Molecular analysis of the LATERAL SUPPRESSOR gene in Arabidopsis reveals a conserved control mechanism for axillary meristem formation

Isolation and structure of the Arabidopsis LATERAL SUPPRESSOR gene

Initially, a PCR-based strategy was used to isolate a Lateral suppressor (Ls)-homologous gene from Arabidopsis thaliana (see Materials and Methods). Sequence analysis revealed that the Ls-homologous gene contains an open reading frame (ORF) encoding a protein with 50.5% sequence identity to the tomato Ls protein. The next most closely related protein encoded by the Arabidopsis genome shows only 34.8% identity. After completion of the Arabidopsis genome sequence, BLAST analysis resulted in the identification of two BAC clones, F20N2 (GenBank accession no. AC002328) and T5A14 (GenBank accession no. AC005223), which both contain the most homologous LATERAL SUPPRESSOR gene (At1g55580). Microsynteny studies between the tomato Ls region and the identified BAC clones demonstrated that the gene repertoire within this region is completely conserved, but two inversions distinguish the arrangements of genes in both species (Rossberg et al. 2001). These results demonstrate that the isolated DNA segment contains the Ls-orthologous gene from Arabidopsis (LAS).

The sequence of the LAS cDNA (GenBank accession no. AY196482) was deduced from PCR products amplified on cDNA that was prepared from shoot tip RNA. The 5′ end (position 1) and the 3′ end (position 1742) of the transcript have been determined in rapid amplification of cDNA ends (RACE) experiments. Comparison of the cDNA to its genomic counterpart revealed that the LAS gene contains an uninterrupted ORF starting with an ATG at position 96 and ending with a stop codon at position 1431, followed by an untranslated 3′ region of 309 bp. The ORF has a coding capacity for 445 amino acids.

The LAS protein is a member of the GRAS family of putative transcriptional regulators (Pysh et al. 1999). LAS, which was previously named SCL18, represents an independent branch of the GRAS protein family in Arabidopsis. The characteristic sequence motifs of this protein family, like the VHIID domain, the two leucine heptad repeats, and a so-called SAW domain (Pysh et al. 1999), were also identified in LAS. In contrast to GAI, RGA, and a few other members of the GRAS family, the LAS protein has a very short N terminus characterized by repetitions of the amino acids serine and threonine and does not contain a DELLA domain.

Identification of las mutants

las mutants were identified by screening a collection of Arabidopsis plants mutagenized by the maize transposable element Enhancer-1 (En-1; Baumann et al. 1998; Wisman et al. 1998). Using a PCR-based screening strategy, three independent En-1 insertions within the LAS ORF were identified (las-1, las-2, las-3; Fig. 1). Insertion points were determined by sequencing diagnostic PCR products.

To establish a causal relationship between the mutant phenotype and a particular En-1 insertion, several derivatives of the insertion alleles that have excised the En-1 element from the LAS locus were analyzed. For this purpose, progeny of outcrosses to the Columbia (Col) wild type were screened for loss of En-1 from the LAS locus using a PCR-based approach. From the insertion alleles las-1 and las-3, four independent derivatives were obtained. Two stable mutants, las-4 and las-5, as well as a wild-type revertant, were derived from las-3. An additional stable mutant, las-6, was isolated from the F1 progeny of las-1 (Fig. 1). Plants homozygous for the different mutant alleles exhibited a very similar phenotype (see below; Fig. 2), whereas the revertant was indistinguishable from the wild type. Sequence analysis revealed that las-4 is characterized by a deletion of 20 nucleotides together with the insertion of 1 nucleotide, whereas the wild-type revertant had the complete wild-type sequence. In the stable mutant las-5, an insertion of a T nucleotide was found at the site of the former En-1 insertion. The las-6 allele is characterized by an insertion of the tetranucleotide AATT. The sequence alterations observed in the stable mutant alleles las-4, las-5, and las-6 gave rise to frameshifts, which result in a premature stop of protein translation. Overall, these data demonstrate a causal relationship between the mutations at the LAS locus and the observed phenotypic alterations (Fig. 2).

las mutants are characterized by an almost complete suppression of axillary shoot formation during vegetative development

