Axillary meristems form in the axils of leaves. After an initial phase of meristematic activity during which a small axillary bud is produced, they often enter a state of suspended growth from which they may be released to form a shoot branch. This post-embryonic growth plasticity is typical of plants and allows them to adapt to changing environmental conditions. The shoot architecture of genotypically identical plants may display completely contrasting phenotypes when grown in distinct environmental niches, with one having only a primary inflorescence and many arrested axillary meristems and the other displaying higher orders of branches. In order to cease and resume growth as required, the plant must co-ordinate its intrinsic developmental programme with the responses to environmental cues. It is thought that information from the environment is integrated throughout the plant using plant hormones as long-distance signals. In the present review, we focus primarily on how two of these hormones, auxin and strigolactones, may be acting to regulate shoot branching.
Auxin and shoot branching
More than 70 years ago, it was proposed that the plant hormone auxin, synthesized in the growing shoot apex, acted as an inhibitor of bud outgrowth . Decapitation removed the auxin source and thus the inhibition; addition of an agar block containing auxin to the cut stump once more restored outgrowth repression. It was initially assumed that the auxin, which is actively transported basipetally within the xylem parenchyma of the stem, moved into the bud to inhibit outgrowth; however, no lateral transport was detected and direct auxin application to the bud did not inhibit growth [2,3].
To explain this indirect inhibitory effect, it was first proposed that auxin regulates the levels/activity of a second upwardly mobile signal which moves to the bud. An excellent candidate for such a signal is the hormone cytokinin which is able to promote bud outgrowth directly . In decapitated pea stems, both xylem-mediated cytokinin export from the root  and transcription of cytokinin biosynthetic genes in the shoot  are increased. These effects could be prevented by addition of auxin to the cut stump. In Arabidopsis, this transcriptional down-regulation of cytokinin synthesis is dependent on AXR1 (AUXIN-RESISTANT1)-mediated auxin signalling . AXR1 forms part of the canonical auxin signalling pathway in which auxin binding stabilizes the interaction between the receptors, the TIR1 (TRANSPORT INHIBITOR RESPONSE1)/AFB (AUXIN RECEPTOR F-BOX) F-box family, and a group of transcriptional repressors called AUX/IAAs, targeting the latter for proteolysis ( and references therein).
Outgrowth and establishment of auxin export from an axillary bud are tightly correlated . On the basis of this observation, a second indirect inhibitory mechanism was proposed, in which downward auxin flow in the main stem blocks the establishment of an auxin-export route from an axillary bud through the process of canalization .
The canalization hypothesis proposes that the initial passive flux of auxin away from a source positively reinforces its own transport and transport polarity, which ultimately confines auxin into narrow canals, transporting auxin away from the source. The polar downward movement of auxin in shoots is mediated by basally localized auxin-efflux facilitators, PIN (PIN-FORMED) proteins, which are highly expressed in xylem-associated cells . Consistent with the canalization hypothesis, auxin appears to be one of the most influential regulators of its own transport, and multiple auxin-regulated feedback loops of PIN-dependent auxin transport have been identified. Auxin up-regulates PIN transcription, PIN incidence at the plasma membrane and also PIN polar localization ( and references therein). Auxin inhibits PIN internalization by an unknown mechanism, thus increasing the amount and the activity of PIN proteins at the cell surface. The regulation of PIN subcellular targeting is an effective way to modulate auxin distribution as it is important for the directionality of auxin fluxes. The extremely rapid changes in PIN protein accumulation and polarization in response to auxin suggest that this may be at least partly independent of the transcriptional machinery. When you couple this with the fact that the PIN proteins are constantly cycling from the plasma membrane to intracellular vesicle compartments, you can see how these dynamic changes could be regulated.
