Nod Factor Requirements for Efficient Stem and Root Nodulation of the Tropical Legume Sesbania rostrata *

Azorhizobium caulinodans ORS571 synthesizes mainly pentameric Nod factors with a household fatty acid, an N-methyl, and a 6-O-carbamoyl group at the nonreducing-terminal residue and with a d-arabinosyl, anl-fucosyl group, or both at the reducing-terminal residue. Nodulation on Sesbania rostrata was carried out with a set of bacterial mutants that produce well characterized Nod factor populations. Purified Nod factors were tested for their capacity to induce root hair formation and for their stability in an in vitro degradation assay with extracts of uninfected adventitious rootlets. The glycosylations increased synergistically the nodulation efficiency and the capacity to induce root hairs, and they protected the Nod factor against degradation. The d-arabinosyl group was more important than the l-fucosyl group for nodulation efficiency. Replacement of the 6-O-l-fucosyl group by a 6-O-sulfate ester did not affect Nod factor stability, but reduced nodulation efficiency, indicating that thel-fucosyl group may play a role in recognition. The 6-O-carbamoyl group contributes to nodulation efficiency, biological activity, and protection, but could be replaced by a 6-O-acetyl group for root nodulation. The results demonstrate that none of the studied substitutions is strictly required for triggering normal nodule formation. However, the nodulation efficiency was greatly determined by the synergistic presence of substitutions. Within the range tested, fluctuations of Nod factor amounts had little impact on the symbiotic phenotype.

Bacteria of the genera Allorhizobium (1), Azorhizobium, Bradyrhizobium, Mesorhizobium, Rhizobium, and Sinorhizobium, can establish a symbiosis with specific leguminous host plants. A compatible interaction leads to the development of nitrogenfixing nodules as the result of a fine-tuned signal exchange (2).
Nod factors consist of a chitooligosaccharide backbone. An acyl group replaces the N-acetyl group at the nonreducingterminal residue, and strain-specific substitutions modify the two terminal residues. NodA, NodB, and NodC proteins are required for the synthesis of the acylated Nod factor backbone, whereas most other Nod proteins are required for specific modifications (6,7). NodL of Rhizobium leguminosarum is an acetyltransferase (8), and nodZ encodes an ␣-1,6-fucosyltransferase (9 -11). Sulfation of Nod factors produced by Sinorhizobium meliloti is carried out by the sulfotransferase NodH that uses 3Ј-phosphoadenosine 5Ј-phosphosulfate, synthesized by the NodPQ complex (12,13).
Nod factors are indispensable for nodulation and cause multiple effects. Plateau-like increases in intracellular free calcium (14), plasma membrane potential changes (15), and actin cytoskeleton disintegrations, which are a prelude for root hair deformation and curling (16), occur within seconds to minutes after Nod factor application. Nod factors also induce the formation of root hairs and preinfection threads, the division of cortical cells, and the expression of plant genes involved in nodule development (17).
In a few interactions, structural requirements for Nod factortriggered responses have been studied. Stokkermans et al. (18), for example, showed that pentameric Nod factors produced by Bradyrhizobium japonicum require a 2-O-methylfucosyl group at the reducing-terminal residue to induce root hair deformation and the formation of nodule primordia on Glycine soja. A nodL and a nodFE mutant of S. meliloti, which synthesize nonacetylated Nod factors and Nod factors with altered acyl chains, respectively, have a reduced capacity to induce infection threads in alfalfa (3). Nonsulfated Nod factors produced by a S. meliloti nodH mutant were biologically inactive on alfalfa (19).
Other studies have dealt with Nod factor stability. Plant chitinases produced during nodule development can degrade Nod factors (20 -24). However, particular decorations contribute to Nod factor stability. Sulfated Nod factors of S. meliloti are more resistant to degradation than the corresponding nonsulfated molecules (21).
