Symbiotic implications of type III protein secretion machinery in Rhizobium

Authors

  • Virginie Viprey,

    1. Laboratoire de Biologie Moléculaire des Plantes Supérieures (LBMPS), Université de Genève, 1 ch. de l'Impératrice, 1292 Chambésy, Geneva, Switzerland.,
    Search for more papers by this author
  • Aixa Del Greco,

    1. Laboratoire de Biologie Moléculaire des Plantes Supérieures (LBMPS), Université de Genève, 1 ch. de l'Impératrice, 1292 Chambésy, Geneva, Switzerland.,
    Search for more papers by this author
  • Wladyslaw Golinowski,

    1. Department of Botany, Institute of Plant Biology, Warsaw Agricultural University, Rakowiecka 26/30, 02-528 Warsaw, Poland.
    Search for more papers by this author
  • William J. Broughton,

    1. Laboratoire de Biologie Moléculaire des Plantes Supérieures (LBMPS), Université de Genève, 1 ch. de l'Impératrice, 1292 Chambésy, Geneva, Switzerland.,
    Search for more papers by this author
  • Xavier Perret

    1. Laboratoire de Biologie Moléculaire des Plantes Supérieures (LBMPS), Université de Genève, 1 ch. de l'Impératrice, 1292 Chambésy, Geneva, Switzerland.,
    Search for more papers by this author

William J. Broughton E-mail broughtw@sc2a.unige.ch; Tel. (22) 906 1740; Fax (22) 906 1741.

Abstract

The symbiotic plasmid of Rhizobium sp. NGR234 carries a cluster of genes that encodes components of a bacterial type III secretion system (TTSS). In both animal and plant pathogens, the TTSS is an essential component of pathogenicity. Here, we show that secretion of at least two proteins (y4xL and NolX) is controlled by the TTSS of NGR234 and occurs after the induction with flavonoids. Polar mutations in two TTSS genes, rhcN and the nod-box controlled regulator of transcription y4xI, block the secretion of both proteins and strongly affect the ability of NGR234 to nodulate a variety of tropical legumes including Pachyrhizus tuberosus and Tephrosia vogelii.

Introduction

Under nitrogen-limiting conditions, soil bacteria of the genera Azorhizobium, Bradyrhizobium and Rhizobium (collectively called rhizobia) form symbiotic associations with leguminous plants. Infectious rhizobia penetrate the root cortex and form nodules. Within the nodules, the bacteria enlarge and differentiate into nitrogen-fixing bacteroids. In return for carbohydrates, bacteroids exchange readily assimilated forms of nitrogen with their hosts. Symbiotic specificity varies from rhizobia that are devoted to one or a few legumes to Rhizobium sp. NGR234, which nodulates more than 110 genera of legumes as well as the non-legume Parasponia andersonii (Trinick, 1980; Lewin et al., 1987; S. Pueppke and W. J. Broughton, unpublished). Among the factors responsible for this promiscuity, the symbiotic regulator NodD1 and the variety of nodulation factors (Nod factors) secreted by this strain are key components (Bender et al., 1988; Price et al., 1992). Although the mechanisms behind the regulation of nod gene expression are highly variable, they all require nodD gene(s) (Fellay et al., 1995a). In NGR234, NodD1 acts both as a sensor of the plant signal (phenolic compounds, such as flavonoids excreted by plant roots) and as a transcriptional activator of the nod genes. NodD proteins belong to the LysR family of transcriptional activators and bind to specific sequences (nod-boxes) in the promoter regions of flavonoid-inducible nodulation genes. In turn, many nodulation genes encode enzymes responsible for the elaboration of Nod factors, which initiate nodule development and permit bacterial invasion.

