Rhizobia are Gram-negative bacteria than can elicit the formation of specialized organs, called root nodules, on leguminous host plants. Upon infection of the nodules, they differentiate into nitrogen-fixing bacteroids. An elaborate signal exchange precedes the symbiotic interaction. In general, both rhizobia and host plants exhibit narrow specificity. Rhizobial factors contributing to this specificity include Nod factors and surface polysaccharides. It is becoming increasingly clear that protein secretion is important in determining the outcome of the interaction as well. This paper discusses our current understanding of the symbiotic role played by rhizobial secreted proteins, transported both by secretion systems that are of general use, such as the type I secretion system, and by specialized, host-targeting secretion systems, such as the type III, type IV and type VI secretion systems.
Rhizobia are Gram-negative bacteria that comprise the genera Allorhizobium, Azorhizobium, Bradyrhizobium, Mesorhizobium, Rhizobium and Sinorhizobium (Van Berkum & Eardly, 1998). They can trigger leguminous host plants to form root nodules, specialized organs that offer the bacteria an exclusive ecological niche in which they reduce atmospheric dinitrogen to ammonia. Ammonia is made available to the plant, which in turn provides carbon sources to the bacteria (Perret et al., 2000; Gage, 2004; Jones et al., 2007). The rhizobial symbiosis with legumes is generally a specialized affair, both partners having narrow host ranges. Exceptions include Rhizobium sp. NGR234, which nodulates over 110 genera of legumes (Pueppke & Broughton, 1999) and Phaseolus vulgaris L. or the common bean plant, which is nodulated by at least 20 species of rhizobia (Michiels et al., 1998). The outcome of the interaction is dependent on an elaborate signal exchange that continues throughout the entire symbiotic process and has been likened to matching locks and keys (Broughton et al., 2000), with only the correct combination giving rise to efficient symbiosis.
Bacteria in the rhizosphere are chemotactically attracted to host plants upon detection of compounds such as flavonoids, phenolics, sugars, dicarboxylic acids and amino acids, which are exuded by the plant roots (Brencic & Winans, 2005). Subsequently, rhizobia adhere to and colonize the root surface. Flavonoids, polycyclic aromatics released by the plant into the rhizosphere, induce expression of nodulation genes (nod, noe, nol and others). The resulting proteins synthesize and export a class of molecules called Nod factors or lipochito-oligosaccharides. They consist of a conserved backbone that is varyingly decorated with accessory groups. Only particular types and mixtures of Nod factors allow a strain to nodulate a certain legume host, thus giving rise to a first rhizobial determinant of host specificity (Spaink, 2000). Early responses to Nod factors include calcium spiking and root hair cytoskeleton modification. Curling of the root hairs occurs due to the localized presence of Nod factor molecules. Simultaneously, root cortex cells are stimulated to reinitiate mitosis, leading to the formation of nodule primordia (Gage, 2004).
Root hair curling results in the attached bacteria becoming entrapped within the deformation. Local lysis of the root hair cell wall is followed by invagination of the plant cell membrane. The rhizobia then enter this invaginated tube-like structure called the infection thread. The infection thread grows inwardly towards the base of the root hair and the dividing cells in the nodule primordium. During infection thread development, rhizobial surface polysaccharides (lipopolysaccharides, exopolysaccharides, capsular polysaccharides and cyclic glucans) interact with the host plant. Successful symbiosis is dependent on their correct composition, making them a second rhizobial determinant of host specificity (Perret et al., 2000; Gage, 2004).
The nodule primordium continues to divide and develops into a novel plant organ, the nodule. After the bacteria exit from the infection thread, they differentiate into bacteroids and are able to fix nitrogen in the microaerobic environment provided by the nodule interior.
In addition to the nonproteinaceous host specificity determinants described above (Nod factors and surface polysaccharides), a third class of rhizobial signals that affects symbiosis consists of secreted proteins. This Minireview focuses on the diverse roles they play in the symbiotic rhizobium–legume interaction.