Here, the influence of Nod factor amounts and substitutions on nodulation efficiency was investigated in the symbiosis between Azorhizobium caulinodans ORS571 and the tropical legume Sesbania rostrata. S. rostrata can be nodulated on the stem and on the root at adventitious rootlets and at the bases of lateral roots, respectively. The Nod factors of A. caulinodans are mainly pentamers that are vaccenoylated (C18:1), palmitoylated (C16:0), or stearoylated (C18:0). The nonreducing-terminal residue carries an N-methyl and a 6-O-carbamoyl group, and the reducing-terminal residue may be substituted with a D-arabinosyl group, an L-fucosyl group, or both (25,26) (Fig. 1). The latter product with a vaccenoyl chain for example, will be referred to as NodARc-V(Carb, 1 Me,C18:1,Ara,Fuc) (17).
Azorhizobial nodulation genes for the synthesis and secretion of Nod factors are clustered in the nodABCSUIJZnoeC operon and the nolK operon (Fig. 1). A collection of bacterial mutants producing well characterized altered Nod factors was used in stem and root nodulation assays. Additionally, purified Nod factors were assayed for biological activity and for stability against degradation. Together, these approaches provide new information about the relative importance of Nod factor amounts and substitutions for the overall effectiveness of the signaling molecules in nodulation at adventitious and lateral root bases of S. rostrata.
Molecular and Genetic Methods-To construct the double mutants ORS571-1.31U-⍀K and ORS571-1.11Z-⍀K, the plasmid pRG901⍀B (Table I) was mobilized into ORS571-1.31U and ORS571-1.11Z, respectively, by triparental mating using the helper plasmid pRK2013 (Table  I). Screening for Tc-sensitive and Sp-and Sm-resistant colonies resulted in the isolation of putative double recombinants. Candidates were checked by Southern hybridization (data not shown). Plasmids pBBRNU and pMP1060 were introduced into ORS571-1.31U-⍀K and plasmid pRTHPQ into ORS571-1.31U-⍀K and ORS571-1.11Z-⍀K by triparental mating (Table I).
Plant Growth and Nodulation Assays-S. rostrata seeds were surface-sterilized and germinated (31). For stem nodulation, three plants were grown per pot. After 4 weeks, a part of the stem containing approximately 100 adventitious rootlets was inoculated with A. caulinodans strains (32). For root nodulation, plants were grown in glass tubes (33). Overnight azorhizobial cultures were washed with sterile water, and approximately 10 8 cells were added to the roots, 1 week after the seedlings were transferred to tubes. Mature stem and root nodules were counted 2 weeks after inoculation. Three independent stem and root nodulation experiments were carried out with 20 plants per strain. A statistical analysis was performed using the t test. Two averages were considered as significantly different when p Ͻ 0.05.
Root Hair Formation Assay-To roots, aseptically grown in tubes, pure Nod factors were applied at concentrations of 10 Ϫ8 , 10 Ϫ9 , 10 Ϫ10 , and 10 Ϫ11 M. Five days later, roots were cleared and stained (25), and the number of dense groups of root hairs per plant was counted under a binocular (Leitz, Heerbrugg, Switzerland). This assay was done twice, independently, on four plants per treatment and per Nod factor concentration. The averages were considered as significantly different when p Ͻ 0.2.
Nod Factor Degradation Assay-Plant material was harvested and immediately frozen in liquid nitrogen, ground, and suspended in sodium acetate buffer (20 mM sodium acetate, pH 5, 0.5% (v/v) dimethyl sulfoxide, 10 g/ml acetylated bovine serum albumin (New England Biolabs, Beverly, MA)). The protein concentration was measured using the D c protein assay kit (Bio-Rad). 14 C-Labeled Nod factors (500 cpm) were incubated at 37°C in a total volume of 150 l with crude extracts of uninoculated adventitious rootlets (100 g of protein) for 5, 10, 15, and 20 h or with crude extracts of 3-day-old developing nodules (7 g of protein) for 15, 30, 45, and 60 min. As controls, Nod factors were mixed with extracts and immediately frozen, or with buffer and incubated for 24 h at 37°C. After incubations, 350 l of water was added and the samples were shaken for 3 h with 500 l of water-saturated n-butanol. The organic phase was kept aside, and the extraction was repeated. Both n-butanol phases were combined and twice washed with 50 l of water, dried, resuspended in 70% methanol, and analyzed by RP-TLC. Nod factor spots were visualized and quantified by PhosphorImager and ImageQuant™ software.