NGR234 contains a single replicon of 536 kb (pNGR234a) that carries most, although not all, symbiotic loci (Perret and Broughton, 1997). Interestingly, 10 of the 416 genes predicted from the nucleotide sequence of pNGR234a encode components of a type III secretion system (TTSS; Freiberg et al., 1997). In certain bacterial pathogens, TTSSs are induced upon contact with host cells and deliver virulence proteins directly into the eukaryotic cytosol (Lee, 1997). Prokaryotic loci responsible for eliciting the hypersensitive response and pathogenesis in plants are called hrp genes (Lindgren et al., 1986) and have been found in four major genera of Gram-negative bacteria: Erwinia, Pseudomonas, Ralstonia (which includes Pseudomonas solanacearum) and Xanthomonas (Willis et al., 1991). Recently, nine conserved hrp genes encoding components of the TTSS machinery of plant pathogens were renamed hrc (hrp conserved) (Bogdanove et al., 1996). Most hrp and hrc genes of the phytopathogenic bacteria E. amylovora, P. syringae pv. syringae, R. solanacearum and X. campestris pv. vesicatoria form clusters of about 25 kb (Alfano and Collmer, 1997). Homologues of all nine hrc genes are clustered on pNGR234a and are adjacent to the plasmid locus, which contains most of the essential nitrogen fixation (nif and fix) genes (Freiberg et al., 1997). Here, we show that the TTSS of NGR234 is inducible by flavonoids, controls the excretion of at least two proteins, NolX and y4xL, and profoundly affects the nodulation of some legumes.

Results

Located between nucleotides 494 000–524 000, the TTSS cluster of NGR234 contains 27 predicted open reading frames (ORFs) delimited by y4xI and y4yS (Fig. 1A). Included within the region are homologues of nolXWBTUV, which were first described in the closely related R. fredii strain USDA257, where they regulate the nodulation of Glycine max (soybeans) in a cultivar-specific manner (Meinhardt et al., 1993). Although NolT and NolW share significant homology with HrpB3 (HrcJ) and HrpA1 (HrcC) of X. campestris, only the complete sequence of pNGR234a clearly established that homologues to all hrc genes were present and physically linked in the genome of NGR234. To distinguish clearly between TTSS components of phytopathogenic bacteria and symbiotic rhizobia, we named the rhizobial ‘Hrc’ homologues ‘Rhc’ (for rhizobia conserved), but retained the last letter code of the unified nomenclature (Bogdanove et al., 1996). Thus, the HrcN counterpart in NGR234 became RhcN, etc. (Table 1).

Figure 1.

. Type III secretion system of NGR234. A. Genomic organization of the TTSS locus of pNGR234a. Predicted ORFs are represented by large coloured arrows oriented according to the sense of transcription. Genes that encode proteins associated with the secretion machinery in (C) are marked in red, while the ORFs coding for the secreted proteins (y4xL and nolX ) and the putative transcription regulator (y4xI) are coloured in green and yellow respectively. Position of the functional nod-box regulatory sequence upstream of y4xI is marked with a yellow triangle. Positions of the Omega interposon within the y4xI (mutant NGRΩy4xI) and rhcN (NGRΩrhcN ) genes are shown above the line. B. High-resolution transcription map of the TTSS locus of pNGR234a. Positions of the amplified products corresponding to genes or intergenic regions reported in (A) are shown as coloured boxes directly below the genetic map. RNA probes were prepared from: RMM, NGR234 cells grown at 28°C in liquid Rhizobium minimal medium (RMM) supplemented with succinate; daidzein 1 h, cells grown in RMM followed by a 1 h induction with 2 × 10−7 M daidzein; daidzein 24 h, RMM-grown rhizobia collected 24 h after induction with daidzein; nod. Vigna, bacteroids purified from Fix+ nodules of V. unguiculata inoculated with NGR234; nod. Cajanus, bacteroids purified from Fix+ nodules of Cajanus cajan inoculated with NGR234. Intensities of hybridization signals detected on autoradiograms are colour coded as follows: violet, strong (level 3); red, medium (level 2); orange, weak (level 1); and white, no signal detected. C. Hypothetical structure of the TTSS apparatus of NGR234. This model was derived from that proposed by Baker et al. (1997), with NGR234 gene products replacing their P. syringae homologues.

