems. As vegetative meristems develop into flowering shoots, CET genes are downregulated as floral meristem identity genes are upregulated. Our results suggest a general model for tobacco, Antirrhinum, and Arabidopsis, whereby the complementary expression patterns of CEN-like genes and floral meristem identity genes underlie different plant architectures. INTRODUCTION Most plants have several phases of growth (Poethig, 1990; Telfer et al., 1997). After germination, the apical meristem goes through a vegetative phase during which it generates leaf primordia bearing axillary meristems. These axillary meristems give rise to side branches or lie dormant until apical dominance is removed (McDaniel et al., 1989; Cline, 1994; Napoli and Ruehle, 1996). When the apical meristem receives the appropriate signals, it switches to reproductive growth (flowering). This switch is tightly controlled by numerous physiological signals and genetic pathways that coordinate flowering with environmental conditions and the developmental stage of the plant (Bernier, 1988; Martinez- Zapater et al., 1994; McDaniel, 1996; Colasanti et al., 1998; Koornneef et al., 1998; Levy and Dean, 1998). When an apical meristem switches from the vegetative to the reproductive phase, it gives rise to an inflorescence, which is the flower-bearing part of the plant. There are two basic types of inflorescences, namely, indeterminate and determinate (Weberling, 1989; Coen and Nugent, 1994). 1 To whom correspondence should be addressed. bradley@ bbsrc.ac.uk; fax Species with indeterminate (or racemose) inflorescences have apical meristems that grow indefinitely, generating floral meristems from their periphery. In contrast, each apical meristem of determinate (cymose) species is eventually transformed into a floral meristem that terminates apical growth, with subsequent growth occurring only from lower axillary meristems. Both Antirrhinum and Arabidopsis have indeterminate inflorescences, whereas species such as tobacco and tomato have determinate inflorescences (Figure 1). The manifestation of these two architectures depends on which meristems give rise to shoots and which to flowers (Weberling, 1989; Coen and Nugent, 1994). This in turn depends on the expression patterns of meristem identity genes. Mutations in the CENTRORADIALIS (CEN) gene of Antirrhinum or its homolog TERMINAL FLOWER1 (TFL1) in Arabidopsis result in the inflorescence meristem being converted into a terminal flower (Figure 1; Shannon and Meeks- Wagner, 1991; Alvarez et al., 1992; Bradley et al., 1996, 1997; Ohshima et al., 1997). Therefore, these genes are required in Antirrhinum or Arabidopsis to maintain an indeterminate shoot identity. Consistent with this role, both genes are expressed at high levels in the inflorescence apex. A complementary set of genes, including FLORICAULA (FLO) and SQUAMOSA (SQUA) in Antirrhinum and LEAFY

2 1406 The Plant Cell Figure 1. Diagrams of Various Types of Plant Architecture. Wild-type (WT) Antirrhinum and Arabidopsis inflorescences are indeterminate. The shoot apical meristem grows indefinitely (arrowheads), and flowers (circles) are produced from the periphery; the first flowers formed are at the most basal position on each shoot. In contrast, the apical meristems of the cen and tfl1 mutants are eventually converted into terminal flowers. Secondary inflorescences also terminate in a flower. In tfl1 mutants, the vegetative and inflorescence phases are shorter than in the wild type. Similar to cen and tfl1 mutants, the wild-type tobacco shoot meristem is determinate and is transformed into a flower; this terminal flower is the first flower formed. Axillary meristems generated below the apex also develop into terminal flowers in a cymose pattern. In wild-type tomato, the vegetative phase is followed by the production of a terminal inflorescence. During the reproductive phase, an axillary meristem forms the new axis of growth and generates three leaves before it terminates in an inflorescence. This process is repeated indefinitely. In self-pruning (sp) mutants, progressively fewer leaves are made in each repeat until two inflorescences terminate growth. (LFY) and APETALA1 (AP1) in Arabidopsis, is required for floral meristem identity (Coen et al., 1990; Irish and Sussex, 1990; Schultz and Haughn, 1991, 1993; Huijser et al., 1992; Mandel et al., 1992; Weigel et al., 1992; Bowman et al., 1993; Gustafson-Brown et al., 1994). In flo or lfy ap1 mutants, floral meristems are replaced by shoot meristems. By contrast, overexpression of LFY or AP1 converts shoots into flowers (Mandel and Yanofsky, 1995; Weigel and Nilsson, 1995). These overexpression phenotypes are similar to those seen in tfl1 mutants. Therefore, indeterminate growth of shoots is associated with CEN or TFL1 activity, whereas expression of floral meristem identity genes promotes flower development. Thus far, the only determinate species in which CEN and TFL1 genes have been analyzed is tomato. In tomato, the SELF-PRUNING (SP) gene was found to be a CEN and TFL1 homolog (Pnueli et al., 1998). Both tobacco and tomato have similar architectures, making terminal structures and generating further shoot growth through lower axillary meristems. In wild-type tomato, the shoot meristem has a vegetative phase that terminates with the formation of an inflorescence (Figure 1). A new axis of growth develops from an axillary meristem and generates three leaves before again terminating in an inflorescence. This process repeats itself indefinitely and represents a sympodial growth pattern (Figure 1). In sp mutants, the vegetative phase is the same as in the wild type, but the sympodial phase is shortened, and eventually the shoot terminates directly in an inflorescence that stops any further growth (Figure 1). Therefore, SP is required to maintain the leaf-producing phase of the sympodial shoots in tomato. Overexpression of SP or CEN rescues the phenotype of the sp mutant, indicating that these two genes are functional homologs. However, in wild-type plants, SP expression appears to occur in all meristems, overlapping with the expression of floral meristem identity genes. Therefore, unlike in Antirrhinum and Arabidopsis, expression patterns give no clues as to how the activities of these two sets of genes might be separated in tomato. To examine whether this difference in expression patterns reflects differences in how determinate and indeterminate architectures are controlled, we analyzed CEN gene function in another determinate species tobacco. The tobacco main shoot terminates in a flower, and no new single axis of growth develops (Figure 1). Below each terminal flower, a number of axillary meristems give rise to additional flowers in a cymose pattern. Secondary shoots develop in the axils of leaves lower on the main shoot. Similar to the main shoot, these branches also have a phase of vegetative growth before terminating in a flower (Figure 1). In most tobacco species, the length of the vegetative phase is independent of day length but is controlled by unidentified developmental signals (McDaniel et al., 1989; McDaniel, 1996). The architecture of tobacco, with the production of a single terminal flower, is similar to cen or tfl1 mutants and could be explained in three ways: (1) CEN and TFL1 function is absent in such species; (2) this function exists, but other

