INTRODUCTION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

When rhizobia (bacteria belonging to the genera Rhizobium, Azorhizobium, Sinorhizobium, Mesorhizobium, and Bradyrhizobium) meet a compatible legume host, root nodules are formed. In these specialized organs, bacteria fix nitrogen that is subsequently assimilated by the plant.

A highly specific signal exchange triggers development of a functional nodule. Flavonoid compounds, exuded from legume roots, induce the production and secretion of the bacterial Nod factors, decorated lipochitooligosaccharides that, in turn, activate the nodulation program in the plant (for review, see Bladergroen and Spaink, 1998).

For a functional nodule to be established, coordinated gene expression in both symbionts is required. The products of plant genes that are expressed before the onset of nitrogen fixation, the early nodulins, are supposed to play a role in nodule ontogeny and bacterial invasion. Several early nodulin genes have been isolated from different legumes but for only one of them sound evidence is available for a specific function in nodulation (Hirsch and LaRue, 1997; Schauser et al., 1999).

In the symbiotic interaction between the tropical legume Sesbania rostrata and the bacterium Azorhizobium caulinodans, root and stem nodules are formed. The nodule development presents several special features (Tsien et al., 1983; Duhoux, 1984; Ndoye et al., 1994; Goormachtig et al., 1997). Root nodules develop at the lateral root bases (Ndoye et al., 1994), whereas stem nodules arise from dormant adventitious root primordia that are positioned in vertical rows along the stem. These dormant root primordia are an adaptation to waterlogging and turn into roots upon submergence. A. caulinodans invades the host via an active Nod factor-dependent process during which intercellular infection pockets (apoplastic accumulations of bacteria) are formed (D'Haeze et al., 1999). This entry mechanism occurs also on other, mostly tropical, legumes, such as Neptunia natans (Subba-Rao et al., 1995) but, so far, has only been extensively studied at the molecular level in the S. rostrata-A. caulinodans interaction. From the infection pockets, the bacteria are guided to the nodule primordia through infection threads that first grow intercellularly and then intracellularly. The nodule primordia develop from inner cortical cells and their initially indeterminate development becomes determinate because of an early arrest in meristematic activity (Goormachtig et al., 1997). Root nodules of S. rostrata have been shown to remain indeterminate under some environmental conditions (Fernández-López et al., 1998).

The stem nodulation is a sensitive system to unravel the early steps of nodule formation. Because of the high abundance of the dormant root primordia and the synchronized development into nodules, very specific plant material can be collected for differential display (Goormachtig et al., 1995). By using this approach, several genes have been isolated that are transiently induced during the early steps of stem nodulation. Some of these genes have been characterized in more detail, such as Srchi13, which codes for a class III chitinase that can degrade A. caulinodans Nod factors (Goormachtig et al., 1998a), and Srchr1, which is similar to genes encoding chalcone reductases (Goormachtig et al., 1999).

We describe the characterization of another gene, Srchi24, whose transcript levels are enhanced as early as 4 h after inoculation with A. caulinodans. In situ hybridization and immunolocalization show that Srchi24 transcripts and the corresponding proteins are located in the outermost cell layers of the developing nodules. The deduced amino acid sequence of Srchi24 is very similar to that of class III chitinases but it lacks an essential Glu residue in the catalytic domain. We demonstrate that Srchi24 lacks hydrolytic activity and we propose a function for this intriguing protein.

	RESULTS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

Sequence Analysis of Srchi24

Srchi24 was first isolated as a short PCR fragment (didi-24) from a differential display experiment in which the RNA pool of uninfected root primordia and young developing stem nodules of S. rostrata were compared (Goormachtig et al., 1995). The full-length cDNA clone was obtained by screening a cDNA library and by 5' RACE (see "Materials and Methods"). The full-length clone of Srchi24 was deposited in the EMBL database (accession no. Y12706). Database searches (EMBL and Swissprot) revealed a 56% to 75% similarity of Srchi24 with class III chitinase genes (Fig. 1), a pI value of 6.3, and the presence of the typical domains for family 18 chitinases (Henrissat, 1991; Henrissat and Bairoch, 1996). However, a Glu residue that is important for hydrolytic activity (Watanabe et al., 1993; Iseli-Gamboni et al., 1998) is changed into Lys (Fig. 1, asterisk). Sequencing of the corresponding genomic region confirmed that the Lys is a specific trait of Srchi24 and not a cloning artifact. An N-terminal signal peptide for endoplasmic reticulum targeting was predicted for the protein (von Heijne, 1983).

View larger version (57K): [in this window] [in a new window]   Figure 1.   Alignment of Srchi24 to class III chitinases. The amino acid sequence of Srchi24 (1; accession no. Y12706) is aligned with that of class III chitinases from S. rostrata (2; Srchi13, accession no. Z48671), Glycine max (3; accession no. AB000097), and Hevea brasiliensis (Hevamine A) (4; accession no. AJ007701). Black and gray-shaded amino acids are similar to four and three sequences, respectively. Asterisks indicate the Lys residue that replaced the catalytic Glu residue.