Table 1. . Products of pNGR234a homologous to proteins of various type III pathways. a. Protein names of the unified nomenclature for conserved TTSS components of plant and animal pathogens.Percentage identity and percentage similarity were calculated for the entire protein except for:b. which used the last 262 amino acids of the C-terminal domain of Yersinia enterocolitica YscC.c. which used the first 229 amino acids of the N-terminal domain of Y. enterocolitica YscC.d. which used the last 142 amino acids of the C-terminal domain of y4yQ.Thumbnail image of

Both experimental and sequence analyses suggest that hrc genes code for an outer membrane protein (HrcC), a lipoprotein-associated outer membrane protein (HrcJ), five inner membrane proteins (HrcR, HrcS, HrcT, HrcU and HrcV), as well as two cytoplasmic proteins (HrcQ and HrcN). Together, these proteins form part of the type III secretion machinery of plant pathogens. Based on this information and the scheme published for the Hrp secretion apparatus of P. syringae pv. syringae (Baker et al., 1997), we propose a model of the TTSS pathway of NGR234 (Fig. 1C). Most components are structurally well conserved except for the outer membrane protein HrcC, two homologues of which were identified in NGR234, RhcC1 (known as NolW in USDA257) and RhcC2 (encoded by y4xJ), corresponding to the N-terminal and C-terminal domains of HrcC respectively. Conversely, the smaller HrpU and HrpU2 products of P. syringae (named HrcQa and HrcQb; Baker et al., 1997) correspond to the single RhcQ protein of NGR234, just as in the E. amylovora, R. solanacearum and X. campestris systems.

TTSS are probably not ubiquitous in rhizobia

Most probably, TTSS systems are present in R. fredii, as many species possess conserved nolXWBTUV loci (Bellato et al., 1997) and export SR proteins into the media (Krishnan et al., 1995). Furthermore, genomic hybridizations have shown that rhc homologues are found in B. elkanii USDA76 and B. japonicum CB756 but not in R. meliloti 2011 and R. loti NZP4010 (data not shown). Sequence analysis of the chromosomal symbiotic region in B. japonicum USDA110 also confirmed the presence of a well-conserved TTSS (Palacios et al., 1998). Thus, although TTSSs are widespread, they are apparently not ubiquitous among phytosymbiotic prokaryotes.

Secretion of two pNGR234a-encoded proteins is controlled by the TTSS

Several lines of evidence suggest that the TTSS of NGR234 is functional and involved in nodulation. Earlier studies have shown that flavonoids that induce the expression of nodulation genes in USDA257 also stimulate the export of proteins into the extracellular medium (Krishnan and Pueppke, 1993). Furthermore, the secretion of five signal-responsive proteins (SR1 to SR5) was abolished when the cultivar specificity locus nolXWBTUV was mutated (Krishnan et al., 1995). To test whether flavonoids play a similar role in NGR234, proteins were isolated from the supernatants of apigenin (or daidzein)-induced and non-induced cultures of the wild-type strain. Separation of the soluble proteins by electrophoresis on polyacrylamide gels revealed two major proteins (SP1 and SP2) that were only present in the supernatants of flavonoid-induced wild-type cultures (cf. Fig. 2A, lanes 1 and 2).

Figure 2.

. Identification of Rhc-dependent extracellular proteins by one- and two-dimensional polyacrylamide gel electrophoresis. A. Extracellular proteins were electrophoretically separated on 12% SDS–PAGE. Lane 1, non-induced culture of NGR234; lane 2, apigenin-induced culture of NGR234; lane 3, apigenin-induced culture of NGRΩrhcN; lane 4, apigenin-induced culture of NGRΩy4xI; lane 5, apigenin-induced culture of NGRΩnodD1. B. Sections of two-dimensional gels of extracellular proteins from apigenin-induced cultures of NGR234 (left) and NGRΩrhcN (right). Labelled arrows indicate the positions of the two secreted proteins SP1 and SP2 that were identified by N-terminal microsequencing as y4xL (open circle) and NolX (black circle) respectively. Gels were silver stained, and the estimated molecular weights (MW) are given in kDa.

To verify that the secretion of SP1 and SP2 is TTSS dependent, polar mutants were constructed in rhcN (NGRΩrhcN) and y4xI (NGRΩy4xI; Fig. 1A). RhcN shares many characteristics of ATPases (Walker box A to C, αβ-subunit signature sequence) and probably energizes the secretion process at the cytosolic face of the inner membrane (Woestyn et al., 1994). Genetic analysis as well as transcription data suggest that a polar mutation within rhcN also affects the expression of six downstream genes, five of which (rhcQ to rhcU ) code for components of the TTSS model (Fig. 1A and B). y4xI is predicted to encode a transcription regulator homologous to HrpG, a key regulatory protein of hrp genes in X. campestris pv. vesicatora (Wengelnik et al., 1996). The absence of both SP1 and SP2 proteins in apigenin-induced supernatants of NGRΩrhcN (Fig. 2A, lane 3; Fig. 2B, right), NGRΩy4xI and NGRΩnodD1 (Fig. 2A, lanes 4 and 5) confirmed that the rhcN locus is required for their secretion. Similarly, these data show that the synthesis and/or export of SP1 and SP2 is both NodD1 and y4xI dependent.