Bacterial Strains with Altered Nod Factor Production-To
complete the set of available A. caulinodans mutants that are affected in Nod factor synthesis (Table II), several additional strains were constructed and characterized and the Nod factor population of the previously described nodS mutant ORS571-1.59S (Table I) was analyzed (Fig. 1). The azorhizobial nodS gene encodes a methyltransferase that is involved in Nod factor methylation (29,34). Nod factors were prepared from 50 liters of ORS571-1.59S culture. The vaccenoylated and stearoylated Nod factors were separated on RP-HPLC and analyzed by LSIMS associated with CID (Table III). Because protonated NodARc-VCarb/C18:0,Fuc) and NodARc-V(Carb,Me,C18: 0,Ara) were isobaric at m/z ϭ 1447, CID-MS was performed on the parental ion with m/z ϭ 1447 (Table III, Fig. 2A). Two series of four characteristic fragments were observed. A loss of D-arabinose and GlcNAc-OH led to a fragment with m/z 1094 and three other consecutive losses of GlcNAc to fragments with m/z 891, 688, and 485 ( Fig. 2A). The latter fragment (m/z ϭ 485) corresponded to the mass of a nonreducing-terminal residue substituted with stearic acid, an N-methyl, and a 6-Ocarbamoyl group (25,26). The second series, with m/z 1080, 877, 674, and 471, differed from the first series in that the m/z value of all the fragments was 14 units lower ( Fig. 2A), demonstrating that this Nod factor was fucosylated and nonmethylated. Thus, the nodS mutation affected, but did not completely abolish, Nod factor methylation. Moreover, the mutation was nonpolar and half of the 22% Nod factors produced by the nodS mutant (Table III) were "wild-type" Nod factors (data not shown).
The mutant strains ORS571-1.31U and ORS571-1.11Z contained a polar Tn5 insertion in nodU and in nodZ, respectively ( Fig. 1). Although the expression of the nodZ gene was knocked out in both strains, a fraction of Nod factors was still fucosylated (Table II), impeding conclusions about Nod factor glycosylations and nodulation. Knock out of the nolK gene, on the other hand, led to a complete absence of fucosylation (10) (Table II). Therefore, an ⍀-cassette was introduced into the nolK gene of ORS571-1.31U and ORS571-1.11Z (see "Experimental Procedures"). The LSIMS spectra of the vaccenoylated and palmitoylated Nod factors produced by ORS571-1.31U-⍀K and a Derivatives of A. caulinodans ORS571 are described in Table I. b Nod factor substitutions correspond to those mentioned in Fig. 1B. c The presence of a substitution on Nod factors synthesized by A. caulinodans derivatives is indicated as follows: ϩ, present as in the wild-type strain; Ϯ, only a minor fraction of the Nod factors contained this modification; Ϫ, Nod factors carrying this modification could not be detected. ORS571-1.11Z-⍀K are presented in Fig. 2 (B and C). Mass spectra of stearoylated Nod factors were similar to those of vaccenoylated Nod factors, except that all masses were 2 units higher because of the saturation of the fatty acid (data not shown). The mass spectra of Nod factors synthesized by ORS571-1.31U-⍀K showed two products with m/z 1270 and 1244, corresponding to N-methylated Nod factors (Fig. 2B) that carried a vaccenoyl and a palmitoyl group, respectively (35). None of the Nod