3 Tobacco Plant Architecture 1407 pathways eventually cause the shoot meristem to enter a floral phase; or (3) the expression patterns of CEN-like genes are altered. We have addressed each of these possibilities in this article. By overexpressing CEN in tobacco, we show that the vegetative phase can be extended. Therefore, tobacco has a phase-change mechanism that is responsive to CEN function, which suggests that such a mechanism has been conserved throughout dicot evolution. Most tobacco plants overexpressing CEN eventually make terminal flowers, suggesting that other factors may override CEN function. However, the observed variation in phenotypes suggests that CEN may be limiting and, when expressed at sufficient levels, could promote vegetative growth indefinitely. We have also analyzed the potential function in meristem determinacy of CEN-like genes from tobacco (CET genes). These genes have novel expression patterns. Those most closely related to CEN are not expressed in the main apical meristem. Instead, their expression is limited to axillary meristems, in which expression occurs only during the vegetative or dormant phases. We further show that the CET genes and the tobacco floral meristem identity genes have complementary expression patterns, similar to the case in Antirrhinum and Arabidopsis. This finding suggests a general model for the molecular control of plant architecture that has been conserved between distantly related indeterminate and determinate species. RESULTS CEN Delays Flowering in Tobacco To overexpress CEN in tobacco, we used the constitutive cauliflower mosaic virus 35S promoter (Odell et al., 1985). After transformation with 35S CEN, 34 kanamycin-resistant tobacco plants were obtained. RNA gel blot analysis revealed that most of these plants strongly expressed CEN (Figure 2A). Seven plants that were kanamycin resistant but did not express CEN had a phenotype indistinguishable from that of the wild type. In contrast, in 90% of the 27 lines overexpressing CEN, flowering was considerably delayed. Wild-type tobacco plants grown under long-day conditions generated 30 to 35 leaves over a period of 3 months before producing a terminal flower. Subsequent flowers developed from axillary meristems below the apex (McDaniel et al., 1989; Figure 1). The majority of 35S CEN primary transformants flowered much later than did the wild type, both in terms of time and the number of leaves made before flowering. The transformants were divided into groups (A to E) based on the number of leaf nodes produced before the terminal flower (Figure 2B). Most transformants did not flower before 5 months; in four extreme transformants, flowering was delayed for 10 months (Figure 2C). The inflorescence architecture of 35S CEN tobacco plants was very similar to that of the wild type, although slightly fewer flowers were generally produced compared with control plants. In one case (AN33), however, the main shoot never flowered and made 120 leaves before senescing. This plant produced many branches that behaved in a manner similar to the main shoot. Eventually, a single branch from this plant, originating from the axil of leaf number 60, generated 130 more leaves before making two flowers. To confirm the effect of 35S CEN, several primary transformants were self-pollinated, and seeds were harvested for analysis in the next generation. Segregation for kanamycin resistance and DNA gel blot analysis of line AN34 showed that the delay in flowering segregated with the presence of the transgene: plants containing 35S CEN generated leaf nodes before the terminal flower, whereas wild-type siblings generated less than half as many ( ; 15 plants analyzed). Also, representative lines from groups C, D, and E (Figure 2B), which contained only one T-DNA insertion, were selected. Seeds from these primary transformants were germinated on kanamycin, and kanamycin-resistant seedlings were transferred to soil and grown under greenhouse conditions next to wild-type tobacco plants. The three selected lines generated approximately twice as many leaf nodes as did wild-type plants: lines AN1 (group C), AN34 (group D), and AN21 (group E) produced 60 7, , and leaves, respectively, compared with leaves for the wild type (nine to 11 plants from each line were scored). Therefore, tobacco plants overexpressing CEN had an extended vegetative phase and a delayed transition to the reproductive phase. Although the conversion of the apical meristem to floral identity was not prevented in most cases, it was strongly delayed. Tobacco plants containing 35S TFL1 were generated at the same time as 35S CEN plants. More than 50 kanamycin-resistant primary transformants were produced and grown under greenhouse conditions next to 35S CEN transformants and wild-type controls. The 35S TFL1 tobacco plants showed no notable differences in flowering time or inflorescence architecture when compared with the wild type: both the wild type and 35S TFL1 flowered after producing 30 to 35 leaves. RNA gel blot analysis of 11 independent primary transformants confirmed that they were expressing TFL1 (data not shown). To analyze any effects of overexpressing TFL1 in tobacco, we germinated seeds from four self-pollinated primary transformants on kanamycin. More than 10 kanamycin-resistant seedlings from each line were grown, and the number of leaf nodes to the terminal flower was counted. Wild-type control plants generated leaves, compared with , , , and generated by 35S TFL1 lines NH7, NH10, NH14, and NH37, respectively. Therefore, in contrast to overexpression of CEN, overexpression of TFL1 did not significantly delay flowering time in tobacco.

