LONELY GUY (LOG) genes encode cytokinin riboside 5′-monophosphate phosphoribohydrolases and are directly involved in the activation of cytokinins.
To assess whether LOG proteins affect the influence of cytokinin on nodulation, we studied two LOG genes of Medicago truncatula.
Expression analysis showed that MtLOG1 and MtLOG2 were upregulated during nodulation in a CRE1-dependent manner. Expression was mainly localized in the dividing cells of the nodule primordium. In addition, RNA interference revealed that MtLOG1 is involved in nodule development and that the gene plays a negative role in lateral root development. Ectopic expression of MtLOG1 resulted in a change in cytokinin homeostasis, triggered cytokinin-inducible genes and produced roots with enlarged vascular tissues and shortened primary roots. In addition, those 35S:LOG1 roots also displayed fewer nodules than the wild-type. This inhibition in nodule formation was local, independent of the SUPER NUMERIC NODULES gene, but coincided with an upregulation of the MtCLE13 gene, encoding a CLAVATA3/EMBRYO SURROUNDING REGION peptide.
In conclusion, we demonstrate that in M. truncatula LOG proteins might be implicated in nodule primordium development and lateral root formation.
Cytokinins are essential for legume plants to establish a nitrogen-fixing symbiosis with rhizobia. During this symbiotic interaction, nodules are formed in which the bacteria reside to fix atmospheric nitrogen for the plant. Central in nodulation signaling are the bacterially produced Nodulation Factors (NFs) that are perceived by LysM-type receptor kinases at the root hair membrane, whereafter a signaling cascade is activated that initiates rhizobial infection as well as nodule organogenesis (Oldroyd et al., 2011). In the nodulation process, the essential role played by cytokinin (Dehio & de Bruijn, 1992; Cooper & Long, 1994; Bauer et al., 1996; Fang & Hirsch, 1998; Mathesius et al., 2000; Lohar et al., 2004; Heckmann et al., 2011) has been demonstrated most prominently by the defective nodule primordium formation in transgenic Medicago truncatula plants, in which the expression of the cytokinin receptor gene CRE1 is reduced, in cre1 mutants, or in knockout mutants for the CRE1 homolog, LOTUS HISTIDINE KINASE1 (LHK1) of Lotus japonicus (Gonzalez-Rizzo et al., 2006; Murray et al., 2007; Plet et al., 2011). Additionally, an L. japonicus gain-of-function mutant for the LHK1 receptor provoked spontaneous nodules, indicating that cytokinin signaling is both necessary and sufficient for nodule formation (Tirichine et al., 2007). Cytokinin signaling is determined by a phosphorelay pathway controlled by A-type and B-type response regulators (RRs). The A-type and B-type RRs, MtRR4 and MtRR1, respectively, are activated by rhizobial inoculation depending on several components of the NF signaling pathway (Gonzalez-Rizzo et al., 2006; Ariel et al., 2012). Moreover, in M. truncatula and L. japonicus, additional A-type RRs have been identified that are induced by NF treatment in the root-susceptible zone and of which gain-of-function mutants triggered the development of nodule primordia-like structures (Op den Camp et al., 2011).
The cytokinin responses during nodulation have been localized in the cortex, where cell divisions are activated for primordium formation (Lohar et al., 2004; Gonzalez-Rizzo et al., 2006; Plet et al., 2011). Cytokinin is sensed by MtCRE1, and via MtRR4 and MtRR1 signaling, stimulates the transcription factors ETHYLENE-RESPONSIVE BINDING DOMAIN FACTOR REQUIRED FOR NODULATION1 (ERN1), NODULATION SIGNALING PATHWAY2 (NSP2) and NODULE INCEPTION (NIN) that are involved in the transcription of early nodulation (ENOD) genes to induce cortical cell divisions (Tirichine et al., 2007; Frugier et al., 2008; Plet et al., 2011; Ariel et al., 2012). MtCRE1 also regulates nodulation-related auxin accumulation by reducing PIN-FORMED protein activity and auxin transport regulation (Plet et al., 2011).
Besides their role in nodule primordium development, cytokinins might also control the meristem differentiation of indeterminate nodules and mediate the root susceptibility for nodulation (Murray et al., 2007; Frugier et al., 2008; Vernié et al., 2008; Plet et al., 2011). Legume plants regulate the number of nodules through various mechanisms, including autoregulation of nodulation (AON), a systemic system that involves root–shoot communication (Mortier et al., 2012b). Central in AON is a shoot-active receptor-like kinase, designated SUPER NUMERIC NODULES (SUNN) in M. truncatula, similar to various receptors that (potentially) bind CLAVATA3/EMBRYO SURROUNDING REGION (CLE) peptides (Okamoto et al., 2013).
Cytokinin also plays an important role in root architecture with a negative effect on the development of lateral roots. This process has been intensively studied in Arabidopsis thaliana, but has been reported in M. truncatula as well (Gonzalez-Rizzo et al., 2006; Del Bianco et al., 2013).
