The root is a dynamic system whose structure is regulated by a complex network of interactions between hormones. The primary root meristem is specified in the embryo. After germination, the primary root meristem grows and then reaches a final size that will be maintained during the life of the plant. Subsequently, secondary structures such as lateral roots and root nodules form via the re-specification of differentiated cells. Cytokinin plays key roles in the regulation of root development. Down-regulation of the cytokinin response is required for the specification of a new stem cell niche, during both embryo and lateral root development. In the root meristem, cytokinin signalling regulates the longitudinal zonation of the meristem by controlling cell differentiation. Moreover, cytokinin regulates radial patterning of root vasculature by promoting protophloem cell identity and by spatially inhibiting protoxylem formation. In this review, an effort is made to describe the known details of the role of cytokinin during root development, taking into account also the interactions between cytokinin and other hormones. Attention is given on the dynamicity of cytokinin signalling output during different developmental events. Indeed, there is much evidence that the effects of cytokinin change as organs grow, underlining the importance of the spatiotemporal specificity of cytokinin signalling.
The plant hormone cytokinin takes its name from its ability to stimulate cytokinesis, that is, cell division. However, cytokinins do more than just stimulate cell division; they have important roles throughout plant development. Cytokinins were first described in 1913, when it was discovered that a compound found in phloem exudates had the ability to stimulate cell division in wounded potatoes (Solanum tuberosum; Haberlandt, 1913). Later, it was shown that coconut (Cocos nucifera) endosperm also had this ability and that various other plant species contained compounds that stimulated cell division (van Overbeek et al., 1941). While searching for factors that influence the development of tobacco (Nicotiana tabacum) calli, it was discovered that adenine, together with phosphate, increased the growth of the callus tissue (Skoog & Tsui, 1948). In 1954, Jablonski and Skoog extended their work showing that vascular tissues contained compounds that promote cell division in explants of tobacco pith (Jablonski & Skoog, 1954). The following year, the first compound that acted as a cytokinin was isolated as a degradation product of herring sperm DNA (Miller et al., 1955) and was named kinetin. The first naturally occurring cytokinin was isolated from maize (Zea mays; Miller, 1961) and was later called zeatin, from the Latin name of the genus of maize, Zea. Since then, many more naturally occurring cytokinins have been isolated and have been found to be ubiquitous to all plant species in one form or another (Salisbury & Ross, 1992; Arteca, 1996).
The ability of cytokinins to regulate cell division has become one of their defining characteristics; however, as said before, cytokinins do more than just stimulate cell division and also play an essential role in cell identity specification. In studies of the cell proliferation response of callus tissue, it was found that high cytokinin:auxin ratios induce shoot formation and high auxin:cytokinin ratios induce root formation, while intermediate concentrations sustain cell division (Miller et al., 1955; Skoog & Miller, 1957). Indeed, despite its clear effect on promoting cell division in general, cytokinin elicits opposite effects, not only on the establishment of the shoot and root meristems, but also on their proliferative activity. In the root, cytokinin deficiency reduces root meristem size and activity (Werner et al., 2003; Higuchi et al., 2004; Miyawaki et al., 2006; Werner & Schmulling, 2009). Conversely, cytokinin application induces an enlargement of the shoot meristem (Lindsay et al., 2006). In contrast, exogenous cytokinin treatment has been shown to inhibit root growth (Beemster & Baskin, 1998), while reduction of endogenous cytokinin concentrations results in increased primary root elongation (Werner et al., 2003). However, several signal transduction and biosynthesis genes are differentially expressed in the root tissues (Higuchi et al., 2004; Mason et al., 2004; To et al., 2004), suggesting that cytokinins indeed play a fundamental, but more complex role during root development.
This review will describe the role of cytokinin in root development. In particular, attention will be focused on the stage-specific role of cytokinin signalling during different root development phases. In the past decade, the molecular details underlying the regulation of root development have become increasingly clear. However, the frontier of research is now moving towards understanding of the dynamics of hormonal signalling to explain how, from a single cell, a complex structure, such as the root meristem, can grow and establish itself. For the characteristics of self-organization and dynamism of auxin signalling and distribution, for which auxin is able to instruct developmental patterning, auxin has always been the centre of such investigations (Del Bianco & Kepinski, 2011). However, cytokinin signalling is emerging as an important player in spatiotemporal regulation during plant development (Muller & Sheen, 2008; Moubayidin et al., 2010).
II. Cytokinin cellular pathway
1. Cytokinin homeostasis
Cytokinins are N6-prenylated adenine derivatives. In higher plants, the predominant cytokinins, known as isoprenoid cytokinins, are isopentenyladenine (iP) trans-zeatin (tZ), cis-zeatin, and dihydrozeatin, iP and tZ being the major forms in Arabidopsis thaliana (Sakakibara, 2006). The maintenance of an optimum cellular concentration of active cytokinin is controlled by the balance among biosynthesis, activation and catabolism. The spatial and temporal regulation of these processes is of key importance for correct growth (Werner et al., 2003; Miyawaki et al., 2004; Sakakibara, 2006; Kuroha et al., 2009), although the precise role of individual members of each family is still to be understood, mainly because of gene redundancy.
The initial, rate-limiting step in cytokinin biosynthesis is catalysed by ATP/ADP-ISOPENTENYLTRANSFERASE (IPT) (Fig. 1). In A. thaliana, there are nine IPT genes (AtIPT1–9). IPTs catalyse the transfer of an isopentenyl group from dimethylallyl diphosphate to an adenine nucleotide (ATP, ADP, or AMP) (Kakimoto, 2001; Takei et al., 2001). From their overall expression patterns it can be deduced that cytokinins are synthesized throughout the plant, including the root, shoot, and immature seeds. However, the expression domain of each IPT gene is tissue- and organ-specific (Miyawaki et al., 2004), suggesting gene-specific functions. Expression of AtIPT3, AtIPT5, and AtIPT7 is relatively high in the vegetative organs. AtIPT4 and AtIPT8 are expressed in immature seeds and AtIPT1 is expressed in ovules and vegetative organs. Nevertheless, no mutant of a single IPT gene exhibits a visible phenotype, suggesting functional redundancy within this gene family (Miyawaki et al., 2006). Combinations of multiple ipt mutants, however, display a drastic decrease in cytokinin concentrations, severely inhibiting shoot growth, while enhancing the production of root tissue (Miyawaki et al., 2006).