4 1408 The Plant Cell Isolation of CEN-like Genes from Tobacco The results obtained from the overexpression of CEN in tobacco indicated that a pathway in tobacco can respond to CEN. Therefore, it is possible that tobacco has CEN-like genes and that these affect phase transition. To assess whether there are CEN-like genes in tobacco (CET genes), we screened a tobacco genomic library at moderate stringency with CEN and obtained the four genes CET1 to CET4. Three more CET genes (CET5 to CET7) were identified in reverse transcription polymerase chain reaction (RT-PCR) experiments using degenerate primers conserved among the CEN, TFL1, and CET genes with RNA from young tobacco inflorescences. The cdnas were isolated and sequenced, and the predicted open reading frames were compared with the available corresponding genomic regions and with CEN. The CET genes have three introns in the same positions as those in CEN and TFL1, but these introns are of different sizes. The CET2 and CET4 putative coding regions show 97% DNA sequence identity, and the 5 and intron regions are also very similar. The predicted proteins differ in only six amino acids and are 83% identical to CEN. Therefore, CET2 and CET4 probably represent single-copy genes in the diploid progenitors of the allotetraploid tobacco (Gray et al., 1974; Okamuro and Goldberg, 1985). CET5 and CET6 are 93% identical to each other and 74% identical to CEN. CET7 is 93% identical to CET1 and 67% identical to CEN. Only partial sequence data were available for CET3, and the predicted exon showed similarity to CEN and the CET genes. However, the predicted length of 71 amino acids was too short for significant comparisons. All predicted proteins were compared with CEN and its homologs from Arabidopsis (TFL1) and tomato (SP) (Figure 3A). This comparison indicated that CET2/CET4, SP, and CEN are more closely related than are any of the others. Use of alignment programs other than CLUSTAL W (such as PAUP) and boot-strapping analyses with rooted and unrooted trees (data not shown), consistently grouped CET2, CET4, CEN, and SP together. These analyses suggest that all other CET gene products are more distantly related, although most trees suggested that CET5 and CET6 are more similar to TFL1 from the distantly related species Arabidopisis. Figure 2. Effect of 35S CEN on the Tobacco Vegetative Phase. (A) RNA gel blot of primary transformants (AN1 to AN34) and wildtype plants (WT). Five micrograms of total RNA from leaves was blotted and probed with a CEN cdna probe. (B) Number of leaves produced before the terminal flower in wild type (WT) and 35S CEN primary transformants (groups A to E). 35S CEN tobacco plants were divided into groups based on the number of leaves produced: group A, 31 to 40 leaves before the terminal flower; group B, 41 to 50 leaves; group C, 51 to 60 leaves; group D, 61 to 70 leaves; and group E, 71 leaves produced. The number inside each bar represents the number of plants grouped. (C) The 35S CEN tobacco line AN33 (left) not flowering 15 months after germination compared with a 4-month-old flowering wild-type plant (right).

5 Tobacco Plant Architecture 1409 An alignment was made of selected CEN-like proteins including both CET2 and CET4, as well as CET1 and a CET5 fragment, the latter two as examples of two other genes duplicated as CET1/CET7 and CET5/CET6 (Figure 3B). The amino acids defined as important by point mutations in tfl1 alleles (Bradley et al., 1997; Ohshima et al., 1997) or in sp mutants (Pnueli et al., 1998) were conserved among all CET genes analyzed, suggesting that these genes are likely to produce functional proteins (Figure 3B). These results together with DNA gel blot analysis of petunia DNA using one CEN-like gene at low stringency (data not shown) and the data of Pnueli et al. (1998) suggest that species in the Solanaceae possess a family of approximately three or four CENlike genes. CET2 and CET4 Are Expressed in Axillary Meristems during the Vegetative Phase Figure 3. Sequences of CEN-like Gene Products Are Highly Conserved among Diverse Plant Species. (A) Distance tree of various CEN-like gene products deduced from the alignment in (B) by using the program CLUSTAL W. The lengths of the horizontal lines are proportional to the similarity between the predicted protein sequences. (B) Comparison of the predicted amino acid sequences for different tobacco CEN-like genes and for those of CEN (Bradley et al., 1996), TFL1 (Bradley et al., 1997; Ohshima et al., 1997), and SP (Pnueli et al., 1998). Identical residues are in black. Dashed lines indicate gaps introduced by the program to achieve maximum alignment. Intron positions conserved among all genes are marked by arrows. Asterisks indicate amino acids in which point mutations have been described for Arabidopsis (Bradley et al., 1997; Ohshima et al., 1997) and tomato (Pnueli et al., 1998). The CET5 sequence is partial and spans only a fragment of the predicted protein. CET2 and CET4 are both 83% identical in amino acids to CEN and 90% to SP, whereas CET1 represents the most distantly related class with 68% identity in amino acids to CEN and SP. To assess the potential function of the CET genes most closely related to CEN, we analyzed the expression pattern of CET2 and CET4 RNA by using in situ hybridization. Tobacco apices were harvested at several developmental stages (time points 0 to 4), from shortly after germination to when the terminal flower was being produced, and longitudinal sections were then examined. Both genes were expressed in the same pattern from very early to late in development (Figure 4; CET4 data only). As with CEN in Antirrhinum and TFL1 in Arabidopsis, CET2 and CET4 RNA appeared to be absent from the epidermal cell layer. In seedlings with only two or three visible leaves (time point 0), CET2 and CET4 RNA was detected in meristems in the axils of leaves (Figures 4A and 4B). At time point 1, when tobacco plants had generated 10 visible leaves, CET2 and CET4 were strongly expressed in axillary meristems; this pattern of expression persisted throughout the vegetative phase to the transition to flowering (Figures 4D, 4F, and 4H). In young axillary meristems recently generated near the apex of plants with 16 leaves (time point 2), CET2 and CET4 expression was strong and could be detected in most of the cells (Figure 4F). However, in older axillary meristems lower in the stem of plants with 20 to 25 leaves, expression was weaker and more restricted to the region below the meristematic dome (Figure 4H). Shortly after the production of the terminal flower (time point 4), CET2 and CET4 expression was not detected in the axillary meristems just below the terminal flower (Figure 4I). These meristems gave rise to flowers. Therefore, in contrast to Antirrhinum and Arabidopsis, in which CEN and TFL1 are expressed in the primary shoot apex, we failed to detect any CET2 or CET4 RNA in the primary shoot apical meristem of tobacco. The expression of both CET2 and CET4 was restricted to axillary meristems during the vegetative phase. In Arabidopsis, TFL1 is also expressed in axillary meristems, which give rise to secondary inflorescences (Ratcliffe et al., 1999). The expression patterns of the other CET gene classes,

6 1410 The Plant Cell Figure 4. Expression Patterns of CET4 and NFL, the Likely Nicotiana FLO/LFY Homolog, in Wild-Type Tobacco Plants. Longitudinal sections of tobacco were harvested at several time points and probed with digoxigenin-labeled antisense CET4 or NFL RNA. The RNA signal is seen as a dark blue black color on a light blue tissue background. (A) to (C) At time point 0, tobacco seedlings had generated 2 or 3 leaves, and CET4 ([A] and [B]) was already expressed in axillary shoot meristems (Ax) that did not express NFL (C). NFL RNA was detected in leaf primordia arising from the apex. NFL is expressed on the periphery of the main apical meristem (AM). (D) At time point 1, the apex had generated 11 visible leaves, and CET4 RNA was strongly expressed in young axillary meristems. (E) NFL expression was restricted to the flanks of the main shoot meristem. (F) At time point 2, the apex had generated 16 leaves. The axillary meristem is older than is the one shown in (D), and expression of CET4 is in the region just below the central dome. (G) An adjacent section of the same axillary meristem shows expression of NFL in leaf primordia arising on the flanks. (H) An even older axillary vegetative meristem, from a position lower in the main shoot, is shown from a plant at time point 3; expression of CET4 is more restricted to the central region. (I) At time point 4, wild-type tobacco plants had produced a terminal flower and several axillary floral meristems below. CET4 expression was not detected in these meristems. (J) In contrast, NFL was strongly expressed in sections adjacent to the meristems shown in (I). Bars in (C), (E), (G), (H), and (J) 200 m for (A) to (J).

