The strigolactone-insensitive max2-1 mutant produces an excessive number of inflorescence branches from rosette leaf axils 54 . To identify novel regulators of shoot branching, we performed a suppressor screen in this genetic background. In one of the isolates, 6-7, a recessive, second-site mutation, significantly reduced rosette branching. In addition, 6-7 shoots were slightly taller than max2-1 and their primary inflorescences had a slightly higher number of cauline, leaf-bearing nodes (Figure 1a, b). We temporarily named the suppressor mutation in this line 6 7. After backcrossing 6-7 to wild-type Columbia, these traits were also detected in the wild-type MAX2 background, although the effect on branching was less striking, and could not be readily used to map the suppressor. A pointed juvenile leaf phenotype that co-segregated with the branching habit was instead used (Figure 1c, d). 6 7 was crossed to Landsberg-erecta, and the locus was mapped to a 126 kb region on chromosome 5 by assessing co-segregation of DNA polymorphisms between Landsberg and Columbia in mutant individuals from the F₂ of this cross. JAtY TAC library clones in pYLTAC17 57 containing large wild-type genomic inserts from the mapping interval were transformed into the mutant and assessed for rescue. This defined six candidate genes, whose coding regions were amplified from 6-7 and sequenced (Figure 1e). The sole divergence from wild type was a G to A transition, which introduced a premature termination codon in At5g41520 (RPS10B), one of three Arabidopsis genes encoding cytoplasmic ribosomal protein S10e. RPS10B transcript level was lower in 6-7 than in the wild type (Figure 1f), suggesting nonsense-mediated decay. Identity of RPS10B as the suppressor gene was confirmed by mutant rescue with a wild-type RPS10B genomic construct (Additional file 1: Table S1), and the mutant allele was named rps10b-1.

With wild-type Columbia plants grown in long photoperiods, floral transition is the trigger for axillary shoot initiation. The axillary shoots activate to form inflorescence branches in an apical-basal wave, i.e. from the cauline leaf axils, situated along the primary inflorescence, towards the rosette leaf axils 58 . In the wild type, only a few of the topmost rosette leaf axils produce branches, while more basal rosette axils carry arrested buds. In max2-1, neither the timing of axillary shoot initiation nor the outgrowth sequence is altered, but nearly all the rosette axils produce a branch 54 .

The rps10b-1 mutation caused a reduction in axillary shoot size at equivalent nodal positions in the rosettes of both MAX2 and max2-1 plants (Figure 2a–d). In addition, one or two axils at the top of the rosette often appeared to be empty. A small proportion of the rps10b-1 cauline leaf axils were also empty (Figure 2e–g, Table 1), and remained so until maturity. This indicates that rps10b-1 affects axillary shoot initiation. Either a delay in axillary shoot formation, or an additional effect on axillary bud growth rate, might cause the reduced size of rps10b-1 buds.

^aThe cauline vegetative nodes along the primary inflorescence of 38–40 plants per genotype were scored by the naked eye.

To quantify these phenotypes, we examined flowering plants under a dissecting microscope and assessed axillary shoot development at consecutive nodal positions throughout the rosette. Four developmental stages were defined, and the proportions of rosette axils at each stage were calculated for ten individual plants per genotype (average proportions ± SEM shown in Figure 2h). The stages were defined as follows: 1. Branches (inflorescence length above 3 mm), 2. Big axillary buds whose inflorescence had not yet significantly elongated. 3. Small buds with leaf primordia clearly visible but shorter than 2 mm. 4. Apparently empty axils lacking visible axillary leaf primordia (it was not possible at the magnification used to determine whether an axillary meristem had been initiated or not). The frequency of class 4 was negligible in both wild-type and max2-1 rosettes, but these genotypes differed with respect to the proportions of the three more advanced classes. Compared to the wild type, max2-1 showed a dramatic increase in the most advanced class, balanced by a decrease of the two intermediate classes. In contrast, for rps10b-1 in both the MAX2 and max2-1 backgrounds, the proportion occupied by the most advanced class decreased, and this was balanced by an increase in the proportion of apparently empty axils, with little change in the intermediate classes. These results indicate that RPS10B promotes axillary shoot development from an early stage, including both axillary bud formation and possibly subsequent bud growth. In contrast, MAX2 represses only the later stages of bud activity 54 , suggesting that RPS10B acts at least in part independently of MAX2.

To assess whether rps10b-1 affects axillary shoot growth independently from initiation, we studied the outgrowth kinetics of axillary inflorescences on isolated cauline nodes. Nodal explants, consisting of a cauline axillary bud smaller than 2 mm and 5–7 mm of the primary inflorescence stem above and below the node, were inserted between two agar slabs in a Petri dish (as described in 59 ). The length of the axillary buds was monitored over a 10 day period. rps10b-1 caused a slight delay in inflorescence outgrowth in both MAX2 and max2-1 backgrounds (Figure 3a, solid lines).