Proteins secreted by secretion systems of general use
Cells from prokaryotes and eukaryotes alike must transport proteins across the membranes that envelop them (Economou & Dalbey, 2004). An inherent problem is the need for long macromolecules with hydrophilic and hydrophobic regions to cross a hydrophobic lipid membrane. In Gram-negative bacteria, the situation is even more complicated due to the presence of two such barriers, the inner and outer membrane (IM and OM). Nevertheless, multiple mechanisms have arisen over the course of evolution to cope with this challenge. Two main groups of secretion systems can be discerned, those that rely on the Sec machinery to cross the IM and those that do not (Papanikou et al., 2007). Substrates of the latter typically traverse both IM and OM in a single step, without the formation of stable periplasmic intermediates. This group includes type I, type III, type IV and type VI secretion systems (T1SS, T3SS, T4SS and T6SS, respectively). The chaperone/usher pathway and type II and type V secretion systems (T2SS and T5SS, respectively) make up the Sec-dependent group. While all these membrane transporters have previously been shown to mediate microbe–host interactions, not all of them do so exclusively. For example, T1SS, T2SS, T5SS and the chaperone/usher pathway have also been implicated in processes such as organelle biogenesis and nutrient acquisition (Economou & Dalbey, 2004; Gerlach & Hensel, 2007). Below, we discuss rhizobial proteins with a (proposed) symbiotic role that are transported by these general secretion systems. A schematic overview is given in Fig. 1.
The first secreted rhizobial protein for which a role in symbiosis could be shown was Rhizobium leguminosarum bv. viciae NodO (de Maagd et al., 1989b). NodO was detected in spent medium of cultures grown in the presence of flavonoids and expression was found to be NodD-dependent (de Maagd et al., 1988). Vlassak (1998) reported a host-specific symbiotic phenotype for a Rhizobium sp. BR816 nodO mutant. The strain is affected in nodulation and nitrogen fixation on P. vulgaris, but not on Leucaena leucocephala. In R. leguminosarum bv. viciae, strains that fail to produce fully decorated Nod factors display a Nod− phenotype in the absence of a functional copy of nodO (Economou et al., 1994; Walker & Downie, 2000). This suggests that NodO can complement deficiencies in Nod factor signalling. In support of this, heterologous expression of Rhizobium sp. BR816 NodO suppresses mutation of R. leguminosarum bv. trifolii nodE, Rhizobium sp. NGR234 nodS and nodU, and Rhizobium tropici nodU (van Rhijn et al., 1996; Vlassak et al., 1998). Moreover, NodO expression confers the ability to nodulate L. leucocephala to Azorhizobium caulinodans, Rhizobium etli, R. leguminosarum bv. trifolii and Sinorhizobium meliloti. Based on these results, Walker & Downie (2000) suggested three possible roles for NodO. The protein may facilitate Nod factor uptake by the host, amplify the perceived Nod factor signal or bypass the host's Nod factor receptor altogether. Observations made by Sutton (1994) favour the amplification hypothesis. NodO was found to insert into liposomes and form cation-selective channels in lipid bilayers, and could therefore enhance the calcium spiking that is observed in root hairs upon Nod factor binding.