7 Tobacco Plant Architecture 1411 represented by CET1 and CET6, also were investigated by using RNA in situ hybridization. No expression could be detected in any tissues, from embryonic stages through flowering. However, expression of both genes was detected by RT-PCR when tissue from vegetative and inflorescence shoots was used but it could not be localized to specific meristematic or tissue domains. NFL and CET2/CET4 Are Expressed in Separate Domains In Antirrhinum and Arabidopsis, floral meristem identity genes, such as FLO and LFY, respectively, are expressed in domains separate from CEN and TFL1, suggesting an antagonistic relationship between these genes (Shannon and Meeks-Wagner, 1993; Weigel and Nilsson, 1995; Bradley et al., 1996, 1997). Whereas CEN and TFL1 are expressed in the apical meristem, FLO and LFY are restricted to peripheral meristems. In cen or tfl1 mutants, FLO or LFY become ectopically expressed throughout the apical meristem. Conversely, ectopic LFY expression prevents TFL1 expression in the shoot apex (Weigel and Nilsson, 1995; Ratcliffe et al., 1999). In tobacco, the expression pattern of NFL (for Nicotiana FLO LFY), the likely FLO/LFY homolog, is different from that of FLO in Antirrhinum and LFY in Arabidopsis (Kelly et al., 1995). NFL is expressed in the shoot apical region throughout development but not in the apical meristem itself; NFL is expressed in a ring of cells outside the central dome in floral meristems and leaf primordia. To assess whether NFL and CET2 and CET4 were coexpressed in axillary meristems or were restricted to separate domains, as occurs in Antirrhinum and Arabidopsis, we compared the expression patterns of NFL with those of CET2 and CET4 by using in situ hybridization (Figure 4). During the vegetative phase, young axillary meristems near the apex of wild-type tobacco plants strongly expressed CET2 and CET4 (Figures 4A, 4B, and 4D), whereas NFL RNA was not detected in these axillary meristems but was restricted to the flanks of the main shoot meristem (Figures 4C and 4E; Kelly et al., 1995). In older axillary meristems, after some leaf primordia had been generated, NFL was detected in a pattern similar to that in the apical meristem (Figure 4G), and CET2 and CET4 were expressed in the central region below the apical dome of these meristems (Figure 4F). In the terminal and axillary floral meristems, NFL expression was strong (Figure 4J), whereas CET2 and CET4 were not expressed (Figure 4I). Therefore, the expression domains of CET2 and CET4 and NFL did not appear to overlap significantly. Overexpression of CEN Delays the Downregulation of CET2 and CET4 In tobacco plants overexpressing CEN, the switch to reproductive development was delayed. To investigate whether the expression of NFL was downregulated or whether CET2 and CET4 were affected, we performed RNA in situ hybridization using 35S CEN tobacco plants and compared the results with those from the wild type (Figure 5; CET4 data only). Apices of 35S CEN lines AN21 and AN34 and wildtype controls were harvested at different time points. At time point 1, 58 days after germination, the tobacco plants had generated 11 or 12 visible leaves, and there were no differences between 35S CEN and control plants; NFL was expressed in the periphery of the apex and developing primordia, whereas CET2 and CET4 were expressed in young axillary meristems (Figures 5E and 5F). At time point 2 (73 days), plants had generated 16 visible leaves, and the expression of the genes remained unchanged. Similarly, at time point 3 (93 days), these expression patterns were maintained in wild-type (Figures 5A and 5B) and in 35S CEN plants (Figures 5G and 5H). At time point 4 (104 days), the plants had 26 to 30 visible leaves. At this time, wild-type apices had formed a terminal flower, and CET2 and CET4 mrnas were absent from axillary buds, giving rise to flowers (Figure 5C). NFL was expressed strongly in these axillary floral meristems (Figure 5D). In contrast, the 35S CEN lines continued to generate leaves with axillary meristems expressing CET2 and CET4 and with NFL transcripts restricted to the periphery of the shoot apex. This expression pattern was maintained at time point 5 (135 days), when 40 to 44 leaves were visible in 35S CEN plants and no terminal flowers had been produced (Figures 5I and 5J). Eventually, after generating 60 leaves, the 35S CEN plants produced a terminal flower, and the expression patterns of CET2 and CET4 and NFL became similar to those in a wild-type inflorescence. These results suggest that overexpression of CEN in tobacco does not delay flowering by directly modifying the expression of NFL. Rather, by extending the vegetative phase, the 35S CEN transgene delayed the downregulation of CET2 and CET4 in axillary meristems. Development of Axillary Meristems Correlates with Downregulation of CET2 and CET4 During growth of tobacco, we observed that shoot development correlated with high NFL and low CET2 or CET4 expression. This pattern was found for the primary shoot meristem and inflorescences developing below the terminal flower. In contrast, vegetative or dormant axillary meristems had high CET2 and CET4 expression. These meristems showed little development during the vegetative phase of the main shoot due to apical dominance (McDaniel and Hsu, 1976; McDaniel et al., 1989). To analyze the development of these buds more directly, we grew wild-type tobacco plants until they generated 12 leaves but had not yet flowered. Primary shoot apices were then excised with a scalpel to remove apical dominance and promote growth of axillary shoots (Figure 6A). Plants were harvested every 2 days for use in RNA in situ hybridization experiments and compared