Six independent lines were established, in which the LAS ORF is disrupted by mutations at different positions. Because all these lines displayed a very similar phenotype, only plants homozygous for the las-4 deletion allele were used in a detailed characterization of the mutant phenotype (Fig. 2). In comparison with Columbia wild-type plants, las mutants showed an almost complete suppression of secondary bolt formation from the axils of rosette leaves (Fig. 2A–D,I). However, side-shoot formation from the axils of cauline leaves of the primary bolt was not inhibited. In some las-4 plants, also the topmost rosette leaves formed axillary shoots. Similar changes in branching patterns were also observed when the las-4 allele was introduced into the Landsberg erecta and Wassilewskija genetic backgrounds (data not shown). The branching phenotype of las mutants was more easily recognized when plants were grown first under short-day conditions to allow the formation of many rosette leaves, and then shifted to long days to induce flowering. Secondary branching was also affected: whereas in the wild type, tertiary shoots developed from the axils of all leaves of secondary shoots, branching from the first two or three leaves of secondary shoots was blocked in the las-4 mutant.

Furthermore, positioning of lateral shoots originating from the axils of cauline leaves was frequently affected in the las-4 mutant. Different from the wild type, in which lateral shoots are positioned in the axils of their subtending leaves, the las-4 mutant showed a tendency toward concaulescence, that is, the point of separation between lateral shoots and the main axis was often displaced acropetally (Fig. 2E,F).

In addition to its defect in side-shoot development, the tomato ls mutant is characterized by the absence of petals from the flower (Williams 1960; Schumacher et al. 1995). Different from tomato, the Arabidopsis las-4 mutant developed a complete whorl of petals (Fig. 2G,H). However, in the mutant, petal abscission seems to be delayed. Root development was analyzed on plantlets grown on vertically oriented agar plates. In contrast to shoot branching, the pattern of root branching did not reveal any significant difference between the las-4 mutant and the Columbia wild type (data not shown).

The Lateral suppressor gene function is conserved between tomato and Arabidopsis

Comparison of the phenotypes of lateral suppressor mutants of tomato and Arabidopsis revealed both similarities and differences. Whereas both mutants showed a severe reduction in the number of axillary shoots, only the tomato mutant was characterized by a failure to develop petals (Williams 1960). In addition, a reduction in male and female fertility was observed in the tomato ls mutant (Groot et al. 1994), but not in the Arabidopsis las-4 mutant. To test for functional complementation between these two distantly related plant species, a 6.3-kb genomic DNA fragment, harboring the Arabidopsis LAS ORF, as well as ∼2.9 kb of 5′ sequence and 2.1 kb of 3′ sequence, was introduced into the tomato lateral suppressor (ls¹/ls¹) mutant. Four independent transgenic lines harboring at least one complete copy of the LAS gene were established. Two transgenic lines developed side shoots in almost every leaf axil and a whorl of petals on all flowers (Fig. 3C,D). The two additional lines developed side shoots in only a fraction of their leaf axils and also showed an incomplete restoration of the wild-type flower phenotype (Fig. 3E,F). Inheritance of the complementation phenotype was analyzed in transgenic line 258 harboring a single-copy T-DNA insertion. Among 19 plants of the self-pollinated progeny of 258, 14 plants showed complementation and 5 plants had the ls phenotype. By Southern analysis, the T-DNA was detected only in those plants showing complementation. This result is consistent with the assumption that a single-copy T-DNA insertion, segregating in a Mendelian fashion, rescues the ls phenotype.

