MAX2 participates in an SCF complex which acts locally at the node to suppress shoot branching


*For correspondence (fax +44 1904 328682; e-mail


The Arabidopsis gene ORE9/MAX2 encodes an F-box leucine-rich repeat protein. F-box proteins function as the substrate-recruiting subunit of SCF-type ubiquitin E3 ligases in protein ubiquitination. One of several phenotypes of max2 mutants, the highly branched shoot, is identical to mutants at three other MAX loci. Reciprocal grafting, double mutant analysis and gene cloning suggest that all MAX genes act in a common pathway, where branching suppression depends on MAX2 activity in the shoot, in response to an acropetally mobile signal that requires MAX3, MAX4 and MAX1 for its production. Here, we further investigate the site and mode of action of MAX2 in branching. Transcript analysis and a translational MAX2–GUS fusion indicate that MAX2 is expressed throughout the plant, most highly in developing vasculature, and is nuclear-localized in many cell types. Analysis of cell autonomy shows that MAX2 acts locally, either in the axillary bud, or in adjacent stem or petiole tissue. Expression of MAX2 from the CaMV 35S promoter complements the max2 mutant, does not affect branching in a wild-type background and partially rescues increased branching in the max1, max3 and max4 backgrounds. Expression of mutant MAX2, lacking the F-box domain, under the CaMV 35S promoter does not complement max2, and dominant-negatively affects branching in the wild-type background. Myc-epitope-tagged MAX2 interacts with the core SCF subunits ASK1 and AtCUL1 in planta. We conclude that axillary shoot growth is controlled locally, at the node, by an SCFMAX2, the action of which is enhanced by the mobile MAX signal.


All plant shoots are composed of simple, repeated units. These are produced by the primary shoot apical meristem (SAM) and consist of an internode (stem) and a node bearing one or several leaves with a secondary SAM in their axil. These secondary, axillary SAMs have the same growth potential as the primary SAM. In reality, however, their activity is tightly controlled and coordinated within the plant, and responsive to developmental and environmental cues. Both the diverse shoot branching patterns of different plant species and the variation within one pattern that individuals can display involve the regulation of secondary SAM activity. It has long been thought that mobile signals control branching, and the role of known mobile plant hormones, especially auxin and cytokinin, has been extensively studied (Cline, 1994; Tamas, 1995).

A genetic approach to shoot branching has provided evidence for a new mobile branching signal. Mutants defective in the repression of secondary SAM activity were identified in petunia (dad mutants; Napoli, 1996; Napoli and Ruehle, 1996; Snowden and Napoli, 2003), in pea (rms mutants; Beveridge, 2000) and in Arabidopsis (max mutants; Booker et al., 2004, 2005; Sorefan et al., 2003; Stirnberg et al., 2002). These mutants can be divided into two classes. For the first class (dad1, dad3; rms1, rms5; max1, max3, max4), the bushy shoot phenotype is rescued by grafting onto a wild-type root (Booker et al., 2005; Morris et al., 2001; Napoli, 1996; Snowden et al., 2005). From a combination of grafting and hormonal analysis with this class of mutants in pea, it was concluded that they lack an acropetally mobile branching inhibitor, unlikely to be auxin or cytokinin (Beveridge et al., 2000; Foo et al., 2001; Morris et al., 2001). This is supported by the cloning of Arabidopsis MAX genes. MAX4 encodes a carotenoid cleavage dioxygenase enzyme (AtCCD8), and RMS1 and DAD1 are its pea and petunia orthologues (Snowden et al., 2005; Sorefan et al., 2003). MAX3 encodes a related enzyme (AtCCD7; Booker et al., 2004) which might act upstream of MAX4 in the sequential cleavage of the same carotenoid substrate (Auldridge et al., 2006; Schwartz et al., 2004). Reciprocal root–shoot grafting experiments in Arabidopsis place MAX3 and MAX4 upstream of MAX1, which encodes a cytochrome P450 enzyme with homology to mammalian thromboxane A2 synthase (Booker et al., 2005). While the substrate and the product of this enzymatic pathway still await identification, recent evidence indicates that it acts as a negative regulator of polar auxin transport (Bennett et al., 2006; Lazar and Goodman, 2006). Furthermore, cytokinin export from the root is reduced in some of the rms mutants (Beveridge et al., 1997a,b; Morris et al., 2001). Therefore, the new pathway does not act in isolation. Rather, it participates in a hormonal network which controls axillary shoot growth.

The second class of mutants (e.g. dad2; rms3 and rms4; max2) is not rescued by grafting onto a wild-type root (Booker et al., 2005; Morris et al., 2001; Snowden et al., 2005). A possible function for this class of gene is to mediate the response to the new signal in the shoot. MAX2 is a strong candidate for acting in this way, because the branching phenotype of max2 is identical to max1, max3 and max4 and is not significantly enhanced in double mutant combinations (Bennett et al., 2006; Stirnberg et al., 2002). MAX2 encodes an F-box leucine-rich repeat protein (Stirnberg et al., 2002). This class of proteins functions as the substrate-recruiting subunit of SCF complexes (Cardozo and Pagano, 2004; Kipreos and Pagano, 2000; Willems et al., 2004). SCF complexes are ubiquitin E3 ligases which can catalyse polyubiquitination of their protein substrates, thereby marking them for degradation by the 26S proteasome (Ciechanover et al., 2000; Smalle and Vierstra, 2004). Thus, repression of axillary growth might require the degradation of as yet unknown protein(s) by an SCFMAX2, downstream of the MAX signal. Recent studies of gibberellic acid, ethylene and auxin signalling indicate that SCF-mediated proteolysis is involved in the response to numerous signals in plants (Thomann et al., 2005).

The cloning of MAX2 revealed its identity to ORE9, which had been isolated earlier as a positive regulator of Arabidopsis leaf senescence (Woo et al., 2001). Furthermore, mutations in ORE9/MAX2 affect the repression of hypocotyl growth by light (Stirnberg et al., 2002), cause resistance to oxidative stress (Woo et al., 2004) and suppress the enhanced response to drought caused by mutation of EDR1, which encodes a CTR1-like protein kinase (Tang et al., 2005). Thus, mutation of ORE9/MAX2 confers phenotypes apparently unrelated to branching. Information about their occurrence in max1, max3 or max4 is incomplete. However, at least one phenotype, the elongated hypocotyl in the light, is not displayed by these mutants (Stirnberg et al., 2002; PS, HMOL unpublished observations). Therefore, it is possible that ORE9/MAX2 has roles outside the MAX pathway.

Here, we examine further the role of MAX2 in branching control. Combining results from a range of different experiments, we conclude that the suppression of branching is dependent only on a small subset of the MAX2 expression domain, and that an SCFMAX2 acts locally, at each node, to suppress growth of its axillary bud in response to the mobile signal produced by MAX3, MAX4 and MAX1.


