Cytokinins were discovered in the 1950s by their ability to promote cell division in cultured plant cells. Recently, there have been significant breakthroughs in our understanding of the biosynthesis, metabolism, perception and signal transduction of this phytohormone. These advances, coupled with physiological and other approaches, have enabled remarkable progress to be made in our understanding of the interactions between cytokinin function and environmental inputs. In this review, we first highlight the most recent advances in our understanding of cytokinin biosynthesis, metabolism and signalling. We then discuss how various environmental signals interact with these pathways to modulate plant growth, development and physiology.
Cytokinins were first identified as factors that promoted the proliferation of cultured plant cells (Miller et al. 1955, 1956). Subsequently, this phytohormone was linked to numerous aspects of plant growth and development, including seed germination, vasculature development, meristem function, apical dominance and leaf senescence (Mok & Mok 2001). Much progress has been made recently in our understanding of cytokinin biosynthesis and signalling, and how environmental cues interact with these components to modulate plant growth, development and physiology. In this review, we will first discuss cytokinin biosynthesis, metabolism and signalling, and with an emphasis on recent findings, we will then discuss the role of cytokinin in the assimilation of various macronutrients, the crosstalk with light signalling and, finally, the interactions of cytokinin with abiotic and biotic factors. For additional information, the reader should see recent detailed reviews of cytokinin biosynthesis and metabolism (Sakakibara 2006; Hirose et al. 2008) and signalling (Ferreira & Kieber 2005; Choi & Hwang 2007; Müller & Sheen 2007; To & Kieber 2007).
The first committed and rate-limiting step of cytokinin biosynthesis, the transfer of an isopentenyl moiety from dimethylallyl diphosphate (DMAPP) to the N6 position of ATP/ADP (Fig. 1), is catalysed by the enzyme isopentenyltransferase (IPT) (Kakimoto 2001; Takei, Sakakibara & Sugiyama 2001a; Sakakibara 2006; Hirose et al. 2008). The immediate products of the IPT reaction are isopentenyladenine (iP) ribotides, the isoprene side chain of which is subsequently trans-hydroxylated by the P450 monooxygenases CYP735A1 and CYP735A2 to yield zeatin ribotides (Takei, Yamaya & Sakakibara 2004b). Cytokinin nucleotides can be converted to their most active free base forms via dephosphorylation and deribosylation. Recently, a gene encoding an enzyme catalysing one such reaction was identified in rice. The lonely guy (log) mutant was isolated in a screen for rice plants that displayed shoot meristem defects (Kurakawa et al. 2007). LOG encodes a phosphoribohydrolase that directly and specifically converts cytokinin-5′-monophosphates to the free base form (Fig. 1). LOG expression is localized to the tip of shoot meristems, and it likely fine tunes the spatial distribution of bioactive cytokinins to regulate meristem activity.
The first plant IPTs were identified in silico by their sequence similarity to the Agrobacterium tumefaciens tmr gene, and their identity was confirmed by the expression and analysis of the proteins in Escherichia coli (Kakimoto 2001; Takei et al. 2001a). IPTs are encoded by a small gene family that displays distinct spatial and sub-cellular distributions (Miyawaki, Matsumoto-Kitano & Kakimoto 2004; Takei et al. 2004a; Hirose et al. 2008). This suggests that cytokinin synthesis is localized in discrete sites throughout the plant, which is at odds with the classic notion that cytokinins are made primarily in roots and transported to shoots. Despite their non-overlapping patterns of expression, single and double loss-of-function mutants of ipt are aphenotypic in Arabidopsis; however, higher-order loss-of-function ipt mutants display reduced rosette development, reduced shoot meristem function and increased root growth (Miyawaki et al. 2006), phenotypes similar to transgenic plants engineered to have a reduced cytokinin content via overexpression of the cytokinin-degrading enzyme cytokinin oxidase/dehydrogenase (CKX) (Werner et al. 2003).
In addition to the regulation of its biosynthesis, the levels of active cytokinins are also regulated by conjugation to sugars and by degradation. Cytokinins can be irreversibly inactivated by conjugation to glucose at the N3, N7 and N9 positions of the adenine ring, and can be reversibly modified by conjugation to glucose and, to a lesser extent xylose, to the hydroxyl group of the side chain (Sakakibara 2006). Several of the enzymes catalysing these reactions have been identified (Martin, Mok & Mok 1999a,b; Martin et al. 2001). Overexpression of a zeatin O-glucosyltransferase in maize leads to phenotypes resembling those of cytokinin-deficient plants, consistent with a role for this enzyme in inactivating cytokinins (Pineda Rodo et al. 2008).