An additional step in the production of bioactive cytokinin is the removal of a ribose 5′-monophosphate group. Initially identified in rice (Oryza sativa), the LONELY GUY (LOG) genes encode enzymes with phosphoribohydrolase activity that releases cytokinin from the inactive precursor cytokinin riboside 5′-monophosphate (Fig. 1). In rice, LOG is specifically expressed in the shoot meristem tips where it regulates stem cell activity through the regulation of cytokinin concentrations (Kurakawa et al., 2007). In A. thaliana, nine homologues of the rice LOG gene have been isolated (AtLOG1–9), the products of seven of which retain enzymatic activity equivalent to that of the rice LOG (Kuroha et al., 2009). Analysis of spatial distribution patterns of AtLOGs revealed that, like the IPTs, these genes are expressed throughout the plant during development, although with different and overlapping expression domains (Kuroha et al., 2009). In some tissues, such as in the lateral root primordia and root vasculature, the expression of AtLOGs is concomitant with the expression of IPTs (Miyawaki et al., 2004; Kuroha et al., 2009). This suggests that cytokinins may be synthesized and subsequently activated by the AtLOGs in the same tissue, and then act as autocrine or paracrine signals. However, some AtLOG genes are expressed in tissues, such as root hairs and the root elongation zone epidermis, in which no expression of IPT genes can be observed (Miyawaki et al., 2004; Kuroha et al., 2009). This could indicate that, in these cases, cytokinin precursors translocate from cell to cell and/or over a long range, before being activated in these tissues. Multiple AtLOG mutants show inhibition of shoot meristem size (Kuroha et al., 2009; Tokunaga et al., 2012), resembling the phenotype of multiple loss-of-function ipt mutants (Miyawaki et al., 2006), strongly suggesting that cytokinin activation is essential to complete different developmental processes. While multiple ipt mutants show an increased production of root tissue, multiple AtLOG mutants display shorter roots compared with wild-type plants as a result of the lack of phloematic tissue, which is still present in the multiple ipt mutants (Matsumoto-Kitano et al., 2008; Tokunaga et al., 2012).
The steady-state concentrations of active cytokinins are negatively regulated by cytokinin catabolic genes. In A. thaliana, cytokinin breakdown is controlled by the activity of seven CYTOKININ OXIDASE/DEHYDROGENASE (CKX) genes (CKX1 to CKX7) (Fig. 1). The N6 side chain of cytokinin, which confers high cytokinin activity (Whitty & Hall, 1974), is the target of CKXs, which catalyse its cleavage (Mok & Mok, 2001; Schmulling et al., 2003). Like IPT and LOG genes, each CKX gene displays a specific expression pattern during plant development, but in general CKX expression can be found in regions of active growth, in the shoot and root meristems, and in emerging leaves (Werner et al., 2003). Overexpression of CKX genes phenocopies the effect of combined ipt and multiple log mutations in the shoot, and ipt mutations in the root, confirming the importance of the fine-tuning of cytokinin concentration for plant development (Werner et al., 2003; Miyawaki et al., 2006).
In order to modulate cytokinin signalling, many phytohormones, including cytokinin itself, auxin and abscisic acid, regulate the expression of IPT and CKX genes. In A. thaliana, cytokinins inhibit the transcription of AtIPT1, AtIPT3, AtIPT5, and AtIPT7 (Fig. 1), whereas auxin promotes the accumulation of the transcripts of AtIPT5 and AtIPT7 in roots (Miyawaki et al., 2004). In maize, CKX genes are up-regulated by cytokinins and abscisic acid (Brugiere et al., 2003) (Fig. 1).
Cytokinin homeostasis varies with time. Promoter-GUS fusion analyses have revealed changes in the expression of members of each family (Kuroha et al., 2009). In the embryo, AtIPT1 is only expressed in mature embryos, while AtIPT4 and AtIPT8 exhibit expression at an early stage of seed development, but while AtIPT4 promoter activity disappears at the early heart stage, AtIPT8 promoter activity decreases when the embryo is at the late heart stage. In the primary root, AtIPT5 is expressed at the transition zone (TZ) and in the columella root cap of the primary root, where its promoter activity decreases with time and is undetectable at 7 d post germination (dpg) (Kuroha et al., 2009). AtCKX4 is also expressed in the root cap, but its expression increases with development of the primary root. LOG8 expression is detectable only in the quiescent centre (QC) of young roots (Kuroha et al., 2009). AtCKX1 and AtCKX6 are expressed at lateral root emergence sites, and the activity of their promoter has been found to gradually increase with increasing elongation of the lateral root (Kuroha et al., 2009). This dynamicity reflects the importance of tight temporal regulation of the concentrations of active cytokinin for plant development.
2. Biochemistry of cytokinin signalling
Cytokinin signalling involves a multistep phosphotransfer cascade similar to the bacterial two-component system (Kakimoto, 2003; Argueso et al., 2010). In its simplest form, a two-component system comprises two elements, a histidine protein kinase and a response regulator protein, that are phosphorylated at conserved His and Asp residues, respectively. Phosphotransfer from the histidine kinase to the response regulator results in activation of the latter. More complex versions of the two-component system, like cytokinin signalling, simply contain multiple phosphotransfer steps between different proteins. The activation of the first element in such a signalling pathway generates a phosphorelay cascade between all the downstream components.
The first step of cytokinin perception requires the binding of cytokinin to a transmembrane histidine kinase receptor (Inoue et al., 2001) (Fig. 1). This binding induces, within the receptor, the autophosphorylation of a conserved His residue in its kinase domain (Perraud et al., 1999). In A. thaliana, there are three histidine kinase receptors: ARABIDOPSIS HIS KINASE 2 (AHK2), AHK3, and AHK4/WOODENLEG (WOL)/CYTOKININ RESPONSE 1 (CRE1 ) (Hwang & Sheen, 2001; Inoue et al., 2001). The three AHKs show largely overlapping expression patterns and partially redundant functions in cytokinin perception (Higuchi et al., 2004; Nishimura et al., 2004; Riefler et al., 2006). However, the lack of embryo lethality in the triple mutant, while calling into question the importance of the role of cytokinin signalling in plant development, leaves open the possibility of the presence of still unidentified cytokinin receptors (Higuchi et al., 2004; Nishimura et al., 2004). Indeed, CYTOKININ INSENSITIVE 1 (CKI1) has been shown to contribute to cytokinin-regulated vascular development and possibly acts as a cytokinin receptor (Kakimoto, 1996; Hejatko et al., 2009).