factors produced by ORS571-1.31U-⍀K were carbamoylated. No fucosylated, arabinosylated, or fucosylated and arabinosylated Nod factors were detected (absence of products at m/z 1416, 1402, and 1548, respectively; Fig. 2B and data not shown). The first glycosidic cleavage resulted in an ion with m/z 1049, due to the loss of GlcNAc-OH. Three consecutive losses of GlcNAc led to fragments with m/z 846, 643, and 440. The mass of the latter fragment corresponded to a vaccenoylated and N-methylated nonreducing-terminal residue (35). Nod factors of ORS571-1.11Z-⍀K differed from those of ORS571-1.31U-⍀K in that their masses were 43 units higher (Fig. 2C). The two products of ORS571-1.11Z-⍀K with m/z 1313 and 1287 were similar to those with m/z 1270 and 1244 of ORS571-1.31U-⍀K, but contained an additional 6-O-carbamoyl group because of an intact nodU gene (Fig. 1). No glycosylated Nod factors could be detected (absence of products with m/z 1459, 1445, and 1591 (Ref. 10); Fig. 2C and data not shown). Cleavage of glycosidic bonds of the parental ion with m/z 1313 resulted in characteristic fragments with m/z 1092, 889, 686, and 483 (Fig. 2C). The mass of the latter fragment corresponded to a nonreducing-terminal residue substituted with a vaccenoyl group, an N-methyl group, and a 6-O-carbamoyl group (26). Upon introduction of pBBRNU (Table I) into ORS571-1.31U-⍀K, Nod factors were identical in structure to those produced by ORS571-1.11Z-⍀K (data not shown).
To address the question of the role of the 6-O-carbamoyl and the L-fucosyl group in nodulation, these groups were replaced by a 6-O-acetyl group and a 6-O-sulfate ester, respectively.
From RP-TLC analysis, Nod factors of ORS571-1.31U-⍀K(pMP1060) ( Table I), which expressed the nodL gene of R. leguminosarum, were found to be more hydrophobic than those of ORS571-1.31U-⍀K(pBBRNU) (data not shown), but production and secretion levels were similar to those of the parental strain ORS571-1.31U-⍀K (Table IV). ORS571-1.31U-⍀K(pRTHPQ) and ORS571-1.11Z-⍀K(pRTHPQ) ( Table I), which expressed the nodHPQ genes of Rhizobium tropici, produced Nod factors that were more hydrophilic than those of the respective parental strains (data not shown), but production and secretion levels were the same as those of strains ORS571-1.31U-⍀K and ORS571-1.11Z-⍀K (Table IV). LSIMS analysis had previously shown that introduction of pRTHPQ into ORS571 led to Nod factors with a 6-O-sulfate ester at the reducing-terminal residue (36).
Nod factor concentrations may influence plant gene expression (37) and the relative importance of specific modifications for biological activity (38,39). Therefore, Nod factor secretion and production (the total amount of secreted plus bacteriaassociated Nod factors) were determined for all A. caulinodans derivatives used in this study (see "Experimental Procedures") ( Table IV). The presence of the NodIJ secretion system had only a minor influence on Nod factor secretion (Table IV), which might be due to the overnight growth of cultures prior to Nod factor preparation (40). Mutations in nodS and nodU strongly diminished the total Nod factor production, whereas introduction of pBBRNU into nodU mutants led to a 2-fold higher production compared with that of the wild type (Table  IV).
