The increasing number of mutations or transgenes that affect auxin content, transport, or sensitivity, especially in Arabidopsis, provides a different approach to investigate the role of auxin in apical dominance, and in particular which stages of lateral shoot development it regulates in vivo. Plants expressing the IAA biosynthetic genes from Agrobacterium tumefaciens have high endogenous IAA levels and increased apical dominance (Klee et al., 1987; Romano et al., 1993, 1995). The iaaL gene from Pseudomonas savastanoi, encoding an enzyme that conjugates IAA to Lys, has been transformed into tobacco, resulting in reduced levels of free IAA and reduced apical dominance (Romano et al., 1991).

Several mutants with altered auxin sensitivity have been produced in Arabidopsis. One such locus, AXR1, is defined by an allelic series of recessive mutations. The phenotype of axr1 mutant plants is pleiotropic, with defects in root, hypocotyl, stamen, and stem elongation, vascular development, lateral root formation, and root gravitropism. Leaf morphology is altered and shoot branching is increased at maturity (Estelle and Somerville, 1987; Lincoln et al., 1990). All axr1 tissues examined showed reduced auxin sensitivity in a variety of assays, including rapid gene induction by auxin (Abel et al., 1995; Timpte et al., 1995). This suggests that the AXR1 protein is required for auxin signaling. The AXR1 gene encodes a protein related to the amino-terminal half of ubiquitin-activating enzyme; sequencing of severe mutant alleles such as axr1-12 indicates complete loss of protein function (Leyser et al., 1993).

Although these mutant and transgenic lines suggest that branching is regulated by auxin in vivo, they were only investigated at a cursory level. Altered branching may be an indirect consequence of other auxin-regulated phenotypes. axr1-12 plants produce more branches than the wild type, and are less fertile (Lincoln et al., 1990). There is evidence that developing fruits limit lateral branching (Tamas et al., 1979; Hensel et al., 1994). Therefore, the increased branching in mature axr1 plants may be a consequence of their infertility.

The in vivo role of auxin is further called into question by examining the pattern of lateral shoot development in Arabidopsis. If apically derived auxin acts as an inhibitor of axillary branching, one might expect the lateral buds closest to the shoot apex to be the most repressed, giving a basal-apical (acropetal) sequence of bud growth. Arabidopsis shows this pattern during prolonged vegetative growth, e.g. in late-flowering ecotypes (Grbić and Bleecker, 1996). However, lateral inflorescences develop in an apical-basal (basipetal) sequence after floral transition (Alvarez et al., 1992; Hempel and Feldman, 1994).

To further assess the role of auxin in the regulation of Arabidopsis shoot branching, we compared axillary shoot development in the wild type and axr1-12 to determine the earliest stage when differences can be detected. In addition, we compared the auxin sensitivity of wild-type and axr1-12 lateral inflorescence outgrowth using nodes excised from the primary inflorescence. The axr1-12 mutation does not affect early stages of axillary shoot development; however, it promotes the subsequent growth of axillary shoots initiated by the acropetal and basipetal pattern. It also renders the outgrowth of isolated lateral inflorescences resistant to auxin inhibition. These results suggest a role for auxin in the stages of lateral shoot development following axillary meristem formation.

For laterals in category (c), inflorescence length was measured and the number of vegetative nodes determined. The presence of visible accessory inflorescences, which may arise between a lateral inflorescence and the subtending leaf in cauline nodes (Talbert et al., 1995), was noted. In total, 40 wild-type and 33 axr1-12 plants were analyzed. Wild-type plants bolted between d 28 and 37 after the end of cold treatment, with a mean ± se of 30.0 ± 0.4 d. axr1-12 plants bolted between d 29 and 35, on average slightly later than wild type (mean 32.0 ± 0.4 d). Wild-type plants produced between 13 and 24 leaves (rosette and cauline) before floral transition, with a mean ± se of 17.0 ± 0.4. With axr1-12, leaf numbers ranged from 16 to 27, with a mean ± se of 23.0 ± 0.5, significantly higher than wild type. Both mean rosette and mean cauline leaf numbers were higher for the mutant. Wild-type plants had on average 3.3 ± 0.2 cauline and 13.7 ± 0.3 rosette leaves. Mutant plants had on average 6.6 ± 0.2 cauline and 16.0 ± 0.4 rosette leaves.

