Systemic control of legume susceptibility to rhizobial infection by a mobile microRNA

See allHide authors and affiliations

Science  12 Oct 2018:
Vol. 362, Issue 6411, pp. 233-236
DOI: 10.1126/science.aat6907

Keeping the doors open for symbiosis

Nitrogen fixation by legumes results from a symbiotic partnership between plant and microbes. These together elaborate nodules on the plant roots that house the bacteria. Tsikou et al. identified a microRNA made in the aboveground shoots of Lotus japonicus that translocates to the plant's roots. In the roots, the microRNA posttranscriptionally regulates a key suppressor of symbiosis, thus keeping the uninfected root susceptible to productive infection by symbiotic bacteria.

Science, this issue p. 233


Nitrogen-fixing root nodules on legumes result from two developmental processes, bacterial infection and nodule organogenesis. To balance symbiosis and plant growth, legume hosts restrict nodule numbers through an inducible autoregulatory process. Here, we present a mechanism where repression of a negative regulator ensures symbiotic susceptibility of uninfected roots of the host Lotus japonicus. We show that microRNA miR2111 undergoes shoot-to-root translocation to control rhizobial infection through posttranscriptional regulation of the symbiosis suppressor TOO MUCH LOVE in roots. miR2111 maintains a susceptible default status in uninfected hosts and functions as an activator of symbiosis downstream of LOTUS HISTIDINE KINASE1–mediated cytokinin perception in roots and HYPERNODULATION ABERRANT ROOT FORMATION1, a shoot factor in autoregulation. The miR2111-TML node ensures activation of feedback regulation to balance infection and nodulation events.

Development of nitrogen-fixing nodules in roots of legumes such as soybean, pea, and Lotus japonicus is induced by rhizobial signal molecules. Nodule numbers are systemically controlled by host plants through a process called autoregulation of nodulation (AON). In the current AON model, the onset of nodulation leads to the transport of root-synthesized CLE peptides to Lotus shoots, where they are perceived by the CLAVATA1-like leucine-rich repeat receptor–like kinase HYPERNODULATION ABERRANT ROOT FORMATION1 (HAR1) (13). The resulting shoot-to-root signaling was proposed to involve cytokinins as mobile molecules (4). Restriction of nodule emergence is effected by TOO MUCH LOVE (TML), a root-active kelch-repeat F-box protein (5), by an unknown mechanism. This model assumes that AON is induced by rhizobia. In this study, we have shown that shoot-derived miR2111 acts in the absence of infection to unlock root responsiveness to rhizobial nodulation (Nod) factors, ensuring susceptibility to infection.

Lotus mature miR2111 levels negatively respond to infection with Mesorhizobium loti in both leaves and roots within 2 days (Fig. 1A and fig. S1). Screening of the Lotus genome (6, 7) revealed three genomic loci representing potential MIR2111 precursor genes (fig. S2A) encoding three miR2111 isoforms, miR2111a to miR2111c (fig. S2A). The MIR2111-1 transcript contains a single copy of miR2111a (fig. S2, A and B), whereas MIR2111-2 and -3 are polycistronic, containing two isoforms each in consecutive stem-loops (fig. S2, A, C, and D).

Fig. 1 miR2111 regulates TML posttranscriptionally.

(A) miR2111 abundance in Lotus leaves (dark bars) and roots (light bars) at 1 to 4 dpi with M. loti. inf., infected; uninf., uninfected. (B) Infection thread (IT) (10 dpi) and (C) nodule (21 dpi) numbers in pUBQ1::MIR2111-2 (-2) and pUBQ1::MIR2111-3 (-3) roots compared to control (cv) roots. (D and E) Nodulation in control (D) and pUBQ1::MIR2111-3 (E) roots (21 dpi). Right-hand panels visualize M. loti DsRED in nodules. Scale bars, 2 mm. (F) miR2111 directs TML cleavage. Bold font marks polymorphisms between miR2111 isoforms a to c (black). Numbers: degradome 5′ ends at arrowhead/total within TML target region (blue). (G) TML mRNA in M. loti–infected roots (1 to 4 dpi). (H to J) miR2111STTM (STTM) expression reduced miR2111, increased TML (H), and reduced nodulation [(I) and (J)] compared to those in control roots (cv). (I) n = 23/26 (miR2111STTM roots/control roots). Green fluorescence [(J), center] shows cotransformation; red [(J), right] indicates nodules with M. loti DsRED. Scale bar, 5 mm. Transgenic roots [(B) to (E) and (H) to (J)] were A. rhizogenes induced. [(A), (G), and (H)] qRT-PCR analyses. RNA levels are relative to those for two reference genes. Dashed lines mark 1 as the reference value in ratio graphs. Error bars [(A) to (C) and (G) to (I)]: SEM of at least three biological replicates. Student’s t test P values: *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001. (I) Results represent one biological replicate (P = 0.006) and were similar in a second (P = 0.001).

