Rhizobium Symbiosis: Nod Factors in Perspective - Plant Cell

The Plant Cell, Vol. 8, 1885-1898, October 1996 O 1996 American Society of Plant Physiologists

Rhizobium Symbiosis: Nod Factors in Perspective Sharon R. Long Howard Hughes Medical Institute, Department of Biological Sciences, Stanford University, Stanford, California 94305

INTRODUCTION

Rhizobium and its allies (Azorhizobium, Bradyrhizobium, and Sinorhizobium) are Gram-negative bacteria that cause the development of root (and sometimes stem) nodules on plant hosts, which the bacteria inhabit as nitrogen-fixing endosymbionts. The early stages of this process, including gene expression in the bacterium and cell growth, division, and differentiation in the host, are mediated by signal exchange between the eukaryotic host and the prokaryotic symbiont (Figure l A , left). The plant produces a signal, usually aflavonoid, that induces gene expression in the bacterium;the bacterium subsequently synthesizes a signal that triggers early nodule development on the plant. The developmental time line for nodulation has been described in several reviews and essays (Sprent, 1989; Truchet et al., 1989; Brewin, 1991; Brewin et al., 1992; Hirsch, 1992; Kijne et al., 1992; Ridge, 1992; Vijn et al., 1993) and is only considered briefly here. Nodules can take on several patterns during development, the form of the nodule being determined by the plant, not the bacterium. One major form is the indeterminate (also called meristematic or cylindrical) type, which develops on alfalfa, clover, and pea roots. A second major type is the spherical or determinate nodule, which is formed by soybean, Phaseolus, and Lofus.A comparison of these symbiotic nodules with those of nonlegumes is presented elsewhere in this issue (see Pawlowski and Bisseling, 1996). The twin hallmarks of early nodulation are its developmental complexity and its specificity. The developmental process in the plant involves architectural changes at the cell and organ levels (for example, root hair morphogenesis, cortical cell enlargement, and vascular patterning) as well as interna1cellular differentiation that includes cytoplasmic activation, cell division, and new gene expression. Structural and developmental studies of nodule formation remain an important part of the overall Rhizobium research picture. Indeed, new views of infection thread formation, cell wall modifications, and intracellular rearrangements in root hairs and elsewhere have recently appeared (Kijne et al., 1992; van Brussel et al., 1992; van Spronsen et al., 1994; DeBoer and Djordjevic, 1995; Ridge, 1995).The exciting tools of video microscopy and image analysis are making it possible to obtain dynamic views of early plant reactionsto Rhizobiumsignals (Allen et al., 1994; Sanchez et al., 1996). Thus, in addition to its inherent interest as a model for understanding plant-microbe interactions, the specificity and timing of early nodulation events make the Rhizo-

bium-plant symbiosis an attractive model system for general plant cell biology studies. The specificity of nodulation is likewise remarkable:with one known exception, the Rhizobiumnodulation habit is restricted to a single plant taxon, the Fabaceae,or legume family. Within this family, individual species, strains, or biovars of bacteria nodulate a restricted set of host plants that are usually but not always related. The signal model (Figure 1) provides an explanation for the species-leve1pattern of host specificity. But we cannot yet answer the larger mechanistic and evolutionary question: Why only legumes? Because a short review cannot catalog complete lists of referentes, even recent ones, the focus of this article is to put selected papers in context: Why is it important for plant biologists to be concerned with bacterial genes, their regulation, and activities? Where do the questions lie in the study of the plant response? What new genetic and cellular methods are needed for their resolution?

THE NOD FACTOR SIGNAL EXCHANGE MODEL

Over the years, many molecules have been proposed as candidate host-specific signals that govern the Rhizobium-plant symbiosis by mediating bacterial invasion. Those proposed include phytohormones, extracellular polysaccharides, and hydrolytic enzymes (among others). Severa1independent lines of evidence, with a strong genetic base, have now convinced most researchers that a nove1 set of molecules, the lipooligosaccharide Nod factors, is at the heart of the symbiosis (de Bruijn and Downie, 1991; DBnariB et al., 1992; Fisher and Long, 1992; Higashi, 1993; Downie, 1994; Spaink, 1995). The genetic analysis was grounded in the identification of Rhizobium nodulation (nod)genes, which define the central functions required for plant invasion and host recognition. The nodgenes can be functionally divided into common nod genes, which are widely conserved in Rhizobium and its allies, and hostspecific nod genes, which are required for nodulation of certain plants but not others. The link between the nod genes and the biosynthesis and delivery of the lipo-oligosaccharidesignal molecules was confirmed in a number of ways. First, nodgene regulation accounts for the production of Nod factors as well as for the symbiotic

1886

The Plant Cell

B Alfalfa

Phaseolus

Lotus (some genotypes)

•v.

fl leguminosarum vlciee

Rlotl /_

Retll \

Figure 1. Schematic of Symbiotic Signal Exchange and Host Specificity. (A) The Nod factor signal exchange model. Successful signal exchange between alfalfa and its compatible symbiont, Rhizobium meliloti, is shown on the left. Alfalfa produces a spectrum of inducers (yellow) that cause transcription of nod genes in R meliloti. The products of the nod genes direct the synthesis of a responding signal, a lipo-oligosaccharide Nod factor, that carries host-specific modifications (red and green symbols). These Nod factors successfully induce host responses. Host specificity based on the bacterial morphogen is shown on the right. The same inducers from alfalfa trigger the expression of a subset of nod genes in a bacterium, R. leguminosarum bv w'c/ae, for which alfalfa is not a host. The resulting R. leguminosarum Nod factor carries different host-specific modifications (red symbol), which are tailored to be active not on alfalfa but on other plants such as We/a (not shown in the figure). Consequently, the R. I. w'c/ae signal causes no response in alfalfa. (B) Host specificity based on bacterial gene expression. This model is inferred from the data of Cardenas et al. (1995) and Lopez-Lara et al. (1995). R. loti, a symbiont of Lotus, and R. etli, a symbiont of Phaseolus, each make a spectrum of Nod factors, of which some have the same structure (bottom; color-coding shows side groups). It is likely that host-specific plant inducers (red vs. green symbols) restrict the host range of bacteria. However, if the expression of the nod genes in each bacterium is driven by means of an inducer-independent NodD variant protein, the bacteria extend their host range to include new plant species or genetic lines. The arrows indicate effective signals on the target organism; the dashed lines with a bar indicate compounds not effective as a signal on the target organism. phenotypes of live bacteria; second, the structure of the lipooligosaccharide molecules correlates with the nod genotype of the bacterium; third, at least some purified nod gene products carry out in vitro synthesis of lipo-oligosaccharide structures; and fourth, purified lipo-oligosaccharide molecules provoke specific Nod-like effects on host plants. Each of these lines of evidence is discussed below.

Regulation of nod Genes Most of the Rhizobium nodulation genes are transcriptionally silent when the bacteria are grown in culture. Their expression is controlled by a transcriptional activator, NodD, which acts together with inducers from the plant, and also by other bacterial regulators, such as NoIR (reviewed in Kondorosi, 1992; Schlaman et al., 1992). NodD acts on promoter elements that include the conserved "nod box" sequence, which is known to be a site of NodD binding and consequent DNA bending (Fisher and Long, 1993). The molecular basis for NodD activation, although outside the scope of this review, is a subject that continues to present several puzzles, not the least of which is the precise role of the plant inducer in triggering nod gene activation. Initial characterization of nodD genes showed that the nodO gene products of various bacterial species appeared to activate nod genes differentially in response to the specific mixtures of inducers derived from the roots of each host plant.

These inducers include luteolin, methoxychalcone, naringenin, and daidzein as well as nonflavonoid inducers such as trigonelline and stachydrine (Phillips et al., 1994). The spectrum of secreted flavonoids may change with conditions, which may have ecological and developmental consequences for the free-living rhizobia (Schlaman et al., 1992; Lawson et al., 1995; see Handelsman and Stabb, 1996, in this issue, for a review of plant-microbe interactions in the rhizosphere). These diverse responses at the level of nod gene induction are mediated in part through the action of specific members of nodD multigene families (Kondorosi, 1992). However, recent studies have provided clues that nod gene regulation involves additional gene regulators and developmental cues, which may have differential effects on Nod factor synthesis, perhaps depending on the host plant or on the subset of nod genes that each induces. For example, Demont et al. (1994) found that Rhizobium meliloti NodDS controls the production of variant acyl groups (18- to 26-carbon A/-acyl groups with omega-1-OH modifications) that were present in Nod factor preparations. Furthermore, in R. meliloti strain 41, Cren et al. (1995) observed that the represser NoIR has different effects on the expression of the common nod gene operon and on other nod box operons. This suggests that the synthesis of the Nod factor core and the modifying side groups can be differentially regulated. Additional regulators seem to operate in Bradyrhizobium japonicum. For example, nodW is a novel nod gene regulator that resembles a two-component response regulator; its ex-

Rhizobium Nod Factors

pression is more important for successful colonization of some plant hosts than for others and, in its absence, the 6.japonicum nod genes are not fully expressed (Stacey et al., 1994). Thus, the NodW circuit adds an additional layer of regulation to the NodD-inducer circuit for nod gene control. The symbiosis gene cluster in B. japonicum also includes a negative nod regulator, nolA, and at least one gene required for signal synthesis, nodZ, that is not controlled by NodD (Dockendorff et al., 1994; Sanjuan et al., 1994; Stacey et al., 1994). Thus, there is evidence from at least two different symbiont-host pairs that nodgene regulation can be differentially tuned in response to specific signals. It is important to determine whether these additional levels of regulation correlate with developmentallysensitive steps in signaling, with extended host range, or with other decision points in the symbiosis. The elucidation of these various circuits for control of nod genes provides not only insight into the regulation of the symbiosis but also a series of useful tools for engineering the expression of nod genes in experimentally manipulable settings. In turn, the elucidation of the structures of the various bacterial Nod factors (see below) now allows an evaluation of how both plant and bacterial signals contribute to specificity. For example, one recent study used combined analyses of Nod factor structure and nod gene regulation to add a new iteration to our understanding of host range (Figure lB). Cardenas et al. (1995) showed that among the Nod factors produced by Rhizobium etli were molecules strikingly similar to some produced by Rhizobium lofi (Lopez-Lara et al., 1995), yet the two bacterial species were not capable of cross-inoculating each other's host plants (Phaseolus and Lotus, respectively). Why? The answer appears to lie in the initial plant induction of the Rhizobium nod genes. If R. etli and R. loti were engineered to produce their Nod factors constitutively, that is, without dependente on plant inducers, cross-inoculation could be achieved. It can be inferred from these observations that the initial signal from the plant root is only capable of inducing Nod factor synthesis in the appropriate bacterial symbiont (Cardenas et al., 1995). Thus, colonization specificity is in some cases directed by the plant signal.