Two-dimensional gel electrophoresis (Fig. 2B) followed by Western blotting was used to purify sufficient amounts of each of the two proteins for N-terminal amino acid sequencing. By comparing the short sequences derived for SP1 (Met-Asp-Ile-Asn-Ser-Thr-Ser-Pro) and SP2 (Ser-Ala-Ser-Asn-Leu-Leu-Pro-Met) with the putative proteins encoded by the 416 predicted ORFs of pNGR234a, we were able to identify y4xL and NolX as the secreted proteins. Gene y4xL is 10.5 kb upstream of nolB (Fig. 1A) and encodes a 338-amino-acid product, with a non-modified molecular mass of 37 kDa and a predicted pI of 5.3 (SWISSPROT P55704). Both predicted pI and molecular mass correlate well with the migration of y4xL on gels (Fig. 2). Database searches revealed no significant similarities, nor were any common motifs detected. Southern blot analysis confirmed that y4xL is also present in USDA257 however (data not shown). As the peptide sequence perfectly matches the first eight predicted amino acids of y4xL, it is exported without N-terminal processing, supporting the hypothesis that this protein is secreted by the TTSS (see Lee, 1997). The peptide sequence of SP2 matches residues 2–9 of NolX, a protein with a predicted molecular mass of 63.9 kDa and a pI of 5.3 (SWISSPROT P55711). Although the loss of the N-terminal methionine is not frequently observed in proteins secreted by the TTSS machinery, it was also found in YopN of Yersinia pseudotuberculosis (Forsberg et al., 1991). Apart from NolX's almost identical counterparts in USDA257 and USDA191 (only six replacements in 596 amino acids), where it is involved in host range determination (Bellato et al., 1997), the only other clear homologue of NolX is HrpF in X. campestris (Huguet and Bonas, 1997), whose function is unknown. Although the secretion of TTSS-dependent proteins other than y4xL and NolX cannot be excluded, we found no indication that y4fR and y4lO, which are homologous to YopM of Y. pestis and AvrPph3 of P. syringae (Freiberg et al., 1997), respectively, are secreted into the extracellular medium.

The TTSS locus is expressed later than most nod genes

Secretion of NolX and y4xL is NodD1 and y4xI dependent, which correlates well with transcription data showing that daidzein and NodD1 regulate the expression of many genes within the TTSS cluster (Fig. 1B; Fellay et al., 1995b). Induction of the TTSS is delayed with respect to most of the other nodulation genes however. One hour after induction with daidzein, few TTSS loci are expressed, whereas most nod genes involved in the synthesis and modification of Nod factors are actively transcribed at this time. One day later, the situation is reversed: all of the 27 ORFs within the TTSS cluster are transcribed, while most nod genes are repressed. The only exception is y4xL, which seems to be constitutively transcribed. Nevertheless, the expression of nolX is clearly flavonoid inducible (Fig. 1B). The absence of nod-box regulatory sequences upstream of the nol and rhc operons indicates that transcriptional regulation of the TTSS genes is mediated by another factor. Yet, the nod-box identified upstream of y4xI was shown to be functional, NodD1 dependent and activated by daidzein, although to a lesser extent than the nod-box found upstream of the nodABCIJnolOnoeI genes (C. Canales and X. Perret, unpublished). Along with the significant homology of y4xI to HrpG and the absence of NolX and y4xL secretion in NGRΩy4xI, these data indicate that the y4xI protein is probably the key intermediary in the regulatory cascade between flavonoids and activation of the TTSS machinery.

Symbiotic properties of the NGRΩrhcN and NGRΩy4xI polar mutants

Inoculation of classic NGR234 hosts, including some that have determinate (Pachyrhizus tuberosus and Vigna unguiculata) and others indeterminate (Leucaena leucocephala and Tephrosia vogelii ) nodules, with the rhcN and y4xI polar mutants gave diverse phenotypes (Table 2). Clearly, there is a host-specific effect, as NGRΩy4xI was virtually without effect on V. unguiculata, and NGRΩrhcN did not influence the nodulation of L. leucocephala. Also, the distinct phenotypes observed when both mutants were tested on the same plant indicate that y4xI probably regulates other functions in addition to the secretion of NolX and y4xL.