When compared with most other class III chitinases, Srchi24 had a C-terminal extension of 35 amino acids (Fig. 1). A similar extension was also present in Srchi13, a functional chitinase from S. rostrata (Goormachtig et al., 1998a), and in a chitinase isolated from soybean (Glycine max) seeds (Yeboah et al., 1998). A helical form for this extention was predicted from the analysis of secondary structure.

For DNA gel-blot analysis, S. rostrata genomic DNA was digested with XbaI, PstI, EcoRI, and HindIII, and hybridized with Srchi24 (see "Materials and Methods"). As shown in Figure 2, one main band was detected in each genomic digest, suggesting that Srchi24 is a single-copy gene. In the PstI and HindIII digestions, a second, much fainter band may reflect the occurrence of a weak homolog.

View larger version (46K): [in this window] [in a new window]   Figure 2.   DNA gel-blot analysis of Srchi24. The enzymes used for digestion of genomic DNA are indicated above each lane. The numbers on the right give the length of the fragments in kilobase pairs.

Srchi24 Is Transiently Expressed during Stem and Root Nodulation

The expression level of Srchi24 was analyzed during root and stem nodulation on S. rostrata. RNA was prepared from carefully excised adventitious root primordia before and 1, 2, 3, 4, 6, 7, and 17 d after inoculation with A. caulinodans, as well as from young roots before and 1, 2, 3, 4, 6, 8, 10, and 30 d after inoculation. The RNA samples were hybridized with an Srchi24 probe (see "Materials and Methods"). Srchi24 transcripts were abundantly present 1 d after stem inoculation (Fig. 3A). To investigate expression at earlier time points, reverse transcription (RT)-PCR was performed with specific primers (see "Materials and Methods"). RT-PCR (Fig. 3B) revealed that Srchi24 transcripts accumulated as early as 4 h after inoculation. A basal Srchi24 transcript level was present in uninfected root primordia. In later stages of stem nodule development, Srchi24 transcript levels remained relatively constant to drop again 7 d after infection (Fig. 3A). The transcript accumulation pattern during root nodulation was approximately the same, but less pronounced: Enhancement was seen 2 d after inoculation with A. caulinodans and the transcript level decreased to basal levels approximately 6 d after inoculation (Fig. 3C).

View larger version (68K): [in this window] [in a new window]   Figure 3.   Temporal expression pattern of Srchi24 during nodulation. A, Srchi24 transcript levels determined by RNA gel-blot analysis in uninfected root primordia () and in developing stem nodules on d 1, 2, 3, 4, 6, 7, and 17 after inoculation with A. caulinodans. To control for equal loading and blotting, the filter was stained with methylene blue (bottom). B, Srchi24 transcript levels determined by semiquantitative RT-PCR in uninfected root primordia () and in developing stem nodules after inoculation with A. caulinodans (2, 4, 8, and 12 h and 1, 2, 3, 4, and 5 d). Upper and lower, represent Srchi24 transcript and ubiquitin RNA levels, respectively. C, Srchi24 transcript levels determined by RNA gel-blot analysis in uninfected roots () and in developing root nodules after inoculation with A. caulinodans (1, 2, 3, 4, 6, 8, 10, and 30 d). To control equal loading, the filter was stained with methylene blue (bottom).

Srchi24 Induction Is Specific for Nodulating Bacteria and Is Induced by Nod Factors

The effect of two bacterial mutants, each differently deficient in nodulation, was tested on Srchi24 expression. Stems were inoculated with the mutant ORS571-V44, which carries a Tn5 insertion in the nodA gene and fails to produce Nod factors and to provoke nodulation (Van den Eede et al., 1987; Mergaert et al., 1993; D'Haeze et al., 1999). The second mutant, ORS571-X15, has a Tn5 insertion in a rhamnose biosynthesis locus and produces altered surface polysaccharides (D'Haeze et al., 1999). This mutant cannot invade the plant properly but produces normal Nod factors and induces pseudo-nodules on S. rostrata (D'Haeze et al., 1999). Stems were infected with both mutants and RNA was isolated after 3 and 6 d. As a control, RNA was prepared from uninfected root primordia and from primordia infected with wild-type A. caulinodans. The RNA samples were hybridized with an Srchi24 probe (see "Materials and Methods").

As shown in Figure 4A, ORS571-X15 provoked Srchi24 transcript accumulation (Fig. 4A; LSP3 and LSP6), although at a level lower than that of wild type (Fig. 4A; WT3 and WT6). In contrast, no Srchi24 induction was observed after application of ORS571-V44 (Fig. 4A; NODA3 and NODA6).

