Plant Peptides Govern Terminal Differentiation of Bacteria in Symbiosis

See allHide authors and affiliations

Science  26 Feb 2010:
Vol. 327, Issue 5969, pp. 1122-1126
DOI: 10.1126/science.1184057


Legume plants host nitrogen-fixing endosymbiotic Rhizobium bacteria in root nodules. In Medicago truncatula, the bacteria undergo an irreversible (terminal) differentiation mediated by hitherto unidentified plant factors. We demonstrated that these factors are nodule-specific cysteine-rich (NCR) peptides that are targeted to the bacteria and enter the bacterial membrane and cytosol. Obstruction of NCR transport in the dnf1-1 signal peptidase mutant correlated with the absence of terminal bacterial differentiation. On the contrary, ectopic expression of NCRs in legumes devoid of NCRs or challenge of cultured rhizobia with peptides provoked symptoms of terminal differentiation. Because NCRs resemble antimicrobial peptides, our findings reveal a previously unknown innovation of the host plant, which adopts effectors of the innate immune system for symbiosis to manipulate the cell fate of endosymbiotic bacteria.

Symbiotic nitrogen fixation by legumes is a major contributor to the combined nitrogen pool in the biosphere. It takes place in specialized root organs called nodules (1). The symbiotic nodule cells are large polyploid cells (2) housing thousands of bacteroids. Bacteroids are differentiated Rhizobium bacteria with specialized metabolic activity, capable of reducing atmospheric nitrogen and supplying the plant with ammonium as a nitrogen source (3). In addition to this metabolic adaptation, the endosymbionts of Medicago truncatula and related legumes undergo striking morphological changes such as cell elongation coupled to genome amplification, membrane modifications, and the loss of reproductive capacity (46). The polyploid state of bacteroids and the induction of bacteroid-like cells by genetic interference with the rhizobial cell cycle (710) suggest that terminal bacteroid differentiation is a cell cycle–related process.

This terminal bacteroid differentiation is specific for legumes belonging to the inverted repeat–lacking clade (IRLC) such as Medicago, Pisum, or Trifolium, whereas bacteroids in the non-IRLC legumes, such as Lotus japonicus, show no sign of terminal differentiation as they maintain their normal bacterial size, genome content, and reproductive capacity (6). The same Rhizobium strains that form symbiosis with both IRLC and non-IRLC legumes have different bacteroid differentiation fates in the two legume types. Therefore, it was concluded that terminal bacteroid differentiation is determined by unknown host factors that are produced by the IRLC legumes and do not exist in the non-IRLC legumes (6). The nodule-specific cysteine-rich (NCR) peptides were likely candidates for these factors (11, 12). NCR genes were found only in the IRLC legumes (11, 12). In the tested cases, the expression of NCR genes was restricted to the Rhizobium-infected plant cells, where different subsets of NCR genes were activated during distinct developmental stages of the symbiotic cells (11). The NCR gene family encodes more than 300, highly divergent peptides in M. truncatula, which are most similar to defensin-type antimicrobial peptides (AMPs) (1113). Increased membrane permeability and definitive inhibition of cell division can be mediated by AMPs (14); therefore, we postulated that NCRs may have AMP-like activities and could be the critical plant factors that mediate terminal bacteroid differentiation.

We tested whether NCRs are translated and targeted to the bacteroids (15). SDS–polyacrylamide gel electrophoresis (SDS-PAGE) analysis of total protein extracts revealed the presence of peptides with molecular weight in the range of 3 to 5 kD in M. truncatula nodules and purified bacteroids, which were absent in roots or free-living Sinorhizobium meliloti, the endosymbiont of M. truncatula (Fig. 1A). This size range corresponded to the predicted molecular weight of the mature NCR peptides, and mass spectrometry confirmed their presence in the bacteroid extracts. Out of the eight sequenced peptides, seven corresponded to five different NCR peptides, whereas one peptide derived from a small protein of S. meliloti (table S1). Likewise, the NCR084 and NCR001 peptides were detected only in the nodule and bacteroid extracts by Western blot analysis (Fig. 1B). Different approaches were undertaken for in situ localization of NCR peptides inside the symbiotic cells. Transgenic M. truncatula nodules expressing NCR035 as a translational fusion with the mCHERRY red fluorescent protein under the control of its own promoter demonstrated that the NCR035 peptides were present only in the infected nodule cells, where they colocalized with the bacteroids (Fig. 1, C and D). Immunofluorescence localization of NCR001 with affinity purified antibodies also labeled the bacteroids in the symbiotic cells (Fig. 1, E and F), in accordance with the predicted role of NCRs in bacteroid differentiation. Immunogold transmission electron microscopy (TEM) demonstrated the localization of NCR001 in the bacteroid membrane but predominantly in the bacteroid cytosol (Fig. 1G and fig. S1), indicating that NCRs may have intracellular rhizobial targets, and their mode of action is not restricted to an interaction with the bacterial membrane.