Nod Genes and Nod Factors Nod Factor Structures The functions of the Rhizobium nod genes can be studied in free living cells, if their expression is induced either by providing transcriptional inducers such as flavonoids or by driving the expression of constitutive forms of NodD. This strategy has revealed that nodgenes are required for the synthesis of Nod factors. The structures of many Nod factors have been determined. They can be most generally described as lipooligosaccharides with an oligomeric backbone of B-1P-linked N-acetylglucosamine (GlcNAc) residues that carry modifications at the reducing and nonreducing end residues (Figure 2). Nod factors are also referred to as chito-lipo-oligosac-

1887

charides and as lipo-chito-oligosaccharides. The species of bacteria for which Nod factor structures have been determined and their corresponding host plants are shown in Table 1. Severa1caveats should be mentioned here. First, most rhizobia produce mixtures of lipo-oligosaccharide molecules. This is particularly true for broad host range symbionts, such as strain NGR 234, in which the production of diverse Nod factors is controlled by a number of nod loci that are dispersed within asymbiosis plasmid regulation region (Price et al., 1992; Fellay et al., 1995). However, even those rhizobia with a more restricted host range can produce a variety of Nod factor structures (Table 1). A second important point is that structural analyses a r e K t complete for all Nod factor molecules. In some cases, for example, the initial assignment of an adduct to a particular residuecan be made (usually through mass spectrometricanalysis), whereas an exact positional assignment awaits further derivatization and spectroscopic analysis (see Figure 2, legend). The chemistry of Nod factor separations continues to be refined and debated. lnitial thin layer chromatography (TLC) methods used to define Nod factors may not resolve all chemical species (Spaink et al., 1995), but new methods are emerging (Price and Carlson, 1995). Finally, most factors studied to date were isolated by using protocols developed for the R. meliloti Nod factors (Lerouge et al., 1990), and it is possible that important signal molecules with very different properties may have escaped identification. Genotype Effects The most favored working model holds that the universal features of Nod factors-the oligosaccharide backbone, and the N-acyl bonds-are constructed by the enzymes encoded in the common nodulation genes, nodA, nodB, and nodC. The side groups that confer specificity on the different bacterial signals are determined by host-specific nod genes (Carlson et al., 1994; see Table 2). However, not all nod genes encode enzymes. For example, the NodO protein of R. leguminosarum bv viciae is a calcium binding hemolysin-like protein. NodO, which creates ion channels in membranes, appears to be able to compensate for suboptimal acyl structures in Nod factors (Downie and Surin, 1990; Sutton et al., 1994). A short review precludes a complete catalog of the nod genes and descriptions of their functions; instead, this section focuses on a few recent highlights and poses the ongoing questions. Two lines of evidence, structural correlation and direct enzymatic assay, connect nod genes to lipo-oligosaccharideNod factor synthesis. The first approach has shown that the actual structure of the Nod factor varies according to nod genotype. This has been demonstrated for a number of features, including the 6-O-sulfate in R. meliloti and R. etli Nod factors, the methylfucose addition in 6.japonicum, and the N-methyl and carbamoyl modifications in Azorhizobium caulinodans and RhizobiumNGR 234 (Figure 2; reviewedin Carlson et al., 1993, 1994; see also Geelen et al., 1995; Jabbourit et al., 1995; FolchMallol et al., 1996):'

1888

The Plant Cell

Sulfate Fucose Methylfucose Sulfo-methyltucose Acetyl-methylfucose D-Arablnose

-

\

O

unsaturation)

Non-reducing end

Reducing end

Figure 2. Generic Structure for Nodulation Factors

The oligomeric backbone of GlcNAc can vary in length from three to six total residues. The backbone is drawn here, by convention, with the reducingend at the right and nonreducingend at the left. Substituentshave been defined on the basis of mass spectroscopy,gas chromatography of derivatized fragments, and NMR. For simplicity, only the 3-Osubstitutionon the nonreducing end residue is shown (*), but either the three or the four positionmay be carryingthe carbamoyl substituent. Nod factors are named by a convention, includingthe initial letters of the species name, the number of GlcNAc residues in the backbone,the length and unsaturationnumber of the acyl chain, and the other substituents(Dénarié et al., 1992). A four-residueR. melilotiNod factor, bearing a 16-carbonfatty acyl moeity with two unsaturations,and a sulfate, is thus designated NodRmlV(C16:2,S). The positions of the unsaturationsare sometimes indicated by standard lipid nomenclature conventions.Methylfucose,2-0methylfucose; acetyl-methylfucose,4-Oacety1, 2-O-methylfucose;sulfo-methylfucose,3-0-sulfo-2-0-methylfucose.

There is some uncertainty concerning the enzymatic synthesis of the Nod factor fatty acyl moieties (Figure 2). Demont et al. (1993) and Spaink et al. (1991) have reported that n o d f and no& genotypes determine the spectrum of acyl structures found in the secreted Nod factors of R. melilotiand R. 1. viciae. Becausethe nodf genotype also determines host range, these data imply a major role for the acyl group in host recognition (reviewed in Spaink, 1995). NodF is homologous to acyl carrier proteins, and NodE is homologousto condensing enzymes (Carlson et al., 1994). One possibility, then, is that these proteins form an apparatus for the synthesis of specialized acyl structures, which are then transferred by the nodA-encoded acyltransferase to the GlcNAc backbone (John et al., 1993; Atkinson et al., 1994; Rohrig et al., 1994). However, reports on the role of nodE in R. leguminosarum bv trifolii are less clear. It is known that nodE is also important for host range in this biovar. Indeed, Bloemberg et al. (1995) and Spaink et al. (1995) found that the TLC and HPLC profiles of Nod factors were different in Rhizobium strains expressing the nodE gene of R. 1. viciae or of R. 1. trifolii. However, these

TLC and HPLC analyses were not capable of distinguishing between Nod factor profiles of no&+ and nodE- strains of R. 1. trifolii. Further spectroscopic analyses did reveal subtle differences between Nod factors from nodf+ and nodfstrains that were ascribed to changes in the degree of saturation of the fatty acyl moieties. A parallel study by Philip-Hollingsworth et al. (1995) also compared Nod factors produced by wild-type and nodf- mutant R. 1. trifolii. However, these authors reported that there were no nodE-relatedeffects on lipo-oligosaccharide structures. One difference between these two studies that may help to reconcile the results is the experimental strategy used to drive the expression of the nodE genes. Less abundant Nod factor species were more likely to be observed in the experiments reported by Spaink et al. (1995) in which a multicopy nodD gene construct was used. This contrasts with the more physiological single-nodD construct used by PhilipHollingsworth et al. (1995). What both studies do show, however, is that multiple molecules are present in the Nod factor mix and that the host-specific nodE-determined factors are at


best a small proportion of the total. We are not sure what the functions (if any) of the remaining components in the mixture may be. Some of the puzzles and inconsistencies described above may relate to the effects of nod gene products on other products of biosynthetic pathways in the bacterial cell. For example, Geiger et al. (1994) found nod5dependent lipids in the phospholipid pool. Furthermore, genes encoding enzymes that direct glucosamine synthesis and sulfate activation exist in the

Table 1. Bacterial-Host Systems with Defined Nod Factors Bacterial Species

Host Plants

Referencesa,b

Rhizobium meliloti Alfalfa Medicago truncatula Melilotus albus R. leguminosarum Pea (Pisum) bv viciae Vetch (Vicia) R. leguminosarum Clover (Trifolium) bv trifoliic

R. etli

Phaseolus

R. tropici

Broad host range: Phaseolus, Leucaena, Medicago Macroptilium Broad host range; 18 genera, including Vigna, Macroptilium, Parasponiad Lotus

Lerouge et al. (1990) Schultze et al. (1992) Spaink et al. (1991) Bloemberg et al. (1995) Philip-Hollingsworth et al. (1995) Spaink et al. (1995) Cardenas et al. (1995) Poupot et al. (1993) Folch-Mallol et al. (1996)

I

R. NGR 234

R. loti

R. fredii

Bradyrhizobium japonicume Azorhizobium caulinodans

Soybean (Glycine max) Glycine soja Soybean (G. max) G. soja Sesbania

Price et al. (1992)

Lopez-Lara et al (1995) Bec-Ferté et al. (1994) Sanjuan et al. (1992) Carlson et al. (1993) Mergaert et al. (1993)

acomprehensive reviews by Dénarié et al. (1992), Dénari6 and Cullimore (1993), Carlson et al. (1994), Relic et al. (1994), Schultze et al. (1994), Spaink and Lugtenberg (1994), and Spaink (1995) provide further references to the original literature and to other reviews. b Some Nod factors are fully defined, with all positions of adducts determined; others have approximate structures. See the original literature for details on each structure. c Differences concerning the identity of Nod factor structures and the interpretation of spectra have been reported. See the primary literature for details. d Parasponia is the only nonlegume host plant known to form symbiotic nodules with Rhizobium. e The type I and type II strains of 13. japonicum have been renamed B. japonicum and 8. elhanii, respectively.