MAX2 is expressed throughout the plant

MAX2 transcript was detected by RT-PCR in all plant organs tested, including cauline and rosette axillary buds of plants at early reproductive stage (Figure 1a). To investigate tissue and subcellular localization of the MAX2 protein, 3.45 kb of MAX2 upstream sequence and the MAX2 coding region, which lacks introns, were fused to the GUS reporter gene. This construct, M2p::M2–GUS, partially complemented the highly branched max2 mutant phenotype. Out of five homozygous, single-insert lines in the max2 background, which showed similar GUS staining patterns, we chose the best-complemented line (Figure 1b) for detailed study. GUS activity was detected in many cell types, but was highest in the vasculature of growing leaves, flowers, siliques and stems (Figure 1c–f). In siliques, the funiculi, which connect the developing seeds to the placenta, stained particularly strongly (Figure 1e). In transverse sections of the vasculature of elongating inflorescence stems, GUS was detected in phloem, cambium and xylem parenchyma cells (Figure 1f). Vascular GUS activity declined when leaves and stems ceased growing, but remained high in the funiculi of ripening siliques (data not shown). Buds developing in rosette leaf axils (Figure 1g) showed uniform GUS staining. In the root, GUS activity was highest in developing vascular, pericycle and endodermal cells at the tip; weak staining was also detected in outer cell types and the root cap (Figure 1h,i). The overall vascular staining intensity decreased towards the base of the root (compare Figure 1i,j; root tip versus root hair differentiation zone). The subcellular pattern of GUS activity, for example in differentiating root cells (Figure 1i,j) and leaf trichomes (Figure 1k), suggests that the MAX2–GUS fusion concentrates in the nucleus. Nuclear accumulation of MAX2–GUS fusion protein was confirmed for dark-grown hypocotyl cells by comparing the patterns of GUS activity and of nuclear staining with the DNA-specific dye 4′,6-diamidino-2-phenylindole (DAPI). In hypocotyls expressing a control construct, a fusion of amino acids 1–26 of MAX2 with GUS expressed from the MAX2 promoter, we observed diffuse GUS staining which differed from the DAPI staining pattern (Figure 2a,b). In contrast, hypocotyls expressing the full-length MAX2–GUS fusion under the MAX2 promoter, showed dots of GUS activity which co-localized with DAPI fluorescence (Figure 2c,d).

Figure 1.

 MAX2 is expressed throughout the plant.
(a) Detection of MAX2 transcript by RT-PCR in total RNA from different organs. Seedlings were harvested after 13 days of growth on vertical agar plates. Vegetative shoot tissues were harvested from soil-grown plants, either when the primary inflorescence was 3–4 cm high or when fully elongated (labelled ‘late’). Flowers were harvested at stage 14 (Smyth et al., 1990) and siliques 4 days past stage 14. Detection of ACTIN2 transcript was used as a cDNA normalization control. cyc, cycle number.
(b) Partial rescue of the max2 branching phenotype in a transgenic max2 line expressing MAX2–GUS fusion protein from the MAX2 promoter (M2p::M2–GUS). Branching was assessed in a decapitation assay (see Experimental procedures). The mean number of rosette branches of at least 2 cm length 10 days after decapitation is shown (error bar = SEM, = 20).
(c–k) Different organs of M2p::M2–GUS (max2) plants stained for GUS activity. If not mentioned otherwise, harvest stages were as in (a): (c) top rosette leaf (bar = 5 mm); (d) flower (bar = 1 mm); (e) silique (bar = 1 mm); (f) stem (part of a transverse section of the second internode of the primary inflorescence, with a vascular bundle in the centre, bar = 50 μm); (g) rosette axillary bud from a plant induced to flower, prior to bolting (bar = 100 μm); (h) primary root, harvested after 5 days of growth on vertical agar plates (transverse section about 0.4 mm above the root tip, bar = 50 μm); (i) root tip and (j) root hair differentiation zone in longitudinal sections of the primary root, harvested after 7 days of growth on vertical agar plates (bars = 50 μm); (k) leaf trichome (bar = 50 μm).

Figure 2.

MAX2–GUS fusion protein is nuclear-localized in dark-grown hypocotyl cells.
Fixed 5-mm segments from 5-day-old dark-grown hypocotyls were stained for GUS activity, and with DAPI to detect nuclei. The figure shows about one-third of the width of the hypocotyls, with the outside on the left. Scale bars = 50 μm.
(a, b) Transgenic control line expressing GUS fused to the 3.45-kb MAX2 upstream sequence plus the sequence encoding N-terminal amino acids 1–26 of the MAX2 protein (M2p::M2(1–26)–GUS) in the wild-type background. GUS activity (a, brightfield illumination) and nuclear staining (b, epifluorescence illumination) do not co-localise.
(c, d) Transgenic line expressing GUS fused to the 3.45-kb MAX2 upstream sequence plus the complete MAX2 coding sequence (M2p::M2–GUS) in the wild-type background. GUS activity (c, brightfield) and nuclear staining (d, epifluorescence) co-localise.

MAX2 is required at each node to suppress axillary bud growth

Root–shoot grafting demonstrates that MAX2 is required in the shoot to repress branching (Booker et al., 2005). To determine the site(s) of action of MAX2 in the shoot, we produced chimeric shoots carrying max2 mutant sectors in a phenotypically wild-type (MAX2/max2) background and studied their axillary bud growth.

As a tool to generate easily detectable, colour-marked sectors, we used the Arabidopsis cell autonomy (CAUT) lines. They employ the CHLORATA-42 gene (CH-42), normally located on chromosome IV, as a marker. chlorata-42 is a recessive, homozygous viable T-DNA insertion allele (Koncz et al., 1990). The yellow ch-42 mutant phenotype manifests in all cells that produce chloroplasts and is cell autonomous (Furner, 1996; Furner et al., 1996). The CAUT lines were generated by transformation of ch-42 homozygous plants with the dominant (green) CH-42 wild-type allele. The individual transformant lines were extensively backcrossed to the yellow (ch-42) parent until they segregated 1:1. Such single-insert lines were made homozygous and the inserts mapped using recombinant inbred lines (Lister and Dean, 1993), gridded bacterial artificial chromosomes (BACs; Choi et al., 1995; Liu et al., 1995, 1999; Mozo et al., 1998) or flanking DNA sequence. Seventy-six lines are available, each with a correcting insert at a single mapped site. In effect this means that the capacity to produce chlorophyll, and hence a cell-autonomous colour marker, is translocated to a different site in each line. To use the CAUT lines to study cell autonomy, the mutant of interest is backcrossed into the homozygous yellow ch-42 mutant background and then to a CAUT line with a correcting insert near the wild-type copy of the locus. X-irradiation of F1 or F2 seeds results in yellow sectors on the plant derived from the L2 and/or L3 layers of the seed SAM. Such sectors have usually lost both the correcting insert and the adjacent wild-type copy of the gene of interest and are sectorial chimeras with marked mutant tissue surrounded by wild-type tissue and overlain by a wild-type (L1-derived) colourless epidermis.

To study the cell autonomy of the MAX2 gene, we generated the ch-42 max2 double mutant. This was crossed with either of two CAUT lines carrying CH-42 inserted close to the MAX2 locus on chromosome II (7F and A24, see Experimental procedures). The resulting F1 (and half of the F2, which is more easily bulked up) has the target genotype for sector generation shown in Figure 3(a). These plants show repressed, wild-type branching and are green. Irradiation-induced deletion of the linked wild-type copies of MAX2 and CH-42 from one chromosome II homologue (red in Figure 3a) in a meristematic cell will result in yellow-marked, max2 somatic sectors. We irradiated the dry F2 seed to induce chromosomal deletions in embryonic SAM cells. Sectors on the first leaf pair are typically frequent but very small, sectors affecting later leaves are typically larger but less frequent and sectors affecting late leaves are typically very large and infrequent (Furner and Pumfrey, 1992; Irish and Sussex, 1992). This is because the cells at the periphery of the dry seed SAM give rise to the very early leaves and later leaves are derived from cells that undergo proliferation in the meristem before leaf initiation. Cells at the centre of the dry seed SAM can take over the meristem over time and the whole layer of the meristem will become of a single marked genotype late in development (Furner, 1996).