Following autophosphorylation, the phosphate group is then transferred to a conserved Asp residue within the receiver domain of the AHK protein, and subsequently transferred to a member of the ARABIDOPSIS HIS PHOSPHOTRANSFER PROTEIN (AHP) family (Hwang & Sheen, 2001; Hutchison et al., 2006) (Fig. 1). While it was previously hypothesized that AHP proteins translocate from the cytosol into the nucleus in response to cytokinin (Hwang & Sheen, 2001), it has been recently observed that in vivo AHPs are continuously transported in and out of the nucleus independently of their phosphorylation status and of the cytokinin response (Punwani et al., 2010). In the nucleus, phosphorylated AHPs transfer the phosphate group to the nuclear-localized ARABIDOPSIS RESPONSE REGULATOR (ARR) proteins (Hutchison et al., 2006) (Fig. 1). AHP6 is an AHP that lacks the conserved histidine residue required for phosphotransfer and therefore acts as a negative regulator of cytokinin signalling (Mahonen et al., 2006) (Fig. 1).
The ARRs are encoded by a gene family. All ARRs share a similar receiver domain with conserved residues targeted for phosphorylation (To & Kieber, 2008). However, ARRs are classified into type A, type B and type C, depending on their protein domains and cytokinin responsiveness (To et al., 2007). Type-A ARRs (ARR3–ARR9 and ARR15–ARR17) possess short C-termini, while type-B ARRs (ARR1, ARR2, ARR10–ARR14 and ARR18–ARR21) have longer C-termini, which contain a conserved GARP (GOLDEN2/ARR/Psr1) DNA-binding domain. Type-A ARRs act as negative regulators of cytokinin responses, via an as yet unknown mechanism (To et al., 2004, 2007; Kim, 2008; Muller, 2011) (Fig. 1). Type-B ARRs, which are transcription factors, play a positive role in mediating cytokinin-regulated gene expression (Fig. 1). Moreover, sequence alignment has revealed stretches of nonconserved sequences, which could be responsible for gene-specific functions (Muller, 2011). The type-C ARRs are more distantly related to type-A and type-B ARRs. They are not transcriptionally regulated by cytokinin like type-A ARRs and do not contain the output domain of type-B ARRs; however, their overexpression results in reduced sensitivity to cytokinin, suggesting a negative role similar to that of type-A ARRs (Kiba et al., 2004).
AHPs also mediate the accumulation in the nucleus of other plant-specific cytokinin-responsive genes, the CYTOKININ RESPONSE FACTORS (CRFs), members of the A. thaliana AP2 (APETALA 2) family of transcription factors (Rashotte et al., 2006). In response to cytokinin, activated type-B ARRs and CRFs act together to regulate gene expression of common and unique targets, including type-A ARRs (Hwang & Sheen, 2001; Rashotte et al., 2006). Indeed, the transcriptional activation of the type-A ARRs represents a negative feedback loop required to dampen the response upon high or prolonged signal input (To et al., 2004). In addition to their transcriptional up-regulation, activated type-A ARR proteins become more stable in response to cytokinin (To et al., 2007).
Although the individual role of each family involved in the two-component cytokinin signalling pathway is well understood, little is known regarding the way in which each member mediates a specific cytokinin-dependent developmental output (Muller, 2011). The fact that every step of the signalling cascade is supported by a small gene family, all members of which present a specific and complex expression pattern and may differ in affinity to the downstream elements, provides a first level of response diversity. In addition, the pre-patterned context of the signal-receiving cell could affect the response triggered by cytokinin. Moreover, temporal variations in the topology of the cytokinin pathway components within a cell or a tissue have been shown to be crucial for the completion of developmental events (Moubayidin et al., 2010).
III. Cytokinin signalling in root development
The core of the root meristem is the stem cell niche (SCN) (Fig. 2b). The SCN contains the organizing centre, a small group of slowly dividing cells known as the QC, and surrounding stem cells. Direct contact between the QC and the stem cells is required for maintenance of cell stemness (van den Berg et al., 1997). Stem cells generate transit-amplifying cells, which divide in the proximal meristem (PM). Cell differentiation is initiated at the TZ, where cells enter the elongation/differentiation zone (EDZ) and terminally differentiate. The maintenance of root meristem size depends on the equilibrium between the rates of cell division in the meristem and cell differentiation at the TZ.
Key to the establishment and function of the root meristem is another plant hormone, auxin. Auxin acts in gradients (Benjamins & Scheres, 2008; Brunoud et al., 2012). The asymmetric distribution of auxin is established through differential polar auxin transport. Polar auxin transport is an active process mediated by both auxin influx and efflux carriers (Band et al., 2012). Auxin influx carriers, such as AUX1 and AUX1-LIKE (LAX), facilitate auxin uptake into cells, while the direction of auxin flux within tissues is driven by the coordinated asymmetrical membrane distribution of efflux carriers (Band et al., 2012). Thus, auxin efflux carriers, like PINs (ARABIDOPSIS THALIANA PIN-FORMED) and PGPs (P-GLYCOPROTEIN), and their dynamic subcellular localization are crucial regulators of patterning. Auxin is transported from the shoot and accumulates at an auxin maximum centred on the QC and columella cells, which act as an auxin sink (Sabatini et al., 1999; Friml et al., 2002). Auxin is then redistributed back up through the lateral root cap and into the proximal meristem (Fig. 2b). Auxin flux through the root meristem is crucial to sustain cell division and inhibit cell differentiation (Vieten et al., 2005). Accordingly, exogenous application of auxin during root growth increases root meristem size (Dello Ioio et al., 2007), while multiple combinations of pin mutants display shorter root meristems and delayed root growth (Blilou et al., 2005).