Because of the absence of a recognizable amino-terminal secretion signal peptide, a Sec-independent secretion pathway was proposed for NodO (de Maagd et al., 1989b; Economou et al., 1990). Subsequently, the protein was shown to be secreted by T1SS of Escherichia coli, Erwinia chrysanthemi and various rhizobia (Scheu et al., 1992). T1SS are a class of ATP-binding cassette transporters that recognize substrates with a carboxy-terminal, noncleavable signal sequence. Through analysis of in-frame deletions, a 24 amino acid carboxy-terminal secretion signal was identified in NodO (Sutton et al., 1996). Almost a decade after its secretion was first described, Finnie (1997) discovered the genes responsible for NodO transport: prsD and prsE. Unusual for a T1SS, these genes are unlinked to nodO. In addition, the PrsDE secretion system transports other substrates as well (Finnie et al., 1998; Russo et al., 2006; Krehenbrink & Downie, 2008). As nodO is only found in R. leguminosarum bv. viciae and Rhizobium sp. BR816 (de Maagd et al., 1989a; van Rhijn et al., 1996), and as the prsDE genes are more widely conserved among rhizobia (York & Walker, 1997), it seems likely that NodO exploits secretion through PrsDE rather than being its original cognate substrate.
Proteins affecting surface polysaccharide composition
Finnie (1997) noted that a R. leguminosarum bv. viciae prsDE mutant has a more severe symbiotic phenotype than a nodO mutant. The former displays alterations in colony morphology and elicits ineffective nodules upon inoculation of pea plants, suggesting a change in exopolysaccharide composition. Analysis of the extracellular protein profile showed that three additional proteins are secreted by PrsDE. At least one of these is required for nitrogen fixation. Subsequently, PlyA and PlyB were identified as polysaccharide lyases secreted by PrsDE (Finnie et al., 1998). Mutation of plyB affects exopolysaccharide processing. However, no role in symbiosis could be attributed. Recently, PlyB was shown to influence R. leguminosarum biofilm formation in vitro, although its involvement is probably indirect and additional PrsDE substrates are likely to be important as well (Russo et al., 2006). In R. etli, a plyB-like gene was shown to be important for swarming (Braeken et al., 2008). The symbiotic properties of the mutant strain appeared to be normal.
Sinorhizobium meliloti produces two types of exopolysaccharide: succinoglycan (exopolysaccharide I) and galactoglucan (exopolysaccharide II) (Glazebrook & Walker, 1989). Production of either type is sufficient for symbiosis with Medicago sativa. Exopolysaccharide I composition is affected by ExoK and ExsH (York & Walker, 1997, 1998). Both proteins were shown to be secreted and to specifically depolymerize nascent succinoglycan chains. ExsH is secreted by a T1SS. No role in symbiosis was yet proposed for either of the proteins, because single and double mutants did not have an altered symbiotic phenotype. ExpE1 is essential for exopolysaccharide II biosynthesis or secretion, as cultures of an expE1 mutant were devoid of exopolysaccharide II (Moreira et al., 2000). Secretion of ExpE1 is mediated by the ExpD1D2 T1SS.
Based on a phenotypic analysis of R. leguminosarum prsD, prsE and plyA plyB mutants, additional substrates of the PrsDE T1SS are thought to influence biofilm formation. Russo (2006) detected nine secreted proteins in total, three of which were previously determined to be NodO, PlyA and PlyB. Three newly identified proteins share similarity with Rhizobium-adhering proteins or Raps (Ausmees et al., 2001) and were named RapA1, RapA2 and RapC. It is as yet unclear what (if any) their role is in biofilm formation and/or symbiosis. RapA1 was previously characterized as a protein with features reminiscent of rhicadhesin and bacterial lectins (Ausmees et al., 2001). Both RapA1 and rhicadhesin bind calcium and agglutinate cells. They differ in size and occurrence, however, as RapA1 is larger (24 vs. 14 kDa) and rap genes are confined to R. leguminosarum and R. etli, whereas rhicadhesin is believed to be common to all rhizobia (Smit et al., 1992; Ausmees et al., 2001). Both RapA1 and Bradyrhizobium japonicum lectin BJ38 localize to a single cell pole after secretion (Loh et al., 1993; Ausmees et al., 2001). Purified BJ38 was shown to specifically bind certain carbohydrates (Ho et al., 1990). Mutants deficient in cell adhesion expressed less or no detectable BJ38 and failed to nodulate soybean roots (Ho et al., 1990, 1994). In a general model for rhizobial attachment to plant roots, rhicadhesin is proposed to function nonspecifically while lectin mediates specific binding (Rodríguez-Navarro et al., 2007). Ideally, the details of this model should be tested using genetically defined mutant strains. However, the unavailability of sequence information concerning both rhicadhesin and lectin BJ38 prevents this.