1889

Table 2. A Brief Guide to Nodulation Genes Gene Names

Occurrencea

Proposed Roleb

nodABC

Common

Synthesis of Nod factor backbone; deacetylation; acylation Export of signals Nod factor export Synthesis of specialized acyl moiety

nodlJ nodT nodFE

Common? Varies Present in all species but have effect on host range Most other nod Varies genes

Synthesis of precursors; modification of backbone on reducing or nonreducing residues

Common nod genes are universally present in Rhizobium, Bradyrhizobium, and Azorhizobium. In at least some cases, they function across species without affecting the host range of the bacteria. Subtle host range effects have been reported in other cases. Suggested roles have been determined in some cases by direct enzymatic assay; in others, by inference from the nature of the secreted factors; and in some cases, by sequence similarity with proteins of known function. For details, see text and Carlson et al. (1994), Schultze et al. (1994), Rivilla et al. (1995), and Spaink (1995). a

nodcluster. Homologs of these genes are also found elsewhere in the bacterial genome (Carlson et al., 1994). These observations suggest that in addition to the primary Nod factors, symbiosis requires other bacterial molecules (Hirsch, 1992). It seems important, then, to b e alert not only for the primary Nod factor signals but also for bacterial molecules that modulate the plant response or act in some other way to tune the sensitivity of the plant to the bacterial signal. Ultimately, to generate a complete picture of how the bacterium signals its plant partner, we need to know how the synthesis of highly specific nodulation factors and possible secondary signals may b e coordinated with the synthesis of housekeeping metabolites such as amino acids, surface structures, and membrane components.

Nod Factor Enzymes: Answering Questions and Providing Experimental Tools Resolution of the Nod factor biosynthetic pathway and confirmation of Nod factor structures will be achieved through further enzymatic studies. Some important steps, such as sulfation, 6-O-acetylation,2-N-methylation,and 2-N-deacetylation, have now been demonstrated in vitro (reviewed in Carlson et al. 1994; Spaink, 1995; see also Bourdineaud et al., 1995; Ehrhardt et al., 1995; Geelen et ai., 1995; Mergaert et al., 1995; and Schultze et al., 1995). Moreover, the NodCgene product, which is hypothesized to be the polymerase for the backbone, may be located in the inner bacterial membrane (Barney and Downie, 1993; Carlson et al., 1994). This protein has been shown to catalyze the synthesis of GlcNAc oligomers in vitro (Kamst et al., 1995; Mergaert et al., 1995).

1890

The Plant Cell

Another highlight from recent biochemical approaches is the in vitro synthesis of active Nod factors. This achievement is importantfor future efforts to identify plant receptors for Nod factors: to study ligand-receptor interactions, it is important to be able to synthesize, modify, and label ligands. The total synthesis of NodRmlV and its derivatives was reported some years ago (Nicolaou et al., 1992), although yields were small. More recently, enzymatic modificationof chitin oligomers followed by chemical acylation provided a means to create a broad variety of lipo-oligosaccharidestructures (Rohrig et al., 1995). Alternatively, a galactosyl p-1,4-transferasehas been used to join tailored glucosamine residues to the nonreducing end of chitin oligomers (Atkinson et al., 1994). Radioactive labeling of Nod factors has been performed both by chemical synthesis (Nicolaou et al., 1992; Bourdineaud et al., 1995) and by using purified enzymes. The side group modifications, such as sulfates, are a particularly promising target for enzymatic modification. For example, sulfation by the R. meliloti NodH enzyme has been used to generate 35S-labeledNod factors in vitro (Bono et al., 1995; Ehrhardt et al., 1995; Schultze et al., 1995). The use of enzymes in the synthesis of radiolabeled ligands has the advantage of creating labeled molecules that have the exact structure and activity of the native signal molecules.

RECOGNITION The diverse responses of plants to Nod factors and the precise nature of side groups needed for activity suggest that molecular recognition is accomplished by one or more plant receptors that interact with the bacterial signals. What are the precise structuralrequirements for Nod factor specificity? The picture is getting more complicated, as shown in two recent reports. Stokkermans et al. (1995), using both synthetic and Bradyrhizobium-producedNod factors, found complex relationships between backbone and side group configurations in nodulationof the Glycine host. For example, GlcNAc tetramers were active without side group modifications on the reducing end residue, but pentameric molecules were inactive unless they carried a side group modification at the reducing end GlcNAc residue. The second study, by Ardourel et al. (1994), made use not of pure Nod factors but of R. meliloti cells with mutations in the nodL ornodEgenes, or both. The products of these genes direct 6-Oacetylationon the nonreducingresidue and the structure of the acyl tail, respectively. Although nodF- or nodLmutantscould stimulate nodule primoridia,infectionthread formation was inefficient. Double nodF-no&- mutants were completely defective at entry and infection thread formation, but they were still able to trigger cell divisions in the plant. On the basis of these experiments, Ardourel et al. (1994) presented a model in which two distinct receptors are involved in the plant response (see also Hirsch, 1992). One, a signaling receptor, is able to interact with non-host-specific Nod

factors, thereby triggering cell division. The second, a stringent entry receptor, must interact with specifically tailored molecules (i.e., the host-specific Nod factors) before formation of infection threads and cell invasion can proceed. This model is appealing because it provides an explanation of the current complexities of Nod factor signaling as well as providing rmm for evolutionary speculation. For example, the low stringency signaling receptor could be more evolutionarilyancient, with the high stringency entry receptor evolving later, perhaps concurrent with the development of the root hair invasion habit. However, there are some weaknesses in this model. It does not fully explain how one would obtain singlegene recessive Nod- plant mutants that lack both the root hair curling and cell division phenotypes. Also, one would predict on the basis of this model that interactions between Rhizobium species and nonhost plants (e.g., R. meliloti placed on pea roots) may induce new cell divisions in roots. These possibilities are testable. Alternatively, it is possible that the different effects of Nod factors on host and nonhost plants are due to different binding affinities of the factors, combined with complex behavior by receptor(s). The receptor-capping model described by Hirsch (1992) provides one possible scenario for how receptors may engage in different levels of interaction. Promoting plant cell division could be possible, for example, at a low threshold of receptor activity, whereas bacterial entry may require a higher leve1 of the same activity. Clearly, identifying one or more plant receptors for specific Nod factors is important. In a situation distinct from certain plant-pathogen systems discussed in other reviews in this issue (see, for example, Bent, 1996), Rhizobium researchers have a ligand but no cloned host genes corresponding to Nodplant mutations. How does one find a receptor? The first principles are clear: a receptor can only be defined as such if it satisfies both of two general criteria. First, it must bind the ligand, and when variant ligands and inhibitors are defined, the receptor should bind these with an affinity proportional to their activity. Second, the receptor must exhibit functional activity, either in vitro (ligand-dependent dimerization, phosphorylation, etc.) or in a reconstituted cellular or artificial membrane system (for example, ligand-stimulated events in a naive cell or liposome).Alternatively, the receptor should have activity in vivo, such as a genetic function for its encoding DNA (for example, complementation of a mutant or conferral of a nove1 specificity in a heterologous background). In the absence of functional activity, a protein that simply binds a Nod factor ligand cannot be considered a receptor, although it automatically becomes a promising candidate for functional studies. Binding proteins that are not true receptors may nonetheless be significant for the plant symbiotic response. For example, Nod factor binding proteins such as those described by Bono et al. (1995) may be involved in Nod factor processing or breakdown (Staehelin et al., 1995), delivery to the plant membrane, or regulatory sequestration. The classic legume seed lectins have also attracted interest as possible receptors. Although they are unlikely to provide


the sugar binding specificity necessary for specific binding of Nod factors, genes for pea lectin were reported to confer novel bacterial nodulation properties on clover plants that had been transformed with a pea lectin gene (see, e.g., van Eijsden et al., 1995). Finally, novel lectinsfrom species such as Dolichos and soybean may prove to be promising candidates to test for interaction with the bacterial signals (M. Etzler and G. Stacey, personal communication). The ultimate tests for necessity and sufficiency of Nod factors, and for putative receptors, will require genetic analyses in the host plant. In particular, plant mutants in the putative receptor genes should be sought to provide confirmation of in vivo function.

PLANT RESPONSES TO NOD FACTORS Assays for nod gene-encoded exudates, and subsequently for purified Nod factors, revealed that these bacterial signals can cause multiple nodulation-like responses in plant hosts in the absence of the producing bacteria. These responses include altered root hair growth, initiation of cell division, and the expression of a class of nodule-specific plant genes, the early nodulins, or ENODs (reviewed in Hirsch, 1992; Kijne et al., 1992; DBnarié and Cullimore, 1993; Schultze et al., 1994; Spaink, 1995). Most recent work relates to three enduring research themes: first, the production of new or upregulated transcripts (ENODs) in the plant during early nodulation; second, the involvement of plant hormones; and third, the pattern of cell division and cell rearrangement early in the nodulation response. In addition, two new areas are emerging: the genetic analysis of the host plant and the cell biology of early plant responses to Nod factors.