8 1412 The Plant Cell with untreated controls. In plants with their primary apices removed, outgrowth of all axillary shoots near the apex was observed by eye at 6 days. Control plants showed little development of similarly positioned meristems (Figures 6B and 6C; CET4 and NFL data only). RNA in situ analysis revealed a reduction in CET2 and CET4 expression in these developing axillary shoots from day 4 (Figure 6D) through day 10 (Figure 6F). In contrast, NFL expression increased from day 4 (Figure 6E) to day 10 (Figure 6G). By 12 days, CET2 and CET4 RNAs were no longer detected in the developing axillary shoot apices but were strongly expressed in axillary meristems generated from their periphery (Figure 6H). NFL was now strongly expressed in the peripheral region of the developing axillary shoot (Figure 6I). Therefore, outgrowth of axillary meristems correlates with a reduction in CET2 and CET4 expression and an increase in that of NFL. DISCUSSION Antirrhinum and tobacco belong to different plant families and represent two forms of plant architecture, those with indeterminate shoots and those forming terminal flowers. We show that these different species can both respond to the overexpression of CEN from Antirrhinum. Therefore, the pathway influenced by CEN may be an ancient and fundamental mechanism that controls the identity of the shoot meristem. By comparing the patterns of endogenous CENlike gene expression in these species, we suggest a molecular basis for the control of different plant architectures. Patterns of CEN Expression May Control Different Plant Architectures Figure 5. Expression Patterns of CET4 and NFL in Wild-Type Tobacco Apices Compared with Those in Plants Expressing 35S CEN. (A) to (D) Expression of CET4 ([A] and [C]) and NFL ([B] and [D]) in wild-type plants at different time points. (A) and (B) are from a plant just before flowering (time point 3), whereas (C) and (D) are apices from plants after the production of the terminal flower (time point 4). (E) to (J) Expression of CET4 ([E], [G], and [I]) and NFL ([F], [H], and [J]) in 35S CEN plants. At time point 1 (11 or 12 leaves), the expression of CET4 (E) and NFL (F) was indistinguishable from that of time point 1 in a wild-type shoot. Similarly, at time point 3 (25 to 27 leaves) in both wild-type and 35S CEN apices, CET4 was strongly expressed in axillary meristems (G), whereas NFL was expressed in a ring of cells around the apical meristem and in leaf primordia (H). Note that by time point 4, the wild type had flowered and only NFL rather than CET4 was expressed in floral meristems. In 35S CEN In Antirrhinum, Arabidopsis, and tobacco, expression of CEN-like genes is localized to specific shoot meristems. In Antirrhinum, CEN expression is limited to the inflorescence meristem, where it maintains the inflorescence phase. Its functional homolog in Arabidopsis, TFL1, acts during all phases of shoot meristem development, with weak expression during the vegetative phase and upregulation upon switching to an inflorescence (Bradley et al., 1997). In both cases, localized expression in the apex acts upon the shoot meristem to prolong the phase in which it is expressed. Ex- plants, the delay in flowering correlated with a delay in the change in pattern and the downregulation of CET4 in axillary floral meristems. At time point 5, when 40 to 44 leaves had been produced, the expression pattern of CET4 (I) and NFL (J) remained unchanged in the 35S CEN apices. Bars in (D) and (J) 200 m for (A) to (J).

9 Tobacco Plant Architecture 1413 pression of the tobacco genes most similar to CEN CET2 and CET4 is restricted to axillary meristems and is absent from the main shoot. Therefore, one role for CET2 or CET4 may be to prolong the vegetative phase of axillary meristems, ensuring that they generate branches rather than flowers. In contrast, expression of SP, the CEN homolog in tomato, occurs in all meristems (vegetative, inflorescence, floral, and axillary) and is also found in leaf and floral organ primordia (Pnueli et al., 1998). How SP influences only the reproductive shoot is therefore unclear. Except for tomato, all of these patterns may reflect a common mechanism of antagonism between CEN genes and floral meristem identity genes. Antirrhinum and Arabidopsis show complementary expression patterns of CEN and TFL1 and their respective floral meristem identity genes, such as FLO and LFY. At the apex, CEN and TFL1 are expressed in a central domain, whereas FLO and LFY are restricted to primordia or meristems at the flanks. These sets of genes are antagonistic, with CEN and TFL1 preventing FLO and LFY expression in the apical meristem. Furthermore, by analyzing expression of TFL1 in plants that make terminal flowers (through ectopic expression of LFY or in tfl1-1 mutants), it was shown that in apical meristems with LFY expression, TFL1 is not expressed (Weigel and Nilsson, 1995; Ratcliffe et al., 1999). In tobacco, the likely CEN homologs CET2 and CET4 also show expression patterns complementary to NFL, the likely tobacco homolog of FLO and LFY. It appears that NFL expression is established first in the main apex of tobacco, which may prevent CET2 and CET4 from being expressed in the primary shoot. This may explain why the apex forms a terminal flower. However, in axillary meristems, CET2 and CET4 expression may be established early to promote a phase of leaf development. As NFL expression increases in these meristems, CET2 and CET4 expression becomes progressively more restricted to the central domain until the genes are downregulated upon flowering. Therefore, CET2 or CET4 may extend the vegetative phase of axillary meristems so that they form leafy branches rather than directly switching to flowers. The antagonism of CET2/CET4 and NFL is further highlighted by expression patterns in axillary Figure 6. Changes in CET4 and NFL Expression Patterns in Developing Axillary Shoots of Wild-Type Tobacco. (A) To promote axillary shoot development, we removed the apex of the main shoot (AM) (dashed line). This removed apical dominance and stimulated outgrowth of the axillary shoots (Ax). (B), (D), (F), and (H) CET4 expression. (C), (E), (G), and (I) NFL expression. Expression was compared in control, uncut apices at day 0 ([B] and [C]) and in axillary shoots at 4 days ([D] and [E]), 10 days ([F] and [G]), and 12 days ([H] and [I]) after the removal of the main apex. Note the decrease in CET4 expression in the developing axillary shoot meristems as NFL expression increased on the periphery of the meristem. After 12 days, CET4 RNA was not detected in the axillary shoot meristem (H). However, by this time, the shoot had generated its own leaves, bearing further axillary meristems, which expressed CET4. Bars in (C) and (I) 200 m for (B) to (I).

