Ectopic expression of miR156 represses nodulation and causes morphological and developmental changes in Lotus japonicus

The effects of microRNA156 overexpression on general plant architecture, branching, flowering time and nodulation were investigated in the model legume, Lotus japonicus. We cloned an miR156 homolog, LjmiR156a, from L. japonicus, and investigated its SQUAMOSA PROMOTER BINDING PROTEIN LIKE (SPL) genes and its biological function at enhancing vegetative biomass yield, extending flowering time, and its impact on nodulation. Thirteen potential targets for LjmiR156 were identified in vitro and their expression profiles were determined in aerial and underground parts of mature plants, including genes coding for eight SPLs, one WD-40, one RNA-directed DNA polymerase, two transport proteins, and one histidine-phosphotransfer protein. Two SPL and one WD-40 cleavage targets for LjmiR156—TC70253, AU089191, and TC57859—were identified. Transgenic plants with ectopic expression of LjmiR156a showed enhanced branching, dramatically delayed flowering, underdeveloped roots, and reduced nodulation. We also examined the transcript levels of key genes involved in nodule organogenesis and infection thread formation to determine the role of miR156 in regulating symbiosis. Overexpression of LjmiR156a led to repression of several nodulation genes during the early stages of root development such as three ENOD genes, SymPK, POLLUX, CYCLOPS, Cerberus, and Nsp1, and the stimulation of NFR1. Our results show that miR156 regulates vegetative biomass yield, flowering time and nodulation by silencing downstream target SPLs and other genes, suggesting that the miR156 regulatory network could be modified in forage legumes (such as alfalfa and trefoils) and in leafy vegetables (like lettuce and spinach) to positively impact economically valuable crop species.

1 3 polymerase II, and they can subsequently form an imperfectly matched hairpin structure. In plants, the precursor is further processed by Dicer-like proteins into mature miR-NAs which are then incorporated into an RNA-induced silencing complex where they negatively regulate target gene expression at the post-transcriptional level by basepairing to complementary targets (Dugas and Bartel 2004;Kidner and Martienssen 2005). In some cases, miRNAs can also silence genes at the transcriptional level by affecting chromatin methylation (Brodersen and Voinnet 2009). Temporal and spatial accumulations of a few highly conserved miRNAs are crucial for maintaining proper plant development. For example, miR165 and miR166 are involved in the determination of leaf patterns (Liu et al. 2009), miR156 and miR157 govern the transition from vegetative to reproductive phase (Wu et al. 2009), and miR172 participates in controlling floral development (Wollmann et al. 2010). The most conserved miRNAs tend to be the most highly abundant in organisms.
MiR156, one of the most conserved and ubiquitous miRNAs in plants, is found in mosses, monocotyledons, and dicotyledons (Arazi et al. 2005;Xie et al. 2006), and has been studied mostly for its role in growth regulation. Although miR156 and miR172 function coordinately in plant development, they actually function in opposite ways, with the expression of miR156 decreasing over time while that of miR172 increasing (Wu et al. 2009). In Arabidopsis thaliana, Zea mays, Oryza sativa, and Brassica napus, overexpression of miR156 results in dramatic morphological alterations, e.g., dwarf and bushy plants, delayed flowering, varied distribution of trichomes, and increased carotenoid and flavonoid contents (Xie et al. 2012;Hultquist and Dorweiler 2008;Wang et al. 2009;Wei et al. 2010).