1891

to be made. There may be some surprises, as shown for the ENOD72 gene; this gene was widely studied because its expression pattern suggests that it is part of a relatively early signal transduction cascade (for example, see Horvath et al., 1993; Lobler and Hirsch, 1993; Bauer et al., 1994; Journet et al., 1994). Recent data bring into question the function of ENODl2, however; a Medicago subspecies lacking the ENOD72gene forms functional nodules (Csanadi et al., 1994). So, although the expression pattern of ENODlP may prove to be a useful marker for early plant responses, it is not clear whether the locus has a required role in nodulation. The sequence of at least one newly identified ENOD, ripl, does provide clues to function. Cook et al. (1995) discovered a gene that is expressed in Medicago root hairs, both in response to Rhizobium and to Nod factor, that shows high sequence similarity to peroxidases.The rapid but transient expression of this gene in root hairs responding to Rhizobium or Nod factors suggests that an oxidative process may occur early in the symbiotic response of the plant. This is especially interesting because of the importance of oxidative processes for cell wall architecture and for defense and developmental responses in plants (see Dangl et al., 1996; Hammond-Kosack and Jones, 1996, in this issue.) Another interesting early nodulin is ENOD40, for which the gene sequence is broadly conserved in legumes but which does not always include a long open reading frame. One possibility is that ENOD40 is a nontranslated regulatory RNA (Asad et al., 1994; Crespi et al., 1994). However, the identification (Vijn et al., 1995b; van de Sande et al., 1996) of a small conserved open reading frame near the 5’ end of ENOD40 transcripts opens the possibility that a small active peptide is encoded by this locus.

Morphogenesis and Molecules Nodulins The overall pattern of néw gene expression in developing nodules, revealed originally through antibody and mRNA analyses, has more recently been explored at the cellular level, both by in situ hybridization and by promoter-b-glucuronidase fusion gene expression (reviewed in Hirsch, 1992; Vijn et al., 1993; de Bruijn et al., 1994; Schultze et al., 1994; see also Horvath et al., 1993; Journet et al., 1994; Pichon et al., 1994; Vijn et al., 1995a). These analyses make it clear-tthe information carried in the Rhizobium Nod factor is sufficient to trigger the plant to mount a complex tissue-specific response at the transcriptional level. However, both the function of these so-called nodulins and the regulatory pathway controlling their induction remain unknown. Severa1 recent studies can be used to illustrate the debates surrounding nodulin function and induction. What functions do nodulins perform? Many ENODs are proline rich and potentially may be located in the cell wall (Hirsch, 1992; Lobler and Hirsch, 1993; Vijn et al., 1993; Schultze et al., 1994; Wilson et al., 1994), although explicit tests have yet

Gene expression in nodule development occurs in the broader context of organogenesis, including cytoplasmic activation and new cell division (van Brussel et al., 1992; Savoure et al., 1994; Stokkermans and Peters, 1994; Yang et al., 1994). However, the pathways and control points for cytoplasmic and cytokinetic activity have not been determined. Because nodulation is developmentally specific yet occurs in response to a facultative signal, nodules provide an experimental opportunity to study cellular events using numerous experimental approaches. For example, cell division and nodule morphogenesis provide an assay system for possible secondary messengers. It has been established that applications of either auxin transport inhibitors or cytokinins cause nodule morphogenesisand nodulin expression(reviewed in Hirsch, 1992).These findings have led to the development of a secondary signal model (Hirsch, 1992; Cooper and Long, 1994; Hirsch and Fang, 1994) in which the initial bacterial signal is transduced by a series of steps that lead to changes in the activities of endogenous hormones. The predictable patterns of cell division in nodules suggest that Nod factors affect root cells accordingto a specific physical

1892

The Plant Cell

pattern. Libbenga and others predictedthat this pattern might be the result of interactions between incoming bacterial signals and gradients of plant substances emanating from the vascular system (Libbengaand Bogers, 1974).Smit et al. (1995) singled out uridine as a component present in the vascular stele that affects cell growth and division rates in root explants. Whether a gradient of uridine exists in roots at concentrations that would correlate with the tissue culture effect is not known. Finally, several lines of evidence point to the involvement of ethylene in the regulation of early nodulation. It is known that plants downregulate nodulation in response to numerous conditions, and endogenous ethylene, based on inspectionof the plant responseand on experimental manipulation of ethylene levels, is a candidate to mediatethis regulatory process(Peters and Crist-Estes, 1989; van Spronsen et al., 1995; van Workum et al., 1995). The examination of cell division patterns provides the opportunity to explore the effect of Nod factors on plant development. Cytoplasmic activation, including phragmosome formation, occurs in Rhizobium-stimulatedroots. In some plants such as Vicia, phragmosome formation is highly site specific and occurs early in response to Nod factors (van Brussel et al., 1992). In other plants, this behavior is less evident (Yang et al., 1994). Using cyclin, histone H4, and cdk gene expression as markersfor cell activation, Yang et al. (1994)found that the cell cycle was activated (by inference, passage into G1 and S phases) in a pattern that included both the inner cortical cells at the site of the future nodule primordium and the outer cortical cells that showed phragmosome formation. The outer cells were invaded by the bacteria and did not divide further; the inner cells continued to divide, forming the nodule cortex and nodule meristem. Therefore, it appears that Rhizobium exploits both a direct result of cell division-an increased number of cells that will form the organ in which the bacteria will ultimately reside-and also an indirect consequence of cell division, namely, increased cell wall synthesis and vesicular traffic. These functions could help cause the usually inactive outer cortical cells to be as conducive to infection thread formation as are the tip-growing root hairs of the epidermis (see also discussions in Hirsch, 1992; Kijne et al., 1992; Ridge, 1992; van Brussel et al., 1992; Yang et al., 1994).

Early Cell Responses

The earliest cell divisions in nodule development occur between 4 2 and 24 hr after bacterial infection, with ENOD expression precedingthis by several hours. Early signal transduction events in epidermal or other cells may account for the coordination of the diverse host responses. Little is known of the early events in root hairs, but several recent studies have explored ion movements as one kind of indicator of cellular activity related to signal perception. Ehrhardt et al. (1992) found that alfalfa root hairs displayed a characteristic depolarization of the cytoplasmic membrane in response to the application

of R. melilotiNod factors. Felle et al. (1995) and Kurkdjian(1995) subsequently found that this cell behavior was specific to the correct structure of the Nod factor and was correlatedwith the developmental stage of the root hair itself. The ionic basis for this behavior is unknown, but Allen et al. (1994), using avibrating probe electrode, found variable currents of two ions, H+ and Ca2+,outside root hairs that had been exposed to Nod factor. More recently, Ehrhardt et al. (1996), using ion-specific reporter dyes to monitor interna1 root hair Ca2+ concentrations, observed a distinctive signal transduction behavior, termed calcium spiking, in alfalfa root hairs exposed to R. meliloti Nod factors. No such responsewas seen after exposure of root hairs to either chitin oligosaccharideor to Nod factors from R. leguminosarum. Furthermore, a Nod- alfalfa mutant that lacks both root hair curling and cell division also showed no calcium spiking. The pattern of spiking, which had a 1-min periodicity, and the lag time after Nod factor presentation, which averaged -10 min, suggest thatthis behavior is not mechanistically related to membrane depolarization. The basis for calcium spiking in plants is not characterized, but in animal cell systems, this behavior is typically associated with inositidetriphosphate signaling. Whether this is the case for calcium spiking in root hairs could be tested in future work. As with receptors, proof of functional involvement of nodulins, early cell behaviors, and signal transduction components must await analyses of plant mutants affected in nodule development (Caetano-Annolbs and Gresshoff, 1991). Plant nodulation mutants that show no response to the bacterial signa1 or that show response to an altered signal are likely to be defective in a receptor or in early stages of signal transduction. One such candidate is the sym2 mutant of Pisum sativum, which interacts differently with R. leguminosarum Nod factors, depending on whether or not they have a 6-O-acetyl group on the nonreducing residue of the oligosaccharide (see Firmin et al., 1993, and papers cited therein). Tests for early cell autonomous responses such as calcium spiking may help sort mutations into those affecting early or middle parts of the signa1transduction pathways. Arabidopsis is not an option for such genetic studies, so research focuses on genetic analysis in crop species such as pea and soybean (Caetano-Annolbsand Gresshoff, 1991; Kneen et al., 1994; Landau-Ellis and Gresshoff, 1994; see also Gianinazzi-Pearson, 1996, in this issue) and in diploid legumes with small genomes, such as Lotus japonicus, Melilotus albus, and M. truncatula (Barker et al., 1990; Miller et al., 1991; Handberg and Stougaard, 1992; Hirsch, 1992).