In both the wild type and axr1-12, all of the cauline leaf axils on the primary inflorescence produced a lateral inflorescence. Some of these were clearly elongating on d 3 after the primary inflorescence started bolting and all were elongating by d 6. Due to the higher total cauline leaf number described above, axr1-12 plants had on average twice as many cauline inflorescences as the wild type.

The time course of first-order lateral development in the rosettes of both genotypes is shown in Figure 1A. As the wild type and axr1-12 differed in the number of leaves (i.e. nodes) in the rosette, the three categories of laterals scored were expressed as proportions of the total rosette node number for each individual. In the wild type, the mean proportion of nodes with no visible bud varied between 19% and 31% without a clear trend between d 3 and 18 after bolting. The same was true for axr1-12, except that the percentages were lower (between 14% and 20%). The percentage of wild-type nodes with visible buds decreased from 81% to 40% between d 3 and 18, accompanied by a rise in the percentage of nodes with elongating inflorescences from 0% to 31%. In axr1-12, the proportion of nodes with visible buds decreased from 81% to 33% and the proportion of elongating inflorescences increased from 0% to 54%. The proportion of elongating inflorescences in the rosette was higher in axr1-12 than in the wild type from d 12 after bolting.

Figure 1 Morphometric analysis of lateral shoot development of wild-type and axr1-12 plants grown in long photoperiods. A, Time course of lateral shoot development in rosette leaf axils. For each genotype, five to eight plants were sampled every 3 d between d (more ...)

First-order lateral development in the rosettes of both genotypes displayed a basipetal gradient. At the time of bolting, a number of leaves at the base of the rosette had no visible buds in their axils. More apical leaves carried visible buds. During the 18 d after bolting, an increasing number of leaves in the most apical part of the rosette carried an elongating lateral inflorescence.

Figure 1B shows one representative plant for each genotype, dissected 6 d after bolting. Their cotyledons and leaves with attached axillary shoots were laid out in the sequence in which they had been produced by the primary shoot apex. The class “visible buds” in Figure 1A included very small buds with leaves just beginning to expand (which are not visible at the magnification of Fig. 1B), up to buds with several expanded leaves and visible flower buds. Although the proportions of “visible buds” did not differ between the wild type and axr1-12 during the 18 d following bolting (Fig. 1A), genotypes differed in the extent of bud growth as early as 6 d after bolting (Fig. 1B). The zone of nodes carrying buds showing significant leaf expansion, clearly seen at the magnification of Figure 1B, was extended to more basal node positions in axr1-12. Overall, axillary leaves had expanded further in the mutant than in the wild type, which was most obvious in the oldest one to two leaves of each bud, but also in the younger leaves in the center of the buds.

To compare inflorescence outgrowth in the wild type and axr1-12, the average first-order lateral inflorescence lengths were calculated for consecutive node positions along the primary shoot axis, starting with the uppermost cauline leaf node and proceeding basipetally. The distribution of inflorescence lengths is shown in Figure 1C for d 18 after bolting. Lateral inflorescence lengths for axils without a visible bud or with a vegetative bud were scored as zero. As the mean number of cauline leaves was three for wild-type and six for axr1-12 plants, plots for the wild type were started three node positions basal from those of axr1-12. Thus, node position 0 represents on average the uppermost rosette leaf for both genotypes. The wild-type inflorescence length distribution showed a characteristic pattern. Branches at the three most apical nodes (on average corresponding to the cauline nodes on the primary inflorescence) were of similar length. Mean inflorescence lengths then decreased progressively through up to four more basal node positions. No elongating inflorescences were found further toward the base. The inflorescence lengths of axr1-12 plants followed a comparable pattern, but the number of apical inflorescences of similar length was increased to six (corresponding to the mean number of cauline inflorescences in axr1-12) and the zone of decreasing length comprised up to six nodes. The apical-basal gradient of inflorescence length was less steep in the mutant. In apical nodes, the mean axr1-12 inflorescence length was clearly below that of wild type, while in the more basal nodes, the mean inflorescence length was slightly higher than that of the wild type. These inflorescence length distributions of the wild type and axr1-12 were established as early as 6 d after bolting and were maintained while the primary and lateral inflorescences elongated (data not shown).