Figure 3.

MAX2 is required at each node to repress bud growth.
(a) Genotype of the irradiated plants. Note the plant is mutant (ch-42) at the authentic locus on chromosome IV and the green phenotype is conferred by the transgene CH-42 insert on chromosome II. Radiation-induced loss of the transgene also results in loss of the wild-type copy of MAX2 (red region) resulting in yellow ch-42 tissue hemizygous for the max2 mutation. Sectors can occur in any tissue but will only be seen if they affect chlorophyll-producing tissue.
(b, c) Sector-free rosettes with a wild-type phenotype. Arrows in (c) indicate the small axillary buds.
(d) Rosette with a large yellow max2 sector on the left. The apex of the plant and the youngest leaves were removed. Axillary buds in the sector have produced leaves about half the length of the rosette leaves (arrowheads). No axillary leaves are visible in the wild-type part of the rosette.
(e) Rosette in which one of the oldest leaves carried a yellow max2 sector which extended into the axil. The yellow leaves (arrowheads) were produced by the bud that developed in this axil.
(f) Rosette in which several parastichious leaves carried a narrow yellow max2 sector which extended into the axil. The photograph shows one of the sectored rosette leaves (white arrow) with its axillary bud. The bud centre and most axillary leaves (white arrowheads) are yellow, but some of the outer axillary leaves are chimeric or green (black arrowheads).

Wild-type and max2 vegetative rosettes show a pronounced difference in axillary bud size over a wide range of node positions along the shoot axis (Stirnberg et al., 2002). Furthermore, large vegetative sectors may be generated from cells at the centre of the embryonic SAM when flowering is delayed (Furner et al., 1996). Therefore, the irradiated F2 was grown in short photoperiods to prolong vegetative growth of the plants. F2 individuals with large yellow ch-42 sectors affecting the (L2-derived) leaf tissue and non-sectored F2 controls (green MAX2 and yellow max2 individuals) were selected and transferred to individual pots for further growth and observation.

Figure 3(b) shows the vegetative rosette of a green MAX2 control 9 weeks after sowing, when phenotypic analysis was carried out. The axillary buds were about 1 mm in length, lacking expanding axillary leaves (Figure 3c). Development of yellow ch-42 max2 control plants was significantly delayed due to chlorophyll deficiency, but the enlarged axillary buds with expanding axillary leaves typical of max2 appeared later. No delay in the development of ch-42 max2 tissue was observed in chimeric plants, where extensive ch-42 max2 sectors were maintained by wild-type tissue (Figure 3d). Table 1 lists the chimeric F2 plants (A–N) for which the phenotypic analysis was performed and for which the max2 mutant sector genotype was confirmed by genetic analysis of sector F3 progeny. In individuals A–E, the sector consisted of many rosette leaves and either the whole SAM (A–C) or part of the SAM (D, E, Figure 3d), which continued to produce yellow max2 leaves. In individuals F–H, several yellow max2 leaves had been produced, but the sector tissue had become eliminated from the SAM, which was only producing green, wild-type leaves at the time of analysis. Individuals I–N carried more narrow sectors which affected the central part of one (K–N, Figure 3e) or several leaves (I, J, Figure 3f). In most individuals, the mutant sector included both the L2 (sub-epidermal) and L3 (central) tissue layer except for C, E and F, where sector leaves with a pale green centre and yellow margins indicated that only the L2 was ch-42 max2. In Arabidopsis, the green tissue of the axillary bud originates from two or more L2 cells at the base of the subtending leaf (Furner and Pumfrey, 1992; Irish and Sussex, 1992). Therefore, a leaf completely included in a mutant sector should carry a mutant bud. If only part of the leaf carries a sector, a central sector is likely to extend into the petiole and include the axillary bud (Furner and Pumfrey, 1992). There are no data on the colourless (L1 derived) epidermis in these situations, but it can reasonably be presumed that these mutant buds have a wild-type epidermis. The recovery of non-L2 gametes (see below) from some plants implies that this is the case.

Table 1.   Genetic mosaic analysis of MAX2 action in the shoot. List of chimeric shoots bearing max2 sectors
F2 individualaShoot organs included in sectorLeaf tissue layers contributing to sectorSector bud colourbSector bud sizeNon-sector bud sizePhenotypic segregation in F3 from a bud in sectorb
LeavesShoot apexg MAX2y max2Others
  1. aMost individuals were selected from an X-irradiated F2ch-42 max2 × CAUT line 7F, except for C, E and L, which were selected from an X-irradiated F2ch-42 max2 × CAUT line A24.

  2. by, yellow, g, green.

AManyWholeL2 and L3yLargeSmall1251
BManyWholeL2 and L3yLargeSmall0220
CManyWholeL2y/few chimericLargeSmall0292
DManyPartL2 and L3yLargeSmall0240
GSeveralL2 and L3y/chimericLargeSmall0260
HSeveralL2 and L3yLargeSmall12260
IParts of severalL2 and L3ChimericLargeSmall0240
JParts of severalL2 and L3ChimericLargeSmall0280
KPart of oneL2 and L3yLargeSmall0190
LPart of oneL2 and L3ChimericLargeSmall0280
MPart of oneL2 and L3yLargeSmall1260
NPart of oneL2 and L3yLargeSmall0240

In all sectored plants, independent of sector size, axillary buds in the max2 sector were large, with several expanding leaves. Axillary buds outside the sector were wild-type (Table 1, Figure 3d–f). Thus sector bud phenotype corresponded to sector genotype, and was not affected by wild-type MAX2 function in the epidermis or other shoot or root tissue. Sector axillary buds subtended by a completely mutant leaf (Figure 3d) were indistinguishable from those subtended by a leaf with part of the lamina green (Figure 3f). Thus, MAX2 activity in the lamina is not sufficient to rescue the max2 bud phenotype. Rather, MAX2 may act in or close to the axillary bud to repress its growth. Some of the non-repressed axillary buds were chimeric, i.e. contained some green MAX2 tissue (Table 1 and Figure 3f). This occurred when the leaf subtending the bud included a sector boundary, and likely mutant and non-mutant cells participated in formation of these axillary buds. These buds had green or chimeric outer axillary leaves, but the younger axillary leaves were yellow (Figure 3f). This further delimits the tissue requiring MAX2 for bud repression to the central tissues of the bud, or tissue of the leaf or primary shoot axis close to the bud, and hints that max2 tissue may exert non-autonomous effects on adjacent wild-type tissue, incorporating adjacent non-mutant tissue into the axillary shoot.

The loss of the wild-type MAX2 copy in the yellow-marked sector tissue was confirmed after phenotypic analysis. Shoots were decapitated above the sector and shifted to long photoperiods to encourage outgrowth of inflorescences from the sector. Loss of the linked CH-42 and MAX2 wild-type copies should result in uniformly yellow max2 F3 sector progeny. This was confirmed for the majority of the 14 chimeric plants (Table 1). However, green wild-type MAX2 segregants were found in F3 sector progeny from three chimeric plants (A, H, M), the proportion being significantly lower than the 75% expected had deletion of MAX2 and CH-42 not occurred. These segregants may have resulted from invasion of the sector L2 lineage by wild-type cells from the epidermal L1 layer during formation of the axillary branch from which the F3 seed was collected. Such non-L2 gametes were also found in earlier studies of cell autonomy (Bouhidel and Irish, 1996; Furner et al., 1996). For a few F3 segregants, branching or colour phenotype could not be classified (individuals A, C). These might have carried additional radiation-induced mutations severely affecting development.