View larger version (78K): [in this window] [in a new window]   Figure 4.   Srchi24 expression upon inoculation with A. caulinodans mutants and purified Nod factors. A, Transcript levels determined by RNA gel-blot analysis in uninfected root primordia () and primordia inoculated with the wild-type A. caulinodans ORS571 (WT3 and WT6), the mutant ORS571-X14 (surface polysaccharide mutant defective in invasion; LPS3 and LPS6), and the mutant ORS571-V44 (NodA, deficient in Nod factor production NodA3; NodA6) on d 3 and 6 after inoculation. The filter was stained with methylene blue to confirm equal loading and blotting (bottom). B, Transcript level determined by RNA gel-blot analysis in root primordia treated for 8 h with 108 M purified Nod factors (NF) and with water (H2O) as control. As a positive control, the expression level is shown in root primordia 3 d after inoculation with A. caulinodans (N) by using the same experimental procedure as for the Nod factor incubation.

To rule out the possibility that the expression is affected indirectly by a Nod factor-dependent bacterial colonization (D'Haeze et al., 1999), the effect of purified Nod factors on Srchi24 expression in dormant root primordia was investigated. Due to drying out of the Nod factor solution, no molecular effect could be detected when the Nod factors were added to the stems in a way similar to that of the bacterial inoculants (data not shown). To keep the humidity high, an in vitro system was developed (see "Materials and Methods"). Nod factors (10⁸ M) were applied to the primordia of stem pieces that were incubated on Norris medium (see "Materials and Methods"). RNA was isolated 8 h after the onset of the treatment. As a control, stem pieces were treated identically with water. A higher Srchi24 transcript level was detected upon Nod factor application than upon water treatment (Fig. 4B). As a positive control, stem pieces were inoculated with A. caulinodans. The Srchi24 transcript level was enhanced in developing stem nodules 3 d after bacterial inoculation (Fig. 4B, N).

Srchi24 Is Expressed in Flowers and Seedlings

Plant chitinases have been best characterized as tools in defense against microbes (Collinge et al., 1993). However, chitinase gene expression has also been correlated with various plant developmental processes (Lotan et al., 1989; Neale et al., 1990; Hanfrey et al., 1996; Kragh et al., 1996). To check whether Srchi24 was expressed in other parts of the plant or at other developmental stages, RNA was prepared from seedlings, shoot apical regions, flowers, and leaves from uninfected plants and RT-PCR was performed with the same specific Srchi24 primers (see "Materials and Methods"). Srchi24 transcripts were detected in young seedlings (Fig. 5, lane 4) and, to a lower extent, in flowers (Fig. 5, lane 6) of S. rostrata. No expression was seen in shoot apical regions and leaves (Fig. 5, lanes 3 and 5).

View larger version (36K): [in this window] [in a new window]   Figure 5.   Srchi24 expression in several plant organs, in seedlings, and upon application of cytokinin. Srchi24 transcript levels determined by semiquantitative RT-PCR 3 d after inoculation with A. caulinodans in developing stem nodules (1), in uninfected root primordia (2), in shoot apices (3), in seedlings (4), in leaves (5), and in flowers (6). Srchi24 transcript levels were treated with 106 M 6-benzylaminopurine for 12 (9), 24 (10), 48 (11), and 72 (12) h. Lanes 7 and 8 correspond to the untreated root controls at 12 and 72 h, respectively. Upper and lower, Srchi24 transcript levels and ubiquitin RNA levels, respectively.

Srchi24 Is Induced by Cytokinin

Different early nodulin genes have been shown to be induced by cytokinin (Bauer et al., 1996; Silver et al., 1996; Coba de la Peña et al., 1997; van Rhijn et al., 1997; Fang and Hirsch, 1998). To test whether Srchi24 is induced by cytokinin, 10⁶ M of 6-benzylamino- purine was added to roots of young (7 to 10 d old) S. rostrata plants. Plant material for RNA preparation was taken 12, 24, 48, and 72 h after application (Fig. 5, lanes 9-12). As shown in Figure 5, the Srchi24 transcript level was induced (Fig. 5, compare lanes 7 and 8 with lanes 9-12). Similar results were obtained with kinetin (data not shown).

Srchi24 Does Not Degrade Lipochitooligosaccharide Molecules

Srchi24 lacks a Glu residue in the catalytic site (see above). To investigate the putative lack of chitinolytic activity, Srchi24 was produced in Escherichia coli as a fusion with the maltose-binding protein (MBP; see "Materials and Methods") and used in assays with the ¹⁴C-labeled peak I (PI) fraction of A. caulinodans Nod factors as substrate (PI corresponds to Nod factors that carry a C18:1 or a C16:0 fatty acid; Mergaert et al., 1993). The reaction products were analyzed by reversed-phase thin-layer chromatography. In Figure 6, lanes 2 and 3 are the reaction products after incubation of PI Nod factors with the enriched MBP and the MBP-Srchi24 protein samples, respectively. No degradation was observed with either protein fraction. As positive control, PI Nod factors were incubated with MBP-Srchi13, a protein sample that is known to have Nod factor-hydrolyzing activity (Goormachtig et al., 1998a; Fig. 6, lane 4). The same results were obtained when the peak II (PII) Nod factor fraction was used as a substrate in the assay (PII corresponds to Nod factors carrying a C18:0 fatty acid; Mergaert et al., 1993; data not shown). Assays were performed under a variety of incubation conditions (see "Materials and Methods"), but no Nod factor-degrading activity was observed for MBP-Srchi24. Also, 4-methylumbelliferyl -d-N,N',N''-triacetylchitotrioside was used as a substrate; again, MBP-Srchi13 showed hydrolytic activity but no activity was associated with MBP-Srchi24 (data not shown).