The homeostasis of the major naturally active cytokinins is maintained by a fine-tuned balance between synthesis, catabolism and storage of inactive forms (Mok & Mok, 2001; Werner et al., 2003; Sakakibara, 2006). Adenosine phosphate-isopentenyltransferases (IPTs) are central in cytokinin synthesis because they catalyze the synthesis of N6-(⊿2-isopentenyl)adenine (iP) and trans-zeatin (tZ), precursors that are converted into active cytokinins (Chen, 1997; Kakimoto, 2001; Takei et al., 2001; Sakamoto et al., 2006; Kurakawa et al., 2007) (Supporting Information Fig. S1). The direct activation pathway is mediated by a cytokinin riboside 5′-monophosphate phosphoribohydrolase, designated LONELY GUY (LOG), that allows immediate conversion to the active cytokinin nucleobases iP and tZ (Kurakawa et al., 2007). IPTs, cytokinin oxidases and LOG proteins are produced at specific locations and time points, especially at active growth regions during plant development (Kuroha et al., 2009; Werner & Schmülling, 2009). Cytokinins also act as long-distance signals to control various plant developmental processes (Kudo et al., 2010) and their transport via the phloem is important for proper root vascular patterning (Bishopp et al., 2011a,b).
The LOG proteins are key players in the release of active cytokinins during plant development (Kurakawa et al., 2007; Kuroha et al., 2009), because the ectopic overexpression of the LOG genes reduced apical dominance, retarded leaf senescence and increased the cell division rate in root vasculature and embryo (Kuroha et al., 2009). Higher-order log mutants had altered root and shoot morphologies and a reduced sensitivity to cytokinins during lateral root formation (Kuroha et al., 2009; Tokunaga et al., 2012). The partially overlapping expression patterns of the LOG genes, as well as the absence of phenotypes in single vs multiples mutants of LOG in Arabidopsis suggested redundant functions for the LOG proteins (Kuroha et al., 2009).
Until now, it has remained not well known whether changes in cytokinin homeostasis due to de novo cytokinin synthesis, cytokinin activation or relocation contribute to nodule development. Because of the importance of LOG genes in cytokinin-dependent developmental processes, we addressed this question by identifying the MtLOG1 and MtLOG2 genes that show enhanced expression during M. truncatula nodulation. The expression of both genes was analyzed in detail and functional analysis suggested an important role during nodulation and lateral root formation.
Materials and Methods
Medicago truncatula Gaertn. cv Jemalong A17, J5, as well as the different mutants were grown and inoculated as described (Mergaert et al., 2003). The Sinorhizobium meliloti 1021 strain, the green fluorescent protein (GFP)-tagged Sm2011 strain with the pHC60 plasmid (Cheng & Walker, 1998), or the monomeric red fluorescent protein (mRFP)-tagged Sm2011 strain with the pBHR plasmid (Smit et al., 2005) were grown as described (Mortier et al., 2010). For RNAi analysis, the S. meliloti strain E65 was maintained on 50 μg ml−1 tetracycline-containing modified Bergensen's medium (Rolfe et al., 1980), the S. meliloti strain WSM1022 was kept on tryptone yeast (TY) medium (Terpolilli et al., 2008) and the Agrobacterium rhizogenes Arqua1 strain (Boisson-Dernier et al., 2001) was preserved on 100 μg ml−1 streptomycin-containing TY medium.
Medicago truncatula J5 plants were grown in vitro in square Petri dishes (12 × 12 cm) on nitrogen-poor SOLi agar (Blondon, 1964). For the temporal expression analysis during nodulation, nodules were harvested at 4–10 d post inoculation (dpi) from plants grown in pouches, watered with nitrogen-poor SOLi medium and inoculated with Sm1021 pHC60-GFP. Infection threads were visible from 4 dpi onward, nodule primordia at 6 dpi, small nodules at 8 dpi and slightly enhanced nodules at 10 dpi. Tissue was collected by visualizing the GFP-tagged bacteria under a stereomicroscope MZFLII (Leica Microsystems, Wetzlar, Germany) equipped with a blue-light source and a Leica GFP Plus filter set (λex = 480/40; λem = 510 nm longpass barrier filter). The zones I of uninoculated roots were isolated at the same developmental stage as the 4-dpi stage. For the MtLOG1 expression analysis in cre1-1 mutants, plants were grown in an aeroponic system. Half of the 6-d-old plants were harvested, whereas the other half were inoculated with Sm2011 pBHR-mRFP and harvested 7 d later. For quantitative reverse-transcription polymerase chain reaction (qRT-PCR) analysis, 1-month-old chimeric plants bearing one of the constructs 35S:GUS, 35S:MtLOG1, 35S:MtCLE12 or 35S:MtCLE13 were placed in an aeroponic system and grown under nitrogen-rich conditions; their roots were harvested 2 wk later for qRT-PCR. RNA extraction, cDNA synthesis and qRT-PCR analysis were done as described previously (Mortier et al., 2010).