LAS transcripts accumulate in the axils of all primordia derived from the SAM

LAS mRNA distribution in different plant tissues was first monitored by reverse transcriptase PCR (RT–PCR) analysis. The LAS transcript was detected in shoot tips, flowers buds, roots, and open flowers, but not in internodes and leaves (Fig. 4). To analyze the distribution of LAS mRNA at the cellular level, RNA in situ hybridization experiments on tissue sections were performed. Columbia and las-4 Arabidopsis plants were grown under short-day conditions and fixed 28 d after sowing. During the vegetative phase of development, LAS transcripts accumulated in the axils of leaf primordia and leaves from P1 to P20/22 (Fig. 5A,E–H). In the axil of P1, signals were observed when the primordium was barely recognizable as a bulge protruding from the SAM. The LAS expression domain was initially about 3–5 cell layers deep, including the L1–L3 layers of the SAM, and extended 1 or 2 cell layers in the adaxial–abaxial dimension. Transverse sections demonstrated that LAS mRNA accumulated in a band-shaped domain at the adaxial side of the leaf primordium, marking the inner boundary of the primordium (Fig. 5E–H). From P5 to P22, LAS expression was restricted to a 1–2-cell-layer-deep domain located within the cleft between the leaf primordium and the inner part of the shoot apex, slightly shifted toward the base of the primordium. In the axils of older leaf primordia, LAS expression faded out until it was no longer detectable in the axils of primordia older than P20/P22. A similar distribution of LAS transcript was observed in vegetative apices of the las-4 mutant, although the intensity of the signal was reduced in comparison to wild-type plants (data not shown).

Because histological studies revealed differences in the timing of axillary meristem formation between Arabidopsis ecotypes (Long and Barton 2000; Stirnberg et al. 2002), LAS expression during the vegetative phase was also analyzed in Landsberg erecta. Different from Columbia, in Landsberg erecta plants a down-regulation of the LAS signal was observed in a small oval area in the center of the LAS expression domain (Fig. 5K,L, arrows) from P6/P7 on, which coincides with the position of the presumptive axillary meristem. Sections of flowering shoots were obtained from plants grown under short-day conditions for 28 d and then induced to flower by two, four, or six successive long days. After the shift to long days, LAS mRNA was found in the axils of all leaf primordia originating from the elongating stem (Fig. 5B,C). Some of the LAS expression domains in cauline leaf axils showed a similar down-regulation in their center (Fig. 5B, arrow) to those observed during Landsberg erecta vegetative development. At the base of outgrowing axillary shoots, residual amounts of LAS transcripts have been detected (Fig. 5C,D, arrows). After the transition to flower formation, LAS transcripts were detected at the adaxial base of flower primordia (Fig. 5D, arrowheads). This LAS expression domain was first detectable at the adaxial border of stage 1 flower primordia with cellular dimensions comparable to those domains observed in the axils of leaf primordia. During flower development, LAS expression was observed in the axils of sepal primordia of stage 3–5 flowers (Fig. 5D). This signal faded out at later stages of flower development.

In contrast to the results of the RT–PCR analyses, LAS transcripts could not be detected in the root by RNA in situ hybridization. These experiments were controlled by using a SCARECROW (SCR; Di Laurenzio et al. 1996) antisense probe on parallel sections, which successfully detected the SCR transcript (data not shown).

las mutants fail to initiate axillary meristems during vegetative development

To test whether the inability to form lateral shoots from the axils of rosette leaves was due to a defect in axillary meristem initiation or to an inhibition of lateral bud outgrowth, axillary shoot formation was analyzed in Columbia wild-type plants and las-4 mutants after removal of the primary bolt. Plants were first grown under short-day conditions for 28 d and then shifted to long days to induce flowering. Primary bolts were removed when the plants reached a height of 10–15 cm. After decapitation, the wild-type plants developed between 5 and 13 axillary shoots, whereas none, one, or two axillary shoots were recorded for the las-4 mutant (Fig. 2I). If the las-4 mutant developed secondary bolts, these always originated from the axils of the topmost rosette leaves. These data suggested that the las-4 mutant phenotype was not a consequence of extreme apical dominance, but more likely the result of a defect in axillary meristem initiation during vegetative development.