Fig. 1

NCR peptides colocalize with bacteroids in M. truncatula nodules. (A) SDS-PAGE analysis and Coomassie blue staining of total protein extracts of roots (R), nodules (N), bacteroids (B), and cultured S. meliloti cells (C) reveal the specific presence of low-molecular-weight peptides (bracket) in nodules and bacteroids. (B) Western blot analysis of identical extracts with antibodies against plant leghemoglobin (Lb), bacterial dinitrogenase reductase (NifH), NCR001, and NCR084 confirms the presence of NCR peptides in nodules and their copurification with bacteroids. (C and D) In a nodule (C) and a symbiotic cell (D) expressing NCR035-mCHERRY, NCR035-mCHERRY (red signal) localizes to bacteroids in symbiotic cells of the interzone (II-III) and fixation zone (III). I, meristem; II, infection zone. (E and F) Immunofluorescence localization of NCR001 with Alexa 633 (red signal) in a nodule (E) and a symbiotic cell (F) demonstrates that NCR001 colocalizes with bacteroids in symbiotic cells of the fixation zone. (G) Immunogold localization (black dots) of NCR001 in a nitrogen-fixing symbiotic cell, confirming the presence of NCR001 in the bacteroids. B, bacteroid; C, cytoplasm of the symbiotic cell. Scale bars, 100 μm in (C) and (E); 10 μm in (D) and (F); 1 μm in (G).

Because of the high complexity of the NCR gene family and the probable redundancy of NCRs, reverse genetics approaches were unlikely to be effective to elucidate NCR functions. Bacteroids in symbiotic nodule cells are not in direct contact with the host cell cytoplasm, because they are surrounded by a plant-derived membrane forming an organelle-like structure, the symbiosome. This suggested that protein trafficking to symbiosomes probably depends on the secretory pathway (16). NCRs have a characteristic signal peptide which, as we showed in onion cells, targets them into the secretory pathway (11). Therefore, it was expected that interference with NCR trafficking may have a general effect and block the function of all NCR peptides. The M. truncatula dnf1-1 mutant is deficient in a nodule-specific component of the signal peptidase complex (SPC) of the secretory pathway (17). The SPC resides in the endoplasmic reticulum (ER) and removes the signal peptide of secretory proteins during their translocation into the ER lumen, which is an essential step for their proper targeting (18). The dnf1-1 mutant forms nonfunctional nodules (17), with comparable structure to wild-type (WT) nodules in which symbiotic cells are infected (fig. S2, A and B), but the bacteroids remain undifferentiated (fig. S2, C to F). This raised the possibility that the lack of bacteroid differentiation in this mutant is caused by improper targeting of NCRs. NCR001 and other NCRs were expressed at similar levels in dnf1-1 and WT nodules (fig. S2G). In the dnf1-1 nodules, low-molecular-weight peptides were present in the nodule extracts but absent in the bacteroid extracts (fig. S2H). Immunoblotting revealed the absence of mature NCR001 peptide in both the dnf1-1 nodules and bacteroids, as well as the presence of a higher-molecular-weight form of NCR001 only in the dnf1-1 nodules, probably corresponding to the full-length unprocessed NCR001 because the SPC is nonfunctional (fig. S2I). Immunofluorescence localization confirmed that NCR001 was not targeted to the bacteroids in the dnf1-1 mutant (Fig. 2, A to F). Instead, it colocalized with the ER (Fig. 2, G to L). Thus, prevention of proper targeting of NCR peptides to the bacteroids correlated with the absence of bacterial differentiation in this plant mutant.

Fig. 2

NCR001 peptides localize to the ER in the secretory pathway mutant dnf1-1. (A to F) Immunofluorescence localization of NCR001 [(A) and (D)], SYTO13 (a nucleic acid marker) staining of bacteroids [(B) and (E)], and an overlay [(C) and (F)] in a WT [(A) to (C)] and dnf1-1 [(D) to (F)] symbiotic cell shows that NCR001 no longer colocalizes with bacteroids in the dnf1-1 mutant background. (G to L) Immunofluorescence localization of NCR001 [(G) and (J)], the KDEL epitope (ER marker) [(H) and (K)], and an overlay [(I) and (L)] in a WT [(G) to (I)] and dnf1-1 [(J) to (L)] symbiotic cell revealing colocalization of NCR001 with the ER in dnf1-1. Scale bars, 10 μm.

