A Gain-of-Function Mutation in a Cytokinin Receptor Triggers Spontaneous Root Nodule Organogenesis

See allHide authors and affiliations

Science  05 Jan 2007:
Vol. 315, Issue 5808, pp. 104-107
DOI: 10.1126/science.1132397


Legume root nodules originate from differentiated cortical cells that reenter the cell cycle and form organ primordia. We show that perception of the phytohormone cytokinin is a key element in this switch. Mutation of a Lotus japonicus cytokinin receptor gene leads to spontaneous development of root nodules in the absence of rhizobia or rhizobial signal molecules. The mutant histidine kinase receptor has cytokinin-independent activity and activates an Escherichia coli two-component phosphorelay system in vivo. Mutant analysis shows that cytokinin signaling is required for cell divisions that initiate nodule development and defines an autoregulated process where cytokinin induction of nodule stem cells is controlled by shoots.

Differentiated plant cells have an unusual capacity for rejuvenating by dedifferentiation and subsequent differentiation to form new organs or complete plants. In the model legume Lotus japonicus (lotus), nodule organogenesis is initiated by dedifferentiation of root cortical cells followed by cell proliferation, establishing a cluster of meristematic cells that give rise to the nodule primordium. The developmental process is triggered by compatible Mesorhizobium loti bacteria synthesizing lipochitin-oligosaccharide nodulation factor (Nod factor) acting as a mitogen and/or morphogen when recognized by the host plant Nod factor receptors, NFR1 and NFR5 (1, 2). Bacterial invasion of primordia occurs via infection threads progressing through root hairs into the root cortex. Ultimately, rhizobia released from infection threads are endocytosed into cells, which become the infected nitrogen-fixing nodule cells. At the same time, pattern formation and cell differentiation specify tissue and cell types of the new specialized organ, which in turn supplies the plant with nitrogen fixed by endocytosed bacteria.

To dissect the genetic regulation of cellular dedifferentiation and meristem formation, we isolated lotus mutants developing root nodules spontaneously. The snf2 (spontaneous nodule formation) mutants develop white rhizobia-free nodules in the absence of M. loti (Fig. 1, A and B). Detailed histological analysis of nodule sections demonstrates that spontaneous nodules are genuine nodules with an ontogeny and physiology similar to rhizobially induced nodules (3). The snf2 allele is monogenic dominant, and inoculation of snf2 mutants with M. loti results in development of normal nitrogen-fixing root nodules, which strongly suggests the presence of a gain-of-function mutation in this allele. Genetic mapping of snf2 and sequencing of bacterial artificial chromosome clones identified a homolog of Arabidopsis histidine kinase genes (AHK) encoding cytokinin receptor proteins (fig. S1A). In light of physiological studies on phytohormones in nodulation (4, 5), this histidine kinase was a likely candidate gene, and the corresponding gene region of snf2 was sequenced. A single nucleotide transition (C to T), resulting in replacement of a conserved leucine 266 by phenylalanine (L266F), identifies snf2 as an allele of a lotus histidine kinase (Lhk1) gene. Alignment of genomic and cDNA sequences defined a primary structure of Lhk1 consisting of 11 exons (fig. S1B). Steady-state levels of Lhk1 transcripts in different plant organs were determined by quantitative reverse transcription polymerase chain reaction. Lhk1 was expressed at the highest level in roots, nodules, and leaves, but transcripts were present in all organs tested. (Fig. 2A).

Fig. 1.

Phenotypic characterization of the snf2 mutant. (A) Wild-type rhizobia induced root nodule (B) spontaneous snf2 root nodule. Arrowheads, 5-week-old nodules. Transverse section of (C) wild-typeand (D) snf2 root at time 0 and (E) wild-type and (F) snf2 root after 6 days on hormone-free medium. Arrowhead, dividing cells in the pericycle; arrows, xylem cells. (G) and (H) Callus growth from hypocotyls of wild-type and snf2 given different concentrations of auxin and cytokinin. Root segments of wild-type (I) and snf2 (J) incubated 3 weeks on hormone-free media. Scale bars: (C to F), 50 μm.

Fig. 2.

Expression of the Lhk1 gene in organs and the Lhk1, Lrr5, and Nin in response to cytokinin. (A) Expression of Lhk1 in different organs. (B) Expression of Lrr5 in wild-type and snf2 root explants incubated on medium with or without 0.5 μg/ml of 6-benzylaminopurine (BAP) for 10 days. (C to E) Expression of Lrr5, Lhk1, and Nin in intact wild-type and snf2 roots in response to 10 μM cytokinin. (F) Expression of Nin in wild-type and snf2 root explants incubated on medium with or without 0.5 μg/ml of BAP for 10 days.