To understand the biological relevance of the decline in miR2111 abundance during rhizobial infection, we expressed the MIR2111-2 and MIR2111-3 precursor genes under the control of the LjUBIQUITIN1 (UBQ1) promoter. pUBQ1::MIR2111-3–expressing roots (fig. S3A) were hyperinfected (Fig. 1B). Overexpression of three other miRNAs by the same system had no phenotypic effect (8, 9), making hypernodulation of pUBQ1::MIR2111-3 roots unlikely to reflect indirect effects of miRNA overexpression. No qualitative defects in root hair responses or infection were apparent (fig. S3, B and C). At 21 days postinfection (dpi), pUBQ1::MIR2111-3 roots had more nodules than controls (Fig. 1C). The mature nodules were smaller than those on wild-type roots but were infected and morphologically normal (fig. S3, D to G), and the proportion of immature nodules was increased to 72.8% from 20.7% in control roots. Infection events (fig. S4, A and B) and nodules (Fig. 1, D and E) were present along a wider fraction of pUBQ1::MIR2111-3 roots, suggesting an expanded susceptibility zone where infections can take place. Overexpression of the 5′ stem-loop of MIR2111-2, encoding a single copy of miR2111b, suggests a dosage-dependent effect of miR2111 (Fig. 1, B and C, and fig. S3A).

To gain insight into the molecular context of miR2111 activity, we queried a degradome dataset prepared from nodule and root tissues (10) for targets of miR2111-directed endonucleolytic cleavage. This revealed TML as an in vivo target of miR2111 (Fig. 1F and fig. S5A). miR2111, as well as target sites in putative TML orthologs, is conserved in dicotyledonous plants (fig. S5, B and C). TML mRNA was rare in pUBQ1::MIR2111-3 roots (fig. S3A), and its abundance increased as miR2111 in roots decreased at 2 to 4 dpi (Fig. 1G and fig. S5D). MIR2111-3 overexpression roots (Fig. 1, B, C, and E) thus phenocopy tml loss-of-function mutants (5, 11) (fig. S5E). TML promoter activity is constitutive in Lotus roots (5), and our data demonstrate miR2111-dependent posttranscriptional regulation of TML mRNA abundance. To investigate miR2111 roles in symbiosis control, we overexpressed a short tandem target mimic (STTM) construct to sequester mature miR2111 (pUBQ1::miR2111STTM). This reduced miR2111 abundance, and TML levels were increased (Fig. 1H). In line with a positive effect of miR2111, pUBQ1::miR2111STTM roots formed fewer nodules than controls (Fig. 1, I and J).

TML mRNA was not detected in leaf tissue (fig. S5F) (12). Yet, the M. loti–induced miR2111 response was systemic (Fig. 1A). We thus hypothesized that miR2111 might undergo shoot-to-root translocation. To locate miR2111 source tissues, we generated stable pMIR2111-3::GUS–expressing plants. Four independent lines showed GUS activity in leaf phloem but none in roots (Fig. 2, A to D, and fig. S6, A to F). GUS signal in pMIR2111-2::GUS lines was below the visual detection limit, but quantitative reverse transcription–polymerase chain reaction (qRT-PCR) detected GUS transcripts (fig. S6F). Agrobacterium rhizogenes–induced pMIR2111-1::GUS, -2::GUS, or -3::GUS roots showed no detectable GUS signal (fig. S6, G to L) and low GUS transcript levels (fig. S6F). These data suggest that whereas mature miR2111 is present in roots, it is synthesized primarily in shoots.

Fig. 2 MIR2111-3 expression and translocation to roots.