Cytokinins with unsaturated isoprenoid side chains (Z- and iP-type cytokinins) can be irreversibly degraded by cleavage of the N6 side chain by the enzyme CKX. CKX is encoded by a multigene family that is differentially regulated by biotic and abiotic factors (Brugiere et al. 2003; Werner et al. 2003, 2006). Major quantitative trait loci (QTL) that are responsible for increasing the number of reproductive organs in the inflorescence meristem, and hence grain number, of the indica variety of rice were identified as a loss-of-function mutation in a CKX gene, OsCKX2 (Ashikari et al. 2005). Reduced expression of this cytokinin oxidase gene causes an increase in the cytokinin content in the inflorescence, which leads to the increase in the number of reproductive organs.
In sum, the steady-state level of active cytokinins is determined by the relative rates of biosynthesis and release of the conjugated forms and the rates of conjugation and degradation.
Elements of the cytokinin signalling pathway share homology to bacterial two-component signalling cascades, by which bacteria perceive extracellular stimuli via a membrane-bound receptor kinase and a cystolic response regulator (RR). In plants, the cytokinin two-component system is a multi-step phospho-relay, consisting of hybrid sensor kinase receptors (HKs), histidine phosphotransfer proteins (HPs) and RRs (Fig. 1) (Ferreira & Kieber 2005; Müller & Sheen 2007; To & Kieber 2007). In Arabidopsis, these proteins are referred to as the Arabidopsis hybrid sensor kinase receptors (AHKs), Arabidopsis histidine phosphotransfer proteins (AHPs) and Arabidopsis response regulators (ARRs), respectively.
The cytokinin receptors
The first cytokinin receptor (AHK4/CRE1/WOL) was isolated as a mutant that failed to form large green calli on shoot-initiation media and/or whose function in heterologous system was dependent on applied cytokinin (Inoue et al. 2001; Suzuki et al. 2001a; Ueguchi et al. 2001). Two additional cytokinin receptors (AHK2 and AHK3) are present in Arabidopsis (Higuchi et al. 2004; Nishimura et al. 2004). The AHKs bind active cytokinins with high affinity and are able to complement yeast and E. coli histidine kinase mutants in a cytokinin-dependent manner (Inoue et al. 2001; Suzuki et al. 2001a; Yamada et al. 2001). The three AHKs display kinase activity when bound to their ligand; however, in the absence of cytokinin, AHK4 (but not AHK2 nor AHK3) displays phosphatase activity (Mähönen et al. 2006b). The ability to regulate kinase and phosphatase activities may provide multiple strategies to modulate cytokinin responses.
The AHPs are a small family of proteins that act as intermediates in cytokinin signalling. The AHPs interact directly with various histidine sensor kinases and type-A and type-B RRs (see further discussion) in yeast two-hybrid assays (Urao et al. 2000; Suzuki et al. 2001b; Tanaka et al. 2004; Dortay et al. 2006). Genetic analyses indicate that the AHPs are partially redundant positive elements in cytokinin signalling (Hutchison et al. 2006). The quintuple ahp1,2,3,4,5 mutant has a stunted primary root phenotype very similar to that of the triple receptor mutant (Hutchison et al. 2006). AHP6, a pseudo HP that lacks the His residue that is the target of phosphorylation, acts as an inhibitor of cytokinin signalling (Mähönen et al. 2006a) that functions to facilitate protoxylem specification by interfering with cytokinin signalling in a spatially defined manner (Mähönen et al. 2006a).
The type-A RRs
The type-A RRs are members of a large gene family whose transcription is rapidly elevated by exogenous cytokinin (Brandstatter & Kieber 1998; D'Agostino, Deruère & Kieber 2000; Asakura et al. 2003; Jain, Tyagi & Khurana 2006). In addition to transcriptional regulation, cytokinin treatment also results in an increase in the half-life of a subset of type-A ARR proteins (To et al. 2007). Type-A RRs contain a single receiver domain that is phosphorylated on a conserved Asp by the upstream AHPs. Genetic analysis in Arabidopsis indicates that type-A ARRs are partially redundant negative regulators of cytokinin signalling (To et al. 2004) that act through phospho-dependent interactions with as yet to be identified target proteins (To et al. 2007).
Disruption of subsets of type-A ARR genes affects multiple aspects of development, including rosette size, root elongation, phyllotaxy, meristem function and embryo development (To et al. 2004; Leibfried et al. 2005; Müller & Sheen 2008). Four type-A ARRs are directly repressed by the meristem identity transcription factor WUSHEL, and the disruption of these type-A ARRs leads to defects in the shoot apical meristem function (Leibfried et al. 2005). Similarly, in maize, the disruption of a single type-A RR leads to alterations in leaf phyllotaxy and an increased size of the meristem (Giulini, Wang & Jackson 2004). Two type-A ARR genes have recently been shown to be repressed by auxin and to play a role in root stem cell specification in the early embryo (Müller & Sheen 2008).