Constructs carrying either the snf2 mutant gene or the wild-type Lhk1 gene were transformed into wild-type roots using Agrobacterium rhizogenes. To assure reliable transfer to transgenic roots of the gene constructs used throughout this study, they were integrated directly into A. rhizogenes transferred DNA (T-DNA) by using a recombination approach (6, 7). Thus, they were transformed into plant cells together with the T-DNA, which gave rise to transgenic roots at the hypocotyl wound site. An Lhk1 gene segment and the corresponding snf2 gene segment were introduced, and nodulation was scored in the absence of rhizobia. Spontaneous nodulation was observed on transgenic roots transformed with the snf2 construct, whereas the Lhk1 wild-type gene was unable to confer spontaneous nodulation (table S1 and fig. S2). This differential response illustrates the effect of the dominant snf2 mutation and confirms that spontaneous nodulation is caused by a single amino acid substitution in the cytokinin receptor. The absence of nodules on the normal root systems, which served as internal controls for the A. rhizogenes–induced snf2 transgenic roots, and the lack of rhizobia in the nodules that were formed on the snf2 transgenic roots show that they were indeed spontaneously formed nodules.

An open reading frame of 2979 nucleotides is predicted in the Lhk1 cDNA clone. The conceptual cytokinin receptor protein (LHK1) consists of 993 amino acids (Fig. 3). At the N terminus, two membrane-spanning segments are located between amino acids 37 and 57 and between amino acids 328 and 357. Located between these segments are motifs characteristic of cytokinin-binding (CHASE) domains. This predicted extracellular domain is followed by a putative intracellular histidine kinase and a receiver domain. These domains are characteristic of two-component regulatory systems operating through phosphorelay.

Fig. 3.

Structure of the Lotus LHK1 protein. (A) Schematic representation of the LHK1 protein domains. (B) The amino acid sequence of LHK1 arranged in protein domains. The extracellular receptor domain is in italics. The predicted CHASE domain within the extracellular receptor domain is underlined. The asterisk marks the amino acid substitution in the mutant. The histidine kinase domain is bold and underlined. The histidine kinase adenosine triphosphatase (ATPase) domain is bold. The receiver domain is bold and italics.

Comparative analysis defines LHK1 as a member of the cytokinin receptor family (fig. S3). Among the three Arabidopsis cytokinin receptors, LHK1 has 68% identity to AHK4/(Cre1), which is important for normal root development and serves a function in perception of externally supplied cytokinin (8). The leucine 266 replaced by a phenylalanine in the snf2 allele is part of a conserved motif shared among the extracellular CHASE domains of histidine kinase receptors (fig. S3). Spontaneous nodulation resulting from an amino acid change located in the CHASE domain suggested a cytokinin-independent function caused by the L266F substitution. To test this hypothesis, we assayed the in vivo activity of lotus wild-type and gain-of-function receptors using the two-component phosphorelay assay developed in E. coli (9). Functional expression of a cytokinin receptor in an E. coli strain lacking the RcsC sensor, which normally regulates extracellular polysaccharide synthesis, allows cytokinin perception to be read out as β-galactosidase activity from a cps::lacZ fusion. Expression of the mutant L266F protein does indeed induce β-galactosidase activity in the absence of cytokinin (Fig. 4A). In contrast, wild-type LHK1 induced β-galactosidase activity in a cytokinin-dependent fashion (Fig. 4A). Quantitative determination of β-galactosidase activity in E. coli cultures shows that L266F-expressing cells have three times the β-galactosidase activity that control cells and cells expressing wild-type LHK1 have (Fig. 4B). Cytokinin addition results in a twofold induction of β-galactosidase activity in LHK1 cells, whereas L266F cells respond with only a marginal increase in activity. These results demonstrate that LHK1 is a cytokinin receptor and that the L266F receptor exhibits cytokinin-independent activity at a level comparable to the cytokinin-induced activity of the wild-type receptor. We propose that the extracellular CHASE domain, normally binding cytokinin to activate the kinase (1012), in the L266F mutant receptor is locked within an active conformation. This hypothesis would explain both the genetic dominant nature of the snf2 allele and the phosphorelay assay results.

Fig. 4.