Nod Factors as Native Plant Growth Regulators? A Cautious View

Two general scenarios can be envisioned for the evolution of the Rhizobium-plant signalingsystem: Rhizobiumsignals could mimic a class of native plant hormones, previously undiscov-


ered, which are active not only in legumes but also in other plants. Alternatively, the Rhizobium signals, with their chitin backbone, could have coopted a plant defense response pathway originally respondingto chitinaceousor other elicitors(see discussions of this and other considerations in Sprent, 1989; Truchet et al., 1989;Brewin et al., 1992;Fisher and Long, 1992; Hirsch, 1992;Kijne et al., 1992;Spaink, 1995). The first model, in which Rhizobium Nod factors mimic native plant growth regulators, has received recent experimental and theoretical attention (Truchet et al., 1991; Dénarié and Cullimore, 1993;Spaink et al., 1993;Rohrig et al., 1995).This is an interesting,lively, and important issue about which there is at present no consensus within the Rhizobium research community. Proof that a compound acts as a true plant hormone requires that (i) the compound is actually found in plants, and (ii) it is active in causing plant responses at levels that correlate with its abundance. Presently, there is no direct biochemical evidente for the production or existence of lipo-oligosaccharide molecules anywhere in plants other than Rhizobium-induced nodules. Nor have there yet been identified any plant homologs of the genes that encode Nod factor-synthesizing enzymes. From these most stringent standpoints, we must conclude that to date, there is no evidence for endogenous Nod factor or other lipo-oligosaccharide molecules in plants. What of indirect evidence? Severa1recent studies report that lipo-oligosaccharidemolecules can be active in tissue culture systems. Based on these observations, it has been suggested that such molecules may function as plant hormones(De Jong et al., 1993;Dénarié and Cullimore, 1993;Spaink et al., 1993; Rohrig, 1995).The tissue culture data may provide the basis for direct experimental tests of lipo-oligosaccharide activities by pointing to likely places of action. Nevertheless,we should recall the lesson learned from many years of phytohormone studies-effects in tissue culture do not always occur because the applied compound closely imitates the structure of a native regulator. Consider auxin: its structure and mode of action would not likely be deducible from an inspection of the structures and activities of functional analogs such as 2,4dichlorophenoxyacetic acid and naphthylphthalamic acid. One study of lipo-oligosaccharideactivity exploitedthe carrot somatic embryogenesis system (De Jong et al., 1993).A wild-type cell line of carrot forms 4 5 0 embryos per 10,000 cells plated; the tsll variant is defective in somatic embryogenesis, forming only 0.3embryo per 10,000cells at 32%. This low number of embryos regenerating at the restrictive temperature could be elevated, although to only 1 to 2% of wild4 per 10,000cells), by treatment of type levels ( ~ 2 . embryos the cultured tsll cells with a purified R. /. viciae Nod factor. The variability of the assay was high, and in the experiments reported to date, the putative inactive compounds (such as Nod factors with altered structure) were evaluated in very few trials comparedwith those for the proposedactive compounds. Future work with a larger number of trials would help to substantiate the hypothesized relationship between Nod factors

1893

and embryogenesis. For example, this link could be further explored through the use of Nod factors in attempts to rescue zygotic embryo-defective mutants, such as those described for Arabidopsis (reviewed in Jürgens et al., 1994;Meinke, 1995). A second study utilized a different series of assays to follow the effects of Nod factor-like lipo-oligosaccharides on cultured plant cells. Rohrig et al. (1995)constructed a series of lipooligosaccharide molecules by combinedenzymatic and chemical modifications of chitin oligomeric backbones (John et al., 1993).This versatile method permittedthe construction of Nod factor-like molecules with structures distinct from those synthesized by Rhizobium. Among the molecules synthesized by these authors were molecules with both cis- and transconformations at the C9 and C11 positions of the N-acyl moiety. The moleculeswith trans conformation consistently showed elevated activity in a series of assays. This result is striking because such molecules have not been found among naturally occurring Rhizobium Nod factors, all of which have the cis conformation at the corresponding position.The active synthetic Nod factors also enhanced tobacco protoplast division rates, independent of auxinkytokinin ratios. Nod factors with cis-conformationacyl groups, like chitin oligomers, had slight effects. When the active compounds were presented to protoplasts along with kinetin, expression from a cauliflower mosaic virus 35s partia1 promoter was stimulated. Moreover, in protoplasts given Nod factor only, transcription from the auxin responsive locus, AXI, was increased. Rohrig et al. (1995) propose from these experiments that auxin and lipooligosaccharides share a signal transduction pathway and that the observed increase in tobacco protoplast division is caused by a mechanism similar to that triggering nodule formation in legumes. If it is indeed the case that lipo-oligosaccharides imitate auxin through a common mechanism, one may also expect to observe effects of Nod factor-like compounds on morphogenesisor transcription in intact plants or to effect the rescue of some categories of auxin mutant using these compounds.

CONCWDING REMARKS One of the most distinctive features of the Rhizobium-legume symbiosis is its specificity. Any model for receptor function and evolution must account for the restriction of specific Rhizobium species or biovars to particular host plants and must explain in general why the Rhizobium symbiosis is restricted'to the Fabaceae (Brewin et al., 1992).The suggestion that Nod factors may be related to or function as native and universal plant growth regulators is somewhat difficult to reconcile with their observed biological specificity in symbiosis. If Nod factors represent a broadly based and universal morphogeneticpathway, one would need to explain how this pathway evolved into a beneficia1 symbiosis in only one plant family that originated 4 5 to 90 million years ago. Perhaps, if this general category

1894

The Plant Cell

of compounds exists and is active in plants, then the native plant compounds have some basic structural difference, as suggested by the data of Rdhrig et al. (1995). At present, evolutionary theories have few data on which to build. New experimental details, arising from biochemical, cellular, and genetic studies of the plant’s role in early nodulation, will help refine our conceptual framework for how nodulation arose in evolution.

ACKNOWLEDGMENTS

Work in my laboratory is supported by the Howard Hughes Medical Institute,the Department of Energy(Gran1No. DE-FG-03-90ER20010), and the National lnstitutesof Health (Grant No. R37-GM30962). I wish to acknowledge my laboratory group for many useful discussions; I am also grateful to W.D. Bauer, T. Bisseling, N. Brewin, A. Hirsch, J.A. Downie, and R. Carlson for numerous helpful comments and suggestions. I thank JoAnne Connellyfor muchassistancein the preparation of the manuscript and Crispin Taylor for editorial comments and review.

REFERENCES

Allen, N.S., Bennett, M.N., Cox, D.N., Shipley, A., Ehrhardt, D.W., and Long, S . R . (1994). Effects of Nod factors on alfalfa root hair Ca++and H+currents and cytoskeletal behavior. In Advances in Molecular Genetics of Plant-Microbe Interactions, M.J. Daniels, J.A. Downie, and A.E. Osbourn, eds (Dordrecht, The Netherlands:Kluwer Academic Publishing), pp. 107-114. Ardourel, M., Demont, N., DebellB, F., Maillet, F., de Billy, F., PromB, J.-C., DénariQ, J., and Truchet, G. (1994). Rhizobium meliloti lipooligosaccharide nodulation factors: Differentstructural requirements for bacterial entry into target root hair cells and induction of plant symbiotic developmental responses. Plant Cell6, 1357-1374. Asad, S., Fang, Y., Wycoff, K.L., and Hirsch, A.M. (1994). lsolation and characterization of cDNA and genomic clones of Msenod40: Transcriptsare detected in meristematic cells of alfalfa. Protoplasma 183, 10-23.

Bec-Fert6, M.P., Krishnan, H.B., PromB, D., Savagnac, A., Pueppke, S.G., and PromB, J.-C. (1994). Structures of nodulation factors from the nitrogen fixing soybean symbiont Rhizobium fredii USDA257. Biochemistry 33, 11782-11788. Bent, A.F. (1996).Plant disease resistancegenes: Function meets structure. Plant Cell 8, 1757-1771. Bloemberg, G.V., Kamst, E.,Harteveld, M., van der Drift, K.M.G.M., Haverkamp, J., Thomas-Oates, J.E., Lugtenberg, B.J.J., and Spaink, H.P. (1995). A central domain of Rhizobium NodE protein mediates host specificity by determining the hydrophobicity of fatty acyl moieties of nodulation factors. MOI.Microbiol. 16, 1123-1136.

Bono, J.J., Riond, J., Nicolaou, K.C., Bockovich, N.J., Estevez, V.A., Cullimore, J.V., and Ranjeva, R. (1995). Characterization of a binding site for chemically synthesizedlipooligosaccharide NodRm factors in particulate fractions prepared from roots. Plant J. 7, 253-260. Bourdineaud, J.-P., Bono, J.J., Ranjeva, R., and Cullimore, J.V. (1995). Enzymatic radiolabeling to a high specific activity of legume lipwligosaccharidic nodulationfactors from Rhizobiummeliloti. Biochem. J. 306, 259-264. Brewin, N.J. (1991). Development of the legume root nodules. Annu. Rev. Cell Biol. 7, 191-226. Brewin, N.J., Downie, J.A., and Young, J. (1992). Nodule formation in legumes. In Encyclopedia of Microbiology, Vol. 3., J. Lederberg, ed (San Diego, CA: Academic Press), pp, 239-248. Caetano-Annol&, G., and Gresshoff, P.M. (1991).Plant genetic control of nodulation. Annu. Rev. Microbiol. 45, 345-382. Cardenas, L., Dominguez, J., Quinto, C., Lopez-Lara, I.,Lugtenberg, B., Spaink, H., Rademaker, G., Haverkamp, J., and ThomasOates, J. (1995). Isolation, chemical structures and biological activity of the lipo-chitin oligosaccharide nodulation signals from Rhizobium etli. Plant MOI. Biol. 29, 453-464. Carlson, R.W., Sanjuan, J., Bhat, U.R., Glushka, J., Spaink, H.P., Wijfjes, A.H.M., van Brussel, A.A.N., Stokkermans, T.J.W., Peten, N.K., and Stacey, G. (1993). The structures and biological activities of the lipo-oligosaccharide nodulation signals produced by type I and type II strains of Bradyrhizobium japonicum. J. Biol. Chem. 268, 18372-18381. Carlson, R.W., Price, N., and Stacey, G. (1994). The biosynthesis of Rhizobial lipo-oligosaccharide nodulation signal molecules. MOI. Plant-Microbe Interact. 7, 684-695.