In Arabidopsis, 11 out of the 17 members of the SQUA-MOSA PROMOTER BINDING PROTEIN LIKE (SPL) family of genes are targeted by miR156 (Rhoades et al. 2002;Xing et al. 2010;Gou et al. 2011). MiR156 and its target SPLs define an essential regulatory module that controls phase transitions, leaf trichome development, male fertility, embryonic patterning, and anthocyanin biosynthesis Yu et al. 2010;Xing et al. 2010;Nodine and Bartel 2010). The SPL family encodes plant-specific transcription factors containing at least one SQUAMOSA-PROMOTER BINDING PROTEIN (SBP) domain. In turn, SPLs control the transcription of other genes by binding to promoters that harbor SBP domain-binding sites (SDBs) (Klein et al. 1996). SPL proteins play critical roles in maintaining normal growth throughout a plant's life cycle. For example, SPL-13 regulates the transition from cotyledons to the appearance of true leaves (Ruth et al. 2010) and SPL-3, SPL-4, and SPL-5 function in determining flowering time through Flowering Locus D (FD)-dependent and -independent pathways (Schmid et al. 2003;). Moreover, SPL-3 participates in the regulation of FLOWERING LOCUS T (FT) under various ambient temperatures in Arabidopsis Hwan Lee et al. 2012). Together with SPL-8, a gene not targeted by miR156, the miR156-targeted SPLs (including SPL-2, -9, and -15) are involved in sporogenous cell formation and cell proliferation, all of which influence plant fertility (Xing et al. 2010(Xing et al. , 2013. In addition to its role in plant growth and phase transition, the miR156/SPL network coordinates plant development with plant secondary metabolism. For example, overexpression of miR156 enhances the levels of carotenoids in seeds of A. thaliana and B. napus (Wei et al. 2010(Wei et al. , 2012) and SPL-9 was shown to reduce the production of anthocyanin in A. thaliana stems by suppressing the expression of anthocyanin biosynthesis genes (Gou et al. 2011).
WD40-like proteins are also involved in many aspects of plant growth, functioning as a platform for protein-protein and protein-DNA interactions, microtubule organization during mitosis (Zeng et al. 2009), andchromatin conformation (Reyes et al. 2002;Verbsky and Richards 2001). Interestingly, Naya et al. (2010) revealed that a WD40 transcript is cleaved by miR156 in the root apices of Medicago truncatula. Shi et al. (2005) demonstrated that the Arabidopsis WD-40 protein, SWA1, participates in maintaining root elongation, indicating a potential role for an miR156/WD-40 network in root development.
Legumes are the third largest plant family and are an important source of forage and food. The ability of leguminous plants to interact symbiotically with rhizobia and mycorrhiza makes them useful in maintaining pastures and soil fertility through nitrogen fixation (Graham and Vance 2003). Lotus, a genus within Fabaceae, has more than 100 species and some, such as bird's foot trefoil (Lotus corniculatus), have been used as economically important forage crops (Escaray et al. 2012). Lotus japonicus is a model plant in legume research due to its relatively small genome, minimal requirements for growing space, and short life cycle (Handberg and Stougaard 1992;Udvardi et al. 2005). Information derived from miRbase (http://www.mirbase. org/, version 20; released in June 2013) revealed 10 and 28 members of miR156 variants in M. truncatula and Glycine max, respectively, but none in L. japonicus.
Aside from their roles in plant development, miRNAs also function in nutrient homeostasis and plant-microbe symbiosis. For example, in L. japonicus, miR171 facilitates rhizobial infection while miR397 is activated in nitrogenfixing nodules (De Luis et al. 2012). Overexpression of M. truncatula miR166 reduces the numbers of both nodules and lateral roots (Boualem et al. 2008). MiR169-regulated MtHAP2-1 was shown to play a crucial role in determining the nodule meristematic zone during nodulation in M. truncatula (Combier et al. 2006). In G. max, miR156 and miR166 have five overlapping predicted gene targets (Zeng et al. 2010), which indicates a potential interaction between the two gene systems in root development and nodulation. Both pre-miR156e and pre-miR156g are slightly up-regulated in the roots of Arabidopsis when the supply of nitrogen is limited (Pant et al. 2009). Expression of miR172 is enhanced in soybean roots at 1 and 3 h post-inoculation (hpi) with Bradyrhizobium japonicum but returns to basal levels by 12 hpi (Subramanian et al. 2008). In soybean, ectopic expression of miR160 resulted in a decrease in nodulation (Turner et al. 2013). On the other hand, elevated expression of miR482, miR1512, and miR1515 caused increased nodulation in soybean (Li et al. 2010). Opposite roles have been described for miR156 and miR172 in controlling the expression of both symbiotic and non-symbiotic hemoglobins to modulate the extent of nodulation in soybean, with enhanced levels of miR156 being consistent with reduced nodule numbers while miR172 acting as a positive regulator of nodule formation (Yan et al. 2013).