To further support the role of NCRs in terminal bacteroid differentiation, we tested whether expression of NCR genes could induce features of the IRLC-specific bacteroid differentiation in L. japonicus nodules, where this bacteroid differentiation does not occur and the NCR genes are absent (11, 12). Different transgenic Lotus lines were created with nodule-specific expression of one of eight NCRs (15), selected on the basis of their expression pattern (11) or the β-glucuronidase (GUS) reporter gene as a control. Expression of each gene was controlled by the M. truncatula leghemoglobin 1 promoter (pMtLb1) that was active in L. japonicus symbiotic cells (fig. S3, A and B). Moreover, targeting of the NCRs to the symbiosomes was also confirmed in transgenic L. japonicus nodules expressing an NCR035-mCHERRY translational fusion under the control of pMtLb1 (fig. S3, C to E). TEM and scanning electron microscopy (SEM) of pMtLb1::GUS control transgenic nodules showed that symbiosomes generally harbored multiple small bacteroids (Fig. 3, A, C, and E), which is indicative of bacteroid division within the symbiosomes, similarly to WT L. japonicus nodules (19). Out of the eight NCR peptides, NCR035 produced the most pronounced effects on the bacteroids in L. japonicus nodules: The majority of the symbiosomes contained a single bacteroid, and many bacteroids were remarkably elongated (Fig. 3, B, D, and E) as in M. truncatula nodules, indicating that these bacteroids, in the presence of NCR035, had lost their capacity to multiply within symbiosomes. Moreover, an increased fraction (25% versus 18% in the control) of isolated bacteroids became permeable by the DNA-staining dye propidium iodide (PI), indicating that NCR035 affected the membrane structure of bacteroids. Thus, expression of NCR genes in L. japonicus nodules was sufficient to induce bacteroid morphologies reminiscent of terminally differentiated M. truncatula bacteroids.

Fig. 3

Expression of NCR genes in L. japonicus leads to features of terminal bacteroid differentiation. (A and B) TEM of a symbiotic nodule cell expressing GUS (A) and NCR035 (B) demonstrates an increase in bacteroid size in the case of NCR035 expression. b, bacteroid; cw, cell wall. (C and D) SEM of bacteroids in a symbiotic cell expressing GUS (C) and NCR035 (D) visualizes the morphological differences of bacteroids upon NCR035 expression. (E) Plotting of bacteroid size as a function of bacteroid number per symbiosome shows the presence of bacteroid populations with enlarged size in NCR035 transgenic lines (encircled); n = 132 bacteroids from two independent transgenic lines. The inset shows the proportion of bacteroids as a function of the number of bacteroids per symbiosome. In the control nodules, most symbiosomes contain more than one bacteroid, whereas in the NCR035-expressing lines, there is most frequently one bacteroid per symbiosome. Gray is the quantification for control lines and black for NCR035-expressing lines. Scale bars, 1 μm.

The observation that NCRs are important for terminal bacteroid differentiation in planta led us to postulate that synthetic NCR peptides may be active in vitro and provoke certain features of terminal bacteroid differentiation when added to S. meliloti cultures. When log-phase bacteria were treated with the peptides, membrane permeabilization was detected by the uptake of the PI dye in a peptide- and concentration-dependent manner (Fig. 4A). NCR035, NCR055, and NCR247 induced bacterial cell death, indicated by the decrease of colony-forming units in plating assays (Fig. 4B) and by the loss of respiratory activity quantified by a decrease in 5-cyano-2,3-di-4-tolyl tetrazolium chloride fluorescence (fig. S4A). These findings show that some NCR peptides possess antimicrobial activity and can therefore be classified as AMPs. Defensins can interact with negatively charged membrane components, and their activity is inhibited by divalent cations (14). Likewise, the activity of NCR peptides was inhibited by Ca2+ (fig. S4B), indicating that NCRs can also interact with negatively charged membrane components such as phospholipids or lipopolysaccharide. A distinctive feature of terminally differentiated bacteroids is their high DNA content, achieved by repeated replication cycles (6). To examine whether NCR peptides affect the bacterial cell cycle, the DNA content was measured by flow cytometry of S. meliloti cells treated with NCR035. Cell populations both in the haploid (1C) and diploid (2C) phase were present in the untreated control sample, whereas in the NCR035-treated sample, haploid cells were not detected; the majority of cells were diploid and a new population appeared with a 4C DNA content (Fig. 4C), indicating that the peptide inhibited cytokinesis while DNA synthesis was ongoing. Investigating further, we followed the uptake and localization of a bioactive fluorescein isothiocyanate (FITC)–labeled NCR035 peptide (fig. S4B) in bacterial cells. Two main patterns could be discerned: labeling of the cell envelope (Fig. 4D, asterisk) and intracellular marking of the bacterial cell division plane (Fig. 4D, arrow and arrowhead), which is in agreement with a role of NCR035 in cytokinesis inhibition. The inhibition of cytokinesis by NCR peptides resulted in enlarged S. meliloti cells, reminiscent of S. meliloti bacteroids (Fig. 4, E and F). Taken together, the in vitro responses of S. meliloti to NCR peptides partially mimicked characteristics of terminally differentiated bacteroids, notably provoking membrane modifications, inhibition of bacterial cytokinesis, and DNA amplification coupled with cell enlargement. However, the effect of NCR peptides on free-living bacteria was more dramatic than on bacteroids in M. truncatula nodules. NCR peptides rapidly abolished respiration and killed the free-living bacteria, in contrast to the nodule, where the bacteroid metabolic activity was maintained for symbiotic nitrogen fixation (6). These observations could be explained by a possibly lower concentration of peptides in the plant than the concentrations used in the in vitro assays. Moreover, the concerted action of many different NCRs produced within the same symbiotic cell or the particular physiological conditions prevalent in nodules (such as low free oxygen concentration or oxidative stress) might also be essential to modulate the bacterial responses in such a way that bacteroids remain alive, although with complete loss of their proliferative ability.