The complex mechanisms underlying auxin transport generate a landscape of auxin distribution throughout the root meristem to which cells respond appropriately (Del Bianco & Kepinski, 2011). The cellular response to auxin involves three different protein families: the TIR1/AFB (TRANSPORT INHIBITOR 1/AUXIN SIGNALLING F-BOX PROTEIN) family of F-box proteins, the Aux/IAA (INDOLEACETIC ACID-INDUCED PROTEIN) family of transcriptional repressors and the ARF (AUXIN RESPONSE FACTOR) family of DNA-binding transcription factors. Auxin binds directly to the TIR1/AFB receptors. In the presence of auxin, TIR1/AFB proteins negatively regulate Aux/IAA repressor concentrations by promoting their degradation by the 26S proteasome, thereby activating ARF-mediated gene expression (Dharmasiri et al., 2005; Kepinski & Leyser, 2005). Many lines of evidence indicate that the auxin response is required for meristem establishment and activity. For example, loss-of-function mutations of ARF5/MONOPTEROS (ARF5/MP) and gain-of-function mutations of IAA12/BODENLOS (IAA12/BDL) interfere with the formation of the root meristem by affecting the orientation of embryonic cell division planes (Berleth & Jurgens, 1993; Hamann et al., 1999, 2002). Moreover, mutations of other members of the Aux/IAA and ARF family members affect primary root length (Leyser et al., 1996; Tian & Reed, 1999; Okushima et al., 2005).
The auxin–cytokinin cross-talk controls root meristem development, pre- and post-embryonically (Dello Ioio et al., 2007; Muller & Sheen, 2008) (Fig. 2). A genetic framework of antagonistic interactions between auxin and cytokinin is responsible for the control of the balance between cell division and cell differentiation in the root meristem (Dello Ioio et al., 2008; Moubayidin et al., 2010). This balance, between proliferation-supporting auxin signalling and differentiation-promoting cytokinin input, determines meristem length (Dello Ioio et al., 2008). However, as will be discuss in the following sections, each stage of root development is accompanied by changes in cytokinin sensitivity (Muller & Sheen, 2008; Moubayidin et al., 2010). These changes are pivotal for the establishment and maintenance of the mature root meristem.
1. The role of cytokinin in the establishment of the SCN during embryo development
The A. thaliana primary root meristem is initiated with the specification of a single cell, the hypophysis (Fig. 2a). The hypophysis divides asymmetrically into an upper lens-shaped cell that gives rise to the QC and a larger basal cell that forms the columella stem cells (Scheres et al., 1994). In addition, adjacent cells from the apical cell lineage are later recruited by the QC to become the upper tier of root stem cells (van den Berg et al., 1997). If the hypophysis is not specified properly, the root meristem will not form, resulting in the rootless phenotype (Hamann et al., 1999).
Hypophysis specification is linked to auxin signalling (Friml et al., 2003) (Fig. 2a). Mutations affecting auxin distribution and signalling interfere with the specification of the hypophysis. Indeed, the ARF transcription factor ARF5/MP drives hypophysis specification by promoting auxin flux from the embryo to the hypophysis precursor (Hamann et al., 1999). This relies on auxin-dependent degradation of the Aux/IAA IAA12/BDL, which releases ARF5/MP from inhibition and allows activation of target genes (Hamann et al., 2002). However, this auxin input is confined to the provasculature of the central region immediately adjacent to the hypophysis and acts non-cell-autonomously on hypophysis specification (Weijers et al., 2006).
In addition to auxin, cytokinin signalling plays an important role in the establishment of the primary root SCN (Fig. 2a). Auxin antagonizes cytokinin signalling in the basal daughter cell of the hypophysis by inducing the expression of two type-A ARR genes, ARR7 and ARR15 (Muller & Sheen, 2008). Auxin-mediated suppression of cytokinin signalling in the embryonic basal cell lineage is required in order to ensure the establishment of the root stem-cell niche (Muller & Sheen, 2008). Moreover, as the lens-shaped cell is prominently marked by activity of the cytokinin-sensitive synthetic promoter Two-Component Output Sensor (TCS) during early embryogenesis, phosphorelay activity could also be important for stem-cell specification. Thus, tissue-specific phosphorelay outputs seem to be required for the successful development of hypophysis-derived daughter cells into an operational root stem-cell system.
2. Cytokinin regulation of cell differentiation in the primary root
During post-embryonic root development, cytokinins control the cell differentiation rate of meristematic cells (Dello Ioio et al., 2007) (Fig. 2b). Indeed, exogenous application of cytokinins inhibits root growth and reduces root meristem size without altering stem-cell activity or cell division in the PM (Dello Ioio et al., 2007). Similarly, mutants in cytokinin biosynthesis genes, in the cytokinin receptor AHK3 gene and in the type-B ARR genes ARR1 and ARR12 display strongly enhanced root growth and enlarged meristems (Dello Ioio et al., 2007, 2008). Moreover, vascular-specific depletion of cytokinin at the TZ is sufficient to affect meristem size, while reduction of cytokinin concentrations in other tissues has no such effect (Dello Ioio et al., 2007). All these data indicate that an AHK3/ARR1/ARR12 two-component cytokinin-signalling module specifically acts in the vascular tissue at the TZ to regulate the differentiation rate (Dello Ioio et al., 2007).
Cytokinins control cell differentiation by negatively regulating the positive effect of auxin on cell division, affecting auxin signalling and distribution (Dello Ioio et al., 2008) (Fig. 2b). Cytokinins, through AHK3/ARR1, directly activate the transcription of the Aux/IAA IAA3/SHORT HYPOCOTYL 2 (IAA3/SHY2) in TZ vascular tissue (Taniguchi et al., 2007; Dello Ioio et al., 2008). IAA3/SHY2 then negatively regulates PIN expression, limiting auxin transport and distribution, and allowing cell differentiation (Dello Ioio et al., 2008). IAA3/SHY2 is necessary and sufficient to control root meristem size in response to cytokinin (Dello Ioio et al., 2008). Indeed, gain-of-function shy2-2 mutants display a short root meristem and impaired root growth, mimicking the effect of cytokinin application. Correspondingly, loss-of-function shy2-31 mutants display an enlarged root meristem, closely resembling the ahk3 and arr1 mutant phenotype. PIN expression is decreased in the shy2-2 mutant, while is enhanced in the shy2-31 background, corroborating the consequential relationship between IAA3/SHY2, PIN expression and meristem length.