Symbiotic Properties of nod Gene Mutants-Stem and root nodulation experiments were performed with the mutant strains and their complemented derivatives listed in Table I. Two weeks after inoculation, mature nodules were counted. Their relative average number per plant is presented in Fig. 3. Regarding the root nodulation efficiencies, three major groups of strains can be distinguished. The first group (a) of most  (26). g ND, not determined. Nod factors were not analyzed by CID-MS when they were also produced by the wild-type strain or when the available quantity was too low. efficient strains contained the wild type, the nonpolar nodS mutant ORS571-1.59S, which was affected in degree of Nod factor methylation as well as in yield (22% total yield of the wild type), but still produced a fraction (half of this population) wild-type Nod factors, and the strain ORS571-4.2K, a mutant that cannot synthesize fucosylated Nod factors. The second group of somewhat less efficient nodulators (60 -70% (b) and 50 -60% (c) of the wild type), basically consisted of strains that produced nonarabinosylated Nod factors with normal (ORS571-1.2), reduced (ORS571-1.39J, ORS571-1.11Z), or no (ORS571-1.11Z-⍀K) fucosylation (Fig. 3). No major effects of Nod factor levels on nodulation efficiency were noticed (Fig. 3). For example, no difference in nodulation efficiency could be observed among ORS571-1.39J, ORS571-1.2, and ORS571-1.11Z, although the two former strains produced approximately 3-fold more Nod factors than the latter strain (Fig. 3). Fig. 3 illustrates as well that the absence of a D-arabinosyl group has a more severe influence than the absence of an L-fucosyl group. The third group, with a more strongly reduced number of root nodules (d; less than 20%) (Fig. 3) contained ORS571-1.31U and ORS571-1.31U-⍀K strains, which do not produce carbamoylated Nod factors and which are affected in the synthesis of glycosylated Nod factors (Fig. 3). Introduction of pBBRNU into the nodU mutants partially restored the root nodulation efficiency (Fig. 3)   a Secreted (medium) and nonsecreted (cells) Nod factors, which were prepared from overnight-grown cultures, are presented as percentages, the sum of which is equal to 100%.
b The total amount of produced Nod factors is indicated with a number relative to the amount of Nod factors produced by ORS571, which is set at 100.
c The values are an average of two independent experiments.
greater impact of the absence of an L-fucosyl group and, in general, a stronger requirement of the glycosylations and the carbamoylation (i.e. for all modifications studied) for efficient nodulation were noticed. Replacement of the 6-O-carbamoyl group by a 6-O-acetyl group as in strain ORS571-1.31U-⍀K(pMP1060) did not affect root nodulation, whereas stem nodulation was reduced compared with ORS571-1.31U-⍀K(pBBRNU) (Fig. 3). Introduction of pRTHPQ into ORS571-1.31U-⍀K and ORS571-1.11Z-⍀K did not alter stem or root nodulation efficiency (Fig. 3).

Biological Activity of Purified Nod Factors-Purified
Nod factors were assayed for their capacity to induce root hair formation at the bases of lateral roots of S. rostrata (25,26). Five different pentameric Nod factors (Fig. 4) were applied at concentrations of 10 Ϫ8 , 10 Ϫ9 , 10 Ϫ10 , and 10 Ϫ11 M. The average number of dense groups of root hairs per plant is presented in Fig. 4. Because no differences in biological activity had been observed previously between vaccenoylated and stearoylated Nod factors (25), only the former type was tested.
Nod Factor Substitutions and Stability against Degradation-An in vitro assay was used to investigate the role of substitutions in protection against Nod factor-degrading enzyme activities of S. rostrata. Purified 14 C-labeled pentameric vaccenoylated Nod factors were incubated with crude plant extracts and assayed by semiquantitative RP-TLC (see "Experimental Procedures"). The degradation assays were carried out at pH 5 to monitor the activity of extracellular hydrolases with a low pH optimum (24), 2 and to mimic the in vivo extracellular pH (41,42).
When extracts of uninoculated adventitious rootlets were used, the RP-TLC pattern contained four different spots (data not shown), corresponding to pentameric Nod factors and tetra-, tri-, and dimeric acylated degradation products as described (23). NodARc-V(Carb,Me,C18:1,Ara,Fuc) was the most stable molecule with a half-life time of more than 20 h (Fig. 5A) (Fig. 5B). The Nod factors were all degraded to yield dimeric acylated degradation products (data not shown). When Nod factors were incubated for 24 h in the absence of extracts, no degradation products could be detected (data not shown).