Bennett et al.  proposed a model in which apical dominance relied on competition between apices for auxin transport through the main stem. The primary apex could thus prevent subtending axillary buds from establishing auxin export into the main stem by saturating its auxin transport capacity, there being an implied assumption that the main stem had a limit to its capacity for auxin transport. This assumption of saturation has been shown to be both not true  and not essential . Using simulation models, it was shown that competition between auxin sources could also occur from the positive feedback between auxin flux and the polarization of active auxin transport seen during canalization. An existing auxin flow from a source to a sink can thus prevent the establishment of sufficient flow from a competing source to the sink, i.e. the same mechanism for regulating auxin flux can also switch auxin transport paths on or off. The primary apex is thus dominant simply because it was the first bud to establish auxin efflux into the primary stem. Prusinkiewicz et al.  have shown that PIN1 protein could be detected in the stems of both dormant and active buds, but there was only polar localization in the actively growing buds, demonstrating that the major difference is due to polarity of auxin transport rather than activity, as shown previously for pea by Morris and Johnson .
Strigolactones and shoot branching
SLs (strigolactones) are a family of plant apocarotenoids, produced and bioactive in tiny amounts . They were first identified as rhizosphere signals that allow arbuscular mycorrhizal symbionts and parasitic plants to recognize a host root . Apart from evidence for an origin from carotenoids , their biosynthetic pathway(s) was unknown. The discovery that SLs may function as plant hormones, and of the first known SL biosynthetic enzymes, is due to a class of branching inhibitor genes identified through screens for highly branched mutants. These genes were hypothesized to act in synthesis or signalling of a novel xylem-mobile branching inhibitor , are conserved in monocotyledons and dicotyledons and called D (DWARF) or HTD (HIGH-TILLERING DWARF) in rice, RMS (RAMOSUS) in pea, DAD (DECREASED APICAL DOMINANCE) in petunia and MAX (MORE AXILLARY GROWTH) in Arabidopsis [19–35] (Figure 1).
Overview of the SL pathway of branching inhibition
Grafting analysis identified candidate biosynthetic genes in this group, whose presence in the root suppressed branching in grafted mutant shoots [27,36]. Two of these encoded plastid-localized carotenoid-cleavage dioxygenases (CCD7 and CCD8) with unknown natural substrates [19,21–23]. The realization that one of these, probably CCD7 [37,38], might catalyse the branchpoint of SL biosynthesis from the carotenoid pathway, initiated a breakthrough in the branching and SL field. The ccd7 and ccd8 mutants of rice, pea and Arabidopsis were found to be SL-deficient and/or rescued by natural SL or the synthetic analogue GR24 [39,40]. This is probably also true for mutants affecting the cytochrome P450-encoding gene MAX1, placed downstream of the CCDs by graft analysis .
The conversion of carotenoids into SLs certainly requires additional enzymes. One of these is probably the D27 protein from rice . As with the ccd mutants mentioned above, d27 is SL-deficient and rescued by GR24, an SL analogue. D27 is an iron-containing plastid-localized protein of unknown function which might act in SL biosynthesis downstream of CCD7/8, but upstream of MAX1, which does not appear to be plastid-localized. A second recent addition to the SL pathway in rice, D14/D88/HTD2, belongs to the large and functionally diverse group of α/β-fold hydrolases [29–31]. This could again indicate a biosynthetic function. However, d14 overproduces SL, and is not rescued by GR24 . This is interesting, as the possibility cannot be excluded yet that the bioactive branching inhibitor is not SL, but rather is an SL derivative. One possibility is therefore an enzymatic action of D14 post-SL.
Alternatively, SL-insensitivity could indicate that D14 mediates SL signalling. A prominent example of an α/β-fold signature protein with a non-enzymatic role is the gibberellin receptor GID1 (GIBBERELLIN-INSENSITIVE DWARF1) . Genetic evidence, plus SL-insensitivity of the mutant also suggests a signalling role for the F-box leucine-rich repeat protein MAX2/D3/RMS4 [21,27,33,40]. The Arabidopsis orthologue, MAX2, forms a SCF (Skp/Cullin/F-box) ubiquitination complex in vivo . Thus SL may be yet another plant hormone that signals via targeted proteolysis of specific regulatory proteins. In the case of auxin and the hormone jasmonoyl-isoleucine, SCF-mediated signalling involves direct binding of hormone to the F-box protein, whereas in ethylene and gibberellin signalling, the F-box component does not bind the hormone directly . In the case of gibberellin, the F-box protein GID2/SLEEPY1 may bind its targets preferentially when in a protein–hormone complex with gibberellin-bound GID1 . This is intriguing because, as mentioned above, GID1 and D14 are distantly related. The substrates of the SCFMAX2/D3/RMS4 are not yet known, and the possibility is open as to whether SLs, like other plant hormones which employ SCFs, signal via proteolysis of transcriptional regulators .