Cytokinin signalling acquires spatial properties along the root meristem by interplaying with the PHABULOSA (PHB) and BREVIS RADIX (BRX) pathways. PHB is a member of the CLASS III HOMEODOMAIN-LEUCINE ZIPPER (HD-ZIP III) of transcription factors, which are involved in patterning processes throughout plant development. Indeed, an increase in PHB protein levels results in a shortening of the root and root meristem. PHB regulates root meristem length by inducing cytokinin biosynthesis via IPT7 in the PM of the root, thus activating ARR1 at the TZ. ARR1, in turn, represses the expression of PHB at the TZ vasculature, thus restricting PHB expression to the distal part of the PM. PHB expression is restricted to the vascular bundle by the activity of microRNA 165A (miR165A) in the endodermis, which is negatively regulated by ARR1. BRX, in contrast, is a transcription co-activator that is expressed in the vasculature and is rate-limiting for transcriptional auxin action (Mouchel et al., 2004). BRX can regulate IAA3/SHY2 expression, probably acting as a cofactor of ARF5/MP (Scacchi et al., 2010). BRX activity is controlled by auxin at both the transcriptional and post-translational levels (Mouchel et al., 2006; Scacchi et al., 2010), while cytokinins are able, through IAA3/SHY2, to control BRX expression and, additionally, polar auxin transport, as PIN3 is downstream of BRX. This regulatory network stabilizes BRX expression in the early PM and IAA3/SHY2 levels at the TZ. The PHB and BRX pathways act, therefore, upstream and downstream of ARR1 to confine cytokinin action to the TZ. Considering the tight interplay between an auxin gradient, miR165 and HD-ZIP III in the regulation of leaf polarity (Pekker et al., 2005), it would be interesting to investigate since BRX is likely an ARF cofactor, these pathways also interact along the root meristem.
A negative feedback loop, additional to the PHB/IPT7 circuit, between IAA3/SHY2 and cytokinin biosynthesis fine-tunes the circuit controlling cell differentiation (Miyawaki et al., 2004; Dello Ioio et al., 2008). Indeed, the cytokinin biosynthetic AtIPT5 gene, which is specifically expressed in the vascular tissue at the TZ, is rapidly induced by auxin (Miyawaki et al., 2004). This induction requires the auxin-dependent degradation of IAA3/SHY2, as the activity of the AtIPT5 promoter is lost in the shy2-2 background (Dello Ioio et al., 2008). By promoting IAA3/SHY2 expression, cytokinins repress auxin signalling and transport, but also negatively regulate their own concentration through the control of IPT5 (Dello Ioio et al., 2008). This circuit of multiple feedback loops guarantees the maintenance of an appropriate root meristem size.
In addition to controlling longitudinal root patterning, cytokinin regulates the radial patterning of the root vasculature (Fig. 3b). In A. thaliana, the root vasculature is a cylinder with inner bisymmetry. Two phloem strands run along the root at opposite poles of the vasculature cylinder. Between them, the xylem occupies the central symmetry plane and is composed of the peripheral protoxylem and central metaxylem. The phloem and xylem are separated by procambial cells. Early in vascular development, pattern is established through a set of asymmetric cell divisions requiring CRE1 activity (Scheres et al., 1995; Mahonen et al., 2000; Nishimura et al., 2004). In plants carrying a version of CRE1 that exhibits a constitutive phosphatase activity, wol, the vasculature is composed of fewer cells, all of which possess protoxylem identity (Scheres et al., 1995; Mahonen et al., 2000, 2006). Moreover, the type-B ARR1, ARR10, and ARR12 control transcription of genes regulating root vascular differentiation, as the triple loss-of-function mutant shows a wol-like phenotype (Yokoyama et al., 2007; Argyros et al., 2008; Ishida et al., 2008). In the procambium, cytokinin signalling regulates PIN1, PIN3, and PIN7 polarity, forcing auxin to accumulate in protoxylem cells (Bishopp et al., 2011a,b). Here, auxin promotes the transcription of AHP6, which acts as a negative regulator of cytokinin signalling (Mahonen et al., 2006). The AHP6-mediated inhibition of cytokinin signalling, prompted by auxin, confines the cytokinin response to the procambial cells, defining vasculature patterning (Mahonen et al., 2006; Bishopp et al., 2011a).
3. The role of cytokinin in lateral root formation
Root branching is a post-embryonic developmental event that involves the production of new roots from a small number of cells of the pericycle. Lateral roots are initiated in the pericycle overlying the xylem pole (Fig. 3b) where, in response to auxin, pairs of cells acquire the attributes of founder cells (De Smet et al., 2007; Dubrovsky et al., 2008). The two founder cells divide synchronously, generating two short central cells and two longer flanking cells (Fig. 3a). Cells will further divide by a series of anticlinal and periclinal divisions, generating primordia that will eventually emerge through the adjacent layers of the primary root.
Dynamic variations in auxin sensitivity and distribution are pivotal for the positioning and development of lateral roots. Priming of pericycle cells for lateral root initiation take place very early, in the EDZ adjacent to the TZ, where local auxin responsiveness oscillates with peaks of expression at regular time intervals (De Smet et al., 2007; Moreno-Risueno et al., 2010). Lateral roots are distributed along the root in a regular left–right alternating pattern, which correlates with gravity-induced waving and depends on the auxin influx carrier AUX1 (De Smet et al., 2007). When a lateral root develops, the first visible event is an increment in auxin concentration, which is necessary for the re-specification of pericycle cells to become founder cells (Dubrovsky et al., 2008). The action of the auxin efflux carriers PINs generates a dynamic auxin distribution that is responsible for the establishment of a new self-sustaining apical meristem (Benkova et al., 2003) (Fig. 3a). For example, PIN1 is initially detected exclusively on the anticlinal sides of the short initial cells, but later it can also be found at the periclinal sides. As the lateral root develops, PIN1 polarity points increasingly towards the primordium tip (Benkova et al., 2003). Finally, in the emerging lateral root, PIN1 localization resembles the distribution in the primary root meristem (Friml et al., 2002; Benkova et al., 2003). It seems that the other PIN proteins, which are normally found in the primary meristem, change distribution within the developing lateral root to sustain auxin accumulation in the founder cells and then establish the looped auxin flux seen in the primary meristem. Moreover, a PIN3-mediated reflux of auxin between the pericycle and the endodermis stabilizes the auxin maximum and promotes lateral root initiation (Marhavy et al., 2013).