Other transcription factors involved in cytokinin responses
In addition to the type-B ARRs, there are several other transcription factors that have been implicated in the response to cytokinin. The cytokinin response factors (CRFs) act, along with the type-B ARRs, to mediate the transcriptional response to cytokinin (Rashotte et al. 2006). The six CRFs are a subset of the APETALA2 (AP2)-like superfamily. Three of the CRFs are induced in response to exogenous cytokinin in a type-B ARR-dependent manner (Rashotte et al. 2003, 2006). A subset of cytokinin-regulated genes are misexpressed in crf loss-of-function mutants. Recently, a subgroup of the plant-specific transcription factor family GLABROUS1 enhancer-binding protein [GeBP, GPL (GeBP-like)] has been implicated in cytokinin responsiveness (Chevalier et al. 2008). The disruption of three closely related members of this family results in an elevated level of type-A ARR gene expression and reduced sensitivity to cytokinin specifically in the shoot. The GeBP/GPL proteins likely act to antagonize the negative feedback of the type-A ARRs.
CYTOKININ AND THE ASSIMILATION OF MACRONUTRIENTS: IT'S ELEMENTAL MY DEAR WATSON
The abilities to sense nitrogen levels and to regulate nitrogen uptake are essential for plant growth and reproduction. Plants obtain nitrogen from inorganic forms present in the soil, the most abundant of which is nitrate (NO3-). The perception and assimilation of nitrate is dynamically regulated in planta (Stitt 1999). Once absorbed from the soil, nitrate ions are translocated to the shoot, where they are reduced to nitrogen-containing organic compounds, including amino acids and proteins. Nitrate levels regulate the expression of genes involved in its assimilation and reduction, such as nitrate transporters, nitrate reductase, glutamine synthetase and glutamate synthase, as well as the enzyme activities of their encoded proteins (reviewed in Sakakibara, Takei & Hirose 2006). The level of nitrate also correlates with the expression of genes involved in sugar metabolism, sulphur assimilation and secondary metabolism, reflecting the importance of nitrate as a principal regulator of nitrogen assimilation and plant metabolism (Forde 2002).
The level of nitrate available to the plant regulates the endogenous concentration of cytokinins. Plants grown on low levels of nitrogen show reduced levels of cytokinin, and the addition of nitrate leads to an increase in the levels of various cytokinin species (Salama & Waering 1979; Samuelson & Larsson 1993; Takei et al. 2001b), which in turn leads to the up-regulation of the cytokinin-responsive type-A RRs (Sakakibara et al. 1998; Taniguchi et al. 1998; Kiba et al. 1999). Nitrate addition up-regulates cytokinin levels in part by inducing expression of cytokinin biosynthetic genes. AtIPT3 and AtIPT5 expression are up-regulated after nitrate treatment, predominantly in the roots (Miyawaki et al. 2004; Takei et al. 2004a). However, the addition of ammonium to N-starved plants leads to an increase in the levels of AtIPT5 but not AtIPT3, indicating that the regulation of cytokinin biosynthesis depends on the forms of nitrogen available to the plant (Takei et al. 2004a). The expression of CYP735A2 (Fig. 1) is also regulated by nitrate levels (Takei et al. 2004b; Wang et al. 2004).
Recent microarray analyses have revealed that the expression of the nitrate- and cytokinin-inducible low-affinity nitrate transporter gene NRT1.3 is compromised in a type-B arr1,10,12 triple mutant (Argyros et al. 2008), suggesting that the expression of this nitrate-responsive gene is, in part, regulated by cytokinin signalling components. Genome-wide microarray analysis revealed that treatment with either cytokinin or nitrate induces the expression of various genes involved in primary metabolism (Wang et al. 2000, 2003, 2004; Rashotte et al. 2003; Brenner et al. 2005; Kiba et al. 2005), and a significant overlap among the sets of genes induced by each treatment is observed (Sakakibara et al. 2006).
The correlation among nitrate, cytokinin levels and their effects on gene expression has led to the suggestion that cytokinin can act as a root to shoot signal to regulate tissue-specific nitrogen metabolism (Takei et al. 2001b, 2002; Gessler, Kopriva & Rennenberg 2004; Sakakibara 2006). Whereas iP is the most abundant cytokinin species present in leaf exudates and in the phloem (Weiler & Ziegler 1981; Emery & Atkins 2002; Corbesier et al. 2003), zeatin riboside is transported in the xylem (Beveridge et al. 1997; Takei et al. 2001b). Along with the localized expression of AtIPT3 in the phloem (Miyawaki et al. 2004) and the high levels of expression of CYP735A2 in the roots and stems, but low CYP735A2 levels in the leaves (Takei et al. 2004b), these data suggest a model in which increased nitrate in the roots leads to the induction of AtIPT3 and CYP735A2 and thus the biosynthesis of zeatin ribosides in the root, which can subsequently be transported to the shoot via the xylem. In the shoot, up-regulation of AtIPT3 and possibly AtIPT5 would lead to the accumulation of iP cytokinins, which are translocated via the phloem to other parts of the plant. The balance of different cytokinin species, perceived by two-component elements, would signal the availability of nitrate forms, and would ultimately lead to the expression of cytokinin-responsive and, potentially, also nitrate-responsive genes to regulate nitrogen metabolism in the plant (Sakakibara et al. 2006).