Table 2. . Symbiotic properties of polar mutants in rhcN and y4xI. Plant tests were performed in Magenta jars. For each test, the standard error of the mean and the total number of plants are shown in brackets. Nodules of all plants fixed nitrogen (Fix+). I, plants that form indeterminate nodules; D, legumes with determinate nodules.Thumbnail image of

Two plants demonstrate the extremes of responses to TTSS proteins. On P. tuberosus, NGRΩrhcN and NGRΩy4xI increased nodulation, while on T. vogelii, both mutants diminished nodule formation in comparison with NGR234. Infection of P. tuberosus with NGRΩrhcN clearly shows the deleterious effects of proteins exported via the TTSS — their absence allows vastly more nodules to form, and the plants fix proportionately more nitrogen (Fig. 3). T. vogelii, on the other hand, appears to need the secreted proteins for optimal nodulation.

Figure 3.

. Symbiotic properties of NGR234 and NGRΩrhcN on Pachyrhizus tuberosus. Shoots (top) and roots (bottom) of 6-week-old P. tuberosus plants grown in Magenta jars and inoculated with NGR234 (B) and NGRΩrhcN (C) are shown. A, non-inoculated controls. Compared with plants inoculated with NGRΩrhcN, wild-type NGR234 provoked very few nodules and an atypical root system.

Discussion

Delayed induction of the secretion pathway loci in comparison with those responsible for Nod factor synthesis suggests that the TTSS machinery is probably assembled after Nod factors have been elaborated. As Nod factors allow rhizobia to enter roots (Relićet al., 1993; 1994), it is probable that protein export begins when intimate contact between bacteria and root hairs has been established. As Nod factors alone are unable to complement infection by NodD1 mutants (B. Relić and W. J. Broughton, unpublished), we invoked a second, NodD1-dependent signal that controls the expression of root hair genes (Arsenijevic-Maksimovic et al., 1997). In this scenario, the most important role of the TTSS machinery would occur after the bacteria have entered the plant, with the secreted proteins (e.g. NolX, y4xL, etc.) being part of the second signal(s). Transcription data show that, of the TTSS genes, only nolBrhcJnolUV are expressed in bacteroids of V. unguiculata nodules (Fig. 1B), whereas in nodules of C. cajan, only nolBrhcJ and rhcN are actively transcribed. Similarly, immunological probing failed to detect SR3 and SR5 proteins of USDA257 in mature G. max and V. unguiculata nodules (Krishnan et al., 1995).

Data supporting the observation that TTSS genes are expressed at an intermediate stage in the symbiosis (i.e. after nod genes but before fix and nif genes are activated) come from the lack of differences in cytological structure between nodules induced by NGRΩrhcN and wild-type NGR234 on P. tuberosus (Fig. 4). This suggests strongly that the TTSS is a determinant of nodule initiation rather than nodule development. In other words, a complete TTSS is probably not required for nitrogen fixation, although it could be involved in the release of rhizobia from infection threads.

Figure 4.

. Electron micrographic study of P. tuberosus nodules inoculated with NGR234 (A and A′) and NGRΩrhcN (B and B′). A and B show the general organization of both infected and uninfected cortical cells, while A′ and B′ show the detail within one infected cell. Significant cytological differences were not observed between nodules formed by wild-type NGR234 or by NGRΩrhcN. Scales are given in microns. Granules of β-hydroxybutarate (PHB) accumulated in bacteroids (B) surrounded by normal peribacteroidal membranes (PBMs). N, nucleus of an infected plant cell.

It has been suggested that the bacterial invasion of plant cells triggers non-specific defence reactions and that successful pathogens overcome these defences, i.e. they are compatible (Gabriel and Rolfe, 1990). Similarly, invading symbionts must have evolved different strategies to lower host plant defences, and TTSS proteins may contribute to this phenomenon. Some plants would perceive these proteins as pathogenic and react to their absence by increased nodulation (e.g. P. tuberosus). Others, such as T. vogelii, would profit from them, while in a third group, they would have no effect (V. unguiculata).

