Azospirillum brasilense, a nitrogen-fixing bacterium found in the rhizosphere of various grass species, was investigated to establish the effect on plant growth of growth substances produced by the bacteria. Thin-layer chromatography, highpressure liquid chromatography, and bioassay were used to separate and identify plant growth substances produced by the bacteria in liquid culture. Indole acetic acid and indole lactic acid were produced by A. brasilense from tryptophan. Indole acetic acid production increased with increasing tryptophan concentration from 1 to 100 ,ig/ml. Indole acetic acid concentration also increased with the age of the culture until bacteria reached the stationary phase. Shaking favored the production of indole acetic acid, especially in a medium containing nitrogen. A small but biologically significant amount of gibberellin was detected in the culture medium. Also at least three cytokinin-like substances, equivalent to about 0.001 ,tg of kinetin per ml, were present. The morphology of pearl millet roots changed when plants in solution culture were inoculated. The number of lateral roots was increased, and all lateral roots were densely covered with root hairs. Experiments with pure plant hormones showed that gibberellin causes increased production of lateral roots. Cytokinin stimulated root hair formation, but reduced lateral root production and elongation of the main root. Combinations of indole acetic acid, gibberellin, and kinetin produced changes in root morphology of pearl millet similar to those produced by inoculation with A. brasilense.

Cross-reactive antigens of clover roots and Rhizobium trifolii were detected on their cell surfaces by tube-agglutination, immunofluorescent, and radioimmunoassay techniques. Anti-clover root antiserum had a higher agglutinating titer with infective strains ofR. trifolii than with noninfective strains. The root antiserum previously adsorbed with noninfective R. trifolii cells remained reactive only with infective cells, including infective revertants. When adsorbed with infective cells, the root antiserum was reactive with neither infective nor noninfective cells. Other Rhizobium species incapable of infecting clover did not demonstrate surface antigens cross-reactive with clover. Radioimmunoassay indicated twice as much antigenic cross-reactivity of clover roots and R. trifolii 403 (infective) than R. trifolii Bart A (noninfective). Immunofluorescence with anti-R. trifolii (infective) antiserum was detected on the exposed surface of the root epidermal cells and diminished at the root meristem. The immunofluorescent crossreaction on clover roots was totally removed by adsorption of anti-R. trifolii (infective) antiserum with encapsulated infective cells but not with noninfective cells. The cross-reactive capsular antigens from R. trifolii strains were extracted and purified. The ability of these antigens to induce clover root hair deformation was much greater when they were obtained from the infective than noninfective strains. The cross-reactive capsular antigen ofR. trifolii 403 was characterized as a high-molecular-weight (>4.6 x 106 daltons), beta-linked, acidic heteropolysaccharide containing 2-deoxyglucose, galactose, glucose, and glucuronic acid. A soluble, nondialyzable, substance (clover lectin) capable of binding to the cross-reactive antigen and agglutinating only infective cells of R. trifolii was extracted from white clover seeds. This lectin was sensitive to heat, Pronase, and trypsin. Inhibition studies indicated that 2-deoxyglucose was the most probable haptenic determinant of the cross-reactive capsular antigen capable of binding to the root antiserum and the clover lectin. A model is proposed suggesting the preferential adsorption of infective versus noninfective cells of R. trifolii on the surface of clover roots by a cross-bridging of their common surface antigens with a multivalent clover lectin.

Quantitative microscope techniques were utilized to examine the adsorption of rhizobial cells to clover root hairs. Adsorption of cells of noninfective strains of Rhizobium trifolii or infective R. meliloti strains to clover root hairs was four to five times less than that of the infective R. trifolii strains. Attachment of the rod-shaped bacteria to clover root cells occurred in a polar, end-on fashion. Viable or heat-killed R. trifolii cells precoated with a clover lectin having 2deoxyglucose specificity had increased adsorption to clover roots. Adsorption of bacteria to roots was not increased if the clover lectin was inactivated by heat or 2-deoxyglucose treatment prior to incubation with R. trifolii. Adsorption of R. trifolii to clover root hairs was inhibited by 2-deoxyglucose (30 mM) but not by 2deoxygalactose or aD glucose. Adsorption ofR. meliloti cells to alfalfa root hairs was not affected by 2-deoxyglucose at that concentration. These results suggest that expression of host specificity in the Rhizobium-clover symbiosis involves a preferential adsorption of infective cells to clover root hairs through a 2-deoxyglucose-sensitive receptor site.