In addition to increased branching, the max2-1 mutant has a range of phenotypes associated with its strigolactone insensitivity. These include an elongated hypocotyl and overexpression of the strigolactone biosynthetic genes CAROTENOID CLEAVAGE DIOXYGENASE7 (CCD7) and CCD8, which are feedback-downregulated by strigolactone signalling 52 55 56 . In a hypocotyl growth inhibition assay, rps10b-1 did not suppress the strigolactone insensitivity of max2-1 (Additional file 2: Figure S1). Furthermore, rps10b-1 did not affect levels of CCD7 or CCD8 transcript characteristic of the MAX2- or max2-1-backgrounds (Additional file 3: Figure S2). Therefore, the suppression of max2-1 by rps10b-1 is specific to axillary shoot growth and does not involve a global restoration of strigolactone responsiveness.

Auxin has been implicated in both axillary meristem initiation and outgrowth. Furthermore, max2-1, in common with other strigolatone mutants, displays a number of auxin-related phenotypes, which led to the hypothesis that strigolactones act by restricting polar auxin transport. We therefore assessed the effect of rps10b-1 on these auxin-related phenotypes. The outgrowth of wild-type buds is strongly delayed by apical supply of the synthetic auxin naphthalene acetic acid (NAA), but max2-1 axillary buds are resistant to this auxin effect 55 59 60 (Figure 3a). In the wild-type background, rps10b-1 delayed outgrowth only slightly, similar to its effect in the absence of auxin. However, in combination with max2-1, rps10b-1 substantially delayed bud outgrowth, such that the outgrowth of double mutant buds on auxin-treated explants was nearly identical to wild-type buds. Thus, rps10b-1 suppresses max2-1 bud auxin resistance.

A second auxin-related phenotype of max2-1 is an increase in basipetal transport of radiolabeled auxin through primary inflorescence stem segments 55 60 . We found that rps10b-1 did not affect this phenotype (Figure 3b). Rather, the mutation slightly increased the amount of auxin transported in both MAX2 and max2-1 backgrounds.

Third, the auxin response reporter construct DR5::GUS 61 has increased activity in the main shoot axis of max2-1 plants 60 , associated with increased amounts of auxin moving in the PATS 44 . We found that this increase in DR5::GUS expression was partially suppressed in rps10b-1 max2-1. This effect was observed in hypocotyls from 2-week old seedlings (Figure 3c–f) as well as hypocotyls from 9-week-old short-day grown plants, which had undergone secondary thickening (Figure 3g–j). In the MAX2 background, rps10b-1 had little effect, with xylem-associated DR5::GUS activity possibly slightly increased. These differences in reporter activity do not simply reflect differences in bud activity, because the 2-week-old seedlings had not yet initiated axillary buds. In summary, rps10b-1 partially rescued some of the auxin-related phenotypes of max2-1, indicating that RPS10B may act by modulating auxin responsiveness or homeostasis.

To learn more about the mode of RPS10B action we assessed its genetic interactions with other known shoot branching regulatory genes. First, the effect of the rps10b-1 mutation in high-branching mutant backgrounds other than max2-1 was assessed (Figure 4a). As expected, the strigolactone biosynthetic mutant max4-1 (ccd8) 62 , was partially suppressed. brc1-2 and brc2-1 are loss of function alleles of bud-specific class II TCP transcription factor genes 28 . As with max2, excessive branching of brc1-2 is strigolactone-insensitive 53 . brc1-2 and the brc1-2 brc2-1 double mutant were also partially suppressed by rps10b-1. In all tested double and triple mutant combinations of rps10b-1 with max4, brc1 or brc2, empty axils were present at apical nodes in the rosette at maturity. Thus, as with max2, at least part of the suppression by rps10b-1 in these backgrounds resulted from a defective or delayed axillary shoot formation.

Perception of auxin by the TIR1/AFB auxin receptors triggers the ubiquitin-mediated degradation of Aux/IAA proteins, which are repressors of the AUXIN RESPONSE FACTOR (ARF) transcriptional regulators 63 . This degradation requires the AUXIN-RESISTANT1 (AXR1) protein 63 64 . Mutation of AXR1 has little effect on bud initiation, but results in increased and auxin resistant bud outgrowth 65 66 . In combination with rps10b-1, the axr1-3 mutant allele surprisingly enhanced the suppression of axillary bud development at both apical rosette (Figure 4b–e) and cauline nodes. In some experiments, as shown in Figure 4b–e, buds in these positions were considerably smaller than those of either single mutant. In other experiments, a large proportion of cauline and apical rosette axils appeared completely empty in rps10b-1 axr1-3 plants. In addition, the primary inflorescence meristem of rps10b-1 axr1-3 plants frequently aborted. Between 20% and 50% of the double mutant individuals, but neither of the single mutants, had this phenotype. These observations suggest that RPS10B and AXR1 interact to promote shoot meristem development. However, this interaction appeared to be positionally restricted. At more basal rosette nodes of double mutant plants, bud behaviour resembled the axr1single mutant; axillary buds initiated and formed inflorescence branches, such that rosette branch numbers of axr1-3 and rps10b-1 axr1-3 at maturity did not differ significantly (Figure 4a).