Lateral inflorescence architecture displayed characteristic differences that were correlated with position along the shoot axis. Apical axillary meristems produced fewer leaves before floral transition than basal meristems. In wild-type plants leaf numbers ranged between two and four for the uppermost cauline inflorescence, and four to five for the lowermost. In axr1-12, the uppermost cauline inflorescences had two or three leaves and the lowermost had between six and 10 leaves. Leaf numbers increased further toward the lowermost rosette inflorescence, which had leaf numbers between nine and 11 for wild type and between 10 and 14 for axr1-12.

Between a lateral shoot and its subtending leaf, an accessory axillary meristem may form and develop into an accessory bud or inflorescence in Arabidopsis (Talbert et al., 1995). Accessory shoots were first seen in the axils of cauline leaves 6 d after bolting, occurring in 6.7% of the 20 wild-type cauline nodes and in 64% of the 39 axr1-12 cauline nodes examined. The percentages increased in both genotypes until d 18 after bolting, but remained clearly lower (27% of 15 nodes) in the the wild type than in axr1-12 (70% of 30 nodes). Thus, differences in accessory development in the two genotypes arose as early as 6 d after bolting. Accessory buds were not detected in rosette leaf axils of either wild type or axr1-12.

Figure 2 Lateral shoot development at consecutive nodes along the primary shoot axis in short photoperiods 49 d after the end of cold treatment. For one wild-type (upper row) and one axr1-12 plant (lower row), the oldest 27 leaves (without the cotyledons that (more ...)

Plants grown in long days were fixed 15 and 20 d after the end of cold treatment, and longitudinal sections of both genotypes were compared. For both genotypes, we found vegetative plants and plants that had recently undergone floral transition among the plants fixed on d 15. Virtually all of the plants fixed on d 20 had undergone floral transition. During vegetative growth, the shoot apical meristem is relatively small and is situated directly on the short primary shoot axis (Vaughan, 1954; Miksche and Brown, 1965; Hempel and Feldman, 1994). For both genotypes, axillary meristems were not detected in the axils of either young or old leaves in vegetative plants (Fig. 3, A and E).

Figure 3 Longitudinal sections of wild-type (A–D) and axr1-12 (E–H) shoots grown in long photoperiods stained with toluidine blue. Plants in A and C and E to G were fixed on d 15, and plants in B, D, and H were fixed on d 20 after the end of cold (more ...)

The early stages of reproductive growth are characterized by enlargement of the shoot apical meristem and by cell divisions in the rib meristem at its base, which is the first step in the formation of the primary inflorescence axis (Vaughan, 1954; Miksche and Brown, 1965; Hempel and Feldman, 1994). In both wild-type and axr1-12 plants at this stage, clusters of very small, strongly staining meristematic cells were observed at the base of young leaf primordia (Fig. 3, B and F), as well as at the base of the petiole of older leaves (Fig. 3, C and G). In plants in which the first flower primordia had been formed by the primary shoot apical meristem, the axillary meristematic zone had increased in size in both genotypes, but axillary leaf formation had not yet started (Fig. 3, D and H). The early stages of axillary shoot formation in axr1-12 (Fig. 3, E–H) were indistinguishable from those of the wild type (Fig. 3, A–D). Accessory axillary meristems at cauline nodes were observed in older plants of both genotypes, fixed at least 25 d after the end of cold treatment, but the frequencies were insufficient for a detailed comparison.