In conclusion, the sector analysis shows that MAX2 is required in the green (L2-derived) tissue at each individual node for repression of its associated axillary bud, and that it acts either in the bud itself or close to it.

MAX2 overexpression partially rescues the branching phenotype of max1, max3 and max4

To investigate whether the level of MAX2 in the plant is limiting for repression of branching, we expressed MAX2 in plants under control of the strong CaMV 35S promoter (35S::M2). First, we tested whether the overexpression construct complements the max2 mutant phenotype. Thirty-nine of 42 35S::M2 (max2) primary transformants had a wild-type phenotype. From these, 10 independent lines containing a single insert were brought to homozygosity. For nine lines, branching was reduced to wild-type level and RT-PCR analysis showed weak to moderate increases of MAX2 transcript level compared with untransformed max2. One line was not rescued, but the endogenous plus transgenic MAX2 transcript level was not increased in comparison with untransformed max2 (see Figure S1a). Thus, the 35S::M2 construct is functional and directs MAX2 expression in those tissues where it is required to repress branching.

For 35S::M2 in the wild-type background, we obtained 30 independent primary transformants. None of these were noticeably different from the wild type. From these, 11 lines were taken to homozygosity. Two homozygous lines showed a significant increase in MAX2 transcript level, but branching did not differ significantly from wild type in long photoperiods (data not shown). The highest-expressing line was further compared with the wild type in a decapitation assay designed to detect small alterations in shoot branching (Figure 4). Plants were grown in short photoperiods for 30 days to increase the number of vegetative nodes, outgrowth from which was encouraged by decapitation of the primary inflorescence after induction of flowering in long photoperiods. Again, branching from the rosette was not significantly different between wild type and the high-expressing 35S::M2 (wt) line (Figure 4a). In leaf samples taken at the end of the experiment, the detection threshold of MAX2 transcript by RT-PCR was reached at least five cycles earlier in this 35S::M2 line than in wild type, indicating a significant rise in transcript level (Figure 4b). This suggests that the level of MAX2 is not limiting for repression of branching in wild-type plants.

Figure 4.

The effect of overexpressing MAX2 in wild-type (wt), max1, max3 and max4 genetic backgrounds.
(a) Branching of wild type, max1, max3 and max4 in the absence (untransformed) or in the presence of MAX2 overexpression from the 35S promoter (35S::M2). Branching was assessed in a decapitation assay (see Experimental procedures). The mean number of rosette branches with a length of at least 2 cm 10 days after decapitation is shown (error bar = SEM, = 11).
(b) Analysis of MAX2 expression for the experiment presented in (a). Reverse transcriptase-PCR from total RNA of rosette leaves, collected after branching had been assessed, using primers that amplify endogenous plus transgenic MAX2. Detection of ACTIN2 transcript was used as a cDNA normalization control. cyc, cycle number.

Grafting and double mutant analysis suggest that the MAX genes define a signalling pathway that results in repression of bud growth, in which MAX2 acts in the downstream response to the signal synthesized by the action of MAX1, MAX3 and MAX4 (Booker et al., 2005). To test this hypothesis further, we asked whether MAX2 overexpression suppresses the branching phenotype of max1, max3 or max4. The highest-expressing 35S::M2 (wt) line (selective marker kanamycin resistance) was crossed with these mutants and with max2. Forty kanamycin-resistant F2 individuals were selected for each cross. Amongst these, we could identify homozygous max mutants, bushier than wild type, for the crosses with max1, max3 and max4. The proportions were not significantly different from 0.25. F3 seed was collected from these max mutant F2 individuals and a family homozygous for the 35S::M2 insert selected. For the cross with max2, all kanamycin-resistant F2 plants were wild type. As MAX2 and MAX3 are closely linked (Booker et al., 2004; Stirnberg et al., 2002), max2 and max3 should recombine with 35S::M2 with similar frequencies. Therefore, the lack of max2 segregants must be due to mutant rescue, confirming that the 35S::M2 transgene used in the crosses was functional. Shoot branching of max1, max3 and max4 in the presence and absence of 35S::M2 was compared in the decapitation assay described above (Figure 4a). 35S::M2 partially rescued the increased branching of max1, max3 and max4. Using leaf samples collected at the end of the assay, we confirmed that MAX2 overexpression in the max, 35S::M2 double homozygotes was at a level similar to the 35S::M2(wt) line (Figure 4b).

This result suggests that the level of MAX2 becomes limiting for repression of branching when the inhibitory signal produced by MAX1, MAX3 and MAX4 is low, and is consistent with an action of MAX2 downstream of MAX1, MAX3 and MAX4. The incomplete rescue of max1, max3 and max4, in spite of a substantial rise in MAX2 transcript level, indicates that the signal produced by MAX1, MAX3 and MAX4 acts to promote MAX2 action in some way other than transcriptional upregulation.

F-box-deleted MAX2 has a dominant-negative effect on shoot branching

SCF-type E3 ubiquitin ligases are multiprotein complexes, in which core subunits with catalytic and scaffold function (ubiquitin-conjugating enzyme E2, Skp1, Cullin, Rbx1) combine with a variable F-box protein subunit that confers the substrate specificity (Cardozo and Pagano, 2004; Willems et al., 2004). Two functional domains enable F-box proteins to mediate substrate ubiquitination. One domain binds the substrate protein. In MAX2, the C-terminal leucine-rich repeat region probably fulfils this role. The second domain, the F-box, is required for assembly of the SCF complex by binding to Skp1. An F-box-homologous region is located at the N-terminus of MAX2 (Stirnberg et al., 2002; Woo et al., 2001).

To test for the significance of the F-box domain for MAX2 protein function, we constructed 35S::ΔF–M2, and introduced it into max2 plants. This construct was identical to 35S::M2, except that the codons for amino acids 9–47 of the protein, spanning the F-box, were deleted. All 15 35S::ΔFM2 (max2) primary transformants had a mutant phenotype. Seven independent, single-insert lines were brought to homozygosity. Despite weak to moderate increases in endogenous plus transgenic MAX2 transcript, their branching was not reduced compared with untransformed max2 (Figure S1b). This suggests that deletion of the F-box abolishes the function of the MAX2 protein.

Deletion of the F-box domain may leave substrate protein binding activity unaffected. It has been reported that co-expression of such ΔF deletion versions with the endogenous F-box-containing protein interferes with ubiquitination of the target protein, resulting in its stabilization and increased activity (Hart et al., 1999; Kitagawa et al., 1999; Marikawa and Elinson, 1998; Wu et al., 2001; Yaron et al., 1998). To test for such dominant-negative action, 35S::ΔFMAX2 was transformed into wild-type plants. We obtained 28 primary transformants, which were either indistinguishable from wild type, or showed somewhat increased branching. Seven single-insert lines, including both phenotypic groups, were made homozygous. The analysis of their branching, and their endogenous and transgenic (ΔF) MAX2 transcript levels, in comparison with wild type and max2, is shown in Figure 5. Three lines (98, 101, 106) showed a significant increase in branching (about twice the number of branches compared to wild type) and had an intermediate level of ΔFMAX2 transcript. Four lines had a branching pattern indistinguishable from the wild type. The two lowest (100, 102), but also the two highest, ΔFMAX2-expressing lines (103, 104) belonged to this group. Similar results were obtained in two further independent repeats of this experiment (data not shown). The highest-expressing lines (103, 104) showed no mutant rescue when crossed into the max2 mutant background, ruling out the possibility that the F-box becomes dispensable for MAX2 function at this level of expression (data not shown).