Long and Barton (2000) have demonstrated that STM expression can be used as an early marker for axillary meristem development. In this study, STM expression was monitored in the wild-type and the las-4 mutant plants to test for initiation of axillary meristems. Transverse sections were prepared from vegetative plants grown under short-day conditions for 28 d and hybridized to an antisense STM probe. These experiments revealed STM transcript accumulation in the SAM and interprimordial regions of both wild-type and las-4 plants (Fig. 6A–D, arrowheads). As reported by other groups (e.g., Long and Barton 2000), leaf primordia and incipient leaf primordia (from P0) were found to be free of STM signal. In the axils of older leaf primordia of the wild type (from P16), STM expression was found to be focused to a group of small, densely cytoplasmic cells localized next to the adaxial center of the primordium border (Fig. 6A,C, arrows, E). These cell groups developed into new meristems, which could first be recognized at the morphological level in axils of P21. However, older leaf axils of the las-4 mutant did not show this focused STM expression (Fig. 6B,D, arrows, F), suggesting that the focusing of STM expression indicative of the onset of meristem organization is dependent on LAS function. Comparison of the patterns of LAS and STM transcript accumulation indicated that the LAS expression domain in the axils of leaf primordia was contained within the STM-positive region (Figs. 5, 6). The developmental timing of STM focusing to the small cell group next to the center of the adaxial primordium boundary seems to be correlated with a down-regulation of the LAS signal, as observed in the Landsberg erecta ecotype (Fig. 5L). Taken together, the results of the decapitation experiments and the analyses of STM expression demonstrate that in the las-4 mutant, axillary meristem initiation is blocked during vegetative development.

LATERAL SUPPRESSOR acts upstream of REVOLUTA in axillary meristem development

In the Arabidopsis revoluta (rev) mutant, shoot branches in the axils of rosette and cauline leaves frequently fail to develop (Talbert et al. 1995). Double-mutant analysis and expression studies suggested that REV is an activator of STM (Otsuga et al. 2001). To unravel the relationship between REV and LAS, RNA in situ hybridization experiments were performed on tissue sections from wild-type (Columbia) and las-4 plants using REV sense and antisense probes. In Columbia, the REV transcript was detected in vascular bundles of leaves, in the vascular traces of leaf primordia, and in an inverse cup-shaped domain in the SAM itself (Fig. 7A,C). From P17 on, REV mRNA was also found in a cell group located next to the center of the adaxial base of leaf primordia (Fig. 7C,E,G, arrows). This expression domain marks the presumptive axillary meristem and shows an extended overlap with the STM expression domain. These findings support the results of expression studies reported by Otsuga et al. (2001). Comparison of the patterns of REV transcript accumulation in wild-type and las-4 plants revealed no differences in the SAM and in the vascular elements of leaf primordia and leaves (Fig. 7C,D). In contrast, the REV expression domain observed in the cell group located at the central adaxial base of older leaf primordia of wild-type plants was absent in las-4 (Fig. 7E–H, arrows).

To test whether LAS expression is dependent on REV function, we have also analyzed LAS transcript accumulation in the rev-5 mutant. Comparison of vegetatively growing wild-type (Columbia) and rev-5 plants did not reveal any difference in the LAS expression pattern (data not shown). Taken together, the results of these expression studies demonstrate that during vegetative development, REV expression in developing axillary meristems is dependent on LAS function.

las is epistatic to axr1 with respect to lateral shoot development during the vegetative phase

The recessive Arabidopsis axr1 mutant shows a pleiotropic phenotype. Beside defects in root, hypocotyl, stamen, and stem elongation, one prominent aspect of its phenotype is an increased branching at maturity (Estelle and Somerville 1987; Lincoln et al. 1990; Stirnberg et al. 1999). All axr1 tissues examined showed reduced auxin sensitivity, suggesting that the AXR1 protein is required for auxin signaling (Abel et al. 1995; Timpte et al. 1995). Phenotypic analyses of loss-of-function mutants suggested that axr1 enhances shoot branching during both the vegetative and the reproductive phases of development (Stirnberg et al. 1999).

To test whether the enhanced shoot-branching potential conferred by the axr1 mutation was able to override the defect caused by las, we have constructed an axr1-12 las-4 double mutant. Homozygous axr1-12 las-4 plants showed the characteristic features of axr1 loss-of-function mutants: for example, reduction in plant height and alterations in leaf morphology (Fig. 8B). The potential for shoot branching was analyzed in a comparison of axr1-12 and axr1-12 las-4 plants. To enhance the outgrowth of lateral buds, primary bolts were removed when the plants reached a height of 10–15 cm. The experiment revealed that, with respect to shoot branching, the axr1-12 las-4 double mutant was similar to las-4: axillary shoot development was blocked in the older rosette leaf axils; however, lateral shoots developed from the axils of cauline leaves and at the topmost rosette leaves (Fig. 8A–D,G). This finding demonstrates that, with respect to lateral shoot development during the vegetative phase, las-4 is epistatic to axr1-12. Different from las-4, the zone of axillary shoot formation extended basipetally into the rosette. axr1-12 las-4 double mutants developed between zero and five shoots from the axils of the topmost rosette leaves (Fig. 8G), whereas only zero to two side shoots were observed in the rosette leaf axils of las-4 mutants (Fig. 2I). The above experiments were controlled by examining las-4 and wild-type populations grown in parallel under the same conditions.