View larger version (11K): [in this window] [in a new window]   Figure 6.   Nod factor degradation assay. The reaction products were analyzed by reverse-phase thin-layer chromatography after overnight incubation of a 14C-labeled Nod factor fraction (peak I; Mergaert et al., 1993) with control (lane 1; identical treatment of Nod factors without protein fractions), enriched MBP fraction (lane 2), enriched MBP-Srchi24 (lane 3), and enriched MBP-Srchi13 fraction (lane 4). F, Solvent front; O, height at which samples are spotted; S, PI Nod factor substrate.

Srchi24 Transcripts and Corresponding Proteins Are Located in the Outermost Cell Layers of Developing Nodules

In situ hybridizations and immunolocalization were performed on sections of developing nodules (see "Materials and Methods"; Fig. 7). In uninfected adventitious root primordia, few Srchi24 transcripts were detected in cells of the outermost cortical cell layer (Fig. 7A, arrows). One day after infection with A. caulinodans, expression was enhanced in this cell layer and remained enhanced during later stages of stem nodule development (Fig. 7, B, C, E, and F). A magnification shows that Srchi24 was expressed in young peridermal cells (Fig. 7D).

View larger version (119K): [in this window] [in a new window]   Figure 7.   In situ localization of Srchi24 and immunolocalization of Srchi24 proteins. A through I, In situ localization of Srchi24 transcripts. Longitudinal sections (10 µm) through infected root primordia and developing stem and root nodules were hybridized with a 35S-labeled antisense RNA probe. Signals are seen as white and black dots in dark-field and bright-field micrographs, respectively. A, Dark-field micrograph of an uninfected root primordium. Arrows indicate cells containing Srchi24 transcripts. B, Dark-field micrograph of an infected root primordium, 1 d after inoculation with A. caulinodans. C, Dark-field micrograph of a developing stem nodule, 4 d after bacterial inoculation. D, Magnification of indicated rectangle in F. E and F, Bright-field micrograph of sections shown in B and C, respectively; asterisks indicate developing nodule parenchyma. G and H, Dark-field micrographs of a developing root and stem nodule 3 d after bacterial inoculation, respectively; arrows indicate cells flanking the infection pocket and containing Srchi24 transcripts. I, Bright-field micrograph of section shown in H. J through L, Immunolocalization of Srchi24 in stem nodules. Signals are seen as bright green (J and K) and bright yellowish green (L). J and K, Confocal fluorescence micrographs of a fluorescein isothiocyanate (FITC)-labeled longitudinal section of an infected root primordium 12 h after inoculation with A. caulinodans and of a developing stem nodule 4 d after bacterial infection, respectively; L, differential interference contrast image of a developing stem nodule 4 d after bacterial infection; overlay (bright yellowish green) with FITC-labeling. c, Cortex; f, fixation zone; fi, fissure; i, infection zone; ip, infection pocket; m, meristem; np, nodule primordium; pe, periderm; r, root. Bars = 100 µm.

Srchi24 transcripts were also detected in cortical cells surrounding infection pockets (Fig. 7, H and I). No signals were seen in the exterior layers of the dormant root meristem (data not shown). Expression disappeared in fully developed stem nodules (20 d; data not shown). From 3 d after infection, very weak signals were found in the developing stem nodule parenchyma (Fig. 7C, asterisks), probably due to some cross-hybridization with Srchi13 transcripts (Goormachtig et al., 1998a). In the empty stem nodules induced by the surface polysaccharide mutant strain, ORS571-X15, the Srchi24 transcripts were localized in a pattern similar to that of wild-type-induced nodules (data not shown).

The expression in root nodules was much lower than that in stem nodules (Fig. 7G) and was restricted to outer cortical cells in the neighborhood of infection pockets (Fig. 7G). The constitutive background detected by RNA gel-blot analysis in uninfected roots could not be localized by in situ hybridization (data not shown).

Immunolocalization revealed an identical pattern for the protein. Rabbit polyclonal antibodies were raised against the MBP-Srchi24 recombinant protein and their specificity was controlled by immunoblot analysis. Protein extracts were prepared from wild-type BY-2 cells (Nagata et al., 1992) and from BY-2 cells that had been transformed with a construct, which contained Srchi24 driven by a 35S promoter (W. Van de Velde and M. Holsters, unpublished data). The extracts were loaded on an SDS-polyacrylamide gel (see "Materials and Methods"). A specific band that corresponded to a protein of approximately 32 kD was observed in the lanes derived from the Srchi24-containing BY-2 cells (Fig. 8). A very faint band corresponding to a protein of approximately 45 kD was detected (Fig. 8). The 32-kD band was also seen in protein extracts of developing stem nodules (data not shown). The antibodies were used for fluorescent immunolocalization (see "Materials and Methods"). As shown in Figure 7J, 12 h after infection with A. caulinodans, strong signals were detected in the outermost cortical cell layer of the adventitious rootlets. This outer pattern remained present during the complete stem nodule development (Fig. 7, K and L). Proteins were also localized in cortical cells that surround the infection pockets (Fig. 7L). The amount of Srchi24 proteins diminished considerably 7 d after infection (data not shown). Approximately 10 d after infection, the Srchi24 signal had disappeared completely (data not shown).