Mutation of AMP1, which encodes a putative carboxypeptidase with unknown molecular function, causes a range of phenotypes related to shoot meristem function including constitutive axillary bud activation, increased shoot meristem size, increased rate of leaf initiation, and increased cytokinin content 67 68 69 70 . The defective axillary bud formation in apical rosette nodes typical of rps10b-1 was completely suppressed in an amp1-1 background (Figure 4b, c, f–h), and at maturity, the average branch number of rps10b-1 amp1-1 plants did not differ significantly from amp1-1 plants. Genetic analysis by Vidaurre and coworkers 71 suggests a major function of ARF-mediated auxin signalling in embryogenic shoot meristem formation and vascularisation might be the downregulation of AMP1 activity. In the light of this finding, the genetic interaction with amp1-1 further supports the idea that reduced ARF-mediated auxin signalling is involved in the rps10b-1 meristematic phenotypes.

We also analysed the effect of rps10b-1 in genetic backgrounds characterised by reduced branching. First, we constructed a double mutant of rps10b-1 with another non-allelic max2-1 suppressor from our screen, far-red elongated hypocotyl3-12 (fhy3-12). This is a loss-of function allele of the transcriptional activator FHY3 72 . This mutation suppresses max2-1 by reducing bud activation, with negligible effects on axillary shoot formation; and our data suggest that auxin might be central to its branching phenotype 73 . rps10b-1 fhy3-12 double mutant plants showed a near-complete loss of rosette axillary buds (Figure 5a–d). Furthermore, the primary inflorescence meristem of double mutant plants often aborted during the reproductive phase, a phenotype not observed with either single mutant (Figure 5g). The frequency of abortion ranged from 30% to 90% in different experiments.

As described earlier, mutation of the HDZIPIII gene REV causes partial defects in axillary meristem formation and floral meristem maintenance. In addition, the HDZIPIII family members redundantly specify adaxial leaf identity, but rev loss-of-function mutant leaves appear normal 7 8 . We generated double mutants between rps10b-1 and a rev T-DNA insertion allele, SALK_102345 (Figure 5f, h). These were highly abnormal. Successive leaves became increasingly needle-like, and axillary shoots were absent. The primary stem was short, pin-like and lacked flowers. Thus, rps10b-1 strongly enhanced the loss of REV function with respect to both leaf polarity and axillary shoot formation. The F₂ analysis also revealed that a single copy of the rev mutant allele strongly enhanced the axillary shoot phenotypes in the rps10b-1 mutant background, while rps10b-1/+ rev/+ axillary shoot development was normal (Figure 6a–e). rps10b-1 rev/+ plants had normal stature and slight defects in floral meristem maintenance. Their leaf polarity appeared largely normal, except that a few leaves had reduced lamina, from which the midvein separated as an abaxial outgrowth at the distal end of the leaf (Figure 6f, g). The strongest effect of REV haploinsufficiency concerned axillary shoot formation. Nearly all the rosette and a substantial proportion of cauline leaf axils were empty (Figure 6d, e). This demonstrates a dosage dependence of REV in the rps10b-1 background, which is not seen in the wild-type RPS10B background, where rev appeared recessive.

The PID protein kinase is required for dynamic changes in plasma-membrane localisation of PIN auxin transporters and thereby auxin transport direction 32 33 34 74 . Plants homozygous for strong pid mutant alleles are defective in flower formation, and the few flowers produced are abnormal and sterile 75 . We crossed rps10b-1 with a pid-14 heterozygote (SALK_049736 34 76 ) and homozygous double mutants were identified in segregating rps10b-1 pid-14/+ F₃ families (Figure 6h–k). In addition to the defect in flower formation, which has been described, homozygous pid-14 segregants from RPS10B pid-14/+ control F₃ families showed mild defects in cauline and axillary bud initiation similar to the rps10b-1 single mutant. pid-14 heterozygotes from the control F₃ were indistinguishable from PID segregants and wild-type controls. The double mutants segregating in the progeny of rps10b-1 pid-14/+ plants had a more severe phenotype than pid-14 alone, as neither cauline leaves nor branches, nor flowers were produced on the primary inflorescence, and the proportion of empty rosette axils was increased. Furthermore rps10b-1 pid-14/+ F₃ individuals also showed slightly enhanced axillary shoot defects when compared with rps10b-1 PID F₃ segregants or rps10b-1 controls. The proportions of empty cauline and rosette axils were increased (Figure 6k). Although less striking than with REV, there is a PID dosage effect in the rps10b-1 background, demonstrating that partial loss of this r-protein increases sensitivity to reduced function of both PID and REV.

As described above, rps10b-1caused failure of the primary shoot meristem or of floral meristems in some mutant backgrounds. This could point to a more general role of RPS10B in supporting shoot meristematic function, which is also indicated by other weakly penetrant traits observed with the rps10b-1 single mutant. In rps10b-1 flowers, the number, identity and separation of lateral organs were affected (Figure 7). Sepal, petal, stamen and carpel numbers were more variable than in the wild type (Table 2). A substantial proportion of rps10b-1 flowers lacked one stamen, while petal and carpel numbers were more often increased than decreased (Figure 7a–d). Fusion between organs in one whorl was sometimes detected, most frequently for the stamens (Figure 7e). Furthermore, some stamens were green and possibly carpelloid (Figure 7f) and/or were partly fused to the gynoeceum (Figure 7g–h).

^a98-100 flowers per genotype were examined under a dissecting microscope.