The ability of miR156 to increase shoot branching, delay flowering, and alter secondary metabolism in several plant species (Fu et al. 2012;Gou et al. 2011;Wei et al. 2010Wei et al. , 2012Wu and Poethig 2006) prompted us to investigate its function in the model forage legume L. japonicus. Therefore, we characterized an L. japonicus homologue, LjmiR156a, focusing on shoot branching, flowering time, and nodulation. Our objective was to investigate the value of the miR156 gene regulatory system and its potential to improve forage and bioenergy crops where high vegetative biomass production is desirable.

Plant material and growing conditions
Lotus japonicus 'Gifu' (accession: B-129) was used as the wild-type (WT) germplasm in all experiments. Seeds were germinated in a Petri dish containing eight layers of filter paper soaked in water. After 1 week, the seedlings were transferred into a mixture of vermiculite, sand, and commercial soil (25:25:50) (Premier Tech Horticulture, Rivière-du-Loup, Quebec, Canada). Plants were grown in a chamber at 23 °C, under a 16-h photoperiod and a light intensity of 250 µE s −1 m −2 . Plants were supplemented biweekly with a Hoagland nutrient solution.
Cloning and overexpression of miR156 in L. japonicus Pre-miRNA sequences of AtmiR156a (accession: AT2G25095) and AtmiR156b (Accession: AT4G30972) from Arabidopsis were used as templates to search for potential pre-miR156 sequences in the L. japonicus genome database (http://www.kazusa.or.jp/lotus/). All positive hits were further examined in silico. First, the candidate sequences were tested with a Basic Local Alignment Search Tool (BLAST) through the National Center for Biotechnology Information (NCBI) Lotus japonicus EST database (http://blast.ncbi.nlm.nih.gov/Blast.cgi) to find those that could be transcripts. These sequences were then assessed for their capacity to form typical miRNA hairpin secondary structures. Sequences that met these two criteria were selected for further study. This analysis identified a 506-bp cDNA sequence, designated pre-LjmiR156a, which was amplified by PCR using LjmiR156a forward and reverse primers (Supplementary Table S1). The cDNA fragment was cloned into the pJET vector (Thermo Fisher Scientific, Waltham, MA) for sequencing, and then sub-cloned into the pBI121 vector (Jefferson 1987) (replacing the GUS gene) between the XbaI and SacI sites. This construct was transformed into Agrobacterium tumefaciens strain LBA4404 (Life Technologies, Burlington, Ontario, Canada). Hypocotyls of L. japonicus were then used for Agrobacterium-mediated transformation according to Márquez and Stougaard (2005). Transformants were selected on media containing G-418 (5 µg/ml), and the presence of the transgene in putative transformants was validated by PCR amplification of a partial 35S promoter sequence linked to the pre-LjmiR156a fragment using 35S promoter forward and pre-LjmiR156a reverse primers.

Small-RNA northern blot analysis
Small RNA was isolated from shoots of mature T 2 plants using the mirVana miRNA Isolation Kit (Life Technologies) following the manufacturer's instructions. RNA probes were synthesized with the mirVana miRNA Probe Construction Kit (Life Technologies) using the primers listed in Supplementary Table 1. DynaMarker for small RNA (BioDynamics Laboratory Inc., Hackensack, NJ, USA) was used to determine the target RNA size. Small RNAs were separated on a denaturing 15 % polyacrylamide gel containing 15 % urea in 0.5× Trix/borate/EDTA buffer. After separation on the gel, the nucleotides were transferred onto a nylon membrane and blotted with an RNA probe overnight at 45 °C. The blotted membrane was treated with CDP-Star Chemiluminescent substrate (Sigma-Aldrich, St. Louis, Missouri, USA) and then exposed to X-ray film. The developed films were then scanned by an EPSON V370 scanner and saved as 1,200 dpi Tagged Image Files. The band intensities were then determined by using ImageJ (http://rsbweb.nih.gov/ij/) to measure the mean grey value. The miR156 probe blotted membrane was then stripped by heating the blot in 1 % SDS solution at 85 °C 30 min. The membrane was then blotted with U6 probe which serves as a loading control.