Generation of 35S:MtLOG1, pMtLOG1 and pMtLOG2 constructs and A. rhizogenes-mediated root transformation
A 2-kb region upstream of MtLOG1 and MtLOG2 was isolated from genomic DNA based on the available genomic data (http://www.ncbi.nlm.nih.gov/). The promoters were fused to the β-glucuronidase (GUS)-encoding (uidA) gene in pKm43GWRolDC1 (Karimi et al., 2002). For the 35S:MtLOG1 construct, the open reading frame (ORF) of MtLOG1 was isolated from genomic DNA. The ORF was cloned behind the 35S promoter in the pB7WG2D (Karimi et al., 2002). Primers used for amplification are presented in Table S1.
The protocol for A. rhizogenes-mediated root transformation, adapted from Boisson-Dernier et al. (2001), was as described previously (Mortier et al., 2010). To measure the effect of 35S:MtLOG1 on root length and lateral root number, uninoculated roots were aeroponically grown in nitrogen-rich medium until 40 d post germination and analyzed by means of the ImageJ program (http://rsbweb.nih.gov/ij/).
In order to check the long-distance effect of MtLOG1, the main root was kept on the juvenile plant and infected by stabbing the hypocotyls with a fine needle containing an A. rhizogenes culture. After cotransformation as described above, the plants were grown for 2 wk at 25°C with a 16-h photoperiod at 70 μmol m−2 s−1 and transferred to an aeroponic system for 7 d. Nodulation was analyzed on the main, untransformed root of plants bearing GFP-positive hairy roots.
Generation of the MtLOG1 RNAi vector and transformation by A. rhizogenes
A 218-bp fragment of the MtLOG1 gene of M. truncatula was amplified with the primers listed in Table S1. The attB sites were inserted into the pDONOR vector (Invitrogen, Carlsbad, CA, USA) and finally into the RNAi vector pHellsgate8 (Helliwell et al., 2002), which drives the hairpin RNA expression with the 35S promoter. The inserts were verified by restriction digests. The verified constructs were transformed into the A. rhizogenes strain Arqua1 with the freeze–thaw transformation method (Höfgen & Willmitzer, 1988). M. truncatula cv Jemalong A17 seeds were scarified on sand paper, sterilized in 6% (w/v) sodium hypochlorite for 10 min and thoroughly rinsed in sterile water. Seeds were vernalized at 4°C overnight and germinated at 25°C in the dark. M. truncatula was transformed with A. rhizogenes as described (Boisson-Dernier et al., 2001), using the empty pHellsgate8 vector as a negative control. Composite plantlets were grown on 15-cm diameter Petri dishes filled with slanted modified Fåhraeus medium-containing agar (Fåhraeus, 1957) supplemented with 25 mg l−1 kanamycin (Boisson-Dernier et al., 2001) and grown at 20°C for 7 d, before transfer to 25°C. A 16-h photoperiod at 150 μmol m−2 s−1 light intensity was used throughout. Afterward, plants were transferred to plates containing Fåhraeus medium without kanamycin and incubated for one more week before inoculation with a fresh culture of S. meliloti strain E65 (OD600 = 0.2), which constitutively expresses NodD3 (Fisher et al., 1988) or with S. meliloti strain WSM1022. Individual roots (one per plantlet) were marked on the plate and inoculated with 10 μl of bacterial culture. Plates were incubated for 2 wk at 25°C with a 16-h photoperiod at 150 μmol m−2 s−1 light intensity, whereafter each of the marked roots was analyzed for nodule number, root length and lateral root numbers. Three lots of 8–10 roots for each treatment were excised, frozen in liquid nitrogen and used for qRT-PCR. Primers used to test the expression of all six MtLOG homologs are listed in Table S1.
A generalized linear model was fitted to nodule number, lateral root number, lateral root density data, which partitioned the phenotypic variation into fixed genotype and experiment effects and random error effects. Because the data followed a Poisson distribution, a logarithmic base e link function was incorporated. Root length was analyzed by one-way ANOVA with the genotype as fixed term. For the qRT-PCR data involving only a genotype effect, a one-way ANOVA model was fitted to the log2-transformed ΔΔ cycle threshold (Ct) values with genotype as fixed term and a random error term. The three biological repeats were added as a random effect, nested under genotype to account for dependencies between samples originating from the same genotype. For the qRT-PCR experiments involving two factors (genotype and time), a two-way ANOVA model was fitted to the data with genotype and time as fixed factors in the model and the interaction term genotype × time and a random error term. The three biological repeats were added as a random effect nested under the interaction term. To estimate all-pairwise comparisons, post-hoc tests were done. The Tukey method was applied to adjust for multiple comparisons. Differences were declared significant when P < 0.05. For all analyses, except for the qRT-PCR, the GenStat software (http://www.vsni.co.uk/software/genstat/) was utilized. Differential expression of qRT-PCR was analyzed with the mixed procedure as described in the SAS/Enterprise Guide 5.1 (SAS Institute Inc., Cary, NC, USA).