The branching inhibitor probably moves in the xylem, and could thus enter the bud from the root/stem below. Is it perceived in the bud? MAX2/D3/RMS4 and D14/D88/HTD2, the candidate signalling genes, are expressed throughout the plant, including the axillary buds, and most strongly in vascular-associated cells [21,29,30,33,34]. Sector analysis in Arabidopsis, and localized SL application to pea, demonstrate that individual buds in the shoot respond autonomously: a tiny radial max2 mutant sector within a wild-type shoot, contributing to stem, bud and leaf tissue of a single node, leads to bud outgrowth from this node only . Similarly, GR24 application to the axil of an SL-deficient rms pea mutant, represses specifically the bud in that axil . However, neither approach can distinguish between an action in the bud and the surrounding nodal tissue, or perhaps either is sufficient.
There is evidence for environmental as well as hormonal control of SL biosynthesis. Root SL levels are increased under phosphate starvation, when attracting mycorrhiza and limiting branching are most crucial . Furthermore, SLs down-regulate their biosynthesis in a feedback loop which involves SL signalling, because SL pathway mutants often show increased CCD7 and CCD8 transcript levels [21,24,43,44], and the rice d3 signalling mutant, which has functional CCD7/8, overproduces SL under some conditions . In Arabidopsis and pea, grafting experiments indicate that this feedback operates locally in roots and in shoots; in addition, an effect of SL signalling in the shoot on CCD7/8 transcript accumulation in the rootstock was detected, which appeared to be more significant in pea than in Arabidopsis [43,44]. Whereas local feedback could be via SLs, long-distance feedback requires an additional downwardly mobile signal that responds to SL signalling. One candidate for this signal is auxin, whose effects on SL biosynthesis are discussed in detail below.
SLs and auxin transport
The Arabidopsis max mutants have increased shoot branching and buds that are resistant to inhibition by auxin [11,23], clearly indicating an interaction between auxin and SL. Mutations in the MAX genes also result in increased stem conductivity for auxin and increased expression of several auxin transporters, including some of the PIN family of exporters [11,12,28]. Reducing auxin transport in these mutants to wild-type levels restored both branching and bud auxin response . Similar data were also obtained from auxin transport studies on the SL biosynthetic rice d27 mutant . Increased expression of DR5:GUS, an auxin-responsive reporter, was observed within the stem vasculature of max mutants. Could the increased auxin levels in the stem purely be due to a homoeostatic mechanism triggered by the increased branching in the max mutants, i.e. do more active buds mean more auxin efflux into the stem? Arguing against this simple explanation are data from the highly branched axr1 mutant, which does not have increased auxin transport, nor can it be rescued by treatment with auxin-transport inhibitors .
The max mutants have increased PIN protein accumulation and increased auxin flux through the stem . In a simulation model, increasing the PIN allocation rate or decreasing PIN turnover rate increased the levels of PIN at the basal end of the cell and mimicked the known phenotypes of the max mutants. These changes promoted simultaneous auxin efflux from multiple buds and increased branching . The simulation predicted a gradient of increasing auxin concentrations from apical to basal in the max mutants and this prediction was verified through auxin export measurements in wild-type and max4 mutant stem segments. It should be noted that the observed increase is in the amount of auxin in transit through the stem and not in the total amount of auxin.
The MAX proteins are required for the biosynthesis and signalling of SLs. As the max mutant phenotypes can also be completely rescued by restoring their auxin transport rates to wild-type levels and are dependent on PIN1 activity , SL could be acting in some way to regulate directly PIN protein activity or polarization. This has already been observed for D’orenone, one of the postulated products resulting from CCD8 cleavage . D’orenone suppresses root-hair formation by interfering with auxin transport and signalling through specifically increasing PIN2 protein abundance and both enlarging and shifting the PIN2 expression domain basipetally . This occurs without altering PIN2 transcript levels and may be through extending the half-life of the PIN2 protein [46,47].