Phosphorus is a macronutrient that is essential for many biochemical reactions, and its availability as orthophosphate (Pi), the form of phosphorus that is preferentially taken up by plants, is limited in most agricultural soils. In response to phosphate starvation, plants undergo various forms of adaptation that are reminiscent of responses to alterations in the levels of cytokinin, including an increase in the number of lateral roots and an increase in the root : shoot ratio (Raghothama 1999).
A role for cytokinins in the Pi-starvation response has been suggested based on evidence that cytokinin levels are reduced in Pi-starved plants (Salama & Waering 1979; Horgan & Waering 1980; Wagner & Beck 1993). Consistent with this, the Arabidopsis pho1 and pho2 mutants, which fail to accumulate and hyper-accumulate Pi in shoots, respectively, show altered sensitivity to cytokinin (Lan, Li & Fischer 2006).
Pi starvation leads to complex changes in gene expression (Hammond et al. 2003; Wu et al. 2003). An initial transient change in the expression of genes encoding general stress response factors is observed, followed by the induction of genes directly involved in the response to Pi starvation (Hammond et al. 2003). In general, cytokinins down-regulate Pi starvation-responsive genes (Martin et al. 2000; Hou et al. 2005). Microarray analysis of rice plants under Pi starvation not only confirmed these results but also identified genes that were up-regulated or unchanged upon cytokinin addition, indicating that the effect of cytokinin in Pi-starvation gene expression is complex (Wang et al. 2006).
The ahk3,4 mutant is defective for the cytokinin repression of gene expression in the local response to Pi starvation but is unaffected in the systemic response (Martin et al. 2000; Franco-Zorrilla et al. 2005). This suggests that while AHK3 and AHK4 are necessary for the local response, AHK2 may play an important role in the systemic response, perhaps redundantly with AHK3. Consistent with this model, AHK4 is primarily expressed in roots and AHK2 and AHK3 most highly in shoots (Higuchi et al. 2004; Nishimura et al. 2004). The expression of AHK4 was found to be repressed by the addition of Pi, indicating a negative feedback regulatory role for cytokinin (Franco-Zorrilla et al. 2002).
Sulphur is assimilated by plant roots in its inorganic, anionic sulphate form. Upon assimilation, sulphate is reduced to sulphide in plastids, giving rise to the sulphur-containing amino acid cysteine, which is either directly incorporated into proteins and glutathionine or converted into the amino acid methionine. In addition, sulphur acts as a cofactor for various enzymes, is involved in the modification of proteins, carbohydrates and lipids, and is essential for the formation of secondary metabolites involved in defence responses to pathogens and herbivores (Leustek et al. 2000).
Sulphate-responsive genes are up-regulated in response to cytokinin in either sulphur-depleted or non-depleted conditions and are only marginally up-regulated by other plant hormones (Ohkama et al. 2002). The expression of APR1, which encodes an enzyme involved in the key step of sulphate assimilation, is induced by cytokinin (Ohkama et al. 2002). The genes encoding sulphate transporters are also regulated by cytokinin (Maruyama-Nakashita et al. 2004), as well as by sucrose and/or nitrate (Ohkama et al. 2002; Rouached et al. 2008). However, the concentration of cytokinins is not altered after sulphate starvation, and the application of cytokinin does not change the concentration of O-acetyl-L-serine (Ohkama et al. 2002), a cysteine biosynthetic precursor that acts as a positive regulator of sulphate starvation-responsive genes (Kim et al. 1999; Hirai et al. 2003), suggesting that the action of cytokinin in the regulation of sulphate uptake and sulphate-responsive genes is most likely indirect.
Iron is a micronutrient required in small quantities for plant growth. In alkaline soils, iron can become limiting, leading to chlorosis and other defects. Iron deficiency triggers the induction of genes involved in iron uptake, including genes encoding high-affinity iron transporters. The expression of these genes is regulated primarily at the level of transcription, and a basic helix-loop-helix (bHLH) transcription factor, called FIT1 in Arabidopsis, is involved in this regulation (Briat, Curie & Gaymard 2007). A recent report provides evidence that cytokinins act to negatively regulate the expression of a subset of iron-responsive genes (Séguéla et al. 2008). This repression requires the AHK3 and AHK4 receptors but is independent of iron status and of FIT1. A transient rise in AtIPT3 and type-A ARR gene expression occurred in response to iron resupply to iron-starved plants (Séguéla et al. 2008), which is reminiscent of the induction of these same genes in response to nitrogen resupply. It was found that other factors that inhibit root growth, such as mannitol and NaCl, also repress the iron-starvation response genes (Séguéla et al. 2008). The authors propose that cytokinin down-regulates iron-responsive gene expression through a growth-dependent pathway (Séguéla et al. 2008), which may underlie the effect of cytokinin on other nutrient assimilation pathways. Further studies that carefully examine the timing of the effects of cytokinin on root growth and nutrient-regulated genes expression, coupled with a better understanding of the transcription circuits regulating iron-responsive gene expression, should help clarify this issue.