Furthermore, patterning defects in addition to the lack of axillary shoots were observed at cauline nodes at low frequencies (Figure 8a–c, Table 1). The topmost cauline branches of rps10b-1 were occasionally not subtended by a cauline leaf (Figure 8c), and fusion of cauline leaf lamina to the inflorescence stem was sometimes detected (Figure 8).

Such phenotypes suggest a role of RPS10B in lateral organ partitioning and separation. To test this hypothesis, we studied the genetic interaction between RPS10B with CUC3, one of three NAC transcription factor family members with partially overlapping roles in organ boundary formation. An rps10b-1 cuc3 double mutant was constructed using a T-DNA knockout allele of cuc3 (GABI-KAT line GK_302G09 77 ). With respect to cauline node development (Table 3 and Figure 8a, d, e), cuc3 was nearly indistinguishable from wild type, consistent with previous reports, demonstrating redundancy in the CUC family for cauline node patterning 11 12 . Very rarely, we observed that accessory axillary shoots, which are often formed at Arabidopsis cauline nodes between the axillary branch and its subtending leaf (Figure 2e), were fused with the stem of the axillary branch (Figure 8e), or that a branch was slightly fused with the base of its subtending cauline leaf. In contrast, in the rps10b-1 cuc3 double mutant, the frequency of obvious cauline node patterning defects was greatly enhanced. There was further loss of either the leaf or the axillary shoot, and increased fusion of organs, such that 76% of the double mutant nodes appeared abnormal (Table 3). The increase in the proportion of nodes showing abnormal leaf development (leaf absent or fused to the stem) in the double mutant, compared with rps10b-1 alone, was highly significant (χ² = 113.1, p < 0.0001). This was also the case when the proportions of nodes lacking an axillary shoot were compared (χ² = 72.3, p < 0.0001).

^aA T-DNA insertion allele of cuc3, GABI-KAT line GK_302G09, was used in this study.

^bThe cauline vegetative nodes along the primary inflorescence of 38–40 plants per genotype were scored by the naked eye.

Loss of CUC3 function has been reported mildly to affect embryonic shoot patterning, with cuc3 seedlings falling into two major classes: phenotypically normal, or showing one-sided cotyledon fusion. Occurrence of the severe cup-shaped phenotype caused by two-sided cotyledon fusion is rare 10 11 . This was also true for the cuc3 allele we used (Table 4). rps10b-1single mutant seedlings did not show cotyledon fusion but rarely, an extra cotyledon was present. Combining rps10b-1 and cuc3 doubled the proportion of seedlings showing cotyledon fusion (13.9%, compared to 7.7% for cuc3 alone, χ² = 6.77, p = 0.01). It also increased the proportion of seedlings showing severe, two-sided cotyledon fusion, but not significantly (Fisher’s exact test, p = 0.06). Thus, the patterning of cotyledonary nodes appeared less sensitive to combined loss of RPS10B and CUC3 function than the patterning of cauline nodes. Our observations suggest that CUC gene-mediated patterning depends on full RPS10B function, but also that this dependence varies with the developmental context.

^aA T-DNA insertion allele of cuc3, GABI-KAT line GK_302G09, was used in this study.

Arabidopsis r-proteins are encoded by small gene families 78 . Two additional RPS10 family members, RPS10A (At4g25740) and RPS10C (At5g52650), show 78% and 74% amino acid identity with RPS10B. RT-PCR from cDNA produced from total RNA of different wild-type tissues showed that all three genes are transcribed and that their relative contributions to transcript level appear invariant for the tissues we analysed (Additional file 4: Figure S3). The AtProteome database 79 points to RPS10B as the most abundant protein isoform. To test for redundancy of protein function, we amplified a cDNA corresponding to the longest annotated protein version for each member, and expressed it under the control of the RPS10B promoter in rps10b-1 plants. As controls we used the wild type, rps10b-1, T₁ plants from transformation of the mutant with the genomic RPS10B construct, and T₂ plants from transformation of the mutant with two JAtY TAC clones, only one of which contained the RPS10B genomic region. Complementation efficiency was scored by counting the stamens of 20 flowers from 8–13 individual plants per genotype or construct (Figure 9). The mean individual stamen numbers ranged between 5.7 and 6 for wild-type plants; but were below 5.4 for the mutant or transformants with the JAtY TAC clone that lacked RPS10B. For the T₂ transformed with the JAtY TAC containing RPS10B, and for 9 out of 10 T₁ transformed with the genomic RPS10B construct, stamen numbers ranged from 5.4 up to the maximum values obtained for wild-type plants. A mean stamen number lower than wild type but still above those of mutant plants may be explained by a lower dose of functional RPS10B in some transformants than in wild type, as the majority of the JAtY T₂ and most of the T₁ are expected to contain one transgene copy. Of the three RPS10B promoter::cDNA fusions, RPS10B::B rescued most efficiently, however with a further reduction compared to the genomic construct, which could indicate a requirement to generate alternative transcripts, or for intronic or untranslated sequences for the proper control of RPS10B gene expression. The RPS10B::C construct complemented the stamen phenotype in about half of the T₁; however, none of the RPS10B::A T₁ was rescued. While the reason for the non-complementation by RPS10A is not clear, the rescue by the RPS10C cDNA argues against a specialised role of RPS10B within the S10e protein family.