10 1414 The Plant Cell meristems just below the terminal flower. These meristems do not express CET2 or CET4 but have strong NFL expression from early in their development. This pattern gives rise to terminal flowers arranged in a cyme (Figure 1). Different architectures in Antirrhinum, Arabidopsis, or tobacco may occur by changing the expression of CEN genes or their antagonists, the floral meristem identity genes, such as FLO and LFY or SQUA and AP1 (Huijser et al., 1992; Mandel and Yanofsky, 1995; Weigel and Nilsson, 1995; Bradley et al., 1996; Ratcliffe et al., 1998, 1999). Floral meristem identity genes appear not to be expressed in the shoot apical meristem of indeterminate species but are restricted to primordia or meristems arising from the periphery. In contrast, many determinate species, such as tobacco, tomato, Impatiens balsimina, and petunia, have FLO- and LFY-like genes that are expressed strongly in the shoot apical meristem from very early in development (Kelly et al., 1995; Pouteau et al., 1997; Pnueli et al., 1998; Souer et al., 1998). Recent data showed that LFY and AP1 repress the expression of TFL1 in Arabidopsis (Ratcliffe et al., 1999). Therefore, in species such as tobacco, CEN-like gene activity may be repressed in the apical meristem, resulting in the production of terminal flowers rather than shoots. This model fits with the analysis of CEN-like and FLO-like genes in tobacco. The CET2 and CET4 genes are not expressed in the apical meristem of the main shoot but are restricted to axillary vegetative meristems. In contrast, the NFL genes are expressed in the main apical meristem from very early in development. Furthermore, when the NFL genes are upregulated in axillary meristems, the CET genes are downregulated. These data suggest a model for Arabidopsis, Antirrhinum, and tobacco in which the production of flowers or shoots depends on the relative expression patterns of CEN-like genes and floral meristem identity genes. In tomato, overexpression of CEN or its endogenous homolog SP can also affect plant architecture. However, it was shown that SP is normally expressed in all meristems, axillary and floral, and most primordia throughout development (Pnueli et al., 1998). This pattern coincides with tomato FLOlike gene expression, rather than being complementary as in other species. How this pattern directs normal development is not clear. CEN Functions in Diverse Species The Antirrhinum CEN gene can affect phase change in tobacco, a species that makes terminal flowers. Overexpression of CEN in tobacco extends the vegetative phase, showing that tobacco has a pathway that can respond to CEN. However, most tobacco plants overexpressing CEN still make terminal flowers. This suggests that determinate species may enter a floral phase that overrides CEN activity. However, in one transformant, flowering was delayed extensively, suggesting that sufficient levels of CEN activity might prevent flowering indefinitely. In contrast to CEN, overexpression of TFL1 has no effect on tobacco development. This may reflect divergence in the CEN and TFL1 proteins. Tobacco and Antirrhinum are much closer to each other phylogenetically than they are to Arabidopsis, so CEN may interact with the phase-change mechanism in tobacco more readily than does TFL1. It is possible that plants have a number of pathways affecting the phasechange mechanism but that tobacco may lack a pathway responsive to TFL1. However, tobacco does retain a pathway that responds to the Antirrhinum CEN gene. Perhaps, by moving CEN-like genes between other indeterminate and determinate species, the nature of this difference in response may be revealed. The SP gene is the functional homolog of CEN in tomato and has high similarity to CEN and TFL1 (Pnueli et al., 1998). Overexpression of SP or CEN in tomato rescues sp mutants. Furthermore, overexpression in some backgrounds can result in more leaves being made in each repeating unit during the sympodial phase. These phenotypes can be interpreted as phase-change effects. In tomato, SP is required to maintain the phase of leaf production in each repeating sympodial unit. Therefore, CEN-like genes in Antirrhinum, Arabidopsis, tomato, and tobacco can function to extend phases of shoot meristems and therefore delay the production of flowers. Overexpression of CEN in tomato did not affect the vegetative phase, only the sympodial phase after production of the first inflorescence. This suggests a difference in the response of the vegetative shoot meristem of tomato compared with tobacco. Alternatively, tobacco may have two phases before the production of the terminal flower. The first is similar to tomato and cannot respond to CEN. However, after the required number of leaf nodes have been made to allow flowering, as shown by grafting experiments (McDaniel et al., 1989), the tobacco shoot may enter a second phase when CEN may be able to delay the switch to flowering. In tomato, there may be only a single phase, after which the switch to the reproductive and sympodial phases occurs, so that SP or CEN may act only after this switch. Arabidopsis and Antirrhinum have similar architectures but are members of widely diverged families of flowering plants. Although CEN and its homolog TFL1 in Arabidopsis both control inflorescence architecture, TFL1 has additional functions during vegetative development. By expressing CEN in tobacco, we have shown that CEN can also function during the vegetative phase. In Antirrhinum, CEN expression is limited to the inflorescence meristem. The effects in tobacco suggest that CEN could also act during vegetative development in Antirrhinum if it was expressed earlier. It remains unclear whether such a pattern of expression was lost during evolution such that CEN became restricted to the inflorescence shoot meristem or TFL1 expression was recruited to earlier development. CEN-like genes affect a common mechanism controlling phase change. This mechanism acts throughout the life cycle of plants in both vegetative and inflorescence shoots.