THE BRIGHT SIDE OF LIGHT
The response pathways for light and cytokinin are intertwined in several contexts. The profound changes that occur in etiolated seedlings in response to light, called photomorphogenesis, can be partially mimicked by growth of seedlings in the presence of exogenous cytokinin or by elevation of endogenous cytokinin (Chory et al. 1994; Lochmanova et al. 2008). The cytokinin-induced de-etiolation is absent in a type-B arr1,10,12 triple mutant (Argyros et al. 2008), indicating that the canonical two-component response pathway is necessary for this response. While these studies demonstrate that increased cytokinins are sufficient to partially mimic photomorphogenesis in etiolated seedlings, they do not address if they are involved in other responses to light.
A key mediator of many light responses is the red light photoreceptor phytochrome. Sweere et al. found that the type-A RR ARR4 stabilized the active, Pfr form of one phytochrome, PhyB, by reducing the rate of dark reversion, and thus ARR4 acts as a positive regulator of the PhyB function (Sweere et al. 2001). This effect required a phosphorylatable form of ARR4 and depended on the AHK cytokinin receptors (Mira-Rodado et al. 2007). In contrast, loss-of-function mutations in arr3, arr4, arr3,4 and the arr3,4,5,6 quadruple mutants were more sensitive to red light as compared to wild-type seedlings (To et al. 2004), suggesting that these type-A ARRs, including ARR4, are negative regulators of the PhyB function. These conflicting data may reflect differences in the growth conditions used in these two studies, as Mira-Rodado et al. found that the arr4 loss-of-function mutant was hyposensitive to red light in their growth conditions (Mira-Rodado et al. 2007). Nevertheless, these results implicate multiple type-A ARRs in the regulation of phytochrome function.
Another point of convergence between light signalling and cytokinin occurs via the HY5 protein. This transcription factor is a positive regulator of photomorphogenesis, acting downstream of multiple families of photoreceptors, including the phytochromes and cryptochromes (Chang et al. 2008). Intriguingly, hy5 mutants are partially insensitive to cytokinin in root elongation and callus initiation assays (Cluis, Mouchel & Hardtke 2004). Recent studies have demonstrated that cytokinin can increase the abundance of HY5 protein but not its transcript levels, suggesting that cytokinin increases HY5 protein stability (Vandenbussche et al. 2007). HY5 was also found to be necessary for the induction of anthocyanin in response to cytokinin in blue light but not for the inhibition of hypocotyl elongation by cytokinin in the dark (Vandenbussche et al. 2007). These results suggest that a subset of cytokinin responses may be mediated through HY5.
Recently, the regulation of cytokinin degradation by CKX has been linked to the response to shade in Arabidopsis. Plants grown under canopy shade receive a low ratio of red/far red light (R/FR), which induces a number of developmental changes including increased hypocotyl elongation and a rapid arrest of leaf primordia growth (Carabelli et al. 2007). Carabelli et al. found that exposure to low R/FR resulted in a rapid increase in auxin signalling in the leaf primordia, which in turn led to the elevation of expression of AtCKX6 expression (Carabelli et al. 2007). This is hypothesized to result in the reduction of cytokinin levels in the leaf primordia, and hence a reduction in cell proliferation. Consistent with this, a loss-of-function ckx6 mutant did not display the arrest of leaf primordia growth in response to low R/FR conditions that was observed in wild-type seedlings (Carabelli et al. 2007).
Elements of the two-component cytokinin response pathway play a role in the circadian clock, which entrains plants to light/dark diurnal cycles. The first evidence for this came from the genetic analysis of type-A ARR mutants. The arr3,4 and arr3,4,5,6 mutants had a longer circadian period and a leading phase change similar to that displayed in phyB mutants (Saloméet al. 2005). The changes in circadian rhythms were absent in the arr3 and arr4 single mutants, suggesting that these two RRs have redundant functions in this response (Saloméet al. 2005). Surprisingly, the altered periodicity of the arr3,4 mutant was suppressed by a reduction in ARR8 and ARR9 function, even though no photoperiod phenotype was observed in the arr8,9 mutant itself. This suggests that these pairs of type-A ARRs can act antagonistically in this response. The finding that the introduction of an ARR5 genomic transgene into the arr3,4,5,6 mutant fully rescued the cytokinin hypersensitivity (To et al. 2004, 2007), but did not affect the altered periodicity of the circadian rhythm (Saloméet al. 2005), suggests that ARR3 and ARR4 act independently of cytokinin in this context. Furthermore, ARR4 displays both PhyB-dependent and -independent effects on the circadian clock (Saloméet al. 2005; Hanano et al. 2006). Intriguingly, the expression of the ARR9 gene (but none of the other two-component genes) displays a strong circadian oscillation, and the timing of this expression is controlled by the major clock components [CIRCADIAN CLOCK ASSOCIATED 1 (CCA1), LATE ELONGATED HYPOCOTYL 1 (LHY1), TIMING OF CAB EXPRESSION 1 (TOC1)] (Ishida, Yamashino & Mizuno 2008a). Cytokinin treatment was found to cause a shift in the phase of the circadian clock (Hanano et al. 2006; Zheng et al. 2006) in an ARR4- and PhyB-dependent manner (Zheng et al. 2006). A mechanistic link for this effect is suggested by the observation that cytokinin induces the expression of LHY and CCA1 genes but represses the expression of TOC1 (Zheng et al. 2006). Thus, there appears to be an interdependent regulatory loop between the clock genes and cytokinin response genes (i.e. ARR9). Further studies should help define the mechanisms underlying this intriguing interaction among light, cytokinin and circadian rhythm.