(A to D) GUS activity in pMIR2111-3::GUS–expressing plants (2 weeks). [(A) and (B)] GUS activity in phloem (p) cells of higher-degree veins of mature leaves (A) and cotyledons (B). No GUS was detected in leaf xylem (x), shoot apices (sa), petioles (pt), and stems (s) (A) or roots [(C) and (D)]. co, cortex; e, epidermis; ml, mature leaflets; vb, vascular bundle; yl, young leaflets. Scale bars, 1 mm [(A) and (D)]; 50 μm [(B) and (C)]. (E) qRT-PCR analyses of miR2111 and TML levels in uninfected roots following root-shoot separation. Error bars: SEM of three biological replicates. **P ≤ 0.01; ***P ≤ 0.001. (F to K) TML loss [(H) and (I)] or pUBQ1::MIR2111-3 expression [(J) and (K)] rescues the asymbiotic phenotype of shootless wild-type roots (G). Nodule counts (F) were at 3 weeks postinoculation with M. loti expressing DsRED on A. rhizogenes–induced wild-type (wt) (G) or tml-1 [(H) and (I)] roots (expressing control vector) and wild-type roots expressing pUBQ::MIR2111-3 [(J) and (K)]. (F) Error bars show SEM of two biological replicates (n = 12, 13, and 11 total root systems, respectively). ox, pUBQ1-mediated overexpression. Scale bars, 1 cm (G); 1 mm [(H) to (K)].

If root miR2111 was of shoot origin, shootless roots should have low miR2111 levels. In line with this hypothesis, 3 days after mechanical separation from shoots, roots showed less miR2111 (Fig. 2E and fig. S7A), whereas induction of the infection-responsive miR172 (9) demonstrated that de novo miRNA synthesis was unimpaired (fig. S7B). Consistent with reduced miR2111 content, shootless roots had more TML mRNA (Fig. 2E and fig. S7C). Infection led to further increase in TML mRNA (fig. S7C), pointing to additional transcriptional regulation of TML mRNA abundance. miR2111 observed in shootless roots (fig. S7A) may reflect miRNA stability or low-level local expression. We further identified miR2111 in flow-out liquid collected from cut-off shoots (table S1), suggesting its presence in phloem sap. These shoots contained miR2111 levels similar to those in shoots of intact plants (fig. S7D), suggesting that miR2111 biosynthesis remained active in cut-off shoots. Thus, miR2111 was translocated from shoots to roots in vivo, and shoot-root separation interrupted translocation (fig. S7E).

In uninfected plants, active TML transcription (5) may be counterbalanced posttranscriptionally by shoot-derived miR2111, ensuring susceptibility. To test whether TML loss or miR2111-mediated transcript reduction can alleviate the TML-imposed blockage, we removed shoots from tml-1 or wild-type roots expressing pUBQ1::MIR2111-3, respectively. Small but infected nodules were formed in the presence of M. loti (Fig. 2, F and H to K), but we never observed nodules on wild-type roots expressing a control vector (Fig. 2, F and G). Together, these results demonstrate that interactions between TML and miR2111 control root susceptibility and that shoot-produced miR2111 contributes to regulating root TML levels.

Like TML loss (11) (fig. S8A) or MIR2111 overexpression (Fig. 1B), knockout of the cytokinin receptor LHK1 results in hyperinfection (13) (fig. S8A). Infected lhk1-1 mutants maintained miR2111 levels upon infection in both roots and shoots (Fig. 3A and fig. S8, B and C), suggesting a systemic negative effect of LHK1-dependent cytokinin signaling on miR2111 accumulation. Grafting revealed root genotype dependence of lhk1-1–associated hyperinfection (Fig. 3B). As cytokinin biosynthesis is first detected in the root cortex rather than in the epidermal cells where infection is initiated (14), these results suggest a connection between the miR2111-TML module and the inducible AON system. The latter involves induction of CLE-RS1, -2, and -3 upon infection (3, 15). In 10-day-old plants that retained seed-derived nutrient supplies and were moderately nitrogen sufficient, we found that lhk1-1 roots showed no infection-dependent CLE-RS3 and impaired CLE-RS1 accumulation (fig. S8D). This suggests a role of LHK1 in inducing autoregulation in response to infection. Supporting this, root expression of selected CLE-RS genes is inducible by cytokinin application in an LHK1/CRE1-dependent manner (16, 17). Dependence of CLE gene expression on LHK1 places root cytokinin perception upstream of HAR1 in AON, in line with nodulation inhibition in lhk1 roots grafted to CLE-RS1 overexpression shoots (4), and does not exclude involvement of shoot-derived cytokinin (4) at a later stage. Consistent with lhk1-1 hyperinfection (13) (Fig. 3B and fig. S8A), TML mRNA levels are reduced in lhk1-1 mutants compared to those in the wild type at 3 dpi (fig. S8E).