In silico prediction of Lj-miR156 cleavage targets The target transcripts for miR156 cleavage were predicted using the psRNATarget web server by following the default parameter (http://plantgrn.noble.org/psRNA-Target/). In addition, the mature sequence of L. japonicus miR156 was used to conduct a manual BLAST search against the L. japonicus genome database and the NCBI EST database. This eliminated any false candidates that could form potential precursors for LjmiR156. After the redundant sequences were manually excluded, potential targets were chosen as candidates for cleavage-site validation using a modified 5′-RACE method (Song et al. 2010). This technique was performed with a FirstChoice RLM-RACE Kit (Life Technologies) according to the manufacturer's instructions, but with a slight modification. Instead of removing 5′PO 4 , the adaptor was ligated directly to the RNA molecules, which were then subjected to reversetranscription. Afterward, nested PCRs were run with outer/inner adaptor-and outer/inner gene-specific primers (Supplementary Table 1). The products were gel-purified and cloned into the pJET vector (Thermo Scientific, Waltham, MA, USA). Eleven (11) clones for TC70253, 12 clones for AU089181, and 16 clones for TC57859 were sequenced to determine the cleavage sites of the two candidate genes.
Quantitative real-time RT-PCR Specific tissues of L. japonicus T 2 plants and control plants (as mentioned in each experiment) were collected, immediately frozen in liquid nitrogen, and stored at −80 °C for further application. Total RNA was extracted with an RNeasy Plant Mini Kit (QIAGEN, Toronto, Ontario, Canada). The integrity of those samples was examined by running the RNA on a 1 % agarose gel and observing the 18S and 28S bands. This RNA was treated with Turbo DNase I (Ambion, Life Technologies) to eliminate genomic DNA contamination, and was then quantified using a NanoVue (GE Healthcare, Mississauga, Ontario, Canada). Reversetranscription reactions (1 μg of total RNA per reaction) were performed with a qScript cDNA synthesis Kit (Quanta Biosciences, Gaithersburg, MD, USA) according to the manufacturer's instructions. QRT-PCR was carried out in a 96-well plate on a C1000 Thermal Cycler and CFX96 Real-Time System (Bio-Rad, Mississauga, Ontario, Canada), with PerfeCTa SYBR Green FastMix (Quanta Biosciences). All primers for qRT-PCR are listed in Supplementary Table 1

Morphological analysis
Seeds from the WT and T 2 transgenic plants that overexpress LjmiR156a (miR156+) were germinated on filter paper and transferred to soil as described above. Beginning 2 months post-germination, heights and branch numbers were recorded every 2 weeks in the two fertile transgenic seed lines #20 and #22 that were recovered from the transformation of L. japonicus. Flowering times were recorded as days post-germination (dpg). Leaves and siliques were photographed under a dissecting microscope, and then measured with ImageJ.

Nodulation analysis
Surface-sterilized seeds from the two T 2 seed lines were germinated on eight layers of filter paper soaked in water. At 7 dpg, the seedlings were transferred to vermiculite-filled pots and watered with B&D solution (Broughton and Dilworth 1971) that was supplemented with nitrogen (by adding KNO 3 to a final concentration of 1 mM). Colonies of Mesorhizobium loti harboring the hemA::lacZ reporter gene (strain: NZP2235) were cultured at 28 °C for 72 h until OD 600 = 0.02. The bacteria were washed twice with distilled water followed by centrifugation. They were then re-suspended in water as 1/10 volume of bacterial culture and used for inoculations in which 300 μL were applied to each plant. Afterward, the plants were transferred to a growth chamber and cultured as described above. To evaluate infection threads (ITs), nodule primordia, and nodules, seedlings were removed from the vermiculite at 7, 14, and 21 days post-inoculation. Seedlings were then rinsed with distilled water to remove residual soil. Root-fixation, LacZ-staining, and clearing procedures were performed as previously described (Karas et al. 2005). For sample sectioning, fixed samples were embedded into 4 % agarose and sectioned with a Leica microtome VT1000S. The samples were cut into 80-µm sections at a speed of 4.5 and frequency of 70 Hz. The sections were then observed under a Leica stereo dissecting microscope.

Statistical analysis
Statistically significant differences among treatments were determined with analysis of variance (ANOVA) followed by post hoc Duncan's test at P value ≤0.05.