Figure 5.

Expression of F-box-deleted MAX2 in the wild-type (wt) background.
(a) Branching in wild type, max2 and seven independent homozygous lines expressing F-box-deleted MAX2 from the 35S promoter (35S::ΔFM2). Branching was assessed as in Figure 4(a), = 9–10.
(b) Analysis of endogenous MAX2 and transgenic, F-box-deleted MAX2 expression for the experiment presented in (a). Reverse transcriptase-PCR from total RNA of rosette leaves, collected after branching had been assessed. Reverse priming in their divergent terminators was used to amplify selectively endogenous and transgenic MAX2 transcripts. Detection of ACTIN2 transcript was used as a cDNA normalization control. cyc, cycle number.

Apart from a dominant-negative action at the protein level, introduction of the 35S::ΔFM2 transgene could have caused an identical phenotypic effect by co-suppressing the endogenous, functional MAX2 copy. None of the 35S::ΔFM2 lines showed a substantial reduction of endogenous MAX2 transcript compared to untransformed wild type (Figure 5b). Endogenous MAX2 expression tended to be highest in the two highest-expressing 35S::ΔFM2 lines, 103 and 104. It tended to be lowest in the three 35S::ΔFM2 lines that showed increased branching, but was not lower than in wild-type plants in two experiments and only slightly lower in the third.

MAX2 interacts with core SCF subunits in vivo

Two experimental approaches have been taken to examine whether ORE9/MAX2 participates in an SCF complex. In a yeast two-hybrid assay, the F-box domain of ORE9 interacted with the Skp1-like Arabidopsis protein ASK1; furthermore, in vitro translated ORE9 bound purified GST-ASK1 fusion protein in an F-box-dependent manner (Woo et al., 2001). We investigated whether MAX2 interacts with core SCF subunits in planta. A C-terminally myc epitope-tagged version of MAX2, under control of the 35S promoter, was expressed in the max2 mutant (35S::M2-myc). In contrast to 35S::M2 primary transformants, few 35S::M2-myc transformants showed complete rescue of the branching phenotype. Analysis of MAX2 transcript levels indicated that the highest-expressing 35S::M2-myc lines showed the best rescue (data not shown), suggesting that the epitope tag reduces protein function but does not abolish it completely. A high-expressing, completely rescued 35S::M2-myc line was used for co-immunoprecipitation analysis (Figure 6). The SCF subunits ASK1 and AtCUL1 (an Arabidopsis Cullin) were detected in protein extracts from non-transgenic max2 and from max2 expressing MAX2-myc. There were no differences in expression levels of these proteins between both lines. Co-immunoprecipitation of ASK1 and AtCUL1 with anti-c-myc antibody depended on the presence of MAX2-myc in the extract. Cycles of covalent attachment of the small protein RUB1 to AtCUL1, followed by its removal, are crucial for optimal SCF ubiquitination activity (Parry and Estelle, 2004). The two bands detected with anti-CUL1 antiserum are therefore probably RUB1-modified and unmodified AtCUL1, which both co-immunoprecipitated with MAX2-myc. The ASK1 polyclonal antiserum we used cross-reacts with ASK2, which is a protein closely homologous and functionally redundant to ASK1 (Gagne et al., 2002; Liu et al., 2004). MAX2-myc appeared to interact with both Skp1-like proteins. This experiment demonstrates the interaction of MAX2 with core SCF components in vivo.

Figure 6.

MAX2 interacts with SCF core components ASK1 and AtCUL1 in vivo.
Anti-myc immunoprecipitates were prepared from total protein extracts of max2 and a transgenic line expressing myc-tagged MAX2 in the max2 background [35S::M2-myc (max2)]. Total protein extracts (lanes 1, 2) and immunoprecipitates (lanes 3, 4) were analysed by Western blotting and probing with polyclonal antiserum raised against AtCUL1 (top) and against ASK1 (bottom).


The site of MAX2 action

Our analysis of expression shows that MAX2 is not limited to axillary shoots or branch points. It is detected throughout the plant, with the highest levels in developing vasculature. This pattern of expression, and the pleiotropic phenotypes that result from mutation of this gene (Stirnberg et al., 2002; Tang et al., 2005; Woo et al., 2001, 2004), indicate that MAX2 might play multiple roles, in different parts of the plant. With regard to bud growth repression, root–shoot grafting demonstrated that MAX2 acts in the shoot (Booker et al., 2005). A genetic mosaic approach allowed us to further delineate the site of MAX2 action.

Mosaics of mutant and wild-type tissues for a gene of interest can be created in a heterozygote by radiation-induced chromosomal deletion (Poethig, 1987). Detection of the mutant tissue is achieved by simultaneous deletion of a cell-autonomous marker gene, located on the same chromosome arm, usually centromere proximal to the gene of interest. This approach has been widely used to study the cell autonomy of genes in maize (e.g. Becraft et al., 2001; Foster et al., 1999; Osmont et al., 2006; Scanlon, 2000). Few mapped and usable marker loci are known in Arabidopsis. The CAUT lines overcome this limitation by providing the marker CH-42 as a transgene at known chromosomal locations, in a ch-42 mutant background. The mapped correcting inserts allow the CH-42 marker to be used for cell-autonomy analysis of genes located almost anywhere in the genome. A CAUT line carrying CH-42 inserted near the gene of interest is crossed with a mutant for that gene in the yellow ch-42 mutant background. The resulting F1 genotype, heterozygous mutant and hemizygous for the CH-42 transgene in a ch-42 mutant background, is used to induce ch-42-marked hemizygous mutant sectors on a green heterozygous plant. In contrast to genetic mosaics created with transposase- or site-specific recombinase-based systems (e.g. Jenik and Irish, 2001; Sieburth et al., 1998), there is no need to generate molecular constructs and transgenic plants. The only information needed to use the CAUT lines is the chromosomal location of the gene of interest and the method can be used for genes whose product is unknown. As the sectors can be scored in living material, continued sector growth can be examined after identification of the sector and the genotype of the sector and parent assessed by collecting seeds set in the tissue.

Our mosaic analysis shows that MAX2 is required at each node to repress its associated axillary bud. Thus MAX2 acts locally, likely in the perception of, or the response to, the long-range inhibitory signal produced by MAX1, MAX3 and MAX4. We used seed irradiation to induce loss of MAX2 function in a single cell of the embryonic SAM. The clonal descendants of this cell produce a radial sector of mutant tissue in the developing shoot. Even a narrow mutant sector restricted to one node and its axillary bud includes part of the shoot axis, the petiole and the leaf lamina. Of these tissues, we were able to exclude the leaf lamina and the outer leaves of the bud as the site of MAX2 action: there were nodes that had wild-type MAX2 present in these tissues, but produced a mutant bud. Thus MAX2 acts either in the centre of bud, or in the petiole or shoot axis close to the bud. Expression analysis does not allow us to limit the site of action further, as MAX2 is expressed in all these tissues. As some of the phenotypically mutant buds included wild-type L2-derived tissue and the L1-derived epidermis of the buds did not normalize the mutant L2-derived tissue, the yellow max2 tissue exerts a non-autonomous effect on the wild-type tissue.