Plants grown in short photoperiods were fixed 37 d after the end of cold treatment, and series of transverse sections through six individual shoots were analyzed for each genotype. Leaves were numbered from apex to base, with 1 being the youngest leaf primordium. In both genotypes, the morphology of axillary shoots observed at consecutive node positions suggested that buds were formed and developed in an acropetal sequence. The axils of a number of the youngest leaf primordia close to the vegetative shoot apical meristem were morphologically indistinguishable from the rest of the leaf primordium. Further basal, the first indication of axillary bud formation was a region of strongly staining meristematic cells at the base of the leaf primordium. This increased in size in leaf primordia further basal until an axillary bud could be distinguished as a zone of meristematic cells bulging out from the petiole of the subtending leaf. Axillary buds at more basal node positions showed increasing numbers of axillary leaf primordia. The most apical node showing axillary cell division, a clear axillary bud, and a bud with at least one leaf primordium was determined for each series of sections (Table I). The mean node positions at which these stages were first observed were not significantly different between the wild type and axr1-12. We also counted axillary leaf primordia at nodes 32, 36, and 40. Node 40 was the oldest node for which an accurate determination could be made. Below this node, axillary buds were often sectioned in an oblique plane and leaf primordia could not be counted. The mean numbers of leaf primordia up to node 40 were slightly but not significantly higher for axr1-12 than for the wild type (data not shown).

Table I Early stages of axillary shoot formation in wild-type and axr1-12 plants grown in short photoperiods

Thus, the pattern and timing of the early stages of axillary bud formation in axr1-12 was indistinguishable from that of the wild type in both long and short photoperiods.

Figure 4 Effect of the synthetic auxin 1-NAA on lateral inflorescence outgrowth from excised nodes of the wild type and axr1-12. The lowermost cauline node was excised from the elongating primary inflorescence of axenically grown plants and inserted between two (more ...)

When excised wild-type nodes were inserted between two slabs of hormone-free agar, the lateral inflorescences elongated and reached a mean length of 5.5 mm on d 4 and 30.6 mm on d 8 after excision (Fig. 4, left). 1-NAA (1 and 10 μm) inhibited elongation when applied to the agar in contact with the apical internode. Elongation was completely inhibited until d 4 after excision. After this time, some buds started to elongate, especially with the lower 1-NAA concentration. 1-NAA (1 or 10 μm) had no effect when contacting the basal internode. When nodes excised from axr1-12 plants (Fig. 4, right), were tested in the absence of auxin or with 1 or 10 μm basal 1-NAA, lateral inflorescence outgrowth showed a time course very similar to that of the wild type under these treatments. However, in contrast to wild type, elongation of axr1-12 inflorescences was not significantly inhibited by apically applied 1-NAA at 1 or 10 μm.

To study the regulation of shoot branching, we characterized lateral shoot development in the Columbia ecotype of Arabidopsis grown in long and short photoperiods. Under either photoperiod, the development in Arabidopsis leaf axils follows a similar sequence from empty axil through vegetative axillary shoot to floral shoot. The branching patterns observed in Arabidopsis plants must reflect the rates of initiation and progress between these states. In this study we investigated the role of the AXR1 gene in these events.

Studies of Arabidopsis ecotypes other than Columbia indicate that the sequence of axillary shoot formation along the primary shoot axis is acropetal during (prolonged) vegetative growth, but basipetal after floral transition (Hempel and Feldman, 1994; Grbić and Bleecker, 1996). Vegetative ecotype Columbia plants showed a clear acropetal pattern of axillary meristem formation, with detectable axillary cell divisions on average 17 nodes from the shoot apical meristem (Table I). After floral transition in long days, axillary cell divisions were seen all along the shoot axis (Fig. 3). Temporal resolution in our experiment may not have been sufficient to detect the basipetal sequence reported by Hempel and Feldman (1994) and Grbić and Bleecker (1996). In any case, floral transition coincides with a switch from the complete inhibition of axillary meristem development near the shoot apical meristem to the formation of axillary meristems in close proximity to it (from 17 nodes to less than three nodes distance in our experiments). This shows that axillary meristem formation is regulated independently of leaf age.