Atkinson, E.M., Palcic, M.M., Hindsgaul, O., and Long, S.R.(1994). Biosynthesis of Rhizobium meliloti lipooligosaccharide Nod factors: NodA is requiredfor an N-acyltransferaseactivity. Proc. Natl. Acad. Sci. USA 91, 8418-8422.

Cook, D., Dreyer,, D., Bonnet, D., Howell, M., Nony, E., and VandenBosch, K. (1995). Transient inductionof a peroxidase gene in Medicagotruncatula precedes infection by Rhizobium me/iloti. Plant Cell 7, 43-55.

Barker, D., Bianchi, S., Blondon, F., Dattee, Y., Duc, G., Essad, S., Flament, T., Gallusci, T., Genier, G., Guy, P.,Muel, X., Tourner, J., DBnari6, J., and Huguet, T. (1990). Medicago truncatula, a model plant for studying the molecular genetics of the Rhizobium-legume symbiosis. Plant MOI. Biol. Rep. 8, 40-49.

Cooper, J.B., and Long, S.R. (1994). Morphogenetic rescue of Rhizobium meliloti nodulation mutants by trans-zeatin secretion. Plant Cell 6, 215-225.

Barney, M., and Downie, J.A. (1993). ldentificationof the NodC protein in the inner but not the outer membrane of Rhizobium leguminosarum. MOI. Plant-Microbe Interact. 6, 669-672. Bauer, I?, Crespi, M.D., Szecsi, J., Allison, L.A., Schultze, M., Ratet, P.,Kondorosi, E., and Kondorosi, A. (1994). Alfalfa fNOD72genes are differentially regulated during nodule development by Nod factors and Rhizobium invasion. Plant Physiol. 105, 585-592.

Cren, M., Kondorosi, A., and Kondorosi, E. (1995). NolR controls expression of the Rhizobium me/i/otinodulation genes involved in the core Nod factor synthesis. MOI. Microbiol. 15, 733-747. Crespi, M.D., Jerkevitch, E., Poiret, Y., dAubenton-Carafa, Y., Petrovics, G., Kondorosi, E., and Kondorosi, A. (1994). enOd40, a gene expressed during nodule organogenesis, codes for a nontranslatable RNA involved in plant growth. EMBOJ. 13,5099-5112. . Csanddi, G., SzBcsi, J., Kald, P., Kiss, P., Endre, G., Kondorosi, A., Kondorosi, E., and Kiss, G.B. (1994). ENOD12, an early nodu-


1895

lin gene, is not required for nodule formation and efficient nitrogen fixation in alfalfa. Plant Cell 6, 201-213.

root hairs are differentially sensitive to structural modifications of the lipo-chitooligosaccharide.Plant J. 7, 939-947.

Dangl, J.L., Dletrlch, R.A., and Richberg, M.H. (1996). Death don't have no mercy: Cell death programs in plant-microbe interactions. Plant Cell 8, 1793-1807.

Firmln, J., Wilson, K., Carlson, R., Davies, A., and Downle, J.A. (1993). Resistance to nodulation of cv. Afghanistan peas is overcome by nodX, which mediates an Oacetylation of the Rhizobium leguminosarum lipo-oligosaccharidenodulationfactor. MOI.Microbiol. 10, 351-360.

DeBoer, M., and Djordjevic, M. (1995). The inhibition dinfection thread development in the cultivar-specific interaction of Rhizobium and subterranean clover is not caused by a hypersensitive response. Protoplasma 185, 58-71. de Bruijn, F.J., and Downie, J.A. (1991). Biochemical and molecular studies: Symbiotic nitrogen fixation. Curr. Opin. Biotechnol. 2, 184-192. de Bruijn, F.J., Chen, R., Fujimoto, S.Y.,Pinaev, A., Silver, D., and Szwyglowski, K. (1994). Regulation of nodulin gene expression. Plant Soil 161, 59-68. De Jong, A.J., Heidstra, R., Spaink, H.P., Hartog, M.V., Meijer, E.A., Hendrlks, T., Lo Schiavo, F., Temi, M., Bisseling, T., Van Kammen, A., and De Vrles, S.C. (1993). Rhizobiumlipo-oligosaccharides rescue a carrot somatic embryo mutant. Plant Cell 5, 615-620. Demont, N., Debellé, F., Aurelle, H., D6nari6, J., and PromB, J.-C. (1993). Role of the Rhizobium melilotinodF and nodE genes in the biosynthesis of lipmligosaccharidicnodulationfactors. J. Biol. Chem. 268, 20134-20142. Demont, N., Ardourel, M., Malllet, F., Prom6, D., Ferro, M., Promé, J.-C., and D6nari6, J. (1994). The RhizobiummelilofiregulatorynOdD3 and syrM genes control the synthesis of a particular class of nodulation factors N-acylatedby (omega-1)-hydroxylatedfatly acids. EMBO J. 13, 2139-2149. DBnari6, J., and Cullimore, J. (1993). Lipo-oligosaccharide nodulation factors: A new class of signaling moleculesmediating recognition and morphogenesis. Cell 74, 951-954. DBnari6, J., DebellB, F., and Rosenberg, C. (1992). Signaling and host range variation in nodulation. Annu. Rev. Microbiol. 46,497-531. andStacey, G. (1994). NolA Dockendorff, T.C., Sanjuan, J., Grob, I?, represes nod gene expression in 8radyrhizobiumjaponicum. MOI. Plant-Microbe Interact. 7, 596-602. Downie, J.A. (1994). Signaling strategies for nodulation of legumes by Rhizobia. Trends Microbiol. 2, 318-324. Downie, J.A., and Surin, 8. (1990). Either of two nod gene loci can complement the nodulation defect of anoddeletion mutant of Rhizobium leguminosarum bv. viciae. MOI. Gen. Genet. 222, 81-86. Ehrhardt, D.W., Atkinson, EM., and Long, S.R. (1992). Depolarization of alfalfa root hair membrane potential by Rhizobium melilofi Nod factors. Science 256, 998-1000. Ehrhardt, D.W., Atkinson, EM., Faull, K.F., Freedberg, D.I., Sutherlin, D.P., Armstmng, R., and Long, S.R. (1995). In vitro sulfotransferase activity of NodH, a nodulation protein of Rhizobium meliloti required for host-specific nodulation. J. Bacteriol. 177, 623-6245, Ehrhardt, D.W., Wais, R., and Long, S.R. (1996). Calcium spiking in plant root hairs responding to Rhizobium nodulation signals. Cell 85, 673-681. Fellay, R., Perret, X., Viprey, V., Broughton, W., and Brenner, S. (1995). Organization of host-inducible transcripts on the symbiotic plasmid of Rhizobium sp. NGR234. MOI. Microbiol. 16, 657-667. Felle, H., Kondoroai, E., Kondorosi, A., and Schultze, M. (1995). Nod signal-induced plasma membrane potential changes in alfalfa

Fisher, R.F., and Long, S.R. (1992). Rhizobium-plant signal exchange. Nature 357, 655-660. Fisher, R.F., and Long, S.R. (1993). lnteractions of NodD at the nod box: NodD binds to two distinct sites on the same face of the helix and induces a bend in the DNA. J. MOI.Biol. 233, 336-348. Folch-Mallol, J.L., Marroqui, S.,Sousa, C., Manyani, H., LopezLara, I.M., van der Drift, K.M.G.M., Haverkamp, J., Quinto, C., Gil-Serrano, A., Thomas-Oates, J., Spaink, H.P., and Meglas, M. (1996). Characterization of Rhizobium tropici CIAT899 nodulation factors: The role of nodH and nodPO genes in their sulfation. MOI. Plant-Microbe Interact. 9, 151-163. Geelen, D., Leyman, B., Mergaert, P., Klarskov, K., Van Montagu, M., Geremia, R., and Holsters, M. (1995). NodS is a S-adenosylL-methionine-dependent methyltransferase that methylates chitooligosaccharides deacetylated at the non-reducing end. MOI. Microbiol. 17, 387-397. Geiger, O., Thomas-Oates, J., Glushka, J., Spaink, H., and Lugtenberg, B.J.J. (1994). Phospholipidsd RhizobiumcontainnodEdetermined highly unsaturated fatty acid moieties. J. Biol. Chem. 269, 11090-11097. Gianinaui-Pearson, V. (1996). Plant cell responses to arbuscular mycorrhizal fungi: Getting to the roots of the symbiosis. Plant Cell 8, 1871-1883. Hammond-Kosack, K.E., and Jones, J.D.G. 1996. Resistance gene-dependent plant defense responses. Plant Cell8,1773-1791. Handberg, K., and Stougaard, J. (1992). Lotus japonicus, an autogamous, diploid legume species for classical and molecular genetics. Plant J. 2, 487-496. Handelsman, J., and Stabb, E.V. (1996). Biocontrol of soilborne plant pathogens. Plant Cell 8, 1855-1869. Higashi, S. (1993). (8rady)Rhizobium-plant communications involved in infection and nodulation. J. Plant Res. 106, 201-211. Hirsch, A.M. (1992). Developmental biology of legume nodulation. New PhytOl. 122, 211-237. Hirsch, A.M., and Fang, Y. (1994). Plant hormones and nodulation: What's the connection? Plant MOI.Biol. 26, 5-9. Horvath, E., Heidstra, R., Lados, M., Moerman, M., Spaink, H.P., Promé, J.-C., Van Kammen, A., and Bisseling, T. (1993). Lipooligosaccharides of Rhizobiuminduce infection-relatedearly nodulin gene expression in pea root hairs. Plant J. 4, 727-733. Jabbourit, S., Fellay, R., Talmont, F., Kamalaprija, P., Burgert, U., Rellc, B., Prom6, J.-C., and Broughton, W. (1995). lnvolvement of nodS in N-methylationand nodU in 6-Ocarbamoylationof Rhizobium sp. NGR234 Nod factors. J. Biol. Chem. 270, 22968-22973. John, M., Rahrig, H., Schmidt, J., Wieneke, U.,and Schell, J. (1993). Rhizobium NodB protein involved in nodulation signal synthesis is a chitooligosaccharide deacetylase. Proc. Natl. Acad. Sci. USA 90, 625-629. Journet, E., Pichon, M., Dedieu, A., de Billy, F., Truchet, G., and Barker, D. (1994). Rhizobium meliloti Nod factors elicit cell-specific