View larger version (83K): [in this window] [in a new window]   Figure 8.   Immunoblot analysis of BY-2 cell lines. 1, Protein extracts from wild-type BY-2 cells; 2, protein extracts of BY-2 cells overexpressing Srchi24. M, Standard molecular mass given in kilodaltons.

	DISCUSSION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

A new early nodulin gene, Srchi24, has been isolated by differential display from developing stem nodules on S. rostrata. Srchi24 is highly similar to class III chitinases but lacks a Glu residue that is important for hydrolytic activity of chitinases (Watanabe et al., 1993; Andersen et al., 1997; Iseli-Gamboni et al., 1998). Expression analyses suggest that Srchi24 may play a role during nodulation.

RNA gel-blot and RT-PCR analyses showed that the Srchi24 transcript level increased already 4 h after A. caulinodans had been applied to the dormant root primordia and decreased approximately 1 week later when stem nodule development was completed. Srchi24 is not an early nodulin in the strict sense of the definition (van Kammen, 1984) because Srchi24 transcripts are also found in dormant root primordia, roots, flowers, and young seedlings. In fact, most early nodulin genes that have been isolated from other legumes (such as Enod40, Enod12, and Enod2) are expressed also in other plant parts, indicating that for nodule development and bacterial infection, the legumes do not contain a completely new set of genes, but isoforms of functions that play a role in other plant processes.

The basal Srchi24 transcript level in the adventitious root primordia is much lower than that in uninoculated roots probably because of the dormant immature nature of the structures. The Srchi24 transcript enhancement is less pronounced during root than stem nodule development perhaps because the dilution of the bacteria is higher in the root nodulation assay. Therefore, in roots, as shown by in situ hybridization, responses might be limited to more densely colonized sites. These observations explain the much higher transcript induction level during stem rather than during root nodulation. The question can be raised whether Srchi24 might have been detected had root nodulation been used for differential display. Hence, the isolation of Srchi24 is a clear example that stem nodulation is a very sensitive system for the isolation of early nodulation markers.

By using different A. caulinodans mutants, we showed that Srchi24 expression depends on nodulating bacteria. Infection with a surface polysaccharide-mutated A. caulinodans enhanced the Srchi24 transcript level, although to a lower extent than with the wild type. This mutant produces normal Nod factors, but is arrested during infection at the infection pocket level and induces empty nodules on S. rostrata (D'Haeze et al., 1999). On the other hand, a nodA mutant that does not produce Nod factors could not enhance the Srchi24 transcript level. The Srchi24 transcript level was enhanced after purified Nod factors had been added to the root primordia, indicating that the observed expression pattern is not an indirect effect of Nod factor-dependent bacterial colonization (D'Haeze et al., 1999). The basal expression level increased slightly under the in vitro conditions used for the Nod factor assay. This observation might be due to the root-associated Srchi24 expression, because after a few days of incubation the root primordia had elongated.

Srchi24 transcript level could be induced by cytokinin treatment of the S. rostrata roots, supporting the idea that nodulation may involve changes in hormone balances (Hirsch and Fang, 1994). Srchi24 is not the first S. rostrata nodulin gene that shows cytokinin-dependent induction. Another example is Enod2 (Dehio and de Bruijn, 1992; Silver et al., 1996). It is remarkable that Srchi24 and Enod2 have a common expression pattern in the outermost cell layers of the developing stem nodules (Goormachtig et al., 1998b). It would be interesting to compare whether the genes are regulated in the same way.

Database searches revealed the significant similarity of Srchi24 with class III chitinases. The most closely related proteins are Srchi13, another class III chitinase from S. rostrata, and a soybean chitinase that was isolated from seeds (Goormachtig et al., 1998a; Yeboah et al., 1998). Although Srchi24 and Srchi13 are induced during nodulation, they probably have completely different functions. First of all, they have a different expression pattern. Whereas Srchi24 is expressed in the outermost cortical cells of the developing nodules, Srchi13 is expressed in the infection center, in the uninfected cells of the central tissue, and in the nodule parenchyma (Goormachtig et al., 1998a). Moreover, Srchi13 shows Nod factor-degrading activity in contrast to Srchi24. The interesting question of whether and where the soybean chitinase is expressed during nodulation is still unanswered.