In vivo assays of receptor-mediated cytokinin signaling. (A) Plate assay of β-galactosidase activity expressed from a cps::lacZ reporter gene in E. coli. The SRC122 strain carrying the cps::lacZ reporter transformed with either the gain-of-function snf2 or Lhk1 expression construct was grown on plates in the absence or presence of cytokinins. The blue color shows β-galactosidase activity (B) Cytokinin induced β-galactosidase activity in liquid cultures of SRC122 cps::lacZ transformed with either snf2 or wild-type constructs. T-z, trans-zeatin. (C) Model for functional role of Lhk1 in nodulation. Recognition of Nod-factor by NFR1 and NFR5 induces Nod-factor signal transduction, including calcium spiking and CCaMK kinase activity. A localized increase in cytokinin levels perceived by the LHK1 receptor then leads to cortical cell dedifferentiation and cell cycle activation. snf2 acts independently of cytokinin but still requires Nin, Nsp2 genes for nodule organogenesis.

Arabidopsis ahk2 ahk3 ahk4 triple mutants display a reduced number of root vascular cell files, because periclinal procambial cell divisions are impaired (13). snf2 mutant roots have the opposite phenotype with extra layers (Fig. 1, C and D). In explants cultivated without phytohormones (Fig. 1, I and J), cell proliferation was even more pronounced. Additional cell layers originating from periclinal divisions were observed, together with an increase in vascular cell numbers (Fig. 1, E and F). In order to examine possible global effects on cell differentiation, we monitored the in vitro performance of snf2 and wild-type hypocotyl and root explants (Fig. 1 and fig. S4). The overall hormone dose response is similar. However, in line with the cytokinin-independent response, snf2 explants survive better at high auxin and develop less callus on cytokinin (Fig. 1, G and H).

Cytokinin-induced changes in cellular processes in plants are accompanied by increased expression of type-A response regulator (ARR) genes (14). Among the type-A genes, ARR5 is a rapidly induced response gene, and an Arabidopsis ARR5 promoter fused with the Gus (β-glucuronidase) reporter gene (GUS fusion) was expressed during nodulation (15). Because the L266F receptor protein has cytokinin-independent activity in the E. coli assay, we determined transcript levels of a lotus ARR5 homolog named Lrr5 (Fig. 2). Following the examples from previous analyses of the complex cytokinin circuitry in Arabidopsis (16, 17), we determined transcript levels in both intact plants and in vitro cultivated plant cells in order to capture the dynamics and the range of cytokinin regulation. Lrr5 transcript in root explants of snf2 mutants incubated without hormones was found to be two times that of wild-type explants, whereas cytokinin addition increased Lrr5 transcript level two- to threefold in both (Fig. 2B). Cytokinin treatment of roots increased the Lrr5 transcript level in snf2 and wild-type roots, but no difference in expression was detected between untreated snf2 and wild-type roots (Fig. 2C). Cytokinin also regulates expression of the Lhk1 gene and induces a rapid increase in Lhk1 gene transcripts after treatment in both wild-type and snf2 mutants (Fig. 2D). We also tested whether the lotus Nin gene known to be required for initiation of nodule primordia was ectopically expressed in snf2 roots. As shown in Fig. 2E, the Nin gene is up-regulated by cytokinin, and the transcript levels in untreated snf2 roots were significantly different from those of wild-type roots. No ectopic expression of Nin in root explants of snf2 mutants incubated on hormone-free medium was detected (Fig. 2F). Attenuation of the cytokinin response pathway, as previously described in Arabidopsis exposed to cytokinin (16, 17), was also observed in lotus wild-type roots and was even more pronounced in snf2 (Fig. 2, C and E). Thirty minutes after exposure to exogenous cytokinin, a sevenfold increase in the steady-state level of Lrr5 transcript was detected in snf2 roots. In spite of the continuous presence of cytokinin, this initial induction was attenuated at later time points, and only a twofold increase in transcript level was found after 8 hours. In Arabidopsis, attenuation of responses to cytokinin is mediated by a complex feedback mechanism. The cytokinin oxidases, which by themselves are cytokinin inducible, and a range of negatively acting response regulators, including ARR5, which also negatively autoregulate their own transcription, were shown to be involved (16, 17). In the gain-of-function snf2, the cytokinin hypersensitivity (fig. S5) and the presence of the LHK2 and LHK3 receptors (18), which remain cytokinin-dependent, appear to reset the balance point of negative regulation at a level where the transcriptional up-regulation of Lrr5 in untreated snf2 roots is relatively small or undetectable. Although transcriptional changes in the snf2 mutants were limited, plant growth is strongly affected by external cytokinin. In line with the cytokinin-independent activity of the gain-of-function receptor observed in the E. coli assay (Fig. 4A) and in the in vitro culture experiments (Fig. 1, H and J; Fig. 2, B, D, and E), snf2 shoot and root growth was hypersensitive to cytokinin (fig. S5). Prolonged exposure of wild-type plants (8 weeks) to lower cytokinin levels than those used in the experiment shown in fig. S5 did lead to development of small “bumps” that resembled nodule primordia.