Other PrsDE-secreted proteins
Very recently, six additional PrsDE-secreted proteins were identified in R. leguminosarum bv. viciae strain 3841 by Krehenbrink & Downie (2008). Based on similarity searches, these could function as a metalloprotease, a glycosyl hydrolase, cadherins and a nucleoside diphosphate kinase. Strikingly, the prsD mutant did not exhibit a symbiotic phenotype, probably due to the use of a different wild-type strain in this study compared with previous reports.
Calcium is known to play an important role in the rhizobium–legume symbiosis (Verhaert et al., 2005). Correspondingly, many of the proteins discussed so far in this section have been found to bind or otherwise interact with Ca2+. This is also the case for CasA or calsymin, a secreted R. etli protein (Xi et al., 2000). Calsymin is a calmodulin-like, Ca2+-binding protein that is essential for normal bacteroid development. It is secreted without cleavage of an amino-terminal signal peptide, although the genes responsible for transport are unknown. As is the case for calmodulin, Ca2+-binding by calsymin depends on the presence of specific domains called EF-hands and induces a conformational change in the protein (Verhaert, 2005). Its precise role in symbiosis remains to be elucidated. Homologues of calsymin are also present in the genome of R. leguminosarum (Young et al., 2007).
Proteins secreted by specialized secretion systems
The secretion pathways that have been discussed thus far are important for processes unrelated to host interactions as well. In contrast, several secretion systems seem to have specialized in mediating such interactions, with the ability to translocate effector proteins into the host cell cytoplasm as a defining feature. These systems include the T3SS, T4SS and T6SS. For a schematic overview, see Fig. 1.
Type III secretion system and effector proteins
T3SS are complex macromolecular structures that span not only the IM and OM, but the host cellular membrane as well. They allow direct translocation of secretion substrates into the host cell cytoplasm. T3SS are structurally related to flagella. Few roles outside of bacteria–host interactions have been described for nonflagellar T3SS. The secretion signal is poorly defined and located in the amino-terminus of secretion substrates (Ghosh, 2004; He et al., 2004).
The first definite evidence of the involvement of a T3SS in the rhizobium–legume symbiosis was provided by Viprey (1998). Taking advantage of the published sequence of the Rhizobium sp. NGR234 symbiotic plasmid, a strain lacking a functional T3SS was constructed through site-directed mutagenesis. The mutation affected symbiosis in a host-specific way. On some hosts, the presence of the T3SS does not influence nodulation (Vigna unguiculata, L. leucocephala), on others the absence of a functional T3SS has a beneficial effect (Pachyrhizus tuberosus), while on a third group of hosts the T3SS is required for effective symbiosis (Tephrosia vogelii). T3SS genes were subsequently identified in B. japonicum USDA110, Sinorhizobium fredii strains HH103 and USDA257, Mesorhizobium loti MAFF303099 and R. etli CNPAF512 (Kaneko et al., 2000; Krause et al., 2002; Krishnan et al., 2003; Hubber et al., 2004; de Lyra et al., 2006; M. Fauvart, unpublished data). In each case, the T3SS affects symbiosis in a host-specific manner. Expression of rhizobial T3SS genes is induced by flavonoids and depends on NodD as well as on a more specific transcriptional regulator, TtsI (Marie et al., 2004). TtsI is believed to bind to conserved promoter elements termed tts boxes (Krause et al., 2002). These are also found in promoter regions of genes unrelated to T3SS, such as those involved in the biosynthesis of rhamnose-rich polysaccharides (Marie et al., 2004). This led to the discovery of a complex interplay between the T3SS and surface polysaccharides in the molecular dialogue of the rhizobium–host interaction (Broughton et al., 2006).