Fig. 3 Systemic regulation of miR2111.

(A) Infection-dependent reduction of mature miR2111 levels in wild-type leaves and roots depends on HAR1 and LHK1. Plants were harvested uninfected or 3 dpi with M. loti and were analyzed by qRT-PCR. (B) Infection thread (IT) counts on roots of grafted wild-type, lhk1-1, and chimeric plants. n = 3, 5, 6, 8 root systems (left to right). [(A) and (B)] Error bars show SEM of at least three biological replicates. Comparisons used Student’s t test (infected versus uninfected; *P ≤ 0.05, **P ≤ 0.01) (A) or analysis of variance (ANOVA) and post hoc Tukey testing (B) (P = 0.011), with distinct letters indicating significant differences.

TML functions downstream of HAR1 in symbiotic autoregulation. Infection-triggered miR2111 regulation was HAR1 dependent (Fig. 3A and fig. S8, B and C), and TML mRNA levels were reduced in har1-3 mutant roots compared with the wild type (fig. S8E). Although har1 mutants are hypernodulating, uninfected har1-3 plants contained less miR2111 than the wild type (fig. S8, B and C). This suggests that HAR1 positively affects miR2111 abundance in uninfected plants and that additional factors contribute to TML regulation.

Thus, the miR2111-TML module operating in uninfected roots connects to the symbiotically triggered AON (fig. S9). This is dependent on LHK1-mediated cytokinin signaling in roots, and CLE-activated shoot HAR1 modifies miR2111 production and/or shoot-root translocation, thereby contributing to the release of root TML from posttranscriptional miR2111 control. These interconnected regulatory systems ensure that transcription of TML in uninfected roots (5) does not prevent symbiosis but restricts its progression immediately upon contact with compatible rhizobial bacteria (fig. S9).

Apart from infection, nitrogen availability also controls symbiosis in a HAR1-dependent manner (18). In line with this, miR2111 abundance was reduced in plants supplied with nitrate (fig. S10A). Whereas nitrogen homeostasis may induce AON once fixing nodules are present, both infection and organogenesis control set in before nodule-derived nitrogen is available. We found that Nod factor receptor nfr1-1 (19) and nfr5-1 (19) mutants failed to downregulate miR2111 (fig. S10B), suggesting that the release of miR2111-mediated TML regulation is triggered by bacterial signaling molecules at the initiation of symbiotic cross talk, resulting in symbiosis control being promptly effective upon infection. In summary, miR2111 is a systemic symbiosis activator constitutively repressing AON in uninfected roots, thereby maintaining susceptibility to rhizobial infection. This mechanism highlights the importance of immediate host control during symbiosis establishment.

Supplementary Materials

Materials and Methods

Figs. S1 to S10

Tables S1 and S2

References (2038)

References and Notes

Acknowledgments: We thank M. Kawaguchi for tml-1 seeds; F. Pedersen and K. A. Kristensen for plant care; M. Nadzieja for help with imaging; and Y. Kawaharada, S. Radutoiu, and C. Gutjahr for discussions. We apologize to authors whose work could not be cited due to space limitations. Funding: This work was funded by the Danish National Research Foundation (grant DNRF79); ERC (advanced grant 268523); German Research Foundation (grant CRC1101, project C07); and Ministry of Science, Research and Art of Baden-Wuerttemberg (Az:7533-30-20/1). Author contributions: D.T., Z.Y., K.M., D.B.H., N.B.A., D.E.R., L.H.M., H.B., and M.S. performed experiments and analyzed data; K.M., D.T., Z.Y., and J.S. conceived of and designed research; K.M. wrote the paper with input from co-authors. Competing interests: The authors declare no competing interests. Data and materials availability: All data are available in the main text or supplementary materials. For material requests please contact the corresponding author.
View Abstract

Stay Connected to Science

Navigate This Article