The phenotype of snf2 mutants suggests that cytokinin signaling acts downstream of Nod-factor signal transduction. To test this hypothesis, the snf2 gene construct was transformed into mutants of the Nod-factor signal transduction pathway and in mutants impaired in downstream genes. snf2-mediated spontaneous nodulation in nfr1-1, nfr5-2 Nod-factor receptor single and double mutants lacking the earliest electrophysiological responses (1, 2) demonstrates a function for Lhk1 downstream of Nod-factor signal perception. The signal transduction symRK mutants lacking Ca2+ spiking (19) and ccamk (Ca2+- and calmodulin-dependent protein kinase) mutants, which have been suggested to be unable to interpret Ca2+ spiking (20, 21), also develop spontaneous nodules in snf2 transgenic roots. Incidentally, these results provide independent evidence for snf2-mediated spontaneous nodulation. The nfr1, nfr5, symrk, and ccamk mutants are unable to form nodules in response to rhizobia inoculation. Thus, nodule formation on the snf2 transgenic roots could not have resulted from contaminating rhizobia. In nin and nsp2 mutants arrested before initiation of cell division induced by Nod-factor signaling, no spontaneous nodules were observed in snf2 transgenic roots. Because A. rhizogenes–induced roots only develop when the hypocotyl wound site infection is used in lotus and because the snf2 gene construct was integrated into the T-DNA, these results show that cytokinin signal perception acts upstream of cell division initiation (table S2). Furthermore, evidence for a central role of cytokinin and cytokinin perception downstream of Nod-factor signal transduction comes from the additive effect of snf1-1 and snf2 mutations. The snf1-1 mutants synthesize a CCaMK protein impaired in autophosphorylation (20, 21) and develop an average of 7 ± 0.9 (95% confidence interval) spontaneous nodules, whereas snf2 mutants develop 3 ± 0.5. The snf1-1 snf2 double mutants exceed both with 17 ± 0.9 spontaneous nodules. Parallel signaling cannot be excluded, but more likely, the deregulated signaling in snf1 results in a local increase in cytokinin levels transcriptionally up-regulating snf2 (Fig. 2D) and amplifying spontaneous nodulation. The previously reported expression of a Nin-GUS promoter fusion in snf1 nodule primordia and the absence of epidermal expression in snf1 roots (20) further suggest cytokinin signaling is a cortical response. Conversion of cortical cells into nodule stem cells or subsequent organ development seem therefore tightly controlled. We tested this in a hypernodulating har1-1 mutant (22). Homozygous snf2 har1-1 double mutants developed an average of 14 ± 1.4 spontaneous nodules, whereas snf2 mutants developed an average of 3 ± 0.5, and har1-1, none (fig. S6). This indicates that only a few cells dedifferentiate or that only a few dedifferentiated cells sustain cell divisions during the snf2 nodule-initiation process. The shoot controlled autoregulation of the root nodule number (22) is thus acting downstream of cytokinin signaling–induced activation of root nodule founder cells (Fig. 4C).

From Arabidopsis and tobacco, there is evidence for cytokinin regulation of cell cycle phase transitions (23) and for overlapping roles for three AHK receptors in maintaining stem cells and cell divisions during organ formation (13). Phytohormones have also been implicated in nodule organogenesis. Applications of auxin transport inhibitors resulted in empty nodule-like structures, which suggested that local inhibition of auxin transport (24) sensitizes cells for division. Other experiments showed that externally supplied cytokinin induces cortical cell division and activation of Enod12, Enod40, and Enod2 genes (4, 25), and expression of a cytokinin biosynthesis tzs gene in a nodulation-deficient Sinorhizobium meliloti resulted in nodule-like structures (5).

Here we show conclusively that cytokinin signaling plays an important role in plant meristem formation and is directly involved in initiating root nodule organogenesis. The opposite phenotype effects of the snf2 gain-of-function and hit1 loss-of-function mutations reported in the accompanying paper (18), together with the reduced nodulation observed after down-regulation of the corresponding gene in Medicago (26), clearly demonstrate that cytokinin signaling is necessary and sufficient for the dedifferentiation and cell proliferation leading to root nodule formation.

Supporting Online Material

Materials and Methods

Figs. S1 to S7

Tables S1 and S2


References and Notes

View Abstract

Navigate This Article