IN TIMES OF TROUBLE
Cytokinin function has been linked to a variety of abiotic stresses (Hare, Cress & van Staden 1997). The examination of public microarray expression data reveals that the genes encoding the proteins in the cytokinin signalling pathway are differentially affected by various abiotic stresses (Fig. 2). For example, cold stress appears to rapidly up-regulate the expression of multiple type-A ARRs and conversely to down-regulate the expression of all three cytokinin receptors (Fig. 2). This suggests a role for cytokinin in the response to cold stress, but there are no reports linking cytokinin to a rapid response to cold stress.
The expression of the AHK2 and AHK3 genes was found to be induced after dehydration (Tran et al. 2007), which is also observed in the public microarray data (Fig. 2). The elevation of these cytokinin receptors could lead to an increase in the sensitivity to cytokinin. The exposure of plants to drought results in a decrease in the level of cytokinins in the xylem sap (Bano et al. 1994; Shashidhar, Prasad & Sudharshan 1996). A recent study has confirmed that isoprene-type cytokinins (zeatin and zeatin riboside) are decreased in the xylem in response to drought stress but surprisingly found that the level of the aromatic cytokinin 6-benzylaminopurine (BAP) was elevated (Alvarez et al. 2008). The increased level of aromatic cytokinin could inhibit leaf senescence during drought conditions and/or might increase the level of proline (Alvarez et al. 2008), which acts as an osmotic protectant in response to water stress.
Consistent with the notion that elevated cytokinin levels may promote survival in drought conditions, a recent study found that the expression of Agrobacterium ipt from a drought/maturation-induced promoter [SENESCENCE ASSOCIATED RECEPTOR PROTEIN KINASE (SARK)] resulted in a remarkable tolerance to extreme drought conditions in tobacco (Rivero et al. 2007). Transgenic plants showed almost complete recovery following a drought regime that killed wild-type plants. Moreover, no loss in yield was reported under water restriction (Rivero et al. 2007). This is consistent with the finding that BAP levels are elevated in conditions of water deficit (Alvarez et al. 2008) and suggests that endogenous cytokinin may play a role in conferring drought tolerance.
The treatment of Arabidopsis with either osmotic or salt stress has a strong effect on the expression of the AHK cytokinin receptors (Fig. 2). AHK2 and AHK4 were down-regulated, both in the root and the shoot, in response to osmotic or salt stress. Conversely, AHK3 was up-regulated in response to these conditions (Fig. 2). Consistent with this, an AHK3 orthologue from Medicago sativa was found to be induced by salt stress (Coba de la Peña et al. 2008). This altered suite of AHK receptor expression may have important effects on receptor output following exposure to cytokinin. Consistent with an important role of these AHKs in the response to salt stress, the disruption of AHK2 and/or AHK3 resulted in increased tolerance to drought and salt stresses (Tran et al. 2007). These data suggest that the AHKs may act as negative regulators in stress signalling. Consistent with this, ahk2,3 double mutants displayed constitutively elevated expression of stress-responsive genes (Tran et al. 2007). Interestingly, AHP4, but none of the other AHPs, is down-regulated in response to salt and osmotic stress (Fig. 2). AHP4 is evolutionarily distinct from the other AHPs, and, in contrast to the other functional HPs in Arabidopsis, may play a negative role in cytokinin signalling in some context (Hutchison et al. 2006). The expression of several of the CRF genes, which were identified as cytokinin-responsive AP2 transcription factors (Rashotte et al. 2006), is down-regulated in response to salt stress, especially in roots. These genes may play an important role in mediating the input of cytokinin into the salt stress response pathway.
CYTOKININ AND PATHOGENS: OUT, DAMN'D SPOT!
After pathogen infection, plants initiate a series of defence responses in order to contain the invader. When defence is not possible, disease develops, leading to a complex and intimate association in which plants try to minimize pathogen damage, while pathogens try to obtain nutrients to grow and complete their life cycle.