Fig. 4

NCR peptides provoke features of terminal bacteroid differentiation in vitro. (A) PI uptake of S. meliloti cells after treatment with different NCRs, showing that NCRs provoke membrane permeabilization of S. meliloti cells. Data are means ± SD (n = 4 measurements). (B) Cell kill assay of S. meliloti cells after treatment with NCRs disposing different cell killing capacities. Data are means ± SD (n = 3). (C) Increased DNA content of S. meliloti cells after NCR035 treatment in comparison to the control. (D) FITC-NCR035 localization in S. meliloti cells, initially at the cell envelope (asterisks), and later, intracellularly at the bacterial division site (arrowheads and arrows). (E and F) Control (E) and NCR-treated (F) S. meliloti cells showing an increase in size upon treatment with NCR peptides (arrowheads). Cells were stained with PI. Scale bars, 1 μm in (D); 10 μm in (E) and (F).

Cysteine-rich peptides are extremely abundant in plants (13, 20, 21). Some of these peptides have signaling roles (22, 23), but many of them are thought to have antimicrobial activity (20, 21). Our results demonstrate that in Medicago and probably in other IRLC legumes, the nodule-specific NCR peptides act as symbiotic plant effectors to direct the bacteroids into a terminally differentiated state and that these peptides have in vitro antimicrobial activity. Therefore, one can speculate that IRLC legumes recruited cysteine-rich AMPs in the context of symbiosis to evolve the nodule-specific NCR family and to dominate the endosymbionts. Although the fitness benefits to the host plant from such a bacteroid differentiation pathway remain to be clarified, the high number and diversity of peptides lead us to propose that NCR peptides interfere with many aspects of the bacteroid metabolism to allow the efficiency of the nitrogen fixation process to be optimized. For example, the peptides could be part of a mechanism to avoid the cheating of rhizobia that could use host resources to accumulate carbon storage compounds instead of fixing nitrogen. Such “freeloading” by bacteroids is often observed in the non-IRLC legumes but not in the IRLC (24). In addition, suppressing bacterial reproduction in conjunction with genome amplification and weakening of bacterial membranes by NCR peptides might be of benefit during nodule senescence, when the fragilized bacteroids can be efficiently digested by the plant cells (5, 25) and their nutrients reused by the host. Alternatively, the enlarged and polyploid bacteroids might have a more efficient metabolism, like polyploid eukaryotic cells (2), resulting in higher nitrogen fixation (24).

Supporting Online Material

Materials and Methods

Figs. S1 to S4

Table S1


References and Notes

  1. Information on materials and methods is available as supporting material on Science Online.
  2. We thank S. Long, T. Bisseling, G. Ferguson, and R. Whitford for helpful comments on the manuscript; S. Brown, M. Yamaura, P. Laporte, P. Durand, É. Klement, E. Magyaródi, and the IMAGIF platform for technical support; and S. Long, D. Wang, D. Barker, P. Ludden, and N. Suganuma for providing materials. This work was supported by the Agence National de la Recherche project ANR-05-BLAN-0129 (W.V.d.V., A.S., E.K. and P.M.), the Japanese Society for the Promotion of Science (G.Z. and T.U.), the Hungarian National Office for Research and Technology Teller program OMFB-00441/2007 (M.D., A.F., K.M., A.N., H.T., A.K., G.M. and E.K.), and a Ph.D. fellowship from the French Ministry of Research (B.A.). E. Kondorosi et al. European patent no. 09305547.3-2107 related to the antimicrobial activity of NCR peptides is pending.
View Abstract

Stay Connected to Science

Navigate This Article