RPS10B belongs to the three-member Arabidopsis gene family encoding the eukaryote-specific protein S10e of the small cytoplasmic ribosomal subunit 78 80 . Like most of the r-proteins, S10e is essential for the biogenesis of its ribosomal subunit 81 . It is positioned at the “beak” of the small subunit, a structure that is formed from protein and rRNA in eukaryotes, but exclusively from rRNA in bacteria 82 . The role of S10e in translation is unknown. Crosslinking experiments indicate that S10e might participate in the interaction of the small subunit with eukaryotic initiation factor 3, which functions in translation initiation 83 84 .

The Arabidopsis rps10b-1 mutant allele is transcribed and can encode a truncated protein; its recessive inheritance is consistent with either reduced or abolished protein function. A knockout allele could not be obtained from T-DNA mutant collections. The fact that cDNAs of RPS10B and RPS10C (driven by the RPS10B promoter) rescued an rps10b-1 mutant phenotype to a similar extent, suggests that the RPS10B protein has not functionally diverged from other family members. We detected transcripts of all RPS10 family members in all tissues tested, with highest transcript levels in young, growing tissues, including axillary buds (Additional file 4: Figure S3).

An increasing collection of ribosomal protein mutants have been recovered from screens for developmental phenotypes in Arabidopsis, with substantial overlap in the suite of phenotypes conferred by these mutations. The phenotypes include altered leaf shape (the first leaves are narrow and pointed) and the ability to enhance the phenotype of mutations that affect leaf adaxial identity, for example asymmetric leaves1 (as1) or as2 85 86 87 88 89 90 91 . However, these r-protein mutations differ substantially in their effects on plant growth, which could reflect variation in the degree of genetic redundancy. In rps10b-1, expression of the pointed first leaf phenotype was mild. Leaf polarity was affected in double mutant combination with rev (Figures 6, 7) and we confirmed that this was also the case with as1 (Additional file 5: Figure S4). Although we observed weak effects on growth rate, for example in axillary buds on isolated nodal segments, the shoot or organ size of mature plants was not noticeably reduced, arguing against a general growth defect. The basis of the developmental defects of r-protein mutants is unclear. Two possibilities seem likely.

First, defective ribosomes may trigger specific developmental defects through their participation in surveillance mechanisms at cell cycle checkpoints. For example in humans, redundancy of r-proteins is less common, and haploinsufficiency of S10e and several other proteins of the large or small ribosomal subunit cause Diamond-Blackfan anemia, a syndrome of specific developmental defects including the failure of red blood cell progenitors 92 93 . According to current understanding of the disease, these mutations perturb ribosome biogenesis via an imbalance in ribosome constituent stoichiometry. This is likely to increase the level of unincorporated r-proteins, several of which can bind and inactivate a ubiquitin ligase which targets the p53 tumor suppressor protein 94 95 , and its resulting stabilization triggers cell cycle arrest in red blood cell progenitors. It may be that this surveillance mechanism operates in certain cell types only, for example cells that proliferate very rapidly 96 97 , which could explain the developmental specificity of the phenotype. It is not known whether similar surveillance systems operate in plants.

Second, ribosome insufficiency, the production of disfunctional ribosomes, or the lack of ribosomes containing a specific r-protein variant could affect the production of specific proteins more than others. For example developmental patterning or cell cycle genes might crucially depend on particularly high translation rates or on a specialized ribosome variant. An interesting case here is the Aux/IAA transcriptional repressors, which are central to auxin-regulated gene expression. Some members of this protein family have extremely short half-lives, in the order of 5 minutes 98 , and are maintained at steady state level in cells with a particular auxin concentration. Upon auxin addition, their half-lives are further reduced 64 99 , resulting in their depletion and hence the up-regulation of transcription by a sub-family of ARFs. Because of the need for continuous replenishment of these proteins, it is possible that developmental events dependent on dynamic changes in auxin signaling are particularly sensitive to inefficient ribosomes. Alternatively, the consequences of reduced or altered ribosome function might be enhanced by specific features of the mRNA encoding a protein, for example by the presence of upstream ORFs, which require translation re-initiation. This is the case for the mRNAs of several ARF transcription factors, including ARF3 (ETTIN) and ARF5 (MONOPTEROS, MP), and was proposed to cause arf-like developmental phenotypes of the r-protein mutant short valve1 (rpl24b) 100 . Another ribosome-dependent process, which might potentially be affected is miRNA-directed translational regulation 101 102 . Many of the genes involved in meristem patterning and adaxial identity are regulated by small RNAs 103 104 .

The work presented here is suggestive of this second set of possibilities, because many of the effects we observe are indicative of a general lack of robustness of the adaxial patterning system, with the rps10b-1 mutation rendering the system sensitive to the dosage of other important regulatory components.