It is intriguing that these three proteins present an approximately 30-amino acid-long carboxyl-terminal extension that is absent in other class III chitinases. This extention is probably not cleaved off because the protein that is recognized by the antibodies in the immunoblot analysis had a size corresponding that of the deduced Srchi24 amino acid sequence without the N-terminal signal peptide. For a tobacco (Nicotiana tabacum) class I chitinase, a short carboxyl-terminal extension is necessary and sufficient to target the protein to the vacuoles (Neuhaus et al., 1991). The Srchi24 extension is not at all similar to that extension nor to other known domains in the databases. Secondary structure analysis predicts a helical structure for the extension, whereas a membrane anchor function could not be assigned by sequence analysis. Reporter fusions and transient expression experiments will clarify its function.

A remarkable difference with active chitinases is the absence in Srchi24 of a Glu residue that is a prerequisite for chitinolytic activity (Watanabe et al., 1993; Andersen et al., 1997; Iseli-Gamboni et al., 1998). This amino acid change is not a cloning artifact. Sequencing of the genomic region confirmed the presence of the mutation. Srchi24 has no hydrolytic activity, as in several other examples of non-hydrolytic chitinase-like proteins: concavalin B, a seed protein from Conavalia ensiformis (Schlesier et al., 1996), secreted glycoproteins from human (gp-39; Hakala et al., 1993) and Drosophila melanogaster (DS47; Kirkpatrick et al., 1995), and another D. melanogaster protein that might be a new growth factor (Kawamura et al., 1999). All these proteins lack both chitinolytic activity and a catalytic Glu residue.

The most obvious substrate for chitinases or chitinase-like proteins during nodulation are the rhizobial Nod factors that are the main signal molecules to trigger the onset of nodulation. Chitinases can control Nod factor activity by regulating the Nod factor concentration (Staehelin et al., 1994b). Several nodulation-induced chitinases or Nod factor-hydrolyzing activities have been isolated (Heidstra et al., 1994; Staehelin et al., 1994a, 1994b, 1995; Goormachtig et al., 1998a; Xie et al., 1999; Ovtsyna et al., 2000; Salzer et al., 2000). A hydrolytic function for Srchi24 is ruled out because of the altered catalytic site, but the protein can possibly bind specific chitin-like molecules. Mutation of the important Glu residue has changed a chitinase-catalytic domain into a chitin-binding domain (Iseli-Gamboni et al., 1998); thus, Srchi24 may have Nod factor-binding capacities and function as a Nod factor trap. Because Srchi24 was found in the outermost cortical cells of the developing nodule, Srchi24-expressing cells are probably in contact with the Nod factors. The subcellular location of Srchi24 is presently unknown and has to be determined by immuno-electron microscopy. No specific membrane domains have been found in Srchi24 by computer analyses.

If Srchi24 binds Nod factors, what is its function? Several lines of evidence suggest that Nod factors act via a receptor-mediated signal transduction mechanism (Stougaard, 2000). Although Srchi24 has no features of a receptor, it could trap Nod factor either to protect or to facilitate interaction with a receptor protein. No obvious protein/protein-binding domains have been found in the primary structure, but two-hybrid screens will be performed to address this question.

So far, two Nod factor-binding sites have been isolated from (NFBS1 and NFBS2, Medicago truncatula and Medicago varia, respectively) and one Nod factor-binding lectin has been reported for Dolichos biflorus (Bono et al., 1995; Niebel et al., 1997; Etzler et al., 1999; Gressent et al., 1999). At present, we do not know whether Srchi24 is similar to one of the two binding sites of alfalfa NFBS2 that needs an important Lys residue for binding. Accidental or not, this is exactly the amino acid change for the catalytic Glu residue in Srchi24. It would be interesting to check whether antibodies against Srchi24 recognize NFBS2.

In conclusion, Srchi24 is an early nodulin with an intriguing putative role during nodulation. Future experiments will be directed to understand its function in symbiosis by expressing sense and antisense constructs in transgenic roots of S. rostrata plants.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

Plant Material

Sesbania rostrata Brem seeds were sterilized as described by Goethals et al. (1989) and germinated on petri dishes in the dark at 28°C for 2 d. The seedlings were transferred to pots containing 50% potting soil and 50% sand. The plants were grown at 28°C with a 16-h light period for 2 to 3 months. The root primordia on the stem were infected by painting them with a bacterial inoculum, harvested by peeling off from the stem, and frozen in liquid nitrogen (Goormachtig et al., 1995). For the root nodulation, the plants were grown in tubes and inoculated as described by Fernández-López et al. (1998).

Nod Factor Application Assay

Stem pieces of S. rostrata were surface sterilized with 10% (v/v) bleach for 30 min, then washed five times with sterile distilled water. The stem pieces were incubated on Norris medium (Vincent, 1970), 3% (w/v) Suc, and 2.7 g L¹ phytagel (Sigma-Aldrich, St. Louis) in a growth chamber at 28°C with a 16-h light period. The next day, 10⁸ M purified S. rostrata Nod factors was added as droplets to the root primordia. This step was repeated every 2 h; the tissue was used for RNA preparation 8 h after the first Nod factor application. As a negative control, stems were treated with water, whereas as a positive control, stems were infected with Azorhizobium caulinodans and developing nodules were taken 3 d after inoculation.