Histochemistry and microcopy
GUS activity in cotransformed roots and nodules was analyzed with 5-bromo-4-chloro-3-indolyl-β-d-glucuronic acid as substrate (Van den Eede et al., 1992). Roots and nodules were vacuum infiltrated for 20 min and subsequently incubated in GUS buffer at 37°C. Incubation lasted 4 h for pMtLOG1:GUS and overnight for pMtLOG2:GUS. After staining, roots and root nodules were fixed, dehydrated, embedded with the Technovit 7100 kit (Heraeus Kulzer, Wehrheim, Germany), according to the manufacturer's instructions, and sectioned with a microtome (Reichert-Jung, Nussloch, Germany). The 3-μm thick sections were mounted on coated slides (Sigma-Aldrich). For tissue-specific staining, sections were submerged in a 0.05% (w/v) ruthenium red solution (Sigma-Aldrich), washed in distilled water and dried. Finally, sections were mounted with Depex (BDH Chemicals, Poole, UK). Photographs were taken with a Diaplan microscope equipped with bright-field and dark-field optics (Leica).
For analysis of the nodule infection of RNAi-transformed roots of MtLOG, root segments inoculated with the GFP-tagged Sm2011 pHC60 (Gage et al., 1996) were embedded in 3% (w/v) agarose, cross-sectioned at 150-μm thickness on a 1000plus vibratome (Vibratome Company, St Louis, MO, USA), and viewed under an epifluorescence microscope DMLB (Leica) with an UV excitation filter (excitation maximum 488 nm, 515 nm longpass filter). Images were taken with a mounted charge-coupled device camera (RT Slider; Diagnostic Instruments, Sterling Heights, MI, USA).
Composite plants carrying 35S:GUS or 35S:MtLOG1 roots were grown under aeroponic conditions in SOLi medium. After 30 d of growth, 10 g of total root material was taken. Samples were divided into five technical replicates (1000 mg). Cytokinins were extracted and separated, essentially as outlined by Novák et al. (2003). Quantification was done by the isotope dilution method (ultra-high performance liquid chromatography–mass spectrometry) according to Novák et al. (2008).
Identification of two nodulation-induced LOG genes in M. truncatula
Based on the sequences of the nine identified members of the Arabidopsis LOG protein family (Kurakawa et al., 2007), homologs were identified in the genome of M. truncatula by tBLASTx analysis against the M. truncatula genomic data (Mt3.5) and their expression patterns were analyzed with the Medicago Gene Expression Atlas (http://mtgea.noble.org/v2/). In total, six LOG homologs were found in the M. truncatula genomic data (Fig. S2; Table S2). As the expression of two of them, designated MtLOG1 (IMGA|Medtr7 g101290.1) and MtLOG2 (IMGA|Medtr1 g064260.1), was upregulated when compared to uninoculated control roots, we concentrated our analyses on these two genes. The similarities between the MtLOG and AtLOG genes were calculated with the supermatcher tool wEMBOSS (Sarachu & Colet, 2005). MtLOG1 and MtLOG2 shared 87.6% similarity. MtLOG1 was most homologous with AtLOG3 (78.7%) and shared between 60.6% and 78.0% similarity with the other AtLOG genes, whereas MtLOG2 was also most homologous with AtLOG3 (78.5%) and shared between 57.3% and 77.5% similarity with the other AtLOG genes. Phylogenetic analysis clearly indicated that MtLOG1 and MtLOG2 are related to each other and to AtLOG3, AtLOG4 and AtLOG6 (Fig. S2).
In order to confirm the expression pattern, the transcript levels were analyzed in nodulated root tissues at 4, 6, 8 and 10 dpi and compared to the expression levels in uninoculated control roots. The elongation zone of uninoculated roots was used as the reference tissue. Whereas the MtLOG1 transcript level steadily increased from 4 dpi on (Fig. 1a), that of MtLOG2 was low at 4 dpi, but increased at 6 dpi, and remained high until 10 dpi (Fig. 1b). The spatial expression of MtLOG1 and MtLOG2 was investigated by promoter:GUS analysis in developing nodules. Transcriptional activation of the uidA gene was visualized by GUS staining. For pMtLOG2:GUS, no GUS staining was observed except very faintly in the center of developing nodules (Fig. 1c). In uninoculated pMtLOG1:GUS roots, no GUS staining was seen in the nodulation-susceptible root zone. At 3 dpi, when incipient nodule primordia were present, pMtLOG1:GUS was visible in the nodule primordium cells (Fig. 1d). Microscopic analysis revealed GUS staining in cells in the center of the nodule primordium and, to a much lesser extent, in dividing cells at the base of the nodule primordium (Fig. 1d,e). At a slightly later stage, GUS staining was detected throughout the cells of the nodule primordium, but not in the outer cortex cells (Fig. 1f). In mature nodules, the expression of pMtLOG1:GUS was visible at the nodule apex, more specifically in the meristem and early differentiating nodule cells (Fig. 1g,h), but did not occur in the outer cortical cells and in cells of the mature infection zone (Fig. 1h). MtLOG1:GUS expression was observed in lateral root primordia (Fig. 1i) and in root tips (data not shown). Together, these data imply a tissue-specific expression of two of the MtLOG genes during nodulation and root development.