Two recent publications conclude that increased branching of the max mutants results from increased polar auxin transport, caused by overexpression of members of the PIN and AUX1 auxin transporter families (Bennett et al., 2006; Lazar and Goodman, 2006). Their models of action of the MAX pathway are based on the established correlation between a bud’s growth activity and its ability to export auxin into the stem (Li and Bangerth, 1999; Morris, 1977; Morris and Johnson, 1990). It is likely that this effect is mediated by competition between the bud and the primary apex for auxin transport capacity in the main stem (Bennett et al., 2006), with the increased transport capacity in the stem of the max mutants allowing active growth of the primary apex as well as multiple axillary buds. However, an effect on auxin transport in the bud cannot be ruled out, since PIN transporters are also overexpressed in the buds of max mutants (Lazar and Goodman, 2006). The main PIN gene overexpressed in buds is PIN4, whereas in the stem it is PIN1. Since loss of PIN1 function is sufficient to suppress the max shoot branching phenotype, it is likely that the stem effect is important for the max phenotype (Bennett et al., 2006). Furthermore, in pea, bud auxin export and growth correlated with efflux polarity in the bud, rather than efflux activity (Morris and Johnson, 1990), suggesting that PIN4 overexpression would not drive increased export of auxin from buds.

In this study, we found that wild-type shoots bearing a narrow max2 sector, limited to the central region of a single node, produced a phenotypically max2 bud at that node. It is likely that in the main stem of these shoots, a segment of mutant cells was present at the node with wild-type cells above and below them in the auxin transporting cell files of the stem. Although it would not result in an overall increase in auxin transport capacity along the shoot axis, this arrangement certainly does not exclude the stem as the site of action. For example, a small group of cells in the stem with high auxin transport capacity might be sufficient to allow the canalization of an auxin export pathway from the bud by creating a local sink, in which auxin supply from above does not fill the capacity available. However, our data do not exclude the possibility that MAX2 acts in the bud to regulate its growth. To determine the relative contributions of these two possible sites of action, mosaic analysis could be exploited further by choosing a method that allows the generation of a wild-type bud on a max2 shoot axis.

The MAX2–GUS fusion under the control of the MAX2 promoter showed highest activity in the vasculature, and was detected in both xylem- and phloem-associated cells. Although it is not clear how closely the construct reports MAX2 promoter activity, the pattern of GUS activity we observed resembles that described for a MAX1 promoter–reporter fusion (Booker et al., 2005), and the presence of MAX1 and MAX2 in xylem parenchyma is consistent with the proposed function of the MAX pathway as a regulator of auxin transport. The reporter constructs also indicate that MAX1 and MAX2 are expressed in the phloem, where their role is unclear, as no phloem mutant phenotype is known in the max mutants.

The mode of action of MAX2

We show that myc-tagged MAX2 interacts, in planta, with two core components of SCF-type ubiquitin E3 ligases, ASK1 and AtCUL1. This provides further evidence that an SCFMAX2 recruits protein(s) for ubiquitination. This widespread post-translational modification has a well-established role in targeting substrate proteins for degradation by the 26S proteasome (Ciechanover et al., 2000). Alternative functions include the regulation of substrate protein localization or activity (Haglund and Dikic, 2005; Sun and Chen, 2004).

Consistent with yeast two-hybrid and in vitro experiments, in which the interaction of ORE9/MAX2 with ASK1 was found to be F-box-dependent (Woo et al., 2001), we found that overexpression of MAX2 rescued the max2 mutant phenotype, whilst overexpression of ΔF–MAX2 did not. These experiments indicate that the F-box domain is required for MAX2 function. Expression of ΔF–MAX2 in the wild-type background indicates that the remaining C-terminal leucine-rich repeat region of MAX2 might function as a substrate-binding domain. We observed an increase in branching in some lines expressing ΔF–MAX2, compared with untransformed wild type. Competition of ΔF–MAX2 and endogenous functional MAX2 for the substrate could lead to inefficient substrate degradation and increased substrate activity. In three independent semiquantitative RT-PCR experiments, the dominant-negative effect on branching was observed in three lines with an intermediate ΔF–MAX2 transcript level, but not in two more strongly expressing lines. This type of dose dependence could occur if, at high expression level, ΔF–MAX2 either became unstable or if it aggregated with its substrate, preventing substrate activity. To examine whether the increased branching resulted from co-suppression of the endogenous, functional MAX2, induced by the ΔF–MAX2 transgene, we monitored endogenous MAX2 expression in the ΔF–MAX2 lines. Endogenous MAX2 transcript levels in the three intermediate ΔF–MAX2-expressing lines were slightly lower than in the two high ΔF–MAX2-expressing lines. However, they were not greatly reduced compared with untransformed wild type, so it seems unlikely that the observed phenotypic effect is due to co-suppression only. Nonetheless, it is possible that the slight reduction in endogenous MAX2 level was significant in the presence of negatively acting ΔF–MAX2, and that this led to the expression of a phenotypic effect.

The activity of many well-studied SCFs involved in ubiquitin-mediated proteolysis is regulated at the level of the F-box-protein–substrate interaction, because efficient binding requires a modification of either substrate (Cardozo and Pagano, 2004; Ciechanover et al., 2000) or F-box protein (Dharmasiri et al., 2005; Kepinski and Leyser, 2005). In this way, the signalling pathway that triggers the modification controls the abundance of the SCF substrate, which is often a key regulatory protein such as a transcription factor. Our data indicate that the MAX signal is required for efficient branching control by the SCFMAX2. We found that MAX2 overexpression in a wild-type background, i.e. in the presence of the MAX signal, did not affect branching. max1, max3 and max4, which lack the signal, were partially, yet inefficiently, rescued by overexpressing MAX2. These results are consistent with a model in which the MAX signal enhances the interaction of MAX2 with its substrate. At least the max3 and max4 alleles used in this experiment were probably null. Therefore, residual MAX signal activity did not contribute to their partial rescue by MAX2 overexpression. The fact that some rescue occurred in these mutant backgrounds might be due to a weak interaction of MAX2 and its substrate in the absence of MAX signal. Such weak interaction might allow sufficient substrate ubiquitination in some processes, other than branching control, which require the SCFMAX2. This provides one possible explanation as to why the phenotypes of max1, max3 and max4 are less pleiotropic than that of max2. Many well-studied F-box proteins have several substrates (Cardozo and Pagano, 2004; Willems et al., 2004). Therefore, an alternative explanation is that the SCFMAX2 has substrates that are not involved in branching control, whose ubiquitination is not regulated by the MAX1/3/4-dependent signal.

In many cell types expressing a translational MAX2–GUS fusion, reporter activity appeared to be nuclear-localized, and we confirmed the co-localization of GUS activity and nuclear staining for hypocotyl cells. Thus, MAX2 may target nuclear protein(s) for degradation, such as transcriptional regulator(s). Candidate transcriptional targets of the MAX pathway are auxin transporter genes (Bennett et al., 2006; Lazar and Goodman, 2006). To better understand the downstream responses to the MAX signal we will need to identify the substrate(s) of SCFORE9/MAX2, and the search for these substrates is now a priority in the analysis of MAX pathway function.

Experimental procedures

Plants and growth conditions

One representative mutant allele at each MAX locus was used in this study: max1-1 (Booker et al., 2005), max2-1 (Stirnberg et al., 2002), max3-9 (Booker et al., 2004) and max4-1 (Sorefan et al., 2003). Allele numbers are omitted in the text. All max alleles were generated in, or backcrossed into (max1), the ecotype Columbia-0, which was used as the wild-type control. In general, conditions for the growth of Arabidopsis on sterile plates and on soil, in 16-h photoperiods, were as described by Bennett et al. (2006). If other conditions were used, they are outlined in the paragraphs below.