LjmiR156 is highly conserved in plants
To clone miR156 of L. japonicus, we used Arabidopsis pre-miR156a and pre-miR156b sequences as templates to search the L. japonicus genome and EST database. One sequence (GenBank accession: GO023849) was identified and selected for further in silico analysis. This candidate sequence was capable of forming a typical miRNA secondary structure. Because it had the highest homology to AtmiR156a, we designated it as pre-LjmiR156a (Supplementary Fig. S1). Its predicted mature sequence was identical to miR156 from various plants, including Arabidopsis thaliana and Glycine max. Furthermore, alignments of pre-LjmiR156a to pre-miR156 sequences from Arabidopsis and G. max showed strong similarity not only between the mature miR156 sequences, but also between the pre-mi156 sequences ( Supplementary Fig. S2).

Generating LjmiR156a overexpression plants
To investigate the function of LjmiR156a, we made a construct for pre-LjmiR156a (506 bp) for expression under the CaMV35S promoter and used it to transform L. japonicus hypocotyl explants (Fig. 1a). After 2 months on B5 medium supplemented with G418 antibiotic, 32 calli were derived from the approximately 150 explants, and of these 19 survived. A total of 17 out of the 19 calli produced shoots after shoot induction and 10 of them regenerated healthy root systems; the presence of the transgene in these plants was confirmed by PCR ( Supplementary Fig. S3), and thus they were selected for analysis. Only six transgenic plants survived after transfer to soil, and these plants manifested a dwarfed phenotype, enhanced branching, and flowered late (Fig. 1b). Southern blot analysis indicated that all of these were independent transgenic events ( Supplementary Fig.  S4). Of the six plants, four were unable to produce enough seeds due to severely delayed flowering, and thus could not be pursued in subsequent generations. The remaining two overexpression plants (lines 20 and 22) maintained a moderate ability to produce seeds and exhibited phenotypes relatively close to those resulting from miR156 overexpression in A. thaliana (Wei et al. 2012), Panicum virgatum (Fu et al. 2012), and Solanum lycopersicum (Zhang et al. 2011).
Enhanced LjmiR156a expression affects multiple aspects of L. japonicus growth Six months after their transfer to soil, T 0 transgenic lines 20 and 22 showed greater branching and fewer flowers compared with the 'Gifu' WT plants (Fig. 2a). Although the leaf shape was similar between the transgenic plants (miR156+) and the WT (Fig. 2b), the surface area was significantly diminished in the former. MiR156+ plants also had shorter siliques (Table 1; Fig. 2b), which resulted in substantially fewer seeds per silique. These T 0 plants were then used in generating T 2 progeny seed lines for further experimentation to reduce the influence of segregation. Both qRT-PCR on pre-LjmiR156a and Small-RNA Northern blot analysis on T 2 plants revealed that the abundance of miR156 in transgenic lines was higher than in WT (Fig. 2c, d).
In general, all organs from the miR156+ transgenic plants were smaller than those from the WT. During the first 3 months, the T 2 miR156+ plants grew slowly, showed a dwarf phenotype, and had fewer branches than the WT (Fig. 2e, f). However, following the WT transition to the reproductive stage (at week 16-18), the transgenic plants gradually out-grew the WT plants and eventually showed an exaggerated branching phenotype. Interestingly, the  Fig. 5B). Subsequently, the lateral shoots developed vigorously and emerged from almost every leaf axil of miR156+, giving those plants an overall bushy phenotype.

Identification of LjmiR156 targets
Based on psRNATarget analysis, we identified 13 candidate target genes of LjmiR156 in the L. japonicus genome (http://www.kazusa.or.jp/lotus/; version 6; released on 18 May 2010) ( Table 2). Among them were eight SPL genes, one WD-40 like gene, one predicted RNA-directed DNA polymerase (RdDP) coding gene, one HISTIDINE-PHOSPHO-TRANSFER PROTEIN gene (HPTP), and two genes encoding trafficking proteins (TP). Their expression profiles were further determined in the leaves, stems, and a mixture of roots and nodules (Fig. 3) from T 2 transgenic lines #20 and #22. Only GO023872 (HPTP) and TC78289 (TP) transcripts were consistently repressed in all three tissue samples of miR156+ plants compared to WT plants from all three tissue samples. AV417559 (SPL) and TC61877 (TP) were shown to be down-regulated in miR156+ plant stems and underground parts. Interestingly, TC57859, a WD-40 like protein coding gene, was only repressed in the underground parts of miR156+ plants. Furthermore, transcript levels of 3 SPL candidate genes, TC70253, TC70719, and TC69981, were only decreased in miR156+ plant stems but not in the other two samples. This investigation revealed that transcripts of these target genes were selectively controlled in miR156+ lines compared to WT. We then used modified 5′-RACE to validate the miR156 cleavage sites for all 13 candidate genes and we were only able to detect cleavage sites in three of the potential targets, two SPL-homologs, AU089181 and TC70253, and one WD-40 homolog TC57859 (Fig. 4a). Phylogenetic tree analysis showed that AU089181 had the highest homology to Arabidopsis SPL-13 (Fig. 4b).