While considering the potential effect(s) that SL may be having on the regulation of auxin transport, it is of interest to note developments on characterization of the AXR4 protein in Arabidopsis. AXR4 functions in the same genetic pathway as the auxin import protein AUX1 and is necessary for its polar cellular localization . The protein sequence of AXR4 contains an α/β-hydrolase fold, a structural domain also found within the D14/D88/HTD2 protein which is proposed to act either in SL signalling or conversion of SL into its bioactive form [29–31].
The above examples offer an intriguing insight into possible mechanisms of SL action on auxin transport. The importance of polarized PIN1 activity in regulating auxin efflux from the bud and thus outgrowth has been discussed above and further investigations will reveal whether one of the mechanisms by which SL inhibits bud outgrowth is to depolarize or prevent polarization of PIN1.
Auxin and SL biosynthesis
A positive correlation between auxin and CCD7/8 transcript levels was originally detected in experiments where endogenous auxin levels in pea were modulated by auxin application, shoot tip removal or auxin-transport inhibitors [21,43]. Now, Arabidopsis genetics indicates that auxin maintains the CCD7/8 expression (and probably SL) levels required for branching suppression, and that SCFTIR1/AFB-mediated degradation of the transcriptional repressor BDL (BODENLOS) (AUX/IAA12) may be involved. First, SL application suppresses the increased branching of the reduced/blocked auxin signalling mutants axr1, the tir1,afb1,2,3 quadruple, and bdl [12,44]. Secondly, CCD7/8 transcript levels were determined in axr1 and bdl and were found to be reduced; and thirdly, auxin-response elements are present in the CCD7/8 promoter regions . In at least some of the above-mentioned experiments, branching of the auxin signalling mutant was rescued completely by SL. However, it is important to note that above-normal SL levels in the rescue might overrule factors which balance out SL at physiological levels. There is good evidence that auxin does not inhibit branching exclusively by promoting SL biosynthesis. For example, axr1,max double mutants show increased branching compared with max single mutants, and at least some of the max alleles used were probable knockouts . Whereas wild-type rootstocks completely rescue branching of grafted SL-deficient mutant shoots, bdl shoots are only partially, and axr1 shoots are not, rescued [27,44,49].
The AUX/IAA12 and CCD7/8 expression domains overlap in the xylem . This suggests that auxin in the xylem-associated polar transport stream regulates a pool of SL required for optimal branching suppression. Combined with the down-regulation of bud auxin export into the polar transport stream by SLs, which was discussed in the previous section, auxin–SL interactions could contribute to local and to shoot-to-root negative-feedback regulation of SL biosynthesis. In Arabidopsis, a comparison of CCD7 and CCD8 expression in the SL signalling mutant max2 and the max2,axr1 double mutant, without and with auxin transport inhibitor, revealed a major contribution of auxin to the up-regulation of CCD7/8 observed in the stem of max2, and possibly also in the hypocotyl, where up-regulation was smaller . These data show that, at least in Arabidopsis, auxin could possibly account for all the feedback observed. In pea, however, feedback is much stronger and has been postulated to involve a non-auxin signal in addition to auxin .
Auxin–SL interactions at multiple levels are critical for branching control, and probably occur in the vasculature (Figure 2). Because of the opposite mobilities of these hormones, signals coming from below and above an axillary bud could be integrated locally. Auxin, basipetally transported through the stem xylem parenchyma, may inhibit axillary bud growth in several ways: by preventing auxin transport canalization from the bud, by repressing cytokinin biosynthetic genes (not shown) and by relieving repression of SL biosynthetic genes. SLs, moving up in the xylem, might inhibit bud growth by preventing auxin transport canalization, but as our knowledge of SL (-derivative) signalling increases, additional mechanism(s) of action may be discovered. Combined, these interactions between auxin and SL also maintain hormonal homoeostasis.
Model of the transport of auxin and SL(-derived) inhibitor in the vasculature, and of their interactions
Experimental Plant Biology: Why Not?!: 4th Conference of Polish Society of Experimental Plant Biology, an Independent Meeting held at Jagiellonian University, Krakow, Poland, 21–25 September 2009. Organized and Edited by Kazimierz Strzałka (Jagiellonian University, Krakow, Poland).
Our research is supported by the Biotechnology and Biological Sciences Research Council and the Gatsby Foundation.
These authors contributed equally to this work