Thus, rhizobial TTSSs play an important role in nodulation, especially of tropical legumes, where they may act as host specificity determinants. The precise biological activity of y4xL and NolX and whether the secreted proteins act at the bacterial–plant interface or inside the host cytosol remain to be elucidated however. Nonetheless, the fact that NGR234 TTSS functions in a manner similar to that of plant pathogens opens up new avenues for research, not only in the ecology of bacteria–eukaryotic interactions, but also in their control.

Experimental procedures

Molecular and microbiological techniques

All bacterial strains and plasmids are listed in Table 3. Escherichia coli recombinants were grown on Luria–Bertani medium or in Terrific broth (Sambrook et al., 1989). Strains of Rhizobium were grown in/on tryptone yeast (TY) medium (Beringer, 1974) or Rhizobium minimal medium (RMM; Broughton et al., 1986). Preparation of Rhizobium genomic DNA was carried out as described in Perret and Broughton (1998). Transcription analysis of the TTSS cluster was carried out as described in Freiberg et al. (1997).

Mutagenesis of rhcN and y4xI

NGRΩrhcN and NGRΩy4xI were constructed as follows. A 3.5 kb BamHI–EcoRV fragment of pXB110 (Perret et al., 1991) containing the rhcN gene was cloned into pBluescript KS+ (Stratagene), and the spectinomycin-resistant (SpR) interposon (Prentki and Krisch, 1984) was inserted into the internal EcoRI site of rhcN. This insert was excised using SpeI and ApaI, purified by agarose gel electrophoresis and subcloned into pJQ200SK (Quandt and Hynes, 1993). The resulting plasmid was mobilized into NGR234 by triparental mating using the helper plasmid pRK2013 (Figurski and Helinski, 1979). Marker exchange was forced by selection on RMM plates containing 5% (w/v) sucrose. A similar procedure was used to generate NGRΩy4xI by inserting the SpR interposon into the internal ApaI of y4xI, carried on a 2.4 kb PstI fragment of pXB110. Mutants were confirmed by probing Southern blots of restricted genomic DNA according to standard methods (Sambrook et al., 1989).

Purification and analysis of extracellular proteins

After induction with 10−6 M apigenin or 0.2 × 10−6 M daidzein, cultures were grown for 40 h at 27°C. Bacteria were removed by two successive centrifugations (5000 × g at 4°C for 20 min). Proteins of supernatants were precipitated with ammonium sulphate (60% final concentration) at 4°C for 2–3 h, recovered by centrifugation (8000 × g at 4°C for 30 min) and resuspended in 500 μl of 50 mM Tris-HCl (pH 7.5). The samples were desalted using Microcon-10 microconcentrators (Amicon) and recovered in a final volume of 30 μl of buffer. Protein concentrations were determined using the Bio-Rad protein assay kit (Bio-Rad Laboratories), and the samples were separated on SDS–polyacrylamide (12% w/v) gels stained with silver (Ausubel et al., 1991). N-terminal amino acid sequencing was performed on products separated by two-dimensional gel electrophoresis (O'Farrell, 1975). In the first dimension, 5–10 μg of proteins were subjected to isoelectric focusing (ampholites of pH 3–10 with increased resolution from pH 5–7), while electrophoresis in the second dimension was performed on SDS–polyacrylamide (12%) gels. Separated proteins were transferred to Millipore Immobilon PVDF membranes by electroblotting, and selected spots were microsequenced.

Plant assays

Nodulation tests were performed in Magenta jars (Lewin et al., 1990). All plants were grown at a day temperature of 26°C, a night temperature of 20°C and a light phase of 16 h. Each plant was inoculated with 109 bacteria and harvested 6–8 weeks after inoculation. Nodules were prepared for electron microscopy as described by Golinowski et al. (1987).

Acknowledgements

We wish to thank S. Relić, C. Canales and D. Gerber for their help in many aspects of this work. We are also grateful to S. Frutiger for microsequencing the N-terminus of NolX and y4xL, K. Schesser for many helpful discussions and Zairi Jaal for providing seeds of Pachyrhizus tuberosus. Financial support for this project was provided by the Fonds National Suisse de la Recherche Scientifique (grant no. 31 45921.95) and the University of Geneva.

Ancillary