TC70253 was not subjected to this analysis because the sequence length was insufficient.

LjmiR156 overexpression affects root development and nodulation
To investigate whether Ljmir156 has any effect on L. japonicus root growth and symbiosis with rhizobia, we examined root length, nodule number, and the development of infection threads (IT) in T 2 plants. At 7 dpi with M. loti, we observed IT on more than 95 % of the 'Gifu' WT roots but on only 10 % of the miR156+ roots ( Table 3). Roots of the transgenic plants were approximately 15 % shorter and had 50 % fewer nodules than the WT (Table 3). Curiously, we observed that the reduced root development at this early stage of miR156+ plants was consistent between M. loti inoculated plants and non-bacterial experiment (Table 3). By 14 days, flattened nodules could be seen in miR156+ plants, whereas nodules of WT were more rounded (Fig. 5b, c). By 21 dpi, the number of nodules on the miR156+ roots was still approximately 50 % of that in the WT (Table 3; Fig. 5a); however, by this stage there were no detectable differences in nodule morphology (data not shown). After 4 months of growth, the root system of miR156+ plants was still underdeveloped compared with the WT, but nodule morphology was the same for both (Fig. 6b). To determine the role of miR156 in regulating symbiosis, we examined the relative transcript levels of several key regulators related to nodulation. At 7 dpi, the transcripts levels of ENOD40-1 and ENOD40-2 were significantly decreased in roots of miR156+ plants compared to WT plants (Fig. 6c). Furthermore, 14 key genes involved in IT formation and nodule organogenesis were examined in plant roots at 14 and 21 dpi. At 7 dpi, no significant differences were detected in transcript levels of candidate genes in the roots of WT and miR156+ (Supplementary Fig. 6). In contrast, by 14 dpi, expression of Nfr1 was up-regulated, but Cerberus, CYCLOPS, POLLUX, nsp1, and Nin were down-regulated in miR156+ plants when compared to WT (Fig. 6d).

Discussion
Legume forages have an inherent advantage over other forage crops owing to their ability to fix nitrogen (Peter and Graham 2003). Here, we used the model legume Lotus japonicus to investigate the role of miR156 in determining legume plant architecture, flowering time, and degree of nodulation. Ectopic expression of miR156 in L. japonicus caused up to a 4-week delay in flowering time in two fertile plants, and most surviving miR156+ plants even failed to set flowers by 12 months after tissue culture (the duration of the experiment). The onset of flowering signals a transition from the vegetative to the reproductive stage, and unleashes significant deposition of lignin in the cell walls of some forage plants, which reduces the energy releasable when digesting cell wall structural carbohydrates (Hatfield et al. 2006). Hence, a delay in flowering time would allow for an extended period of vegetative growth, leading to enhanced yield and quality of vegetative biomass. MiR156+ plants also showed a two-to three-fold increase in the number of lateral shoot branches since overexpression of miR156 appeared to cause the primary shoots to lose apical dominance. Lateral shoots emerged from almost every axil, and these new lateral shoots also produced secondary laterals. These shoot phenotype changes gave miR156+ L. japonicus plants an extensive branching phenotype and a bushy appearance. Enhanced miR156 expression in L. japonicus also resulted in smaller size organs, including smaller leaves and siliques compared to WT. Other plant species with enhanced miR156 expression generally exhibited similar phenotypes, including enhanced branching, delayed flowering, dwarfed growth, and small leaves. Such phenotypes have been reported as a result of ectopic expression of miR156 in the dicot species, Arabidopsis thaliana (Wei et al. 2012), Brassica napus (Wei et al. 2010), Oryza sativa (Xie et al. 2012), Zea mays (Chuck et al. 2007), Solanum lycopersicum (Zhang et al. 2011), and a monocot species Panicum virgatum (Fu et al. 2012).