11 Tobacco Plant Architecture 1415 This effect on phase change is shown by the action of CEN in Antirrhinum and tobacco and TFL1 in Arabidopsis. Also, this mechanism has been highly conserved across species because CEN can function in both Antirrhinum and tobacco, diverse species that have very different architectures. Furthermore, these genes appear to act via a similar molecular mechanism that indirectly delays the upregulation of floral meristem identity genes. In Arabidopsis, TFL1 delays the upregulation of LFY and AP1 in meristems initiated from the flanks of the inflorescence shoot meristem. In tobacco, expression of CEN delays the change in pattern of NFL and its upregulation in axillary meristems. Further analysis of CEN may show how this common mechanism operates in different plant species to control natural variation in plant architectures. RNA in Situ Hybridization In situ hybridization with digoxigenin-labeled antisense RNA was performed on 7- m longitudinal tissue sections as described by Coen et al. (1990) and Jackson (1991). RNA probes were generated using either T3 or T7 polymerases from appropriate plasmids. The CEN antisense RNA probe was generated from pjam2020 (Bradley et al., 1996), the TERMINAL FLOWER1 (TFL1) RNA probe was from pjam2045 (Bradley et al., 1997), the LEAFY (LFY) RNA probe was from plasmid pdw122 (kindly provided by D. Weigel, Salk Institute, La Jolla, CA; Weigel et al., 1992), and APETALA1 (AP1) was from pky89 (kindly provided by M. Yanofsky, University of California at San Diego; Mandel et al., 1992). NFL RNA probes were generated from plasmids pjam2233 and pjam2234 (kindly provided by R. Elliott and E. Coen, John Innes Institute). Signal was detected as a dark blue color on a light blue background when viewed under the light microscope. METHODS Nomenclature For consistency, we have used the Arabidopsis thaliana system of gene nomenclature for all species discussed in this article. Tobacco Transformation The CENTRORADIALIS (CEN) cdna was released from the plasmid pjam2021 by digestion with NcoI and XbaI and ligated into the NcoI- XbaI sites of SLJ4D4 containing the cauliflower mosaic virus 35S promoter (Jones et al., 1992). The product of this ligation, pjam2190, was digested with EcoRI and HindIII to release the 35S CEN OCS (octopine synthase terminator) fragment. This fragment was cloned into the EcoRI-HindIII sites of the binary vector SLJ44024A (Jones et al., 1992), which confers kanamycin resistance, to give pjam2075. Tobacco plants (Nicotiana tabacum cv Samsun) were transformed with pjam2075 (for CEN overexpression) and pjam2076 (for TFL1 overexpression; Ratcliffe et al., 1998), according to standard procedures using Agrobacterium tumefaciens (Horsch et al., 1985). Plant Growth Conditions Tobacco plants were grown in soil in a greenhouse at 20 and 18 C, day and night temperatures, respectively. During the months of October to April, 16-hr photoperiods were maintained with supplementary lighting. The number of leaf nodes in the tobacco time-course experiments was established by counting the number of leaves on the primary shoot from the last leaf (1 cm long) produced to the most basal leaf. When the terminal flower was visible to the naked eye, all of the leaf nodes on the primary shoot below the terminal flower were counted. In experiments to promote axillary shoot development, apices were cut with a scalpel under a dissection microscope. In each case, the apical region that included the first three to five leaves was removed. All values showing ( ) represent the standard error of the mean with 95% confidence limits. Cloning Techniques and Materials A EMBL4 genomic library, which was made from a partial MboI digest of tobacco (cv Samsun) DNA, was kindly provided by C. Martin (John Innes Institute). Approximately 600,000 recombinants were screened at a moderate stringency of 60 C, with washes at 60 C in 2 SSC (0.3 M NaCl and 0.1 M sodium citrate, ph 7.4) and 0.5% (w/ v) SDS using the full-length CEN cdna. Four positive clones were isolated, and CEN-hybridizing bands were subcloned into the pbluescript SK vector (Stratagene, La Jolla, CA) or puc118 (Clontech, Palo Alto, CA). Each clone contained a different sequence with similarity to CEN, and they were named CEN-like tobacco clones 1 to 4 (CET1 to CET4). The CET1, CET2, and CET4 clones contained fulllength genes, whereas CET3 contained a fragment with similarity to only the fourth exon of CEN. Oligonucleotides flanking the CET1, CET2, and CET4 gene open reading frames were designed and used for reverse transcription polymerase chain reaction (RT-PCR) to isolate the corresponding cdna fragments. An alternative strategy to isolate CEN-like genes employed primers conserved between CEN of Antirrhinum and its homolog, TFL1, of Arabidopsis. Primer G1745 (5 -AATGGCCATGAGCTCTTTCCTTC- 3 ) was used in combination with primer G3039 (5 -TG G / T ATC- CCTATG C / T T C / T GGCCTTGG-3 ) or G1750 (5 -CT T / C CTGGCAGC G / A - GT T / C TC G / T C G / T CTG-3 ) for PCR with genomic DNA as well as RT- PCR with RNA from young tobacco inflorescences. Genomic fragments of two novel homologs, CET5 and CET6, and a fragment identical to the corresponding region of CET4 were isolated. From tobacco inflorescence RNA, we isolated fragments corresponding to CET2, CET4, and CET6. New degenerate primers, S41 and S42, that were conserved between CEN, TFL1, and the above-mentioned CET clones were designed to screen for any further genes. Primers S41 (5 -AATGGICATGA G / A CTCTT T / G CC-3 ) and S42 (5 -AAGAA G / A / T AC A / T / G - GCIGC A / G AC A / T GG-3 ) were used for RT-PCR with tobacco inflorescence RNA. Fragments corresponding to CET1, CET2, and novel CET7 were isolated. Primer annealing steps in the PCR programs used were performed at 50 to 55 C. Preparation of RNA and DNA and PCR techniques were conducted according to standard procedures (Frohman et al., 1988; Ausubel et al., 1995). Sequences were determined using the ABI Prism system (Perkin-Elmer), and sequence analysis and alignments were produced using programs from the Genetics Computer Group package (Madison, WI) and the program CLUSTAL X (Thompson et al., 1997). The sequence data