DISCUSSION
The Importance of Nod Factor Levels for Nodulation Efficiency-Loss of the nodS or the nodU gene activity diminished severely Nod factor production (1.59S, 1.31U, and 1.31U-⍀K in Fig. 3). These results are similar to reports on decreases in Nod factor production of a nodSU mutant of Rhizobium sp. NGR234 (43), and nodL mutants of S. meliloti (3) and R. leguminosarum (44,45). Introduction of pBBRNU in the nodU mutant strains ORS571-1.31U or ORS571-1.31U-⍀K, not only restored Nod factor carbamoylation but led to a Nod factor production that was 2-fold higher than wild type (Fig. 3). This phenomenon was not observed when the nodL gene from R. leguminosarum was expressed in the nodU mutants. Acetylated Nod factors were produced but still at reduced levels. NodU may act relatively early in the Nod factor biosynthesis pathway. NodS is known to methylate deacetylated chitopentameric substrates (34). Enzymes downstream in the pathway, such as the acyltransferase NodA (46, 47) of A. caulinodans may be specific for methylated 2 M. Schultze, personal communication.

FIG. 3. Stem and root nodulation of S. rostrata by A. caulinodans and derivatives.
A, relative average number of mature root (black bars) and stem (white bars) nodules per plant induced by A. caulinodans and derivatives. The nodule number induced by the wildtype strain is set at 100%, which corresponded to 20 root or 88 stem nodules. Nodules were counted 2 weeks after inoculation. Two averages that are not significantly different (p Ն 0.05) are indicated with the same letter (a-d for root nodulation and aЈ-hЈ for stem nodulation). The relative Nod factor production (see Table IV) is presented under the bars. B, concise description of the Nod factors produced by the different strains used in the nodulation assays. Substitutions at the nonreducing-and the reducing-terminal residue are indicated. Underlined modifications indicate that the relative amount of this Nod factor fraction is lower than that in the wild-type strain and "Ϫ" that the substitution is absent in the Nod factor population of that particular Azorhizobium derivative. Mutant strains are indicated with their name as in Table I, but without the prefix "ORS571."

FIG. 4. Nod factor-induced root hair formation on S. rostrata.
The average value per plant of dense groups of root hairs formed at the basis of lateral roots upon Nod factor application is plotted in function of the Nod factor concentration and the Nod factor type. Root hair groups were counted 5 days after application. For Nod factor nomenclature, see the introduction. Two averages that are not significantly different (p Ն 0.2) are indicated with the same letter (a-m). and/or carbamoylated (but not for acetylated) substrates. It is equally possible that Nod factors are assembled in a multienzyme complex (7). When particular key Nod proteins are absent (NodS, NodU), or when a Nod protein is exchanged for a nonnative one such as NodL, the complex ("Nod factor factory") may be disturbed and intermediate metabolites could leak out in the cytoplasm, or the overall complex might be less stable or less active.
What can be concluded about the impact of Nod factor production levels on the nodulation phenotype? The drastic reduction of Nod factors produced by nodS and nodU mutants has no or only minor consequences for nodulation efficiency. Several examples are given in Fig. 3 of strains producing Nod factors with identical modifications but altered levels and very similar nodulation efficiencies. The wild type and the nodS mutant are good examples. The nodS mutant produces only 10% wild-type Nod factors yet nodulates as efficiently. The higher Nod factor levels of the strain ORS571-1.31U(pBBRNU) do not alter nodulation efficiency in comparison to ORS571-1.39J and ORS571-1.11Z, whereas all three strains produce Nod factor with the same modifications. Only in the case of ORS571-1.11Z-⍀K and ORS571-1.31U-⍀K(pBBRNU), which produce identical Nod factors, a higher production seems to influence positively stem nodulation but negatively root nodulation. Predictive calculations of Nod factor production at the initial site of bacterial colonization during nodulation assays, suggested that, for any mutant tested, super-saturation levels may be reached. 3 Moreover, an auto-regulatory mechanism (48) could also strongly modulate any expected concentration dependence of Nod factor action in planta. The overall conclusion from our experimental observations (important for the subsequent discussion of the role of specific Nod factor modifications in nodulation) is that differences in Nod factor amounts had no great impact on nodulation efficiencies.