Isolation of the Srchi24 cDNA Clone

Srchi24 was first isolated as a short 550-bp cDNA fragment (didi24) from a differential display experiment (Goormachtig et al., 1995). This fragment was used as a probe to screen 10⁵ plaques of a ZAP cDNA library of developing nodules (Sambrook et al., 1989; Goormachtig et al., 1997). ³²P-labeled probe was generated by the T7 Quick Prime kit (Amersham Pharmacia Biotech, Little Chalfont, UK). Phages from positive plaques were transferred to their corresponding plasmid form according to the manufacturer's protocol (Stratagene, La Jolla, CA). Plasmid DNA was prepared (Sambrook et al., 1989) and the plasmid with the largest insert was designated pdidi24fl54.

To isolate the full-length clone, 5' RACE reactions were performed (Gibco BRL, Gaithersburg, MD). The antisense primer 5'-AGGTAATTGGCGACTTGTTTTG-3' was used to obtain the specific cDNA. For the amplification, Vent polymerase (New England Biolabs, Beverly, MA) was used with the specific antisense primer 5'-CAGTTT-GATGCCTTTTTGTTGG-3' together with the G-rich sense primer provided by the manufacturer (Gibco BRL). Twenty-six cycles were performed at 95°C for 1 min, at 54°C for 1 min, and at 72°C for 2 min. The 5' RACE products were made blunt ended with the Klenow fragment of DNA polymerase (Roche Diagnostics, Brussels), according to standard protocols. The fragments were cloned in a pBluescript KS() vector (Stratagene). The complete cDNA was reconstructed in a PCR experiment with the sense primer 5'-ATGATGTCTTCCGAAAGG-CAAGC-3', the antisense primer 5'-CACAAGGCAAT-GTGCAATCTTTATTG-3', and Vent polymerase (New England Biolabs). PCR fragments were made blunt ended and cloned in pBluescript KS() (Stratagene).

DNA Gel-Blot Analysis

S. rostrata genomic DNA was extracted from young leaves as described by Dellaporta et al. (1983). DNA gel-blot hybridization was performed as described by Goormachtig et al. (1997).

RNA Gel-Blot Analysis

RNA was prepared as described by Goormachtig et al. (1995). Ten micrograms of RNA was separated on a 1% (w/v) agarose gel containing 2% (v/v) formaldehyde and transferred to Hybond-N filters (Amersham Pharmacia Biotech). Hybridization was carried out as described by Goormachtig et al. (1995). ³²P-labeled probes were made by the T7 Quick Prime kit (Amersham Pharmacia Biotech). Hybridized filters were exposed to films (Fuji, Tokyo) or analyzed by a phosphorimager (Amersham Pharmacia Biotech). To control equal loading and transfer of RNA, filters were stained with methylene blue (Sambrook et al., 1989).

RT-PCR

cDNA was synthesized from 3 µg of total RNA using the Superscript Preamplification System (Gibco BRL). A one-tenth volume (2 µL) was used subsequently in a PCR with 2 µL GeneAmp 10× PCR buffer (Perkin-Elmer Cetus, Norwalk, CT), 1.2 µL of 25 mM MgCl₂, 0.5 µL of 10 mM dNTPs, 1 unit of AmpliTaq DNA Polymerase (Perkin-Elmer Cetus), 200 ng of sense primer (5'-CGGTAACTTCGGAGGCCATTGTGGT-3'), and 200 ng of antisense primer (5'-CCCCCGTTCCAATAGCTCTGTCAAG-3'). After 20 cycles of amplification (30 s at 94°C, 30 s at 60°C, and 1 min at 72°C) samples were separated on a 1% (v/v) agarose gel, transferred to Hybond-N filters (Amersham Pharmacia Biotech), and hybridized with a ³²P-labeled Srchi24 probe at 65°C according to standard procedures (Sambrook et al., 1989). As a constitutive control, the same procedure was performed to amplify a ubiquitin fragment with 5'-GTATCCCACCAGACCAGCAGAGG-3' and 5'-CACAGACCCATTACACATCCACAAG-3' as sense and antisense primers, respectively. The results were visualized and analyzed with a phosphorimager (Amersham Pharmacia Biotech).

In Situ Hybridization

In situ hybridizations were performed as described by Goormachtig et al. (1997). ³⁵S-Labeled sense and antisense probes were generated with T7 and T3 polymerase, respectively (Gibco BRL). The plasmid pdidi24fl54 was digested with EcoRI or XhoI to produce antisense and sense probes, respectively. Photographs were taken with a Diaplan microscope (Leitz, Wetzlar, Germany) equipped with bright-field (signals seen as black dots) and dark-field (signals seen as white dots) options.

DNA Sequence Analysis

DNA sequence procedures were as described by Sanger et al. (1977). Data were assembled and analyzed with the GCG package (version 7; Genetics Computer Group, Madison, WI). Percentages of similarity were determined with the GAP program (Genetics Computer Group).