Despite the intuitive lack of specificity expected from a ribosomal protein mutation, it is clear that mutation of RPS10B causes a syndrome of phenotypes that can be attributed to patterning events at the shoot apical meristem, and particularly to the establishment of boundaries between the meristem and the leaf, and to a lesser extent, within the leaf.

rps10b-1 suppresses excessive shoot branching in the max2-1 mutant background. A reduced ability to initiate or maintain axillary shoot meristems is a major cause of this suppression. The axillary shoot defects of rps10b-1 were enhanced in double mutant combination with axr1, fhy3, rev, and pid, and were sensitive to reductions in the dose of REV and PID. Moreover, maintenance of the primary shoot apical meristem was partially affected in combination with axr1 and fhy3; a phenotype not observed in the single mutants. Finally, rps10b-1 enhanced the floral meristem defects of rev and pid. This indicates a general role of RPS10B in shoot meristem function. In addition, the rps10b-1 rev double mutant phenotype revealed that RPS10B is involved in leaf polarity, like many other r-protein genes.

While axillary meristem defects have not yet been reported for r-protein gene single mutants (perhaps because they are relatively weak), introgression of piggy1 (rpl10ab) into a rev mutant, stv1 (rpl24b) into an arf3 mutant and rpl4d into an as1 mutant background resulted in striking axillary and/or floral meristem defects 88 100 105 . Formation of the embryonic shoot meristem; and shoot meristem, vascular and leaf patterning crucially depend on an interaction between HDZIPIII and KAN genes 2 19 106 . Furthermore, axillary and embryonic shoot meristem formation are similar in many respects and likely to share the HDZIPIII /KAN patterning mechanism. With respect to leaf patterning, r-protein genes were found to promote genetically the adaxialising role of the HDZIPIII genes, and to antagonise the action of the abaxialising KAN genes 105 107 . RPS10B genetically promoted the action of REV both in shoot meristem function and leaf polarity. Interestingly, axillary meristem formation appeared more sensitive to halving the REV dose in the rps10b-1 background, than did leaf polarity. This supports the notion that RPS10B acts at least partly via meristem establishment itself, and not only via the specification of leaf adaxial fate, which is a prerequisite for axillary meristem initiation in Arabidopsis 5 7 8 16 . Despite the strong genetic interactions between r-protein genes and the HDZIPIII/KAN pathway, further analysis did not implicate any of the ad- or abaxial polarity genes examined as direct targets of ribosomal regulation 88 91 105 107 .

The rps10b-1 mutant displayed other shoot meristem-related phenotypes that were not enhanced in combination with rev. Cauline nodes lacked a leaf, or the leaf was rudimentary. Sometimes, the cauline leaf margin was fused to the stem. Floral organ numbers were more variable than in the wild type and, organ fusion within and between whorls occurred. Such phenotypes indicate misregulated organ separation. Some rps10b-1 phenotypes resemble loss of function, while others resemble gain of function phenotypes described for the three partially redundant CUC genes 9 10 11 12 108 109 110 . Furthermore, combining rps10b-1 and cuc3 enhanced organ separation defects which were rare in the single mutants, most dramatically at cauline nodes. In the r-protein mutant rpl27ac-1d, CUC2 was mislocalised during embryonic shoot meristem formation 91 . Interestingly, the leaf polarity regulators AS1 and AS2 have been implicated in CUC gene regulation and organ boundary formation 111 112 113 . Conversely, mutant phenotypes suggest a role for CUC genes in leaf polarity 11 . This suggests that the well-known ribosomal regulation of leaf polarity, and the organ boundary role we describe here for RPS10B could have a shared molecular basis.

A common feature of many of the developmental events described above is their dependence on, or interaction with auxin and its directed transport. The formation of both the leaf-meristem boundary and the leaf abaxial-adaxial boundary involve the specific and dynamic reorientation of auxin transport paths and hence auxin distribution patterns 14 . Consistent with the importance of auxin in these events, the general reduction in the robustness of the patterning of these boundaries in the rps10b-1 mutant is associated with a range of auxin-related phenotypes.

First, mutations affecting auxin signalling or transport enhanced some of these defects. The auxin signalling mutation axr1-3, which does not affect axillary shoot formation in a wild-type background 66 , enhanced axillary bud loss in combination with rps10b-1. In addition, a novel phenotype of primary inflorescence meristem arrest was displayed by some axr1-3 rps10b-1 double mutant plants. The effects of reduced or abolished function of the auxin transport regulator PID on lateral organ formation were enhanced in the rps10b-1 mutant background. The effect of both these auxin-related mutations may be to interfere with ARF–regulated developmental programmes either globally (axr1) or through altered auxin distribution (pid). Mutation of the transcriptional activator FHY3, another max2 branching suppressor from our screen, very strongly enhanced axillary meristem failure when combined with rps10b-1, and also caused inflorescence meristem arrest. We hypothesise that FHY3 also regulates branching via auxin signalling or homeostasis 73 .

Second, the amp1-1 mutation suppressed the axillary meristem failure of rps10b-1 in the double mutant. Although the molecular function of AMP1 is not known, loss-of-function mutant phenotypes suggest that it restricts shoot meristematic growth 70 114 . Increased levels of cytokinins have been detected in amp1 plants 67 68 , which might cause their increased meristematic stem cell activity 115 . Interestingly, a link between AMP1 and ARF-mediated auxin signaling has recently been proposed. amp1 suppresses the effect of loss of ARF5 (MP) in embryonic shoot meristem development and vascularisation, indicating that one important activity of MP might be to antagonise AMP1 71 . In this way, auxin signalling in the shoot meristem could sustain the stem cell pool required for future lateral organ formation. The genetic interactions of rps10b with axr1, pid, and amp1 are consistent with RPS10B supporting stem cell production indirectly by maintaining ARF-mediated auxin signalling.