MtLOG1 and MtLOG2 expression in cre1-1 mutants
As nodule primordium development depends on the CRE1 function and because the MtLOG1 and MtLOG2 expression had been correlated with nodule primordium development, we investigated the transcript levels in the cytokinin receptor-affected cre1-1 mutant by qRT-PCR, before and after inoculation (Fig. 2). Upon inoculation of the wild-type M. truncatula, the expression of MtLOG1 and MtLOG2 was significantly increased compared to control roots (Fig. 2a–d), whereas this increase was not seen in cre1-1 roots (Fig. 2a–c). These results suggest that the expression of MtLOG1 and MtLOG2 is only activated late in the nodulation signaling cascade, downstream of CRE1. To confirm these data, we generated transgenic roots of pMtLOG1:GUS in a wild-type and cre1-1 mutant background, and analyzed the root-susceptible zones harboring infection threads at 5 dpi. pMtLOG1:GUS expression at the nodule primordium stage was observed only in wild-type plants and not in cre1-1 (Fig. 2c), but still in lateral root primordia (Fig. 2c). These results indicate that during nodulation the expression of MtLOG1 and MtLOG2 depends on CRE1, which is essential for nodule organ formation, although MtLOG1 does not need CRE1 for expression in lateral root primordia.
Influence of ectopic expression of MtLOG1 on cytokinin homeostasis
In order to test whether LOG1 is a true cytokinin riboside 5′-monophosphate phosphoribohydrolase, composite plants with 35S:MtLOG1-expressing roots were made by A. rhizogenes transformation and concentrations of different cytokinin metabolites were measured and compared between 35S:MtLOG1 and 35S:GUS roots. qRT-PCR confirmed the ectopic overexpression of the 35S:MtLOG1 constructs (Fig. 3a). The concentrations of the LOG enzyme substrates, the cytokinin nucleotides N6-(⊿2-isopentenyl)adenosine-5′-monophosphate) (iPMP) and tZ riboside-5′-monophosphate (tZMP), as well as N6-(⊿2-isopentenyl)adenosine) (iPR), tZ riboside (tZR) and tZ were statistically lower in 35S:MtLOG1 roots than those in 35S:GUS roots (Fig. 3b; Table S3), but the concentrations of iP did not significantly differ. In addition, the concentration of cis-zeatin (cZ), which is not produced by a LOG-mediated pathway, and most of its metabolites did not differ significantly (data not shown). The reduced concentrations of iPMP and tZMP suggest metabolization by MtLOG1. Because the measurements did not show unequivocally the accumulation of active cytokinins, we followed the expression of a subset of M. truncatula A-type RRs that are induced by exogenous cytokinin or activated in transgenic roots expressing the Mt35S:CRE1*(L267F) construct (Gonzalez-Rizzo et al., 2006; Op den Camp et al., 2011). The mRNA levels of the cytokinin and nodulation-induced RR4 were significantly higher in 35S:MtLOG1 plants than those in control plants, as was also the case for the expression levels of RR5, RR9 and RR11, but not for RR8 (Fig. 4).
Next, we reasoned that the absence of active cytokinin accumulation in the 35S:MtLOG1 plant could be due to a rapid turnover or conjugation. In fact, the expression of the only studied cytokinin oxidase (CKX1) gene (Ariel et al., 2012) was not enhanced (Fig. S3a) and, additionally, among the cytokinin conjugates, the concentrations of N6-(⊿2-isopentenyl)adenine 7-glucoside (iP7G), N6-(⊿2-isopentenyl)adenine 9-glucoside (iP9G) and tZ 7-glucoside (tZ7G) did not differ significantly between 35S:MtLOG1 and 35S:GUS roots, whereas the concentrations of tZ 9-glucoside (tZ9G) and tZ riboside O-glucoside (tZROG) were statistically reduced in 35S:MtLOG1 (Fig. S3b). Thus, these data revealed that cytokinin homeostasis was modified and that cytokinin signaling was activated in 35S:MtLOG1 plants.
Effect of MtLOG1 RNAi on root and nodule development
In order to unravel the role of MtLOG1, an RNAi hairpin construct was created that targeted 218 bp of the MtLOG1 sequence. Due to the homology between MtLOG1 and the other MtLOG gene candidates, this region also produced identical overlaps with a maximal length of 29, 23, 25, 12 and 17 bp of MtLOG2, Medtr4 g058740.1, Medtr1 g105240.1, Medtr1 g015830.1 and Medtr3 g113710.2, respectively. Composite plants were inoculated with S. meliloti strain WSM1022 and the nodule number was assessed at 14 dpi. The number was reduced from an average of 9.2 nodules per root in empty-vector hairy roots to 4.7 nodules in MtLOG1 RNAi hairy roots. (P < 0.005) (Fig. 5a). A significant reduction in nodule numbers was also seen after infection of MtLOG1 RNAi roots with S. meliloti E65 compared to controls (data not shown). qRT-PCR analysis showed that the RNAi vector significantly decreased the MtLOG1 expression to c. 44% of the empty-vector controls, but that it did not alter significantly the expression of any of the other five MtLOG homologs (Fig. 5b). After microscopic analysis of the vibratome sections, no visible difference in nodule structure and infection with GFP-producing rhizobia was observed between MtLOG1-silenced roots and empty-vector controls (Fig. 5c,d).