Constructs and plant transformation

Polymerase chain reactions for cloning were performed using Pfu DNA polymerase and PCR-derived clones confirmed by sequencing. Primer sequences are listed in Table 2. Plasmids were purified from Escherichia coli and electroporated into Agrobacterium tumefaciens GV3101. A transformed clone was selected to transform plants by floral dipping (Clough and Bent, 1998).

Table 2.   Primers used in this study
  1. Bold letters indicate point mutations introduced to create restriction sites.

Cloning primers
RT-PCR primers

Fusion of the CaMV 35S promoter with the N-terminal part of MAX2 (35S::NT-M2).

The 35S promoter sequence was amplified from a 4327-bp fragment purified from SphI-digested vector pCAMBIA1303 (GenBank accession AF234299) with primers 35S-F and 35S-R. Point mutations in 35S-F created HindIII/AvrII sites upstream of the promoter. The N-terminal part of the MAX2 sequence was amplified with primers MAX2-A and MAX2-B from a genomic MAX2 clone in pCAMBIA2300 (clone a, a subclone of BAC F14N22 in pCAMBIA2300; Stirnberg et al., 2002). Point mutations in MAX2-A created a SalI site between 35S promoter and MAX2. An equimolar mixture of these overlapping PCR products was used as a template with primers 35S-F and MAX2-B to generate the 35S::NT–M2 fusion. The product was digested with HindIII and XbaI and ligated into plasmids pBin19 and pBluescript, generating pBin(35S::NT–M2) and pBlue(35S::NT–M2).

Fusion of the C-terminal part of MAX2 with the Nopaline synthase (Nos) terminator (CT-M2::Nos).

The C-terminal MAX2 sequence was amplified with primers MAX2-C and MAX2-D from the genomic MAX2 clone described above. Point mutations in MAX2-D generated ApaI and PstI sites between the MAX2 coding region and the Nos terminator, which was amplified with primers Nos-F and Nos-R from the SphI-digested pCAMBIA1303 fragment described above. Point mutations in Nos-R created a KpnI site downstream of the Nos terminator. An equimolar mixture of these overlapping PCR products was the template to generate the CT-M2::Nos fusion with primers MAX2-C and Nos-R. It was digested with XbaI and KpnI and cloned into pBluescript, generating pBlue(CT-M2::Nos).


The CT-M2::Nos fragment was cut out from pBlue(CT-M2::Nos) with XbaI and KpnI and ligated into pBin(35S::NT–M2) opened with the same enzymes. In the resulting clone, pBin(35S::M2), the complete MAX2 coding sequence is reconstructed by linking N-terminal and C-terminal parts at the unique XbaI site.


The bases encoding amino acids 9–47 of the MAX2 protein were deleted. Using pBlue(35S::NT–M2) as a template, two PCR products upstream and downstream of the F-box were generated, one with FDEL1 and FDEL2; the other with FDEL3 and FDEL4. An equimolar ratio of the overlapping PCR products provided the templates to fuse these sequences in a PCR with primers FDEL1 and FDEL4. The product was digested with EcoRV and NcoI and used to replace the F-box-containing fragment in pBlue(35S::NT–M2). The insert of the resulting clone, pBlue(35S::ΔF–NT–M2), was digested out with HindIII and XbaI and transferred into pBin19 [pBin(35S::ΔF–NT–M2)]. The MAX2 C-terminus and Nos terminator were added from pBlue(CT–M2::Nos) as for 35S::M2 to generate pBin(35S::ΔF–M2).


The MAX2 stop codon was replaced by a ScaI site for c-myc cassette insertion by generating a PCR product with primers MAX2-Nco and MAX2-Sca using pBlue(CT-M2::Nos) as a template. The product was digested with NcoI and PstI to replace the original MAX2 3′-end in the template clone. The resulting clone, pBlue(CTM2ΔSTOP::Nos), was digested with XbaI and KpnI, and the insert was transferred into pBin(35S::NT–M2) digested with the same enzymes to generate pBin(35S::M2ΔSTOP::Nos). A c-myc epitope cassette was amplified from ABRC clone CD3-128 [six c-myc epitopes in pGem 7z f(+)] with primers MYC-F and MYC-R. The product was digested with DraI and SmaI and ligated into pBin(35S::M2ΔSTOP::Nos) opened with ScaI. A clone with the myc-cassette in the correct orientation was selected: pBin(35S::M2-myc).

MAX2promoter::MAX2–GUS (M2p::M2–GUS): full-length MAX2 translational fusion to GUS under the control of the MAX2 promoter.

The MAX2 coding sequence was cut out from pBin(35S::M2ΔSTOP::Nos) with SalI and ScaI and cloned into pBI101.3 opened with SalI and SmaI, upstream of and in frame with the GUS reporter sequence, with the sequence linking MAX2 and the GUS coding region predicted to encode SGYGQSLM. The resulting clone was cut with SalI, and after filling the overhang, cut with XhoI, for insertion of a SmaI/XhoI fragment spanning 3450 bp of MAX2 promoter sequence plus MAX2 5′-coding sequence up to the unique XhoI site, which was excised from MAX2 genomic clone a (Stirnberg et al., 2002).

MAX2promoter::MAX2(1–26)–GUS (M2p::M2(1–26)–GUS): MAX2 amino acids 1–26 translationally fused to GUS under the control of the MAX2 promoter.

A MAX2(1–26)–GUS–GFP translational fusion was originally constructed in pCAMBIA1303. A 3450-bp MAX2 promoter plus a short stretch of N-terminal MAX2 sequence were cut out from MAX2 genomic clone a (Stirnberg et al., 2002) with BamHI and SacI, and cloned into pCAMBIA1303 opened with BamHI and SpeI using an adaptor. This placed amino acids 1–26 of MAX2 in frame with GUS encoded by the vector via a stretch of sequence predicted to encode amino acids MVDLTS. To obtain a control for the full-length MAX2 translational fusion in pBI101.3, a fragment containing MAX2 promoter, N-terminal MAX2 plus the GUS N-terminal sequence was cut out from the construct in pCAMBIA1303 with SmaI and SnaBI, and cloned into pBI101 cut with the same enzymes, upstream of and in frame with the GUS C-terminus.

Semiquantitative RT-PCR

Total RNA was extracted using TRIzol reagent according to the manufacturer’s instructions (Invitrogen, For tissue where co-precipitation of polysaccharides with the RNA occurred, either high-salt RNA precipitation was carried out (siliques, leaf samples from branching assays) or the partially resuspended sample was purified using RNeasy (Qiagen, columns (seedling roots and shoots). Following DNaseI digestion (RQ1, Promega,, total RNA was reverse-transcribed with Superscript II and oligo-dT primer according to the manufacturer’s protocol (Invitrogen). As MAX2 lacks introns, three RNA samples of each experiment were randomly selected and also processed without adding reverse transcriptase. For semiquantitative RT-PCR reactions, cDNA aliquots corresponding to 50 ng total RNA were used. Aliquots loaded onto the gel varied between 1/7 and 1/9 of the PCR reaction volume for different experiments. Reaction aliquots were taken at intervals over a range of cycles and run on the gel to select a cycle number where the plateau phase was not yet reached. Primers designed to quantify endogenous and transgenic MAX2 expression either separately (M2-for and either E-M2-rev or T-M2-rev) or non-selectively (NS-M2-for and NS-M2-rev) are listed in Table 2. ACTIN2 was used as a normalization control. Minus reverse transcriptase and water controls were negative in all the RT-PCR experiments presented.