The enhanced shoot-branching potential of axr1 plants leads also to an increase in the formation of accessory side shoots from the axils of cauline leaves (Stirnberg et al. 1999). Comparing axr1-12 with axr1-12 las-4 plants, we observed a strong reduction in the development of accessory side shoots in the double mutant (Fig. 8E,F,H). This finding indicates that, despite the formation of lateral shoots observed in the cauline leaf axils of the las-4 mutant, the LAS gene has an influence on the shoot-branching potential during reproductive development. This conclusion is also supported by the pattern of LAS transcript accumulation.

LAS function is required for axillary shoot formation during vegetative development

In Arabidopsis, shoot branching is differently controlled during the vegetative and reproductive phases of development. Under prolonged short-day conditions, initiation of axillary meristems starts in the axils of the oldest leaves and then progresses toward younger leaf axils. Subsequently, axillary shoot development progresses in parallel with development of the subtending leaf, resulting in an acropetal gradient of lateral shoot formation (Hempel and Feldman 1994; Grbic and Bleecker 1996; Stirnberg et al. 1999, 2002). After the transition to flowering, axillary meristems are initiated evenly (Stirnberg et al. 1999, 2002) or in a basipetal sequence (Hempel and Feldman 1994; Grbic and Bleecker 2000) along the flowering stem. During this phase, bud outgrowth follows a basipetal gradient starting with the youngest leaf axils and progressing toward older leaf axils (Hempel and Feldman 1994; Grbic and Bleecker 1996; Stirnberg et al. 1999, 2002). Under ambient light conditions, both gradients may overlap with each other. Characterization of las mutants has shown that axillary meristems do not form during vegetative development. After the floral transition, however, axillary shoots develop from the axils of cauline leaves and sometimes from the topmost rosette leaves. This phenotype suggests that in Arabidopsis, the LAS gene is an essential component of the genetic mechanism governing the acropetal gradient of axillary meristem formation. On the other hand LAS function is not required for the initiation of primary axillary meristems during the reproductive phase of development. In this period, either its function is not needed or it is provided by a different gene, maybe a different member of the GRAS gene family.

LAS function is, however, not completely dispensable in the axils of cauline leaves. When comparing the number of accessory shoots in cauline leaf axils of axr1-12 and axr1-12 las-4 plants, we observed a strong reduction in the double mutant. In addition, in situ hybridization experiments revealed that LAS mRNA accumulates also in the axils of cauline leaves. After outgrowth of axillary buds, residual amounts of LAS transcripts have been detected at the base of axillary shoots. These findings suggest that LAS has an influence on the potential of cauline leaf axils to form lateral shoots, which is not reflected in the pattern of primary side-shoot development.

LAS is a marker for axil identity

LAS transcripts accumulate in the axils of all primordia derived from the SAM. During vegetative development, LAS transcripts can be detected in the axils of leaf primordia from P1 to P21 in a narrow domain extending 3–5 cell layers in the apical–basal and 1–2 cell layers in the adaxial–abaxial dimension. After floral transition, LAS mRNA was found in the axils of flower primordia and sepals, where the expression domains show similar dimensions. The pattern of LAS transcript accumulation in the axils of leaf primordia is similar to those of the genes LATERAL ORGAN BOUNDARIES (LOB), CUP-SHAPED COTYLEDON1 (CUC1), and CUP-SHAPED COTYLEDON2 (CUC2). Transcripts of all these genes are detected at the boundary of two tissues, the SAM and the new leaf primordium (Aida et al. 1999; Takada et al. 2001; Shuai et al. 2002). It is tempting to speculate that these expression domains are established at this boundary as a result of the influence of transcriptional regulators from both tissue types. The finding that transcripts of different genes accumulate in the axils of leaf primordia in very similar patterns suggests that the axillary tissue has a special identity. Based on the pattern of LAS transcript accumulation, one can assume that axil identity is established together with, or immediately after, the formation of the corresponding leaf primordium. Future experiments have to show whether or not LAS, LOB, CUC1, and CUC2 are related to each other in a hierarchical order of gene interactions.