1896

The Plant Cell

transcription of the ENOD72 gene in transgenic alfalfa. Plant J. 6, 241-249. Jürgens, G., Torres Ruiz, R.A., and Berleth, T. (1994). Embryonic pattern formation in flowering plants. Annu. Rev. Genet. 26,351-371. Kamst, E., van der Drift, K.M.G.M., Thomas-Oates, J.E., Lugtenberg, B.J.J., and Spaink, H.P. (1995). Mass spectrometric analysis of chitin oligosaccharides produced by Rhizobium NodC protein in Escherichia coli. J. Bacteriol. 177, 6282-6285. ,

-

Kijne, J.W., Bakhuizen, R., van Biussel,,A.A.N.,..CanterCremers, H.C.J., Diaz, C.L., de Pater, B.S'., Smit, G., Spaink, H.P., Swart, s., Wijffelman, C.A., and Lugtenberg, B.J.J. (1992). The Rhizobium trap: Root hair curling in rwt-nodule symbiosis. In Perspectives in Plant Cell Recognition, Society for Experimental Biology Seminar Series, Vol. 48 (Cambridge, UK: Cambridge University Press), pp. 267-284. Kneen, B., Weeden, N.F., and LaRue, T.A. (1994). Non-nodulating mutants of Pisum safivum (L) cv. Sparkle. Heredity J. 85, 129-133. Kondorosi, A. (1992). Regulation of nodulation genes in Rhizobia. In Molecular Signals in Plant-Microbe Communication, D.P.S. Verma, ed (Boca Raton, FL: CRC Press), pp. 325-340. Kurkdjian, A.C. (1995). Role of differentiationof root epidermal cells in Nod factor (from Rhizobium meliloti) induced root hair depolarization of Medicago sativa. Plant Physiol. 107, 783-790. Landau-Ellis, D., and Gresshoff, P. (1994). The RFLP molecular marker closely linked to the supernodulation locus of soybean contains three inserts. MOI. Plant-Microbe Interact. 7, 432-433.

U

Lawson, C., Rolfe, B., and Bjordjevic, M. (1995). Rhizobium inoculation induces condition-dependentchanges in the flavonoid composition of root exudates in Triolium subferraneum. Aust. J. Plant Physiol. 23, 93-101. Lerouge, P., Roche, P., Faucher, C., Maillet, F., Truchet, G., PromB, J.-C., and Denari&, J. (1990). Symbiotic host-specificity of Rhizobium melilotiis determined by a sulfated and acylated glucosamine oligosaccharide signal. Nature 344, 781-784. Libbenga, K.R., and Bogers, R.J. (1974). Root-nodule morphogenesis. In The Biology of Nitrogen Fixation, A. Quispel, ed (Amsterdam: North-Holland Publishing Co.), pp. 430-472. Lobler, M., and Hirsch, A. (1993).A gene that encodes a proline rich nodulinwith limited homology to PsENOD12 is expressed in the invasion zone of Rhizobiummelilotiinduced alfalfa root nodules. Plant Physiol. 103, 21-30. Lopez-Lara, I.M., van den Berg, J.D. J., Thomas-Oates, J.E., Glushka, J., Lugtenberg, B.J.J., and Spaink, H.P. (1995). Structural identificationof the lipochitin oligosaccharids nodulation signals oí Rhizobium loti. MOI.Microbiol. 15, 627-638. Meinke, D.W. (1995). Molecular genetics of plant embryogenesis. Annu. Rev. Plant Physiol. Plant MOI. Biol. 46, 369-394. Mergaert, J., Van Montagu, M., P n " , J.&, and Holstem, M. (1993). Three unusual modifications, a o-arabinosyl, an N-methyl, and a carbamoyl group, are present on the Nod factors of Azorhizobium caulinodans ORS571. Proc. Natl. Acad. Sci. USA 90, 1551-1555. Mergaert, P., D'Haeze, W., Geelen, D., PromB, D., Van Montagu, M., Geremia, R., Promb, J.-C., and Holsters, M. (1995). Biosynthesis of Azorhizobium caulinodans Nod factors: Study of the activity of the NodABCS proteins by expression of the genes in Escherichia coli. J. Biol. Chem. 270, 29217-29223. Miller, J.E., Viands, D.R., and Larue, T.A. (1991). lnheritanceof nonnodulating mutants of sweetclover. Crop Sci. 31, 948-952.

Nicolaou, N., Bockovich, N.J., Carcanague, DA., Hummel, C.W., and Even, L.F. (1992). Total synthesis of the NodRm-IV factors, the Rhizobium nodulation signals. J. Am. Chem. SOC.114, 8701-8702. Pawlowski, K., and Bisseling, T. (1996). Rhizobial and actinorhizal symbioses: What are the shared features? Plant Cell 8, 1899-1913. Peters, N.K., and Crlst-Estes, D.K. (1989). Nodule formation is stimulated by the ethylene inhibitor aminoethoxyvinylglycine.Plant Physiol. 91, 690-693. Philip-Hollingsworth, S.,Orgambide, G.G., Bradford, J.J., Smith, D.K., Hollingsworth, R.I., and Dauo, F.B. (1995). Mutation or increased copy number of nodE has no effect on the spectrum of chitolipooligosaccharide Nodfactors made by Rhizobiumleguminosarum bv trifolii, J. Biol. Chem. 270, 20968-20977. Phillips, D.A., Dakora, F.D., Sande, E., Joseph, C.M., and Zon, J. (1994). Synthesis, release and transmission of alfalfa signals to rhizobial symbionts. Plant Soil 161, 69-80. Pichon, M., Journet, E., de Biliy, F., Huguet, T., Truchet, G., and Barker, 0. (1994). ENODl2 gene expression as a molecular marker for comparing Rhizobium-dependentand 4ndependent nodulation in alfalfa. MOI. Plant-Microbe Interact. 7, 740-747. Poupot, R., Martinez-Romero, E., and Prom6, J.-C. (1993). Nodulation factors from Rhizobium tropici are sulfated or nonsulfated chitopentasaccharides containing an N-methyl-N-acylglucosamine terminus. Biochemistry 32, 10430-10435. Price, N., and Carbon, R. (1995). Rhizobial lipo-oligosaccharide nodulation factors: Multidimensionalchromatographic analysis of symbiotic signals involved in the development of legume root nodules. Glycobiology 5, 233-242. Price, N.P.J., Relic, B., Talmont, F., Lewin, A., PromB, D., Pueppke, S.G., Maillet, F., DBnariB, J., PromB, J.-C., and Broughton, W.J. (1992). Broad-host-rangeRhizobium species strain NGR234 secretes afamily of carbamoylated, and fucosylated, nodulationsignals that are O-acetylated or sulfated. MOI. Microbiol. 6, 3575-3584. Relic, B., Perret, X., Estrada-Garcia, M.T., Kopclnska, J., Gollnowski, W., Krishnan, H.B., Pueppke, S.G., and Broughton, W.J. (1994). Nod factors of Rhizobium are a key to the legume dwr. MOI.Microbiol. 13, 171-178. Ridge, R. (1992). A model of legume root hair growth and Rhizobium infection. Symbiosis 14, 359-373. Ridge, R. (1995). Micro-vesicles, pyriform vesicles and macro-vesicles associated with the plasma membrane in the root hairs of Vicia hirsuta after freeze-substitution. J. Plant Res. 108, 636-638. Rivilla, R., Sutton, J., and Downie, J.A. (1995). Rhizobiumleguminosarum NodT is related to a family of outer-membranetransport proteins that includes TolC, PrtF, CyaE and AprF. Gene 161, 27-31. Rohrig, H., Schmidt, J., Wieneke, U., Kondorosi, E., Barlier, I., Schell, J., and John, M. (1994). Biosynthesis of lipo-oligosaccharide nodulation factors: Rhizobium NodA protein is involved in Kacylation of the chitooligosaccharide backbone. Proc. Natl. Acad. Sci. USA 91, 3122-3126. Rohrig, H., Schmidt, J., Walden, R., Czaja, I., Miklasevics, E., Wieneke, U., Schell, J., and John, M. (1995). Growth of tobacco protoplasts stimulated by synthetic tipo-chitooligosaccharides. Science 269, 841-843. Sanchez, F., Cardenas, L., van Spronsen, P., Villanueva, M.A., Valdes, V., Guillen, G., Rongcai, M., Noguez, R., Dantan, E., Vidali, L., Perez, H., Hepler, P., and Quinto, C. (1996). Actin