Srchi24 Synthesis in Escherichia coli

Plasmid pdidi24fl54 was digested with PstI and BamHI restriction enzymes and the insert was directionally cloned into pMal-c2 (Sambrook et al., 1989), yielding the clone pMal-c2-Srchi24. The production and enrichment of the MBP-Srchi24 fusion protein was performed as described by Goormachtig et al. (1998a). The MBP-Srchi24 protein was eluted in phosphate-buffered saline (PBS) medium (pH 7.4) or new column buffer, containing 20 mM Tris-Cl (pH 7.4), 200 mM NaCl, 1 mM EDTA, and 1 mM dithiothreitol (New England Biolabs).

Immunolocalization

Rabbit polyclonal antibodies were raised against enriched MBP-Srchi24 and tested for their titer and specificity by ELISA and protein gel-blot analysis. Uninfected root primordia and stem nodules at different developmental stages were embedded in paraffin. After the paraffin had been removed by soaking in Histoclear (National Diagnostics, Atlanta), 7-µm sections were immunolocalized according to the method of De Wilde et al. (1998). Sections were incubated with a 1:100 dilution of the different sera overnight at 4°C. After several washing steps, the sections were incubated for 2 h with a 1:80 dilution of an anti-rabbit IgG antibody (whole molecule), coupled to FITC (Sigma-Aldrich), at room temperature. Control immunolocalization was done at a 1:100 dilution of a rabbit anti-MBP serum (New England Biolabs). Sections were incubated with the primary or the secundary antisera only. FITC complexes were visualized with an Axioskop fluorescence microscope (Zeiss, Jena, Germany). Images were recorded with an LSM510 confocal scanning laser microscope system (Zeiss) that was equipped with an Axiovert 100 M inverted microscope (Zeiss) and a Plan-neofluar 40×/1.30 oil immersion lens (Zeiss). The 488-nm line of an Ar⁺-ion laser and the 543-nm line of a HeNe laser were used for excitation of fluorescein and recording of differential interference contrast images, respectively. The filter setup for FITC consisted of a primary dichroic mirror (488/543), a secondary dichroic mirror (545), and a BP 505-550 emission filter. A 488/543 dichroic mirror was used for the differential interference contrast images.

Enzymatic Activities

For Nod factor degradation, ¹⁴C-labeled pI (containing a C18:1 or C16:0 fatty acid) and pII (containing C18:0 fatty acid) Nod factor fractions (Mergaert et al., 1997) of A. caulinodans were prepared and purified as described by Mergaert et al. (1993). Radioactive incorporation was measured by scintillation counting. Different buffers with different pHs were tested: 50 mM sodium acetate buffers at pH 4.5, 5.2, and 6.3 and PBS buffer at pH 7.4. Incubation conditions were as follows: 50 µg of enriched MBP-Srchi13, MBP-Srchi24 fusion protein, or purified MBP2 protein (New England Biolabs) were mixed with ¹⁴C-labeled Nod factors (3,000 cpm) in 200 µL buffer and incubated at 37°C overnight. After n-butanol extraction, Nod factors and their degradation products (nonreducing end) were analyzed by reverse thin-layer chromatography as described by Mergaert et al. (1993). Visualization and analysis of the reaction products were done with a phosphorimager (Amersham Pharmacia Biotech).

Immunoblot Analysis

The proteins were extracted from 1-week-old (exponential growth fase) BY-2 cells by filtrating the cells on a filter (Whatman, Maidstone, UK) and grinding in liquid nitrogen in a chilled mortar. The resulting powder was resuspended in an equal volume of PBS buffer (pH 7.4) and centrifuged at 10,000g for 30 min at 4°C, separating the soluble proteins from the remainder. The protein content in the supernatant was determined by using the DC protein assay kit (Bio-Rad, Hercules, CA) with bovine serum albumin as a standard. For immunoblot analysis, 100 µg of total protein was precipitated by overnight incubation at 4°C with trichloric acid and subsequently washed with 100% (v/v) ethanol and ether. SDS-PAGE was performed according to Laemmli (1970) with 10% (w/v) polyacrylamide-resolving gels and 5% (w/v) polyacrylamide-stacking gels. Precipitated samples were resuspended in 1× Tris-Gly-SDS sample buffer (Novex, San Diego), supplemented with 100 mM -mercaptoethanol and boiled for 5 min. The molecular mass was estimated by co-electrophoresis of prestained low-range M_r standards (Bio-Rad). After SDS-PAGE, separation proteins were transferred by wet blotting onto an Immobilon-p membrane (Millipore, Bedford, MA). Blocking of the membrane was done overnight at 4°C in 3% (v/v) skimmed milk in PBS buffer. Srchi24 proteins were specifically detected by incubation of the membrane with a rabbit polyclonal antibody raised against the recombinant MBP-Srchi24 fusion protein (see above). After washing, membranes were incubated with a secondary antibody, coupled to alkaline phosphatase, and revealed with nitroblue tetrazolium and 5-bromo-4-chloro-3-indolyl phosphate (Bio-Rad).