Several lines of evidence support a possible link between cytokinin and pathogenicity. One line of evidence comes from observations that cytokinin levels in plants are altered after pathogen infection. For example, in beans (Phaseolus vulgaris), the levels of active cytokinins are decreased after inoculation with the viral pathogen white clover mosaic potexvirus (Clarke et al. 1999). In wheat plants, infection with the fungal pathogen Tilletia caries leads to an increase in the cytokinin levels to twice the levels seen in uninfected plants (Maksimov, Ganiev & Khairullin 2002). Moreover, the application of exogenous cytokinin leads to changes in pathogen susceptibility. For example, when bean seedlings are treated with exogenous cytokinin, the level of replication of white clover mosaic potexvirus is reduced, accompanied by an induction of defence-response genes, such as those encoding pathogenesis-related (PR) proteins (Clarke, Burritt & Guy 1998). Up-regulation of stress- and defence-related genes also occurs in cytokinin-treated seedlings and plant cell cultures (Schäfer et al. 2000; Rashotte et al. 2003; Brenner et al. 2005) and in tomato plants overexpressing the Agrobacterium ipt gene (Martineau et al. 1994). Further evidence of crosstalk between cytokinin and defence-response pathways was provided by the characterization of the uni1-d mutant in Arabidopsis. uni1-d carries a mutation in a Coiled Coil-Nucleotide Binding domain-Leucine-Rich Repeat (CC-NB-LRR) resistance protein and shows increased cytokinin sensitivity and up-regulation of defence-related genes (Igari et al. 2008).
While there appears to be a link among cytokinin levels, pathogenicity and defence-related gene expression, the sources and species of cytokinins involved in specific plant–pathogen interactions is uncertain. Several pathogens, such as the bacteria A. tumefaciens and the biotrophic actinomycete Rhodococcus fascians, are able to synthesize cytokinins, which is important for their pathogenicity (Crespi et al. 1992; Sakakibara et al. 2005). The analysis of Arabidopsis plants overexpressing cytokinin oxidases indicates that both plant- and pathogen-derived cytokinins play a role in disease development in the R. fascians interaction (Depuydt et al. 2008). Plasmodia of the obligate biotrophic pathogen Plasmodiophora brassicae, causal agent of clubroot disease, can produce cytokinin (Muller & Hilgenberg 1986), and the infection of Arabidopsis plants with this pathogen leads to the up-regulation of ARR5 and differential regulation of genes encoding cytokinin oxidases, suggesting that pathogen-derived cytokinins are involved in disease development (Devos et al. 2006; Siemens et al. 2006). Moreover, transgenic plants overexpressing cytokinin oxidases show decreased disease symptoms indicating a role for cytokinins in P. brassicae pathogenicity (Siemens et al. 2006).
Several physiological functions attributed to cytokinins may explain their role in plant–pathogen interactions. The interaction of some biotrophic pathogens and their hosts leads to the formation of green bionissia (formerly known as green islands), which are sites of green living tissue surrounding the sites of active pathogen growth (Walters, McRoberts & Fitt 2008). The formation of these green bionissia is correlated with elevated levels of cytokinins in these tissues, as for example, in the interaction of barley and the fungal pathogen Blumeria graminis (Coghlan & Walters 1992). It is commonly thought that cytokinins delay the onset of senescence in green bionissia, allowing pathogens to obtain more nutrients from the plant (Walters et al. 2008). The effects of cytokinin in nutrient assimilation and allocation within the plant, previously described in this review and elsewhere (Ashby 2000; Walters & McRoberts 2006), might also play an important role, especially for biotrophic pathogens, which obtain their nutrients from living plant cells.
Nevertheless, necrotrophic pathogens, or those that obtain their nutrients from dead plant cells, are also able to cause the formation of green tissue around the sites of infection, which have been named green necronissia (Walters et al. 2008). For instance, the inoculation of the fungal pathogen Pyrenophora teres on barley plants can cause the formation of green necronissia, and the application of exogenous cytokinin is able to mimic this effect (Angra & Mandahar 1991). Similar results were observed in maize leaves infected with Dreschlera maydis (anamorph of Cochliobolus heterostrophus) (Angra-Sharma & Sharma 1999). In contrast to its role in the formation of green bionissia, a role for cytokinin in these interactions might occur through the regulation of cell death (Carimi et al. 2003). A specific role for cytokinin in the pathogenicity of necrotrophic pathogens might also be exerted through the control of senescence. Because these pathogens use toxins and cell wall-cleaving enzymes to cause cell damage and death, the control of plant cell senescence might be correlated to the pathogen's ability to cause disease. This could be the case in the relationship between Arabidopsis and the necrotrophic fungus Botrytis cinerea, which, although unable to synthesize cytokinins, induces senescence in plant cells in a manner that can be reverted by the overexpression of a bacterial ipt (Swartzberg et al. 2008).
CYTOKININ AND NODULATION (NOD): FIXING TO GROW?
Nitrogen-fixing rhizobacteria play an important role in the control of available nitrogen in the soil. Through the process of nitrogen fixation, these bacteria are able to convert N2 into organic forms that can be assimilated by the plants. The bacteria obtain dicarboxylic acids from the plant in exchange for ammonia produced during nitrogen fixation, an energetically costly process that is tightly regulated in legume plants.