Auxin controls lateral root formation through successive auxin response modules. Early events in lateral root initiation are regulated by the IAA14/SOLITARY ROOT (IAA14/SLR)–ARF7–ARF19-dependent auxin response module (Fukaki et al., 2002; Vanneste et al., 2005). Indeed, plants harbouring a gain-of-function version of IAA14/SLR, slr1, or null mutations in both ARF7 and ARF19 show a lateral rootless phenotype (Okushima et al., 2005; Wilmoth et al., 2005). It has been shown that the IAA14/SLR–ARF7–ARF19 auxin response module is required for cell cycle activation of the pericycle cells that will become founder cells (Fukaki et al., 2002; Vanneste et al., 2005). However, cell cycle stimulation in slr1 mutants is insufficient to stimulate lateral root organogenesis (Vanneste et al., 2005). Indeed, after cell cycle activation, the IAA12/BDL–ARF5/MP-mediated auxin response is required for the development of an organized lateral root (De Smet et al., 2010). Lateral root outgrowth is then regulated by the small RNA miR390, which controls the production of the tasiARFs (TAS3-derived trans-acting short-interfering RNA) that repress the expression of ARF2, ARF3, and ARF4 (Marin et al., 2010; Yoon et al., 2010).
Cytokinins are negative regulators of lateral root formation. Indeed, application of exogenous cytokinins inhibits lateral root initiation in both A. thaliana (Li et al., 2006; Laplaze et al., 2007; Kuderova et al., 2008) and O. sativa (Rani Debi et al., 2005). Moreover, plants carrying loss-of-function mutations in several type-B ARR and AHK genes, as well as overexpressing CKX, show enhanced lateral root formation (Werner et al., 2001, 2003; Lohar et al., 2004; Mason et al., 2005; Riefler et al., 2006). Accordingly, several type-A ARR mutants exhibit a reduced number of lateral roots (To et al., 2004).
Down-regulation of cytokinin sensitivity in pericycle cells is required for lateral root priming at early stages in the EDZ of the primary meristem. Early phases of lateral root organogenesis, including priming and initiation, take place just above the TZ (De Smet et al., 2007). However, this zone shows elevated levels of biologically active cytokinins and is an important target for cytokinin signalling (Dello Ioio et al., 2008; Bielach et al., 2012). Nevertheless, in the pericycle cells primed for lateral root formation, the cytokinin response is repressed (Bielach et al., 2012). Indeed, induction of IPT expression in the zone of lateral root initiation greatly affects the formation of lateral roots at later stages (Bielach et al., 2012).
Cytokinin acts in the xylem-pole pericycle cells to inhibit lateral root initiation. Tissue-specific expression of IPT genes or of the CKX1 gene in xylem-pole pericycle cells perturbs lateral root initiation (Laplaze et al., 2007). Cytokinin acts through the CRE1 receptor and AHPs to down-regulate PIN expression in the initials of the lateral root founder cells (Hutchison et al., 2006; Li et al., 2006; Laplaze et al., 2007). Accordingly, overexpression of IPT genes affects early primordium patterning, as well as auxin distribution during lateral root organogenesis (Kuderova et al., 2008). However, an enhanced cytokinin response in the xylem-pole pericycle cells between existing lateral root primordia seems to be functionally important to prevent lateral root initiation near the existing primordia (Bielach et al., 2012) (Fig. 3a). Indeed, auxin application to ipt3 ipt5 ipt7 triple mutant roots induces a massive lateral root initiation, not seen in wild-type roots (Bielach et al., 2012).
4. Cytokinin signalling in nodule formation
Legumes have evolved a specialized structure for association, under nitrogen-limiting conditions, with symbiotic nitrogen-fixing bacteria: the root nodule (Crespi & Frugier, 2008). Nodulation, that is, the formation of root nodules, typically begins with soil rhizobium bacteria attaching to the tip of a growing root-hair cell, causing root-hair curling and branching (Fournier et al., 2008). Simultaneously, protoxylem-pole cortical cells are activated for division, underneath the infection point, forming the nodule primordium. Bacteria proliferate within the root-hair curl, locally degrading the plant cell walls, and the infection thread develops, growing towards the root cortex, then branches and ramifies in the nodule primordium. Finally, rhizobia are internalized into the nodule primordium cells, where they differentiate into nitrogen-fixing bacteroids (Stacey et al., 2006). Bacteroids convert atmospheric nitrogen into ammonium, which is then exported to the plant cell cytosol and exploited by the host.
Both plant and bacterial signals are involved in the regulation of root nodule formation. In the pre-infection stage, specific flavonoids released by roots serve as chemoattractants for the rhizobia. Moreover, perception of flavonoids by bacteria triggers the production of the Nodulation (Nod) factors, which are perceived by the host root and act as signal molecules to initiate nodule formation (Crespi & Frugier, 2008).
Cytokinin signalling coordinates cortical cell division with infection. In the legume model plants Lotus japonicas and Medicago truncatula, impairment of cytokinin receptor activity results in blocked nodule formation (Gonzalez-Rizzo et al., 2006; Murray et al., 2007). Lotus japonicas plants harbouring a gain-of-function mutation in the LOTUS HISTIDINE KINASE 1 (LHK1) cytokinin receptor spontaneously developed empty nodules in the absence of rhizobia (Tirichine et al., 2007). The role of cytokinin in M. truncatula is further supported by the reduced infection phenotype in plants with reduced expression of 3-hydroxy-3-methylglutaryl CoA reductase 1 (HMGR1), an enzyme required for cytokinin biosynthesis (Kevei et al., 2007). Cytokinin influences nodule development by inducing the expression of the putative transcription factor NODULE INCEPTION (NIN) in the root cortex (Heckmann et al., 2011). However, NIN acts downstream of Nod signalling, and the absence of cytokinin-induced primordia in nin mutants shows that cytokinin also acts downstream of Nod signalling (Madsen et al., 2010; Heckmann et al., 2011). Indeed, in M. truncatula NIN is required for the induction of MtARR4, which acts downstream of MtCRE1 during nodulation (Gonzalez-Rizzo et al., 2006).
Cytokinin and flavonoid pathways converge on auxin distribution to regulate nodule organogenesis. During nodule formation, auxin specifically accumulates in dividing cortical cells and silencing of PIN genes results in reduced nodulation (Hirsch et al., 1989; Rightmyer & Long, 2011; Suzaki et al., 2012). MtPIN expression is up-regulated in the cre1 mutant of M. truncatula, suggesting that one role of cytokinin signalling in nodulation might be to establish local auxin accumulation through control of the expression of auxin transporters (Plet et al., 2011). However, Nod factors positively feed back on the synthesis of flavonoids, which will then act on auxin transport (Hirsch, 1992; Wasson et al., 2006). Furthermore, the key transcription factor in nodule development, NIN, has been shown to positively regulate auxin accumulation in the cortical cells (Suzaki et al., 2012).