Modifications at the Reducing-terminal Residue-The results show that for root as well as for stem nodulation, the presence of an arabinosyl group is more important than that of a fucosyl group. On roots, however, even in the absence of all glycosylations, approximately 50% nodulation still occurred. For stem nodulation, requirements were more stringent and in the absence of glycosylations the relative average nodule number dropped to a mere 20%. The positive role of the glycosylations for efficient nodulation of S. rostrata is supported by the observation that Sinorhizobium saheli ORS611 and Sinorhizobium teranga bv. sesbaniae ORS604, two other symbionts of S. rostrata (49), produce Nod factors identical to those of A. caulinodans (50). Replacement of the 6-O-L-fucosyl group by a 6-O-sulfate ester interfered with nodulation efficiency without changing the Nod factor stability. The presence of the 6-Osulfate ester did not actively hinder Nod factor function, because introduction of the nodHPQ genes did not decrease severely the nodulation efficiency compared with that of the parental mutant strain (Fig. 3). This situation is different from that in R. leguminosarum bv. viciae where the nodulation of the host plant Vicia sativa was dramatically reduced upon expression of the S. meliloti nodHPQ genes (51). A bio-assay led to very analogous conclusions concerning the impact of Nod factor modifications. The influence of substitutions at the reducing-terminal residue on root hair induction capacity and on nodulation efficiency may be partly explained by the greater stability of glycosylated Nod factors against degradation by hydrolytic enzymes of S. rostrata. Nod factor-degrading enzymes are present in uninfected, stem-located, adventitious rootlets and in developing nodules (23). The degradation kinetics showed an inverse relationship between the number of glycosylations at the reducing-terminal residue and the degradation rate when Nod factors were incubated with extracts of uninfected adventitious rootlets. Tetrameric Nod factors of S. meliloti were more stable against degradation by root hydrolytic enzymes when they carried a sulfate ester at the reducingterminal residue (21). Recently, Ovtsyna et al. (52) proposed that the 6-O-acetyl group at the reducing-terminal residue of Nod factors of R. leguminosarum bv. viciae may play a role in increasing the stability toward Afghan pea chitinases, but not in a specific receptor-ligand interaction, because a fucosyl group could functionally replace the structurally different 6-Oacetyl group. However, in the A. caulinodans-S. rostrata interaction, modifications at the reducing-terminal residue seem to have a dual function: protection of the Nod factor against degradation as well as recognition. Indeed, a 6-O-L-fucosyl group can be replaced by a 6-O-sulfate ester for protection, but not for nodulation (compare ORS571-1.2 and ORS571-1.11Z-⍀K(pRTHPQ) in Fig. 3), whereas the D-arabinosyl group protects the Nod factor against degradation to the same extent as the 6-O-L-fucosyl group, but is more important for nodulation (compare ORS571-4.2K to ORS571-1.2 in Fig. 3). When extracts of developing nodules were used, all Nod factors tested were quickly degraded, suggesting that genes encoding hydrolytic enzymes are up-regulated. Indeed, Goormachtig et al. (23) showed that a Nod factor-degrading chitinase is produced during stem nodule development on S. rostrata. These findings resemble those of Staehelin et al. (22), who demonstrated that Nod factors of S. meliloti induced the production of Medicago sativa chitinases that rapidly degrade S. meliloti Nod factors. The biochemical characterization of the azorhizobial chitinase and its substrate specificity are subject of a parallel study in our laboratory.