When the root architecture between control and MtLOG1 RNAi plants was compared, on average, the root length was 72.00 ± 9.05 mm (Fig. 6a), the number of lateral roots 3.65 ± 1.02 (Fig. 6b), giving a lateral root density of 0.50 ± 0.074 lateral roots cm−1 in control plants (Fig. 6c) vs 56.82 ± 4.79 mm root length, 4.96 ± 0.60 lateral roots and 1.0 ± 0.1 lateral roots cm−1 in the MtLOG1 RNAi roots (Fig. 6a–c). Thus, the lateral root density of MtLOG1 RNAi roots was two-fold higher than that of the control roots (P = 0.02), in agreement with a negative role for cytokinins in lateral root development.
Effect of ectopic expression of MtLOG1 on root development
We examined the potential role of LOG proteins in root and nodule development by analyzing 35S:MtLOG1 roots in detail for root structure, lateral root development and nodulation. The lengths of 35S:MtLOG1-expressing roots of composite plants made by A. rhizogenes transformation were measured 38 d post transformation. On average, the length of control roots was 40.52 ± 3.55 cm vs 22.88 ± 3.55 cm for 35S:MtLOG1 roots (Fig. 7a). In accordance, the lateral root number for control 35S:GUS and 35S:MtLOG1 transgenic roots was on average 40.07 ± 3.31 and 18.23 ± 2.45 (Fig. 7b), resulting in a lateral root density of 1.36 ± 0.13 and 1.38 ± 0.14 lateral roots cm−1, respectively (Fig. 7c). Except for the lateral root density, all differences in measurements were statistically significant (P < 0.001).
The 35S:MtLOG1-expressing roots seemed to be more robust than the 35S:GUS roots. To analyze this observation in more detail, we sectioned transversally the root tissue located 1–2 cm above the root tip of 1-month-old plants. The vascular tissue of MtLOG1 transgenic roots was more expanded than that of control roots (Fig. 7d–f). The diameter of the vascular bundle, comprising xylem tissue, phloem tissue, pericycle and endodermal cell layer, was measured. On average, the diameter of 35S:MtLOG1 roots was 276.2 ± 6.0 μm vs only 182.0 ± 5.0 μm for control roots (P < 0.001, Student's t-test) (Fig. 7f). Careful analysis of the sections indicated enlarged, but also numerous, cells in the vascular tissues of 35S:MtLOG1 (Fig. 7e,f). The same measurements done on the cortical tissue of these roots revealed changes neither in size nor in cell number in radial sections of 35S:MtLOG1 roots when compared to control roots. The data suggest that the thickening of transformed 35S:MtLOG1 results from vascular tissue expansion.
Effect of 35S:MtLOG1 on nodule development
Analysis of nodulation of the 35S:MtLOG1 roots revealed that ectopic expression of MtLOG1 resulted in only a few nodules compared to those counted on control roots (Fig. 8a). The structure of these limited nodules was examined by microscopy on 3-μm–thick Technovit-embedded sections of 14-d-old nodules. Infection threads were clearly distinguishable on sections of 35S:GUS and 35S:MtLOG1 transgenic nodules (Fig. 8b–e), indicative for normal rhizobial infection. However, the meristematic tissue of 35S:MtLOG1-transformed nodules seemed to disappear and the fixation zone was disorganized (Fig. 8b–e).
Because of the reduced number of nodules in 35S:MtLOG1 roots, we wondered whether the observed phenotype might interact with AON, a long-distance, nodule number-controlling mechanism. Previously, MtCLE13, a CLE peptide gene that putatively activated the AON pathway, had been shown to be induced by cytokinin (Mortier et al., 2010). Therefore, the expression of MtCLE13 was analyzed by qRT-PCR on cDNA samples of uninoculated root tissues transformed by A. rhizogenes with 35S:MtLOG1. In addition, MtCLE12 was tested, which does not show increased expression in response to cytokinin (Mortier et al., 2010). The MtCLE13 expression was indeed higher in 35S:MtLOG1 roots than that in control roots, but not significant for MtCLE12 (Fig. 9a,b). To investigate whether MtCLE13 overexpression had an effect on MtLOG1 expression, we analyzed the cDNA of roots ectopically expressing MtCLE13 by qRT-PCR and found no consistent differential expression of MtLOG1 and MtLOG2 between control and 35S:MtCLE12 or 35S:MtCLE13 roots (Fig. S4). Together, these data might imply that MtLOG1 expression contributes to the MtCLE13 expression during nodulation.