T3SS substrates come in two flavours: effectors and helper proteins. Effectors are translocated into eukaryotic host cells, while helper proteins assist in the translocation process. Proteins secreted through rhizobial T3SS are called nodulation outer proteins or Nops. Three helper proteins have been identified so far: NopA, NopB and NopX (Lorio et al., 2004; Deakin et al., 2005; Saad et al., 2005, 2008). NopA and NopB constitute major and minor components of the T3SS pilus, respectively. NopX is predicted to be involved in the formation of a pore-like translocon structure in the host cell membrane, thereby stimulating effector translocation into the host cell cytosol. Although Nop translocation has yet to be confirmed experimentally, three T3SS substrates stand out as candidate effectors: NopL, NopP and NopT. These proteins affect symbiosis in a host-specific manner without influencing the secretion of other Nops (Marie et al., 2003; Ausmees et al., 2004; Skorpil et al., 2005; Dai et al., 2008). Interestingly, NopL and NopP are phosphorylated by plant kinases (Bartsev et al., 2003; Skorpil et al., 2005). Although the biological relevance of this observation is unknown, it could point to a function in modulating host signalling pathways. Specifically for NopL, a role in the downregulation of host plant defences is proposed (Bartsev et al., 2004). NopT is a functional cysteine protease of the YopT family with a predicted myristoylation site (Dai et al., 2008). It is as yet unclear whether these properties are essential for its function in symbiosis.
Other candidate effectors include the S. fredii T3SS-secreted proteins NopD and NopM, which are similar to T3SS effector proteins XopD of Xanthomomas spp. and YopM of Yersinia spp., respectively (Rodrigues et al., 2007). XopD and YopM are targeted to host cell nuclei where they are thought to interfere with the regulation of host proteins during infection (Skrzypek et al., 1998; Hotson et al., 2003). The cysteine protease XopD achieves this through hydrolysis of small ubiquitin-like modifier-conjugated proteins (Hotson et al., 2003), while YopM probably acts as a scaffold for recruiting and stimulating other proteins (McDonald et al., 2003). Secretion of NopC, a protein of unknown function, and GunA2, an endoglucanase, was also shown to be T3SS-dependent (Deakin et al., 2005; Süβ, 2006). In addition, Rhizobium sp. NGR234 y4lO is predicted to encode a protein homologous to the Yersinia T3SS effector YopJ, a cysteine protease that inhibits mitogen-activated protein kinase signalling pathways (Freiberg et al., 1997; Orth, 2002). Furthermore, R. etli possesses hrpW and yha00035, genes expected to encode proteins homologous to Pseudomonas syringae T3SS substrates HrpW and HopG1 (González et al., 2006). HrpW was recently shown to promote T3SS effector translocation into plant leaves, making it a helper protein rather than a true effector (Kvitko et al., 2007). HopG1 suppresses basal resistance in tobacco (Oh & Collmer, 2005). These homology-based findings significantly expand the pool of potential rhizobial T3SS effectors.
Type IV secretion system
T4SS are functionally similar to T3SS, although the structural proteins that make up both systems do not share sequence similarity. Like T3SS, T4SS can translocate proteins directly into host cells, are related to a macromolecular structure of general use (in casu conjugation machineries) and are primarily implicated in bacteria–host interactions. Transport across the IM is independent of the Sec pathway, although Sec-dependent IM transport has been reported. T4SS substrates are characterized by a carboxy-terminal secretion signal, while an additional motif might be involved as well (Christie et al., 2005; Vergunst et al., 2005).