In situ hybridization experiments with transverse sections demonstrated that LAS transcripts accumulate along the whole adaxial border of leaf primordia. However, axillary meristems are only initiated in the middle of this expression domain close to the center of the adaxial leaf base. This observation suggests that additional positional information is required to determine the specific position of lateral meristem initiation. The close proximity between the developing vascular elements and developing axillary meristems, combined with the observation that ectopic meristems form preferentially over underlying vascular bundles (Chuck et al. 1996; Takada et al. 2001), suggests that vascular bundles contribute to the induction of meristems.

LAS function is required to establish axillary meristems at a distance to the SAM

In Arabidopsis, the first lateral organ that is formed after the shift from short-day to long-day conditions is a flower primordium; cauline leaves develop from the youngest leaf primordia initiated before the floral transition (Hempel and Feldman 1994). With respect to the las mutant phenotype, this observation implies that the youngest leaf axils, which will develop axillary shoots in the absence of LAS function, are either predisposed for axillary meristem formation, or acquire this ability as a consequence of the floral transition. The fact that las-4 plants at different stages of shoot development respond in the same manner when shifted from short-day to long-day conditions indicates that lateral meristem induction is not affected by the number of leaves initiated before the transition. The zone of axillary meristem formation in las mutants may be defined by the range of a signal from the apical meristem or a region of competence of axillary cells extending several nodes from the SAM. Formally, the signal could either be an activator released from the inflorescence meristem or the loss of a repressor of lateral meristem initiation produced by the vegetative SAM. The latter would be very much reminiscent of the phenomenon of correlative inhibition of axillary bud outgrowth known to be exerted by a repressing signal from the shoot apex. In this context, it is interesting to note that in las-4 axr1-12 plants, the zone of side-shoot development is extended basipetally into the rosette. In this double mutant, the zone of axillary meristem formation may be larger because of a modified signal from the SAM or of an alteration in competence of axillary cells. The above observation also indicates that the AXR1 gene does not only repress lateral bud outgrowth but may also negatively control the initiation of axillary meristems. In summary, the youngest leaf axils of vegetatively growing Arabidopsis plants have always the potential to form axillary shoots independent of the LAS function. This potential is realized when the vegetative SAM undergoes the transition into an inflorescence meristem.

It is still a matter of debate, whether axillary meristems are detached from the SAM of the main shoot or develop de novo from fully or partially differentiated cells (Steeves and Sussex 1989; Long and Barton 2000). Because this study demonstrates that axillary meristems develop from LAS-positive cell groups, LAS transcript accumulation in leaf axils can be considered as a marker for meristem formation competence. Comparisons of the patterns of LAS and STM transcript accumulation in leaf axils revealed that LAS-positive cells are also positive for STM mRNA, until the time point when the STM signal is focused to the center of the leaf axil (from P16 in Columbia) and REV expression is turned on. The coexpression of LAS and STM in the axils of P1 to P15 primordia suggests that the identity of axillary meristem progenitor cells is determined together with or shortly after the initiation of the corresponding leaf primordium and that the meristematic character of these cells is retained, whereas neighboring cells lose their meristematic identity. These findings support the “detached meristem” concept. However, meristematic identity by itself, as revealed by STM mRNA accumulation, is not always sufficient to allow axillary meristem initiation, because in the las-4 mutant many STM-positive axils do not develop axillary meristems. During the vegetative phase of development, an additional function, which requires the LAS protein, is needed to initiate axillary meristems. The view that an additional function is required to trigger the formation of axillary meristems at a distance to the SAM supports the argument in favor of a de novo formation of axillary meristems. Taken together, these results suggest that the process of axillary meristem formation during Arabidopsis vegetative development shows characteristics of both, the “detached meristem” and the “de novo formation” concepts.