Previously, nodulation had been found to be totally inhibited on 35S:MtCLE13 roots and that this effect involved long-distance mechanisms (Mortier et al., 2010). To test whether the ectopic expression of MtLOG1 might also result in long-distance effects, we generated composite plants with small 35S:GUS or 35S:MtLOG1 transgenic roots in addition to the wild-type main root. At 7 dpi with the Sm2011 carrying the pBHR-mRFP construct, the nodule number was assessed on the primary wild-type root. On average, 17.36 ± 1.72 nodules were counted on the primary root of the wild-type 35S:GUS plants vs 14.62 ± 1.59 nodules on that of 35S:MtLOG1 plants (Fig. 9c). Regression analysis (Poisson distribution, logarithmic link) by means of the GenStat software revealed that these differences were not statistically significant (P > 0.05). Thus, because ectopic MtLOG1 expression could only reduce the nodule number in the roots in which the construct was expressed and not in nontransgenic roots of composite plants that additionally carried transgenic 35S:LOG1 roots, the effect of the ectopic overexpression of MtLOG1 on nodulation had to be local and not systemic. This observation is in contradiction with what had been seen for MtCLE13, where the ectopic expression affected nodulation not only on the transgenic roots, but also on the nontransgenic roots of the same plant (Mortier et al., 2010).
This negative effect of 35S:MtCLE13 on nodulation is abolished in sunn-4 mutants that are affected in the AON pathway due to a mutation in a leucine-rich repeat receptor-like protein kinase, possibly perceiving CLE peptides (Mortier et al., 2012a). To study the link between MtCLE13 and MtLOG1, we tested the effect of ectopic overexpression of MtLOG1 on nodulation of a sunn-4 mutant. Nodulation was assessed at 7 dpi with Sm2011 pBHR-mRFP and resulted in 41.50 ± 4.41 nodules on 35S:GUS roots (control) and in 8.67 ± 1.75 nodules on 35S:MtLOG1 roots (Fig. 9d). These differences were statistically significant (P < 0.001) in the regression analysis (Poisson distribution, logarithmic link) by means of the GenStat software. 35S:MtLOG1 expression caused a similar decrease in nodule numbers in sunn-4 mutant plants compared to the wild-type (compare Fig. 9d with Fig. 8a). These results indicate that the MtCLE13 peptide signaling might be involved in the nodule number reduction caused by the 35S:MtLOG1 expression, but that the inhibition is local and independent of SUNN.
Many experiments have shown that cytokinin signaling is essential for the development of the nodular organ. Here, we identified two LOG genes in Medicago truncatula that showed increased expression upon nodulation. The two MtLOG genes share a high degree of similarity with AtLOG1, AtLOG2, AtLOG3, AtLOG4, AtLOG5, AtLOG7 and AtLOG8 that are involved in the release of the active cytokinin nucleobases from the cytokinin riboside 5′-monophosphates, as reported for the LOG gene of rice (Oryza sativa) (Kurakawa et al., 2007; Kuroha et al., 2009). Measurements of cytokinin metabolites have shown that 35S:MtLOG1 roots contain fewer cytokinin nucleotides than 35:GUS roots, indicating that MtLOG1 might be a true 5′-monophosphate phosphoribohydrolase, but without increase in free iP and tZ. This lack of increase might be due to activation of negative feedback mechanisms or to a rapid turnover for a tight control of the cytokinin levels (Sun et al., 2003; Miyawaki et al., 2006); however, no elevated expression of a gene encoding a cytokinin-degrading oxidase was measured, nor were elevated concentrations of metabolites detected, indicative for the activation of conjugating mechanisms. By contrast, a decrease in some glycosylated cytokinin forms in the 35S:MtLOG1 plants was observed, implying a compensatory action of other parts of the cytokinin pool. Whatever the reason for these results, the cytokinin signaling increased in 35S:MtLOG1 roots because several cytokinin-responsive genes were activated.
Expression of MtLOG1 and MtLOG2 had been shown to depend on the cytokinin receptor CRE1. Accordingly, the earliest MtLOG1 and MtLOG2 expression occurred in cells of developing nodule primordia. Hence, we propose that the cytokinins released via MtLOG1 and MtLOG2 might probably not act as a primary cytokinin pool that is sensed by MtCRE1 to reinitiate cell division in the cortex for nodule development, but rather are part of a positive feedback to trigger cell division. Indeed, silencing of MtLOG1 expression reduced the nodule number, supporting the positive effect of cytokinins on nodulation as suggested previously (Lohar et al., 2004; Gonzalez-Rizzo et al., 2006; Murray et al., 2007; Tirichine et al., 2007; Plet et al., 2011), although the nodules that were formed did not differ structurally from those observed on control nodules. Thus, the cytokinins possibly produced by the MtLOG proteins might account, at least partly, for the cytokinin pool needed to maintain nodule formation (Fig. 10).