Comparative sequence analysis of the symbiosis islands of M. loti strains MAFF303099 and R7A showed that the latter encodes a T4SS whereas the former encodes a T3SS. Strikingly, the genes flanking the corresponding region are well conserved between the two strains (Sullivan et al., 2002). T3SS and T4SS mutants displayed similar, host-specific phenotypes, suggesting that both secretion systems are, at least in part, functionally interchangeable (Hubber et al., 2004). Two candidate effector proteins were identified. Msi059 shares similarity with VirF-related proteins from Agrobacterium spp., while Msi061, like NopD, is similar to the Xanthomonas T3SS effector XopD. Recently, expression of the M. loti T4SS genes was shown to be NodD- and VirA-dependent (Hubber et al., 2007). This is in agreement with the observed effect of the T4SS on symbiosis.
Type VI secretion system
T6SS have been discovered only recently (Pukatzki et al., 2006) and much of the information available is therefore still of a speculative nature. They form a third distinct pathway through which proteins are believed to traverse directly into host cells and affect bacteria–host interactions. No recognizable amino-terminal secretion signal was identified in secreted proteins (Bingle et al., 2008). Before its formal discovery and naming, Bladergroen et al. (2003) identified an R. leguminosarum T6SS locus (named imp, impaired in nodulation) as a determinant of host specificity on Pisum sativum. In vitro secretion was found to be temperature-dependent and a single secreted protein was identified as an RbsB-like protein. RbsB is an E. coli periplasmic ribose-binding protein, encoded by a locus that is involved in ribose transport into the cell (Park et al., 1999). Its relationship with the T6SS and its possible role in symbiosis are unclear.
Based on sequence analysis, Bingle (2008) recently reported the presence of T6SS in at least three genera of rhizobia. In addition to R. leguminosarum, conserved T6SS genes are also present in the sequenced genomes of B. japonicum and M. loti. Following the recent surge of interest in T6SS of pathogenic bacteria, this presents another opportunity to compare the functional conservation of secretion pathways in pathogenesis and symbiosis.
Conclusions and perspectives
Effective, nitrogen-fixing rhizobium–legume symbiosis requires an intricate molecular dialogue between the two interaction partners before and during invasion by the microsymbiont (Broughton et al., 2000; Perret et al., 2000). Host specificity is determined by several factors. From the bacterial side, the main signalling molecules are Nod factors, surface polysaccharides and secreted proteins. The importance of the latter is becoming increasingly clear, and judging by developments in the area of host–pathogen interactions, secretion pathways specialized in the direct delivery of effector proteins into the eukaryotic host's cell cytoplasm offer the most potential for exerting control over host responses. This is not only the case for bacterial pathogens, which appear to favour T3SS, but also for eukaryotic pathogens such as oomycetes and fungi, which transport proteins into host cells by means of a specialized structure called the haustorium (Ellis et al., 2007). It should therefore come as no surprise that rhizobia use the exact same secretion mechanisms as their pathogenic counterparts in trying to persuade prospective hosts to allow rhizobial invasion.
Sequence analysis suggests that pathogens and rhizobia share not only secretion pathways, but some effector proteins as well. This offers the enticing prospect of being able to extrapolate findings made in one domain to the other.
In spite of the many relevant lessons that may be gleaned from the analysis of pathogenic organisms, it is important to remember that there is still a long way to go. It is telling that formal evidence of the translocation of rhizobial effector proteins into leguminous root cells through any secretion system is still lacking, let alone the identification of the molecular targets of these effectors. Such fundamental issues need to be resolved before starting to extrapolate and speculate. From there, the daunting challenges that abound for scientists investigating the symbiotic role of rhizobial protein secretion can be taken one at a time.
The authors would like to thank the anonymous reviewers for their helpful suggestions. M.F. is a recipient of a fellowship from the Institute for the Promotion of Innovation through Science and Technology in Flanders (IWT-Vlaanderen). This work was supported by grants from the Research Council of the K.U. Leuven (GOA/2003/09) and from the Fund for Scientific Research-Flanders (G.0108.01 and G.0287.04).