Modifications at the Nonreducing-terminal Residue-The presence of the carbamoyl group contributes to nodulation efficiency, carbamoylated Nod factors have a higher biological activity than their noncarbamoylated counterparts, and the carbamoyl group enhances Nod factor stability in the in vitro Nod factor degradation assay with extracts of uninfected adventitious rootlets (Figs. 3-5). Staehelin et al. (22) showed that a 6-O-acetyl group at the nonreducing-terminal residue of S. meliloti Nod factors increased their stability against degrada-3 W. D'Haeze and M. Holsters, unpublished data. tion by an N-acyl chitobiose-forming Nod factor hydrolase. The impact of the carbamoyl group in Nod factor protection is comparable to that of the fucosyl group at the reducing-terminal residue (Fig. 5); however, strains that are affected in carbamoylation are more severely altered in their nodulation efficiency (Fig. 3), suggesting an additional role of the carbamoyl group in recognition. Nevertheless, the carbamoyl group can be replaced by an acetyl group for root nodulation and partially also for stem nodulation. Perhaps the keto function, which is common for both groups, or a blocked C-6-hydroxyl group at the nonreducing-terminal residue contribute equally to recognition. As discussed above, the reduced Nod factor production by nodU mutant strains is not the major cause of the reduced nodulation efficiencies. Similar to the azorhizobial nodU mutant, a nodL mutant of S. meliloti was affected in its Nod factor production and nodulation. An increased production did not resolve the symbiotic defects, indicating that the defects were due to the absence of the acetyl group rather than to the diminished Nod factor production (3).
As for methylation, we cannot reach conclusions about the necessity of the methyl group for nodulation as strain ORS571-1.59S still produced 10% wild-type Nod factors. A lack of polarity of a Tn5 insertion has also been observed, for instance, for certain chromosomal Tn5 insertion mutants in the E. coli lac operon (53). The fact that a mutation in nodS did not completely abolish methylation was seen in Rhizobium sp. NGR234 as well (43). The residual methyltransferase activity may be due to a yet unknown Nod protein or to a household methyltransferase. The location of the insertion at the carboxyl-terminal part of the nodS gene does probably not cause a partial inactivation of the NodS protein because the Rhizobium sp. NRG234 nodS mutant did carry an ⍀-cassette in the middle of the nodS gene and still produced a fraction of methylated Nod factors (43).
Nod Factor Requirements for Stem and Root Nodulation-Although the overall tendencies regarding Nod factor requirements were quite similar for both stem and root nodulation, the structural requirements of Nod factors for root nodulation were overall less stringent than for stem nodulation. Root nodulation takes place in an aqueous environment; stem nodulation occurs under aerial conditions that are probably more restrictive, regarding inoculation and bacterial growth conditions, perhaps explaining the lower efficiency of several mutants on the stem.
No macroscopic differences were seen between nodules induced by the wild-type strain and those induced by any of the mutant strains, either on the stem or on the root (data not shown), neither were other phenotypes, such as the formation of pseudo-nodules, observed. Furthermore, with none of the A. caulinodans mutants that were studied in the Sesbania interaction, an uncoupling of epidermal and cortical responses took place, as described by Ardourel et al. (3) in the S. melilotialfalfa interaction. These results are thus more in line with observations by Stokkermans et al. (18), who showed that in G. soja natural and synthetic Nod factors that were biologically active induced both root hair deformations and nodule initiations.
Nod factor decorations are major contributors to host specificity by interaction with a specific receptor (3)(4)(5). This hypothesis is supported by recent indications that heterotrimeric GTP-binding regulator proteins may mediate Nod factor signal transduction mechanisms in the S. meliloti-Medicago interaction (54). In addition, the Nod factor substitutions may allow efficient nodulation by protecting against degradation (21). Perhaps Nod factor modifications evolved to protect an elementary lipochitooligosaccharide from degradation by plant-de-rived hydrolytic enzymes rendering them more effective. Subsequently, receptors might have coevolved to bind more efficiently to these modified lipochitooligosaccharides.