Early axillary shoot development in axr1-12 was indistinguishable from that of the wild type. In particular, the experiment in short photoperiods suggests that the axillary meristem and the first axillary leaf primordia were formed at a similar rate up to 40 nodes from the shoot apical meristem. Thus, AXR1 is not required to specify either of the opposing patterns of axillary meristem formation or the switch between them. AXR1 is also not involved in regulating the onset of axillary shoot formation in either pattern. Our observations were not sufficient to compare the formation of accessory axillary meristems in detail. They were detected in sections of both genotypes, which does not contradict the hypothesis that their formation is also unaffected by the axr1-12 mutation.

In short photoperiods, mutant buds were more advanced than those of the wild type before visible bolting of the primary shoot apex. Furthermore, differences between wild-type and axr1-12 lateral shoot development in long-day-grown plants were detected soon after floral transition and before significant fruit development (data not shown). Therefore, the effects of the axr1-12 mutation on lateral branching cannot just be the result of reduced fertility.

Involvement of AXR1 in the control of apical dominance had been proposed by Estelle and Somerville (1987) and Lincoln et al. (1990), based on the bushy phenotype of mature axr1 mutant plants. The AXR1 gene has been isolated (Leyser et al., 1993) and our present understanding of the mechanism of action of the AXR1 protein suggests that it modulates the ubiquitin-mediated degradation of regulatory proteins (del Pozo et al., 1998; Leyser, 1998; Ruegger et al., 1998). Although all of the phenotypes of axr1 loss-of-function mutants may be explained in terms of reduced auxin sensitivity, AXR1 might regulate the degradation of other proteins not involved in the auxin response. However, it has been reported that increased branching conferred by the axr1-3 mutation is epistatic to the IAA-M auxin over-producing transgene (Romano et al., 1995). We show that detectably enhanced lateral growth after axillary meristem formation in vivo correlates with auxin resistance of isolated lateral inflorescences in axr1. This suggests an AXR1-mediated role for auxin in regulating Arabidopsis shoot branching after axillary meristem formation.

No differences in the timing or the pattern of axillary meristem formation between axr1-12 and wild type were detected. This may be because axillary meristem formation is auxin independent. Alternatively, auxin may regulate axillary meristem formation in an AXR1-independent manner. Analogy with yeast signaling systems suggests that the AXR1 protein may increase the efficiency of the degradative pathway, rather than being absolutely required for it (Leyser, 1998). AXR1 may therefore only be necessary for a subset of auxin signaling events, for example, those triggered by low levels of auxin.

The axr1-12 mutation affects axillary shoot growth independently of the prevailing pattern of axillary shoot formation and growth along the shoot axis. This suggests that growth is regulated by a common factor, and the fact that the axr1-12 mutant is auxin resistant suggests that this factor may be auxin. If this is the case, then the switch in the pattern of lateral shoot growth with floral transition might be due to a change in auxin distribution. At floral transition, the primary shoot apex ceases producing leaves and it is these that are thought to be the source of axillary growth inhibition and the site of auxin production rather than the apical meristem itself (White et al., 1975; Weiss and Shillo, 1988). With technical advances in auxin analysis (Uggla et al., 1996), detailed information on the temporal and tissue distribution of auxin in Arabidopsis may be obtained in the future.

Previous studies indicate that auxin does not act directly in the axillary bud, since apically applied radiolabeled IAA is not transported into inhibited buds (Hall and Hillman, 1975; Morris, 1977). To address the question of where auxin acts, we are currently expressing the wild-type AXR1 gene in an axr1-12 mutant background under the control of a variety of tissue-specific promoters. We hope to establish which tissues require the wild-type AXR1 protein to repress axillary shoot growth.

We thank Dr. Voijslava Grbić for helpful discussions, Dr. Karin van de Sande, Dr. Jon Booker, and Stephen Day for critical reading of the manuscript, Megan Stark for photographic work, and the horticultural technicians at the University of York for excellent plant care.