LATERAL SUPPRESSOR function is conserved during evolution

Microsynteny analysis between the chromosomal segments harboring the LATERAL SUPPRESSOR genes in tomato and Arabidopsis has demonstrated that both genes are orthologous (Rossberg et al. 2001). However, the LATERAL SUPPRESSOR proteins of both species show only 50.5% identity. The defects observed in the corresponding loss-of-function mutants display both similarities and differences. In both cases, side-shoot development is abolished during the vegetative phase of development and not affected during the reproductive phase. Tomato ls mutants develop lateral shoots only in the two leaf axils preceding an inflorescence, whereas the number of leaf axils with lateral shoots in Arabidopsis las-4 plants is ∼3–13, depending on growing conditions. This difference in the length of the zone, where axillary shoots are formed, may be explained by the more elongated stature of the tomato stem during vegetative development or by the slower leaf initiation rate in tomato. In addition, tomato ls mutants fail to develop petals on the flower and are characterized by a reduced male and female fertility. Such defects have not been observed in Arabidopsis plants homozygous for LAS loss-of-function alleles. In this study, we have demonstrated that the Arabidopsis LAS gene can fully complement the tomato lateral suppressor mutant. This result demonstrates that both genes are functionally equivalent and that an essential control mechanism in the process of side-shoot development is conserved between two distantly related species, tomato and Arabidopsis. Differences in the mutant phenotypes of both species may be explained by redundant gene activities during flower development in Arabidopsis leading to a masking of a possible mutant phenotype.

Plant materials

Tomato seed material of Lycopersicon esculentum L. cv. Antimold B-ls¹ (LA0329) was obtained from the Tomato Genetics Resource Center, Davis, CA. Seeds of Arabidopsis thaliana (Heynh.), ecotypes Columbia (Col-O) and Landsberg erecta, were obtained from R. Schmitt, Max-Delbrück-Laboratorium, Köln, Germany. Tomato plants were grown under standard glasshouse conditions with additional artificial light (16-h photoperiod) during the winter period. For cultivation under short-day conditions, Arabidopsis plants were grown in a controlled environment room with 8 h of light (16 h of dark), 20°C–25°C day temperature, and 10°C–15°C night temperature. Short-day-grown plants were induced to flower after 24 or 26 d by successive 16-h photoperiods. Cultivation under long-day conditions was done in the greenhouse with additional artificial light when needed.

Isolation of las mutants

LAS∷En insertion lines were identified in an En-1-mutagenized population of Arabidopsis plants (Wisman et al. 1998) following the protocol of Baumann et al. (1998). PCR screens were performed using the gene-specific primers LSEN2 (5′-GTAC GAATCAACCGTCAGCAACAAG-3′) and LSEN3 (5′-GCCA CAACATCCAAAATCTCCTTACC-3′) in combination with the En-1-specific primer En205 (Baumann et al. 1998). Candidate lines were grown first under short-day conditions (e.g., 26 d), then shifted to long days and inspected for alterations in their shoot-branching pattern. To obtain stable mutant alleles, the En-1 insertion lines were outcrossed to the Columbia (Col) wild type, and progeny plants were screened by PCR for loss of the En-1 element. Candidates that had lost the En-1 element were self-pollinated, and their progenies were analyzed for segregation of the mutant phenotype and sequence alterations within the LAS gene. Mutant alleles were back-crossed six times into the Columbia genetic background, without noticeable changes in phenotypic expression. For phenotypic analyses we used las-4 lines selected after at least three back-crosses into the Columbia background.

DNA isolation

Plant DNA preparation and Southern blot analysis have been described (Brandstädter et al. 1993). Standard techniques were carried out according to Sambrook and Russell (2001).