The MtLOG1 gene also contributes to the cytokinin effect on root development, similarly to the AtLOG genes that negatively control lateral root emergence (Kuroha et al., 2009). Indeed, pMtLOG1:GUS activity was detected in lateral root primordia and the lateral root density was increased on MtLOG1 RNAi plants. Moreover, on transgenic 35S:MtLOG1 roots, root length was reduced, indicative of a possibly enhanced production of root growth-inhibiting cytokinins. In Arabidopsis, the length of the primary root in plants ectopically expressing an AtLOG gene was similar to that of control plants (Kuroha et al., 2009). Possibly, morphological or molecular differences between Arabidopsis and M. truncatula, but also between the activities of the LOG proteins, might explain this discrepancy. Alternatively, the modified auxin sensitivity of transgenic roots generated by A. rhizogenes could influence the cytokinin response of the root. However, as observed for the ectopic expression of LOG genes in Arabidopsis (Kuroha et al., 2009), the root vascular tissue had expanded.
Surprisingly, despite the positive regulation of nodulation by cytokinins, the number of nodules of 35S:MtLOG1 roots was much lower than that of controls and the nodule meristem disappeared in the few produced nodules. These observations are in contrast with a positive cytokinin role in nodule development and might result from a secondary effect caused by changes occurring in the root. Indeed, it is well known that hormones act in a strict spatiotemporal pattern (Vanstraelen & Benková, 2012) and the general ectopic expression of MtLOG1 might influence the hormonal landscape in the root so as to inhibit nodule development. Other studies reported similar unexpected effects, such as the RNAi of RR9, an A-type RR that negatively influences cytokinin signaling, reduced nodulation, in spite of the expected opposite result (Op den Camp et al., 2011). Ectopic expression of RR9 formed primordia that could either be arrested lateral roots or de novo nodule primordia (Op den Camp et al., 2011). Interestingly, the expression of RR9, together with other A-type RRs, was enhanced in 35S:MtLOG1 roots, but no ectopic primordia were observed in our study.
The MtCLE13 peptides might act as a downstream factor of MtLOG1, because the expression of the corresponding gene was higher in transgenic 35S:MtLOG1 roots than that in control roots. MtCLE13 negatively regulates nodulation and interacts with the AON pathway, which systemically controls the nodule number and is activated at the onset of primordium formation (Li et al., 2009; Okamoto et al., 2009; Mortier et al., 2010; Reid et al., 2011). Although this hypothesis is tempting, the nodulation phenotype of roots overexpressing MtLOG1 was milder than that of the overexpressed CLE genes, because, instead of a total inhibition, the nodule number was only reduced and, in addition, the effect of 35S:MtLOG1 was local and independent of SUNN. Thus, the inhibitory effect seen on the nodulation of MtLOG1 plants might be (in part) due to the upregulation of MtCLE13. As many aspects are still to be resolved, it would be interesting to control the localized expression of MtCLE13 in the 35S:MtLOG1 plants. Moreover, the analysis of the MtCLE13 RNAi in 35S:MtLOG1 roots might reveal whether the CLE peptides play an active role in reducing the nodulation induced by 35S:MtLOG1.
In summary, the MtLOG genes might substantiate the proper nodule primordium development and they might negatively regulate lateral root formation. Further studies will reveal the individual contributions of both MtLOG proteins studied here, but stable mutants will be necessary. Moreover, via the MtCLE13 activation, MtLOG1 might link to the AON pathway mechanism that is well known to be activated when the first cell divisions appear in the cortex (Li et al., 2009). Nevertheless, additional experiments are required to clearly demonstrate the potential link between cytokinin production and the AON mechanism.
The authors thank Rene Geurts (Wageningen University, Wageningen, the Netherlands), Pascal Gamas (Institut de la Recherche Agronomique, Toulouse, France), Doug Cook (University of California, Davis, CA, USA), Giles Oldroyd (John Innes Institute, Norwich, UK), Florian Frugier (Institut des Sciences du Végétal, Centre National de la Recherche Scientifique, Gif-sur-Yvette, France), Sharon Long (Stanford University, Stanford, CA, USA) and Julia Frugoli (Clemson University, Clemson, SC, USA) for S. meliloti strains and M. truncatula mutants, their colleagues Christa Verplancke and GuangLing Cui for skillful assistance, Wilson Ardiles for sequence analysis, Marnik Vuylsteke and Véronique Storme for help with the statistical analysis and Stephane Rombauts for help with the bioinformatics. This work was supported by grants from the European Commission Marie Curie International Research Staff Exchange Scheme (IRSES) (grant no. PIRSES-GA-2008-230830), the Ministerie van de Vlaamse Gemeenschap (grant no. CLO/IWT/020714), and the Research Foundation-Flanders (grant nos. G.0350.04N and G.0066.07N) (to M.H. and S.G.), the Centre of the Region Haná for the Biotechnological and Agricultural Research, Faculty of Science (ED 0007/01/01) (to P.T.), and a Future Fellowship from the Australian Research Council (FT100100669) (to U.M.).