In addition to axillary meristem specification defects, which likely underlie the poor axillary shoot formation phenotype of the rps10b-1 mutant, we also detected defects in axillary meristem activity, which may contribute to the suppression of shoot branching in max2. Because of the effects on axillary bud formation, it was difficult to ascertain the effect of rps10b on bud outgrowth in intact plants. Therefore, we used excised cauline nodes, which were selected for approximately equal bud size at the start of the experiment. Except for one specific situation, which is discussed below, the effect of rps10b on bud outgrowth rate was surprisingly small, given the transcriptional evidence for high r-protein synthesis in active buds 25 26 . This could indicate that loss of RPS10B was compensated by functional family members. Mechanisms that ensure that ribosomal components are produced in stochiometric amounts are better studied in other organisms, but they are likely to operate in plants as well 116 117 . We detected at most a slight upregulation of RPS10A or C transcripts in rps10b by semi-quantitative RT-PCR (Figure 1f). However, the example of the Arabidopsis rpl4a and rpl4d mutants shows that compensation at the protein level can occur in the absence of detectable compensation at the transcript level 118 .

The F-box protein MAX2 is required for normal strigolactone responsiveness, and is thought to act in an E3 ubiquitin ligase, selecting unknown protein targets for degradation 119 120 . Strigolactones are negative regulators of PIN protein levels, and of polar auxin transport in the vasculature 55 . Recent studies with excised axillary buds, to which a synthetic strigolactone was applied via the basal internode yielded two interesting observations. First, the ability of strigolactone to inhibit single excised buds required apical auxin; second, if buds on two consecutive nodes were excised, basal strigolactone enhanced the growth differential or competition between them, rather than inhibiting both 55 121 . This fits with a model of bud regulation via auxin transport canalisation, where bud activation requires the export of auxin via a shared auxin transport route in the stem, and strigolactones inhibit this process by restricting PIN protein accumulation 45 . Consistent with this, axillary buds of strigolactone mutants, including max2, are moderately resistant to apically applied auxin 55 60 . Interestingly, we found that the growth-inhibiting effect of rps10b on auxin-treated max2 buds was much stronger than for other genotypes and treatment combinations, such that bud outgrowth kinetics of the auxin-treated double mutant were restored to wild-type. This could indicate that rps10b specifically suppresses a downstream effect of the max2 mutation in bud outgrowth. This effect might be auxin-related, as rps10b specifically suppressed auxin responsive gene expression, as reported by DR5::GUS activity, in the shoot axis of rps10b max2, while it did not have this effect in the MAX2 background. A mode of action different from strigolactone / MAX2 is suggested by the fact that rps10b did not antagonise the effect of max2 on stem polar auxin transport; and the fact that rps10b did not restore the altered shoot vascular architecture of max2 back to wild type (compare sections of older plants in Figure 3g-j). The vasculature of max2 stems shows increased activity of the PIN1::PIN1-GFP reporter 55 60 . In a recent evaluation of the vascular role of the HDZIPIII and KAN genes, both contributed to focused and canalised auxin movement during vascular differentiation; it was proposed that KAN genes act by downregulating PIN activity, and that HDZIPIIIs promote the differentiation of xylem tissues, including the auxin-conducting xylem parenchyma 42 . A relatively subtle change in the HDZIPIII / KAN activity balance characteristic for the r-protein mutants, with lowered HD/ZIPIII or increased KAN activity, might not be critical for bud auxin export and activation in wild type, but might prevent buds of max2 from activating when there is a higher auxin load in the main stem.

Ecotype Col-0 was used as the wild-type control, and unless stated otherwise mutant lines were in this genetic background. The following lines were described previously: amp1-1 67 69 , axr1-3 65 123 , brc1-2 and brc2-1 (SALK_091920 and SALK_023116 28 ), fhy3-12 73 , max2-1 54 , max4-1 62 and pid-14 34 76 . Two lines obtained from T-DNA mutant collections were characterised by sequencing from both T-DNA borders: SALK_102345, an insertion in the last exon of REV (At5g60690) upstream of the termination codon, and GABI-KAT line GK_302G09, an insertion affecting the second exon of CUC3 (At1g76420). Multiple mutants which we constructed were confirmed by genotyping, using wild-type and T-DNA allele-specific PCR for insertional alleles, and CAPS 124 or dCAPS 125 markers for point mutation alleles; except for max4-1, where homozygosity was confirmed by testing progeny for uniform BASTA-resistance. As REV and RPS10B are linked, a reduced frequency of double mutant individuals was expected in the F₂ of the rps10b x rev cross. Therefore, 36 rps10b-1 homozygous F₂ were selected based on their seedling leaf phenotype, genotyped for RPS10B and REV, and their leaf and lateral shoot development was observed. For the cross rps10b x pid-14, genotyping was used in the F₂ to identify rps10b-1 pid-14/+ individuals expected to segregate the double mutant in the F₃, with RPS10B pid-14/+ individuals used as controls. About 40 F₃ progeny each were then genotyped and phenotyped.