12 1416 The Plant Cell have been submitted to GenBank as accession numbers AF for CET1, AF for CET2, AF for CET4, and AF for CET5. ACKNOWLEDGMENTS We thank Cathie Martin for generously providing the tobacco genomic DNA library, Mark Chadwick for help in screening, Steven Rothstein for help with 35S CEN cloning, Robert Elliott and Enrico Coen for tobacco NFL probes, Patrick Bovill and David Baker for sequencing, and Alan Cavill and Coral Vincent for technical advice. For critical discussions, we thank Enrico Coen and Pilar Cubas. For continuous support, advice, and comments on the manuscript, we especially thank Enrico Coen and Rosemary Carpenter. We are grateful for the generous support of a European Community Marie Curie Fellowship to I.A., a John Innes Foundation Studentship to O.J.R., and a Biotechnology and Biological Sciences Research Council (BBSRC) Research Fellowship to D.J.B. I.A. and D.J.B. were further supported by the Sainsbury Laboratory. Received February 4, 1999; accepted May 12, REFERENCES Alvarez, J., Guli, C.L., Yu, X., and Smyth, D.R. (1992). terminal flower: A gene affecting inflorescence development in Arabidopsis thaliana. Plant J. 2, Ausubel, F.M., Brent, R., Kingston, R.E., Moore, D.D., Seidman, J.G., Smith, J.A., and Struhl, K., eds (1995). Current Protocols in Molecular Biology (New York: John Wiley and Sons). Bernier, G. (1988). The control of floral evocation and morphogenesis. Annu. Rev. Plant Physiol. Plant Mol. Biol. 39, Bowman, J.L., Alvarez, J., Weigel, D., Meyerowitz, E.M., and Smyth, D. (1993). Control of flower development in Arabidopsis thaliana by APETALA1 and interacting genes. Development 119, Bradley, D., Carpenter, R., Copsey, L., Vincent, C., Rothstein, S., and Coen, E.S. (1996). Control of inflorescence architecture in Antirrhinum. Nature 376, Bradley, D., Ratcliffe, O., Vincent, C., Carpenter, R., and Coen, E. (1997). Inflorescence commitment and architecture in Arabidopsis. Science 275, Cline, M. (1994). The role of hormones in apical dominance: New approaches to an old problem in plant development. Physiol. Plant. 90, Coen, E.S., and Nugent, J.M. (1994). Evolution of flowers and inflorescences. Development (suppl.) 120, Coen, E.S., Romero, J.M., Doyle, S., Elliott, R., Murphy, G., and Carpenter, R. (1990). FLORICAULA: A homeotic gene required for flower development in Antirrhinum majus. Cell 63, Colasanti, J., Yuan, Z., and Sundaresan, V. (1998). The indeterminate gene encodes a zinc finger protein and regulates a leaf-generated signal required for the transition to flowering in maize. Cell 93, Frohman, M.A., Dush, M.K., and Martin, G.R. (1988). Rapid production of full length cdnas from rare transcripts: Amplification using a single gene-specific oligonucleotide primer. Proc. Natl. Acad. Sci. USA 85, Gray, J.C., Kung, S.D., Wildman, S.G., and Sheen, S.J. (1974). Origin of Nicotiana tabacum detected by polypeptide composition of fraction 1 protein. Nature 252, Gustafson-Brown, C., Savidge, B., and Yanofsky, M.F. (1994). Regulation of the floral homeotic gene APETALA1. Cell 76, Horsch, R.B., Fry, J.E., Hoffmann, N.L., Eichholtz, D., Rogers, S.G., and Fraley, R.T. (1985). A simple and general method of transferring genes into plants. Science 227, Huijser, P., Klein, J., Lonnig, W.E., Meijer, H., and Sommer, H. (1992). Bracteomania, an inflorescence anomaly, is caused by the loss of function of the MADS-box gene squamosa in Antirrhinum majus. EMBO J. 11, Irish, V.F., and Sussex, I.M. (1990). Function of the apetala-1 gene during Arabidopsis floral development. Plant Cell 2, Jackson, D. (1991). In situ hybridisation in plants. In Molecular Plant Pathology: A Practical Approach, D.J. Bowles, S.J. Gurr, and M. McPhereson, eds (Oxford, UK: Oxford University Press), pp Jones, J.D.G., Shlumukov, L., Carlan, F., English, J., Scofield, S.R., Bishop, G.J., and Harrison, K. (1992). Effective vectors for transformation, expression of heterologous genes, and assaying transposon excision in transgenic plants. Transgenic Res. 1, Kelly, A.J., Bonnlander, M.B., and Meeks-Wagner, D.R. (1995). NFL, the tobacco homolog of FLORICAULA and LEAFY, is transcriptionally expressed in both vegetative and floral meristems. Plant Cell 7, Koornneef, M., Alonso-Blanco, C., Peeters, A.J.M., and Soppe, W. (1998). Genetic control of flowering time in Arabidopsis. Annu. Rev. Plant Physiol. Plant Mol. Biol. 49, Levy, Y.Y., and Dean, C. (1998). The transition to flowering. Plant Cell 10, Mandel, M.A., and Yanofsky, M.F. (1995). A gene triggering flower development in Arabidopsis. Nature 377, Mandel, M.A., Gustafson-Brown, C., Savidge, B., and Yanofsky, M.F. (1992). Molecular characterisation of the Arabidopsis floral homeotic gene APETALA1. Nature 360, Martinez-Zapater, J., Coupland, G., Dean, C., and Koornneef, M. (1994). The transition to flowering in Arabidopsis. In Arabidopsis, E.M. Meyerowitz and C.R. Somerville, eds (Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press), pp McDaniel, C.N. (1996). Developmental physiology of floral initiation in Nicotiana tabacum L. J. Exp. Bot. 47, McDaniel, C.N., and Hsu, F.C. (1976). Position-dependent development of tobacco meristems. Nature 259, McDaniel, C.N., Sangrey, K.A., and Singer, S.R. (1989). Node counting in axillary buds of Nicotiana tabacum cv. Wisconsin 38, a day-neutral plant. Am. J. Bot. 76, Napoli, C.A., and Ruehle, J. (1996). New mutations affecting meristem growth and potential in Petunia hybrida Vilm. J. Hered. 87, Odell, J.T., Nagy, F., and Chua, N.-H. (1985). Identification of DNA sequences required for the activity of the cauliflower mosaic virus 35S promoter. Nature 313,