The process of nitrogen fixation occurs in specialized structures in the legume root called nodules. In the course of nodule organogenesis, the bacteria induce morphogenic alterations in the roots of their legume plant hosts, a process that requires extensive modification of the root epidermis for infection initiation and subsequent activation of cell division leading to nodule formation, and in which Nod factors produced by rhizobial bacteria play a crucial role.
Cytokinins have been suggested to play a role in Nod based on various lines of evidence, including their ability to induce cell division. Some nitrogen-fixating bacteria, such as Rhizobium leguminosarum and Bradyrhizobium japonicum, produce compounds with cytokinin-like activity that could act to induce cell division in the host during Nod (Phillips & Torrey 1972; Sturtevant & Taller 1989). Application of exogenous cytokinin leads to the induction of cortical cell divisions (Torrey 1961) and up-regulation of early nodulin genes (Dehio & Debruijn 1992; Mathesius et al. 2000). The expression of the Agrobacterium ipt gene in a Rhizobium Nod- (Nod defective) strain conferred the ability to induce nodule-like structures on alfalfa roots. Moreover, the expression of the same gene in E. coli led to the formation of cell division clusters indicative of nodule formation, suggesting that cytokinin is sufficient for some aspects of Nod, at least in some species (Cooper & Long 1994).
The expression of two-component elements involved in cytokinin signalling is correlated to the Nod process. An ARR5 promoter-GUS reporter was rapidly induced in response to rhizobial infection in Lotus japonicus in the deformed root hairs, as well as in nodule primordia, but declined as nodule development progressed (Lohar et al. 2004). In Medicago truncatula, the expression of MtCRE1 (an orthologue of CRE1/AHK4) and type-A MtRR genes, is elevated upon inoculation in the areas of nodule development (Gonzalez-Rizzo, Crespi & Frugier 2006). Moreover, the disruption of the MtCRE1 gene results in a failure to initiate cortical cell divisions that are necessary for Nod in M. truncatula (Gonzalez-Rizzo et al. 2006) and L. japonicus (Murray et al. 2007), while a gain-of-function mutation in the same receptor leads to the spontaneous formation of root nodules in the absence of rhizobia (Tirichine et al. 2007). Together, these studies demonstrate that cytokinin function is necessary and sufficient for Nod.
The mode of action of cytokinin in Nod is still unclear and most likely involves interplay among different factors. Cytokinin could play an important role in the regulation of the transcription of genes involved in the re-differentiation of cortical cells, as genes encoding transcription factors with roles in Nod, such as NIN and NSP2 (Schauser et al. 1999; Oldroyd & Long 2003), are regulated by cytokinin, with NIN showing a kinetics and spatial pattern of expression similar to that of RRs (Grønlund et al. 2005; Gonzalez-Rizzo et al. 2006). Another interesting possibility is that cytokinin might play a role in the regulation of the Ca++ spiking in response to Nod (Ehrhardt, Wais & Long 1996), possibly through the action of cytokinin-regulated calcium-dependent protein kinases (CDPKs) such as MtCPK3 in M. truncatula (Gargantini et al. 2006). Furthermore, cytokinin might also play an important role in regulating the sink/source relationships and nitrogen metabolism in nodules to coordinate the exchange of nutrients between the rhizobial cells within the nodule and the host plant.
In sum, while it is clear that cytokinin likely plays a role in the response to many environmental signals, much remains to be clarified. Physiological studies have correlated changes in cytokinin levels to responses to environmental cues. Genome-wide microarray studies reveal overlapping transcriptional responses between cytokinin and various environmental inputs. The components of the cytokinin biosynthetic and signalling pathway are, in turn, transcriptionally altered by environmental perturbations. While many of the links of cytokinin to various environmental stimuli have been simply correlative, recent studies using mutants that alter cytokinin biosynthesis or signalling have begun to demonstrate an important role for cytokinins in these responses. However, because of the wide range of outputs of the cytokinin signalling pathway, dissecting the role of cytokinin in the response to a particular stress remains challenging. For example, ethylene has been implicated in the response to both biotic and abiotic stresses, and cytokinin elevates ethylene biosynthesis (Vogel et al. 1998). This is a complicating factor in dissecting the role of cytokinin in the response to a particular stress, as it is unclear if the effects of cytokinin are direct or the result of altered ethylene levels. Nevertheless, it is evident that cytokinin plays an important role in the response to a diverse array of environmental inputs. Continued analysis using the large number of tools now available to alter cytokinin levels and responsiveness will continue to shed light on this subject. As we further our understanding of the circuitry underlying the input of cytokinin into the response to various environmental signals, we should be able to engineer these pathways to produce plants with increased tolerance to biotic and abiotic stresses.
We thank Jayson Punwani and Jennifer To for helpful discussions and comments on the manuscript. This work was supported by funding from the National Science Foundation (NSF) (#IOS-0618286) and Department of Energy (DOE) (#DE-FG02-05ER15669).