IV. Spatiotemporal variations of cytokinin signalling
1. Requirement of transient suppression of cytokinin signalling for SCN specification in both embryo and lateral root development
The establishment of a new SCN requires the down-regulation of cytokinin signalling. As seen in the previous sections, the auxin-mediated suppression of cytokinin signalling in the embryonic basal cell lineage is required for the establishment of the primary root stem-cell niche (Muller & Sheen, 2008). It seems that, during lateral root organogenesis, cytokinin inhibits lateral root initiation by blocking the division of pericycle cells that will acquire founder cell identity (Li et al., 2006). However, the effects of cytokinin input on the establishment of a root SCN change according to the developmental stage, in both embryo and lateral root development.
Manipulation of cytokinin signalling after the embryonic heart stage does not affect root stem-cell organization (Muller & Sheen, 2008) (Fig. 4a). Mimicking the cytokinin response in the basal cell of early globular embryos disrupts the established embryo pattern. However, manipulations of cytokinin signalling initiated later, at the embryonic heart stage, seem to have no effect on root stem-cell organization. The suppression of cytokinin signalling in the basal cell at the early globular stage could therefore be instructive for the establishment of a functional SCN.
In the formation of lateral root founder cells, the down-regulation of cytokinin is required for founder cell priming. The advantage of the lateral root system is that the primary root represents a longitudinal axis of time and development, along which all the different stages of lateral root development can be traced. Remarkably, exogenous induction of IPT activity in the area of lateral root priming and early development, but not in the area of lateral root emergence, impaired lateral root initiation (Bielach et al., 2012). Moreover, young lateral root primordia are more sensitive to perturbations in cytokinin activity than are developmentally more advanced primordia (Bielach et al., 2012). The ability of cytokinin to affect lateral root development is therefore strongly stage-dependent.
2. The switch in cytokinin signalling defines the growth phase and the final size of the primary root meristem
The root meristem has to increase in size in the first phases after germination to reach its final dimensions, and then its growth must stop and meristem size be maintained. This is a very important step for the future fitness of the plant and has probably been forged by natural selection, varying between species adapted to different environments. Indeed, a meristem that is not active enough or too active could generate a root system not suited to reach the water supply or exploit the soil nutrients. Moreover, a meristem that is not maintained properly would use valuable energy to continue to grow, or differentiate completely and not be active for the entire lifespan of the plant. To allow the meristem to lengthen, initially cell proliferation has to exceed cell differentiation. Then the ceasation of meristem growth can be achieved by varying either the differentiation or the proliferation rate. It has been shown recently that, after seed germination, root meristem growth is controlled by an increase in the rate of cell differentiation relative to cell division, and that cytokinin signalling plays a pivotal role in this event (Moubayidin et al., 2010) (Fig. 4b).
After seed germination, root meristem growth is controlled by the modulation of cytokinin signalling (Moubayidin et al., 2010). As described in the previous section, the molecular network involving both cytokinin and auxin controls the balance between cell division and cell differentiation in the root meristem. To achieve this, auxin signalling and cytokinin signalling interact to control the transcription and activity of the transcription factor IAA3/SHY2 (Gray et al., 2001; Dello Ioio et al., 2007; Scacchi et al., 2010) (Fig. 4b). In the first days post germination (dpg), IAA3/SHY2 mRNA levels increase over time, reaching a maximum at 5 dpg. Moreover, transient expression of the gain-of-function version of IAA3/SHY2, shy2-2, during the root meristem growth phase, is sufficient to reduce and prematurely set root meristem size (Moubayidin et al., 2010). IAA3/SHY2 expression is under the control of two type-B ARRs, ARR1 and ARR12, which exhibit different patterns of temporal expression. Both transcription factors are expressed at the TZ; however, ARR12 is already present following germination, whereas ARR1 is activated only at 5 dpg (Dello Ioio et al., 2007; Moubayidin et al., 2010). Molecular as well as genetic analyses of double mutant combinations have demonstrated that in the first 5 dpg, during meristem growth, expression of IAA3/SHY2 is controlled only by ARR12 (Moubayidin et al., 2010). However, at 5 dpg, ARR1 transcription is activated, prompting up-regulation of IAA3/SHY2 by cytokinin (Moubayidin et al., 2010). The increase in IAA3/SHY2 expression modulates PIN expression, causing auxin redistribution and an increase in the cell differentiation rate (Dello Ioio et al., 2008; Moubayidin et al., 2010).
In accordance with variation in cytokinin signalling, the downstream targets of cytokinin vary between 3 and 5 dpg. BRX mediates the cytokinin effect and is necessary to enhance PIN3 expression during early root development to promote meristem growth and determine final meristem size (Scacchi et al., 2010). Moreover, cytokinin-dependent auxin redistribution may affect the downstream targets of auxin, such as the PLETHORA (PLT) genes. The PLT transcription factors are master regulators of root meristem development (Aida et al., 2004; Blilou et al., 2005; Galinha et al., 2007). PLTs are part of a finely tuned feedback network, together with auxin and PINs, that is necessary to maintain stem cell activity and meristem growth (Blilou et al., 2005).
The plant hormone gibberellin is responsible for selective repression of ARR1 expression at early stages of meristem development (Moubayidin et al., 2010) (Fig. 4b). Gibberellins (GAs) are diterpene phytohormones that de-repress their signalling pathway by inducing degradation of DELLA domain containing (DELLA) proteins, which are master growth repressors, via ubiquitin–proteasome (Sun, 2008). Similarly to auxin, GAs have been shown to act as positive regulators of root growth and meristem size by sustaining cell division (Ubeda-Tomas et al., 2008, 2009; Achard et al., 2009). Exogenous application of GAs increases root meristem size and down-regulates ARR1 expression, without affecting ARR12 expression (Moubayidin et al., 2010). The DELLA protein REPRESSOR OF GA1–3 (RGA) has been shown to positively control ARR1 expression. Indeed, loss-of-function rga mutants display lower expression levels of both ARR1 and IAA3/SHY2, and present an enlarged root meristem, similar to arr1 and shy2-31 mutants (Moubayidin et al., 2010). Activation of ARR1 transcription at 5 dpg is mediated by RGA and results from a decrease in GA concentration (Moubayidin et al., 2010). The decrease in GA activity has been inferred to occur from the finding of a reduction in the expression of genes encoding rate-limiting enzymes of GA biosynthesis (Moubayidin et al., 2010). As early inhibition of GA biosynthesis allows ARR1 expression at 3 dpg, GA signalling is necessary and sufficient to regulate root meristem growth between 3 and 5 dpg.