The common phenotypic alterations in these diverse species demonstrate that our LjmiR156 precursor is functional in L. japonicus, and suggests that the role of miR156 is highly conserved among different plant species. We also observed some other important impacts from overexpression of miR156 in L. japonicus. Most of the miR156+ plants failed to transition to the flowering stage during the 1-year period of our experiment. Moreover, those that did set flowers (only lines 20 and 22) produced a significantly smaller number of flowers when compared to WT (data not shown). Together with our data, this suggests a role for LjmiR156 in fertility and flowering through its SPL targets. In Arabidopsis, SPL-8 (a non-miR156 targeted SPL) and miR156-targeted SPLs (SPL-2, SPL-9, and SPL-15) function in regulating plant fertility (Xing et al. 2010(Xing et al. , 2013. Radically reduced flowering could force plant breeders to rely on vegetative propagation of crops or to rely on conducting seed increases in countries with longer growing seasons, factors which may increase the cost of maintaining and storing varieties. Therefore, from a practical perspective, lines should be selected with only a moderate delay in flowering onset when deploying miR156 in forage breeding. The root length of miR156+ plants was 10-20 % shorter than WT plants, but the roots of miR156+ plants had ~50 % fewer nodules than the WT L. japonicus plants indicating that ectopic expression of LjmiR156 affects plant symbiosis with rhizobia. To confirm this phenomenon was not caused by rhizobia directly, a non-bacterial inoculation experiment  Table 3 Characterization of symbiosis between WT and miR156+ T 2 L. japonicus plants For 7, 14, and 21 dpi (14-, 21-, and 28-day-old seedlings, respectively) experiment, at least 10 seedlings were collected and measured. Values are mean ± SE. Numbers in brackets were derived from a non-bacterial inoculated control experiment. Root length and nodule numbers from five 4-month-old inoculated seedlings for each line were also counted and measured, respectively. Values are mean ± SE (n ≥ 5) * Significant difference between miR156+ and WT plants at P < 0.05 (one-way ANOVA, Duncan's test) i These numbers indicate how many infection threads or nodule primordia were observed out of each 10 plants Here, we observed that the differences in root length and branching patterns between miR156+ and WT plants were consistent between these two conditions. So we believe that the impacts on root system were caused by miR156 rather than by rhizobia. Symbiosis with soil organisms has always been a valuable attribute for maintaining soil fertility and reducing fertilizer input. In other plants, functional characterization of miR156 focused mainly on above-ground traits, i.e., shoot branching, flowering, and seed composition (Fu et al. 2012;Wei et al. 2010;Wu and Poethig 2006). Here, we demonstrated the role for LjmiR156a in the symbiotic relationship between L. japonicus and rhizobia. The impact of this reduction in nodulation should now be examined in greater detail in miR156+ plants to determine whether resources (e.g., carbohydrates) were diverted away from symbiotic rhizobia to allocate more energy toward developing vegetative biomass.
To further elucidate the relationship between miR156 and nodulation, we examined 14 nodule organogenesisrelated genes in roots of miR156+ and WT plants at 7 and 14 dpi. The genes we tested have been shown earlier to affect many aspects of L. japonicus nodule development (Madsen et al. 2010), including Nfr1 and Nfr5 (for nodfactor perception); SymRK, Nup85, Nup133, Castor, and POLLUX (for signal transduction); Cyclops and CCaMK (for Ca 2+ signal interpretation); Cerberus (potential protein degradation); Lhk1 (for cytokinin signaling); and Nsp1, Nsp2, and Nin (for transcriptional regulation). At 7 dpi, Lhk1 and Nin were down-regulated while Nup133, POL-LUX, and CCamK were slightly up-regulated. At 14 dpi, only NFR1 was up-regulated, while POLLUX, CYCLOPS, Cerberus, nsp1, and Nin were down-regulated. Within these five genes, NIN has been shown to be involved in both infection thread formation and nodule organogenesis. As a nodulation-specific transcription factor, NIN is associated with the development of root nodule primordia (Schauser et al. 1999;Marsh et al. 2007), and its loss-of-function was shown earlier to inhibit root nodule organogenesis (Soyano et al. 2013).