Isolation of the Arabidopsis LAS gene

Alignment of the tomato Ls amino acid sequence with the related A. thaliana proteins GAI (Peng et al. 1997), RGA (Silverstone et al. 1998), SCARECROW (Di Laurenzio et al. 1996), and SCL13 (Pysh et al. 1999) revealed regions of high and low sequence conservation. Degenerated primers deduced from different regions were tested in PCR experiments under reduced stringency conditions on Arabidopsis genomic DNA. A fragment of ∼700 bp could be amplified with the primer combination CD61-38 (5′-CARTGGCCNCCNYTNATGCA-3′) and CD61-41 (5′-TGRTTYTGCCANCCNARRAA-3′). The PCR reaction was performed using the following regime: 95°C for 30 sec, 50°C for 1 min, 72°C for 1 min (35 times). This PCR product was used to isolate a clone from a genomic Arabidopsis cosmid library by colony hybridization.

RNA isolation, reverse transcription (RT)-PCR Analysis, and rapid amplification of cDNA ends (RACE)

Total RNA was isolated using the RNeasy plant mini kit (QIAGEN) following the manufacturer's instructions. For RT–PCR analysis, 1 μg of total RNA was reverse transcribed using the Superscript II polymerase (Life Technologies) according to the manufacturer's instructions. The product of the first-strand cDNA synthesis reaction was amplified by PCR using the LAS-specific primers AtLs2949F (5′-GTCTCCAATTCCAGTTTCA CAC-3′) and AtLs3532R (5′-TTGCCAGCCAAGAAACAAAG-3′). Amplification of actin cDNA by using the primers ActinATfor (5′-TGGTGTCATGGTTGGGATG-3′) and ActinATrev (5′-CACCACTGAGCACAATGTTAC-3′) was performed as a control to ensure that equal amounts of cDNA were added to each PCR reaction.

RACE experiments were performed using the RACE system of Life Technologies according to the manufacturer's instructions. For 5′ RACE, oligo(dT)-primed cDNA was tailed and amplified using the general amplification primers and the LAS-specific primers AtLs2717R (5′-CCGTTAAATGACCGAAC CGA-3′), AtLs2618R (5′-ACTGTTCTTTGCATACACTGCT-3′), and AtLs2535R (5′-GCCGTCTGATCTTGTTGCT-3′).

For 3′ RACE, first-strand cDNA was amplified using the general amplification primer and the LAS-specific primers AtLs2858F (5′-CCTCCATCTCTCCGCATAAC-3′), AtLs2949F (5′-GTCTCCAATTCCAGTTTCACAC-3′), and AtLs3339F (5′-CGGAAGAGACGGAGAGAAAG-3′). Amplified fragments were cloned into the pGEM-T (Promega) plasmid vector and sequenced.

Tomato transformation

For complementation experiments, a 6.3-kb NheI–BanI fragment of the genomic cosmid clone harboring the Arabidopsis LAS gene was cloned into the binary vector pGPTV-Kan (C28.1; Becker et al. 1992) and transferred into the Agrobacterium tumefaciens strain GV3101 (Koncz and Schell 1986). Transformation of tomato leaf explants was done as described previously (Knapp et al. 1994).

RNA in situ hybridizations

Sample preparations and in situ hybridizations of 8-μm sections were done as described by Coen et al. (1990) with slight modifications. Tween 20 (0.03%) was added to the fixative, and dewatering of the fixed material was done without NaCl. LAS probes contained the nucleotides 2–1348, REV probes the nucleotides 682–2611, in both cases relative to the ATG. STM probes included base pairs 78–1122 of the ORF. The sequences were cloned into pGEM vectors in sense and antisense orientation, respectively, relative to the T7 promoter. Linearized plasmids were used as templates for probe synthesis with T7 RNA polymerase. The time for probe hydrolysis was calculated for an average length of 150 nucleotides as described (Jackson 1992). After the color reaction, slides were mounted in 30% glycerol and photographed using differential interference contrast microscopy.

DNA sequencing and analysis

DNA sequences were determined by the DNAcore facility of the Max-Planck-Institut für Züchtungsforschung (Cologne, Germany) on Applied Biosystems (Weiterstadt, Germany) Abi Prism 377 and 3700 sequencers using BigDye-terminator chemistry. Premixed reagents were from Applied Biosystems. Oligonucleotides were purchased from Life Technologies or Metabion. Analysis of sequences was performed using the Wisconsin Package (version 10.0-UNIX; Genetics Computer Group). Database searches were performed using BLAST programs (Altschul et al. 1990) against NCBI sequence databases.