cytoskeleton expression and organization during root-noduledevelopment. Plant Physiol. 111, S3. Sanjuan, J., Carlson, R.W., Spaink, H.P., Bhat, U.R., Barbour, W.M., Glushka, J., and Stacey, G. (1992). A 2-O-methylfucosemoiety is present in the lipo-oligosaccharide nodulation signal of Bradyrhizobium japonicum. Proc. Natl. Acad. Sci. USA 89, 8789-8793. Sanjuan, J., Grob, P., Gottfert, M., Hennecke, H., and Stacey, G. (1994). NodW is essential for full expression of the common nodulation genes of Bradyrhizobium japonicum. MOI. Plant-Microbe Interact. 7, 364-369. Savoure, A., Magyar, A., Pierre, M., Brown, S., Schultze, M., and Kondorosi, A. (1994). Activation of the cell cycle machinery and the isoflavonoid biosynthesis pathway by active Rhizobium meliloti Nod signal moleculesin Medicago microcallussuspensions. EM60 J. 13, 1093-1102. Schlaman, H.R.M., Okker, R.J.H., and Lugtenberg, B.J.J. (1992). Regulation of nodulation gene expression by NodD in Rhizobia. J. Bacteriol. 174, 5177-5182. Schultze, M., Quiclet-Sire, B., Kondomsi, E., Virelizier, H., Glushka, J.N., Endre, G., Gero, S.D., and Kondorosi, A. (1992). Rhizobium meliloti produces a family of sulphated lipo-oligosaccharidesexhibiting different degrees of plant host specificity. Proc. Natl. Acad. Sci. USA 89, 192-196. Schultze, M., Kondorosi, E., Ratet, P., Buire, M., and Kondorosi, A. (1994). Cell and molecular biology of Rhizobium-plant interactions. Int. Rev. Cytol. 156, 1-75. Schultze, M., Staehelin, C., Rohrig, H., John, M., Schmidt, J., Kondorosi, E., Schell, J., and Kondorosi, A. (1995). ln vitro sulfotransferase activity of Rhizobium meliloti NodH protein: Lipochitooligosaccharide nodulation signals are sulfated after synthesis of the core structure. Prw. Natl. Acad. Sci. USA 92,2706-2709. Smit, G., de Koster, C.C., Schripsema, J:, Spaink, H.P., van Brussel, A.A., and Kijne, J. (1995). Uridine, a cell division factor in pea roots. Plant MOI. Biol. 29, 869-873. Spaink, H.P. (1995). The molecular basis of infection and nodulation by Rhizobia: The ins and outs of sympathogenesis. Annu. Rev. Phytopathol. 33, 345-368. Spaink, H.P., and Lugtenberg, B.J.J. (1994). Role of rhizobial lipochitin oligosaccharide signal molecules in root nodule organogenesis. Plant MOI. Biol. 26, 1413-1422. Spaink, H.P., Sheeley, M., van Brussel, A.A.N., Glushka, J., York, W.S., Tak, T., Geiger, O., Kennedy, E.P., Reinhold, V.N., and Lugtenberg, B.J.J. (1991). A nove1highly unsaturated fatty acid moiety of Iipo-oligosaccharidesignals determines host specificity of Rhizobium. Nature 354, 125-130. Spaink, H.P., Wijfjes, A.H.M., van Vliet, T.B., Kijne, J.W., and Lugtenberg, B.J.J. (1993). Rhizobium lipo-polysaccharidesignals and their role in plant morphogenesis: Are analogouslipophilic chitin derivatives producedby the plant? Aust. J. Plant Physiol.20,381-392. Spaink, H.P., Bloemberg, G.V., van Brussel, A.A.N., Lugtenberg, B.J.J., van der Dritt, K.M.G.M., Haverkamp, J., and ThomasOates, J.E. (1995). Host specificity of Rhizobiumleguminosarumis determined by the hydrophobicity of highly unsaturated fatty acyl moieties of the nodulation factors. MOI.Plant-Microbe Interact. 8, 155-164.

1897

cific nodulation gene, is involved in the fucosylation of the lipooligosaccharidenodulation signal of Bradyrhizobiumjaponicum. J. Bacteriol. 176, 620-633. Staehelin, C., Schultze, M., Kondomsi, E., and Kondorosi, A. (1995). Lipochitooligosaccharidenodulation signals from Rhizobium meliloti induce their rapid degradation by the host plant alfalfa. Plant Physiol. 108, 1607-1614. Stokkermans, T., and Peters, N.K. (1994). Bradyrhizobium elkanii lipooligosaccharidesignals induce complete nodule structures on Glycine soja Siebold et Zucc. Planta 193, 413-420. Stokkermans, T., Ikeshita, S., Cohn, J., Carlson, R., Stacey, G., Ogawa, T., and Peten, N.K. (1995). Structural requirementsof synthetic and natural product lipo-chitin oligosaccharidesfor induction of nodule primordia on Glycinesoja. Plant Physiol. 108, 1587-1595. Sutton, J., Lea, E., and Downie, J.A. (1994). The nodulation-signaling protein NodO from Rhizobium leguminosarum biovar viciae forms ion channels in membranes. Proc. Natl. Acad. Sci. USA91,9991?-9994. Truchet, G., Camut, S., de Billy, F., Odorico, R., and Vasse, J. (1989). The Rhizobium legume symbiosis: Two methods to discriminate between nodules and other root derived structures. Protoplasma149, 82-88. Truchet, G., Roche, P., Lerouge, P., Vasse, J., Camut, S., de Billy, F., PromB, J.-C., and DBnariB, J. (1991). Sulphated lipooligosaccharide signals of Rhizobium melilofi elicit root nodule organogenesis in alfalfa. Nature 351, 670-673. van Brussel, A., Bakhuizen, R., van Spronsen, P., Spaink, H., Tak, T., Lugtenberg, B., and Kijne, J. (1992). lnduction of pre-infection thread structures in the leguminous host plant by mitogenic lipooligosaccharides of Rhizobium. Science 257, 70-72. van de Sande, K., Pawlowski, K., Czaja, I., Wieneke, U., Schell, J., Schmidt, J., Walden, R., Matvienko, M., Wellink, J., van Kammen, A., Franssen, H., and Bisselling, T. (1996). Modification of phytohormone response by a peptide encoded by ENOD40 of legumes and a nonlegume. Science 273, 370-373. van Eijsden, R., Diaz, C., de Pater, B.S., and Kijne, J. (1995). Sugarbinding activity of pea (Pisum sativum) lectin is essential for heterologous infectionof transgenic white clover hairy roots by Rhizobium leguminosarum biovar viciae. Plant MOI. Biol. 29, 431-439. van Spronsen, P.V., Bakjuizen, R., van Brussel, A.A.N., and Kijne, J. (1994). Cell wall degradation during infection thread formation by the root nodule bacterium Rhizobium leguminosarum is a twostep process. Eur. J. Cell Biol. 64, 88-94. van Spronsen, P.V., van Brussel, A.A.N., and-Kijne, J. (1995). Nod factorsproduced by Rhizobium leguminosarum biovar viciae induce ethylene-related changes in root cortical cells of Vicia sativa ssp. nigra. Eur. J. Cell Biol. 68, 463-469. van Workum, W., van Brussel, A., Tak, T., Wijffelman, C., and Kijne, J. (1995). Ethylene prevents nodulation of Vicia sativa ssp. nigra by exopolysaccharide-deficient mutants of Rhizobium leguminosarum bv. viciae. MOI.Plant-Microbe Interact. 8, 278-285. Vijn, I., das Neves, L., Van Kammen, A., Franssen, H., and Bisseling, T. (1993). Nod factors and nodulation in plants. Science 260, 1764-1765.

Sprent, J. (1989). Which steps are essential for the formation of functional legume nodules? New Phytol. 111, 129-153.

Vijn, I., Martinez-Abarca, F., Yang, W.C., das Neves, L., van Brussel, A., Van Kammen, A., and Bisseling, T. (1995a). Early nodulin expression during Nod factor induced processes in Vicia sativa. Plant J. 8, 111-119.

Stacey, G., Luka, S., Sanjuan, J., Banfalvi, Z.,Nieuwkoop, A.J., Chun, J.Y., and Forsberg, L.S. (1994). NodZ, a unique host spe-

Vijn, I.,Yang, W.-C., Pallisgard, N.E.O., Jensen, E.O., Van Kammen, A., and Bisseling, T. (1995b).VsENODS, VsENOD12 and VsENOD40

1898

The Plant Cell

expression during Rhizobium induced nodule formation on Vicia sativa roots. Plant MOI. Biol. 28, 1111-1119.

Wilson, C.W., Long, F., Maruok, EM., and Cooper, J.B. (1994).A new proline-rich early nodulin from Medicago truncatula is highly expressed in nodule meristematic cells. Plant Cell 6, 1265-1275.

Yang, W.-C., de Blank, C., Meskiene, I.,Hirt, H., Bakker, J., Van Kammen, A., Franssen, H., and Bisseling, T. (1994). Rhizobium Nod factors reactivate the cell cycle during infection and nodule primordium formation, but the cycle is only completed in primordium formation. Plant Cell 6, 1415-1426.