Decapitation assay

Following incubation in water for 3 days at 4°C, seeds were sown onto soil in 35-cm2 pots (three seeds per pot later thinned to one) and grown at 20°C/17°C in 8-h light/16-h dark photoperiods at a light intensity of 120 μmol cm−2 sec−1. After 30 days, plants were shifted to 16-h light photoperiods (same light intensity) to induce flowering. Pots were fertilized regularly with mineral nutrient solution (Lincoln et al., 1990) until bolting. The primary inflorescence of each plant was removed when it was 2–4 cm long to encourage branching from the rosette. The number of rosette branches with a length of at least 2 cm was determined 10 days later. Tissue for RT-PCR analysis was harvested after branching of all individuals had been scored. One rosette leaf per individual was taken and pooled for each genotype. Sample size varied between 9 and 11 for different experiments.

GUS staining and histology

Samples from soil-grown plants were immersed in heptane for 10 min prior to GUS staining. Tissue was incubated in GUS buffer (0.5 mg ml−1 5-bromo-4-chloro-3-indolyl-β-glucuronide, 50 mm potassium phosphate pH 7, 0.1 mm potassium ferrocyanide, 0.1 mm potassium ferricyanide, 10 mm EDTA, 0.1% Triton X-100) at 37°C overnight. For direct examination, tissue was cleared in 70% ethanol. For sectioning, tissue was fixed in 4% formaldehyde/1% glutaraldehyde in 50 mm potassium phosphate buffer (pH 7) for 3 h (roots) or overnight (stems), dehydrated in an ethanol series and embedded in Technovit 7100 (Heraeus–Kulzer, Six-μm sections were fixed to slides and either mounted in water (roots), or counterstained with 0.05% Ruthenium Red and mounted in DPX (stems). For the nuclear localization experiment, sterilized, cold-treated seed on agar plates was incubated in the light for 1 day to induce germination, followed by 4 days in the dark. Five-mm segments were cut from the middle region of the hypocotyl, and fixed for 15 min under vacuum in buffer (100 mm potassium phosphate pH 7, 1 mm EDTA) containing 2% formaldehyde. After a buffer wash, segments were incubated at 37°C in buffer containing the DNA-specific dye DAPI at 0.5 μg ml−1 and GUS substrate and potassium ferro- and ferricyanide at the concentrations listed above. Samples were taken after 4–7 h of staining and were observed under a microscope with brightfield and epifluorescence optics.

Sector analysis

max2 (Columbia-0 background) was crossed with the ch-42 T-DNA insertion mutant that had been generated in the Columbia background (Koncz et al., 1990) and backcrossed into the Landsberg erecta background (Furner et al., 1996). A max2 ch-42 double mutant line that did not segregate for the erecta phenotype was selected in F2. This was crossed to CAUT lines A24 and 7F, in which a wild-type CH-42 transgene is inserted on the bottom arm of chromosome II, in a homozygous ch-42 mutant background. Map distances between MAX2 and the CH-42 transgene of 0.18 cM for A24 and of 0.34 cM for 7F were calculated from phenotypic segregation ratios in the resulting Fs. Selecting for max2 in creating the ch-42 max2 double mutant must result in a Columbia DNA island around the MAX2 locus, while the CAUT lines are in the ch-42 backcrossed to Landsberg erecta background. This allowed us to map the CH-42 insertion relative to MAX2 by genotyping DNA of recombinant F2 individuals from ch-42 max2 × A24 and ch-42 max2 × 7F crosses for Columbia/Landsberg polymorphisms around the MAX2 locus. CH-42 is centromere distal to MAX2 in A24 and centromere proximal in 7F. This is in broad agreement with the sequence data on MAX2 and the flanking sequence of the 7F insert (data not shown).

To generate shoots with marked max2 sectors, 80 mg of F2 seed each from the crosses of ch-42 max2 with A24 and with 7F were irradiated with X-rays at a dose of 16 krad as previously described (Furner and Pumfrey, 1992). Seeds were sown evenly onto soil and grown in 8-h photoperiods. After 4 weeks, individuals with yellow sectors, as well as green MAX2, and yellow max2 controls were selected and transplanted to 16-cm2 pots. Their axillary bud development was scored after 9–10 weeks. Plants were then shifted to 16-h photoperiods and decapitated to encourage formation of inflorescence branches from the sector. F3 sector progeny was grown to confirm the sector genotype. A pdf outlining the theory and practice of the use of the CAUT lines is available from NASC. CAUT lines (stock numbers N5800–N5875) and the parental ch-42 mutant (stock number N5876) can also be ordered online from NASC (


Vapour-sterilized seeds ( were incubated in water for 5 days at 4°C and then germinated in mineral nutrient solution (Lincoln et al., 1990) containing 10 g l−1 sucrose for 7 days at 21°C with slow shaking. Seedlings were blotted, frozen in liquid nitrogen and ground to a fine powder. For protein extraction, the powder was thawed in an equal volume of ice-cold IP buffer (50 mm TRIS–HCl pH 7.5, 150 mm NaCl, 0.5% NP-40, 1 mm PMSF, 1 μg ml−1 pepstatin A, 1 μg ml−1 leupeptin and 1 μg ml−1 aprotinin; TRIS = 2-amino-2-(hydroxymethyl)-1,3-propanediol; PMSF = phenylmethylsulphonylfluoride; NP-40 = Nonidet P40), vortexed, and spun at 14 000 g for 30 min at 4°C. The supernatant was recovered and passed through a 0.2-μm filter, and its protein concentration adjusted to 2 mg ml−1 with IP buffer. For immunoprecipitation, 25 μl bed volume of protein G-agarose beads (Kirkegaard & Perry Laboratories, Inc., were washed three times with 1 ml of IP buffer. The last wash was removed to leave about 100 μl of beads and buffer, and 1 μl of anti-c-myc antibody 9E10 (Covance, was added. After 4 h of rotation at 4°C, 2 mg of protein extract was added and rotation continued for 2 h. Beads were then washed three times with IP buffer for 20 min. The last wash was removed completely using a plastic pipette tip heated and pulled out to a fine end. Beads were eluted by adding 50 μl of 0.1 m glycine·HCl pH 2.5 and rotating for 10 min. The eluate was collected using a pipette tip pulled out to a fine end, neutralized with 5 μl of 1 m TRIS–HCl pH 8, and 18 μl 4 ×  protein gel loading dye added. Running of NuPAGE Novex 4–12% Bis-Tris protein gels (Invitrogen), blotting onto Invitrolon PVDF membrane (Invitrogen), and probing with primary antibodies and peroxidase-conjugated secondary antibody followed by ECL Plus chemiluminescent detection (Amersham Biosciences, were carried out according to the manufacturers’ instructions.


We would like to acknowledge Mazhar Sheikh, Saleha Bahkt, Bushra Mirza, Carley Pullen and Louise Ellis for work making the CAUT lines, and Sean May and Emma Wigmore at NASC for help with setting up the CAUT line PDF and ordering. We thank Mark Estelle and Sunethra Dharmasiri for anti-ASK1 and anti-AtCUL1 antisera and advice on Western analysis, Stefan Kepinski and Tobias Sieberer for helpful discussions on immunoprecipitation, and the York University horticultural team for excellent plant care. The myc epitope cassette clone was provided by the ABRC. This work was funded by grants from the Biotechnology and Biological Sciences Research Council.