The involvement of Rhizobium enzymes that degrade plant cell wall polymers has long been an unresolved question about the infection process in root nodule symbiosis. Here we report the production of enzymes from Rhizobium leguminosarum bv. trifolii that degrade carboxymethyl cellulose and polypectate model substrates with sensitive methods that reliably detect the enzyme activities: a double-layer plate assay, quantitation of reducing sugars with a bicinchoninate reagent, and activity gel electrophoresis-isoelectric focusing. Both enzyme activities were (i) produced commonly by diverse wild-type strains, (ii) cell bound with at least some of the activity associated with the cell envelope, and (iii) not changed appreciably by growth in the presence of the model substrates or a flavone that activates expression of nodulation (nod) genes on the resident symbiotic plasmid (pSym). Equivalent levels of carboxymethyl cellulase activity were found in wild-type strain ANU843 and its pSym-cured derivative, ANU845, consistent with previous results of Morales et al. (V. Morales, E. Martínez-Molina, and D. Hubbell, Plant Soil 80:407-415, 1984). However, polygalacturonase activity was lower in ANU845 and was not restored to wild-type levels in the recombinant derivative of pSym- ANU845 containing the common and host-specific nod genes within a 14-kb HindIII DNA fragment of pSym from ANU843 cloned on plasmid pRt032. Activity gel electrophoresis resolved three carboxymethyl cellulase isozymes of approximately 102, 56, and 33 kDa in cell extracts from ANU843. Isoelectric focusing activity gels revealed one ANU843 polygalacturonase isozyme with a pI of approximately 7.2. These studies show that R. leguminosarum bv. trifolii produces multiple enzymes that cleave glycosidic bonds in plant cell walls and that are cell bound.

A central event of the infection process in the Rhizobium-legume symbiosis is the modification of the host cell wall barrier to form a portal of entry large enough for bacterial penetration. Transmission electron microscopy (TEM) indicates that rhizobia enter the legume root hair through a completely eroded hole that is slightly larger than the bacterial cell and is presumably created by localized enzymatic hydrolysis of the host cell wall. In this study, we have used microscopy and enzymology to further clarify how rhizobia modify root epidermal cell walls to shed new light on the mechanism of primary host infection in the Rhizobium-legume symbiosis. Quantitative scanning electron microscopy indicated that the incidence of highly localized, partially eroded pits on legume root epidermal walls that follow the contour of the rhizobial cell was higher in host than in nonhost legume combinations, was inhibited by high nitrate supply, and was not induced by immobilized wild-type chitolipooligosaccharide Nod factors reversibly adsorbed to latex beads. TEM examination of these partially eroded, epidermal pits indicated that the amorphous, noncrystalline portions of the wall were disrupted, whereas the crystalline portions remained ultrastructurally intact. Further studies using phase-contrast and polarized light microscopy indicated that (i) the structural integrity of clover root hair walls is dependent on wall polymers that are valid substrates for cell-bound polysaccharide-degrading enzymes from rhizobia, (ii) the major site where these rhizobial enzymes can completely erode the root hair wall is highly localized at the isotropic, noncrystalline apex of the root hair tip, and (iii) the degradability of clover root hair walls by rhizobial polysaccharide-degrading enzymes is enhanced by modifications induced during growth in the presence of chitolipooligosaccharide Nod factors from wild-type clover rhizobia. The results suggest a complementary role of rhizobial cell-bound glycanases and chitolipooligosaccharides in creating the localized portals of entry for successful primary host infection.