Still to be elucidated are the molecular mechanisms that determine the switch in GA activity between 3 and 5 dpg. It is known that auxin signalling is capable of modulating GA response and biosynthesis (Fu & Harberd, 2003; Frigerio et al., 2006). The influence of cytokinin on auxin distribution could therefore feed back on the GA response, generating a positive feedback loop that would ultimately define root meristem size.
V. Concluding remarks
There are many lines of evidence demonstrating that the precise tuning of cytokinin signalling, which generates an elaborate spatiotemporal signalling landscape, is of great importance during plant development. The examples considered in this review underline the importance of the context specificity of the cytokinin response. Indeed, there are many lines of evidence showing that the effects of the cytokinin response change as organs develop, probably because the cellular and tissue environment where cytokinin is perceived has also changed.
In the establishment of a new SCN, cytokinin signalling switches from instructive to permissive according to the developmental stage (Muller & Sheen, 2008; Bielach et al., 2012). This shift has been connected to the ability of cytokinin to influence auxin distribution (Bielach et al., 2012). Indeed, it has been suggested that the stage-dependent effect of cytokinin on lateral root organogenesis could be attributable to the increasing stability of the auxin gradient during primordium development (Bielach et al., 2012). However, still to be elucidated are the possible instructive roles of positive cytokinin signalling, for example in the specification of stem cells during embryo development (Muller & Sheen, 2008).
Root meristem size is regulated by a complex network of interactions between hormones. Cytokinin and auxin act in opposite ways to establish the balance between cell differentiation and proliferation. Indeed, the cytokinin response factor ARR1 activates the expression of IAA3/SHY2, a repressor of auxin signalling, causing auxin redistribution and prompting cell differentiation. Moreover, GA antagonizes the cytokinin cell differentiation input by repressing ARR1. A decrease in GA concentration between 3 and 5 dpg results in ARR1 accumulation, increasing cell differentiation and allowing the root meristem to reach its final length. The differential expression of cytokinin signalling components could cause switches in other plant organs. For example, in the shoot apical meristem, several mechanisms operate in parallel to up-regulate cytokinin sensitivity to maintain meristem size (Leibfried et al., 2005; Gordon et al., 2009; Zhao et al., 2010). Moreover, similarly to the root meristem, the shoot meristem undergoes changes in terms of growth rate when it transits to the reproductive phase (Grandjean et al., 2004; Kwiatkowska, 2004, 2006; Reddy et al., 2004; Smith et al., 2004). It is known that cytokinins positively regulate the expression of the gene encoding the master regulator of shoot meristem stem cell identity, WUSCHEL (WUS) (Gordon et al., 2009). Although this is pure speculation, it is possible that, as in the establishment of root meristem length, a switch in cytokinin signalling could determine the change in shoot meristem size.
In recent years, interest in understanding the dynamicity of the role of cytokinin during plant development has grown. We continue to uncover genetic frameworks of tight interplay between cytokinin and other signalling pathways. These feedbacks could preside over the dynamicity of cytokinin function. The complexity of studying the integration of multiple varying signalling responses in growing organs is driving research towards the use of computational modelling to describe developmental events. Indeed, computational models have been used, for example, to test the outcome of the so-termed incoherent feedforward loop involving PHB/miR165A/ARR1, wherein cytokinin both represses and prevents the repression of PHB, which in turn feeds back to promote cytokinin biosynthesis (Dello Ioio et al., 2012). Indeed, the model shows that cytokinin-dependent miR165A regulation is needed to both dampen PHB reduction and accelerate PHB expression recovery in response to a temporary increase in cytokinin.
Great effort has been made to model the interaction between cytokinin and auxin. A subcellular model of the effects of auxin on the negative feedback loop of a generic Aux/IAA gene was developed and successively extended to account for cytokinin signalling (Middleton et al., 2010; Muraro et al., 2011). Such mechanistic models generate testable predictions regarding the emergence of periodic dynamics in hormonal concentrations along the root apex, and the effect of particular mutants. In particular, it has been suggested that high auxin concentrations can trigger a Hopf bifurcation, promoting the periodic variations observed at the TZ (Middleton et al., 2010). The subsequent increase in cytokinin concentrations, beyond the TZ, would cause a second Hopf bifurcation, leading the system to a stable steady state (Muraro et al., 2011). Moreover, it has been shown that, although gain-of-function mutations of IAA3/SHY2 can qualitatively reproduce the effect of varying auxin and cytokinin on their response genes, some elements of the network, such as Aux/IAA and phosphorylated type-A ARR protein levels, respond differently to changes in hormonal supply and to genetic mutations.
Subcellular models can be integrated at the macroscale level in a multidimensional layout aiming to represent the three-dimensional structure of the growing root. First, a two-dimensional model of auxin gradient formation in the root tip was developed, based on the tissue-specific distribution of PIN transporters (Grieneisen et al., 2007). This model establishes that a stable auxin maximum can be determined and maintained by a looped transport of auxin in which a basipetal reflux of auxin feeds back to the acropetal flow along the meristem, the so-called ‘reverse fountain’ flow. Cytokinin signalling was later integrated in a mono-dimensional model of the root, to investigate the cross-talk between auxin and cytokinin distributions in meristem zonation (Muraro et al., 2013). According to the authors, consistent with experimental evidence (Beemster & Baskin, 2000), this model shows that auxin application reduces the total length of the root by reducing that of the EDZ, while increasing the length of the PM. In contrast, an increase in cytokinin concentrations reduces the overall length of the root by decreasing cell growth rate and number. Interestingly, overexpression of PIN proteins is predicted to generate a longer root, with a longer EDZ, but a smaller PM and therefore fewer cells than the wild type.
Improved, more complex computational models are eagerly awaited, in order to gain further insights into the effects of hormonal cross-regulation on the spatiotemporal patterning of plant roots.
We would like to apologize to those authors whose valuable contributions could not be included because of space constraints. M.D.B. and L.G. are supported by the ERC.