Arabidopsis seeds were sown onto Levington F2 compost pretreated with systemic insecticide (Intercept 70WG, Everris Limited, Ipswich, UK). Trays were chilled at 4°C for 3 days and then incubated in a greenhouse with 16-h supplemental lighting. These conditions were used for all soil-grown plants except for the hypocotyls examined by histology (Figure 3). These were from 14-day-old plants grown in continuous low light (40 μmoles m^-2 sec^-1 from white fluorescent tubes, 21°C) and from 60-day-old plants grown in short (8-h) photoperiods (160 μmoles m^-2 sec^-1 from fluorescent white tubes, 21°C day / 17°C night temperature). Except for the mutant screen described below, individual plants were grown at a density of 1 per 16 cm² in trays consisting of 40 x 16 cm² compartments.

max2-1 seeds were mutagenised with 0.3% ethyl methanesulfonate. 18 000 seeds from the resulting M₂ generation were sown at densities of one plant per 3 or 5 cm² and screened for reduced rosette branching at maturity. One of the suppressor mutations isolated, 6-7, was recessive and segregated independently from max2-1 after backcrossing to Columbia wild-type. The suppressor locus was mapped to a 126-kbp interval on Chromosome 5 using about 1600 mutant individuals from the F₂ of a cross between Ler plants and the 6 7 mutant in the MAX2 background. End-sequenced TAC clones from the Arabidopsis wild-type Columbia genomic JAtY library in pYLTAC17 57 with inserts spanning the mapping interval were obtained from the John Innes Genome Laboratory, and transformed into Agrobacterium strain GV3101 for floral dipping of 6-7 MAX2. This was done according to Clough and Bent 126 , except that the infiltration medium contained glucose instead of sucrose. T₁ selected for BASTA-resistance under sterile conditions were further cultivated on soil. Their phenotypic rescue was scored; and they were genotyped to confirm the presence of the left and right vector – genomic insert borders specific to the TAC clone.

Total RNA was extracted using the RNeasy plant miniprep kit with on-column DNaseI digestion (Qiagen, Hilden, Germany) from about 100 mg tissue powder, obtained from 10 pooled 1-cm primary inflorescence stem base segments per genotype, from bolting plants of about 25 cm height. cDNA synthesis was performed from 1 μg total RNA in a total volume of 10 μl with SuperscriptII (Invitrogen, Life Technologies, Carlsbad, CA) and oligo-dT primer. After diluting each sample by adding 70 μl of water, 2 μl were used in 50 μl semi-quantitative PCR reactions with 26 cycles, unless stated otherwise. Gene-specific RPS10A-, RPS10B- and RPS10C primer pairs were used. RT-PCR for ACTIN2 (At3g18780) was used as RNA input control. Primer sequences are listed in Table 5.

A genomic RPS10B construct was produced by amplifying a 3.5 kb fragment spanning the RPS10B genomic region from Columbia wild-type genomic DNA with primers RPS10BgenomicF and RPS10BgenomicR (Table 5). This was digested with SpeI and HindIII and cloned into binary vector pCAMBIA2300 (http://www.cambia.org) opened with XbaI and HindIII for plant transformation.

To express RPS10A, RPS10B, RPS10C cDNAs under the RPS10B promoter, RPS10B promoter region was amplified from Columbia wild-type genomic DNA, and the RPS10A (At4g25740.1), RPS10B (At5g41520.1) and RPS10C (At5g52650.1) coding regions were amplified from Columbia wild-type cDNA using the primers specified in Table 5. The three forward primers for the RPS10 coding regions introduced an overlap with the RPS10B promoter amplicon, which was then fused upstream of each cDNA by overlap extension in a second round of polymerase chain reaction. Furthermore, the primers introduced a BamHI site followed by a NotI site just upstream of the promoter and an XbaI site just downstream of the termination codon. The products were digested with BamHI and XbaI and ligated into the cloning vector pART7 127 opened with the same enzymes. Inserts were confirmed by sequencing. From these plasmids, NotI releases a fragment consisting of the RPS10B promoter, the RPS10A, B or C coding region and the plasmid-encoded octopine synthase gene terminator, which was transferred into a NotI-digested derivative of the plant transformation vector pART27 127 which confers BASTA-resistance in plants. Confirmed constructs were shuttled into Agrobacterium strain GV3101 and used for plant transformation 126 .

Axillary bud outgrowth assays were performed with cauline nodes excised from the primary inflorescence of plants grown in sterile conditions, as described 59 . 2-¹⁴C-indoleacetic acid transport assays were conducted with 1.5-cm stem segments from the basal internode of the primary inflorescence of 6-week old soil-grown plants 55 60 .

2 mm of apical hypocotyl tissue and the cotyledonary node of 14 day-old seedlings germinated under continuous illumination and 2-mm segments from the thickened hypocotyls of 60-day-old short-day-grown plants were stained for β-glucuronidase (GUS) activity at 37°C overnight, fixed for 5 h, and embedded in Technovit (Heraeus Kulzer, Hanau, Germany); 10 μm transverse sections were prepared, mounted to slides, counterstained with ruthenium red, and permanently mounted as described 119 .