Nodules of WT and miR156+ L. japonicus plants were morphologically similar at mature nodule stages (but had a flattened phenotype early on), while miR156+ plants had reduced numbers of infection threads and nodule primordia. Reduced transcription of the five key symbiotic Fig. 6 Root phenotype and relative transcript levels of nodulationrelated genes in roots (nd, not detectable). Means (±standard error) with same letter for the same gene are not significantly different. a 4-month-old WT and T 2 miR156+ plants. b Mature nodule structures. c, d Transcript levels of key genes in inoculated roots ◂ genes above at 14 dpi, but not at 7 dpi suggests that the reduced nodule number in the transgenic plants might not be directly related to these regulators at the transcriptional level. On the other hand, three ENOD genes were downregulated at 7 dpi in the miR156+ plants relative to the WT controls, with more than a twofold reduction observed for ENOD40-2 in both transgenic lines. These early nodulin (ENOD) genes were shown earlier to be induced during nodulation in L. japonicus (Takeda et al. 2005). Hence, we hypothesize that miR156 may function at the early stages of nodule biogenesis by directly or indirectly targeting ENOD40 genes. In the future, the three SPL genes that were down-regulated in the roots of miR156+ plants, AU089181, AV417559, and TC70253, could be tested to determine whether they are the direct link between miR156 and Nin, ENOD2, ENOD40-1 and -2. The fact that miR156 expression is induced under conditions of nitrogen deficiency (Pant et al. 2009) indicates a possible role in nutrient homeostasis, especially nitrogen fixation through plantmicrobe symbiosis. A recent study in soybean showed that overexpression of miR172 enhanced nodule numbers, whereas elevated miR156 expression repressed nodule formation (Yan et al. 2013). This is consistent with our findings on reduced nodulation in miR156+ L. japonicus plants. Taken together, our results suggest that overexpression of LjmiR156 inhibits nodule formation, and that miR156 temporally regulates nodulation-related genes.
Several plant hormones, including auxin and gibberellins can positively or negatively regulate nodulation (Maekawa et al. 2009;Deinum et al. 2012). These hormones function at different stages in the nodulation process and may participate in the coordination of epidermal and cortical cell development (Ding and Oldroyd 2009). The recently discovered phytohormone, strigolactone which is derived from carotenoids, has a positive impact on nodulation in Pisum sativum (Foo and Davies 2011) and inhibits shoot branching downstream of auxin in Arabidopsis (Brewer et al. 2009). Our previous results showed that miR156 affects carotenoid accumulation in B. napus and Arabidopsis (Wei et al. 2010;. Given that strigolactones are carotenoid catabolism products and that changes were observed in nodulation and shoot branching in miR156+ L. japonicus plants, strigolactones should now be tested in this species to determine whether they are impacted (directly or indirectly) by miR156 and the SPL regulatory system.
In summary, our data clearly show that LjmiR156 controls a network of downstream genes likely through the silencing of its target SPLs in the model legume L. japonicus, and that this system in turn regulates the expression of other downstream developmental and quality trait genes. Thus, by fine-tuning the control of this regulatory network, it may be possible to target specific desirable aspects of plant growth and development. For example, SPL-9 regulates the expression of TCL1 and TRY, which determine trichome density and distribution in Arabidopsis . As well, SPL-9 regulates dihydroflavonol reductase which is involved in flavonoid biosynthesis in Arabidopsis (Gou et al. 2011). Some quality traits, e.g., increased branching, delayed flowering, and higher biomass yield, are highly sought after by the forage industry; hence, our findings in L. japonicus are highly applicable to the world's major forage crops, such as Medicago sativa (alfalfa). As with other species, we anticipate that different target genes alone (or possibly in combination) will affect only one or a limited number of plant traits. Once their downstream target genes are identified and characterized, it may be possible to influence only desirable traits by manipulating the expression of specific target genes rather than the entire miR156 gene regulatory network. In the future, we intend to increase the level of miR156 genes in forage crops and leaf vegetable crops where extended vegetative growth is highly desirable. We will also investigate the role of its downstream targets to determine which genes can be selectively targeted to improve crop yield and quality with little or no impact on root growth and nodulation.